Three β-Glucuronosyltransferase Genes Involved in Arabinogalactan Biosynthesis Function in Arabidopsis Growth and Development

Arabinogalactan proteins (AGPs) contain arabinogalactan (AG) polysaccharides that are biologically relevant to plant growth processes. Here, the biochemical and physiological roles of three Golgi localized β-glucuronosyltransferase genes (GLCAT14A, GLCAT14B and GLCAT14C) in Arabidopsis thaliana, responsible for the addition of glucuronic acid to AG chains, were further investigated using single, double and triple glcat14 mutant plants. These proteins were localized to the Golgi apparatus when transiently expressed in Nicotiana benthamiana. Sugar analysis of AGP extracts from Arabidopsis stem, leaf and siliques showed a consistent reduction in glucuronic acid in glcat14 mutants relative to wild type, with concomitant effects resulting in tissue-specific alterations, especially in arabinose and galactose sugars. Although we observed defects in trichome branching in glca14a/b and glca14a/b/c mutants, scanning electron microscope analysis/energy dispersive microanalysis (SEM/EDX) showed no difference in the calcium content of trichomes in these mutants relative to wild type. Immunoblot analyses of the stem and leaf showed a reduction in AGPs as detected with the LM2 antibody in glcat14a/b and glcat14a/b/c mutants relative to wild type. The current work exemplifies the possibility of conducting structure-function assessment of cell wall biosynthetic genes to identify their physiological roles in plant growth and development.


Introduction
Plant cell walls are composed of structurally complex heteropolymers that are essential to plant development. Arabinogalactan-proteins (AGPs) are a family of complex proteoglycans found in the plant cell walls of all higher plants [1] and contain a large amount of arabinogalactan polysaccharide (90% w/w) attached to hydroxyproline residues in their protein cores [2]. AGPs belong to a superfamily of structural macromolecules known as the hydroxyproline-rich glycoproteins (HRGPs), of which extensins and proline-rich proteins are members with unique carbohydrate moieties central to their biological functions [3]. Out of all the HRGPs, AGPs are the most complex, and their complexity is driven by the heterogeneity of their type II arabinogalactans (AG) and the variety of their protein backbones [4]. AGPs have N-terminal signal peptides, which are cleaved to allow for secretion, and about half of AGP family members also have C-terminal glycosylphosphatidylinositol (GPI) anchors to allow for their retention and attachment to the outer leaflet of the plasma membrane [5].
AGPs are implicated in an array of plant growth and development processes, including cell expansion, somatic embryogenesis, root and stem growth, salt tolerance, hormone signaling, programmed cell death, male and female gametophyte development, and wounding/defense [4,6]. Structurally, AGs are O-linked to hydroxyproline residues Arabidopsis. Previous work provided enzymatic evidence that three GLCATs (GLCAT14A, GLCAT14B, GLCAT14C) out of the 11 putative Arabidopsis GLCAT genes/enzymes in the CAZy GT14 family function in glucuronidation of AGPs [11]. More recently, GLCAT14D and GLCAT14E were reported to be involved in AG glucuronidation [12]. A common feature in verified and putative GLCATs in Arabidopsis is the presence of a highly conserved GLCAT domain [17]. Additionally, the intron-exon genetic architecture of Arabidopsis CAZy GT14 family members is conserved and includes the presence of four exons and three introns ( Figure 1B). Phylogenetic analysis showed that GLCAT14A and GLCAT14B belong to the same phylogenetic clade, while GLCAT14C, D and E are phylogenetically distant from GLCAT14A and GLCAT14B, consistent with earlier reports [11,12]. Sequence analysis showed differences in similarities among the GLCAT14 genes, with the lowest sequence similarity observed to be 39% (between AT1G53100 and ATGLCAT14D) and highest being between AT4G03340 and AT1G03520); others are nestled between these two extremes (Table S2). Interestingly, these five enzymes, ATGLCAT14A/B/C/D/E, demonstrated to be involved in the glucuronidation of type II AGs, have unique substrate preferences; ATGLCAT14A, ATGLCAT14B and ATGLCAT14C transfer GlcA onto both β-1,3 and β-1,6 galactan chains of shorter length, while ATGLCAT14D and ATGLCAT14E have preferences for longer galactan side chains [11,12].

glcat14 Mutant Generation and Characterization
Arabidopsis glcat14 T-DNA insertion mutant lines were confirmed for the presence of T-DNA insertions ( Figure S1A) and were subsequently used for genetic crosses to obtain higher order homozygous mutants (Figure S1B,C). Additionally, results obtained using qRT-PCR showed the absence of transcripts in these mutants ( Figure 1C). We could not verify the presence of a T-DNA insertion in the SALK_051810 line belonging to GLCAT14C, but a CRISPR knock-out mutant (glcat14c-2) has a 178 bp deletion [13] in the first exon ( Figure 1A) and biological phenotypes were comparable to those obtained in the glcat14c-1 mutant (SALK_005705), as subsequently described.

