A new role for Zinc limitation in bacterial pathogenicity: modulation of α-hemolysin from uropathogenic Escherichia coli

Metal limitation is a common situation during infection and can have profound effects on the pathogen’s success. In this report, we examine the role of zinc limitation in the expression of a virulence factor in uropathogenic Escherichia coli. The pyelonephritis isolate J96 carries two hlyCABD operons that encode the RTX toxin α-hemolysin. While the coding regions of both operons are largely conserved, the upstream sequences, including the promoters, are unrelated. We show here that the two hlyCABD operons are differently regulated. The hlyII operon is efficiently silenced in the presence of zinc and highly expressed when zinc is limited. In contrast, the hlyI operon does not respond to zinc limitation. Genetic studies reveal that zinc-responsive regulation of the hlyII operon is controlled by the Zur metalloregulatory protein. A Zur binding site was identified in the promoter sequence of the hlyII operon, and we observe direct binding of Zur to this promoter region. Moreover, we find that Zur regulation of the hlyII operon modulates the ability of E. coli J96 to induce a cytotoxic response in host cell lines in culture. Our report constitutes the first description of the involvement of the zinc-sensing protein Zur in directly modulating the expression of a virulence factor in bacteria.

During infection, pathogens often encounter metal limitation. Alterations in the concentrations of certain metal ions have a great impact in cell physiology and gene expression. For instance, changes in the concentration of iron regulate the expression of virulence factors through metalloregulatory proteins such as Fur, the ferric uptake regulator, and the closely related PerR, the peroxide-stress regulator 1,2 . Recent studies revealed that another metal ion, zinc, also has an effect on bacterial physiology during infection 3,4 . Zinc is an essential element as a catalytic or structural cofactor of key enzymes and proteins involved in many processes such as DNA replication and protein synthesis and turnover 5,6 . Conversely, zinc can be toxic to bacteria when present in excess and so intracellular levels must be tightly controlled 5,7 . Zinc release from host tissue has been proposed to be an important innate defense mechanism 7 . In the case of S. pneumoniae infection in mice, elevated total zinc concentrations have been reported in host tissue and serum samples. The authors propose that the ability of zinc excess to inhibit pathogen growth arises from zinc competition with Mn uptake 7 . Zinc concentration in host fluids can rise in response to bacterial infection and inflammation by its release from damaged or apoptotic cells, and from sequestering proteins such as metallothionein 7 .
Studies in both E. coli and B. subtilis have shown that, in response to variations in zinc concentration, cells regulate gene expression using the metalloregulatory protein Zur (zinc uptake regulator) as a sensor. Zur forms a dimeric complex with zinc acquiring the ability to bind DNA and to repress gene expression 8 . Characterized Zur regulons mainly consist of genes coding for proteins involved in zinc homeostasis and paralogs of ribosomal proteins which, under zinc limitation conditions, replace ribosomal proteins that contain zinc 9 . Although most known Zur-regulated genes are involved in zinc uptake and adaptation to low concentrations of the metal 8-10 , a number of questions about how Zur impacts overall cell physiology and pathogenesis remain unanswered.
Uropathogenic Escherichia coli (UPEC) is the main causal agent of community-acquired urinary tract infections. In order to colonize its host, it expresses a battery of virulence factors including adhesins, siderophores, flagella, and toxins 11 . One of these factors is the pore-forming toxin α-hemolysin, whose expression induces tissue damage and more severe infections 12 . Approximately 40-50% of UPEC isolates are able to produce α-hemolysin 13 . Remarkably, a recent study in mice linked zinc concentrations to the pathogenic effect of α-hemolysin during gastrointestinal disorders, showing that treatment with zinc protects against α-hemolysin-induced intestinal barrier disfunction 14 .
