Premise of Study

Plant endophytic bacterial strains can influence plant traits such as leaf area and root length. Yet, the influence of more complex bacterial communities in regulating overall plant phenotype is less explored. Here, we conducted two complementary experiments to test if we can predict plant phenotype response to changes in microbial community composition. 

Methods

In the first study, we inoculated a single genotype of Populus deltoideswith individual root endophytic bacteria and measured plant phenotype. Next, single inoculation data were used to predict phenotypic traits in mixed three-member community inoculations, which we tested in the second experiment. 

Key Results

When in isolation, each bacterial endophyte significantly but weakly altered plant phenotype relative to non-inoculated plants. In mixture, bacterial strain BurkholderiaBT03, constituted at least 98% of community relative abundance. Yet, plant resource allocation and tissue nutrient concentrationswere disproportionately influenced by Pseudomonas sp.GM17, GM30, and GM41. We found a 10% increase in leaf mass fraction and a 11% decrease in root mass fraction when replacingPseudomonas GM17 with GM41 in communities containing both Pseudomonas GM30 and BurkholderiaBT03. 

Conclusions

Our results indicate that interactions among endophytic bacteria may drive plant phenotype over the contribution of each strain individually. Additionally, we have shown that low-abundant strains contribute to plant phenotype challenging the assumption that the dominant strains will drive plant function.

Plant propagation - We propagated Populus deltoidesin tissue culture following standard procedures (see Henning et al., 2016). Briefly, surface-sterilized shoot tips were cultured in a magenta box (Sigma-Aldrich, St. Louis, MO, USA) containing 80 mL of 1× Murashige & Skoog (MS) basal medium (Murashige and Skoog, 1962)supplemented with MS vitamins (Caisson Labs, North Logan, UT, USA), 0.05% 2-(N-morpholino) ethanesulfonic acid (Sigma-Aldrich), 3% sucrose, 0.1% PPM™ (plant protective mixture, Plant Cell Technology, Washington, DC, USA), 0.5% activated charcoal (Sigma-Aldrich, St. Louis, MO, USA), and 0.15% Gelzan (Plantmedia, bioWORLD, Dublin, OH, USA). Plants were sub-cultured three generations to reduce natural microbiome communities. Plant cultures were maintained in a growth room at 25 °C under a 16 h photoperiod. To measure if plant cultures had reduced endophyte communities, we plated macerated plant material on Reasoner’s 2A agar (R2A) medium (Sigma-Aldrich, St. Louis, MO, USA) and used a dissecting microscope (Leica Microsystems, Wetzlar, Germany) to search for colony forming units after three generations of tissue culture. While this method would miss un-culturable endophytic strains, we were able to determine that three successive generations of culturing reduced the culturable bacterial colony forming units to zero. After root establishment, we selected plants that were similar in size and at a similar development stage to use in experiments. We weighed each individual plant prior to establishing the experimental treatment to account for variation in initial plant size within our statistical models.

Bacterial isolates - Pseudomonas sp.strains (GM17, GM30, and GM41) and Burkholderiasp. (BT03), hereafter termed PseudomonasGM17, PseudomonasGM30, PseudomonasGM41, and BurkholderiaBT03 were isolated from Populusroot endosphere, for isolate descriptions, see Brown et al., 2012; Weston et al., 2012). Comparative genomics of Pseudomonas GM30, GM41, GM17 and BurkholderiaBT03 strains can be found in Brown et al., 2012; Timm et al., 2015, 2016; Henning et al., 2016.

Experimental design- We constructed closed mesocosms by interlocking two sterile Magenta boxes (Sigma-Aldrich, St. Louis, MO, USA) with a coupler (Sigma-Aldrich) (for full description, see Henning et al., 2016). We added 150 cc calcined clay (Pro’s choice Sports Field Products, Chicago, IL, USA) and 70 ml of 1× Hoagland’s nutrient solution (Sigma-Aldrich, St. Louis, MO, USA) into each mesocosm. We drilled two 7 mm holes on adjacent sides of the upper magenta box to allow air flow into and out of the mesocosm. We covered the air holes with adhesive microfiltration discs to prevent microbial contamination of the mesocosm yet allow air exchange (Tissue Quick Plant Laboratories, Hampshire, United Kingdom). Prior to bacterial addition, we double-sterilized each closed mesocosm with two 60 m dry autoclave cycles on consecutive days (STERIS Laboratories, Mentor, OH, USA). We grew the bacterial strains in isolation and at a constant temperature (25 °C) in 5 ml of R2A medium. After growing strains overnight we pelleted and re-suspended them in sterile water to an OD600 of 0.01 (~1.0×E7 cells ml^-1).

Experiment 1: Individual bacterial impacts on plant traits- We grew Populus individuals in isolatedmesocosms. Each mesocosm was inoculated with individual PseudomonasGM17, Pseudomonas GM30, Pseudomonas GM41, or BurkholderiaBT03. We also grew Populusindividuals in a bacteria-free control treatment. Prior to plant addition to the mesocosm, we weighed each individual to account for initial plant size differences among treatments. In total, there were 40 mesocosms with five treatments (n = 8). The experiment was divided into four different establishment dates. In 2014, Pseudomonas GM30, PseudomonasGM41, or BurkholderiaBT03 were established at three sampling points (1 March, three replicated blocks; 25 March, two replicated blocks; and 2 April, three replicated blocks) because microbiome-free plant tissues were difficult to propagate. On 25 March 2019, all eight replicates of Pseudomonas GM17 were established. Prior to plant addition, we pipetted 10 ml of each bacterial strain (10⁷cells¹ml^-1) into the clay substrate and stirred the slurry for 30 s to distribute the bacteria. Non-inoculated controls received 10ml of sterile water that was stirred into the clay substrate for 30 s. Next, we transferred an individual Populusinto each mesocosm. We grew each experimental block for five weeks under a 16 hr photoperiod, at 21°C, and at 80% relative humidity (Percival Scientific, Perry, Iowa, USA). Mesocosms were randomly sorted in the growth chamber within block and were re-randomized weekly to minimize growth chamber artifacts (e.g., light differences). Closed mesocosms were not re-opened until the end of the experiment to maintain sterility within the mesocosms. We did not provide plants with additional water during the experiment because our clay medium remained hydrated within our closed system.

