Contribution of plasma membrane Ca2+ ATPase to cerebellar synapse function

Abstract

The cerebellum expresses one of the highest levels of the plasma membrane Ca²⁺ ATPase, isoform 2 in the mammalian brain. This highly efficient plasma membrane calcium transporter protein is enriched within the main output neurons of the cerebellar cortex; i.e. the Purkinje neurons (PNs). Here we review recent evidence, including electrophysiological and calcium imaging approaches using the plasma membrane calcium ATPase 2 (PMCA2) knockout mouse, to show that PMCA2 is critical for the physiological control of calcium at cerebellar synapses and cerebellar dependent behaviour. These studies have also revealed that deletion of PMCA2 throughout cerebellar development in the PMCA2 knockout mouse leads to permanent signalling and morphological alterations in the PN dendrites. Whilst these findings highlight the importance of PMCA2 during cerebellar synapse function and development, they also reveal some limitations in the use of the PMCA2 knockout mouse and the need for additional experimental approaches including cell-specific and reversible manipulation of PMCAs.

Key Words: Plasma membrane Ca²⁺ ATPase; Cerebellum; Calcium; Purkinje neuron

THE CEREBELLUM, THE PURKINJE NEURON AND THE IMPORTANCE OF CALCIUM DYNAMICS AT CEREBELLAR SYNAPSES

The cerebellum is a major centre for the integration of sensory and motor information in the brain and plays a central role in our ability to learn and refine motor tasks; the specialised function of the cerebellum allows us to execute motor tasks in a finely controlled but, at the same time, “unaware” manner that can still be improved by learning. For a detailed review of cerebellar function see[1]. In brief, the function of the cerebellum relies upon the ability of the main output neurons of the cerebellar cortex, the Purkinje neurons (PNs), to integrate the motor and sensory information that they receive through two types of synaptic inputs. The first is provided by the parallel fibre (PF) synapse and the second by the climbing fibre (CF) synapse (Figure 1A). The control of the dynamics of calcium levels within the pre- and post-synaptic compartments at these two very different synapses is critical for their function. In particular, the control of pre-synaptic calcium is important for the controlled release of the excitatory chemical neurotransmitter glutamate. Adequate control of calcium is also important within the post- synaptic compartment for the control of neuronal excitability and the initiation of events that change the strength and efficiency of the synaptic connections during learning[1]. We therefore set out to test the hypothesis that a calcium extrusion mechanism, such as plasma membrane calcium ATPase 2 (PMCA2), could influence calcium dynamics and synaptic communication at these cerebellar synapses. The hypothesis relied upon supportive evidence in the literature for PMCA2 expression in the cerebellum.

Figure 1 Expression of PMCA2 in the cerebellar cortex. A: Schematic of cerebellar cortex circuitry and the synapses therein. PN: Purkinje neuron; PF: Parallel fibre; CF: Climbing fibre; GC: Granule cells; IO: Inferior olive; B: Basket cell molecular layer interneuron; S: Stellate cell molecular layer (ML) interneuron; DCN: Deep cerebellar nuclei; B: PMCA2 expression in PNs and GCs from mouse. Left panel is a low magnification image of a sagittal section of the cerebellar cortex, showing rich expression of PMCA2 (red) in the ML and the weaker level of PMCA2 expression in the granule cell layer (GCL). The right panel shows at higher magnification the same pattern of PMCA2 expression in the ML and GCL (red), now with the PNs identified by their expression of parvalbumin, shown in green, and the overlapping expression of PN PV and PMCA2 as yellow, scale bar is 100 microns; C: Increased expression of PMCA2 during increasing post natal days of rat development as detected by western blot (i.e. protein immunoblot) from cerebellar tissue. Arrows represent the approximate position of the molecular weight markers; D: A quantitative estimate of PMCA2 expression from Western blots from cerebellar tissue from up to 4 animals at each time point; values are normalized to maximal expression of PMCA2 in the cerebellum and bars represent mean and error bars ± SE.

PMCA2 EXPRESSION AT CEREBELLAR SYNAPSES

The rich expression of PMCA2 within the cerebellum[2], and particularly within the PNs[3], has been known since the 1990’s, particularly as isoform specific antibodies became available. Although this early work identified PMCA2 as the most highly expressed PMCA in the cerebellum, PMCA3 is also highly expressed, whilst PMCAs 1 and 4 are much less abundant, although present. Later, similar findings were also confirmed by immunohistochemistry in the rat brain[4].

