Functional Genomic Characterization of Virulence Factors fromNecrotizing Fasciitis-Causing Strains of Aeromonas hydrophila

Christopher J. Grim,a Elena V. Kozlova,b Duraisamy Ponnusamy,b Eric C. Fitts,b Jian Sha,b Michelle L. Kirtley,b Christina J. van Lier,b

Bethany L. Tiner,b Tatiana E. Erova,b Sandeep J. Joseph,c Timothy D. Read,c Joshua R. Shak,c Sam W. Joseph,d,e Ed Singletary,f

Tracy Felland,g Wallace B. Baze,h Amy J. Horneman,i Ashok K. Choprab

Food and Drug Administration, Laurel, Maryland, USAa; Department of Microbiology and Immunology, University of Texas Medical Branch, Galveston, Texas, USAb;Department of Medicine, Division of Infectious Diseases, Emory University School of Medicine, Atlanta, Georgia, USAc; Department of Cell Biology and Molecular Genetics,University of Maryland, College Park, Maryland, USAd; Maryland Institute of Applied Environmental Health, School of Public Health, University of Maryland, College Park,Maryland, USAe; Doctors Hospital, Augusta, Georgia, USAf; Mercy Hospital and Trauma Center, Janesville, Wisconsin, USAg; Department of Veterinary Sciences, M. D.Anderson Cancer Center, Bastrop, Texas, USAh; Pathology and Laboratory Medical Services, VA Maryland Health Care System, Baltimore, Maryland, USAi

The genomes of 10 Aeromonas isolates identified and designated Aeromonas hydrophila WI, Riv3, and NF1 to NF4; A. dhakensisSSU; A. jandaei Riv2; and A. caviae NM22 and NM33 were sequenced and annotated. Isolates NF1 to NF4 were from a patientwith necrotizing fasciitis (NF). Two environmental isolates (Riv2 and -3) were from the river water from which the NF patientacquired the infection. While isolates NF2 to NF4 were clonal, NF1 was genetically distinct. Outside the conserved core genomesof these 10 isolates, several unique genomic features were identified. The most virulent strains possessed one of the followingfour virulence factors or a combination of them: cytotoxic enterotoxin, exotoxin A, and type 3 and 6 secretion system effectorsAexU and Hcp. In a septicemic-mouse model, SSU, NF1, and Riv2 were the most virulent, while NF2 was moderately virulent.These data correlated with high motility and biofilm formation by the former three isolates. Conversely, in a mouse model ofintramuscular infection, NF2 was much more virulent than NF1. Isolates NF2, SSU, and Riv2 disseminated in high numbersfrom the muscular tissue to the visceral organs of mice, while NF1 reached the liver and spleen in relatively lower numbers onthe basis of colony counting and tracking of bioluminescent strains in real time by in vivo imaging. Histopathologically, degen-eration of myofibers with significant infiltration of polymorphonuclear cells due to the highly virulent strains was noted. Func-tional genomic analysis provided data that allowed us to correlate the highly infectious nature of Aeromonas pathotypes belong-ing to several different species with virulence signatures and their potential ability to cause NF.

Aseminal paper regarding Aeromonas species as a cause of hu-man infection was first published in 1968 (1). Since then,

water-associated infections with aeromonads have been on therise (2, 3). Further, the high resistance of Aeromonas species toboth water chlorination, especially in biofilms, and multiple anti-biotics (4) squarely resulted in the placement of this organism onthe EPA’s Contaminant Candidate List 2 (5) and its categorizationas an emerging human pathogen. Indeed, Aeromonas species wereisolated as pure cultures from 22% of the wounds of infected pa-tients during the 2004 tsunami in southern Thailand (6). In addi-tion to causing severe wound infections, this pathogen is associ-ated with hemolytic-uremic syndrome and necrotizing fasciitis(NF) (7–11). The two earliest human cases of NF with Aeromonaswere published in 1997 (12, 13), while the latest reported case ofNF resulted from trauma and freshwater contact following a mo-torcycle accident (14).

Over the past 2 decades, several virulence factors of Aeromo-nas have been extensively characterized. For example, an aeroly-sin-related cytotoxic enterotoxin (Act) is a type 2 secretion system(T2SS)-secreted exotoxin with the ability to cause both diarrheaand severe tissue damage in the host (15–19). The toxin leads toapoptosis or necrosis of host cells, depending upon the dose (16–18, 20, 21). A functional T3SS and two potent and diverged T3SSeffectors, namely, AexT and AexU, have been identified in aero-monads (22), with GTPase-activating protein (GAP) and ADP-ribosyltransferase (ADP-RT) activities, leading to host cell death.Likewise, two T6SS effectors, hemolysin-coregulated protein(Hcp) and the valine-glycine repeat G (VgrG) family of proteins

(VgrG1,-2, and -3) have been reported and characterized (23, 24).The biological effects of translocated VgrG proteins are associatedwith their specific C-terminal extensions. In Aeromonas hydro-phila SSU (recently reclassified as Aeromonas dhakensis sp. nov.comb. nov. (25), we have shown that VgrG1 carries a C-terminalvegetative insecticidal protein 2 (VIP-2) domain with ADP-RTactivity, and its translocation into host cells leads to the ADP-ribosylation of actin, resulting in cytotoxic effects (24). Likewise,the T6SS-dependent translocation of Hcp was first shown by ourgroup in A. dhakensis SSU-infected human colonic epithelial cells,which led to caspase 3 activation and subsequent apoptosis of hostcells (23). In addition to its ability to be translocated into hostcells, the secreted form of Hcp inhibited bacterial phagocytosis(26).

Aeromonads have various quorum-sensing (QS) systems

Received 10 February 2014 Accepted 26 April 2014

Published ahead of print 2 May 2014

Editor: H. Nojiri

Address correspondence to Ashok K. Chopra, achopra@utmb.edu.

C.J.G., E.V.K., D.P., E.C.F., and J.S. contributed equally to this report.

Supplemental material for this article may be found at http://dx.doi.org/10.1128/AEM.00486-14.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

doi:10.1128/AEM.00486-14

4162 aem.asm.org Applied and Environmental Microbiology p. 4162– 4183 July 2014 Volume 80 Number 14

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

that could modulate bacterial virulence genes (27, 28). Like-wise, the ability of bacteria to swim and swarm contributes tothe overall virulence of aeromonads (29, 30). It is unclearwhether any of these virulence factors/mechanisms of Aeromo-nas species contribute to the pathogenesis of NF or whether NF isa host-mediated disease. During NF caused by group A Strepto-coccus (GAS), superantigens produced by the bacteria interactwith major histocompatibility complex II of the antigen-present-ing cells, as well as the �-chain of the T-cell receptor and thereceptor of the CD28 costimulatory molecule (31). However, suchsuperantigens have not yet been identified in aeromonads.Clearly, individuals who are somehow predisposed to NF geneti-cally, physiologically, or immunologically are at risk (32). Withthis variation in host susceptibility, exposure to a large number ofA. hydrophila bacteria of a particular pathotype in a unique envi-ronment could likely lead to a rapid, fulminant, and necrotic in-fection, requiring rapid surgical and chemotherapeutic interven-tion. To provide some answers, we performed genome-widesequencing of NF-causing strains of A. hydrophila and comparedthem with other clinical and environmental isolates of aeromon-ads to identify any unique or pathotype-specific genes that couldbe implicated in NF. In addition, we developed an NF mousemodel of infection and monitored the dissemination of bacteria inthis model in real time by in vivo imaging.

MATERIALS AND METHODSBacterial strains. The sources of the Aeromonas strains and plasmids usedin this study are listed in Table 1. Two A. hydrophila isolates successivelycultured from a patient with NF were designated NF1 and NF2. These NFstrains were recovered from the patient during and immediately afteramputation surgery (culture of stump tissue) and preserved in Trypticasesoy broth with 30% glycerol at �80°C. Following extended cultivation,two additional colony types, NF3 and NF4, were isolated from the samesheep blood agar (SBA) plate as NF2. These two colonies appeared muchsmaller than the dominant isolate, NF2.

Since major pathogenic aeromonads form beta-hemolytic coloniesrelatively similar in colony size, morphology, and color on SBA plates, it ishighly conceivable that the original culture from the amputated leg was a“mixed culture” of similar-looking Aeromonas strains, namely, A. hydro-phila NF1 and NF2. No other Gram-positive or Gram-negative organismswere isolated from the cultures. During a second culture of tissue takenlater from the amputation stump, a single colony of the persistent strain,NF2, was chosen for identification. Two environmental strains (A. jandaeiRiv2 and A. hydrophila Riv3) were isolated from the river water fromwhich the above-described patient acquired the infection. Two isolatesthat belonged to the A. caviae-media group (A. caviae NM22 and NM33)had similar DNA signatures and virulence factor-encoding genes andwere isolated, respectively, from well water (NM33) and from a child withdiarrhea (NM22) who consumed that contaminated well water (33).These isolates were confirmed in this study to be strains of A. caviae on thebasis of phylogenetic analysis of the rpoD gene (data not shown). A. hy-

TABLE 1 Strains and plasmids used in this study

Strain or plasmid Relevant characteristic(s) Source or reference

A. dhakensis CDCa

SSU-Rifr Rifr strain of A. dhakensis SSU Laboratory stockSSU-Rifr-luc Strain with Tn7-luciferase operon This studySSU �ahyRI mutant ahyRI gene deletion mutant of A. dhakensis SSU; Smr Spr Rifr Laboratory stockSSU �luxS mutant luxS gene deletion mutant of A. dhakensis SSU; Kmr Rifr Laboratory stockSSU �qseB mutant qseB gene deletion mutant of A. dhakensis SSU; Kmr Rifr Laboratory stock

A. hydrophilaATCC 7966T Canned milk isolate ATCCb,c

ATCC 7966T-Rifr-luc Rifr strain of A. hydrophila ATCC 7966T and Tn7-luciferase operon This studyNF1 Initial isolate from patient wound site This studyc

NF1-Rifr-luc Strain with spontaneous rifampin resistance and Tn7-luciferase operonNF2-NF4 Morphologically distinct isolates obtained successively from NF1 wound

site following amputation and surgical debridementThis studyc

NF2-Rifr-luc Rifr strain of A. hydrophila NF2 with Tn7-luciferase operon This studyWI Tracheal aspirate This studyc

Riv3 River water isolate This studyc

A. jandaeiRiv2 River water isolate This studyc

Riv2-Rifr-luc Rifr strain of A. jandaei Riv2 with Tn7-luciferase operon This study

A. caviaeNM22 Stool sample isolate 33c

NM33 Well water isolate 33c

P. aeruginosaPA103 ExoA-positive strain Laboratory stock�exoA mutant exoA gene deletion mutant of P. aeruginosa PA103 Laboratory stock

Plasmid pUTmini-Tn5::luxKm2 Kmr Laboratory stocka Origin, Philippines.b ATCC, American Type Culture Collection.c Origin, United States.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4163

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

drophila strain WI, resistant to multiple antibiotics (ampicillin-sulbac-tam, amoxicillin-clavulanic acid, aztreonam, cefazolin, cefuroxime, cip-rofloxacin, ertapenem, gentamicin, imipenem, and levofloxacin), wasfrom the tracheal aspirate of a patient with acute adult respiratory andkidney failure. A. dhakensis SSU was obtained from the Centers for Dis-ease Control and Prevention (CDC) and was originally obtained from apatient with diarrhea during a cholera-like outbreak in the Philippines.

Biochemical identification of Aeromonas isolates. All of the bacterialstrains were identified by conventional methods, specifically, Aerokey II(34) and additional biochemical tests (35). River water samples were col-lected and transported in Whirl-Pak bags (Nasco, Fort Atkinson, WI), and10-ml aliquots were passed through 47-mm paper filters (pore size, 0.45�m). The filters were placed on ampicillin-dextrin agar with vancomycinto select for aeromonads (this medium detects all Aeromonas species ex-cept [potentially] A. trota) (36).

Genome sequencing and annotation. Genome sequencing of A. hy-drophila NF1 and NF2 and A. dhakensis SSU was performed at the EmoryGenome Center with GS-FLX (454 Life Sequencing, Branford, CT). Thenumber of reads was 519,429 for strain NF1, 418,086 for NF2, and 261,221for strain SSU. The estimated average coverage of the NF1, NF2, and SSUgenomes was 23, 19-, and 11-fold, respectively. Genomes were assembledwith Newbler software (37). The draft genome of strain NF1 was4,809,530 bp on 170 contigs, that of NF2 was 4,789,684 bp on 154 contigs,and that of strain SSU was 4,879,288 bp on 281 contigs.

Genome sequencing of A. hydrophila Riv3 and WI, A. caviae NM22and NM33, and A. jandaei Riv2 was performed at the University of TexasMedical Branch (UTMB) Molecular Genomics Core Facility with a HiSeq1000 (Illumina, San Diego, CA) by using the 2 � 50-cycle paired-endprotocol. Trimmed reads were assembled into contigs by using AbYSS(38). The draft genome of strain WI was 5,172,023 bp on 112 contigs, thatof Riv2 was 4,478,089 bp on 43 contigs, and that of strain Riv3 was4,861,432 bp on 58 contigs. The draft genomes of strains NM22 andNM33 contained 4,419,058 bp on 250 contigs and 4,470,246 bp on 162contigs, respectively.

Genome sequencing of A. hydrophila NF3 and NF4 was performed ona MiSeq (Illumina, San Diego, CA) by using paired-end version 2 chem-istry. The number of reads was 2,169,752 for strain NF3 and 6,762,904 forstrain NF4. Paired-end FASTQ data sets were trimmed and assembledinto contigs by using CLC Genomics Workbench, version 6.0.5 (CLC bio,Aarhus, Denmark). A draft genome of strain NF3 was 4,757,900 bp on 197contigs, and that of NF4 was 4,779,984 bp on 156 contigs. The estimatedaverage coverage of NF3 and NF4 genomes was 42- and 50-fold, respec-tively. Genomic contigs were annotated by using the RAST annotationserver (39) to identify RNAs and protein-coding genes.

Comparative genomic analyses. Comparative genomic analysis wasperformed to identify shared and dispensable genetic traits among theAeromonas strains that were sequenced. In addition to the 10 genomessequenced during this study, the closed genomes of A. hydrophila ATCC7966T (GenBank accession no. CP000462.1), from a tin of milk with a fishyodor (40), and A. salmonicida A449 (GenBank BioProject PRJNA58631),from a brown trout (41), as well the draft genomes of A. hydrophila E1 andE2 (SRA063950), from a patient with a wound infection (42), A. aquari-orum AAK1 (GenBank accession no. BAFL00000000.1), recently reclassi-fied as A. dhakensis sp. nov. comb. nov. (25) and from the blood of apatient with septicemia and NF (43), A. veronii B565 (GenBank accessionno. CP002607.1), from aquaculture pond sediment (44), and A. caviaeAe398 (GenBank Whole-Genome Shotgun project no. CACP00000000)from a child with profuse diarrhea (45), were used.

