1. Expression profiles of eight Hyp-GALTs in the GT31 family
Expression of the eight Hyp-GALTs genes was examined in various vegetative and floral organs/tissues using publicly available RNA-seq data sets with Araport (Supplemental Fig. 1). This in silico analysis indicated mostly overlapping tissue expression patterns for the eight genes throughout the plant except pollen which showed unique expression pattern, corroborating previous studies based on microarray and quantitative RT-PCR data (Basu et al., 2015; Ogawa-Ohnishi and Matsubayashi, 2015; Basu et al., 2016). All the Hyp-GALTs genes were expressed in multiple plant organs/tissues including stem/aerial, carpel, inflorescence, leaf, pollen, root, seedling, root and shoot apical meristem. Among the eight genes, GALT7 generally displayed the highest expression levels, particularly in inflorescence and root tissues.
In addition, transcriptomic analysis data of laser-capture micro-dissected seeds as depicted in the BAR eFP browser  showed high but distinct expression patterns for the Hyp-GALT genes during various stages of seed development (Supplemental Fig. 2). GALT2, GALT3, GALT5 and GALT8 were highly expressed in the seed coat at the pre-globular and globular stages of seed development, while GALT4, GALT7, GALT8 and GALT9 were highly expressed in the micropylar endosperm. These expression patterns during seed development pointed towards a possibility of germination and seed coat defects in Hyp-GALT mutants.
2. Generation of higher-order Hyp-GALT knock out mutants
T-DNA insertion alleles GALT2, GALT5  as well as GALT7, GALT8 and GALT9  were used to generate multiple higher-order mutants (Fig. 1A and Supplemental Table 1). All the single alleles were reported to be T-DNA insertions in the introns, exons or UTRs: galt2-2, galt8 (hpgt2-1) and galt9 (hpgt3-1) had a T-DNA insertion in exon seven, four and one respectively, galt5-1 had T-DNA insertion in 5’UTR whereas galt7 (hpgt1-1) had a T-DNA insertion in intron six. galt2 galt5 double mutant and the galt7 galt8 galt9 (hpgt1 hpgt2 hpgt3) triple mutant, which were reported in previous studies by Basu et al. (2016) and Ogawa-Ohnishi and Matsubayashi (2015), were crossed to generate a line heterozygous for all these Hyp-GALT genes, namely galt2 galt5 galt7 galt8 galt9. This line was self-pollinated and the resulting F2 and F3 plants were subjected to PCR genotyping to isolate various triple, quadruple, and quintuple homozygous mutants (Fig. 1B). All five genes used in this study are located on different chromosomes or are far apart on the same chromosome in the Arabidopsis genome in one case. GALT2, GALT5, GALT7, GALT8, and GALT9 are located on chromosome IV, I, V, IV and II, respectively. The GALT2 and GALT8 genes located 4.27 Mb apart on the long arm of the chromosome IV according to The National Center for Biotechnology Information .
The resulting mutants obtained from the F2 and F3 screening used in this study were galt5 galt8 galt9, galt2 galt5 galt7 galt8, galt2 galt5 galt7 galt9 and galt2 galt5 galt7 galt8 galt9 (see Key in Fig. 1D) along with WT (Col-0), galt2 galt5 and galt7 galt8 galt9 as controls. Figure 1C provides an example of genotype conformation for the quintuple mutant.
3. Relative expression profiles of the Hyp-GALT genes in higher-order mutants
Quantitative reverse transcription (qRT-PCR) was performed to assess the expression of the five Hyp-GALT genes: GALT2, GALT5, GALT7, GALT8 and GALT9 genes from inflorescences collected from the plants grown on soil for 40 d (Fig. 2) which is in good agreement with qRT-PCR and RT-PCR of single allelic mutants and the 25 double mutant in previous studies [30, 31, 34]. In our study, the relative expression of the Hyp-GALT genes in the various mutants confirmed the near absence of mutant allele expression in all higher-order mutants (marked with asterisks) along with a concomitant increase or similar amount of expression of the normal alleles (with no mutation). Multiple mutations in the Hyp-GALTs generally lead to the upregulation of normal Hyp-GALTs for compensation. For example, GALT8 in 2579 was upregulated up to 2.3-fold in comparison to relative gene expression of 1 in WT.
