Glycosylphosphatidylinositol anchor lipid remodeling directs proteins to the plasma membrane and governs cell wall mechanics

Abstract

Glycosylphosphatidylinositol (GPI) anchoring is a common protein modification that targets proteins to the plasma membrane (PM). Knowledge about the GPI lipid tail, which guides the secretion of GPI-anchored proteins (GPI-APs), is limited in plants. Here, we report that rice (Oryza sativa) BRITTLE CULM16 (BC16), a membrane-bound O-acyltransferase (MBOAT) remodels GPI lipid tails and governs cell wall biomechanics. The bc16 mutant exhibits fragile internodes, resulting from reduced cell wall thickness and cellulose content. BC16 is the only MBOAT in rice and is located in the endoplasmic reticulum and Golgi apparatus. Yeast gup1Δ mutant restoring assay and GPI lipid composition analysis demonstrated BC16 as a GPI lipid remodelase. Loss of BC16 alters GPI lipid structure and disturbs the targeting of BC1, a GPI-AP for cellulose biosynthesis, to the PM lipid nanodomains. Atomic force microscopy revealed compromised deposition of cellulosic nanofibers in bc16, leading to an increased Young’s modulus and abnormal mechanical properties. Therefore, BC16-mediated lipid remodeling directs the GPI-APs, such as BC1, to the cell surface to fulfill multiple functions, including cellulose organization. Our work unravels a mechanism by which GPI lipids are remodeled in plants and provides insights into the control of cell wall biomechanics, offering a tool for breeding elite crops with improved support strength.

BRITTLE CULM16 is involved in glycosylphosphatidylinositol anchor lipid remodeling, which is required for targeting modified proteins to the cell surface and governs cell wall biomechanics.

IN A NUTSHELL.

Background: Glycosylphosphatidylinositol (GPI) anchoring is an important post-translational modification, which tethers proteins to the outer leaflet of the plasma membrane. Approximately 1% of plant proteins are thought to be modified with a GPI anchor and involved in many biological processes via facilitating signal perception, cell adhesion, transportation, and metabolism. Mature GPI moieties of eukaryotes usually contain a conserved glycan core structure and a variable lipid tail. The lipid portion is crucial for delivery of GPI-anchored proteins (GPI-APs) to specific lipid microdomains to carry out specialized functions. However, contrary to their known functional importance, plant GPI lipid composition and the biosynthetic pathway for GPI modification in plants remain to be explored.

Question: How are the lipid tails of GPI-APs formed in plants and what are their effects on cell wall formation?

Findings: Through characterization of the rice (Oryza sativa) brittle culm16 (bc16) mutant, we identified a membrane-bound O-acyltransferase (MBOAT) required for GPI lipid modification. We found that BC16 is located in the endoplasmic reticulum and Golgi apparatus. BC16 acts as an acyltransferase in GPI lipid remodeling, based on the findings that BC16 rescued the growth defects of a yeast mutant deficient in a MBOAT homolog, and that disruption of BC16 in rice causes an obvious reduction in GPI-AP lipids. Using BC1, a previously reported GPI-AP, as a reporter, we clarified that BC16-mediated lipid remodeling is required for targeting GPI-APs, including BC1, to plasma membrane lipid nanodomains. We further performed atomic force microscopy and nanoindentation analyses to reveal the impacts of BC16 on cell wall nanofiber organization and elastic moduli.

Next steps: We will identify more enzymes involved in plant GPI lipid maturation, which will shed light on mechanisms of GPI modification and offer a comprehensive understanding of plant GPI-AP functions in cell wall organization.

Introduction

The plasma membrane (PM) is an important interface between the extracellular matrix and the interior of eukaryotic cells. Proteins embedded in or associated with the PM usually play roles in signal perception, cell–cell communication, cell adhesion, transportation, and metabolism at the cell surface, mediating many biological processes such as embryogenesis, organ development, immune responses, and fertilization (Nagamune et al., 2000; Lalanne et al., 2004; Gillmor et al., 2005; Almeida et al., 2006; Ueda et al., 2007). Proteins can be tethered to the outer leaflet of the PM via glycosylphosphatidylinositol (GPI) anchoring, a class of post-translational modification (Fujita and Kinoshita, 2012; Yeats et al., 2018). The attachment of the GPI anchor occurs in the endoplasmic reticulum (ER), which is followed by lipid remodeling mediated by a series of enzymes in the ER and/or Golgi apparatus. Studies in mammalian cells, yeast (Saccharomyces cerevisiae), and parasites have identified approximately 30 genes for GPI modification (Yeats et al., 2018). Based on the sequence similarity of the encoded proteins, the biosynthetic pathway for GPI modification appears to be conserved. Homologs were also found in Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa), as well as in Archaea. The evolution of the GPI anchor likely predates the origin of eukaryotes (Eichler and Adams, 2005; Zhao et al., 2020).

The mature GPI moieties of eukaryotes usually contain a conserved glycan core structure and a variable phosphatidylinositol (PI) tail (Ferguson, 1999; Fujita and Kinoshita, 2012; Yeats et al., 2018). The glycan core is composed of a glucosamine (GlcN) and three α-linked mannose (Man) residues, and it is synthesized and bound to ethanolamine phosphate via nearly 10 reaction steps. Each step is catalyzed by at least one specific enzyme (Kinoshita, 2016; Yeats et al., 2018). Attachment of the GPI anchor to the carboxyl (C)-terminus of target proteins is mediated by a GPI transamidase in the ER (Kinoshita and Fujita, 2016; Desnoyer et al., 2020). Five enzymes responsible for glycan core synthesis have been identified in Arabidopsis through mutant characterizations (Lalanne et al., 2004; Gillmor et al., 2005; Dai et al., 2014; Bundy et al., 2016), demonstrating that the core glycan structure of plant GPI-anchored proteins (GPI-APs) is conserved. By contrast, the PI tail varies across species. The tail is formed through a four-step reaction in the ER in S. cerevisiae and in the ER and Golgi in mammals (Bagnat et al., 2000; Hemsley, 2015). Eventually, yeast GPI-APs acquire either a diacyl-PI (diacylglycerol moiety) or an inositolphosphoceramide (IPC, ceramide moiety) tail, directing them to the microdomains of ER-exit sites for secretion, whereas mammalian GPI-APs bear a 1-alkyl-2-acyl PI or diacyl-PI (C18:0) lipid tail involved in different secretion processes (Brown and London, 1998; Bagnat et al., 2000; Muniz and Zurzolo, 2014; Desnoyer et al., 2020). The lipid portion of the GPI anchor is hence required for the delivery of GPI-APs to specific lipid microdomains (Fujita et al., 2006). Because plant mutants deficient in lipid remodeling are rarely reported (Bernat-Silvestre et al., 2021), we have very limited knowledge about the lipid structure of plant GPI-APs.

The GPI anchor can guide the transfer of GPI-APs to specific functional domains at the cell surface, such as lipid nanodomains that are rich in sphingolipids or sterols. Our knowledge about GPI-AP sorting largely comes from the studies in animals and yeast (Paladino et al., 2006; Zurzolo and Simons, 2016). Pectin methylesterase inhibitor protein1 (PMEI1) is one of the few plant GPI-APs for which trafficking has been characterized (De Caroli et al., 2011). GPI modification was identified as the primary sorting signal for targeting the plant GPI-APs to the membranous channels of plasmodesmata (Zavaliev et al., 2016). While GPI-AP sorting is thought to be conserved between plants and other eukaryotes, plants may have unique secretion processes for locating GPI-APs to the lignocellulose-consisting cell wall (Konrad and Ott, 2015; Yeats et al., 2018).

