Why Does Cronobacter sakazakii Survive for a Long Time in Dry Environments? Contribution of the Glass Transition of Dried Bacterial Cells

ABSTRACT To investigate the mechanism of adaptation of Cronobacter sakazakii to desiccation stress, the present study focused on the glass transition phenomenon of dried bacterial cells, using a thermomechanical technique. The mechanical glass transition temperature (Tg) of dried C. sakazakii cells per se, prepared by different drying methods (air drying and freeze-drying) and with different water activity (aw) levels (0.43, 0.57, 0.75, and 0.87), were determined. In addition, we investigated the survival of two strains of C. sakazakii (JCM 1233 and JCM 2127) prepared by different drying methods under different storage temperatures (4, 25, and 42°C) and aw conditions (0.43 and 0.87). While the Tg of the air-dried C. sakazakii cells increased as the aw decreased, the freeze-dried C. sakazakii cells showed an unclear aw dependency of the Tg. Air-dried C. sakazakii cells showed a higher Tg than freeze-dried C. sakazakii cells at an aw of <0.57. Freeze-dried C. sakazakii cells were more rapidly inactivated than air-dried cells regardless of the difference in aw and temperature. The difference between the Tg and storage temperature was used as an index that took into consideration the differences in the drying methods and aw levels. As the difference between the Tg and storage temperature increased to >20°C, the dried C. sakazakii cells survived stably regardless of the drying method. In contrast, when the difference between the Tg and storage temperature was reduced to <10°C, the viable cell numbers in dried C. sakazakii cells were quickly decreased. Thus, the Tg is a key factor affecting the desiccation tolerance of C. sakazakii. IMPORTANCE The mechanical glass transition temperature (Tg) of dried Cronobacter sakazakii cells varied depending on differences in drying methods and water activity (aw) levels. Because the Tg of the dried bacterial cells varied depending on the drying method and aw, the Tg will play an important role as an operational factor in the optimization of dry food processing for controlling microbial contamination in the future. Furthermore, the differences between the Tg and storage temperature were introduced as an integrated index for survival of bacterial cells under a desiccation environment that took into consideration the differences in the drying methods and aw levels. As the difference between the Tg and storage temperature decreased to <10°C, the dried C. sakazakii cells were inactivated quickly, regardless of the drying methods. The relationship between Tg and storage temperature will contribute to understanding the desiccation tolerance of bacterial cells.

IMPORTANCE The mechanical glass transition temperature (T g ) of dried Cronobacter sakazakii cells varied depending on differences in drying methods and water activity (a w ) levels. Because the T g of the dried bacterial cells varied depending on the drying method and a w , the T g will play an important role as an operational factor in the optimization of dry food processing for controlling microbial contamination in the future. Furthermore, the differences between the T g and storage temperature were introduced as an integrated index for survival of bacterial cells under a desiccation environment that took into consideration the differences in the drying methods and a w levels. As the difference between the T g and storage temperature decreased to ,10°C, the dried C. sakazakii cells were inactivated quickly, regardless of the drying methods. The relationship between T g and storage temperature will contribute to understanding the desiccation tolerance of bacterial cells.
KEYWORDS desiccation tolerance, freeze-dry, glass transition temperature, storage temperature, water activity pathogens responsible for foodborne disease outbreaks related to dried foods (1)(2)(3)(4)(5)(6)(7). There are some reports showing that most of these bacteria can survive for long periods, such as more than a year, under desiccation (1)(2)(3). For example, C. sakazakii, Salmonella spp., and E. coli O157:H7 can persist for more than 12 months in powdered infant formula (4,5). Although storage temperature and water activity (a w ) have been thought to be critical factors for the long-term survival of bacteria in dry environments, the governing environmental or intrinsic factors underpinning the survival of bacteria have not been elucidated thus far (3,6).
