Comparison of Escherichia Coli and Campylobacter Jejuni Transport in Saturated Porous Media

Published online May 31, 2006

Comparison of Escherichia coli and Campylobacter jejuni Transport in Saturated Porous Media

Carl H. Bolster,* Sharon L. Walker, and Kimberly L. Cook

ABSTRACT Although the presence of pathogenic microorganisms Due to the difficulties in testing for specific pathogens, water sam- in drinking water supplies is a threat to public health, ples are tested for the presence of nonpathogenic indicator organisms rarely are concentrations of these particular organisms to determine whether a water supply has been contaminated by fecal measured. Rather, water samples are tested for the pres- material. An implicit assumption in this approach is that where patho- ence of nonpathogenic microorganisms commonly genic microorganisms are present fecal indicator organisms are pres- found in fecal material, referred to as indicator organ- ent as well; yet surprisingly few studies have been conducted that isms. The presence of these indicator organisms does not directly compare the transport of indicator organisms with pathogenic mean necessarily that the water is unsafe for consump- organisms in ground water environments. In this study we compared tion; rather it indicates that the water has likely been the cell properties and transport of Escherichia coli, a commonly used contaminated by fecal material and thus potentially indicator organism, and Campylobacter jejuni, an important entero- pathogen commonly found in agricultural wastes, through saturated contains pathogens (Leclerc and Mosel, 2001). Com- porous media. Differences in cell properties were determined by mea- monly used indicator organisms include total coliforms, suring cell geometry, hydrophobicity, and electrophoretic mobility. fecal coliforms, E. coli, and enterococci. To be effective, Transport differences were determined by conducting miscible dis- an indicator organism must occur in a sample when fecal placement experiments in laboratory columns. Under the experimen- pathogens are present, but in far greater numbers than tal conditions tested, C. jejuni was much more negatively charged the pathogen(s) of interest. Additionally, the indicator and more hydrophobic than E. coli. In addition, C. jejuni cells were should respond to natural environmental conditions and slightly longer, narrower, and less spherical than E. coli. The variations water treatment processes in a manner similar to the in cell properties, primarily surface charge, resulted in significant pathogen(s) of concern. In ground water environments differences in transport between these two microorganisms, with the transport of C. jejuni exceeding that of E. coli when conditions fa- this requires that the indicator organism have similar or vored low attachment rates, thus calling into question the usefulness of greater transport and survival characteristics than the using E. coli as an indicator organism for this important pathogen. pathogenic microorganisms of concern to ensure detect- able quantities of the indicator organism in fecally contaminated waters. NFILTRATION of fecal material into the subsurface can Given the diversity of potentially pathogenic micro- Iresult in contamination of soils and ground water organisms found in animal manures, it is reasonable to by enteropathogenic microorganisms such as bacteria, expect diversity in the cell surface properties of these viruses, and protozoa. Recent studies have shown that microorganisms. For instance, Lytle et al. (2002) have fecal contamination has been detected in nearly half of shown that several important microorganisms to public all the drinking water wells tested in the United States health such as E. coli O157:H7, Cryptosporidium parvum (Macler and Merkle, 2002). Some of the most cited oocysts, and Vibrio cholorae all have significantly dif- sources of fecal contamination of ground water supplies ferent charges (zeta potential) on their outer surface. are agricultural activities (USEPA, 1994) including use Given that the movement of bacteria through the sub- of manure for crop fertilization (Gagliardi and Karns, surface is governed, in part, by cell properties such as 2002), pasturing of livestock, animal feeding operations surface charge (Busscher et al., 1984; van Loosdrecht (Gerba and Smith, 2005), wash water from animal hous- et al., 1987a), cell geometry (Fontes et al., 1991; Gannon ing (Mawdsley et al., 1995), and leaky manure storage et al., 1991; Weiss et al., 1995), and hydrophobicity lagoons (USEPA, 1993). Several recent studies have (Stenstrom, 1989; van Loosdrecht et al., 1987a, 1990), found a strong link between human illness and land ap- it is likely that there exists a range of transport char- plication of agricultural waste (Clark et al., 2003; Stanley acteristics for the variety of microorganisms commonly and Jones, 2003; Valcour et al., 2002). found in fecal material. Therefore, it seems unlikely that a single organism, for example E. coli, can adequately serve as a surrogate for the subsurface transport of all C.H. Bolster and K.L. Cook, USDA-ARS, 230 Bennett Lane, Bowling Green, KY 42104. S.L. Walker, Department of Chemical and En- pathogenic microorganisms. Indeed, the effectiveness vironmental Engineering, University of California, Riverside, B355 of E. coli as an indicator of enteric virus transport in Bourns Hall, Riverside, CA 92521. Mention of trade names or com- ground water has already been questioned (Borchardt mercial products is solely for the description of experimental proce- et al., 2003; Meschke and Sobsey, 2003). dures and does not imply recommendation or endorsement by the U.S.

Reproduced from Journal of Environmental Quality. PublishedDepartment by ASA, CSSA, and SSSA. All copyrights reserved. of Agriculture. Received 4 June 2005. *Corresponding The objective of our study was to compare key cell author ([email protected]). properties and transport characteristics of C. jejuni,an important pathogen commonly found in agricultural Published in J. Environ. Qual. 35:1018–1025 (2006). wastes (Wesley et al., 2000) and a leading cause of gas- Technical Reports: Ground Water Quality doi:10.2134/jeq2005.0224 troenteritis in the United States (Allos, 2001), and E. coli, ª ASA, CSSA, SSSA a commonly used indicator of fecal contamination. Al- 677 S. Segoe Rd., Madison, WI 53711 USA though Campylobacter spp. has been detected in ground 1018 BOLSTER ET AL.: COMPARISON OF E. COLI AND C. JEJUNI TRANSPORT IN POROUS MEDIA 1019

