AtEXT3 is not essential for early embryogenesis or plant viability in Arabidopsis

EXTENSINs (EXTs) are a subclass of the very diverse family of cell wall proteins known as hydroxyproline-rich glycoproteins. The EXT subclass is defined by the presence of at least two repeats of the Ser-Pro3-5 motif in which the proline residues can be hydroxylated and, subsequently, O-glycosylated with arabinose containing oligosaccharides leading to the formation of hydrophilic domains (Cannon et al., 2008; Velasquez et al., 2011; Ogawa-Ohnishi et al., 2013). In addition,many EXTs from land plants also contain distinct and characteristic hydrophobic Tyr-containing motifs (Schnabelrauch et al., 1996; Held et al., 2004; Cannon et al., 2008). These have recently been termed cross-linking (CL)EXTs, because of their ability to undergo peroxidase-mediated covalent cross-linking in vitro through their Tyr residues, a phenomenon that has been proposed to underlie important structural functions in plant cell walls (Marzol et al., 2018). EXTENSINs have a marked ability to self-assemble into dendritic networks in vitro (Cannon et al., 2008). This capacity, proposed to be based on self-ordering due to the alternation of hydrophobic and hydrophilic domains (Lamport et al., 2011),maymediate observed interactions with other cell wall components, including pectins (Valentin et al., 2010). Despite an abundance of information regarding the biochemical behaviour of EXTs in vitro, information on their biological functions in vivo remainsmarkedly scarce, possibly due to extensive redundancy between the multiple genes encoding EXTs and EXTlike proteins within plant genomes. Work in Arabidopsis has identified potential roles for AtEXTs in promoting root-hair elongation (Velasquez et al., 2011), pollen development (Choudhary et al., 2015) and early embryogenesis (Hall & Cannon, 2002; Saha et al., 2013; Chen et al., 2015). This last function is our focus in this Letter. Following an enhancer-trap (transposon)-based mutagenesis screen, Hall & Cannon (2002) isolated a seedling-lethal mutant called root-shoot-hypocotyl-defective (rsh) which, when homozygous, produces disorganised embryos that arrest either before (Cannon et al., 2008) or just after (Hall & Cannon, 2002) germination. Mutants show clear defects in cell-plate formation. A transposon insertion in the promoter of the AtEXT3 gene was stated to be the causal lesion in rsh mutants. Consistent with this, complementation of the rsh phenotype using an AtEXT3 genomic DNA fragment was reported. However, in two subsequent studies, ‘selfrescue’ of the rsh phenotype to a fully wild-type phenotype was reported in up to 20% of genetically atext3 homozygous plants (Saha et al., 2013; Chen et al., 2015). No intermediate phenotypes were reported, and ‘rescue’ was transmitted stably to all the progeny in subsequent generations. This calls into question the validity of genetic complementation experiments, as the reported ‘self-rescue’ phenomenon would interfere with this approach. A mechanism involving compensatory regulation of other AtEXT-encoding genes was evoked to explain this phenomenon. Nonetheless it appears difficult, based on the published data, to exclude the possibility that the reported rsh phenotype is caused by a lesion linked to the original transposon insertion inAtEXT3, but affecting a separate locus, and lost by segregation in ‘rescued’ lines. We became interested in the AtEXT3 gene as a result of our discovery that the developing embryo of Arabidopsis is covered with an endosperm-derived ‘sheath’ rich in epitopes detected by anti-EXT antibodies (Moussu et al., 2017). In silico data (Le et al., 2010) concur with published in situ hybridisation data (Francoz et al., 2016) in showing that AtEXT3 is expressed primarily in the embryo-surrounding endosperm during Arabidopsis seed development, rather than in the embryo itself. We were therefore keen to further investigate the function of AtEXT3 and we generated a series of new alleles in the Col-0 background using the CRISPR-Cas9 based technique. We selected four alleles containing deletions or insertions within the 50 coding sequence of the AtEXT3 gene (Fig. 1a). All are predicted to produce either strongly truncated or frame-shifted proteins (Fig. 1b). Being aware of recent work suggesting that 50 mutations generated by CRISPR can lead to the production of functional proteins due to the use of downstream in-frame ATGs (Smits et al., 2019), we verified the AtEXT3 coding sequence (which lacks introns) for such sequences. None are present. We therefore conclude