α-hemolysin is encoded in the polycistronic operon hlyCABD, which contains four open reading frames (ORFs) 15 . The hlyC gene codes for an acyltransferase required for the activation of pro-α-hemolysin. Pro-α-hemolysin is encoded by hlyA, while the hlyB and hlyD genes encode two components of the secretory system that transports α-hemolysin out of the cells 16,17 . The hlyCABD operon in E. coli is tightly regulated and strongly affected by several environmental and physiological signals. A number of regulators, including RfaH, CpxRA and H-NS, modulate α-hemolysin expression [18][19][20][21] . In hemolytic UPEC isolates from humans, the hly-CABD operon is present in the chromosome within pathogenicity islands, together with genes coding for other virulence factors 22,23 . Although most hemolytic UPEC strains carry a single copy of the operon, the presence of two functional copies of the hlyCABD operon has been detected in two pyelonephritis isolates, J96 and 536 23 .
In this article, we show that the two hemolytic operons present in the chromosome of E. coli J96 are differentially regulated. The hly I operon does not respond to zinc limiting conditions whereas the hly II operon is silenced in the presence of zinc in a Zur-dependent manner. A highly conserved Zur binding site was found in the promoter sequence of the hly II operon. Direct binding of Zur to this promoter was confirmed by electrophoretic mobility shift assay (EMSA). Although a role of Zur-mediated regulation in bacterial pathogenicity has previously been suggested 6,24 , our research represents the first description of the involvement of zinc and the zinc-sensing transcription factor Zur in regulating the expression of a virulence factor in bacteria directly.

Results
The two hemolytic operons of J96 are differentially expressed. Two UPEC isolates, J96 and 536, extensively used as model organisms in pathogenicity studies, have two functional hlyCABD operons (designated I and II) coding for α-hemolysin (HlyA) 23 . The full sequences of the two operons from 536 and the hlyCABD II operon (hly II ) from J96 are known; however, only a fragment of the hlyCABD I operon (hly I ) from J96, carrying part of the hlyC gene (224 bp) and its upstream sequence, is available. While the ORFs of the sequenced operons are almost identical, the hlyC upstream sequences carrying the promoter and the regulatory motifs diverge extensively between operons (Fig. 1a). Consistent with a previous report 23 , pairwise alignments showed two different types of sequence upstream of hlyC and indicated that the hly II-J96 operon is identical to the hly I-536 operon and that hly I-J96 is identical to hly II-536 . The hly II-J96 operon was the first hlyCABD operon cloned from human isolates and used to substantiate the contribution of HlyA to UPEC virulence 25,26 .
J96 derivative mutant strains carrying partial deletions of either hly I or hly II were obtained by gene replacement. The hemolytic phenotype of the mutant derivatives was monitored on Columbia Blood Agar (CBA) plates (Fig. 1b). The mutant J96ΔII (Δhly II ), carrying only the hly I operon, displays a phenotype identical to the J96 strain. However, the J96ΔI (Δhly I ), carrying only the hly II operon, exhibits a very thin hemolytic halo, suggesting that very little HlyA is produced from hly II operon under optimal conditions for HlyA detection 27 . The levels of secreted HlyA by J96 and its derivatives J96ΔI and J96ΔII were determined in bacterial cultures grown in LB 0 (with no NaCl added) at 37 °C up to late-logarithmic phase. SDS-PAGE separation of secreted protein extracts from J96 revealed a single band at 110 kDa, corresponding to HlyA. Consistent with the hemolytic phenotype, the amount of HlyA secreted by the J96ΔII strain (hly I operon) was very similar to the amount detected in the extract from the wild-type strain, whereas the amount of HlyA secreted by the J96ΔI strain (hly II operon) was much lower (Fig. 1c). These results suggest that in the assessed experimental conditions, the hly I operon is the main contributor to the HlyA production and the hly II operon is somehow silenced. The extracts analyzed in Fig. 1c were obtained from cultures grown in LB 0, however, the described differential expression of the two hly operons of J96 does not depend on the external osmolarity. Secreted protein extracts from cultures grown in LB (5 g l −1 of NaCl) were also analyzed. Although expression of HlyA was lower in all the strains, hly I expressed most of the HlyA, while expression of HlyA from the hly II operon was barely detected (Fig. S1).
