In food industry Escherichia coli strains are common contaminants and indicator of poor hygiene and sanitary conditions. Beyond sanitary issues, Esch. coli serotype such as O157:H7 and others are a leading cause of outbreaks associated with foods including dairy products. The journey of raw material such as milk, from farm to product processing facility provides multiple portals for the organism to enter the supply chain. At a farm level, it is a common practice to use fluoroquinolones such as ciprofloxacin (Cip) to contain microbial risks. However, their usage at sub-judicious concentrations may adapt the organism even to higher concentrations of antibiotics (van der Horst et al. Reference van der Horst, Schuurmans, Smid, Koenders and Ter Kuile2011). In this regard, a previous report has documented prevalence of Cip resistance in 21·5% of food origin Esch. coli in Republic of Ireland, which was the highest in Europe (url: http://www.safefood.eu/en/Publication/Research-reports/The-problem-of-Antimicrobial-Resistance-in-the-food-chain/ accessed on 27 July 2011). The next level in a food chain where food microflora is subjected to antimicrobial compounds is clean-in-place protocols in a food processing facility. A routine sanitary regime may involve application of biocides such as benzalkonium chloride (BC) and sodium hypochlorite to disinfect processing equipment and surfaces (Langsrud et al. Reference Langsrud, Sundheim and Holck2004). However, various circumstances such as carry over contamination of water, improper rinse and protective action of food might reduce their effective concentrations. These sub-lethal concentrations have been reported to induce adaptations against related and unrelated antimicrobial compounds in the organism (van der Horst et al. Reference van der Horst, Schuurmans, Smid, Koenders and Ter Kuile2011). In this context an earlier study has shown that BC adapted strains of Pseudomonas aeruginosa of dairy origin could develop cross resistance against Cip (Pagedar et al. Reference Pagedar, Singh and Batish2011).
A previous study has shown strong biofilm forming clinical isolates of Esch. coli to exhibit reduced susceptibility to Cip (Naves et al. Reference Naves, del Prado, Huelves, Gracia, Ruiz, Blanco, Dahbi, Blanco, del Carmen Ponte and Soriano2008). Similarly, the biofilms of Esch. coli have also been found to be resistant to biocides such as BC and other disinfectants (Ntsama-Essomba et al. Reference Ntsama-Essomba, Bouttier, Ramaldes, Dubois-Brissonnet and Fourniat1997). Overall, it is a well established fact that the metabolic changes and protective effect of biofilm matrix create a conducive environment for biofilm inhabitants to adapt to higher concentrations of antimicrobial compounds (Brown & Gilbert, Reference Brown and Gilbert1993). However, there is hardly any report discussing the phenomenon in reverse order i.e. the effect of antimicrobial adaptation on biofilm formation potential of the organism. Such a study may provide useful information from a food safety as well as a clinical point of view if such adaptations confer any survival advantage to microorganisms.
In Esch. coli, resistance-nodulation division-type efflux pumps [AcrB, AcrF, MdtB (YegN), MdtC (YegO), YhiV (MdtF)] are widely distributed and play a significant role in both intrinsic and acquired resistances against compounds such as detergents, ethidium bromide (EtBr) and antibiotics (Bohnert et al. Reference Bohnert, Schuster, Fahnrich, Trittler and Kern2007). In Esch. coli efflux mediated fluoroquinolone and BC resistance has already been reported (Bore et al. Reference Bore, Hébraud, Chafsey, Chambon, Skjaeret, Moen, Møretrø, Langsrud, Rudi and Langsrud2007; van der Horst et al. Reference van der Horst, Schuurmans, Smid, Koenders and Ter Kuile2011). Apart from antimicrobial resistance, EPA has also been held responsible for enhancing the virulence of enteroaggregative Esch. coli by promoting aggregation (Imuta et al. Reference Imuta, Nishi, Tokuda, Fujiyama, Manago, Iwashita, Sarantuya and Kawano2008). In our previous study, we established the role of EPA in imparting adaptive and cross resistance against BC and Cip to food origin Ps. aeruginosa (Pagedar et al. Reference Pagedar, Singh and Batish2011). However, in reference to food origin Esch. coli we could not come across any study on these lines.
