Biochemical characterization of xyloglucan galactosyltransferases MUR3 and XLT2 from Spirodela polyrhiza

SUMMARY

Glycosyltransferases (GTs) are the primary enzymes responsible for the biosynthesis of the complex polysaccharides in plant cell walls. Given the important role of GTs in plants, it is necessary to undertake their functional characterization to better understand plant cell wall synthesis pathways to develop improved feedstocks for efficient conversion into fuels and products to support the emerging bioeconomy. The GT47 family in plants represents a unique target for characterization due to the substantial diversity of donor and acceptor substrates observed within a single family. Here, we have carried out the biochemical characterization of MUR3 and XLT2 orthologs from the aquatic monocot Spirodela polyrhiza. Our findings support existing genetic and phylogenetic data classifying these enzymes as regio‐specific galactosyltransferases involved in xyloglucan (XyG) sidechain biosynthesis. In addition, we have identified novel characteristics for both enzymes, such as in vitro arabinopyranosyltransferase activity and distinctiveness in xyloglucan reducing end specificity.

Significance Statement

An in‐depth biochemical analysis of two prominent plant GT47 family members, MUR3 and XLT2, supports existing in planta data with biochemical evidence and reveals biochemical insights into the function of both enzymes.

INTRODUCTION

Duckweed is an aquatic monocot in the Lemnaceae family that is divided across five genera: Landoltia, Lemna, Spirodela, Wolffiella, and Wolffia (Bog et al., 2019). Duckweeds are known for their designation as one of the fastest‐growing plants, with a population doubling time of as little as 48 h in ideal conditions (Leng, 1999). A defining feature of all plant cells is their encasement by a cell wall composed of polysaccharides, proteins, and polyphenols. Fractionation and composition analysis of Spirodela polyrhiza biomass revealed a cell wall highly enriched in matrix polysaccharides with a composition of 13% cellulose, 28% hemicellulose, and 59% pectin (Pagliuso et al., 2022). Additionally, S. polyrhiza contains a relatively small and non‐redundant genome whose size is on par with that of Arabidopsis thaliana and is smaller than that of other energy crops (Acosta et al., 2021), comprising around 158 Mbp (Wang et al., 2014).

Plant cell wall polysaccharides form an interconnected network providing structural support, protection, and flexibility to plants (Somerville et al., 2004; Voiniciuc et al., 2018; Zhang et al., 2023). To gain insight into the biosynthesis, structural diversity, and specific functions of plant cell wall polymers, it is essential to identify and functionally characterize the enzymes responsible for their synthesis. With this goal in mind, we have set out to characterize duckweed‐specific cell wall biosynthetic pathways due to this species' appeal as a next‐generation energy crop and its utility as a rising model organism for investigating plant cell wall biosynthesis. The primary enzymes involved in plant cell wall biosynthesis are glycosyltransferases (GTs), carbohydrate‐active enzymes (Drula et al., 2022) (CAZymes) that catalyze the formation of glycosidic bonds by transferring a sugar from an NDP‐sugar donor onto an acceptor substrate (Sinnott, 1990). GT‐encoding genes are highly abundant in plants, often representing 2–4% of a plant's genome (Zhang et al., 2023). Functional knowledge of these diverse gene families largely originates from predictive phylogenetic data or in vivo experimentation. Despite the importance of these enzymes in synthesizing the complex carbohydrate structures that make up the majority of the cell wall, very few polysaccharide‐synthesizing GTs from any species have been biochemically characterized. One GT family of particular interest is the CAZy GT47 family which is highly expanded in plants and is proposed to harbor multiple donor and acceptor substrate specificities within a single family (Tehrani & Corulli, 2024; Zhang et al., 2023). Plant genomes contain at least 30 or more diverse GT coding sequences from the GT47 family and play a crucial role in the biosynthesis of hemicellulose (xyloglucan, xylan), mannan, and pectin (Zhang et al., 2023). Xyloglucan is found in all plant cell walls, including charophycean green algae, the closest land plant relatives (Mikkelsen et al., 2021). This cell wall polymer is often used to develop methods for biochemical analysis, as workflows to easily study its structure have made xyloglucan into an analytically tractable model saccharide (Julian & Zabotina, 2022). This is largely driven by the identification of xyloglucan‐specific endo‐β‐1,4‐glucanohydrolase (XEG), which has been used extensively for the fragmentation of xyloglucan into structurally distinct oligosaccharides that are amenable for isolation and analysis (Pauly et al., 1999). Inspired by a large number of structural studies, a nomenclature system utilizing a one letter designation for xyloglucan sidechain structures was introduced by the cell wall community (Fry et al., 1993), and has been expanded to include new sidechains as they are discovered (Tuomivaara et al., 2015). In this system, G represents an unsubstituted glucose (Glc) residue of the β1,4‐Glc‐linked polymer backbone, X represents a backbone Glc residue that is substituted with an α‐xylose (Xyl) at the O‐6 position, and L represents a X sidechain further substituted with a galactose resulting in a (Gal)‐β1,2‐Xyl‐α1,6‐ disaccharide. Finally, F represents an L sidechain further substituted with a fucosyl residue: (Fuc)‐α1,2‐Gal‐β1,2‐Xyl‐α1,6‐. When the single letter code is used, a XXXG structure refers to a backbone consisting of four β1,4‐Glc residues where the first three Glc are Xyl‐modified followed by one unsubstituted Glc residue. X, L, and F sidechain structures are the most common in xyloglucans from eudicots and gymnosperms. For example, the xyloglucan in Arabidopsis features four recurring XXXG‐type core subunits (S1s): XXXG, XXFG, XXLG, and XLFG (Fry et al., 1993). Here, we will refer to a linear assembly of two S1 motifs as an S2 motif and degalactosylated S2 motifs as S2 ΔGal XyG (Figure 1). Likewise, S3 and S4 subunits are composed of linear assemblies of three and four S1 motifs, respectively.

Figure 1.

Schematic of the hypothesized xyloglucan (XyG) sidechain modifications synthesized by MUR3 and XLT2.

(a) MUR3 and XLT2 facilitate the addition of β‐d‐Gal onto XyG to form the β‐d‐Gal‐(1,2)‐α‐d‐Xyl (“L”) sidechains motifs. Both enzymes differ in regiospecificity, with MUR3 and XLT2 catalyzing the addition of galactose to the 3rd and 2nd xylose from the non‐reducing end of the XyG “S1” motif, respectively.

(b) Schematic of XyG XXXG‐type core subunits XXXG (S1 ΔGal XyG) and XXXGXXXG (S2 ΔGal XyG).

(c) XyG sidechain motifs with corresponding single letter sidechain nomenclature.

