|Home | About | Journals | Submit | Contact Us | Français|
Learning is accompanied by modulation of postsynaptic signal transduction pathways in neurons. Although the neuronal protein kinase cyclin-dependent kinase 5 (Cdk5) has been implicated in cognitive disorders, its role in learning has been obscured by the perinatal lethality of constitutive knockout mice. Here we report that conditional knockout of Cdk5 in the adult mouse brain improved performance in spatial learning tasks and enhanced hippocampal long-term potentiation and NMDA receptor (NMDAR)-mediated excitatory postsynaptic currents. Enhanced synaptic plasticity in Cdk5 knockout mice was attributed to reduced NR2B degradation, which caused elevations in total, surface and synaptic NR2B subunit levels and current through NR2B-containing NMDARs. Cdk5 facilitated the degradation of NR2B by directly interacting with both it and its protease, calpain. These findings reveal a previously unknown mechanism by which Cdk5 facilitates calpain-mediated proteolysis of NR2B and may control synaptic plasticity and learning.
The cellular and molecular mechanisms underlying declarative learning and memory constitute a predominant area in basic neurobiological research and are of considerable clinical relevance. A generally accepted hypothesis is that memories are formed in specific circuits of the brain through changes in the strength of neuronal synapses1. One form of synaptic plasticity, long-term potentiation (LTP), seems to be a critical component of information processing and memory storage in the hippocampus1,2. Information storage at the behavioral and cellular levels depends on neurotransmitter receptor complexes and down-stream intracellular signaling cascades. Obtaining a better understanding of these biochemical pathways is a crucial step in delineating the processes controlling learning and synaptic plasticity.
The proline-directed serine/threonine protein kinase Cdk5, together with its neuronal-specific activating cofactors p35 and p39, have been implicated in a plethora of normal and pathological processes in the mammalian CNS3-12. However, assessment of Cdk5’s function in learning and synaptic plasticity has been limited by congenital abnormalities in constitutive knockout mice9 and by pharmacological nonspecificity13.
To circumvent these confounds, we examined the effects of conditional Cdk5 knockout on contextual and spatial learning and synaptic plasticity. Loss of Cdk5 in the adult brain improved performance in hippocampal behavioral learning tasks, enhanced synaptic plasticity and elevated currents through NMDARs. Biochemical and neurophysiological analyses showed that the enhanced plasticity in Cdk5 knockout mice was due to changes in NR2B subunit degradation. These findings reveal a previously unknown pathway through which Cdk5 may regulate ionotropic glutamate receptor constituency via protein-protein interactions.
We generated an inducible conditional Cdk5 knockdown model to study the role of Cdk5 in learning and synaptic plasticity. This Cre/loxP system allowed temporal control of Cdk5 gene deletion in the adult brain. Using homologous recombination, exons encoding vital Cdk5 catalytic-domain components were flanked with loxP elements (floxed; Fig. 1a). Homozygous floxed Cdk5 mice were crossed with animals bearing an inducible Cre-ERTrecombinase transgene under the control of the prion protein promoter14.
Conditional Cdk5 knockdown in homozygous floxed mice carrying the Cre transgene was achieved by administration of hydroxytamoxifen, an estrogen receptor antagonist that induces nuclear translocation of the Cre-ERT recombinase14. Hydroxytamoxifen-dosed homozygous floxed mice lacking the Cre-ERT transgene showed no loss of Cdk5 and served as controls (wild type; Supplementary Fig. 1 online). Floxed alleles, transgenic alleles and drug treatment had no effects on Cdk5 levels. Furthermore, Cdk5 knockout did not alter basic hippocampal cytoarchitecture (Supplementary Fig. 1). Empirically optimized drug treatment reduced Cdk5 mRNA levels in whole hippocampus, striatum, cortex, cerebellum, hippocampal CA1 and CA3, and dentate gyrus layers by 45.7 ± 8.8%, 59.3 ± 8.3%, 75.9 ± 7.4%, 73.8 ± 6.8%, 71.4 ± 3.0%, 55.7 ± 3.2% and 49.0 ± 2.6%, respectively, compared with wild type (Fig. 1b and Supplementary Fig. 1). Real-time polymerase chain reaction (RT-PCR) revealed a 43.6 ± 6.8% reduction in whole hippocampus Cdk5 mRNA (Supplementary Fig. 1). Decreases in hippocampal Cdk5 protein level (59.2 ± 4.9%, Fig. 1c) and activity (44.1 ± 0.2%, Fig. 1d) were detected by quantitative immunoblots and immunoprecipitation kinase assays, respectively. Pseudocolor quantification of Cdk5 mRNA also showed a robust knockdown of Cdk5 message in hippocampal layers (Fig. 1e). High-resolution confocal analyses of fluorescent in situ hybridizations (FISH) were conducted to assess Cdk5 knockout in CA1 neurons. FISH showed that 90.9 ± 7.7% of cells in the stratum pyramidale of Cdk5 knockout lacked Cdk5 mRNA (Fig. 1f). An equivalent proportion of CA1 cells in Cdk5 knockout mice expressed the Cre-ERT mRNA (Fig. 1f). Thus, temporally controlled, conditional loss of Cdk5 was achieved, allowing for the study of hippocampus-dependent learning and synaptic plasticity.
