Ferric Uptake Regulator Fur Coordinates Siderophore Production and Defense against Iron Toxicity and Oxidative Stress and Contributes to Virulence in Chromobacterium violaceum

Maintenance of iron homeostasis, i.e., avoiding both deficiency and toxicity of this metal, is vital to bacteria and their hosts. Iron sequestration by host proteins is a crucial strategy to combat bacterial infections. In bacteria, the ferric uptake regulator Fur coordinates the expression of several iron-related genes. Sometimes, Fur can also regulate several other processes. In this work, we performed an in-depth phenotypic characterization of fur mutants in the human opportunistic pathogen Chromobacterium violaceum. We determined that fur is a conditionally essential gene necessary for proper growth under regular conditions and is fully required for survival under iron and oxidative stresses. Fur also controlled several virulence-associated traits, such as swimming motility, biofilm formation, and siderophore production. Consistent with these results, a C. violaceum fur null mutant showed attenuation of virulence. Therefore, our data established Fur as a major player required for C. violaceum to manage iron, including during infection in the host.

ABSTRACT

Iron is a highly reactive metal that participates in several processes in prokaryotic and eukaryotic cells. Hosts and pathogens compete for iron in the context of infection. Chromobacterium violaceum, an environmental Gram-negative bacterial pathogen, relies on siderophores to overcome iron limitation in the host. In this work, we studied the role of the ferric uptake regulator Fur in the physiology and virulence of C. violaceum. A Δfur mutant strain showed decreased growth and fitness under regular in vitro growth conditions and presented high sensitivity to iron and oxidative stresses. Furthermore, the absence of fur caused derepression of siderophore production and reduction in swimming motility and biofilm formation. Consistent with these results, the C. violaceum Δfur mutant was highly attenuated for virulence and liver colonization in mice. In contrast, a manganese-selected spontaneous fur mutant showed only siderophore overproduction and sensitivity to oxidative stress, indicating that Fur remained partially functional in this strain. We found that mutations in genes related to siderophore biosynthesis and a putative CRISPR-Cas locus rescued the Δfur mutant growth defects, indicating that multiple Fur-regulated processes contribute to maintaining bacterial cell fitness. Overall, our data indicated that Fur is conditionally essential in C. violaceum mainly by protecting cells from iron overload and oxidative damage. The requirement of Fur for virulence highlights the importance of iron in the pathogenesis of C. violaceum.

IMPORTANCE Maintenance of iron homeostasis, i.e., avoiding both deficiency and toxicity of this metal, is vital to bacteria and their hosts. Iron sequestration by host proteins is a crucial strategy to combat bacterial infections. In bacteria, the ferric uptake regulator Fur coordinates the expression of several iron-related genes. Sometimes, Fur can also regulate several other processes. In this work, we performed an in-depth phenotypic characterization of fur mutants in the human opportunistic pathogen Chromobacterium violaceum. We determined that fur is a conditionally essential gene necessary for proper growth under regular conditions and is fully required for survival under iron and oxidative stresses. Fur also controlled several virulence-associated traits, such as swimming motility, biofilm formation, and siderophore production. Consistent with these results, a C. violaceum fur null mutant showed attenuation of virulence. Therefore, our data established Fur as a major player required for C. violaceum to manage iron, including during infection in the host.

INTRODUCTION

Iron is a redox-active metal with a central role in biology. It is a cofactor of several enzymes involved in key biochemical processes, such as respiration, nitrogen fixation, DNA biosynthesis, and gene regulation (1, 2). Given that iron is an essential micronutrient and has low solubility, organisms have evolved strategies to acquire and store this metal for the supply of their physiological demands (1, 3, 4). However, it is also of vital importance that iron is not in excess inside cells because it can react through the Fenton reaction with cell metabolites and generate reactive oxygen species (ROS), which cause damage to all cellular macromolecules (5, 6). Thus, both prokaryotic and eukaryotic cells have evolved mechanisms to maintain nondamaging homeostatic iron levels (1, 3, 4, 7).

In bacteria, the ferric uptake regulator (Fur) family of transcription factors exerts a major role in controlling metal homeostasis. These regulators sense intracellular levels of metals such as iron (Fur), zinc (Zur), manganese (Mur), and nickel (Nur) and regulate the expression of proteins related to the acquisition or efflux of these metals (8, 9). Concerning iron homeostasis, the Fur regulator mostly represses the expression of genes related to iron acquisition when the intracellular levels of this metal are high (8–10), although Fur can also act as an activator in some cases (11–13). Determination of the Fur regulon in several bacteria has identified some classically Fur-regulated genes, such as genes encoding proteins for siderophore biosynthesis, uptake of iron-siderophore complexes (TonB-dependent receptors [TBDRs]), and several other iron-related and nonrelated processes (12–15). The role of Fur as a global regulator fits with the findings that fur mutants have pleiotropic phenotypes (sometimes nonrelated to iron), including changes in biofilm and motility, sensitivity to acid stress, and virulence attenuation in some bacteria (10, 16–18). The reasons why fur is important and even an essential gene in some bacteria remain elusive but possibly involve multiple factors (18–20).

Iron plays an important role during bacterial infection. It has been shown that iron scavenging by host proteins such as transferrin is a key strategy employed by mammalian hosts to deprive bacteria of iron, a process known as nutritional immunity (21, 22). The bacterial counterattack involves the production of siderophores to chelate iron and the expression of TBDRs for iron uptake directly from heme or host iron-binding proteins (7, 22, 23). Environmental pathogenic bacteria deal with iron limitation in both free-living and host-associated conditions. Recently, our group found that such an environmental pathogen, the Gram-negative betaproteobacterium Chromobacterium violaceum, relies on endogenous catecholate siderophores for iron uptake and full virulence in mice (24). C. violaceum is an opportunistic human pathogen that causes infections with high mortality rates (35% to 53% of the cases) in part due to its high intrinsic antibiotic resistance (25, 26). Although C. violaceum has emerged as a model to study bacterial infection (26, 27), little is known about how C. violaceum copes with the limitation or excess of iron (24, 28) and the impact of iron-sensing regulators on its virulence.

In this work, we determined the role of Fur in C. violaceum by performing a deep phenotypic characterization of spontaneous and null fur mutant strains. Fur absence decreased growth and cell fitness, reduced swimming motility and biofilm production, and increased siderophore release and sensitivity to iron and oxidative stresses. We also demonstrated that the fur null mutant strain is highly attenuated for virulence. Using a transposon mutagenesis strategy, we selected a strain from the fur null mutant background rescued from fitness disturbance but still attenuated for virulence in mice, reinforcing the role of Fur as a key virulence determinant in C. violaceum.

