Arabidopsis (Arabidopsis thaliana) Wassilewskija (Ws) wild type seeds were purchased from Lehle Seeds. The anp double mutants (in the Ws background) mutants were kindly provided by Patrick J. Krysan (University of Wisconsin). Generation of conditional triple mutants [amiR1 (line 2.5), and amiR3 (line 5.3)] and of plants overexpressing ANP1 (OE1, line 2.15) or ANP3 (OE3, line 3.1) has been reported previously (Savatin et al., 2014). For gene expression analysis, seeds were surface sterilized and germinated in multiwell plates containing 0.5X Murashige and Skoog medium (1 ml/well; Sigma–Aldrich) supplemented with 0.5% (w/v) sucrose. Triple mutant seedlings were grown in 0.5X MS containing β-estradiol (1 μM for wild type and amiR1, 10 nM for amiR3) or DMSO (mock; 0.01% v/v) for 9 days. The culture medium was then replaced with fresh medium containing β-estradiol at the same concentrations and seedlings were grown for further 24 h before the analysis. For phenotypical analysis of the triple mutants roots, 5-day-old seedlings of wild type, amiR1 and amiR3 grown in presence/absence of estradiol at different concentrations (0.5–1 μM for amiR1, 2.5–50 nM for amiR3) in agar plates containing 0.5X Murashige and Skoog medium supplemented with 0.5% (w/v) sucrose. In the case of ISX treatment for Jasmonate analysis, 6 day-old seedlings of Ws, anp1 anp2, anp1 anp3, anp2 anp3 or Col-0, OEANP1 and OEANP3 were grown in liquid 0.5X Murashige and Skoog medium supplemented with 1% (w/v) sucrose. Cell wall analysis was performed in roots collected from 10-day-old seedlings of wild type, amiR1 and amiR3 grown in 0.5X MS plates in the presence of 1 μM estradiol or DMSO, and wild type seedlings treated for 24 h with 20 nM ISX (N-[3-(1-ethyl-1-methylpropyl)-1,2-oxazol-5-yl]-2,6-dimethoxybenzamide-PESTANAL^®, Sigma–Aldrich). For the latter treatment, wild type seedlings were grown for 9 days in plates containing DMSO and then transferred for 24 h to medium containing 20 nM ISX.

Roots were collected, washed extensively with distilled water to remove sugars contained in the medium, weighed, frozen and milled using a Retschmill MM300 for 60 s at 18 Hz. The root powder was suspended in 1 ml of pre-warmed (70°C) 70% ethanol and left for 30 min at room temperature (25°C). Samples were centrifuged at 25000 × g for 10 min (Beckman Allegra) to pellet the Alcohol Insoluble Solid (AIS). The supernatant containing pigments were discarded. In order to eliminate proteins and nucleic acids the AIS was suspended in 1 ml chloroform:methanol (1:1) and incubated again for 30 min at room temperature. Samples were centrifuged at 25000 × g for 10 min (Beckman Allegra) and supernatant was discarded. This step was repeated twice. To remove lipids and facilitate subsequent dehydration the pellet was washed in 1 ml 80% acetone, recovered by centrifugation at 25000 × g for 5 min. This step was repeated twice. The pellet was dried O/N at room temperature under a fume hood.

The AIS material obtained as described above was weighed and transferred in 2-ml screw-capped Eppendorf tubes. Buffer 1 (100 mM phosphate buffer, 5 mM NaCl, sodium azide 0.02% w/v, pH 7.0; 500 μl) was added and samples were heated at 70°C for 10 min to gelatinize starch granules. The suspension was cooled at 4°C for 10 min and 1.2 μl of α-amylase Type I-A (Sigma–Aldrich; 5U/mg AIS) was added to each sample. To improve digestion, samples were incubated in rotary shaker at 37°C for 24 h as previously described (Zablackis et al., 1995). After 24 h, samples were centrifuged at 25000 × g for 10 min and supernatants were discarded. The de-starched AIS (ds-AIS) was washed twice with 1 ml of water, centrifuged at 25000 × g and the pellet was washed once with 80% acetone. Samples were centrifuged at 25000 × g for 5 min and supernatant was eliminated. Pellet was dried O/N at room temperature under a fume hood.

