Introduction

Glycosaminoglycans (GAGs) are a class of linear polyanionic polysaccharides ubiquitously distributed on the cell surface and in the extracellular matrix (ECM) of animal tissues (Cohen and Merzendorfer,2019), and they participate in various physiological and pathological processes through interacting with a series of chemokines (Dyer, et al.,2015; Irie, et al.,2008), growth factors (Sirko, et al.,2010; Zhang, et al.,2019) or other ECM components. Except for keratan sulfate (KS), which does not contain hexuronic acid (HexUA) residues, other GAGs are composed of repeating disaccharides consisted of HexUA (D-glucuronic acid (GlcUA) or L-iduronic acid (IdoUA)) and hexosamine (HexNAc) (D-N-acetylgalactosamine (GalNAc) or D-N-acetylglucosamine (GlcNAc)). Based on the disaccharide composition and the type of glycosidic bonds between disaccharides units, HexUA-containing GAGs are classified into three classes: hyaluronan (HA), chondroitin sulfate (CS)/dermatan sulfate (DS) and heparan sulfate (HS)/heparin (Hep) (Cohen and Merzendorfer,2019). HA is the only non-sulfated GAG composed of repeating -GlcUAβ1-3GlcNAc-disaccharides linked by β1-4 glycosidic bonds. In contrast, the structures of CS/DS and Hep/HS are highly complex due to sulfation and epimerization modifications. The backbone of CS is composed of repeating -GlcUAβ1-3GalNAc-disaccharides linked by β1-4 glycosidic bonds, and is further sulfated at C-2 of GlcUA and C-4/C-6 of GalNAc by sulfotransferases; meanwhile, the GlcUA residues in CS can be epimerized into IdoUA residues by glucuronyl C-5 epimerase to form DS (Kusche-Gullberg and Kjellen,2003). Similarly, HS/Hep composed of repeating −4GlcUAβ1-4GlcNAcα1-units can be sulfated at C-2 of GlcUA and N/C-2/C-3/C-6 of GlcNAc/deacetylated GlcN, and GlcUA residues are often epimerized to IdoUA residues, especially in Hep (Cohen and Merzendorfer,2019). Such complex modification through sulfation and epimerization endows sulfated GAGs with diverse biological functions.

GAG lyases, belonging to the superfamily of polysaccharide lyases (PLs), specifically catalyze the degradation of HexUA-containing GAGs (Charnock, et al.,2002; Garron and Cygler,2010). These reactions generate a plethora of oligosaccharides that contain an unsaturated double bond between C-4 and C-5 on uronic acids at their nonreducing ends, which is generated via a β-elimination mechanism (Garron and Cygler,2010). As a complimentary mechanistic strategy to glycoside hydrolases (GHs), GAG lyases degrade HexUA-containing GAGs without the participation of a water molecule and involved in the polysaccharide metabolism of microorganisms (Garron and Cygler,2010; Ndeh, et al.,2020). For the past many years, an increasing number of gene sequences containing PL catalytic modules and some ancillary modules have been annotated as PLs (Lombard, et al.,2010). Based on their sequence homologies, 43 families (and an unclassified PL0 family) have been hierarchically classified in the CAZy database (http://www.cazy.org/), and at least one member of each family has been subjected to activity identification and biochemical property analysis (Drula, et al.,2022). The identified GAG lyases are widely distributed in the PL6, PL8, PL12, PL13, PL15, PL16, PL21, PL23, PL29, PL30, PL33, PL35 and PL37 families (Drula, et al.,2022). Based on substrate specificity, GAG lyases are classified as HA-specific lyases, which can degrade HA only, CS/DS lyases (chondroitinase, CSases), which degrade CS/DS as well as HA in most cases, and Hep/HS lyases (heparinases, Hepases), which specifically cleave Hep/HS. All identified GAG lyases have substrate specificity to strictly distinguish one or two kinds of GAGs based on saccharide composition and glycosidic bonds between each HexUA-HexNAc disaccharide units, while other uronic acid-containing polysaccharides are usually unable to serve as substrates for these enzymes.

