Regulation of Fatty Acid Metabolism by FadR Is Essential for Vibrio vulnificus To Cause Infection of Mice

Abstract

The opportunistic bacterial pathogen Vibrio vulnificus causes severe wound infection and fatal septicemia. We used alkaline phosphatase insertion mutagenesis in a clinical isolate of V. vulnificus to find genes necessary for virulence, and we identified fadR, which encodes a regulator of fatty acid metabolism. The fadR::mini-Tn5Km2phoA mutant was highly attenuated in a subcutaneously inoculated iron dextran-treated mouse model of V. vulnificus disease, was hypersensitive to the fatty acid synthase inhibitor cerulenin, showed aberrant expression of fatty acid biosynthetic (fab) genes and fatty acid oxidative (fad) genes, produced smaller colonies on agar media, and grew slower in rich broth than did the wild-type parent. Deletion of fadR essentially recapitulated the phenotypes of the insertion mutant, and the ΔfadR mutation was complemented in trans with the wild-type gene. Further characterization of the ΔfadR mutant showed that it was not generally hypersensitive to envelope stresses but had decreased motility and showed an altered membrane lipid profile compared to that of the wild type. Supplementation of broth with the unsaturated fatty acid oleate restored wild-type growth in vitro, and infection with oleate in the inoculum increased the ability of the ΔfadR mutant to infect mice. We conclude that fadR and regulation of fatty acid metabolism are essential for V. vulnificus to be able to cause disease in mammalian hosts.

Vibrio vulnificus is a gram-negative opportunistic pathogen responsible for two serious diseases. Septicemia occurs after ingestion of raw, contaminated seafood in susceptible individuals, and wound infection occurs after contact with contaminated seawater or seafood products, leading to necrotizing fasciitis in otherwise healthy hosts (reviewed in reference 17). V. vulnificus infections are characterized by extremely rapid replication of the bacteria in the host and extensive tissue damage, with mortality rates as high as 75% and 50% for septicemia and wound infection, respectively (17). Although there are only about 50 confirmed cases of V. vulnificus infections annually in the United States (46), this bacterium has the potential for increased disease incidence during times of natural disasters. During hurricane Katrina in 2005, 14 cases of wound infection were attributed to V. vulnificus, and 3 resulted in death (4).

The means by which V. vulnificus causes such rapid, destructive, and lethal infection remain uncertain, despite a considerable amount of published research (17). The known virulence factors include the antiphagocytic capsule (43, 50, 51); iron sequestration via siderophores (28); flagella and motility (24, 26); type IV pili (36); the fibronectin-binding outer membrane protein OmpU (15); the HlyU protein, which upregulates expression of some toxin genes (22); and the RtxA1 toxin (23, 25, 29). Although some of the virulence factors discovered to date aid in growth and tissue damage, the levels of attenuation seen in these mutants vary, with some showing only marginal decreases in virulence. This led us to believe that important virulence factors remained to be discovered.

Our goal is to identify and characterize virulence factors of V. vulnificus that are involved with rapid growth, evasion of host defenses, and tissue damage. We used alkaline phosphatase (PhoA) mutagenesis (31) with a transposon-based delivery system for phoA based on mini-Tn5Km2 (7) (R. N. Brown and P. A. Gulig, unpublished results). PhoA mutagenesis takes advantage of the fact that the alkaline phosphatase enzyme is active only when transported outside the cytoplasm of a bacterial cell (19). A truncated phoA gene lacking a promoter and secretion signal sequence is randomly inserted into genes. An in-frame fusion of ′phoA with a gene that encodes a secreted protein can be screened by using a chromogenic substrate for alkaline phosphatase, 5-bromo-4-chloro-3-indolyl phosphate (BCIP) (19). As part of our initial studies using PhoA mutagenesis, we obtained several mutants with decreased virulence (Brown and Gulig, unpublished). As described here, one of the insertions was in the fadR gene, encoding a regulator of fatty acid metabolism.

There is considerable published work on the roles of FadR in fatty acid metabolism and the glyoxylate shunt in Escherichia coli, but far less is known about FadR in other bacteria. The first reported role for FadR in E. coli was negative regulation of the fatty acid degradation (fad) genes (42). Nunn et al. (35) suggested a role for FadR in unsaturated fatty acid synthesis and noted that FabA enzyme (β-hydroxydecanoyl-thioester dehydrase) activity was decreased in the fadR mutant. Later, the fabB gene was also shown to be positively regulated by FadR in E. coli (3). E. coli FadR also activates iclR, encoding the repressor of the glyoxylate shunt enzymes (16), and represses uspA, encoding a universal stress protein (10). In E. coli, fadR mutants produce smaller amounts of unsaturated fatty acids than do wild-type stains, but this was reported to be “phenotypically asymptomatic” (35). There have been no reports of FadR being essential for bacterial infection, and there is a scarcity of literature concerning the role of fatty acid metabolism during infection.

In the presently described work, we identified a transposon insertion in the fadR gene that did not result in an in-frame fusion to ′phoA but greatly attenuated the ability of V. vulnificus to cause infection and death in iron dextran-treated mice. A fadR deletion was then constructed for better genetically controlled studies. The fadR mutants were defective in causing local skin lesions and lethal systemic liver infection compared with the wild-type parent. We also confirmed various FadR phenotypes in terms of altered gene expression, fatty acid composition, and sensitivity to certain drugs. We propose that FadR and proper coordination of fatty acid metabolism are important for the ability of V. vulnificus to establish lethal infection of mammalian hosts. These results could open the door for new avenues of antibacterial chemotherapy.

MATERIALS AND METHODS

Bacterial strains and media.

The bacterial strains are listed in Table 1. V. vulnificus FLA399 is a spontaneous, rifampin-resistant derivative of clinical isolate CMCP6 (22). FLA602 is a mini-Tn5Km2phoA insertion mutant of FLA399 with the insertion in VV1_2233 (fadR), and FLA614 is CMCP6 ΔfadR::aph. E. coli Top10 (Invitrogen Corporation, Carlsbad, CA) was used for routine cloning procedures. TransforMax EC100D pir⁺ electrocompetent E. coli (Epicentre Biotechnologies, Madison, WI) allowed for lacZα complementation blue-white screening and maintenance of pir-dependent suicide plasmids. E. coli S17-1λpir (41) was used as the donor strain for conjugations with V. vulnificus.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant characteristic(s)	Reference or source
Strains
E. coli
EC100D pir+	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80dlacZΔM15 ΔlacX74 recA1 endA1 araD139 Δ(ara, leu)7697 galU galK λ−rpsL (Strr) nupG pir+(DHFR)	Epicentre
Top10	F−mcrA Δ(mrr-hsdRMS-mcrBC) φ80dlacZΔM15 ΔlacX74 recA1 araD139Δ(ara-leu)7697 galUgalK λ−rpsL (Strr) endA1 nupG	Invitrogen
S17-1λpir	λpir thipro hsdR hsdM+recA RP4-2 Tc::Mu-Km::Tn7(Tpr Smr)	41
V. vulnificus
CMCP6	Clinical isolate	22
FLA399	Spontaneous rifampin-resistant derivative of CMCP6	This study
FLA602	FLA399 fadR::mini-Tn5Km2phoA	This study
FLA612	FLA602 reverted to fadR+ by allelic exchange	This study
FLA614	CMCP6 ΔfadR::aph	This study
FLA1000	CMCP6 ptsI::mini-Tn5Km2phoA	F. Donoso and P. A. Gulig, unpublished results
Plasmids
pCR2.1	T/A cloning vector; lacZα multiple cloning site; Apr Kmr	Invitrogen
pCVD442	R6K ori-based suicide plasmid; mob RP4; sacB; Apr	9
pGTR201	pUTmini-Tn5Tag3phoA (′phoA delivery vector)	This study
pGTR349	V. vulnificus fadR cloned into pRK437 for complementation	This study
pGTR1129	pCVD442 with lacZα from pUC19 with USER Friendly cloning oligonucleotide linker from pNEB206A (1); cat	Gulig et al., unpublished
pGTR2007	fadR cloned by USER into pGTR1129 for reversion	This study
pGTR2009	Sequences 500 bp upstream and 500 bp downstream flanking fadR cloned by USER into pGTR1129 for deletion of fadR	This study
pGTR2010	pGTR2009 with aph from pUC4K cloned between the fadR upstream and downstream sequences	This study
pRT291	IncP1; TnphoA; Kmr Tcr	48
pRK437	Expression vector; mob RK2; lacZα multiple cloning site; Tcr	40
pUC4K	Kmr derivative of pUC4	47
pUTmini-Tn5Tag3	Mini-Tn5Tag3 delivery vector; R6K ori; mob RP4; Apr Kmr	27

