Monomeric α-synuclein activates the plasma membrane calcium pump

aSN co-localizes with PMCA and stimulates calcium extrusion

Co-immunoprecipitation experiments with brain homogenate from aSN-knockout mice supplemented with purified aSN monomer revealed aSN binding, and a similar experiment with aSN oligomer showed the same result indicating no discrimination on aSN conformation (Fig 1A, left). Co-immunoprecipitation and western blotting of endogenous aSN and PMCA from detergent extracts of total brain homogenates from wild-type C57BL/6 mice show that endogenous PMCA interacts with endogenous aSN (Fig 1A, right). PMCA appears with multiple bands, or a smear on western blots from the two co-IPs, likely due to the varying amounts of post-translational modifications of PMCA (Tolosa de Talamoni et al, 1993). Moreover, a pull-down assay performed with purified proteins showed aSN binding to PMCA immobilized on Calmodulin sepharose (Fig 1B).

The interaction between aSN and PMCA was investigated further in primary hippocampal neurons derived from newborn C57BL/6 mice and with neurons from aSN-KO mice as negative controls. Primary hippocampal neurons were fixed and analyzed after 14 days in culture by immunofluorescence labeling of synapses by synaptophysin and an aSN/PMCA proximity ligation assay (PLA). The PLA is used for in situ detection of protein interactions and is based on two primary antibodies from different species, here ASY-1, a polyclonal aSN antibody generated in rabbit, and 5F10, a monoclonal pan PMCA antibody generated in mouse. The secondary antibodies were labeled with oligonucleotides that enzymatically can be ligated when in proximity and then amplified, hence generating concatemeric sequences. These sequences can bind fluorescently labeled oligonucleotides yielding a red fluorescent signal when aSN and PMCA are located within approximately 40 nm. PMCA and aSN were found to be in this proximity of each other and located to the synapses (Fig 1C).

The functional effect of PMCA calcium transport was investigated in primary hippocampal neurons from aSN-KO neurons transiently transfected with either aSN and mCherry, or mCherry alone as negative control. SERCA was inhibited by thapsigargin (4 μM) before recording. Cellular calcium responses were monitored by the calcium-sensing dye Fura2-AM upon depolarization by 8 mM KCl in the extracellular medium. Neurons expressing aSN expelled calcium to a markedly higher degree than the mCherry expressing neurons (Fig 1D and E). Vanadate is a well-known and potent inhibitor of P-type ATPases (Cantley et al, 1978; Bond & Hudgins, 1980; Dupont & Bennett, 1982) and was used in this experiment as a PMCA inhibitor, since SERCA was inhibited by thapsigargin from the start. Vanadate decreased the calcium expulsion of aSN mCherry neurons to the same level as mCherry-only neurons treated with vanadate. This supports that increased efflux induced by aSN is driven by PMCA.

Calcium extrusion experiments similar to those in neurons were also attempted in three different cell lines, SH-SY5Y, PC12, and Chromaffin cells, to investigate further variations of background conditions. In the human neuroblastoma cell line SH-SY5Y, with doxycycline controllable wild-type aSN expression, the response to K⁺ dependent depolarization had great variability, as have been observed previously (Morton et al, 1992). For cells with a clear response to K⁺ dependent depolarization, the SH-SY5Y cells overexpressing aSN exhibited increased calcium expulsion compared to SH-SY5Y with no aSN expression (Appendix Fig S1A–C). In PC12 cells, the response to K⁺-dependent depolarization gave a good signal, but no difference between control cells and cells overexpressing aSN was observed (data not shown). This can have many causes, but we did not explore it further. For Chromaffin cells, we did not detect a significant response to K⁺-dependent depolarization. Hence, in particular primary hippocampal neurons, and to some extend also SH-SY5Y cells with inducible overexpression of wild-type aSN, show aSN-stimulated PMCA activity.

Monomeric aSN activates human PMCA in a lipid-dependent manner and independent of the autoinhibitory domain of PMCA

The measurements of calcium-dependent ATPase activity of PMCA were performed with two human PMCA—ubiquitous PMCA1 and neuron-specific PMCA2 isoforms (specifically the splice variants PMCA1d and PMCA2w/a). The experiments were performed with PMCAs relipidated either with porcine brain phosphatidylcholine (BPC, neutral lipids) or with bovine brain lipid extract Folch fraction I (Sigma-Aldrich, from here on referred to as brain extract lipids, BE), of which 60% are negatively charged phosphatidylinositol and phosphatidylserine lipids (Folch et al, 1957; Boura & Hurley, 2012). We observed monomeric aSN causing a strong increase in the activity of both PMCA isoforms, however only for PMCAs relipidated with BE, not BPC. Ca²⁺ titration revealed that the elevation of ATPase activity is also accompanied by a three- to five-fold increase in the apparent calcium affinity (Fig 2A, left, center; Table 1). Additionally, the activation is gradually abolished by increasing the content of neutral BPC in the relipidation mixture (Fig 2A, right). To rule out potential, unspecific effects of other ATPases or impurities in the PMCA preparations, we prepared an inactive variant of PMCA (PMCA1d_D475N), in which the reactive aspartyl residue in the active site was replaced by asparagine. In terms of purity, these preparations were no different from the wild type, as confirmed by SDS–PAGE and size exclusion chromatography (Appendix Fig S2A–D), and they showed no ATPase activity, neither with nor without aSN (Fig 2A left, 2B).

To further investigate the acidic phospholipid dependence, we performed aSN titration comparing effects of wild-type aSN and the A30P variant with reduced membrane-binding ability (Jensen et al, 1998). In BE, the activation of PMCA was significantly attenuated by the A30P mutation. As could be expected, the aSN variant was ineffective in the neutral environment of BPC (Fig 2B).

