Given the crucial role of NMDA receptor synaptic activation in neuronal plasticity, simplistic interpretation of the observation that a NMDA receptor antagonist such as memantine can improve cognition and neuronal plasticity under pathological conditions might seem to be paradoxical. It should, however, be stressed that divalent cation Mg2+ is an endogenous NMDA receptor channel blocker and its absence leads both to an impairment in neuronal plasticity [49, 74] and neuronal death [81]. Any dysfunction of postsynaptic neurons leading to weakened blockade by Mg2+, e.g. due to partial depolarization as a consequence of an energy deficit, may also trigger prolonged Ca2+ influx and structural (neuronal loss) deficits [54, 230]. Because memantine is more potent and slightly less voltage-dependent than Mg2+ due to its simple monovalent charge, it may thus serve as a more effective surrogate for Mg2+ [198]. As a result of its somewhat less pronounced functional voltage-dependency, memantine is more effective than Mg2+ in blocking tonic pathological activation of NMDA receptors at moderately depolarized membrane potentials. However, following strong synaptic activation, memantine, like Mg2+, can leave the NMDA receptor channel with voltage-dependent, fast unblocking kinetics. In turn, memantine not only suppresses tonic pathological synaptic “noise” but also allows the relevant physiological synaptic signal to be detected. This provides both neuroprotection and symptomatic restoration of synaptic plasticity by one and the same mechanism [54, 196]. Antagonists that have “too high” affinity for the channel or “too little” functional voltage-dependence, such as dizocilpine ((+)MK-801), thus produce numerous side effects, since they essentially act as an irreversible plug of the NMDA receptor channel and block both pathological and physiological function.
So from what I understand from this, Memantine is a noncompetitive blocker of the NMDA receptor that only blocks the NMDA receptor in a noncompetitive manner during hyperpolarization, and leaves the receptor about as quickly as Magnesium, our endogenous NMDA blocker, would?
In that case, it does not seem that there would be any reason for NMDA receptor upregulation to occur. GIven the information, it seems upregulation would only occur during competitive NMDA blockage, especially that which is pathological.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2645549/