Moderate overexpression of hok does not suppress bacterial growth. A hok overexpression construct, pOE-hok, was previously generated by cloning the E. amylovora Ea1189 hok gene into the lac promoter-containing plasmid pEVS143 (27). The lac promoter allows low levels of transcription in the absence of the inducer isopropyl b-D-1-thiogalactopyranoside (IPTG) (35). We did not observe any growth defect in E. amylovora Ea1189 cells transformed with pOE-hok (Fig. S1), suggesting that E. amylovora is able to tolerate leaky hok expression without inhibiting growth. Therefore, we hypothesized that Ea1189(pOE-hok) grown in the absence of IPTG induction may provide a useful system to identify the physiological roles of Hok separate from those caused by its toxicity. We used quantitative real-time PCR (qRT-PCR) to measure the expression levels of hok in Ea1189(pEVS143) and Ea1189(pOE-hok) without IPTG and in four progressively increasing doses of IPTG, and monitored the growth of the cultures in the same conditions. In the absence of IPTG, expression of hok was approximately 40-fold higher in Ea1189(pOE-hok) compared to Ea1189(pEVS143), and expression of hok increased by another ~130-fold when 1mM IPTG was added to the Ea1189(pOE-hok) culture (Fig. 1A). The expression levels of the small RNA antitoxin sok remained almost unchanged in these conditions (Fig. 1A). Induction of hok did not result in cell death until expression reached about 60-fold induction or greater, induced by the addition of 0.01 mM IPTG (Fig. 1B). Henceforth, we will define hok expression from the lac promoter with 0.01 mM, 0.1 mM or 1 mM IPTG as the “toxic” expression conditions for this study, while expression from the lac promoter with 0.001 mM or no IPTG will be defined as the “subtoxic” expression conditions.
Induction of hok causes PMF collapse and membrane rupture. Membrane-associated type I toxins of E. coli, including HokB and TisB, form membrane pores (8, 10, 19), and cause collapse of the PMF (9, 10, 17). We therefore wondered if the transmembrane domain-containing E. amylovora Hok, sharing 48% and 14% amino acid identity to HokB and TisB, respectively, also causes membrane depolarization and rupture. To assess this possibility, we measured membrane potential using DiBAC4(3) (bis-(1,3-dibutylbarbituric acid) trimethine oxonol), a membrane potential-sensitive fluorescent dye. Fluorescence level negatively correlates to membrane potential, meaning that higher fluorescence indicates a greater level of PMF collapse. Carbonyl cyanide-m-chlorophenylhydrazone (CCCP), a protonophore that uncouples the PMF, was used as a positive control for the DiBAC4(3) staining (Fig. S2). Propidium iodide (PI) was used as an indicator of membrane rupture, which binds to nucleic acid and generates fluorescence in membrane integrity compromised cells. Ethanol disturbs the physical structure of cell membranes and was used as a positive control for the PI staining (Fig. S2). Fluorescence was measured in single cells using a flow cytometer. We found that induction of hok to subtoxic levels caused membrane depolarization and rupture in a subpopulation of cells, though many cells remained unchanged in their membrane states (Fig. 2A). More drastic membrane depolarization and rupture was observed when hok was induced to toxic levels (Fig. 2A). At the highest level of hok induction, almost the entire population was shifted to the membrane depolarization state, with varied levels of membrane rupture. We next asked whether mannitol, a bacterial metabolite that feeds into glycolysis and was shown to stimulate the PMF in E. coli (36), was able to restore the collapsed PMF and rupture of cell membrane due to the toxicity of Hok in E. amylovora. In cells expressing hok with 0.1 mM IPTG induction, mannitol partially relieved the membrane stress (Fig. 2A). Similarly, addition of mannitol significantly alleviated the inhibitory effect of bacterial growth during 0.01 or 0.1 mM induction of hok (Fig. 2B). However, when 1 mM IPTG was supplemented, the protective effect of mannitol was not observed in any of these phenotypes (Fig. 2A and Fig. 2B). Arabinose, which does not contribute to the PMF (36), was used a negative control for the assays (Fig. S3).
Transcriptomic analysis reveals that hok overexpression affects genes involved in stress responses and energy generation/consumption. While overexpression of E. amylovora hok causes extreme disturbance of essential membrane functions, it is not clear how the membrane disruption capacity of these toxins may affect bacterial physiology when hok is expressed in subtoxic or native expression conditions. To distinguish potential downstream effects of E. amylovora Hok from those resulting from toxicity, we compared the transcriptomes of E. amylovora cultures expressing hok at wild-type levels (i.e., wild-type strains carrying the empty vector) with cultures expressing hok at subtoxic (IPTG untreated) and toxic (1 mM IPTG treated) levels. Expression of each gene was quantified as counts per million reads (CPM), and differentially-expressed genes (DEGs) were defined as those having greater than 2-fold change of CPM values and less than 0.05 of the corresponding false discovery rate (FDR) values (Fig. S4 and Table S1).
