Many central excitatory synapses undergo developmental alterations in the molecular and biophysical characteristics of postsynaptic ionotropic glutamate receptors via changes in subunit composition. Concerning AMPA receptors (AMPARs), glutamate receptor 2 subunit (GluR2)-containing, Ca2+-impermeable AMPARs (CI-AMPARs) prevail at synapses between mature principal neurons; however, accumulating evidence indicates that GluR2-lacking, Ca2+-permeable AMPARs (CP-AMPARs) contribute at these synapses early in development. Here, we used a combination of imaging and electrophysiological recording techniques to investigate potential roles for CP-AMPARs at developing hippocampal mossy fiber–CA3 pyramidal cell (MF–PYR) synapses. We found that transmission at nascent MF–PYR synapses is mediated by a mixed population of CP- and CI-AMPARs as evidenced by polyamine-dependent inwardly rectifying current–voltage (I–V) relationships, and partial philanthotoxin sensitivity of synaptic events. CP-AMPAR expression at MF–PYR synapses is transient, being limited to the first 3 postnatal weeks. Moreover, the expression of CP-AMPARs is regulated by the PDZ (postsynaptic density-95/Discs large/zona occludens-1) domain-containing protein interacting with C kinase 1 (PICK1), because MF–PYR synapses in young PICK1 knock-out mice are philanthotoxin insensitive with linear I–V relationships. Strikingly, MF–PYR transmission via CP-AMPARs is selectively depressed during depolarization-induced long-term depression (DiLTD), a postsynaptic form of MF–PYR plasticity observed only at young MF–PYR synapses. The selective depression of CP-AMPARs during DiLTD was evident as a loss of postsynaptic CP-AMPAR-mediated Ca2+ transients in PYR spines and reduced rectification of MF–PYR synaptic currents. Preferential targeting of CP-AMPARs during DiLTD is further supported by a lack of DiLTD in young PICK1 knock-out mice. Together, these findings indicate that the transient participation of CP-AMPARs at young MF–PYR synapses dictates the developmental window to observe DiLTD.
The glutamate receptor 2 (GluR2) subunit dictates several AMPAR biophysical properties and interacts with various molecules implicated in receptor trafficking (Dingledine et al., 1999; Collingridge et al., 2004; Isaac et al., 2007). Typically GluR2 mRNA undergoes efficient editing at the “Q/R site” (Hollmann et al., 1991; Sommer et al., 1991; Verdoorn et al., 1991; Burnashev et al., 1992; Seeburg et al., 1998), and incorporation of these edited subunits into AMPARs results in low-conductance, Ca2+-impermeable channels (CI-AMPARs) with relatively linear I–V relationships. In contrast, AMPARs lacking edited GluR2 have higher conductance, are readily Ca2+ permeable (CP-AMPARs), and exhibit inwardly rectifying I–V relationships caused by voltage-dependent channel block by intracellular polyamines (Bowie and Mayer, 1995; Donevan and Rogawski, 1995; Kamboj et al., 1995; Koh et al., 1995). Thus, the presence or absence of GluR2 greatly influences AMPAR and, hence, synaptic function.
The ubiquitous expression and efficient editing of GluR2 within principal neurons ensures that CI-AMPARs dominate transmission between excitatory neurons throughout the nervous system. However, recent findings indicate GluR2-lacking, CP-AMPARs are assembled in principal neurons and become synaptically incorporated under certain conditions (Aizenman et al., 2002; Kumar et al., 2002; Eybalin et al., 2004; Ju et al., 2004; Terashima et al., 2004; Bagal et al., 2005; Harms et al., 2005; Ogoshi et al., 2005; Shin et al., 2005; Thiagarajan et al., 2005; Clem and Barth, 2006; Plant et al., 2006; Sutton et al., 2006). One compelling line of evidence suggests that CP-AMPARs are transiently expressed by principal neurons early in development. For example, synaptic CP-AMPARs occur in layer 5 pyramids until postnatal day 16 after which CI-AMPARs dominate transmission (Kumar et al., 2002). Similar developmental switches happen at inner hair cell synapses (Eybalin et al., 2004), Xenopus retinotectal synapses (Aizenman et al., 2002), and chicken forebrain synapses (Migues et al., 2007). Moreover, developmental increases in the ratio of GluR2 to other AMPAR subunits occur throughout the CNS (Pellegrini-Giampietro et al., 1992; Pickard et al., 2000; Zhu et al., 2000). Such transient CP-AMPAR expression could provide a potentially important route for Ca2+ entry during synapse maturation and impart distinct short/long-term plastic properties to developing networks (Rozov and Burnashev, 1999; Liu and Cull-Candy, 2000).
In the hippocampus, mossy fiber–CA3 pyramidal cell (MF–PYR) connections mature entirely postnatally providing an attractive model to probe for AMPAR changes during synapse development in a defined central pathway (Amaral and Dent, 1981; Henze et al., 2000). Moreover, MF–PYR synapses constitutively insert overexpressed GluR1 homomers in contrast to synapses between PYRs in which such recombinant CP-AMPARs are excluded unless driven by CaMKII (calcium/calmodulin-dependent protein kinase II) (Kakegawa et al., 2004), suggesting that MF–PYR synapses are inherently permissive sites for CP-AMPARs. Hence, in the current study we investigated whether native CP-AMPARs contribute to basal transmission and synaptic plasticity at developing MF–PYR synapses.
Materials and Methods
Hippocampal slice preparation.
Slices were prepared as described previously (Pelkey et al., 2005; Topolnik et al., 2005) using postnatal day 10 (P10) to P30 Sprague Dawley rats or C57BL/6 mice as indicated throughout the paper, except for experiments involving transgenic animals, which were created on a hybrid genetic background of 129 and C57BL/6 strains (Gardner et al., 2005). Briefly, animals were anesthetized with isoflurane and decapitated allowing removal of the brain into ice-cold saline solution containing the following (in mm): 130 NaCl, 24 NaHCO3, 3.5 KCl, 1.25 NaH2PO4, 0.5 CaCl2, 5.0 MgCl2 (or MgSO4), and 10 glucose, saturated with 95% O2 and 5% CO2, pH 7.4. In some cases, NaCl was substituted with sucrose (250 mm). After dissection of the brain, individual hemispheres were transferred to the stage of a VT-1000S vibratome (Leica Microsystems, Bannockburn, IL) and sectioned to yield transverse hippocampal slices (300 μm), which were incubated in the above solution at 35°C for recovery until use (minimum 1 h recovery). All animal procedures conformed to the National Institutes of Health and Université de Montréal animal welfare guidelines.
