Hippocampal mossy fibers comprise the axons of dentate gyrus granule cells, which project to form synapses with CA3 pyramidal cells and interneurons. The mossy fiber pathway is characterized by a relatively low basal probability of transmitter release (Jonas et al.,1993), which is maintained by the action of inhibitory presynaptic receptors, including A1 adenosine receptors (Moore et al.,2003), and possibly group II metabotropic glutamate receptors (mGluRs) (Kwon and Castillo,2008). Group II mGluRs (mGluR2; mGlu2, encoded by Grm2 and mGluR3; mGluR3, encoded by Grm3) negatively modulate the adenylate cyclase/protein kinase A (PKAw signaling cascade and are believed to play a key role in the induction of mossy fiber long-term depression (LTD) (Nicoll and Schmitz,2005). This is based on reports that an mGluR2/3 antagonist significantly reduced mossy fiber LTD (Kobayashi et al.,1996; Tzounopoulos et al.,1998), while mossy fiber LTD was largely absent in Grm2 knockout mice (Yokoi et al.,1996).
A recent study using the newer, more potent and selective group II mGluR antagonist, LY341495, reported, however, that mossy fiber LTD was unaffected by this antagonist (Wostrack and Dietrich,2009). This led the authors to suggest that group II mGluRs are, in fact, neither necessary nor sufficient for mossy fiber LTD induction, and that the reduction in LTD observed in earlier studies using less selective antagonists may actually reflect nonspecific blockade of alternative mGluRs.
In addition to a decrease in cyclic adenosine monophosphate (cAMP) induced by presynaptic mGluR activation, induction of mossy fiber LTD also requires an activity-dependent rise in presynaptic intracellular Ca2+ concentration (Kobayashi et al.,1999; Tzounopoulos et al.,1998). Presynaptic kainate receptors (Negrete-Díaz et al.,2006,2007) and A1 adenosine receptors (Hagena and Manahan-Vaughan, 2010) have also been suggested to play a role in mossy fiber LTD. Nevertheless, the possible roles of these additional mechanisms have not been well explored.
We have therefore used Grm2/3 double knockout (dko) mice to directly assess the extent to which group II mGluRs are necessary for mossy fiber LTD and the possible involvement of alternative mechanisms. Surprisingly, we were able to induce mossy fiber LTD of normal magnitude in these mice when recording in 4 mM external Ca2+, whereas LTD was greatly reduced in 2 mM Ca2+. Using whole-cell recordings, we further demonstrated that LTD of the N-methyl-D-aspartic acid (NMDA) receptor-mediated component of the mossy fiber excitatory postsynaptic current (EPSC) in Grm2/3 dko mice was not blocked by postsynaptic application of a Ca2⊃ chelator, indicating that the LTD was presynaptic in origin. LTD of the isolated NMDA receptor-mediated EPSC in Grm2/3 dko mice could, however, be blocked using the α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainate receptor antagonist, NBQX. This study has thus revealed that group II mGluRs are not obligatory for mossy fiber LTD, since a Ca2+ sensitive form of mossy fiber LTD can be induced in Grm2/3 dko mice. This result calls for a possible re-interpretation of the role of the cAMP cascade in induction of mossy fiber LTD.
MATERIALS AND METHODS
All experiments were conducted under the auspices of the UK Home Office project and personal licenses held by the authors. Age-matched, male, wild-type, and Grm2/3 dko mice were obtained from GlaxoSmithKline, Harlow, UK. To generate the dko animals, Grm2−/− mice (Yokoi et al.,1996) were crossed with Grm3−/− mice (Corti et al.,2007a) to generate Grm±Grm± double heterozygous offspring. The double heterozygotes were then crossed to generate 1:16 dko, 1:16 wt, and 14:16 single/double heterozygous mice. To avoid the prohibitive wastage of animals that would occur if all dko and wt mice were derived from double heterozygous crosses, separate lines of true breeding wt and dko mice were established. Mice aged 12–22 months were used in extracellular field recording experiments, and 5-month-old mice in whole-cell experiments. Animals were housed in groups of 2–4 and kept on a 12-h light-dark cycle (lights on at 07:00 and off at 19:00) with ad libitum access to food and water.
