Infectious diseases caused by bacteria that are resistant to antimicrobial agents are a serious public health concern (Neu, Reference Neu1992Barbosa & Levy, Reference Barbosa and Levy2000b; Levy & Marshall, Reference Levy and Marshall2004). Members of the Enterobacteriaceae family of bacteria are causative agents of infectious disease, and their resistance to antimicrobial agents compromises chemotherapeutic efforts (Paterson, Reference Paterson2002). Antimicrobial agents are used in agriculture for the treatment of infection, prophylaxis and growth promotion (Levy, Reference Levy2002; Levy & Marshall, Reference Levy and Marshall2004).
Bacterial resistance to antimicrobial agents may be of clinical significance due to dissemination of pathogenic bacteria through a population of food animals (McDermott et al. Reference McDermott, Zhao, Wagner, Simjee, Walker and White2002; Angulo et al. Reference Angulo, Baker, Olsen, Anderson and Barrett2004; Silbergeld et al. Reference Silbergeld, Graham and Price2008). In isolated instances, outbreaks of food-borne infectious disease from contaminated milk and occurrences of milk-derived isolates of members from the Enterobacteriaceae family of bacteria have been studied (Gillespie et al. Reference Gillespie, Adak, O'Brien and Bolton2003). Although relatively well-studied within dairy cattle (Hershberger et al. Reference Hershberger, Oprea, Donabedian, Perri, Bozigar, Bartlett and Zervos2005), bacteria that are resistant to antimicrobial agents in other dairy farm environments, such as soil (Burgos et al. Reference Burgos, Ellington and Varela2005), water (Biyela et al. Reference Biyela, Lin and Bezuidenhout2004) and milk (Makovec & Ruegg, Reference Makovec and Ruegg2003) are less well-characterized. Dissemination mechanisms for drug resistant bacterial pathogens within dairy farms are also poorly understood (Hershberger et al. Reference Hershberger, Oprea, Donabedian, Perri, Bozigar, Bartlett and Zervos2005).
The marRAB locus is a well studied genetic element in Escherichia coli and mediates bacterial resistance to multiple antimicrobial agents such as β-lactams, chloramphenicol, quinolones, and tetracycline (Randall & Woodward, Reference Randall and Woodward2002). Several organic compounds are known to modulate resistance to antimicrobial agents via the marRAB locus (Alekshun & Levy, Reference Alekshun and Levy1999). Previously, milk-containing foods were shown to modulate the marRAB locus (Rickard et al. Reference Rickard, Lindsay, Lockwood and Gilbert2004); the effects of milk, however, were not directly examined. Thus, we tested the hypothesis that milk confers reduced susceptibilities to antimicrobial agents in Esch. coli cells harboring the marRAB locus. The objective of this study was to examine the relationship between the culturing of Esch. coli in milk and the subsequent development of resistance to antimicrobial agents. Here, we found that upon culturing in pasteurized milk, antimicrobial agent-sensitive Esch. coli becomes less susceptible to multiple antimicrobial agents, a phenotype conferred by the marRAB locus.
Materials and Methods
Bacterial strains
Esch. coli K-12 bearing a deletion of the marRAB operon (strain JHC1096), was a generous gift from Dr. Jean Greenberg (Harvard University, MA; Greenberg et al. Reference Greenberg, Chou, Monach and Demple1991). The Esch. coli strain harbouring a deletion of marR (strain AG112) (George & Levy, Reference George and Levy1983; Moken et al. Reference Moken, McMurry and Levy1997), and a wild-type strain harbouring an intact marRAB locus (strain GC4468) (Carlioz & Touati, Reference Carlioz and Touati1986; Oethinger et al. Reference Oethinger, Podglajen, Kern and Levy1998) were generous gifts from the laboratory of Dr. Stuart Levy (Tufts University, MA).
Growth determination in Luria-Bertani broth and milk
Cells JHC1096, AG112, and GC4468 were grown overnight at 37°C while shaking in Luria-Bertani (LB) broth (Difco), which is routinely used for maintenance and propagation of Esch. coli (Luria et al. Reference Luria, Adams and Ting1960). Then 20 μl of culture was added to 20 ml fresh LB broth and grown to the mid-log phase of growth (O.D.600=0·35). Next, 10 μl of the mid-log culture was used to inoculate 10 ml of fresh LB broth or commercially available pasteurized whole milk. This procedure was repeated as above except that the LB broth and the pasteurized whole milk each contained 5 mm-Salicylate. The cultures were incubated while shaking at 37°C for the times indicated in Figs. 1 & 2. Samples (10 μl) of the cultures were removed at the time points indicated, plated onto MacConkey agar, and incubated for 12 h. Colony forming units (CFU/ml) were then determined. The means plotted were the results of 4 independent experiments, each in triplicate.
