Evaluating the lettuce metatranscriptome with MinION sequencing for future spaceflight food production applications

Future long-duration space missions require reliable plant growth for food, life support, and crew well-being. Plant-associated microbiomes are integral to crop health, yet plant–microbe interactions can be altered by spaceflight and microgravity, sometimes increasing disease susceptibility. Prior observations include increased susceptibility of soybean seedlings to Phytophthora sojae in spaceflight, reduced nodulation in Medicago truncatula under clinorotation, and shifts in endophytic bacterial diversity in wheat under simulated microgravity. Space-based crop systems (e.g., Veggie, Advanced Plant Habitat) have produced safe leafy greens on the ISS, but episodes such as Fusarium oxysporum infection in Zinnia highlight the need to characterize and routinely monitor plant microbiomes during flight. Traditional protocols rely on returning frozen samples to Earth, which is impractical for exploration missions with long resupply intervals. The study’s objective is to develop and demonstrate an on-site, rapid, metatranscriptomic methodology—compatible with the portable Oxford Nanopore MinION—to characterize the active microbial community on lettuce grown under ISS-like conditions, enabling functional monitoring and early detection of potential pathogens to guide mitigation strategies and ensure crop health and crew safety.

The paper summarizes evidence that microgravity and spaceflight environments can perturb plant–microbe relationships at genetic and physiological levels, often increasing disease susceptibility (e.g., soybean susceptibility to Phytophthora sojae; reduced Medicago–rhizobium/mycorrhizal symbioses under clinorotation; altered wheat endophyte diversity under simulated microgravity with shifts in genera such as Pseudomonas, Paenibacillus, Bacillus). Space crop systems (Veggie, Advanced Plant Habitat, LADA) have successfully produced edible crops on ISS; however, opportunistic contamination events (e.g., Fusarium oxysporum on Zinnia in Veggie) underscore risk. Prior plant microbiome metatranscriptomic studies have generally used Illumina platforms, unsuitable for space due to mass/footprint, while nanopore sequencing has been validated aboard ISS for DNA but is emerging for metatranscriptomics. The authors position MinION-based metatranscriptomics as a feasible, rapid, hypothesis-free approach to monitor transcriptionally active crop microbiomes in-flight, complementing prior metagenomic surveys that may include DNA from non-viable organisms.

Study design and growth conditions: Red romaine lettuce (Lactuca sativa cv. Outredgeous) was grown from surface-sanitized seeds (bleach–HCl gas sterilization; sterility validated by plate assays) in 4-inch pots containing greens-grade arcillite with time-release fertilizer. Plants were grown at Kennedy Space Center in environmental chambers cleaned of human-associated microbes and set to ISS-like conditions: 50% RH, 3000 ppm CO2, 23 °C, 16 h light/8 h dark (200–300 µE m−2 s−1), automatic daily watering. Pots were thinned to one plant at day 7 and harvested at day 28. Three biological replicates (L93, L96, L101) were collected and stored in RNAlater at −80 °C. RNA extraction and microbial RNA enrichment: Total RNA from leaf tissue was extracted (Qiagen RNeasy Plant Mini Kit) with on-column DNase treatment. An RNAlater control was processed in parallel (no RNA recovered). To enrich microbial RNA from host–microbe total RNA (7.33–23.56 µg input), plant poly(A)+ RNAs were removed using Dynabeads poly(A) capture (some fungal mRNA may also be removed due to polyadenylation). The poly(A)-depleted material (microbial mRNA/rRNA plus plant rRNA) was purified (Zymo RNA Clean & Concentrator) and plant rRNA was depleted using RiboMinus Plant Kit. The host-depleted RNA (1–10 µg) was A-tailed with poly(A) polymerase to enable cDNA library construction. RNA quantity and quality were assessed by Qubit and Agilent Bioanalyzer; purity by A260/280. Library preparation and nanopore sequencing: For each replicate, 50 ng enriched RNA was reverse-transcribed (Maxima H Minus RT; 42 °C 90 min, 85 °C 5 min). Full-length cDNA was amplified with ONT SQK-PCS109 rapid attachment primers using LongAmp Taq (PCR: 95 °C 5 min; 14 cycles of 95 °C 15 s, 62 °C 15 s, 65 °C 1 min; 65 °C 6 min; 4 °C hold). Products were purified with AMPure XP beads (1.8x). Approximately 100 fmol cDNA with adapters was loaded on FLO-MIN106D (R9) flowcells; each replicate sequenced separately on a MinION for 48 h. Basecalling and read processing: Real-time basecalling used Guppy in MinKNOW v3.3.2 via MinIT v19.05.2. Reads were processed with Porechop (adapter trimming), filtlong (Q-score ≥5, length ≥100 bp; reads >30,000 bp removed), and NanoPlot for statistics. Host removal: Reads mapped with Bowtie2 to Asteraceae genomes (Lactuca sativa and Helianthus annuus); mapped reads were removed. rRNA removal: SortMeRNA v2.1 used to remove rRNA for non-rRNA functional analysis; microbial 16S and non-host 18S rRNA were retained for taxonomy. Taxonomic profiling: Kraken2 (kraken2-microbial database from NCBI RefSeq release 89) classified microbial 16S/18S rRNA reads. Read counts were normalized across samples; taxa with >500 normalized reads (across all three) were visualized. Cyanobacteria and unclassified reads were excluded to avoid chloroplast misassignments. Functional annotation: Non-rRNA reads were annotated with Trinotate v3.0 using BLAST+ v2.9.0 against UniProt (Swiss-Prot). Only reads with e-value ≤0.05 were kept. Transcripts assigned to viruses and other eukaryotes (except fungi) were excluded; Cyanobacteria-like and photosynthesis-associated transcripts were also excluded to minimize chloroplast confounds. KEGG Orthology (KO) assignments were obtained via KEGGREST; abundances normalized using gene-length corrected TPM followed by TMM (edgeR). Heatmaps were generated with Morpheus; hierarchical clustering used one minus Pearson’s correlation. Fungal signal transduction co-expression networks were inferred using Bray–Curtis dissimilarity (cutoff 0.5) in Phyloseq. Reproducibility was assessed via three biological replicates, each sequenced on separate flowcells.

