ABSTRACT
Microbial populations residing in close contact with plants can be found in the rhizosphere, in the phyllosphere as epiphytes on the surface, or inside plants as endophytes. Here, we analyzed the microbiota associated with Espeletia plants, endemic to the Páramo environment of the Andes Mountains and a unique model for studying microbial populations and their adaptations to the adverse conditions of high-mountain neotropical ecosystems. Communities were analyzed using samples from the rhizosphere, necromass, and young and mature leaves, the last two analyzed separately as endophytes and epiphytes. The taxonomic composition determined by performing sequencing of the V5-V6 region of the 16S rRNA gene indicated differences among populations of the leaf phyllosphere, the necromass, and the rhizosphere, with predominance of some phyla but only few shared operational taxonomic units (OTUs). Functional profiles predicted on the basis of taxonomic affiliations differed from those obtained by GeoChip microarray analysis, which separated community functional capacities based on plant microenvironment. The identified metabolic pathways provided insight regarding microbial strategies for colonization and survival in these ecosystems. This study of novel plant phyllosphere microbiomes and their putative functional ecology is also the first step for future bioprospecting studies in search of enzymes, compounds, or microorganisms relevant to industry or for remediation efforts.
INTRODUCTION
Andean high-mountain environments have been reported as diversity hot spots, mainly because of their endemic species (1). The Paramos ecosystems within the Neotropical Andes consist of isolated, high-elevation areas that are reported to be the world's fastest-evolving biodiversity hot spot (2). These ecosystems are exposed to harsh environmental conditions, such as high incidence of UV radiation (3) and daily shifts in temperatures that impose selective pressure on native plants and their associated microbiota (4). In particular, the phyllosphere of endemic plants from Paramos represents a unique ecosystem for microbial communities with diverse and distinctive abilities to survive under conditions considered extreme for other forms of life.
The phyllosphere refers to all aboveground surfaces of any plant, including leaves, stems, buds, flowers, and fruits (5). It acts as a landing stage where spores or other propagules can develop and multiply (6) and has been reported as probably the largest ecosystem on earth colonized by microorganisms, mainly bacteria and fungi (7). Interest in studying the phyllosphere microbiota is growing due to its potential in terms of microbial interactions, survival under harsh environmental, nutrient or humidity conditions, and bioprospecting. The most emblematic plant in the Colombian Paramos is known as “frailejón,” a plant endemic to the region and belonging to the genus Espeletia (8, 9). These plants have unique adaptations that enable them to resist exposure to UV light and daily temperature changes; they are in close relation with more than 125 animal species (10) and are important for soil health and the capacity of these ecosystems to retain and regulate water availability and to store carbon (11). Based on the developmental stage, these plants can be separated into different “tiers” (12). The upper tier is composed of young leaves somewhat protected from the environment, the middle tier (midtier) is composed of fully mature leaves exposed to environmental conditions, and the necromass tier is composed of senescent leaves (Fig. 1). Finally, the root soil environment, which is humid, tends to have an acidic pH, and is rich in carbon (11), can be very different from that of the plant phyllosphere.
Overview of sampling site and Espeletia sp. morphology. (A) Sampling site (El Coquito Hot Spring, 04°52′27″N, 75°15′51.4″W). (Adapted from Google Earth [copyright 2015 DigitalGlobe and Google, Image Landsat].) (B) Espeletia sp. morphology. (C) Sampling distribution per individual collected. Y, young leaves; M, mature leaves; N, necromass; R, roots; EP, epiphyte; ED, endophyte.
Both environmental conditions and the host must influence the functional ecology of plant microbial communities (13), driving their composition and interactions. Microbial communities associated with plants such as Espeletia (i.e., epiphytes and endophytes) should therefore reflect the adaptations to the environmental conditions to which they are exposed and have the metabolic plasticity required for them to thrive. The different plant tiers also represent various microenvironments in which microbial communities should be taxonomically diverse or at least metabolically differentiated. Thus, the ecology and molecular and functional diversity of microbial populations associated with Espeletia plants may present key insights into understanding how microorganisms interact with and adapt to these extreme habitats. Based on these hypotheses, we analyzed Espeletia plant-associated microbial communities, which remain largely unknown. Some studies done by culturing bacteria and fungi, including mycorrhizae, indicate that many microorganisms are commonly associated with these plants and are probably important for nutrient availability and decomposition of biomass (14–16). Other work has focused on endophytic fungi and their biocontrol and biotechnological potential (12, 17). In this work, we used culture-independent means, 16S rRNA gene sequencing and GeoChip 5.0 functional microarrays, to address community structure, diversity, and functional potential using samples from different plant tiers. The description of bacterial communities allowed us to compare microbial structures across the plant and to highlight microbial contributions to particular geobiological processes and the potential of these communities in terms of metabolic plasticity and adaptation.
