Phytophthora agathidicida isolate NZFS 3770 and P. cinnamomi isolate NZFS 3910 were sourced from Scion (Rotorua, New Zealand). The two species were routinely cultured at 22°C in darkness on clarified 20% V8-juice agar (V8A). For antibiotic susceptibility testing using PM plates (Biolog Inc, CA, United States), isolates were cultured on potato-dextrose agar (PDA; P. agathidicida) or cornmeal agar (CA; P. cinnamomi) (Becton, Dickinson & Co, NJ, United States). EC₅₀ assays were carried out using CA for both species.

The activities of potential anti-oomycete compounds were tested using a modified version of the Kirby–Bauer disk diffusion assay (Bauer et al., 1966). Test compounds were prepared by adding 20 μL of sterile water to each well of a PM plate. Each plate (PM 21D, 22D, 23A, 24C, and 25D) contains 24 different compounds at four different concentrations. Five microliters of each resuspended compound were added to sterile 6 mm filter disks (Whatman number 1 paper). After sufficient drying, the four filter disks containing a single test compound at different concentrations were placed at regular intervals around the edges of a PDA or CA plate, along with a control disk (sterile water). Each plate was then inoculated by placing a 3 mm agar plug of mycelium (taken from the leading edge of an actively growing mycelial mat) in the center. Each plate was incubated at 20°C in the dark and radial growth was measured after 5 days (P. cinnamomi) or 7 days (P. agathidicida). Three biological replicates (from independently grown cultures) were performed for each compound screened. Any compound that reproducibly gave zones of inhibition that were greater than the control disk and were dose-dependent was scored as a positive.

The EC₅₀ values for radial growth inhibition were determined for each compound that inhibited Phytophthora growth in the disk diffusion assays. CA was amended with eight twofold dilutions of the compound of interest, with concentrations ranging from 0.125 to 512 μg/ml. Concentration ranges were chosen based on preliminary growth inhibition assays and differed among compounds. Control plates (CA only) were included for each compound tested, and three biological replicates were performed for each compound and concentration. Each plate was inoculated by placing a 3 mm agar plug of mycelium (taken from the edge of an actively growing mycelial mat) in its center; plates were then incubated at 20°C for 4 days (P. cinnamomi) or 6 days (P. agathidicida). Two perpendicular measurements were taken for each plate and averaged, and the size of the agar plug was subtracted to obtain the final measurement of radial growth. Measurements from test compound plates were subtracted from control plate measurements and converted to inhibition percentages. To estimate EC₅₀ values, compound concentrations were log-transformed, and non-linear regression with curve fitting (by least squares) was carried out using GraphPad Prism version 6.0. All chemicals were purchased from Sigma-Aldrich (MO, United States) except for copper sulfate pentahydrate (Ajax Chemicals, Sydney, Australia) and kanamycin sulfate (Applichem GmbH, Darmstadt, Germany).

In order to produce sporangia in P. cinnamomi and P. agathidicida, approximately 2 cm² of agar from the edge of an actively growing mycelial mat was excised and transferred to a Petri dish containing a 1:10 dilution of carrot broth (P. agathidicida) or clarified 20% V8 broth (P. cinnamomi), cut into ∼1 mm² pieces and then incubated at 25°C in darkness for 24 h. The following day, the broth was removed and each dish was washed four times in sterile Chen–Zentmyer salt solution (for P. cinnamomi; Chen and Zentmyer, 1970) or sterile pond water (for P. agathidicida) for 45 min at a time on the bench (P. agathidicida) or on a rocking table (P. cinnamomi), followed by overnight incubation (without rocking) at 22°C under light. Following sporangium production, zoospore release was induced by removing the liquid and washing each dish three times with sterile Milli-Q water that had been cooled to 4°C. Each wash was for 20 min, with the first two at room temperature and the final wash at 4°C. Wash volumes were 15 ml per dish for the first two washes, and 10 ml for the final wash. Following the final wash, dishes were returned to room temperature for 30–90 min until sufficient numbers (a final concentration of approximately 1 × 10³ zoospores/ml) of zoospores had been released.

To assess the effect of each putative anti-oomycete compound on zoospore germination, 200 μl of zoospore suspension were spread on triplicate water agar plates amended with the test compounds. Concentrations of the compounds were as for the mycelial growth inhibition assays, and ranged from 0.125 to 512 μg/ml. Plates were incubated for 12–16 h, and 100 zoospores per plate were counted under a Nikon C-DS dissecting microscope at 40× magnification to determine the percentage that had germinated. Spores were defined as having germinated if the germ tube length was at least twice as long as the spore diameter, as described previously (Liu et al., 2014). The level of germination inhibition was calculated by subtracting the percent germinated at each concentration from that of three control (water agar only) plates.

