Identification of Differentially Abundant Proteins of Edwardsiella ictaluri during Iron Restriction

Edwardsiella ictaluri is a Gram-negative facultative anaerobe intracellular bacterium that causes enteric septicemia in channel catfish. Iron is an essential inorganic nutrient of bacteria and is crucial for bacterial invasion. Reduced availability of iron by the host may cause significant stress for bacterial pathogens and is considered a signal that leads to significant alteration in virulence gene expression. However, the precise effect of iron-restriction on E. ictaluri protein abundance is unknown. The purpose of this study was to identify differentially abundant proteins of E. ictaluri during in vitro iron-restricted conditions. We applied two-dimensional difference in gel electrophoresis (2D-DIGE) for determining differentially abundant proteins and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI TOF/TOF MS) for protein identification. Gene ontology and pathway-based functional modeling of differentially abundant proteins was also conducted. A total of 50 unique differentially abundant proteins at a minimum of 2-fold (p ≤ 0.05) difference in abundance due to iron-restriction were detected. The numbers of up- and down-regulated proteins were 37 and 13, respectively. We noted several proteins, including EsrB, LamB, MalM, MalE, FdaA, and TonB-dependent heme/hemoglobin receptor family proteins responded to iron restriction in E. ictaluri.

Iron is an essential micro element for almost all living organisms and is involved in various metabolic processes like sugar, protein, energy, and DNA metabolism, growth, and response to oxidative stress [21]. Reduced availability of iron may cause significant stress for bacterial pathogens and is considered a signal that leads to significant changes in gene expression [22] Vertebrate hosts tend to chelate free iron using high affinity proteins like ferritin, transferrin, and heme proteins, which restricts iron availability for bacteria [23,24]. This innate mechanism of iron-restriction by the host is an important host defense mechanism against bacterial infection [25,26]. In turn, low levels of iron in the environment often trigger virulence factor expression in pathogens [27]. In many Gram-negative bacteria, iron associates with ferric uptake regulator (Fur) to regulate expression of virulence genes [28]. Based on this phenomenon, a significant number of potential virulence genes have been identified in E. coli [29,30], E. ictaluri [31], Shigella dysenteriae [32], Vibrio cholera [33][34][35][36], Neisseria meningitidis [37], and Pseudomonas aeruginosa [38][39][40].
High throughput proteomics methods have the potential to accelerate discovery of virulence determinants of E. ictaluri. Previously, we analyzed and annotated the sub-proteome of E. ictaluri strain 93-146 [41]. We now report how the E. ictaluri sub-proteome responds when grown under iron-restricted conditions. This information has the potential to elucidate mechanisms of ESC pathogenesis at the molecular level.
Six bacterial pellets (three control and three treatment) were washed three times using standard cell wash buffer (10 mM TRIS hydrochloride (Tris-HCl) and 5 mM magnesium acetate) at 30°C and were suspended in 750 uL of urea-CHAPS buffer (8 M urea, 30 mM Tris-HCl, 4% 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 8 mM phenylmethanesulfonyl fluoride pH 8.0). Bacteria were lysed on ice by applying ten intermittent pulses of 10 s with a sonicator, and cellular debris was removed by centrifugation at 4°C at 20,817 x g for 5 min.
Proteins from supernatant were precipitated by trichloroacetic acid/acetone, and the resultant protein pellets were suspended in urea-CHAPS buffer. The pH of the lysates was adjusted to 8.5 using 50 mM sodium hydroxide. Protein concentrations were estimated using a 2-D Quant Kit (GE Healthcare, Piscataway, NJ) following the manufacturer's instructions.

Labeling of proteins
Protein samples were labeled using a CyDye difference in-gel electrophoresis (DIGE) Fluor minimal labeling kit (GE Healthcare) according to the manufacturer's manual. Briefly, 50 μg of protein from an internal standard (equal mixture (8.33 μg) of all 6 samples), control, and treatment were mixed with 400 pmol of Cy2, Cy3, or Cy5 dyes, respectively, and protein-dye mixtures were incubated on ice in the dark for 30 min. Labeling reaction was terminated by adding 1 μl 10 mM lysine, mixing well, and incubating samples in the dark for 10 min.
