Oxygenase-Catalyzed Desymmetrization of N,N-Dialkyl-piperidine-4-carboxylic Acids**

γ-Butyrobetaine hydroxylase (BBOX) is a 2-oxoglutarate dependent oxygenase that catalyzes the final hydroxylation step in the biosynthesis of carnitine. BBOX was shown to catalyze the oxidative desymmetrization of achiral N,N-dialkyl piperidine-4-carboxylates to give products with two or three stereogenic centers.


Initial MS screening
. The set of cyclic GBB analogues used in initial substrate screening with BBOX. Only (1) was observed to be hydroxylated. The results imply that BBOX requires a quaternary ammonium centre for reactivity, since neither piperidine-4-carboxylic acid (5) nor N-methyl-piperidine-4carboxylic acid (6) are substrates. Analogues with additional modifications in the piperidine ring (e.g. unsaturated analogue (7), or methylated analogue (8)) were not substrates. BBOX has an apparent requirement for a C-4 carboxylate, since (13) with a C-3 carboxylate was not hydroxylated. The 5membered ring analogues were not hydroxylated.  (1). Spectra were recorded in mass range 120-300 Da (scan mode), single ion mode used in parallel to monitor masses: 158 (substrate, retention time = 9.68 min) and 174 (product, retention time = 8.02 min).

Conformational assignement of N,N-dimethyl isonipecotic acid
The 1 H NMR spectrum of (1) in D 2 O at room temperature displayed only one distinctive set of peaks, likely corresponding to a single conformer (Fig. S5). The position of H4, adjacent to the carboxylic group, was assigned as axial, based on its coupling constant analysis (J aa = 11.2 Hz, J ae = 4.6 Hz). The overall signal pattern was characteristic for a chair conformation (The slight phase distortions seen for the multiplets arises from the excitation sculpting water suppression scheme employed). The identity of signals was further confirmed by assignment of 13 C NMR shifts for (1) using 1 H-13 C HSQC spectra ( Fig. S6). Figure S5. Conformation of N,N-dimethyl isonipecotic acid (1) as assigned by 1

S8
The methyl group position was assigned by nOe studies (Fig. S8). The H4 proton (i.e. adjacent to the carboxylate) was assigned to be axial in both isomers of (3) by the coupling constants pattern in the 1 H NMR spectra (isomer A: J aa = 10.2 Hz, J ae = 5.1 Hz; isomer B: J aa = 11.5 Hz, J ae = 4.8 Hz).

