Metal Ion Promiscuity and Structure of 2,3‐Dihydroxybenzoic Acid Decarboxylase of Aspergillus oryzae

Abstract Broad substrate tolerance and excellent regioselectivity, as well as independence from sensitive cofactors have established benzoic acid decarboxylases from microbial sources as efficient biocatalysts. Robustness under process conditions makes them particularly attractive for preparative‐scale applications. The divalent metal‐dependent enzymes are capable of catalyzing the reversible non‐oxidative (de)carboxylation of a variety of electron‐rich (hetero)aromatic substrates analogously to the chemical Kolbe‐Schmitt reaction. Elemental mass spectrometry supported by crystal structure elucidation and quantum chemical calculations verified the presence of a catalytically relevant Mg2+ complexed in the active site of 2,3‐dihydroxybenoic acid decarboxylase from Aspergillus oryzae (2,3‐DHBD_Ao). This unique example with respect to the nature of the metal is in contrast to mechanistically related decarboxylases, which generally have Zn2+ or Mn2+ as the catalytically active metal.

Difference map (Fobs-Fcalc) drawn at 5 r.m.s.d shows the excess of model electrons for Mn 2+ and Zn 2+ (red/green). Both models were refined by PDB redo. [1] The B factors in chain A at full occupancy refined to 42.3 for Zn 2+ and to 28.7 for Mg 2+ compared to the envelope B factor of 27.6 and 25.9, respectively.

Expression and purification
The synthetic gene for 2,3-DHBD_Ao (accession number XP_001817513) was cloned into pET29a(+) using NdeI/Xho1. E. coli BL21-DE3 cultures transfected with this plasmid were grown to an OD600 of 0.6 in LB medium at 37 °C before being induced with 1 mM IPTG and left to shake at 180 rpm over night at 20 °C. Pellets harvested by centrifugation were lysed by sonication and purified by Ni NTA purification as described previously. [ 2 ] Pooled and concentrated fractions of the Ni purification were polished by preparative SEC on a Superdex 200 Increase 10/300 column (Cytiva) with 150 mM NaCl 10 mM MES pH 6.5 as the mobile S9 Figure S4. Michaelis-Menten fit for 2,3-DHBD_Ao sample treated with Mn 2+ . "S" refers to substrate concentration in mM and "v" to the measured decarboxylation velocity in s -1 . S10 Figure S5. Michaelis-Menten fit for non-treated 2,3-DHBD_Ao. "S" refers to substrate concentration in mM and "v" to the measured decarboxylation velocity in s -1 .   [11] SAD_Tm Zn 2+ salicylic acid 1080 0.34 0.3 [12] 2,3-DHBD_Fo Zn 2+ 2,3-dhba 8190 27 3.3 [13] 2,3-DHBD_Ao Zn 2+ 2,3-dhba 420 3 [a] 7.1 [14] S12 Phe296, Glu297). In addition, four crystallographic water molecules were also included in the model. Amino acids were truncated as shown in the figures, and hydrogen atoms were added manually. In the geometry optimizations, a number of atoms were fixed at their crystallographic positions to keep the structure of the active site pocket. These are indicated in green color in the figures below. The model consists of 306 atoms and has a total charge of 0.

Geometries of transition states and intermediates of the Mg-dependent enzyme
Optimized structures for the transition states (TS1Mg and TS2Mg) and intermediates (IntMg and E:PMg) in the reaction pathway of the Mg-enzyme of 2,3-DHBD_Ao. The energies relative to the E:SMg are given in the parentheses (kcal/mol).

Geometries of transition states and intermediates of the Mn-dependent enzyme
Optimized structures for the enzyme-substrate complex (E:SMn), transition states (TS1Mn and TS2Mn) and intermediates (IntMn and E:PMn) in the reaction pathway of the Mn-enzyme of 2,3-DHBD_Ao. The energies relative to the E:SMn are given in the parentheses (kcal/mol).

Alternative monodentate binding mode of the substrate
We examined in the present study the binding mode of the substrate in 2,3-DHBD_Ao. In a previous work on the Mn-containing 2,6-DHBD_Ps ( -RSD_Ps), two binding modes were considered for the γ-resorcylate substrate, one of which is similar to the bidentate binding of 2,3-dihydroxybenzonate to the metal in 2,3-DHBD_Ao as shown in Figure 3 in the paper. The other one is an unproductive monodentate binding mode with only the carboxylate group being coordinated to the metal. In the case of 2,6-DHBD_Ps, these two binding modes were found to be comparable in energy, only 0.4 kcal/mol in favor of the monodentate mode. As 2,3-DHBD_Ao and 2,6-DHBD_Ps belong to the same family, we considered here also the monodentate binding mode for 2,3-DHBD_Ao. In this case, the monodentate mode is less favored, with calculated energies of 2.8 kcal/mol and 1.8 kcal/mol higher than the bidentate one for Mg-and Mn-enzymes, respectively (see Figures S16 and S17). S18 Figure S16. Optimized structure of enzyme-substrate complex with a monodentate binding mode of the 2,3-dihydroxylbenzonate substrate for the Mg-enzyme of 2,3-DHBD_Ao. The energy relative to E:SMg is given in kcal/mol. Figure S17. Optimized structure of enzyme-substrate complex with a monodentate binding mode of the 2,3-dihydroxylbenzonate substrate for the Mn-enzyme of 2,3-DHBD_Ao. The energy relative to E:SMn is given in kcal/mol.

Cartesian coordinates
The transition states and intermediates involved in the reaction pathway of the Mgenzyme of 2,3-DHBD_Ao