Root cell wall material was subjected to total hydrolysis (according to Saeman et al., 1954). Briefly, the de-starched AIS was weighed and treated firstly with 175 μl of 72% H₂SO₄ at 30°C for 90 min. The sample was diluted to a sulphuric acid concentration of 13% by adding 825 μl of H₂O in screw capped tubes and incubated at 121°C for 90 min. For monosaccharide composition, samples were neutralized with 0.1 M Ba(OH)₂, centrifuged twice at 8000 × g for 10 min to remove BaSO₄. Supernatants were filtered and analyzed directly by HPAEC-PAD. Sugar content of the ChASS, 0.1K-H/P and 4K-H fractions was determined according to (Dubois et al., 1956).

The de-starched AIS was incubated with the Chelating Agent Solution (CAS) [50 mM 1,2-diaminocyclohexanetetraacetic acid (CDTA), 50 mM ammonium oxalate, in 50 mM ammonium formiate, (pH 5.2)] for 2 h at 70°C. To improve the extraction, samples were vortexed every 20 min. After 2 h, samples were centrifuged at 14000 × g for 5 min. These steps were repeated twice. The supernatants obtained from these subsequent extractions were combined to obtain the ChASS fraction. To remove the CAS from the ChASS fraction, samples were precipitated with EtOH at final concentration of 30% (v/v) under continuous shaking O/N at 4°C. Samples were centrifuged at 25000 × g for 20 min and pellet was dissolved the in 100 μl of ultrapure water. Aliquots of 10 and 20 μl were used to quantify the amount of sugars contained in the ChASS fraction (Dubois et al., 1956).

The residual pellet that contains cellulose, hemicellulose and lignin was treated sub-sequentially with alkali solution (0.1N KOH and 4N KOH) to solubilize hemicellulose contained in the cell wall material as previously described (Peng et al., 2000). After the extraction the pectin/hemicellulose [0.1K-H/P (0.1N KOH)] and the hemicellulose [4K-H (4N KOH)] enriched fraction were dialyzed (vs. water), lyophilized, suspended in water and filtered. Ten and Twenty microliter were used to quantify the amount of sugars contained in these fractions as previously described (Dubois et al., 1956). The ChASS, 0.1K-H/P and 4K-H fractions were firstly hydrolyzed for 16 h with 3N methanolic HCl (Sigma–Aldrich CAS Number 7647-01-0) at 80°C and then with 2 M TFA. After evaporation of the TFA under a stream of N₂ by adding 2-propanol, samples were dissolved in distilled water and used for HPAEC-PAD analysis.

Crystalline cellulose was determined as previously described (Updegraff, 1969). The AIS prepared as described above, was treated firstly with 2 M TFA, then with Updegraff reagent (acetic acid: nitric acid: water, 8:1:2 v/v) at 100°C for 30 min. Samples were cooled on ice for 20 min and then centrifuged at 14000 × g for 10 min. The pellet was washed twice with water and once with 1 ml 80% acetone, and allowed to dry completely on the bench O/N. Dry pellets were subjected to a complete Saeman hydrolysis (Saeman et al., 1954). Glucose in the hydrolysate was analyzed by using HPAEC-PAD after H₂SO₄ neutralization with Ba(OH)₂ as described above.

Basal accumulation of H₂O₂ was analyzed in 10-day-old seedlings grown in liquid medium in the presence of 1 μM β-estradiol or DMSO, by using a colorimetric assay based on xylenol-orange as previously described (Galletti et al., 2008).

Jasmonate (JA) was extracted as previously described (Savatin et al., 2014) from 10-day-old seedlings of wild type, amiR1 and amiR3 grown in multiwells containing liquid medium in the presence of 1 μM β-estradiol (wild type and amiR1) or 10 nM β-estradiol (amiR3) or DMSO as a control and quantified using a calibration curve. Statistical differences resulted from the mean value of JA level of three different biological replicates each composed by 40 seedlings (n = 40).

