Golgi-localized membrane protein AtTMN1/EMP12 functions in the deposition of rhamnogalacturonan II and I for cell growth in Arabidopsis

Abstract

Appropriate pectin deposition in cell walls is important for cell growth in plants. Rhamnogalacturonan II (RG-II) is a portion of pectic polysaccharides; its borate crosslinking is essential for maintenance of pectic networks. However, the overall process of RG-II synthesis is not fully understood. To identify a novel factor for RG-II deposition or dimerization in cell walls, we screened Arabidopsis mutants with altered boron (B)-dependent growth. The mutants exhibited alleviated disorders of primary root and stem elongation, and fertility under low B, but reduced primary root lengths under sufficient B conditions. Altered primary root elongation was associated with cell elongation changes caused by loss of function in AtTMN1 (Transmembrane Nine 1)/EMP12, which encodes a Golgi-localized membrane protein of unknown function that is conserved among eukaryotes. Mutant leaf and root dry weights were lower than those of wild-type plants, regardless of B conditions. In cell walls, AtTMN1 mutations reduced concentrations of B, RG-II specific 2-keto-3-deoxy monosaccharides, and rhamnose largely derived from rhamnogalacturonan I (RG-I), suggesting reduced RG-II and RG-I. Together, our findings demonstrate that AtTMN1 is required for the deposition of RG-II and RG-I for cell growth and suggest that pectin modulates plant growth under low B conditions.

Arabidopsis thaliana, boron, cell elongation, cellular attachment, endomembrane protein (EMP), Golgi apparatus, pectin deposition, rhamnogalacturonan I, rhamnogalacturonan II, transmembrane nine protein (TMN)

Introduction

In multicellular organisms, tissue and organ formation is mediated by cellular adhesion. In plants, the stiffness and plasticity of cell walls support cellular growth and organization. Pectin, which is found within primary cell walls as a gel matrix, is critical for cell adherence (Daher and Braybrook, 2015).

Pectin consists of galacturonic acid (GalA)-rich polysaccharides, including homogalacturonan (HG), rhamnogalacturonan-I (RG-I), and rhamnogalacturonan-II (RG-II) domains (Voragen et al., 2009). HG is a major pectin component composed of a polymer of (α-1,4)-linked d-galacturonic acids. De-methylesterified HG chains are crosslinked through ionic bonding between calcium ion (Ca²⁺) and carboxyl residues. This crosslinking of HG and Ca²⁺ is important for normal pectin network maintenance, which controls pectin gel properties. RG-I represents the second largest fraction of pectin; it contains a backbone consisting of repeating (α-1,2)-l-rhamnose-(α-1,4)-d-galacturonic acid and side chains. RG-II is a minor but highly conserved component in plants; it consists of a HG backbone linked to four distinct side chains composed of more than 12 different monosaccharides (O’Neill et al., 2004). RG-II is crosslinked with borate between the two RG-II monomers, to form borate-dimerized RG-II through the formation of cis-diol ester bonds with apiosyl residues in side chain A. Boron (B) depletion has been reported to cause cellular adhesion and cell elongation disorders in plants through a reduction in the quantity of crosslinked RG-II (Ishii et al., 2001; Iwai et al., 2002; Chatterjee et al., 2014; Camacho-Cristobal et al., 2015).

Several proteins involved in pectin biosynthesis have been characterized. RG-II structure is synthesized in the Golgi apparatus by a variety of glycosyltransferases, using nucleotide diphosphate-linked sugars as activated donor substrates (Driouich et al., 2012). Most nucleotide diphosphate-linked sugars are synthesized in the cytosol (Ahn et al., 2006; Reboul et al., 2011) and then transported into the Golgi (Sechet et al., 2018). However, thus far, the overall molecular mechanism underlying RG-II polysaccharide synthesis in the Golgi and subsequent secretion to the apoplast remains poorly understood.

Recent studies have shown that the transmembrane nine (TMN) protein family is involved in cellular attachment among eukaryotes including Drosophila melanogaster, Dictyostelium discoideum, Saccharomyces cerevisiae, and Homo sapiens (Cornillon et al., 2000; Bergeret et al., 2008; Froquet et al., 2008; Paolillo et al., 2015). TMNs belong to a superfamily of nine multi-spanning endomembrane proteins; they are highly evolutionarily conserved (Hegelund et al., 2010). TMN proteins contain a long N-terminal domain with variable amino acid sequences, followed by nine transmembrane domains and a short tail on the C-terminus. TMN proteins in Dictyostelium and Drosophila interact with integrin-like protein and peptidoglycan recognition protein, respectively, for transport into the plasma membrane (Froquet et al., 2012; Perrin et al., 2015a, b). TMN proteins are localized to the Golgi apparatus in Dictyostelium and to intracellular vesicles or plasma membrane in Drosophila. TMN proteins may be involved in vesicular trafficking systems; they commonly contribute to cellular attachment among eukaryotes, despite their distinct interacting proteins and sub-cellular localization in different organisms. The presence of TMN is also conserved in plants; the Arabidopsis and Oryza sativa genomes have 12 and 17 paralogs, respectively. AtTMN1/EMP12 is localized to the Golgi apparatus (Gao et al., 2012); however, its physiological function in plants has not yet been revealed.

In this study we applied forward genetics to identify a new component required for RG-II deposition or dimerization in cell walls, focusing on the effects of altering B-dependent growth patterns, especially under low B conditions. This is because some RG-II synthesis mutants have shown impaired growth in response to changes in B concentrations caused by changes in the efficiency of borate-dimerized RG-II formation (O’Neill et al., 2001; Voxeur et al., 2011; Sechet et al., 2018). Through characterization of the isolated mutants, we revealed that loss of function of AtTMN1 partially relieved inhibition in root elongation under severely low B, and constantly reduced overall biomass compared with wild type plants, regardless of B conditions. It was suggested that the amounts of RG-II and RG-I were reduced in the cell wall of tmn1 mutants, highlighting a role of Arabidopsis TMN1 in deposition of pectin in cell walls, for normal plant growth.

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana (L.) Heynh. accession Columbia-0 (Col-0) was used as the wild type (WT). Plants were grown in solid or liquid media (Fujiwara et al., 1992). Solid media contained 1% (w/v) sucrose and 1% (w/v) gellan gum (FUJIFILM Wako Pure Chemical, Japan). Boron (B) concentrations were adjusted with boric acid. For growth in solid media, surface-sterilized seeds were incubated in ultrapure water at 4 °C for stratification. Seeds were sown on solid media containing 0.1 μM (severely low B) or 100 μM (normal B) boric acid. They were incubated in a vertical position at 22 °C under a 16 h light/8 h dark cycle; light was provided by white fluorescent lamps (7 000–8 000 lux). To analyse vegetative growth, total B concentration in plant tissues and cell wall properties, Arabidopsis plants were germinated on rockwool with ultrapure water, and then incubated for 7 d. Seedlings were transferred into liquid media containing 0.1 µM (severely low B), 0.3 µM (mildly low B), 100 µM (normal B), and 500 μM (high B) boric acid; they were incubated for 38 d for vegetative growth and total B concentration analysis, and for 37 d for cell wall analysis. During the initial 14 d after the transfer, liquid medium was replaced weekly; subsequently, they were replaced at intervals of 3 d. Plants were grown at 70% relative humidity at 22 °C under short days (10 h light/14 h dark) to continue the vegetative growth stage. For reproductive growth tests, after plants had been incubated in water on rockwool for 7 d, they were grown for 68 d in liquid media supplemented with 0.1 µM, 0.3 µM, 100 µM, and 500 μM boric acid under long days (16 h light/8 h dark). Liquid medium was replaced weekly during the initial 14 d after transfer, and subsequently at intervals of 3 d or 4 d.

Mutant screening and genotyping of tmn1-1, tmn1-2, and tmn1-3 mutants

For mutant screening, Col-0 seeds were mutagenized with ethyl methanesulfonate; 12 400 mutagenized M₂ seeds were grown on solid media containing 0.03 μM boric acid, representing severe B deficiency and 114 plants showing longer roots were initially selected. Then, M₃ seeds derived from self-pollination of the selected M₂ plants, were grown on solid media under 0.03 μM and 100 μM B and mutant plants were screened for those showing longer roots compared with Col-0 under 0.03 μM (low B), but did not exhibit longer roots under 100 µM boric acid (i.e., sufficient B) supply. Among 22 mutant candidates, two mutant lines, numbers 19 (tmn1-1) and 45 (tmn1-2), were investigated in this study. Isolated mutants 19 and 45 were crossed with Landsberg erecta (Ler) and genetic mapping was conducted. The whole genome of each mutant was re-sequenced using the HiSeq2000 platform (Illumina, USA). The data were analysed using the DNA Data Bank of Japan Read Annotation Pipeline (Nagasaki et al., 2013) and the candidates responsible for mutations were found.

