Expression of plasma membrane calcium ATPases confers Ca2+/H+ exchange in rodent synaptic vesicles

Ca2+ transport into synaptic vesicles (SVs) at the presynaptic terminals has been proposed to be an important process for regulating presynaptic [Ca2+] during stimulation as well as at rest. However, the molecular identity of the transport system remains elusive. Previous studies have demonstrated that isolated SVs exhibit two distinct Ca2+ transport systems depending on extra-vesicular (cytosolic) pH; one is mediated by a high affinity Ca2+ transporter which is active at neutral pH and the other is mediated by a low affinity Ca2+/H+ antiporter which is maximally active at alkaline pH of 8.5. In addition, synaptic vesicle glycoprotein 2 s (SV2s), a major SV component, have been proposed to contribute to Ca2+ clearance from the presynaptic cytoplasm. Here, we show that at physiological pH, the plasma membrane Ca2+ ATPases (PMCAs) are responsible for both the Ca2+/H+ exchange activity and Ca2+ uptake into SVs. The Ca2+/H+ exchange activity monitored by acidification assay exhibited high affinity for Ca2+ (Km ~ 400 nM) and characteristic divalent cation selectivity for the PMCAs. Both activities were remarkably reduced by PMCA blockers, but not by a blocker of the ATPase that transfers Ca2+ from the cytosol to the lumen of sarcoplasmic endoplasmic reticulum (SERCA) at physiological pH. Furthermore, we rule out the contribution of SV2s, putative Ca2+ transporters on SVs, since both Ca2+/H+ exchange activity and Ca2+ transport were unaffected in isolated vesicles derived from SV2-deficient brains. Finally, using a PMCA1-pHluorin construct that enabled us to monitor cellular distribution and recycling properties in living neurons, we demonstrated that PMCA1-pHluorin localized to intracellular acidic compartments and recycled at presynaptic terminals in an activity-dependent manner. Collectively, our results imply that vesicular PMCAs may play pivotal roles in both presynaptic Ca2+ homeostasis and the modulation of H+ gradient in SVs.

Synaptic vesicles (SV) are storage organelles for neurotransmitters and play a key role in synaptic transmission. Vesicular transport of classical neurotransmitters including glutamate, GABA/glycine, biogenic monoamines, acetylcholine, and adenosine triphosphate (ATP) is mediated by the respective vesicular transporters for these neurotransmitters. Vesicular transport is driven by a proton electrochemical gradient generated by the vacuolar-type H + ATPase (V-ATPase) 1 . The V-ATPase generates both a pH gradient (ΔpH) and a membrane potential (internal positive voltage, ΔΨ) across the SV membrane. Depending on the chemical properties of neurotransmitters and the intrinsic characteristics of vesicular transporters, the dependence of neurotransmitter uptake on either driving force varies among the neurotransmitters. In principle, the uptake of cationic transmitters such as monoamines and acetylcholine depends primarily on ΔpH, whereas the uptake of anionic transmitters such as glutamate and ATP depends more on ΔΨ, although pivotal roles of protons in glutamate transport have also been suggested [2][3][4] . The uptake of zwitter-ionic transmitters such as GABA and glycine depends equally on both ΔpH and ΔΨ (but see Juge et al. 5 who proposed a predominant role of ΔΨ). Therefore, a balance between ΔpH and ΔΨ across individual SVs differentially regulate transport rates and storage of neurotransmitters in SVs 1 .

Results
A Ca 2+ /H + exchanger in sVs operates at neutral pH. Evidence that SVs isolated from sheep brain cortex contain low affinity Ca 2+ /H + antiport activity relied essentially on an acidification assay in which fluorescence quenching of acridine orange due to V-ATPase-dependent acidification was reversed by addition of Ca 2+ 21-23 . This activity was characterized by its low affinity (Km > 200 µM) and operated at pH 8.5. However, chelation of Ca 2+ by EGTA would result in rapid liberation of H + from EGTA, and the ability of Ca 2+ chelation by EGTA depends critically on pH. As such, addition of Ca 2+ in the presence of EGTA during the measurement may have caused a change in pH, raising the possibility that the affinity and pH dependence may have been disturbed. Therefore, we decided to use BAPTA whose K D is similar to that of EGTA, but the ability to chelate Ca 2+ is much less affected by pH changes. First, we tested the effect of Ca 2+ in the presence of BAPTA at neutral pH of 7.2, and compared with that in the presence of EGTA (Fig. 1a). Acidification of crude SV fraction (LP2) was induced by adding 2 mM ATP in the presence of 100 mM Cl − , and 600 µM Ca 2+ was then added. As observed in previous studies at pH 8.5 21,23 , slow alkalization of SVs was also observed in the presence of EGTA (Fig. 1a, red trace in the left panel). Notably, the effect of Ca 2+ was more pronounced when the assay was performed in the presence of BAPTA (Fig. 1a, red trace in the right panel), indicating that the Ca 2+ /H + exchange activity was also operative at neutral pH. Essentially, the same results were obtained under conditions where acidification was monitored in SVs pre-acidified by VGLUT-mediated glutamate transport 3 (Fig. 1b), confirming that the activity originated from SVs rather than contaminating organelles such as microsomes and mitochondria. To further confirm if the de-quenching represented a transporter-mediated process, we measured de-quenching velocities upon the addition of Ca 2+ at various temperatures (32.3 °C and 37.3 °C) under the same conditions as those in Fig. 1b (right  panel). The temperature co-efficient (Q 10 ) was estimated to be ~1.6 ( Fig. 1c), which is within the range of physical properties of transporter-mediated processes rather than simple diffusion 32,33 . the vesicular Ca 2+ /H + exchanger exhibits high affinity for Ca 2+ . To estimate the affinity of the Ca 2+ /H + exchanger in SVs at neutral pH, we measured alkalization of pre-acidified SVs by various concentrations of CaCl 2 in the presence of a fixed concentration of BAPTA (50 µM). To this end, we first estimated free Ca 2+ concentrations www.nature.com/scientificreports www.nature.com/scientificreports/ in our assay buffers using fura-2 as a Ca 2+ indicator and found it to be ~1 µM ( Supplementary Fig. 1). Taking this into account, free Ca 2+ concentrations in the assay buffer were set by adding CaCl 2 (see Materials and Methods for details). After complete acidification of SVs in the presence of 100 mM KCl was achieved, we added a series of Ca 2+ concentrations ranging from 0 to 75 µM that were equivalent to 3.63 to 6410 nM free [Ca 2+ ] in the assay medium. With the ranges of Ca 2+ added, the rate of alkalization increased as a function of free Ca 2+ concentrations (Fig. 2a). A plot of initial velocities of alkalization as a function of free Ca 2+ concentrations revealed that K m of Ca 2+ -induced SV alkalization was ~400 nM (Fig. 2b), a value that appeared to be several magnitudes lower than that reported previously 21 . To further characterize the Ca 2+ /H + antiport activity at neutral pH, we tested a series of divalent cations for the ability to alkalize pre-acidified SVs. Previous studies demonstrated that, at alkaline pH, Zn 2+ and Cd 2+ are substrates of the same Ca 2+ /H + exchanger, whilst Ba 2+ and Sr 2+ are not 29 . In our assay conditions in the presence of BAPTA at pH 7.2, Zn 2+ and Cd 2+ induced alkalization of pre-acidified SVs in the presence of 100 mM KCl to a similar extent as that of Ca 2+ , whereas Ba 2+ induced slower but significant alkalization (Fig. 3). Consistent with previous results, Sr 2+ showed marginal alkalization, indicating that the Ca 2+ /H + exchanger in SVs shows substrate preference as follows: Ca 2+ ≈ Zn 2+ ≈ Cd 2+ > Ba 2+ ≫ Sr 2+ . The same trend was observed when LP2 was pre-acidified by glutamate, indicating that the alkalization by divalent cations originated from VGLUT-containing SVs ( Supplementary Fig. 2). It should be noted here that it remains unknown if these divalent cations are transported by distinct transporters with similar proton coupling, or if the alkalization resulted from the blockade of the V-ATPase activity (see also Supplementary Fig. 4    were analyzed by western blotting using isoform-specific antibodies. Antibodies used for western blots are indicated at the left of the images. Rabbit polyclonal antibodies against SV2A, SV2B, and SV2C, and a mouse monoclonal antibody that recognizes all SV2 isoforms (pan-SV2) were used. For loading controls, a mouse monoclonal antibody against synaptophysin (Syp) (Cl7.2) was used. Note that expression of the SV2 isoforms was completely abolished in the respective DKO samples. A faint band revealed by the pan-SV2 antibody in the SV2A/2B-DKO sample indicated that SV2C content was much less than that of SV2A or SV2B. The images were cropped from four independent blots for presentation, and the original digital images of the full-length blots are presented in Supplementary Fig. 3. (b) Ca 2+ -induced alkalization in vesicles derived from SV2A/B-DKO (red) compared to wild-type mice (black). 50 µM CaCl 2 was added after vesicles were pre-acidified in the presence of 100 mM KCl. (c) Effect of levetiracetam (LEV, 30 µM) on Ca 2+ -induced alkalization in vesicles from wild-type mice. Vesicles were pre-acidified in the presence of 100 mM KCl, and 50 µM CaCl 2 was then added. LEV pre-www.nature.com/scientificreports www.nature.com/scientificreports/ th elumen of sarcoplasmic endoplasmic reticulum (SERCA) 34 . However, contributions of these ATPases to SV acidity due to a H + coupling have not been directly investigated. Furthermore, according to synaptic phenotypes related to changes in presynaptic Ca 2+ concentrations and in presynaptic calcium buffering in SV2-deficient neurons 26,27 , it has been hypothesized that SV2s may be involved in Ca 2+ transport into SVs. We therefore sought to clarify which protein was responsible for the observed Ca 2+ /H + exchange activity in SVs at neutral pH.
First, in order to verify the contribution of SV2s in Ca 2+ /H + exchange, we generated SV2A/2B-and SV2B/2C-double knockout (DKO) mice using the CRISPR/Cas9 system. We then tested if Ca 2+ /H + exchange activity was affected. Western blot analysis of LP2 fractions purified from SV2A/2B-DKO or SV2B/2C-DKO mice using isoform-specific antibodies confirmed efficient gene knockout by the CRISPR/Cas9 system (Fig. 4a). Moreover, western blot analysis using a pan-SV2 antibody that equally recognizes all SV2 isoforms revealed that SV2A and SV2B represented the major SV2 isoforms in brain, and expression of SV2C was minor (Fig. 4a). As a control, we confirmed that expression of synaptophysin, an abundant SV protein 34 , was unaltered in SVs derived from SV2-deficient mice. As shown in Fig. 4b, Ca 2+ /H + exchange activity was normal in SVs from SV2A/2B-DKO mice, and the extent of alkalization induced by 50 µM CaCl 2 was comparable to that observed in SVs from wild-type mice of the same age (Fig. 4b).
