Taking Advantage of Bacterial Adaptation in Order to Optimize Industrial Production of Dry Propionibacterium freudenreichii

Propionibacterium freudenreichii is a beneficial bacterium, used both as a probiotic and as a cheese starter. Large-scale production of P. freudenreichii is required to meet growing consumers’ demand. Production, drying and storage must be optimized, in order to guarantee high P. freudenreichii viability within powders. Compared to freeze-drying, spray drying constitutes the most productive and efficient, yet the most stressful process, imposing severe oxidative and thermal constraints. The aim of our study was to provide the tools in order to optimize the industrial production of dry P. freudenreichii. Bacterial adaptation is a well-known protective mechanism and may be used to improve bacterial tolerance towards technological stresses. However, the choice of bacterial adaptation type must consider industrial constraints. In this study, we combined (i) modulation of the growth medium composition, (ii) heat-adaptation, and (iii) osmoadaptation, in order to increase P. freudenreichii tolerance towards technological stresses, including thermal and oxidative constraints, using an experimental design. We further investigated optimal growth and adaptation conditions, by monitoring intracellular compatible solutes accumulation. Glucose addition, coupled to heat-adaptation, triggered accumulation of trehalose and of glycine betaine, which further provided high tolerance towards spray drying and storage. This work opens new perspectives for high quality and fast production of live propionibacteria at the industrial scale.


Introduction
Diet plays a pivotal role in the maintenance of physiology and of health, with a prominent effect on shaping the structure and activity of the gut microbiota. Dysbiosis, i.e., rupture of the gut microbiota homeostasis, is involved in inflammatory, allergic and atopic disorders [1][2][3]. Over the past decades, the diet of European countries evolved due to changes in life style, and this coincided with of P. freudenreichii growth and adaptation with an experimental design, in order to improve cell viability upon spray drying. We quantified compatible solutes to better understand P. freudenreichii cross-protections. We further checked the starter capacities of P. freudenreichii after adaptations and drying. This work opens new avenues for the optimization of the industrial production of dry P. freudenreichii.

Growth and Adaptation Conditions in the Experimental Design
The JMP Software from SAS was used to study the impact of glucose addition, heat-adaptation, and osmoadaptation on P. freudenreichii resistance to heat and oxidative stress. More precisely, glucose was added in YEL medium with concentrations ranging from 0 to 30 g·L −1 . P. freudenreichii was grown at 30 • C over 32 h to 46 h with gentle agitations (40 rpm) to decrease the growth rate [34,35]. At the beginning of the stationary phase, NaCl was added to cultures at a concentration ranging from 0 to 0.9 M during 2 h. Then, the cultures were heated at 42 • C during 0, 1, or 2 h. Adaptations were defined by the experimental design (Table 1). Table 1. Arrangement of the experimental design for the three independent variables used and their level.  −1  50  15  0  25  1  0  0  50  15  1  26  1  0  1  50  15  2  27  1  1  −1  50  30  0  28  1  1  0  50  30  1  29  1  1  1  50  30  2

Stress Challenges
Heat and oxidative challenges were applied to cultures after adaptations as previously described [25]. Heat challenge was performed by placing 2 mL (in a 15 mL Falcon tube) of P. freudenreichii cultures in a water bath at 60 • C over 10 min. Oxidative challenge was performed by adding 1.25 mM of hydrogen peroxide (Labogros, France) to 2 mL of culture during 1 h at 30 • C. CFU counting was made on untreated bacteria and after challenges in order to calculate survival percentage. The samples tested are reported in Table 1 and tests were performed in duplicate.
Statistics analyses were done to describe the effect of glucose concentration, heat adaptation and osmoadaptation on P. freudenreichii viability during heat and oxidative challenges. The software JUMP was used to fit the second order model to the independent variable. Only variables with a significance higher than 95% (p < 0.05) were included in final models. Surface response were drawn to illustrate the main and interactive effects of the independent variables on survival to heat and oxidative challenges. These challenges simulated the stressful spray drying process. Cells were then harvested by centrifugation (8000 g 10 min) and were washed twice in a NaCl solution with the same osmolality than the culture medium. Cells were then re-suspended in 2 mL of distilled water, then 8 mL of absolute ethanol were added. The suspension was homogenized and centrifuged (8000 g, 10 min) in order to remove cell fragments. The supernatant extract was evaporated over 7 h with a rotary evaporator. Dried extracts were then solubilized in deuterium oxide (Sigma-Aldrich, Saint Louis, MO, USA) as previously described [25].