RESULTS

Conditional essentiality of Fur in C. violaceum and selection of a spontaneous fur mutant.

Our group has been able to obtain null mutant strains in C. violaceum using a well-established allelic exchange mutagenesis protocol (24, 29, 30). However, we failed here to obtain a C. violaceum Δfur mutant strain using the same approach. Approximately 200 colonies were PCR screened, and all of them presented a genotype that had reverted to that of the wild-type (WT) strain (representative example in Fig. 1A). These findings raise the possibility of fur being an essential gene in C. violaceum. To test this hypothesis, we tried to delete fur in the WT(pMR20fur) strain, in which extra episomal copies of fur are provided in the replicative vector pMR20 (Table 1). Using this strategy, we promptly obtained strains with chromosomal deletion of fur, with five events detected in 13 PCR-tested colonies (data not shown). Another possibility is that the absence of fur renders C. violaceum susceptible to iron intoxication. To test this hypothesis, we repeated the allelic exchange procedure in the WT strain, but now the second recombination event was selected on sucrose agar plates supplemented with the iron chelator 2,2′-dipyridyl (2,2′-DP). Using this approach, we obtained six Δfur mutant strains from 24 PCR-screened colonies (representative example in Fig. 1B), indicating that fur is essential for C. violaceum only under certain conditions.

FIG 1.

Genetic confirmation of C. violaceum fur mutants. (A and B) Generation of a fur null mutant strain. PCR to detect the first and second recombination events (1st rec and 2nd rec, respectively) in the allelic exchange protocol. The expected PCR product sizes were 1,633 bp for the WT strain and 1,249 bp for the Δfur mutant strain. Representative reactions from 2nd rec colonies selected in LB medium (A) or LB medium depleted of iron with 100 μM 2,2′-DP (B). WT, genomic DNA from wild-type C. violaceum. NC, negative control. MM, molecular marker Thermo Scientific GeneRuler 1 kb plus DNA ladder. (C) Confirmation of a manganese-selected spontaneous fur mutant. DNA sequencing of the fur gene from the Fur^R40S strain indicated a C118A mutation that resulted in an R40S substitution in the Fur protein. Alignments show only the indicated segments numbered from the ATG and the initiation codon of Fur. Asterisks indicate conservation. The mutation is highlighted in blue (WT) and green (Fur^R40S). (D) Detection of Fur in mutant and complemented strains. Fur was detected by Western blot of the indicated strains with an anti-Fur polyclonal antibody. For additional clarity about the Fur^R40S levels, the Western blot was repeated using a large amount of whole-cell extract (bottom panel). Similar protein loading was confirmed by Ponceau staining of the nitrocellulose membrane.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Descriptiona	Reference or source
Escherichia coli
DH5α	E. coli strain for cloning purposes	51
S17-1	E. coli strain for plasmid mobilization	52
BL21(DE3)	E. coli strain for heterologous expression of proteins	Novagen
SM10λpir	E. coli strain for transposon mobilization	36
C. violaceum
WT	C. violaceum ATCC 12472 WT strain with sequenced reference genome	53
WT(pMR20)	WT control strain harboring the empty pMR20 plasmid	This work
WT(pRKlacZ290)	WT control strain harboring the empty pRKlacZ290 plasmid	This work
WT(pRKlacZ290_pcbaF)	WT strain with cbaF (CV_1486)-lacZ fusion	This work
Δfur	WT strain with fur (CV_1797) gene deleted	This work
Δfur(pMR20)	Δfur mutant with empty pMR20 plasmid	This work
Δfur(pMR20fur)	Δfur mutant complemented with WT copy of fur	This work
Δfur(pMR20furC118A)	Δfur mutant complemented with mutated copy of fur	This work
ΔfurNALR	Δfur mutant NALr with gyrA spontaneous mutation, for transposon mutant selection	This work
Δfur CRISPR-Cas::T8	ΔfurNALR mutant strain with random insertion of T8 transposon	This work
Δfur(pRKlacZ290_pcbaF)	Δfur strain with cbaF-lacZ fusion	This work
Δfur(pRKlacZ290_pcbaF/pSEVA221fur)	Δfur strain with cbaF-lacZ fusion, complemented with WT copy of fur	This work
FurR40S	WT strain with spontaneous fur mutation (C118A) selected in MnCl2 excess	This work
FurR40S(pMR20)	FurR40S mutant with empty pMR20 plasmid	This work
FurR40S(pMR20fur)	FurR40S mutant complemented with WT copy of fur	This work
FurR40S(pMR20furC118A)	FurR40S mutant complemented with mutated copy of fur	This work
FurR40S(pRKlacZ290_pcbaF)	FurR40S mutant with cbaF-lacZ fusion	This work
JF1485848382	WT strain with the CV_1485, CV_1484, CV_1483, and CV_1482 genes deleted (ΔcbaCEBA)	24
ΔcbaCEBA Δfur	WT strain with combined mutations of cbaCEBA and fur	This work
Plasmid
pNPTS138	Suicide vector containing oriT, sacB; Kanr	M. R. K. Alley
pMR20	Broad-host-range low-copy vector containing oriT, Tetr	54
pSEVA221	Broad-host-range vector containing oriT, oriRK2, Kanr	55
pET15b	Expression of proteins with N-terminal His-tag; Ampr	Novagen
pIT2	Plasmid harboring T8 transposon (ISlacZ/hah); Ampr, Tetr	36
pGEM-T easy	Cloning plasmid; Ampr	Promega
pRKlacZ290	pRK2-derived vector with promoterless lacZ gene, Tetr	56

^aKan, kanamycin; Tet, tetracycline; Amp, ampicillin; NAL, nalidixic acid; r, resistance.

We also tried to select spontaneous fur mutants through the cultivation of WT C. violaceum in manganese excess. The rationale for this selection is that manganese excess mimics Fe²⁺-Fur, rendering repression of iron-uptake genes (in WT but not in a fur mutant) even when iron levels within the cell are not sufficient (31). We screened 192 manganese-resistant colonies on peptone-sucrose agar-chrome azurol S (PSA-CAS) plates for siderophore overproduction as an indicator of fur mutation (data not shown). DNA sequencing of 10 of these colonies indicated a C118A missense mutation in the fur gene in one colony that we named the Fur^R40S strain because this gene encodes a Fur protein with an R40S substitution (Fig. 1C). Both Δfur and Fur^R40S mutant strains were trans-complemented with fur or fur^C118A (indicating the C-to-A change at position 118 encoded by fur) cloned into the low-copy-number plasmid pMR20. Western blot analysis with an anti-Fur polyclonal antibody revealed that the Fur protein was undetected in the Δfur strain, and the Fur levels were decreased in the Fur^R40S strain and elevated in the complemented strains harboring fur or fur^C118A in comparison with those of the WT strain (Fig. 1D). Since we used a low-copy vector (Table 1), it is not clear why Fur was found at a high level in the complemented strains.