Subcellular Localization of GLCAT14A, GLCAT14B and GLCAT14C
In the subcellular localization experiment involving ATGLCAT14A, ATGLCAT14B and ATGLCAT14C, we detected punctate vesicles that co-localized with the Golgi marker sialyltransferase short cytoplasmic tail and single transmembrane domain fused to enhanced GFP (STtmd-GFP) ( Figure S3), indicating localization of ATGLCAT14A, ATGLCAT14B and ATGLCAT14C in the Golgi apparatus.

glcat14a/b and glcat14a/b/c had Reduced Immunolabelling of LM2 bound AGPs in Stem and Leaf
Structural modifications can affect the abundance of glycan moieties that make up the plant cell wall. These structural changes can be delineated using antibodies raised C Figure 1. ATGLCAT14A (AT5G39990), ATGLCAT14B (AT5G15050), and ATGLCAT14C (AT2G37585) gene/genetic mutant structure, phylogenetic relationships, and expression. (A) Schematic drawing of the ATGLCAT14A (AT5G39990), AT-GLCAT14B (AT5G15050) and ATGLCAT14C (AT2G37585) gene structure and sites of T-DNA insertions in two independent lines: atglcat14a-1 and atglcat14a-2 are allelic mutants for ATGLCAT14A; atglcat14b-1 and atglcat14b-2 are allelic mutants for ATGLCAT14B; atglcat14c-1, and atglcat14c-2, generated by CRISPR-Cas9 strategy targeting exon 1 by two guide RNAs gRNA1 (C1) and gRNA2 (C2) as indicated above [12] are allelic mutants for ATGLCAT14C. Exons are represented by blue boxes; introns are represented by thin gray lines and non-coding regions are represented by orange boxes. The T-DNA insertion sites for each mutant are indicated, with primer binding sites indicated by black arrows. The branch domain corresponds to PF02485 from the Pfam database is indicated above. Red arrowheads indicate primer binding sites for qRT-PCR. (B) Phylogenetic analysis of the GT14 gene family along with their exon-intron structure. (C) Relative gene expression of GLCAT14A, GLCAT14B, and GLCAT14C in Arabidopsis leaves in various genetic mutant backgrounds as determined by qRT-PCR. Transcript levels were normalized to the mean of one reference gene, the Arabidopsis Actin 2 gene, AtACT2. Averages of three biological replicates ± SE are shown.

glcat14 Mutant Generation and Characterization
Arabidopsis glcat14 T-DNA insertion mutant lines were confirmed for the presence of T-DNA insertions ( Figure S1A) and were subsequently used for genetic crosses to obtain higher order homozygous mutants (Figure S1B,C). Additionally, results obtained using qRT-PCR showed the absence of transcripts in these mutants ( Figure 1C). We could not verify the presence of a T-DNA insertion in the SALK_051810 line belonging to GLCAT14C, but a CRISPR knock-out mutant (glcat14c-2) has a 178 bp deletion [13] in the first exon ( Figure 1A) and biological phenotypes were comparable to those obtained in the glcat14c-1 mutant (SALK_005705), as subsequently described.

Subcellular Localization of GLCAT14A, GLCAT14B and GLCAT14C
In the subcellular localization experiment involving ATGLCAT14A, ATGLCAT14B and ATGLCAT14C, we detected punctate vesicles that co-localized with the Golgi marker sialyltransferase short cytoplasmic tail and single transmembrane domain fused to enhanced GFP (STtmd-GFP) ( Figure S3), indicating localization of ATGLCAT14A, ATGLCAT14B and ATGLCAT14C in the Golgi apparatus.

glcat14a/b and glcat14a/b/c Had Reduced Immunolabelling of LM2 Bound AGPs in Stem and Leaf
Structural modifications can affect the abundance of glycan moieties that make up the plant cell wall. These structural changes can be delineated using antibodies raised against glycan epitopes in an immunoblot analysis. The LM2 antibody, which specifically binds to β-glucuronic acid in Type II AGs in AGP protein cores, was used to immunolabel wild type, glcat14a/b and glcat14a/b/c leaf and stem protein extracts separated by SDS-PAGE prior to membrane transfer. As expected, LM2 immunolabelling generated broad smears, reflective of the heavily glycosylated nature of AGPs, and the labelling intensities of LM2-bound AGPs were greater in leaves than in stems (Figure 2A,B). In contrast to the wild type, weak signals with LM2-bound AGPs were observed in glcat14a/b and glcat14a/b/c extracts from leaves ( Figure 2A) and stems ( Figure 2B). Moreover, the molecular size of LM2-bound AGPs in the stem and the leaf were similar, and signal intensities were in the high molecular mass (100-250 kDa) region of the Western blot. against glycan epitopes in an immunoblot analysis. The LM2 antibody, which specifically binds to β-glucuronic acid in Type II AGs in AGP protein cores, was used to immunolabel wild type, glcat14a/b and glcat14a/b/c leaf and stem protein extracts separated by SDS-PAGE prior to membrane transfer. As expected, LM2 immunolabelling generated broad smears, reflective of the heavily glycosylated nature of AGPs, and the labelling intensities of LM2-bound AGPs were greater in leaves than in stems (Figure 2A,B). In contrast to the wild type, weak signals with LM2-bound AGPs were observed in glcat14a/b and glcat14a/b/c extracts from leaves ( Figure 2A) and stems ( Figure 2B). Moreover, the molecular size of LM2-bound AGPs in the stem and the leaf were similar, and signal intensities were in the high molecular mass (100-250 kDa) region of the Western blot.

Quantification of β-D-Gal-Yariv Precipitated AGPs in glcat14 Mutants
Given the differential expression of GLCATs in different organs and across different developmental stages [17], we quantified the amount of Yariv precipitable AGPs from different organs of WT and glcat14 mutants. The ability of β-D-Gal-Yariv to bind to the β-1,3-galactose backbone on AGPs was exploited to isolate and quantify AGP content as described previously [18]. Across the organs examined (rosette leaves, stems and siliques), the amounts of Yariv precipitable AGPs in the stems and siliques were greater than those observed in the rosette leaves. Specifically, in contrast to WT, we observed an increase in Yariv precipitable AGPs in glcat14a, glcat14a/c, glcat14a/b and glcat14a/b/c mutants in the rosette leaves and stems; however, in the siliques, the glcat14a mutants along with glcat14c, glcat14a/c, glcat14a/b and glcat14a/b/c mutants showed an increase in Yariv precipitable

