Secretome analysis of diarrhea-inducing strains of Escherichia coli

Abstract

Secreted proteins constitute a major part of virulence factors that are responsible for pathogenesis caused by Gram-negative bacteria. Enterohemorrhagic Escherichia coli, O157:H7, is the major pathogen often causing outbreaks. However, studies have reported that the significant outbreaks caused by non-O157:H7 E. coli strains, also known as “Big-Six” serogroup strains, are increasing. There is no systematic study describing differential secreted proteins from these non-O157:H7 E. coli strains. In this study, we carried out MS-based differential secretome analysis using tandem mass tags labeling strategy of non-O157:H7 E. coli strains, O103, O111, O121, O145, O26, and O45. We identified 1241 proteins, of which 565 proteins were predicted to be secreted. We also found that 68 proteins were enriched in type III secretion system and several of them were differentially expressed across the strains. Additionally, we identified several strain-specific secreted proteins that could be used for developing potential markers for the identification and strain-level differentiation. To our knowledge, this study is the first comparative proteomic study on secretome of E. coli Big-Six serogroup and the several of these strain-specific secreted proteins can be further studied to develop potential markers for identification and strain-level differentiation. Moreover, the results of this study can be utilized in several applications, including food safety, diagnostics of E. coli outbreaks, and detection and identification of bio threats in biodefense.

1 Introduction

The presence of Shiga toxin producing Escherichia coli (STEC) strains in food or water sources are well known to cause various diseases in humans [1]. The pathogenesis of E. coli STEC has been reported to involve “attaching and effacing” mechanisms to the host cells [2, 3]. The recent studies have revealed that various genes associated with the virulence vary among E. coli strains resulting disease outbreaks [4]. The STEC pathogenicity in the host cells is manifested by the pathogen secreting various proteins that compromise the cytoskeletal reorganization of the host cell [5]. The O157:H7 and non-O157:H7 STEC strains known to cause diarrhea, hemolytic uremic syndrome, hemorrhagic colitis, and death if not treated [6]. The STEC including O157:H7 and non-O157:H7 serotypes share common O-group designation and virulence features. While O157 has been implicated in most common food outbreaks, there is mounting evidence that other non-O157 STEC E. coli strains are responsible for significant illness and serious outbreaks across the globe [7, 8]. The top six non-O157:H7 STEC strains that are commonly reported in food outbreaks are O26, O45, O103, O111, O121, and O145, also known as “Big-Six” E. coli group, which can cause over 80% of the total reported non-O157 illnesses [9]. The infection is mainly caused by Shiga toxin, which is encoded by two Shiga toxin producing genes, Stx1 and Stx2, in which Stx2 shares 60% sequence homology with Stx1 [10]. In the United States, it is estimated that non-O157:H5 STEC causes more illness than STEC O157:H7 and it is estimated that 231,157 cases annually caused by these Big-Six strains [11, 12].

The increased incidents of food outbreaks caused by the E. coli Big-Six group of strains have led to the implementation of government regulations for zero tolerance of these strains in food matrices. Studies have addressed the development of effective detection and identification methods to provide practical solution for elimination and spread of STEC strains in food chain of consumers [11]. There are several detection methods for the Big-Six STEC strains that had been reported in the literature, which include optical spectroscopy [13], genomic-based PCR [14], loop-mediated isothermal amplification [15], and antibody-based high-throughput microarray platforms were employed for rapid detection of Big-Six group E. coli strains [16]. Such techniques were effective in detection of the Big-Six STEC strains in various food samples, however, they did not provide a differentiation power among the STEC strains. Fewer proteomic studies have been carried to identify STEC isolates using whole cell lysis method [17]. Recently, we reported E. coli strain level differentiation method in which we used extracellular proteins to differentiate between enterohemorrhagic E. coli (EHEC) and enteroaggregative E. coli (EAEC) strains [18].

