Inactivation of Lactococcus lactis ssp. cremoris cells in a droplet during convective drying

Biochemical Engineering Journal 79 (2013) 46–56

Contents lists available at SciVerse ScienceDirect

Biochemical Engineering Journal journal homepage: www.elsevier.com/locate/bej

Inactivation of Lactococcus lactis ssp. cremoris cells in a droplet during convective drying Nan Fu a,b,∗ , Meng Wai Woo b , Cordelia Selomulya b , Xiao Dong Chen a,b,∗∗ a b

College of Chemistry, Chemical Engineering and Materials Science, Soochow University, Suzhou city, Jiangsu 215123, PR China Department of Chemical Engineering, Monash University, Clayton, Victoria 3800, Australia

a r t i c l e

i n f o

Article history: Received 27 February 2013 Received in revised form 15 June 2013 Accepted 21 June 2013 Available online xxx Keywords: Viability Kinetic parameters Modelling Food engineering Inactivation kinetics Culture preservation

a b s t r a c t Spray drying is a less costly alternative to freeze drying in the mass-production of active dry microorganisms, if the drying conditions could be optimized to preserve cell viability. As spray drying is akin to a black-box process, in this study we used an alternative approach of a single droplet drying to study how drying conditions affect the inactivation of bacterial cells. The inactivation histories of Lactococcus lactis ssp. cremoris were investigated at air temperatures of 70, 90, and 110 ◦ C. It was found that the viability of L. cremoris cells could be maintained at approximately the original level for extended drying durations (60–210 s), despite the high air temperatures. When plotted against droplet temperature Td , the inactivation rate kd at six drying conditions formed a general trend. An inactivation model was proposed to describe different inactivation histories under varied drying conditions. The description closely followed the experimental data, reported for the first time in literature. kd increased rapidly after Td passed a transition temperature range of 50–65 ◦ C, coinciding with the onset temperature for denaturation of bacterial ribosomes. Other environmental parameters affecting inactivation are discussed to better understand the integrated effects of multiple stresses experienced by bacterial cells during convective drying. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Active dry microorganisms are important food ingredients for a range of products with high trading values [1,2]. These dry cultures have prolonged shelf-life, more stable properties, and are easier to transport compared to liquid cultures, while ideally still retaining similar bioactivity. Examples of commercially available dry cultures include: dry starter cultures for the dairy industry, mainly from lactic acid bacteria [3]; dry probiotics in health care and pharmaceutical products [4]; dry yeasts for fermentations in breweries and bakers’ industries [5]; and bio-preservatives inhibiting the growth of other microorganisms to replace antibiotics in feed products [6]. Traditionally, dry cultures are produced by freeze drying, which requires relatively long processing times and high energy consumption [7,8]. Recently, spray drying as a conventional dehydration approach to process heat-sensitive materials into

∗ Corresponding author at: College of Chemistry, Chemical Engineering and Materials Science, Soochow University, 199 Ren-Ai Road, Suzhou Industrial Park, Suzhou, Jiangsu 215123, PR China. Tel.: +86 512 65883267; fax: +86 512 65883267. ∗∗ Corresponding author at: College of Chemistry, Chemical Engineering and Materials Science, Soochow University, 199 Ren-Ai Road, Suzhou Industrial Park, Suzhou, Jiangsu 215123, PR China. Tel.: +86 512 65882767; fax: +86 512 65882767. E-mail addresses: [email protected], [email protected] (N. Fu), [email protected] (M.W. Woo), [email protected] (C. Selomulya), [email protected] (X.D. Chen). 1369-703X/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bej.2013.06.015

powders has attracted interest for the production of active dry microbial cultures [1,9]. In reported studies, bacteria and yeast could survive a spray drying process with high survival ratios of up to 80–100%, which compares favorably with the freeze drying process [10–13]. During spray drying, a feed solution containing microorganisms and often a carrier material to protect these microorganisms is first atomized into droplets and then sprayed into a drying tower. Hot air flows, often more than 200 ◦ C, are used in the drying tower to quickly remove the moisture from the droplets. Evaporation occurs rapidly as a result of the high air temperature and the large contact area, producing dried particles of the carrier where the microorganisms are encapsulated. During the simultaneous heating and dehydration processes, microbial cells would suffer from multiple stresses such as thermal [14], osmotic [15], and oxidative stresses [16], which can lead to irreversible cell death. Fu and Chen [1] discussed nine extrinsic parameters affecting the survival of microbial cells during convective drying. Briefly, they can be grouped as (i) drying air properties (air temperature and humidity); (ii) kinetic drying parameters (material temperature Td , initial moisture content X0 , moisture removal rate −dX/dt, temperature variation rate dT/dt and exposure time t); and (iii) carrier properties (carrier composition and the location of cells inside the carrier). A reliable mathematical model correlating the inactivation of microbial cells to these parameters would not only optimize individual spray drying operations, but also contribute to understandings of

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

Nomenclature Ed k0 kd N N0 t Td X

activation energy for deactivating cells (kJ/mol) a pre-exponential factor rate constant of cell deactivation number of viable cells (cfu/mL) initial number of viable cells (cfu/mL) time of drying (s) droplet temperature (◦ C or K) average moisture content on dry basis (kg/kg)