Next, we examined the potential interplay between aSN and the well-known protein activator of PMCA, namely Calmodulin (CaM). For both PMCA isoforms, the susceptibility for activation by CaM and aSN appears to be lipid-dependent, with the acidic lipids promoting the effect of aSN and neutral lipids promoting the effect of CaM (Fig 2C). To confirm that the action of aSN was independent of the autoinhibitory domain, we designed a truncated construct (PMCA1-ΔC), lacking 184 residues of the C-terminus including the autoinhibitory domain with the CaM-binding site(s) and a long, intrinsically disordered linker region (Fig 2D, top). Foreseeably, loss of the C-terminal tail resulted in a constitutively active protein, insensitive to CaM (Appendix Fig S3A) and with basal activity higher than that of the full-length wild type (Fig 2D, bottom left). However, the stimulation by aSN was preserved and appeared like for the full-length pump, indicating the aSN effect to be independent of the autoinhibitory mechanism associated with the C-terminal tail (Fig 2D, bottom right). The apparent Ca²⁺ affinity of PMCA1d was increased by the truncation itself, and aSN strengthened that effect (Table 1). Again, the activation was lipid-dependent, occurring in the PMCA1-ΔC variant reconstituted in BE and not in BPC (Appendix Fig S3A).

Additionally, we examined the oligomeric form of the full-length aSN, which was previously established to have a great activating effect on SERCA (Betzer et al, 2018). For PMCA, however, the effect of oligomers was smaller than that of the monomer, occurring only at higher concentrations and to a lower fold of activation (Appendix Fig S3B). The activation might simply be explained by dissociating monomer from the oligomer, which we have found to be present in oligomer preparations (Appendix Fig S4). Moreover, we measured the PMCA activity in presence of two other synuclein isoforms, beta- and gamma-synuclein. Both stimulated the pump, but to a lesser extent (Fig EV1A and B).

Click here to expand this figure.

PMCA alternative splicing events correlate with aSN expression level

PMCA is extensively regulated by alternative splicing, which results in over 20 variants, differing in distribution, kinetic properties, and CaM sensitivity (Strehler & Zacharias, 2001; Krebs, 2015). We wondered if PMCA splice variants could vary also in how they respond to aSN. We used available transcriptomics data from the VastDB database, http://vastdb.crg.eu/ (Tapial et al, 2017), containing information on alternative splicing events. The level of alternative splicing events in different tissues are indicated by PSI values (“percent spliced in”), and we asked whether specific splicing events correlate with aSN expression levels. First, we analyzed the splice site C, which is located at the C-terminal region of PMCA and contains the autoinhibitory domain that interacts with CaM. Alternative splicing at this site gives variants that may differ in the degree of autoinhibition or rate of activation by CaM. The splice variant “a,” with a shortened autoinhibitory domain, is known to have higher basal activity and to be less sensitive to CaM stimulation (Caride et al, 2007). The available data regarded PMCA1, PMCA3, and PMCA4, and the analysis displayed in Fig 3A shows that in the case of PMCA1 and PMCA4, incorporation leading to the variant “a” positively correlates with the expression level of aSN with only a few outliers. Only PMCA3 did not correlate and the PSI value of the “a” variant was relatively high regardless of aSN expression interval. The results suggest that in tissues expressing aSN at high levels, PMCA is alternatively spliced in a way that favors less CaM-sensitive variants. The PMCA “a” variant is mainly expressed in brain tissues and indicates a positive correlation between PMCA and aSN. Besides brain tissues, aSN is also expressed in high amounts in melanocytes and bone marrow, which account for the two outliers in the plot.

The other splice site—termed A—is located at the loop between the A-domain and TM3 (from here on referred to as “A-TM3 loop”). This loop is much longer in PMCAs compared to other P-type ATPases. Among isoforms, PMCA1 is not being spliced in this region, while in PMCA2 alternative exon composition leads to four different variants named “z,” “x,” “y,” and “w,” with all of them except “y” having been detected in humans (Strehler et al, 2007b; Di Leva et al, 2008). Figure 3B displays the analysis of the PMCA2 splicing events, where three exons can be incorporated leading to variants “w” (full length containing sequences marked in orange, yellow, and blue) and “x” (containing the blue sequence). The “z” variant occurs when none of the three exons are incorporated (Fig 3C). We observed incorporation of the exons leading to variants with a longer A-TM3 loop insert in tissues with low aSN expression and barely any incorporation of these exons for tissue with high expression of aSN. This suggests that low aSN expressing tissues would preferentially transcribe the longest PMCA2w, and tissues richer in aSN would express more PMCA2x and z.

The extended, variable A-TM3 loop of PMCA is predicted to be structurally disordered; a trait which was confirmed using circular dichroism (CD) spectroscopy analysis of a recombinantly expressed PMCA2w_A-TM3 loop (Fig 3D). Near TM3, the loop contains several positively charged residues that mediate the activation by acidic phospholipids (Brodin et al, 1992; Pinto Fde & Adamo, 2002; Brini et al, 2010). We speculated about the importance of the loop length in the interaction with disordered aSN and performed the free calcium titration experiment with PMCA2 variants w/a, x/a, and z/a. All PMCA2 constructs used in the experiment were transformed anew into yeast cells derived from one colony and were expressed, purified, and tested simultaneously. The activity assays show that all three variants are activated by aSN (Fig 3E). The most remarkable difference is that PMCA2w/a has the lowest maximum activity among variants when activated by aSN. It was also the only one with a substantial increase in the apparent calcium affinity induced by aSN, the other having high affinity also in the basal activity state. PMCA2z/a, having the shortest loop region, showed the highest basal activity. The results implicate the A-TM3 loop in aSN interactions.

Indirect interaction between aSN and PMCA occurs between the N-terminal region of aSN and the acidic phospholipid binding site of PMCA

To determine whether the activation of PMCA was triggered by a direct protein–protein interaction of aSN with the PMCA_A-TM3 loop, we performed interaction studies using NMR spectroscopy. We incubated recombinant PMCA2w_A-TM3 loop with ¹⁵N-labeled aSN, however observing no chemical shift perturbations (CSPs; Fig EV2A), in both the absence (Fig EV2B) and presence of Ca²⁺ (Fig EV2C). Further, we found no indications of a direct interaction between full-length PMCA2w/a and aSN, as no CSPs (Fig EV2D) or changes in NMR peak intensities (Fig EV2E) could be seen in the aSN spectra. This suggests the interaction between aSN and PMCA is dependent on a more complex environment and likely is occurring via a mechanism facilitated by the membrane.