Compared with Ea1189(pEVS143), which was also untreated with IPTG, 321 DEGs were identified in IPTG-untreated Ea1189(pOE-hok), of which 234 had increased expression and 87 had decreased expression (Fig. 3A). After 1 mM IPTG treatment of Ea1189(pOE-hok), a much larger set of 541 and 560 genes were up- and down-regulated, respectively (Fig. 3A). Approximately 83% of the DEGs identified in the subtoxic condition were differentially expressed in the same direction and to a greater extent in the toxic condition. Expression of representative genes in Ea1189(pOE-hok) in subtoxic and toxic conditions was validated through qRT-PCR (Fig. 3B). The housekeeping gene recA was used as an endogenous control, that had negligible differences in expression among E. amylovora cultures expressing wild-type, subtoxic, or toxic levels of hok in our transcriptomic analysis. Based on the read count, the ratio of hok to sok was approximately 18 in the wild-type condition, that increased to ~200 in the subtoxic condition and ~6,000 in the toxic condition (Fig. S5). Gene ontology (GO) enrichment analysis of the DEGs further revealed that hok exerts substantial effects in the essential metabolism of E. amylovora (Fig. 4 and Table S2). Oxidative phosphorylation-related genes (GO:0006119), that include NADH-coenzyme Q oxidoreductase (complex I), Succinate-Q oxidoreductase (complex II), Cytochrome c oxidase (complex IV) and F1Fo-ATPase (complex V), were enriched among the higher expressed genes in both toxic and subtoxic conditions. Specifically, in the toxic condition, higher expressed genes were also significantly associated with the “tricarboxylic acid cycle” GO term (GO:0006099).
Several genes with demonstrated importance to bacterial plant pathogenesis were negatively affected by elevated hok expression. Specifically, hrpA and flhD, encoding a T3SS protein and a flagellar transcriptional activator, respectively, decreased in expression at both levels of hok induction. In toxic but not subtoxic conditions, down-regulated genes were primarily comprised of flagellar genes and “protein secretion” (GO:0009306) genes, which included type II secretion system (T2SS) and type III secretion system (T3SS)-related genes.
Induction of hok also activated multiple genes involved in stress responses. Several genes with known roles in antibiotic persistence and other stress responses, i.e. groS, groL, dnaK, dnaJ, skp, surA, sucB and lon (37-42), were consistently more highly expressed in both hok induction conditions. Also upregulated were genes in the “response to virus” ontology (GO:0009615), including genes encoding phage shock proteins, i.e. pspABCD, and CRISPR-associated proteins. The catalase gene katA showed increased expression in the subtoxic condition, consistent with our previous observation that catalase activity is significantly compromised in a hok-sok deletion mutant (27). The stress-induced ATP-dependent chaperone gene clpB was also more highly expressed in the subtoxic but not the toxic condition. Together, these results show that different hok expression levels exert diverse and overlapping effects on the E. amylovora transcriptome, enhancing expression of metabolic and stress-related traits while suppressing genes required for infection.
hok positively affects ATP biosynthesis. Membrane-associated type I toxins have been shown to cause leakage of cellular ATP as indicated by either decrease level of intracellular ATP or increase level of extracellular ATP (7, 10, 19). In this study, we found that genes associated with oxidative phosphorylation, the process of ATP generation through electron transfer, were higher expressed in the subtoxic condition and were higher expressed to a greater extent in the toxic condition (Fig. 5). We hypothesized that the upregulation of ATP biogenesis-related genes could be part of a response to compensate for the possible leakage of intracellular ATP through increased ATP synthesis in Ea1189(pOE-hok) cultures in both subtoxic and toxic conditions. To determine whether ATP leakage was occurring, we performed simultaneous measurements of both the intracellular and the extracellular levels of ATP in both subtoxic and toxic conditions. When induced with 0.1 or 1 mM IPTG, conditions causing more than 70% dieoff (Fig. 1B), E. amylovora Hok caused dramatic leakage of ATP from the cells, indicated by the decreased level of intracellular ATP and increased level of extracellular ATP (Fig. 5 and Fig. S6). In contrast, a significant increase in intracellular ATP was measured after induction with 0.01 mM or less IPTG (Fig. 5 and Fig. S6), expression conditions that were associated with minimal or no cell death of E. amylovora (Fig. 1B). No ATP leakage was observed in these subtoxic conditions.
