Leaf Gas Film 1 promotes glycerol ester accumulation and formation of a tight root barrier to radial O₂ loss in rice

Abstract

Rice (Oryza sativa L.) and many other wetland plants form an apoplastic barrier in the outer parts of the roots to restrict radial O₂ loss to the rhizosphere during soil flooding. This barrier facilitates longitudinal internal O₂ diffusion via gas-filled tissues from shoot to root apices, enabling root growth in anoxic soils. We tested the hypothesis that Leaf Gas Film 1 (LGF1), which influences leaf hydrophobicity in rice, plays a crucial role in tight outer apoplastic barrier formation in rice roots. We examined the roots of a rice mutant (dripping wet leaf 7, drp7) lacking functional LGF1, its wild type, and an LGF1 overexpression line for their capacity to develop outer apoplastic barriers that restrict radial O₂ loss. We quantified the chemical composition of the outer part of the root and measured radial O₂ diffusion from intact roots. The drp7 mutant exhibited a weak barrier to radial O₂ loss compared to the wild type. However, introducing functional LGF1 into the mutant fully restored tight barrier function. The formation of a tight barrier to radial O₂ loss was associated with increased glycerol ester levels in exodermal cells, rather than differences in total root suberization or lignification. These results demonstrate that, in addition to its role in leaf hydrophobicity regulation, LGF1 plays an important role in controlling the function of the outer apoplastic barriers in roots. Our study suggests that increased deposition of glycerol esters in the suberized root exodermis establishes a tight barrier to radial O₂ loss in rice roots.

Introduction

Cultivated rice (Oryza sativa L.) and many other wetland plants can develop apoplastic barriers in the outer parts of their roots to impede radial O₂ loss to the anoxic rhizosphere. These barriers facilitate longitudinal internal O₂ diffusion via gas-filled tissues from shoot to root apices (Armstrong 1979; Colmer 2003a, 2003b). The presence of an O₂-tight outer apoplastic barrier in roots is of evident importance for plants growing in flooded soils; without it, insufficient molecular O₂ reaches the root apices and roots stop growing (Armstrong and Webb 1985). However, this root trait has recently gained additional attention as these barriers can also restrict the entry of ions or gases into the roots [i.e. Na⁺, Cl⁻ (Ranathunge et al. 2011); H₂, H₂S (Peralta Ogorek et al. 2021, 2023], or even restrict radial water loss in the opposite direction from the proximal part of the root to a bone dry topsoil (Peralta Ogorek et al. 2021; Song et al. 2023).

In cultivated rice, the root barrier to radial O₂ loss is an inducible trait, but in some wild rice relatives, the barrier is formed constitutively (Ejiri et al. 2020). The barrier in cultivated rice can be induced by multiple factors including soil flooding (Colmer et al. 1998) and low water potential around the roots (Song et al. 2023). Moreover, various compounds produced by bacteria in anoxic soils can act as environmental cues for barrier formation such as hydrogen sulfide (H₂S) (Armstrong and Armstrong 2005; Peralta Ogorek et al. 2023), high concentrations of reduced iron (Fe²⁺) (Mongon et al. 2014), low nontoxic concentrations of low molecular mass organic acids (Colmer et al. 2019), and low nitrate (NO₃⁻) concentrations (Shiono et al. 2024). Due to the benefits of these outer apoplastic barriers in multiple and often contrasting environmental situations, their induction, function, and chemical compositions have recently been thoroughly reviewed (Liu and Kreszies 2023; Peralta Ogorek et al. 2024a, 2024b).

Histological and chemical characterization of the outer part of roots with outer apoplastic barriers capable of restricting radial O₂ loss have shown that the barrier formation coincides with increased depositions of suberin in exodermal cell walls and lignification of sclerenchyma cell layers (De Simone et al. 2003; Ranathunge et al. 2011; Shiono et al. 2014; Peralta Ogorek et al. 2023). Interestingly, recent research has shown that depositions of suberin and lignin in the outer parts of the roots can be delayed by Si (Tong et al. 2024). Nevertheless, these cell wall depositions of suberin and/or lignin in the outer parts of the root have long been associated with apoplastic barriers restricting the entry of dissolved gases, ions or molecules from the rhizosphere (Hose et al. 2001; Enstone et al. 2003; Ma and Peterson 2003; Schreiber et al. 2005; Ranathunge et al. 2011b; Kreszies et al. 2018).

In addition to suberin and lignin, soluble lipids associated with the suberin polymer are also known to be present in roots, but their function, if any, had not previously been examined. Monoacylglycerols are the main components of these soluble lipids in suberin, and they have also been referred to as “root waxes.” They are deposited in the suberin-rich periderm forming the outer cell layers of the Arabidopsis (Arabidopsis thaliana) root (Li et al. 2007). The amounts of root waxes in Arabidopsis are substantial with 3.6-fold higher concentrations than the classical waxes on leaf surfaces as determined by chloroform dipping (Li et al. 2007). In contrast to dicots (e.g. Arabidopsis) roots of monocots do not form a periderm but as mentioned above, the exodermis of rice can be heavily suberized and this is also the case for maize (Zea mays) (Enstone and Peterson 2005). Using the chloroform dipping approach, it was found that wax extracts of the outer part of maize and rice roots also possessed monoacylglycerols in high amounts, and comparing the total amounts of waxes, rice exhibited the fourth-highest concentration level, following Arabidopsis, salt cress, and maize (Kosma et al. 2015).

In aerial tissues, soluble lipids (waxes) deposited on the cuticles are important for hydrophobicity and the associated self-cleansing effect (Neinhuis and Barthlott 1997). In rice, the dripping wet leaf 7 (drp7) was isolated as a mutant with very low leaf hydrophobicity, and the drp7 mutant therefore failed in forming persisting leaf gas films as a response to submergence (Kurokawa et al. 2018). The causal gene involved in the reduced hydrophobicity was named Leaf Gas Film 1 (LGF1) (Kurokawa et al. 2018), and its involvement in leaf hydrophobicity had previously been described in the oshsd1 mutant (Zhang et al. 2016). LGF1/OsHSD1 encodes a hydroxysteroid dehydrogenase (HSD) influencing leaf wax composition, and its function was suggested to occur via interactive effects on the expression of Eceriferum (CER) genes (Zhang et al. 2016). Reanalyzing the gene expression data of Kulichikhin et al. (2014), we found that LGF1/OsHSD1 is highly expressed in the root apex of rice when growing in stagnant, deoxygenated nutrient solution (Supplementary Fig. S1), a treatment mimicking soil flooding (Wiengweera et al. 1997).

