Long-term depression (LTD) of Purkinje cell–parallel fiber synaptic transmission is a critical determinant of normal cerebellar function. Impairment of LTD through, for example, disruption of the metabotropic glutamate receptor–IP3–calcium signaling cascade in mutant mice results in severe deficits of both synaptic transmission and cerebellar motor control. Here, we demonstrate that selective genetic deletion of the calcium-binding protein calbindin D-28k (calbindin) from cerebellar Purkinje cells results in distinctly different cellular and behavioral alterations. These mutants display marked permanent deficits of motor coordination and sensory processing. This occurs in the absence of alterations in a form of LTD implicated in the control of behavior. Analysis of synaptically evoked calcium transients in spines and dendrites of Purkinje cells demonstrated an alteration of time course and amplitude of fast calcium transients after parallel or climbing fiber stimulation. By contrast, the delayed metabotropic glutamate receptor-mediated calcium transients were normal. Our results reveal a unique role of Purkinje cell calbindin in a specific form of motor control and suggest that rapid calcium buffering may directly control behaviorally relevant neuronal signal integration.
- calbindin D-28k
- conditional null mutant
- Purkinje cell
- motor coordination
- long-term depression
- synaptically evoked calcium transients
One of the hallmarks of cerebellar Purkinje cells is their ability to express a characteristic form of activity-dependent synaptic plasticity named long-term depression (LTD). LTD is induced by the joint activity of afferent climbing fibers (CFs) and parallel fibers (PFs) and represents a persistent reduction of the efficacy of parallel fiber-mediated excitatory postsynaptic potentials (Ito, 1986). A key event in LTD induction is the activation of postsynaptic metabotropic glutamate receptors (mGluRs) (Rose and Konnerth, 2001). Mice null mutant for mGluR1, a subtype of mGluRs that is expressed in Purkinje cells (Martin et al., 1992; Gorcs et al., 1993; Ryo et al., 1993), exhibit severe cerebellar symptoms such as ataxia and motor discoordination and lack LTD (Aiba et al., 1994;Conquet et al., 1994). The behavioral relevance of Purkinje cell mGluR1 for proper cerebellar function was elegantly and unequivocally confirmed by the selective rescue of mGluRs in Purkinje cells of mGluR1 null mutant mice (Ichise et al., 2000), which restored both LTD and locomotor activity.
Further confirmation and extension of these findings came from studies showing that a perturbation of the mGluR1-activated signaling cascade at virtually every level [e.g., G-protein Gαq (Offermanns et al., 1997), phospholipase Cβ (Kano et al., 1998), and IP3 (Matsumoto et al., 1996; Miyata et al., 2000)] results in impaired LTD and disturbed cerebellar motor control. IP3 causes Ca2+release from intracellular stores (Mikoshiba et al., 1994; Finch and Augustine, 1998; Takechi et al., 1998), and synaptically induced dendritic Ca2+ signaling has been known for a long time to be a critical step in LTD induction (Sakurai, 1990;Konnerth et al., 1992). On the basis of earlier evidence, it seemed that not merely the lack of synaptically mediated Ca2+ signaling in Purkinje cells, but also a change in the temporal dynamics of postsynaptic Ca2+ transients after general genetic deletion of the calcium-binding protein calbindin D-28k (calbindin), might cause impairment of motor coordination (Airaksinen et al., 1997). However, despite the suggestion of an involvement of Purkinje cells, it remained unclear which cell types were responsible for the behavioral phenotype, because calbindin is expressed by many types of neurons throughout the brain (Celio, 1990). Even in the cerebellum, calbindin is expressed abundantly not only in Purkinje cells but also in climbing fibers (Celio, 1990; Scotti, 1995), and previous studies demonstrated that Ca2+-binding proteins influence presynaptic transmitter release (Chard et al., 1995; Klapstein et al., 1998; Caillard et al., 2000). Finally, it was not tested whether the calbindin-mediated change in Ca2+signaling affected cerebellar LTD. To study the specific role of calbindin in a single defined neuronal cell type, we generated a Purkinje cell-specific calbindin null mutant mouse strain and performed an analysis on the cellular and behavioral level.
