Expression and pharmacology of endogenous Cav channels in SH-SY5Y human neuroblastoma cells.

SH-SY5Y human neuroblastoma cells provide a useful in vitro model to study the mechanisms underlying neurotransmission and nociception. These cells are derived from human sympathetic neuronal tissue and thus, express a number of the Cav channel subtypes essential for regulation of important physiological functions, such as heart contraction and nociception, including the clinically validated pain target Cav2.2. We have detected mRNA transcripts for a range of endogenous expressed subtypes Cav1.3, Cav2.2 (including two Cav1.3, and three Cav2.2 splice variant isoforms) and Cav3.1 in SH-SY5Y cells; as well as Cav auxiliary subunits α2δ1-3, β1, β3, β4, γ1, γ4-5, and γ7. Both high- and low-voltage activated Cav channels generated calcium signals in SH-SY5Y cells. Pharmacological characterisation using ω-conotoxins CVID and MVIIA revealed significantly (∼ 10-fold) higher affinity at human versus rat Cav2.2, while GVIA, which interacts with Cav2.2 through a distinct pharmacophore had similar affinity for both species. CVID, GVIA and MVIIA affinity was higher for SH-SY5Y membranes vs whole cells in the binding assays and functional assays, suggesting auxiliary subunits expressed endogenously in native systems can strongly influence Cav2.2 channels pharmacology. These results may have implications for strategies used to identify therapeutic leads at Cav2.2 channels.

Introduction

Voltage-gated Ca2+ channels (Cav) are membrane proteins essential for the control of calcium signaling events, such as muscle contraction, gene expression, and neurotransmitter and hormone release. Dysfunction of Cav channels is related to a variety of heart, circulatory and neurological diseases; including arrhythmias, hypertension, some forms of epilepsy, migraine and other chronic diseases such as cancer, diabetes, ischemic brain injury and neuropathic pain [1], [2]. The Cav α subunit contains the voltage sensor and gating machinery and is the binding site for most inhibitors. This subunit comprises 4 domains each with six transmembrane segments. The pore is formed by the S5/S6 segments and the connecting pore loop, with channel opening gated by bending of the S6 segments at a hinge glycine or proline residue [3], [4]. The voltage sensor domain consists of the S1–S4 segments, with positively charged residues in S4 serving as gating charges [5] (for review see: [3], [6], [7]).

Based on the distinct pharmacological and electrophysiological properties of Cav channels, ten different gene subfamilies have been identified in vertebrates and classified as high voltage activated (HVA) Cav1.1–4 (L-type), Cav2.1 (P/Q-type), Cav2.2 (N-type), Cav2.3 (R-type); and low voltage activated (LVA) Cav3.1–3 (T-type). The α subunit includes channels containing α1S, α1C, α1D, and α1F, which mediate L-type Ca2+ currents. The Cav2 subfamily (Cav2.1 to Cav2.3) includes α subunits α1A, α1B, and α1E, which mediate P/Q-type, N-type, and R-type Ca2+ currents, respectively. The Cav3 subfamily (Cav3.1 to Cav3.3) includes α subunits α1G, α1H, and α1L, which mediate T-type Ca2+ currents ( ) (for reviews see: [3], [8], [9]). Of these, Cav2.2 has been of particular interest as a therapeutic target given the central role it plays mediating neurotransmitter release in nociceptive pathways such as presynaptic nerve terminals and dendrites [10]. Cav α subunits are co-expressed in native systems together with two or three auxiliary subunits (β, α2δ and γ), which undergo alternative splicing (for review see: [9]) and dramatically influence Cav channel function, intracellular trafficking and posttranslational modifications [11]. Indeed, when expressed alone in recombinant system, the α1B subunit, for example, encodes a voltage-dependent calcium channel with kinetic properties different from those of native Cav2.2 channels [9], [12]. In contrast, when co-expressed with auxiliary β and α2δ, increased current amplitudes are observed and the kinetics of activation and inactivation are closer to those of native channels [12].

Table 1

Primers used to identify Cav channels α subunits in SH-SY5Y cells.

Subtype	Accession Number	Primer Forward/Reverse	Size (bp)	Annealing T (°C)
Cav1.1	NM_000069.2	CGCATCGTCAATGCCACCTGGTTTA/AGCACATTGTCGAAGTGGAAGTCGC	623	(?) ND
Cav1.2	[19]	CTGCAGGTGATGATGAGGTC/GCGGTGTTGTTGGCGTTGTT	502	58 [19]
Cav1.3	EU 363339.1	ACCCCCACCTGTAGGATCTCTCTCC/TCCTGACACTAGTCGAAGTGGTCGC	541	68
Cav1.3	NM_001128840.1	GCTGCTGTGGAAGTCTCTGTCAAGC/TCAGTGATTCCACCACACACCACGA	343	68
Cav1.4	NM_005183.2	AGGGACCCCTAAGCGAAGAAACCAG/ACCCCATGGCATCTTGCATCCAGTA	899	(?) ND
Cav2.1	FJ040507.1	AGGACGAGGACAGTGATGAA/GCAGAGGAAGATGAAGGA AA	365	(?) ND
Cav2.2	NM_000718.2	GGAACTGACTTCGACCTGCGAACAC/CCTCCTCTGCGTGGATCAGGTCATT	754	60
α1BΔ1	Bp 22+34	AGGAGATGGAAGAAGCAGCCAATCA/CCTTTCTGGTGTTTCATCTGGTGCA	900	58
α1BΔ1	Bp 23+33 [16]	CCAGAGGATGCAGACAATCAGCGGA/GCATCTTCTACCTGTCGAGGTACGC	900	60
α1BΔ2	Bp 21+31	CAGCCAATCAGAAGCTTGCTCTGCAAAAG/CTTTCGTTTGCGGTGGTCCCGCGGT	700	(65)
α1BΔ2	Bp 24+33 [16]	CAAGGATGAAGAGGAGATGGAAGAA/GCGTACCTCGACAGGTAGAAGATGC	1300	(?) ND
Cav3.1	BC110995.1	GCTGCTGGAGACACAGAGTACAGGT/CTCGTGGTATTCGATGCCCATGCTG	397	60
Cav3.2	NM_021098.2	CCTGATCCCTACGAGAAGATCCCGC/CACGGCTGAAGTACTTGCTGTCCAC	433	60
Cav3.3	AF393329.1	AGATGCCCTTCATCTGCTCCCTGTC/AAGATCTCCTCGTAGCAGTCGCCAG	526	60

