Figures and data
Development of RiboD-PETRI and validation of its technical performance in studying population heterogeneity.
(A) Graphic summary of the RiboD-PETRI method illustrating the incorporation of RiboD after cell pooling and lysis in PETRI-seq. The RiboD protocol is represented by the dashed-line box. (B) Comparison of non-rRNA (tRNA, mRNA and other non-rRNA) and rRNA UMI counts ratio among different bacterial scRNA-seq methods. Data from PETRI-seq (E col/)8, MicroSPLIT-seq (E coli)9, M3-seq (E coli)11 cited from from previous studies. Error bars represent standard deviations of biological replicates. (C) Comparison of UMI counts per cell between RiboD-PETRI and PETRI at the same unsaturated sequencing depth. (D) Assessment of the effect of rRNA depletion on transcriptional profiles. The Pearson correlation coefficient (r) of UMI counts per gene (log2 UMIs) between RiboD-PETRI and PETRI was calculated for 3790 out of 4141 total genes, excluding those with zero counts in either library. Each point represents a gene. (E) Evaluation of the correlation between RiboD-PETRI data and bulk RNA-seq results. The Pearson correlation coefficient (r) of UMI counts per gene (log2 UMIs) among RiboD-PETRI data and the reads per gene (log2 reads) of bulk RNA-seq data was calculated for 3814 out of 4141 total genes, excluding those with zero counts in either library. Each point represents a gene. (F-I) RiboD-PETRI data from E coli in exponential phase (E coli 3h). (F) Distributions of mRNA UMIs captured per cell in RiboD-PETRI data of E coli in exponential phase. (G) Uniform Manifold Approximation and Projection (UMAP) visualization of E coli bacteria during the exponential phase. Data were filtered for cells with UMIs between 200 and 5,000, resulting in 1,464 cells. Each dot represents a cell. (H) Heatmap illustrating the normalized gene expression levels of marker genes in different clusters of E. coli during the exponential phase. Marker genes with relatively high expression levels are depicted in yellow, while lower expression levels are shown in purple. Each row represents a gene, and each column represents a cell. (I) Functional enrichment analysis of marker genes in cluster 2. Marker genes were selected based on screening criteria of p-value < 0.001 and log2 fold change (FC) > 0.2.
RiboD-PETRI Resolves Biofilm Heterogeneity and Identifies New Marker Genes.
(A-D) RiboD-PETRI data from 24-hour static biofilms of E coli (E coli 24h). (A) UMI counts per cell in RiboD-PETRI data of 24-hour static biofilms of E coli. Two replicates were screened for cells with UMIs between 100 and 2000, resulting in 1621 and 3999 cells, with median UMIs per cell of 283.5 and 239, respectively. (B) UMAP visualization of 24-hour static biofilms of E. coli, revealing two small populations of heterogeneous cells in clusters 2 and 3. (C) Inferred expression levels of marker genes from 24-hour static biofilms of E. coli across different clusters. (D) Enrichment pathways for marker genes in cluster 2, selected based on screening criteria of p-value < 0.001 and log2 fold change (FC) > 0.2. (E & F) Dot plot displaying scaled expression levels of marker genes in different clusters of E. coli in exponential phase (E) and E. coli in 24- hour static biofilms (F). These genes were markers of static E. coli biofilms in cluster 2, identified with screening criteria of p-value < 0.001 and logFC > 3. Dot size represents the percentage expression of the gene in the cluster, while color indicates the average expression level normalized from 0 to 1 across all clusters for each gene. (G & H) UMAP plots showing the distribution of pdel in single-cell data of E. coli in exponential phase (G) and E. coli in 24-hour static biofilms (H). Each dot represents a cell colored by normalized expression levels of genes. (I) Subcellular localization of Pdel-GFP. (J) c-di-GMP levels (R-1 score) in E. colicells with different PdeI-GFP expression levels (low or high) in 24-hour static biofilms. c-di-GMP levels are measured using the c-di-GMP sensor system15 integrated into E. coli cells. Scale bar, 2 μm. R-1 score was determined using the fluorescent intensity of mVenusNB and mScarlet-I in the system. The fluorescent intensity is measured by sorted by fluorescence-activated cell sorting (FACS). (K) Determination of cellular concentrations of c-di-GMP by HPLC-MS/MS in cells overexpressing Pdel under the control of arabinose promoter, with 0.002% arabinose induction for 2 h (n=3). (L) Localization of Pdel-high cells in the biofilm matrix. Cells expressing Pdel-BFP under the control of the pdel native promoter were grown in a glass-bottom cell culture dish and stained with SYTO™ 24 for bacterial DNA. (M & N) Heterogeneous expression of Pdel in single-cell data of E. coli in exponential phase (M) and E. coli in 24- hour static biofilms (N). Biofilm cells with high or low expression levels of Pdel-BFP were sorted by FACS. (O) Persister counting assay using 150 pg/ml ampicillin on cells with high or low expression levels of Pdel-BFP from 24h static biofilm E. coli, sorted by FACS. (P) Time-lapse images of the persister assay observed under a microscope. Static biofilm cells of the Pdel-GFP strain were spotted on a gel pad and treated with 150 μg/ml ampicillin in LB broth. lmages were captured over 6 hours at 37°C, followed by the replacement of fresh LB broth to allow persister cell resuscitation. Scale bar, 2 μm. Error bars represent standard deviations of biological replicates. Significance was ascertained by two-tailed Student’s t test. Statistical significance is denoted as *P <0.05, **P < 0.01, ***, P<0.001, ****, P<0.0001.