Individual slices were transferred to a recording chamber and perfused (2–3 ml/min) with extracellular solution composed of the following (in mm): 130 NaCl, 24 NaHCO3, 3.5 KCl, 1.25 NaH2PO4, 2.5 CaCl2, 1.5 MgCl2 (or MgSO4), 10 glucose, 0.0025 bicuculline methobromide, and 0.05–0.100 dl-AP5, saturated with 95% O2 and 5% CO2, pH 7.4, 32–35°C. Whole-cell patch-clamp recordings using multiclamp 700A or Axopatch 200B amplifiers (Molecular Devices, Foster City, CA) in voltage-clamp mode (Vh, −60 mV) were made from individual CA3 pyramidal neurons, visually identified with infrared video microscopy and differential interference contrast optics. For Ca2+ imaging recordings, electrodes (4–5 MΩ) pulled from borosilicate glass (World Precision Instruments, Sarasota, FL) were filled with intracellular solution (ICS) composed of the following (in mm): 130 CsMeSO3, 5 CsCl, 2 MgCl2, 5 diNa-phosphocreatine, 10 HEPES, 2 ATP2Na, 0.4 GTPNa, 0.1 spermine, 1 lidocaine N-ethyl bromide (QX314), and 0.05 OGB1 (Oregon Green 488 BAPTA-1, hexapotassium salt) (Invitrogen, Eugene, OR), pH 7.2–7.3, 275–285 mOsm. For all other recordings, electrodes were filled with ICS of the following composition (in mm): 100 Cs-gluconate, 5 CsCl, 0.6 EGTA, 5 MgCl2, 8 NaCl, 2 Na2ATP, 0.3 NaGTP, 40 HEPES, 1 QX314, and 0.1 spermine, pH 7.2–7.3 and 290–300 mOsm. In experiments in which reagents were delivered postsynaptically through the recording pipette, we waited a minimum of 15 min after establishing the whole-cell configuration before any attempt was made to induce plasticity to ensure adequate diffusion of reagents into proximal dendritic compartments. The generation of the EVKI (KKEGYNVYGIEEVKI) and SGKA (KKEGYNVYGIESGKA) peptides has been described previously (Gardner et al., 2005). These peptides (200 μm) were applied together with a peptidase inhibitor mixture containing leupeptin, bestatin, and pepstatin (all at 0.1 mm) to prevent degradation. The EVKI peptide serves as a GluR2 c-tail phosphomimic and selectively disrupts GluR2–protein interacting with C kinase 1 (PICK1) postsynaptic density-95/Discs large/zona occludens-1 (PDZ) domain-mediated interactions, and the SGKA peptide is a GluR2 c-tail mutant peptide that does not function as a PDZ binding motif. Data were not corrected for liquid junction potentials. Uncompensated series resistance (8–15 MΩ) was rigorously monitored by the delivery of small voltage steps at regular intervals, and recordings were discontinued after changes of >15%.
Synaptic responses were typically evoked at 0.33 Hz by low-intensity microstimulation (100 μs duration; 20–40 μA intensity) via a constant-current isolation unit (A360; World Precision Instruments, Sarasota, FL) connected to a patch electrode filled with oxygenated extracellular solution. For MF–PYR synaptic recordings, the stimulating electrode was placed in either the dentate gyrus or stratum lucidum. Except in Ca2+ imaging recordings, MF origin of EPSCs was always confirmed by perfusion of slices with the group II metabotropic glutamate receptor (mGluR) agonist 2S,2′R,3′R-2-(2′,3′-dicarboxy-cyclopropyl)glycine (DCGIV) (1 μm) (Kamiya et al., 1996). For PYR collateral fiber stimulation, the stimulating electrode was placed in CA3 stratum radiatum and evoked events did not exhibit DCGIV sensitivity. MF–PYR depolarization-induced long-term depression (DiLTD) was induced by depolarizing the postsynaptic CA3 pyramidal cell from −60 to −10 mV for 5 min in the absence of stimulation (Lei et al., 2003). Data for I–V curves were obtained by varying the postsynaptic membrane potential between −60 and +40 mV in steps of 20 mV.
For each recording, EPSC amplitudes from individual sweeps were measured during a 1–2 ms window around the peak of the waveform. Then, for incorporation into group data, the EPSC amplitudes for a given recording were binned as consecutive 1 min averages and normalized to the mean amplitude obtained during the baseline period (first 3–10 min). Data are presented as means ± SEMs unless otherwise indicated. Statistical significance was assessed using both parametric (paired and unpaired t tests) and nonparametric (Mann–Whitney or Wilcoxon signed-rank tests) statistical analyses as appropriate. Unless otherwise indicated, p values reported throughout the manuscript correspond to values obtained using nonparametric analyses. For some recordings, paired-pulse ratios (PPRs) were obtained by calculating mean P2/mean P1, where P1 represents the amplitude of the first evoked current and P2 the amplitude of the second for 20 consecutive individual traces. For I–V relationships, a minimum of five events was averaged for each holding potential. To determine rectification indexes (RIs), the negative portion of individual I–V relationships (between Vh of −60 and 0 mV) was fit by a linear regression, and then RI was calculated as the ratio of the actual current amplitude observed at +40 mV to the predicted value at +40 mV based on the linear regression. Alternatively, in experiments in which the RI was monitored before and after DiLTD, RI was determined as the absolute value of the ratio of the mean current amplitude at +40 mV to the mean current amplitude at −60 mV (RIEPSC+40/EPSC−60).
Ca2+ imaging of thorny excrescences.