(?) ND: isoform not detected, unknown annealing temperature.

Cell-based systems are desirable in the field of high-throughput screening assays due to their similarity to in vivo environment. SH-SY5Y human neuroblastoma cells are derived from human sympathetic neuronal tissue. This cell line maintains in culture many of the properties of nerve cells, providing a useful model for the characterisation of molecules affecting human neuronal function, including endogenously expressed Cav channels [13]–[15]. In particular, SH-SY5Y cells have been an attractive model system for the study of Cav2.2 function [13]. Although heterologous expression models provide control of subunit expression, native systems provide potentially more complex models which, when characterized, can help to determine the pharmacology of drugs in a native context and the physiology and pathophysiology of endogenously expressed receptors and channels. However, little is known about the Cav subtypes and auxiliary subunits endogenously expressed in SH-SY5Y cells, limiting the interpretation of pharmacological data. Here we report a detailed characterisation of endogenously expressed Cav channels expressed in SH-SY5Y cells using PCR and pharmacological approaches, with particular emphasis on the nociceptive target Cav2.2.

Results

SH-SY5Y Cells Endogenously Express Multiple Cav Subtypes, Cav2.2 Isoforms and Auxiliary Subunits

We assessed expression of mRNA transcripts for Cav subtypes and auxiliary α2δ, β and γ subunits isoforms in SH-SY5Y cells by performing RT-PCR using specific primers ( ). Bands with the predicted sizes were detected for Cav1.3, Cav2.2, and Cav3.1, while Cav1.1, 1.2, 1.4, 2.1, 2.3, 3.2 and 3.3 were not detected ( ). In addition, bands of expected sizes for ( ) β1, β3, β4, α2δ1–3, γ1, γ4, γ5 and γ7 auxiliary subunits ( ) were also identified. Since splice variants can be generated by alternative RNA processing, which can influence function and pharmacology [16], we also investigated the expression of some human splice variants [16], [17]. PCR bands with the predicted sizes for Cav1.3 isoforms 1 and 2; full length Cav2.2, α1B1 (Gene bank accession number M94172.1), shorter α1B variant, α1B2 (Gene bank accession number M94173.1) [17]; and Δ1 (but not Δ2) [16], [18] were detected for the first time in the SH-SY5Y cells ( , ).

Figure 1

RT-PCR to identify the Cavα and auxiliary subunit isoforms expressed in SH-SY5Y cells.

Expression of Cavα subtypes, auxiliary β, α2δ and γ subunits, as well as Cav2.2 splice variant isoforms were determined in SH-SY5Y cells using standard RT-PCR and specific primers for each isoform. (A) SH-SY5Y cells endogenously express Cav1.3 isoform 1, Cav1.3 isoform 2, Cav2.2 and Cav3.1, but not Cav1.1 and Cav1.2, Cav1.4, Cav2.3, Cav3.2 and Cav3.3. Expected band sizes were (bp): Cav1.3 isoform 1, 541; Cav1.3 isoform 2, 343; Cav2.2, 754; and Cav3.1, 397, as indicated with arrows (B) SH-SY5Y cells endogenously express different Cav2.2, α1B splice variant isoforms. Bands with predicted sizes were (bp): α1B1, 728; α1B2, 854; Δ1, 900 bp. No band was detected for splice Δ2. (C–D) SH-SY5Y cells express the auxiliary β1, β3, and β4 but not β2; in addition to α2δ1–3, but not α2δ4; and γ1, γ4–5 and γ7 but not γ2–3 and γ8 subunits. Expected band sizes were (bp, base pairs): β1, 331; β3, 594; β4, 731; α2δ1, 252; α2δ2, 878; α2δ3, 132 and γ1, 367; γ4, 909; γ5, 257; and γ7, 910.

Table 2

Primers used to identify Cav channel auxiliary subunits in SH-SY5Y cells.