Protein separation using two-dimensional gel electrophoresis  For isoelectric focusing (IEF), precast 17 cm pH 3-10 NL immobilized pH gradient (IPG) strips (Bio-Rad, Hercules, CA) were used. Each of the labeled protein samples from control, treatment, and internal standard were combined with rehydration buffer containing 7 M urea, 2 M thio urea, 4% CHAPS, 1:50 carrier ampholyte, and 2% DTT. Mixed samples were loaded onto each IPG strip for in-gel rehydration. IEF was performed in a Protean IEF cell (Bio-Rad, Hercules, CA) in the dark at 23°C, 500 V for 15 min; linear ramp to 10,000 V for 3 h; and 10,000 V until a total of 70,000 Vh was reached. After IEF, IPG strips were equilibrated in 6 M urea, 30% glycerol, 50 mM Tris-HCl, 2% sodium dodecyl sulfate (SDS), 2% dithiothreitol (DTT), at pH 8.8, and with a trace of bromophenol blue for 15-20 min followed by equilibration containing 2.5% iodoacetamide (IAA) instead of 2% DTT for 15-20 min. Once equilibrated, strips were transferred onto 12% SDS-polyacrylamide gel electrophoresis gels (Jule Inc., Milford, CT) and sealed with 0.5% agarose in electrophoresis buffer. Electrophoresis was performed using a PROTEAN II XL system (Bio-Rad) at a constant current of 10 mA/gel for the first 15 min followed by 24 mA/gel at 20°C until the dye front reach the lower end of the gel.

Analysis of 2-D DIGE gel images
After electrophoresis, DIGE gels were scanned using a Typhoon 9410 imager (GE Healthcare). Excitation and emission filters used for each dye were as follows: Cy2 (488 nm/520 nm), Cy3 (532 nm/580 nm), and Cy5 (633 nm/670 nm). Acquired images were analyzed using DeCyder 5.0 software (GE Healthcare). Briefly, spots were detected using the differential in-gel analysis (DIA) module. Spot matching between gels and statistical analysis of protein-abundance changes were conducted using the biological variation analysis (BVA) module. Among the three replicates, the gel with the highest number of spots was assigned as the master gel. All the spots that were matched automatically were also manually compared among all 3 replicate gels to minimize false spot matching. Statistical significance was calculated using the Student's ttest with applied false discovery rate and a significance threshold of p < 0.05. Only spots showing at least 2-fold change in spot intensity and were consistent in all three replicate gels were considered as differentially abundant and chosen for protein identification.

Preparative gel electrophoresis and protein identification
Preparative 2-DE gels were prepared exactly as described above, except that the IPG strips were loaded with 500 μg of protein. Resultant gels were stained using Deep Purple Total Protein Stain (GE Healthcare) according to the manufacturer's protocols. Briefly, gels were fixed overnight in 15% v/v ethanol and 1% w/v citric acid followed by staining for 1 h in 1:200 parts of Deep Purple and 100 mM sodium borate solution at pH 10.5. Gels were then washed for 30 min with 15% v/v ethanol in water, acidified for 30 min using a solution containing 15% v/v ethanol and 1% w/v citric acid. Stained gels were scanned using a Typhoon 9410 imager using a 532 nm laser and a 610 nm BP30 emission filter. In-gel trypsin digestion and MALDI peptide mass fingerprinting (PMF) was performed as previously described [46] with slight modifications (mass tolerance value was 150 ppm and E. ictaluri protein database was used).

Functional modeling of differentially abundant proteins
We used gene ontology (GO) resources GORetriever and GOanna (available at AgBase) [47] for obtaining biological process and molecular functional annotations of differentially abundant proteins. Using GORetriever, we obtained all existing GO annotations for proteins. Proteins with no existing GO annotations but with a sequence similarity of >80% with presumptive orthologs were annotated using Goanna. Obtained GO biological process and molecular function annotations were manually summarized to more generalized categories based on the ancestor chart for GO terms at QuickGO [48]. The subcellular locations of differentially abundant proteins were predicted using PSORTb v3.0.0 [49]. To gain insight into various biological pathways that were significantly represented by our differentially abundant proteins, we used Pathway Studio 6.0 (Ariadne, Rockville, MD) as previously reported [41]. In addition, "build pathway" function was used to build a biological interactions network of up-and down-regulated proteins.