Zoospore motility assays were based on a method described previously (Hu et al., 2007). One milliliter of zoospore suspension (approximately 1,000 zoospores) was added to triplicate wells in a 24-well plate containing 10 μl of the test compound solution at different concentrations, resulting in five twofold dilutions of the test compound. Control wells were amended with an equal volume of sterile water. Zoospores were kept at room temperature, and observed with the dissecting microscope to determine the time required for all zoospore motility to stop. Observations were made at 5 min intervals for the first hour, then at 30 min intervals until motility had ceased. Wells that had no motile zoospores by the time of the first observation (5 min post-treatment) were scored as 0 (i.e., complete inhibition). Subsequent loss of motility was calculated by dividing the length of time zoospores remained motile by the average motility time in the controls.

The activities of potential anti-oomycete compounds were tested using a disk diffusion assay (Bauer et al., 1966), where the zone of inhibition of mycelial growth around filter disks containing the test compounds was assessed (Figure 1). Each species was screened against a panel of Biolog PM plates. In total, the panel of five PM plates contained 120 potential antimicrobial compounds, with each compound present at four concentrations. The panel contained common anti-yeast and antifungal antibiotics, as well as other types of chemicals with antimicrobial activity such as metals, detergents, and inhibitors. The complete list of compounds tested is available in Supplementary Table 1.

Of the 16 anionic toxins tested, only one (arsenite) had an inhibitory effect. Similarly, only two of the 15 cationic toxins tested, copper (II) chloride and copper (II) sulfate, showed inhibitory effects. None of the chelating agents tested (including dipyridyl, EDTA, and EGTA) showed any inhibitory effects.

Several common antibiotics successfully inhibited growth, particularly at higher concentrations. The majority were aminoglycosides (hygromycin B, kanamycin, neomycin, paromomycin, and tobramycin), which act by inhibiting protein synthesis. Chlortetracycline (a tetracycline antibiotic which also inhibits protein synthesis) and D-cycloserine (an amino acid derivative which inhibits pyridoxal 5′-phosphate dependent enzymes) also inhibited mycelial growth. Of the antibiotics identified here, only neomycin and paromomycin have been previously reported to have anti-oomycete activity against other Phytophthora species (Lee et al., 2005).

In addition to antibiotics, the PM screen included seven common antifungals: three triazoles (fluconazole, myclobutanil, propiconazole), an imidazole (miconazole), a polyene antimycotic (nystatin), a nucleoside analog (5-fluorocytosine) and cycloheximide. None of these showed any inhibitory effects.

Benzethonium chloride, a quaternary ammonium salt often used as a disinfectant, showed strong inhibition of P. agathidicida mycelial growth, and a low level of inhibition against P. cinnamomi.

Thallium (I) acetate and sodium arsenite both showed promising inhibitory effects; however, their non-selective toxicity (Hughes, 2002; Lennartson, 2015) makes them unsuitable for use in the environment so they were not tested further. Similarly, fluorodeoxyuridine (an oncology drug) and trifluoperazine (an antipsychotic) inhibited mycelial growth, but were excluded from further analysis due to their unsuitability for use in the real-world treatment of oomycete infections.

Based on the results of our qualitative high-throughput screen, we selected the eight most promising inhibitors for more quantitative testing. To begin, we determined EC₅₀ values for inhibition of mycelial growth. This is important, because mycelial growth assays approximate both systemic growth within the plant, and across root-grafts between plants (i.e., root-to-root contact). Regression analysis resulted in well-supported inhibition curves, with all R² values >0.9 (Figure 2). The calculated EC₅₀ values for all compounds are presented in Table 1. These values varied widely among the compounds tested, from 0.79 to 180 μg/ml.

The most effective inhibitor of mycelial growth in both Phytophthora species was chlortetracycline hydrochloride, with EC₅₀ values of 0.79 and 1.1 μg/ml in P. agathidicida and P. cinnamomi, respectively. Benzethonium chloride was also an effective inhibitor, though more so for P. agathidicida, as was seen in the initial high-throughput screen. Both of the copper compounds tested had relatively low EC₅₀ values, and were more effective against P. cinnamomi. EC₅₀ values for the aminoglycoside antibiotics varied widely, with paromomycin by far the most effective. D-cycloserine was a relatively ineffective inhibitor, although it showed the greatest species specificity of any compound tested (almost fivefold more effective against P. cinnamomi than P. agathidicida).

