Variation in detection limits between bacterial growth phases and precision of an ATP bioluminescence system

Letters in Applied Microbiology ISSN 266-8254 ORIGINAL ARTICLE Variation in detection limits between bacterial growth phases and precision of an ATP bioluminescence system S.J. Vogel 1,3, M. Tank 2,4 and N. Goodyear 1 1 Department of Clinical Laboratory and Nutritional Sciences, University of Massachusetts Lowell, Lowell, MA, USA 2 Department of Biological Sciences, University of Massachusetts Lowell, Lowell, MA, USA 3 Present address: Rapid Micro Biosystems Inc., One Oak Park Drive, Bedford, MA 173, USA 4 Present address: Shire Human Genetic Therapies, 2 Shire Way, Lexington, MA 2421, USA Significance and Impact of the Study Surface hygiene is a critical component of food safety and infection control; increasingly, ATP detection by bioluminescence is used to evaluate surface hygiene and effective cleaning. This is the first study to show that the number of living and potentially infectious bacteria remaining when the device reads zero varies between the different bacterial life cycle phases: lag, log, stationary and death. ATP device users need to be aware of this information to use the devices appropriately. Keywords ATP bioluminescence, bacterial growth curve, limits of detection, precision, surface hygiene. Correspondence Nancy Goodyear, Department of Clinical Laboratory and Nutritional Sciences, 3 Solomont Way, Suite 4, University of Massachusetts Lowell, Lowell, MA 1854, USA. E-mail: Nancy_Goodyear@uml.edu 213/24: received 2 October 213, revised 6 November 213 and accepted 22 November 213 doi:1.1111/lam.12199 Abstract To determine the detection limits of the SystemSure Plus, Escherichia coli and Staphylococcus aureus growth curve samples were taken in lag (1 h), log (6 h), stationary (12 h) and death phases (E. coli 144 h, Staph. aureus 72 h). At each time point, the log 1 CFU ml 1 was determined for the dilution where the SystemSure read relative light units (RLU). Average detection limits were E. coli: lag 627, log 588, stationary 745 and death 688; Staph. aureus: lag 437, log 515, stationary 788 and death 757. Between-run precision was determined with positive control; within-run precision with positive control, lag and log growth for each bacteria. Within-run precision mean RLU (CV): positive control 274 (12%), E. coli lag 1 (63%), log 2173 RLU (19%), Staph. aureus lag 2 (58%) and log 5535 (18%). Between-run precision was 232 (16%). The precision is adequate with most values within the 95% confidence interval. The detection limit varied by 351 log 1 for Staph. aureus and 147 log 1 for E. coli. The lowest detection limits were during E. coli log and Staph. aureus lag phases; the highest was during stationary phase. These results suggest that organism identification and growth phase both impact ATP RLU readings. Introduction Proper surface sanitation and hygiene is a critical component of infection control and food safety. In addition to removing pathogenic micro-organisms, organic matter such as body fluids or food residue must also be removed as they can provide a source of nutrients for microorganisms. Traditionally, culture-based methods have been used to determine bacterial contamination of a surface. Classical methods for detection and/or enumeration of bacteria include contact agar plates, flexible agar films and various swabbing methods. Culture-based methods have the advantage of providing organism identification and are known to be effective; however, they take time, expertise and laboratory facilities. Immunological and nucleic acid-based methods are also available. Drawbacks of these methods include specificity concerns with immunological methods, the cost of materials and instrumentation, and for molecular methods, interferences from food substances and the inability to differentiate viable from dead cells (Jasson et al. 21). Methods that detect ATP by bioluminescence have become increasingly popular as rapid, easy to use alternatives to culture-based methods (Whitehead et al. 28; Guh et al. 21; Jasson et al. 37 Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology

S.J. Vogel et al. ATP detection limits 21; Cunningham et al. 211; Sciortino and Giles 212; Alfa et al. 213a,b). These methods make use of the firefly luciferase enzyme to produce measurable light emissions, which are interpreted as relative light units (RLU). ATP bioluminescence methods provide real-time results in seconds to minutes and require fewer resources than culture, immunological or molecular methods. They lack the specificity of plate counting as ATP is produced by all living cells, including bacterial cells and other organic cells such as those from fruits and vegetables (Fujikawa and Morozumi 22; Guh et al. 21). As ATP is present in many types of cells, the terms bio-load or organic soil may be used to describe the source of ATP. The interpretation of organic material present rather than strictly microorganisms is advantageous in some settings but may be problematic in others (Malik et al. 23; Willis et al. 27; Whitehead et al. 28; Guh et al. 21; Jasson et al. 21; Turner et al. 21; Cunningham et al. 211; Sciortino and Giles 212; Luick et al. 213). The presence of nonmicrobial organic soil may indicate inadequate cleaning and the presence of essential nutrients for microorganism growth (Whitehead et al. 28; Guh et al. 21; Shama and Malik 213). Limitations of monitoring surface hygiene by detection of ATP by bioluminescence include possible interference of chemical residues left behind from cleaning, the lack of direct correlation between RLU and the quantity of microorganisms present, and concerns about detection limits and interpretation of RLU readings (Willis et al. 27; Aiken et al. 211; Sciortino and Giles 212; Luick et al. 213; Shama and Malik 213). Manufacturer-defined detection limits are generally based on detection of purified ATP rather than living bacteria or other organic material. Reported limits of detection in living bacteria generally range from 1 to 1 5 colony-forming units (CFU), depending on the organism tested (Fujikawa and Morozumi 22; Willis et al. 27; Turner et al. 21; Aiken et al. 211; Alfa et al. 213a; Fushimi et al. 213; ). Some manufacturers of ATP bioluminescence systems provide recommendations for interpretation of RLU, while others do not, preferring users to determine their own limits of acceptability. Several interpretative guidelines and cut-offs have been proposed for applications including hospital surfaces, flexible gastric endoscopes and surfaces in schools (Griffith et al. 2; Lewis et al. 28; Alfa et al. 213a; ISSA 213). The proposed limits for an interpretation of clean or acceptable range from one to more than 1 RLU depending on which device is used and specific application. For example, using the 3M Clean-Trace luminometer (3M, St. Paul, MN) for hospital surfaces, Griffith et al. (2) proposed that readings <5 RLU would be considered clean, whereas Lewis et al. (28) proposed <25 RLU as the cut-off. Using the same device for monitoring the cleaning of flexible gastric endoscopes, Alfa et al. (213a) proposed <2 RLU as the cut-off. A 2 July 213 draft K 12 School cleaning standard from ISSA The Worldwide Cleaning Industry Association (213) proposed widely varying cutoffs for three different ATP devices. For example, highly effective cleaning for cafeteria tables was proposed to be 4951 RLU using the NovaLUM (Charm Sciences, Inc, Lawrence, MA), 141 for the Uni-Lite NG (3M), and 9 for the Hygiena SystemSure Plus. While such a great difference may accurately reflect differences in the systems, the lack of consistency as to what constitutes an acceptable reading may result in confusion and the potential for misinterpretation or inappropriate use of the devices. An additional consideration with ATP-based detection of bacteria is the fact that bacterial cells do not contain a consistent quantity of ATP. The concentration of ATP per cell varies between bacterial species, from cell to cell within the same species, and at different points in the bacterial life cycle. The life cycle, or growth curve, consists of lag, log, stationary and death phases (Fujikawa and Morozumi 22, 23, 25; Shama and Malik 