This study attempted to expand on the applicability of the proteomic-based strain differentiation method and to provide the relative quantitation of the secreted proteins expression in the ATCC Big-Six STEC strains, that is, O26, O45, O103, O111, O121, and O145 strains. It is imperative to understand the distribution of those secreted proteins in the STEC strains, especially the ones with virulence functions in order to provide effective medical counter measures for clinical treatment. The identification of strain-specific proteins in general can be done using genomic approaches. However, this does not provide complete knowledge about how many proteins are indeed expressed and translated. Thus, the identification of strain-specific proteins using high-resolution MS-based approach is an effective strategy for the strain-level identification and quantification of proteins. We employed isobaric tag based multiplexing strategy for the relative proteomic comparison of the ATCC E. coli Big-Six strains [19]. We also carried out bioinformatics analysis to identify strain-level proteins by mapping the MS-derived data to the genome sequence of the studied E. coli strains. Using this strategy, we quantitatively identified 1241 proteins, of which 565 proteins were found to be secreted as predicted by PSORTb and SecretomeP. We enriched 68 proteins pertaining to type III secretion system (T3SS) and have shown the differential expression of these T3SS proteins within the ATCC Big-Six STEC strains. Further, using proteogenomic analysis, we found several strain-specific secreted proteins that could serve as potential diagnostic markers for this group. To our knowledge, this study is the first comparative proteomic study on secretome of E. coli Big-Six serogroup and the several of these strain-specific secreted proteins can be further studied to develop potential markers for identification and strain-level differentiation. Moreover, the results of this study can be utilized in several applications, including food safety, diagnostics of E. coli outbreaks, and biodefense surveillance in the detection and identification of biothreats.

2 Materials and methods

2.1 Escherichia coli strains

A panel of Big-Six group of ATCC E. coli O111:H8, O145, O26:H11, O103:H11, O45:H2, O121:H19, and K12 MG1655 strains were purchased from ATCC. The complete strain details, genome size, and predicted protein numbers are represented in Supporting Information Table 4. All reagents were purchased from Sigma, unless otherwise specified.

2.2 Culture and harvesting secretome

The E. coli strains O111, O145, O26, O103, O45, O121, and K12 obtained from ATCC (Manassas, VA, USA). The culturing has been carried out in an approved biosafety laboratory level II. A single colony of each E. coli strain was inoculated in a pre-LB medium (MP Biomedicals, Santa Ana, CA, USA) and was allowed to grow overnight until the OD600 reading reaches to 1.0. The preinoculum was subcultured in a 250 mL of DMEM medium (Gibco, Langley, OK, USA) and was agitated at 180 rpm at 37°C until the OD 600 reaches 0.8. The culture supernatant was centrifuged at 2500 rpm for about 10 min to pellet down the bacterial cells and the supernatant, secretome, was filtered through a 0.22 µm filter membrane (Millipore, USA) to get rid of any bacteria, and the filtrate was further concentrated using 3 kDa molecular weight cutoff filters (Millipore) to collect the secreted proteins. The protein amount was measured using bicinchoninic acid assay (Pierce, Waltham, MA, USA) and stored at −80°C until further analysis.

2.3 Protein digestion and TMT labeling

A 100 µg of protein lysate from each strain was subjected to reduction using dithiothreitol at a final concentration of 5 mM at 60°C for 20 min and alkylated using 10 mM iodoacetamide for 20 min at room temperature in dark. Then, the processed protein samples were digested using tryps in (Promega, Madison, WI, USA) in a 1:20 ratio and incubated overnight at 37°C. The tryptic peptides were purified using Sep-Pak^® C18 cartridges (Waters, MA, USA) and lyophilized. The lyophilized tryptic peptides were subjected to tandem mass tag (TMT) labeling according to manufacturer instructions with minor modifications. The peptide digest was dissolved in 50 mM triethyl ammonium bicarbonate buffer pH 8.0. The dissolved peptide was incubated with 0.8 mg of TMT reagent at room temperature for 1 h. The labeling efficiency was determined by pooling 1 µg of amount from each condition and analyzing it on mass spectrometer. The labeled peptides were pooled and fractionated by high-pH basicRPLC (bRPLC) on an Agilent 1100 LC system as described earlier [19]. A total of 96 fractions were collected and concatenated to 12 fractions and then vacuum-dried. The dried peptides were subjected to LC-MS/MS analysis.

2.4 LC-MS/MS analysis

A total of 12 bRPLC fractions analyzed as technical replicates on an LTQ-Orbitrap Elite mass spectrometer (Thermo Electron, Bremen, Germany) interfaced with Easy-nLC II nanoflow LC system (Thermo Scientific, Odense, Denmark). The peptide digests were reconstituted in 0.1% formic acid and loaded onto a trap column (75 µm × 2 cm) packed in-house with Magic C₁8 AQ (Michrom Bioresources, Inc., Auburn, CA, USA). Peptides were resolved on an analytical column (75 µm × 20 cm) at a flow rate of 300 nL/min using a linear gradient of 10–35% solvent B (0.1% formic acid in 95% ACN) over 85 min. The total runtime including sample loading and column reconditioning was 120 min. Data-dependent acquisition with full scans in 350–1700 m/z range was carried out using an Orbitrap mass analyzer at a mass resolution of 120 000 at 400 m/z. Fifteen most intense precursor ions from a survey scan were selected for MS/MS fragmentation using higher energy collisional dissociation fragmentation with 37% normalized collision energy and detected at a mass resolution of 30 000 at 400 m/z. Dynamic exclusion was set for 30 s with a 10 ppm mass window. Internal calibration was carried out using lock mass option (m/z 445.120025) from ambient air [20].

