Abstract
The opening-duration of the NMDA receptors implements Hebb’s synaptic coincidence-detection and is long thought to be the rate-limiting factor underlying superior memory. Here, we investigate the molecular and genetic determinants of the NMDA receptors by testing the ‘‘synaptic coincidence-detection time-duration’’ hypothesis vs. ‘‘GluN2B intracellular signaling domain’’ hypothesis. Accordingly, we generated a series of GluN2A, GluN2B, and GluN2D chimeric subunit transgenic mice in which C-terminal intracellular domains were systematically swapped and overexpressed in the forebrain excitatory neurons. The data presented in the present study supports the second hypothesis, the ‘‘GluN2B intracellular signaling domain’’ hypothesis. Surprisingly, we found that the voltage-gated channel opening-durations through either GluN2A or GluN2B are sufficient and their temporal differences are marginal. In contrast, the C-terminal intracellular domain of the GluN2B subunit is necessary and sufficient for superior performances in long-term novel object recognition and cued fear memories and superior flexibility in fear extinction. Intriguingly, memory enhancement correlates with enhanced long- term potentiation in the 10–100 Hz range while requiring intact long-term depression capacity at the 1–5 Hz range. Citation: Jacobs S, Cui Z, Feng R, Wang H, Wang D, et al. (2014) Molecular and Genetic Determinants of the NMDA Receptor for Superior Learning and Memory Functions. PLoS ONE 9(10): e111865. doi:10.1371/journal.pone.0111865 Editor: Ya-Ping Tang, Louisiana State University Health Sciences Center, United States of America Received July 22, 2014; Accepted October 6, 2014; Published October 31, 2014 Copyright: 2014 Jacobs et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: The authors confirm that all data underlying the findings are fully available without restriction. All relevant data are within the paper and its Supporting Information files. Funding: This work was supported by funds from the National Institute of Mental Health (MH060236), National Institute on Aging (AG024022, AG034663 & AG025918), USAMRA00002, and Georgia Research Alliance (all to JZT). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors can confirm that JZT is a member of the PLOS ONE Editorial Board. This does not alter the authors’ adherence to PLOS ONE editorial policies and criteria. * Email: [email protected] . These authors contributed equally to this work. and memory in both transgenic mice and rats [14–18]. The Introduction prevalent view in the field is that memory enhancement by N-methyl-D-aspartate (NMDA) receptors are known to be the GluN2B up-regulation is due to its longer opening duration in key modulators of synaptic plasticity in the forebrain regions [1–4] comparison to that of the GluN2A subunit. This enables the and act as the molecular gating switch for learning and memory GluN2B-containing NMDA receptors to be better coincidence [5,6]. It is widely accepted that their unique coincidence detection detectors. property allows them to impart Hebb’s rule on synapses, by However, the different structural motifs of GluN2 subunits are 2+ requiring the simultaneous pre-synaptic release of glutamate and known to regulate the NMDA receptors’ Mg dependency, 2+ the depolarization of the postsynaptic membrane to remove the channel opening duration, magnitude of Ca influx, as well as, 2+ extracellular Mg block [7]. NMDA receptors are composed of intracellular signaling cascades [10]. For example, the GluN2A 2+ two GluN1 subunits, as well as two GluN2 subunits [8]. In the and GluN2B subunits have high Mg dependency, whereas the 2+ adult forebrain regions, GluN2A and GluN2B subunits are the GluN2C and GluN2D subunits have much less Mg dependency 2+ main subunits available in excitatory synapses for receptor [13]. Thus, the extracellular Mg blockade of the GluN2A or complex formation [8,9], and are ideal for coincidence detection GluN2B-containing NMDA receptors suppresses NMDA-mediat- 2+ 2+ due to their strong Mg dependency [10–12]. influx at voltages close to the resting membrane potential ed Ca During postnatal brain development, the GluN2B subunits are allowing the cell to differentiate between correlated synaptic input the predominate subunits expressed specifically in excitatory and uncorrelated activity [7,19,20]. Recent studies have further neurons of the forebrain regions, such as the cortex and demonstrated that the C-terminals of GluN2A and GluN2B bind hippocampus [9,13]. As the animal develops into adulthood, to different downstream signaling molecules. This has led to a GluN2B expression decreases while GluN2A expression increases, greater appreciation for their contribution to synaptic plasticity resulting in an overall decrease in synaptic plasticity. Previously, and behavior [21,22]. To examine the effects of GluN2A on we have shown that an overexpression of the GluN2B subunit in learning and memory, we recently generated CaMKII promoter- the forebrain excitatory neurons enhances many forms of learning driven GluN2A transgenic mice and found profound long-term PLOS ONE | www.plosone.org 1 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement memory deficits in these mice, while their short-term memories Results remain unaffected [23]. Therefore, overexpression of GluN2A or Generation of transgenic mice expressing chimeric GluN2B in the mouse forebrain leads to impaired or enhanced 2B(CTR) 2A(CTR) 2B(CTR) memory function, respectively. These observations have raised GluN2A , GluN2B , or GluN2D subunits in several key questions, as to whether enhanced memories in the forebrain principal neurons GluN2B transgenic mice or impaired memories in GluN2A To investigate the potentially distinct roles of the C-terminal transgenic mice were due to their differences in NMDA receptor domains vs. the N-terminal and membrane domains in GluN2 channel-opening durations or their distinct intracellular signaling subunits in mediating memory enhancement, we created con- processes. Answers to this crucial question can be highly valuable structs encoding chimeric receptors based on GluN2B and for developing therapeutic strategies for preventing memory loss in GluN2A but with their respective CTDs replaced (denoted as 2A(CTR) 2B(CTR) patients. CTR) with each other’s (GluN2B and GluN2A , Currently, two hypotheses have been postulated to explain the respectively. We have created three new chimeric GluN2 observed memory enhancement in the GluN2B transgenic animals transgenic mouse lines. We used the same aCaMKII promoter or memory impairment in GluN2A transgenic mice [4]. One for driving transgene expression in forebrain excitatory neurons as we did for producing the GluN2B [16] and GluN2A transgenic hypothesis, known as the ‘‘coincidence-detection’’ hypothesis, 2A(CT) posits that because the GluN2B subunit makes the channel mice [23]. In the first transgenic line, termed Tg-GluN2B chimeric transgenic mice, the C-terminal domain of the GluN2B opening duration longer than that of the GluN2A subunit, the subunit has been swapped for the counterpart C-terminal domain GluN2B overexpression allows a greater coincidence detection of the GluN2A and overexpressed in the forebrain excitatory window, thereby leading to superior memory functions. The neurons (Figure 1A). This effectively pairs the opening duration of shorter coincidence-detection, such as in the GluN2A transgenic the GluN2B subunit with the signaling domain of the GluN2A mice, underlies impaired long-term memories. The second 2B(CT) subunit. In the second transgenic line, termed Tg-GluN2A , hypothesis is that the distinct intracellular domain of the GluN2B the C-terminal domain of the GluN2A subunit has been swapped subunit is responsible for the enhancements observed in the for the counterpart C-terminal domain of the GluN2B (Fig- GluN2B transgenic mice [4]. Several recent key observations ure 1A). This chimeric subunit possesses the GluN2A opening support this ‘‘intracellular domain hypothesis’’. Biochemical duration but with the signaling domain from the GluN2B subunit. studies have shown that the intracellular C-terminal domains of 2+ Additionally, to investigate the requirement of the Mg depen- the GluN2A and GluN2B subunits preferentially interact with dent synaptic coincidence-detection function for producing different downstream molecules and play distinct roles in synaptic GluN2B-mediated intracellular signaling, we created a third functions [24]. Conversely, studies using genetically truncated 2B(CT) transgenic mouse line, namely, Tg-GluN2D mice, in which GluN2A or GluN2B subunits demonstrate that the C-terminal the C-terminal domain of the GluN2B subunit has been fused to connections are essential for NMDA receptor function. The the N-terminal and membrane domain of the GluN2D subunit truncated subunits often act as functional knockouts of the whole 2+ 2B(CTR) (which is less Mg -dependent) denoted as GluN2D subunit [21]. Although several truncated C-terminal studies have (Figure 1A). focused on the mechanisms by which the GluN2 subunits mediate We confirmed the transgene integration into the genome of the NMDA receptor functions, the structural motifs crucial for their off-spring by Southern Blot analysis using Poly(A) probes learning