Long-term potentiation (LTP) is a long lasting activity-dependent increase in the strength of synaptic transmission that can persist for many hours (Bliss and Lomo, 1973; Bliss and Collingridge, 1993; Malenka and Nicoll, 1999). This process is considered as a possible physiological substrate for long-term memory (Martin et al., 2000). At Schaffer collateral-CA1 synapses of hippocampus, LTP is NMDA receptor-dependent (Bliss and Collingridge, 1993; Malenka and Nicoll, 1999; Malenka and Bear, 2004). We investigated whether along with changes in short-term plasticity, there is also an alteration in long-term plasticity at Schaffer collateral-CA1 syanpses in α-neurexin KO mice. Our experiments with induction of LTP by high frequency stimulation demonstrated that LTP is enhanced in neurexin 1α/2α double KO as compared to control mice. This result was surprising in the first instance because of two reasons. Firstly, the impaired basal release in α-neurexin double KO would be expected to cause reduced release of glutamate during the trains, thereby causing reduced induction of LTP at the postsynaptic apparatus. Secondly, it was shown in an earlier study that
in neocortical slice cultures, the NMDA/AMPA ratio at the excitatory synapses was reduced in α-neurexin triple KOs as compared to controls (Kattenstroth et al., 2004). The first possibility does not take into account the release properties during the whole train of stimuli.
We showed that in α-neurexin double KOs, there is enhanced facilitation during the stimulus train. So the overall release during the train in α-neurexin double KOs might be similar to or even higher than in control mice. This has been shown for RIM1α KO mice in which the postsynaptic Ca2+ signal in the CA1 dendritic spine in response to trains of stimuli was similar to that in control mice (Huang et al., 2005), even though the basal release probability was reduced (Schoch et al., 2002). Regarding NMDA/AMPA ratio, two factors have to be kept in mind. 1) The investigations in neocortical slice cultures for NMDA/AMPA ratio were done on α-neurexin triple KOs. The effect on NMDA/AMPA ratio is expected to be lesser in α -neurexin double KO mice which were used in the present study. 2) The induction protocol for LTP that we used was strong (4 x 100 Hz, 1s each) and could have overridden a small impairment of NMDA receptors. The NMDA/AMPA ratio will be considered in more detail below.
Our experiments on LTP in the presence of DL-APV showed that the enhanced LTP in neurexin 1α/2α double KO is NMDA receptor-dependent. The Schaffer collateral-CA1 synapses show NMDA receptor-dependent LTP. However at some synapses LTP is NMDA receptor-independent e.g. at hippocampal mossy fibre-CA3 synapses (Harris and Cotman, 1986; Johnston et al., 1992; Nicoll and Malenka, 1995), cerebellar parallel fibre synapses (Salin et al., 1996; Linden and Ahn, 1999), and corticothalamic synapses (Castro-Alamancos and Calcagnotto, 1999). The experiments with DL-APV confirmed that the enhanced LTP in neurexin 1α/2α double KO was not because of the appearance of an NMDA receptor-independent component.
The primary cause of enhanced LTP in neurexin 1α/2α double KO is found to be a reduction in the inhibitory synaptic transmission. In the presence of GABAA receptor antagonist bicuculline, we found that the LTP in neurexin 1α/2α double KOs was similar to that in control mice. The activity of inhibitory synapses is known to keep neuronal excitability under control (Alger, 1991). Any change in inhibitory transmission can have profound effects on the synaptic plasticity at excitatory synapses. The blockade of GABAergic synapses using GABAA receptor antagonists enhances the induction of LTP in the CA1 region (Wigstrom
matrix molecule tenascin-R reduced perisomatic inhibition and enhanced LTP (Saghatelyan et al., 2000). A recent report showed that repeated exposure to cocaine enhanced the induction of LTP in rat midbrain slices through a reduction in GABA-mediated inhibition of neurons (Liu et al., 2005). Conversly, an enhanced GABAergic inhibition has been shown to underlie reduced LTP in Ts65Dn mice (transgenic model of Downs syndrome) (Kleschevnikov et al., 2004) and in transgenic mice with overexpression of extracellular matrix heparin-binding growth-associated molecule (Pavlov et al., 2006). Similarly antagonists of acetylcholine (Ye et al., 2001) or opioid receptors µ and δ-1 (Bramham and Sarvey, 1996) suppress the induction of LTP by enhancing GABAergic transmission. All these observations imply that the level of inhibitory activity can have profound impact on LTP. The facilitated induction of LTP is probably related to the higher levels of dendritic depolarization attained in the presence of inhibitory blockade, leading to facilitated activation of NMDA receptors.
α-neurexin KO mice have been shown to have impaired inhibitory GABAergic synaptic transmission in neocortex and brain stem (Missler et al., 2003). In addition, the number of inhibitory synapses are reduced in the neocortex of α-neurexin KOs. A direct assessment of GABAergic activity in acute hippocampal slices has not been done in the present study, partly because the extracellular recordings are not suitable for this analysis. However all three α -neurexins are expressed in CA1 interneurons (Ullrich et al., 1995), and their deletion is therefore expected to reduce release at GABAergic synapses onto CA1 pyramidal cells, similar to that in neocortex (Missler et al., 2003). The reduced inhibitory synaptic transmission in α-neurexin double KO mice would facilitate the NMDA receptors by dendritic depolarization, leading to enhanced induction of LTP at excitatory Schaffer collateral-CA1 synapses. Pharmacological blockade of the GABAergic activity in control animals provides the same mechanism, thereby bringing the LTP in control animals to the similar level of neurexin 1α/2α double KO mice.
Our LTP experiments also suggest a mechanism behind the reduced NMDA/AMPA ratio in α-neurexin triple KOs as measured by whole-cell patch-clamp of neocortical pyramidal cells (Kattenstroth et al., 2004). Enhanced facilitation of NMDA receptors due to reduced inhibitory tone is potentially harmful for the cells, as NMDA receptors are permeable to Ca2+, and Ca2+ influx through NMDA receptors is believed to be neurotoxic (Choi et al., 1995).
Therefore, homeostatic mechanisms exist to regulate the NMDA receptor activity (Perez-Otano and Ehlers, 2005). The enhanced activity of NMDA receptors observed under reduced
inhibitory transmission could therefore cause a compensatory reduction in NMDA receptor number/activity at the synapses. The support for this hypothesis comes from a study in which immunotoxin-mediated cell targeting was done to eliminate inhibitory Golgi cells in the cerebellum of transgenic mice. Golgi cell ablation led to a reduced NMDA/AMPA ratio at the granule cells (which recieve inhibitory input from Golgi cells) after chronic depletion of GABA neurotransmitter (Watanabe et al., 1998). In addition, pharmacological blockade of GABAergic activity for a week decreased synaptic targeting of NMDA receptors in hippocampal cultures (Crump et al., 2001). This indicates that the NMDA receptor number/activity is regulated in the long-term by the level of inhibitory tone in the network.
The reduced NMDA/AMPA ratio in α-neurexin triple KO neocortical slices could therefore reflect a compensatory decrease in the number or activity of NMDA receptors, measured independent of the inhibitory tone by voltage-clamp.