Journal list menu

Volume 4, Issue 10 art125 p. 1-12
Article
Open Access

Leaf endophytes and Populus genotype affect severity of damage from the necrotrophic leaf pathogen, Drepanopeziza populi

Posy E. Busby

Corresponding Author

Posy E. Busby

Stanford University, Department of Biology, 371 Serra Mall, Stanford, California 94305 USA

Present address: Biology Department, University of Washington, Seattle, Washington 98115 USA.

E-mail:[email protected]Search for more papers by this author
Naupaka Zimmerman

Naupaka Zimmerman

Stanford University, Department of Biology, 371 Serra Mall, Stanford, California 94305 USA

Search for more papers by this author
David J. Weston

David J. Weston

Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 USA

Search for more papers by this author
Sara S. Jawdy

Sara S. Jawdy

Biosciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 USA

Search for more papers by this author
Jos Houbraken

Jos Houbraken

CBS-KNAW Fungal Biodiversity Centre, Utrecht, The Netherlands

Search for more papers by this author
George Newcombe

George Newcombe

College of Natural Resources, University of Idaho, Moscow, Idaho 83844-1133 USA

Search for more papers by this author
First published: 24 October 2013
Citations: 30

Corresponding Editor: M. Allen.

Abstract

Fungal leaf endophytes—nonpathogenic microfungi that live within plant leaves—are ubiquitous in land plants. Leaf endophytes and host plant genotypes may interact to determine plant disease severity. In a greenhouse inoculation experiment, we found that leaf endophyte species and Populus angustifolia genotypes both affected disease outcomes in plants inoculated with the necrotrophic leaf pathogen Drepanopeziza populi. Contrary to many studies showing endophytes conferring defense, all plant genotypes inoculated with the endophyte Penicillium sp. prior to inoculation with the pathogen D. populi were characterized by greater pathogen symptom severity than plants inoculated with the pathogen only. We quantified defense gene expression via qRT–PCR, but found no evidence that increased pathogen damage was related to differential expression of the assayed genes. A second endophyte, Truncatella angustata, which was previously found to reduce symptom severity of the biotrophic pathogen Melampsora in Populus trichocarpa, did not affect symptom severity of the necrotrophic pathogen D. populi or defense gene expression. Overall, our study highlights the variable effects of endophytes on pathogen symptom severity, and illustrates that plant genotypic variation can remain important for disease outcomes even in the presence of endophytes altering disease. Additional work is needed to elucidate the mechanism by which fungal leaf endophytes alter disease in their host plants.

Introduction

Plant leaves host diverse fungal communities, comprised of species that reflect a spectrum from mutualist to pathogen (Lewis 1985, Carroll 1988). Fungi that inhabit asymptomatic plant tissues, known as fungal endophytes, have been shown to exhibit a range of life history strategies (Rodriguez et al. 2009). Our focus here is specifically on nonclavicipitaceous endophytes, those that are restricted to aboveground plant tissues and are transmitted horizontally (labeled Class 3 endophytes). Unlike the well-known clavicipitaceous endophytes that are found only in grasses (e.g., Neotyphodium endophytes), nonclavicipitaceous endophytes are found in most of the world's land plants, yet remain largely unstudied.

Controlled inoculation studies have shown that particular fungal leaf endophyte species can reduce disease in their plant hosts (Arnold et al. 2003, Ganley et al. 2008, Raghavendra and Newcombe 2013). Variation in host plant genotype is also well known to influence disease: genetic resistance, which can vary among plant genotypes, strongly determines pathogen symptom severity (Jones and Dangl 2006). A key question is whether antipathogen defense properties of endophytes depend on the interaction between particular host plant genotypes and the endophyte species involved. Answers to this question should contribute to an understanding of how endophytes potentially alter the evolutionary ecology of plant-pathogen interactions, and should also help us to draw inferences about the underlying mechanism responsible for endophyte effects on disease outcomes. A mechanistic understanding should in turn inform disease management strategies utilizing endophytes.

