Methylobacterium-Induced Endophyte Community Changes Correspond with Protection of Plants against Pathogen Attack

Plant inoculation with endophytic bacteria that normally live inside the plant without harming the host is a highly promising approach for biological disease control. The mechanism of resistance induction by beneficial bacteria is poorly understood, because pathways are only partly known and systemic responses are typically not seen. The innate endophytic community structures change in response to external factors such as inoculation, and bacterial endophytes can exhibit direct or indirect antagonism towards pathogens. Earlier we showed that resistance induction by an endophytic Methylobacterium sp. in potato towards Pectobacterium atrosepticum was dependent on the density of the inoculum, whereas the bacterium itself had no antagonistic activity. To elucidate the role of innate endophyte communities in plant responses, we studied community changes in both in vitro and greenhouse experiments using various combinations of plants, endophyte inoculants, and pathogens. Induction of resistance was studied in several potato (Solanum tuberosum L.) cultivars by Methylobacterium sp. IMBG290 against the pathogens P. atrosepticum, Phytophthora infestans and Pseudomonas syringae pv. tomato DC3000, and in pine (Pinus sylvestris L.) by M. extorquens DSM13060 against Gremmeniella abietina. The capacities of the inoculated endophytic Methylobacterium spp. strains to induce resistance were dependent on the plant cultivar, pathogen, and on the density of Methylobacterium spp. inoculum. Composition of the endophyte community changed in response to inoculation in shoot tissues and correlated with resistance or susceptibility to the disease. Our results demonstrate that endophytic Methylobacterium spp. strains have varying effects on plant disease resistance, which can be modulated through the endophyte community of the host.


Introduction
Plant inoculation with nonpathogenic bacteria may induce faster defense reactions towards bacterial, fungal, or viral pathogens and environmental stresses [1,2]. The most studied resistance-inducing bacteria are soil-dwelling rhizobacteria, which usually do not enter plant tissues, but live in close proximity to, or as epiphytes on the plant root surface. These bacteria may induce systemic resistance (ISR) via jasmonic acid (JA) and/or ethylene (ET) signaling pathways of the plant [1,3]. Overall, the molecular mechanisms behind induction of resistance by rhizobacteria remain unclear. There are no consistent changes in the expression of plant genes encoding pathogenesis-related proteins, which are activated in the systemic acquired resistance (SAR) triggered by necrotizing pathogens [1,3]. Because systemic changes are not detected in the plant expression profile before challenge inoculation with the pathogen, the plant responses have been characterized as ''priming'' of the plant defense system [1,3].
Some rhizobacteria can also enter and colonize the root and above-ground tissues internally as endophytes. Endophytic bacteria are generally defined to colonize the plant interior without harming the host [4], and a number of endophytes are able to establish a mutualistic relationship with the host by promoting plant growth [2,5,6]. Each plant individual contains communities of endophytic populations in each tissue, and for example the root and shoot communities typically differ from each other [4,[6][7][8][9]. Some members of these innate (or resident) endophyte communities are transmitted through the seeds or vegetative parts from parent to progeny, and others enter the plant during the life cycle [4,6].
Results of several studies suggest that innate endophytic communities can be directly involved in plant defense [5,[10][11][12][13]. Endophytes can reduce pathogen invasion by various mechanisms, from outcompeting phytopathogens to the production of a wide range of compounds against the invader, and by induction of plant resistance. Numerous reports indicate that endophytes have direct antagonistic potency towards other microbes. This trait is essential for defense of the endophyte itself and can increase plant defense, depending on the species and cultivar [5,10,14,15]. Endophytic bacteria also have the potential to complement the inefficient antioxidative systems of some plant species with their own reactive oxygen species (ROS) -eliminating mechanisms, and/or activate the antioxidant system of the plant [16][17][18]. The outcome of the concerted action of the plant and the endophytes depends on the structure of the endophyte population.
Methylobacterium spp. are frequently encountered as endophytes and have the capacity for methylotrophy, biofilm formation, production of quorum-sensing signals, heavy metal and other stress resistance, and ISR [11,18,[28][29][30]. Recently, we found that Methylobacterium sp. IMBG290 can induce resistance in potato towards Pectobacterium atrosepticum by activation of the antioxidant system in an inoculum density-dependent manner [17]. Resistance was induced when a low density of endophytes was used for inoculation, whereas high density resulted in susceptibility to the pathogen. The inoculation with Methylobacterium sp. IMBG290 itself caused no symptoms of disease or defense reaction [17], but inoculation of endophytic M. extorquens DSM13060 has been shown to induce the expression of plant defense genes in pine [31]. The interaction between the plant, the inoculated endophyte and the environment could be very complex and might further affect the existing endophyte communities. The aim of this study was to investigate how the inoculation of Methylobacterium spp. endophytes to potato (Solanum tuberosum L) and pine (Pinus sylvestris L.) affects plant responses to different phytopathogens, and how these interactions shape the existing endophytic microflora. The endophyte community changes were analyzed by terminal restriction fragment length polymorphism (T-RFLP) -based community fingerprinting, where changes in the relative abundance of community members were detected as corresponding terminal fragments (T-RF) [32,33].