Materials and Methods
Conditional null mutant mice. A mouse strain carrying a floxed calbindin allele was used (Barski et al., 2002). In brief, a pgk promoter-neo-pA-pgk promoter-TK-pA cassette flanked by loxP sites in the same orientation was inserted into the Eco47III site of a 6 kb XhoI–SalI genomic fragment (Airaksinen et al., 1997), and a third loxP was ligated into the ClaI site upstream of the first coding exon. ES cell clones were generated in R1 embryonic stem cells (a gift from A. Nagy, University of Toronto, Toronto, Canada), and correctly targeted clones underwent a second electroporation with the Cre expression plasmid pMCCre (a gift from H. Gu and K. Rajewsky, University of Cologne, Cologne, Germany). Clones with the desired recombination were injected into C57Bl/6 (C57Bl/6NCrl BR; Charles River, Sulzbach, Germany) blastocysts. Resulting chimeras were crossed with C57Bl/6 mice. Animals of heterozygote intercrosses were used for analysis. The Cre transgenic strain used in this study has previously been characterized (Barski et al., 2000); here, we used substrain L7Cre-2.
Western blot and immunohistochemistry. Western blotting was as previously described (Airaksinen et al., 1997) using a mouse monoclonal antibody to calbindin (1:10,000; SWant). For immunohistochemistry (Airaksinen et al., 1997) on 30-μm-thick free-floating sections, we used rabbit antibodies to parvalbumin (1:500; SWant) and mouse monoclonal calbindin antibodies (1:500; SWant) together with fluorescein or Texas Red-coupled secondary antibodies (Jackson ImmunoResearch, West Grove, PA). Images were taken on a Leitz DM IRB confocal microscope (Leica, Nussloch, Germany) using the graphics program Image-Space (Molecular Dynamics, Sunnyvale, CA).
Reverse transcription-PCR. The reverse transcription (RT) reaction was performed on 1 μg of total RNA from the indicated brain regions, as described earlier (Barski et al., 2002).
Locomotion analysis. Open-field analysis was performed on the TruScan activity monitor (Coulbourn Instruments), as described (Barski et al., 2000).
The runway and horizontal bar tests were performed essentially as before (Airaksinen et al., 1997). The number of assay days and width of the runway were as indicated. Maximal time on the horizontal bar was 120 sec. Data were analyzed with the ANOVA single-factor test.
Compensatory eye movements. Measurement of compensatory eye movements was performed essentially as described before (De Zeeuw et al., 1998a). The eye coil had an outside diameter of 0.5 mm and was made of 40 windings of 25-μm isolated brass wire (resistance, 25–35 Ω). A servo-controlled turntable and drum (Benedict) delivered optokinetic and vestibular stimuli. The stimulus paradigms included sinusoidal optokinetic stimulation and sinusoidal table stimulation in the dark and light at frequencies from 0.1 to 1.6 Hz and with peak velocities from 3 to 96°/sec. All gain and phase values were analyzed off-line using the Ced-Anal program for Windows according to standard procedures in our laboratory. Eye movements were recorded from 11 adult recombined and 10 adult control mice.