Research efforts are underway to evaluate plant-endophyte-pathogen interactions in Populus, a genus including P. trichocarpa and P. angustifolia, woody plant models for genomics (Tuskan et al. 2006) and ecology (Whitham et al. 2008), respectively. One such recent study evaluated how four different fungal leaf endophytes influenced resistance to the biotrophic rust pathogen Melampsora x columbiana in six different Populus host genotypes differing in quantitative resistance to the rust (Raghavendra and Newcombe 2013). Quantitative resistance does not completely prevent infection, but rather involves many genes working together to reduce pathogen latent period, infection efficiency and/or spore production (Geiger and Heun 1989). Raghavendra and Newcombe (2013) found that different endophytes reduced rust locally (i.e., not systemically) by varying degrees (0.6 to 16 times). Overall, variation among endophyte-inoculated and control leaves explained 54% of the observed variation in rust symptom severity, whereas host genotype explained only 5.0%, and host genotype-by-endophyte interaction only 1% (i.e., the endophytes were consistently ranked in the magnitude of their effects across genotypes). Therefore, disease outcomes depended more on the particular endophyte species than the particular plant genotypes involved. In fact, endophyte effects almost completely swamped plant genotypic effects on pathogens, leading to the hypothesis that endophytes are a second line of pathogen defense in plants—after major genes for resistance (Flor 1955), but before quantitative genetic resistance (Raghavendra and Newcombe 2013).

In addition to elucidating the relative importance of endophytes and plant genotypes for disease outcomes, controlled inoculation studies can also yield mechanistic insights. There are two hypotheses for how endophytes reduce pathogen symptom severity in their host plants: direct interaction between endophytes and pathogens, or local induction of quantitative genetic resistance. Systemic induction is not consistent with observed patterns of local resistance conferred by endophytes in the aforementioned study in Populus and in studies of foliar endophytes of the tropical tree Theobroma cacao (Arnold et al. 2003). While there have been no direct tests of these two hypotheses, results of Raghavendra and Newcombe (2013) are more consistent with direct interaction than with local induction. More specifically, they argue that the consistent ranking of endophytes across plant genotype-rust combinations is compatible with direct interaction between endophytes and rust (Raghavendra and Newcombe 2013). In contrast, consistent ranking is not easily explained by endophytes varying in their capacity to induce resistance yet insensitive to the differences in plant genotype-rust combinations (Raghavendra and Newcombe 2013). Despite these observations, the number of studies in this area is limited and general inferences regarding the mechanism of interaction cannot yet be drawn. Moreover, direct tests of hypothesized mechanisms are needed to confirm inferences.

Our study has two aims: (1) to evaluate the interactions between different genotypes of the host plant, Populus angustifolia, different fungal endophytes, and the necrotropic foliar pathogen, Drepanopeziza populi; and (2) to make a preliminary assessment of the possible mechanism of interaction, based on host gene expression across a range of defense-related genes. Our full factorial greenhouse inoculation experiment included six genotypes of Populus angustifolia, two endophyte species (Truncatella angustata and Penicillium sp.) plus a water control, and a single necrotrophic pathogen species (Drepanopeziza populi) plus a water control. T. angustata was previously identified as a leaf endophyte of Populus (Raghavendra and Newcombe 2013). In this study, we identify Penicillium sp. as an endophyte of Populus angustifolia; Penicillium sp. was isolated from field-collected asymptomatic leaves and showed no evidence of pathogenicity during greenhouse inoculation and re-isolation studies (see Methods and Results).

Our study design allowed us to test the relative importance of endophyte, host plant genotype, and their interaction for disease outcomes. To investigate the mechanism underlying endophyte effects on disease outcomes, we used qRT–PCR to characterize gene expression of 15 putative defense-related genes at two time-points during our experiment: three days following endophyte inoculation, then again three days after subsequent pathogen inoculation. If endophytes reduce disease severity by inducing plant genetic resistance, we would expect enhanced defense gene expression in endophyte-plus plants compared to control plants following pathogen inoculation. If endophytes increase disease severity by suppressing plant genetic resistance, we would expect suppressed expression in endophyte-plus plants compared to endophyte-minus plants following pathogen inoculation. Results consistent with the direct interaction hypothesis would include either an increase or a decrease in disease severity without any endophyte effects on defense gene expression.