Experimental Design
The disease resistance inducing capacity of two Methylobacterium spp. strains, IMBG290 and DSM13060, was studied on their native hosts potato and pine in in vitro (both species; experiment 1) and greenhouse conditions (potato; experiment 2). The effect of Methylobacterium spp. inoculation on the innate endophytic bacterial community of the plant host was studied by T-RFLP and compared with the effect of pathogen infection. Additionally, the genotype-dependent changes were assessed in potato. Summary of the experiments is shown in Table S1. All experiments were performed with five biological replicates (n). Each replicate represents all plants grown in one pot, e.g. five plants (potato in vitro experiment), seven plants (potato greenhouse experiment) and fifteen plants (pine in vitro experiment).

Plant Material and Culture Conditions
Potato plants (Solanum tuberosum L. cvs. Blue Congo, Timo, Pito, Matilda; The Seed Potato Centre, Tyrnävä , Finland) were propagated for the in vitro experiment by micrografting and cultivated on Murashige and Skoog (MS) [34] agar medium without phytohormones in the growth chamber, where the growth conditions were 16/8 hours light/dark photoperiod (34-406mmol/m 2 s), 22uC and relative humidity of 80-85%. For the greenhouse experiment, micrografts of in vitro potato cvs. Bellarosa and Javir (Institute for Potato Research, Nemishaeve, Ukraine) were grown on perlite saturated with MS salts solution in the growth chamber before transfer to greenhouse.
For Scots pine (Pinus sylvestris L.), open-pollinated seed material (EY/FIN/M24-97-0203) was obtained from Siemen Forelia Oy, Rovaniemi, Finland. To remove epiphytic microbes, seeds were kept at 55uC for 3 days, soaked in sterile water for 12 h, then sterilized with 4% calcium hypochlorite (Sigma-Aldrich, USA) for 12 min and rinsed with sterile water three times. The seeds were cultivated in the growth chamber on sterile vermiculite saturated with deionized water.

Microbial Strains and Culture Conditions
Pseudomonas syringae pv. tomato DC3000, Pectobacterium atrosepticum, Methylobacterium sp. IMBG290 and Methylobacterium extorquens DSM13060 were cultured as described previously [17]. Phytophthora infestans was cultivated according to the Eucablight protocol (http://www.eucablight.org/EucaBlight.asp) and Gremmeniella abietina according to Petä istö & Kurkela [35]. In general, selective media were used for microbial cultivation and methanol was used as the carbon source for the Methylobacterium spp. strains. P. infestans was cultured on detached potato leaves to induce sporulation.

Plant Inoculation with Endophytic Methylobacterium spp. Strains
The plants were inoculated with endophyte strains originally derived from the same plant species, Methylobacterium sp. IMBG290 for potato and M. extorquens DSM13060 for pine, by standard procedures found most efficient for each species. For the in vitro experiment, potato cuttings were treated with a suspension of Methylobacterium sp. IMBG290 in 10 mM MgSO 4 (10 5 , 10 6 , 10 7 or 10 8 CFU mL {1 ) for 20 min. For the greenhouse experiment, Methylobacterium sp. IMBG290 cells (10 3 and 10 5 CFU g {1 in individual treatments) were re-suspended in MS salts solution and mixed with perlite (grain size: 1.5-5 mm). For pine, two-month old seedlings were watered with a suspension of M. extorquens in 10 mM MgSO 4 (10 4 , 10 5 , 10 6 , 10 7 or 10 8 CFU mL {1 ). Controls were mock-treated with 10 mM MgSO 4 . Lower inoculum densities were used for the greenhouse experiment because Methylobacterium spp. cells were inoculated directly to the nutrient-rich substrate, assuming better bacterial growth in the ambient conditions and continuous plant colonization from the substrate.

