Fig 2: Structural similarity between the human HSD17B14 protein and bacterial L-fucose dehydrogenase. Ribbon representations of (A) human HSD17B14 (PDB: 5hs6), (B) Burkholderia multivorans L-fucose dehydrogenase (PDB: 4gvx), and (C) the superposition of both enzyme structures, highlighting a clearly similar fold architecture (a root mean square deviation of 2.43 Å). A, HSD17B14 is shown in complex with NAD (black sticks) and estrone (red sticks), whereas (B) L-fucose dehydrogenase is illustrated with bound NADP (magenta sticks) and L-fucose (cyan sticks). C, the estrone structure is omitted in the superposition for simplifying. D, close-up of the active sites of HSD17B14 and L-fucose dehydrogenase. Dinucleotide coenzymes superimpose very well, with a clear exception for the 2′-phosphate of NADP that is absent in NAD. In bacterial L-fucose dehydrogenase, the His40 residue plausibly facilitates the preferential biding of NADP over NAD, while the Asp42 residue present in HSD17B14 determines the enzyme preference for NAD. The catalytic triads consisting of Ser 140/141, Tyr153/154, and Lys 157/158 are also shown. All models were prepared using UCSF ChimeraX (46).

Fig 3: Q-TOF fragmentation spectra of L-fuconate and the product formed by human HSD17B14 protein. The homogenous recombinant human enzyme was incubated for 10 min with 2 mM L-fucose and 0.75 mM NAD+, and the progress of the reaction was followed spectrophotometrically at λ = 340 nm. The reaction mixture was then deproteinized by adding methanol and acetonitrile (1:1:1), chromatographed on an anion exchange Dionex IonPac AS11HC column, and analyzed by tandem mass spectrometry. Mass spectra, covering the mass range m/z 50 to 300, (A) of commercial L-fuconate and (B) the product biocatalyzed by human HSD17B14 enzyme were acquired. The structure of L-fuconic acid and the assignments of some of its fragment ions are also shown.

Fig 4: Test of the enzymatic activity of the purified recombinant WT (rbHSD17B14) and mutated (Y154F) rabbit HSD17B14. The enzyme activity was followed spectrophotometrically by measuring the conversion of NAD+ into NADH (λ = 340 nm). The reaction was performed with 1.1 μg of Y157F protein as described in the "Experimental procedures" section. The addition of 1.1 μg of WT rbHSD17B14 initiated L-fucose oxidation, concomitant with the reduction of NAD+ to NADH, as evidenced by the change in absorbance at 340 nm. The figure shows the results of a single representative assay. Human HSD17B14 showed similar enzymatic activity, while rat enzyme was less active. The results for human and rat enzymes as well as control reactions carried out in the absence of L-fucose are shown in Fig. S1.

Fig 5: SDS-PAGE and Western blot analysis of recombinant rabbit HSD17B14 purification. Lysate of HEK-293T cells overexpressing the recombinant enzyme (LOAD) was applied to the HisTrap Crude column, and flow-through (FT) was collected. The column was washed with a buffer without imidazole (Wash). Retained proteins were eluted by applying the buffer with indicated concentrations of imidazole. To remove imidazole, 150-mM and 300-mM fractions were subjected to dialysis. Prior to the dialysis, fractions were supplemented with BSA (1 mg/ml) to improve the stability of purified enzyme. The purification process was analyzed by SDS-PAGE (A), whereas the presence of recombinant rbHSD17B14 protein was verified by Western blot analysis (B) using an antibody against the His6 tag. Analogous results were obtained for human and rat enzymes. Note that the 65 kDa band visible on Western blot likely corresponds to a dimer of the recombinant enzyme, suggesting incomplete denaturation during sample preparation. In dialyzed enzyme preparations, this band disappears, possibly due to the masking effect of BSA. The purity of both recombinant proteins was above 95%. M, prestained protein ladder.