Single-nucleotide polymorphisms in AtTMN1 were detected by dCAPS and CAPS markers for tmn1-1 and tmn1-2, respectively. DNA fragments of the AtTMN1 sequence were amplified by PCR using Primer 1 (P1) and P2 for tmn1-1, and P3 and P4 for tmn1-2 (Supplementary Table S5). The resulting fragments were then digested with TaqI for tmn1-1 and MboI for tmn1-2.

tmn1-3 (WiscDsLoxHS217_03C), a T-DNA insertion line of AtTMN1, was originated from Col; the seeds were obtained from the Arabidopsis Biological Resource Center (USA). From the segregating population, plants homozygous for T-DNA were established by PCR using P5 and P6 for T-DNA insertions, and P6 and P7 for the WT genome (Supplementary Table S5).

Detection of the full-length transcript of AtTMN1

Full-length transcripts of AtTMN1 were detected by reverse transcription PCR. Plants were grown in solid media containing 100 μM boric acid for 11 d. Total RNA was extracted from leaves and roots using the RNeasy Plant Mini Kit (Qiagen, Germany). cDNA was synthesized using Prime Script RT Enzyme Mix (Takara, Japan). cDNA of AtTMN1 and Actin1 (cDNA quality control) were amplified using KOD-Plus-Neo (Toyobo, Japan) with specific primers P8 and P9 for AtTMN1, and P10 and P11 for Actin1 (Supplementary Table S5). cDNA from 5 ng (for AtTMN1 and Actin1 in rosette leaves), 10 ng (for AtTMN1 in roots), and 1 ng (for Actin1 in roots) of total RNA was used as template; 40 cycles of amplification were performed.

Measurement of cell lengths and cell numbers in roots

To obtain fully elongated cortical cell length measurements and meristematic cell counts, the root tips of plants grown on solid media were stained and observed. Under severely low B (0.1 μM), primary roots (PR) of 4-d-old plants were used, in which Col-0 had shorter roots than the tmn1 mutants, but in which root tip cells had not severely collapsed. PRs were collected directly from solid media and stained with 10 μg ml^–1 propidium iodide (PI; FUJIFILM Wako Pure Chemical, Japan) for 15 min, then rinsed twice in ultrapure water. Under normal B (100 μM), 11-d-old plants were used, in which root growth inhibition was obvious in the mutants. PRs were collected directly from solid media and stained with PI as described above for 40 min. Images were obtained using a confocal laser scanning microscope (LSM510, Zeiss, Germany). The excitation and detection wavelength windows were set at 488 and >540 nm, respectively. Up to three fully elongated cortical cells were selected from each PR; their longitudinal lengths were measured using ImageJ software (https://imagej.nih.gov/ij/). The numbers of cortical cells between the quiescent centre and the first elongating cortical cell were counted in the PI-stained PRs. Counting was performed using the Cell Counter ImageJ plugin (https://imagej.nih.gov/ij/plugins/cell-counter.html).

Observation of columella cells in root tips

To examine root morphology, PR tips were stained as described above, with a slight modification to avoid root cap detachment during sampling. Plants were grown on solid media containing 100 μM boric acid for 14 d. PRs were cut; several drops of ultrapure water were placed on PRs while they remained on the media to ensure columella cells remained intact. Floating PRs were stained with PI for 5 min and observed by confocal laser scanning microscopy (LSM510, Zeiss, Germany).

Plasmid construction and plant transformation

In preparation for the complementation test and observation of the sub-cellular localization of GFP-AtTMN1 protein, we generated new transgenic lines carrying proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome referring to a design described by Gao et al., (2012). This construct was to express GFP-AtTMN1 under the control of the endogenous AtTMN1 promoter. Genomic nucleotides 3657117–3664330 of chromosome 1 sequences consisting of the putative promoter region (2.1 kb), ORF (4.3 kb), and terminator (0.7 kb), were amplified using Prime STAR HS DNA Polymerase (Takara, Japan) with primers P12 and P13 (Supplementary Table S5). The fragments were cloned into the pENTR/D-TOPO vector using Gateway technology (Invitrogen, USA), and the plasmid was named pAH2. To construct proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome, three PCR fragments were assembled. The first fragment, which included a sequence encoding the AtTMN1 signal peptide (SP), was amplified from pAH2 using PCR, with primers P14 and P15 (Supplementary Table S5). The second fragment, which included GFP, was amplified using PCR, with primers P16 and P17 (Supplementary Table S5). The third fragment, which included AtTMN1 coding sequences, was amplified from pAH2 using PCR, with primers P18 and P19 (Supplementary Table S5). The three fragments were mixed and then amplified using primers P14 and P19 (Supplementary Table S5). The sequences encoding GGGGS between SP (26 aa) and GFP, and GGGSGGGS between GFP and AtTMN1 (27 aa) were inserted as linkers using the underlined primer sequences shown in Supplementary Table S5. The resulting fragments were digested with ScaI and BamHI; they were then ligated into the pAH2 vector, which had been treated with both enzymes. The resulting plasmid, proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome in pENTR/D-TOPO, was named pAH3. The insert in pAH3 was cloned into the destination vector pMDC99 (Curtis and Grossniklaus, 2003) by means of an LR reaction using Gateway technology (Invitrogen, USA), and the plasmid was named pAH6. pAH6 was introduced into Agrobacterium tumefaciens strain GV3101 (C58C1Rif^r) pMP90 (Gm^r), and tmn1-1 plants were transformed using the floral dipping method (Clough and Bent, 1998). proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome was confirmed to be functional by testing the restoration of root growth inhibition of tmn1-1 in the T₂ generation under 0.1 μM B. Among 30 independent transgenic lines, 20 exhibited full or partial complementation. For detailed analysis, a T₃ generation homozygous for T-DNA was established and four independent lines were used: pAH6-90-2, pAH6-100-1 (#1), pAH6-102-3, and pAH6-151-2 (#2).

Preparation of cell wall fractions from Arabidopsis rosette leaves and roots

For cell wall property analysis, alcohol-insoluble residues (AIRs) were prepared as previously described (Matsunaga and Ishii, 2006). Rosette leaves and roots were harvested from 44-d-old plants and then frozen. For each analysis, the following number of plants were collected as a single sample: B concentration, six for rosette leaves and four to nine for roots; ratio of dRG-II-B to total RG-II, six each for rosette leaves and roots; 2-keto-3-deoxy monosaccharide concentration, three to six for rosette leaves and six to 12 for roots; monosaccharide composition, six for rosette leaves. The frozen materials were homogenized in 80% (v/v) ethanol. After the homogenates had been centrifuged, all insoluble pellets were washed with 80% (v/v) ethanol twice, 99.5% (v/v) ethanol once, chloroform/methanol (1:1, v/v) twice, acetone once, and ultrapure water twice. The AIRs were freeze-dried with FDU-1200 (EYELA, Japan) and treated as cell wall fractions. For monosaccharide composition analysis, rosette leaves were homogenized in 80% (v/v) ethanol. After the homogenates had been centrifuged, insoluble debris were washed with 80% (v/v) ethanol twice, 80% (v/v) ethanol overnight, 99.5% (v/v) ethanol once, chloroform/methanol (1:1, v/v) three times, acetone twice, and ultrapure water twice. The AIRs were freeze-dried as described above.

B and Ca concentration measurements

To measure total B concentrations, rosette leaves and roots of 45-d-old plants were harvested and rinsed with ultrapure water. The harvested rosette leaves and roots were dried at 60 °C for longer than 2 d and the dry weights were measured. After the samples had been dried, they were submerged in concentrated HNO₃ (60–61%) at 22–24 °C for 2 d, digested at 110 °C, then completely digested with H₂O₂ (30–35.5%) at 80 °C. The digested samples were dissolved in 2% HNO₃. B concentrations were measured by inductively coupled plasma mass spectrometry (ELAN 6100 DRC-e, PerkinElmer, USA). To measure B and Ca concentrations in cell walls in rosette leaves and roots, 0.7–2.3 mg of AIR samples were submerged in concentrated HNO₃ at 22–24 °C for 2–4 d. The samples were digested as described above and then dissolved in 2% HNO₃. B and Ca concentrations were then determined by inductively coupled plasma mass spectrometry.

Determination of RG-II-B dimer formation in cell walls

To determine the ratio of dRG-II-B (RG-II-B dimer) to total RG-II in rosette leaf and root cell walls, the RG-II dimers and monomers were analysed as previously described (Matsunaga and Ishii, 2006) with slight modifications; 2.0–2.2 mg and 1.0–2.1 mg of AIRs were used for rosette leaves and roots, respectively. AIRs were saponified with 300 μl of 0.1 M NaOH at 4 °C for 4 h to remove methyl and acetyl esters; the supernatant pH was then adjusted to 5.0 with 10% (v/v) acetic acid. The suspensions were treated with six units (U) of endo-polygalacturonase (EPG) M1 for 24 h at 4 °C to release RG-II from AIRs. Before use, EPG from Aspergillus niger (Megazyme, Ireland) was dialysed with 0.1 M sodium acetate buffer. After the suspensions had been centrifuged, the supernatants were filtered through 0.45 μm membranes and subjected to size-exclusion HPLC/refractive index detection (HITACHI High Technologies Corporation, Japan) using a Diol-120 column (8 mm × 300 mm, YMC Co., Kyoto, Japan). The analyses were performed as follows: eluent, 0.2 M ammonium formate (pH 6.5); flow rate, 1.0 ml min^–1; and injection volume, 10 μl or 100 μl for rosette leaves and 100 µl, 150 µl, or 200 μl for roots. Relative proportions of dRG-II-B to total RG-II were calculated from the peak area of dRG-II-B and RG-II monomer.

