Brachypodium distachyon UNICULME4 and LAXATUM-A are redundantly required for development

Abstract

In cultivated grasses, tillering, leaf, and inflorescence architecture, as well as abscission ability, are major agronomical traits. In barley (Hordeum vulgare), maize (Zea mays), rice (Oryza sativa), and brachypodium (Brachypodium distachyon), NOOT-BOP-COCH-LIKE (NBCL) genes are essential regulators of vegetative and reproductive development. Grass species usually possess 2–4 NBCL copies and until now a single study in O. sativa showed that the disruption of all NBCL genes strongly altered O. sativa leaf development. To improve our understanding of the role of NBCL genes in grasses, we extended the study of the two NBCL paralogs BdUNICULME4 (CUL4) and BdLAXATUM-A (LAXA) in the nondomesticated grass B. distachyon. For this, we applied reversed genetics and generated original B. distachyon single and double nbcl mutants by clustered regularly interspaced short palindromic repeats – CRISPR associated protein 9 (CRISPR-Cas9) approaches and genetic crossing between nbcl targeting induced local lesions in genomes (TILLING) mutants. Through the study of original single laxa CRISPR–Cas9 null alleles, we validated functions previously proposed for LAXA in tillering, leaf patterning, inflorescence, and flower development and also unveiled roles for these genes in seed yield. Furthermore, the characterization of cul4laxa double mutants revealed essential functions for nbcl genes in B. distachyon development, especially in the regulation of tillering, stem cell elongation and secondary cell wall composition as well as for the transition toward the reproductive phase. Our results also highlight recurrent antagonist interactions between NBCLs occurring in multiple aspects of B. distachyon development.

UNICULME4 and LAXATUM-A are involved in tillering, tiller elongation, stem cell wall formation, leaf patterning, inflorescence and flower development, and grain yield.

Introduction

Plant organogenesis relies on pools of dividing pluripotent cells called the plant stem cell niches that reside in the meristems (Žádníková and Simon, 2014). The shoot apical meristem is organized into a central zone, composed of slowly dividing stem cells, a peripheral zone where lateral organs initiate, and a rib zone that provides cells for internodes (Aichinger et al., 2012). The initiation of organs from the peripheral zone requires the creation of meristem-to-organ boundaries (hereafter boundaries) that separate two groups of cells with very distinct gene expression programs and morphologies, repress cell proliferation, and promote initiation and differentiation of adjacent lateral organs (Aida and Tasaka, 2006a, 2006b; Barton, 2010; Žádníková and Simon, 2014; Hepworth and Pautot, 2015; Wang et al., 2016). Among the molecular actors participating in boundary regulation, the members of the NOOT-BOP-COCH-LIKE (NBCL; Couzigou et al., 2012) clade encode highly conserved and key developmental transcriptional co-factors containing BROAD COMPLEX, TRAM TRACK, and BRICK A BRACK/POXVIRUSES and ZINC FINGER (BTB/POZ) and ANKYRIN domain repeats.

In dicots, through their roles in boundary regulation, NBCL genes are associated with multiple aspects of developmental processes, such as differentiation and patterning of stipules and leaves (Yaxley et al., 2001; Ha et al., 2003, 2004, 2007; McKim et al., 2008; Couzigou et al., 2012), floral meristem identity acquisition, internode elongation, and flower patterning and identity (Yaxley et al., 2001; Ha et al., 2003 2004; Norberg et al., 2005; Ha et al., 2007; Xu et al., 2010; Couzigou et al., 2012; Khan et al., 2012a, 2015). NBCLs act as positive regulators of the lignin production (Khan et al., 2015; Shen et al., 2021; Zhang et al., 2019). They also play a role in the establishment and functioning of abscission zones (Hepworth et al., 2005; Norberg et al., 2005; McKim et al., 2008; Ietswaart et al., 2012; Couzigou et al., 2016; Frankowski et al., 2015; Du et al., 2021), as well as in inflorescence architecture and fruit patterning (Hepworth et al., 2005; Khan et al., 2012a; Xu et al., 2016). In addition, NBCL genes were recently shown to be involved in photo/thermo-morphogenesis and root development (Woerlen et al., 2017; Zhang et al., 2017) and may help to restrict fungal susceptibility of the Arabidopsis (Arabidopsis thaliana) rosette core (Dai et al., 2019). The role of these genes in multiple aspects of development and biotic interactions make them interesting targets for crop improvement. Overexpression of A. thaliana BLADE-ON-PETIOLE (BOP) genes shows an increased branching phenotype, producing extra paraclades in leaf nodes (Ha et al., 2007). In A. thaliana, BOP1/2 function downstream of BREVIPEDICELLUS and PENNYWISE in the stem and have a reciprocal function associated with lignin biosynthesis (Khan et al., 2012a).

In monocotyledons, NBCL genes are also conserved, but their roles have been less studied despite their potential role in crop breeding. Two NBCL genes HvUniculme4 (HvCul4) and its paralog HvLaxatum-a (HvLax-a) are present in barley (Hordeum vulgare). The HvCul4 gene acts at leaf axil and leaf boundary regions to control axillary bud outgrowth and tillering as well as the development of the ligule and the HvLax-a gene controls internode length, floral organ identity, and rachis development (Tavakol et al., 2015; Jost et al., 2016). In contrast to H. vulgare, modern maize (Zea mays ssp. mays) possesses four NBCL paralogs: ZmTASSELS REPLACE UPPER EARS1 (ZmTRU1) and ZmTASSELS REPLACE UPPER EARS1-LIKE1 (ZmTRL1) are homologous to HvCul4 and ZmTRU2 and ZmTRL2 are homologous to HvLax-a (Dong et al., 2017). Zea mays, which was domesticated from teosinte (Z. mays ssp. parviglumis) is characterized by the suppression of axillary branching through an increased apical dominance. Branches suppression in Z. mays was achieved through the selection of a gain of function allele of the ZmTEOSINTE BRANCHED1 (ZmTB1) transcription factor. The loss of function of ZmTB1 overproduces tillers and presents long aerial branches tipped by male tassels that replace the normally female ears (Doebley et al., 1995). ZmTB1 functions as a repressor of both axillary bud growth and inflorescence sexual fate (Hubbard et al., 2002) and encodes a class II TB1, CYCLOIDEA, PCF1 transcription factor that is orthologous to A. thaliana BRANCHED1 (Doebley et al., 1997; Aguilar-Martínez et al., 2007). The ZmTRU1 NBCL gene is directly activated by ZmTB1 to mediate axillary branch suppression (Dong et al., 2017). In rice (Oryza sativa), there are three NBCL genes. OsBOP1 is homologous to HvCul4 and OsBOP2/3 are homologous to HvLax-a. They act as regulators of leaf proximal–distal patterning through the activation of proximal sheath differentiation and suppression of distal blade differentiation. They control the temporal changes in the sheath–blade ratio of O. sativa leaves and are also essential for ligule and auricle differentiation (Toriba et al., 2019). The mutant phenotypes of the Hvcul4, Zmtru1, and Osbops suggest that the roles of NBCL genes might be different in these plants for tillering and for the development of ligule and auricles at the sheath–blade boundary region (Tavakol et al., 2015; Jost et al., 2016; Dong et al., 2017; Toriba et al., 2019). In addition, the interaction(s) between these genes was not yet studied in grasses.

The nondomesticated grass model Brachypodium (Brachypodium distachyon) is phylogenetically close to wheat and H. vulgare and is prone to genetic transformation (Draper et al., 2001; Vogel et al., 2006; Opanowicz et al., 2008; Brkljacic et al., 2011; Scholthof et al., 2018). The B. distachyon NBCL paralogs, BdUNICULME4 (CUL4) and BdLAXATUM-A (LAXA) genes are orthologous to HvCul4 and HvLax-a, respectively, and belong to two distinct monocot-specific NBCL clades (Magne et al., 2020). A previous study of Targeting Induced Local Lesions IN Genomes (TILLING) nbcl mutants in B. distachyon, showed that the CUL4 and LAXA genes were individually participating in plant development. The loss-of-function of CUL4 caused a reduced tillering, ligule and auricles developmental defects, and the loss of spikelet determinacy, while missense LAXA mutations caused an increased tillering, modified spikelet architecture, and impaired floral organization and identity (Magne et al., 2020). Grass NBCL genes are agronomically relevant since they contribute to tillering and spike architecture, two essential traits for cereal crops.

