RalA GTPase Tethers Insulin Granules to L- and R-Type Calcium Channels Through Binding α2δ-1 Subunit

Authors


Corresponding author: Herbert Y. Gaisano, herbert.gaisano@utoronto.ca

Abstract

RalA GTPase has been implicated in the regulated delivery of exocytotic vesicles to the plasma membrane (PM) in mammalian cells. We had reported that RalA regulates biphasic insulin secretion, which we have now determined to be contributed by RalA direct interaction with voltage-gated calcium (Cav) channels. RalA knockdown (KD) in INS-1 cells and primary rat β-cells resulted in a reduction in Ca2+ currents arising specifically from L-(Cav1.2 and Cav1.3) and R-type (Cav2.3) Ca2+ channels. Restoration of RalA expression in RalA KD cells rescued these defects in Ca2+ currents. RalA co-immunoprecipitated with the Cavα2δ-1 auxiliary subunit known to bind the three Cavs. Moreover, the functional molecular interactions between Cavα2δ-1 and RalA on the PM shown by total internal reflection fluorescent microscopy/FRET analysis could be induced by glucose stimulation. KD of RalA inhibited trafficking of α2δ-1 to insulin granules without affecting the localization of the other Cav subunits. Furthermore, we confirmed that RalA and α2δ-1 functionally interact since RalA KD-induced inhibition of Cav currents could not be recovered by RalA when α2δ-1 was simultaneously knocked down. These data provide a mechanism for RalA function in insulin secretion, whereby RalA binds α2δ-1 on insulin granules to tether these granules to PM Ca2+ channels. This acts as a chaperoning step prior to and in preparation for sequential assembly of exocyst and excitosome complexes that mediate biphasic insulin secretion.

RalA is a member of the Ras superfamily of GTP-binding proteins that localizes to the plasma membrane (PM) as well as to endocytic and exocytotic granules. Targeting of exocytotic granules to the PM is thought to occur in part through specific interactions between RalA and effector proteins. RalA has previously been reported to regulate the trafficking of protein to the basolateral membrane in polarized epithelial cells [1], and the secretion of human growth hormone and norepinephrine from neuronendocrine cells [2], but also has a role in multiple cellular processes [3-6]. The identification of the mammalian tethering complex, the exocyst, as a direct effector of RalA [7], prompted investigations into the role of Ral in granule tethering at the PM. RalA is activated by Ca2+ [8, 9] and phospholipase D1, the latter being essential for catecholamine release in PC12 cells [9]. RalA activation occurs through its GTP-loading and this is thought to initiate binding of two exocyst subunits, Sec5 and Exo84 [8, 10, 11], which then act to recruit additional exocyst subunits. The RalA-dependent assembly of the exocyst complex is proposed as a rate limiting step in granule tethering at the PM and thus exocytosis.

Voltage-gated calcium (Cav) channels regulate secretion in neurons and neuroendocrine cells, including insulin-secreting pancreatic islet β-cells. Cav channels pore-forming α1 subunits, Cav1 and Cav2, exist as heteromeric complexes by their association with two auxiliary subunits, the β and α2δ subunits, which influence the trafficking of the channels to the PM and fine tune the channel biophysical properties [12-16]. In islet β-cells, L-type Cav1 channels, Cav1.2 and Cav1.3, can effect first phase insulin secretion by acting on the readily releasable pool (RRP) of insulin secretory granules (SGs) in mouse, rat and human islets [17-19]. However, genetic deletion of R-type Cav2.3 in mouse β-cell selectively suppressed second phase insulin secretion without influencing the early component of depolarization-induced insulin exocytosis [20, 21]. Thus, first and second phase insulin secretion are believed to be mediated by L- and R-type Cav channels, respectively.

In our previous study, we reported that depletion of endogenous RalA inhibited biphasic insulin secretion in mouse islets [22]. Now we reveal a major underlying mechanism by which RalA modulates this biphasic insulin release. We found that RalA can bind the α2δ-1 auxiliary subunit on the insulin SG, this complex we postulate serves to chaperone insulin SGs to PM-bound L- and R-type Ca2+ channels. RalA may then serve additional functions in the assembly of the exocyst and excitosome complexes [23, 24] required to mediate the biphasic insulin secretion.