Determination of 2-keto-3-deoxy sugars in cell walls

To estimate the quantities of RG-II in rosette leaf and root cell walls, 2-keto-3-deoxy (RG-II specific) sugars were measured using a modified thiobarbituric acid method (York et al., 1985). Approximately 2.0 mg and 1.5 mg of AIRs were used for rosette leaves and roots, respectively. AIRs were saponified with 190.5 μl of 0.1 M NaOH at 4 °C for 8 h; the supernatant pH was adjusted to 5.5 with 10% (v/v) acetic acid. Cell walls were treated with 5 U of EPG M2 from Aspergillus aculeatus (Megazyme, Ireland) for 89 h at 35 °C for complete digestion. Before use, the enzyme was dialysed with 0.1 M sodium acetate buffer. The suspensions were centrifuged and the supernatants were collected. This process was repeated three times to completely remove insoluble residues. The supernatant (200 μl) was mixed with 100 μl of 0.5 M H₂SO₄, then incubated for 30 min at 100 °C to hydrolyse polysaccharides into monosaccharides. The solutions were cooled for 10 min at 22–24 °C. Following this, 250 µl of 40 mM HIO₄ dissolved in 62.5 mM H₂SO₄, was added to the solution, which was then incubated for 20 min at 22–24 °C to generate formylpyruvic acid from oxidized 2-keto-3-deoxy sugars. Subsequently, 500 µl of 2% Na₂SO₃ dissolved in 0.5 M HCl, was added to the solutions to neutralize excess HIO₄. Then 500 µl of 25 mM thiobarbituric acid was added to the solutions, which were then incubated for 15 min at 100 °C to generate pigments. DMSO (99.5%, 1 ml) was added to the solutions, which were then incubated for 6–7 min at 22–24 °C to stabilize the pigments. The 548 nm absorbance of the pigments was measured using a spectrophotometer (U-3900/3900H, HITACHI High Technologies Corp., Japan). The quantities of 2-keto-3-deoxy sugars were calculated using 2-keto-3-deoxyoctonate ammonium salt (Sigma-Aldrich, USA) as a standard.

Observation of pectin distribution in cross-sections of roots

To observe pectin distribution, the PRs were embedded in LR White resin (Nisshin EM Corp., Japan); pectin was visualized with an indirect immunofluorescence method. Plants were grown on solid media containing 0.1 µM and 100 μM boric acid for 5 d. The PR tips were cut and then immersed with 4% paraformaldehyde dissolved in 20 mM sodium cacodylate buffer (pH 7.4) for 2 h at 4 ºC to fix cellular structures. The roots were immersed with 50% ethanol for 30 min at 4 ºC to be dehydrated. The process was performed with 60, 70, 80, and 90% ethanol for 30 min at each step, followed by 95% ethanol for 21 h at 4 ºC. Finally, the roots were immersed with absolute ethanol for 1 h at 4 ºC. Then the roots were infiltrated with 50% and 100% LR White resin for 22.5 and 47 h, respectively, at 4 ºC. After transferring the roots into 100% LR White resin in 1.5 ml plastic tubes, the resins were incubated for 32 h at 60 ºC to polymerize completely. Serial root sections (0.99 μm) were prepared using an ultramicrotome (ULTRACUT N, Reichert-Nissei, Germany) and placed on coated glass slides (MAS-01, Matsunami Glass Industry, Japan ). Section samples were treated with 2 N HCl for 15 min at 22–24 °C for epitope activation and then washed three times with phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, pH 7.4; FUJIFILM Wako Pure Chemical, Japan) for 42–6 (anti-RG-II antibody; Zhou et al., 2018) or T/Ca/S buffer (20 mM Tris-HCl, pH 8.2, 0.5 mM CaCl₂, 150 mM NaCl) for 2F4 (anti-Ca²⁺ cross-linked HG antibody; PlantProbes, UK). For blocking, the sections were incubated with 10% normal goat serum (Thermo Fisher Scientific, USA) for 1 h at 22–24 °C and were then labelled with primary antibodies: 20 μg ml^–1 anti-RG-II antibody, 42–6, and 1/10 anti-Ca²⁺ cross-linked HG antibody, 2F4 dissolved in 10% normal goat serum. They were incubated for 20 h at 4 ºC. After washing with the buffers three times, the sections were incubated with 20 μg ml^–1 (1/100) goat anti-rabbit IgG (H+L) and 20 μg ml^–1 (1/100) goat anti-mouse IgG1 antibodies conjugated to Alexa Fluor 488 (Thermo Fisher Scientific, USA) dissolved in 10% normal goat serum against 42–6 and 2F4, respectively, for 2 h at 22–24 °C. Following washing with the buffers three times, the sections were mounted with 50% (v/v) glycerol to be observed with a fluorescence microscope (DM2500, Leica, Germany). The Alexa Fluor 488 was excited with a mercury fluorescence source (EBQ 100-04-L, Leistungselektronik JENA GmbH, Germany) and observed through a 515 nm long-pass emission filter.

Cell wall monosaccharide composition analysis

To determine the monosaccharide composition of cell walls, extracted rosette leaf cell walls were analysed using an ultra-performance liquid chromatography–p-aminobenzyl ethyl ester system (Sakamoto et al., 2015). Approximately 2.00–3.00 mg of AIR were added to a 2 ml microtube, and starch contained in the AIR was digested with 1 ml of an amylase solution containing 500 U ml^–1 of α-amylase from porcine pancreas (Megazyme, Ireland) and 0.33 U ml^–1 of amyloglucosidase from Aspergillus niger (Megazyme, Ireland) in 0.1 M sodium malate buffer (pH 6.0) at 37 °C for 18 h. Insoluble residues in the amylase suspension were rinsed with absolute ethanol and ultrapure water, then dried completely at 60 °C overnight. The dried pellet was depolymerized with 50 μl of 72% (w/w) H₂SO₄ for 1 h with shaking at 1700 rpm. After the addition of 1.4 ml of ultrapure water to the AIR suspension, the AIR was hydrolysed at 121 °C for 1 h. The hydrolysed supernatant was neutralized with calcium carbonate powder and adjusted to around pH 5.0. Monomerized sugars in the neutralized supernatant were labelled with aminobenzyl ethyl ester solution containing 330 mg ml^–1 of p-aminobenzyl ethyl ester (FUJIFILM Wako Pure Chemical, Japan), 66 mg ml^–1 of sodium cyanoborohydride, 8% (v/v) acetic acid, and 75% (v/v) methanol at 80 °C for 30 min. Chromatographic separation and detection were conducted using an ACQUITY UPLC H-Class system equipped with an ACQUITY UPLC BEH C18 column (100 mm × 2.0 mm, 1.7 μm particle size, Waters Corp., USA) and fluorescence detector (ACQUITY UPLC FLR Detector, Waters Corp., USA) as previously described (Sakamoto et al., 2015). For samples without amylase treatment, AIR powder was hydrolysed with H₂SO₄; monosaccharide composition in hydrolysate was analysed as described above.

Observation of GFP-AtTMN1 localization

To examine the molecular function of AtTMN1 protein, GFP-AtTMN1 localization was observed in the presence of exocytosis and endocytosis inhibitors using a confocal laser scanning microscope (TCS SP5, Leica, Germany). Transgenic plants carrying proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome were grown for 5 d on solid media supplemented with 100 μM boric acid. In each treatment, chemicals were dissolved in liquid media containing 100 μM boric acid. Roots were incubated with 2 μM FM4-64 (Thermo Fisher Scientific, USA) for 5 min and then with 50 μM cycloheximide (Sigma-Aldrich, USA; a protein synthesis inhibitor) for 30 min. Subsequently, roots were treated with 50 μM brefeldin A (BFA; Sigma-Aldrich, USA; an exocytosis inhibitor) and cycloheximide for 60 min. The liquid media for BFA treatment contained 1.2% DMSO. To examine the possibility of AtTMN1 localization in the plasma membrane, 5-d-old transgenic plants grown under 100 μM B were treated with 100 μM Dynasore (Sigma-Aldrich, USA; an endocytosis inhibitor) for 90 min and then with 1 μM FM4-64 (Thermo Fisher Scientific, USA) for 30 s. The Dynasore liquid medium contained 1.7% DMSO. The laser excitation/detection wavelength bandwidths were 488/650–700 nm for FM4-64 and 488/500–530 nm for GFP.