MATERIALS AND METHODS
Study site and sampling.Sampling was performed at El Coquito hot spring (04°52′27″N, 75°15′51.4″W) in the Natural National Park Los Nevados in Colombia (http://www.parquesnacionales.gov.co). Leaves were sampled from Espeletia hartwegiana according to reported methodologies (6, 18) with some modifications. Briefly, leaves (50 to 100 g) from three individuals were taken from three different tiers: (i) upper tier, young leaves; (ii) midtier, mature and fully developed leaves; and (iii) lower tier, senescent leaves or necromass. Because of their ecological importance, only three individuals were sampled, in close proximity (within 10 m), to avoid possible environmental effects. Two sets of leaves were taken from each individual, one for the epiphyte community analysis and one for the endophyte community. Roots (1 to 5 g) were taken from two different plants with a sterile scalpel (Fig. 1).
DNA extraction.Endophyte DNA was isolated according to previously reported methodologies, with some modifications (12, 19). Briefly, the plant tissue was surface sterilized by washing with sterile H2O to remove dirt, placed in NAP buffer (124 mM Na2HPO4), and vortexed for 1 min to dislodge epiphytes. Leaves were then shaved to remove the pubescence on their surface, which facilitates the subsequent sterilization process (12), washed with sterile H2O, submerged in 90% ethanol (60 s), 5.25% sodium hypochlorite solution (6 min), and 70% ethanol (30 s), and finally rinsed with sterile distilled H2O. Sterilization was checked by taking an imprint of the leaf on malt extract medium (12) and incubating at 25°C. One gram of the previously treated material was cut into 0.1- to 0.5-mm sections, placed in a 1.5-ml Eppendorf tube containing 1 g of sterile 0.1-mm-diameter glass beads and 1 ml TE (10 mM Tris, 10 mM EDTA, pH 8.0), and homogenized in a Mini-BeadBeater (BioSpec Products) for 5 min. DNA was extracted using the PowerSoil DNA isolation kit (Mobio Laboratories, Carlsbad, CA, USA), according to the manufacturer's instructions.
We obtained epiphyte DNA by first releasing bacteria from the surface of leaves by submerging 10 to 20 g of healthy plant tissue in 100 ml of release buffer (0.1 M potassium phosphate, 0.1% glycerol, 0.15% Tween 80, pH 7.0) and vortexing for 7 min (13, 20). The remaining bacteria were dislodged from the leaves with the help of a sterile swab, and the buffer was then filtered through a 0.2-μm-pore filter. DNA was extracted using the PowerSoil DNA isolation kit.
Combined epiphyte and endophyte DNA was extracted from root and necromass samples by cutting the tissue into 0.5- to 1-cm fragments, which were placed in 25 ml of release buffer in a 50-ml tube and homogenized by vortexing for 10 min. The buffer was filtered through a 0.2-μm-pore filter, and the filters were used for DNA extraction using the PowerSoil DNA isolation kit. All DNA extractions were quantified using a Qubit 2.0 fluorometer (Life Technologies Corporation, Carlsbad, CA, USA). In total, we obtained six epiphyte and six endophyte DNA extractions, corresponding to the upper and middle tiers from three plant replicates, three DNA extractions for the necromass tier, one for each replicate, and two for the roots.