Results

KD of endogenous RalA expression inhibits L- and R-type Ca2+ channels activities

We previously reported that KD of RalA expression in INS-1 cells could inhibit depolarization-induced insulin exocytosis of both primed RRP and refilling of the RRP [22]. This raised the possibility that RalA might be able to alter Cavs activity per se rather than only the exocytotic machinery in these RalA KD [RalA knockdown (shRNA, small hairpin RNA or short hairpin RNA)] INS-1 cells. We first show the level of reduced expression of RalA in the acute RalA KD INS-1 cells was 56% compared to control (Figure 1Ai,ii) as previously reported [22]. Confocal imaging shows the INS-1 cells transfection efficiency [tagged with green fluorescent protein (GFP)] is ∼50% (Figure 1Bi). To determine the transfected cells (tagged with GFP) are RalA KD cells, we performed immunostaining with anti-RalA antibody (secondary antibody is anti-mouse Texas Red) on INS-1 cells cotransfected with RalA shRNA and GFP. We found GFP cells did not colocalize with RalA stained cells, which means that GFP cells are depleted of RalA, whereas non-GFP cells retained abundant endogenous RalA (Figure 1Bii). We separately also generated a stable RalA KD cell line. We performed patch clamp studies on the greenest single cells from the acute RalA KD cells and the stable RalA KD cell line, which generated very similar results. Employing patch clamp recording of whole cell Cav current in stable RalA KD INS-1 cells line, we found the Cav currents amplitudes were reduced (Figure 1Ci,ii) by 39% (26.4 ± 3.1 pA/pF, n = 18; p < 0.05) compared to control stable INS-1 cells (43 ± 5 pA/pF, n = 18). L- and R-type calcium channels have been postulated to act on first and second phases of glucose-stimulated insulin exocytosis, respectively [17-21]. To identify the specific Cav channel(s) modulated by RalA KD, selective blockade of L-type (Cav1.2 and Cav1.3) and R-type (Cav2.3) Cav channels by nifedipine and SNX482, respectively, were employed. We found Cav currents in control INS-1 cells were reduced by 60% after adding nifedipine (n = 20, p < 0.05; Figure 1Di,ii,iii); whereas in acute RalA KD INS-1 cells, there were no differences in Cav currents in the presence or absence of nifedipine treatment (n = 18; Figure 1Ei–iii). R-type Ca2+ channel inhibitor SNX482 blocked Cav current by 25% reduction in control INS-1 cells (n = 19; p < 0.05; Figure 1Di–iii). However, in the acute RalA KD INS-1 cells, there were again no differences in Cav currents in the presence or absence of SNX482 treatment (n = 18; Figure 1Ei–iii). These results indicate that RalA KD-induced inhibition of insulin exocytosis as previously reported [22] is in part attributed to inhibition of L- and R-type Cav channels, which have been postulated by others to mediate first and second phase insulin secretion, respectively [19].

Figure 1.

RalA KD in INS-1 cells reduces L- and R-type voltage-gated Ca2+ currents. Ai), Representative blots showing protein expression in acute RalA KD INS-1 cells (KD). INS-1 cells were used as control. Aii) RalA blots in A quantified by densitometry scanning followed by analysis with Scion Image (n = 3, *p < 0.01). Bi) Confocal imaging shows acute RalA shRNA transfection, co-transfected with GFP to identify the RalA-shRNA transfected cells (for patch clamp), observed in 10× objective. Bii) Confocal imaging shows RalA KD (as in Bi) immunostained with anti-RalA (secondary antibody is anti-mouse Texas Red). Note GFP cells have no RalA staining. Ci) Representative traces showing Cav currents recorded in the whole-cell mode from control and RalA KD INS-1 cells. Cii) Current–voltage relationship of Cav channels. Currents were normalized to cell capacitance to yield current density. Statistical analysis shows Cav currents from RalA KD cells are strongly blocked compared to control condition. Values are means ± SEM. *p < 0.05 for control vs. RalA KD (n = 30 cells). Di) Representative Cav currents with nifedipine (10 µm) or SNX482 (100 nm) treated from a control INS-1 cell. Dii) Current–voltage relationship of Cav channels from control INS-1 cells (n = 19–20 cells). Statistical analysis shows Cav current mean amplitudes from control cells are reduced by nifedipine and SNX482. Diii) Bar chart showing maximum increase in current densities in control INS-1 cells. Results were normalized to the percentage of control value. *p < 0.05 for control vs. nifedipine-treated condition and control vs. SNX482-treated conditions respectively. Ei) Representative Cav currents with nifedipine or SNX482 treated from a RalA KD INS-1 cell. Eii) Current–voltage relationship of Cav channels from RalA KD INS-1 cells. Statistical analysis shows Cav current from RalA KD cells cannot be inhibited by nifedipine or SNX482. Values are means ± SEM (n = 18 cells). Eiii) Bar chart showing maximum increase in current densities in RalA KD INS-1 cells. Results were normalized to the percentage of control value. The patch clamp data were generated from both acute RalA KD (C) and stable RalA KD cells (D,E).

Restoration of RalA rescued the reduced Cav currents through RalA interaction with L- and R-type Cav channels

To confirm that the reduced Cav channel activities in RalA KD cells were indeed attributed to RalA per se, we infused GST-RalA (1 µm) into RalA KD INS-1 cells and rat pancreatic islet β-cells. Figure 2 shows that the maximum Cav currents density in control stable INS-1 cells (treated with control shRNA) was 37.2 ± 3.7 pA/pF (n = 20); in stable RalA KD INS-1 cells (RalA shRNA), maximum Cav currents density was 18.1 ± 1.8 pA/pF (n = 16). After restoration of RalA expression into RalA KD cells, the Cav currents were rescued and increased by 77% (32.1 ± 2.6 pA/pF, n = 14; p < 0.05), or to 86% of control INS-1 cells (Figure 2A–C). We obtained similar results with rat pancreatic islet β-cells (Figure S1). To prove functional connection between RalA protein to L- and R-type Cav channels, we infused RalA-GST into acute RalA KD INS-1 cells to restore RalA KD-reduced Cav current, and then further perfused cells with L-type Cav channels blocker, nifedipine. Here, the current was blocked by 60% again; and when perfused with R-type blocker, SNX482, the current was reduced by 25% (Figure 2D–F). These results indicate that RalA regulates Cav currents likely by direct or at least functional interaction with L- and R-type Cav channels.

Figure 2.

