Comparison of pulsed-field gel electrophoresis & repetitive sequence-based PCR methods for molecular epidemiological studies of Escherichia coli clinical isolates

Pulsed-field gel electrophoresis (PFGE), repetitive extragenic palindromic PCR (rep-PCR), and multilocus sequence typing (MLST) are widely used to identify related bacterial isolates and determine epidemiologic associations during outbreaks1234. Among these molecular methods, PFGE is generally considered the gold standard because of its high discriminatory ability, and is, therefore, used routinely to subtype many different bacterial species including Escherichia coli. However, PFGE is a laborious, time-consuming procedure that needs to be performed by a well-trained technician5. Furthermore, if there is degradation of genomic DNA during plug preparation, banding patterns may not be observed. Brolund et al5 reported that PFGE band patterns were interpretable in only 88 per cent of E. coli isolates. Another major limitation of PFGE is its low inter-laboratory reproducibility.

Recently, rep-PCR using the DiversiLab system (bioMerieux, Grenoble, France) has been used to characterize the genomic relatedness of bacterial isolates56. This technique requires less time to perform than PFGE, is technically simple, and has been shown to have similar discriminative power to PFGE7.

In this study, the ability of rep-PCR and PFGE methods was compared to determine the genetic relationships among E. coli isolates assigned to various sequence types (STs) by two MLST schemes, namely Achtman's scheme (www.mlst.ucc.ie) and Whittam's scheme (www.shigatox.net). The two MLST schemes were further studies to determine whether these yielded compatible results.

Material & Methods

This study was performed in the Research Institute of Bacterial Resistance, Yonsei University College of Medicine, Seoul, Republic of Korea.

Bacterial strains: A total of 41 extended-spectrum β-lactamase- (ESBL-) and/or AmpC β-lactamase-producing E. coli clinical isolates were evaluated in this study. Of these isolates, 34 harboured ESBL, four harboured AmpC β-lactamase, and three harboured both ESBL and AmpC β-lactamase (Table). The isolates were identified using either a Vitek 2 GN card (bioMérieux, France) or by 16S rRNA sequencing. Detection of genes coding for plasmid-borne ESBLs and AmpC β-lactamases was performed by PCR amplification as described previously89. The PCR products were directly sequenced using an automated DNA sequencer (Applied Biosystems model 3730xl, Weiterstadt, Germany).

MLST by Achtman's MLST scheme: The E. coli isolates were typed by amplifying partial fragments of seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, and recA) following the protocols available at the website: http://mlst.ucc.ie/mlst/dbs/Ecoli.

MLST by Whittam's MLST scheme: The E. coli isolates were also typed by amplifying partial fragments of seven housekeeping genes (aspC, clpX, fadD, icdA, lysP, mdh, and uidA) following the protocols available at the website: http://www.shigatox.net/ecmlst.

PFGE: Plugs containing whole genomic DNA of the E. coli isolates were digested with the XbaI restriction enzyme. DNA fragments were separated by PFGE using a CHEF-DRII device (Bio-Rad, USA). The conditions used for XbaI-macrorestriction analysis by PFGE were 6 V/cm for 20 h with pulse times ranging from 0.5 to 60 sec at 14°C. A lambda ladder (Bio-Rad) was used as a DNA size marker. Gel images in TIFF format were exported to Molecular Analyst Fingerprinting Software Ver. 3.2 (Bio-Rad) for analysis. E. coli isolates were compared using the band-based dice coefficient. Dendrograms were generated by the unweighted pair group method with arithmetic averages and a 1.0 per cent position tolerance, and DNA relatedness was calculated based on the criteria described by Tenover et al10. Isolates showing >80 per cent similarity were considered to be related.

Rep-PCR: Rep-PCR experiments were performed using the DiversiLab system (bioMérieux, Grenoble, France) according to the manufacturer's instructions. Results were analyzed with DiversiLab software using the Kullback-Leibler method to determine distance matrices and with the unweighted pair group method with arithmetic averages to create a dendrogram. Isolates showing >95 per cent similarity were considered to be related7.

Results

Achtman's scheme vs. Whittam's scheme: Of the 22 E. coli isolates identified as ST131 using Achtman's MLST scheme, 16 were identified as ST34, one each as the single locus variants ST872 and ST876, and four as the double-locus variants ST856 (n = 3) and ST873 (n = 1) using Whittam's MLST scheme. The four E. coli isolates identified as ST171 by Whittam's MLST scheme were identified as ST10 (n = 1), its single-locus variants ST44 (n = 1) and ST744 (n = 1), and its double-locus variant ST617 (n = 1) by Achtman's MLST scheme. In addition, two E. coli isolates identified as ST857 using Whittam's MLST scheme were identified as ST354 and its single-locus variant ST2037 by Achtman's MLST scheme (Table). These results indicated that the two schemes yielded compatible results.

