Single-stranded oligodeoxynucleotides induce plant defence in Arabidopsis thaliana

Abstract

Background and Aims

Single-stranded DNA oligodeoxynucleotides (ssODNs) have been shown to elicit immune responses in mammals. In plants, RNA and genomic DNA can activate immunity, although the exact mechanism through which they are sensed is not clear. The aim of this work was to study the possible effect of ssODNs on plant immunity.

Key Results

The ssODNs IMT504 and 2006 increased protection against the pathogens Pseudomonas syringae pv. tomato DC3000 and Botrytis cinerea but not against tobacco mosaic virus-Cg when infiltrated in Arabidopsis thaliana. In addition, ssODNs inhibited root growth and promoted stomatal closure in a concentration-dependent manner, with half-maximal effective concentrations between 0.79 and 2.06 µm. Promotion of stomatal closure by ssODNs was reduced by DNase I treatment. It was also diminished by the NADPH oxidase inhibitor diphenyleneiodonium and by coronatine, a bacterial toxin that inhibits NADPH oxidase-dependent reactive oxygen species (ROS) synthesis in guard cells. In addition it was found that ssODN-mediated stomatal closure was impaired in bak1-5, bak1-5/bkk1, mpk3 and npr1-3 mutants. ssODNs also induced early expression of MPK3, WRKY33, PROPEP1 and FRK1 genes involved in plant defence, an effect that was reduced in bak1-5 and bak1-5/bkk1 mutants.

Conclusions

ssODNs are capable of inducing protection against pathogens through the activation of defence genes and promotion of stomatal closure through a mechanism similar to that of other elicitors of plant immunity, which involves the BAK1 co-receptor, and ROS synthesis.

INTRODUCTION

Nucleic acids trigger defence against pathogens in all kingdoms of life. Given that they are present in both the pathogen and the host during an infective process, these molecules have been proposed to act either as microbe-associated molecular patterns (MAMPs) or as damage-associated molecular patterns (DAMPs) (Heil, 2009; Gallucci and Maffei, 2017). In bacteria, recognition of foreign DNA by both restriction enzymes and clustered regularly interspaced short palindromic repeats (CRISPR)-associated endonuclease 9 (CRISPR-Cas) are involved in defence against bacteriophages (Garneau et al., 2010). The mechanisms of DNA recognition have been best characterized in mammals, where DNA perception promotes immunity against bacteria and viruses. The Toll-like receptor 9 (TLR9) is located in the endoplasmic reticulum and was shown to recognize DNA present in endosomes, which activates the production of pro-inflammatory cytokines (Latz et al., 2004a, b). This receptor preferentially recognizes single-stranded unmethylated CpG-rich DNA, although methylated single-stranded CpG DNA and double-stranded DNA (dsDNA) can also activate it less effectively (Hartmann and Krieg, 2000; Krieg, 2002; Ohto et al., 2015; Hartmann, 2017; Ohto et al., 2018). The CpG-rich single-stranded DNA oligodeoxynucleotide (ssODN) 2006 is a potent activator of TLR9 (Hartmann et al., 1999), while mammalian self-DNA bound to antibodies can be endocytosed and subsequently activates TLR9 (Ohto et al., 2015). It was also demonstrated that RNase 7, a 14.5-kDa antimicrobial ribonuclease constitutively expressed in the human epidermis, promotes self-DNA recognition in keratinocytes and activates antiviral immune responses in the skin (Kopfnagel et al., 2018). In addition, association mapping studies have linked mutations in the TLR9 promoter to inflammatory diseases (Demirci et al., 2007). DNA can also activate inflammatory and local antiviral responses in the cytoplasm of mammalian cells, where it can be present as a result of viral infection or upon extensive extracellular endocytosis, through the activation of different DNA receptors, such as cyclic GMP-AMP synthase-stimulator of interferon genes (sGAS-STING), Z-DNA-binding protein, the receptor for advanced glycation end-products (RAGE) and several receptors belonging to the PYHIN family (Sirois et al., 2013; Sun et al., 2013; Collins et al., 2015; Li and Chen, 2018). Non-CpG oligonucleotides containing a PyNTTTTGT motif, such as IMT504, have also been shown to stimulate B lymphocytes and plasmacytoid dendritic cells of the human immune system (Elias et al., 2003; Rodriguez et al., 2006). Both CpG- and non-CpG-containing oligonucleotides have been shown to improve the activity of vaccines targeting infectious diseases and cancer (Elias et al., 2005; Jahrsdörfer and Weiner, 2008; Bode et al., 2011; Montaner et al., 2012; Adamus and Kortylewski, 2018).

