Callose deposition assays
For analysis of callose deposition after chitosan treatment, material from tomato and Arabidopsis plants with different concentrations of chitosan were collected 1 day after treatment (dat) and placed in 96% (v/v) ethanol in order to destain leaves. Aniline blue was used to stain callose deposits as described previously (Luna et al., 2011). Analysis of callose associated with the infection by B. cinerea in tomato leaves was performed as described (Rejeb et al., 2018) with some modifications. Briefly, infected tomato leaf samples were collected and placed in 96% (v/v) ethanol 1 day after infection with B. cinerea and allowed to destain. Destained material was hydrated with 0.07 M phosphate buffer (pH 9.0) for 30 min and then incubated for 15 min in 0.1% (w/v) aniline blue (Sigma-Aldrich) and 0.005% (w/v) fluorescent brightener (Sigma-Aldrich). Solutions were then replaced with 0.1% (w/v) aniline blue and incubated for 24h in the dark prior to microscopic analysis. All observations were performed using an UV-epifluorescence microscope (GXM-l2800 with GXCAM HiChrome-MET camera). Callose was quantified from digital photographs by the number of yellow pixels (callose intensity). Infection-associated callose was scored and analysed in a similar way but callose intensity was expressed relative to fungal lesion diameters. Image analyses were performed with Photoshop CS5 and ImageJ.

Chitosan antifungal activityin vitroassay

B. cinerea mycelial growth assessment was performed using Potato Dextrose Agar (PDA) as culture media with different concentrations of chitosan (1%, 0.1%, 0.01% w/v). PDA was autoclaved and then chitosan and the fungicide Switch (as positive fungicide control) (1%, 0.1%, 0.01% w/v) were added directly to PDA as it cooled. One 5 mm diameter agar plugs of actively growing B. cinerea mycelium was added per plate. Five plates per treatment were sealed with parafilm and then incubated under controlled conditions (darkness and 24°C). After 4 days, the mean growth of the fungus was determined by measuring two perpendicular diameters and calculating the mean diameter.

High-Pressure Liquid Chromatography (HPLC) - Mass Spectrometry (MS)

Healthy and infected tomato leaf tissues were harvested in liquid nitrogen and subsequently freeze-dried for 3 days. Freeze-dried samples were ground in 15 mL Falcon tubes containing a tungsten ball in a bead beater. Ten mg of each sample was used for hormone extraction. Sample extraction, HPLC-MS quantitative analysis of plant hormones and data analysis were performed as described (Forcat et al., 2008). Accurate quantification of ABA, SA and JA used the deuterated internal standards added during sample extraction (Forcat et al., 2008) and concentrations were calculated using standard concentration curves. Relative accumulation of jasmonic acid-isoleucine (JA-Ile) were obtained by calculations of % peak areas among samples.

Transcriptome analysis

Four conditions were analysed using microarrays: (i) ddH2O-treated and non-infected plants (Water + Mock); (ii) Chitosan-treated and non-infected plants (Chitosan + Mock); (iii)ddH2O-treated and B. cinerea -infected plants (Water + B. cinerea ); (iv) Chitosan-treated and B. cinerea -infected plants (Chitosan + B. cinerea ). Inoculations were performed four days after treatment (dat) with chitosan, and leaf discs from four independent plants (biological replicates) per treatment were sampled at 6, 9 and 12 h post-inoculation (hpi) with mock orB. cinerea spores. Total RNA was extracted with an RNeasy Plant Mini Kit (Qiagen) as recommended. A custom 60-mer oligonucleotide microarray was designed using eArray (https://earray.chem.agilent.com/earray/; A-MTAB-667 and E-MTAB-8868;www.ebi.ac.uk/arrayexpress/) from predicted transcripts (34,616 in total) of the S. lycopersicum (ITAG 2.3) genome. Experimental design is detailed at E-MTAB-8868;www.ebi.ac.uk/arrayexpress/. Two-channel microarray processing was utilised, according to the Low Input Quick Amp Labelling Protocol v. 6.5 (Agilent). Microarray images were imported into Feature Extraction software (v. 10.7.3.1; Agilent) and data extracted using default parameters. Data were subsequently imported into Genespring software (v. 7.3; Agilent) for subsequent pre-processing and statistical analysis. Following Lowess normalisation, data were re-imported as single-colour data. Data were filtered to remove probes that did not have detectable signal in at least 3 replicates, leaving 22,381 probes for statistical analysis.
Analysis of Variance (2-way ANOVA; p-value ≤0.01, Benjamini-Hochberg false discovery rate correction) was used to identify differentially expressed genes (DEGs) for the factors ‘Treatment’ (3,713 DEGs), ‘Time’ (6,920), and ‘Treatment-Time interaction’ (186). Subsequently, pairwise Student’s T-tests were performed (Volcano plots: P-value ≤0.05, 2-fold cut-off) on the global set of 8,471 DEGs for each of the three test treatments (Chitosan + Mock, Water + Mock and Chitosan + B. cinerea ) compared to control (Water + Mock) at each time point. Venn diagrams were used at each time point to identify common and specific DEGs.
Panther gene ontology (GO) term enrichment analysis
Panther software (Thomas et al., 2003) was used to visualise DEG products in the context of biological pathways and/or molecular functions, using default settings. Functional enrichment analysis was performed using DEG lists for Chitosan + B. cinerea and Water +B. cinerea treatments at 6 hpi. ‘Biological processes’ and ‘molecular functions’ were selected using PANTHER Overrepresentation Test (release 20170413) against S. lycopersicum (all genes in database) and Bonferroni correction for multiple testing.
DEG transcript co-expression analysis
Two-way ANOVA was performed on the filtered microarray dataset at increased stringency (p-value ≤0.01, Bonferroni false discovery rate correction) to identify 1,722 highly-significant DEGs. Pearson’s correlation was used with default settings in Genespring (v 7.3) to generate a heatmap to help identify co-expressed transcripts (Fig 3b).
Gene expression analysis
Validation of S. lycopersicum transcriptomic analysis was performed by qRT-PCR of nine candidate differentially expressed genes (DEGs), comparing gene expression values with microarray. RNA samples were DNAse-treated with TurboDnase (ThermoFisher) and complementary DNA (cDNA) was synthesized from 2.5 µg total RNA using Superscript III reverse transcriptase (Invitrogen) as recommended with random hexamer/oligo dT primers. RT-qPCR reactions were performed with specificS. lycopersicum oligonucleotide primers (Table S4) purchased from Sigma-Aldrich. Gene primers and probes were designed using Universal Probe Library (UPL) assay design centre (Roche Diagnostics Ltd.). RT-qPCR was performed using FastStart Universal Probe Master Mix (Roche) and expression was calculated against two reference genes (SlActin-like and SlUbiquitin ) using the Pfaffl method (Pfaffl, 2001).