Quantification of β-D-Gal-Yariv Precipitated AGPs in glcat14 Mutants
Given the differential expression of GLCATs in different organs and across different developmental stages [17], we quantified the amount of Yariv precipitable AGPs from different organs of WT and glcat14 mutants. The ability of β-D-Gal-Yariv to bind to the β-1,3-galactose backbone on AGPs was exploited to isolate and quantify AGP content as described previously [18]. Across the organs examined (rosette leaves, stems and siliques), the amounts of Yariv precipitable AGPs in the stems and siliques were greater than those observed in the rosette leaves. Specifically, in contrast to WT, we observed an increase in Yariv precipitable AGPs in glcat14a, glcat14a/c, glcat14a/b and glcat14a/b/c mutants in the rosette leaves and stems; however, in the siliques, the glcat14a mutants along with glcat14c, glcat14a/c, glcat14a/b and glcat14a/b/c mutants showed an increase in Yariv precipitable AGPs ( Figure 3). The glcat14b and glcat14b/c mutants, however, did not show an increase in AGP content and were comparable to WT. AGPs ( Figure 3). The glcat14b and glcat14b/c mutants, however, did not show an increase in AGP content and were comparable to WT. Statistical differences were determined by two-way ANOVA, followed by the Tukey's honestly significant difference test (** p < 0.01); results were averages of three independent experiments.

Monosaccharide Composition Analyses and Calcium Content
AGP glycans were evaluated in stems, siliques and leaves to provide insights into organ-specific glycan composition and to understand to what extent the AGP structural modifications would impact the abundance of other AGP sugars in glcat14 mutants relative to WT. When the plant cell wall integrity (CWI) system is compromised, adaptive changes in cellular and cell wall metabolism can alter wall composition and structure [19]. Previous work on glcat14a mutants reported a slight increase in galactosylation, while the GlcA content was comparable to WT in AG extracts from 14-day-old seedlings [7]. In acidhydrolyzed AGP extracts investigated by HPAE-PAD, we observed significant changes in sugar amounts relative to WT. Specifically, in the leaves, and with the exception of glcat14a and glcat14c, we observed an increase in the amounts of galactose and minor changes in arabinose relative to the wild type (Table S3). While reductions in the amount of GlcA were observed across all genotypes in leaves, we focused on glcat14a/b and glcat14a/b/c mutants because the leaf trichome branching phenotype was only found in these mutants (discussed below) and was absent in the glcat14 T-DNA insertion lines. In Figure 3. Quantification of AGPs in the various glcat mutants in different organs of 40-day-old Arabidopsis plants. Leaves, siliques, and stems were obtained from 40-day-old Arabidopsis plants. glcat14a, glcat14a/c, glcat14a/b and glcat14a/b/c showed significant increases in AGP content in the leaves and stems, while glcat14a, glcat14c, glcat14a/c, glcat14a/b and glcat14a/b/c showed increases in AGP content in siliques. AGPs were measured in micrograms per gram fresh weight. Statistical differences were determined by two-way ANOVA, followed by the Tukey's honestly significant difference test (** p < 0.01); results were averages of three independent experiments.

Monosaccharide Composition Analyses and Calcium Content
AGP glycans were evaluated in stems, siliques and leaves to provide insights into organ-specific glycan composition and to understand to what extent the AGP structural modifications would impact the abundance of other AGP sugars in glcat14 mutants relative to WT. When the plant cell wall integrity (CWI) system is compromised, adaptive changes in cellular and cell wall metabolism can alter wall composition and structure [19]. Previous work on glcat14a mutants reported a slight increase in galactosylation, while the GlcA content was comparable to WT in AG extracts from 14-day-old seedlings [7]. In acidhydrolyzed AGP extracts investigated by HPAE-PAD, we observed significant changes in sugar amounts relative to WT. Specifically, in the leaves, and with the exception of glcat14a and glcat14c, we observed an increase in the amounts of galactose and minor changes in arabinose relative to the wild type (Table S3). While reductions in the amount of GlcA were observed across all genotypes in leaves, we focused on glcat14a/b and glcat14a/b/c mutants because the leaf trichome branching phenotype was only found in these mutants (discussed below) and was absent in the glcat14 T-DNA insertion lines. In the leaf AGPs, we observed a 47% and 49% reduction in the amounts of GlcA in glcat14a/b and glcat14a/b/c mutants, respectively, relative to wild type ( Figure 4A), while for stem AGPs, we observed a 35% and 73% reduction in GlcA content in glcat14a/b and glcat14a/b/c mutants, respectively, relative to wild type; the remaining genotypes had comparable GlcA amounts to WT in stem AGPs (Table S4). With the exception of glcat14a/c and glcat14b/c, which were not significantly different from WT, a reduction in galactose content in glcat14a/b/c in the stem was unexpected, given the significant increases in AGP content of glcat14a/b/c mutants in stem tissues. In addition, minor changes were observed in the amount of arabinose in glcat14 mutants relative to WT. For the siliques, we observed a reduction in the GlcA content in glcat14 mutants, coupled with a corresponding increase in galactose content in both single and higher order glcat14 mutants (Table S5A). It should be noted that some residue amount of GlcA remains associated with the AGPs in all of these organs, regardless of the mutant. This suggest the involvement of other GLCATs, whose roles remain to be elucidated.
the leaf AGPs, we observed a 47% and 49% reduction in the amounts of GlcA in glcat14a/b and glcat14a/b/c mutants, respectively, relative to wild type ( Figure 4A), while for stem AGPs, we observed a 35% and 73% reduction in GlcA content in glcat14a/b and glcat14a/b/c mutants, respectively, relative to wild type; the remaining genotypes had comparable GlcA amounts to WT in stem AGPs (Table S4). With the exception of glcat14a/c and glcat14b/c, which were not significantly different from WT, a reduction in galactose content in glcat14a/b/c in the stem was unexpected, given the significant increases in AGP content of glcat14a/b/c mutants in stem tissues. In addition, minor changes were observed in the amount of arabinose in glcat14 mutants relative to WT. For the siliques, we observed a reduction in the GlcA content in glcat14 mutants, coupled with a corresponding increase in galactose content in both single and higher order glcat14 mutants (Table S5A). It should be noted that some residue amount of GlcA remains associated with the AGPs in all of these organs, regardless of the mutant. This suggest the involvement of other GLCATs, whose roles remain to be elucidated.   Calcium is well known as a universal signaling molecule and acts as a 'second messenger' in plants [20,21]. The ability of AGP to bind and release calcium in a pHdependent manner in vitro gave rise to the AGP Ca 2+ capacitor hypothesis [16]. The intramolecular calcium binding abilities by paired GlcA residues of AGPs was exploited to estimate the calcium content of AGPs. We investigated whether there were significant alterations in calcium bound by AGPs, possibly mediated by the reduction in the amounts of GlcA across tissues investigated using a colorimetric assay [22]. We observed that the calcium content of AGP extracts in glcat14 mutants were similar to the WT in leaves ( Figure 4B) and siliques (Table S5B); however, that was not the case for stems. While the calcium content of glcat14a was similar to WT, we observed a significant reduction in calcium content in glcat14b and glcat14c single mutants, and also for the higher order glcat14 mutants in stem ( Figure 4D)