Identification of differentially abundant proteins
The DIGE analysis detected approximately 2,200 spots in each replicate, and after automatic matching and manual verification of each spot, only those spots that were matched in all 3 replicate gels were subjected to statistical analysis. Analysis of these spots revealed that 131 spots (92 up-and 39 down-regulated) were differentially abundant with a minimum of 2-fold, p < 0.05 in iron-restricted conditions compared to bacteria grown in regular BHI media. Among the 131 spots, 71 spots (54 up-and 17 down-regulated) matched to a preparative gel were cut for mass spectrometric analysis, and 65 (91.54%) positive identifications with confidence intervals > 99% were identified (Fig 1). Fifteen proteins were represented in more than of Edwardsiella ictaluri grown in iron-rich and iron-restricted conditions. Fifty μg of soluble protein from E. ictaluri grown in regular brain heart infusion (BHI) media was labeled with Cy3, grown in BHI with chelator 2, 2'-dipyridyl was labeled with Cy5, and the pooled internal standard labelled with Cy2. Spots shown in red and green arrow head are up-and down-regulated (2 fold), respectively. 3D images of 2 spots with maximum and minimum fold up-regulated proteins were shown on left top and bottom corners of gel image, respectively. 3D images of 2 spots with maximum and minimum fold down-regulated proteins were shown on right top and bottom corners of gel image, respectively. doi:10.1371/journal.pone.0132504.g001 one spot, probably due to migration of abundant proteins to more than one spot or post-translational modifications and processing. In conclusion, we were able to determine 50 (37 up-and 13 down-regulated) unique differentially abundant E. ictaluri proteins under in vitro ironrestricted conditions (Table 1). Notable among these were EsrB, LamB, MalM, MalE, Fda, AspA, DsbA, OmpA, OppA, and TonB-dependent heme/hemoglobin receptor family protein.
Functional modeling of differentially abundant proteins GO annotation of the 50 unique differentially abundant proteins and manual slimming based on GO terms resulted in 14 biological process (Fig 2) and 14 molecular function (Fig 3) categories. Up-regulated proteins were represented in 12 biological processes, whereas downregulated proteins were represented only in 7 biological processes. The top three biological process categories represented by higher numbers of up-regulated proteins were carbohydrate metabolic process, oxidation reduction, and cellular metabolic process. Two of these categories (cellular metabolic process and oxidation reduction) were also among the top three biological processes involving down-regulated proteins. Interestingly, carbohydrate metabolic processes, which included the highest number of up-regulated proteins, did not include any down-regulated proteins.
Up-regulated proteins were represented in all 14 molecular functional categories, whereas down-regulated proteins were represented only in 9 molecular functional categories. In the molecular function grouping, transferase activity, metal ion binding, and hydrolase activity were the top three categories represented by up-regulated proteins. Only four down-regulated proteins were in these groups, while most (11/13) down-regulated proteins were categorized under oxidoreductase activity, metal ion binding, and nucleotide binding.
Subcellular locations of differentially abundant proteins were predicted using PSORTb (Fig 4). Higher numbers of up-and down-regulated proteins were predicted to be located in the cytoplasm and periplasm, excluding those proteins of unknown location.
Pathways with significant representation of differentially abundant proteins were determined (p < 0.05). Ten pathways related to carbohydrate, amino acid, lipid, and nucleotide metabolism were significantly represented ( Table 2). We used a pathway reconstruction algorithm, "Build Pathway" available in Pathway Studio, to analyze the shortest paths of up-and down-regulated proteins with biological interactions such as binding interactions, post-translational regulation, and abundance regulation. Cellular processes such as pathogenesis, virulence, secretion, biofilm, motility, regulation of signal transduction, protein folding, glycolysis, gluconeogenesis, growth rate, catabolism, transcription termination, respiration, proteolysis, apoptosis, and cell survival were predominantly represented in the differential protein abundance interaction network (Fig 5).

Discussion
The purpose of the present study was to identify differential abundance in proteins of E. ictaluri grown under iron-restricted and normal growth conditions and investigate their possible role in pathogenesis. We identified 50 unique E. ictaluri proteins with altered abundance (37 upand 13 down-regulated) in response to iron-restriction. It is known that iron is an essential micronutrient that acts as a cofactor for enzymes involved in oxidative and electron transport processes. Hence, iron is essential for pathogenic bacteria to establish an infection.
