Wild type (WT) A. thaliana seeds, ecotype Col-0, were purchased from Lehle Seeds (Round Rock, TX, United States). Arabidopsis seeds were grown in pots containing commercial soil or in artificial soil (Jiffy-7, 44 mm Ø, distributed by InterGrow, Aalter) in a growth chamber at 21°C with a 16/8 h light/dark photoperiod after a 3 days stratification period at 4°C in the dark. The light intensity in the controlled growth chamber was approximately 100 μmol/m².s [Radium Spectralux plus white (58W) lamps]. Alternatively, seeds were grown in vitro, therefore seeds were surface sterilized in 70% ethanol for 2 min followed by 5% bleach for 10 min. Afterward seeds were rinsed with sterile distilled water. The sterilized Arabidopsis seeds were sown in vitro on solid MS medium (4.3 g/L MS salts with vitamins and nutrients (Duchefa), 30 g/L sucrose [Applichem), pH 5.7-5.8 (adjusted with 0.5 M NaOH) and 8 g/L plant agar (Duchefa)]. After a 3 days stratification period at 4°C in the dark, the plates were transferred to a growth chamber at 21°C with a 16/8 h light/dark photoperiod.

Wild type N. benthamiana seeds were supplied by Dr. Verne A. Sisson (Oxford Tobacco Research Station, Oxford, NC, United States). For transient transformation, the tobacco seeds were sown in pots containing commercial soil and cultivated in a growth chamber at 25°C with a 16/8 h light/dark photoperiod.

Plant samples were homogenized using a mortar and a pestle, and total RNA was extracted using TRI Reagent^® according to the instructions of the manufacturer (Sigma–Aldrich). The RNA samples were treated with DNase I (Life Technologies) according to the manufacturer’s instructions. The RNA concentration was measured with a Nanodrop 2000 spectrophotometer (Thermo Scientific). cDNA was synthesized from 1 μg of total RNA using the moloney murine leukemia virus RT and oligo(dT)₂₅ primer (Life Technologies).

The full length cDNA sequences encoding AN3, AN4, and AN5 were retrieved by RT-PCR reactions with gene specific primers (Supplementary Table 1). The PCR reaction mixture was as follows: 2 μl cDNA, 2 μl 10 mM dNTPs (Thermo Fisher Scientific), 2.5 μl 10 × RxN buffer (VWR), 1 μl 5 μM forward and 1 μl 5 μM reverse primer (Life Technologies), 0.75 μl 50 mM MgCl₂, 0.125 μl Platinum^® Pfx DNA Polymerase (Life Technologies) and water up to the volume of 25 μl. The PCR conditions used were: 2 min 95°C – 30–35 × (15 s 94°C – 30 s 47–50°C – 1 min 72°C) – 5 s 72°C. After cloning these sequences in the pJET2.1 vector with the CloneJET PCR Cloning kit (Life Technologies), the constructs were checked by agarose gel electrophoresis and sequenced (LGC Genomics, Berlin, Germany) to confirm the correct cDNA sequence.

cDNA quality was checked by RT-PCR using primers specific for the PP2A gene (Supplementary Table 2). The PCR reaction mixture was the same as previously mentioned except for the buffer (10 × EXTRA buffer, VWR) and the enzyme (Taq DNA polymerase, VWR). The PCR conditions were as follows: 5 min 95°C – 45 × (45 s 94°C – 45 s 55°C – 30 s 72°C) – 5 min 72°C. PCR amplification products were checked by agarose gel electrophoresis.

Coding sequences for the ArathNictabas were N- and C-terminally fused to EGFP using the Gateway^® Cloning Technology (Life Technologies, Carlsbad, CA, United States). The cloned full length cDNA sequences (in the pJET1.2 vector) were used as a template to amplify the open reading frames with primers to attach attB sites. In the first PCR, the first part of the attB site is attached using Platinum^® Pfx DNA Polymerase (Life Technologies) and primers with a gene specific part (with or without stop codon) and the first part of the attB site (Supplementary Table 3). In the second PCR, a 1:5 dilution of the first PCR product was used as a template in combination with primers to complete the attB sites (Supplementary Table 3). The PCR conditions for the first PCR were as follows: 2 min 94°C – 30 × (15 s 94°C – 30 s 50°C – 1 min 72°C) – 5 min 72°C. The conditions for the second PCR were: 2 min 94°C – 5 × (15 s 94°C – 30 s 48°C – 1 min 72°C) – 25 × (15 s 94°C – 30 s 55°C – 1 min 72°C) – 5 min 72°C. After checking the PCR products by agarose gel electrophoresis, the PCR fragments were used in a BP recombination reaction with the pDONR221 vector. The attB PCR products and the pDONR221 vector were incubated overnight in equimolar amounts with the BP Clonase^® II enzyme mix. The next day, the resulting entry clones were transformed into heat-shock competent Escherichia coli cells (TOP10) and transformants were selected on LB agar plates with 50 μg/mL kanamycin. Subsequently, transformants were checked with colony PCR and agarose gel electrophoresis. The entry clones were extracted using the GeneJET Plasmid Miniprep kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. After sequencing (LGC Genomics, Berlin, Germany) the entry clones were used in a LR recombination reaction with the destination vectors pK7WGF2,0 or pK7FWG2,0 to make the N-terminal and C-terminal fusion constructs with EGFP, respectively (Karimi et al., 2002). This recombination reaction was incubated overnight according to the Gateway^® manual and expression clones were transformed into E. coli TOP10 cells using heat shock. Transformants were selected on LB agar plates with 75 μg/mL spectinomycin and screened with colony PCR using gene specific and EGFP primers (Supplementary Table 3).

The expression vectors containing the different EGFP fusion constructs were introduced into Agrobacterium tumefaciens C58C1 pMP90 cells using triparental mating. Briefly, a donor strain (E. coli containing the expression vectors), a helper strain and the Agrobacterium were mixed together on solid YEB medium (5 g/L beef extract, 5 g/L peptone, 1 g/L yeast extract, 5 g/L sucrose, and 15 g/L bacterial agar) containing 2 mM MgSO₄. After incubation, dilution series were made and transformants were selected on YEB medium containing 75 μg/mL spectinomycin and 20 μg/mL gentamicin. After purification of the expression clones from the Agrobacterium, screening of the clones was done by PCR (Supplementary Table 3).