Levetiracetam (LEV, also known as Keppra ® or E Keppra ® ) is an established second-generation antiepileptic drug, whose molecular target is SV2A 35 . LEV selectively binds to the cytoplasmic portion of SV2A 35,36 and potentially inhibits SV2A function 36 . The inability of LEV to inhibit Ca 2+ /H + exchange (Fig. 4c) in SVs from wild-type mice provided further support that SV2A was not responsible for the activity, although SV2B, another isoform of SV2s, was present in the SV preparation. To completely rule out the contribution of SV2A to Ca 2+ /H + exchange activity, we examined the effect of LEV on transport activity in SV2B/2C-DKO samples (note that only SV2A is present in SV2B/2C-DKO mice). Again, LEV did not reduce Ca 2+ -induced alkalization (Fig. 4d). Taken together, although SV2s have been proposed to function as Ca 2+ transporters in SVs, SV2s do not contribute to high affinity Ca 2+ /H + exchange activity at neutral pH. the plasma membrane Ca 2+ Atpases (pMCAs) mediate Ca 2+ /H + exchange in sVs. Next, we examined the effect of blockers that selectively inhibit the P-type Ca 2+ ATPases, including PMCAs and SERCA. Among the available inhibitors, 500 µM vanadate, a PMCA blocker, potently inhibited Ca 2+ -induced alkalization of SVs (Fig. 5a). On the other hand, 15 µM cyclopiazonic acid (CPA), a SERCA inhibitor 37 , did not affect alkalization by Ca 2+ (Fig. 5b), indicating that the high affinity Ca 2+ /H + exchange activity observed with this assay was attributed to PMCAs. Notably, whereas Ba 2+ -and Sr 2+ -induced alkalization of pre-acidified SVs by glutamate was largely abolished in the presence of vanadate, alkalization induced by Zn 2+ or Cd 2+ was not completely blocked by vanadate ( Supplementary Fig. 4). These observations implied that, while Ba 2+ and Sr 2+ are recognized by PMCAs, there are other divalent cation/H + exchangers that transfer Zn 2+ or Cd 2+ . Of note, ZnT-3, a vesicular Zn 2+ transporter, is present on a subpopulation of SVs that contain glutamate [38][39][40] , and therefore could contribute to vanadate-insensitive Zn 2+ /H + exchange activity observed here, although the precise mechanism of Zn 2+ transport into SVs, e.g. proton coupling during transport cycles, remains to be determined.
It was previously reported that in sheep SVs, vanadate-sensitive Ca 2+ transport activity was largely absent in the presence of >200 µM Ca 2+ 22 . However, Ca 2+ -induced alkalization at neutral pH was evident by 500 µM Ca 2+ under our assay conditions (Fig. 5c, black trace). More importantly, 500 µM vanadate effectively impeded the alkalization (Fig. 5c, red trace), indicating that even in the presence of higher concentrations of Ca 2+ , Ca 2+ -induced alkalization was mediated by PMCAs, irrespective of the Ca 2+ concentrations within this range.
The PMCAs, but not SV2s, are responsible for the majority of Ca 2+ transport into sVs at neutral pH. To verify if Ca 2+ -induced alkalization observed in our assay conditions was correlated with Ca 2+ transport by PMCAs, we next examined radioactive 45 Ca 2+ uptake in the presence of various inhibitors. For this purpose, LP2 fraction was further purified by applying it to sucrose gradient centrifugation to minimize the contamination of other organelles and membranes. Consistent with the previous report 22 , 45 Ca 2+ uptake was suppressed by PMCA inhibitors (10 µM eosin and 500 µM vanadate) to background levels measured in the absence of ATP (Fig. 6a). In contrast, the V-ATPase inhibitor, bafilomycin (500 nM), did not affect 45 Ca 2+ uptake (Fig. 6b). Collectively, these results indicate that PMCAs are responsible for the majority of Ca 2+ transport into SVs. In agreement with the results from acidification assays using SV2-KO vesicles, 45 Ca 2+ uptake into SVs derived from SV2A/ SV2B-DKO mice (which lack the majority of SV2 proteins) was not significantly different from that derived from wild-type mice (Fig. 6c). Likewise, addition of 30 µM LEV did not reduce Ca 2+ transport into SVs derived from SV2B/2C-DKO mice (Fig. 6d), excluding the possibility that SV2s play a role in Ca 2+ transport into SVs.

