Use of miniaturized protein arrays for Escherichia coli O serotyping

Abstract

Serological typing of Escherichia coli O antigens is a well-established method used for differentiation and identification of O serotypes commonly associated with disease. In this feasibility study, we have developed a novel somatic antibody-based miniaturized microarray chip, using 17 antisera, which can be used to detect bound whole-cell E. coli antigen with its corresponding immobilized antibody, to assess the feasibility of this approach. The chip was tested using the related 17 control strains, and the O types found by the microarray chip showed 100% correlation with the O types found by conventional typing. A blind trial was performed in which 100 E. coli isolates that had been O serotyped previously by the conventional assay were tested by the array approach. Overall, the O serotypes of 88% of isolates were correctly identified by the microarray method. For several isolates, ambiguity of O-type designation by microarray arose due to increased sensitivity of this method, allowing signal intensities of cross-reactions to be quantified. Investigation of discrepancies between conventional and microarray O serotyping indicated that some isolates upon storage had become untypeable and, therefore, gave poor signal intensity when tested by the microarray or retested by conventional means. For all 20 serotype O26 and O157 isolates, the apparent discrepancy in O serotyping was analyzed further by a third independent test, which confirmed the microarray results. Therefore, the use of miniaturized protein arrays increases the speed and efficiency of O serotyping in a cost-effective manner, and these preliminary findings suggest the microarray approach may have a higher accuracy than those of traditional O-serotyping methods.

FIG. 1.

Layout of the miniaturized protein array used in this study. The position of each somatic antibody probe is shaded on the array.

FIG. 2.

Reactions for detecting E. coli somatic antigens using miniaturized arrays. Antibodies raised against 17 different somatic antigens were immobilized on the array surface. Addition of boiled cultured E. coli was followed by sequential addition of anti-E. coli core LPS antibody, biotinylated anti-mouse IgG, and horseradish peroxidase (HRP) conjugate. The bound cell sandwich was finally detected by the addition of tetramethylbenzidine substrate derivative (TMB derivative), which resulted in a blue precipitate that was detectable using the ArrayTube reader.

FIG. 3.

Optimization of the primary and secondary antibody concentrations used in the assay. The primary and secondary antibodies were used in a range of concentrations/dilutions, including 20 μg/ml of WN1222-5 and 1:500 dilution of biotin-conjugated secondary anti-mouse IgG antibody (a), 52 ng/ml of WN1222-5 and 1:10,000 dilution of biotin-conjugated secondary anti-mouse IgG antibody (b), and 10 ng/ml WN1222-5 and 1:10,000 dilution of biotin-conjugated secondary anti-mouse IgG antibody (c) to determine the optimal concentration that would result in high signal intensities for spots with minimal nonspecific cross-hybridization, using a control E. coli strain. The control spots at the sides are clearly distinguishable for panels b and c but not for panel a. (d) Result of analyzing the array image from panel b with the IconoClust software. Abbreviation: B, blank.

FIG. 4.

Comparison of conventional serotyping with the array system. The percentages of correlation of results from a blind trial using the AT system to serotype strains that had previously been serotyped using a conventional method are shown. A total of 100 strains were included. The result of the total initial correlation and correlation for each individual serotype is shown. The total number of isolates and number in each serotype are given in parentheses above the bars.