Development of heat-shock resistance in Legionella pneumophila modeled by experimental evolution

Abstract

Because it can grow in buildings with complex hot water distribution systems (HWDS), healthcare facilities recognize the waterborne bacterium Legionella pneumophila as a major nosocomial infection threat and often try to clear the systems with a pasteurization process known as superheat-and-flush. After this treatment, many facilities find that the contaminating populations slowly recover, suggesting the possibility of in situ evolution favoring increased survival in high-temperature conditions. To mimic this process in a controlled environment, an adaptive laboratory evolution model was used to select a wild-type strain of L. pneumophila for survival to transient exposures to temperatures characteristic of routine hot water use or failed pasteurization processes in HWDS. Over their evolution, these populations became insensitive to exposure to 55°C and developed the ability to survive short exposures to 59°C heat shock. Heat-adapted lineages maintained a higher expression of heat-shock genes during low-temperature incubation in freshwater, suggesting a pre-adaptation to heat stress. Although there were distinct mutation profiles in each of the heat-adapted lineages, each acquired multiple mutations in the DnaJ/DnaK/ClpB disaggregase complex, as well as mutations in chaperone htpG and protease clpX. These mutations were specific to heat-shock survival and were not seen in control lineages included in the experimental model without exposure to heat shock. This study supports in situ observations of adaptation to heat stress and demonstrates the potential of L. pneumophila to develop resistance to control measures. IMPORTANCE As a bacterium that thrives in warm water ecosystems, Legionella pneumophila is a key factor motivating regulations on hot water systems. Two major measures to control Legionella are high circulating temperatures intended to curtail growth and the use of superheat-and-flush pasteurization processes to eliminate established populations. Facilities often suffer recolonization of their hot water systems; hospitals are particularly at risk due to the severe nosocomial pneumoniae caused by Legionella. To understand these long-term survivors, we have used an adaptive laboratory evolution model to replicate this process. We find major differences between the mutational profiles of heat-adapted and heat-naïve L. pneumophila populations including mutations in major heat-shock genes like chaperones and proteases. This model demonstrates that well-validated treatment protocols are needed to clear contaminated systems and-in an analog to antibiotic resistance-the importance of complete eradication of the resident population to prevent selection for more persistent bacteria.

Keywords: Legionella pneumophila; adaptive laboratory evolution; experimental evolution; heat resistance; hot water distribution systems; pasteurization.

Fig 1

Adaptive laboratory evolution increased the resistance of L. pneumophila to heat shock. (A) Brief schematic showing the workflow of each passage in the evolutionary model for both the control and heat-adapted branches (created with BioRender). Each panel depicts 1 of 70 cycles, with highlighted rectangles showing the samples used to propagate the next passage. In brief, after collection from post-exponential growth in AYE and 24 h suspension in Fraquil, control lineages were propagated by subsampling the suspended population to re-inoculate AYE while heat-adapted lineages were propagated by re-inoculating AYE with the population surviving heat shock. (B) Survival of heat-adapted L. pneumophila lineages following 20 min of exposure to 55°C, 57°C, and 59°C heat shock for 20 min (lines, left axis). Decimal reduction times (DRT) were calculated for each temperature as the amounts of time required for a 90% reduction in population count and are plotted for each tested passage (bars, right axis; omitted for passages 65 and 70 at 55°C because there was an insignificant level of cell death). Survival was measured independently (with three technical replicates) for each of six heat-adapted lineages at five passage intervals. Data show mean of all lineages with error bars showing ±SD, n = 6. (C) Direct competition between green-fluorescent isolates and non-fluorescent isolates growing in AYE. Significance calculations show one-sample t-test against 1.0 (neutral relative fitness), n = 3. (D) Infection and replication within V. vermiformis in modified PYNFH without FBS, n = 3. The avirulent DotA(−) strain is used as a negative control.

Fig 2

qPCR analysis of RNA expression levels in the ancestor Philadelphia-1 following 5 min of exposure to 55°C heat shock for heat-shock response genes (A) and sigma factors (B). Ct values were normalized using 16S rRNA levels and converted to arbitrary expression levels, n = 3. Relative difference in expression in RNA samples collected from control lineages (C-) and heat-adapted lineages (HA-) before heat shock (C) and after heat shock (D) compared to Philadelphia-1, n = 3.

Fig 3

(A) Derived mutations identified with Breseq fixed in replicate control and heat-adapted lineages aligned to the genome of the ancestral strain Philadelphia-1 (68). The outermost concentric circle shows the single chromosome of L. pneumophila, as well as a reversibly integrated pPh38 element which was not retained in our lab strain. Each band shows the mutations observed across control lineage (orange) and heat-adapted (green), with circles representing point mutations and triangles representing insertions and deletions. Genes of interest in the interior are labeled by color with red genes mutated in multiple heat-adapted lineages and blue genes mutated in multiple control lineages. (B) Key participants in the bacterial heat-shock response include chaperones, proteases, and disaggregases, which act in a protein quality control network to direct proper protein function in nascent or damaged polypeptides. Proteins highlighted by a pink box acquired mutations in this adaptive laboratory evolution system. Multimeric proteins are depicted as monomeric structures for simplicity (created with BioRender). (C) Rates of population decline during exposure to 55°C in a water bath do not significantly differ between wild-type KS79 and a lidA deletion mutant, n = 3. (D) Rates of population decline of mutant strain carrying dnaK alleles observed in heat-adapted lineages, wild-type KS79 (WT), and heat-adapted population (HA-1 and HA-2) during exposure to 55°C, n = 6 (WT) or n = 3 (the remainder). (C) and (D) Point and error bars show mean ± SEM of the linear regression between log₁₀ population values and time of exposure measured over 15 min (C) or 30 min (D). Statistical significance computed from slope and SEM using (C) Student’s t-test or (D) ANCOVA with Holm-Šídák correction for multiple comparisons; * = P < 0.05,*** = P < 0.0001 .

Fig 4

(A) Anti-htpG Western blots visualizing the expression of HtpG in lineages HA-3, 5, and 6 and in constructed merodiploid mutants (representative image). (B) Survival relative to initial population size of knock-in exchange mutants expressing derived alleles of htpG in place of the wild-type allele over a 30-min challenge of heat shock at 55°C, n = 3. (C) Survival relative to initial population size of an htpG deletion mutant heterologously expressing different alleles of htpG found in treated lineages over a 30-min challenge of heat shock at 55°C, n = 3. (D) Relative fitness of the htpG deletion mutant complemented with an empty plasmid, derived, or ancestral htpG alleles against the wild-type and deletion mutant strains over a 24-h incubation in AYE, n = 3. (B and C) Data shown represent mean population surviving relative to population size prior to heat shock (i.e., time 0) ± SD. (D) Data show mean relative fitness ± SD. Statistical significance calculated with two-way ANOVA in Prism 5.3.1 with Holm-Šídák correction for multiple comparisons; * = P < 0.005, ** = P < 0.01.