New insights into the role of plasmids from probiotic Lactobacillus pentosus MP-10 in Aloreña table olive brine fermentation

In silico analysis of Lactobacillus pentosus MP-10 plasmids (pLPE-1 to pLPE-5) suggests that plasmid-borne genes mediate the persistence of lactobacilli during olive fermentation and enhance their probiotic properties and their competitiveness in several ecological niches. The role of plasmids in the probiotic activities of L. pentosus MP-10 was investigated by plasmid-curing process which showed that plasmids contribute in increased metal tolerance and the biosequestration of several metals such as iron, aluminium, cobalt, copper, zinc, cadmium and mercury. Statistically significant differences in mucin adhesion were detected between the uncured and the cured L. pentosus MP-10, which possibly relied on a serine-rich adhesin (sraP) gene detected on the pLPE-2 plasmid. However, plasmid curing did not affect their tolerance to gastro-intestinal conditions, neither their growth ability under pre-determined conditions, nor auto-aggregation and pathogen co-aggregation were changed among the cured and uncured L. pentosus MP-10. These findings suggest that L. pentosus MP-10 plasmids play an important role in gastro-intestinal protection due to their attachment to mucin and, thus, preventing several diseases. Furthermore, L. pentosus MP-10 could be used as a bioquencher of metals in the gut, reducing the amount of these potentially toxic elements in humans and animals, food matrices, and environmental bioremediation.

fermentation is the oldest practice by our ancestors to preserve vegetables and to also produce different flavours and textures. Additionally, fermented table olives remain an important economy for many production countries and a component of the Mediterranean diet (and recommended as part of the Healthy Eating Pyramid published in 2010, https://dietamediterranea.com/). The high nutritional value of fermented table olives (e.g., their content of carbohydrates, fiber, minerals, vitamins, fatty acids, and amino acids) and their role as potential source of probiotic lactobacilli of vegetable origin 1-5 make them very attractive from an economic and social point of view. Lactobacillus genus is the most representative and heterogeneous member of lactic acid bacteria (LAB) group currently consisting of 237 species (as of December 2018 in www.bacterio.net) since they harbour in their genome a plethora of genes involved with a wide array of functional properties 6,7 . Lactobacillus spp. are principal bacteria in olive fermentation processes, possessing many biochemical and physiological traits to ferment several carbohydrates and tolerate stress 8 . These phenotypes are important as the brine environment represent harsh conditions for bacterial growth with low nutrient availability, saltiness, low pH and the presence of antimicrobials (e.g., phenolic compunds and oleuropein); thus, highly robust L. plantarum and L. pentosus are frequently isolated from the end of olive fermentation 1,8,9 . Furthermore, Perpetuini, et al. 10 demonstrated by transposon mutagenesis that the high capacity of L. plantarum and L. pentosus to survive in the hostile, brine environments was due to critical genes encoding proteins involved in carbohydrate metabolism, membrane structure and function, and
In silico analysis of plasmid properties in L. pentosus MP-10. Analysis of the annotated CDs of each L. pentosus MP-10 plasmid revealed the presence of five genes involved in mobilization (mobA gene) distributed in all plasmids except the pLPE-2 plasmid (Tables 2-6). These genes are likely required for plasmid relaxation and mobilization by conjugative plasmids. Also, conjugation-related genes were found, e.g., traG in pLPE-4 (traG_1 and traG_2) and pLPE-5 (traG_3) plasmids (Tables 5 and 6). A gene encoding for a bacteriophage peptidoglycan hydrolase that may have been involved in growth was found in pLPE-4 (XX999_00013 and XX999_00049) and pLPE-5 (XX999_03566) plasmids (Tables 5 and 6).
Chloride-(clcA_2) and sodium-(nhaS3_4) transport genes harboured by pLPE-2 plasmid ( Continued copy of the same genes clcA_1, nhaS3_1, nhaS3_2 and nhaS3_3 were also found in L. pentosus MP-10 chromosome with the aim to potentiate chloride and sodium tolerance in brines. Genes related to carbohydrate metabolism were found on plasmids (besides on the chromosome) such as L-Lactate dehydrogenase in pLPE-5 plasmid (ldh_7 and ldh_8 genes) ( Table 6), genes involved in glucose uptake and metabolism such as glcU_1 and gdhIV_1 genes in pLPE-3 plasmid (Table 4), and a gene involved in xylan catabolic process (axeA1_3) in pLPE-5 (Table 5). However, another gene involved in xylan catabolic process (XX999_00089) was only detected in pLPE-3 plasmid, but not on the chromosome (Table 4).
Toxins reported in L. pentosus MP-10 plasmids include mazF-toxin encoding gene (XX999_03521) detected in pLPE-1 plasmid, genes coding for Zeta toxins in pLPE-3 (XX999_00053) and pLPE-4 (XX999_00024) plasmids, and also for antitoxins such as RelB antitoxin (XX999_00026) in pLPE-4 plasmid and the bifunctional antitoxin/ transcriptional repressor RelB in pLPE-5 plasmid (XX999_03554) (Tables 2, 4-6). MazF toxin is a desirable property in probiotic bacteria, and is only detected in plasmid DNA of L. pentosus MP-10, not in the chromosome. However, L. pentosus MP-10 has to protect itself from the MazF toxin without any MazE antitoxin. On the other hand, RelB antitoxins were found both on plasmids and on the chromosome; however, no RelB toxins were detected. Zeta toxins were detected both on the chromosome (one gene) and also on plasmid DNA (two genes); however, no antitoxin was detected.
Other coding genes for several functions, such as a serine-rich adhesin for platelets precursor (sraP gene), were detected in pLPE-2 plasmid but not on the chromosome (Table 3); genes coding for vitamin biosynthesis such as panE_1 and panE_2 genes coding for 2-dehydropantoate 2-reductase (biosynthesis of vitamin B5), a gene XX999_00068 coding for prephenate dehydratase (biosynthesis of phenylalanine, tyrosine and tryptophan), were detected on the pLPE-3 plasmid (Table 4) and also on the chromosome.
Regarding their responses to stress, in-silico analysis of plasmid sequences revealed the presence of yhdN_1 gene coding for a general stress protein 69 (in pLPE-3, Table 4) and several genes coding for metal tolerances, such as cadmium [cadmium resistance transporter (XX999_03594) and a putative positive regulator of cadmium resistance (cadC)] and two operons of arsenic resistance (in pLPE-5, Table 6). One ars operon consists of arsR_3 (arsenical resistance operon repressor ArsR) and arsB [arsenical pump membrane protein (ArsB)], but lacks arsC gene (arsenate reductase ArsC); the other ars operon contains arsA [arsenical pump-driving ATPase (ArsA)] and arsD gene [arsenical resistance operon trans-acting represor (ArsD)] in pLPE-5 ( Table 6). The synteny of arsenic-resistance genes was examined by comparing the annotated sequences of pLPE-5 and pWCFS103 plasmids (aligned by MAUVE algorithm) from L. pentosus MP-10 and L. plantarum WCFS1, respectively. Comparison revealed that the synteny of genes was similar (Fig. 2), being arsenic operons in pLPE-5 of L. pentosus MP-10 composing of two copies each gene: arsB [coding for trivalent As(III) efflux permease ArsB], arsA [coding for trivalent As(III)-stimulated ATPase ArsA], arsD [coding for trivalent As(III) metallochaperone ArsD] and arsR_3 gene [a trivalent As(III)-responsive repressor (ArsR)]. On the other hand, arsC gene (arsC2 coding for reductase ArsC), as a part of ars operon with arsB and arsR genes, was found in L. pentosus MP-10 chromosome, as well as two arsR gene copies (arsR_1 and arsR_2).  Continued absence of 6.5% NaCl, different pH ranges (1.5 to 7.0), nor the presence of bile salts (1.8 or 3.6%) -no differences in 600 nm absorbances were detected over 24 h of incubation-(Figs S1-A,B, S2). In a similar manner, pH monitoring during their incubation also did not exhibit any significant differences between cured and uncured strains in regards to their acidification capacity ( Fig. S1-C). Furthermore, no differences in the growth were detected between the cured and uncured L. pentosus MP-10 strains in the presence of xylan as the only carbohydrate source ( Fig. S1-D). However, at high salt concentration of 8% usually found in brine, significant differences were detected between the cured and uncured L. pentosus MP-10 strains, with the uncured strain being the most tolerant ( Fig. S1-E). Table 7 shows that curing had no significant effect on the growth of uncured and cured L. pentosus MP-10 in the presence of phenolic compounds naturally present in the brines; both the cured and uncured strains tolerated more than 200 mg/ml of olive-leaf extract.