All strains were grown in Luria-Bertani broth containing 0.85% (wt/vol) NaCl (LB-N) or on LB-N plates containing 1.5% (wt/vol) agar. When required, antibiotics were used as follows: rifampin, kanamycin, chloramphenicol, and tetracycline at 50 μg/ml, 40 μg/ml, 30 μg/ml, and 12.5 μg/ml, respectively, for E. coli and 50 μg/ml, 300 μg/ml, 3 μg/ml, and 6.25 μg/ml, respectively, for V. vulnificus. Colistin (10⁵ U/liter) was used to select against E. coli after conjugations. VVM (5), a selective and differential medium for Vibrio spp., was used to verify transconjugants as being V. vulnificus. M9 minimal salts (39) containing 0.2% (wt/vol) glucose or 0.01% (wt/vol) fatty acids was used to assess auxotrophy and the abilities of strains to use fatty acids as sole carbon sources.

Strains were stored at −80°C in LB-N with 35% (vol/vol) glycerol. For mouse infection experiments, static overnight starter cultures of bacteria were grown in culture tubes at room temperature. Before infection, starter cultures were diluted 1:20 into prewarmed LB-N and shaken at 37°C until the optical density at 600 nm (OD₆₀₀) reached 0.4 to 0.6 (exponential-phase growth). The bacteria were diluted in phosphate-buffered saline containing 0.01% (wt/vol) gelatin (BSG) (38) to the appropriate inoculum for infection. Further dilution in BSG and plating were used to confirm the number of CFU/ml inoculated.

Alkaline phosphatase mutagenesis and analysis of disease in mice.

mini-Tn5Km2phoA (essentially a recreation of miniTn5phoA (7) was made by inserting ′phoA from TnphoA (31) into pUTmini-Tn5Tag3 (27) to form pGTR201 (Brown and Gulig, in preparation). E. coli S17-1λpir carrying mini-Tn5Km2phoA on pGTR201 was conjugated with V. vulnificus FLA399 by filter mating. Transconjugants were selected on LB-N agar plates containing rifampin, kanamycin, 40 μg/ml BCIP (Sigma-Aldrich), and 0.2% (wt/vol) glucose. Glucose reduced the endogenous alkaline phosphatase activity of V. vulnificus, allowing for screening of PhoA⁺ fusions on BCIP agar (50). Colonies appearing bluer than FLA399 on BCIP-containing plates were considered PhoA⁺. The initial screen for virulence involved infection of two mice with each PhoA⁺ mutant at an inoculum of 1,000 CFU (three times the wild-type minimum lethal dose). PhoA mutants that caused obvious signs of disease in mice (scruffy fur, lethargy, decreased body temperature, visible skin lesion upon necropsy) were considered virulent and were not studied further. Strains with decreased virulence compared to that of the wild type were examined by a repeat infection of five mice, followed by quantitative analysis of infection, described immediately below. Genomic DNA was extracted (DNeasy blood and tissue kit; Qiagen) from attenuated PhoA⁺ strains for direct sequencing of the mutation by using oligonucleotide primer phoA5′-rev (Table 2) at the University of Florida Interdisciplinary Center for Biotechnology Research DNA Sequencing Core Laboratory.

TABLE 2.

Oligonucleotides used in this study

Oligonucleotide function and primer	Sequence (5′→3′)a
Cloning and deletion of genes
TnPhoA5′-NotI-3	CGCGGCCGCCCTGTTCTGGAAAACCGGGCTGC
PhoA3′-NotI-2	GGCGGCCGCGGTTTTATTTCAGCCCCAGAGCGGC
vv12233-5′-rbs	CGGATCCTGAGTGCCATTCGACCCAAAAC
vv12233-3′	GGGATCCGTCAATTATTAGCTATTAGCAGTCG
FadR5′USER	GGAGACAUGACGACTTCCAGATTCCGCAA
FadR3′USER	GGGAAAGUCTATTAGCAGTCGTCTTCTGTG
sacB5′	AAGTTCCTGAATTCGATTCGTCC
sacB3′	CCTTTCGCTTGAGGTACAGCG
delfadRup5	GGAGACAUCTTGCCAAGTTACTTCCCTTGAA
delfadRup3	ACCCGGGUGGCTCTTTGCCTTAATGACCATT
delfadRdn5	ACCCGGGUAGACGACTGCTAATAGCTAATAATT
delfadRdn3	GGGAAAGUCCGATCCCGCGACCTTCTTG
DNA sequencing
phoA5′-rev	CTGATCACCCGTTAAACG
qRT-PCR
fabA Fwd	TTTGACTGCCATTTCCCTGG
fabA Rev	AACTGCCACATCGCATCCAA
fabB Fwd	TGTTTGCAGGTGGTGGTGAA
fabB Rev	TATTTGGTTGACAGTGCGCC
fadB Fwd	TCGTTGCTTACGCAGCCAAA
fadB Rev	AGCACGCGGTTCACAAAGAA
fadD Fwd	TGATGCCAAACCTGCTGCAA
fadD Rev	TCACAGCAATACAGCCAGCA
16S Fwd	TCGTCAGCTCGTGTTGTGAA
16S Rev	ACTCGCTGGCAAACAAGGAT

^aRestriction sites are underlined, and USER cloning sequences are bolded.

Infection of mice.

The subcutaneously (s.c.) inoculated, iron dextran-treated mouse model of Starks et al. (44, 45) was used. All animal experiments were approved by the University of Florida IACUC. Seven- to 10-week-old female ICR mice (Harlan Sprague-Dawley, Indianapolis, IN) housed under specific-pathogen-free conditions were used. Mice were injected intraperitoneally with 250 μg of iron dextran (Sigma-Aldrich) per gram of body weight at least 45 min before inoculation. Inocula consisting of 0.1 ml of bacteria suspended in BSG were injected s.c. in the lower back. In some experiments, as indicated, BSG was supplemented with 0.225% (wt/vol) sodium oleate. Mice were euthanized by carbon dioxide asphyxiation when rectal temperatures dropped below 33°C, indicating that mice were moribund, or at up to 22 h postinoculation.

After euthanization, the skin was peeled back to reveal the s.c. lesions. Samples of skin lesions and liver were aseptically removed from mice, homogenized in 5 ml of BSG by using glass tissue homogenizers, diluted, and plated on LB-N agar. For analysis of complementation of V. vulnificus FLA614 with pGTR349, samples were plated both nonselectively (LB-N) and selectively (LB-N with 6.25 μg/ml tetracycline). Samples were taken only from mice with visible skin lesions, and a minimum detectable number of CFU/g (10⁴ CFU/g for skin and 10^2.5 CFU/g for liver) was assigned for mice without lesions for statistical analysis.

Complementation of fadR mutations.

The wild-type fadR gene, VV1_2233, was PCR amplified from V. vulnificus FLA399 genomic DNA by using the primers vv12233-5′-rbs and vv12233-3′ (Table 2) with Taq polymerase. The primers were designed to incorporate the putative fadR ribosome-binding site and to introduce flanking BamHI restriction sites for use in cloning. The 898-bp product was T/A cloned into pCR2.1-TOPO and electroporated into E. coli Top10 for lacZα blue-white screening on LB-N plates containing tetracycline and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) (40 μg/ml). The VV1_2233 fragment was excised by BamHI digestion and cloned into the pRK437 cloning vector (40), digested with BamHI so that fadR could be expressed from the lac promoter. The complementing plasmid, pGTR349, was electroporated into E. coli S17-1λpir for conjugation into V. vulnificus FLA602 and FLA614 by filter mating. Transconjugants were selected on LB-N plates containing kanamycin and tetracycline, and the presence of the plasmid was confirmed by agarose gel electrophoresis.

