Surfactin, mycosubtilin, and plipastatin (LPPs) were produced in 5-L shake flasks filled at 20% with Landy media, under orbital agitation at 160 rpm, as previously reported (Coutte et al., 2010a). Surfactin and plipastatin were respectively produced by B. subtilis BBG131 strain (Coutte et al., 2010a) and B. subtilis Bs2504 strain (Ongena et al., 2007). BBG131 and Bs2504 were grown for 72 h at 37°C in a modified Landy MOPS culture medium (composition: 20 g·L⁻¹ of glucose, 5 g·L⁻¹ of glutamic acid, 1 g·L⁻¹ of yeast extract, 1 g·L⁻¹ of K₂HPO₄, 0.5 g·L⁻¹ of MgSO₄·7H₂O, 0.5 g·L⁻¹ of KCl, 16 mg·L⁻¹ of l-tryptophan, 1.6 mg·L⁻¹ of CuSO₄·5H₂O, 1.2 mg·L⁻¹ of MnSO₄·H₂O, 0.4 mg·L⁻¹ of Fe₂(SO₄)₃·7H₂O, and 21 g·L⁻¹ of 3-morpholinopropane-1-sulfonic acid MOPS, buffered at pH 7.0 using KOH 3 M). Mycosubtilin was produced by B. subtilis BBG125 strain (Béchet et al., 2013). BBG125 was grown for 72 h at 30°C in another modified Landy MOPS culture medium (Guez et al., 2021) (composition: 20 g·L⁻¹ of glucose, 2 g·L⁻¹ of glutamic acid, 1 g·L⁻¹ of yeast extract, 1 g·L⁻¹ of K₂HPO₄, 0.5 g·L⁻¹ of MgSO₄·7H₂O, 0.5 g·L⁻¹ of KCl, 2.3 g·L⁻¹ of (NH₄)₂SO₄, 32 mg·L⁻¹ of CuSO₄·5H₂O, 100 mg·L⁻¹ of MnSO₄·H₂O, 30 mg·L⁻¹ of Fe₂(SO₄)₃·7H₂O, and 21 g·L⁻¹ of 3-morpholinopropane-1-sulfonic acid MOPS, buffered at pH 6.5 using KOH 3 M). In these culture conditions, BBG131, Bs2504, and BBG125 were able to produce respectively 1.2 ± 0.2 g·L⁻¹ of surfactin, 0.4 ± 0.1 g·L⁻¹ of plipastatin, and 0.05 ± 0.02 g·L⁻¹ of mycosubtilin.

The different steps undertaken for the production and the purification of LPP from a B. subtilis fermentation are shown in Figure 1. At the end of the culture, the broth medium was centrifuged in an Avanti J-E centrifugation unit (Beckman Coulter, Krefeld, Germany) at 8,000 g for 40 min at 10°C, and the cells were discarded. The supernatant was then 10-times concentrated through a 10-kDa Hydrosart ultrafiltration membrane (Sartorius, Goettingen, Germany). The resulting retentate was then supplemented with demineralized water (4 volumes of water for 1 volume of retentate) in order to perform four successive diafiltration processes through the ultrafiltration membrane described above. The de-salted resulting retentate was subsequently supplemented with ethanol 100% (3 volumes of ethanol for 1 volume of diafiltered retentate) to reach a final concentration in ethanol at 75%. On the one hand, the addition of ethanol precipitates most of the proteins included in the retentate, and on the other hand, it leads to the disruption of the LPP micelles into free LPP molecules (with a molecular weight <10 kDa). The ethanolic fraction obtained was subsequently filtrated through the same membrane described above in order to separate the precipitate fraction (retentate) from the pre-purified LPP fraction (permeate). Finally, the ethanol contained in the pre-purified LPP fraction was evaporated using a rotary evaporator (Rotavapor^® R-300, Büchi Labortechnik AG, Flawil, Switzerland).

