Figure 5. Cell-to-cell variation in mitochondria potential predicts single cell drug resistance.

(A) Heatmap of expression of respiration genes in cells sorted by their TMRE signal intensity (bins HI-LO). (B) Heatmap of expression of DNA damage response and DNA repair genes. (C) Heatmap of expression of genes associated with iron deficiency (Puig et al., 2005; Veatch et al., 2009). Data are from four independent experiments. (D) Sorted bins from WT cells were grown in a commonly used antifungal drug fluconazole and were observed under microscope for growth over 7 days. The images show growth of cells in bins HI, M1, M2 and LO in 50 μg/ml of fluconazole after 7 days. (E) Cells of HI bin showed significantly higher survival compared to other bins in both 50 μg/ml (three independent experiments) and 60 μg/ml fluconazole (four independent experiments). Cells were grown in liquid medium supplemented with fluconazole on microscopy plates and viability was calculated from microscopic observations over 7 days. Colonies showing growth rate above 0.02 h⁻¹ after first time point were considered to be survivors. Error bars show ±1 s.d. values. (F) Percentage survival of high and low TMRE cells on fluconazole plates. High TMRE cells showed higher survival than low TMRE cells (Mann-Whitney U test). A substantial fraction of surviving high TMRE cells were respiration competent. The error bars represent ±1 s.d. values from six technical replicates for each bin. X-axis shows fluconazole concentrations used from three independent experiments. (G) From RNA sequencing data, cells from HI bin showed significantly higher expression of multidrug transporter PDR5 gene and its transcriptional activator PDR3 compared to cells from bins M1, M2 and LO. Results are from four independent experiments.

Figure 5—figure supplement 1. Expression patterns of diverse gene functional classes in cells with low, medium and high mitochondrial membrane potential.

(A) Functional classes showing significant enrichment among genes over-expressed in cells of bin HI compared to LO (hypergeometric test, p<0.01) and significantly enriched functional classes among genes showing lower expression in cells of bin HI compared to bin LO and the corresponding fold-change in expression. (B) Heatmap of expression of genes involved in cytoplasmic translation. (C) MA plot (Langmead and Salzberg, 2012) showing mean expression of genes vs. fold change in expression for comparison between cells in bin LO and bin M1. Differentially expressed genes are marked in red (FDR < 0.1). (D) Functional classes showing significant enrichment among genes over-expressed in cells of bin LO compared to M1 (p<0.01) and significantly enriched functional classes among genes showing lower expression in cells of bin LO compared to bin M1.

Figure 5—figure supplement 2. Stress survival and expression of stress response genes in cells with different levels of mitochondrial membrane potential.

(A) Left - % survival of cells from sorted bins HI-LO from WT strain after heat shock at 50°C for 2 min, 3 min and 4 min (Mann-Whitney U test). Right - Growth rate of cells in sorted bins HI, M1, M2 and LO from WT strain in SCD medium without any heat shock, after heat shock at 50°C for 2mins, 3mins and 4mins. (B) % survival of cells from HI-LO bins in 0.6 mM, 1 mM and 1.2 mM hydrogen peroxide (Mann-Whitney U test) (C) Expression of some key heat shock and stress response genes in bins HI-LO from RNA sequencing (data from four independent experiments).

Figure 5—figure supplement 3. Molecular pathways that underlie increased drug resistance in cells with high mitochondrial membrane potential.

(A) Percentage survival of sorted cells from bins HI, M1, M2, and LO from diploid BY4743 strain in 50 µg/ml of antifungal drug fluconazole. Cells from bin HI exhibited higher drug survival (results from two independent experiments). (B) PDR5-GFP strain was stained with TMRE and a sub-population of cells with high TMRE signal (corresponding to bin HI) also showed higher expression of PDR5 gene. (C) Deletion of the transcriptional activator PDR3 in PDR5-GFP strain wiped out the sub-population of cells showing higher expression of PDR5 gene. (D) Heatmap depicting expression of ergosterol biosynthesis genes in bins HI-LO sorted from WT cells.

Figure 5—figure supplement 4. Growth lag and slowdown in high-throughput microscopy assay.

(A) Example of microcolonies showing lag phase in growth. (B) Example of microcolonies showing growth slow down near the end of observation (C) Example of cells showing no lag and no growth slowdown.

Figure 5—figure supplement 5. Correlation of growth rates of deletion mutants within experimental replicates and with published datasets.

