CaV2.1 α1 R192Q mutant KI and WT littermates were used for the experiments. Genotyping was routinely performed as previously described [25, 27]. Homozygous R192Q KI and WT littermates used for this study share the same genetic background (~97% C57Bl6J background), as described in van den Maagdenberg et al , and originally backcrossed for five generations followed by further in-house backcrossing to preserve the original phenotype. All procedures were approved by the local ethical committee for animal experimentation in accordance with guidelines of the International Association for the Study of Pain. Animals were maintained in accordance with the Italian Animal Welfare Act and their use was approved by the Local Authority Veterinary Service.
Cell culture preparation of TG neurons
Trigeminal ganglia primary cultures from KI and WT mice were prepared as previously described (P10-14), and used after 24 h from plating [29, 55].
Membrane protein extraction
For total membrane protein extraction, trigeminal ganglia were lysed in phosphate saline buffer (PBS) containing Triton X-100 1%, 100 mM NaF, 20 mM orthovanadate plus the protease inhibitors cocktail (Sigma, Milan, Italy), incubated on ice for 45 min and centrifuged for 20 min at 10,000 g at 4°C. Total ganglia membranes were subjected to ultracentrifugation at 100,000 g for 1 h (4°C) with a fixed-angle rotor [56, 57]. Triton-insoluble membrane pellets were dissolved in sample buffer (100 mM Tris-HCl pH 6.8, 200 mM dithiothreitol, 4% SDS, 20% glycerol, 8 M urea) and separated on 8% polyacrylamide gel, and represented the raft fraction. The remaining supernatant (Triton-soluble) was defined as non raft fraction.
To ensure correct equal loading for neuronal cell content in different lysates, protein extracts were quantified with bicinchonic acid (Sigma) and normalized for the neuronal specific β-tubulin III. The amount of loaded proteins was in the 20-50 μg/mL range. Immunoprecipitation of P2X3 receptors from the two fractions (raft and non-raft) was performed as described earlier . β-tubulin or β-actin were used for gel loading reference.
Sucrose density gradient preparation
Sucrose density gradients were prepared as described previously . Briefly, cells were lysed in TNE buffer (10 mM Tris HCl pH 7.4, 150 mM NaCl, 2 mM EDTA, 20 mM NaF, 20 mM orthovanadate) plus protease inhibitors (30 min on ice) followed by sonication; insoluble membrane proteins were purified by ultracentrifugation (100,000 g for 60 min at 4°C). Pellets were resuspended in 500 mM Na2CO3, and added to 25 mM 2-N-morpholino ethanesulfonic acid (MES buffer), 150 mM NaCl, at pH 6, with 90% sucrose and protease inhibitors to reach a final dilution of 45% sucrose. The lysate was transferred to a TLS55 ultracentrifuge tube with a swing-bucket rotor to separate lipid rafts , and layered sequentially with 90, 45 and 5% sucrose in MES buffer. Samples were centrifuged at 200,000 g for 16 h and 100 μl fractions were collected from the top. 1% CHAPS was added to each fraction (to allow correct migration in western blotting) that was diluted with TNE buffer (1:5) and kept for 45 min on ice. Proteins were concentrated by centrifugation at 13,000 g for 30 min at 4°C. Final pellets were resuspended in sample buffer (100 mM Tris-HCl pH 6.8, 200 mM dithiothreitol, 4% SDS, 20% glycerol, 8 M urea) and separated on 8% polyacrylamide gel.
Western immunoblotting was performed using the following antibodies: rabbit anti-P2X3 (1:300; Alomone, Jerusalem. Israel), mouse anti-flotillin 1 (1:250, BD Biosciences, Franklin Lakes, NJ, USA), mouse anti β-actin (specific HRP-conjugated; dilution 1:1000; Sigma) and mouse β-tubulin III (1:1000, Sigma).
As secondary antibodies, to avoid detection of immunoglobulin heavy chains in Western blot, previously-validated HRP-conjugated antibodies (Jackson ImmunoResearch, Suffolk, UK) were used . Signals were detected with the enhanced chemiluminescence light system ECL (Amersham Biosciences, Piscataway, NJ). For quantification of intensities of the immunoreactive protein bands we used Scion Image software (NIH, Bethesda, USA) or the digital imaging system UVTEC (Cambridge, UK).