With the evolution of host polysaccharides (Csoka and Stern,2013; Popper, et al.,2011), the bacterial enzymes that degrade these polysaccharides have also evolved to varying degrees in response to complex environments. As a kind of unbranched anionic extracellular polysaccharide produced by lower organisms algae and bacteria (such as brown alga (Popper, et al.,2011) and bacteria belonging to Pseudomonas (Evans and Linker,1973) and Azotobacter (Clementi,1997), alginate containing no hexosamine and sulfation possesses a simpler structure than GAGs in animals and should emerge earlier than GAGs during the course of evolution. Like GAGs, alginate is a linear polyanionic polysaccharide that contains HexUA residues; however, alginate possesses a completely different chemical structure, which is composed of D-mannuronate (M) and its C5 epimer L-guluronate (G), and M and G residues often alternate randomly to form heteropolyuronic blocks, including polyM composed of M residues linked by β1-4 bonds, polyG composed of G residues linked by α1-4 bonds, and polyMG/GM composed of alternating M and G linked by β/α1-4 bonds (Pawar and Edgar,2012). Likewise, a series of alginate lyases essential for the metabolism of alginate were identified from microorganisms and algae as well as lower marine animals, which belong to the PL5, PL6, PL7, PL8, PL14, PL15, PL17, PL18, PL31, PL32, PL34, PL36, PL38, PL39 and PL41 families (Drula, et al.,2022). Based on their specificity, these lyases are also divided into the following categories: M block-specific lyases (Itoh, et al.,2019; Zhu, et al.,2015), G block-specific lyases (Matsubara, et al.,1998; Yamasaki, et al.,2005) and MG-specific alginate lyases (Jagtap, et al.,2014; Yamasaki, et al.,2004). From an evolutionary perspective in previous reports, GAG lyases are thought to have originated from alginate lyases by a divergent evolution through adaptation to the evolution of substrate polysaccharides, which is mainly attributed to some homology in the sequences and folds of these two types of enzymes, such as the (α/α)_n toroid and antiparallel β-sheet folds of GAG lyases in PL8 (hyaluronate lyase and chondroitin lyase), PL12 (heparinase III), PL21 (heparinase II) and PL23 (chondroitin lyase) families and alginate lyases in PL15, PL17 and PL39 families, the β-jelly roll fold of heparinase I in PL13 family and alginate lyases in PL7 and PL18 families, and the β-helix fold of chondroitinase B and alginate lyase in PL6 family (Garron and Cygler,2014). Considering that the alginate with simpler structure from brown alga or bacteria might appear earlier, it is possible that the ancestral enzymes originally used alginate as a substrate. However, enzymes with transitional characteristics in activity and structure have not been discovered to support this speculation.

Recently, a new class of GAG lyases belonging to PL35 family was discovered with an ultrabroad substrate spectrum (Wei, et al.,2024). Unlike the identified GAG lyases (HA-specific lyases, CSases and Hepases) with quite strict substrate specificities, these novel GAG lyases can degrade three types of uronic acid-containing GAGs (HA, CS and HS), and thus were named GAGases. More interestingly, one of the eight identified GAGases (GAGase III) can even degrade alginate. Analysis of substrate structure preference represented by GAGase I showed that GAGases selectively act on GAG domains composed of non/6-O-/N-sulfated hexosamines and D-glucuronic acid, among which GAGase III can also selectively degrade polyM blocks composed of D-mannuronic acids but not polyG blocks composed of L-guluronic acids in alginate. Furthermore, the primary structure alignment and phylogenetic analysis showed that GAGases have considerable degree of homology with not only GAG lyases from PL15, PL21 and PL33 families but also alginate lyases from PL15, PL17 and PL34 families (Wei, et al.,2024), indicating that GAGases may exhibit some transitional features from alginate lyases to GAG lyases in structure.

In this study, crystal structures of two GAGases (GAGase II and GAGase VII) were determined for the first time, and their catalytic mechanism was further elucidated through the site-directed mutagenesis of their crucial site residues and molecular docking. Structural alignment showed that GAGases are more structural similarity to some alginate lyases rather than GAG lyases, suggesting that PL35 family GAGases might originate from alginate lyases possessing an N-terminal (α/α)_n toroid domain and a C-terminal antiparallel β-sheet domain. Moreover, one of the reasons why GAGase III can additionally degrade M-block alginate are also resolved by structural modeling, structural alignment and site-directed mutagenesis. This study not only helps to resolve the catalytic mechanism of the PL35 family lyases in particular GAGases, but also provides a potential structural evidence for the divergent evolution of GAG lyases.

Results

Overall crystal structures of GAGase II and GAGase VII

To explore the substrate recognition and catalytic mechanism of the GAGases, the eight enzymes with different sequence identity (Supplemental Table S1) were individually used to prepare their crystals for structural analysis. Finally, the structures of GAGase II from Spirosoma fluviale (GenBank accession number: SOD82962.1, 81.2% sequence identity with GAGase I), GAGase VII from Bacteroides intestinalis (GenBank accession number: EDV05210.1, 44.8% sequence identity with GAGase I) and selenomethionine (SeMet)-labelled GAGase II (Se-GAGase II) in their ligand-free states were solved at 1.9 Å, 2.4 Å and 2.0 Å resolution, respectively (Table 1). Both GAGase II and GAGase VII are monomers consisting of two domains: a N-terminal (α/α)₇ toroid domain and a C-terminal two-layered β-sheet domain (Figure 1A).

Figure 1.

Table 1.

