Conditional Restoration of Hippocampal Synaptic Potentiation in GluR-A-Deficient Mice

See allHide authors and affiliations

Science  29 Jun 2001:
Vol. 292, Issue 5526, pp. 2501-2504
DOI: 10.1126/science.1059365


Plasticity of mature hippocampal CA1 synapses is dependent onl-α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptors containing the glutamate receptor A (GluR-A) subunit. In GluR-A–deficient mice, plasticity could be restored by controlled expression of green fluorescent protein (GFP)–tagged GluR-A, which contributes to channel formation and displayed the developmental redistribution of AMPA receptors in CA1 pyramidal neurons. Long-term potentiation (LTP) induced by pairing or tetanic stimulation was rescued in adult GluR-A–/– mice when GFPGluR-A expression was constitutive or induced in already fully developed pyramidal cells. This shows that GluR-A–independent forms of synaptic plasticity can mediate the establishment of mature hippocampal circuits that are prebuilt to express GluR-A–dependent LTP.

Of the four AMPA receptor subunits (GluR-A to GluR-D) constituting one family of glutamate-gated ion channels (1–3), GluR-A is essential for adult hippocampal LTP but not for spatial learning in a water maze task (4). Studies on mice lacking GluR-A provided evidence that after tetanic stimulation, increased transmission at Schaffer collateral (SC/CA1) synapses is established by an augmented response of AMPA receptors. The selective, strong reduction of somatic AMPA receptor currents in GluR-A–deficient mice (4) further suggests that lack of AMPA receptors at nonsynaptic sites is linked to absence of LTP. GFP-tagged AMPA receptor expression in organotypic slice cultures supports the idea that GluR-A and GluR-D subunits facilitate rapid AMPA receptor delivery in SC/CA1 synapses after LTP induction (5–8).

We now investigate the contribution of GluR-A–dependent LTP in development of hippocampal connections. We analyzed whether in adult GluR-A–deficient mice, synaptic plasticity can be restored by controlled expression of GluR-A. First, we generated mice with regulated GFPGluR-A expression and characterized the function of the fluorescent subunit in the genetic background of wild-type mice. In a second step, the GFPGluR-A expression system was transferred to GluR-A–/– mice, and the restoration of synaptic plasticity was studied in the presence and absence of GFPGluR-A.

Regulated GFPGluR-A expression (Fig. 1A) by the doxycycline (dox) system (9) provided strong GFPGluR-A immunoreactivity to the hippocampal formation at postnatal day 14 (P14) and P42 (Fig. 1B and Web fig. 1B). However, the somata of hippocampal pyramidal cells displayed different GFPGluR-A labeling intensities, indicating variable amounts ofGFPGluR-A in distinct pyramidal cells (Web fig. 1C). As estimated by immunoblots (Fig. 1C), theGFPGluR-A levels were about 10% of total GluR-A, which showed that GFPGluR-A–expressing mice should not be compromised by overexpression of AMPA receptors.GFPGluR-A was incorporated in active receptor channels as evidenced by current-voltage relations (Web fig. 1A) and could be copurified with GluR-B and GluR-C from hippocampi ofGFPGluR-A–expressing mice (Fig. 1D). GFPGluR-A was also associated with endogenous GluR-A (Fig. 1D), a clear indication of AMPA receptor populations with more than a single GluR-A subunit.

Figure 1

GFPGluR-A expression and AMPA receptor assembly. (A) Schematic drawing of the αCaMKII/tTA transgene of line Tg(CaMKIItTA)Mmay(17) and the construct of line Tg(nlacZtetOGFPGluR-A)A1.1 (18) for tTA-controlled expression of a nuclear β-gal and ofGFPGluR-A. In the presence of dox, tTA is inactive. (B) Enzymatically visualized β-gal and anti-GFP–immunostained (16) GFPGluR-A in coronal forebrain sections of wild-type mice positive for both transgenes shown in (A). (C) Expression of GluR-A,GFPGluR-A, and β-gal was monitored by immunoblots of hippocampal proteins (16) of wild-type (b, d) andGFPGluR-A–expressing (a, c, e) mice at P14 (a, b) and P42 (c, d), and at P42 in GluR-A–/– mice (e). (D) Coimmunoprecipitations with the indicated antibodies of solubilized AMPA receptors (19) from brain homogenates ofGFPGluR-A–expressing wild-type mice. Additionally, 5% of the homogenate (Input) used for coimmunoprecipitation was loaded. The blot was probed with anti-GluR-A. (E) GFPGluR-A fluorescence in CA1 neurons of wild-type and GluR-A–/–mice at P14 and P42.