and memory enhancement remain undefined. It is (Figure S1A) and Western Blot (Figure S1B). Next, we performed completely unknown as to whether and what degree the C- a series of in situ hybridization experiments to determine the terminal domain of the GluN2B would contribute to memory expression pattern of the transgenes in the mouse brains. As enhancement. shown, the transgenes are highly enriched in the cortex, striatum, In the present study, we set out to examine the above two and hippocampus, but not in hindbrain regions such as the hypotheses aimed at determining how and whether the molecular cerebellum (Figure 1C). The high expression transgenic mice were motifs underlying coincidence-detection time duration, or intra- crossed with C57BL/6J wild-type mice for at least 8 generations. cellular signaling cascades play a role in enhancing learning and These chimeric GluN2 transgenic offspring were found to grow memory. Our strategy is to swap or replace the C-terminal and breed normally, having similar adult weights to their wild-type cytoplasmic domain of the GluN2B subunit with the C-terminal littermates (Figure 1D) (Wt: n = 11, 29.1860.985 g; Tg-Glu- domain of the GluN2A subunit, or vice versa. Additionally, we 2B(CT) 2A(CT) N2A : n = 7, 30.0960.990 g; Tg-GluN2B : n = 10, have replaced the C-terminal domain of the GluN2D subunit with 2B(CT) 30.3460.648 g; Tg-GluN2D : n = 9, 29.4160.940 g), and the C-terminal domain of the GluN2B subunit effectively reducing being visually indistinguishable among them. Additionally, we also 2+ the Mg dependency of the receptor. We have produced and produced transgenic GluN2A overexpression mice (Tg-GluN2A analyzed five different GluN2 transgenic mouse lines, namely, mice) [23] and transgenic GluN2B overexpression mice (Tg- 2A(CT) GluN2A transgenic mice (Tg-GluN2A), Tg-GluN2B trans- GluN2B mice) [14–17] for comparisons on learning and memory genic mice, GluN2B transgenic mice (Tg-GluN2B), Tg-Glu- tests. These two transgenic mouse lines were also maintained on 2B(CT) 2B(CT) N2A transgenic mice, and Tg-GluN2D transgenic the same genetic background. mice. Our experiments suggest that the C-terminal domain of the 2B(CT) 2A(CT) The Tg-GluN2A , Tg-GluN2B , and Tg- GluN2B subunit is necessary and sufficient to produce memory 2B(CT) GluN2D mice showed no differences in the open field 2+ enhancement, as long as it is coupled to Mg dependent forms of behavioral paradigm, either in time spent in the center verses the GluN2 subunits such as GluN2A or GluN2B, while coupling of the periphery (Figure 1E) (center: Wt: n = 7, 117.91617.006 s; Tg- 2B(CT) 2A(CT) GluN2A’s C-terminal domain to GluN2B N-terminal and GluN2A : n = 6, 95.95612.893 s; Tg-GluN2B :n=5, 2B(CT) transmembrane domains lead to profound memory impairment. 78.74615.154 s; Tg-GluN2D : n = 6, 79.15613.483 s; pe- 2B(CT) Moreover, coupling of the C-terminal domain of the GluN2B riphery: Wt: 481.86616.994 s; Tg-GluN2A : 2A(CT) subunit to GluN2D subunit’s N-terminal and transmembrane 503.98612.901 s Tg-GluN2B : 520.96615.098 s; Tg- 2B(CT) domains lead to memory deficits. GluN2D : 520.65613.527 s), or in locomotor activity 2B(CT) (Figure 1F) (center: Wt: 1184.876218.32 cm; Tg-GluN2A : PLOS ONE | www.plosone.org 2 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement Figure 1. Constructs and basic behavioral assays of the GluN2 chimeric mice. (A) Illustration of the constructs used to create the 2B(CTR) 2A(CTR) 2B(CTR) GluN2A , GluN2B , and GluN2D chimeric subunits. (B) A point mutation was made on the cloning vector to induce an Aat II cutting sites to link the N-terminal and membrane domain to the C-terminal domain. After successfully joining the domains, the point mutation was restored to the original sequence. The arrow indicates the fusion position located in trans-membrane domain. (C) In situ hybridization of the transgene 2B(CT) 2A(CT) 2B(CT) expression in the wild-type mice (Wt), the Tg-GluN2A mice, the Tg-GluN2B mice and the Tg-GluN2D mice using SV-40 probes with a schematic of the receptor subunit expressed in the excitatory neurons. (D) No differences were found in the average adult body weight of the wild- 2B(CT) 2A(CT) 2B(CT) type mice, the Tg-GluN2A , Tg-GluN2B , and Tg-GluN2D mice. (E) The chimeric transgenic mice spent similar amounts of time as the wild-type mice in the center verses the periphery of the open field arena. (F) The chimeric transgenic mice and the wild-type mice showed similar 2B(CT) 2A(CT) 2B(CT) locomotion in the open field. (G) The Tg-GluN2A , Tg-GluN2B and Tg-GluN2D mice spent similar amounts of time in the closed arms and the open arms of the elevated plus maze as the wild-type mice. doi:10.1371/journal.pone.0111865.g001 PLOS ONE | www.plosone.org 3 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2A(CT) 2B(CT) 2A(CT) 945.796158.233 cm; Tg-GluN2B : 824.156147.123 cm; GluN2A : n = 10, 51.8563.192%; Tg-GluN2B : n = 10, 2B(CT) 2B(CT) Tg-GluN2D : 819.966195.070 cm; periphery: Wt: 49.8560.932%; Tg-GluN2D : n = 20, 52.5262.097%). 2B(CT) 3864.836460.03 cm; Tg-GluN2A : 3675.256196.950 cm; At the one hour retention session, the Tg-GluN2A, Tg- 2A(CT) 2B(CT) 2B(CT) Tg-GluN2B : 3294.386378.288 cm; Tg-GluN2D : GluN2B and Tg-GluN2A mice showed similar interest in 3458.346249.227 cm). This suggests that these transgenic mice the novel object as compared to the wild-type mice (Wt: n = 10, were normal in locomotor activity and anxiety. Additionally, no 60.9064.913%; Tg-GluN2A: n = 11, 59.8064.270%; Tg- 2B(CT) differences were found in the elevated plus maze paradigm, which GluN2B: n = 7, 57.3862.76%; Tg-GluN2A : n = 14, also measured for anxiety-like behavior (Figure 1G). Therefore, 59.1565.294%) demonstrating no changes in short-term recog- 2A(CT) the chimeric GluN2 mice were indistinguishable from their wild- nition memory. Whereas the Tg-GluN2B and Tg- 2B(CT) type littermates in growth, body weights, and these basic GluN2D mice show no preference for the novel object 2A(CT) 2B(CT) behaviors. (Tg-GluN2B : n = 12, 51.1564.808%; Tg-GluN2D : n = 22, 53.8562.535%), suggesting memory impairment in this test. Enhancement of long-term object recognition memory 2B(CT) At the 24 hour retention session, the Tg-GluN2A mice showed in Tg-GluN2A mice but impairments in long-term 2A(CT) 2B(CT) no preference for the novel object and significantly less interest in memory of the Tg-GluN2B and Tg-GluN2D it than the wild-type mice (Figure 2A) (GluN2A: n = 10, mice 50.0363.860%; F(2, 24) = 7.45, p = 0.003) as noted previously To investigate recognition memory functions in the transgenic [23], demonstrating their inability to form a long-term recognition mice, we tested the mice in a novel object recognition task for both 2B(CT) memory. As expected, the Tg-GluN2D mice also showed no short-term and long-term memory domains. During training, all preference for the novel object at the 24 hour retention session transgenic mouse groups showed comparable exploratory behav- spending significantly less time exploring the novel object than the ior and motivation for the task, exploring each object to a similar 2B(CT) wild-type mice (Wt: n = 10, 66.5663.610%; Tg-GluN2D : degree (Figure 2A) (Wt: n = 10, 51.5563.65%; Tg-GluN2A: 25 n = 20, 43.8062.566%; F(5, 63) = 6.36, p = 7.7610 ). The Tg- n = 10, 50.1560.932%; Tg-GluN2B: n = 9, 49.1761.611%; Tg- 2A(CT) GluN2B mice spent only slightly more time with the novel object than the familiar object (n = 12, 55.1064.821%). However, 2B(CT) Figure 2. Enhanced long-term recognition memory of the Tg-GluN2A mice and impaired long-term memory on the Tg- 2B(CT) GluN2D mice. (A) All groups of mice tested showed similar exploratory behavior in the training session. At the one hour retention session, the 2B(CT) 2A(CT) Tg-GluN2A, Tg-GluN2B and Tg-GluN2A mice showed similar interest in the novel object as the wild-type mice. Whereas the Tg-GluN2B and 2B(CT) 2B(CT) Tg-GluN2D mice show almost no preference for the novel object. At the 24 hour retention test, as expected the Tg-GluN2A and Tg-GluN2D 2B(CT) 2A(CT) mice showed no preference for the novel object. The Tg-GluN2B, The Tg-GluN2A and Tg-GluN2B mice all spent similar amounts of time 25 2B(CT) with the novel object. *p = 0.003, **p = 7.7610 .(B) In addition to the enhancement seen in the Tg-GluN2A mice at the 24 hour recall session, 2B(CT) these mice also showed enhanced recognition memory even at 3 days post-training over the wild-type mice. *p = 0.003. Whereas the GLUN2A mice show no preference for the novel object at 3 day or 7 days. doi:10.1371/journal.pone.0111865.g002 PLOS ONE | www.plosone.org 4 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2B(CT) the Tg-GluN2B and Tg-GluN2A mice showed similar the deficit in converting short-term contextual fear memory into memory of the novel object to the wild-type mice (Tg-GluN2B: long-term contextual fear memory due to expression of GluN2A. 2B(CT) n = 7, 66.4462.417%; Tg-GluN2A : n = 14, 59.7763.418%). 