PMCA1-pHluorin localizes in acidic compartments at presynaptic terminals. The biochemical data
described have demonstrated that PMCAs were responsible for the majority of Ca 2+ transport into SV-rich membrane fractions, but it remains unclear whether they localized in functionally competent SVs in living neurons. In fact, isolated SVs from rodent brains contain various endosomal proteins such as endosomal SNAREs and rab proteins that may preferentially localize non-recycling vesicles at presynaptic terminals 34 . To clarify whether treatment (red traces) shows little impact on alkalization by Ca 2+ compared to the respective controls without LEV pre-treatment (black traces). (d) Effect of levetiracetam (LEV, 30 µM) on Ca 2+ -induced alkalization in vesicles from SV2B/2C-DKO mice. Measurements were performed as in (c). LEV pre-treatment (red traces) shows little impact on alkalization by Ca 2+ compared to the respective controls without LEV pre-treatment (black traces). Traces in (b-d) are the representative data from two to three measurements in each condition.
www.nature.com/scientificreports www.nature.com/scientificreports/ PMCAs localized at recycling or non-recycling vesicles in living neurons, we constructed a fluorescent reporter in which a pH-sensitive green fluorescent protein (super-ecliptic pHluorin; SEP 30 ) was conjugated to the first luminal loop of PMCA1 (Fig. 7a). A SEP fused to the second luminal loop of synaptophysin (SypHy) and a SEP fused to the C-terminal end of Syntaxin1a (Syntaxin1a-SEP) were used as controls for SV residents and plasma membrane residents, respectively (Fig. 7a). Due to the pH sensitivity of the reporter, fluorescence is quenched in acidic compartments whereas it rises when the reporter is exposed to the neutral pH solution in the extracellular space 30 . When PMCA1-SEP was transiently expressed in cultured hippocampal neurons derived from embryonic mouse brains, immunostaining of the fixed cells with an anti-GFP antibody revealed ubiquitous expression of PMCA1-SEP in cell bodies and dendrites (Fig. 7b, left), as well as in presynaptic bouton-like structures (Fig. 7b, right). Co-immunostaining with an antibody against synaptophysin further confirmed the presynaptic localization of PMCA1-SEP (Fig. 7b, right panels). Estimation of inner/surface distribution of PMCA1-SEP by sequential applications of an acidic solution (pH 5.5) and a 50 mM NH 4 Cl solution 41 indicated that ~75% of PMCA1-SEP was localized in the acidic compartments whose average pH was 6.47 ± 0.03 (Fig. 7d,e, Supplementary Fig. 5). Control experiments with SypHy revealed that ~80% of SypHy was localized in the acidic compartments whose average luminal pH was ~5.91 ± 0.02. In contrast, control experiments with Syntaxin1a-SEP revealed that it was localized almost exclusively on the plasma membrane (99.5 ± 2.2%) (Fig. 7c-e, Supplementary Fig. 5). The higher luminal pH of PMCA1-pHluorin-positive vesicles indicates that a portion of PMCA1-pHluorin may localize in endosomal compartments. Alternatively, exogenous expression of PMCA1-pHluorin may exert additional Ca 2+ /H + antiport, which would facilitate alkalization of the vesicular lumen. Collectively, although PMCA1-SEP expression was not restricted to the presynaptic terminals in cultured hippocampal neurons, a fraction of PMCA1-SEP did seem to be localized in the acidic compartments whose luminal pH was similar to that of typical SVs.

PMCA1-SEP recycles at presynaptic terminals.
To examine whether PMCA1-SEP-positive vesicles would recycle at presynaptic terminals in an activity-dependent manner, cultured hippocampal neurons Measurements were performed as in (a) except that a high concentration of CaCl 2 (500 µM) was added. All traces are the representative data from at least two to three measurements in each condition. Since vanadate slightly reduced AO quenching signals in the presence of 5 mM glutamate and 3 mM KCl, fluorescence traces were normalized by amplitudes of acidification, and the starting points of alkalization by CaCl 2 were adjusted for comparison.
www.nature.com/scientificreports www.nature.com/scientificreports/ expressing PMCA1-SEP were exposed to repetitive electrical stimulation (Fig. 8). At the end of recordings, 50 mM NH 4 Cl solution was applied to estimate the total fluorescence of the SEP molecules (Fig. 8a). Notably, PMCA1-SEP fluorescence increased upon repetitive stimulation at 20 Hz for 10 sec in a manner similar to that of SypHy (Fig. 8a), although the fraction of PMCA1-SEP molecules engaged in exocytosis during stimulation was significantly less than that of SypHy (Fig. 8b). In accordance with the slightly higher vesicular pH of the PMCA1-carrying vesicles (Fig. 7e), this indicated that compared to synaptophysin, more PMCA1 was localized in non-recycling acidic compartments such as endosomes. The kinetics of the rise phase (mainly reflecting exocytosis) and decay phase (reflecting endocytosis and subsequent re-acidification of endocytosed vesicles) were identical to those of SypHy (Fig. 8c-e). Furthermore, when neurons were treated with 10 nM tetanus toxin (TeNT), which cleaves the major vesicular SNARE (v-SNARE) synaptobrevin/VAMP2 42 , PMCA1-SEP responses were completely abolished (Fig. 8f, Supplementary Fig. 6). This effect was also observed for SypHy (Fig. 8f,  Supplementary Fig. 6), indicating that PMCA1 was present on synaptobrevin/VAMP2-positive vesicles and followed the same fate as that of synaptobrevin/VAMP2 during activity-dependent SV recycling. Figure 6. PMCAs predominantly contribute to Ca 2+ uptake into SVs independent of H + electrochemical gradient generated by V-ATPase. (a) Ca 2+ transport was measured using 45 Ca 2+ as a tracer. The assay was performed in the absence of inhibitors (Control) or in the presence of vanadate (Van, 500 µM) or eosin (10 µM). Both PMCA blockers effectively inhibited Ca 2+ transport into the SV fraction. Error bars indicate s.e.m. of three independent measurements. *p < 0.0025, unpaired equal variance Student's t-test. (b) Ca 2+ transport was measured as in (a) in the absence (Control) and presence of the V-ATPase inhibitor, bafilomycin A (500 nM) (Baf