Double-stranded RNA (dsRNA) can also activate antiviral responses in the cytosol upon recognition by RIG1 and MDA5 proteins (Takeuchi and Akira, 2008a, b; Ablasser et al., 2009). RNA can also be recognized in the endolysosomal compartment by other members of the Toll-like family, TLR3, TLR7 and TLR8 (Gasser et al., 2017), and, based on genomic structure and sequence similarity, it has been proposed that TLR9 forms a subfamily with TLR7 and TLR8 (Chuang and Ulevitch, 2000; Du et al., 2000). It was demonstrated that single-stranded RNAs (ssRNAs) derived from human immunodeficiency virus-1 (HIV-1) are able to activate TLR7 and TLR8, stimulating dendritic cells and macrophages to secrete interferon α and cytokines as part of the defence response (Heil et al., 2004).

Like in animals, plant immunity is also activated by both DNA and RNA. Bacterial DNA was demonstrated to act as an effective elicitor in Arabidopsis thaliana plants, triggering reactive oxygen species (ROS) generation and callose deposition, an effect that was reduced by DNA CpG methylation (Yakushiji et al., 2009). It was also shown that the extracellular DNA present in pea root-cap slime is required for defence against a necrotrophic fungus (Wen et al., 2009). In agreement with this observation, the absence of a single gene encoding a candidate extracellular DNase from the fungal phytopathogen Cochliobolus heterostrophus results in reduced virulence (Park et al., 2019). Huang et al. (2019) have also reported a salivary DNase II from the insect Laodelphax striatellus that acts as an effector that suppresses plant defence.

Several studies have found evidence that self-DNA, or DNA of related species, is more active in triggering immunity than foreign DNA. Autotoxicity of aged litter has been linked to its content of fragmented DNA, since inhibition of root growth by self-DNA was stronger than inhibition by DNA from unrelated species (Mazzoleni et al., 2015a). Inhibition of growth by self-DNA but not by foreign DNA was also observed in bacteria, fungi, algae, protozoa and insects (Mazzoleni et al., 2015a, b). Similarly, it was found that leaf homogenates from common bean protected this species from the bacterial pathogen Pseudomonas syringae and induced H₂O₂ and extrafloral nectar production. Non-self homogenates also elicited common bean defences, but to a lesser extent (Duran-Flores and Heil, 2014). Similar results were obtained when fragmented self-DNA was used instead of leaf homogenates (Duran-Flores and Heil, 2018) or DNA. In lettuce it was also observed that self-DNA induced plant defences, induced changes in CpG methylation and inhibited seed germination and root growth to a higher extent than non-self-DNA (Vega-Muñoz et al., 2018). By contrast, non-self-DNA from phylogenetically distant plant species did not exhibit a protective role in lettuce. It was also shown that self-DNA was more effective than heterologous plant or insect DNA in inducing plasma membrane depolarization and calcium signalling in lima bean and maize (Barbero et al., 2016). RNA has also been shown to activate plant defences, since pre-infiltration of total RNAs from Pseudomonas syringae pv. tomato DC3000 (Pst DC3000), but not from A. thaliana, activated innate immunity in this plant, reducing susceptibility to both bacterial and fungal pathogens (Lee et al., 2016).

Unlike in mammals, the molecular basis of DNA perception in plants is not clear. It was found that nucleotide-binding leucine-rich repeat (NB-LRR) receptor I2 from tomato (Fenyk et al., 2016) and Rx1 from potato (Fenyk et al., 2015) can bind both double- and single-stranded DNA as well as ssRNA in vitro. Rx1 was shown to regulate the DNA-binding activity of a Golden2-like transcription factor (Townsend et al., 2018) and triggered immune responses leading to cell death after activation by a PVX virus capsid (Knip et al., 2019). However, neither I2 nor Rx1 has been linked to nucleic acid sensing in vivo.