To facilitate comparisons, the genomes of these other seven aeromon-ads were also annotated by using the RAST annotation server. Genome-to-genome comparisons were performed primarily with a SEED viewer(46), which uses bidirectional protein-protein BLAST (blastp) sequencecomparison of translated open reading frames. For all draft genomes,genes at the end of a contig or interrupted by contig gaps were analyzed byusing bidirectional BLASTN analysis against all other genomes. Genomic

regions (GRs), defined as regions present in one (unique) or more ge-nomes and missing from at least one other genome (dispensable), wereidentified as previously reported (42) for the genomes of A. hydrophilaATCC 7966T, WI, Riv3, NF1 to NF4, E1, and E2; A. dhakensis SSU; and A.jandaei Riv2 (47, 48). Average nucleotide identity (ANI) by BLAST wascomputed with Jspecies (49). Evolutionary analyses were conducted inMEGA5 (50). The act (cytotoxic enterotoxin) gene locus was comparedamong different Aeromonas strains by using the Artemis ComparisonTool (51).

Core cluster alignment and phylogenetic inference. The completepredicted proteome from the 16 Aeromonas genomes was searched againstitself by using BLASTP with an E value cutoff of 1e�05. The best blastscores were converted into a normalized similarity matrix with theOrthoMCL (52) algorithm, which uses an additional step of Markov Clus-tering algorithm (MCL) to improve the sensitivity and specificity of theorthologous sequences identified. Core genes were identified as the pro-tein-coding gene clusters that were shared by all of the 16 Aeromonasisolates used in this study. Multiple-sequence alignments of the core pro-tein-coding genes were generated with the program MUSCLE (53) byusing default settings. These core alignments were filtered for uninforma-tive characters by GBLOCKS (54) by using default settings. Whole-ge-nome phylogeny was determined with a concatenation of aligned individ-ual core protein sequences, followed by neighbor joining (NJ) withNEIGHBOR in the PHYLIP package (55, 56). Distances were identifiedwith the PROTDIST program within PHYLIP. The support of the data foreach of the internal node of the phylogeny was estimated by using 100bootstraps.

In vitro characterization of Aeromonas strains for various virulencetraits. On the basis of the literature and our own earlier studies, biofilmformation, swimming and swarming motility, protease production, andtoxin/effector secretion via different secretion systems are correlated withthe virulence of aeromonads in animal models. We have refined several ofthese in vitro assays to correlate gene expression with the functionality ofthe target gene to obtain reliable and quantitative data.

Crystal violet (CV) biofilm assay. To measure solid-surface-associ-ated biofilms, a modified biofilm ring assay (57, 58) was used. Biofilmformation was then quantified (59), and the results were normalized to1 � 109 CFU to account for any minor differences in the growth rates ofthe various bacterial strains used. The CFU count was determined by serialdilution of the samples, followed by plating (33, 58).

Motility assay. Luria-Bertani (LB) medium with 0.35% Difco Bactoagar (Difco Laboratories, Detroit, MI) was used to characterize swimmingmotility, while Difco nutrient broth with 0.5% Eiken agar (Eiken Chem-ical Co., Ltd., Tokyo, Japan) was used to measure the swarming motility ofaeromonads (30, 42, 58). These methods provided better quantification ofthe motility data.

Measurement of protease activity. Protease activity was measured inculture filtrates of Aeromonas isolates grown overnight (60). The hidepowder azure substrate was used to measure protease activity (42). Pro-tease activity was calculated per milliliter of culture filtrate per 108 CFU asdetermined by serial dilution of the samples and colony counting.

Measurement of hemolytic activity. To measure hemolytic activityassociated with Act, the culture filtrates from various Aeromonas isolateswere first treated with trypsin (final concentration, 0.05%) at 37°C for 1 hand then subjected to a hemolytic activity assay with rabbit erythrocytes(61). Hemolytic activity titers were calculated as the absorbance at 540 nmof the hemoglobin release multiplied by the dilution factor of the culturefiltrates. Units of hemolytic activity were reported per milliliter of cellfiltrate/1 � 108 CFU (58). Although Act is active against erythrocytes ofother animal and human species, rabbit red blood cells are more sensitiveto the action of Act and hence were used in this study.

For the neutralization assay, culture filtrates of the strains studied weremixed with either preimmune (control) or hyperimmune rabbit serum(laboratory stock, 1:10 dilution) containing antibodies to Act (42, 62) andincubated at 37°C for 1 h before use.

Grim et al.

4164 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

Lactone production. N-Acyl-homoserine lactone (AHL) productionwas detected by cross-streaking various Aeromonas strains against the bio-sensor Chromobacterium violaceum CV026 on LB agar plates (27, 42).Violet pigment production from C. violaceum CV026 by AHLs producedby Aeromonas isolates was scored on the basis of color intensity afterovernight incubation of the plates at 37°C. The result was recorded as nolactone production (�) or a low (�), moderate (��), or high (���)level of AHLs. An isogenic ahyRI mutant of A. dhakensis SSU (Table 1) wasused as a negative control because of its inability to produce AHLs (62).

Western blot analysis. Western blot analysis was used to detect theproduction and secretion of Hcp (a T6SS effector), AexU (a T3SS effec-tor), and exotoxin A (ExoA) by various Aeromonas isolates. Briefly, thesupernatants and cell pellets from cultures grown overnight were sepa-rated and the pellets were directly lysed in SDS-PAGE loading buffer. Theproteins in the supernatants were first precipitated with 10% (vol/vol)trichloroacetic acid and then dissolved in the loading buffer before beingsubjected to SDS-PAGE. For Western blot analysis, specific antibodies toHcp and AexU (available in the laboratory) (22, 26, 63) and to ExoA ofPseudomonas aeruginosa (LSBio, Inc., Seattle, WA) were used. At theamino acid level, ExoA of A. hydrophila is 65% homologous to that of P.aeruginosa (64). The bacterial strains were grown overnight in the LBmedium at 37°C. The yield of ExoA produced by bacteria was influencedby the concentration of iron in the culture medium (65, 66).

Generation of bioluminescent Aeromonas strains. The reporterstrains were generated following triparental conjugation of rifampin-re-sistant (Rifr) Aeromonas isolates with Escherichia coli SM10 �pir carryingthe pTNS2 plasmid and SM10 harboring the pUC18-mini-Tn7T-Km2-lux plasmid (67). The minitransposon system contains a lux luminescenceoperon with the native promoter and a kanamycin resistance (Kmr) cas-sette for transposon selection. This system allows site-specific transposi-tion downstream of the glmS gene, which encodes a conserved glucosa-mine-6-phosphate synthetase, with the helper plasmid pTNS2 providingthe transposase complex (67). The mutants were screened for biolumi-nescence by using an ImageQuant LAS4000 bioluminescence and fluores-cence imaging workstation (GE Healthcare Sciences, Pittsburgh, PA). Thereporter strains that were oxidase positive and emitted bioluminescencesignals were selected for further characterization. Insertion sites were con-firmed by PCR with primers PTn7R:5=-CACAGCATAACTGGACTGATTTC-3= and GlmSFwd: 5=-GCCAGTATCCCATTGCCATG-3=, which,respectively, corresponded to the 5= end of the Tn7 minitransposon andthe 3= end of the glmS gene in Aeromonas, followed by DNA sequencing.Rifr Aeromonas strains were generated by spontaneous mutation in re-sponse to antibiotic selection (200 �g/ml) to aid in the selection of trans-poson mutants.

Animal experiments. (i) Septicemic-mouse model of infection. In afirst set of experiments, groups (n 6 to 23) of healthy female SwissWebster mice (Taconic Farms) were infected via the intraperitoneal (i.p.)route with various aeromonads. The animals were infected with a dose of5 � 107 CFU, and deaths were recorded for 14 days postinfection (p.i.).Before each study, a pilot experiment was performed with three differentdoses of SSU (8 � 106, 2 � 107, and 5 � 107 CFU). The dose chosen wasthat which was 100% lethal. For infection studies, Aeromonas culturesgrown overnight were centrifuged and the pellets were washed three timesin sterile phosphate-buffered saline (PBS) before being suspended in 1/10of the original culture volume of PBS. Subsequently, each culture wastitrated and inocula were prepared such that a 50-�l volume contained theintended infectious dose of the organism.

(ii) Mouse model of i.m. infection to mimic NF. Mice (n 5/group)were anesthetized with isoflurane, 50 �l of various Aeromonas cultureswas injected intramuscularly (i.m.) into one of the legs at doses of 5 � 106

to 5 � 108 CFU, and deaths were recorded for 14 days p.i. The animalswere also observed for the development of possible necrotic lesionsaround the injection site. The animals used for the time point studies wereeuthanized by carbon dioxide narcosis, followed by cervical dislocation.

Evaluation of bacterial dissemination after i.m. infection. Animals(n 5) infected as described above were euthanized at 24 or 48 h p.i. Fromeach animal, the spleen and liver were collected aseptically and transferredto disposable tissue grinders (Fisher Scientific, Pittsburgh, PA) containing1 ml of PBS for the spleen and 2 ml for the liver. These tissues werehomogenized and serially diluted, and aliquots were subjected to bacterialcounting after incubation on LB agar plates at 37°C for 24 h.

Gross pathological examination of lesions and processing of muscletissues for histopathological analysis. At the time of collection of inter-nal organs for CFU determination, as well as from those mice that sur-vived for longer times, the lesions associated with NF were grossly exam-ined. In addition, leg muscle tissues adjacent to the injection site werecollected and fixed in 10% buffered formalin. After 48 h of fixation, thetissues were processed and sectioned at 5 �m before staining with hema-toxylin and eosin (H&E). The tissue sections were evaluated by light mi-croscopy in a blinded fashion.

IVIBB in mice. As we sought to characterize the i.m. model of mouseinfection to mimic NF, we used in vivo imaging of bioluminescent bacteria(IVIBB) to evaluate colonization and dissemination of bacteria to periph-eral organs. Consequently, strains SSU, ATCC 7966T, NF1, and Riv2 har-boring the Tn7 minitransposon system with a lux operon were used.

Animals (n 5) infected by the i.m. route with various biolumines-cent Aeromonas strains were lightly anesthetized with isoflurane (24- and48-h time points) at the indicated doses and subjected to in vivo imagingon an IVIS 200 bioluminescence and fluorescence whole-body imagingworkstation (Caliper Life Sciences, Alameda, CA). At the same timepoints, mice were sacrificed after imaging and their livers, spleens, and legmuscles were subjected to CFU determination.

Finally, quantitation of radiance by regions of interest (ROIs), corre-sponding to the anatomical regions of liver, spleen, and leg muscle, wasaccomplished by using Living Image software (Caliper Life Sciences).ROIs of the same shape and area were replicated across images measuringtotal flux (/s) to determine radiance as we previously described (68).

Statistical analysis. All of the experiments were performed in tripli-cate, and statistical significance was analyzed by one-way analysis of vari-ance (ANOVA). All of the animal data were subjected to Kaplan-Meiersurvival estimates, and a P value of �0.05 was considered significant.

Accession numbers. The whole-genome shotgun projects and theirassociated short-read archive (SRA) files for the genomes described in thisstudy have been deposited at NCBI under the following Biosample num-bers: A. hydrophila WI, SAMN02597475; A. hydrophila NF1, SAMN02597476; A. hydrophila NF2, SAMN02597477; A. hydrophila NF3, SAMN02597478; A. hydrophila NF4, SAMN02597479; A. hydrophila Riv3, SAMN02597480; A. dhakensis SSU, SAMN02597481; A. jandaei Riv2, SAMN02597482; A. caviae NM22, SAMN02597483; and A. caviae NM33, SAMN02597484.

RESULTSGenome-wide comparisons of various Aeromonas species. FourA. hydrophila isolates, NF1 to NF4, two environmental river waterisolates, Riv2 and Riv3, tracheal aspirate isolate WI of A. hydro-phila, and diarrheal isolate SSU of A. dhakensis were sequenced,assembled, and annotated. Likewise, A. caviae NM22 (stool) andNM33 (water) isolates were sequenced and used for comparisonanalysis along with sequences from two recently described A. hy-drophila strains, E1 and E2, from a case of human wound infection(42, 69). Additional comparisons were made with a number ofpreviously published Aeromonas genomes (33, 40, 44, 45, 70).

ANI comparisons. Pairwise ANI comparisons of the NF-caus-ing strains revealed that NF2 to NF4 were very similar to eachother, compared to the general level of ANI between isolates thesame species (Table 2). However, NF3 and NF4 appeared to bemore similar to each other than to NF2 (Table 2), and in general,the values were below what we would expect for “true” clonal

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4165

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

relationships, i.e., identical strains. The ANI values are sensitive todifferences in the sequencing and assembly technologies used, i.e.,Roche 454 GS Junior versus Illumina MiSeq, as well as differencesin coverage and the number of assembled contigs (which probablyreflects differences in the number of repeat elements betweenstrains). For example, the genome of strain NF3 was composed of197 contigs, while those of NF2 and NF4 contained 154 and 156contigs, respectively. Differences in gene content, measured bypairwise core genome size (Table 2), also suggested that the ge-nomes of NF2 to NF4 were highly similar. Indeed, the core ge-nome of these three strains (NF2 to NF4) contained 3,964 codingsequences (CDSs) at the 100% identity level, while the total coregenome contains 4,230 CDSs. Visual inspection confirmed thatmany genes below 100% identity were the result of single-nucle-otide polymorphisms (SNPs) altering the reading frame and trun-cating or lengthening the translated product, which is a commonerror found in pyrosequencing.

In order to correct for this problem, we built a robust tree byusing only the 2,541 conserved full-length protein sequences (Fig.1). This tree suggested that while isolates NF2 to NF4 were closelyrelated, NF2 was a significantly more distant relative. In 696,126positions in the concatenated protein alignment, there were 63amino acid substitutions between NF3 and NF4, whereas there

were 297 and 320 substitutions between NF2 and NF3 and be-tween NF2 and NF4, respectively. The high numbers of substitu-tions may be, to some extent, an artifact of the use of differentsequencing technologies (454 for NF2 and MiSeq for NF3 andNF4). However, this level of variation suggests that the threestrains had a last ancestor in common well before they infected thepatient or coinhabited the environmental niche directly beforeinfection. For comparison, a recent simulation of within-hostvariation of the 2.9-Mbp Staphylococcus aureus genome showedthat a diversity of four or five SNPs per genome should be expectedfrom long-term clonal expansion with a constant population sizeof 5,000 (71).