The role of Cdk5 in associative learning and memory was evaluated via contextual and cued fear-conditioning behavioral tests. Animals were placed in a novel context, analyzed for baseline behavior, and exposed to cue/foot shock pairings. Memory to context was assessed by examining freezing behavior 24 h after training. Cdk5 knockout mice showed significantly increased contextual memory (61.4 ± 4.1% for knockout versus 42.1 ± 3.8% for wild type; Fig. 2a). No significant differences were observed in baseline freezing, cue-conditioned memory, nociception, locomotion, anxiety or preknockout memory (Fig. 2a, Supplementary Fig. 2 online). The increased freezing upon re-exposure to context indicates enhanced hippocampus-dependent memory.
We examined the rate and extent of contextual fear-extinction by re-exposing the animals to the conditioned context over several trials. Cdk5 knockout mice showed greater contextual extinction, as evidenced by significantly less freezing during the fourth re-exposure, 96 h after training (29.8 ± 4.4% for knockout versus 71.1 ± 12.6% for wild type, Fig. 2b; Supplementary Fig. 2). Thus, conditional Cdk5 knockout resulted in increased hippocampus- and NMDAR-dependent15 cognitive flexibility.
To assess the effect of conditional Cdk5 knockout on spatial learning and memory, we used the hippocampus- and NMDAR-dependent water-maze task16,17. Animals were trained to find a submerged platform using distant visual cues, after which the platform was removed to test spatial memory. We then assessed hippocampus-dependent17-19 cognitive flexibility by retraining the mice to find a platform in the opposite quadrant (reversal). Although both genotypes showed equally successful learning during initial training sessions (Fig. 2c), knockout mice showed shortened escape latencies during the reversal task (Fig. 2d). The enhancement of reversal learning was apparent as early as the second trial of day 1 and first trial of day 2, indicating single-trial and long-term spatial learning enhancements (Fig. 2d). Reversal differences were not due to disparities in velocity (Supplementary Fig. 2). Better reversal learning was accompanied by superior spatial memory, as assessed by a platform-removal probe trial. Although both genotypes showed significant preferences for the target quadrant after 11 d of training (Fig. 2c), only knockout mice showed a preference for the new target region after reversal training (Fig. 2d). Therefore, conditional Cdk5 knockout resulted in improved performance in the hippocampus-dependent17 reversal spatial learning task.