Methods

Plant material

Leaves emerging from dormant buds are generally thought to be endophyte-free (Stone et al. 2000), and are useful for controlled inoculation experiments. In March 2011, we collected sixty replicate dormant cuttings from each of six mature Populus angustifolia genotypes growing in a common garden in Ogden, Utah (n = 360). These trees were cloned in 1991 from randomly selected trees growing in natural stands along the nearby Weber River, and vary in Drepanopeziza populi symptom severity (Busby et al. 2013).

Cuttings were standardized by size (20 cm in length), then rooted and raised in planters containing Sunshine Mix 4 potting soil at Stanford University in Stanford, California. For the duration of the experiment, aboveground plant tissues were kept dry to minimize infection by extraneous greenhouse fungi, and light conditions in the greenhouse were maintained at 16 hours daylight/8 hours darkness.

Endophyte material

We employed two fungal endophyte species for our experiment. The first study species was selected from a survey of leaf fungi of Populus angustifolia. In summer 2010, fungi were cultured and isolated from asymptomatic P. angustifolia leaves collected from trees along the Weber River, Utah. We surface-sterilized leaves by washing them in deionized water, soaking them in 1% sodium hypochlorite (NaOCl) solution for 2 minutes, then in two rinses in sterile distilled water for one minute each (Raghavendra and Newcombe 2013). Surface sterilized leaf discs (1cm diameter) were placed onto culture plates containing sterile malt extract agar (MEA) with the antibiotic gentamicin sulfate. The efficacy of our surface-sterilization protocol was confirmed by imprinting randomly selected surface-sterilized leaf discs on MEA in separate plates (Schulz and Boyle 2005). Fungi cultured in this way were then sub-cultured and maintained as pure cultures. Of the fungi isolated, we selected a single species (Penicillium sp., hereafter Penicillium) that was characterized by profuse sporulation under laboratory conditions and high re-isolation frequency (see below).

We used morphological and DNA sequence data to determine the identity of the Penicillium species isolated from Populus angustifolia. Total genomic DNA was extracted using the MoBio PowerSoil or MoBio Ultraclean Microbial DNA isolation kit according to the manufacturer's instructions. The ITS regions and a part of the β-tubulin gene (BenA) were PCR amplified using the primer pairs ITS1/ITS4 (White 1990) and Bt2a/Bt2b (Glass and Donaldson 1995), respectively. PCR products were cleaned with QiaGen QiaQuick PCR cleanup kits, and Sanger sequenced bidirectionally. Sequences were checked for quality and contigs were assembled. The obtained sequences were compared to the NCBI sequence database and internal databases of the CBS-KNAW Fungal Biodiversity Centre. Newly generated sequences were deposited in GenBank under accession numbers JX999057 (ITS) and JX999058 (BenA). A phenotypic characterization was performed in order to support the results of the sequence-based identifications. Penicillium strains were inoculated at three points on 90-mm Petri dishes containing malt extract agar (Oxoid) (MEA) and incubated for 7 days at 25°C. After incubation, macro- and microscopical characters on MEA were studied according the method described by Houbraken et al. (2011).

The second endophyte used in our experiment was Truncatella angustata (hereafter Truncatella). Truncatella infects a wide range of plants (Farr and Rossman 2013). We used a particular isolate (GenBank Accession number: HM122046) that was previously identified as a fungal leaf endophyte of Populus trichocarpa, and found to reduce the severity of the biotrophic rust pathogen Melampsora × columbiana in Populus trichocarpa and hybrids (Raghavendra and Newcombe 2013).

Endophyte re-isolation

In a separate experiment, we evaluated the re-isolation frequency of Truncatella and Penicillium by culturing leaf discs from plants experimentally inoculated with each of the species. We introduced sterile deionized water into endophyte cultures with sporulating conidiophores to generate inoculum. Conidia were dislodged from conidiophores by gently scraping the culture with a sterile bent glass rod. Spore solutions were transferred to spray bottles and diluted with additional sterile deionized water. We then inoculated three replicates of two of the Populus angustifolia genotypes with Truncatella, three replicates of the same two genotypes with Penicillium, and three replicates of the same two genotypes with sterile dionized water. Two weeks later, using the culturing methods described above, we calculated the re-isolation frequency of each species in three leaf discs per plant. Successful re-isolation from leaf tissues would confirm one of the two major criteria for endophyte classification: residence within leaves.