Plant cells are surrounded by cell walls outside the PM. Some plant GPI-APs could provide connections between the PM and the cell wall, giving these proteins distinct functions not found in animals. These functions include monitoring cell wall integrity and synthesizing and remodeling the cell wall, thereby regulating plant growth and mechanical strength (Schindelman et al., 2001; Hayashi et al., 2008; MacMillan et al., 2010; Hou et al., 2016; Yeats et al., 2018). The rice protein BRITTLE CULM1 (BC1, also termed COBRA-like5 [COBL5]) is a major player of multiple plant GPI-APs regulating cellulose biosynthesis (Li et al., 2003). We previously reported that BC1 can be released to the cell wall after reaching the PM and bind to cellulose, facilitating cellulose crystallization (Liu et al., 2013). However, how BC1 proteins are delivered to the PM is unknown.

In this study, we characterized a rice membrane-bound O-acyltransferase (MBOAT) that is involved in the GPI lipid modification. Disruption of BC16 compromises the lipid moieties of GPI-APs. Transformation of BC16 into a yeast mutant deficient in the homolog Glycerol uptake protein1 (Gup1) largely restored the GPI lipid defects. We further established that lesion in this remodeling interferes with targeting BC1, one of the GPI-APs required for cellulose biosynthesis, to the PM lipid nanodomain, resulting in abnormal cellulose deposition and reduced cell wall elasticity in the bc16 mutant. Our work hence offers insights into the modification of the GPI lipid tail in plants and outlines a mechanism for plant cell wall biomechanics control.

Results

BC16 regulates cell wall biosynthesis and mechanical force

The brittle culm phenotype is an indicator of abnormal cell walls in rice plants and has been used successfully as a screen to identify genes involved in secondary cell wall biosynthesis (Zhang and Zhou, 2011; Zhang et al., 2021). bc16 is a spontaneous mutant isolated from a japonica rice variety, Zhonghua11 (ZH11). In addition to dwarfism, we found that the bc16 mutant had fragile internodes (Supplemental Video S1), as indicated by a reduced breaking force (Figure 1, A–C). The dwarfism was caused by a 20%−30% reduction in the length of each internode (Supplemental Figure S1A). Moreover, bc16 plants exhibited more tillers but significantly reduced seed set and grain weight compared with the wild type (Supplemental Figure S1B).

Figure 1.

BC16 mutation causes cell wall defects. A, Three-month-old wild type (WT), bc16, and complemented (Com^bc16) plants. Bars = 10 cm. B, Broken internodes of the indicated genotypes. The arrowhead indicates the breaking point. C, Measurement of the breaking force for internodes of the indicated genotypes. Data represent the mean ± sd of 14 internodes from individual plants of the indicated genotypes. D, Map-based cloning of BC16. M2−M56 indicate molecular markers for mapping. Numbers indicate the number of recombinant individuals. Arrows represent the ORFs. The arrowhead indicates the bc16 mutation site. The dashed line and box in the diagram of the BC16 protein indicate the portions truncated in bc16. Diagram of the construct for the complementation test (Com). E, SEM images of cross sections of fiber cells in the indicated genotypes. Bars = 2 μm. F, Boxplot of the area ratios of the cell wall to the whole cell in fiber cells of the indicated genotypes. Box boundaries represent the 25th and 75th percentiles, the center line represents the median, × indicates the mean, and whiskers represent the 25th percentile − 1.5 × the interquartile range and the 75th percentile + 1.5 × the interquartile range. One hundred fiber cells from three individual plants of the indicated genotypes were used for measurement. G and H, Cellulose content and cell wall composition in the internodes of the indicated genotypes. Data represent the mean ± sd of five biological replicates. Significant differences in this figure were calculated with two-tailed Welch’s unpaired t tests (*P < 0.05; **P < 0.01).

To identify the causal gene for the bc16 phenotypes, we performed map-based cloning in a F₂ population generated by crossing bc16 with an indica variety. Subsequently, the position of the causal gene was narrowed to a 47-kb region on chromosome 5 (Figure 1D) that contains six open-reading frames (ORFs). Sequencing analysis revealed a one-base pair insertion at the 675-bp site of ORF3 (Os05g05200) in bc16, which introduces a stop codon, resulting in a truncated translational product containing 225 amino acids (Figure 1D). To confirm that ORF3 is BC16, we transferred the wild-type genomic fragment containing the ∼2-kb upstream promoter sequence, the BC16 coding region, and a ∼0.5-kb terminator (Com, pBC16:BC16) into the bc16 mutant (Figure 1D). The transgenic plants (Com^bc16) had a wild-type-like phenotype, with rescued height and mechanical strength (Figure 1, A–C and Supplemental Figure S2), suggesting that ORF3 corresponds to BC16.

To investigate the alterations in fragile internodes, we examined the wall thickness of fiber cells and xylem vessels, as these cells constitute the mechanical tissues in plants. Using scanning electron microscopy (SEM), we found that bc16 fiber cells had thinner walls than those of the wild type (Figure 1, E and F). Transmission electron microscopy (TEM) revealed that the mutant also displayed decreased cell wall thickness in vessels (Supplemental Figure S1, C and D). Analysis of cell wall composition in the internodes demonstrated that the mutation in BC16 causes a reduction in cellulose level and an increased abundance of xylose, arabinose, galactose, and lignin compared with levels in the wild type (Figure 1, G and H), in agreement with the changes in cell wall composition previously reported for other rice bc mutants (Li et al., 2003; Zhang et al., 2009; Wu et al., 2012). The increases in the cell wall components, such as arabinoxylan (indicated by the higher xylose and arabinose content), pectin (indicated by the higher galactose content), and lignin, may be a feedback response to compensate for the deficient cellulose biosynthesis. These cellular and chemical abnormalities were fully restored by the introduction of the BC16 genomic fragment into the bc16 plants (Figure 1), indicating that BC16 contributes to cell wall biosynthesis and mechanical strength.

BC16 is an ER- and Golgi-localized protein

We next investigated the expression profile of BC16. Reverse transcription-quantitative PCR (RT-qPCR) analysis revealed that BC16 is ubiquitously expressed in rice (Figure 2A), consistent with the expression data available online (Supplemental Figure S3A). We further examined the BC16 expression pattern at a cellular level. Fiber cells and parenchyma cells were harvested from cross sections of wild-type young internodes by laser microdissection and subjected to RT-qPCR analysis. BC16 transcripts were detected in both cell types, with relatively higher abundance in fiber cells (Figure 2B), coinciding with the cell wall defects revealed by microscopy (Figure 1, E and F). BC16 is a ubiquitously expressed gene with certain cell-type preferences.

Figure 2.

Expression profile and localization pattern of BC16. A, RT-qPCR analysis of BC16 transcripts in the indicated tissues of the WT (Nipponbare), showing the relative level to rice HNR. S1–S8 indicate internode segments cut from the base to the tip of the developing 2^nd internode. Data represent the mean ± sd of three biological replicates. R, root; L, leaf; LS, leaf sheath; P, panicle. B, RT-qPCR analysis of BC16 transcripts in fiber cells (FC) and parenchyma cells (PC) collected by laser microdissection from internode sections (left panel). Data represent the mean ± sd of four biological replicates. Rice HNR was used as an internal control. The expression level in FC was set as 1. Bar = 100 μm. C and D, Confocal images and intensity profile plots (the dashed lines) of tobacco epidermal cells coexpressing BC16-GFP together with a mCherry-tagged ER marker (HDEL) or Golgi marker (Man49). Bars = 5 μm. E and F, Confocal images and intensity profile plots (the dashed lines) of rice protoplast cells coexpressing BC16-GFP together with an ER marker (HDEL) or a Golgi marker (Man49). Bars = 5 μm. G and H, Measurement of the degree of colocalization of the GFP-BC16 fusion with ER (HDEL) and Golgi (Man49) markers in tobacco epidermal cells (G) and rice protoplasts (H) using the Pearson correlation coefficient and Mander’s coefficient. The arrow schemes above the bars indicate the intensity overlap of HDEL/Man49 with BC16 and the reversals, according to the arrow directions. Data represent the mean ± sd of 10 leaf cells or 5 protoplast cells.

