Towards understanding inter-strain attachment variations of Escherichia coli during transport in saturated quartz sand - Transport of Multiple Escherichia Coli Transport in Saturated Porous Media - page 39

essential for the attachment of E. coli to the quartz grains. We considered the flu gene to be
present, if both primer sets, amplifying different parts of the gene, yielded correctly sized PCR
product. So, of the 20 strains considered for PCR, eight strains (UCFL-94, UCFL-131, 1923,
1991, 2621, 1625, 2059, and 2041) were positive for the flu gene. Besides the lack of relation
between presence/absence of the flu gene and attachment efficiency, there was also no relation (
0.67 p = ; Kruskal-Wallis test) between presence/absence of the gene and cell aggregation.

S fimbriae and F1C fimbriae ( sfa/focDE gene cluster, focG , and sfaS genes), the porcine
associated gene ( paa ), the non-fimbrial adhesin ( nfaE ), and bmaE were present in (less than)
10% of the strains tested, while genes aah , AIDA-I, AIDA-C, fanC , faeG , fedA , and Saa were
completely absent. We concluded that those genes amplified only once or twice with PCR
analyses could not have played a significant role in initial attachment to the quartz sediment.

3.4 Discussion

Our work indicated that in a column of 7 cm in height, under identical flow conditions, there was
a two log variation in maximum breakthrough concentrations of the 54 E. coli strains we used.
Attachment efficiencies varied between 0.3 and 1, which was rather high, and in a number of
cases, attachment efficiencies were above 1. When values were between 1 and 1.25, we did not
consider those to be problematic. Values above unity might occur due to the fact that the TE
correlation equation (Tufenkji and Elimelech, 2004a), which we used to determine the single
collector contact efficiency, applies to spherical collectors. Our sand grains were somewhat
rounded, not spherical. In such case, the single collector contact efficiency might be under
estimated, giving rise to attachment efficiencies in excess of unity. Others (Shellenberger and
Logan, 2002; Morrow et al., 2005; Paramonova et al., 2006; Yang et al., 2006; Lutterodt et al.,
2009a) have also reported on attachment efficiencies above 1. For the remaining three strains
(2041, 1514, and 2049), it was difficult to find an explanation for their very high attachment
efficiencies of 1.42, 1.61 and 2.10, other than to assume that the bacteria size that was used in
calculating the single collector contact efficiency had been wrong, most likely due to the
presence of surface structures sticking out of the surface of these strains, thereby actually acting
as a bigger colloidal particle than used in the TE correlation equation. As we indicated in the
Methods Section, we wanted to create an environment inside the column which would enhance
the attachment of cells. We succeeded in this (attachment efficiencies ranged between 0.3 and 1),
because of the presence of divalent Ca and Mg in the solution: Foppen and Schijven (2006)
reported similar attachment efficiencies for E. coli in suspensions of low ionic strength
containing Ca and Mg. The two log variation in maximum breakthrough concentrations
reinforced the idea that, although E. coli is an indicator organism, there is not one single
attachment efficiency for all E. coli strains. There is variation, even at a very small scale. These
variations are often neglected, especially when studying few E. coli strains (e.g. Foppen et al.,
2007a,b; Tufenkji, 2006; Bradford et al., 2006; Walker et al., 2004; Powelson and Mills, 2001;
Levy et al., 2007; Jiang et al., 2007).