Ca 2+ Permeable AMPA Receptor Induced Long-Term Potentiation Requires PI3/MAP Kinases but Not Ca/CaM- Dependent Kinase II

Transcription

1 Ca 2+ Permeable AMPA Receptor Induced Long-Term Potentiation Requires PI3/MAP Kinases but Not Ca/CaM- Dependent Kinase II Suhail Asrar 1,2, Zikai Zhou 1,2, Wei Ren 3, Zhengping Jia 1,2 1 Neurosciences and Mental Health, Hospital for Sick Children, Toronto, Ontario, Canada, 2 Department of Physiology, University of Toronto, Toronto, Ontario, Canada, 3 College of Life Science, Shaanxi Normal University, Xi an, China Abstract Ca 2+ influx via GluR2-lacking Ca 2+ -permeable AMPA glutamate receptors (CP-AMPARs) can trigger changes in synaptic efficacy in both interneurons and principle neurons, but the underlying mechanisms remain unknown. We took advantage of genetically altered mice with no or reduced GluR2, thus allowing the expression of synaptic CP-AMPARs, to investigate the molecular signaling process during CP-AMPAR-induced synaptic plasticity at CA1 synapses in the hippocampus. Utilizing electrophysiological techniques, we demonstrated that these receptors were capable of inducing numerous forms of long-term potentiation (referred to as CP-AMPAR dependent LTP) through a number of different induction protocols, including high-frequency stimulation (HFS) and theta-burst stimulation (TBS). This included a previously undemonstrated form of protein-synthesis dependent late-ltp (L-LTP) at CA1 synapses that is NMDA-receptor independent. This form of plasticity was completely blocked by the selective CP-AMPAR inhibitor IEM-1460, and found to be dependent on postsynaptic Ca 2+ ions through calcium chelator (BAPTA) studies. Surprisingly, Ca/CaM-dependent kinase II (CaMKII), the key protein kinase that is indispensable for NMDA-receptor dependent LTP at CA1 synapses appeared to be not required for the induction of CP-AMPAR dependent LTP due to the lack of effect of two separate pharmacological inhibitors (KN-62 and staurosporine) on this form of potentiation. Both KN-62 and staurosporine strongly inhibited NMDA-receptor dependent LTP in control studies. In contrast, inhibitors for PI3-kinase (LY and wortmannin) or the MAPK cascade (PD98059 and U0126) significantly attenuated this CP-AMPAR-dependent LTP. Similarly, postsynaptic infusion of tetanus toxin (TeTx) light chain, an inhibitor of exocytosis, also had a significant inhibitory effect on this form of LTP. These results suggest that distinct synaptic signaling underlies GluR2-lacking CP-AMPAR-dependent LTP, and reinforces the recent notions that CP- AMPARs are important facilitators of synaptic plasticity in the brain. Citation: Asrar S, Zhou Z, Ren W, Jia Z (2009) Ca 2+ Permeable AMPA Receptor Induced Long-Term Potentiation Requires PI3/MAP Kinases but Not Ca/CaM- Dependent Kinase II. PLoS ONE 4(2): e4339. doi: /journal.pone Editor: Georges Chapouthier, L université Pierre et Marie Curie, France Received September 15, 2008; Accepted December 18, 2008; Published February 3, 2009 Copyright: ß 2009 Asrar 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. Funding: This work was supported by grants from the Canadian Institutes of Health Research (CIHR MOP-42396) and Ontario Mental Health Foundation. Suhail Asrar was supported by the Ontario Graduate Scholarship (OGS). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. zpjia@hotmail.com Introduction The a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) subtype glutamate receptors are the principal mediators of the fast excitatory synaptic transmission in the mammalian CNS and are important for the expression of various forms of longlasting synaptic plasticity, including long-term potentition (LTP) [1 3]. AMPA receptors (AMPARs) are heteromeric complexes assembled from four distinct subunits (GluR1 4), of which GluR2 is particularly interesting because it dictates a number of important biophysical and biochemical properties [4 7]. Hence, AMPARs lacking edited GluR2 are Ca 2+ permeable (CP-AMPAR) with higher conductance and inwardly rectifying I/V relationships. These GluR2-lacking CP-AMPARs are widely expressed in the CNS (including interneurons, stellate and glial cells) where they can contribute to synaptic transmission and changes in synaptic efficacy [8] as well as induce multiple forms of synaptic plasticity, including LTP [9 15]. Subunit composition switching from GluR2-lacking to GluR2-containing AMPARs was demonstrated as fundamental to plasticity in cerebellar stellate cells [13] and the ventral tegmental area [16]. CP-AMPARs were also shown to mediate the induction and expression of LTP at neuron-glia synapses [17]. At interneuron synapses, CP-AMPARs are believed to play a crucial role in an unusual form of anti-hebbian LTP [18]. Recent studies have also indicated that CP-AMPARs are expressed in cortical and hippocampal pyramidal neurons [8]. At developing hippocampal mossy fiber-pyramidal synapses, the selective loss of CP-AMPARs underlies a depolarization-induced form of LTD [19]. Additionally, mossy fiber-interneuron synapses were shown to demonstrate concomitant forms of LTD from either NMDARs or CP-AMPARs that were dependent on Ca 2+ influx [20], suggesting that both types of calcium permeable receptors could work in parallel to collectively contribute to synaptic plasticity in regions where they coexist. Of particular relevance to the present study is the finding that CP-AMPARs are transiently recruited to CA1 synapses by LTP-inducing stimulations where they are involved in the consolidation of this NMDAR-dependent LTP [22 24, but see 21, 25]. Finally, the PLoS ONE 1 February 2009 Volume 4 Issue 2 e4339

2 CP-AMPAR Dependent LTP expression of CP-AMPARs and the resultant Ca 2+ influx are also associated with a number of pathophysiological states, including ischemia, epileptic seizures and drug addiction [15,26 29]. Despite the importance of CP-AMPARs in synaptic regulation and pathology, the molecular processes activated by Ca 2+ influx through these receptors is unknown. In this study, we took advantage of genetically altered mice lacking GluR2 (GluR22/2) or having a reduced level of GluR2 (GluR2+/2) to present evidence that a distinct synaptic signaling underlies this CP- AMPAR-dependent LTP. Results CP-AMPAR-dependent LTP at CA1 synapses We have previously demonstrated that GluR2 mutants exhibit high Ca 2+ permeability and inward rectification as well as an enhanced form of plasticity at CA1 synapses facilitated by Ca 2+ influx through both NMDARs and CP-AMPARs [10,30]. The utilization of an NMDAR antagonist such as D,L-AP5 allowed us to specifically isolate plasticity induced through CP-AMPARs, and thus investigate the molecular mechanisms underlying long-lasting synaptic increases induced by Ca 2+ influx through these receptors. Therefore, in the present study we used hippocampal slices prepared from these mice to investigate CP-AMPAR-induced synaptic plasticity by performing both field and whole-cell patch-clamp recordings at the CA1 synapses. In wild-type animals, a brief high frequency stimulation (HFS, 2 trains of 100 Hz lasting 1 second) produced a long-lasting increase in field excitatory postsynaptic potentials (fepsps) that could be completely blocked by application of 100 mm D,L-AP5 (vehicle = %; D,L-AP5 = %; P,0.001) during the induction phase, indicating that this form of LTP was completely NMDAR-dependent (Figure 1A). In contrast, a significant amount of LTP was generated with the same induction protocol in GluR22/ 2mice (16668%; P,0.001) despite the presence of 100 mm D,L- AP5 (Figure 1B). To test whether this NMDAR-independent LTP could be induced by other stimulation protocols, we utilized theta burst stimulation (TBS), which is considered to be more physiologically relevant. As shown in Figure 1C,D, this protocol generated a significant amount of LTP during both extracellular ( %; P,0.001) and whole cell recordings ( %; P,0.001) in knockout slices in the presence of the NMDAR antagonist, in stark contrast to wild-type animals (Figure S1; control = %; D,L- AP5 = %; P,0.001). To determine the persistence of LTP induced through CP-AMPARs, we delivered multiple trains of HFS (4 trains of 100 Hz at 20 second intervals), which are commonly used induce a long-lasting (or a late phase) LTP (L-LTP). Utilizing this protocol, a CP-AMPAR dependent L-LTP (Figure 2B) was prominent in D,L-AP5-perfused GluR22/2slices (228620%; P,0.001). One potential problem with the above experiment was that all the AMPARs in GluR22/2mice lack the GluR2 subunit, which may rarely occur under normal physiological or pathological conditions. GluR22/2mice may also suffer developmental compensations that could lead to changes in neuronal signaling processes. Therefore, we utilized the GluR2+/2(heterozygous) mice, where the level of total GluR2 protein is reduced and both GluR2-containing and GluR2-lacking AMPARs are expressed at CA1 synapses. In addition, GluR2+/2mice are completely indistinguishable from the wild-type animals in growth and behavioral responses as opposed to GluR22/2mice, which have multiple deficits [10]. As shown in Figure 2B, long-lasting L-LTP was also clearly generated in GluR2+/ 2mice in the presence of 100 mm D,L-AP5 ( %; P,0.001). This CP-AMPAR dependent L-LTP shared the characteristic dependence of longer-lasting forms of plasticity on the formation of new proteins [31], where plasticity induced in both wild-type (vehicle treated = %; anisomycin = %; P = 0.006) and GluR2+/2slices (D,L-AP5+vehicle = %; D,L-AP5+anisomycin = %; P = 0.002) was significantly reduced (Figure 2C,D) under the administration of the protein synthesis inhibitor anisomycin (25 mm). These results indicate that CP- AMPARs can induce various types of long-lasting synaptic plasticity at CA1 synapses, including a previously undemonstrated form of protein synthesis-dependent L-LTP that is NMDAR-independent. Induction of CP-AMPAR-dependent plasticity exclusively requires CP-AMPARs To exclude the possibility that other receptor subtypes (such as high voltage activated calcium channels) may play a role in the induction of CP-AMPAR-dependent plasticity, we decided to test whether this form of potentiation was susceptible to the selective CP-AMPAR inhibitor IEM-1460 [21,32]. Administration of 100 mm IEM-1460 significantly reduced basal transmission in GluR22/2slices (pre-treatment = %; treated = %; P,0.001) and completely blocked the subsequent induction of CP-AMPAR-dependent LTP by 2 trains of 100 Hz (Figure 3A) in the presence of D,L-AP5 (D,L-AP5+IEM-1460 = %). Accordingly, administration of 100 mm IEM-1460 in GluR2+/ 2slices completely inhibited CP-AMPAR-dependent L-LTP induced by 4 trains of 100 Hz (Figure 3B) in the presence of D,L-AP5 (D,L-AP5 = %; D,L-AP5+IEM-1460 = %; P,0.001). These results confirm that CP-AMPAR-dependent plasticity is induced exclusively through CP-AMPARs in our GluR2 knockout mouse model. Requirement of calcium ions in CP-AMPAR-dependent LTP To investigate the mechanisms underlying this CP-AMPARdependent form of potentiation, we compared paired pulse facilitation (PPF) before and during 2 trains of 100 Hz LTP (GluR22/2with D,L-AP5; P = 0.91) and 4 trains of 100 Hz L- LTP (GluR2+/2with D,L-AP5; P = 0.97), but found no significant differences (Figure 4A,B), suggesting that presynaptic involvement was not altered following the induction of this plasticity. Therefore, we concentrated our analyses on postsynaptic mechanisms. To test whether postsynaptic calcium ions are important, we performed whole-cell patch-clamp recordings with or without the high affinity Ca 2+ chelator BAPTA (30 mm) in the intracellular solution. As shown in Figure 4D, TBS induced a persistent increase in the amplitude of excitatory postsynaptic currents (EPSCs) that could last during the entire recording period. However, in the presence of BAPTA, the CP-AMPAR dependent LTP was completely blocked (D,L-AP5 = %; D,L- AP5+BAPTA = %; P = 0.001). These results indicate that Ca 2+ ions in the postsynaptic neurons are crucial triggers for CP-AMPAR-dependent LTP, similar to their role in traditional NMDAR-dependent forms of plasticity (Figure 4C; control = %; BAPTA = %; P,0.001). Independence of Ca/CaM-dependent kinase II (CaMKII) In NMDAR-dependent LTP at CA1 synapses, Ca 2+ influx from NMDARs activate CaMKII to trigger a number of downstream events, including AMPAR trafficking to the synapse that ultimately result in an increase in synaptic transmission [1,33 35]. Therefore, CaMKII is the key Ca 2+ -activated protein kinase indispensable for the induction of NMDAR-dependent LTP. To test whether CaMKII also plays a role in CP-AMPAR-dependent LTP induced by 2 trains of 100 Hz, we first utilized the broad spectrum CaMKII inhibitor staurosporine [36,37]. Administration PLoS ONE 2 February 2009 Volume 4 Issue 2 e4339

3 CP-AMPAR Dependent LTP Figure 1. GluR2-lacking mice are capable of robust long-lasting LTP in the presence of the NMDA antagonist D,L-AP5. (A, B) D,L-AP5 completely inhibited NMDAR-dependent LTP induced by 2 trains of 100 Hz (as indicated by arrow) in (A) GluR2+/+slices (vehicle, n = 5; D,L-AP5, n = 5; P,0.001) but not CP-AMPAR-dependent LTP in (B) GluR22/2slices (D,L-AP5, n = 6; P,0.001). (C, D) Robust LTP induced in GluR22/2slices by TBS (as indicated by arrow) in the presence of D,L-AP5 in (C) field EPSP recordings (D,L-AP5, n = 6; P,0.001) and (D) whole-cell recordings (D,L-AP5, n = 6; P,0.001). All field EPSP recordings of CP-AMPAR-dependent LTP in GluR2 mutants involved the addition of 100 mm D,L-AP5 to perfusate 15 minutes prior to induction, lasting until 5 minutes post-induction, while all whole-cell studies of CP-AMPAR dependent LTP involved 100 mm D,L-AP5 being perfused during the entire recording period. Error bars represent SEM. doi: /journal.pone g001 of 100 nm staurosporine drastically reduced NMDAR-dependent LTP (Figure 5A) in the wild-type animals (vehicle treated = %; staurosporine = %; P = 0.002), but surprisingly had no effect on the amount of CP-AMPAR dependent LTP (Figure 5B) induced in the presence of D,L-AP5 (D,L- AP5+vehicle = %; D,L-AP5+staurosporine = 14863; P = 0.97). To confirm these findings, we performed further investigations by including the CaMKII specific inhibitor KN-62 in the perfusion solution. Accordingly, we found that KN-62 (15 mm) also had no effect on CP-AMPAR-induced LTP in GluR22/2mice. As shown in Figure 5D, the magnitude of LTP was indistinguishable with or without KN-62 (D,L-AP5+vehicle = %; D,L-AP5+KN-62 = %; P = 0.42). The lack of KN-62 effect on CP-AMPAR-dependent LTP was not due to the ineffectiveness of the drug because it could effectively block NMDAR-dependent LTP (Figure 5C) in wild-type animals (vehicle treated = %; KN-62 = %; P,0.001) and also significantly inhibited the enhanced LTP in GluR22/ 2mice (Figure 5E) in the absence of D,L-AP5 (vehicle treated = %; KN-62 = %; P = 0.02). These results indicate that CaMKII is not required for CP-AMPAR-dependent LTP, and suggest that a distinct synaptic signaling cascade must be operating during this form of LTP. Requirement for mitogen-activated kinase (MAPK) cascade and phosphoinositide 3-kinase (PI3-kinase) Recent studies suggest that both MAPK and PI3-kinases are involved in NMDAR-dependent LTP and AMPAR trafficking in PLoS ONE 3 February 2009 Volume 4 Issue 2 e4339