BC16 is predicted to have multiple transmembrane domains similar to its homologs in yeast and mammal (Supplemental Figure S3B). The yeast homolog Gup1 is an ER protein and mammalian homolog PGAP2 is in the Golgi apparatus (Fujita and Kinoshita, 2012). To determine the subcellular location of BC16, we fused green fluorescence protein (GFP) to the amino (N)-terminus or C-terminus of BC16 and introduced these constructs into Nicotiana benthamiana leaves and rice protoplasts with red fluorescent protein (RFP) mCherry fused to ER or Golgi markers, respectively. In the leaf epidermal cells that expressed either C terminal or N terminal fusion constructs, we observed reticular and particulate GFP signals that separately overlapped with those of the ER and Golgi markers (Figure 2, C and D and Supplemental Figure S4, A and B). These results, taken together with the similar localization patterns obtained in rice protoplasts and results of the colocalization assay using the Pearson correlation coefficient and Mander’s coefficient (Figure 2, E–H and Supplemental Figure S4, C–F), led us to conclude that BC16 locates in both organelles. Therefore, GPI lipid remodeling in plants likely occurs in the ER and Golgi apparatus.

BC16 functions as an acyltransferase in GPI lipid remodeling

Different from S. cerevisiae, which contains two MBOAT members, BC16 is the only MBOAT in the rice genome and shares sequence and structure (predicted by AlphaFold) similarity with yeast Gup1, an MBOAT for GPI modification (Supplemental Figure S5, A and B). Phylogenetic analysis grouped BC16 with plant homologs into one subclade (Supplemental Figure S5C). This, taken together with the co-expression of BC16 with genes involved in GPI modification (Supplemental Figure S6A and Supplemental Data Set S1), indicates that BC16 is likely a plant GPI acyltransferase. To test this hypothesis, we expressed BC16 in a S. cerevisiae mutant that is deficient in Gup1 (Bosson et al., 2006). The mutation in Gup1 disrupts cell wall structure and makes the yeast mutant cells unable to grow on medium containing calcofluor white (CFW), a cell wall dye (Bosson et al., 2006; Ram and Klis, 2006). The mutant yeast cells harboring BC16 or BC16-GFP grew normally on the medium containing CFW, whereas those expressing the empty vector were unable to survive on the same medium (Figure 3A). To confirm that the surviving yeast cells indeed contained the transgenes, we spotted the yeast cells on SD medium without uracil (SD-Ura). Only the transgenic cells grew normally on SD-Ura medium due to the presence of the URA3 gene in the vector (Figure 3B). Adding CFW to the SD-Ura medium suppressed the growth of cells expressing the empty vector or GFP alone but not that of cells expressing BC16 or BC16-GFP (Figure 3B), demonstrating that rice BC16 plays a similar role to yeast Gup1.

Figure 3.

BC16 remodels the lipid tail of GPI-APs. A, Yeast complementation assay. Images show growth status of WT cells (BY4742), gup1Δ cells, and gup1Δ cells containing the indicated constructs on YPD or YPGal media containing 50 µg/mL CFW. B, Validation of transgenic yeast strains. Growth status of the indicated yeast cells on SD-Ura medium or that plus CFW. C, Lipid content of ConA-enriched GPI-APs extracted from the indicated yeast strains. pG1 (PI containing saturated long-chain fatty acids) is the sum of PI 40:0 and PI 42:0 in the middle panel. IPCs is the sum of IPC (including the hydroxylated forms) in the right panel. D, Lipid content of ConA-enriched GPI-APs extracted from rice internodes. PI (phosphatidyl inositol with a saturated long chain fatty acid) is the sum of PI 40:0 and PI 42:0 in the middle panel. IPCs is the sum of IPC (including the hydroxylated forms) in the right panel. Data in this figure represent the mean ± sd of three biological replicates. Significant differences in comparison to the WT were calculated with two-tailed Welch’s unpaired t tests (*P < 0.05; **P < 0.01).

GUP1 was identified as the second step enzyme that mediates the transfer of a saturated long chain fatty acid to the sn‐2 position in GPI lipid remodeling (Bosson et al., 2006), responsible for generating mature GPI lipid tails (pG1 or IPCs) (Supplemental Figure S6B). To determine the GPI lipid structure formed by Gup1, we used Concanavalin A-Sepharose (ConA) to extract GPI-APs from gup1Δ yeast, the BC16 complemented cells (gup1Δ-BC16), and wild-type strain (BY4742) and subjected them to lipid structure analysis via liquid chromatography-triple quadrupole mass spectrometry (LC-MS/MS). Compared with the GPI-APs from wild-type yeast cells, those from the gup1Δ mutant had a much lower amount of pG1 and IPCs, in agreement with the lipid defects reported for this mutant (Bosson et al., 2006; Fujita et al., 2006). The defects mainly included a ∼90% reduction in PI 40:0, a ∼76% reduction in PI 42:0 (termed pG1), and a ∼75% reduction in IPCs that consist of phytosphingosine (PHS; t18:0/t20:0) and a C26:0 or C24:0 fatty acid (Figure 3C and Supplemental Data Set S2). The levels of IPCs with varied hydroxylation were consistently lower in the gup1Δ mutant than in the wild type (Supplemental Figure S7A). More importantly, all those defects were almost restored by expressing rice BC16 in gup1Δ cells, as shown by LC-MS/MS analysis (Figure 3C and Supplemental Figure S7A). Hence, BC16 can function as a GPI lipid acyltransferase like Gup1.

To examine the lipid profile of plant GPI-APs, we applied the same strategy to isolate GPI-APs from wild-type and bc16 plants. The lipid composition analysis revealed that the GPI-APs extracted from the wild-type rice plants mainly have PI-type or IPC-type (containing varied hydroxylated forms) lipid tails (Figure 3D and Supplemental Figure S7B and Supplemental Data Set S2), similar to in yeast. More importantly, mutation in BC16 results in GPI lipid defects similar to those in gup1Δ. In bc16, the contents of PI tails, including PI 40:0 and PI 42:0, and the levels of IPCs, including IPC (t18:0/24:0), IPC (t18:0/26:0), IPC (t20:0/26:0), and the hydroxylated forms, were reduced to only ∼20% of the levels in the wild type (Figure 3D and Supplemental Figure S7B). Furthermore, we overexpressed BC16-GFP in N. benthamiana leaves and collected the transformed leaves for GPI lipid profile analysis. We found that overexpression of BC16-GFP exclusively augmented the amount of PI 42:0 and certain IPCs in the extracted GPI-APs compared with expression of GFP alone (Supplemental Figure S7, C and D). These analyses demonstrated that BC16 is a lipid acyltransferase in GPI anchor modification.