Subunit	Accession Number	Primer Forward/Reverse	Size (bp)	Annealing T (°C)
β1	NM_000723.3	ATGCACGAGTACCCAGGGGAG/CAGCGCAGTAGCGGGCCTTATT	331	60
β2	NM_000724.3	TCGCTTGCCAAACGCTCGGT/ATGACGGCTGCGCTGCTTGT	909	(?) ND
β3	NM_000725.2	GCAGCAGCTCGAAAGGGCCA/ATGCTGGAGCGGGCAGAGGA	594	65
β4	NM_001005747.2	TGAAGACTCGGAGGCTGGTTCAGC/TGGACCGGGTGTTCGAACGT	731	(?) ND
α2δ1	NM_000722.2	TGCTCATCGGCCCCTCGTCG/CCAGGCGCACCAGGGCTTTAG	252	60
α2δ2	NM_001174051.1	AGCCTAGGCAGGCGCACACT/TCTGCACTAGCTCACACTGCTCCGG	878	60
α2δ3	NM_018398.2	GGACGAGAGGCTGCGTTTGCA/GGGCCGGCTAAGCACGTGAA	132	65
α2δ4	NM_172364.4	TGGCCTGGGCCTTTGTGCAG/GCCTCCTCGGCAGCTTCCAC	328	(?) ND
γ1	NM_000727.3	TGCTGGCCATGACAGCCGTG/AACATGGACGCGGGTCGCAG	367	60
γ2	NM_006078.3	TCTCTGGGCCTTAATTTTCCCC/TTTTCACAGACCCCCAAAGACA	439	(?) ND
γ3	NM_006539.3	CTCCCCTTCCCCTTTCCTTAAC/AGCTGGGATTTCCTTTCTGGAG	840	(?) ND
γ4	NM_014405.3	TTTGCACGAAGGTTGTGCTG/TTGCTCTCCTGGCGTTGATT	909	62–64
γ5	NM_145811.2	GATCAAGATGTCCCTGCACTCA/CAGAGACAAAGGCCAGTATCGT	257	64
γ6	NM_145814.1	TGCTCAGTAAAGGTGCAGAGTT/CTCGGTGGTTGCTTAGAGAAGT	334	(?) ND
γ7	NM_031896.4	ACTGGCTGTACATGGAAGAAGG/TGAAATAAGGGAGTCTGTGGGC	910	65
γ8	NM_031895.5	TGCTGAAGCATAGTCATGGTGT/CCTCTGCCTTCTCAGTGAACTT	987	(?) ND

(?) ND: isoform not detected, unknown annealing temperature.

The best annealing temperature for each gene analysed (see – ) was determined using a gradient PCR protocol in rounds of control experiments prior to testing each Cav gene-specific primer. Target-specific primers for the housekeeping gene GAPDH were designed as previously described [19] and GAPDH was detected in all PCRs, indicating amplifications were cDNA specific. PCR master mix using random primers without cDNA was used as negative gDNA control in all PCRs. Specificity of primers was demonstrated in a range of control experiments (data not shown), including detection of Cav2.2 plasmid but not other Cav subtypes by Cav2.2 primers; and absence of Cav2.2 in HEK cells. β1 and α2δ1 primers were positive for β1 and α2δ1 plasmids, while the same primers were negative for β2–4 and α2δ2–4 (data not shown), indicating primers were selective for β1 and α2δ1 auxiliary subunits. The identity of each of these PCR products, including γ1, γ4, γ5 and γ7, was confirmed by sequencing analysis (data not shown).

Displacement of 125I-GVIA Binding from SH-SY5Y Cell Membranes

GVIA is a highly selective Cav2.2 blocker [20] and 125I-GVIA binding assays have been well established using rat brain membranes [21]–[23]. We performed binding assays and confirmed SH-SY5Y cells contain 125I-GVIA binding sites which can be fully displaced by Cav2.2 selective inhibitors ω-conotoxins CVID, GVIA and MVIIA. Affinities of ω-conotoxins for human and rat Cav2.2 channels were next compared using these assays. CVID, GVIA and MVIIA each fully displaced 125I-GVIA binding to crude rat brain membranes with similar affinities (pIC50± SEM values; CVID 10.53±0.15, GVIA 10.43±0.16, and MVIIA 10.19±0.04) ( , ), consistent with earlier studies [21]. Intriguingly, the affinity of GVIA (pIC50 10.55±0.15) to displace 125I-GVIA binding to SH-SY5Y membranes was similar to that shown in rat brain, while both CVID and MVIIA had significant higher affinity for the human cell membranes (pIC50s of 11.51±0.12 and 11.29±0.23, respectively) than for rat brain membranes ( , ). In addition, the affinities of these ω-conotoxins dramatically decreased when determined on the intact SH-SY5Y cells instead of membranes, with GVIA affinity shifted ∼10-fold, and CVID and MVIIA affinity shifted ∼100-fold (see , ).

Figure 2

Displacement of 125I-GVIA from SH-SY5Y whole cell and membranes byω-conotoxins.

Displacement of 125I-GVIA binding to Cav2.2 expressed in rat brain and SH-SY5Y intact/whole cell and membranes. (A) Displacement of 125I-GVIA from rat brain membranes. (B) Displacement of 125I-GVIA from human SH-SY5Y cell membranes. (C) Displacement of 125I-GVIA from human SH-SY5Y whole cell. (D) ω-Conotoxins affinity (Kd ± SEM) to displace 125I-GVIA from rat brain membranes and human SH-SY5Y cell membranes. Data are mean ± SEM of triplicate data from a representative experiment best fitted to a single-site competition model using GraphPad Prism.

Table 3

ω-Conotoxin affinities (IC50± SEM) to displace 125I-GVIA binding.

ω-Conotoxin	Rat membrane Kd (nM)	SH-SY5Y membrane Kd (nM)	Whole SH-SY5Y Kd (nM)
CVID	0.034±0.013	0.0034±0.009	3.2±0.2
GVIA	0.043±0.013	0.033±0.012	0.27±0.084
MVIIA	0.064±0.007	0.0065±0.0019	10±0.085