Electrophysiology in slices and LTD. Whole-cell recordings were obtained from Purkinje cells of 300-μm thick cerebellar slices from 18- to 30-d-old mice. The animals were decapitated after anesthesia with CO2, and the cerebella were rapidly removed and placed in ice-cold artificial CSF composed of (in mm): 125 NaCl, 2.5 KCl, 2 CaCl2, 1 MgCl2, 1.25 NaH2PO4, 26 NaHCO3, and 20 glucose, bubbled with 95% O2 and 5% CO2. After cutting, slices were kept for 45–60 min at 37°C and then for up to 8 hr at 25°C until they were used for experiments. Somatic whole-cell recordings were obtained with an EPC8 or EPC9 amplifier (HEKA). PULSE software (HEKA) was used for data acquisition. Pipettes (2–4 MΩ resistance) were pulled from borosilicate glass (Hilgenberg) and coated with silicon (RTV 615; GE Silicons). The pipette solution contained (in mm): 148 potassium gluconate, 10 HEPES, 10 NaCl, 0.5 MgCl2, 4 Mg-ATP, and 0.4 Na3-GTP, pH 7.3. For Ca2+ imaging experiments, 100 μm Oregon green BAPTA-1 (Molecular Probes, Eugene, OR) was added to the pipette solution. During the experiments, the slices were continuously perfused with artificial CSF bubbled with 95% O2 and 5% CO2 (at 22–23°C, except for LTD experiments) containing 10 μm bicuculline (Sigma, St. Louis, MO). Synaptic stimulation was performed by using a standard patch pipette filled with 1 mm NaCl (1 MΩ resistance) placed in the molecular layer. The stimulus pulse amplitude (150 μsec duration) was 2–20 V for PF stimulation and 20–55 V for CF stimulation. PF-EPSCs were identified by their characteristic features (graded response amplitude and paired pulse depression). Without moving the stimulus pipette, the intensity was increased until a climbing fiber response was observed, distinguished by its all-or-none character and paired pulse depression (Konnerth et al. 1990, 1992), and the CF stimulus threshold was determined. Parallel fibers were stimulated at 0.2 Hz, and EPSCs were recorded in the voltage-clamp mode until a stable baseline amplitude was obtained for at least 10 min. To induce LTD, the stimulus intensity was raised to a value at least 20% over CF threshold, and 240 stimuli were repeated at 1 Hz in conjunction with a depolarizing pulse (200 msec, −60 to 0 mV). After pairing, the stimulus intensity was set to the initial value, and the recording of PF-EPSCs at 0.2 Hz was resumed for 60 min. Passive membrane properties of Purkinje cells were monitored by applying 3 mV hyperpolarizing pulses. The series resistance was monitored and actively compensated throughout the measurement to ∼10–20 MΩ, and only cells fulfilling our stability criteria were used in the analysis (see Fig. 4). The bath temperature was 32°C.
Calcium imaging. A confocal laser-scanning microscope (Odyssey; Noran), attached to an upright microscope [Axioskop2, 63× water immersion objective, numerical aperture (NA) 0.9; Zeiss, Thornwood, NY) was used to acquire fluorescence images at 30 Hz in parallel to the whole-cell recordings. Ca2+ imaging was started at least 20 min after establishment of whole-cell configuration.
Full-frame images were recorded at 30 Hz on an optical disk (TQ-FH224; Panasonic) using the Image-1 software ( Universal Imaging Corporation, West Chester, PA) and analyzed off-line with custom-made software on the basis of LABVIEW (National Instruments). Ca2+ transients (see Fig. 5 D) were recorded in regions of interest in active dendritic regions. For measurement of calcium transients in dendritic spines, two-photon imaging was performed in parallel to whole-cell patch-clamp recordings. We used a custom-built two-photon laser-scanning microscope on the basis of a mode-locked Ti:sapphire laser system operated at 790 nm center wave length, 80 MHz pulse repeat, and <100 fsec pulse width (Tsunami and Millenia, Spectra Physics) and a laser-scanning system (MRC 1024; Bio-Rad, Hercules, CA) coupled to an upright microscope (BX50WI; OlympusOptical, Tokyo, Japan) equipped with a 60× 0.9 NA water immersion objective (Olympus Optical). To achieve high temporal resolution, fluorescence was acquired in the line scan mode at a sampling rate of 160 Hz (control animals) or 480 Hz (recombined). Background-corrected line scan images were analyzed off-line with a custom-written routine in LABVIEW (National Instruments). The Ca2+-dependent fluorescence signals were expressed as increases in fluorescence divided by the prestimulus fluorescence values (ΔF/F 0) and further analyzed using Igor Pro (Wavemetrics) or Origin (Microcal) software.
