Microbial occurrence and symbiont detection in a global sample of lichen metagenomes

Abstract

In lichen research, metagenomes are increasingly being used for evaluating symbiont composition and metabolic potential, but the overall content and limitations of these metagenomes have not been assessed. We reassembled over 400 publicly available metagenomes, generated metagenome-assembled genomes (MAGs), constructed phylogenomic trees, and mapped MAG occurrence and frequency across the data set. Ninety-seven percent of the 1,000 recovered MAGs were bacterial or the fungal symbiont that provides most cellular mass. Our mapping of recovered MAGs provides the most detailed survey to date of bacteria in lichens and shows that 4 family-level lineages from 2 phyla accounted for as many bacterial occurrences in lichens as all other 71 families from 16 phyla combined. Annotation of highly complete bacterial, fungal, and algal MAGs reveals functional profiles that suggest interdigitated vitamin prototrophies and auxotrophies, with most lichen fungi auxotrophic for biotin, most bacteria auxotrophic for thiamine and the few annotated algae with partial or complete pathways for both, suggesting a novel dimension of microbial cross-feeding in lichen symbioses. Contrary to longstanding hypotheses, we found no annotations consistent with nitrogen fixation in bacteria other than known cyanobacterial symbionts. Core lichen symbionts such as algae were recovered as MAGs in only a fraction of the lichen symbioses in which they are known to occur. However, the presence of these and other microbes could be detected at high frequency using small subunit rRNA analysis, including in many lichens in which they are not otherwise recognized to occur. The rate of MAG recovery correlates with sequencing depth, but is almost certainly influenced by biological attributes of organisms that affect the likelihood of DNA extraction, sequencing and successful assembly, including cellular abundance, ploidy and strain co-occurrence. Our results suggest that, though metagenomes are a powerful tool for surveying microbial occurrence, they are of limited use in assessing absence, and their interpretation should be guided by an awareness of the interacting effects of microbial community complexity and sequencing depth.

Fig 1. Bioinformatic pipeline used in the study.

(A) Flowchart of the bioinformatic analysis of this study. (B) Sankey plot showing the source of 437 metagenomes used in the analysis and the progressive reduction of the dataset in the course of the study. LFS, lichen fungal symbiont; MAG, metagenome-assembled genome.

Fig 2. Maximum likelihood phylogenetic trees of the MAGs.

The trees are calculated using IQ-TREE and are based on alignments of marker genes. Clade labels and color bars indicate broad groups the MAGs were assigned to. Reference genomes (interleaved) are not colored or labeled. (A) Tree of recovered fungal MAGs (colored sectors) interleaved with reference genomes (listed in S5 Table), based on 709 BUSCO genes. (B) Tree showing recovered algal MAGs interleaved with reference genomes based on 1296 BUSCO genes. Bars represent the number of occurrences of the MAG in Dataset 1. (C) Tree of the bacterial MAGs obtained from lichen metagenomes interleaved with reference genomes based on 120 marker genes from GTDB-Tk. The 4 most frequent bacterial families are denoted by colored sectors. Radiating bars represent the number of occurrences for each given MAG in Dataset 1. Reference genomes belong to bacteria previously isolated from lichens and are shown as numbers: 1: Lichenicola cladoniae PAMC 26569; 2: Lichenicoccus roseus KEBCLARHB70R; 3: Lichenihabitans minor RmlP026; 4: Lichenihabitans ramalinae RmlP001; 5: Lichenihabitans psoromatis PAMC 29128 and PAMC 29148; 6: Lichenifustis flavocetrariae BP6-180914; 7: Methylobacterium planeticum YIM 132548; 8: Aureimonas leprariae YIM 132180; 9: Rubellimicrobium rubrum YIM 131921; 10: Paracoccus lichenicola YIM 132242; 11: Polymorphobacter megasporae PAMC 29362; 12: Subtercola lobariae CGMCC 1.12976; 13: Luteimicrobium album NBRC 106348; 14: Streptomyces lichenis LCR6-01; 15: Nakamurella leprariae YIM 132084. Full versions of the phylogenomic trees are available in FigShare (10.6084/m9.figshare.27054937). GTDB, Genome Taxonomy Database; MAG, metagenome-assembled genome.

Fig 3. Organismal occurrence by major microbial group.