Pathogen material

Drepanopeziza populi (hereafter Drepanopeziza) is a major leaf pathogen of Populus angustifolia (Busby et al. 2012, 2013). This necrotrophic pathogen kills Populus host tissue, resulting in premature leaf abscission (Ostry 1987). Reduced plant growth, shoot and branch death, and stem cankers caused by the disease can even cause tree mortality (Ostry and McNabb 1986). In July 2011, leaves infected with Drepanopeziza were collected from mature trees of the six Populus angustifolia genotypes growing in the common garden. Leaves were incubated in plastic bags at room temperature for 72 hours to initiate sporulation. The inoculum solution was prepared by suspending leaves in sterile deionized water in a closed container and shaking vigorously to dislodge conidia. Differences in pathogenicity might be observed if inoculum were prepared from widely separated field sites, but within-site uniformity is typically seen (Spiers 1983, Busby et al. 2012).

Greenhouse inoculation experiment

Our fully factorial experiment included a total of six treatment groups: −endophyte/−pathogen; −endophyte/+pathogen; +Truncatella/−pathogen; +Truncatella/+pathogen; +Penicillium/−pathogen; +Penicillium/+pathogen. Each treatment group was populated by 4–10 replicates of each of the six Populus angustifolia genotypes. Uneven replication was a constraint of differential survival among cuttings of different genotypes.

In June 2011, +endophpyte and −endophyte plants were inoculated with one of the two endophytes or sterile deionized water, respectively. The conidia concentrations of the inoculum solutions were sufficiently concentrated to ensure coverage (3 × 106 mL−1 for Truncatella and 6 × 107 mL−1 for Penicillium). The upper and lower surfaces of leaves of each +endophyte and −endophyte plant was sprayed with 10 mL of the endophyte solution (Truncatella or Penicillium) or sterile deionized water, respectively. Inoculations were conducted in isolation to avoid cross-contamination. Plants were covered with plastic bags for 24 hours to maintain moisture at the leaf surface.

Two weeks after endophyte inoculations, +pathogen and −pathogen plants were inoculated with Drepanopeziza populi or sterile deionized water, respectively. Using spray bottles, we applied 10 mL of inoculum solution or sterile water to the upper and lower surfaces of leaves of each +pathogen and −pathogen plant, respectively. The spore concentration of the inoculum solution was 1 × 105 mL−1. Plants were covered with plastic bags for 24 hours to maintain moisture at the leaf surface.

We measured disease severity both in terms of necrotic leaf area and premature leaf abscission. Two weeks after pathogen inoculation, we quantified the percent of leaf area damaged by pathogens using image analysis. The adaxial surfaces of a total of five leaves per plant, standardized by age (leaf positions 3–7), were photographed. Images were processed with a custom MATLAB script, where we calculated the percentage of necrotic leaf area for each leaf, and the mean level of damage across leaves imaged, for each plant. We measured premature leaf abscission by counting the total number of leaves abscised over the course of the experiment (up to four weeks following pathogen inoculation) as a fraction of a plant's total leaf number.

Gene expression analysis

We used qRT–PCR to evaluate and compare plant genetic resistance responses to endophyte and pathogen inoculation, and to evaluate whether endophytes interact with host plant genetic resistance. For these analyses, we selected a total of fifteen genes (Table 1) associated with quantitative genetic resistance in Populus (Azaiez et al. 2009). The selected genes are involved in a range of putative functions, including: cell wall modification, defense, hormone metabolism, secondary metabolism, signal perception, stress and detoxification, and transcription and transport. Two genes of particular interest—PtATOSM34 and PtGlyHy17—code for defense–related proteins. A previous study found all fifteen genes were up-regulated in response to inoculation with the rust pathogen Melampsora (Azaiez et al. 2009).

Table 1. Genes used in expression analysis, from (Azaiez et al. 2009), and fold change differences in defense gene expression. Fold changes greater than ±3 are shown; empty cells represent comparisons where fold changes are less than ±3. All comparisons were less than ±3 for plants inoculated first with an endophyte and then with the pathogen, so those columns are not shown. An asterisk (*) represents t-test comparisons with P < 0.05, prior to correction for multiple hypothesis testing. No comparisons were significant at P < 0.05 after correction for multiple hypothesis testing.
table image

Our qRT–PCR analysis of gene expression was limited to 3–5 biological replicates of two different Populus angustifolia genotypes. At two time points during the experiment a single leaf (position 4 or 5) from each of these plants was sampled: three days following endophyte inoculation, and three days following pathogen inoculation. Leaves were frozen by immersion in liquid nitrogen within 30 seconds of collection. All samples were shipped in a liquid nitrogen cryoshipper to Oak Ridge National Laboratory in Oak Ridge, Tennessee for processing.