2A(CT) 2B(CT) This demonstrates impaired long-term recognition memory in the Tg-GluN2B and Tg-GluN2D mice showed 2B(CT) Tg-GluN2D mice. significant impairments in long-term cued fear memory, To determine the extent of the enhancement in the Tg- 2B(CT) whereas Tg-GluN2A showed enhanced memory 2B(CT) GluN2A mice, separate cohorts of mice were further used to To assess whether and how hippocampal-independent forms of test in their ability to retain the memory of the object over three- memories are affected by the N-terminal and C-terminal domain day and seven-day periods. Remarkably, at the three day retention properties, we used a new cohort of the mice and tested them in 2B(CT) tests, the Tg-GluN2A mice, like the Tg-GluN2B mice, spent cued fear conditioning task which required the mouse to associate significantly more time investigating the novel object (Figure 2B) an unconditioned stimulus (a shock) with a conditioned stimulus (a 2B(CT) (Tg-GluN2B: n = 7, 67.6464.337%; Tg-GluN2A : n = 14, tone). In the cued fear conditioning paradigm, all of the mice 62.7662.968%) than the wild-type mice (Wt: n = 10, exhibited little pre-tone freezing responses during the retention 47.9364.045%, F(2, 28) = 7.05, p = 0.003). However, the Tg- tests as they entered a novel chamber (Figure 3B) (Wt: n = 10, 2A(CT) GluN2B mice showed no interest in the novel object, 2.2260.997%; Tg-GluN2A: n = 15, 6.1761.734%; Tg-GluN2B: spending approximately equal time with both objects (Tg- 2B(CT) n = 7, 6.1761.251%; Tg-GluN2A : n = 11, 2.0260.758%; 2A(CT) GluN2B : n = 8, 52.7462.924%). At the seven day retention 2A(CT) 2B(CT) 2A(CT) Tg-GluN2B : n = 11, 1.2660.576%; Tg-GluN2D : session, the Wt, GluN2B, Tg-GluN2B , and Tg-Glu- 2B(CT) n = 12, 3.4761.373%). Upon the recall tone, all five types of N2A mice spent similar amounts of time investigating the transgenic mice exhibited significant amounts of freezing responses novel object (Wt: n = 10, 55.0564.096%; Tg-GluN2B: n = 6, 2B(CT) at the one hour retention session, comparable to that of wild-type 53.1263.373%; Tg-GluN2A : n = 14, 56.3663.344%; Tg- 2A(CT) mice (Wt: n = 10, 71.6764.843%; Tg-GluN2A: n = 18, GluN2B : n = 8, 54.7561.928%; F(2,35) = 0.21, p = 0.93). 58.0864.710%; Tg-GluN2B: n = 7, 62.0866.801%; Tg-Glu- This demonstrates the significant enhancement in long-term 2B(CT) 2A(CT) 2B(CT) N2A : n = 10, 67.2264.83%; Tg-GluN2B : n = 10, recognition memory in the Tg-GluN2A mice, to a similar 2B(CT) 60.0064.833%; Tg-GluN2D : n = 9, 61.1164.856%). This degree as the Tg-GluN2B mice did. shows that the transgenic mouse lines have normal short-term hippocampal-independent emotional memory and all are able to Normal contextual fear memory in the chimeric form an association between the tone (CS) and the shock (US). transgenic mice For one-day cued fear memory retention tests, a second cohort To investigate the emotional memory in the chimeric transgenic of similarly trained mice was placed into a novel enclosure and an mice, we tested the mice in a contextual fear conditioning task. identical tone to the training tone was presented. Interestingly, the This type of fear conditioning is hippocampal-dependent and is 2A(CT) Tg-GluN2A mice, the Tg-GluN2B mice and the Tg- often used to test short-term (one-hour) and long-term (one-day) 2B(CT) GluN2D mice demonstrated significantly less freezing than time points. In the training session, all of the mice displayed similar the wild-type mice (Figure 3B) (Wt: n = 9, 55.0263.030%; Tg- freezing responses immediately after the shock was delivered 2A(CT) GluN2A: n = 10,14.7565.189%; Tg-GluN2B : n = 10, (Figure 3A) (Wt: n = 13, 26.9266.126%; Tg-GluN2A: n = 15, 2B(CT) 14.7268.477%; Tg-GluN2D : n = 12, 16.6765.772%; F(5, 25.7563.146%; Tg-GluN2B: n = 30.0063.637%; Tg-Glu- 26 2B(CT) 56) = 27.03, p,1.0610 ). The Tg-GluN2A mice froze 2B(CT) 2A(CT) N2A : n = 13, 34.6165.155%; Tg-GluN2B : n = 12, significantly more than the wild-type mice (n = 13, 2B(CT) 25.0067.812%; Tg-GluN2D : n = 13, 23.1867.045%). At 71.5862.727%). Consistent with the previous studies (Tang et al, the one-hour retention session, the wild-type mice, the Tg- 1999), the Tg-GluN2B mice also showed enhanced cued memory 2A(CT) 2B(CT) GluN2A, Tg-GluN2B and Tg-GluN2D mice all over the wild-type mice (Tg-GluN2B: n = 8, 65.7565.918%, p, displayed similar freezing responses when they were returned to 0.05). These data demonstrate that the Tg-GluN2A, Tg- the shock chamber in the absence of footshock (Wt: n = 13, 2A(CT) 2B(CT) GluN2B and Tg-GluN2D mice have impaired long- 26.2864.444%; Tg-GluN2A: n = 11, 32.2264.789%; Tg-GluN2- term hippocampal-independent fear memory, whereas the Tg- 2A(CT) 2B(CT) B : n = 10, 25.2866.286%; Tg-GluN2D : n = 13, 2B(CT) GluN2A and Tg-GluN2B mice exhibited similarly enhanced 37.5165.415%). Interestingly, both the Tg-GluN2B and the Tg- long-term cued fear memories. 2B(CT) GluN2A mice spent significantly more time freezing than the wild-type mice (Tg-GluN2B: n = 7, 45.0662.823%; Tg-Glu- 2B(CT) 2B(CTR) Enhanced cued fear extinction in Tg-GluN2A mice N2A : n = 13, 52.5663.672%; F(6, 61) = 4.98, 2B(CT) over the wild-type mice p = 0.0007). This suggests that Tg- GluN2A mice, similar Fear extinction has been widely used as a test for assessing to Tg-GluN2B, exhibited enhanced 1-hr contextual fear memory. flexible learning behaviors. The extinction of learned fear requires At the one-day retention session, the Tg-GluN2B mice still the formation of new flexible relations, instead of forgetting or displayed significantly more freezing than the wild-type mice as erasing the established fear memories [25]. Because the Tg- previously reported (Figure 3A) (n = 8, 55.9464.911%, p = 0.039), 2A(CT) 2B(CT) GluN2A, Tg-GluN2B , and Tg-GluN2D mice were suggesting greater long-term contextual fear memory in these 2B(CT) 2A(CT) impaired in the one-day retention session, they were not used for mice. However, Tg-GluN2A , Tg-GluN2B and Tg- 2B(CT) the fear extinction experiment. Instead, we focused our investiga- GluN2D mice displayed similar freezing responses as those 2B(CT) tion of this form of learning on the Tg-GluN2A mice, the of the wild-type mice, indicating that all of these chimeric Tg-GluN2B mice, as well as the wild-type mice. transgenic mice have the normal 1-day hippocampal-dependent 2B(CT) In this fear extinction task, we used a five trial extinction contextual fear memories. (Figure 3A) (Tg-GluN2A : n = 13, 2A(CT) 44.8764.779%; Tg-GluN2B : n = 12, 41.9167.003%; Tg- paradigm in which the animals were repeatedly exposed to the 2B(CT) training chamber (contextual extinction) or the tone in a novel GluN2D : n = 12, 47.9267.050%). On the contrary, the Tg- GluN2A mice demonstrated reduced freezing responses during the context (cued extinction) without the delivery of the shock in either contextual recall (Wt: n = 14, 43.6564.382%; Tg-GluN2A: context. We first tested the mice in the contextual fear extinction n = 13, 23.5463.811%, F(5, 66) = 3.65, p = 0.005), suggesting paradigm. We found that the Tg-GluN2B mice initially showed PLOS ONE | www.plosone.org 5 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2A(CT) 2B(CT) Figure 3. Selectively impaired emotional memory in the Tg-GluN2B and Tg-GluN2D mice. (A) The mice showed similar 2A(CT) freezing responses immediately following the US. At the one hour retention session, the wild-type mice, the Tg-GluN2A, Tg-GluN2B and Tg- 2B(CT) 2B(CT) GluN2D mice all displayed similar freezing responses. Interestingly, both the Tg-GluN2B and the Tg-GluN2A mice spent significantly more time freezing. At the 24 hour recall session only the Tg-GluN2A mice demonstrated a diminished freezing response to the context in which the shock was delivered. *p = 0.005, **p = 0.0007. (B) The mice tested also showed similar pre-tone freezing responses and similar freezing at the one hour 2A(CT) 2B(CT) contextual recall. At the 24 hour recall session the Tg-GluN2A mice, the Tg-GluN2B mice and the Tg-GluN2D mice demonstrated 2B(CT) significantly less freezing than the wild-type mice, whereas the Tg-GluN2A mice and Tg-GluN2B froze significantly more than the wild-type mice. 26 2B(CT) *p,0.05, **p,1.0610 , *** p = 0.0007. (C) The Tg-GluN2A mice showed quicker fear extinction than the wild-type mice in the contextual fear 2B(CT) extinction paradigm *p,0.05, **p,0.01. (D) The Tg-GluN2A mice showed quicker fear extinction to the CS than the wild-type mice in the contextual fear extinction paradigm *p,0.05, **p,0.01, ***p,0.001. doi:10.1371/journal.pone.0111865.g003 significantly more freezing than the wild-type mice and the Tg- the wild-type mice did not significantly decrease their freezing 2B(CT) GluN2A mice 24 hours after the training session (Figure 3C) response from the first to the second exposure session. It is noted 2B(CT) (1: Wt: n = 15, 40.1963.852%; Tg-GluN2B: n = 8, that the Tg-GluN2A mice exhibited significantly less 2B(CT) 55.9464.911%; Tg-GluN2A : n = 13, 44.8764.779%). In- freezing, as determined by ANOVA analysis, than both the Tg- terestingly, over the extinction trials both the Tg-GluN2B and the GluN2B mice and their wild-type littermates (Wt: 38.7065.161%; 2B(CT) Tg-GluN2A mice significantly decreased their freezing F(2, 33) = 3.56, p = 0.04). All groups of mice spent significantly responses as early as the second session 2 hours later in less time freezing in the third exposure than the second exposure comparison to the first trial (2: Tg-GluN2B: 38.2965.161%; (3: Wt: 27.4165.034%, p = 0.01; Tg-GluN2B: 29.1264.294%, 2B(CT) 2B(CT) p = 0.014; Tg-GluN2A : 23.2964.409%; p = 0.002), whereas p = 0.03; Tg-GluN2A : 10.2662.632%, p = 0.007). The mice PLOS ONE | www.plosone.org 6 