Fig. 1. Effects of salicylate on bacterial growth (CFU) in LB broth.
Esh. coli cells of strains AG112 (filled circles), JHC1096 (empty circles) and GC4468 (filled triangles) were incubated in LB broth without inducer (i.e., no salicylate), panel A, or in LB broth with inducer (5 mm-salicylate), panel B. See methods for details
Fig. 2. Effects of salicylate on bacterial growth (CFU) in milk.
Esch. coli cells of strains AG112 (filled circles), JHC1096 (empty circles) and GC4468 (filled triangles) were incubated in pasteurized milk without inducer (i.e., no salicylate), panel A, or in pasteurized milk containing inducer (5 mm-salicylate), panel B. See methods for details
Antimicrobial susceptibility testing
Prior to antimicrobial susceptibility testing via the E-Test assay using current CLSI guidelines (2006), Esch. coli cells JHC1096, AG112, and GC4468 were treated with LB broth (Difco) with and without 5 mm-salicylate, commercially available pasteurized whole milk, or milk count agar (OXIOD) in the following manner: Esch. coli strains JHC1096, AG112 and GC4468 were inoculated in LB Broth (Difco), incubated overnight at 37°C with aeration and then used to inoculate refresher cultures in LB broth (Difco) for growth to the mid-log phase (O.D.600≈0·35). The mid-log cells were used for incubation in LB broth (Difco) (with and without 5 mm-salicylate) at 37°C with aeration for 8 h or for incubation in commercially available pasteurized whole milk at 30°C with aeration for 8 h. An additional 10 ml pasteurized whole milk was inoculated with 10 μl sterile distilled water as a negative control. For the treatment in LB broth (Difco) with or without 5 mm-salicylate, a total of 50 μl of each culture was plated onto LB agar medium (Difco) plates and incubated overnight at 37°C. For the treatment in pasteurized whole milk, a total of 50 μl of each culture was plated onto either LB agar (Difco) or milk count agar (OXIOD) and incubated at 30°C for 24 h.
For the antimicrobial agent susceptibility measurements, three single colonies of each strain were randomly picked from each of the plates, resuspended in 1 ml per colony of 0·85% NaCl and each immediately spread uniformly onto cation-adjusted Muller-Hinton II Agar (Difco) plates or Milk Count Agar. A portion of the 0·85% NaCl resuspension (0·1 ml) was plated onto LB agar, and cells were identified by BD BBL Crystal Enteric/Nonfermentor Identification Kits (Becton Dickinson Microbiology Systems, Cockeysville, MD) according to the manufacturer specifications. E-Test strips (AB Biodisk) were placed on the surface as per the manufacturer's instructions, and the plates were incubated at 37°C for 16 h. The minimal inhibitory concentrations (MICs) were read and interpreted according to current CLSI guidelines (2006). The means±sd (ranges) were the result of three independent experiments, each in triplicate, n=3. Break points for susceptibility interpretations were MICs ⩽8 μg/ml for trimethoprim, ⩽2 μg/ml for gatifloxacin, ⩽8 μg/ml for cefotaxime, and ⩽4 μg/ml for tetracycline. The MIC break point for resistance to tetracycline was ⩾16 μg/ml.
Statistics
All data represent the mean plus or minus the standard deviation and range of at least three independent experiments (n=3) conducted under standardized conditions. The nonparametric Holm-Sidak method was used to determine the significance of mean MIC data compared with mean baseline MICs within each strain. Results were considered significant at P⩽0·05.
Results and Discussion
Baseline determinations of antimicrobial susceptibilities
Prior to culturing Esch. coli strains GC4468 (intact marRAB locus), JHC1096 (marRAB deleted) AG112 (marR deleted) in pasteurized whole milk or milk count agar, the minimal inhibitory concentrations (MICs) were determined by the standard E-test method (2006). All three Esch. coli strains were susceptible to the antimicrobial agents trimethoprim, gatifloxacin, cefotaxime and tetracycline (Table 1). These antimicrobial agents represent distinct drug classes and have different mechanisms of actions (McDermott et al. Reference McDermott, Walker and White2003). These agents also have significance in clinical therapy, such as trimethoprim (Juckett, Reference Juckett1999; Casewell et al. Reference Casewell, Friis, Marco, McMullin and Phillips2003), gatifloxacin (Lutsar et al. Reference Lutsar, Friedland, Jafri, Wubbel, Ng, Ghaffar and McCracken Jr1999), and cefotaxime (Wittmann et al. Reference Wittmann, Jones, Malledant and Privitera1997) or significance in agriculture for the purposes of growth promotion or prophylaxis in dairy calves, such as tetracycline (Constable & Morin, Reference Constable and Morin2002).