A robotic qRT–PCR system (7900HT Real Time PCR detection system; Applied Biosystems) was used to evaluate marker gene expression along defined metabolic and defense pathways as described in (Miranda et al. 2007, Azaiez et al. 2009). Total RNA was isolated, DNase treated and checked for quality using an Experion RNA StdSens analysis kit and chip (BioRad) as described in (Weston et al. 2012). cDNA was synthesized using a SuperScript III First-Strand Synthesis SuperMix for qRT–PCR (Invitrogen) according to provided protocols. The resulting 21 μl of cDNA was diluted in 100 μl H2O and used in subsequent qRT–PCR reactions. For each sample, we quantified gene expression levels relative to a UBQ gene. Using the ddCT method (Livak and Schmittgen 2001), we calculated fold change differences in relative expression for each gene between paired treatments of interest (e.g., +/− endophyte at the first time point).

Statistical analyses

To assess the significance of changes in necrotic leaf area, we modeled our response variable—the percentage of necrotic leaf area—using beta regression with a logit link function, an extension of generalized linear modeling (Ferrari and Cribari-Neto 2004), implemented in R in the ‘betareg' package (Cribari-Neto and Zeileis 2010). This analysis method is robust to heteroscedasticity and unbalanced designs. For these analyses we included only data from the pathogen treatment plants, as control plants showed little or no leaf damage. We included endophyte treatment, Populus angustifolia genotype, and their interaction as factors in models, and tested for their significance using hierarchical fitting of all possible models and assessing the contribution and significance of each term via likelihood ratios, using the ‘lmtest' package in R (Zeileis and Hothorn 2002). We were unable to satisfactorily fit any parametric models to the fraction of leaves prematurely senesced, as the data were both proportional and highly zero-inflated (which precluded the use of beta regression). Therefore, to quantify the significance of the main fixed effects on this variable, we used the non-parametric Kuskal-Wallis test to assess the effects of plant genotype and endophyte treatment individually, but did not assess their interaction. Pairwise post hoc comparisons between treatments of interest used the non-parametric Mann-Whitney U statistic.

For each tree genotype, we used fold change differences in gene expression relative to the UBQ housekeeping gene to assess plant genetic resistance responses to endophyte, pathogen, and endophyte plus pathogen inoculation. We first evaluated endophyte effects on the relative expression of individual defense genes by using a t-test to compare fold change differences for +Truncatella, +Penicillium and −endophyte plants 3 days after endophyte inoculation. Second, we evaluated pathogen effects on the relative expression of individual defense genes by comparing data for −endophyte+pathogen and –endophyte−pathogen plants 3 days after pathogen inoculation. Finally, we evaluated endophyte effects on the relative expression of individual defense genes by comparing data for +Truncatella+pathogen, +Penicillium+pathogen, and −endophyte+pathogen plants 3 days after pathogen inoculation. This represents a total of 5 comparisons for each of the 2 genotypes sampled for each of the 15 genes of interest (150 comparisons total). After evaluating normality assumptions via the Shapiro test, we conducted t-tests for each of the comparisons that showed a fold-change of ±3. We also assessed significance following a Holm correction for multiple hypothesis testing (Holm 1979).

Results

Endophyte species and their re-isolation

A BLAST search of the NCBI nucleotide database with the ITS and BenA sequence obtained from the Penicillium species used for inoculation of the plants did not retrieve any close matches. Comparison of these sequences with internal sequence database at CBS–KNAW Fungal Biodiversity Centre showed that this strain represents a new undescribed species in Penicillium section Aspergilloides. Strains of the same species were previously isolated from leaves of a Dodonaea sp. in Tasmania, leaves in an unknown plant in Kangeroo Island, Australia and soil in Tunisia and Portugal. Colonies of the investigated Penicillium species grow quickly on MEA and have a velvety texture. Microscopical analysis showed monoverticillate conidiophores with smooth walled stipes and (sub)globose conidia. The combination of these characters is in agreement with other species belonging to Penicillium section Aspergilloides (Houbraken and Samson 2011). The description of this species is deferred until a more detailed analysis is performed.

