Arabinogalactan proteins: focus on carbohydrate active enzymes

Arabinogalactan proteins (AGPs) are a highly diverse class of cell surface proteoglycans that are commonly found in most plant species. AGPs play important roles in many cellular processes during plant development, such as reproduction, cell proliferation, pattern formation and growth, and in plant-microbe interaction. However, little is known about the molecular mechanisms of their function. Numerous studies using monoclonal antibodies that recognize different AGP glycan epitopes have shown the appearance of a slightly altered AGP glycan in a specific stage of development in plant cells. Therefore, it is anticipated that the biosynthesis and degradation of AGP glycan is tightly regulated during development. Until recently, however, little was known about the enzymes involved in the metabolism of AGP glycans. In this review, we summarize recent discoveries of carbohydrate active enzymes (CAZy; http://www.cazy.org/) involved in the biosynthesis and degradation of AGP glycans, and we discuss the biological role of these enzymes in plant development.


INTRODUCTION
Arabinogalactan proteins (AGPs) are a family of proteoglycans found on the plasma membrane and in the cell walls of diverse species of plants. AGPs are synthesized by several posttranslational modifications of proteins in the secretory pathway. The proteins generally contain repetitive dipeptide motifs, e.g., Ala-Pro, Ser-Pro, Thr-Pro, and Val-Pro, which are distinguished from the sequence motifs for extensin type glycosylation [e.g., Ser-(Pro) 2-3 ] known as another major class of O-glycosylation in plants (Kieliszewski, 2001). The Pro residues are hydroxylated by prolyl 4-hydroxylases and further O-glycosylated by glycosyltransferases (GTs). Moreover, many AGPs are attached by a glycosylphosphatidylinositol anchor, which attaches AGPs to the plasma membrane, but can be cleaved by phospholipases (Wang, 2001;Schultz et al., 2004). AGPs on the plasma membrane and cell wall may also be processed by proteolytic activities and glycosyl hydrolases or transported by endocytotic multivesicular bodies to the vacuole where they are degraded (Herman and Lamb, 1992).
The glycan moiety of AGPs accounts for more than 90% of their total mass, which has been suggested to play an essential role in the function of AGPs, based on studies using synthetic phenylglycoside dyes (β-Yariv reagents) that specifically binds to the β-1,3-galactan moiety of AGPs (Kitazawa et al., 2013) as well as various monoclonal antibodies that recognize different AGP glycan epitopes (Seifert and Roberts, 2007). However, because of its complexity and heterogeneity, little is known about the structurefunction relationship of AGP glycans. In fact, various structures have been reported for AGP glycans depending on samples and analytical methods. The common structural feature is a backbone of β-1,3-galactan, which is often substituted at O6 with side chains of β-1,6-galactan decorated further with arabinose, and less frequently also with fucose, rhamnose, and (methyl) glucuronic acid (Figures 1A,B). Tan et al. (2010) proposed that the backbone is composed of a repeat of a β-1,3-galactotriose unit with or without side chains, which is connected by β-1,6-linkages (kinks). This model is based on the AGPs synthesized onto synthetic peptides expressed in tobacco cells and analyzed by NMR (Tan et al., 2004(Tan et al., , 2010. In this model, the side chains are rather short and composed of a single Gal decorated by 1-5 other sugars. However, longer β-1,6-galactan side chains have been reported for AGPs from radish root (Haque et al., 2005), wheat flour (Tryfona et al., 2010) and Arabidopsis leaf (Tryfona et al., 2012) based on the linkage and mass spectroscopy analysis.