Construction of a fadR::mini-Tn5Km2phoA revertant.

A genomic region of 1,009 nucleotides encompassing VV1_2233 was PCR amplified from V. vulnificus CMCP6 chromosomal DNA by using Taq polymerase and primers FadR5′USER and FadR3′USER to incorporate the USER sites. pGTR1129, a USER capture vector constructed using the allelic exchange vector pCVD442 (9) as a backbone, was used to clone the FadR-USER PCR product according to the USER Friendly cloning procedure (1) (New England Biolabs). DNA from the ligation reaction was purified using a DNA clean and concentrator kit (Zymo Research Corporation, Orange, CA) and electroporated into EC100D pir⁺ for lacZα blue-white screening on LB-N plates containing chloramphenicol and X-Gal. Plasmid mini-extracts of the resulting clones were resolved on an agarose gel to confirm the correct size. The insert of the reversion construct, pGTR2007, was sequenced to confirm the presence of the intact wild-type fadR gene. pGTR2007 was electroporated into E. coli S17-1λpir for conjugation into V. vulnificus FLA602 by filter mating. Transconjugants were selected on LB-N plates containing chloramphenicol and colistin. Genomic DNA was tested by PCR using primers FadR5′USER and FadR3′USER to confirm the presence of the wild-type fadR gene and sacB5′and sacB3′ to confirm the integrated vector. After confirmation of plasmid integration, negative selection on LB-N agar plates containing 6% (wt/vol) sucrose was used to isolate cells that had undergone a second recombination event. This was confirmed by plating to observe chloramphenicol and kanamycin sensitivity and by PCR with primers FadR5′USER and FadR3′USER to show the presence of wild-type fadR, PhoA3′-NotI-2 and TnPhoA5′-NotI-3 to show loss of ′phoA, and sacB5′ and sacB3′ to show loss of sacB. The resulting strain was named V. vulnificus FLA612.

Deletion of fadR from wild-type CMCP6.

A three-part USER deletion method was used (P. A. Gulig et al., unpublished results). Briefly, 500-bp regions flanking VV1_2233 were PCR amplified using oligonucleotides delfadRup5/delfadRup3 and delfadRdn5/delfadRdn3. The upstream and downstream flanking DNA sequences were USER cloned into allelic exchange vector pGTR1129 (pCVD442::lacZα::USER cat) (Gulig et al., unpublished), which was engineered to generate 3′ overhangs for capturing USER Friendly PCR amplicons, yielding pGTR2009. pGTR2009 was linearized by SmaI digestion and ligated with a blunt-ended aph (Kan^r) fragment from pUC4K (47), forming plasmid pGTR2010 carrying ΔfadR::aph. pGTR2010 was introduced into V. vulnificus CMCP6 by conjugation from E. coli S17-1λpir, and a single crossover event was selected by resistance to chloramphenicol and kanamycin. After growth in the absence of chloramphenicol and the presence of kanamycin, double crossover events that replaced the wild-type fadR gene with the ΔfadR::aph allele were selected using sucrose resistance and chloramphenicol resistance. The proper allelic exchange event was confirmed by PCR, and the strain was named FLA614.

Growth rate.

For measurement of growth rate in rich broth, static overnight cultures in LB-N were diluted 1:20 into prewarmed LB-N and grown with shaking at 37°C to exponential phase (OD₆₀₀ = 0.4). Exponential-phase cultures were diluted to approximately 2.5 × 10⁵ CFU/ml in 20 ml of LB-N or LB-N supplemented with 0.005% (wt/vol) sodium oleate and shaken at 37°C. Aliquots were removed for measurement of OD₆₀₀ immediately and every 30 min thereafter for up to 5 h. To determine if strains could use glucose or fatty acids as a sole carbon source, static overnight cultures in M9-glucose were pelleted, washed once in M9 minimal salts without a carbon source, and suspended in 5 ml of M9, M9 plus 0.2% (wt/vol) glucose, M9 plus 0.01% (wt/vol) sodium decanoate (Sigma-Aldrich Inc.), or M9 plus 0.01% (wt/vol) sodium oleate (Sigma-Aldrich). Turbidity after overnight incubation at room temperature indicated growth in the specified medium. Strains were also plated on LB-N agar to confirm viability.

Cerulenin MIC.

A modification of the method of Campbell and Cronan (3) was used. Overnight cultures of V. vulnificus FLA602, CMCP6, and FLA602 (pGTR349) were diluted 1:200 into LB-N. Cerulenin (MP Biomedicals, Inc.) was dissolved in 100% (vol/vol) ethanol at 1 mg/ml and was added to culture tubes containing 1 ml of the diluted bacterial cultures to achieve concentrations of 20, 15, 10, 5, 2.5, 2, and 1 μg/ml. The tubes were incubated with shaking at 37°C overnight. The MIC was the concentration of cerulenin that resulted in no visible growth.

Quantitative reverse transcription-PCR (qRT-PCR) of FadR-regulated genes.

A Qiagen RNeasy mini kit (Qiagen, Inc.) was used to isolate RNA from exponential-phase cultures (OD₆₀₀ = 0.4 to 0.6) grown in LB-N. The RNA was treated with DNase I by using a Qiagen RNase-free DNase set during RNA extraction and was reverse transcribed using an iScript cDNA synthesis kit (Bio-Rad Laboratories, Inc.). To check for DNA contamination, purified RNA without reverse transcriptase was used as a negative control. A standard curve was plotted for each primer set by using known quantities of purified PCR products from CMCP6 genomic DNA. iQ Sybr green Supermix (Bio-Rad) was used for detection via the iQ real-time PCR detection system (Bio-Rad). The oligonucleotide primer sets are listed in Table 2. The PCR cycling conditions were 1 cycle at 95°C for 30 s, 40 cycles at 95°C for 10 s and 60°C for 45 s, 100 cycles at 60°C for 10 s, and a 4°C hold. The PCRs were done in triplicate. Relative expression levels were determined by the 2^−ΔΔCt method, using the 16S rRNA gene as an internal control. At least three separate experiments were performed for each gene.

Fatty acid analysis.

Strains were grown to exponential phase (OD₆₀₀ = 0.4 to 0.6) in LB-N at 37° with shaking. Cells were pelleted by centrifugation and used for fatty acid extraction, methyl esterification, and gas chromatographic analysis with the Sherlock system (MIDI Inc., Newark, DE) at the University of Florida Bacterial Identification & Fatty Acid Analysis Laboratory. Four independent experiments were done.

Assays of membrane stress sensitivity and motility.

The MICs of ethanol, sodium dodecyl sulfate (SDS), and polymyxin B were determined essentially as done for cerulenin (described above). For ethanol, solutions of 6, 5, 4, 3, 2, and 1% (vol/vol) were made using 95% (vol/vol) ethanol in LB-N. The concentrations of SDS were 0.5, 0.45, 0.4, 0.35, 0.3, 0.25, 0.2, 0.15, and 0.1% (wt/vol) in LB-N. The polymyxin B concentrations were 350, 300, 250, 200, 150, and 100 U/ml in LB-N. For the assay of motility, static overnight cultures grown in LB-N at room temperature were stabbed in the centers of motility plates consisting of LB-N containing 0.3% (wt/vol) agar. Motility was measured as the diameter of spread after 15 to 17 h of incubation at 37°C.

Serum sensitivity assay.

Static overnight starter cultures were diluted 1:20 in LB-N and grown to an OD₆₀₀ of 0.4 to 0.6. Cultures were adjusted to 10⁸ CFU/ml and serially diluted for plating to quantify input CFU. An aliquot of 20 μl from the 10⁻¹ dilution (10⁷ CFU/ml) was mixed and then incubated with 180 μl of untreated rat serum or heat-inactivated (30 min at 56°C) rat serum for 2 h at 37°C. The mixture was diluted and plated to quantify survival. The difference in log-transformed number of CFU between untreated and heat-inactivated sera was measured. Three independent assays were performed.