Click here to expand this figure.

To map the potential interaction site on the aSN sequence (1–140), we performed a comparison of shortened aSN variants, truncated N-terminally (Δ14, Δ29) to compromise membrane binding (Cholak et al, 2020) or C-terminally (1–95, 1–61) to compromise low-affinity calcium binding and to remove the NAC binding region (Skaanning et al, 2020) respectively (Fig 4A). Most notably, the Δ14 variant had a greatly weakened effect on PMCA and Δ29 had almost lost its ability to activate PMCA, displaying only minor effects above 5 μM concentration. The C-terminally truncated (1–95) variant generally maintained function although to a lesser extent, activating to a lower fold (Fig 4B, Appendix Fig S3C). The (1–61) variant of aSN did not reach saturation within the examined concentration range, which suggests greatly reduced apparent affinity to the PMCA association.

The disordered C-terminal tail of aSN has calcium-binding properties (Lautenschlager et al, 2018) with a relevant affinity of ~ 20–50 μM (Newcombe et al, 2021), and thus we asked if it would impact the apparent calcium affinity of PMCA. Here, we compared activation of PMCA by the full length and C-terminally truncated (1–95) aSN in the Ca²⁺ titration experiment. However, despite a lower-fold activation, the C-truncated construct did not differ in shifting PMCA's apparent K_d for Ca²⁺ (Appendix Fig S3C).

Within certain limitations, deletions of large fragments of the lengthy A-TM3 loop of PMCA, including its binding site for acidic, negatively charged lipids, can retain basal activity (Pinto Fde & Adamo, 2002). Guided by this, we designed constructs and expressed and purified deletion variants of PMCA2, where we removed large parts of the loop. Sequences of the targeted region of PMCA2w/a and three designed variants (Δ292–383, Δ298–383, Δ298–372) are displayed in Fig 4C. The activity assay in the presence of BE (Fig 4D) showed differences in the basal activity and response to aSN. We observed a complete loss of ATPase activity of the Δ292–383 variant. Keeping the six N-terminally located residues of the loop resulted in the Δ298–383 variant with a basal activity preserved (0.84 ± 0.02 nmol Pi/μg PMCA/min.), but not response to aSN titration. Preserving further the 11-residue-long binding site for negatively charged lipids at the C-terminal end of the loop close to TM3 (Δ298–372) boosted basal activity (3.4 ± 0.5 nmol Pi/μg PMCA/min) and reinstalled the activation by aSN to the level observed in wild type PMCA2w/a (which has however a lower basal activity of 2.3 ± 0.3 nmol Pi/μg PMCA/min). This points to the binding site for negatively charged lipids as crucial for the interaction with aSN, and we propose the aSN-PMCA interaction to be mediated by acidic phospholipids involving the N-terminal region (1–95) of aSN and the PMCA binding site for negatively charged lipids (Fig 4E).

The activation by aSN is specific to mammalian calcium pump

PMCA activation by aSN is observed both for the ubiquitous PMCA1 and the more tissue-specific PMCA2 and was also verified for the C-terminally truncated PMCA4x isoform (Appendix Fig S3D), which likely colocalizes with aSN in synaptic terminals of retinal neurons (Krizaj et al, 2002; Bodis-Wollner et al, 2014; Cali et al, 2017). The broad impact on PMCA isoforms raised the question whether the stimulatory effect of aSN is at all specific to PMCA.

Interestingly, we found that rabbit SERCA1a also seems to be activated by the aSN monomer in the same lipid-dependent manner as PMCA (Fig 5A). From the two experiments performed, the effect was only predominant in low (60 nM) free Ca²⁺ concentration, whereas at high 1.8 μM free Ca²⁺, SERCA was activated to a lower fold. This indicates that aSN may also modulate the apparent Ca²⁺ affinity of SERCA. With continuous ER extending into presynaptic compartments (Wu et al, 2017; Singh et al, 2021), this is also of relevance to aSN function, and most likely it is also related to the previously reported activation by aSN oligomers (Betzer et al, 2018).

Next, we turned to a plant homolog of PMCA, namely the autoinhibited calcium ATPase 8 of Arabidopsis thaliana (ACA8), belonging to the same P2B subfamily of P-type ATPase (Axelsen & Palmgren, 2001), but not coexisting with aSN, which is only found in vertebrates. Structurally, ACA8 and PMCA differ in the localization of the autoinhibitory domain, which is situated on the N-terminus of ACA8 and the C-terminus in PMCAs, and ACA8 is lacking the long, disordered loop for the A-TM3 loop with the putative lipid-binding site in PMCA (Brodin et al, 1992; Pinto Fde & Adamo, 2002; Brini et al, 2010). In the presence of BE, ACA8 is not activated by aSN, but rather inhibited, and the effect is reversed by CaM. This may suggest an interplay between aSN, CaM, and acidic lipids, that results in a different effect, albeit weak, for the plant pump (Fig 5B). In the presence of soy PC, the pump is autoinhibited and can be activated by CaM, but not aSN. Furthermore, the human Na,K-ATPase (α1 isoform) was not activated by aSN (Fig 5C). These results suggest the specificity of the aSN interaction for mammalian Ca²⁺-ATPases.

A mathematical model describes the effect of aSN activation of PMCA on Ca²⁺ levels

To conceptualize the findings on lipid-mediated aSN binding to PMCA and to assess the influence on the calcium concentration in the presynaptic terminal cytosol, we developed a mathematical model for the presynaptic calcium ion regulation. It was based on a quantitative model for presynaptic calcium regulation, which integrates the classical Hodgkin-Huxley model for action potential propagation in order to simulate calcium dynamics in response to membrane potential changes (Hodgkin & Huxley, 1952; Erler et al, 2004; Keener & Sneyd, 2009). The model presented by Erler et al considers Ca²⁺ fluxes through PMCA (ATP-driven Ca²⁺ pump), the sodium-calcium exchanger (NCX), voltage-gated calcium channels (VGCC), and passive fluxes through unknown transport pathways and leakage across the membrane (Fig 6A). We adapted the model such that the PMCA-mediated Ca²⁺ flux depends on the cytosolic aSN concentration, by fitting a kinetic equation for non-essential activation as described in the Supplements (Appendix Fig S8, Equation S1) (Baici, 2015).