Transverse hippocampal slices (350 μm) were prepared from animals that had been deeply anaesthetized by isoflurane inhalation. Slices were prepared in artificial cerebrospinal fluid (aCSF) containing (in mM): NaCl, 126; KCl, 3; NaH2PO4, 1.25; MgSO4, 2; CaCl2, 2; NaHCO3, 26; glucose, 10, chilled to 0–2°C and saturated with carbogen gas (95% O2/5% CO2) to pH 7.2–7.4. Slices were cut using a vibrating blade microtome (Leica VT 1000S). They were then maintained in either an interface chamber at 35–37°C for use in extracellular field potential recordings, or in a submerged-style recording chamber at room temperature (22–27°C) for whole-cell patch-clamp recordings. Slices were allowed to recover for 1–2 h before the start of experiments.
Stimulation and recording conditions
Monopolar stainless steel stimulating electrodes were used (0.01 mm focal tip, 5 MΩ; A-M Systems, Carlsborg, WA), while recording electrodes consisted of silver chloride wires in aCSF surrounded by glass capillaries (4–6 MΩ; Harvard Apparatus, Edenbridge, Kent, UK). Signals from the recording electrode were amplified (Axoclamp-2A, Axon Instruments, Molecular Devices, Foster City, CA), filtered at 1 kHz, and digitized at 2 kHz (Instrutech ITC-16, Instrutech Corporation, Port Washington, NY). A Master-8 stimulator unit (A.M.P.I., Jerusalem) was used to program timing and duration of stimulation current pulses delivered by a stimulus isolation unit (IsoFlex, A.M.P.I.); the amplitude of current delivered could be adjusted manually. Customized procedures within Igor Pro software (version 5.0, WaveMetrics, Lake Oswego, OR) were used for the collection, storage, and analysis of data.
Extracellular fEPSP recordings
A stereoscope was used to position the stimulation and recording electrodes in the stratum lucidum of CA3. Baseline stimulation comprised pairs of pulses (pulse duration 50 μs) with an interpulse interval of 40 ms (25 Hz), delivered every 5 s (0.2 Hz). Paired pulses were used so that the paired-pulse ratio (PPR) could be continuously monitored. Mossy fiber synapses display higher PPRs than the associational-commissural (AC) synapses that are also present in CA3. To ensure that relatively pure mossy fiber responses were recorded, electrodes were re-positioned until the PPR was ≥1.5 (and preferably ≥2). The middle one third of the field excitatory postsynaptic potential (fEPSP) slope was used as an index of synaptic response.
A stimulus-response curve (10–70 μA, mean of five fEPSPs at each stimulation strength) was compiled and input strength subsequently adjusted until the first fEPSP in each pair was approximately 50% of maximal amplitude. LTD induction was only attempted whenever basal synaptic transmission had been stable (≤5% change in fEPSP slope) for at least 10 min. Pairs of stimuli (25 Hz) were delivered at 1 Hz for 15 min (a total of 900 stimulus pairs) and recording continued for 45–60 min. In approximately 40% of slices from wild-type mice, 1 μM (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine (DCG-IV; Tocris Bioscience, Bristol, UK), a group II mGluR agonist, was applied to the perfusing aCSF at the end of the experiment to assess the purity of the mossy fiber response. To block NMDA receptors, in some experiments, 20 μM D-(-)-2-amino-5-phosphonopentanoic acid (D-AP5; Tocris) was added to the perfusate 45 min before the start of baseline recording. To block group II mGluRs in wild-type mice, either the nonselective group I/group II mGluR antagonist (RS)-α-methyl-4-carboxyphenylglycine (MCPG; Tocris) or the selective group II mGluR antagonist (2S)-2-amino-2-[(1S,2S)-2-carboxycycloprop-1-yl]-3-(xanth-9-yl) propanoic acid (LY341495; Tocris) was used.