RalA rescues the reduced Cav currents in RalA KD INS-1 cells. INS-1 cells were transduced with either control shRNA (control) or RalA shRNA lentivirus (RalA KD) for 48 h. Cav currents were then recorded from control, RalA KD and RalA rescued INS-1 cells. In the latter, GST-RalA (1 µm) was infused by patch pipette into the RalA KD INS-1 cells. A) Representative Cav currents from control, RalA KD and RalA rescue INS-1 cells. B) Current–voltage relationship of Cav channels. Data normalized to respective cell size. C) Bar chart showing Cav current maximum amplitudes are reduced by RalA KD and recovered by infusing GST-RalA into cells. Values are means ± SEM (n = 14–20 cells), *p < 0.05 for control vs. RalA KD; RalA KD vs. RalA rescue, respectively. D) Representative of RalA rescued Cav current with or no nifedipine. E) Representative of RalA rescued Cav current with or no SNX482. F) Bar chart showing maximum increase of RalA rescued Cav current in RalA KD INS-1 cells. Results were normalized to the percentage of control value (n = 10 cells). *p < 0.05 for the RalA rescued Cav current with or no nifedipine; the RalA rescued Cav current with or no SNX482, respectively. These data were generated from both acute RalA KD cells (A,B,C) and stable RalA KD cells (D,E,F).

RalA is physically associated with the auxiliary subunit α2δ-1

We postulated that the functional effects of RalA on the Cav currents could be by their physical interactions. We employed protein binding and pull down assays of RalA with Cav1.2, Cav1.3, Cav2.3, α2δ-1 and β3 subunits expressed in HEK293 (Figure 3A) and INS-1 cells (Figure 3B). As shown in Figure 3A with transfected HEK293 cells, GST-RalA (bound to agarose glutathione beads) was able to pull down the expressed α2δ-1, but not other Cav subunits (β3, Cav1.2, Cav1.3 and Cav2.3). Figure 3B shows the same result with INS-1 cells, wherein GST-RalA pulled down endogenous α2δ-1 but not β3, Cav1.2, Cav1.3 or Cav2.3. We did not observe any significant interaction with any subunits using GST alone (Figure 3C). These results show that RalA predominantly binds to the α2δ-1 subunit. To further confirm if RalA is physically associated with α2δ-1 subunit, we performed immunoprecipitation experiments on INS-1 (Figure 4A) and HEK293 cells (Figure 4B). pcDNA3.1/Myc-RalA transfected INS-1 cells were kept in either non-stimulatory condition (0.8 mm glucose) or stimulated with 16.7 mm glucose plus 10 nm GLP-1. 400 µg of proteins from each condition was then subjected to immunoprecipitation (left panels) with antibody against Myc and probed with anti Cav channel subunits antibodies. These results show that RalA interacts with α2δ-1 and Cav1.2 Ca2+ channel subunits upon glucose stimulation in INS-1 cells. To further confirm if RalA interacts Cav1.2 directly or through α2δ−1 subunit, Myc-RalA and α2δ-1 (Figure 4Bi) or Myc-RalA and Cav1.2 (Figure 4Bii) were coexpressed in HEK293 cells, and the Myc-RalA proteins immunoprecipitated, and co-precipitated proteins probed with anti α2δ-1 or Cav1.2 antibodies (left panels). The results show that RalA interacts with α2δ-1 subunit not with Cav1.2 subunit in the HEK293 cells. Taken together, these results indicate that RalA is physically associated with the α2δ-1 auxiliary subunit, which in turn suggests that RalA-α2δ-1 binding could be acting to chaperone insulin SGs to L- (Cav1.2, Cav1.3) and R-type (Cav2.3) calcium channels on the PM.

Figure 3.

RalA binds to α2δ-1 subunit. A) HEK293 cells were transfected with Cav1.2, Cav1.3, Cav2.3, α2δ-1 and β3 plasmid DNAs, respectively. GST (as a negative control) and GST-RalA (both bound to agarose glutathione beads, 350 pM protein each) were used to pull down the overexpressed proteins from transfected HEK293 cell lysate extracts (400 µg protein). HEK293 cell lysate extracts (20 µg protein) were used as positive controls. B) GST and GST-RalA were used to pull down the endogenous Cav1.2, Cav1.3, Cav2.3, α2δ-1 and β3 proteins from INS-1 cell lyaste extracts (450 µg protein), as indicated. INS-1 cell lysate extract (20 µg protein) was loaded as positive control. In (A) and (B), shown are representative blots (n = 3). Molecular mass markers (kDa) are indicated on the left. C) Ponceau S staining of the blot to demonstrate the amounts of GST and GST-RalA loaded.

Figure 4.

Myc- RalA co-immunoprecipitation with α2δ-1 subunit. A) RalA interacts with Cav-α2δ-1 and Cav1.2 Cav channel subunits upon glucose stimulation in INS-1 cells. pcDNA3.1/Myc-RalA transfected INS-1 cells were kept in either non stimulatory condition (0.8 mm glucose) or stimulated with 16.7 mm glucose + 10 nm GLP-1 as described in Materials and Methods. 400 µg of proteins from each condition were then subjected to immunoprecipitation (left panels) with antibody against Myc and probed with anti Ca2+ channel subunit antibodies. Corresponding right panels show ‘Input’ controls (25 µg protein, total INS-lysates), confirmed the expression of similar levels of RalA-Myc and other Ca2+ channel subunit proteins. Results shown are representative of 3 independent experiments. B) RalA interacts directly with α2δ-1 but not with Cav1.2 Ca2+ channel subunit. Myc-RalA and α2δ-1 or Myc-RalA and Cav1.2 were coexpressed in HEK293 cells by transfecting (i) pcDNA3.1/Myc-RalA and pcDNA3.1-Cavα2δ-1 or (ii) pcDNA3.1/Myc-RalA and pRcCMV-Cav1.2. Myc-RalA proteins were immunoprecipitated from 48 h post-transfected cells and probed with anti α2δ-1 or Cav1.2 antibodies (left panels). Corresponding right panels show ‘Input’ controls (25μg protein, total HEK293 lysates), confirmed the expression of similar levels of RalA-Myc, Cav2δ-1 and Cav1.2 and other Ca2+ channel subunit proteins. Results shown are representative of 3 independent experiments.

