1 Center for Biofilm Engineering, Montana State University, Bozeman, MT 59717-3980, USA
2 Biofilm Laboratory, Division of Healthcare Quality Promotion, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA
3 Scientific Resources Program, Centers for Disease Control and Prevention, Atlanta, GA 30333, USA
Correspondence
Martin A. Hamilton
marty_h{at}erc.montana.edu
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In this study, preliminary experiments were performed to generate biofilms of three clinically relevant micro-organisms, Pseudomonas aeruginosa, Klebsiella pneumoniae and Streptococcus pneumoniae. Then a standard operating procedure (SOP) was created for growing a P. aeruginosa biofilm. This procedure was incorporated into an intra-laboratory evaluation to determine the repeatability and ruggedness of the method. We present a description of the CBR system, preliminary results on biofilms grown in that system, an SOP for P. aeruginosa and the results of a study designed specifically to evaluate the repeatability and ruggedness of the SOP.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
S. pneumoniae (clinical isolate from the Boston's Children's Hospital, Boston, MA, USA, provided by Paul Edmonds, Georgia Institute of Technology, Atlanta, GA, USA) was transferred from a frozen stock onto Trypticase soy agar containing 5 % sheep's blood (blood agar) (BD Microbiology Systems) and incubated at 35 °C in a 5 % CO2 incubator overnight. A single colony was picked and inoculated into a 10 ml tube of brain heart infusion (BHI) broth (Difco), and incubated for 12 h at 35 °C in a 5 % CO2 incubator. For the S. pneumoniae experiments, the CBR contained 1·27 cm diameter Teflon coupons. The reactor was sterilized in an autoclave, then filled with 400 ml of filter-sterilized full strength BHI broth. Nine millilitres of the 12 h culture was added to the reactor, providing 2·13x107 c.f.u. ml1 in the reactor as determined by plating on blood agar incubated at 35 °C in a CO2 incubator. A filter-sterilized mixture of 85 % N2, 10 % CO2 and 5 % O2 was continually supplied to the CBR to provide an atmosphere of supplemental CO2. The entire reactor was placed into a heated water bath to maintain a temperature of approximately 35 °C for the duration of the experiment. The system was operated in batch mode for 12 h, then in continuous-flow by pumping a 1/10 dilution of BHI broth at a flow rate of 0·5 ml min1 (providing a residence time of 13·3 h) for 24 h prior to sampling coupons.
In each preliminary experiment, all 24 coupons in the reactor were sampled and analysed. Each rod was placed into a biological safety cabinet and three coupons from one rod were aseptically removed. Each coupon was rinsed twice in Butterfield buffer, pH 7·2 (cat. no. 298267, containing 0·4 g monobasic potassium phosphate l1, 0·1 % peptone, 2·0 % polysorbate 80; Becton Dickinson) to remove planktonic cells and placed into a tube containing 10 ml of PBS. Biofilm bacteria were recovered by subjecting coupons to three alternating 30 s cycles of sonication at a frequency of 42 kHz (model 2510 sonicating water bath; Branson), followed by vortexing (Vortex Genie 2; Scientific Products). The removed biofilm was disaggregated by homogenizing the suspension with a tissue homogenizer (Polyscience Tissue Homogenizer model K-120; Polysciences) at 16 000 r.p.m. for 60 s. The disaggregated biofilm was then processed to quantify the number of viable cells. For P. aeruginosa and K. pneumoniae, this entailed spread-plating serial dilutions of the suspension onto R2A agar (Difco), incubating the plates at 35 °C for 48 h and counting the colonies. For S. pneumoniae, the diluted suspension was spread-plated onto blood agar and colonies were counted after incubating for 24 h at 35 °C in a CO2 incubator.
Following the preliminary experiments, the experimental protocol was modified to provide conditions optimal for the P. aeruginosa strain used in the ruggedness tests. The sample and analysis steps were also evaluated and modified (see the standard operating procedure below).
Standard operating procedure (SOP).