Whole-cell patch-clamp recordings were made from CA3 pyramidal cells under visual guidance using infrared differential interference contrast microscopy. Current-clamp recordings were made with patch pipettes containing (in mM): potassium gluconate, 110; HEPES (4-[2-hydroxyethyl]piperazine-1-ethanesulfonic acid), 40; NaCl, 4; ATP-Mg, 4; GTP, 0.3; pH 7.2–7.3). To block Ca2+ and G protein-mediated effects postsynaptically, in some experiments 30 mM 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N″-tetraacetic acid (BAPTA) was added to the pipette solution, and potassium gluconate replaced with CsF. Series resistance was monitored regularly throughout the experiment, and cells were rejected if the series resistance changed by more than 15%. Baseline stimulation frequency was 0.1 Hz, and LTD was induced using 15 min of 1 Hz single-pulse stimulation. 2 μM DCG-IV (Tocris) was applied at the end of the experiment to assess the purity of the mossy fiber response in wild-type animals. NMDA receptor-mediated EPSCs were recorded at +40 mV in the presence of 2 μM 6-imino-3-(4-methoxyphenyl)-1(6H)-pyridazinebutanoic acid (SR95531) hydrobromide (gabazine) (Tocris) to block A type γ-aminobutyric acid (GABAA) receptors and 30 μM 1-(4-aminophenyl)-3- methylcarbamyl-4-methyl-3,4-dihydro-7,8-methylenedioxy-5H-2,3-benzodiazepine (GYKI53655), a selective AMPA receptor antagonist. When required, kainate receptors were blocked by the additional application of 3 or 5 μM 2,3-dioxo-5-nitro-1,2,3,4-tetrahydrobenzo[f]quinoxaline-7-sulfonamide (NBQX; Tocris), a mixed AMPA/kainate receptor antagonist.
Statistical analysis was performed using SPSS (v15.0). Paired-pulse facilitation (PPR > 1) was calculated by dividing the slope of the second fEPSP (extracellular) or EPSC (whole-cell) by that of the first. For LTD experiments, the average slope of the first fEPSP/EPSC in each pair for each of the last 5min of baseline recording before the start of 1 Hz stimulation was calculated. The average of these five values was then used as a reference, against which all other data points were normalized. The magnitude of frequency facilitation after 1 min of 1 Hz paired-pulse stimulation was compared across groups using a two-sample t-test. The magnitude of LTD was examined 30–50 min after the end of 1 Hz paired-pulse stimulation, using the average response over the last 5 min of recording. Groups were again compared using a two-sample t-test. In all statistical analysis of fEPSPs, “n” is the number of animals: whenever more than one slice was used from a single animal, an average value for that animal was calculated. In all analysis of whole-cell recordings, “n” is the number of cells. Values are given as mean ± sem unless otherwise indicated. Error bars indicate sem.
Extracellular field recordings
Confirmation of recording from mossy fiber synapses
To ensure that relatively pure mossy fiber responses were obtained, stimulating and recording electrodes were positioned in the stratum lucidum in which the majority of mossy fibers terminate (Blackstad et al.,1970). Given the rapid kinetics of mossy fiber responses compared with their AC equivalents, fEPSPs with a relatively rapid time course (<1.0 ms 20–80% rise time; ∼5.0 ms decay time constant) were chosen. Mossy fibers display stronger paired-pulse facilitation than AC fibers; electrodes were therefore repositioned until the PPR was ≥1.5 (and preferably ≥2) at an interstimulus interval of 40 ms. In approximately 40% of wild-type slices, 1 μM DCG-IV was applied to the perfusate at the end of the experiment and the percentage of blockade calculated. This was positively correlated with the PPR (r = 0.806; P = 0.029; Supporting Information Fig. S1). Note that the recording with a PPR <1.5 (which was therefore excluded from further analysis) displayed relatively little DCG-IV blockade (66.7%) and relatively little frequency facilitation (24.4%), consistent with an impure mossy fiber response. Of the six experiments with PPR ≥1.5, one displayed <80% DCG-IV blockade (73.5%) plus relatively little frequency facilitation (54.1%), suggesting that this response might have been misclassified. This experiment was retained in the analysis, however, as selection on the basis of DCG-IV blockade cannot be used in Grm2/3 dko mice. To ensure comparability between wild-type and dko experiments, care was taken to use the same electrode positions across slices.
