Determining the Impact Changing Host Metabolism has on Leaf Associated Microbiomes for Improved Efficacy of Foliar Biopesticides

General Experimental Procedures

Experiments were conducted in an indoor growth chamber at the Penn State University's Plant Pathology and Environmental Microbiology department. This growth chamber was a single, isolated, full environment Conviron chamber which allowed for robust control over all abiotic conditions. The utilized growth chamber contained two compartments, with equal lighting in the top growing compartment and the lower growing compartment. This was validated utilizing a light meter software app called 'photone'. Independent testing with comparisons to hardware light meter (Extech LT40 LED Light Meter) and Photone showcased near identical readings between attempts (~.5 +/-) and photone was utilized for the rest of the study. Table 1 below provides details on the growth chamber conditions, and seed sterilization protocols. In order to assure that plants had minimal disturbance by a nascent microbial community which could act as bio stimulants, mountain fresh plus seeds (MFP) were sterilized and grown in glass petri dishes for 1 week, typically until tap roots were ~ one (1) inch long. After a week, plantlets were transplanted into ~ five (5) inch pots with growing media. Labels were distributed randomly to each plant, and then each biological replicate per treatment was randomly selected and put into a tray with biological replicates of other treatments. This ensured all treatments were distributed equally across the top and bottom of the growth chamber and maintained randomization that could be included for later statistical analysis as a 'block'. There were in total five (5) biological replicates across five (5) 'blocks' and two (2) 'levels' (the top growing compartment or the lower growing compartment in the growth chamber). For both objectives in this experiment, plants were allowed to grow two weeks post transplanting before initial inoculations. The second youngest fully mature leaf for each plant was selected and considered the 'treated' leaf marked with a white band (Figure 1, red; https://projects.sare.org/?attachment_id=1072763) and the next youngest fully mature leaf was considered the 'distal' leaf (Figure 1, white; https://projects.sare.org/?attachment_id=1072763). Inoculations were done by suspending the bacterial species of interest into a liquid suspension of KB at one (1) OD (a measurement of bacterial mass in suspension, one (1) OD is ~ two (2) million cells) and submerging the 'treated leaf' in the suspension for twenty (20) seconds.

Table 1. Details on growth chamber conditions and seed sterilization protocols.

This work was conducted in two parts that subsequently filtered into each other (Figure 2 and Figure 3). Each experiment will be explained in further detail below. Experiments and their timelines, treatments and controls, and subsequent notes for both experiments can be found in table 2.

Table 2. Experiments and their timelines, treatments and controls, and subsequent notes

Experimental Procedure For Objective 1     

The first portion of the experiment was to validate if leaf surface metabolites changed as a function biotic stress. Stress was induced by infecting the treated leaf with the pathogen Pseudomonas syringae pv. tomato DC3000 (DC3K), which is known to cause bacterial speck (Picture 1)

Seventy-two (72) hours after inoculation with the pathogen, the treated leaf and distal leaf were collected into separate 50 mL falcon tubes, submerged in MeOH/H2O (70%/30%), vortexed for ten (10) seconds, sonicated for ten (10) minutes, and vortexed again for ten (10) seconds before being being discarded. For each sample, the MeOH/H2O solution was then filtered through a .22 um PES single use filter (150 mL) and poured onto two petri dishes to dry (Figure 4).

All samples finished drying twenty-four (24) hours after pouring onto the petri dishes. One set of dried Metabolites were resuspended in .5 mL of deuterated H2O while the other half was frozen and saved for use in the second objective of this study. The reconstituted metabolites in .5 mL D2O were aliquoted into a 5 mm NMR tube. Single proton nuclear magnetic resonance imaging allows comparison of hydrogen composition across samples where differences in the position of the hydrogen on the spectrum indicates a different hydrogen (Figure 5).