Polycarbonate coupons were sonicated for 30 s in a detergent solution (a 12 % solution of Micro-90; International Products Corporation). Then each was rinsed and sonicated for another 30 s in reagent-grade water. Alternate rinsing and sonication was repeated until no soap was left on the coupon surface. The coupons were then soaked for 2 h in 2 M HCl, rinsed, and allowed to air dry prior to use. One coupon was positioned into each hole of the reactor rods so that the face of the coupon was flush with the rod surface that faced the baffle. A set screw was tightened to hold the coupon in place. The reactor system was then assembled and 500 ml of 300 mg tryptic soy broth (TSB) l1 (Difco) was added. The assembled system, minus the pump and stir plate, was autoclaved.
After the system had cooled to 23 °C, the glass vessel was set on a digital stir plate and inoculated with 1 ml of a 108 c.f.u. ml1 suspension of P. aeruginosa (ATCC 700888). The P. aeruginosa suspension was prepared by inoculating 100 ml of broth (300 mg TSB l1) with a single colony collected from a bacterial isolation plate. The suspension was incubated for 1824 h at 37 °C in a shaker. Immediately after the addition of P. aeruginosa to the reactor, the stir plate was set to rotate at 125 r.p.m. The biofilm was allowed to establish for 24 h in batch phase while the baffle rotated. A 24 h continuous-flow phase followed immediately. While the baffle continued to rotate, a 100 mg TSB l1 nutrient broth was pumped into the reactor at a rate equal to 11·7 ml min1, resulting a 30 min residence time.
The stir plate and pump were turned off after 24 h of continuous flow and the biofilm was harvested from a predetermined number of randomly selected coupons. For each selected coupon, the rod holding that coupon was removed through the top of the reactor. The set screw that held the coupon in place was loosened and the coupon was removed using a flame-sterilized haemostat. Care was taken not to disturb the surface of the coupon facing the baffle because that surface held the biofilm sample that was analysed. Following the procedures described by Zelver et al. (1999), the biofilm was then scraped from the surface, homogenized to create a uniform cell suspension, serially diluted, plated on R2A agar, incubated for 1824 h at 37±2 °C and viable cell numbers (c.f.u.) were counted. In brief, to remove the biofilm from the coupon, the coupon was held with a sterile clamp. Using a sterile applicator stick, the surface of the coupon was thoroughly scraped for about 1 min. During that time, the stick was occasionally stirred in 9 ml sterile buffered water to remove attached material. After sufficient scraping, the coupon surface was rinsed with 1 ml sterile buffered water. The final volume in the sample test tube was 10 ml. Then a sterile homogenizer probe was inserted into the homogenizer and the sample tube was homogenized at
20 500 r.p.m. for 60 s. The disaggregated biofilm was then processed to quantify the number of viable cells. This entailed drop-plating serial dilutions of the suspension onto R2A agar (Difco), incubating the plates at 37 °C for 24 h, and counting the colonies (Herigstad et al., 2001
).
The key measurement was the log10 density, where density was expressed in units of c.f.u. cm2, for each sampled coupon, calculated with the following equation:
log density=log10(mean c.f.u. per plate)+log10(vol. scraped into)+log10(dilution)log10(vol. plated)log10(area of coupon face)
The area of coupon face was 1·267 cm2. For this and all subsequent statistical calculations performed on the log10 scale, five or more significant figures past the decimal point were carried; rounding occurred only at the conclusion of the calculations.
Ruggedness and repeatability evaluation study.
A series of experiments were conducted to estimate the repeatability standard deviation, denoted by sr, and the regression coefficients that measure the ruggedness of the CBR SOP. In these experiments, the SOP was not followed exactly; instead, some of the operational factors were purposely varied slightly. For example, some experiments were conducted at 20 °C rather than at the SOP temperature of 23 °C. For the ruggedness tests, we purposely altered the settings of four operational factors temperature, r.p.m. of the rotating baffle, time in batch, and nutrient concentration during continuous flow. Three settings were selected for each factor (Table 1).
|
Ruggedness was quantified by the regression coefficients in a least-squares multiple regression analysis where the response variable was log density and the predictor variables were temperature, r.p.m., log10-transformed time in batch, and log10-transformed nutrient concentration during continuous flow. To do the analysis, the four predictor variables were entered as covariates into the General Linear model component of the analysis of variance (ANOVA) module in the computer software package Minitab (Release 13; Minitab, State College, PA, USA).