Characterization of Purkinje cell-specific null mutant mice
For orientation, the targeting strategy for the floxed calbindin allele (referred to as Calbtm2 ) (Barski et al., 2002) is depicted in Figure1 A. Because the properties of the floxed and recombined alleles are critical to the work presented here, we performed additional experiments to this point and repeated some of the earlier work. Western blotting and RT-PCR experiments confirmed our previous conclusion that the introduced loxP sites are without effect on calbindin levels or distribution in homozygous Calbtm2 mice (Fig.1 B). We had previously shown thatCalbtm2/Calbtm2 mice are also indistinguishable from their wild-type littermates in the open field and runway behavioral assays (Barski et al., 2002). The L7Cre-2 transgenic mouse strain, which expresses Cre recombinase under the control of the Purkinje cell-specificL7/pcp-2 minigene, has previously been shown to allow highly selective and efficient recombination in Purkinje cells (Barski et al., 2000).
Intercrossing of double heterozygous mice gave offspring at the expected Mendelian ratios, which was divided into two groups. The control group (not recombined) comprises all genotypes lacking either the Calbtm2 or the L7Cre-2allele. The recombined group consists of all mice homozygous for theCalbtm2 allele combined with one or twoL7Cre-2 alleles. Within each group, behavioral, biochemical (besides genotypes), histological, and immunocytochemical assays did not reveal differences between animals. Mice of both groups were similar with respect to growth, life span, and fertility. Tissue-specific recombination was verified by RT-PCR (Fig.1 C) and Western blotting (Fig. 1 D). In the cerebellum, calbindin transcripts and protein were readily detected in the control group but almost absent in recombined mice. Residual calbindin was estimated to be <5% of control levels. No differences were detected in cortex and hippocampus. Cell type specificity was investigated by immunohistochemistry in sagittal brain sections. We found that <2% of Purkinje cells of recombined mice displayed residual calbindin immunoreactivity (Fig. 1 E–H; fields selected for a maximum number of not-recombined cells are shown). Purkinje cells of normal morphology could, however, be rendered visible by immunostaining for parvalbumin, another major cytosolic calcium-binding protein of these cells (Fig.1 G,H). Calbindin immunoreactivity was also reduced in basal parts of the cerebellum, where Purkinje cell axons terminate in the deep cerebellar nuclei (Fig.1 I,J). The presence of few calbindin-positive fibers is consistent with the low number of Purkinje cells in which recombination did not occur. No differences in calbindin immunoreactivity were observed for other brain regions (Fig.1 K–P; dentate gyrus and CA1 pyramidal cells; data not shown). Thus, efficient and specific deletion of calbindin from Purkinje cells has been achieved.
Our previous work has revealed impaired motor coordination in the general calbindin null mutant mouse (Airaksinen et al., 1997). In that study, however, it could not be tested whether changes in multiple normally calbindin-expressing populations caused these deficits, or whether calbindin is critically required within a single cell type. To resolve this question, we examined motor coordination of Purkinje cell-specific null mutants.
General locomotor abilities and explorative behavior as tested in the open field did not differ between control and Purkinje cell-specific null mutant groups (data not shown). However, severe locomotor deficits in recombined mice were revealed in a standardized limb coordination test in which foot slips were counted during passage over a narrow runway (Airaksinen et al., 1997). Recombined mice made significantly more errors throughout the entire testing period and during prolonged training (Fig. 2 A). Although their slip rate decreased during the first days of the test, they never reached the performance levels of controls. Rather, their error rate appeared to asymptotically level off at values clearly higher than those of the control group. Thus, lack of calbindin in Purkinje cells results in a permanent impairment of coordination that cannot be compensated even by prolonged training. Moreover, the deficit was correlated with the difficulty of the task (Fig. 2 C). Within groups, decreasing the width of the runway by a factor of two (from 2 to 1 cm) resulted in twofold higher initial slip rates, with the recombined group making more errors on both runways than the not-recombined animals. An interesting additional difference became apparent at the end of testing. Control mice reached similar levels of performance irrespective of the width of the runway used, whereas recombined mice made two times more errors on the narrow than on the wide runway. Furthermore, groups differed in their ability to stay on a stationary horizontal bar (Fig. 2 B). Whereas many mice of the not-recombined control group reached the test criterion (120 sec on the horizontal bar) on the fourth day, for recombined animals, only a small increase in the time spent on the bar occurred between days 1 and 3, after which no further improvement was observed.