Pattern-triggered immunity (PTI) allows plants to recognize MAMPs or DAMPs through pattern recognition receptors (PRRs), which leads to activation of basal resistance. PRRs are all cell-surface-localized receptors, receptor-like kinases (RLKs) or receptor-like proteins (RLPs). RLKs and RLPs differ in that the former possess a cytoplasmic kinase domain, while the latter do not (Saijo et al., 2018). BAK1 [BRASSINOSTEROID INSENSITIVE1 (BRI1)-ASSOCIATED RECEPTOR KINASE1] is an RLK belonging to the family of leucine-rich repeat like kinase (LRR-RLK) receptors and act as co-receptor, or adaptor, of several LRR-RLK receptors involved in MAMP and DAMP perception, such as FLAGELLIN SENSING2 (FLS2) flagellin receptor (Chinchilla et al., 2007a, b) and PEP RECEPTOR1 (PEPR1), which recognizes endogenous PROPEP peptide-derived DAMPs, including PEP1 (Huffaker et al., 2006). BAK1 belongs to a subfamily of LRR-RLKs that includes BAK1-LIKE1 (BKK1), which can also heterodimerize with FLS2 or PROPEP1 (Roux et al., 2011). BAK1 is shared by several signalling pathways controlling developmental as well as defence responses (reviewed in Chinchilla et al., 2009). Consistent with a role in plant immunity, plants depleted in BAK1 showed reduced responses to flg22 and exhibited more symptoms in response to virulent bacteria (Heese et al., 2007; Chinchilla et al., 2007b). BAK1 might be involved in sensing of dsRNA, since this nucleic acid has been shown to activate plant defence against oilseed rape mosaic virus (ORMV) in a BAK1-dependent manner (Niehl et al., 2016).

While the immunostimulatory effect of single-stranded DNA (ssDNA) has been well characterized in mammals, it has not been investigated in plants. In this work we found that ssODNs IMT504 and 2006, previously shown to activate immunity in mammalian cells, can protect A. thaliana from Botrytis cinerea and from Pst DC3000, but not from crucifer-infecting tobacco mosaic virus (TMV-Cg). Both IMT504 and 2006 promoted stomatal closure, activated expression of defence genes and inhibited root growth, in a BAK1-dependent manner.

MATERIALS AND METHODS

Plant material and growth conditions

Plants were grown in Petri dishes containing half-strength Murashige and Skoog (MS) medium with 1 % sucrose under a 12 h:12 h light/dark cycle (photon flux density 90 μE) at 22–23 °C. The relative humidity was maintained at 60–700 %. After a week plants were transferred to a mixture of vermiculite, peat moss and perlite (1:1:1). Experiments were performed using Arabidopsis thaliana ecotype Columbia-0 (Col-0) or the following mutants: npr1-3 (Cao et al., 1997), mpk3 (SALK_151594), bak1-5 (Nekrasov et al., 2009; Schwessinger et al., 2011) and bak1-5/bkk1 (Schwessinger et al., 2011).

Bacterial strain

Strain Pst DC3000 (Ma et al., 1991) was grown overnight at 28 °C in Luria–Bertani medium supplemented with kanamycin 50 μg/mL and rifampicin 50 μg/mL.

Bacterial infection assays

Oligonucleotides (4 μm), flg22 (1 μm) or sterile water (mock) were infiltrated into leaves from 4-week-old A. thaliana plants using a needleless syringe. Twenty-four hours after pretreatment, Pst DC3000 suspensions were sprayed on A. thaliana leaves as previously described (Macho et al., 2012). Briefly, cells were harvested by centrifugation, and pellets were resuspended in 10 mm MgCl₂ at an OD₆₀₀ of 0.1. Immediately prior to spraying, Silwet L-77 0.02 % (v/v) was added to the bacterial suspension. Bacteria were sprayed onto leaf surfaces, and plants were not covered. Four hours and 4 d post-inoculation (dpi), six 0.6-cm² leaf discs (three discs per plant) per treatment were harvested and surface-sterilized (30 s in 70 % ethanol followed by 30 s in sterile distilled water). Leaf discs from two different plants were ground in 10 mm MgCl₂ in a tube using a plastic pestle. After grinding, samples were serially diluted 1:10 and plated on PYM solid medium (Cadmus et al., 1976) supplemented with the appropriate antibiotics. Plates were placed at 28 °C for 2 d, after which the colony-forming units were counted.

Botrytis cinerea infection assay

Oligonucleotides (4 μm) or sterile water were infiltrated into 4-week-old A. thaliana leaves with a needleless syringe. Twenty-four hours after pretreatment, leaves were detached, placed on MS agar plates (without sucrose) and inoculated with conidial suspensions of B. cinerea isolate B05.10 (Rossi et al., 2011). The inoculum consisted of 5-µL droplets of a conidial suspension (5 × 10⁵ conidia/mL) or medium alone (for mock treatment) prepared as previously described (Real et al., 2004; Flors et al., 2007). The adaxial surface of three fully expanded leaves per plant from four plants was inoculated with one droplet on the right side of the central vein and incubated in MS agar (without sucrose) in Petri dishes. Plates were sealed to maintain humidity. Symptoms were recorded and measured 48 h post-inoculation and the lesion area was measured using ImageJ software (Analyze tool).