In this work, we first generated null alleles for the LAXA locus by CRISPR–Cas9 genome edition (laxa^CR). laxa^CR mutants showed similar but stronger phenotypes than the laxa missense TILLING mutants (laxa^TI) previously characterized in Magne et al. (2020). In addition, we describe original functions for this gene in leaf positioning, and seed and root development. Next, in order to study the potential functional redundancy that might exist between the B. distachyon NBCLs, we generated B. distachyon cul4laxa double mutants by crossing CUL4^TI and LAXA^TI TILLING mutants (cul4^TIlaxa^TI) and by directly applying CRISPR–Cas9 on the LAXA locus in the CUL4^TI background (cul4^TIlaxa^CR). As a result we showed that CUL4 and LAXA were redundantly required for B. distachyon development. We found that the cul4^TIlaxa^CR double mutants were strongly impacted in their development, they were especially dwarf and showed aberrant flowers leading to complete sterility, making these transgenics not suitable for characterization. However, through the use of weaker cul4^TIlaxa^TI double mutants, we were able to show that, in B. distachyon, NBCL genes positively contribute to tillering, to internode cell elongation, as well as to the production of cell wall polysaccharide and lignin. In this study, we also unveiled original and antagonistic interactions between CUL4 and LAXA.

Results

Generation of laxa CRISPR–Cas9 null alleles and of cul4laxa double mutants

In this study, in order to help distinguish between the BdUNICULME4 (CUL4, Bradi2g60710) and BdLAXATUM-A (LAXA, Bradi4g43150) TILLING and CRISPR–Cas9 mutant alleles, we named the TILLING mutants alleles cul4^Q127* or cul4^W203* and laxa^T381I or laxa^L365F from Magne et al. (2020) as cul4^TI and laxa^TI, respectively. Furthermore, in this work, we generated null alleles for the LAXA locus using CRISPR–Cas9 editing approaches in both wild-type (WT) and cul4^TI backgrounds. In B. distachyon, the use of this technology resulted in the recovery of several mutant alleles in the two backgrounds. In these backgrounds, respectively, the different laxa CRISPR–Cas9 mutant alleles showed identical phenotypes and we thus named these mutations laxa^CR through the rest of this study independently of the laxa CRISPR–Cas9 null allele used (Figure 1A;Supplemental Figure S1).

Figure 1.

Description of the cul4 and laxa mutant alleles and aerial architecture of the TILLING and CRISPR–Cas9 nbcl single and double mutants. A, Schematic representation of the CUL4 and LAXA proteins. Unhatched and hatched blocks represent BTB/POZ domains and ankyrin repeat domains, respectively. cul4 and laxa mutations are indicated above and below the protein scheme, respectively. The TILLING mutant alleles cul4^Q127* (Bd4982), cul4^W203* (Bd7965), laxa^L365F (Bd5998), and laxa^T381I (Bd3615) are indicated by asterisks (KO mutations) or by black triangles (missense mutations). The CRISPR-Cas9 mutant alleles laxa^CR are indicated by asterisks (KO mutations). B–G, Aerial phenotype of WT B. distachyon (WT, B), cul4^TI TILLING KO mutant (C), laxa^TI TILLING missense mutant (D), laxa^CR CRISPR–Cas9 KO mutant (E), cul4^TIlaxa^TI (Bd7965 x Bd3615) double mutant (F), and cul4^TIlaxa^CR (Bd4982, laxa^CR) double mutant (G). Scale bars B–G: 1 cm. H, Primary and I, secondary tillers quantification in the different genotypes (n: 40 except for cul4^TIlaxa^CR with n: 10). J, Height of fully mature dry plants. For WT, cul4^TI, laxa^CR, cul4^TIlaxa ^TI, cul4^TIlaxa^CR, n: 20, 18, 22, 15, 20, and 10 plants, respectively. For the box plot, the median (middle horizontal line), upper and lower quartiles (boxes), as well as minimum and maximum values (whiskers) are indicated. Individual values are indicated by dots. Asterisks indicate significant difference compared to WT (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; Mann–Whitney U test).

In B. distachyon WT background, we targeted the LAXA locus by CRISPR–Cas9 using two different RNA guides and obtained two independent mutant alleles for LAXA (laxa^CR;Figure 1A;Supplemental Figure S1). The first allele consisted a single base pair (bp) deletion in the first exon of LAXA resulting in a predicted truncated protein of 99 aa long and the second allele consisted a 4-bp deletion in the second exon of LAXA resulting in a predicted truncated protein of 387 aa long instead of the 503 aa of the full-length WT protein (Supplemental Figure S1). Gene expression analysis in different organs from WT and laxa^CR mutants showed that the expression of LAXA was significantly downregulated in laxa^CR mutant organs as expected for a mutant gene with a premature stop codon (Supplemental Figure S1). The mutant progeny of these two independent laxa^CR alleles showed similar but stronger phenotypes than the laxa^TI missense mutants previously described in Magne et al. (2020) (see detailed phenotyping of laxa mutants in the sections below).

Next, in the B. distachyon cul4^TI (cul4^Q127*) background, by using the same strategy, we obtained five different mutations for the LAXA locus consisting of 1-bp addition or 1-, 2-, 3-, or 4-bp deletions (Figure 1A;Supplemental Figure S1). The mutant alleles with 3-bp deletion in the LAXA gene showed no additional phenotype as compared to the cul4^TI parental line, indicating that this in-frame deletion has no detectable effect on the LAXA gene function. In contrast, the edited plants containing either 1-bp addition or 1-, 2-, and 4-bp deletions in the first exon of the LAXA gene generated premature stop codons resulting in predicted truncated proteins (98, 99, 149, and 150 aa truncated proteins; Supplemental Figure S1). These four independent homozygous cul4^TIlaxa^CR mutant lines were extremely dwarf and completely sterile. These additional phenotypes were not present in the cul4^TI parental line (see detailed phenotyping of cul4^TIlaxa^CR in the sections below).

Unfortunately, because of the dwarf phenotype and this complete sterility, the cul4^TIlaxa^CR transgenic lines were not suitable for further studies. Therefore, we generated weaker double mutant alleles by crossing cul4^TI (cul4^W203*) null allele and laxa^TI (laxa^T381I) missense weak allele (Figure 1A). The corresponding cul4^TIlaxa^TI double mutants were less affected than the cul4^TIlaxa^CR double mutants, produced seeds and enable to pursue the study. This confirms that laxa^T381I is a weak mutant allele (missense) and not a null (KO) allele. In this study, most subsequent analyses were therefore performed using cul4^TIlaxa^TI.

CUL4 and LAXA are redundantly required for tillering and tiller elongation

As previously described in Magne et al. (2020), here we show and confirm that cul4^TI and laxa^TI were, respectively, negatively and positively impacted for tillering compared to WT (Figure 1). In addition, we show that laxa^CR also presents an increased number of tillers similarly as laxa^TI, with laxa^TI and laxa^CR having twice more secondary or tertiary tillers than the WT with a substantial increase in total tillers for the laxa^CR mutant (Figure 1; Supplemental Figure S2) indicating an increased severity of the KO mutation. In both cul4^TIlaxa^TI and cul4^TIlaxa^CRdouble mutants the tillering was drastically reduced compared to WT plants. This tillering defect was stronger when compared with the single cul4 mutant and accentuated in cul4^TIlaxa^CR compared to cul4^TIlaxa^TI (Figure 1).

Besides their tillering modifications, both the cul4^TIlaxa^TI and cul4^TIlaxa^CR double mutants were dwarf while no significant difference in plant height was recorded between cul4^TI, laxa^TI, and laxa^CR relative to WT (Figure 1). In addition, while cul4^TI showed no significant difference relative to WT, laxa^TI, and laxa^CR had shorter roots compared to WT (Supplemental Figure S3). As the LAXA gene is apparently not expressed in roots (Winter et al., 2007; Sibout et al., 2017; Magne et al., 2020; http://bar.utoronto.ca/efp_brachypodium/cgi-bin/efpWeb.cgi), it suggests that in B. distachyon, LAXA could play an indirect and positive effect on root development, maybe through resource allocation between the roots and the aerial parts. In addition, as expected for a dwarf line, the cul4^TIlaxa^TIroot length was also reduced by ∼40% compared to the WT (Supplemental Figure S3).