One of the outstanding features of PMCA2 expression in the cerebellum is its rich expression in the elaborate dendrites of the PNs and also in their specialised post-synaptic compartments, the dendritic spines[3]. PMCA2 is also unique in showing clear membrane located expression within the large PN soma (Figure 1). Interestingly PMCA3 seems to be only very weakly expressed in the somatic membrane of the PN[4], instead recent electron microscopy (EM) ultrastructural evidence indicates that PMCA3 is concentrated within the pre-synaptic compartment of the PFs[5], and it is this that gives rise to the immunohistochemical signal from anti PMCA3 antibodies in the molecular layer (ML) of the cerebellar cortex, where both PN dendrites and PF terminals reside. The same, very elegant EM study, also indicated that PMCA2 was relatively enriched within the post-synaptic PN spines compared with the pre-synaptic compartment[5]; in the latter compartment PMCA2 was rarely detected with the EM immunogold approach. This was curious, especially since the granule cells, whose axons give rise to the PFs, do express PMCA2[4] (Figure 1B). Furthermore, another study in the pig, and more recently in mouse cerebellum, colocalised the pre-synaptic protein synaptophysin with PMCA-positive punctae on PNs[6,7]. Subsequent biochemical fractionation approaches have revealed that whilst not enriched, PMCA2 is present in the pre-synapse membrane web fraction, prepared from cerebellar cortex. Furthermore, PMCA2 makes molecular interactions within synapse-enriched protein complexes that contain syntaxin 1A, one of the components of the pre-synaptic, calcium-dependent release machinery[8]. The molecular and functional Ca²⁺ imaging evidence[9] also points to an important contribution from PMCA2 to pre-synaptic PF function in the cerebellum (see below).

In contrast to the pre-synaptic side, the presence of PMCA2 in the post-synaptic PN spines is very clear[3]. More recently, antibodies to the PMCA2 “a” isoform have revealed that PMCA2a is present in the PN dendrites and spines[10] confirming the presence of the shorter PMCA2 isoform detected in the earlier Western blot studies from cerebellar tissue[2]. This is functionally significant as the PMCA2a splice form lacks the full auto-inhibitory C-terminus of PMCA2. This feature endows PMCA2a with one of the fastest activation kinetics of the PMCAs (second only to PMCA3“f”, where the C-terminus is even shorter than in the “a” splice isoform)[11]. In a small compartment, like a post-synaptic spine, the fast activation of PMCA2a could be critical for cerebellar synapse function and the integrative properties of the PN. PMCA2b, the longer PMCA2 isoform that possesses the full length autoinhibitory domain and the C-terminal PDZ domain interacting sequence, is also present in cerebellar tissue[2]. It therefore seems tempting to suggest that PMCA2b co-exists with PMCA2a to provide a fine-tuning of the speed of calcium extrusion kinetics from the PN. Further studies will be required to determine this. However, the longer C-terminus of the PMCA2b isoform seems to be functionally important in the cerebellum as its PDZ-binding sequence is capable of interacting with the synaptic PDZ-domain-containing protein, PSD95 (and also PSD97), but not SAP102[8]. This interaction, and the promiscuity of the PMCA2b PDZ sequence, was originally identified in the forebrain[12] and provided the first evidence that PMCA2b could play an important signalling role at synapses. Subsequently, we have identified subunits (NR1 and NR2a) of the NMDA receptor subtype of glutamate receptor as important components of the complex of PMCA2 and PSD95 both in cerebellum and forebrain[8]. The recent confirmation of the functional contribution of NMDA receptors to excitatory post-synaptic transmission in the PN dendrites[13,14] raises the possibility that a PMCA2, PSD95, NMDA receptor tripartite interaction plays a role during PN (and/or other cerebellar) synapse plasticity. Such plasticity, for example at the PF-PN synapse, involves nitric oxide (NO) signalling[15], and given the fact that PMCA4b is a negative regulator of NO synthase in the heart and endothelium[16,17], we might speculate that PMCA2b could perform a similar role in the cerebellum.