Using this model, we investigated the effect of different levels of aSN on the steady state Ca²⁺ concentration (Fig 6B) and the increase in cytosolic calcium in response to one or several action potentials (Fig 6C and D). The simulations showed an increase in the steady-state calcium concentration from 0.1 μM to around 0.25 μM, with the increase being most prominent at aSN concentrations below 10 μM. During an action potential, not only the height but also the duration of the calcium wave were increased. The predominant increase of the calcium wave duration (or width) was caused by an increase of the calcium clearance time (post-peak width, Fig 6D).

To investigate the influence of aSN on the calcium concentration during an action potential spike train, we analyzed the mean calcium concentration during this process. In general, these simulations resulted in an oscillating increase of the calcium concentrations as observed in vitro (Chamberland et al, 2020) (Fig 6C). Figure 6D (lower right panel) shows the maximum of the mean calcium concentration during an action potential spike train (as visualized in Fig 6C). In response to decreased aSN concentrations, the average calcium concentration increased because of the larger calcium waves at aSN concentrations below 10 μM (Fig 6D). Hence, the presence of monomeric aSN not only stabilizes the steady-state concentration of Ca²⁺, but also affects the calcium dynamics in response to one or several action potentials.

Co-immunoprecipitation assay

Brains from C57BL/6 mice (Janvier Labs) or aSN-KO (Sncatm1Rosl [C57BL/6, The Jackson Laboratory]) were homogenized in 7× w/v homogenization buffer (320 mM sucrose, 4 mM HEPES–NaOH, 2 mM EDTA, and complete protease inhibitor mix [Roche], pH 7.4) using a loose-fitting glass-Teflon homogenizer (10 up-and-down strokes, 700 rpm). Debris was removed from the homogenate by centrifugation for 10 min at 1,000 g in a Sorvall RC 5C plus centrifuge. The resulting supernatant was centrifuged for 1 h at 100,000 g. The supernatant was removed, and the pellet was resuspended in the original volume (7×) of RIPA (50 mM Tris pH 7.4, 159 mM NaCl, 1% Triton X-100, 2 mM EDTA, 0.5% sodium deoxycholate, 0.1% SDS) for 3 h, whereafter samples were spun at 20,000 g, for 30 min at 4°C. Protein concentration was measured using the bicinchoninic acid assay. aSN oligomer or monomer (2 μg/ml) was mixed with 0.5 mg/ml aSN-KO mouse brain homogenate diluted in PBS, 0.5% Triton X-100 and incubated overnight at 4°C.

aSN binding (ASY-1) or control (non-immune IgG) was performed as previously described (Betzer et al, 2015). The samples were incubated for 2 h with rotation. The Sepharose beads were isolated and washed twice with PBS, 0.5% Triton X-100, and Co-IP proteins were eluted by incubation in a non-reducing SDS loading buffer at room temperature. Proteins were resolved on 10–16% gradient SDS–PAGE under reducing conditions followed by immunoblotting for PMCA (primary antibody: 5F10 anti-pan-PMCA, Abcam, ab2825, secondary antibody: anti-mouse-HRP, Dako) and aSN (primary antibody: anti-Syn-1, BD Transduction Laboratory, 610787 secondary antibody: anti-mouse-HRP, Dako, P0260). The interaction between endogenous aSN and the endogenous PMCA was studied in the extracts from C57BL/6 mice as described above.

Primary hippocampal neuronal cultures and cytosolic Ca²⁺ measurements

Primary hippocampal neurons were cultured from newborn (P0) aSN-KO mice (Sncatm1Rosl [C57BL/6], The Jackson Laboratory). Hippocampi were dissected in ice-cold Hank's balanced salt solution, dissociated in 20 U/ml papain in Hibernate A medium (Gibco) supplemented with 1xB27 and 0.3 g/l L-glutamine for 20 min at 37°C, washed twice, and triturated in plating medium (MEM [Gibco] supplemented with 5 g/l glucose, 0.2 g/l NaHCO₃, 0.1 g/l transferrin, 0.25 g/l insulin, 0.3 g/l L-glutamine, and 10% fetal bovine calf serum [heat-inactivated]). Hippocampal neurons were seeded on Matrigel® matrix (Corning®)-coated coverslips. After 24 h, the medium was changed to growth medium (MEM supplemented with 5 g/l glucose, 0.2 g/l NaHCO₃, 0.1 g/l Transferrin, 0.075 g/l L-glutamine, 1× B-27 supplement, 2 μM cytosine arabinoside and 5% fetal bovine calf serum [heat-inactivated]).

At culture day 6 (DIV 6), the neurons were transiently transfected by lipofectamine 3000 with Bicistronic vectors coding for mCherry and aSN or mCherry under synapsin promotor according to manufacturer's instructions yielding varying expression levels in individual neurons. At culture day 8, cytosolic Ca²⁺ levels in primary neurons were determined using the Ca²⁺-sensitive fluorescent indicator Fura-2-AM (Molecular Probes/Invitrogen). Cells were loaded with Fura-2 in sterile-filtered HEPES-buffered saline (HBS: 20 mM HEPES, 150 mM NaCl, 5 mM KCl, 1 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, pH 7.4) containing 2.5 μM Fura-2 AM, 0.04% pluronic acid, F127 for 30 min at 37°C, 5% CO₂. The Fura-2-containing medium was replaced with fresh HBS without Fura-2 and incubated additionally for 30 min. After dye loading and prior microscopic analysis, the culture was moved into a recording buffer with thapsigargin to inhibit contribution from SERCA for 5 min., and an area containing 1–3 transfected neurons was found. The fluorescence was measured on an Olympus Scan^R high-content microscope using excitation wavelengths at 340 and 380 nm and emission at 510 nm. The cytosolic Ca²⁺ levels in single cells were measured by placing a region of interest (ROI) outside the nucleus. The recording was started and at the recording time of 2 min KCl was added to depolarize the neurons. The calcium response was followed over time until 12 min. The response upon K⁺-induced depolarization was quantified as the area under curve (AUC) from each measured neuron from the 2–12 min. measurement interval.

