The inhibition of functional expression of calcium channels by prion protein demonstrates competition with α2δ for GPI-anchoring pathways

It has been shown recently that PrP (prion protein) and the calcium channel auxiliary α2δ subunits interact in neurons and expression systems [Senatore, Colleoni, Verderio, Restelli, Morini, Condliffe, Bertani, Mantovani, Canovi, Micotti, Forloni, Dolphin, Matteoli, Gobbi and Chiesa (2012) Neuron 74, 300–313]. In the present study we examined whether there was an effect of PrP on calcium currents. We have shown that when PrP is co-expressed with calcium channels formed from CaV2.1/β and α2δ-1 or α2δ-2, there is a consistent decrease in calcium current density. This reduction was absent when a PrP construct was used lacking its GPI (glycosylphosphatidylinositol) anchor. We have reported previously that α2δ subunits are able to form GPI-anchored proteins [Davies, Kadurin, Alvarez-Laviada, Douglas, Nieto-Rostro, Bauer, Pratt and Dolphin (2010) Proc. Natl. Acad. Sci. U.S.A. 107, 1654–1659] and show further evidence in the present paper. We have characterized recently a C-terminally truncated α2δ-1 construct, α2δ-1ΔC, and found that, despite loss of its membrane anchor, it still shows a partial ability to increase calcium currents [Kadurin, Alvarez-Laviada, Ng, Walker-Gray, D’Arco, Fadel, Pratt and Dolphin (2012) J. Biol. Chem. 1287, 33554–33566]. We now find that PrP does not inhibit CaV2.1/β currents formed with α2δ-1ΔC, rather than α2δ-1. It is possible that PrP and α2δ-1 compete for GPI-anchor intermediates or trafficking pathways, or that interaction between PrP and α2δ-1 requires association in cholesterol-rich membrane microdomains. Our additional finding that CaV2.1/β1b/α2δ-1 currents were inhibited by GPI–GFP, but not cytosolic GFP, indicates that competition for limited GPI-anchor intermediates or trafficking pathways may be involved in PrP suppression of α2δ subunit function.