β-GALACTOSYLTRANSFERASES
The first step in the glycosylation of AGPs is the transfer of Gal to hydroxyproline residues present in the peptide backbone. The Arabidopsis enzyme catalyzing this step was identified (At4g21060, AtGALT2, Basu et al., 2013). This enzyme belongs to the CAZy family GT31 and the recombinant protein expressed in Pichia pastoris demonstrated GalT activity transferring a Gal to hydroxyproline residues in the synthetic AGP peptides. Arabidopsis T-DNA knockout mutants contained reduced levels of Yariv precipitable AGPs and microsomes purified from mutants exhibited reduced levels of GalT activity compared to wild type. The mutant lines showed no detectable growth phenotype under normal growth conditions. Since the GalT activity was not completely abolished in the mutant microsomes, redundant activities encoded by other genes most likely exist. Nevertheless, based on this study, the quantity of AGP-glycans appears to not be crucial for plant development.
Another Arabidopsis GT from family GT31 encoded by At1g32930 was also characterized. Recombinant enzyme expressed in Escherichia coli and Nicotiana benthamiana demonstrated β-1,6-GalT activity elongating β-1,6-galactan side chains of AGP glycans in in vitro assays (AtGALT31A; Geshi et al., 2013). AtGALT31A is expressed specifically in the suspensor cells of the embryo proper and T-DNA insertion lines showed abnormal cell division in the hypophysis and arrested further development of embryos. Therefore, functional AtGALT31A is essential for normal plant embryogenesis. How AtGALT31A that is expressed in suspensor cells influences cell division in hypophysis remains unknown.
AtGALT29A (At1g08280) was identified as a gene co-expressed with AtGALT31A. Recombinant enzyme expressed in Nicotiana benthamiana demonstrated β-1,6-GalT activities elongating β-1,6-galactan and forming 6-Gal branches on β-1,3-galactan of AGP glycans . Moreover, Förster resonance energy transfer analysis revealed an interaction between AtGALT29A and AtGALT31A when both proteins are expressed as C-terminal fluorescent fusion proteins in Nicotiana benthamiana . The protein complex containing heterologously expressed AtGALT29A and AtGALT31A were purified and demonstrated increased levels of β-1,6-GalT activities by the AtGALT29A single enzyme. These results suggest cooperative action between AtGALT31A and AtGALT29A by forming an enzyme complex, which could be an important regulatory mechanism for producing β-1,6-galactan side chains of type II AG during plant development.

β-GLUCURONOSYLTRANSFERASE
An Arabidopsis GT from family GT14 encoded by At5g39990 was identified as a glucuronosyltransferase involved in the biosynthesis of AGP glycans (AtGlcAT14A; Knoch et al., 2013). The enzyme expressed in Pichia pastoris demonstrated β-GlcAT activity by adding GlcA to both β-1,6and β-1,3-galactan. Arabidopsis possesses 11 proteins in the GT14 family, of which two additional proteins encoded by At5g15050 and At2g37585 also demonstrated the same β-GlcAT activities and were named AtGlcAT14B and AtGlcAT14C, respectively . The T-DNA insertion lines contained reduced levels of GlcA substitution of β-1,6-galactobiose and β-1,3-galactan in their AGPs compared to wild type. In addition to the altered levels of GlcA, a marked increase of Gal and a decrease of Ara were detected in the mutant AGPs. Mutant lines showed an increased cell elongation rate in dark grown hypocotyls and light grown roots during seedling growth compared to wild type. Since several sugars were altered in the mutant AGPs lines, it is unlikely that the observed phenotype  (Gille et al., 2013). 3 Chracterized Arabidopsis enzymes demonstrated β-xylosidase activity toward xylan (Goujon et al., 2003;Minic et al., 2004). 4 Characterized Arabidopsis enzymes demonstrated xyloglucan endo-transferase activity (Rose, 2002). 5 Characterized Arabidopsis enzymes demonstrated α-1,2-fucosidase activity specifically toward xyloglucan (Léonard et al., 2008;Günl et al., 2011). is solely a consequence of the reduced levels of GlcA, but most likely related to the dynamic conformational changes of AGP glycans caused in the mutants.

