Effects of memantine on the excitation-inhibition balance in prefrontal cortex

Highlights

• Novel actions of the Alzheimer's Disease (AD) drug memantine were investigated.

• Memantine disinhibits cortex by shifting excitation/inhibition balance.

• Memantine reduces NMDAR responses in PV + interneurons more than in pyramidal neurons.

• Memantine reduces spiking in PV + interneurons, increases spiking in pyramidal neurons.

• Disinhibition may be an effective mechanism of action for new AD therapeutics.

Abstract

Memantine is one of the few drugs currently approved for treatment of Alzheimer's disease (AD). The clinical effects of memantine are thought to be associated with inhibition of NMDA receptors (NMDARs). Surprisingly, other open-channel NMDAR blockers have unacceptable side effects that prevent their consideration for AD treatment. One of the mechanisms proposed to explain the therapeutic benefits of memantine involves preferential decrease of excitatory drive to inhibitory neurons in the cortical circuitry and consequent changes in balance between excitation and inhibition (E/I). In this study we addressed effects of memantine on E/I balance in the prefrontal cortex (PFC). We found that a moderate concentration of memantine shifted E/I balance away from inhibition in the PFC circuitry. Indeed, memantine decreased the frequency and amplitude of spontaneous inhibitory postsynaptic currents in pyramidal neurons while leaving spontaneous excitatory postsynaptic currents unaffected. These circuitry effects of memantine were occluded by the competitive NMDAR inhibitor AP-5, and thus are associated with NMDAR inhibition. We also found that memantine decreased feed-forward disynaptic inhibitory input to pyramidal neurons, which is thought to be mediated by parvalbumin (PV)-positive interneurons. Accordingly, memantine caused a greater decrease of the amplitude of NMDAR-mediated synaptic responses in PV-positive interneurons than in pyramidal neurons. Finally, memantine reduced firing activity in PV-positive interneurons while increasing firing in pyramidal neurons. This study elucidates a novel mechanism of action of memantine associated with shifting of the E/I balance away from inhibition in neocortical circuitry, and provides important insights for AD drug development.

Introduction

Alzheimer's disease (AD) is a devastating brain disorder that heavily burdens the aging American population. Although a daunting challenge, creating new AD therapeutics can be aided by better understanding of the mechanisms of action of existing AD drugs.

Memantine has been used to treat AD and other dementias for more than two decades. Although memantine binds to a number of receptors, it is generally believed that the therapeutic effects of memantine are mainly associated with N-methyl-d-aspartate receptor (NMDAR) channel block (Lipton, 2006, Parsons et al., 2007, Johnson et al., 2015). There is no general agreement on how NMDAR channel block by memantine slows cognitive decline in AD patients. Several mechanisms of memantine action have been proposed, the majority of which are based on memantine's neuroprotective action. The neuroprotective effect of memantine could result from prevention of excitotoxicity due to excessive NMDAR excitation in pathological brain conditions (Lipton, 2006), or from preferential inhibition of extrasynaptic NMDARs (Leveille et al., 2008, Xia et al., 2010) based on the hypothesis that extrasynaptic NMDAR activation can lead to cell death (Hardingham and Bading, 2010) (but see (Wroge et al., 2012)). However, there is evidence arguing against therapeutic effects of memantine based solely on neuroprotection, e.g. the lack of beneficial effects in early-stage AD, and the rapidity of memantine's effects (Johnson and Kotermanski, 2006).

The ability of memantine to slow cognitive decline in AD patients has also been proposed to result from partial correction of an AD-induced alteration of cortical excitation/inhibition (E/I) balance (Schmitt, 2005). A delicate balance of excitatory and inhibitory elements in the cortex is essential to circuit function, and disturbances of this balance can lead to pathological conditions (Homayoun and Moghaddam, 2007, Haider and McCormick, 2009). There is strong evidence that the E/I balance is shifted away from excitation in AD. A decrease in cortical activity (Rombouts et al., 2000) and PFC hypometabolism in the prefrontal cortex (PFC) (Schroeter et al., 2012) were reported in AD patients. Data from AD postmortem brains indicate that, whereas excitatory pyramidal neurons in the PFC are prone to neurodegeneration (Hof and Morrison, 2004), PFC inhibitory neurons that express the Ca^2 + binding protein parvalbumin (PV) are spared (Hof et al., 1991). In transgenic AD model mice, despite substantial neuronal loss in the PFC, no changes in PV-positive and calretinin-positive interneurons were detected (Lemmens et al., 2011). There is substantial loss of cortical spines, the principal site of excitatory synaptic input onto pyramidal neurons, in both the neocortex and hippocampus of AD patients (Cochran et al., 2014). Note, however, that some data do not support decreased excitation in AD. For instance, in the parietal cortex and hippocampus of mice expressing human amyloid precursor protein, nonconvulsive seizure activity resulting from an aberrant increase in network excitability was detected (Palop et al., 2007), and increased excitability of hippocampal CA1 pyramidal neurons was detected in APP/PS1 AD model mice (Siskova et al., 2014).

In this study we explore the hypothesis that memantine can shift the E/I balance in cortical circuitry away from inhibition (Johnson and Kotermanski, 2006) by preferentially reducing NMDAR-mediated excitation of inhibitory neurons. This hypothesis is based on two observations: (1) in physiological Mg^2 +, memantine at therapeutic concentrations preferentially inhibits NMDARs that contain the GluN2C or GluN2D subunits (Kotermanski and Johnson, 2009); (2) in adult cortex and hippocampus, the GluN2D subunit is preferentially expressed in inhibitory neurons (Monyer et al., 1994, Standaert et al., 1996). If this hypothesis is correct, then memantine at a concentration that preferentially inhibits GluN2C and GluN2D-containing NMDARs should be more effective at reducing the action potential frequency of inhibitory neurons than of excitatory neurons. To test this prediction we explored in mouse PFC the effects of 10 μM memantine on inhibitory and excitatory synaptic inputs to pyramidal neurons, on NMDAR-mediated synaptic inputs to PV-positive interneurons, and on synaptically-activated action potentials in pyramidal neurons and PV-positive interneurons.