Some Observations on Bacterial Thermal Death Time Curves'

Some Observations on Bacterial Thermal Death Time Curves' J. J. LICCIARDELLO AND J. T. R. NICKERSON I)epartment of Nutrition and Food Science, Mllassachusetts Institute of Technology, Cambridge, Mfassachusetts Received for publication 6 June 1963 ABSTRACT LICCIARDELLO, J. J. (AMassachusetts Institute of Technology, Cambridge) AND J. T. R. NICKERSON. Some observations on bacterial thermal death time curves. Appl. MIicrobiol. 11:476-48. 1963. Thermal death rate data were obtained for spores of Clostridium. sporogenes P.A. 3679 and Bacillus subtilis var. niger, and for cells of Salmnonella senftenberg 775W. The survival curves for P.A. 3679 were approximately linear, but for B. subtilis var. niger or S. senftenberg 775W they were sigmoidal. Decimal reduction times were derived from the regression slopes of the apparent linear portion of the survival curves, and from these a phantom thermal death time (TDT) curve was constructed. In general, the phantom TDT curves were linear for B. subtilis var. niger and S. senftenberg 775W and nonlinear for P.A. 3679. Bigelow (1921) conducted his classical study on thermal death time (TDT) curves for bacterial spores and concluded that the relationship between the logarithms of the TDT and the corresponding temperatures was linear. On the basis of much corroborating evidence reported since then, this conclusion has come to be accepted by many inivestigators. It should be pointed out, however, that in most inistances these latter determinations were carried out over a comparatively narrow temperature range. Occasionially, there have been results reported which indicate that, when TDT curves are determined over a broad temperature range, there is a deviation from linearity, particularly at the extremes of temperature. Reports of this type have not usually received serious conisideration and have been dismissed with the belief that the discrepanicy was due to experimental error. On the assumption that the order of death of bacteria is logarithmic or first order, it has been shown mathematically by applying the Arrhenius equation that, theoretically, the TDT curve should have a systematic curvature, being asymptotic to both the ordinate and abscissa (Gillespy, 1951). Within a narrow range of temperature, the TDT curve should be approximately linear. Although Bigelow (1921) concluded that the TDT curves for spores of 15 different thermophiles were of a linear nature, the actual curves that he obtained were ' Contribution number 527 from the Department of Nutrition and Food Science, AMassachusetts Institute of Technology, Cambridge. sigmoidal. Over the temperature interval of 1 to 15 C, the curve exhibited a lag. From 15 to 125 C, there was a linear relationship between log time and temperature, and from 125 to 14 C the curve began to tail off. Bigelow attributed the tailing off to experimental error, but he did seem to accept the deflection of the TDT curve at the lower temperatures as a valid experimental observation. Halverseni and Hays (1936) reported TDT for Clostridiumn botulinum1 spores in various foods at 5-degree increments from 9 to 115 C. A plot of their data yielded TDT curves with a lag at the low end of the temperature range. In a previous study with spores of C. sporogenes P.A. 3679, Licciardello and Nickerson (1962) found the phantom TDT curve to be concave down in the temperature range of 9 to 11 C but, since the curve was based on determinations made only at three different temperatures, no definite conclusioni could be reached. The purpose of this investigation was to conduct a more detailed study of the TDT curves for C. sporogenes and other organisms to determine the shape of the curve over a broad temperature range. MATERIALS AND MWETHODS Pr opagation of test or)ganisms. An initial inoculum of actively growing C. sporogenes cells was prepared from a previously harvested spore crop of P.A. 3679 (ATCC 7955). This was added to Erlenmeyer flasks containing a beef liver infusion medium together with cooked ground liver and several iron nails. The following ingredients were also added (in g per liter): peptone (Difco), 5; Tryptone (Difco), 1.5; dextrose, 1; K2HPO4, 1.25; with the ph adjusted to 7.2 to 7.4 with KOH. After incubation for 1 week at 37 C and 2 weeks at 3 C, the spores were harvested, washed several times with distilled water, and resuspended in phosphate buffer (ph 7.). Samples of a 24-hr nutrient broth culture of Bacillus subtilis var. niger which had been propagated from a single isolated colony, were spread over the surface of a solid medium contained in Roux flasks. The medium had the following composition:.1 % Casamino Acids,.25% glucose,.5 % yeast extract,.1 % MnSO4,.1 % FeSO4, and 3 %/c agar. After 5 days of incubation at 37 C, the spores were harvested, washed several times, and resuspended in phosphate buffer (ph 7.). Salmonella senftenberg 775W was obtained from a culture collection maintained within the department, but the 476