INTRODUCTION
Ca V (voltage-gated Ca 2 + ) channels contain a pore-forming α 1 subunit, which determines the main properties, both biophysical and pharmacological, of the channels. For the Ca V 1 and Ca V 2 subfamilies, the α 1 subunit is associated with a membraneanchored predominantly extracellular α 2 δ subunit (for a review, see [1]) and a cytoplasmic β subunit (for reviews, see [2,3]). The β subunits have been shown to enhance calcium channel trafficking by binding to the I-II linker of the α 1 subunits [4,5], and thus inhibiting proteasomal decay as has been shown for the Ca V 2.2 [6] and Ca V 1.2 [7] channels. In contrast, the mechanism of enhancement of the Ca V channel functional expression by α 2 δ subunits has not been firmly established, although we have found that it involves their von Willebrand Factor A domain [8], and results in increased plasma membrane expression and active zone targeting of the channel complex [8,9].
The major mechanism whereby α 2 δ subunits increase the functional expression of calcium channels is due to an increase in the amount of channel protein at the plasma membrane ( [8,10] and J.S. Cassidy and A.C. Dolphin, unpublished work). Furthermore, it has been found that the α 2 δ subunits do not increase single channel conductance or open probability [11,12], two other mechanisms whereby macroscopic current could be increased. Nevertheless, there are effects of α 2 δ subunits to increase voltage-dependent inactivation and to hyperpolarize the voltagedependence of steady-state inactivation, providing evidence that these subunits do remain associated with the functional channel complex on the plasma membrane [8,13,14].
Mammalian genes encoding four α 2 δ subunits have been identified (for reviews, see [3,15]). The topology of the α 2 δ protein was first determined for α 2 δ-1, and is thought to generalize to all four α 2 δ subunits (for reviews, see [1,16]). They all have predicted N-terminal signal sequences, indicating that the Nterminus is extracellular. The α 2 and δ subunits are the product of a single gene, encoding the α 2 δ pre-protein, which is posttranslationally cleaved into α 2 and δ [17]. The extracellular α 2 subunit then remains disulfide-bonded to the membrane-bound δ subunit. Although α 2 δ subunits were originally described as transmembrane proteins, we have shown previously that they can form GPI (glycosylphosphatidylinositol)-anchored proteins [13].
The PrP C [normal cellular PrP (prion protein)] is a GPIanchored sialoglycoprotein [18] exposed on the outer leaflet of the plasma membrane, that is widely distributed throughout the nervous system [19,20]. Its normal physiological function is still being elucidated [20] and genetic ablation of PrP KO (knockout) only produces subtle phenotypes in mice [21,22]. However, it appears to play a role in signal transduction [23], copper binding [24], synaptic function [25] and myelin maintenance [26]. Furthermore, misfolding of PrP C results in the formation of the partially protease-resistant isoform PrP Sc (PrP scrapie), which accumulates in the brain and results in neurodegeneration, giving rise to Creutzfeldt-Jakob disease [19,27]. PrP Sc is both neurotoxic and infectious. In the absence of PrP C , PrP Sc cannot be generated and PrP-null mice do not propagate infectivity or develop pathology on infection with PrP Sc [28][29][30]. Thus it has been postulated that the presence of PrP C in neurons is essential for the initiation of a cascade of signalling events leading to prion disease pathology.
Initial evidence that the α 2 δ subunits are closely associated with PrP came from the finding that the α 2 δ-1, α 2 δ-2 and α 2 δ-3 subunits were co-immunopurified with PrP from the brains of transgenic mice expressing Myc-tagged PrP [31]. The α 2 δ subunits are expressed strongly in cholesterol-rich detergentresistant membranes [13,32], as is PrP [33,34], which may represent the basis for their association. It has been shown recently that α 2 δ-1 and PrP can co-immunoprecipitate from both brain tissue and cell lines [35], and that the interaction between pathogenic mutant PrP and α 2 δ-1 results in the intracellular retention of both species [35].
We were interested to determine whether PrP co-expression has any direct effect on calcium channel currents via a direct or indirect effect on α 2 δ subunit processing or function. We therefore utilized full-length and C-terminally truncated PrP and α 2 δ-1 to examine the effect of these species on calcium channel functional expression.

Mice
Cerebella were obtained from C57BL/6J (WT) mice or from Tg(WT-E1) mice expressing WT mouse PrP with an epitope for the monoclonal antibody 3F4 at approximately 4-fold the endogenous PrP level, referred to throughout the text as PrP Tg(WT) [42], or from PrP-KO mice with a pure C57BL/6J background (European Mouse Mutant Archive, Monterotondo, Rome; EM:01723) [21].

Preparation of Triton X-100-insoluble membrane fractions [DRMs (detergent-resistant membrane)]
All steps were performed on ice. One cerebellum was used as the starting material from WT, PrP Tg(WT) and PrP-KO mice. The cerebella were homogenized using a Teflon homogenizer in MBS [Mes-buffered saline; 25 mM Mes (pH 6.5), 150 mM NaCl and Complete TM protease inhibitor cocktail (Roche)], containing 1 % (v/v) Triton X-100 (Thermo Scientific), and left on ice for 1 h. An equal volume of 90 % (w/v) sucrose in MBS was then added. The sample was transferred into a 13 ml ultracentrifuge tube and overlaid with 10 ml of discontinuous sucrose gradient, consisting of 35 % (w/v) sucrose in MBS (5 ml) and 5 % (w/v) sucrose in MBS (5 ml). The sucrose gradients were centrifuged at 138 000 g for 18 h at 4 • C (Beckman SW40 rotor). Fractions (1 ml) were subsequently harvested from the top to the bottom of the tube. When necessary, protein fractions from the gradient were washed free of sucrose by dilution in 25 volumes of icecold PBS and ultracentrifugation (150 000 g for 1 h at 4 • C) to pellet the cholesterol-enriched microdomain material. Triton X-100-insoluble protein was resuspended in deglycosylation buffer and treated with PNGase F (peptide N-glycosidase F; Roche), as described below.