To determine when the growth of the double mutants starts to be affected, we monitored the development of cul4^TIlaxa^TI and cul4^TIlaxa^CR from seed germination until fully mature plants (Supplemental Figure S2). Two weeks after germination, cul4^TIlaxa^TI had a WT stature but started to phenotypically diverge right after, when internodes began to elongate. In contrast, 2 weeks after germination, cul4^TIlaxa^CR already exhibits serious developmental defects, which are clearly visible from one leaf stage until the plant dies. These observations show that the impact of the mutation in cul4^TIlaxa^CR are more severe than in cul4^TIlaxa^TI and indicate that the penetrance of the laxa^CR mutation is more detrimental than the one of laxa^TI.

We next investigated the cause of this dwarf phenotype using the cul4^TIlaxa^TI double mutant. In these cul4^TIlaxa^TI double mutants the internode lengths were visibly reduced in size (Figure 2A). To quantify this, we measured the length of the first three internodes, counted from the apex, in WT, cul4^TI, laxa^TI, and cul4^TIlaxa^TI plants. Our results showed that in cul4^TIlaxa^TI, the internode length was tremendously reduced relative to the WT and single mutants indicating that the dwarf phenotype of the double mutants is due to a defect in internode elongation (Figure 2, A and F). We next histologically examined longitudinal sections from the first internodes and nodes from WT, cul4^TI, laxa^CR, and cul4^TIlaxa^TI plants (Figure 2, B–E). Our histological analysis together with cell length measurements revealed that the length of the internode and node cells was significantly reduced in cul4^TIlaxa^TI compared to WT while not affected in cul4^TI and laxa^CR single mutants (Figure 2, B–E, G, and H). These results show that CUL4 and LAXA redundantly and positively contribute to internode and node cell elongation in the stem.

Figure 2.

Internode and node cell elongation in the singles and the double nbcl mutants. A, Stem elongation phenotype of WT, cul4^TI, laxa^TI, and cul4^TIlaxa^TI. B–E, Longitudinal sections of stem stained with toluidine blue showing the first internode (top) and the first node (bottom) of flowering stems from WT (B), cul4^TI (C), laxa^CR (D), and cul4^TIlaxa^TI (E). F, Length of first internode (I1, blue bars), secondary internode (I2, red bars) and third internode (I3, yellow bars) from fully elongated stem of WT, cul4^TI, laxa^TI and cul4^TIlaxa^TI. Error bars represent standard deviation. Thirty stems of each line were measured for each experiment. Blue, red, and yellow asterisks indicate significant differences for internodes 1, 2, and 3, respectively. G, First internode and H, first node cell length measurement from WT, cul4^TI, laxa^CR, and cul4^TIlaxa^TI flowering stems. A total of 200 cells were measured for each replicate. For the box plot, the median (middle horizontal line), upper, and lower quartiles (boxes), as well as minimum and maximum values (whiskers) are indicated. Individual values are indicated by dots. F–H, Asterisks indicate significant difference relative to WT (***P < 0.001; ****P < 0.0001; Mann–Whitney U-test). Scale bars: A, 1 cm; B–E, 100 µm.

CUL4 and LAXA positively contribute to cell wall formation in stem and to the regulation of cellulose and lignin-related gene expression

In the dicots, A. thaliana, Gossypium hirsutum, and Parasponia andersonii, NBCLs have been reported as promoting the lignin biosynthesis (Khan et al., 2015; Zhang et al., 2019; Shen et al., 2021). Lignin is an important component of secondary cell walls providing rigidity and hydrophobicity to the cell wall, and reduced lignin production can result in dwarf phenotypes in different plants (Ha et al., 2021). Thus, to further investigate if grass NBCLs are also involved in the regulation of the lignin biosynthesis pathway, we performed phloroglucinol–HCl staining on stem cross-sections from the WT and the cul4^TIlaxa^TI double mutant. Phloroglucinol–HCl staining colors lignified cell walls in red while nonlignified cell walls remain colorless. We observed that the interfascicular fibers of cul4^TIlaxa^TI double mutant were yellow or white relative to the red ones of the WT plants. In contrast, the vascular bundles of cul4^TIlaxa^TI were stained in red suggesting that vascular bundle tissues were not or less impacted than interfascicular fibers (Figure 3, A–D). These results indicate that, in cul4^TIlaxa^TI, a very low amount of lignin is deposited onto the interfascicular fiber cell wall. Subsequent quantification of the acetyl bromide soluble lignin content in cul4^TIlaxa^TI supported our findings and showed a significant decrease of 32% compared to the WT (Figure 3E). Not unexpectedly, thioacidolysis analyses revealed a significant reduction of uncondensed guaiacyl (G), syringyl (S), and p-hydroxy-phenyl (H) lignin units content in cul4^TIlaxa^TI compared to the WT (Figure 3F).

Figure 3.

Internode cell wall composition, lignin, and monosaccharide contents. A–D, Representative pictures of the lignin deposition (phloroglucinol–HCl staining, red coloration) in the first internodes of fully senesced WT and cul4^TIlaxa^TI mutants. A and B, Whole stem and (C and D), stem magnification focusing on fascicular and interfascicular bundles of B. distachyon from WT (A and C) and cul4^TIlaxa^TI (B and D). In cul4^TIlaxa^TI, the lignin staining is less intense compared to WT, especially at the interfascicular regions. Scale bars: 100 μm. E, Quantification of acetyl bromide soluble lignin content. F, Quantification of b-O-4 linked H (circles), G (triangles), and S (crosses) lignin units by thioacidolysis. G, Main cell wall sugar contents in dried B. distachyon stem. ara, arabinoxylan; xyl, xylose and glc, glucose. For each line, pulverized stem tissue from 6 to 16 individuals from three independent replicates was analyzed. E–G, Asterisks indicate significant differences relative to WT (*P <0.05; ***P < 0.001; ****P < 0.0001; Mann–Whitney U-test), n = 3 and error bars represent standard deviation. %DM stands for percentage of dry mass. H, Relative transcript abundance of the lignin biosynthesis-related genes BdCAD1 and BdCOMT6, and of the cell wall-related genes BdCESA4, BdCESA7, BdCESA8, and BdSWAM1 in WT (white bars) and cul4^TIlaxa^TI (purple bars). Gene expression analysis was performed by RT-qPCR on stem samples. Gene expression was normalized against the constitutively expressed BdUBQ4 and BdUBC18 reference genes. Results represent means ± sem (standard error of mean) of three biological repeats and three technical replicates. Primers used for RT-qPCR analysis are given in Supplemental Table S2. For the box plot, the median (middle horizontal line), upper and lower quartiles (boxes), as well as minimum and maximum values (whiskers) are indicated. Individual values are indicated by dots. Asterisks indicate significant differences relative to WT (*P < 0.05; **P < 0.01; one-way annalysis of variance (ANOVA)).

This striking change in lignin content led us to investigate the content of other cell wall compounds in the double mutant lines. Similarly, the sugar composition, including glucose (mainly extracted from cellulose) was significantly affected in the double mutants with a reduction of the xylose and glucose contents (Figure 3G). In addition, our study suggested that the content in arabinoxylans, the main hemicellulose in B. distachyon stems, was also impacted but in a positive manner.

Together, our histological and chemical approaches highlighted that cell wall composition is overall impacted in the cul4^TIlaxa^TI background. Our data suggest that, in B. distachyon, CUL4 and LAXA positively contribute to cell wall composition. These changes in the cell wall composition may participate in the overall aerial development defect and in the reduction in the above-ground biomass (Supplemental Figure S4).

To better understand the role of the NBCL genes in grass cell wall biosynthesis, we analyzed the expression of cell wall formation-marker genes in WT and nbcl mutant backgrounds. The B. distachyon stems are particularly enriched in BdCINNAMYL ALCOHOL DEHYDROGENASE 1 (BdCAD1) and BdCAFFEIC ACID O-METHYLTRANSFERASE 6 (BdCOMT6) transcripts which encode two key enzymes of the lignin monomer pathway (Bouvier d’Yvoire et al., 2013; Dalmais et al., 2013; Ho-Yue-Kuang et al., 2016), as well as in BdCELLULOSE SYNTHASE A (BdCESAs) transcripts which are involved in cell wall synthesis (Handakumbura et al., 2013; Sibout et al., 2017). Also, BdSECONDARY WALL-ASSOCIATED MYB1 (BdSWAM1) acts as a positive regulator of cell wall biosynthesis (Handakumbura et al., 2018). Our results showed that the expression level of BdCAD1, BdCOMT6, BdCESA4, BdCESA7, and BdCESA8 was significantly downregulated in cul4^TIlaxa^TI compared to WT, while the expression of BdSWAM1 was significantly upregulated (Figure 3H;Supplemental Figure S5).