The N-terminal domain (Met³⁴-Ala³⁶⁵) of GAGase II is composed of 16 large or small α-helices (α1-α16) to form an (α/α)₇ toroid domain. Specifically, the N-terminal 16 α-helices constitute seven complete α-helix pairs (α1-2, α3-4, α5-7, α8-9, α10-12, α13-14 and α15-16), forming seven hairpin structures, which eventually embrace into the (α/α)₇ toroid structure. The transition between α6 to α8 and between α10 to 12 is discontinuous, and both transitions possess a small α-helix α7 and α11, respectively. In each hairpin structure, two antiparallel α-helices are connected by loops composed of 9-12 residues, and each hairpin contains a turn structure composed of 1-4 residues. The N-terminal of this toroid domain is connected to a signal peptide that has been removed during cloning expression, and its C-terminal is connected to the two-layered β-sheet domain through a 17-residue linkage (Trp³⁵⁰-Met³⁶⁵). Its inner layer is formed by helices α1, α3, α5, α8, α10, α13 and α16, while its outer layer is formed by α2, α4, α6+α7, α9, α11+α12, α14 and α15. Moreover, the α-helix pairs α13-14 and α15-16 are significantly shorter than the first five sets, and α15 and α16 are connected by a short β-turn consisting of only 3 residues, which makes the toroid structure closed (Figure 1A).

The C-terminal domain (Met³⁶⁶-Ser⁶¹²) of GAGase II is composed of 17 β-strands (β1-β17) to form a two-layered β-sheet domain with 4 small α-helices located between β5-β6 and β8-β9. Both layers of this C-terminal β-sheet domain are composed of eight groups of β-strands, namely β1-β5, β11 and β15-β16 and β7-β10, β12-β14 and β17. The two layers are roughly parallel to each other and twisted to form an angle of approximately 60°. The first five β-strands (β1-β5) of the C-terminal domain are arranged in anti-parallel configuration. Subsequently, there is an Ω-loop structure wrapping around a metal ions (Mn²⁺), which consists of three small α-helices and a small β-strands, with many positively charged amino acid residues (such as His⁴⁰¹, His⁴⁰³, Tyr⁴²⁷, Glu⁴³¹, Trp⁴³⁸, Arg⁴⁴⁶ and so on) on top, which might serve as potential substrate-binding sites. Finally, the C-terminal domain terminates at β17 in the second layer of the bilayer structure. Taken together, these structural elements constitute the C-terminal anti-parallel bilayer β-sheet domain of GAGase II (Figure. 1A).

Similar to GAGase II, GAGase VII also has an N-terminal (α/α)₇ toroid domain (Leu³⁵-Arg³⁶⁰) composed of 16 α-helices, but its α1 helix consists of only three residues (Glu⁴²-Val⁴⁴) and a long loop chain (Figure. 1B). Moreover, the C-terminal two-layered β-sheet domain (Met³⁶¹-Met⁶¹⁶) of GAGase VII is also composed of 17 β-strands (β1-β15, β16a, β16b and β17). However, unlike GAGase II, its β16 strands is interrupted into two small β-strands (β16a, β16b) by a randomly coiled loop chain (Figure. 1B). Compared with GAGase II, these additional loop regions may make GAGase VII less structurally stable than GAGase II, and may be an important factor in why the activity of GAGase VII is far lower than GAGase II. Nevertheless, structural alignment further shows that GAGase II has good superposition with GAGase VII and other GAGases, indicating that GAGases are highly conserved in structures (Supplemental Figure S1).

Mn binding site of GAGase II and GAGase VII

Unlike the GAG lyases identified from PL8, PL15 and PL21 families, which contain Ca²⁺ or Zn²⁺ in their overall crystal structures (Shaya, et al.,2008; Shaya, et al.,2006; Zhang, et al.,2021), a Mn²⁺ was detected in C-terminal domains of both GAGase II and VII by inductively coupled plasma-mass spectrometry (ICP-MS) analysis (Figure 1B). Structurally, the Mn²⁺ is located inside the long Ω loop after the β1-β5 antiparallel β-sheet, and coordinates with the nearby nitrogen and oxygen atoms of several residues such as Asp and His in both GAGase VII and GAGase II (GAGase II also has coordination with three surrounding water molecules), with distances of 2.0-2.5 Å. Such interaction should stabilize the nearby loop structures and may form a substrate-binding site relevant for substrate selectivity. To investigate the roles of these adjacent residues interacting with the Mn²⁺, His⁴⁰³, Asp⁴⁰⁵, Asp⁴²¹ and His⁴⁵⁷ in GAGase II were individually mutated to alanine (Ala), and the results showed that these mutations resulted in complete or substantial inactivation of GAGase II (Figure 2A). In addition, the inhibitory effects of the chelator (EDTA) and the stimulation of Mn²⁺ on the enzymatic activities also support the critical role of Mn²⁺ on GAGases (Figure 2B).

Figure 2.

