Development of Three Multiplex PCR Assays Targeting the 21 Most Clinically Relevant Serogroups Associated with Shiga Toxin-Producing E. coli Infection in Humans

Escherichia coli serogroups O5, O15, O26, O45, O55, O76, O91, O103, O104, O111, O113, O118, O121, O123, O128, O145, O146, O157, O165, O172, and O177 are the O-antigen forms of the most clinically relevant Shiga toxin-producing E. coli (STEC) serotypes. In this study, three multiplex PCR assays able to specifically detect these 21 serogroups were developed and validated. For this purpose, the O-antigen gene clusters of E. coli O5 and O76 were fully sequenced, their associated genes were identified on the basis of homology, and serogroup-specific primers were designed. After preliminary evaluation, these two primer pairs were proven to be highly specific and suitable for the development of PCR assays for O5 and O76 serogroup identification. Specific primers were also designed for serogroups O15, O45, O55, O91, O104, O113, O118, O123, O128, O146, O157, O165, O172, and O177 based on previously published sequences, and previously published specific primers for serogroups O26, O103, O111, O121, and O145 were also included. These 21 primer pairs were shown to be specific for their target serogroup when tested against E. coli type strains representing 169 known O-antigen forms of E. coli and Shigella and therefore suitable for being used in PCR assays for serogroup identification. In order to validate the three multiplex PCR assays, 22 E. coli strains belonging to the 21 covered serogroups and 18 E. coli strains belonging to other serogroups were screened in a double-blind test and their sensitivity was determined as 1 ng chromosomal DNA. The PCR assays developed in this study could be a faster, simpler, and less expensive strategy for serotyping of the most clinically relevant STEC strains in both clinical microbiology and public health laboratories, and so their development could benefit for clinical diagnosis, epidemiological investigations, surveillance, and control of STEC infections.


Introduction
Shiga toxin (Stx)-producing Escherichia coli (STEC) are important food-borne zoonotic pathogens responsible for a broad spectrum of clinical symptoms in humans, ranging from mild diarrhea to hemorrhagic colitis (HC) and the life-threatening hemolytic uremic syndrome (HUS) [1]. Although serotype O157:H7 has been implicated in most outbreaks and in most cases of HUS, there is growing concern about the risk to human health associated with non-O157 STEC serotypes [2,3], which may be as well responsible for important outbreaks, such as the renowned 2011 STEC O104:H4 German outbreak [4].
E. coli serotyping is typically performed by agglutination reactions using antisera raised in rabbits against the different O and H standard reference strains [14]. However, traditional serotyping is both laborious and time consuming and it often generates equivocal results due to cross-reaction between different serogroups, and even no results when testing rough strains, which are refractory to typing. The reference technique requires the previous thermal inactivation, at different temperatures, of the capsule in order to expose the O antigens, and also the use of a wide collection of antisera, which is too costly for most laboratories and can only be generated by specialized laboratories with animal facilities. Thus, rapid, less expensive, and more specific molecular methods for identifying different E. coli serotypes are strongly needed.
Much of the O-antigen variation in E. coli is a consequence of the extensive genetic diversity within the rfb (O-antigen) gene cluster, which encodes many of the enzymes involved in Oantigen biosynthesis and assembly [15]. The rfb region maps flanked by the two housekeeping genes galF and gnd on the E. coli chromosome. Indeed, the JUMPstart sequence, which is a 39bp conserved element located in the intergenic region between galF and the O-antigen gene cluster, and the gnd sequence, which is present downstream the cluster, have been used to successfully amplify the entire O157 O-antigen gene cluster by PCR [16]. The cluster typically includes three different types of genes: (i) genes encoding enzymes involved in the synthesis of the sugars that form the O subunit; (ii) genes encoding transferases, which assemble sugar substituents into the O subunit; and (iii) genes encoding proteins involved in processing and assembly steps to build the O antigen from the O subunit, such as wzx (enconding the O-antigen transporter or flippase) and wzy (enconding the O-antigen polymerase) [17]. Several genes in the O-antigen gene cluster, in particular wzx and wzy, show relatively low similarity among different E. coli serogroups, and therefore primers targeting wzx and wzy are generally used to develop serogroup-specific PCR assays [18][19][20][21].
In this study, the O-antigen gene clusters of E. coli O5 and O76 reference strains were fully sequenced, their associated genes were identified on the basis of homology, and specific primers targeting wzx were designed for each serogroup. Specific primers targeting wzx or wzy were also designed for serogroups O15, O45, O55, O91, O104, O113, O118, O123, O128, O146, O157, O165, O172, and O177 based on previously published sequences from each serogroup. Previously published specific primers targeting wzx of serogroups O26, O103, O111, O121, and O145 were added to the designed primers to further develop three serogroup-specific multiplex PCR assays able to detect these 21 serogroups. These PCR assays were shown to be highly specific and sensitive, and suitable for the detection of the most clinically relevant STEC serogroups.