The transcriptional changes observed for these cell wall-related marker genes are in line with the phenotypic changes observed in cell wall composition. These important changes in the cell wall composition may participate in the extreme dwarf phenotype of the double nbcl mutants, however, while a disturbed secondary cell wall composition can affect the general growth of plants (Ha et al., 2021), here we cannot fully disentangle if these changes in secondary cell wall are the cause or the consequence of the observed dwarf phenotype.

CUL4 and LAXA antagonistically contribute to the patterning of the leaf laminar junction and modulate leaf angle

In grasses, the leaf is composed of a basal sheath, of a sheath–blade junction (also called “laminar junction”) and of a distal blade (Becraft et al., 1990; Sylvester et al., 1990). The sheath and leaf blade are separated and articulated by a laminar junction containing the ligule and two wedge-like structures called auricles acting as a hinge allowing the leaf blade to project at an angle from the vertical stem.

We first analyzed the gene expression level of CUL4 and LAXA genes in sheath, laminar junction, and blade tissues by RT-qPCR and showed that CUL4 and LAXA were highly expressed in the sheath and the laminar junction, while a very low expression was detected in the blade, suggesting that these genes mainly mediate their functions in those tissues (Supplemental Figure S6A). The low level of expression detected here is in agreement with our previous study (Magne et al., 2020) and with public data (http://bar.utoronto.ca/efp_brachypodium/cgi-bin/efpWeb.cgi; Winter et al., 2007; Sibout et al., 2017) showing that the two genes are not expressed in the leaf blade at different stages of their development.

Then, in agreement with previous findings from Magne et al. (2020), we confirmed that cul4^TI mutants did not develop ligule and auricles (Figure 4, A, E, I, B, F, and J), and showed that, as described for laxa^TI mutants, laxa^CR null alleles also do form ligule and auricles (Figure 4, A, E, I, C, G, and K).

Figure 4.

Developmental characterization of the laminar junction anatomy and determination of leaf angle. Representative pictures of the laminar junction in 40-d-old WT (A, E, and I); cul4^TI (B, F, and J), laxa^CR (C, G, and K), and cul4^TIlaxa^TI (D, H, and L). The first, second, and third rows represent side, adaxial, and abaxial views, respectively. WT laminar junction presents a ligule and two auricles (A, E, and I). cul4^TI lacks ligule and auricles. Instead of the ligule, there is a fringe of tissues (B, F, and J). laxa^CR looks like WT and forms ligule and auricles (C, G, and K). cul4^TIlaxa^TI lacks ligule but does form auricles (D, H, and L). In E, F, G, and H, the culms have been manually removed in order to display the ligule. Missing auricles and ligules are indicated by white arrowheads and white asterisks, respectively. M–P, Representative pictures of the leaf angle in WT (M), cul4 (N), laxa^CR (O), and cul4^TIlaxa^TI (P). Scale bars: A–P, 0.5 cm. Q, Determination of the flag leaf angle in the different genotypes. n: 40 plants for each genotype except n: 15 for cul4^TIlaxa^TI. For the box plot, the median (middle horizontal line), upper and lower quartiles (boxes), as well as minimum and maximum values (whiskers) are indicated. Individual values are indicated by dots. Asterisks indicate significant difference compared to the WT (**P < 0.01; ***P < 0.001; Mann–Whitney U-test). ND, not determined.

We next wanted to investigate the consequences of the loss-of-function of both CUL4 and LAXA genes in leaf development and especially at the laminar junction. In cul4^TIlaxa^TI, the ligule does not develop, however, the auricles developed normally (Figure 4, A, E, I, D, H, and L). This suggests that the formation of auricles can take place independently of these genes and that the development of the auricles is repressed by LAXA when CUL4 is absent. We hypothesized a possible model of action, in which CUL4 may antagonize LAXA function to promote the formation of auricles (Supplemental Figure S6B).

The leaf angle is an important parameter of grass architecture and is determined by the anatomy of the laminar junction (Kong et al., 2017; Zhou et al., 2017). Since the loss-of-function of CUL4 and LAXA alters the development of the laminar junction, we examined the flag leaf angle between leaf blade and the internode below the spike in WT, cul4^TI, laxa^CR, and cul4^TIlaxa^TI. In cul4^TI, the flag leaf angle was significantly reduced while significantly increased in laxa^CR (Figure 4, M–O and Q). In cul4^TIlaxa^TI, the entire leaf twisted at almost 180° and tended to split from the stem, positioning the entire leaf (including the sheath) nearly horizontally. As a consequence, leaf angle could not be measured in cul4^TIlaxa^TI (Figure 4, P and Q).

These results indicate that CUL4 and LAXA antagonistically contribute to leaf angle determination in B. distachyon.

LAXA and CUL4 contribute to spikelet architecture, flower patterning, and to floral identity gene expression

CUL4 and LAXA were previously shown to be involved in B. distachyon spikelet architecture and determinacy (Magne et al., 2020). In the present study, our findings support previous observations and confirmed that cul4^TI and laxa^TI mutants present an exaggerated spikelet length and a short-spikelet phenotype, respectively (Figure 5, A–C, G–I and S). This results in an increased and a reduced number of florets per plant, in cul4^TI and laxa^TI, respectively (Figure 5T). In addition, here we showed that laxa^CR consistently presents a short spikelets phenotype which also results in a reduced number of florets per plant (Figure 5, D, J, S, and T). Furthermore, in laxa^CR the development of spikelet was even more affected compared to the laxa^TI allele since laxa^CR presented an increased number of sterile florets per plant (Figure 5U). Thus, here, by using a true laxa^CR null allele, we confirmed the function of LAXA in spikelet development and architecture.

Figure 5.

Spike, spikelet, and seed phenotypes of the TILLING and CRISPR–Cas9 nbcl single and double mutants. A–F, 49-d-old entire spikes, G–L, 60-d-old spikelets, and M–R, dry seeds from WT (A, G, and M), cul4^TI (B, H, and N), laxa^TI (C, I, and O), laxa^CR (D, J, and P), cul4^TIlaxa^TI (E, K, and Q), and cul4^TIlaxa^CR (F, L, and R). B and H, The cul4^TI spikelets are exaggeratedly long. C, I, D, and J, The laxa^TI and laxa^CR spikelets are compact and short. E and K, cul4^TIlaxa^TI spikelets are exaggeratedly long and looked-like the cul4^TI spikelets. F and L, cul4^TIlaxa^CR produces leaf-like organs instead of spikelets (white asterisks). WT and supplementary florets are indicated by white and yellow arrowheads, respectively. S, Spikelet length in the different nbcl mutant backgrounds. For WT, cul4^TI, laxa^TI, laxa^CR, and cul4^TIlaxa^TI spikelets, n: 200, 150, 180, 190, and 120, respectively. T, Number of florets per plant in the different nbcl mutant backgrounds. For WT, cul4^TI, laxa^TI, laxa^CR, and cul4^TIlaxa^TI, n: 17 plants. U, Number of sterile florets per plant in the different nbcl mutant backgrounds. For WT, cul4^TI, laxa^TI, laxa^CR, and cul4^TIlaxa^TI, n: 17 plants. V, Number of seeds per plant in the different nbcl mutant backgrounds. For WT, cul4^TI, laxa^TI, laxa^CR, and cul4^TIlaxa^TI, n: 18 plants. W, Seed weight per plant in the different nbcl mutant backgrounds. For WT, cul4^TI, laxa^TI, laxa^CR, and cul4^TIlaxa^TI, n: 36, 31, 29, 36, and 31 plants, respectively. X, Seeds weight for 1.000 seeds in the different nbcl mutant backgrounds. n: 1.000 seeds per genotype. S–X, for the box plot, the median (middle horizontal line), upper and lower quartiles (boxes), as well as minimum and maximum values (whiskers) are indicated. Individual values are indicated by dots. Asterisks indicate significant differences relative to WT (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; Mann–Whitney U-test). Scale bars: A–F, 1 cm; G–L, 0.5 cm; M–Q, left, 0.5 cm; right, 1 cm.