213). The length of each phase is dependent on the specific organism and the medium used for growth. In addition, bacteria found in foods or on surfaces may be stressed to the point of sublethal injury (Wu 28; Ukuku et al. 29; Shama and Malik 213). The ATP content of each bacterial cell may be quite low as the bacteria are not growing and dividing actively in this stressed state, thereby decreasing the sensitivity of the method. The purpose of this study was to (i) evaluate withinrun (intra-assay) precision and between-run (interassay) precision of one ATP bioluminescence system (System- Sure Plus, Hygiena, Camarillo, CA) and (ii) determine the detection limits of the device at each phase of the bacterial growth curve for one representative Gram-negative (Escherichia coli) and one representative Gram-positive (Staphylococcus aureus) bacteria. Stakeholders are in need of this information to utilize these tools effectively and appropriately. Results and discussion Within-run and between-run precision The means and coefficients of variation (CVs) for withinrun and between-run precision are shown in Table 1. Most values for the between-run and within-run precision of the SystemSure fell within the 95% confidence level. The CVs varied between the growth phases; this variation can be accounted for by the ATP concentration differences along the curve. The CVs for lag phase bacteria (63% for E. coli and 58% for Staph. aureus), when ATP concentration is lowest, were above 2%, generally Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology 371

ATP detection limits S.J. Vogel et al. Table 1 Means and coefficients of variation (CVs) for within-run (positive control only) and between-run precision Sample defined as unacceptable; however, at such low levels a variation of just one or two RLU will drive the CV above 2%. All readings for lag phase bacteria were below five RLU, which would be interpreted as clean, by most users. In log phase, when the ATP concentration was much higher (average of 2173 RLU and 5535 RLU for E. coli and Staph. aureus, respectively), our CVs were less than 2%. Readings greater than 1 RLU would be interpreted as dirty by most users. Growth curves Mean RLU Averages of CFU ml 1 and ATP RLU for three replicate growth curves for E. coli and Staph. aureus are shown in Figs 1 and 2, respectively. The growth curves and ATP concentrations for both Staph. aureus and E. coli followed typical patterns although there are differences between the two organisms. During lag phase ( 4 h), bacteria are introduced to a fresh source of nutrients and they begin to increase in size and synthesize proteins and nucleic acids necessary for cell division. ATP concentration begins low but increases more rapidly than the CFU per ml as the cells prepare for cell division. The next phase is log phase (6 8 h for Staph. aureus and 6 1 h for E. coli), during which bacterial cells are dividing at an exponential rate. ATP concentration continues to be high to support rapid cell division, SD Number > 2 SD CV (%) Within-run 232 36 16 Between-run 274 34 1 12 Escherichia coli lag phase 1 82 1 63 Staphylococcus aureus 2 116 1 58 lag phase E. coli log phase 2173 44 19 Staph. aureus log phase 5535 97 3 18 but at lower concentrations than during the late stages of lag phase. As nutrients are consumed and metabolic waste products build up, the cells enter stationary phase (1 12 h for Staph. aureus and 8 24 h for E. coli). As cell division slows, ATP concentration drops steadily but more rapidly than cell numbers. In the final phase, death phase, nutrients are depleted and cell death occurs. ATP concentration levels out as the number of viable cells falls (Fujikawa and Morozumi 22, 23, 25). In our growth curves, Staph. aureus grew to a higher CFU ml 1, but entered stationary and death phases sooner than E. coli. ATP readings dropped sooner in the log phase for Staph. aureus as well, with the peak at 4 h in contrast to the E. coli, where the peak ATP concentration was at 6 h. Limits of detection The detection limits for each growth phase are shown in Table 2. In our growth curves, the SystemSure