The C. violaceum fur null mutant has pleiotropic phenotypes under regular in vitro growth conditions.

To assess the fitness of the fur mutants under unstressed conditions, we calculated CFU (strains grew for 20 h in liquid and were plated on solid medium) and determined growth curves in both rich (LB) and minimal (M9 minimal medium supplemented with 0.1% casein hydrolysate [M9CH]) liquid media. In comparison to the WT and Fur^R40S strains, the Δfur strain presented fewer CFU in each medium (Fig. 2A) as well as a growth delay, as shown by colony size measurement of the strains grown for 24 h on LB and M9CH agar plates (Fig. 2B). The impaired growth of the Δfur mutant was also confirmed in liquid media in growth curves carried out in LB (Fig. 2C) and M9CH (Fig. 2D). In both cases, the Δfur mutant growth defect was reverted after complementation (Fig. 2C and D). However, the growth rate and doubling time of the fur mutants remained similar to those of the WT strain in unstressed conditions (see Table S1 in the supplemental material). We also tested the phenotypes of swimming motility and biofilm formation of the fur mutants. While the Fur^R40S mutant showed no difference in swimming motility and static biofilm in comparison to that of the WT strain, the Δfur strain was less motile (Fig. 2E) and produced less biofilm (Fig. 2F). The wild-type phenotypes were restored in the complemented strains (Fig. 2E and F). Altogether, these results show that the fur null mutant has pleiotropic phenotypes even under normal growth conditions. In contrast, the Fur^R40S mutant presents only the phenotype of siderophore overproduction as detected in the screening of this strain on PSA-CAS after manganese selection. These data indicate that the absence of fur is detrimental to C. violaceum physiology and suggest that the Fur^R40S protein is still partially functional.

FIG 2.

The Δfur mutant, but not the Fur^R40S mutant, has pleiotropic phenotypes. (A) Counting of CFU. The indicated strains were grown for 20 h in LB or M9CH medium, and CFU were determined by plating serial dilutions in the same medium. ***, P = 0.0007; ****, P < 0.0001. Two-way analysis of variance (ANOVA) followed by Sidak’s multiple-comparison test. (B) Growth rate on solid medium. The same colonies of the CFU assay (n ∼ 50) were measured after 24 h of incubation in the indicated medium. Colony photos were acquired with a magnifying glass, and measurements were manually performed with ImageJ (represented as a.u., arbitrary units). Representative images are included at the top of the figure. (C and D) Growth curves. The indicated strains were grown in LB (C) or M9CH (D), and aliquots were withdrawn for OD₆₀₀ measurement. All curves were determined in three independent biological replicates. Data are shown as the mean and standard deviation. (E) Swimming motility assay. The indicated strains were grown on M9CH 0.3% agar plates, and the diameter of swimming motility was measured after 48 h postinoculation. Experiments were performed in three biological replicates; ****, P < 0.0001. Two-way ANOVA followed by Dunnett’s multiple-comparison test. (F) Static biofilm assay. After 24 h of cultivation in LB medium, the biofilm formation of the indicated strains was determined by crystal violet staining. Data were plotted as the ratio of biofilm quantification by cell density of the cultures (OD₆₀₀); ****, P < 0.0001. Two-way ANOVA followed by Holm-Sidak’s multiple-comparison test.

The absence of fur has little effect on cell death in liquid media.

It has been shown in Pseudomonas aeruginosa that the absence of fur is lethal on solid medium but not in liquid medium (18). To investigate whether the reduced CFU of the Δfur mutant observed on agar plates (Fig. 2A) would correlate with high cell death in liquid media, we performed LIVE/DEAD cell viability assays (Fig. 3A and B). Quantification of fluorescence images indicated that Δfur and Fur^R40S mutant strains were as viable as the WT strain regardless of medium (LB or M9CH) or growth phase (exponential or stationary). As expected, all strains were less viable in the stationary phase than in the exponential phase (Fig. 3A and B). These data indicate that the decreased CFU of the Δfur mutant on solid medium could not be attributed to increased cell death in the liquid cultures. In P. aeruginosa, the requirement of Fur for colony formation is dependent on the biosynthesis of the siderophore pyochelin (18). To test whether the same phenomena occur in C. violaceum, we generated a ΔcbaCEBA Δfur mutant strain in which the cbaCEBA genes, required for siderophore biosynthesis (24), were also deleted. The CFU counting on LB agar plates of the ΔcbaCEBA Δfur strain was similar to that of the WT strain and higher than the CFU of the Δfur strain (Fig. 3C), suggesting that the impaired colony growth of the Δfur strain was, in part, dependent on siderophore production.

FIG 3.

The viability of the fur mutants is unaffected in liquid medium. (A and B) LIVE/DEAD cell viability assay of the indicated strains grown in different liquid medium and growth phases. (A) Representative fluorescence images showing live (SYTO 9 dye, green) and dead (propidium iodide, red) cells. The white bar in the last image indicates 10 μm. (B) Quantification of live and dead cells using Fiji software. We used at least 60 cells for each strain/condition. (C) The absence of siderophore production in the Δfur mutant increases its fitness. CFU of the indicated strains grown for 24 h on LB medium, determined as stated in the legend of Fig. 2. **, P = 0.001; ***, P = 0.0008; ****, P < 0.0001. One-way ANOVA followed by Tukey’s multiple-comparison test.

Fur is required for protection against iron toxicity and oxidative stress.