Discussion
In this study, we characterized the biochemical properties of Ca 2+ transport into SVs isolated from rodent brains. We found that PMCAs, the plasma membrane Ca 2+ ATPases, confer the majority of Ca 2+ transport into SVs. Previous studies have indicated that there are two Ca 2+ transport systems in isolated SVs; one is mediated by PMCAs at neutral pH, and the other is mediated by an independent Ca 2+ /H + exchanger at alkaline pH. Our results strongly support a united model indicating that PMCAs are solely responsible for both activities in the majority of the SV population at neutral pH. Novel findings extending beyond those of previous studies [21][22][23]29 are as follows. First, even at neutral pH of the extra-vesicular space, Ca 2+ alkalized pre-acidified SV lumen. This phenomenon became evident when BAPTA was used to control free Ca 2+ concentrations, although we currently do not understand why the use of EGTA attenuated alkalization signals in the acidification assays. Since the SV alkalization by Ca 2+ was restored by vanadate which concomitantly impeded 45 Ca uptake into SVs, we conclude that PMCAs are responsible for Ca 2+ /H + exchange activity. This is consistent with more recent findings that counter transport of H + is associated with Ca 2+ transport by PMCAs 43 . Second, the Ca 2+ /H + exchange at neutral pH exhibited high affinity for Ca 2+ (K m ~ 400 nM) that was similar to the reported values for PMCAs 44 , whereas previously reported Ca 2+ /H + exchange activity at alkaline pH was characterized as a low affinity Ca 2+ /H + exchanger (K 0.5 ~ 217 µM) 22 . Third, although PMCAs have been characterized as residing in the plasma membrane, our imaging analysis revealed that PMCAs were potentially capable of recycling in an activity-dependent manner at presynaptic terminals. This observation was further substantiated by the fact that recycling of PMCA1-SEP depended exclusively on the presence of synaptobrevin/VAMP2 on the same vesicles, which is essential for stimulus-dependent exocytosis of SVs 45 .
Our results are highly consistent with biochemical evidence that, in addition to V-type H + ATPases, vanadate-sensitive P-type ATPases constitute the major ATPases in cholinergic SVs from Torpedo electric organs 46 as well as clathrin-coated vesicles from bovine brains 47 . Although PMCAs, which are P-type ATPases, have been characterized as plasma membrane residents, our results demonstrate that PMCAs constitute the major ATPase on recycling SVs as well. This is also consistent with proteomic analyses that identified PMCAs within SV fractions 34,48 . Thus, besides the main role of PMCAs in excluding Ca 2+ from the presynaptic cytosol to the extracellular space at the plasma membrane, vesicular PMCAs may also contribute to Ca 2+ clearance from the presynaptic cytosol. Furthermore, since Ca 2+ transport into SVs is coupled to H + efflux, vesicular PMCAs may be involved in the regulation of ΔµH + of SVs, which drives neurotransmitter uptake into SVs. Indeed, external (cytosolic) Ca 2+ strongly inhibited dopamine uptake into isolated SV fraction 49 , which is reasonable given that ΔpH, the driving force for dopamine uptake, is attenuated by Ca 2+ /H + exchange. In contrast, glutamate uptake, which is predominantly driven by ΔΨ, was only marginally facilitated under the condition where ΔΨ predominated in the presence of 17 mM Cl −49 . This subtle effect by Ca 2+ on glutamate transport indicates that PMCAs on SVs are electro-neutral pumps (Ca 2+ :H + = 1:2), which is consistent with observations in snail neurons 43 . Alternatively, the fact that glutamate transport in the presence of Ca 2+ retained a typical biphasic Cl − dependency 49 (note that increasing Cl − concentrations build up greater ΔpH) may indicate an important contribution of protons (or vesicular pH) on VGLUT function [2][3][4] . Further investigations will be needed to clarify how Ca 2+ / H + exchangers in conjunction with vesicular Cl − carriers, either VGLUTs or the ClC-type Cl − transporters, affect glutamate transport into SVs.
The plasma membrane PMCAs are responsible for transient acidification of the cytoplasm during sustained stimulation at mouse motor nerve terminals, which may regulate endocytosis 31 . Our imaging analysis indicated that a substantial portion of vesicular PMCAs are inserted into the plasma membrane by exocytosis and retained there until endocytosis is completed, indicating that the total PMCA expression on the plasma membrane increases, especially during high activity. Although the contribution of the vesicular PMCAs that are inserted into the plasma membrane to acidification of the cytoplasm seems to be negligible 31 , it remains to be determined to what extent Ca 2+ extrusion from the cytoplasm is accelerated by the vesicular PMCAs that translocate into the plasma membrane during high neural activity.
