Are human cytochromes P450 crystal structures "wrong"?

A recent paper by Denisov et al. "Structural differences between soluble and membrane bound cytochrome P450s", have shown that there may be discrepancies between the crystal structures of human cytochromes P450 enzymes and reality. In this post I will discuss what they actually have shown, and it's possible implication on computational work done on membrane-bound cytochrome P450s.

First, they have found that there are some structural differences with regard to the heme propionate side chains of CYPs that are membrane bound vs. CYPs that are not membrane bound. This is highly interesting as all crystal structures of human CYPs are made of solubilized variants (while in fact, they are membrane bound in vivo).

So the two major question then becomes, do these structural differences affect the binding site? and are all the crystal structures of human CYPs bad representations of reality?

To study this, they took the 1TQN structure of CYP3A4 and run a 50 ns molecular dynamics simulation with the protein bound to a membrane. Interestingly, they find that the access channels to the active site changes compared to a MD simulation in water. After 40 ns one of the channels have closed, and remain closed for the remaining 10ns. Of course, this is only a 10ns time frame, but the result is still intriguing. Unfortunately, they did not analyze the properties of the binding site. So we don't know yet if this has any effects on the shape and properties of the binding site.

The harsh reality for studies of access channels for human CYPs is that all made without a membrane might be wrong (or maybe not...), I guess the future will tell.

There is currently labs researching human P450s with longer MD simulations in membranes (up to 1 mikrosecond), so it will be interesting to see the results and if there actually are any effects on access channels or the active site during a longer period of time.