Both Single and Double glcat14 Mutant Display Pleiotropic Growth Defects
Seeds obtained from WT and glcat14 mutant lines were grown on MS plates supplemented with or without ABA to investigate the potential roles of GLCAT14A, B, and C in seed germination, given that GLCAT14A and GLCAT14B are both highly expressed in the micropylar endosperm, and the endosperm cell wall's weakening is repressed by ABA [13]. The glcat14 T-DNA mutant lines germinated like WT seeds under standard growth conditions. However, when the growth media were supplemented with 1 µ M ABA, we observed that the glcat14 mutants exhibited a delay in germination ( Figure  5). Similarly, seeds grown under light conditions for 9 d showed no obvious differences in root growth relative to WT ( Figure S2A). Additionally, dark grown etiolated seedlings in both glcat14 T-DNA mutant lines and the CRISPR lines showed no obvious differences in root growth relative to WT ( Figure S2B). Previous work reported an increased length of hypocotyls and roots in glcat14a mutants of 5-day old etiolated seedlings [7]; however, it is noteworthy that our observation agrees with earlier work that showed no difference in hypocotyl length in etiolated seedlings except for glcat14a/b/e, which were remarkably For the sugar analysis, values are relative to total sugar composition (expressed as mol %) of triplicate assays ± SE. For the calcium assay, values are the mean ± standard deviation from three biological replicates. (E) Growth phenotypes of 40-day-old glcat14 mutants and WT showed growth reductions in the glcat14 double and triple mutants relative to WT; Scale bar = 12 cm. (F) Inflorescence stem lengths from 40-day-old wild-type and glcat14 mutant plants. Asterisks indicate significant differences between glcat14 mutants and WT as defined by one-way ANOVA followed by Tukey's multiple comparison test (Student's t-test, p < 0.05 for single asterisks, p < 0.01 for double asterisks), n = 15 per line per replicate.
Calcium is well known as a universal signaling molecule and acts as a 'second messenger' in plants [20,21]. The ability of AGP to bind and release calcium in a pH-dependent manner in vitro gave rise to the AGP Ca 2+ capacitor hypothesis [16]. The intramolecular calcium binding abilities by paired GlcA residues of AGPs was exploited to estimate the calcium content of AGPs. We investigated whether there were significant alterations in calcium bound by AGPs, possibly mediated by the reduction in the amounts of GlcA across tissues investigated using a colorimetric assay [22]. We observed that the calcium content of AGP extracts in glcat14 mutants were similar to the WT in leaves ( Figure 4B) and siliques (Table S5B); however, that was not the case for stems. While the calcium content of glcat14a was similar to WT, we observed a significant reduction in calcium content in glcat14b and glcat14c single mutants, and also for the higher order glcat14 mutants in stem ( Figure 4D).