had the highest limit of detection during stationary phase (745 log 1 CFU ml 1 for E. coli and 788 for Staph. aureus) when there are still many viable cells, but the ATP concentration has fallen significantly. In each growth phase, four seven logs of viable organisms were present in the dilution that gave a reading of RLU. The greatest difference between growth phases was seen with Staph. aureus, with a 351 log difference between lag and stationary phases. The difference in detection limits for each growth phase is important to consider when determining appropriate use of ATP bioluminescence in real-world setting. It may be impossible to determine whether a bacterial contaminant is in a growth phase or a nongrowth or sublethal state. If a contaminant is introduced into a food source, it may quickly enter a log phase, making it easier to detect smaller numbers. If a contaminant has been dried on a surface for hours, for example overnight, then it might be in a stationary or death phase, or in a sublethal state, especially if limited nutrients are available and conditions are not 3 8 E. coli CFU ml 1 x 1 7 25 2 15 1 5 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 Time (h) 7 6 5 4 3 2 1 ATP reading (RLU) Figure 1 Average of three growth curves for Escherichia coli showing lag, log, stationary and death phases. CFU ml 1 9 1 7 ( ) are shown on the left side Y-axis and ATP measurement in RLU ( ) on the right side Y-axis. Error bars indicate the standard deviations for three replicate growth curves (1 SD = 17 CFU ml 1 and 292 RLU). 372 Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology

S.J. Vogel et al. ATP detection limits 16 14 8 7 Figure 2 Average of three growth curves for Staphylococcus aureus showing lag, log, stationary and death phases. CFU ml 1 9 1 8 ( ) are shown on the left side Y-axis and ATP measurement in RLU ( ) on the right side Y-axis. Error bars indicate the standard deviations for three replicate growth curves (1 SD = 5 CFU ml 1 and 7 RLU). S. aureus CFU ml 1 x 1 8 12 1 8 6 4 2 1 2 3 4 5 6 7 8 Time (h) 6 5 4 3 2 1 ATP reading (RLU) favourable. In that case, more viable organisms could be present, but they would be less easily detected. Other considerations in environmental testing are initial bacterial concentration and temperature (Theron et al. 1987; Fujikawa and Morozumi 22, 23). The lower the starting concentration of bacteria, the longer the lag phase, and the slower the rise in ATP. The same situation occurs with lower temperatures. Theron et al. (1987) showed that the ATP concentration of Enterobacter cloacae and Bacillus cereus dropped rapidly within 1 minutes of a temperature drop to 4 6 C. When held at 4 6 C for up to 48 hours, the ATP concentrations remained low. Organisms tested were from 24-h cultures, which would most likely be in stationary phase to begin with, where ATP concentration is low relative to cell number. This study did not evaluate bacteria held at room or refrigerator temperature; however, testing of food products or surfaces at lower temperatures may also result in decreased sensitivity of ATP reading due to a drop in ATP concentration. A limitation of this study is that only two representative organisms were tested. Although different strains of these organisms and different organisms will differ in the Table 2 Detection limits (log 1 CFU ml 1 at which RLU reading was obtained) for each growth phase for Escherichia coli and Staphylococcus aureus Growth Phase* Staph. aureus Log 1 E. coli Log 1 CFU ml 1 CFU ml 1 Average (Range) SD Average (Range) SD Lag 627 (618 635) 25 437 (432 443) 6 Log 588 (568 599) 87 515 (494 534) 21 Stationary 745 (735 758) 33 788 (778 798) 1 Death 688 (669 696) 9 757 (719 768) 28 *Lag phase = 1 h; log phase = 6 h; stationary phase = 12 h; and death phase = 144 h for E. coli and 72 h for Staph. aureus. quantity of ATP present at each phase of the growth cycle, this study was restricted to these two organisms to determine a baseline degree of variation between growth phases. The differences between the two organisms tested here, the phase in which the greatest sensitivity was seen and the