(d) Standardized symbol used for graphical representation glycan structures based on the Symbol Nomenclature for Glycans (SNFG) standardized methodology.

In a recent study of monocotyledon xyloglucans of Araceae species, which includes duckweeds in the subfamily Lemnoideae, S. polyrhiza was found to have a XXXG‐type core motif similar to that in Arabidopsis (Hsiung et al., 2023). Furthermore, the xyloglucan sidechains of S. polyrhiza contain both arabinose (15%, XXSG/XXDG) and galactose (3%, XLXG/XXLG; 13%, XXFG+XLFG). These include S and D sidechains where xylose is substituted with arabinofuranose (Araf) or arabinopyranose (Arap), respectively (Hsiung et al., 2023). Combining mutant studies and functional complementation analyses in Arabidopsis led to the identification of genes that encode enzymes in the GT47 family that transfer Gal residues to X sidechains: MURUS3 (MUR3) (Madson et al., 2003) and XYLOGLUCAN L‐SIDE CHAIN GALACTOSYLTRANSFERASE POSITION 2 (XLT2) (Jensen et al., 2012). They are hypothesized to act as regiospecific GTs involved in catalyzing the addition of β‐d‐Gal to the Xyl residues on the third and second β1,4‐Glc units in a XXXG core subunit, forming XXLG and XLXG, respectively (Figure 1); however, their biochemical activities have never been confirmed in vitro, which is an important step in clarifying enzyme activity outside of factors found in vivo such as the availability of donor substrates, cofactors, other regulatory proteins, inhibitors, activators, and the cellular environment. Here, we present the biochemical characterization of MUR3 and XLT2 orthologs from S. polyrhiza. We show that both proteins are predominantly galactosyltransferases involved in the synthesis of Gal‐β1,2‐Xyl‐α1,6 sidechains on xyloglucan, forming XXLG and XLXG, respectively. However, in the in vitro enzymatic assays, both enzymes show a trace amount of arabinopyranosyltransferase activity, synthesizing Arap‐α1,2‐Xyl‐α1,6‐sidechains on xyloglucan, forming XXDG or XDXG motifs. Detailed biochemical analyses have uncovered novel characteristics of these enzymes, such as their requirement for larger xyloglucan oligosaccharides (>DP8) as acceptors with further subunit selectivity.

RESULTS

Heterologous expression of SpMUR3 and SpXLT2

The full‐length coding sequences of SpMUR3 (Spipo1G0014600) and SpXLT2 (Spipo17G0019700) were derived from the S. polyrhiza (v2) genome via the Phytozome 13 database. The expressed fusion proteins consisted of a signal sequence, an 8X‐His tag, AviTag, “superfolder” GFP, the 7‐amino acid recognition sequence of the tobacco etch virus (TEV) protease followed by the indicated catalytic domain regions of SpMUR3 (amino acids 36–544) or SpXLT2 (amino acids 86–537). SpMUR3 and SpXLT2 were expressed as soluble secreted fusion proteins by transient transfection of suspension culture HEK293‐F cells and purified using immobilized metal affinity chromatography (IMAC) as described for plant GTs (Prabhakar et al., 2020). Sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) analysis indicated that the purified proteins are the expected size (Figure S1). Size exclusion chromatography (SEC) coupled with Small and Wide‐Angle X‐ray Scattering (SAXS/WAXS) was used to understand SpMUR3 structure, protein aggregation, oligomeric state, and stability, as it is a good target for structural study. The monodisperse SEC peak indicates SpMUR3 is stable. From one frame of the peak, we generated an ab initio structure of SpMUR3, which had a Χ² value of 1.116. We overlaid a monomeric SpMUR3 AlphaFold model with the SAXS SpMUR3 model using ChimeraX, which fit well, apart from the flexible N‐terminus (Figure S1). By contrast, SpXLT2 was determined to be inadequate for structural characterization due to its low expression and loss of function after SEC purification (data not shown).

SpMUR3 and SpXLT2 donor and acceptor substrate utilization

To biochemically determine the donor and acceptor substrate specificities of SpMUR3 and SpXLT2, both GTs were screened using a donor substrate library containing various UDP‐sugars to examine hydrolytic activity (defined here as the tendency for a GT to transfer a sugar from a UDP‐sugar to water in the absence of an acceptor, producing UDP as a by‐product) followed by subsequent analysis using the preferred donor (UDP‐Gal) and a library of plant‐specific acceptor substrates (Figure 2). Substrate screening was accomplished by quantifying UDP, a by‐product of GT activity following successful monosaccharide transfer from an activated nucleotide sugar donor (Sheikh et al., 2017). Hydrolytic and transferase activities of SpMUR3 and SpXLT2 were determined at different pH ranges and temperatures. The optimal activity for both proteins was found to be pH 7 (Figure S2) and 30°C (Figure S3). These conditions were subsequently used for the remaining assays. The NDP‐sugar donor screen with SpMUR3 showed significant hydrolytic activity in the presence of UDP‐β‐l‐arabinopyranose (UDP‐Arap) and UDP‐β‐l‐arabinofuranose (UDP‐Araf). By contrast, in the absence of an acceptor, SpXLT2 demonstrated little to no hydrolytic activity for any NDP‐sugar donor (Figure 2a). Our experience with plant GT analysis has suggested that many of these enzymes do not hydrolyze the preferred donor in the absence of a suitable acceptor.

Figure 2.

The relative activity of SpMUR3 (blue) and SpXLT2 (orange) in the presence of various UDP‐sugar donor and acceptor substrates.

Enzymatic activity was measured by quantifying the UDP by‐product after enzymatic transfer of a UDP‐sugar donor to either water (hydrolysis) or an acceptor substrate.

(a) The hydrolytic activity of SpMUR3 and SpXLT2 in the presence of a UDP‐bound sugar donor library.

(b) Transferase activity of SpMUR3 and SpXLT2 in the presence of various acceptor substrates using UDP‐Gal as a donor.

(c, d) Transferase activity of (c) SpMUR3 and (d) SpXLT2 in the presence of ΔGal XyG polymer with the donor substrates UDP‐Gal, UDP‐Arap, or UDP‐Araf.

(e, f) Transferase activity of (e) SpMUR3 and (f) SpXLT2 in the presence of an XXXG (S1 ΔGal XyG) or an XXXGXXXG (S2 ΔGal XyG) acceptor using UDP‐Gal as a donor. Error bars indicate Mean ± SD with an n of 3 biological replicates. The letters above indicate statistically significant differences between all experimental conditions accomplished by a two‐way ANOVA followed by a Tukey's multiple comparison test (P ≤ 0.05) (a–d) or a one‐way ANOVA followed by a Tukey's multiple comparison test (P ≤ 0.05) (e–f).