Both Single and Double glcat14 Mutant Display Pleiotropic Growth Defects
Seeds obtained from WT and glcat14 mutant lines were grown on MS plates supplemented with or without ABA to investigate the potential roles of GLCAT14A, B, and C in seed germination, given that GLCAT14A and GLCAT14B are both highly expressed in the micropylar endosperm, and the endosperm cell wall's weakening is repressed by ABA [13]. The glcat14 T-DNA mutant lines germinated like WT seeds under standard growth conditions. However, when the growth media were supplemented with 1 µM ABA, we observed that the glcat14 mutants exhibited a delay in germination ( Figure 5). Similarly, seeds grown under light conditions for 9 d showed no obvious differences in root growth relative to WT ( Figure S2A). Additionally, dark grown etiolated seedlings in both glcat14 T-DNA mutant lines and the CRISPR lines showed no obvious differences in root growth relative to WT ( Figure S2B). Previous work reported an increased length of hypocotyls and roots in glcat14a mutants of 5-day old etiolated seedlings [7]; however, it is noteworthy that our observation agrees with earlier work that showed no difference in hypocotyl length in etiolated seedlings except for glcat14a/b/e, which were remarkably shorter [12]. We also investigated the role of GLCAT14A, B, and C in plant height, and our results show that glcat14a/c and glcat14b/c and CRISPR lines (glcat14a/b and glcat14a/b/c) were significantly shorter than the wild type plants ( Figure 4E,F). shorter [12]. We also investigated the role of GLCAT14A, B, and C in plant height, and our results show that glcat14a/c and glcat14b/c and CRISPR lines (glcat14a/b and glcat14a/b/c) were significantly shorter than the wild type plants ( Figure 4E,F). Previous work observed that CRISPR-Cas9 triple knockouts of GLCAT14A-C [13] showed defects in trichome branching. A recent study similarly identified ATGLCAT14D and ATGLCAT14E as being involved in ensuring normal trichome development; this study also found that trichome defects in glcat14 mutants were suppressed in a Ca 2+ concentration-dependent manner [12]. Using SEM, we examined the trichome defects of glcat14a/b and glcat14a/b/c and conducted an elemental composition analysis directed at estimating the amount of calcium using the energy dispersive X-ray (EDX) microanalysis technique [23]. Energy dispersive X-ray (EDX) microanalysis is an analytical technique commonly used to give semiquantitative measurements of surface elements within the surface layer of 1-2 microns in thickness [24]. While the significant majority of the trichomes in glcat14a/b and glcat14a/b/c are one-branched instead of the two-branched trichomes characteristics of WT ( Figure 6A; Table 1), EDX spectra of investigated genotypes for elemental calcium showed no significant differences in the normalized mass of calcium relative to WT ( Figure 6D). It is noteworthy that none of the single and Germination percentages of glcat14 mutants showed delayed germination compared to wild type in normal 1 /2 MS supplemented with 1 µM ABA, which were grown for 12 days following 3 days of stratification at 4 • C (See methods). Twenty-five seeds of each genotype were used for germination, with three replicates.
Previous work observed that CRISPR-Cas9 triple knockouts of GLCAT14A-C [13] showed defects in trichome branching. A recent study similarly identified ATGLCAT14D and ATGLCAT14E as being involved in ensuring normal trichome development; this study also found that trichome defects in glcat14 mutants were suppressed in a Ca 2+ concentration-dependent manner [12]. Using SEM, we examined the trichome defects of glcat14a/b and glcat14a/b/c and conducted an elemental composition analysis directed at estimating the amount of calcium using the energy dispersive X-ray (EDX) microanalysis technique [23]. Energy dispersive X-ray (EDX) microanalysis is an analytical technique commonly used to give semiquantitative measurements of surface elements within the surface layer of 1-2 microns in thickness [24]. While the significant majority of the trichomes in glcat14a/b and glcat14a/b/c are one-branched instead of the two-branched trichomes characteristics of WT ( Figure 6A; Table 1), EDX spectra of investigated genotypes for elemental calcium showed no significant differences in the normalized mass of calcium relative to WT ( Figure 6D). It is noteworthy that none of the single and double glcat14 T-DNA insertion mutants had defects in trichome branching, as the trichome defects were only observed in glcat14a/b and glcat14a/b/c mutants. double glcat14 T-DNA insertion mutants had defects in trichome branching, as the trichome defects were only observed in glcat14a/b and glcat14a/b/c mutants.

glcat14a/c Mutants Had Reduced Pollen Germination and Misshaped Pollen
Scanning electron micrographs and in vitro pollen germination experiments were conducted to examine the pollen morphology and pollen germination of various glcat14 T-DNA mutants. Previous work showed that glcat14a/b and glcat14a/b/c mutants displayed significant increases in defective pollen with concomitant effects on pollen germination [13]. In this study, we observed that in addition to glcat14a/b and glcat14a/b/c mutants, pollen from the glcat14a/c mutant also demonstrated a significantly lower germination rate compared to WT ( Figure 7B), which could be reflective of the significant increase in defective pollen in glcat14a/c mutants in both the in vitro pollen germination ( Figure 7A,C) and SEM ( Figure 7E). Although the pollen tube lengths of all glcat14 mutants were comparable to WT ( Figure 7D), an observation that is consistent with earlier studies [13], some pollen tubes of glcat14a/c mutants displayed tip swelling relative to WT ( Figure 7A) coupled with slight alterations in the reticulate pattern of Arabidopsis pollen ( Figure 7E). The role of AGPs in pollen germination was demonstrated in previous work on agp6 and agp11 and agp6/11 double mutants, which displayed reduced pollen germination [25],

glcat14a/c Mutants Had Reduced Pollen Germination and Misshaped Pollen
Scanning electron micrographs and in vitro pollen germination experiments were conducted to examine the pollen morphology and pollen germination of various glcat14 T-DNA mutants. Previous work showed that glcat14a/b and glcat14a/b/c mutants displayed significant increases in defective pollen with concomitant effects on pollen germination [13]. In this study, we observed that in addition to glcat14a/b and glcat14a/b/c mutants, pollen from the glcat14a/c mutant also demonstrated a significantly lower germination rate compared to WT ( Figure 7B), which could be reflective of the significant increase in defective pollen in glcat14a/c mutants in both the in vitro pollen germination ( Figure 7A,C) and SEM ( Figure 7E). Although the pollen tube lengths of all glcat14 mutants were comparable to WT ( Figure 7D), an observation that is consistent with earlier studies [13], some pollen tubes of glcat14a/c mutants displayed tip swelling relative to WT ( Figure 7A) coupled with slight alterations in the reticulate pattern of Arabidopsis pollen ( Figure 7E). The role of AGPs in pollen germination was demonstrated in previous work on agp6 and agp11 and agp6/11 double mutants, which displayed reduced pollen germination [25], while another report in Torenia fournieri demonstrated that a disaccharide sugar, β-methyl-glucuronosyl galactose (4-Me-GlcA-β-1,6-Gal), present on AGPs, makes pollen tubes competent for ovule targeting/guidance [26]. Considering that large amounts of pollen grains are produced during Arabidopsis sexual reproduction, there appears to be a significant proportion that are competent to ensure fertilization, given that the silique lengths and seed set in glcat14a/c mutants were comparable to WT (data not shown). while another report in Torenia fournieri demonstrated that a disaccharide sugar, βmethyl-glucuronosyl galactose (4-Me-GlcA-β-1,6-Gal), present on AGPs, makes pollen tubes competent for ovule targeting/guidance [26]. Considering that large amounts of pollen grains are produced during Arabidopsis sexual reproduction, there appears to be a significant proportion that are competent to ensure fertilization, given that the silique lengths and seed set in glcat14a/c mutants were comparable to WT (data not shown).