Treatment of Triton X-100-insoluble protein fractions with PI-PLC (phosphatidylinositol-specific phospholipase C)
Triton X-100-insoluble fractions from brain tissue were collected, washed free of sucrose and centrifuged as described above. The resultant pellet of Triton X-100-insoluble material was resuspended in an appropriate volume of PI-PLC reaction buffer [10 mM Tris/HCl (pH 7.4) and 150 mM NaCl containing Complete TM protease inhibitor cocktail (Roche)], to a final protein concentration of ∼ 2 mg/ml. The samples were sonicated and treated with 25 units of PI-PLC enzyme (Sigma) for 3 h at 37 • C.

Phase separation of PI-PLC-treated proteins using Triton X-114
Membrane-associated proteins were separated from soluble proteins in two phases of Triton X-114 as described previously [43]. Briefly, the pellet of detergent-insoluble material was resuspended in an appropriate volume of reaction buffer (final concentration of ∼ 2 mg/ml of protein) and incubated with PI-PLC as described above. Control experiments omitting the enzyme were also performed. After PI-PLC incubation the samples were supplemented with Triton X-114 (Thermo Scientific) to a final concentration of 1 %. A cushion of 6% (w/v) sucrose, 10 mM Tris/HCl (pH 7.4), 150 mM NaCl and 0.06 % Triton X-114 was placed at the bottom of a 1.5 ml Eppendorf tube. The protein sample was then overlaid on this sucrose cushion and the tube incubated for 3 min at 30 • C and centrifuged at 300 g for 4 min at room temperature (20 • C) in a swinging bucket rotor. Following centrifugation, the detergent phase was present as an oily droplet at the bottom of the tube. Fresh Triton X-114 was then added to the upper aqueous phase to 0.5 % and the procedure was repeated using the same sucrose cushion. In the last step the aqueous phase was removed from the cushion, supplemented with fresh Triton X-114 to 2 % and subjected to another centrifugation. The detergent phase of this last procedure was discarded. The aqueous and detergent phases from this procedure were adjusted to the equal volume with 10 mM Tris/HCl (pH 7.4) and 150 mM NaCl plus protease inhibitors.

Acetone precipitation and PNGase F deglycosylation
To remove the remaining Triton X-114 in the detergent and aqueous phases from the phase separation experiment the proteins were precipitated by the addition of 4 volumes of ice-cold acetone and subsequent incubation for 1 h at − 20 • C. The precipitated material was centrifuged at 16 000 g for 10 min and the pellet was washed once with an acetone/water (4:1) mixture ( − 20 • C). The pellets of the precipitated proteins were then resuspended in 45 μl of PNGase F buffer [10 mM Tris/HCl (pH 7.5) and 150 mM NaCl supplemented with 75 mM 2-mercaptoethanol, 0.5 % Triton X-100, 0.1 % SDS and protease inhibitors]. A total of 1 unit of PNGase F was added per 10 μl volume followed by incubation at 37 • C for 5-12 h. The samples were then resuspended in an appropriate volume of SDS gel loading buffer and heated for 10 min at 56 • C in order to terminate the reaction.