Thus, since previous data on Arabidopsis mur3 and xlt2 mutants suggest MUR3 and XLT2 as xyloglucan galactosyltransferases, we performed a subsequent acceptor substrate screen utilizing UDP‐Gal as a donor substrate (Figure 2b). The acceptor screen included three different XyG substrates: ΔGal XyG polymer, ΔGal XyG oligos that included a mixture of S1 and larger XXXG motifs, and native tamarind XyG polymer. Several common plant cell wall saccharides were also included as negative controls in our assays. Production of UDP as a by‐product of the GT reaction in the presence of the three XyG‐derived substrates indicates that they can serve as effective acceptors for both enzymes (Figure 2b). Matrix‐assisted laser desorption ionization time‐of‐flight mass spectrometry (MALDI‐TOF MS) analysis of the reaction products using ΔGal XyG oligos as acceptors indicated that both recombinant SpMUR3 and SpXLT2 catalyze the transfer of galactosyl residues (Figure S4). These results suggest that both enzymes are galactosyltransferases with specificity for polymerized ΔGal XyG. However, shorter ΔGal XyG oligosaccharides can still be utilized as suitable acceptor substrates.

We also tested the activity of both enzymes using UDP‐Arap and UDP‐Araf donors, with polymerized ΔGal XyG as the acceptor. Our results indicate that both enzymes can transfer Arap to the polymer. Despite the ability of SpMUR3 to hydrolyze both UDP‐Arap and UDP‐Araf donors in the absence of an acceptor, neither enzyme exhibited transferase activity for UDP‐Araf in the presence of a XyG acceptor (Figure 2c,d). To further investigate the acceptor substrate selectivity of SpXLT2 and SpMUR3 on longer chain XyG oligos, we examined transferase activity using purified S1 or S2 ΔGal XyG oligosaccharide acceptors with UDP‐Gal as a donor. Both SpXLT2 and SpMUR3 preferred the S2 ΔGal XyG motif over the shorter S1 ΔGal XyG motif, indicating that both enzymes prefer longer XyG oligosaccharides (Figure 2e,f). Based on these results, the S2 ΔGal XyG motif was used in subsequent experiments due to solubility issues with polymerized XyG after galactose removal.

SpMUR3 and SpXLT2 donor and acceptor kinetics

Since SpMUR3 and SpXLT2 are remarkably similar in sequence (Figure S5) and function, we decided to examine whether there were differences in both enzyme donor and acceptor affinity. For kinetic analyses, we measured the generation of the UDP by‐product to determine reaction rates with UDP‐Gal as a donor and pure S2 ΔGal XyG as an acceptor. SpMUR3 displayed a substantially lower K _M of 9.1 μm compared to 26.8 μm for SpXLT2 with UDP‐Gal as a donor substrate (Figure 3). Likewise, SpMUR3 displayed a lower K _M of 32 μm compared to 88.5 μm for SpXLT2 with a S2 ΔGal XyG acceptor (Figure 3).

Figure 3.

Kinetic analyses of SpXLT2 and SpMUR3.

(a–d) The analysis of reaction rate in the presence of 0.1 μm of SpMUR3 and 5 μm of SpXLT2 in 250 mm Tris/MOPS/MES pH 7, 30°C with indicated concentrations of UDP‐Gal and XXXGXXXG (S2 ΔGal XyG). Each point represents a mean of three replicates (n = 3), with error bars representing the standard deviation. V _max and K _m values were calculated using non‐linear regression of the Michaelis–Menten equation in GraphPad Prism 6.

Analysis of saccharide product formation

In the interest of understanding the substrate selectivity and regiospecificity of SpMUR3 and SpXLT2, MALDI‐TOF MS and High‐Performance Anion Exchange Chromatography with Pulse Amperometric Detection (HPAEC‐PAD) were used to formulate a detailed analysis of saccharide product formation. The results obtained from the initial high‐throughput donor and acceptor screening assays support the assumption that SpMUR3 and SpXLT2 function predominantly as xyloglucan galactosyltransferases, with lesser arabinosyltransferase activity, but this data alone is insufficient for a comprehensive biochemical study. A detailed examination of the XyG sidechain modifications synthesized by SpMUR3 and SpXLT2 was conducted to obtain information on substrate utilization, sugar addition, and regiospecificity. To get an initial indication of the products formed by SpMUR3 and SpXLT2, the S2 ΔGal XyG acceptor was subject to MALDI‐TOF MS analysis after being treated with SpMUR3 or SpXLT2 in the presence of UDP‐Gal at various time points. The resulting spectra revealed mass shifts indicative of galactose additions on the acceptor in the presence of both enzymes. Furthermore, SpMUR3 consistently added two galactose residues to the S2 ΔGal XyG acceptor while SpXLT2 only added one (Figure 4a). In this regard, SpMUR3 appeared to add the second Gal residue only after fully galactosylating the ΔGal acceptor with one Gal residue. Due to SpMUR3 and SpXLT2 exhibiting signs of transferase activity with UDP‐Arap (Figure 2c,d), we repeated these experiments using UDP‐Araf and UDP‐Arap donors. Based on MS analysis, Araf‐containing products were not generated by either enzyme. However, both enzymes transferred one Arap to a S2 ΔGal XyG motif (Figure 4b). Both enzymes appear to be prominently galactosyltransferases in vitro as SpMUR3 and SpXLT2 show a more substantial increase in UDP production with UDP‐Gal than with UDP‐Arap in the presence of XyG (Figure 2c,d).

Figure 4.

Examination of sidechain additions catalyzed by SpMUR3 and SpXLT2 onto XXXGXXXG using MALDI‐TOF MS.

(a) Time course showing galactosylation of a S2 ΔGal XyG acceptor in the presence of either SpMUR3 or SpXLT2 using UDP‐Gal as a donor.

(b) Modification of a S2 ΔGal XyG acceptor in the presence of either SpMUR3 or SpXLT2 using UDP‐Arap as a donor. Reaction times are indicated. Masses corresponding to an increase of 162 Da or 132 Da correspond to the addition of one hexose (Gal) or pentose (Arap) residue to the acceptor, respectively. Schematic representations of acceptor and hypothetical product structures are depicted using SFNG representation (Figure 1d).