that all four alleles are null. Homozygous plants were all fully phenotypically wild-type in terms of root growth, plant growth and reproductive development (Fig. 1c–e). No defective seeds or seedlings were detected for any line, in any generation. Our results suggest that loss of AtEXT3does not lead to detectable defects in development in the Col-0 background. However, the rsh mutant originally describedbyHall andCannonwas identified in the Landsberg erecta (L-er) background. To eliminate the possibility that a suppressor exists inCol-0 that is not present in theL-erbackground, we carried out CRISPR experiments in the L-er background, this time using guides predicted to interrupt the EXTENSIN 2 domains of the AtEXT3 protein. Of multiple null alleles brought to homozygosity we selected three deletions predicted to produce either strongly truncated or frame-shifted proteins (Fig. 2). We were again unable to detect any defects in root growth, plant growth or reproductive development in any of these lines. In addition, we confirmed the absence of phenotypes during seed development, and germination (Supporting Information Fig. S1). Taken together, our results indicate that AtEXT3 is not essential for embryogenesis and plant development in Arabidopsis,

EXTENSINs (EXTs) are a subclass of the very diverse family of cell wall proteins known as hydroxyproline-rich glycoproteins. The EXT subclass is defined by the presence of at least two repeats of the Ser-Pro 3-5 motif in which the proline residues can be hydroxylated and, subsequently, O-glycosylated with arabinose containing oligosaccharides leading to the formation of hydrophilic domains (Cannon et al., 2008;Velasquez et al., 2011;Ogawa-Ohnishi et al., 2013). In addition, many EXTs from land plants also contain distinct and characteristic hydrophobic Tyr-containing motifs (Schnabelrauch et al., 1996;Held et al., 2004;Cannon et al., 2008). These have recently been termed cross-linking (CL)-EXTs, because of their ability to undergo peroxidase-mediated covalent cross-linking in vitro through their Tyr residues, a phenomenon that has been proposed to underlie important structural functions in plant cell walls (Marzol et al., 2018). EXTENSINs have a marked ability to self-assemble into dendritic networks in vitro (Cannon et al., 2008). This capacity, proposed to be based on self-ordering due to the alternation of hydrophobic and hydrophilic domains (Lamport et al., 2011), may mediate observed interactions with other cell wall components, including pectins (Valentin et al., 2010).
Despite an abundance of information regarding the biochemical behaviour of EXTs in vitro, information on their biological functions in vivo remains markedly scarce, possibly due to extensive redundancy between the multiple genes encoding EXTs and EXTlike proteins within plant genomes. Work in Arabidopsis has identified potential roles for AtEXTs in promoting root-hair elongation (Velasquez et al., 2011), pollen development (Choudhary et al., 2015) and early embryogenesis (Hall & Cannon, 2002;Saha et al., 2013;Chen et al., 2015). This last function is our focus in this Letter.
Following an enhancer-trap (transposon)-based mutagenesis screen, Hall & Cannon (2002) isolated a seedling-lethal mutant called root-shoot-hypocotyl-defective (rsh) which, when homozygous, produces disorganised embryos that arrest either before (Cannon et al., 2008) or just after (Hall & Cannon, 2002) germination. Mutants show clear defects in cell-plate formation. A transposon insertion in the promoter of the AtEXT3 gene was stated to be the causal lesion in rsh mutants. Consistent with this, complementation of the rsh phenotype using an AtEXT3 genomic DNA fragment was reported. However, in two subsequent studies, 'selfrescue' of the rsh phenotype to a fully wild-type phenotype was reported in up to 20% of genetically atext3 homozygous plants (Saha et al., 2013;Chen et al., 2015). No intermediate phenotypes were reported, and 'rescue' was transmitted stably to all the progeny in subsequent generations. This calls into question the validity of genetic complementation experiments, as the reported 'self-rescue' phenomenon would interfere with this approach. A mechanism involving compensatory regulation of other AtEXT-encoding genes was evoked to explain this phenomenon. Nonetheless it appears difficult, based on the published data, to exclude the possibility that the reported rsh phenotype is caused by a lesion linked to the original transposon insertion in AtEXT3, but affecting a separate locus, and lost by segregation in 'rescued' lines.