2.5 MS data analysis

A total of 24 LC-MS/MS raw files that are acquired as technical runs were used for MS data analysis for the peptide identification and quantification. The raw data were processed through Proteome Discoverer 2.0.0.802 software suite to generate peak list files for the database searches. A combined strain-specific E. coli database (O111:H8, O145, O26:H11, O103:H11, O45:H2, O121:H19, and K12 MG1655) was created by downloading it from the NCBI containing 20712 protein sequences. Combined SEQUEST and MASCOT search algorithms were used for the peptide identification, and reporter ion quantifier node was enabled for peptide quantification. A dual workflow, processing, and consensus workflows were created with the following parameters: (i) minimum and maximum precursors were selected as 350 and 8000 da, respectively; (ii) trypsin is selected as protease with a maximum of two missed cleavages allowed; (iii) minimum peptide length of seven amino acid residues; (iv) precursor and fragment mass tolerance were set as 20 ppm and 0.1 Da, respectively; (v) oxidation of methionine residue as variable modification and carbamidomethylation of cysteine, peptide N-terminus, and lysine side chain as TMT-reported tag were selected as static modification. Percolator node was used for calculating q-value of identified peptide spectrum matches (PSMs) and peptides for statistical significance. One percent protein-level and peptide-level FDR was used in consensus workflow. The peptide quantification was carried out using reporter ion quantifier, and isolation interference cutoff was set as 20% in order to account for the interference from co-eluting peptides [21]. Only the peptides that fall below this criterion were considered for peptide quantification and further analysis.

2.6 Bioinformatics analysis

In order to identify strain-specific proteins, we generated six-frame translation of all the E. coli genomes used in the current study. Using custom python scripts, identified peptides were mapped to the genome in order to ensure if a given protein exists in the genome of a given strain. This analysis was used to validate the strain-specific proteins identified from the TMT reporter tags. SecretomeP 2.0 (http://www.cbs.dtu.dk/services/SecretomeP/) and PSORTB (http://www.psort.org/) programs were used for the prediction of secreted proteins. GeneE (http://www.broadinstitute.org/cancer/software/GENE-E/) and Perseus software suites were used for the clustering and enrichment analysis [22].

2.7 Data availability

The MS raw data result files have been deposited to the ProteomeXchange consortium, PRIDE (http://proteomecentral.proteomexchange.org) with the dataset identifier“PXD004372”.

3 Results and discussion

3.1 Proteomic analysis of the secreted proteins from the Big-Six E. coli strains

Our goal was to study the secretome differences among the Big-Six E.coli STEC group strains. We employed isobaric-labeling strategy by multiplexing the Big-Six group of strains to compare the relative proteomic expression with the non-pathogenic E. coli K12 using high-resolution Fourier transform MS as shown in Fig. 1. We carried out the deep fractionation using high-pH RP chromatography and prepared 12 fractions by concatenating as described earlier [23]. A total of 12 bRPLC fractions were run twice as technical replicates on Orbitrap Elite mass spectrometer to attain the deeper coverage. Although each bacterial cell can express different number of proteins per cell, we believe that a rough estimate to 1.0E07–1.0E09 cfu will be needed to identify biomarkers with high confidence using a targeted MRM approach. The raw MS data were searched against a combined Refseq ATCC E. coli Big-Six group and K12 database using SEQUEST and MASCOT search algorithms on Proteome Discoverer software suite. We identified a total of 1241 proteins, 4635 unique peptides with 22 470 PSMs from both technical replicates. We applied 1% FDR level for both proteins and peptides. In order to reduce the impact of interference of co-eluting peptides, we only considered peptide spectrum matches with less than 20% isolation interference cutoff for peptide quantification. However, we retained the peptides with more than 20% isolation interference for identification purpose but not for quantification. The complete list of proteins and peptides identified in this study can be seen in Supporting Information Tables 1A and B, and 2A and B. We obtained a substantial coverage of 75% of the identified secreted proteins among the technical replicates (Fig. 2A), which suggests reproducible MS data. Further bioinformatics analysis was carried out to identify the potential secreted protein. SecretomeP [24] and PSORTb [25, 26] predicted 565 of the identified proteins as secreted proteins. Of these secreted proteins, 310 and 255 were categorized into classical and nonclassical proteins, respectively (Fig. 2B).

Figure 1.