Given that SpXLT2 adds one galactose to a S2 ΔGal XyG motif as compared to the two galactose residues synthesized by SpMUR3 (Figure 4a), we set out to determine whether SpXLT2 prefers the reducing versus non‐reducing end of the acceptor by labeling the reducing end with 2‐aminobenzamide (2AB). Labeled substrates were then incubated with SpXLT2 and reaction products were subsequently cleaved in half with XEG to yield labeled and non‐labeled S1 motifs. The 2AB‐labeled products enhance mass spectrometry by chromatographic sensitivity and are therefore predominantly detected over non‐2AB‐labeled products (Martial et al., 2008). In the case of SpMUR3, galactosylated and non‐galactosylated labeled S1 products were visible, showing no preferred specificity for the reducing versus non‐reducing end of a S2 ΔGal XyG motif (Figure S6b). In contrast, SpXLT2 shows only one fully galactosylated, 2AB‐labeled S1 motif (Figures S6a and S6b), implying that the galactose addition synthesized by SpXLT2 prefers only the reducing end block of the S2 ΔGal XyG motif. Since SpMUR3 and SpXLT2 can also use UDP‐Arap as a sugar donor in vitro, we investigated whether SpMUR3 and SpXLT2 may have a broader activity or if these results are due to the structural similarity between Gal and Arap. We tested donor ratios of 1:16, 1:4, and 1:1 of UDP‐Gal to UDP‐Arap with SpMUR3 and SpXLT2. Both enzymes use UDP‐Gal more readily in the presence of XyG, even with higher ratios of UDP‐Arap in solution, further indicating that SpMUR3 and SpXLT2 are primarily galactosyltransferases in vitro, substantially preferring the UDP‐Gal donor over UDP‐Arap (Figure 5).

Figure 5.

Examining the donor preference (UDP‐Arap versus UDP‐Gal) of SpMUR3 (top) and SpXLT2 (bottom) by MALDI‐TOF MS.

Modification of a XXXGXXXG (S2 ΔGal XyG) acceptor by either SpMUR3 or SpXLT2 with varying ratios of UDP‐Arap:UDP‐Gal donor substrates. Donor substrate ratios used were 16:1, 4:1, and 1:1 (UDP‐Arap: UDP‐Gal) with the final concentration of total donor in reactions equaling 1 mm. Masses corresponding to an increase of 162 Da or 132 Da correspond to the addition of one hexose or pentose residue to the acceptor. Schematic representations of acceptor and hypothetical product structures are depicted using SFNG representation (Figure 1d).

Previous research additionally implicates MUR3 and XLT2 as regiospecific enzymes, that is, MUR3 and XLT2 are reported to catalyze the addition of galactose to the third and second xylose from the non‐reducing end of the core XyG S1 motif, respectively, forming the XLXG and XXLG XyG structures (Figure 1). To biochemically examine the regiospecific properties of both SpMUR3 and SpXLT2, HPAEC‐PAD was used to quantify the formation of specific XyG motifs after a S2 ΔGal XyG substrate was modified by SpMUR3, SpXLT2, or SpMUR3 + SpXLT2 and then digested with XEG to generate S1 motif products. HPAEC‐PAD results reveal the formation of XXLG and a small amount of XLXG with SpMUR3 and only XLXG with SpXLT2 (Figure 6a). When combining both enzymes, we observed the formation of XXLG, XLXG, and XLLG motifs. HPAEC‐PAD was additionally used to characterize the XyG motifs generated by SpMUR3 and SpXLT2 when provided with UDP‐Arap as a donor. SpMUR3 generated products characteristic of XXDG motifs (Zhu et al., 2018), while SpXLT2 products produced a unique peak, which we attribute to be most likely XDXG (Figure S7).

Figure 6.

Examining the substrate regiospecificity of SpMUR3 and SpXLT2 on XXXGXXXG (S2 ΔGal XyG) using HPAEC‐PAD and xyloglucan product fucosylation by AtFUT1.

(a) HPAEC‐PAD chromatograms of S2 ΔGal XyG incubated with SpMUR3, SpXLT2, or SpMUR3 + SpXLT2 in the presence of UDP‐Gal. Products were digested with Xyloglucan‐specific endo‐β‐1,4‐glucanase (XEG) before HPAEC‐PAD analysis.

(b) MALDI‐TOF MS spectra of S2 XyG products generated by SpMUR3 or SpXLT2 in the presence of UDP‐Gal and subsequent spectra of those respective products after being fucosylated with AtFUT1 in the presence of GDP‐Fuc. Masses corresponding to an increase of 162 Da or 146 Da correspond to the addition of one hexose or deoxyhexose residue to the acceptor. Schematic representations of saccharide structures are depicted using SFNG representation (Figure 1d).

AtFUT1 is a fucosyltransferase that specifically catalyzes the addition of fucose onto galactosylated XyG residues, forming fuc‐α‐l‐(1,2)‐Gal‐β‐d‐(1,2)‐Xylα‐d‐(F) sidechains on the third xylose from the non‐reducing end of the structure (XLLG ➔ XLFG) (Urbanowicz et al., 2017). This is the same third xylose proposed to be galactosylated by MUR3. To further support the results obtained from HPAEC‐PAD, we used the GT37 AtFUT1 to examine the regiospecificity of these enzymes. To achieve this, a S2 ΔGal XyG acceptor was treated with SpMUR3 or SpXLT2 in the presence of UDP‐Gal before being further fucosylated by AtFUT1 in the presence of GDP‐Fuc. MALDI‐TOF MS analysis of the saccharide products of each reaction indicated that fucosylation of S2 products was only observed in samples previously treated with SpMUR3 and not with SpXLT2 (Figure 6b). Taken together, the analysis of the reaction products by both HPAEC‐PAD and MALDI‐TOF MS provides strong evidence for the regiospecific nature of both SpMUR3 and SpXLT2 in synthesizing the additions of galactose to the third and second xylose of the core XXXG motif, respectively.

Functional complementation with SpMUR3

XyG is a major hemicellulose in the plant cell wall, particularly in dicots, and plays a crucial role in cell wall structure and expansion. In Arabidopsis XyG, the “L” or “F” sidechain motifs are major components, and much research encompassing plant cell wall modifying GTs has been carried out in the Arabidopsis model. Furthermore, a distinct “cabbage” phenotype is observed in Arabidopsis mur3‐3 mutant plants where XyG with only XXXG and XLXG motifs are observed (Kong et al., 2015; Madson et al., 2003). By contrast, XLT2‐deficient plants exhibit no prominent phenotypes (Jensen et al., 2012). Given the unique MUR3‐deficient phenotype observed in Arabidopsis, MUR3 has been previously examined in planta numerous times (Immelmann et al., 2023; Jensen et al., 2012; Kong et al., 2015; Liu et al., 2015; Schultink et al., 2013; Wang et al., 2020; Wilson et al., 2023; Zhu et al., 2018).