Technical Application of RiboD-PETRI in S. aureus and C. crescentus. (A-E) Single-cell RiboD-PETRI results for Staphylococcus aureus strain ATCC 25923 (S. aureus), cultured for 9 hours in MHB medium at 37°C. (F-J) Single-cell RiboD-PETRI results for Caulobacter crescentusstrain NA1000 (C. crescentus), incubated at 37°C for 3 hours. (A & F) Distribution of mRNA UMIs captured per cell in RiboD-PETRI data of(A) S. aureus and (F) C. crescentus, presented as violin plots showing the upper quartile, median, and lower quartile lines. “SA” denotes S. aureus, and “CC” denotes C. crescentus. For S. aureus, 1,000, 5,000, and 8,000 cells were selected in descending order of UMI counts per cell, with median UMIs per cell of 273,122, and 94, respectively. For C. crescentus, 1,000, 5,000, and 10,000 cells were selected with median UMIs per cell of 2190, 662, and 225, respectively. (B & G) Screening of single-cell data for (B) S. aureus and (G) C. crescentus, resulting in 9386 and 5728 cells, respectively. Screening criteria included UMIs between 15 and 1000 for S. aureus and UMIs between 200 and 5000 for C. crescentus, along with gene counts greater than 30. (C & H) Distribution of UMIs on the UMAP results for (C) S. aureus and (H) C. crescentus. UMAP results reveal heterogeneity among populations, with each point representing a cell and color shading indicating UMI counts. (D & I) UMAP visualization of (D) S. aureus and (I) C. crescentus, demonstrating the ability of RiboD-PETRI to distinguish population heterogeneity. (E & J) Expression of major marker genes for (E) S. aureus and (J) C. crescentus overlaid on the UMAP plot, highlighting cells with high expression levels in blue.
Technical performance of RiboD-PETRI. (A) Scatterplot illustrating the relationship between reads per cell and counts of UMIs per cell detected from exponential phase E. coli data. Each dot represents a cell. (B) Scatterplot demonstrating the relationship between reads per cell and counts of UMIs per cell detected from static biofilm E. coli data. Two replicates of the sample are included. (C) Calculation of the Pearson correlation coefficient (r) of UMI counts per gene between replicate 1 and replicate 2 of static biofilm E. coli. The analysis involved 4,062 out of 4,141 total genes, with a significant correlation (p-value < 0.0001, r = 0.96), indicating good replication between samples. Each dot represents a gene. (D & F) Normalized and Principal Component Analysis (PCA) performed on screened data of (D) exponential phase E. coli, resulting in 1,464 cells, and (F) two replicates of static biofilm E. coli, resulting in 1,621 and 3,999 cells, respectively. The resulting scatterplots show heterogeneity among the populations, with each point representing a cell. (E & G) UMAP plots generated based on the distribution of UMI counts of cells, illustrating the distribution of UMI counts for different cells in (E) and (G). The distribution of clusters in the UMAP results is independent of UMI counts. (H) UMAP plot based on the original identity of static biofilm E. co//samples (replicate 1 and replicate 2). Each dot represents a cell, with red indicating replicate 1 and green indicating replicate 2.
Marker Genes Identified in E.coli 3h Single-Cell Data by RiboD-PETRI. Expression levels of various marker genes during the 3-hour exponential period of E. coli overlaid on the UMAP plot. Cells with high expression levels are depicted in blue. Marker genes were selected based on a p-value greater than 0.001 and a log2FC greater than 3.