After obtaining the whole-cell configuration, 20–30 min were allowed for intracellular diffusion of the fluorophore. Imaging was performed using a two-photon laser scanning microscope LSM 510 (Carl Zeiss, Kirkland, Quebec, Canada) based on a mode-locked Ti:sapphire laser operating at 800 nm, 76 MHz pulse repeat, <200 fs pulse width, and pumped by a solid-state source (Mira 900 and 5 W Verdi argon ion laser; Coherent, Santa Clara, CA). A long-range water-immersion objective (40×; numerical aperture, 0.8) was used. Fluorescence was detected through a short-pass filter (cutoff, 680 nm) in non-descanned detection mode and images were acquired using the LSM 510 software (Carl Zeiss). Ca2+ transients (average of four responses) evoked by local electrical stimulation were measured at 20–40 μm from the soma by scanning a line through the thorny excrescence. Changes in fluorescence were calculated relative to the baseline and expressed as %ΔF/F = [(F − Frest)/Frest] × 100. Summary data are expressed as mean ± SE.
Immunogold electron microscopy.
Immunogold studies were done as described previously (Petralia and Wenthold, 1999; Petralia et al., 1999; Rouach et al., 2005). Briefly, mice were perfused with 4% paraformaldehyde plus 0.5% glutaraldehyde and sections were frozen in liquid propane in a Leica CPC and freeze-substituted into Lowicryl HM-20 in a Leica AFS. Ultrathin sections were incubated in 0.1% sodium borohydride plus 50 mm glycine in Tris-buffered saline plus 0.1% Triton X-100 (TBST), and then in 10% normal goat serum (NGS) in TBST, and then in primary antibody in 1% NGS/TBST, followed by immunogold in 1% NGS/TBST/0.5% polyethylene glycol (molecular weight, 20,000), and then stained with uranyl acetate and lead citrate. Figures were processed with Adobe Photoshop with minimal use of levels, brightness, and contrast used uniformly over the images. Double immunogold labeling used a rabbit polyclonal antibody to GluR1 (Petralia and Wenthold, 1992) with 5 nm immunogold, and a mouse monoclonal antibody to GluR2 (Steinberg et al., 2006) with 15 nm immunogold. Control sections where primary antibodies were omitted did not produce significant 5 and 15 nm gold labeling. Micrographs were taken in the CA3 stratum lucidum at 20,000× for identification of mossy terminal/spine associations and at 40,000× for quantitative analysis. Mossy fiber terminals were identified as large, irregular structures with many round synaptic vesicles contacting complex spines (thorny excrescences) originating from CA3 pyramidal cell apical dendrites. Identification and characterization of these structures have been described in great detail (Petralia and Wenthold, 1992; Darstein et al., 2003). Gold particles in both the synaptic cleft and postsynaptic density were considered membrane associated and synaptic. At extrasynaptic sites of spine profile membranes, only gold particles within 20 nm of the membrane were considered membrane associated [based on the resolution of the immunogold method (Rouach et al., 2005)]. We examined total surface labeling by combining the synaptic and extrasynaptic counts on each spine profile. Then, for each spine profile, we compared synapses in which 5 nm gold labeling for GluR1 was >100 nm from any 15 nm gold particle (“5 nm only”) to those in which 5 and 15 nm gold particles were <100 nm apart (“5 + 15 nm”). The former group was intended to represent synaptic AMPA receptors that were more likely to lack GluR2 than the latter group; 100 nm was selected arbitrarily for the sake of comparison between the wild-type and mutant mice. Immunogold labeling is expected only to represent some fraction of total receptor labeling (Petralia et al., 1999).
Transient expression of CP-AMPARs at developing MF–PYR synapses
AMPARs lacking edited GluR2 subunits exhibit pronounced inward rectification caused by voltage-dependent block of the channel pore by intracellular polyamines, providing a convenient electrophysiological signature to assess the contribution of CP-AMPARs to synaptic transmission (Bowie and Mayer, 1995; Donevan and Rogawski, 1995; Kamboj et al., 1995; Koh et al., 1995). Thus, to initially probe for the presence of CP-AMPARs at developing MF–PYR synapses, we examined I–V relationships of pharmacologically isolated (50–100 μm dl-APV) AMPAR-mediated MF–PYR EPSCs in slices from young (P10–P17) mice. At this stage of development, using an intracellular recording solution (ICS) supplemented with the polyamine spermine (100 μm), we consistently observed significant rectification in MF–PYR EPSC I–V relationships (Fig. 1A, top panel). The degree of rectification for each recording was quantified by calculating the ratio of the actual EPSC value observed at a holding potential (Vh) of +40 mV to that predicted by a linear fit of the I–V relationship at holding potentials between −60 and 0 mV (see Materials and Methods). For MF–PYR synapses in slices obtained from P10–P17 mice, this RI ranged from 0.12 to 0.69 (mean ± SEM, 0.45 ± 0.05; n = 13) (Fig. 1C,D). Importantly, synaptic events were inhibited by >90% during treatment with the AMPAR-specific antagonist GYKI-53655 [1-(4-aminophenyl)-3-methylcarbamyl-4-methyl-7,8-methylenedioxy-3,4-dihydro-5H-2,3-benzodiazepine] (40 μm) (Fig. 1A, top panel, inset) eliminating the possibility that rectification resulted from KAR (kainate receptor) contamination of EPSCs. It is unlikely that rectification resulted from inadequate voltage clamp of MF–PYR synapses as AMPAR-mediated EPSCs at more distally located collateral fiber–PYR synapses displayed near-linear I–V relationships, yielding significantly greater RIs (1.29 ± 0.08; n = 5) in slices of comparable age (Fig. 1D). Moreover, linear I–V relationships with RI values close to unity were obtained in control MF–PYR recordings from young mice using ICS not supplemented with spermine (RI, 0.99 ± 0.08; n = 7) (Fig. 1D), consistent with a specific role for polyamines in mediating the rectification. The spermine-dependent rectification of EPSCs suggests that CP-AMPARs participate in transmission at developing MF–PYR synapses. Such a contribution of CP-AMPARs to transmission at nascent MF–PYR contacts was subsequently confirmed when we observed partial inhibition of MF–PYR EPSCs by the CP-AMPAR specific antagonist philanthotoxin-433 (PhTx) (2 μm): with 15 min of PhTx perfusion, EPSCs were significantly depressed to 73 ± 4.9% of control (n = 16; p = 0.004) (Fig. 1B, top panel; C). Considered together the range of RI values observed and the partial PhTx sensitivity of MF–PYR EPSCs in slices from P10–P17 mice demonstrates that transmission is mediated by a mixed population of both CP- and CI-AMPARs early in development.
