UDP-GLYCOSYLTRANSFERASE 72E3 Plays a Role in Lignification of Secondary Cell Walls in Arabidopsis

Lignin is present in plant secondary cell walls and is among the most abundant biological polymers on Earth. In this work we investigated the potential role of the UGT72E gene family in regulating lignification in Arabidopsis. Chemical determination of floral stem lignin contents in ugt72e1, ugt72e2, and ugt72e3 mutants revealed no significant differences compared to WT plants. In contrast, the use of a novel safranin O ratiometric imaging technique indicated a significant increase in the cell wall lignin content of both interfascicular fibers and xylem from young regions of ugt72e3 mutant floral stems. These results were globally confirmed in interfascicular fibers by Raman microspectroscopy. Subsequent investigation using a bioorthogonal triple labelling strategy suggested that the augmentation in lignification was associated with an increased capacity of mutant cell walls to incorporate H-, G-, and S-monolignol reporters. Expression analysis showed that this increase was associated with an up-regulation of LAC17 and PRX71, which play a key role in lignin polymerization. Altogether, these results suggest that UGT72E3 can influence the kinetics of lignin deposition by regulating monolignol flow to the cell wall as well as the potential of this compartment to incorporate monomers into the growing lignin polymer.


Introduction
Lignin is a major structural component of many plant cell walls. The appearance of this polymer is associated with terrestrialization since it is responsible for the mechanical support, essential for upright growth, but also for providing a hydrophobic environment to the cells specialized in the conduction of water [1]. As well as playing a central role in the biology of vascular plants, lignin also has a major impact on the quality of a wide range of economically important products derived from plant lignocellulosic biomass [2]. The composition of the lignin polymer can be very complex but the major constituents derive from the oxidative coupling of three p-hydroxycinnamyl alcohols (monolignols), p-coumaryl, coniferyl and sinapyl alcohols, differing in their degree of methoxylation.
although whether these molecules can be further transported through the phloem towards other parts of the plant remains an open question.
When considering these different data, the regulatory role of the glycosylation on the availability of lignin precursors and more generally on the lignification process still remains unclear and deserves further investigation. Indeed, UGTs have been mentioned as actors of lignification [26,27] but to date, there are only very few experimental data supporting this hypothesis. In tobacco, the ectopic expression of a Populus tomentosa UGT gene controlled by the CaMV35S promoter increased the amounts of Klason lignin by approximately 60% [28]. This unexpected result showed that UGTs may have indirect effect on lignin deposition. In more recent studies, the Arabidopsis UGT72B1 gene was knocked-out, leading to an ectopic lignification phenotype [29] that was more consistent with the hypothesis of a role played by UGTs in the aglycone lignin precursor homeostasis.
In this work, we investigated the role of the 3 UGT72E family members in the regulation of lignification in Arabidopsis thaliana. We first showed that the expression of UGT72E3 is independent of that of the 2 others and that the dynamics of lignin deposition in the corresponding mutant was modified compared to the WT. This was assessed by Raman microspectroscopy and a histological ratiometric approach. We then evaluated the cell wall capacity to integrate monolignols with a bioorthogonal reporter incorporation technique and clearly showed that it was increased in the ugt72e3 mutant and associated with peroxidase and laccase gene upregulation.

Impact of UGT72E Mutations on Plant Growth
To obtain information about the biological role of the different UGT72E genes, the corresponding insertion mutants were investigated for modified phenotypic characters. Although no marked phenotype was visible, specific measurements of the rosette surface of 28-day-old plants (corresponding to the 3.20-3.50 stage [30]) revealed a slight but significant reduction in rosette area for ugt72e1, but not for ugt72e2 and ugt72e3 mutants when compared to WT plants ( Figure 1A) suggesting a possible negative effect of the ugt72e1 mutation on vegetative growth. Phenotypic differences were also observed at stage 5.1 corresponding to the emergence of the floral stem ( Figure 1B). The number of days necessary to reach this stage was significantly increased in the ugt72e1 mutants, unchanged in ugt72e2 and decreased in ugt72e3 compared to WT plants. These differences were also reflected in the distribution of floral stem appearance times observed for the majority of plants within compared genotypes ( Figure 1C). Both WT and ugt72e2 populations show a progressive flowering profile with a maximum at day 44 whereas the two other mutants exhibit a narrower flowering peak occurring either prior to WT (ugt72e3) or after (ugt72e1), followed by a rapid decline. Taken together these results would suggest that the knock-out of the UGT72E1 and UGT72E3 genes have a discrete but significant effect on plant development.

Expression of the UGT72E Genes in the WT and Mutant Floral Stems
To better understand the possible involvement of the UGT72E genes in floral stem development and to see whether the inactivation of one of the 3 genes may affect the expression of other UGT72E family members, expression analyses were performed on the top and bottom regions of this organ corresponding respectively to "young" and "old" tissues ( Figure 2).
In WT floral stem tissues, all three genes were expressed albeit with different profiles. UGT72E1 was more highly expressed (approx. 4×) in old tissues compared to young tissues while the expression of UGT72E2 and UGT72E3 were similar in both stem regions. In the ugt72e2 and ugt72e3 mutants, UGT72E1 transcript accumulation was slightly but significantly increased in the old part, but not the young part of floral stems as compared to WT. In contrast the accumulation of UGT72E2 transcripts was only significantly increased in the ugt72e1 mutant (both young and old zones), but not ugt72e3 as  Expression analyses of UGT72E genes in wild type (WT) and ugt72e mutants. RT-qPCR was used to measure the expression of the genes in the old and young parts of the stem. The values are expressed relative to the WT old stem fragments for each analyzed gene. The expression of the genes corresponding to the mutants is null so was not tested statistically. PP2A (AT1G13320) and PEROXIN4 (AT5G25760) were used as internal reference genes. Means with the same lowercase letters are not significantly different according to Duncan test (p-value < 0.05, n = 3).