In Esch. coli, haemolysin is one of the major virulence factors along with shiga-like toxin, attaching and effacing lesions (Skals et al. Reference Skals, Jorgensen, Leipziger and Praetorius2009). Alpha haemolysin releasing Esch. coli have been reported to colonise and cause urinary tract infection (Rijavec et al. Reference Rijavec, Muller-Premru, Zakotnik and Zgur-Bertok2008). A similar report in context to prostitis causing Esch. coli strains established a close association (P = 0·03) between in vitro BFP and haemolysin activity (Soto et al. Reference Soto, Smithson, Martinez, Horcajada, Mensa and Vila2007). The underlying basis for the association of susceptibility to antibiotics and virulence is still unclear. There are contrary reports regarding acquisition of virulence before (Johnson et al. Reference Johnson, Kuskowski, Gajewski, Soto, Horcajada, Jimenez de Anta and Vila2005) or after development or acquisition of resistances (Horcajada et al. Reference Horcajada, Soto, Gajewski, Smithson, Jimenez de Anta, Mensa, Vila and Johnson2005; Piatti et al. Reference Piatti, Mannini, Balistreri and Schito2008). In the present study, the affect of adaptation to BC and Cip was studied on haemolysin activity of Esch. coli strains of dairy background.
As evident from the literature review, generic Esch. coli isolates of food origin have hardly been explored in reference to adaption to antimicrobial agents and subsequent effect on BFP, EPA, and haemolytic activity. Such a study may provide useful information on the role of food microflora in emergence and spread of antimicrobial resistance. Moreover, food processing conditions and parameters might modulate the organism to adapt some specific phenotypes that might not be encountered in a clinical setup. Hence, the present investigation was undertaken to investigate the effect of adaptation to Cip and BC on BFP, EPA and haemolysin activity and correlation if any, in Esch. coli isolates of dairy origin.
Materials and Methods
Test Organisms, growth conditions and biofilm formation assay
Biofilm samples (slime like substance) were collected after clean–in-place regime from raw milk line (E1), pasteurizer inlet (E2), outlet (E3) and, milk tank (E4) of a small scale and commercial scale dairy handling 50 000 and 1 million litres of milk per day, respectively. The clean-in-place protocol that was being followed in both the dairies involved routine application of NaOH (1%) at 75 °C for 5 min, warm water rinse, HNO3 (0·5%) at 60 °C for 5 min and, water rinse. The weekly cleaning protocol involved use of NaOCl (200 ppm) at 60 °C for 5 min and warm water rinse. The sample swabs from each isolation point were immersed immediately in TSB and transported to the laboratory in an ice box for further analysis. Presumptively, all the isolates which, (a) fermented lactose with gas production within 48 h at 35 °C, (b) appeared as Gram-negative non-sporeforming rods and (c) showed IMViC patterns of + + −− were considered to be Esch. coli strains. These isolates were further confirmed by streaking onto eosine methylene blue agar and only colonies with typical sheen were retained. A total of 81 Esch. coli isolates was obtained and these were designated on the basis of their isolation source (I/II)-sampling point (E-1/2/3/4) and number during the process of isolation. All the isolates (n = 81) were subjected to biofilm formation assay as previously described and categorised as ‘strong’, ‘moderate’, ‘weak’ or ‘no biofilm’ formers (Pagedar et al. Reference Pagedar, Singh and Batish2010). Prior to each experiment, desired cell counts were obtained by adjusting absorbance (620 nm) to 0·12 (approx. 108 cfu/ml) and subsequent dilution and plating onto plate count agar. All the assays were carried out in triplicate and means were considered for subsequent analysis. Reagents and media were procured from HiMedia Labs, Mumbai, India unless specified otherwise.