BC16 directs BC1 to the cell surface

The bc16 phenotypes, including brittleness and abnormal cell wall composition, are reminiscent of a previously reported rice brittle culm mutant bc1, which is deficient in a GPI-AP COBL5 (Li et al., 2003; Liu et al., 2013). Therefore, BC1 could be one of the BC16-modified GPI-APs contributing to cell wall mechanics. We next investigated the retention of BC1 at the PM in wild-type and bc16 protoplast cells, since GPI anchoring usually serves as a signal for targeting GPI-APs to the cell surface. We coexpressed GFP-tagged BC1 (GFP-BC1) and a mCherry-tagged PM marker (PIP2A) in rice protoplasts and found that most of the BC1 signal was on the PM of the wild-type cells, whereas the BC1 signal was mostly retained inside of the bc16 mutant cells (Figure 4, A–C). Quantification of protoplast cells bearing GFP signal further revealed that approximately 93% of wild-type cells had the PM-localized signal, compared with only 23% of bc16 cells (Figure 4D). To determine whether this reduced PM localization was due to the defects in GPI modification, we expressed GFP-tagged and C-terminally truncated (from the GPI attachment site) BC1 (BC1^ΔGPI) in wild-type protoplasts. In this case, most of the GFP signal was located inside of the cells (83%), similar to the bc16 cells expressing full-length BC1 (Figure 4, A–D). Similar intracellular signal retention patterns were observed in bc16 cells expressing either BC1 or BC1^ΔGPI (Figure 4D), indicating that delivering BC1 to the PM is largely dependent upon GPI modification, including lipid remodeling. We carried out this analysis for two additional GPI-APs, Arabidopsis Arabinogalactan protein4 (AGP4) and rice Fasciclin-like arabinogalactan-protein9 (FLA9). Both proteins showed a similar reduction in PM localization in bc16 cells (Figure 4, A–C), suggesting that the influence of BC16 on PM localization is not limited to BC1 and is applicable to other GPI-APs. To investigate whether BC16 can facilitate BC1 release to the apoplast and targeting to the secondary cell walls, we performed TEM immunolabeling of cross sections of wild-type and bc16 internodes using BC1-specific antibodies (Liu et al., 2013). We detected more gold particles in the wild-type fiber cell walls than in the mutant fiber cell walls (Supplemental Figure S8). BC16 mutation disturbs the targeting of BC1 to the PM and secondary walls.

Figure 4.

BC16 influences BC1 targeting to the PM. A, Representative images of WT and bc16 protoplasts coexpressing GFP-tagged BC1, BC1^△GPI, Arabidopsis AGP4, or rice FLA9 with mCherry-tagged PIP2A (a PM marker). Bars = 5 μm. B and C, Quantification of the degree of colocalization of the examined GPI-APs with the PM marker PIP2A in WT and bc16 protoplasts using the Pearson correlation coefficient (B) and Mander’s coefficient (C). The arrow schemes above the bars indicate the intensity overlap of PIP2A with BC1/BC1^△GPI/FLA9/AGP4 and the reversals, according to the arrow directions. Data represent the mean ± sd of 5 cells. D, Quantifying the proportion of BC1 and BC1^△GPI localized inside and on the PM of WT and bc16 protoplasts. The data represent the mean proportion of three transfection replicates. In each replicate, at least 25 cells were subjected to examination. E, Immunoblot of BC1 in total microsome proteins (TM) extracted from young internodes of the WT and bc16, and in detergent-soluble fraction and DRM fraction of TM upon 1% Triton treatment. PIP1s and CESA9 were used as controls. F, Relative abundance of BC1 present in the detergent-soluble and DRM fractions of WT and bc16 plants according to the band intensity of immunoblots. Data represent the mean ± sd of three biological replicates. G and H, High-resolution confocal images of N. benthamiana cells coexpressing Arabidopsis VND6, mRFP-HIR1 & mNG-BC1 (G), and mRFP-HIR1& mNG-CESA4 (H). Bar = 2 μm. I, Colocalization coefficients of HIR1 & BC1 signal (G) and HIR1 & CESA4 signal (H) using the Pearson correlation coefficient and Mander’s coefficient analysis. The arrow schemes above the bars indicate the intensity overlap of HIR1 with BC1/CESA4 and the reversals, according to the arrow directions. Data represent the mean ± sd of 10 cells from three individual plants. Significant differences were calculated with two-tailed Welch’s unpaired t tests (**P < 0.01).

The PM is usually heterogeneous, with multiple functional domains. Next, we determined whether the BC1 proteins locate in specialized lipid nanodomains. As detergent resistance is a common property of these domains (Simons and Ikonen, 1997), we treated microsomes extracted from wild-type and bc16 plants with a detergent. We found that bc16 displayed reduced BC1 signal relative to the wild type. Moreover, the proportion of BC1 retained in the detergent-resistant membrane (DRM) fraction was decreased in bc16 compared with that in the wild type, but the proportion of PIP1s, another kind of DRM-resident protein, was not affected in bc16 (Figure 4, E and F).

We then used the lipid nanodomain marker HYPERSENSITIVE-INDUCED REACTION1 (HIR1) (Lv et al., 2017) to discern the PM nanodomains in which BC1 resides. We transiently coexpressed RFP (mRFP)-tagged HIR1 with GFP-BC1 in the wild-type and bc16 protoplasts (Supplemental Figure S9A). In contrast to the wild-type cells, in which most of the BC1 signal was at the PM and colocalized with HIR1, the BC1 signal was mainly inside in the bc16 cells (Supplemental Figure S9, B, D, and E). Cellulose synthase complex (CSC), which catalyzes cellulose synthesis at the PM, was also found in DRMs (Figure 4E; Bessueille et al., 2009). To test whether CSCs are also in HIR1-harboring nanodomains, we introduced GFP-HIR1 and mScarlet-CESA4 (a secondary cell wall cellulose synthase as CESA9) into wild-type and bc16 protoplasts (Supplemental Figure S9A). The CESA4 signal overlapped with the HIR1 signal at the PM in the wild type, and this localization was not affected in bc16 (Supplemental Figure S9, C–E), indicating that BC1 and CESA4 can position at the PM lipid nanodomains harboring HIR1, but likely via different delivery processes. We further expressed mRFP-HIR1 and mNeonGreen (mNG)-BC1 or mNG-CESA4 in N. benthamiana leaf epidermal cells and observed the overlapped signals using high-resolution confocal microscopy (Supplemental Figure S9, F−I). To verify these associations in cells undergoing secondary cell wall synthesis, we coexpressed Arabidopsis VASCULAR-RELATED NAC DOMAIN6 (VND6) in N. benthamiana leaf epidermal cells to ectopically induce secondary cell wall formation (Supplemental Figure S10). In these cells, HIR1 also largely colocalized with BC1 or CESA4; the overlapped signals tended to have a band-like pattern that roughly coincided with the growing secondary cell wall bands (Figure 4, G–I). Therefore, BC16-mediated lipid remodeling is involved in targeting BC1 to the PM lipid nanodomains that also likely gather CSCs.

Cellulosic nanofiber organization is altered in bc16

BC16 regulates BC1 localization onto the PM, prompting us to investigate whether both mutants have common effects on cell wall structure. We examined the deposition of cellulosic nanofibers in fiber cells because that is where the major cell wall defects display. By atomic force microscopy (AFM) analysis, we observed a broader distribution of nanofiber orientation (less alignment of microfibrils) in bc16 compared with in the wild type (Figure 5, A and B). The aligned cellulosic fibers in the wild type could be bundled into thicker nanofibers than those in bc16 (Figure 5C), indicating that cellulose assembly is altered in the bc16 mutant. As we expected, similar defects in nanofiber orientation and thickness were found in bc1 compared with in the corresponding wild-type plants, although the cellulosic nanofibers in bc1 were even more scattered than those in bc16 (Figure 5, A–C). Considering that CESA4 can reside in the HIR1-harboring nanodomains like BC1, bc11 (a missense mutant deficient in CESA4) was also subject to AFM analysis (Zhang et al., 2009). The bc11 fiber cell walls displayed similar scattered and thin nanofibers as bc1 (Figure 5, A–C), indicating a potential connection between both proteins. Taken together, these results indicate that BC16 affects cellulosic nanofiber organization, comparable to the impacts of BC1.