In Situ Analysis Reveals Increased Cell Wall Lignin Content Only in Ugt72e3 Mutant Stems
To see whether the inactivation of different UGT72E genes modified lignification, we used a dual approach consisting of (i) a global quantification by acetyl bromide and (ii) an evaluation of relative cellular lignin quantities by a histological ratiometric approach. In both cases, the analyses were performed on the young and old regions of 24 cm-high floral stems. When using the acetyl bromide procedure, no significant differences in lignin contents were observed between the different lines in  Expression analyses of UGT72E genes in wild type (WT) and ugt72e mutants. RT-qPCR was used to measure the expression of the genes in the old and young parts of the stem. The values are expressed relative to the WT old stem fragments for each analyzed gene. The expression of the genes corresponding to the mutants is null so was not tested statistically. PP2A (AT1G13320) and PEROXIN4 (AT5G25760) were used as internal reference genes. Means with the same lowercase letters are not significantly different according to Duncan test (p-value < 0.05, n = 3).

In Situ Analysis Reveals Increased Cell Wall Lignin Content Only in Ugt72e3 Mutant Stems
To see whether the inactivation of different UGT72E genes modified lignification, we used a dual approach consisting of (i) a global quantification by acetyl bromide and (ii) an evaluation of relative cellular lignin quantities by a histological ratiometric approach. In both cases, the analyses were performed on the young and old regions of 24 cm-high floral stems. When using the acetyl bromide procedure, no significant differences in lignin contents were observed between the different lines in  Figure 2. Expression analyses of UGT72E genes in wild type (WT) and ugt72e mutants. RT-qPCR was used to measure the expression of the genes in the old and young parts of the stem. The values are expressed relative to the WT old stem fragments for each analyzed gene. The expression of the genes corresponding to the mutants is null so was not tested statistically. PP2A (AT1G13320) and PEROXIN4 (AT5G25760) were used as internal reference genes. Means with the same lowercase letters are not significantly different according to Duncan test (p-value < 0.05, n = 3).

In Situ Analysis Reveals Increased Cell Wall Lignin Content Only in Ugt72e3 Mutant Stems
To see whether the inactivation of different UGT72E genes modified lignification, we used a dual approach consisting of (i) a global quantification by acetyl bromide and (ii) an evaluation of relative Int. J. Mol. Sci. 2020, 21, 6094 5 of 22 cellular lignin quantities by a histological ratiometric approach. In both cases, the analyses were performed on the young and old regions of 24 cm-high floral stems. When using the acetyl bromide procedure, no significant differences in lignin contents were observed between the different lines in both parts of the stems ( Figure 3). The lignin amounts (g/g CWR) were roughly 2-fold higher in bottom stem fragments compared to the top ones whatever the genotype considered. both parts of the stems (Figure 3). The lignin amounts (g/g CWR) were roughly 2-fold higher in bottom stem fragments compared to the top ones whatever the genotype considered. Since global wet chemistry measurements can mask subtle, but real changes in cell wall lignification, we used a histological approach based on our recent development of a quantitative in situ technique coupling fluorescent microscopy and safranin O staining [31]. There were no significant differences in the lignin content of both interfascicular fibers and xylem from young and old regions of the ugt72e1 and ugt72e2 mutant stems compared to the WT (Figure 4 and Figure 5). In contrast, the lignin content of the ugt72e3 mutant was significantly higher (approx. 40 %) in both tissues from the young part of the stems compared to WT stems. In the old part of the ugt72e3 stem, the lignin content was similar to the WT.
These results suggest the existence of a functional link between UGT72E3 gene expression (but not UGT72E1 or UGT72E2) and lignification. To confirm the specificity of the UGT72E3 gene mutation effect and evaluate a possible functional redundancy with UGT72E1 and UGT72E2, lignin was quantified in the double ugt72e1,2 mutant used as a control. Indeed, mutant complementation is often used to gain knowledge on the role of a particular gene, but has many disadvantages such as position effect of the T-DNA [32] carrying the non-mutated gene along with the variability of transgene expression by RNA sensing mechanisms [33] but also possible influences of environmental parameters [34].
We first showed that the expression level of UGT72E3 was the same as in the WT ( Figure S1). Concerning lignin dosage, our results revealed no significant differences in floral stem lignin content between WT and ugt72e1,2 samples when evaluated by either the acetyl bromide method ( Figure S2) or the safranin O technique ( Figure S3). Taken together, these data demonstrate that the observed modifications in cell wall lignin content are most likely caused by UGT72E3 gene knock-out. Since global wet chemistry measurements can mask subtle, but real changes in cell wall lignification, we used a histological approach based on our recent development of a quantitative in situ technique coupling fluorescent microscopy and safranin O staining [31]. There were no significant differences in the lignin content of both interfascicular fibers and xylem from young and old regions of the ugt72e1 and ugt72e2 mutant stems compared to the WT (Figures 4 and 5). In contrast, the lignin content of the ugt72e3 mutant was significantly higher (approx. 40 %) in both tissues from the young part of the stems compared to WT stems. In the old part of the ugt72e3 stem, the lignin content was similar to the WT.
These results suggest the existence of a functional link between UGT72E3 gene expression (but not UGT72E1 or UGT72E2) and lignification. To confirm the specificity of the UGT72E3 gene mutation effect and evaluate a possible functional redundancy with UGT72E1 and UGT72E2, lignin was quantified in the double ugt72e1,2 mutant used as a control. Indeed, mutant complementation is often used to gain knowledge on the role of a particular gene, but has many disadvantages such as position effect of the T-DNA [32] carrying the non-mutated gene along with the variability of transgene expression by RNA sensing mechanisms [33] but also possible influences of environmental parameters [34].
We first showed that the expression level of UGT72E3 was the same as in the WT ( Figure S1). Concerning lignin dosage, our results revealed no significant differences in floral stem lignin content between WT and ugt72e1,2 samples when evaluated by either the acetyl bromide method ( Figure S2) or the safranin O technique ( Figure S3). Taken together, these data demonstrate that the observed modifications in cell wall lignin content are most likely caused by UGT72E3 gene knock-out.