As iron overload could play an important role in the reduced fitness of the Δfur mutant strain, we investigated the effect of iron in the fur mutants. In LB medium, the growth of the Δfur mutant was severely impaired with iron supplementation (Fig. 4A) but was less affected by iron limitation (Fig. 4B). These findings were further evidenced in the growth rate and doubling time measurements calculated from all growth curves (Table S1). Accordingly, iron decreased the growth rate of the Δfur mutant on LB agar plates while stimulating the growth of a siderophore-defective mutant used as a control (see Fig. S1A in the supplemental material). Interestingly, when the growth curves were determined in M9CH medium supplemented with iron, the absorbance of Δfur mutant cultures presented a sudden drop in the late exponential phase (Fig. 4C), possibly indicating cell lysis. These Δfur mutant cultures showed a brownish color in the presence of iron (Fig. 4D) and released DNA to the medium, as measured by precipitation of genomic DNA of the culture supernatants (Fig. S1B). To confirm these results, we performed an ethidium bromide fluorescence assay (32) using bacterial cells and culture supernatants. In both cases, we verified an increased fluorescence emission of the Δfur mutant in M9CH (Fig. 4E), and this effect was exacerbated in the presence of iron (Fig. 4F), indicating that the Δfur cells are more permeable and release more DNA. As a control, the same assays performed without ethidium bromide showed only background fluorescence (Fig. S1C and D). Overall, the Fur^R40S strain presented absence or weakly detectable phenotypes related to iron toxicity, and all iron-related growth defects observed in the Δfur mutant were reverted in the complemented strains (Fig. 4). These results indicated that iron intoxicates the Δfur mutant but has little effect in the Fur^R40S mutant.

FIG 4.

Iron intoxicates the Δfur mutant. (A and B) Growth curves in LB with distinct iron levels. Strains were grown in LB supplemented with 250 μM FeSO₄ (A) or LB supplemented with 150 μM 2,2′-DP (B). (C) Growth curves in M9CH supplemented with 50 μM FeSO₄. Growth curves were determined in three independent biological replicates. Data are shown as the mean and standard deviation. (D) Photos of the indicated strains acquired after 24 h of cultivation in M9CH in the presence of different concentrations of FeSO₄. (E and F) Ethidium bromide fluorescence assay. Fluorescence emission of bacteria (filled lines) and supernatants (dashed lines) of the indicated strains grown in M9CH (E) or M9CH supplemented with 50 μM FeSO₄ (F) in the presence of ethidium bromide. The assay was performed in three biological replicates.

Iron reacts with hydrogen peroxide (H₂O₂), producing hydroxyl radicals that are highly toxic to cells (5, 6). As the Δfur mutant was sensitive to iron (Fig. 4), we tested the sensitivity of fur mutants to H₂O₂ at concentrations previously determined for C. violaceum (33). In comparison to that of the WT strain, the Δfur mutant was more sensitive to H₂O₂ treatment at concentrations as low as 5 mM, while for the Fur^R40S mutant, the effect was observed from 10 mM. In both mutant strains, the sensitivity phenotype was restored in the complemented strains (Fig. 5A). Fluorescence assays using the ROS-reactive probe 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA) indicated that the Δfur mutant was under oxidative stress regardless of the presence or absence of iron, while in the Fur^R40S mutant, the level of ROS increased only in the presence of iron. The level of ROS in the complemented strains was comparable to that of the WT strain. As expected, the fluorescence was high for all strains treated with H₂O₂ as a control (Fig. 5B). These data indicated that Fur is required for protection against iron toxicity and oxidative stress in C. violaceum. Curiously, in previous work, we found that exposure to organic hydroperoxides caused upregulation of several genes related to iron uptake, presumably as a consequence of Fur oxidation (33).

FIG 5.

The fur mutants are susceptible to oxidative stress. (A) H₂O₂ sensitivity determined by disc diffusion assays. Measurements of inhibition halos of the indicated strains in the presence of increasing concentrations of H₂O₂. a.u., arbitrary units. *, P = 0.016; **, P = 0.042; ***, P = 0.0001; ****, P < 0.0001. Two-way ANOVA followed by Dunnett’s multiple-comparison test. (B) Fluorescence-based ROS detection using the H₂DCFDA probe. Cells were left untreated or were treated for 40 min with 0.5 mM FeSO₄ or 20 mM H₂O₂ before H₂DCFDA treatment followed by fluorescence measurement. Both experiments were performed in three independent replicates. *, P = 0.028; **, P = 0.0023; ****, P < 0.0001. Two-way ANOVA followed by Dunnett’s multiple-comparison test.

Fur represses siderophore production.

As C. violaceum produces catecholate siderophores (24) and considering that Fur is the transcriptional repressor of siderophore biosynthetic pathways in many bacteria (9, 10), we analyzed the siderophore production of the fur mutants using PSA-CAS plates and nutrition tests (24, 34, 35). Both fur mutant strains released more siderophores than the WT strain over the evaluated time course (Fig. 6A and B). Siderophore overproduction in the Δfur mutant was more evident at 48 h (Fig. 6A and B) and confirmed using spent culture supernatants (Fig. 6C) since Δfur mutant growth was impaired on PSA-CAS plates. We verified that the yellow zones in the CAS plates of the fur mutants were due to siderophore overproduction because these mutants were more effective at stimulating the growth of a siderophore-defective mutant than the WT strain (Fig. 6D). Notably, the Δfur mutant produced more siderophores than the Fur^R40S strain (Fig. 6). Indeed, complementation with fur restored siderophore production in the fur mutants to a level similar to that of the WT strain, but complementation with fur^C118A was not able to fully decrease siderophore production (Fig. 6A and B). Altogether, these results show that siderophore production/release was increased in fur mutants, indicating that Fur represses siderophore production in C. violaceum and that this repression was only partial in the Fur^R40S strain.

FIG 6.

The fur mutants release an increased amount of siderophores. (A) Time course evaluation of siderophore release. The indicated strains produced siderophores (orange halos) on PSA-CAS plates. Experiments were performed in three independent biological replicates, and representative images of one assay are shown. (B) Quantification of siderophore release. The area of the orange halos (represented as a.u., arbitrary units) was measured using Fiji software, discounting the growth area of bacteria, to eliminate interference of growth spreading between strains. (C) Detection of siderophores released in spent culture supernatants. Culture supernatants of the indicated strains grown in M9CH medium for 48 h were placed in PSA-CAS plates. (D) Nutrition tests. The ΔcbaCEBA strain was agar embedded in LB plates supplemented with 125 μM 2,2′-DP. The purple halos indicate the growth of ΔcbaCEBA stimulated by the indicated strains. In all panels, the strains are indicated by the same numbers.

Fur represses cbaF expression in iron sufficiency.