Cytosolic Ca²⁺ measurements in SH-SY5Y cells

A doxycycline controlled SH-SY5Y human cell line that stably overexpresses wt human aSN (SH-SY5Y ASYN) was used (Vekrellis et al, 2009). In the presence of doxycycline, aSN expression is suppressed. SH-SY5Y ASYN cells were cultured in the presence of 1 μM doxycycline similarly to previously described (Reimer et al, 2022).

SH-SY5Y ASYN cells were split and incubated with or without 1 μM doxycycline for 3 days, then split and seeded onto polylysine-coated coverslips still with or without 1 μM doxycycline and cultured for an additional 2 days, whereafter calcium extrusion experiments were performed using the fluorescent calcium binding dye Fluo-8 AM (ATT Bioquest).

The cells on the coverslips were loaded with 2.5 μM Fluo-8 in loading buffer (140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.8 mM CaCl₂, 1.4 mM MgCl₂, 1.2 mM NaH2PO4, 5 mM NaHCO₃, 10 mM Hepes pH 7.4, 2.5 μM Fluo-8 AM) for 30–60 min. at 37°C, 5% CO₂. Following Fluo-8 loading, coverslips were washed once in experiment buffer (140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.8 mM CaCl₂, 1.4 mM MgCl₂, 1.2 mM NaH₂PO₄, 5 mM NaHCO₃, 10 mM Hepes pH 7.4).

Coverslips were then mounted in an imaging chamber and covered with 300 μl pre-heated (37°C) recording buffer (140 mM NaCl, 11.5 mM glucose, 5.9 mM KCl, 1.8 mM CaCl₂, 1.4 mM MgCl₂, 1.2 mM NaH₂PO₄, 5 mM NaHCO₃, 10 mM Hepes pH 7.4, 4 μM thapsigargin) and placed in a 37°C pre-heated OkoLab imaging chamber on the microscope stage.

Imaging was performed on a Nikon Ti Eclipse inverted microscope equipped with an OkoLab heating chamber, Perfect Focus 3 system, a Plan Apo 60× (NA 1.40) oil objective, a Zyla sCMOS5.5 Megapixel camera (Andor), fluorescence illumination system CoolLED-pE-300^white, and the fluorescence filter set for GFP. The system was controlled by NIS Elements software from Nikon. Imaging was performed similar to previously described (Ernstsen et al, 2022). Images were acquired every second for 8 min in total. Binning of 2 was used. Following baseline imaging for 30 s, 600 μl depolarization buffer (140 mM NaCl, 11.5 mM glucose, 132 mM mM KCl, 5.7 mM CaCl₂, 1.4 mM MgCl₂, 1.2 mM NaH₂PO₄, 5 mM NaHCO₃, 10 mM Hepes pH 7.4, 4 μM thapsigargin) was added into the center of the imaging chamber, while imaging continued. Two technical replicates were performed. A square region of interest was placed in the cytoplasm of cells (2–4 cells per replicate) for all frames in the time-lapse sequences (481 frames) of the different conditions and mean fluorescence intensity was measured using ImageJ (Schindelin et al, 2012). The response to K+ induced depolarization was quantified as the Area Under Curve (AUC) for each cell.

Proximity ligation assay (PLA)

Proximity ligation assay (PLA) enables in situ detection of protein–protein interaction using single-stranded oligonucleotides conjugated to specie-specific antibodies. If the oligonucleotides are in close proximity (within 40 nm), they can be ligated to form circular DNA, which can be amplified by multiplying the binding site for fluorescently labeled complementary oligonucleotides and thus multiplying the fluorescent readout.

Primary hippocampal neurons were cultured from newborn (P0) C57BL/6 mice (Janvier Labs) as described for aSN-KO. At culture day 14 (DIV14), the neurons were fixed in 4% paraformaldehyde for 30 min at room temperature (RT) followed by a wash in PBS and 10 min permeabilization in 0.1% Triton X-100, 50 mM glycine, 3 mM CaCl₂, 2 mM MgCl₂, pH 7.4. Unspecific binding was blocked by 3% bovine serum albumin in PBS for 1 h followed by incubation with the primary antibody for 1.5 h [anti-AS (1 μg/ml, ASY-1, Lindersson et al, 2004; Kragh et al, 2009), and anti-PMCA 5F10 (1 μg/ml, Abcam, ab2825)]. The Duolink procedure was conducted according to the manufacturer's instructions with the Duolink® In Situ Red Starter Kit Mouse/Rabbit (Duolink®, Sigma-Aldrich), with secondary antibodies contained in the kit. After PLA staining, the neurons were labeled with synaptophysin (primary—antibody—guinea pig anti-synaptophysin 1, (Synaptic Systems #101004), secondary: goat anti-guinea pig, Alexa Fluor® 488 nm (Abcam, ab150185) to visualize the synapses where synuclein normally is located and DAPI for nuclei. Images were obtained using a Zeiss Observer Z1 inverted microscope equipped with ApoTome.2.

Expression and purification of α-synuclein

Recombinant human aSN wild type and variants were produced in Escherichia coli, and purified as previously described (Huang et al, 2005). Monomeric and oligomeric forms of aSN were produced and isolated as previously described (Betzer et al, 2015). Concentration of all aSN preparations was confirmed by BCA assay. The proteins were stored in −80°C in a buffer containing 20 mM Tris–HCl, pH 7.5, 150 mM KCl. Recombinant ¹⁵N-aSN for NMR experiments and aSN(Δ1-14) was prepared as previously described (Cholak et al, 2020) and aSN(1–61) as in (Skaanning et al, 2020). Prior to the experiments, proteins were subjected to size exclusion chromatography on Superdex 200 increase 10/300 column (GE Healthcare) in the buffer containing 50 mM Tris–HCl pH 7.2, 150 mM KCl. Fractions were pooled, concentrated to 1–1.5 mg/ml 5 kDa MWCO Vivaspin protein concentrator (Sartorius), flash-frozen in liquid nitrogen, and stored in −80°C for further experiments.