Previous investigations did not observe significant rectification (Jonas et al., 1993) or PhTx sensitivity (Toth et al., 2000) of MF–PYR synaptic events in acute rat hippocampal slices. Importantly, this discrepancy with the current findings is not attributable to species differences (Bellone and Luscher, 2005, 2006) because we also observed rectification of MF–PYR EPSCs in acute slices from young (P15–P17) rats (RI, 0.53 ± 0.04; n = 8). Although a lack of spermine supplementation may have contributed to the previous failure to observe rectification (Jonas et al., 1993), it is also possible that the contribution of CP-AMPARs at MF–PYR synapses decreases with development as the age range used in the previous studies (P15–P25) was skewed toward animals older than those used in our recordings described above. Indeed, when we examined MF–PYR transmission in slices from older mice (>P17), I–V relationships became progressively more linear with increasing age, yielding significantly larger RI values (Fig. 1A, bottom, C). Similarly, slices from older rats (>P23) yielded significantly greater MF–PYR RIs (0.80 ± 0.06; n = 7; p = 0.006 vs recordings from P15–P17 rat slices). Furthermore, we did not observe any PhTx sensitivity of MF–PYR synaptic events in slices from older mice (Fig. 1B, bottom panel). Thus, the contribution of native CP-AMPARs to MF–PYR transmission is transient, being limited to early developmental time points with subsequent replacement by CI-AMPARs as reported for various other synapses between principal neurons (Aizenman et al., 2002; Kumar et al., 2002; Eybalin et al., 2004).
As an independent test for the participation of CP-AMPARs at developing MF–PYR synapses, we attempted to observe postsynaptic CP-AMPAR-mediated Ca2+ transients (CaTs) at MF–PYR contacts in acute hippocampal slices obtained from young (P10–P16) rats (Fig. 2). PYRs were filled with the fluorescent Ca2+ indicator OGB-1 through a whole-cell recording electrode and imaged using two-photon laser-scanning microscopy (Fig. 2A). After an initial dye loading period (20–30 min), MF–PYR postsynaptic elements were identified as large, complex, spine-like protrusions (thorny excrescences) along the proximal apical dendrite (Fig. 2B) (Chicurel and Harris, 1992; Reid et al., 2001, 2004). Subsequently, postsynaptic CaTs evoked by local microstimulation within stratum lucidum were monitored within these developing thorny excrescences in line-scanning mode (Fig. 2B–D). To maximize the likelihood of observing postsynaptic CaTs, recordings were started with both AMPAR- and NMDAR-mediated transmission intact. Strikingly, in slices from young rats (<P17), MF–PYR CaTs were not significantly affected by NMDAR inhibition but were strongly depressed by subsequent inhibition of CP-AMPARs: CaTs were 89 ± 11% of control after 10 min of d-APV (50 μm) perfusion and further decreased to 21 ± 4% of control after 15 min of PhTx (2 μm) applied with constant synaptic stimulation (0.5Hz) to facilitate use-dependent block of CP-AMPARs (Fig. 2C,E). Similarly, simultaneously monitored EPSCs from these recordings were not significantly affected by d-APV (85 ± 7% of control; p = 0.9) but exhibited partial sensitivity to PhTx (70 ± 7% of control; p = 0.04) (Fig. 2C). Although these compound EPSCs represent not only activity at the imaged synapses but also activity at other synapses not under observation, the findings are consistent with those reported above for young mice (compare Fig. 1B). In slices from older animals (>P23), MF–PYR CaTs were almost entirely inhibited by d-APV (17 ± 5% of control), indicating that NMDARs are the primary mediators of postsynaptic Ca2+ influx at more mature MF–PYR synapses with little or no contribution of CP-AMPARs (Fig. 2D,E). Together, the results from our imaging experiments confirm that CP-AMPARs transiently participate in transmission greatly contributing to postsynaptic Ca2+ dynamics at developing MF–PYR synapses.
CP-AMPARs are preferentially downregulated during MF–PYR DiLTD
In cerebellar stellate cells and dopaminergic neurons of the ventral tegmental area (VTA), CP-AMPARs impart a distinct form of long-term plasticity in which CP-AMPAR-mediated transmission is persistently lost and synapses become dominated by CI-AMPARs (Liu and Cull-Candy, 2000; Bellone and Luscher, 2005; Gardner et al., 2005). In both cases, plasticity is triggered by a rise in postsynaptic Ca2+ mediated by influx through the CP-AMPARs themselves or group I mGluR activation during high-frequency stimulation. Recently, we described a novel postsynaptic form of LTD at MF–PYR synapses that is activity independent, being induced simply by transient postsynaptic depolarization (Lei et al., 2003). This DiLTD is expressed as a persistent depression of AMPAR function and, like the plasticities described above, is triggered by a rise in postsynaptic Ca2+, although in this case via influx through L-type voltage-gated Ca2+ channels (VGCCs). Importantly, this DiLTD exhibits a developmental profile similar to that observed for the expression of CP-AMPARs at MF–PYR synapses: DiLTD is robust at nascent MF–PYR synapses during the second postnatal week, gradually declines during the third postnatal week, and is absent beyond P35 (Lei et al., 2003). Thus, given the overlapping developmental windows for CP-AMPAR expression and DiLTD competence as well as the role of postsynaptic Ca2+ in triggering DiLTD, we next investigated a potential role for modulation of CP-AMPARs in DiLTD.