CbaF (CV_1486) is the nonribosomal peptide synthetase (NRPS) enzyme involved in the assembly of the siderophore chromobactin in C. violaceum (24). To test whether cbaF is regulated by iron and Fur, we cloned the promoter region of cbaF into the pRKlacZ290 reporter plasmid, generating a cbaF-lacZ fusion. We generated WT, Δfur, and Fur^R40S strains harboring the cbaF-lacZ fusion to perform β-galactosidase activity assays in iron sufficiency (FeSO₄) or iron limitation (2,2′-DP) in both LB (Fig. 7A) and M9CH (Fig. 7B) media. In the WT strain, the expression of cbaF was higher in iron-chelated conditions (LB 2,2′-DP, M9CH 2,2′-DP) than in iron sufficiency. However, cbaF expression was high even in untreated M9HC, indicating that C. violaceum is under iron limitation in this minimal medium. In the Δfur strain, cbaF expression was derepressed regardless of the medium or iron availability but decreased in iron sufficiency in a fur-complemented strain, indicating that Fur represses cbaF expression in iron sufficiency. Lastly, in the Fur^R40S strain, cbaF expression was partially derepressed in the presence of iron, indicating that the R40S substitution compromises but does not abolish Fur activity (Fig. 7). Altogether, these results indicate that Fur represses cbaF expression in C. violaceum and provides an explanation for the siderophore overproduction phenotype of the fur mutants.

FIG 7.

Fur represses cbaF expression in iron sufficiency. The cbaF expression in response to iron and Fur was determined by β-galactosidase activity assays. The indicated strains harboring a cbaF-lacZ fusion were grown in LB (A) or M9CH (B) medium, and the cultures were either untreated or treated with FeSO₄ or 2,2′-DP. Experiments were performed in three independent biological replicates. ***, P = 0.0003; ****, P < 0.0001. Two-way ANOVA followed by Dunnett’s multiple-comparison test.

Fur is required for C. violaceum virulence in vivo in mice.

In previous work, we determined that siderophores are key virulence determinants in C. violaceum (24). Considering that the fur mutants overproduced siderophores (Fig. 6), we investigated the role of Fur in C. violaceum virulence in BALB/c mice (Fig. 8). Survival curves of animals intraperitoneally (i.p.) injected with C. violaceum strains indicated that the Δfur mutant, but not the Fur^R40S mutant, was attenuated for virulence in comparison to that of the WT strain. This decreased virulence phenotype was reversed when the Δfur mutant was complemented with fur (Fig. 8A). Concordantly, the bacterial burden in the liver was markedly decreased for the Δfur mutant in comparison to that of the WT and complemented strains (Fig. 8B). Collectively, these results show that Fur plays an important role in C. violaceum virulence.

FIG 8.

The Δfur mutant is attenuated for virulence in mice. (A) Survival curves of infected BALB/c mice. Animals (n = 16 for WT and mutants or n = 8 for complemented strains) were i.p. injected with 10⁶ CFU of the indicated strain. Animal survival was monitored for up to 10 days. Statistical analysis was performed by the log rank (Mantel-Cox) test; ****, P < 0.001; n.s. = not significant. (B) Bacterial burden in the liver. BALB/c mice were infected with 10⁶ CFU, and after 16 h of infection, livers were aseptically collected, homogenized, diluted, and plated for CFU quantification; **, P < 0.005. One-way ANOVA followed by Dunnett’s multiple-comparison test.

fur null mutant without fitness disturbance remains attenuated for virulence in mice.

To understand the basis of virulence attenuation of the Δfur mutant strain, we employed transposon mutagenesis screening to isolate mutations that rescued the Δfur mutant growth defects on LB agar plates, a strategy used in P. aeruginosa (18). First, we determined that the vector pIT2 developed for P. aeruginosa (36) is useful to generate unique, random insertions of the transposon ISlacZ/hah (T8) into the chromosome of WT C. violaceum (see Fig. S2 in the supplemental material). By screening colonies with T8 insertions from the Δfur background rescued for regular growth on LB plates, we identified one clone, hereafter named Δfur CRISPR-Cas::T8, in which transposon inserted in the first out of four CRISPR-Cas loci found in the C. violaceum genome (Fig. 9A to D; see also Fig. S3 in the supplemental material). Further characterization indicated that the Δfur CRISPR-Cas::T8 strain (i) releases more siderophores than the Δfur strain, probably due to its improved growth on PSA-CAS plates, reinforcing that fur mutation causes derepression of siderophore production (Fig. 9E) and (ii) shows better growth and fitness than the Δfur strain in LB medium (Fig. 9F to H). These data suggest that mutation of a CRISPR-Cas locus increases the fitness of the Δfur mutant strain, although the reasons remain unknown. The Δfur CRISPR-Cas::T8 strain remained less motile than the WT strain (Fig. 9I) and attenuated for virulence in mice (Fig. 9J), although to a lesser extent than the Δfur strain (Fig. 8A), indicating that more than a fitness disturbance is required to explain the involvement of Fur in C. violaceum motility and virulence.

FIG 9.

Insertion of the transposon ISlacZ/hah (T8) into a CRISPR-Cas locus improves the fitness of the Δfur mutant strain. (A) Identification of T8 insertion into a CRISPR-Cas locus in the chromosome of the Δfur strain (red arrow), generating a Δfur CRISPR-Cas::T8 strain (indicated as Δfur T8). Light blue bars indicate CRISPR array repeats, and dark blue bars indicate CRISPR array spacers. (B) Confirmation of T8 insertion by PCR using the primers CRISPR_ID_Fw and CRISPR_ID_Rv (black arrows indicate primer positions in genome). The predicted PCR product in the WT strain is 635 bp. (C) Additional confirmation of T8 insertion by PCR. Primers anneal in the transposon (lacZ-148) and in the genome (CRISPR_ID_Rv). Predicted PCR product in the Δfur T8 strain is 678 bp. (D) PCR confirming that the selected T8 mutant was still a Δfur mutant. WT, Δfur and Fur^R40S were included as controls. The predicted PCR product of WT is 1,633 bp, and that of the Δfur mutant is 1,249 bp. NC, negative control. DNA fragment sizes based on the molecular marker GeneRuler 1 kb plus DNA ladder. (E) The Δfur CRISPR-Cas::T8 strain releases more siderophores than the Δfur strain. Siderophore evaluated on PSA-CAS plates. (F, G, H) The Δfur CRISPR-Cas::T8 strain shows better growth and fitness than the Δfur strain. (F) Growth on LB agar plates of culture with an OD₆₀₀ of 0.01 of the indicated strains documented after 24 h. (G) Growth curves in LB medium determined in three independent biological replicates. Data are shown as the mean and standard deviation. (H) CFU counting after 24 h on LB. *, P = 0.0344; **, P = 0.0011. One-way ANOVA followed by Dunnett’s multiple-comparison test. (I) Swimming motility assays. Experiments were performed in two biological replicates; ****, P < 0.0001. Two-way ANOVA followed by Tukey’s multiple-comparison test. (J) The Δfur CRISPR-Cas::T8 strain has attenuated virulence. Virulence assay in BALB/c mice (n = 11 for WT or n = 20 for Δfur CRISPR-Cas::T8 strain). Statistical analysis was performed by the log rank (Mantel-Cox) test. ****, P < 0.0001.