16S rRNA gene amplification and sequencing.The V5-V6 hypervariable region of the 16S rRNA gene of Bacteria and Archaea was amplified with primer 799F (5′-AACMGGATTAGATACCCKG-3′), which minimizes contamination from chloroplast DNA and amplifies a mitochondrial product that is larger and thus easier to separate and differentiate from the microbial amplified products (21), and the reverse primer 1050R (5′-AGYTGDCGACRRCCRTGCA-3′) (22). DNA concentration was adjusted as previously reported (13) and used in 25-μl PCR mixtures containing DNA (10 ng for endophytic fraction or 1 ng for epiphytic fraction), 2.5 μl 10× AccuBuffer [600 mM Tris-HCl, 60 mM (NH4)2SO4, 100 mM KCl, 20 mM MgSO4, pH 8.3], 2 μl 10 mM deoxynucleoside triphosphate (dNTP) mix, 0.5 μM of each primer, and 5 units of Accuzyme DNA polymerase (Bioline USA Inc., Taunton, MA, USA). Cycling conditions were 94°C for 2 min, followed by 30 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 45 s and a final extension of 72°C for 10 min. PCR products were separated on a 3% agarose gel and purified based on amplicon size difference using the QIAquick gel extraction kit (Qiagen, Valencia, CA, USA) when necessary. The remaining samples were purified using the QIAquick PCR purification kit (Qiagen, Valencia, CA, USA). DNA was quantified using a Qubit 2.0 fluorometer (Life Technologies Corporation, Carlsbad, CA, USA), and amplicons were sequenced on an Illumina MiSeq machine with a paired-end protocol (PE-250; Molecular Research MR DNA, Shallowater, TX, USA).
Sequence analysis.Forward and reverse Illumina paired-end sequence reads were assembled using the FastqJoin software (23) to obtain longer sequences, using a minimum overlap of 150 bp, given the average length of our reads (∼250 bp) and the total length of the V5-V6 region (∼300 bp). Assembled reads were then analyzed using UPARSE (24) and QIIME 1.8 (25). The reads were first filtered by quality at a mean quality level of >25 and then based on the maximum expected error (<0.5) (24) and separated according to the barcodes, and sequences with mismatches in the primer or barcode or with ambiguous bases were excluded. Operational taxonomic units (OTUs) were picked at 97% sequence similarity level using the UPARSE-OTUref algorithm, which includes a de novo and reference-based chimera-checking step. In this step, singletons were removed to avoid including sequencing errors as recommended in the UPARSE pipeline. Taxonomy assignment was performed with QIIME 1.8, using the Greengenes reference database and taxonomy (version 13_8) (26). The phylogenetic overlap of 97% between samples was computed using the core microbiome script in QIIME 1.8. Sequences assigned to “chloroplast” or “mitochondria” were removed from the data set (approximately 2.8%). For tree-based analysis, representative sequences for each OTU were aligned against the Greengenes core data set (27) using PyNAST (28). The approximately maximum likelihood phylogenetic tree was built using FastTree (29).
Diversity analysis.Alpha and beta diversity analyses were performed using both QIIME 1.8 (25) and the R package Phyloseq (30). We assessed the richness in the different samples using the Chao1 index. For diversity estimates, we computed Shannon entropy and Gini-Simpson indices, which take into account richness and abundance. These indices are sensitive to abundant species and to abundant and rare species, respectively (31). They were transformed into effective numbers of species for comparison among samples (32). In order to measure and compare diversity indices (i.e., beta diversity), we normalized the OTU tables using the variance stabilization function from DESeq2 to account for different library sizes and sequencing depth (33) wrapped in Phyloseq (30), as reported earlier (34). Normalized tables were then used to estimate the differential abundance of OTUs between microbial populations using DESeq2 (33). Principal coordinates analysis (PCoA) and hierarchical clustering were computed based on UniFrac distances. Alpha and beta diversity plots were made using Phyloseq and SPSS Statistics 22 (IBM, New York, NY, USA).
Functional analysis.Metabolic pathways were predicted using the software PICRUSt (35) after performing an additional closed-reference OTU picking pipeline using QIIME 1.8. The resulting KEGG orthologies were further processed using HUMAnN (36), which transforms 16S rRNA-based predictions into gene and pathway summaries. This summary was then visualized using the Galaxy web-based application GraPhlAn (37–39). Total metagenomic DNA was also analyzed using the GeoChip 5.0 functional microarray (40, 41) (Glomics, Norman, OK, USA). The normalization was performed as previously described (42). Briefly, (i) spots with a signal-to-noise ratio of <2 were removed, (ii) the normalized intensity of each spot was calculated by dividing the signal intensity of each sport by the mean intensity of all the spots of the array, and (iii) at least two of three spots were required for a gene to be positive (singleton removal). The normalized hybridization output was organized based on functional categories, singletons were removed, and data were analyzed using the multivariate statistical software package PRIMER-E v6 (Plymouth Marine Laboratory). Principal coordinate ordinations were used to visualize Bray-Curtis similarities. Analysis of similarity (ANOSIM) was used to assess the confidence in the similarities observed. Functional categories in each sample were compared using analysis of variance (ANOVA) (F statistic) and plotted using SigmaPlot v12 (Systat Software, Inc., San Jose, CA, USA).