RalA KD disrupts α2δ-1 localization to the insulin SG and reduces α2δ-1 expression in insulin SGs

If RalA binding to α2δ-1 mediates localization of this auxiliary Cav subunit to insulin SGs, then reduced expression of RalA should relieve the trafficking of α2δ-1 to the SGs. Indeed, in RalA KD rat β-cells (Figure 5A, confocal microscopy), α2δ-1 colocalization with insulin SGs was reduced according to Pearson's coefficient analysis with the values reduced from 0.75 ± 0.03 (n = 12) to 0.47 ± 0.03 (n = 13; p < 0.001). We also noted the fluorescence intensity of α2δ-1 to have become reduced by 27% (p < 0.001), with all antibody labeling and imaging parameters always held constant. In contrast, insulin SG colocalization with Cav1.2, Cav1.3 and Cav2.3 were lower at ∼0.4, and may indicate the subpopulations of insulin SGs docked on these PM-bound Cavs. The auxiliary β3 subunit colocalization with insulin granules was very low at 0.25. Importantly, RalA KD did not affect the colocalization of β3, Cav1.2, Cav1.3 and Cav2.3 with insulin SGs, and had no effect on their relative fluorescence intensities (Figure S2). To examine this more quantitatively and unequivocally, we performed subcellular fractionation of the control and RalA KD INS-1 cells and western blotting analysis of the RalA and α2δ-1 distributions in PM and insulin SG fractions. First, we assessed the levels of RalA and other proteins of interest in whole cell lysates. As shown in Figure 5B i,ii, RalA expression in KD cells reduced to 51.6% of that in control cells (n = 3, p < 0.01) but the levels of Cav1.2, Cav1.3, Cav2.3 and α2δ-1 did not change. After fractionation, we validated the purity of the PM and SG fractions by enrichment of PM marker proteins (Na+-K+ ATPase and Syn-1A) in the PM fraction, and SG marker protein VAMP2 in the SG fraction (Figure 5C), their low levels in the non-cognate subcellular compartments, and that their levels in these compartments were not affected by RalA KD. Next, we examined if RalA KD would affect α2δ-1 levels in the SG and PM compartments. Figure 5Ei,ii demonstrates that RalA levels were reduced by 24.4% (n = 3, p < 0.05) in PM and more so by 43.8% (n = 3, p < 0.01) in SGs from RalA KD cells, respectively, as compared with control cells. From the same RalA KD cells, α2δ-1 level did not change in the PM fraction but was decreased by 36.9% in the SG fraction (n = 3, p < 0.01), which is closed to the 44% reduction of RalA levels. Also, we assessed the levels of α1 subunits proteins (Cav1.2, Cav1.3 and Cav2.3, in Figure 5D) levels in the PM fraction (no signal could be detected in SG fraction). None of these proteins changed in the PM from RalA KD cells as compared with control cells (Figure 5D). Taken together, these results suggest that RalA binding to α2δ-1 may in part serve to direct trafficking of insulin SGs toward PM Cav-α1 subunits which would then engage the SG α2δ-1, thus assist in tethering insulin SGs to PM-bound Cavs.

Figure 5.

RalA KD disrupts α2δ-1 localization to insulin granules in rat β-cells and reduces α2δ-1 expression in insulin granules in INS-1 cells. Correlation of the colocalization of α2δ-1 (labeled with anti-rabbit Cy5 antibody) and insulin granules (tagged with intracellular marker Ad-IAPP-mCherry) in control and RalA KD rat islet β-cells. Purple in Merge and PMC (positive merged channel) images indicate the sites of colocalization of α2δ-1 with insulin granules. Ai) This shows representative images in control and RalA KD rat β-cells; scale bar, 5 µm; (Aii) shows the quantification of colocalization values of α2δ-1 subunit with insulin granules using Pearson's coefficient analysis; and (Aiii) shows the quantification of the relative fluorescence intensity values. Bi) Western blotting analysis of the proteins of interest, as indicated, in whole cell lysates (28 µg protein/lane) from the control and RalA KD cells. Shown are representative blots of three separate experiments. Bii) Quantitative summary of the RalA blots in Bi by densitometry scanning followed by analysis with Scion Image (n = 3). C) The same cells as in Bi were fractionated and the prepared plasma membrane (PM) and secretory granule (SG) fractions were validated by immunodetection of PM marker proteins (Na+-K+ ATPse and Syn-1A) and SG protein (VAMP2). D) Representative blots showing the α1 subunit protein expression in PM fraction (12 µg protein/lane). Ei) Representative blots showing RalA and α2δ-1 levels in PM and SG fractions (12 µg protein/lane). Eii) Quantitative summary of the blots in (Ei) (n = 3). *, p < 0.05 as compared with control; **, p < 0.01. These data were generated from acute RalA.