FUNCTIONAL IMPACT OF PMCA2 FOR CEREBELLAR SYNAPSE FUNCTION

The high levels of expression of PMCA2 at cerebellar synapses, by themselves, provided strong support for a role for PMCA2 during cerebellar synaptic transmission. This hypothesis is further strengthened by the behavioural phenotype of the PMCA2 knockout (PMCA2^-/-) mouse[18]. The mouse exhibits profound hearing loss and shows vestibular disturbances and a cerebellar “ataxic” phenotype. Furthermore, given the paucity of specific or effective pharmacological inhibitors of PMCA activity, the PMCA2 knockout mouse currently offers the best way to remove PMCA2 from a pre- or post-synaptic compartment for physiological studies. Our initial studies with the PMCA2^-/- mice revealed that the ataxic phenotype only becomes obvious at around 12 d old, and this is consistent with the time taken for postnatal maturation of the PNs, their synapses and normal PMCA2 expression[7]. Our observations confirmed that development of the cerebellum is altered in the PMCA2^-/- mice as revealed by the smaller size of the PMCA2^-/- cerebellum (even though the forebrain of these mice appears to develop normally) and the reduced thickness of the ML, as also reported by Kozel et al[18].

Contribution of PMCA2 to pre-synaptic calcium signalling at cerebellar synapses

Despite these changes to the cerebellar cortex, electrophysiological recordings from the PMCA2^-/- PNs revealed that they respond to stimulation of the PFs with excitatory post-synaptic currents as a consequence of calcium dependent release of glutamate from the pre-synaptic PF terminals. The unique properties of the PF-PN synapse also provided an opportunity to investigate the functional role of PMCA2 for the control of pre-synaptic calcium (Ca²⁺) dynamics. Stimulation of the PFs activates pre-synaptic P/Q type voltage gated calcium channels that allow calcium entry into the PF terminal[19], thereby raising terminal (Ca²⁺) and triggering the release of glutamate. Extrusion of elevated (Ca²⁺) occurs at the timescale of 100 ms. Because of this rather slow decay, (Ca²⁺) remains elevated so that it can add together with any additional closely timed calcium influx. This means that if a second stimulation is applied rather soon after the first, a greater release of transmitter will occur as a consequence of raised “residual” calcium levels[20] (see Figure 2). This mechanism explains the paired-pulse and frequency facilitation observed at PF synapses. We therefore expected that the rate of removal of the “residual” calcium at this synapse by a slower calcium extrusion mechanism, such as PMCA2, would be ideally placed to influence the properties of the synapse. Experimental evidence for this was provided by direct recordings of PF pre-synaptic calcium transients at the PF-PN synapse. These revealed a slower recovery in the PMCA2^-/- PFs presumably because PMCA2 was not present to provide for fast extrusion of the calcium. Furthermore, as a consequence of the slower recovery of [Ca²⁺] in the pre-synaptic terminal we observed an enhancement of paired-pulse facilitation of post-synaptic PN currents. This enhancement presumably arose as a consequence of the raised “residual” calcium in the absence of PMCA2 (shown schematically in Figure 2). Whilst these results provided the first indication that PMCA2 could influence synapse plasticity, they also provided functional evidence for PMCA2 at the pre-synaptic PFs (see also above).

Figure 2 Schematic of the influence of deletion of PMCA2 on “residual” calcium at the PF to PN synapse. Upper solid traces represent the calcium concentration dynamics in the pre-synaptic terminal in response to a paired stimulation of the PFs; calcium rises once and then again a few milliseconds later. Lower dotted line trace represents the excitatory post synaptic current (EPSC) on the post-synaptic side as a consequence of glutamate released during the rise in calcium in the pre-synaptic terminal; note the larger second EPSC coinciding with the larger peak calcium, the paired pulse facilitation (PPF). In red, in the PMCA2 knockout tissue, calcium rises in the pre-synaptic terminal but then decays more slowly, meaning that on the second stimulation, calcium rises even higher; this is seen as an increase in the peak calcium and also the residual calcium. As a consequence, the second EPSC, dotted lines, lower trace, is larger in amplitude, since more glutamate was released from the PF terminal as a consequence of increased calcium. The recovery of the facilitated EPSCs is also slowed in the PMCA2^-/- cerebellum, as residual calcium stays elevated for a longer time in the absence of the PMCA2.

