Dissecting a two-domain alginate lyase of family PL6 reveals a mechanistic basis for substrate specificity and enzyme activity

TL;DR

This study reveals how two parts of an enzyme work together to break down a specific type of sugar polymer in the gut.

Contribution

The study identifies the functional roles of the N- and C-terminal domains in a two-domain PL6 alginate lyase.

Findings

The N-terminal domain contains the catalytic site and cleaves M-G bonds in endo-mode.

The C-terminal domain binds polyguluronate and enhances activity on polyG when combined with the N-terminal domain.

A CTD mutant disrupts the WT structure and reduces activity on polyG but improves polyMG processing.

Abstract

Alginate lyases (ALs) cleave 4-O-glycosidic linkages in alginate, composed of mannuronate (M) and guluronate (G) residues via β-elimination with preference for either one or several M-M, M-G, G-M, G-G linkages. ALs in polysaccharide lyase family 6 (PL6) present different specificities and modes of action and contain either one or two parallel β-helix domains. About half of almost 600 PL6 sequences are of the two-domain type, all located in the phyla Pseudomonadota and Bacteroidota. Here, functional roles are described for the N- and C-terminal domains (NTD and CTD) using BoPL6, a two-domain AL from the human gut bacterium Bacteroides ovatus CP926, which is specific for G in subsite +1. The NTD contains the catalytic site, but BoPL6-NTD markedly preferred the model substrate polyMG and cleaved M-G bonds in endo-mode, whereas the NTD + CTD mixture, similarly to BoPL6, acted with highest activity on model substrate polyG in exo-mode, verified by time-resolved 1H-NMR. The CTD was not catalytically active but bound polyguluronate and, when mixed with BoPL6-NTD, promoted activity on polyG, yielding products of DP 1‒3, similarly to BoPL6. This defines a crucial role of the CTD in shaping the active site in BoPL6, as validated by substrate docking. The BoPL6 E634A mutant in the conserved CTD DEST loop, interacting with the active site in two-domain PL6 enzymes, was inactive, while the corresponding CTD mutant mixed with the NTD did not form the WT structure and had highly reduced activity on polyG but acted on polyMG in endo-mode with improved rate and conversion.

Figures

Full text

Alginate is a linear anionic polysaccharide constituting up to 55% of the cell wall in brown seaweed (1, 2, 3). It consists of 1,4-linked β-d-mannuronate (M) and its C-5 epimer α-l-guluronate (G) organized in M-, G-, and alternating MG-blocks (Fig. 1A). The biosynthesis involves initial formation of polymeric M (polyM), which is converted to the final alginate by the action of epimerases at the polymer level (4) where the resulting M/G ratio depends on factors such as environment and seaweed species (5). Tailor-made alginate having defined composition and organization of M and G residues can be prepared by epimerase-engineered Azotobacter (5) or Pseudomonas (6) strains. The request for specific alginates arises from the different characteristics displayed by M and G, as exemplified by the strong gelling properties and calcium ion-mediated cross-linking of G (7) and M’s enhancement of protein binding (8). Seaweed also contains other valuable polysaccharides, such as carrageenans, ulvan, starch, agarose, laminarin, and fucoidans (9, 10, 11). The latter is found in brown algae, where alginate, hampering their extraction and purity (12, 13, 14), can be removed through depolymerization catalyzed by alginate lyases (ALs) (15, 16, 17).Figure 1Alginate structures and reaction mechanism of alginate lyases. A, schematics showing 1,4-linked guluronate (orange, G-block), mannuronate (green, M-block), and alternating mannuronate‒guluronate (MG-block) residues. B, β-elimination reaction of alginate lyases. An amnio acid or a salt cation (purple sphere) stabilizes the uronate carboxylate group, the proton is then abstracted from the C-5 by the catalytic base (blue sphere) causing electron displacement with formation of a double bond between C-4 and C-5 and a proton is donated by the acid catalyst (red sphere) to the glycosidic oxygen, here leading to an unsaturated monosaccharide (4-deoxy-l-erythro-4-hexenopyranuronate) and a new non-reducing end (blue Δ).

ALs are categorized into 16 polysaccharide lyase (PL) families in the Carbohydrate Active enZyme database (CAZy; www.cazy.org) (18). Although highly diverse with regard to structural fold (19, 20), the ALs share catalytic mechanism and depolymerize alginate via a β-elimination reaction in which 4-O-glycosidic bonds are proposed to be cleaved in three steps (Fig. 1B); (i) neutralization of the C-5 carboxylate at subsite +1, (ii) abstraction of the C-5 proton by a catalytic base, and (iii) formation of a double bond between C-4 and C-5 at the non-reducing end of the product. An acid catalyst donates a proton to the C-1 oxygen of the new reducing end (21, 22). To fully describe the substrate preference of PL6 ALs, substrate cleavage, e.g., G↓G, is related to binding at subsites −1 and +1, with the β-elimination occurring at the +1 subsite (Fig. 1B).

No AL has been reported to produce unsaturated monosaccharides by exo-mode of attack on all four M-M, G-M, M-G and G-G linkages. For complete alginate depolymerization, microbial enzymes evolved targeting the linkages in alginate with complementary preference and mode of action to work in unison (23, 24, 25). Evolutionary pressure is not on the fitness of the individual enzymes, but rather on their synergy (26, 27). In the bacterial phylum Bacteroidota, ALs are encoded within gene clusters called polysaccharide utilization loci (PULs) (20, 28, 29, 30). Enzymes of polysaccharide lyase family 6 (PL6) are predicted in PULs of many Bacteroides, represented by 266 hits in the database PULDB of CAZy (18, 31). So far, different ALs of PL6 have been shown to cleave either M-G (32, 33) or G-G (23) linkages; notably, those acting on M-G often have dual preference with minor activity on G-G (32, 34). Further, PL6 ALs can release mono- and oligosaccharides (AOSs) (32), with product profiles depending on an open cleft, dead-end cleft, or pocket architecture of the active site (20, 34, 35).

PL6s are either single- or double-parallel β-helix fold domain proteins. In two-domain PL6s, the N-terminal domain (NTD) contains the catalytic site and shares structural similarity with the one-domain enzymes (34). Truncation of the C-terminal domain (CTD) in AlyGC from the marine bacterium Glaciecola chathamensis and BcAlyPL6 from the human gut bacterium Bacteroides clarus resulted in 6% and 1.5% activity towards polyG and polyMG, respectively, of the corresponding parent enzymes (36, 37). Moreover, the homologous AlyGC D641A and BcAlyPL6 D617A mutants on the CTD situated in spatial proximity to the active site on the NTD retained 5.7% and no activity on polyG, respectively, while BcAlyPL6 D617A showed 0.3% activity on polyMG (36, 37).

Here, NTD (BoPL6-NTD) and CTD (BoPL6-CTD) are produced from the two-domain PL6 from Bacteroides ovatus CP926 (BoPL6) preferring polyG as substrate and generating products of DP 1 to 3 from polyguluronate (CS-PG) (23). BoPL6 is encoded in a PUL together with the exo-acting M-M specific BoPL17, that releases 4-deoxy-l-erythro-4-hexenopyranuronate (Δ), a 4,5-unsaturated monosaccharide, from polyM and a unique multispecific endo-acting BoPL38 equally efficiently degrading polyM, polyG, and polyMG to oligosaccharides of DP 5 and 6 (23). Markedly, BoPL6-NTD preferred polyMG, had trace activity on polyG, and produced longer AOSs than BoPL6. BoPL6-CTD had no catalytic power of its own, but when mixed with BoPL6-NTD, the exo-action on polyG was regained, while activity on polyMG of the NTD was retained. The NTD and CTD clearly interacted, and docking of ΔGG and the (MG)3 to BoPL6, respectively, BoPL6-NTD indicated differences in substrate orientation at the catalytic site, suggesting a change of substrate preference in the subsites in the presence and absence of the CTD.