If MF–PYR DiLTD proceeds as a loss of synaptic CP-AMPARs, similar to CP-AMPAR-associated LTD in the cerebellum and VTA (Liu and Cull-Candy, 2000; Bellone and Luscher, 2005; Gardner et al., 2005), then stimulus evoked synaptic Ca2+ influx should be reduced, resulting in a persistent depression of MF–PYR CaTs. Therefore, we initially investigated the effects of DiLTD induction on MF–PYR CaTs in slices from P10–P16 rats (Fig. 3). Synaptic events were evoked at 0.33 Hz and associated CaTs within thorny excrescences were monitored at 1–5 min intervals to minimize potential photodamage. After a stable baseline recording period, presynaptic stimulation was interrupted and the postsynaptic cell was transiently depolarized from a holding potential of −60 to −10 mV for 5 min (Lei et al., 2003) (see Materials and Methods). Subsequently, the postsynaptic membrane potential was returned to −60 mV, presynaptic stimulation was resumed, and the resulting MF–PYR CaTs were recorded for comparison to baseline responses. Using this protocol, we obtained stable CaTs during the baseline recording period; however, immediately after DiLTD induction, a persistent depression of MF–PYR CaTs was evident: CaTs were 36 ± 7.6% of baseline responses immediately after DiLTD induction and ultimately depressed to 6.5 ± 3.4% of baseline responses within 25 min after induction (n = 5) (Fig. 3A). This LTD of MF–PYR CaTs was paralleled by a persistent inhibition of simultaneously recorded EPSCs (to 73 ± 8.8% of control at 25 min after induction; p = 0.03) (Fig. 3B). Importantly, in interleaved control recordings, stimulus interruption without concomitant PYR depolarization did not significantly affect MF–PYR CaTs, indicating that CaT LTD did not result from nonspecific rundown of MF–PYR CaTs or photodamage (Fig. 3A, inset). Because MF–PYR CaTs measured in thorny excrescences of slices from P10–P16 rats are mediated primarily by CP-AMPARs (Fig. 2), these findings indicate that CP-AMPARs are persistently depressed during DiLTD.
The dramatic depression of MF–PYR CaTs associated with DiLTD is consistent with a long-term inhibition of CP-AMPAR-mediated transmission at MF–PYR synapses allowing transmission to become dominated by CI-AMPARs. If this is the case, then we should be able to detect a DiLTD-induced change in the rectification properties of MF–PYR EPSCs as reported for LTD in cerebellar stellate cells and VTA dopaminergic neurons (Liu and Cull-Candy, 2000, 2005; Bellone and Luscher, 2005; Gardner et al., 2005). Thus, to complement our imaging studies, we next probed for DiLTD-induced changes in RIs of MF–PYR EPSCs. For these experiments, we returned to slices from young mice because we would be examining transgenic animals in later experiments (see below). Moreover, this return to mouse slices allowed us to confirm that the phenomenon of DiLTD, like the expression of CP-AMPARs at MF–PYR synapses, is not species specific. Thus, we first confirmed that mice express DiLTD with properties similar to those observed in rat (Lei et al., 2003). In slices obtained from mice aged P10–P17, transient PYR depolarization from −60 to −10 mV for 5 min produced a depression of MF–PYR EPSCs assayed at −60 mV that persisted for the duration of recordings: 15 min after induction, EPSCs were 63 ± 3.4% of control responses obtained before the depolarization protocol (Fig. 4A,D). In contrast, transient depolarization did not affect MF–PYR transmission in slices from animals older than P22: EPSCs were 95 ± 17% of control at 15 min after induction (Fig. 4B,D). As in the rat, DiLTD observed in slices from young mice (P10–P17) did not require synaptic activity during induction and proceeded independent of changes in paired-pulse ratio consistent with postsynaptic induction and expression loci (Fig. 4A, inset, bottom panel). Furthermore, loading PYRs with BAPTA significantly inhibited DiLTD in slices from young mice revealing a central role for Ca2+ in mouse DiLTD similar to that previously observed in rats (Fig. 4C,D).
Having confirmed the presence of DiLTD in slices from young mice we proceeded to examine whether rectification of AMPAR-mediated MF–PYR EPSCs is altered by DiLTD by comparing RIs before and after induction (Fig. 5). To avoid potentially confounding influences of the prolonged depolarization periods required to obtain full I–V relationships, we adopted an abbreviated RI measure using the absolute value of the ratio of EPSC amplitudes obtained at holding potentials of +40 mV and −60 mV (RIEPSC40/EPSC−60) (see Materials and Methods). Importantly, the brief monitoring of EPSCs at +40 mV (average of 5–10 events obtained at 0.33 Hz, thus 15–30 s) during the baseline period did not significantly affect MF–PYR transmission (data not shown), consistent with a minimal 1–3 min depolarization requirement to observe any depression (Lei et al., 2003). In 9 of 10 recordings, we observed that DiLTD produced a greater depression of EPSCs monitored at −60 mV compared with those obtained at +40 mV (Fig. 5A). Accordingly, DiLTD was associated with a significant increase in MF–PYR RIEPSC40/EPSC−60 values to 157 ± 16% of control at 10 min after induction (Fig. 5B). This RI increase indicates that the contribution of inwardly rectifying CP-AMPARs to MF–PYR EPSCs is depressed after DiLTD induction, leaving transmission to be primarily supported by CI-AMPARs.