NMR (Nuclear Magnetic Resonance) Analyses
All 1 H and 13 C NMR spectra were recorded at 298 K on a Bruker Advance 500 spectrometer equipped with a 5 mm TCI triple-resonance cryoprobe (PRISM core facility, Rennes, France). 1 H spectra were acquired with a 6 kHz spectral width, 32 K data points and a total repetition time of 6.73 s. 13 C spectra were acquired using a proton power-gated decoupling sequence with a 30 • flip angle, a 30 kHz spectral width, 64 K data points and a total repetition time of 3.08 s. The data were processed with Topspin software (Bruker Biospin, Wissembourg, France). Before applying the Fourier transform, free induction decays of 1 H spectra were treated with an exponential broadening of 0.3 Hz.
The relative concentration of trehalose, glutamate, proline, and glycine betaine in the samples was determined by integration of their 1 H signals. Results are expressed as NMR relative units (RU) reported to one unit of optical density (equivalent to 1 mL of culture at an OD of 1).

Spray Drying and Powder Storage
P. freudenreichii was cultivated as previously either in YEL medium, or in YEL medium containing glucose, with 1 or 2 h of heat adaptation at 42 • C. (YEL + 15 g glucose+ 42 • C, 1 h and YEL + 30 g Glucose+ 42 • C, 2 h). 20% (w/w) of maltodextrine (DE = 6-8) was added to the cultures at the beginning of the stationary phase or after heat adaptation. These different mixtures (∼2 L) were shaken for 10 min prior to delivery to the dryer by a peristaltic pump (520S, Watson-Marlow, France).
A two-fluid nozzle with a diameter of 0.8 mm was used for atomization. The inlet air temperature was fixed at 160 • C. The temperature outlet air was controlled at 60 ± 2 • C, by adjusting the feed rate. The bacterial viabilities were estimated by numeration on YEL agar plates before and after spray drying as described previously [36]. The powders were collected and sealed in sterilized polystyrene bottles (Gosselin, France), stored at a controlled temperature of 4, 20 or 37 • C, and kept away from light over 145 days. One gram of powder was solubilized in 9 g of sterile water and bacterial viability was tested by numeration on YEL agar plates incubated at 30 • C.
2.6. Revivification of dried P. freudenreichii in a cheese-like dairy medium 2.6.1. Culture with Dried P. freudenreichii in a Cheese-Like Dairy Medium Cow milk ultrafiltrate was prepared by milk ultrafiltration (cut-off 8 kDa) as previously described [37]. The two major lactic acid starters, Lactobacillus delbrueckii CIRM-BIA 209 and Streptococcus thermophilus CIRM-BIA 67, were co-cultivated at 43 • C over 12 h in milk ultrafiltrate containing 5 g·L −1 of casein peptone. Cultures were centrifuged in order to eliminate lactic acid bacteria. The supernatants were collected, the pHs were readjusted to 7 using a solution of NaOH 5 M, then the supernatants were filter-sterilized (Top filter PES, 0.45 µm, Nalgene Company, NY, USA). Then, 1.5 g of P. freudenreichii powder dried previously was inoculated in 400 mL of this preconditioned milk ultrafiltrate (named later cheese-like medium). A control was performed using a fresh liquid P. freudenreichii culture realized in the cheese-like medium supplemented by 1.5 g of maltodextrine. Cultures were incubated at 24 • C over 14 days in order to mimic growth conditions in the warm room during Emmental cheese making [38]. Growth curves were established by CFU numeration on YEL agar plates.

