Spatial transcriptomics of planktonic and sessile bacterial populations at single-cell resolution

Abstract

Capturing the heterogeneous phenotypes of microbial populations at relevant spatiotemporal scales is highly challenging. Here, we present par-seqFISH (parallel sequential fluorescence in situ hybridization), a transcriptome-imaging approach that records gene expression and spatial context within microscale assemblies at a single-cell and molecule resolution. We applied this approach to the opportunistic pathogen Pseudomonas aeruginosa, analyzing about 600,000 individuals across dozens of conditions in planktonic and biofilm cultures. We identified numerous metabolic- and virulence-related transcriptional states that emerged dynamically during planktonic growth, as well as highly spatially resolved metabolic heterogeneity in sessile populations. Our data reveal that distinct physiological states can coexist within the same biofilm just several micrometers away, underscoring the importance of the microenvironment. Our results illustrate the complex dynamics of microbial populations and present a new way of studying them at high resolution.

Fig. 1.. Parallel and sequential mRNA-FISH in bacteria.

(A) seqFISH probe design scheme. Primary probes contain unique sequences (Si) that are read by secondary probes (colored wands). Each gene is read by a unique probe and its fluorescence can be turned “ON” or “OFF”. (B) mRNA-FISH applied sequentially to the same sample. In each cycle, a new set of secondary readout probes are introduced. Raw fluorescence data is shown on the right and the detected local spot maxima are shown in the spot detection image. Merged spots for many genes shown in shuffled colors. (C) Combinatorial labeling can be used to encode species taxonomy using 16S rRNA, or to enable the parallel study of (D) bacteria grown in different conditions.

Fig. 2.. Parallel seqFISH (par-seqFISH) of an LB growth curve experiment

(A) The sampled LB growth curve. Collected time points are indicated with gray circles. A zoom-in shows the sampled lag phase. The presented cfu/ml were estimated using OD₆₀₀ values (OD₆₀₀ = 1.0 reporting on ~10⁹ cfu/ml (104)). The OD₆₀₀ values are indicated over each time-point. (B) Demultiplexed bacteria and their mRNAs. The merged, raw Ribo-Tag 16S rRNA fluorescence is shown for a representative region. Different barcodes (16S combinations) result in unique colors that visually report the condition of which they originated from (indicated with the corresponding OD₆₀₀ value). Ellipses fitted to the segmented cell boundaries are shown. The mRNA spots (fitted position of maximal intensity) for all genes per cell are shown in unique colors per gene. Each spot may represent more than one mRNA copy. (C-F) Condition specific distributions of nucleoid length, chromosome copy, ribosome levels and total mRNAs detected across our gene set. Distributions contain all demultiplexed cells per condition and are significantly different from their previous time-point unless otherwise noted (Wilcoxon, p < 0.001). (G) Heatmap showing average gene expression normalized to the maximal value for each gene across all conditions. Highlighted gene groups and their functions are indicated on the right.

Fig. 3.. Single-bacterium analysis reveals physiologically distinct dynamic sub-populations.

(A) UMAP analysis using cells from all 11 time points. Identified clusters are shown in different colors and are indexed by group size. Specific group and their enriched functions are shown to the right. (B) Gene expression overlays for four genes that report on metabolic state, stationary phase progression and exoproduct biosynthesis. The color map shows the normalized expression scaled to unit variance. Cells from all 11 time points are displayed in the plot (C-F) Density scatter plots of cells from individual conditions in a zoom-in of the UMAP (dashed box in panel A). The clusters are indicated by their index. (G-J) Gene expression overlays shown as in B.

Fig. 4.. Spatial transcriptomics in P. aeruginosa biofilms at a single cell resolution.

(A) A representative field of view collected during a 10h surface colonization experiment showing cells via 16S rRNA fluorescence (gray). A zoom-in (orange box) shows the cell segmentation masks depicted as white ellipses. The 16S rRNA signal and mRNA-FISH data for several genes are shown in different colors. (B) A 35h experiment field is shown in an identical manner to panel A. Scale bar length is annotated within the figure. (C) Joint UMAP cluster analysis of biofilm and planktonic experiments. Planktonic cells are shown for all time points collected (D) UMAP scatter plots showing cells from either planktonic or biofilm experiments as indicated. Below, a highlighted set of UMAP clusters associated with each experiment is annotated with enriched functions. (E-HGHH) UMAP overlay with specific gene data. The color map shows the normalized expression scaled to unit variance. Cells from the liquid experiment and both 10h and 35h biofilms are displayed together.

Fig. 5.. Spatial expression patterns for motility and pyocin related genes.

(A-B) Representative regions from the 10h and 35h biofilm experiments, cells are shown in via 16S rRNA fluorescence (gray) and overlayed with raw mRNA-FISH fluorescence for different genes as indicated. (C) planktonic cells from the pair liquid experiments. Cells are shown via DAPI and expression as indicated (D-E) 10h aggregate showing R2-pyocin expression. (F) Enrichment of R2-Pyocin mRNA near strong induction sites (cell with 99.5^th percentile pyocin expression). X-axis shows the number of cells closest to an induction site that were analyzed (neighborhood size; center cell was excluded). Y-axis shows the enrichment in each neighborhood relative to the total population. A non-pyocin control gene is shown (rpoA). (G) Examples of mRNA R-pyocin transcript and ribosome polar localization as indicated in the legends.

Fig. 6.. Oxygen availability shapes microscale metabolic heterogeneity in biofilms

(A-E) Representative 10h biofilms. Cells are shown via 16S rRNA FISH fluorescence (gray) and overlayed with raw mRNA-FISH fluorescence for different genes as indicated in each panel. White circles highlight regions of interest. (F) Cells painted according to their UMAP derived metabolic state as indicated in the panel legends (also see fig. S7 clusters, 0, 8, 12 and 15), showing co-localization of multiple metabolic states within a given region.

Fig. 7.. Functional zonation in a single microaggregate.

(A-D) A P. aeruginosa 35h aggregate. Bacteria are shown via 16S rRNA FISH fluorescence (gray) and are overlaid with raw mRNA-FISH fluorescence for different genes as described in the panel legends.