Figure 5.

BC16 regulates cellulose nanofiber assembly. A, Representative AFM images of cell wall structure in the indicated genotypes. The bc mutants and their corresponding WT varieties are indicated. Bars = 100 nm. B, Distribution of cellulosic nanofiber orientation in fiber cells of the indicated genotypes. The number of nanofibers (snakes) at each orientation is represented as a percentage of the total number of nanofibers (snakes) identified using SOAX software (n > 18,000 snakes from images of three fiber cells from three individual plants). C, Boxplots of cellulosic nanofiber width in fiber cell walls of the indicated genotypes. Box boundaries represent the 25th and 75th percentiles, the center line represents the median, × indicates the mean, and whiskers represent the 25th percentile − 1.5 × the interquartile range and the 75th percentile + 1.5 × the interquartile range. Data represent the mean ± sd of 100 nanofibers in three fiber cells from the indicated genotypes. Significant differences were calculated with two-tailed Welch’s unpaired t tests (**P < 0.01).

BC16 regulates cell wall mechanics

Cellulosic nanofibers constitute the cellular load-bearing network. To investigate whether the altered nanofiber organization affects cell wall mechanical properties, we isolated fiber cells from the internodes of wild-type and bc16 plants (Supplemental Figure S11) and used AFM nanoindentation to measure the elastic modulus of individual fiber cells. For these experiments, we applied a 100-nN nano-indentation force with a triangular pyramid AFM tip and then rapidly retracted the AFM tip for scanning a 3 µm × 3 µm region of the cell wall surface in the central region of each fiber cell. The force curves during indentation and retraction were recorded and the Young’s moduli were calculated. We repeated the measurements at three different sites in each fiber cell and found that the slopes of the indentation curves were higher for the mutant fiber cells than for the wild-type fiber cells (Figure 6A). The discrepancy suggested that the mutant cell walls underwent less elastic deformation than the wild-type cell walls when external forces were applied. Surprisingly, Young’s moduli for the bc16 cell walls ranged from 10 to 270 MPa, whereas that for wild-type walls ranged from 10 to 130 MPa, resulting in an average modulus for the bc16 fiber cell walls twice as high as that of the wild-type fiber cell walls (Figure 6, B and C). Therefore, the bc16 fiber cells have significantly higher cell wall rigidity than the wild-type cells.

Figure 6.

bc16 plants have abnormal mechanical properties. A, Representative force distance curves generated by AFM-based nano-indentation on WT and bc16 fiber cells. Approach and retract data were fitted as a curve. B and C, Measurement of Young’s modulus of WT and bc16 fiber cell walls using AFM-based nano-indentation. The data are expressed in a frequency plot (B) and bar chart (C). Data represent the mean ± se of five fiber cells. Significant differences were calculated with two-tailed Welch’s unpaired t tests (**P < 0.01). D and E, Tensile strength of three internodes of the indicated genotypes. The load-displacement curves (D) and the calculated stress-strain curves (E) were determined on a tensile tester.

To investigate mechanical properties at the tissue level, we subjected wild-type and bc16 internodes to a tensile test. Under gradually increasing load force, the rice internodes were stretched and displaced until they finally ruptured. Based on the generated load-deformation curves, we recorded a maximum load force of 239.3 ± 33.0 (N) for the wild-type internodes and 41.2 ± 14.2 (N) for bc16 internodes, approximately one-fifth of that of the wild type (Figure 6D). Meanwhile, the extension length of the wild-type internodes at the point of rupture was 2.2 ± 0.4 (mm), two to four times that of bc16 internodes (Figure 6D). The stress–strain diagram further revealed that the maximum tensile strength of wild-type internodes was significantly greater than that of bc16 internodes (34.9 versus 8.2 MPa, Figure 6E). The strain of the wild-type internodes at the point of rupture was also two times higher than that of the bc16 internodes (4.3% versus 1.3%, Figure 6E). These data indicate that bc16 internodes are fragile and less flexible compared with wild-type internodes. Hence, the abnormal nanofiber organization in bc16 results in unusually weak mechanical properties.

Discussion

GPI-anchoring is an important post-translational modification that targets proteins to the outer surface of the PM. The lipid portion of GPI anchors contributes to localization of GPI-APs to the functional destination at the cell surface (Yoko-o et al., 2018). Plants have around 300 GPI-APs (Yeats et al., 2018), but the GPI constituents and lipid moieties are largely unknown. Furthermore, how the GPI lipid portion is synthesized in plants is poorly understood. In this work, we characterize BC16 as a GPI lipid remodelase that regulates the lipid structure of GPI-APs in rice. We also clarify the role of BC16 in directing BC1 (a GPI-AP) to the PM, as well as its influence in governing cell wall mechanics.

Based on previous studies in yeast and mammals, lipid remodeling of the GPI anchor includes four to five deacylation and reacylation reactions (Fujita and Kinoshita, 2012). Arabidopsis PGAP1 has been identified as a GPI inositol-deacylase, responsible for the first-step reaction of lipid remodeling in plants (Bernat-Silvestre et al., 2021). Here, we identify BC16 as an acyltransferase for GPI remodeling in plants. Its ortholog in yeast, Gup1, acts as an enzyme that transfers a long-chain saturated fatty acid to the sn-2 position of diacylglycerol, converting lyso-PI tail to pG1 (Bosson et al., 2006; Jaquenoud et al., 2008). The activity of BC16 and Gup1 is conservative because BC16 fully restored the growth defects of the yeast mutant gup1Δ and expressing BC16 in gup1Δ elevated the dramatically reduced amount of pG1 (including PI 40:0 and PI 42:0) and IPCs almost to the wild-type level. Furthermore, the AlphaFold-predicted protein structure models for BC16 and Gup1 are nicely superimposed. Disruption of rice BC16 displayed similar defects in PI 40:0 and PI 42:0 as the yeast gup1Δ mutant. The nearly 80% reduction in PI indicated that PI 40:0 and PI 42:0 may be the direct products of the BC16 reaction. In yeast, pG1 is usually replaced by a ceramide moiety consisting of PHS through the next step reaction (Fujita et al., 2006; Yoko-o et al., 2013). The decrease in IPCs is likely a secondary effect of the deficiency in BC16/Gup1. Although further studies will need to validate the authentic activity of BC16, our data support the notion that BC16 is a lipid remodelase like Gup1. BC16 is the only MBOAT in rice. The remaining PI in bc16 (20%) indicates that an alternative GPI lipid-remodeling pathway may exist in rice. In fact, genetic studies have validated alternative lipid-remodeling pathways in yeast (Yoko-o et al., 2013).

More importantly, through LC-MS/MS analysis, the data regarding to the alterations of GPI lipids in the gup1Δ mutant were qualitative and quantitative, providing clear evidence for Gup1 function, in comparison with that obtained by thin-layer chromatography (Bosson et al., 2006). Furthermore, we demonstrated that the lipid portion of ConA-enriched plant GPI-APs is yeast-like, possessing a PI-type or IPC-type lipid tail, in agreement with the previous reports for AGP1 in pear cells and Yariv enriched AGPs in rose (Oxley and Bacic, 1999; Svetek et al., 1999). This finding is also in agreement with the fact that both plant and yeast cells harbor cell walls. In fact, the lipid moieties of plant GPI-APs have long been unknown. Here, lipid profiles in yeast, rice, and tobacco provide unprecedented insights into GPI lipid structure, which furthers our understanding of GPI modification in different species.