4 CP-AMPAR Dependent LTP Figure 2. CP-AMPARs are capable of inducing long-lasting and protein-synthesis dependent forms of L-LTP. (A, B) D,L-AP5 completely blocked NMDAR-dependent L-LTP induced by 4 trains of 100 Hz (as indicated by arrow) in (A) GluR2+/+slices (vehicle, n = 5; D,L-AP5, n = 5; P,0.001) but not CP-AMPAR-dependent L-LTP in (B) GluR22/2(D,L-AP5, n = 5; P = 0.002) and GluR2+/2slices (D,L-AP5, n = 5; P,0.001). (C, D) L-LTP induced by 4 trains of 100 Hz (as indicated by arrow) is dependent on protein synthesis in both (C) GluR2+/+slices (vehicle, n = 5; anisomycin n = 5; P = 0.006) and (D) GluR2+/2slices (vehicle, n = 5; D,L-AP5+anisomycin, n = 5; P = 0.002). CP-AMPAR-dependent L-LTP recordings in GluR2 mutants involved the addition of 100 mm D,L-AP5 to perfusate 15 minutes prior to induction, lasting until 5 minutes post-induction. 25 mm anisomysin was added to perfusate minutes prior to L-LTP induction and washed away 5 minutes post-induction. Error bars represent SEM. doi: /journal.pone g002 the hippocampus [38 44]. Therefore, we tested whether they are also important in CP-AMPAR-dependent LTP induced by 2 trains of 100 Hz by comparing the effects of MEK and PI3-kinase inhibitors. Utilization of the PI3-kinase inhibitor LY (20 mm) completely blocked both NMDAR-dependent LTP (Figure 6A) in the wild-type animals (vehicle treated = 16367%; LY = %; P,0.001) and CP-AMPAR-dependent LTP (Figure 6B) in GluR22/2mice (D,L-AP5+vehicle = ; D,L-AP5+ LY = ; P,0.001). Accordingly, administration of wortmannin (1 mm), another PI3-kinase inhibitor that is structurally unrelated to LY294002, also attenuated CP-AMPAR dependent LTP in knockout slices (D,L-AP5+vehicle = ; D,L-AP5+wortmannin = ; P,0.001; Figure 6B). In a similar manner, the use of MEK (MAPKK or ERK kinase) inhibitor PD98059 (50 mm) strongly suppressed LTP (Figure 7A,B) in both wild-type (vehicle treated = %; PD98059 = %; P,0.001) and GluR22/2mice (D,L-AP5+vehicle = %; D,L- AP5+PD98059 = %; P,0.001). The involvement of the MAPK cascade in CP-AMPAR dependent plasticity was further demonstrated by the suppression of LTP in knockout slices (D,L- AP5+vehicle = %; D,L-AP5+U0126 = 12566%; P = 0.02; Figure 7B) by the PD98059-structurally unrelated MEK inhibitor U0126 (35 mm). These results indicate that both PI3-kinase and MAPK signaling pathways are required for CP-AMPAR-dependent LTP. Role of MAPK in the induction of plasticity To further elucidate the role of the MAPK signaling cascade in CP-AMPAR dependent plasticity, we tested whether the MEK inhibitor PD98059 (50 mm) would attenuate potentiation when perfused during the maintenance phase of CP-AMPAR dependent L-LTP induced by 4 trains of 100 Hz. We first demonstrated that the administration of this inhibitor (Figure 8A) prior to and during the induction of L-LTP in wild-type slices largely inhibited potentiation (vehicle treated = %; PD98059 = %; P = 0.002), while having no significant effect when introduced during the maintenance phase (Figure 8C) of this NMDAR-dependent plasticity (vehicle treated = %; PD98059 = %; P = 0.9). In a similar manner, the presence of PD98059 in GluR2+/2slices significantly blocked CP-AMPAR dependent L-LTP in the inductory (D,L-AP5+vehicle in induction phase = %; D,L-AP5+PD98059 in induction phase = %; P = 0.009) but not the maintenance phase (D,L-AP5+vehicle in maintenance phase = %; D,L- AP5+PD98059 in maintenance phase = %; P = 0.83) of this form of potentiation (Figure 8B,D). These results suggest that PLoS ONE 4 February 2009 Volume 4 Issue 2 e4339

5 CP-AMPAR Dependent LTP Figure 3. Induction of CP-AMPAR-dependent plasticity is blocked by the CP-AMPAR inhibitor IEM (A, B) Administration of IEM in (A) GluR22/2slices significantly reduced basal synaptic response as well as completely attenuated CP-AMPAR-dependent LTP (D,L-AP5+IEM- 1460, n = 6; P = 0.44) induced by 2 trains of 100 Hz (as indicated by arrow). (B) In a similar fashion, CP-AMPAR-dependent L-LTP induced by 4 trains of 100 Hz (as indicated by arrow) in GluR2+/2slices was also completely blocked by IEM-1460 (D,L-AP5, n = 5; D,L-AP5+IEM-1460, n = 5; P,0.001). All CP- AMPAR-dependent LTP field EPSP studies in GluR2 mutants involved adding 100 mm D,L-AP5 to ACSF perfusate 15 minutes prior to induction until 5 minutes post-induction. 100 mm IEM-1460 was added to the ACSF perfusate 25 minutes prior to LTP induction in GluR22/2slices and was present throughout the entire recording period, while 100 mm IEM-1460 was added to the ACSF perfusate minutes prior to L-LTP induction in GluR2+/ 2slices up until 5 minutes post-induction. Error bars represent SEM. doi: /journal.pone g003 PLoS ONE 5 February 2009 Volume 4 Issue 2 e4339

6 CP-AMPAR Dependent LTP Figure 4. Plasticity induced through CP-AMPARs is completely dependent on Ca 2+ influx. (A, B) Paired pulse facilitation (PPF) revealed no significant difference in presynaptic involvement in CP-AMPAR-dependent plasticity induced by 2 trains of 100 Hz (as indicated by arrow) in (A) GluR22/2slices (D,L-AP5, n = 3; P = 0.91) and by 4 trains of 100 Hz (as indicated by arrow) in (B) GluR2+/2slices (D,L-AP5, n = 5; P = 0.97) in a manner similar to NMDAR-dependent potentiation induced in wild-type controls. (C, D) Potentiation induced by TBS (as indicated by arrow) in whole cell recordings (as indicated by arrow) is completely dependent on Ca 2+ influx in both NMDAR-dependent LTP in (C) GluR2+/+slices (untreated, n = 5; BAPTA, n = 5; P,0.001) and CP-AMPAR-dependent LTP in (D) GluR22/2slices (D,L-AP5, n = 6 ; D,L-AP5+BAPTA, n = 8; P = 0.001). CP-AMPARdependent LTP field EPSP studies in GluR2 mutants involved adding 100 mm D,L-AP5 to ACSF perfusate 15 minutes prior to induction until 5 minutes post-induction. Whole-cell recordings of CP-AMPAR dependent LTP involved the presence of 100 mm D,L-AP5 throughout the recording period. 30 mm BAPTA was included in the intracellular solution for Ca 2+ studies. Error bars represent SEM. doi: /journal.pone g004 the MAPK signaling cascade is an essential component in the induction but not maintenance of both NMDAR- and CP- AMPAR dependent forms of plasticity. Partial requirement of receptor trafficking Ample results have indicated that AMPAR trafficking at the synapse is critical for synaptic changes [1,34,35]. Thus, expression of NMDAR-dependent LTP is dependent on increased AMPAR insertion through regulated exocytosis. To determine whether this process plays a role in CP-AMPAR-dependent LTP, we examined the effect of the exocytosis-inhibiting tetanus toxin light chain fragment (TeTx) on plasticity by including the toxin in the intracellular recording solution. Consistent with previous reports [45], TeTx (75 nm) completely blocked NMDAR-dependent LTP (Inactive TeTx = %; active TeTx, 75 nm = %; P = 0.003) induced by TBS (Figure 9A) in the wild-type animals. In GluR22/2slices, the magnitude of CP-AMPAR dependent TBS- LTP in the presence of D,L-AP5 was also significantly reduced (D,L- AP5+inactive TeTx = %; D,L-AP5+active TeTx, 75 nm = %; P,0.001) by 75 nm TeTx (Figure 9B). However, unlike NMDAR-dependent LTP, plasticity induced through CP-AMPARs was not completely abolished in the presence of the toxin, including when the concentration of TeTx was increased to 250 nm (D,L- AP5+active TeTx, 75 nm = %; D,L-AP5+active TeTx, 250 nm = %; P = 0.62; Figure 9B,C). These results suggest that exocytosis plays at least a partial role in plasticity induced through CP-AMPARs. Discussion In this study, by using GluR22/2and GluR2+/2mice, we were able to induce and investigate the underlying mechanisms of synaptic plasticity caused by CP-AMPARs in the CA1 region of the hippocampus. We demonstrated that these receptors can induce a PLoS ONE 6 February 2009 Volume 4 Issue 2 e4339

7 CP-AMPAR Dependent LTP Figure 5. CaMKII is not involved in long-term potentiation induced through CP-AMPARs. (A, B) Contrasting effects of the broad spectrum inhibitor staurosporine where NMDAR-dependent LTP induced by 2 trains of 100 Hz (as indicated by arrow) was significantly attenuated in (A) GluR2+/+slices (vehicle, n = 6; staurosporine, n = 5; P = 0.002) but not CP-AMPAR-dependent LTP in (B) GluR22/2slices (vehicle, n = 5; D,L- AP5+staurosporine, n = 5; P = 0.97). (C, D) Contrasting effects of the CaMKII-specific inhibitor KN-62 where LTP induced by 2 trains of 100 Hz (as indicated by arrow) was completely blocked in (C) GluR2+/+slices (vehicle, n = 5; KN-62, n = 5; P,0.001) but not in (D) GluR22/2slices (vehicle, n = 5; D,L-AP5+KN-62, n = 5; P = 0.42). (E) However, KN-62 significantly inhibited LTP induced in GluR22/2slices (vehicle = 6; KN-62 = 5; P = 0.02) in the absence of D,L-AP5. (F) Summary graph of the means of the last 10 minutes of potentiation seen in KN-62 treatment studies. All CP-AMPARdependent LTP recordings in GluR2 mutants involved the administration of 100 mm D,L-AP5 to perfusate 15 minutes prior to induction, lasting until 5 minutes post-induction. 100 nm staurosporine and 15 mm KN-62 were added to perfusate minutes prior to LTP induction and washed away 5 minutes post-induction. Error bars represent SEM. denotes P,0.05. doi: /journal.pone g005 long-lasting enhancement in synaptic strength that is Ca 2+ dependent, but surprisingly independent of CaMKII, a crucial regulator of NMDAR-dependent LTP. Instead, this form of LTP requires both the PI3-kinase and MAPK signaling cascades, as well as exocytosis. These results indicate a unique synaptic signaling process during CP-AMPAR-induced synaptic plasticity. PLoS ONE 7 February 2009 Volume 4 Issue 2 e4339

8 CP-AMPAR Dependent LTP PLoS ONE 8 February 2009 Volume 4 Issue 2 e4339

9 CP-AMPAR Dependent LTP Figure 6. PI3-kinase is required for CP-AMPAR-dependent long-term potentiation. (A) The specific PI3-kinase inhibitor LY significantly attenuated NMDAR-dependent LTP induced by 2 trains of 100 Hz (as indicated by arrow) in GluR2+/+slices (vehicle, n = 6; LY294002, n=5; P,0.001). CP-AMPAR-dependent LTP elicited by 2 trains of 100 Hz (as indicated by arrow) in (B) GluR22/2slices was also strongly suppressed in the presence of the structurally unrelated PI3K inhibitors LY (vehicle, n = 5; D,L-AP5+LY294002, n = 5; P,0.001) and wortmannin (vehicle, n = 5; D,L-AP5+wortmannin, n = 5; P,0.001). (C) Summary graph of the means of the last 10 minutes of potentiation seen in LY and wortmannin treatment studies. CP-AMPAR-dependent LTP recordings in GluR2 mutants involved the addition of 100 mm D,L-AP5 to perfusate 15 minutes prior to induction, ceasing at 5 minutes post-induction. 20 mm LY and 1 mm wortmannin were added to ACSF perfusate minutes prior to LTP induction up until 5 minutes post-induction. Error bars represent SEM. denotes P,0.05. doi: /journal.pone g006 We also demonstrated that a number of protocols were able to induce CP-AMPAR-dependent LTP. These include TBS (Figure 1, 4, 9) that is often used to induce NMDAR-dependent LTP at resting membrane potentials, suggesting that CP-AMPARdependent synaptic plasticity may occur under physiological conditions. Additionally, we also presented the first ever demonstration of the capability of CP-AMPARs to induce and sustain a longer-lasting protein synthesis dependent form of L-LTP in the CA1 region of the hippocampus, previously thought to be a hallmark of NMDARs alone. CP-AMPAR-dependent LTP was completely blocked by postsynaptic inclusion of 30 mm BAPTA, indicating that calcium ions in the postsynaptic neurons are required (Figure 4). Consistent with this, we and others have previously shown that GluR2-lacking AMPARs exhibit significant Ca 2+ permeability [7,10] that is similar to that of NMDARs; therefore, the Ca 2+ influx from these receptors is likely sufficient to trigger Ca 2+ -dependent events in the postsynaptic neurons. Ca 2+ transients resulting from synaptically activated CP-AMPARs have also been demonstrated in the cortex and shown to amplify the transient amplitude when co-expressed with NMDARs, suggesting the presence of mechanisms that allow the proportional scaling of each receptor type independent of presynaptic miniature-release factors [46]. Since CP-AMPAR-dependent LTP was also inducible in the presence of high voltage activated calcium channel blockers [10], the most likely calcium source in this form of plasticity is the Ca 2+ influx from CP-AMPARs. This notion was confirmed by our results showing that CP-AMPAR-dependent plasticity induced by either 2 trains or 4 trains of 100 Hz was completely blocked by the selective CP-AMPAR inhibitor IEM-1460 in GluR22/2and GluR2+/2mutants respectively (Figure 3). In contrast to GluR22/2animals, the same concentration of IEM-1460 (100 mm) had a small but non-significant effect on the basal synaptic response in GluR2+/2slices, suggesting that CP- AMPARs may not play an important role in basal transmission at CA1 synapses, and reinforces the idea that Ca 2+ influx from newly recruited forms of these receptors may help consolidate previously induced plasticity in the region [22 24, but see 21, 25]. Given the critical role for Ca 2+ in CP-AMPAR-dependent LTP, it is therefore surprising to discover that CaMKII is not required for this form of LTP as the CaMKII-specific inhibitor KN-62 had no effect (Figure 5). In addition, the general kinase inhibitor staurosporine, which is also known to block the actions of PKC, PKA, and protein tyrosine kinases in addition to CaMKII [36,37], also had no inhibitory effects on CP-AMPAR-dependent LTP. The lack of effect of these two inhibitors is not due to the ineffectiveness of the drugs because they strongly inhibited NMDAR-dependent LTP in slices prepared from wild-type littermate animals (Figure 5A,C). Additionally, the enhanced LTP (comprising of both NMDAR- and CP-AMPAR-dependent components in the absence of D,L-AP5, Figure 5E) seen in GluR22/2mice [10] was also significantly, but not completely, attenuated in the presence of KN-62. This would suggest that the CaMKII inhibitor worked only towards blocking the NMDARdependent component of the enhanced LTP induced in GluR22/ 2mice when D,L-AP5 was not present, while not affecting the CP- AMPAR-dependent component, corresponding with our pharmacological inhibitor results in wild-type and mutant animals respectively. It is also important to note that the expression and targeting of CaMKII is not altered in GluR22/2or GluR2+/ 2mice (data not shown). These results indicate that Ca 2+ influx from CP-AMPARs initiates a distinct synaptic signaling process that is different from the activation of NMDARs. Interestingly, CaM and CaMKII are associated with the NMDAR but not AMPAR complex [47 49]. In contrast to CaMKII, our results provide evidence for the involvement of the PI3-kinase (Figure 6) and MAPK signaling cascades (Figure 7) in CP-AMPAR dependent plasticity. This was accomplished through the utilization of two separate pairs of structurally unrelated inhibitors geared towards each of these kinase systems (LY and wortmannin for PI3K; PD98059 and U0126 for MAPK respectively). The varying degrees of attenuation of CP-AMPAR-dependent LTP by the above-mentioned drugs was probably due to differences in concentration and the modalities of action of these individual inhibitors. It is interesting to note that CP- AMPAR dependent L-LTP was not completely inhibited by the same concentration of PD98059 (Figure 8B) that abolished CP- AMPAR dependent LTP in knockout slices (Figure 7B). This could be explained by the possibility that the stronger stimulation protocols utilized for L-LTP need a longer period of time for plasticity to completely decay, requiring an extended period of post-inductory recording. Additionally, mechanistic differences between LTP and L-LTP have been previously described [31], suggesting that the induction of the latter form of plasticity may recruit additional biochemical pathways that are distinct and insensitive to MAPK inhibitors. These notions are supported by the fact that L-LTP induced in wild-type slices was also not completely blocked by the same concentration of PD98059 in a similar time period (Figure 8A), in comparison to the total abolishment of LTP induced by 2 trains of 100 Hz in the same animals (Figure 7A). The involvement of the PI3-kinase and MAPK cascades in this form of plasticity is of particular interest as both have also been shown to be important for NMDAR-dependent LTP [39,40]. We therefore hypothesize that these two signaling pathways may serve as a common target for both NMDAR- and CP-AMPARdependent LTP. Since CP-AMPAR-dependent LTP is independent of CaMKII, our model suggests that CP-AMPARs may act downstream of NMDARs and CaMKII, consistent with the idea that CP-AMPAR-activated signaling process may serve as a mechanism for the consolidation of NMDAR-dependent LTP [22]. To support this notion, a recent study implicated another member of the CaMK family (CaMKI) in the specific synaptic recruitment of CP-AMPARs during TBS-LTP in the CA1 region [24]. Furthermore, the shared dependence of both NMDAR- and CP-AMPAR-induced types of L-LTP on new protein synthesis may suggest that Ca 2+ influx through newly recruited GluR2- lacking receptors could also be an important facilitator of traditional longer-lasting forms of plasticity in the hippocampus. Our results are consistent with the hypothesis that the expression mechanisms of CP-AMPAR-dependent LTP are PLoS ONE 9 February 2009 Volume 4 Issue 2 e4339