Identification of the critical and governing environmental or intrinsic factors associated with the desiccation tolerance of bacteria is expected to contribute significantly to the control of foodborne pathogens contaminating and surviving in dry foods for long periods (6). The governing factors can then be applied for optimization of dry food processing and/or distribution. Although there have been some reports on evidence for factors that are protective to C. sakazakii cells under desiccation, such as trehalose accumulation (8) and the presence of capsules (9)(10)(11), the mechanism underpinning bacterial survival in dry environments could be partly elucidated by the identification of previously unknown critical factors. Thus, the present study aimed to identify the factors governing the long-term survival of pathogenic bacteria in dry environments.
Storage temperature and water activity (a w ) cannot sufficiently explain bacterial desiccation tolerance, which has long been studied by many researchers (6,7,(11)(12)(13)(14)(15). These factors alone could not sufficiently explain properties like the fluidity of the bacterial cytoplasm. Thus, we introduced the concepts of glassy state and glass transition phenomenon as a third evaluation axis related to cryptobiosis (16). The cryptobiosis state has been observed in some organisms, such as tardigrades and African chironomids (17,18). Cryptobiosis is a metabolic state that confers a high survival rate on organisms by transforming them under harsh environments, such as in deserts. For the efficient preparation of probiotics that show some health effects, such as lactic acid bacteria, the glass transition phenomenon and control of the glass transition temperature (T g ) of freeze-dried lactic acid bacteria containing protectants have been considered additional and/or alternative control factors to temperature and a w (19). Accordingly, we hypothesized that the glass transition phenomenon might explain the survival of pathogenic bacteria under desiccation. In addition, we further focused on the relevance of the effect of the drying method on the survival and/or glass transition of bacterial cells. Because drying and glass transition are both highly related to the movement of water molecules in the material, the drying method might also have an impact on the desiccation tolerance of bacteria.
In the present study, we focused on C. sakazakii, which has long been a concern with regard to the safety of powdered infant formula, as a representative desiccationtolerant bacterium (20)(21)(22)(23)(24)(25)(26)(27). The objectives of this study were to determine and compare the mechanical T g of dried C. sakazakii cells prepared by different drying methods and with different a w levels, using a thermomechanical technique. In addition, the survival of C. sakazakii cells during storage under different a w levels and temperatures was investigated, and the relationship between storage temperature and T g in the survival of C. sakazakii is discussed.

RESULTS
Glass transition temperature of C. sakazakii cells. The relationship between the a w and T g of C. sakazakii cells differed significantly between the drying methods, as shown by the results in Fig. 1. While air-dried C. sakazakii cells exhibited a decrease in T g with increasing a w , freeze-dried C. sakazakii cells exhibited an unclear a w dependency of the T g . The T g values were significantly lower than those of air-dried samples at lower a w values, such as a w 0.43 (P = 0.00012 and 0.00031 for strains JCM 1233 and JCM 2127, respectively) and a w 0.57 (P = 0.00045 and 0.00021 for strains JCM 1233 and JCM 2127, respectively). The T g of air-dried C. sakazakii cells at an a w of ,0.57 was approximately 10 to 15°C higher than that of freeze-dried C. sakazakii cells. No significant difference in T g between the two drying methods was observed under high a w levels (.0.75; P . 0.05 for both strains). These trends were observed regardless of differences between the C. sakazakii strains, and there was no significant difference in the T g values at the same a w between the two strains.