Bacterial strains
Reference strains for E. coli O5 and E. coli O76 from Statens Serum Institute (SSI, Copenhagen, Denmark) were used for nucleotide sequence analysis of their O-antigen gene clusters. A collection of 10 E. coli O5 and 15 E. coli O76 strains isolated from feces of humans and animals and from food at different time periods in Spain [22][23][24][25] was used for the preliminary evaluation of the O5 and O76 serogroup-specific PCR primers designed (S1 Table). A collection of E. coli type strains representing 169 known O-antigen forms of E. coli and Shigella was used for testing of primer specificity and validation of the serogroup-specific multiplex PCR assays developed (S2 Table).
Amplification, sequencing, and sequence analysis of the E. coli O5 and O76 O-antigen gene clusters Chromosomal DNA from E. coli O5 and E. coli O76 reference strains was prepared with a QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA). Published oligonucleotides 482 (5 0 -CAC TGC CAT ACC GAC GAC GCC GAT CTG TTG CTT GG-3 0 ) and 412 (5 0 -ATT GGT AGC TGT AAG CCA AGG GCG GTA GCG T-3 0 ) complementary to the JUMPstart sequence and to the proximal end of gnd [11], respectively, were used to amplify their entire O-antigen gene clusters in a long PCR assay carried out with the Expand Long Template PCR System (Roche Diagnostics, Mannheim, Germany) in an ABI 2720 thermal cycler (Applied Biosystems, Foster City, CA, USA) as follows: denaturation at 94°C for 10 s, annealing at 64°C for 30 s, and extension at 68°C for 15 min, repeated ten times. For the next 20 cycles, the extension step was increased by 20 s each time. One initial denaturing step (94°C for 2 min) and one final elongation step (72°C for 7 min) were added. Sequencing of long PCR products was achieved by a primer-walking approach in an ABI 3730xl DNA Analyzer (Applied Biosystems). DNA sequence assembly and analysis, and primers design were performed with the Lasergene software 7.0 (DNAstar, Madison, WI, USA). Within the resulting sequences, putative coding regions were identified by using ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/). BLAST and PSI-BLAST were used for searching databases, including GenBank, COG, and Pfam (http:// www.ncbi.nlm.nih.gov/blast/).

Evaluation of the E. coli O5 and O76 serogroup-specific PCR primers
Serogroup-specific primer pairs were designed based on the wzx sequences determined before from E. coli O5 and E. coli O76 reference strains (Table 1). In order to preliminarily evaluate the E. coli O5 and O76 serogroup-specific PCR primers designed, chromosomal DNA from 50 E. coli strains including 10 E. coli O5, 15 E. coli O76 (S1 Table), and 25 E. coli strains belonging to other serogroups were screened in a double-blind test. For DNA extraction, a 1-μl loop of bacterial growth was suspended in 0.5 ml of sterile distilled water, boiled for 5 min, and centrifuged at 10,000 rpm for 5 min. The supernatant was used directly as template DNA in the PCR assays, without previous dilution. Conventional PCR amplifying a single target gene was performed using DreamTaq DNA Polymerase (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's instructions, in an ABI 2720 thermal cycler (Applied Biosystems) as follows: denaturation at 94°C for 30 s, annealing at 58°C for 30 s, and extension at 72°C for 1 min, repeated 25 times. Each reaction contained 400-800 nM of each primer (Table 1) and 5 μl of template DNA in a final volume of 25 μl. Fragments were separated in 2% agarose (MS8 type, Pronadisa, Madrid, Spain) gels by unidirectional electrophoresis using TAE 1x buffer and visualized by staining with ethidium bromide. Fragment size was determined by comparison with 100 bp DNA ladders (Thermo Fisher Scientific).