Size exclusion chromatography profiles and SDS–PAGE analysis are presented in Appendix Fig S2.

PMCA expression constructs and site-directed mutagenesis

Codon-optimized genes encoding for human PMCA1d and PMCA2w/a (Genscript) were cloned into pEMBL-yex4 expression plasmids by homologous recombination in S. cerevisiae as described (Drew et al, 2008).

PMCA1d and PMCA2w/a constructs were subjected to site-directed mutagenesis using QuikChange site-directed mutagenesis kit according to the manufacturer's protocol (Agilent). To obtain an inactive variant D475N (the autophosphorylation site), mutagenetic primers were designed using PrimerX online tool (https://www.bioinformatics.org/primerx). To obtain C-terminally truncated PMCA1, an internal TEV cleavage site was introduced to the original construct. Primers for the insertion of codon-optimized TEV site sequence and the large deletions in PMCA2 were designed according to described methods (Liu & Naismith, 2008). The integrity of all constructs was verified by DNA sequencing (Eurofins).

Expression of recombinant PMCAs in S. cerevisiae

All circular vectors were introduced into yeast cells using PEG/LiAc/ssDNA-mediated transformation (Gietz & Schiestl, 2007). All recombinant PMCAs were expressed in S. cerevisiae strain K616, depleted of the native calcium ATPases (MATα pmr1::HIS3 pmc1::TRP1 cnb::LEU2 ura3-1) (Cunningham & Fink, 1994). All culture media were supplemented with 10 mM CaCl₂, as it is a necessary survival condition for the strain to maintain internal stores. For expression culture, 100 ml of uracil-deficient 2% glucose SD medium was inoculated with a colony of transformed cells and allowed to grow for 24 h in 30°C in a 120 rpm shaking incubator. The preculture was used for inoculation of 9 L of -Ura SD medium with 1% glucose and 40 mg/L adenine hemisulfate, starting from OD₄₅₀ of 0.05. After 22 h, when all glucose in the medium was used and the culture reached OD₄₅₀ of ~ 5, the expression was induced by 2% galactose, and the culture was supplemented with YP medium and 40 mg/L adenine hemisulfate. Cells were harvested after 18–20 h by centrifugation, washed with water and TEKS buffer (50 mM Tris–HCl pH 7.6, 2 mM EDTA, 100 mM KCl, 0.6 M D-sorbitol), and resuspended in TESin buffer (50 mM Tris–HCl, 2 mM EDTA, 0.6 M D-sorbitol with the addition of PMSF and protease inhibitor. The typical yield was ~ 15 g of cells per liter of culture.

Purification of PMCAs

The cells were disrupted in a pulverisette grinder with an equal volume of 0.5 mm glass beads (4°C, 450 rpm, 4 cycles of 3 min millings, and 1 min pause). The material was centrifuged for 20 min at 2,000 g to remove beads and uncracked cells. The supernatant (S1) was centrifuged for 30 min at 20,000 g for further removal of cell debris. The supernatant (S2) was subjected to ultracentrifugation for 3 h at 150,000 g. Pelleted membranes were suspended in solubilization buffer (20% glycerol, 50 mM Tris–HCl pH 7.2, 150 mM KCl, BME, PMSF, protease inhibitors) and homogenized in a Dounce glass homogenizer. Membranes were flash-frozen in liquid nitrogen and stored at −80°C until further processing.

Membranes were diluted with solubilization buffer to 5 mg/ml total protein concentration (measured with Bradford assay) and solubilized in 1% DDM and 0.2% cholesteryl hemisuccinate for 1 h by stirring in 4°C, followed by ultracentrifugation at 150,000 g for 1 h to pellet unsolubilized material. The supernatant was supplemented with 4 mM CaCl₂ and stirred for 2 h in 4°C with CaM-Sepharose beads (GE Healthcare), pre-equilibrated with binding buffer (50 mM Tris–HCl pH 7.2, 150 mM KCl, 0.017% w/v DDM, 10% glycerol, 2 mM CaCl₂). The beads were packed onto a gravity flow column and washed with 10–15 column volumes of binding buffer. The protein was eluted a buffer containing 50 mM Tris–HCl pH 7.2, 150 mM KCl, 0.017% w/v DDM, 10% glycerol, 2 mM EGTA. EGTA stock solution (200 mM) was beforehand pH-adjusted with KOH to ~ 7.5. The fractions containing the protein of interest were pooled and concentrated to ~ 3 mg/ml using a 100 kDa MWCO Vivaspin protein concentrator (GE Healthcare). The concentrated protein was subjected to size exclusion chromatography on Superose 6 increase 10/300 column (GE Healthcare) in the buffer containing 50 mM Tris–HCl pH 7.2, 150 mM KCl, 0.017% w/v DDM, 10% glycerol. Fractions were pooled, concentrated to 1–1.5 mg/ml, flash-frozen in liquid nitrogen, and stored in −80°C for further experiments. Size exclusion chromatography profiles and SDS–PAGE analysis of all PMCA variants used in the study are presented the Appendix Fig S2.

To produce C-terminally truncated PMCA1, the PMCA1d construct with a tobacco etch virus (TEV) protease cleavage site inserted next to H1074 just before the autoinhibitory domain was used. The protein was cleaved with TEV protease while bound to the CaM sepharose beads. After overnight incubation with the protease at 4°C, the C-terminally truncated PMCA was collected in the flow-through fraction and run over a Ni²⁺ affinity column to bind the His-tagged protease. The flow-through was concentrated to ~ 3 mg/ml (100 kDa MWCO) and subjected to size exclusion chromatography on Superdex 200 increase 10/300 column (GE healthcare) in size exclusion buffer containing 50 mM Tris–HCl pH 7.2, 150 mM KCl, 0.017% w/v DDM, 10% glycerol. Peak fractions of PMCAs were pooled, concentrated to 1–1.5 mg/ml using a 100 kDa MWCO Vivaspin protein concentrator (Sartorius), aliquoted, and flash-frozen in liquid N₂ and stored at −80°C for further procedures. The purity of the proteins was analyzed by SDS–PAGE (Appendix Fig S2).

