A recent paper in Neuron provides a note of caution when interpreting gene knockout studies, a hint of the impact RNAi technology has on a field of study, and some new tidbits about AMPA receptor trafficking. Previous studies has shown that overexpressing a protein called PSD-95 (post-synaptic density - 95 kDa) led to an increase in synaptic AMPA receptors. AMPA receptors are responsible for a major portion of the excitatory neurotransmission in the brain. They normally respond to pre-synaptic glutamate release by opening a pore through their center and allowing sodium into the cell, thereby depolarizing the post-synaptic neuron and increasing the probability that it will fire an action potential. Increases in synaptic AMPA receptor content are likely to explain the increased synaptic response observed in long-term potentiation (LTP). Since PSD-95 seems capable of modulating this factor it has attracted much interest.

PSD-95 is part of a larger family of proteins called PSD-MAGUKs that also includes PSD-93, SAP102, and SAP97. MAGUKs don't seem to have a catalytic activity. Rather, they are likely to serve as scaffolding proteins, connecting one protein to another through multiple binding sites up and down the MAGUK sequence. All PSD-MAGUKs share similar binding domains but vary in the details. Part of the impetus for the present paper is that a transgenic mouse with targeted disruption of the portion of PSD-95 that ought to interact with AMPA receptors shows no signs of difficulty in synaptic transmission. Now, thanks to these authors, we have a mouse that lacks PSD-95 altogether and still no deficit in AMPA receptor delivery.

Long story short, PSD-93 is picking up the slack when PSD-95 is knocked out. It's compensation. Redundancy and degeneracy are the rule rather than the exception in biological systems. This set of experiments makes it so clear that a knockout mouse is never what you expected you were making. There is no deficit in a PSD-93 knockout either. Only in the double knockout (mice lacking PSD-95 and PSD-93) do you get a significant decrease in AMPA receptor transmission, and in that case SAP-102 still pops up and does its damnedest to keep the synapses running on time.

The way all of this was discovered was through the use of more acute gene knockdown using short hairpin RNAs expressed via a viral vector. You can create a virus that will code for something that looks tasty to the RNA interference machinery like an shRNA and infect a slice of brain tissue with it. RNAi machinery will grab the shRNA, chop out a 22 base sequence from it, and go around translationally repressing or destroying any RNA that matches that sequence. So you can get gene specific knockdown quick enough in the adult preparation and not allow the neuron time to compensate by upregulating a redundant gene. As opposed to knocking out PSD-95 from day 1, shRNA targeted against PSD-95 drastically decreased excitatory synaptic transmission. The same goes for PSD-93. Knocking down SAP-102 with shRNAs in the PSD-93/PSD-95 double knockout almost completely eliminated any sign of synaptic AMPA receptors. One of the most interesting control experiments in the study was to use shRNA against a gene that was already knocked out (i.e. PSD-95-shRNA in the PSD-95-KO). This invariably produced no effect, showing that the shRNA manipulation was specific.

When SAP-102 isn't rescuing PSD-93/95 double knockout mice from oblivion, it has a day job. It is normally expressed earlier in development (peaking at 10 days post-natal). There is a developmental switch in which MAGUKS are the prominent AMPA receptor traffickers, so after SAP-102 for a couple weeks, PSD-93/95 come on and take over synapse scaffolding duty. In normal mice, shRNA against SAP-102 only has an effect in the early developmental timepoint.

Also, in the normal mouse, PSD-95 shRNA doesn't completely block AMPA receptor currents and neither does PSD-93 knockdown. This suggests that normal mice have a heterogeneous population of synapses, some of which utilize 93 and others that use 95. It would be interesting to know if there is any functional or localization difference between a 93 and a 95 synapse. While one of the major messages of this paper is that these proteins can compensate for each other seamlessly at least in the assays performed thus far, it seems unlikely that they will perform the exact same function in their natural setting.