1. Introduction
Xyloglucan endotransglycosylases (XETs) are fundamental glycosidic bond-forming biocatalysts that operate during the biogenesis of plant cell walls (CWs) and fulfill the structural and mechanistic roles in CW formation and remodelling. The action of XETs is irreplaceable in physiological cellular processes that underlie CW expansion and reconstruction [
1,
2,
3,
4,
5]. The discovery of XETs from various plant sources was reported independently by three research groups [
6,
7,
8], and since then, significant knowledge of the structure, function, biochemistry, biophysics, and evolutionary relationships of XETs has been acquired [
9,
10,
11].
According to the Enzyme Commission (EC), XETs are classified among glycosyl transferases, with their systematic name xyloglucan:xyloglucosyl transferases (EC 2.4.1.207) [
12]. This EC description recognises xyloglucan (XG) as a donor substrate and XG or XG-derived oligosaccharides (XG-OS) as acceptor substrates, which are utilised in homo-transglycosylation reactions. Concerning the focus of this work, it is important to note that the definition of XETs by EC still includes a comment ‘does not use cello-oligosaccharides as either donor or acceptor’. As it will be shown in this work, and in light of the current knowledge, this remark is obsolete [
13,
14,
15,
16,
17,
18,
19].
The Carbohydrate-Active enZYmes Database (CAZy) [
20] and CAZypedia [
21] clarify the issue of the XET nomenclature and classify entries based on their tertiary structures, substrate specificity, phylogenomic relationships, and evolutionary history. According to CAZy, XETs are members of the glycoside hydrolase family 16 (GH16), while the transferase groups specifically contain the enzymes utilising ‘activated’ sugar phosphates as the glycosyl donors. The GH16 family is subdivided into 23 subfamilies according to the features in their tertiary structures [
22]. XET enzymes are allocated together with xyloglucan endohydrolases (XEHs, EC 3.2.1.151) into the GH16_20 subfamily [
23]. The structure-based studies of these enzymes [
24] showed their close similarity in tertiary structures and provided evidence for the evolution of XEHs from XETs [
23,
25,
26]. These findings were supported by a cross-genome survey of the evolutionary origin of endoglucanases within the GH16 family and XET/XEH enzymes [
27]. The latest phylogenomic and comparative structural analyses [
10] derived the origin of the GH16_20 subfamily from the non-plant alphaproteobacteria ExoK biocatalysts involved in the loosening of biofilms in icy environments [
28], rather than from the previously suggested bacterial lichenases [
29,
30], which are classified in the GH16_21 subfamily [
22]. Predicted intermediates between ExoKs and XETs are the charophycean EG16-2 enzymes, which originated due to the horizontal gene transfer event during the Cryogenian geological period [
31,
32,
33].
In addition to homo-transglycosylation reactions catalysed by XETs, a new subtype of transglycosylation reactions was identified in 2006, when XETs from a crude nasturtium (
Tropaeolum majus) extract were found to recognise poly- and oligosaccharide substrates other than XG-derived [
13]. These reactions were in 2007 demonstrated in near-homogenous barley HvXET5 [
14], and later in other plant enzymes [
11,
34] and crude extracts [
9,
16,
17]. Currently, there are only a few XETs with defined primary structures for which enzyme activities on substrates other than XG were described [
34]. In principle, the reactions catalysed by XETs are subdivided into three types, as described below.
(i) Homo-transglycosylation reactions with XG-derived substrates: The most appropriate representative of strictly specific XETs is the poplar PttXET16A [
25,
35], which exclusively recognises XG-derived substrates and is the best characterised XET due to its tertiary structure derived from X-ray crystallography. An enzyme with analogous substrate specificity, the
Pinus radiata PrXTH1 [
36], showed in-silico a weak interaction with cello-oligosaccharides (Cello-OS) as substrate donors [
37]. Nevertheless, both PttXET16A and PrXTH1, according to phylogenomic analyses of the GH16 family [
34], clustered with the HvXET5 enzyme.
(ii) Hetero-transglycosylation reactions with cellulose-derived or (1,4;1,3)-β-d-glucan (mixed-linkage glucan; MLG)-derived substrates: As determined experimentally, HvXET5 in a near-homogenous form in-vitro catalysed the transfer of hydroxyethyl cellulose (HEC) fragments on XG-derived oligosaccharides (XG-OS; 44% efficiency) or XG fragments on Cello-OS. Here, the reaction rates were comparable to those of XG with XG-OS, while those with the MLG donor substrate were low (efficacy 0.2%) [
14]. The formation of hybrid products was confirmed by mass spectrometry.