Virus infection assays

The third oldest expanded leaf of each plant (1.08 stage) was dusted with carborundum. Subsequently, 5 μL of semipurified TMV-Cg virus (Asurmendi et al., 2004) diluted in 20 mM phosphate buffer (pH 7) was added and the surface of the leaf and was gently abraded. Mock-inoculated plants were buffer-rubbed. Samples of systemic leaves (half of rosette upper third oldest expanded leaf) were taken at 5 and 7 dpi. Leaves were frozen in individual tubes in liquid nitrogen and stored at −80 °C until RNA extraction. The TMV-Cg infection level was determined as relative accumulation of TMVCg-CP transcript by RT-qPCR. The following primers were used for qRT-PCR: TMVCg-CP FW, 5′ TGTCGCAATCGTATCAAAC 3′; TMVCg-CP RV, 5′ CTGTATCTGGAAACCGCTG 3′. Plants were infiltrated with distilled water or ssODN IMT504 (4 μm) 24 h before virus infection with TMVCg. Total RNA was isolated from frozen A. thaliana leaf tissues using Trizol Reagent (Thermo Fisher, USA) and subsequently treated with DNAse I (Thermo Fisher, USA). For mRNA detection, first-strand cDNA was synthesized using MMLV reverse transcriptase (Thermo Fisher, USA) according to the manufacturer’s instructions. All qRT-PCR experiments were carried out using an ABI Prism 7500 Real-Time PCR System (Applied Biosystems, USA), following MIQE (Minimum Information for Publication of Quantitative Real-Time PCR Experiments) requirements. Ubiquitin5 (UBQ5, NM-116090) was used as an internal reference gene (FW, 5′ CGGACCAGCAGCGATTGATT 3′; RV, 5′ ACGGAGGACGAGATGAAGCG 3′).

Chemicals

Oligonucleotides (IMT 504, CATCATTTTGTCATTTTGT CATT; 2006, TCGTCGTTTTGTCGTTTTGT) were synthesized by IDT (Integrated DNA Technologies, USA) in both phosphodiester and phosphorothioate versions. In the latter case, all bonds between nucleotides were phosphorothioate. All ssODNs were solubilized in water. Abscisic acid (ABA, mixed isomers), coronatine and diphenyleneiodonium (DPI) were purchased from Sigma (USA), while flg22 was synthesized by GL Biochem (China) and PEP1 by Abbiotec (USA). PEP1 and ABA were dissolved in ethanol, coronatine in methanol and DPI in DMSO, taking care to maintain the final concentration of all solvents below 1 % (v/v). For root length and qRT-PCR experiments, phosphorothioate ssODNs were used, while phosphodiester ssODNs were used for the rest of the experiments.

RNA isolation and qRT-PCR

Gene expression assays were performed on RNA extracted from five 7-d-old seedlings grown in 1× MS liquid medium supplemented with 0.5 % sucrose (w/v) (Millet et al., 2010) (four biological replicates per treatment). Seedlings were incubated with H₂O (control), 100 nm flg22 or phosphorothioate versions of the oligonucleotides (4 µm) for 4 h. Plant material was frozen in liquid nitrogen, and total RNA was extracted using Trizol Reagent (Thermo Fisher, USA) according to the manufacturer’s instructions. RNA samples were treated with RQ-1 DNA-free DNase I (Promega, USA) and quantified with a Nanodrop spectrophotometer (Biophotometer Plus, Eppendorf). Complementary DNA synthesis was performed using MMLV reverse transcriptase (Promega, USA) according to the manufacturer’s instructions. qRT-PCR experiments were carried out with a StepOnePlus Real-Time PCR System (Applied Biosystems, USA). Elongation factor 1 (EF1) was used as an internal reference gene. Primers used for qRT-PCR were as follows: EF1c (At1g18070) FW, 5′ AGCACGCTCTTCTTGCTTTC 3′; RV, 5′ GGGTTGTATCCGACCTTCTTC 3′; MPK3 (At3g45640) FW, 5′ TCCCTGGTAAAGACCATGTTCA 3′; RV, 5′ TCGGTGTGCCAAGCAACTC 3′; WRKY33 (At2g38470) FW, 5′ TTCGTATGGCTGCTTCTTTTC 3′; RV, 5′ TGAGGTTTAGGATGGTTGTGG 3′; PROPEP 1 (At5g64900) FW, 5′ TGGCTCTACTCGTCACAACG 3′; RV, 5′ CAATGTAACTTAAAGTGCCTAATTATG 3′; and FRK1 (At2g19190) FW, 5′ CAGAAACAGCGCGAAACGA 3′; RV, 5′ GGTCGGGCGGTCTGAAA 3′. qRT-PCR data analysis was performed and primer efficiencies were obtained using LinReg PCR software (Ramakers et al., 2003). The EF1c gene was used to standardize the expression of a given target gene; a ratio between treatments was then calculated using the algorithm developed by Pfaffl (2001). Relative expression ratios were determined and statistical analysis was performed using the fgStatistics software interface (J. A. Di Rienzo, pers. comm., National University of Córdoba, Argentina). The cut-off for statistically significant differences was set at P < 0.05.