MS-based quantitative profiling of E. coli Big-Six group strains secretome. A schematic illustration of E. coli Big-Six group strains. Bacteria were harvested and filtered using 0.22 µm filters and concentrated using 3 kDa cutoff filters. Equal amount of proteins from each bacteria were trypsin-digested, cleaned using Sep-Pak C18 cartridges, and TMTs labeling was carried out. Labeled peptides were pooled and fractionated using bRPLC fractions and analyzed on Orbitrap Elite mass spectrometer.

Figure 2.

Result summary of E. coli Big-Six group strains secretome: (A) Venn diagram depicting the strains of secreted proteins from technical replicates. (B) The detailed summary of protein identifications from combined technical replicate analysis.

3.2 Identification and quantification of secreted proteins in the Big-Six E. coli strains

3.2.1 Classical secretory proteins

In general, pathogenic E. coli bacteria have specialized secretory pathways to mediate cellular signaling by secreting several effectors and virulence factors into the host cells. There are at least seven different secretory systems in prokaryotes, which have been reported [27]. These secretion systems are categorized based on the localization of secretory proteins in Gram-negative bacteria. Some of the proteins span both inner membrane and outer membrane (OM) and few to the OM alone. Of these, the double membrane spanning systems were categorized as T1SS, T2SS, T3SS, T4SS, and T6SS, while the proteins that span the OM alone was referred as T5SS secretion system. Among these secretion systems, T3SS is well studied in several of Gram-negative bacteria including pathogenic E. coli. In this study, by applying high-resolution MS and TMTs labeling, we studied proteomic changes of secretome of Big-Six E. coli group. We identified a total of 1241 proteins based on the PSORTb and SecretomeP results, we categorized proteins that contain cleavable signal peptide sequence as classical secreted proteins [28], the remaining proteins as nonclassical secreted proteins. Of the 565 identified secreted proteins, there were 310 classical secreted proteins that are predicted to contain cleavable signal peptide, of those, 148 proteins were quantified. The complete list of proteins and peptides is provided in Supporting Information files (Supporting Information Tables 1A and B, and 2A and B). Further, we categorized these proteins into various subcategories based on their localization and functional roles. The relative expression profile of the identified proteins would serve the E. coli scientific community for further characterization and better understanding of these proteins in the disease pathogenicity caused by E. coli.

3.2.2 Nonclassical secretory proteins

In Gram-negative bacteria, the secretory proteins are well characterized by the presence of signal peptide sequence that leads the activation of secretory pathway. However, some of the recent studies in both prokaryotes and eukaryotes have reported that proteins that lack signal peptides sequence could still govern the secretion [24, 29, 30]. The first non-classical protein secretion in bacteria was the glutamine synthase secretion reported in Mycobacterium tuberculosis [31]. In this study, we identified 255 proteins as nonclassical secretory proteins, and they lack classical signal peptide cleavage site in the protein sequence (Supporting Information Table 1A and B). We have excluded 676 proteins that were not predicted by SecretomeP and PSORTb analysis and their presence can be explained by the cell lysis before the secretome collection, thus we have not considered them for further analysis [18]. The analysis of protein expression across strains of the nonclassically secreted proteins revealed enrichment of several virulence factors that were overrepresented in E. coli O121:H19 and O45:H2 strains. Proteins related to flagellar assembly are found to be overexpressed in O121:H19 strain as compared to the other strains. The flagellar proteins are primarily exported to their assembly destination in conjunction with the members of type III superfamily [32]. We identified well-known flagellar assembly proteins that include FlgE (Flagellar hook protein), FlgD (Flagellar hook capping protein), Flagellin (FliC), and the rod proteins (FlgB, FlgC, FlgF, and FlgG). We identified another cluster of overexpressed nonclassical secreted proteins in E. coli O45:H2 and O121:H19 strains. This cluster includes intimin, tail protein, secretion protein EasA, tail fiber protein, head decoration protein, and EaeB. The selective enrichment or the overexpression of these proteins suggests that these proteins play a major role in the pathogenesis of O121:H19 and O45:H2.