Determination of minimum inhibitory concentrations (MIC)
From each isolation point of two sources, two each of the strongest and the weakest biofilm formers were selected for further antimicrobial resistance studies (n = 32) [(2 + 2)4 × 2=32] against Cip and BC. Susceptibility to Cip was determined in terms of MIC by macrobroth dilution method. The stock solution of antibiotic (10 000 mg/l) was prepared in sterile water, filter sterilized and aseptically diluted in tubes containing 3 ml Mueller Hinton broth to obtain a suitable dilution. The respective culture counts for the experiment (105 cfu/ml) were obtained as discussed in previous section. The Mueller Hinton broth tubes with varying antibiotic concentrations were inoculated with 105 cfu/ml of respective cultures and incubated at 37 °C for 24 h. MIC was recorded as the lowest concentration of antibiotic at which OD-620 nm of the test was equivalent to that of the uninoculated control tube. To determine the resistance against BC, inoculum was prepared as described above and added at a final concentration of 105 cfu/ml to Mueller Hinton broth containing 0–500 ppm of BC (in steps of 50). The biocide exposure was for 15 min at room temperature and then samples were neutralised by 10 times dilution in neutralising solution [0·3% lecithin, 3% Tween 80, 0·5% sodium thiosulphate, 0·1% l-histidine, 0·034% potassium dihydrogen phosphate] (Van der veen & Abee, Reference Van der Veen and Abee2010). The tubes were subsequently incubated at 37 °C for 24 h and MIC was determined against uninoculated control tube. The MIC of antibiotic Cip was tested in geometric progression from 0·25 to 1024 μg/ml for parent strains of AR category and in steps of 0·25 μg/ml for parent strains of ANR category. For adapted strains of both the categories (AR & ANR), MIC of Cip was determined in steps of 5 μg/ml. The MICs were median value of experiments performed three times in triplicate, which were repeated until a median value was obtained. On the basis of their respective MIC of Cip and BC these isolates were arbitrarily categorised into antibiotic/biocide resistant (AR/BR), and non-resistant (ANR/BNR). The AR category comprised the isolates exhibiting MIC ⩾ 4, while conventionally sensitive and intermediate isolates (having MIC ⩽ 2) were grouped together in ANR category (CLSI, 2011). For BC, the isolates exhibiting MIC ⩾ 200 ppm were characterised as resistant while other as non-resistant (Velázquez et al. Reference Velázquez, Barbini, Escudero, Estrada and de Guzmán2009). Throughout the study, Esch. coli MTCC 739110 (ATCC 10536) was used as a reference strain. Four each of AR, BR, ANR and BNR, total 16 out of 32 isolates, were randomly selected for further investigation as box and whisker plots [XL-statistics v4.5 (http://www.man.deakin.edu.au/rodneyc/XLStats.htm)] indicated no significant effect of sources and isolation point on antimicrobial resistance pattern.
Development and stability of the acquired adaptive resistance
Adaptation of AR and ANR isolates was initiated by growing them in nutrient broth containing Cip at respective sub-MICs level (first well showing growth below MIC) and then increasing concentrations of Cip in steps of 5 mg/l. Similarly, BR and BNR isolates were adapted to BC in steps of 10 ppm. The procedure was repeated every 24 h until the cultures adapted to higher concentrations of respective antimicrobials. The stability of adapted resistance was evaluated by MIC determination up to 30 d by daily subculturing in nutrient broth devoid of Cip and BC.
Estimation of Efflux pump activity and haemolysin activity at pre and post adapted stages
For EPA evaluation of the selected isolates (n = 16) at pre and post-adaptation level, EtBr was used as model systems as per method described previously (Pagedar et al. Reference Pagedar, Singh and Batish2011). Briefly, preselected isolates were grown overnight at 37 °C on tryptic soy agar (TSA) plates containing 0·5 μg/ml of EtBr (Banglore Genei, India). On irradiation of these plates on an ultra violet transilluminator, cells that accumulated EtBr fluoresced pink (negative for EPA), whereas cells with partial accumulation appeared pale pink. Cells that did not accumulate EtBr were white and considered positive for EPA. Prior to EPA quantitation, EtBr sub-MICs were determined by appropriately diluting the overnight grown isolates (108 cfu/ml) of AR, ANR, BR, & BNR categories and spread plating onto TSA (37 °C/48 h) containing gradient of EtBr (0–100 μg/ml in steps of 10 μg/ml). The lowest concentration of EtBr in TSA plate with no colonies was considered as EtBr MIC. To quantify EPA, overnight grown cultures (108 cfu/ml) were inoculated into TSB with respective sub-MIC (MIC-1) EtBr concentrations (μg/ml) at room temperature for 0·5 h. Subsequently, cells were pelleted, washed, excited at 530 nm and fluorescence emission was observed at 600 nm (Raherison et al. Reference Raherison, Gonzalez, Renaudin, Charron, Bébéar and Bébéar2002). Accumulation of EtBr was calculated using standard curve equation after subtraction of natural fluorescence of the cells.