the effects of each factor. Previous reported inactivation models often relate the cell death rate to the dryer outlet temperature [17–19]. However, the activation energy of the cell inactivation calculated in this manner would be inaccurate, as droplets/particles inside the dryer usually experience wide temperature and moisture variations [1,18]. Typically average values for all the droplets are considered in previous studies, which inhibits the accurate understandings of drying conditions on cell viability. The sub-cellular mechanism of spray drying to deactivate microorganisms is also poorly understood and constitutes object of further studies [20]. Droplets during spray drying would experience different drying histories due to the differences in initial sizes and droplet trajectories [21,22]. This wide thermal and moisture history variability is a challenge to interpret, making the development of inactivation models for individual droplets a very difficult task. Single droplet drying (SDD) is an established experimental technique to study droplet drying behaviour in a controlled environment mimicking spray drying conditions [23,24]. The technique is capable of accurately measuring changes in the droplet weight, temperature and diameter as drying progresses. This kinetic data is essential for developing models [25] as well as to validate CFD simulations of spray drying processes [26–28]. SDD has been shown to produce powders with the same morphological properties to spray dryers [29–31]. For these reasons, SDD technique has hence become a powerful tool to study any droplet-drying-related phenomenon, such as those occurring during spray drying. Previously, SDD systems have been employed to study the inactivation kinetics of different bacteria [32] and effects of various carriers [33]. In the present study, the inactivation of a model microorganism under different drying conditions was monitored to investigate the correlation between cell inactivation and droplet drying histories. The model microorganism used was Lactococcus lactis ssp. cremoris, a typical Gram-Positive cocci used as a starter culture in the cheese industry. An inactivation model was developed to describe inactivation histories under varied conditions, focusing also on the effects of extrinsic parameters on cell survival. 2. Materials and methods 2.1. Microorganism and culture conditions The culture of Lactococcus lactis ssp. cremoris was maintained on standard M17 (Oxoid CM0817) agar plate at 4 ◦ C with subculture to fresh media every seven days. For single droplet drying (SDD) experiments, 10 mL of M17 culture medium were inoculated with the strain and then incubated in a 30 ◦ C stationary incubator for 24 h prior to drying experiments. 2.2. Preparation of cells for drying experiments After 24 h incubation, L. cremoris cells were collected by centrifugation at 10,000 rpm, 25 ◦ C for 10 min. The resultant cell pellets were re-suspended in 10 mL carrier solution, in order for the viable

47

cell concentration in the carrier solution to be equivalent to that of the 24 h culture. The carrier used was reconstituted skim milk (RSM), with 10 or 20 wt% solids content. Skim milk powder was purchased locally, consisting of 37.97 wt% protein, 1.34 wt% fat, 58.96 wt% sugar, and 1.73 wt% minerals according to the product specification. The powder was reconstituted in Milli-Q water (QGARD00R1 Milli-Q system, Millipore, Australia) and sterilization was effected by autoclaving at 105 ◦ C for 10 min. The RSM carrier solution with the re-suspended cells was placed in an ice bath to prevent undesired cell proliferation during the SDD experiments. The viable cell concentration in the carrier solution was checked immediately after preparation and after 4 h of preparation, to ensure that there was no cell reproduction and that the viable cell concentration was similar to freshly prepared samples. 2.3. Single droplet drying (SDD) experiment The working principle and experimental set-up of the SDD system used in the present study have been described elsewhere [34,35]. Briefly, an air stream with controlled temperature, velocity and humidity was used for drying of a suspended single droplet in a confined chamber. During drying, the single droplet was suspended at the tip of a specially-made fine glass filament; while changes in droplet temperature and mass were measured using different droplet suspension modules in separate drying runs with identical conditions. A 5 ␮L GC micro-syringe was used to generate single RSM droplets containing L. cremoris cells for each run. The initial size of each droplet was 2 ± 0.05 ␮L, where the error was taken as half of the minimal graduation of the micro-syringe. After each drying run, the syringe was rinsed three times, with Milli-Q water, 70% (v/v) ethanol, and sterilized Milli-Q water, respectively, to avoid cross-contamination. Drying conditions used in the present study are summarized in Table 1. Under these drying conditions, Reynolds number was around 55–65 and the Nusselt number was around 6.2–6.6 at the beginning of the drying experiment. The value of both dimensionless numbers decreased as drying progressed, approaching 22–28 and 4.7–5.0, respectively, at the final stage of drying. For each drying condition, droplet mass, temperature and L. cremoris viability were monitored as drying progressed. Each experiment was repeatedly carried out twice, and results reported here are the averages of duplicate experiments. The droplet weight and temperature were recorded following a previously described procedure [34,35]. Viability measurements were performed on repeated SDD runs under identical drying conditions, but stopped at different times: at 0, 15, 30, 45, 60, 90, 180, 210, 240, 270, and 300 s. The resulting semi-dried particle, with an initial volume of 2 ␮L, was dissolved in situ with 2 mL of M17 diluents without removing it from the suspending glass filament. The residual viable cells in each semi-dried particle were estimated using the standard plate count procedure, where M17 diluents were used for serial dilutions. The M17 diluents were 1/10 of the strength of the original M17 medium without lactose. As such, each viable cell data corresponded to a separate SDD process. Duplicate experiments of viability measurement were independently carried out using a fresh 24 h culture. 2.4. Visualisation of dried bacterial cells SEM imaging was conducted for RSM particles with an initial solids content of 10 wt%. These particles were dried for 6 min at 70 ◦ C, 5 min at 90 ◦ C, and 4 min at 110 ◦ C, to achieve a relatively complete water removal. At the end of each drying run, the formed particles were removed from the suspending glass filament and carefully divided into smaller pieces for SEM imaging. At least two specimens were used for each drying temperature. Samples were

48

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

Table 1 Drying conditions used in the current study. Air flow condition

Temperature: Velocity: Moisture content:

70 ◦ C, 90 ◦ C, 110 ◦ C 0.75 m/s 0.0001 kg/kg

Droplet parameters

Carrier material: Initial carrier solids contenta : Initial droplet size: Initial viable cell concentrationb :

Reconstituted skim milk 10 wt%, 20 wt% 2 ␮L 3 ∼ 5 × 108 cfu/mL, the same as the 24 h culture

a b

The unit of solids content is mass percentage (wt%). The unit of viable cell concentration is colony-forming unit per milliliter (cfu/mL).

fixed on an aluminum sample stub using a conducting carbon tape and then sputter-coated with ∼ 2 nm gold-palladium to produce a conductive surface. Secondary electron SEM images were recorded using a JEOL 7001F (Jeol Co. Ltd., Japan) operating at 5 kV. 2.5. Calculation of inactivation rate constant kd and activation energy Ed The inactivation of microbial cells by thermal processes is usually described with first order kinetics, as follows [32,36]: d(N/N0 ) = −kd (N/N0 ) dt

(1)

where N is the number of viable cells (cfu/mL) at time t(s), N0 is the initial number of viable cells (cfu/mL), and kd is the inactivation rate constant under the given conditions. In the present study, Eq. (1) was used to calculate kd with the experimentally determined N, N0 at t. Assuming that heat is the sole lethal factor that deactivates cells, kd can be correlated to droplet temperature Td (K) using the Arrhenius equation [32]:

 E  d

kd = k0 exp −

RTd

(2)

where k0 is a pre-exponential factor, Ed is the activation energy for deactivating cells, and R is the gas constant. Eq. (2) indicates that the plot of ln(kd ) versus 1/Td is expected to form a straight line with a slope of (–Ed /R), which is used to estimate Ed in the present study. 3. Results 3.1. Drying kinetics of bacteria-loaded RSM droplets and the inactivation of L. cremoris cells The drying kinetics of RSM droplets containing L. cremoris at different conditions are shown in Fig. 1, together with the corresponding inactivation curves of bacterial cells. The viable cell concentration data are plotted on logarithmic scale in the figure. Under all conditions, the viability of L. cremoris was maintained at approximately the initial level for extended durations (60–210 s) after the start of drying. At a drying air temperature of 70 ◦ C (Fig. 1a and A), a decrease of viable cells number was only observed in the final stage of drying, where the removal of droplet moisture (X) achieved a minimal rate. The loss of cell viability was around one order of magnitude for 10 wt% RSM carrier and half order of magnitude for 20 wt% by the end of the drying experiment (at 270 and 300 s, respectively). When the drying temperature was increased to 90 ◦ C (Fig. 1b and B), the viable cell concentration remained similar to the initial level during the initial 120 s of drying. Subsequently, it showed a significant decrease on the log scale, with around three orders of magnitude of cells deactivated during 120 and 250 s of drying. The deactivation rate was faster with 10 wt% RSM than that of 20 wt% at 90 ◦ C (Fig. 1b and B). A noticeable tailing effect was observed after the viable cell concentration reached 105 cfu/mL. At a drying temperature of 110 ◦ C, extensive cell death

rapidly occurred after 90 s at 10 wt% RSM carrier and 60 s at 20 wt% (Fig. 1c and 1C). The viable cell level in the 10 wt% RSM droplets was reduced from 108 to 105 cfu/mL in 60 s (from 90 to 150 s of drying). For 20 wt% RSM droplets, the viable cell concentration dropped to 105 cfu/mL at 120 s. This trend was different to that observed at lower temperatures, 70 and 90 ◦ C, whereby the 10 wt% RSM carrier resulted in more cell inactivation than with 20 wt% carrier. Observing cell inactivation and drying kinetics at each condition demonstrated that inactivation was closely related to the temperature of the semi-dried RSM particles. At the initial stage of drying, the droplet/particle temperature was still low due to evaporative cooling effect [21,37], in accordance to the minor loss of cell viability. As drying progressed, the droplet/particle temperature rose faster, accompanied with a significant decrease of viable cells. 3.2. The correlation between droplet temperature and cell inactivation It has been suggested that droplet temperature Td and droplet moisture content X are two major factors affecting the inactivation of microorganisms during convective drying [33,36]. To investigate the effect of each factor, the inactivation rate constant kd was calculated at each sampling time-point (15, 30, 45. . .300 s) using Eq. (1) and compared to the Td and X of the time. As cell death was assumed to occur from the commencement of drying and was a continuous process, the viability curves in Fig. 1 were then curve-fitted to determine d(N/N0 )/dt. The fitting curves are shown in Fig. 2 on a normal scale, while experimental data are also included for comparison. The tailing effects observed at 90 ◦ C were not considered during fitting, as the focus of the present study was to elucidate the reason for transition from the initial plateau of cell viability maintenance to the rapid cell inactivation at the later stage of drying. Besides the effects of drying temperature reported in the present study, the effects of carrier composition had also been investigated (unpublished results). The only condition showing notable tailing effect was 90 ◦ C with RSM as carrier. The mechanisms governing this phenomenon are not yet understood. More work is required to elucidate the tailing effect, as it poses an important challenge for pathogen inactivation in the food industry. Fig. 3a and b shows the correlation when kd is plotted against the corresponding Td for all six processing conditions studied. A general trend is observed with low kd values at a low temperatures <50 ◦ C, a transition point at 50–65 ◦ C followed by a dramatic increase at higher temperatures above 65 ◦ C. The fact that the six correlations from different conditions could form a single trend indicated that Td was a main factor deactivating L. cremoris cells at the tested droplet drying conditions. In Fig. 3c, ln(kd ) is plotted against 1/Td to determine the activation energy Ed of the cell inactivation process. Data from different drying conditions form one trend with clear linearity: 1 R2 = 0.959 (3) Td Hence, the value of ln(k0 ) was 33.28 ± 2.22 (95% confidence interval) and the value of (–Ed /R) was −12.7 ± 0.72 (95% confidence

ln(kd ) = 33.28 + (−12.7)

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

49

Fig. 1. Drying kinetics of reconstituted skim milk droplets carrying L. cremoris and the corresponding inactivation kinetics of the bacteria. (a)–(c) Droplets containing 10 wt% initial solids dried at 70, 90 and 110 ◦ C, respectively; (A)–(C) Droplets containing 20 wt% initial solids dried at 70, 90 and 110 ◦ C, respectively. : Droplet moisture;  droplet temperature; 䊉 viable cell number.

interval). The resulting activation energy is 105.59 ± 5.99 kJ/mol (95% confidence interval). According to Eq. (2), the correlation between kd and Td can be described using the following equation:



kd = 2.84 × 10

14

exp

1.0559 × 105 − RTd



RSM as carrier at various drying conditions. To verify the accuracy of this model, N at each second of drying was calculated and compared to the measured data at different conditions (Fig. 4). N was calculated as followed:

(4)

The equation description compared favorably to experimental results as shown in Fig. 3a (R2 = 0.908). The increase of kd occurred at the temperature range of 50–65 ◦ C was well described. As the model was developed from a range of conditions with different drying air temperatures and initial solids contents, Eq. (4) could potentially describe the inactivation of L. cremoris cells with

1. With a measured Td , kd was calculated at 1 s internal using Eq. (4). 2. With a known initial N/N0 = 1, d(N/N0 )/dt was determined using Eq. (1). Then the survival ratio N1 /N0 at t = 1 s can be calculated. 3. Repeating step 2, the N/N0 history at each drying conditions was updated; then the viable cell number N at each second was determined with a known N0 .

50

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

◦

Fig. 2. Comparison of measured cell inactivation kinetics to fitting results. (a) 70 C; (b) 90 ◦ C; (c) 110 ◦ C. Closed symbols: 10 wt% RSM; open symbols: 20 wt% RSM; straight line: fitting for 10 wt% curve; dotted line: fitting for 20 wt% curve.