water samples (Stanley et al., 1998), we are unaware of trophotometer (BioSpec-mini; Shimadzu, Kyoto, Japan). Hy- drophobicity was determined by: any studies that have looked specifically at the transport of C. jejuni in porous media and compared its transport C0 2 C45 to that of an indicator organism. The results of our re- hydrophobicity 5 3 100% [1] C search will provide much needed information for deter- 0 21 mining whether E. coli is a useful indicator for assessing where C0 is the initial concentration (cells mL ) and C45 is the potential ground water contamination by C. jejuni. concentration in the aqueous phase following mixing and a 45-min separation period (cells mL21). Electrophoretic mobility of the bacterial cells was deter- MATERIALS AND METHODS mined using a ZetaPALS analyzer (Brookhaven Instruments, Holtsville, NY). The bacteria were prepared by the same Bacteria Preparation methods as used for transport experiments and diluted to a 5 6 21 Cultures of E. coli O157:H7 strain L-2 were maintained at final concentration between 10 and 10 cells mL in a 10 mM 2808C in Luria broth (LB) with 50 mg ampicillin mL21 and KCl electrolyte solution. Electrophoretic mobility measure- 15% glycerol. The L-2 strain is derived from E. coli O157:H7 ments were conducted at 258C and were repeated five times strain B6-914 (ATCC 43888), which does not produce Shiga- for each microorganism. like toxins I or II and does not possess the genes for these Average cell diameter, cell width, and cell shape were mea- toxins (Siragusa et al., 1999). One week before the experi- sured using Image Pro Plus software (Media Cybernetics, 2004). ments the culture was resuscitated by two successive passages Cell shape, defined as the ratio between cell width and length, on LB agar with 50 mg ampicillin mL21 and incubated for 24 h was used as a way to assess each cell’s departure from spheric- at 378C. One isolated colony from a fresh LB agar plate was ity. The closer the shape value is to 1 the more spherical it is. inoculated into 50 mL of LB broth and the culture was incu- Three hundred cells were measured for each microorganism. bated at 378C for 24 h to achieve stationary phase. Cultures of C. jejuni (ATCC 49943) were maintained on Campylobacter Selective Agar (CSA) prepared as recommended by the manu- Sand Preparation facturer and transferred weekly. Briefly, campylobacter agar Quartz sand (Unimin, New Caanan, CT) was sifted through base (Oxoid, Basingstoke, England) was autoclaved for 15 min 350- and 250-mm sieves (U.S.A. Standard Testing Sieves; and cooled, and modified Preston campylobacter selective sup- ATM, New Berlin, WI). The fraction of sand between 250 and plement (Oxoid) and 5% lysed horse blood (Hemostat Labs, 350 mm in size was collected and utilized for the experiments. Dixon, CA) were added. Inoculated plates were incubated at To remove metal oxide coatings and organic matter associated 378C for 48 h under microaerophilic conditions using anaero- with the quartz, the sand was boiled in a 2-L flask containing bic gas jars equipped with the CampyPak Plus microaerophilic 1 M hydrochloric acid (Fisher Scientific) for 2 h and then system (BBL, Cockeysville, MD). Before the experiments, one rinsed with deionized water until the rinse water pH was equal C. jejuni colony from a CSA plate was used to inoculate 500 mL to the pH of the deionized water. The sand was then dried of tryptic soy agar (BD, Sparks, MD). To achieve stationary overnight at 1058C, re-rinsed in deionized water the following phase C. jejuni cultures were incubated at 378C for 48 h in day, and dried again overnight. anaerobic gas jars under microaerophilic conditions. Bacterial Portions of the acid-washed sand were coated with metal cells were pelleted and washed three times in KCl solution by oxides by the method of Bolster et al. (2001). Sand was placed centrifugation at 9000 3 g for 10 min and resuspended in the in a concentrated solution containing either FeCl3 or AlCl3 appropriate KCl solution for approximately 18 h to ensure that (Fisher Scientific). The solution pH was adjusted so that it the bacteria were in a resting state before conducting column exceeded 9 by the addition of 10 M NaOH (Fisher Scientific). experiments and characterization of surface properties. Sand was equilibrated in the solution for 36 to 48 h. One molar Cell culturability was evaluated following 18-h incubation NaOH was added to the solution when the pH dropped below in KCl solution. Culturability of E. coli was determined by plat- 9. Following equilibration of the sand with the solution, the ing onto LB agar with no antibiotic. Plates were incubated at sand was rinsed in deionized water until the water was clear 378C for 24 h. Final culturable cell numbers for E. coli were and baked again at 1058C for 24 h. The final step of rinsing and 1.0 3 108 cells mL21. Culturability of C. jejuni was determined drying was repeated. Before column experiments, the sand was by plating onto TSA with 5% sheep’s blood (Hemostat sterilized by autoclaving at 1218C and 103 kPa (15 psi) for Labs). Plates were incubated microaerophilically at 378C for 20 min. 48 h. Final culturable cell numbers for C. jejuni were 2.9 3 108 cells mL21. Column Preparation Cell Properties Columns were wet packed by slowly pouring autoclaved sand into 2.5-cm-diameter Chromaflex Chromatography Col- Hydrophobicity was measured using the microbial adhesion umns (Kontes Glass Co., Vineland, NJ) filled with electrolyte to hydrocarbons (MATH) assay (Rosenberg et al., 1980). Four solution (10 or 1 mM KCl; Fisher Scientific). Columns were milliliters of a bacterial solution containing approximately repeatedly tapped during the addition of sand to prevent any 2 Reproduced from Journal of Environmental Quality. Published by10 ASA, CSSA,8 and SSSA.cells All copyrights reserved. mL 1 was added to 13-mL test tubes containing 1 mL entrapment of air bubbles within pore spaces. Column lengths n-dodecane (Fisher Scientific, Hampton, NH). The test tubes ranged from 9.8 to 10.2 cm. After packing was completed, were vortexed for 2 min and allowed to stand for 45 min to columns were operated in a downflow direction using a peri- allow the complete separation of the hydrocarbon and water staltic pump, and approximately 10 pore volumes (where a phases. Afterward, 1 mL of the water phase was removed from pore volume represents the volume of water contained within each test tube and the bacterial concentration in the aqueous the sand pack) of the appropriate electrolyte solution were phase was measured. Concentrations of E. coli and C. jejuni passed through each column to equilibrate the sand pack. were determined by measuring the optical density of the Darcian velocity was approximately 3 m d21. All column samples at a wavelength of 546 nm with a UV–visible spec- experiments were run in duplicate. 1020 J. ENVIRON. QUAL., VOL. 35, JULY–AUGUST 2006