Strain NF1 was cultured from the initial surgical site and wasdistinctly different from the NF2 to NF4 group, supported bypairwise ANI values, gene content comparisons (Table 2), and thecore protein phylogeny (Fig. 1). Interestingly, all four isolates(NF1 to NF4) were of the same molecular serotype on the basis ofcomparisons of the gene content and similarity of CDSs locatedbetween the waaL and rmlA genes. Further, the O-antigen regionof A. dhakensis SSU was identical in terms of gene content to NF1to NF4 but divergent in nucleotide identity at a level comparableto that of the overall genome nucleotide identity between these

TABLE 2 ANIs and shared gene contents in pairwise comparisons of Aeromonas genomes

Strain (total no. ofCDSs)

ANI (%) or no. of CDSs shareda with strain:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1. A. hydrophila ATCC7966T (4,279)

96.78 96.65 96.96 96.92 96.96 96.78 97.12 96.86 92.99 86.48 84.76 85.5 86.45 85.64 85.66

2. A. hydrophila WI(4,660)

3,986 96.31 96.84 96.81 96.86 96.63 96.95 96.71 92.78 86.34 84.64 85.29 86.29 85.55 85.57

3. A. hydrophila Riv3(4,358)

3,984 3,975 96.92 96.9 96.92 96.92 96.99 96.97 92.95 86.42 84.64 85.42 86.38 85.61 85.63

4. A. hydrophila NF2(4,366)

3,935 3,946 3,923 99.95 99.98 96.92 97.02 96.81 92.83 86.36 84.79 85.39 86.52 85.66 85.69

5. A. hydrophila NF3(4,373)

3,903 3,923 3,900 4,253 99.96 96.86 96.98 96.76 92.78 86.24 84.75 85.43 86.43 85.67 85.67

6. A. hydrophila NF4(4,375)

3,938 3,953 3,933 4,285 4,296 96.91 97.02 96.81 92.78 86.31 84.74 85.33 86.41 85.6 85.6

7. A. hydrophila NF1(4,361)

3,912 3,951 3,923 4,021 3,984 4,013 96.83 96.77 92.85 86.32 84.67 85.45 86.33 85.59 85.6

8. A. hydrophila E2(4,241)

3,897 3,867 3,901 3,835 3,804 3,831 3,807 96.88 92.98 86.35 84.83 85.36 86.39 85.67 85.69

9. A. hydrophila E1(4,373)

3,895 3,892 3,917 3,955 3,922 3,950 3,979 3,814 92.8 86.32 84.71 85.34 86.42 85.62 85.64

10. A. dhakensis SSU(4,527)

3,759 3,809 3,765 3,835 3,820 3,842 3,850 3,681 3,816 85.88 84.54 85.14 86.47 85.65 85.68

11. A. salmonicidaA449 (4,570)

3,564 3,606 3,561 3,626 3,602 3,630 3,639 3,475 3,598 3,605 82.97 84.17 84.19 83.55 83.62

12. A. jandaei Riv2(4,107)

3,471 3,472 3,455 3,518 3,500 3,524 3,540 3,392 3,500 3,475 3,366 88.76 83.53 82.78 82.81

13. A. veronii B565(4,045)

3,402 3,426 3,373 3,378 3,363 3,386 3,420 3,308 3,352 3,407 3,290 3,433 83.74 82.8 82.82

14. A. caviae Ae398(4,043)

3,339 3,432 3,361 3,334 3,319 3,337 3,369 3,275 3,324 3,340 3,285 3,124 3,192 97.17 97.18

15. A. caviae NM33(4,172)

3,318 3,362 3,333 3,310 3,291 3,316 3,331 3,262 3,334 3,265 3,266 3,110 3,119 3,600 99.94

16. A. caviae NM22(4,134)

3,300 3,349 3,316 3,298 3,283 3,305 3,310 3,250 3,318 3,251 3,249 3,093 3,103 3,588 3,966

a ANIs (upper right) and shared gene contents (lower left) revealed by pairwise comparisons are shown. Values in bold indicate that the strains are of the same species on the basisof a threshold of 95.0%.

Grim et al.

4166 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

two groups of organisms (see Fig. S1 in the supplemental mate-rial).

Pairwise ANI comparisons of all of the A. hydrophila genomesexamined in this study revealed a heterogeneous collection of iso-lates of this species (Table 2). This is not surprising, given thatsome strains were from environmental sources, while clinical iso-lates were from different patients, sites, and sources. However, it isinteresting that clinical isolates from different patients and diseasesyndromes were just as divergent from environmental isolates asclinical isolates were, according to ANI. Genomic comparison ofgene contents (Table 2), on the other hand, revealed that the ge-nomes of clinical strains had more genes in common than did aclinical isolate and an environmental isolate (see below).

Core and pangenome. In terms of a core genome, amongseven A. hydrophila strains (NF2 was used to represent NF2 toNF4); we estimated a core genome of 3,617 CDSs. In terms of coregenome stability, on average, 110 CDSs were subtracted from thecore genome upon the addition of each A. hydrophila genome (seeFig. S2 in the supplemental material). Variation from this meanwas high because of a large number of dispensable genes, as well asthe geotemporal relatedness of strains from the same patient, suchas NF1 to NF2 and E1 to E2. As expected, the core genome sizedecreased sharply when a genome from a different species wasincluded in the analysis (see Fig. S2). To identify genetic featuresrelated to or indicative of virulent pathotypes, we identified dis-pensable and unique GRs from 11 sequenced genomes (9 of A.hydrophila, 1 of A. dhakensis, and 1 of A. jandaei) and determinedthe distribution of these features (Table 3; see Table S1 in thesupplemental material).

Surface appendages. Like those of other Gram-negative bac-teria, Aeromonas genomes harbored genes for an assortment offimbriae and pili. All of the genomes harbored the Tap type IVpilus (TFP) locus, whose pilin and subunit gene clusters were un-der considerably stronger evolutionary pressure than those fromother Tap TFP gene clusters or the overall genomes (see Fig. S3 inthe supplemental material). All of the genomes also containedgenes for the mannose-sensitive hemagglutinin (MSHA) TFP lo-cus (Table 3). The genome of A. salmonicida A449 was found to bemissing a seven-gene internal cluster, mshNEGF2BAC. The ge-nomes of both A. caviae NM22 and NM33 revealed a deletion ofthe majority of the msh locus genes because of the insertion of amobile element, resulting in a truncated gene arrangement ofmshH-acd-(mobile element)-mshDOPQ (Table 3).

The tight adherence (TAD)/fimbrial low-molecular-weightprotein (Flp) pilus was present in A. hydrophila ATCC 7966T, aswell as in A. salmonicida A449 and A. veronii B565. Three fimbriaewere widely distributed among Aeromonas genomes, two � (class5) alternate chaperone-usher fimbriae and one � (P) chaperone-usher fimbria. Additionally, the genome of A. hydrophila Riv3 con-tained a second � fimbria that was encoded on a prophage ele-ment (Table 3).

All Aeromonas genomes contained five genetic loci encoding asingle polar, unsheathed flagellum, with one exception. The ge-nomes of A. caviae NM22 and NM33 did not contain the flaABGfliDS locus, which encodes flagellins (flaABG), the cap protein(fliD), and the essential fliS gene product (Table 3). Accordingly,these two strains were nonmotile (data not shown). A 35-kb lat-eral flagellar gene cluster was present in the genomes of NF1 to

FIG 1 NJ tree of strains built by using 2,514 conserved full-length predicted proteins concatenated together. The distance was calculated with PROTDIST (55,56). The scale bar shows 0.001 substitution per site. There was strong bootstrap support for the tree, with all but three branches having 100/100 replicatecongruence. The three branches with 50% but �100% support were all internal branches.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4167

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

TABLE 3 Distribution of GRs of A. hydrophila and A. dhakensis strains among Aeromonas species genomes

Virulence trait(s)

Presence in or absence from strain:

ATCC 7966T WI Riv3 NF2 NF3 NF4 NF1 E2 E1 SSU A449 Riv2 B565 Ae398 NM33 NM22

Surface appendages and featuresCFA/I (� C/U) fimbriae, AHA_0060 � � � � � � � � � � � � � � � �CFA/I (� C/U) fimbriae, AHA_1021 � � � � � � � � � � � � � � � �P (� C/U) fimbriae, AHA_0521 � � � � � � � � � � � � � � � �Phage-encoded P (� C/U) fimbriae � � � � � � � � � � � � � � � �Tap type IV pilus � � � � � � � � � � � � � � � �MSHA BFP type IV pilus, mshaH acd

mshaIIJK3LM_DOPQ� � � � � � � � � � � � � � � �

mshaNEGF2BAC � � � � � � � � � � � � � � � �TAD Flp pilus � � � � � � � � � � � � � � � �Polar flagellum � � � � � � � � � � � � � � � �flaABG-fliDS cluster, AHA_1698–AHA_1702 � � � � � � � � � � � � � � � �Lateral flagella � � � � � � � � � � � � � � � �

Putative virulence factors/toxinsCytotoxic enterotoxin/hemolysin, act � � � � � � � � � � � � � � � �Cytotonic enterotoxin, ast � � � � � � � � � � � � � � � �Pore-forming cytolysin/hemolysin, hlyA � � � � � � � � � � � � � � � �TPS hemolysin � � � � � � � � � � � � � � � �Cytotonic enterotoxin/lipase, alt � � � � � � � � � � � � � � � �Enolase � � � � � � � � � � � � � � � �Elastase � � � � � � � � � � � � � � � �Phospholipase/lecithinase/hemolysin

(GCAT)� � � � � � � � � � � � � � � �

Extracellular protease precursor,AHA_2712–AHA_2714

� � � � � � � � � � � � � � � �

ExoA homolog � � � � � � � � � � � � � � � �syp exopolysaccharide � � � � � � � � � � � � � � � �Group II capsule � � � � � � � � � � � � � � � �RTX toxin, AHA_1359, transporter cluster � � � � � � � � � � � � � � � �FHA family, RTX toxin � � � � � � � � � � � � � � � �Ferrichrome iron uptake cluster,

AHA_1951–AHA_1954� � � � � � � � � � � � � � � �

Ferric hydroxamate uptake (fhu) operon � � � � � � � � � � � � � � � �Pyoverdine siderophore operon,

ASA_4368–ASA_4378� � � � � � � � � � � � � � � �

Ferric siderophore-TonB cluster,AHA_3433–AHA_3439

� � � � � � � � � � � � � � � �

T3SSs and T6SSshcp (T6SS) � � � � � � � � � � � � � � � �T3SS � � � � � � � � � � � � � � � �yopH and chaperone (T3SS) � � � � � � � � � � � � � � � �aexTU (T3SS) � � � � � � � � � � � � � � � �yopT and chaperone (T3SS) � � � � � � � � � � � � � � � �

Resistance, including antimicrobialsABC-type multidrug transport,

AHA_0484–AHA_0486� � � � � � � � � � � � � � � �

MATE pump, AHA_3203–AHA_3204 � � � � � � � � � � � � � � � �DMT efflux, AHA_3820–AHA_3821 � � � � � � � � � � � � � � � �Multidrug resistance protein B,

AHA_4116–AHA_4117� � � � � � � � � � � � � � � �

Acriflavin RND transporter,AHA_1320–AHA_1323

� � � � � � � � � � � � � � � �

NodT family RND efflux pump � � � � � � � � � � � � � � � �RND efflux pair, AHA_2959–AHA_2960 � � � � � � � � � � � � � � � �Macrolide efflux pump, macAB,

AHA_0758–AHA_0761� � � � � � � � � � � � � � � �

Tripartite MRS � � � � � � � � � � � � � � � �Polymyxin B resistance (arn) � � � � � � � � � � � � � � � �cepS, cephalosporinase � � � � � � � � � � � � � � � �

(Continued on following page)

Grim et al.

4168 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

NF4, E1, SSU, A449, and A. jandaei Riv2, all of which were highlypathogenic species or strains (Table 3). The gene content of thislocus was highly conserved among Aeromonas species, and thegene cluster had the same chromosomal location in all of thestrains; downstream from the gene encoding rRNA small-subunitmethyltransferase I (RsmI, AHA_3894). Phylogenetic reconstruc-tion of the evolutionary history of this locus argues for verticaltransmission (see Fig. S4 in the supplemental material).

Toxins. Several other virulence factor-encoding genes wereidentified in the genomes analyzed and included lipases, hemoly-sins, siderophore clusters, elastase, cytotonic/cytotoxic enterotox-ins, T3SS, and the capsule polysaccharide-encoding loci (Table 3).The distribution of three of these factors, act (cytotoxic entero-toxin), exoA (ExoA), and a T3SS, correlated with virulent patho-types. The act gene was present in ATCC 7966T, WI, NF1 to NF4,E1, SSU, A449, Riv2, and B565 but absent from all three of the A.caviae genomes and Riv3 (Table 3).

Closer inspection of the act genetic locus revealed variability interms of gene content (Fig. 2A). For example, in the genomes ofNF2 to NF4, a gene encoding a putative lipoprotein and a group 1glutathionylspermidine synthase (EC 6.3.1.8) was inserted down-stream of the act gene. Additionally, the act gene was truncated inA. hydrophila NF2 to NF4 (567 bp), as opposed to other A. hydro-phila and A. dhakensis act gene sequences, which were 1,482 bp inlength (Fig. 2A). A putative arylsulfate sulfotransferase-encodinggene was inserted in this locus in the genomes of A. hydrophilaRiv3, WI, and E2, of which only A. hydrophila WI possessed the actgene (Fig. 2A). Additionally, the act locus of Riv2 was approxi-mately 600 bp longer than that of strain ATCC 7966T because ofthe insertion of a 516-bp MerR family transcriptional regulator-

encoding gene between the act and thiC genes (data not shown).These alterations in the act gene locus might affect the expressionof the gene in various Aeromonas isolates. The sizes of the act genesfrom other species of Aeromonas were 1,479 bp for A. salmonicidaand 1,464 bp for A. jandaei Riv2 and A. veronii B565. Despite thesefindings, phylogenetic analysis of the act locus supported the hy-pothesis that this gene was vertically transmitted in Aeromonasgenomes (Fig. 2B and C).

Five genomes (A. hydrophila NF2 to NF4 and E1 and A. dhak-ensis SSU) contained a gene encoding an ExoA homologue, whichis a NAD-dependent ADP-RT commonly associated with patho-genic Pseudomonas species (Table 3). The toxin leads to ADP-ribosylation of host elongation factor 2, resulting in the shutdownof protein synthesis and cell death (72). Importantly, this feature isa distinguishing trait among the three NF isolates, namely, NF2,NF3, and NF4. All of these isolates had this gene (100% identityamong the isolates); however, this gene was absent from the ge-nome of NF1. The exoA gene was also present in the genome of ahighly virulent E1 strain isolated from a case of wound infectionand a diarrheal A. dhakensis isolate, SSU (Table 3).