3.3 Identification of type III secreted proteins in Big-Six E. coli strains

Gram-negative bacteria use specialized secretory systems to deliver toxins and effector proteins into the host cells [27]. T3SS and T4SS are well studied in Gram-negative bacteria. These toxins and effector proteins indeed affect the host cellular functions by attaching and effacing mechanism [33]. T3SS system secretes a large number of diverse proteins, which can be classified based on their function into three groups, early substrates (T3SS needle and inner rod components), intermediate substrates (translocators), and late substrates (effectors) [33]. Due to the limited knowledge of the known T3SS effectors, we used a computationally derived prediction algorithm to identify the potential T3SS effectors for the proteins that are identified in our data [34]. After processing through SIEVE server (http://www.sysbep.org/sieve), we have shown the quantitative relative expression for 68 T3SS proteins in E. coli Big-Six group. The relative abundance of these proteins across the studied E. coli strains is depicted in Fig. 3. Of the 68 T3SS proteins, 30 secreted proteins were predicted to contain signal peptides, while 38 secreted proteins do not have predicted signal peptides. Our findings are in agreement with the reported knowledge about the T3SS proteins in E. coli strains, in which the majority of them do not contain defined signal peptide sequences [35]. This observation could be attributed to the fact that most of the effectors are known to interact with the host cellular machinery during infection. The secreted proteins that are associated with the pathogenesis of the infection are Shiga toxin, flagellin, EaeB, flagellar capping protein, EspA (Supporting Information Fig. 1A), EspD, EspB, EspF, FlgK, intimin receptor Tir (where Tir is translocated intimin receptor), exotoxin, and accessory colonization factor (AcfD) (Supporting Information Fig. 1D). EscF is a locus of enterocyte effacement encoded protein of T3SS system, the core structure of needle proteins (Table 1). The null mutant of EscF was shown to fail to disrupt host cell spreading and attachment to substratum [36]. EscF null mutant was also unable to translocate effector proteins or produce EspA filaments [37]. In this study, EscF protein is found to be overexpressed in O26:H11 and O103:H11, which demonstrates the indispensable role of EscF in pathogenic E. coli. EscF, EspA, and EspD in the secretome are overrepresented in Big-Six group compared to K12 strain. We also identified Tir protein as highly secreted in the extracellular space of Big-Six group. Formation of the actin “pedestals” is required for the attachment of enterohemorrhagic E. coli to intestinal epithelial cells. Tir is an effector, playing a critical role in the pedestal formation [38]. We also identified BolA as part of T3SS protein, which is a transcriptional regulator involved in biofilm formation [39]. HdeB is periplasmic chaperon used as biomarker for the discrimination of E. coli serovar [40]. In the stress condition, HdeB is known to provide survival advantage to non-O157 STEC [41]. In our study, we identified higher expression of BolA and HdeB in four strains of Big-Six group. The Tol–Pal system is a periplasmic protein complex required for maintenance of OM integrity and proper septation during cell division. YbgF plays a critical role in regulation of Tol complex components [42]. We found increase in abundance of YbgF in Big-Six group compared to E. coli K12. Escherichia coli infects epithelial cells by suppressing host inflammatory response through T3SS. Two important T3SS effector proteins involved in this mechanism are NleE and NleC. NleE block NF-κB signaling by inhibiting the kinase enzyme complex IκB degradation and nuclear translocation of the p65 subunit of nuclear factor kappa B cells (NF-κB) [43]. We identified higher levels of NelE in the secretome of five of the six strains. The comparative analysis of Big-Six E. coli strains using the quantitative proteomics approach enabled us to examine the relative expression profiles of T3SS proteins. Interestingly, E. coli O121:H19 and O45:H2 strains appeared to have large number of virulence factors compared to other strains. This could attribute to their relative degree of pathogenicity. Alternatively, strain-specific proteins could be of major benefit in terms of providing strain differentiation and enhancing diagnostic determination. Further investigation into the comparative expression of the virulent factors and their correlation with infection could provide mechanistic understanding of their degree of pathogenicity in host cells.

Figure 3.

Hierarchical clustering of E. coli Big-Six group strains T3SS. The relative protein expression of T3SS system was shown in a heat map. Each row value for a given strain is a log₂ value of fold change with K12. Higher expression of T3SS-secreted proteins was observed in O45:H2 and O121:H19 strains.

Table 1.

Partial list of proteins identified in type III secretion system in E. coli Big-Six group of strains

Accession	Protein description	Unique peptides	PSMs	Biological significance
564848394	Secretion protein EspA	4	7	EspA is a major member of T3SS family along with EspB, EspD, and Tir; these proteins are essential for attaching and effacing (A/E) lesion formation
564832654	Secretion protein EspD	13	13	EspD is essential protein which plays an important role in the formation of translocation pore on the host cellular membrane during infection
566113444	Translocated intimin receptor (Tir)	3	16	Tir also plays an essential role in the formation of A/E lesions in the host cells along with T3SS machinery during the infection
566122287	Type III secretion apparatus protein	2	2	This protein is a YscI/HrpB, C-terminal domain containing protein which is a well-conserved protein in bacterial species which has T3SS system
566115143	Accessory colonization factor (AcfD)	4	8	This protein has peptidase 360 domain that has an homology with Vibrio cholera, which plays a role in intestinal colonization during the infection
566113794	Flagellar hook associated protein (FlgK)	12	12	This protein is essential for flagellar export
566115665	Flagellar capping protein	22	22	This protein has FliD domain that is essential for mucin adhesion process
16129041	Flagellar basal body rod protein FlgG	5	5	This protein has a C-terminal domain with unknown function