For pre and post-adaption estimation of haemolysin activity, supernatant of preselected overnight grown cultures (108 cfu/ml) were diluted in 1:1 ratio using sterile saline and an aliquot of 100 μl was mixed with 100 μl of freshly prepared defibrinated, 3% citrated red blood cell solution in 96 well microtitre plate. The plates were incubated at 37 °C/1 h and respective absorbances were measured at 540 nm using microplate reader (ECIL, Microscan, India). The haemolysin activity (μg of haemolysin/ml) was calculated in reference to control using the standard curve equation obtained by pure haemolysin (H9395, Sigma Aldrich, Missouri, USA).
Statistical analysis
The statistical analysis (Two way ANOVA, box and whisker plots, median, arithmetic mean and sd) was performed using XL-statistics v4.5 (http://www.man.deakin.edu.au/rodneyc/XLStats.htm) ‘Significance’ expressed at the 5% level (P < 0·05) or mentioned otherwise.
Results
Biofilm formation and MIC against Ciprofloxacin and Benzalkonium Chloride
Evaluation of the selected isolates for their BFP revealed that the originally resistant (AR and BR) phenotypes were strong biofilm formers. Post-adaptation change in BFP of the resistant category was not significant; hence it was not considered for statistical analysis in terms of correlation with other traits evaluated (MIC, EPA, haemolysin and stability). On the contrary, as a result of adaptation, non-resistant categories (ANR and BNR) exhibited a significant increase in BFP (P ⩽ 0·1), which led to upgrading their categorisation status from ‘weak/moderate’ to ‘moderate/strong’ biofilm former (Table 1a & b).
Table 1. Biofilm formation, adaptive resistance, resistance stability, efflux pump and haemolytic activity of ciprofloxacin resistant (AR) and non-resistant (ANR) (1a), and benzalkonium chloride resistant (BR) and non-resistant (BNR) (1b) isolates at pre- and post-adaptive stages
† Indicates the median value of three experiments
The MIC of antibiotic Cip was tested in geometric progression from 0·25 to 1024 μg/ml for parent strains of AR category and in steps of 0·25 μg/ml for parent strains of ANR category (due to very high susceptibility of this group to Cip). For adapted strains MIC of Cip was determined in steps of 5 μg/ml for both categories
‡ EPA was determined in steps of 10 μg/ml of EtBr
The MIC of biocide BC was evaluated in steps of 50 ppm for parent strains (both categories) and in steps of 10 ppm for adapted strains (both categories)
Development and stability of the acquired adaptive resistance
In the current investigation, irrespective of initial susceptibility of Esch. coli isolates, slow growth was observed during initial passages while developing adaptive resistance to both BC and Cip. In addition to a general trend, strain specific behaviours were also observed. Two isolates I-E3-8 and II-E4-82 exhibited the same pre-adaptive MIC to Cip (8 mg/l) and BC (250 ppm), however post-adaptation resistance to Cip was greater in I-E3-8 (50 mg/lin comparison with 25 in II-E4-82) while that against BC was greater in II-E4-82 (340 ppm in comparison with 320 ppm for I-E3-8). For all the adapted isolates, the stability of developed adaptive resistances was analysed by passage in antibiotic- or biocide-free medium. From Table 1a & b, it is evident that, the extent of developed adaptive resistance was greater in ANR followed by AR, BNR and BR isolates. Regarding the stability of the adaptive resistance, the strain dependent variation resulted in wide variation and thus, no trend in the stability of adaptive resistance could be established for any particular category.