Multiple structural alignments of GAGase II and its structurally similar GAG lyases and alginate lyases

The ultrabroad substrate degradation spectra of GAGases indicate that they should be structurally homologous with some other PLs with different substrate specificities. By using GAGase II as a representative, a structural alignment assay showed that the PL35 GAGases are structurally similar to various N-terminal (α/α)_n toroid domain and C-terminal antiparallel β-sheet domain-containing GAG/alginate lyases from PL8 (Fethiere, et al.,1999; Huang, et al.,2003; Li and Jedrzejas,2001; Lunin, et al.,2004; Shaya, et al.,2008), PL12 (Hashimoto, et al.,2014; Ulaganathan, et al.,2017), PL15 (Ochiai, et al.,2010; Zhang, et al.,2021), PL17 (Park, et al.,2014), PL21 (Shaya, et al.,2006), PL23 (Sugiura, et al.,2011) and PL39 (Ji, et al.,2019). Notably, GAGase II shares very low sequence identity with PL21 heparinase II from Pedobacter heparinus (Shaya, et al.,2006) (23.1% identity with 29% query coverage), PL12 heparinase III from Bacteroides thetaiotaomicronin (Ulaganathan, et al.,2017) (23.6% identity with 32% query cover) and PL39 alginate lyase from Defluviitalea phaphyphila (Ji, et al.,2019) (26.5% identity with 31% coverage), and no significant sequence identity with PL17 alginate lyases (Park, et al.,2014), PL15 exoHep (Zhang, et al.,2021), PL8 chondroitin ABC lyase (Shaya, et al.,2008) and chondroitin AC lyase (Lunin, et al.,2004); however, all are relatively well superimposable in their overall structures and highly conserved in their triplet active site residues (Figure 3). Among them, GAGases show more structural similarity to alginate lyases, such as those from the PL15 (PDB code: 3a0o) (Ochiai, et al.,2010), PL17 (PDB code: 4NEI) (Park, et al.,2014) and PL39 (PDB code: 6JP4) (Ji, et al.,2019) families, than various GAG lyases (Table 2), indicating that GAGases might originate from alginate lyase rather than GAG lyase. Reasonable support for this speculation could be the significant alginate-degrading activity of GAGase III. Therefore, the broad substrate spectrum of GAGases may be because they are evolutionarily transitional types with the enzymatic features of alginate lyases and GAG lyases.

Figure 3.

Table 2.

Catalytic center and substrate binding sites of GAGases

Based on the sequence and structural alignment with homogenous alginate/GAG lyases, the conserved residues Tyr²⁴⁶, His⁴⁰¹, Asn¹⁹² of GAGase II, Tyr²⁴³, His³⁹⁸, Asn¹⁸⁹ of GAGase III, and Tyr²⁴¹, His³⁹⁷, Asn¹⁸⁷ of GAGase VII may be the key triplet residues involved in their β-elimination catalysis active sites (Figure 3). To confirm this hypothesis, a site-directed mutagenesis assay was performed in which these residues were individually replaced by Ala; as a result, the activity of all of mutants toward CS-C was completely lost (Figure 4A). These results indicate that GAGases act through a general acid‒base catalytic mechanism by using His/Tyr as a Brønsted acid/base, similar to that observed for various other identified GAG/alginate lyases (Table 3). Compared with their homologous GAG/alginate lyases, GAGases contain a shorter catalytic cavity, which may explain why they can accommodate and degrade a variety of substrates with very different structures. However, we were unsuccessful in preparing the cocrystals of GAGases/their inactive mutants with substrates to reveal their substrate binding and degradation mechanisms. Alternatively, molecular docking experiments of GAGase II with several structurally defined hexasaccharide (Hexa) substrates were performed as shown in figure 4B. Based on the docking results obtained for GAGase II with an HA Hexa (GlcUA1–3GlcNAc)₃ (PDB code: 1HYA) (Winter, et al.,1975) (Figure 4B), the nonreducing end of the substrate at the “-1” and “-2” subsites is near some positively charged amino acids, including Arg⁸⁷, Arg⁸⁸, Arg⁹⁴, His¹³⁶, Asn¹⁹¹ and Asn¹⁹², in the vicinity of the catalytic cavity of GAGase II, which could facilitate the binding of negatively charged substrates. Besides, several acidic amino acids (Glu²⁴², Asp²⁹⁹ and Glu⁴³¹) are very close to the “+1” and “+2” subsites, which could promote the release of the reducing end product via charge repulsion. There are some other residues surrounding around the catalytic cavity, such as Tyr²⁴⁰, Gln³⁰⁷, Arg³⁴², Tyr⁴²⁷, Glu⁴²⁸, Glu⁴³¹, Trp⁴³⁸, Met⁴⁴⁰ and Arg⁴⁴⁶, which may play a key role in stabilization of the catalytic cavity and overall structure. Site-directed mutagenesis of these residues individually replaced with Ala showed that all resulting mutants were partially or completely inactivated (Figure 4C), confirming that these residues play important roles in the binding and release of substrate and structural stabilization.

Figure 4.

Table 3.