While monitoring RIEPSC40/EPSC-60, we observed that DiLTD is associated with an increase in the PPR for EPSCs monitored at +40 mV, in contrast to events recorded at −60 mV at which no DiLTD-induced change in PPR was detected (Figs. 4A, 5A, traces). This seemingly contradictory observation likely reflects the inability of spermine blocked CP-AMPARs to contribute to transmission at +40 mV despite the increased release of transmitter on the second pulse of two closely timed stimuli. Indeed the PPR of naive MF–PYR EPSCs measured at +40 mV was significantly less than that obtained at −60 mV with spermine in the recording pipette (1.6 ± 0.12 vs 1.9 ± 0.11 at Vh of +40 and −60, respectively; n = 10; p = 0.04) (Fig. 5C); however, no difference in PPRs at the two holding potentials was evident under spermine-free recording conditions (1.8 ± 0.16 vs 1.7 ± 0.16 at Vh of +40 and −60, respectively; n = 7; p = 0.6). At naive MF–PYR synapses, use-dependent relief from polyamine block of CP-AMPARs (Rozov et al., 1998; Rozov and Burnashev, 1999; Toth et al., 2000) might be expected to contribute to an increased PPR at +40 mV compared with that at −60 mV (i.e., the larger the degree of initial block the greater opportunity for use-dependent unblock). However, use-dependent unblock does not occur at positive holding potentials between 0 and +60 mV (Rozov et al., 1998; Rozov and Burnashev, 1999). After DiLTD induction, the increased PPR at +40 mV approximates the PPR observed at −60 mV (Fig. 5C) consistent with a switch to transmission being dominated by spermine-insensitive CI-AMPARs. Thus, together, our imaging and electrophysiological data provide compelling evidence for the preferential inhibition of CP-AMPARs during DiLTD at developing MF–PYR synapses, explaining the limited developmental window for this form of synaptic plasticity.
A role for PICK1 in regulating the GluR2 content of MF–PYR synapses
A number of investigations have described a role for the PDZ domain-containing protein PICK1 in regulating the CP-AMPAR content of various central synapses (Terashima et al., 2004; Gardner et al., 2005; Liu and Cull-Candy, 2005; Bellone and Luscher, 2006). Thus, we investigated whether PICK1 participates in controlling the levels of CP-AMPARs at developing MF–PYR synapses.
In cerebellar stellate cells, LTD associated with a loss of CP-AMPAR-mediated transmission is prevented by disruption of PICK1–GluR2 interactions (Gardner et al., 2005; Liu and Cull-Candy, 2005). Because our findings thus far indicate that DiLTD similarly leads to a loss of synaptic CP-AMPARs, we first investigated potential roles for PICK1 in DiLTD using a combination of peptide interference and genetic deletion strategies. Initially, we postsynaptically applied the small interfering peptide EVKI, which selectively disrupts GluR2–PICK1 interactions, through the recording pipette and attempted to induce DiLTD (Daw et al., 2000; Xia et al., 2000; Kim et al., 2001; Gardner et al., 2005; Liu and Cull-Candy, 2005). Because of the variability in time required to find an appropriate MF input after establishing whole-cell configuration, and the close proximity of MF–PYR synapses to the somatic recording location (facilitating rapid dialysis), no effort was made to examine potential peptide influences on transmission during the baseline period. In these recordings, ICS supplemented with the peptide EVKI significantly inhibited DiLTD: 15 min after induction, EPSCs remained at 80 ± 5.7% of control responses (n = 5) (Fig. 6A,D). This inhibition of DiLTD did not result from nonspecific effects of peptide infusion because typical DiLTD was observed in recordings performed with a control peptide SGKA that does not affect PDZ domain-mediated interactions (60 ± 5.3% of control; n = 5) (Fig. 6B,D). Next, we examined DiLTD in slices from young (P10–P17) PICK1 knock-out (PICK1−/−) mice (Gardner et al., 2005; Steinberg et al., 2006). Consistent with the data from our peptide interference experiments, recordings in slices from PICK1−/− mice revealed a severe deficit in DiLTD: at 15 min after induction, EPSCs were 97 ± 8.6% of control responses (n = 10) (Fig. 6C,D). In contrast, slices from age-matched wild-type (PICK1+/+) littermates exhibited robust DiLTD with EPSCs depressing to 63 ± 5.3% of control (n = 4) (Fig. 6D) comparable with that observed in young C57BL/6 mice used throughout the study to this point (compare Fig. 4D).
The blockade of DiLTD by disrupting PICK1 function could be interpreted as evidence for an acute active role of PICK1 in the loss of synaptic CP-AMPARs during DiLTD induction. Alternatively, PICK1 disruption could occlude DiLTD by causing the removal of CP-AMPARs from MF–PYR synapses before induction. Indeed, previous studies have reported alterations in basal synaptic GluR2 content after manipulations of PICK1 function by peptide perfusion or overexpression (Terashima et al., 2004; Gardner et al., 2005). To determine whether EVKI perfusion and PICK1 knock-out alter the basal contribution of CP-AMPARs to MF–PYR synapses, we examined the rectification properties of MF–PYR EPSCs after peptide treatment and in PICK1−/− mice. Surprisingly, in slices from young C57BL/6 mice, recordings with EVKI-supplemented ICS yielded MF–PYR EPSCs with near-linear I–V relationships resulting in RI values (0.90 ± 0.10; n = 5) significantly greater than those obtained with SGKA-supplemented ICS (0.53 ± 0.08; n = 5) (Fig. 7A,C. Similarly, in slices from young PICK1−/− mice, MF–PYR synapses exhibited RIs significantly greater than those obtained in slices from littermate control PICK1+/+ mice (PICK1−/−: 0.84 ± 0.08, n = 11; PICK1+/+: 0.54 ± 0.06, n = 5) (Fig. 7B,C). Furthermore, MF–PYR EPSCs from PICK1−/− mice were not significantly affected by PhTx treatment (EPSCs were 92 ± 8.7% of control after 15 min of treatment). Together, these data indicate that loss of PICK1 function acutely by peptide interference, or chronically by genetic ablation, reduces the contribution of CP-AMPARs to developing MF–PYR synapses. Thus, rather than revealing an acute active role for PICK1 in DiLTD, the lack of DiLTD observed in PICK1−/− mice or after EVKI treatment likely reflects a deficit in the contribution of CP-AMPARs at developing MF–PYR synapses, indicating a critical role for PICK1 in regulating the GluR2 content of developing MF–PYR synapses.