These experiments show that specific motor deficits are indeed induced by lack of calbindin from a single cell type, the cerebellar Purkinje cell.
Compensatory eye movement
We sought to further dissect calbindin-dependent motor coordination by measuring optokinetic reflex (OKR) and vestibular-ocular reflex (VOR) (DeZeeuw et al., 1998a). Both reflexes use sensory information [visual for the OKR and vestibular for the VOR in the dark (VOR-D)] to compensate for movements of visual scene or head. Both types of sensory input act synergistically in the light (VOR-L). The parameters gain and phase are measures for the amplitude and timing of eye movements relative to those of the visual scene or head (Koekkoek et al., 1997).
The gain of the OKR was significantly lower in recombined than control animals at all stimulus frequencies tested (0.1–1.6 Hz; average significance level, p = 0.02), whereas the phase values were not different (Fig.3 A,B). Similarly, gain but not phase of the VOR-L was significantly reduced at all frequencies (0.2–1.6 Hz; average significance level,p = 0.03; Fig. 3 C,D). VOR-D parameters were not altered; the average gain and phase values of both animal groups varied from 0.1 to 0.3 and a lead from 123 to 18°, respectively (data not shown). Thus, the motor pathways involved in these cerebellar reflexes seem not affected by loss of calbindin, whereas the visual control of the extent of compensatory eye movement is clearly impaired. This could be because of either an alteration in visual input to Purkinje cells or a problem with Purkinje cell-dependent processing of this input. Because of the specificity of the generated conditional mutation, the former possibility is unlikely. Specifically, an alteration of the properties of retinal bipolar cells, which are known to express L7/pcp-2 and the Cretransgene used in this study (Barski et al., 2000), can be excluded, because calbindin is not expressed in these cells in mice (Celio, 1990;Pochet et al., 1991). These data show that absence of calbindin from cerebellar Purkinje cells affects eye (as well as limb) coordination and suggest that calbindin in these cells is important for differential regulation of calcium signals involved in processing of visual and vestibular information.
LTD at the parallel fiber → Purkinje cell synapse
Calcium release from internal stores by PF-mediated mGluR activation is essential for physiologically relevant forms of LTD at the PF → Purkinje cell synapse, and supralinearity of postsynaptic calcium signals seems to be the prevalent mechanism of coincidence detection that underlies this form of synaptic plasticity (Wang et al., 2000). LTD and its associated behavioral processes are, therefore, likely to be influenced by the action of endogenous calcium-binding proteins. We studied LTD using whole-cell patch-clamp recordings in acute cerebellar slices of Purkinje cell-specific null mutant mice in which postsynaptic effects of calbindin deficiency can be studied independently from presynaptic alterations. The stimulation protocol we applied produces reliable, nearly saturating depression. The same and similar protocols have revealed impaired LTD in other mutant mouse models tested in our and other laboratories (Aiba et al., 1994; Conquet et al., 1994; Kashiwabuchi et al., 1995; De Zeeuw et al., 1998b).
In both control and Purkinje cell-specific null mutant mice, the amplitude of Purkinje cell EPSCs evoked by PF stimulation decreased by 40% after performing the LTD induction protocol and remained at this level for at least 60 min (Fig. 4). There were no differences in the time course of depression of EPSCs. These results indicate that the buffering of calcium signals by calbindin is not relevant for induction and maintenance of a form of LTD, which is often impaired in cerebellar mutants.
Calcium transients evoked by synaptic stimulation
We measured synaptic calcium transients in spines and adjacent dendrites of Purkinje cells in cerebellar slices (Fig.5 A,B) using a combination of whole-cell patch-clamp recordings with two-photon imaging. As reported earlier, single-shock PF activation produced fast calcium transients in dendrites and spines because of an AMPA receptor (AMPAR)-mediated local depolarization associated with Ca2+ entry (Eilers et al., 1995). These early synaptic calcium transients (ESCTs) were two to three times larger in amplitude, and their decay time constants were three to four times shorter in recombined than in control animals.