Substrate selectivity of GAGases

As mentioned above, GAGase III can degrade not only HA, CS and HS but also alginate, even though the activity on alginate is relatively low (about 1/50 of the activity on HA). To investigate the reason why this enzyme is capable to act on alginate, its structure was predicted using Robetta, and aligned with GAGase II and GAGase VII. Results showed that GAGase III has a unique His¹⁸⁸ residue in the catalytic cavity of GAGase III, which is conserved in alginate lyases from PL17 (Park, et al.,2014), PL39 (Ji, et al.,2019), PL38 (Ronne, et al.,2023), PL5 (Pandey, et al.,2021) and PL15 (Ochiai, et al.,2010) families, while other GAGases have a conserved asparagine residue such as the Asn¹⁹¹ of GAGase II and the Asn¹⁸⁶ of GAGase VII in the corresponding position (Figure 5A-B). To verify the function of this histidine, His¹⁸⁸ was replaced with alanine and asparagine, respectively, and the results of substrate-specificity analysis showed that the alginate-degrading activity of GAGase III-H188A and H188N was completely abolished but only partially affect its GAG-degrading activity, indicating that the His¹⁸⁸ residue is essential for the alginate-degrading activity of GAGase III (Figure 5C-D). However, when the asparagine residue at the corresponding position of other GAGases was muted to histidine residue the obtained mutants did not exhibit any alginate-degrading activity (Supplemental Figure S2), suggesting that other structural factors that are vital to the alginate-degrading activity of GAGase II remains to determine excepting for the His¹⁸⁸ residue.

Figure 5.

GAGases show preference for some specific sulfation patterns (Wei, et al.,2024). For example, GAGases prefer to degrade CS domains composed of non-/6-O-sulfated but not 4-O-sulfated GalNAc residues and D-GlcA residues. To further investigate the reason why these enzymes have such structural preference for substrates, we tried to prepare the co-crystal of GAGase II or VII with various structure-defined oligosaccharide substrates, but ultimately failed. Therefore, we used molecular docking analysis to preliminarily explore the possible reasons why GAGases have substrate structural preference. To this end, docking assays of GAGase II with a nonsulfated CS Hexa (GlcUAβ1–3GalNAc)₃ (PDB code: 2KQO) and a tri-4-O-sulfated CS Hexa (GlcUAβ1–3GalNAc(4S))₃ (PDB code: 1C4S) (Sattelle, et al.,2010) were carried out, respectively. In the docked model, the monosaccharide residues of the substrate were named “+1” and “+2” toward reducing end, and “-1” and “-2” toward non-reducing end from the β1-4 cleavage site following the nomenclature (Davies, et al.,1997). The results show that for the nonsulfated CS Hexa, the GlcUA residue at the “+1” subsite is closely approached by the Tyr²⁴⁶ residue as a Brønsted base (Figure 6 left), which facilitates the uptake of the C5 proton of GlcUA by Tyr²⁴⁶, and in the case of tri-4-O-sulfated CS Hexa, the 4-O-sulfate group in GalNAc residue at the “+2” subsite interacts with the neutralizer Asn¹⁹² and the catalytic residue Tyr²⁴⁶ via hydrogen bonds etc., the C5 proton of GlcUA at the “+1” subsite is away from the key catalytic residue Tyr²⁴⁶ (Figure 6 right, Supplemental Figure S3), which hinders the degradation of CS with such sulfation pattern, as we detected in the biochemical analysis above. Of course, the results from docking assay need to be further confirmed by more structural and biochemical evidences.

Figure 6.

Catalytic mechanism of GAGases

In summary, taking GAGase II as representative, the catalytic process of GAGases is proposed as follows: the negatively charged GAG substrate binds to several positively charged residues, such as Arg⁸⁷, Arg⁸⁸, Arg⁹⁴, His¹³⁶, Asn¹⁹¹ and Asn¹⁹², near the catalytic triad residues in the catalytic cavity; the Asn¹⁹¹ and Asn¹⁹² residues interact with the carboxyl group of GlcUA at the +1 subsite to reduce the pKa and promote the protonation of the C5 proton; His⁴⁰¹ works as a proton receptor (Brønsted base) to abstract protons from the C5 position of GlcUA and create an enolate tautomeric; the protonated Tyr²⁴⁶ (Brønsted acid) provides a proton to the C-4–O-1 glycosidic bond between the “-1” and “+1” subsite; the glycosidic bond is cleaved as an electron transfer occurs from the carboxyl group to form a double bond between C4 and C5 of GlcUA; and a new reducing end and a new unsaturated nonreducing end are recreated; moreover, the degradation of polyM by GAGase III requires the involvement of His¹⁸⁸ residue, which may involve in the neutralization of carboxyl groups in M-blocks in alginate (Figure 7).

Figure 7.

Discussion

GAG lyases derived from microorganisms, as GAG-specific degrading enzymes, are classified into 13 PL families in the CAZy database (Drula, et al.,2022). They are not only critical for microorganisms to degrade and utilize GAGs as carbon sources, but also become useful tools for structural and functional studies of GAGs. In contrast, alginate lyases, as alginate-specific degrading enzymes, are classified into 15 PL families in the CAZy database (Drula, et al.,2022). They are essential for the metabolism of alginate in microorganisms and algae as well as lower marine animals, and have completely different substrate specificities than GAG lyases. Due to the similarity between GAG lyases and alginate lyases in folds, GAG lyases are thought to have originated from the divergent evolution of alginate lyases (Garron and Cygler,2014), but enzymes that exhibit transitional features in function and structure, which would provide direct evidence to support this view, are lacking. On the basis of previous functional studies (Wei, et al.,2024), structural studies of GAGases with HA, CS, HS and even alginate-degrading activities should further provide substantial evidence for the evolution of GAG lyases.