Penicillium was re-isolated from inoculated leaves at a frequency of 100%; i.e., Penicillium was re-isolated from 3/3 leaves in each of the six plants in which it was inoculated. Truncatella was re-isolated at a frequency of 83% across all leaves in which it was inoculated. Neither Truncatella nor Penicillium was isolated from control plants. However, we were able to isolate fungi that were neither Truncatella nor Penicillium from one third of control leaves. There was no effect of Populus genotype on re-isolation frequency.

Endophyte effects on disease severity

Control plants and plants inoculated only with either endophyte, but not the pathogen, did not present disease symptoms (Fig. 1). This confirms that the endophytes themselves are not pathogens under the range of conditions in this experiment. For plants inoculated with the pathogen, prior inoculation with Penicillium resulted in a two-fold increase in both necrotic leaf area and premature leaf abscission from Drepanopeziza (Fig. 1; Table 2). In contrast, Truncatella had no significant effect on disease severity. Overall, endophytes explained 11% of the variation in necrotic leaf area, whereas Populus genotype explained 37% (Table 2). In the full model, endophyte-by-plant genotype interaction was not statistically significant for necrotic leaf area (Table 2). This suggests that the effects of the endophytes used here, notably Penicillium (which significantly altered disease severity), were consistent across genotypes that varied in resistance to the pathogen.

figure image

Treatment effects (mean ± SE) on pathogen symptom severity. Disease severity, expressed as leaf area damaged and premature leaf abscission, for six different genotypes of Populus angustifolia (A–F) under endophyte and pathogen treatments. Prior inoculation with Penicillium resulted in two times greater leaf area damaged and premature leaf abscission. Truncatella had no effect on pathogen symptom severity.

Table 2. Results showing the effects of endophyte, genotype, and their interaction on disease severity for plants inoculated with the pathogen Drepanopeziza populi. Hierarchical model fitting and assessment of log-likelihood ratios determined P values for factors correlating with necrotic leaf area, and the Kruskal Wallis test was used to assess the significance of the main factors related to premature leaf abscission. The Penicillium treatment differed significantly from the control group for both necrotic leaf area (Mann-Whitney U, P < 0.001) and premature leaf abscission (Mann-Whitney U, P < 0.001). The Truncatella treatment did not differ from the control group for either measure of disease severity.
table image

Endophyte effects on gene expression

Of the 150 comparisons we made (2 plant genotypes × 15 genes × 5 comparisons of interest), 27 showed fold changes of greater than ±3 (Table 1). However, after correction for multiple comparisons using the Holm method, we did not have enough statistical power to assess significance in our expression data. Despite this, the trends we observed in gene expression are informative. Supporting previous work on quantitative resistance in Populus (Miranda et al. 2007, Azaiez et al. 2009), several defense-related genes were strongly up regulated (>5 fold) following pathogen inoculation (e.g., Ptgluc, PtATOSM34, PtGSTU25, PtWRKY51, PtCAD9) (Table 1). In particular, the expression of PtATOSM34, a gene coding for a defense protein, was more than 38 times greater in plants inoculated with Drepanopeziza than in control plants (Table 1). In general, gene expression results were similar for the two tree genotypes assayed. This is consistent with their similar level of resistance to the pathogen (Fig. 1).

The selected defense genes did not respond as strongly to endophyte inoculation as they did to pathogen inoculation (Table 1). In both tree genotypes, Penicillium and Truncatella inoculation resulted in modest up-regulation in a few genes that were also up regulated by the pathogen (e.g., Ptgluc). Two genes involved in secondary metabolism (PtZOG1, PtCYP92A) showed possible down regulation in both tree genotypes following Penicillium inoculation (Table 1).

For plants inoculated first with an endophyte and then with the pathogen, we found low levels of up or down-regulation of the selected defense genes in either genotype (all fold changes less than ±3) compared to plants inoculated only with the pathogen (Table 1; Fig. 2). Overall, these qRT–PCR results are generally not consistent with fungal leaf endophytes inducing or suppressing quantitative resistance.

figure image

Treatment effects (mean ± SE) on the relative expression of two defense-related protein genes, PtATOSM34 (panels a and b) and PtGlyHy17 (panels c and d), relative to the housekeeping ubiquitin gene (UBQ). Prior inoculation with endophytes did not alter the expression of either defense gene in plants responding to pathogen inoculation.