Stomatal aperture bioassays

Stomatal bioassays were performed as previously described (Gudesblat et al., 2009). Epidermal peels from Col-0 or mutant leaves of 4-week-old plants were floated on 10:10 buffer under light (10 mm KCl and 10 mm MES-KOH, pH 6.15) for 2.5 h, then ABA (20 µm), flg22 (5 µm), PEP1 (100 nm) or ssODNs IMT504 and 2006 at indicated concentrations were added to the incubation medium and peels were incubated for a further 1.5 h. DPI (20 µm) or coronatine (1.56 µm) was applied 30 min prior to addition of flg22, IMT504 or 2006. For the DNase I experiment, IMT504 (1 µg) was incubated with or without 1 µL (1 U) of RQ-DNase I (Promega, USA) in the reaction buffer supplied by the manufacturer for 30 min at 37 °C. Aliquots of the reactions were diluted in 10:10 buffer with the epidermal peels so that the final concentration of the initial input of IMT504 was 4 µm. Peels were also incubated with an identical volume of reaction buffer with DNase I as a control. EC₅₀ values from dose–response curves were calculated using GraphPad Prism software (v5.01).

Root length measurements

Seedlings were grown on MS agar plates for 2 d and subsequently transferred to 1× MS liquid medium supplemented with 0.5 % sucrose (w/v) for a further 5 d in the presence of phosphorothioate versions of the ssODNs. Root length was measured with ImageJ software. Half-maximal effective concentration (EC₅₀) values from concentration–response curves were calculated using GraphPad Prism software (v5.01).

RESULTS

ssODNs elicit plant innate immunity in A. thaliana

To investigate whether ssODNs are able to trigger plant defence, we performed a bacterial infection assay with Pst DC3000. Leaves from 4-week-old A. thaliana were infiltrated with the ssODNs IMT504 and 2006, the flagellin-derived peptide flg22 (Felix et al., 1999), or water as control. Plants were sprayed with Pst DC3000 24 h later. We observed that pretreatment with IMT504 or 2006 reduced pathogen growth to a similar extent as flg22 (Fig. 1A). A similar 24-h pretreatment with IMT504 or 2006 also conferred some protection against the necrotrophic fungus B. cinerea (Fig. 1B), thus showing that the assayed ssODNs elicit a defence response against these pathogens. Next, we studied whether IMT504 protects against TMV-Cg infection, by monitoring TMV-Cg coat protein (CgCP) mRNA accumulation through qPCR. Pretreatment with the ssODN failed to protect the plant against TMV-CG (Fig. 1C). Thus, the tested ssODNs confer protection against bacterial and fungal infections, but not against TMV-Cg.

Fig. 1.

ssODNs induce immunity against Pst DC3000 and B. cinerea but not against TMV-Cg in A. thaliana. (A) ssODNs protect from Pst DC3000. Col-0 plants were infiltrated with H₂O (mock), flg22 (1 μm), IMT504 (4 μm) or 2006 (4 μm) 24 h prior to infection with Pst DC3000. Bacterial growth was measured at 0 and 4 dpi. CFU, colony-forming units. (B) ssODNs protect from B. cinerea. Col-0 plants were infiltrated with H₂O (mock), IMT504 (4 μm) or 2006 (4 μm) 24 h prior to infection with B. cinerea, and lesion areas were measured 48 h after inoculation. (C) TMV-Cg accumulation 5 and 7 dpi in Col-0 plants pre-infiltrated 24 h before infection with IMT504 (4 μm) relative to H₂0 (mock) control. TMV-Cg infection was determined as relative accumulation of TMVCg-CP transcript by RT-qPCR. Different letters indicate statistically significant differences [P < 0.05, (A) two-way ANOVA, (B, C) one-way ANOVA]. Error bars represent the standard error from two biological replicates per treatment in three independent experiments (A), from 12 lesion replicates in three independent experiments and 30 replicates in one additional independent experiment (B), and from eight replicates in two independent experiments (C).