To examine cell-type specific expression of AtTMN1 protein in roots, transgenic plants carrying proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome were grown for 5 d on solid media supplemented with 100 μM boric acid. PRs were collected directly from solid media and stained with 10 μg ml^–1 PI for 1 min, then rinsed twice in ultrapure water. GFP and PI fluorescence in PRs was observed by the confocal laser scanning microscope. The laser excitation/detection wavelength bandwidths were 488/500–530 nm for GFP and 488/650–700 nm for PI. Their fluorescence was enhanced using Leica Application Suite Advanced Fluorescence Lite (2.6.0 build 7266; Leica, Germany).

Co-expression analysis

Co-expression analysis was performed using publicly available transcriptome data including many tissues from various parts of plants (NCBI SRA Accession no.: PRJNA268115 [Klepikova et al., 2015]; PRJNA314076 [Klepikova et al., 2016]); leaf treated with abiotic stress (NCBI SRA Accession no.: PRJNA324514 [Klepikova et al., 2016]), and many cell types of root (NCBI SRA Accession no.: PRJNA323955 [Li et al., 2016]). Raw data was mapped to Arabidopsis TAIR10 CDS dataset by STAR software (Dobin et al., 2013) and normalized globally by DESeq2 software (Love et al., 2014) with default setting. Co-expressed genes showing Pearson’s correlation value r>0.6 with AtTMN1 were collected (488 genes; Supplementary Table S3). Enrichment analysis was performed by binomial test of R statistical software.

Statistical analyses

Statistical analyses were performed using R software ver. 4.0.2. PR lengths were compared among plant lines using the Tukey–Kramer test. Comparison between Col-0 and tmn1 mutants in the other experiments were performed using Dunnett’s multiple comparison test.

Results

AtTMN1 mutations altered primary root growth in response to B nutrition

To isolate novel components of RG-II deposition or dimerization, Arabidopsis Col-0 seeds were mutagenized with ethyl methanesulfonate; 22 mutant candidates with an altered response to B were isolated, which showed longer roots only under low B. Two independent mutants (nos.19 and 45) showed alleviation of root growth inhibition under low B, but exhibited reduced root growth under normal B conditions compared with WT Col-0 (Fig. 1A). The PR lengths of mutants nos.19 and 45 were 3.2-fold greater than the length of WT Col-0 under severely low (0.1 μM) B at 11 d after sowing (DAS; Fig. 1B). Under normal (100 μM) B, the PR lengths of mutants nos.19 and 45 were reduced by 13.4% and 20.4%, respectively, compared with Col-0; this indicated that B-dependent root growth was impaired in both mutants (Fig. 1B).

Figure 1.

Impaired B nutrient response caused by loss of AtTMN1 function. (A) Eleven DAS seedlings of Col-0, tmn1-1 (no.19), tmn1-2 (no.45), tmn1-3, and tmn1-1 carrying proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome (lines #1 and #2) grown on solid media supplemented with 0.1 µM and 100 μM boric acid. Scale bars =20 mm. (B) Primary root lengths of seedlings 11 DAS under 0.1 µM and 100 μM B. Values are means ± SD of 16–20 individual plants. Different letters indicate significant differences among plant lines for each B condition (P<0.05, Tukey–Kramer test). (C) Gene structure of AtTMN1 and mutation positions in tmn1 alleles. Blue and white boxes indicate exons and untranslated regions, respectively. Black lines across exons indicate introns. Triangle indicates T-DNA insertion. Numbers correspond to the position of the genomic sequence from a transcriptional start site. (D) Insertion region of the fifth intron (81 bp) in tmn1-2. Uppercase and lowercase letters indicate exons and introns, respectively. A bold red letter represents a single-base substitution in tmn1-2. Numbers correspond to the position of the coding sequence from a translation start site. (E) AtTMN1 transcripts (1.77 kb) amplified from cDNA in rosette leaves and roots using a AtTMN1-specific primer. Actin1 full-length transcripts (1.69 kb) were amplified as a reference. M, DNA size marker.

In the F₂ population derived from F₁ by crossing the mutants and WT Ler, 19.7% and 14.8% of F₂ plants exhibited the mutant phenotype (i.e. alleviation of inhibition of root elongation under low B) in mutants nos.19 and 45, respectively (Supplementary Table S1). Based on the segregation ratio, the phenotype appears to have been caused by a recessive mutation in a single genetic locus. Using a map-based cloning approach and genome analysis of the mutants, we found that both mutants carried a single-base substitution (G to A), but in a different position in Transmembrane nine 1 (AtTMN1/EMP12; AT1G10950), which encodes a membrane protein of unknown function (Fig. 1C). AtTMN1 contains a long N-terminal domain followed by nine transmembrane domains, and is localized at the Golgi apparatus (Gao et al., 2012). However, its mutant phenotypes have not been reported, and its physiological functions remain unknown.

Mutant no. 19 (tmn1-1) possesses a non-sense mutation in the 10th exon, which results in W355stop. Mutant no.45 (tmn1-2) carries a mutation in the 5’ splice donor site in the fifth intron (Fig. 1D). tmn1-3, a line carrying a T-DNA insertion in the eighth intron of AtTMN1, showed similar growth to tmn1-1 and tmn1-2 under 0.1 µM B and 100 μM B (Fig. 1A, B), indicating that AtTMN1 mutations were responsible for the phenotype.

To examine AtTMN1 mRNA in the tmn1 mutants, full-length transcripts of AtTMN1 were investigated using reverse transcription PCR. In rosette leaves, AtTMN1 cDNA was detected in tmn1-1 and tmn1-2, but not in tmn1-3 (Fig. 1E). The detected PCR product in tmn1-1 corresponded with the size of the AtTMN1 PCR product in Col-0, whereas the size of the major product in tmn1-2 was slightly larger. Sequence analysis showed that the fifth intron (81 bp), where the mutation was present at the 5’ splice donor site, was not spliced out; it remained within tmn1-2 cDNA, resulting in a premature stop codon (Fig. 1D). Because proteins that are potentially translated from tmn1-1 and tmn1-2 mRNAs lack a large portion of the transmembrane domain at the C-terminus, functional AtTMN1 proteins are unlikely to be produced in tmn1-1 and tmn1-2 leaves. No transcript was detected in tmn1 mutant roots (Fig. 1E), indicating that AtTMN1 mRNA expression was below the detection limit in all three mutants, and that non-sense-mediated mRNA decay was induced in tmn1-1 and tmn1-2 in roots. Collectively, these results suggest that the tmn1 mutants were loss-of-function mutants and that the mutant phenotypes were caused by loss of AtTMN1 function.

To confirm that the loss of AtTMN1 function was responsible for the mutant phenotypes, we conducted a complementation test by introducing a proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome construct into tmn1-1 (no. 19), thereby expressing GFP-AtTMN1 fusion protein under the control of the native AtTMN1 promoter. When T₃ homozygous lines (#1 and #2) were grown under 0.1 and 100 μM B conditions, abnormal root growth in tmn1-1 (no. 19) was rescued, further supporting the notion that impaired B-dependent growth was caused by the loss of AtTMN1 function (Fig. 1A, B).

According to an Arabidopsis Electronic Fluorescent Pictograph Browser (Winter et al., 2007), AtTMN1 mRNA was universally detected in roots, rosette leaves, internodes and siliques during vegetative and reproductive growths. Publicly available transcriptome data suggests that AtTMN1 is predominantly expressed in the mature xylem pole and cortex in roots (Supplementary Table S3). To examine cell-type expression of AtTMN1 in roots, GFP-AtTMN1 fluorescence was observed. GFP-AtTMN1 fluorescence was entirely detected in PR tips and weakly observed in the transition and mature zone where root hairs initiated. The fluorescence at the stele was relatively strong in any zone of the roots (Supplementary Fig. S1A, B, C, D). Furthermore, in PR tips, GFP-AtTMN1 was observed at the columella root cap; the intensity at a proximal end of root cap (Dolan et al., 1993; Kumar and Iyer-Pascuzzi, 2020) tended to be higher compared with the other columella cells of distinct stages (Supplementary Fig. S1A, B). These results support that AtTMN1 is involved in root growth.

Root cell elongation was altered in tmn1 mutants

To understand which process was altered in tmn1 mutant root growth, we measured the lengths of elongated cells in the root hair zone and counted cells in the meristematic zone at four DAS under 0.1 μM B and at 11 DAS under 100 μM B. Because incubation for longer than 4 d dramatically deformed root cell morphology under severely low B (0.1 μM), especially in Col-0, we observed the root cells at four DAS. Under 0.1 μM B, longitudinal lengths of fully elongated cortical cells were significantly increased by 31.2% and 27.3% in tmn1-1 and tmn1-2 (P<0.001; Fig. 2A, B), respectively, compared with Col-0. Conversely, under 100 μM B, cell lengths were significantly reduced by 11.0% and 10.6% in tmn1-1 and tmn1-2, respectively (P<0.05; Fig. 2C, D). No significant differences in root meristem cell numbers were observed between Col-0 and the tmn1 mutants under 0.1 µM or 100 μM B treatments (P>0.05; Fig. 2E–H). Our observations suggested that cell division was not primarily changed, but that cell elongation was affected in tmn1 mutants under low and normal B conditions. Changes in cell length were positively correlated with changes in PR length in the tmn1 mutants (Figs 1, 2); this suggested that the abnormal root growth observed in tmn1 mutants was caused by changes in cell elongation.