Because alterations in the strength of synaptic connection between hippocampal neurons are believed to underlie contextual and spatial memories1,16, we analyzed synaptic plasticity in the hippocampal Schaffer collateral pathway (SC-CA1). No differences between wild-type and knockout groups were observed in AMPA receptor–mediated synaptic transmission or paired-pulse facilitation (PPF; Fig. 3a), as measured by extracellularly recorded field excitatory postsynaptic potentials (fEPSPs). A weak theta-burst stimulus (2× TBS) had no effect on wild type (2.0 ± 6.9% change from baseline), but induced robust LTP in knockout slices (34.0 ± 4.0%; Fig. 3b,c). We obtained similar results after a weak 50-Hz tetanus (0.1 ± 2.3% for wild type versus 31.1 ± 5.9% for knockout; Fig. 3c). A stronger theta-burst stimulation (3× TBS) induced LTP of similar magnitude in wild-type and knockout slices (34.2 ± 6.3% versus 49.7 ± 1.0%, respectively; Fig. 3b,c), and elevated post-tetanic potentiation in knockout slices relative to wild type (123.4 ± 8.9% versus 52.4 ± 14.7%, respectively; Fig. 3b). Strong 100-Hz tetanus also produced equivalent LTP in wild-type and knockout slices (48.3 ± 5.4% versus 49.1 ± 4.9%, respectively; Fig. 3c). These results indicated that Cdk5 knockout reduced the induction threshold for LTP in the SC-CA1 pathway.
To investigate the LTP enhancement mechanism, we assessed the relative contribution of synaptically evoked NMDAR-mediated current. The relationship of stimulus intensity to fiber-volley magnitude was equivalent between groups, indicating a lack of differential axonal stimulation (data not shown). In addition, there were no disparities in AMPA receptor-mediated synaptic transmission (Fig. 3a) or whole-cell currents (3,601 ± 400 versus 3,800 ± 313 charge for wild type and knockout; P = 0.71, Student’s t-test). Voltage measurements in field recordings from Cdk5 knockout slices revealed increased NMDAR-mediated synaptic transmission and a 9.0 ± 2.4-fold larger NMDA/AMPA fEPSP ratio versus wild type (Fig. 3d). Whole-cell voltage-clamp synaptically evoked NMDAR-mediated excitatory postsynaptic currents (EPSCNMDA) in CA1 pyramidal neurons revealed no difference in NMDAR reversal potential, but did show a 1.4 ± 0.2-fold larger NMDA/AMPA EPSC ratio (Fig. 3e and Supplementary Fig. 3 online). Thus, loss of Cdk5 resulted in larger NMDAR-mediated fEPSPs and currents.
NMDAR-mediated current in the adult hippocampal SC-CA1 pathway is predominantly mediated by receptor complexes consisting of NR1 and NR2A and/or NR2B subunits20. We determined the relative subunit contributions to the increased EPSCNMDA by measuring sensitivity to an NR2B-selective inhibitor, ifenprodil21,22. Ifenprodil-sensitive EPSCNMDA charges were 3.2 ± 0.5-fold greater in Cdk5 knockout slices compared with wild type (Fig. 4a). Consistent with this observation, the decay constant for the EPSCNMDA was longer in knockout (74.2 ± 7.2 ms) than wild type (51.5 ± 4.7 ms, Fig. 4a). Notably, immunoblot analysis of Cdk5 knockout hippocampal homogenates revealed a 1.33 ± 0.07-fold increase in NR2B levels, whereas NR2A, NR1, and GluR1 levels were unaltered (0.86 ± 0.06, 0.96 ± 0.11, and 0.99 ± 0.18 versus wild type, respectively; Fig. 4b). Furthermore, cell-surface receptor-labeling experiments showed a 1.9 ± 0.2-fold increase in NR2B surface-to-total ratio in knockout slices (Fig. 4c). This correlated with a hippocampus-specific 2.0-fold increase in synaptosomal NR2B (Supplementary Fig. 4 online), indicating that the increase in surface NR2B was manifested at synapses. These findings suggest that Cdk5 knockout altered NMDAR constituency owing to increased synaptic and surface levels of NR2B. Consequently, total and ifenprodil-sensitive NMDAR-mediated currents were both elevated.
We also evaluated the contribution of increased NR2B-containing NMDAR-mediated current to the enhancement of plasticity in conditional Cdk5 knockout mice. As previously reported23, the NR2B-selective inhibitor ifenprodil had no effect on LTP induction in wild-type slices after a sufficient tetanus (48.8 ± 3.5% change from baseline, n = 3), confirming that NR2B-containing receptors are not obligatory for SC-CA1 LTP induction in normal adult mice. The LTP enhancement in Cdk5 knockout mice was reversed by ifenprodil (2.3 ± 3.5%) or a general NMDAR antagonist (AP5, 0.5 ± 0.9%; Fig. 4d). Therefore, the increase in synaptic plasticity resulting from Cdk5 knockout may be attributed to increased current through NR2B-containing receptors.