The cell surface is the final destination of GPI-APs. The factors affecting GPI-AP trafficking are crucial for their functions. According to studies of yeast gup1Δ and cwh43Δ mutants (Bosson et al., 2006; Yoko-o et al., 2018), GPI lipid moieties appear to be important for GPI-AP sorting in yeast. However, their impacts in plants remain unclear. In this work, through the expression of several GPI-APs in wild-type and bc16 protoplasts, we revealed that BC16 mediates the delivery of the examined GPI-APs, especially BC1, to the PM. BC1 is one of the GPI-APs regulating cell wall mechanical strength. We previously reported that the function of BC1 is dependent upon its cell wall localization and GPI modification (Liu et al., 2013). Here, a series of examinations in rice protoplasts corroborated the notion that GPI lipid remodeling is essential for localization of BC1 to the PM and cell wall.

The association between GPI-APs and DRMs is known in yeast and largely relies on GPI lipid remodeling (Fujita et al., 2006; Yoko-o et al., 2013). Here, we found BC1 in DRMs, as revealed by biochemical analysis and high-resolution confocal microscopy. Disruption of BC16 alters BC1 localization in DRMs and in the HIR1-harboring nanodomains. We further found that the DRM-associated CESA proteins (Colombani et al., 2004; Bessueille et al., 2009) were also present in the HIR1-harboring lipid nanodomains, albeit the delivery pathways seem different. AFM revealed comparable alterations in cellulosic nanofibers in bc16, bc1, and cesa4/bc11, suggesting a correlation of these proteins in nanofiber organization. However, the relationship between these proteins remains elusive. In addition to the effect on BC1 localization, BC16 may regulate BC1 protein abundance since a low BC1 amount was detected in the bc16 plants, but the underlying mechanism is unclear. Further studies are necessary to address these issues.

Regulation of cell wall mechanics is one of the functions executed by plant GPI-APs (Li et al., 2013; Liu et al., 2013; Ben-Tov et al., 2015; Yeats et al., 2018; Bernat-Silvestre et al., 2021). However, how GPI-APs orchestrate cell wall mechanical property is still largely unclear. Stem fracture is a common phenotype of cell wall defective mutants in rice, maize, and wheat (Sindhu et al., 2007; Zhang and Zhou, 2011; Deng et al., 2019). Although alterations in cell wall composition have been extensively studied, the mechanical property underlying brittleness is poorly understood. Here, by measuring nanoscale elastic moduli, we found that the fiber cell wall of bc16 exhibited increased rigidity and decreased elasticity relative to that of the wild type. It is worth noting that the elastic moduli analyzed here were applied on the isolated fiber cells, which revealed the mechanical differences, but might not fully indicate the characteristics of native cell walls. Furthermore, tensile tests verified this mechanical property at the tissue level. Our study hence offers insights into the physics of brittleness in plants. In addition to the impact on mechanical support, BC16 is likely involved in other biological processes, such as plant height control.

In summary, we report that BC16 functions as an O-acyltransferase to modify the lipid moieties of GPI anchors in plants. This remodeling is required for the associations of GPI-APs, especially BC1, to the PM nanodomains, to facilitate cellulose synthesis and organization likely together with CSCs, thereby regulating cell wall biomechanical properties and plant erect growth (Figure 7).

Figure 7.

Working model of how BC16 regulates cellulose assembly and cell wall mechanical strength. BC16 functions as an O-acyltransferase that remodels the GPI lipid tail in plants. BC16 is localized in the ER and Golgi. BC1 is one of the GPI-APs modified by BC16. The GPI-anchored BC1 proteins are delivered to the PM via vesicles (v) and reach the lipid nanodomains, where CSC is also present. PM-localized BC1 and CSC together facilitate assembly of cellulosic nanofibers in cell walls (CW), thereby determining stem strength and supporting an erect growth habit. The grey letters and dashed arrows indicate the enzymes and steps that have not yet been reported in plants.

Materials and methods

Plant materials and growth condition

The rice (O. sativa) brittle culm mutants, including bc16/gup1, bc1/cobl5, and bc11/cesa4, and the corresponding wild-type accessions (ZH11, Wuyujing3, and Nipponbare), used in this study were planted in the fields at the Institute of Genetics and Developmental Biology in Beijing (China), College of Agriculture, Yangzhou University in Yangzhou (Jiangsu Province, China), or in Lingshui (Hainan Province, China) in the growing seasons.

To clone the BC16 gene, 1,200 mutant plants in an F₂ population generated by crossing bc16 with an indica wild-type cultivar Kasalath were subjected to construction of the linkage map using molecular markers (Supplemental Table S1). After narrowing down the cloning region within a 47-kb region on chromosome 5, the DNA fragments of the six ORFs in the mutant and wild-type plants were amplified and sequenced using a 3730 sequencer (ABI). For complementation, a 9,300-bp wild-type genomic fragment containing a 2,080-bp promoter, the BC16 (Os05g05200) gene, and a 532-bp terminator was cloned into the pCAMBIA1300 vector (pBC16:BC16). The resulting construct was introduced into Agrobacterium tumefaciens strain EHA105 to transfect calli generated from bc16 seeds.

Electron microscopy

The internodes from mature wild-type, bc16, and complementary plants were collected and fixed in 4% paraformaldehyde (Sigma-Aldrich). After dehydration through a gradient of ethanol and critical point drying, the samples were sprayed with gold particles and observed with a scanning electron microscope (S-3000N, Hitachi). The wall thickness of fiber cells was quantified as the ratio of cell wall area to that of the whole cell. At least 100 fiber cells from three individual plants were recorded for each genotype. For the TEM assay, the fixed internodes of wild-type and bc16 plants were embedded in butyl methyl methacrylate (Sigma-Aldrich). Approximately 90-nm-thick sections were cut using a microtome (EM UC6, Leica). After post-staining using 2% uranyl acetate (w/v), the samples were observed under a transmission electron microscope (HT7700, Hitachi) equipped with a charge-coupled device camera (Gatan 832). Cell wall thickness of metaxylem vessels was calculated based on 30 measurements in at least 7 vessels harvested from three individual internodes. For immunogold electron microscopy, ∼100-nm-thick sections from three individual plants were probed with affinity chromatography-purified anti-BC1 antibodies at 1:300 dilution (Liu et al., 2013). Secondary antibody, 10-nm colloidal gold-conjugated goat anti-rat IgG (G7035, Sigma), was applied at a 1:20 dilution. After post-staining with 2% uranyl acetate in 70% ethanol, the images were acquired using a Hitachi HT7700. Gold particles were counted in more than 45 cell wall areas in 10 cells from the internode sections of three individual plants.

Phylogenetic analysis

The BC16 sequence was obtained from a rice database (https://rice.uga.edu) and aligned using ClustalW (https://www.ebi.ac.uk/Tools/msa/clustalw2). The alignment data were viewed using Jalview version 2.11 (https://www.jalview.org/). MBOAT members in O. sativa, Arabidopsis thaliana, S. cerevisiae, Homo sapiens, and Drosophila melanogaster were analyzed according to the sequences retrieved from the NCBI database. An unrooted tree was established using the Neighbor-joining method with 1,000 bootstrap replications in MEGA5 (https://megasoftware.net). Detailed information for the sequence alignment and phylogenetic tree analysis are provided in Supplemental Files S1 and S2.