3.4 ABC transporters and translocase secreted proteins

We identified several ATP-binding transporters across the studied strains (Supporting Information Table 1A and B). ATP-binding cassette (ABC) transporters are the family of integral membrane proteins that are involved actively in the transport of molecules across membranes [44]. It is well known that bacteria express both prokaryotic- and eukaryotic-type ABC transporters. Prokaryotic-type transporters function as importers and eukaryotic-type transporters function as exporters in bacteria [45]. Most of the proteins that are substrates for these exporters are known to lack N-terminal signal sequence but have an uncleavable signal sequence on their C-termini. We identified 48 ABC transporters in this study. We observed an overexpression of molybdenum ABC transporter substrate proteins in five of six strains and similar observation was seen in spermidine/putrescine ABC transporter binding protein. Glutathione ABC transporter protein is overexpressed in E. coli O111, O26, and in O123. Another ABC family transporter protein, long-chain fatty acid OM transporter, is found to be overexpressed in E. coli O111, O145, O26, and O121. Polyamine transporter subunit is found to be overexpressed only in E. coli O45:H2 of Big-Six group. We observed that several of the identified proteins are periplasmic-binding proteins, which suggest that these ABC transporters might play a major role in shuttling several essential molecules across the periplasmic region of Gram-negative bacteria. Mutations associated with this class of proteins result in impaired function, and thus serve as potential targets for drugs [46]. We also identified several translocase proteins that are well-known substrates for bacterial twin-arginine translocation pathway (Tat system pathway) and Sec-mediated pathways. In our study, we observed upregulation of SecG and SecD subunits in all the Big-Six group strain except in O45:H2. In addition, we have identified two of the Tat-mediated pathway subunits, TatA, TatB, and TatE proteins, whose expressions are observed to be downregulated in the Big-Six group. This observation may suggest that the Sec-mediated pathway is overrepresented in this group than in the Tat-mediated pathway. However, systematic future functional studies are warranted to target these pathways in this pathogenic E. coli group.

3.5 Outer membrane and extracellular proteins

OM proteins (OMPs) play an essential role in the regulation of transport of small molecules and metabolites between bacteria and environment, maintenance of osmotic pressure, and drug resistance [47, 48]. In this study, we identified 47 different OMPs across the Big-Six group E. coli strains. Of the 47 proteins, 31 proteins are found to be overexpressed across the Big-Six group, which include rcsF, OM porin C, BamD, slyB, pal, btuB, tolC, fadL, FepA, ompA, ompT, ompX, ompF, ompW, fhuE, and bamB. The relative expression of the OMPs is shown as a heat map in Fig. 4. We observed that ompT, ompW, ompX, putative TonB-dependent receptor, fepA, fadL, tsx have shown higher expression in O145, O111:H8, and O121:H19. Recent evidence shows that the downregulation of Tsx and ompW and upregulation of ompX are essential for iron homeostasis in E. coli [48]. We also observed overexpression of ompC in O145 and O103:H11 strains. ompC has been shown to be differentially regulated in response to various antibiotics in E. coli [49]. Also, high salt concentration enhances the expression of ompC for the transport of small hydrophilic molecules across OM in E. coli [50]. Catecholate siderophores are important for conferring virulence in Gram-negative bacteria and have antioxidant properties for tight control of iron metabolism [51]. We observed higher expression of cirA in O121:H19 than other strains of Big-Six group, which suggests that the further functional studies are required to characterize its role in O121:H19, and this can be a potential diagnostic marker for bacterial infection. We also observed higher expression of an uncharacterized protein, Q455_0216665, in O145 and O45. This protein contains a domain named “extended signal peptide of type V secretions system” ESPR, and have reported that this domain only exists in Gram-negative bacteria and has originated from the beta and gamma proteobacteria. In this study, we identified 26 different proteins that are categorized as extracellular. We observed a diverse expression of this category of proteins (Fig. 5). As shown in the heat map, we identified two unique clusters that we named as cluster A and cluster B (Table 2). Proteins pertaining to cluster A has a higher expression in E. coli O103:H11, whereas in cluster B we have observed higher expression in E. coli O121:H19. Cluster A includes autotransporter, Tir, peptidase M66, eaeB, and tail protein. Cluster B includes flagellar component of cell-distal portion of basal-body rod, flagellar hook protein (FlgE), flagellar hook length control protein (FliK), flagellar capping protein, flagellin, flagellar hook associated protein (FlgK), flagellar filament structural protein, flagellar hook associated protein (FlgL), and flagellar basal body rod modification protein flgD. In prokaryotes, the flagellar transport is a unique export system localized extracellularly and is ATP-driven [52]. Interestingly, an enrichment of flagellar proteins in this cluster and higher expression in E. coli O121:H19 has been observed, which again suggests that these extracellular proteins can be further studied for better understanding of flagellar protein transport during the pathogenesis.