A 0.5-pore-volume pulse of either E. coli or C. jejuni at a The attachment rate, k, is a function of the physical, chemi- concentration of approximately 108 cells mL21 was injected at cal, and biological factors which affect a cell’s ability to collide the top of each column followed by bacteria-free electrolyte with, and remain attached to, a collector surface; in our case a solution. Effluent was collected in bulk (i.e., effluent was col- quartz sand grain. Based on theoretical considerations we can lected for each column as a single sample) for approximately define the attachment rate as (Tien et al., 1979): 3.5 pore volumes to ensure the main pulse of bacteria had exited the column. Enumeration of effluent after the 3.5-pore- 3(1 2u)m k 5 h0a [4] volume collection time yielded negligible numbers of bacteria 2dc indicating minimal tailing. Column influent and effluent was where d is the collector (sand grain) diameter (0.3 mm); u is enumerated for total cell concentrations by staining the ap- c porosity (ranging from 0.36 to 0.39 in our study); h0 is the propriate dilutions with a 0.1% solution of Acridine Orange single-collector contact efficiency, which is a measure of the (AO) (Sigma, St. Louis, MO) and filtering onto 0.22-mm black physical mechanisms that bring a bacterial cell to the surface of polycarbonate filters, using a modification of the AO direct the grain; and a is the attachment or “sticking” efficiency, which count method (Hobbie et al., 1977). Stained cells were ob- is a measure of the successful attachment of bacterial cells to served using a BX-41 microscope (Olympus, Melville, NY) the sediment surface (Yao et al., 1971). Although equations equipped with the appropriate lamp and filter set (100-W exist for calculating the single-collector contact efficiency, h0, mercury lamp, a 460–490-nm excitation filter, and a 590-nm the derivation of these equations is based on spherical particles cut-off filter). A minimum of 10 fields were counted to ensure (Logan et al., 1995; Rajagopalan and Tien, 1976; Tufenkji and the enumeration of a statistically significant number of cells. Elimelech, 2004). Because our bacteria are not spherical in shape and deviations from sphericity have been observed to Data Analysis affect bacterial transport in porous media (Weiss et al., 1995), we do not follow the common approach of calculating a from Assuming low surface coverage and negligible entrainment estimates of h0. Rather, we calculated the combination of h0 the one-dimensional transport of bacteria through laboratory and a, hereafter referred to as the removal efficiency (Tufenkji columns can be described by (Harvey and Garabedian, 1991; and Elimelech, 2004), because we believe that the biological Hornberger et al., 1992): differences between these two microorganisms will affect both the single-collector efficiency, h , and the sticking efficiency, a. ]c ]2c ]c 0 5 2 m 2 Combining and rearranging Eq. [3] and [4] allows us to calcu- D 2 kc [2] ]t ]x ]x late the removal efficiency (h0a) from the relative recovery (R ) in the column effluent by: where c is the aqueous concentration of bacteria (cells mL21), r "# x is the distance from the column inlet (cm), D is the hydro- 2 2 21 2dc ln(Rr) dynamic dispersion coefficient (cm h ), v is the interstitial h0a5 2ln(Rr) 1 [5] pore water velocity (cm h21), and k is the bacterial attachment 3(1 2u)L Pe rate (h21) which controls the rate at which bacteria are removed from the aqueous phase. High rates of attachment Because our experiments were all conducted under similar result in reduced transport in ground water environments. physicochemical conditions (e.g., flow rate, porosity, ionic strength, Integrating the analytical solution to Eq. [2] over time yields grain size) we assume that any observed differences in our re- an expression for the relative recovery of bacteria eluted at moval efficiency can be attributed to differences in cell proper- the column outlet (Bolster et al., 1998; Harvey and Garabe- ties such as size and shape, surface charge, and hydrophobicity. dian, 1991). The bacterial attachment (or removal) rate, k, Hydrophobicity, surface charge, relative recovery (Rr), and can then be calculated from the relative recovery by (Bolster removal efficiency (h0a) values were statistically analyzed et al., 1998): using one-sided t tests assuming equal variances whereas dif- "# ! ferences in cell geometry (length, width, and shape) were ana- 2 lyzed using one-sided t tests assuming unequal variances. All ln(Rr) m k 52ln(Rr) 1 [3] statistical analyses were performed using SAS Version 9.1 Pe L (SAS Institute, 2003).

where Rr is the relative recovery defined as the number of bacteria recovered in the column effluent normalized by the RESULTS AND DISCUSSION total number of bacteria introduced into the system, L is column length, and Pe is the Peclet number (dimensionless), Cell Properties which represents the ratio of advective to dispersive forces. We estimated the Peclet number in our columns to be 110 based on The electrophoretic mobility of the two bacteria bromide tracer tests conducted on larger diameter columns tested in our study varied markedly, with C. jejuni hav- that contained the same sand and were of the same length as ing a significantly ( p , 0.001) greater negative charge the columns used in this study. than the E. coli strain used in our study (Table 1). We

Table 1. Average values and standard deviations (shown in parentheses) of measured biological properties of C. jejuni and E. coli Reproduced from Journal of Environmental Quality. Published by ASA, CSSA,O157:H7. and SSSA. All copyrights reserved. Isolate Cell mobility*** (n 5 5) MATH*** (n 5 6)† Cell length (n 5 300) Cell width*** (n 5 300) Cell shape (W/L)** (n 5 300) mmcmV21 s21 % mm C. jejuni 22.7 (0.10) 43 (2.2) 2.2 (0.89) 1.0 (0.30) 0.53 (0.21) E. coli O157:H7 20.19 (0.11) 26 (6.2) 2.1 (0.47) 1.2 (0.19) 0.57 (0.14) ** Significant differences between microorganisms at the 0.01 probability level. *** Significant differences between microorganisms at the 0.001 probability level. † MATH, microbial adhesion to hydrocarbons. BOLSTER ET AL.: COMPARISON OF E. COLI AND C. JEJUNI TRANSPORT IN POROUS MEDIA 1021