The increase in the amount of NR2B could be due to increased gene expression or reduced protein degradation. Analysis of Cdk5 knockout hippocampal tissue by RT-PCR showed no increase in NR2B mRNA (Supplementary Fig. 4), suggesting a post-transcriptional mechanism. NMDA is known to induce degradation of NR2B by calpain, a Ca2+-dependent protease24-26. We detected significantly less calpain-mediated NR2B degradation after NMDAR activation in hippocampal knockout slices than in controls (3.6 ± 12.8% and 22.5 ± 8.2% of wild type after 30 and 60 min, respectively; Fig. 4e). Furthermore, Cdk5/p25 markedly enhanced calpain-mediated cleavage of the NR2B C terminus in vitro (Fig. 4f). These ex vivo and in vitro findings indicate that Cdk5 normally facilitates calpain-mediated cleavage of NR2B.
Cdk5 could affect calpain-mediated cleavage of NR2B via either phosphorylation or direct binding. However, Cdk5 phosphorylated neither calpain (data not shown) nor NR2B in vitro (Supplementary Fig. 4), and it activated calpain in the absence of Mg2+ and ATP (Fig. 4f). Thus, cleavage enhancement was independent of kinase activity. On the other hand, NR2B, Cdk5, p35 and calpain coimmunoprecipitated with one another and the postsynaptic density marker PSD-95 in wild-type hippocampal lysates (Fig. 4g). Furthermore, Cdk5 knockout reduced the coimmunoprecipitation of NR2B with Cdk5 to 23.1 ± 2.6% that in wild-type controls (Supplementary Fig. 5 online). Although equal amounts of the Cdk5 cofactor p35 immunoprecipitated from wild-type and knockout lysates, the amount of NR2B associated with p35 was reduced to 48.4 ± 4.8% compared with wild type (Supplementary Fig. 5). These data indicate that NR2B, Cdk5, p35 and calpain are in a postsynaptic complex and Cdk5 mediates the interactions between NR2B and p35. NR2B–PSD-95 interaction ratios were unaffected by Cdk5 knockout (Supplementary Fig. 5). Finally, in vitro pull-down assays demonstrated that resin-conjugated NR2B bound Cdk5 and calpain, and resin-conjugated Cdk5 bound NR2B and calpain (Fig. 4h). These findings suggest that Cdk5 binds calpain and NR2B, thereby facilitating cleavage of NR2B. Thus, elevated NR2B levels in the Cdk5 knockout hippocampus likely resulted from the loss of these critical protein-protein interactions.
The effects of conditional Cdk5 knockout reveal a previously unknown role for Cdk5 in regulating glutamate receptor degradation. Previous reports separately implicate both Cdk5 (refs. 5-7,9) and CA1 hippocampal NMDARs1,15,16,20,22,23,27-29 in spatial learning and synaptic plasticity. The present study demonstrates that conditional loss of Cdk5 caused an increase in the amount of NR2B-containing NMDARs and related currents. Notably, transgenic overexpression of the NR2B subunit15 resulted in learning and plasticity enhancements that were similar to those of Cdk5 knockout mice. Our data also show that NR2B degradation was facilitated by interactions between Cdk5, calpain and NR2B. These findings are consistent with previous reports showing that calpain is a major component of the NMDAR complex30 and that calpain-mediated cleavage of NR2B decreases the number of functional NMDARs25,26. Cdk5 may function structurally to increase interactions between NR2B and calpain and to facilitate NR2B degradation in response to synaptic activity. Calpain also cleaves Cdk5’s activating cofactors31 and numerous Cdk5 substrates32. In addition, PSD-95 has been suggested to regulate NMDAR degradation by calpain33. Cdk5 has also been reported to modulate PSD-95 levels34 and may thus contribute to the modulation of NMDARs via multiple pathways.