Bioinformatics

The domain structure of BC16 was analyzed using TMHMM2.0 (http://www.cbs.dtu.dk/services/). A coexpressional network of BC16 was acquired using ATTED-II (http://atted.jp) and listed in Supplemental Data Set S1. For structural model comparison, AlphaFold models of BC16 (AF-Q0DKT8) and SceGUP1 (AF-P53154) were retrieved from the AlphaFold Protein Structure Database (https://alphafold.com/) and aligned using the PDB pairwise alignment tool (https://www.rcsb.org/alignment). The superimposed models were visualized using UCSF ChimeraX (https://www.cgl.ucsf.edu/chimerax).

RT-qPCR analysis

The expression profile of BC16 in different organs was obtained from online expression data (http://ricexpro.dna.affrc.go.jp/). Roots, leaf sheathes, leaves, young panicles, and 9-cm-long internodes were collected from the wild-type accession Nipponbare at proper growth stages. The 9-cm-long internodes were further cut into nine segments. All segments except the ninth segment were subjected to extraction of total RNA using Plant RNA Reagent (Invitrogen). RT-qPCR was performed in 20 μL of FastStart universal SYBR green master qPCR mix (4913914001, Roche) in a CFX96 RT-qPCR system (Bio-Rad) using the primers shown in Supplemental Table S2. The PCR program included a 5-min pre-degeneration at 95°C, followed by 45 cycles of 10 s denaturation at 95°C, 25 s annealing at 58°C, and 30 s extension at 72°C. To investigate BC16 expression at the cellular level, young internodes were embedded in paraffin and subjected to laser microdissection (Zhang et al., 2018). The 12-μm-thick sections were prepared and used to harvest epidermal fiber cells and parenchyma cells by the LMD 7000 laser microdissection system (Leica). Total RNA was isolated from the collected samples using an RNeasy micro kit (QIAGEN) and were analyzed using qPCR. Rice HNR (Os01g71770), which encodes the heterogeneous nuclear ribonucleoprotein 27C, was used as an internal control in this analysis (Huang et al., 2015). Three or four biological replicates were included in the experiment.

Subcellular localization

GFP was inserted into the amino and carboxyl of BC16 via in-frame fusion in the pCAMBIA1300 or pUC19 vector between the Cauliflower mosaic virus (CaMV) 35S promoter and the NOS terminator (Supplemental Table S2). For transient expression in rice protoplasts, 2-week-old Nipponbare seedlings were used to generate protoplast cells. The resulting constructs were infiltrated into 4-week-old N. benthamiana leaves or transfected into rice protoplast cells together with a mCherry-tagged Golgi (Man49-mCherry) or ER (HDEL-mCherry) marker. After incubation, the fluorescent signal was analyzed with an Airyscan2 confocal microscope (LSM 980, Zeiss). Fluorescence was detected at 488 nm for excitation and 505–550 nm for emission (GFP channel), and 561 nm for excitation and 600–650 nm for emission (mCherry channel). The Pearson correlation coefficient and Mander’s coefficient were used to quantify the degree of colocalization in 10 leaf cells or 5 protoplast cells.

Cell wall composition

The second internodes of mature wild-type, bc16, and complementary plants were dried and ball milled into fine powders. Alcohol insoluble residues were prepared, destarched, and hydrolyzed in 2 M trifluoroacetic acid. The supernatant was used to generate alditol acetate derivatives. The monosaccharide composition was determined by mass spectrometry-coupled gas chromatography (Agilent). The pellet was treated with Updegraff reagent and then used to analyze the cellulose content (Updegraff, 1969). The lignin content was examined according to the acetyl bromide method (Foster et al., 2010). Five biological replicates were included in these examinations.

Yeast complementation assay

Saccharomyces cerevisiae yeast used in this study, including the wild-type strain and gup1Δ mutant, was purchased from EUROSCARF (http://www.euroscarf.de/). BC16 and BC16-GFP were in-frame inserted into the expression vector pYES2 between the galactose-inducing GAL1 promoter and the CYC1 terminator. The resulting constructs, as well as the empty vector and the vector containing GFP alone, were introduced into the gup1Δ mutant using the lithium acetate/polyethylene glycol method. The transgenic strains and wild-type yeast were spotted on YPD medium (10 g/L yeast extract, 20 g/L Bacto peptone, 20 g/L glucose, and 20 g/L agar) to observe their growth. To induce BC16 expression, 20 g/L galactose was added into YPD to replace glucose (YPGal). The transgenic lines were validated by growing them on SD-Ura medium (37 g/L SD base, 0.77 g/L -Ura dropout supplements, 20 g/L galactose, and 20 g/L agar). Drug sensitivity was tested by spotting 3 μL of the diluted yeast suspensions onto the relevant medium containing 50 µg/mL CFW. After cultivation at 30°C for 3 days, the growth of yeast cells was recorded.

Lipid structure analysis of GPI-APs

Young internodes were harvested from wild-type and bc16 plants at the heading stage and subjected to isolation of total membrane fractions. After homogenizing in an extraction buffer (25 mM Tris–HCl pH 7.5, 2 mM EDTA, 2 mM DTT, 15 mM mercaptoethanol, 0.25 M sucrose, 10% glycerol, and protease inhibitor cocktail), the samples were centrifuged at 10,000 × g for 10 min, followed by ultracentrifugation at 100,000 × g for 1 h. To isolate the membrane-associated glycoproteins, the pellets were resuspended in a lysis buffer (50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 2% Triton X-100 [v/v], 2% DDM [w/v], and proteinase inhibitor cocktail) at 4°C for 1 h. After centrifugation at 20,000 × g for 30 min, the supernatant was diluted in a binding buffer (20 mM Tris, 500 mM NaCl, 1 mM MnCl₂, 1 mM CaCl₂, pH 7.4) and applied into a HR column with ConA-sepharose 4B beads (17-0440-03, GE Healthcare) to concentrate GPI-APs (Guillas et al., 2000). The unbound proteins were removed using a wash buffer (20 mM Tris–HCl, 0.5 M NaCl, pH 7.4).

The resulting GPI-APs were eluted with 1 M methyl-α-d-mannopyranoside and quantified using a protein assay kit (Bio-Rad). GPI lipid moieties were released from 50 μg eluted proteins by deamination using sodium nitrite (Yoko-o et al., 2018). The released lipids were analyzed using 1290 HPLC (Agilent Technologies) coupled with Agilent 6490 triple quadrupole using head group- and fatty acyl-specific multiple reaction monitoring (MRM) (Lam et al., 2014). In brief, individual GPI lipids were separated using an ACQUITY HSS T3 column (internal diameter 100 × 2.1 mm) with mobile phase A (acetonitrile: H₂O, 65: 35) and mobile phase B (isopropanol: acetonitrile, 90: 10, with 10 mM ammonium formate). To eliminate the contaminants derived from free lipids, the extraction aliquots prior to releasing GPI lipids (using sodium chloride instead) were also subjected to LC-MS/MS examination in parallel. The lipid content of released GPI lipids minus that of the counterparts prior to release was referred to as the content of examined GPI lipids. The quantification was carried out using soy PI (840044, Avanti) as a reference. The resulting unsaturated PIs were not counted. Three biological repeats were included in this examination.