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement continued to decrease their freezing responses in the fourth (4: Wt: Enhanced 10 Hz induced LTP observed in the Tg- 2B(CT) 2B(CT) 25.7462.125%; Tg-GluN2B: 12.6962.576%; Tg-GluN2A : GluN2A CA1 region 10.4762.816%, p = 0.002) and fifth exposures (5: Wt: We first performed LTP and LTD studies on the Tg- 2B(CT) 2B(CT) 2B(CT) 21.4863.325%; Tg-GluN2B: 6.5761.300%; Tg-GluN2A : GluN2A mice. In the Tg-GluN2A mice, LTP can be 9.1962.276%). ANOVA analysis indicated that in both the fourth readily induced by 100 Hz stimulation (Figure 4A) (Wt: n = 6/ and fifth exposures, the wild-type mice had significantly higher 2B(CT) 3(# of slices/# of animals), 135.267.6%; Tg-GluN2A : freezing responses than both the Tg-GluN2B and Tg-Glu- n = 7/4, 146.568.7%). Interestingly, a significant increase in 2B(CT) N2A mice (4: F(2, 33) = 11.91, p = 0.0001; 5: F(2, 33) LTP was observed in the transgenic mice, compared to that of = 8.08, p = 0.001). These data demonstrate that the Tg-GluN2B wild-type slices, in response to the 10 Hz frequency stimulation 2B(CT) mice and the Tg-GluN2A mice had better hippocampal- 2B(CT) (Figure 4B) (Wt: n = 7/4, 103.2613.0%; Tg-GluN2A : dependent fear flexibility learning ability than the wild-type mice. n = 5/3, 150.1617.0%). Additionally, a significant difference Next, we exposed the mice to the tone in a novel environment was further observed at 5 Hz stimulation (Figure 4C) (Wt: 2B(CT) for cued fear extinction learning. The Tg-GluN2A mice, 2B(CT) n = 4/3, 94.461.8%; Tg-GluN2A : n = 6/4, 115.563.9%). similar to the Tg-GluN2B mice, showed significantly faster cued There is no statistical difference at the 3 Hz stimulation fear extinction than the wild-type mice (Figure 3D). In the first 2B(CT) (Figure 4D) (Wt: n = 7/3, 70.3611.3%; Tg-GluN2A : exposure to the tone 24 hours after training, the Tg-GluN2B and n = 5/3, 79.2613.3%) or 1 Hz stimulation (Figure 4E) (Wt: 2B(CT) Tg-GluN2A mice showed significantly higher freezing 2B(CT) n = 7/5, 82.461.6%; Tg-GluN2A : n = 6/3, 95.4615.6%). responses than the wild-type mice (1: Wt: n = 12, 2B(CT) Overall, we found that the Tg-GluN2A mice show little 58.1063.323%; Tg-GluN2B: n = 8, 70.8765.017%; Tg-Glu- 2B(CT) difference in the LTD, except at 5 Hz, but show significantly N2A : n = 13, 71.5862.727%, F(2, 30) = 4.97, p = 0.01). enhanced LTP around at 10 Hz frequency (Figure 4F). These At the second exposure, two hours after the first exposure, the 2B(CT) 2B(CT) data indicate that the GluN2A overexpression produced Tg-GluN2B mice and the Tg-GluN2A mice significantly synaptic changes that were more similar to that of GluN2B reduced their freezing responses to the presentation of the tone, overexpression in the transgenic mice and rats [14,16,18]. whereas the wild-type mice did not (2: Wt: 57.4164.436%; Tg- 2B(CT) GluN2B: 41.4662.836%, p = 0.002; Tg-GluN2A : Enhanced 10 Hz LTP and diminished 1–3 Hz LTD in the 48.7165.34%, p = 0.003), again suggesting the faster fear 2A(CT) extinction in these transgenic mice. Remarkably, both the Tg- Tg-GluN2B CA1 region 2B(CT) 2A(CT) GluN2B and Tg-GluN2A mice further decreased freezing We then measured synaptic plasticity in the Tg-GluN2B 2A(CT) from the second to the third exposures as well (3: Wt: CA1 slices. Overexpression of GluN2B significantly in- 49.5464.323%; Tg-GluN2B: 20.1461.278%, p = 0.0004; Tg- creased LTP versus their wild-type littermates at both 100 Hz 2B(CT) GluN2A : 35.0464.958%, p = 0.0005). The wild-type mice (Figure 5B) (n = 13/6, 176.6616.2%; Figure 5A) and 10 Hz spent significantly more time freezing in the third exposure than (n = 5/3, 180.7633.0%) frequencies. Interestingly, while 10 Hz 2B(CT) the Tg-GluN2B and Tg-GluN2A mice (F(2, 30) = 9.87, response did not differ, LTD was also significantly impaired as 2B(CT) p = 0.0005). The Tg-GluN2B mice and the Tg-GluN2A compared to the wild-type hippocampal slices at 5 Hz (n = 4/2, mice continued to decrease their freezing responses in the fourth 121.662.2%; Figure 5C), 3 Hz (n = 6/3, 103.6613.8%; Fig- (4: Wt: 45.1464.340%; Tg-GluN2B: 7.7561.175%; Tg-Glu- ure 5D) and 1 Hz (n = 21/13, 114.266.6%; Figure 5E). This 2B(CT) 2A(CT) N2A : 30.7764.118%) and fifth exposures (5: Wt: shows that although the Tg-GluN2B mice have significantly 2B(CT) 42.8264.972%; Tg-GluN2B: 5.3460.939%; Tg-GluN2A : increased LTP, they also have significantly blocked 1 Hz and 3 Hz 29.2764.342%). It is worth noting that the Tg-GluN2B had the induced LTD (summarized in Figure 5F). This decrease in LTD in 2A(CT) faster extinction learning. ANOVA analysis revealed that the Tg- Tg-GluN2B slices was more similar to that seen in the Tg- GluN2B mice demonstrated significantly less freezing than the GluN2A mice [23]. 2B(CT) Tg-GluN2A mice and the wild-type mice in both the fourth (F(2, 30) = 19.32, p = 4.0610 ) and fifth exposures (F(2, 30) 2B(CT) Impaired 5 Hz responses in the Tg-GluN2D CA1 = 16.22, p = 1.7610 ). region 2B(CT) Finally, we examined the effects of GluN2D overexpres- Basic electrophysiological properties in the chimeric sion on CA1 plasticity. Since GluN2D has very weak magnesium GluN2 mice dependency but much greater opening duration, it would lead to GluN2A and GluN2B subunits’ contribution to synaptic 2+ significantly more Ca influx into the postsynaptic sites. We plasticity has been intensely investigated in the CA1 region using 2B(CT) performed LTP and LTD measurements on the Tg-GluN2D both pharmacological and genetic methods [26–29]. We took hippocampal slices. Interestingly, we found no differences at either advantage of the existing knowledge in the literature and the 100 Hz (Figure 6A) (n = 8/6, 12868.6%) or the 10 Hz investigated and compared how various chimeric transgenic frequency between the transgenic and control littermates (Fig- overexpressions would affect the bidirectional control of synaptic ure 6B) (n = 6/3, 125.5612.6%). However, at the 5 Hz frequency, plasticity in the CA1 region. To investigate the basic electrophys- a small, but significant LTP was observed in the transgenic slices, 2B(CT) iological properties in the hippocampus of the Tg-GluN2A , in comparison to the LTD induced in the slices from the control 2A(CT) 2B(CT) Tg-GluN2B , and Tg-GluN2D mice, we recorded from littermates (Figure 6C) (n = 5/4, 123.563.6%). However, there the CA1 Schaffer collaterals of the mouse hippocampus. We found were no significant differences observed in LTD at 1 Hz the input-output properties (Figure S2A), as well as the paired stimulation (Figure 6E) (n = 8/4, 88.667.6%), 3 Hz (Figure 6D) pulse facilitation (Figure S2B) from each genotype were similar to (n = 7/5, 87.6612.4%). The summary graphs show the overall those of the wild-type controls, thereby demonstrating normal 2B(CT) similarities between the Tg-GluN2D mice and their wild- presynaptic function and basal transmissions in these transgenic type counterparts, except in its 5 Hz frequency response mice. We then systematically measured the long-term potentiation (Figure 6F). (LTP) and long-term depression (LTD) in the CA1 slices from each mouse line. PLOS ONE | www.plosone.org 7 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2B(CT) 2B(CT) Figure 4. Enhanced LTP in the Tg-GluN2A mouse hippocampal slices. A. Slightly enhanced LTP seen in the Tg-GluN2A mice with a 2B(CT) 1 s 100 Hz stimulation. B. Significantly enhanced LTP was seen in the Tg-GluN2A mice when a 10 Hz stimulation was applied from 10 s. C–E. No changes in LTD were seen in the 5 Hz, 3 Hz, or 1 Hz stimulation protocols. F. A summary plot of the % change in fEPSP slope versus the frequencies. doi:10.1371/journal.pone.0111865.g004 systematically analyzing three chimeric GluN2 transgenic mice Discussion together with Tg-GluN2A and Tg-GluN2B mice, we have tested The NMDA receptor is widely known as the key coincidence- two major hypotheses, namely, synaptic coincidence-detection/ detector at central synapses to implement Hebb’s learning rule. calcium influx hypothesis vs. GluN2B C-terminal intracellular The channel opening-duration and the level of membrane signaling hypothesis in gating memory enhancement. Our 2+ depolarization, determines the amount of Ca that influxes into experiments have revealed several novel insights into the the cell [30,31]. In this study, we have identified the critical relationships between GluN2 subunit motifs, synaptic plasticity, molecular motifs of the GluN2 subunits essential for achieving and memory enhancement. learning and memory enhancement in the adult mouse brain. By PLOS ONE | www.plosone.org 8 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2A(CT) Figure 5. Enhanced LTP and diminished LTD in the Tg-GluN2B mouse hippocampal slices. (A) A slight increase in LTP was seen in 2A(CT) the 100 Hz stimulation protocol in the Tg-GluN2B mice. (B) When a 10 Hz stimulation was applied for 10 s a significant increase in LTP was seen 2A(CT) 2A(CT) in the Tg-GluN2B mice over their wild-type littermates. (C) LTD was diminished at 5 Hz stimulation in the Tg-GluN2B mice. (D) LTD was 2A(CT) significantly diminished at the 3 Hz stimulation protocol. (E) At the 1 Hz stimulation, the Tg-GluN2B mice show significantly diminished LTD. (F) A summary plot of the % change in fEPSP slope versus the frequencies. doi:10.1371/journal.pone.0111865.g005 The ‘‘synaptic coincidence-detection’’ hypothesis reflects the subunits possessed the GluN2B intracellular signaling capability predominant view in the field as the rate-limiting factor in but with the other key properties such as the shorter opening determining learning and memory capability. It posits that duration and the voltage-dependency from the GluN2A and because the GluN2B subunit makes the channel opening duration GluN2D, respectively. 