Allelochemicals of the phenoxazinone class act at physiologically relevant concentrations

ABSTRACT Plants compete with their neighbors via the release of chemical compounds into the rhizosphere. These phytotoxins originate from a series of secondary metabolites and can be processed further by soil-living microorganisms before exerting their activity on the target plant. To determine the molecular mode of action and the physiological relevance of potential phytotoxins, it is important to simulate true-to-life conditions in laboratory experiments, for example by applying physiologically relevant concentrations. Here, we report on an improved experimental setting to study the function of allelochemicals of the benzoxazolinone class. By adjusting the solvent and the application of the chemicals, we reduced by more than 2fold the concentration that is necessary to induce growth defects in the model plant Arabidopsis thaliana.

In the struggle to survive and reproduce in their natural habitat, plants as sessile organisms compete with neighboring plants for limited resources such as light, water, nutrients and space. One strategy to gain competitive advantage is the release of phytotoxic biochemicals into the surrounding soil, either via exudation from the root or via decomposing plant material. These compounds interfere with molecular and cellular processes and thus impair the growth of the neighbors, decreasing their competitiveness. 1 This type of biochemical interference, also referred to as allelopathy, has been described for numerous species, and many of the responsible chemical compounds have been characterized. 1 However, the molecular mechanisms by which these substances act on the target plants have largely remained elusive.
We have recently reported on the biochemical mode of action of the phenoxazinones APO (2-amino-3H-phenoxazin-3-one) and AMPO (2-amino-7-methoxy-3H-phenoxazin-3one). 2 This particular class of allelochemicals derives from the cyclic hydroxamic acids DIBOA (4-dihydroxy-2H-1,4-benzoxazin-3(4H)-one) and DIMBOA (2,4-dihydroxy-7-methoxy-2H-1,4-benzoxazin-3(4H)-one), which are contained in root exudates from many grasses, among them several major crops such as wheat, rice and maize. 3,4 We were able to show that APO and AMPO have the potential to bind to histone deacetylases (HDACs), 2 an evolutionarily highly conserved class of enzymes that govern the accessibility of the DNA by removal of acetyl groups from various amino acid residues of histone proteins. Histone acetylation is a dynamic epigenetic mark that correlates with relaxed chromatin conformation, accessible DNA and transcriptional activity (Fig. 1). APO, and to a lesser extent AMPO, act as HDAC inhibitors with broad specificity in vitro. When applied on Arabidopsis thaliana seedlings, APO and AMPO cause genome-wide hyper-acetylation of histones, presumably by binding to and inhibiting HDACs in the nucleus. This results in altered chromatin configuration, which in turn leads to large-scale transcriptional changes (Fig. 1). Ultimately, misregulation of a large number of genes leads to the reduction and inhibition of growth of the target plant. Our study has been the first characterization, at the molecular level, of the mode of action of a potent allelochemical.
One major difficulty of in planta assays with purified allelochemicals is to simulate physiologically relevant conditions. 5 In particular, to make assumptions on the allelopathic potential of a compound the concentrations applied in the experiment should reflect those that realistically occur in natural environments. Allelopathic compounds are often bound to soil particles; moreover, many of them are hydrophobic. Both these factors potentially increase their biological activity, because retention by soil particles might lead to higher accumulation in the rhizosphere, and because lipophilic compounds might enter root cells more easily. However, association with soil particles and low solubility in water impede the correct assessment of allelochemical concentrations in soil. 6 Reliable indications on naturally occurring allelochemical concentrations therefore can be best derived from experiments in hydroponic culture. Experiments on durum wheat and rye showed that DIBOA accumulated in the leachate at concentrations of 30 mM and 300 mM, respectively. 7,8 Taking into account a 10:1 conversion rate of DIBOA to APO, 9,10 the resulting APO concentration reaches 3 mM to 30 mM. In our previous experiments we had determined the half-maximal-effect concentration (EC 50 ) of APO to be 75 mM, 2 and thus slightly above the concentrations found in the root leachates. It should be noted that APO, and even more AMPO, has low solubility in water. In our experiments we had therefore used ethanol as solvent for the APO stock solution, which we had then diluted in the aqueous plant growth medium.
Here, we report on an improved experimental setup that allows studying the effects of APO at lower concentrations, and thus potentially with less side effects. We found that APO was better soluble in the frequently used solvent dimethylsulfoxide (DMSO) compared to the ethanol stocks. Stock solutions were thus set up in DMSO at a concentration of 100 mM. In a dilution series, this primary stock was diluted in DMSO to generate secondary stock solutions at lower concentrations. Ultimately, 5 ml of the corresponding secondary stock solution were added to 1 ml of culture medium to achieve final concentrations between 0.05 mM to 500 mM APO. Because DMSO itself has a negative effect on root growth (Fig. S1), we added the same volume of pure DMSO to the control. In contrast to our previous experiments, in which we had grown seedlings in liquid medium, we used solid growth medium, to restrict exposure to APO mainly to the root and to transfer the experimental setting to more realistic conditions. We performed dose response analyses using root growth inhibition as readout: concentrations above 50 mM were fully inhibiting root growth, indicating that APO dissolved in DMSO had substantially higher activity compared to the ethanol stocks. In an attempt to define the active range of APO, we performed a dose response assay with concentrations ranging from 0.05 mM to 50 mM and estimated the EC 50 DMSO to 3.25 mM (Fig. 2). The EC 50 DMSO of APO was thus more than 20-fold reduced compared to the EC 50 EtOH .
To validate whether, at such low concentrations, we could detect transcriptional effects of APO similar to those identified in our previous experiments, we measured gene expression levels of APO-responsive genes by quantitative real-time PCR. There was strong overall correlation of expression fold changes between the two experimental conditions (Pearson's correlation coefficient D 0.92; P D 0.009). For five out of six tested genes we were able to confirm the change in expression, independent of the solvent. The remaining gene (AT3G14420), which had been moderately down-regulated in the ethanol-based assay, showed little response in the DMSO-based assay. We noticed that the magnitude of the effect in most cases was higher when using ethanol as solvent. One possible explanation for the discrepancy between phenotypic (root growth) and molecular (gene expression) effects of the different APO treatments is that due to the higher absolute concentration in the ethanol-based approach, APO might penetrate more deeply into the plant tissue. This could lead to transcriptional changes in a larger number of cell layers, while phenotypic effects might be caused by APO activity in the outer cell layers or the meristematic zone alone. Little is still known about the uptake of APO by the plant, or about its transport within the root and from the root to the shoot; further experiments are needed to reveal the location of cellular activity of this allelochemical.
In conclusion, our study supports our previous claim that phenoxazinone allelochemicals inhibit root growth of A. thaliana in a dose-dependent manner. 2 Making use of the higher solubility of APO in DMSO, we have established experimental conditions that better reflect physiologically relevant conditions. This setup will not only allow improved realism in future experiments, but will also drastically reduce costs, as APO is not commercially available and needs to be custom synthesized. It should be noted that the present experimental conditions do not take into account the activity of microorganisms that under natural conditions are present on or around the plant roots. Future experiments carried out in sterile versus non-sterile soil are necessary to address the role of these microbial communities in modulating the toxicity of allelochemicals and to simulate an ecologically representative scenario.