The entry clones containing the AN4 and AN5 coding sequences with stop codon (see section “Construction of EGFP Fusion Constructs”) were used in an LR reaction to generate the overexpression constructs. The destination vector pK7WG2,0 (Karimi et al., 2002) containing the 35S promoter was combined with the entry clones and the Gateway LR Clonase II to get the desired expression clones (35S::AN4 and 35S::AN5). These expression clones were heat shock transformed in TOP10 E. coli cells and transformants were selected on LB agar plates containing 75 μg/mL spectinomycin. Transformants were screened with colony PCR using a forward primer in the 35S promoter and a reverse primer in the 35S terminator sequence (Supplementary Table 4). Electrocompetent A. tumefaciens GV3101 cells were transformed with these expression clones (300 ng) using electroporation with the following parameters: 2.0 kV, 25 μF, and 200 Ω. Immediately after the pulse, YEB medium was added and the cells were grown on a shaker (200 rpm) at 28°C for 2 h. Transformants were selected on YEB agar plates with 75 μg/mL spectinomycin and screened with colony PCR using primers located in the 35S promoter and terminator sequences (Supplementary Table 4).

Transient expression of the EGFP fusion constructs was obtained by infiltration of the transformed Agrobacterium in leaves of 4- to 6-week-old N. benthamiana plants as described by Sparkes et al. (2006). First, the Agrobacterium strains were grown in liquid YEB medium containing 75 μg/mL spectinomycin and 20 μg/mL gentamicin for 2 days at 25°C on a rotary shaker (200 rpm). Agrobacterium cells were harvested by centrifugation and resuspended in infiltration medium (50 mM MES, 2 mM Na₂HPO₄, 0.5% glucose, pH 5.6). Centrifugation and resuspension were repeated twice, the second time using infiltration medium with 100 μM acetosyringone. After washing, the cells were diluted to a final optical density at 600 nm of 0.01, 0.05, 0.1, and 0.2 and infiltrated in the leaf epidermal cells. Two or three days post-infiltration, microscopic analysis was performed.

Stably transformed Arabidopsis plants were created using the floral dip transformation method (Clough and Bent, 1998). Transformed seeds were selected on MS medium containing 75 μg/mL kanamycin (Duchefa) using the fast selection protocol according to Harrison et al. (2006). Green plantlets were transferred to new selective MS medium and afterward to artificial soil. For the ArathNictaba EGFP overexpression lines integration of the T-DNA was checked by PCR on gDNA using gene specific and EGFP primers (Supplementary Table 3). T2 generation Arabidopsis plants were used for all analyses. For the ArathNictaba overexpression lines, selection was done until the transformed plants were homozygous for the T-DNA integration (T3 and T4). Integration of the T-DNA was checked by PCR on gDNA using kanamycin resistance gene primers (Supplementary Table 4) using the following PCR program: 10′ 94°C – 45 × (30′′ 94°C – 30′′ 48°C – 1′ 72°C) – 5′ 72°C. The quality of the gDNA was checked with actin primers (Supplementary Table 4) using the same PCR program. Overexpression levels of the ArathNictaba genes in 15-day-old seedlings were quantified by RT-quantitative (q)PCR. Three independent homozygous single insertion lines were selected for each construct (35S::AN4 and 35S::AN5).

For the aerial plant tissues, WT Arabidopsis seeds were sown in vitro on MS medium (until 22 days) or in artificial soil (Jiffy-7, 44 mm Ø). The Arabidopsis plants grown in artificial soil were watered regularly and fertilizer was added once after 25 days. Whole plantlets were collected at 6, 15, and 22 days after sowing. Rosette leaves from at least five plants were harvested and pooled after 31 days. After 39 and 54 days, rosette leaves, cauline leaves, stems and flowers from at least five plants were sampled and pooled. For the root samples, WT Arabidopsis seeds were sown in expanded clay granules (Ø < 4 mm). The plants were watered regularly and fertilizer was added once per week. Root samples from at least 20 plants were collected and pooled after 34, 46, and 59 days. All samples were immediately frozen in liquid nitrogen and stored at -80°C prior to RNA extraction. Two biological replicates were performed and analyzed, each with two technical replicates.

Sixteen-day-old Arabidopsis seedlings grown in vitro on a filter paper on top of MS medium were used for the treatment with the following solutions: 100 μM MeJA, 100 μM ABA, 300 μM SA, and 150 mM NaCl. Prior to use, stock solutions of the hormones (MeJA, ABA, and SA) were made in 100% ethanol and water was used in case of the salt solution. Control plants were kept on liquid MS medium containing an equal concentration of the corresponding solvent (ethanol or water). For each treatment, the filter papers with the germinated seedlings were transferred to Petri dishes filled with liquid MS medium containing either the hormone or the salt solution, and incubated at 21°C. Heat stress was applied by incubating the plates with seedlings in the dark at 37°C, controls were incubated at 21°C in the dark. For every treatment, 50 seedlings were collected at several time points (1, 3, 5, 10, and 24 h) after stress initiation. Samples were immediately frozen in liquid nitrogen and stored at -80°C prior to use. Four independent biological replicates were performed for MeJA, SA and NaCl stress, two biological replicates were performed for ABA and heat stress.

Pseudomonas syringae and Botrytis cinerea infection experiments were performed with 5-week-old WT Arabidopsis plants of the Col-0 ecotype grown in artificial soil (Jiffy-7). The P. syringae pv. tomato DC3000 strain and the B. cinerea B05.10 strain were supplied by Prof. Dr. M. Höfte of the Phytopathology Lab (Ghent University, Belgium). Infection assays were performed according to Pieterse et al. (1996), Audenaert et al. (2002), and Katagiri et al. (2002), with minor modifications.

Pseudomonas syringae pv. tomato DC3000 was grown in liquid King’s B medium (20 g/L peptone, 1% glycerol, 1.5 g/L KH₂PO₄, 1.5 g/L MgSO₄⋅7H₂O, pH 7.2) at 28°C on a rotary shaker (200 rpm) until the culture reached the mid to late log growth phase (OD₆₀₀ = 0.6–1.0). After centrifugation of the culture (10 min, 2500 g), bacterial cells were resuspended in 10 mM MgSO₄ to obtain a solution of bacteria with an OD₆₀₀ of 0.05 (corresponding to 2.5 × 10⁷ cfu/mL). Prior to use, 0.05% Silwet-77 (GE Specialty Materials, Switzerland) was added to the infection solution. The mock solution consisted of 10 mM MgSO₄ containing 0.05% Silwet-77. The rosette leaves of the Arabidopsis plants were sprayed until run-off with either the infection or the mock solution. To increase the efficiency of the infection, the plants were maintained at 100% relative humidity 1 day before the treatment until 2 days after the start of the bacterial infection. Three biological replicates were performed and analyzed.

The Botrytis strain was kept on regular potato dextrose agar plates at 21°C. Sporulation was stimulated by incubation for 10 days at 21°C under a 12/12 h UV/dark light regime. After 10 days, Botrytis spores were harvested by washing the plates with distilled water containing 0.01% Tween-20 (VWR). This suspension was filtered through a nylon membrane (20 μm Ø) and an inoculation suspension (5 × 10⁵ conidia/mL) was prepared in half strength potato dextrose broth medium. The mock solution consisted of the same medium without spores. Droplets (10 μL) of the infection suspension or the mock solution were applied on the upper side of three rosette leaves from each plant. Plants were maintained in 100% relative humidity during the entire experiment. Two biological replicates were performed and analyzed.