RESULTS
Microbial community composition and structure.The phylogenetic composition of microbial communities associated with Espeletia sp. was analyzed using 17 DNA samples, isolated from four different plant tiers, by amplifying and sequencing the V5-V6 region of the 16S rRNA gene (Fig. 1). A total of 3,041,094 16S rRNA paired-end sequences were obtained, assembled into single sequences, demultiplexed, and quality filtered, yielding a total of 1,762,044 high-quality sequences. After the OTU clustering and chimera-checking steps, we obtained 6,744 to 102,266 sequence reads per sample and a total of 1,548 OTUs. Samples had an approximate coverage in terms of OTUs of 79.8%, calculated using observed OTUs and Chao1 richness, estimates that ranged from 656 to 1,165 (see Table S1 in the supplemental material). OTU accumulation curves showed that most samples were well characterized with our sequencing efforts, in particular the epiphyte leaf communities and the necromass and root samples (see Fig. S1 in the supplemental material), the last two including combined endophyte and epiphyte communities. The endophyte leaf communities were less well characterized, even though coverage was good (Good's coverage, >96%) (see Table S1 in the supplemental material). Shannon and Simpson diversity indices (see Table S1 in the supplemental material) were transformed into effective numbers of species to compare samples (32). Both transformations showed that diversity was significantly higher for the necromass and root fractions (P < 0.05) than even the combined leaf endophyte and epiphyte fractions (Fig. 2).
Community diversity indices. Shapes indicate the means for the transformed Shannon (A) and Simpson (B) indices for each community (YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root), and of the combined tiers YED + YEP (upper tier) and MED + MEP (middle tier). Variations among communities were tested with one-way ANOVA. Pairwise testing was corrected using Tukey's post hoc test; letters (a, b) represent statistically significant differences (P < 0.05). The error bars show the standard errors (SE) for the replicates after the transformation to effective number of species.
Taxonomic affiliation was examined using the uclust-based consensus taxonomy classifier. There were few Archaea (0.1 to 16%) in our data set, i.e., Crenarchaeota (marine benthic group A [MBGA]), Euryarchaeota (Thermoplasmata), and Thaumarchaeota, and a low percentage of unclassified OTUs (<4% across samples) that might represent new microorganisms. There was also a marked difference in OTU abundances across samples (see Fig. S2 in the supplemental material), and of the 1,548 OTUs identified, only 174 were shared by all samples (see Fig. S3 in the supplemental material). Among these core community OTUs, the most common corresponded to Acinetobacter sp., “Candidatus Baumannia” species, Burkholderia sp., Erwinia sp., Hymenobacter sp., Klebsiella sp., Pseudomonas sp., Propionibacterium sp., and Sphingomonas sp. The most abundant phyla in all plant samples were Acidobacteria, Actinobacteria, Bacteroidetes, Crenarchaeota, Firmicutes, and Proteobacteria even when subsampling at the lowest number of reads obtained, 6,700 sequences per sample (data not shown) (Fig. 3). The phylum Proteobacteria was more abundant in endophyte and epiphyte communities of both young and mature leaves (71.38% to 85.83%) than in the necromass and root fractions (51.94% and 45.40%, respectively). Within this phylum, there were marked differences in relative abundances across samples at the class level, such as the higher abundance of Gammaproteobacteria in young and mature leaves (both endophytes and epiphytes) than in the root or necromass tiers (Fig. 3). However, the necromass and root fractions had a higher relative abundance of the phylum Acidobacteria (11.81% and 20.28%, respectively) than did the leaf habitat (2.40% to 3.66%), and there were more Bacteroidetes and Alphaproteobacteria in the necromass fraction (21.05%) and Crenarchaeota in the root fraction (13.93%). Actinobacteria showed a relative abundance ranging from 4.07% to 12.02% across samples.
Relative abundance of bacterial phyla associated with Espeletia sp. YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root. The Proteobacteria phylum has been replaced by the corresponding classes (Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, and Deltaproteobacteria).