Fig. 3. The correlation between inactivation constant kd and droplet temperature Td when L. cremoris cells were dried with reconstituted skim milk as carrier. (a) Normal scale of ordinate from −0.01 to 0.29; (b) Reduced scale of ordinate from −0.01 to 0.0; (c) Ln(kd ) vs. reciprocal Td . Squares: 70 ◦ C; triangles: 90 ◦ C; cycles: 110 ◦ C; closed symbols: 10 wt% RSM; open symbols: 20 wt% RSM.

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

51

Fig. 5. The correlation between the inactivation constant kd and droplet moisture content X when L. cremoris cells were dried with reconstituted skim milk as carrier. Squares: 70 ◦ C; triangles: 90 ◦ C; cycles: 110 ◦ C; closed symbols: 10 wt% RSM; open symbols: 20 wt% RSM.

Fig. 4. Description of cell inactivation kinetics using Eq. (4). Squares: 70 ◦ C; triangles: 90 ◦ C; cycles: 110 ◦ C; closed symbols: experimental data; open symbols: predicted data using Eq. (4).

It can be seen from Fig. 4 that the predicted cell inactivation histories closely followed experimental data at all six conditions. The initial plateau of cell viability maintenance was well described, together with the transition to the rapid loss of cell viability at a later stage of drying. Nevertheless, when developing Eq. (4), the tailing effects were not taken into account; as a result, Eq. (4) was unable to describe the tailing effects for both initial solids contents at 90 ◦ C, showing the limitation of the first order kinetics. 3.3. The effects of other extrinsic parameters on cell survival The effect of droplet moisture content X on cell survival was also investigated by plotting kd against the corresponding X at each sampling time-point in Fig. 5. kd remained low at high X range (from 9 to 2 kg/kg) for all processing conditions, corresponding to the initial plateau of cell viability in Fig. 1. Along with further moisture removal, kd started to increase; the drying temperature of 110 ◦ C showed the highest kd values amongst the three temperatures, followed by the 90 ◦ C. For each temperature, the kd curve of 20 wt% carrier is located slightly below that of 10 wt%, indicating that at the same moisture content, the inactivation rate was higher with 10 wt% than with 20 wt%. Examining Fig. 1, at 70 and 90 ◦ C, the use of 20 wt% RSM resulted in a slightly higher cell survival ratio than 10 wt% at the later stage of drying. In contrast, at 110 ◦ C the viable cell level in 20 wt% RSM carrier reached a level of 105 cfu/mL

at 120 s, earlier than 10 wt% carrier (the viable cell level reached 105 cfu/mL at 150 s), indicating a more efficient inactivation. To explore the mechanisms behind these observations, droplet temperature histories at different conditions were compared (Fig. 6), plotted against the time of drying and the corresponding X. Fig. 6a shows that Td of 20 wt% RSM droplets at the initial drying stage was higher than that of 10 wt%, in accordance with a faster cell viability loss in Fig. 2 (viability loss was generally within 0.3 order of magnitude). When Td passed the transition range (50–65 ◦ C), the Td of 10 wt% exceeded that of 20 wt%, causing a faster cell inactivation. Consequently, a higher survival ratio after drying was achieved with 20 wt% carrier at 70 and 90 ◦ C. At a drying temperature of 110 ◦ C, Td of 20 wt% RSM reached 60 ◦ C after approximately 58 s, in comparison to 73 s for 10 wt% droplet (Fig. 6a). Td of 10 wt% then exceeded that of 20 wt% at around 120 s, although at that stage Td of both solids contents had reached >90 ◦ C. Therefore, before Td of 10 wt% exceeded that of 20 wt%, the semi-dried particle of 20 wt% had been exposed to the high temperature range (60–90 ◦ C) for more than 60 s; three orders of magnitude of cells had been deactivated as a result (Fig. 1). It is possible that the kd of 10 wt% would exceed that of 20 wt% after 120 s due to the higher Td . However, this cannot be verified from the current data since residual viable cells in 20 wt% particles after 120 s were too low to be detected. In addition, Fig. 6b, could further explain the protection given by 20 wt% carrier at the lower X range (<1.8 kg/kg), which had a lower Td than 10 wt% carrier at the same moisture content. Hence, the enhanced protective effects of 20 wt% compared to 10 wt% RSM carrier on cell survival at 70 and 90 ◦ C can be attributed to its ability to keep Td at a comparatively lower value at the final stage of drying. On the other hand, the heating-up period of 20 wt% droplets was earlier than that of 10 wt% droplets, adversely affecting cell viability. Besides Td , X and exposure time t, rate changes of droplet temperature and moisture content were also considered of great importance affecting cell inactivation rate during a heat treatment [38,39]. The temperature variation rate (dT/dt) and moisture removal rate (−dX/dt) curves of the six tested conditions are compared in Fig. 7, plotted against the droplet moisture content (X). The highest values of both −dX/dt and dT/dt were achieved at the initial drying stage (the right end of abscissa), where the loss of cell viability was minimal. In other words, high values of −dX/dt and dT/dt at this stage did not lead to much mortality. When drying was just

52

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

Fig. 6. The droplet temperature histories of reconstituted skim milk droplets carrying L. cremoris cells dried under the six conditions, (a) when plotted against the time of drying; (b) when plotted against droplet moisture content X. Squares: 70 ◦ C; triangles: 90 ◦ C; cycles: 110 ◦ C; closed symbols: 10 wt% RSM; open symbols: 20 wt% RSM.

started, L. cremoris cells were present in an aqueous environment with a comparatively low Td (approximately 18–33 ◦ C). Then the moist environment gradually dried up. It is possible that during the transition of moist to dry, bacterial cells might become more sensitive to the changing rates of Td and X, as pointed out previously [20]. The lower values of −dX/dt and dT/dt attained with the 20 wt% carrier than with 10 wt% (Fig. 7) could be another reason for the better protection.

Fig. 7. Changes in (a) moisture removal rate and (b) temperature variation rate of RSM droplets carrying L. cremoris cells subjected to drying under the six conditions. Squares: 70 ◦ C; triangles: 90 ◦ C; cycles: 110 ◦ C; closed symbols: 10 wt% RSM; open symbols: 20 wt% RSM.