4. Hyp-GALT mutants have reduced β-Yariv-precipitable AGPs
To investigate the effect of GALT mutations on glycosylation of AGPs, we performed quantification of β-Gal-Yariv precipitable AGPs. β-Gal-Yariv, a specific binding agent to detect, quantify, and purify AGPs (Yariv et al., 1967) via binding with β-1,3-galactan chains . AGPs were precipitated and quantified from rosette leaf, cauline leaf, stem, roots, siliques, and flowers of Hyp-GALT mutant plants.
Overall, significant reductions in β-Gal-Yariv precipitable AGPs were observed in all organs examined for the mutants (Fig. 3), which is in agreement with the expression patterns of the Hyp-GALTs in virtually all plant organs examined. The 25 double mutant produced a 10–42% decrease in precipitable AGP content compared to the WT in various plant organs. In contrast, 789 displayed a much higher reduction in β-Gal-Yariv precipitable AGPs in flowers, roots, and rosette leaves with an average of 69, 59 and 75% reductions respectively compared with WT. The 789 triple mutant also displayed a much greater reduction in AGP content in comparison to the other triple mutant 589.
The quadruple mutants 2578 and 2579 exhibited nearly similar amounts of reductions of precipitable AGPs such that stem, silique and flowers ranged between 45–50%, while cauline leaves and rosette leaves reductions ranged between 60–75%. These quadruple mutants 2578 and 2579 did not exhibit AGP reductions as high as in the stems, siliques and flowers of 789. Introducing mutation of two genes galt2 galt5 in place of one gene galt7 in combination with galt8 galt9 did not produce as much of a reduction in precipitated AGPs as in 789, indicating that 789 caused a greater effect on AGPs than other GALTs in stem, roots, silique and flower. These results are corroborated by transcript levels depicted by in silico gene expression profiles of Hyp-GALTs among different organs where GALT2 and GALT5 have lower transcript levels than GALT7 in carpel and inflorescence (Supplemental Fig. 1). However, the cauline and rosette leaves of 789, 2578, 2579 had similar levels of reduction in precipitable AGPs.
Clearly, the 25789 quintuple mutant exhibited the highest reductions in AGP precipitations such that flower, root and rosette leaf exhibited about an 80% decrease whereas other organs (stem, silique and cauline leaf) showed an ~ 70% reduction in precipitable AGPs.
5. Monosaccharide composition analysis of the Hyp-GALT mutants
To investigate the effect of introducing Hyp-GALT mutations on AGP sugar compositions, β-Gal-Yariv purified AGPs were extracted from silique and root tissues of Hyp-GALT mutants and were subjected to monosaccharide composition analysis (Fig. 4). The data showed that AGPs from all organs were mainly composed of Gal and Ara residues in approximately 1.5-2:1 molar ratio. AGPs obtained from the roots and siliques generally demonstrated a decrease in Gal content in the Hyp-GALT mutants compared to the WT.
The 25789 mutant siliques displayed a maximum reduction in Gal content of 12.5% in siliques (Fig. 4A) and 18.6 % in roots (Fig. 4B). For silique tissues, 25789 was followed by less severe Gal content reductions in the three lesser mutants (789, 2578 and 2579); while 589 and 25 showed Gal content similar to WT. In root tissues, 25789 was followed by smaller reductions in Gal content in four mutants (25, 789, 2578 and 2579); while 589 showed Gal content similar to WT. Unlike roots, the Rha in siliques also decreased significantly in higher-order mutants specifically in 25789. A concomitant increase in the percentages of other sugars like Xyl and/or Man were observed in root and silique with subtle variations, which is likely a result of expressing the data as mol percentages.
Interestingly, the calcium content bound to extracted AGPs displayed a significant reduction (12–31%) in the silique, flower, stem and root tissues of the quintuple mutant 25789 only but not in any other Hyp-GALT mutant compared to the WT (Fig. 5).
6. Hyp-GALT mutants exhibit germination defects
A significant delay in germination of the quintuple 25789 was observed under normal conditions in post-harvested (> 6–8 months) seeds. At 36 hr, only 22% germination was seen in the quintuple 25789 mutant, followed by quadruples, 2578 and 2579 demonstrating 30–33% germination rates in comparison to WT, while 25, 589 and 789 exhibited maximum germination rates (40–43%) similar to WT. By 48 hr, all genotypes germinated to their maximum germination percentages. At 48 hr, the quintuple 25789 mutant showed a maximum germination rate of 54% whereas quadruples 2578 and 2579 displayed a maximum germination rate of 80–83% in comparison to WT, 25, 589 and 789, which attained 95–98% germination (Fig. 6A and 6B).