Before proceeding to LPP spray-drying, a preliminary experiment has been carried out to determine the resistance of surfactin, mycosubtilin, and plipastatin to high-temperature treatment. LPPs (surfactin, mycosubtilin, and plipastatin) formulated in powder form were supplied by Lipofabrik (Lipofabrik, Villeneuve d'Ascq, France). LPP powders were solubilized with demineralized water at a concentration of 1 g·L⁻¹ for surfactin and plipastatin and 0.5 g·L⁻¹ for mycosubtilin. LPPs were held in an aqueous solution, in a closed container avoiding evaporation, in a sand bath (VWR, Fontenay-sous-Bois, France) heated at 20°C, 75°C, 92°C, 109°C, 126°C, 134°C, or 143°C for 1 min. Before and after the thermal treatment, the number of LPPs was quantified by ultra-high-performance liquid chromatography (UPLC) according to the method described in Lipopeptide Quantification and Purity Determination. LPPs were considered to be stable at the tested temperature when the same amount of LPPs was quantified before and after the thermal treatment.

The characteristics of all the spray-dried LPP samples (LPP concentration and volume of the solution) of this study are presented in Supplementary Table S1. For each modality, the experiment (production, purification, and spray-drying) was performed in triplicate. Two types of LPP samples were spray-dried: either already formulated LPPs in powder forms (Supplementary Table S1A) or LPPs in solution sampled at different steps of the LPP production and purification using membrane ultrafiltration workflow (Figure 1; Supplementary Table S1B). LPPs (surfactin, mycosubtilin, and plipastatin) formulated in powder form, supplied by Lipofabrik (Lipofabrik, Villeneuve d'Ascq, France), were resuspended with demineralized water before the spray-drying step. Surfactin, plipastatin, and mycosubtilin samples of various purities were sampled at different steps of the LPP purification process (A, untreated supernatant; B, concentrated and diafiltered LPP solution; C, enriched LPP solution after ethanol evaporation) before the spray-drying process. The addition of maltodextrin as an aid for spray-drying was also evaluated with mycosubtilin supernatant. Thus, supernatant from fermentation broth and containing 1 g·L⁻¹ of mycosubtilin was also spray-dried after the addition of 2.5% and 5% w/v maltodextrin (Glucidex IT 12) supplied by Roquette (Lestrem, France).

The spray-drying of the solutions was carried out in the Büchi Mini Spray Dryer B-290 (Büchi Labortechnik AG, Switzerland) configured with an inlet temperature of 140°C and exhaust temperature of 70°C. The pressure of the spraying air was at 40-mm water column, and the pumping flow was maintained at 8.5 ml·min⁻¹. During the spray-drying procedure, inlet and outlet air characteristics (temperature, percentage of humidity, dew temperature, and wet bulb temperature) were analyzed using the Testo 625 psychrometer (Testo, Titisee-Neustadt, Germany). These values were used to establish a water balance as well as an energy balance.

In order to calculate the yield of the spray-drying process, the dry matter of the obtained powder was determined with a desiccator (Presica XM60, Presica Instruments AG, Dietikon, Switzerland) set on 102°C. The yield (Y) and the specific LPP yield (Y _LPP ), expressed in percent, were respectively calculated according to Eqs 1, 2: Y =   D M p o w d e r   ∗   m p o w d e r D M s o l u t i o n   ∗   m s o l u t i o n (1) Y L P P =   P l i p o p e p t i d e   ∗   m p o w d e r C l i p o p e t i d e   ∗   V s o l u t i o n (2) with the following parameters:

D M p o w d e r   : the dry matter of the obtained spray-dried powder (%);

In order to ensure that the spray-drying process is not altering LPP properties, several activity assays were carried out on the obtained spray-dried products. The surfactant properties of the three kinds of spray-dried LPPs (surfactin, mycosubtilin, and plipastatin) were evaluated by measuring the surface tension according to the Du Nouÿ method using a tensiometer TD1 (Lauda, Lauda-Königshofen, Germany) as described previously (Leclère et al., 2006). This was carried out in distilled water at pH 8. The values were compared with the surface tension measured on LPP solutions before spray-drying using the same methodology. The surfactant properties of LPPs are considered to be conserved if the values measured after spray-drying fit with those measured before spray-drying.