Finally, as an independent assay to probe whether PICK1 regulates the subunit composition of MF–PYR synapses, we performed immunogold electron microscopy (EM) in PYR thorny excrescences of PICK1+/+ and PICK1−/− mice. Tissue from three different P14 animals for each genotype was double labeled using a polyclonal GluR1-specific antibody and a monoclonal GluR2-specific antibody, and then anti-GluR1 and anti-GluR2 signals were revealed with 5 and 15 nm gold particles, respectively. Consistent with our electrophysiological data, PYR spine profile membranes of PICK1−/− mice showed a modest reduction in GluR1 signal and a modest increase in GluR2 signal compared with PICK1+/+ littermate controls yielding a significantly larger GluR2/GluR1 signal ratio in PICK1−/− mice: 0.62 ± 0.1 and 1.0 ± 0.2 for PICK1+/+ and PICK1−/−, respectively (p = 0.04) (Fig. 7D). Moreover, we observed an increase in the number of synaptic profiles doubly labeled for GluR1 and GluR2 in the absence of PICK1. The ratios of synapses labeled with 5 nm gold particles only, to those labeled with 5 and 15 nm gold particles were 0.98 and 0.64 for PICK1+/+ and PICK1−/− MF–PYR synapses, respectively (Fig. 7D, micrographs) (see Materials and Methods). Thus, considered together, our immuno-EM and electrophysiological data indicate a critical role for PICK1 in the developmental expression of CP-AMPARs at MF–PYR synapses.
In the mature CNS, excitatory transmission between principal neurons is dominated by CI-AMPARs because of prominent expression of edited GluR2 subunits within postsynaptic AMPARs. However, GluR2-lacking, CP-AMPARs participate at principal cell synapses early in development in various central structures (Pellegrini-Giampietro et al., 1992; Pickard et al., 2000; Zhu et al., 2000; Aizenman et al., 2002; Kumar et al., 2002; Eybalin et al., 2004; Migues et al., 2007). Here, we examined whether CP-AMPARs contribute at developing MF–PYR synapses using imaging and electrophysiological approaches. We found that MF–PYR transmission within the first 2–3 postnatal weeks uses a mixed population of CP- and CI-AMPARs. CP-AMPAR participation at MF–PYR synapses is developmentally regulated as CI-AMPARs dominate beyond P17. This temporal profile parallels the period during which MF–PYR synapses are DiLTD competent, and our findings suggest that CP-AMPARs are selectively targeted during DiLTD. Consistent with preferential removal of CP-AMPARs during DiLTD, young MF–PYR synapses lacking CP-AMPARs because of PICK1 disruption were DiLTD deficient. These data also revealed that PICK1 regulates the GluR2 content of MF–PYR synapses as reported previously (Terashima et al., 2004; Gardner et al., 2005; Liu and Cull-Candy, 2005; Bellone and Luscher, 2006). We conclude that expression of CP-AMPARs dictates the developmental window for DiLTD and additionally propose that DiLTD proceeds as the PICK1-dependent exchange of CI-AMPARs for CP-AMPARs reminiscent of plasticity in cerebellar stellate cells and VTA neurons (Bellone and Luscher, 2005; Gardner et al., 2005; Liu and Cull-Candy, 2005).
Previous studies did not observe rectifying I–V curves or PhTx sensitivity of MF–PYR synaptic events in P15–P25 rat slices (Jonas et al., 1993; Toth et al., 2000). This age range biases toward developmental stages where MF–PYR transmission is entirely supported by CI-AMPARs, likely explaining the differences with our findings. Additionally, Jonas et al. (1993) performed their study before the discovery that CP-AMPAR rectification results from intracellular polyamines, and so it was not common practice to supplement intracellular solutions with polyamines. Because spermine supplementation is critical for MF–PYR rectification (Fig. 1D), this may also have contributed to the lack of rectification in the previous investigation. Subsequently, Jonas and colleagues examined rectification properties of PYR somatic membrane patches from P13–P15 rat slices and again observed linear I–V relationships even with spermine present (Koh et al., 1995). This difference with our findings cannot be attributed to age differences or lack of spermine supplementation and, thus, indicates that young PYR extrasynaptic membranes are rich in CI-AMPARs. Similarly, CI-AMPARs dominate extrasynaptic membranes of cerebellar stellate cells and developing layer 5 pyramids despite synaptic expression of CP-AMPARs (Liu and Cull-Candy, 2000, 2002; Kumar and Huguenard, 2001; Kumar et al., 2002; Gardner et al., 2005). We additionally found that associational/commissural-PYR transmission is mediated solely by CI-AMPARs during early development. Considered together, the evidence indicates that during the first 2–3 postnatal weeks PYRs express CP-AMPARs in a synapse-specific manner.
Species-dependent differences in synaptic expression of CP-AMPARs also exist (Bellone and Luscher, 2005, 2006); therefore, we considered whether our observations in mouse differed from previous reports because of choice of experimental animal. However, both imaging and electrophysiological recordings confirmed participation of CP-AMPARs at developing rat MF–PYR synapses. Our imaging experiments also provided direct evidence for Ca2+ permeability of MF–PYR AMPARs, an important consideration because Ca2+ permeability is more sensitive to GluR2 content than is rectification or PhTx sensitivity (Dingledine et al., 1992; Washburn et al., 1997). A previous study in slice culture found that MF evoked CaTs in PYR spines were insensitive to the CP-AMPAR antagonist HPP-spermine [N-(4-hydroxyphenylpropanoyl)-spermine] (Reid et al., 2001). Again, this discrepancy with our findings likely reflects the limited developmental expression of CP-AMPARs at MF–PYR synapses as slices were obtained at P8 and cultured for 10–21 d.
CP-AMPARs impart distinct short- and long-term plasticity to various central synapses (Rozov et al., 1998; Rozov and Burnashev, 1999; Liu and Cull-Candy, 2000; Toth et al., 2000; Lei and McBain, 2002; Bellone and Luscher, 2005; Gardner et al., 2005; Pelkey et al., 2005, 2006; Shin et al., 2005). The loss of CaTs and decreased rectification after DiLTD induction indicate that CP-AMPARs are selectively depressed during DiLTD. Accordingly, MF–PYR synapses devoid of CP-AMPARs, in slices from older animals or after PICK1 disruption, do not exhibit DiLTD. The CI-AMPARs that support transmission after DiLTD appear to principally reflect the population initially present at naive synapses. However, the occasionally observed EPSC growth at +40 mV, particularly for the second EPSC in pairs of events (Fig. 5A, traces), suggests that CI-AMPARs may also be exchanged for CP-AMPARs after DiLTD, similar to LTD in cerebellar stellate cells and dopaminergic VTA neurons (Liu and Cull-Candy, 2000; Bellone and Luscher, 2005; Gardner et al., 2005). The initial wide range of RIs at naive MF–PYR synapses combined with the large variability in presynaptic release and low number of events collected at +40 mV in the recordings may have confounded reliable detection of EPSC growth at positive potentials. The greater success in observing such growth on the second EPSC of evoked pairs could reflect increased release probability combined with perisynaptic localization of new CI-AMPARs that will diffuse laterally to postsynaptic sites.