Table 1. Minimal inhibitory concentrations of antimicrobial agents for Esch. coli in pasteurized milk, milk count agar or salicylate
MIC values (CLSI) are presented as the mean±sd (range), n=3. See methods for details
* P⩽0·05, **P⩽0·01 and ***P⩽0·001 compared to baseline MICs for the same Esch. coli strain
Antimicrobial susceptibilities after culturing in pasteurized milk and milk count agar
We found that after culturing Esch. coli GC4468 (intact marRAB locus) in pasteurized milk, the antimicrobial agent MICs increased by 1·4-fold for trimethoprim (P=0·05), by 1·5-fold for gatifloxacin (P⩽0·05), and by 2·0-fold for cefotaxime (P=0·008), as shown in Table 1. The 1·4-fold increase in mean MIC for tetracycline was not significantly different in GC4468 (P⩾0·05). The mean MIC in AG112 (marR deleted) increased (P=0·003) by 2·8-fold for trimethoprim, but not for gatifloxacin, cefotaxime, and tetracycline (P=0·10). The MICs for the same antimicrobial agents in Esch. coli AG112 were not significantly different (P⩾0·05) compared with the MIC values obtained prior to culturing in milk and milk count agar, implying that the marRAB locus in AG112 was already intrinsically induced, as expected, and that another inducer (milk) would not further modulate susceptibility. The MICs of the antimicrobial agents tested in this study (ranged between 0·01 and 0·017 μg/ml) for cells of JHC1096 (marRAB locus deleted) after culturing in pasteurized whole milk were indistinguishable (P⩾0·05) from the MICs determined prior to culturing (Table 1) in the same type of milk.
As shown in Table 1, we found that after culturing Esch. coli GC4468 on milk count agar, the antimicrobial agent MICs increased by 3·4-fold for trimethoprim (P=0·036), by 10-fold for gatifloxacin (P=0·001), by 7·1-fold for cefotaxime (P=0·011), and by 40·5-fold for tetracycline (P=0·074) compared with the mean MICs derived prior to culturing on milk count agar. Interestingly, GC4468 cells on milk count agar with a mean MIC 74·67 μg/ml are considered resistant to tetracycline, if not susceptible. As expected, cells of AG112 (marR deleted) were not less susceptible to the antimicrobial agents tested compared with baseline MICs (P⩾0·05), regardless of whether milk or milk count agar was used, further implying that within induced cells, susceptibilities are not further reduced. For instance, AG112 cells were considered tetracycline resistant regardless of the variable tested (Table 1), as would be expected if the marRAB locus was induced. The MICs of the antimicrobial agents for JHC1096 cells (ranging between 0·01–0·017 μg/ml) after culturing in pasteurized whole milk or milk count agar were indistinguishable (P⩾0·05) from the MICs determined prior to culturing (Table 1) in the same milk-type, except for an observed 5-fold increase in the MIC of gatifloxacin (P<0·05) for JHC1096 cultured on milk count agar (but not pasteurized milk) compared with the baseline MIC in gatifloxacin. Prior to culturing in milk or milk agar, Esch. coli strains GC4468 and JHC1096 were susceptible to all antimicrobial agents tested. The reductions in antimicrobial agent susceptibilities in milk were more pronounced in GC4468 after sub-culturing on milk count agar compared with sub-culturing in pasteurized milk. We conclude that culturing in pasteurized milk or milk agar, Esch. coli cells harboring an intact marRAB locus become less susceptible to multiple antimicrobial agents.
Measurement of antimicrobial susceptibilities with salicylate
As a positive control experiment for the induction of the marRAB locus, the antimicrobial agent MICs were determined in 5 mm-salicylic acid, a well-known marRAB locus inducer (Alekshun & Levy, Reference Alekshun and Levy1999). As shown in Table 1, in the presence of salicylate, the MICs increased for trimethoprim (P=0·01) and cefotaxime (P=0·008) in GC4468 cells. AG112 cells (marR deleted) with or without salicylate had tetracycline MICs of 213 and 256 μg/ml, respectively, while JHC1096 cells (marRAB locus deleted) remained susceptible to tetracycline and the other antimicrobial agents tested, consistent with the known marRAB locus resistance properties (Alekshun & Levy, Reference Alekshun and Levy1999).