21 21 measured average cell mobilities of 22.7 mmcmV s 0.1 21 21 for C. jejuni and 20.19 mmcmV s for E. coli O157: E. coli H7 suspended in 10 mM KCl. Our measurements are C. jejuni ) α

consistent with other reported values for E. coli strains 0 belonging to the serogroup O157, including reported η mobility values ranging from 20.22 to 20.46 mmcmV21 s21 for seven strains of O157:H7 (Lytle et al., 1999) and 0.01 20.31 mmcmV21 s21 for another O157:H7 strain (Lytle et al., 2002). We are unaware of any reported zeta potential or electrophoretic mobility measurements for C. jejuni but our results are in agreement with those of Removal Efficiency ( Walan and Kihlstrom (1988) who reported a highly negative surface charge based on ion-exchange chromatography measurements for all C. jejuni strains tested in 0.001 their study. In addition to surface charge, hydrophobicity differed significantly (p , 0.001) between the two microorgan-

isms. On average 43% of the C. jejuni cells partitioned in Uncoated 1 mM Uncoated 10 mM Al-coated 10 mM the hydrocarbon, whereas only 26% of the E. coli cells Fe-coated 10 mM did the same (Table 1). Our results indicate that the more Treatment Type strongly charged C. jejuni cells are significantly more Fig. 1. Removal efficiency for E. coli and C. jejuni in replicate 10-cm hydrophobic than the less negatively charged E. coli columns under the following conditions: uncoated quartz sand with cells. Our observation that hydrophobicity and mobility 10 mM KCl, uncoated quartz sand with 1 mM KCl, Fe-coated sand are positively correlated is not uncommon (McCaulou with 10 mM KCl, and Al-coated sand with 10 mM KCl. Error bars represent one standard deviation of the mean. et al., 1994; van Loosdrecht et al., 1987a). It is worth noting, however, that the microbial adhesion to hydrocarbons (MATH) assay is an indirect measure of hy- efficiencies were greater by 38 to 52% for C. jejuni as drophobicity and it likely measures a combination of compared to E. coli (Fig. 1). Due to the nonlinear re- hydrophobic and electrostatic interactions (Geertsema- lationship between bacterial recovery and attachment Doornbusch et al., 1993), thus our observed correlation rates, however, relative recovery of E. coli exceeded that may not hold for other organisms or test conditions. of C. jejuni by over an order of magnitude, although Size and shape were also observed to vary between recovery for both microorganisms was quite low (Fig. 2). the two microorganisms (Table 1). Consistent with other In contrast, in both experiments with uncoated quartz studies, we observed E. coli to be rod-shaped and C. sand, removal efficiencies for E. coli exceeded that for jejuni to have two distinct morphologies: one being a C. jejuni. Thus, there was greater transport of the patho- coccoid shape and the other being a slender, spirally genic C. jejuni as compared to the indicator organism curved rod with a characteristic “seagull-wing” shape (Smibert, 1978). The two distinct cell morphologies for 1 C. jejuni have been hypothesized to be a result of dif- E. coli ferences in cell viability (Boucher et al., 1994). On C. jejuni ) average, C. jejuni was slightly longer ( p 5 0.51) and sig- r 0.1 R nificantly narrower (p , 0.001) than E. coli with average lengths of 2.2 and 2.1 mm and average widths of 1.0 and 1.2 mm for C. jejuni and E. coli, respectively. Although 0.01 the difference appears minor, C. jejuni was, on average, significantly ( p , 0.01) less spherical than E. coli with average cell shapes of 0.53 for C. jejuni and 0.57 for 0.001 E. coli. In looking at the size distribution, C. jejuni had a Relative Recovery ( noticeably greater range in length and width than E. coli (data not shown) with the smaller C. jejuni cells repre- 0.0001 senting the coccoid morphology. No attempt was made to compare cell size, shape, and distribution between the column influent and effluent. Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.

Uncoated 1 mM Transport Experiments Uncoated 10 mM Fe-coated 10 mMAl-coated 10 mM Treatment Type The observed differences in cell properties between these two microorganisms resulted in significant differ- Fig. 2. Relative recovery of E. coli and C. jejuni from replicate 10-cm columns under the following conditions: uncoated quartz sand with ences (p , 0.05) in relative recovery and removal effi- 10 mM KCl, uncoated quartz sand with 1 mM KCl, Fe-coated sand ciency for all experimental treatments (Fig. 1 and 2). In with 10 mM KCl, and Al-coated sand with 10 mM KCl. Error bars columns packed with metal-oxide-coated sands, removal represent one standard deviation of the mean. 1022 J. ENVIRON. QUAL., VOL. 35, JULY–AUGUST 2006