Survival kinetics of C. sakazakii cells during storage. Storage temperature greatly influenced the survival kinetics of dried C. sakazakii cells regardless of differences in drying methods (air or freeze-dried), a w levels (0.43 or 0.87), and strains (JCM 1233 and 2127), as shown by the results in Fig. 2 and 3. The higher the storage temperature applied was, the faster the death rate. The fitted death rate parameters (b) of the survival kinetics clearly illustrated the trend of faster death at higher storage temperatures ( Table 1). The drying method also affected the survival kinetics of dried C. sakazakii cells. The surviving cell numbers of freeze-dried C. sakazakii decreased more rapidly than those of air-dried C. sakazakii regardless of the storage temperature, a w , and strain. In particular, marked differences in the survival kinetics of C. sakazakii cells stored at 25°C and a w 0.43 were observed between the drying methods for both strains. The survival kinetics of C. sakazakii cells stored at a w 0.87 did not exhibit a large FIG 1 Relationship between the water activity (a w ) and observed mechanical glass transition temperature (T g ) of air-dried (l) and freeze-dried (~) Cronobacter sakazakii strains JCM 1233 (a) and JCM 2127 (b). The results are expressed as the mean values 6 standard deviations from three independent experiments. Values with different letters at the same a w between the air-drying and freeze-drying methods were significantly different (P , 0.05). difference between the drying methods at any of the tested temperatures compared with the differences observed under storage at a w 0.43. The influence of storage temperature and drying method was pronounced at lower a w levels, such as a w 0.43. In summary, the results demonstrated that the higher the a w and storage temperature were, the faster bacterial death occurred, regardless of the difference in the drying methods.
Effect of the relationship between storage temperature and T g on the survival of C. sakazakii cells. We demonstrated the effect of the relationship between storage temperature and T g on the survival (log-cycle reduction after 21 days of storage) of dried C. sakazakii cells (Fig. 4). As the difference between T g and storage temperature increased to .20°C, slight reductions in viable cell numbers (1-to 2-log-cycle reductions) were observed regardless of differences in the drying methods and a w levels. In contrast, when the difference between T g and storage temperature decreased to ,10°C, the viable cell numbers of C. sakazakii were markedly decreased (8-to 9-logcycle reductions) regardless of differences in the drying methods and a w levels. The decreases in the viable cell numbers of C. sakazakii were dependent only on the relationship between T g and storage temperature and were independent of the drying method and a w level. The results illustrated in Fig. 4 clearly indicate that the difference between T g and storage temperature plays a key role in explaining the survival of dried C. sakazakii cells during storage at different temperatures and under different a w conditions, regardless of the drying methods.

DISCUSSION
Differences in the drying methods used for C. sakazakii cells resulted in differences in the T g (Fig. 1) and survival kinetics ( Fig. 2 and 3). The differences in the T g , as shown by the results in Fig. 1, could be attributed to the drying mechanism. The present study used air drying and freeze-drying as representative methods for dry food production. Since both methods are widely used for the production of various forms of dry food, the effect of those drying methods on the survival of dried bacterial cells would be applicable to various kinds of dry food. While the freeze-drying method removes solid water from bacterial cells by sublimination and results in a porous state, the air-drying method gradually removes water by evaporation. The desiccation speed, process, and phase transition of water might influence the amorphous components, and slow desiccation by air drying might result in a higher T g . Although the spray-drying procedure is also used in powder production, especially in the production of powdered infant formula, rapid evaporation by spray drying might have results similar to those of freezedrying. The effect of spray drying on the survival of dried bacterial cells should also be investigated in the future. Furthermore, the effect of the desiccation rate on the glass transition needs to be investigated to further our understanding of the mechanism underlying the differences in T g among different drying methods.
