Increased intracellular persulfide levels attenuate HlyU-mediated hemolysin transcriptional activation in Vibrio cholerae

Abstract

The vertebrate host's immune system and resident commensal bacteria deploy a range of highly reactive small molecules that provide a barrier against infections by microbial pathogens. Gut pathogens, such as Vibrio cholerae, sense and respond to these stressors by modulating the expression of exotoxins that are crucial for colonization. Here, we employ mass spectrometry-based profiling, metabolomics, expression assays, and biophysical approaches to show that transcriptional activation of the hemolysin gene hlyA in V. cholerae is regulated by intracellular forms of sulfur with sulfur-sulfur bonds, termed reactive sulfur species (RSS). We first present a comprehensive sequence similarity network analysis of the arsenic repressor superfamily of transcriptional regulators, where RSS and hydrogen peroxide sensors segregate into distinct clusters of sequences. We show that HlyU, transcriptional activator of hlyA in V. cholerae, belongs to the RSS-sensing cluster and readily reacts with organic persulfides, showing no reactivity or DNA dissociation following treatment with glutathione disulfide or hydrogen peroxide. Surprisingly, in V. cholerae cell cultures, both sulfide and peroxide treatment downregulate HlyU-dependent transcriptional activation of hlyA. However, RSS metabolite profiling shows that both sulfide and peroxide treatment raise the endogenous inorganic sulfide and disulfide levels to a similar extent, accounting for this crosstalk, and confirming that V. cholerae attenuates HlyU-mediated activation of hlyA in a specific response to intracellular RSS. These findings provide new evidence that gut pathogens may harness RSS-sensing as an evolutionary adaptation that allows them to overcome the gut inflammatory response by modulating the expression of exotoxins.

Keywords: Vibrio cholerae; bacterial pathogenesis; bacterial toxin; bacterial transcription; homocysteine; host–pathogen interaction; sulfur; thiol; transcription regulation.

Figure 1

Sequence similarity network analysis of the ArsR superfamily of bacterial repressors.A, results of an SSN clustering analysis of 168,163 unique sequences belonging to the Pfam PF01022 and Interpro IPR001845 using genomic enzymology tools and visualized using Cytoscape. Clusters were functionally annotated as arsenic, transition metal ions, persulfide, or hydrogen peroxide sensors are presented here, along with a representation of the inducer recognition site as an inset mapped onto a representative ArsR structure (PDB codes: 6j0e, 2kjc, 6o8n, 7txm, and 6o8l, respectively). All the determined main clusters are presented in Fig. S1. All clusters are designated by a number and ranked according to the number of unique sequences (Table S2), color-coded. Each node corresponds to sequences that are 50% identical over 80% of the sequence, using an alignment score of 22 (see Experimental procedures). Functionally characterized members in each are indicated with species name and trivial name, except for vcBigR, which is included for clarity and has not been characterized biochemically. B, sequence logo representations of sequence conservation defined by the indicated cluster of sequences derived from panel A. The residues that coordinate metals/metalloid ions or undergo redox chemistry in ArsR are marked by stars. We note that coordinating residues in variable regions in the N terminus or C terminus do not always appear in the sequence logos. ArsR, arsenic repressor; SSN, sequence similarity network.

Figure 2

LC-ESI-MS analysis of HlyU in vitro reactivity upon the addition of GSSH or H₂O_2.A, HlyU reacts with IAM but not with dimedone, showing that the protein is fully reduced. B, HlyU reacts with GSSH to form a tetrasulfide link between its two cysteines, as seen in the mass shift of +62, corresponding to the addition of two sulfur atoms and the subtraction of two hydrogens. C, HlyU does not react with H₂O₂, as seen by the absence of a peak corresponding to a dimedone adduct (127). IAM, iodoacetamide.

Figure 3

DNA-binding isotherms of VcHlyU over its DNA operator in different oxidation states at 100 mM NaCl.A, reduced (black) versus GSSH pretreated (red) and (B) reduced (black) versus H₂O₂ pretreated (blue). Anisotropy changes of the fluorescein-labeled HlyO operator with VcHlyU after addition of a 10-fold excess of either (C) GSSH or (D) H₂O₂. After addition of oxidant the anisotropy was followed over time until a new equilibrium condition was reached. Then a final concentration of 5 mM TCEP was added to the solution to test the reversibility of the oxidation. DNA-binding isotherms were obtained using the DynaFit software after a global fit of at least two replicates in each case. TCEP, tris(2-carboxyethyl)phosphine.

Figure 4

HlyU mediated phlyA activation followed by quantitative RT-PCR.A, model of the mechanism of HlyU regulation of the hlyA gene, HlyU DNA dissociation leads to H-NS (green)–mediated repression. B, quantitative RT-PCR performed over a ΔCTX Δfur ΔhapR Vibrio cholerae strain with the addition of Na₂S or H₂O₂. The bar chart shows the fold changes of induction of Vc hlyA after addition of Na₂S (red) and H₂O₂ (blue), with transcript values normalized relative to the transcription level of recA. The values correspond to transcript levels relative to unstressed (UN) (middle sand bar) and are shown as mean ± SD from three replicate cultures. Statistical significance was established using a paired t test relative to UN under the same conditions (∗∗p < 0.01, ∗p < 0.05). Lines on the top of the chart show statistical significance relative to ΔCTX Δfur ΔhapR V. cholerae mutant strain UN.

Figure 5

LMWT and LMW persulfide metabolite profiling of Vibrio cholerae strains. A, cartoon representation of the scheme for LMWT and LMW persulfide profiling. Growth of a ΔCTX Vibrio cholerae strain until A of ∼0.2 is followed by the addition of Na₂S or H₂O₂ to a final concentration of 0.2 mM. Cultures were centrifuged at 0 (prior to addition of the stressor), 15, 30, and 60 min. In all cases 1 ml of sample was withdrawn for protein quantitation. The metabolite profiling was generally carried out using HPE-IAM as labeling agent. The ratiometric LC-ESI-MS experiments were performed with the dilution of isotopically labeled internal standards of known concentration, which were used for quantitation of the organic and inorganic species. B, endogenous concentrations of hydrogen sulfide before and after addition of stress (Na₂S, red; H₂O₂, blue) to midlog-phase cultures. C, endogenous concentrations of hydrogen disulfide before and after addition of stress (Na₂S, red; H₂O₂, blue) to midlog-phase cultures. D, endogenous concentrations of CysSSH, GSSH, and h-CysSSH before and after addition of stress at different timepoints to midlog-phase cultures. Statistical significance was established using a paired t test relative to UN under the same conditions (∗∗p < 0.01, ∗p < 0.05). HPE, β-hydroxyphenyl-ethyl; IAM, iodoacetamide.