We set out to determine whether SpMUR3 can functionally complement the Arabidopsis mur3‐3 mutant. It is well established that mur3‐3 plants have small, curled leaves, that are often referred to as “cabbage‐like” and lack the ability to add galactose to xylose residues on the third position of the XXXG subunit (Kong et al., 2015). This phenotype can be rescued when the plant is transformed with a MUR3 gene. To achieve this with SpMUR3, mur3‐3 (Salk_141953) plants were transformed with SpMUR3 driven by the 35S promoter via Agrobacterium infiltration. Arabidopsis mur3‐3 plants transformed with SpMUR3 exhibit a partial phenotypic rescue of the wild‐type phenotype (Figure 7a,b). SpMUR3 transformed plants are larger but retain some “cabbage‐like” characteristics, including curled rosette leaves (Figure 7d and Figure S8). Chemotypic analysis of the plant tissue by XyG fingerprinting and HPAEC‐PAD analysis revealed that rescued plants were able to produce xyloglucan with a motif composition similar to that of wild‐type plants (Figure 7c,e,f), suggesting a rescue of the cell wall chemotype. Quantification of XXFG and XXLG oligosaccharides, which are known to be galactosylated on the reducing‐end xylose by MUR3, showed that xyloglucan from wild‐type Arabidopsis plants contained 24.98% and 21.42% of these subunits, while xyloglucan from mur3‐3 SpMUR3 complemented plants that show a clear intermediate phenotype contained 15.36% and 8.43%, respectively. Taken together, our xyloglucan analysis indicated that SpMUR3 was able to partially complement the mur3‐3 xyloglucan chemotype (restoring about ~50% of the reducing end “L” sidechains relative to WT) and is consistent with the phenotypic data (i.e., partial rescue). However, the lack of complete phenotypic complementation and galactosylated XyG suggests that differences in activity or expression may be occurring between the native AtMUR3 and SpMUR3 transgene in the Arabidopsis model.

Figure 7.

Examination of SpMUR3 functional equivalency in Arabidopsis.

(a) phenotypic characterization of mur3‐3 KO, SpMUR3 rescued Gen‐1, and Wild‐Type Arabidopsis plants 14 days after gemination.

(b) PCR of Wild‐Type, mur3‐3 KO, and SpMUR3 rescued Arabidopsis Gen‐1 plants. Plants were examined for the mur3‐3 (Salk_141953) T‐DNA insertion in the native AtMUR3 gene, AtMUR3, and SpMUR3 respectively.

(c) xyloglucan fingerprinting of Wild‐Type, Gen‐3 SpMUR3 rescue, and mur3‐3 KO plant seedlings.

(d) Rosette leaf comparison of Wild‐Type, Gen‐3 SpMUR3, and mur3‐3 KO Arabidopsis plants 26 days after germination.

(e) HPAEC‐PAD chromatograms of xyloglucan motifs extracted from mur3‐3, Wild‐Type, and Gen‐3 SpMUR3 rescued plants and (f) the relative percent abundance of each xyloglucan motif extracted from each plant line. The numbers 6 and 16 represent two independent SpMUR3 rescued plant lines. Statistical significance between the XyG motif composition of plant lines was accomplished by a two‐way ANOVA followed by a Tukey's multiple comparison test, n.s. not significant, ****P ≤ 0.001).

DISCUSSION

Understanding the biochemical function of plant cell wall synthesizing glycosyltransferases is important in elucidating their roles in planta, especially for the diverse members of the GT47 family, whose members are predicted to play a role in the synthesis of a multitude of matrix polysaccharides. Similar to Arabidopsis, S. polyrhiza possesses a small genome of approximately 158 Mb that contains around 19 000 annotated genes (Michael et al., 2017). This conserved set of angiosperm genes, with minimal genetic redundancy, offers a unique opportunity to explore gene functions in a simplified monocot plant model (Acosta et al., 2021). Here, we biochemically characterized the GT47 family enzymes MUR3 and XLT2 from the greater duckweed S. polyrhiza. Our resultant data not only support previously obtained in planta data from their Arabidopsis counterparts, but also establish novel findings about these enzymes, such as their capacity to function as arabinopyranosyltransferases in vitro. SpMUR3 and SpXLT2 catalyze the addition of galactose onto xyloglucan to create the typical “L” sidechain motif observed on the polysaccharide. Further, we show that SpXLT2 modifies the second (X X XG) and SpMUR3 predominantly modifies the third (XX X G) xylose of a XXXG motif. Based on activity and kinetics, it appears that SpMUR3 exhibits a greater degree of activity in vitro in comparison to SpXLT2. These results may suggest activity differences between SpMUR3 and SpXLT2 within S. polyrhiza. This is supported by the XyG composition of S. polyrhiza where it was found that the percentage of XXXG‐type XyG (94% in total) which is galactosylated exclusively by SpMUR3 (XXFG) is 13%, whereas XyG galactosylated by only SpXLT2 or both SpMUR3 + SpXLT2 (XLXG/XXLG, XLSG/XLDG, XLLG, XLFG) is only 3% (Hsiung et al., 2023).

SpXLT2 additionally appears to only add one Gal residue to the reducing end of a S2 ΔGal XyG acceptor. This may indicate that the frequency of modification by SpXLT2 on the second Xyl is not as complete along the XyG polymer chain as modification by SpMUR3. SpMUR3 is also shown to functionally complement mur3‐3 plants, although full phenotypic rescue was not completely achieved. There are numerous reasons why this may be the case such as gene silencing or some attribute of SpMUR3 itself, but the partial rescue of plants, xyloglucan fingerprinting, and HPAEC‐PAD analysis of XyG motifs demonstrate the capacity of SpMUR3 to have functionality in the Arabidopsis model. Interestingly, given that SpMUR3 restored ~50% of the XyG galactosylated by wild‐type AtMUR3 and still possessed a partial mur3‐3 phenotype, it may be the case that the total quantity of galactosylated sidechains plays a direct role in the resulting cause of the mur3‐3 phenotype. This may support the notion that XyG structure is necessary for cellular trafficking, as the mur3‐3 phenotype has been linked to XyG precipitation and decreased cellular traffic from the Golgi (Hoffmann & Mcfarlane, 2024).

We also determined that both SpMUR3 and SpXLT2 can use UDP‐Arap as a sugar donor in vitro. However, the arabinopyranosyltransferase function of SpMUR3 was not observed in SpMUR3 complemented Arabidopsis plants, implying that this activity is absent in vivo. Additionally, both enzymes prefer to use UDP‐Gal more readily in the presence of XyG, even in conditions where both donors are present with UDP‐Arap existing as the predominant donor. This donor preference indicates that SpMUR3 and SpXLT2 have a greater affinity for transferring UDP‐Gal to XyG. Another puzzling part of our study is that SpMUR3 can hydrolyze UDP‐Araf in the absence of an acceptor, relative to no enzyme controls, but cannot use it for transfer onto XyG. This suggests that UDP‐Araf binds to the active site of SpMUR3 but is not proficient for transfer. Taken together, these data suggest that both enzymes primarily function as galactosyltransferases.