Specificity of the serogroup-specific PCR primers
Apart from the primers specific for wzx genes of E. coli O5 and O76 described before, specific primers targeting wzx or wzy were also designed for serogroups O15, O45, O55, O91, O104, O113, O118, O123, O128, O146, O157, O165, O172, and O177 based on previously published sequences from each serogroup (Table 1), and previously published specific primers targeting wzx of serogroups O26, O103, O111, O121, and O145 [18] (Table 1) were also considered to further develop multiplex PCR assays. The specificity of the 21 serogroup-specific PCR primer pairs was evaluated by testing them against 169 E. coli type strains representing a broad range of O antigens of E. coli and Shigella. For this purpose, chromosomal DNA was prepared as described before from each of the 169 type strains and used to make DNA pools. A total of 18 pools were made, each containing DNA from 4 to 10 strains (S2 Table), and the 21 serogroupspecific PCR primer pairs were used to screen the DNA pools. Conventional PCR was performed as described before, with each reaction containing 200-800 nM of each primer (Table 1).

Validation and sensitivity of the serogroup-specific multiplex PCR assays
The 21 serogroup-specific primer pairs were combined in three multiplex 5 0 -nuclease PCR assays (multiplex 1 to 3) (Table 1). In order to validate the three multiplex PCR assays, 22 E. coli strains belonging to the 21 covered serogroups and 18 E. coli strains belonging to other serogroups selected from the E. coli type strains collection were screened in a double-blind test. For this purpose, chromosomal DNA from these 40 E. coli strains was prepared as described before and screened with the three multiplex PCR assays. PCR was performed as described before but in a multiplex way, with each multiplex reaction containing 7 primer pairs in total and 200-800 nM of each primer (Table 1). Additionally, the three multiplex PCR assays were used to test 10-fold serially diluted chromosomal DNA prepared from E. coli strains belonging to the 21 covered serogroups selected from the E. coli type strains collection.

Nucleotide Sequence Accession Number
The DNA sequences of the E. coli O5 and O76 O-antigen gene clusters have been deposited in GenBank under the accession numbers KM881565 and KM881564, respectively.

Results and Discussion
Nucleotide sequence analysis of the E. coli O5 and O76 O-antigen gene clusters A sequence of 9,915 bases between the JUMPstart sequence and gnd was obtained from E. coli O5, and eleven open reading frames (ORFs) were found (Fig. 1). All the ORFs were assigned functions and shown to be related to O-antigen biosynthesis on the basis of their similarity to related genes in nucleotide sequence databases (Table 2). A sequence of 7,134 bases was obtained from E. coli O76, and nine ORFs were found (Fig. 1). Likewise, all the ORFs were assigned functions and shown to be related to O-antigen biosynthesis on the basis of their similarity to related genes in nucleotide sequence databases ( Table 3). As expected for O-antigen gene clusters, the sequences obtained had a significantly lower G+C content than those in the E. coli genome (Tables 2 and 3) [26]. The E. coli O5 O-antigen gene cluster was shown to be highly related to the Salmonella Pomona O28 O-antigen gene cluster (accession number EU805803), with the same organization and 78% DNA identity. E. coli and Salmonella are closely related, and several cases in which the O-antigen structures are identical or highly similar in the two species have been documented [27,28]. The sequence similarity level between Salmonella and E. coli O-antigen gene clusters that express identical O-antigen backbones is close to the lower end of the range for their housekeeping genes (between 76% and 100% DNA identity), indicating that O-antigen gene clusters for each structure originate from a common ancestor [28].
Evaluation of the E. coli O5 and O76 serogroup-specific PCR primers Primer pairs specific for wzx genes of E. coli O5 and O76 were designed (Table 1) and evaluated in a double-blind test with 50 E. coli strains including 10 E. coli O5, 15 E. coli O76 (S1 Table), and 25 E. coli strains belonging to other serogroups. All the E. coli O5 and O76 strains gave the expected PCR products corresponding to primer pairs used (Table 1), and no PCR products were obtained from strains belonging to other serogroups. Thus, the primers were proven to be highly specific and suitable for the development of PCR assays for O5 and O76 serogroup identification.