The ANOVA also provided a variance component analysis to assess the variance within experiments and the variance between experiments. The square root of the sum of those two variances was sr, which was interpreted as the typical difference, sign neglected, between the log density for a single (randomly chosen) experiment and the mean log density across many independent, identical (same operational factor settings) experiments.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Coupon position
Each rod in the CBR held three coupons in vertical alignment. Data from the preliminary experiments with K. pneumoniae, S. pneumoniae and P. aeruginosa were submitted to an analysis of variance. Neither the eight rods nor the three coupon positions significantly affected the mean log density (P>0·50 for each of the three organisms). For the SOP ruggedness test experiments, the mean differences in log densities were 0·18 (bottom minus middle), 0·10 (bottom minus top) and 0·08 (middle minus top). The mean log densities for the three positions were not significantly different (P=0·22). These results indicate that the 24 coupons in the reactor were equally representative of the log density.
Repeatability
For the SOP ruggedness test experiments, replicate experiments were conducted to allow assessment of repeatability of the SOP. The mean log10 P. aeruginosa density (c.f.u. cm2) was 7·0590, independent of coupon position within the reactor. The estimated within-experiment variance was 0·1884 (estimated with 103 degrees of freedom) and the estimated between-experiment variance was 0·1656 (estimated with 8 degrees of freedom). Therefore, , of which 53 % was attributable to within-experiment variation and 47 % to between-experiment variation. This sr pertained to a protocol that sampled only one coupon per experiment. The repeatability standard deviation for a protocol that requires sampling n coupons per experiment is
![]() |
According to this equation, the sr for the biofilm mean log density based on three coupons (n=3) would equal 0·48, of which 27 % is attributable to within-experiment variation and 73 % to between-experiment variation. For the most intense sampling protocol possible, where all n=24 coupons in the reactor are sampled, sr=0·42, which is 95 % attributable to between-experiment variation.
Because there is an important between-experiment variance, it would be reasonable to choose an operating procedure that requires m independent experiments, n coupons per experiment. In this case, the log density estimate would be the mean log density across all n·m coupons, where the repeatability standard deviation is
![]() |
Ruggedness
Equation (1) is the least-squares regression model for log density. It shows the estimated log density associated with any pattern of operating conditions, as long as those operating conditions are near the SOP specifications. Fig. 2 shows the relationship between the predicted and observed log density for each coupon, where the predictions are based on equation (1). The correlation coefficient between the predicted and observed values (the multiple correlation coefficient) is 0·89, indicating that the regression model is a good fit to the log densities.
![]() |
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
A method is reproducible if the same result occurs when the same experiment is run independently by different researchers in different laboratories. A collaborative study involving several laboratories is required to assess reproducibility. The results of the collaborative study are summarized conventionally by a standard deviation, called the reproducibility standard deviation, which can be no smaller, and is usually significantly larger, than the repeatability standard deviation (AOAC, 1995; ASTM, 2002a
; Feldsine et al., 2002
).
A rugged method is one for which the outcome is insensitive to minor perturbations of critical factors or conditions. There is no conventional quantitative measure for ruggedness, though several have been suggested (Thompson et al., 2002; AOAC, 1998
Appendix C; ASTM, 2002b
). One reason for conducting a ruggedness test is to provide a single laboratory approximation to a collaborative study. If the operational factors are changed by the amounts that one expects them to vary among laboratories, the standard deviation of the results should be similar to the reproducibility standard deviation. That expectation may be too optimistic. Although there could be considerable savings of resources if in fact a single laboratory ruggedness test could be substituted for a collaborative study, it is not possible at present to design such a test, due to lack of knowledge about all relevant factors (Thompson, 2000
). A ruggedness test can highlight the critical components of a laboratory method so that practitioners know which steps or conditions require special attention or which parameters should be optimized before a collaborative study is performed. Our informal examination of the literature indicates that ruggedness evaluations of new microbiological methods are seldom conducted. Moreover, we are unaware of other studies describing a quantitative assessment of the influence of individual operational factors on microbiological methods in general and biofilm methods in particular. The work presented here demonstrates that quantitative ruggedness evaluations of microbiological methods can be feasible and informative. The statistical design and analysis strategy presented here could also be used to evaluate other microbiological methods.