Results

Phylogenetics and structural analysis in family PL6

Three crystal structures are published of two-domain PL6s, BcAlyPL6 from B. clarus YIT 12056 (PDB: 7DMK), AlyGC from Paraglaciecola chathamensis S18K6T (PDB: 5GKQ), and Patl_3640 from Paraglaciecola sp. T6c (PDB: 7O7T), with C_α_ root mean square deviations below 1.7 Å (Table S1). These enzymes have less than 50% sequence identity but share conserved residues in the active sites and important loops of the CTD (Fig. 2A, Table S2). Further, there was a high structural similarity of the NTDs of the two-domain with the one-domain PL6 enzymes. RMSD was 0.6 Å for the structural alignment of Pedsa0632 from Pseudopedobacter saltans (34) (PDB: 7O7A) to a two-domain PL6, BcAlyPL6 NTD (37) (PDB: 7DMK) (Fig. S1), which made us pose the question if characteristic differences exist between the two groups of the PL6 family. Sequence alignment of NTDs of two-domain and one-domain separated these PL6 types into two clades in the phylogenetic tree (Fig. S2). The majority of PL6 sequences reported in CAZy belong to Proteobacteria and Bacteroidetes (>75%), where the remaining originate from Actinomycetota, Bacillota, Planctomycetota, Verrucomicroba, and Ignavibacteriae. Remarkably, the two-domain PL6s all belong exclusively to the two major phylae Pseudomonadota and Bacteroidota and when aligned based on CTD sequences cluster by phyla (Fig. S3). Although the CTD was required for retaining the activity level of AlyGC from Glaciecola chathamensis and BcAlyPL6 from B. clarus (36, 37), the distinct molecular and functional roles of the CTDs remained elusive.Figure 2Two domain PL6 similarities. A, sequence alignment of BcAlyPL6, AlyGC, Patl3640, and BoPL6 showing conserved residues in light green, active site residues in red, conserved CTD loop residues in cyan, and the linker regions in olive. B, aligned structures of the AlphaFold2 model of BoPL6 (yellow), BcAlyPL6 (PDB: 7DMK, green), AlyGC (PDB: 5GKD, salmon), Patl3640 (PDB: 7O77, light blue), and Unknown PL6 (8BWK, light orange). A zoom of the active site residues Asn201 Arg207, Lys240, Arg261, and His262 is shown in the red stippled oval on the left. Loop residues are shown in the orange stapled box on the left. All numbers refer to BoPL6 numbering.

Identification and production of BoPL6-NTD and BoPL6-CTD

The NTD and CTD of BoPL6 are connected by a 39-residue linker. In the AlphaFold2 model (38), the first 23 linker residues (407–430) interacted with the backside relative to the active site of the NTD, while the succeeding residues (431–446) interacted with neither NTD nor CTD. On this basis, the two domains were produced recombinantly and separated in the linker between His434 and Pro435 (38). BoPL6, BoPL6-NTD and BoPL6-CTD were obtained at 17, 50 and 75 mg per L E. coli culture and migrated corresponding to molecular masses of 83 kDa, 48 kDa and 36 kDa respectively in SDS-PAGE, which matched the theoretical values of 83,032, 48,133 and 36,308 Da (39) (Fig. S4). From the prep grade size exclusion chromatography (SEC) column, the domains elute as monomers, unlike the full-length BoPL6 that elutes as a dimer (Fig. S5).

Enzymatic action and stability of BoPL6-NTD and BoPL6-CTD

BoPL6-NTD degraded polyG with 0.14% of the initial rate of BoPL6 (Fig. 3, A and B) under optimal conditions (150 mM NaCl and pH 7.75; Fig. S6). Notably, BoPL6-NTD showed 12-fold higher initial rate on polyMG and did not act on polyM (Fig. 3C). Only the polyG degradation by BoPL6 was accompanied by loss of absorbance at 235 nm (Fig. 3A), indicative of formation of Δ (25, 40, 41). BoPL6-CTD was not active on any of these three model substrates (Fig. 3D), confirming that catalytic residues responsible for the β-elimination mechanism reside on the NTD. The domains have the highest T_m_ in a broad range from pH seven to eight, which from 0 mM to 1 mM NaCl increased by 3 to 4 °C from 44 °C for BoPL6-NTD and 49 °C for BoPL6-CTD (Figs. S7 and S8). Addition of 2 mM Mg^2+^, Ca^2+^, or EDTA did not affect T_m,_ of the domains; BoPL6-NTD but not BoPL6-CTD precipitated in the presence of 2 mM Mn^2+^ and Co^2+^ (Fig. S9). Under assay conditions, BoPL6 has a melting temperature (T_m_) of 45 °C and comparable values of 52 and 48 °C for BoPL6-NTD and BoPL6-CTD, respectively, suggesting that both domains retained their conformational integrity.Figure 3Activity of BoPL6, BoPL6-NTD and BoPL6-CTD on alginate model substrates. A, progress of degradation by 50 nM BoPL6 (black), 1 μM BoPL6-NTD (orange), and 1 μM BoPL6-CTD (blue) of 1 mg/ml polyG (≈DP10) in 20 mM HEPES, 150 mM NaCl, 2 mM Ca^2+^ pH 7.75 at 37 °C monitored at 235 nm. Negative control is 1 mg/ml polyG in buffer. B, initial rates (ΔUV_235 nm_/min) are determined for the first 4 min and normalized setting 100% for BoPL6. C, 1 μM BoPL6-NTD reacting on 1 mg/ml polyG (black), polyM (orange), and polyMG (blue). D, 1 μM BoPL6-CTD mixed with model substrates as in (C). Negative control: 1 mg/ml of model substrates in buffer.

Recovery of function in BoPL6-NTD and BoPL6-CTD mixtures

The NTDs and CTDs in two-domain PL6s share a large interface, as determined for the BoPL6 AlphaFold2 model (Fig. 2B–1151 Å^2^ by a PISA analysis (42)). We hypothesized that the two domains interact spontaneously upon mixing and reconstitute structural and functional features of the full-length BoPL6. Therefore, substrate preference and mode of action was evaluated by combining the two domains (Fig. 4). Remarkably, a mixture of 1 μM BoPL6-NTD and 4 μM BoPL6-CTD (Fig. 4B) showed essentially the same high initial rate as BoPL6-NTD alone (Fig. 3C) for degradation of polyMG and produced progress curves similar to BoPL6 degrading polyG (Fig. 4A). The clear release of more Δ and oligosaccharides with Δ non-reducing ends from polyG (Fig. 4B) demonstrated higher catalytic efficiency of the domain mixture than NTD alone. LC-ESI-MS analysis of the progress of formation of AOSs showed products of DP 2 to 6 (mostly DP 2, 5, and 6) from CS-PG by BoPL6-NTD (Fig. S10A), whereas the domain mixture, similar to BoPL6 (23), generated shorter products dominated by monomers and trimers (Fig. S10B). Where, the BoPL6-CTD modulated substrate accommodation and catalysis at the active site in BoPL6-NTD, reminiscent of the action of full-length BoPL6 (23) as illustrated in Figure 4C. Convincingly, the action on polyG and polyMG by 1 μM BoPL6-NTD varied with the addition of 0‒12 μM BoPL6-CTD (Fig. S11). Thus, the initial rate of degradation by an equimolar domain mixture was slightly higher on polyG (0.04 AU/min) than on polyMG (0.03 AU/min) and reached a maximum at 8-fold molar excess of BoPL6-CTD, showing >5 times higher initial rate on polyG than polyMG, moving towards the strict preference of BoPL6 for polyG. Notably, adding BoPL6-CTD in up to 12-fold molar excess did not change the initial rate of polyMG degradation by the domain mixture (Fig. S11C).Figure 4Substrate preference and illustration of mode of action of the BoPL6-NTD + BoPL6-CTD mixture. A, cleavage by 0.5 μM BoPL6 of 1 mg/ml polyG, polyM, and polyMG. B, cleavage by 1 μM BoPL6-NTD + 4 μM BoPL6-CTD of 1 mg/ml polyG, polyM, and polyMG. All reactions were in 20 mM HEPES, 150 mM NaCl, 2 mM Ca^2+^, pH 8.0, at 37 °C. C, illustration of the mode of action on polyG for BoPL6, BoPL6-NTD, and the mixture of BoPL6-NTD + BoPL6-CTD.