Ultimately, the increase in NR2B-containing NMDARs may lengthen the integration time window for NMDAR coincidence detection15 and decrease the threshold to induce the long-term synaptic changes. Consistent with previous reports35,36, the increase in NMDAR-mediated current and reduction in plasticity threshold may facilitate information processing and learning. Although these findings advance our understanding of Cdk5’s role in learning and plasticity, it is important to note that to further understand the molecular mechanism of memory, it will be necessary to clarify the relationship between plasticity and learning suggested here. Furthermore, although we have focused on the role of Cdk5 in regulating functional plasticity via NMDARs, Cdk5 knockout likely affects learning via additional pathways. Nonetheless, these observations reveal a previously unknown molecular mechanism through which Cdk5 directly interacts with the NR2B NMDAR subunit and calpain to facilitate NR2B degradation. Through such interactions, Cdk5 may influence NMDAR constituency, synaptic plasticity and learning.
The floxed line was generated as previously described37. We crossed floxed Cdk5 and Cre-ERT mice14 and treated F3 male offspring with 4-hydroxytamoxifen (15 d, 66.67 mg per kg of body weight, intraperitoneal). All experiments were carried out with 10–12-week-old males, 2–4 weeks after treatment. PCR genotyping with primers encompassing loxP sites or within the Cre-ERT transgene were used to distinguish genotypes. Primers for Cdk5 allele genotyping had the following nucleotide sequences: 5′-GCAGGCCTTCGTTCCTCCC-3′ and 5′-CCTGACACGCTTCAGAGCC-3′. Primers for the Cre-ERT transgene had the following sequences: 5′-CAACCGAGCTGAAGCATTCTGCC-3′ and 5′-CATCAACGTTTTCTTTTCGGATC-3′. The template for the Cdk5 in situ hybridization probe was created from mouse brain cDNA using the primers 5′-GCGCGGTACCACCTACGGAACTGTGTT-3′ and 5′-GTCGCCGCGGCTTCACAATCTCAGGG-3′. Radiolabeled in situ hybridizations were carried out as described38,39. Quantitative RT-PCR was performed as described39 on whole hippocampus using the Cdk5 primers 5′-GGCTAAAAACCGGGAAACTC-3′ and 5′-CCATTGCAGCTGTCGAAATA-3′, the NR2B primers 5′-TCTGCCTTCTTAGAGCCATTCAG-3′ and 5′-AGACAGCTACAGCAGAGAC-3′, the NR2A primers 5′-CCTCGAACC CTTCAGCGCCT-3′ and 5′-AATGGCAGAGACAATGAGCAG-3′ and the NR1 primers 5′-GCAGGTGGAGTTGAGCACCATGTAC-3′ and 5′-AGCTTGTTGTCCCGCACA-3′. Cdk5 from glia and presynaptic terminals from nonhippocampal regions may have contributed to the Cdk5 signal measured using RT-PCR, immunoblot and activity analyses. For more accurate and higher-resolution analysis, FISH was carried out using digoxigenin-labeled mRNA probes and analyzed with confocal microscopy, essentially as decribed39.
Immunoblots, immunoprecipitations, immunoprecipitation-kinase assays, in vitro phosphorylation reactions and surface-labeling experiments were carried out using standard methodologies4,6,29,38. All immunoblot quantification is shown relative to tubulin loading controls.
Synaptosomal preparations were prepared as follows: hippocampi or cortex and striatum from three knockout or wild-type mice were dissected and homogenized with buffer A containing 5 mM Tris-HCl (pH 7.4), 1 mM MgCl2, 0.5 mM CaCl2, 1 mM NaF, 0.1 mM PMSF and protease cocktail inhibitor. The homogenized sample was centrifuged at 1,400g for 10 min to obtain the P1 and S1 fractions. The P1 fraction was resuspended with buffer A and centrifuged at 700g for 10 min to obtain the S1′ fraction. The S1 and S1′ fractions were mixed and centrifuged at 13,800g for 10 min to obtain the P2 fraction (pellet). The P2 fraction was resuspended in buffer B containing 6 mM Tris-HCl (pH 8.0), 1 mM NaF, 0.1 mM PMSF and 0.32 mM sucrose containing protease cocktail inhibitor, and layered on the top of a discontinuous sucrose gradient consisting of 0.85 M, 1.0 M and 1.25 M sucrose. The gradient was centrifuged at 82,500g for 2 h. The synaptosome fraction was collected and centrifuged at 38,200g for 20 min. The pellet was lysed with lysis buffer containing 1% SDS and 50 mM NaF, and subjected to SDS-PAGE and immunoblot analysis.