The trafficking of the GFPAMPA receptors seemed undisturbed because the distribution of green fluorescence mimicked the development-dependent redistribution of the AMPA receptors from somato/dendritic to dendritic location in hippocampal CA1 neurons (10). At P14, principal neurons in CA1 were labeled in somata and in dendritic fields, whereas at P42, somaticGFPGluR-A fluorescence disappeared (Fig. 1E). Expression levels of the transgene in the hippocampus measured by immunoblots and β-galactosidase (β-gal) activity did not change from P14 to P42 (Fig. 1C, Web fig. 1C). However, within CA1 dendrites,GFPGluR-A could be observed in spines and shafts in young animals (Fig. 2A), whereas in older mice (P42) it was difficult to find GFPGluR-A in dendritic shafts (Fig. 2A). This indicates that in adult mice, the majority of GluR-A–containing AMPA receptors is located in spines. The age-dependent difference in the GluR-A distribution pattern was not simply caused by a different amount of AMPA receptors. When AMPA receptor expression was reduced by transfer of theGFPGluR-A expression system into the genetic background of GluR-A–deficient mice, we observed a GFPGluR-A fluorescence pattern that was very similar to that detected in the wild-type background (Fig. 2B).

Figure 2

GFPGluR-A in CA1 pyramidal cells: mosaic expression, spine location, and contribution to somatic AMPA currents. (A and B) Confocal images (16) of pyramidal cell bodies (left) and spines on dendritic shafts (middle and right) visualized by GFP fluorescence inGFPGluR-A–expressing wild-type and GluR-A–/–mice. At P42, for both genotypes, GFPGluR-A was dominant in spines and hardly visible in shafts. (C) For recordings, pyramidal cell bodies were monitored for GFPGluR-A fluorescence (left) and patched in the infrared mode (right) to determine (D) AMPA (fast) and NMDA (slow) receptor-mediated currents of nucleated patches (4). (E) Bar diagram depicting AMPA/NMDA current ratios in wild-type (gray,n = 5), GluR-A–/– (white,n = 8), and GFPGluR-A–expressing GluR-A–/– mice (green, n = 7).

We next investigated if the GFPGluR-A subunit was able to replace the function of the depleted endogenous GluR-A in GluR-A–deficient mice. Nucleated patch recordings from fluorescent CA1 pyramidal cells (Fig. 2, C through E) ofGFPGluR-A–expressing GluR-A–/– mice showed a relatively small increase in the AMPA/N-methyl-d-aspartate (NMDA) receptor current ratio (0.92 ± 0.51) over that of GluR-A–deficient mice (0.27 ± 0.12). However, the current ratio remained far below that determined in wild-type mice (5.41 ± 1.01). This result is in agreement with immunological data, which suggested that in the mosaic CA1 cell population the strongest GFPGluR-A–expressing cells have less GFPGluR-A than endogenous GluR-A. Despite the low AMPA receptor–mediated soma currents, pathway specific cellular LTP induced by pairing (11, 12) was evoked inGFPGluR-A positive CA1 neurons and was not detectable in CA1 cells of GluR-A–deficient mice at P42 (Fig. 3A). The mosaic GFPGluR-A expression in the CA1 cell population (Fig. 2, A and B) caused partial recovery of field LTP (Fig. 3B). The average field excitatory postsynaptic potential (fEPSP) slope 40 to 45 min after tetanization (100 Hz, 1 s) in adult GFPGluR-A–expressing GluR-A–/– mice was 125 ± 3%, clearly between that of adult GluR-A–deficient (108 ± 3%) and wild-type (147 ± 5%) mice.

Figure 3

Recovery of SC/CA1 LTP inGFPGluR-A–expressing GluR-A–/– mice. (A) Summary graphs of SC-evoked EPSC amplitudes in the paired (filled circles) and unpaired control pathway (open circles) before and after pairing in CA1 pyramidal cells of wild-type (top,n = 7), GluR-A–/– (middle,n = 5), and GFPGluR-A–epressing GluR-A–/– mice (bottom, n = 5 GFP fluorescent cells) (12). (B) Summary graphs of extracellular fEPSP slopes evoked in the tetanized (closed circles) and untetanized (open circles) pathways (4) in slices of wild-type (top, n = 35), GluR-A–/–(middle, n = 34), andGFPGluR-A–expressing GluR-A–/– mice (bottom, n = 21). Data were obtained in Ringer's solution containing 2 mM Ca2+ and 2 mM Mg2+. The LTP level of GFPGluR-A–expressing GluR-A–/– mice differed significantly from LTP of GluR-A–/– (P = 0.0001) and wild-type (P = 0.004) mice. In all genotypes, the untetanized control pathway showed no improved response (102 ± 2%, 103 ± 1%, and 104 ± 1%). Vertical bars indicate SEM.