During the infection assays, plants were kept in a Conviron growth chamber at 21°C with a 12/12 h light/dark photoperiod. Control plants were kept separately from infected plants. Rosette leaves of 8-10 randomly selected plants were sampled in liquid nitrogen at different time points post-infection and samples were stored at -80°C prior to use.

Myzus persicae was kindly provided by Prof. Dr. Guy Smagghe (Agrozoology Lab, Ghent University, Belgium) and kept on sweet pepper plants under lab conditions (Shahidi-Noghabi et al., 2009). Aphid infestation was performed on 5-week-old Arabidopsis plants sown in round plastic pots (Ø 11 cm) with soil on which a transparent ventilated cage (Novolab) was placed. Sixty aphids were placed on the rosette leaves of each plant. Control plants were grown in cages without aphids. During the assay, plants were kept in a Conviron growth chamber at 21°C with a 12/12 h light/dark photoperiod. At indicated time points, two leaves of nine randomly chosen plants were sampled, frozen in liquid nitrogen and stored at -80°C prior to use. Four biological replicates were performed and analyzed, each with two technical replicates.

Real time RT-qPCR analyses of the gene expression during development and after P. syringae infection, were performed using the Rotor-Gene 3000 (Corbet Life Science) and the Rotor Discs (Qiagen, Hilden, Germany). The program was as follows: 10 min 95°C – 45 × (25 s 95°C – 25 s 60°C – 20 s 72°C) – 5 min 72°C followed by generation of a melting curve (gradual increase from 72 to 95°C with 1°C/step). The Rotor Gene 6 software generated the raw output data (Cq values), these were statistically analyzed using the REST-384 software (Corbett Research). REST-384 uses a pair wise fixed reallocation randomization test as a statistical model (Pfaffl et al., 2002).

RT-qPCR analyses of the other stress experiments and overexpression experiments were performed using the 96-well CFX Connect^TM Real-Time PCR Detection System (Bio-Rad). The program was as follows: 10 min 95°C – 45 × (15 s 95°C – 25 s 60°C – 20 s 72°C) followed by generation of a melting curve (gradual increase from 65 to 95°C with 0.5°C/step). The CFX Manager 3.1 software (Bio-Rad) generated the raw output data which were statistically analyzed using the REST-384 software.

All reactions were conducted in a total volume of 20 μl containing 1 × SensiMix^TM SYBR^® No-ROX One-Step mix, 2 μl undiluted cDNA template, 500 nM gene specific forward and reverse primers (Supplementary Table 2). All gene specific qPCR primers were designed using Primer3¹. Specificity of the primers was tested in silico by BLAST search and amplicons were cloned and verified using agarose gel electrophoresis and sequencing (LGC Genomics, Berlin). Amplification efficiency of all primer pairs was determined in the CFX Manager 3.1 (Bio-Rad) and qBASE^PLUS software (Hellemans et al., 2007). All expression data were normalized using three reference genes: PP2A, TIP41, and UBC9 (Czechowski et al., 2005). All melting curves were checked and reference gene stability and quality control of the samples were validated in the qBASE^PLUS software (Hellemans et al., 2007).

Four-week-old Arabidopsis WT plants and three independent homozygous transgenic lines for each construct (35S::AN4 and 35S::AN5) were inoculated with the infection (1.6 × 10⁷ cfu/mL) or mock solutions. Rosette leaves of three individual randomly selected plants were sampled at different time points post-infection (1 – 5 dpi). Two independent biological replicates were performed.

To measure leaf damage, six leaves per line per time point were scanned with a flatbed scanner at the highest resolution. The percentage of leaf damage was determined using the Image Analysis Software for Plant Disease Quantification Assess 2.0 (APS, St. Paul, MN, United States) using a self-written macro adjusted to our sampled leaves. The data were tested for normal distribution with the Shapiro–Wilkinson test. The Mann–Whitney U test was used for not-normally distributed data, supplemented with a non-parametric equivalent of the Levene’s test to check homogeneity of variances. The Bonferroni–Holm correction was used for multiple testing.

Using trypan blue staining (Sigma–Aldrich, Diegem, Belgium), plant cell death was visualized. One leaf of three plants per line for each time point was submerged in trypan blue solution (0.02%) and boiled for 2 min. Afterward, the leaves were incubated overnight at room temperature on a rotary shaker (50–100 rpm). Next, the trypan blue solution was replaced with a chloral hydrate solution (100 g/40 mL water) to destain the leaves. The destained leaves were placed on a microscopy slide in 50% glycerol and pictures were taken with a Leica S8APO microscope (DFC400 camera) and Leica Plan APO 1.6× objective. The trypan blue staining was scored by estimating the percentage of blue staining or cell death. Leaves without trypan blue staining (0%) were assigned a score 1. Score 2 was assigned to leaves for which the percentage of cell death was 1–30%. Leaves with 31–60% of cell death were assigned a score 3. Score 4 was assigned to leaves for which the percentage of cell death was 61–100%. Each transgenic line was statistically compared with the WT using a Mann-Whitney U test supplemented with a non-parametric equivalent of the Levene’s test to check homogeneity of variances. The Bonferroni–Holm correction was used for multiple testing.

To determine the P. syringae biomass, gDNA was first extracted from the infected and mock treated leaves using CTAB. The CTAB buffer (2% CTAB, 0.1 M Tris-HCl pH 7.5, 1.4 M NaCl, 20 mM EDTA) was added to 100 mg of plant material and this mixture was incubated at 65°C for 90 min in a shaking heat block. After chloroform:isoamylalcohol (24:1) extraction the gDNA was precipitated with 100% isopropanol, washed with 76% ethanol/0.2 M NaOAc and 76% ethanol/10 mM NH₄OAc and dissolved in water. Quantification of P. syringae biomass was performed with RT-qPCR using oprf primers targeting the outer membrane porin protein F gene of P. syringae (Brouwer et al., 2003) (Supplementary Table 2). ACT2 and PEX4 primers were used as reference genes for A. thaliana (Supplementary Table 2). The REST-384 software was used to calculate the ratio of P. syringae gDNA to A. thaliana gDNA (Pfaffl et al., 2002).

Gibson assembly was used to assemble the AN4-HIS sequence with the pET-21a(+) vector (Novagen) (Gibson et al., 2009). The AN4 open reading frame followed by a Gly3-linker and a His6-tag was used as a template for PCR. Forward and reverse primers were used to add Gibson assembly sites (Supplementary Table 5). The vector backbone of the pET21a vector was amplified in a PCR reaction with the following components: 5 × Q5 reaction buffer, 2 mM dNTP mix, 0.1–1 ng pET21a, Q5 high fidelity DNA polymerase, water, forward and reverse primer (10 μM) (Supplementary Table 5). Linearization of the vector backbone was performed by DpnI restriction (1 h at 37°C). PCR products were purified with the InnuPREP PCR pure kit (Analytik Jena, Germany). For the Gibson assembly reaction, equimolar amounts of linear vector backbone and AN4 expression construct were mixed together with 15 μl of Gibson master mix and incubated at 50°C for 1 h.