Discussion
A large number of Arabidopsis genes are predicted to contain glycosyltransferases that are involved in cell wall biosynthesis [27]. Despite these predictions, very few candidate genes have been functionally characterized. Genetic strategies deployed to elucidate glycosyltransferase functions have been hampered by functional redundancy, as many genes have been identified to play similar roles in the same and/or different organs/tissues. As a result, multiple genes have to be disrupted before biochemical or physiological roles of these cell wall genes can be deduced. Previous work identified eleven GLCAT genes in Arabidopsis that belong to the CAZy GT14 family [17,28]; these genes may be involved in the glucuronidation of Type II AGs, given their conserved genetic architecture and the conserved GLCAT domain present in genes [17]. While the functional roles of some GLCAT genes have been identified (i.e., GLCAT14A-E; Figure 1B), the functional roles of the remaining putative GLCATs remain to be verified and studied further. Work done previously using CRISPR-Cas9 knockouts of GLCAT14A-C showed severe pleiotropic growth defects, especially in the higher order mutants [13], while subsequent work by another group extended this work and identified developmental defects in additional Arabidopsis glcat14 mutants that were partially suppressed by supplementing the growth medium with calcium [12]. In this report, T-DNA insertion mutants and CRISPR lines of GLCAT14A-C were utilized to corroborate previous findings and to obtain additional insights into the biochemical and physiological functions of GLCAT14A-C in Arabidopsis. Specifically, we sought to understand the degree to which the alterations in the amount of GlcA and calcium content accounts for the mutant phenotypes using biochemical and molecular genetic approaches.
Plant cell walls are important dynamic structures that need to be monitored properly to ensure their functional integrity [29]. Alterations in the polymer structure often trigger compensatory changes in the cell wall during normal growth or in changing environments. Strong support for the continuous cellular surveillance mechanism aimed at optimizing plant cell function exists [30]. The cell wall integrity maintenance mechanism constantly monitors and perceives changes in biological processes and coordinates activities to promote plant growth through the restructuring of glycan moieties [31,32]. In order to examine whether there are alterations in AGP abundance in glcat14 mutants relative to WT, we quantified the amount of AGPs in rosette leaves, stems, and siliques using the β-Gal Yariv reagent that selectively binds to AGPs [18]. We observed that the amount of glycosylated AGPs was altered in the various organs of the glcat14 mutants, with increased amounts of Yariv-precipitable AGP detected in siliques and stems compared to rosette leaves ( Figure 3). Moreover, an increase in AGP content was detected in glcat14a, glcat14a/c, glcat14a/b and glcat14a/b/c mutants in the rosette leaves and stems, while in the siliques, glcat14a, glcat14c, glcat14a/c, glcat14a/b and glcat14a/b/c mutants showed an increase in AGP content. Although previous work reported an increase in Yariv-precipitable AGPs in CRISPR-Cas9 generated glcat14ab and glcat14abc [13], our findings indicate that the genetic knock out of GLCAT14A plays a role in the increased Yariv precipitable AGP content observed for glcat14a/c, glcat14a/b and glcat14a/b/c mutants. In addition, the increase in Yariv-precipitable AGP observed among glcat14 mutants relative to the wild type might be a compensatory response that is necessary for maintaining cell wall integrity and establishing normal plant physiological functions.
Considering the observed alterations in AGP content across the investigated organs in the glcat14 mutants, we conducted monosaccharide composition and Western blot analyses to gain additional insights into changes in the glycosylation processes in these mutants. We observed organ-specific glycosylation responses, reflected by organ-specific alterations in the galactose content in glcat14 mutants investigated. Previous work on the sugar analysis of AGP extracts of the aerial parts of glcat14 mutants showed an increase in the galactose content in glcat14ab and glcat14abc [13], but our findings indicate that this pattern is not the same for all organs. While we observed an increase in galactose contents in the leaves ( Figure 4A) and siliques (Table S3) for glcat14a/b and glcat14a/b/c, we observed a significant reduction in galactose content in the stem for glcat14a/b/c mutants relative to WT ( Figure 4C). Overall, the observed increase in galactose content in the glcat14 T-DNA mutant and CRISPR mutant lines (Tables S3-S5A) centers around the idea that GlcA residues serve as a "cap" that can terminate the elongation of galactan chains and hence the absence of such GlcA residues could lead to an increase in the length of galactan chains [7]; however, this idea was challenged elsewhere [12]. Our data indicate that organ-specific cell wall changes do occur in response to structural changes, to ensure that plant cell walls can perform their biological functions in different situations [32].
Immunolabelling of β-GlcAs of Type II AGs showed weak immunolabelling with LM2, an antibody that targets β-GlcAs of Type II AGs, in glcat14a/b and glcat14a/b/c stem and leaf extracts compared to the wild type which had increased LM2 binding. This finding corroborates the observed reduction in GlcA in the monosaccharide composition analysis, supporting the argument that other GLCATs may be involved in the AG glucuronidation processes. The LM2 signal was detected in the high molecular mass region (Figure 2A,B), which might be a due to AGPs' interaction with other cell wall polymers, as demonstrated in APAP1 mutants [33]. We observed weak binding of LM2 antibody in stem compared to the leaf, and this differences in LM2 binding may be due to the micro-heterogeneity of the AGP glycan in stem and leaf.
