Chick CaV2.1 C-terminal constructs C1-C2 (residues 1748–2127), C1 (residues 1743–1898), and C2 (residues 1899–2100) were generated using RT-PCR. Gene-specific forward and reverse primers (GSPs) were designed using the chick CACNA1A (CaV2.1 alpha 1 subunit) C-terminal nucleotide sequence (Supplementary Table S1). The primers ranged from 17 to 41 nucleotides, contained at least five nucleotides that were different than the chick CACNA1B (CaV2.2 alpha 1 subunit) C-terminal sequence, and had a guanine-cytosine (GC) ratio of ~3:2. Each RT-PCR reaction consisted of 10 μL primer solution and 90 μL of double-distilled H₂O.

Fusion protein constructs encoding the CaV2.1 C-terminal regions were sub-cloned into the pGEX-KG vector as described (Chan et al., 2007). All constructs were transformed into E. Coli BL21 (DE3) cells (New England Biolabs, Ipswich, MA, USA), as previously described (Chan et al., 2007). In frame DNA sequence of each construct was confirmed by dideoxy sequencing (ACGT, Toronto, ON, Canada). All fusion protein constructs were transformed into DH5α competent cells (Invitrogen, Carlsbad, CA, USA) to confirm that the DNA sequence was in frame (ACGT, Toronto, ON, Canada).

The CaV2.2 distal C terminal fusion protein, C3 (Figure 1A), was used as a positive control (see Wong et al., 2014; where it was termed C3_WildF).

Fusion proteins were purified using standard procedures (Chan et al., 2007; Gardezi et al., 2013). Briefly, proteins were expressed in BL21(DE3) cells and induced using isopropyl-β-D-thiogalactopyranoside (IPTG). Bacterial cell pellets were lysed using 1× PBS (Thermo Fisher Scientific, Waltham, WA, USA) supplemented with 1% Tween-20 (Bio-Rad), 0.01% Beta-mercaptoethanol, protease inhibitor cocktail (Sigma Aldrich), and 0.1 mM phenylmethanesulfonyl fluoride (PMSF) followed by sonication on ice using cycles of 10 s burst and 10 s rest. Cell lysate was centrifuged at 2400× g for 15 min. The supernatant was incubated with glutathione sepharose 4B beads (GE Healthcare, Little Chalfont, UK) for 2–3 h at 4°C. Immobilized fusion proteins were washed 3× with PBS supplemented with 0.05% Tween-20, 2× with PBS, 1× with HB (0.32 M sucrose, 2 mM EDTA, 10 mM HEPES; pH 7.4) and 2× with SV-PD buffer (0.32 M sucrose, 2 mM EDTA, 5 mM EGTA, 0.2857 mM CaCl2 (to give 10 nM Ca²⁺), 10 mM HEPES pH 7.4). Fusion proteins were used on bead for SV-PD assays. All the steps were carried out on ice and buffers were supplemented with protease inhibitor cocktail and 0.01 mM PMSF.

Two new antibodies were generated for this study: PmidC2 and PC2var. PmidC2 antibody was generated against a LGTDLSVTTQSGDLPS peptide replicating a sequence in the middle region of the chick CaV2.1 C-terminal, while PC2var antibody was generated against a RRKVRPRGNNL peptide replicating the E44 alternative splice sequence (see “Results” section). These peptides were made with a terminal cysteine to permit conjugation to a carrier protein, KLH (Biomatik, Cambridge, Canada). Immunization of rabbits were done as previously described (Chen et al., 2017). All antibodies used in the present study are listed in Table 1.

The synaptosome and SV fractionation method has been described in detail (Gardezi et al., 2016). Briefly, E14-E17 chick brains (typically 105 brains per preparation) were homogenized. Synaptosomes were isolated by differential sucrose density gradient centrifugation and then ruptured by osmotic shock to release their contents. List of Antibodies used in this study. SVs were then isolated and purified by differential sucrose density gradient centrifugation. The SVs were maintained intact in detergent-free buffer for all experiments. Key buffers were: homogenization buffer (HB), 0.32 M sucrose, 10 mM HEPES, 2 mM EDTA, pH 7.4; and osmotic rupture buffer, 50 mM HEPES, 2 mM EDTA, pH 7.4 (Wong et al., 2014; Gardezi et al., 2016).