However, it was unlikely for any viable cells to survive drying temperatures of 90 and 110 ◦ C, although the cell morphology was still maintained. The fines holes on the cell wall of bacteria dried at 90 and 110 ◦ C were possibly an indication of cell death (Fig. 8B and C). Excessive heat would simultaneously damage multiple cellular sites, while the formation of such holes could be due to the destruction of peptidoglycan, a major composition of cell wall of Gram-Positive bacteria, under such high temperatures.

3.4. Surface morphologies of dried L. cremoris cells

4. Discussion

To further understand the effects of air temperatures on L. cremoris cells subjected to convective droplet drying, the morphology of dried bacterial cells at different drying conditions was captured using SEM imaging. Fig. 8 shows that these dry cells were encapsulated in the dried carrier, i.e., skim milk particles. At all temperatures, there were dried L. cremoris cells identified with characteristics coccus shape and chain arrangement. According to results shown in Fig. 1, these dried cells may still be viable upon rehydration when the drying temperature of 70 ◦ C was used.

4.1. Comparison of cell inactivation in the present study to spray drying results Heat inactivation of pathogenic microorganisms in food environment has been extensively studied [40]. A droplet drying process, as in the single droplet drying processes or in spray drying, differs from conventional heat treatments mainly due to the rapid change of droplet moisture content and the comparatively short residence time (usually around tens to hundreds seconds in

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

53

Fig. 8. SEM images of L. cremoris cells dried with 10 wt% reconstituted skim milk as carrier. Images were taken at the cross-section of broken milk particles. (a)–(c) Magnification of 20,000–25,000 times; (A)–(C) Magnification of 70,000–100,000 times. The photos were colored with adobe photoshop to improve clarity.

comparison to over 30 min). The short time that microbial cells are exposed to the hot air may protect cells from being over-heated, although the rapid change rate of moisture content may be detrimental to cellular integrity [38]. In addition, droplets being dried would also experience large temperature variations while being heated up. Changes in droplet moisture content and temperature are interrelated in this process. Hence, there is a need to establish an understanding on cell inactivation behaviour associated with such a fast dehydration process, if one is to form a clear understanding of approaches that could potentially protect the microbial cells during spray drying. In this study, the inactivation of a model organism, Lactococcus lactis ssp. cremoris, in suspended single droplets were evaluated

against drying kinetics they experienced under different drying conditions. L. cremoris is a mesophile microorganism with optimum growth temperature between 25 and 30 ◦ C. All three drying temperatures used, viz., 70, 90 and 110 ◦ C, could cause irreversible heat damage to these bacterial cells. However, data showed that cell viability could be well maintained at approximately the initial level as long as the droplet temperature was kept low. The lowest drying temperature of 70 ◦ C only incurred viability loss of around one order of magnitude when drying approached the final stage. This result was in good agreement with previous spray drying studies. In spray drying studies, the outlet temperature was assumed as the main factor affecting cell survival [17,41,42]. A higher outlet temperature led to lower cell survival ratio, while an outlet

54

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

temperature range from 40 to 60 ◦ C generally maintained cell viability after spray drying [11,12,43]. During spray drying operations, the outlet temperature is a partial indication of the temperature distribution of the dryer. Therefore, the utilization of a low outlet temperature could protect cells from being over-heated to a large extent, as reported in previous literature investigating spray drying of microorganisms. On the other hand, the temperature history of a droplet during spray drying is affected by the drying air humidity, flow rate, inlet temperature and droplets’ trajectories [1,33]. The outlet temperature is only one parameter among many that determine the temperature inside the droplets. Estimating the cell survival ratio based solely on the outlet temperature may introduce errors with poor reproducibility. The inactivation histories at 90 and 110 ◦ C in the present study demonstrated the possibility of maintaining the bacterial viability at high air temperatures, as long as the droplet temperature remained low. Before the occurrence of extensive viability loss, the total viability loss was less than 0.3 order of magnitude at all four conditions tested (Fig. 2). This observation explained why the outlet temperature is considered a more important factor than the inlet temperature affecting cell inactivation, a common observation made in previous spray drying studies [19,44,45]. At the initial drying stage, heat loss in the droplet due to water evaporation was compensated by heat convection from the hot air, keeping Td at around the wet-bulb temperature range. At this stage Td was usually lower than 50 ◦ C, depending on the dry-bulb temperature and the relative humidity of air [21,37]. As drying progressed and droplets approached the exit of spray dryer, the evaporative cooling effect of the semi-dried particle was no longer able to keep the particle temperature at the low temperature range, associated with an increase of Td (Fig. 6a). The extent of temperature rise at this stage was affected by the outlet temperature used. Therefore, controlling the over-heating at the final stage of drying is of paramount importance to optimize spray drying conditions to preserve microbial cells. The results in the present study also suggest that two-stage drying would be a good approach to produce active dry microorganisms. Spray drying could be firstly employed to remove the majority of moisture. In the second stage, other drying approaches with milder drying conditions could be adopted, to slowly dry these particles to a lower water activity while maintaining a high cell survival. The activation energy of cell deactivation process reported in previous spray drying studies is generally in the range of 28–75 kJ/mol [17–19], lower than the value obtained in the present study (105.59 kJ/mol). It has been suggested that the Ed calculated with the outlet temperature is useful for comparison between spray drying studies, but is not an absolute value [18]. Ed reported in the present study, taking into account the Td histories at different conditions, should be closer to the true value. Some previous studies have attempted to correlate the inactivation of microbial cells to drying histories of cells, using similar SDD concepts [32,33] or an infrared drying device [36]. In the study of Ghandi et al. [33], the Ed to deactivate Lactococcus lactis was reported to be 181.3 kJ/mol, but the tracking of cell inactivation histories was not carried out. In other studies [32,36], Ed was correlated to Td , X, −dX/dt and dTd /dt using a number of modelling approaches. The changing rates of Td and X were reported to affect the inactivation of bacterial and yeast cells to some extent; however the trends predicted by the models still had relatively large discrepancies from experimental results. 4.2. Extrinsic factors influencing cell inactivation during droplet drying Amongst the extrinsic factors to be considered for cell viability maintenance during spray drying [1], the initial moisture content X0 (also known as initial water activity aw,0 ) and the composition