ISX-induced JA was analyzed as previously described (Forcat et al., 2008). Seedlings of anp double mutants and OE1 or OE3 lines were treated with 20 nM ISX for 7 h, flash-frozen in liquid nitrogen and freeze-dried for 24 h. Extraction buffer (10% methanol, 1% acetic acid) containing internal standards (10 ng jasmonic-d5 acid; CDN Isotopes, Pointe-Claire, QC, Canada) was used for JA extraction. Supernatants obtained from two sequential extractions were combined and centrifuged to remove all debris prior to LC-MS/MS analysis. Chromatographic separation was carried out on a Shimadzu UFLC XR, equipped with a Waters Cortecs C18 column (2.7 mm, 2.1 × 100 mm). The solvent gradient (acetonitrile (ACN)/water with 0.1% formic acid each) was adapted to a total run time of 7 min: 0–4 min 20% to 95% ACN, 4–5 min 95% ACN, 5–7 min 95% to 20% ACN; flow rate 0.4 ml/min.

Ethylene produced by 10-day-old seedlings of wild type, amiR1 and amiR3 grown in 10-ml flasks containing 2 ml of 0.5X MS containing DMSO or β-estradiol (1 μM for wild type and amiR1, 10 nM for amiR3) was quantified as previously described (Brutus et al., 2010).

Gene expression analysis was performed on 10-day-old seedlings of Ws, amiR1 and amiR3 grown in the presence/absence of 1 μM β-estradiol, as previously described in Savatin et al. (2014).

Lignin detection was performed on 5-day-old seedling roots [Ws and amiR3, grown in the presence of β-estradiol (10 nM) or in its absence (DMSO)] cleared overnight in 70% EtOH. A saturated solution of phloroglucinol (in 20% HCl) (Liljegren et al., 2002) was applied on microscope slides, roots incubated for 2 min and covered with coverslips before microscope imaging.

Five-day-old seedlings of Ws, amiR1 and amiR 3 were grown on 0.5X MS in the presence/absence of β-estradiol (1 μM for Ws and amiR1, 50 nM for amiR3). Cotyledons were stained with propidium iodide (P4170- Sigma Aldrich) at a final concentration of 1 mg/ml for 2 min, washed twice with Milli-Q water and imaged with Leica SP8 confocal microscope, excited with a 514-nm laser/610-627 emission. Z-stacks were transformed into Z-projections and cell outlines analyzed by using Fiji (ImageJ).

Aniline blue stain was used to visualize callose dots in 10-day-old seedling roots of Ws, amiR1 and amiR3 grown and germinated on 0.5xMS Agar plates containing β-estradiol (1 μM). For callose staining, roots were incubated in 0.07 M sodium phosphate buffer pH 9 for 30 min and in 0.005% (w/v) aniline blue (in 0.07 M sodium phosphate buffer pH 9) for 1 h. After two washes with water, samples were analyzed under a Nikon Eclipse E800 microscope using a UV-2A filter (EX 330–380 nm, DM 400 nm, BA 420 nm). Roots were stained with calcofluor white (in 10% KOH) (18909 Sigma–Aldrich) for 2 min before imaging. For radial sections, roots have been embedded in 3% low melting agarose after calcofluor white staining and hand cut with a razor blade. Images were taken with LSCM Zeiss 800 [405 nm (EX)/450–465 nm (EM)]. Pectin staining was performed on radial sections by using the HG-specific probe COS⁴⁸⁸ (Mravec et al., 2014). Buffer (50 mM MES, pH 5.7) containing COS⁴⁸⁸ was applied on top of the sections for 15 min after which they were washed with 50 mM MES buffer (pH 5.7) prior to visualization. Images were taken with LSCM Zeiss 800, was excited with a 488-nm laser line and emission analyzed between 517 and 527 nm.