SSODNs inhibit root growth and promote stomatal closure in a BAK1-dependent manner

Elicitors capable of triggering PTI, such as flg22 (Zipfel et al., 2006; Bartels and Boller, 2015) as well as dsRNA (Niehl et al., 2016) and exogenous DNA (Mazzoleni et al., 2015b; Duran-Flores and Heil, 2018; Vega-Muñoz et al., 2018), have been shown to inhibit root growth. Therefore, we tested the effect of IMT504 and 2006 on root growth in A. thaliana seedlings. We found no effect on root growth with regular ssODNs; however, when IMT504 and 2006 with phosphorothioate instead of phosphodiester bonds were used we found that both inhibited root growth in a concentration-dependent manner, with EC₅₀ 1.037 and 2.06 µm, respectively (Fig. 2A). Phosphorothioate bonds render ssODNs resistant to nucleases (Wickstrom, 1986). Lack of effect of regular ssODNs with phosphodiester bonds might be due to nuclease degradation after prolonged incubation in culture medium. Promotion of stomatal closure is another common response triggered by different elicitors of plant defence, such as flg22 (Melotto et al., 2006; Zhang et al., 2008), chitin, chitosan (Lee et al., 1999; Gust et al., 2007; Amborabé et al., 2008) and PEP1(Zheng et al., 2018). For this reason, we tested the effect of ssODNs on stomata, and found that IMT504 and 2006 promoted stomatal closure in a concentration-dependent manner, with EC₅₀ 1.05 and 0.79 µm, respectively (Fig. 2B). Promotion of closure by IMT504 was strongly reduced by treatment with DNase I but not by the enzyme buffer alone (Fig. 2C), showing that ssODN integrity is required to elicit stomatal immunity. Next, we tested whether BAK1, an LRR-RLK protein that has been shown to act as co-receptor of multiple proteins involved in MAMP perception (Yasuda et al., 2017), is required for the effect of ssODNs. For this purpose we measured inhibition of root growth and promotion of stomatal closure in bak1-5 and bak1-5/bkk1 mutants, since BKK1 is a receptor similar to BAK1 and with partially redundant function (Roux et al., 2011). We found a partial reduction in inhibition of root growth by IMT504 in both mutants (Fig. 3A), and a strong reduction in promotion of stomatal closure (Fig. 3B). These results indicate that ssODNs activate responses similar to those of other MAMPs and require BAK1 to activate immune responses.

Fig. 2.

ssODNs inhibit root growth and promote stomatal closure in a concentration-dependent manner. (A) IMT504 and 2006 inhibit Col-0 root growth. Two-day-old seedlings were transferred for 5 d to liquid MS medium supplemented with IMT504 or 2006 at the indicated concentrations. (B) IMT504 and 2006 promote stomatal closure in Col-0. Stomatal apertures were measured 1.5 h after incubation with IMT504 or 2006 at the indicated concentrations. (C) DNase I treatment reduces promotion of stomatal closure by IMT504. A 30-min pretreatment of IMT504 with DNase I, but not with DNase I buffer alone, reduced its capacity to promote stomatal closure. IMT504 was used at 4 μm for the stomatal assay. Different letters indicate significant differences (P < 0.05, one-way ANOVA). Error bars represent the standard error from three independent trials (A, B); n = 5 (A) or n = 40 (B, C) per trial.

Fig. 3.

ssODNs inhibit root growth and promote stomatal closure in a BAK1-dependent manner. (A) bak1-5 and bak1-5/bkk1 mutants are less sensitive than Col-0 in inhibition of root growth by IMT504 and 2006. Two-day-old seedlings were incubated for 5 d in liquid MS medium supplemented with phosphorothioate IMT504 (4 μm) or 2006 (4 μm). (B) bak1-5 and bak1-5/bkk1 mutants fail to close stomata in response to IMT504 and 2006. Stomatal apertures were measured 1.5 h after incubation with H₂O (mock), ABA (20 µm), flg22 (5 µm), PEP1 (100 nm), IMT504 (4 μm) or 2006 (4 μm). Different letters in (A) and (B) indicate significant differences (P < 0.05, two-way ANOVA). Error bars represent the standard error from three independent trials; n = 5 (A) or n = 40 (B) per trial.

Signalling elements involved in induction of defence by ssODNs

Reactive oxygen species produced by NADPH oxidase RBOHD are required for stomatal closure triggered by MAMPs (Torres and Dangl, 2005; Kadota et al., 2014; Toum et al., 2016). To find out whether NADPH oxidases are also involved in signalling downstream of ssODNs, we conducted stomatal assays in the presence of the NADPH oxidase inhibitor DPI. As shown in Fig. 4A, we found that DPI significantly reduced closure induced by IMT504 and 2006, consistent with an involvement of ROS in ssODN signalling. This possibility was further strengthened by the observation that coronatine, a bacterial toxin previously shown to inhibit ROS synthesis in guard cells (Toum et al., 2016), significantly inhibited ssODN-induced stomatal closure (Fig. 4B). We next investigated sensitivity to ssODNs of npr1-3 and mpk3 mutants, previously shown to be affected in promotion of stomatal closure by MAMPs such as flg22 (Macho et al., 2012), and found that both mutants were only partially insensitive to ssODNs. By contrast, they were completely insensitive to flg22, as previously reported. Together, these results indicate that ssODNs share signalling components with other MAMPs, such as flg22; however, the fact that npr1-3 and mpk3 mutants retain some sensitivity suggests that signalling cascades downstream of ssODNs and flg22 in guard cells are not identical.