10 CP-AMPAR Dependent LTP PLoS ONE 10 February 2009 Volume 4 Issue 2 e4339

11 CP-AMPAR Dependent LTP Figure 7. The MAPK signaling cascade plays an essential role in long-term potentiation induced through CP-AMPARs. (A) The specific MEK inhibitor PD98059 attenuated LTP elicited by 2 trains of 100 Hz (as indicated by arrow) in GluR2+/+slices (vehicle, n = 5; PD98059, n = 6; P,0.001). CP-AMPAR-dependent LTP induced by 2 trains of 100 Hz (as indicated by arrow) was also significantly inhibited in (B) GluR22/2slices by the structurally unrelated MEK inhibitors PD98059 (vehicle, n = 6; D,L-AP5+PD98059, n = 5; P,0.001) and U0126 (vehicle, n = 6; D,L-AP5+U0126, n = 5; P = 0.02). (C) Summary graph of the means of the last 10 minutes of potentiation seen in PD98059 and U0126 treatment studies. CP-AMPARdependent LTP recordings in GluR2 mutants involved the addition of 100 mm D,L-AP5 to perfusate 15 minutes prior to induction, ceasing at 5 minutes post-induction. 50 mm PD98059 and 35 mm U0126 were added to ACSF perfusate minutes prior to LTP induction lasting until 5 minutes post-induction. Error bars represent SEM. denotes P,0.05. doi: /journal.pone g007 Figure 8. The MAPK signaling cascade plays a role in the induction but not maintenance of plasticity induced through CP-AMPARs. (A, B) Administration of the MEK inhibitor PD98059 during the induction phase of L-LTP significantly attenuated potentiation induced by 4 trains of 100 Hz (as indicated by arrow) during both NMDAR-dependent L-LTP in (A) GluR2+/+slices (vehicle, n = 5; PD98059, n = 6; P = 0.002) and CP-AMPARdependent L-LTP in (B) GluR2+/2slices (vehicle, n = 5; D,L-AP5+PD98059, n = 5; P = 0.009). (C, D) However, administration of PD98059 during the maintenance phase of L-LTP induced by 4 trains of 100 Hz (as indicated by arrow) had no significant effect on both (C) GluR2+/+slices (vehicle, n = 5; PD98059, n = 5; P = 0.9) and (D) GluR2+/2slices (vehicle, n = 5; D,L-AP5+PD98059, n = 5; P = 0.83). All CP-AMPAR-dependent L-LTP recordings in GluR2 mutants involved adding 100 mm D,L-AP5 to perfusate 15 minutes prior to induction until 5 minutes post-induction. For MAPK L-LTP induction studies, 50 mm PD98059 was added to ACSF perfusate minutes prior to induction lasting until 5 minutes post-induction. For MAPK L-LTP maintenance studies, 50 mm PD98059 was added to the ACSF perfusate minutes post-induction. Error bars represent SEM. doi: /journal.pone g008 postsynaptic. This is based on the findings that PPF does not change after LTP induction (Figure 4) and that postsynaptic injection of exocytosis-inhibiting tetanus toxin (75 nm) largely blocks this form of LTP (Figure 9). However, since the inhibiting effect of tetanus toxin is not complete even at higher concentration levels (250 nm; Figure 9B,C), other mechanisms independent of receptor trafficking may also contribute to CP-AMPAR-dependent LTP. These may include changes in receptor protein phosphorylation and subunit composition. It is known that multiple postsynaptic mechanisms exist for NMDAR-dependent LTP at CA1 synapses [35]. The requirement of AMPAR insertion for CP-AMPAR-dependent LTP is also consistent with the above model that CP-AMPAR-induced LTP is downstream of NMDAR-dependent LTP. This mechanism has been postulated to consolidate synaptic enhancement as newly inserted GluR2- lacking CP-AMPARs are gradually replaced by GluR2-containing receptors [22 24], a process that likely depends on exocytosis. In summary, we have identified a unique signaling pathway underlying long-lasting synaptic enhancement triggered by Ca 2+ influx through CP-AMPARs at CA1 synapses. We suggest that this synaptic signaling process may provide an important and general mechanism for synaptic plasticity at synapses where the expression of GluR2 is dynamically regulated under various circumstances, including development, synaptic plasticity and pathological conditions. PLoS ONE 11 February 2009 Volume 4 Issue 2 e4339

12 CP-AMPAR Dependent LTP PLoS ONE 12 February 2009 Volume 4 Issue 2 e4339

13 CP-AMPAR Dependent LTP Figure 9. Receptor trafficking plays an important role in plasticity induced through CP-AMPARs. (A) Synaptic plasticity induced by TBS (as indicated by arrow) was significantly attenuated in the presence of the exocytosis-inhibiting tetanus toxin (TeTx, 75 nm) during NMDARdependent LTP in GluR2+/+slices (inactive toxin, n = 6; 75 nm TeTx, n = 6; P = 0.003). (B) CP-AMPAR-dependent LTP elicited by TBS (as indicated by arrow) in GluR22/2slices was also significantly reduced in the presence of 75 nm TeTx (D,L-AP5+inactive toxin, n = 7; D,L-AP5+75 nm TeTx, n = 5; P,0.001) and 250 nm TeTx (D,L-AP5+inactive toxin, n = 7; D,L-AP5+250 nm TeTx, n = 5; P,0.001) respectively to statistically similar levels (D,L- AP5+75 nm TeTx, n = 5; D,L-AP5+250 nm TeTx, n = 5; P = 0.62). (C) Summary graph of the means of the last 5 minutes of potentiation seen in tetanus toxin treatment studies. Whole-cell recordings of CP-AMPAR dependent LTP involved the presence of 100 mm D,L-AP5 throughout the recording period. 75 nm and 250 nm TeTx were included in the intracellular solution for exocytosis inhibition studies. Error bars represent SEM. denotes P,0.05. doi: /journal.pone g009 Materials and Methods GluR2 knockout mice GluR2 knockout mice were created and bred as previously described [10,30]. For genotyping of GluR2 knockout mice, a standard protocol for Taqman PCR was used. A common reverse primer 59-TCGCCCATTTTCCCATATAC-39 and forward primers 59-GGTTGGTCACTCACCTGCTT-39 and 59-TCGCCC- ATTTTCCCATATAC-39 were used to detect wild-type allele and the neomycin resistance cassette in knockout mice respectively. All studies with mutant animals (GluR22/2or GluR2+/2) were performed alongside wild-type littermates (GluR2+/+) for controls. Experimental protocols were approved by The Hospital for Sick Children Animal Care Committee. Extracellular fepsp electrophysiological recordings The preparation of brain slices has been previously described [10,30]. Briefly, hippocampal slices (400 mm) were obtained from 3 to 6 month old adult mice and allowed to recover in a submerged holding chamber for at least 1 hour. A single slice was then transferred to the recording chamber and submerged and superfused with 95% O 2-5% CO 2 saturated artificial cerebral spinal fluid (ACSF, 2 ml/min) at a temperature of 28uC. The ACSF contained (in mm) 120 NaCl, 2.5 KCl, 1.3 MgSO 4, 1.0 NaH 2 PO 4, 26 NaHCO 3, 2.5 CaCl 2, and 11 D-glucose. For field EPSPs, the recording pipette (3 MV) was filled with ACSF solution. Synaptic responses were evoked by bipolar tungsten electrodes placed mm from the cell body layer in the CA1 area. fepsps were measured by taking the slope of the rising phase between 5% and 60% of the peak response. All data acquisition and analysis were done using pclamp 7 software (Axon instruments). After a stable baseline period, LTP was induced by high-frequency stimulation (HFS) using 2 trains of 100 Hz at 10 second intervals (with each train lasting 1 s) or a TBS protocol (total 60 pulses) while late-phase LTP (L-LTP) was induced by 4 trains of 100 Hz at 20 second intervals (with each train lasting 1 s). Paired-pulse facilitation (PPF) was recorded prior to and following induction of LTP (with the first and second responses being separated by an interval of 50 ms). The ratio of the slope of the second response to the slope of the first response was subsequently calculated. All drugs for field EPSP recordings were purchased from Sigma Aldrich (Oakville, Canada), Tocris (Missouri, U.S.A.) and LC Laboratories (Massachusetts, U.S.A.). D,L-AP5 and IEM-1460 were dissolved in distilled water. Staurosporine, KN-62, PD98059, U0126, LY294002, wortmannin and anisomycin were dissolved in DMSO. For tests of inhibition of LTP, drugs were added to ACSF perfusate minutes before the LTP induction protocol was initiated (with staurosporine, KN-62, PD98059, U0126, LY294002, wortmannin and anisomycin having a maximum final concentration of 0.1% DMSO or less after being added to ACSF). For experiments concerning the maintenance phase of L-LTP, PD98059 was added to the ACSF perfusate minutes post-induction. Vehicle treatments were performed with 0.1% DMSO and/or distilled water depending on the drug/drugs perfused. All inhibition studies with mutant animals (GluR22/2 or GluR2+/2) were performed alongside wild-type littermates (GluR2+/+) for controls. For comparison of the magnitude of LTP between different groups, the last 5 10 minutes of recordings were compared statistically. n represents the number of hippocampal slices used in each experiment. Normally, one slice per mouse was used for experiments. When average data was plotted, data was normalized to the average of the baseline responses unless indicated otherwise. The representative traces were averages of four successive sweeps during recording. All data was statistically evaluated by Student s t-test. Experimental protocols were approved by The Hospital for Sick Children Animal Care Committee. Whole-cell voltage-clamp recordings Acute hippocampal slices of 300 mm thickness were prepared from GluR2 knockout mice and littermates at days of age. Slicing and recovery procedures are the same as in fepsp recordings. Electrodes (3 4 M V) contained (in mm) 120 Csmethanesulfonate, 5 NaCl, 1 MgCl 2, 0.5 EGTA, 2 Mg-ATP, 0.1 Na 3 GTP, 20 HEPES, ph 7.2, mosm. Picrotoxin (100 mm) was used in ASCF to eliminate GABAergic transmissions. D,L-AP5 (100 mm) was administrated in perfusate throughout all CP-AMPAR dependent LTP experiments. The internal solution for BAPTA (30 mm) experiments was separately made. For tetanus toxin (TeTx) experiments, stock solution was prepared at 1,0006of final concentration (75 nm, 250 nm) and was added to internal solution just prior to recordings. The same amount of tetanus toxin was boiled for 5 minutes prior and used as inactive control. Test pulses were given at 0.1 Hz and neurons were clamped at 265 mv during basal level recording and LTP recording. Whole-cell mode was briefly switched to current-clamp mode to allow depolarization induced by theta burst stimulation (15 bursts with 200 msec interval of 4 pulses at 100 Hz). A 23 mv step was given 250 msec after each test pulse to monitor membrane and access resistances. Data acquisition and analysis were done using pclamp 8. Only recordings with a drift of access and membrane resistances less than 20% were included for statistical analysis by Student s t-test. For comparison of the magnitude of LTP between different groups, the last 5 minutes of recordings were compared statistically. n represents the number of hippocampal slices used in each experiment, with normally only one slice per mouse used for experiments. The representative traces were averages of five successive sweeps during recording. All drugs for whole-cell recordings chemicals were purchased from Sigma Aldrich (Oakville, Canada). Tetanus toxin light chain fragment was a generous gift from Dr. William Trimble. Experimental protocols were approved by The Hospital for Sick Children Animal Care Committee. Supporting Information Figure S1 Figure demonstrating that the administration of D,L- AP5 in the ACSF perfusate blocks TBS-LTP in wild-type slices during whole-cell recordings. Found at: doi: /journal.pone s001 (0.05 MB PDF) PLoS ONE 13 February 2009 Volume 4 Issue 2 e4339