In this study we used previously described conditional anp triple mutants, in which the expression of the third gene family member was silenced in both the anp1 anp2 and anp2 anp3 double mutants, by individually expressing ANP3- and ANP1-specific artificial microRNAs (amiRNAs), respectively, under the control of a β-estradiol-inducible promoter (Savatin et al., 2014). The transgenic lines of each background, homozygous for a single insertion, when induced by β-estradiol, are hereon indicated as triple mutants amiR1 and amiR3 (line 2.5 and line 5.3, respectively). The two types of mutants respond to different concentrations of the inducer (Figure 1A), which may be due to different permeability of β-estradiol in the different mutant backgrounds. The overall phenotype of 5-day-old triple anp mutant seedlings, germinated and grown in β-estradiol (0.5/1 μM for amiR1 and 2.5/50 nM for amiR3) is shown in Figure 1A. While amiR1 control seedlings (labeled as DMSO, β-estradiol solvent) displayed a global development defect (as previously described in Krysan et al., 2002; Savatin et al., 2014), both triple anp mutants displayed reduction in primary root elongation when grown in the presence of β-estradiol compared to their corresponding controls (Figure 1A). To investigate closely, we analyzed the primary root phenotype in seedlings germinated and grown for 5 days in the presence of intermediate doses of the β-estradiol (0.5 μM for amiR1 and 2.5 nM for amiR3). Primary roots of both amiR1 and amiR3 mutant seedlings showed cell bulging in the epidermis of the transition zone (TZ) (Figure 1B, black arrows), suggesting that root length and consequently elongation might be linked to these defects. Nonetheless no alteration was detected in organization of root tip tissues compared to the wild type by using these doses of β-estradiol (Figure 1B). However, at higher doses, defects in the root phenotype were more severe and development of the entire root tip was completely compromised in both amiR lines (Figure 1C, black arrow). In the latter situation, radial swelling and tissue disorganization have been observed not only in the TZ but also in the meristem; moreover, vascular differentiation was observed in close proximity to the stem cell niche (Figure 1C). A similar phenotype was observed in triple mutant lateral roots, where, after 15 days of growth in β-estradiol, cell bulging was visible in the epidermis approximately in the same area as previously shown for triple anp mutant primary roots (Figure 1D).

We also investigated the impact of the loss of ANPs on cotyledon development, by analyzing epidermal cells in 5-day-old seedling cotyledons of Ws, amiR1 and amiR3 germinated and grown in the presence or absence of 1 μM (Ws and amiR1), or 50 nM β-estradiol (amiR3). We found that, although the lack of these proteins did not affect the number of epidermal cells (Figure 1E) thus did not significantly influence the cotyledon size, it altered the number of cell wall stubs (Figure 1F). It has been previously reported that the simultaneous loss of ANP2 and ANP3, but not of ANP1 and ANP2, strongly disturbs the formation of complete cell walls, leaving remnant cell wall stubs and multinucleated cells (Krysan et al., 2002). As expected, we observed cell wall stubs in the control DMSO-grown amiR1 (anp2 anp3) seedlings, but not in the amiR3 (anp1 anp2) ones (Figure 1F, upper panel). In the amiR1 seedlings, β-estradiol-induced ANP1 silencing enhanced the formation of cell wall stubs, which became detectable also in the β-estradiol-induced amiR3 seedlings (Figure 1F, lower panel).

Because the cell bulging phenotype of the triple mutant strongly suggests alterations in the cell walls, we investigated cell wall composition in roots. In the analysis we included wild type roots treated with isoxaben (ISX), because mild treatment with this cellulose biosynthesis inhibitor induces a strikingly similar phenotype (Supplementary Figure S1A). We also observed that 48 h of treatment with β-estradiol was sufficient to trigger the phenotypical defects of the triple anp mutants (Supplementary Figure S1B).