Raman Microspectroscopy Reveals Modifications in Ugt72e3 Cell Walls
To validate the results obtained by the safranin O methodology and to further explore effects of UGT72E3 knock-out on cell wall phenotype we used Raman microspectroscopy.
We firstly acquired spectra of interfascicular fiber cell walls from floral stem sections prepared from WT and ugt72e3 genotypes and generated mean spectra for analysis ( Figure S4). Spectra obtained from xylem cell walls were found to be highly variable due to the heterogeneity of the cell types (e.g., vessels, fibers, parenchyma) and sizes within this tissue and it was therefore impossible to generate suitable mean spectra for meaningful comparison of WT and ugt72e3 xylem cell walls within the context of this study. The comparison of WT and ugt72e3 baseline-corrected spectra of interfascicular fiber cell walls from young stem regions ( Figure 6A) revealed important differences between the two spectra. Four lignin-associated peaks (1269 cm −1 , 1333 cm −1 , 1595 cm −1 , and 1652 cm −1 ) were higher in the ugt72e3 spectra compared to WT. Similar results were also obtained for major peaks that could be assigned to cellulose (375 cm −1 , 1092 cm −1 , and 1120 cm −1 ). In contrast only slight/no differences were observed for peaks related to other major cell wall polymers such as xylans (432 cm −1 , 454 cm −1 , 489 cm −1 , and 516 cm −1 ). In WT spectra 3 peaks (1005 cm −1 , 1158 cm −1 , and 1521 cm −1 ) that were assigned to carotenoids and/or xanthophyll pigments [35] were more intense compared to the mutant spectra.
In order to confirm that the observed differences were statistically relevant we performed principal component analysis (PCA) on the spectra corresponding to the different WT and mutant samples analyzed. Our results show that the first principal component (PC1) represented 88.89% of the total differences observed ( Figure 6B). Analyzing scores on PC1 showed that all WT and mutant

Raman Microspectroscopy Reveals Modifications in Ugt72e3 Cell Walls
To validate the results obtained by the safranin O methodology and to further explore effects of UGT72E3 knock-out on cell wall phenotype we used Raman microspectroscopy.
We firstly acquired spectra of interfascicular fiber cell walls from floral stem sections prepared from WT and ugt72e3 genotypes and generated mean spectra for analysis ( Figure S4). Spectra obtained from xylem cell walls were found to be highly variable due to the heterogeneity of the cell types (e.g., vessels, fibers, parenchyma) and sizes within this tissue and it was therefore impossible to generate suitable mean spectra for meaningful comparison of WT and ugt72e3 xylem cell walls within the context of this study. The comparison of WT and ugt72e3 baseline-corrected spectra of interfascicular fiber cell walls from young stem regions ( Figure 6A) revealed important differences between the two spectra. Four lignin-associated peaks (1269 cm −1 , 1333 cm −1 , 1595 cm −1 , and 1652 cm −1 ) were higher in the ugt72e3 spectra compared to WT. Similar results were also obtained for major peaks that could be assigned to cellulose (375 cm −1 , 1092 cm −1 , and 1120 cm −1 ). In contrast only slight/no differences were observed for peaks related to other major cell wall polymers such as xylans (432 cm −1 , 454 cm −1 , 489 cm −1 , and 516 cm −1 ). In WT spectra 3 peaks (1005 cm −1 , 1158 cm −1 , and 1521 cm −1 ) that were assigned to carotenoids and/or xanthophyll pigments [35] were more intense compared to the mutant spectra. When compared to the WT samples, the interfascicular fiber cell walls from young stem regions of the ugt72e3 mutant incorporated significantly higher amounts of all three reporters ( Figure 7A,B) suggesting that the lignification process was modified in this mutant. Examination of xylem from the same region also revealed changes in the lignification process ( Figure 7C,D). Here, mutant cell walls incorporated significantly higher amounts of Galk-and Scp-reporters compared to WT samples. Only the incorporation of Haz was not significantly different between WT and mutant xylem cell walls.
In older parts of the stem, reporter incorporation into interfascicular fiber cell walls was altered in ugt72e3 samples indicating that the lignification process was also modified ( Figure 8A,B). However, in contrast to the pattern observed in the young part, here significantly lower amounts of In order to confirm that the observed differences were statistically relevant we performed principal component analysis (PCA) on the spectra corresponding to the different WT and mutant samples analyzed. Our results show that the first principal component (PC1) represented 88.89% of the total differences observed ( Figure 6B). Analyzing scores on PC1 showed that all WT and mutant samples were well discriminated on this component. Examination of the PC1 loading vector ( Figure 6C) indicated that the peaks making positive contributions (1005 cm −1 , 1158 cm −1 , and 1521 cm −1 ) correspond to pigments while peaks making negative contributions correspond to lignin (1269 cm −1 , 1333 cm −1 , 1595 cm −1 , and 1652 cm −1 ) and cellulose (375 cm −1 , 1092 cm −1 , and 1120 cm −1 ) in agreement with the average spectra ( Figure 6A). The detection of carotenoids during our analyses is probably due to our technical procedure during which the young tissues containing high amounts of photosynthetic pigments were only fixed during a short period in ethanol. As a consequence, some cellular contents, including pigments, probably interacted with the cell wall material after cutting the tissues with the vibratome. The differences in the degree of contamination between the WT and the ugt72e3 mutant may possibly be related to a general modification of the cell wall architecture.
The mean spectra ( Figure 6D) obtained from old WT and ugt72e3 floral stem regions were much more similar than corresponding spectra obtained from younger samples ( Figure 6A). Nevertheless, when analyzing scores of different samples on the PC1 that accounted for 72.3% of the total variability ( Figure 6E), we observed that the WT and ugt72e3 mutant could still be discriminated (but less so than in younger samples). Examination of the corresponding PC1 loading vector ( Figure 6F) indicated that peaks making positive contributions could be assigned to lignin (1141 cm −1 , 1268 cm −1 , 1331 cm −1 , 1596 cm −1 , and 1654 cm −1 ). Altogether the data obtained by Raman microspectroscopy are in agreement with the results obtained using the safranin O approach. Both methods indicate that cell wall lignin content is significantly increased in the ugt72e3 mutant compared to WT and that the difference is greater in samples from young floral stem regions.