Pull-down assay

CaM-Sepharose beads (GE Healthcare) were pre-equilibrated with binding buffer (40 mM BisTris/HEPES pH 7.2, 100 mM KCl, 0.017% w/v DDM, 3 mM MgCl₂, 2 mM CaCl₂). PMCA was preincubated with aSN in the mixture of 80 μl total volume containing 0.6 mg/ml PMCA2w/a, 1.5 mg/ml aSN, and 2.5 mg/ml brain lipid extract (BE). Both PMCA and aSN stock solutions were in their final storage buffers, BE stock solution contained 50 mM Tris–HCl pH 7.2, 150 mM KCl, 1% DDM. Simultaneously, the preincubation mixture without lipids was also prepared. Control samples without PMCA or aSN were prepared. After 30 min. preincubation in room temperature, pre-equilibrated CaM-sepharose beads were added and the mixture was diluted to 1 ml with the binding buffer. For the samples where lipids were present, the binding buffer contained 100 μg/ml brain lipid extract. Samples were incubated with agitation in the room temperature for 60 min and washed four times with respective binding buffer. Each wash cycle consisted of 2 min. centrifugation at 900 g to pellet the beads, removal of the supernatant, and addition of 1 ml of the fresh buffer. Finally, 40 μl of the elution buffer (40 mM BisTris/HEPES pH 7.2, 100 mM KCl, 0.017% w/v DDM, 3 mM MgCl₂, 2 mM EGTA) was added to the beads and after short centrifugation, supernatant was collected, and analyzed by SDS–PAGE and western blotting. We used primary antibodies: rabbit ASY-R1 (Jensen et al, 2000), mouse 5F10 anti-pan-PMCA (Abcam, b2825); secondary antibodies: anti-rabbit-HRP (DAKO–P0217), anti-mouse-HRP (DAKO–P0260).

Expression and purification of CaM

Mammalian CaM in the pET15b vector was transformed into E. coli strain C41 and grown in LB media at 37°C. Induction was performed with IPTG at an OD₆₀₀ of 0.6. After 16 h of growth at 21°C, cells were harvested by centrifugation at 3,000 g, resuspended in lysis buffer (20 mM Tris–HCl pH 7.5, 1 mM EDTA, complete protease inhibitor), and lysed using sonication. The lysate was centrifuged for 1 h at 20,000 g. The supernatant was mixed 1:1 with wash buffer (20 mM Tris–HCl pH 7.5, 10 mM CaCl₂) and loaded onto a pre-equilibrated (wash buffer) gravity flow phenyl sepharose column. The column was washed with 5 CV wash buffer, 5 CV high salt wash buffer (20 mM Tris–HCl pH 7.5, 500 mM NaCl, 10 mM CaCl₂), and 5 CV wash buffer. CaM was eluted in elution buffer (50 mM Tris–HCl pH 7.5, 150 mM NaCl, 10 mM EGTA). Before use in activity assays, CaM buffer was exchanged to 20 mM Tris–HCl pH 7.5, 150 mM KCl using PD MiniTrap G25 desalting column (GE Healthcare), aliquoted, and stored in −80°C.

Expression and purification of ACA8

ACA8 expression and membrane isolation was performed similarly to the PMCA isoforms. ACA8 was solubilized in a buffer containing 50 mM Tris–HCl pH 7.6, 100 mM KCl, 20% glycerol, 3 mM MgCl₂, 5 mM β-mercaptoethanol (βME), 7.5 mg/ml DDM, 1 mM PMSF, and 1 μg/ml of chymostatin, pepstatin A, and leupeptin. Purification was achieved through Ni²⁺ affinity followed by size exclusion chromatography. ACA8 was bound to a Ni²⁺-affinity column in low imidazole buffer (50 mM Tris–HCl pH 7.5, 10 mM imidazole, 200 mM KCl, 10% glycerol, 3 mM MgCl₂, 5 mM βME, 0.2 mM LMNG). On the column, the ACA8 His-tag was cleaved off with thrombin and ACA8 was eluted in low imidazole buffer. ACA8 was subjected to size exclusion chromatography on a Superose 6 increase 10/300 column in a SEC buffer (50 mM Tris–HCl pH 7.5, 200 mM KCl, 3 mM MgCl₂, 10% glycerol, 0.02 mM LMNG, 5 mM βME). The protein was stored in −80°C in SEC buffer for further experiments.

Expression and purification of PMCA2wA-TM3 loop

Escherichia coli cells (BL21 DE3) were transformed with a modified pET24b plasmid containing N-terminal His₆-SUMO tagged PMCA2w_A-TM3 loop (L289-K383) using heat shock transformation. Pre-cultures (10 ml LB medium, 50 μg/ml kanamycin) were grown overnight and used to inoculate 1 L of ZYM-5052 media (Studier, 2005), supplemented with 50 μg/ml kanamycin. Cells were grown for 3 h at 37°C with shaking, then transferred to a shaking incubator at 16°C, and grown for a further 24 h. The cells were pelleted at 4,000 g and kept at −20°C for at least 24 h prior to lysis in column binding buffer (50 mM Tris pH8, 150 mM NaCl, 10 mM Imidazole), supplemented with complete protease inhibitor (Sigma), at 20 kpsi using a French pressure cell disruptor (Constant Systems, Daventry, UK). Lysate was cleared by centrifugation at 20,000 g, then passed through a 0.45 μm filter before being incubated with equilibrated Ni-Sepharose fast flow resin (GE Healthcare) for 30 min. Lysate was passed through the column via gravity flow, followed by wash buffer (5× CV) (50 mM Tris pH8, 1 M NaCl, 10 mM Imidazole) 5× CV binding buffer, and eluted with elution buffer (10 ml; 50 mM Tris pH 8, 150 mM NaCl, 250 mM Imidazole). Purified His₆-SUMO-PMCA2w was made up to 50 ml (with 50 mM Tris pH 8, 150 mM NaCl) and cleaved with ULP1 (0.1 mg) ON at 4°C. The cleaved protein was purified by collecting the flow-through from a second Ni-Sepharose column and concentrated using Amicon spin filters (Millipore). Protein was stored at −20°C.