α-FUCOSYLTRANSFERASE
Two Arabidopsis GTs from family GT37 were identified as α-1,2fucosyltransferases involved in the biosynthesis of AGP glycans (AtFUT4 and AtFUT6, encoded by At2g15390 and At1g14080, respectively; Wu et al., 2010). AGPs from tobacco BY2 cells contain no fucose, but heterologous expression of AtFUT4 and AtFUT6 in BY2 cells resulted in fucosylated AGPs. The recombinant enzymes purified from BY2 cells demonstrated fucosyltransferase activity to endogenous AGPs. Single Arabidopsis T-DNA insertion lines, atfut4 and atfut6, contained reduced levels of fucose in AGPs, and the double T-DNA insertion line fut4/fut6 contained no detectable fucose in its AGPs ; however, no obvious phenotype was observed in both types of mutants when they were grown under normal conditions. Differences between wild type and mutants were only seen in seedlings grown under salt stressed condition, where the mutant lines showed reduced root growth compared to wild type Tryfona et al., 2014). This was somewhat surprising because the Arabidopsis mur1 mutant, which is defective in a GDP-mannose-4,6-dehydratase (Bonin et al., 1997), contained a 40% reduction of fucose in root extracts  and showed a 50% reduction of cell elongation rate in roots (Van Hengel and Roberts, 2002). The decrease of root cell elongation in mur1 was previously attributed to the lack of fucose in AGPs ( 2002;Vanzin et al., 2002). The mur1 phenotype might be due to under-fucosylated rhamnogalaturonan II, or to the combination of several cell wall polysaccharides deficient in fucose.

α-ARABINOFURANOSYLTRANSFERASE
An Arabidopsis GT encoded by At1g70630 (named REDUCED ARABINOSE YARIV1, RAY1; Gille et al., 2013), was characterized as a putative arabinofuranosyltransferase since the mutation caused a reduced level of arabinofuranose (Araf ) in its AGPs. This GT belongs to the GT77 family, which also contains XEG113, the mutation of which results in the reduction of β-linked arabinose in extensin (Gille et al., 2009). Therefore, Araf transferase activity that makes β-linkages was expected for RAY1, and indeed, microsomes isolated from Nicotiana benthamiana after expression of recombinant RAY1 demonstrated β-Araf transferase activity to methyl β-Gal. The T-DNA insertion lines contained reduced levels of 3-linked Ara in its AGP fractions compared to AGPs from wild type, and the mutant plants exhibited slower root growth as well as a reduced rosette size and inflorescence. However, β-1,3-Linked Araf has not been reported in AGPs, therefore the involvement of RAY1 in the biosynthesis of AGP glycans remains unclear.

GLYCOSIDE HYDROLASES
Glycoside hydrolases (GHs) acting on AGPs are potentially very important for the metabolism of these glycoproteins. AGPs from tobacco stylar transmitting tissue are degraded as the pollen tube grows and the released sugars are considered to be used as the carbohydrate resource necessary for the elongation of pollen tubes (Cheung et al., 1995). Similarly, rapid turnover of AGPs are observed in suspension cell culture and millet seedlings (Gibeaut and Carpita, 1991) and a substantial amount of AGPs are considered to be hydrolyzed to free sugars and recycled in the cytosol for the synthesis of new glycans (Gibeaut and Carpita, 1991) or degraded in the vacuole (Herman and Lamb, 1992). The appearance of distinct AGP epitopes in a developmentally regulated manner might be controlled by GHs in the cell walls. Additionally, the occurrence of free AG glycans detached from proteins observed in the cell walls may be a result of GH actions.
For the hydrolysis of AGP glycans, several GHs are required, e.g., β-galactosidases, β-galactanases, α-arabinofuranosidases, β-arabinopyranosidases, β-glucuronidases, α-fucosidases, and α-rhamnosidases. AGP degrading GHs from microbial origin have been relatively well characterized, while only a few plant GHs have been reported to degrade AGP glycans. Below is an overview for those GHs reported to possess hydrolase activity of AGP glycans from both microbial and plant origins ( Figure 1B and Table 1).
In plants, β-1,3-galactosidase has been purified from radish (Raphanus sativus) seed extracts . Based on the deduced protein sequence, the enzyme RsBGAL1 was classified to the GH35 CAZy family. This enzyme was expressed heterologously in Pichia pastoris and demonstrated GH activity by degrading β-1,3and β-1,6-galactan in an exo manner, but not β-1,4-galactan. The efficiency of degradation of AGP glycans by RsBGAL1 alone was limited, but co-treatment with arabinofuranosidase and glucuronidase resulted in the release of up to 90% of the bound sugars from AGPs, indicating the synergy of those GHs in the degradation of AGP glycans.
β-Arabinopyranose generally represents only a minor part of Ara in AGP glycans, but has been reported from acacia, larch and wheat flour AGPs (Aspinall et al., 1958;Groman et al., 1994;Odonmazig et al., 1994;Tryfona et al., 2010). β-Arabinopyranosidase has been identified from Streptomyces avermitilis . The enzyme, named SaArap27A, belongs to family GH27, and the recombinant enzyme expressed in Streptomyces demonstrated the release of L-arabinopyranoside (Arap) from p-nitrophenyl-β-Larabinopyranoside, as well as the release of L-arabinose from gum Arabic and larch AG. Arabidopsis contains four uncharacterized proteins in the GH27 family.