Statistical analyses.

The Student t test was used to examine significant differences between pairs of means. Analysis of variance (ANOVA) with Bonferroni's posttest was used to identify significant differences when more than two pairs of means were analyzed. The χ² test was used in mouse experiments to determine if the number of mice with detectable CFU was significantly changed in mutant versus wild-type infections. P values of <0.05 were considered significant. Statistical analyses were done using Microsoft Excel or GraphPad Prism5 software. All quantitative experiments were repeated at least once.

RESULTS

A V. vulnificus mini-Tn5Km2phoA insertion mutant is severely attenuated for skin and liver infection in iron dextran-treated mice.

Using methods detailed in Materials and Methods, we constructed a library of mini-Tn5Km2phoA insertions in V. vulnificus FLA399 to find mutants with decreased virulence in our mouse model of disease. Colonies that were bluer than the wild type on chromogenic BCIP agar plates were assumed to be PhoA⁺ and were screened for virulence in mice. Colonies of V. vulnificus isolate FLA602 had a slightly bluer color than did the wild-type parent on BCIP plates. FLA602 also formed smaller colonies than did the wild type on LB-N agar plates and grew with a longer doubling time than did the wild type in LB-N (see below). However, the slow growth of FLA602 was not due to auxotrophy, since it could grow in M9 minimal medium. The initial screen for virulence involved s.c. inoculation of two iron dextran-treated mice with 1,000 CFU of V. vulnificus FLA602. In the initial screen and in three subsequent infections with inocula ranging from 2,000 to 4,000 CFU, V. vulnificus FLA602 failed to cause skin lesions in mice over a 22-h course of infection (data not shown). Mice inoculated with 300 CFU of wild-type V. vulnificus become moribund at 15 to 18 h postinfection. The FLA602-infected mice remained healthy in appearance, and the rectal temperatures were in the normal range for uninfected mice (37°C). To assess the extent of attenuation, we infected mice with increasing inocula of V. vulnificus FLA602. Inoculation of 10⁴ CFU failed to cause skin lesions; the mice remained healthy in appearance, and body temperatures were similar to those for uninfected mice throughout the 22-h course of infection. Skin lesions were observed only with inocula at or above 10⁵ CFU, and liver infection was detectable only at an inoculum of 10⁷ CFU (Fig. 1). With an inoculum of 10⁵ CFU of FLA602, only three of five mice had visible skin lesions and the mean number of bacteria recovered from the skin lesions (10^5.8 CFU/g skin tissue) was significantly lower than that for the wild type at an inoculum of 10³ CFU (10⁸ CFU/g skin tissue) (P = 0.01). Similarly, the number of bacteria recovered from liver samples of mice inoculated with 10⁵ CFU of the mutant (10^2.4 CFU/g liver) was significantly lower than that for the wild type (10^4.7CFU/g) (P = 0.0007). Even when 10⁶ CFU of the mutant was inoculated, levels of skin infection (10^6.1 CFU/g) and liver infection (10^2.5 CFU/g) were significantly lower than those for the wild type at a 10³ CFU inoculum (P = 0.04 and P = 0.0008, respectively). An inoculum of 10⁷ CFU of FLA602 resulted in wild-type levels of skin infection (10^7.8 CFU/g skin) but low levels of systemic infection (10^2.6 CFU/g; P = 0.002 for comparison to the wild type). FLA602 caused full wild-type levels of infection only when 10⁸ CFU was inoculated. The mean temperatures of mice infected with 10⁵, 10⁶, or 10⁷ CFU of FLA602 were also higher than those of mice inoculated with 10³ CFU of the wild type. Additionally, the proportions of liver samples of mice that yielded bacteria from the mutant infections at 10⁵, 10⁶, and 10⁷ CFU showed statistically significant decreases compared to the proportion of liver samples of mice that yielded bacteria from the wild-type infection (P = 0.002, P = 0.002, and P = 0.04, respectively; χ² test) (Fig. 1).

FIG. 1.

Skin and liver infection caused by wild-type and fadR V. vulnificus in iron dextran-treated mice. Mice were inoculated s.c. with approximately 10³ CFU of wild-type V. vulnificus FLA399 (WT) or 10⁴, 10⁵, 10⁶, 10⁷, or 10⁸ CFU of fadR::mini-Tn5Km2phoA mutant FLA602. At up to 22 h postinfection, mice were euthanized, and skin and liver were sampled for quantification of V. vulnificus. Skin lesions were not observed with inocula of 10⁴ CFU for the fadR mutant. Liver infection was detected only with an inoculum of 10⁷ CFU. For samples with no detectable bacteria, the minimum detectable value was assigned for statistical analysis. Fractions beneath the bars indicate the numbers of mice that yielded detectable numbers of bacteria from the skin or liver samples over the number of inoculated mice. Asterisks indicate statistically significant differences in number of CFU/g tissue or temperature between mutant and wild-type infections (*, P < 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ANOVA with Bonferroni's posttest). Daggers indicate statistically significant differences in number of samples yielding bacteria between mutant and wild-type infections (†, P = 0.04; ††, P = 0.002; χ² test).

The FLA602 mini-Tn5Km2phoA insertion is in the fadR gene.

The blue color of colonies of FLA602 on LB-N agar containing BCIP and glucose suggested that the mini-Tn5Km2phoA formed an in-frame gene fusion of a V. vulnificus gene to ′phoA. However, DNA sequence analysis of genomic DNA from the mutant identified a backward insertion of ′phoA into V. vulnificus CMCP6 gene VV1_2233, annotated as a “fatty acid metabolism regulator.” PSORT (14) prediction based on the amino acid sequence suggested a cytoplasmic protein localization. The amino acid sequence contained the conserved domain FadR and had 52% identity to the well-studied E. coli FadR protein. As such, we named the mutated gene fadR. The mini-Tn5Km2phoA insertion occurred at nucleotide 346 of 840 in the fadR gene, meaning that the N-terminal 40% of the FadR protein was possibly translated.

Because FadR is a cytoplasmic protein, we were puzzled by the blue color of colonies of FLA602 on BCIP plates. We did not resolve the cause but noted that complementation and reversion of the mutation resulted in colonies that were similar in color to the wild type (see below), so the blue color of the fadR::mini-Tn5Km2phoA mutant on BCIP plates was not likely due to a secondary mutation. Furthermore, as detailed below, a ΔfadR mutation did not yield a similar blue color. Hence, the partially translated FadR protein of FLA602 might have caused the blue color.

Confirmation of the fadR phenotype.

Although the gene interrupted by the mini-Tn5Km2phoA insertion possessed significant homology to fadR of E. coli, we further examined FLA602 to confirm the phenotype. The fatty acid synthase inhibitor cerulenin has been used as an indicator of the fadR mutant phenotype in E. coli (2, 3) and to inhibit fatty acid synthesis in V. vulnificus (6). Because FadR positively regulates genes involved in unsaturated fatty acid biosynthesis in E. coli, one aspect of the fadR mutant phenotype is decreased synthesis of unsaturated fatty acids (35). This renders the mutant hypersensitive to cerulenin because the drug further decreases fatty acid synthesis. Similar to E. coli fadR mutants, the V. vulnificus fadR::mini-Tn5Km2phoA mutant showed increased sensitivity to cerulenin compared to the level for the wild-type parent. The mutant had a mean MIC of 2.8 μg/ml cerulenin, in contrast to the wild-type parent, which had a mean MIC of 16.7 μg/ml, representing an approximately sixfold difference (P = 0.001; Student's t test). These values are the means for three biological replicates. This was similar to the 10-fold increase in sensitivity reported for an E. coli fadR mutant examined by Campbell and Cronan (3).