A continued developmental expression of GluR-A–mediated synaptic plasticity might be important for the establishment of LTP in synapses of adult animals. We determined whether “delayed” expression of GluR-A—i.e., expression only in the adult brain—can restore LTP in hippocampal SC/CA1 connections of GluR-A–/– mice. For expression delay, transcription of the GFPGluR-A transgene was switched off by dox until P21. Figure 4A and Web fig. 1D illustrate that in the continued presence of dox, neither GFPGluR-A nor β-gal was detected at P14. When dox was removed at P21, GFPGluR-A and β-gal expression was induced, and both proteins were readily expressed in CA1 at P42. The delayed induced expression ofGFPGluR-A and β-gal did not reach the levels seen in mice that had both genes activated during development (Fig. 4B). We estimate that the level of hippocampal GFPGluR-A was, at most, 20% of that in mice raised without dox. The decreased delayed expression was accompanied by a reduced number of β-gal–positive cells in CA1 (Fig. 4C). In acute slices, we identified by fluorescence CA1 pyramidals that expressed GFPGluR-A and determined cellular LTP. In neurons with GFPGluR-A, cellular LTP was expressed (Fig. 4, D and E). The averaged normalized excitatory postsynaptic current (EPSC) amplitudes 20 min after pairing were 1.67 ± 0.15 in GluR-A–deficient, constitutively GFPGluR-A–expressing mice and 1.78 ± 0.14 after GFPGluR-A induction. Wild-type and GluR-A–/– mice had normalized EPSC amplitudes of 1.96 ± 0.27 and 1.09 ± 0.07, respectively.

Figure 4

Rescue of LTP in GluR-A–deficient mice by delayed GFPGluR-A expression induced at P21 and analyzed at P42. (A) GFP immunoreactivity seen in the hippocampus of P14 GFPGluR-A–expressing mice (left) was not detected in mice nursed under dox (75 μg/ml drinking water) (middle) (16). When dox was removed at P21, GFP immunoreactivity was observed at P42 (right). (B) From immunoblots, GFPGluR-A (upper blot) and β-gal expression (lower blot) at P42 was high in hippocampi of GFPGluR-A–expressing wild-type (a) and GluR-A–/– kept without dox (b), but lower in GluR-A–/– mice when dox was removed at P21 (c). (C) Cryostat sections showing that at P42 the number of blue, β-gal–positive CA1 pyramidal cells was higher (left) compared to mice with delayedGFPGluR-A and β-gal expression induced at P21 (right) (20). (D) Summary graphs of SC-evoked EPSC amplitudes as described in Fig. 3A, but now analyzed in CA1 cells ofGFPGluR-A–expressing GluR-A–/– mice nursed under dox until P21 (n = 3 fluorescent cells). (E) Twenty minutes after pairing, the averaged normalized EPSC amplitudes are significantly increased in paired versus control pathway of wild-type (n = 8) and GluR-A–/– mice with constitutive (n = 8; green) or P21-inducedGFPGluR-A expression (n = 6; green/white) and unchanged in GluR-A–deficient mice (n = 7). The LTP level in GluR-A–/– mice with constitutive or P21-induced GFPGluR-A differed significantly from LTP in in GluR-A–/– mice (P = 0.002 andP = 0.0006, respectively), whereas they did not differ significantly from each other (P = 0.59) nor from that in wild-type mice (P = 0.34 andP = 0.59, respectively).

In adult GluR-A–deficient mice, the lack of LTP induced at SC/CA1 synapses by pairing or tetanic stimulation could be restored by the expression of GFPGluR-A. This supports a strict GluR-A dependence of rapidly increased synaptic efficacy in both protocols of LTP induction. At the cellular level, LTP was rescued to its full extent, although the expression of GFPGluR-A did not reach wild-type levels. Somatic AMPA receptor current amplitudes were still limited to about 10% of those of wild-type mice, indicating that an AMPA receptor density generating only 10% of extrasynaptic currents was sufficient for expression of potentiation.

In GluR-A–deficient mice, functional synaptic SC/CA1 connections were formed in the absence of GluR-A and GluR-A–dependent plasticity. Synapses were made modifiable when the block of GFPGluR-A gene expression was unlocked by removing dox, and GFPGluR-A was readily synthesized and incorporated into dendritic spines. This implies that mature synapses were prebuilt for GluR-A–mediated potentiation even when GluR-A was never expressed. It is possible that in wild-type mice early in development, other GluR subunits (GluR-C and GluR-D) contribute to AMPA receptor dependent plasticity and that in mature synapses, one of these subunits is replaced by GluR-A. The GluR-D subunit is discussed as being important for neonatal plasticity (5). However, the mechanism of activity-induced AMPA receptor delivery to the synapse differs for GluR-D- and GluR-A–containing receptors (8). Therefore, other mechanisms of neuronal plasticity are more likely to establish functional hippocampal circuitry. Recently, we found a GluR-A–independent form of LTP in juvenile mice that might be important for the regular development of hippocampal connections (10). The fact that the induction and establishment of LTP in adult mice can be repressed or restored shows that regulated expression of individual AMPA receptor subunits, especially of GluR-A, can substantially alter the functional properties of hippocampal synaptic connections, in particular their ability to undergo long-term changes in synpatic efficacy.

  • * Present address: Department of Neurophysiology, Faculty of Biology, Vrije University, 1081 HV, Amsterdam, Netherlands.

  • To whom correspondence should be addressed. E-mail: sprengel{at}


View Abstract

Stay Connected to Science

Navigate This Article