Iron uptake in bacteria is controlled tightly by the ferric uptake regulator (fur) gene [28,50], and it has been shown that the E. ictaluri fur gene has a similar regulatory function [31]. Transport proteins, especially cation transporters, are highly expressed in iron-restricted conditions. TonB-dependent heme/hemoglobin receptor family protein, with its 18-fold higher abundance in iron-restricted growth conditions, may act as a crucial factor in iron uptake as part of the E. ictaluri hemPRSTUV operon. Recently, an up-regulation of the E. ictaluri TonB-dependent heme/hemoglobin receptor in iron limited conditions has also been reported [31]. TonBdependent heme/hemoglobin receptor family protein in E. ictaluri might have both receptor and transporter activity along with its involvement in transduction of environmental signals, and a possible role in pathogenicity similar to several bacterial pathogens [31,51,52]. Research in Vibrio alginolyticus demonstrated that mutants of TonB complex exhibited attenuation in virulence compared to wild-type in zebrafish (Danio rerio) [53]. Similarly, maltoporin (LamB), a member of the sugar porin family, aid in transport of maltose and other maltodextrins across the outer membrane in E. coli [54]. Mutational studies of the lamB gene of enteropathogenic E. coli showed that mutants were deficient in adherence to HEp-2 cells [55]. It was also shown that LamB is an important outer membrane protein in E. coli for obtaining tetracycline resistance [56]. Based on our understanding, proteins involved in the acquisition of iron are closely associated with virulence of several bacteria; it is likely that transporter proteins identified in the present study might be important in E. ictaluri pathogenesis. Translational proteins like elongation factor Tu (EF-Tu), a three-domain GTPase, is crucial during the elongation phase of mRNA translation. The EF-Tu in complex with GTP and aminoacyl-tRNA delivers tRNA to the ribosome. It is known that EF-Tu might play a role in protein-folding during stress [57]. It is also proposed that EF-Tu might sense and respond to stress [58]. Hence, EF-Tu may assume the role of translational regulation allowing it to trigger the synthesis of stress-induced proteins and to thwart the translation of unnecessary proteins. During starvation/stress in E. coli, EF-Tu was shown to be methylated and become membrane associated [59]. Previous research has also shown that EF-Tu might act as a virulence factor in P. aeruginosa [60]. During those conditions EF-Tu may play a possible role in the organism's response to stress and growth regulation, in addition to its primary role in regulation of translation.
N-acetylmuramoyl-L-alanine amidase is an outer membrane lipoprotein which catalyzes cleavage of the bond between muramic acid and L-alanine of murein. Similarly, ADP-Lglycero-D-mannoheptose-6-epimerase is involved in the lipopolysaccharide (LPS) biosynthesis pathway and is responsible for synthesis of the ADP-heptose precursor of core LPS [61]. Glucose-6-phosphate isomerase, phosphoenolpyruvate carboxykinase, hypothetical protein NT01EI_2846, which is also named as Phosphoglyceromutase (GpmA), and Fructose 1, 6-bisphosphate aldolase (FbA) are known to be involved in the glycolysis/gluconeogenesis pathway. Vassinova and Kozyrev, (2000) [62] suggested that transcription of gpmA is regulated by Fur in E. coli. It has also been shown that FbA of E. ictaluri has antigenic properties and is regulated during the infectious process of ESC in catfish [63]. Similarly, research conducted by Ling et al. (2004) [64] also revealed that on respiratory challenge with virulent Streptococcus pneumonia in mice, FbA is able to elicit significant levels of immune response. Succinate dehydrogenase catalyzes the oxidation of succinate to fumerate in the tricarboxylic acid cycle. Aspartate ammonia-lyase is an anerobic enzyme which catalyzes the amination of fumarate to generate L-aspartate. Mutational studies conducted by Jacobsen et al. (2005) [65] showed that aspartate ammonia-lyase plays an important role in the pathogenesis of Actinobacillus pleuropneumoniae in pigs [65].