Figure 2.

Increased and reduced cell elongation in the tmn1 mutants under severely low and normal B conditions, respectively. (A, C) Representative confocal images of mature root cells stained with PI and (B, D) longitudinal lengths of fully elongated cortical cells of Col-0, tmn1-1, and tmn1-2 at four DAS under 0.1 μM B (A, B) and at 11 DAS under 100 μM B (C, D). Values are means ± SD from 29–67 independent cells in 10–19 individual plants. (E, G) Representative confocal images of root meristematic cells stained with PI and (F, H) cortical cell numbers between the quiescent centre and the first elongating cortical cell in seedlings of Col-0, tmn1-1, and tmn1-2 at four DAS under 0.1 μM B (E, F) and at 11 DAS under 100 μM B (G, H). Values are means ± SD from six to 15 individual plants. Scale bars =50 μm. Asterisks indicate significant differences between Col-0 and the tmn1 mutants (*P<0.05, ***P<0.001; Dunnett’s multiple comparison test).

The root tip morphology of seedlings grown in solid media under 100 μM B at 14 DAS showed that detaching root cap layers remained suspended from root tips in the tmn1 mutants, but not in Col-0 (Supplementary Fig. S2). This finding suggested a defect in the process of root cap maturation in tmn1 mutants, in addition to altered cell elongation.

Leaf and root dry weights were reduced in tmn1 mutants in hydroponic culture

To investigate long-term vegetative growth in the tmn1 mutants, plants were grown hydroponically under short days. Rosette leaf sizes were markedly reduced in tmn1 mutants, compared with Col-0, under all B treatments (Fig. 3A). Roots were longer in the tmn1 mutants than in Col-0 under severely low B (0.1 μM), whereas the roots of tmn1 mutants were shorter than the roots of Col-0 under 0.3 µM (mildly low), 100 µM (normal), and 500 μM (high) B conditions (Fig. 3B); these findings were consistent with mutant root phenotypes in solid media (Fig. 1A, B). Root bundles of the tmn1 mutants appeared thinner than those of Col-0 under all B treatments (Fig. 3B). Thickness of individual root tips in tmn1 mutants seemed to be reduced compared with Col-0 under 0.1 µM and 100 μM B (Supplementary Fig. S3). Leaf and root dry weights were significantly reduced by 53.0–76.0% and 47.3–73.8% in the tmn1 mutants under all B treatments including 0.1 μM B (P<0.01), when longer PRs were observed in the tmn1 mutants than in Col-0 (Fig. 3C, D). This result indicated that overall biomass production in the vegetative stage was inhibited in the tmn1 mutants, despite the presence of increased root elongation under severely low B treatment in the tmn1 mutants.

Figure 3.

Defective vegetative growth in the tmn1 mutants. Rosette leaves (A) and roots (B) of Col-0, tmn1-1, tmn1-2, and tmn1-3 plants grown hydroponically for 44 d under 0.1 µM, 0.3 µM, 100 µM, and 500 μM B treatments under short days. Scale bars =20 mm. Dry weights of rosette leaves (C) and roots (D) in Col-0, tmn1-1, tmn1-2, and tmn1-3 grown for 45 d. Values are means ± SD from four independent plants. Asterisks indicate significant differences between Col-0 and the tmn1 mutants under each B treatment (**P<0.01, ***P<0.001; Dunnett’s multiple comparison test).

To examine reproductive growth, plants were grown under long days. Although different growth responses to B deficiency by day length have been classically described (Warington, 1933), B-deficient symptoms in rosette leaf expansion and root elongation of Col-0 under low B did not appear to be largely affected by day length under our experimental regime. Similar to the vegetative growth results, the tmn1 mutants displayed shorter roots under 0.3 µM (mildly low) B to 500 μM (high) B treatments, but longer roots under 0.1 μM (severely low) B treatment, compared with Col-0 (Supplementary Fig. S4C). Inhibited internode elongation and sterility were observed in Col-0 under severely low (0.1 µM) B and mildly low (0.3 μM) B treatments, whereas these disorders were partially recovered in the tmn1 mutants, compared with Col-0 (Supplementary Fig. S4A, B), which suggested that B-deficiency symptoms were relieved in the tmn1 mutants. However, main stem length and branch numbers were reduced in the tmn1 mutants, compared with Col-0, under 100 µM (normal) B and 500 μM (high) B treatments (Supplementary Fig. S4A). Together, these results showed that AtTMN1 is necessary for normal growth in the vegetative and reproductive stages, irrespective of B conditions; moreover, AtTMN1 mutations increase growth in specific tissues under low B conditions.

Cell wall B concentrations were lower in rosette leaves and roots of the tmn1 mutants

To explore the mechanisms of impaired growth in the tmn1 mutants, total B concentration per dry weight was determined in plants hydroponically grown under short days. In rosette leaves, no significant differences in total B concentration were observed between Col-0 and the tmn1 mutants under 0.1 µM, 0.3 µM, or 100 μM B treatments (P>0.05; Supplementary Fig. S5A). Under 500 μM B, leaf total B concentrations were significantly higher in tmn1-1 compared with Col-0 (P<0.01), but not in tmn1-2 or tmn1-3 (P>0.05; Supplementary Fig. S5A). Because we observed no consistent changes in B concentrations in mutant leaves, changes in total B concentration among leaves were presumably not a primary cause of the biomass reduction observed in Fig. 3C. In roots, total B concentration significantly decreased in tmn1-3 under 0.1 μM B, tmn1-2 under 0.3 μM B, and all three tmn1 mutants under 100 μM B, compared with Col-0 (P<0.05; Supplementary Fig. S5B). No significant differences were observed between Col-0 and the tmn1 mutants under 500 μM B (P>0.05; Supplementary Fig. S5B). Reduced B concentrations in roots under normal B conditions may have caused the reduction in root growth observed in the tmn1 mutants (Fig. 3D); however, reductions in root biomass observed under 500 μM B is not likely to be explained by changes in total B concentration in roots.

Because a primary physiological function of B in plants is pectic polysaccharide RG-II crosslinking in cell walls (O’Neill et al., 2004), we next examined the AtTMN1 function in cell wall B concentration in plants hydroponically grown under short days. In rosette leaves, no significant differences in cell wall B concentration were observed between Col-0 and the tmn1 mutants under 0.1 μM B (P>0.05; Fig. 4A). However, B concentrations were significantly (20.4–43.2%) lower in the tmn1 mutants than in Col-0 in leaf cell walls under 0.3 µM, 100 µM, and 500 μM B treatments (P<0.01; Fig. 4A). In root cell walls, there were no significant differences between Col-0 and the tmn1 mutants under 0.1 μM B (P>0.05; Fig. 4B). Under 0.3 μM B, B concentrations in root cell walls were significantly lower in tmn1-1 (P<0.05); they also showed lower tendencies in tmn1-2 and tmn1-3 (Fig. 4B). Root cell wall B concentrations were significantly reduced by 24.9–36.3% in the tmn1 mutants, compared with Col-0, under 100 µM and 500 μM B (P<0.01; Fig. 4B). Because B is predominately distributed to RG-II and nearly all RG-II is presumed to be crosslinked by borate in Col-0 under 100 µM B and 500 μM B treatments, the reduction of cell wall B concentration suggested that reduced quantities of borate-crosslinked RG-II were present in the tmn1 mutants.

Figure 4.

Reduced cell wall B concentrations in tmn1 mutants under mildly low to high B conditions. B concentrations in rosette leaf (A) and root (B) cell walls of Col-0, tmn1-1, tmn1-2, and tmn1-3 plants grown hydroponically under 0.1 µM, 0.3 µM, 100 µM, and 500 μM B treatments for 44 d under short days. For each cell wall sample, six individual plants for rosette leaves and four to nine plants for roots were harvested and homogenized. Values are means ± SD from three and four independent cell wall samples for rosette leaves and roots, respectively. Asterisks indicate significant differences between Col-0 and the tmn1 mutants under each B condition (*P<0.05, **P<0.01, ***P<0.001; Dunnett’s multiple comparison test). AIR, alcohol-insoluble residue.