14 CP-AMPAR Dependent LTP Acknowledgments We are thankful to Guijin Lu, Shouping Zhang, Zarko Todorovski and members of the Jia lab for technical assistance. References 1. Malinow R, Malenka RC (2002) AMPA receptor trafficking and synaptic plasticity. Annu Rev Neurosci 25: Song I, Huganir RL (2002) Regulation of AMPA receptors during synaptic plasticity. Trends Neurosci 25: Bredt DS, Nicoll RA (2003) AMPA receptor trafficking at excitatory synapses. Neuron 40: Hestrin S (1993) Different glutamate receptor channels mediate fast excitatory synaptic currents in inhibitory and excitatory cortical neurons. Neuron 11: Jonas P, Racca C, Sakmann B, Seeburg PH, Monyer H (1994) Differences in Ca2+ permeability of AMPA-type glutamate receptor channels in neocortical neurons caused by differential GluR-B subunit expression. Neuron 12: Bowie D, Mayer ML (1995) Inward rectification of both AMPA and kainate subtype glutamate receptors generated by polyamine-mediated ion channel block. Neuron 15: Geiger JR, Melcher T, Koh DS, Sakmann B, Seeburg PH, et al. (1995) Relative abundance of subunit mrnas determines gating and Ca2+ permeability of AMPA receptors in principal neurons and interneurons in rat CNS. Neuron 15: Isaac JTR, Ashby M, McBain CJ (2007) The role of the GluR2 subunit in AMPA receptor function and synaptic plasticity. Neuron 54: Gu JG, Albuquerque C, Lee CJ, MacDermott AB (1996) Synaptic strengthening through activation of Ca2+-permeable AMPA receptors. Nature 381: Jia Z, Agopyan N, Miu P, Xiong Z, Henderson J, et al. (1996) Enhanced LTP in mice deficient in the AMPA receptor GluR2. Neuron 17: Mahanty NK, Sah P (1998) Calcium-permeable AMPA receptors mediate longterm potentiation in interneurons in the amygdala. Nature 394: Rozov A, Burnashev N (1999) Polyamine-dependent facilitation of postsynaptic AMPA receptors counteracts paired-pulse depression. Nature 401: Liu SQ, Cull-Candy SG (2000) Synaptic activity at calcium-permeable AMPA receptors induces a switch in receptor subtype. Nature 405: Kullmann DM, Lamsa KP (2007) Long-term synaptic plasticity in hippocampal interneurons. Nat Rev Neurosci 8: Liu SJ, Zukin RS (2007) Ca2+-permeable AMPA receptors in synaptic plasticity and neuronal death. Trends Neurosci 30: Bellone C, Lüscher C (2005) mglurs induce a long-term depression in the ventral tegmental area that involves a switch of the subunit composition of AMPA receptors. Eur J Neurosci 21: Ge WP, Yang XJ, Zhang Z, Wang HK, Shen W, et al. (2006) Long-term potentiation of neuron-glia synapses mediated by Ca2+-permeable AMPA receptors. Science 312: Lamsa KP, Heeroma JH, Somogyi P, Rusakov DA, Kullmann DM (2007) Anti- Hebbian long-term potentiation in the hippocampal feedback inhibitory circuit. Science 315: Ho MT, Pelkey KA, Topolnik L, Petralia RS, Takamiya K, et al. (2007) Developmental expression of Ca2+-permeable AMPA receptors underlies depolarization-induced long-term depression at mossy fiber CA3 pyramid synapses. J Neurosci 27: Lei S, McBain CJ (2002) Distinct NMDA receptors provide differential modes of transmission at mossy fiber-interneuron synapses. Neuron 33: Gray EE, Fink AE, Sariñana J, Vissel B, O Dell TJ (2007) Long-term potentiation in the hippocampal CA1 region does not require insertion and activation of GluR2-lacking AMPA receptors. J Neurophysiol 98: Plant K, Pelkey KA, Bortolotto ZA, Morita D, Terashima A, et al. (2006) Transient incorporation of native GluR2-lacking AMPA receptors during hippocampal long-term potentiation. Nat Neurosci 9: Lu Y, Allen M, Halt AR, Weisenhaus M, Dallapiazza RF, et al. (2007) Agedependent requirement of AKAP150-anchored PKA and GluR2-lacking AMPA receptors in LTP. EMBO J 26: Guire ES, Oh MC, Soderling TR, Derkach VA (2008) Recruitment of calciumpermeable AMPA receptors during synaptic potentiation is regulated by CaMkinase I. J Neurosci 28: Adesnik H, Nicoll RA (2007) Conservation of glutamate receptor 2-containing AMPA receptors during long-term potentiation. J Neurosci 27: Pellegrini-Giampietro DE, Gorter JA, Bennett MV, Zukin RS (1997) The GluR2 (GluR-B) hypothesis: Ca(2+)-permeable AMPA receptors in neurological disorders. Trends Neurosci 20: Author Contributions Conceived and designed the experiments: SA ZJ. Performed the experiments: SA ZZ. Analyzed the data: SA ZZ. Contributed reagents/ materials/analysis tools: WR ZJ. Wrote the paper: SA ZZ WR ZJ. 27. Bellone C, Lüscher C (2006) Cocaine triggered AMPA receptor redistribution is reversed in vivo by mglur-dependent long-term depression. Nat Neurosci 9: Mameli M, Balland B, Luján R, Lüscher C (2007) Rapid synthesis and synaptic insertion of GluR2 for mglur-ltd in the ventral tegmental area. Science 317: Conrad KL, Tseng KY, Uejima JL, Reimers JM, Heng LJ, et al. (2008) Formation of accumbens GluR2-lacking AMPA receptors mediates incubation of cocaine craving. Nature 454: Meng Y, Zhang Y, Jia Z (2003) Synaptic transmission and plasticity in the absence of AMPA glutamate receptor GluR2 and GluR3. Neuron 39: Lynch MA (2004) Long-term potentiation and memory. Physiol Rev 84: Buldakova SL, Kim KK, Tikhonov DB, Magazanik LG (2007) Selective blockade of Ca2+ permeable AMPA receptors in CA1 area of rat hippocampus. Neuroscience 144: Malenka RC, Nicoll RA (1999) Long-term potentiation a decade of progress? Science 285: Collingridge GL, Isaac JTR, Wang YT (2004) Receptor trafficking and synaptic plasticity. Nat Rev Neurosci 5: Derkach VA, Oh MC, Guire ES, Soderling TR (2007) Regulatory mechanisms of AMPA receptors in synaptic plasticity. Nat Rev Neurosci 8: Rüegg UT, Burgess GM (1989) Staurosporine, K-252 and UCN-01: potent but nonspecific inhibitors of protein kinases. Trends Pharmacol Sci 10: Sanchez-Martinez C, Shih C, Faul MM, Zhu G, Paal M, et al. (2003) Aryl[a]pyrrolo[3,4-c]carbazoles as selective cyclin D1-CDK4 inhibitors. Bioorg Med Chem Lett 13: English JD, Sweatt JD (1996) Activation of p42 mitogen-activated protein kinase in hippocampal long term potentiation. J Biol Chem 271: English JD, Sweatt JD (1997) A requirement for the mitogen-activated protein kinase cascade in hippocampal long term potentiation. J Biol Chem 272: Sanna PP, Cammalleri M, Berton F, Simpson C, Lutjens R, et al. (2002) Phosphatidylinositol 3-kinase is required for the expression but not for the induction or the maintenance of long-term potentiation in the hippocampal CA1 region. J Neurosci 22: Opazo P, Watabe AM, Grant SGN, O Dell TJ (2003) Phosphatidylinositol 3- kinase regulates the induction of long-term potentiation through extracellular signal-related kinase-independent mechanisms. J Neurosci 23: Man HY, Wang Q, Lu WY, Ju W, Ahmadian G, et al. (2003) Activation of PI3- kinase is required for AMPA receptor insertion during LTP of mepscs in cultured hippocampal neurons. Neuron 38: Qin Y, Zhu Y, Baumgart JP, Stornetta RL, Seidenman K, et al. (2005) Statedependent Ras signaling and AMPA receptor trafficking. Genes Dev 19: Passafaro M, Piëch V, Sheng M (2001) Subunit-specific temporal and spatial patterns of AMPA receptor exocytosis in hippocampal neurons. Nat Neurosci 4: Lu W, Man H, Ju W, Trimble WS, MacDonald JF, et al. (2001) Activation of synaptic NMDA receptors induces membrane insertion of new AMPA receptors and LTP in cultured hippocampal neurons. Neuron 29: Wang S, Jia Z, Roder J, Murphy TH (2002) AMPA receptor-mediated miniature synaptic calcium transients in GluR2 null mice. J Neurophysiol 88: Strack S, Colbran RJ (1998) Autophosphorylation-dependent targeting of calcium/ calmodulin-dependent protein kinase II by the NR2B subunit of the N- methyl- D-aspartate receptor. J Biol Chem 273: Gardoni F, Caputi A, Cimino M, Pastorino L, Cattabeni F, et al. (1998) Calcium/calmodulin-dependent protein kinase II is associated with NR2A/B subunits of NMDA receptor in postsynaptic densities. J Neurochem 71: Leonard AS, Lim IA, Hemsworth DE, Horne MC, Hell JW (1999) Calcium/ calmodulin-dependent protein kinase II is associated with the N-methyl-Daspartate receptor. Proc Natl Acad Sci U.S.A 96: PLoS ONE 14 February 2009 Volume 4 Issue 2 e4339

15 Neuropharmacology 56 (2009) Contents lists available at ScienceDirect Neuropharmacology journal homepage: A critical role of Rho-kinase ROCK2 in the regulation of spine and synaptic function Zikai Zhou a,b,1, Yanghong Meng a,1, Suhail Asrar a,b, Zarko Todorovski a,b, Zhengping Jia a,b, a Neurosciences and Mental Health, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8 b Department of Physiology, University of Toronto, Toronto, Ontario, Canada article info abstract Article history: Received 25 April 2008 Received in revised form 30 June 2008 Accepted 16 July 2008 Keywords: Rho GTPase ROCK2 knockout mice Dendritic spine Actin filaments Hippocampal long-term potentiation The actin cytoskeleton is critically involved in the regulation of the dendritic spine and synaptic properties, but the molecular mechanisms underlying actin dynamics in neurons are poorly defined. We took genetic approaches to create and analyze knockout mice specifically lacking ROCK2, a protein kinase that directly interacts with and is activated by the Rho GTPases, the central mediator of actin reorganization. We demonstrated that while these knockout mice were normal in gross brain anatomy, they were impaired in both basal synaptic transmission and hippocampal long-term potentiation (LTP). Consistent with the electrophysiological deficits, the ROCK2 knockout neurons showed deficits in spine properties, synapse density, the actin cytoskeleton, and the actin-binding protein cofilin. These results indicate that ROCK2/cofilin signaling is critical in the regulation of neuronal actin, spine morphology and synaptic function. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Dendritic spines are highly specialized actin-rich protrusions where the majority of excitatory synapses are formed in the mammalian CNS. Changes in spine properties are closely linked to both normal synaptic physiology and mental disorders (Sorra and Harris, 2000; Hering and Sheng, 2001; Van Galen and Ramakers, 2004; Carlisle and Kennedy, 2005). However, the molecular mechanisms by which spines are regulated are poorly understood. Recent studies have indicated that the Rho family small GTPases, including Rho, Rac and Cdc42, play a central role in the regulation of both spine structure and plasticity (Hall, 1998; Luo, 2000; Lamprecht and LeDoux, 2004; Govek et al., 2005; Tada and Sheng, 2006). Thus, manipulations of Rho proteins have been shown to clearly affect basal spine properties (Nakayama et al., 2000; Tashiro et al., 2000; Luo, 2000; Tada and Sheng, 2006), activity-dependent neuronal structural remodeling (Sin et al., 2002; Li et al., 2002; Van Aelst and Cline, 2004) and many forms of synaptic plasticity, including long-term potentiation (LTP) (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999; O Kane et al., 2003, 2004). Consistent with these molecular and cellular studies, genetic deficits in Rho signaling have been found in patients with a number of Corresponding author. Neurosciences and Mental Health, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8. Tel.: þ ; fax: þ address: zhengping.jia@sickkids.ca (Z. Jia). 1 These authors contribute equally to this work. forms of mental retardation (Chelly and Mandel, 2001; Ramakers, 2002) and functional perturbations of Rho signaling impair learning and memory in animal models (Dash et al., 2004; Meng et al., 2005; Diana et al., 2007). The mechanisms by which the Rho GTPases regulate spine and synaptic properties remain unclear; however, it is known that the actin network is involved (Luo, 2002; Govek et al., 2005). This is consistent with the observations that the actin cytoskeleton is highly enriched in the dendritic spine and that it is critical for various aspects of spine properties and synaptic plasticity (Kim and Lisman, 1999; Krucker et al., 2000; Matus, 2000; Smart and Halpain, 2000; Luo, 2002; Lisman, 2003; Dillon and Goda, 2005). Although a wide range of effector proteins have been identified that could mediate the effect of the Rho GTPases on actin regulation (Govek et al., 2005), the protein kinases that are directly associated with and activated by the GTPases are of exceptional importance because they are highly expressed in spiny neurons and are potent neuronal actin regulators (Riento and Ridley, 2003; Zhao and Manser, 2005). In this study, we investigated the role of one of these kinases, Rho-kinase 2 (ROCK2) that is specifically activated by Rho, by generating and analyzing knockout mice deficient in the expression of ROCK2. We demonstrated that while the knockout mice were normal in gross brain anatomy, they were altered in spine morphology, basal synaptic transmission and hippocampal LTP. Accordingly, the knockout mice were altered in the actin cytoskeleton and the actin-binding protein cofilin. These results provide strong genetic evidence that ROCK2 is critical for both spine and synaptic function via cofilin and actin regulation /$ see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi: /j.neuropharm

16 82 Z. Zhou et al. / Neuropharmacology 56 (2009) Methods 2.1. Generation of ROCK2 knockout mice To generate ROCK2 knockout mice, a BAC clone containing exon 5 region was isolated and a targeting vector was constructed by inserting a pgk-neomycin cassette in the Nco1 site of exon 5 upstream of the predicted kinase domain of ROCK2 (Fig. 1A). The vector was used to transfect R1 embryonic stem (ES) cells (derived from the 129 sv mouse strain) and G418 resistant colonies were screened for a targeted event by Southern blot analysis. Two positive ES clones were used to generate chimeric mice by aggregation. The chimeras were crossed with CD1 mouse strain to generate F1 populations for each targeted line. The CD1 strain was chosen for present studies because in this genetic background the effect of ROCK2 deletion on the embryonic development was minimal compared to the C57/B6 strain that was reported previously (Thumkeo et al., 2003). The knockout offspring (ROCK2 / ) and control littermates (ROCK2þ/þ) from F1 (ROCK2þ/ ) interbreedings were used for this study. Analyses of the numbers of various genotypes of F2 population indicated that the number of ROCK2 knockout mice was lower than expected (48 out of 250 recorded F2 progeny), suggesting an increase in embryonic lethality compared to the wild-type littermates. However, the surviving ROCK2 knockout mice showed normal postnatal viability and life span. No home-cage behavioral abnormalities were observed in ROCK2 knockout mice Histology, electron microscopy and hippocampal culture The procedures for histology and electron microscopy (EM) of fixed brain tissues were described previously (Jia et al., 1996; Meng et al., 2002). For EM analysis, asymmetric synapses were defined by the presence of a prominent postsynaptic density and associated presynaptic vesicles. To compare synapse density and spine area, thin-section electron micrographs were randomly collected and analyzed from equivalent CA1 regions. For each genotype, a total of images covering approximately (mm) 2 from two animals were used for quantification. For low-density cultures, hippocampal neurons were prepared from postnatal pups (P0-1) and maintained according to a procedure recommended for Neurobasal-A medium (Invitrogen, San Diego, CA). For F-actin labeling, fixed cells were permeabilized with 0.1% Triton X-100/PBS and stained for 30 min with 1 mg/ml tetramethylrhodamine isothiocyanate (TRITC)-conjugated phalloidin (Sigma, St. Louis, MO). For immunostaining of cultured neurons, fixed cells were permeabilized for 10 min with 0.1% Triton X-100/PBS, blocked for 1 h with 5% BSA, incubated with appropriate primary and then secondary antibodies, followed by TRITC-conjugated phalloidin staining. For immunohistochemical staining of hippocampal sections, the freshly isolated brain was frozen rapidly and sliced to 15 mm thickness sections with a Leica cryostat. The sections were fixed with 4% paraformaldehyde/sucrose, permeabilized with 0.02% Triton, blocked with 5% fetal bovine/donkey serum, incubated with the primary antibodies (Synapsin I, Santa Cruz, USA, 1:100 or VGLUT1, Sigma, MO, USA, 1:300) and then appropriate secondary antibodies. Neuronal images were acquired using a confocal microscope with 20, 40 or 60 objectives under identical gain and contrast settings for the wild-type and knockout samples. For immunohistochemical experiments, the intensity of the fluorescence signals was analyzed using the Volocity software (Improvision) and expressed as a percentage of the green signals (Synapsin I). For cultured neurons, dendritic spines were defined as protrusions on the dendrites (identified as MAP2 positive processes) between 0.5 and 5 mm in length. Spine length and density were measured using images with MAP2/phalliodin double staining. Spine length was defined from the base (the edge of MAP2 signal) to the tip of the spine. Only linear primary and secondary dendrites and associated protrusions were used for analyses. For each genotype, a total of neurons from three independent hippocampal cultures and approximately mm dendritic segments per neuron were used for quantification. Whenever possible, knockout neurons that were plated along with the wild-type littermates on the same 24-well plates were used to maintain similar conditions. Both spine and synapse properties were analyzed with NIH Image software. To quantify the distribution of F-actin, the confocal images stained with phalliodin were converted to grayscale and the intensity of the fluorescence was measured and compared between the spine head and an adjacent dendritic area using a Northern Eclipse program (Meng et al., 2002). All images were collected by one person and analyzed A R K N H S N N R Wild-type Rock2 Locus KR Exon 5 Exon 6 10Kb probe R Targeting Vector Targeted Rock2 Locus pgk-neo R K R R H S N R 7.5Kb B +/- -/- +/+ 10Kb 7.5Kb C +/+ -/- ROCK2 PAK1 D Rock2+/+ Rock2-/- Fig. 1. Creation of ROCK2 knockout mice. (A) Schematic representations of the wild-type Rock2 genomic locus, the targeting vectors, and the targeted Rock2 locus. The solid boxes indicate the positions of the exons where the insertion of pgk-neo-polya cassette (grey box) was made in the targeted locus. (B) Representative Southern blot analysis of tail DNA of ROCK2 knockout mice. Genomic DNA was isolated from tails, digested by EcoRI (R), and hybridized with the external probe shown in (A). As predicted, the probe detected a 10 kb fragment in the wild-type (ROCK2þ/þ), a 7.5 kb fragment in the knockout (ROCK2 / ) and both of these fragments in the heterozygous (ROCK2þ/ ) mice. (C) Absence of ROCK2 protein in the ROCK2 knockout mice. Total brain lysates (20 mg) were isolated from adult animals of various genotypes, immunoblotted, and probed by ROCK2 antibodies. No signals were detected in ROCK2 knockout mice. The level of PAK1 was not affected in ROCK2 knockout mice. (D) Normal hippocampal formation in ROCK2 knockout mice. Nissl staining of fixed brain sections showed normal anatomy of hippocampus. Other brain regions, including the cortex and cerebellum, also showed no obvious deficits.