Given that both enzymes more readily transfer UDP‐Gal to XyG and the notable absence of arabinosylated XyG in complemented SpMUR3 Arabidopsis plants, the observed UDP‐Arap transferase activity in vitro may be a consequence of the close structural resemblance between Arap and Gal. Another possibility may be that SpMUR3 and SpXLT2 have a high degree of phylogenetic similarity to other GT47s within their respective clade which may function as key arabinosyltransferases. In the interest of identifying possible arabinosyltransferases within S. polyrhiza, we constructed a phylogeny of GT47‐A enzymes derived from S. polyrhiza, A. thaliana, Populus trichocarpa, Thuja pilicata, and Oryza sativa. We additionally included functionally characterized GT47‐A enzymes within our phylogeny such as PpXDT from Physcomitrella patens (Zhu et al., 2018), SlXST1/SlXST2/SlMUR3 from Solanum lycopersicum (Schultink et al., 2013), VcXBT/VcXLT2/VcMUR3 from Vaccinium corymbosum (Immelmann et al., 2023), BrMUR3 from Brassica rapa, and BoMUR3 from Brassica oleraces (Wang et al., 2020) (Figure 8a). SpMUR3 and SpXLT2 were found within the respective MUR3 and XLT2 subclades. Excluding SpMUR3 and SpXLT2, the only other GT47‐A S. polyrhiza sequences, Spipo1G00860000 and Spipo26G0017500, were surprisingly distinct from the characterized GT47‐A clade members, sharing 29% and 32% amino acid sequence identity with the arabinopyranosyltransferase PpXDT along with 30% and 35% amino acid sequence identity with the arabinofuranosyltransferase SlXST1. In contrast, SpMUR3 and SpXLT2 possess a greater sequence identity to PpXDT and SlXST1 in comparison to the previously mentioned unknown Spirodela sequences (Table S1).

Figure 8.

Phylogenetic tree and alignment of GT47‐A sequences.

(a) Phylogenetic tree representing GT47‐A clade enzymes in Arabidopsis thaliana, Populus trichocarpa, Thuja plicata, Oryza sativa, and Spirodela polyrhiza. Characterized GT47‐A clade members from other species have also been included such as PpXDT from Physcomitrella patens (Zhu et al., 2018), SlXST1/SlXST2/SlMUR3 from Solanum lycopersicum (Schultink et al., 2013), VcXBT/VcXLT2/VcMUR3 from Vaccinium corymbosum (Immelmann et al., 2023), BrMUR3 from Brassica rapa, and BoMUR3 from Brassica oleracea (Wang et al., 2020). The MUR3 and XLT2 subclades have been highlighted in blue and orange respectively. XyG motif structures have been included next to characterized members of the clade.

(b) sequence alignment of Spirodela GTs along with every characterized enzyme within the GT47‐A clade. Five distinct regions of interest associated with donor sugar specificity identified by (Wilson et al., 2023) have been highlighted. SFNG representations for sugar donors used by each enzyme are also shown.

We additionally examined the residues of our S. polyrhiza sequences and all characterized GT47‐A enzymes in five distinct regions identified in (Wilson et al., 2023) which correlate with distinctiveness in donor specificity (Figure 8b). SpMUR3, SpXLT2, PpXDT, and every other characterized galactosyltransferase shared similar residues within these regions. Spipo26G0017500 additionally shared most of the residues associated with the characterized galactosyltransferases. On the other hand, Spipo1G00860000 was the most distinct out of the aligned sequences as residues within the five regions were the most variable. Given that nearly 15% of S. polyrhiza XyG contains XXSG/XXDG arabinosylated sidechain structures (Hsiung et al., 2023), it is still possible that either of the two unidentified S. polyrhiza genes within the GT47‐A clade could function as arabinosyltransferases involved in the synthesis of “D” and “S” motifs in S. polyrhiza, though a thorough biochemical investigation of both genes is necessary to confirm this assumption.

Given that SpMUR3 and SpXLT2 have nearly identical functional and structural characteristics (Figure S5), the distinct functional differences between both enzymes present a unique opportunity for an in‐depth structural comparison to uncover key structural features involved in acceptor substrate selectivity. This can be leveraged to elucidate the GT‐B fold structural features responsible for substrate binding. This also calls for careful biochemical characterization coupled with phylogenetic and structural studies of other genes from the GT47 family.

MATERIALS AND METHODS

MUR3 and XLT2 gene identification, expression, and purification

Using pFam03016 and Arabidopsis thaliana GT47 genes, we have identified two genes, Spipo1G0014600 (SpMUR3) and Spipo17G0019700 (SpXLT2), based on the sequence similarity (Zhang et al., 2023). The DNA of the catalytic domain of MUR3 (36–544 residues) and XLT2 (86–537 residues) without the transmembrane domain as predicted by DeepTMHMM—1.0 (Hallgren et al., 2022) were generated via gene synthesis (Twist Biosciences) in the pTwist Gateway ENTR Kozak vector and transferred into the mammalian expression vector pGEn2‐DEST (Moremen et al., 2018) by Gateway™ recombination. SpMUR3 and SpXLT2 were expressed as soluble secreted fusion proteins by transient transfection of suspension culture HEK293‐F cells (FreeStyle™ 293‐F cells, Thermo Fisher Scientific, Waltham, MA, USA). Protein production was carried out for 5 days after transfection, and then cell cultures were harvested, clarified by sequential centrifugation at 1200 rpm for 10 min and 3500 rpm for 15 min at 4°C, and passed through a 5 μm filter (Millipore, Billerica, MA, USA). The proteins were purified using an AKTA 25 L system (Cytiva, Marlborough, MA, USA) at 4°C using immobilized metal affinity chromatography (IMAC) with HisTrap FF prepacked columns (Cytiva) (Moremen et al., 2018; Prabhakar et al., 2020), and stored at 4°C for short‐term or −80°C for long‐term storage according to published protocols. Purified proteins were loaded on the SDS‐PAGE to confirm the expected molecular weight.

Small and wide‐angle X‐ray scattering (SAXS/WAXS)

SAXS/WAXS data were collected at the National Synchrotron Light Source II (NSLSII) Life Science X‐ray scattering (LIX) beamline. SAXS/WAXS data were collected simultaneously employing Size Exclusion Chromatography coupled with SAXS (SEC‐SAXS), described previously (Yang et al., 2020). Briefly, a bioinert Agilent Infinity 1260 system with a multisampler injected 50 μl of sample onto the Cytiva Superdex 200 Increase 5/150GL column at a flow rate of 0.3 ml/min. Eluate from the column was split between SAXS/WAXS data, which were collected at 0.5 Hz and UV 280 in a roughly 2:1 ratio. Initial data processing employed LIX beamline software, py4xs for detector image processing, merging, and scaling, and lixtools to visualize and perform background subtraction. Buffer frames 0–40 were averaged and used for background subtraction and frames 170–195 were averaged, corresponding to the Mur3 peak. The Pair Distance Distribution Function was calculated using GNOM, part of the ATSAS suite (Svergun et al., 2001). GASBOR was used to generate the ab initio reconstruction of Mur3 from individual frames.