T3SSs, T6SSs, and their effectors. The genomes of eight Aero-monas strains harbored a T3SS and included A. hydrophila NF1 toNF4 and E1, A. dhakensis SSU, A. salmonicida A449, and A. jandaeiRiv2 (Table 3). The T3SS was inserted at the same locus, a serinetRNA (TGA) (at nucleotide positions 2983703 to 87 of A. hydro-phila ATCC 7966T) (40), in all of the genomes of A. hydrophila andA. dhakensis. The T3SS is located on plasmid pAsa5 in A. salmoni-cida A449 and is flanked by transposases, indicating horizontalacquisition, while the T3SS was inserted between genes encodinga short-chain-specific acyl coenzyme A dehydrogenase (EC

TABLE 3 (Continued)

Virulence trait(s)

Presence in or absence from strain:

ATCC 7966T WI Riv3 NF2 NF3 NF4 NF1 E2 E1 SSU A449 Riv2 B565 Ae398 NM33 NM22

ampS(H), class D beta-lactamase � � � � � � � � � � � � � � � �ompK-ampG � � � � � � � � � � � � � � � �imiS(H), class 3 metallo-�-lactamase � � � � � � � � � � � � � � � �LPS modification, cephalosporin

hydroxylase, AHA_4152–AHA_4169� � � � � � � � � � � � � � � �

LPS modification, aminoglycoside 3=-N-acetyltransferase,AHA_4170–AHA_4182

� � � � � � � � � � � � � � � �

Aminoglycoside 6=-N-acetyltransferase � � � � � � � � � � � � � � � �Mercury, arsenic, heavy metal resistance

tnp� � � � � � � � � � � � � � � �

Heavy metal resistance tnp, TEM-1beta-lactamase

� � � � � � � � � � � � � � � �

Aminoglycoside 3=-PTS � � � � � � � � � � � � � � � �Chloramphenicol acetyltransferase,

mercury resistance, OXA-1 beta-lactamase transposon

� � � � � � � � � � � � � � � �

Spectinomycin 9-O-adenylyltransferase,sulfonamide, OXA beta-lactamase

� � � � � � � � � � � � � � � �

Macrolide 2=-PTS � � � � � � � � � � � � � � � �Tetracycline � � � � � � � � � � � � � � � �Integrative conjugative element (46 kb),

organic solvents� � � � � � � � � � � � � � � �

Quaternary ammonium compound,ethanolamine, propanediol

� � � � � � � � � � � � � � � �

Organic hydroperoxide resistance,AHA_3609–AHA_3611

� � � � � � � � � � � � � � � �

Accessory arsenic resistance operon � � � � � � � � � � � � � � � �

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4169

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

1.3.99.2) involved in isoleucine degradation and a cystathioninebeta-lyase (EC 4.4.1.8) in the genome of A. jandaei Riv2.

In terms of gene content, the 26.7-kb T3SSs (hypotheticalascUTSRQPO ascN aopN acr12 ascXY ascV acrRGVH aopBDaxsCBAD ascCDEFGHIJKL) of the A. hydrophila strains analyzedin this study were identical (Table 3). The T3SS of A. dhakensisSSU was slightly larger, approximately 27.5 kb, because of theinsertion of an aopX effector protein-encoding gene downstreamof ascU (which is transcribed in the opposite direction). This genewas also present in the T3SS of A. salmonicida A449. The T3SS ofA. jandaei Riv2 was considerably larger, at 30 kb, having accessorygenes at both ends of the gene cluster. There was a 1,182-bp zincmetalloprotease-encoding gene located downstream of ascU andan exoU effector protein homologue and chaperone locateddownstream of the ascL gene.

Three putative T3SS effector protein gene clusters wereidentified outside the T3SS locus (Table 3). The aexTU effector-chaperone pair was inserted between a tryptophan-specific-per-mease-encoding gene and a phenylalanyl-tRNA synthetase betasubunit-encoding gene (AHA_0123 and AHA_0124, respec-tively). This effector protein was the most widely distributed of the

three. Also present in this three-gene cluster was a homologue ofthe gene for Txp40, a 40-kDa insecticidal toxin. Interestingly, thisgene cluster was present in the genome of A. hydrophila Riv3, eventhough the main T3SS gene locus was absent.

The yopH sycH effector-chaperone pair (Table 3) was locatedbetween a hexose phosphate transport protein (UhpT)-encodinggene and a phosphotransferase system (PTS) fructose-specificIIABC component-encoding gene. Its distribution was more re-stricted; it was present in the five strains of A. hydrophila with aT3SS locus and A. salmonicida A449 (Table 3). The genome of A.jandaei Riv2 also contained a yopT sycT effector-chaperone genepair (Table 3) inserted between a siroheme synthase-encodinggene and a magnesium transporter-encoding gene (homologuesof AHA_4121 and AHA_4122, respectively). Additionally, the ge-nome of A. salmonicida A449 contained a yopO effector homo-logue and a putative chaperone-encoding gene pair.

In contrast to the T3SS, a gene cluster encoding a T6SS wasmore widely distributed; it was present in 12 of the 16 Aeromonasgenomes that were compared (Table 3). These included the ge-nomes of all nine A. hydrophila isolates, A. dhakensis SSU, A. sal-monicida A449, and A. jandaei Riv2. All of the genomes that har-

FIG 2 (A) Genetic organization of the act gene locus from Aeromonas species. Gene locus schematics are drawn to relative scale. Genes are depicted as follows.Gray arrows represent conserved flanking genes, i.e., that for the hypothetical protein FIG 00361276 (5= or upstream) and thiC (3= or downstream). Diagonallystriped arrows represent the act gene. The white arrow represents hypothetical protein-encoding genes (nonhomologous). The vertically striped arrows representputative arylsulfate sulfotransferase-encoding genes. The horizontally striped arrow represents the gene for a putative glutathionylspermidine synthase, group 1(EC 6.3.1.8). The regions between the gene schematics show homology, with darker red representing higher homology and lighter red indicating regions of lowerhomology, while colorless regions are not homologous and indicate insertions or deletions. The gene locus of strain ATCC 7966T is also representative of SSU,NF1, and E1, while Riv3 is also representative of E2. (B) Phylogenetic relationship of the act gene locus, including conserved flanking genes, i.e., that forhypothetical protein FIG 00361276 (5= or upstream) and thiC (3= or downstream), act, and other gene insertions, in Aeromonas. (C) Phylogenetic relationshipof the act gene. The evolutionary reconstructions were inferred by using the NJ method. The bootstrap consensus tree was inferred from 1,000 replicates.Evolutionary distance is shown as the number of nucleotide differences.

Grim et al.

4170 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

bored the cluster had an 18-CDS region, impB to vgrG, incommon. The 5= region of the T6SS cluster was more variable,with most of the genomes containing accessory genes in compar-ison to ATCC 7966T, WI, and Riv3. Among the A. hydrophila andA. dhakensis genomes, the cluster was located between two tRNAgenes, tRNA-Ser-GGA (AHA_1825) and tRNA-His-GTG(AHA_1851). The T6SS cluster was also adjacent to tRNA-His-GTG in the genome of A449, but there was a rearrangement in the5= region adjacent to the cluster. On the contrary, the cluster wasinserted at a different chromosomal location in the genome of A.jandaei Riv2; however, its exact location could not be deducedbecause of a repetitive DNA (vgrG) sequence present at both endsof the gene cluster, which led to its assembly into a distinct contig.

In the genome of A. hydrophila ATCC 7966T, a single hcp geneand two vgrG genes were present in this main cluster. A smalleraccessory T6SS-related gene cluster, at a separate chromosomallocus, contained a second hcp gene and one additional vgrG gene.This number of T6SS effector protein-encoding genes was appar-ently conserved among other Aeromonas genomes; however, be-cause of the high homology among paralogues of each type ofeffector-encoding gene, the exact gene content could not alwaysbe concluded. The genomes of A. veronii B565 and A. caviaeAe398, NM22, and NM33 did not contain an effector-encodinggene, hcp or vgrG.

Resistance to antimicrobials. The cepS(H) cephalosporinasegene was absent only from the genome of A. veronii B565, and theimiS(H) gene was missing only from the three A. caviae genomes(NM22, NM33, and Ae398) (Table 3). Two different aminoglyco-side-modifying enzymes were found in some genomes of A. hy-drophila, namely, ATCC 7966T, Riv3, E1, and E2, while severalAeromonas genomes contained at least one of these. The genomeof A. hydrophila strain WI also possessed a number of modifyingenzymes that contributed to antibiotic resistance (Table 3).Among these were genes that imparted resistance to aminoglyco-sides, chloramphenicol, streptomycin, sulfonamides, macrolides,tetracycline, ampicillin, and oxacillin. At least one of the oxacillin�-lactamases appeared to be an extended-spectrum �-lactamase(Table 3). Most of these genes were flanked by mobile-elementgenes.

In addition, several clusters of genes that impart resistance

to heavy metals, quaternary ammonium compounds, organichydroperoxide, and organic solvents were also identified, aswell as genes for several putative multidrug efflux systems, in-cluding members of the RND, ABC, MATE, and MFS super-families (Table 3).

Other dispensable GRs. In addition to the genomic featuresassociated with virulence or survival in the clinical setting, therewere a number of other dispensable GRs in A. hydrophila and A.dhakensis genomes (see Table S1 in the supplemental material). Anumber of these features were known operons involved in thecatabolism or transport of specific substrates, such as N-acetylmu-ramic acid, N-acetylgalactosamine, thiamine, malate, arabinose,DL-lactate, L-cystine, taurine, sucrose, tungstate, N-ribosylnicoti-namide, tetrathionate, phosphonate, and L-fucose (see Table S1).However, many of the dispensable features contained mostly hy-pothetical-protein-encoding genes, with perhaps one or two oth-erwise annotated genes. In general, the distribution of many ofthese dispensable genomic features correlated with species delin-eation. For example, many GRs were not present in the three A.caviae strains. Additionally, each genome contained a number ofunique and, in rare cases, homologous prophage and prophage-like elements (data not shown). Of interest was the presence of asmall integron, 8 to 13 kb, in the genomes of the four NF A.hydrophila strains, E1, and A. jandaei Riv2, of the zona occludenstoxin prophage type (see Table S1).

In vitro evaluation of virulence characteristics. We focusedour studies on NF-causing strains (NF1 and NF2) of Aeromonas,as the mechanisms associated with the pathogenesis of NF arepoorly understood. Among the environmental isolates, we stud-ied two that were from river water (Riv2 and Riv3) through whichthe NF patient acquired the infection. These strains were com-pared to an environmental isolate (ATCC 7966T) of A. hydrophila(as a negative control), the genome of which we first sequencedand annotated (40), and that of highly virulent A. dhakensis SSU,whose sequence was annotated during this study. Our purposewas to discern if there were any pathogenic features unique tothese strains, especially those associated with NF. A. hydrophila WIwas chosen because of its resistance to multiple antibiotics.

Biofilm formation. As shown in Fig. 3, the NF1 and SSU iso-lates formed comparable and significant biofilms. However, all ofthe other isolates produced statistically significantly lesser bio-films, which were minimal with the ATCC 7966T and Riv3 iso-lates.

Swimming and swarming motility. All of the isolates testedexhibited swimming motility (Fig. 4), albeit to various degrees.Among these isolates, NF1 and Riv2 showed a hypermotility phe-notype compared to SSU. Both ATCC 7966T and WI were mini-mally motile. To more accurately measure the migration zonesizes of hypermotile strains (NF1 and Riv2), we used larger(100-mm instead of 60-mm) swimming agar plates. Our data in-dicated that while the swimming motility of SSU remained unal-tered, one could clearly visualize the hypermotility phenotype ofthe NF1 and Riv2 isolates because of the larger surface area avail-able to them for swimming (data not shown).

As noted in Fig. 5, NF1 and Riv2 swarmed to a higher degreethan SSU, a difference that became statistically significant when100-mm agar plates were used (data not shown). While ATCC7966T, NF2, and Riv3 had moderate levels of swarming motility,WI did not swarm.

Protease and hemolytic activities. As shown in Fig. S5 in the

FIG 3 Biofilm formation by Aeromonas strains was quantified on polystyreneplastic after 24 h of incubation at 37°C by CV staining. Three independentexperiments were performed, and the arithmetic means � standard deviationswere plotted. An asterisk indicates a P value of �0.0001 as determined byone-way ANOVA. OD, optical density.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4171

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

supplemental material, all of the Aeromonas strains studied pro-duced a protease(s) to various degrees; however, no statisticallysignificant difference between them and SSU was found. In termsof hemolytic activity associated with Act, we noted that the hemo-globin release from rabbit erythrocytes caused by NF2, WI, Riv2,Riv3, and ATCC 7966T was at a level much lower than that of SSUand NF1 (see Table S2 in the supplemental material). To demon-strate that most this hemolytic activity was associated with Act, weneutralized the toxin by using anti-Act polyclonal antibodies. In-deed, 75 to 98% of the Act-associated hemolytic activity of thebacterial strains tested was neutralized by these antibodies (seeTable S2).

AHL production. As shown in Fig. S6A in the supplementalmaterial, all of the Aeromonas isolates (clinical versus water) pro-duced similar levels of AHLs within 24 h, except for ATCC 7966T,as judged by the production of purple pigment by C. violaceumCV026 (73) (see Fig. S6B). We used an AHL-negative mutant ofSSU (�ahyRI) (62) as a negative control in the assay.

Production and secretion of Hcp and AexU. Hcp secretion isT6SS dependent and a reliable indicator of the presence of a func-tional T6SS in bacteria (74). On the basis of Western blot analysis,Hcp production and secretion were noted in A. jandaei Riv2, aswell as in A. dhakensis SSU (Fig. 6A). In contrast to ATCC 7966T,which did not synthesize Hcp, isolates NF1, NF2, Riv3, and WIsynthesized Hcp but were unable to secrete it into the medium(Fig. 6A). As another positive control, we used the production andsecretion of Hcp by Yersinia pestis, which also suggested the func-tionality of the T6SS in the plague bacterium.

The AexU toxin is a T3SS effector identified in SSU, whichpossesses GAP and ADP-RT activities (63, 75). Both isolates NF1and NF2 produced AexU, as was also noted in SSU (Fig. 6B). Noneof the other strains tested produced AexU.

Existence of ExoA homolog. On the basis of genome-widesequencing, a homolog of P. aeruginosa ExoA was found in SSUand NF2 isolates. Comparison of the DNA sequence of the exoAgene of P. aeruginosa with that of Aeromonas species revealed 70%homology. At the amino acid level, the sequence identity ranged

from 64 to 65%. A sequence identity of 97 to 98% was noted forisolates SSU, NF2, and E1. To demonstrate that the exoA gene inthe Aeromonas strains tested was expressed, we performed West-ern blot analysis with antibodies to P. aeruginosa ExoA. As notedin Fig. 6C, a secreted version of ExoA was seen in SSU and NF2isolates. As a positive control, we used P. aeruginosa PA103, whichalso demonstrated a correct-size ExoA band of 66 kDa (76).

By analyzing the ExoA protein sequences at the NCBI con-served-domain database, a Pfam protein sequence search (http://pfam.sanger.ac.uk/), and SMART analysis (http://smart.embl-heidelberg.de/), it was revealed that the ExoA sequence fromaeromonads possessed three domains similar to that found in P.aeruginosa. The ExoA sequences of SSU, NF2, and E1 had 639amino acid residues. Interestingly also, the amino acid sequencesof ExoA homologs in more recently sequenced A. hydrophila ge-nomes, according to the NCBI database, were shorter than that ofExoA of SSU. Prediction of signal peptides in Archaea with thehidden Markov model revealed that the first 25 amino acid resi-dues represent the leader peptide in SSU, NF2, and E1, similar tothat seen in ExoA of P. aeruginosa PA103 (76).

Virulence associated with clinical and environmental aero-monads in a mouse model of infection. (i) Septicemic-mousemodel. To test the virulence of aeromonads, animals were injectedwith various A. hydrophila strains (SSU, ATCC 7966T, NF1, NF2,WI, and Riv3) or A. jandaei Riv2 via the i.p. route at a dose of 5 �107 CFU. As shown in Fig. 7, 100% of the animals infected with A.dhakensis SSU and 90% of the mice challenged with NF1 or Riv2died within 2 days p.i. The mortality rates were 35 and 45%, re-spectively, when the mice were infected with NF2 and Riv3 at thesame infectious dose. Both strains ATCC 7966T and WI of A.hydrophila were the least virulent in a septicemic-mouse model,allowing the survival of 90 to 100% of the animals.