The survival kinetics of dried C. sakazakii cells in the present study were overall consistent with those from previous studies (5,(22)(23)(24). As the storage temperature and a w increased, the viable cell numbers of C. sakazakii decreased more rapidly. In contrast, a difference in the survival kinetics between air-dried and freeze-dried samples was observed, even at the same storage temperature and a w . Freeze-dried C. sakazakii cells were more rapidly inactivated than air-dried C. sakazakii cells ( Fig. 2 and 3). A previous study showed that fluidized bed drying, an air-drying method, supports better survival of Lactobacillus paracasei than freeze-drying because it causes reduced absorption  of water, which is attributed to the lower porosity and larger agglomerates (28). In contrast, fluidized bed drying significantly reduced the survival rates of Enterococcus faecium and Lactobacillus plantarum compared to freeze-drying (29). In addition, some viable bacteria were detected even after freeze-drying, and their survival rates varied from 8.0 to 96.7% over 20 years (16,30). Since the differences in bacterial survival between the drying methods have not been sufficiently investigated in detail thus far, there has been a paucity of research on the effect of the drying method on the viability of foodborne pathogens. The present study demonstrated the possibility of using T g as an alternative and integrated index for evaluating multiple effects of temperature, a w , and drying method on the survival of dried pathogenic bacterial cells. The importance of the difference between the environmental temperature and T g rather than T g itself was suggested as a critical factor affecting the physical state of amorphous polymers (19,31). The socalled Williams-Landel-Ferry (WLF) equation is one example of such a relationship (31). Measurement of the viscosity of simple mono-and disaccharides revealed that the viscosity decreased by approximately 4 to 6 orders of magnitude when the temperature increased to 20°C above the T g (32)(33)(34). The stability of freeze-dried Lactobacillus paracasei subsp. paracasei cells was demonstrated at storage temperatures that were more than 20°C above the T g (19). The effect of the difference between temperature and T g on the survival of C. sakazakii cells was demonstrated in the present study, as well as in previous studies, as mentioned above. When the difference between the T g of the dried bacterial cells and storage temperature is sufficiently large, such as .20°C, the dried bacterial cells survive stably for a long period. In contrast, when the difference between the T g of the dried bacterial cells and storage temperature becomes small, such as ,10°C, that is, if the T g is close to the storage temperature, the dried bacterial cells are inactivated quickly. The difference between the T g and storage temperature indicates whether the bacterial cells are in a stable glassy state or active rubbery state. Thus, the T g indicates whether the bacterial cells are tolerant of desiccation.
The protective effects of different kinds of sugars on the stability of bacterial survival were reported in previous studies (35,36). The studies on the relationship between T g and the stability of bacterial cells so far included different kinds of sugars with dried bacterial cells. Those studies might not evaluate the net response of bacterial cells to desiccation, because of the protective effect of added sugars. On the other hand, the present study demonstrated the response of dried bacterial cells themselves. The relationship between the T g and a w of C. sakazakii cells demonstrated in the present study would appropriately reflect the net physical response of bacterial cells to desiccation stress. This difference from the previous studies would contribute to further understanding of the mechanism of acquisition of desiccation tolerance of bacterial cells.
Because the T g of the dried bacterial cells varied depending on the drying method and a w , the T g will play an important role as an operational factor in the optimization of dry food processing for controlling microbial contamination in the future. Further investigation of the T g of dried microbial cells for other bacterial species and strains will be needed for generalization of the glass transition phenomenon in microorganisms. In addition, the glass transition of microbial cells in dry food environments should be examined. Furthermore, the mechanism underlying changes in the glassy state and the effect of drying speed on the glass transition and T g should be investigated. Nevertheless, the present study illustrated that the difference between the T g and storage temperature could be a useful index for prediction of the survival kinetics of dried bacterial cells during storage.

MATERIALS AND METHODS
Preparation of dried bacterial cells. C. sakazakii strains (JCM 1233 and JCM 2127) were obtained from Japan Collection of Microorganisms (Tsukuba, Ibaraki, Japan). The details of the strains used are available at https://www.jcm.riken.jp/cgi-bin/jcm/jcm_number?JCM=1233 and https://www.jcm.riken.jp/ cgi-bin/jcm/jcm_number?JCM=2127. The strains were maintained at 280°C in tryptic soy broth (TSB, Merck, Darmstadt, Germany). They were recovered from a frozen stock and cultured on tryptic soy agar (TSA; Merck, Darmstadt, Germany) plates. Since the objective of this study was to investigate the response to desiccation stress of dried bacterial cells per se, considerable quantities of bacterial cells were needed for physical and microbiological analyses. To efficiently prepare and collect bacterial cells in quantity and exclude an effect of specific nutrition, we used tryptic soy broth (TSB) as a representative and standard culturing medium. The cultures were incubated at 37°C for 24 h, and an isolated colony of each bacterium was transferred to 5 ml of fresh TSB at 37°C for 24 h; thereafter, 100 ml of cultured C. sakazakii cells was added to 800 ml of TSB and incubated at 37°C for 48 h. The cultured cells were centrifuged (3,000 Â g, 10 min), and the resulting pellets were washed with 30 ml of pure water. Then, the bacterial cell suspension with pure water was centrifuged once more with the same procedure as described above to obtain purified bacterial cell pellets.