Combining intercellular and extracellular ATP measurements allowed us to assess the total ATP concentration under each expression condition. In the absence of IPTG, total ATP was greater in Ea1189(pOE-hok) cultures than Ea1189(pEVS143). Total ATP in Ea1189(pOE-hok) increased with IPTG addition at concentrations up to 0.1 mM (Fig. 5). At the highest concentration of IPTG tested, 1 mM, the total ATP in Ea1189(pOE-hok) cultures started to decrease compared with lower levels of inducer, likely due to the massive kill-off of ATP-generating cells at this induction level. Taken together, our results suggest that hok positively affects the biosynthesis of ATP, and leakage of ATP only occurs when hok was induced at toxic levels.
Overexpression of the ATP synthase gene atpB is toxic to E. coli cells; it allows leakage of protons through the F0 sector of F1Fo-ATPase (43-46). Given that hok positively affects ATP synthase gene expression and ATP biosynthesis in subtoxic conditions, we wondered if the toxicity of Hok was increased by the upregulation in ATP synthase genes. To test this hypothesis, we generated ATP synthase gene deletion mutants, Ea1189ΔatpB and Ea1189ΔatpBEFHAGDC. The growth of Ea1189ΔatpB and Ea1189ΔatpBEFHAGDC mutants was severely reduced, as overnight cultures only reached OD600 ≈ 0.3 compared with OD600 ≈ 1.5 in the wild-type Ea1189 strain (data not shown). pOE-hok was transformed into the ATP synthase mutants to generate Ea1189ΔatpB(pOE-hok) and Ea1189ΔatpBEFHAGDC(pOE-hok), respectively. Hok expression was induced in the wild-type and ATPase mutant backgrounds with 1 mM IPTG, and survival rates were measured. Hok killing efficiency was not changed between the wild-type and the mutants (Fig. S7), suggesting that the toxicity of Hok is not affected by the increased expression of ATP biosynthesis genes.
Expression of pspA is induced in known PMF dissipation conditions and relieves the toxicity of Hok. Our transcriptome results indicated that psp genes were upregulated in both expression conditions. The psp genes are induced on exposure to conditions that dissipate the PMF, such as bacteriophage infection, alkaline pH, and addition of uncoupling agents, in both Gram-negative and -positive bacteria (reviewed in (47)). The protective roles of PspA in managing membrane stresses have been validated in E. coli and Salmonella enterica serovar Typhimurium (48-50). As the functions of psp genes have not been previously investigated in E. amylovora, we constructed a transcriptional fusion of the promoter region of the pspABCD operon to a green fluorescence protein (gfp) reporter. As expected, the promoter activity of the pspABCD operon was significantly increased in E. amylovora cells after exposure to bacteriophage, and was increased to a lesser extent in the presence of CCCP, ethanol, or Triton X-100 (Fig. S8). To examine the possible protective role of pspA under the condition of membrane stress in E. amylovora, we generated the pspA-overexpression construct, pBAD33-pspA, through cloning the pspA gene into the pBAD33 plasmid, containing the arabinose-inducible PBAD promoter. Compared with Ea1189(pBAD33), Ea1189(pBAD33-pspA) cultures were ~100 times more tolerant to CCCP (Fig. 6A). Interestingly, without supplementing any IPTG, Ea1189(pOE-hok) cultures survived at significantly higher rates than Ea1189(pEVS143) (Fig. 6A), suggesting that induction of hok at subtoxic levels protect E. amylovora cells from further membrane damage by activating the expression of pspA. Interestingly, pspA overexpression significantly alleviated the toxicity due to high levels of hok induction (Fig. 6B), further validating the defensive role of pspA in response to membrane stress in E. amylovora.
Subtoxic expression of hok increases tolerance of stationary-phase E. amylovora cells to the aminoglycoside antibiotic streptomycin. Transcriptome results showed that hok expression upregulated several genes previously associated with antibiotic persistence, so we next asked whether hok has a role in antibiotic tolerance during stationary phase. Without addition of IPTG, stationary phase E. amylovora cultures expressing hok had 10 times the number of survivors to streptomycin exposure than the vector control strain (Fig. 7). concentration that is routinely used for management of fire blight and screening of streptomycin-resistant E. amylovora isolates (51-53). Of note, we did not observe altered tolerance of E. amylovora cultures overexpressing pspA, suggesting that hok does not affect antibiotic tolerance through overproduction of PspA.