BoPL6-NTD followed Michaelis-Menten kinetics on polyMG, reacting 50 times more efficiently (kcat/Km) than on polyG and with equal catalytic efficiency as BoPL6 on polyG (Table 1; Fig. S12). Data at polyG > 1 mg/ml does not fit Michaelis-Menten kinetics due to gel formation (Fig. S14, fitted red curves), as seen in previous studies (23). Truncation of the CTD reduced kcat, Km and kcat/Km by 320-, 6- and 50-fold on polyG, compared to BoPL6. For an equimolar NTD + CTD mixture, kcat and catalytic efficiency on polyG were about 9 times higher (Table 1) compared to BoPL6-NTD alone, while Km was unaffected. But neither kcat nor the catalytic efficiency for the domain mixture reached the level of BoPL6, whereas Km was more favorable and reduced by 7-fold.Table 1. Kinetic parameters of BoPL6-NTD and BoPL6 determined using model substratesEnzymeSubstratekcat [s^‒1^]Km [mg/ml]kcat/Km [mL s^‒1^ mg^‒1^]BoPL6PolyG41.92 ± 5.881.02 ± 0.2541.10PolyMGn.r.n.r.n.r.BoPL6-NTDPolyG0.13 ± 0.010.16 ± 0.050.81PolyMG2.06 ± 0.030.05 ± 0.0141.22BoPL6-NTD + BoPL6-CTDaPolyG1.08 ± 0.050.15 ± 0.027.20Michaelis-Menten curves are shown in Figure S14. n.r. = no reaction. All values originate from the black fitted curve (Fig. S14).aMolar ratio was 1:1 for this sample.

Function of the conserved BoPL6-CTD loop in alginate depolymerization

The conserved loop (residues DEST) from the CTD protrudes into the active site [28,30] (Figs. 2A and S13), and the AEST alanine mutants of both AlyGC and BcAlgPL6 were concluded by Xu et al. (36) and Wang et al. (37) to abolish activity on a substrate similar to CS-PG, whereas the DEAT mutant in AlyGC only decreased the activity (36). To assess loop and substrate interactions, the ΔGG trisaccharide was docked into the active site of BoPL6 (Fig. S13E), which revealed an interaction between Glu634 (DEST) and the unsaturated non-reducing terminal residue in ΔGG. In contrast to previous studies on BcAlgPL6 by Wang et al. (37) we found no interaction with Asp633 (DEST) (Fig. S13E). The BoPL6 E634 A mutant was inactive on polyG and polyMG (Fig. S13A,C), and mixing BoPL6-NTD with the corresponding BoPL6-CTD mutant (here referred to as E201A, CTD-numbering) did not recover BoPL6 activity and mode of action on polyG (Fig. S13B). Notably, the combination of BoPL6-CTD E201A with BoPL6-NTD led to higher final absorbance at 235 nm by polyMG degradation and increased initial rate compared to BoPL6-NTD, both alone and when mixed with BoPL6-CTD (Fig. S13D). This may stem from a difference in the formation of product lengths, as shorter products will result in higher final absorbance.

Molecular interaction of BoPL6-CTD with BoPL6-NTD and polyguluronate

The interaction between BoPL6-NTD and BoPL6-CTD was further investigated in the absence of substrate by isothermal titration calorimetry (ITC). The dissociation constant was Kd = 6.9 ± 1.6 μM (0.2 ± 0.05 mg/ml), with 0.95 ± 0.04 binding sites for BoPL6-CTD per BoPL6-NTD molecule at pH 8.0, 150 mM NaCl and 25 °C (Fig. 5A). ITC was also performed to identify the effect of the CTD E201A mutation on the NTD CTD interaction. In stark contrast to the interaction measured for the WT domains, no interaction was measured by ITC (Fig. 5A), suggesting that E201 (BoPL6-WT E634) serves an essential role in the NTD CTD complex formation. NTD was mixed with either CTD or CTD E201A in a 1:2 ratio (40:80 μM) and analyzed by SEC, which separated molecules depending on hydrodynamic diameter (Fig. 5, B and C). The resulting chromatograms, supported by SDS-PAGE, confirmed that the mixture of NTD with CTD WT formed complexes of dimer (AB) and tetramer (A2B2) size, whereas the mixture of NTD with CTD E201A eluted as monomers and some NTD dimer (A2) (Fig. 5D). Further investigation into the properties of BoPL6 E634A revealed an increase in polydispersity with the width of the peak on SEC being 2.48 ml vs 0.73 ml for the WT. This suggested that even in the full-length enzyme where NTD and CTD were covalently linked, the E634A mutation imposes a change in tertiary/quaternary structure (Fig. S14A). By peak convolution, 66% of BoPL6 E634A elutes like the WT, while 34% elutes at a higher hydrodynamic diameter (Fig. S14B). Crystal structures of two-domain PL6 enzymes and the AlphaFold2 model of BoPL6 revealed that E634 is involved in charge-charge interaction with arginines 305 (R305) and 353 (R353) located on the NTD (Fig. S15). Though R305 was conserved in the two single-domain PL6 crystal structures (PDBid: 7O79 and 7O7A) of P. saltans DSM 1214, it was not conserved in the single-domain PL6 crystal structure (6QPS) of Bacteroides cellulosilyticus DSM 14838. Neither R353 nor N350 were conserved in the structures of single-domain PL6s, reflecting the difference in evolution between one- and two-domain PL6 enzymes.Figure 5Mutual binding of BoPL6-NTD and BoPL6-CTD and substrate binding to BoPL6-CTD. A, ITC raw data of 80 μM BoPL6-NTD titrated with 800 μM BoPL6-CTD (top, left) or BoPL6-CTD E201 A (top, right) in 20 mM HEPES, 150 mM NaCl, pH 8.0, 25 °C; enthalpogram of the data with one-binding site model (bottom). B, chromatogram of a preincubated NTD (40 μM) and CTD (80 μM) mixture. C, Chromatogram of a preincubated NTD (40 μM) and CTD E201 A (80 μM) mixture. All SEC was performed at 4 °C with a flow of 0.5 ml/min, 1 ml fractions were collected and symbolized as the colors under the curves. D, SDS-PAGE of fractions from chromatograms A and B, with matching color coding. Numbers represent the median molecular weight of each fraction. E, illustration of how NTD (blue) and CTD (green) can organize into quaternary structures of different sizes. F, intrinsic tryptophan fluorescence of 4 μM BoPL6-CTD titrated with CS-PG (stock 50 mg/ml) in 20 mM HEPES, 150 mM NaCl, pH 8.0, at 25 °C using excitation at 295 nm and monitoring emission at 310‒370 nm (left). Kd,app calculated from a non-linear curve fit to intrinsic fluorescence intensity at 317 nm versus CS-PG (mg/ml) (right).