In cul4^TIlaxa^TI, the spikelets length and the number of florets per plant were significantly increased compared to WT and the cul4^TIlaxa^TI spikelets were reminiscent of the cul4^TI spikelets (Figure 5, E, K, S, and T). In cul4^TIlaxa^TI, most florets were sterile and consequently empty (Figure 5U). The altered fertility of cul4^TIlaxa^TI was associated with a drastic reduction of the seed yield per plant (Figure 5, V and W).

Despite the strong developmental defect of the cul4^TIlaxa^CR double mutants and the difficulty to work with, we succeeded in maintaining a few transgenics until the flowering phase. cul4^TIlaxa^CR produced extremely short and aberrant inflorescences, looking-like vegetative spikes, in which the entire spikelets were replaced by a single leaf-like organ (Figure 5, F and L). Sporadically, we could observe rare flower formation in cul4^TIlaxa^CR (Supplemental Figure S7). These results indicate that CUL4 and LAXA are required for spikelet development and suggest that CUL4 and LAXA redundantly promote spikelet meristem fate acquisition. The phenotypes of the double mutants showed additive defects compared to the single mutants and highlighted that in B. distachyon, CUL4 and LAXA contribute to both spikelet development and to the reproductive transition.

Brachypodium distachyon flowers are composed of a feathery stigma, an ovary, two abaxial lodicules, and two adaxial stamens (Supplemental Figure S7). In Magne et al. (2020), we reported that cul4^TI was not altered in floral patterning while laxa^TI displayed additional stamens and partial lodicule-to-stamen identity switches called stamenoid lodicules. Consistently, here we found that laxa^CR shared similar floral organization defects with laxa^TI (Magne et al., 2020) consisting of adaxial additional stamens and abaxial stamenoid lodicules (Supplemental Figure S7). These results definitely support that in B. distachyon LAXA controls floral organ identity and patterning.

Surprisingly, cul4^TIlaxa^TI flowers looked like WT and did not present any laxa-associated flower modifications (Supplemental Figure S7). Consistently, in the rare cul4^TIlaxa^CR flowers that develop, we did not observe any particular floral modification (Supplemental Figure S7). These results suggest that LAXA is suppressing ectopic stamen formation but that CUL4 antagonizes LAXA action.

To better understand how CUL4 and LAXA act during flower development, we quantified the expression of key regulators of the floral development in WT, cul4^TI, laxa^CR, and cul4^TIlaxa^TI spikelets. We monitored the gene expression of BdAPETALA1 (BdAP1), a key regulator of the floral meristem identity (Xu et al., 2010), of BdINDETERMINATE SPIKELET1 (BdIDS1), and BdSISTER OF INDETERMINATE SPIKELET1 (BdSID1), two BdAPETALA2-LIKE floral identity repressors (Chuck et al., 2008), and of BdKNOTTED1-LIKE-FROM-ARABIDOPSIS-THALIANA6 (BdKNAT6), a class I KNOX transcription factor important for inflorescence formation which is also a downstream target of AtBOP in A. thaliana (Khan et al., 2012b). Our gene expression analysis showed that the expression of BdAP1 and BdKNAT6 was significantly downregulated in all mutant backgrounds compared to the WT and that the expression of BdIDS1 and BdSID1 was significantly reduced in the laxa^CR and cul4^TIlaxa^TI backgrounds (Supplemental Figure S8).

These results suggest that CUL4 and LAXA positively contribute to the correct expression of BdAP1 and BdKNAT6 and that LAXA contributes to the expression of the AP2/ERF, BdIDS1, and BdSID1. In B. distachyon, the NBCL genes are thus required for the regulation of floral identity-related genes.

CUL4 and LAXA control seed size but are not required for seed shattering

In dicots, the abscission of above-ground organs is an NBCL-dependent process (McKim et al., 2008; Wu et al., 2012; Couzigou et al., 2016). However, it is unknown if the NBCL-dependent abscission applies in grass species. In B. distachyon, the abscission ability was not compromised in cul4^TI KO mutant and laxa^TI missense mutants, independently (Magne et al., 2020). At this time an eventual gene redundancy between CUL4 and LAXA was not tested because of the lack of the genetic material.

Here we thus tested the abscission ability of laxa^CR and cul4^TIlaxa^TI. As observed in WT plants, seeds detached easily from the spike at maturity in the two mutant lines. In both laxa^CR and cul4^TIlaxa^TI, at the seed abscission point, we observed sharp scars without any sign of mis-formation (Supplemental Figure S9). Subsequent histological analysis of the seed abscission zones of the WT, laxa^CR, and cul4^TIlaxa^TI revealed typical layers of small differentiated cells corresponding to the abscission zone (Supplemental Figure S9).

The abscission ability of both laxa^CR and cul4^TIlaxa^TI was thus not compromised suggesting that, in contrast to their dicot homologs, CUL4 and LAXA are not essential for B. distachyon seed abscission zone formation and subsequent shattering.

We next studied seed size in the different nbcl mutant backgrounds. We observed that cul4^TI seeds were longer and wider relative to WT (Figure 5, M and N). In agreement with this observation, the total weight for 1.000 cul4^TI seeds was higher relative to WT (Figure 5X). In contrast, laxa^TI, laxa^CR, and cul4^TIlaxa^TI seeds were shorter and narrower relative to WT (Figure 5, M, O, P, and Q), and the total weight for 1.000 seeds was reduced by 50% compared to WT (Figure 5X). Thus, in the genetic backgrounds containing a laxa mutation, the seed size was reduced while in cul4^TI mutant the seed size was increased. This led us to conclude that in B. distachyon, LAXA and CUL4 have antagonist functions in the control of seed size. However, due to a partial sterility, the seed yield per plant was not increased in the cul4^TI mutant line (Figure 5, U, V, and W). In the laxa^TI and laxa^CR mutants, the reduced number of seeds per spike and the reduced size of these seeds resulted in a lower seed yield per plant (Figure 5, V and W). The seed production per plant was even lower in the cul4^TIlaxa^TI double mutant due to the high proportion of sterile florets produced and to the reduced size of the seeds (Figure 5, U, V, and W).

Discussion

In this study, we made use of the CRISPR–Cas9 technology in the nondomesticated grass model B. distachyon and succeeded in generating null alleles for the LAXA locus. In addition, we succeeded to produce B. distachyon cul4laxa double mutants by two different approaches, first, by knocking out LAXA using CRISPR/Cas9 in the cul4 mutant background, and second, via genetic crossing between cul4 and laxa TILLING alleles. The studies of this genetic material complement our previous study (Magne et al., 2020) and show that the NBCL genes CUL4 and LAXA are essential for multiple aspects of B. distachyon development. Interestingly, several of these NBCL-dependent traits including tillering, leaf angle, cell wall composition, and seed yield are of agronomic interest.

As an important finding, we found that the loss-of-function of both CUL4 and LAXA drastically affected the development of B. distachyon and that the corresponding double mutants were dwarf, a phenotype associated with an inhibition of the cell elongation in stems. Our results are in agreement with the dwarf phenotype reported for the Osbop triple mutant (Toriba et al., 2019). Interestingly, contrary to the very strong dwarf phenotype of the double KO mutant (cul4^TIlaxa^CR) that did not allow for a detailed analysis due to sterility, the use of a weaker laxa mutant allele (missense) in the cul4^TIlaxa^TI double mutant also resulted in dwarf plants that were less affected in seed production and which allowed studying different aspect of development.

In A. thaliana and tomato, the double (Atbop1bop2) and triple (Slbop1bop2bop3) mutants do not present such a dwarf phenotype (Norberg et al., 2005; Xu et al., 2016). This suggests that in dicot and monocot, NBCL function and/or regulation might be different.