Cholesterol and rafts labelling
Cholesterol distribution in cultured trigeminal neurons was analyzed using a cell based cholesterol assay kit (Cayman, Ann Arbor, MI) [59, 60], based on filipin staining (5 μg/mL, 10-20 min), a compound that forms fluorescent complexes with unesterified cholesterol .
For detection of lipid rafts, paraformaldehyde fixed trigeminal neurons were washed with PBS and incubated for 30 min with blocking solution (5% FCS, 5% BSA, 0.1% Triton-X) followed by 3 times washout with PBS, and then incubated with FITC conjugated cholera toxin B subunit (3 μg/mL; Sigma) for 10 min and then washed thrice with PBS. Labelled cells were viewed with a Zeiss microscope and acquired with MetaView software (Molecular Devices, Downingtown, PA, USA) in non-saturation mode. To quantify data, basal threshold was arbitrarily set to zero, and grey values were then analyzed with MetaMorph software (Molecular Devices, Downingtown, PA, USA). Response intensity was evaluated for each cell from each region of interest (0.075 mm2) and expressed as arbitrary units (AU). An average of fifty cells was analyzed in each test; data are the mean of at least three independent experiments.
Patch Clamp Recording
Currents were recorded from mouse trigeminal neurons in culture as previously described  under whole cell voltage clamp mode at a holding potential of -60 mV. Cells were continuously superfused with control solution containing (in mM): 152 NaCl, 5 KCl, 1 MgCl2, 2 CaCl2, 10 glucose, 10 HEPES; pH 7.4 adjusted with NaOH. Patch pipettes had resistance of 3-4 MΩ when filled with (in mM): 140 KCl, 2 MgCl2, 0.5 CaCl2, 2 ATP-Mg, 2 GTP-Li, 20 HEPES, 5 EGTA; pH 7.2 adjusted with KOH. Data were acquired and analyzed with the pCLAMP software Clampex 9.2 (Molecular Devices, Palo Alto, CA, USA).
The agonist α,β-meATP (Sigma) was applied for 2 s by rapid solution changer system (Perfusion Fast-Step System SF-77B, Warmer Instruments, Hamden, CT, USA). The cholesterol depleting agent methyl-β-cyclodextrin (MβCD; Sigma) was dissolved in water and then applied to the cells in culture at the concentration of 10 mM for 30 min in the cell incubator in accordance with previous reports [22, 61]. Electrophysiological recordings started after 10 min of wash with control solution and continued for approximately 1 h. This protocol of MβCD application increased the holding current of patch-clamped neurons from -50 ± 9 to -85 ± 9 pA (n = 65 and 49; p = 0.01).
The currents were analyzed in terms of their peak amplitude and current onset (10-90% of the response rise-time). The decay of the current during agonist application due to receptor desensitization was fitted with a biexponential function using Origin 6.0 (Microcal, Northampton, MA, USA) which provided the desensitization time constants (τfast and τslow). Recovery from desensitization was assessed by a paired pulse protocol over 30 s intervals in accordance with previous reports [27, 55].
All data are presented as mean ± standard error of the mean (S.E.M); n is number of cells. Statistical significance was evaluated with unpaired Student's t-test (for parametric data) or Mann-Whitney-Wilcoxon test (for nonparametric data), p ≤ 0.05 was considered significant.
Kinetics of the P2X3 receptor mediated currents was simulated using the cyclic Markov state model  shown in Figure 5A. It assumes the binding of the three molecules of the agonist to the receptor in the resting state R, the isomerisation of the receptor into the open state Ro, followed by transition into the rapidly developing desensitized state Df, and the slower occurring desensitized-bound state A3D with slow agonist dissociation. Standard desensitization is inferred to develop from open bound channels A3Ro.
In total, in the model there are 10 states and 26 rate constants. The original model used rate constants applied to membrane currents recorded from rat dorsal root ganglion neurons. To adapt this scheme to trigeminal mouse sensory neurons that possess distinct time-course of desensitization and recovery , manual optimization of the rate constants was performed, using a PC, with our in-house developed program  written in Delphi. Hence, the peak amplitude, onset and decay time constants and recovery of the α,β-meATP induced P2X3 receptor mediated currents in control conditions and after cholesterol depletion with MβCD were obtained with the rate constant values indicated in Table 1 to reproduce the experimentally-obtained records. As we observed a comparable experimental change in WT and KI responses after MβCD, only simulation of the WT currents is shown.