E. coli (Fig. 1 and 2). Given that C. jejuni and E. coli like-charged surfaces is governed by a combination of differed in size, shape, size distribution, surface charge, electrostatic double layer repulsion (Hogg et al., 1966) and hydrophobicity, all of which have been shown to and attractive van der Waals forces (Gregory, 1981). affect bacterial attachment to surfaces (Gannon et al., Because bacteria and uncoated quartz sand are both 1991; Stenstrom, 1989; van Loosdrecht et al., 1987a; negatively charged particles, the electrical double layer Weiss et al., 1995), it is difficult to determine which of of counter ions around their surfaces results in electro- these factors was most important in controlling these static repulsion. As ionic strength is lowered, the thick- differences in transport. We can, however, make some ness of this electrical double layer increases as does the inferences based on our results. magnitude of electrostatic double layer repulsive forces. The recovery of bacteria in the column effluent was Depending on the thickness of the double layer, most strongly affected by surface coatings on the quartz bacterial cells may be prohibited from approaching the sand grains. For both microorganisms, the relative surface at a sufficiently close distance that would allow recovery greatly decreased for the metal-oxide-coated for attractive van der Waals forces to overcome electro- sands as compared to the uncoated sand (Fig. 2); how- static repulsion. Thus, decreasing ionic strength is expected ever, the effect was more pronounced for C. jejuni.The to decrease bacterial attachment to like-charged sur- average relative recovery for C. jejuni decreased from faces and has been confirmed in numerous studies 0.35 for the uncoated sand (10 mM KCl) to 0.00034 and (Fontes et al., 1991; Martin et al., 1991; Mills et al., 1994; 0.0013 for the Fe- and Al-coated sand, respectively. For Scholl et al., 1990). Given the stronger negative charge E. coli, the average relative recovery decreased from of C. jejuni, it is not surprising that changes in ionic 0.11 in uncoated sand (10 mM KCl) to 0.0033 and 0.012 strength would have a more pronounced influence on for the Fe- and Al-coated sand, respectively (Fig. 2). deposition of this organism as compared to the near Observed reductions in relative recovery for oxide- neutrally charged E. coli. coated sands were statistically significant at the 0.05 We can also use DLVO theory to better understand probability level for both C. jejuni and E. coli. the observed differences in attachment rates to the un- The presence of the metal-oxide coatings on the coated quartz sand for E. coli and C. jejuni. Using DLVO quartz sand grains resulted in an electrostatic attraction theory to calculate the total interaction energy between between the positively charged metal-oxide coating and our bacterial cells and a quartz surface indicates that at the negatively charged bacteria leading to increased an ionic strength of 10 mM KCl, repulsive electrostatic attachment rates; this observation is in agreement with forces are suppressed and no energy barrier exists for other published studies (Bolster et al., 2001; Johnson E. coli attachment to negatively charged quartz sand. and Logan, 1996; Mills et al., 1994; Scholl and Harvey, This suggests that irreversible attachment in the primary 1992; Scholl et al., 1990). This increase in attachment minimum is occurring at this ionic strength (Fig. 3). On rates, leading to a decrease in bacterial transport, is evi- the other hand, C. jenuni experiences a substantial re- dent when comparing the removal efficiency (h0a) be- pulsive energy barrier to deposition at this ionic strength tween coated and uncoated sands (Fig. 1). For both due to the strong negative charge of this organism mak- microorganisms, the removal efficiency greatly in- ing deposition in the primary minimum unlikely; how- creased (p , 0.05) for the metal-oxide-coated sands as ever, a shallow energy well, or secondary minimum, compared to the uncoated sand but the increase in re- exists at a separation distance of approximately 17 nm moval efficiency was greatest for C. jejuni. This ob- between the bacterium and quartz surface (Fig. 3). This servation is not surprising as the C. jejuni cells were energy minimum is approximately 2.7 kT, which is suf- considerably more negatively charged, therefore, the ficiently deep to allow a bacterium to be retained (Hahn presence of positively charged metal-oxide coatings and O’Melia, 2004). This provides an explanation for would be expected to have a more pronounced influence why experiments with uncoated quartz sand resulted in on deposition for this organism. some removal of C. jejuni even though considerable re- Changing the ionic strength of the carrier fluid was pulsion was expected, as these cells likely were captured also identified as playing a role in the extent of bacterial in the secondary energy minimum (Redman et al., 2004). attachment to the uncoated quartz sand. Decreasing ionic The effect of surface coatings and ionic strength on strength of the carrier solution from 10 to 1 mM resulted attachment rates for our two organisms demonstrates in slight increases in relative recovery from columns con- that surface charge is likely an important factor causing taining uncoated sands for both types of bacteria (Fig. 2), the observed differences in transport between E. coli although the increase for E. coli was not statistically sig- and C. jejuni. To fully evaluate the relative transport of nificant (p 5 0.28). Decreasing ionic strength resulted in a E. coli and C. jejuni, however, we must account for other statistically significant (p 5 0.049) decrease in removal ef- surface properties and the potential coupled influence Reproduced from Journal of Environmental Quality. Publishedficiency by ASA, CSSA, and SSSA. All copyrights reserved. (0.0033 to 0.0021) for C. jejuni but for E. coli the of these properties on cell transport and deposition. In influence of lowered ionic strength on removal efficiency columns containing metal-oxide-coated quartz, when was insignificant (p 5 0.22). greater hydrophobicity and more negative surface The impact of changing ionic strength on attachment charge would suggest greater deposition for C. jejuni, behavior can be explained by the Derjaguin–Landau– removal efficiencies were greater for C. jejuni than for Verwey–Overbeek (DLVO) theory of colloidal stability E. coli (Fig. 1). In this case, the combined hydrophobic (Derjaguin and Landau, 1941; Verwey and Overbeek, and electrostatic forces resulted in anticipated trends. 1948). According to this theory, bacterial deposition to In experiments utilizing negatively charged uncoated BOLSTER ET AL.: COMPARISON OF E. COLI AND C. JEJUNI TRANSPORT IN POROUS MEDIA 1023

50 hydrophobic, highly negatively charged bacterium in E. coli 40 columns containing uncoated quartz sand. In addition to surface properties, cell geometry has 30 Vtotal (kT) been shown to be a major factor affecting bacterial 20 VEDL (kT) transport. Theoretical considerations indicate cell size V (kT) 10 VDW should be important in attachment rates (Harvey and Garabedian, 1991; Yao et al., 1971) and this has been 0 observed in laboratory studies. Gannon et al. (1991) -10 observed the transport of a variety of bacterial strains -20 through a loam soil to be correlated with cell size rather

