Most intriguingly however there was no change in basal gluta

Most intriguingly, however, there was no change in basal glutamatergic neurotransmission despite a loss of about half of the neuronal AMPAR population. Highest levels of ammonia did neither affect the amplitude nor the frequency of AMPAR mediated mEPSCs. As revealed by our recordings from somatic outside-out patches, the severe loss of neuronal AMPARs induced by high ammonia translated into their selective reduction within the extrasynaptic plasma membrane domain. Thus, neurons suffering a loss of half of their AMPARs due to a rather unspecific disease stimulus maintained synaptic transmission at the expense of extrasynaptic receptors. Current models of postsynaptic receptor localization assume lateral surface traffic between synaptic and extrasynaptic membranes as an important link between internal recycling and receptor trapping at the postsynaptic density (PSD) via interaction with scaffold proteins (Choquet and Triller, 2013, Opazo et al., 2012). Applied to our data, we expect receptor diffusion rates into or out of the synapse to change significantly in response to the dramatic loss of AMPARs in order to maintain basal neurotransmission. Given a loss of 60% AMPARs, their synaptic dwell time should increase by 2.5-fold to ensure stable mEPSC amplitudes (Czöndör et al., 2012), which reflect postsynaptic receptor numbers. How could synaptic dwell time of AMPARs increase? Most likely, receptor trapping at the PSD may be enhanced. Auxiliary TARP proteins are thought to link GluA subunits to the scaffold protein PSD-95, a process that can be bi-directionally regulated by posttranslational modification of the TARP C-terminal tail (Sumioka et al., 2010, Tomita et al., 2005). Whereas phosphorylation of a cluster of serine residues in TARPs mobilizes its C-terminus for interaction with PSD-95 by interfering with an otherwise inhibiting TARP–membrane phospholipid interaction (Sumioka et al., 2010), phosphorylation of their very C-terminal PDZ-ligand motif has been reported to disrupt PSD-95 binding (Chetkovich et al., 2002, Stein and Chetkovich, 2010). Indeed, significant changes in the CAY10603 phospho-proteome have been described in animal models of HE (Brunelli et al., 2012). Changes in the molecular composition of AMPARs towards a higher TARP/GluA ratio, which could also improve receptor trapping at the PSD by increasing multi-valence of TARP–PSD-95 interactions (Sainlos et al., 2011), are rather unlikely, as we found the predominant hippocampal TARPs γ-8 and γ-2 to be reduced to similar extent as the main pore-lining GluA subunits. Moreover, mEPSC kinetics were accelerated after ammonia treatment exhibiting shorter decay time constants, which argues against an increase in the TARP/GluA ratio that would be expected to slow receptor gating (Milstein and Nicoll, 2008). The faster mEPSC decay times in ammonia treated neurons might be due to a different stoichiometry of GluA subunits with a bias for the fast gating GluA3 and GluA4 subunits; in contrast to GluA1 and GluA2, GluA3 and GluA4 expression remained grossly unaffected in our model. Finally, we cannot exclude ammonia-induced morphological changes in synapse geometry also known to be a determinant of synaptic strength and hence maybe at least supportive in maintaining mEPSC amplitudes (Freche et al., 2011). Despite the ability of neurons to maintain basal neurotransmission under high ammonia conditions, synaptic plasticity was severely impaired. Whereas a brief activation of NMDARs by their co-agonist glycine elicited an increase in mEPSC amplitude in control neurons, this form of cLTP was completely abolished by prior ammonia treatment. Similar to AMPAR expression, also NMDAR expression might have been significantly affected in ammonia treated neurons explaining the lack of plasticity. However, application of the selective agonist NMDA readily induced cLTD in ammonia treated neurons indicating functional NMDAR expression. Such selective impairment of plasticity with synaptic potentiation being abolished and depression being preserved is highly reminiscent of the electrophysiological GluA1 knock out phenotype (Granger and Nicoll, 2014, Granger et al., 2013, Selcher et al., 2012, Zamanillo et al., 1999). The common characteristic of the genetic deletion of GluA1 and our disease model is the severe reduction in the size of the extrasynaptic population of AMPARs, whereas the size of the synaptic population quantified by AMPAR mediated EPSC amplitudes remains stable. In line with the previous interpretations of GluA1 deletion (Granger and Nicoll, 2014, Granger et al., 2013), we therefore conclude from our data that ammonia treatment constrains synaptic plasticity by reducing the number of extrasynaptic AMPARs. Our study represents the first pathophysiological setting strongly supporting the hypothesis that a sufficiently large reserve pool of extrasynaptic receptors is a prerequisite for LTP induction.