Plant growth
A. thaliana Columbia-0 seeds were sterilized in 75% ethanol for 1 min and rinsed several times with pure ethanol, then dried on a sterile filter paper. For dose response assays, seeds were stratified at 4 C for 3 days; seedlings were grown at 23 C and 16 h/8 h light/dark cycle on half-strength Murashige-Skoog medium, supplemented with 0.8% Agar and either pure DMSO or variable concentrations of APO dissolved in DMSO. For each of the three replicate experiments, 25 seedlings were grown per well in six-well plates, each well containing 3 ml solidified medium. We measured root length 6 d after germination (DAG) using the software package rootdetection v.0.1.2 (http://www.labutils.de/rd.html). For expression analysis, seedlings were germinated in sterile water; DMSO or variable concentrations of APO were added 6 DAG. Seedlings were harvested for RNA extraction 24 h later. Model of benzoxazolinone activity. Histone acetylation is controlled by histone acetyltransferases (HATs) and histone deacetylases (HDACs), respectively. APO and AMPO are derived from root-exuded cyclic hydroxamic acids. Once they enter the root cells of target plants, APO and AMPO bind to and inhibit HDACs. As a consequence, histone acetylation levels are increased, and locus-specific condensation of chromatin is inhibited. Alteration of local chromatin accessibility results in transcriptional mis-regulation.

Statistical analysis
Dose-response analysis was done using the "nplr" package (https://github.com/fredcommo/nplr) in R v3.1. We normalized root length data with the convertToProp function. We ran the nplr function to test all possible models; as expected in the case of a classical dose response, the 5-parameter model showed best performance. EC50 was calculated using the getEstimates function.

Quantitative real-time PCR
For details on qPCR protocol and primer sequences, see ref. 2

Disclosure of potential conflicts of interest
No potential conflicts of interest were disclosed