Monolignol Reporter Incorporation is Increased in Ugt72e3 Stems
In the light of the observed local increases in ugt72e3 cell wall lignin content revealed by both the safranin O approach and Raman spectroscopy, we compared the capacity of the WT and ugt72e3 mutant to incorporate monolignol reporters into the growing lignin polymer in situ using a recently developed bioorthogonal triple labelling strategy [36]. This technology gives detailed information about the potential of the cell wall machinery to incorporate tag-bearing monolignol surrogates (chemical reporters) and therefore provides dynamic information about the lignification process in cell walls.
The use of three different chemical reporters H az , G alk and S cp ( Figure S5) and independent bioorthogonal reactions allowed us to specifically evaluate the incorporation of lignin units corresponding to p-coumaryl alcohol, coniferyl alcohol, and sinapyl alcohol molecules respectively. The results are presented as normalized fluorescence intensities extracted from confocal microscopy images for the young (Figure 7) and old ( Figure 8) regions of WT and ugt72e3 mutant floral stems.
When compared to the WT samples, the interfascicular fiber cell walls from young stem regions of the ugt72e3 mutant incorporated significantly higher amounts of all three reporters ( Figure 7A,B) suggesting that the lignification process was modified in this mutant. Examination of xylem from the same region also revealed changes in the lignification process ( Figure 7C,D). Here, mutant cell walls incorporated significantly higher amounts of G alk -and S cp -reporters compared to WT samples. Only the incorporation of H az was not significantly different between WT and mutant xylem cell walls.
In older parts of the stem, reporter incorporation into interfascicular fiber cell walls was altered in ugt72e3 samples indicating that the lignification process was also modified ( Figure 8A,B). However, in contrast to the pattern observed in the young part, here significantly lower amounts of G alk -and S cp -reporters (but not H az -reporters) were incorporated compared to WT samples. Monolignol reporter incorporation was also significantly increased for H az -, G alk -, and S cp -reporters in ugt72e3 xylem cell walls in comparison with the WT (Figure 8C,D). Taken together, these results clearly suggest that the knock-out of the UGT72E3 gene induces modifications in the lignification process.
Galk-and Scp-reporters (but not Haz-reporters) were incorporated compared to WT samples. Monolignol reporter incorporation was also significantly increased for Haz-, Galk-, and Scp-reporters in ugt72e3 xylem cell walls in comparison with the WT (Figure 8C,D). Taken together, these results clearly suggest that the knock-out of the UGT72E3 gene induces modifications in the lignification process.

Ugt72e3 Mutation has no Influence on the Expression of the UGT72B Genes
Our results obtained by three independent methodological approaches (safranin O, Raman, reporter chemistry) all indicated changes to cell wall phenotype in the ugt72e3 mutant suggesting that UGT72E3 plays a role in cell wall metabolism. However, it was recently shown that glycosyltransferases from the UGT72B subfamily shared common biochemical functions with UGT72E2 in Arabidopsis thaliana [29]. Like the UGT72E genes, UGT72B1 is expressed in stems and its knock-out modified the expression of not only UGT72B3 but also UGT72E2, suggesting an interconnection between the UGT72E and UGT72B subfamilies. In order to rule out the possibility that the observed ugt72e3 phenotype was due to an indirect effect via the UGT72B genes, we investigated the effect of knocking-out the UGT72E3 gene on their expression in floral stems. As shown in Figure 9, the transcript accumulation of all three UGT72B genes is not significantly modified in the ugt72e3 mutant thereby demonstrating that the observed phenotype is not caused by an indirect effect on other lignin-related UGT72 genes.

Ugt72e3 Mutation has no Influence on the Expression of the UGT72B Genes
Our results obtained by three independent methodological approaches (safranin O, Raman, reporter chemistry) all indicated changes to cell wall phenotype in the ugt72e3 mutant suggesting that UGT72E3 plays a role in cell wall metabolism. However, it was recently shown that glycosyltransferases from the UGT72B subfamily shared common biochemical functions with UGT72E2 in Arabidopsis thaliana [29]. Like the UGT72E genes, UGT72B1 is expressed in stems and its knock-out modified the expression of not only UGT72B3 but also UGT72E2, suggesting an interconnection between the UGT72E and UGT72B subfamilies. In order to rule out the possibility that the observed ugt72e3 phenotype was due to an indirect effect via the UGT72B genes, we investigated the effect of knocking-out the UGT72E3 gene on their expression in floral stems. As shown in Figure 9, the transcript accumulation of all three UGT72B genes is not significantly modified in the ugt72e3 mutant thereby demonstrating that the observed phenotype is not caused by an indirect effect on other lignin-related UGT72 genes.