To perform the analysis a fraction corresponding to the de-proteinized and de-starched total cell walls (de-starched Alcohol-Insoluble Solid, ds-AIS) was prepared from whole roots of triple mutants and wild type seedlings grown in the presence of β-estradiol (1 μM) or DMSO (mock), as well as of wild type seedlings grown for 24 h in the presence of 20 nM ISX. Part of the ds-AIS was subjected to complete hydrolysis (Saeman et al., 1954) for monosaccharide composition (MSC) analysis. Both the triple mutants and ISX-treated wild type roots displayed a significant reduction in the relative amount of glucose [expressed as molar ratio (%)] compared to the corresponding untreated plants (Figure 2A). β-estradiol-treated anp triple mutant roots were also characterized by an increase of galactose (Gal) (Figure 2A). To obtain the different cell wall components, the remaining part of the ds-AIS was subjected to sequential extractions (Peng et al., 2000; Bischoff et al., 2009; Socha et al., 2014). After extracting the fraction mainly composed of pectins (named Chelating-Agent Soluble Solid or ChASS) (Figure 2B), a fraction containing both loosely bound hemicellulose and strongly bound pectins (solubilized by using KOH 0.1 M, and named 0.1K-Hemicellulose/Pectin (or 0.1K-H/P)) was extracted (Supplementary Figure S2). Finally, a fraction containing mainly hemicellulose was obtained by solubilization with 4 M KOH (named 4K-Hemicellulose or 4K-H) (Figure 2C). In the triple mutants, ChASS showed a significantly decreased relative amount of arabinose (Ara) and increased amount of galacturonic acid (GalUA); after ISX treatment, the relative content of GalUA was instead reduced (Figure 2B). The 0.1K-H/P fractions of β-estradiol treated triple mutant roots did not show any significant differences from the control (DMSO treated triple mutants) (Supplementary Figure S2A) whereas the 4K-H fractions of both the triple mutants and the ISX-treated wild type roots exhibited a significant reduction in relative amount of Xyl compared to the corresponding control seedlings (Figure 2C). No significant difference in glucose was detected in these two fractions between the triple mutants and the wild type. We then assessed whether the relative lower content of glucose observed in the ds-AIS MSC reflects a decrease content of crystalline cellulose. Indeed, the content of crystalline cellulose was reduced both in ISX-treated wild type roots and in the induced triple mutants when compared with their respective DMSO-treated controls (Figure 2D). The total abundance of pectins and loosely bound hemicelluloses/strongly bound pectins (ChASS, 0.1K-H/P) or hemicelluloses (4K H) was also examined. Both the triple mutant and the ISX-treated wild type root cell walls showed a significant reduction in the abundance of pectins contained in the ChASS fraction (Figure 2E). ISX-treated seedlings and induced amiR3 seedlings also showed an increased abundance of the 0.1K-H/P fraction, while the amiR1 mutant had an already increased abundance in the control treatment and no further increase after induction with β-estradiol (Supplementary Figure S2B). No significant decrease in the amount of hemicelluloses contained in the 4K-H fraction was observed in the triple mutants and in the ISX-treated wild type roots (Figure 2F).

In order to study the localization of the cell wall defects of the triple anp mutants, we analyzed the distribution of callose and cellulose by using aniline blue and calcofluor white, respectively (Supplementary Figures S3A,B). Aniline blue appeared to penetrate into the β-estradiol-grown triple anp mutants more easily than into the corresponding DMSO-treated controls. However, the presence of small bright dots (likely representing callose deposits) was observed only upon silencing of ANP1 or ANP3 (Supplementary Figure S3A, white arrows). Apart from the bulging phenotype typical of the anp mutants, calcofluor white staining did not reveal significant differences between the mutant and the wild type roots in terms of cellulose abundance or patterning (Supplementary Figure S3B). Indeed, the fluorescence intensity of the different stained roots was similar in all the genotypes (Ws, amiR1 and amiR3) and conditions analyzed (DMSO, β-estradiol). We also tested the HG-specific probe COS⁴⁸⁸ on the same seedling roots, to detect the cellular location of HG with different esterification states (Mravec et al., 2014). Interestingly, we found, in the triple anp mutant root sections, a reduction of demethylesterified HG, which normally localizes mostly at the corners of intercellular spaces in proximity of cells connection (Supplementary Figure S4, middle panels white arrows), compared to the DMSO-treated amiR controls.

Taken together, these results show that the composition of root cell walls in the β-estradiol-treated triple mutants is severely altered, with an altered pectin composition and a lower content of glucose, as detected in the total cell wall preparations. The latter defect may be ascribed to a reduction in crystalline cellulose content, although it was not revealed by calcofluor white, a stain that, however, binds β-glucans besides cellulose (Pradhan and Loqué, 2014) and, in any case, does not detect subtle changes in cellulose content. It is also possible that an altered organization of cellulose, rather than a decrease abundance, may account for the dramatic phenotype of the triple anp mutants.