Prior to this study, no member of the PL35 family had been structurally characterized. Here, the structures of GAGase II and VII were determined at resolutions of 1.9 Å and 2.4 Å, respectively, and both exhibit a highly homologous structure of “N-terminal (α/α)₇ toroid and C-terminal two-layer antiparallel β-sheet”, which is also the fold adopted by most structurally known GAG lyases except CSase B in PL6 (Huang, et al.,1999), Hepase I in PL13 (Han, et al.,2009) and hyaluronan lyase in PL16 (Martinez-Fleites, et al.,2009). Structural alignment has shown that GAGases share considerable degrees of homology with all GAG/alginate lyases with the (α/α)_n toroid and antiparallel β-sheet fold in PL8 (Fethiere, et al.,1999), PL12 (Hashimoto, et al.,2014), PL15 (Ochiai, et al.,2010; Zhang, et al.,2021), PL17 (Park, et al.,2014), PL21 (Shaya, et al.,2006), PL23 (Sugiura, et al.,2011) and PL39 (Ji, et al.,2019) families, especially the key catalytic triplet residues (Tyr, His and Asn) in their active centers, which are highly conserved; thus, they may share a similar Brønsted base/acid catalytic mechanism (Garron and Cygler,2010; Garron and Cygler,2014). Notably, GAGases are more structurally homologous to alginate lyases from PL15 (Ochiai, et al.,2010), PL17 (Park, et al.,2014) and PL39 (Ji, et al.,2019) family than to various GAG lyases, which further structurally supports the speculation that GAGases originated from alginate lyases.

As mentioned previously, many different GAG/alginate lyases share similar catalytic center sites and the same catalytic machinery (Garron and Cygler,2010; Garron and Cygler,2014), but their differences in substrate selectivity and endo-/exolytic manner are mainly attributed to subtle differences in the substrate-binding sites. The recognition of appropriate substrates by PLs mainly depends on the interaction between the positively charged side chains of basic amino acid residues in their substrate-binding sites and the negatively charged carboxyl groups of HexUA residues in substrates. Moreover, compared to the structurally and biochemically characterized members of the PL8, PL12, PL15, PL21 and PL39 families, PL35 family proteins possess a more open and shorter catalytic cavity; thus, PL35 family proteins can more easily accommodate various substrates with different structures but may exhibit a weaker ability to bind and degrade specific substrates. The activities of GAGases towards various GAGs and alginate are much lower than those of GAG or alginate-specific lyases.

In addition, GAGases tend to act on GlcUA-containing HA, CS and HS but not IdoUA-containing DS and Hep, which should result from the absence of a His residue functioning as an anti-base in the catalytic cavity (Garron and Cygler,2010; Shaya, et al.,2008; Shaya, et al.,2006). The His residue as an anti-base can substitute Tyr to absorb the C5 proton in the IdoUA and ensure that β-elimination is carried out successfully, which is detected in chondroitin sulfate ABC lyase (2Q1F) from PL8 family (His⁴⁵⁴) (Shaya, et al.,2008) and heparinase II (2FUQ) from PL21 family (His²⁰²) (Shaya, et al.,2006), and endows these enzymes with a bifunctional activity against GlcUA- and IdoUA-containing GAGs. Notably, GAGase III, the only identified GAGase with alginate-degrading activity, has a unique His¹⁸⁸ residue also conversed in various alginate lyases from PL5, PL15, PL17, PL38 and PL39 families. Mutation assay showed that the His¹⁸⁸ is a key residue for the alginate-degrading activity of GAGase III. However, when a conversed asparagine residue at the corresponding position of other GAGases was replaced by histidine, the muted GAGases still could not act on alginate, indicating that some other unidentified structural factors are also involved in the alginate degradation of GAGase III. Overall, while the substrate-degrading features of GAGases can be preliminarily elucidated structurally, more direct evidence remains to be obtained by detailed structural and biochemical approaches.

In addition, the identification of GAGases may help clarify an inappropriate naming issue regarding the functional module “Hepar_II_III superfamily”. This which functional module is commonly found in not only GAG lyases but also alginate lyases with an “N-terminal (α/α)_n toroid + C-terminal β-sandwich” structure. Interestingly, however, a large number of identified alginate lyases with the functional module “Hepar_II_III superfamily” do not have the ability to degrade Hep/HS. Historically, due to the importance of Hep as an anticoagulant in clinical application, Hep/HS-related studies including heparinases have received much higher attention than alginate-related studies including alginate lyases, which may be the reason why some sequences in alginate lyases that are homologous to heparinase are annotated named the “Hepar_II_III superfamily” module. From an evolutionary perspective, this “Hepar_II_III superfamily” module might be a conserved functional domain in the divergent evolution process from alginate lyase to GAG lyase. Therefore, it might be more reasonable and accurate for this functional domain to be named as the “alginate lyase superfamily” module, and the presence of this functional module in GAGases with evolutionary transitional characteristics may well support this view.