The Expression of PRX and LAC Genes is Up-Regulated in Ugt72e3 Mutants
Results obtained with safranin O, Raman and chemical reporter approaches all indicated modified lignification profiles in the ugt72e3 mutant compared to WT plants. In particular, the chemical reporter strategy suggested that mutants incorporated tagged monolignols to higher levels in the young stem region. Since the formation and extension of the lignin polymer depends mainly on the activities of class III PRX and LAC enzymes we performed RT-qPCR analyses of selected lignin-related redox genes in ugt72e3 and WT samples. PRX25 and PRX71 genes were chosen because their knock-out mutants are characterized by a decrease in total lignins, thinner cell walls, and/or altered lignin structures in stems [37,38]. PRX64 was also selected since it showed a specific expression in the stems [39] and the corresponding protein was localized in the middle lamella and cell corners of cell walls [40]. We also selected LAC4 and LAC17 that have been proposed as key actors in stem lignification [41].
In the WT, no differences in the accumulation of the selected PRX and LAC transcripts were revealed when comparing young and old stem regions ( Figure 10). When comparing WT and ugt72e3 mutant plants in older stem regions, PRX25 and PRX71 showed significant increases in transcript accumulation whereas in young stem samples, the accumulation was higher for PRX71 and LAC17. These data clearly indicate that the expression of several key lignin-related redox genes is upregulated in the floral stems of ugt72e3 mutants and that the genes possibly involved in monolignol activation are different depending on the age of the stem portion.

The Expression of PRX and LAC Genes Is Up-Regulated in Ugt72e3 Mutants
Results obtained with safranin O, Raman and chemical reporter approaches all indicated modified lignification profiles in the ugt72e3 mutant compared to WT plants. In particular, the chemical reporter strategy suggested that mutants incorporated tagged monolignols to higher levels in the young stem region. Since the formation and extension of the lignin polymer depends mainly on the activities of class III PRX and LAC enzymes we performed RT-qPCR analyses of selected lignin-related redox genes in ugt72e3 and WT samples. PRX25 and PRX71 genes were chosen because their knock-out mutants are characterized by a decrease in total lignins, thinner cell walls, and/or altered lignin structures in stems [37,38]. PRX64 was also selected since it showed a specific expression in the stems [39] and the corresponding protein was localized in the middle lamella and cell corners of cell walls [40]. We also selected LAC4 and LAC17 that have been proposed as key actors in stem lignification [41].
In the WT, no differences in the accumulation of the selected PRX and LAC transcripts were revealed when comparing young and old stem regions ( Figure 10). When comparing WT and ugt72e3 mutant plants in older stem regions, PRX25 and PRX71 showed significant increases in transcript accumulation whereas in young stem samples, the accumulation was higher for PRX71 and LAC17. These data clearly indicate that the expression of several key lignin-related redox genes is up-regulated in the floral stems of ugt72e3 mutants and that the genes possibly involved in monolignol activation are different depending on the age of the stem portion.

Discussion
Glycosyl-transferase (GT) mediated reactions are considered as among the most important biotransformations on Earth [42]. To date, 110 different GT families are referenced in the CAZy database among which are the UGTs (uridine diphosphate glucosyltransferases) belonging to the GT1 family. There is an important mismatch between the number of glycosylated specialized metabolites in plants, potentially more than 100,000 [43], and the limited number of UGT genes, meaning that the corresponding enzymes can very certainly act on many different substrates. In Arabidopsis thaliana, this high functional diversity is reflected by the existence of 21 gene families based on sequence homology (UGT71-92 but not 77). The phylogenetic analysis of the 123 UGT protein sequences showed that they could be divided in 14 groups (A to N) [44]. Since many of the corresponding proteins are still insufficiently characterized, it remains difficult to know whether each group reflects common functions.
Over the last two decades UGT72E and, more recently UGT72B proteins belonging to the group E have received attention because of their possible role in regulating not only phenylpropanoid metabolism, but also deposition of the important cell wall polymer lignin. In terms of biochemical functions, in vitro tests of enzymatic activities indicate that both UGT72E1 and UGT72E2 can transfer a glucosyl group onto coniferaldehyde and sinapaldehyde and that UGT72E2 and UGT72E3 can transform sinapic acid but also sinapyl and coniferyl alcohols [19]. Ferulic acid can also be