of carrier could influence the intrinsic stress tolerance of microbial cells to some extent [46–48]. Both air temperature and humidity, as well as the location of cells inside the carrier, would firstly alter the droplet drying parameters (Td , X, −dX/dt, dT/dt and t) which then impact on cell inactivation. Therefore, a feasible way to evaluate the effects of these three parameters is to establish the correlation between the inactivation of microbial cells and droplet drying kinetics at first, and then explore how the droplet drying kinetics is affected by these parameters. The results in the present study showed that Td was the most important factor governing the survival of L. cremoris cells during droplet drying with an elevated air temperature (Figs. 3 and 6). In contrast, the effect of X on cell inactivation appeared minor (Fig. 5). The negligible effect of X, to some extent, could be a result of the high air temperature used in the present study. Teixeira et al. [49] suggested that heating at 64 ◦ C and 65 ◦ C may lead to different sites of cellular damage. At 64 ◦ C, the damage at cytoplasmic membrane was the main reason for cell inactivation, whereas cell wall and proteins would be damaged at 65 ◦ C causing cell death. Hence it is possible that microbial cells exhibit different inactivation mechanisms at different air temperatures during drying. The mortal effects of dehydration and osmotic stress may play a more important role on deactivating cells when a low air temperature is used. As far as the effect of –dX/dt was concerned, it was found that at the final drying stage a higher –dX/dt value coincided with a high inactivation rate (Fig. 7b); however, the high –dX/dt values at the initial drying stage barely showed any mortal effect towards cell viability. The six dT/dt curves all showed a peak at the middle stage of drying (Fig. 7a), corresponding to the heating-up period of the semi-dried particles. Such peak was not observed for kd changes during drying. Finally, a longer exposure time t to lethal environmental conditions undoubtedly led to more severe inactivation of cells. In droplet drying processes, a shorter exposure time t is usually accompanied with more rapid changes in moisture content (–dX/dt) and droplet temperature (dT/dt). For example, the drying rate in a spray drying process could be considerably higher than that in SDD processes. It has been suggested in heat treatment studies that the kinetics of temperature variation and osmotic variation affect the inactivation of yeast and bacterial cells to a large extent [38,39]. The studies of Li et al. [32] and Huang et al. [36], which correlated the inactivation of microorganisms to the drying histories that cells experienced, also reported a similar observation. However, in the present study, the high –dX/dt and dT/dt values at the initial stage of drying were associated with minimal viability loss of bacterial cells. It is possible that the detrimental effects of high –dX/dt and dT/dt values were concealed by the overwhelming high temperature effects at the last stage of drying. Therefore, there could be different mechanisms involved for the inactivation of microorganisms during convective drying and during heat treatment. From such understandings on the relationship between droplet drying and bacterial inactivation, the effects of other extrinsic parameters on cell survival can be more easily interpreted. For example, the enhanced protection on cell viability shown by 20 wt% RSM over 10 wt% was attributed to the comparatively low Td at the same moisture content when X was reduced to less than 1.8 kg/kg (Fig. 6b). 4.3. The sub-cellular mechanism of cell inactivation during droplet drying The identification of a possible critical site responsible for cell death upon lethal heat has been a topic of long-term research interest since 1960s [50,51]. When drying is involved, the first cellular site to be considered is the cytoplasmic membrane and/or cell wall. Previous studies found that the removal of water molecules could

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

lead to package of polar groups of phospholipids on cell membrane, causing leakage of cellular contents upon rehydration [46,52]. The analysis of lipid peroxidation for Saccharomyces cerevisiae cells before and after drying in a fluidized bed showed that damage in cell membrane due to oxidative stresses is one of the main causes of yeast death [16]. Nonetheless, the evaluation of cellular damage in these studies was carried out at moderate environmental temperatures. In the current study, drying air temperatures up to 110 ◦ C were used and the results showed that heat damage is dominant during droplet drying with elevated air temperature (Fig. 3). Since industrial spray drying operations commonly utilize inlet temperatures over 200 ◦ C, the results in the present study should provide a more accurate description to the cell inactivation inside a spray dryer. The superimposition of kd trends at different drying conditions in Fig. 3 could possibly be explained by the fact that it was essentially the same strain dried within the same carrier. Comparing the transition temperature rage (50–65 ◦ C) in the present study to the denaturation temperature of different cellular components may provide a preliminary insight on the cellular component that was mainly responsible for cell inactivation. Previous studies investigated the denaturation of various cellular components during heat inactivation using differential scanning calorimetry (DSC) [48,53–55]. As heat damage was found to be the main cause to deactivate L. cremoris cells in the present study, results reported in those DSC studies could provide references for interpreting the current results. Different bacterial species showed slight divergences in the exact denaturation temperature for the same cellular component. Nevertheless, it has generally been agreed that 60–65 ◦ C is the onset of denaturation of bacterial ribosomes [49,53]. In addition, the denaturation of 30S ribosomal subunit starts at even lower temperature range of around 47–52 ◦ C [54]. The transition temperature (50–65 ◦ C) observed in the present study was in accordance with the onset temperature of denaturation of bacterial ribosomes, while the onset of the transition temperature fell in the temperature range of 30S ribosomal subunit denaturation (Fig. 3b). The results may hence be used as an independent argument for supporting ribosomes as the critical cellular sites of heat inactivation of prokaryotic cells. This hypothesis is in agreement with the study of Tolker-Nielsen and Molin [56], where it was reported that the death of Salmonella typhimurium cells during heat treatment coincided with the decline of 16S rRNA content, a part of 30S ribosomal subunit. Nonetheless, the present hypothesis should be interpreted with caution due to a couple of reasons. Firstly, no DSC data of L. cremoris have ever been reported in literature as direct evidence. Secondly, it is unclear whether the carrier composition would affect the transition temperature range. Future SDD experiments on denaturation kinetics of extracted ribosomes of L. cremoris would have to be carried out to confirm the correlation between cell inactivation and ribosomes denaturation during droplet drying of bacteria.