The presence of two surface exposed tryptophan’s in the CTD, of which Trp692 is 5.3 Å from Glu634 involved with the active site in BoPL6, allowed for observing substrate binding via intrinsic fluorescence (Fig. 5F). Thus, addition of CS-PG to BoPL6-CTD under the same conditions as above, decreased the intrinsic fluorescence and caused a red shift of the peak at 317 to 345 nm, reflecting a change in the chemical environment of tryptophans with an apparent binding constant (Kd,app) of 3.94 ± 0.12 mg/ml (Figs. 5F and S16). CTD was 60 Å long, which translates to a binding surface length of 15 interlinked uronic acid residues, corresponding to a molecular weight of 2912.1 g/mol. On this basis, the resulting Kd,app in molar units was 1.4 ± 0.1 mM. Furthermore, NMR 1D ^1^H T1ρ relaxation filtered experiments (43) corroborated interactions between G residues of polyG and BoPL6-CTD (Fig. S17). All signals in polyG showed a substantial >90% decrease in signal intensity between the reference spectrum and 400 ms spin-lock (Fig. S17A). Although BoPL6-CTD also interacted with polyMG, this binding was less pronounced, and the G signals were more affected, showing a 60% to 80% decrease versus a 30% to 60% decrease for M residues (Fig. S17B). The results indicate that CTD binds guluronate and interacts more prominently with G-blocks than MG-blocks.

Molecular basis of substrate bond preference and mode of action

A small amount of the polyG (DP 20–22) substrate used in this project has 1 to 2 consecutive M-residues at the non-reducing end. This stems from the production of G-blocks by acid hydrolysis of alginate, in which the rates are highest for M-M and G-M linkages, slower for M-G linkages, and slowest for G-G linkages (44, 45, 46, 47). Thus, the M-residues remaining in the G-blocks are most likely located at the non-reducing end.

The degradation of polyG (DP 20–22) by BoPL6, monitored by time-resolved NMR, revealed formation of Δ monomer (Fig. S18, A and C). While there was activity detected on polyG, the substrate was only partially degraded, with the signal for Δ accounting for at most ≈ 2.5% (relative to internal -GG- bonds in starting material) and the internal -GG- H-1 signal was only slightly decreased (Fig. S18D). This indicated that BoPL6 did not cleave G-G linkages. Rather, an initial increase in M reducing ends (M_α_; Fig. S18, B and C) indicated that M residues present at the non-reducing end of the substrate are first cleaved with formation of ΔM- and/or ΔG-. Formation of M reducing ends and unsaturated residues at non-reducing ends of AOSs (ΔG- and ΔM-; Fig. S18, A and C) indicated cleavage from the non-reducing end of the polymeric substrate, as no Δ-containing oligomers would be formed upon cleavage from the reducing end. Notably, BoPL6 cleaved Δ-G bonds, releasing Δ monomers, as indicated by decreasing ΔG- (Fig. S18C). We hypothesized that this activity was the primary source of the Δ monomer.

When acting on polyMG (DP 12), BoPL6 again cleaves M-G bonds, producing solely ΔM-oligosaccharides and no dimers (Fig. 6A). This means that BoPL6 accepted M in the −1 subsite and was unable to further process the ΔM-products, suggesting that it was exo-acting without the ability to cleave Δ-M bonds.Figure 6PolyMG (DP 12) depolymerization by BoPL6, BoPL6-NTD, and BoPL6-NTD + BoPL6-CTD monitored by time-resolved ^1^H NMR. A, polyMG (9.2 mg/ml) and BoPL6 (200 nM), (B) polyMG (9.2 mg/ml) and BoPL6-NTD (1.65 μM), (C) polyMG (8.8 mg/ml), and BoPL6-NTD (1.63 μM) + BoPL6-CTD (6.42 μM). The time-resolved spectra show H-4 of the Δ products (top) and normalized integrals for H-1 of the unsaturated products plotted through time (bottom). Integrals are normalized relative to the internal -GM- signal by calibrating the H-1 -GM- signal to 1.0 in the first spectrum. All reactions were in 10 mM HEPES, pH 8.0, 200 mM NaCl, 530 μM CaCl_2_ in 99.9% D_2_O at 25 °C. One spectrum (number scans = 32) collected every 5 min for a total of 14 h and 10 min (170 spectra). ΔM-: Δ1,4 linked β-d-mannuronate of DP three or larger, ΔM_α_: dimer of Δ 1,4 linked α-d-mannuronate, ΔM_β_: dimer of Δ 1,4 linked β-d-mannuronate, -GM-: α-l-guluronate 1,4 linked β-d-mannuronate. D, summary of which substrates (polyG and polyMG) will lead to a reaction, and which do not react in BoPL6 WT and BoPL6-NTD.

The removal of the CTD enabled NTD to act endolytically. BoPL6-NTD cleaved M-G bonds of polyMG (DP 12), producing ΔM-ends in DP two and larger AOSs (Fig. 6B). In fact, BoPL6-NTD processed a larger portion of the polyMG substrate (ΔM- ≈ 60% relative to internal -GM- in the starting material) compared to BoPL6 and a molar ratio of NTD:CTD of 1:4 still permitted endolytic activity but slowed the action on the substrate (Figs. 6C and S15D) compared to the NTD alone. Taken together, BoPL6 required G in the +1 subsite and could accept either M or Δ, but not G, in subsite −1. BoPL6-NTD also required G in subsite +1, cleaving M-G bonds but was unable to act on Δ-M and Δ-G bonds (Fig. 6D). Additionally, while BoPL6 was exo-acting from the non-reducing end, BoPL6-NTD alone appeared endolytic.

Upon docking of the hexasaccharide (MG)3 into the active site of BoPL6-NTD (centered around Asn201, Lys240, Arg261, and His262), it became apparent that a groove fitting the substrate chain was present in the BoPL6-NTD fold (Fig. S19, B and D). Looking closer at the active site, previous studies suggested that Lys240 was the catalytic base for the PL6 family based on mutation scans of the active site (34, 35, 36). Structural studies have, however, given little insight into the mechanism, as all crystal structures with substrate found the substrate located in a non-catalytically active pose (Fig. S20). As AlphaFold models are uncertain on the sidechain placement, we compared the active site residues of the BoPL6 model to those found in the PL6 crystal structures (Fig. S21). This revealed that the catalytic residues from the BoPL6 model lie within the catalytic residues of the crystal structures, reflecting an accurate prediction of the active site. Though the crystal structures generally agree on the location of active side residues, there was an overall variation of 1 Å from structure to structure. Crystal structures were highly affected by their crystallization condition and potentially do not match experimentally confirmed active conditions. To match our experimental conditions for activity, Ca^2+^ was added to the retained calcium binding pocket (Glu204, Glu233, Glu235) (48). Docking of ΔGG in BoPL6 showed that Δ fell in −1 subsite and GG (G-2, G-3) fell in subsites +1 and + 2 (Fig. 7A). The Ca^2+^ coordinated the second uronate group (G-2) at subsite +1 ensuring that the C5 proton was located near Lys240 N_ζ_ (Fig. 7C). Asn201 N_δ2_ was the closest residue to the 1_O^4^, suggesting it as the possible proton donor (acid) in this reaction, strongly supporting the hypothesis by Feingold (49) that Δ**^(─1)^↓G^(+1)^** cleavage is performed in anti when G is in subsite +1. The (MG)3 AOS was found in a slightly different coordination when docked into BoPL6-NTD as compared to ΔGG docked into BoPL6 (Fig. S17, A, C and E). The second, third and fourth uronate (M-2, G-3, M-4), located in the −1, +1 and + 2 subsites were slightly turned compared to the ΔGG pose in BoPL6 (Fig. 7B). Resultingly, the C5 proton of G-3 in subsite +1 was located close to Lys240 N_ζ and Asn201 N_δ2_ was near the _1_O^4^ (Fig. 7D). Thus, the reaction mechanism was not affected by the removal of the CTD as Lys240 and Asn201 were the acid base pair for both BoPL6 and BoPL6-NTD.Figure 7Docking of ΔGG and (MG)3 to BoPL6. Substrate orientation, distances to BoPL6 active site residues Asn201, Lys240, Arg261, and His262, and hypothetical mechanisms. A and B, orientation of docked substrates (ΔGG, left, and (MG)3, right), subsites are labeled with numbers. The backbone is shown as cartoon in black and white, with active site residues shown in orange. C, the guluronate at subsite +1 of docked ΔGG. D, the fourth uronate residue of (MG)3 at subsite +1. Distances to _1_O^4^ (red boxes) and to the C-5 proton (blue boxes). The corresponding proposed mechanisms are shown in the center of panels A and B.

