The results described here uncovered a new role for ERK signaling in GABA release, plasticity and learning. We also show that disruption of this mechanism accounts for the learning deficits associated with NF1. Our results demonstrate that the learning deficits in a mouse model of Neurofibromatosis type I are caused by increased hippocampal GABA-release which dampens hippocampal synaptic plasticity and consequently leads to hippocampal-dependent learning deficits. Importantly, we also showed that spatial learning triggers a lasting increase in GABA release, and demonstrate that neurofibromin modulates ERK-dependent phosphorylation of synapsin I and that this is critical for GABA release, LTP and learning. Thus, our findings suggest that the mechanisms uncovered in our NF1 studies are of general importance for learning and memory, and not just specific to the pathology of NF1.
A key finding reported here is that neurofibromin in inhibitory neurons regulates ERK dependent phosphorylation of synapsin I and consequently GABA release. First, we reported that Nf1+/−
mice, as well as mice with either an Nf1
deletion in inhibitory neurons (Nf1Dlx5/6+/−
) or in inhibitory and excitatory neurons (Nf1Syn I+/−
), showed increased frequency of mIPSCs a traditional marker of pre-synaptic effects, without changes in either mIPSC amplitude, rise time or decay constant, three common electrophysiological tags of post-synaptic changes in neurotransmission. In contrast, deletion of Nf1
in excitatory neurons (Nf1αCaMKII+/−
) or astrocytes (Nf1GFAP+/−
did not affect mIPSCs. These results indicate that neurofibromin has a critical role in GABA release. Second, two different MEK inhibitors rescued the increase in mIPSC frequency observed in Nf1+/−
mice, implicating the Ras/MEK/ERK signaling pathway in GABA release. In addition, we showed that MEK inhibition, although less effectively, also affected mIPSC frequency in controls, indicating that these effects are not specific to Nf1
mutants. However, importantly, the Nf1+/−
mutation did not affect mEPSCs, indicating that neurofibromin does not modulate ERK’s role in hippocampal glutamatergic release(Kushner et al., 2005
). Previous work had indicated that the ERK signaling pathway modulates learning by regulating post-synaptic LTP mechanisms(Sweatt, 2004
). Our results are not in contradiction with these earlier findings, and instead they suggest that other GAPs (Bernards and Settleman, 2004
), control ERK signaling in excitatory neurons during learning.
How does neurofibromin/ERK signaling modulate GABA release? Previous reports implicated ERK signaling in synapsin I phosphorylation, vesicle docking and glutamate release under conditions of high stimulation frequency. Consistent with previous models of synapsin I function in excitatory neurons (Chi et al., 2003
), our findings suggest that the higher ERK activation found in inhibitory neurons leads to higher levels of synapsin I phosphorylation and therefore to greater GABA release. We show that behavioral training known to engage the hippocampus(Frankland et al., 1998
; Kim and Fanselow, 1992
) results in more robust hippocampal ERK phosphorylation and ERK-dependent phosphorylation of synapsin I in Nf1
mice, including those with the mutation restricted to inhibitory neurons (Nf1Dlx5/6+/−
). Since the deletion in Nf1Dlx5/6+/−
is restricted to inhibitory neurons, we can attribute the increase in ERK and synapsin I phosphorylation to processes in these cells. Additionally, we showed that deletion of the Nf1
gene in excitatory neurons does not affect either ERK, synapsin I phosphorylation or various measures of glutamate release, indicating that neurofibromin does not play a critical role in the modulation of ERK/synapsin I function in excitatory pre-synaptic terminals.
Importantly, mutations (Nf1+/−
and Nf1Syn I+/−
) that affected GABA release also disrupted LTP, demonstrating that the LTP deficits previously reported for Nf1+/−
mice are caused by increased GABA release. Additionally, the very same lines with increased hippocampal GABA release and LTP deficits also showed abnormal learning in a hippocampal-dependent task (i.e. Morris water maze), while the Nf1αCaMKII+/−
mice did not show either inhibition, LTP or learning abnormalities. Abnormally high levels of GABA released during learning, could result in increased hyperpolarization of excitatory neurons and consequently in deficits in LTP. In contrast, up-regulating ERK signaling pathway in excitatory neurons leads to enhancements of both LTP and learning in mice(Kushner et al., 2005
), while down regulating it in excitatory neurons leads to LTP and learning deficits(Chen et al., 2006
). Altogether these findings demonstrate that ERK signaling in both excitatory and inhibitory neurons is critical for synaptic plasticity and learning.
The results discussed above demonstrate that increases in GABA release can affect learning. But, does learning also affect GABA release? Our results demonstrate that spatial training in the Morris water maze caused a lasting increase in GABA release, a result consistent with the idea that changes in GABAergic function may have a role in orchestrating the circuit changes involved in memory. Although these findings suggest that learning normally involves increases in GABA release, our results demonstrate that the abnormally high GABA release documented for Nf1+/− mice disrupts LTP and learning.
Our findings also impact on the development of treatments for the learning disabilities associated with NF1; we demonstrated that increased GABA release is responsible for the LTP and learning deficits associated with Nf1+/−
mice. This result is consistent with previous findings that GABAergic transmission is important in the regulation of synaptic plasticity and learning(Crestani et al., 2002
; Introini-Collison et al., 1994
; McElroy and Korol, 2005
; Wiltgen et al., 2005
). Based on our finding that Nf1
mutants have increased inhibition, we were able to reverse the learning deficits of the Nf1+/−
mice with a dose of a GABAA
antagonist (picrotoxin) that was shown not to enhance learning in WT mice. These results confirm the role of inhibition in the learning deficits associated with NF1 and suggest that safe strategies that decrease GABAmediated inhibition will be useful to treat the learning deficits associated with NF1. It is important to note that GABAergic drugs like picrotoxin have a limited clinical usefulness due to their dangerous convulsant properties. However, it may be possible to use safer GABAergic drugs that preserve the beneficial cognitive effects without the proconvulsant properties (Atack et al., 2006
). Interestingly, a recent study also found that changes in GABA inhibition underlie the learning deficits of an animal model of Down syndrome(Kleschevnikov et al., 2004
; Rueda et al., 2008
) suggesting that GABA inhibition may have a prominent role in the pathophysiology of cognitive disorders.