Electrophysiology
Calcium channel expression in tsA-201 cells was investigated by whole-cell patch-clamp recording essentially as described previously [46]. The internal (pipette) and external solutions and recording techniques were similar to those described previously [47]. The patch pipette solution contained: 140 mM caesium aspartate, 5 mM EGTA, 2 mM MgCl 2 , 0.1 mM CaCl 2 , 2 mM K 2 -ATP and 10 mM Hepes, pH 7.2 at 310 mM mOsm, with sucrose. The external solution for recording Ba 2 + currents contained: 150 mM tetraethylammonium Br, 3 mM KCl, 1 mM NaHCO 3 , 1 mM MgCl 2 , 10 mM Hepes, 4 mM glucose and 1 mM BaCl 2 , pH 7.4 at 320 mOsm, with sucrose. Pipettes of resistance 2-4 M were used. An Axopatch 1D amplifier (Axon Instruments) was used, and data were filtered at 1-2 kHz and digitized at 5-10 kHz. Current records were subjected to leak and residual capacitance current subtraction (P/8 protocol). Analysis was performed using Pclamp 9 (Molecular Devices) and Origin 7 (Microcal Origin). All comparisons between different groups of transfected cells were performed in parallel on the same experimental days.
When stated, current-voltage (I-V) plots were fit with a modified Boltzmann equation for determination of the voltage for 50 % activation (V 50 , activation) [48], and steady-state inactivation curves were fit with a single Boltzmann function for determination of the voltage for 50 % inactivation (V 50 , inactivation). Where data are given as means + − S.E.M., statistical comparisons were performed using either Student's t test or ANOVA with the appropriate post-hoc test.
Xenopus oocytes were prepared, injected and utilized for electrophysiology as described previously [49], with the following exceptions. Plasmid cDNAs for the calcium channel α1, α 2 δ-1 and β1b subunits and for PrP were mixed at 2:1:2:x ratios at 1 μg/μl, where x = 2, 1 or 0.5 unless otherwise stated, and the empty vector was included to maintain the total cDNA mixture constant. A 9 nl volume was injected intranuclearly, after 2-fold dilution of the cDNA mixtures. Recordings in Xenopus oocytes were performed at 18 • C as described previously [48], and all recordings were performed 48-60 h after injection for Ca V 2.2. The Ba 2 + concentration was 10 mM.
It is well-established that PrP is GPI-anchored [18], and our previous findings indicate that the α 2 δ subunits can form GPI-anchored proteins [13]. One important piece of evidence supporting this premise is that following PI-PLC treatment, both endogenous α 2 δ subunits from the rat brain and heterologously expressed α 2 δ proteins redistribute from the detergent phase into the aqueous phase upon Triton X-114 phase separation [13]. This demonstrates that PI-PLC treatment has converted the protein into a hydrophilic species, most probably by removing the bulky hydrophobic GPI anchor. To assess if the behaviour of α 2 δ proteins in this phase-separation assay was affected by the presence or absence of PrP, we concentrated the DRM fractions enriched in both PrP and α 2 δ and subjected them to PI-PLC treatment as described previously [13]. An aliquot of the initial DRM material from the three mouse genotypes, before PI-PLC treatment, was deglycosylated with PNGase F to demonstrate the presence of α 2 δ-1 and α 2 δ-2 ( Figure 1B).
PI-PLC treatment caused α 2 δ-1 and α 2 δ-2 to redistribute into the aqueous phase upon Triton X-114 phase separation in all three of the mouse genotypes tested ( Figure 1C). As a control, the PrP Tg(WT) samples were also blotted for PrP, and redistributed similarly upon PI-PLC treatment ( Figure 1D).
Importantly, the response of α 2 δ-1 and α 2 δ-2 to PI-PLC treatment was not affected by the absence of PrP, in the PrP KO cerebellar material ( Figure 1C). It was also not affected by the overexpression of PrP in the PrP Tg(WT) cerebellar material ( Figure 1C). These data indicate that although PrP has been reported to interact with α 2 δ subunits [35], interaction with PrP was not responsible for the presence of α 2 δ subunits in DRM fractions or their redistribution into the aqueous phase following PI-PLC cleavage of GPI-anchor sites.

Co-expression of PrP decreased calcium channel currents containing Ca V 2.1/β4/α 2 δ-2
We then examined whether PrP would influence the ability of α 2 δ to increase calcium channel currents. We expressed Ca V 2.1/β4 and α 2 δ-2 to mimic the calcium channel combination present in cerebellar Purkinje cells, either with or without PrP. We found that PrP co-expression produced a moderate, but consistent, reduction in peak calcium channel currents at + 10 mV of approximately 34 %, although the currents in the presence of PrP and α 2 δ remained significantly larger than those in the absence of α 2 δ (Figures 2A-2C). The voltage-dependence of activation of the Ca V 2.1/β4/α 2 δ-2 currents in the presence of PrP was also significantly depolarized, compared with in its absence, although not to the same extent as in the absence of α 2 δ ( Figure 2D). Similarly, the voltage-dependence of steady-state inactivation of Ca V 2.2/β4/α 2 δ-2 was depolarized significantly in the additional presence of PrP (Figures 2E and 2F). All these effects are indicative of a reduced enhancement by α 2 δ-2 of Ca V 2.1/β4 calcium channel currents when PrP was co-expressed.