Discussion
Glycosyl-transferase (GT) mediated reactions are considered as among the most important biotransformations on Earth [42]. To date, 110 different GT families are referenced in the CAZy database among which are the UGTs (uridine diphosphate glucosyltransferases) belonging to the GT1 family. There is an important mismatch between the number of glycosylated specialized metabolites in plants, potentially more than 100,000 [43], and the limited number of UGT genes, meaning that the corresponding enzymes can very certainly act on many different substrates. In Arabidopsis thaliana, this high functional diversity is reflected by the existence of 21 gene families based on sequence homology (UGT71-92 but not 77). The phylogenetic analysis of the 123 UGT protein sequences showed that they could be divided in 14 groups (A to N) [44]. Since many of the corresponding proteins are still insufficiently characterized, it remains difficult to know whether each group reflects common functions.
Over the last two decades UGT72E and, more recently UGT72B proteins belonging to the group E have received attention because of their possible role in regulating not only phenylpropanoid metabolism, but also deposition of the important cell wall polymer lignin. In terms of biochemical functions, in vitro tests of enzymatic activities indicate that both UGT72E1 and UGT72E2 can transfer a glucosyl group onto coniferaldehyde and sinapaldehyde and that UGT72E2 and UGT72E3 can transform sinapic acid but also sinapyl and coniferyl alcohols [19]. Ferulic acid can also be glycosylated by UGT72E2. The characterization of the UGT72B1 protein also showed that while the enzymatic activity was redundant with that of UGT72E2 [29], the Km value was higher in the former.
While a recent study [29] demonstrated that UGT72B1 is essential for normal cell wall lignification in Arabidopsis, the situation is less clear for members of the UGT72E family. Analyses of mutants over-/under-expressing UGT72E1/2/3 genes indicated changes in the levels of monolignol glucosides however, no significant effect on lignin deposition was observed [45]. In this paper we decided to reevaluate the potential involvement of the UGT72E family in lignification using chemical quantification together with several powerful in situ imaging approaches to characterize the Arabidopsis ugt72e1, ugt72e2, and ugt72e3 mutants.
Lignin contents were estimated using acetyl bromide. This dosage protocol was first developed for small wood samples [46] and later adapted for herbaceous species. It is particularly appropriate for lignin estimation in the floral stem of Arabidopsis because of the low amounts of cell wall xylans, which can potentially interfere with the spectrometric absorption [47] when using perchloric acid during the lignin extraction. In our protocol, the absence of artifactual furans [48] was also shown by checking the absence of a shift in the absorption spectrum to wavelengths lower than 280 nm. In our hands, this approach did not reveal any differences in lignin content between the ugt72e mutants and wild type. Chemical estimations of lignin, however, are generally destructive, providing an average value and resulting in the loss of detailed spatial information about cell wall lignin levels [49]. We therefore decided to use complementary in situ imaging techniques to investigate whether UGT72E gene knock-out affected cell wall lignin.
We firstly implemented a new quantification approach that we have developed based upon the potential of fluorescent microscopy coupled to safranin O staining [31]. Using this technique, it is now possible to generate high-definition spatial information about lignin localization as well as quantitative information based on ratiometric emission measurements. Our results showed that lignin deposition was significantly changed only in the ugt72e3 mutant, but not in ugt72e1 or ugt72e2. Lignin levels in both the xylem and interfascicular fibers from the younger part of ugt72e3 stems were higher when compared to the WT. In contrast, no statistical differences were observed in the bottom part of the organ that contains more mature cells.
It seems therefore that the knock-out of the UGT72E3 gene accelerates cell wall lignification in younger tissues but does not influence the final amounts at the end of cell wall development. In order to confirm this observation we then used Raman microspectroscopy and a chemical reporter strategy [49]. Both of these approaches confirmed that cell wall lignin levels were significantly increased in ugt72e3 mutants compared to WT plants and provided strong evidence for the idea that the UGT72E3 gene is involved in the regulation of cell wall lignification. Expression analyses showed that the accumulation of transcripts for the 3 UGT72B family members were not significantly modified in the mutant thereby ruling out the possibility that the phenotype was due to an indirect effect on the UGT72B genes.
The existence of a possible relationship between UGTs and lignification has been often mentioned but is only poorly supported by experimental data [13,26,27,50]. In tobacco plants, the ectopic expression of the Populus tomentosa PtGT1, homologous to the 3 UGT72E Arabidopsis genes, led to the accumulation of higher lignin levels in the xylem [28]. The interpretation of this result is not really intuitive when considering that UGTs may reduce the flux of monolignols towards the cell wall. It is possible that the use of the strong CaMV35S promoter to control UGT gene expression led to significant disruption of gene function in this case. It should also be taken in account that no enzymatic activity towards monolignols was associated with the corresponding recombinant protein, suggesting that the impact of this UGT on lignin biosynthesis is not directly through the glycosylation of monolignols, In Arabidopsis, the ugt72b1 mutant was characterized by ectopic lignification in areas around pith tissues and interfascicular fibers [29]. The increase in the quantities of lignins in the cell walls does agree with the fact that this UGT could catalyze the glucose conjugation of monolignols but still, the very strong impact of the mutation is surprising for this gene.
The ectopic expression of PtGT1 in tobacco resulted in an early flowering phenotype [28]. In our work, the development analysis of the ugt72e3 mutant also showed that it had an advanced flowering phenotype. This relationship between the development of the floral stem and its lignin contents is difficult to explain. In different plant species, the flowering time has often been linked to lignification when searching for a marker of the biomass yield. In Medicago truncatula, a delayed flowering mutant had higher biomass and lower lignin amounts compared to the wild type at the same age [51]. Nevertheless, the correlation between flowering time and lignification may dependent on the clade. In maize, the brown midrib bm1 heterozygote lines flowered significantly earlier than the wild-type plants but have a reduced content and altered subunit composition lignin polymer [52]. It is possible that the anatomical differences including the absence of secondary tissues in monocots may explain the observed difference with the dicots.
In the light of our results it therefore seems that the mutation of the UGT72E3 gene impacts lignin deposition, showing that glycosylation could act as a regulator of the monolignol flux. Indeed, when considering the recent suggestion that monolignols cross the plasma membrane by passive diffusion, it is likely that the decreased lipophilic nature of monolignol glycosides resulting from glycosylation would reduce monolignol flow to the cell wall [17]. The role of glycosylation was also highlighted during a study on the potential role of ABC transporters [53] suggesting that the glycoconjugates are transported into the vacuoles and aglycones go towards the cell wall through the plasma membrane. If the UGT72E gene is indeed involved in regulating monolignol flux to the cell wall then one would expect there to be an inverse relationship between lignin levels and amounts of glycosylated monolignols. Based on a recent metabolomic study on the inner seed coats of pomegranate [54], it was proposed that the high amounts of monolignols accumulated in the inner seed coats could either be transported to the cell wall and used as lignin precursors leading to a hard-inner seed coat phenotype, or else conjugated into monolignol glucosides and stored, causing a soft-seeded cultivar. In flax, the same inverse correlation was described in a comparative study of the amounts of glysosylated phenolic compounds in outer stem tissues containing fibres with very low amounts of cell wall lignins, and the lignin-rich xylem tissues [55]. In this case, phenolic glycoside levels were higher in the lignin-poor outer tissues compared to the lignin-rich inner tissues. In this same species, a transcriptomic approach showed that the expression of a UGT gene close to the UGT72E family, was downregulated in the highly lignified bast fibre lbf mutant [50] further supporting the idea of this relationship between the lignification status and monolignol glycosylation activity.
Given the hypothesis that the knock-out of the UGT72E3 gene leads to increased monolignol flow to the cell wall also raises the question of the capacity of cell walls to incorporate these precursors. Before monolignols can be incorporated into the growing lignin polymer, they must first be oxidized by cell wall located peroxidases and/or laccases. If these enzymes are not working close to their maximum efficiency then increasing monolignol flow to the cell wall should lead to increased incorporation and higher cell wall lignin levels. On the other hand, if lignin redox enzymes are biochemically saturated then increasing the monolignol flow will not change lignin levels. To gain insight into this question we used a novel bioorthogonal triple labelling strategy based on the use of chemical reporters to compare lignification dynamics in the floral stem of ugt72e3 mutants and WT plants [36]. In addition, we also determined expression levels of selected lignin-related peroxidase and laccase genes.
Our results showing that UGT72E3 knock-out stimulated the capacity of mutant cell walls to incorporate monolignol chemical reporters in interfascicular fibre and xylem cell walls of young stem regions is coherent with, not only the data obtained by the safranin O ratiometric approach and Raman microspectroscopy, but also the hypothesis of increased monolignol flow to the cell wall. The observation that the accumulation of PRX71 and LAC17 transcripts also increased in this stem region would suggest that these genes are up-regulated in response to the increased monolignol flow to the cell wall resulting from decreased glycosylation. The fact that PRX25, PRX64, and LAC4 expression were unchanged could indicate that the corresponding proteins are present in a constitutive way and are not saturated by increased monolignol flow or alternatively they may be less involved in the lignification of cell walls in the younger part of the stem. In tissues from old stem regions, a more complex pattern emerged with decreased lignin G alk -and S cp -reporter incorporation in ugt72e3 interfascicular fibre cell walls compared to WT associated with increased incorporation of all 3 reporters in xylem tissue. The observed reduction of G alk -and S cp -reporter incorporation in interfascicular fibre cell walls is potentially related to the fact that in this region of the stem, the lignin levels in these cell walls are very similar in both WT and mutant plants as demonstrated by the safranin O and Raman microspectroscopy results. This could suggest that the lignification process is winding down in these cells in agreement with reduced reporter incorporation. In contrast, the formation of new cells in the xylem is associated with continued lignification as shown by the increased incorporation of all 3 reporters. The gene expression data for the old stem region showed that PRX71 and PRX25, but not LAC17, were both more highly expressed in the ugt72e3 mutant, most likely representing a developmentally related switch in the nature of the lignin related redox enzymes. The analysis of the corresponding mutants with our in vivo triple labelling strategy could provide more information on the precise role of these different PRX and LAC genes. Further investigations using a high throughput transcriptomics approach on the ugt72e3 line will also provide information on the expression of lignin biosynthetic genes and the genes encoding transcription factors known to be involved in the regulation of the phenylpropanoid pathway. It is indeed possible that a higher amount of monolignols were synthesized due to the mutation.
In conclusion, this study has generated several key results supporting the hypothesis that monolignol glycosylation plays a regulatory role in controlling cell wall lignification in the Arabidopsis floral stem, especially in the young tissues. The use of three independent high-performance imaging techniques combined with gene expression data provides clear evidence for a role of the UGT72E3 protein in this process.