Standard Western blotting (WB) method was carried out as described previously (Wong et al., 2013, 2014) and immunoblots were imaged using a ChemiDoc™ XRS System (Bio-Rad, Hercules, CA, USA) as described (Gardezi et al., 2016).

It was essential to ensure that the concentrations of fusion protein on the beads were comparable. As a first step, for each fusion protein we added a range of concentrations to the precipitation beads and used Coomassie staining to identify a concentration that did not saturate the gel and was reasonably comparable to that of our other proteins (data not shown). In initial SV-PD experiments we fine-tuned the method by running several (2–4) concentrations of the new fusion protein. SV-PD analysis was only carried out on lanes where the Coomassie fusion protein bands were comparable in intensity. Thus, each experiment had up to 12 actual SV-PD trials (for example see Supplementary Figure S1) but only the lanes with comparable fusion protein concentration were used for analysis. The blot was imaged at a range of exposure durations and the duration that had the darkest bands without saturation of the SV-PD lanes was selected for further analysis. Any contrast adjustments for final figures were applied to the entire blot. The last step was to crop the gel image to present only the lanes that were relevant to the experiment in question and that contributed to the quantitative analysis. Hence, the immunoblots in each figure are all from the same experiment and care was taken to ensure that their relative intensities were maintained for display.

The SV pull-down method (SV-PD) has been described in detail (Wong et al., 2013, 2014). Briefly, purified SVs were incubated with immobilized fusion protein or control GST in a detergent free, SV-PD buffer (HB with 5 mM EGTA and (Ca²⁺) clamped to 10 nM by addition of CaCl₂ as calculated with MaxChelator, maxchelator.stanford.edu). Forty microliter of the SV suspension, containing the SV sample used for pull-down assays, was reserved for WBs. SV-PD samples were washed four times with SV-PD buffer and solubilized in 2× Laemmli sample buffer (Bio-Rad) with 5% β-mercaptoethanol. Samples were boiled for 5 min at 100°C, cooled on ice and then run on 8%–12% discontinuous polyacrylamide gels for WBs. Immunoblots were probed for integral SV membrane proteins as markers for vesicle capture. SV-PD was considered positive if the band intensity of at least two SV integral membrane marker proteins, generally SV2, STG or v-ATPase, were more intense than for control samples. Our decision to only use freshly-prepared SVs (to prevent storage artifacts) made these experiments time-demanding and limited us to typically one experiment per week. It should also be noted that detergent-free biochemical methods, using intact lipid-bound structures are prone to higher levels of non-specific SV capture, dictating repeated trials and statistical analysis to form a conclusion. For unknown reasons GST-probed blots were particularly variable with respect to non-specific binding and could not be used for quantification. Criteria for acceptance for further analysis were based on a rejection of blots with a significant level of non-specific binding (“Materials and Methods” section described in Wong et al., 2014; Gardezi et al., 2016).

Immunoblots were analyzed as described in Gardezi et al. (2016). Immunoblots were imaged with the ChemiDoc (Bio-Rad, Hercules, CA, USA) with a broad range of exposure times. For each experiment, protein band intensities were quantified by densitometry from a common blot at a single exposure selected for clear bands without saturation. Background counts were subtracted using an automated rolling disk subtraction. Protein band intensities were normalized to a single control condition for each experiment. For experiments comparing different fusion proteins, intensities were normalized to the C2 fusion protein intensity. For peptide experiments, intensities were normalize to the control peptide condition as described previously (Gardezi et al., 2016).

Fusion protein concentrations were visualized using Coomassie stain (Sigma-Aldrich) and imaged with the ChemiDoc (Bio-Rad) Coomassie stain protein gel function. Concentrations were quantified by densitometry and background was subtracted using an automated rolling disk subtraction. Fusion protein concentration was used as a loading control so that SV2 (I_SV2) and STG (I_STG) protein intensities were normalized to the fusion protein concentration (I_FP) from the same lane, hence %SV-PD was calculated as I_SV2/I_FP or I_STG/I_FP respectively.