A mutation in zur, encoding the zinc uptake regulator, causes derepression of the hly II operon. To identify putative regulators of the hly II operon responsible for its low expression under the assessed conditions, random mutagenesis using a gentamycin resistance mariner transposon was performed on strain JFV3, a J96 derivative that carries a chromosomal hly II ::lacZ reporter fusion. Mutants were selected according to their Lac phenotype on LB agar plates with X-gal, to detect hly II ::lacZ expression. The clone #13 was selected for its deep blue colour compared to the parental strain, which indicates increased expression of hly II ::lacZ (Fig. 2a). Quantitative expression studies showed that the selected Gm R mutant causes a drastic derepression (47-fold) of the hly II operon transcriptional expression (Fig. 2b). Using primers for the mariner transposon, the nucleotide sequence adjacent to the insertion was determined in clone #13. The transposon was inserted within the zur gene, in codon 142 of 172.
The zur gene encodes the Zur protein (zinc uptake regulator), a transcriptional repressor that binds to DNA in the presence of zinc and thereby represses specific genes 28 . To corroborate the association of Zur with the observed derepression, a new zur mutant was generated in J96 by gene replacement. The sequence of the zur gene between positions 43 and 381 bp (position 1 corresponds to the first nucleotide of the first codon) was deleted, and a Cm R cassette was inserted. The zur::Cm R allele causes the same effect on hly II transcriptional expression as the Gm R allele obtained by random mutagenesis both in late-logarithmic phase (Fig. 2b) and in mid-logarithmic phase (Fig. S2) of bacterial growth. The effect of the zur mutation on hly II expression was also determined by measuring the amount of the secreted HlyA in culture supernatants of Zur proficient and deficient derivatives of J96, J96ΔI and J96ΔII strains. Both Coomassie-staining and immunodetection using specific HlyA antibodies showed a clear increase in the amount of HlyA in the presence of the Δzur mutation in the two strains carrying the hly II operon (J96 and J96ΔI) (Fig. 2c). By contrast, derepression of the hly I operon (J96ΔII strain) was not detected in the zur mutant ( Fig. 2c and Fig. S1). Due to the low expression level of the hly I operon in LB, HlyA immunodetection in J96ΔII extracts was performed on 20-fold concentrated samples. Furthermore, the severity of the hemolytic phenotype correlates with the amount of HlyA in each strain. The strain carrying only the operon hly II showed minimal hemolysis in the presence of an intact zur gene and a higher hemolytic phenotype when the zur gene was mutated. Strains carrying only the hly I operon showed the same phenotype, regardless of the presence or absence of an intact zur gene (Fig. 2d). These results clearly indicate that Zur specifically regulates the hly II operon. Zur-mediated regulation of hly II transcription was confirmed by trans-complementation experiments (Fig. 2b). The strong derepression of the hly II operon observed in the zur mutant was fully complemented when a pBR322 derived plasmid containing the J96 zur gene with its own promoter was introduced in the mutant strain. Our findings on the Zur-mediated regulation of the hly II operon of J96 are the first description of a regulatory factor that modulates differently the two hemolytic operons present in J96.
Hemolysin has a role in cell toxicity and urothelial damage during infection 19,29 . In order to determine the impact of Zur-mediated regulation of α-hemolysin on bacterial pathogenicity, we assessed the ability of different derivatives of the J96 strain to induce hemoglobin release after infection of blood cell suspensions and cell detachment after infection of bladder epithelial cell monolayers. A slight increase in hemolytic activity was detected after infection of the Δzur derivatives of J96 and J96ΔII as compared to the zur + counterparts (Fig. 3A). Remarkably, the zur mutation caused a very drastic effect when blood cell suspensions were infected with derivatives of the J96ΔI strain. Hemolytic activity was more than 12-fold higher in the Δzur mutant as compared to wild-type strain, consistent with the described derepression of the hly II operon in strains lacking Zur. As a control, J96 derivative strains lacking both hly I and hly II operons showed very low hemolysis in either zur + or Δzur background. These results clearly demonstrate that the effect observed was caused by α-hemolysin.