17 Z. Zhou et al. / Neuropharmacology 56 (2009) by a second person blind to the genotypes. The averaged data were assessed with Student s t-test Slice protein phosphorylation assay The protein lysates were extracted from acutely prepared hippocampal slices and analyzed with following antibodies: anti-phospho-cofilin (Santa Cruz Biotechnology, Santa Cruz, CA), anti-phospho-creb (Santa Cruz Biotechnology), antiphospho-erk1/2 (P44/42), anti-phospho-mek (Cell Signaling Technology, Beverly, MA), anti-map2 (Upstate Biotechnology), anti-pak1 (Santa Cruz Biotechnology), anti-pak3 (Santa Cruz Biotechnology), anti-cofilin (Cytoskeleton), anti-creb (06-863; Upstate Biotechnology), anti-rock2 (Santa Cruz Biotechnology), anti-erk1/2 and anti-mek (Cell Signaling Technology), anti-slingshot (Abcam, USA) and anti-a/ b Tubulin (Cell Signaling Technology). The conditions for preparing and maintaining slices were the same as for electrophysiological recordings. Under these conditions, the levels of both phosphorylated and total proteins analyzed in this study remained stable during the course of the experiments (up to 6 h). For each experiment, protein samples (10 mg of total proteins) of each genotype were loaded on SDS gel (whenever possible, samples from the knockout mice were loaded side by side with those from the wild-type littermates for better comparisons) and analyzed with Western blot analysis. The amount of each protein detected by the enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ) was estimated by scanning the optical density of the blot, and the averaged normalized data were evaluated with the Student s t-test Electrophysiology All electrophysiological recordings were conducted at the Schaffer/Collateral pathway in the hippocampus as previously described (Meng et al., 2002, 2003b). For LTP and LTD field recordings, the mice ranged from 2 to 5 months of age and, for whole-cell patch clamping, from 6 to 8 weeks of age. The extracellular solution contained the following (in mm): 120 NaCl, 2.5 KCl, 1.3 MgSO 4, 1.0 NaH 2 PO 4,26 NaHCO 3, 2.5 CaCl 2, and 11 D-glucose. For field EPSPs (fepsps), the recording pipette (3 MU) was filled with extracellular solution, and for whole-cell voltage-clamp recordings, the patch pipette (3 5 MU) contained the following (in mm): 140 CsMeSO 3, 5 NaCl, 1 MgCl 2, 0.2 EGTA, 10 HEPES, 3 Mg-ATP, 0.3 Na-GTP, and 5 N-ethyl bromide quaternary salt, ph 7.2, ( mosm). All data acquisition and analysis were done using pclamp 8 software (Molecular Devices, Union City, CA). n represents the number of hippocampal slices, and at most two slices from each animal were used. All averaged recording data were statistically evaluated with Student s t-test. 3. Results 3.1. Normal brain anatomy in ROCK2 knockout mice Although at least two members have been identified for the ROCK family protein kinases, ROCK2 is the predominant form that is expressed in the brain (Leung et al., 1996; Riento and Ridley, 2003; Zhao and Manser, 2005). Therefore, we generated knockout mice deficient in the expression of ROCK2 (Fig. 1A and B). The knockout mice showed no detectable amount of ROCK2 protein, but the expression of a related protein kinase PAK1 was not affected (Fig. 1C). Since ROCK2 is expressed in both the developing and mature nervous system, we performed histological analysis of fixed brain sections but found no apparent abnormalities in the gross anatomy of the CNS, including the hippocampus (Fig. 1D). The adult ROCK2 knockout mice were slightly smaller than the wild-type littermates, but exhibited no changes in home-cage behavior (data not shown). These results suggest that ROCK2 is not essential for the development of the gross structure of the CNS Deficits in hippocampal synaptic function To investigate whether ROCK2 was important in synaptic regulation, we examined the properties of synaptic physiology by recording field excitatory postsynaptic potentials (fepsps) in the CA1 region of the hippocampus. First, we tested if the basal synaptic strength was affected by comparing fepsps evoked by various stimulation intensities (input/output curve). As shown in Fig. 2A, the sizes of fepsps were significantly reduced in ROCK2 knockout mice compared to the wild-type littermates. The differences were obvious over a wide range of stimulation intensities indicating that ROCK2 is critical for maintaining a sufficient amount Fig. 2. Reduced basal synaptic strength in ROCK2 knockout mice. (A) fepsp slopes as a function of stimulation intensities showing significant reductions in both maximal synaptic responses and stimulation sensitivity in ROCK2 knockouts compared to the wild-type littermates. (B) PPF as a function of interpulse interval showing a limited reduction in response facilitation at ms intervals in ROCK2 knockout mice. of basal synaptic transmission. To test whether ROCK2 was involved in presynaptic function, we compared paired-pulse facilitation (PPF), a form of presynaptic short-term plasticity. As shown in Fig. 2B, no differences were significant for interpulse intervals except for 25 ms interval point, suggesting a mild deficit in presynaptic regulation in ROCK2 knockout mice. To further investigate whether ROCK2 is involved in synaptic plasticity, we assessed hippocampal LTP, the most extensively studied form of synaptic plasticity believed to underlie learning and memory (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999). As shown in Fig. 3A and B, LTP induced by high frequency stimulation (four trains of 100 Hz lasting 1 s each with 20 s intertrain intervals) was significantly reduced in ROCK2 knockout mice compared to the wild-type littermates. The averaged LTP was % for ROCK2þ/þ and % for ROCK2 / (p < 0.05). Another

18 84 Z. Zhou et al. / Neuropharmacology 56 (2009) A B C Last 10 min LTP ROCK2+/+ ROCK2-/- Fig. 3. Impairments in hippocampal CA1 LTP. (A) A significant reduction in CA1 LTP induced by high frequency stimulation (arrow, four trains of 100 Hz, 1 s each with a 20 s intertrain interval) in ROCK2 knockout mice. (B) Summary graph showing the averaged mean LTP of last 10 min. The magnitude of fepsps 2 h after the induction of LTP was % for ROCK2þ/þ and % for ROCK2 / (p ¼ 0.02). (C) Normal CA1 LTD induced by a brief application of metabotropic glutamate receptor agonist DHPG (100 mm, 10 min) in the ROCK2 knockout slices. The magnitude of LTD min after DHPG application was % for ROCK2þ/þ and % for ROCK2 / (p ¼ 0.34). form of hippocampal synaptic plasticity, long-term depression (LTD) induced by application of metabotropic glutamate receptor agonist (RS)-3,5-DHPG was not altered in these knockout mice (Fig. 3C). To test if LTP deficits in the knockout mice were related to changes in the channel properties of NMDA or AMPA subtype receptors, we carried out whole-cell voltage-clamp recordings from CA1 pyramidal neurons. Activation of NMDA receptors and subsequent changes in AMPA receptors are required for the induction and expression of CA1 LTP, respectively (Malinow and Malenka, 2002; Bredt and Nicoll, 2003; Collingridge et al., 2004). As shown in Fig. 4, no differences were evident in current voltage relation in either NMDA-EPSCs (excitatory postsynaptic currents) or AMPA- EPSCs, suggesting that LTP deficits in the knockout mice were not likely caused by changes in NMDA or AMPA receptor channel properties Synapse and spine deficits in ROCK2 knockout mice To investigate whether the deficits in synaptic physiology in ROCK2 knockout mice were related to synaptic structures, we analyzed the electron micrographs of hippocampal CA1 synapses. We counted the total number of asymmetric synapses and found that the density was significantly lower in ROCK2 (Fig. 5B). The averaged synapse density was /100 mm 2 for ROCK2þ/þ and /100 mm 2 for ROCK2 / (p < 0.005). Interestingly, in addition to a reduction in synapse density, the averaged postsynaptic area or spine area was significantly increased in ROCK2 knockout mice (Fig. 5C). The averaged cross section area of spines was mm 2 for the wild-type and mm 2 for ROCK2 / (p < 0.005). To further confirm that there is a reduction in excitatory synapse density, we performed immunohistochemical staining using fixed hippocampal sections with both the synaptic mark synapsin and glutamatergic synaptic marker vesicular glutamate transporter 1 (vglut1). As shown in Fig. 6, the co-localization of these two proteins was also significantly reduced in ROCK2 knockout mice. These results together suggest that the reduced basal fepsps in ROCK2 knockout mice are caused by a decrease in the density of the excitatory synapse. To investigate the mechanisms that may be responsible for the reduced synaptic density, we turned to cultured hippocampal neurons for the examination of neuronal morphology. Changes in the dendritic branches and/or spines could lead to a reduction in the number of synapses formed per neuron. In addition, previous in vitro studies have shown that perturbations of proteins involved in Rho signaling affect neuronal structures, including spines (Luo, 2000; Tashiro et al., 2000; Govek et al., 2004; Tada and Sheng, 2006). First, we compared the complexity of the dendrites by analyzing MAP2 positive processes but found no significant differences in either the number or the total length between all genotypes (Fig. 7A). Similarly, the axonal growth was not affected in young neurons in these mice (data not shown). We then analyzed spine properties by comparing spine length and density. As shown in Fig. 7C, the spine length was significantly increased in ROCK2

19 Z. Zhou et al. / Neuropharmacology 56 (2009) Fig. 4. Normal AMPA and NMDA receptor-mediated synaptic currents in ROCK2 knockout mice. Averaged normalized amplitudes of evoked AMPA receptor- (A) or NMDA receptor- (B) mediated EPSCs recorded from CA1 pyramidal neurons showing no differences in current voltage relation between the wild-type and ROCK2 knockout mice. AMPA peak currents were measured in the presence of 100 mm picrotoxin plus 100 mm APV and normalized to the EPSC at 60 mv. The NMDA currents were measured with peak amplitudes in the presence of 100 mm picrotoxin and 10 mm NBQX and normalized to the EPSC at þ60 mv. The representative traces at various holding potentials were averages of four successive sweeps. A ROCK2+/+ ROCK2-/- B Synapses/100µm C Spine area (µm 2 ) ROCK2+/+ ROCK2-/- ROCK2-/ ROCK2+/+ Fig. 5. Excitatory synapse density and cross-sectional spine areas in ROCK2 knockout mice. (A) Representative electron micrographs of hippocampal CA1 areas. Asterisks indicate structures counted as asymmetric synapses. Scale bar: 500 nm. (B,C) Summary graphs showing a significant reduction in synapse density (B) and an increase in spine area (C) in ROCK2 knockout mice. For each genotype, thin-section micrographs from two animals covering neuropil regions totaling mm 2 were used for quantification. p < with Student s t-test.

20 86 Z. Zhou et al. / Neuropharmacology 56 (2009) A a b c R2+/+ R2-/- B Fig. 6. Immunohistochemical analysis of hippocampal sections. (A) Representative images of rapidly fixed hippocampal sections stained with the synaptic marker synapsin I (a, green), vesicular glutamate transporter 1 (b, red) and merged images (c, yellow). Insets: higher magnification images (40) of CA1 areas. Scale bar: 10 mm. R2þ/þ: ROCK2þ/þ;R2 / : ROCK2 /. (B) Summary graph showing a significant reduction in the co-localization of synapsin I and vesicular transport 1 (yellow). The yellow fluorescence intensity was normalized with that of the green signals. A total of 20 equivalent hippocampal sections (two animals for each genotype) and mm 2 per section were used for quantification. p < with Student s t-test. knockout mice. The averaged spine length was mm for ROCK2þ/þ and mm for ROCK2 / (p < 0.001). Many of the ROCK2 knockout spines appeared to be filopodium-like structures that are characteristic of immature spines. Changes in spine morphology in ROCK2 knockout mice may provide the simplest explanation for the reduction in the synapse density, synaptic strength and LTP found in these knockout mice Deficits in the actin cytoskeleton and cofilin It is known that the actin cytoskeleton is a primary cellular target of Rho/ROCK signaling and that actin regulation is required for proper spine development, spine morphology and spine motility (Hall, 1998; Matus, 2000; Govek et al., 2005; Dillon and Goda, 2005). In addition, actin dynamics is intimately involved in various aspects of LTP expression, including glutamate receptor trafficking (Bredt and Nicoll, 2003; Collingridge et al., 2004; Derkach, et al., 2007) and structural remodeling at the synapse, including spine plasticity (Matus, 2000; Fukazawa et al., 2003; Okamoto et al., 2004; Zito et al., 2004; Lamprecht and LeDoux, 2004; Lin et al., 2005; Dillon and Goda, 2005; Chen et al., 2007). Therefore, we reasoned that spine and LTP deficits in ROCK2 knockout mice might be associated with actin deficits. To test this possibility, we examined the distribution of actin filaments (Factin) in spines and dendrites using cultured hippocampal neurons stained with the fluorescent F-actin dye phalloidin. We measured spine/dendrite fluorescence intensity and found that the spine/ dendrite intensity ratio was significantly reduced in ROCK2 knockout neurons (Fig. 7D), suggesting that ROCK2 is required for normal F-actin accumulations in the spine. To further elucidate the molecular mechanisms responsible for the actin changes, we analyzed the expression and activities of a wide range of proteins that may be affected by ROCK2, but found no differences between genotypes (Fig. 8A). These included: PAK1, PAK3, CREB, components involved in mitogen-activated protein kinase cascade (extracellular signal-regulated kinase-1 and 2 (ERK1/2) and MAPK/ERK kinase) and cofilin phosphatase Slingshot. Previous studies have indicated that regulation of actin depolymerization factor ADF/cofilin activity by LIM (for the three gene products Lin-11, Isl-1, and Mec-3) kinase (LIMK) is a major

21 Z. Zhou et al. / Neuropharmacology 56 (2009) A a ROCK2+/+ d ROCK2-/- b e c f B C D Spines/micron ROCK2+/+ Spine length (microns) ROCK2+/+ Spine/dendrite F-actin ratio ROCK2+/+ ROCK2-/- ROCK2-/- ROCK2-/- Fig. 7. Spine and actin properties in ROCK2 knockout mice. Cultured hippocampal neurons (20 DIV) stained with phalloidin (red) and MAP2 (green) showing normal dendritic complexity (a,b,d,e) and spine density (c,f, arrow heads), but elongated spine shapes in ROCK2 knockout neurons. (B,C) Summary graphs showing normal spine density (B) but increased spine length (C) in ROCK2 knockout mice. (D) Summary graph showing averaged mean fluorescence intensity ratios (the ratio was for the wild-type and for ROCK2 / neurons). Scale bars: 5 mm. p < with Student s t-test. The summary data were collected from three independent cultures at 20 DIV. For each genotype, a total of neurons and mm linear dendrites per neuron were used for quantification.