Organic Acids Quantification
Supernatants of the different cultures were harvested by centrifugation during the exponential phase, the beginning of the stationary phase and after 14 days of growth. Culture supernatants were half-diluted with 0.02 M H 2 SO 4 and centrifuged at 8000 g during 30 min at 4 • C to discard the protein pellet prior to filter-sterilization and analysis. Organic acids and sugar were separated using a Minex A-6 ion exchange column (Dionex, Sunnyvale, CA, USA) at 55 • C with 0.01 M H 2 SO 4 as eluent at a flow rate of 1 mL min −1 . Both UV (210 nm) and refractometric detectors were used. Appropriate standards of lactose, acetate, propionate, and lactate were used as described previously [39].

Statistical Analysis
The data were from triplicate samples. All the results are presented as mean values with standard deviation. Statistical significance was set at p < 0.05. Calculations were performed using GraphPad Prism Software (Prism 7 for Windows, Redmond, WA, USA).

Optimization of P. freudenreichii Stress Tolerance with an Experimental Design
We analyzed P. freudenreichii survival rates upon heat and oxidative challenges in order to estimate individual and interactive coefficients effect on P. freudenreichii tolerance towards heat challenge (Table 2) and oxidative challenge (Table 3). Salt addition after the beginning of the stationary phase had a negative impact on P. freudenreichii survival during heat and oxidative challenges ( Figure 1) (coefficient estimated: -7% and -10%, in Tables 2 and 3, respectively), whereas glucose addition and heat adaptation had a positive impact on P. freudenreichii survival during these challenges ( Figure 1). Indeed, glucose addition increased P. freudenreichii viability during heat challenge (coefficient estimated: +24%) more than heat adaptation (coefficient estimated: +9%). During oxidative challenge, heat-adaptation had a higher impact than glucose addition (coefficient estimated: +19% and +10%, respectively). Best survival during both challenges was obtained when P. freudenreichii was grown in the presence of 30 g·L −1 of glucose and heat-adapted (42 • C, 2 h) at the beginning of stationary phase ( Figure 1).

P. freudenreichii Viability after Spray Drying and Storage
Control cultures (YEL), in the best conditions defined by the experimental design (YEL + 30 g glucose + 42 • C, 2 h) and in the intermediate tolerance conditions (YEL + 15 g glucose + 42 • C, 1 h) were spray dried. P. freudenreichii grown in YEL with 15 g of glucose and heat-adapted during 1 h at 42 • C survived better to spray drying than unadapted cells (74% and 30% survival, respectively) ( Figure 2A). A higher glucose concentration in the culture medium with a longer thermal adaptation further increased the survival during spray drying (96.4%). The addition of glucose and/or the thermal adaptation had a positive impact on P. freudenreichii viability during spray drying. During storage at 4 and 20 • C, P. freudenreichii grown with glucose and then heat-adapted displayed enhanced viability ( Figure 2B,C). The highest the glucose concentration and the longest the heat-adaptation, the lowest P. freudenreichii death occurred during storage at 4 and 20 • C. During storage at 37 • C, these adaptations had no impact on cells' death ( Figure 2D). As a quality control, P. freudenreichii colonies were isolated before and after adaptation and spray drying, prior to RAPD analysis. Figure 3 shows that RAPD profiles were identical in the three conditions; original culture, spray dried after adaptation (YEL + 15 g glucose + 42 • C, 1 h), and spray dried after (YEL + 30 g glucose + 42 • C, 2 h).     . RAPD analysis of P. freudenreichii cultures before and after adaptation and drying P. freudenreichii. Colonies were isolated from the original culture and from powders after adaptation and spray drying. RAPD was performed using two RAPD primers (left: primer 1: AAGAGCCCGT and right: primer 2: AACGCGCAAC) as described in "Materials and Methods". Amplified products were resolved by electrophoresis (100 V, 1 h) on 1% agarose gels. First and last lines correspond to 1 kb DNA ladder.