To examine NR2B biochemically, we purified the recombinant full-length C terminus of NR2B-GST (plasmid kindly provided by K.U. Bayer, University of Colorado Health Sciences Center), GST and protein phosphatase inhibitor-1 using standard protein purification techniques. Cdk5 purification and in vitro kinase reactions were carried out as described4. For coimmunoprecipitations, hippocampal lysates in RIPA buffer were incubated with primary antibodies, and then with protein-A–conjugated resin at 4 °C before washes and SDS-PAGE and immunoblot analysis. For binding assays, purified NR2B-GST or GST was rebound to glutathione-Sepharose resin. Cdk5 was reconjugated to Ni2+ agarose. Protein/resin and controls were washed, incubated in Ca2+-free buffer with potential binding partners (NR2B, Cdk5 and/or calpain-1) for 1 h at 4 °C, washed again, and then analyzed by immunoblotting. In vitro calpain-mediated NR2B degradation assays were carried out similarly as described24. Recombinant full-length C terminus of NR2B (0.5 μM) was incubated at 4 °C for 90 min in the presence of inactive Cdk5/p25 to allow for binding. The time course of NR2B degradation was conducted at 37 °C after the addition of Ca2+-activated calpain-1 (Sigma, 1:1,000). Cdk5 and calpain activities were blocked with ATP-free and Ca2+-free conditions, respectively.
Hippocampal slices were prepared as follows: 400-μm transverse hippocampal slices were prepared in ice-cold Krebs buffer and then allowed to recover in oxygenated Krebs buffer (30 °C) for 30 min, similarly as described40. Slices were then incubated in 50 μM NMDA for 30 or 60 min or in Krebs buffer for 60 min. Additional slices were incubated for 60 min in 50 μM NMDA and 20 μM calpeptin to confirm calpain dependence of cNR2B generation.
Mice were housed four or five per cage in a colony maintained at 23 °C with a 12-h light/dark cycle (lights on from 7:00 a.m. to 7:00 p.m.) and ad libitum food and water. All procedures were approved by the University of Texas Southwestern Institutional Animal Care and Use Committee. Behavior was carried out with the experimenter blind to genotype. Contextual and cue fear conditioning were performed using one or two 1-s, 0.8-mA foot shocks, respectively, as described41. For extinction trials, freezing to the conditioned context was assessed for 5 min, 24 h, 48 h, 72 h and 96 h after training and normalized to the 24-h-post-training value. Nociceptive responses were determined by measuring the stimulus threshold to elicit flinching, jumping and vocalizing as described41. Hotplate latency to hind-paw lick, jumping or vocalization was assessed after animals were placed on a 52 °C hotplate. Locomotor activity and elevated plus anxiety maze was carried out, similarly as described41. For the open-field anxiety test, each animal was placed in the center of a Plexiglas box under standard room-lighting conditions. Locomotor activity was quantified using the number of photoreceptor beam breaks and open-field anxiety was measured as the time spent in the center of the box using a standard computer tracking system. The water-maze task was carried out essentially as described41.