Anti-Legionella activity of surfactin was performed in a microtiter plate to evaluate the minimal inhibitory concentration (MIC) of surfactin against Legionella pneumophila strains. The strains L. pneumophila 2-15-1 and 2-15-2 used in this study were isolated and identified from sanitary water by the Institut Scientifique de Service Public of Liège and kindly made available to us for this study. The MIC has been determined following the standard protocol (Loiseau et al., 2015). Briefly, 100 µl of sterile liquid Buffered Yeast Extract (BYE) broth have been dispensed in each well of a 96-well plate. Then, 9 two-fold dilutions of surfactin (before spray-drying and after spray-drying) in BYE broth were made starting at a concentration of 50 mg·L⁻¹. Wells were inoculated by dispensing 5 µl of a 10⁵ CFU·ml⁻¹ of bacterial culture in the growth phase. Wells containing only 100 µl of BYE broth and wells inoculated BYE broth were used respectively as negative and positive controls. Plates were incubated at 37°C for 48 h. The MIC has been determined at the lowest concentration where no visible growth is observed in the wells. The mean value of MIC was presented.

The antifungal activities of plipastatin samples were determined against V. inaequalis S575 by measuring the EC₅₀ as recently described (Desmyttere et al., 2019). The mean value of EC₅₀ was presented.

The antifungal activity of mycosubtilin was evaluated using the microwell dilution method in a 96-well microplate. Saccharomyces cerevisiae DSM1333 strain was grown overnight at 37°C on Potato Dextrose Agar (PDA), dispersed in water, adjusted to a final OD_600nm of 0.04, and used as an inoculum. Mycosubtilin samples were dissolved in dimethyl sulfoxide (DMSO) and then diluted in Roswell Park Memorial Institute (RPMI) 2× media to reach a final concentration of 0.4 g·L⁻¹. Ninety-six-well microplates were prepared by dispensing 100 µl of RPM 2× media in all wells except in the negative control. Initially prepared mycosubtilin samples measuring 200 µl were added to the first wells. Two-fold serial dilutions of 100 µl were made from columns 1 to 10 in a concentration range from 0.8 to 400 mg·L⁻¹. The inoculum measuring 100 µl was added to each well except in the negative control. Distilled sterile water measuring 100 µl was added to the negative control. The final volume of each well was 200 µl. The plate was incubated at 37°C for 12 h. After incubation, the lowest concentration of each extract showing no yeast growth or turbidity was taken as its MIC value. The mean value of MIC was presented.

For these three antimicrobial assays, mean values of MIC or EC₅₀ and CIs (α = 0.05) were calculated from three technical and three biological replicates using R software nlstools package (R version 3.5.3, R Core Team, 2019).

The determination of the LPP concentration and purity in the samples before and after the spray-drying process was performed by UPLC. The LPPs in powder form, either those before spray-drying (surfactin, plipastatin, and mycosubtilin provided by Lipofabrik, Villeneuve d’Ascq, France) or those obtained after each spray-drying experiment, were solubilized in 100% methanol at a concentration of 1 g·L⁻¹. Prior to being injected, all samples were centrifuged (10,000 g, 10 min). The high-concentrated samples such as those sampled at the late phase of the LPP purification process (diafiltered fraction and enriched LPP fraction) were diluted with demineralized water in order to be properly quantified by the apparatus. All the measurements have been done with an injection volume of 20 µl into an Interchim C18 column (UP5TP18-250/030 C18, Interchim, Montluçon, France) equipped on an ACQUITY UPLC system (Waters, Milford, MA, USA) coupled to a UV detector (detection at 215 nm). The solvents used, water of high-performance liquid chromatography (HPLC) grade with 0.1% trifluoroacetic acid (Solvent A) and acetonitrile of HPLC grade with 0.1% trifluoroacetic acid (Solvent B), were provided by Sigma Aldrich (St. Louis, MO, USA). The solvent flow was 0.6 ml·min⁻¹ throughout the chromatography column.