Half of the Gibson assembly mixture was transformed into heat shock competent E. coli strain Rosetta(DE3) cells (Novagen). Transformants were grown on LB agar plates supplemented with 100 μg/mL ampicillin and 25 μg/mL chloramphenicol, and screened by colony PCR using primers that contain the Gibson assembly sites (Supplementary Table 5). The pET21a plasmids containing the AN4-HIS were purified using the GeneJET Plasmid Miniprep kit (Life Technologies) and sequenced by LGC Genomics (Berlin, Germany) with a forward sequencing primer on pET21a (Supplementary Table 5).

Recombinant E. coli Rosetta(DE3) cells were grown overnight in 5 mL LB supplemented with 20 μg/mL ampicillin and 25 μg/mL chloramphenicol at 37°C on a rotary shaker (185 rpm). The next morning, the E. coli cells were diluted in 50 ml LB supplemented with 200 μg/mL carbenicillin and 25 μg/mL chloramphenicol, and grown at 30°C on a rotary shaker (185 rpm). After the cells reached an OD_{600 nm} of 0.5–0.6, the expression of recombinant protein was induced with 0.2 mM IPTG. After application of IPTG, the E. coli cells were grown overnight at 14°C on a rotary shaker (185 rpm). When growing bacterial cultures in larger volumes (300 mL), the time of induction was prolonged to 3 days to obtain more recombinant protein. Next, the E. coli cell cultures were harvested by centrifugation at 8,000 rpm for 15 min. The cell pellets were kept overnight at -20°C and solubilized in 10 mL of 1 × PB (20 mM NaH₂PO₄⋅2H₂O, 230 mM Na₂HPO₄) with 1 M NaCl and 1 mg/mL lysozyme at pH 8.0. Additionally, the cells were sonicated three times for 2.5 min. After sonication, the solutions were centrifuged at 4°C for 45 min (9,000 rpm) to separate the soluble from the insoluble fractions. The insoluble fractions were resuspended in 8 M urea. Both fractions were checked for the presence of recombinant protein by SDS - PAGE and Western blot analysis.

Escherichia coli cell extracts containing the recombinant AN4-HIS were used for further purification. Therefore extracts adjusted to 10 mM IZ and pH 8.0, were loaded on a Ni-NTA agarose column (MCLAB, South San Francisco, CA, United States) equilibrated with PB pH 8.0 containing 1 M NaCl. The column was washed with PB containing 1 M NaCl, 50 mM IZ, pH 8.0 and PB containing 1 M NaCl, 75 mM IZ, pH 8.0 to remove a specifically bound proteins. Elution of the column was performed with PB containing 1 M NaCl, pH 8.0 with increasing concentrations of IZ ranging from 100 to 500 mM IZ. Several fractions of 500 μL were collected for each elution buffer. The OD_{280 nm} of each fraction was measured with the Nanodrop 2000 spectrophotometer (Thermo Scientific). The purity of the protein samples was verified by SDS-PAGE and Western blot analysis.

Protein samples were analyzed by SDS-PAGE on 15% acrylamide gels as described by Laemmli (1970). After separation, proteins were visualized by gel staining with Coomassie Brilliant Blue R-250 or blotted onto polyvinylidene fluoride transfer membranes (FluoroTrans^® PVDF, Pall Laboratory, United States). Membranes were blocked with TBS (10 mM Tris, 150 mM NaCl, 0.1% (v/v) Triton X-100, pH 7.6) containing 5% (w/v) non-fat milk powder. Subsequently, membranes were incubated for 1 h with a mouse monoclonal anti-His6 antibody (Thermo Fisher Scientific) diluted 1:3000 in TBS. After washing three times with TBS, membranes were incubated for 1 h with the 1:1000 diluted rabbit anti-mouse IgG secondary antibody labeled with horseradish peroxidase (Dako, Glostrup, Denmark). After washing twice with TBS and once with 0.1 M Tris-HCl buffer (pH 7.6), immunodetection was achieved using a colorimetric assay with 0.1 M Tris-HCl buffer (pH 7.6) containing 700 μM 3,3′-diaminobenzidine tetrahydrochloride (Sigma–Aldrich, St. Louis, MO, United States) and 0.03% (v/v) hydrogen peroxide. The detection reaction was stopped after 2–10 min by washing the membrane with distilled water. All washes and incubations were performed at room temperature on a gently shaking platform.

Pull-down assays were performed on column. First, the soluble fractions of the two different E. coli cultures harboring the AN4-HIS pET expression vector (treated or non-treated with IPTG) were incubated with the Ni matrix overnight. After washing, the Ni-agarose beads were incubated with leaf extracts from 16-day-old Arabidopsis plants subjected or not subjected to NaCl stress for 5 h, as described in Section “Hormone and Abiotic Stress Treatments,” resulting in four different combinations. For each pull-down experiment three replicates were performed and analyzed by mass spectrometry.

First, the soluble fraction (50 mL) of the E. coli culture producing recombinant AN4-HIS was incubated overnight with the equilibrated Ni matrix (100 μL). The soluble fraction originated from a 300 mL E. coli culture induced or non-induced with IPTG to produce the recombinant AN4 (bait). After overnight incubation, the mixture of Ni-agarose beads and soluble E. coli protein solution was transferred in a Poly-Prep purification column (Bio-Rad). The column was washed first with PB containing 1 M NaCl and 50 mM IZ at pH 8.0 followed by a second wash with the same buffer, but with 75 mM IZ. Then the Ni-agarose beads were transferred to an Eppendorf tube, 1 mL plant extract (prey) was added to the beads and this mixture was incubated for 30–35 min on a turning wheel. The plant extract was obtained from untreated or salt-stressed Arabidopsis plant material (0.5 mL) with 1 mL extraction buffer (1 × PB, 1 M NaCl, 25 mM IZ, 10% glycerol, 0.1% Tween-20, 1% β-mercaptoethanol and complete^TM, Mini, EDTA-free Protease Inhibitor Cocktail at pH 8.0). After centrifugation and removal of the plant extract, the beads were washed with the first wash buffer containing 50 mM IZ. Finally, the beads were washed three times with 20 mM Tris-HCl containing 2 mM CaCl₂ at pH 8.0 and resuspended in 150 μL of the same buffer to store at -20°C prior to mass spectrometry. All steps described above were performed in a cold room at 4°C.