The observation that the antimicrobial agent susceptibilities were unchanged in the Esch. coli strain that lacks MarR and which constitutively expresses MarA, a multiple antibiotic resistance activator (Hachler et al. Reference Hachler, Cohen and Levy1991; Barbosa & Levy, Reference Barbosa and Levy2000a), confirms the role of the marRAB locus in conferring reduced susceptibilities to antimicrobial agents by milk in GC468 cells, which possess an intact marRAB locus. The component within milk that modulates the marRAB locus remains to be identified. Although chromosomal elements are directly involved in the reduced antimicrobial susceptibilities observed here, we are nonetheless not able to definitively rule out the possibility that extra-chromosomal elements (i.e., plasmids, etc.) may be involved in naturally-derived environmental or agriculture bacterial isolates (Burgos et al. Reference Burgos, Ellington and Varela2005). Future studies are necessary to distinguish between these alternatives. Along these lines, we are presently examining the antimicrobial agent susceptibility profiles for previously published Esch. coli isolates from dairy soil (Burgos et al. Reference Burgos, Ellington and Varela2005) and from dairy water for a new panel of antimicrobial agents.
Growth in LB medium and milk
The growth of Esch. coli strain AG112 cells reached higher growth plateaus when compared with those of GC4468 and JHC1096 strains in both LB broth and in milk (see Figs 1 & 2). All three strains, however, reach higher growth plateaux in LB broth than in milk. Our investigation of the relationship of Esch. coli growth indicates that the total growth yield in LB broth is better than that in liquid milk while the antimicrobial agent susceptibilities are reduced, which is consistent with the contention that milk itself, and not growth in milk, is the factor conferring the reduced susceptibilities that we observe here.
In summary, pasteurized milk or milk agar media confers reduced susceptibilities to multiple antimicrobial agents in Esch. coli. Our studies provide evidence to support the contention that the mode of acquisition of reduced susceptibilities to antimicrobial agents by Esch. coli in pasteurized milk is due to modulation of the marRAB locus. We examined Esch. coli, which is a model bacterium and member of the Enterobacteriaceae family of bacteria, because of its relevance to physiology, its biomedical role as a causative agent of infectious disease and its antimicrobial agent resistance properties, which in turn compromise therapy (Sanders & Sanders, Reference Sanders and Sanders Jr1992; Baquero et al. Reference Baquero, Negri, Morosini and Blazquez1998; Alekshun & Levy, Reference Alekshun and Levy2006). Esch. coli microorganisms have profound significance as a tool for molecular biology, microbial genetics, and as indicators of environmental sewage contamination, plus pathogenic Esch. coli variants have considerable clinical importance (Neidhardt, Reference Neidhardt1996). Because of their demonstrated transferability of antimicrobial agent resistance determinants among bacteria of distinct species (Neu, Reference Neu1984; Nord, Reference Nord1993; Leverstein-van Hall et al. Reference Leverstein-van Hall, Blok, Donders, Paauw, Fluit and Verhoef2003), human consumption of raw milk contaminated with multidrug resistant microorganisms may contribute to the dissemination of multidrug resistant bacteria in agriculture and could be a source of potentially untreatable zoonotic infectious disease (Tollefson et al. Reference Tollefson, Fedorka-Cray and Angulo1999; McEwen & Fedorka-Cray, Reference McEwen and Fedorka-Cray2002; Angulo et al. Reference Angulo, Baker, Olsen, Anderson and Barrett2004). Lastly, because the marRAB locus is present in other members of the Enterobacteriaceae family of bacteria, such as Enterobacter, Klebsiella, Shigella, and Salmonella (Cohen et al. Reference Cohen, Yan and Levy1993), the findings presented here have relevance to these medically important microorganisms and their relationships with milk.
This publication was made possible by National Institutes of Health grants 1 R15 GM070562-01 and P20 RR016480, the latter of which is from the NM-INBRE program of the National Center for Research Resources, an internal research grant awarded by Eastern New Mexico University, and a contribution from Calton Research Associates in honor of George and Clytie Calton. E.F.C. was supported by NIH grants RR-16480, R25-GM56626 Bridges fellowship and by NSF grants NM-AGEP and HRD-0086701. R.R.C. was supported by an American Society for Microbiology Undergraduate Research Fellowship. We thank Dr. Sanath Kumar, Jamey F. Licata and Johnny Sena for expert technical assistance. We thank Dr. Jeffrey K. Griffith (University of New Mexico, Albuquerque, NM) and Dr. Tomofusa Tsuchiya (Okayama University, Okayama, Japan) for critical reading of the manuscript before publication.