In the present study, we therefore tested the hypothesis that LGF1 gene expression plays a crucial role in the sealing of the outer apoplastic barrier of rice. To test the hypothesis, we used the drp7 mutant (in which the functional LGF1 gene is deficient), its wild type (cv. Kinmaze), and an overexpression line containing the functional LGF1 gene. We diagnosed the outer apoplastic root barrier and its ability to restrict radial O₂ loss and analyzed the chemical composition of enzymatically isolated sclerenchyma/exodermal root cells. We found that the drp7 mutant formed a weak barrier to radial O₂ loss, but introducing the LGF1 gene into the mutant background fully restored the tight function of the outer apoplastic barrier. The formation of a tight barrier to radial O₂ loss was associated with increased amounts of glycerol esters located in the suberized exodermal cells. These results suggest that LGF1 plays an important role not only in the shoot, where it controls leaf hydrophobicity, but also in the root controlling the function of the outer apoplastic barrier serving to restrict O₂ loss to the rhizosphere and impede entry of phytotoxins from flooded soils. The identification of the LGF1 gene as partly responsible for the formation of a tight outer apoplastic barrier, along with the specific role of glycerol esters in sealing the root to impede radial O₂ loss, provides an opportunity to develop crops with enhanced tolerance to soil flooding.

Results

Responses of the wild type and mutant to simulated soil flooding

We obtained the expression profiles of the LGF1 gene along the root to help understand the LGF1 function in the root. We found that the gene was expressed under both flooded and drained conditions, but its expression level was significantly higher 10 to 20 mm behind the root apex under flooded conditions (Fig. 1A), which is the location where functional depositions of biopolymers occur (Shiono et al. 2011) when an outer apoplastic barrier restricting radial O₂ loss begins to form (Colmer 2003). Histochemical localization of LGF1 promoter GUS activity revealed that LGF1 expression was induced slightly in cortical cells, and strongly in the outer part of the root at 10 to 20 mm behind the root apex (Fig. 1B). Consequently, we investigated if LGF1 had a function in cortical cells and/or the outer part of the root and used O₂ microsensors to measure cortical O₂ status of both the wild type and mutant grown under contrasting conditions, simulating either flooded soil or well-oxygenated drained soil.

Figure 1.

Root responses to simulated soil flooding. The wild type (cv. Kinmaze) and the mutant (drp7) were both grown for 21 days in aerated nutrient solution (to simulate a drained soil) or in stagnant, deoxygenated nutrient solution (to simulate soil flooding). A shows the expression level relative to TFIIE along the root of the LGF1 gene using qRT-PCR in the wild type in drained or flooded conditions, n = 4. B shows the LGF1 promoter GUS assay at 2 positions behind the root apex for LGF1 expression in the wild type under drained or flooded conditions (scale bars = 100 µm). In C, cortical pO₂ was measured at discrete positions along the root on intact plants with their roots in a severely hypoxic medium using O₂ microsensors, n = 5. In D, O₂ consumption (respiration) was measured on the apical part of the root, (n = 5). In E, radial O₂ loss was measured at discrete positions along roots of the 2 genotypes raised under drained or flooded conditions (data are mean ± SE, n = 8 to 11). Box-whisker plots in A, C, and D show the median (horizontal line), the upper and lower quartiles (box) and minimum or maximum (whiskers). In A, the 2-way ANOVA showed significant effects of distance and distance × treatment (P < 0.0001) but not of treatment. The post hoc test (Tukey) showed significant differences at some root positions, with asterisks indicating the level of significance (** P < 0.01, **** P < 0.0001). In C, the 2-way ANOVA showed significant effects of distance and genotype (P < 0.0001) but no interaction effect. In D, the 1-way ANOVA showed no significant effects of the combinations of genotype and growing condition. In E, the 2-way ANOVA showed significant effects of distance, genotype/growing conditions, and distance × genotype/growing conditions (P < 0.0001). To enhance clarity, the results of the post hoc tests (Tukey) have been omitted in panels C, D, and E.

Cortical O₂ status was assessed under conditions of low external pO₂, revealing that the wild type grown under flooded conditions maintained the highest tissue O₂ status along the entire root length (Fig. 1C). Conversely, the mutant exhibited significantly lower cortical pO₂ levels compared with the wild type raised under drained or flooded conditions. Nevertheless, in all cases, cortical pO₂ decreased toward the root apex, with the mutant grown in drained conditions showing the most pronounced decline in internal O₂ status.

The decline in internal O₂ status along the root can be attributed to a combination of tissue O₂ consumption and radial O₂ loss to the hypoxic environment. However, tissue O₂ consumption (respiration) was similar between the wild type and the mutant. Additionally, tissue O₂ consumption remained unaffected by growth conditions (Fig. 1D), thus failing to explain the observed differences in cortical O₂ status depicted in Fig. 1C. Instead, the divergent internal O₂ status is more likely a consequence of varying rates of radial O₂ loss, with both the wild type and the mutant exhibiting similarly high rates when grown in drained conditions (Fig. 1E). In contrast, the wild type formed a tight barrier to radial O₂ loss when grown under flooded conditions, exhibiting very low rates of radial O₂ loss in the proximal part of the root, followed by a sharp increase approximately 30 mm behind the apex (Fig. 1E). Intriguingly, the mutant raised under flooded conditions displayed a weak barrier with significantly higher rates in the proximal part of the root compared with the wild type (Fig. 1E).

Overexpression of LGF1 in the mutant restored barrier function

To establish a connection between the observed lack of function in the outer part of the root and the LGF1 gene, we generated an overexpression line containing the functional LGF1 gene in the mutant background. Overexpression of LGF1 fully restored the tightness of the barrier to radial O₂ loss, evidenced by similar and very low rates of radial O₂ diffusion in the proximal part of the roots, with rates not differing from those of the wild type (Fig. 2A). Methylene blue staining corroborated the measurements of radial O₂ loss, showing that O₂ primarily leaked from the root apices of the wild type and the overexpression line, as indicated by blue staining. In contrast, the roots of the mutant remained leaky, exhibiting blue staining, albeit weaker, along the entire root length, with the most pronounced staining occurring around the apices (Fig. 2B). These findings provide compelling evidence that the LGF1 gene is involved in the development of outer apoplastic barriers in roots.

