Ferulic acid decarboxylase

Figure 1. Proposed mechanism for AnFDC.

In the same paper in which the structure of prFMN was deduced in the active site of AnFdc1 there was a proposal for the mechanism by which Fdc1 decarboxylates α,β-unsaturated carboxylic acids.[5] Not all UbiD enzymes decarboxylate acrylic acid substrates and other mechanisms may be at play for alternative substrates.[9] In the case of AnFdc1 it was noted that prFMN displays an azomethine ylide characteristic C4a-N5+=C1’(Figure 1). This is a well-known 1,3-dipole in organic chemistry, positioned in the enzyme active site near to the α,β-unsaturated carboxylic acid substrate which contains a 1,3-dipolarophile. Thus, it was proposed that a 1,3-dipolar cycloaddition mechanism was responsible for the enzymatic decarboxylation. This was confirmed in a later paper.[8]

The mechanism proposed in [5] for 1,3-dipolar cycloaddition by Fdc1 is as follows (intermediates represented in Figure 1):

1. 1,3-dipolar cycloaddition between prFMN^iminium and the α,β-unsaturated substrate leads to a pyrrolidine cycloadduct (Int1)
2. This pyrrolidine cycloadduct supports simultaneous decarboxylation and ring opening, resulting in the formation of a distinct prFMN-alkene adduct (Int2)
3. A conserved glutamic acid residue (E282) donates a proton to the alkene moiety, resulting a second pyrrolidine cycloadduct (Int3)
4. The reaction concludes with cycloelimination of Int3 and the release of the alkene product and CO₂

A study went on to present evidence for the 1,3-dipolar cycloaddition,[8] due to suspected turnover of cinnamic acid a crystal structure of AnFdc1 in complex with α-fluorocinnamic acid revealed the substrate Cα and Cβ carbons are located directly above the prFMN^iminium C1’ and C4a respectively (shown as Sub in Figure 1 - with cinnamic acid as opposed to α-fluorocinnamic acid). Cinnamic acid was confirmed to bind in a similar manner using inactive AnFdc1 crystals containing FMN. The AnFdc1 E282Q mutant crystallised with cinnamic acid revealed a structure corresponding to the Int2 species, this was taken to mean that progression through the 1,3-dipolarcycloadition cycle was halted as E282 is unable to donate a proton to the alkene moiety.

In order to observe the Int1 and Int3 structures alkyne analogues were used. Like alkenes these compounds can also act as dipolarophiles but cycloaddition would yield a cycloadduct containing a double bond. An inactive AnFdc1 enzyme (with prFMN^radical bound) co-crystallised with the phenylpropiolic acid revealed binding in a similar manner to the α-fluorocinnamic acid AnFdc1 and cinnamic acid AnFdc1 with FMN bound (Inhib). An active AnFdc1 enzyme co-crystallised with phenylpropiolic acid revealed clear density for a 3-pyrroline cycloadduct (Int3’) between the alkyne and prFMN^iminium. Int3’ represents a structure post decarboxylation, so it was assumed that over the time it took for crystallisation (~24h) the decarboxylation had occurred. Using a rapid soaking procedure, a different cycloadduct was observed that retained the carboxyl moiety (Int1’).