We became interested in the AtEXT3 gene as a result of our discovery that the developing embryo of Arabidopsis is covered with an endosperm-derived 'sheath' rich in epitopes detected by anti-EXT antibodies (Moussu et al., 2017). In silico data (Le et al., 2010) concur with published in situ hybridisation data (Francoz et al., 2016) in showing that AtEXT3 is expressed primarily in the embryo-surrounding endosperm during Arabidopsis seed development, rather than in the embryo itself. We were therefore keen to further investigate the function of AtEXT3 and we generated a series of new alleles in the Col-0 background using the CRISPR-Cas9 based technique. We selected four alleles containing deletions or insertions within the 5 0 coding sequence of the AtEXT3 gene (Fig. 1a). All are predicted to produce either strongly truncated or frame-shifted proteins (Fig. 1b). Being aware of recent work suggesting that 5 0 mutations generated by CRISPR can lead to the production of functional proteins due to the use of downstream in-frame ATGs (Smits et al., 2019), we verified the AtEXT3 coding sequence (which lacks introns) for such sequences. None are present. We therefore conclude that all four alleles are null. Homozygous plants were all fully phenotypically wild-type in terms of root growth, plant growth and reproductive development . No defective seeds or seedlings were detected for any line, in any generation.
Our results suggest that loss of AtEXT3 does not lead to detectable defects in development in the Col-0 background. However, the rsh mutant originally described by Hall and Cannon was identified in the Landsberg erecta (L-er) background. To eliminate the possibility that a suppressor exists in Col-0 that is not present in the L-er background, we carried out CRISPR experiments in the L-er background, this time using guides predicted to interrupt the EXTENSIN 2 domains of the AtEXT3 protein. Of multiple null alleles brought to homozygosity we selected three deletions predicted to produce either strongly truncated or frame-shifted proteins (Fig. 2). We were again unable to detect any defects in root growth, plant growth or reproductive development in any of these lines. In addition, we confirmed the absence of phenotypes during seed development, and germination (Supporting Information Fig. S1).
Taken together, our results indicate that AtEXT3 is not essential for embryogenesis and plant development in Arabidopsis,   suggesting that its function is likely to be redundant with other related proteins. We feel that it is important that our result be communicated to the scientific community, as the role of AtEXT3 has been very widely cited as a proof of the important function of EXTs in plant development and particularly in cytokinesis. Although fully convinced that EXTs are of primary importance in plants, we propose that the roles of AtEXT3 may be almost fully redundant with those of other related proteins. Supporting this hypothesis, phenotypeless atext3 mutants were shown to have extensive changes in the expression of other AtEXT-encoding genes, or the accumulation of AtEXT proteins (Saha et al., 2013;Chen et al., 2015). It is possible that a similar phenomenon occurs in the CRISPR generated atext3 mutants. Finally, our study does not call into question the presence of AtEXT3 in cell walls post germination, or the beautiful biochemical studies of AtEXT3 behaviour in vitro that have been published (Cannon et al., 2008).
For the generation of the CRISPR/Cas9 alleles of EXT3 on the Ler background the method described by Stuttmann et al. (2021) was used. Two guides (first guide, 5 0 -GGTGGGGAGTGGTA TACCGG-3 0 ; second guide, 5 0 -ATACAAATCTCCACCTC CAC-3 0 ) were inserted into the pDGE332 and pDGE334 shuttle vectors and recombined into the pDGE347 recipient vector.