This study used a response-surface experimental design (ASTM, 2002b; NIST/SEMATECH, 2003
), starting with a fractional factorial design (Youden & Steiner, 1975
), to identify the smallest number of experiments necessary for a multiple regression analysis. This two-step approach was efficient and provided the desired measure of the influence of each factor.
The choice of which operational factors to study and the range of settings for each factor were subjective steps in the ruggedness evaluation. It was unfeasible to study all operational factors; therefore, we relied on experience gained during the development of the reactor and SOP to choose the important factors for evaluation. The relevant settings for each factor were those that fell within the range expected when competent researchers faithfully followed the SOP. We observed responses at three or more levels of each factor. This made it possible to inspect the data for a potential nonlinear association between log density and the factor. There was no evidence of nonlinearity within the range of settings used for any of the four factors. The most reliable regression coefficient estimates occur when one chooses lower and upper settings of the operational factor as far apart as possible within the range of linearity (Thompson et al., 2002).
The interpretation of the influence of operational factors is complicated if there is an interaction, i.e. if the regression coefficient for one factor is dependent on one of the other factors. We did not expect interactive effects to occur within the ranges of factor settings investigated. However, it was unfeasible to run the large number of experiments required to check whether all potential interactions were in fact negligible. Note that the model of equation (1), which does not contain interaction terms, fits the data (Fig. 2).
The CDC system
Laboratory-grown biofilms are engineered to emulate a specific real-world environment. By altering parameters, such as flow dynamics and reactor configuration, it is possible to grow biofilms differing in structure (e.g. thickness) and function (e.g. nutrient consumption). A case in point is the effect of shear conditions on biofilm structure. Stoodley et al. (1999) showed that under high shear, biofilms consisted of dense elongated cell clusters while under low shear, biofilms were composed of less dense round cell clusters. Neither of these is the uniformly correct laboratory biofilm; each emulates a different growth environment. Because the choice of reactor affects the laboratory biofilm, it is important for the investigator to choose the appropriate reactor and growth conditions. The biofilm reactor (CBR) described in this paper provided a useful tool for growing repeatable biofilms under constant shear using a variety of organisms. By standardizing the design of the reactor, establishing an SOP, and performing multiple experiments, it was possible to conduct a statistical evaluation.
The ruggedness results showed that the baffle rotation speed could be an influential factor if it is not carefully controlled. For example, if the protocol was followed except that the stir plate was set so that the baffle rotated at a speed of 180 r.p.m. instead of 125 r.p.m. as specified by the protocol, the mean log density for a coupon would be estimated to be 6·58, a decrease of 0·47. We recommend that a digital stir plate be used with the CBR to provide that control.
The CBR can be operated under a wide range of controllable conditions. With modifications to the SOP, we believe it can be used to grow a standard biofilm for addressing diverse research questions.
![]() |
ACKNOWLEDGEMENTS |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
AOAC (1995). Guidelines for collaborative study procedures to validate characteristics of a method of analysis. J AOAC Int 78, 143A160A.
AOAC (1998). AOAC Peer-verified Methods Program Manual on Policies and Procedures. Gaithersburg, MD: AOAC International.
ASTM (2002a). E456-02: Standard terminology relating to quality and statistics. In Annual Book of ASTM Standards, vol. 14.02. West Conshohocken, PA: ASTM International.
ASTM (2002b). E1169-02: Standard guide for conducting ruggedness tests. In Annual Book of ASTM Standards, vol. 14.02. West Conshohocken, PA: ASTM International.