In conclusion, the first determination of the structure of GAGases not only facilitates the elucidation of the catalytic mechanism of PL35 family enzymes, especially the substrate selection mechanism of GAGases, but also provides potential evidence for the hypothesis that GAG lyase evolved from alginate lyase.

Materials and methods

Materials

HA from Streptococcus equi, CS-C from shark cartilage, alginate sodium from brown algae, polyethylene glycol (PEG) 6000, PEG 400, calcium chloride (CaCl₂), tris-(hydroxymethyl) aminomethane (Tris) and hydroxyethyl piperazine ethanesulfonic acid (HEPES) were purchased from Sigma‒Aldrich. Lysine, phenylalanine, threonine, leucine, isoleucine, valine and L-selenomethionine were purchased from Solarbio Co.,Ltd. (Beijing, China).

Heterologous expression and purification of GAGases and GAGase II selenomethionine derivant

The cloning and induced expression of PL35 family GAGases were described in previous study (Wei, et al.,2024). Briefly, the synthesized expression plasmid pET-30a-GAGases were transformed into E. coli BL21 (DE3) cells and culture in LB broth at 37 °C. Recombinant enzymes with a 6His-tag were expressed at 16 °C for 16 h by inducing with isopropyl 1-thio-β-D-galactopyranosid (IPTG) (final concentration of 0.05 mM) at the cell density of OD₆₀₀=0.6–0.8. Furthermore, to obtain the Se-GAGase II, cultured cells harboring the pET-30a-GAGase II were centrifuged at 3,700 g at 4 °C for 5 min, washed with M9 culture medium (17.2 g/L Na₂HPO₄, 3 g/L KH₂PO₄, 2.5 g/L NaCl and 5 g/L NH₄Cl), transferred into M9 culture medium supplemented with 1 mM MgSO₄, 0.1 mM CaCl₂, 0.4% (w/v) glucose and cultured at 37 °C until the OD₆₀₀ reached 0.6. Then, the culture medium was cooled to 22 °C. Additional essential amino acids, including lysine (100 mg/L), phenylalanine (100 mg/L), threonine (100 mg/L), leucine (50 mg/L), isoleucine (50 mg/L), valine (50 mg/L), and L-selenomethionine (L-SeMet) (50 mg/L), were added and incubated at 22 °C for 15 min. The proteins contained L-SeMet were also induced and expressed at 16 °C for 16 h by supplementing with 5 mM IPTG.

After further induced cultivation, cells were harvested, resuspended and disrupted by sonication, the recombinant proteins in the supernatant of cell lysates are primarily purified by loading on a nickel affinity column, washing with buffer A containing 10 mM imidazole to remove impurities, and finally eluting using buffer A containing 250 mM imidazole. After nickel affinity chromatography, the elutes of GAGase II and GAGase VII were diluted and loaded on a Q-Sepharose FF column (GE Healthcare) eluted with a gradient concentration of NaCl from 0 to 1 M, and the target proteins were desalted and concentrated through ultrafiltration and further sub-fractionated by gel filtration on a Superdex G-200 column (GE Healthcare) for crystal culture. The concentration of each protein was determined using BCA Protein Assay Kit with bovine serum albumin (BSA) as reference protein (Cwbio, Shanghai).

Activity assay of GAGase II, III, VII and their mutants toward various GAGs and alginate

To evaluate the activity of GAGase II, III, VII and their mutants, CS-C or alginate (30 μg) was treated with each wild-type enzyme or its mutant (6 μg) in the optimal buffer (50 mM Tris-HCl, pH 7.0) for 12 h at 40 °C. After inactivated by boiling for 10 min, the products were analyzed by gel filtration HPLC using a Superdex^TM Peptide 10/300 GL column eluted with 0.20 M NH₄HCO₃ at a flow rate of 0.4 ml/min and monitored at 232 nm using a UV detector, and online analysis was conducted using LCsolution version 1.25.

To evaluate the activity of GAGase II and GAGase III mutants from the site-directed mutagenesis of some potential substrate-binding residues, HA, CS-C, or HS (150 μg) was treated with each wild-type enzyme or its mutant (6 μg) in the optimal buffer (50 mM Tris-HCl, pH 7.0) for 1 h at 40 °C, and the absorbance of each resultant were determined at 232 nm using a UV spectrophotometer.

