Fig 1.
Protocol for the serial passage of M. tb pellicle biofilms and subsequent transcriptomic comparisons.
Ancestral (clinical) populations were grown in planktonic cultures, then grown as pellicle biofilms (passage 1) before passaging 8–12 times. This passaged (evolved) population was taken from a biofilm culture and grown planktonically again. We performed RNA sequencing on all ancestral and evolved populations, grown under both biofilm and planktonic conditions.
Fig 2.
Differentiation of transcriptomes between closely related populations.
Left: Whole-genome sequence phylogeny of ancestral populations used for passaging. Populations fall into two major L4 sub-lineages: L4.9 (purple) and L4.4 (orange) with SNP distances shown along each branch. Right: Number of significantly differentially expressed genes (DEGs) between biofilm and planktonic conditions for each ancestral population. Note that the analysis of ancestral MT49 is not included because only one biological replicate was available under planktonic conditions.
Fig 3.
Genetic background affects transcriptional mechanisms of biofilm growth.
A) Experimental diagram highlighting comparator populations: ancestral populations grown as pellicle biofilms are compared to the same populations grown in planktonic culture. Ancestral populations from freezer stocks of low-passage clinical isolates were used to seed planktonic cultures, and subsequently biofilm cultures. B) Principal component analysis (PCA) of variance stabilizing transformed gene expression for ancestral populations. Each point is a biological replicate. Results demonstrate diversity among ancestral transcriptomes, with intermingling of planktonic and biofilm samples. C) Volcano plot summarizing differential expression between ancestral populations grown as biofilms and as planktonic cultures. Log transformed adjusted p-values plotted against the log2 fold change for each gene. Genes that did not have significant differential expression are shown in grey. Very few genes are significantly differentially expressed when all populations are analyzed together. A zoomed in version of this figure is available in S4D Fig) Matrix of individual DEGs shared across ancestral populations. A total of 3874 DEGs are plotted according to whether the gene is upregulated (red), downregulated (blue) or not significantly differentially expressed (white) in each population. *Only 4% of downregulated genes are shared by at least 4 of the analyzed populations while no upregulated genes are shared across more than 3 populations. Note that analyses of ancestral MT49 are not included as only one biological replicate was available under planktonic conditions.
Fig 4.
Transcriptional responses to biofilm growth following biofilm passaging.
A) Experimental diagram highlighting comparator populations: evolved populations grown as pellicle biofilms are compared to the same populations grown in planktonic cultures. Analyses are parallel to those of ancestral populations in Fig 3. B) Principal component analysis (PCA) of variance stabilizing transformed gene expression for evolved populations grown as biofilms and planktonic cultures. Biofilm passaging resulted in clear separation of biofilm and planktonic transcriptomes. These transcriptomes were not clearly separated at baseline (Fig 3B) C) Volcano plot summarizing differential expression between evolved populations grown as biofilms and as planktonic cultures. Log transformed adjusted p-values plotted against the log2 fold change for each gene. Genes that did not have significant differential expression are shown in grey. Biofilm transcriptomes were characterized by broad-scale downregulation of gene expression. D) Matrix of individual DEGs shared across evolved populations. A total of 3568 DEGs are plotted according to whether that gene is upregulated (red), downregulated (blue) or not significantly differentially expressed (white) in each population. 17% of downregulated genes are shared by at least 5 populations (*), while 6% of upregulated genes are shared by at least 5 populations (**); see S2 Data for details. This contrasts with ancestral populations, which did not share any upregulated genes and only shared 4% of downregulated genes across 5 or more populations (Fig 3D). Note that low concordance between biological replicates of evolved MT31 grown under planktonic conditions may affect these results. See text for further discussion.
Fig 5.
Genetic background shapes adaptive trajectories.