Mossy fiber LTD in Grm2/3 dko mice in 4 mM Ca2+/4 mM Mg2+
Presynaptic mGluR2 has been reported to be essential for induction of mossy fiber LTD (Yokoi et al.,1996). However, mGluR2 activation is not sufficient for mossy fiber LTD induction, and must be coupled to an increase in presynaptic Ca2+ (Kobayashi et al.,1999; Tzounopoulos et al.,1998). Moreover, previous work has indicated that mossy fiber plasticity is sensitive to the concentration of divalent cations in the recording medium (Lauri et al.,2003). To maximize the likelihood of uncovering additional induction mechanisms for mossy fiber LTD, we therefore tested whether it might be possible to induce mossy fiber LTD in Grm2/3 dko mice in high Ca2+ conditions (4 mM Ca2+ and 4 mM Mg2+). Surprisingly, 15 min of 1 Hz paired-pulse stimulation induced robust LTD (Fig. 1A). Indeed, in 4 mM Ca2+/4 mM Mg2+ aCSF, the magnitude of LTD did not differ between wild-type and Grm2/3 dko mice (wt mean: 37.8 ± 5% depression, n = 8 slices/8 animals; Grm2/3 dko mean: 42.6 ± 7% depression, n = 9 slices/5 animals; t <1; P > 0.20).
Mossy fiber LTD is reduced in Grm2/3 dko mice in 2 mM Ca2+/2 mM Mg2+
The presence of mossy fiber LTD in Grm2/3 dko mice of the same magnitude as that in wild-type animals was surprising, given the report that mossy fiber LTD was largely absent in Grm2 single knockout mice (Yokoi et al.,1996). We therefore repeated the experiments in 2 mM Ca2+/2 mM Mg2+. Under these conditions, the magnitude of LTD was reduced in Grm2/3 dko mice compared with wild-type control animals (wt mean: 39.7 ± 5.4% depression, n = 8 slices/7 animals; Grm2/3 dko mean: 9.3 ± 3.7% depression, n = 8 slices/6 animals; t(11) = 4.469; P < 0.001) (Fig. 1B), in agreement with reduced mossy fiber-LTD reported under similar conditions in Grm2−/− mice (Yokoi et al.,1996), and with the partial blockade of mossy fiber-LTD by mGluR2/3 antagonists (Kobayashi et al.,1996; Tzounopoulos et al.,1998).
The presence of 4 mM Ca2+ and 4 mM Mg2+ during induction alone is sufficient to rescue mossy fiber LTD in Grm2/3 dko mice
To examine whether the higher divalent cation concentration is required during induction and/or expression of mossy fiber LTD, we established baseline recordings in Grm2/3 dko mice in 2 mM Ca2+ and 2 mM Mg2+, then switched to 4 mM Ca2+ and 4 mM Mg2+ for the duration of the 1 Hz paired-pulse stimulation, before switching back to the 2 mM conditions. This protocol induced robust depression (42.5 ± 5% depression, n = 5 slices/5 animals; t(4) = 13.69; P < 0.001) of a magnitude similar to that seen with 4 mM Ca2+ and 4 mM Mg2+ aCSF present throughout. The presence of the high divalent cation solution during induction alone is therefore sufficient to rescue mossy fiber LTD (Fig. 2).
To control for the possibility that the 4 mM condition could by itself induce LTD, we repeated the same 2–4–2 mM aCSF exchanges in interleaved experiments but omitted the 1 Hz stimulation, continuing instead at baseline frequency. In the absence of 1 Hz stimulation, however, no depression was seen (mean fEPSP slope 41–45 min after the transition from 2 mM to 4 mM: 105.0 ± 13.5% of baseline response, n = 5 slices/4 animals; t(7) = 4.24; P = 0.004 compared with response over the same time period in the presence of 1 Hz stimulation, see above; Fig. 2). This shows that the Ca2+ sensitive depression induced by 1 Hz paired-pulse stimulation is likely to be genuine activity-dependent mossy fiber LTD.
Grm2/3 dko LTD is NMDAR-independent
To verify that the LTD induced in the Grm2/3 dko mice did not reflect NMDA receptor-dependent LTD in a contaminating AC response, we applied 20 μM D-AP5 to slices before beginning baseline recording in 4 mM Ca2+ and 4 mM Mg2+ aCSF. The magnitude of LTD did not differ in the presence and absence of D-AP5 (D-AP5 mean: 47.9 ± 5.0% depression, n = 4 slices/4 animals; control mean: 55.6 ± 7.8% depression, n = 5 slices/4 animals; t <1; P > 0.20) (Supporting Information Fig. S2), confirming that the response is likely to comprise a genuine mossy fiberLTD.