(ii) Developing NF by Aeromonas species in a mouse model.Previously, the subcutaneous injection route was used on the foot-pads of humanized CD46 transgenic mice to induce Streptococcuspyogenes-associated NF (77). However, Aeromonas-related NF has

FIG 4 Swimming motility assay of various Aeromonas strains. We used60-mm petri plates to determine swimming motility. Three independent ex-periments were performed, and the arithmetic means � standard deviationswere plotted. An asterisk indicates a P value of �0.0001 as determined byone-way ANOVA.

FIG 5 Swarming motility assay of various Aeromonas strains. We used 60-mmpetri plates to determine swarming motility. Three independent experimentswere performed, and the arithmetic means � standard deviations were plot-ted. An asterisk indicates a P value of �0.0001 as determined by one-wayANOVA.

Grim et al.

4172 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

always been linked to infections acquired through nonintact skinand exposure of wounds or skin lacerations to contaminated soilor water (9, 78, 79). To mimic this infection route, we challengedmice i.m. with strains SSU, NF1, NF2, and Riv2 or a mixture ofNF1 and NF2 and used strain ATCC 7966T as a negative control.As depicted in Fig. 8A, A. jandaei Riv2 was highly virulent, result-ing in 100% mortality in mice within 24 h p.i. at a dose of 5 � 108

CFU. Isolates SSU and NF2 also resulted in 100% mortality but at48 h p.i.

Interesting were our findings that while NF2 was still 100%lethal to mice at a lower dose of 5 � 107 CFU by the i.m. route(data not shown), 65% of the animals survived when injectedby the i.p. route with NF2 (Fig. 7). On the other hand, strain

NF1, which was relatively more virulent than strain NF2 in asepticemic-mouse model of infection (Fig. 7), with 10% versus65% survival, exhibited greatly reduced virulence (100% sur-vival) when injected via the i.m. route at the same dose of 5 �107 CFU (data not shown). At a higher dose of 5 � 108 CFU,animals infected with NF1 also succumbed to infection by day6 (Fig. 8A). As expected, none of the animals infected withstrain ATCC 7966T died.

As shown in Fig. 8B, while NF2 was still highly virulent (100%mortality) at doses of 1 � 108 to 2 � 108 CFU when given by thei.m. route, NF1 resulted in death in only 40% of the mice infectedwith a dose of 2 � 108 CFU. Since the NF1 and NF2 isolates wereobtained at different times (NF1 from the initial wound site and

FIG 6 Production and secretion of Hcp, AexU, and ExoA by various strains of Aeromonas and P. aeruginosa as determined by Western blotting. While only SSUand Riv2 secreted Hcp into the supernatant, the Hcp protein was identified in the cell pellets of all of the other strains studied, except for A. hydrophila ATCC7966T (A). Isolates SSU, NF1, and NF2 produced AexU (B), while ExoA was secreted by isolates SSU and NF2, as well as by P. aeruginosa PA103 (C). Y. pestis (YP)served as a control. The molecular sizes of Hcp, AexU, and ExoA are indicated.

FIG 7 Virulence of various Aeromonas strains in a septicemic-mouse model of infection. Swiss-Webster mice (n 6 to 23) were infected with a dose of 5 � 107

CFU of the strains shown by the i.p. route. The animals were monitored for death for 14 days. The data were statistically analyzed by using the Kaplan-Meiersurvival estimate, and the actual P values are presented. The number of animals used for each strain is shown in parentheses.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4173

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

NF2 after amputation) from the same patient, we determinedwhether mixing these two cultures in equal numbers could aug-ment their virulence in a mouse model. Our data indicated thatthe virulence of NF2 predominated in the mixed culture, as 80%of the animals died by day 2 (Fig. 8B).

Dissemination of Aeromonas species after i.m. injection in amouse model. To evaluate the dissemination pattern of variousAeromonas isolates, infected mice (n 5) were sacrificed at 24 or48 h p.i. and their spleens and livers were subjected to CFU deter-mination. In agreement with the animal survival data, greater bac-

terial loads were noted in the livers and spleens of mice in thosegroups that were infected with SSU (1 � 108 CFU), Riv2 (1 � 108

CFU), and NF2 (2 � 108 CFU) (Fig. 9A to C). The number ofbacteria in these organs increased over the course of infectionfrom 24 to 48 h. Because of the highly lethal nature of the Riv2isolate, some of the animals died between 24 to 48 h after infection(Fig. 9B). Strain NF1 disseminated to the liver and spleen in lowernumbers, even at a slightly higher infective dose (5 � 108 CFU)(Fig. 9D), than strains SSU, NF2, and Riv2 (Fig. 9A to C), withstrain ATCC 7966T disseminating minimally to the livers and

FIG 8 Percent survival of mice after infection with various Aeromonas isolates by the i.m. route. (A) Animals were infected with isolate SSU, NF1, NF2, Riv2, orATCC 7966T at 5 � 108 CFU/mouse and monitored for death for 14 days. (B) Animals were infected with NF1 and NF2 at the lower doses indicated or with amixture of these two isolates at 1 � 108 CFU each per mouse. Isolates SSU, NF1, NF2, and Riv2 were statistically significantly more virulent than avirulent A.hydrophila strain ATCC 7966T (A). Likewise, NF2 at both the doses was statistically significantly more virulent than NF1 (B). The data were statistically analyzedby using Kaplan-Meier survival estimates, and the actual P values are presented.

Grim et al.

4174 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

spleens of mice (1/5) over the course of 48 h of infection(Fig. 9E).

Gross and histopathological evaluations of tissues fromAeromonas species-associated NF. We initially focused on twoisolates, namely, SSU and NF1. At 24 h p.i. with isolate SSU (5 �

108 CFU), muscle tissues around the injection site revealed severeedematous and hemorrhagic swelling, unlike those of animals in-jected with PBS or lipopolysaccharide (20 �g) (Fig. 10A). Theseinflammatory changes were more progressive in nature at 48 h p.i.(Fig. 10B1 and B2). In addition, the leg tissues of mice infected

FIG 9 Bacterial dissemination patterns for various Aeromonas isolates in a mouse model of NF. Mice (n 5) were infected by the i.m. route with the dose ofisolate SSU (A), Riv2 (B), NF2 (C), NF1 (D), or ATCC 7966T (E) indicated. At 24 or 48 h p.i., the spleen and liver were aseptically removed from each mouse,homogenized, and subjected to bacterial colony counting. M1 to M5 represent individual mice, and some groups have fewer animals because of deaths p.i.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4175

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

with the NF1 isolate (2 � 108 CFU) were liquefied and showedgangrenous necrosis after 7 days, as shown in Fig. 10C.

Histopathological examination of skeletal muscle tissuesfrom thigh regions adjacent to the injection site revealed vari-ous levels of multifocal and diffuse necrotic changes in striated

muscle bundles, accompanied by edematous exudation and in-filtration of polymorphonuclear cells (PMNs), specifically,neutrophils (Fig. 11A). Specifically, tissues collected from ani-mals infected with strains ATCC 7966T and NF1 showed morePMN infiltration and edematic fluid but limited multifocal ne-

FIG 10 Gross pathological examination of leg muscle tissues after infection with A. dhakensis SSU and A. hydrophila NF1. At 24 (A) or 48 (B1 and B2) h p.i., legtissues from animals infected with isolate SSU at 5 � 108 CFU/mouse via the i.m. route were examined macroscopically. Note that the B2 animal expired at thetime of examination. The other leg of the animal was given either PBS or LPS. Also, at 7 days p.i. (C), the animals that survived infection with strain NF1 at 2 �108 CFU/mouse were sacrificed and the leg tissues were grossly observed for lesions. Only representative animals are shown.

FIG 11 Histopathological examination of mouse leg muscle tissues after infection by the i.m. route with various Aeromonas isolates. (A) Leg muscle tissuescollected at 48 h p.i. from mice injected with strain ATCC 7966T, NF1, SSU, or Riv2 (at a dose of 5 � 108 CFU/mouse for ATCC 7966T and NF1 or 1 � 108 forSSU and Riv2) or with PBS (negative control) were processed for histopathological analysis and observed under a microscope after staining with H&E. A starsymbols denote infiltration of PMNs, while squares, rectangles, and long or short arrows show necrotic lesions in skeletal muscle tissues with loss of characteristicstriations. Triangles indicate edematous swelling. Each of these images represents a group of five samples. (B) Animals were infected with highly virulent isolateNF2 at 5 � 106 or 1 � 107 CFU/mouse. An arrow indicates skeletal muscle necrosis, and an asterisk indicates infiltration of PMNs. An image representative offive mice is shown. Magnifications (for panels A and B), 100� (top rows) and 400� (bottom rows).

Grim et al.

4176 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

crotic changes (Fig. 11A). On the other hand, tissues from an-imals infected with strains SSU and Riv2 exhibited extensiveand diffuse necrotic changes with moderate levels of edema andinfiltration of PMNs (Fig. 11A). Since strain NF2 was highlylethal to mice by the i.m. route, we reduced the infectious doseto 5 � 106 to 5 � 107 CFU. These tissues also revealed variouslevels of multifocal and local, as well as diffuse, necroticchanges in the skeletal muscle. In addition, infiltration ofPMNs (specifically, neutrophils) and various severities of ede-matic fluid accumulation were noticed. As depicted in Fig. 11B,necrotic and degenerated myofibers could be seen (arrow) withsignificant inflammatory and edematous responses (asterisk)in animals infected with NF2.

Dissemination of bioluminescent bacteria in a mouse modelof NF. We infected mice (n 10) by the i.m. route in the righthind leg with a challenge dose of 5 � 108 CFU of each of theabove-mentioned strains, except Riv2. Since Riv2 exhibited highervirulence than the other strains tested, two lower infectious doses(5 � 107 and 1 � 108 CFU) were used. After infection, we mea-sured bioluminescence in the legs, liver, and spleen. In addition,we determined bacterial counts in the livers and spleens of fiveanimals per time point at 24 and 48 h p.i.

At 24 h p.i. for no strain SSU bioluminescence was observedbeyond the baseline level, while two animals were positive forbacterial colonies in the liver and one was positive for bacterialcolonies in the spleen (Fig. 12A). It should be noted that for ani-mals to show bioluminescence, the number of organisms shouldbe �105 CFU (68). At 48 h p.i., three mice were positive for bio-luminescence in the leg at the injection site and two had lumines-cence extending into the peritoneal cavity. The bioluminescenceobserved also corresponded to the bacterial counts in the liver andspleen (Fig. 12A). Both of the animals that had strong lumines-cence in the peritoneal cavity showed bacterial counts in the rangeof 106 to 107 CFU in the liver and spleen. For each of the otherstrains (NF1, ATCC 7966T, and Riv2), strong bioluminescencewas observed in the infected leg at 24 h p.i. (Fig. 12B to E), with theexception of one mouse in the ATCC 7966T group (Fig. 12C).

At 48 h p.i., a decrease in luminescence was observed in the legsacross all of the mice infected with strain ATCC 7966T, with amajority of the animals (4/5) having dissemination into the liverbut remaining below 105 CFU. No bacteria were detected in thespleen (Fig. 12C). With strain NF1, luminescence was observed inall of the mice at both the 24- and 48-h time points, with dissem-ination of bacteria to the spleen and liver; however, luminescencewas absent from the peritoneal cavity (Fig. 12B). In mice infectedwith strain Riv2 at a dose of 5 � 107 CFU, luminescence wasobserved at the injection site in the hind leg, as well as extendinginto the peritoneal cavity of one mouse, corresponding to viableplate counts of �107 CFU in the liver and spleen at 48 h (Fig. 12D).In mice infected with 1 � 108 CFU of Riv2, strong luminescencewas observed at the injection site, as well as extending into theperitoneal cavities of two mice at 48 h, representing disseminatinginfection even at the lower dose (Fig. 12E). At both doses of Riv2,one mouse died by 24 h and two additional animals died at 48 hand were not included for imaging.

DISCUSSION

In this study, genetically diverse isolates of Aeromonas (belongingto A. hydrophila, A. jandaei, and A. caviae) were sequenced andanalyzed to identify virulence factors associated with their distinct

pathotypes. Our purpose was to discern if there were any patho-genic features unique to these strains, especially those associatedwith NF. NF is part of a family of diseases referred to as necrotizingsoft tissue infections (NSTIs). Even milder NSTIs often requireseveral surgical debridement (80). There are three major classifi-cations of NSTIs, type I, involving mixed infections of anaerobesand aerobic bacteria (polymicrobial), type II, involving S. pyo-genes, and type III, gas gangrene or clostridial myonecrosis, withall three requiring intensive care and rapid treatment. A variant oftype I is salt water NF, in which an apparently minor skin woundis contaminated with salt water containing a Vibrio species. Evenwith rapid surgical intervention (81, 82), case fatalities can be ashigh as 50% (83) and leave surviving patients with lifelong disabil-ities and disfigurement (81, 82, 84).

Whole-genome sequencing of isolates NF1 to NF4 from a pa-tient revealed that NF1 was genetically distinct and that strainsNF2 to NF4 were not true clones in the sense that they likely didnot derive from a last common ancestor either during the courseof infection or in the environment directly prior to infection.Comparative genomic analysis, coupled with functional assays, ofstrains NF1 and NF2 and other clinical and environmental Aero-monas strains allowed us to investigate pathovar-specific virulencefactors or markers.

Although isolates Riv2 and Riv3 were cultured from the samewater source through which the patient with NF contracted theinfection, they were genetically different (e.g., Riv3) from isolatesNF1 to NF4 of A. hydrophila. We believe that since these watersamples were collected at a later time subsequent to the infectionof the patient, water microbiota changed because of fluctuationsin the environmental conditions (85, 86). Nevertheless, our stud-ies reinforce the idea that both benign and highly virulent strainscan be present at all times in fresh and brackish waters, whereaeromonads flourish. In fact, the advent of Aeromonas species as acause of NF leads us to suggest that this condition is another vari-ant of type 1, which can be termed “estuarine fasciitis.”

In the same vein, recent studies of Skwor et al. (85) recoveredtetracycline- and ciprofloxacin-resistant Aeromonas strains fromLake Erie in Pennsylvania. Importantly, the prevalence of aeroly-sin and serine protease-encoding genes and their correspondinghemolytic and proteolytic/cytotoxic activities was also high in an-tibiotic-resistant strains of some Aeromonas species (e.g., A. hy-drophila and A. veronii), suggesting that they contribute to bacte-rial virulence (85). Since the antibiotics of choice for the treatmentof aeromonads are sulfamethoxazole-trimethoprim and cipro-floxacin (87), the presence of antibiotic-resistant Aeromonasstrains in natural water becomes even more significant in the man-agement of serious infections such as NF.

