Endo-fucoidan hydrolases from glycoside hydrolase family 107 (GH107) display structural and mechanistic similarities to α-l-fucosidases from GH29

Abstract

Fucoidans are chemically complex and highly heterogeneous sulfated marine fucans from brown macro algae. Possessing a variety of physicochemical and biological activities, fucoidans are used as gelling and thickening agents in the food industry and have anticoagulant, antiviral, antitumor, antibacterial, and immune activities. Although fucoidan-depolymerizing enzymes have been identified, the molecular basis of their activity on these chemically complex polysaccharides remains largely uninvestigated. In this study, we focused on three glycoside hydrolase family 107 (GH107) enzymes: MfFcnA and two newly identified members, P5AFcnA and P19DFcnA, from a bacterial species of the genus Psychromonas Using carbohydrate-PAGE, we show that P5AFcnA and P19DFcnA are active on fucoidans that differ from those depolymerized by MfFcnA, revealing differential substrate specificity within the GH107 family. Using a combination of X-ray crystallography and NMR analyses, we further show that GH107 family enzymes share features of their structures and catalytic mechanisms with GH29 α-l-fucosidases. However, we found that GH107 enzymes have the distinction of utilizing a histidine side chain as the proposed acid/base catalyst in its retaining mechanism. Further interpretation of the structural data indicated that the active-site architectures within this family are highly variable, likely reflecting the specificity of GH107 enzymes for different fucoidan substructures. Together, these findings begin to illuminate the molecular details underpinning the biological processing of fucoidans.

Keywords: X-ray crystallography; enzyme; enzyme catalysis; enzyme mechanism; glycoside hydrolase; polysaccharide.

Figure 1.

Modular schematics showing the variation in currently classified GH107 enzymes and the Psychromonas enzymes P5AFcnA and P19DFcnA. A, the modular structures of full-length MfFcnA and the truncations used in this study. B, schematics of P5AFcnA and P19DFcnA. The proteins were recombinantly expressed without their predicted secretion signal peptides (shown in “yellow”). C, the modulare structures of the remaining currently classified GH107 enzymes. Fda1 and Fda2 are the entries from Alteromonas sp. SN-1009 and SVI_0379 is from Shewanella violacea DSS12. AXE80_07420, AXE80_07425, AXE80_07310, and AXE80_07305 are from Wenyingzhuangia fucanilytica CZ1127. In all panels, the domains are: yellow, signal peptide; blue, GH107 (β/α)₈ D1 domain; gray, immunoglobulin-like R domain (IPR013783); green, secretion system C-terminal sorting domain (IPR026444); purple, concanavalin A-like lectin/glucanase domain superfamily (IPR013320); light blue, archaeal/bacterial C-terminal peptidase (IPR007280); and red, conserved all β-strand domain. Module/domain assignments were based on a combination of manual assignment by BLAST and automated assignment by InterProScan (14).

Figure 2.

Fucoidanolytic substrate specificity screen of recombinant MfFcnA, P5AFcnA and P19DFcnA by C-PAGE. A, activity screen against class one fucoidan (C. filum, α(1–3)-linked fucopyranose polysaccharide backbone) and unclassified fucoidan (M. pyrifera). B, activity screen against class two fucoidan (A. nodosum, repeating α(1–3)- and α(1–4)-linked fucopyranose backbone) and unclassified fucoidan (D. antarctica). C, activity screen against a heterogeneous fucoidan that contains both class 1 and 2 backbone structures (S. japonica) and unclassified fucoidan (L. hyperborea). In A–C, the lanes include a carrageenoligosaccharide standard (lanes S), untreated fucoidan (lanes 1), and fucoidan treated with MfFcnA2 (lanes 2), P5AFcnA (lanes 3), and P19DFcnA (lanes 4). D, a complete summary of activity, including examples were no activity was seen, is given.

Figure 3.

The structure of P5FcnA shown as a cartoon representation. A bound calcium atom is shown as a green sphere, and a bound malonate molecule is shown as green sticks.

Figure 4.