Before mass spectrometry, the samples were treated with trypsin (1 μg) for 4 h at 37°C to cleave all proteins from the Ni beads. After removal of the Ni beads, a second trypsin treatment was performed overnight at 37°C. Trifluoroacetic acid (1%) was added to deactivate the trypsin and the samples were subsequently desalted. Next, the samples were dried completely and re-dissolved in 2% acetonitrile and 0.1% trifluoroacetic acid. Afterward, mass spectrometry was performed with the Q Exactive^TM HF Hybrid Quadrupole-Orbitrap^TM Mass Spectrometer. Database searches were achieved with the MaxQuant software and statistical analysis with the Perseus software. Both analyses were outsourced to the Proteomics Expertise Center (Center for Medical Biotechnology, VIB, UGhent).

Microscopy analysis was performed using the confocal laser scanning microscope Nikon A1R (Nikon Belux). The lower epidermis of the leaf disks (spots that are infiltrated) were visualized using the 40 × S Plan Fluor ELWD objective lens (NA 0.6). All images are a combination of different fluorescent images acquired along the z-axis, as such the complete epidermis cell could be visualized. EGFP was excited with a 488 nm argon ion laser, fluorescent emission filters were 525 nm for EGFP and 650 nm for autofluorescence of chlorophyll. All images were created by the software package NIS-Elements (Nikon) and image analysis was performed using Fiji (Schindelin et al., 2012).

Multiple sequence alignments were performed with Clustal Omega² and pairwise sequence alignments were obtained using EMBOSS Water³. Prediction of the presence of a signal peptide was conducted using Phobius and SignalP 4.1 (Käll et al., 2004; Petersen et al., 2011), whereas the presence of a NLS was predicted by NucPred (Brameier et al., 2007). Using SUBA3, protein subcellular localization was predicted (Hooper et al., 2014). BLASTn and tBLASTn searches against the expressed sequence tag (EST) database were performed using the NCBI website⁴. Using the eFP browser (Winter et al., 2007) and Genevestigator (Hruz et al., 2008), microarray expression data were analyzed to study the developmental expression as well as the gene expression under different stress conditions.

This study focuses on three Nictaba homologs from Arabidopsis, namely AN3 (AT4G19850.2), AN4 (AT1G31200), and AN5 (AT4G19840). These ArathNictaba sequences contain next to their Nictaba domain (approximately 145 AA), an N-terminal sequence of different length (Figure 1A). Protein BLAST with these N-terminal sequences concedes no homology to any known protein domain. A search for signal peptides and transmembrane regions using the Phobius and SignalP 4.1 server predicted no signal peptides or transmembrane regions. Judging from the absence of signal peptides, AN3, AN4, and AN5 are probably synthesized on free ribosomes in the cytoplasm.

Assuming that the primary transcripts undergo no further processing, AN3, AN4, and AN5 will differ from each other in their molecular masses and isoelectric points. AN5 is the largest protein with a molecular mass of 28.1 kDa. AN3 has a molecular mass of 25.6 kDa and AN4 is the smallest protein, being 20.3 kDa. AN3 is an acidic protein with an isolectric point of 6.11 whereas both AN4 and AN5 have an isoelectric point around 9 (9.44 and 9.18, respectively).

Sequence alignment of the Nictaba domains for the ArathNictaba 3–5 and Nictaba from tobacco (encoded by AF389848) allowed to calculate the sequence identities and similarities (Figure 1B). AN3 and AN5 show the highest percentage of sequence identity (48.4% sequence identity; 67.5% sequence similarity) in the lectin domain sequence. Moreover, their sequence identity with Nictaba is very similar, 31.8 and 31.5%, respectively. The lowest percentage of sequence identity with Nictaba is observed for AN4 (22.9%). The tryptophan residues reported to be important for carbohydrate binding activity of Nictaba are conserved in all ArathNictaba sequences (Figure 1B, Schouppe et al., 2010).

To validate the in silico results with respect to the cellular localization, N- and C-terminal EGFP fusion constructs were created. Microscopy analysis of the T2 generation of Arabidopsis plants stably transformed with the C-terminal EGFP fusion constructs for each ArathNictaba sequence showed a nucleocytoplasmic localization for AN4 and AN5 whereas AN3 only resided in the cytoplasm (Figure 2). The localization of AN4 and AN5 in the cytoplasm and the nucleus was confirmed for the N-terminal EGFP fusion constructs. The N-terminal EFGP fusion construct of AN3 never showed a fluorescent signal (Supplementary Figure 1A).

In addition, leaves from N. benthamiana were transiently transformed with the N- and C-terminal EGFP fusion constructs for each ArathNictaba sequence and confirmed the results obtained with A. thaliana (Supplementary Figure 1B). In both experiments, free EGFP was used as a positive control and was observed in the nucleus and the cytoplasm (Supplementary Figures 1A,B).

Using RT-qPCR, the expression level of the ArathNictabas was investigated in different tissues from Arabidopsis during development under standard growth conditions. Plant material was collected at different developmental stages starting from 6-day-old plantlets to 54-day-old plants and the normalized relative expression for the three ArathNictabas was quantified throughout the development of the plant relative to the expression of these ArathNictabas in 6-day-old plantlets (Figure 3). Transcripts for the three ArathNictaba genes are detected in every tissue at all developmental stages tested. The expression level of AN3 is significantly higher in the stems and the flowers compared to the expression in 6-day-old plantlets. The expression level of AN4 is slightly but significantly higher in the roots and significantly lower in the flowers of the plant compared to its expression in 6-day-old plantlets. The expression level of AN5 is significantly higher in the rosette- and cauline leaves at all developmental stages tested. Moreover, the expression level of AN5 is slightly but significantly lower in the flowers compared to its expression in 6-day-old plantlets.

The normalized expression of AN3, AN4, and AN5 compared to the expression of one of the three reference genes (PP2A, TIP41, or UBC9) is much lower for AN3 and AN4 than for AN5, indicating that the expression level of AN5 is higher than the expression levels of AN3 and AN4 in all tissues throughout the development of the plant (data not shown).

To further investigate the biological importance of the ArathNictabas in the stress response, A. thaliana was transformed with the overexpression constructs 35S::AN4 and 35S::AN5, and transgenic lines were selected. Transcript levels for AN4 and AN5 were determined in the homozygous AN4 and AN5 overexpression lines using RT-qPCR on cDNA obtained from 15-day-old Arabidopsis seedlings. The 35S::AN4 transgenic lines show 50- to 800-fold expression of AN4 compared to WT plants. However, for the 35S::AN5 transgenic lines a much lower overexpression level is observed, ranging between 15- and 45-fold compared to WT plants. Based on their expression level relative to the ArathNictaba expression in WT plants, three independent overexpression lines exhibiting different overexpression levels were selected (Figure 7).