In mammals, four different genes encode the four isoforms of PMCA 50 . PMCA1 and PMCA4 are expressed ubiquitously, whereas PMCA2 and PMCA3 are expressed predominantly in the central nervous system. Furthermore, each isoform has multiple splice variants, comprising more than 30 spliced isoforms, although little is known about the functional significance of multiple splice variants 50 . In this study, we show that one of the PMCAs, the full length PMCA1 (1,249 a.a.) fused to pHluorin at its luminal region, is sorted preferentially to recycling SVs rather than to the plasma membrane at presynaptic terminals, and its distribution and recycling behaviors are very similar to those of the SV marker, synaptophysin (Figs 7 and 8). It remains unknown whether other PMCA isoforms or respective splice variants show similar properties to those of the full length PMCA1-pHluorin at presynaptic terminals, which may necessitate more complex regulation of Ca 2+ and H + dynamics at presynaptic terminals. Additionally, it should be kept in mind that we still lack concrete evidence for the existence of PMCA proteins on SVs, as the evidence we provide in this study relies on the usage of pharmacological blockade of Ca 2+ transport by PMCA blockers and also on the exogenous expression of PMCA1 fused to fluorescent proteins. Although the SV proteome supports the existence of PMCA in the SV fraction isolated from native brains 32 , previous fractionation experiments, combined with western blot analysis using isoform-specific PMCA antibodies, have provided controversial results concerning the localization of PMCA isoforms in synaptic PMCA1-SEP, and Syntaxin1a-SEP between cell surface (black bars) and acidic intracellular compartment (red bars) deduced from the traces in (c  www.nature.com/scientificreports www.nature.com/scientificreports/ vesicles 49,50 , probably due to their predominant expression at the plasma membrane of the cell body. Thus, direct demonstration of PMCA isoforms on synaptic vesicles, e.g. via immuno-gold labeling of isolated SV fractions or immuno-gold labeling of brain sections, will be essential to confirm their vesicular localization in the future. SV2s are postulated to function as a Ca 2+ transporter, since synaptic phenotypes observed in SV2-deficient mice could be well explained if SV2s function as Ca 2+ transporters 26,27 . However, our results from direct tests of Ca 2+ transport into SV2-deficient vesicles rule out this hypothesis (Figs 4 and 6). Therefore, the Ca 2+ -related phenotypes observed previously in SV2-deficient mice may be indirect consequences of unknown functions of SV2. Of note, it has recently been shown that SV2A mediates galactose/H + symport when heterologously expressed in yeast cells 51 . Although the functional significance of galactose in the nervous system, particularly its role in SVs, has been enigmatic, changes in either metabolism of carbohydrates or composition of glycans attached to proteins or lipids may indirectly regulate Ca 2+ homeostasis at presynaptic terminals, which would lead to the observed Ca 2+ -related phenotypes in SV2-deficient synapses. Uncovering the link between SV2s' function and presynaptic [Ca 2+ ] regulation will help to elucidate the role of SV2s in epileptogenesis caused by SV2A gene knockout 26,52 .
In summary, our present results collectively suggest that Ca 2+ transport across SV membranes is predominantly, if not exclusively, mediated by PMCAs. Due to their property as Ca 2+ /H + exchangers, PMCAs may contribute to the regulation of ΔµH + and the dynamic control of cytosolic Ca 2+ and H + at presynaptic terminals. Since various single nucleotide polymorphisms (SNPs) have been identified in PMCA genes that are associated with neuronal disorders such as autism and deafness 53 , it will be crucial to establish how these SNPs affect the function and distribution of PMCAs in neurons to elucidate the mechanisms underlying these diseases.

Materials and Methods
synaptic vesicle isolation. Crude synaptic vesicle fraction (LP2) was isolated from mouse whole brains with slight modifications 54 . Briefly, whole brains from C57BL/6 mice were homogenized in homogenization buffer (320 mM sucrose, 4 mM MgSO 4 , 4 mM HEPES-NaOH, pH 7.3). The homogenate was centrifuged fro 10 min at 1,000 × g, the resulting supernatant (S1) was centrifuged for 15 min at 12,000 × g. The resulting pellet (P2) was washed with homogenization buffer and re-centrifuged for 15 min at 13,000 × g to obtain crude synaptosomes (P2′). To release SVs from the synaptosomes, P2′ fraction was subjected to an osmotic shock by the addition of 9 volume of ice-cold water and the subsequent homogenization. The resulting suspension was centrifuged for 20 min at 33,000 × g. After centrifugation, the supernatant (LS1; lysate supernatant) was centrifuged for 2 hours at 260,000 × g. The final membrane pellet (LP2) was resuspended in standard acidification buffer (300 mM sucrose, 4 mM MgSO 4 , 10 or 20 mM MOPS-KOH, pH 7.2), and stored at −80 °C until use. Essentially, all fluorometric assays, unless indicated otherwise, were performed with this fraction. For determination of apparent affinity of Ca 2+ /H + exchange activity and radioactive 45 Ca 2+ uptake, LP2 obtained from Wistar rat brains was further purified by sucrose gradient centrifugation (50-800 mM continuous gradient). After centrifugation at 65,000 × g max for 4 hours, turbid materials visible in the middle of the gradient (in the range of 200 to 400 mM sucrose) were pooled and sedimented by centrifugation at 260,000 × g max for 90 min. The resulting pellet (SV) was