Pharmacology of the Endogenously Expressed Cav Channels

To investigate if the Cav channels endogenously expressed in SH-SY5Y cells were functional, and to further study the pharmacology of these channels, we assessed KCl-evoked Cav responses using a fluorescent high-throughput Ca2+ imaging assay on the FLIPRTetra (Fluorescent plate image reader, Molecular Devices, Sunnyvale, CA) ( ). CaCl2 (5 mM) was added to the KCl stimulation solution in all experiments to maximize the Ca2+ influx signal. Co-addition of 90 mM KCl and 5 mM CaCl2 evoked a large transient response indicating increase in intracellular Ca2+ ( ). Concentration-response curves for KCl-mediated stimulation showed activation of Cav responses with an EC50 of 17.3 mM (pEC50 1.88±0.06, Hill slope of 2.5) ( , ), similar to previously described values [24]–[26]. To assess the contribution of each Cav channel expressed in SH-SY5Y cells to the KCl-evoked Ca2+ responses, we determined concentration-response curves for KCl/Ca2+ stimulation in the presence of subtype-specific inhibitors. The Cav1 (L-type) inhibitor nifedipine was used at a concentration (10 µM) that does not affect responses of other Cav subtypes (N, R, P/Q or T-type) [27], to isolate non-L-type responses. The KCl concentration-response curve was shifted to the right in the presence of nifedipine (EC50 of 20.4 mM, pEC50 1.69±0.12, Hill slope of 2.9) ( ). Conversely, the Cav2.2 (N-Type) inhibitor ω-conotoxin CVID was used at a concentration that does not affect responses of other Cavs (up to 3 µM) [21], [28], to isolate non-N-type responses. Compared to responses in the presence of nifedipine, the KCl concentration-response curve was shifted to the left in the presence of CVID (EC50 of 18.6 mM, pEC50 1.86±0.10, Hill slope of 3.5) ( , ). These differences can be accounted for by the electrophysiological properties of each Cav channel subtype identified, since L-type requires a larger depolarization than N-type to be activated [28], and the control KCl responses is a result of activation of both channel types. These results confirm that SH-SY5Y cells express functional Cav subtypes, including Cav1 and Cav2.2, which can be pharmacologically isolated using selective inhibitors. The observed pharmacology is consistent with the subtypes identified in our PCR experiments and with previous reported electrophysiological data [14], [15].

Figure 3

Cav2.2 and Cav1 channels endogenously expressed in SH-SY5Y cells are functional.

Data obtained from fluorescent Ca2+ imaging assays of KCl-evoked Ca2+ responses in SH-SY5Y cells. (A) Cav1 and Cav2.2 activation in the presence of CVID (open ball) and nifedipine (filled ball), respectively, shifted control KCl-evoked Ca2+ responses (quadrilateral) significantly in SH-SY5Y cells (p>0.05). (B) Time course of Ca2+ responses is shown for control KCl 90 mM (black), KCl in the presence of nifedipine (blue) and KCl in the presence of CVID (green). (C) Concentration-response curve for nifedipine inhibition of Cav1 responses (D) Concentration-response curves for CVID, GVIA and MVIIA inhibition of Cav2.2 responses. The responses were normalized using controls: positive KCl and negative PSS buffer; and plotted across increasing concentrations of antagonists (E) Comparison of ω-conotoxins CVID, GVIA and MVIIA potencies (IC50/Kd ± SEM of n = 3–4 replicates for each experiment, n = 3 experiments) in displacing 125I-GVIA from SH-SY5Y whole cell and SH-SY5Y cell membranes with the functional assays data.

Table 4

Potency (IC50± SEM) of Cav channel modulators on functional assays.

Cav Activator/Inhibitor	Ca2+ Stimulation EC50 (mM)	Ca2+ Inhibition IC50 (µM)
KCl	17.28±3.41	−
KCl+CVID	18.61±3.22	−
KCl+NIFEDIPINE	20.35±3.17	−
CVID	−	0.16±0.025
GVIA	−	0.15±0.09
MVIIA	−	0.024±0.005
NIFEDIPINE	−	0.23±0.046
MIBEFRADIL	−	3.0±0.031
PIMOZIDE	−	1.3±0.097
ω-AGATOXIN TK	−	NDR
SNX 482	−	NDR

NDR: Non-detectable response.

Since 90 mM KCl/5 mM CaCl2 elicits maximal Cav1 and Cav2.2 responses ( ), we used this combination to further characterize the Cav channel subtypes expressed in SH-SY5Y cells. Concentration-response curves for nifedipine at Cav1 channels were generated in the presence of saturating concentration of CVID (3 µM). Under these conditions, nifedipine inhibited KCl evoked Ca2+ responses with an IC50 of 0.28 µM (pIC50 6.5±0.052) ( , ), consistent with reports for nifedipine block of L-type responses in neuronal cells [29]. To characterize Cav2.2 pharmacology, inhibition by ω-conotoxins was determined in the presence of a near saturating concentration of nifedipine (10 µM). Under these conditions, the potency of CVID was IC50 0.16 µM (pIC50 6.87±0.078), GVIA 0.15 µM (pIC50 6.84±0.06) and MVIIA 0.024 µM (pIC50 7.7±0.13) ( , ). These results are consistent with previous studies on MVIIA [22]–[24], [26] and CVID [22]–[24] inhibition of N-type responses in native and recombinant systems, when Cav2.2 was co-expressed with β and α2δ subunits [22]–[24]. In contrast, GVIA potency at Cav2.2 expressed in SH-SY5Y cells (IC50 of 0.15 µM; pIC50 6.8±0.072) was consistently lower than previously described for heterologous expressed rat [24], [26] and human [25] α1B co-expressed with α2δ1 and β3, but similar to data obtained using native expression systems such as dissociated rat DRG cells [30] and chicken synaptosomes [31].