overall difference between the growth phases for each organism, indicate that the target organism(s) will influence the sensitivity of ATP devices. Additional studies should also include detection limits in stressed organisms and organisms on surfaces as opposed to liquid media as the growth rates and ATP concentrations may differ. Other ATP devices should also be tested for detection limits across the growth curve because there are significant technological differences between them. Bacteria contamination is of concern in many settings because of the risk of infectious diseases and food spoilage. This is the first report of the limits of detection of such a device during each bacterial growth phase. The high limit of detection of the SystemSure for bacterial ATP, especially during stationary and death growth phases, is a limitation that must be taken into consideration when determining how the device will be used. Materials and methods Bacterial strains Escherichia coli (ATCC 29214) and Staph. aureus (ATCC 6538) were purchased from the American Type Culture Collection (Manassas, VA). Within-run and between-run precision Within-run precision of the SystemSure was determined for five substrates: (i) positive control consisting of lyophilized ATP (Hygiena, Camarillo, CA), (ii) lag phase E. coli Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology 373

ATP detection limits S.J. Vogel et al. culture, (iii) log phase E. coli culture, (iv) lag phase Staph. aureus culture and (v) log phase Staph. aureus culture. Lag phase cultures were tested at time 1 h and log phase at 6 h. Each was tested 1 times in immediate succession on the same day with the SystemSure, and the RLU reading was recorded. Mean, standard deviation and coefficient of variation (CV) were calculated. For between-run precision, positive control was tested with the SystemSure once a day on 1 successive days and the RLU recorded. Mean, standard deviation and coefficient of variation were calculated. Growth curves A single colony was inoculated in 1 ml nutrient broth (Oxoid LTD, Basingstoke, Hampshire, UK) and incubated overnight for 16 18 h to achieve stationary phase. At time zero, one ml of stationary phase culture was added to 1 ml nutrient broth. This culture was incubated at 37 C with shaking through the entire growth curve. At designated time points, an aliquot was removed from the culture and serially diluted in phosphate buffer solution (PBS, Fisher Scientific International Inc., Waltham, MA). Undiluted and serial 1-fold dilutions were tested in duplicate with the SystemSure Plus using Ultrasnap swabs, and 1 ll was spread-plated in duplicate onto tryptic soy agar plates (TSA, Becton, Dickenson and Company, Sparks, MD), which were incubated overnight, counted the next day and CFU ml 1 calculated. Samples were taken every 2 h for 12 h; then, every 24 h until death phase was achieved. Growth curves were run in triplicate, and averages and standard deviations at each growth phase were calculated. Limits of detection The growth curves generated above were used to select time points during each growth phase (lag phase: 1 h, log phase: 6 h, stationary phase: 12 h and death phase: 72 h for Staph. aureus and 144 h for E. coli). For these time points, the log 1 CFU ml 1 was calculated for the dilution at which the ATP reading was zero RLU. Three log 1 CFU ml 1 (one for each growth curve) were averaged to determine the limit of detection, and standard deviations were calculated at each growth phase. Acknowledgements This work was supported by an Academic Research Grant from the Toxics Use Reduction Institute at the University of Massachusetts, Lowell, MA. The authors would like to thank Amritpreet Birdi and Robert DeMatteo for their assistance in performing growth curves. Conflict of Interest No conflict of interest declared. References Aiken, Z.A., Wilson, M. and Pratten, J. (211) Evaluation of ATP bioluminescence assays for potential use in a hospital setting. Infect Control Hosp Epidemiol 32, 57 59. Alfa, M.J., Fatima, I. and Olson, N. (213a) Validation of adenosine triphosphate to audit manual cleaning of