VOL. 11, 1963 BACTERIAL THER-MAL DEATH TIME CURVES 477 original culture had been obtained from the Western Regional Research Laboratory, Albany, Calif. The cells were grown and harvested under the same conditions as described for B. subtilis var. niger, except that they were grown on nutrient agar and incubated for 3 days. All spore or cell suspensions were kept in screw-cap jars containing glass beads, and were held at 2 to 4 C until used. Prior to determining thermal resistance, spores of P.A. 3679 were heat-shocked for 1 min at 1 C, and spores of B. subtilis were heat-shocked for 1 min at 8 C. M1ethod for deternining thermnial resistance. Thermal death rate was determined by a capillary TDT technique which has been previously described (Licciardello and Nickerson, 1962). In essence,.25 ml of spore suspenision was introduced within melting-point capillary tubes by means of a Warburg-manometer calibrator (MIicro-Metric Instrument Co., Cleveland, Ohio) fitted with a glass adapter and a 24-gauge hypodermic needle. The tubes were sealed and heated for various intervals of time in an apparatus that was described by Farkas (1962). The number of survivors was determined by a direct plating procedure, anid a survival curve was thus established. Four capillary tubes were usually heated at each different time period. Plating procedure. Quantitative counts of P.A. 3679 were made by inoculating (in triplicate) deep-culture Pyrex tubes (12 by 2 mm), containing.5 ml of a sterile 5 %c sodium bicarbonate solution, with decimal dilutions of the heated spore suspension. Approximately 12 ml of an anaerobic culture medium were then added, followed by stratification with a 2 %O agar solution. Colonies were counted after 18 to 26 hr at 37 C. Composition of the culture medium was: 25 ml of beef liver infusion (prepared by simmering 1 lb of ground beef liver for 1 hr with 1 liter of water, filtering, and adjusting the ph of the infusion to 7.1 to 7.3 with KOH); 75 ml of distilled water; 15 g of agar; 29.3 g of dehydrated fluid Thioglycollate Mledium (Difco) with resaurin indicator. Spores of B. subtilis var. niger and cells of S. senftenberg 775W were counted by a spread surface technique. Portionis (.1 ml) of appropriate dilutions were spread over the surface of prepoured agar plates (prepared 48 hr in advance) by means of glass rods shaped in the form of a hockey stick. Tryptone Glucose Extract Agar was used for B. subtilis var. niger, anid Trypticase Soy Agar supplemented with 1 g of yeast extract per liter was used for S. senftenberg. Colonies of B. subtilis var. niger could be easily distinguished from occasional contaminants by the typical orange color of the colonies, and S. sen-ftenberg 775W could be distinguished by the morphological characteristics of the colonies. Serial dilutions of spore suspenisions of P.A. 3679 or B. subtilis var. niger were made with distilled water at room temperature but, in the case of S. senftenberg 775W, refrigerated phosphate buffer (ph 7.) was used. RESULTS Figure 1 depicts some of the survival curves at various temperatures for spores of C. sporogenes P.A. 3679 in phosphate buffer (ph 7.). Similarly shaped survival curves were obtained at intermediate temperatures. Each point represents the average number of survivors of four replicate capillary TDT tubes. The straight lines drawn through the points are regression lines as determined by the method of least squares. The relationship between log survivors and heating time is fairly linear, although in some instances there appears to be a slight tailing off. Decimal reduction time (D value) was determined as the negative reciprocal of the regression slope of the survivor curve and by plotting log D as a function of heating temperature; a phantom TDT curve was obtained. This phantom TDT curve for P.A. 3679 is obviously nonlinear (Fig. 2). The 95 % confidence limit of each regression slope was calculated (Hald, 1957), anid these are shown in Fig. 2 as upper and lower limits. It is evident that these 2 INUTES 24 HEATING TIME FIG. 1. Survival curves at variouis temnperatures for spores of Clostridium sporogenes P.A. 3679 in phosphate buffer (ph 7.). w - 3a -J 1 Uvv a I- _ I.1 8 85 9 95 1 15 11 115 12 1- FIG. 2. Phantom thermal death time curve for Clostridium sporogenes in phosphate buffer (ph 7.).