(A) Correlation between replicate measurements of mode growth rate of all strains and between replicate measurements of percentage slow fraction as well as between mode growth rate and percentage slow fraction. The panels below the diagonal show plots with the actual data points and the numbers above the diagonal show correlation values (generated using R package corrgram). (B) Correlation of mean growth rate for deletion mutants from our experiment and published data. The pie charts below the diagonal show magnitude of correlation and the numbers above the diagonal show the actual correlation values.

Figure 5—figure supplement 6. Test for recovery of growth rate, mtDNA copy number and respiration capability in slow-growing sub-population.

(A) Proliferation distribution measurement after 48 hr growth of sorted bins HI-LO from WT strain. (B) mtDNA copy number (using quantitative PCR) of colonies from sorted bins (HI-LO) from WT strain after 7 day growth on SCD plate. (C) Growth rate distribution measurement of small (respiration deficient) and big (respiration competent) colonies from high TMRE bin of WT strain after 48 hr of growth in SCD. Colonies were picked from SCDG plates after 7 days of growth and each curve represents data from one clone. (D) Percentage of colonies that regained capability to respire from the sub-population in the TMRE sorted bins HI, M1, M2 and LO in WT strain. Each data point shows result from an independent experiment. (E) Percentage of respiration deficient cells in sorted bins HI-LO from WT strain right after sorting (left), following 24 hr of growth in SCD medium (middle) and after 48 hr of growth in SCD medium (right).

Figure 5—figure supplement 7. Slow to fast switching in sorted sub-populations and in gene deletion mutants.

(A) Figure showing the microcolony growth rate distribution for the WT strain. The dotted line represents the growth rate of 0.3 h⁻¹ which was used as cutoff for defining slow and fast cells and to estimate slow to fast and fast to slow switching in microcolonies. (B) Percentage of microcolonies showing slow to fast switching and fast to slow switching in the sub-population of cells in TMRE sorted bins HI, M1, M2 and LO from WT strain. (C) Data showing the growth rates of microcolonies showing slow to fast switching (in red), fast to slow switching (in blue) and no switching (in black) from sorted bins in WT strain (D) Percentage of microcolonies showing slow to fast and fast to slow switching in WT strain and other deletion strains using the growth rate cutoff of 0.3 h⁻¹. (E) Correlation between percentage of cells showing fast to slow switching in WT and deletion strains and percentage of cells showing slow to fast transition with percentage of slow growing sub-population.

Figure 5—figure supplement 8. Percentage of microcolonies showing slow to fast switching and fast to slow switching in the sub-population of cells in TMRE sorted bins HI, M1, M2 and LO from WT strain using different growth rate cutoffs (0.24 h⁻¹, 0.26 h⁻¹,0.28h⁻¹,0.30h⁻¹,0.32h⁻¹,0.34h⁻¹, and 0.36 h⁻¹).

Each data point shows result from an independent experiment.

Figure 5—figure supplement 9. Percentage of microcolonies showing slow to fast switching and fast to slow switching in the sub-population of cells in TMRE sorted bins HI, M1, M2 and LO from WT strain using alternative growth rate cutoffs as shown in the figure.

Figure 5—figure supplement 10. Measurement of TMRE in cells of sorted bins HI-LO from WT strain.

(A) right after sorting (B) after 24 hr and 48 hr growth in SCD medium.

Figure 5—figure supplement 11. Microcolony growth rate distribution for cells in HI bin from TMRE sorted WT strain for calculation of switching rate from high to low TMRE state.

80% of the cells were estimated to be slow-growing. The mean growth rate (geometric mean) of the slow-growing sub-population was estimated to be 0.20 h⁻¹ and the mean growth rate of the fast-growing sub-population was estimated to be 0.37 h⁻¹.

Figure 5—figure supplement 12. Reproducibility of RNAseq experiments and growth rate measurements.

(A) Heatmap depicting distance between replicate RNA-seq samples of the sorted bins HI-LO of the WT strain. One replicate of LO bin showed high distance from all clusters. (B) PCA analysis of expression level of all genes for all replicates of bin HI-LO. Again, one replicate of bin LO was an outlier which was discarded before subsequent analyses. (C) Growth distributions for mip1Δ strain from the deletion collection and for two remade clones. The solid and the dotted lines represent replicate measurements on different days. The data clearly show that the growth rate in the strain from the collection is compensated. (D) Growth distributions for mrpl8Δ strain from the deletion collection and for two remade clones.