After 35 days of growth, we measured relative leaf chlorophyll content on three fully opened leaves on each plant using standard protocol with a non-destructive, handheld SPAD meter prior to harvest (Konica Minolta Chlorophyll Meter SPAD-S02, Ramsey, NJ, USA). Next, individual plants were removed from the mesocosm. We submerged the entire root system in sterile deionized H₂O and carefully removed clay clumps on the root system by hand. Next, we collected total plant biomass of leaves, roots, and stems by clipping plants into various tissues and drying the plant for 48 hours at 70°C. Next, we measured plant biomass allocation patterns by calculating leaf, root, and stem mass fractions (leaf, root, or stem biomass (mg) / total biomass (mg) ×100).

Experiment 2: Assembling mixed bacterial communities- Using a similar experimental design as above, we inoculated three-member bacterial communities using PseudomonasGM17, PseudomonasGM30, PseudomonasGM41, BurkholderiaBT03 strains, in a factorial design. We had four unique treatments: PseudomonasGM17, GM30, GM41; Pseudomonas GM17, GM30, BurkholderiaBT03; PseudomonasGM17, GM41, BurkholderiaBT03, and PseudomonasGM30, GM41,BurkholderiaBT03. We selected three Pseudomonasstrains to explore changes in composition among closely-related but functionally different microbes(Weston et al., 2012; Timm et al., 2015; Henning et al., 2016; Jun et al., 2016). In each mesocosm, we pipetted (10 ml) and evenly distributed the assigned bacterial community (3.33 ml of each strain diluted to 10⁷cells ml^-1). In total, we had a total of 28 mesocosms (n=7). We transferred plants into the mesocosms and incubated them in the growth chambers for 4 weeks on April 28, 2015 (see description above). After 28 days of growth, the plants in experiment 2 reached the top of the mesocosm. Thus, we harvested them seven days earlier than our first experiment. At harvest, we measured chlorophyll content, leaf biomass, stem biomass, root biomass, leaf mass fraction, stem mass fraction, and root mass fraction as described in Experiment 1. 

To measure total Kjeldahl N and phosphorus (P) concentration by mass of the dried leaf and root tissues (Pérez-Harguindeguy et al., 2013), we used a modified micro-Kjeldahl digestion (Parkinson and Allen, 1975). First, we ground the tissue samples with a mortar and pestle. Next, we weighed 75 mg, of the ground sample and folded it into a piece of adhesive-free cigarette paper. We digested the sample for 5 h at 350° C in 5 mL H₂SO₄in a Kjeldatherm digestion block (Gerhardt, Königswinter, Germany). After the digested samples cooled, we added 45 mL deionized water to each digestion. We used a SmartChem 200 discrete analyser (Unity Scientific, Brookfield, CT, USA) to measure total Kjeldahl N and P, expressed as a proportion of total tissue mass.

Bacterial community composition- To measure bacterial community composition, we extracted DNA from 50 mg ground subsamples of each flash-frozen root sample using the PowerPlant kit (MO BIO Laboratories; Carlsbad, CA, USA). We quantified each bacterial strain in our community using qPCR primers designed to detect PseudomonasGM17 (forward 5’-TGTCACTATTATCAGCCATTGTAGA-3’ and reverse 5’-AACAGTGGATGAGGTCTAATAACAA-3’), PseudomonasGM30 (forward 5′-CAGGAAACAGCTATGACCATYGAAATCGCCCAARCG-3′ and reverse 5′-TGTAAAACGACGGCCAGTCGGTTGATKTCCTTGA-3′)PseudomonasGM41 (forward 5′-ATCCGTACCATTTATGTTGATGAGT-3′ and reverse 5′-GAAACACATCCTCTTCGTTCTGTAT-3′) and Burkholderia BT03 (forward 5′-AGACTTCTTTGATTGAGGTGAAGTA-3′ and reverse 5′-CATATAGTCGAGATGGTCATTTAGG-3′) as in (Weston et al., 2012; Timm et al., 2016). We performed qPCR with the iTaq Supermix (Biorad, Hercules, CA, USA) on a Biorad CFX96 instrument using gDNA as a standard according to manufacturer’s instructions. We corrected qPCR results to account for the initial wet root weight and corrected by dry biomass to convert our measurements to bacterial cells g^-1 of dry root biomass. We found no relationship between total number of cells and total plant biomass (F_(1,25)= 0.002, p= 0.97). We include additional details on methodology and quality control within Appendix S1 (see the Supplementary Data with this article).

This dataset includes data from 2 experiments. The first included single-strain inoculations of bacterial endophytes (Burkholderia BT03, Pseudomonas sp. GM17, Pseudomonas sp.GM30, Pseudomonas sp. GM41 and experiment 2 mixes the strains within 3-member endophyte communities. As response variables, we measured cholorophyll content as SPAD, and biomass allocation toward leaves, roots, and stems.  For more information about the headings of each column, please refer to the metadata file.