Plant Material and Sampling
All the Arabidopsis thaliana plants were in the Col-0 background. The three T-DNA insertion mutant lines GK-340H02, SM_3_20654 and SAIL_1279_D02 corresponding respectively to the UGT72E1 (At3g50740), UGT72E2 (At5g66690), and UGT72E3 (At5g26310) genes were ordered from NASC (http://arabidopsis.info) and the homozygotes were selected by PCR. Double mutants ugt72e1,2 were obtained by crossing ugt72e1 with ugt72e2. In each mutant, the T-DNA was positioned at the 5 end of the unique exon.
The seeds were stratified during 72 h at 4 • C in the dark. Plants were grown in GroBank chambers (CLF Plant Climatics GmbH, Wertingen, Germany) under 12 h light (100 PAR light intensity) at 22 • C and 12 h dark at 20 • C. After 6 weeks, the photoperiod was changed to 16 h light, 8 h dark. For the phenotyping analyses, the rosette surfaces of at least 70 plants were measured on 28-day-old plants corresponding to the 3.20-3.50 stage [30] using FIJI v2.0 [56]. The age of the plants reaching the 5.1 stage, corresponding to the emergence of the floral stem, was also annotated. For lignin analyses, floral stems were collected when they reached 24 cm. The first bottom centimeter was discarded and a sample including either the above centimeter was used for safranin and biorthogonal labelling analyses or the 8 first centimeters used for lignin dosage by acetyl bromide or gene expression approaches (old stem regions). A second 8 cm fragment was collected for lignin dosage and gene expression analysis above the 11 first centimeters from the bottom of the stem. The central part of this fragment measuring 1 cm was used for safranin and biorthogonal labelling approaches (young stem regions).

Determination of Gene Expression by RT-qPCR
Top (young) and bottom (old) stem fragments from three 6-week-old plants for each biological replicate were pooled, frozen in liquid nitrogen and stored at −80 • C. Total RNA was extracted from ground tissues using Tri-Reagent (MRC, Inc., Dundee, Scotland, UK) according to manufacturer's instructions, treated with TURBO DNA-free DNase (Thermo Fisher, Courtaboeuf, France) and quantified by Nanodrop. First strand cDNA was obtained from 1µg total RNA using iScript™ cDNA Synthesis Kit (Bio-Rad, Marnes-la-Coquette, France) according to manufacturer's instructions. PCR was performed on Mx3005P (Agilent, Les Ulis, France) in duplicate on cDNA pooled from two independent reverse transcription reactions on three independent biological replicates. Reaction mixtures containing 10 µL of iQ SYBR Green Supermix (Bio-Rad, Marnes-la-Coquette, France), 375 nM each primer (Table S1) and 5 µL of cDNA diluted 6 times were incubated for 3 min 30 s at 95 • C followed by 40 cycles at 95 • C for 30 s and 60 • C for 30 s. Gene expression was normalized to the Protein Phosphatase 2A subunit A3 gene PP2A (AT1G13320) and PEROXIN4 (AT5G25760) [57]. Relative expression was determined as previously described [58]. One-way analyses of variance were performed with R, then grouping was done with Duncan multiple range test (p-value < 0.05).