RalA modulates Cav channel activity through RalA interactions with α2δ-1

We next determined the functional coupling of RalA and α2δ-1 in regulating Cav currents. We first examined the effect of α2δ-1 KD (α2δ-1 shRNA) on Cav currents in INS-1 cells (Figure 6), which caused Cav currents to be reduced by 54% from 46.8 ± 7.1 pA/pF (control cells, n = 14) to 21.5 ± 3.7 pA/pF (n = 13; p < 0.01). However, in stable RalA KD INS-1 cells (27.2 ± 3.5 pA/pF, n = 13), double KD of α2δ-1 expression (DbKD) by ∼54% (Figure 6A,B) did not further reduce the Cav currents (23.2 ± 2.3 pA/pF, n = 16; Figure 6C–E). Notably, intracellular infusion of GST-RalA rescued RalA KD-induced inhibition of Cav currents in single RalA KD INS-1 cells (41.6 ± 5.8 pA/pF in single RalA KD cells, n = 11; Figure 6C–E), but not in double α2δ-1 and RalA KD cells (25.8 ± 3.7 pA/pF, n = 13; Figure 6C–E). To determine the direct interaction of RalA with α2δ-1 subunit, we performed fluorescence resonance energy transfer (FRET) imaging assays (Figure 7). INS-1 cells were co-transfected with RalA-mCherry and α2δ-1-pHluorin, in which α2δ-1-pHluorin was used as FRET donor and RalA-mCherry as FRET acceptor in each FRET experiment. Figure 7A shows the representative recordings of FRET signals on the PM of INS-1 cells in non-stimulatory condition with 0.8 mm glucose (left) and the same cell which was stimulated by 16.7 mm glucose plus 10 nm GLP-1 for 10 min (right). Figure 7B shows the FRET efficiency from 3.87 ± 0.41% in control resting condition increased to 19.4 ± 2.6% after stimulation with glucose and GLP-1 (n = 11; p < 0.001). The FRET efficiency indicates this very close proximity and biochemical interactions between RalA and α2δ-1 proteins could be induced by physiologic stimulation and thus their presumed physiologic interactions. Taken together, these results are strong evidence confirming that RalA modulates Cav channel activity likely through direct interactions with α2δ-1 auxiliary subunit, and such interactions may be required for subsequent trafficking and engagement of α2δ-1 with PM-bound Cavα1 subunits to fully influence Cav channel gating properties.

Figure 6.

RalA and α2δ-1 are functionally coupled in regulating Cav channel activity. Depletion of α2δ-1 subunit greatly reduced Cav currents in control INS-1 cells, but not in stable RalA KD INS-1 cells; and RalA KD-induced inhibition of Cav currents cannot be recovered in α2δ-1 simultaneously KD in RalA KD INS-1 cells. A) Representative western blotting shows α2δ-1 level in shRNA control and shRNA α2δ-1 INS-1 cells. β-actin was used as protein loading control. B) The blots were quantified by densitometry scanning followed by analysis with Scion Image (n = 3). Molecular mass markers (kDa) are indicated on the left. *p < 0.01 as compared with control. C) Representative Cav currents from INS-1 cells of control, α2δ-1 KD, RalA KD, RalA KD + GST-RalA, α2δ-1 KD + RalA KD (DbKD) and α2δ-1 KD + RalA KD + GST-RalA (DbKD+GST-RalA), respectively. D) I–V relationship of Cav channels in (C). E) Summary data of the current densities of (D). *p < 0.05 for the indicated comparisons.

Figure 7.

RalA interacts with auxiliary subunit α2δ-1 in living cells analyzed by FRET. A) Representative recordings of FRET signals on the plasma membrane of INS-1 cells expressing RalA-mCherry and α2δ-1-pHluorin under the resting condition with 0.8 mm glucose (Left) and the same cell which was stimulated by 16.7 mm glucose and 10 nm GLP-1 for 10 min (Right). The excitation laser wavelength was 488 nm and the corresponding FRET signal was collected at 605–655 nm. The vertical scale bar indicates 5 µm. Vertical scale bar indicates FRET efficiency in pseudocolor. B) Summary of FRET efficiency before and after stimulation. Bar graphs shown as mean ± SEM. n = 11. ***p < 0.001.

Discussion

There has been much effort to define the mechanism by which docked SGs become primed and undergo exocytotic fusion. However, little is known about how SGs are recruited to become morphologically docked and tethered to the PM. SG tethering at the PM occurs following the transport of SG along cytoskeletal motors. This tethering step is thought to involve a physical but reversible interaction between SGs and proteins associated with the PM, in a step that precedes SNARE complex assembly [11, 25, 26]. RalA has previously been shown to play a central role in SG tethering through its regulated association with the exocyst complex. We had previously reported that RalA-GTP associates with the exocyst in mouse pancreatic β-cells and that depletion of endogenous RalA protein expression in mouse pancreatic islets dramatically inhibited biphasic insulin secretion. Furthermore, patch clamp capacitance experiments of RalA KD in INS-1 cells showed that the inhibition of insulin secretion is directly attributed to the reduction in the exocytosis of the RRP and RRP refilling [22]. Depolarization by KCl or patch clamp reduced insulin exocytosis in RalA KD cells, demonstrating that RalA-dependent exocytosis requires Ca2+. We have now defined an additional mechanism for RalA function in insulin secretion. In this study, we found RalA KD inhibited Ca2+ channels activity (Figure 1), suggesting that RalA interacts with Ca2+ channels. Notably, residual Ca2+ currents in RalA KD cells could not be further reduced by nifedipine (inhibits L-type Cav1.2 and Cav1.3 Ca2+ channels that mediates first phase insulin secretion) [17-19] or SNX482 (inhibits R-type Cav2.3 Ca2+ channels that mediates second phase insulin secretion) [20, 21]. Moreover, RalA reconstitution could rescue Cav currents (Figures 2 and S1), and restoration of RalA KD-induced Cav current could be further blocked by nifedipine and SNX482 at the same level as control conditions (Figure 2). We propose that RalA acts to chaperone insulin SGs to L- and R-type Ca2+ channels on the PM, where SGs become tethered, in preparation for the formation of excitosome complexes with vesicle- and PM SNARE proteins [23, 24]. RalA would then interact with the exocyst components on the SG and PM to complete the process of SG tethering to the PM. Much further work will be required to prove this postulated sequence of events.