Aeromonads readily form biofilms; however, it is unclearwhether there are any variations in biofilm formation amongclonally diverse clinical and environmental isolates. The dynamicsof biofilm formation depend on the environmental niches wherethe microbes hide during an infection and dictate the pathogenicnature of a given organism by modulating virulence gene expres-sion (88). In this vein, we described the dynamics of mature,three-dimensional biofilm formation by A. dhakensis strain SSU(89). Indeed, highly virulent isolates NF1 and Riv2 formed bio-films comparable to that of SSU, and these likewise were betterswimmers and swarmers, which overall contributed to their viru-lence (Fig. 3 to 5).

The AI-1-based QS system consists of the ahyRI genes, with

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4177

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

FIG 12 Bioluminescent imaging and corresponding bacterial loads of harvested livers and spleens infected with luminescent Aeromonas strains SSU (A), NF1(B), ATCC 7966T (C) Riv2 (5 � 107 CFU) (D), and Riv2 (1 � 108 CFU) (E). Infected mice (n 5) were imaged at 24 and 48 h p.i. and used for colony counting.For isolates SSU, NF1, and ATCC 7966T, the challenge dose was 5 � 108 CFU/mouse by the i.m. route. The CFU are reported for whole organs derived fromcolony plate counts.

Grim et al.

4178 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

ahyR encoding a transcription activator and the ahyI gene synthe-sizing an autoinducer AHL. The role of this system in biofilmregulation has been well established (27, 62, 90). However, in thepresent study, no direct correlation between AHL production andbiofilm formation in the strains studied was noted (Fig. 3; see Fig.S6 in the supplemental material). It is plausible that in some Aero-monas strains, the regulation of biofilm formation is very tightlyregulated, as dictated by specific environmental and nutritionalniches found only under in vivo conditions and needing furtherexploration. Indeed, the expression of the locus that codes forlateral flagella and allows bacteria to swarm is linked to biofilmformation (91). Further, overlapping sets of regulators governswarming motility and biofilm formation in a reciprocal fashion,depending upon the environmental cues that challenge microbesduring an infection (92, 93).

In P. aeruginosa, it was noted that swarming isolates had di-minished biofilm formation capabilities and that swarming mo-tility occurred in the absence of either twitching or swimmingmotility. Further, the best swarmers secreted increased levels ofproteases (94). On the contrary, the best E. coli biofilm formersdisplayed the greatest swimming motility (95). The above-men-tioned scenario with E. coli seemed to fit with the SSU, NF1, andRiv2 isolates, as they exhibited greater swarming and swimmingmotilities (Fig. 4 and 5), formed better biofilms (Fig. 3), and werehighly virulent in mouse models of infection (Fig. 7 and 8). How-ever, on the basis of this study, no correlation seemed to existbetween swarming motility (Fig. 5) and protease production (seeFig. S5 in the supplemental material).

The high hemolytic activity associated with Act, as well as thepresence of the T3SS effector AexU also correlated with increasedvirulence of SSU and NF1 in mouse models of infection (Fig. 6 to8; see Table S2 in the supplemental material). These results werecorroborated by our earlier findings indicating that deletion of theact and aexU genes from A. dhakensis SSU significantly reducedtheir respective virulence in the septicemic-mouse model of infec-tion (61, 63, 75).

Interesting, however, were our findings that although Riv2 andNF2 were highly virulent in one or both in vivo models of bacterialinfection (Fig. 7 and 8), they produced low levels of Act (see TableS2 in the supplemental material). For NF2, it is likely related to thetruncated nature of the act gene, while in the Riv2 isolate, the toxingene might be regulated differently from that of isolates SSU andNF1. These isolates were still highly virulent in an NF mousemodel (Fig. 8), despite low Act-associated hemolytic activity. Thiscould be explained by the presence of a functional ExoA in NF2and secretion of T6SS-associated Hcp or differences in T3SS effec-tor proteins in Riv2, which could compensate for low Act-associ-ated hemolytic activity (Fig. 6). The reverse seemed to be true forthe NF1 isolate, as it had high hemolytic activity and producedAexU but did not contain the exoA gene or secrete Hcp. In thisregard, isolate SSU had all of the weapons (e.g., the presence ofAct, ExoA, AexU, and Hcp) (Fig. 6) necessary to inflict seriousdamage on the host.

The reduced hemolytic activity of Act in the WI isolate could bethe result of an insertion of a putative arylsulfate-encoding geneupstream of the act gene that might have affected the expression ofthe gene (Fig. 2A). Finally, it is not clear why Riv3 had Act toxin-neutralizable hemolytic activity (see Table S2 in the supplementalmaterial), as the genome of this isolate did not harbor the act gene.One plausible explanation is that the hemolytic activity seen in this

isolate is contributed by another hemolysin for which we mighthave antibodies in our Act antitoxin preparation. Another possi-bility is that there might be a neutralizable cross-reactiveepitope(s) among different hemolysins. Since the WI isolate didnot secrete Hcp or produce AexU and ExoA, it was the least viru-lent in a mouse model of infection. The same logic applies to strainATCC 7966T (Fig. 6 and 7).

Our earlier study indicated the presence of an exoA gene in thegenome of wound isolate E1 (42); however, the functionality ofthe gene was not confirmed. Indeed, ExoA was secreted by theSSU, NF2, and PA103 isolates (Fig. 6), as well as by E1 (data notshown). While secretion of ExoA was demonstrated in the SSUand NF2 isolates when they were grown in LB medium at 37°C,ExoA was secreted from the E1 isolate at a low iron concentrationof 0.05 �g/ml at 32°C (65). The low level of ExoA secretion de-tected in PA103 was probably related to the complex nutritionalconditions needed for its synthesis (96).

In a mouse model of NF, the pathodymanics of lesion devel-opment by isolates SSU, NF1, NF2, and Riv2 (Fig. 11) closelymimicked the time course events reported for human NF. WhenNSTIs were first characterized, an overabundance of neutrophilsat the infection site was recognized as a hallmark of the disease(97). In the mouse model of NF, we also noted abundant neutro-phils at the infection site (Fig. 11). Therefore, it is plausible thatneutrophils might have a role in the pathogenesis of NSTIs and inthe extensive tissue damage that is characteristic of the disease.

The mechanism and significance of enhanced neutrophil re-cruitment observed in NSTIs are unknown. However, increases inthe level of resistin, a proinflammatory peptide derived from neu-trophils, have been correlated with disease severity in both NSTIsand septic shock (98). It is known that degranulation events inneutrophils lead to the release of azurophilic and specific granulesinto the extracellular space, exposing the surrounding tissue totoxic enzymes and antimicrobial peptides (99).

Strain NF2 was less lethal to mice when given by the i.p. routebut was highly virulent when injected by the i.m. route (Fig. 7 and8). These data raised the possibility that the microenvironmentprevailing at the i.m. injection site would favor efficient adapta-tion of the bacteria to the host immune system and thus later led tosystemic infection because of the presence of unique virulencefactors. On the other hand, A. hydrophia strain ATCC 7966T,which lacks a T3SS and many other virulence factors (e.g., Hcp,ExoA) (40), remained avirulent in mice although the site of i.m.injection showed infiltration of PMNs (Fig. 11).

Substantiating the above-mentioned paradigm, the animalmortality pattern caused by isolates SSU, NF1, NF2, and Riv2 (Fig.8) closely matched their dissemination profiles (Fig. 9) and corre-lated with the presence of key virulence factors. As alluded toabove, a hallmark characteristic of NF is the dissemination of bac-terial infection along fascial planes of the limbs, followed by, in asetting of severe infection, sepsis (81). To study this phenomenonin more detail, we used various bioluminescent reporter Aeromo-nas strains. In most of the strains tested, we observed that infectionwas limited to the hind leg, the site of bacterial injection, althoughdissemination and colonization of that region were extensive, withfurther dissemination of some strains (e.g., SSU, Riv2) to otherorgans occurring by 48 h. Unfortunately, since NF2 was not highlyvirulent in a septicemic-mouse model of infection, we did notconstruct this reporter strain. However, significant dissemination

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4179

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

of this isolate to other organs was revealed by plate counting (Fig.9C).

A. hydrophila ATCC 7966T had no lethality in this mousemodel, and the infection was confined to the injection site, withlittle or no dissemination, even within the region of infection (Fig.12C). In both strains SSU and Riv2, dissemination above thethreshold for imaging was noted, which was confirmed by viableplate counts (Fig. 12A, B, D, and E).

In contrast, mice infected with NF1 showed more uniformdissemination, albeit at lower levels at 24 and 48 h (Fig. 12B), butthis strain exhibited larger necrotic lesions in the infected legs ofmice. Despite being isolated directly from the initial wound sitefrom a patient with NF, the NF1 isolate seemed to show a lowerpropensity to disseminate and grow rapidly outside the fascialtissue of the infected leg (Fig. 12B). In fact, infection seemed to beconfined to the leg, and only after an aggressive necrotic lesion wasformed did the infection spread significantly (data not shown).One week p.i., several of the mice infected with NF1 developedlesions that encompassed the entire leg (Fig. 10C) before advanc-ing across the peritoneum. Histologically, strain NF1 also seemedto localize at the injection site and thus to favor the infiltration ofPMNs, eventually leading to localized necrotic changes ratherthan causing systemic dissemination (Fig. 11).

On the basis of the genomic and functional data presented inthis paper, the role of virulence factors such as Act, ExoA, andT3SS and T6SS and their effectors and the functionality of thelateral flagella in causing NF are evident. However, one should becautious in data interpretation, as the unique environment cuesthe aeromonads are exposed to during NF may trigger bacterialgenes differently to contribute to the severity of the disease. Whilewe did not identify a single, highly unique virulence factor in theNF-associated strains, nevertheless, it was quite evident that thesestrains were extremely virulent and were representative of manyclinical strains described in our previous studies (33, 42).

Our findings are consistent with previous suggestions thatthere are particular virulent subsets of pathotypes within the A.hydrophila group. Recent findings obtained with the most preva-lent NF-causing organism, GAS, suggested that NF is the by-prod-uct of superantigen genes responsible for toxins that are capable ofbypassing normal intracellular antigen processing and presenta-tion by major histocompatibility complex class II (100). In theabsence of evidence of these or similar genes in Aeromonas, wepropose that Aeromonas-related NF is the result of either a triad orquartet of pathovar-specific virulence factors that result in NFprimarily in susceptible individuals. In sum, it appears that secre-

tion/translocation of multiple toxins via different secretion sys-tems along with other virulence mechanisms as summarized inTable 4 could be responsible for causing NF by aeromonads andneeds further investigation.

While this hypothesis remains to be proven, future studies onAeromonas NF cases will provide information regarding the mo-lecular and genetic features of these organisms. Studies such asthose performed with GAS should be instrumental in revealing theintricacies of Aeromonas NF in vulnerable hosts.

ACKNOWLEDGMENTS

The financial support provided to A.K.C. through Leon Bromberg andRobert E. Shope and John S. Dunn Distinguished Chair in Global Healthendowments, UTMB, is gratefully acknowledged.

We thank the Genomics Core (specifically, Tom Wood and StevenWiden), UTMB, for sequencing several of the aeromonad genomes.

REFERENCES1. Von Graevenitz A, Mensch AH. 1968. The genus Aeromonas in human

bacteriology—report of 30 cases and review of the literature. N. Engl. J.Med. 278:245–249. http://dx.doi.org/10.1056/NEJM196802012780504.

2. Presley SM, Rainwater TR, Austin GP, Platt SG, Zak JC, Cobb GP,Marsland EJ, Tian K, Zhang B, Anderson TA, Cox SB, Abel MT,Leftwich BD, Huddleston JR, Jeter RM, Kendall RJ. 2006. Assessmentof pathogens and toxicants in New Orleans, LA following HurricaneKatrina. Environ. Sci. Technol. 40:468 – 474. http://dx.doi.org/10.1021/es052219p.

3. Chen PL, Wu CJ, Chen CS, Tsai PJ, Tang HJ, Ko WC. 16 November2013. A comparative study of clinical Aeromonas dhakensis and Aeromo-nas hydrophila isolates in southern Taiwan: A. dhakensis is more predom-inant and virulent. Clin. Microbiol. Infect. http://dx.doi.org/10.1111/1469-0691.12456.

4. Palú AP, Gomes LM, Miguel MA, Balassiano IT, Queiroz ML, Freitas-Almeida AC, de Oliveira SS. 2006. Antimicrobial resistance in food andclinical Aeromonas isolates. Food Microbiol. 23:504 –509. http://dx.doi.org/10.1016/j.fm.2005.07.002.

5. Chauret C, Volk C, Creason R, Jarosh J, Robinson J, Warnes C. 2001.Detection of Aeromonas hydrophila in a drinking-water distribution sys-tem: a field and pilot study. Can. J. Microbiol. 47:782–786. http://dx.doi.org/10.1139/w01-070.

6. Hiransuthikul N, Tantisiriwat W, Lertutsahakul K, Vibhagool A,Boonma P. 2005. Skin and soft-tissue infections among tsunami survi-vors in southern Thailand. Clin. Infect. Dis. 41:e93–96. http://dx.doi.org/10.1086/497372.

7. Figueras MJ, Aldea MJ, Fernandez N, Aspiroz C, Alperi A, Guarro J.2007. Aeromonas hemolytic-uremic syndrome. A case and a review of theliterature. Diagn. Microbiol. Infect. Dis. 58:231–234. http://dx.doi.org/10.1016/j.diagmicrobio.2006.11.023.

8. Figueras MJ. 2005. Clinical relevance of Aeromonas sM503. Rev. Med.Microbiol. 16:145–153. http://dx.doi.org/10.1097/01.revmedmi.0000184410.98677.8a.

TABLE 4 Virulence characteristics of various Aeromonas strains

StrainBiofilmformation

Motility Hcp (T6SS)

AexU (T3SS) ExoA

Cytotoxicenterotoxin(Act)a

Virulence in mice

Swimming Swarming Expression Secretion i.p. i.m.

A. dhakensis SSU High High High � � � � High High HighA. hydrophila ATCC 7966T Low Moderate Moderate � � � � Moderate Low LowA. hydrophila NF1 High Hypermotile Hypermotile � � � � High High HighA. hydrophila NF2 Moderate Moderate Moderate � � � � Low Moderate HighA. hydrophila WI Moderate Low Low � � � NDb Moderate Low NDA. jandaei Riv2 Moderate Hypermotile Hypermotile � � � � Low High HighA. hydrophila Riv3 Low Moderate Moderate � � � ND Moderate Moderate NDa Hemolytic activity associated with Act.b ND, not determined.

Grim et al.

4180 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

9. Monaghan SF, Anjaria D, Mohr A, Livingston DH. 2008. Necrotizingfasciitis and sepsis caused by Aeromonas hydrophila after crush injury ofthe lower extremity. Surg. Infect. 9:459 – 467. http://dx.doi.org/10.1089/sur.2007.028.

10. Tsai YH, Huang KC, Huang TJ, Hsu RW. 2009. Case reports: fatal necro-tizing fasciitis caused by Aeromonas sobria in two diabetic patients. Clin.Orthop. Relat. Res. 467:846–849. http://dx.doi.org/10.1007/s11999-008-0504-0.