22 R2+/+ 88 Z. Zhou et al. / Neuropharmacology 56 (2009) A R2+/+ R2-/- B C ROCK2 PAK3 PAK1 R2+/+ MEK P-MEK P-CREB CREB 20 P-ERK1/2 0 ERK1/2 R2-/- P-Cofilin R2+/ R2-/- R2-/ P-Cofilin Cofilin Cofilin Relative pcofilin/ cofilin (%) Relative pcofilin/ cofilin (%) R2+/+0min R2+/+10min R2-/- 0min R2-/- 10min SLINGSHOT TUBULIN Fig. 8. Alterations in cofilin phosphorylation. (A) Western blot analysis of hippocampal slice lysates of ROCK2 knockout mice showing no differences in various proteins. (B) Western blot analysis of hippocampal slice lysates showing a significant reduction of p-cofilin under basal synaptic conditions in ROCK2 knockout mice. (C) Normal NMDA-induced increases in p-cofilin in acute hippocampal slices of ROCK2 knockout mice. The amount of p-cofilin 2 h after hippocampal slice recovery and before the NMDA (50 mm)/glycine (10 mm) treatment (0 min) was defined as 100%. Addition of NMDA/glycine for 10 min (in extracellular recording solution at room temperature) significantly increased the amount of p- cofilin in both the wild-type and ROCK2 knockout slices. The level of total cofilin was not altered by the treatment in either genotype. The data were from at least six independent experiments. Error bars in all graphs indicate SEM. R2þ/þ: ROCK2þ/þ; R2 / : ROCK2 /. p < 0.05 with Student s t-test. mechanism by which actin and synaptic plasticity are regulated (Bamburg, 1999; Meng et al., 2003a). ROCK2 can directly phosphorylate and activate LIM kinases, which in turn phosphorylate and inactivate ADF/cofilin (Bamburg, 1999; Maekawa et al., 1999). As shown in Fig. 8B, although the total protein level of cofilin was not altered, the phosphorylated form of cofilin was significantly reduced in ROCK2 knockout slices. However, NMDA-induced cofilin changes were not altered in ROCK2 knockout slices (Fig. 8C). These results suggest that ROCK2 is only required for basal cofilin phosphorylation/dephosphorylation regulation. 4. Discussion The Rho family small GTPases can exert dramatic effects on neuronal structures and synaptic function, but the underlying molecular mechanisms are poorly defined. In this study, we addressed this issue by analyzing genetically altered mice deficient in the expression of protein kinase ROCK2 that is specifically activated by Rho. Electrophysiological, morphological and biochemical analyses indicate that these knockout mice are affected in basal synaptic function, the actin-binding protein cofilin and the actin cytoskeleton. These results provide strong in vivo and genetic evidence that Rho/ROCK2 signaling is critical in synaptic function through regulation of cofilin and actin. Histological analysis indicates that ROCK2 knockout mice exhibit no detectable abnormalities in the gross anatomy of the CNS, indicating that ROCK2 is not required for CNS formation. The lack of anatomical deficits in these knockout mice is not likely caused by functional redundancy or developmental compensation by other proteins because the expression of these proteins is not significantly altered in the knockout mice. In addition, we were able to create double knockout mice lacking both PAK1 and ROCK2 and show that these mice are viable and also anatomically normal (data not shown). However, despite the apparently normal brain anatomy, the ROCK2 knockout mice display clear deficits in basal synaptic function with a profound reduction in evoked fepsps (Fig. 2). It is important to note that despite a dramatic reduction in basal synaptic responses the magnitude of LTP is only mildly affected, indicating that the primary function of ROCK2 is to regulate basal synaptic function rather than synaptic plasticity. Furthermore, both EM and immunohistochemical results obtained from hippocampal CA1 areas are consistent with the notion that the reduction in the number of excitatory synapses is responsible for the reduced synaptic response in the knockout mice (Figs. 5 and 6). However, it is interesting to note that spine length but not the spine density is altered in ROCK2 knockout mice (Fig. 7), suggesting that proper spine morphology is critical for synapse formation and/or maintenance. In ROCK2 knockout slices, there are significant changes in the amount of phosphorylated cofilin and F-actin under basal conditions, indicating that ROCK2 is critical for tonic cofilin and actin regulation (Fig. 8). Given the importance of cofilin and actin in spine and synaptic regulation and their potent regulation by ROCK2 (Meng et al., 2002, 2003a), it is reasonable to postulate that changes in cofilin/actin are responsible for or at least contribute to spine and synaptic deficits in ROCK2 knockout mice. It is interesting to stress that the effect of ROCK2 deletion on cofilin is very different from that of PAK1 deletion where only NMDA-induced cofilin phosphorylation is abolished (Asrar et al., 2009). These results suggest that ROCK2 is mainly important for basal cofilin regulation, whereas PAK1 is specifically important for activity-dependent cofilin regulation. This idea is consistent with the electrophysiological findings that ROCK2 and PAK1 knockout mice are impaired in basal synaptic function and hippocampal LTP, respectively. In summary, the results presented in this study provide in vivo and genetic evidence that ROCK2 plays a critical role in the regulation of spine and synaptic properties in the hippocampus. Our results are consistent with previous studies showing that pharmacological perturbations of Rho-kinases in general affect learning and memory in mice (Lamprecht et al., 2002; Dash et al., 2004). We propose that the synaptic function of ROCK2 is accomplished at least in part by regulation of cofilin activity and the actin cytoskeleton. Future studies to investigate the effect of ROCK2 deletions

23 Z. Zhou et al. / Neuropharmacology 56 (2009) on behavioral responses, the mechanisms by which ROCK2/cofilin signaling regulates synaptic responses, and ROCK2 regulation by synaptic activity will be important to understand the in vivo function of ROCK2 and the Rho GTPases in general. Acknowledgments This work was supported by grants from the Canadian Institutes of Health Research (CIHR MOP-42396) and Ontario Mental Health Foundation. We are grateful to Guijin Lu, Shouping Zhang, Wayne Huang and Y.M. Heng for technical assistance. References Asrar, S., Meng, Y., Zhou, Z., Todorovski, Z., Huang, W.W., Jia, Z., Regulation of hippocampal long-term potentiation by p21-activated protein kinase 1 (PAK1). Neuropharmacology 56, Bamburg, J.R., Proteins of the ADF/cofilin family: essential regulators of actin dynamics. Annu. Rev. Cell Dev. Biol. 15, Bliss, T.V.P., Collingridge, G.L., A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361, Bredt, D.S., Nicoll, R.A., AMPA receptor trafficking at excitatory synapses. Neuron 40, Carlisle, H.J., Kennedy, M.B., Spine architecture and synaptic plasticity. Trends Neurosci. 28, Chelly, J., Mandel, J., Monogenic causes of X-linked mental retardation. Nat. Rev. Genet. 2, Chen, L.Y., Rex, C.S., Casale, M.S., Gall, C.M., Lynch, G., Changes in synaptic morphology accompany actin signaling during LTP.J. Neuroscience 27, Collingridge, G.L., Isaac, J.T., Wang, Y.T., Receptor trafficking and synaptic plasticity. Nat. Rev. Neurosci. 5, Dash, P.K., Orsi, S.A., Moody, M., Moore, A.N., A role of hippocampal Rho ROCK pathway in long-term spatial memory. Biochem. Biophys. Res. Commun. 322, Derkach, V.A., Oh, M.C., Guire, E.S., Soderling, T.R., Regulatory mechanisms of AMPA receptors in synaptic plasticity. Nat. Rev. Neurosci. 8, Diana, G., Valentini, G., Travaglione, S., Falzano, L., Massimo, P., Zona, C., Meschini, S., Fabbri, A., Fiorentini, C., Enhancement of learning and memory after activation of cerebral Rho GTPases. Proc. Natl. Acad. Sci. U.S.A. 104, Dillon, C., Goda, Y., The actin cytoskeleton: integration form and function at the synapse. Annu. Rev. Neurosci. 28, Fukazawa, Y., Saitoh, Y., Ozawa, F., Ohta, Y., Mizuno, K., Inokuchi, K., Hippocampal LTP is accompanied by enhanced F-actin content within the dendritic spine that is essential for late LTP maintenance in vivo. Neuron 38, Govek, E.E., Newey, S.E., Akerman, C.J., Cross, J.R., Van der Veken, L., Van Aelst, L., The X-linked mental retardation protein oligophrenin-1 is required for dendritic spine morphogenesis. Nat. Neurosci. 7, Govek, E.E., Newey, S.E., Van Aelst, L., The role of the RHO GTPases in neuronal development. Genes Dev. 19, Hall, A., Rho GTPases and the actin cytoskeleton. Science 279, Hering, H., Sheng, M., Dendritic spine: structure, dynamics and regulation. Nat. Rev. Neurosci. 2, Jia, Z.P., Agopyan, N., Miu, P., Xiong, Z., Henderson, J., Gerlai, R., Taverna, F., Velumian, A., MacDonald, J.F., Carlen, P., Abranow-Newerly, W., Roder, J., Enhanced LTP in mice deficient in the AMPA receptor, GluR2. Neuron 17, Kim, C.H., Lisman, J.E., A role of actin filaments in synaptic transmission and long-term potentiation. J. Neurosci. 19, Krucker, T., Siggins, G.R., Halpain, S., Dynamic actin filaments are required for stable long-term potentiation (LTP) in area CA1 of hippocampus. Proc. Natl. Acad. Sci. U.S.A. 97, Lamprecht, R., LeDoux, J., Structural plasticity and memory. Nat. Rev. Neurosci. 5, Lamprecht, R., Farb, C.R., LeDoux, J.E., Fear memory formation involves p190 RhoGAP and ROCK proteins through a GRB2-mediated complex. Neuron 36, Leung, T., Chen, X.Q., Manser, E., Lim, L., The p160 RhoA-binding kinase ROKa is a member of a kinase family and is involved in the reorganization of the cytoskeleton. Mol. Cell. Biol. 16, Li, Z., Aizenman, C.D., Cline, H.T., Regulation of rho GTPases by crosstalk and neuronal activity in vivo. Neuron 33, Lin, B., Kramár, E.A., Bi, X., Brucher, F.A., Gall, C.M., Lynch, G., Theta stimulation polymerizes actin in dendritic spines of hippocampus. J. Neurosci. 25, Lisman, J., Actin s action in LTP-induced synapse growth. Neuron 38, Luo, L., Rho GTPases in neuronal morphogenesis. Nat. Rev. Neurosci. 1, Luo, L., Actin cytoskeleton regulation in neuronal morphogenesis and structural plasticity. Annu. Rev. Cell Dev. Biol. 18, Maekawa, M., Ishizaki, T., Boku, S., Watanabe, N., Fujita, A., Iwamatsu, A., Obinata, T., Ohashi, K., Mizuno, K., Narumiya, S., Signaling from Rho to the actin cytoskeleton through protein kinases ROCK and LIM-kinase. Science 285, Malenka, R.C., Nicoll, R.A., Long-term potentiation-a decade of progress? Science 285, Malinow, R., Malenka, R.C., AMPA receptor trafficking and synaptic plasticity. Annu. Rev. Neurosci. 25, Matus, A., Actin-based plasticity in dendritic spines. Science 290, Meng, Y.H., Zhang, Y., Tregoubov, V., Janus, C., Cruz, L., Jackson, M., Lu, W.Y., MacDonald, J.F., Wang, J.Y., Falls, D.L., Jia, Z.P., Abnormal spine morphology and enhanced LTP in LIMK-1 knockout mice. Neuron 35, Meng, Y.H., Zhang, Y., Tregoubov, V., Falls, D.L., Jia, Z.P., 2003a. Regulation of spine morphology and synaptic function by LIMK and the actin cytoskeleton. Rev. Neurosci. 14, Meng, Y.H., Zhang, Y., Jia, Z.P., 2003b. Synaptic transmission and plasticity in the absence of AMPA glutamate receptor GluR2 and GluR3. Neuron 39, Meng, J.S., Meng, Y.H., Hanna, A., Janus, C., Jia, Z.P., Abnormal long-lasting synaptic plasticity and cognition in mice lacking the mental retardation gene Pak3. J. Neurosci. 25, Nakayama, A.Y., Harms, M.B., Luo, L., Small GTPases Rac and Rho in the maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J. Neurosci. 20, O Kane, E.M., Stone, T.W., Morris, B.J., Activation of Rho GTPases by synaptic transmission in the hippocampus. J. Neurochem. 87, O Kane, E.M., Stone, T.W., Morris, B.J., Increased long-term potentiation in the CA1 region of rat hippocampus via modulation of GTPase signaling or inhibition of Rho kinase. Neuropharmacology 46, Okamoto, K., Nagai, T., Miyawaki, A., Hayashi, Y., Rapid and persistent modulation of actin dynamics regulates postsynaptic reorganization underlying bidirectional plasticity. Nat. Neurosci. 7, Ramakers, G.J.A., Rho proteins, mental retardation and the cellular basis of cognition. Trends Neurosci. 25, Riento, K., Ridley, A., ROCKS: multifunctional kinases in cell behavior. Nat. Rev. Mol. Cell Biol. 4, Sin, W.C., Hass, K., Ruthazer, E.S., Cline, H.T., Dendritic growth increased by visual activity requires NMDA receptor and Rho GTPases. Nature 419, Smart, F.M., Halpain, S., Regulation of dendritic spine stability. Hippocampus 10, Sorra, K.E., Harris, K.M., Overview on the structure, composition, function, development, and plasticity of hippocampal dendritic spines. Hippocampus 10, Tada, T., Sheng, M., Molecular mechanisms of dendritic spine morphogenesis. Curr. Opin. Neurobiol. 16, Tashiro, A., Minden, A., Yuste, R., Regulation of dendritic spine morphology by the Rho family small GTPases: antagonistic roles of Rac and Rho. Cereb. Cortex 10, Thumkeo, D., Keel, J., Ishizaki, T., Hirose, M., Nonomura, K., Oshima, H., Oshima, M., Taketo, M.M., Narumiya, S., Targeted disruption of the mouse rho-associated kinase 2 gene results in intrauterine growth retardation and fetal death. Mol. Cell. Biol. 23, Van Aelst, L., Cline, H.T., Rho GTPases and activity-dependent dendrite development. Curr. Opin. Neurobiol. 14, Van Galen, E.J., Ramakers, G.J., Rho proteins, mental retardation and the neurobiological basis of intelligence. Prog. Brain Res. 147, Zhao, Z.S., Manser, E., PAK and other Rho-associated kinases-effectors with surprisingly diverse mechanisms of regulation. Biochem. J. 386, Zito, K., Knott, G., Shepherd, G.M., Shenolikar, S., Svoboda, K., Induction of spine growth and synapse formation by regulation of the spine actin cytoskeleton. Neuron 44,

