James R. Johnson, Parissa Delavari, and Adam L. Stell are at the Medical Service, VA Medical Center, and Department of Medicine, University of Minnesota, Minneapolis, Minnesota; Guillem Prats is at the Departament de Microbiologia, Hospital de la Santa Creu i Sant Pau, Universitat Autònoma, 08025 Barcelona, Spain; and Ulrike Carlino and Thomas A. Russo are at the Medical Service, VA Medical Center, and Department of Medicine, and Center for Microbial Pathogenesis, State University of New York at Buffalo.  

Links to Other ASM Pages:



• Table of Contents
• ASM News Issues

Integrity of Archival Strain Collections: the ECOR Collection 

James R. Johnson, Parissa Delavari, Adam L. Stell, Guillem Prats, Ulrike Carlino, and Thomas A. Russo

Papers that use all or part of the ECOR Collection

Figure 1
Figure 2
Figure 3

Investigators depend on the integrity of archival strain collections. Yet humans are inherently error-prone, and microorganisms are inherently genetically unstable (Blum et al. Infect. Immun. 62:606-614, 1994). Consequently, maintaining the integrity of archival strain collections presents a daunting challenge. How to optimally maintain archival strain collections, and whether extant collections already have become corrupted, are questions of vital significance to the microbiology community. A case in point is the Escherichia coli Reference (ECOR) collection (Ochman and Selander, J. Bacteriol. 157:690-693, 1984), a representative subset of the total E. coli population that has facilitated numerous phylogenetic studies. We noted 33 discrepancies (Table 1) between our virulence genotype results for the ECOR strains (Johnson et al., J. Infect. Dis. 183:78-88, 2001) and those from previous publications (Boyd and Hartl, J. Bacteriol. 180:1159-1165, 1998; Marklund et al., Mol. Microbiol. 6:2225-2242, 1992). Consequently, we assessed the phylogenetic authenticity of multiple versions of the 11 implicated ECOR strains as provided by three internationally recognized archival repositories (X, Y, and Z) and by laboratories P and Q, which contributed to one of the previous studies. We compared multiple versions of each strain with the phylogenetically closest ECOR strain(s) (Fig. 1) by random amplified DNA polymorphism (RAPD) analysis (Figures 2 and 3) and used O antigens and pulsed-field gel electrophoresis (PFGE) to confirm strain identities (Johnson et al., Infect. Immun. 68:3327-3336, 2000). 

Discrepancies Found 

Figure 4
Figure 5
Figure 6

Of the 11 ECOR strains, 6 (ECOR 23, 32, 39, 43, 57, and 59) exhibited one or more false versions among their various representatives, evidence of strain substitutions involving two repositories of the ECOR collection (Fig. 3 and 4). For two strains, one repository provided mixed cultures, each containing the authentic ECOR strain plus an unrelated strain. For ECOR 43, two repositories provided the same false strain. One repository supplied a false version of ECOR 59 on two occasions. For five of the six unique false strains, we identified a source within the ECOR collection itself (Fig. 4 and 5), evidence of internal substitutions. papA from the false versions of ECOR 23 and 59, respectively, exhibited 99.2% nucleotide identity with papA from the corresponding suspected source strains, i.e., (authentic) ECOR 62 and 52 (Fig. 4 and 5). 

Table 1
Table 2

Overall, strain substitutions explained 22 of the 33 initially noted discrepancies (Tables 1-2). Two additional discrepancies were explained by probable virulence gene deletions, as suggested by differences in virulence genotype between the several authenticated versions of certain ECOR strains, sometimes accompanied by subtle alterations in PFGE profile (Table 2, Fig. 4). Multiple other probable deletions were documented involving virulence genes not previously assayed (Table 2, Fig. 4). Nine of the initially noted discrepancies remained unexplained; all represented unconfirmed positive results for hly or pap from one previous study (Table 2). 