Stable transformation of Arabidopsis thaliana (L.) Heynh
The plasmids generated were first transformed into the C58PMP90 Agrobacterium strain by electroporation at 2.2 kV in a 1-ml cuvette (Eurogentec, Liege, Belgium). Agrobacteria were then grown at 28°C for 2 h in LB liquid medium without antibiotics before being spread on Petri dishes containing YEB solid medium with rifampicin (50 mg l À1 ), gentamicin (20 mg l À1 ) and kanamycin (50 mg l À1 ) for the pHEE401-based plasmid or rifampicin (50 mg l À1 ), gentamicin (20 mg l À1 ) and spectinomycin (100 mg l À1 ) for the pDGE347 based plasmid. After 2 d, the agrobacteria were used to transform Col-0 plants using the floral-dip method (Logemann et al., 2006). Selection of transgenic plants was performed on Murashige & Skoog (MS) plates with 30 mg l À1 of hygromycin.

Identification of the mutant alleles and generation of homozygous plants
Amplification of the region around the cutting sites was performed by PCR on DNA extracted from transgenic Col-0 plants using the 5 0 -GGGTGTGAAGGGAAGGCACTAAATC-3 0 and 5 0 -GTAA ACGTAGTGCTTCTTTGGTGG-3 0 primers. For the transgenic L-er plants the primers used were 5 0 -GGGTGTGAAGGGAAGG CACTAAATC-3 0 and 5 0 -AGGTGGGGGTGGGGAATGGTA-3 0 . Sequencing was carried out with the 5 0 -GTAAACGTAGTGC TTCTTTGGTGG-3 0 (Col-0) or and 5 0 -GGGTGTGAAGG GAAGGCACTAAATC-3 0 (L-er) primers. Homozygous mutants for atext3 were selected from the T2 plants. In the Col-0 background the CRISPR/Cas9 cassette was removed by selection of hygromycin sensitive plants and validated by the absence of cassette amplification by PCR with the 5 0 -TGTCCCAGGATTA GAATGATTAGGC-3 0 and 5 0 -AGCCCTCTTCTTTCGATC CATCAAC-3 0 primers. Homozygous mutants were confirmed by sequencing. In L-er transgenic plants nonfluorescent seeds were selected in the T2 and homozygous mutants were confirmed by sequencing.

Plant growth
For the selection processes and in vitro growth (Figs 1c, 2c), seeds were gas sterilised with chlorine gas (3 ml of HCl (33%) in 100 ml of bleach) for at least 3 h in a hermetic box. Seeds were then sown on MS medium with 0.5% of sucrose and, for the selection of Col-0 mutants, with 30 mg l À1 of hygromycin. Stratification was carried out 2 d at 4°C in darkness. Plates were then transferred into growth chambers in long day conditions (16 h light, 21°C). For the in vitro growth experiments, plants were grown for 14 d (Col 0) or 10 d (Ler) and pictures were taken with an Epson perfection V300 photo scanner. For germination tests, plants were grown for 12 d (100 seeds per plate). In all the other cases, seedlings were transferred after 7 d onto soil (Argile 10 (favorit)) and grown in long day conditions (16 h light, 21°C). Images of plants at 29 d after stratification (DAS) and 46 DAS (Fig. 1d,e) or 28 DAS and 46 DAS (Fig. 2d,e) were taken with a Canon EOS 450D camera with a Sigma 50 mm f/2.8 DG Macro objective. For analysis of silique filling, individual siliques were opened at the mature green stage, and seeds were counted under a binocular microscope.

Seed clearing
To visualise developing seeds, the siliques at 7 d after pollination (DAP) were dissected with a needle and forceps on adhesive tape on a microscope slide and mounted in the clearing solution (1 : 7, glycerol : chloral hydrate liquid solution, v/v; VWR Chemicals, Rosny-sous-Bois, France). After 48 h incubation at 4°C samples were imaged under a Zeiss Axio Imager M2 microscope.