resuspended in acidification buffer and stored at −80 °C until use. To set up the equations above, the overall apparent association constants, K′s (K′ CaBAPTA , K′ MgBAPTA , K′ MgATP and K′ CaATP ) were converted from the absolute association constants, Ks, which were determined for standard conditions (see details in Marks and Maxfield 55 ). Supplementary Table 1 lists the published association constants for BAPTA and ATP, as well as the ΔH values for BAPTA and ATP. ΔH values were necessary for the conversions from the absolute association constants to the overall apparent association constants. Acidification assay. Acidification measurements were performed according to previous publications using acridine orange (AO, Molecular Probes) as a ΔpH reporter 3 . Changes in AO fluorescence (excitation at 492 nm and emission at 530 nm with slit lenghs with 2.5 nm, HMT 700 V) were monitored in a Hitachi F2500 fluorometer www.nature.com/scientificreports www.nature.com/scientificreports/ (Hitachi, Japan) at 32 °C, unless otherwise stated 3 . Typically, 20 µg of LP2 or SV fraction was preincubated in 1 mL of assay buffer (300 mM sucrose, 4 mM MgSO 4 , 1.5 µM AO, 10 or 20 mM MOPS, pH 7.2) with varying composition of 5 mM K-glutamate, 3 mM or 100 mM KCl, 50 µM EGTA, and 50 µM BAPTA as indicated in the figures or figure legends. After a stable baseline was achieved (usually within 10 min), 2 mM ATP was added to start acidification. Various concentrations of CaCl 2 or 50 µM other divalent cations were added at 10 min where indicated. At the end of recordings, a V-ATPase inhibitor, bafilomycin A 1 (500 nM) was added to ensure that quenching of AO was due to proton translocation by the V-ATPase. For Figs 4 and 5, 15 µM cyclopiazonic acid, 500 µM vanadate, or 30 µM levetiracetam was pre-incubated for 5 min before measurements. Representative traces from multiple measurements are shown in the figures. For estimation of temperature co-efficient (Q 10 ) for the Ca 2+ -dependent AO de-quenching, acidification assays were performed at two different temperatures. The Q 10 was calculated from an equation:

Calculation of free Ca
( 10 T T ) 2 1 in which τ 1 and τ 2 are time constants of recovery phases of acridine orange fluorescence after the addition of Ca 2+ at temperature T 1 and T 2 . τ 1 and τ 2 were obtained by a first order exponential fitting using a Solver function in Excel software. T 1 and T 2 (T 2 > T 1 ) are 32.3 and 37.3 °C, respectively.
Animals. C57BL/6NJcl mice were purchased from CLEA, Japan. ICR mice were purchased from SLC, Japan.
All mice were given food and water ad libitum. Animals were kept in an SPF facility with a 12-hour light and 12-hour dark cycle. The ambient temperature was maintained around 21 °C with a relative humidity of 50%. ICR mice (12 to 20 weeks old) were used as recipients. A combination anesthetic (0.75 mg/kg of medetomidine, 4.0 mg/kg of midazolam, and 5.0 mg/kg of butorphanol) was used for surgery. The anesthetics were administered to recipient mice by intraperitoneal injection. All animal experiments were approved by the Institutional Animal Care and Use Committee of the RIKEN Kobe branch (approval number: QA2013-04-6) and the Institutional Animal Care and Use Committee of Doshisha University.
One-cell embryo microinjection. C57BL/6N females (4-6 weeks old) were superovulated and mated with C57BL/6N males. Fertilized eggs were collected from the ampulla of the oviduct of plugged C57BL/6N females by micro-dissection and kept in KSOM medium (Merck Millipore) in a 5% CO 2 incubator at 37 °C. Cas9 mRNA (100 ng/µL) and six gRNAs (50 ng/µL each, 300 ng/µL total) were co-injected into the cytoplasm of fertilized eggs in M2 medium (Merck Millipore) at room temperature. Details of the cytoplasmic injection procedure have been described previously 56 . After microinjection, the injected embryos were cultured for 1 hr in KSOM medium (Merck Millipore) in a 5% CO 2 incubator at 37 °C, then 15-30 embryos were transferred to the oviducts of recipient ICR female mice.
Ca 2+ transport assay. The concentrations of SV or LP2 fractions were adjusted to 0.2 mg/mL in sucrose buffer. 100 µL aliquots were preincubated for 10 min at 32 °C. The reaction was started by addition of 10 µM (final concentration) 45 Ca 2+ dissolved in sucrose buffer containing ATP (2 mM), vanadate (500 µM), eosinY (10 µM), cyclopiazonic acid (15 µM) and levetiracetam (30 µM) when indicated. All incubations contained 0.125% (v/v) DMSO and 0.167% (v/v) ethanol in a final volume of 150 µL. After incubation for 10 min at 32 °C, the reaction was stopped by addition of 3 mL of ice-cold buffer, followed by filtration through nitrocellulose filters. The incubation tubes were washed out with 3 mL of ice-cold buffer, and the wash out buffer was poured through the filter. The filters were washed out twice with 2.5 mL of ice-cold buffer. Radioactivity retained on filters was measured by liquid scintillation counting using an ALOKA LSC-6100 liquid scintillation counter (ALOKA, Japan). www.nature.com/scientificreports www.nature.com/scientificreports/ antibodies for SV2s (Synaptic Systems, Germany) or with a pan-SV2 monoclonal antibody (a kind gift from Reinhard Jahn, Göttingen, Germany). For detection, the appropriate secondary antibodies conjugated to horseradish peroxidase were used. After washing steps, the horseradish peroxidase was detected by enhanced chemiluminescence using a commercially available kit (Perkin Elmer, Inc., MA). As loading controls, anti-synaptophysin monoclonal antibody (Cl7.2, 'a gift from Reinhard Jahn' instead of 'Synaptic Systems, Germany') was used.