Control and putative blocking peptides were synthesized by Biomatik. SV binding-motif blocking and control peptides have been described previously (Gardezi et al., 2016). In the present study, HQARRAPNGA was used as the motif peptide and HQAGGAGGGA was used as a control peptide. Peptides were reconstituted in SV-PD buffer at 10 mM and added to SVs to obtain a final concentration of 0.6 mM. SVs were incubated with the peptides for 2–3 h at 4°C prior to pull-down with the fusion proteins.

To facilitate experimental analysis, we divided the CaV2.2 C-terminal into three segments: C1, C2 and C3 (Figure 1A) and created corresponding normal and mutant fusion proteins (Wong et al., 2013, 2014; Gardezi et al., 2016). One of these, C3, was used for comparison with CaV2.1 fusion proteins in this study (Figure 1B). To simplify discussion, we renamed these fusion proteins according to their C-terminal region (Figure 1B). However, fusion protein laboratory names are also listed for cross-reference.

At the onset of this project only a very short segment of the chick CACNA1A had been sequenced (NCBI Reference Sequence: XM_004950371.1). The library did, however, contain a fairly extensive but incomplete, “predicted” and hence, unconfirmed sequence for a distantly related bird, the ground pecker (Pseudopodoces humilis; NCBI Reference Sequence: XM_005533612.1). As neither library sequence included the C-terminal region, we cloned the full-length channel (Supplementary Figure S2; NCBI Reference Sequence: KY353011).

The chick CaV2.1 has a high homology with crocodile (the closest phylogenic preceding species) and the incomplete predicted ground pecker (Avian) sequence with, as expected, a lower homology to mammals (data not shown). However, we found that the C-terminal was truncated, lacking virtually the entire C3 region (Figure 1B). Attempts to identify a splice variant with a longer C-terminal were unsuccessful and we presume that the predominant chick CaV2.1 terminates at the HCMNRNN sequence (aa 2127). Thus, it appears that the chick CaV2.1 channel lacks our previously reported SV binding site just proximal to the C-terminal tip (see “Discussion” section). To facilitate study of our fusion proteins we raised an antibody, PmidC2, against a peptide (LGTDLSVTTQSGDLPS; aa 2068–2083) within the CaV2.1 C-terminal C2 region. This antibody was characterized by Western blot (WB) and immunocytochemistry (Supplementary Figures S3Bi,ii,iii). It identified our CaV2.1 C2, but not CaV2.1 C1 or CaV2.2 C2 region fusion proteins (Supplementary Figure S3Aii) and a band corresponding approximately to a CaV2 channel weight by WB (note, however, that molecular weights vary markedly between blots for these very large proteins; Supplementary Figure S3Ai). Immunocytochemistry demonstrated staining of the large chick cerebellar Purkinje somata and dendrites, the cells in which the channel was first identified (Llinás et al., 1989; Supplementary Figure S3Aiii).

By alignment with CaV2.2 we divided the CaV2.1 C-terminal into C1 and C2 regions, with the C2 region containing the terminus (Figure 1). During the cloning, we discovered a splice variant that lacked a short segment within the C2 region: aa sequence RRKVRPRGNNLS (aa 2009–2020; Figure 2). This splice variant has been described previously in mammalian CaV2.1 channels and corresponds to omission of exon 44 (Zhuchenko et al., 1997). We refer to fusion proteins that contain these two forms as the “C2” and “C2ΔE44” variants, respectively.

To analyze SV binding to the CaV2.1 C-terminal we generated five fusion proteins, including the full-length terminal (C1+C2), the C1 region (C1), C2 region (C2) and C2 without the splice (C2ΔE44) variants (Figure 1B).

In our earlier reports, we demonstrated and characterized a SV binding site on the C3 segment of CaV2.2 (Wong et al., 2014; Gardezi et al., 2016). Since our cloned CaV2.1 channel has a truncated C-terminal lacking this region, we initially hypothesized that it would fail to also capture SVs. To test this, we carried out SV-PD using a fusion protein that comprised the entire CaV2.1 C-terminal, C1+C2 (Figure 1B). The finding of robust SV capture (N = 4; Figure 3A) contradicted this hypothesis and demonstrated that there had to be a novel SV binding site on the truncated CaV2.1 C-terminal.