Transverse hippocampal slices were prepared in cutting saline (200 mM sucrose, 3 mM KCl, 1.4 mM NaH2PO4, 26 mM NaHCO3, 2 mM MgCl2, 2 mM CaCl2 and 10 mM glucose) and maintained in an interface chamber containing artificial cerebrospinal fluid42. Extracellular SC-CA1 voltage measurements in field recordings (fEPSPs) were synaptically evoked at 0.033 Hz and performed at 27 °C. Weak TBS (2×) and strong TBS (3×) were defined as 10 and 15 bursts, respectively (four pulses per burst at 100 Hz, 5-Hz burst frequency). Tetani were delivered at 50 and 100 Hz for 1 s. Basal input/output measurements, PPF and LTP experiments were carried out in the absence of any drugs, whereas fEPSPNMDA and fEPSP NMDA/AMPA measurements were performed in the presence of 2 μM SR95531, 1 μM atropine, 50 μM NiCl2, 10 μM nimodipine and 50 nM tetrodotoxin. NMDAR-mediated fEPSPs were recorded in the presence of 20 μM 6,7-dinitroquinoxaline-2,3-dione (DNQX) and blocked with 50 μM AP5. fEPSP data acquisition and analysis were carried out using the multielectrode MED64 hardware and software packages, essentially as described43. Additional analyses were performed using custom macros running in Igor Pro and Microsoft Excel. Whole-cell voltage-clamp SC-CA1 EPSCs were recorded, using instruments similar to those described42, to more selectively isolate NMDAR-mediated currents and eliminate threshold nonlinearities. Whole-cell recording electrodes (2–3 MΩ) were filled with a 7.3 pH intracellular cell solution containing 110 mM d-gluconic acid, 110 mM CsOH, 20 mM CsCl, 10 mM disodium phosphocreatine, 0.3 mM sodium GTP, 2 mM magnesium ATP, 10 mM HEPES, 0.5 mM EGTA and 5 mM lidocaine N-ethyl bromide. EPSCs were evoked at 0.05 Hz at 32 °C, and stimulus intensity was chosen to elicit 125–150-pA EPSCs at −70-mV holding potential. Before each sweep, a −5-mV step pulse was injected to assess and control for series resistance. τ for EPSCs was calculated with a single-exponential fit. Whole-cell recordings were carried out in the presence of 2 μM SR95531, 1 μM atropine, 50 μM NiCl2 and 10 μM nimodipine. NMDAR-mediated EPSCs were isolated at +40 or +60 mV in the presence of 20 μM DNQX and blocked with 50 μM AP5. NMDA/AMPA EPSC ratio was calculated under +40 mV/20 μM DNQX and −70 mV conditions. An alternative whole-cell SC-CA1 EPSC NMDA/AMPA ratio was calculated in the absence of DNQX: the NMDA component was measured as the area of the +40 mV EPSC from 45 ms to 350 ms in 2 μM SR95531, 1 μM atropine, 50 μM NiCl2 and 10 μM nimodipine. Ifenprodil-sensitive NMDAR-mediated currents were recorded at a +60 mV holding potential in the presence of 20 μM DNQX. Quantification for Iifen-sensitive was calculated as the area between INMDA and Iifen-resistant, normalized to INMDA. Whole-cell data acquisition and analysis were carried out using custom macros running in Igor Pro, Microsoft Excel and GraphPad Prism.
We thank B. Potts, S. Gold, J. Pick and M. Waung for assistance with experiments, D. Metzger for characterization of Cre-ERT line, C. Steffen for assistance with animal husbandry and K. Bayer for the NR2B clone. We thank A. Nairn and E. Nestler for comments and discussion and are grateful to the Medical Scientist Training Program at the University of Texas Southwestern Medical Center. This work was made possible by the US National Institutes on Drug Abuse National Research Service Award training grant (A.H.H.), US National Institutes of Health individual National Research Service Awards (D.R.B., C.N.), US National Alliance for Research on Schizophrenia and Depression Young Investigator awards (C.M.P. and D.C.C.), grant funding from the US National Institutes on Drug Abuse (P.G., D.C.C. and J.A.B.) and Mental Health (P.G., C.M.P. and J.A.B.) and the Ella McFadden Charitable Trust Fund at the Southwestern Medical Foundation (J.A.B).
Note: Supplementary information is available on the Nature Neuroscience website.
A.H.H. performed knockout optimization, histology, learning behavior testing, extracellular recordings, slice pharmacology, biochemistry and data analysis. D.R.B. established the mouse colony, optimized genotyping and histology and performed anxiety behavior testing. C.N. performed whole-cell voltage-clamp recordings. J.W.K. assisted in genotyping and histology. K.H. prepared synaptosomes and recombinant Cdk5. P.C. and P.G. provided mouse lines. C.M.P., D.C.C. and J.A.B. designed and supervised the experiments conducted in their laboratories. A.H.H. and J.A.B. prepared the manuscript. All authors contributed to experimental design, discussed the results and commented on the report.
COMPETING INTERESTS STATEMENT
The authors declare no competing financial interests.