478 LICCIARDELLO AND NICKERSON APPL. MICROBIOL. limits are too narrow to satisfy a single straight line drawn through the points. Figure 3 illustrates some of the survival curves at various temperatures for spores of B. subtilis var. niger suspended in phosphate buffer (ph 7.). Survival curves at intermediate temperatures were similar in form, and are not shown to avoid overcrowding the figure. The sigmoidal nature of these survival curves made it difficult to construct a phantom TDT curve based on the concept of a D value. Some investigators, when dealing with sigmoidal survivor curves, determined the D value from the regression slope of the linear portion of the data. Reynolds and Lichtenstein (1952) showed that a phantom TDT curve derived in this manner had the same slope as the TDT curve constructed from end-point determinations It should be realied, however, that when D values derived in this manner are converted into TDT, a factor should be included which takes into consideration the initial lag in the survivor curve. The survival data in Fig. 3 appear to become linear after having gone through approximately 1 log cycle of reduction. Thus, regression slopes were determined from this point on. The D values derived from these slopes are plotted in Fig. 4 as a function of temperature. The phantom TDT curve appears to be linear, and it may be that at the low end of the temperature range the curve is becoming convex as may be predicted by application of the Arrhenius equation. Baker and McClung (1939) constructed phantom TDT curves from sigmoidal survival curves of Escherichia coli by plotting the log of the time required for 1 or.1 % survival as a function of temperature. Phantom TDT curves drawn in this fashion from the survival data of B. subtilis var. niger are presented in Fig. 4. These curves are concave down. Schmidt (1957) suggested the following empirical relation between the thermal death time, F, and D: F = D (log a + 2) Since the initial number of spores (a) in our study was approximately five million, the equality is reduced to: F = 8.7 D. The curve obtained by plotting log of the time required for each survival curve to traverse 8.7 log cycles (approximately 1-7% survival) as a function of temperature is presented in Fig. 4. This curve is considerably flattened in relation to the curves derived from the 1 % or the.1 % survival levels, and for all practical purposes has the same slope as the D curve. It is obvious that as one plots times for smaller and smaller levels of per cent survival, the shape of the curve approaches that of the phantom TDT curve constructed from D values. Survival curves for cells of S. senftenberg 775W heated in phosphate buffer (ph 7.) are presented in Fig. 5. These curves are sigmoid in form and an attempt to straighten them out by plotting the data on log probability paper was not successful. Therefore, TDT data were obtained in an 1 \1 D" CURVE 1 X DLiJ w~~~~~~~~~~~ E 1 o O )1 F.1% 1 _j ~SURVIVAL o D ~~~~1% U) 1 SUVVA Il I 65 7 75 8 85 9 95 1 15 > L~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C) FIG. 4. Phantom thermal death time curves for Bacillus subtilis var. niger in phosphate buffer (ph 7.). 57.2eC (SCALE B) D)"OM 15 18 21 24 27 3 HEATING TIME FIG. 3. Survival curves at various temperatures for spores of Bacillus subtilis var. niger in phosphate buffer (ph 7.). 1 2 4 6 8 1 12 MINUTES SCALE A 2 4 6 8 1 MINUTES SCALE B 2 4 6 8 1 HOURS SCALE C 15 3 45 6 75 HOURS 9 SCALE D!5 FIG. 5. Survival curves at various temperatures for Salmonella senftenberg 775W in phosphate buffer (ph 7.).

VOL. 11 i 1963 BACTERIAL THERMAL DEATH TIME CURVES 479 analogous manner as for B. subtilis var. niger. The D values were calculated from the regression slopes of the apparent linear portion of the survival curves. The phantom TDT curve constructed from these D values was linear (Fig. 6). In Fig. 6, one can also see the concave phantom TDT curve that resulted when the exposure time for 1 % survival was plotted against temperature. A graph of the exposure time for.1 % survival more closely approximated the D curve. DIscUSSION The nonlinear phantom TDT curve for C. sporogenes may be difficult to accept, as the shape is concave rather than convex such as would be predicted by the Arrhenius equation. However, similarly shaped TDT curves were obtained by Halversen and Hays (1936) for C. botulinum over the temperature range of 9 to 115 C and by Bigelow (1921) for thermophilic spores in the temperature range of 1 to 125 C. Bigelow suggested that certain biological changes may occur within bacterial spores at mild heating temperatures which render the spores more susceptible to inactivation. In the case of C. sporogenes, it is quite possible that exposure of the spores to temperatures below 1 C induced the initial stage of germination within the spore in which condition the spore was more vulnerable to heat. Evidence along this line was reported by Evans and Curran (1943), who found that germination of Bacillus spores was accelerated by incubation at temperatures from 65 to 95 C. The TDT curve appeared to be linear over the range of 15 to 12 C, and it should be pointed out that much of the previous work on TDT curves of clostridia has been done in this temperature range. In the case of B. subtilis var. niger and S. senftenberg 46.2 51.7 572 62.7 65.5 FIG. 6. Phantom thermal death time curves for cells of Salmonella senftenberg 776W in phosphate buffer (ph 7.). 