We compared the substrate binding surface of the BoPL6-NTD to crystalized WT PL6 enzymes (34, 48) and found that the binding surface of the BoPL6-NTD was curved, as compared to the linear binding surfaces for the WT PL6s (Fig. S22). The negative subsites of the docked AOS are located at the NTD-CTD interaction surface (Fig. S23). The architecture of the binding surface at the negative subsites has not evolved to bind AOSs but rather evolved to interact with the CTD, resulting in the non-linear binding site. The placement of CTD thus blocks access to subsites beyond −1, explaining the inherent exolytic activity of all two-domain PL6’s.

As the two-domain PL6 enzymes BoPL6, Bacteroides eggerthii PL6 (BePL6), AlyGC, and BcAlyPL6 all prefer oligosaccharides over polysaccharides (23, 25, 36, 37) and BoPL6 preferred shorter over longer alginates (Fig. S24), we analyzed the effect of recombining the two domains of BoPL6 against three alginates, AlgM, AlgMG, and AlgG of F_G_ 0.35, 0.48 and 0.65 (Table 2 and Table S3). While the BoPL6-NTD had similar product formation curves on AlgM and AlgMG, this was increasingly reduced by equimolar and four-fold molar excess addition of CTD (Fig. S25). BoPL6 showed a high initial rate, but low product formation for AlgG releasing Δ, linearizing spontaneously as monitored by loss of absorbance at 235 nm (Fig. S25). However, on AlgG, BoPL6-NTD alone and in combination with the CTD produced oligosaccharides rather than Δ, as judged from the progress curves.Table 2. Substrate characteristicsSubstrateMw (kDa)Mn (kDa)DPFGFM_PolydispersityPolyMn.a.n.a.24‒300.001.00n.a.PolyGn.a.n.a.≈ 100.950.05n.a.PolyMGn.a.n.a.≈ 260.500.50n.a.Polyguluronate (CS-PG)6–8n.a.n.a.0.820.18n.a.AlgM260139n.a.0.350.651.9AlgMG300148n.a.0.480.522.0AlgG269110n.a.0.650.352.4PolyG (NMR)n.a.n.a.20‒22>0.95<0.05n.a.PolyMG (NMR)n.a.n.a.120.450.55n.a.PolyG (NMR)n.a.n.a.10‒300.900.10n.a.Weight average molecular weight (M_w), number average molecular weight (M_n_), polydispersity (M_w_/M_n_), and M-, G-fractions. n.a. not applicable.

The inherent role in the activity of different PL6 CTDs

To examine if CTDs generally enforce activity, substrate preference and mode-of-action in two-domain PL6 enzymes, NTD and CTD were also prepared for the two-domain BePL6 from the gut bacterium B. eggerthii DSM 20697, which has 49% sequence identity to BoPL6. BePL6, similarly to BoPL6, was only active on polyG and releases monosaccharides, oligosaccharides of DP3–6, and small amounts of disaccharides (25). The estimated domain interaction area of BePL6 was 1012 Å^2^ and 10% smaller than that of BoPL6 (1151 Å^2^). Like BoPL6-NTD, BePL6-NTD preferred polyMG as substrate, and the BePL6 function on polyG was recovered by addition of 4-fold molar excess of its CTD (Fig. S26), but with a lower initial rate than that of BoPL6-NTD + BoPL6-CTD, albeit leveling off at the same amount of product (Fig. 8). The activity on polyG of BePL6-NTD improved in a concentration-dependent manner up to about 8–10-fold molar excess of BePL6-CTD, again resembling the effect of BoPL6-CTD on the activity of its NTD (Figs. S11 and S26). However, combining BePL6-NTD with BoPL6-CTD did not raise its inherent low initial rate of polyG degradation (Fig. 8B), whereas combining BoPL6-NTD with BePL6-CTD did increase the initial rate by 2.6-fold (Fig. 8A). This effect of mixing the BePL6 domains was modest, in light of the BoPL6-NTD + BoPL6-CTD combination increasing the initial rate on polyG 63 times over the NTD alone (Fig. 8B). The addition of the CTDs from BoPL6 and BePL6 to the previously characterized one-domain enzymes, BcelPL6 (48) and Pedsa0632 (32), neither affected activity nor changed their substrate preference (data not shown). Thus, the two-domain and single-domain PL6s may have distinct evolutionary trajectories, which agree with the phylogenetic trees (Figs. S2 and S3).Figure 8Combination of PL6 NTDs and CTDs from Bacteroides ovatus CP926 and Bacteroides eggerthii DSM 20697. Progress is monitored for degradation of 1 mg/ml polyG by (A) 1 μM BoPL6-NTD (black), 1 μM BoPL6-NTD + 4 μM BoPL6-CTD (orange), and 1 μM BoPL6-NTD + 4 μM BePL6-CTD (blue). B, 1 μM BePL6-NTD (black), 1 μM BePL6-NTD + 4 μM BePL6-CTD (orange), 1 μM BePL6-NTD + 4 μM BoPL6-CTD (blue), and 4 μM BePL6-CTD (green). All reactions were performed in 20 mM HEPES, 150 mM NaCl, 2 mM Ca^2+^, pH 8.0, at 37 °C.

Discussion

In the phylogenetic tree comprising 791 one-domain and 360 two-domain PL6 enzymes (Figs. S2 and S3), 67 of the one-domain PL6s were linked to other domains, e.g., PL7, GH16, and different CBMs. While the one-domain enzymes belonged to several phyla, mostly Bacteroidota, but also Pseudomonadota, Actinobacteria, Firmicutes, and Planctomycetes, the two-domain PL6s were only from Bacteroidota and Pseudomonadota, suggesting different phyla are associated with different approaches for alginate degradation (23). Some PL6s catalyze the release of Δ monomers, and others form oligosaccharides with a Δ residue at the non-reducing end (32). Notably, many organisms encoding several putative PL6 enzymes were incapable of growing on alginate, and alginate utilization across phyla was not well understood (20). B. ovatus CP926 has overcome this challenge through synergy between its three ALs (23), and organisms producing single-domain PL6 enzymes probably need a different enzyme for monosaccharide production (20).