2+ longer than that of the GluN2A subunit, GluN2B overexpression As predicted, because GluN2D has greatly reduced Mg allows a greater coincidence detection window, thereby leading to dependency, which renders synaptic coincidence-detection inef- 2B(CT) superior memory functions [4]. Because the N-terminal and fective, GluN2D transgenic mice indeed exhibited memory transmembrane domains of the GluN2 subunits are known to be deficits in novel object recognition (in both the short-term and 2+ crucial for controlling voltage-gating and ion (Ca ) influx long-term form) and long-term cued fear conditioning memory duration, we replaced the GluN2B N-terminal domains with (although the contextual fear memory seemed to be normal). either GluN2A or GluN2D while retaining its wild-type C- These observations have provided evidence that synaptic coinci- terminal intracellular domain. As such, these two chimeric GluN2 dence-detection is necessary for producing memory enhancement PLOS ONE | www.plosone.org 9 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 2B(CT) Figure 6. Diminished LTD at the 5 Hz range in the Tg-GluN2D mouse hippocampal slices. (A) No changes from the wild-type 2B(CT) hippocampal slices were seen in the LTP of the Tg-GluN2D mouse hippocampal slices when a 100 Hz stimulation was applied. (B) When a 10 Hz 2B(CT) stimulation was applied for 10 s there, again was no significant change in LTP observed in the Tg-GluN2D mice over their wild-type littermates. 2B(CT) (C) LTD was diminished at 5 Hz stimulation in the Tg-GluN2D mice. (D) LTD was not significantly diminished at the 3 Hz stimulation protocol. (E) 2B(CT) At the 1 Hz stimulation, the Tg-GluN2D mice showed no significant differences in LTD. (F) A summary plot of the % change in fEPSP slope versus the frequencies. doi:10.1371/journal.pone.0111865.g006 via the GluN2B intracellular signaling cascades. Without proper enhancement, as long as the GluN2B C-terminal domain is magnesium dependent voltage gating, the presence of the transducing the signaling. It is important to note here that 2B(CTD) overexpressed GluN2B domain from the chimeric GluN2D Punnakkal et al. found little differences in the whole cell currents subunit still could not produce optimal synaptic changes for of similar chimeric constructs, with only a slight decrease in the memory enhancement. peak amplitude of a similar GluN2AB construct from that of the On the other hand, we were quite surprised that the Tg- GluN2A wildtype subunit [32]. No changes in the peak 2B(CT) GluN2A transgenic mice exhibited a very similar memory amplitude of a similar GluN2BA construct over that of the enhancement phenotype to those of the Tg-GluN2B mice. The GluN2B wildtype subunit. Additionally, deactivation times different channel opening-durations derived from GluN2A and remained unchanged between the wildtype and chimeric GluN2B subunits’ N-terminal and transmembrane domains are receptors. Importantly, they also concluded that the peak not the most critical factor in determining the memory opening probability appeared to be determined by the GluN2 PLOS ONE | www.plosone.org 10 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement N-terminal domain [32]. Therefore, these genetic experiments beyond 60 minutes [16]. This indicates the longer time duration of 2+ have shown that Mg -dependent coincidence-detection function, channel opening (thereby more calcium influx) via the GluN2B N- but not necessarily the opening-duration difference between terminal and pore does make larger and more stable LTP in GluN2A and GluN2B, is prerequisite for achieving learning and response to 100 Hz stimulation [42,43]. However, at the 10 Hz memory enhancement in the adult brain. stimulation range, the amount of calcium influx via the GluN2A N-terminal and pore domains (coupled to GluN2B C-terminal Interestingly our present study has provided clear evidence supporting the second hypothesis that is known as the ‘‘GluN2B region) can produce the similarly larger LTP in the Tg- 2B(CT) intracellular domain’’ hypothesis [4,33]. Two separate pieces of GluN2A mice as that of the Tg-GluN2B mice in comparison evidence came from our behavioral analyses of the Tg-GluN2- to that of the wild-type controls. This 10 Hz stimulation frequency 2A(CT) 2B(CT) B and Tg-GluN2A transgenic mice. First, we found can be particularly interesting because we have observed that fear 2B(CT) that the Tg-GluN2A mice had enhanced object recognition conditioning-induced firing increase in CA1 pyramidal cells is memory and emotional memory. These phenotypes are very mostly in the range of 5,30 Hz [6,44]. This behaviorally relevant similar to those of the Tg-GluN2B mice (and also Tg-GluN2B rats) frequency range deserves special investigation for memory (Figure 7). On the contrary, when the C-terminal domain of the enhancement in future experiments both in the hippocampus GluN2B subunit was replaced by that of the GluN2A subunit, as and other brain regions such as the prefrontal cortex and 2A(CT) we did in Tg-GluN2B mice, this swap led to profound amygdala. In addition, contrary to LTD produced by 5 Hz 2B(CT) memory deficits in novel object recognition test and long-term stimulation in the wild-type slices, Tg-GluN2D slices cued fear memories. These memory deficits mirrored those of Tg- exhibited a significant switch to LTP. Taken together, these GluN2A mice [23]. These subunits-swap experiments, by extend- findings have provided additional support for the notion that the ing to learning and memory enhancement, are consistent with GluN2B C-terminal domain plays a key role in regulating LTP other reports that the intracellular domains of the GluN2 subunits [26,45–48], and more importantly, our study has further defined, play critical roles in mediating different functions, such as synaptic for the first time, its essential link to memory enhancement. localization, clustering, signal transduction, and behaviors [22,33– While it is evident that the C-terminal of the GluN2B subunit 39]. Therefore, our studies suggest that both ‘‘synaptic coinci- plays a crucial role in producing synaptic potentiation, we found 2A(CT) dence-detection’’ hypotheses and ‘‘GluN2B intracellular signal- that Tg-GluN2B mice had larger, more robust, LTP not ing’’ hypothesis are mutually complementary in term of explaining only at 10 or 100 Hz. Intriguingly, such a swap also promoted an the molecular determinants for memory enhancement. overall shift toward potentiation even in response to lower Two additional conceptual insights have also been obtained on frequencies. As a result, the ability to produce LTD at 1–3 Hz 2A(CT) how different molecular motifs of the overexpressed GluN2 frequency range is greatly impaired in Tg-GluN2B slices. subunits regulate the levels and degrees of LTP or LTD over a These findings show that longer opening duration achieved by the wide range of stimulation frequencies. Despite multiple pharma- overexpressed GluN2B N-terminal and pore domains, but coupled cological and knockout approaches to analyzing GluN2A and to the GluN2A intracellular signaling cascade, brings a greater GluN2B on regulating LTP and LTD [40], few were done under potentiation but at the cost of losing synaptic depression capacity. the context of examining its relationship with cognitive enhance- This is in stark contrast with the normal LTD in Tg-GluN2B or 2B(CT) ment [16,18,41]. Here, we consistently found that GluN2B and Tg-GluN2A in response to 1 or 3 Hz stimulation. This 2B(CT) GluN2A overexpression enhanced LTP in the range of strongly suggests that increased calcium influx (via the GluN2B N- 10 Hz and/or 100 Hz range without significantly affecting 1 Hz terminal and core domains) is useful to produce bigger LTP, but its or 3 Hz LTD. It is noteworthy to point out that that Tg- effect on 1 Hz LTD critically depends on whether the downstream 2B(CT) GluN2A seemed to produce larger 100 Hz LTP in the initial signaling cascade is mediated by the GluN2A C-terminal tail or 20,30 minutes range than that of wild-type slices, but become GluN2B C-terminal tail. In other words, under such circumstance, indistinguishable by the 40-minutes time points. Interestingly, the the presence of chimeric GluN2A C-terminal domain, but not Tg-GluN2B overexpression tended to produce much larger chimeric GluN2B C-terminal domain, can override LTD. This 100 Hz induced LTP in comparison to the wild-type mice well 2B(CT) Figure 7. Summary of LTP, LTD and behavioral tasks results. The Tg-NRA mice had enhanced LTP, as well as enhanced long-term 2A(CT) recognition memory and contextual fear conditioning. The Tg-GluN2B mice have significantly impaired LTD resulting in impaired short-term 2B(CT) recognition memory and impaired long-term cued fear conditioning. The Tg-GluN2D mice have decreased LTD at 5 Hz and impaired long-term recognition memory and