We also investigated synaptic calcium transients elicited by climbing fiber stimulation that results in a standard all-or-none electrical response, the so-called “complex spike” (Llinas and Sugimori, 1980b). Single-shock activation of climbing fibers evoked a complex spike that consisted of a burst of five to six action potentials. As observed earlier (Airaksinen et al., 1997), the electrical responsiveness of the Purkinje neurons was indistinguishable between control and calbindin-deficient mice. Postsynaptic calcium transients were detected throughout the entire dendritic tree and in adjacent dendritic spines. The decay of these calcium transients was best fitted with a monoexponential function (Fig.5 B,C). Although the electrical response was similar in control and mutant animals, striking differences between the amplitudes and time courses of fast synaptic calcium transients were observed. In mutant mice, peak amplitudes were enhanced by ∼80% in both spines and dendrites, and decay time constants were reduced by >50% (Fig.5 B,C). These data clearly show that calbindin deficiency of Purkinje neurons results in a considerable enhancement and sharpening of fast postsynaptic calcium transients.
To address the question of whether calbindin also affects the amplitude and kinetics of slow postsynaptic calcium signals, a third type of calcium transients that can be evoked by repetitive PF stimulation was investigated (Finch and Augustine, 1998; Takechi et al., 1998). These transients start only ∼200 msec after stimulation and peak within 500 msec and were thus termed “delayed synaptic calcium transients” (DSCTs) (Takechi et al., 1998). DSCTs represent IP3-mediated Ca2+release signals from intracellular stores after activation of mGluRs. Recent work suggests that these are essential for behaviorally relevant forms of LTD at the PF-Purkinje cell synapse (Wang et al., 2000). We analyzed dendritic DSCTs in the presence of the AMPAR antagonist CNQX by means of conventional confocal microscopy and whole-cell patch-clamp recording. In contrast to ESCTs, DSCTs were not significantly altered in recombined mice (Fig. 5 C,D). These data indicate that calbindin selectively buffers only fast synaptically evoked Ca2+ transients, and that it acts close to postsynaptic sites.
Our results directly demonstrate that calbindin expressed in Purkinje cells is an essential determinant of normal motor coordination and sensory integration. The absence of calbindin from this neuron, which exclusively provides an output from the cerebellar cortex, results in a novel mouse phenotype with distinct deficits in the precision of motor coordination and in the processing of coordination-relevant visual information.
The observed deficit is permanent and cannot be fully compensated by learning-related or other plasticity mechanisms. Decreased gain of VOR-L and OKR (both of which of are regulated by visual sensory input) is also observed in the lurcher mouse, a mutant deficient in Purkinje cells (De Zeeuw et al., 1998a). There are, however, clear differences between both mutants in that the phases (i.e., the relative timing of eye movements) were normal in mice lacking calbindin, whereas large alterations in all three experimental paradigms were reported for lurcher mice. This difference may reflect true calbindin-independent phase regulation by Purkinje cells. Furthermore, the gain of the VOR-D is increased in lurcher mice, possibly reflecting secondary plastic changes in the vestibular nuclei because of Purkinje cell loss. Absence of any VOR-D alterations in our mutants suggests that loss of calbindin alone is not sufficient to cause these secondary changes and that the observed behavioral alterations are likely caused by a functional deficit within Purkinje cells.
Differential effects of Purkinje cell-specific calbindin deficiency on visual versus vestibular processing may also be because of recruitment of different extracerebellar brain structures. For example, visually guided limb coordination depends on output from the deep cerebellar nuclei to premotor areas in the cerebral cortex (Passingham et al., 1988; Mushiake and Strick, 1993). In this scenario, genetic changes in all Purkinje cells could invoke differential effects depending on the sensitivity of their extracerebellar targets to altered input from Purkinje cells via the deep cerebellar nuclei.