The collected pellets of C. sakazakii cells were dried by one of two methods widely used for various kinds of foods in the food industry. One group of bacteria was subjected to air drying, which is based on the principle of diffusion. The bacterial cell pellets were placed in a biological safety hood at room temperature for 48 h to dry them completely. Another group of C. sakazakii cells was freeze-dried for 24 h in a freeze dryer (FDU-2200; EYELA, Tokyo, Japan) at 280°C and a pressure of ,10 Pa. Both groups of dried samples were ground and powdered manually using a sterile mortar and pestle.
Determination of the mechanical glass transition temperature (T g ). The powdered C. sakazakii cells were stored in airtight containers at four levels of relative humidity (% RH) and 4°C for 48 h to stabilize their a w (3). The % RH was adjusted using a saturated aqueous salt solution (43% RH, potassium carbonate; 57% RH, sodium bromide; 75% RH, sodium chloride; and 87% RH, potassium chloride). The RH and temperature in the airtight container were checked continuously using a thermorecorder (TR-72wf; T and D, Nagano, Japan). The a w of the bacteria was confirmed by a water activity meter (Aqualab 4TE; Decagon Devices, Pullman, WA, USA).
The T g of the a w -adjusted C. sakazakii cells was determined by thermal rheological analysis (TRA) as described in previous studies (37)(38)(39)(40) and as shown in Fig. 5. A dried bacterial cell sample (ca. 100 mg) was placed in the forming die (w = 3 mm) and compacted with a rheometer ca. 10 MPa. Subsequently, the sample was compressed at 7 MPa for 3 min and then heated at a rate of approximately 1.5°C/min until the temperature reached 120°C. Pressure-time data were collected with software attached to the rheometer. In parallel, a thermocouple was attached to the bottom of the forming die, and time-temperature data were collected every second using a datalogger. Accordingly, the relationship between pressure and temperature data during heating was determined. Since pressure reduction begins at the point at which the bottom temperature of the sample reaches the mechanical T g , the onset temperature of pressure reduction could be regarded as the T g of the sample. It is known that the onset T g determined by TRA is ;5 to 10°C higher than that determined by differential scanning calorimetry (39).
Survival kinetics of C. sakazakii cells under the selected temperature and relative humidity conditions. Changes in the viable cell numbers of dried C. sakazakii cells stored in airtight containers maintained at different relative humidity levels (43% and 87%) at 4°C, 25°C, and 42°C were determined over time. The surviving cell numbers of C. sakazakii were determined periodically, after 1, 2, and 3 weeks of storage. Triplicate trials were conducted. Fifty milligrams of dried C. sakazakii cells was mixed with 500 ml of 0.1% peptone water, and then the bacterial cell suspension was serially diluted 10-fold. The diluted samples at each dilution level were plated onto duplicate TSA plates and incubated at 37°C for 24 h, and then the number of colonies was determined.
Triplicate samples were collected at each sampling time. The colony counts for the triplicate samples of each bacterium at each sampling interval were transformed to log CFU/g values, and the values for the triplicate samples were averaged to represent the number of viable cells at each sampling time.
The Weibull model, one of the most frequently used survival models (3,6,41), was used as a representative model. This model is described by the following formula: where S(t) is the momentary ("instantaneous") survival ratio, b and n are parameters representing the survival rate and kinetics curve, respectively, and N(t) and N 0 are the momentary and initial cell counts, respectively. The initial cell counts (N 0 ) were assumed to represent the state of the samples immediately