Interaction Energy (kT) than with hydrophobicity or surface charge. Fontes et al. -30 (1991) also observed differences in bacterial transport -40 due to differences in cell size. Cell shape has also been -50 shown to affect bacterial transport with more spherical 0 20 40 60 80 100 cells being favored for transport (Weiss et al., 1995). If cell geometry (size, shape, and distribution) was the 50 primary factor controlling bacterial transport in our C. jejuni 40 study we would expect that the disparity in removal efficiency between the two microorganisms would remain 30 Vtotal (kT) relatively unchanged between experimental treatments. 20 VEDL (kT) Instead, we observed that the removal efficiency for V (kT) 10 VDW each microorganism responded differently to changes in 0 physicochemical conditions such as surface coatings and ionic strength. Thus, it is unlikely that the measured dif- -10 ferences in cell size and shape between our two micro- -20 organisms were substantial enough to have a noticeable Interaction Energy (kT) -30 impact on attachment rates. When viewed in their entirety, our results strongly -40 suggest that surface charge was the primary factor de- -50 termining the extent of deposition and transport of these 0 20 40 60 80 100 two microorganisms. This conclusion is supported by the Separation Distance (nm) following observations: (i) in column experiments con-

Fig. 3. Total interaction energy (Vtotal) as calculated with Derjaguin– taining positively charged metal-oxide-coated sands, Landau–Verwey–Overbeek (DLVO) theory as a combination of attachment was greater for the more negatively charged electrostatic double layer (VEDL) and van der Waals (VVDW) forces C. jejuni; (ii) in columns packed with negatively charged between E. coli and C. jejuni and the quartz sand surface at 10 mM uncoated quartz sand, greater attachment was observed KCl. Calculations assume sphere–plate interactions (Gregory, 1981; Hogg et al., 1966) and a Hamaker constant of 6.5 3 10221 J for the near neutrally charged E. coli, due to less elec- (Redman et al., 2004), and utilized a zeta potential value for quartz trostatic repulsion as compared to the more negatively previously reported (Walker et al., 2002). charged C. jejuni; and (iii) the sensitivity of the removal efficiency to changes in ionic strength with the nega- quartz sand, removal efficiencies for the more hydro- tively charged uncoated sand was greater for the more philic and near neutrally charged E. coli exceeded that negatively charged C. jejuni. This is not to say that dif- of C. jejuni. If hydrophobicity were the driving force in ferences in cell geometry and hydrophobicity do not our experiments we would expect greater attachment of play a role in transport. Rather, our results indicate that the more hydrophobic microorganism, C. jejuni, to the for the experimental conditions used in this study these uncoated quartz sand than the hydrophilic organism, E. factors played a secondary role, if any, and are consistent coli, yet we observed the opposite (Stenstrom, 1989; van with the findings of others who have reported that bac- Loosdrecht et al., 1987b). This suggests that electrostatic terial attachment to hydrophilic surfaces is primarily repulsion between uncoated quartz sand and C. jejuni is controlled by electrostatic interactions (Gross and substantial enough to overcome the hydrophobic inter- Logan, 1995; McCaulou et al., 1994; van Loosdrecht actions that might otherwise control the extent of depo- et al., 1990). sition. Additionally, the near neutrally charged, more hydrophilic E. coli cells deposited on the like-charged Reproduced from Journal of Environmental Quality. Published byquartz, ASA, CSSA, and SSSA. All copyrights reserved. indicating that hydrophobic forces were sur- CONCLUSIONS passed by the forces incorporated in traditional DLVO Although the ability of Campylobacter spp. to move theory. Our findings are similar to those reported by through ground water environments has been inferred McCaulou et al. (1994) who also observed greater re- from ground water sampling studies (Stanley et al., covery of a hydrophilic bacterium with a low negative 1998), we are unaware of any studies that have looked charge compared to a hydrophobic bacterium with a specifically at the transport of C. jejuni through porous high negative surface charge in columns containing media. Our study suggests that the transport of C. jejuni Fe-coated sands but observed greater recovery of the may exceed that of the commonly used bacterial 1024 J. ENVIRON. QUAL., VOL. 35, JULY–AUGUST 2006

