Composition and liquid-to-solid maturation of protein aggregates contribute to bacterial dormancy development and recovery

Abstract

Recalcitrant bacterial infections can be caused by various types of dormant bacteria, including persisters and viable but nonculturable (VBNC) cells. Despite their clinical importance, we know fairly little about bacterial dormancy development and recovery. Previously, we established a correlation between protein aggregation and dormancy in Escherichia coli. Here, we present further support for a direct relationship between both. Our experiments demonstrate that aggregates progressively sequester proteins involved in energy production, thereby likely causing ATP depletion and dormancy. Furthermore, we demonstrate that structural features of protein aggregates determine the cell's ability to exit dormancy and resume growth. Proteins were shown to first assemble in liquid-like condensates that solidify over time. This liquid-to-solid phase transition impedes aggregate dissolution, thereby preventing growth resumption. Our data support a model in which aggregate structure, rather than cellular activity, marks the transition from the persister to the VBNC state.

Fig. 1. Dormancy and aggregation are dynamic processes that are often highly correlated.

The strains analyzed for their effect on dormancy are the wild-type E. coli strain (WT), E. coli ΔrpoS, ΔsdhC and ΔrecA and E. coli with pBAD33Gm (Vector), pBAD33Gm-obgE (ObgE) and pBAD33Gm-era (Era). a, b The absolute number of persister cells that could grow after ofloxacin treatment was measured at different time points during incubation. Data are represented as averages ± SEM (n ≥ 4) (See Supplementary Table 2 for exact values). Relative values of persisters, VBNC and dead cells are shown in Supplementary Fig. 1. c–f Quantitative microscopy analysis was used to determine the fraction of cells with IbpA (c, d) and Ph (e, f) aggregates at different time points during incubation. A minimum of 50 cells was analyzed for every repeat and every time point. A nonlinear logistic growth model was fitted on the data (solid line). Data are represented as averages ± SEM, (n ≥ 3) (See Supplementary Table 3 and 4 for exact values). g For each strain, the moment of maximal persistence and the start and maximal levels of VBNC cell formation and IbpA and Ph aggregation were determined and visualized. The moment of the start and maximum of IbpA and Ph aggregation were determined by fitting a nonlinear logistic growth model on the aggregation data and calculating the time between reaching 10% and 90% of the maximal value. h Heatmap of the Pearson’s R values that were calculated based on the timing of maximal persistence, the onset of VBNC cell formation and the peak in IbpA and Ph aggregation across the selected strains (n = 7). All correlations were high and significant (P < 0.05) (See Supplementary Table 5 for confidence intervals and p-values). Source data are provided as a Source Data file.

Fig. 2. Aggregates preferably sequester proteins involved in energy production, translation and metabolism.

Aggregate composition was quantified by MS using label-free quantitation values based on spectral counting after 40 hours of incubation for E. coli WT, E. coli ΔrpoS, ΔsdhC and ΔrecA and E. coli with pBAD33Gm (Vector) and pBAD33Gm-obgE (ObgE). a This graph shows the number of unique proteins that were aggregated in all 6 strains, in only one strain, and everything in between. b The number of unique aggregated proteins in each strain is indicated. Bar graphs and error bars represent the mean ± SEM (n = 3, with n the number of samples per strain from which the aggregates were purified). Mean and SEM for WT, ΔrpoS, ΔsdhC, ΔrecA, Vector and ObgE strains are 239.3 ± 65.71, 304.7 ± 10.65, 254 ± 18.81, 265.7 ± 50.34, 259.7 ± 28.59 and 190.7 ± 14.19. c Aggregate composition data of all samples were combined to determine which COG categories were overrepresented using the one-sided Fisher’s exact test with Bonferroni correction for multiple testing. d Aggregate composition data of all the strains were combined to determine which KEGG pathways were enriched using a hypergeometric test with the Benjamini-Hochberg multiple testing correction. Overrepresented categories are shown (P < 0.05). Exact p-values are (top to bottom): 0.0024, 0.0041, 0.0067, 0.00026, 0.0029, 1.01 × 10⁻⁸, 0.0046, 0.0057, 1.16 × 10⁻⁵, 3.41 × 10⁻¹⁰, 0.0046, 0.0104, 0.00036, 8.72 × 10⁻⁵, 0.0096, 1.58 × 10⁻⁸, 6.73 × 10⁻⁵ and 8.35 × 10⁻¹⁹. The absolute number of aggregated proteins that belong to each category is indicated next to the bars. COG categories represented in c are: A—RNA processing and modification, C—energy production and conversion, D—cell cycle control, cell division, and chromosome partitioning, E—amino acid transport and metabolism, F—nucleotide transport and metabolism, G—carbohydrate transport and metabolism, H—coenzyme transport and metabolism, I—lipid transport and metabolism, J—translation, ribosomal structure, and biogenesis, K—transcription, L—replication, recombination, and repair, M—cell wall/membrane/envelope biogenesis, N—cell motility, O—posttranslational modification, protein turnover, and chaperones, P—inorganic ion transport and metabolism, Q—secondary metabolites biosynthesis, transport, and catabolism, S—function unknown, T—signal transduction mechanisms, U—intracellular trafficking, secretion, and vesicular transport, V—defense mechanisms. Source data are provided as a Source Data file.