A small portion (5–15%) of the KCl-evoked responses was insensitive to block by co-application of 10 µM nifedipine and 3 µM CVID ( ). To pharmacologically characterize these remaining responses, we assessed the effects of Cav2.1 and Cav2.3 subtype-specific inhibitors, as well as of compounds with activity at Cav3, on the Ca2+ responses evoked by 90 mM KCl/5 mM CaCl2, in the presence of both CVID and nifedipine. The Cav2.1 blockers ω-agatoxin IVA (data not shown) and ω-agatoxin TK did not significantly affect KCl-evoked Ca2+ responses at concentrations up to 10µM ( , ). The Cav2.3 antagonist SNX 482 also had no significant inhibitory effect at concentrations up to 10µM ( , ). On the other hand, mibefradil (30 µM), a benzimidazolyl-substituted tetraline reported to inhibit Cav3 responses in different systems with weak affinity [32], [33] fully inhibited these remaining responses with an IC50 of 3 µM (pIC50 5.3±0.035) ( , ). Similar IC50 values for mibefradil block of T-type responses in native systems have been previously reported (see [32]–[35]). In addition, another Cav3 inhibitor, the antipsychotic pimozide, also fully inhibited the remaining responses with an IC50 of 1.3 µM (pIC50 5.2±0.097) ( , ), similar to previously reported literature values [34]. These findings are in agreement with our PCR ( ) which detected mRNA transcripts for Cav3.1, but neither Cav2.1 nor Cav2.3 was identified.

Figure 4

Characterization of resistant Ca2+ responses in SH-SY5Y cells.

Data obtained from fluorescent Ca2+ imaging of KCl-evoked Ca2+ responses in SH-SY5Y cells. (A) Concentration-response curves for mibefradil, pimozide, ω-agatoxin TK and SNX 482 in inhibiting resistant KCl-evoked Ca2+ responses in SH-SY5Y cells, pretreated with CVID (3 µM) plus nifedipine (10 µM) (B–D) Time course of transient Ca2+ responses activated by 90 mM KCl/5 mM CaCl2, in the presence of CVID (3 µM) and nifedipine (10 µM) and following the addition of agatoxin TK, SNX-482 and mibefradil.

Discussion

Cav2.2 channels play a key role in regulating nociception. Inhibition of Cav2.2 at the spinal cord produces analgesia in animal models of pain [23], [36] and in humans [37], with direct (eg. Prialt) and indirect (eg. gabapentin) inhibitors among some of the most recently developed analgesics [18]. Neuroblastoma cells, including the sympathetically derived human neuroblastoma cell line SH-SY5Y, provide excellent model systems to study Cav2.2 channels in a native context [14], [15]. However little is known about the Cavα and auxiliary subunits expressed, limiting interpretation of pharmacological data from these cells. To address this limitation, we have characterized the expression and pharmacology of Cav channels in SH-SY5Y cells and investigated mechanisms likely to influence the pharmacology of ω-conotoxins at Cav2.2 channels.

Previous electrophysiological studies have identified L- and N- currents from high voltage activated channels Cav1 and Cav2.2 in SH-SY5Y cells, but not low voltage activated T- type currents from Cav3 channels [13]–[15], [38]. In contrast, we detected mRNA transcripts for the N-type (Cav2.2), two L-type (Cav1.3 isoform 1 and 2) and one T-type isoform (Cav3.1). In addition, we also detected mRNA transcripts for Cav2.2 splice variants, including α1B2 (74 amino acid shorter) [16], [17], [39] and the splice α1BΔ1 (382 amino acid shorter) [16].

Functional Cav responses elicited by addition of KCl/CaCl2 were assessed using a fluorescent high-throughput Ca2+ imaging assay on the FLIPRTetra. KCl has been used extensively to activate Cav responses in a diversity of functional assays ([24], [26], [40]). Addition of high concentrations of KCl causes a change in membrane potential, which in turn leads to opening of Cav channels, influx of Ca2+ and a resultant increase in intracellular fluorescence. While the change in membrane potential elicited by addition of KCl at the concentrations used here is approximately linear, accumulation of intracellular Ca2+ is saturable and fits a sigmoidal concentration-response curve because a change in membrane potential leads to a finite change in channel open probability and thus Ca2+ influx.

Cav2.2 channels expressed in SH-SY5Y cells were functional and generated KCl activated responses that were inhibited by ω-conotoxins CVID, GVIA and MVIIA in the presence of saturating concentrations of nifedipine. As expected, when L-type responses were isolated by addition of saturating concentrations of CVID, nifedipine concentration-dependently blocked KCl responses. However, a small response remained (5–15%) in the presence of a combination of Cav2.2 and Cav1 inhibitors. This resistant response was completely abolished by the Cav3 inhibitors mibefradil and pimozide. While mibefradil and pimozide are not specific inhibitors of T-type currents and also inhibit L-type channels [32], [41], inhibition of the residual Ca2+ response was also observed in the presence of saturating concentrations of nifedipine, suggesting that activity of these compounds at L-type channels did not contribute to inhibition of residual response. Based on our observations that this resistant response was not blocked by inhibitors of L-type (nifedipine), N-type (CVID), R-type (SNX 482) or P/Q-type channels (ω-agatoxin), but was completely abolished by compounds with known activity at T-type channels, it seems plausible that this response may be mediated by Cav3.1, which mRNA expression was detected in SH-SY5Y. Alternatively, it is known that the Δ1 splice variant, which mRNA expression was detected in SH-SY5Y cells, is significantly more resistant to the blockade by MVIIA and GVIA [42]. While inhibition by CVID of the Cav2.2 splice variants detected in SH-SY5Y cells has not been characterised, it is possible that, akin to inhibition of Nav channels by the µ-conotoxin GIIIA, complete current inhibition by CVID cannot be achieved for these splice variants. Alternatively, the response remaining in the presence of nifedipine and CVID could represent another undefined resistant current, or an artifact of the KCl/Ca2+ activation buffer used in this study.