Research conducted by Wang et al. (2009) [66] confirmed that the conserved hypothetical protein (EsrB) and iron concentrations regulate the E. tarda virulence proteins. Thiol: disulfide interchange protein DsbA, which was -3.35 fold down-regulated, was a protein-folding catalyst which aids in correct folding of surface-presented virulence factors like adherence factors, toxins, and components of the type III secretory system [67]. The observed down-regulation of periplasmic oligopeptide-binding protein (oppA), an ATP-dependent ABC Superfamily of transporters involved in oligopeptide uptake, is in agreement with reduced metabolism of bacteria due to stress caused by the limitation of available iron [68]. Research conducted by Lee et al. (2009) [69] revealed that the dsbA mutant of Pseudomonas putida exhibited enhanced  extracellular matrix production and biofilm formation. Down-regulation of OmpA (4.79 fold), is consistent with findings of Clamydia pneumonia an iron-limitation model [70]. Similarly, superoxide dismutase (Sod_Fe) was shown to be positively regulated by the Fur-Fe + complex in many bacterial species [71,72]. Hence in the present study, down regulation of superoxide dismutase was expected due to growth of E. ictaluri in iron-restricted conditions [73]. GO annotation and manual slimming of up-regulated proteins resulted in a higher number of biological processes (12) and molecular functional categories (14) compared to downregulated proteins (7 and 9, respectively). This was expected as the number of unique proteins that were up-regulated (37) was high compared to those that were down-regulated (13). Upregulated proteins were highly represented in the carbohydrate metabolic process. This might be due to a higher abundance of proteins involved in glycolysis/gluconeogenesis, pyruvate metabolism, and in synthesis of cell wall structures like peptidoglycan and LPS. It has been shown that, bacteria in general, alter their metabolic activity particularly increase their ability to metabolize variety of carbon sources followed by changes in expression pattern of the virulence factors [74,75]. Similarly, during iron scarce situation, iron-dependent pathways in microbes are diminished while iron-independent enzymes and metabolic pathways are enhanced [76]. Down-regulation of few proteins involved in cellular metabolic processes, oxidation reduction, and translational processes could also indicate a possible reduction in the metabolism of E. ictaluri due to iron starvation stress.
A higher number of up-regulated proteins were predicted to be located in cytoplasm as they are hydrophilic and thus do not interfere in 2-DE separation techniques. DsbA, Sod_Fe, periplasmic oligopeptide-binding protein, and conserved hypothetical protein were the four downregulated proteins predicted to be located in periplasm.
Protein interaction networks of differentially abundant proteins were built using Pathway Studio. Differentially abundant proteins were involved in cellular processes like virulence, pathogenesis, secretion, biofilm, and regulation of signal transduction. Up-regulated proteins like FbaA and ChuA and down-regulated proteins like OppA, OmpA, and DsbA were involved in virulence and pathogenesis processes suggesting that differentially abundant proteins during iron-restriction may play an important role in E. ictaluri pathogenesis. Down-regulation of several proteins involved in cellular processes like cell survival, motility, and growth rate may be expected with reduced metabolism due to iron-limitation stress. Furthermore, up-regulation of proteins involved in carbohydrate metabolism, nucleoside/nucleotide metabolism, TCA cycle, and transport process might be due to a part of global iron homeostatic response of E. ictaluri replacing iron dependent enzymes with iron-independent alternatives as exhibited by E. coli and many other bacteria [77][78][79]. Few proteins involved in biosynthesis process, oxidation reduction, translation, and other processes are down regulated. This might be due to E. ictaluri's engagement in an iron-sparing process, similar to E. coli, to conserve its limited iron resources [80,81] It is always a challenge to elucidate how bacteria employs various adaptive mechanisms to invade, colonize, and successfully establish a disease in the host. It is likely that E. ictaluri will encounter several environmental stresses in the gastric environment of catfish like iron starvation and fluctuation in pH during the initial course of its pathogenesis. With an objective to thoroughly elucidate the response of E. ictaluri to iron-restriction, we used 2D-DIGE technology to investigate changes protein abundance. We therefore hypothesized that analysis of E. ictaluri response towards iron-restriction conditions may aid in enlightening the possible mechanisms of pathogenesis of E. ictaluri. In the present study, we noted several differentially abundant proteins that were previously shown to be involved in the pathogenesis of other Gram-negative bacteria including E. ictaluri. Future experiments determining the role of these differentially abundant proteins should provide important information regarding the mechanisms used by E. ictaluri during colonization and establishment of ESC in catfish.