Fig. 3. Aggregates progressively sequester proteins involved in energy production and translation.

Aggregates were isolated and their composition was quantified for E. coli pBAD33Gm (Vector) (a) after 8, 16, 32, 48 and 72 hours of incubation and for E. coli pBAD33Gm-obgE (ObgE) (b) after 8, 16, 24, 40 and 72 hours of incubation (n = 4 for V32 and n = 5 for all the other samples). For the last 4 sampling points, protein abundances in the aggregate were compared to the abundances in the previous sampling point to identify enriched proteins and their COG categories using the one-sided Fisher’s exact test with Bonferroni correction for multiple testing. Source data are provided as a Source Data file.

Fig. 4. Early-stage IbpA aggregation precedes ATP depletion, membrane depolarization and VBNC cell formation in stationary phase.

All measurements were performed with E. coli pBAD33Gm (Vector) and pBAD33Gm-obgE (ObgE). a Based on the single-cell ATP measurements and the threshold for ATP depletion indicated in Supplementary Fig. 4, the fraction of cells with ATP depletion was determined at different time points. A nonlinear logistic growth model was fitted onto the data (solid line). b The fraction of cells with membrane depolarization was determined at different time points by measuring the proportion of the bacterial population that stained with DiBAC₄(3), a fluorescent dye that penetrates depolarized cells. A nonlinear logistic growth model was fitted onto the data (solid line). c Translation was measured using the Click-iT® HPG Alexa Fluor® Protein Synthesis Assay Kit. As a negative control, the Vector was treated with the translation inhibitor chloramphenicol (Cm). For each repeat and each time point, the sum of the fluorescence intensity in 10,000 bacterial cells is shown. Translation levels were compared between the Vector and ObgE at the different time points by unpaired two-sided t-tests with the Holm-Šídák correction for multiple comparisons (for the comparison after 48 hours of incubation: ****P = 0.000082). Data in graph a–c are represented as averages ± SEM, (n = 4) (See Supplementary Table 7 for exact values). d The timing of dormancy (persistence and VBNC cell formation), energy depletion (ATP depletion and membrane depolarization) and aggregation (IbpA and Ph aggregation) is represented. For persistence, the peak moment is indicated, while for the other processes, both the initiation and the peak moment are shown. Source data are provided as a Source Data file.

Fig. 5. DnaK decreases dormancy and is important for regrowth.

Aggregation, persistence and VBNC cell formation were measured for E. coli WT, E. coli ΔdnaK (a-d), E. coli pBAD33Gm (Vector), and E. coli pBAD33Gm-dnaK (DnaK) (e–h). a, e Quantitative analysis of microscopy images shows the fraction of cells containing Ph aggregates. Data are represented as averages ± SEM, (n ≥ 3, > 50 cells analyzed for every repeat) (Supplementary Tables 2-4 and 9-12 for exact values). b, f Persistence was measured at different time points. Data are represented as averages ± SEM, (n ≥ 4) (Supplementary Tables 2-4 and 9-12 for exact values). c, d, g, h The number of CFUs (solid line) and viable cells (dashed line) were measured at different time points. The area between both curves represents the number of VBNC cells. Data are represented as averages ± SEM, (n ≥ 4) (See Supplementary Tables 2-4 and 9-12 for exact values). Note that viable cells and CFUs are measured differently and are therefore not perfectly comparable. The discrepancy observed at some early timepoints suggests that the number of viable cells may be underestimated. i The number of CFUs and viable cells was determined after 72 hours of obgE overexpression in WT and deletion strains. The logarithm of the number of viable and culturable cells was compared between WT and deletion strains using a two-way ANOVA with Dunnett’s correction for multiple comparisons (ΔdnaK: q ratio = 9.771, 42 degrees of freedom, ****P < 0.0001). Bar graphs and error bars represent averages ± SEM, (n ≥ 3) (See Supplementary Table 13 for exact values). j The percentage of proteins that are aggregated after 40 hours of incubation and that are a substrate of DnaK was calculated. Bar graphs and error bars represent averages ± SEM, (n = 3, with n the number of samples per strain from which the aggregates were purified). For WT, ΔrpoS, ΔsdhC, ΔrecA, Vector and ObgE strains, the mean ± SEM values are 43.56 ± 2.123, 42.27 ± 0.5844, 44.07 ± 1.067, 44.55 ± 1.785, 42.59 ± 0.4752 and 48.01 ± 0.5392. Source data are provided as a Source Data file.