11. Cui H, Hao S, Arous E. 2007. A distinct cause of necrotizing fasciitis:Aeromonas veronii biovar sobria. Surg. Infect. 8:523–528. http://dx.doi.org/10.1089/sur.2006.046.

12. Furusu A, Yoshizuka N, Abe K, Sasaki O, Miyazaki K, Miyazaki M,Hirakata Y, Ozono Y, Harada T, Kohno S. 1997. Aeromonas hydrophilanecrotizing fasciitis and gas gangrene in a diabetic patient on haemodi-alysis. Nephrol. Dial. Transplant. 12:1730 –1734. http://dx.doi.org/10.1093/ndt/12.8.1730.

13. Joseph SW, Daily OP, Hunt WS, Seidler RJ, Allen DA, Colwell RR.1979. Aeromonas primary wound infection of a diver in polluted waters.J. Clin. Microbiol. 10:46 – 49.

14. Sever R, Lee Goldstein A, Steinberg E, Soffer D. 2013. Trauma with atouch of fresh water: necrotizing fasciitis caused by Aeromonas hydro-philia after a motorcycle accident. Am. Surg. 79:E326 –E328.

15. Xu XJ, Ferguson MR, Popov VL, Houston CW, Peterson JW, ChopraAK. 1998. Role of a cytotoxic enterotoxin in Aeromonas-mediated infec-tions: development of transposon and isogenic mutants. Infect. Immun.66:3501–3509.

16. Ferguson MR, Xu XJ, Houston CW, Peterson JW, Coppenhaver DH,Popov VL, Chopra AK. 1997. Hyperproduction, purification, andmechanism of action of the cytotoxic enterotoxin produced by Aeromo-nas hydrophila. Infect. Immun. 65:4299 – 4308.

17. Galindo CL, Fadl AA, Sha J, Pillai L, Gutierrez C, Jr, Chopra AK.2005. Microarray and proteomics analyses of human intestinal epi-thelial cells treated with the Aeromonas hydrophila cytotoxic entero-toxin. Infect. Immun. 73:2628 –2643. http://dx.doi.org/10.1128/IAI.73.5.2628-2643.2005.

18. Galindo CL, Gutierrez C, Jr, Chopra AK. 2006. Potential involvementof galectin-3 and SNAP23 in Aeromonas hydrophila cytotoxic enterotox-in-induced host cell apoptosis. Microb. Pathog. 40:56 – 68. http://dx.doi.org/10.1016/j.micpath.2005.11.001.

19. Ribardo DA, Kuhl KR, Boldogh I, Peterson JW, Houston CW, ChopraAK. 2002. Early cell signaling by the cytotoxic enterotoxin of Aeromonashydrophila in macrophages. Microb. Pathog. 32:149 –163. http://dx.doi.org/10.1006/mpat.2001.0490.

20. Galindo CL, Fadl AA, Sha J, Gutierrez C, Jr, Popov VL, Boldogh I,Aggarwal BB, Chopra AK. 2004. Aeromonas hydrophila cytotoxic en-terotoxin activates mitogen-activated protein kinases and induces apop-tosis in murine macrophages and human intestinal epithelial cells. J. Biol.Chem. 279:37597–37612. http://dx.doi.org/10.1074/jbc.M404641200.

21. Rosenzweig JA, Chopra AK. 2013. The exoribonuclease polynucleotidephosphorylase influences the virulence and stress responses of yersiniaeand many other pathogens. Front. Cell. Infect. Microbiol. 3:81. http://dx.doi.org/10.3389/fcimb.2013.00081.

22. Sha J, Wang SF, Suarez G, Sierra JC, Fadl AA, Erova TE, Foltz SM,Khajanchi BK, Silver A, Graf J, Schein CH, Chopra AK. 2007. Furthercharacterization of a type III secretion system (T3SS) and of a new effec-tor protein from a clinical isolate of Aeromonas hydrophila—part I. Mi-crob. Pathog. 43:127–146. http://dx.doi.org/10.1016/j.micpath.2007.05.002.

23. Suarez G, Sierra JC, Sha J, Wang S, Erova TE, Fadl AA, Foltz SM,Horneman AJ, Chopra AK. 2008. Molecular characterization of a func-tional type VI secretion system from a clinical isolate of Aeromonashydrophila. Microb. Pathog. 44:344 –361. http://dx.doi.org/10.1016/j.micpath.2007.10.005.

24. Suarez G, Sierra JC, Erova TE, Sha J, Horneman AJ, Chopra AK. 2010.A type VI secretion system effector protein, VgrG1, from Aeromonashydrophila that induces host cell toxicity by ADP ribosylation of actin. J.Bacteriol. 192:155–168. http://dx.doi.org/10.1128/JB.01260-09.

25. Beaz-Hidalgo R, Martinez-Murcia A, Figueras MJ. 2013. Reclassifica-tion of Aeromonas hydrophila subsp. dhakensis Huys et al. 2002 and Aero-monas aquariorum Martinez-Murcia et al. 2008. as Aeromonas dhakensissp. nov. comb nov. and emendation of the species Aeromonas hydrophila.Syst. Appl. Microbiol. 36:171–176. http://dx.doi.org/10.1016/j.syapm.2012.12.007.

26. Suarez G, Sierra JC, Kirtley ML, Chopra AK. 2010. Role of Hcp, a type6 secretion system effector, of Aeromonas hydrophila in modulating acti-vation of host immune cells. Microbiology 156:3678 –3688. http://dx.doi.org/10.1099/mic.0.041277-0.

27. Swift S, Karlyshev AV, Fish L, Durant EL, Winson MK, Chhabra SR,Williams P, Macintyre S, Stewart GS. 1997. Quorum sensing in Aero-monas hydrophila and Aeromonas salmonicida: identification of theLuxRI homologs AhyRI and AsaRI and their cognate N-acylhomoserinelactone signal molecules. J. Bacteriol. 179:5271–5281.

28. Kozlova EV, Khajanchi BK, Sha J, Chopra AK. 2011. Quorumsensing and c-di-GMP-dependent alterations in gene transcripts andvirulence-associated phenotypes in a clinical isolate of Aeromonashydrophila. Microb. Pathog. 50:213–223. http://dx.doi.org/10.1016/j.micpath.2011.01.007.

29. Kirov SM, Castrisios M, Shaw JG. 2004. Aeromonas flagella (polar andlateral) are enterocyte adhesins that contribute to biofilm formation onsurfaces. Infect. Immun. 72:1939 –1945. http://dx.doi.org/10.1128/IAI.72.4.1939-1945.2004.

30. Kirov SM, Tassell BC, Semmler AB, O’Donovan LA, Rabaan AA,Shaw JG. 2002. Lateral flagella and swarming motility in Aeromonasspecies. J. Bacteriol. 184:547–555. http://dx.doi.org/10.1128/JB.184.2.547-555.2002.

31. Ramachandran G, Tulapurkar ME, Harris KM, Arad G, Shirvan A,Shemesh R, Detolla LJ, Benazzi C, Opal SM, Kaempfer R, Cross AS.2013. A peptide antagonist of CD28 signaling attenuates toxic shock andnecrotizing soft-tissue infection induced by Streptococcus pyogenes. J. In-fect. Dis. 207:1869 –1877. http://dx.doi.org/10.1093/infdis/jit104.

32. Baker C, Antonovics J. 2012. Evolutionary determinants of geneticvariation in susceptibility to infectious diseases in humans. PLoS One7:e29089. http://dx.doi.org/10.1371/journal.pone.0029089.

33. Khajanchi BK, Fadl AA, Borchardt MA, Berg RL, Horneman AJ,Stemper ME, Joseph SW, Moyer NP, Sha J, Chopra AK. 2010. Distri-bution of virulence factors and molecular fingerprinting of Aeromonasspecies isolates from water and clinical samples: suggestive evidence ofwater-to-human transmission. Appl. Environ. Microbiol. 76:2313–2325.http://dx.doi.org/10.1128/AEM.02535-09.

34. Carnahan AM, Behram S, Joseph SW. 1991. Aerokey II: a flexible keyfor identifying clinical Aeromonas species. J. Clin. Microbiol. 29:2843–2849.

35. Abbott SL, Cheung WK, Janda JM. 2003. The genus Aeromonas: bio-chemical characteristics, atypical reactions, and phenotypic identifica-tion schemes. J. Clin. Microbiol. 41:2348 –2357. http://dx.doi.org/10.1128/JCM.41.6.2348-2357.2003.

36. Havelaar AH, During M, Versteegh JF. 1987. Ampicillin-dextrin agarmedium for the enumeration of Aeromonas species in water by mem-brane filtration. J. Appl. Bacteriol. 62:279 –287. http://dx.doi.org/10.1111/j.1365-2672.1987.tb02410.x.

37. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, Bemben LA,Berka J, Braverman MS, Chen YJ, Chen Z, Dewell SB, Du L, Fierro JM,Gomes XV, Godwin BC, He W, Helgesen S, Ho CH, Irzyk GP, JandoSC, Alenquer ML, Jarvie TP, Jirage KB, Kim JB, Knight JR, Lanza JR,Leamon JH, Lefkowitz SM, Lei M, Li J, Lohman KL, Lu H, MakhijaniVB, McDade KE, McKenna MP, Myers EW, Nickerson E, Nobile JR,Plant R, Puc BP, Ronan MT, Roth GT, Sarkis GJ, Simons JF, SimpsonJW, Srinivasan M, Tartaro KR, Tomasz A, Vogt KA, Volkmer GA, etal. 2005. Genome sequencing in microfabricated high-density picolitrereactors. Nature 437:376 –380. http://dx.doi.org/10.1038/nature03959.

38. Simpson JT, Wong K, Jackman SD, Schein JE, Jones SJ, Birol I. 2009.ABySS: a parallel assembler for short read sequence data. Genome Res.19:1117–1123. http://dx.doi.org/10.1101/gr.089532.108.

39. Aziz RK, Bartels D, Best AA, DeJongh M, Disz T, Edwards RA,Formsma K, Gerdes S, Glass EM, Kubal M, Meyer F, Olsen GJ, OlsonR, Osterman AL, Overbeek RA, McNeil LK, Paarmann D, Paczian T,Parrello B, Pusch GD, Reich C, Stevens R, Vassieva O, Vonstein V,Wilke A, Zagnitko O. 2008. The RAST server: rapid annotations usingsubsystems technology. BMC Genomics 9:75. http://dx.doi.org/10.1186/1471-2164-9-75.

40. Seshadri R, Joseph SW, Chopra AK, Sha J, Shaw J, Graf J, Haft D, WuM, Ren Q, Rosovitz MJ, Madupu R, Tallon L, Kim M, Jin S, Vuong H,Stine OC, Ali A, Horneman AJ, Heidelberg JF. 2006. Genome sequenceof Aeromonas hydrophila ATCC 7966T: jack of all trades. J. Bacteriol.188:8272– 8282. http://dx.doi.org/10.1128/JB.00621-06.

41. Michel C. 1979. Furunculosis of salmonids: vaccination attempts in

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4181

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

rainbow trout (Salmo gairdneri) by formalin-killed germs. Ann. Rech.Vet. 10:33– 40.

42. Grim CJ, Kozlova EV, Sha J, Fitts EC, van Lier CJ, Kirtley ML,Joseph SJ, Read TD, Burd EM, Tall BD, Joseph SW, Horneman AJ,Chopra AK, Shak JR. 2013. Characterization of Aeromonas hydrophilawound pathotypes by comparative genomic and functional analyses ofvirulence genes. mBio 4:e00064-13. http://dx.doi.org/10.1128/mBio.00064-13.

43. Wu CJ, Wang HC, Chen CS, Shu HY, Kao AW, Chen PL, Ko WC.2012. Genome sequence of a novel human pathogen, Aeromonasaquariorum. J. Bacteriol. 194:4114 – 4115. http://dx.doi.org/10.1128/JB.00621-12.

44. Li Y, Liu Y, Zhou Z, Huang H, Ren Y, Zhang Y, Li G, Wang L. 2011.Complete genome sequence of Aeromonas veronii strain B565. J. Bacte-riol. 193:3389 –3390. http://dx.doi.org/10.1128/JB.00347-11.

45. Beatson SA, das Gracas de Luna M, Bachmann NL, Alikhan NF,Hanks KR, Sullivan MJ, Wee BA, Freitas-Almeida AC, Dos Santos PA,de Melo JT, Squire DJ, Cunningham AF, Fitzgerald JR, Henderson IR.2011. Genome sequence of the emerging pathogen Aeromonas caviae. J.Bacteriol. 193:1286 –1287. http://dx.doi.org/10.1128/JB.01337-10.

46. Overbeek R, Begley T, Butler RM, Choudhuri JV, Chuang HY, Co-hoon M, de Crecy-Lagard V, Diaz N, Disz T, Edwards R, Fonstein M,Frank ED, Gerdes S, Glass EM, Goesmann A, Hanson A, Iwata-ReuylD, Jensen R, Jamshidi N, Krause L, Kubal M, Larsen N, Linke B,McHardy AC, Meyer F, Neuweger H, Olsen G, Olson R, Osterman A,Portnoy V, Pusch GD, Rodionov DA, Ruckert C, Steiner J, Stevens R,Thiele I, Vassieva O, Ye Y, Zagnitko O, Vonstein V. 2005. The sub-systems approach to genome annotation and its use in the project toannotate 1000 genomes. Nucleic Acids Res. 33:5691–5702. http://dx.doi.org/10.1093/nar/gki866.

47. Chun J, Grim CJ, Hasan NA, Lee JH, Choi SY, Haley BJ, Taviani E,Jeon YS, Kim DW, Brettin TS, Bruce DC, Challacombe JF, Detter JC,Han CS, Munk AC, Chertkov O, Meincke L, Saunders E, Walters RA,Huq A, Nair GB, Colwell RR. 2009. Comparative genomics revealsmechanism for short-term and long-term clonal transitions in pandemicVibrio cholerae. Proc. Natl. Acad. Sci. U. S. A. 106:15442–15447. http://dx.doi.org/10.1073/pnas.0907787106.

48. Grim CJ, Kotewicz ML, Power KA, Gopinath G, Franco AA, Jarvis KG,Yan QQ, Jackson SA, Sathyamoorthy V, Hu L, Pagotto F, Iversen C,Lehner A, Stephan R, Fanning S, Tall BD. 2013. Pan-genome analysisof the emerging foodborne pathogen Cronobacter spp. suggests a species-level bidirectional divergence driven by niche adaptation. BMC Genom-ics 14:366. http://dx.doi.org/10.1186/1471-2164-14-366.

49. Richter M, Rossello-Mora R. 2009. Shifting the genomic gold standardfor the prokaryotic species definition. Proc. Natl. Acad. Sci. U. S. A.106:19126 –19131. http://dx.doi.org/10.1073/pnas.0906412106.

50. Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S. 2011.MEGA5: molecular evolutionary genetics analysis using maximum like-lihood, evolutionary distance, and maximum parsimony methods. Mol.Biol. Evol. 28:2731–2739. http://dx.doi.org/10.1093/molbev/msr121.

51. Carver TJ, Rutherford KM, Berriman M, Rajandream MA, Barrell BG,Parkhill J. 2005. ACT: the Artemis comparison tool. Bioinformatics21:3422–3423. http://dx.doi.org/10.1093/bioinformatics/bti553.