A PCoA using weighted UniFrac distances was used to assess if microbial communities clustered according to their plant microenvironment (Fig. 4). The root fraction was separated from the phyllosphere (leaf endophyte and epiphyte) and necromass samples. All leaf samples clustered together, indicating that they were similar and differed from the necromass and root fractions, as confirmed by hierarchical clustering (see Fig. S4 in the supplemental material). The similarity between epiphyte and endophyte leaf communities was also evident by the fact that the only differences found corresponded to changes in abundance of few OTUs belonging to Acidobacteria, Bacteroidetes, and Proteobacteria (see Fig. S5 in the supplemental material). Despite being part of the phyllosphere, the necromass had increased abundance of some OTUs from the phyla Acidobacteria, Actinobacteria, Bacteroidetes, Firmicutes, Proteobacteria, and candidate divisions FBP and TM7 that distinguished this fraction from leaf epiphyte and endophyte communities (see Fig. S5 in the supplemental material). The differences observed for the root fraction were driven by OTUs mainly from the phyla Acidobacteria, Actinobacteria, AD3, Armatimonadetes, Bacteroidetes, Chlamydiae, Chloroflexi, Crenarchaeota, Firmicutes, Proteobacteria, TM6, TM7, and Verrucomicrobia. These phyla were significantly more abundant in the root fraction than in the phyllosphere communities (i.e., leaves and necromass; P < 0.01) (see Fig. S5 in the supplemental material), a finding consistent with the aforementioned higher diversity measurements for the root fraction. Interestingly, Corynebacterium sp., Pseudomonas sp., and Rothia sp. were more abundant in both the epiphyte and endophyte fractions than in the necromass and root fractions.
Principal coordinate analysis (PCoA) plot based on weighted UniFrac distances. The two axes represent 51.9% of the variation in the samples. Individual points represent the replicates for each sample. YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root.
Metabolic profiles vary according to plant microenvironment.To further understand how Espeletia plant microhabitats affect microbial communities, we analyzed the samples using a bioinformatics 16S rRNA-based prediction approach and a functional microarray. We predicted the metabolic potential of each of the communities using PICRUSt (35). Based on this information, we constructed a tree depicting the main metabolic pathways, categorized by the KEGG database (43), combined with a heat map of gene “abundances” for each category in every sample (see Fig. S6 in the supplemental material). The predicted functional categories showed genes involved in xenobiotic degradation, in carbohydrate, lipid, amino acid, and cofactor metabolism, in biosynthesis of secondary metabolites, and in energy metabolism. Interestingly, genes involved in replication and repair pathways, including base excision repair (alkB), DNA replication, homologous recombination (radA, radB, radC, recF, recO, recN), mismatch repair (mutS, mutS2, mutL, mutH, vsr), and nucleotide excision repair (mfd) systems, were also predicted to be present in the samples. All these mechanisms are involved in repair of DNA damage caused by UV (44). The degradation of xenobiotics and the biosynthesis of secondary metabolites such as antibiotics, alkaloids, polyketides, and terpenoids were also a relevant feature of this prediction (see Fig. S6 in the supplemental material). Although the 16S rRNA-based predictions did not show any clear separation between the communities, we found statistically significant differences between the endophyte community and the root and necromass communities (ANOSIM, global R = 0.359, P < 0.05, n = 17, permutation = 9,999) (Fig. 5A).
Analysis of microbial metabolic potential. PCoA based on Bray-Curtis similarities calculated for metabolic profiles predicted with PICRUSt (the two axes represent 85.5% of the variation) (A) and metabolic profiles derived from the GeoChip analysis (the two axes represent 65.6% of the variation) (B). Individual points represent the replicates for each sample. YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root.
To obtain information regarding the actual functional profiles of these microbial communities, we used the GeoChip 5.0 functional microarray, which contains over 167,000 probes covering more than 395,000 coding sequences from approximately 1,500 functional gene families involved in several biogeochemical, cellular, and ecological processes. All microbial communities had genes involved in carbon fixation and degradation pathways, nitrogen, phosphorus, and sulfur metabolism, organic remediation, secondary metabolism, virulence-related genes, and environmental stress responses. Protection against high UV exposure, which can occur in the exposed microenvironment of the leaves, was evident by the presence of genes involved in pigment production (acsF, bacteriorhodopsin gene, bchG, blh, ctrW, pebB, among others) and in repair of DNA damage, as mentioned above, caused directly by UV or by UV-induced production of agents such a reactive oxygen species (ROS) (44). There were statistically significant differences among the communities analyzed (ANOSIM, global R = 0.524, P < 0.05, n = 16, permutation = 9,999). Pairwise comparisons showed that the epiphyte, root, and necromass communities shared greater similarity, while the endophyte communities were significantly different from the rest of the communities (P < 0.05). Compared to the ordination analysis performed using the 16S rRNA-based predictions, the GeoChip analysis separated the endophyte communities from the rest (Fig. 5B).