Next, we performed infection assays of bladder epithelial cells (T24 monolayers) using the same bacterial strains (Fig. 3B). After 3.5 hours of infection with the J96ΔI strain, the monolayers were intact, showing a percentage of attached cells similar to the uninfected monolayers. In contrast, when monolayers were infected with the Δzur derivative of J96ΔI less than 20% of the cells remained attached after 3.5 hours. Again, the effect can be attributed to the production of α-hemolysin since cell attachment did not decrease after infection with J96ΔIΔIIΔzur. The specific effect of α-hemolysin production on cell monolayers could be detected 2 hours post-infection as cells became round-shaped when infected with strain J96ΔIΔzur (Fig. 3C), an α-hemolysin-induced phenotype previously described 30 . Both wild-type and Δzur derivatives of the J96 and J96ΔII strains, carrying the Zur-independent hly I operon, disrupt cell adhesion at different extents ( Fig. 3B). In line with the aforementioned expression studies on the hly II operon, the Δzur mutation triggers virulence-associated features such as hemoglobin release and disruption of epithelial cell adhesion.

Expression of the hly II operon responds to the presence of zinc.
In the presence of zinc, the transcriptional regulator Zur forms a complex that binds to DNA, causing transcriptional repression 8 . We hypothesized that the hly II operon should be repressed by the presence of zinc in a Zur-dependent manner. To test this hypothesis, transcriptional expression of the hly II operon was monitored in cultures of strain JFV3 grown in either LB depleted from zinc (M-LB) or the same medium replenished with 1 μM and 10 μM ZnCl 2 . In the zur + strain, hly II transcriptional expression is high in the absence of zinc and severely drops in medium replenished with this metal ion (Fig. 4a). Remarkably, the response of hly II expression to zinc was dose-dependent. Furthermore, zinc-mediated regulation is strictly dependent on the presence of Zur, since no repression is observed in a zur mutant strain by the presence of zinc. The growth kinetics of the JFV3 strain was not altered by the absence or presence of different zinc concentrations (Fig. 4b). The effect of zinc on hly II transcriptional expression was corroborated by monitoring the level of secreted HlyA. In cultures of the strain J96ΔI, a high amount of secreted HlyA was detected in the absence of zinc, whereas the HlyA amount was very low in cultures grown in the presence of ZnCl 2 (10 μM) (Fig. 4c). The amount of secreted HlyA from hly I operon was also monitored in both the absence and presence of zinc. Consistent with the fact that hly I operon expression was not altered by the zur mutation, the expression of hly I operon was not affected by the presence of zinc. In cultures of the J96 strain, carrying both hly I and hly II operons, a slight effect on HlyA production by the presence of zinc was detected.
Zur directly represses hly II expression by binding to its promoter. The effect of sudden addition of ZnCl 2 (10 μM) on hly II transcriptional expression was monitored in cultures of strains JFV3 and its ∆zur derivative grown in M-LB medium up to an OD 600 nm of 0.3. Bacterial growth was not substantially affected by adding ZnCl 2 (10 μM) to the zinc-free medium cultures (Fig. 5a). The ZnCl 2 exposure caused a rapid decrease of the hly II transcriptional expression in the Zur proficient strain, as compared to cultures where no ZnCl 2 was added (Fig. 5b). After 15 minutes, a drop in hly II expression was already detected, falling to 53% of the levels measured in cultures grown with no ZnCl 2 addition. Expression progressively declined to 20% after one hour exposition to ZnCl 2 . In agreement with all previous results, exposure to ZnCl 2 during one hour did not significantly alter the expression of the hly II operon in a zur mutant strain. As a control, the level of hly II operon expression was very low in cultures grown in M-LB replenished with ZnCl 2 from the start.  The fast response to ZnCl 2 of hly II transcriptional expression suggested that Zur directly represses the hly II promoter. Zur regulates transcription by cooperative binding of two dimers to a well-characterized DNA sequence located in the vicinity of the promoter of Zur-regulated genes 9,31 . Using the consensus sequence reported 31 for the E. coli Zur box (RNNNYRNNRYNNYRNNNY), the J96 hly II promoter sequence was scrutinized, revealing a putative Zur binding site ( Fig. 6a and b). The bases of the Zur box involved in establishing hydrogen-bonds with Zur have been identified 31 and are indicated in Fig. 6b. The motif found in the J96 hly II promoter sequence was highly conserved except for the base G19′ (see Fig. S3). It must be noted that G19′ is one of the least determinant bases for Zur binding 31 .  A transcriptional start for the hly II operon was previously proposed based on primer extension analyses on RNA samples from K12 strains carrying the hly II operon in a multicopy plasmid 32 . In our study, 5′-RACE with RNA samples from J96 was performed and a hly II operon transcription start was defined 9 nucleotides upstream of the previously described +1 position. This transcriptional start site is properly positioned to a putative promoter identified with the Neural Network Promoter Prediction software (Fig. 6a and c). Consequently, the putative Zur binding site is located at positions −25 to −6. Zur binding would overlap the −10 sequence, blocking the binding of RNA polymerase and repressing hly II transcription. The relative position of the putative Zur binding site found in the hly II operon and other genes of the E. coli Zur regulon was compared after searching the Zur consensus sequence described 31 within the upstream regions of Zur-regulated genes 9 . The relative location of the putative Zur binding site in the hly II operon turned out to resemble the position of other members of the Zur regulon (Fig. 6b). For instance, the location of the Zur box in znuC and zinT is −22 to −3 and −23 to −4, respectively.