The next near-homogenous XET, AtXTH3 from
Arabidopsis thaliana L. Heynh, recognised cellulose as the donor substrate [
38] and cellulose, Cello-OS, and XG-OS as acceptors, in addition to XG-derived substrates. Moreover, this enzyme formed cello-oligomers from the aminopyridyl derivative of cellohexaose with higher degrees of polymerisation (DPs) than the original substrate, as confirmed by mass spectrometry. In the absence of other substrates, insoluble cellulose-like material was formed. Notably, barley HvXET5 [
14] and AtXTH3 [
38] clustered within the same phylogenetic XTH I clade, as presumably XG-specific PttXET16A and PrXTH1 [
25,
36], although they segregated to different sub-clades [
34].
Further XETs with a defined primary structure capable of transferring besides XG, also cellulose, or MLG fragments were described in the
Equisetum fluviatile L. and were named EfXTH-A, EfXTH-H, and EfXTH-I [
39]. The homo-transglycosylation activity (with the XG/XG-OS pair) was a dominant reaction for all acidic isoforms, whereas the efficiencies of transfers with cellulose and MLG fragments differed. Similarly, as HvXET5 [
14], EfXTH-A showed a comparable transfer of MLG fragments to XG-OS (efficiency 0.2–0.3%), while the transfer with cellulose was incomparably higher with the barley HvXET5 isoform [
14]. The hetero-transglycosylation activities of both EfXTH-H and EfXTH-I were equivalent to those of the MLG substrates, and the activity with the cellulose donor was around one order of magnitude higher than that of EfXTH-A [
39]. On the contrary, other transglycosylating enzymes from
Equisetum, the hetero-transglucanase (HTG) and MLG:xyloglucan endo-transglucosylase, preferred cellulose and MLG substrates with XG-OS as respective donors and acceptor substrates [
40,
41]. The predicted function of the latter enzyme was to reconstruct hemicelluloses in horsetail shoots [
40]. Similarly, as EfXTHs, HTG was also subjected to molecular modelling [
41]. Regardless of its donor specificity, HTG as a member of the GH16_20 subfamily, clustered within the XTH II clade [
10,
34]. It was suggested that XETs from the XTH II clade evolved from the XTH I clade catalysts [
10]. The representatives of this clade first appeared in lycophytes, but HTG- and MLG:xyloglucan endotransglycosylase-like activities were also found in charophytic algae [
9,
42,
43]. The atomic structures of these XET enzymes have yet to be determined. Among others, HEC and Cello-OS substrates served as respective donors and acceptor substrates, also for partially purified XETs from parsley roots [
44] or for XETs isolated from parsley stems and leaves, and nasturtium stems, leaves, and roots [
17]. The efficiency of transglycosylation did not exceed 5% with the HEC/Cello-OS pair compared to the XG/XG-OS pair.
(iii) Hetero-transglycosylation reactions with acceptors other than XG-, cellulose-, or MLG-derived: Unlike XG substrate donors, HEC or carboxymethyl cellulose derivatives, and MLG substrates, i.e., donors derived from polysaccharides with a backbone made of Glc moieties connected mainly by (1,4)-β-glycosidic linkages, the structure of substrate acceptors differed significantly both in terms of the saccharide moieties and glycosidic linkages that interconnect them. The broad acceptor specificity of XETs isolated from nasturtium germinating seed extracts [
13] initiated the structural studies linked to the substrate specificity of the major
Tropaeolum majus TmXET6.3 isoform (named according to its isoelectric point of 6.3), that clustered in the XTH II clade [
17,
34]. Recombinant and near-homogenous TmXET6.3 did not utilise polysaccharides other than XG or HEC as donors [
17], but it was able to transfer their fragments to a whole spectrum of structurally different neutral acceptor substrates derived from cellulose (Cello-OS), MLG (MLG-OS), laminarin (La-OS), pustulan (Pu-OS), xylan (Xyl-OS), arabinoxylan (AraXyl-OS), arabinan (Ara-OS), arabinogalactan (AraGal-OS), mannan (Man-OS), glucomannan (GlcMan-OS), and galactomannan (GalMan-OS) [
17]. Reaction rates with acceptors varied in the following order: MLG-OS > Cello-OS > Pu-OS > AraXyl-OS > La-OS > Xyl-OS > GlcMan-OS > Ara-OS. Minimal activities were seen with AraGal-OS, Man-OS, and GalMan-OS. Other factors influencing the activity of TmXET6.3 were DPs of Cello-OS or the positions of (1,4)-β- and (1,3)-β-linkages in MLG-OS [
17]. TmXET6.3 could not catalyse the transfer of XG or HEC fragments on ionic (charged) acceptors.
As predicted from the alignments of TmXET6.3 with other XETs [
17], including barley HvXET3, HvXET4, and HvXET6 isoforms [
15,
45,
46], which showed the presence of residues identified in TmXET6.3 and were responsible for a broad acceptor specificity [
17], all these XETs were able to catalyse the transfer of XG or HEC fragments to a wide panel of neutral acceptors [
18]. As expected, there were only small differences in the hetero-transglycosylation efficacies between these isoforms, probably due to a residue variation in the acceptor binding sites. However, unlike TmXET6.3, the barley isoforms catalysed a novel type of hetero-transglycosylation reaction with negatively charged oligosaccharide acceptors, i.e., they catalysed the reaction between XG, cellulose, and the penta-galacturonide acceptor (fragment of a linear part of pectin) [
18].