Increased cerulenin sensitivity in the V. vulnificus fadR mutant suggested decreased levels of the 3-ketoacyl-acyl carrier protein synthase I enzyme (FabB), the main target of the inhibitor (2, 37). To examine this possibility and to assess some aspects of FadR-regulated gene expression in FLA602, we performed qRT-PCR using genes known to be regulated by FadR in E. coli. In E. coli, FadR positively regulates at least two genes involved in unsaturated fatty acid biosynthesis, fabA and fabB, and negatively regulates genes involved in β oxidation (reviewed in reference 13). To determine if V. vulnificus FadR functioned similarly, we tested the relative expression levels of fabA and fabB, as well as two genes involved with β oxidation, fadB and fadD, in fadR::mini-Tn5Km2phoA mutant FLA602 and the wild type by using qRT-PCR. All gene expression levels were normalized to 16S rRNA levels as an endogenous control. Consistent with the expected results for a fadR mutation, the expression levels of fabA and fabB were lower and the fadB levels were higher than those of the wild type in FLA602. fadD expression was not significantly changed in FLA602 under the conditions tested (Table 3).

TABLE 3.

Expression of putative FadR-regulated genes in V. vulnificus fadR::mini-Tn5Km2phoA mutant FLA602^a

Gene	Fold change (mean ± SD) in mutant	P
fabA	−8.4 ± 5.3	0.005
fabB	−6.9 ± 1.5	0.0002
fadD	−1.6 ± 2.1	0.1
fadB	4.2 ± 2.4	0.04

^aFLA399 and FLA602 were grown to exponential phase in LB-N, and RNA was extracted. qRT-PCR was used to measure changes with the 2^−ΔΔCt method, using 16S rRNA as an endogenous control between fadR mutant FLA602 and wild-type FLA399. P values are for paired Student t tests comparing normalized threshold cycle values for the wild-type and mutant strains.

In vitro growth of fadR::mini-Tn5Km2phoA mutant FLA602.

Increased sensitivity to cerulenin and changes in fad and fab gene expression suggested that the fadR mutant may be impaired for fatty acid metabolism. E. coli fadR mutants are able to grow using the medium-chain fatty acid decanoate as a sole carbon source, while wild-type cells cannot (20). In contrast, both wild-type and fadR mutant E. coli can use long-chain fatty acids, such as oleate, as carbon sources. We tested the growth of the V. vulnificus fadR::mini-Tn5Km2phoA mutant FLA602 and parental FLA399, using decanoate and oleate as sole carbon sources. Both V. vulnificus strains grew using both fatty acids as sole carbon sources. Unexpectedly, the V. vulnificus fadR mutant grew slower in rich LB-N than did the wild type (29-min versus 18-min doubling times, respectively) (Table 4), a phenotype that was not reported for E. coli fadR mutants. To test if the slow in vitro growth of FLA602 could be responsible for the observed decreased infectivity, we compared FLA602 with another slow-growing mutant generated by phoA mutagenesis, FLA1000, which contains a ptsI::mini-Tn5Km2phoA mutation. FLA1000 had a doubling time of 25 min in LB-N, compared to 18 min for the wild type; however, FLA1000 retained full virulence in mice (Table 4). Therefore, the slow in vitro growth of fadR FLA602 did not necessarily explain its severe attenuation in our mouse model of disease.

TABLE 4.

In vitro growth rates and virulence levels of V. vulnificus strains

Strain	Genotype	Doubling time (min) in LB-N	Virulencea
			Log no. of CFU/g		% of mice with detectable infection
			Skin	Liver
FLA602	fadR::mini-Tn5Km2phoA	29	ND	ND	0b
FLA1000	ptsI::mini-Tn5Km2phoA	25	8.4 ± 0.2	6.3 ± 0.3	100
CMCP6	Wild type	18	8.5 ± 0.2	6.5 ± 1.5	100

^aFor virulence, five iron dextran-treated mice were inoculated s.c. with approximately 1,000 CFU, and infection was measured between 14 and 22 h later. ND, not detected.

^bCompared with levels for FLA1000 and CMCP6 (P = 0.002; χ² test).

Complementation and reversion of the fadR::mini-Tn5Km2phoA mutation.

To prove that the attenuated infectivity phenotype of FLA602 was due to the knockout of the fadR gene by the transposon insertion and not a polar effect or secondary mutation, we introduced the wild-type fadR gene in trans into V. vulnificus FLA602 via plasmid pGTR349 (pRK437 containing the V. vulnificus fadR gene). Complementation of the fadR::mini-Tn5Km2phoA mutant restored the wild-type size to the colonies and reduced cerulenin sensitivity to wild-type levels (data not shown). The complemented strain looked identical to the wild type (pale blue) on BCIP plates, indicating that the blue color of FLA602 was somehow related to the genotype of fadR::mini-Tn5Km2phoA. Complementation also resulted in a significant increase in virulence (Fig. 2). At an inoculum of 1,000 CFU, skin lesions were observed in all mice (P = 0.002 for comparison to the fadR::mini-Tn5Km2phoA mutant; χ² test). These skin lesions yielded a mean of 7.7 ± 0.53 log CFU/g, similar to the level for a wild-type infection at this inoculum. Although high levels of skin infection were restored, liver CFU were detected in only one of five mice (P = 0.29 for comparison to the fadR::mini-Tn5Km2phoA mutant; χ² test). With an inoculum of 10⁴ CFU, however, both skin and liver infections were detected at wild-type levels: 7.8 ± 0.28 log CFU/g skin and 4.8 ± 1.6 CFU/g liver.

FIG. 2.

Quantification of infection by fadR::mini-Tn5Km2phoA mutant FLA602 after complementation in trans with wild-type (WT) fadR or reversion by allelic exchange with wild-type fadR. Iron dextran-treated mice were infected s.c. with approximately 1,000 CFU of each V. vulnificus strain as detailed in Materials and Methods. Enumeration of the complemented strain on nonselective medium is shown. There was not a statistically significant difference between CFU counts on selective and nonselective media. Because the fadR::mini-Tn5Km2phoA strain was avirulent at this inoculum and did not yield CFU, the minimum detectable values for skin and liver, 10⁴ and 10^2.5 CFU/g, respectively, were assigned. Asterisks indicate statistically significant differences in number of CFU/g tissue or temperature between fadR::mini-Tn5Km2phoA infections and complemented, reverted, or wild-type infections (*, P ≤ 0.006; **, P ≤ 0.001; ***, P ≤ 3 × 10⁻⁷; ANOVA with Bonferroni's posttest.) Daggers indicate statistically significant differences in number of samples yielding bacteria between the mutant skin infection and the complemented, reverted, or wild-type strain or between the mutant liver infection and the reverted or wild-type strain (†, P ≤ 0.01; χ² test).

Because complementation of the fadR::mini-Tn5Km2phoA mutant did not completely restore wild-type infection, we reverted the mutation by allelic exchange with the wild-type fadR gene. This would preclude the possibility of dominant-negative effects of the mutant FadR protein, in which the N-terminal 40% of the protein was intact. The truncated FadR protein would contain the critical N-terminal DNA-binding domain with residues that are predicted to form alpha helices that interact during dimerization and DNA binding (49). Thus, the truncated FadR protein could potentially dimerize with wild-type FadR provided in trans, thereby reducing the efficiency of the complementation. In contrast to the complementation, reversion of the fadR::mini-Tn5Km2phoA mutation completely restored virulence in terms of skin and liver infection (Fig. 2). Mean bacterial yields from skin (10^7.6 CFU/g) and liver (10^5.2 CFU/g) tissues of infected mice were similar to those for the wild type (10^8.2 CFU/g skin and 10⁴ CFU/g liver) at the same inoculum (1,000 CFU). Therefore, we concluded that the attenuated phenotypes of FLA602 were due to the fadR mutation.

Deletion of fadR from wild-type V. vulnificus CMCP6.