What are the cellular mechanisms underlying the behavioral malfunction? To assess possible changes in plasticity at this synapse, we used a well established model of LTD, which is thought to be behaviorally relevant. The protocol used to induce LTD was comparable with methods previously applied to other mutants (Aiba et al., 1994; Conquet et al., 1994; Kashiwabuchi et al., 1995; De Zeeuw et al., 1998b). Whereas the calcium-binding protein calbindin is highly expressed in Purkinje cells and acts here as a powerful Ca2+ buffer (Fierro and Llano, 1996; Airaksinen et al., 1997; Maeda et al., 1999), and although Ca2+ signaling is essentially needed for the induction process (Sakurai, 1990; Konnerth et al., 1992), LTD is not affected in the Purkinje cell-specific calbindin null mutants. This preservation of LTD is most likely a consequence of the fact that mGluR-mediated Ca2+ signaling, known to be critical for LTD induction (Daniel et al., 1998; Inoue et al., 1998; Wang et al., 2000; Ito, 2001), seems to be normal in the mutant mice. Thus, at least under experimental conditions, the distortion of AMPAR-mediated Ca2+ signals does not significantly hinder LTD induction. We cannot exclude alterations, particularly abnormally enhanced acquisition of LTD, with other LTD protocols, or in other forms of cerebellar synaptic plasticity. Interestingly, a recent report revealed conditions under which postsynaptic, nitric oxide-dependent long-term potentiation (LTP) is observed at the parallel fiber → Purkinje cell synapse (Lev-Ram et al., 2002). Because this LTP is enhanced by postsynaptic calcium chelators with properties similar to calbindin, a decrease in its strength might be expected in the Purkinje cell-specific null mutants. However, before comparisons between genotypes are meaningful, the conditions required for this form of LTP need to be firmly established.
In contrast to LTD, kinetics and dynamics of synaptically evoked calcium transients are differentially regulated by calbindin. The explanation for the relative calbindin independency of the slow mGluR-mediated Ca2+ signaling is straightforward. Our recordings of ESCTs in calbindin-deficient mice reveal decay time constants of the calcium signals in the range of 80–100 msec. These decay time constants are a direct measure of the calcium-clearing mechanisms in dendrites in the absence of the dominating fast endogenous calcium buffer. Because the mGluR-mediated calcium signal has much slower kinetics, with a time to peak of >500 msec and decay time constants in the range of 800–900 msec, calcium clearance as well as calcium buffering by calbindin, which prolongs the decay time constant of ESCTs by just 200–300 msec (Fig. 5), will not interfere in a major way with the waveforms of DSCTs. Thus, because calcium mobilization and clearance seem to be the major determinants of the mGluR-mediated calcium response, DSCTs were less affected in calbindin-deficient Purkinje cells.
Because calbindin is a rapid endogenous calcium-buffering protein (Roberts, 1993; Airaksinen et al., 1997; Maeda et al., 1999), at least four types of Ca2+ signals can be affected by calbin-din deficiency. First, as shown above, the AMPAR-mediated calcium transients produced by parallel fiber stimulation in spines and dendrites are distorted. In spines and the adjacent dendrites, the amplitudes of the Ca2+ transients are enhanced, and the time course is markedly faster. This distortion is predicted by the absence of a rapid Ca2+buffer (Wagner and Keizer, 1994).
Second, climbing fiber-mediated postsynaptic Ca2+ signals exhibit a similar modification in amplitude and time course in the absence of calbindin (Fig. 5, compare B, C), whereas the electrical responsiveness of the Purkinje cells is apparently not changed. The distortion is observed in dendrites and interestingly also in spines. These results are similar to those described by Airaksinen et al. (1997), who reported amplitude changes that are in the same range. However, whereas the decay phase of synaptic calcium transients was fitted with biexponential functions in the foregoing study, the decay phase of the vast majority of calcium transients analyzed in the present study is best fitted with monoexponential functions. Both the measurement in small compartments and the use of a low dye concentration implicated a reduced signal-to-noise ratio, which might have masked a small additional fast decay time constant. Moreover, Airaksinen et al. found that although the amplitude of the calcium transients was increased in calbindin-deficient mice, the time constants were unchanged (compare Airaksinen et al., 1997, their Figure 4). If the data of Airaksinen et al. (1997) are fitted with only one exponential, the results are identical to those obtained in the present study. We believe that the measurements performed in the present study are more accurate, because we monitored the calcium signals directly at the site of calcium entry. In contrast, the earlier study was performed in thick dendrites relatively close to the cell body.