To quantify surfactin, plipastatin, and mycosubtilin, the same analysis method was applied with the following gradient: from 0 to 5 min, 95% A/5% B; from 5 to 40 min, a linear gradient from 95% A/5% B to 0% A/100% B; from 40 to 45 min, 0% A/100% B; and from 45 to 56 min, 95% A/5% B. The concentration of the sample was determined by comparison with 98% surfactin, 95% iturin A, and 90% fengycin standards, all provided by Sigma Aldrich (St. Louis, USA). The elution time of surfactins, mycosubtilins (and iturin A), and plipastatins, under the above-described analysis conditions, were respectively 36–41, 24–26, and 27–33 min. The retention time and second derivative of the absorption spectrum between 200 and 400 nm were used to identify the eluted molecules (Empower Software, Waters).

The purity of the analyzed powders was determined by analogy with the corresponding LPP standard of known purity. Both the analyzed samples and the corresponding standard were solubilized at a concentration of 1 g·L⁻¹ in 100% methanol. After analysis according to the method described above, the purity (P _sample in %) was determined according to the following equation (with P _standard as the purity of the standard in %; A _sample as the total specific area of the sample in µV·s⁻¹; and A _standard as the total specific area of the standard in µV·s⁻¹): P s a m p l e =   P s t a n d a r d   ∗   A s a m p l e A s t a n d a r d

Prior to considering spray-drying as an appropriate method to formulate LPPs in powder form, it is important to ensure that this operation is not negatively impacting LPP activities. First of all, evaluation of the thermal resistance of the LPPs of B. subtilis was carried out. Commercial powders of the three LPPs (i.e., surfactin, plipastatin, and mycosubtilin, provided by Lipofabrik (Lipofabrik, Villeneuve d'Ascq, France)) were exposed 1 min to different temperatures (up to 143°C) in order to evaluate their thermal resistance and then were quantified by UPLC. Regarding the applied temperature, any degradation of the three LPPs tested was observed after exposure (data not shown). This result enables us to confirm that the structures of these LPPs are preserved after a short-time exposure at high temperatures (up to 143°C). This observation is in line with other previous studies that have demonstrated that the surfactant properties of surfactin for instance are preserved after a short-time exposure at temperatures of up to 170°C (Barcelos et al., 2014). Then, these different commercial LPPs in powder form have also been resuspended in demineralized water and subsequently spray-dried. Activities of the obtained dried LPPs have then been evaluated and compared with the activities of the non-spray-dried initial samples. The surfactant activity has been monitored for each LPP, while anti-Legionella activity was measured for surfactin samples, and antifungal activities were evaluated for both mycosubtilin and plipastatin samples.

After demonstrating, on highly pure molecules, that spray-drying does not affect the activities of LPPs (surfactant and antimicrobial), we were interested in the performances of this drying process on different LPP preparations resulting from the different steps of the ultrafiltration purification process, namely, the cell-free supernatant, the diafiltered fraction, and finally the purified fraction. These different steps are presented in Figure 1. The dry matter of the spray-dried LPP solution and of the obtained powder, as well as the LPP purity in the dried product, is depicted in Table 3. From these first results, it can be concluded that spray-drying is a very efficient process to obtain LPPs in powder form with a high dry matter of 93.6% ± 2.2% whatever the considered family of LPPs is; nevertheless, the state of purification of the latter has an important impact.