In addition to mycelial growth, a useful antimicrobial agent could also inhibit one or more of the other stages in the Phytophthora life cycle (see, for example, Mircetich, 1970 and Judelson and Blanco, 2005). The first step in initiating a new infection is the encystment and germination of a zoospore on the host plant (Judelson and Blanco, 2005). We tested the ability of our eight candidate antimicrobials to inhibit zoospore germination, following a previous protocol (Liu et al., 2014).

As no previous studies have reported laboratory methods for producing P. agathidicida zoospores, several variations on the standard P. cinnamomi protocol were trialed in order to maximize the number of sporangia formed and zoospores released. It was found that excised pieces of mycelial mats grew best when grown in diluted, clarified carrot broth (Erwin and Ribeiro, 1996) rather than V8 broth. Sporangia production was increased when the subsequent washing steps were carried out using sterile (filtered and autoclaved) pond water (pH ∼8.5) rather than the salt solution used for P. cinnamomi. The chemical composition of the pond water used can be found in Supplementary Table 2. Pond water samples from two sources (Rotorua, North Island, New Zealand and Dunedin, South Island, New Zealand) were trialed, with no noticeable differences in sporangium numbers. Adjusting the pH of the pond water up or down, to values between 5 and 10, reduced sporangia production, emphasizing that unadjusted pond water was optimal. As the mycelial mats of P. agathidicida were less adherent than those of P. cinnamomi, carrying out the wash steps on a rocking table tended to dislodge the mats and resulted in decreased numbers of sporangia; therefore wash steps were carried out without rocking. Methods for inducing zoospore release were as for P. cinnamomi, though P. agathidicida often required a longer final incubation step at room temperature before zoospores began to be released.

Germination rates on unamended agar plates ranged from 85 to 98% for P. agathidicida and 57–91% for P. cinnamomi (dashed lines in Figure 3). Germination of P. agathidicida zoospores was inhibited to a greater or lesser extent by five of the eight compounds tested (Figure 3), with the strongest inhibitors being benzethonium chloride (Figure 3A), copper (II) chloride (Figure 3C), and copper (II) sulfate (Figure 3D). All of these compounds inhibited germination completely at concentrations ranging from 10 to 80 μg/ml, with the two copper salts proving the most effective. On the other hand, neomycin and paromomycin only effected partial inhibition of germination at the highest concentrations tested (Figures 3G,H). The other antibiotics (chlortetracycline, D-cycloserine and kanamycin) did not inhibit germination of P. agathidicida zoospores.

The pattern was broadly similar for inhibiting germination of P. cinnamomi zoospores (Figure 3). As with P. agathidicida, the most effective inhibitors were benzethonium chloride (Figure 3A), copper chloride (Figure 3C), and copper sulfate (Figure 3D), all of which prevented germination completely at concentrations between 10 and 80 μg/ml. Paromomycin was the most effective of the antibiotics, although it only reduced the germination rate by fourfold (Figure 3H).

A third way of disrupting Phytophthora pathogenicity is to inhibit zoospore motility. Once released from sporangia, zoospores exhibit chemotaxis toward the roots of a new host plant, and, in the case of P. cinnamomi, can swim at approximately 0.5 m/h through water (Allen and Newhook, 1973; Allen and Harvey, 1974). Rendering them immotile would offer a novel containment approach that prevents spread from infected trees.

In control experiments, the untreated zoospores of P. agathidicida and P. cinnamomi remained motile in sterile water for approximately 17 and 20 h on average, respectively. We tested our set of eight anti-oomycete compounds over a range of concentrations. In general, we observed dose dependent responses, with higher compound concentrations leading to faster zoospore encystment. The only exception was D-cycloserine, which had no effect on the motility of P. cinnamomi zoospores at any concentration tested.

In order to quantify the effectiveness of the compounds, we compared the lowest concentration of each that was required to cause complete loss of motility within 5 min of treatment. The concentration required for this rapid inhibition varied widely among compounds (Table 2). Copper chloride, copper sulfate, and benzethonium chloride proved the most effective, with concentrations ≤0.1 μg/ml causing almost immediate loss of motility. At the other extreme, 256 μg/ml D-cycloserine was required to effect such immediate inhibition of P. agathidicida zoospore motility. Similarly, at the highest concentration tested (8 μg/ml), kanamycin failed to inhibit motility within 5 min. Instead, it took 25 min for P. agathidicida zoospores to lose motility at this concentration of kanamycin, and 60 min for P. cinnamomi zoospores. The general trends observed for both Phytophthora species in response to the eight compounds were similar, although P. agathidicida zoospores were over an order of magnitude more sensitive to neomycin than P. cinnamomi (Table 2).