Figure 2.

Root responses to simulated soil flooding in the wild type, the mutant, and the overexpression line. The wild type (cv. Kinmaze), the mutant (drp7), and the overexpression line (pUb::LGF1) were all grown for 21 days in a stagnant, deoxygenated nutrient solution to simulate soil flooding. A shows the radial O₂ loss to a severely hypoxic medium as a function of distance behind the root apex (data are means ± SE, n = 8 to 9. B visualizes radial O₂ loss from the roots of the 3 genotypes with the blue color from oxidized methylene blue indicating sites of O₂ diffusion to the severely hypoxic medium (scale bars = 15 mm). C and D show cortical aerenchyma (n = 4 to 5) and root diameter (n = 3 to 8), respectively, of the 3 genotypes. Box-whisker plots show the median (horizontal line), the upper and lower quartiles (box), and minimum or maximum (whiskers). In A, the 2-way ANOVA showed significant effects of genotype and distance (P < 0.0001) but not of the interaction. In C, the 2-way ANOVA showed significant effects of genotype (P < 0.001), distance (P < 0.0001), and genotype × distance (P < 0.05). In D, the 2-way ANOVA showed significant effects of genotype (P < 0.001) and distance (P < 0.05) but not of the interaction. To enhance clarity, the results of the post hoc tests (Tukey) have been omitted.

Still, other key anatomical traits exhibited differences among the wild type, the overexpression line, and the mutant when grown under flooded conditions. Aerenchyma formation varied longitudinally along the roots; however, roots of the wild type consistently developed more aerenchyma in the proximal parts of the root, with differences exceeding 10% at certain positions (Fig. 2C). Additionally, the wild type displayed significantly thicker adventitious roots compared with both the overexpression line and the mutant (Fig. 2D). Both of these parameters can contribute to preserving O₂ within the root. The relative differences in cortical aerenchyma and root thickness were also present under drained conditions (Supplementary Fig. S2). Importantly, the overexpression line and the mutant did not differ in terms of longitudinal aerenchyma formation and root thickness, suggesting that the LGF1 gene does not control these root traits.

Not only was the barrier to radial O₂ loss fully restored in the overexpression line, but we also assessed the vegetative growth of both the wild type and the mutant under flooded and drained conditions. This evaluation was conducted because LGF1 expression is strongly induced in the roots of the wild type under flooded conditions (Fig. 1, A and B). Compared with the wild type, the mutant showed reduced root growth under drained conditions. These differences were also present under flooded conditions. Flooding led to a significant 21% reduction in the length of the longest root (from 260 to 204 mm) in the mutant and 15% in the wild type (from 345 to 294 mm). These growth reductions were similar, but not statistically significant, in the overexpression line (15% from 193 to 164 mm) in flooded conditions (Supplementary Fig. S3). This suggests that the growth penalty was primarily related to the deficient barrier to radial O₂ loss (Figs. 1C and 2A), rather than to lower amounts of aerenchyma (Fig. 2C) and the more slender roots (Fig. 2D), as these 2 key root traits remained similar in the mutant and the overexpression line.

Qualitative and quantitative analyses of suberin, soluble lipids and lignin

Suberin and lignin are considered to be the primary functional constituents of the outer apoplastic barriers restricting radial O₂ loss (Liu and Kreszies 2023; Peralta Ogorek et al. 2024a, 2024b). Hence, we analyzed the chemical composition of the cell walls comprising the outer part of the roots (i.e. sclerenchyma/exodermis) by enzymatically digesting the cortical cells and mechanically removing the endodermis, followed by analytical chemistry (GC-FID (Flame Ionization Detector) and GC-MS). The outer part of the roots from the 3 genotypes evaluated contained suberin monomers of distinct chain length (Supplementary Table S1), primarily composed of fatty acids, diacids, alcohols, and ω-hydroxy acids (Supplementary Table S2). Soluble lipids extracted from the suberin polymer included fatty acids, alcohols, alkanes, sterols, triterpenoids, and glycerol esters (Supplementary Table S3).

Generally, drained conditions led to significantly lower concentrations of suberin, soluble lipids, and lignin in the outer part of the roots compared with flooded conditions (Supplementary Fig. S4 and Fig. 3). Under simulated soil flooding, however, the wild type, the overexpression line, and the mutant all exhibited similar amounts of the 2 categories of suberin (aromatic and aliphatic) in the outer part of the root. Consequently, the total amount of suberin also did not differ among them (Fig. 3A). A similar pattern was observed for soluble lipids, with no differences in sterols, triterpenoids, or total aliphatic wax between the 3 genotypes. These quantitative findings were corroborated by qualitative analyses using histochemical staining with Fluorol Yellow 088 at discrete positions along the roots. Consistent with the quantitative analyses, the histochemical staining did not reveal any differences among the wild type, the overexpression line, and the mutant when grown under flooded conditions (Fig. 3C).

Figure 3.

Cell wall depositions of biopolymers in the outer part of roots grown in flooded conditions. The wild type (cv. Kinmaze), the mutant (drp7), and the overexpression line (pUb::LGF1) were all grown for 21 days in a stagnant, deoxygenated nutrient solution to simulate soil flooding. A shows the concentrations of aromatic and aliphatics suberins (mean ± SE, n = 6), B shows the concentrations of soluble lipids (sterols, triterpenoids, and total waxes; mean ± SE, n = 6). C shows histochemical staining of suberin using Fluorol Yellow 088 (scale bars = 100 µm, overexpress. = overexpression). D shows total, uncondensed lignin (mean ± SE, n = 3) and E shows histological staining of lignin using the Mäule reaction (scale bars = 100 µm, overexpress. = overexpression). The 2-way ANOVA analysis in A and B showed a significant effect of the types of suberins and soluble lipids (P < 0.0001) but no effect of genotype, and no interaction effect. Since there was no genotype effect, the results of the post hoc test (Tukey) have been omitted for clarity. In D, the 1-way ANOVA showed no significant effects of genotype on uncondensed lignin.

