Genetic Diversity within E. coli

The results shown for the 13 strains from the Escherichia coli reference (ECOR) collection that Souza et al. included “as a reference” in their recent allozyme analysis of diverse natural E. coli isolates (8) fail to validate the newly derived phylogeny (Fig. 1 in reference 8), leave mysterious the correspondence between this phylogeny and that of the ECOR collection (1–5, 7), and raise serious concerns regarding the validity of the investigators' allozyme data and the conclusions derived therefrom (8).

In the dendrogram of Souza, et al. (Fig. 1 in reference 8), the eight ECOR group A control strains are variously placed in the new “ancestral cluster” (along with the four group B1 strains) and in clusters G and C (together with the sole ECOR group D control strain), i.e., across the breadth of the tree and in association with members of the two other ECOR groups studied. This conflicts with previous analyses of the ECOR strains, which with few exceptions have placed the group A strains close together, apart from representatives of other ECOR groups, irrespective of cluster analysis method (e.g., principal-component analysis, UPGMA, neighbor joining, or parsimony) or type of data set (e.g., starch or cellulose acetate gel allozyme analysis, comparative DNA sequencing, or PCR fingerprinting) (1–5, 7).

A direct comparison of allozyme data for the 13 ECOR controls as provided by Souza et al. (8), and as obtained from the ECOR database at the Thomas Whittam laboratory web site (http://www.bio.psu.edu/People/Faculty/Whittam/Lab/), for the six loci at which these two data sets coincide is informative (Table 1). Souza et al. report more polymorphisms at five of the six loci than are documented in the ECOR database. These discrepancies are most marked for the eight group A ECOR strains, among which Souza et al. list twice as many allelic variants (not counting nulls) over the six loci as does the ECOR database (Table 1). According to Souza et al., but not the ECOR data base, ECOR strains 5 and 8 exhibit distinct alleles for IDH, MDH, MPI, and/or PGM as compared with the other six group A strains (data not shown), evidence suggesting that different strains may have been tested as ECOR 5 and 8 in the different analyses. Still, Souza et al. report more diversity even among the remaining six group A ECOR strains than is found in the ECOR database (data not shown).

TABLE 1.

Allozyme polymorphisms among 13 ECOR control strains^a according to two different sources

Locus	No. of alleles per locus among ECOR control strains
	All ECOR controls (n = 13)		Group A controls (n = 8)
	Souza et al.b	ECOR databasec	Souza et al.b	ECOR databasec
ADH	3	2 (+ 1 null)	2	1 (+ 1 null)
G6PD (6Gb)	1	4	1	2
IDH	4	2	3	2
MDH	3	2	2	1
MPI	5 (+1 null)	3	4 (+ 1 null)	1
PGM	4	3	4	1
Total	20	16	16	8

^aThe ECOR controls included ECOR strains 1, 5, 7, 8, 10, 11, 12, and 23 (group A), 26, 27, 33, and 45 (group B1), and 41 (group D). 

^bPer reference 8. 

^cPer Thomas Whittam web site (http://www.bio.psu.edu/People/Faculty /Whittam/Lab/). 

That these discrepancies are not simply a matter of cellulose acetate gels (8) versus starch gels (ECOR database) is shown by the data of Pupo et al. (5). Using cellulose acetate gel allozyme analysis, these investigators found the group A ECOR strains to be quite homogeneous across 10 enzyme loci, including the six shown in Table 1, and derived a neighbor joining tree quite similar to that obtained by Herzer et al. using starch gels and 38 loci (3, 5). Thus, the data of Souza et al. (8), at least for the group A ECOR strains, are of uncertain validity. Confidence in the remainder of these investigators' allozyme data set is weakened by the inclusion of an enzyme (ARK, arginine kinase) that is without precedent in the E. coli allozyme literature and the exceptional finding of two loci for ME (malic enzyme) (3, 6). Consequently, the tree (Fig. 1 in reference 8), the genetic diversity calculations (Table 2 in reference 8), and all inferences drawn from these analyses are suspect and provide no support for the revised understandings of phylogenetic relationships within E. coli proposed by the authors (8).