long-term cued fear memory. doi:10.1371/journal.pone.0111865.g007 PLOS ONE | www.plosone.org 11 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement novel insight adds to the notion that GluN2A may have a general be valuable to the current efforts in developing and optimizing ability to drive toward LTP [49–54]. memory enhancement strategies. In addition, by taking advantage of the correlational analysis between synaptic changes and memory performances, our present Methods study has uncovered two detailed insights into the memory Production of Transgenic Mice enhancement strategy: first, bigger LTP would lead to better We have produced three chimeric GluN2 subunit constructs learning and memory, however, only if the LTD ability remains for the present study. In the first two constructs, the N-terminal intact. This is supported by the observation that bigger CA1 LTP and transmembrane domains of GluN2A or GluN2D subunit is associated with better memory in Tg-GluN2B and Tg- 2B(CT) were fused with the C-terminal domain from the GluN2B GluN2A mice while their LTD was not altered. Second, if 2B(CTR) 2A(CTR) subunit, termed GluN2A and GluN2D , respec- LTP enhancement results in overriding or diminishing LTD 2A(CT) tively. In the third construct, we also fused the N-terminal and capacity, such as those observed in Tg-GluN2B , it would 2A(CT) transmembrane domain of GluN2B subunit with the C-terminal also lead to memory deficits. The Tg-GluN2B phenotypes 2A(CTR) domain from the GluN2A subunit, termed GluN2B (see are more similar to the knockout of PSD-95 which also leads to Figure 1A). The fusion site was located near the end of the larger LTP, lack of 1 Hz LTD, and memory deficits (i.e. [55,56]). fourth transmembrane domain, just before the C-terminal Our recent characterization of Tg-GluN2A mice showed that domains begin. For making the constructs, we first introduced overexpression of GluN2A results in no change in 100 Hz LTP or a point mutation to create a unique Aat II cutting site for the 1 Hz, but greatly impaired 3 or 5 Hz LTD. These mice also fusing of the given C-terminal domain (Figure 1B). Upon exhibited long-term memory deficits, while short-term memories 2B(CT) successful ligation, the point mutation was mutated back to its remained mostly normal. Our Tg-GluN2D mice, which also original sequence. These chimeric transgene constructs were showed 5 Hz LTP responses instead of either no change or LTD, driven by the forebrain-specific aCaMKII promoter for as in the wild-type mice, were also profoundly impaired in long- targeting their expression to the excitatory neurons in the term memory. These observations support the ‘‘LTD-memory forebrain regions such as the cortex and hippocampus. trace sculpting’’ hypothesis, that the weakening of uncorrelated The chimeric constructs were created by first introducing a synaptic connections would reduce the background ‘‘noise’’ while point-mutation at the site to create a unique Aat II cutting site for enabling the stabilization (or crystallization) of the learning-related swapping the NT and CT domains. Upon successful ligation the synaptic patterns [23]. However, it is important to note that our point mutation was swapped back to the original sequence. The current electrophysiological recordings were limited to the CA1 modified subunit was targeted for forebrain expression by the region. Given the fact that we used the CaMKII promoter to drive CaM-kinase II (CaMKII) promoter as previously described the transgenes, electrophysiological analyses should be extended in [16,67]. The founding line of transgenic animals was produced future experiments into other brain regions such as the amygdala by pronuclear injection of a linearized chimeric transgene vector and prefrontal cortex from which cued fear learning and fear into C57BL/6J zygotes similar to previously described [16,68]. A extinction are processed. Clearly, simple correlation between CA1 total of seven independent mouse founder lines (three lines for synaptic plasticity and memory are likely not sufficient for 2B(CTR) 2B(CT) GluN2A termed ‘‘Tg-GluN2A ’’ mice, two lines for counting memory enhancement and thus, any extrapolation 2A(CTR) 2A(CT) GluN2B termed ‘‘Tg-GluN2B ’’ mice, and two lines should be only taken with great caution. In addition, little is 2B(CTR) 2B(CT) for the GluN2D termed ‘‘Tg-GluN2D ’’ mice). All known about how any of the artificial stimulation paradigms for these lines gave successful germline transmissions. The genotypes producing LTP or LTD can be translated into real-time memory of the transgenic mice were determined by PCR analysis of a tail patterns. Recent successful decoding of real-time fear memory biopsy. The transgene was detected using the SV40 poly(A) traces in the hippocampal CA1 from the wild-type mice and the sequence, as previously described [16,18,41]. Southern blotting forebrain excitatory neuron-specific NMDA receptor inducible was used to confirm the transgene integration in to the transgenic knockout mice, have revealed many fundamental insights how the mouse line. Western blotting of the forebrain regions (cortex and NMDA receptors regulate real-time memory code and memory hippocampus) was visualized with either a polyclonal GluN2A C- engrams [6,44]. It would be of great interest to use such brain terminal antibody (Upstate/Millipore) or a polyclonal GluN2B C- decoding technologies to investigate the various transgenic mice terminal antibody (Millipore). For the present electrophysiological described here. and behavioral experiments, we used a high expression line chosen In summary, our above experiments have identified the key 2B(CT) 2A(CT) from the Tg-GluN2A mice, the Tg-GluN2B mice, and molecular and genetic determinants that would be necessary 2B(CT) the Tg-GluN2D mice that have been crossed with C57BL/ and sufficient for achieving superior learning and memory 6J wildtype mice for at least 8 generations. For in situ ability in the adult brain. Although transgenetic methods are hybridizations, brains from the transgenic mice and wild-type unlikely to be used for human clinical settings, the C-terminal littermates were isolated and 20 mm sections were prepared using region of the GluN2B subunit contains many important sites for a cryostat. The slices were hybridized to the [a S] oligonucleotide various molecular interaction including with CaMKII, cdk5, probe which hybridized to the untranslated artificial intron region and Kinesin superfamily protein 17 (KIF17) [57–60]. Indeed, in the transgene similarly to previously described [16,23]. manipulations of cdk5 and KIF17 which result in upregulation of GluN2B also resulted in memory enhancement [61,62]. More recently, researchers have taken a novel, dietary approach to Behavioral Experiments up-regulate GluN2B expression in the brain via elevating brain Mice were maintained in a temperature and humidity magnesium [63]. They showed that the compound, magnesium controlled vivarium with a 12:12 light-dark cycle. All testing was threonate, can cross the blood brain barrier efficiently and boost done during the light phase with 3–5 month old animals. Mice GluN2B expression in the neurons, and subsequent memory were allowed free access to food and water, except during improvement in both aging and wild-type mice [64,65]. This experimental procedures. Mice were extensively handled prior to compound is currently under clinical trials [66]. Therefore, it is any testing paradigm. Separate cohorts were used for each study conceivable that knowledge gained from the present study will and each recall time point unless otherwise stated. All testing PLOS ONE | www.plosone.org 12 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement procedures were conducted in sound dampened, dimly lit To test the contextual freezing exhibited by the animal, at the behavioral rooms. Experimenters were blind to the genotype of described time (1 hour or 24 hours) the trained mice were placed the animals. This study was carried out in strict accordance with back into the shock chamber for 5 minutes while their freezing the recommendations in the Guide for the Care and Use of response was monitored. The mice were then placed into a novel Laboratory Animals of the National Institutes of Health. The chamber and monitored for their freezing response (pre-tone) for protocols were approved by the Institutional Animal Care and Use 3 minutes before the onset of the CS tone for 3 minutes. During Committee of the Georgia Regents University. the tone the animal’s freezing response was monitored to test the cued fear retention. Open Field Freezing was judged as the complete immobility of the animal, 2B(CT) except for movement necessary for respiration. The mice were One cohort of Tg-GluN2A, Tg-GluN2B, Tg-GluN2A , 2A(CT) 2B(CT) then returned to their home cage for either 1 hour or 24 hours. At Tg-GluN2B , and Tg-GluN2D mice and their wild-type littermates were individually placed into a 50 cm L650 cm the described time the mice were returned to the chamber for measurement of the contextual freezing. The mice were then W625 cm H white Plexiglas open field arena. The mouse was placed in a novel chamber and the tone was delivered for 3 min, allowed to explore for ten minutes. The time that the mouse spent in the center and