The ECOR Collection May Not Be a Stable Entity 

These findings suggest that strain substitutions, mixed cultures, and/or virulence gene deletions affect many copies of the ECOR collection. Since we investigated only those ECOR strains with known discrepancies for pap, kps, sfa, or hly, our data provide a minimal estimate of the extent of strain substitutions. To our knowledge, there has been no prior explicit analysis of this problem, although serotype discrepancies have suggested its existence (Johnson et al., J. Infect. Dis. 183:78-88, 2001). Our data suggest that “the” ECOR collection per se may no longer exist. Instead, there may be many different versions of the collection, all differing from one another (and from the original) to varying degrees, much as clonal progeny differ from a parent due to both point mutations (here, isolated virulence gene deletions) and recombination (here, strain substitutions). It would appear that as microbiologists are using the ECOR collection to study the evolution of E. coli they also are experiencing, and even contributing to, the evolution of the collection itself.

Strain substitutions are problematical if they lead to false conclusions regarding the phylogenetic distribution of specific traits or confound the validation of alternative phylotyping methods. The possibility of strain substitutions thus introduces uncertainty regarding the interpretation of previous studies involving the ECOR collection and begs the question of how (and which version of) the collection can be validly used in future studies. It may be that a phylogenetic reanalysis of a “master copy” of the entire ECOR collection is needed to reestablish a reliable comparison standard for future use. In the meantime, and even following such a reanalysis, investigators should remain alert to the possibility of preexisting or new strain substitutions. They should be wary of drawing spurious conclusions from analysis of false strains, and should strive not to introduce additional substitutions. In addition, reports of studies involving ECOR strains should identify the source of the strains used. 

These considerations may be broadly applicable to other strain collections as well. We recently discovered that 8 (11%) of 75 E. coli blood isolates from patients with urosepsis exhibited internally inconsistent typing results, evidence that certain strains had undergone substitutions during their multiple evaluations or transfers between laboratories (Johnson and Stell, J. Infect. Dis. 181:2122, 2000).

Virulence gene deletions, which we documented for 5 of the 11 authentic ECOR strains and for one of the false strains (Fig. 4), are irrelevant for studies of other traits. However, it is unlikely that deletions have been limited to virulence genes. Multiple unrecognized deletions of both virulence- and non-virulence-associated traits probably have occurred in different versions of the ECOR strains. 

The conditions under which ECOR strains have been stored in the various archival repositories may be an important determinant of the likelihood of new deletions. Room-temperature storage of archetypal extraintestinal pathogenic strain CP9 (Russo et al. Infect. Immun. 68:2854-62, 2000) is associated with the appearance of diverse spontaneous mutations in the kps region, some of which result in alterations in the strain's PFGE profile (unpublished data). This is analogous to the subtle PFGE profiles differences between different versions of certain ECOR strains that also exhibited minor virulence genotype alterations (Fig. 4). 

Conclusion 

The multiple strain substitutions and probable spontaneous deletions of virulence genes we documented for certain ECOR strains suggest a need for a comprehensive phylogenetic reevaluation of the ECOR collection and urge caution in interpreting the results of previous studies. More generally, they illustrate the genetic instability of stored bacterial isolates and the fallibility of current systems for maintaining pure and authentic archival cultures. 

All workers in the field depend on and are grateful for the generosity and good faith of those maintaining reference collections, and it is not unexpected that over the years collections in individual laboratories may be compromised. For professional reference collections with their not negligible fees, concerns are heightened. It is not clear to us how the process of maintaining the integrity of such collections can best be done, so that natural biological variation can be distinguished from human error and in vitro mutation. 

ACKNOWLEDGMENTS 

This material is based upon work supported by Office of Research and Development, Medical Research Service, Department of Veterans Affairs VA Merit Review (J.R.J., T.A.R.), National Institutes of Health grants DK-47504 (J.R.J) and AI-42059 (T.A.R.), and grant 95/1379 from the Fondo de Investigaciones Sanitarias de la Seguridad de España (G.P.)