5. Conclusions This study experimentally investigated the inactivation of L. cremoris cells during droplet drying processes at six conditions. The resultant kd and Td under the six conditions constitute a single trend, suggesting that heat damage due to a high droplet temperature is the dominating factor responsible for cell death. The considered first order inactivation model provided a close description for all six inactivation histories. Droplet drying is a comparatively fast dehydration approach where the entire process can be completed in tens to hundreds seconds. The short duration may limit the effects of other environmental stresses that induce major cellular damage at lower temperatures. Future studies could

55

compare cell inactivation to the denaturation of different cellular components to explore the sub-cellular inactivation mechanisms. It will also be important to study the inactivation of other strains of bacteria to investigate whether the current model can be applied to other cases.

Acknowledgments The authors acknowledge use of the facilities and the scientific and technical assistance of the Monash Centre for Electron Microscopy (a facility funded by the University, Victorian and Commonwealth Governments).

References [1] N. Fu, X.D. Chen, Towards a maximal cell survival in convective thermal drying processes, Food Res. Int. 44 (2011) 1127–1149. [2] Y.-C. Wang, R.-C. Yu, C.-C. Chou, Viability of lactic acid bacteria and bifidobacteria in fermented soymilk after drying, subsequent rehydration and storage, Int. J. Food Microbiol. 93 (2004) 209–217. [3] E.E. Tymczyszyn, R. Diaz, A. Pataro, N. Sandonato, A. Gomez-Zavaglia, E.A. Disalvo, Critical water activity for the preservation of Lactobacillus bulgaricus by vacuum drying, Int. J. Food Microbiol. 128 (2008) 342–347. [4] T. Bansal, S. Garg, Probiotics: from functional foods to pharmaceutical products, Curr. Pharm. Biotechnol. 9 (2008) 267–287. [5] D. Bayrock, W.M. Ingledew, Fluidized bed drying of baker’s yeast: moisture levels, drying rates, and viability changes during drying, Food Res. Int. 30 (1997) 407–415. [6] G. Prabakaran, S.L. Hoti, Optimization of spray-drying conditions for the largescale preparation of Bacillus thuringiensis var. israelensis after downstream processing, Biotechnol. Bioeng. 100 (2008) 103–107. [7] C.A. Morgan, N. Herman, P.A. White, G. Vesey, Preservation of micro-organisms by drying; a review, J. Microbiol. Meth. 66 (2006) 183–193. [8] C. Santivarangkna, U. Kulozik, P. Foerst, Alternative drying processes for the industrial preservation of lactic acid starter cultures, Biotechnol. Prog. 23 (2007) 302–315. [9] S.H. Peighambardoust, A. Golshan Tafti, J. Hesari, Application of spray drying for preservation of lactic acid starter cultures: a review, Trends Food. Sci. Tech. 22 (2011) 215–224. [10] G.E. Gardiner, P. Bouchier, E. O’Sullivan, J. Kelly, J.K. Collins, G. Fitzgerald, R.P. Ross, C. Stanton, A spray-dried culture for probiotic Cheddar cheese manufacture, Int. Dairy J. 12 (2002) 749–756. [11] W.-C. Lian, H.-C. Hsiao, C.-C. Chou, Survival of bifidobacteria after spray-drying, Int. J. Food Microbiol. 74 (2002) 79–86. [12] K.B.P.K. Reddy, A.N. Madhu, S.G. Prapulla, Comparative survival and evaluation of functional probiotic properties of spray-dried lactic acid bacteria, Int. J. Dairy Technol. 62 (2009) 240–248. [13] B.C.S. To, M.R. Etzel, Spray drying, freeze drying, or freezing of three different lactic acid bacteria species, J. Food Sci. 62 (1997) 576–578. [14] D. Bayrock, W.M. Ingledew, Mechanism of viability loss during fluidized bed drying of baker’s yeast, Food Res. Int. 30 (1997) 417–425. [15] Y. Mille, J.-P. Girard, L. Beney, P. Gervais, Air drying optimization of Saccharomyces cerevisiae through its water–glycerol dehydration properties, J. Appl. Microbiol. 99 (2005) 376–382. [16] E. Garre, F. Raginel, A. Palacios, A. Julien, E. Matallana, Oxidative stress responses and lipid peroxidation damage are induced during dehydration in the production of dry active wine yeasts, Int. J. Food Microbiol. 136 (2010) 295–303. [17] H. Elizondo, T.P. Labuza, Death kinetics of yeast in spray drying, Biotechnol. Bioeng. 16 (1974) 1245–1259. [18] S.S. Kim, S.R. Bhowmik, Survival of lactic acid bacteria during spray drying of plain yogurt, J. Food Sci. 55 (1990) 1008–1010. [19] X. Zhou, J. Dong, J. Gao, Z. Yu, Activity-loss characteristics of spores of Bacillus thuringiensis during spray drying, Food Bioprod. Process. 86 (2008) 37–42. [20] X.D. Chen, K.C. Patel, Micro-organism inactivation during drying of small droplets or thin-layer slabs—a critical review of existing kinetics models and an appraisal of the drying rate dependent model, J. Food Eng. 82 (2007) 1–10. [21] N. Fu, Z. Zhou, T.B. Jones, T.T.Y. Tan, W.D. Wu, S.X.Q. Lin, X.D. Chen, P.P.Y. Chan, Production of monodisperse epigallocatechin gallate (EGCG) microparticles by spray drying for high antioxidant activity retention, Int. J. Pharm. 413 (2011) 155–166. [22] W.D. Wu, K.C. Patel, S. Rogers, X.D. Chen, Monodisperse droplet generators as potential atomizers for spray drying technology, Drying Technol. 25 (2007) 1907–1916. [23] D.E. Walton, C.J. Mumford, Spray dried products—characterization of particle morphology, Trans. IChemE. 77 (1999) 21–38. [24] N. Fu, M.W. Woo, X.D. Chen, Single droplet drying technique to study drying kinetics measurement and particle functionality: a review, Drying Technol. 30 (2012) 1771–1785. [25] X.D. Chen, S.X.Q. Lin, Air drying of milk droplet under constant and timedependent conditions, AIChE J. 51 (2005) 1790–1799.