52. Li L, Stoeckert CJ, Jr, Roos DS. 2003. OrthoMCL: identification ofortholog groups for eukaryotic genomes. Genome Res. 13:2178 –2189.http://dx.doi.org/10.1101/gr.1224503.

53. Edgar RC. 2004. MUSCLE: a multiple sequence alignment method withreduced time and space complexity. BMC Bioinformatics 5:113. http://dx.doi.org/10.1186/1471-2105-5-113.

54. Talavera G, Castresana J. 2007. Improvement of phylogenies after re-moving divergent and ambiguously aligned blocks from protein se-quence alignments. Syst. Biol. 56:564 –577. http://dx.doi.org/10.1080/10635150701472164.

55. Felsenstein J. 1988. Phylogenies from molecular sequences: inferenceand reliability. Annu. Rev. Genet. 22:521–565. http://dx.doi.org/10.1146/annurev.ge.22.120188.002513.

56. Felsenstein J. 1989. Mathematics vs. evolution: mathematical evolution-ary theory. Science 246:941–942. http://dx.doi.org/10.1126/science.246.4932.941.

57. O’Toole GA, Kolter R. 1998. Initiation of biofilm formation in Pseu-domonas fluorescens WCS365 proceeds via multiple, convergent signal-ling pathways: a genetic analysis. Mol. Microbiol. 28:449 – 461. http://dx.doi.org/10.1046/j.1365-2958.1998.00797.x.

58. Khajanchi BK, Kozlova EV, Sha J, Popov VL, Chopra AK. 2012. Thetwo-component QseBC signalling system regulates in vitro and in vivovirulence of Aeromonas hydrophila. Microbiology 158:259 –271. http://dx.doi.org/10.1099/mic.0.051805-0.

59. Morohoshi T, Shiono T, Takidouchi K, Kato M, Kato N, Kato J, IkedaT. 2007. Inhibition of quorum sensing in Serratia marcescens AS-1 bysynthetic analogs of N-acylhomoserine lactone. Appl. Environ. Micro-biol. 73:6339 – 6344. http://dx.doi.org/10.1128/AEM.00593-07.

60. Erova TE, Pillai L, Fadl AA, Sha J, Wang S, Galindo CL, Chopra AK.2006. DNA adenine methyltransferase influences the virulence of Aero-monas hydrophila. Infect. Immun. 74:410 – 424. http://dx.doi.org/10.1128/IAI.74.1.410-424.2006.

61. Sha J, Kozlova EV, Chopra AK. 2002. Role of various enterotoxins inAeromonas hydrophila-induced gastroenteritis: generation of entero-toxin gene-deficient mutants and evaluation of their enterotoxic activity.Infect. Immun. 70:1924 –1935. http://dx.doi.org/10.1128/IAI.70.4.1924-1935.2002.

62. Khajanchi BK, Sha J, Kozlova EV, Erova TE, Suarez G, Sierra JC,Popov VL, Horneman AJ, Chopra AK. 2009. N-Acylhomoserine lac-tones involved in quorum sensing control the type VI secretion system,biofilm formation, protease production, and in vivo virulence in a clin-ical isolate of Aeromonas hydrophila. Microbiology 155:3518 –3531. http://dx.doi.org/10.1099/mic.0.031575-0.

63. Sierra JC, Suarez G, Sha J, Baze WB, Foltz SM, Chopra AK. 2010.Unraveling the mechanism of action of a new type III secretion systemeffector AexU from Aeromonas hydrophila. Microb. Pathog. 49:122–134.http://dx.doi.org/10.1016/j.micpath.2010.05.011.

64. Lory S, Strom MS, Johnson K. 1988. Expression and secretion of thecloned Pseudomonas aeruginosa exotoxin A by Escherichia coli. J. Bacte-riol. 170:714 –719.

65. Liu PV. 1973. Exotoxins of Pseudomonas aeruginosa. I. Factors that in-fluence the production of exotoxin A. J. Infect. Dis. 128:506 –513.

66. Bjorn MJ, Iglewski BH, Ives SK, Sadoff JC, Vasil ML. 1978. Effect ofiron on yields of exotoxin A in cultures of Pseudomonas aeruginosa PA-103. Infect. Immun. 19:785–791.

67. Choi KH, Gaynor JB, White KG, Lopez C, Bosio CM, Karkhoff-Schweizer RR, Schweizer HP. 2005. A Tn7-based broad-range bacterialcloning and expression system. Nat. Methods 2:443– 448. http://dx.doi.org/10.1038/nmeth765.

68. Sha J, Rosenzweig JA, Kirtley ML, van Lier CJ, Fitts EC, Kozlova EV,Erova TE, Tiner BL, Chopra AK. 2013. A non-invasive in vivo imagingsystem to study dissemination of bioluminescent Yersinia pestis CO92 ina mouse model of pneumonic plague. Microb. Pathog. 55:39 –50. http://dx.doi.org/10.1016/j.micpath.2012.09.011.

69. Shak JR, Whitaker JA, Ribner BS, Burd EM. 2011. Aminoglycoside-resistant Aeromonas hydrophila as part of a polymicrobial infection fol-lowing a traumatic fall into freshwater. J. Clin. Microbiol. 49:1169 –1170.http://dx.doi.org/10.1128/JCM.01949-10.

70. Reith ME, Singh RK, Curtis B, Boyd JM, Bouevitch A, Kimball J,Munholland J, Murphy C, Sarty D, Williams J, Nash JH, Johnson SC,Brown LL. 2008. The genome of Aeromonas salmonicida subsp. salmoni-cida A449: insights into the evolution of a fish pathogen. BMC Genomics9:427. http://dx.doi.org/10.1186/1471-2164-9-427.

71. Worby CJ, Lipsitch M, Hanage WP. 2014. Within-host bacterial diver-sity hinders accurate reconstruction of transmission networks fromgenomic distance data. PLoS Comput. Biol. 10:e1003549. http://dx.doi.org/10.1371/journal.pcbi.1003549.

72. Iglewski BH, Liu PV, Kabat D. 1977. Mechanism of action of Pseudomo-nas aeruginosa exotoxin A: adenosine diphosphate-ribosylation of mam-malian elongation factor 2 in vitro and in vivo. Infect. Immun. 15:138 –144.

73. McClean KH, Winson MK, Fish L, Taylor A, Chhabra SR, Camara M,Daykin M, Lamb JH, Swift S, Bycroft BW, Stewart GS, Williams P.1997. Quorum sensing and Chromobacterium violaceum: exploitation ofviolacein production and inhibition for the detection of N-acylhomoserine lactones. Microbiology 143(Pt 12):3703–3711. http://dx.doi.org/10.1099/00221287-143-12-3703.

74. Pukatzki S, McAuley SB, Miyata ST. 2009. The type VI secretion sys-tem: translocation of effectors and effector-domains. Curr. Opin. Micro-biol. 12:11–17. http://dx.doi.org/10.1016/j.mib.2008.11.010.

75. Sierra JC, Suarez G, Sha J, Foltz SM, Popov VL, Galindo CL, GarnerHR, Chopra AK. 2007. Biological characterization of a new type IIIsecretion system effector from a clinical isolate of Aeromonas hydro-

Grim et al.

4182 aem.asm.org Applied and Environmental Microbiology

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from

phila—part II. Microb. Pathog. 43:147–160. http://dx.doi.org/10.1016/j.micpath.2007.05.003.

76. Gray GL, Smith DH, Baldridge JS, Harkins RN, Vasil ML, Chen EY,Heyneker HL. 1984. Cloning, nucleotide sequence, and expression inEscherichia coli of the exotoxin A structural gene of Pseudomonas aerugi-nosa. Proc. Natl. Acad. Sci. U. S. A. 81:2645–2649. http://dx.doi.org/10.1073/pnas.81.9.2645.

77. Matsui H, Sekiya Y, Nakamura M, Murayama SY, Yoshida H, Taka-hashi T, Imanishi K, Tsuchimoto K, Uchiyama T, Sunakawa K, Ubu-kata K. 2009. CD46 transgenic mouse model of necrotizing fasciitiscaused by Streptococcus pyogenes infection. Infect. Immun. 77:4806 –4814. http://dx.doi.org/10.1128/IAI.00577-09.

78. Borger van der Burg BL, Bronkhorst MW, Pahlplatz PV. 2006. Aero-monas hydrophila necrotizing fasciitis. A case report. J. Bone Joint Surg.Am. 88:1357–1360. http://dx.doi.org/10.2106/JBJS.C.00923.

79. Chern CH, How CK, Huang LJ. 2006. Images in emergency medi-cine. Necrotizing fasciitis caused by Aeromonas hydrophila. Ann.Emerg. Med. 48:216 –225. http://dx.doi.org/10.1016/j.annemergmed.2005.12.027.

80. Psoinos CM, Flahive JM, Shaw JJ, Li Y, Ng SC, Tseng JF, Santry HP.2013. Contemporary trends in necrotizing soft-tissue infections in theUnited States. Surgery 153:819 – 827. http://dx.doi.org/10.1016/j.surg.2012.11.026.

81. Salcido RS. 2007. Necrotizing fasciitis: reviewing the causes and treat-ment strategies. Adv. Skin Wound Care 20:288 –293, quiz 294 –295. http://dx.doi.org/10.1097/01.ASW.0000269317.76380.3b.

82. Keung EZ, Liu X, Nuzhad A, Adams C, Ashley SW, Askari R. 2013.Immunocompromised status in patients with necrotizing soft-tissue in-fection. JAMA Surg. 148:419 – 426. http://dx.doi.org/10.1001/jamasurg.2013.173.

83. Al Shukry S, Ommen J. 2013. Necrotizing fasciitis—report of ten casesand review of recent literature. J. Med. Life 6:189 –194.

84. Johansson L, Thulin P, Low DE, Norrby-Teglund A. 2010. Gettingunder the skin: the immunopathogenesis of Streptococcus pyogenes deeptissue infections. Clin. Infect. Dis. 51:58 – 65. http://dx.doi.org/10.1086/653116.

85. Skwor T, Shinko J, Augustyniak A, Gee C, Andraso G. 2014. Aeromo-nas hydrophila and Aeromonas veronii predominate among potentiallypathogenic ciprofloxacin- and tetracycline-resistant Aeromonas isolatesfrom Lake Erie. Appl. Environ. Microbiol. 80:841– 848. http://dx.doi.org/10.1128/AEM.03645-13.

86. Ibekwe AM, Lyon SR. 2008. Microbiological evaluation of fecal bacterialcomposition from surface water through aquifer sand material. J. WaterHealth 6:411– 421. http://dx.doi.org/10.2166/wh.2008.058.

87. Horneman AJ, Morris JG. 2012. Aeromonas. In Rose BD (ed), UpTo-Date: clinical reference library for physicians. Wolters Kluwer Health,Wellesley, MA.

88. Moormeier DE, Endres JL, Mann EE, Sadykov MR, Horswill AR, RiceKC, Fey PD, Bayles KW. 2013. Use of microfluidic technology to ana-lyze gene expression during Staphylococcus aureus biofilm formation re-

veals distinct physiological niches. Appl. Environ. Microbiol. 79:3413–3424. http://dx.doi.org/10.1128/AEM.00395-13.

89. Kozlova EV, Popov VL, Sha J, Foltz SM, Erova TE, Agar SL, Horne-man AJ, Chopra AK. 2008. Mutation in the S-ribosylhomocysteinase(luxS) gene involved in quorum sensing affects biofilm formation andvirulence in a clinical isolate of Aeromonas hydrophila. Microb. Pathog.45:343–354. http://dx.doi.org/10.1016/j.micpath.2008.08.007.

90. Lynch MJ, Swift S, Kirke DF, Keevil CW, Dodd CE, Williams P. 2002.The regulation of biofilm development by quorum sensing in Aeromonashydrophila. Environ. Microbiol. 4:18 –28. http://dx.doi.org/10.1046/j.1462-2920.2002.00264.x.

91. Canals R, Altarriba M, Vilches S, Horsburgh G, Shaw JG, Tomas JM,Merino S. 2006. Analysis of the lateral flagellar gene system of Aeromonashydrophila AH-3. J. Bacteriol. 188:852– 862. http://dx.doi.org/10.1128/JB.188.3.852-862.2006.

92. Caiazza NC, Merritt JH, Brothers KM, O’Toole GA. 2007. Inverseregulation of biofilm formation and swarming motility by Pseudomonasaeruginosa PA14. J. Bacteriol. 189:3603–3612. http://dx.doi.org/10.1128/JB.01685-06.

93. Shrout JD, Chopp DL, Just CL, Hentzer M, Givskov M, Parsek MR.2006. The impact of quorum sensing and swarming motility on Pseu-domonas aeruginosa biofilm formation is nutritionally conditional. Mol.Microbiol. 62:1264 –1277. http://dx.doi.org/10.1111/j.1365-2958.2006.05421.x.

94. Murray TS, Ledizet M, Kazmierczak BI. 2010. Swarming motility,secretion of type 3 effectors and biofilm formation phenotypes exhibitedwithin a large cohort of Pseudomonas aeruginosa clinical isolates. J. Med.Microbiol. 59:511–520. http://dx.doi.org/10.1099/jmm.0.017715-0.

95. Wood TK, Gonzalez Barrios AF, Herzberg M, Lee J. 2006. Motilityinfluences biofilm architecture in Escherichia coli. Appl. Microbiol. Bio-technol. 72:361–367. http://dx.doi.org/10.1007/s00253-005-0263-8.

96. Daddaoua A, Fillet S, Fernandez M, Udaondo Z, Krell T, Ramos JL.2012. Genes for carbon metabolism and the ToxA virulence factor inPseudomonas aeruginosa are regulated through molecular interactions ofPtxR and PtxS. PLoS One 7:e39390. http://dx.doi.org/10.1371/journal.pone.0039390.

97. Stamenkovic I, Lew PD. 1984. Early recognition of potentially fatalnecrotizing fasciitis. The use of frozen-section biopsy. N. Engl. J. Med.310:1689 –1693.

98. Johansson L, Linner A, Sunden-Cullberg J, Haggar A, Herwald H,Lore K, Treutiger CJ, Norrby-Teglund A. 2009. Neutrophil-derivedhyperresistinemia in severe acute streptococcal infections. J. Immunol.183:4047– 4054. http://dx.doi.org/10.4049/jimmunol.0901541.

99. Lacy P. 2006. Mechanisms of degranulation in neutrophils. AllergyAsthma Clin. Immunol. 2:98 –108. http://dx.doi.org/10.1186/1710-1492-2-3-98.

100. Lintges M, van der Linden M, Hilgers RD, Arlt S, Al-Lahham A, ReinertRR, Plucken S, Rink L. 2010. Superantigen genes are more important thanthe emm type for the invasiveness of group A Streptococcus infection. J. Infect.Dis. 202:20–28. http://dx.doi.org/10.1086/653082.

Necrotizing Fasciitis and A. hydrophila

July 2014 Volume 80 Number 14 aem.asm.org 4183

on April 10, 2021 by guest

http://aem.asm

.org/D

ownloaded from