DISCUSSION

In this work, we determined the role of Fur in the physiology and virulence of C. violaceum by characterizing distinct fur mutant strains. Our data pointed to the conditional essentiality of fur in C. violaceum, mainly attributed to cell intoxication by iron and ROS in the absence of Fur. In addition to protecting C. violaceum against iron and ROS toxicity, Fur controlled swimming motility, biofilm formation, and siderophore production, all phenotypic traits with the potential to explain the decreased virulence observed in a C. violaceum fur null mutant strain.

Although fur has been described as an essential gene in a few bacteria (19, 37), recent data support the notion of fur as a conditional essential gene (18, 20), as we are proposing for fur in C. violaceum. Indeed, suppressor or transposon mutations in genes related to heme uptake (20), defense against ROS (38), and siderophore production (18) rescued the fitness of fur null mutants in distinct growth conditions in P. aeruginosa and Burkholderia multivorans. Our data, indicating that the isolation of a fur null mutant was possible only in iron-deprived conditions and the phenotype of the high toxicity of iron and ROS for this Δfur mutant, suggest that a key role of Fur in C. violaceum is to avoid iron intoxication. Siderophore overproduction could increase iron uptake in the Δfur mutant, contributing to iron toxicity. Similarly, susceptibility to ROS or elevated ROS levels is a common feature in fur mutants characterized in several bacteria (12, 17, 39–42). Additional factors might contribute to the fitness disturbance of the Δfur mutant, as we were able to obtain mutants in genes related to siderophore biosynthesis and a putative CRISPR-Cas locus in which the Δfur growth defects were rescued. While a similar effect was previously described for the siderophore pyochelin in P. aeruginosa (18), the reasons why a mutation in a CRISPR-Cas locus improved the fitness of Δfur in C. violaceum remain to be determined. Proteomics analysis has identified another CRISPR-Cas locus induced by hydrogen peroxide stress in C. violaceum (43).

In addition to the phenotypes related to iron overload and ROS generation, the C. violaceum fur null mutant showed impaired swimming motility and biofilm formation. A similar effect was described in fur mutants of Helicobacter pylori (44) and Pectobacterium carotovorum (40), while in uropathogenic Escherichia coli, the fur mutant had increased adhesion, biofilm, and motility due to upregulation of fimbrial and flagellar genes (45). Whether genes related to adhesion and motility are part of the Fur regulon in C. violaceum remains to be determined. Although manganese-selected spontaneous fur mutants have been used to study the Fur regulon or the role of fur in bacterial physiology in several bacteria (16, 17, 19, 31, 37), the Fur^R40S mutant selected in C. violaceum by this method shared with the Δfur mutant only the phenotypes of siderophore overproduction and sensitivity to ROS, indicating that the Fur^R40S protein remained partially functional. Indeed, the R40S mutation was located in the N-terminal DNA binding domain of Fur within the stabilization helix (amino acids 35 to 45) but not directly in the recognition helix (amino acids 50 to 64) of the helix-turn-helix (HTH) motif (data not shown).

The role of Fur in virulence has been investigated in several bacterial pathogens (10). For instance, fur mutants have been described as attenuated for virulence in P. carotovorum (40), Xanthomonas vesicatoria (46), H. pylori (47), Campylobacter jejuni (48), Vibrio vulnificus (39), and Francisella tularensis (42). In agreement, our data indicated that a C. violaceum fur null mutant exhibited reduced virulence and liver colonization in mice. The involvement of Fur in bacterial virulence seems to be multifactorial as follows: (i) growth defects of the fur mutants in the host due to their metabolic inabilities, (ii) intoxication of the fur mutants by host-derived ROS, and (iii) Fur acting as a direct regulator of virulence factors (10, 39, 45). The role of Fur in C. violaceum virulence could involve indirect and direct mechanisms since the fur null mutant had fitness defects and pleiotropic phenotypes, but a fur null mutant without fitness disturbance (Δfur CRISPR-Cas::T8) remained attenuated for virulence. One possible direct mechanism could be the increased production/release of siderophores observed in the fur mutant, as previous work determined that extracellular siderophore accumulation markedly decreased virulence in C. violaceum (24). Further global investigation of the Fur regulon will clarify the regulatory basis of the multiple phenotypes shown here, including the involvement of Fur in C. violaceum virulence.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this work are indicated in Table 1. E. coli strains were cultivated in Luria-Bertani (LB) medium at 37°C. C. violaceum strains were cultivated at 37°C in LB medium or M9 minimal medium (49) supplemented with 0.1% casein hydrolysate (M9CH) (24). For the Δfur mutant strain, solid M9 was supplemented with 1% casein hydrolysate. When necessary, cultures were supplemented with kanamycin (50 μg/ml), ampicillin (100 μg/ml), or tetracycline (10 μg/ml). Conditions of iron sufficiency by FeSO₄ supplementation or iron deficiency by 2,2′-dipyridyl (2,2′-DP) treatment were previously defined for C. violaceum (24).

Construction of C. violaceum mutant strains.

An allelic exchange mutagenesis protocol (24, 29, 30) was used to generate the in-frame Δfur (from the WT strain) and ΔcbaCEBA Δfur (from the ΔcbaCEBA strain) deletion mutants (24). For the Δfur mutant strain, selection in sucrose plates included the addition of 100 μM 2,2′-DP. The primers used for cloning, sequencing, and mutant confirmation are listed in Table 2. A spontaneous fur mutant strain (Fur^R40S) was obtained by manganese (MnCl₂) selection in LB, as previously described (31). C. violaceum MnCl₂-resistant colonies were isolated in a range between 25 and 30 mM MnCl₂ and screened on PSA-CAS plates for siderophore overproduction. Mutations in the fur gene were detected by DNA sequencing.

TABLE 2.