Colonization Assay
To study Methylobacterium spp. colonization by a culturing assay, a kanamycin-resistant strain of Methylobacterium sp. IMBG290, produced by transposon mutagenesis, was used for treatment of potato cv. Blue Congo plants, as described above. After four weeks, whole plants were surface sterilized with 70% ethanol and 5% sodium hypochlorite and flamed [32] prior to maceration in 10 mM MgSO 4 . Serial dilutions were plated on a selective medium [17] with kanamycin, and pink-pigmented colonies were counted.

Pathogen Inoculation
A challenge inoculation with the pathogen was performed four weeks after endophyte inoculation. Inoculation with the pathogens P. syringae pv. tomato and P. atrosepticum was performed as described earlier [17]. In the case of the greenhouse plants, the pathogen suspension was sprayed. The foliage blight test and the detached leaf test of P. infestans were performed on in vitro-grown and greenhouse plants, respectively, according to the Eucablight protocol (http://www.eucablight.org/Protocol/Protocol.asp). With respect to G. abietina infection, pine shoots were soaked in a suspension of fungal conidia (10 8 conidia mL {1 ) in 10 mM MgSO 4 with 0.025% (v/v) Silwet L-77. The typical disease symptoms (necrotic spots on potato leaves, chlorotic and necrotic leaf symptoms, sporulating lesions with leaf necrosis, and needle necrosis for P. atrosepticum, P. syringae pv. tomato, P. infestans and G. abietina, respectively) were recorded five days (except nine days for G. abietina) after pathogen infection. The percentage of symptomatic leaves per plant (or the percentage of symptomatic seedlings per pot on G. abietina) was calculated.

DNA Extraction and T-RFLP
Endophytic microbial communities were analyzed from plants four weeks after endophyte inoculation (prior to pathogen application) and three days after pathogen inoculation. Whole in vitro-grown plants and 5-7 cm long shoot tips of greenhouse-grown plants were surface sterilized as described for the colonization assay. In experiment 1, the plant material was used for a beadbeating step prior to DNA isolation [32]. In experiment 2, the plant material was macerated in a mortar with glass beads because a bead mill was not available, and the liquid extract was used for the DNA isolation. The DNA was extracted by the CTAB method [36]. For pine, the extraction buffer was supplemented with 2% polyvinylpolypyrrolidone (Serva, Germany) (w/v) and 2% b- Bacteria-specific primers, as well as restriction enzymes, were tested for their suitability for the T-RFLP analysis using the MICA software [37]. Bacterial 16S rRNA gene was amplified by PCR using the primers 799f [9] and 1520r [38], which generate different sizes of bacterial and mitochondria-derived amplicons, whereas chloroplast and other plant-derived rRNA genes are not amplified. For experiment 1, the forward primer was 59-labeled with 6FAM TM , while the reverse primer was unlabeled. For experiment 2, both primers were labeled at 59 end to increase resolving power of the method, the forward primer with NED TM and the reverse primer with VICH (primers were purchased from Applied Biosystems, Espoo, Finland). The DNA amplification, extraction of the PCR product from the gel, restriction digestion and processing for T-RFLP analysis were performed according to Sessitsch & Rasche [32]. PCR products from three individual reactions from each sample were pooled together and 50 ng of the pooled PCR product was digested with AluI and HhaI (Fermentas, Lithuania) for experiments 1 and 2, respectively. The fragments were separated on an ABI3100 sequencer using POP4 polymer and an internal size standard GS500 ROX (Applied Biosystems, UK) in experiment 1 and ABI3130 sequencer, POP6 polymer and GS600 LYZ size standard in experiment 2 (depending on which instrument was in use at each research site).

Sequence Analysis
The 16S rRNA genes were amplified as described above with unlabeled primers. Cloning and sequencing were performed as described by Sessitsch & Rasche [32]. The T-RFLP results indicated a low number of bacterial species in in vitro-grown plants, as there were only small differences between T-RFs of the cultivars and treatments, and therefore several in vitro-grown potato and pine samples were pooled together and used for DNA amplification and sequencing. To avoid sequencing of identical clones, the clones (120 for potato and 16 for pine) were first analyzed by amplified ribosomal DNA restriction analysis (ARDRA) using AluI as the restriction enzyme. Unique bands were sequenced using Big Dye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA, USA). Extension products were then purified by the ethanol/EDTA precipitation protocol and analyzed on a ABI 3100 Avant Genetic analyzer (Applied Biosystems, Foster City, CA, USA) as recommended by the manufacturer. Chimera check was done with Pintail software v.1.1 (Cardiff School of Biosciences, UK). Sequences were subjected to Basic Local Alignment Search Tool (BLAST) analysis with the National Center for Biotechnology Information (NCBI, Bethesda, MD) database.