Effect of adaptive resistance on EPA and haemolysin activity
As a result of adaptation to antimicrobials, highly significant increase in EPA could be observed for the non-resistant isolates (ANR and BNR) (P = 0·005) and BR category (P < 0·05). Only AR category exhibited an insignificant post-adaptive increase in EPA (P = 0·36) which is indicative of some specific mechanism responsible for conferring antibiotic resistance rather than a broad spectrum mode as in biocide resistance. The above mentioned observation was reinforced when EPA and adaptive resistances of the isolates were compared. During adaptation to Cip, the moderate percentage increase in EPA (range of 33–67% in ANR; 16–50% in AR) and yet, the remarkable adaptive resistance (range of 900–4900% in ANR; 212–525% in AR) again indicates that EPA certainly could not account as sole mechanism for the adaptive resistance. Interestingly, adaptation to BC presented altogether a different scenario. The BR and BNR isolates exhibited almost similar percentage increase in EPA (range of 50–85% in BR; 57–80% in BNR) and a marginal difference in developed adaptive resistance (28–35% in BR; 40–160% in BNR). Overall in resistant categories, post adaptive increase in EPA of AR was significantly lower than that of BR (P < 0·05), still AR developed greater adaptive resistance than BR (P < 0·02).
The percentage increase in EPA was compared statistically with the percentage increase in adaptive resistance of ANR and BR only (P < 0·05). In reference to haemolysin activity, strain specific variation was observed and therefore this trait was not considered for statistical analysis.
Discussion
Emergence of antimicrobial resistance is a global problem and findings of the current study suggest a role for the food chain in it. Strong BFP of all the parent isolates of resistant phenotypes (AR & BR) indicates that there may be a direct relationship between BFP and antimicrobial resistance. These findings are in line with previous reports which have established that biofilm formation leads to increased tolerance to antimicrobial agents (Drenkard & Ausubel Reference Drenkard and Ausubel2002; Pagedar et al. Reference Pagedar, Singh and Batish2011). Post-adaptation, only a marginal increase in the BFP status of resistant phenotypes (AR & BR) was observed. However, due to paucity of the published literature on adaptation of resistant phenotypes, we could not discuss these findings. Our findings regarding substantially improved BFP of non-resistant category after adaptation can be corroborated with similar observations in earlier studies with coagulase positive staphylococci (Qu et al. Reference Qu, Daley, Istivan, Garland and Deighton2010) and Ps. aeruginosa (Drenkard & Ausubel Reference Drenkard and Ausubel2002) of clinical origin.
Apparently, exposure to antimicrobial agents induces stress response in microorganisms. In Esch. coli a master regulator RpoS is known to regulate stress responses (Hengge, Reference Hengge2009). It has been reported to up-regulate the expression of csgD operon responsible for curli production, which is an important factor in biofilm formation and surface adhesion (Gualdi et al. Reference Gualdi, Tagliabue and Landini2007). In the present study, substantially enhanced biofilm production by non-resistant isolates (post-adaptation), may be due to the over expression of curli gene through RpoS regulation, in response to antimicrobial stress. This speculation seems to be feasible as originally resistant isolates (AR & BR) probably experienced lesser stress during adaptation, and hence displayed only marginal increase in biofilm formation.
These findings imply that in a food industry, non judicious usage of antimicrobial compounds may facilitate transformation of non-resistant isolates to resistant ones, thus imposing a food safety threat. Our assumptions draw support from a recent study which concluded that biofilms of BC adapted Esch. coli become thicker on re-exposure to the biocide (Machado et al. Reference Machado, Lopes, Sousa and Pereira2012). Contrary to our observations, the effect of BC adaptation on BFP of Listeria monocytogenes has been reported to be insignificant (Romanova et al. Reference Romanova, Gawande, Brovko and Griffiths2007).