Development of non-electrophysiological HTS Cav3 channel assays has been hampered by some of the properties of this channel, including their low voltage threshold for activation and inactivation and rapid inactivation kinetics. However, although T-type currents inactive rapidly, fluorescence Ca2+ assays detect accumulation of intracellular Ca2+ rather than currents, and are thus not subject to the same temporal resolution constraints. In addition, compared to heterologous systems, SH-SY5Y cells have a relatively hyperpolarised resting membrane potential [43], which would be conducive to channels being present in the resting state. Accordingly, Ca2+ assays at Cav3 channels using the FLIPR have been successfully developed [40] and it is clearly conceivable that functional responses of Cav3.1 expressed in SH-SY5Y cells could be elicited using KCl/Ca2+ stimulation.

In addition to functional characterization, we also confirmed Cav2.2 expression at the protein level using 125I-GVIA binding assays. The ω-conotoxins CVID, GVIA and MVIIA each fully displaced 125I-GVIA binding to SH-SY5Y cell membranes with high affinity. Interestingly, while the affinity of GVIA was not significantly different between species, CVID and MVIIA affinities were ∼10-fold higher in human SH-SY5Y membranes compared to rat brain membranes. These results support the findings that MVIIA and CVID interacts with Cav2.2 human channels through a different pharmacophore, as compared with GVIA [44].

Variation in the affinity of ω-conotoxins between species is likely influenced by Cavα splice variants, with differences in toxin sensitivity, time course and voltage-dependence of inactivation, single channels conductance, gating behavior and sensitivity to G-protein-mediated modulation reported for splice isoforms endogenously expressed in neuronal cells of rat, mouse, rabbit and humans 16,17,39,42,45–48 (for review see: [46]). In pain, the Cav2.2 splice variant 37a replaces the usual variant 37b in a specific subset of nociceptive neurons, and thus may represent a potential therapeutic target [42], [46], [49]. However, this variant has to date only been described in rat dorsal root ganglion neurons, and is not known to be present in human tissue.

Additional human splice variants include two α1B isoforms that have long or short C-termini [17], and two human forms that lack large parts of the domain II-III linker region, including the synaptic protein interaction site. These splice variants, termed Δ1 and Δ2, have been previously isolated from IMR32 human neuroblastoma cell line and human brain cDNA libraries [16]. We have identified mRNA transcripts for the full length α1B1, α1B2 (74 amino acid shorter) [16], [17], [39] and the splice variant Δ1 (382 amino acid shorter) [16] in SH-SY5Y cells. The α1B1 is an axonal/synaptic isoform, while α1B2 is restricted to neuronal soma and dendrites [39], [50], however, apart from differential susceptibility to Gαi/Gαo-versus Gαq-mediated inhibition, little is known regarding its biophysical and pharmacological properties. On the other hand, the Δ1 splice variant has lost part of the synaptic protein interaction (synprint) site and is thus unlikely to play a role in fast synaptic transmission, with shifts in the voltage dependence of steady-state inactivation and a more rapid recovery from inactivation compared to full length α1B1 [16]. Importantly and clinically relevant, Δ1 variant was significantly more resistant to the blockade by MVIIA and GVIA; however the degree of effect varied for each toxin [16]. Thus, expression of the Δ1 variant in SH-SY5Y cells may contribute to the reduced ω-conotoxin affinity observed. While expression of these splice variants in SH-SY5Y cells was detected using gene specific primers, which have been extensively validated in the literature [17], further confirmation of expression at the protein level is warranted.

Cav channel auxiliary subunits can also influence the pharmacology of Cav inhibitors, with ω-conotoxins displaying reduced affinity in the presence of the α2δ subunit [11], [22]–[24], [27], [51]. Specifically ω-conotoxins GVIA, MVIIA and CVID had reduced affinity when α2δ1 subunit was co-expressed with the Cav α1B [23]. α2δ up-regulation has been associated with chronic pain and epilepsy, with gabapentin and pregalin binding to α2δ reducing Cav2.2 trafficking and the symptoms of pain [11]. The α2δ1–3, β1, β3 and β4, γ1, γ4–5 and γ7 subunits were detected in SH-SY5Y cells and potentially contribute to the differences in ω-conotoxins potency in whole cell vs. membrane assays.

The γ1 subunit was originally identified in skeletal muscle in complex with Cav1 channels [52], but effects of this subunit on the ω-conotoxins affinity at Cav2.2 have not been determined. In contrast, co-expression of the γ7 subunit almost abolished the functional expression of CaV2.2 in either Xenopus oocytes or COS-7 cells [53], [54]. The neuronal γ2 is associated with epileptic and ataxic phenotypes of stargazer mouse [55], but was not detected in SH-SY5Y cells. The γ5 and γ7 subunits represent a distinct subdivision of the γ subunit family of proteins identified by structural and sequence homology to stargazing. The γ4 subunit affected only the Cav2.1 channel [55], [56]. The γ5 subunit may be a regulatory subunit of Cav3.1 channels (for review see: [57]). These subunits may also potentially contribute to differences in ω-conotoxins binding affinities observed in whole cell vs. membrane assays.

While auxiliary subunits affect ω-conotoxin affinity in functional studies, this quaternary complex is likely to be disrupted upon preparation of homogenized membranes for the binding assays [58]. To examine this possibility, we studied the ability of GVIA to displace 125I-GVIA from whole SH-SY5Y cells compared to homogenized membranes. Interestingly, ω-conotoxins CVID, MVIIA and GVIA had higher affinity to displace 125I-GVIA from the homogenized membranes compared to the whole cells, an effect that was most pronounced for CVID and MVIIA (∼100-fold) compared to GVIA (∼10-fold). We have previously reported a similar trend for both CVID and MVIIA in heterologous expression system with and without the α2δ subunit [22]. Potency estimates obtained with the functional assays were significantly lower than estimates obtained in whole cell radioligand binding assays. The relatively high level of Ca2+ in the physiological saline used traditionally for functional assays compared to binding assays could contribute to these differences, since Ca2+ non-competitively inhibits ω-conotoxin binding [21]. However, our whole cell data was also obtained by incubating ω-conotoxins in a Ca2+-free physiological saline solution and the origin of these differences is unclear. Interestingly, this effect was most marked for GVIA, intermediate for CVID and insignificant for MVIIA.