Nucleotide Accession Numbers
The bacterial sequences were deposited in the NCBI database under accession numbers GU939191-GU939194, JN897354.

Data Processing and Statistical Analyses
The presented data of biotests are mean values +SD. The statistical significance of the differences between mean values was determined by the Student's t-test. T-RFLP data were collected using the Peak Scanner Software v1.0 (Applied Biosystems, UK). The peaks were determined at 40-450 bp using Local Southern methods for size calling [39] and the baseline was set to 5 fluorescent units. Each fluorogram was additionally checked visually to ensure proper peak capture, as well as discreteness of the closely sized peaks. Further statistical analysis was performed using the T-REX software [40]. True peaks were determined according to Abdo et al. [33]. The data were normalized and the data matrix was constructed using peak height averaged over replicates (Table S2). T-RFs occurring in less than 5% of the samples were omitted. ANOVA was performed on the data matrix, and the beta diversity and the percentages of the main and interaction effects were determined. The data were analyzed by Additive Main Effects and Multiplicative Interaction Model [40].

Results
Capacity of the Methylobacterium spp. endophytes to induce resistance was tested in different conditions towards different pathogens (Table S1). The treatments resulting in specific and profound profiles of Methylobacterium -triggered resistance were selected to study the endophytic bacterial communities by T-RFLP.