periphery was determined. The periphery of the during which their cued freezing response was monitored. To test the fear extinction of the animals the same recall testing paradigm open field was considered to be the first four inches along the wall, while the center of the open field was the square inside this area was repeated at 2 hour intervals for four additional trials. [69]. Additionally, the distance traveled by the mouse was determined using Biobserve Viewer II software. Statistical analysis of behavioral data All behavioral data are presented as mean 6 SEM. Significance Elevated Plus Maze was determined by ANOVA analysis with Tukey-Kramer, or a The elevated plus maze consisted of a black Plexiglas ‘‘plus’’ Student’s t-test. P values of ,0.05 were considered significant. maze approximately 60 cm above the floor, with each arm measuring 30 cm in length and 10 cm wide. Two opposite arms Hippocampal Slice Recordings were left open, with the other two arms being enclosed on three Transverse slices of the hippocampus were rapidly prepared 2B(CT) 2A(CT) sides. The ambient room lighting was 75 lux. The amount of time from wild-type and Tg-GluN2A , Tg-GluN2B , and Tg- 2B(CT) the mice spent within the enclosed arms was recorded, as well as GluN2D mouse lines (3,6 months old) and maintained in an the amount of time the animal spent in the open arms [69]. The interface chamber at 28uC and were subfused with artificial cerebral times were used to determine a preference index. spinal fluid (ACSF, 124 mM NaCl, 4.4 mM KCl, 2.0 mM CaCl , 1.0 mM MgSO , 25 mM NaCHO , 1.0 mM Na HPO and 4 3 2 4 Novel Object Recognition 10 mM glucose) and bubbled with 95% O and 5% CO . Slices 2 2 The behavioral paradigm was the same as previously described were kept in the recording chamber for at least two hours. A bipolar [16,70]. The mice were individually habituated to a 50 cm tungsten stimulating electrode was placed in the stratum radiatum L650 cm W625 cm H open field apparatus for 10 minutes a day in the CA1 region. A glass microelectrode (3–12 MV) filled with for three days. On the first testing day, the mice were placed into ACSF was used to measure the extracellular field potentials in the the open field with two identical objects for 5 minutes. The time stratum radiatum. Test response elicited at 0.02 Hz. Current they spent exploring each object was recorded. At the described intensity (0.5–1.2 mA) which produced 30% of maximal response retention time the mice were placed back into the open field arena was used for studies of PPF and synaptic plasticity at different with one of the familiar objects used in training, and one novel frequencies. Various interpulse intervals (20–400 msec) were used object, and allowed to explore for 5 minutes. The time they spent for measuring PPF. Low-frequency stimulation of (5 Hz for 3 min, with each object was recorded and used to determine a preference 3 Hz for 300 s, or 1 Hz for 900s) was then used to produce index. Different groups of mice were used for the each retention depotentiation [73]. Long term potentiation was induced by tetanic session. stimulation (100 Hz for 1 s and 10 Hz stimulation for 10 s). Data are expressed as mean 6 SEM. One-way ANOVA (with Duncan’s Fear Conditioning multiple range test for post hoc comparison) and Student’s t-test were used for statistical analysis. The detailed procedures were the same An operant chamber (25 cm L625 cm W638 cm H) equipped with activity monitors and camera was used. The flooring was a 24 as described (Tang et al. 1999; Shimizu, et al. 2000; Wang et al. bar shock grid with a speaker, shock generator, and photo-beam 2003; Wang, et al. 2008). scanner (MedAssociates). The chamber was located in a sound damping isolation box. The apparatus was thoroughly cleaned Supporting Information with 70% ethanol between mice to avoid any olfactory cues. Figure S1 Conformation of the integration of the Freezing was monitored by the software and confirmed by the transgene. (A) Southern Blot analysis of Tg-GluN2(A/B), Tg- experimenter. GluN2(B/A), and Tg-GluN2(D/B) mice. The numbers indicate Testing procedures were similar to those previously described the positive control and the copy number. (B) Western Blot [16,71,72]. Animals were habituated to the testing environment analysis of the chimeric animals showing enhanced expression of for 5 minutes one day before testing. On the day of training the the GluN2A C-terminal domain in the Tg-GluN2A, Tg- mice were placed into the chamber and allowed to explore for GluN2(B/A) mice and no enhancement of the GluN2A C- 5 minutes. Then the mice were exposed to a conditioned stimulus terminus in the Tg-GluN2(A/B) and Tg-GluN2(D/B) mice. (C) (CS, 85 dB tone at 2800 Hz), with the unconditioned stimulus Western Blot analysis of the chimeric animals showing enhanced (US, a scrambled foot shock at 0.75 mA) occurring the last expression of the GluN2B C-terminal tail in the Tg-GluN2(A/B), 2 seconds of the CS. The mice were allowed to stay in the and Tg-GluN2(D/B) mice of the expression in the wild-type mice. chamber for 30 seconds after the CS/US pairing to monitor (TIF) immediate freezing. PLOS ONE | www.plosone.org 13 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement Figure S2 Electrophysiology of hippocampal slices. (A) Acknowledgments There were no significant differences in the basal synaptic The authors would like to thank Philip Wang and Shuqin Zhang for his transmission as seen in the CA3-CA1 input-output curve between valuable assistances with behavioral experiments, as well as Fengying the wildtype mice and the transgenic mice. (B). The paired-pulse Huang for animal colony maintenance. facilitation was unchanged between the wildtype and the chimeric transgenic mice indicating that the presynaptic function is Author Contributions unchanged. Conceived and designed the experiments: ZC JZT. Performed the (TIF) experiments: SJ ZC RF HW DW. Analyzed the data: SJ ZC RF HW DW. Contributed reagents/materials/analysis tools: JZT. Wrote the paper: SJ JZT. References 1. Bliss TV, Collingridge GL (1993) A synaptic model of memory: long-term 27. Hrabetova S, Serrano P, Blace N, Tse HW, Skifter DA, et al. (2000) Distinct potentiation in the hippocampus. Nature 361: 31–39. NMDA receptor subpopulations contribute to long-term potentiation and long- 2. Stevens CF, Sullivan J (1998) Synaptic plasticity. Curr Biol 8: R151–153. term depression induction. J Neurosci 20: RC81. 3. Bear MF, Malenka RC (1994) Synaptic plasticity: LTP and LTD. Curr Opin 28. Xu Z, Chen RQ, Gu QH, Yan JZ, Wang SH, et al. (2009) Metaplastic Neurobiol 4: 389–399. regulation of long-term potentiation/long-term depression threshold by activity- 4. Tsien JZ (2000) Linking Hebb’s coincidence-detection to memory formation. dependent changes of NR2A/NR2B ratio. J Neurosci 29: 8764–8773. Curr Opin Neurobiol 10: 266–273. 29. Peng Y, Zhao J, Gu QH, Chen RQ, Xu Z, et al. (2010) Distinct trafficking and 5. Wang H, Hu Y, Tsien JZ (2006) Molecular and systems mechanisms of memory expression mechanisms underlie LTP and LTD of NMDA receptor-mediated consolidation and storage. Prog Neurobiol 79: 123–135. synaptic responses. Hippocampus 20: 646–658. 6. Zhang H, Chen G, Kuang H, Tsien JZ (2013) Mapping and Deciphering Neural 30. Wigstrom H, Gustafsson B (1986) Postsynaptic control of hippocampal long- Codes of NMDA Receptor-Dependent Fear Memory Engrams in the term potentiation. J Physiol (Paris) 81: 228–236. Hippocampus. PLoS One 8: e79454. 31. Lynch MA, Littleton JM (1983) Possible association of alcohol tolerance with 7. Mayer ML, Westbrook GL, Guthrie PB (1984) Voltage-dependent block by increased synaptic Ca2+ sensitivity. Nature 303: 175–176. Mg2+ of NMDA responses in spinal cord neurones. Nature 309: 261–263. 32. Punnakkal P, Jendritza P, Kohr G (2012) Influence of the intracellular GluN2 C- terminal domain on NMDA receptor function. Neuropharmacology 62: 1985– 8. Monyer H, Sprengel R, Schoepfer R, Herb A, Higuchi M, et al. (1992) Heteromeric NMDA receptors: molecular and functional distinction of subtypes. Science 256: 1217–1221. 33. Halt AR, Dallapiazza RF, Zhou Y, Stein IS, Qian H, et al. (2012) CaMKII binding to GluN2B is critical during memory consolidation. EMBO J 31: 1203– 9. Sheng M, Cummings J, Roldan LA, Jan YN, Jan LY (1994) Changing subunit composition of heteromeric NMDA receptors during development of rat cortex. 1216. Nature 368: 144–147. 34. Sheng M (1996) PDZs and receptor/channel clustering: rounding up the latest suspects. Neuron 17: 575–578. 10. Cull-Candy SG, Leszkiewicz DN (2004) Role of distinct NMDA receptor 35. Kennedy MB (1997) The postsynaptic density at glutamatergic synapses. Trends subtypes at central synapses. Sci STKE 2004: re16. Neurosci 20: 264–268. 11. Erreger K, Dravid SM, Banke TG, Wyllie DJ, Traynelis SF (2005) Subunit- specific gating controls rat NR1/NR2A and NR1/NR2B NMDA channel 36. Kennedy PR, Bakay RA (1997) Activity of single action potentials in monkey motor cortex during long-term task learning. Brain Res 760: 251–254. kinetics and synaptic signalling profiles. J Physiol 563: 345–358. 12. Dingledine R, Borges K, Bowie D, Traynelis SF (1999) The glutamate receptor 37. Kornau HC, Seeburg PH, Kennedy MB (1997) Interaction of ion channels and ion channels. Pharmacol Rev 51: 7–61. receptors with PDZ domain proteins. Curr Opin Neurobiol 7: 368–373. 