flexible endoscope channels. Am J Infect Control 41, 245 248. Alfa, M.J., Fatima, I. and Olson, N. (213b) The adenosine triphosphate test is a rapid and reliable audit tool to assess manual cleaning adequacy of flexible endoscope channels. Am J Infect Control 41, 249 253. Cunningham, A.E., Rajagopal, R., Lauer, J. and Allwood, P. (211) Assessment of hygienic quality of surfaces in retail food service establishments based on microbial counts and real-time detection of ATP. J Food Prot 74, 686 69. Fujikawa, H. and Morozumi, S. (22) New estimation methods of bacterial concentration by measuring ATP changes during incubation. J Food Hyg Soc Japan 43, 155 159. Fujikawa, H. and Morozumi, S. (23) Estimation of bacterial concentrations in commercial foods by measuring ATP changes during incubation. J Food Hyg Soc Japan 44, 83 88. Fujikawa, H. and Morozumi, S. (25) Modeling surface growth of Escherichia coli on agar plates. Appl Environ Microbiol 71, 792 7926. Fushimi, R., Takashina, M., Yoshikawa, H., Kobayashi, H., Okubo, T., Nakata, S. and Kaku, M. (213) Comparison of adenosine triphosphate, microbiological load, and residual protein as indicators for assessing the cleanliness of flexible gastrointestinal endoscopes. Am J Infect Control 41, 161 164. Griffith, C.J., Cooper, R.A., Gilmore, J., Davies, C. and Lewis, M. (2) An evaluation of hospital cleaning regimes and standards. J Hosp Infect 45, 19 28. Guh, A. and Carling, P. and the Environmental Evaluation Workgroup. (21) Options for evaluating environmental cleaning. Available at: http://www.cdc.gov/hai/toolkits/ Evaluating-Environmental-Cleaning.html. (Accessed 29September 213). ISSA (213) Standard for measuring the effectiveness of cleaning in K 12 schools. Available at: www.issa.com. (Accessed 29 September 213). Jasson, V., Jacxsens, L., Luning, P., Rajkovic, A. and Uyttendaele, M. (21) Alternative microbial methods: an overview and selection criteria. Food Microbiol 27, 71 73. Lewis, T., Griffith, C., Gallo, M. and Weinbren, M. (28) A modified ATP benchmark for evaluating the cleaning of some hospital environmental surfaces. J Hosp Infect 69, 156 163. 374 Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology

S.J. Vogel et al. ATP detection limits Luick, L., Thompson, P.A., Loock, M.H., Vetter, S.L., Cook, J. and Guerrero, D.M. (213) Diagnostic assessment of different environmental cleaning monitoring methods. Am J Infect Control 41, 751 752. Malik, R.E., Cooper, R.A. and Griffith, C.J. (23) Use of audit tools to evaluate the efficacy of cleaning systems in hospitals. Am J Infect Control 31, 181 187. Sciortino, C.V. and Giles, R.A. (212) Validation and comparison of three adenosine triphosphate luminometers for monitoring hospital surface sanitization: a Rosetta Stone for adenosine triphosphate testing. Am J Infect Control 4, e233 e239. Shama, G. and Malik, D.J. (213) The uses and abuses of rapid bioluminescence-based ATP assays. Int J Hyg Environ Health 216, 115 125. Theron, D.P., Prior, B.A. and Lategan, P.M. (1987) Effect of minimum growth temperature on the adenosine triphosphate content of bacteria. Int J Food Microbiol 4, 323 329. Turner, D.E., Daugherity, E.K., Altier, C. and Maurer, K.J. (21) Efficacy and limitations of an ATP-based monitoring system. J Am Assoc Lab Anim Sci 49, 19 195. Ukuku, D.O., Zhang, H., Bari, M.L., Yamamoto, K. and Kawamoto, S. (29) Leakage of Intracellular UV materials of high hydrostatic pressure-injured Escherichia coli O157:H7 strains in tomato juice. J Food Prot 72, 247 2412. Whitehead, K.A., Smith, L.A. and Verran, J. (28) The detection of food soils and cells on stainless steel using industrial methods: UV illumination and ATP bioluminescence. Int J Food Microbiol 127, 121 128. Willis, C., Morley, R., Westbury, J., Greenwood, M. and Pallett, A. (27) Evaluation of ATP bioluminescence swabbing as a monitoring and training tool for effective hospital cleaning. Br J Infect Control 8, 17 21. Wu, V.C.H. (28) A review of microbial injury and recovery methods in food. Food Microbiol 25, 735 744. Letters in Applied Microbiology 58, 37--375 213 The Society for Applied Microbiology 375