56

N. Fu et al. / Biochemical Engineering Journal 79 (2013) 46–56

[26] Y. Jin, X.D. Chen, Numerical study of the drying process of different sized particles in an industrial scale spray dryer, Drying Technol. 27 (2009) 371–381. [27] M.W. Woo, W.R.W. Daud, A.S. Mujumdar, M.Z.M. Talib, W.Z. Hua, T.S. Masrinda, Comparative study of droplet drying models for CFD modelling, Chem. Eng. Res. Des. 86 (2008) 1038–1048. [28] M.W. Woo, A.S. Mujumdar, W.R.W. Daud, Spray Drying: Operation, Deposition and CFD Simulations, VDM Verlag Dr. Müller Aktiengesellschaft & Co. KG, Saarbrücken, 2010. [29] T.M. El-Sayed, D.A. Wallack, C.J. King, Changes in particle morphology during drying of drops of carbohydrate solutions and food liquids. 1. Effects of composition and drying conditions, Ind. Eng. Chem. Res. 29 (1990) 2346–2354. [30] H. Schiffter, G. Lee, Single-droplet evaporation kinetics and particle formation in an acoustic levitator. Part 2: Drying kinetics and particle formation from microdroplets of aqueous mannitol, trehalose or catalase, J. Pharm. Sci. 96 (2007) 2284–2295. [31] N. Fu, M.W. Woo, F.T. Moo, X.D. Chen, Microcrystallization of lactose during droplet drying and its effect on the property of the dried particle, Chem. Eng. Res. Des. 90 (2012) 138–149. [32] X. Li, S.X.Q. Lin, X.D. Chen, L. Chen, D. Pearce, Inactivation kinetics of probiotic bacteria during the drying of single milk droplets, Drying Technol. 24 (2006) 695–701. [33] A. Ghandi, I. Powell, X.D. Chen, B. Adhikari, Drying kinetics and survival studies of dairy fermentation bacteria in convective air drying environment using single droplet drying, J. Food Eng. 110 (2012) 405–417. [34] N. Fu, M.W. Woo, S.X.Q. Lin, Z. Zhou, X.D. Chen, Reaction engineering approach (REA) to model the drying kinetics of droplets with different initial sizes—experiments and analyses, Chem. Eng. Sci. 66 (2011) 1738–1747. [35] N. Fu, M.W. Woo, C. Selomulya, X.D. Chen, K. Patel, P. Schuck, R. Jeantet, Drying kinetics of skim milk with 50 wt% initial solids, J. Food Eng. 109 (2012) 701–711. [36] H. Huang, M.S.-L. Brooks, H.-J. Huang, X.D. Chen, Inactivation kinetics of yeast cells during infrared drying, Drying Technol. 27 (2009) 1060–1068. [37] L.C. Lievense, K. van’t Riet, Convective drying of bacteria I: the drying processes, Adv. Biochem. Eng. Biot. 50 (1993) 45–63. ˜ [38] P.A. Marechal, I. Martínez de Marnanón, I. Poirier, P. Gervais, The importance of the kinetics of application of physical stresses on the viability of microorganisms: significance for minimal food processing, Trends Food Sci. Tech. 10 (1999) 15–20. [39] I. Poirier, P.-A. Marechal, P. Gervais, Effects of the kinetics of water potential variation on bacteria viability, J. Appl. Microbiol. 82 (1997) 101–106.

[40] A.H. Geeraerd, C.H. Herremans, J.F. Van Impe, Structural model requirements to describe microbial inactivation during a mild heat reatment, Int. J. Food Microbiol. 59 (2000) 185–209. [41] Y. Boza, D. Barbin, A.R.P. Scamparini, Effect of spray-drying on the quality of encapsulated cells of Beijerinckia sp, Process Biochem. 39 (2004) 1275–1284. [42] W.-Y. Fu, M.R. Etzel, Spray drying of Lactococcus lactis ssp. lactis C2 and cellular injury, J. Food Sci. 60 (1995) 195–200. [43] A. Horaczek, H. Viernstein, Beauveria brongniartii subjected to spray-drying in a composite carrier matrix system, J. Microencapsul. 21 (2004) 317–330. [44] K. O’Riordan, D. Andrews, K. Buckle, P. Conway, Evaluation of microencapsulation of a Bifidobacterium strain with starch as an approach to prolonging vaibility during storage, J. Appl. Microbiol. 91 (2001) 1059–1066. [45] B. Riveros, J. Ferrer, R. Borquez, Spray drying of a vaginal probiotic strain of Lactobacillus acidophilus, Drying Technol. 27 (2009) 123–132. [46] J.H. Crowe, F.A. Hoeskstra, L.M. Crowe, Anhydrobiosis, Annu. Rev. Physiol 54 (1992) 579–599. [47] Y. Mille, J.-P. Obert, L. Beney, P. Gervais, New drying process for lactic bacteria based on their dehydration behavior in liquid medium, Biotechnol. Bioeng. 88 (2004) 71–76. [48] K. Obuchi, H. Iwahashi, J.R. Lepock, Y. Komatsu, Calorimetric characterization of critical targets for killing and acquired thermotolerance in yeast, Yeast 16 (2000) 111–119. [49] P. Teixeira, H. Castro, C. Mohacsi-Farkas, R. Kirby, Identification of sites of injury in Lactobacillus bulgaricus during heat stress, J. Appl. Microbiol. 83 (1997) 219–226. [50] C.A. Miles, Relating cell killing to inactivation of critical components, Appl. Environ. Microbiol. 72 (2006) 914–917. [51] B. Rosenberg, G. Kemeny, R.C. Switzer, T.C. Hamilton, Quantitative evidence for protein denaturation as the cause of thermal death, Nature 232 (1971) 471–473. [52] L.M. Crowe, Lessons from nature: the role of sugars in anhydrobiosis, Comp. Biochem. Physiol. A: Mol. Integr. Physiol. 131 (2002) 505–513. [53] J. Lee, G. Kaletunc, Evaluation of the heat inactivation of Escherichia coli and Lactobacillus plantarum by differential scanning calorimetry, Appl. Environ. Microbiol. 68 (2002) 5379–5386. [54] B.M. Mackey, C.A. Miles, S.E. Parsons, D.A. Seymour, Thermal denaturation of whole cells and cell components of Escherichia coli examined by differential scanning calorimetry, J. Gen. Microbiol. 137 (1991) 2361–2374. [55] C.A. Miles, B.M. Mackey, S.E. Parsons, Differential scanning calorimetry of bacteria, J. Gen. Microbiol. 132 (1986) 939–952. [56] T. Tolker-Nielsen, S. Molin, Role of ribosome degradation in the death of heatstressed Salmonella typhimurium, FEMS Microbiol. Lett. 142 (1996) 155–160.