Table Sequence types of E. coli clinical isolates according to Achtman's and Whittam's MLST schemes

MLST vs. rep-PCR: The 22 E. coli isolates identified as ST131 using Achtman's MLST scheme were >95 per cent similar by rep-PCR (Fig. 1A). However, the five isolates of ST38, ST46, ST95, ST354, and ST648 also showed >95 per cent similarity to the ST131 isolates (I in Fig. 1B). Furthermore, two isolates typed as different ST isolates using Achtman's MLST scheme, ST44 and ST393, were also >95 per cent similar to one another (II in Fig. 1B). The remaining ST isolates showed <95 per cent similarity to each other (Fig. 1B). These results demonstrate that rep-PCR can accurately group E. coli isolates belonging to the same ST together, but this method demonstrated limited ability to discriminate between E. coli isolates belonging to different STs.

Close

Fig. 1

Banding patterns of 22 E. coli isolates identified as ST131 using Achtman's MLST scheme (A) and 20 E. coli isolates identified as non-ST131 according to Achtman's MLST scheme, (B) obtained by rep-PCR. The dotted line in Fig. 1B indicates 95 per cent similarity.

Export to PPT

PFGE vs. MLST: In the XbaI-macrorestriction experiments, of the 22 E. coli isolates identified as ST131 by Achtman's MLST scheme, most of the isolates showed <80 per cent similarity (Fig. 2A), except for a few isolates (I-IV in Fig. 2A) that showed >80 per cent similarity. However, most of the isolates classified as belonging to different STs by MLST showed <80 per cent similarity by PFGE (Fig. 2B), although there were some isolates (I and II in Fig. 2B) belonging to different STs that showed >80 per cent similarity. These results indicate that PFGE can segregate E. coli isolates belonging to different STs into different types, but is not optimal for grouping E. coli isolates of the same ST.

Close

Fig. 2

Banding patterns of 22 E. coli isolates identified as ST131 using Achtman's MLST scheme (A) and 20 E. coli isolates identified as non-ST131 according to Achtman's MLST scheme, (B) obtained by PFGE. The dotted lines in Figures indicate 80 per cent similarity.

Export to PPT

Discussion

Achtman's MLST scheme is based on seven housekeeping genes (adk, fumC, gyrB, icdA, mdh, purA, and recA) and has been used in epidemiological studies of antimicrobial resistance, whereas Whittam's MLST scheme, which is also based on a set of seven housekeeping genes (aspC, clpX, fadD, icdA, lysP, mdh, and uidA), has been used mainly in the field of food microbiology. E. coli isolates harbouring the same antimicrobial resistance genes have been detected in food, the gastrointestinal tract of food-producing and companion animals, and in humans11. In other words, food-producing and companion animals may play a significant role in the dissemination of antibiotic resistant bacteria in humans12. Therefore, comparing the epidemiology of bacteria isolated from humans and food-producing animals or food is crucial to understand the mechanisms and components underlying the dissemination of antimicrobial resistance genes. Unfortunately, using different MLST schemes to subtype E. coli isolates from humans and food-producing animals complicated epidemiological comparisons.

This study showed that that the two schemes yielded compatible results, even though employing different loci. These results could be used in the understanding of reports for E. coli epidemiology. For example, six enterotoxigeneic E. coli isolates identified as ST34 by Whittam's MLST scheme in a study by Steinsland et al13 would be considered ST131 E. coli isolates according to Achtman's MLST scheme. And CTX-M-14 extended-spectrum β-lactamase-producing E. coli isolates from Spain identified as ST10 by Achtman's MLST scheme in a study by Valverde et al14 would be considered ST171 E. coli isolates according to Whittam's MLST scheme.

PFGE showed higher discriminative power than rep-PCR in grouping E. coli isolates belonging to different STs, whereas rep-PCR accurately grouped E. coli isolates belonging to the same ST together, but PFGE did not. This study showed that the appropriate molecular typing method for an epidemiological study should be chosen based on the particular characteristics of that study. PFGE is more suited for investigating outbreaks in a limited space such as a specialty hospital or an intensive care unit due to its high discriminatory abilities, whereas rep-PCR should be used in nationwide or worldwide epidemiology studies because it is more efficient at grouping strains belonging to the same STs together.