We also analyzed the lignin content to investigate whether the functional differences in radial O₂ loss could be attributed to variations in lignin deposition in the cell walls of the outer part of the roots. Similar to suberin, the 3 genotypes showed no differences in total uncondensed lignin when grown under flooded conditions (Fig. 3D). Likewise, histochemical staining for lignin at 8 positions along the root also failed to reveal any differences among the 3 genotypes (Fig. 3E). Therefore, suggesting that the distinct patterns of radial O₂ loss observed in the wild type, the overexpression line, and the mutant cannot be attributed to overall depositions of total suberin, total soluble lipids, or lignin, but rather to specific groups of such compounds deposited in root cell walls.

Glycerol esters are low in the outer parts of the roots of the mutant

Quantitative and qualitative analyses of total suberin, total soluble lipids, and lignin in the outer parts of the roots did not reveal any significant differences between the wild type, the mutant, and the overexpression line. However, the amount of glycerol esters in the outer part of the mutant's root was considerably lower than that of the wild type or the overexpression line. We identified 5 very long-length glycerol esters found in 2 isomers (Fig. 4, A and B), and both types were present in significantly lower concentrations in the mutant. Interestingly, the reduction in glycerol esters in the mutant was notably more pronounced for type 2 compared with type 1; the amounts of type 2 glycerol esters in the mutant were typically only 25% to 40% of those in the wild type and the overexpression line (Fig. 4B). We suggest that the weak barrier to radial O₂ loss formed in the mutant under flooded conditions lies in these differences in glycerol ester accumulation in the outer parts of the roots, as this reduction was only observed in roots grown under flooded conditions (Fig. 4C). Consequently, we propose a conceptual model illustrating the possible deposition sites of these glycerol esters in the outer parts of the roots, thereby reinforcing other barrier elements consisting of suberin and lignin (Fig. 4D), based on the histochemical visualization of suberin and lignin (Fig. 3) and the spatial expression of the LGF1 gene (Fig. 1, A and B).

Figure 4.

Concentration of glycerol esters in the outer part of roots grown in flooded or drained conditions. The wild type (cv. Kinmaze), the mutant (drp7), and the overexpression line (pUb::LGF1) were all grown for 21 days in stagnant, deoxygenated nutrient solution to simulate soil flooding, or in aerated nutrient solution (to simulate a drained soil). Glycerol esters were found in 2 isomers (A and B) with carbon chain lengths from 24 to 32. C shows total concentrations of glycerol esters for roots formed under drained or flooded growth conditions. D shows a proposed conceptual model incorporating the positioning of glycerol esters as part of a tight, outer apoplastic barrier in roots of rice, with larger font size indicating higher amounts of glycerol esters (Ep = epidermis; Ex = exodermis; Scl = sclerenchyma; Co = cortex). The cells in D are not drawn to scale; for clarity, exodermis is enlarged. In A, the 2-way ANOVA showed significant effects of chain length and genotype (P < 0.05) but no effect of interaction. The post hoc test (Tukey) revealed no significant effects in the pairwise comparisons. In B, the 2-way ANOVA showed significant effects of chain length and genotype (P < 0.0001) but no effect of interaction. The post hoc test (Tukey) revealed significant effects in some of the pairwise comparisons with asterisks indicating level of significance (*** P < 0.001, **** P < 0.0001). In C, the 2-way ANOVA showed significant effects of genotype, treatment, and genotype × treatment (P < 0.01, 0.0001, and 0.05, respectively). The post hoc test (Tukey) revealed significant effects in some of the pairwise comparisons with asterisks indicating level of significance (* P < 0.05, ** P < 0.01). Data in A, B, and C are means ± SE, n = 6. Panel D created with Biorender.com.

Discussion

In the present study, we tested the hypothesis that the expression of the LGF1 gene plays a crucial role in sealing the outer apoplastic barrier of rice roots. We found that the expression of the LGF1 gene was induced at the outer part of the root, 10 to 20 mm behind the root apex, in response to soil flooding. This expression pattern is similar to expression patterns of the genes related to suberin biosynthesis (Nishiuchi et al. 2021), suggesting that the LGF1 gene is functionally involved in the formation of outer apoplastic root barriers restricting radial O₂ loss. Roots of the drp7 mutant were examined in hydroponics mimicking soil flooding and assessed for the root barrier to radial O₂ loss. Compared with the wild type, we found that the mutant formed a weak barrier to radial O₂ loss, but the introduction of the functional LGF1 gene into the mutant background fully restored the tight barrier function. Importantly, the formation of a tight barrier to radial O₂ loss was associated with specifically increased amounts of glycerol esters in the suberized exodermal cells, rather than an overall increase in suberin, soluble lipids, or lignin. These results demonstrate that the LGF1 gene plays an important role not only in the shoot, where it controls leaf hydrophobicity (Zhang et al. 2016), but also in the root, controlling the function of the outer apoplastic barriers. Below, we discuss these findings in the context of root acclimation to soil anoxia, and we provide an outlook for further research on the chemical composition of outer apoplastic barriers in roots.

The dual role of the LGF1 gene

We discovered that the LGF1 gene of rice not only is involved in the functioning of the leaf cuticle but also contributes to the effective functioning of the outer apoplastic barrier in roots. The involvement of the LGF1/OsHSD1 gene in cuticle function is a relatively recent finding (Zhang et al. 2016; Kurokawa et al. 2018). It has been suggested that OsHSD1 plays a key role in cuticle formation and lipid homeostasis, likely by mediating sterol signaling (Zhang et al. 2016). Later, it was demonstrated that expression of LGF1 in a mutant background (in which the functional LGF1 gene is deficient) restored C30 primary alcohol synthesis and wax platelet abundance on the cuticle, resulting in the leaf becoming superhydrophobic (Kurokawa et al. 2018). Leaf hydrophobicity is primarily regarded as a self-cleansing feature (Neinhuis and Barthlott 1997), where raindrops are repelled by the leaf surface while picking up dirt particles and spores of pathogens in the process (Barthlott et al. 2016). However, in rice, the superhydrophobic leaves serve an additional role. They retain a thin gas film upon submergence (Raskin and Kende 1983), and these gas films greatly enhance underwater photosynthesis (Pedersen et al. 2009), making this trait important for flood tolerance.