Nutrient utilization and survival.Even though gene families for several metabolic pathways were detected with GeoChip (see Fig. S7 in the supplemental material), we focused on functions relevant to microbial growth and survival that could provide insight regarding strategies for adaptation to the harsh high-mountain plant environment of Espeletia sp. Pathways associated with carbon cycling were abundant and present in all communities associated with the plant. There was evidence of autotrophic capacity due to genes involved in six carbon fixation pathways (see Fig. S8 in the supplemental material), of the capacity to form carboxysomes (bacterial microcompartments that contain enzymes involved in carbon fixation), and of the ability to carry out C1 metabolism such as methanogenesis, mainly in root and epiphyte fractions, and methane oxidation (see Fig. S8 in the supplemental material). Heterotrophic metabolism was also found within our samples. The pathways involved in carbon degradation were highly abundant in all samples, but more so in the epiphyte, necromass, and root fractions (P < 0.05) (Fig. 6). Functions ranged from the utilization of labile carbon sources such as starch, the most abundant pathway overall, to the metabolism of recalcitrant polymers such as lignin and other plant-derived compounds such as cellulose, hemicellulose, pectin, and terpenes. There were also genes for the degradation of organic aromatic compounds that, in some cases, had a higher relative abundance than those involved in the metabolism of the more common cellulose, hemicellulose, and pectin plant components (see Fig. S9 in the supplemental material). Taxa abundant in the necromass fraction, Methylobacterium sp., Pedobacter sp., Sphingomonas sp., and Spirosoma sp., and in roots, “Candidatus Koribacter” and “Candidatus Solibacter,” were associated with increased carbon utilization genes. Nitrogen cycling pathways, such as assimilatory nitrate reduction, denitrification, nitrification, nitrogen fixation, ammonification, and anammox pathways, were also present in all fractions, although at lower levels than pathways for carbon cycling (see Fig. S10 in the supplemental material). The root fraction had a higher abundance of genes for assimilatory and dissimilatory nitrate reduction, nitrification, and nitrogen fixation, while pathways for ammonification, denitrification, and anammox were more abundant in the necromass and epiphyte fractions.
Mean normalized signal intensity of several carbon-degrading genes. Genes are clustered by major carbon sources: starch, hemicellulose, cellulose, camphor, chitin, cutin, inulin, lignin, pectin, terpenes. The signal intensity was normalized by the mean intensity of the microarray. Mean values of samples were plotted with their respective SE. Variations among communities were tested with one-way ANOVA. Pairwise testing was corrected using Tukey's post hoc test; letters (a, b, c) represent statistically significant differences (P < 0.05). YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root.
The functional microarray analysis also detected genes indicative of microbial capacity to respond to adverse environmental conditions. These include responses to acidic and alkaline shock, cold and heat shock, glucose, phosphate, and nitrogen limitation, osmotic stress, oxidative stress, oxygen limitation, and the stringent response (data not shown). Genes involved in the production of bacterial pigments and of enzymes involved in the removal of ROS were also present (Fig. 7). Finally, genes involved in the production of antibiotics, associated mainly with Actinobacteria and Proteobacteria, were also evident in all samples. Moreover, all samples had a remarkable abundance of antibiotic resistance genes belonging to several phyla, as well as the presence of plant hormone production genes.
Mean normalized signal intensity of UV resistance-related genes. Genes are clustered by major functions: antioxidant enzymes, oxidative stress response, and pigment production. The signal intensity was normalized by the mean intensity of the microarray. Mean values of samples were plotted with their respective SE. Variations among communities were tested with one-way ANOVA, and letters (a, b) indicate statistically significant differences (P < 0.05). Pairwise testing was corrected using Tukey's post hoc test; letters represent statistically significant groups. YED, young endophyte; MED, mature endophyte; YEP, young epiphyte; MEP, mature epiphyte; N, necromass; R, root.