Structural analysis of MfFcnA. A, identification of dynamic regions of MfFcnA2 by HDX. Hydrogen deuterium incorporation levels observed for MfFcnA2 after 3 s of deuterium exposure at 0 °C. Every point represents the central residue of a peptide, generated during immobilized pepsin digestion, on the x axis versus HDX on the y axis. The experiments were conducted in triplicate, and error bars are shown. The deuterium incorporation levels are colored according to the legend where amino acids that have low deuterium incorporation are colored blue, and amino acids that have high deuterium incorporation are colored red (see also Fig. S1). B, the MfFcnA4 structure colored by secondary structure and orientated to show the three consecutive C-terminal Ig-like domains (R1, R2, and R3) that wrap around the large N-terminal (β/α)₈-barrel catalytic domain (D1) that coordinates an ethylene glycol (green sticks). C, the solvent-accessible surface of MfFcnA4. D, the MfFcnA9 structure (yellow) superimposed over the MfFcnA4 (gray) structure to show the similar structural arrangements of the composite domains in different crystallographic conditions. In all panels, calcium atoms are shown as green spheres, a sodium atom is shown as a purple sphere, and bound ethylene glycol and malonate ion are green sticks.

Figure 5.

Comparative structural analysis of MfFcnA4 and P5AFcnA with the α-fucosidase BiAfcB from family GH29. A, the MfFcnA4 (pink) and P5AFcnA (green) structures superimposed over the BiAfcB (blue; PDB code 3UET) structure in complex with lacto-N-fucopentaose II Le^a antigen (LNFP II; Galβ1–3(Fucα1–4)GlcNAcβ1–3Galβ1–4Glc) to show structural similarities between (β/α)₈-barrel catalytic domains. B, comparison of catalytic sites between MfFcnA4 (pink sticks), P5AFcnA (green sticks), and BiAfcB (blue sticks). Because the BiAfcB–LNFP II complex was generated with a D172A mutant, we overlapped the apo-BiAfcB structure (PDB code 3MO4) with the BiAfcB–LNFP II complex, and it is the unmutated active site of BiAfcB that is shown with the ligand from the complexed structure. In A and B, the substrate bound to BiAfcB is shown as yellow sticks. C, C-PAGE analysis of the activity of MfFcnA4_H294Q on fucoidan from C. filum and A. nodosum.

Figure 6.

Electrostatic surface representations of the MfFcnA4 (A) and P5AFcnA active sites (B). Using the structural overlay of MfFcnA4 and P5AFcnA with the BiAfcB–LNFP II complex (see Fig. 4), α-l-fucopyranosyl-2-sulfate was overlapped with the fucose in the −1 subsite of the BiAfcB–LNFP II complex. The α-l-fucopyranosyl-2-sulfate is shown in the context of the solvent-accessible surface colored by electrostatic potential. The putative sulfate-group binding site is circled in red. O3 and O4 are labeled for reference and would represent the points from which the intact polysaccharide may extend. The dashed black line represents the approximate direction of the active-site groove.

Figure 7.

The catalytic mechanism of fucoidan hydrolysis by MfFcnA2. A, capturing the first formed anomer with 2-mercaptoethanol. Top panel, free disaccharide and disaccharide acetal and their ¹H spectra. Isolated disaccharide acetal has a J_1,2 = 3.8 Hz, indicating α configuration and thus retention of stereo chemistry at the cleavage site. Bottom panel, NOESY and HMBC spectra of isolated disaccharide acetal showing the linkage of 2-mercaptoethyl group to C-1 of the disaccharide. B, a schematic of the proposed catalytic mechanism of MfFcnA.

Figure 8.

Amino acid conservation in the GH107 family catalytic domains. A, an amino alignment of relevant portions of the (β/α)₈ D1 domain of all 10 GH107 members illustrating the conserved putative aspartate nucleophile (indicated by a green arrow), the conserved putative histidine acid/base (indicated by yellow arrow), and the conserved tryptophan residue in the −1 subsite. The numbering above the alignment represents that of the MfFcnA4 structure. B, the conservation of amino acids of the full D1 domain from A mapped onto the solvent-accessible surface of the D1 domain structure from MfFcnA4 calculated by ConSurf (38). Amino acids colored turquoise are considered variable through to amino acids colored maroon, which are considered highly conserved. The inset focuses on the putative −1 subsite with the highly conserved residues shown as sticks and colored as in A. The bound ethylene glycol molecule is shown as green sticks.