• Microbial RNA enrichment and host depletion: Plant RNA depletion was effective, with a mean loss of ~93% of total RNA after enrichment (L93: 96.28%; L96: 88.89%; L101: 94.62%), leaving a residual rRNA peak in depleted samples indicating enrichment of microbial RNA.
• Sequencing output and quality (MinION cDNA): Mean base-called reads per replicate pot ~10.0 million. After trimming/QC (Q≥5, length ≥100 bp), retained reads were: L93 4,061,653 (58.34%), L96 10,661,802 (84.95%), L101 7,361,100 (70.49%). Mean read lengths: 378.9 bp (L93), 526.4 bp (L96), 423.7 bp (L101); N50: 417, 616, 512 bp; mean Q-scores: 8.0, 8.5, 7.0; percentage reads >Q5: 86.2%, 93.3%, 81.8%. Total bases: 2.64, 6.65, 4.42 Gb for L93, L96, L101, respectively.
• Host and rRNA filtering: Reads mapping to lettuce were removed (L93 19.17%; L96 22.72%; L101 7.23%) and sunflower (L93 3.53%; L96 5.52%; L101 1.71%). Remaining rRNA reads removed for functional analysis: L93 63.10%; L96 93.75%; L101 69.34%. Retained for taxonomy (microbial 16S and non-host 18S rRNA): 521,345 (L93), 2,453,876 (L96), 1,315,470 (L101).
• Functional annotation: UniProt-annotated non-rRNA reads (e-value ≤0.05): 57,315 (L93), 104,111 (L96), 88,850 (L101). KEGG Orthology assignments: Bacteria with KO annotation: 2,247 (L93), 1,075 (L96), 4,276 (L101); Fungi with KO annotation: 442 (L93), 208 (L96), 711 (L101).
• Taxonomic composition (rRNA-based): A total of 43 microbial phyla across Archaea, Bacteria, and Eukarya were detected. After excluding Cyanobacteria and unclassified reads, the five most abundant phyla were four bacterial (Firmicutes, Proteobacteria, Actinobacteria, Bacteroidetes) and one fungal (Ascomycota), collectively comprising 15.48% (L93), 13.13% (L96), and 15.04% (L101) of reads. Among 1,240 genera detected within these phyla, the top 10 (across samples) were: Clostridium (31.89%), Peptoniphilus (29.28%), Frondihabitans (5.74%), Lactococcus (3.64%), Bacillus (3.08%), Algoriphagus (2.96%), Fusarium (2.68%), Colletotrichum (2.43%), Actinomyces (1.91%), Nitrosococcus (1.88%). The top eight bacterial genera accounted for 78.34–81.67% of sequences; the two fungal genera (Fusarium, Colletotrichum) comprised 4.04–5.96%.
• Bacterial functional pathways: 106 KEGG level-3 pathways observed; the top 15 accounted for 94.28% of assigned reads. Dominant pathways: glyoxylate and dicarboxylate metabolism (mean 47.64% of bacterial assigned transcripts) and oxidative phosphorylation (mean 40.33%). Additional representation included glycolysis/gluconeogenesis, pyruvate metabolism, carbon fixation in prokaryotes, nucleotide and amino acid metabolism, ribosome, aminoacyl-tRNA biosynthesis, RNA polymerase/degradation, two-component system, and plant–pathogen interaction.
• Fungal functional pathways: 85 KEGG level-3 pathways observed; top 15 accounted for 84.11% of assigned reads. Dominant pathways: thermogenesis (mean 30.43%) and oxidative phosphorylation (mean 30.25%). Multiple signal transduction pathways were detected, including MAPK signaling (4.10%), phospholipase D (1.20%), two-component system (0.74%), FoxO (0.63%), and Rap1 (0.63%).
• Network analysis: Co-expression network of fungal signal transduction genes showed two clusters; cluster 2 was enriched in MAPK signaling genes with short edges indicating highly similar expression patterns across replicates, suggesting coordinated MAPK pathway activity.
• Operational feasibility: The protocol from tissue to analysis can be completed in ~2–5 days (sample prep to library ~8 h; sequencing up to 48 h; bioinformatics thereafter), contrasting with ~12 months needed to identify a prior ISS plant pathogen incident using traditional approaches.