The ability of Zur to bind to the hly II promoter sequences in vitro was assessed using EMSA. The Zur protein used in this assay is derived from the E. coli strain MG1655, which shares 97.7% identity and 100% similarity with the Zur protein from J96. The results clearly demonstrate that Zur binds the hly II promoter in vitro (Fig. 6d).
Remarkably, E. coli Zur binds to the hly II operon even more tightly than it does to the E. coli MG1655 Zur promoter region for the L31p promoter (0.025 ± 0.01 × 10 −18 M 2 ), which corresponds to the highest affinity of any of the operator sites evaluated 31 . While the binding is too tight to calculate a reliable dissociation constant under these experimental conditions, the Zur concentration at half maximal binding (100 nM Zur dimer) allow us to estimate an upper limit for this dissociation constant, i.e. <0.02 × 10 −18 M 2 . Together with the in vivo studies on gene expression, these data reveal that Zur likely represses transcription of the hly II operon directly and confirm that the hly II operon is part of the Zur regulon in the UPEC strain J96. Interestingly, the putative Zur binding site was also found in the hly I operon from the UPEC strain 536 that is equivalent to hly II from J96. In 536 two mismatches from the consensus sequence were found: the mismatch already found in J96 hly II and a second base which does not directly interact with Zur (in position 13, a G instead of A or T is found). The upstream intergenic sequence of the hly I operon of J96 was also analyzed for a Zur binding site and, as expected, none was found. No Zur binding sites were found in upstream intergenic regions of hly operons from other UPEC isolates such as CFT073, UTI89 and F11. Pairwise alignments showed that these operons and their promoters are J96 hly I -like. Accordingly, the Δzur mutation in CFT073 does not cause an increase of HlyA production and hemolysis after infection of blood cells (Fig. S4).

Discussion
The presence of two hlyCABD operons in the UPEC isolates J96 and 536, with more than 7 Kb DNA sequence sharing 99% identity, is intriguing considering that the presence of large homologous DNA sequences in a chromosome may cause genome instability and result in detrimental effects on bacterial fitness. Interestingly, the sequences upstream of the hlyC gene from both operons differ greatly, indicating that their regulatory sequences are unrelated and suggesting the existence of different mechanisms of regulation. Indeed, in this report we demonstrate that the hly II operon is silenced by the Zur regulator allowing expression only under zinc limiting conditions. In contrast, the operon hly I does not respond either to zinc or this regulator (Zur). Interestingly, most hlyCABD operons from human isolates described in databases showed homology to the Zur-insensitive operon hly I , whereas the hly II operon was only detected in two pyelonephritis isolates, 536 and J96, and in the environmental isolate MRE600. Recent genomic and phylogenetic studies revealed that MRE600 is an E. coli strain with distinctive genomic properties and more similar to Shigella than to other E. coli strains 33 .
One may hypothesize that expressing hemolysin under a wide variety of environmental conditions, including zinc limitation, may be beneficial to ensure success during the process of host colonization. In this scenario, the versatility provided by having two operons that respond to different conditions may constitute an adaptive advantage that outweighs the detrimental effect in fitness caused by genome instability due to the coexistence of large and highly homologous DNA sequences.