A) Principal component analysis (PCA) of variance stabilizing transformed expression counts from ancestral and evolved biofilm populations. Arrows are drawn between corresponding ancestral and evolved populations, indicating the trajectory of evolution across passaging. Points are colored by sub-lineage of the ancestral population. B) PCA of variance stabilizing transformed expression counts from ancestral and evolved populations grown under planktonic conditions. Like panel A, arrows are drawn between corresponding ancestral and evolved populations and are colored by sub-lineage. Biofilm passaging reduced transcriptome diversity, with populations converging on a sub-lineage-specific signature (A); this pattern was not observed under planktonic growth conditions (B). N.B. While the positions of ancestral and evolved states are known (dots on the PCA), the trajectories between them are not, and connecting lines shown here are schematic. C) Bar plot of DEGs comparing evolved and ancestral biofilm populations broken down by sub-lineage. There are 680 DEGs shared between sub-lineages, and 405 and 200 DEGs unique to L4.4 and L4.9, respectively. Opacity of the bar indicates the subset of DEGs that are either up (dark) or downregulated (light) after passaging. The pellicle biofilm transcriptome is characterized by broad scale downregulation of gene expression with a smaller complement of genes that are upregulated in a sub-lineage specific manner. D) Differential expression of DEGs unique to each sub-lineage, from the comparison of evolved to ancestral biofilm populations. Log transformed adjusted p-values plotted against the log2 fold change for each gene. Genes that did not have significant differential expression are shown in grey. Lineage 4.9 evolved a larger complement of upregulated genes in response to biofilm passaging.
Fig 6.
Genome duplication results in complex patterns of gene expression.
A) Experimental diagram highlighting comparator populations: evolved and ancestral biofilm populations. B) Top: large tandem duplication that evolved in populations MT31 and MT55 under selection to grow as a biofilm. Bottom: coordinates of the duplication vary slightly between populations, with an average duplicated length of 175 kb. C) Left: log2 fold changes in gene expression between evolved and ancestral biofilm populations plotted against the position in the genome. Each point is a single gene, colored according to whether that gene lies inside of the duplicated region for that population. Right: Boxplots of log2 fold changes in gene expression, for genes outside (light blue) and inside (dark blue) the duplication. Genes within the duplication were significantly more likely to be upregulated in both populations (Mann Whitney U test with Benjamini-Hochberg correction, p < 0.0001).
Fig 7.
Expression of ncRNAs and sORFs mirror coding regions after passaging.
A) Volcano plot summarizing differential expression of ncRNAs and sORFs in evolved populations. Log transformed adjusted p-values plotted against the log2 fold change for each gene. Genes that did not have significant differential expression are shown in grey. B) Heatmap of normalized, variance stabilizing transformed expression counts for ncRNAs and sORFs with significant differential expression in evolved populations. Each column is a single sample from an evolved population, grown either as a biofilm or in a planktonic culture. Each row is a gene labeled either ncRNA or sORF. Expression values for each gene are normalized to the mean across samples. Samples are clustered by Euclidean distance and plotted as a tree at the top of the heatmap. C) Principal component analysis (PCA) of variance stabilizing transformed expression counts of ncRNAs and sORFs from ancestral and evolved biofilm populations. Arrows are drawn between corresponding ancestral and evolved populations, indicating the trajectory of evolution across passaging. Points are colored by sub-lineage of the ancestral population. The effects of biofilm passage on ncRNAs and sORFs mirror that of coding regions: reduced transcriptome diversity and widespread downregulation with populations converging on a sub-lineage-specific signature. N.B. While the positions of ancestral and evolved states are known (dots on the PCA), the trajectories between them are not, and connecting lines shown here are schematic.
Table 1.
Most significantly downregulated (< -5 L2FC) and upregulated (> 2 L2FC) ncRNAs in evolved populations grown as biofilms (compared to planktonic cultures).
Features included in this list are significantly differentially expressed when all populations were analyzed together, as well as in at least five individual populations. Log2 fold change (L2FC) values given for analysis of all populations together.
Fig 8.
Upregulation of ncRNA is a common feature of M. tb biofilm growth.
A) Log2 fold change (L2FC) values from ancestral populations by feature type. ncRNA have significantly higher L2FC values than ORFs and sORFs (Mann Whitney U Test with Benjamini-Hochberg correction, p < 0.0001) indicating they are more likely to be upregulated in biofilms. B) L2FC values between evolved and ancestral biofilms by feature type. ncRNA expression increased significantly more than either ORFs or sORFs after biofilm passage (Mann Whitney U Test with Benjamini-Hochberg correction, p < 0.0001).