In summary, we have characterized functional Cav channels expressed in SH-SY5Y human neuroblastoma cell line. Our studies have shown expression of different Cavα splice variants, in conjunction with auxiliary subunits in a native context, can modulate the pharmacology of Cav2.2 channel inhibitors. SH-SY5Y cell line provides a useful model for the investigation of novel human Cav2.2 inhibitors and is amenable to the establishment of high-throughput assays [59], which can be adapted to detect endogenously expressed human Cav1.3, Cav2.2 and possibly Cav3.1, in the presence of appropriate inhibitors. These assays are expected to prove useful for the discovery and pharmacological characterization of novel Cav channel modulators targeting human Cav related diseases.

Materials and Methods

Reverse Transcriptase Polymerase Chain Reaction (RT-PCR)

Cav channel subtype and auxiliary subunits mRNA expression profiles were investigated in SH-SY5Y cells using standard RT-PCR and specific primers. The primers were designed using The Basic Local Alignment Search Tool (BLAST) [60], [61], or otherwise specified as, previously described in the literature. Primer sequences, Gene Bank reference numbers, predicted PCR product sizes, and optimum annealing temperatures are shown in . The primers used to identify Cav subtypes and auxiliary subunits were designed so that all splice variants of specific isoforms would be amplified. On the other hand, primers to amplify Cav2.2 splice variants isoform were designed to be specific to each isoform. PCR conditions to detect splice variants were set as previously described [16], [17], with gradient PCR performed for all sets of primers, allowing the identification of optimal annealing temperatures. Different sets of primers were used to identify the full length and isoforms Δ1 and Δ2 (see table 1) [16]. These primers were designed based on the region of the domain II-III linker of Cav2.2 channels, as previously described [16]. Primers used to identify the full length α1B1 and short α1B2 isoforms were designed based on the C-terminus region [17]. Data is representative of at least three independent experiments.

SH-SY5Y cells (1×106) were harvested and total RNA isolated using Trizol® Reagent (Invitrogen, Carlsbad, CA). The isolated RNA was subsequently treated with RNase-free DNase to remove any genomic DNA contamination. RNA concentration was determined by absorbance measurements at 260 nm and its purity/integrity was accessed by analyzing the ratio 260/280 nm with a Nanodrop® (Thermo Scientific). Synthesis of first strand cDNA was performed using 1 µg of the extracted RNA and the Omniscript Reverse Transcription Kit (Qiagen), according to the manufacturer’s instructions. cDNA amplifications were performed using Taq Polymerase (New England Biolabs, US). The reaction mix (total 25 µL) included (µL): 1 cDNA (100 ng), 0.125 of the enzyme, 0.5 reverse and 0.5 forward primers (10 µM), 0.5 dNTPs (10 mM), 2.5 Thermopol reaction buffer (10×) and nuclease free water. RT-PCR was carried through as an initial denaturation step at 95°C for 3 min followed by 35 cycles of the steps: 95°C for 30 s, optimal annealing temperature as previously determined ( ) for 60 s, 68°C extension for 60 s, plus an extra 5 min elongation step at 68°C. PCR products were analyzed by 1% agarose gel and predicted sizes estimated by comparison with DNA molecular weight makers (50 and 100 bp ladder, New England Biolabs). Target-specific primers for the housekeeping gene GAPDH were designed as previously described [19]. PCR master mix using random primers without cDNA was used as negative gDNA control in all PCRs. Specificity of primers was demonstrated in a range of control experiments (data not shown), including detection of Cav2.2 plasmid but no other Cav subtypes by Cav2.2 primers; and absence of detectable levels of Cav2.2 in HEK cells. β1 and α2δ1 primers were positive for β1 and α2δ1 plasmids, while the same primers were negative for β2–4 and α2δ2–3 (data not shown), indicating primers were selective for β1 and α2δ1 auxiliary subunits. In addition, identity of PCR products was further confirmed by sequencing analysis (data not shown). Figures 1A–D is representative of the average of 3–10 individual experiments.

Sequencing

PCR amplicons were first separated on agarose gels and bands of expected sizes identified. PCR products were purified using the Wizard SV Gel and PCR clean-up system (Promega), and a sample of each purified PCR product was sent for sequencing at the Australian Genome Research Facility. cDNA sequences of human Cav subtypes and auxiliary subunits were retrieved from GenBank (http://www.ncbi.nlm.nih.gov/Entrez/) and BLASTn [62] was used for confirmation of the identity of human Cav subtypes and auxiliary subunits.

Cell Culture

The human neuroblastoma SH-SY5Y cells (Victor Diaz, Goettingen, Germany) were cultured and routinely maintained at 37°C and 5% CO2 in RPMI 1640 antibiotic-free medium (Invitrogen) supplemented with 10% heat-inactivated FBS and 2 mM GlutaMAX™ (Invitrogen). Trypsin/EDTA was used to detach the cells from the T-75 or T175 flasks and cells were split in a ratio of 1∶5–1∶10 every 3–4 days or when ∼80% confluent.