Baseline transmission and short-term plasticity in Grm2/3 dko mice
To investigate which mechanisms might contribute to the different induction requirements of mossy fiber LTD in Grm2/3 dko mice, we compared baseline transmission and short-term plasticity in Grm2/3 dko mice and wild-type control animals in 2 mM and 4 mM conditions. Basal synaptic transmission as estimated with input/output curves did not differ between wild-type and Grm2/3 dko in either condition (Figs. 3A,B). A small but significant reduction in PPR was observed in Grm2/3 dko mice compared with wild-type control animals in 2 mM, but not 4 mM conditions (2 mM Ca2+/2 mM Mg2+: wt mean PPR, 1.94 ± 0.14, n = 9 slices/7 animals; Grm2/3 dko mean PPR, 1.61 ± 0.04, n = 8 slices/6 animals; t(11) = 2.15; P = 0.041. 4 mM Ca2+/4 mM Mg2+: wt mean PPR, 1.80 ± 0.08, n = 9 slices/7 animals; Grm2/3 dko mean PPR, 1.73 ± 0.04, n = 8 slices/6 animals; t(11) < 1; P > 0.20) (Figs. 3C,D). Frequency facilitation was strongly reduced in Grm2/3 dko mice in both 2 mM and 4 mM conditions (2 mM Ca2+/2 mM Mg2+: wt mean: 180.3 ± 46.1%, n = 8 slices/7 animals; Grm2/3 dko mean: 37.3 ± 11.7%, n = 8 slices/6 animals; t(11) = 2.79; P = 0.018; Fig. 1B; 4 mM Ca2+/4 mM Mg2+: wt mean, 110.5 ± 4.3%, n = 8 slices/8 animals; Grm2/3 dko mean, 26.8 ± 1.2%, n = 9 slices/5 animals; t(11) = 3.28; P = 0.008; Fig. 1A). This reduced frequency facilitation in both 2 mM and 4 mM conditions appears to dissociate frequency facilitation from the induction of LTD, which was absent in 2 mM, but present in 4 mM, conditions.
Antagonists at group II mGluRs fail to block mossy fiber LTD in wild-type mice
To test whether this group II mGluR-independent form of mossy fiber LTD can also be induced in wild-type animals, we first tested the nonspecific group I/group II mGluR antagonist (RS)-MCPG. 1.5 mM MCPG blocked the effect of DCG-IV in both 2 mM and 4 mM conditions (94 ± 8%, n = 8 and 93 ± 10%, n = 7, respectively). Similar to mossy fiber LTD in Grm2/3 dko mice, at 4 mM Ca2+ and Mg2+, MCPG failed to block the induction of mossy fiber LTD (control mean, 28 ± 6% depression, n = 8; MCPG mean, 20 ± 7% depression, n = 7; n.s.; Fig. 4A), whereas at 2 mM Ca2+ and Mg2+, MCPG significantly reduced mossy fiber LTD (control mean, 31 ± 7% depression, n = 7; MCPG mean, 7 ± 6% depression, n = 7; t(12) = 7.477; P < 0.01; Fig. 4B). In contrast, the selective group II mGluR antagonist 3 μM LY341495 (Kingston et al.,1998), which also blocked the effect of DCG-IV (110 ± 8%, n = 9 and 114 ± 10%, n = 7), failed to block mossy fiber LTD under both conditions (4 mM Ca2+/4 mM Mg2+: control mean, 27 ± 5% depression, n = 7; LY mean, 22 ± 6%, n = 7; n.s.; Fig. 4C; 2 mM Ca2+/2 mM Mg2+: wt mean, 23 ± 11%, n = 7; LY mean, 22 ± 6%, n = 7; n.s.; Fig. 4D), consistent with the recent report by Wostrack and Dietrich (2009). These results support the conclusion that group II mGluRs are not necessary for the induction of mossy fiber LTD.