To address the failure to fully complement the fadR::mini-Tn5Km2phoA mutation in trans, we deleted the fadR gene from wild-type V. vulnificus CMCP6, as described in Materials and Methods. The deletion mutant, FLA614, shared colony morphology characteristics with mini-Tn5Km2phoA insertion mutant FLA602; colonies were small and slightly yellow compared to those of the wild type. FLA614 also grew slower than the wild type in rich broth (discussed below; see Fig. 6) and was hypersensitive to cerulenin (not shown). Surprisingly, ΔfadR FLA614 did not appear bluer than the wild type on BCIP-containing media. This result suggested that the blue color seen for FLA602 was unique to the fadR::mini-Tn5Km2phoA mutation. Most notably, the deletion mutant was attenuated for infectivity in the mouse model of disease. At an inoculum of 1,000 CFU, FLA614 did not cause skin lesions and mice remained healthy throughout the 22-h course of infection (Fig. 3). Surprisingly, infections with increasing inocula showed that the deletion mutant was not as highly attenuated as the mini-Tn5Km2phoA insertion mutant, FLA602. An inoculum of 10⁴ CFU of FLA614 was sufficient to cause skin lesions, although the level of skin infection (10^5.8 CFU/g) was more than 2 logs lower than the mean number of CFU/g recovered from an infection with the wild-type parent, CMCP6, at 1,000 CFU (10^8.4 CFU/g) (P = 0.04). Liver infection was detectable in only one of five mice inoculated with 10⁴ CFU of FLA614 (mean, 10^2.2 CFU/g liver). This value was also significantly lower than that for wild-type recovery (10^4.2 CFU/g liver) (P = 0.003), representing a 100-fold decrease in liver CFU for FLA614 with an inoculum that was 10-fold greater than that used for the wild-type infection. An inoculum of 10⁶ CFU of FLA614 was sufficient to cause wild-type levels of skin and liver infection, with mean recoveries of 10^8.3 CFU/g skin and 10^5.6 CFU/g liver. Most importantly, complementation in trans with the fadR gene expressed from plasmid pGTR349 restored full wild-type virulence to ΔfadR FLA614. One thousand CFU of the complemented deletion strain caused wild-type levels of skin infection (10^8.3 CFU/g) and liver infection (10^5.1 CFU/g) in mice (Fig. 3). Because ΔfadR FLA614 was phenotypically similar to fadR::mini-Tn5Km2phoA FLA602 and was fully complemented in trans, we concluded that the fadR mutation was indeed responsible for the morphology, growth, and virulence phenotypes seen in both mutants, and FLA614 was used as a definitive fadR mutant for subsequent experiments.

FIG. 6.

Supplementation with oleate in vitro restores wild-type growth to V. vulnificus ΔfadR FLA614. A concentration of 2.5 × 10⁵ CFU of wild-type CMCP6 or ΔfadR mutant FLA614 was inoculated into LB-N or LB-N plus 0.005% (wt/vol) sodium oleate and grown with shaking at 37°C. OD₆₀₀ was measured every 30 min. Supplementation with oleate increased the growth rate and final yield of the ΔfadR mutant to wild-type levels. Data shown are means ± standard deviations for two independent experiments. Asterisks indicate significant differences between the ΔfadR mutant grown in LB-N (fadR-LBN) and the ΔfadR mutant grown with oleate (fadR-OL), the wild type grown in LB-N (WT-LBN), or the wild type grown with oleate (WT-OL) at each time-point (*, P < 0.01; **, P < 0.001; two-way ANOVA followed by Bonferroni's posttest).

FIG. 3.

Virulence of V. vulnificus ΔfadR and its complemented strain. Mice were inoculated with 1,000 CFU of ΔfadR::aph mutant FLA614 (ΔfadR), fadR-complemented strain FLA614(pGTR349) [ΔfadR(fadR)], or wild-type CMCP6 (WT). Enumeration of the complemented strain on nonselective medium is shown. There was not a statistically significant difference between CFU counts on selective and nonselective media. The ΔfadR strain was avirulent at 1,000 CFU and was assigned the minimum detectable values. The minimum detectable numbers of CFU for skin and liver samples are 10⁴ and 10^2.5 CFU/g, respectively. Fractions beneath the bars indicate the proportion of samples that yielded bacteria. Daggers indicate statistically significant differences in number of samples yielding bacteria between the mutant infection and the wild-type or complemented strain (†, P ≤ 0.01; ††, P = 0.002; χ² test). Asterisks indicate significant differences in mean number of CFU/g tissue or temperature between the ΔfadR mutant and the wild-type or complemented strain (*, P = 0.01; **, P = 4 × 10⁻³; ***, P ≤ 2 × 10⁻¹⁰; ANOVA with Bonferroni's posttest).

Altered fatty acid content of V. vulnificus ΔfadR.

Given that the fadR insertion or deletion mutations caused hypersensitivity to cerulenin and that decreased fab gene expression was evident, we hypothesized that a fadR mutant should exhibit decreased synthesis of unsaturated fatty acids that could lead to an altered membrane lipid profile (essentially all fatty acids and lipids in bacteria are found in the cell envelope [8]). Gas chromatographic analysis of fatty acid methyl esters derived from ΔfadR FLA614 and wild-type CMCP6 revealed a small (13%) but statistically significant (P = 0.005) decrease in unsaturated fatty acids and a concurrent 12% increase in saturated fatty acids in the ΔfadR mutant compared to the levels for the wild type (P = 0.006) (Table 5). Taken together, the increase in cerulenin sensitivity, the decrease in V. vulnificus fab gene expression, and the decrease in cellular unsaturated fatty acids suggested that fadR mutants are deficient in synthesis of unsaturated fatty acids.

TABLE 5.

Changes in the fatty acid profile of V. vulnificus ΔfadR relative to that of the wild type^a

Fatty acid type	Relative abundance (mean ± SD) in sample		% Change in mutant	P
	Wild type	ΔfadR
Saturated	41.4 ± 1.9	46.9 ± 1.8	12	0.006
Unsaturated	57.8 ± 2.5	51.1 ± 1.8	−13	0.005

^aWild-type CMCP6 and ΔfadR mutant FLA614 were grown to exponential phase in rich broth, pelleted, and washed. Equal wet masses of cells of each strain were used for methyl esterification and gas chromatography-mass spectrometry. Relative abundance values are means ± standard deviations for four independent experiments. P values are for unpaired Student t tests comparing mutant and wild-type fatty acid abundance values.

Envelope stress sensitivity and motility of V. vulnificus ΔfadR.

Because FLA614 showed slightly lower membrane unsaturated fatty acids than did the wild type, it was possible that the unbalanced fatty acid composition might render the mutant more susceptible to external stresses, possibly explaining the attenuated infection of mice. To test the susceptibility of the ΔfadR mutant to envelope stresses, we performed MIC experiments using chemicals that disrupt membrane integrity: ethanol, a membrane perturbant; SDS, a detergent that solubilizes the lipid bilayer; and polymyxin B, which binds lipopolysaccharide and disrupts membrane integrity. The MICs of ethanol and of polymyxin B for the ΔfadR mutant were similar to those for the wild type (Table 6). The ΔfadR mutant was slightly, but significantly, more sensitive to SDS than was the wild type, with 0. 22% (wt/vol) SDS causing growth inhibition of the mutant, compared to 0.33% for the wild type (P = 0.02) (Table 6). In light of the small magnitude of the change in SDS sensitivity and the fact that FLA614 was not hypersensitive to other membrane-perturbing agents, FLA614 was generally not more sensitive than the wild type to envelope stresses.

TABLE 6.

Sensitivities of wild-type and ΔfadR V. vulnificus strains to envelope stresses^a

Envelope stress	MICb (mean ± SD)		P
	Wild type	ΔfadR
Ethanol	5 ± 0.8	4 ± 0.8	0.13
Polymyxin B	192 ± 38	217 ± 61	0.41
SDS	0.33 ± 0.06	0.22 ± 0.06	0.02

^aWild-type CMCP6 and ΔfadR mutant FLA614 were grown to exponential phase in rich broth. The MICs for ethanol, SDS, and polymyxin B were measured. Values are means ± standard deviations for at least four independent experiments. P values are for unpaired Student t tests comparing mutant and wild-type MICs. Except for a slight increase in sensitivity to SDS, the ΔfadR mutant was not more sensitive to envelope stresses than was the wild type.