Crystallization, X-ray diffraction and data collection of GAGase II, GAGase VII and Se-Met-GAGase II

The purified GAGase II (20 mg/ml) was crystallized in 0.1 M HEPES (pH 7.0), 12% (w/v) PEG 6000 and 0.2 M CaCl₂. The purified Se-Met-GAGase II (5 mg/ml) was crystallized in 0.1 M Tris (pH 7.5), 20% (w/v) PEG 6000 and 0.2 M CaCl₂. The purified GAGase VII (10 mg/ml) was crystallized in 0.1 M HEPES (pH 7.0) and 36% (v/v) PEG 400. The crystallization of all the above proteins was carried out at 18 °C using hanging drop vapor diffusion method in 24-well plates. Crystals were harvested using nylon loops and soaked in cryoprotectant consisting of 20% glycerol and 80% mother liquor. X-ray diffraction data were collected at Shanghai Synchrotron Radiation Facility (SSRF) BL18U1 beamline. Diffraction datasets were processed using XDS (Kabsch,2010). Data collection statistics are summarized in Table 1.

Structure determination and refinement

The structure of GAGase II was solved by selenium single-wavelength anomalous dispersion (Se-SAD) using AutoSol from Phenix suite (Adams, et al.,2011). The structure of GAGase VII was determined by molecular replacement using Phaser (McCoy, et al.,2007) with the structure of GAGase II as the search model. Initial model building was carried out using AutoBuild from Phenix suite. Then several rounds of refinement and manual building were performed alternately using Phenix.Refine and Coot (Emsley and Cowtan,2004), respectively. The ion type was confirmed by inductively coupled plasma-mass spectrometry (ICP-MS) analysis. For each sample, 10 ml of GAGase II or GAGase VII (10 mg) was mixed with 10 ml of nitric acid in equal proportions. The mixture was then placed in a metal bath at 120 °C and heated until clarified. Subsequently, the clarified mixture was filtered through a 0.22 µm membrane and analyzed using ICP-MS (NexlON, PerkinElmer). The structures of other GAGases were predicted using Robetta (Baek, et al.,2021). Structure alignments were performed using ChimeraX with point accepted mutation (PAM)-120 mtrix (Pettersen, et al.,2021). All of the structure illustrations were generated using ChimeraX (https://www.cgl.ucsf.edu/chimerax/) and Pymol (https://www.pymol.org/2/). A structure-based homology search for GAGase II and VII were performed using the DALI server (http://ekhidna2.biocenter.helsinki.fi/dali/) (Holm and Rosenstrom,2010). The classification of identified enzyme were confirmed based on the database of carbohydrate-active enzymes (CAZy) (http://www.cazy.org/) (Drula, et al.,2022) and the structures of identified PLs were download from the RCSB Protein Data Bank (PDB) (https://www.rcsb.org/) (Berman, et al.,2000).

Site-directed mutagenesis of GAGases

Some potentially important residues located in and around the active center were individually replaced with alanine (Ala) by using the Fast Mutagenesis Kit V2 from Vazyme Biotech Co., Ltd (Nanjing, China) and these mutants were expressed and examined for activity against various substrates (HA, CS-C, HS, and alginate) to confirm their crucial role in enzyme catalysis. The primer pairs of these mutants are shown in the Table S2. The activities of the mutants were evaluated as described above.

Molecular docking

Tertiary structure-defined GAG or alginate hexasaccharide was used to dock into the active cavity of GAGase II by using the Autodock Vina program (Eberhardt, et al.,2021; Trott and Olson,2010). Docking was performed on a search space of 15.00×18.19×19.53 Å covering the active centre (x, y, z = −2.60, −12.57, 34.13). The binding energies using non-sulfated and 4-O-trisulfated hexasaccharides are calculated as −5.874 and −5.163 kcal/mol, respectively.

Data availability

All data supporting the findings of this study are available within the paper (and supporting information files). All relevant data generated during this study or analysed in this published article are available from the corresponding author on reasonable request. The atomic coordinates and structure factors of the structures in this study have been deposited in the Protein Data Bank (PDB codes: 8KHV and 8KHW). The sequences of GAGase II (GenBank: SOD82962.1) and GAGase VII (GenBank: EDV05210.1) have already existed in NCBI database.

Acknowledgements

This work was supported by the National Key R&D Program of China (No. 2021YFC2103100), National Natural Science Foundation of China (Nos. 31971201, 32330001, 31570071, 31800665), Major Scientific and Technology Innovation Project (MSTIP) of Shandong Province (No. 2019JZZY010817), Natural Science Foundation of Shandong Province (No. ZR2023MC017) and the Program of Shandong for Taishan Scholars (tspd20181203).

Author contributions

Wei Lin, Investigation, Validation, Visualization, Writing-Original Draft; Cao Hai-Yan, Investigation, Validation, Visualization, Writing-Original Draft; Zou Ruyi, Investigation, Validation; Du Min, Investigation, Validation; Zhang Qingdong, Investigation; Lu Danrong, Resources; Xu Xiangyu, Investigation; Xu Yingying, Investigation; Wang Wenshuang, Investigation; Chen Xiu-Lan, Investigation; Zhang Yu-Zhong, Conceptualization, Methodology, Writing-Reviewing and Editing; Li Fuchuan, Conceptualization, Methodology, Writing-Reviewing and Editing.