DISCUSSION
The present study of the phylogenetic and functional profiles of Espeletia sp.-associated microbial communities examines novel host-associated microbiomes and reveals features of their functional ecology that can provide useful insight regarding strategies for microbial plant colonization and survival. The Espeletia sp. microbial communities showed the dominance of few bacterial phyla, which also tend to dominate in phyllosphere microbiomes from Arabidopsis thaliana (13), Thlaspi geosingense (45), potato (46), soybean (47), almond drupes (48), and several tree species (18, 49), among others (50). In addition to this similarity at the phylum level, Espeletia communities also contained taxa, such as Bacillus, Burkholderia, Methylobacterium, Pseudomonas, Sphingomonas, and Xanthomonas, that are an important fraction of the core community in several plants (7, 13, 50). It is difficult to assess, however, if these bacterial communities are stable (51–53), and sampling at different months throughout the year could improve our capacity to discriminate between permanent and temporary phyllosphere residents. Despite this difficulty, the fact that the most abundant phyla and bacterial species are shared across plants suggests that there is a widespread “global” core community adapted to life in the phyllosphere. In addition, the presence of unclassified OTUs, both in this and previous studies, also suggests that there is still more diversity to be found in the plant phyllosphere (54–57).
Differences in plant-associated microbiomes are probably determined by changes in relative abundances or the presence of rare bacterial taxa with important ecological functions. Given the different leaf microenvironment of endophytic and epiphytic microorganisms, we expected the corresponding community structures to vary. However, these two communities were very similar in taxonomy, they shared most OTUs, and were no different in richness or diversity indices, in contrast to what has been reported for A. thaliana leaf endophyte and epiphyte communities (13). These inconsistent results could be attributed to the use of a small number of plant samples or of a different plant species or to differences in the protocols used for obtaining phyllosphere community DNA. However, the differences in terms of diversity and relative abundance of taxa between Espeletia leaf phyllosphere communities and necromass and root communities suggest niche-selective properties. Overall, the presence of shared taxa across plant tiers indicates that these communities are interconnected and that bacteria can travel from the rhizosphere toward the leaves and vice versa, as has been reported previously (13, 58, 59). The high diversity observed in terms of bacterial composition and functional potential for both the necromass and root fractions is consistent with the fact that these communities are shaped by factors that contrast with those affecting phyllosphere communities. While the leaf surface is considered a water- and nutrient-limited environment (6), both the necromass, which consists of senescent material, and the more stable and exudate-rich root microenvironment contain more readily available nutrients (13, 60). In addition, both the necromass and soils environments are more protected from environmental conditions such as UV exposure and desiccation (50).
The functional analysis of the various samples obtained from Espeletia plants revealed metabolic capabilities relevant for microbial survival in these ecosystems. Both the 16S rRNA-based prediction and the GeoChip 5.0 analysis, which has been used to assess the metabolic pathways in other ecosystems (40, 42, 61–63), showed that these communities possess a diverse and versatile metabolic potential. However, communities were separated based on plant microenvironment only via GeoChip analysis, indicating that predictions can yield valuable global information but may not be an adequate reflection of the community variations detected only when analyzing samples directly. The GeoChip analysis also revealed differences between communities in terms of the relative abundance of some functional groups. Epiphytes, for example, had a greater abundance of genes involved in utilization of carbon sources such as starch, hemicellulose, and cellulose, among others, and in nitrogen utilization pathways, which were also abundant in the necromass fraction, such as ammonification, denitrification, and anammox. These differences could suggest adaptations to niches and nutrient availability, particularly when comparing the necromass and root samples with the leaf communities. However, differences could also be due to variations in biomass recovery, particularly from the less abundant endophyte population.
Carbon cycling pathways, which were among the most abundant detected using GeoChip analysis (see Fig. S7 in the supplemental material), revealed that plant-associated microbial communities can use diverse growth strategies. There was evidence of autotrophy (mainly represented by Calvin cycle genes) and C1 metabolism, but based on pathway abundances, microorganisms probably use predominantly heterotrophic growth on plant- or insect-derived carbon sources such hemicellulose, chitin, starch, pectin, and lignin. The pathways for carbon degradation were more abundant in the epiphyte, necromass, and root fractions and could reflect differences in nutrient availability. Many of the identified genes were associated with taxa abundant in the necromass tier, which is characterized by senescent plant material, that have been reported in other plants to be important for carbon utilization, such as Bacillus sp. (64), Klebsiella sp. (65), Propionibacterium sp. (66, 67), and Pseudomonas sp. (68, 69). Interestingly, the root community had increased abundance of “Candidatus Koribacter” and “Candidatus Solibacter,” two Acidobacteria members that have been isolated from Arctic soils and are reported to have metabolic versatility and many genes involved in breakdown of starch, hemicellulose, and pectin, among others (70). Genes involved in the degradation of aromatic compounds were also identified in our samples, consistent with the presence of compounds in Espeletia sp. plants, such as α-pinene, β-pinene, α-thujene, and longipilin acetate, among others (71–73), that can shape these populations, particularly in the necromass, where they may be used for the degradation of senescent leaves. Finally, the presence of genes for degradation of organic contaminants suggests that these plants may be more influenced by human intervention than previously thought and opens the possibility of exploiting these communities for bioremediation purposes or in industrial production systems.