Compatible Solutes Accumulation Depends on P. freudenreichii Growth Medium and Adaptation Conditions.
We used NMR as the standard used in microbial physiology for evidence accumulation of compatible solutes as a result of osmoadaptation [26,[40][41][42]. The accumulated compatible solutes were analyzed by NMR as described [43][44][45][46]. As shown in Figure 4A-D), the spectra corresponding to glycine betaine (yellow), trehalose (purple), glutamate (green), and proline (red) allowed us to identify and quantify these compounds in cytoplasmic hydroalcoholic extract (blue). P. freudenreichii CIRM-BIA 129 was able to accumulate glycine betaine, trehalose, glutamate, and proline during growth and adaptations ( Figure 4E). In the control medium (YEL), P. freudenreichii accumulated high amounts of glutamate (66.7%), and to a lesser extent proline (24.2%) and little amounts of glycine betaine (9.1%). This medium is isotonic with an osmolality of 0.308 osmol. The addition of 15 g of glucose to this culture medium (0.433 osmol), as well as heat-adaptation over 1 h, drastically affected compatible solutes accumulation. First, they decreased the total amount of compatible solute accumulated. Second, they changed the proportions of compatible solutes accumulated. Compared to the control medium, glycine betaine and trehalose were accumulated in higher amount (49.4% and 306 % respectively), with lower amounts of proline (13.0%) and glutamate (7.0%). Higher Colonies were isolated from the original culture and from powders after adaptation and spray drying. RAPD was performed using two RAPD primers (left: primer 1: AAGAGCCCGT and right: primer 2: AACGCGCAAC) as described in "Materials and Methods". Amplified products were resolved by electrophoresis (100 V, 1 h) on 1% agarose gels. First and last lines correspond to 1 kb DNA ladder.

Compatible Solutes Accumulation Depends on P. freudenreichii Growth Medium and Adaptation Conditions
We used NMR as the standard used in microbial physiology for evidence accumulation of compatible solutes as a result of osmoadaptation [26,[40][41][42]. The accumulated compatible solutes were analyzed by NMR as described [43][44][45][46]. As shown in Figure 4A-D), the spectra corresponding to glycine betaine (yellow), trehalose (purple), glutamate (green), and proline (red) allowed us to identify and quantify these compounds in cytoplasmic hydroalcoholic extract (blue). P. freudenreichii CIRM-BIA 129 was able to accumulate glycine betaine, trehalose, glutamate, and proline during growth and adaptations ( Figure 4E). In the control medium (YEL), P. freudenreichii accumulated high amounts of glutamate (66.7%), and to a lesser extent proline (24.2%) and little amounts of glycine betaine (9.1%). This medium is isotonic with an osmolality of 0.308 osmol. The addition of 15 g of glucose to this culture medium (0.433 osmol), as well as heat-adaptation over 1 h, drastically affected compatible solutes accumulation. First, they decreased the total amount of compatible solute accumulated. Second, they changed the proportions of compatible solutes accumulated. Compared to the control medium, glycine betaine and trehalose were accumulated in higher amount (49.4% and 306 % respectively), with lower amounts of proline (13.0%) and glutamate (7.0%). Higher concentration of glucose (30 g·L −1 : 0.455 osmol) in the YEL medium and a longer heat-adaptation (2 h) further increased the amount of compatible solutes accumulated by P. freudenreichii, but with a profile similar to that of P. freudenreichii grown with 15 g of glucose and heat-adapted for 1 h. Indeed, P. freudenreichii accumulated in this case trehalose (40%) and glycine betaine (42.5%) in high amounts, with lower amounts of proline (10%) and of glutamate (10%). concentration of glucose (30 g . L -1 : 0.455 osmol) in the YEL medium and a longer heat-adaptation (2 h) further increased the amount of compatible solutes accumulated by P. freudenreichii, but with a profile similar to that of P. freudenreichii grown with 15 g of glucose and heat-adapted for 1 h. Indeed, P. freudenreichii accumulated in this case trehalose (40%) and glycine betaine (42.5%) in high amounts, with lower amounts of proline (10%) and of glutamate (10%).