Removal of the CTD by truncation of the two-domain AlyGC from the marine bacterium Glaciecola chathamensis S18K6 (36) and BcAlyPL6 from the human gut bacterium B. clarus (37) led to 5.7% and 1.6% residual activity of the resulting NTD toward polyG and polyMG, respectively. Similar to BoPL6, these enzymes were exo-acting, producing monosaccharides and short oligosaccharides. AlyGC was not characterized for a change in mode of action or product profiles, but the CTD was proposed to be essential for the dimeric state in solution and for catalytic activity (36). By contrast, the CTD in BcAlyPL6 was suggested to be essential for the exolytic mode of action without further investigation into possible mechanisms. We speculate that the ability to degrade polyMG by NTD stems from its active site being more accessible. Structures of three one-domain PL6s, all reported to be endolytic, show an open-cleft active site (20, 34, 48). The change in preference of NTD to polyMG from polyG of BoPL6 correlates with polyMG fitting better into the active site of one-domain enzymes (Fig. S17). Another one-domain PL6 AlyF from Vibrio splendidus OU02 has a more closed pocket-shaped active site but was reported to be endolytic, preferring polyG despite this active site architecture (20, 35, 50). In AlyGC as well as BcAlyPL6, a loop from the CTD seemed to take part in shaping the active site and positioning of substrate by interaction with the uronate residue at the −1 subsite similar to BoPL6 (Fig. S13E). Previously, the aspartic acid and serine in the conserved DEST loop of the CTD (Fig. 2A) were hypothesized to be important [28,30]. Mutational analysis in AlyGC and BcAlyPL6 showed in particular the aspartic acid to have strong impact on activity, which in the AlyGC mutant was reduced to a similar level as by removing CTD and lost completely in BcAlyPL6. The mutation of the serine had less influence on activity. By docking G4 into BcAlyPL6, the Asp and Ser were shown to shape the catalytic pocket, playing a role in enforcing the favorable conformation of the substrate (37). Our docking of ΔGG to BoPL6 showed interaction with the glutamic acid of the DEST loop, and the E634A mutant showed E634 was as important for activity as Asp633, as a 20-times higher concentration of E634A had negligible activity (Fig. S13, A and C). Further, the alanine mutant (DAST) of the loop in BoPL6-CTD, improved the initial rate of polyMG degradation by the NTD + CTD mixture (Fig. S16D). Thus, the DEST loop influenced the positioning of the substrate in subsite −1 where lack of sufficient coordination abolishes enzyme activity. E634 proved crucial for binding of CTD to NTD, and even in the full-length enzyme, the E634A mutation inferred changes to the tertiary and quaternary structure. Peak deconvolution revealed that 66% migrated similarly to the WT, while 34% eluted earlier, suggesting a larger hydrodynamic volume. The importance for binding was corroborated by the CTD E201A not being able to bind to NTD mutation and none of the mutants had activity on polyG.

CTD truncation of BcAlyPL6, reduced kcat on polyG from 54.75 ± 0.83 s^—1^ for the two-domain enzyme to 0.85 ± 0.01 s^—1^ for the NTD, corresponding to 1.6% residual activity (37) and comparable to the reduction in kcat for BoPL6-NTD to 0.31%. In this case, BoPL6-NTD has 15.8 times higher kcat towards polyMG than on polyG, in agreement with its altered substrate preference compared to BoPL6. Reconstitution of BcAlyPL6 by mixing the NTD and CTD was not performed, and the maintained substrate preference for BcAlyPL6-NTD, may be interrelated with the lack of information on purity and structure for the used uncharacterized substrates. Importantly, as substrate preference and Km differ between BoPL6-NTD and BoPL6, and the CTD is not catalytically active. The function of CTD and E634 in BoPL6 was multifaceted as it influences both, binding, kinetics, and mode of action on the substrate. The catalytic rate of the BoPL6-NTD does, however, not compare to WT one domain PL6 enzymes, having a kcat 20x lower than BcelPL6 (2.06 s^‒1^ versus 43.4 s^‒1^). The Km also does not compare, being 39x lower (0.05 mg/ml vs 1.96 mg/ml), suggesting that BoPL6-NTD binds polyMG tightly. As the negative subsites were optimized for binding by the CTD, it suggested that the cleaved substrate may linger here, as the area did not evolve to release cleaved substrates. The distinct evolutionary path is also observed in the phylogenetic tree, where two domain PL6 enzymes form a separate clade based on sequence alignment of the NTDs.

Violot et al. (34) mapped the function of active site residues in several PL6 enzymes and described lysine and arginine as catalytic residues, albeit without suggesting a reaction mechanism. The substrate poses obtained from docking of ΔGG and (MG)3 into BoPL6 and BoPL6-NTD showed that Lys240 was favored as the catalytic base in BoPL6 for C-5 proton abstraction in ΔGG, whereas Arg261 did not appear to be located near the C-5 proton. We hypothesize that the positioning of the C-5 proton and _1_O^4^ relative to the residues Asn201 and Lys240 seen in BoPL6 in the +1 subsite determined the rate of the reaction, whereas coordination in −1 and +2 affects binding and thus M-G or G-G specificity in the PL6 family.

Both BoPL6 and BoPL6-NTD accept M in subsite −1, must have G in +1, and allow for both M or G in +2. While BoPL6 acts exolytically with Δ in subsite −1 and G in +1, this reaction was not observed for BoPL6-NTD. With the removal of the CTD, the substrate is likely more favorably bound in the negative subsites, thus Δ was not accommodated in subsite −1. BoPL6-NTD showed low activity while BoPL6 did not act on G-G bonds, which fits with its WT in vivo activity being the last step of a digestion pathway where other ALs will generate ΔG-rich ends (23). We hypothesize that these findings hold true for other one and two-domain PL6s but will require further NMR studies.

The functionalities of the NTD and CTD of BoPL6 were inherent in BePL6 by its regaining mode of action by domain mixing. Furthermore, the substrate preference of the NTD from both BoPL6 and BePL6 changed from polyG to polyMG. By addition of CTD, an increase in initial velocity was observed for polyG, which was greater for the BoPL6 than the BePL6 NTD. The lack of effect of mixing BePL6-NTD and BoPL6-CTD showed that the domains were not interchangeable, but specific for the respective enzymes. Differences in oligosaccharide products between BoPL6 and BePL6 point to functional differences in the two enzymes at the NTD–CTD interface, which can also account for the lack of combinatorial ability among the domains.

Conclusion

Active sites in ALs from PL6 can be categorized into three structural architectures: open-cleft, dead-end-cleft, and pocket. Here we have shown that the isolated N-terminal domain, which contains the active site of a PL6 preferring G in the +1 subsite, when separated from its C-terminal domain, was specific towards M^(─1)^↓G^(+1)^M^(+2)^ linkages producing ΔM ends. We have discovered that both the WT and truncated version allowed for M in the −1 subsite, and while BoPL6-NTD also allowed for G in the −1 subsite, this was not possible in the WT. The CTD of intact two-domain PL6 members regulated the substrate in the active site. Specifically, the CTD blocked access to minus subsites and altered the orientation of the substrate in relation to active site residues Asn201, Lys240, and His262 (BoPL6 numbering), where Lys240 is the suggested catalytic base in the absence of CTD and Asn201 is suggested as the catalytic acid. Mixing the two individual domains of BoPL6 reconstituted the activity, but this can vary for other two-domain PL6 enzymes. The evolution of the two-domain PL6 arose from its strong genetic connection to, and synergistic action with, other lyases, where it served to degrade oligosaccharides to monosaccharides, as opposed to one-domain PL6 enzymes that release oligosaccharides. Several bacteria were found to encode more than one PL6 enzyme, suggesting that a combination of a one-domain with a two-domain PL6 was an alternative approach to combining ALs of different families with clear synergistic capability for efficient alginate degradation and utilization.

Experimental procedures

Materials

Alginates vary a lot, and ALs are very substrate sensitive, requiring the use of well-defined different model substrates and alginates. PolyM was obtained from an epimerase-negative AlgG mutant of Pseudomonas fluorescens (6), polyG and polyMG were prepared as previously described (51). Polyguluronic acid (CS-PG) was purchased from Carbosynth. Sodium alginates (AlgM, AlgMG, and AlgG) (8) were kind gifts of International Flavors & Fragrances (IFF, now Roquette, Sandvika, Norway) (Table S2).

Full-length proteins studied were presented in previous studies (23, 25). BoPL6 (GenBank acc. no. WII03612.1) originates from B. ovatus CP926, and BePL6 (GenBank acc. no. EEF87057.1) from B. eggerthii DSM 20697.