Wild type and transgenic 35S::AN4 and 35S::AN5 A. thaliana plants were infected with the virulent hemibiotrophic P. syringae to investigate the role of the ArathNictaba genes in the defense against this pathogen. Infection of A. thaliana plants with P. syringae results in yellow lesion areas on the rosette leaves of the plant. Leaf damage was measured daily on scanned leaves and the percentage of the lesion area relative to the total leaf area was calculated. First bacterial lesions started to appear at 3 dpi, but only at 4 dpi differences in leaf damage were observed for the overexpression lines compared to WT plants (Figure 8A). Statistically significant differences in leaf damage compared to the leaf damage in WT plants were observed especially for lines AN4_B1, AN4_L4, AN5_D5, and AN5_G3. These four overexpression lines reveal a significantly lower percentage of leaf damage compared to WT plants suggesting they are more tolerant to P. syringae infection. The leaf damage in mock treated plants was also determined but was never higher than 6.5% (data not shown). Furthermore, the leaf damage observed for the 35S::AN5 lines is significantly (p ≤ 0.05) negatively correlated with the expression level of the different overexpression lines (Pearson correlation, SPSS23). No such correlation between the level of leaf damage and the expression level of AN4 was observed for the different overexpression lines for AN4.

To strengthen these results, P. syringae biomass on the A. thaliana WT and transgenic plants was quantified in leaves at 3 and 4 dpi. Transgenic lines AN4_B1, AN5_D5, and AN5_G3, which showed a significantly lower percentage of leaf damage compared to WT plants, also had a significantly lower P. syringae biomass (Figure 8B) indicating that these overexpression lines are more tolerant than WT plants to infection with P. syringae. Transgenic line AN5_A2, which had less (though not significantly) leaf damage compared to WT, also revealed a significantly lower P. syringae biomass. Transgenic line AN4_L4 had a lower (but not significant) P. syringae biomass compared to WT plants.

A third analysis included the visualization and quantification of cell death in the leaves using trypan blue staining (Figures 8C,D). Transgenic line AN4_B1 showed a significantly lower level of leaf damage and P. syringae biomass compared to WT plants (Figures 8A,B) which is in agreement with the differences (not significant) observed for cell death (Figure 8C) concluding that line AN4_B1 is more tolerant to P. syringae infection than WT plants. Surprisingly, line AN4_J4 showed a significantly reduced amount of cell death compared to WT plants (Figure 8C) but the percentage of leaf damage was almost equal to that of WT plants and the P. syringae biomass was significantly higher than in WT plants (Figures 8A,B). Transgenic line AN4_L4 showed comparable results to WT plants in all analyses except for the leaf damage analysis which revealed a significantly lower leaf damage (Figure 8A). As such it cannot be concluded that line AN4_J4 and AN4_L4 are more tolerant to P. syringae infection. P. syringae biomass for line AN5_A2 is significantly lower than for the WT plants (Figure 8B) which is in agreement with the lower percentages of leaf damage and cell death, suggesting AN5_A2 is more tolerant to P. syringae infection than WT plants. Lines AN5_D5 and AN5_G3 show significantly lower levels of leaf damage and P. syringae biomass but the amount of cell death is not significantly lower. Therefore lines AN5_D5 and AN5_G3 are probably also more tolerant to P. syringae infection than WT plants. The trypan blue staining in mock treated plants was not quantified since no signs of cell death were observed (Figure 8D).

To search for interacting partners, recombinant protein was produced and used for pull-down assays. Recombinant protein production was attempted for AN3, AN4, and AN5 in two eukaryotic systems namely the yeast Pichia pastoris and a tobacco cell culture of bright yellow 2 cells. However, no recombinant protein could be detected for either of the ArathNictabas. Recombinant protein production in E. coli resulted in a small amount of AN4 protein, but no protein for AN3 and AN5 was detected in the soluble fraction. The pull-down experiments were performed with recombinant AN4 produced in E. coli and Arabidopsis extracts from WT seedlings or seedlings treated with 150 mM NaCl for 5 h. All E. coli and A. thaliana proteins were identified using mass spectrometry. Comparative analyses for pull-down assays performed with the IPTG induced E. coli culture producing AN4 and a non-induced E. coli culture allowed to exclude the E. coli proteins that bound aspecifically to the Ni matrix and the Arabidopsis proteins that bind aspecifically to these E. coli proteins or to the Ni matrix. Results from the pull-down assays with extracts from non-treated and salt treated plants were also compared with each other. Since the transcript levels for AN4 in Arabidopsis are downregulated after salt stress (see section “The Expression of the ArathNictabas Showed Dissimilar Patterns after Abiotic Stress Treatments”) this can also be the case for the interaction partners for AN4. Possible interaction partners for which the expression is downregulated by salt stress are expected to be absent in the analysis using the plant extract from the salt stressed plants compared with the analysis using the extract from the non-treated Arabidopsis seedlings.

Volcano plots allow to quickly identify significant differences between two large datasets by plotting significance (p-value) vs. fold change. A first comparative analysis was made for the pull-down assays performed on Ni matrix with the IPTG induced and non-induced E. coli culture, using protein extracts of Arabidopsis seedlings grown under normal conditions (Figure 9A). As expected, AN4 is significantly more present in the pull-down experiments performed with the induced E. coli culture. Next to AN4, three A. thaliana proteins originating from the plant extract are also significantly more present in particular TGG1, ESM1, and PTAC16, suggesting these proteins are possible interaction partners for AN4. The volcano plot in Figure 9B shows a second analysis resulting from the comparison between the proteins identified in the pull-down assays performed with the IPTG induced and non-induced E. coli culture, using a plant extract of salt-treated Arabidopsis seedlings. AN4 and two A. thaliana proteins TGG1 and BGLU23 are significantly more present in the pull-down experiments with the induced E. coli culture (Figure 9B), suggesting that these two proteins are possible interaction partners for AN4. TGG1 was also retrieved in the first analysis (Figure 9A) making it more conclusive to be a true interaction partner for AN4. Compared to TGG1, the data for BGLU23 had a much lower p-value (Figure 9B). A comparison between the datasets resulting from the pull-down assays performed using the IPTG induced E. coli culture and protein extracts from non-treated Arabidopsis seedlings and salt-treated Arabidopsis seedlings did not yield any significant hits, suggesting that there are no significant differences in AN4-bound proteins from plant extracts of the salt stress treated Arabidopsis seedlings compared to WT seedlings (data not shown). All significant hits from the pull-down experiments can be found in Supplementary Tables 6, 7.

Alignment of the sequences encoding Nictaba from tobacco and the Nictaba domains from the ArathNictaba sequences showed that the tryptophan residues important for carbohydrate binding activity in Nictaba are conserved in all Nictaba domains. However, based on this observation, no conclusions can be drawn with respect to the carbohydrate binding specificity of the ArathNictabas. Stefanowicz et al. (2012) analyzed the carbohydrate specificity of an F-box Nictaba (AT2G02360) protein from A. thaliana. The F-box Nictaba sequence contains the tryptophan residues important for sugar binding activity but glycan array analyses yielded results that were very different compared to the GlcNAc binding Nictaba. F-box Nictaba was shown to specifically bind to N-acetyllactosamine, Lewis A, Lewis X, Lewis Y and blood type B motifs (Stefanowicz et al., 2012).