Given the significance of LGF1 in rice flood tolerance, our discovery that the same gene is necessary for the formation of a tight outer apoplastic barrier is indeed intriguing. In the past, the emphasis on functional cell wall components in the outer apoplastic barrier of rice has primarily been on suberin in the exodermis (Shiono et al. 2024), with some consideration also given to lignin in the sclerenchyma (Colmer et al. 2019). Here, we demonstrate that the drp7 mutant (in which the functional LGF1 gene is deficient) forms an incomplete, weak barrier to radial O₂ loss (Fig. 1C). However, both quantitative and qualitative analyses of suberin and lignin revealed similar total amounts of these biopolymers in the outer part of the root across the wild type, the mutant, and the overexpression line (Fig. 3, A to E). On the other hand, the function of the barrier was fully restored in the overexpression line, and hence, the weak barrier could not be attributed to differences in suberin or lignin deposition in the cell walls of the exodermis and sclerenchyma, respectively.

Research on functional biopolymer root cell wall depositions has focused solely on suberin and lignin. The reason is likely that these can be visualized using histochemical staining (examples in Fig. 3, C and E), but such staining remains qualitative and with relatively low specificity, sometimes yielding contrasting results when compared with chemical quantifications (Soukup et al. 2007; Abiko et al. 2012). However, some studies on the barrier to radial O₂ loss have attempted to quantify suberin (e.g. Shiono et al. 2022) and lignin (e.g. Ranathunge et al. 2016), and these studies clearly show that suberization of exodermis and lignification of the sclerenchyma are enhanced when roots have been formed under soil flooding conditions. Similarly, we also found that roots grown under soil flooding showed significantly higher concentrations of both suberin and lignin in the wild type, the mutant, and the overexpression line compared with these genotypes grown under drained conditions (Fig. 3, A to E and Supplementary Fig. S4, Supplementary Tables S1 and S2). Importantly, the overall amounts of the respective biopolymers did not differ between the 3 genotypes under soil flooding, even if the mutant possessed a weak barrier to radial O₂ (Fig. 1C). This indicates that the formation of a tight barrier to radial O₂ loss is not solely established by the suberin or lignin polymers themselves but rather by increased deposition of the suberin-associated glycerol esters, acting as an additional functional component of the outer apoplastic root barrier (see below).

The role of soluble lipids in the function of tight outer apoplastic root barriers

Considering the role of LGF1 in the production of epicuticular wax in rice leaves (Kurokawa et al. 2018) in combination with the fact that LGF1 is highly expressed in the zone where the root barrier to radial O₂ loss begins to form (Fig. 1, A and B and Supplementary Fig. S1), we would have expected differences in concentration of all root soluble lipids (e.g. waxes) in the outer part of the root between the wild type and the mutant. However, our data showed only a weak and statistically insignificant tendency to lower concentrations of total aliphatic soluble lipids in the outer part of the root of the mutant (Fig. 3B). Hence, we are unable to functionally link the weak barrier to radial O₂ loss of the mutant to differences in total amounts of soluble lipids deposited in the outer cell walls of rice roots.

However, our chemical analytical analyses of soluble lipids showed significant differences in the substance class of glycerol esters, with the mutant having significantly lower concentrations in the outer part of the root. In fact, total glycerol ester concentrations did not differ significantly in the mutant regardless of growth conditions (Fig. 4C), and the flooded roots of the mutant had similar concentrations of glycerol esters to those of the drained roots of the wild type (Fig. 4C). Moreover, the glycerol esters were found in 2 isoforms (Fig. 4, A and B), and almost all of the carbon chain lengths (C24 to C32) displayed a strong tendency for lower concentrations in the mutant compared with the wild type and the overexpression line. Therefore, we are unable to nominate one of them as particularly important in sealing the outer apoplastic barrier.

The exact role of glycerol esters in the outer apoplastic barriers involved in radial O₂ loss is not known. The finding that LGF1 specifically controls the abundance and chain length of glycerol esters in roots provides insight into soluble lipids biosynthesis, but further studies are required to identify the specific gene function. In contrast to the general role of leaf waxes in preventing cuticular water loss (Qin et al. 2011; Jian et al. 2022; Zhang et al. 2022), or increasing leaf hydrophobicity (Zhang et al. 2016; Kurokawa et al. 2018), the role of root waxes is still largely unknown. Here, we found that under flooding conditions, increased suberization and lignification of the outer part of the roots restricted radial O₂ loss but resulted in a weak barrier in the mutant. However, increased levels of glycerol esters deposited in the suberized exodermis further prevented radial O₂ loss from roots, forming a tight barrier in the wild type and overexpression line. This effect mirrors the action of cuticular barriers in aerial tissues, which consist of deposits of cutin along with their associated waxes. In leaves, cutin without associated waxes results in an ineffective transport barrier against water and dissolved substances (Schreiber 2010; Suresh et al. 2022). Cuticular wax seals the cuticle, but increased depositions of waxes on the cuticle do not necessarily lead to lower cuticle permeabilities to water (Grünhofer and Schreiber 2023). In roots, their complex characteristics prevent the drawing of oversimplified conclusions, such as assuming a direct association between suberin levels in the endodermis and root water transport properties (Grünhofer et al. 2024). In summary, we found that the specific accumulation of glycerol esters in the outer part of the root, rather than total amounts of suberin, soluble lipids, or lignin were related to a tight barrier to radial O₂ loss.

We suggest here that the primary factor contributing to the mutant's weak barrier to radial O₂ loss is the difference in glycerol ester accumulation in the outer part of the root, evident in roots cultivated under flooded conditions. In our conceptual model, we have depicted the glycerol ester depositions in the cell walls of the exodermis, along with the known position of suberin (Fig. 4D; Liu and Kreszies (2023)). However, at present, we cannot fully rule out the possibility that glycerol esters might also be deposited in the sclerenchyma ring below the exodermis, although it seems unrealistic. Glycerol has also been shown to be an important part of the suberin polymer acting as a crosslinker for the aliphatic suberin monomers (Graça and Pereira 1997) and similar monoacylglycerols, as they were identified here as soluble lipids, have also been identified after partial suberin depolymerization (Graça and Pereira 1999). Thus, we hypothesize that the unknown function of the LGF1 gene in root suberization is involved in regulating or coordinating the suberin polymerization in parallel with the deposition of the soluble lipids (root waxes), but this needs to be investigated in more detail in the future.