Grm2/3 dko mossy fiber LTD depends on activation on non-NMDA ionotropic glutamate receptors
Since mossy fiber LTD in Grm2/3 dko mice is independent of NMDA receptors, but is Ca2+ dependent and requires synaptic activity, we asked whether non-NMDA ionotropic glutamate receptors might be necessary during induction. To this end, we monitored the NMDA receptor-mediated component of the EPSC during mossy fiber LTD, using whole-cell voltage-clamp recording from CA3 neurons. To distinguish between AMPA and kainate receptors, we used the selective AMPA receptor antagonist, 30 μM GYKI53655, to block the AMPAR component of the EPSC, and the combined AMPA/kainate receptor antagonist, 3 or 5 μM NBQX, toadditionally block kainate receptors. Mossy fiber LTD of the NMDAR-mediated EPSC was seen in both wild-type and Grm2/3 dko mice in the presence of GYKI53655, showing that AMPA receptors are not necessary for induction of group II mGluR-independent mossy fiber LTD in 4 mM Ca2+/4 mM Mg2+ conditions (dko mean: 32 ± 4% depression, n = 8) (Fig. 5A). Similar to the results with fEPSPs, however, mossy fiber LTD of the NMDA receptor-mediated EPSC was greatly reduced in Grm2/3 dko mice in 2 mM Ca2+ and 2 mM Mg2+ (wt mean: 30 ± 7% depression, n = 6; Grm2/3 dko mean: 5 ± 4% depression, n = 6; P < 0.01). To additionally block kainate receptors in Grm2/3 dko mice in 4 mM conditions, we added the AMPA/kainate receptor antagonist, 3 or 5 μM NBQX. In the presence of this antagonist, mossy fiber LTD of the NMDA receptor-mediated EPSC was no longer induced in Grm2/3 dko mice (dko mean: 2 ± 7% depression, n = 7) (Fig. 5B), suggesting that Grm2/3 dko mossy fiber LTD is likely to be mediated by kainate receptors. Mossy fiber LTD in Grm2/3 dko mice was not, however, blocked by postsynaptic BAPTA (dko mean: 38 ± 4% depression, n = 5) (Fig. 5B), confirming that the LTD is presynaptic in origin. As with field EPSPs, frequency facilitation was reduced in Grm2/3 dko mice compared with wild-type controls under both sets of conditions (2 mM: wt mean, 400 ± 47%, n = 6; dko mean, 189 ± 29%, n = 6; P < 0.01; 4 mM: wt mean, 423 ± 56%, n = 10; dko mean, 210 ± 35%, n = 8, P < 0.01).
This study has revealed that group II mGluRs are not obligatory for the induction of mossy fiber LTD, since a Ca2+ sensitive form of mossy fiber LTD can be induced in Grm2/3 dko mice and antagonists at group II mGluRs fail to block the induction of mossy fiber LTD. Robust mossy fiber LTD could be induced in 4 mM Ca2+, while mossy fiber LTD was greatly reduced in Grm2/3 dko animals in 2 mM Ca2+. This group II mGluR-independent, Ca2+ sensitive form of LTD was not blocked by D-AP5, confirming that it did not comprise NMDA receptor-dependent LTD in contaminating AC fibers. Using whole-cell recordings, we confirmed that the LTD in Grm2/3 dko mice was not blocked by postsynaptic BAPTA, indicating that it was presynaptic in origin. Finally, we revealed that the Grm2/3 dko mossy fiber LTD was dependent on non-NMDA ionotropic glutamate receptors, most likely kainate receptors, since LTD of the NMDA receptor-mediated component of the EPSC was blocked by the AMPA/kainate receptor antagonist, NBQX, but not by the selective AMPA receptor antagonist, GYKI53655.
Mossy fiber LTD induction has been attributed to group II mGluRs largely on the basis of pharmacological data. Thus, mGluR2/3 antagonists significantly reduced mossy fiber LTD (Kobayashi et al.,1996; Tzounopoulos et al.,1998), while mGluR agonists induced depression of mossy fiber responses that was not reversed by subsequent application of an mGluR antagonist (Tzounopoulos et al.,1998). Moreover, mossy fiber LTD was greatly reduced in Grm2 knockout mice (Yokoi et al.,1996). On the other hand, even broad-spectrum mGluR antagonists did not fully block mossy fiber LTD, and a recent study using the potent and selective mGluR2/3 antagonist, LY341495, reported no significant effect of this drug on mossy fiber LTD under 2 mM conditions (Wostrack and Dietrich,2009). Our results confirm the failure of LY341495 to block mossy fiber LTD under both 2 mM and 4 mM conditions, whereas MCPG blocked mossy fiber LTD under 2 mM but not 4 mM conditions. We have no explanation for this difference in effect of LY341495 and MCPG. Our data show that robust mossy fiber LTD can be induced in Grm2/3 dko mice in 4 mM Ca2+ and Mg2+, although they display greatly reduced mossy fiber LTD under 2 mM Ca2+, in agreement with Yokoi et al. (1996). The LTD observed in Grm2/3 dko and wild-type mice was not blocked by D-AP5, and is therefore unlikely to reflect LTD in recurrent collaterals, which evidence suggests is NMDA receptor dependent (Debanne et al.,1998; Unni et al.,2004).