Enhanced green fluorescent protein fusion proteins for AN4 and AN5 localized to the nucleus and the cytoplasm whereas AN3 only resided in the cytoplasm (Figure 10). The nucleocytoplasmic localization for AN5 was previously also reported by Cayla et al. (2015). No other experimental data are available with respect to the subcellular localization of AN3 and AN4, but a cytoplasmic localization is also predicted by the SUBA3 server (Hooper et al., 2014).

The results from the microscopy analyses are in agreement with the absence of a signal peptide in the three ArathNictaba sequences and suggest that translation of the ArathNictaba transcripts takes place on free ribosomes in the cytoplasm. AN4 and AN5 locate to the nucleus although the putative NLS reported for the Nictaba sequence from tobacco (K102–K105) is not conserved in the ArathNictaba sequences. Lange et al. (2007) reported that proteins larger than 40 kDa are too big to diffuse passively into the nuclear compartment, as already shown in Paine et al. (1975). Görlich (1998) reported that the nuclear pore complex allows passive diffusion from proteins up to 60 kDa. Wang and Brattain (2007) extended this limit, stating that the size of the proteins that diffuse passively into the nucleus can be even larger than 60 kDa. The calculated molecular mass of the EGFP fusion proteins for AN4 and AN5 is 48.5 and 56.3 kDa, respectively. Considering the most recent information about the passive diffusion limit, both proteins could enter the nucleus by passive diffusion through nuclear pore complexes (Wang and Brattain, 2007). Alternatively, AN4 and AN5 possibly contain a non-classical NLS sequence recognized by importin α (Kosugi et al., 2009). Furthermore, additional nuclear import pathways, independent of importin α, have been characterized (Suh and Gumbiner, 2003; Ziemienowicz et al., 2003; Pemberton and Paschal, 2005).

Quantitative analyses of the transcript levels for the ArathNictabas in different tissues from A. thaliana revealed that the three ArathNictaba genes under study are expressed in all tissues throughout development. However, it should be mentioned that lectin expression is generally low, especially for AN3 and AN4. However, specific stress factors can enhance the transcript levels for each lectin. The low expression observed for the Arathnictabas is unlike the expression patterns observed for Nictaba from tobacco. The tobacco lectin was not detected in plants grown under non-stress conditions (Chen et al., 2002; Lannoo et al., 2007). Similar to AN3, AN4, and AN5, F-box Nictaba from A. thaliana and the Nictaba-like lectins from soybean were present in plant cells at low levels when grown under optimal growth conditions (no stress), suggesting that the function of the Nictaba homologs is more complex (Stefanowicz et al., 2016; Van Holle et al., 2016).

Comparative analyses of the expression patterns for the different Nictaba homologs in response to stress revealed that transcript levels for the ArathNictabas are specific and vary depending on the different abiotic or biotic stress treatments performed (Figure 10).

The genome of A. thaliana is fully sequenced and annotated. For many proteins, interaction partners are predicted and/or experimentally proven, revealing part of the interactome of A. thaliana. Until now, no interaction partners are reported for AN4. Pull-down experiments using AN4-HIS coupled to Ni agarose beads as a bait protein and an Arabidopsis extract as prey proteins, followed by mass spectrometry revealed several possible interaction partners for AN4 (Figure 9). Two interaction partners were identified in the pull-down assays with the plant lysate originating from the salt stressed Arabidopsis seedlings. TGG1 (AT5G26000) or myrosinase 1 is a thioglucoside glucohydrolase localized in the vacuole. BGLU23 (AT3G09260) or PYK10 is a β-O-glucosidase localized in the ER bodies. A similar experiment with the plant lysate originating from non-treated Arabidopsis seedlings yielded three possible interaction partners, one which was also a significant hit in the first analysis, in particular TGG1. In addition, GDSL esterase/lipase ESM1 (AT3G14210) and protein plastid transcriptionally active 16 (PTAC16; AT3G46780) were identified as possible interaction partners (Supplementary Table 6). ESM1 is a secreted protein with a role in glucosinolate hydrolysis as a myrosinase-associated protein. PTAC16 is known as a chloroplast protein which regulates the membrane-anchoring functions of the nucleoid (Zhang et al., 2006; Burow et al., 2008; Ingelsson and Vener, 2012). It should be mentioned that the latter analysis was the result of two rather than three replicates of the soluble fraction of the non-induced E. coli cultures.

Myrosinase 1 belongs to the glycosyl hydrolase 1 family and is a β-thioglucoside glucosidase (Andersson et al., 2009). Glycosyl hydrolases hydrolyse the glycosidic bond between carbohydrates or a carbohydrate and non-carbohydrate moiety (Ahn et al., 2010). The myrosinases from A. thaliana, referred to as TGG1 - TGG6 are all grouped in the myrosinase gene family (Barth and Jander, 2006), and catalyze the hydrolysis of glucosinolates, thereby initiating the formation of isothiocyanates, nitriles, thiocyanates, epithionitriles and other reactive products (Thangstad et al., 2004; Andersson et al., 2009). Glucosinolates or thioglucosides consist of a glucose residue linked to an AA derived R-group of aliphatic, aromatic or indole types by a thioglucoside bond (Thangstad et al., 2004). In A. thaliana, different glucosinolates with side chains derived from methionine, tryptophan, phenylalanine and isoleucine were found (Barth and Jander, 2006). Myrosinases and glucosinolates are known to be localized in separate plant cells and consequently only make contact after tissue disruption. Myrosinases are present in the myrosin phloem idioblasts in phoem parenchyma while glucosinolates are present in the S-cells adjacent to the phloem (Thangstad et al., 2004; Barth and Jander, 2006; Andersson et al., 2009). After pathogen invasion or insect herbivory, the myrosinases can degrade glucosinolates and as such produce toxic products to protect the plant against these invaders. Thus, myrosinases play a role in plant defense against microbes and herbivores (Barth and Jander, 2006). Next to the myrosin phloem idioblasts, TGG1 was also reported in the guard cells (Thangstad et al., 2004).