^bValues for ethanol are percentages (vol/vol), values for polymyxin B are numbers of U/ml, and values for SDS are percentages (wt/vol).

We also tested sensitivities to heat and cold as indications of membrane integrity. ΔfadR mutant FLA614 and wild-type CMCP6 were grown to exponential phase in LB-N, and serial dilutions were spotted onto LB-N plates and incubated at 37°C, 42°C, or 4°C for 18 h. FLA614 grew as well as CMCP6 at 37°C and 42°C (Fig. 4), but neither strain showed signs of growth after 18 h of incubation at 4°C (not shown). Further incubation of the 4°C plates overnight at room temperature (23°C) resulted in equal growth for both strains (Fig. 4). Overall, the ΔfadR mutant did not exhibit increased sensitivity to envelope stresses compared to that of the wild type.

FIG. 4.

Sensitivities of wild-type CMCP6 (WT) and ΔfadR FLA614 to heat and cold. Strains were grown to exponential phase at 37°C, and serial dilutions were aliquoted onto LB-N plates and incubated at the temperatures indicated for 18 h. The plates incubated at 4°C showed no visible growth after 18 h and were photographed after an additional overnight incubation at room temperature. Images are representative of triplicate plates. FLA614 was not more sensitive to heat or cold than was the wild type.

As a further test of membrane defects, we tested the ΔfadR mutant for sensitivity to serum complement as an indicator of decreased resistance to host defenses. ΔfadR mutant FLA614, wild-type CMCP6, and complement-sensitive E. coli MG1655 bacteria were incubated in intact or heat-inactivated 90% rat serum for 2 h and then plated to quantify survival. E. coli MG1655 was completely killed by incubation with rat serum. The ratio of MG1655 bacteria that survived in untreated versus heat-inactivated sera was log −6.6 ± 0.4 (P = 10⁻⁵ for comparison to V. vulnificus CMCP6; n = 3 biological replicates). In contrast, neither the ΔfadR mutant nor wild-type V. vulnificus was sensitive to complement-mediated killing. The ratio of CMCP6 bacteria that survived in untreated versus heat-inactivated sera was log 0 ± 0.1, and that for FLA614 was log −0.2 ± 0.2 (P = 0.19 for comparison to the wild type; n = 3 biological replicates). Because the ΔfadR mutant survived the bactericidal activity of serum complement as well as did the wild type, we concluded that the ΔfadR mutation did not cause a significant defect in the integrity of the cell envelope.

In addition to the possibility that loss of fadR could cause changes in envelope stress sensitivity, it was also possible that the ΔfadR mutant could be affected for motility. Motility is a complex function that requires the assembly and activity of the flagellum and motor apparatus across the cell envelope as well as ion gradients across the inner membrane to provide energy. Furthermore, flagella of the Vibrionaceae possess a sheath that appears to be an extension of the cell outer membrane (reviewed in reference 33). We therefore hypothesized that the altered fatty acid profile of the ΔfadR envelope could lead to changes in motility. We tested the motility of ΔfadR FLA614 and wild-type CMCP6 by measuring the diameter of spread through 0.3% motility agar. FLA614 was only 45% as motile as the wild type (Fig. 5) (P = 3 × 10⁻⁷). Motility was fully restored by complementation. We considered the possibility that the apparent decreased motility of ΔfadR FLA614 could have been due to its decreased growth rate. However, slow-growing V. vulnificus FLA1000 (Table 4) was fully motile compared to the wild type (data not shown).

FIG. 5.

Decreased motility of the ΔfadR mutant. Static overnight cultures of FLA614 (ΔfadR), complemented FLA614(pGTR349) [ΔfadR(fadR)], and wild-type CMCP6 (WT) were stabbed into the centers of 0.3% agar motility plates. The diameter of spread of bacteria was measured 15 h to 17 h later, and the mean for the wild type was assigned a value of 100%. Six plates per strain were used. At least two independent experiments were performed. Asterisks indicate significant differences between the ΔfadR mutant and the wild-type or complemented strain (*, P = 3 × 10⁻⁷; **, P = 3 × 10⁻⁸; ANOVA with Bonferroni's posttest).

Supplementation with unsaturated fatty acid in vitro and in vivo restores growth and infectivity.

Deletion of V. vulnificus fadR resulted in a decreased growth rate in vitro and decreased levels of membrane unsaturated fatty acids. To test if the decreased in vitro growth rate of the V. vulnificus ΔfadR mutant was related to decreased unsaturated fatty acid synthesis, we supplemented LB-N with an unsaturated fatty acid derivative, sodium oleate, and measured the growth of ΔfadR FLA614 and wild-type CMCP6. The addition of 0.005% (wt/vol) sodium oleate (half the concentration used as a sole carbon source in M9 minimal medium) facilitated the growth of the ΔfadR mutant to wild-type levels (Fig. 6). Interestingly, wild-type CMCP6 did not grow any better when oleate was added to the growth medium, suggesting that the availability of fatty acids in LB-N was not a limiting factor for wild-type growth. The addition of 0.005% (wt/vol) sodium oleate to LB-N plates also significantly increased the mean diameter of colonies formed by the ΔfadR mutant. While the mean colony diameter of the ΔfadR mutant grown on LB-N plates was 2.4 ± 0.2 mm, the mean colony diameter for the mutant grown on LB-N plates supplemented with sodium oleate was 3.7 ± 0.4 mm (P = 4 × 10⁻⁴). Similar to the trend seen for wild-type CMCP6 grown in LB-N with oleate, the addition of oleate to LB-N plates did not significantly alter the mean colony diameter of wild-type CMCP6 (5.0 ± 0 mm on LB-N and 5.2 ± 0.4 mm on LB-N plus oleate; P = 0.35). Because supplementation with oleate functionally complemented the growth defect of the ΔfadR mutant in vitro, we inferred that the slow growth of the mutant in vitro was due to insufficient synthesis of unsaturated fatty acids.

To test if a relative lack of unsaturated fatty acids was responsible for the attenuation of the ΔfadR mutant, mice were infected with FLA614 suspended in sodium oleate or in BSG as a control. We used a concentration of sodium oleate (0.225% [wt/vol]) that was nearly 50-fold higher than that used for in vitro supplementation to compensate for equilibration with the fluid phase of the mouse. Mice were also infected with an equal inoculum of wild-type CMCP6 suspended in BSG for comparison. Control mice inoculated with sodium oleate alone showed no skin pathology upon necropsy (not shown). The addition of oleate to the inoculum allowed 1,000 CFU of the ΔfadR mutant to establish wild-type levels of skin infection (10^8.2 CFU/g skin tissue) and near-wild-type levels of systemic infection (10^4.8 CFU/g liver tissue), compared to what was found for CMCP6 (10^8.4 CFU/g skin [P = 0.35] and 10^5.8 CFU/g liver [P = 0.01]) (Fig. 7). In contrast, the ΔfadR mutant without oleate supplementation established infection in only one of the five mice inoculated. Given that the oleate supplementation significantly increased the infectivity of the ΔfadR mutant in mice, we concluded that the major defect of the ΔfadR mutant in vivo was insufficient synthesis of unsaturated fatty acids.

FIG. 7.

Oleate potentiates infection of mice with ΔfadR V. vulnificus. Mice were inoculated s.c. with 1,000 CFU of ΔfadR mutant FLA614 suspended in BSG plus 0.225% (wt/vol) sodium oleate, FLA614 suspended in BSG alone, or wild-type CMCP6 (WT) suspended in BSG. Skin lesion and liver samples were plated to quantify infection. Daggers indicate statistically significant differences in number of samples yielding bacteria between the mutant infection and the oleate-supplemented mutant or the wild type (P = 0.01; χ² test). # indicates significant differences in number of CFU/g liver between oleate-supplemented ΔfadR and the WT (P = 0.01). Asterisks indicate significant differences in number of CFU/g tissue or temperature between the ΔfadR mutant and the wild type or oleate-supplemented ΔfadR (*, P = 0.01; **, P ≤ 0.003; ***, P ≤ 3 × 10⁻⁴; ANOVA with Bonferroni's posttest).