13. Monyer H, Burnashev N, Laurie DJ, Sakmann B, Seeburg PH (1994) 38. Steigerwald F, Schulz TW, Schenker LT, Kennedy MB, Seeburg PH, et al. Developmental and regional expression in the rat brain and functional (2000) C-Terminal truncation of NR2A subunits impairs synaptic but not properties of four NMDA receptors. Neuron 12: 529–540. extrasynaptic localization of NMDA receptors. J Neurosci 20: 4573–4581. 14. Cui Y, Jin J, Zhang X, Xu H, Yang L, et al. (2011) Forebrain NR2B 39. Martel MA, Ryan TJ, Bell KF, Fowler JH, McMahon A, et al. (2012) The overexpression facilitating the prefrontal cortex long-term potentiation and subtype of GluN2 C-terminal domain determines the response to excitotoxic enhancing working memory function in mice. PLoS One 6: e20312. insults. Neuron 74: 543–556. 15. Jacobs SA, Tsien JZ (2012) Genetic overexpression of NR2B subunit enhances 40. Shipton OA, Paulsen O (2014) GluN2A and GluN2B subunit-containing social recognition memory for different strains and species. PLoS One 7: e36387. NMDA receptors in hippocampal plasticity. Philos Trans R Soc Lond B Biol 16. Tang YP, Shimizu E, Dube GR, Rampon C, Kerchner GA, et al. (1999) Sci 369: 20130163. Genetic enhancement of learning and memory in mice. Nature 401: 63–69. 41. Cao X, Cui Z, Feng R, Tang YP, Qin Z, et al. (2007) Maintenance of superior 17. Tang YP, Wang H, Feng R, Kyin M, Tsien JZ (2001) Differential effects of learning and memory function in NR2B transgenic mice during ageing. enrichment on learning and memory function in NR2B transgenic mice. Eur J Neurosci 25: 1815–1822. Neuropharmacology 41: 779–790. 42. Zhang XL, Sullivan JA, Moskal JR, Stanton PK (2008) A NMDA receptor 18. Wang D, Cui Z, Zeng Q, Kuang H, Wang LP, et al. (2009) Genetic glycine site partial agonist, GLYX-13, simultaneously enhances LTP and enhancement of memory and long-term potentiation but not CA1 long-term reduces LTD at Schaffer collateral-CA1 synapses in hippocampus. Neurophar- depression in NR2B transgenic rats. PLoS One 4: e7486. macology 55: 1238–1250. 19. Nowak L, Bregestovski P, Ascher P, Herbet A, Prochiantz A (1984) Magnesium 43. Berberich S, Jensen V, Hvalby O, Seeburg PH, Kohr G (2007) The role of gates glutamate-activated channels in mouse central neurones. Nature 307: 462– NMDAR subtypes and charge transfer during hippocampal LTP induction. 465. Neuropharmacology 52: 77–86. 20. Gielen M, Siegler Retchless B, Mony L, Johnson JW, Paoletti P (2009) 44. Chen G, Wang LP, Tsien JZ. (2009). Neural population-level memory traces in Mechanism of differential control of NMDA receptor activity by NR2 subunits. the mouse hippocampus. PLoS One. 4(12): e8256. Nature 459: 703–707. 45. Clayton DA, Mesches MH, Alvarez E, Bickford PC, Browning MD (2002) A 21. Sprengel R, Suchanek B, Amico C, Brusa R, Burnashev N, et al. (1998) hippocampal NR2B deficit can mimic age-related changes in long-term Importance of the intracellular domain of NR2 subunits for NMDA receptor potentiation and spatial learning in the Fischer 344 rat. J Neurosci 22: 3628– function in vivo. Cell 92: 279–289. 3637. 22. Ryan TJ, Kopanitsa MV, Indersmitten T, Nithianantharajah J, Afinowi NO, et 46. Kohr G, Jensen V, Koester HJ, Mihaljevic AL, Utvik JK, et al. (2003) al. (2013) Evolution of GluN2A/B cytoplasmic domains diversified vertebrate Intracellular domains of NMDA receptor subtypes are determinants for long- synaptic plasticity and behavior. Nat Neurosci 16: 25–32. term potentiation induction. J Neurosci 23: 10791–10799. 23. Cui Z, Feng R, Jacobs S, Duan Y, Wang H, et al. (2013) Increased NR2A: 47. Foster KA, McLaughlin N, Edbauer D, Phillips M, Bolton A, et al. (2010) NR2B ratio compresses long-term depression range and constrains long-term Distinct roles of NR2A and NR2B cytoplasmic tails in long-term potentiation. memory. Sci Rep 3: 1036. J Neurosci 30: 2676–2685. 24. Kennedy MB, Beale HC, Carlisle HJ, Washburn LR (2005) Integration of 48. Gardoni F, Mauceri D, Malinverno M, Polli F, Costa C, et al. (2009) Decreased biochemical signalling in spines. Nat Rev Neurosci 6: 423–434. NR2B subunit synaptic levels cause impaired long-term potentiation but not 25. Falls WA, Miserendino MJ, Davis M (1992) Extinction of fear-potentiated long-term depression. J Neurosci 29: 669–677. startle: blockade by infusion of an NMDA antagonist into the amygdala. 49. Liu HN, Kurotani T, Ren M, Yamada K, Yoshimura Y, et al. (2004) J Neurosci 12: 854–863. Presynaptic activity and Ca2+ entry are required for the maintenance of NMDA 26. Bartlett TE, Bannister NJ, Collett VJ, Dargan SL, Massey PV, et al. (2007) receptor-independent LTP at visual cortical excitatory synapses. J Neurophysiol Differential roles of NR2A and NR2B-containing NMDA receptors in LTP and 92: 1077–1087. LTD in the CA1 region of two-week old rat hippocampus. Neuropharmacology 50. Erreger K, Chen PE, Wyllie DJ, Traynelis SF (2004) Glutamate receptor gating. 52: 60–70. Crit Rev Neurobiol 16: 187–224. PLOS ONE | www.plosone.org 14 October 2014 | Volume 9 | Issue 10 | e111865 Molecular Determinants of Memory Enhancement 51. Berberich S, Punnakkal P, Jensen V, Pawlak V, Seeburg PH, et al. (2005) Lack 63. Slutsky I, Abumaria N, Wu LJ, Huang C, Zhang L, et al. (2010) Enhancement of of NMDA receptor subtype selectivity for hippocampal long-term potentiation. learning and memory by elevating brain magnesium. Neuron 65: 165–177. J Neurosci 25: 6907–6910. 64. Abumaria N, Yin B, Zhang L, Li XY, Chen T, et al. (2011) Effects of elevation 52. Li YH, Han TZ (2008) Glycine modulates synaptic NR2A- and NR2B- of brain magnesium on fear conditioning, fear extinction, and synaptic plasticity containing NMDA receptor-mediated responses in the rat visual cortex. Brain in the infralimbic prefrontal cortex and lateral amygdala. J Neurosci 31: 14871– Res 1190: 49–55. 53. Santucci DM, Raghavachari S (2008) The effects of NR2 subunit-dependent 65. Abumaria N, Luo L, Ahn M, Liu G (2013) Magnesium supplement enhances NMDA receptor kinetics on synaptic transmission and CaMKII activation. spatial-context pattern separation and prevents fear overgeneralization. Behav PLoS Comput Biol 4: e1000208. Pharmacol 24: 255–263. 54. Gerkin RC, Lau PM, Nauen DW, Wang YT, Bi GQ (2007) Modular 66. Cyranoski D (2012) Testing magnesium’s brain-boosting effects. Nature News: competition driven by NMDA receptor subtypes in spike-timing-dependent Nature Publishing Group. plasticity. J Neurophysiol 97: 2851–2862. 67. Tsien JZ, Chen DF, Gerber D, Tom C, Mercer EH, et al. (1996) Subregion- and 55. Migaud M, Charlesworth P, Dempster M, Webster LC, Watabe AM, et al. cell type-restricted gene knockout in mouse brain. Cell 87: 1317–1326. (1998) Enhanced long-term potentiation and impaired learning in mice with 68. Tsien JZ, Huerta PT, Tonegawa S (1996) The essential role of hippocampal mutant postsynaptic density-95 protein. Nature 396: 433–439. CA1 NMDA receptor-dependent synaptic plasticity in spatial memory. Cell 87: 56. Carlisle HJ, Fink AE, Grant SG, O’Dell TJ (2008) Opposing effects of PSD-93 1327–1338. and PSD-95 on long-term potentiation and spike timing-dependent plasticity. 69. Wang LP, Li F, Wang D, Xie K, Shen X, et al. (2011) NMDA receptors in J Physiol 586: 5885–5900. dopaminergic neurons are crucial for habit learning. Neuron 72: 1055–1066. 57. Gho M, McDonald K, Ganetzky B, Saxton WM (1992) Effects of kinesin 70. Wang H, Feng R, Phillip Wang L, Li F, Cao X, et al. (2008) CaMKII activation mutations on neuronal functions. Science 258: 313–316. state underlies synaptic labile phase of LTP and short-term memory formation. 58. Setou M, Nakagawa T, Seog DH, Hirokawa N (2000) Kinesin superfamily Curr Biol 18: 1546–1554. motor protein KIF17 and mLin-10 in NMDA receptor-containing vesicle 71. Cui Z, Wang H, Tan Y, Zaia KA, Zhang S, et al. (2004) Inducible and reversible transport. Science 288: 1796–1802. NR1 knockout reveals crucial role of the NMDA receptor in preserving remote 59. Hawasli AH, Bibb JA (2007) Alternative roles for Cdk5 in learning and synaptic memories in the brain. Neuron 41: 781–793. plasticity. Biotechnol J 2: 941–948. 72. Rampon C, Jiang CH, Dong H, Tang YP, Lockhart DJ, et al. (2000) Effects of 60. Yin X, Takei Y, Kido MA, Hirokawa N (2011) Molecular motor KIF17 is environmental enrichment on gene expression in the brain. Proc Natl Acad fundamental for memory and learning via differential support of synaptic Sci U S A 97: 12880–12884. NR2A/2B levels. Neuron 70: 310–325. 73. Brandon EP, Zhuo M, Huang YY, Qi M, Gerhold KA, et al. (1995) 61. Hawasli AH, Benavides DR, Nguyen C, Kansy JW, Hayashi K, et al. (2007) Hippocampal long-term depression and depotentiation are defective in mice Cyclin-dependent kinase 5 governs learning and synaptic plasticity via control of NMDAR degradation. Nat Neurosci 10: 880–886. carrying a targeted disruption of the gene encoding the RI beta subunit of 62. Wong RW, Setou M, Teng J, Takei Y, Hirokawa N (2002) Overexpression of cAMP-dependent protein kinase. Proc Natl Acad Sci U S A 92: 8851–8855. motor protein KIF17 enhances spatial and working memory in transgenic mice. Proc Natl Acad Sci U S A 99: 14500–14505. PLOS ONE | www.plosone.org 15 October 2014 | Volume 9 | Issue 10 | e111865
Journal
PLoS ONE
– Unpaywall
Published: Oct 31, 2014