BGLU23 or PYK10 is a β-O-glucosidase that, like myrosinase 1, belongs to the GH1 family (Nagano et al., 2008). In total 47 β-glucosidases have been identified in A. thaliana, called BGLU1 - BGLU47. BGLU23 (or PYK10) and TGG1 (or BGLU38), the two interaction partners identified for AN4, represent two of these β-glucosidases. BGLU23 is a root and hypocotyl specific β-glucosidase which is synthesized with an endoplasmic reticulum (ER) retention signal at its C-terminus (Nitz et al., 2001; Xu et al., 2004). BGLU23 is known to hydrolyze the natural substrate scopolin and other coumarin glucosides similar in structure to scopolin. Scopolin is one of the most abundant secondary metabolites in the Arabidopsis roots (Ahn et al., 2010). The resulting scopoletin is a fungitoxic compound and can be polymerized by peroxidase in the presence of H₂O₂ (Reigh et al., 1973). This can protect plant cells from the oxidative damage caused by pathogens. BGLU23 localizes to ER bodies in the roots, which are not present in rosette leaves under normal growth conditions. However, ER bodies can be induced by MeJA treatment and wounding in rosette leaves (Nagano et al., 2005; Ahn et al., 2010). Upon disruption of cells, BGLU23 forms large complexes with its binding proteins. One of these binding proteins is PBP1, a jacalin homolog from A. thaliana (Ahn et al., 2010). BGLU23 is a glycoprotein with three high-mannose oligosaccharides which can be recognized by PBP1, the jacalin homolog. It is possible that PBP1 participates in the BGLU23 and ER body-mediated defense systems against herbivores and pathogens (Matsushima et al., 2004). PBP1 is thought to act as a molecular chaperone that helps the correct polymerization of BGLU23 when tissues are damaged. This polymerization of BGLU23 is necessary for its activity (Nagano et al., 2005, 2008).

At first sight, there is no obvious link between AN4 and the plant defense-involved enzymes TGG1 and BGLU23. However, interesting observations were made when the localization, the expression pattern and the expression levels were compared in more detail. Protein localization and expression under normal growth conditions can be compared at subcellular level, according to cell type and at tissue level. AN4 resides in the nucleus and the cytoplasm of the cell and its expression under normal growth conditions is significantly higher in the roots compared to 6-day-old seedlings (Figures 2, 3). Jakoby et al. (2008) identified AN4 as one of the 5% most highly expressed genes in mature Arabidopsis trichomes. Moreover, they reported a strong enrichment of genes involved in root atrichoblast differentiation in the trichome transcriptome (Jakoby et al., 2008). The sequence encoding TGG1 contains a signal peptide and the protein localizes to the vacuole (Thangstad et al., 2004). Xue et al. (1995) reported the expression of TGG1 in the leaves, stems and floral organs. At cell level, TGG1 was especially found in the guard cells and in phloem myrosin cells. Myrosin cells or idioblasts are specific cells that differ greatly from the neighboring cells in size, structure and content. The morphology of these specific cells can vary in different organs, tissues and developmental stages (Thangstad et al., 2004). BGLU23, like TGG1, is synthesized with a signal peptide, but also contains an ER-retention signal (KDEL sequence). This β-O-glucosidase, also called PYK10 is specifically localized in ER bodies and is root specific (Nitz et al., 2001; Matsushima et al., 2003; Ahn et al., 2010). BGLU23 was also reported in the Arabidopsis plasmodesmal proteome, but probably represents a cytoplasmic contaminant in this fraction (Fernandez-Calvino et al., 2011).

A striking similarity between AN4 and BGLU23 at tissue level is their expression in the roots of Arabidopsis. However, according to microarray data (eFP browser) the expression level of BGLU23 in the roots is much higher than the expression level of AN4 under normal growth conditions (Winter et al., 2007). Furthermore, it is interesting to note that BGLU23 was also identified as one of the 5% most highly expressed genes in mature Arabidopsis trichomes (Jakoby et al., 2008). Although trichomes are present on the leaves, it is reported that trichome development in leaves and atrichoblast development in roots share a network of transcription factors (Pesch and Hülskamp, 2004; Schellmann et al., 2007). However, it is not known whether trichomes and atrichoblasts really share common patterns of gene expression (Jakoby et al., 2008). Jakoby et al. (2008) reported high activity of genes involved in the glucosinolate pathways in trichomes, indicating the role of trichomes in plant defense. Possibly AN4 and BGLU23 can interact in the roots or trichomes when the plant is subjected to stress. To do so, at least one of the two proteins would have to change its subcellular localization. Changes in subcellular localization after stress have been reported before (García et al., 2010; Moore et al., 2011). Alternatively AN4 and BGLU23 can interact as a result of the cell damage provoked by, e.g., pathogens.

Another similarity at tissue level is seen between TGG1 and AN4. TGG1 was especially abundant in the phloem myrosin cells whereas AN4 was reported as a PP2 of Arabidopsis. PP2 proteins are most abundant in the phloem sap. Though the sequence of AN4 shows high similarity to the PP2 domain, a protein domain that is conserved among many species in the plant kingdom, there is no evidence for the presence of AN4 in the phloem, except for the microarray data from eFP browser (Dinant et al., 2003; Thangstad et al., 2004; Winter et al., 2007).

The expression level of AN4 is low under normal growth conditions and its expression remains low or is even downregulated after all investigated stress treatments, except for the P. syringae infection. P. syringae infection of 5-week-old rosette leaves resulted in approximately two times upregulation of AN4 transcript levels at the latest timepoints (after 5 and 7 days). According to the microarray data of the eFP browser, BGLU23 shows a sixfold upregulation 24 h after P. syringae infection, but TGG1 shows a small downregulation (Winter et al., 2007). It has to be mentioned that our data also first show a twofold significant downregulation of AN4 about 24 h after P. syringae infection. Although TGG1 and AN4 show a similar regulation 24 h after P. syringae infection, their absolute expression levels are very different, TGG1 is present in much higher levels in infected Arabidopsis leaves than AN4 (Figure 10). At present, it remains unclear if the expression of BGLU23 and AN4, and their interaction in the trichomes is part of the defense response against Pseudomonas.

Since the pull-down analysis is an in vitro analysis and the plant extract contains a mixture of proteins derived from different cell compartments, it is uncertain whether the interaction partners retrieved from the pull-down assay represent true in vivo interaction partners for AN4. Additionally, TGG1 and BGLU23 are two proteins that are quite abundant in Arabidopsis seedlings, whereas AN4 is expressed at a much lower level. Future experiments are thus necessary to confirm these protein interactions in vivo.

To conclude, similar to Nictaba from tobacco the ArathNictabas under study are also localized to the nucleus and/or the cytoplasm (Chen et al., 2002). ArathNictaba 3–5 are expressed at very low levels in all tissues during the development of A. thaliana under normal growth conditions, but their expression is clearly stress-inducible, though specific stresses trigger differential expression of ArathNictaba 3–5. Thus, it can be concluded that expression patterns for the ArathNictabas are specific and vary for different abiotic or biotic stress treatments (Figure 10). Our data suggest that ArathNictabas could play a role in the stress response of A. thaliana. Future research is needed to investigate the subcellular localization of the ArathNictabas after plants that have been exposed to stress situations. In addition, analyses with mutant lines can help to decipher the role of the ArathNictabas in the plant stress response. Finally, the interaction of AN4 with TGG1 and BGLU23 has to be confirmed in vivo and it has to be shown whether this binding is a protein–protein or a protein–carbohydrate interaction.