Discussion

Evolutionary ecology of plant–endophyte–pathogen interactions

While the total number of fungal leaf endophyte species globally is unknown, some estimates suggest that they could represent as many as 1 out of every 14 species of life on Earth (Ganley et al. 2004). They occur in all lineages of land plants (Stone et al. 2000), and reach impressive levels of species diversity within individual plants (Arnold et al. 2002). Whereas molecular sequencing methods have enabled major advances in our understanding of endophyte species diversity and distribution (Arnold and Lutzoni 2007, Zimmerman and Vitousek 2012), knowledge of their ecological functions lags far behind.

Controlled inoculation experiments where host plants, endophytes and/or pathogens have been manipulated provide insights into the evolutionary ecology of this three-way-interaction. Studies have primarily reported that fungal leaf endophytes reduce pathogen symptom severity in host plants (Arnold et al. 2003, Ganley et al. 2008, Raghavendra and Newcombe 2013). Further, one particular study found that these effects were consistent across host genotypes varying in quantitative resistance to the rust pathogen Melampsora (Raghavendra and Newcombe 2013). Because endophyte effects on disease overwhelmed the effect of underlying plant genotypic variation in resistance, Raghavendra and Newcombe (2013) hypothesized that fungal leaf endophytes trump quantitative resistance in the plant defense hierarchy.

Our controlled inoculation study also manipulated both endophytes and host plant genotypes, but yielded different results than those of Raghavendra and Newcombe (2013). One leaf endophyte, Penicillium, significantly increased pathogen symptom severity, while another, Truncatella, had no effect on pathogen symptom severity. This same strain of Truncatella was previously found to reduce Melampsora rust severity in Populus trichocarpa (Raghavendra and Newcombe 2013). A broader range of endophyte effects on disease (from positive, to neutral, to negative), and the differential responses of Drepanopeziza and Melampsora to the same endophyte species, are not consistent with fungal leaf endophytes as generally anti-pathogenic, but rather suggest that endophyte effects on disease will depend on the particular host, endophyte and/or pathogen species involved in this three-way interaction.

In addition to evaluating the effect of different endophytes on disease outcomes, we also varied host plant genotype. But in contrast to the result of Raghavendra and Newcombe (2013), in our study, both endophytes and plant genotype contributed significantly to pathogen symptom severity. This suggests that plant genotypic variation (i.e., quantitative resistance) can remain important for disease outcomes even in the presence of endophytes altering disease. With just a few studies in this area, clearly additional work is needed before any general inferences regarding the role of fungal endophytes in the plant defense hierarchy can be drawn.

Direct versus indirect interaction hypotheses

The mechanism by which fungal leaf endophytes alter disease outcomes is not known. One hypothesis is that leaf endophytes reduce disease severity by priming plant resistance to pathogens, as has been shown for rhizobacteria and mycorrhizal fungi (Wang et al. 2005, Weston et al. 2012). The idea of plant symbionts as “vaccines” was first suggested long ago (Beauverie 1901), and is now referred to as systemic acquired resistance (SAR) or induced systemic resistance (ISR; Van Wees et al. 2008). This mechanism could also potentially work in the opposite direction: leaf endophytes could increase disease severity by suppressing plant genetic resistance (Houterman et al. 2008). Similarly, systemic acquired susceptibility (SAS) to one pathogen can be caused by prior infection by another pathogen (Bonello et al. 2008).

Despite a widely held view of plant microbes as generalist inducers that elicit broad-spectrum resistance to pathogens and even herbivores in their plant hosts (Van Wees et al. 2008), the SAR and SAS hypotheses have not been explicitly tested for fungal leaf endophytes. Our gene expression analysis, which relied on previous microarray work on Populus and pathogens to design the marker gene assay and then used robotic qRT–PCR to assay those genes (Miranda et al. 2007, Azaiez et al. 2009), was not consistent with the hypothesis that endophytes induce plant genetic resistance. We observed no significant endophyte effect on defense-gene expression in plants responding to pathogen inoculation (Table 1). More specifically, we found no evidence that greater disease severity in Penicillium-inoculated plants was caused by systemic induced susceptibility. We acknowledge that we assessed only a small subset of defense genes in Populus (Tuskan et al. 2006), and those particular genes may not be indicative of larger trends in host resistance. It is possible that Penicillium may have suppressed a plant defense response that was unrelated to the suite of defense genes that we selected for our gene expression analysis.