Previous studies suggest that zinc may play a relevant role during host colonization since zinc uptake systems are important for fitness in different pathogens, including UPEC 6,34 . However, the exact role of the Zur protein in the control of virulence factors has not yet been elucidated 35 . In S. aureus, Zur does not have an apparent role in pathogenicity, as concluded from assays using an in vivo infection model 36 . In M. tuberculosis and S. suis, Zur regulates genes coding for proteins belonging to the early secretory antigen target and metalloproteases, respectively, although the role of those proteins in virulence is unclear 37,38 . In A. baumannii, the zur mutant exhibits a defect in dissemination in a pneumonia mouse model, and RNA-seq assays showed that several genes encoding secreted proteins and putative virulence factors were upregulated in the mutant strain 24 . Our study is the first description of a direct Zur-mediated regulation of the expression of a virulence factor, UPEC's α-hemolysin. More precisely, Zur represses the transcriptional expression of the hly II operon of J96, which consequently is only expressed under zinc limiting conditions. This finding raises thought-provoking questions. Is α-hemolysin expression a mechanism used by the bacterium to release or mobilize zinc from cells? If so, this may facilitate growth of bacterial pathogens in spite of host-induced zinc limitation. In this scenario, hly II expression would be acting as a bacterial factor involved in zinc homeostasis. Is the response of the J96 hly II operon to zinc, and therefore the acquisition of the Zur box, required for proper sensing of specific environments? If so, the presence of hly II may facilitate efficient colonization of sites within the host.
As exemplified by iron, metal limitation is a common situation during host-pathogen interactions 1 . Zinc concentration in most human tissues is extremely low 39 . Within the cytosol of human cells available zinc is particularly scarce, due to its sequestration by metallothioneins 40 . Furthermore, during infection phagocytes and epithelial cells express other zinc-chelating proteins, such as calprotectin [41][42][43] . In the case of UPEC, which invades the urinary tract, it inevitably faces zinc-limiting conditions since zinc concentration in human urine was reported to range between 0.17 and 0.5 ppm (2.6 to 7.6 µM) in healthy subjects, being one of the human corporal fluids with lowest zinc concentration reported 39,44,45 . Furthermore, UPEC probably encounters zinc starvation during intracellular growth, where bioavailable zinc is limited due to sequestration by host proteins like metallothioneins.
On the other hand, zinc concentration in the intestine, the reservoir of UPEC, is considerably high, due to the predominant fecal excretion of this metal 46 . One may speculate that during intestinal transit hly II expression would be repressed by Zur, and therefore very little HlyA would be produced from this operon. Interestingly, during acute gut inflammation, calprotectin released by neutrophils sequesters zinc and, during this zinc-limiting conditions, Salmonella enterica Typhimurium grows better than gut commensal bacteria due to increased expression of the Zur-regulated ZnuABC transporter 3 . Moreover, anomalously high expression of hlyA by the UPEC isolate 536 in the intestine has been detected during active ulcerative colitis, which produces acute inflammation 47 . Therefore, it is reasonable to hypothesize that intestinal zinc limitation due to inflammation could promote the production of HlyA by UPEC strains residing in the gut and carrying hemolytic operons responsive to zinc, thus producing further inflammatory lesions in ulcerative colitis. A recent report described zinc treatment to be efficient against α-hemolysin-induced intestinal leakage in mice colonized by the pyelonephritis isolate 536 14 . It is known that zinc is an important factor for intestinal barrier integrity 48 . Although the authors showed that zinc has a protective effect on α-hemolysin-mediated damage per se, our data indicate that zinc may have an additional protective effect by promoting downregulation of α-hemolysin expression from the 536 strain, which contains a homolog of the Zur-regulated operon here reported. Our study describes the direct involvement of Zur in the control of virulence expression and suggests that zinc limitation may play a pivotal signaling role during the infectious process.

