This paper describes the detailed structure of a membrane-embedded aspartyl protease called Flak, a preflagellin peptidase (PFP) from an extremophile that has been categorized as a "GXGD aspartyl protease," along with presenilin and SPP. As this is the first structure of a membrane-embedded aspartyl protease, it is quite important, and Ha and his coauthors deserve considerable credit, as working with such proteases and elucidating their structure is highly challenging. The FlaK structure should be useful for testing specific hypotheses about how this and (truly) related enzymes work, just as the structure of the E. coli Rhomboid GlpG has allowed testing of specific ideas for how transmembrane substrates gain access to the interior Ser-His active site dyad.

The structure shows two aspartates close, but not close enough to be catalytically active. The conditions used for crystallization (e.g., pH 9.5) are apparently capturing an inactive conformation, and so it is unclear exactly what the active conformation looks like. One can only guess about the changes needed to...  Read more

This paper describes the detailed structure of a membrane-embedded aspartyl protease called Flak, a preflagellin peptidase (PFP) from an extremophile that has been categorized as a "GXGD aspartyl protease," along with presenilin and SPP. As this is the first structure of a membrane-embedded aspartyl protease, it is quite important, and Ha and his coauthors deserve considerable credit, as working with such proteases and elucidating their structure is highly challenging. The FlaK structure should be useful for testing specific hypotheses about how this and (truly) related enzymes work, just as the structure of the E. coli Rhomboid GlpG has allowed testing of specific ideas for how transmembrane substrates gain access to the interior Ser-His active site dyad.

The structure shows two aspartates close, but not close enough to be catalytically active. The conditions used for crystallization (e.g., pH 9.5) are apparently capturing an inactive conformation, and so it is unclear exactly what the active conformation looks like. One can only guess about the changes needed to assume an active conformation. The structure does, however, make it clear that this multi-pass membrane aspartyl protease is fundamentally different from soluble or membrane-tethered aspartyl proteases such as pepsin, renin, BACE1, and HIV protease. This is a clear example of mechanistic convergence, in which nature has found the same solution to the problem of proteolysis for membrane proteins as for soluble aspartyl proteases (i.e., using two active site aspartates to activate water).

That being said, the parallels drawn between this PFP and presenilin are highly speculative, and the FlaK structure cannot be used for homology modeling of presenilin. The sequences of the two proteins are not similar (i.e., they are not evolutionarily related), and the transmembrane domains with the active site aspartates are different (TMs 1 and 4 for FlaK and TMs 6 and 7 for presenilin). Moreover, FlaK cleaves outside the substrate transmembrane domain, and, consistent with this, the active site aspartates are at the interface or just outside the membrane. In contrast, presenilin cleaves within the substrate transmembrane domain, and the active site aspartates apparently reside within the confines of the lipid bilayer. We will need a structure of an actual presenilin or presenilin-like protease (e.g., SPP) to better understand how presenilin works and to make possible structure-based inhibitor/modulator design.

View all comments by Michael Wolfe