Seasonal activities of the phyllosphere microbiome of perennial crops

Abstract

Understanding the interactions between plants and microorganisms can inform microbiome management to enhance crop productivity and resilience to stress. Here, we apply a genome-centric approach to identify ecologically important leaf microbiome members on replicated plots of field-grown switchgrass and miscanthus, and to quantify their activities over two growing seasons for switchgrass. We use metagenome and metatranscriptome sequencing and curate 40 medium- and high-quality metagenome-assembled-genomes (MAGs). We find that classes represented by these MAGs (Actinomycetia, Alpha- and Gamma- Proteobacteria, and Bacteroidota) are active in the late season, and upregulate transcripts for short-chain dehydrogenase, molybdopterin oxidoreductase, and polyketide cyclase. Stress-associated pathways are expressed for most MAGs, suggesting engagement with the host environment. We also detect seasonally activated biosynthetic pathways for terpenes and various non-ribosomal peptide pathways that are poorly annotated. Our findings support that leaf-associated bacterial populations are seasonally dynamic and responsive to host cues.

Fig. 1. Phyllosphere microbiome field sampling strategy at the Great Lakes Bioenergy Research Center Bioenergy Cropping System Experiment (BCSE) in 2016 and 2017.

A The study site is at Kellogg Biological Station (KBS), a Long-Term Ecological Research site focused on agroecosystems located in southwest Michigan. B Four replicate randomized cropping system blocks from the Biofuel Cropping System Experiment (BCSE) were sampled at each time point for switchgrass (green) and/or miscanthus (blue). Within each plot, a fertilized main plot and unfertilized subplot were sampled. C Symbol fill shows which sequencing was performed on samples from which time points, for which crop. Filled squares show samples that were collected and open squares show samples that were not collected at a particular time point, with a particular crop. In 2016, both switchgrass and miscanthus were sampled, and in 2017 only switchgrass was sampled. Switchgrass leaves were flash-frozen in liquid nitrogen for RNA extraction and metatranscriptome analysis at a subset of time points in 2016 and at all points in 2017. Notably, the new metagenome and metatranscriptome datasets presented here overlap with the samples of the 16S rRNA amplicon time series presented in Grady et al.and the ITS amplicon time series presented in Bowsher et al..

Fig. 2. Overview of bioinformatic processing of the metagenome (solid green arrows) and metatranscriptome (solid red arrows) datasets.

Switchgrass reads are shown in light green, and miscanthus is dark green. The solid grey arrow represents the common analysis step for both datasets. Figure was made with Lucidchart. Note that metatranscriptome data are from switchgrass only and not miscanthus.

Fig. 3. Summary of leaf-associated MAGs.

A Abundance and occupancy of genomes assembled and binned from switchgrass and miscanthus phyllosphere metagenomes. Quality and contamination assessment were determined using checkM. 40 focal MAGs were selected. Occupancy is the proportion of the 192 metagenomes in which a MAG was detected, and abundance is the average normalized by housekeeping gene (HKG) MAG coverage across the 192 metagenomes. Symbol size shows the percent completeness of the MAG and color shows the percent contamination, with cooler colors indicating lower contamination. B Taxonomy of focal MAGs, as annotated with GTDB-tk, and taxonomic overlap with the persistent taxa detected in our previous 16S rRNA gene amplicon survey, which was conducted on the same samples and time series. Source data are provided as a Source Data file.

Fig. 4. Seasonal patterns of the 40 focal MAG metagenome and metatranscriptome read recruitment.

Abundance is indicated by color intensity, with blue indicating high and white indicating low abundance. MAG abundances for: A Miscanthus 2016 metagenomes (metaG); B Switchgrass 2016 metagenomes; C Switchgrass 2017 metagenomes; D Switchgrass 2016 metatranscriptomes (metaT) and E Switchgrass 2017 metatranscriptomes. Abundances of metagenome contigs were estimated with the median base pair of read recruitment divided by the average median base pair coverage of housekeeping genes. Abundances of metatranscriptome ORFs were estimated based on the median basepair coverage of all reads mapped to ORFs and divided by the median basepair coverage of housekeeping genes. The same dendrogram is applied to all panels, and it is the result of hierarchical clustering (see Fig. S2) of metatranscriptome diversity and abundances in switchgrass. Source data are provided as a Source Data file.

Fig. 5. 2016 (circle) and 2017 (triangle) switchgrass leaf transcript dynamics of KEGG metabolism classifications to the 40 focal MAGs.

The y-axis is scaled for each classification. Sample sizes are provided in Table 1 and included 22 and 56 metatranscriptomes for switchgrass in 2016 and 2017, respectively. Data are presented as mean values +/− standard error of the mean. Source data are provided as a Source Data file.

Fig. 6. Key functional gene pathways detected in the 40 focal MAGs and their activities (mapped transcripts) during the 2016–2017 switchgrass growing season.

Functional gene pathways were curated using antiSMASH for biosynthetic gene clusters, gapseq for general metabolic pathways, and manual selection for plant-associative functions reported in the literature. Pathways that were discovered in the MAGs but not detected in the transcripts are represented by open circles, and pathways detected in the MAGs and mapped by transcripts are represented by filled circles. Colors categorize different functional groups of pathways. Source data are provided as a Source Data file.

Fig. 7. Switchgrass microbiome transcripts detected over two growing seasons.

A 2016 (green, circle) and 2017 (blue, triangle) switchgrass leaf transcript dynamics of KEGG metabolism classifications associated with terpene metabolism. B Transcripts in MAGs associated with terminal enzymes in the non-mevalonate isoprene biosynthesis, gcpE and lytB. MAG IDs include predicted taxonomy at the genus level: Methyl = Methylobacterium, Frigor = Frigoribacterium, Pseudokineo = Pseudokineococcus, Microbac = Microbacterium, Amnibac = Amnibacterium, Hymeno = Hymenobacter, Sphingo = Sphingomonas, and Pseudo = Pseudomonas. Sample sizes are provided in Table 1 and included 22 and 56 metatranscriptomes for switchgrass in 2016 (green circles) and 2017 (blue triangles), respectively. Data are presented as mean values ± standard error of the mean. Source data are provided as a Source Data file.

Fig. 8. Detection of 36 focal MAGs (out of 40) in publicly available metagenomes of bioenergy grasses based on the average number of reads mapped to each MAG.

The sample size was 55 public metagenomes. Data are presented as mean values ± standard error of the mean. Source data are provided as a Source Data file.