Membrane Preparation for the Radioligand Binding Assay

Radioligand binding assays were performed using rat brain or SH-SY5Y cell membranes prepared as described by Wagner, et al., 1988 [63] with slight modification. For rat brain membranes, male Wistar rats weighing 175–250 g were sacrificed by cervical dislocation and the whole brain was rapidly removed and dissected on ice. At 4°C, tissue was re-suspended in 50 mM HEPES, pH 7.4 (50 mg wet weight tissue/ml buffer), homogenized using a Brinkmann Polytron homogenizer and centrifuged for 15 min at 40,000×g. The pellet was re-suspended in 50 mM HEPES and 10 mM EDTA at pH 7.4, incubated on ice for 30 min and centrifuged at 40,000×g for 10 min. The pellet was then re-suspended in 50 mM HEPES pH 7.4 containing 10% glycerol, aliquots were made and kept at –80°C prior to use. Bicinchoninic acid (BCA) assay reagent (Pierce Rockford, IL) was used for protein quantification.

SH-SY5Y cell membranes were harvested using trypsin/EDTA, washed once with DPBS, and centrifuged for 4 min at 500×g. After centrifugation, the supernatant was discarded and the pellet re-suspended in 10 ml binding assay buffer at pH 7.2 containing (mM): 20 HEPES, 75 NaCl, 0.2 EDTA, 0.2 EGTA and complete protease inhibitor (Roche Diagnostics, AU) and sonicated. The homogenates were then centrifuged for 30 min at 40,000×g and 4°C. The supernatant was discarded and the pellet dissolved in aliquots of binding assay buffer containing 10% glycerol stored at –80°C prior to use. BCA was used for protein quantification.

Whole Cell Preparation for the Radioligand Binding Assay

Whole cells were prepared as described for SH-SY5Y cell membranes with the following modifications: after cells were harvested and centrifuged, the supernatant was discarded and the pellet re-suspended in sufficient volume of binding buffer to plate 50 µL/well in triplicates in 96 well plates. Specific ω-conotoxins binding was determined using the same concentration of protein as used for SH-SY5Y cell membranes (20 µg/50 µL), corresponds to 600.000 cells per well.

Radioligand Binding Assay

Tyr22-[125I]-GVIA, was prepared using IODOGEN, as previously described by Ahmad [64], purified using reverse phase HPLC and stored at 4°C for use within 3 weeks. On the day of the assay, membranes were thawed on ice and reconstituted to 10 µg/50 µL (rat) or 10–20 µg/50 µL (SH-SY5Y) in binding assay buffer containing 2% complete protease inhibitor and 0.1% bovine serum albumin. Stock [125I]-GVIA was diluted to 20000 cpm/50 µL or 30 pM. For displacement studies, [125I]-GVIA was incubated with rat brain or SH-SY5Y membranes or whole cells and varying concentrations of the competing ligand in triplicates in 96 well plate formats. The plates were incubated with shaking for 1 h at room temperature and vacuum filtered through a glass fiber filter pre-soaked in 0.6% polyethyleneimine (PEI), to reduce non-specific binding and washed with buffer containing (mM) 20 HEPES and 125 NaCl at pH 7.2 using a vacuum system (Tomtec harvester). The filters were then dried at 37°C before being placed in sample bags and soaked in liquid scintillant. Radioactivity was counted using a Microbeta Jet (Wallac, Finland). The non-specific binding was determined in the presence of 50 µL of unlabeled peptides.

Intracellular Ca2+ Response Measurement Using the FLIPR

SH-SY5Y cells were seeded onto 96-well or 384-well flat, clear bottom, black-walled imaging plates (Corning, Lowell, MA, US) at a density of 160,000 or 40,000 cells/well, respectively, resulting in 90–95% confluent monolayer after 48 h. On the day of the Ca2+ imaging assays, cells were loaded for 30 min in the dark at 37°C with 5 µM Fluo-4 acetomethoxyester (Fluo-4-AM), in physiological salt solution (PSS composition: NaCl 140 mM, glucose 11.5 mM, KCl 5.9 mM, MgCl2 1.4 mM, NaH2PO4 1.2 mM, NaHCO3 5 mM, CaCl2 1.8 mM, HEPES 10 mM, pH 7.4) containing in addition 0.3% BSA and 10 µM nifedipine. After the incubation period, the cells were washed once with 100 µL assay buffer (no Fluo-4-AM or BSA), and replaced with 100 µL of the same buffer. Plates were then transferred to the FLIPRTETRA (Molecular Devices, Sunnyvale, CA) fluorescent plate image reader, camera gain and intensity were adjusted for each plate to yield between 800–1000 arbitrary fluorescence units (AFU) baseline fluorescence, and Ca2+ responses measured using a cooled CCD camera with excitation at 470–495 nM, and emission at 515–575 nM. Ten baseline fluorescence readings were taken prior to the addition of antagonists, and then fluorescent readings every 2 s for 300 s before 90 mM KCl/5 mM CaCl2 buffer was added and fluorescence readings again recorded each second for further 300 s. To ensure full inhibition of Cav1 responses, the cells were pre-incubated for 40 min with 10 µM nifedipine. To ensure full inhibition of Cav2.2 responses, the cells were pre-incubated for 10 min with 1–3 µM CVID.

Statistical Analysis

Concentration-response curves were determined following nonlinear regression analysis using a 4-parameter Hill equation, with variable Hill slope fit to the functional assays data and one site fit to the radioligand binding assays; and normalized using GraphPad Prism (Version 5.00, San Diego, California). Negative and positive controls (PSS buffer and KCl 90 mM +5 mM CaCl2, respectively) were used to normalize functional data. All data is presented as mean ± SEM of 6–10 independent experiments performed in triplicate, unless otherwise stated. Statistical significance was determined using analysis of variance (ANOVA) or student’s t-test, with statistical significance defined as p<0.05, unless otherwise stated.