OK, as much as I believe in theory, experimental confirmation would be nice. This is hard to come by, since the formation of the C=O is a substantial driving force for collapse of the tetrahedral intermediate.

• Hence, the intermediate does not normally accumulate during the reaction

However, it is possible to trick the enzymes, and get them to make a tetrahedral intermediate that can't break down.

Meyer [J. Am. Chem. Soc., 1989, 111, 3368; Biochemistry, 1989, 28, 7610] has done this in two ways, one more successful than the other.

The first trick uses a fluoroketone substrate:

Note that the carbonyl the arrows point to is a ketone, not an amide, carbonyl.

• The fluorines enhance its reactivity toward nucleophiles, and it is positioned in the middle of a peptide chain.

• The compound turns out to be an irreversible inhibitor of elastase. The crystal structure of the inhibitor-enzyme complex is shown below:

Subtilisin with Covalent Inhibitor (4est)

• The fluorines (fuschia) also stabilize the hemiacetal

The other trick was to provide a boron, in the form of Cbz-Ala-Ile-B(OH), which is sp², and thus resembles a carbonyl carbon. In this case, Meyer got more than he bargained for; the His57 also binds to the boron:

Subtilisin with Boric Acid Inhibitor (5est)

• The boron is the magenta atom in the center

Where next? Here's the tetrahedral intermediate, bound to Ser195 with the former carbonyl oxygen in the "oxyanion hole".

The next step is the reconstruction of the carbonyl double bond, with expulsion of the leaving group - in this case, the rest of the protein.

This is the stage at which the protein chain actually is cleaved, and it produces an "acyl enzyme": the acyl part of the peptide that was cleaved, bound as an ester to Ser195.