Visualization of mRNA Expression in Pseudomonas aeruginosa Aggregates Reveals Spatial Patterns of Fermentative and Denitrifying Metabolism

Abstract

Gaining insight into the behavior of bacteria at the single-cell level is important given that heterogeneous microenvironments strongly influence microbial physiology. The hybridization chain reaction (HCR) is a technique that provides in situ molecular signal amplification, enabling simultaneous mapping of multiple target RNAs at small spatial scales. To refine this method for biofilm applications, we designed and validated new probes to visualize the expression of key catabolic genes in Pseudomonas aeruginosa aggregates. In addition to using existing probes for the dissimilatory nitrate reductase (narG), we developed probes for a terminal oxidase (ccoN1), nitrite reductase (nirS), nitrous oxide reductase (nosZ), and acetate kinase (ackA). These probes can be used to determine gene expression levels across heterogeneous populations such as biofilms. Using these probes, we quantified gene expression across oxygen gradients in aggregate populations grown using the agar block biofilm assay (ABBA). We observed distinct patterns of catabolic gene expression, with upregulation occurring in particular ABBA regions both within individual aggregates and over the aggregate population. Aerobic respiration (ccoN1) showed peak expression under oxic conditions, whereas fermentation (ackA) showed peak expression in the anoxic cores of high metabolic activity aggregates near the air-agar interface. Denitrification genes narG, nirS, and nosZ showed peak expression in hypoxic and anoxic regions, although nirS expression remained at peak levels deeper into anoxic environments than other denitrification genes. These results reveal that the microenvironment correlates with catabolic gene expression in aggregates, and they demonstrate the utility of HCR in unveiling cellular activities at the microscale level in heterogeneous populations. IMPORTANCE To understand bacteria in diverse contexts, we must understand the variations in behaviors and metabolisms they express spatiotemporally. Populations of bacteria are known to be heterogeneous, but the ways this variation manifests can be challenging to characterize due to technical limitations. By focusing on energy conservation, we demonstrate that HCR v3.0 can visualize nuances in gene expression, allowing us to understand how metabolism in Pseudomonas aeruginosa biofilms responds to microenvironmental variation at high spatial resolution. We validated probes for four catabolic genes, including a constitutively expressed oxidase, acetate kinase, nitrite reductase, and nitrous oxide reductase. We showed that the genes for different modes of metabolism are expressed in overlapping but distinct subpopulations according to oxygen concentrations in a predictable fashion. The spatial transcriptomic technique described here has the potential to be used to map microbial activities across diverse environments.

Keywords: ABBA; HCR; Pseudomonas aeruginosa; aggregates; biofilms; denitrification; gene expression; heterogeneity; hybridization chain reaction.

FIG 1

Regulatory diagram of target genes. This diagram was compiled from a literature search of transcriptomics reviews in P. aeruginosa (34, 37, 42).

FIG 2

mRNA probes are target specific. (A) Micrographs of single-cell deletion validation controls for mRNA probes. (B) Micrographs of aggregate deletion validation controls for mRNA probes. Images reflect the region 50 to 100 μm from the air-agar interface. (C) Quantification of mean mRNA probe intensities for single cells grown in liquid culture of wild-type and deletion mutants for target genes under upregulating growth conditions. Mean intensity was 10-fold higher in the wild type than in the deletion for all probe sets. Each boxplot summarizes approximately 10 images per replicate. Three replicates were performed per condition, and each colored box represents a different biological replicate. Whiskers represent 1.5 times the interquartile range, while diamonds on the boxplots represent outliers. (D) Quantification of mean mRNA probe intensities for aggregates grown in agar blocks wild-type and deletion mutants for ackA under upregulating growth conditions. Mean intensity was 2-fold higher in the wild type, and each boxplot summarizes 3 to 5 images per replicate. Three replicates were performed per condition.

FIG 3

Metabolic genes show distinct patterns across three-dimensional oxygen gradients. (A) Mean oxygen levels of ABBA samples grown with LB plus 40 mM nitrate for 12 h. The dark line is the mean, while the shading represents the standard deviation of 8 biological replicates grown under identical conditions. (B) Three-dimensional micrographs of probe signal in LB plus 40-mM nitrate ABBAs. Each image represents a 50-μm slice of agar, compiled from 8 individual z-slices with an interslice distance of 6.24 μm, viewed from the top of the block, with each sequential image from the top of the figure representing the section directly below the slice above it. The rRNA signal is colored cyan, while the mRNA signal is colored magenta. (C) Mean mRNA channel intensity per aggregate (x axis) plotted by depth. Each plot represents four images of one replicate each of an experimental and control condition, and each point represents one aggregate. Filled points represent the experimental condition, while the open circles represent the background autofluorescence intensity of aggregates imaged in the mRNA channel in a control condition where only rRNA probes were used.

FIG 4

Metabolic genes show distinct patterns compared to rRNA expression. Mean mRNA intensity per aggregate (x) plotted against mean rRNA intensity per aggregate (y) colored by depth. Each plot represents four z-stacks of one replicate, and each point represents one aggregate. Large points represent the experimental condition, and small points represent the control condition without mRNA probes. The Pearson correlation coefficient (PCC) of ccoN1 with rRNA was 0.956; for ackA with rRNA, it was 0.896; for narG with rRNA, it was 0.612; for nirS with rRNA, it was 0.080; and for nosZ with rRNA, it was 0.678.