Figure 4.

Hierarchical clustering of E. coli Big-Six group strains outer membrane proteins: The relative protein expression of outer membrane proteins was shown in a heat map. The fold-change value with K12 is log₂-transformed.

Figure 5.

Hierarchical clustering of E. coli Big-Six group strains extracellular proteins: The relative protein expression of extracellular proteins was shown in a heat map. Two unique clusters were shown in clusters A and B in O103:H11 and O121:H19 strains, respectively.

Table 2.

The list of extracellular proteins identified in E. coli Big-Six group strains

Accession	Protein description	Fold change
		O111:H8/ K12	O145/ K12	O26:H11/ K12	O45:H2/ K12	O103:H11/ K12	O121:H19/ K12
Extracellular proteins: cluster A
564848160	Autotransporter	5.5	3.4	3.4	10.7	41.1	36.6
564832648	Translocated intimin receptor (Tir) secretion protein EspD	5.7	2.8	1.7	4.4	56.7	3.3
564849256	Peptidase M66	3.8	2.4	1.4	2.1	59.6	3.4
564832655	eaeB	5.6	3.9	1.6	2.7	41.0	6.5
564836611	Tail protein	5.4	2.1	5.9	4.0	41.7	3.5
Extracellular proteins: cluster B
16129041	Flagellar basal body rod protein FlgG	0.8	2.1	1.1	0.7	0.9	67.5
566113788	Flagellar hook protein (FlgE)	4.1	5.9	2.8	6.2	1.2	58.9
564846303	Flagellar hook length control protein (FliK)	2.6	2.0	1.1	1.9	2.3	48.2
566115665	Flagellar capping protein	5.2	3.0	1.6	3.6	1.0	73.7
566115664	Flagellin	3.4	2.6	1.6	3.6	1.4	78.5
566113794	Flagellar hook associated protein (FlgK)	3.4	3.1	1.9	4.7	1.1	83.5
16129870	Flagellar filament structural protein	2.8	2.2	1.6	3.1	1.3	55.8
566113795	Flagellar hook associated protein (FlgL)	2.6	2.2	0.9	2.0	1.2	72.2
564838562	Flagellar basal body rod modification protein	3.2	2.4	1.2	2.5	1.2	80.0

3.6 Hypothetical proteins

Hypothetical proteins are significant fraction of protein class found in many MS-based proteomic studies whose existence is predicted based on the computational algorithms [53, 54]. MS-based protein identification of these uncharacterized proteins provides a direct translational evidence, which can be further studied to ascertain their functional role. We identified 94 hypothetical proteins that are predicted to be secreted in our study, of which 48 proteins contained signal peptide. Majority of these proteins showed a diverse expression among the Big-Six group (Supporting Information Table 1A and B and Supporting Information Fig. 1C). We observed a higher expression of a hypothetical protein ydel among the Big-Six group which is a RPA_2b–aaRSs_OBF_like superfamily domain containing protein that is predicted to localize in periplasmic region of the Gram-negative bacteria. This protein has been reported to confer resistance to antimicrobial peptide in association with other porin OmpD protein in salmonella [55]. We identified Sel1A repeat domain containing protein in five of six strains, which is overexpressed with 7.0- and 5.2-fold in O103:H2 and O121:H19 strains, respectively. We could not detect this protein in O45:H2 strain. Sel1A repeat domain containing proteins are known to involve in cellular signaling in both prokaryotes and eukaryotes. The expression of SLR repeat domain containing proteins was reported in Gram-negative Helicobacter pylori [56], Pseudomonas aeruginosa, Rhizobium leguminosarum, and in Vibrio parahaemolyticus. These proteins reported to be expressed during the stress conditions and possess immunomodulatory functions, thus might involve in establishing host–pathogen interactions during the infection [57]. We identified another uncharacterized protein named pliG, periplasmic inhibitor of g-type lysozyme, which is a PPC (prepeptidase C) domain containing protein and whose expression is observed to be high in O103:H11, 22-fold increase compared to the other Big-Six group of E. coli. This observation suggests that the O103:H11 might be sensitive to pliG protein, as it has been previously reported that deletion of pliG gene resulted increase in sensitivity to g-type lysozyme [58]. We also identified CreA family protein whose function is not yet characterized. In our study, this protein is found to be overexpressed in O111, O145, and O26 strains. This protein has been identified to be greatly overexpressed with more than 40-fold across the Big-Six group and suggest that further studies are necessary to characterize and understand their role in pathogenicity.