DISCUSSION

Identification of a V. vulnificus fadR mutation that is defective for infection of mice.

We used alkaline phosphatase fusion-insertion mutagenesis to identify secreted virulence factors of V. vulnificus and isolated a mini-Tn5Km2phoA insertion in fadR (fatty acid metabolism regulator) that encodes a cytoplasmic protein. The transposon insertion was backward relative to the fadR open reading frame; hence, there was no phoA fusion. We do not understand the increased blue color of colonies of the fadR::mini-Tn5Km2phoA mutant on BCIP-containing agar plates, but it should be noted that, due to the presence of other alkaline phosphatase genes in the genome, V. vulnificus CMCP6 has a low level of background blue color on BCIP plates. Inclusion of glucose reduces the alkaline phosphatase activity but hinders growth. fadR::mini-Tn5Km2phoA FLA602 was one of the first mutants we isolated using PhoA mutagenesis; hence, it had no positive control for comparison. We subsequently used FLA602 as a negative control for blue color because it did not represent an in-frame phoA fusion. In any case, the fadR::mini-Tn5Km2phoA mutant was significantly attenuated for its ability to infect s.c. inoculated iron dextran-treated mice (Fig. 1), leading us to investigate the role of fadR and fatty acid metabolism in virulence of V. vulnificus. Part of this analysis included the construction of a fadR deletion mutant that generally exhibited the same phenotypes as the mini-Tn5Km2phoA mutant. The lack of blue color on BCIP plates and slightly milder attenuation of the deletion mutant suggested that the mini-Tn5Km2phoA mutation could have been polar on the downstream gene, ribA, encoding GTP cyclohydrolase II. However, ribA is 221 bp downstream, and there is a strong stem-loop structure resembling a transcriptional terminator immediately following fadR. In any case, we based our conclusions about fadR on the deletion mutant.

Comparison of V. vulnificus FadR with known FadR properties.

Although the V. vulnificus fadR mutant was not a fatty acid auxotroph, it grew slower than did the wild type in rich LB-N (Table 4) and produced smaller colonies. These phenotypes were not reported for E. coli fadR strains (35), indicating that mutation of fadR in V. vulnificus has more severe phenotypic consequences. We confirmed expected fadR phenotypes, including cerulenin sensitivity and regulation of fatty acid-related genes (Table 3); moreover, the amplitudes of transcriptional regulation of fab and fad genes that we observed were within the range of those reported for E. coli (3, 10). Therefore, we are confident that our annotation of fadR to VV1_2233 is correct. Both E. coli and V. vulnificus wild-type strains grow using oleate as a sole carbon source, and fadR mutants of both species can grow using decanoate as a sole carbon source. However, wild-type V. vulnificus, but not wild-type E. coli, can grow using decanoate as a sole carbon source. A growth trend similar to that for V. vulnificus was reported for Salmonella enterica serovar Typhimurium (20). A recent study that compared various aspects of FadR proteins of several pathogens noted that the Vibrio cholerae FadR protein contains a 40-residue insertion relative to the E. coli protein (21). The amino acid sequence of V. vulnificus FadR is 85% identical to the V. cholerae sequence, and V. vulnificus FadR contains an identical 40-residue insertion relative to the E. coli protein. Because V. vulnificus FadR shares only 52% identity with the E. coli protein, there may be differences in protein function yet to be discovered.

Role of FadR in infection of mice by V. vulnificus.

To our knowledge, the results reported herein are the first to demonstrate a role for fadR in the ability of a pathogen to infect animal hosts. Most of the literature on the role of fatty acid metabolism in bacterial virulence focuses on degradation (β oxidation). For example, fadB was identified in serovar Typhimurium as a gene specifically induced during infection (30). The authors suggested that fatty acid oxidation may be an important protective mechanism against proinflammatory and toxic host fatty acids. Fatty acid degradative defects are not likely the explanation for the attenuation of the V. vulnificus fadR mutant, as our results support previous reports that fad genes are expressed at higher levels in fadR mutants than in the wild type. To our knowledge, there has been no published work on the deleterious effects of constitutive fatty acid degradation during infection. Fatty acid synthesis was identified as important for acid tolerance and virulence of the cariogenic bacterium Streptococcus mutans (11, 12). A fabM (trans-2, cis-3-decenoyl-acyl carrier protein isomerase) mutant of S. mutans grew slower than did the wild type in vitro and had several altered enzymatic activities, including reduced glycolytic capability and altered glucose-phosphotransferase system activity (11). FabM in gram-positive bacteria functions similarly to FabA in gram-negative bacteria; both enzymes can isomerize the trans-2-enoyl bond of a nascent fatty acyl chain to the cis-3 isomer during fatty acid biosynthesis (reviewed in reference 32). This isomerization is essential for diversion of the growing acyl chain into the unsaturated fatty acid synthetic pathway. Therefore, loss of this isomerase activity results in decreased unsaturated fatty acid synthesis. Since a V. vulnificus ΔfadR mutant showed decreased expression of fabA (Table 3), we were not surprised that the mutant had some of the phenotypic characteristics reported for a fabM mutant (slow growth and decreased virulence). It remains to be seen if V. vulnificus fadR mutations also have the far-reaching enzymatic defects seen in the S. mutans fabM mutant.

We expected that a fadR mutation might severely impair membrane integrity and function, but we found that sensitivity to envelope stress was not increased in the ΔfadR mutant (Table 6 and Fig. 4) and that complement resistance was not impaired in the mutant. Therefore, it seemed that the altered membrane fatty acid profile of the ΔfadR mutant (Table 5) did not result in fragile membranes. On the other hand, the mutant was defective in motility (Fig. 5). The 55% decrease in the motility of the ΔfadR mutant compared to that of the wild type was similar to that seen in a V. vulnificus strain lacking the flagellin genes flaCDE (M. S. Tucker, P. C. Thiaville, and P. A. Gulig, unpublished results). This flagellar mutant showed decreased systemic infection but not decreased local infection. Even completely nonmotile V. vulnificus mutants (ΔflaFBA-ΔflaCDE and ΔmotAB) retain the capacity to cause local infection (Tucker et al., unpublished), so the decrease in motility of the fadR mutant was not sufficient to explain the severe attenuation of infectivity.

While it can be argued that slow growth could contribute to decreased virulence, we observed that another slow-growing mutant of V. vulnificus was capable of causing wild-type infection (Table 4). Therefore, slow growth in vitro is not necessarily a predictor of attenuated virulence. On the other hand, we noted that the growth rate of the V. vulnificus ΔfadR mutant could be functionally complemented in vitro by adding the fatty acid oleate to rich broth (Fig. 6), suggesting that the growth defect may be due to decreased unsaturated fatty acid synthesis. Likewise, inclusion of oleate in the inoculum caused a significant increase in skin infection during infection of mice (Fig. 7), indicating that a defect in fatty acid biosynthesis was the main factor causing decreased infectivity in the V. vulnificus fadR mutant.

During s.c. infection, wild-type V. vulnificus thrives in the host environment, with a doubling time of 15 to 28 min in iron dextran-treated mice (44, 45). We hypothesize that this bacterium, normally a resident of shellfish and estuarine water, causes lethal opportunistic infections in humans by overwhelming the host innate immune defenses by rapidly dividing to reach a critical mass while destroying host tissues with various toxins and enzymes. Our results reported here underscore the importance of FadR for regulating unsaturated fatty acid biosynthesis during the infectious process of V. vulnificus. These findings are likely to extend to additional bacterial pathogens. Because the bacterial fatty acid system differs from the mammalian system, this has long been an attractive target for antibiotics (reviewed in reference 18). However, some compounds identified for this purpose have had the drawback of interactions with the mammalian fatty acid metabolic system. For example, cerulenin interacts with mammalian fatty acid synthases (34). There is no FadR homolog in mammals. As such, FadR may be potential target for chemotherapy for V. vulnificus and other bacterial pathogens.