Arabidopsis GLCAT14A-C are highly expressed during seed germination [17]. Moreover, GLCAT14A and GLCAT14B are primarily expressed in the micropylar endosperm during seed development [34]. We observed significant delays in seed germination in the presence of 1 µm ABA in glcat14 mutants relative to WT ( Figure 5), a finding that is consistent with an earlier report [13]. ABA is reported to influence the cytosolic calcium oscillations and impact many physiological processes, such as seed germination and environmental stress responses [35,36], and such oscillations in cytoplasmic calcium concentrations could be mediated by calcium that is bound by GlcA residues in AGPs [16]. In any case, further research is needed to verify this idea.
Following the localization of ATGLCAT14A in the Golgi apparatus [7], we investigated the subcellular localizations of ATGLCAT14B and ATGLCAT14C. We observed that ATGLCAT14B-EYFP and ATGLCAT14C-EYFP signals, in addition to that of ATGLCAT14A-EYFP, were detected in punctate vesicles that co-localized with STtmd-GFP ( Figure S3A-C), indicating localization of ATGLCAT14A, ATGLCAT14B and ATGLCAT14C in the Golgi apparatus. These observations are consistent with the idea that that ATGLCAT14A, AT-GLCAT14B and ATGLCAT14C function in the glucuronidation of AGPs at unique positions in the type II glycan structure [12], and therefore are expected to be localized to the Golgi.
The importance of GLCATs in leaf development is reflected by the significant reduction in GlcA content in glcat14a/b and glcat14a/b/c mutants relative to WT, as revealed by monosaccharide composition and Western blot analyses (Figures 2A and 4A). Using SEM, we observed defects in trichome branching in glcat14a/b and glcat14a/b/c mutants ( Figure 6A), and this finding is consistent with earlier work [12,13]. While most of the glcat14a/b and glcat14a/b/c trichomes were single-branched trichomes (Table 1), a much more severe trichome branching defect was reported for glcat14a/b/e mutants, which could be partially suppressed by supplementing the growth media with calcium [12]. Energy dispersive X-ray (EDX) analysis is an analytical technique commonly used to provide semi-quantitative measurements of surface elements within a surface layer of 1-2 microns in thickness [24]. We utilized this technique to estimate whether there were differences in the amount of calcium in trichomes of glcat14a/b and glcat14a/b/c mutants. We found that the normalized calcium masses of trichomes of glcat14a/b and glcat14a/b/c mutants were comparable to wild type based on the SEM-EDX technique ( Figure 6D). Similarly, the calcium contents of leaf AGP extracts obtained from glcat14a/b and glcat14a/b/c mutants were similar to WT ( Figure 4B). Despite this finding, we cannot exclude the possibility that calcium plays a role in establishing normal trichome branching and may be involved during the earlier stages of trichome development, as demonstrated earlier [12]. It is worth mentioning that most GlcA residues in leaf AGPs are methylated [10], and it remains to be determined whether certain GlcA residues added by particular GLCATs are preferentially methylated or involved in calcium binding and how such modifications impact trichome development in Arabidopsis.
This investigation into understanding the roles of GLCAT14A, B, and C in plant growth and development identified significant reductions in plant height in glcat14a/c, glcat14b/c T-DNA insertion mutant lines relative to WT ( Figure 4E,F). Previous work showed that glcat14bc, glcat14ab and glcat14abc were shorter than WT [13], while severe growth defects were also reported for glcat14a/b/d and glcat14a/b/e [12]. We found that glcat14a/c mutants were also significantly shorter than the WT but were not as severe as those observed in glcat14a/b and glcat14a/b/c mutants ( Figure 4E,F). The importance of calcium was recently demonstrated by the observed hypersensitivity of the inflorescence stem lengths of some glcat14 mutants to low calcium concentrations [12]. In addition, the significant reductions in calcium were only observed in the stem, while the calcium contents of glcat14 mutants in leaves and siliques were comparable to WT. Considering the complex framework of plant cell wall glycans, the maintenance of cell wall structure and expansion may not be limited to the interaction of Ca 2+ with AGP and/or pectins, as other possible interactions of calcium may exist with cell wall-modifying enzymes [20,21].
Homozygous glcat14a-1 and glcat14c-1 were crossed, and the F1 generation was selfed before screening the F2 generation to obtain glcat14a/c. Similarly, glcat14b-1 and glcat14c-1 were crossed, and the F1 generation was selfed before screening the F2 generation to obtain glcat14b/c ( Figure S1). CRISPR-Cas9 generated glcat14a/b and glcat14a/b/c mutants (henceforth, referred simply as glcat14a/b and glcat14a/b/c) were previously generated in our lab [13] and included in this study to gain broader insight into the biological roles of the GLCATs. For the purpose of this work, glcat14a/b and glcat14a/b/c, generated by the CRISPR-Cas9 approach, are referred to as glcat14 CRISPR lines, while the other genotypes involving T-DNA insertion mutants are referred to as glcat14 T-DNA insertion lines.