Detergent treatment

Total membrane proteins were extracted from the wild-type and bc16 internodes in the above protein extraction buffer. After centrifuging at 12,000 × g for 10 min, the supernatant was pelleted by ultra-centrifuging at 100,000 × g for 1 h. Approximately 1 mg of the pellet was resuspended in an extraction buffer containing 1% Triton X-100 at 4°C for ∼30 min. The resulting samples were centrifuged at 16,000 × g for 20 min to separate the solubilized fraction and pellets. The resulting pellets were resuspended in a buffer (60 mM β-octylglucoside, 10 mM Tris HCl pH 7.5, and 150 mM NaCl) and subjected to blotting using primary antibodies against BC1 (Liu et al., 2013) and PIP1s (Agrisera, AS09 505, a PM marker) and CESA9 (generated in rabbit immunized with a polypeptide [105–245 aa] of rice CESA9) at a 1:1,000 dilution. The secondary antibody, peroxidase-conjugated goat anti-rabbit IgG (H + L) (IH-0011, Dingguo), was applied at a 1:3,000 dilution. The relative amount of BC1 was calculated based on the band intensity of immunoblots from three independent experiments.

Microscopy

To examine the influence of BC16 on sorting GPI-APs to the PM, BC1 (Os03g30250), Arabidopsis AGP4 (AT5g10430), and rice FLA9 (Os05g07060) were cloned using the primers shown in Supplemental Table S2. GFP was in-frame fused to these proteins after the signal peptide and inserted into the pUC19 or pCAMBIA1300 vector. BC1 truncated from the ω site (BC1^ΔGPI) was used as a negative control. The plasmids of the resulting constructs (in the pUC19 vector) were purified using a DNA Purification kit (QIAGEN). To obtain the protoplasts from rice seedlings, 2-week-old wild-type and bc16 seedlings growing in 1/2 MS agar were subjected to isolation of protoplasts. The above plasmids and mCherry-fused PIP2A (a PM marker) were transiently introduced into the protoplasts via the PEG method (Zhang et al., 2018). After overnight incubation, the protoplasts were observed using an Airyscan2 confocal microscope (LSM 980, Zeiss). The fluorescent signals were detected at 488 nm for excitation and 505–550 nm for emission (GFP channel), and 561 nm for excitation and 600–650 nm for emission (mCherry channel). At least five cells were analyzed for the degree of colocalization.

To specify the localization to the PM lipid nanodomains, BC1, CESA4 (Os01g54620), and HIR1 (AT01g69840, a lipid nanodomain marker) were cloned and fused with mNeoGreen (mNG), mScarlet, or mRFP at the N terminus, and inserted into the pCAMBIA1300 vector between the CaMV 35S promoter and the NOS terminator. For ectopically inducing secondary cell walls, VND6 (AT5g62380) was cloned into the pCAMBIA1300 vector and introduced into A. tumefaciens strain EHA105. The resulting constructs were infiltrated into 4-week-old N. benthamiana leaves. Fluorescent signals were observed using an Airyscan2 confocal microscope (LSM 980, Zeiss) and detected at 488 nm for excitation and 505–550 nm for emission (GFP channel); 561 nm for excitation and 600–650 nm for emission (RFP channel); and 405 nm for excitation and 440–480 nm for emission (UV channel). Ten cells were analyzed for colocalization efficiency.

AFM

The mature second internodes of bc16, bc1, and bc11 mutants, as well as the corresponding wild-type plants, were treated in 11% peracetic acid solution. Fiber cells were isolated from the softening tissues by micromanipulation. The fibers were scanned at a 1-μm scale at 512 × 512 pixels using the ScanAsyst-Air probe on a MultiMode scanning probe microscope (MM-SPM, Bruker) with an advanced NanoScope V Controller (Veeco). At least five different cells from three individual plants were scanned. AFM images were analyzed with SOAX3.6.1 (https://omictools.com/soax-tool) to determine the orientation of cellulosic nanofibers. The cellulosic nanofiber segments detected by SOAX were defined as “snakes” using 1 pixel of snake point space and 20 pixels of minimum snake length. An orientation histogram was calculated based on the snakes after cutting at the junctions. Thousands of snakes were analyzed and shown as a frequency percentage in the histogram (Zhang et al., 2019).

Biomechanics analyses

Breaking force was examined using the mature second internodes from the wild-type, bc16, and the complementary lines. Internode segments with equal length were analyzed for breaking force using a digital dynamometer (NK-200). For the tensile test, 15-cm second internode segments from wild-type and bc16 plants were prepared and fixed for examination using a digital force/length tester (DZ-107). The sample extension length during load force application was recorded. Stress–strain curves were calculated using the obtained data. At least three internodes from three individual plants were analyzed.

For nano-indentation analysis, the fiber cells harvested from wild-type and bc16 internodes were probed under a Nanowizard 3 AFM with PeakForce Quantitative Nanomechanical Mapping (JPK). An RTESPA-525 probe with a triangular pyramid shape (Bruker) was used for indentation examination. The used tips bear a resonance frequency of ∼2,048 Hz and a spring constant of ∼20 N m⁻¹. AFM tip was calibrated by engaging the tip onto a clean glass slide in single ramp mode before each indentation experiment. Deflection sensitivity was automatically recorded and calculated using JPK Data Processing software (version 7.097, Bruker). Average deflection sensitivity was used from at least 400 ramp curves. A summary of the statistical analysis is shown in Supplemental Data Set S3.

Accession numbers

Sequences from this study can be downloaded from the rice genome annotation project (http://rice.uga.edu/) or the Arabidopsis information resource (https://www.arabidopsis.org) with the following accession numbers: BC16, LOC_Os05g05200; BC1, LOC_Os03g30250; FLA9, LOC_Os05g07060; CESA4, LOC_Os01g54620; CESA9, LOC_Os09g25490; HNR, LOC_Os01g71770; HIR1, AT01g69840; AGP4, AT5g10430; and VND6, AT5g62380.

Supplemental data

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

Supplemental Figure S1. Phenotypes of the bc16 mutant.

Supplemental Figure S2. Validation of the genotypes.

Supplemental Figure S3. Characteristics of BC16.

Supplemental Figure S4. Subcellular localization of GFP-BC16.

Supplemental Figure S5. Sequence alignment of BC16 and MBOAT members.

Supplemental Figure S6. Biochemical pathway of BC16.

Supplemental Figure S7. BC16 regulates lipid structure of GPI-APs.

Supplemental Figure S8. BC16 affects BC1 release.

Supplemental Figure S9. Localization of HIR1 with BC1 and CESA4.

Supplemental Figure S10. Expression of Arabidopsis VND6 induces vessel-like secondary cell wall bands in N. benthamiana.

Supplemental Figure S11. Isolation of fiber cells from rice internodes.

Supplemental Table S1. List of primers for BC16 cloning.

Supplemental Table S2. List of primers used in this study.

Supplemental Data Set S1. List of genes in the BC16 coexpression network.

Supplemental Data Set S2. LC-MS/MS examination of GPI lipid profiles.

Supplemental Data Set S3. Summary of statistical analyses.

Supplemental Video S1. Bending test of second internodes from the plants of indicated genotypes, showing brittleness in bc16.

Supplemental File S1. Sequence alignments of BC16 and homologs.

Supplemental File S2. Phylogenetic tree built using BC16 and homologs.

 

Y.Z. and B.Z. designed the experiments. Z.X. performed the map-based cloning, genetic complementation, and field trials with the assistance of J.M. and Y.G. Y.G. and S.W. performed the biochemical and lipidomics analyses with the assistance of Y.W. C.G. performed AFM and Young’s modulus analyses. J.Ma. and H.Y. performed the gene expression and microscopy analyses. S.C. performed the tensile analysis. X.L. performed the gene transformation. F.Z. assisted in the lipidomics analysis. Y.Z., B.Z., Y.G., C.G., Z.X., H.Y., S.C., and Q.L. analyzed the data. Y.Z. and B.Z. wrote the manuscript. All authors read and commented on the article.

The authors 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/plcell) are: Baocai Zhang (bczhang@genetics.ac.cn) and Yihua Zhou (yhzhou@genetics.ac.cn).