Usually, quantal size and release probability were determined from the amplitude and frequency miniature EPSCs(mEPSC), I wonder why does NMDAR-mediated EPSCs by MK-801 can be used to examine basal release probability?And what is the mechanism?
It is very difficult to measure release probability. I would not consider mini frequency a reliable indicator. You can assess release probability indirectly using the paired-pulse ratio of AMPA receptor EPSCs, but it depends on the characteristics of the receptors a little (do they desensitize, do they saturate), and it might be nice to know whether an effect is presynaptic.
Using the NMDAR EPSC is an approach. The basic idea is that MK-801 is a use-dependent blocker. It gets stuck in the pore when the channel is activated, is my understanding. The logic is that when release probability is high, lots of glutamate is released in response to a presynaptic action potential, lots of NMDA receptors activate, and a high proportion get blocked by MK-801, so the next time you look at the NMDAR EPSC amplitude, it will be smaller. When release probability is low, less glutamate is released, not that many NMDA receptors activate, and only a small proportion get blocked, so the rundown of the NMDAR EPSC will be slower.
What you compare then is the rate of rundown of the NMDAR EPSC. Faster rundown = higher release probability. Then you don't have to worry about baseline differences in synapse size, or saturation. However, it will not report to you an actual value of release probability, just a relative number for comparing synapses.
The mechanism of MK-801 is well described by Dr. Xu-Friedman above.
I will just add a point.
As said above, MK-801 can not report the actual probability. An important point of the method is that it is "pathway-specifc". It can be helpful to know the change of release probability of synaptic pathway you are stimulating.
A neuron recieves many synaptic inputs, and the change in miniature EPSC parameters reflects an "overall" change of synapses which a neuron recieves.
Therefore, you may not detect a significant change in mEPSC even if your manipulation actually changes a population of synapses.
I think the first answer explains it all. The faster NMDAR responses decrease, more would be the release probability as more NMDARs would have been conducting and thereafter blocked by MK-801.. Higher NMDAR activation implies more presynaptic glutamate release and therefore higher release probability.