Identification of serogroup-specific genes by PCR
Primer pairs specific for wzx or wzy genes of E. coli O5, O15, O26, O45, O55, O76, O91, O103, O104, O111, O113, O118, O121, O123, O128, O145, O146, O157, O165, O172, and O177 (Table 1) were used to screen the 18 DNA pools containing representatives of 169 known Oantigen forms of E. coli and Shigella (S2 Table) in order to test their specificity. The pools containing strains of any of the 21 covered serogroups gave PCR products of the expected size (Table 1), and no PCR products were obtained from the remaining pools, with the only exceptions of pools 1 and 15. Pool 1 gave a positive PCR result when tested with the primer pair specific for wzx gene of E. coli O104 (272 bp). Such a result was not surprising, since pool 1 contained an E. coli O9 strain and it is well known that the gene cluster encoding for serogroup O104 has the same genes in the same order as the K9 gene cluster [29], and the K9 antigen is generally present in E. coli strains belonging to serogroups O8, O9, and O9a [30], which was exactly the case of the E. coli O9 strain included in pool 1. Likewise, pool 15 gave a positive result when tested with the primer pair specific for wzx gene of E. coli O118 (409 bp). This result was also expected, since pool 15 contained an E. coli O151 strain and the O-antigen gene clusters of both E. coli O118 and E. coli O151 have been shown to be organized in the same manner and to share high level identity (> 99% DNA identity) [28]. Indeed, the sequences of wzx genes of E. coli O118 and E. coli O151 are identical [28]. Thus, apart from these limitations, the 21 primer pairs were proven to be specific for their target serogroup when tested against E. coli type strains representing 169 different serogroups and therefore suitable for being used in PCR assays for serogroup identification.

Development of serogroup-specific multiplex PCR assays
In order to develop a less laborious PCR method, the 21 serogroup-specific primer pairs were combined in three multiplex 5 0 -nuclease PCR assays (multiplex 1 to 3) (Table 1) aiming to detect the most clinically relevant STEC serogroups. For this purpose, the primer pair efficiency for the 21 serogroups was determined on the basis of the amplicons of expected sizes by testing different primer concentrations. The primer concentration resulting in high-signal products was used as described in the Methods section. At the optimized primer concentration ratio, the DNA of 22 strains belonging to the 21 covered serogroups produced the expected PCR products (Fig. 2) in the double-blind test for validation of the three multiplex PCR assays. DNA from 17 strains belonging to other serogroups did not produce any other PCR products and one E. coli O9 strain gave a positive PCR result (272 bp) when tested with multiplex 3, due to the presence of the primer pair specific for wzx gene of E. coli O104 on this multiplex PCR assay, as discussed before. To test the sensitivity of the three multiplex PCR assays, they were carried out to amplify serially diluted chromosomal DNA (100 ng, 10 ng, 1 ng, 100 pg, 10 pg, and 1 pg) prepared from strains belonging to the 21 covered serogroups, and positive PCR results were obtained from as little as 1 ng of DNA for each of the strains.
In conclusion, the three serogroup-specific multiplex PCR assays developed in this study were found to be highly specific and sensitive, and suitable for serogroup identification in E. coli. The combination of these three multiplex PCR assays enables the reliable detection of genes encoding the O antigen in E. coli strains belonging to the most clinically relevant STEC serotypes, including typical, atypical, and emerging EHEC serotypes. This method of molecular serotyping is a faster, simpler, and less expensive technique than traditional serotyping, also enabling the detection of E. coli O antigens even when they cannot be expressed by the bacteria. As a consequence, these PCR assays could be an efficient and convenient strategy for serotyping of the most clinically relevant STEC strains in both clinical microbiology and public health laboratories, especially in those where PCR is already a routine tool, and so their development could benefit clinical diagnosis, epidemiological investigation, surveillance, and control of STEC infections.
Supporting Information S1