Furthermore, the radicle length was measured at 48 hr after sowing. The 789 mutant showed a slightly negative effect on radicle length growth, though statistically non-significant. Interestingly, the quintuple mutant 25789 showed a significantly smaller radicle length (41% smaller radicle length) compared to the WT and all other Hyp-GALT mutants (Fig. 6C). In addition, for the post-harvested (> 6-8months) seeds, germination percentages reduced in quadruple (2578, 2579) by 12% and 15% which were not statistically significant though. The quintuple (25789) mutants showed significant decrease of 45% in germination percentage (Fig. 6D and 6E) in comparison to WT. For the WT, 25, 789, and 589, the germination percentage ranged between 94–99%.
7. Hyp-GALT mutants exhibit stunted plant growth when grown on soil and plates
To investigate the effect of Hyp-GALT mutations on plant vegetative growth, mutants and WT were grown on ½ MS media. The higher-order Hyp-GALT mutants exhibited pronounced pleiotropic morphological alterations in vegetative growth and bolting (Figs. 7 and 8) whereas single Hyp-GALT mutants showed no obvious phenotypes in previous studies [31, 32, 34]. The 789 triple mutant, as well as the 2578 and 2579 quadruple mutants showed significant reductions in plant height by 14–25%, while the quintuple mutant 25789 demonstrated 40% reduction in plant height compared to WT at 50 DAG (Fig. 7A and 7D). The 589 triple and 25 double mutants, however, did not show any significant effect with respect to this growth phenotype.
The days to flowering was also delayed significantly in the 789, 2578 and 2579 mutants by 3–5 d and in the quintuple mutant 25789 by 10–11 d (Fig. 7A and 7C). The higher-order Hyp-GALT mutants also demonstrated retarded growth of rosette leaves as seen in Fig. 7B; this is evident in 789, 2578, 2579, and is the most severe in 25789 (Supplemental Fig. 3).
The 2578 and 2579 quadruple mutants showed no significant primary root growth. The 789 mutants showed significant reduction (21%) in primary root growth while the 25789 quintuple mutants displayed the most retarded primary root growth (38%) (Fig. 8A and B). In contrast, the root hair density and root-hair length were not affected much (Fig. 8C and Supplemental Fig. 4) in these mutants with the exception of the 25789 mutant which showed a small increase in root hair density and a small decrease in root hair length.
The higher-order Hyp-GALT mutants showed much more pronounced salt hypersensitivity response in form of root tip swelling compared to previously reported in single and double Hyp-GALT mutants . All of the higher-order Hyp-GALT mutants displayed root tip swelling and decreased root elongation compared to WT plants; however, 589, 2578, 2579 showed the less severe root growth defects in comparison to 789 and 25789 (Fig. 8D and Supplemental Fig. 5). β-Yariv reagent is known to inhibit root growth by binding to AGPs [8, 39]. Seedlings of all genotypes were grown in the presence of β-Gal-Yariv reagent such that WT seedlings displayed reduced root growth as expected (Fig. 8E). In contrast, all Hyp-GALT mutants showed a β-Yariv insensitive root growth phenotype; 789 and 25789 displayed the highest β-Yariv insensitivity with respect to root growth (Fig. 8F) as quantified 7d and 14d after transfer to β-Yariv.
8. Effect on cell wall structure and AGP profiling of 25789 quintuple mutant stems
Since the 25789 mutants exhibited stunted seedling growth and shortened inflorescence stems (Fig. 7), transverse sections of fresh tissue from the base of the 8-week-old inflorescence stem were stained with toluidine blue for cell wall polysaccharides, to visualize differences in primary and/or secondary cell wall morphology (Fig. 9). The results revealed that 25789 mutant stems (Fig. 9A, 9B) have smaller vascular bundles (Fig. 9C, 9D) with reduced thickness of fiber cells, xylem fibers, vessels and interfascicular fibers stems (Fig. 9F, 9H) compared to the WT (Fig. 9E, 9G). Moreover, transmission electron microscope (TEM) analysis of cross sections confirmed thinner secondary cell walls in the vessels, vascular fibers and interfascicular fibers of the 25789 mutants (Fig. 9H, 9L) relative to WT (Fig. 9G, 9K).