Methods
Bacterial strains, plasmids and growth conditions. Bacterial strains used in this work, shown in Table 1, are all derived from the pathogenic isolates J96 and CFT073. All strains were grown routinely in LB (Lennox broth, tryptone 10 g l −1 , yeast extract 5 g l −1 and NaCl 5 g l −1 ) at 37 °C with aeration. LB without NaCl and with 10 g l −1 of NaCl was used and denoted LB 0 and LB 10 , respectively. In most experiments, bacterial cultures were grown up to an OD 600 nm of 1.0, denoted late-log phase of growth. The zinc-free medium (M-LB) was prepared as indicated 49 . Briefly, LB medium was treated with 50 mg l −1 of chelating resin Chelex 100 (Bio-Rad) one hour at room temperature while stirring. After removing the resin by filtration (0.2 μm), trace elements were replenished by adding 1 μM FeCl 3 , 1 μM CaCl 2 , 20 μM MgCl 2 and 1 ml l −1 SL-7 trace element solution (without zinc). Finally, 20 μM TPEN (N,N,N′,N′-Tetrakis(2-pyridylmethyl)ethylenediamine, Sigma-Aldrich) was added to further chelate zinc. Replenishment solutions were prepared out of concentrated stock solutions and sterilized by filtration using 0.2 μm filters when necessary. SL-7 trace element solution without zinc has the following composition: 10 ml 25% HCl, 1. O and deionized water to a final volume of 1 L. In order to minimize contamination by zinc, all cultures with M-LB were performed in 5 ml culture medium using 50 ml conic centrifuge plastic tubes. Any other used material was also made of plastic and handled so as to avoid metal contamination. To monitor hemolytic phenotype Columbia Blood Agar plates (Scharlau) were used.
Genetic techniques. The primers used in this work are listed in Table S1 (supplementary material). The strain J96Δlac is a ΔlacZ derivative of J96 strain obtained as previously described 50 . The strain JFV16 (J96ΔII), carrying a hlyA deletion from codon 20 to 434 was constructed using the primers HlyA-P1 and HlyA-P2 and the method described by Datsenko  to the codon 422 of hlyA was constructed using the same methodology and the primers Hly1-P1 and Hly1-P2-2. Strain EV64 (J96ΔIΔII) was obtained using the same methodology and the primers Hly1-P1 and Hly1-P2-2 on the JFV16 strain. Similarly, zur mutants were generated using primers ZurP1 and ZurP2.2 in J96 and ZurP1.2 and ZurP2.1 in CFT073. The strain JFV3 carries a promoter less lacZ gene linked to a kanamycin resistance cassette inserted in the hlyA gene of the hly II operon. The lacZ-Km R cassette from plasmid pKG137 was inserted into the hlyA gene of JFV16 following the method described previously 52 . The zur gene of J96 together with the intergenic region located upstream of the zur ORF was PCR-amplified using the primers Zur-BamHI and Zur-EcoRI. The PCR-amplified fragment was cloned in pBR322 resulting the plasmid pBRzur.

SDS-PAGE and Western immunoblotting analyses.
The standard SDS-PAGE procedure was used to monitor HlyA production. Gels were routinely stained with Coomassie blue. For immunoblotting, proteins were transferred to PVDF membranes and detected with the monoclonal anti-α-hemolysin H10 53 and a horseradish peroxidase-conjugate antibody (Promega) using the ECL Plus Western Blotting Detection System (GE Healthcare). Gels were analyzed on a Chemidoc System (BioRad) equipped with the QuantityOne ® Software.
β-Galactosidase assay. β-Galactosidase activity was monitored by the method described earlier 54 . Data shown are mean values and standard deviation of duplicate determinations from three independent experiments.
Assay of hemolytic activity. Quantitative hemolytic assay was performed essentially as previously described 55 . Briefly, a 10% defibrinated sheep blood suspension was prepared in 0.9% NaCl containing 10 mM CaCl 2 , centrifuged and resuspended in the same volume of the same solution. The wash step was repeated three times in order to eliminate debris from broken cells. Bacterial cultures were grown in LB 10 at 37 °C to an OD 600 nm of 1.0, centrifuged to discard supernatant and the bacterial pellet resuspended in the same volume of LB 10 . In 96-well microtiter plates 50 µl of blood suspension was mixed with an equal amount of a 1/125 dilution of the bacterial cell suspension, centrifuged (400 g, 10 min, 4 °C) and incubated for 60 min at 37 °C. Thereafter, 100 µl of ice-cold 0.9% NaCl was added and the microtiter plate was centrifuged (400 g, 10 min, 4 °C). A 100 µl aliquot of the supernatant was removed to another plate and the release of haemoglobin measured spectrophotometrically at 545 nm. In all plates non-infected wells were used as controls and the resulting OD 545 nm from those controls was subtracted to the value obtained for the different samples. A bar shows the arithmetic mean of experimental results and the error bar indicates the standard deviation from three independent experiments. Control experiments using blood suspension prepared in 0.9% NaCl, in the absence of CaCl 2 , were performed. Consistent with the fact that the activity of hemolysin is calcium-dependent no hemoglobin release was detected. LB 10 was used since regular LB (5 g/L NaCl) causes unspecific release of haemoglobin due to osmotic pressure alterations.