Potato (Solanum tuberosum L.)
In vitro experiment. The Methylobacterium sp. IMBG290 was tested in potato cultivars Blue Congo, Timo, Pito and Matilda in vitro against two pathogens, the bacterial soft-rot pathogen P. atrosepticum and the oomycete P. infestans. Against P. infestans only Pito cultivar exhibited some resistance (Fig. S1A), but a range of resistance levels was observed towards P. atrosepticum (Fig. 1A) depending on the cultivar and on the inoculation density of Methylobacterium sp. IMBG290. Enhanced resistance was found in the cultivars Blue Congo and Pito when Methylobacterium sp. IMBG290 was applied at low densities (10 5 to 10 6 CFU ml 21 ), but not in the cultivar Timo. Enhanced susceptibility to the pathogen was observed in the cultivar Matilda (Fig. 1A). These experiments were selected for T-RFLP analysis of the innate endophyte communities before and after endophyte and challenge inoculations. The comparison of shoot endophyte communities was made between potato genotypes treated with 10 5 -inoculation density of Methylobacterium sp. IMBG290 (Fig. 1B), and shoot and root communities were compared in Blue Congo cultivar treated with low (10 5 ) and high (10 8 ) inoculation densities (Fig. 1C).
Prior to T-RFLP analysis, the bacterial 720-bp 16S rDNA amplicons were cloned and sequenced to confirm the quality of PCR products, and bacterial sequences identical or similar with strains of e.g. Bacillus pumilus, Pseudomonas fluorescens and Ralstonia taiwanensis were identified. We attempted to identify the corresponding T-RFs, and four-bp offset was observed between the theoretical and experimental sizes of T-RFs. The main T-RFs responsible for data ordination could be matched with the following most closely related bacterial species: 62 bp (P. fluorescens), 74 bp (Cupriacidius metallidurans), 77 bp (P. atrosepticum) and 78 bp (B. pumilus). T-RFs corresponding to the Methylobacterium spp. inoculants were not among the principal components responsible for data ordination in any experiment. The average T-RF richness in all T-RFLP analyses varied from 4.4 to 7.0. In all T-RFLP profiles analyzed, there were at least two significant principal component axes (P,0.001) identified by F-test at 5% level. These first two axes captured the total variance from 88.0 to 96.99%. Beta diversity and the percentages of the main and interaction effects are shown in Fig. 1B, C, Fig. 2B and Fig. 3B.
In the in vitro experiment on Methylobacterium sp. IMBG290 inoculation of potato, tested against P. atrosepticum, the most important factor for treatment grouping was the pathogen infection, followed by the plant genotype and the Methylobacterium sp. IMBG290 inoculation (Fig. 1B). The challenge inoculation with P. atrosepticum induced significant changes in the bacterial communities. The difference between the treatments before versus after challenge inoculation was determined by the appearance of new T-RFs (72 bp, 286 bp, P,0.001 and 73 bp, P,0.01) along with the 77 bp (P,0.001) fragment originating from P. atrosepticum (Table S2). In addition, significant changes were observed for the majority of the T-RFs present, as the relative abundance of the 44bp fragment (Methylobacterium sp. IMBG290) decreased (P,0.05) and the ratio of T-RFs 62 bp (P. fluorescens) and 78 bp (B. pumilus) increased (P,0.01 and P,0.001, respectively) after challenge inoculation. These changes were dependent on the potato genotype and on the pre-treatment with Methylobacterium sp. IMBG290. The shift in the community structure observed after Methylobacterium sp. IMBG290 treatment and challenge inoculation was associated with disease resistance: in plants showing resistance (cvs. Blue Congo and Pito) a 62-bp T-RF (P. fluorescens) was detected, while a 44 bp T-RF (Methylobacterium sp. IMBG290) disappeared (Table S2).
Because the induction of resistance by Methylobacterium sp. IMBG290 was dependent on the inoculation density, the innate endophyte community structures were studied in more detail in Blue Congo shoots and roots. Blue Congo plants were previously shown to consistently express enhanced disease resistance associated with the low inoculation density [17] and therefore were selected for the T-RFLP analysis. The analysis demonstrated that the root endophyte communities of all treatments grouped together. The shoot endophyte communities differed before and after the challenge inoculation in plants treated with the low and high inoculation densities of 10 5 and 10 8 CFU ml 21 (Fig. 1C). Two T-RFs (62 bp [P. fluorescens], P,0.01 and 78 bp [B. pumilus], P,0.05) were significantly different between these treatments (P,0.001 and P,0.01 before and after challenge inoculation, respectively). The shoot endophyte communities of plants treated with the high inoculation density were similar to that of the roots, and hosted a lower bacterial richness based on the number of T-RFs (Fig. 1C, Table S2). To assess the colonization by the Methylobacterium sp. IMBG290 endophyte using the two inoculation densities, the colonies growing from the plant tissue were counted after four weeks. When the low inoculation density was used, colonization was 3.45+0.86 Ã10 3 CFU g {1 in the shoots and 1.74+0.12 Ã10 3 CFU g {1 in the roots. At the high inoculation density, the colonization was 1+0.13 Ã10 3 CFU g {1 in the shoots and 0.3+0.05 Ã10 3 CFU g {1 in the roots. The colonization using the two inoculation densities differed significantly in the roots (P,0.001) and shoots (P,0.01).
Greenhouse experiment. Since the endophyte communities of in vitro-grown plants are generally small, the changes in the community structures were further analyzed in greenhouse-grown plants. The cultivars Bellarosa and Yavir were inoculated with Methylobacterium sp. IMBG290 at two densities, 10 3 and 10 5 CFU mL 21 . No significant effect on resistance towards P. infestans was observed (Fig. S1B), but resistance towards both  Resistance of in vitro-grown Scots pine to Gremmeniella abietina induced by Methylobacterium extorquens DSM13060 (m) applied at densities 10 4 , 10 5 , 10 6 , 10 7 and 10 8 CFU ml 21 (4, 5, 6, 7 and 8) and (B) analysis of the corresponding bacterial communities in shoots and roots at 10 4 and 10 8 densities. Disease resistance data are means 6 SD (n = 5), letters indicate significant difference between treatments and control by Student's t-test (a, b and c indicate P,0.05, 0.01 and 0.001, respectively). Cluster plots generated by Additive Main Effects and Multiplicative Interaction (AMMI) analysis are constructed from five T-RFLP replicates and contain the information on beta diversity (Beta), the percentage of the main (M) and interaction (I) effects, the principal T-RFs responsible for the data ordination for each of the interaction principal components axes (IPCA1 and 2), and the percentage of variance captured by each of the axes. Different shapes indicate grouping patterns. doi:10.1371/journal.pone.0046802.g003 bacterial pathogens P. syringae pv. tomato and P. atrosepticum was enhanced in Bellarosa plants regardless of the inoculation density ( Fig. 2A, Fig. S1B). The experiment against P. syringae pv. tomato was selected for T-RFLP analysis of the innate endophyte communities. According to the analysis, the Methylobacterium sp. IMBG290 inoculation induced distinct changes in the bacterial community structure of shoots but not roots (Fig. 2C, Table S2), regardless of the fact that inoculations were done on the roots. Three T-RFs; 41 bp (F), 166 bp (F) (both at P,0.05) and 184 bp (F) (P,0.01) were significantly different in these treatments. The bacterial community of Methylobacterium sp. IMBG290-treated shoots grouped together with the root community, but after challenge inoculation these treatments segregated, T-RF 41 bp (R) (P,0.05) being significantly different between these treatments.