RalA is physically associated with the auxiliary subunit α2δ-1 (Figures 3 and 4). RalA KD would disrupt α2δ-1 localization to SGs and reduce α2δ-1 expression in insulin SGs (Figure 5) and disrupt their functional coupling in regulating Cav currents (Figure 6). RalA reconstitution could rescue and fully restore Cav currents across the PM (Figures 2 and S1) but not when both RalA and α2δ-1 are knocked down (Figure 6). FRET efficiency assay shows that RalA could interact with α2δ-1 (Figure 7). These results taken together indicate that the RalA-α2δ-1 complex on SGs likely engage and traffic SGs to the PM-bound Cavα1 subunits (Cav1.2, Cav1.3 and Cav2.3) to be able to then conduct Ca2+ entry across fully functioning L- and R-type Cav channels (Figures 2 and S1) at precisely the sites of SG exocytosis with PM. This may also be a mechanism by which RalA on insulin SGs acts to assist in chaperoning Ca2+ channel subunits to the PM to complete Ca2+ channel assembly. Since L- and R-type Cav channels mediate first and second phase insulin secretion, respectively, this would potentially explain how RalA influenced biphasic insulin secretion as we previously reported [22]. Diabetic metabolic insults such as high fat diet and hyperglycemia recently shown to cause functional uncoupling of Ca2+ channels from insulin SGs [27], could in part be explained by a possible disruption of RalA–exocyst interactions that follows after RalA had already completed its actions on Ca2+ channel assembly, the latter thus preserving Ca2+ channel activity [27]. RalA-based strategies could therefore provide potential novel therapies directed at optimizing these sequential steps of the secretory process to treat the deficient biphasic insulin secretion in type 2 diabetes.

The auxiliary subunit α2δ plays a key role in trafficking Cav1 and Cav2 channels to the PM and modulating channel gating. The α2δ subunits promote trafficking of the calcium channel complex by a mechanism that involves their Von Willebrand factor-A (VWA) domain [15]. Topological analysis supports a model in which α2 is entirely extracelluar and δ has a single transmembrane region with a short intracellular domain [28]. α2 is extensively glycosylated, a post-translational modification important in maintaining the stability of the interaction with α1 and is a major determinant of the protein's ability to stimulate the current amplitude [28, 29]. More work will be required to see how RalA influences the assembly of α2δ with the α1-subunits to form functional ion channels on PM. Nonetheless, this work provides evidence that a contributing mechanism is RalA binding to α2δ-1 on insulin SGs which facilitate trafficking of insulin SGs to tether to PM-bound L- and R-type calcium channels to form excitosome complexes [23] ready to evoke first and second phase insulin release. Nonetheless, the full demonstration of the model we have proposed here will require much future studies directed at identifying the precise interacting sites within RalA and α2δ-1 by mutational analysis, and the functional sequelae on Cav kinetics and biphasic insulin secretion.

Materials and Methods

Islets isolation and cells culture

Islets from male Sprague–Dawley rats (300–325 g) were isolated by collagenase digestion method as previously described [30]. The islets were dispersed into single cells using a Ca2+/Mg2+−free phosphate-buffered saline (at 5 mm EDTA) with 0.25 mg/mL trypsin at 37°C for 5 min with gentle shaking and then resuspended in enriched RPMI-1640 media containing 11 mm d-glucose. The resulting cell suspensions were plated on glass coverslips and allowed to adhere ∼48 h before experiments. INS-1 832/13 cells were cultured in RPMI-1640 medium (GIBCO) with 10% FBS and penicillin/streptomycin at 37°C in an atmosphere of 5% CO2. Rats were housed on a 12-h light/dark cycle and were allowed free access to standard rat food and water. All experimental procedures have been approved by the Animal Care Committee of the University of Toronto.

Lentiviral transduction

Lentivirus construction was previously described [22]. Rat pancreatic β-cells and INS-1 832/13 cells were transduced with concentrated control or RalA shRNA lentivirus (RalA knockdown or KD) for 48 h. INS-1 stable RalA KD cell lines were established by selection of GFP-positive cells using flow cytometry (BD Biosciences). Entry of viral particles after transduction was determined by GFP expression observed by epiflorescence imaging on Nikon TE2000U inverted microscope.

Electrophysiology

Recording pipettes were from 1.5-mm borosilicate glass capillary tubes using a programmable micropipette puller. Pipettes were then heat-polished and tip resistances ranged from 2 to 3 MOhm when filled with intracellular solution. For measurement of Cav currents, barium was used as charge carrier, pipettes were filled with (mm): 120 CsCl, 20 tetraethylammonium chloride, 5 EGTA, 5 MgATP and 5 HEPES (pH 7.2). The external solution containing in mm: 100 NaCl, 20 BaCl2, 20 tetraethylammonium chloride, 4 CsCl, 1 MgCl2, 10 glucose and 5 HEPES (pH 7.4). L-type Cav channel inhibitor nifedipine (10 µm) and R-type Cav channel inhibitor SNX482 (100 nm) are from Sigma (Sigma-Aldrich). Cells were held at −70 mV for 2 min after formation of whole-cell mode, and currents elicited by step 300  milliseconds depolarizations from −60 to +60 mV in 10 mV increments. Recordings were conducted using an EPC10 patch clamp amplifier and the Pulse and X-Chart software programs (HEKA Electronik).