Quantitative RT-PCR
To verify whether gene transcripts exist in Arabidopsis glcat14 T-DNA insertion lines, total RNA was extracted from Arabidopsis leaves using Trizol (Life Technologies, Grand Island, NY, USA). RNA (1 µg) was used for first-strand cDNA synthesis along with an oligo (dT20) primer and SuperScript III reverse transcriptase (Thermo Scientific). The qPCR was performed using PerfeCTa SYBR Green SuperMix (Quanta Biosciences, Gaithersburg, MD, USA) following methods described elsewhere [13]. Expression levels were calculated relative to the Arabidopsis ACTIN2 gene.

AGPs Quantification Using β-D-Gal-Yariv Reagent
AGPs were extracted from rosette leaves, stems, and siliques of 40-day-old WT, glcat14 T-DNA insertion and CRISPR lines, using the β-D-Gal-Yariv precipitation method described elsewhere [38]. The dissolved AGPs were quantified by measuring absorbance at OD 420 . Different concentrations of gum arabic (G9752, Sigma-Aldrich, St. Louis, MO, USA) dissolved in 1% CaCl 2 were used to make the standard curve. Measurements for each genotype were done in triplicate.

Monosaccharide Composition Analysis by High Performance Anion Exchange Chromatography with Pulsed Amperometric Detection (HPAE-PAD)
Following the extraction of AGPs as described above, fifty microliters of 10 mg/mL AGP were hydrolyzed using 2 N TFA at 121 • C for 90 min. TFA was removed by evaporation with N 2 gas. Samples were dissolved in 500 µL milli-Q water containing 0.2 mM cellobiose as an internal standard. A standard sugar mixture (fucose, rhamnose, arabinose, galactose, glucose, xylose, mannose, galacturonic acid, and glucuronic acid) was used for making the standard curve. Monosaccharide compositions were calculated as molar percentages (mol %). All samples and standards were subjected to high pH anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD) using a Dionex PA-20 column (Thermo Fisher Scientific, Sunnyvale, CA, USA), essentially as described previously [39].

Calcium Binding Assay
Following the extraction of AGPs, calcium binding assays were conducted as previously described [13]. Briefly, 10 µL of 10 mg/mL AGP extracted from glcat14 T-DNA insertion and glcat14 CRISPR lines were used for the calcium binding assay. A commercial calcium colorimetric assay kit (MAK022, Sigma-Aldrich, St. Louis, MO, USA) was used for calcium measurement following the manufacturer's protocol. In this assay, calcium ions from the AGP extracts form a complex with the o-cresolphthalein, which causes a color change from transparent to pink. The amount of calcium was determined using a UV spectrometer at OD 575 and a standard curve made with different concentrations of CaCl 2 .

Seed Germination
One-month post-harvested seeds of WT and allelic glcat14 mutant lines (glcat14a-1, glcat14a-2, glcat14b-1, glcat14b-2, glcat14c-1, glcat14c-2, glcat14a/c, glcat14b/c, glcat14a/b, and glcat14a/b/c) were used for germination experiments. Following seed sterilization, seeds were stratified at 4 • C for 3 d in the dark and were sown on 1 /2 MS and 1% sucrose agar plates supplemented with or without 1 µM ABA. Germination percentages were counted from 1 to 12 d after sowing and approximately 25 seeds were sown for each genotype with three replicates. In another experiment to evaluate the growth of etiolated seedlings of respective genotypes, stratified seeds sown on MS plates were exposed to 4 h of fluorescent white light at 20 • C to synchronize germination before wrapping the plates in aluminum foil for growth in the darkness for 5 d at 20 • C. Root and hypocotyl length measurements at 5 d were taken and values were compared to wild type.

Trichome Morphology
The first two leaves of 14-day-old WT, glcat14a/b and glcat14a/b/c mutants were immersed in methanol and leaf tissues were critical point dried following methods described earlier [40] before examining using scanning electron micrography (SEM). Trichomes were examined using a JEOL JSM-6390 scanning electron microscope (Hitachi High-Technologies, Tokyo, Japan) and leaf tissues were mounted on aluminum stubs using double adhesive tapestubs and sputter coated with a palladium alloy using a Cressington 208C high-resolution sputter coater (Ted Pella Inc., Redding, CA, USA). Scanning electron microscope energy dispersive X-ray (SEM/EDX) analysis was used as a semi quantitative approach to estimate the calcium content in the trichomes of WT, glcat14a/b and glcat14a/b/c mutants following methods described previously [23].

Cloning Procedures and Subcellular Localizations of GLCAT14A, GLCAT14B and GLCAT14C
Full-length sequenced GLCAT14A, GLCAT14B and GLCAT14C CDS, flanked by the sites KpnI were subcloned into pSAT6 vector to generate pSAT6-GLCAT14A-EYFP and pSAT6-GLCAT14C-EYFP constructs containing PI-PSPI restriction sites. Following the digestion of the new constructs (pSAT6-GLCAT14A-EYFP, pSAT6-GLCAT14B-EYFP and pSAT6-GLCAT14C-EYFP) with PI-PSPI, the fragment was introduced into the pPZP-RCS-nptII binary vector [43]. The binary vector containing the constructs was transformed into Agrobacterium strain GV3101 before infiltration using 5-week old tobacco leaves grown at 22-24 • C [44]. Golgi co-localization was confirmed by using the Golgi marker sialyltransferase short cytoplasmic tail and single transmembrane domain fused to enhanced GFP (STtmd-GFP). Transformed plants were incubated under normal growth conditions and imaged 3 days post-infiltration using an upright Zeiss LSM 510 META laser scanning confocal microscope (Jena, Germany), with a 40× oil immersion lens and an argon laser. Both GFP and E-YFP were imaged using 458 and 488 nm excitation wavelengths, respectively.

Conclusions
This study corroborates and extends our understanding of the molecular dynamics of GlcA and calcium in AGPs and how they influence plant physiological processes. Although we observed a significant reduction in glucuronidation that differs across organs in glcat14 mutants, it remains to be determined which of the GlcA residues added by the various GLCATs are modified by calcium or methyl groups or are cross-linked with other structurally complex heteropolymers such as pectins [33] to perform their cellular function. An understanding of which GlcA residues are modified and/or unmodified and how such modifications (or lack of modifications or crosslinking with other cell wall polymers) impact plant physiological functions remains to be established. Future research aiming to address such questions could potentially increase our understanding of AGP-Ca 2+ function and its possible multifunctional role in plant growth processes.