indicator, E. coli, under conditions where removal rates Derjaguin, B.V., and L. Landau. 1941. Theory of stability of strongly are expected to be low (e.g., when bacteria and col- charged lyphobic soils and of the adhesion of strongly charged particles in solutions of electrolytes. Acta Physicochim. URSS 14: lectors are similarly charged). The practical significance 633–662. of this finding is that C. jejuni may be present in ground Fontes, D.E., A.L. Mills, G.M. Hornberger, and J.S. Herman. 1991. water samples in which E. coli is not detected. Ad- Physical and chemical factors influencing transport of microorgan- mittedly, this study was developed with a simplified ap- isms through porous media. Appl. Environ. Microbiol. 57:2473–2481. Gagliardi, J.V., and J.S. Karns. 2002. Persistence of Escherichia coli proach and does not address important factors such as O157:H7 in soil and on plant roots. Environ. Microbiol. 4:89–96. physical, chemical, and biological heterogeneities (Bol- Gannon, J.T., V.B. Manilal, and M. Alexander. 1991. Relationship be- ster et al., 2000); limitations with colloid-filtration theory tween cell surface properties and transport of bacteria through soil. (Bradford et al., 2002); and recent observations that Appl. Environ. Microbiol. 57:190–193. different strains of E. coli transport differently through Geertsema-Doornbusch, G.I., H.C. van der Mei, and H.J. Busscher. 1993. The involvement of electrostatic interactions in microbial ad- porous media (Morrow et al., 2005). As such, the values hesion to hydrocarbons (MATH). J. Microbiol. Methods 18:61–68. for the removal efficiencies obtained in this study may Gerba, C.P., and J.E. Smith, Jr. 2005. Sources of pathogenic micro- be viewed as effective values representative for the con- organisms and their fate during land application of wastes. J. ditions of this study. Nevertheless, our results are helpful Environ. Qual. 34:42–48. Gregory, J. 1981. Approximate expressions for retarded Vanderwaals in understanding the potential problems and limitations interaction. J. Colloid Interface Sci. 83:138–145. in using E. coli as a surrogate for the transport of C. Gross, M.J., and B.E. Logan. 1995. Influence of different chemical jejuni and other pathogenic microorganisms in ground treatments on transport of Alcaligenes paradoxus in porous media. water environments. Indeed, these results bring into Appl. Environ. Microbiol. 61:1750–1756. question the validity of using E. coli as an indicator or- Hahn, M.W., and C.R. O’Melia. 2004. Deposition and reentrainment of brownian particles in porous media under unfavorable chemical ganism for this important pathogen. Clearly, further conditions: Some concepts and applications. Environ. Sci. Technol. study is needed to establish the correlation between the 38:210–220. presence of C. jejuni and indicator organisms in envi- Harvey, R.W., and S.P. Garabedian. 1991. Use of colloid filtration ronmental water samples, and to determine if the use theory in modeling movement of bacteria through a contaminated sandy aquifer. Environ. Sci. Technol. 25:178–185. of indicator organisms such as E. coli is a meaningful Hobbie, J.E., R.J. Daley, and S. Jasper. 1977. Use of Nucleopore filters tool to assess potential ground water contamination by for counting bacteria by fluorescence microscopy. Appl. Environ. C. jejuni. Microbiol. 33:1225–1228. Hogg, R., T.W. Healy, and D.W. Fuerstenau. 1966. Mutual coagulation of colloidal dispersions. Trans. Faraday Soc. 62:1638–1651. ACKNOWLEDGMENTS Hornberger, G.M., A.L. Mills, and J.S. Herman. 1992. Bacterial This research was part of USDA-ARS National Program transport in porous media: Evaluation of a model using laboratory 206: Manure and Byproduct Utilization. This manuscript was observations. Water Resour. Res. 28:915–938. Johnson, W.P., and B.E. Logan. 1996. Enhanced transport of bacteria greatly improved by the thoughtful comments of Scott Brad- in porous media by sediment-phase and aqueous-phase natural or- ford and three anonymous reviewers. ganic matter. Water Res. 30:923–931. Leclerc, H., and D.A.A. Mosel. 2001. Advances in bacteriology of the REFERENCES coliform group: Their suitability as markers of microbial water safety. Annu. Rev. Microbiol. 55:201–234. Allos, B.M. 2001. Campylobacter jejuni infections: Update on emerg- Logan, B.E., D.G. Jewett, R.G. Arnold, E.J. Bouwer, and C.R. ing issues and trends. Clin. Infect. Dis. 32:1201–1206. O’Melia. 1995. Clarification of clean-bed filtration models. J. Bolster, C.H., G.M. Hornberger, A.L. Mills, and J.L. Wilson. 1998. A Environ. Eng. 121:869–873. method for calculating bacterial deposition coefficients using the Lytle, D.A., C.H. Johnson, and E.W. Rice. 2002. A systematic fraction of bacteria recovered from laboratory columns. Environ. comparison of electrokinetic properties of environmentally impor- Sci. Technol. 32:1329–1332. tant microorganisms in water. Colloids Surf. B 24:91–101. Bolster, C.H., A.L. Mills, G.M. Hornberger, and J.S. Herman. 2000. Lytle, D.A., E.W. Rice, C.H. Johnson, and K.R. Fox. 1999. Elec- Effect of intra-population variability on the long distance transport trophoretic mobilities of Escherichia coli O157:H7 and wild-type of bacteria. Ground Water 38:370–375. Escherichia coli strains. Appl. Environ. Microbiol. 65:3222–3225. Bolster, C.H., A.L. Mills, G.M. Hornberger, and J.S. Herman. 2001. Macler, B., and J.C. Merkle. 2002. Current knowledge on groundwater Effect of surface coatings, grain size, and ionic strength on the maxi- pathogens and their control. Hydrogeol. J. 9:29–40. mum attainable coverage of bacteria on sand surfaces. J. Contam. Martin, R.E., L.M. Hanna, and E.J. Bouwer. 1991. Determination of Hydrol. 50:287–305. bacterial collision efficiencies in a rotating disk system. Environ. Borchardt, M.A., P.D. Bertz, S.K. Spencer, and D.A. Battigelli. 2003. Sci. Technol. 25:2082–2087. Incidence of enteric viruses in groundwater from household wells in Mawdsley, J.L., R.D. Bardgett, R.J. Merry, B.F. Pain, and M.K. Wisconsin. Appl. Environ. Microbiol. 69:1172–1180. Theodorou. 1995. Pathogens in livestock waste, their potential for Boucher, S.N., E.R. Slater, A.H.L. Chamberlain, and M.R. Adams. movement through soil and environmental-pollution. Appl. Soil 1994. Production and viability of coccoid forms of Campylobacter Ecol. 2:1–15. jejuni. J. Appl. Bacteriol. 77:303–307. McCaulou, D.R., R.C. Bales, and J.F. McCarthy. 1994. Use of short- Bradford, S.A., S.R. Yates, M. Bettahar, and J. Simunek. 2002. Physical pulse experiments to study bacteria transport through porous factors affecting the transport and fate of colloids in saturated porous media. J. Contam. Hydrol. 15:1–15. Reproduced from Journal of Environmental Quality. Published by ASA, CSSA,media. and SSSA. All copyrights reserved. Water Resour. Res. 38:1327, doi:10.1029/2002WR001340. Media Cybernetics. 2004. Image Pro Plus start-up guide. Media Busscher, H.J., A.H. Weerkamp, H.C. van der Mei, A.W.J. van Pelt, Cybernetics, Silver Springs, MD. H.P. de Jong, and J. Arends. 1984. Measurement of the surface free Meschke, J.S., and M.D. Sobsey. 2003. Comparative reduction of Nor- energy of bacterial cell surfaces and its relevance for adhesion. walk virus, poliovirus type 1, F+RNA coliphage MS2 and Escherichia Appl. Environ. Microbiol. 48:980–983. coli in miniature soil columns. Water Sci. Technol. 47:85–90. Clark, C.G., L. Price, R. Ahmed, D.L. Woodward, P.L. Melito, F.G. Mills, A.L., J.S. Herman, G.M. Hornberger, and T.H. DeJesus. 1994. Rodgers, F. Jamieson, B. Ciebin, A. Li, and A. Ellis. 2003. Charac- Effect of solution ionic strength and iron coatings on mineral grains terization of waterborne outbreak-associated Campylobacter jejuni, on the sorption of bacterial cells to quartz sand. Appl. Environ. Walkerton, Ontario. Emerg. Infect. Dis. 9:1232–1241. Microbiol. 60:3300–3306. BOLSTER ET AL.: COMPARISON OF E. COLI AND C. JEJUNI TRANSPORT IN POROUS MEDIA 1025