The cul4laxa mutants were also severely impacted for cell wall composition. In grasses, stem cell wall polymers mainly consist of cellulose, hemicellulose (arabinoxylan), and lignin (Rancour et al., 2012; Bouvier d'Yvoire et al., 2013; Zhang et al., 2014a, 2014b; Penning et al., 2019). The latter is an important component of the cell wall and one of the major factors impacting stiffness in land plants (Rogers and Campbell, 2004). It also contributes to primary plant defenses by providing a physical barrier to pathogens and its composition is highly related to digestibility by animals (Ha et al., 2021). In A. thaliana, G. hirsutum and P. andersonii, previous researches showed a promotive role for NBCL genes on the lignin production (Khan et al., 2012a; Zhang et al., 2019; Shen et al., 2021). In agreement with the literature, here we show that in B. distachyon, the simultaneous inactivation of the two NBCL genes resulted in a drastic decrease of the lignin content characterized by a profound impact on the lignification of the interfascicular fiber cells. We show that all lignin units are impacted irrespective of their chemical constitution (H, G, and S units) in the double mutant. Gene expression analysis of key lignin-related genes revealed that the expression of BdCAD1 and BdCOMT6 were significantly downregulated in cul4^TIlaxa^TI double mutant while not impacted in the single mutant backgrounds. This indicates that CUL4 and LAXA redundantly contribute to the expression of lignin-related gene expression. Besides the altered lignin content, we also demonstrated that the loss-of-function of CUL4 and LAXA altered the expression of CELLULOSE SYNTHASE A genes and led to reduced polysaccharides content in B. distachyon internodes.

Taken together, these results make CUL4 and LAXA two major activators of cell wall biosynthesis in B. distachyon stem and these modifications in cell wall modifications may correspond with the dwarf phenotype of the double mutant (Ha et al., 2021).

Tillering is a key determinant of grass architecture directly contributing to grain yield (Kebrom et al., 2013). In Magne et al. (2020), we previously demonstrated that CUL4 was essential for B. distachyon tillering. Such a function is supported by another study in H. vulgare in which HvCul4 promotes tillering and control axillary shoot outgrowth (Tavakol et al., 2015). In H. vulgare, the loss-of-function of HvLax-a did not impact tillering (Jost et al., 2016). However, in B. distachyon, both the missense laxa TILLING (laxa^TI) mutants and the null laxa CRISPR–Cas9 (laxa^CR) mutants showed a significantly increased tillering (Magne et al., 2020 and this study). Together, these findings confirm the opposite role of the two NBCL paralogs in B. distachyon tillering. Interestingly, cul4^TIlaxa^TI and cul4^TIlaxa^CR were extremely impacted for tillering. In Particular, cul4^TIlaxa^CR was nearly deprived of secondary tillers. The cul4laxa tillers were often bent and distorted, possibly as a result of difficulties in emergence from the leaf sheaths that enclosed them or of the lack of rigidity induced by poor secondary cell wall deposition.

In conclusion, although the molecular mechanisms involving NBCL genes in the regulation of tillering are not yet known, our studies highlighted the essential functions of both CUL4 and LAXA in grass tillering.

In this study, we observed a reduction of the root length associated with laxa mutations. This effect does not seem to be directly associated with the above-ground biomass and/or plant height because the single laxa^TI and laxa^CR mutants that were not significantly different relative to WT also showed this root phenotype. This phenotype is difficult to explain as the LAXA gene expression is not detectable in roots (Winter et al., 2007; Sibout et al., 2017; Magne et al., 2020; http://bar.utoronto.ca/efp_brachypodium/cgi-bin/efpWeb.cgi). We can hypothesize that LAXA positively but indirectly contributes to the root development from the aerial part of the plant where it is expressed and from where it may modify hormonal homeostasis and/or resource allocation between aerial and root organs.

In O. sativa, the three OsBOP genes participate in ligule and auricle formation as these organs are absent in the triple mutant (Toriba et al., 2019). In the H. vulgare cul4 mutant, only ligules are absent (Tavakol et al., 2015). In contrast in the B. distachyon cul4 mutant, ligules and auricles are absent while the laxa missense mutant line ligules and auricles were present (Magne et al., 2020). The construction of the KO mutant confirmed that ligules and auricles are present in laxa mutants; however, here we show that the cul4laxa double mutant do form auricles while having a nonfunctional CUL4 allele. This shows that contrary to the conclusion done using the single cul4 mutant, CUL4 and LAXA are not absolutely required for auricles formation. We therefore hypothesize that during auricles development, CUL4 inhibits the function of LAXA which itself represses the formation of auricles (see model Supplemental Figure S6). Such complex phenotypes suggest the existence of antagonistic interactions between these two NBCL proteins in the control of this organ formation.

The leaf angle is influenced by leaf morphology, especially by the anatomy of the laminar joint. This leaf angle directly influences the structure of the canopy. The Z. mays liguleless (lg) mutants, such as lg1, lg2, lg3, and lg4, or the O. sativa lg1 are deficient in the formation of the ligule and auricles, resulting in a reduced leaf angle (Moreno et al., 1997; Walsh et al., 1998; Mantilla-Perez and Salas Fernandez, 2017). In addition, the diversity of auricle development in a large population of Z. mays inbred lines also correlates with leaf angle (Kong et al., 2017).

In B. distachyon, CUL4 and LAXA also participate in the development of the laminar junction and therefore in leaf angle. We found that CUL4 and LAXA positively and negatively contribute to leaf angle, respectively. Changes in cul4 could result from the modification of the laminar junction but leaf angle increase in the laxa mutant lines is not related to visible modification of the laminar junction. Subtle modifications like cell size changes in this region may affect leaf angle but this was not studied here. In grass crops, erected leaves result in increased grain yield (Sakamoto et al., 2006; Liu et al., 2019; Yu, 2019). In agreement with this finding, in B. distachyon, seed size was increased in cul4 showing a reduced leaf angle and decreased in laxa showing an increased leaf angle. Our results are thus also pointing to an association between small leaf angle and seed size in B. distachyon. In addition to leaf angle, tillering is also negatively related to seed size, with reduced tillering resulting in increased seed size and increased tillering to reduced seed size. Despite the increased seed size and increased spikelet length in the cul4^TI mutant, the seed yield per plant was not increased because of the sterility observed in this mutant. In the case of the laxa mutants the reduced number of florets per spike, associated with a reduction in seed size, also resulted in a reduced seed production per plant. Thus in the B. distachyon nbcl mutants, seed yield per plant was not related to tillering or plant biomass.

In the double mutant, the leaf sheath no longer wrapped around the stem making the entire leaf growing nearly horizontally and separately from the stem. This strong phenotype demonstrates the crucial role of NBCL in grass leaf organogenesis.

Several phenotypes described above for our nbcl mutants are similar to those described for hormonal mutants. For example, the O. sativa clustered primary branch 1 brassinosteroid mutant is affected in growth (semi dwarf), leaf angle, spikelet architecture, and seed yield (Wu et al., 2016). Similarly, in Z. mays, the brassinosteroid-deficient dwarf1 mutant is dwarf, showing reduced fertility and modification of the auricle-leaf region (Makarevitch et al., 2012). It will be interesting in future works to study the link between B. distachyon NBCL genes and the brassinosteroids which represent major actors of leaf angle determination in plants (Li et al., 2020). However, other hormones, like auxin, can also be involved in leaf angle determination (Zhang et al., 2014a, 2014b).

Inflorescence architecture is the most prominent part of small-grain cereal plants, it has a direct effect on yield and represents a key aspect for selection in breeding programs aiming to improve yield. In H. vulgare, while the loss-of-function of HvCul4 does not affect spike or spikelet development, the mutation of HvLax-a causes pleiotropic developmental alterations in the inflorescence (Tavakol et al., 2015; Jost et al., 2016). The HvLax-a mutants produced elongated spike rachis leading to more relaxed inflorescence and produced thinner grains that are exposed at spike maturity due to a reduced marginal growth of the palea and lemma (Jost et al., 2016).

In B. distachyon, cul4^TI mutants produced longer terminal and lateral spikelets due to an exaggerated and continuous production of florets (Magne et al., 2020) while laxa^TI and laxa^CR mutants produced characteristic short spikelets with a reduced number of florets. In addition, laxa^CR showed even stronger defects than the previously described laxa^TI alleles suggesting that laxa^TI were weak mutants. In these laxa mutants, the floret rachilla is bent and tend to grow perpendicularly from the abaxial/adaxial floret axis leading to a shorter and opened spikelet architecture. The spikelet phenotype of the cul4^TIlaxa^TI double mutants carrying a laxa weak mutation is similar to the one of cul4^TI. However, in the cul4^TIlaxa^CR double KO mutants, we observed a strong homeotic modification of the spikelet into a leaf-like structure, looking-like vegetative spikes, in which the entire spikelets were replaced by a single leaf-like organ. This double mutant was consequently sterile and did not produce any seeds.