Primers used in this work

Purpose and primer name	Sequence (5′→3′)a	Description
Construction of mutant strains
Fur_del1	ATTATTGGGCCCCCGAATCCGCGCGGTCTATC	ApaI/HindIII upstream flanking fragment with 613 bp
Fur_del2	ATTATTAAGCTTGTGACTGGCTTTGCTCATCGTG
Fur_del3	ATTATTAAGCTTTACGGCGAGTGCCCGGATTG	HindIII/EcoRI downstream flanking fragment with 630 bp
Fur_del4	ATTATTGAATTCTGGCCAGATGCTCGGTATGC
M13Fw	GTAAAACGACGGCCAGT	Sequencing of DNA fragments cloned into plasmids
M13Rv	AGCGGATAACAATTTCAC
Construction of complemented strains
Fur_comp_Fw	ATTATTGAGCTCGGTACTCCGATTGACTCAAACGT	SacI/BamHI 556-bp product with fur and its promoter region
Fur_comp_Rv	ATTATTGGATCCAGACAGGCTCGGGACCGAGC
Heterologous expression
Fur_exp_Fw	ATTATTCTCGAGATGAGCAAAGCCAGTCACCT	XhoI/BamHI 432-bp product with fur open reading frame. These primers were also used for sequencing of fur spontaneous mutants.
Fur_exp_Rv	ATTATTGGATCCTCATCCGCGGCGGCCGCT
T7 promoter	TAATACGACTCACTATAGGG	Sequencing of DNA fragments cloned into plasmids
T7 terminator	GCTAGTTATTGCTCAGCGG
β-galactosidase assay
Promot_1486FW	CCTAGCGGATCCGCTGAAGGCCGGCAAGCTG	Amplification of 415 bp of promoter region of cbaF with BamHI/PstI sites for pRKlacZ290 cloning
Promot_1486RV	GGCCTACTGCAGGATGGCTTCCGCGGTCCAG
Transposon mutant generation and insertion identification
GyrA-QRDR-Fw	ATGACCGATAACCTGTTCGCC	419-bp fragment used to sequence the quinolone resistance-determining region of gyrA gene from C. violaceum
GyrA-QRDR-Rv	ATGTCGGCCAACAGCTCGTG
ISTet_Fw	TGGACAGCATGGCCTGCAAC	560-bp fragment of Tet resistance gene to detect the presence of the transposon
ISTet_Rv	TTTCGGCGTGGGTATGGTGG
CEKG2A	GGCCACGCGTCGACTAGTACNNNNNNNNNNAGAG	Random primers used to identify the transposon insertion site
CEKG2B	GGCCACGCGTCGACTAGTACNNNNNNNNNNACGCC
CEKG2C	GGCCACGCGTCGACTAGTACNNNNNNNNNNGATAT
CEKG4	GGCCACGCGTCGACTAGTAC	Specific primers used to identify the transposon insertion site
lacZ-211	TGCGGGCCTCTTCGCTATTA
lacZ-148	GGGTAACGCCAGGGTTTTCC
CRISPR_ID_Fw	TACCGGGGTACCCCAGGCTTCCAGATACTGCG	635-bp fragment used to confirm transposon insertion
CRISPR_ID_Rv	TACCGGGGATCCACTGGCTAGCAATGCGTTCAG

^aUnderlined sequences indicate the restriction enzyme recognition sites used for cloning purposes.

Construction of complemented strains.

For complementation of the fur mutant strains, the entire fur gene (CV_1797), in WT (fur) or mutated (fur^C118A) versions, was PCR-amplified from the WT or Fur^R40S strains and cloned into the replicative plasmid pMR20. The primers used for cloning are listed in Table 2. The constructs were introduced in the fur mutants by conjugation (24, 29).

Expression and purification of recombinant Fur protein.

The coding region of fur was PCR amplified (Table 2) and cloned into the pET15b plasmid (Table 1). The recombinant His-tagged Fur protein was overexpressed from E. coli BL21(DE3) after induction with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 2 h at 37°C in LB. The His-Fur protein was purified from the soluble fraction by nitrilotriacetic acid (NTA)-resin affinity chromatography in phosphate buffer according to manufacturer recommendations (Qiagen). After concentration (Vivaspin 6 concentrator; Sartorius Stedim Biotech) and desalting (PD-10 desalting columns; GE Healthcare), the purified Fur protein was resolved by 15% SDS-PAGE and quantified by Bradford assay (49).

Whole-cell extracts, SDS-PAGE, and Western blot.

Cultures at the exponential phase grown in LB were resuspended in loading buffer to obtain whole-cell extracts (49). Proteins in cell extracts were resolved by 15% SDS-PAGE and transferred to nitrocellulose membranes (GE Healthcare). Protein loading was assessed by staining in Ponceau S 0.1% (wt/vol). Western blot was carried out with a Protein Detector LumiGLO Western blotting kit (KPL) according to the manufacturer’s instructions using anti-Fur antibodies at a 1:1,000 dilution. Anti-Fur polyclonal antibodies were generated by subcutaneous inoculation of purified Fur protein in 6-week-old female BALB/c mice according to a protocol (number 181/2017) approved by the Local Ethical Animal Committee (CEUA) of FMRP-USP.

CFU quantification.

Bacterial cultures were grown for 20 h in LB or M9CH medium starting from an optical density at 600 nm (OD₆₀₀) of 0.01. Cells were serially diluted in phosphate-buffered saline (PBS), and 100 μl of 10⁻⁷ dilution was plated on LB or M9CH according to the medium in which the cells were initially grown. After 24 h of incubation at 37°C, CFU were counted, and colonies were photodocumented with a magnifying glass.

Growth curves.

Growth curves were determined in LB and M9CH without or with 2,2′-DP or FeSO₄, using the concentrations indicated in the figures. Overnight cultures of C. violaceum were diluted in 15 ml of each medium at an OD₆₀₀ of 0.01. Cultures were grown at 37°C under agitation (250 rpm), and aliquots were withdrawn for OD₆₀₀ measurement in an Eppendorf BioPhotometer. Experiments were performed in three independent biological replicates. Growth rate (μ) was calculated using the growth curve data and is shown as reciprocal hours (h⁻¹). We used OD₆₀₀ values at the exponential growth phase (between 3 and 5 h of cultivation). The time required for bacterial cultures to increase by a factor of 2, the doubling time, was calculated using the growth rate (μ) and is shown in minutes.

Swimming motility.

The swimming motility assay was performed in M9CH 0.3% agar plates as previously described (45). A volume of 5 μl of stationary-phase cultures, adjusted to an OD₆₀₀ of 0.1, was inoculated in the center of M9CH plates. Motility was evaluated as the halo diameter of spreading 48 h postinoculation. Experiments were performed in three independent biological replicates.

Static biofilm.

Quantification of biofilms was performed using a crystal violet staining protocol (30) with some modifications. Cultures were grown in LB medium at 37°C for 24 h in glass tubes without agitation before staining with 0.1% (wt/vol) crystal violet. The stained material was resuspended in 1 ml of 100% ethanol, followed by OD₆₀₀ measurement. Experiments were performed in six biological replicates.

LIVE/DEAD cell viability assays.