Post-adaptation, a remarkable resistance developed in all the isolates irrespective of initial resistant phenotypes. Various mechanisms (specific and non-specific) are involved in conferring antimicrobial resistances to microorganisms. Intrinsic resistance against fluoroquinolone like Cip is due to several specific mechanisms (Arias et al. Reference Arias, Seral, Gude and Castillo2010), whereas only few non-specific mechanisms like modification of membrane composition, and EPA are responsible for conferring biocide (BC) resistance (Kendall & Sperandio, Reference Kendall and Sperandio2007). Interestingly, our findings depict that although the post-adaptive increments in EPA of BR isolates were larger than those of AR, the AR exhibited greater adaptive resistance against Cip than BR against BC. Probably simultaneous action of multiple mechanisms along with EPA contributed towards notable adaptive resistance against Cip in AR isolates. On the other hand, intrinsic EPA seems to be the only predominant mechanism conferring resistance against BC in BR isolates. We tried to understand this unexpected behaviour of BR isolates during adaptation, on the basis of findings of a previous study by Ma et al. (Reference Ma, Alberti, Lynch, Nikaido and Hearst1996). They observed that during adaptation and/or general stress conditions, efflux pump encoded by acrAB was up-regulated by only four times whereas its repressor (acrR) by 10 times in Esch. coli. In such a scenario it seems possible that even an increase in EPA may not be reflected in phenotypic resistance pattern as observed in the current study. In this study, a significant increase in EPA observed for non-resistant phenotypes (ANR & BNR), implies a pivotal role of EPA in conferring adaptive resistance to the isolates of this category. Moreover, beyond adaptive resistance, EPA has also been shown to confer cross resistance against Cip and BC in Ps. aeruginosa of dairy origin (Pagedar et al. Reference Pagedar, Singh and Batish2011).
Previous studies have shown that during adaptation, the expression of acrAB was up-regulated but only in the initial phase of adaptation whereas the expression of stress response genes continued to increase throughout the adaptation and render the strains resistant against oxidising agents (Bohnert et al. Reference Bohnert, Schuster, Fahnrich, Trittler and Kern2007; Bore et al. Reference Bore, Hébraud, Chafsey, Chambon, Skjaeret, Moen, Møretrø, Langsrud, Rudi and Langsrud2007). Such findings further imply that the residuals or sub-lethal concentrations of antimicrobials (e.g. BC and Cip) in industries would aid the microorganisms to develop resistance against not only the strong detergents like BC used in dairy plant cleaning regime but routinely used oxidising agents also. Further, when such adaptively biocide resistant strains are implicated in foodborne illness, the therapeutic treatment becomes difficult as BC adapted strains have previously been reported to develop cross resistance to fluoroquinolones and other clinically relevant antibiotic (Langsrud et al. Reference Langsrud, Sundheim and Holck2004; Bore et al. Reference Bore, Hébraud, Chafsey, Chambon, Skjaeret, Moen, Møretrø, Langsrud, Rudi and Langsrud2007). Our findings regarding presence of antibiotic resistant isolates signifies that antimicrobial resistance mechanisms adopted by microorganisms in response to stresses imposed by food chain conditions seems to be conferring resistance against clinically relevant antimicrobial compounds. Surprisingly, we could not come across any report wherein EPA and antimicrobial resistance pattern of dairy or food origin Esch. coli stains were studied. However, any such correlation between BFP and antimicrobial resistance was ruled out in clinical origin Esch. coli (Marhova et al. Reference Marhova, Kostadinova and Stoitsova2010). Strain specific behaviour and lack of a correlation with other factors studied signifies that adaptation has a limited and/or strain specific effect on the haemolysin production. A number other factors such as growth conditions and genetic makeup of the organism has been shown to play important role in determining and regulation of haemolysin production (Tiwari et al. Reference Tiwari, Deol, Rishi and Grewal2002).
In conclusion, the present study showed development of higher and more stable adaptive resistance against BC and Cip by Esch. coli isolates exposed to sub-lethal concentrations of antimicrobial agents. Some adaptive strains also exhibited enhanced BFP, EPA and virulence potential (haemolysin activity). The findings emphasize that antimicrobials should be used at specified doses in appropriate manner to avoid emergence of resistance which may subsequently lead to increased virulence. A significant threat is imposed when such virulence factors or resistances are horizontally transferred to high risk pathogens.
Ankita Pagedar duly acknowledges the Director, National Dairy Research Institute, Karnal, India, for the lab facilities and The Indian Council of Agricultural Research, India for the financial assistance in the form of senior research fellowship. The authors are thankful to Mr Snehal Ingale, Asst. Prof., ARIBAS, New V.V. Nagar, Gujarat, India, for his expert advice regarding statistical analysis.