Co-expression of PrP decreased calcium channel currents containing Ca V 2.1/β1b/α 2 δ-1
We then examined whether PrP would also inhibit calcium channel combinations containing α 2 δ-1, and tested whether any effect could be due to the demonstrated interaction at the level of the two polypeptides [35] or a result of lipid anchoring and co-localization in the same cholesterol-rich microdomains. We therefore expressed Ca V 2.1/β1b/α 2 δ-1 with either WT PrP or anchorless PrP, truncated at the GPI-anchor site ( GPI-PrP) [38]. We observed that co-expression with WT PrP reduced the peak calcium channel currents to a similar extent (39 %) to that shown for the Ca V 2.1/β4/α 2 δ-2 combination (Figures 3A-3C), whereas GPI-PrP had no effect on the Ca V   δ-1 calcium channel currents (Figures 3A-3C). In these experiments WT PrP did not produce a significant depolarization of the voltage-dependence of activation of the currents ( Figure 3D). In another expression system, Xenopus oocytes with microinjected cDNAs and where the ratio of the different constructs can be controlled accurately, the effect of PrP to reduce Ca V 2.2/β1b/α 2 δ-1 calcium channel currents was observed at an α 2 δ-1/PrP ratio of 1:1, but not at 2:1 or 4:1 ( Figures 3E and 3F), indicating that it was dependent on the amount of PrP expressed relative to α 2 δ. Taken together, these results suggest that the effect of PrP to reduce calcium channel currents is concentrationdependent, and the presence of a GPI anchor on PrP may be essential for the observed reduction in the functional expression of calcium currents. with either WT α 2 δ-1 or anchorless α 2 δ-1 (α 2 δ-1 C-HA), together with either WT PrP or anchorless PrP truncated at the GPI-anchor site ( GPI-PrP). We have shown previously that an anchorless form of α 2 δ-1, although being mainly secreted, is still able to promote an increase in calcium channel currents compared with no α 2 δ, although to a smaller extent than WT α 2 δ-1 [36]. In the present study we obtained the surprising result, shown in Figures 4(A)-4(C), that, whereas WT PrP inhibited calcium channel currents in the presence of WT α 2 δ-1, it did not produce any reduction in the presence of anchorless α 2 δ-1 C-HA ( Figure 4C). In agreement with the data shown in Figure 3, GPI-PrP did not produce any inhibition of calcium   [53]. (E) Western blot of α 2 δ-1 C in medium (upper panels; 3-8 % Tris/acetate gel) and cell lysate (lower panels; 4-12 % Bis-Tris gel) co-expressed (1:1) with PrP (left-hand panels), GPI-PrP (middle panel) or empty vector (right-hand panels; con). In both (D) and (E), all three lanes are from the same blots from which irrelevant lanes have been excised and the molecular mass is given on the right-hand side in kDa.
channel currents formed with WT α 2 δ-1 ( Figure 4C). Similarly GPI-PrP did not produce any reduction in the currents formed with anchorless α 2 δ-1 C-HA ( Figure 4C). We found from Western blotting that the WT and truncated α 2 δ and PrP species used are all well-expressed in tsA-201 cells ( Figures 4D and  4E), and lack of expression therefore could not account for the lack of effect of the truncated constructs when expressed together.
This result suggests either that association of the α 2 δ and PrP proteins in the plasma membrane, and potentially in the same cholesterol-rich membrane microdomains, facilitated by their GPI anchoring, is essential for the inhibition of calcium currents observed or that there is competition between α 2 δ-1 and PrP for the GPI-anchor intermediates, resulting in less maturation of α 2 δ subunit in the presence of PrP. We therefore examined whether the effect of PrP might be related to the GPI anchor and tested the c The Authors Journal compilation c 2014 Biochemical Society effect of an unrelated protein engineered to contain a GPI anchor (GFP-GPI). We found that co-expression of GFP-GPI markedly reduced calcium channel currents formed by Ca V 2.1/β1b/α 2 δ-1 compared with the co-expression of cytosolic GFP, and in this experiment GPI-PrP again had no effect ( Figure 5A). This result is in agreement with the hypothesis that overexpression of a GPI-anchored protein can produce this inhibition. We then compared the level of α 2 δ-1 expression in the presence of coexpressed GFP-GPI and cytosolic GFP ( Figure 5B). In all cases substantial α 2 δ-1 was co-expressed.