The presence of genes for methanogenesis, especially in the epiphyte and root fractions, suggests that this C1 metabolic pathway could occur in the plant environment, consistent with a recent report indicating that methanogenesis can take place in plants in the presence of oxygen by not-yet-identified mechanisms (74). In our work, these genes were associated mainly with archaeal genes from members of the class Methanomicrobia, in which these pathways have been described (75, 76). It is possible that methanogenesis could occur in oxygen-limited biofilm structures on leaves, consistent with the high abundance of Pseudomonas spp., microorganisms long recognized as plant colonizers and biofilm formers (13, 53, 68). Finally, methane produced by methanogenic microorganisms, besides being used by methanotrophs as a carbon source, can be coupled to the nitrogen cycle, as has been reported recently (77, 78). Our GeoChip results also indicated that these plant microbial communities could perform almost every transformation of nitrogen, especially in the root fraction, where these pathways were more abundant. The abundance of carbon and nitrogen functional groups in plant samples suggests that microbial communities in the roots might be more active in providing nitrogen in a usable form whereas phyllosphere communities provide carbon sources as nutrients.
In addition to nutrient acquisition, plant microbial communities must also be able to survive under the environmental conditions of the particular habitat and as part of a complex community of bacteria, fungi, insects, and nematodes, using strategies that should be revealed in the functional profiles obtained. Several mechanisms for resistance to solar radiation were identified in our samples, such as production of pigments and genes involved in protection from ROS, consistent with the high exposure to UV radiation (3). The presence of genes involved in antibiotic production, particularly by Actinobacteria, can be indicative of microbial competition for nutrients and space. Although these metabolites could be used as signaling molecules, as suggested previously (79), the widespread presence of genes involved in antibiotic resistance across several taxa could also indicate ongoing competition. Finally, the production of plant hormones suggests an important role for these microorganisms in their association with the plant host. Microbial production of plant hormones has been reported to promote growth and development (80) and to prevent the entry of plant pathogens by modulating the plant's immune system (57, 81). The relation between bacteria and host growth and mortality has been previously described (49), suggesting that microbial communities may indeed be important for plant development. The present study therefore represents a starting point for uncovering possible interactions between microbial communities and Espeletia sp. development and health that become particularly relevant given the importance of these plants to these strategic yet threatened ecosystems and the recent reports of their increased mortality from causes that are still unclear but related to the presence of insects and/or fungi (82).
This study provides both taxonomic and functional information regarding Espeletia plant microbiomes. Taxonomic analysis indicated a continuum of microorganisms throughout the plant and communities with functional profiles shaped by specific niche characteristics. The presence of genes involved in growth and survival and in interactions with other species indicates that there might be both functional complementation and competition among microbial communities that, in addition, might also be important for host health. Although no new mechanisms for adaptation were identified, given the limitations of the metagenomics approaches used, this survey of microbial communities and their encoded functions provides a starting point for future studies aimed at understanding adaptations of the Espeletia phyllosphere microbiota, the roles played by these microorganisms, and their relevance in these ecosystems, which hopefully may lead to strategies for conservation of these plants. In addition, the data can also be used for comparing microbiomes between locally restricted plants, such as Espeletia, with more globally distributed species. Lastly, this work also opens the possibility of bioprospecting for microbial processes such as nutrient utilization, remediation, and antimicrobial compound production in the phyllosphere microbiome of plant species endemic to high Andean mountains.
ACKNOWLEDGMENT
We declare that we have no competing interest in relation to the work presented in this paper.
FOOTNOTES
- Received 28 August 2015.
- Accepted 30 December 2015.
- Accepted manuscript posted online 8 January 2016.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.02781-15.
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