What is the mechanism underlying the activity-dependent Ca2+-sensitive form of mossy fiber LTD in Grm2/3 dko mice? Our evidence suggests that presynaptic non-NMDA ionotropic glutamate receptors, most likely kainate receptors, mediate this form of LTD. One possibility is that these receptors could couple to the adenylate cyclase/PKA signaling cascade implicated in mossy fiber LTD (see Nicoll and Schmitz,2005), for which there is some evidence (Rodríguez-Moreno and Sihra,2007). However, the Ca2+-sensitivity of this form of LTD might rather suggest that Ca2+ permeable presynaptic kainate receptors and/or voltage-activated Ca2+ channels are involved. Such mechanisms have been shown to contribute to the kainate receptor-dependent component of frequency facilitation at this synapse (Lauri et al.,2003; Schmitz et al.,2003), and it is known that presynaptic Ca2+ dependent mechanisms contribute to LTD (Kobayashi et al.,1999; Tzounopoulos et al.,1998). Thus, it is possible that a Ca2+-dependent presynaptic process, such as activation of a Ca2+-dependent phosphatase, could act in concert with reduced PKA activity during induction of LTD, and that this Ca2+ dependent process might be sufficient during elevated Ca2+ in the absence of mGluR2/3. Another possibility is that presynaptic Ca2+ is the primary mediator of mossy fiber LTD, and that mGluR signaling has a permissive role, gating the Ca2+ dependent process. Further experiments are required to elucidate the nature of this Ca2+ dependent process.
Functional redundancy between mGlu2/3 and kainate receptors might also operate at the level of short-term plasticity. Frequency facilitation was reduced in Grm2/3 dko mice. Given that basal synaptic transmission was unchanged under both sets of recording conditions, this may indicate that mGluR2/3 are predominantly phasically rather than tonically activated (Kew et al.,2001,2002). Antagonism of presynaptic kainate receptors has likewise been shown to reduce frequency facilitation (Breustedt and Schmitz,2004; Lauri et al.,2001) with knockout mouse studies suggesting that the effect may be mediated specifically by GluK2 (Breustedt and Schmitz,2004; Contractor et al.,2001). Antagonism of adenosine A1 receptors also reduces frequency facilitation (Klausnitzer and Manahan-Vaughan,2008; Moore et al.,2003); these receptors might contribute to the residual frequency facilitation observed in Grm2/3 dko mice in the presence of a kainate receptor antagonist. The reduced frequency facilitation in both 2 mM and 4 mM conditions is in contrast to the induction of LTD, which was absent at 2 mM, but present at 4 mM Ca2+.
Non-NMDA glutamate receptor-dependent Ca2+ sensitive mossy fiber LTD might reflect a compensatory response in Grm2/3 dko mice to the absence of group II mGluRs. Consistent with this possibility, Grm2 and Grm3 knockout mice display changes in glutamate receptor and transporter expression (Lyon et al.,2008). Nevertheless, data indicating that group II mGluRs are not sufficient for mossy fiber LTD induction were first obtained in wild-type animals (Kobayashi et al.,1999; Tzounopoulos et al.,1998; Wostrack and Dietrich,2009), as were data suggesting redundancy of mGluR2/3 and kainate receptor function (Negrete-Díaz et al.,2006,2007). The present results show that group II mGluRs are not obligatory for induction of mossy fiber LTD, since robust LTD could be induced in Grm2/3 dko mice. This result may also help explain why the behavioral phenotype of Grm2 knockout mice is mild (Yokoi et al.,1996).
The authors thank Prof. Shigemoto Nakanishi for generous provision of the Grm2 knockout mice.