NMR experiments

All NMR spectra were recorded at 5°C on a Bruker 800 MHz spectrometer equipped with a cryogenic probe and Z-field gradient using Bruker Topspin v4.0.7, recording ¹⁵N,¹H-HSQC spectra to address interaction and using the backbone assignments obtained in (Cholak et al, 2020). ¹⁵N-HSQC spectra were obtained in HEPES buffer (20 mM HEPES pH 7.4, 150 mM NaCl) of 50 μM ¹⁵N-aSN alone and in the presence of PMCA2w_A-TM3 loop (molar ratio 1:5) and with further addition of 25 mM CaCl₂ (molar ratios 1:4:500). Spectra of 9.5 μM ¹⁵N-aSN alone and with full-length PMCA2w/a (molar ratio 1:1) were obtained in Tris buffer (50 mM Tris pH 7.5, 150 mM KCl, 1.5% (w/v) DDM). Combined amide (N, H^N) chemical shift perturbations (CSPs) were measured by CCP-analysis version 2.5. The spectral analyses were done in CcpNMR analysis (Vranken et al, 2005).

Far-UV CD spectropolarimetry

Far UV CD spectra were recorded at 5°C on a JASCO J-815 spectropolarimeter on 12.8 μM PMCA2w_A-TM3 loop dissolved in 10 mM Na₂HPO₄, 200 mM NaF pH 7.4. The spectrum was recorded from 260 to 190 nm, averaging 10 scans (D.I.T 2 s; data pitch 0.2 nm; band width 1 nm; scan rate 50 mm/min; path length 1 mm). A spectrum of the buffer was recorded using identical settings and subtracted.

SERCA and Na, K-ATPase preparations

SERCA1a from rabbit skeletal muscle was kindly donated by Thomas Lykke-Møller Sørensen, Anne Lillevang, and Claus Olesen. The protein preparation was performed according to previously established procedures (Andersen et al, 1985; Sorensen et al, 2004). Na, K-ATPase was kindly donated by Michael Habeck and Natalya Fedosova. The protein was expressed and purified as described earlier (Habeck et al, 2017).

ATPase activity assay

The ATPase activity was tested by monitoring the release of free inorganic phosphate from ATP in a molybdenum blue assay (Baginski et al, 1967).

Purified PMCA requires relipidation to regain activity; therefore, the samples were preincubated with phospholipids before activity assay. Lipid stock solutions were prepared from powdered lipids as follows: brain lipid extract (from bovine brain Type I, Folch Fraction I, Sigma-Aldrich), brain PC (Avanti), or soy PC (Avanti) was dissolved to a concentration of 20 mg/ml in a buffer containing 50 mM Tris pH 7.5, 150 mM KCl, and 1.5% (w/v) DDM. After cycles of heating (45–50°C), sonicating, and vortexing, a translucent gel was obtained. Stocks were aliquoted and stored in −20°C. Shortly before the assay lipid solution was added to PMCA sample in PMCA:lipid w/w ratio 1:5 and mixed by gentle pipetting. After 15 min, the relipidated protein was added to a final concentration of 2.5 μg/ml to a reaction buffer containing 50 mM KNO₃, 5 mM NaN₃, 0.25 mM NaMob, 40 mM BisTris/HEPES pH 7.2, 3 mM MgSO₄, 1 mM EGTA, and CaCl₂ in concentration needed for obtaining the desired concentration of free calcium ions. The free calcium concentration was calculated using Maxchelator Ca/Mg/ATP/EGTA Calculator v1.0 using constants from NIST database (https://somapp.ucdmc.ucdavis.edu/pharmacology/bers/maxchelator). aSN and/or CaM were added to the reaction mixture in desired concentrations and corresponding protein storage buffer was added to control samples in equal volume. The reactions were started by adding 3 mM ATP, carried out at 37°C, and stopped after 7 or 11 min by adding a stop solution. Stop solution was prepared shortly before every assay by mixing ice-cold solutions A (17 μM ascorbic acid, 0.1% SDS in 0.5 M HCl) and B (0.7 μM ammonium molybdate VI tetrahydrate) in 5:1 proportion. The absorbance was measured with Victor 3 96-well plate reader (PerkinElmer) at 860 nm.

Mathematical modeling of aSN-dependent calcium regulation

The solve_ivp function from the scipy (Virtanen et al, 2020) package was used for numerical integration of the model equations to simulate the model behavior. Simulation results were visualized using the matplotlib (Hunter, 2007) library to create informative plots and figures. To optimize the parameters of the kinetic equation used for PMCA, the scipy.minimize function was employed, minimizing the discrepancy of the equation to the experimental data. With respect to the modeling work presented here, data handling tasks were performed using the pandas (McKinney, 2010; Data ref: Reback et al, 2022) and numpy (Harris et al, 2020) libraries.

A detailed method with parameter table (Appendix Table S1) can be found in the Appendix Supplementary Methods.

Transcriptomics data analysis

Data on gene expression for PMCA1-4 and aSN were used. Data on tissue expression and exon incorporation from VastDB for the genes ATP2B1, ATP2B2, ATP2B3, ATP2B4, and SNCA were imported in R. For the PMCA exons of relevance, the “percent spliced-in” (PSI) was correlated to aSN tissue expression. For tissues in specified aSN expression intervals (ranging from 0 to 190 interval size of 10), the average incorporation of the exons was determined, and the mean PSI was plotted as a function of the aSN expression interval. The code written in R for the analysis is included in the Appendix S1. Links to analyzed VastDB datasets are provided in the Expanded View Table EV1.

Protein disorder prediction

Protein disorder prediction was performed using NetSurfP 2.0 software developed by the Technical University of Denmark and available at https://services.healthtech.dtu.dk/service.php?NetSurfP-2.0.

Statistical analysis

Statistics were performed using GraphPad Prism and P values < 0.05 were considered significant. The statistical method used is described in the figure legends.