T24 cell culture assays. T24 bladder epithelial cells were maintained in RPMI supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 1% penicillin-streptomycin. Bacterial cultures were grown in LB at 37 °C. Infection assays were performed in 24-well plates containing RPMI-10% FBS where 10 µL of bacterial growth (OD 600 nm of 0.5) were added to a monolayer of T24 cells. After infection, the cells were incubated at 37 °C, 5% CO2 for 2 h for microscopy analysis and for 3.5 h for measurements of cell detachment. The T24 monolayers were visualized after 2.5 h with an Axiovert 40 C inverted optical microscope (Carl Zeiss) and images were captured with EOS 1000D Canon camera at 25X increase. After 3.5 h infection, the cells were washed three times with ice-cold PBS, fixed in 10% formalin for 15 min at room temperature and stained in 0.1% Crystal Violet, 0.2% ethanol for 10 min at room temperature. After 2 washes with water, the cells were allowed to dry at room temperature and then lysed in 2% sodium dodecyl sulphate (SDS) for 30 min. The absorbance was quantified by spectrophotometry at 590 nm as an indicator of crystal violet staining. The data is given as percentage of attached cells. For each experiment the OD 590 nm of non-infected samples was given arbitrarily the value 100%. A bar shows the arithmetic mean of experimental results and the error bar indicates the standard deviation from three biological replicates.
Random mutagenesis with pBT20 plasmid. Mutagenesis experiments were performed using the mariner-based transposon system carried on plasmid pBT20 56 . Bacterial suspensions of the donor (E. coli S17-λpir containing pBT20) and the recipient (JFV3) were recovered from overnight plates and resuspended in LB. These suspensions were adjusted to OD 600 nm of 40 and 20, respectively, and mixed in a 1:1 ratio. Aliquots of 50 μl were spotted on a LB agar plate and incubated at 37 °C for 6 h. The mating mixtures were subsequently recovered in LB and cultured on LB agar plates containing kanamycin, gentamicin, and X-gal at 37 °C. Clones depicting darker blue color were selected, and hly II transcriptional expression was monitored by β-galactosidase quantification. Clone #13 genomic DNA was extracted using the Blood & Cell Culture DNA Midi kit (Qiagen) with the Genome DNA Buffer Set (Qiagen) and sequenced using the primer 526.
5′RACE assay. The 5′ end of the hly II operon transcript was determined using the 5′ RACE System for Rapid Amplification of cDNA Ends (Invitrogen). Total RNA was isolated from cultures of the J96 strain grown in LB at 37 °C up to an OD 600 nm of 1.0. Retrotranscription was done using the oligo hlyIIBamIII, specific for the hly II operon. Next, a polyC tail was added to the 3′ end of the cDNA by using terminal deoxynucleotidyl transferase. Two rounds of PCR amplification were performed using specific primers for the hly II operon, hlyII-GSP2 and hlyII-GSP3, and primers provided by the commercial kit. The final PCR product was isolated and sequenced.
Electrophoretic Mobility Shift Assay. Zur binding to the Zur box of the J96 hly II operon was assessed as previously reported 31 . A fluorescence-labeled 50-bp fragment of the hly II promoter containing the putative Zur binding site was prepared by annealing two single strand DNA oligos. An oligo with a 5′ end Cy5 label (5′-/5Cy5/ TCTTTATTTATTGTTTGTGTTATGTTATAACATATAAAAGAGTATTGTTT-3′) and its reversed complement