Conclusions and outlook

In conclusion, our results suggest that a tight outer apoplastic barrier restricting radial O₂ loss in roots of rice is established by increased depositions of glycerol esters in the suberized root exodermis. Here, glycerol esters act as an extra diffusion barrier restricting O₂ loss to the rhizosphere. The identification of the function of glycerol esters in sealing the outer apoplastic barrier to restrict O₂ loss warrants further study, given the increasing emphasis on the importance of such barriers. They not only reduce O₂ loss but also impede water loss (Song et al. 2023), confer drought tolerance (Peralta Ogorek et al. 2024a, 2024b), limit the entry of soil phytotoxins into the root, and potentially decrease root permeability to greenhouse gases (Jiménez and Pedersen 2023). Nevertheless, further studies are needed to unravel the coordinated effects and functions of the LGF1 gene in both leaves and roots, as well as the regulatory factors controlling gene expression in these tissues. Moreover, the identification of the LGF1 gene as partly responsible for the formation of a tight outer apoplastic barrier adds to our current understanding of the genetic basis of flood and drought tolerance in rice and provides an opportunity in the future to develop climate-resilient rice to sustain, or even increase, production in areas that are prone to abiotic stresses.

Materials and methods

In this study, we characterized the formation of barriers to impede radial O₂ loss in a mutant, in which the functional LGF1 gene is deficient, and its wild type. We also examined the expression of the LGF1 gene in roots and O₂ consumption by these genotypes when grown in drained or flooded conditions. After discovering that the mutant genotype formed a weak barrier to radial O₂ loss, and the wild type a tight barrier, we developed an overexpression line containing the functional LFG1 gene. We then evaluated radial O₂ loss, root anatomy, and the chemical composition of isolated sclerenchyma and exodermal cell layers from the 3 genotypes (see Supplementary Fig. S5 for an overview of the experimental approach).

Plant materials and growth conditions

Plants of rice (Oryza sativa L.) cv. Kinmaze, a rice mutant (dripping wet leaf 7; drp7), and an overexpression line of LGF1 expressed in the drp7 mutant background were used in the present experiments. The drp7 mutant presents a single nucleotide substitution in LOC_Os11g30560.1 of a “T” instead of an “A” in the wild type, located in the third intron of the LGF1/OsHSD1 gene, which is a member of the short-chain dehydrogenase reductase (SDR) family (Zhang et al. 2016; Kurokawa et al. 2018). The overexpression line of LGF1 had previously been produced by Kurokawa et al. (2018).

Seeds of the 3 genotypes were disinfected in 10% (w/v) sodium hypochlorite for 5 min, washed thoroughly with deionized (DI) water, and imbibed in DI water for 24 h at 4 °C before being germinated in dark conditions on a floating mesh over a 25% (v/v) concentration aerated nutrient solution (see below for growing conditions and composition of full-strength nutrient solution). On day 3, seeds were exposed to light in a growth chamber at a controlled temperature of 30 °C. Seven-day-old plants were transferred into full-strength nutrient solutions (4 plants per 3-L pots) for 7 days. Nutrient solution composition was (mM): K⁺, 5.95; Ca²⁺, 1.5; NH₄⁺, 0.625; Mg²⁺, 0.40; Na⁺, 0.20; NO₃−, 4.375; SO₄²⁻, 1.905; H₂PO₄⁻, 0.20; SiO₂²⁻, 0.10; Mn²⁺, 0.0020; Zn²⁺, 0.0020; Ni²⁺, 0.0010; Cu²⁺, 0.00050; Cl⁻, 0.050; BO₃³⁻, 0.025; MoO₄²⁻, 0.00050; and Fe-EDTA, 0.05. The solution also contained 2.5 mol m⁻³ MES, and the pH was adjusted to 6.5. Continuous forced aeration was used in the aerated nutrient solutions. Plants (now 14 days old) were then transferred to 3-L pots containing either full-strength aerated nutrient solutions or full-strength stagnant deoxygenated solutions. The stagnant deoxygenated solution was prepared by dissolving 0.1% (w/v) agar in the nutrient solution to prevent convection and simulate changes in gas composition as commonly occur in flooded soils (Wiengweera et al. 1997). These solutions were prepared using the same nutrient composition as the aerated solutions and were flushed with high-purity N₂ gas to purge out O₂ of the solution before the plants were transferred. Treatments lasted for 21 days, and pots were re-randomized when all nutrient solutions were renewed every 7 d.

Radial O₂ loss and root tissue O₂ status along intact roots

Radial O₂ loss along intact adventitious roots when in a deoxygenated medium was measured using root-sleeving platinum electrodes as described in detail in (Jiménez et al. 2021; Nishiuchi et al. 2021). For each plant, radial O₂ loss (ROL) was measured along one root (80 to 110 mm in length) at positions 10, 20, 30, 40, 50, 60, 70, and 80 mm behind the root apex. Measurements using the root-sleeving electrode were taken in a growth chamber with fixed temperature of 30 °C and 750 µmol m⁻² s⁻¹ photosynthetically active radiation. Root diameters at each position were measured using a digital caliper. ROL was also measured using the approach of Colmer et al. (2020) with an O₂ microsensor OX-25, Unisense A/S, Aarhus, Denmark) at 2 different radial positions above the root and applying the equations of Henriksen et al. (1992).

Cortical O₂ status along intact adventitious roots when in a deoxygenated medium was measured using O₂ microsensors (Pedersen et al. 2020). The shoot and the target roots were gently fixed on a metal mesh with rubber bands. The mesh holding the plant was fixed in an aquarium (200 × 100 × 100 mm, in total 2 L) and the O₂ microsensor (OX-10, Unisense A/S, Aarhus, Denmark) was inserted 150 to 200 µm into the cortex. Stagnant, deoxygenated nutrient solution was added to the aquarium, and cortex O₂ concentration was logged using Logger (SensorTrace Suite v. 3.2, Unisense A/S) at a rate of 1 Hz. The O₂ concentration and temperature of the external medium were monitored using an O₂ minioptode (Opto-MR, Unisense A/S) and a temperature probe (ZNTC, Unisense A/S). The positioning of the microsensor inside the root cortex was aided by a motorized micromanipulator controlled by Logger and visually aided with a dissection microscope (Leica WILD M3B, WILD LEITZ, Heerbrugg, Switzerland).