Notably, this created a 'fingerprint' capable of comparing changes in the metabolic composition between the control and pathogen treatment via differences in peak position and intensity. Generated spectra from HNMR on samples were aligned by EtOH peak position (ppm ~2.3, not shown above) and normalized by leaf weight (data not shown here). Spectra were then filtered by a noise value threshold of 100 to establish a 'floor' for relevant peaks. Peaks that persisted past this filtering step were then assigned as either 'overlapping', 'control associated', or 'DC3K associated' via manual alignment. Peaks that overlapped were given the same number and peaks that did not were given a unique number. For example, hydrogen peak at ppm 3.49 was consistent in all spectra regardless of sample and was assigned the character 1 to indicate character 1 relates to the ppm 3.49. At character 1, all samples had a peak at ppm 3.49. Peak 7.43 was only identified in pathogen samples and was given the character 2. No control samples had a peak at character 2, so each control sample had a 0 value, and each pathogen sample had a value of 7.43. So on and so forth, until all peaks had been given a character and that character was either considered  to be an overlapping peak, a control associated peak, or a DC3K associated peak. A value of 0 was input for empty values where the peak was not present. This created a character by value data frame which allowed comparisons of peak composition without the expectation of them being additive values. Notably if not done this way, most statistical programs either 1) added the values to compare means which has no relationship to the experiment as these are discreet positions, or 2) assumed that the value was not present if not aligned within the same row. The aligned and binned peak values were input into metaboanalyst's one factor statistical analysis module and generated a PCA of identified peaks across control and pathogen plants, in both treated and distal leaves. These significantly different peaks were then input into a natural products dereplication process wherein the peaks are compared to known spectra to isolate series of peak combinations that can be identified within the fingerprint and resolved to a metabolite or other product.

Experimental Procedure For Objective 2

The second portion of this experiment was to understand how the changes in leaf surface metabolites affect the pathogen DC3K and a closely related species that is considered a plant growth promoter, Pseudomonas fluorescens A506 (A506). The first step for validation was an in vitro 96-well plate experiment observing changes in bacterial growth to understand if leaf surface metabolite changes observed in Objective 1 were related to metabolites with antibiotic capacities or if the altered metabolite profile promoted pathogen growth. The second step for validation was a two-stage in vitro experiment, wherein seventy-two hours after initial biotic elicitor inoculation (DC3K or A506) the distal leaf was inoculated with either DC3K and A506 and CFUs were recovered from distal leaves seventy-two hours post distal inoculation. 

in vitro validation

DC3K and A506 were grown from single colony streaks for twenty-four 24 hours prior to application. On day of application, each species was suspended in M9 minimal media with no carbon source (recipe Table 3) to an OD of .2. The frozen metabolites collected from Objective 1 were reconstituted to 1.5 mL utilizing the same M9 no carbon media. 500 ul of M9-bacterial solution was placed in each well, and 500 ul of M9-metabolite solution were placed in each well. Growth of bacteria over 18 hours was recorded, and this experiment was repeated 3 times for a total N of 15 for each treatment/species.

in vivo validation

Stress was induced by inoculating the treated leaf with either pathogen DC3K or growth promoter A506. Seventy-two (72) hours after inoculation with either bacteria, the distal leave was treated with either DC3K or A506. For controls, there was a fully negative control (inoculated distal leaf and treated leaf with buffer), a DC3K control (buffer treated leaf and DC3K inoculated distal leaf), and an A506 control (buffer treated leaf and A506 inoculated distal leaf). 72 hours after distal inoculation, only the distal leaves were collected across all treatments, and samples were placed into separate 50 mL falcon tubes filled with KB buffer. Submerged leaf samples were vortexed for ten (10) seconds, sonicated for ten (10) minutes, and vortexed again for ten (10) seconds before being being discarded. The resulting KB+CFUs suspension was diluted in 1/5 dilutions in subsequent KB for a total of 8 times and pipetted onto a rifampicin antibiotic plate. A506 and DC3K strains used in this experiment were rifampicin resistant. This procedure allowed for comparisons between what an initial infection would look like (control DC3K or control A506) and a subsequent infection on the distal leaves would look like, while maintaining equal time for the infection to initiate. This resulted in CFUs that could be corroborated back to the treatment for comparison and normalized by plant surface area.