Confocal microscopic Imaging

Immunostaining cells mounted on glass coverslips were examined using a Leica DMIRE2 inverted fluorescence microscope (Leica Microsystems) equipped with a Hamamatsu Back-Thinned EM-CCD camera (Hamamatsu Crop.) and spinning disk confocal scan head. The unit is equipped with four separate diode-pumped solid state laser lines (405, 491, 561 and 638 nm, from Spectral Applied Research, Spectral Applied Research Inc.), an ASI (Applied Scientific Instrumentation, Inc.) motorized XY stage and an Improvision Piezo Focus Drive (for Z-scan, 0.1 µm step) (PerkinElmer). Data acquisition and analysis were performed using Volocity software (PerkinElmer). In our experiments, we chose the 63× objective and the 491, 561, and 638 nm laser to excite the Fluorescein isothiocyanate (FITC), Texas Red and Cy5 dyes, respectively. All the images were subjected to deconvolution to remove the background noise. The following primary antibodies were used: rabbit anti-α2δ-1 (GenWay); rabbit anti-β3, rabbit anti-Cav1.2 and rabbit anti-Cav1.3 (all from Alomone Labs); rabbit anti-Cav2.3 (Millipore); guinea pig anti-insulin (DakoCytomation) and mouse anti-RalA (BD Biosciences). The second antibodies were sheep anti-mouse FITC (Serotec); goat anti-rabbit Texas Red (Molecular Probes); donkey anti-guinea pig Texas Red (Bio/Can Scientific) and goat anti-rabbit Cy5 (Bio/Can Scientific). Adenovirus IAPP-mCherry labeling insulin granules was a gift from Dr. M. Hoppa (University of Oxford, Oxford, UK).

A quantitative assessment of fluorophore colocalization in confocal optical sections can be obtained from the selected regions of interest. Two values are generated: one is Pearson's correlation coefficient, which is one of the standard techniques applied in pattern recognition for matching one image to another in order to describe the degree of overlap between the two patterns [31]. Pearson's correlation coefficient is calculated according to the equation [32], where Ch1i is the signal intensity of pixels in the first channel and Ch2i is the signal intensity of pixels in the second channel. Ch1ave and Ch2ave are average values of pixels in the first and second channel, respectively. Another parameter is the overlap coefficient [31], which is defined as:

display math

Constructs and recombinant GST-fusion proteins

pGEX-KG-RalA and RalA shRNA were from Dr. S. Sugita (a co-author, [2]), pcDNA3.1-Cavα2δ-1,pcDNA3.1-Cavβ3, pcDAN6-Cav1.3 and α2δ-1-pHluorin from Dr. D. Lipscombe (co-author, Brown University, Providence, RI, USA), pRcCMV-Cav1.2 from Dr. M. Hosey (Northwestern University, Chicago, IL, USA), pcDNA3.1-Cav2.3 from Dr. B. Adams (Utah State University, Logan, UT, USA) and Myc-RalA from Dr. D. James (co-author, Garvan Institute of Medical Research, Sydney, New-South Wales, Australia). RNA interference (RNAi)-mediated KD of endogenous Cavα2δ-1 subunit was from GeneCopoeia. Scrambled shRNA control was used as a negative control. INS-1 cells were transfected with RalA shRNA and Cavα2δ-1 shRNAs using Lipofectamine 2000 Transfection Reagent according to manufactures instructions (Invitrogen). GST-fusion protein expression and purification were performed following manufacturer's instructions (Amersham Biosciences).

In vitro binding assay and western blotting

In vitro binding assays were performed according to the method we previously described [33]. Briefly, GST (as a control), GST-RalA (350 pmol protein each) were bound to glutathione agarose glutathione beads and incubated with INS-1 cells lysate extract (450 µg protein) or transfected HEK293 cells lysate extract (400 µg protein) in cell lysate buffer [25 mm HEPES (pH7.4), 100 mm KCl, 1.2 mm MgCl2, 1.5% Triton X-100, 1 µg/mL leupeptin, 10 µg/mL aprotinin and 1 µg/mL pepstatin A] at 4°C for 2 h with constant agitation. The beads were then washed three times with cell lysate buffer. The samples were then separated on 10 or 12% SDS–PAGE, transferred to nitrocellulose membrane and identified with specific primary antibodies against Cavα2δ-1 (1:200, Genway Biotech, Inc.), Cavβ3 (1:200, Alomone Labs), Cav1.2 (1:200, Alomone Labs), Cav1.3 (1:200, Alomone Labs) or Cav2.3 (1:200, Millipore). The blots were quantified by densitometry scanning followed by analysis with Scion Image (release beta 4.0.2) (Scion Corp.).