24 Neuropharmacology 56 (2009) Contents lists available at ScienceDirect Neuropharmacology journal homepage: Regulation of hippocampal long-term potentiation by p21-activated protein kinase 1 (PAK1) Suhail Asrar a,b,1, Yanghong Meng a,1, Zikai Zhou a,b, Zarko Todorovski a,b, Wayne Wenyin Huang a,b, Zhengping Jia a,b, a Neurosciences and Mental Health, The Hospital for Sick Children, 555 University Avenue, Toronto, ON M5G 1X8, Canada b Department of Physiology, University of Toronto, Toronto, ON, Canada article info abstract Article history: Received 25 April 2008 Received in revised form 19 June 2008 Accepted 20 June 2008 Keywords: Rho GTPase PAK1 knockout mice Actin cytoskeleton Cofilin Hippocampal long-term potentiation The Rho family small GTPases are critically involved in the regulation of spine and synaptic properties, but the underlying mechanisms are poorly defined. We took genetic approaches to create and analyze knockout mice deficient in the expression of the protein kinase PAK1 that is directly associated with and activated by the Rho GTPases. We demonstrated that while these knockout mice were normal in both basal and presynaptic function, they were selectively impaired in long-term potentiation (LTP) at hippocampal CA1 synapses. Consistent with the electrophysiological deficits, the PAK1 knockout mice showed changes in the actin cytoskeleton and the actin binding protein cofilin. These results indicate that PAK1 is critical in hippocampal synaptic plasticity via regulating cofilin activity and the actin cytoskeleton. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction The Rho family small GTPases are guanine nucleotide binding proteins that function as molecular switches cycling between the active GTP-bound state to GDP-bound inactive state. There are at least 14 distinct family members in the mammalian cells, the best characterized being RhoA, Rac1 and Cdc42. These are involved in many cellular processes but best known for their effects on various aspects of cell morphology and motility through the regulation of the actin cytoskeleton (Hall, 1998; Govek et al., 2005). In neurons, the function of Rho proteins is especially prominent because of the complex nature of the neuronal structures. Of particular importance is their effects on the dendritic spine, a highly specialized actin-rich protrusion where the majority of excitatory synapses are formed (Sorra and Harris, 2000; Carlisle and Kennedy, 2005). Thus, manipulations of Rho proteins and their regulators have been shown to potently affect spine properties, synaptic transmission and plasticity, as well as learning and memory (Luo, 2000; Tashiro et al., 2000; O Kane et al., 2003, 2004; Dash et al., 2004; Van Aelst and Cline, 2004; Tada and Sheng, 2006; Diana et al., 2007). Consistent with these animal studies, genetic deficits in Rho signaling have been found in patients with mental retardation (Chelly and Mandel, 2001; Ramakers, 2002). Corresponding author. Tel.: þ ; fax: þ address: zhengping.jia@sickkids.ca (Z. Jia). 1 These authors contributed equally to this work. The exact mechanisms by which Rho GTPases regulate spine and synaptic properties remain unknown. However, the actin network is thought to play an important role in this process (Luo, 2002; Dillon and Goda, 2005; Govek et al., 2005). This is consistent with the observations that the actin cytoskeleton is highly enriched in the dendritic spine and that changes of actin dynamics correlate with or affect numerous aspects of spine properties and synaptic plasticity (Kim and Lisman, 1999; Krucker et al., 2000; Matus, 2000; Meng et al., 2002; Fukazawa et al., 2003; Zito et al., 2004; Dillon and Goda, 2005; Lin et al., 2005; Chen et al., 2007). Although many effector proteins have been identified that could relay signals from the Rho GTPases to actin (Govek et al., 2005), the protein kinases that are directly activated by the GTPases have attracted increasing attention as they are highly expressed in the CNS and are potent regulators of actin dynamics (Bokoch, 2003; Zhao and Manser, 2005). In this study, we employed genetic approaches to investigate the role of one of these kinases, PAK1 (p21-activated kinase 1) that is specifically activated by Rac and Cdc42. We demonstrated that while PAK1 knockout mice were normal in both basal synaptic transmission and presynaptic function, they were selectively impaired in long lasting synaptic plasticity in the hippocampus. Consistent with these electrophysiological deficits, PAK1 knockout mice exhibited abnormalities in the actin cytoskeleton and activity-dependent regulation of the actin binding protein cofilin. These results provide strong genetic and in vivo evidence that PAK1 is critical in synaptic plasticity through the regulation of cofilin /$ see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi: /j.neuropharm

25 74 S. Asrar et al. / Neuropharmacology 56 (2009) Materials and methods 2.1. Generation of PAK1 knockout mice To create PAK1 knockout mice, a BAC clone containing the ATG exon was isolated from 129sv genomic library and sequenced to confirm the identity of the PAK1 gene. A PAK1 targeting vector was constructed by replacing a part of the ATG exon and adjacent upstream intronic sequence by a pgk-neomycin resistant gene cassette (Fig. 1A). The vector was designed to completely eliminate the PAK1 protein. The vector was used to transfect R1 embryonic stem (ES) cells (derived from the 129sv mouse strain) and G418 resistant colonies were screened for a targeted event by Southern blot analysis. DNA probes external to the targeting vector was used for analyzing both ES and mouse tail DNA. Two positive ES clones were used to generate chimeric mice by aggregation. The chimeras were crossed with CD1 mouse strain to generate F1 populations. The knockout offspring ( / ) and control littermates (þ/þ) from F1 (þ/ ) inter-breedings were used for this study. Analyses of the numbers of various genotypes of F2 populations indicated that the PAK1 knockout mice were normal in viability, fertility and life span. No home-cage behavioral abnormalities were observed in PAK1 knockout mice Histology, electron microscopy and hippocampal culture The procedures for histology and electron microscopy (EM) of fixed brain tissues were described previously (Jia et al., 1996; Meng et al., 2002). For EM analysis, asymmetric synapses were defined by the presence of a prominent postsynaptic density and associated presynaptic vesicles. To compare synapse density and spine area, thin section electron micrographs were randomly collected and analyzed from equivalent CA1 regions. For each genotype, a total of images covering approximately mm 2 from two animals were used for quantification. For low-density cultures, hippocampal neurons were prepared from postnatal pups (P0 1) and maintained according to a procedure recommended for Neurobasal-A medium (Invitrogen, San Diego, CA). For F-actin labeling, fixed cells were permeabilized with 0.1% Triton X-100/PBS and stained for 30 min with 1 mg/ml tetramethylrhodamine isothiocyanate (TRITC)-conjugated phalloidin (Sigma, St. Louis, MO). Neuronal images were acquired using a confocal microscope with 20 or 60 objectives under identical gain and contrast settings for the wild-type and knockout samples. For each genotype, a total of neurons from three independent hippocampal cultures and approximately mm 2 dendritic segments per neuron were used for quantification. To quantify the distribution of F-actin, the confocal images were converted to grayscale and the intensity of the fluorescence was measured and compared between the spine head and an adjacent dendritic area using a Northern Eclipse program (Toronto) as described previously (Meng et al., 2002). The averaged data were assessed with Student s t-test Slice protein phosphorylation assay In order to correlate electrophysiological and biochemical data, we analyzed protein lysates extracted from acutely prepared hippocampal slices. The following antibodies were used: anti-phospho-cofilin (Santa Cruz Biotechnology, Santa Cruz, CA), anti-phospho-creb (Santa Cruz Biotechnology), anti-phospho-erk1/2 (P44/42) (Cell Signaling Technology, Beverly, MA), anti-phospho-jnk (Cell Signaling Technology), anti-phospho-mlc (#3671, Cell Signaling Technology), anti-phospho-mlck (Santa Cruz Biotechnology), anti-glua2 and anti-glua3 (Chemicon, Temecula, CA), anti-pak1 (Santa Cruz Biotechnology), anti-pak2 (Santa Cruz Biotechnology), anti- PAK3 (Santa Cruz Biotechnology), anti-cofilin (Cytoskeleton), anti-creb (06 863; Upstate Biotechnology), anti-rock2 (Santa Cruz Biotechnology), anti-erk1/2 (Cell Signaling Technology), anti-mlc (#3672, Cell Signaling Technology), and anti-mlck (Santa Cruz Technology). The conditions for preparing and maintaining slices were the same as for electrophysiological recordings (see below). For each experiment, protein samples (10 mg of total proteins) of each genotype were loaded on SDS gel A B B H Wildtype PAK1 Locus Targeting Vector probe 6.0Kb Exon1 B B B pgk-neo B B B B H Targeted PAK1 Locus 4.7Kb B +/- -/- +/+ C 6.0Kb 4.7Kb PAK1 PAK3 D PAK1+/+ +/+ +/- -/- PAK1-/- Fig. 1. Creation of PAK1 knockout mice. (A) Schematic representations of the wild-type Pak1 genomic locus, the targeting vector, and the targeted Pak1 locus. The solid box indicates the position of the exon where the insertion of pgk-neo-polya cassette was made in the targeted locus. (B) Representative Southern blot analysis of tail DNA of PAK1 knockout mice. Genomic DNA was isolated from mouse tails, digested by BamHI (B), and hybridized with the external probe shown in (A). As expected, the probe detected a 6 kb fragment in the wild-type (PAK1þ/þ), a 4.7 kb fragment in the knockout (PAK1 / ) and both fragments in the heterozygous (PAK1þ/ ) mice. (C) Absence of PAK1 protein in the PAK1 knockout mice. Total brain lysates (20 mg) were isolated from adult animals of various genotypes, immunoblotted, and probed by PAK1 antibodies. No signals were detected in PAK1 knockout mice, confirming the absence of PAK1 protein in the knockout mice. The level of PAK3 was not affected. (D) Normal hippocampal formation in PAK1 knockout mice. Nissl staining of fixed brain sections showed normal anatomy of hippocampus. Other brain regions also showed no obvious abnormalities.

26 S. Asrar et al. / Neuropharmacology 56 (2009) (whenever possible, samples from the knock-out mice were loaded side by side with those from the wild-type littermate for better comparisons) and analyzed with Western blot analysis. The amount of each protein detected by the enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ) method was estimated by scanning optical density of the blot, and the averaged normalized data were evaluated with Student s t-test Electrophysiology All electrophysiological recordings were conducted at the Schaffer/Collateral pathway in the hippocampus as previously described (Meng et al., 2002, 2003b). For LTP and LTD field recordings, the mice ranged from 2 to 8 weeks of age and, for whole-cell patch clamping, from 2 to 5 weeks of age. The extracellular solution contained the following (in mm): 120 NaCl, 2.5 KCl, 1.3 MgSO 4, 1.0 NaH 2 PO 4,26 NaHCO 3, 2.5 CaCl 2, and 11 D-glucose. For field EPSPs (fepsps), the recording pipette (3 MU) was filled with extracellular solution, and for whole-cell voltage-clamp recordings, the patch pipette (3 5 MU) contained the following (in mm): 140 CsMeSO 4, 5 NaCl, 1 MgCl 2, 0.2 EGTA, 10 HEPES, 3 Mg-ATP, 0.3 Na-GTP, and 5 N-ethyl bromide quaternary salt, ph 7.2, ( mosm). All data acquisition and analyses were done using pclamp 8 software (Molecular Devices, Union City, CA). n represents the number of hippocampal slices, and at most two slices from each animal were used. All averaged recording data were statistically evaluated with Student s t-test. 3. Results 3.1. Normal brain anatomy in PAK1 knockout mice Although there are multiple members of the PAK family protein kinases, PAK1 appears to be the predominant form that is expressed in brain tissues, including neurons (Manser et al., 1994; Ong et al., 2002; Zhao and Manser, 2005). In addition, a number of previous studies (Boda et al., 2004; Hayashi et al., 2004; Zhang et al., 2005) have suggested that PAK1 is a likely candidate required for neuronal structure and synaptic regulation. Therefore, we generated knockout mice deficient in the expression of PAK1 (Fig. 1, see Section 2 for details). The knockout mice showed no detectable amount of PAK1, but the expression of their closest family members was not affected (Fig. 1C). Since PAK1 is expressed in the developing nervous system, we performed histological analysis of fixed-brain sections but found no apparent abnormalities in the gross anatomy of the CNS, including the hippocampus (Fig. 1D). The PAK1 knockout mice also showed no changes in fertility, viability, lifespan, or home-cage locomotor activities as assessed by daily behavioral observations. These results suggest that PAK1 is not required for the development of gross brain structure Selective deficits in hippocampal LTP To investigate the role of PAK1 in hippocampal synaptic function, we examined the properties of synaptic physiology by recording field excitatory postsynaptic potentials (fepsps) in the CA1 region of the hippocampus. First, we assessed the basal synaptic strength by comparing the sizes of fepsps evoked by various stimulation intensities, but found no differences between the wildtype and PAK1 knockout mice in either the maximal responses or stimulation sensitivity (Fig. 2A,B). Similarly, no differences were significant in paired-pulse facilitation (PPF), a form of presynaptic short-term plasticity (Fig. 2D). These results suggest that both basal synaptic strength and presynaptic function were unaltered in PAK1 knockout mice. We then examined hippocampal LTP, the most extensively studied form of synaptic plasticity believed to underlie learning and memory (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999). As shown in Fig. 3A,B, LTP was dramatically reduced in PAK1 knockout mice compared to the wild-type littermates. The averaged LTP was % for PAK1þ/þ and % for PAK1 / (p ¼ ). Another form of hippocampal synaptic plasticity, long-term depression (LTD) induced either by application of metabotropic glutamate receptor agonist (RS)-3,5-DHPG (Fig. 3C) or by low frequency stimulation (1 Hz lasting 15 min) (Fig. 3D) was not altered in these knockout mice. Therefore, PAK1 appears to be selectively important in hippocampal LTP. To test if LTP deficits in PAK1 knockout mice were related to the channel properties or synaptic expression of glutamate receptors, we carried out whole-cell voltage-clamp recordings from CA1 pyramidal neurons. Both NMDA and AMPA receptors are important in the generation of CA1 LTP (Malinow and Malenka, 2002; Bredt and Nicoll, 2003; Collingridge et al., 2004) and therefore any changes in these receptors could potentially impair LTP. As shown in Fig. 4A,B, no differences were evident in current voltage relation in either excitatory postsynaptic currents mediated by NMDA receptors (NMDAR-EPSCs) or by AMPA receptors (AMPAR-EPSCs) between genotypes. Similarly, no differences were found in AMPA-/NMDA- EPSC ratios (Fig. 4C), suggesting that glutamate receptor trafficking or synaptic expression was not significantly altered in these knockout mice. These results together suggest that LTP deficits in PAK1 knockout mice were not likely caused by changes in NMDA or AMPA receptors Normal synaptic and spine structures in PAK1 knockout mice To further investigate whether the deficits in hippocampal LTP in PAK1 knockout mice were related to their synaptic structures, we analyzed the electron micrographs of hippocampal CA1 areas of fixed brain sections. We counted the total number of asymmetric synapses but found no significant differences between the Fig. 2. Normal basal synaptic strength and presynaptic function in PAK1 knockout mice. (A, B) fepsp slopes as a function of stimulation intensity (A) and presynaptic fiber volley (B) in PAK1 knockout mice, showing no differences in either maximal synaptic responses or stimulation sensitivity between the wild-type and PAK1 knockout mice. The representative traces in A were averages of three successive sweeps at stimulation intensity 25, 50, and 100 ma. (C) PPF as a function of interpulse interval showing no changes in PAK1 knockout mice. The traces were representatives of three successive paired responses at 50 ms interpulse interval.

