Balancing trade-offs between biotic and abiotic stress responses through leaf age-dependent variation in stress hormone cross-talk

Significance Plants are exposed to conflicting stresses simultaneously in nature. As stress responses are costly, plants likely coordinate these responses to minimize fitness costs. The nature and extent to which plants employ inducible mechanisms to cope with combined physical and biological stresses remains unknown. We identify a genetic mechanism by which leaves of distinct ages differentially control stress-response cross-talk. At the organism level, this mechanism balances stress-response trade-offs to maintain plant growth and reproduction during combined stresses. We also show that this leaf age-dependent stress-response prioritization influences the establishment of plant-associated leaf bacterial communities. This study illustrates the importance of active balancing of stress-response trade-offs for plant fitness maintenance and for interaction with the plant microbiota.

frozen in liquid nitrogen, and stored at -80°C. Total genomic DNA was extracted and Hpa biomass was quantified by qPCR as described previously (11).

Performance assay
To measure performance under single salt stress, seedlings were grown vertically for seven days on ½ MS plates supplemented with 1% sucrose (0.8% agar) at 22°C with a 10-h light period. Seedlings were then transferred to new ½ MS plates containing 1% sucrose as well as 100 mM NaCl, and grown at 22°C with a 10-h light period.
Shoot fresh weight was measured ten days after transfer. For survival rate, two-week-old plants were soildrenched with 300 mM NaCl for 14 d. Then, plants were recovered by soil drenching with water and survivors were counted seven days later. Survival rate was calculated as the percentage of survivors.
Combined Hpa and salt stress treatments were performed with 2.5-week-old plants. Two days prior to the infection with Hpa, plants were soil-drenched with 100 mM NaCl. Spore preparation and spray infection with Hpa isolate Noco2 (4 x 10 4 conidiospores/ml) were performed as described previously (10). Plants were kept with a clear cover to increase the relative humidity to 100%. Seven days after Hpa infection, plants were transferred to another chamber at 19°C with 90% relative humidity and a 16-h light period. After that, plants were soil-drenched with 50 mM NaCl, and shoot fresh weight was quantified seven days later.
Combined Pto DC3000 cor-and salt stress treatments were performed with four to five-week-old-plants.
Plants were grown in pots that were covered with a mesh to enable infection via vacuum infiltration. Two days prior to bacterial infection, plants were transferred to a 16-h light period to induce flowering and soil-drenched with 50 mM NaCl. Vacuum infiltration (OD600 = 0.0002) was performed as described previously (12). Twenty days later, the frequency of watering with water or NaCl was reduced to induce fruit maturation and the numbers of siliques per plant were counted ten days later.

Quantitative PCR
For gene expression, four to five-week-old plants were sprayed with ABA (Sigma A1049, 500 µM in 0.5% EtOH) or mock (0.5% EtOH) for 24 h or 48 h, or pre-sprayed with ABA (Sigma A1049, 500 µM in 0.5% EtOH) or mock (0.5% EtOH) for 24 h, followed by a spray with SA (500 µM) or water for 24 h. Total RNA was extracted with TriFast (peqlab, Erlangen, Germany) and cDNA was synthesized with superscript II (Life Technologies). For oomycete biomass quantification, the amount of oomycete DNA, estimated by primers against the conserved region ITS5.8S was normalized to the amount of plant DNA, estimated by primers against ACTIN2. Quantitative PCR was performed as described previously (13). The primers used in this study are presented in Dataset S3.
The log2-transformed gene expression or biomass data were normalized to ACTIN2 and fit to the following model: Ctyre = Yy + Rr + eyr or Ctgyre = GYgy + Rr + egyr, where GY, genotype:treatment interaction and random factors; R, biological replicate; e, residual.

RNA-seq
Four to five-week-old Col-0 leaves were sprayed with ABA (Sigma A1049, 500 µM, 0.5% EtOH) or mock (0.5% EtOH) for 48 h. Total RNA was extracted with the TRIzol reagent (Invitrogen). Extracted RNA was treated with DNase I (Roche) and purified using the RNeasy MinElute Cleanup Kit (Qiagen). Library preparation after PolyA enrichment was performed with NEBNext Ultra™ Directional RNA Library Prep Kit for Illumina (New England Biolabs). Construction of libraries and sequencing were done at the Max Planck-Genome-centre Cologne (http://mpgc.mpipz.mpg.de/home/). Briefly, libraries were sequenced using HiSeq v3 chemistry on a HiSeq2500 (Illumina) system. Strand-specific sequences were mapped to the A. thaliana genome (TAIR 10) using Tophat2 software with default settings. Mapped reads per library were counted using HTSeq software (14). Differentially expressed genes were determined using the edgeR and limma package (15). Only genes with ten read counts per sample on average were used for analysis. Data were normalized via trimmed mean of M-values (TMM) normalization (16) and normalized values were log2-transformed using the voom function (17). After fitting a linear model containing the parameters treatment, leaf age and replicate, differentially expressed genes were selected based on a log2 fold change > 1 and q-value < 0.01. q-values were obtained by the qvalue R package (18). Uncentered correlation clustering was done in Cluster3.0 and visualized with Java TreeView (19). The RNAseq data used in this study were deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (accession no. GSE114645).