The PF-PN synapse proved to be a very good model to test the contribution from the PMCA2 to synaptic transmission in the PMCA2^-/- mouse, firstly, because we could directly measure the pre-synaptic calcium decay kinetics, and secondly, because the facilitatory properties (and low release probability) of the synapse relies on a slower build-up of pre-synaptic residual calcium. But is PMCA2 only important at synapses where residual calcium decay kinetics are important? We therefore tested the other major synapse to the PN, the CF synapse in the PMCA2^-/- mouse. In contrast to the PF-PN synapse, the CF-PN synapse is a very high release probability synapse; this means that a large release of glutamate follows stimulation in order to create a fail-safe synapse[21]. In response to a second stimulation, rather than facilitating, as occurs at the low release probability PF synapse, the CF synaptic response depresses. This is thought to arise as a consequence of depletion of readily available neurotransmitter[21]. In the PMCA2^-/- mouse, CF responses and the extent of depression at the CF synapse was very similar to that seen in wild type mice (Figure 3). Our result suggests that PMCA2 is either not present, or not required, at the CF pre-synapse, and indicates an important segregation of the mechanisms available to control pre-synaptic (Ca²⁺) depending upon the type and behaviour of the synapse. Interestingly, in motor terminals of Drosophila larvae, PMCA2 makes little contribution to the function of the high release probability Is terminals compared with the lower output Ib terminals[22].

Figure 3 Intact CF evoked excitatory post-synaptic currents in PMCA2^-/- PNs. A: Representative traces of CF-evoked excitatory post-synaptic currents, EPSCs (stimulation shown by downward arrows) from individual mouse PNs voltage clamped at -20 mV. Patch electrodes contained 5mM QX314 to prevent the occurrence of Na⁺ dependent action potentials. Note that the second response, EPSC was smaller than the first, indicating a relative depression of the second EPSC (paired pulse depression) consistent with the high release probability of glutamate at the CF synapse[21]; B: There was no significant difference in the amplitude of the first EPSC; C: No significant difference in the extent of paired pulse depression. Values are mean ± SE error bars.

Contribution of PMCA2 to post-synaptic calcium signalling at cerebellar synapses

The rich expression of PMCA2 in the PN dendrites and spines pointed towards a contribution of PMCA2 to the kinetics of decay of calcium transients in these PN compartments. An initial indication that PMCA2 in the PN dendrites is physiologically important came from our findings that the PN dendrites in the PMCA2^-/- mice are stunted[9], as if they do not fully develop in the absence of progressively increased expression of PMCA2 that normally takes place during cerebellar development[7,23] (in the rat, Figure 1C).

A recent report also indicated that PMCA2 contributes to the post-synaptic molecular complex formed between the InsP₃ receptor, metabotropic glutamate receptors (mGluR1) and homer, a signalling adaptor scaffolding protein[24]. This is a very interesting finding and provided the first evidence of a signalling role for a PMCA2 molecular interaction in a post-synaptic cerebellar compartment, additional to its role in calcium clearance[25]. Subsequent electrophysiological analysis shows that the mGluR1 current is markedly reduced in the PMCA2^-/- cerebellar PNs (Figure 4), even though the mGluR1 receptors could be detected, if disorganized, in these cells[24,27]. The reduction in mGluR1 mediated synaptic signalling could provide one of the most important clues as to the behavioural cerebellar ataxia in the PMCA2^-/- mice. Metabotropic glutamate receptors are critically involved in cerebellar learning (LTD or long term depression) are depleted in two mouse models of ataxia[28,29] and are also lost in a rare form of human ataxia[30].