Root tissue O₂ consumption

Rates of O₂ consumption by excised root apices were measured following the procedure described in Colmer et al. (2019). Two 10-mm long root apices were inserted into a 4-mL glass vial containing aerated nutrient solution. Each vial was fitted with a stir bar and a mesh to separate the stir bar and tissue, and the stirring rate was set to 600 rpm (stirrer controller unit MR2-St-Co, Unisense A/S, Denmark). The sealed vials were placed in a rack and submerged into a constant temperature bath (30 °C), and left to stabilize for 15 min. O₂ was then measured in each vial using an O₂ optode (OP-MR, Unisense A/S) every 10 to 20 min, during which the O₂ concentration within the vials declined from approximately 230 to 200 μmol O₂ L⁻¹. Vials with incubation medium, but without tissues, served as blanks. O₂ consumption rates were calculated as the difference in O₂ concentration (μmol O₂ L⁻¹) between 2 time points multiplied by the vial volume using Rate (Sensortrace Suite version 3.1.50, Unisense A/S) and finally divided by the fresh mass (FM) of the tissue to obtain the O₂ consumption rate in nmol O₂ L⁻¹ g⁻¹ FM s⁻¹.

Root anatomy

Randomly selected roots (90 to 110 mm in length) formed under each treatment were excised from plants and cut into segments (10 mm long) at distances of 10, 20, 30, 40, 50, 60, 70, and 80 mm behind the root apex, and then embedded in 5% (w/v) agar. Root cross sections (approximately 100 µm) were taken using a vibrating microtome (VT 1200S, Leica Biosystems, Nussloch, Germany) and viewed under white light microscope (BX3, Olympus Optical, Japan). Cross-sections were photographed with a CCD camera (Olympus DP73, Japan), and root diameters and aerenchyma formation (% gas-filled large spaces in the root cortex) were determined using ImageJ software (National Institutes of Health, Bethesda, USA).

Histochemical staining for suberin and lignin

Suberization of the cell walls of the outer part of the roots was visualized using Fluorol Yellow 088. Approximately 100-µm thick cross sections were prepared as above and suberin lamellae were detected after staining with 0.01% (w/v) Fluorol Yellow 088 as described in Brundrett et al. (1991). Suberin lamellae were visualized as a yellow-green fluorescence in cross-sections evaluated under UV excitation light (U-FUW, Ex 340–390, dichroic mirror DM-410, emission filter 420IF) in an epifluorescence microscope (BX3, Olympus, Japan) and photographed using a digital camera (DP74, Olympus Optical, Japan).

Similarly, lignin depositions in root cross-sections were detected using the Mäule reaction (Kutscha and Gray 1972). Lignified cell walls (syringil group) appeared brown-orange, while non-lignified tissues remained unstained under white light microscopy (BX3, Olympus Optical, Japan). For the Mäule reaction, cross sections were treated with a 1% (w/v) KMnO₄ solution for 30 s, then washed thoroughly with DI water, and incubated in 12% HCl before these were treated with 32% ammonia.

Chemical analysis from enzymatically isolated root hypodermal/exodermal cells

Histochemical analysis indicated that suberization of root exodermal cells and lignification of root sclerenchyma occurred in the 3 genotypes investigated (Fig. 3, C and E). We enzymatically isolated the sclerenchyma and exodermal cell layers and analyzed their chemical composition. Whole roots of c. 90 to 120 mm were infiltrated for 10 min with an enzymatic solution containing 0.5% (w/v) pectinase and 0.5% (w/v) cellulase in 10 mm citric acid monohydrate at pH 3.0 (Baales et al. 2021). Root sections were then incubated in the same enzymatic solution under low-motion shaking for 1 to 2 weeks. The enzymatic solution was replaced 3 times during this period. This treatment was sufficient to digest cortical cells without digesting outer modified cells with suberin and/or lignin deposition in the sclerenchyma and exodermis. The central cylinder including the endodermis of the roots was then mechanically removed using fine-tipped forceps under a stereo microscope (S6E, Leica, Germany). Isolated root sclerenchyma/exodermal cells were washed in 0.01 m borate buffer at pH 9.2 for 1 day under continuous shaking, and finally, the root tissues were washed 3 times with DI water and dried.

Soluble lipids (root waxes) were extracted from enzymatically isolated sclerenchyma/exodermal cell walls using chloroform. As an internal standard for lipid quantification, every sample was spiked with 20 µg tetracosane (100 µl of a solution of 10 mg tetracosane in 50 ml chloroform; Fluka). Subsequently, chloroform-extracted sclerenchyma/exodermal cell walls were incubated in boron trifluoride/methanol (BF3/MeOH, Fluka) for suberin analysis, as described by Franke et al. (2005). For lignin analysis, the materials were depolymerized by thioacidolyisis, as described by Foster et al. (2010). After transesterification and thioacidolysis, depolymerized monomers were extracted with chloroform and spiked with dotriacontane for both suberin and lignin monomers as an internal standard (10 µg of dotriacontane in 50 µl of solution; Fluka). Using a moderate stream of nitrogen, the final chloroform volume for all the extracted compounds was lowered to a final volume of 200 µl. For quantification, gas chromatography coupled to flame ionization detection (GC-FID; Hewlett Packard 5890 series H, Agilent) was used. Gas chromatography coupled to mass spectrometry (GC-MS; quadrupole mass selective detector HP 5971, Hewlett Packard, Agilent) analysis was employed for the identification of soluble lipid, suberin, and lignin monomers. Using 1 column (for soluble lipids) or split/splitless injection (for suberin and lignin), 1 µl of each sample was run on 30 m GC columns (DB-1 Columns, Agilent). The chemicals from each group were then identified using molecular ion fragment patterns using a custom-made MS library for soluble lipids and suberin. For lignin, reference tables with molecular ion fragments from Rolando et al. (1992) and Zeier and Schreiber (1997) were used.