3.7 Identification of strain-specific proteins

Identification of genes that are specific to a particular strain is important, as they could be served as potential candidate genes in pathogenic E. coli strain identification and differentiation among the Big-Six group. We translated the genomes of Big-Six group and K12 E. coli strains to verify if the identified strain-specific peptides indeed exist in the genome of a given strain. Using custom-made python scripts, we mapped all the identified peptides against the translated genome sequence of each of the strain. This strategy also validates the MS-based peptide identification and the presence of a reporter ion tags in a given strain but not in others, thus categorizing them as strain-specific proteins. From this rigorous analysis, we identified several strain-specific proteins across the studied E. coli strains. Upon manual verification of MS/MS spectra, we considered 24 strain-specific proteins in the Big-Six group (Fig. 6A). The validation of all the identified peptides/proteins by genome comparison is listed in Supporting Information Table 3. MS/MS spectra of identified strain-specific proteins have been provided in Supporting Information Fig. 2. Among the identified strain-specific proteins, we observed 11 secreted proteins alone identified in E. coli O111 and O121 strains, which include eaeB, ACfD, conjugal transfer proteins TraA, TraF, and TraK, and a hypothetical protein whose function is not determined in E. coli O111. Similarly, catecholate siderophore receptor Cir, ectoin, membrane proteins, and hypothetical proteins were identified in E. coli O121. The MS/MS spectrum of an identified peptide for membrane protein is shown in Fig. 6B. We also identified conjugal transfer proteins that are F-plasmid proteins; they are categorized under T4SS and are known for DNA transfer [59, 60]. These strain-specific secreted proteins could potentially serve as potential biomarkers for strain-level differentiation and for diagnostic applications of non-O157 E. coli strains that are encountered recently in several food outbreaks [61, 62].

Figure 6.

Identification of strain-specific proteins in E. coli Big-Six group strains. (A) The strain-specific proteins in E. coli Big-Six group strains are shown. The genome of each strain was translated and mapped to each of the identified peptide. The presence of each strain-specific protein is shown as red. (B) Representative MS/MS spectrum of an identified peptide, DEAWVILEGHIVK of membrane protein (yhha) in E. coli O121:H19.

4 Concluding remarks

We demonstrated the differential expression profile of secreted proteins in Big-Six STEC strains using quantitative proteomics approach to gain insight on the distribution of virulent factors associated with the pathogenicity of the studied E. coli strains. We have shown the diverse protein expressions across the Big-Six STEC strains in the T3SS. Importantly, two of the studied strains, that is, E. coli O45:H2 and O121:H19, were found to be expressing higher titers of secreted proteins that are known in causing the diarrhea in humans. We identified several proteins known to be associated with the Big-Six STEC strains infection including OM, periplasmic and several nonclassical secreted proteins. Using proteogenomics approach, we demonstrated the utility of MS-derived data for the identification of strain-specific secreted proteins. Our data can serve as a proof of principle for identifying strain-specific secreted proteins that can be further studied in developing potential diagnostic markers for the Big-Six STEC strains.

Until recently, much of the focus was on E. coli O157:H7 serogroup but not on Big-Six STEC strains. Our comparative proteomic analysis of Big-Six group would enhance the knowledge gap that eventually aid in enhancement of strain-level differentiation and accurate identification for their presence in outbreaks. The future goal is to investigate those potential markers in terms of developing diagnostic tools for biosurveillance, biothreats reduction, and for food safety and public health monitoring against such pathogenic strains involved in increasing number of outbreaks.

Significance of the study.

Secretome analysis of the Big-Six group of pathogenic E. coli strains have been investigated using multiplexed quantitative proteomic analysis. The Big-Six group of pathogenic E. coli has been more frequently implicated in food outbreak and requires thorough examination and characterization to develop efficient medical countermeasures. This proteogenomic study identified and quantified 565 secreted proteins in the Big-Six group. Several secreted proteins pertaining to T3SS pathway have been identified and their quantitative expression has been shown in Big-Six group. In addition, this study also identified several strain-specific proteins that could be used for developing targeted quantitative assays for better understanding their pathogenicity, food safety, accurate diagnostics of E. coli outbreaks, and biodefense.

Abbreviations

ABC

ATP-binding cassette

bRPLC

basicRPLC

OM

outer membrane

OMPs

OM proteins

STEC

Shiga toxin producing Escherichia coli

Tat

twin-arginine translocation

Tir

translocated intimin receptor

TMT

tandem mass tag

T3SS

type III secretion system