Calcium is another element in cell walls and it functions to cross-link de-methylesterified HG chains, a major domain of pectic polysaccharides. To verify the effects of AtTMN1 on Ca concentrations in cell walls, we analysed cell wall Ca concentrations. Rosette leaf cell wall Ca concentrations did not differ greatly between the tmn1 mutants and Col-0 (Supplementary Fig. S6A). Although significant increments were observed in tmn1-1 under 0.1 µM, 0.3 µM, and 500 μM B and in tmn1-2 under 500 μM B (P<0.05; Supplementary Fig. S6A), these changes were not consistently observed in the leaf cell walls of tmn1 mutants. In root cell walls, the tmn1 mutants showed significant reductions in Ca concentration, compared with Col-0, under 0.1 µM and 0.3 μM B (P<0.05; Supplementary Fig. S6B). Under 100 µM and 500 μM B, there were no significant differences between Col-0 and the tmn1 mutants (P>0.05; Supplementary Fig. S6B). In immunostaining with 2F4 which recognizes Ca²⁺ cross-linked HG, the signal at the lamella of cortex periphery tended to be decreased in 5-d-old tmn1-1 and tmn1-2 grown on solid media under 0.1 μM B compared with Col-0, suggesting potential reduction of Ca²⁺ cross-linked HG in the mutants under low B (Supplementary Fig. S6C).

Unlike the observed reduction in cell wall B concentrations, consistent patterns of changes in cell wall Ca concentrations were not found in the tmn1 mutants, which suggested that the AtTMN1 mutations predominately caused the observed reduction in cell wall B concentrations. Thus, reduced borate-cross-linked RG-II could be involved in the development of these impaired growth patterns.

Cell wall RG-II was reduced in tmn1 mutant rosette leaves and roots

We considered two explanations for the reduction of borate-dimerized RG-II based on lowered cell wall B concentrations in tmn1 mutants (Fig. 4): (i) reduced efficiency of dimeric RG-II-B formation, and (ii) diminished total quantity of cell wall RG-II in the tmn1 mutants. To test the first possibility, we determined the relative proportion of dRG-II-B to the proportion of total RG-II in cell walls of plants hydroponically grown under short days. In leaves, relative rates of RG-II cross-linking in cell walls were significantly higher in the tmn1 mutants than in Col-0 under 0.1 μM B (P<0.05; Table 1). Under 0.3 µM, 100 µM, and 500 μM B, there were no significant differences in cross-linking rates between leaves of Col-0 and leaves of tmn1 mutants (P>0.05; Table 1). In roots, cross-linking rates were significantly higher in tmn1-3 under 0.1 μM B and in the three mutants under 0.3 μM B, compared with Col-0 (P<0.05; Table 1). Under 100 µM and 500 μM B, there were no significant differences between Col-0 and the tmn1 mutants (P>0.05; Table 1). The lack of reduction in the relative proportions of dRG-II-B to total RG-II in the mutant cell wall likely refutes the hypothesis of lowered efficiency in dRG-II-B formation in the tmn1 mutants.

Table 1.

No reduction in the relative proportion of RG-II-B dimer formation in the tmn1 mutants.

	Relative proportion of RG-II-B dimer in rosette leaf (%)				Relative proportion of RG-II-B dimer in root (%)
Plant line	0.1 μM B in media	0.3 μM B in media	100 μM B in media	500 μM B in media	0.1 μM B in media	0.3 μM B in media	100 μM B in media	500 μM B in media
Col-0	78.0±3.5	85.9±1.5	91.1±5.5	93.8±1.7	62.1±4.2	72.2±3.2	92.5±0.4	90.4±4.1
tmn1-1	86.8±3.8*	84.6±3.7	87.8±2.0	91.3±2.2	71.3±5.8	81.7±1.9**	90.7±2.6	95.0±0.4
tmn1-2	85.1±1.8*	89.0±0.6	89.9±2.7	91.9±2.1	73.8±4.5	83.5±3.2**	92.9±3.3	93.3±3.7
tmn1-3	85.6±2.3*	89.5	91.9±2.2	91.9±3.6	75.3±6.2*	82.8±1.2**	90.9±2.1	92.7±0.3

Relative proportions of RG-II-B dimer in rosette leaf and root cell walls of Col-0, tmn1-1, tmn1-2 and tmn1-3. The plants were grown hydroponically under 0.1 µM, 0.3 µM, 100 µM, and 500 μM B for 44 d under short days. For one individual cell wall sample, six plants for rosette leaves and roots were harvested to be homogenized. RG-II-B dimers and RG-II monomers released from AIRs by EPG were detected by size-exclusion HPLC/refractive index detector. Relative proportions represent percentages of RG-II-B dimer peak area which is divided by total area of RG-II-B dimers and RG-II monomers. Values represent means ± SD from three and four independent cell wall samples for rosette leaves and roots, respectively. The value of tmn1-3 rosette leaf under 0.3 μM B represents a mean from two independent cell wall samples. Significant differences between Col-0 and the tmn1 mutants under each B condition are indicated as *P<0.05, **P<0.01 (Dunnett’s multiple comparison test). AIR, alcohol-insoluble residue.

To determine whether total quantities of RG-II were reduced in the tmn1 mutants, 2-keto-3-deoxy sugars were measured in cell walls. In plants, 2-keto-3-deoxy sugars, (2-keto-3-deoxy-d-lyxo-heptulosaric acid and 3-deoxy-d-manno-oct-2-ulosonic acid) have been found only in RG-II; thus, their quantities represent the entire quantity of RG-II. The concentration of 2-keto-3-deoxy sugars in leaf cell walls were significantly reduced by 19.9–32.7% in the tmn1 mutants, compared with Col-0, under all B treatments (P<0.001; Fig. 5A). Similarly, the concentration of 2-keto-3-deoxy sugars in root cell walls were significantly reduced by 25.5–32.8% in the tmn1 mutants, compared with Col-0, under all B treatments (P<0.01; Fig. 5B). The reduced amount of 2-keto-3-deoxy sugars in cell walls indicates that total quantities of cell wall RG-II are diminished in both rosette leaves and roots of the tmn1 mutants compared with Col-0, irrespective of B conditions.

Figure 5.

Reduced cell wall RG-II content in the tmn1 mutants. Concentration of 2-keto-3-deoxy sugars in rosette leaf (A) and root (B) cell walls of Col-0, tmn1-1, tmn1-2, and tmn1-3 plants grown hydroponically under 0.1 µM, 0.3 µM, 100 µM, and 500 μM B treatments for 44 d under short days. For one individual cell wall sample, rosette leaves of three to six plants and roots of six to 12 plants were harvested for homogenization. RG-II was solubilized from AIRs by EPG treatment and 2-keto-3-deoxy sugar concentrations were measured using a modified thiobarbituric acid method. Values are means ± SD from four independent cell wall samples. Asterisks indicate significant differences between Col-0 and the tmn1 mutants under each B condition (**P<0.01, ***P<0.001; Dunnett’s multiple comparison test). The concentration of 2-keto-3-deoxy sugars was estimated based on the standard curve of 2-keto-3-deoxyoctonate ammonium salt. AIR, alcohol-insoluble residue. Immunohistochemistry of root cross-sections in the meristematic (C) and elongation zones (D) of PRs with an anti-RG-II antibody, 42–6. Plants were grown on solid media containing 100 μM boric acid for 5 d. Scale bars =50 μm.

To examine RG-II distribution, RG-II was visualized in cross-sections of PRs grown on solid media by labelling with an RG-II antibody. In meristematic zone of roots, signal intensities in the stele and the cortex periphery were decreased in the tmn1 mutants compared with Col-0, whereas there were no apparent differences in the epidermis (Fig. 5C). In elongation zone of the roots, the signals in the cross-sections were entirely reduced in the tmn1 mutants compared to Col-0 (Fig. 5D). This observation supports that RG-II amounts were reduced in both the meristematic and elongation zones of the roots in the tmn1 mutants.

These results demonstrate that AtTMN1 is necessary for maintenance of the quantity of RG-II in cell walls. Because these reductions are consistent with the reductions of dry weight under all B conditions, it is highly likely that reduced quantities of RG-II, which cause a reduction in borate-dimerized RG-II, are at least partly responsible for the biomass reductions and altered growth responses observed in the tmn1 mutants under severely low B conditions.

Quantities of cell wall RG-I were also reduced in the rosette leaves of tmn1 mutants

To investigate whether cell wall components other than RG-II were affected by the lack of AtTMN1, the quantities of 10 monosaccharides were determined in leaf cell walls of plants grown with 100 μM B under short days. Regardless of the presence of amylase treatment, rhamnose content was significantly reduced by 21.2–27.3% in the tmn1 mutants, compared with Col-0 (P<0.001; Table 2, Supplementary Table S2). The quantity of GalA (a substantial component of HG) was significantly reduced by 6.5% in tmn1-1 cell walls treated with amylase, and by 14.8% and 17.3% in tmn1-1 and tmn1-2 cell walls without amylase treatment, respectively (P<0.05 ; Table 2, Supplementary Table S2). Glucose, xylose, mannose and glucuronic acid contents were significantly increased in the tmn1 mutants, compared with Col-0, in cell walls treated with amylase (P<0.05; Table 2, Supplementary Table S2). Because rhamnose in cell walls is largely derived from RG-I, these reductions in rhamnose contents suggested that quantities of RG-I were also reduced in the cell walls of the tmn1 mutants. Additionally, quantities of HG may have decreased slightly in the tmn1 mutants. This observation suggests that AtTMN1 functions in the deposition of pectic polysaccharides in cell walls, especially RG-II and RG-I.