The results presented in Table 3 show that the purification process has a significant impact on the purity of the dried LPPs obtained. This result was expected from our previous work on this purification process (Coutte et al., 2010b; Jauregi et al., 2013). The purity obtained with diafiltered solution of surfactin (between 50% and 60%) using the same ultrafiltration procedure is in accordance with that previously published by our team (Coutte et al., 2013). Nevertheless, a small decrease in the performance of the last purification step (enriched LPP fraction) can be noticed for surfactin and mycosubtilin in comparison with the ones previously obtained (82.4% ± 1.0% and 78.3% ± 12.6%, respectively) (Coutte et al., 2013; Deravel et al., 2014), but this remains in very high values compared with other studies using the ultrafiltration process (Wei et al., 2010). Regarding the purification of plipastatin alone, there is little work in the scientific literature because it is often combined with surfactin (Coutte et al., 2010b; Coutte et al., 2017). Our results obtained using a cut-off of 10 kDa are satisfactory with respect to the literature, not only at the diafiltration step where the purity of more than 30% is obtained (Ben Ayed et al., 2015) but also after the last purification step. Indeed, a work using a combination of ultrafiltration, acid precipitation, and nanofiltration has shown that it was possible to obtain fengycin with a purity close to 95%, which is close to the purity of our enriched plipastatin fraction (97.7%) (Sen and Swaminathan, 2005). It should also be noted here that there is a significant difference in the purity of surfactin and plipastatin (or mycosubtilin) after the diafiltration step. This can be explained by the fact that the concentrations of the solutions are very different between these molecules (4.73 g·L⁻¹ for surfactin, 1.22 g·L⁻¹ for mycosubtilin, and 1.46 g·L⁻¹ for plipastatin).

The results presented in Figures 3A,B reveal, for the first time, that the yield of the spray-dried process is also related to the purity of the treated product. This is verified with spray-drying cell-free supernatant only, where the yield in LPP is quite low in the case of supernatant containing surfactin. For the latter two, a thick yellow paste is formed and sticks in an uncontrolled way all the parts of the spray-drier (data not shown). This phenomenon seems to be close to the one described by Barcelos et al. (2014), using crude LPP extract produced by B. subtilis LBBMA RI4914. A fermentation broth, even without cells, remains a complex matrix composed of numerous molecules that are more or less soluble in an aqueous solution (cell proteins, organic acids, biopolymers, sugars, etc.). This set of molecules (and particularly proteins) and the interactions they may have with each other when the solvent is removed can explain the formation of this sticky paste. Interestingly, in our study, this phenomenon does not appear when the supernatant contains only surfactin. Surfactin and plipastatin were produced by two strains derived from B. subtilis 168, thus showing a very similar potential of primary and secondary metabolite production profile (which is not the case for the mycosubtilin producing strain). Nevertheless, the overproduction of surfactin certainly limits the production of other metabolites by this strain, which may interfere with the spraying process. For surfactin, the yield of LPP recovered after the diafiltration step (diafiltered sample) is similar to the one obtained after the ethanol evaporation step (±80%). For plipastatin a slight decrease in the yield of LPP can be observed between the diafiltration and last purification steps (100%–78%). Nevertheless, for mycosubtilin, the yield of LPPs of the sample “purified fraction,” obtained after the ethanol evaporation step, is two times lower than the one obtained before this last step of purification (95% compared with 52%). This result seems to correlate with a decrease in the global spray-drying yield after this step of purification (Figure 3A). This is a surprising result, which led us to investigate deeper the case of mycosubtilin.