DISCUSSION
In the present study we found that when PrP is co-expressed with Ca V 2.1 calcium channels together with a β-subunit and either α 2 δ-1 or α 2 δ-2, it produces a modest reduction in the expressed calcium channel currents. This reduction in calcium channel currents involves the membrane anchoring of PrP, as it is absent when PrP is truncated prior to its GPI-anchor site. Thus GPI-PrP did not inhibit currents formed by the expression of Ca V 2.1/β1b/α 2 δ-1. This result suggests that the process might involve competition for pathways associated with the GPI anchoring of PrP, which might relate to our finding that α 2 δ subunits form GPI-anchored proteins [13]. The results of the present study reinforce that view, since native cerebellar α 2 δ-1, α 2 δ-2 and PrP all behave similarly following PI-PLC treatment and phase separation, being concentrated in the aqueous phase (Figure 1), which is supportive of evidence for GPI anchoring.
We therefore utilized an α 2 δ-1 subunit truncated at the predicted membrane attachment site (α 2 δ-1 C), which, as we have shown recently, is not an integral membrane protein [36]. We have characterized α 2 δ-1 C and found that, despite being mainly secreted, it still shows a partial ability to enhance calcium channel currents and to be associated extrinsically with the plasma membrane [36]. It is therefore possible that α 2 δ-1 enhances calcium currents by two mechanisms, one involving a trafficking process that is still maintained by anchorless α 2 δ-1 C, and another mechanism that requires its membrane anchoring, for example a process that stabilizes channel complexes in the plasma membrane.
Importantly, we also found in the present study that WT PrP does not inhibit calcium channel currents when α 2 δ-1 C was substituted for WT α 2 δ-1. This suggested to us two possibilities. First, PrP and α 2 δ-1 may compete for GPI-anchor intermediates or subsequent trafficking processes specific for GPI-anchored proteins when both are overexpressed. Secondly, the interaction between PrP and α 2 δ-1 with other protein and lipid components in cholesterol-rich membrane microdomains [50,51] might be involved in the suppression of currents. Thus when either PrP or α 2 δ-1 is devoid of the lipid anchor no suppression occurs.
It is possible that WT α 2 δ-1 and PrP interact indirectly via an intermediary trafficking protein and compete for binding to that protein, whereas this does not occur when one of them is not membrane-anchored. Nevertheless, the additional finding that Ca V 2.1/β1b/α 2 δ-1 currents were also inhibited by GPI-GFP (when compared with the control GFP) suggests that the presence of a GPI anchor might be an important common denominator in this process. Only approximately 150 proteins are known to be GPI-anchored, and the GPI synthetic pathway, involving multiple enzymatic steps in the ER (endoplasmic reticulum) and Golgi, is likely to have limited capacity in mammalian cells [52]. Overexpression of multiple GPI-anchored proteins in the same cell might lead to saturation of limited resources. Further studies will be necessary to determine whether overexpression 1/β1b/α 2 δ-1 with GFP (; n = 9), GPI-GFP (᭡; n = 9), or GFP and GPI-PrP (᭝; n = 6). The broken line indicates the level of I Ba observed for Ca V 2.1/β1b/α 2 δ-1 plus WT PrP from Figure 3(B). ***P < 0.001 between the peak I Ba when GPI-GFP was co-expressed compared with when GFP was co-expressed (Student's t test). (B) Western blot of α 2 δ-1 (upper panels) co-expressed (1:1) with GFP-GPI (lane 1), GFP (lane 2) or empty vector (lane 3; con). GFP expression is shown in the lower panel. All three lanes are from the same blots from which irrelevant lanes have been excised and the molecular mass is given on the right-hand side in kDa.
of other GPI-anchored proteins disrupts the function of α 2 δ within membrane microdomains, or possibly competes with it for GPI transamidase enzymes, GPI-anchoring intermediates or trafficking proteins subsequent to GPI anchoring. Taken together, the results of the present study provide supporting evidence that α 2 δ-1 can be GPI-anchored. More generally the results point to a competitive interaction between GPI-anchored proteins for resources that determine their level of post-translational processing or expression on the plasma membrane.