Fig. 6. Early-stage condensates show liquid-like properties, while solidified late-stage aggregates contain more intermolecular β-sheets.

After 8 (a, c) and 40 (b, c) hours of incubation, E. coli ibpA-msfGFP with pBAD33Gm (Vector) or pBAD33Gm-obgE (ObgE) and E. coli ibpA-msfGFP ΔdnaK were imaged with and without 10% 1,6-hexanediol. Quantitative microscopy analysis shows the fraction of cells with IbpA foci in each condition. For the Vector, ObgE and ΔdnaK strains, the fraction of cells with IbpA foci was compared between cells treated with and without hexanediol using a two-way ANOVA with Holm-Šídák correction for multiple comparisons. Bar graphs and error bars represent averages ± SEM, (n = 4, ≥ 50 cells were analyzed for each repeat) (see Supplementary Table 13 for exact values). c Representative microscopy images are shown. Scale bar, 5 µm. FRAP was performed on a small part of the IbpA foci of E. coli ibpA-msfGFP pBAD33Gm-obgE after 8 (d) and 40 (e) hours of incubation. Representative images before and right after bleaching, and after recovery are presented as well as the relative fluorescence compared to the starting point. Scale bar, 1 µm. f The percentage recovery after bleaching is shown. The recovery was compared using a two-sided Wilcoxon signed-rank test. Data are represented as averages ± SEM. E. coli pBAD33Gm-obgE was incubated for 2 (g), 8 (h) and 40 (i) hours and 3 repeats of the samples were imaged using AFM-IR spectroscopy. The results of the first repeat are presented here. The IR absorbance at 1625 cm⁻¹ for the complete sample is shown on the left. Scale bar, 5 µm. Different foci within the sample were selected for which a full IR spectrum was measured (indicated by crosses). The complete IR spectra of these foci are shown on the right. k-means clustering on the second derivative of the amide I band (1580 cm⁻¹–1700 cm⁻¹) was performed to discriminate between spectra with high and low β-sheet signal (red and blue, respectively). j For each complete IR spectrum (Fig. 6g–i and Supplementary Fig. 6d–i), the fraction of β-sheets compared to α-helices was determined. Source data are provided as a Source Data file.

Fig. 7. Model describing how protein aggregation induces different categories of dormancy.

In response to stress such as nutrient exhaustion, bacteria progressively sequester proteins in liquid-like condensates. The condensation of proteins active in energy production precedes—and thereby likely induces—a depletion of cellular ATP and dormancy. The chaperone DnaK is able to dissolve these condensates due to their liquid-like structure. This dissolution allows the recovery and regrowth of the persister cells. Restart of growth is supported by dissoluted and refolded proteins that are released from the biomolecular condensate. Indeed, the condensates in dormant cells contain a variety of proteins involved in translation, energy production and metabolism which could be repurposed for growth. However, if stress persists over an extended period of time, the liquid-like condensates solidify into aggregates by accumulating intermolecular β-sheet structures. This transition complicates their removal by DnaK and consequently the exit from dormancy. The solidification of the aggregates thereby pushes the shallowly dormant persister cells in the deeper dormant VBNC state. Under specific conditions, solidified aggregates from VBNC cells can presumably be converted back to liquid-like structures (dashed arrow), which is thought to allow disaggregation by DnaK and subsequent regrowth.