In grasses, AP2/ERF transcription factors represent key regulators of spike architecture, controlling the number and the development of spikelets, and floret meristem fate. Loss-of-function of AP2/ERF such as ZmBRANCHED SILKLESS1 (Chuck et al., 2002), ZmIDS1/SID1 (Chuck et al., 2008), TaFRIZZY PANICLE (Dobrovolskaya et al., 2015), HvCompositum2 (Poursarebani et al., 2015), OsFRIZZY PANICLE/BRANCHED FLORETLESS1 (Komatsu et al., 2003), and BdMORESPIKELETS1Derbyshire and Byrne, 2013; Dobrovolskaya et al., 2015), triggers spikelet-to-branch homeosis with each branch looking like an entire spike. In our work, we found that the gene expression of the AP2/ERF BdIDS1 and BdSID1 was significantly reduced in laxa^CR and moderately decreased in cul4^TIlaxa^TI suggesting that BdIDS1 and BdSID1 expressions are partly LAXA-dependent. In addition, the expression level of BdAP1 and BdKNAT6 was significantly decreased in all nbcl mutant lines studied here. This control of flower identity gene expressions by the CUL4 and LAXA genes may explain the different phenotypes observed in the mutants and especially the strong modifications of the flower development (homeosis) observed in the double KO mutant cul4^TIlaxa^CR. In B. distachyon, CUL4 and LAXA are thus playing important roles in spikelet development and the phenotype of cul4^TIlaxa^CR clearly highlights the essential roles of CUL4 and LAXA in the reproductive transition and spikelet identity acquisition.

In H. vulgare, Hvlax-a mutant alleles are also altered in floral organ identity and displayed homeotic conversion of lodicules into stamenoid structures (Jost et al., 2016). Lodicule-to-stamen transformations were also reported in the B. distachyon laxa^TI mutants and additional stamens were often found (Magne et al., 2020). In this study, we used laxa^CR null alleles and confirmed previous results obtained using missense mutant alleles suggesting that LAXA positively contributes to floret patterning possibly through repressing male sexual fate and controlling the number of stamens. This also indicates that LAXA participates in the balance between class A and class B functions (Irish, 2017) during flower development. Interestingly, in the cul4^TIlaxa^TI double mutants, we did not observe lodicule-to-stamen homeosis or additional stamens. Once again, this suggests possible antagonist actions between CUL4 and LAXA during flower development. In A. thaliana, BOP1/2 regulate the expression of AP1 leading to the down-regulation of AGAMOUS (AG)-LIKE24, a homolog of AG (Xu et al., 2010) and promoting A class floral patterning. In contrast, in O. sativa, ectopic expression of OsMADS3, the ortholog of AG, leads to the homeotic transformation of lodicules into stamens and thus C class floral patterning (Kyozuka and Shimamoto, 2002). We can thus hypothesize that in laxa, the conversion of lodicules into stamens could be caused by the downregulation of AP1 and possibly upregulation of C class genes in this mutant. Future studies will define more precisely the role of this gene in flower formation in B. distachyon.

Natural seed shattering leads to seed loss in the field, limiting grain harvesting and causing drastic reduction in yields (Ji et al., 2006). Therefore, seed shattering represents a key agronomic trait. In dicots, NBCL genes are involved in abscission zone formation and functioning (Hepworth et al., 2005; McKim et al., 2008; Wu et al., 2012; Ichihashi et al., 2014; Couzigou et al., 2016; Frankowski et al., 2015). In grasses, such a role for the NBCL genes in seed shattering was not investigated yet and cultivated grasses are already selected for a nonshattering phenotype.

B. distachyon has not been selected for a nonshattering phenotype and represents an ideal model to determine whether NBCLs genes participate in the abscission ability in grasses. In a previous study, we showed that cul4^TI KO and laxa^TI missense mutants were not affected for seed abscission zone establishment and functioning (Magne et al., 2020). In this study, laxa KO mutants and cul4laxa double mutants were studied for abscission ability. These mutants did not show visible defects in seed shattering. Thus, our work suggests that, in B. distachyon, the NBCLs may not participate in the seed abscission process. These results might also mean that in grass species, abscission would be NBCLs-independent.

In conclusion, by using B. distachyon nbcl double mutants, we showed that the B. distachyon BTB/POZ and ANKYRIN domain proteins, CUL4 and LAXA, are redundantly required for B. distachyon development and that they also often show antagonist functions in development. Our studies also unravel a series of functions for these NBCLs associated with development, architecture, reproduction, and cell wall biosynthesis that represent agronomic characters of interest for cereals and make them candidate targets for breeding and neo-domestication of wild grasses.

Materials and methods

Plant material

The community standard diploid inbred line Bd21-3 (WT) was used for transformation and as a control in all experiments. The TILLING mutant alleles used in this work are described in Magne et al. (2020).

Genetic crosses

Brachypodium (B. distachyon) genetic crosses were performed following methods from Garvin lab (2009, 2010) and Vogel lab (2010). To generate the weak double mutant cul4^TIlaxa^TI, we crossed the cul4^W203* and laxa^T381I TILLING mutants. We succeeded to generate a single F1 line and to retrieve six cul4^TIlaxa^TI (cul4^W203*laxa^T381I) homozygous double mutant plants within a progeny of 137 individuals in which segregated both cul4 and laxa single mutant phenotypes.

Growth conditions

After the removal of lemma and palea, B. distachyon WT and mutant seeds were surfaces sterilized in a solution of sodium hypochlorite (half a pellet, 1.5 g per 1 L of sterile water; NOTILIA group, ref:156104) with one droplet of liquid soap for 10 min, shaking at room temperature. Surface sterilized seeds were washed 3 times with sterile water before placement on 7% (w/v) Kalys Agar plates. Sealed plates were stratified for 7 d at 4°C under darkness and then transferred to 19°C for 48 h under darkness for acclimatization. Seedlings were transferred in 1.5 l pots in a loam–sand–perlite mixture (2:1:1; v/v/v) or in a sand–perlite mixture (1:2, v/v) in a controlled growth chamber with a 20-h light/4-h dark cycle, 19/17°C day–night temperature, relative humidity 60% and photosynthetic photon flux density (200 µmol m⁻² s⁻¹ at 10 cm above the ground). Plants were watered 3 times per week.

CRISPR–Cas9 strategy

For CRISPR–Cas9-mediated genome edition in B. distachyon, we used Gateway binary T-DNA vectors designed for rice (O. sativa) transformation co-expressing CAS9 and guide RNA (Miao et al., 2013). Two sequence-specific single guides RNA (sgRNA) located in the two exons of the LAXA gene were used (Supplemental Table S1). In order to produce laxa^CR single mutant and cul4^TIlaxa^CR double mutant, the LAXA locus was targeted in WT B. distachyon and cul4^Q127* KO mutant backgrounds, respectively. The sgRNAs were designed using the http://crispor.tefor.net/ web site. The sgRNAs were expressed from the polymerase III-type promoter of U3 small nuclear RNA and the CAS9 coding sequence was expressed from the maize (Z. mays) Ubiquitin (Ubi) promoter (Miao et al., 2013). The two annealed oligos containing BsaI restriction sites were inserted into the BsaI-digested pOssgRNA gateway plasmid, checked by sequencing and combined into the binary vector pH-Ubi-cas9-7. The constructs were introduced by electroporation into Agrobacterium tumefaciens strain AGL1 (Lazo et al., 1991) and subsequently used for B. distachyon embryo-derived calli transformation.

Brachypodium distachyon calli culture

Tissue culture was performed according to the procedures described in Bragg et al. (2015). Embryos (<0.3 mm) were dissected out from sterilized WT and cul4^Q127*mutant seeds ∼14 d after anthesis (Draper et al., 2001). Embryo was then placed with the scutellar surface in contact with the callus induction medium (CIM: 4.4 g/L Murashige and Skoog [MS] minimal organics and Skoog medium including vitamins [Duchefa Biochemie, Netherlands, Haarlem; M0222.0050], 0.07% [w/v] Solution MES [+BCP], and 0.0008% [w/v] Solution BCP, 30 g/L sucrose, 0.6 mg/L CuSO₄, 2.5 mg/L 2,4-D, pH 5.7, and 0.2% [w/v] Phytagel [Sigma, St Louis, MO, USA; P8169]). Sealed plates were incubated at 28°C under darkness for 3–4 weeks. Calli were then transferred to a fresh callus induction medium and cultured for two additional weeks. The embryogenic calli from the second subculture were grown for one additional week before being used for transformation.