A third type of Ca2+ signals that might be distorted by calbin-din deficiency are action potential-associated Ca2+ transients in the cell bodies (Llinas and Sugimori, 1980a), dendrites (Llinas and Sugimori, 1980b), axons (Callewaert et al., 1996), and terminals. Thus, action potential-evoked Ca2+ transients in dendrites and spines will also have a higher amplitude and a more rapid decay time course and may alter signal integration, perhaps through a reduced efficacy of coincidence detection (Wang et al., 2000).
Finally, transmitter release from Purkinje cell nerve terminals in the deep cerebellar nuclei may be altered. It has been demonstrated that another calcium-binding protein, parvalbumin, regulates the paired-pulse ratio through modulation of transmitter release (Caillard et al., 2000). Furthermore, overexpression or deletion of calbindin in hippocampal neurons resulted in the selective modification of posttetanic potentiation (Chard et al., 1995; Klapstein et al., 1998), presumably by changing the global Ca2+concentration increase in presynaptic terminals necessary for mediating potentiation of neurotransmitter release on a second to minute time scale.
Thus, although calbindin is absent in only a single, defined cell type in the cerebellum, a wide range of cellular actions involving rapid Ca2+ signaling may contribute to the behavioral deficits.
Our study suggests a new mechanism for an LTD-independent model of cerebellar malfunction caused by the specific deficiency of calbindin in Purkinje cells. Although cerebellar malfunction persisted even after prolonged training, initial rates of improvement of runway performance were similar in calbindin-deficient and control mice, suggesting that some forms of motor learning are intact in the Purkinje cell-specific null mutant. A more rigorous evaluation of motor learning in our mutant by assessment of the adaptability of eye coordination is hindered by their impaired basal eye coordination. Interestingly, different forms of motor learning have been implicated in the phenotype of another Purkinje cell-specific mutant with normal basal eye coordination and impaired LTD (De Zeeuw et al., 1998b; Goossens et al., 2001). Animals with reduced PKC activity are severely impaired in short-term but much less so in long-term adaptability of eye movements. This has led to the hypothesis that LTD may be more important for short-term motor learning (van Alphen and De Zeeuw, 2002). The situation in calbindin-deficient mice seems complementary: normal LTD with normal initial performance improvement is combined with long-term deficits. Thus, rapid Ca2+ signaling regulated by endogenous Ca2+-buffering proteins such as calbin-din might contribute to long-term learning involving as yet unknown cellular targets and pathways.
It needs to be stressed that rapid Ca2+signaling, for example in association with action potential firing, is a general property of all central neurons. A large variety of endogenous calcium-binding proteins control the dynamics of rapid intracellular Ca2+ signals in a highly cell type-specific manner. Therefore, the impact of our results may extend far beyond Purkinje cell and cerebellar function, indicating a widespread role of rapid Ca2+ signaling in neuronal computation.
This work was supported by Deutsche Forschungsgemeinschaft Grant SFB391 (to A.K.), Deutsche Forschungsgemeinschaft Grant ME1121/3 (to M.M.), and grants from the Human Frontier Science Program, European Economic Community, and Nederlandse Organisatie voor Wetenschappelijk Onderzoek (to C.I.D.Z.). We thank B. Kunkel, A. Steinberg, and E. Barska for excellent assistance; and Hans Thoenen, Chris Yeo, and Michael Häusser for discussion and critical reading of this manuscript.
Correspondence should be addressed to Dr. Michael Meyer, Division of Molecular Genetics, Institute of Ophthalmology, University College London, 11-43 Bath Street, London EC1V 9EL, UK. E-mail:.