Figure 4 Metabotropic glutamate receptor currents are reduced in PNs from PMCA2^-/- mice. Given the interaction between metabotropic glutamate receptors (mGluR1) and PMCA2[24] we tested mGluR1 function using an electrophysiological assay. A: Representative traces of mGluR1 dependent inward currents in response to increasing numbers of 100 Hz stimulation to the PFs[36]. The timing of the stimulation is shown by the vertical arrows. As the number of stimulations increased, the amplitude of the evoked inward current increased, as is evident in the wild type PMCA2^+/+ example. In the wild type, the responses to 0, 3, 5, 7 and 9 stimuli are shown for a single cell and for the PMCA2^-/- cell the responses to 0, 3, 5, 7, 8 and 9 stimuli are shown. The responses from PMCA2^-/- PNs were much reduced both in size and peak response; note the smaller vertical amplitude scale bar in the lower panel. These recordings were made with other ionotropic excitatory and inhibitory synaptic events, pharmacologically blocked, and also in the presence of TBOA to enhance available glutamate by preventing its uptake[26]; as expected the inward currents were abolished by the application of the mGluR1 antagonist, 10 μmol/L CPPCOEt, n = 6. mGluR currents were unaffected by the application of the PMCA inhibitor 10 μmol/L carboxyeosin, n = 3, P = 0.6 paired t-test, data not shown; B: The combined mean data where the increase in the amplitude of the mGluR current in wild type PMCA2^+/+ PNs is evident as the number of stimuli increased. Values are mean and error bars are SE, error bars are within the size of the open symbols in the case of the PMCA2^-/- data.

At a physiological level, given its rich expression of PMCA2, the PN provides a good model to test the contribution of PMCA2 to cytosolic post-synaptic (Ca²⁺) clearance kinetics. To do this, we took advantage of the CF synapse where activation of the synapse leads to a reproducibly large rise and fall of cytosolic calcium in the PN post-synaptic dendrites. This post-synaptic PN calcium transient is initiated by a glutamatergic depolarisation following pre-synaptic CF activation accompanied by opening of post synaptic P/Q type voltage gated calcium channels[31] where post-synaptic (Ca²⁺) decay kinetics rely upon the removal of cytosolic calcium by binding to the endogenous PN calcium buffers calbindin and parvalbumin[32].

For technical reasons, the stunted nature of the PMCA2^-/- PN dendrites made analysis and interpretation of fluorescence based calcium measurements from these cells difficult, although qualitatively it was very clear that the shape and behaviour of the calcium transients in the absence of PMCA2^-/- were severely perturbed and that there was little calcium control in the dendrites (unpublished data). Basal calcium was also very high in these PMCA2^-/- PNs, a finding consistent with the raised basal calcium in sensory neurons when PMCA activity is removed[33]. Whilst these results indicated the importance of PMCA2 in the PN dendrites for adequate calcium control, a more quantitative approach was required. To do this, we took advantage of the PMCA2^+/- heterozygous mouse where the PNs were not visibly stunted or disordered, but where PMCA2 expression in the PNs was reduced to about half its wild type levels[34] (as also predicted from Northern analysis[18]). Under these conditions the decay kinetics of CF-induced calcium transients are significantly slowed. This provided a clear indication of the importance of PMCA2 for the recovery of cytosolic (Ca²⁺) in the PN dendrites under normal conditions[34]. The result contrasted to the rather small contribution the PMCA makes to the clearance of cytosolic (Ca²⁺) from the PN soma in a thorough study by Fierro et al[35]. The greater reliance on PMCA2 for calcium control in dendrites compared with the soma probably lies with the restricted diffusional space for calcium in the dendrites. The disturbance to the control of cytosolic (Ca²⁺) in the PMCA2^+/- dendrites had important consequences for the excitability and output properties of the PNs. The electrical output or action potential firing frequency behaviour of the PMCA2^+/- PNs was significantly slower than normal[34]. Since the timing of the electrical output of the PNs is critical for motor control[36], we hypothesized that the slowing might disrupt the motor behaviour of the PMCA2^+/- mice. However, the behaviour of the PMCA2^+/- mice in a normal cage environment is almost indistinguishable from their wild type litter mates, so to test the behavioural hypothesis we used a discriminating test of motor performance based upon the ability of the mice to traverse a narrow beam, a task also known to rely heavily on cerebellar function[37,38]. The PMCA2^+/- mice performed at a level well below their wild type litter mates[34] and always made more slips from the beam, although they did learn to traverse the beam over several trials. This result strongly suggests that a reduction (not even a complete removal) of PMCA2 expression in PNs is sufficient to perturb cerebellar function. Of course, other processes that involve PMCA2 function could also contribute to the behavioural impairment of the PMCA2^+/- mice, and possibilities include the potential for a decrease in motor unit number estimate[39], as has been reported in the deafwaddler PMCA2 heterozygous mutant mouse[40], some mild hearing loss and possibly some mild vestibular deficits. Nevertheless, the PMCA2^+/- mouse provides a useful model to study the contribution of PMCA2 to PN synapse function (and perhaps other physiological functions) without the rather dramatic anatomical alterations of the PMCA2^-/- cerebellum.