The critical trigger for DiLTD induction is an increase in postsynaptic Ca2+ by influx through L-type VGCCs and release from intracellular stores (Lei et al., 2003). Similarly, loss of synaptic CP-AMPARs in cerebellar stellate cells and VTA dopaminergic neurons is triggered by increased intracellular Ca2+ (Liu and Cull-Candy, 2000; Bellone and Luscher, 2005). CP-AMPAR removal may serve a protective feedback mechanism to combat additional cytosolic Ca2+ increases that can be deleterious for the postsynaptic cell (Sattler and Tymianski, 2000; Arundine and Tymianski, 2003). The physiological trigger for the developmental loss of CP-AMPARs at MF–PYR synapses remains unknown. However, PYRs exhibit intense burst firing early in development and such firing could effectively activate L-type VGCCs, expression of which is upregulated between P8 and P21 and localizes to proximal dendrites near MF inputs (Westenbroek et al., 1990; Glazewski et al., 1993; Hell et al., 1993; Elliott et al., 1995). Alternatively, excess Ca2+ influx through the CP-AMPARs themselves could provide a self-regulating mechanism triggering CP-AMPARs removal (Liu and Cull-Candy, 2000).
The loss of CP-AMPAR-mediated MF–PYR transmission after PICK1 disruption was surprising because cerebellar stellate cells maintain, or even increase, the synaptic complement of CP-AMPARs after EVKI treatment or PICK1 knock-out (Gardner et al., 2005; Liu and Cull-Candy, 2005). However, our findings are consistent with observations in CA1 pyramids in which PICK1 overexpression promotes CP-AMPAR accumulation at synapses, whereas EVKI overexpression increases GluR2 content (Terashima et al., 2004). Similarly, small interfering RNA knockdown of PICK1 promotes membrane insertion of GluR2-containing receptors (Sossa et al., 2006). Development of a unified model for PICK1 in regulating synaptic AMPAR composition is hampered by conflicting findings at distinct synapses, suggesting that PICK1 functions in a cell- and synapse-specific manner. For example, a well defined role for PICK1 in GluR2-containing AMPAR internalization exists at cerebellar parallel fiber–Purkinje neuron synapses (Xia et al., 2000; Steinberg et al., 2006), whereas stellate cells require PICK1 for synaptic delivery of GluR2-containing AMPARs (Gardner et al., 2005; Liu and Cull-Candy, 2005). Thus, evidence supports roles for PICK1 in AMPAR internalization and surface delivery. Indeed, PICK1 supports bidirectional changes in surface AMPARs (Sossa et al., 2006). This duality of function likely reflects the Ca2+ binding properties of PICK1, which yield bimodal regulation of PICK1–GluR2 interactions (Hanley and Henley, 2005; Sossa et al., 2006). Our findings suggest a model in which PICK1 maintains intracellular pools of GluR2-containing AMPARs (Jin et al., 2006), which are trafficked to the surface after appropriate stimuli like increased intracellular Ca2+. Loss of PICK1–GluR2 interactions by highly elevated intracellular Ca2+, EVKI treatment, or PICK1 knock-out could promote mobilization of this CI-AMPAR intracellular pool to the surface, which subsequently may trigger removal of synaptic CP-AMPARs, allowing for synaptic incorporation of the CI-AMPARs through lateral diffusion. Although additional investigation is needed to characterize molecular events controlling AMPAR trafficking at MF–PYR synapses, our findings further implicate PICK1 as an important regulator of synaptic GluR2 (Terashima et al., 2004; Bellone and Luscher, 2005; Gardner et al., 2005; Liu and Cull-Candy, 2005).
In conclusion, we found that CP-AMPARs contribute to nascent MF–PYR transmission within the first 3 postnatal weeks. This early participation of CP-AMPARs may provide a postsynaptic Ca2+ signal important for synapse maturation. Developing synapses are believed to proceed from silent, with transmission solely mediated by NMDARs, to functional by the stepwise acquisition of AMPARs (Isaac, 2003; Poncer, 2003). Our findings along with other reports (Pellegrini-Giampietro et al., 1992; Pickard et al., 2000; Zhu et al., 2000; Aizenman et al., 2002; Kumar et al., 2002; Eybalin et al., 2004; Migues et al., 2007) indicate the transient incorporation of CP-AMPARs may represent a previously unappreciated step at many central synapses in this model. Interestingly, a similar phenomenon may occur during various forms of synaptic plasticity in the mature CNS, although on more rapid timescales (Thiagarajan et al., 2005; Clem and Barth, 2006; McCormack et al., 2006; Plant et al., 2006; Sutton et al., 2006), further linking the processes of synapse development and plasticity.
This research was supported by National Institutes of Health (NIH) intramural awards [National Institute of Child Health and Human Development (C.J.M.) and National Institute on Deafness and Other Communication Disorders (R.S.P.)], NIH extramural and Howard Hughes Medical Institute awards (R.L.H.), as well as Canada Research Chairs, Canadian Institutes of Health Research (CIHR), and Fonds de la Recherche en Santé du Québec awards (J.-C.L.). K.A.P. is an NIH Visiting Fellow and a CIHR Fellow. L.T. received support from the Savoy Foundation. We thank Xiaoqing Yuan, Brian Jeffries, and Dr. Ya-Xian Wang for expert technical assistance.
- Correspondence should be addressed to Kenneth A. Pelkey, Laboratory on Cellular and Synaptic Neurophysiology, Building 35, Room 3C705, National Institute of Child Health and Human Development, Bethesda, MD 20892.