Heatmaps showing occurrence rates of mapped MAGs, and detection of canonical symbionts and major additional organismal groups using rRNA in assemblies and raw read sets. (A) For class-level taxa against major LFS (lichen fungal symbiont) groups and major photobiont groups; (B) for 9 of the 13 key genus-level bacterial groups against major LFS groups and major photobiont groups. We grouped metagenomes by the photobiont that is expected in a given lichen based on literature sources (see Supporting information for Spribille and colleagues [9]). Only metagenomes from the Dataset 2 (n = 330) are included in this analysis, and only groups represented by 4 or more metagenomes are shown. The number of metagenomes in each category is shown in the “n” column. The data underlying this figure can be found in S1 Data. LFS, lichen fungal symbiont; MAG, metagenome-assembled genome.

Fig 4. Organism detection as a function of sequencing depth.

(A) Number of recovered MAGs as a function of sequencing depth (bp). Each dot represents a metagenome colored based on the recovery of the LFS (lichen fungal symbiont) and the photobiont. The curve indicates a generalized additive model (GAM) smoothing. The data underlying this figure can be found in S9 Table. (B) Detection of the key microbial groups in lichen metagenomes based on 2 methods of screening: presence of MAGs and presence of the SSU rRNA in the raw, unassembled reads as a function of sequencing depth. Each vertical bar represents a single metagenome, with the bar heights representing sequencing depth and the color representing the screening outcome. The metagenomes are sorted along the x-axis based on the outcome of the screening and the sequencing depth. The data underlying this figure can be found in S1 Data. GAM, generalized additive model; LFS, lichen fungal symbiont; MAG, metagenome-assembled genome; SSU, small subunit.

Fig 5. Functional annotation of key bacterial MAGs.

(A) Selecting MAGs for annotation. The bar graph on the left shows the total number of occurrences per bacterial “genus” (after GTDB), with genera listed in decreasing order. Red bars represent the genera selected for functional annotation, which were the 13 most frequent genera together accounting for 53% of all bacterial occurrences. The waffle graph on the right shows the most frequent bacterial genera and MAGs assigned to them. Red circles represent MAGs selected for annotation: all MAGs from the selected genera that had >95% completeness. Filled gray circles represent MAGs with >95% completeness that belonged to less frequent genera and were not annotated. The open gray circles represent MAGs with ≤95% completeness, also not annotated. (B) Presence of selected pathways and protein complexes in the MAGs of most common lichen bacteria. Each column represents one of the 63 annotated MAGs, grouped by their taxonomy. We reconstructed pathways using KEGG; to detect carotenoid BGCs, we used antiSMASH. Here, we show the presence of pathways and protein complexes potentially relevant to the symbiosis. For 3 pathways (biosynthesis of bacteriochlorophyll, biotin, and cobalamin), we also show partial completeness (allowing one missing gene). (C) Number of genes assigned to each CAZy class per MAG. We annotated CAZymes in the MAGs selected for an in-depth annotation using dbcan. The data here are grouped on the family level. The CAZy classes are: Auxiliary Activities (AA), Carbohydrate-Binding Modules (CB), Carbohydrate Esterases (CE), Glycoside Hydrolases (GH), Glycosyl Transferases (GT), and Polysaccharide Lyases (PL). The data underlying this figure can be found in S1 Data. GTDB, Genome Taxonomy Database; MAG, metagenome-assembled genome.

Fig 6. Completeness of modules involved in cofactor biosynthesis in the high-quality eukaryotic MAGs.

We selected all ≥95% complete and ≤10% contaminated eukaryotic MAGs, and used KEGG to determine completeness of pathways involved in synthesis of biotin, thiamine, and cobalamin. The trees on the left represent the phylogenomic trees constructed for the MAGs; only tips corresponding to the high-quality MAGs are shown. Color on the heatmap represents the percentage of present blocks in a given pathway. The data underlying this figure can be found in S1 Data. MAG, metagenome-assembled genome.

Fig 7. Co-occurrence of cofactor synthesis modules within lichen metagenomes.

Each panel represents a highly complete lichen metagenome (defined as a metagenome that contains the MAGs of the LFS, photobiont, and at least 1 high-frequency bacterium, all at ≥95% complete and ≤10% contaminated). Each column within a panel represents a MAG, labeled based on the inferred role it plays in the symbiosis and colored based on its taxonomic assignment. For each genome, we show the presence/absence and completeness of several KEGG modules for synthesis of biotin and thiamine. The data underlying this figure can be found in S1 Data. LFS, lichen fungal symbiont; MAG, metagenome-assembled genome.