Phylogenetics and multiple sequence alignment analyses

Accession numbers and enzyme classification data for sequences of the PL6 family were extracted from the CAZy database (18) and downloaded from GenBank (52). Redundancy removal was performed using CD-hit (53) with a 0.95 similarity threshold. Signal peptides and potential N-terminal transmembrane regions were predicted by Phobius (54) and removed before alignment and truncation in Qiagen CLC main workbench 20, to only include either the N-terminal catalytic domain (791 distinct patterns) or the C-terminal domain (360 distinct patterns). Obvious fragments were removed during this process. The alignments were used to generate Maximum Likelihood (ML) phylogenetic trees using RaxML-HPC BlackBox (v. 8.2.10) (55) at the CIPRES science gateway (56) using the LeGascuel substitution matrix (LG) (57) and otherwise default parameters. RaxML stopped the rapid bootstrap search for both trees at 300 after meeting the MRE-based Bootstrapping criterion. For visualization of the multi-modular nature of PL6 sequences on the tree, putative domain boundaries were predicted using the dbCAN meta server (58) and visualized in ITOL (59).

AlphaFold2

BoPL6, BoPL6-NTD, BoPL6-CTD, BePL6, BePL6-NTD, and BePL6-CTD were modelled using the online version of AlphaFold2 (60) available on Google Colab server (61) using the standard settings, without using the multimer model. All models were evaluated by inspecting the predicted local distance difference test used as color code for the models (Fig. S24). Domains folded separately were superimposed to the full-length model and aligned to reveal possible structural changes caused by the truncations.

Production of protein domains and variants

pET28a(+) plasmids purchased from GenScript harbored the predicted BoPL6-NTD, BoPL6-CTD, BoPL6-CTD E201A, and BoPL6 E634A cloned in by NheI/BamHI and XbaI/BamHI and BePL6-NTD and BePL6-CTD cloned in by NdeI/BamHI and NcoI/BamHI. The BoPL6 E634A, BoPL6-NTD, and BePL6-NTD contained an N-terminal His-tag and BoPL6-CTD, BoPL6-CTD E201A, and BePL6-CTD a C-terminal His-tag. Plasmid encoding pesda0632 (Previously described by Mathieu et al. (32)) was also purchased from GenScript, cloned in by NcoI/XhoI into pET28a(+) by NcoI/XhoI, and plasmid encoding BcelPL6 was obtained from a previous study by the authors of this study (48). The plasmids were transformed by heat shock into E. coli. BL21 was selected on LB agar plates containing 50 μg/ml kanamycin. A single colony was picked, inoculated in 10 ml LB medium with 50 μg/ml kanamycin, and grown overnight (37 °C, 170 RPM). The culture was used to seed 800 ml LB medium (10 mM glucose, 50 μg/ml kanamycin), grown (37 °C, 170 RPM) to an OD_600_ of 0.6‒0.8 in shake flasks, and then added isopropyl β-D-thiogalactopyranoside (IPTG) to 0.5 mM for induction at 18 °C (170 RPM). Cells were harvested after approximately 18 h (5000g, 20 min, 4 °C), and pellets were kept at −20 °C until protein purification. Full-length BoPL6 and BePL6 (also referred to as WT) were produced as previously described (23, 25).

Protein purification

Cell pellet (2‒4 g) corresponding to approximately 400 ml culture was resuspended in 20 ml (4 °C) HisTrap binding buffer (50 mM HEPES, 300 mM NaCl, pH 8), supplemented with 2 μl Benzonase Nuclease (Sigma-Aldrich), lysed at 1 bar by high-pressure homogenization (Pressure cell homogenizer, Stansted Fluid Power), and centrifuged (20,000g, 20 min, 4 °C). HisPur nickel-nitrilotriacetic acid resin (2 ml; Thermo Fisher Scientific) equilibrated with 20 ml binding buffer, applied supernatant, gently mixed (4 °C, 30 min), and transferred to a gravity flow column. The column was washed with 20 ml binding buffer, followed by 15 ml binding buffer containing 20 mM imidazole, and eluted with 5 ml binding buffer containing 300 mM imidazole. The eluted protein was further purified by size exclusion chromatography using a HiLoad 16/600 Superdex 200 pg column (GE Healthcare) equilibrated with 20 mM HEPES, 150 mM NaCl, pH 8, at a flow rate of 0.8 ml/min. Fractions containing protein of expected size were pooled, and the purity verified by SDS-PAGE (Fig. S4). Protein concentrations were determined spectrophotometrically at 280 nm using molar extinction coefficients of BoPL6-NTD = 39,350, BoPL6-CTD = 29,380, BePL6-NTD = 40,840 and BePL6-CTD = 34,755 M^– 1^cm^– 1^ as predicted by ProtParam (39).

Biochemical characterization

The pH, NaCl and metal ion dependency on activity was determined for 1 μM BoPL6-NTD on 2 mg/ml CS-PG. This substrate is available in ample amounts, whereas the other model substrates are available in limited amounts. The pH optimum was determined in 20 mM UB4 buffer (62) pH 6 to 9 supplemented with 150 mM NaCl. The effect of 0 to 1 M NaCl was determined in 20 mM HEPES, pH 7.75. Dependence of divalent cations was determined at the optimal pH and NaCl concentration. Substrates were incubated with 100 mM EDTA and enzymes with 10 mM EDTA for 1 h at 4 °C. Substrate was dialyzed against 3x 100 volume ddH_2_O (4 °C, 1 kDa cutoff, SpectrumLabs, Greece) and freeze-dried (Scanvac Coolsafe, Holm and Halby). Enzymes were dialyzed 3 times against x100 volume of appropriate buffer (4 °C, 12–14 kDa cutoff, SpectrumLabs, Greece). Enzyme and substrate were incubated with 2 mM CaCl_2_, MgCl_2_, MnCl_2_ and NiCl_2_ and activity was measured under standard conditions (20 mM HEPES, 150 mM NaCl, pH 7.75) spectrophotometrically (Bio-Tek Powerwave XS; Holm and Halby) by following formation of Δ at 235 nm (63, 64) every 10 s at 37 °C. Enzyme and substrate were pre-incubated (5 min at 37 °C) before mixing in 96-well UV-star chimney well plate (In Vitro). All measurements were performed in triplicate.

Substrate preference and product profile analyses

Substrate specificities were investigated for 1 μM BoPL6-NTD, BoPL6-CTD, BePL6-NTD, BePL6-CTD and BoPL6-NTD + BoPL6-CTD towards 1 mg/ml AlgM, AlgG, AlgMG, polyM, polyG and polyMG under optimal reaction conditions as determined for BoPL6-NTD, and measuring activity under standard conditions for up to 2 h. Product profiles were analyzed on mixtures of 1 μM BoPL6-NTD, and BoPL6-NTD + CTD (1 μM + 4 μM) with 5 mg/ml CS-PG incubated (37 °C, 600 RPM; Eppendorf ThermoMixer C). The reactions were terminated at 0, 1, 2, 5, 30, 60 and 120 min by heating (95 °C, 10 min), centrifuged (10,000g, 4 °C, 10 min), and the supernatants were stored at −20 °C. The samples were thawed, added acetonitrile (1:1, v:v), centrifuged (10,000g, 4 °C, 10 min) and the supernatants were analyzed the same day by LC-ESI-MS as previously described (24) (Amazon SL iontrap (Bruker Daltonics) coupled to UltiMate 3000 UHPLC equipped with GlycanPac AXH-1 column, 150 x 2.1 mm (Thermo Fisher Scientific). Relative quantification of compound intensities was performed in TASQ 2.2 (Bruker Daltonics). Detected DP1 products comprise uronate monomer as well as the downstream products arising from spontaneous ring opening, DEH (4-deoxy-l-erythro-5-hexoseulose), DEH hydrates and DHF (4-deoxy-d-manno-hexulofuranosidonate) epimers, which cannot be distinguished by LC-ESI-MS.