Molecular biology.
To construct PMCA1-SEP, a full length mouse PMCA1 (accession no. NM_001359506.1) was amplified by PCR from mouse adult brain cDNA generated using SuperScript RT-PCR system (Invitrogen) and subcloned into a StuI site of pCR-Blunt vector (Thermo Fisher Scientific). A DNA fragment encoding the N-terminal region of PMCA1 (a.a. 1-139) with linker sequence (STSGGSGGTGGS) and a fragment of super-ecliptic pHluorin (SEP) amplified from SypHy plasmid 58,59 (a kind gift from Leon Lagnado) were amplified by PCR and cloned into an EcoRV site of pcDNA3.1(+) using In-Fusion Cloning Kit (Clontech). A DNA fragment encoding the C-terminal region of PMCA1 (a.a. 140-1,220) was PCR-amplified and cloned into a NotI site of pcDNA3.1(+) which contained the N-terminal region of PMCA1 and SEP using In-Fusion Cloning Kit. To construct Syntaxin1a-SEP, a full length mouse Syntaxin-1a (accession no. NM_016801.3) without the stop codon and a fragment of SEP were amplified by PCR and simultaneously cloned into a NheI/EcoRV site of pcDNA3.1(+). A DNA fragment encoding SypHy 59 was amplified by PCR and subcloned into a NheI/XbaI site of pcDNA3.1(+).
Neuronal cultures. Primary hippocampal cultures were prepared from embryonic day 16 ICR mice as described previously 60 with slight modifications. Briefly, hippocampi were dissected and were incubated with papain (90 units/mL, Worthington) for 20 min at 37 °C. After digestion, hippocampal cells were plated onto poly-D-lysine-coated coverslips in 24-or 12-well plates (Falcon) at a density of 20,000 cells/cm 2 and kept in a 5% CO 2 humidified incubator. At 2-4 days in vitro (DIV), 40 μM FUDR (Sigma) and 100 μM uridine (Sigma) were added to inhibit the growth of glial cells. One-fifth of the culture medium was replaced with fresh medium every 2-4 days. Cultures were transfected with plasmids encoding either PMCA1-SEP, SypHy, or Syntaxin-1a-SEP at 5-7 DIV using CalPhos TM mammalian transfection kit (Clontech) in accordance with a calcium phosphate transfection method which is optimized for neuronal cultures 61 , and were subjected to experiments at 12-14 DIV. Animals for the primary neuron cultures were treated according to our institutional guidelines for the care and use of animals (Doshisha University).
Live imaging. Cells cultured on a glass coverslip were placed in a custom-made imaging chamber on a movable stage and continuously perfused with standard extracellular solution containing (in mM): 140 NaCl, 2.4 KCl, 10 HEPES, 10 glucose, 2 CaCl 2 , 1 MgCl 2 , 0.02 CNQX, and 0.025 D-APV (pH 7.4). A solution containing 50 mM NH 4 Cl (pH 7.4) was applied directly onto the area of interest with a combination of a fast flow exchange microperfusion device and a bulb controller, both of which were controlled by Clampex 10.2. To estimate luminal pH and surface fraction of the SEP probes, a MES-buffered solution at pH 5.5 and 50 mM NH 4 Cl (pH 7.4) were successively applied to cultured neurons as described previously 35 . Briefly, SEP fluorescence during acid quenching (F Q ) and during subsequent NH 4 Cl application (F NH4Cl ) was measured. The observed fluorescence in a given terminals is thought to be the sum of the fluorescence derived from the surface fraction of the probes (S) that experiences extracellular pH and from the vesicular fraction (1 − S) that is exposed to luminal pH (pHv). The fluorescence during the baseline (F 0 ) and the fluorescence during F Q are expressed as where F pH 5.5 and F pHv are the total fluorescence values predicted when all probe molecules in a terminal are exposed to pH 5.5 and pH v , respectively. By solving Equations (1) and (2), S and F pHv /F pH 7.4 were calculated as follows: where the pK a value of SEP estimated previously by using SypHy was 7.1 53 . Finally, SV pH (pH v ) was then calculated as follows: Fluorescence imaging was carried out at room temperature (~27 °C) on an inverted microscope (Olympus) equipped with a 60× (1.35 NA) oil immersion objective and 75 W Xenon lamp. Images (1024 × 1024 pixels) were acquired with a scientific cMOS camera (ORCA-Flash 4.0, Hamamatsu Photonics) with 100 ms exposure time under the control of MetaMorph software (Molecular Devices). Fluorescence of SypHy, PMCA1-SEP or Syntaxin-1a-SEP was imaged with 470/22 nm excitation and 514/30 nm emission filters.
For live imaging, images were acquired in time-lapse mode at 1 Hz under control of MetaMorph software. Field stimulation was delivered via bipolar platinum electrodes with 1 msec constant voltage pulses (50 V). A solution containing 50 mM NH 4 Cl was applied at the end of recordings as described above. study approval. All experiments in this study were carried out under the approval of Doshisha University Animal Committee and Recombinant DNA Experiments Committee, and of the Institutional Animal Care and Use Committee of the RIKEN Kobe branch (approval number: QA2013-04-6). All experiments were performed in accordance with the guidelines and regulations of the respective institutions.

Data Availability
The data that support the findings of this study are available from the corresponding authors upon reasonable request.