Growth of Dried P. freudenreichii in a Cheese-Like Medium
Powders containing P. freudenreichii, or a fresh culture (P. freudenreichii in stationary phase) in YEL medium as a control, were inoculated in a cheese-like medium consisting of milk ultrafiltrate previously fermented by a mixture of L. delbrueckii and a S. thermophilus, the major lactic starters implemented together with P. freudenreichii in Emmental cheese. In the absence of adaptation, P. freudenreichii was not able to grow in the cheese-like medium and its population remained constant throughout the time of incubation in this medium. Conversely, adapted cultures and dried P. freudenreichii powders were able to grow in the cheese-like medium ( Figure 5A), indicating revivification of the propionic starter, so that it reached final propionibacterial populations equal to the control. At the end of the fermentation, all cultures had a pH between 4.72 and 4.95, but the pH decreased more slowly with the fresh culture, when compared to dried propionibacteria ( Figure 5B).

Fate of Saccharides and Organic Acids
Utilization of lactose and of lactate, as well as production of propionate and of acetate, the main end products of propionic fermentation, were monitored as described in materials and methods. Lactose utilization was nearly the same between dried P. freudenreichii and the fresh culture ( Figure 5C-E). Dried P. freudenreichii consumed lactate and produced propionate and acetate less rapidly than the fresh culture. Final consummation of lactate and lactose and final production of acetate and propionate were non-significantly different between fresh preculture and dried P. freudenreichii. Propionate final concentration was twice the acetate final concentration. Interestingly, in the case of dried propionibacteria, lactose was consumed as soon as the beginning of growth, while lactate remained first constant before being consumed after 100 h of growth. In the case of fresh precultures, both lactose and lactate were used throughout the culture. were used to ferment a cheese-like medium. Powders obtained previously were inoculated in a cheese-like medium. P. freudenreichii growth was monitored by CFU counting in YEL agar plates (A) and acidification were followed (B). Lactate and lactose consumption, as well as propionate and acetate production, were quantified by HPLC (C-E) as described in material and methods.

Addition of Glucose and Heat-Adaptation Both Induce Cross-Protections to P. freudenreichii
Glucose addition and heat adaptation both improved P. freudenreichii tolerance towards oxidative and heat challenges (Figure 1). Heat adaptation haas already been described to increase P. freudenreichii, Enterococcus faecalis, and Escherichia coli thermotolerance [44,47,48]. Addition of glucose during growth also improved P. freudenreichii viability during heat and oxidative stress. Accordingly, growth in hyper-concentrated sweet whey, a medium rich in the saccharide lactose, was previously shown to induce such cross-protection, in parallel with trehalose accumulation [26,36]. We showed here that addition of the monosaccharide glucose in combination with a thermo-adaptation also triggers cross-protections and the accumulation of trehalose and of glycine betaine. Glucose is known to be a trehalose building block in a wide range of microorganisms [49]. The experimental design highlighted the positive effect of interactions between the addition of glucose and the heat adaptation. The condition allowing the best survival during oxidative and heat challenges was the YEL medium containing 30 g·L −1 of glucose and heat-adaptation during 2 h at 42 • C ( Figure 1). Following glucose addition and thermal adaptation, we also tested addition of NaCl after growth and observed afterwards a reduced tolerance. Growth in the presence of NaCl may induce tolerance [25,44,50], while NaCl addition after growth does not.