Cells were assayed for viability in liquid medium using the LIVE/DEAD BacLight bacterial viability kit (Thermo Scientific) according to the manufacturer’s instructions. Cells were grown to mid-exponential or stationary phases in LB or M9CH, washed in 0.85% (wt/vol) NaCl, and stained with SYTO 9 dye and propidium iodide. Cell suspensions were dropped in slides containing 1% agarose pads, and a coverslip was added. Live and dead cells were visualized in a Leica DMI6000 B fluorescence microscope and quantified using FIJI software (https://imagej.net/Fiji).

Ethidium bromide fluorescence.

This assay was performed similarly to a protocol previously described (32). Briefly, cultures were grown in M9CH or M9CH plus 50 μM FeSO₄ until mid-exponential phase. Cells were washed in PBS, normalized to an OD₆₀₀ of 0.3, and added to a Corning 96-well microplate. Immediately before fluorescence reading (excitation, 530 nm; emission, 600 nm) in a SpectraMax i3× fluorescence reader (Molecular Devices), ethidium bromide was added to a final concentration of 1 μg/ml. This assay was also performed using supernatants of cultures grown for 24 h in the growth conditions stated above. As a control, the fluorescence levels of both cells and supernatants without ethidium bromide were recorded. The experiment was performed in three biological replicates.

Hydrogen peroxide sensitivity.

A disc diffusion assay for H₂O₂ was performed in M9 plus 1% CH to allow proper growth of the Δfur mutant strain. Stationary-phase cultures were diluted to an OD₆₀₀ of 0.1 and plated on M9 agar plates containing 1% CH. Sterile paper discs containing different concentrations of H₂O₂ (0, 5, 10, 20 mM) were placed over the bacteria. After 24 h of incubation at 37°C, photos were taken, and the growth inhibition zones were measured using ImageJ software.

H₂DCFDA staining.

The cell permeant 2′,7′-dichlorodihydrofluorescein diacetate (H₂DCFDA; Thermo Scientific) was used as a probe for the measurement of reactive oxygen species (ROS) inside the cells according to the manufacturer’s instructions. Bacterial cells were grown in M9CH until mid-exponential phase and were either untreated or treated for 40 min with 0.5 mM FeSO₄ or 20 mM H₂O₂. After treatment with dimethyl sulfoxide (DMSO)-diluted H₂DCFDA (final concentration of 10 μM for 45 min), the cells were washed and plated in a Corning 96-well microplate. Fluorescence (excitation, 490 nm; emission, 525 nm) was read in a SpectraMax i3× fluorescence reader (Molecular Devices). The experiment was performed in three biological replicates.

Siderophore production and cross-feeding nutrition tests.

Siderophore production was detected with the universal chrome azurol S (CAS) agar plate assay (34), modified by the replacement of MM9 medium with peptone-sucrose agar (PSA) (24, 35). C. violaceum stationary-phase cultures (5 μl) were dropped on PSA-CAS plates, and siderophore production was evaluated by the orange halos that appeared after incubation for 18, 24, or 48 h at 30°C. To measure siderophore release in supernatants, cells were grown in M9CH for 48 h, and supernatants were obtained by centrifugation and filtration in a 0.22-μm pore filter. Approximately 80 μl of cell-free supernatants was placed in PSA-CAS plates. A cross-feeding nutrition test was performed with the ΔcbaCEBA strain embedded in LB agar chelated with 125 μM 2,2′-DP. Over the solidified agar, 5 to 10 μl of cultures from the tested strains were dropped. Cross-feeding was assessed by the lighter purple halos indicative of growth stimulation of the agar-embedded ΔcbaCEBA strain.

Construction of transcriptional lacZ fusions and β-galactosidase activity assay.

The upstream region of cbaF (CV_1486) was PCR amplified with proper primers (Table 2) and cloned into the pGEM-T easy plasmid (Promega). A BamHI/PstI fragment was subcloned into the pRKlacZ290 vector, generating a cbaF-lacZ transcriptional fusion. C. violaceum cultures harboring this reporter plasmid were grown until mid-exponential phase in LB or M9CH and were either untreated or treated with the indicated concentrations of 2,2′-DP or FeSO₄ for 1 h. Cells were assayed for β-galactosidase activity based on a previously described protocol (50). The experiment was performed in three biological replicates.

Mouse virulence assay.

Virulence assays were performed in a mouse intraperitoneal (i.p.) model of C. violaceum infection as previously described (27, 33). Bacteria were cultivated and diluted according to the CFU quantification stated above. After serial dilution, a dose of 10⁶ CFU was i.p. injected into 6-week-old female BALB/c mice. Animal survival was monitored for up to 10 days postinfection. For counting the bacterial burden in the liver, mice were infected as stated above and euthanized 16 h postinfection. Livers were aseptically collected, homogenized in PBS, diluted, and plated for CFU quantification.

Transposon mutagenesis and insertion site identification.

To obtain transposon mutants in C. violaceum, we used the plasmid pIT2 containing the transposon ISlacZ/hah (T8) developed for P. aeruginosa (36). To allow antibiotic selection after pIT2 conjugation, we isolated C. violaceum nalidixic acid-resistant colonies (WT CV^NALR or Δfur^NALR strain) containing a point mutation in the gyrA gene as previously described (30). E. coli SM10λpir(pIT2) was conjugated with the CV^NALR or Δfur^NALR strain, and C. violaceum pIT2-recipient colonies were selected in LB plates with tetracycline and nalidixic acid. The success of pIT2 in achieving transposon mutants in C. violaceum was verified by (i) PCR detection of the tetracycline resistance gene (Table 2), (ii) the absence of replicative pIT2 in C. violaceum confirmed by minipreparation of plasmid DNA, and (iii) random insertion of the transposon ISlacZ/hah (T8) into the chromosome of C. violaceum WT verified by Southern blotting. Briefly, genomic DNA was extracted, digested with EcoRI, resolved in a 0.8% agarose gel, and transferred to a nylon membrane (49). The membrane was hybridized with a radiolabeled probe for the tetracycline resistance gene (Table 2), and the signal was detected by autoradiogram. To isolate mutations that improve the fitness of the Δfur mutant, colonies with transposon insertion in the Δfur^NALR background were screened by visual inspection for regular growth in LB plates. The transposon insertion site in one such clone (Δfur CRISPR-Cas::T8) was determined using a protocol of semidegenerative PCR (Table 2), followed by Sanger sequencing, as previously described (36).

Statistical analysis.

Statistical analysis was performed using GraphPad Prism version 8 (GraphPad, San Diego, CA). Data distribution was tested using a Shapiro-Wilk normality test, assuming statistical significance at the 5% level. The comparison tests performed in each case are indicated in the figure legends.