Morrow, J.B., R. Stratton, H.-H. Yang, B.F. Smets, and D. Grasso. 2005. predicting single-collector efficiency in physicochemical filtration Macro- and nanoscale observations of adhesive behavior for in saturated porous media. Environ. Sci. Technol. 38:529–536. several E. coli strains (O157:H7 and environmental isolates) on USEPA. 1993. The report of the EPA/State feedlot workgroup. mineral surfaces. Environ. Sci. Technol. 39:6395–6404. USEPA, Washington, DC. Rajagopalan, R., and C. Tien. 1976. Trajectory analysis of deep-bed USEPA. 1994. National water quality inventory: 1992 Report to filtration with the sphere-in-cell porous media model. AIChE J. Congress. USEPA, Washington, DC. 22:523–533. Valcour,J.E.,P.Miche,S.A.McEwen,andJ.B.Wilson.2002. Redman, J.A., S.L. Walker, and M. Elimelech. 2004. Bacterial ad- Associations between indicators of livestock farming intensity and hesion and transport in porous media: Role of the secondary energy incidence of human Shiga toxin-producing Escherichia coli infec- minimum. Environ. Sci. Technol. 38:1777–1785. tion. Emerg. Infect. Dis. 8:252–257. Rosenberg, M., D. Gutnick, and E. Rosenberg. 1980. Adherence of van Loosdrecht, M.C.M., J. Lyklema, W. Norde, G. Schraa, and A.J.B. bacteria to hydrocarbons: A simple method for measuring cell- Zehnder. 1987a. Electrophoretic mobility and hydrophobicity as a surface hydrophobicity. FEMS Microbiol. Lett. 9:29–33. measure to predict the initial steps of bacterial adhesion. Appl. SAS Institute. 2003. SAS user’s guide: Statistics. SAS Inst., Cary, NC. Environ. Microbiol. 53:1898–1901. Scholl, M.A., and R.W. Harvey. 1992. Laboratory investigations on the van Loosdrecht, M.C.M., J. Lyklema, W. Norde, G. Schraa, and A.J.B. role of sediment surface and groundwater chemistry in transport Zehnder. 1987b. The role of bacterial cell wall hydrophobicity in of bacteria through a contaminated sandy aquifer. Environ. Sci. adhesion. Appl. Environ. Microbiol. 53:1893–1897. Technol. 26:1410–1417. van Loosdrecht, M.C.M., W. Norde, J. Lyklema, and A.J.B. Zehnder. Scholl, M.A., A.L. Mills, J.S. Herman, and G.M. Hornberger. 1990. 1990. Hydrophobic and electrostatic parameters in bacterial The influence of mineralogy and solution chemistry on the attach- adhesion. Aquat. Sci. 52:103–114. ment of bacteria to representative aquifer materials. J. Contam. Verwey, E.J.W., and J.T.G. Overbeek. 1948. Theory of the stability of Hydrol. 6:321–336. lyophobic colloids. Elsevier Publ., Amsterdam. Siragusa, G.R., K. Nawotka, S.D. Spilman, P.R. Contag, and C.H. Walan, A., and E. Kihlstrom. 1988. Surface charge and hydrophobicity Contag. 1999. Real-time monitoring of Escherichia coli O157:H7 of Campylobacter jejuni strains in relation to adhesion to epithelial adherence to beef carcass surface tissues with a bioluminescent HT-29 cells. APMIS 96:1089–1096. reporter. Appl. Environ. Microbiol. 65:1738–1745. Walker, S.L., S. Bhattacharjee, E.M.V. Hoek, and M. Elimelech. 2002. Smibert, R.M. 1978. The genus Campylobacter. Annu. Rev. Microbiol. A novel asymmetric clamping cell for measuring streaming 32:673–709. potential of flat surfaces. Langmuir 18:2193–2198. Stanley, K.N., R. Cunningham, and K. Jones. 1998. Isolation of Cam- Weiss, T.H., A.L. Mills, J.S. Herman, and G.M. Hornberger. 1995. The pylobacter jejuni from groundwater. J. Appl. Microbiol. 85:187–191. effect of cell size and cell shape on bacterial retention in sand Stanley, K.N., and K. Jones. 2003. Cattle and sheep farms as reservoirs columns. Environ. Sci. Technol. 29:1737–1740. of Campylobacter. J. Appl. Microbiol. 94:104S–113S. Wesley, I.V., S.J. Wells, K.M. Harmon, A. Green, L. Schroeder-Tucker, Stenstrom, T.A. 1989. Bacterial hydrophobicity, an overall parameter M. Glover, and I. Siddique. 2000. Fecal shedding of Campylobacter for the measurement of adhesion potential to soil particles. Appl. and Arcobacter spp. in dairy cattle. Appl. Environ. Microbiol. 66: Environ. Microbiol. 55:142–147. 1994–2000. Tien, C., R.M. Turian, and H. Pandse. 1979. Simulation of deep bed Yao, K.-M., M.T. Habibian, and C.R. O’Melia. 1971. Water and waste filters. AIChE J. 25:385–395. water filtration: Concepts and applications. Environ. Sci. Technol. Tufenkji, N., and M. Elimelech. 2004. Correlation equation for 5:1105–1112. Reproduced from Journal of Environmental Quality. Published by ASA, CSSA, and SSSA. All copyrights reserved.