An alternative mechanistic explanation for leaf endophyte effects on disease severity observed in this and other studies is direct interaction with pathogens. Leaf endophytes could reduce disease severity via direct competition for limited nutrient resources, by parasitizing pathogens, or by producing mycotoxins. For example, the fungus Eudarluca caricis has been shown to parasitize rust fungi on the outer surfaces of leaves (Nischwitz et al. 2005), and the fungal leaf pathogen Stachybotrys cylindrospora produces mycotoxins that can negatively affect other pathogens (Ayer and Miao 1993). Direct effects could also occur in the opposite direction (Kurose et al. 2012). For example, leaf endophytes could increase disease severity by producing a compound that would otherwise be limiting to pathogen growth. Lastly, we acknowledge the possibility that extraneous greenhouse fungi could have influenced the direct or indirect effects between plants, endophytes and pathogens. Indeed, some of our plants were infected with such fungi, although any incidental colonization would have been constant across all treatments. In addition, given the overwhelmingly high inoculum spore concentrations used in our experiment, we believe our study species greatly outweighed greenhouse fungi in infection success, and potential interactions with plants and/or pathogens.

While our study did not investigate possible direct effects of endophytes on Drepanopeziza, we propose a few hypotheses for how Penicillium could have directly increased disease severity. Penicillium could produce secondary compounds that facilitate pathogenesis by Drepanopeziza, or make otherwise inaccessible nutrients available to Drepanopeziza. For example, Penicillium could degrade different cell wall constituents such that Drepanopeziza and Penicillium complement or even synergize with each other in degrading the apoplast more thoroughly. These potential interactions between fungal leaf endophytes, pathogens, and their plant hosts have not been explicitly tested, but a recent study has also highlighted the ability of endophytes to increase pathogenic aggressiveness (Kurose et al. 2012).

Endophyte effects on defense gene expression

While we found no evidence that endophytes altered the expression of selected defense genes when plants were challenged with the pathogen, we did observe plant genetic responses to the endophyte alone (Table 1). Perhaps this should not come as a surprise, since endophytes produce microbe-associated molecular patterns (MAMPs) that are recognized by plant genes (Boller and Felix 2009). The most biologically significant response to endophytes was up-regulation of Ptgluc, a gene associated with cell-wall modification. Both endophytes and the pathogen elicited this response, revealing some similarity in the way the plant perceives these fungi. Future research that includes a broader range of plant genes will be poised to explicitly evaluate and compare plant genetic responses to leaf pathogens versus leaf endophytes.

Conclusion

In summary, we found that endophytes and host plant genotypes differentially affected disease outcomes. In particular, one endophyte increased disease symptoms while another had no effect on disease, and plant genotypes displayed varying levels of resistance to the pathogen. Our preliminary analysis of defense gene expression yielded no evidence that endophyte effects on pathogen symptom severity resulted from endophytes altering plant defense gene expression. Overall, these results are neither consistent with fungal leaf endophytes as generally anti-pathogenic, or the hypothesis that endophytes overwhelm plant quantitative resistance to pathogens. Instead, fungal leaf endophytes appear to have complex, species-specific effects on host-pathogen interactions. Additional work is needed to elucidate the mechanism(s) underlying endophyte effects on disease in their host plants.

Acknowledgments

We are grateful to Dave Wilson and Raymond Von Itter for their assistance at the Stock Farm Greenhouses at Stanford University. This research was supported by a DOE GCEP fellowship to PEB, a Heinz Fellowship to PEB, a NSF Graduate Research Fellowship to NBZ, and the US DOE, Office of Science, Biological and Environmental Research under contract DE-AC05-00OR22725, as part of the Plant Microbe Interfaces Scientific Focus Area. P. E. Busby and N. Zimmerman contributed equally to this work.