To localize the SV binding site, we next created separate fusion proteins of the C1 and C2 regions (Figure 1B) and tested these for SV-PD. Consistently, the C1 region failed to capture SVs compared to strong capture by the C2 region (Figure 3B; C1 expressed as a percentage of C2: SV2 = 21.5 ± 4.5% N = 3, p < 0.01; STG = 41.6 ± 6.7% N = 3, p < 0.05). Thus, the SV capture observed with the C1+C2 fusion protein could be attributed to the C2 region.

As discussed above, we previously used mimetic blocking peptides to identify an SV binding motif, +(+)x RR (where + denotes a positively charged and x an unspecified residue, respectively) in the distal C3 region of the C-terminal (Gardezi et al., 2016). To test if SV capture in the C2 region involved the same motif we carried out SV-PD in the presence of a previously confirmed blocking peptides, HQARRAPNGA, or an inert control, HQAGGAGGGA (see Gardezi et al., 2016). Interestingly, pre-incubation of the SVs with the test peptide markedly reduced C2 fusion protein SV-PD compared to the control (Figure 4; C2+ blocking peptide SV-PD expressed as a percentage of the control peptide: SV2 = 59.8 ± 9.8% N = 14, p < 0.01; STG = 68.0 ± 10.3% N = 14, p = 0.01). Based on this finding we concluded that SV capture involved the same SV binding motif in the C2 region of CaV2.1 channels as previously characterized for the CaV2.2 C3 region.

We next scrutinized the amino acid sequences of the C2 region fusion protein to search for putative SV-binding motifs (Figure 2). As we have no evidence that these motifs are within structurally restricted regions (Gardezi et al., 2016) we included both ortho- (N-to-C) and antidromic (C-to-N) matches and identified one high-homology site: RRKVRPRGNNL (aa 2009–2020) plus one additional site (aa2059–2067) that exhibited motif-like sequences in both orthodromic, RHHPRRGHR, and antidromic, RHHPRRGHR directions. We were fascinated by the realization that the first of these sites, lay within the alternatively spliced E44 region and, hence, is present in the C2 but not the C2ΔE44 fusion proteins. Thus, we could evaluate SV binding to this site by comparing fusion proteins of the two natural C2 splice variants. We observed markedly weaker SV capture using C2ΔE44 compared to the C2 fusion protein (Figure 5A; C2ΔE44 SV-PD as a percentage of C2: SV2 = 65.6 ± 5.2 N = 25, p < 0.0001; STG = 61.7 ± 4.5 N = 23, p < 0.0001). Our data indicates, therefore, that a SV binding site is located within the alternatively spliced region of the CaV2.1 channel C-terminal. However, we also noted that although SV-PD was reduced with C2ΔE44, the blots suggested that it was not eliminated. To explore this possibility, we tested if SV protein capture using C2ΔE44 could be inhibited using the mimetic motif-blocking peptide, as above. This was the case (Figure 5B; C2ΔE44 with blocking peptide SV-PD as a percentage C2ΔE44 with control peptide: SV2 = 56.5 ± 4.6% N = 11, p < 0.0001; STG: = 68.1 ± 7.2% N = 11, p < 0.001), favoring involvement of the second, +(+)x RR, motif binding site within the C2 region.

To test if the other SV binding motif could account for the residual SV-PD we created a mutant fusion protein that also lacked aa 2058–2068, C2ΔE44 Δ58–68 (Figure 1B). The deleted sequence, RHHPRRGHR, contains the antidromic SV-binding motif RRGHR as well as a possible (but perhaps unlikely due to the proline) orthodromic motif of HHPRR. SV protein capture using this fusion protein was reduced even further, as compared to the C2ΔE44 fusion protein. While this finding was only significant for SV2, we could detect a trend towards significant with STG as well (Figure 6A, C2ΔE44Δ58–68 SV-PD as a percentage of C2: SV2 = 28.7 ± 7.0%, N = 10; STG = 38.5 ± 8.5%, N = 11, p < 0.0001, N = 11, C2ΔE44 compared to C2ΔE44Δ58–68 SV2: p < 0.05; STG: p < 0.1). The double mutant effectively eliminated motif-dependent SV capture as the blocking peptide was now without effect (Figure 6B; C2ΔE44Δ58–68 with blocking peptide as a percentage of C2ΔE44Δ58–68 with control peptide: SV2 = 125.3 ± 25.4%, p > 0.1, N = 7; STG = 131.9 ± 39.4%, p > 0.1, N = 7).