In XET sequences, enzymes could have either the Q102/R116 or H102/Q116 residue combinations (numbering of residues according to PttXET16A). The first combination is dominant and considered to be ancestral [
10]. The Q108 residue of TmXET6.3 and the matching residues in non-specific barley HvXET isoforms corresponded to R116 in PttXET16A and are considered a signature residue for the XTH clade II [
23]. Similarly, the H94 residue in TmXET6.3 and barley isoforms corresponded to Q102 in PttXET16A and is regarded as an additional signature residue of this XTH clade [
10]. The shift from Q102/R116 (signature residues of the XTH clade I) to the H102/Q116 combination occurred at least five times during evolution and led to the convergent co-evolution of these residues, with the last event leading to the XTH clade II origin. Members of this clade, such as TmXET6.3 [
17] and EfHTG [
41,
47], have broad substrate specificity, which can be considered an evolutionary advantage induced by the co-evolution of residues binding different saccharides [
10].
The key contributions to the clarification of substrate specificity in XETs were made through experimental measurements combined with computational investigations and bioinformatics. This was possible due to the structural knowledge resulting from the atomic structure of PttXET16A [
35]. Among the most important tools are the descriptions of the recognition mechanisms of substrate-enzyme complexes that are important for enzyme design. One example of a computational approach is the exploration of the dimeric XG nonasaccharide binding using molecular dynamics (MD) simulations [
24], where one of XG nonasaccharides occupied the donor site creating a stable intermediate with an enzyme while the second XG nonasaccharide occupied the acceptor site. In both PttXET16A and TmNXG1 (which is XEH; EC 3.2.1.151), the Glc moiety of the nonreducing end of the XG nonasaccharide was located closest to the catalytic residues (which occupied the donor binding site) and altered its low-energy
4C
1 into the
1S
3 skew-boat conformation at the beginning of the MD simulation and maintained it [
24].
Further, the benefit of computational methods can be illustrated by the fact that the substrate promiscuity in XETs from
Poaceae was predicted by the molecular modelling of the GH16 family [
48] before the first experimental evidence of broad substrate specificity in XETs was obtained [
13,
14]. Later, the specifics of these XG-OS interactions in the acceptor binding sites of several barley isoforms were demonstrated computationally and through enzyme kinetics [
45]. The next valuable contribution was brought by the molecular modelling of HTG, where it revealed the residues that were responsible for the distinct substrate specificity of HTG [
41]. It was shown that P10 and S34 participated in the donor substrate binding while L245 bound the acceptor substrate. It was notable that in other XETs, P10 is substituted by tryptophan and S34 by glycine residues [
39]. Barley XET5 [
14] and AtXTH3 [
38], which exhibit high catalytic rates using cellulose as the donor, also contained in equivalent positions proline and serine residues as HTG, indicating the validity of this rationale. It was also suggested that the R246L mutation in HTG underlies the differences in Cello-OS binding [
39]. However, this has yet to be verified because TmXET6.3 [
17], EfXTHs [
39], and barley HvXET3, HvXET4, and HvXET6 isoforms [
18] catalysing transfers with Cello-OS have Arg in the equivalent positions, while AtXTH3 [
38] has the R246K variation.
An additional contribution of the joined efforts of computational chemistry and bioinformatics provided fundamental information on the residues responsible for the differences in the acceptor specificity of the XTH clade I and II enzymes [
17]. The signature residues mentioned above, specifically H94 and Q108, and certain lysine residues at the C-terminal end of TmXET6.3, such as K234, and K237, were identified [
17]. Further, the residues responsible for the differences in substrate specificity amongst the XTH clade II members were identified, e.g., in certain non-specific barley XET isoforms that transferred XG or cellulose fragments onto charged acceptors [
18], but not in others such as TmXET6.3 [
17]. The H75 and R110 residues in barley HvXET3 and HvXET4 were identified to be responsible for these novel acceptor substrate specificities. In both cases, the accuracy of theoretical findings was verified and confirmed by mutational analyses [
17,
18].
In this work, the acceptor substrate specificities of specific PttXET16A and non-specific TmXET6.3 were studied using computational chemistry tools. The complexes of enzymes with the donor XG heptaoligosaccharide (XG-OS7) and a variety of acceptor substrates were obtained by molecular docking followed by MD simulations and combined with binding free energy calculations. We found that the stabilities of enzyme-substrate complexes were broadly in agreement with the experimental activity assays. Here, the instability of certain acceptors in the active site of PttXET16A was observed, while all tested oligosaccharide substrates were stable in the acceptor binding site of TmXET6.3. These findings are situated in the context of reaction mechanisms of specific and non-specific XETs and their functional roles during biogenesis and re-structuring of plant CWs.