Table 2.

Reduced rhamnose contents in tmn1 mutant cell walls treated with amylase.

Plant line	Mol% of monosaccharides in rosette leaf cell wall
	Rha	GalA	Glc	Xyl	Man	Ara	Gal	Fuc	mGlcA	GlcA
Col-0	7.75±0.11	41.6±0.67	24.2±0.60	5.84±0.07	1.91±0.05	9.41±0.23	7.63±0.08	1.23±0.02	0.27±0.02	0.14±0.01
tmn1-1	5.65±0.14***	38.9±1.12*	27.9±0.74***	6.26±0.17**	2.16±0.12**	9.73±0.22	7.48±0.21	1.41±0.05***	0.33±0.06	0.20±0.02***
tmn1-2	5.83±0.13***	40.1±1.75	27.7±0.59***	6.22±0.15**	2.13±0.10*	8.99±0.54	7.16±0.37	1.33±0.08*	0.32±0.04	0.18±0.01*
tmn1-3	5.80±0.14***	40.9±1.65	27.4±0.57***	6.27±0.08**	2.09±0.07*	8.77±0.61	7.04±0.27*	1.30±0.04	0.32±0.18	0.17±0.01*

The 10 monosaccharides contents were determined in rosette leaf cell walls treated with amylase. The molar percentages were calculated by dividing with total detected monosaccharide contents. Plants were grown hydroponically under 100 μM B for 44 d under short days. Rosette leaves of six plants were harvested and homogenized to create one independent cell wall sample. After amylase treatment, the quantities of 10 monosaccharides were determined. Data are means ± SD from four independent rosette leaf cell walls treated with amylase in Col-0, tmn1-1, tmn1-2, and tmn1-3. Asterisks indicate significant differences between Col-0 and the tmn1 mutants (*P<0.05, **P<0.01, ***P<0.001; Dunnett’s multiple comparison test). Rha, l-rhamnose; GalA, d-galacturonic acid; Glc, d-glucose; Xyl, d-xylose; Man, d-mannose; Ara, l-arabinose; Gal, d-galactose; Fuc, l-fucose; m-GlcA, 4-O-methyl-d-glucuronic acid; GlcA, d-glucuronic acid.

GFP-AtTMN1 BFA responses were largely similar to those of Golgi proteins

It has been reported that AtTMN1 is localized in the Golgi apparatus in Arabidopsis (Gao et al., 2012). RG-I and RG-II structures are synthesized at the Golgi apparatus by a variety of glycosyltransferases (Driouich et al., 2012); packaged pectic polysaccharides are then delivered into the apoplast matrix via two trans-Golgi network derived pathways (Sinclair et al., 2018). Therefore, we hypothesized that AtTMN1 functions in pectin biosynthesis in the Golgi apparatus and/or in secretion to the apoplast. To gain insight into the molecular function of AtTMN1, GFP-AtTMN1 localization was observed in the presence of exocytosis and endocytosis inhibitors. Under treatment with 50 μM BFA (an exocytosis inhibitor) accompanied by cycloheximide for 60 min, ring-like localization of GFP-AtTMN1 was mainly observed on the BFA compartment (Robinson et al., 2008) of FM4-64 (an endocytosis tracer; Fig. 6A). The distribution pattern of GFP-AtTMN1 was similar to the pattern of ST-mRFP (Naramoto et al., 2014), confirming the localization of GFP-AtTMN1 mainly in the Golgi apparatus. Furthermore, under 90 min treatment with 100 μM Dynasore (a dynamin-dependent endocytosis inhibitor), GFP-AtTMN1 fluorescence signals did not show co-localization with FM4-64 on the plasma membrane (Fig. 6B). This suggested the likelihood that most of the GFP-AtTMN1 was neither localized to trafficking systems nor targeted to the plasma membrane. From these observations of GFP-AtTMN1 localization, we inferred that AtTMN1 plays a role in pectin biosynthesis in the Golgi apparatus, rather than in trafficking pectin to the apoplast.

Figure 6.

Golgi localization of GFP-AtTMN1 in the presence of BFA and Dynasore. (A) Sub-cellular localization of GFP-AtTMN1 under BFA treatment. Five-d-old transgenic plants carrying proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome (#1) under 100 μM B were incubated with 2 μM FM4-64 for 5 min and then with 50 μM cycloheximide for 30 min, followed by 50 μM BFA and cycloheximide for 60 min. Arrowheads indicate ring-like signals of GFP-AtTMN1. (B) Sub-cellular localization of GFP-AtTMN1 under Dynasore treatment. Five-d-old proAtTMN1:SP(AtTMN1)-GFP-AtTMN1genome transgenic plants (#1) under 100 μM B conditions were treated with 100 μM Dynasore for 90 min and then with 1 μM FM4-64 for 30 s. Scale bars =10 μm. Similar patterns were observed in the four independent transgenic lines.

Furthermore, genes co-expressed with AtTMN1 (r>0.6; 488 genes; Supplementary Table S3) include at least 20 published genes proven to be related to pectin biosynthesis and modification (Supplementary Table S3). Enrichment analysis revealed that terms related to the Golgi apparatus, pectin, and primary cell wall were over-represented among the co-expressed genes (Supplementary Table S4). These data support our conclusion that AtTMN1 is involved in the pectin deposition pathway.

Discussion

AtTMN1 plays a role in pectic polysaccharide deposition in cell walls

In this study, AtTMN1 was identified as a novel component required for pectin deposition, which governs cell elongation and consequently overall plant growth. We screened Arabidopsis mutants with impaired growth responses to low B nutrition, based on prior observations that deficits in RG-II synthesis lead to altered responses to B nutrition (O’Neill et al., 2001; Sechet et al., 2018). In the tmn1 mutants, quantities of RG-II, represented by 2-keto-3-deoxy sugar concentrations, and RG-I, represented by rhamnose contents, were decreased by 20–30% in leaf and root cell walls under all B treatments (Fig. 5; Table 2), which demonstrated that AtTMN1 was essential for normal pectin deposition in cell walls.

Considering the importance of RG-II and RG-I in cell growth (O’Neill et al., 2001; Oomen et al., 2002; Delmas et al., 2008; Kobayashi et al., 2011; Liu et al., 2011), these results suggested that reductions of RG-II and RG-I contents were the main causes of cell elongation inhibition (Fig. 2C, D). Meanwhile, the tmn1 mutants exhibited no apparent abnormalities of cell division (Fig. 2E-H) despite the reduced RG-II amounts in the meristematic zone of the PRs (Fig. 5C), implying that the RG-II reduction in the tmn1 mutants was not so severe as to cause defects in cell division. This view is supported by the observation that reduced B-crosslinked RG-II under B deficiency primarily impairs cell elongation (Camacho-Cristóbal et al., 2015). Thus, it is assumed that the cell elongation inhibition rather than the defect of cell division resulted in whole-plant growth and developmental impairment in the tmn1 mutants. This is demonstrated by a reduction of dry weight in the vegetative growth stages under all tested B conditions, as well as impaired reproductive growth under sufficient B supply (100 µM and 500 μM; Fig. 3; Supplementary Fig. S4). The assumption is supported by the prior observation that a 20% reduction in borate dimerized RG-II likely caused dwarfism in Arabidopsis mur1 and GDP-d-mannose 3,5-epimerase-silenced tomato plants with impaired RG-II structure (O’Neill et al., 2001; Voxeur et al., 2011).

Altered pectin affects growth under limited B conditions

Under severely low B, the tmn1 mutants displayed alleviation of root elongation inhibition and reproductive growth failure compared with WT plants (which exhibited reduced growth) (Figs 1, 3Supplementary Fig. S4). In roots, this phenomenon was caused by increment of cell elongation in the tmn1 mutants (Fig. 2A, B), in contrast to inhibition of cell elongation under B sufficiency (Fig. 2C, D). Cell elongation is generally inhibited under low B conditions, and pollen tube elongation is suppressed by B depletion, resulting in plant sterility (Dell and Huang, 1997; Huang et al., 2000; Wang et al., 2003). Therefore, we conclude that the alleviation of cell elongation inhibition in the corresponding tissues explained the tmn1 mutant phenotypes under severely low B conditions.