Reverse transcription quantitative PCR (RT-qPCR)

Randomly selected adventitious roots (80 to 90 mm in length) formed under stagnant, deoxygenated conditions were excised, and cut into 10 mm long segments at distances of 10, 20, 30, 40, 50, 60, 70, and 80 mm behind the root apex. The root segments were immediately frozen in liquid nitrogen and ground with a multibeads shocker (MB2200, Yasuikikai, Osaka, Japan). The total RNA was extracted from the root segments using NucleoSpin RNA Plant and Fungi (Macherey-Nagel, Germany) according to the manufacturer's protocol. Gene-specific primers are listed in Supplementary Table S4. Ten ng total RNA was used for the qRT-PCR. qRT-PCR was conducted using OneStep TB Green PrimeScript RT-PCR Kit II (Takara, Japan) in QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific, USA). The qRT-PCR cycling conditions consisted of a reverse transcription step at 42 °C for 5 min and an initial denaturation step at 95 °C for 10 s, followed by 40 cycles of denaturation at 95 °C for 5 s, annealing and extension at 60 °C for 30 s. Standard curves for absolute quantification were constructed using purified and quantified PCR fragments of the LGF1 gene. Transcript levels were normalized to those of the gene encoding TFIIE (transcription initiation factor IIE).

LGF1 promoter and GUS analysis

The DNA fragment encoding GUS was amplified from pBI121, and the resulting PCR fragment of GUS contained SpeI linkers at the 5′ and 3′ ends. The GUS DNA fragment and pCAMBIA1380 were digested with SpeI, respectively. The DNA fragment was inserted into the SpeI site of pCAMBIA1380, generating GUS-pCAMBIA1380. The 2135 bp upstream region of LGF1 and the DNA fragment were amplified from O. sativa cv. Kinmaze genomic DNA as promoter sequence and the resulting PCR fragment of LGF1 promoter contained BglII linkers at the 5′ and 3′ ends. The DNA fragment of the LGF1 promoter and GUS-pCAMBIA1380 were digested with BglII, respectively. The DNA fragment was then inserted into the BglII site of GUS-pCAMBIA1380, generating LGF1pro::GUS-pCAMBIA1380, and LGF1pro::GUS-pCAMBIA1380 was transformed into Agrobacterium tumefaciens strain EHA105 by electroporation. Rice transformation was performed as previously described (Toki et al. 2006).

The plants were cultured in aerated or stagnant, deoxygenated nutrient solutions and adventitious roots (100 to 120 mm long) were collected and fixed with 90% cold acetone. After fixation, roots were washed with 50 mm sodium phosphate buffer (pH 7.0) and were then incubated in GUS staining solution containing 0.5 mg/mL X-Gluc (5-bromo-4-chloro-3-indolyl-beta-D-glucuronic acid), 50 mm sodium phosphate buffer (pH 7.0), 0.5 mm potassium ferricyanide, 0.5 mm potassium ferrocyanide, 10 mm EDTA, and 0.1% Triton X-100. The staining reaction was carried out within 2 to 3 h at 37 °C in the dark. GUS staining patterns were analyzed qualitatively on root cross-section using a light microscope (BX3, Olympus Optical) connected to a CCD camera (Olympus DP73).

Statistical analyses

Data are presented as means ± S.e. Statistical differences between data values for treatments (aerated, stagnant) or genotypes (wild type, mutant, overexpression line) were evaluated through 1- or 2-way ANOVA, see details on the legend of each graph. The assumptions for the ANOVA test (homogeneity of variance, normal distribution, and independence) were tested, and all requirements were met. Statistical analyses were carried out using the Agricolae library (Mendiburu 2023) of the statistical software R.

Accession numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers LC363889 (WT LGF1) and LC363890 (mutant lgf1).

Acknowledgments

We thank Drs. Atsushi Yoshimura, Hideshi Yasui, and Yoshiyuki Yamagata from the School of Agriculture, Kyushu University, for providing seeds of cv. Kinmaze and the drp7 mutant. We also thank Dr. Yusuke Kurokawa from the Faculty of Agriculture, Meijo University, for stimulating discussions.

Author contributions

J.C.J., H.T., O.P., L.S., and M.N. conceived and designed the study. J.C.J., S.N., K.S., V.Z.-D., L.L.P.O., M.H., E.P., K.N., M.A., H.T. and O.P. conducted the experiments. J.C.J., H.T., O.P., L.S., and M.N. analyzed and interpreted the data. J.C.J., H.T., and O.P. drafted the manuscripts, and all authors commented on and approved the final version for submission.

Supplementary data

The following materials are available in the online version of this article:

Supplementary Table S1. Chain-length distribution of diacids and ω-hydroxyacids.

Supplementary Table S2. Monomeric composition of suberin.

Supplementary Table S3. Monomeric composition of soluble lipids.

Supplementary Table S4. Primer list for RT-qPCR.

Supplementary Figure S1. Expression of LGF1 in rice root tips.

Supplementary Figure S2. Root responses to a simulated drained soil in the wild type, the drp7 mutant, and the pUb::LGF1 overexpression line.

Supplementary Figure S3. Maximum root length of the wild type (cv. Kinmaze), the drp7 mutant, and the pUb::LGF1 overexpression line.

Supplementary Figure S4. Cell wall deposition of biopolymers in the outer part of roots grown in drained conditions.

Supplementary Figure S5. Experimental flow.

Funding

The following funding bodies kindly support this study: The European Union under a Marie Skłodowska-Curie Postdoctoral Fellowship (grant No. 101061962 to JdlCJ, grant No. 839542 to EP, grant no. 801199 to LLPO); Japan Society for the Promotion of Science (to JdlCJ); Independent Research Fund Denmark (grant no. 8021-00120B to OP); Danish International Development Agency (grant No. 19-03-KU to OP and MH); German Research Foundation (grant No. 518270674 to LS); Grant-in-Aid for Transformative Research Areas (A) (MEXT KAKENHI, Japan) (grant No. JP20H05912 to MA and MN).

Data availability

All data are available from the UCPH data repository, ERDA, at https://doi.org/10.17894/ucph.ad87b142-cbee-4923-a43a-73436c506bc7.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• root AmiGo: PO:0009005

• glycerol CHEBI: CHEBI:17754

• SUBERIN CHEBI: CHEBI:60973

• aerenchyma AmiGo: PO:0005702

Author notes

The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (https://academic.oup.com/plphys/pages/General-Instructions) is Mikio Nakazono.

Conflict of interest statement. None declared.