The study addresses the need for on-orbit monitoring of plant-associated microbiomes by demonstrating a MinION-compatible metatranscriptomic workflow that enriches microbial RNA from plant tissues and characterizes both taxonomic composition and functional activity. Effective depletion of host RNAs led to recovery of microbial transcripts sufficient for taxonomic (rRNA-based) and functional (mRNA-based) analyses, with reproducible outputs across three independent lettuce plants. Results revealed transcriptionally active bacterial communities dominated functionally by glyoxylate/dicarboxylate metabolism and oxidative phosphorylation—consistent with phyllosphere metabolism of short carbon compounds—and fungal communities exhibiting strong oxidative phosphorylation and thermogenesis, along with coordinated MAPK signaling indicative of active cellular signaling potentially relevant to stress responses and fungal–plant interactions. Detection of transcriptionally active genera such as Peptoniphilus and Clostridium (bacteria) and Fusarium and Colletotrichum (fungi) highlights the method’s capacity to flag taxa with potential relevance to plant health, enabling prioritization for further investigation. The approach thus provides a rapid, hypothesis-free surveillance tool that can inform mitigation strategies during spaceflight, complementing metagenomic surveys by focusing on the viable, active microbiome.

This proof-of-concept establishes a feasible, reproducible MinION-based metatranscriptomic workflow for on-site monitoring of the active microbiome of a key space crop (red romaine lettuce) grown under ISS-like conditions. The method effectively enriches microbial RNA from plant tissues, enables rRNA-based taxonomic profiling and mRNA-based functional analysis, and reveals core bacterial and fungal activities relevant to phyllosphere ecology and plant health (e.g., glyoxylate metabolism, oxidative phosphorylation, MAPK signaling). The pipeline can deliver actionable insights within days, offering a substantial improvement over current return-to-Earth diagnostics. Future work should: optimize species-specific host rRNA/mRNA depletion (including organelle sequence removal), refine fungal RNA retention during poly(A) depletion, automate sample prep and library construction for microgravity (e.g., microfluidics, VolTRAX), expand temporal and spatial sampling in-flight, and integrate with imaging/sensor data to build predictive plant health monitoring systems.

• Residual host RNA: Despite ~93% depletion, substantial plant-derived reads remained, likely due to suboptimal performance of commercial plant rRNA depletion kits for lettuce and incomplete probe coverage (e.g., 28S rRNA). Chloroplast-derived transcripts can confound Cyanobacteria-like assignments.
• Fungal RNA loss: Poly(A) depletion used to remove plant mRNA can also remove fungal mRNA, potentially biasing fungal functional profiles.
• Taxonomic exclusions: To avoid chloroplast misassignments, Cyanobacteria-like reads and photosynthesis-related transcripts were excluded, which may remove true cyanobacterial signals.
• Limited sampling: Only one crop species, one cultivar, a single developmental time point (day 28), and three biological replicates were analyzed; findings may not generalize across time, cultivars, tissues, or flight conditions.
• Platform and annotation constraints: Nanopore read error profiles and short average read lengths (~379–526 bp) limit annotation sensitivity; many non-rRNA reads lacked confident UniProt hits (e-value ≤0.05). Detection is constrained by reference database completeness and curation.
• Operational readiness: Current protocol requires manual multi-step workflows; additional automation and hardware adaptation are needed for routine microgravity operations.