Fig. 4.

Promotion of stomatal closure by ssODNs involves ROS production and is reduced in mpk3 and npr1-3 mutants. (A) DPI inhibits ssODN-induced stomatal closure. DPI (20 μm) was added to the incubation buffer 10 min prior to flg22 (5 μm), IMT504 (4 μm) or 2006 (4 μm). (B) Coronatine (COR, 1.56 μm) partially prevented stomatal closure induced by IMT504 (4 μm) and 2006 (4 μm). (C) npr1-3 and mpk3 mutants displayed reduced susceptibility to IMT504 and 2006. Epidermis was incubated with IMT504 (4 μm) and 2006 (4 μm) or flg22 (5 μm). Different letters indicate significant differences [P < 0.05 (A), (B) one-way ANOVA, Tukey’s test; (C) two-way ANOVA]. Error bars represent the standard error from three independent trials; n = 40 per trial in all experiments. Stomatal apertures were measured 1.5 h after application of the respective treatments.

Early defence-related genes are upregulated by ssODNs

Since ssODNs protected A. thaliana plants from pathogen attack, we investigated the induction of some genes involved in early plant defence. For this purpose, we treated 7-d-old wild-type, bak1 and bak1/bkk1 seedlings with ssODNs and measured expression of the pathogen-inducible genes MPK3 (Beckers et al., 2009), PROPEP1 (Huffaker et al., 2006), WRKY33 (Birkenbihl et al., 2012) and FRK1 (Boudsocq et al., 2010) by qRT-PCR 4 h post-incubation. All four genes were induced in response to IMT504 and 2006 in wild-type and to a lesser extent in bak1 and bak1/bkk1 mutants, albeit not in all cases with a statistically significant difference, consistent with a possible role of BAK1 in ssODN signalling (Fig. 5).

Fig. 5.

ssODNs induce defence-related genes MPK3, PROPEP1 and WRKY33. Arabidopsis thaliana 7-d-old seedlings were incubated in liquid MS medium with IMT504 (4 μm), 2006 (4 μm), flg22 (1 μm) or H₂O (control, mock). Plants were harvested 4 h post-incubation and expression of MPK3, PROPEP1, WRKY33 and FRK1 was measured by qRT-PCR. Asterisks indicate significant differences relative to mock-treated plants (P < 0.05, permutation test, fgStatistics). Error bars represent the standard deviation from three independent trials with two biological replicates each. FC, fold change.

DISCUSSION

Previously it was demonstrated that DNA extracted from plants (Wen et al., 2009; Duran-Flores and Heil, 2014; Mazzoleni et al., 2015a, b; Barbero et al., 2016; Duran-Flores and Heil, 2018; Vega-Muñoz et al., 2018) or bacteria (Yakushiji et al., 2009) can activate plant defence against bacterial or fungal pathogens. In this work we showed that two ssODNs, CpG-rich 2006 and CpG-poor IMT504, can also activate plant immunity at concentrations in the low micromolar range, similar to those reported for other activators of plant immunity, such as several MAMPs and DAMPs. To our knowledge, this is the first report showing that synthetic ssDNA can activate plant immunity against bacterial and fungal pathogens. However, we failed to observe protection by the tested ssODNs against TMV-Cg, an ssRNA virus.