β-Gal-Yariv precipitable AGPs of 25789 mutant stems were profiled by reverse phase-HPLC (RP-HPLC). A significant difference in the AGP profile was observed between mutant and WT, suggesting either the mutant contained a different subset of AGPs and/or that AGP glycosylation was significantly altered (Fig. 9N, 9O).
9. Hyp-GALT mutants display reduced seed set and abnormal seed morphology
The higher-order Hyp-GALT mutants displayed a significant reduction in total siliques/plant in the 25789 mutants (25%) in comparison to WT unlike the other mutants (Fig. 10C). The 25789 mutant also demonstrated most dramatic reduction in seed set (~ 70%) compared to 14–15 % reductions in the average seed set for the 789, 2578, 2579 mutants; (Fig. 10A, 10B and 10E). The analysis of basal (oldest) fifteen siliques on the main inflorescence stem also showed the reduction in seed set of 25789 mutant compared to the WT (Fig. 10A and S6). The average silique length was also reduced in the higher-order Hyp-GALT mutants (Fig. 10A, 10B and 10D). The average silique length was affected more for 789 (22% reduction) than both 2578 and 2579, which showed 11% reductions; however, 25789 exhibited a 60% reduction (Fig. 10E and Supplemental Fig. 7).
SEM seed morphology analysis revealed that the quintuple mutant exhibited altered seed shape and disfigured remnants of columellae. No such discernable differences were found in other mutant seeds (Fig. 11A and 11B). To examine the involvement of AGPs and the Hyp-GALTs in modifying seed coat mucilage, ruthenium red staining which stains acidic biopolymers such as pectin, and calcofluor staining, which stains cellulose as well as β-glucans, were done with the Hyp-GALT mutants (Fig. 11C and 11D). All Hyp-GALT mutants displayed reduced cellulose ray staining and reduced pectin staining in the mucilage adhering to the seeds compared to the WT with 25789 displaying the strongest reduction in seed mucilage pectin and cellulose staining. To examine and quantify the alterations in the outer, non-adherent mucilage versus the adherent mucilage, sequential extraction of seeds with ammonium oxalate and 0.2 N NaOH (for extraction of soluble and weakly attached pectins) followed by 2 N NaOH (for extraction of strongly linked pectins and cross-linking glycans/hemicelluloses) was performed  (Supplemental Table 4). The higher-order Hyp-GALT mutant seeds had a significant increase in the total sugar present in the ammonium oxalate and 0.2 N NaOH extracts with a concomitant decrease in the adherent mucilage compared to WT seeds. These observations corroborate the results of ruthenium red staining which also suggested a decrease in adherent mucilage of the higher-order Hyp-O-GALT mutants.
10. Hyp-GALT mutants display anther and pollen defects
Hyp-GALT genes are highly expressed in the inflorescence and in the pollen (Supplemental Fig. 1). Moreover, as previous studies on single Hyp-GALT mutants also demonstrated pollen tube growth defects , we microscopically examined our higher order Hyp-GALT mutants for phenotypic differences in male reproductive tissues. As shown in the Supplemental Fig. 8, the arrangement of floral reproductive organs appeared indistinguishable from the WT plants, although the overall flower size was smaller in the 25789 mutant.
In vitro pollen germination was affected only for 25789, which showed 50% germination in comparison to 77% germination in the WT (Fig. 12A and 12C). For pollen tube lengths, an 8–10% reduction was observed for mutants, 789, 2578, and 2579, whereas 25789 exhibited a 44% reduction compared to WT (Fig. 12D). A defective pollen phenotype was also observed in the Hyp-GALT mutants (Fig. 12G, Fig. 13). Around 10–12% of defective pollen was observed in the 789, 2578, and 2579 mutants, whereas the 25789 mutant exhibited 20% defective pollen. Further, SEM analysis revealed defects in 25789 with abnormal exine structure with smaller lacunae and abnormal reticulate structure (Fig. 13).
789, 589, 2578, 2579, and 25789 mutants also displayed reduced inhibition of pollen tube growth in response to β-Gal-Yariv reagent compared to WT (Fig. 12B, 12E and 12F); maximum pollen tube elongation was observed in 25789 which was 44% longer than WT.