Donor and acceptor screening and kinetic analysis

The UDP‐Glo™ Glycosyltransferase Assay (Promega, Madison, WI, USA) was used to quantify the UDP bi‐product of glycosyltransferase reactions according to the manufacturer's instructions. UDP‐Glo™ Glycosyltransferase Assays, NDP‐sugar donors were CIAP‐treated (Calf Intestinal Alkaline Phosphatase) to achieve low background (Sheikh & Wells, 2006). All reactions (10 μL) consisted of 250 mm Tris+MOPS+MES pH 7, 0.2 mg/ml of acceptor substrates, 0.5 mm of NDP‐sugar (0.1 mm for donor screens), and 5 μm of the enzyme. Assays were carried out at 30°C for 120 min. 5 μL of each reaction was mixed with 5 μL of GDP‐Glo™ Detection Reagent in white polystyrene, low‐volume, 384‐well assay plates (Corning Incorporated, Corning, NY, USA) and incubated for 60 min at room temperature. Luminescence was recorded with a multifunctional microplate reader BioTek Synergy LX Multi‐Mode Reader (Agilent, Santa Clara, CA, USA). A standard curve was generated using UDP dilutions with a detection limit ranging from 0.122 to 125 pmol. Michaelis–Menten Kinetics of SpMUR3 and SpXLT2 were obtained using the UDP‐Glo assay mentioned above with varying donor and acceptor concentrations. The K _m and V _max were calculated using GraphPad Prism 6 software. Statistical analyses were also accomplished using the same software.

Substrates

UDP‐Xyl, UDP‐Arap, and UDP‐Gal were purchased from Carbosource Services (Athens, GA, USA). UDP‐Araf was obtained from the Peptide Institute Inc. (Saito‐Asagi, Ibaraki‐shi, Osaka, Japan). All other UDP‐sugars were purchased from Promega. Tamarind seed xyloglucan (XyG) polysaccharide used in these studies was previously generated at the CCRC from tamarind seed flour and further modified as described previously (Tuomivaara et al., 2015). ΔGal XyG polymer was generated by treating tamarind seed xyloglucan with β‐Galactosidase (Aspergillus niger, Megazyme, Chicago, IL, USA). ΔGal XyG oligos were generated by treating tamarind seed xyloglucan with xyloglucan‐specific endo‐β‐1,4‐glucanase (XEG) (Pauly et al., 1999) and β‐Galactosidase. XyG fragments were purified by High‐Performance Liquid Chromatography (HPLC) using a Superdex™ Peptide 10/300 GL column (York et al., 1993) to yield a mixture of S1 and larger XXXG motifs.

Briefly, to de‐galactosylate Tamarind XyG, 200 mg was dissolved in 50 mm sodium acetate pH 5 (200 ml) before being treated with galactosidase (400 units) from Megazyme and incubated overnight at 40°C. XEG (0.5 mg) was added to the solution and incubated at 37°C for 5 min before being heat deactivated. The XyG solution was then dialyzed in water with a 3500 Dalton molecular weight cut‐off (MWCO) dialysis membrane overnight before being centrifuged at 15000 RPM for 30 min. The supernatant was taken and treated with sodium acetate pH 5 to make the final concentration 50 mm. This solution was treated again with galactosidase and dialyzed, before being lyophilized and solubilized in water. S2 ΔGal XyG oligosaccharides were purified from this XyG solution via HPLC with a Superdex™ Peptide 10/300 GL column.

Preparation of 2AB‐labeled S2 ΔGal XyG xyloglucan

2AB (2‐aminobenzamide) labeling was carried out in a 1 ml (5 mg/ml) solution of S2 ΔGal XyG in a glass tube. 2AB and NaBH₃CN were added to the XyG solution to achieve final concentrations of 0.2 m (2AB) and 1 m (NaBH₃CN). This solution was solubilized at 65°C before the pH was adjusted to 6 with 10% acetic acid (AcOH). This solution was then kept at 65°C for 2 h. The reaction was dialyzed overnight in water using a 1000 Dalton MWCO dialysis membrane before the final product was lyophilized.

MALDI‐TOF MS and HPAEC‐PAD

XyG substrates were treated with SpMUR3 and/or SpXLT2 in 20 μL reactions containing 50 mm HEPES pH 7, 0.4 mg/ml S2 ΔGal XyG, 1 mm UDP‐Gal or UDP‐Arap, and 5 μm of the respective enzyme. To examine the preference for UDP‐Arap to UDP‐Gal donor substrate, 16:1, 4:1, and 1:1 ratios were tested, with the final concentration of both donors totaling 1 mm. Reactions were heat deactivated at the indicated time points and the saccharide reaction products were analyzed by matrix‐assisted laser desorption/ionization mass spectrometry (MALDI‐TOF MS). Fucosylation of SpMUR3 and SpXLT2 S2 XyG reaction products was carried out after heat‐deactivating the initial SpMUR3 and SpXLT2 overnight reactions. AtFUT1 (1 μm) and GDP‐Fuc (1 mm) were added and allowed to proceed for 1 h, heat deactivated and analyzed by MALDI‐TOF MS in positive ion mode as described below.

For MALDI‐TOF MS, reactions were diluted to 0.04 mg/ml, treated with Dowex‐50 cation exchange resin for 1 h, and centrifuged. The resulting supernatant (1 μL) was mixed with a matrix solution (2 μL; 20 mg ml⁻¹ 2,5‐dihydroxybenzoic acid (DHB) in 50% methanol) on a target plate and left to dry. Reaction products were analyzed using a smartfleX spectrometer (Bruker, Billerica, MA, USA) in positive ion mode. The theoretical and observed mass‐to‐charge ratios (m/z) of the XyG units examined in this study can be found in Table S2.

Xyloglucan oligosaccharides were analyzed by high‐performance anion exchange chromatography with pulsed amperometric detection (HPAEC‐PAD) analysis. To prepare samples for analysis, S2 ΔGal XyG or enzyme reaction products were converted to S1 fragments using XEG, diluted to a concentration of 0.2 mg/ml, and then analyzed with a Dionex™ ICS‐6000 DC chromatography system (Thermo Scientific) equipped with an autosampler and a pulse amperometric detector (PAD) and CarboPac PA1 column (Thermo Scientific, Sunnyvale, CA, USA). Oligosaccharides were eluted at 1.0 ml/min with a multi‐step gradient of NaOAc (0–3 min: 0 mm; 3–5 min: 0–40 mm; 5–27 min: 40–80 mm; 27–50 min: 80–430 mm; 50–52 min: 430 mm; 52–53 min: 430–1000 mm; 53–56 min: 1000–1000 mm 56–57 min: 1000–0 mm) in 100 mm NaOH.