Induction of immunity in A. thaliana by IMT504 and 2006 has similarities with that elicited by other MAMPs, such as flagellin (Chinchilla et al., 2007a, b; Melotto et al., 2006), since both ssODNs promoted stomatal closure, inhibited root elongation and induced the transcription of defence genes. The three responses were reduced in bak1-5 and bak1-5/bkk1 mutants, in which co-receptors involved in the perception of various elicitors of plant defence are affected (Yasuda et al., 2017). These mutants were more markedly affected in promotion of stomatal closure and inhibition of root elongation, but their effect was not apparent in all cases in induction of gene expression. While the mechanism of DNA perception in plants is not clear, our results suggest that ssODNs are perceived by an LRR-RLK receptor that requires association with BAK1 co-receptor in order to activate a downstream signalling cascade. Consistent with this possibility, similar EC₅₀s were observed for inhibition of root elongation and promotion of stomatal closure for both ssODNs, ranging from 0.791 to 2.06 µm. While flg22, a BAK1-dependent elicitor of plant immunity, has been shown to induce extracellular alkalinization with an EC₅₀ close to 1 nm (Chinchilla et al., 2006), the same ligand only promoted stomatal closure at concentrations of 1 µm and higher (Melotto et al., 2006) and inhibited seedling growth with an EC₅₀ close to 0.1 µm (Gomez-Gomez et al., 1999). As a reference, an EC₅₀ close to 1 µm has been observed for NF-κB activation by 2006 in HEK293 cells expressing TLR9 (Latz et al., 2004a), although different reports describing activation of different immune responses in mammalian cells by ssDNA or dsDNA report concentrations from low nanomolar to over 100 µm (reviewed in Heil and Vega-Muñoz, 2019). In plants, the potato NB domain of the NLR R1 receptor was shown to bind to ssDNA, dsDNA and ssRNA in vitro, with dissociation constants (K_d) of 3.34, 2.75 and 12.45 µm respectively (Fenyk et al., 2015); however, this receptor has not been linked to DNA perception in vivo. Like ssODNs, dsRNA also stimulates immunity against ORMV, a dsRNA virus, in a BAK1-dependent manner (Niehl et al., 2016), indicating that, like ssODNs, dsRNAs could also be perceived by an LRR that requires BAK1 as co-receptor. The integrity of ssODNs is required for their effect, since DNase treatment abolished most of its activity.

Previously it has been shown that A. thaliana sonicated genomic self-DNA almost completely stopped root growth, while Lepidium sativum heterologous DNA only reduced it by around 10 % in the same plant (Mazzoleni et al., 2015b). In our experiments both ssODNs reduced growth by around 50 % at the highest concentration used, but only when phosphorothioate versions were used. No inhibition of growth was observed with phosphodiester ssODNs, suggesting that they could be degraded during incubation. It is difficult to make comparisons between the two experiments, since the mechanism of perception of genomic DNA and ssODNs could not be identical. Nevertheless, it would be interesting in the future to study the effect of synthetic double and single DNAs of diverse sequence to better understand the molecular mechanism underlying the differential perception of self and heterologous DNA.

It has been proposed that dsDNA acts as a DAMP in plants (Yakushiji et al., 2009; Gallucci and Maffei, 2017; Vega-Muñoz et al., 2018). ssDNA is not normally found inside plants; however, given that the TLR9 mammalian receptor can be activated by both ssDNA and dsDNA (Rigby et al., 2014; Gasser et al., 2017), it could be possible that a similar receptor is responsible for the perception of both kinds of molecule in plants. Alternatively, ssDNA perception could have evolved as a mechanism of defence against ssDNA viruses such as geminiviruses, members of the Geminiviridae family.

Signalling downstream of MAMPs whose perception depends on LRR receptors that associate with BAK1, such as flg22 and elf18 (Chinchilla et al., 2007b; Nekrasov et al., 2009), is completely dependent on ROS production by RBOHD (Kadota et al., 2014). Promotion of stomatal closure by ssODNs was only partially reduced by DPI, an NADPH oxidase inhibitor, or in mpk3 and npr1-3, two mutants completely insensitive to flg22 that are presumably affected downstream of ROS in the signalling cascade (Toum et al., 2016). Further evidence that ROS are involved in ssODN signalling arises from the fact that promotion of closure by these molecules is reduced by coronatine, a bacterial toxin that affects ROS production in guard cells (Toum et al., 2016). Therefore, it seems likely that signalling cascades for flg22 and ssODNs share some elements, but do not overlap completely. More research is needed to find out which are the ROS-independent signalling events triggered by ssODNs.

In this work, we reported that two ssODNs previously shown to stimulate immunity in mammals, CpG-rich 2006 and CpG-poor IMT504, could also elicit plant defence. Further research using ssODNs with different sequences should reveal whether there is any specificity in sequence recognition, as suggested by previous findings showing that plants and other organisms preferentially activate their defences in response to foreign DNA (Vega-Muñoz et al., 2018). The use of synthetic oligonucleotides could shed light on the mechanism of stimulation of immunity by DNA, which might be a promising tool for crop protection and plant disease management.

FUNDING

This work was supported by Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) (grants PICT 2013 no. 1045, PICT 2015 no. 3286, PICT 2017 no. 2320 and PICT 2017 no. 2075) and by Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET) (grants PIP 2012 no. 00677 and PIP 2015 no. 00903).