SA measurements
Leaves of four to five-week-old plants were sprayed with ABA (Sigma A1049, 500 µM in 0.5% EtOH) or mock (0.5% EtOH) and samples were harvested 48 h later. For flg22-triggered SA accumulation, leaves were infiltrated with 10 µM flg22 and samples were harvested 9 h later. Samples were stored at -80°C before analysis. SA extraction and quantification were performed as described previously (20). The following model was fitted to metabolic data: Xgyr = GYgy + Rr + egyr, where GY, genotype:treatment interaction and random factors; R, biological replicate; e, residual.

Proline quantification
For quantification of salt stress-induced proline, four to five-week-old plants were soil-drenched with 100 mM NaCl, and leaves were harvested after five days of stress. Proline was extracted and quantified as described previously (21). The following model was fitted to metabolic data: Xgyr = GYgy + Rr + egyr, where GY, genotype:treatment interaction and random factors, R, biological replicate; e, residual.

Parallel quantification of multiple phytohormones
Multiple phytohormones were measured in the leaves of four to five-week-old plants 6 h, 12 h, 24 h, and 72 h after 100 mM NaCl soil drench treatment. Parallel quantification of stress-induced ABA, ACC, IAA, JA, OPDA, and SA levels was done as described previously (22). The following models were fitted to metabolic data: Xgyr = GYgy + Rr + egyr (Fig. 4f), Ctgytr = GYTgyt + Rr + eytr ( Supplementary Fig. 6 harvested as "old" and the youngest fully developed leaves were harvested as "young". The unplanted soil sample was taken from the middle of the pot, approximately 2 cm below the soil surface. Both leaf and soil samples were frozen in liquid nitrogen directly after harvesting. All harvested samples were stored at -80°C until further processing. DNA for bacterial 16S rRNA gene profiling was prepared as described previously (23), with a few minor changes. DNA was extracted with the FastDNA Spin Kit for Soil (MP Biomedicals) following the manufacturer's protocol with modifications: incubation from step 6 was omitted and in the final step the DNA was eluted with 50 µl of nuclease-free water (Qiagen). DNA concentration was measured with a Nanodrop™.
The final elongation step of the PCR reaction was 10 min and the amplicon library contained 5 ng of DNA per sample. Quality-controlled reads were mapped as described before (24). Shannon index of α-diversity, Principal Coordinates Analysis (PCoA) with Bray-Curtis distances and enrichment analysis on the single OTU level were performed in R on the normalised OTU table with previously published R scripts (24). All the plots were made in R with the use of the ggplot2 package (25). The bacterial 16S profiling data was deposited in the European Nucleotide Archive (accession no. PRJEB26793).

Statistical analysis
Statistical analysis was performed in R using mixed linear model function (lmer) from the package lme4 unless otherwise described. Standard errors were calculated from variance and covariance values after model fitting.
The Benjamini-Hochberg method was applied for correction of multiple testing in figures showing all pairwise comparisons of the mean estimates.

Accession numbers
The accession numbers for the genes discussed in this article are as follows: AtACTIN2 (AT3G18780), AtPR1 (AL7G25380), Hpa ITS-5.8S (EU049263.1). Fig. S1. ABA influences a subset of responses in a leaf age-dependent manner. (A) RAB18 expression levels in OL and YL of Col-0 plants treated with 500 µM ABA or mock at the indicated time points. Data represent means ± SEM of at least three biological replicates using a mixed linear model. Different letters indicate significant differences (adjusted P < 0.05). (B) Bacterial growth after ABA treatment. OL and YL of 4-5 week-old Col-0 plants were infiltrated with Pto DC3000 hrcC (OD600 = 0.0002) 24 h after ABA (500, 50, or 5 µM) spray or mock treatment. Bacterial growth was measured at 2 dpi. Data represent means ± SEM of two independent experiments each with at least five biological replicates using a mixed linear model. Different letters indicate significant differences (adjusted P < 0.005). (C and D) The expression changes in L7 and L12 of 4-5 week-old Col-0 plants 48 h after 500 µM ABA spray compared to mock (top and middle panels) or the expression changes in L12 compared to L7 of 4-5 week-old Col-0 plants 48 h after mock treatment (bottom panels). Expression was measured by RT-qPCR (C) or represents data from the RNAseq (D). Data represent means ± SEM of at least three biological replicates. *, P < 0.05; **, P < 0.01; two-tailed Student's t-tests.