Cell Wall Residue Preparation and Lignin Quantification
For each biological repeat, 4 or 8 fragments were collected from the bottom or top parts of the floral stems. Cell wall residues (CWR) were prepared as previously described [59]. The extensive solvent extraction method was chosen to remove proteins. It globally consisted in successive treatments of the plant material with the addition of 80% (v/v) ethanol, 100% acetone and CHCl 3 /methanol (1:1). The procedure of cell wall isolation also included amylase and amyloglucosidase treatments (the equivalent of 20 and 40 units respectively, per gram of cell wall sample) for starch removal. Finally, cell wall samples were ground in a bead miller (Mixer Mill MM 301; Retsch, Haan, Germany) at 30 Hz for 2 × 3 min of milling with 5 min pauses in between. The lignin quantification of CWR was performed using a modified acetyl bromide procedure [60]. Ten mg of cell wall sample were incubated in 5 mL of 25% (w/w) acetyl bromide in glacial acetic acid (extemporized preparation) at 70 • C for 30 min to achieve complete dissolution of residues. After cooling on ice, 2 mL of the mixture were incubated for 30 min in 0.5 M NaOH in glacial acetic acid. Lignin content was determined by measuring the OD at 280 nm and using ε = 20 g −1 ·L·cm −1

Sample Preparation for Imaging Analyses
Stem fragments were collected as described in 4.1 and embedded in agarose 4% before sectioning with a vibroslicer (VT-1000S, Leica, Wetzlar, Germany) to a thickness of 80 µm. The sections were placed in Murashige and Skoog half-strength (MS1/2) solution when used for biorthogonal labelling chemistry or fixed for 3 h. in 70% ethanol at room temperature under shaking and then kept at 4 • C before safranin O staining and Raman microspectroscopy.

Safranin O Ratiometric Imaging
Safranin O staining and analyses were performed as previously described [31]. Images from 3 independent biological replicates for each genotype and tissue were acquired. A Nikon A1R confocal equipped with a 60×/1.4 aperture oil immersion objective (Plan APO VC) and the NIS Element AR3.0 software was used (Nikon, Tokyo, Japan). Acquisition parameters were the following: track 1 (ex: 488 nm, BP: 525/50), track 2 (ex: 561 nm, BP: 595/50). Images were treated using FIJI v2.0 [56] and analyzed with the associated macro developed for this method as previously described [31]. Lignin appears in colors ranging from purple (low amounts) to red (high amounts).

Raman Microspectroscopy
Confocal Raman hyperspectral analyses were performed with a Labram HR Evolution (Horiba Scientific, Villeneuve-d'Ascq, France) equipped with a two-dimensional EMCCD detector (1600 × 200 pixels sensor), a 50× Fluotar LWD objective (Leica, Wetzlar, Germany) and a 545-nm laser with a power of 45 mW at the sample. The spectral range of 300 to 2000 cm −1 was used for acquisitions. Ten accumulations of 1 s with 100% laser power were used.
The spectra were collected with the LabSpec 6 software (Horiba Scientific, Lille, France). A minimum of 100 spectra (interfascicular fibres) and 80 spectra (xylem) per genotype from 6 to 8 plants distributed on 3 independent biological replicates were acquired. For data processing, once acquired, the hyperspectral images were unfolded and the spectra were baseline corrected using the Whittaker filter method (lambda = 10,000, λ= 0.01), normalized on the 590-630 cm −1 (top stem) and 375 cm −1 (bottom stem) bands and mean-centered with MatLab R2018b software. PCA were applied on the treated spectra of both genotypes from top and bottom stem samples.

Chemical Reporter Imaging
Chemical reporters H az , G alk and S cp were synthesized as previously described [36,61,62]. Natural and tagged monolignols were metabolically incorporated in the plant samples at a final concentration of 5 µM for 20 h ( Figure S5). Bioorthogonal ligation of a specific fluorescent probe to each incorporated reporters H az , G alk , and S cp was carried out according to our previously described triple labelling technology with modifications [36], using a combination of inverse electronic demand Diels-Alder chemistry, Copper-catalyzed alkyne-azide cycloaddition and Strain-promoted alkyne-azide cycloaddition. Cyclopropane-reactive Tetrazine-Cy5 (Jena Bioscience, Jena, Germany) and alkyne-reactive Azidefluor 545 (Sigma-Aldrich, Saint-Louis, MO, USA) probes were used at 5 µM to label incorporated S cp -and G alk -units, respectively. DBCO-PEG 4 -Rhodamine Green (Jena Bioscience, Jena, Germany) was used at 2.5 µM to label incorporated H az -units. Confocal microscopy was done with the same equipment used for safranin imaging with the following acquisition parameters: track 1-lignin autofluorescence (ex: 405, BP: 450/50); track 2-H units (ex: 488, BP: 525/50); track 3-G units (ex: 561, BP: 595/50); track 4-S units (ex: 561, BP: 700/75). An additional channel was acquired with transmitted line detector to obtain morphological images. Images from natural or tagged monolignol units from the different plant lines were acquired with the same acquisition parameters on the same day for valid comparison. Pictures were treated using FIJI v2.0 [56]. Quantitative measurements were performed using an in-house developed macro. Image processing consisted in two phases: the threshold background determination and the analysis. Threshold backgrounds were determined for the channels 2, 3, and 4 on samples treated with natural monolignols. A mean background threshold was determined for each sample from 3 images. Images from samples treated with tagged monolignols were converted in 32-bit images and the channels 2, 3, and 4 were extracted. The determined signal background was subtracted for each channel and pixels below the threshold background were classified as "NaN background". A region of interest was determined for each sample and the protocol was applied on the three obtained images. Finally, signal intensity was measured and comparison between the different plant lines was performed.