Enzyme kinetics

Initial velocities within the first 100 s of release of Δ products (under standard conditions) by 500 nM BoPL6-NTD acting on 0.01 to 4.8 mg/ml polyG, and 100 nM BoPL6-NTD acting on 0.006 to 3.2 mg/ml polyMG were obtained by linear regression. Absorbance units [AU] were converted to μM using the extinction coefficient of 6150 M^−1^ cm^−1^ (65, 66). The kinetic parameters kcat and Km were determined by plotting initial velocities [μM/s] against initial substrate concentrations [mg/ml] by fitting the Michaelis-Menten equation v_0_ = V_max_/(1+(Km/[S]0)) to the data using GraphPad Prism 9.3.1 (GraphPad Software)

Thermostability

Thermostability was determined for 3 μM BoPL6-NTD or BoPL6-CTD (in triplicate) at different pH (20 mM UB4 buffer (62), 150 mM NaCl, pH 4–9); salt concentration (20 mM HEPES, pH 7.75, NaCl concentrations 0–1 M) and divalent cations (20 mM HEPES, 150 mM NaCl, pH 7.75) using nano differential scanning fluorimetry loaded into Prometheus capillary tubes, placed in the Prometheus Panta (Nanotemper technologies) (Figs. S23–S25). The fluorescence at 330 nm and 350 nm was measured continuously from 20 to 95 °C with a linear gradient of 1.5 °C/min. The data were analyzed using the manufacturer’s software (PR.Stability Analysis software) to obtain the protein melting temperature.

Analytical size-exclusion chromatography

The five enzyme variants, BoPL6, BoPL6-NTD, BoPL6-CTD, BoPL6-CTD E201A, and BoPL6 E634A, were diluted in SEC buffer (20 mM HEPES, 150 mM NaCl, pH 8.0) to a final concentration of 60 μM. For interaction studies 40 μM BoPL6-NTD was mixed with 80 μM BoPL6-CTD or BoPL6-CTD E201A and incubated for 3 h 500 μl were loaded onto the Superdex 200 Increase 10/300 Gl column (Cytiva) and eluted isocratically at a 0.5 ml/min flow. Fractions of 1 ml were collected throughout the elution. A high molecular weight gel filtration calibration kit (Cytiva) was used for molecular weight estimations. The standard mix included ovalbumin (5 mg/ml), conalbumin (3 mg/ml), aldolase (2 mg/ml), and ferritin (1 mg/ml) and was analyzed under identical conditions.

Fractions corresponding to chromatographic peaks were analyzed by SDS-PAGE using Novex 4 to 20% Tris-Glycine Plus WedgeWell gels (Invitrogen). PageRuler Plus Prestained Protein Ladder (Thermo Scientific) was used as a molecular weight marker.

Isothermal titration calorimetry

BoPL6-NTD, BoPL6-CTD and BoPL6-CTD E201A were dialyzed (3x 100-fold volume, 3000 MWCO Spectra/Por membrane, Spectrumlabs) against 20 mM HEPES, 150 mM NaCl, pH 8.0 and adjusted to 80 and 800 μM, respectively. Degassed (2 min) NTD (200 μl) and CTD (40 μl) were loaded into the cell and syringe, respectively. Titration at 25 °C (with stirring at 750 rpm) involved an initial 0.4 μl injection followed by 19 injections of 2 μl (ITC200, Thermo Scientific, USA). Raw data were corrected for heat of dilution by subtracting a blank titration (Fig. S26). Normalized, integrated raw data, plotted as enthalpograms were fitted with an independent one-site binding model to determine dissociation constant (Kd), stoichiometry (n) and enthalpy (ΔH) using the instruments Origin-based software (Microcal Analysis).

Intrinsic fluorescence

The binding of polyguluronate to 4 μM BoPL6-CTD was analyzed via changes in intrinsic tryptophan fluorescence by adding CS-PG (50 mg/ml) in 0.5 mg/ml steps to final 4 μM BoPL6-CTD and 10 mg/ml CS-PG. Tryptophan fluorescence emitted at 310‒370 nm by excitation at 295 nm was measured at 37 °C and 250 RPM stirring (Jasco 8500 spectrophotometer; Jasco International Co). Excitation and emission bandwidths were 5 nm and the response time 0.5 s. Five spectra were averaged for samples measured in triplicate to produce the resulting spectra. Kd,app was calculated by fitting the following formula to the data: $[eqn]$ , where Y_0_ is the fluorescence in the absence of ligand, X is the ligand concentration in mg/ml or mM, A is the slope at $[eqn]$ and Y is the fluorescence at 317 nm.

Nuclear magnetic resonance spectroscopy

NMR spectra were acquired at 25 °C on a Bruker Avance III HD 800 MHz spectrometer using a 5 mm z-gradient CP-TCI (H/C/N) cryogenic probe at the NV-NMR-Center/Norwegian NMR Platform (NNP) at the Norwegian University of Science and Technology (NTNU). The ^1^H chemical shift was internally referenced to the residual water signal (4.75 ppm) and the ^13^C chemical shift was indirectly referenced to DSS using a ^13^ C/^1^H frequency ratio = 0.251449530 (67). Spectra were recorded, processed, and analyzed using TopSpin 3.5 or 4.1.4 (Bruker BioSpin). Reactions were run in 10 mM HEPES, 200 mM NaCl, 0.53 mM CaCl_2_, pH 8.0 in 99.9% D_2_O in 3 mm LabScape Stream NMR tubes (Bruker LabScape).

For time-resolved experiments, the substrate (polyG (NMR) and polyMG (NMR)) dissolved in buffer was preheated to 25 °C in the NMR instrument and a 1D ^1^H spectrum recorded. The reaction was started by adding the enzyme (preincubated in buffer with calcium) and mixed by inverting three times. Immediately thereafter, the sample was re-inserted into the NMR instrument and the experiment started. The recorded spectrum is a pseudo-2D experiment recording a 1D ^1^H spectrum every 5 min. After each time-resolved experiment, a ^1^H-^13^ C heteronuclear single quantum coherence (HSQC) spectrum with multiplicity editing was recorded. Signals were assigned according to previous data (32, 68, 69).

1D ^1^H T1ρ relaxation-filtered experiments were recorded for BoPL6-CTD mixed with each substrate. First, a reference experiment with a spin-lock of 10 ms was recorded, followed by a second experiment with a spin-lock of 400 ms. Intensity of each spectrum pair is standardized using the signals of HEPES. To estimate the percentage decrease, all signals were integrated relative to the HEPES signal (=1.0). The difference was calculated as 1-(Integral_400 ms_/Integral_Ref_) expressed as a percentage. Signals were assigned according to (47, 70).

Docking

The AlphaFold2 structures of BoPL6 and BoPL6-NTD were prepared using the Protein Preparation Wizard within the Schrödinger suite 2016-1 (71) for pH 7.75 and using the default settings. The ligands (ΔGG and (GM)3, (MG)3, M(GM)2, G(MG)2, (GM)2 and (MG)2) were prepared using the ligand preparation function in the Schrödinger suite 2016-1. According to amino acid substitutions and crystal structures of other PL6s, the center of the docking region (grid) was defined by Asn201, Lys240, Arg261 and His262. Docking calculations were performed using Glide within the Schrödinger suite 2016-1 (71) and applying Induced-Fit Docking (72) (IFD), where the option “sample ring conversions” was deactivated. All other parameters corresponded to the default settings.

Data availability

Data is available upon request.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.