Cell enlargement in specific tissues of the tmn1 mutants, compared with Col-0, may have been caused by reduction of overall pectin content, through cell wall loosening under severely limited B, when cell elongation was severely inhibited in Col-0. In roots, reduction of Ca²⁺ cross-linked HG (Supplementary Fig. S6C) could cause cell wall loosening and result in increased cell elongation in the mutants under severely low B. However, based on the assumption that RG-II cross-linking by borate is a primary determinant for growth under low B, reduced RG-II content might be a major contributing factor in the increment of cell elongation in the tmn1 mutants under limited B supply. In mutant cell walls under severely low B, B concentrations and total RG-II content were reduced, compared with WT plants (Figs 4, 5); this suggested that absolute quantities of dRG-II-B and monomeric RG-II had decreased. Considering that B deficiency inhibits cell elongation in WT through the reduction of dRG-II-B and increase in monomeric RG-II contents compared with plants under sufficient B supply, the reduction in monomeric RG-II, but not in dRG-II-B, in the tmn1 mutants likely resulted in the alleviation of cell elongation inhibition under severely low B conditions. This hypothesis is consistent with the findings in a previous study, where differences in tolerance to low B among rapeseed genotypes were reportedly defined by low quantities of pectin, presumably in the form of monomer RG-II (Zhou et al., 2017). Our results further suggest that altered pectin, mainly RG-II, modifies plant growth under limited B supply.

Several homologs of the Catharanthus roseus receptor-like kinase subfamily have been shown to play regulatory roles in cell expansion by binding cell wall polysaccharides or glycosylated proteins (Nissen et al., 2016). These proteins (e.g., FER, a receptor kinase that interacts with the pectin polysaccharide backbone) may sense the disruption of pectin cross-linking by B and Ca²⁺ (Feng et al., 2018); these proteins may perceive monomeric RG-II and modulate cell elongation under limited B conditions.

The TMN protein family shares cell adhesion functions among eukaryotes

TMN proteins are conserved as the transmembrane 9 superfamily (TM9SF) among eukaryotes; some of these proteins have been characterized in organisms other than plants. Down-regulation of human TM9SF4, which belongs to the other cluster from AtTMN1 in the phylogenetic tree separated into two clusters (Hegelund et al., 2010), has been shown to reduce adhesion of myelomonocytic cells to fibronectin (Paolillo et al., 2015). In Drosophila, a null mutant of putative phagocytic receptor 1a (Phg1a; an ortholog of TM9SF4) showed defective phagocytosis against wasp eggs (Bergeret et al., 2008). Given that haemolymph cells can recognize and attach to invaders, the Drosophila phg1a mutant may lose the capacity for cellular attachment under these conditions.

In Dictyostelium, a unicellular organism, the number of cells adhering to hydrophilic glass surfaces was reduced in a Phg1A-knockout mutant (Cornillon et al., 2000). Similarly, the triple mutant S. cerevisiae tmn1 tmn2 tmn3 displayed reduced adhesion to solid yeast extract–peptone–dextrose medium surfaces and a concurrent filamentous growth defect (Froquet et al., 2008), such that single yeast cells assembled into a multicellular form in response to nitrogen starvation (Mösch and Fink, 1997; Cullen and Sprague, 2012). Together, these data show that TMN family members are essential for intercellular attachment among eukaryotes, including both multicellular and unicellular organisms, although the regulation systems and molecules involved in cell attachment are distinct among organisms.

The AtTMN1 mutations impaired the detachment of the proximal end of root cap, consistent with its expression (Supplementary Figs S1, S2). The previous studies on loss-of-function mutants in pectin biosynthesis and degradation demonstrated that proper pectin metabolism is required for root cap maturation (Durand et al., 2009; Bennett et al., 2010; Kamiya et al., 2016). Although the tmn1 phenotype does not agree with a hypothesis that reduced amounts of pectin including HG promotes the root cap detachment, it is suggested that Arabidopsis TMN1 contributes to appropriate intercellular attachment of columella cells through proper pectin deposition. Considering that normal pectin is required for plant cell adhesion (Bouton et al., 2002; Iwai et al., 2002), it is believed that AtTMN1 in plants also contributes to the biological processes involved in cellular attachment mediated by pectin deposition.

AtTMN1 might be involved in vesicular trafficking for pectin biosynthesis

Based on analyses of TMN knockout mutants and overexpression lines, it is suggested that TMN families control the amounts of cargo molecules at the plasma membrane in Dictyostelium, Drosophila and human (Froquet et al., 2012; Perrin et al., 2015a, b). Furthermore, molecular interactions between the cargo molecules and TMN proteins suggest that TMN proteins function as cargo receptors in the vesicular trafficking pathway to secrete them into the cell surface.

In the current study, we found that Arabidopsis TMN1 controls RG-II and RG-I deposition in the apoplast and that a major portion of TMN1 displayed the pattern of Golgi-localized proteins. The cytoplasmic tail of the C-terminus in AtTMN1 possesses the KXD/E motif, a Golgi retention signal, which is conserved among other TMN homologs in eukaryotes (Gao et al., 2014). Furthermore, AtTMN1 interacts with Sec21 (a coat protein complex I subunit) and Sec24 (a coat protein complex II subunit) to mediate retrograde and anterograde transport, respectively, between the Golgi apparatus and endoplasmic reticulum (Gao et al., 2012); thus, AtTMN1 is presumably involved in a vesicular trafficking system. Given the Golgi localization of a series of proteins, including glycosyltransferases and nucleotide sugar transporters required for pectin biosynthesis (Mohnen, 2008; Nikolovski et al., 2012), and the interaction of AtTMN1 with coat protein complex subunits, AtTMN1 may transport these proteins required for RG-II and RG-I synthesis from the endoplasmic reticulum to the Golgi. On the other hand, we do not rule out a possibility that Arabidopsis TMN1 is involved in trafficking of biosynthesized pectin from the Golgi to apoplast, considering that ECHIDNA, YPT/RAB GTPase interacting Protein 4a and b are localized on the trans-Golgi network but not plasma membrane; they function in secretion of HG, RG-I and xyloglucan (Gendre et al., 2013; McFarlane et al., 2013). Therefore, it is possible that a minor portion of AtTMN1 localized at the trans-Golgi network contributes to pectin transport.

In conclusion, the results of this study demonstrate that the plant TMN family plays roles in RG-II and RG-I pectic polysaccharide deposition for normal cell elongation. We also found that the pectin synthesis mutant exhibited reduced sensitivity to low B supply in terms of cell elongation. Our findings shed light on the mechanisms of growth regulation, which are dependent on changes in pectin content under low B conditions.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Expression of GFP-AtTMN1 in root tips.

Fig. S2. Failed cellular detachment of root caps in the tmn1 mutants under sufficient B condition.

Fig. S3. Reduction tendency of root tip thickness in tmn1-1 and tmn1-2.

Fig. S4. Restoration of tmn1 mutant reproductive growth under severely low B condition.

Fig. S5. No substantial reduction in total B concentration in the tmn1 mutants under sufficient B conditions.

Fig. S6. No consistent patterns of changes in cell wall Ca concentrations of the tmn1 mutants.

Table S1. F₂ segregating population derived from F₁ of mutants nos.19 or 45 crossed with Ler.

Table S2. Reduced rhamnose and galacturonic acid contents in tmn1 mutant cell walls without amylase treatment.

Table S3. Genes co-expressed with AtTMN1.

Table S4. Enrichment analysis results for genes co-expressed with AtTMN1.

Table S5. Primers used in this study.

Abbreviations

Abbreviations
 
• AIR

  alcohol-insoluble residue

 
• B

  boron

 
• BFA

  brefeldin A

 
• DAS

  day after sowing

 
• GFP

  green fluorescent protein

 
• HG

  homogalacturonan

 
• Phg1a

  putative phagocytic receptor 1a

 
• PI

  propidium iodide

 
• PR

  primary root

 
• RG-I

  rhamnogalacturonan I

 
• RG-II

  rhamnogalacturonan II

 
• TMN1

  transmembrane nine 1

 
• TM9SF

  transmembrane 9 superfamily

Acknowledgements

We thank Makoto Kimura and Michiko Tsukamoto for technical assistance. We are grateful to Hiroko Yamamoto (Hokkaido University) for helpful instruction on equipment. We give special thanks to Toshiaki Ito and Dr. Masanori Yasui (Hokkaido University) for technical assistance with an ultramicrotome and a confocal laser scanning microscope at Electron Microscope Laboratory in Research Faculty of Agriculture. We thank Dr. Toshihiro Watanabe (Hokkaido University) for assistance in ICP-MS and Dr. Hironori Takasaki (Saitama University) for downloading and mapping of RNA-seq data used in co-expression analysis. We thank Dr. Masaru Kobayashi for providing the RG-II antibody. We acknowledge that this work was supported by Japan Society for the Promotion of Science (JSPS) Grants-in-Aid for Scientific Research Grant Number 15H01222,18H05490,19H05637 to K.M. and the Advanced Low Carbon Technology Research and Development Program (ALCA) of the Japan Science and Technology Agency to N.M. (grant no. JPMJAL1107).

Author contributions

A.H. and K.M. designed the research; A.H., S.S. and K.M. conducted the experiments and analysed the data; N.M. performed co-expression analysis; A.H. and K.M. wrote the article with contributions from all the authors.

Data availability

All data supporting the findings of this study are available within the paper and within its supplementary data published online.