Scots Pine (Pinus sylvestris L.)
To explore the community changes in response to endophyte inoculation and pathogen challenge in another plant species, we performed a similar in vitro experiment on Scots pine. When pine seedlings were inoculated with the endophyte M. extorquens DSM13060, resistance against G. abietina was observed at the high inoculation density (10 8 CFU ml 21 ), whereas seedlings treated with low density (10 4 CFU ml 21 ) became more susceptible to the pathogen (Fig. 3A). When the innate endophyte population structures were studied in these samples, treatment with M. extorquens DSM13060 significantly decreased the relative abundance of T-RF 44 bp (P,0.05) in the shoots (Fig. 3B, Table S2). Furthermore, the T-RF 62 bp (P. fluorescens) significantly decreased (P,0.05) when the low density of M. extorquens DSM13060 cells was used. After challenge inoculation, the T-RF of 221 bp disappeared (P,0.05) and the T-RF of 74 bp (C. metallidurans, P,0.05) disappeared in the root community from the samples treated with the low density of M. extorquens DSM13060 cells.

Discussion
The present study was aimed to elucidate the changes occurring in innate endophytic communities associated with inoculation of beneficial endophytes, pathogens, and the resulting disease tolerance or susceptibility. The Methylobacterium spp. inoculant strains had a variable effect on plant disease resistance. These results can be explained largely by the fact that the tested host plant species, as well as pathogens used, differ from each other by defense mechanisms and disease strategies, respectively. Potato is an annual dicotyledonous crop plant that is propagated clonally, and relies largely on inducible defense responses [41]. Scots pine is a wind-pollinated coniferous tree having an efficient constitutive defense system, such as preformed resins and polyphenols [42,43]. Pectobacterium atrosepticum is a necrotrophic pathogen, which elicits both salicylic acid (SA)-and JA-ET-dependent defense responses [44], whereas Pseudomonas syringae pv. tomato is a hemibiotroph that elicits the SA-defense signaling cascade and inhibits the JA(ET)-dependent defense responses [45], and Gremmeniella abietina is a fungal pathogen. Such large scale of plant-pathogen combinations were used to reveal whether endophyte inoculation induces a change in the innate endophyte community universally with respect to resistance or susceptibility to a pathogen.
Presence of a pathogen affects the innate endophyte community structure of a plant [8,10,23,46], and in this paper, we show that endophyte inoculation can modulate the communities, to result in a divergent structure after pathogen challenge. Even though results of T-RFLP analysis are semi-quantitative due to PCR bias [47], statistically significant changes in the structures were observed in different combinations of plant species, pathogen, and Methylobacterium spp. inoculation. The endophyte community analysis indicated that the structures were dependent on Methylobacterium spp. inoculation density, plant genotype, and pathogen challenge. Most importantly, Methylobacterium spp.inoculated plants challenged with the pathogen showed highly different endophyte communities compared to uninoculated control plants. Typically, the Methylobacterium spp. inoculation alone induced only small changes in the endophyte communities. However, the changes in the community structures induced by pathogen inoculation were different between controls and Methylobacterium spp.-treated plants, indicating that Methylobacterium spp. inoculation had modified the endophyte community responses towards pathogen challenge. Specifically, some of the appearing or disappearing T-RFs could be associated with resistance or susceptibility of the plant to the disease. For example, the relative abundance of P. fluorescens increased significantly after Methylobacterium sp. IMBG290 treatment and correlated with enhanced resistance towards P. atrosepticum. In the greenhouse experiment against P. syringae pv. tomato, three T-RFs were significantly different in endophytic bacterial communities of potato inoculated with Methylobacterium sp. IMBG290, coinciding with less leaf disease symptoms. When pine seedlings were inoculated with the endophyte M. extorquens DSM13060, one significantly different T-RF correlated with pathogen resistance.
