RNA Quality Control Enables Antibiotic Tolerance

Description

Antibiotic resistance poses a significant clinical challenge, yet the mechanisms underlying adaptation to antibiotic pressure within the complex host environment remain incompletely understood. By experimentally evolving Streptococcus pneumoniae in mice subjected to various antibiotics and immune states, we demonstrate that the high fitness costs of canonical resistance mutations severely restrict their emergence in vivo. Instead, populations adopt distinct adaptive strategies depending on the specific selective context: while neutrophil-replete environments select for immune-evasive, loss-of-function mutations in the nicotinamidase SP_1583, general antibiotic stress drives convergent mutations in rny, encoding the RNA degradosome scaffold RNase Y. Isogenic rny mutants exhibit broad-spectrum tolerance and accelerated recovery. Single-cell transcriptomics reveals that antibiotic-induced death in wild-type bacteria is driven by transcriptional collapse, a lethal surge of uncontrolled transcription followed by a catastrophic loss of RNA quantity and integrity. In contrast, rny mutants avert this fate via a bet-hedging strategy: a resilient minority maintains a near baseline transcriptional profile, while a quiescent majority undergoes rapid, selective RNA degradation to preserve transcript fidelity. Upon stress removal, these populations execute a prioritized transcriptional ribosomal reboot, facilitating accelerated recovery. These findings identify RNA turnover as a tunable master regulator of stress tolerance that pathogens can exploit to survive the combined pressures of antibiotics and immunity.

Steps to reproduce

This repository contains all scripts, intermediate data, and processed input files required to reproduce the single-cell RNA-seq downstream analyses and figures described in the associated manuscript. SCRIPTS 1. data_preprocessing.py All dependencies needed to run this script are listed in 'scrna_env.yml'. This script performs the following steps: 1) Produces Scanpy .h5ad files and corresponding cell count summaries 2) Calculates RNA content per cell 3) Quantifies the number of cells per cluster per sample 4) Computes transcriptional entropy 5) Estimates cellular transitions using CellRank 6) Calculates normalized degradation and recovery rates Run the individual cell for each analysis. Required data files are available in this repository. 2. plotting.py All dependencies needed to run this script are listed in 'scrna_env.yml'. This script is organized according to the order of figures in the study. Each code cell corresponds to a specific figure (e.g., Fig 1C). 3. run_cellrank.py This script generates the random walk plots using "rnyL77S_tmk.pkl" and "/scanpy_h5ad_files/rnyL77S_louvain_01.h5ad". All dependencies needed to run this script are listed in 'cellrank.yml'. FOLDERS AND FILES 1. Input gene-by-cell matrix files in gene_cell_matrix/ 2. Preprocessed .h5ad files in scanpy_h5ad_files/ 3. Per-cell RNA content files in per-cell_RNA_content/ (related to Figure 1D, 2C, S1A) 4. Normalized RNA degradation/recovery rates files in normalized_degradation_recovery_rate/ (related to Figure 3C-J, S3A-F, S4A-F) 5. ALL 3 Rep CFU Cells per mL FINAL both strains.csv CFU values at each time point for both WT and mutant (Figure 1A) 6. Cef_all_samples_louvain_resolution0.1_cell_count.tsv includes the number of cells in each cluster per sample (Figure 2B) 7. DE_NDC_clusters_louvain01.tsv includes the resulting log2FC values used in entropy calculation (Figure 2D) 8. rnyL77S_tmk.pkl includes the saved trained CellRank RealTime Kernel (related to Figure 2F, S1B) 9. Annotation_TIGR4.tsv includes the gene annotations (used for GSEA analysis in Figure S3GH)

Institutions

• Boston Children's Hospital

  MA, Boston
• Harvard Medical School

  MA, Boston

Categories

Licence