As shown in Figures 3A,B, a decrease in the yield of spray-drying of mycosubtilin after the last step of the purification process (ethanol evaporation) is observed. For a better understanding of this phenomenon, we studied the impact of the initial concentration of mycosubtilin in the solution to be spray-dried. Samples from 0.9 to 10 g·L⁻¹ of mycosubtilin were spray-dried to mimic the concentration conditions obtained after the diafiltration or after the ethanol evaporation steps of purification. The results are presented in Figure 4. Interestingly, the total yield and the LPP yield decrease with the increase in the concentration of mycosubtilin, until a limit concentration of approximately 3.5 g·L⁻¹. Indeed, results obtained for solutions below 3.5 g·L⁻¹ are 90% ± 10% for LPP yield and 83% ± 5% for total yield. For solutions above 3.5 g·L⁻¹, these results decrease to 50% ± 10% and 48% ± 10%. The ethanol evaporation process, which generates a high concentration, obviously has an important impact on the agglomeration phenomenon. The behavior of mycosubtilin in an aqueous solution is totally different from that of the other families of LPPs. It was shown that in the case of iturin A, micelles are formed at the CMC and larger vesicular structures at higher concentrations (Grau et al., 2001). In a more recent study, authors have investigated the behavior of the three families of LPPs using cryo-transmission electron microscopy, X-ray diffraction, and small-angle X-ray scattering. Their results confirm that mycosubtilin at high concentration has a distinct mode of self-assembly into extended nanotapes based on the stacking of LPP bilayers (Hamley et al., 2013). Moreover, Jauregi et al. (2013) have shown that mycosubtilin can interact with protein during the concentration/purification process by ultrafiltration (Jauregi et al., 2013). The amphiphilic, tensioactive, and uncharged nature of mycosubtilin and its ability to interact with proteins are important parameters to consider in a drying process. Indeed, it is likely to adsorb at the air–water droplet interface, where inadequate surface energies may expose the hydrophobic region and induce LPPs to aggregation, as it is well known for proteins (Lee, 2002; Ameri and Maa, 2007). To prevent this phenomenon, excipients/surfactants are commonly used for the drying process (especially for protein), but they also make the preparation stickier, which leads to a higher wall deposition and yield losses (Ameri and Maa, 2007). Knowing the propensity of mycosubtilin to agglomerate and its surfactant properties, it does not seem surprising to observe this wall deposition. Wall deposition is a key processing problem during spray-drying. This phenomenon is greatly impacted by the technical parameters of the apparatus (size and geometry of spray dryer, wall surface energy, etc.) and the operational conditions (inlet–outlet temperature, feed rates, and excipients) but not exclusively (Keshani et al., 2015). The impact of the concentration of the feed solution on the spray-drying efficiency has already been observed in the case of carbohydrates (Elversson and Millqvist-Fureby, 2005). Indeed, the concentration influences the particle size, with an increase in particle size until a plateau is reached. The authors also showed that the solubility of the carbohydrates had an impact on the particle size, the less soluble they were and the larger the particle size (Elversson and Millqvist-Fureby, 2005). Thus, these results show that the success of the spray-drying of this particular LPP is therefore also dependent on its concentration. It is advisable to limit its concentration below 3.5 g·L⁻¹ to obtain a good yield.

Finally, to deepen your knowledge on the direct spray-drying of the supernatant containing mycosubtilin (at low concentration), some additional trials were run to determine if the addition of charge (i.e., maltodextrin) can enable to obtain a dry product from a mycosubtilin-based supernatant. Maltodextrin is commonly used in spray-drying as a drying aid; it is used during the process to reduce not only the thermoplasticity and hygroscopicity but also the stickiness and product deposition (Keshani et al., 2015; Lee et al., 2018). Two proportions of maltodextrin were evaluated: 2.5% and 5% (w/v). In both cases, an analyzable dry product was obtained with the addition of maltodextrin. With 5% w/v of maltodextrin, the mycosubtilin supernatant containing 6.64% of dry matter was dried into a powder at 96.57% dry matter, with 1.39% LPP purity, and with a process yield of 70.18% and an LPP yield of 67.1%. With 2.5% w/v of maltodextrin, the mycosubtilin supernatant containing 4.33% of dry matter was dried into a powder at 91.32% dry matter, with 2.15% LPP purity, and with a process yield of 88.05% and an LPP yield of 89.76%. In both cases, the antifungal properties of mycosubtilin were only slightly impacted (compared with the value obtained with purified mycosubtilin, i.e., 4.1 mg·L⁻¹ [2.1–6.3]) with a MIC against S. cerevisiae of 5.2 mg·L⁻¹ [3.1–7.3]. This confirms the results obtained by Barcelos et al. (2014), who showed that the addition of maltodextrin at 10% is effective for the spray-drying of supernatant of B. subtilis-containing LPPs. Nevertheless, we demonstrated that 4 times less maltodextrin is sufficient as filler to make the spray-drying of these supernatant-containing LPPs easier.