The perception of the cell wall damage induces complex responses, hereon collectively referred as CWDS, as described in several cell wall mutants or upon cellulose biosynthesis inhibition (Caño-Delgado et al., 2000; Ellis et al., 2002; Denness et al., 2011; Tsang et al., 2011). The CWDS includes ROS accumulation, ectopic lignification of tissues and phytohormone [jasmonic acid (JA) and salicylic acid (SA)] accumulation (Denness et al., 2011). Here, we tested whether the cell wall defects occurring in plants that lack ANPs activate the CWDS, by examining extracellular hydrogen peroxide production, lignification and JA levels. Indeed, higher levels of extracellular hydrogen peroxide were produced by amiR1 and amiR3 triple mutant seedlings, germinated and grown in the presence of β-estradiol (Figure 3A). Control DMSO-treated amiR3 seedlings showed hydrogen peroxide levels that were comparable to those detected in the wild type, whereas DMSO-treated amiR1 control seedlings, which exhibit morphological defects, showed higher levels (Figure 3A), in agreement with previous results (Savatin et al., 2014). Next, the presence of lignin was analyzed in wild type and amiR3 triple mutant seedlings by staining with phloroglucinol-HCl solution. Triple mutant roots displayed a patchy distribution of staining that was undetectable in the non-induced controls (Figure 3B). The ectopic lignification occurred in proximity of the root tip. JA quantification showed that JA was undetectable in non-induced amiR3 plants and increased in the presence of β-estradiol (Figure 3C). In contrast, JA content was already high in non-induced amiR1 seedlings compared to both the wild type and non-induced amiR3 seedlings, and did not increase further upon β-estradiol treatment (Figure 3C).

Consistent with the increased JA levels observed, the expression of several JA-regulated genes, such as PLANT DEFENSIN 1.2 (PDF1.2) and ETHYLENE RESPONSIVE FACTOR1 (ERF1), was increased in the β-estradiol-induced amiR3 mutants, compared to both the non-induced controls and the wild type (Figure 3D). The expression of these genes was constitutively high in the anp2 anp3 double mutant (Supplementary Figure S5), suggesting that lack of both ANP2 and ANP3 is sufficient for the induction of the CWDS to some extent.

Expression of both PDF1.2 and ERF1 is positively regulated not only by JA, but also by ethylene (Moffat et al., 2012); however, ethylene production in the triple mutant seedlings was comparable to that detected in the β-estradiol-treated wild type seedlings (Supplementary Figure S6), suggesting that only the higher levels of JA lead to increased expression of these genes in the amiR lines. ISX treatment, in which JA induction but not ethylene accumulation occurs, was used as control for the induction of the analyzed genes (Supplementary Figure S7).

Given the similarity between the ISX-induced phenotypes and the triple anp mutants, we investigated the possibility that ANPs play a role in the cell wall integrity (CWI) maintenance pathway. Accumulation of absolute JA levels, a hallmark of the ISX response (Denness et al., 2011), was analyzed in the three combinations of anp double mutant seedlings. As shown in Figure 4A, basal JA levels and PDF1.2 expression were wild type-like in anp1 anp2, and increased to different degrees in anp1 anp3 and anp2 anp3, which are characterized by developmental defects of differing severity (Krysan et al., 2002). This suggests a positive correlation between the degree of developmental defects and JA accumulation. While anp1 anp2 showed a response to ISX similar to that of the wild type, anp1 anp3 seedlings showed an increased expression of PDF1.2, and anp2 anp3 showed both increased JA accumulation and increased PDF1.2 expression compared to the wild type (Figure 4A). We therefore checked whether, conversely, overexpression of ANP3 or ANP1 affects the ISX-induced JA levels in the opposite direction. OEANP1 and OEANP3 plants, overexpressing ANP1 and ANP3, respectively, have previously been generated to elucidate the function of these kinases in the responses to DAMPs and MAMPs (Savatin et al., 2014). They displayed wild type-like basal JA levels and PDF1.2 transcripts, but reduced accumulation of both in response to ISX, with both alterations being more pronounced in OE3 plants (Figure 4B). These results indicate that the overexpression of ANP reduces the CWD responses and suggest that these kinases play a role in CWI maintenance.