Assignments of product of BBOX catalysed hydroxylation of N,N-ethylmethylpiperidine-4-carboxylic acid (3)
Mixtures of isomers (3a) and (3b) (mixture 1: (3a):(3b) = 2:1, mixture 2: (3a):(3b) = 1:3) were subjected to BBOX catalysed oxidation. MS based assay indicated the formation of new species with a mass shift of +16, corresponding to a single hydroxylation (Fig. S9). In case of both Mixture 1 and 2 a formation of new peaks in the 1 H NMR spectra was observed. In both cases, the presence of two new peaks, characteristic for axial protons adjacent to hydroxyl and carboxylate moiety was observed (4.17 and 2.32 ppm, respectively) ( Fig. S10 and S11). In case of mixture 1, where isomer (3a) was the major component only one set of new peaks was observed, indicating that (3a) is the preferred substrate. Interestingly, for mixture 2, where isomer (3b) was the major component, the presence of second set of product peaks was detected; the major product was the same as that observed with Mixture 1 (Fig. S10 and S11). The result suggest that (3b) is hydroxylated by BBOX, but is a less preferred substrate than (3a) (even though (3b) was in excess in Mixture 2, the major product arises from hydroxylation of the minor component (3a)). Due to the low level of (3b) hydroxylation and signal overlap the stereochemistry of minor product has not been assigned with confidence. Figure S10. Overlay of the 4.10-4.25 ppm region of the 1 H NMR spectra of cyclic analogue reaction mixtures. (a) H 3 peak of BBOX catalysed oxidation of (1). H 3 was assigned to be in axial position (J aa = 10.8 Hz × 2, J ae = 4.7 Hz). (b) When a mixture of (3a):(3b) = 2:1 was subjected to BBOX, H 3 was also assigned to be axial (J aa = 10.4 Hz × 2, J ae = 4.4 Hz). Only one hydroxylation product was observed. (c) When a mixture of (3a):(3b) = 1:3 was subjected to BBOX, two peaks corresponding to H 3 were observed, both assigned as axial protons (major: J aa = 10.5 Hz × 2, J ae = 4.4 Hz, minor: J aa = 11.1 Hz × 2, J ae = 5.4 Hz). The major product is same as that product obtained for the mixture in (b) and corresponds to the hydroxylation product of (3a) to give (4a). The minor product likely corresponds to hydroxylation of (3b) to give (4b), which was the major component of substrate mixture used in (c). (d) Reaction mixture containing substrate mixture as in (b), before addition of BBOX. (e) Reaction mixture containing substrate mixture as in (c) before addition of BBOX. Figure S11. Overlay of the 2.25-2.32 ppm region of 1 H NMR spectra of cyclic analogues reaction mixtures. (a) H 4 peak of BBOX catalysed oxidation of (1). H 4 was assigned to be in axial position (J aa = 12.4 Hz, J aa = 10.5 Hz, J ae = 4.7 Hz). (b) When a mixture of (3a):(3b) = 2:1 was subjected to BBOX oxidation, H 4 of the product was assigned as axial (J aa = 12.2 Hz, J aa = 10.2 Hz, J ae = 4.8 Hz). Only one hydroxylation product was observed. (c) When a mixture of (3a):(3b) = 1:3 subjected to BBOX a mixture of signals corresponding to H 4 was observed, the major signal assigned as corresponding to an axial proton (J aa = 12.4 Hz, J aa = 10.2 Hz, J ae = 4.8 Hz), identical that signal obtained for mixture as in (b). The J values of the H 4 signal of the minor product could not be assigned due to signal overlap. Starred peaks likely correspond to the minor product. (d) Reaction mixture containing substrate mixture as in (b), before addition of BBOX. (e) Reaction mixture containing substrate mixture as in (c), before addition of BBOX. The multiplets observed are partially masked by other signals in the spectra.   [2] . **yield is given as a percentage conversion of substrate into hydroxylated product (as calculated by 1 H NMR) after 20 min of the reaction.

Enzyme production
Recombinant human BBOX was produced and purified as described [1] .

Initial MS screening
Initial screens for BBOX substrates were performed on Waters LCT Premier Instrument, employing Electron impact Chemical Ionisation, fitted with time of flight (ToF) analyser. Samples were measured using direct injection (no column attached) and analysed for the presence of a +16 peak.
Enzymatic assays were run in the following conditions: 100 µM substrate,

LC-MS method
Chromatographic separation of the GBB analogues was performed using mixed mode chromatography. Chromatographic separation was performed using an Aquity UPLC system (Waters

Hydroxylation product assignment
Hydroxylation reactions were performed using the following conditions: Crystals were harvested in nylon loops and flash cooled in liquid nitrogen using 25% glycerol in mother liquor as a cryoprotectant. Data were collected on a single crystal in-house using a Rigaku FRE+ SuperBright X-ray diffractometer equipped with Osmic HF optics and a Saturn944+ CCD detector. Data were integrated and scaled using HKL3000 [3] . The structure was solved by molecular replacement using PHASER [4] (search model PDB ID: 3O2G). Iterative cycles of model building in COOT [5] and refinement using PHENIX [6] proceeded until the converging R cryst and R free no longer decreased.

General procedure for the synthesis of cyclic GBB analogues
The appropriate amino acid (1 equiv.) was dissolved in methanol (5 mL