COMPENSATORY ALTERATIONS AT SYNAPSES WITHIN THE PMCA2^-/- CEREBELLAR CORTEX

Complete lack of PMCA2 clearly has major consequences for the PN dendrites, not only for their ability to control cytosolic (Ca²⁺) but for a very obvious reduction in the complexity of dendritic branching (Figure 5A). However, although substantially impaired, the PMCA2^-/- mouse still has sufficient motor control to successfully breed (albeit at a reduced success rate, unpublished observations), although it is not clear how much the cerebellum contributes to the remaining motor skills in the mouse. If so, it implies that adaptive mechanisms must operate within the PMCA2^-/- cerebellar cortex during development even though cerebellar PN development is critically dependent upon strict control of dendritic and spine calcium[41]. Given the elevated (Ca²⁺) in the dendrites and the sensitivity of dendritic spines to excessive rises in intracellular (Ca²⁺)[42], we expected the number of post-synaptic spines to be markedly reduced in the PMCA2^-/- dendrites. However, reconstruction of PN dendrites and manual counting of the spine density along the dendrite length revealed only a small reduction in spine number, as shown in Figure 5. This was consistent with our ability to record excitatory synaptic responses from the PMCA2^-/- PNs following both PF[34] and CF stimulation (Figure 3).

Figure 5 Dendritic morphology of PMCA2^-/- PN dendrites is severely disrupted. A: Representative PNs from wild type PMCA2^+/+ (left) and PMCA2^-/- (right) cerebellum. Note the reduced thickness of the ML in PMCA2^-/- cerebellar cortex, as indicated by the black dashed line and the disorganized dendritic tree. PNs had been previously filled with biocytin during electrophysiological patch clamp recordings. Images shown here are after post-hoc identification of the cells with an anti-biocytin antibody; B: Higher resolution images show the tertiary dendrites in more detail from PNs collected in the same way as in (A). Images are reconstructed from a stack of 20 consecutive confocal slices at 0.1 μm intervals. Spines are visible along these tertiary dendrites in both examples, but the spines on the PMCA2^-/- dendrites are less ordered and have lower density. C: The mean values from combined data from more than 90 segments of dendrite are shown, error bars are SE, ^bP < 0.001.

FUTURE PERSPECTIVES FOR THE CONTRIBUTION OF PMCA2 TO NEURONAL CA²⁺ DYNAMICS IN THE CEREBELLUM

Studies over the last few years have highlighted the importance of PMCA2 and its contribution to synaptic calcium dynamics for cerebellar function. Whilst the PMCA2^-/- mouse has been critical for these studies, it is becoming clear that complicated adaptive changes also arise in the PMCA2^-/- cerebellum most likely as a consequence of inadequate PMCA2 and calcium signalling during cerebellar development. This makes the search for new approaches to study the physiological contribution of PMCA2 and its isoforms both timely and necessary. In vivo viral-based delivery of timed siRNA PMCA2 knockdown might offer a more useful approach[43,44] together with the development of transgenic mice approaches for knock-in or knock-out, especially if expression could be limited to specific cell types within the cerebellum, such as the PNs[45]. In addition, improvement of PMCA-specific pharmacological tools beyond the advances made by the development of the caloxins[46] would further aid physiological studies.

In all our considerations of calcium dynamics we must also consider how the actions of PMCA2 work together with other members of the calcium toolkit in order to shape calcium signals in neurons (and all cells)[47]. In particular we continue to consider how PMCA interacts with other calcium extrusion mechanisms (and even between its own isoforms with their different kinetic properties[11]) in order to successfully and fully understand how calcium extrusion is controlled in different neuronal compartments. In particular, the sodium calcium exchanger, NCX[48], the potassium dependent NCX, NCKX[49], the sarcoplasmic reticulum calcium ATPase, SERCA[6] and the secretory pathway calcium ATPase SPCA[50] are all expressed within the cerebellum and all can be predicted to work with PMCAs to influence cerebellar synapse calcium dynamics.

In summary, there is still much to be understood about how PMCAs contribute to calcium dynamics and signalling events at cerebellar synapses and how this influences cerebellar network function, motor learning and behaviour.