Plant growth, transformation, phenotypic, and genotypic analysis

The SpMUR3 coding sequence was amplified with the primers listed in Table S3. SpMUR3 gDNA was then inserted into the pENTR/D‐TOPO entry vector via the pENTR/D‐TOPO™ cloning kit (Thermo Scientific) before being subsequently cloned into the pMDC32‐HPD vector (Qi & Katagiri, 2009) using Gateway™ LR Clonase™ II Enzyme mix (Thermo Scientific). Agrobacterium tumefaciens strain GV3101 was used to transform mur3‐3 (At2g20370; Salk_141953) plants with the generated constructs via the floral dip method (Clough & Bent, 1998). Seeds were collected and sterilized before being grown on Murashige and Skoog (MS) media containing 25 μg/mL hygromycin for transgenic plant selection. Selected plants were transferred to soil, and plants from generations T1 and T3 were used for phenotypic comparison as indicated. Plants were placed in soil and set in a ventilated growth chamber with a 14/10‐h light/dark cycle. Pictures of plants were taken with a Samsung Galaxy Z Flip 4 camera 14 days after germination and 26 days after germination.

Plant genotyping was accomplished by extracting DNA from Arabidopsis leaf tissue using a mixture of 10 mm Tris pH 8, 100 mm EDTA pH 8, and 0.5% SDS. PCR was used to identify native AtMUR3, SALK_141953 T‐DNA, and SpMUR3 in wild‐type, mur3‐3, and SpMUR3‐transformed plants. PCR was carried out with PrimeSTAR HS DNA Polymerase (TaKaRa). The primers used are listed in Table S3.

XyG fingerprinting was carried out by grinding homozygous generation T3 seedlings with a mortar and pestle in liquid nitrogen before being washed with 80% ethanol. Plant tissue was then centrifuged, and the supernatant was removed before being washed with acetone, methanol, methanol/chloroform (1:1 ratio), and methanol once again. After centrifugation, the pellets were left to dry for 2 days at 40°C. The seedling alcohol insoluble residue (AIR) samples were then de‐starched with Liquozyme SC DS and Spirizyme Excel (Novozymes, Franklinton, NC, USA) in sodium acetate (pH 5) at 50°C for 24 h. De‐starched AIR samples were then incubated in 4 m KOH overnight, after which the supernatant was extracted and subsequently dialyzed in water. The dialyzed sample was finally lyophilized. Lyophilized KOH fractions (5 mg) were dissolved in 500 μL of 50 mm ammonium formate (pH 5) before adding two units of Xyloglucanase (Megazyme) to the mixture. The reactions were incubated at 37°C for 24 h before being filtered with a 0.45 μm nylon filter plate. The samples were diluted 1:10 before being examined by MALDI‐TOF MS as described previously.

Phylogenetic tree

Protein sequences that contain PF03016 were downloaded from Phytozome 14 using these databases: Arabidopsis thaliana Araport 11, Oryza sativa v7.0, Populus trichocarpa v4.1, Spirodela polyrhiza v2, and Thuja plicata v3.1. Protein sequences of BrMUR3, BoMUR3 and PpXDT were obtained by searching Brassica rapa FPsc v1.3, Brassica oleracea capitata v1.0, and Physcomitrium patens v3.3 databases on Phytozome 14. Protein sequences of SIXST1, SIXST2, and SIMUR3 were downloaded from Sol Genomics Network. Protein sequences of VcXBT, VcXLT2, and VcMUR3 were obtained from Genebank using these accession numbers: OQ851746, OQ851744, and OQ851743 respectively.

An initial protein sequence alignment was performed in Geneious Prime (version 2022.2.1) using MUSCLE with default settings. Based on this alignment, a preliminary phylogenetic tree was constructed using the Neighbor‐Joining method, also with default settings. Using the SpMUR3 protein as a reference, the N‐terminal sequences of proteins grouped in clade A of the initial GT47 tree were removed, and the remaining sequences were extracted. These sequences were then realigned using MUSCLE with default settings. Finally, a Neighbor‐Joining tree including only the proteins from clade A was generated using the same default parameters.

AUTHOR CONTRIBUTIONS

CC, AG, DC, and PKP performed the experiments. CC, BRU, PKP, and LZ wrote the manuscript. LZ guided CC in doing plant experiments. SJZ, JB, and YJB performed Small and Wide‐Angle X‐ray Scattering experiments, analysis and edited the manuscript. BRU, KWM, and PKP made additional contributions and edited the manuscript. CC, BRU and PKP planned and designed the research. KWM, YJB, PKP and BRU were responsible for funding acquisition.

CONFLICT OF INTEREST

The authors declare that they have no conflicts of interest in relation to the content of this manuscript.

Supporting information

Figure S1. Analysis of purified SpMUR3 and SpXLT2.

Figure S2. Screening SpMUR3 and SpXLT2 transferase and hydrolytic activity at different pH ranges.

Figure S3. Screening SpMUR3 and SpXLT2 transferase activity at different temperature ranges.

Figure S4. MALDI‐TOF MS spectra of ΔGal XyG oligosaccharides (S2–S4) before and after reaction with SpMUR3 or SpXLT2 with UDP‐Gal as a donor.

Figure S5. Sequence and structural alignment of SpMUR3 (Spipo1G0014600) or SpXLT2 (Spipo17G0019700).

Figure S6. Examining the reducing‐end specificity of SpMUR3 and SpXLT2 on 2‐AB XXXGXXXG (2AB S2 ΔGal XyG) oligosaccharides using MALDI‐TOF MS.

Figure S7. HPAEC‐PAD comparison of SpMUR3 and SpXLT2 XyG (xyloglucan) reaction products using UDP‐Gal and UDP‐Arap donors.

Figure S8. Phenotypic examination of Gen‐3 SpMUR3 rescue, mur3‐3 KO, and Wild‐Type Arabidopsis plants 26 days after germination.

Table S1. Sequence identity of GT47‐A clade sequences from S. polyrhiza and functionally characterized GT47‐A enzymes from other species.

Table S2. Theoretical and observed mass‐to‐charge ratios (m/z) of xyloglucan oligosaccharides detected using MALDI‐TOF MS analysis.

Table S3. List of primers used in this study.

DATA AVAILABILITY STATEMENT

All the relevant data can be found within the manuscript and its supporting information and/or are available from the corresponding author upon reasonable request.