Camper, A., Jones, W. L. & Hayes, J. T. (1996). Effect of growth conditions and substratum composition on the persistence of coliforms in mixed-population biofilms. Appl Environ Microbiol 62, 40144018.[Abstract]
Ceri, H., Olson, M. E., Stremick, C., Read, R. R., Morck, D. & Buret, A. (1999). The Calgary Biofilm Device: a new technology for rapid determination of antibiotic susceptibilities of bacterial biofilms. J Clin Microbiol 37, 17711776.
Characklis, W. G. & Marshall, K. C. (1990). Biofilms. New York: Wiley.
Donlan, R. M., Murga, R., Carpenter, J., Brown, E., Besser, R. & Fields, B. (2002). Monochloramine disinfection of biofilm-associated Legionella pneumophila in a potable water model system. In Legionella, pp. 406410. Edited by R. Marre and others. Washington, DC: American Society for Microbiology.
Donlan, R. M., Piede, J. A., Heyes, C. D., Sanii, L., Murga, R., Edmonds, P., El-Sayed, I. & El-Sayed, M. A. (2004). A model system for growing and quantifying Streptococcus pneumoniae biofilms in situ and in real time. Appl Environ Microbiol 70, 49804988.
Feldsine, P., Abeyta, C. & Andrews, W. H. (2002). AOAC International methods committee guidelines for validation of qualitative and quantitative food microbiological official methods of analysis. J AOAC Int 85, 11871200.[Medline]
Herigstad, B., Hamilton, M. & Heersink, J. (2001). How to optimize the drop plate method for enumerating bacteria. J Microbiol Methods 44, 121129.[CrossRef][Medline]
ISO (1993). Guide to the Expression of Uncertainty in Measurement. Geneva: ISO.
Kharazmi, A., Giwercman, B. & Hoiby, N. (1999). Robbins Device in biofilm research. Methods Enzymol 310, 207215.[Medline]
Mittelman, M. W., Kohring, L. L. & White, D. C. (1992). Multipurpose laminar-flow adhesion cells for the study of bacterial colonization and biofilm formation. Biofouling 6, 3951.
Murga, R., Forster, T. S., Brown, E., Pruckler, J. M., Fields, B. S. & Donlan, R. M. (2001). Role of biofilms in the survival of Legionella pneumophila in a model potable-water system. Microbiology 147, 31213126.[Medline]
Nickel, J. C., Wright, J. B., Ruseska, I., Marrie, T. J., Whitfield, C. & Costerton, J. W. (1985). Antibiotic resistance of Pseudomonas aeruginosa colonizing a urinary catheter in vitro. Eur J Clin Microbiol 4, 213218.[Medline]
NIST/SEMATECH (2003). e-Handbook of Statistical Methods. http://www.itl.nist.gov/div898/handbook/.
O'Toole, G. A. & Kolter, R. (1998). Initiation of biofilm formation in Pseudomonas fluorescens WCS365 proceeds via multiple, convergent signaling pathways: a genetic analysis. Mol Microbiol 28, 449461.[CrossRef][Medline]
Stoodley, P., Dodds, I., Boyle, J. D. & Lappin-Scott, H. M. (1999). Influence of hydrodynamics and nutrients on biofilm structure. J Appl Microbiol Symp Suppl 85, 19S28S.
Thompson, M. (2000). Towards a unified model of errors in analytical measurement. Analyst 125, 20202025.[CrossRef]
Thompson, M., Ellison, S. & Wood, R. (2002). Harmonized guidelines for single-laboratory validation of methods of analysis. Pure Appl Chem 74, 835855.
Youden, W. & Steiner, E. (1975). Statistical Manual of the Association of Official Analytical Chemists. Gaithersberg, MD: AOAC International.
Zelver, N., Hamilton, M., Pitts, B., Goeres, D., Walker, D., Sturman, P. & Heersink, J. (1999). Measuring antimicrobial effects on biofilm bacteria: from laboratory to field. Methods Enzymol 310, 608628.[Medline]
Received 15 October 2004;
revised 6 December 2004;
accepted 10 December 2004.
HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
INT J SYST EVOL MICROBIOL | MICROBIOLOGY | J GEN VIROL |
J MED MICROBIOL | ALL SGM JOURNALS |