Immunoprecipitation

INS-1 and HEK cells were grown in antibiotic free media to ≈60% confluence in 65 mm culture plates and were transfected with the 3 µg of pCDNA3.1-RalA with Lipofectamine reagent according to the manufacturer's instructions. For stimulation, 48 h transfected INS-1 cells were first washed with 1× PBS and incubated for 30 min at 37°C in Krebs–Ringer HEPES buffer (KRH, 125 mm NaCl, 5.6 mm KCl, 1.28 mm CaCl2, 5.0 mm Na2CO3, 25 mm HEPES, pH 7.4. with 0.1% BSA) at 0.8 mm glucose to obtain uniform basal conditions. The INS-1 cells were then subjected to two conditions. For the stimulated condition, INS-1 cells were first preincubated for 30 min with 10 nm GLP-1 at basal glucose concentration (0.8 mm glucose) and then stimulated with 16.7 mm glucose with 10 nm GLP-1 for 1 h. For the control basal condition, INS-1 cells were subjected to parallel changes of KRH buffer containing basal (0.8 mm) glucose concentration. Treated cells were then harvested and lysed by sonication in lysis buffer (25 mm HEPES, 100 mm KCl, 1.5% Triton X 100 with protease inhibitors). For immunoprecipitation, 400 µg of protein extract from each condition were initially precleared with 50 μL of protein G-sepharose beads (Molecular Probes) for 2 h at 4°C and then subjected to immunoprecipitation with 2 µg of mouse monoclonal Myc antibody (Sigma) bound to 50 μL protein G-sepharose beads incubated overnight at 4°C. Beads were washed twice with lysis buffer and the co-precipitated proteins separated on SDS–PAGE and identified by western blotting.

Subcellular fractionation of INS-1 cells

Subcellular fractionation of INS-1 cells was performed using the method previously described by Dotta et al. [34]. Briefly, the cells were washed three times with ice-cold homogenization buffer (HB; 5 mm HEPES ,250 mm sucrose, 0.5 mm EGTA, 0.1 mm PMSF, 2 µg/mL leupeptin and 10 µg/mL aprotinin, adjusted to pH 7.4 with KOH) and then harvested in HB. The cells were disrupted by 10 strokes through 27 G needles. The homogenates were centrifuged at 700 ×g for 15 min at 4°C to remove the nuclei and unbroken cells. The supernatants were mixed with Percoll (GE Healthcare) and sucrose to give a final concentration of 15% v/v Percoll and 250 mm sucrose. After centrifugation of the samples at 48 000 ×g for 25 min at 4°C in a fixed angle rotor (Beckman, TI 50), two opaque bands were visible at the top and bottom corresponding to the PM and SGs, respectively. These bands were collected with a pipette and then washed twice with 4 volumes HB and re-centrifuged at 150 000 ×g at 4°C for 30 min. The fractions were re-suspended in lysis buffer (25 mm HEPES, 100 mm KCl, 1.5% Triton X-100, 2 µg/mL leupeptin and 10 µg/mL aprotinin, adjusted to pH 7.4 with NaOH) and sonicated for western blotting analysis.

FRET imaging

FRET imaging assay was performed as we recently reported [35, 36]. INS-1 cells were sequentially transfected with RalA-mCherry for 72 h and α2δ-1-pHluorin (from D. Lipscombe and A. Andrade, Brown University, Providence, RI, USA) for 12 h, then the cells were seeded the on the autoclaved glass coverslips and imaged in intracellular buffer (IB) (20 mm HEPES, 5 mm NaCl, 140 mm potassium gluconate and 1 mm MgCl2, and pre-equilibrated with 95:5 O2:CO2, pH 7.4) by using the total internal reflection fluorescent microscope (TIRFM). This TIRFM system was composed by Nikon TE-2001U inverted microscope, a Plan apo × 60 oil immersion objective (1.49 NA), an argon laser unit (480 ± 10 nm; Spectral-Physics), a helium-neon laser units (545 ± 10 nm; Melles Griot), and a cooled EM-CCD (Quantum 512SC, Photometrics) and was driven by NIS software (Nikon). The penetration depth of the evanescent field for both lasers was adjusted to around 150 nm, which allowed us to only record the molecular interactions on the surface of the PM and avoid contamination from intracellular FRET signals. This TIRFM system, combined with the Dual-View imaging system (Optical-Insights by Photometrics) which contains a dual-band dichroic splitter (565dcxr) and an emission filter (HQ530/30 for pHluorin and HQ630/50 for mCherry, enabled simultaneous monitoring of two-channels of emission fluorescence. All images were acquired at room temperature. For each FRET experiment α2δ-1-pHluorin was used as FRET donor and RalA-mCherry as FRET acceptor. Four images including donor excitation/donor emission (Dd), donor excitation/acceptor emission (Da), acceptor excitation/ acceptor emission (Aa), and acceptor excitation/donor emission (Ad) were acquired under absolutely the same conditions. And donor only and acceptor only samples were acquired with the same setting at the beginning of each experiment for bleed through calculation. FRET efficiency was used to indicate the interaction of the two proteins, calculated as: FRET efficiency % = {[(FRETraw − CoB × DdFRET) − CoA × AaFRET]/DdFRET} × 100%, where CoB is the amount of donor bleed through in the absence of an acceptor and CoA is the amount of acceptor bleed through in the absence of a donor. For statistical analysis of FRET efficiency, we draw ROIs (region of interests) around the entire area of the PM surface expressing any FRET signal (blue to green to red, see pseudocolor bar) as indicated, and calculated the average FRET efficiency; then we tracked the changes of FRET efficiency in the same ROI area under the indicated different conditions.

Statistical analysis

Statistical comparison was performed with unpaired two-tailed Student's t-test. All data are presented as mean ± SEM and considered significant if p < 0.05.

Acknowledgments

This work was supported by a grant from the Canadian Institutes for Health Research (CIHR MOP 86544) to H. G., and a Postdoctoral Fellowship from the Canadian Diabetes Association to L. X. We are very grateful to Diane Lipscombe and Arturo Andrade from Brown University for the kind gift of the α2δ-1-pHluorin construct and advice. We thank Yunfeng Liu for rat islets isolation.

Ancillary