27 76 S. Asrar et al. / Neuropharmacology 56 (2009) Fig. 3. Selective deficits in hippocampal CA1 LTP. (A) A dramatic reduction in CA1-LTP induced by high frequency stimulation (arrow, 4 trains of 100 Hz, 1 s each with a 20 s intertrain interval) in PAK1 knockout mice. (B) Summary graph showing the magnitudes of LTP of the last 10 min following LTP induction. The magnitude of fepsps 2 h after the induction of LTP was % for PAK1þ/þ and % for PAK1 / (p ¼ ). (C) Normal CA1-LTD induced by a brief application of metabotropic glutamate receptor agonist DHPG (100 mm, 10 min) in the PAK1 knockout slices. The magnitude of LTD min after DHPG application was % for PAK1þ/þ and % for PAK1 / (p ¼ 0.4). (D) Normal CA1 LTD induced by low frequency stimulation (LFS, 1 Hz lasting 15 min). The magnitude of LTD min after low frequency stimulation was % for PAK1þ/þ and % for PAK1 / (p ¼ 0.6). The representative traces were averages of three successive sweeps at indicated time points before (1) and after LTP or LTD (2, 3). wild-type and PAK1 knockout mice (Fig. 5). The averaged synapse density was /100 mm 2 for PAK1þ/þ and / 100 mm 2 for PAK1 / (p ¼ 0.58). Similarly, no differences were seen in the postsynaptic area, an indicator of normal spine morphology. The averaged cross-section area of spines was mm 2 for the PAK1þ/þ and mm 2 for PAK1 / (p ¼ 0.55). These results suggest that neither spines nor synapses were altered in the knockout mice. The absence of structural deficits in the synapse was consistent with the electrophysiological findings that the basal synaptic efficacy was not impaired in PAK1 knockout mice Deficits in spine actin filaments and cofilin Since actin can potentially regulate synaptic plasticity through mechanisms that are independent of structural regulation (Matus, 2000; Derkach et al., 2007), we examined actin properties using cultured hippocampal neurons stained with the fluorescent F-actin dye phalloidin. Consistent with many previous studies, we showed that actin filaments were highly enriched in the spine compared to adjacent dendritic areas in the wild-type neurons, but this spine enrichment of F-actin was significantly reduced in PAK1 knockout neurons (Fig. 6A,B). The spine/dendrite fluorescence intensity ratio

28 S. Asrar et al. / Neuropharmacology 56 (2009) Fig. 4. Normal AMPA and NMDA receptors-mediated synaptic currents. Averaged normalized amplitudes of evoked AMPA receptor-mediated (A) or NMDA receptor-mediated (B) EPSCs recorded from CA1 pyramidal neurons showing no differences in current voltage relation between the wild-type and PAK1 knockout neurons. AMPAR peak currents were estimated in the presence of 100 mm picrotoxin plus 100 mm APV and normalized to the EPSC at 60 mv. The NMDA currents were measured with peak amplitudes in the presence of 100 mm picrotoxin and 10 mm NBQX and normalized to the EPSC at þ60 mv. The representative traces at various holding potentials were averages of three successive sweeps. (C) Normal AMPAR-/NMDAR-EPSC ratios in PAK1 knockout mice. AMPAR-EPSCs were first recorded at þ40 mv in the presence of 100 mm picrotoxin plus 100 mm APV, followed by APV washout (15 min) and addition of 10 mm NBQX to reveal NMDAR-EPSCs. AMPAR-/NMDAR-EPSC ratio was for PAK1þ/þ and for PAK1 / (p ¼ 0.9). The representative traces were averages of three successive sweeps of either AMPA-EPSCs (black) or NMDA-EPSCs (gray). was for PAK1þ/þ and for PAK1 / neurons (p ¼ ). To identify the molecular targets that might be responsible for the actin and synaptic changes in PAK1 knockout mice, we analyzed the expression of a number of proteins, including glutamate receptors and PAK related kinase ROCK2, but found no differences (Fig. 6C). Mitogen-activated protein kinase (MAPK) cascade is known to be an important substrate of PAK1 phosphorylation in cultured cell lines and it is shown to be critically involved in both spine and synaptic regulation (Bokoch, 2003; Thomas and Huganir, 2004). However, no differences were found between genotypes in any of the components tested, including extracellular signal-regulated kinase-1 and -2 (ERK1/2), MAPK/ERK kinase (MEK), and JNK (Fig. 6D). The transcription factor camp-responsive element binding protein (CREB) is also known to be a target for Rho activation (Meng et al., 2005; Laumonnier et al., 2007), but its level or activity was not altered in PAK1 knockout mice (Fig. 6D). We then turned to effector proteins that are directly involved in the regulation of the actin cytoskeleton. Previous studies have indicated that regulation of the actin depolymerization factor cofilin by LIM (for the three gene products Lin-11, Isl-1, and Mec-3) kinases (LIMK) represents a key mechanism by which actin and synaptic plasticity are regulated (Bamburg, 1999; Meng et al., 2003a). PAK1 can directly phosphorylate and activate LIM kinases, which in turn phosphorylate and inactivate cofilin (Edwards et al., 1999; Bokoch, 2003). As shown in Fig. 6D, under basal conditions no differences were found in either total or phosphorylated forms of cofilin between the wild-type and PAK1 knockout slices. However, NMDA-induced increase in phosphorylated cofilin was significantly attenuated in PAK1 knockout slices (Fig. 6E), suggesting that PAK1 is specifically required for activity-dependent cofilin regulation. Another well-studied pathway shown to be regulated by PAK1 and implicated in actin and spine regulation is the myosin regulatory light chain (MLC) and its kinase (MLCK) (Zhang et al., 2005; Ryu et al., 2006). However, we found no differences in either total or phosphorylated forms of MLC or MLCK (data not shown). These results are consistent with the idea that synaptic and actin deficits in PAK1 knockout mice are most likely caused by abnormal cofilin regulation. 4. Discussion As a key effector of the Rho GTPases, PAK1 perturbations have been previously shown to affect both spine properties and cortical synaptic function (Hayashi et al., 2004; Zhang et al., 2005); however, the underlying mechanisms responsible for these effects are not clear. In addition, whether PAK1 plays a role in hippocampal synaptic function remains unknown. In this study, we investigated these issues by generating and analyzing knockout mice deficient in the expression of PAK1. Electrophysiological, morphological and biochemical analyses indicate that while these knockout mice are normal in spine/synaptic structures, basal synaptic transmission and presynaptic function, they are specifically impaired in hippocampal LTP. The knockout mice are also altered in the actin cytoskeleton and activity-dependent cofilin regulation. These results suggest that PAK1 regulates hippocampal synaptic plasticity via mechanisms involving cofilin and actin but independent of structural changes. Histological analysis indicates that PAK1 knockout mice exhibit no detectable abnormalities in the gross anatomy of the CNS. The lack of deficits in gross brain anatomy in these knockout mice is not likely caused by the developmental compensation by other members of the PAK family because the expression of these proteins is not significantly altered in the knockout mice. In addition, double knockout mice lacking both PAK1 and its closest family member PAK3 (Meng et al., 2005) are viable and show no detectable changes in brain gross anatomy (data not shown). Furthermore, we were able to create double knockout mice lacking both PAK1 and Rho kinase ROCK2 and show that these mice are viable and also anatomically normal (data not shown). These results indicate that the Rho GTPase-activated protein kinases in general do not play critical roles in the development of gross brain structure. Our results also indicate that PAK1 is not critical for the ultrastructures of the neurons as both spines and synapses are normal in PAK1 knockout mice. These results are inconsistent with those obtained from the dominant negative PAK1 (dnpak1) transgenic mice where both spine and synaptic properties appear to be altered (Hayashi et al., 2004). There are a number of possibilities that may

29 78 S. Asrar et al. / Neuropharmacology 56 (2009) A PAK1+/+ PAK1-/- B C Synapses/100μm Spine area (μm 2 ) PAK1+/ PAK1-/- PAK1+/+ PAK1-/- Fig. 5. Normal synapse and spine density in PAK1 knockout mice. (A) Representative electron micrographs of hippocampal CA1 areas. Asterisks indicate structures counted as asymmetric synapses. Scale bar: 500 nm. (B, C) Summary graphs showing no differences in either synapse density (B) or spine/postsynaptic areas (C) between the wild-type and PAK1 knockout mice. For each genotype, thin-section micrographs from two animals covering neuropil regions totaling mm 2 were used for quantification. account for this inconsistency. One is that the effect of dnpak1 expression may be more drastic than the PAK1 deletion of this study as the dnpak1 not only affects PAK1 but also other members of the PAK family. It would therefore be important to examine spines and synapses in PAK1/3 double knockout mice. It is also possible that the differences in the results of these two studies are related to the intrinsic differences between the cortex and hippocampus, thus it would be interesting to analyze spine and synaptic properties in the cortex of the PAK1 knockout mice. It should be pointed out that in dnpak1 transgenic mice, the inhibitory effect of dnpak1 expression is largely restricted to the cortex without significant inhibition of hippocampal PAK activities; therefore, the role of PAK1 in the hippocampus cannot be studied using this animal model. Lastly, we cannot completely rule out the possibility that PAK1 knockout mice may suffer developmental deficits or compensations that are not detected and contribute to the differences between this study and those using dnpak1 transgenic mice. In PAK1 knockout mice, neither the amplitude of evoked fepsps nor PPF are altered, suggesting that PAK1 is not critical for either basal synaptic responses or presynaptic function. These electrophysiological observations are consistent with the findings that both spines and synapses are normal in CA1 areas of the knockout mice. However, hippocampal CA1 LTP is dramatically reduced in PAK1 knockout mice, indicating that PAK1 is selectively important in this form of synaptic plasticity. It is again important to note the differences between our results and those obtained from the dnpak1 transgenic mice where both basal synaptic responses and LTP are enhanced in the cortex (Hayashi et al., 2004). While the reasons for this discrepancy remain unclear, the factors discussed above may also contribute towards these disparities. It would also be especially important to assess electrophysiological properties of cortical neurons in PAK1 and PAK1/3 double knockout mice to determine whether they have similar deficits as those observed in dnpak1 transgenic mice. The mechanisms underlying the impaired LTP in PAK1 knockout mice is unknown; however, based on the findings that both spine F-actin and phosphorylated cofilin are reduced in the PAK1 knockout mice, it is likely that PAK1 regulates hippocampal LTP through regulation of cofilin activities and actin dynamics without morphological involvement. Previous studies have indicated that perturbations of actin dynamics affect hippocampal synaptic transmission and LTP without obvious effects on spine structures (Kim and Lisman, 1999). Indeed, actin has been shown to directly participate in the regulation of receptor trafficking and channel properties (Matus, 2000); therefore, it would be important to analyze glutamate receptors and associated proteins to determine whether the activity-dependent synaptic expression and trafficking properties are affected in PAK1 knockout mice. In addition to cofilin and actin, other molecular targets of PAK1 may also play a role in mediating its effect on hippocampal LTP because changes in cofilin activities alone by deleting its kinase LIMK-1 resulted in both altered spine morphology and enhanced LTP (Meng et al., 2002). Since NMDA-induced but not basal cofilin activity is altered in PAK1 knockout mice, it is reasonable to suggest that PAK1 is specifically important in activity-dependent regulation

30 P1-/-0min S. Asrar et al. / Neuropharmacology 56 (2009) A a PAK+/+ b c PAK-/- B Spine/shaft actin intensity ratio PAK+/+ PAK1-/- C PAK1+/+ PAK1-/- D PAK1+/+ PAK1-/- E PAK1 Cofilin PAK1+/+ PAK1-/ P-Cofilin P-Cofilin Relative pcofilin / cofilin (%) Cofilin PAK3 PAK2 ROCK2 GluA3 GluA2 P-CREB P-JNK P-ERK1/2 P-MEK Actin Actin P1+/+0min P1+/+10min P1-/-10min Fig. 6. Altered F-actin and cofilin activity. (A) Cultured hippocampal neurons (20 DIV) stained with phalloidin, showing a higher accumulation of F-actin in virtually all spines (arrowheads) than the adjacent dendritic shaft (arrows) in the wild-type (a). In PAK1 knockout neurons (b, c), many spines failed to accumulate F-actin. Circled areas indicate representative sample regions where the fluorescence intensity was measured. (B) Summary graph showing a significant reduction in the averaged mean fluorescence intensity ratio (spine/dendrite) in PAK1 knockout neurons. The ratio was for the wild-type and for PAK1 / (p < 0.001). (C) Western blot analysis of hippocampal slice lysates showing no changes in the protein levels of PAK3 (96 7%, n ¼ 5, p > 0.05), PAK2 (92 7%, n ¼ 6, p > 0.05), ROCK2 (102 6%, n ¼ 6, p > 0.05), GluA2 (105 4%, n ¼ 6, p > 0.05) and GluA3 (98 5%, n ¼ 6, p > 0.05). (D) Western blot analysis of hippocampal slice lysates of PAK1 knockout mice showing no differences in total cofilin (97 3%, n ¼ 6, p > 0.05), phosphorylated cofilin (p-cofilin, 93 5%, n ¼ 6, p > 0.05), active CREB (p-creb, 108 6%, n ¼ 6, p > 0.05), active MAPK component JNK (p-jnk, 94 4%, n ¼ 5, p > 0.05), active ERK1/2 (p-erk1/2, 92 5%, n ¼ 6, p > 0.05) and active MAPK/ERK kinase (p-mek, 94 7%, n ¼ 5, p > 0.05). (E) Abolition of NMDA-induced changes in cofilin activity in PAK1 knockout hippocampal slices. The amount of p-cofilin 3 h after hippocampal slice recovery and before the NMDA (50 mm)/glycine (10 mm) treatment (0 min) was defined as 100%. Addition of NMDA/glycine for 10 min increased the amount of p-cofilin in the wild-type but not in PAK1 knockout slices. The level of cofilin was not altered by the treatment in either genotype. For quantitative analysis of protein levels in C and D, the immunoreactivities of each protein from the knockout genotype was normalized with that of actin of the same sample and expressed as the percentage of the wild-type animals. The averaged data were from 5 6 independent experiments. p < 0.05 with Student s t-test. of LIMK-1/cofilin and its absence leads to selective deficits in hippocampal LTP but not basal spine or synaptic properties. It would be important to assess whether the structural changes of the spine in response to synaptic activities are affected in these mice because these changes are thought to contribute to long-lasting synaptic plasticity (Fukazawa et al., 2003; Meng et al., 2003a; Lamprecht and LeDoux, 2004; Okamoto et al., 2004; Zito et al., 2004; Dillon and Goda, 2005; Lin et al., 2005; Chen et al., 2007). Clearly, more studies are needed to elucidate the mechanisms by which PAK1 regulates hippocampal LTP. In summary, the results presented in this study and the accompanying paper (Zhou et al., submitted for publication) provide in vivo and genetic evidence that PAK1 and ROCK2 play distinct roles in the regulation of spine and synaptic properties in the hippocampus. While ROCK2 is important for basal spine and synaptic transmission, PAK1 is more specifically involved in synaptic plasticity. We propose that these distinct synaptic functions are accomplished at least in part by the differential regulation of cofilin activity and the actin cytoskeleton, which may contribute to the complex neuronal and cognitive effects of the Rho GTPases. Acknowledgments This work was supported by grants from the Canadian Institutes of Health Research (CIHR MOP-42396) and Ontario Mental Health Foundation. We are grateful to Guijin Lu, Shouping Zhang, and Y.M. Heng for technical assistance. References Bamburg, J.R., Proteins of the ADF/cofilin family: essential regulators of actin dynamics. Annu. Rev. Cell Dev. Biol. 15, Bliss, T.V.P., Collingridge, G.L., A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361, Boda, B., Alberi, S., Nikonenko, I., Node-Langlois, R., Jourdain, P., Moosmayer, M., Parisi-Jourdain, L., Muller, D., The mental retardation protein PAK3 contributes to synapse formation and plasticity in hippocampus. J. Neurosci 24,