Agrobacterium tumefaciens-mediated transformation of B. distachyon

The B. distachyon transformations were performed as described in Christiansen et al. (2005) and Bragg et al. (2015) with minor modifications. For the A. tumefaciens-callus co-culture, the CIM-based medium was supplemented with 60 mg/L acetosyringone and 0.1% pluronic (F-68 solution, Sigma-life science, CAS: 9003-11-6). The transgenic tissues were selected on the CIM medium supplemented with 200 mg/L hygromycin (Sigma H3274-1g) and 250 mg/L Timentin (Duchefa T0190.0025). For shoot regeneration, the selection medium was supplemented with 0.2 mg/L Kinetin (Sigma, K3378), 2,4-D was not used (Sigma D7299-100G) and the sucrose was replaced by maltose. Plates were incubated under light condition at 28°C, calli started to turn green and shoots appeared within 2–4 weeks. Transgenic shoots were then transferred on rooting medium based on CIM medium supplemented with 2.5 mg/L Indole-3-butyric acid (Sigma I5386), 10 g/L saccharose, and 2,4-D was not added. Four and three independent T0-generation transgenic laxa^CR and cul4^TIlaxa^CR were obtained, respectively. The T0 transgenic plants contained a mixture of laxa^CR mutations in the different tillers with some tillers containing bi-allelic mutations. This suggested that the CRISPR–Cas9 machinery was operating during plant regeneration, creating independent mutations in the different tillers of a single plant (see chimeric plants in Supplemental Figure S2C). Up to five different types of mutations could be isolated in one plant (Supplemental Figure S1). In total, we obtained six different types of mutations for LAXA in these CRISPR–Cas9 transgenic plants (Supplemental Figure S1). To make the analysis easier the mono allelic mutant plants were selected from the progeny of heterozygous or bi-allelic mutant T0 plants for further studies.

Brachypodium distachyon genomic DNA extraction

Brachypodium distachyon DNA of individual plants was isolated from young leaves using a phenol/chloroform procedure (Theologis et al., 1985). DNA was precipitated using 3M cold sodium acetate: isopropanol (0.1:1) and washed by 70% (v/v) ethanol. DNA samples were dried, resuspended in sterile water and RNase treated (Roche).

Genotyping of the transgenic plants

Plant genotyping was performed by PCR using AccuPrime GC-Rich DNA Polymerase (Invitrogen, Waltham, MA, USA) or goTaq DNA polymerase (Promega Madison, WI, USA). The AccuPrime GC-Rich DNA Polymerase was used to amplify the GC-rich region of the first exon. With other enzymes, the PCR products consisted in a mixture of deletions making the sequencing impossible. PCR products were sequenced by Sanger (http://www.eurofinsgenomics.eu) and analyzed with A plasmid Editor software version 2.0.50. Transgenic plants were also checked by PCR for the presence of the selection marker and CAS9 sequences. Respectively, four and three independent transgenic lines were used for laxa^CR and cul4^TI laxa^CR mutant subsequent phenotyping. Information related to the primers used for the genotyping is provided in Supplemental Table S1.

RT-qPCR gene expression analysis

RNA extraction, DNAse treatment, and cDNAs reverse transcription were performed as described in Magne et al. (2020). Gene expression analysis was performed on three independent collects of 40 d old leaves, nodes, internodes, stems, and young spikelets samples of 40-d old WT and mutant B. distachyon. Young spikelets were collected and flash-frozen when the inflorescence was just visible from the flag leaf of the tallest stem with equivalent development. RT-qPCR reactions were performed on a LightCycler96 (ROCHE, Basel, Switzerland). The final threshold cycle (Ct), efficiency, and initial fluorescence (R0) were calculated with the Miner algorithm (Zhao and Fernald, 2005). Relative expression levels were obtained from the ratio between R0 of the reference gene and R0 of the target gene. RT-qPCR primers information is given in Supplemental Table S2.

Histochemical staining of lignin

Sections were stained with 1% (w/v) phloroglucinol for 2 min followed by a wash in 50% (v/v) HCl and were mounted onto microscope slides for observation as described in Bouvier d’Yvoire et al. (2013). All the histochemical stainings were performed on sections cut in the middle of the second internode from the base of developmentally equivalent transgenic and control plants. Samples were embedded in 4% (w/v) agarose and transversely sectioned at a thickness of 70 µm using a vibratome (Leica VT1200S, Leica, Wetzlar, Germany). Sections were observed using an Apotome II microscope system (Zeiss Dublin, CA, USA) with automatic exposure times.

Lignin and monosaccharide composition of stem secondary cell wall

Main stems without spikelets were collected and ground. Ground samples were washed 3 times at 60°C with 50 mL of ethanol. At each step, the samples were vortexed. Samples were dried for 12 h at 40°C. About 5 mg of extract-free samples were used for both lignin content measurement and sugar analysis. Lignin content was measured using the acetyl bromide method according to Dence and Lin (1992). The quantification of uncondensed lignin units was performed by thioacidolysis as described in Méchin et al. (2014) and analyzed with an Agilent 5973 Gas Chromatography–Mass Spectrometry system. Neutral monosaccharide composition was performed by Gas Chromatography after sulfuric acid degradation of alcohol insoluble residue (Hoebler et al., 1989) and conversion of the monomers to alditol acetates (Blakeney et al., 1983). Chromatography was performed on a Perkin Elmer Gas Chromatography Gas 580. Calibration was made with standard sugar solution and inositol as internal standard.

Technovit section

Thin tissue sections were performed using technovit resins as described in Van de Velde et al. (2006). In brief, fixed samples were infiltrated 15 min under vacuum (≈500 mm of Hg) in sodium cacodylate buffer (0.05 M, pH: 7), 1% (v/v) glutaraldehyde, and 4% (v/v) formaldehyde and incubate at 4°C overnight. Once dehydrated by successive ethyl-alcohol bathes, samples undergo three successive ethyl-alcohol/Technovit stock solutions (3:1, v/v), (1:1, v/v), and (1:3, v/v) bathes and three 100% Technovit stock solution bathes at 4°C under agitation. Samples were included in Technovit resin thanks to a Teflon Histoform S embedding mold (Heraeus Kulzer). Technovit sections were carried out using a microtome RM 2165 (LEICA) and tungsten disposable blade (TC-65, LEICA), 8-µm thickness sections were stained 10 min by immersion in toluidine blue 0.02% (w/v). Pictures were acquired using an Apotome II microscope (Zeiss) and exploited using ZEN (blue edition) software.

Data availability statement

The mutants described in this manuscript are available on request by contacting Pascal Ratet (pascal.ratet@cnrs.fr). All relevant data can be found within the manuscript and its supporting materials.

Accession numbers

BdUNICULME4 (BdCUL4, Bradi2g60710); BdLAXATUM-A (BdLAXA, Bradi4g43150).

Supplemental data

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

Supplemental Figure S1. CRISPR–Cas9 editions of the LAXA locus in WT and cul4^Q127* backgrounds, predicted LAXA mutant proteins and LAXA gene expression.

Supplemental Figure S2. Tiller phenotypes of laxa and cul4laxa mutants.

Supplemental Figure S3. Primary root growth of B. distachyon nbcl mutants.

Supplemental Figure S4. Brachypodium distachyon nbcl mutant biomasses.

Supplemental Figure S5. Cell wall- and lignin-related gene expression in B. distachyon nbcl mutant stems.

Supplemental Figure S6. CUL4 and LAXA gene expression and a regulatory model for ligule and auricles formation.

Supplemental Figure S7. Flowers phenotypes in B. distachyon nbcl mutants.

Supplemental Figure S8. Floral marker gene expression in B. distachyon nbcl mutants.

Supplemental Figure S9. Shattering phenotypes of laxa^CR and cul4^TIlaxa^TI mutants.

Supplemental Table S1. Primers used for genotyping and sgRNA used for CRISPR–Cas9.

Supplemental Table S2. Primers used for RT-qPCR gene expression analysis.

 

P.R. and S.L. conceived the project. S.L. performed the CRISPR–Cas9 approaches, the gene expression analysis, the mutant genotyping, and the histological analysis. K.M. performed the TILLING mutant isolation and the crossing of the double mutant. S.D. and R.S. performed the cell wall composition analysis. S.L., K.M., and P.R. wrote the article.

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 Pascal Ratet (pascal.ratet@cnrs.fr).