Combining Glucose Aaddition and Thermo-Adaptation Increases P. freudenreichii Viability during Spray Drying and Storage
Glucose addition in the growth medium and heat adaptation at the beginning of the stationary phase increased P. freudenreichii survival after drying (Figure 2A). Increasing the glucose concentration and/or the duration of heat adaptation led to higher survival as predicted by the experimental design. Heat adaptation has already been reported to improve Lactobacillus paracasei and Candida sake survival during spray drying [28,51]. Over storage at 4 and 20 • C, dried P. freudenreichii grown with glucose and heat-adapted exhibited a better survival rate than the control ( Figure 2B,C). In other words, a higher glucose concentration and a longer heat-adaptation increased survival during storage as during spray-drying. Similarly, heat adaptation was reported to increase L. plantarum viability during storage [31]. Sucrose addition can increase Lactobacillus sakei survival during storage after spray drying [52]. However, the impact of saccharides on bacteria viability during storage is not well known yet.

Combining Glucose Addition and Thermo-Adaptation Leads to Compatible Solutes Accumulation and Increases P. freudenreichii Viability to Spray Drying
Control cultures accumulated a high amount of the compatible solutes glutamate and proline. This can be explained by the culture agitation and the presence of oxygen. Indeed, such accumulation was not observed in the same YEL medium in the absence of agitation (data not shown). In this study, proline was accumulated under the three different conditions. This amino acid is accumulated by Brevibacterium flavum in the presence of dissolved oxygen and is biosynthesized from glutamate [53]. In the presence of glucose and under heat-adaptation, P. freudenreichii accumulated high amounts of glycine betaine and of trehalose. Glycine betaine is known to play a role during thermo-protection [54,55]. Trehalose accumulation is well known during heat-adaptation in yeast [56,57], and P. freudenreichii is known to accumulate trehalose [41]. Increased glucose concentration in the growth medium led to increased trehalose accumulation, as described previously [41]. Lactose reportedly leads to higher trehalose accumulation by P. freudenreichii than glucose [26,41]. Addition of glucose coupled to heat adaptation increase trehalose accumulation by Saccharomyces cerevisiae [57]. When P. freudenreichii was grown with high glucose concentration and was heat-adapted, its trehalose accumulation and its resistance to heat challenges increased. Accordingly, trehalose accumulation is determinant for thermo-tolerance for S. cerevisiae and P. freudenreichii [26,58].

Combining Glucose Addition and Heat Adaptation Leads to Production of Efficient Propionic Starter
P. freudenreichii, in the cheese-like medium, exhibited similar growth parameters, whether inoculated as a fresh culture, or as a powder obtained from an adapted culture. Compared to the control fresh culture, dried P. freudenreichii acidified more and faster the cheese-like dairy medium. Lactate concentration decreased later in cultures inoculated with powders containing adapted P. freudenreichii, which delayed the production of propionate and acetate. In all cultures, P. freudenreichii consumed both lactate and lactose and produced acetate and propionate with a 2:1 ratio, that is characteristic of propionic fermentation according to the Fitz equation [59]. The amount of lactose, lactate, propionate, and acetate were similar for all cultures at the end of fermentation. However, P. freudenreichii used both lactose and lactate as soon as the beginning of growth in the case of fresh cultures. By contrast, P. freudenreichii preferentially used lactose at the beginning of growth, in accordance with faster acidification, in the case of dried bacteria. P. freudenreichii grown in presence of glucose and heat-adapted displayed enhanced resistance to spray drying and enhanced acidification ability. Such powders containing adapted P. freudenreichii can thus be used as efficient cheese starters.

Conclusions
To conclude, we modulated the growth medium composition and we used thermal adaptation to increase P. freudenreichii viability during spray drying and during storage. In the presence of glucose, P. freudenreichii accumulated trehalose and glycine betaine, whatever the osmotic pressure. Compared to osmotic adaptation, thermal adaptation is easier to implement in industry, and allows faster production of P. freudenreichii, with high viability during spray drying and storage. This work thus opens up new avenues for industrial production of propionic starters.