775W, construction of phantom TDT curves from survival rate data was complicated by the nonlinear survival rates. It probably should be mentioned at this point that for years there has been a controversy among bacteriologists as to the shape of bacterial survival curves. Regarding this, there are two schools of thought. One group maintains that bacteria die at a logarithmic rate based on a uniformity of resistance among all cells in a population, whereas the second group believes that a bacterial population usually consists of a graded distribution of resistance among the individual cells and, for this reason, the survival curves are nonlinear. In support of this latter theory, Withell (1942) showed that the.individual resistances of cells in a population very closely approximated a log normal distribution, and he- indicated the various forms of survival curves that could be obtained from a bacterial population of log normally graded resistance. The shape of the curve was governed by the mean and standard deviation of the resistance distribution and, for various values of these parameters, linear or nonlinear survival curves could be obtained. Nevertheless, an attempt was made in this study to construct phantom TDT curves from the nonlinear survival curves of B. subtilis var. niger and S. senftenberg 775W by use of different methods which have been suggested. It is considered that the phantom TDT curves obtained by plotting time for 1 or.1 % survival against temperature are not valid because, as the value for per cent survival became smaller, the resulting TDT curve tended to parallel the D curve. Therefore, it is concluded that the TDT curve for spores of B. subtilis var. niger and for cells of S. senftenberg 775W is linear over the temperature range investigated. Angelotti, Foter, and Lewis (196) found log D to be a linear function of temperature over the range of 54.5 to 65.6 F for S. senftenberg 775W in various food substrates. The explanation offered for the concave TDT curve for C. sporogenes was that, at low heating temperatures, the spores may have become more susceptible to heat, possibly owing to an induction of the initial stage of germination. However, this proposed induced germination at low heating temperatures did not seem -to have occurred with spores of B. subtilis. It is entirely possible that the concept of decimal reduction time, as derived from the survival curves of this investigation, does not accurately characterie thermal death time. We feel that if the existence of nonlinear survival curves is to be recognied, then there is a need for further study concerning the determination of the parameters necessary for thermal process calculation. It would be interesting to compare phantom TDT curves that were determined from survival curves of bacterial populations of different degrees of graded resistances. ACKNOWLEDGMENT This research was supported by National Institutes of Health grant EFO6-5.

48 LICCIARDELLO AND NICKERSON APPL. 1\IICROBIOL. LITERATURE CITED ANGELOTTI, R., M. J. FOTER, AND K. H. LEWIS. 196. Time-temperature effects on salmonellae and staphylococci in foods. II. Behavior at warm holding temperatures. Thermal death time studies. Robert A. Taft Sanitary Eng. Center Tech. Rept. F 6-5. BAKER, E. E., AND L. S. MCCLUNG. 1939. Determination of the heat resistance of non-spore-forming bacteria. Food Res. 4 :21-29. BIGELOW, W. D. 1921. The logarithmic nature of thermal death time curves. J. Infect. Diseases 29:528-536. EVANS, F. R., AND H. R. CURRAN. 1943. The accelerating effect of sublethal heat on spore germination in mesophilic aerobic bacteria. J. Bacteriol. 46:513-523. FARKAS, D. F. 1962. Apparatus and nmethod for rapidly heating and cooling biological materials. Food Technol. 16:16. GILLESPY, T. G. 1951. Estimation of steriliing value of processes as applied to canned foods. I. Packs heating by conduction. J. Sci. Food Agr. 2:17-125. HALD, A. 1957. Statistical theory with engineering applications. John Wiley & Sons, Inc., New York. HALVERSEN, W. V., AND G. L. HAYS. 1936. The thermal death time of Clostridium botulinum spores at temperatures and ph values commonly encountered in home canning. J. Bacteriol. 32:466-467. LICCIARDELLO, J. J., AND J. T. R. NICKERSON. 1962. Effect of radiation environment on the thermal resistance of irradiated spores of Clostridiutm sporogenes P.A. 3679. J. Food Sci. 27: 211-218. REYNOLDS, H., AND H. LICHTENSTEIN. 1952. Evaluation of heat resistance data for bacterial spores. Bacteriol. Rev. 16:126-143. SCHMIDT, C. F. 1957. Thermal resistance of microorganisms, p. 831-884. In G. F. Reddish [ed.]. Antiseptics, disinfectants, fungicides and steriliation. Lea and Febiger, Philadelphia. WITHELL, E. R. 1942. The significance of the variation in shape of time-survivor curves. J. Hyg. 42:124-183.