β-GLUCURONIDASE
Microbial β-glucuronidases are found in family GH79. Two fungal GH79 β-glucuronidases from Neospora crassa (NcGlcAase) and Aspergillus niger (AnGlcAase) have been cloned and recombinant proteins expressed in Pichia pastoris demonstrated β-glucuronidase activity (Konishi et al., 2008). AnGlcAase and NcGlcAase share high homology in their amino acid sequences, but possess slightly different substrate specificity. Both enzymes recognize unsubstituted and 4-methyl substituted β-GlcA on AGPs, but AnGlcAase cleaves both GlcA and 4-methyl GlcA with an equal efficiency, while NcGlcAase preferably cleaves GlcA and only small amounts of 4-methylGlcA. Arabidopsis contains three proteins in the GH79 family.
The T-DNA knockout insertion lines exhibited increased levels of GlcA, whereas plants overexpressing AtGUS2 lacked detectable levels of GlcA in their AGP fractions. The T-DNA insertion mutant of AtGUS2 showed no clear changes in the elongation rate of plant organs, whereas the overexpression lines exhibited increased elongation of roots and stems. The increase of cell elongation observed in the overexpression lines of AtGUS2 resembles similar observations of the atglcat14a T-DNA insertion lines. Although reduced levels of GlcA is observed in both types of plants, the altered profiles of other sugars present in AGP glycans are inconsistent. Therefore, it is unlikely that the increase of cell elongation is solely caused by the reduction of GlcA levels in AGPs.

CONCLUSIONS
Microbial GHs working on the degradation of plant AGPs have been reported in several studies, but little was known about the enzymes working on the biosynthesis and degradation of AGPs in plants. Recent discovery of plant GTs/GHs working on AGPs, together with the technical development of in-depth structural analysis of complex AGP glycans, has broadened our knowledge for AGP metabolism significantly. On the other hand, the attempt to elucidate the biological role of each sugar moiety or a particular part of AGP glycan structure by investigating knockout mutants or overexpressors of those enzymes did not result in straightforward answers. For instance, a mutation in AtGlcAT14A did not result in a sole reduction of GlcA but also exhibited an increase of Gal and a reduction of Ara in the AGP glycan as well as enhanced cell elongation in seedlings. Furthermore, the overexpression of AtGUS2 resulted in a reduction of GlcA, Gal and Ara in AGP glycans and seedlings showed increased cell elongation, similarly to atglcat14a. The developmentally regulated appearance of different AGP glycan epitopes is well known, but the results available thus far are inconclusive concerning the molecular role of a particular part of AGP glycans in plant growth and development.
The carbohydrate active enzymes involved in the AGP metabolism have just begun to be identified and characterized. Further investigation of the remaining members in the AGP glycosylation pathway and their role in vivo is needed to understand the role of CAZy enzymes in relation to AGP glycans, the cell wall architecture, and in plant growth and development.