The endophyte-induced changes occurred mainly in the shoot communities regardless of endophyte inoculation being done on the roots, indicating systemic effects. Methylobacterium sp. IMBG290 is an active colonizer of potato, residing in the leaves and stems of potato shoots [11], and M. extorquens DSM13060 has been isolated from pine shoot tips [48]. Neither of the Methylobacterium spp. endophyte strains used for plant inoculation have direct antagonistic activity towards the pathogens tested [49]. Microorganisms, irrespective of their mode of action, can modulate plant responses [16][17][18] and thereby affect innate endophyte communities. We have earlier analyzed the defense responses of in vitro-grown potato cultivar Blue Congo when inoculated at high and low inoculation densities with Methylobacterium sp. IMBG290 against P. atrosepticum [17]. In those studies the same results were obtained, low inoculation density resulting in resistance and high density leading to susceptibility towards the pathogen, but no obvious mechanism behind the phenomenon was identified. The antioxidant system of potato was moderately activated by the endophyte inoculation specifically at low inoculation density [17]. Modulation of the antioxidant system is important for plant defense [50], but it is not a direct defense mechanism, i.e. it does not kill or inhibit the growth of the pathogen. The infection by microbes induces production of reactive oxygen species, which in turn, activates the antioxidant system, including ascorbate and glutathione reducing enzymes [51]. Generally, the plant redox status is considered a candidate factor maintaining status quo in the plant-endophyte interaction (see e.g. [52]). Rocking this balance could affect the growth of individual, innate endophyte species inside the plant, seen as changes in the community structure by T-RFLP.
As endophyte communities were significantly different in resistant or susceptible plants in both potato and pine, there could be a (direct or indirect) link between the presence or absence of certain community members and disease tolerance. We have earlier seen that an inoculated bacterial strain can dramatically increase the numbers of specific endophytes inside the plant, even to a point of inducing outgrowth [5,11,26]. In vitro-grown plants typically host a far lower diversity of endophytes than field-grown plants, and the innate endophytic community is cultivar-specific, potentially including beneficial bacteria as well as latent pathogens [4]. Therefore, changes occurring in the plant due to external factors can affect growth of both beneficial and potentially harmful microorganisms, either positively or negatively [13,21,22]. If the populations of latent pathogens grow, the defense system of the plant host becomes compromised [53][54][55]. On the other hand, favorable conditions for growth of beneficial or protective endophytes due to endophyte inoculation can increase plant resistance. In potato cv. Blue Congo, sequence analysis indicated the presence of strains closely related to P. fluorescens and Bacillus pumilus that have earlier been reported as plant growth-promoting bacteria capable of triggering ISR [1,3,4].
Recently Doornbos et al. [56] reported that Arabidopsis mutants compromised in ISR differ from wild-type plants by their rhizosphere bacterial microflora, suggesting a potential link between the plant-associated bacterial community and the development of ISR. As endophytes have several ways of increasing plant resistance towards pathogens, the outcome of their concerted action defines the type of plant responses. Whether the endophyte-induced changes in the population structures of innate endophytic communities are responsible for plant resistance or susceptibility towards the pathogen, or simply a reaction to the plant responses, cannot be exclusively determined in this study. However, the study shows that changes in the endophyte communities correspond with plant responses and therefore the significance of the innate plant endophyte microbiome should be considered in future studies on plant defense. Figure S1 Effect of Methylobacterium inoculation on disease resistance of potato. Resistance of (A) in vitro-grown potato cvs. Blue Congo, Timo, Pito, Matilda (B, T, P, M) to Phytophthora infestans and (B) greenhouse-grown potato cvs. Bellarosa (Br) and Yavir (Y*) towards P. infestans (Pi) and Pectobacterium atrosepticum (Pa). Methylobacterium sp. IMBG290 was applied at densities of 10 5 , 10 6 , 10 7 and 10 8 CFU ml 21 (5, 6, 7 and 8 respectively) to in vitro-grown plants (A) and at densities of 10 3 and 10 5 CFU mL {1 (3 and 5 respectively) to greenhouse-grown plants (B). Control -mock-treated plants. Data are means 6 SD (n = 5). Letters indicate significant difference between the treatments and control by Student's t-test (a, b and c indicate P,0.05, 0.01 and 0.001 respectively). (TIF)  (Fig 1B, 1C, 3B) and five (Fig. 2B)