In vitro
Effect of plasmid curing on metal tolerance. Plasmid annotations predicted gene clusters involved in arsenateand/or arsenite-, and cadmium resistance. First, we precisely determined metal concentrations that inhibit the visible growth of the wildtype L. pentosus MP-10; results showed that this strain tolerated high concentrations of metals depending on the metal with 1 < MIC < 4096 μg/ml, and tolerances were observed to be in order Fe > [Al/ Cu/Co] > Zn > Cd > Hg (Table 8). When we compared the uncured and the cured L. pentosus MP-10, we found that mercury and cadmium exibited different MICs among strains by 2-8 fold increase (Table 8) in those uncured; as such, plasmids have a key role in mercury and cadmium tolerances.
The removal of different metals was shown in Table 8, which demonstrated that L. pentosus MP-10 was able to remove different metals, thus exhibiting high removal capacity of mercury (81.74% ± 2.04), cadmium (67.10% ± 0.88) and aluminium (57.14% ± 0.99). However, the cured L. pentosus MP-10C demonstrated statistically significant reduced performance. Metal removal differences between the uncured and the cured L. pentosus MP-10 highlight the role of plasmids to remove iron, cadmium, aluminium, cobalt, copper, zinc and mercury (Table 8).
To understand how L. pentosus MP-10 interact with selected metals, SEM analysis was performed and showed the biosorption potential of the uncured L. pentosus MP-10 ( Fig. 3). The micrographs and EDX spectra obtained before and after the biosorption process showed clearly that the cell morphology of the uncured L. pentosus MP-10 changed and exhibited the presence of bright particles on the surface of the bacteria exposed to some metals. Regarding cadmium, mercury and zinc, we couldn´t detect these metals by EDX analysis. Furthermore, in the presence of either aluminium, cobalt, copper, mercury or zinc, higher potential for biofilm formation was observed. These results, confirmed by EDX analyses, support that these metals remained adsorbed entirely on the cell surface.
Effect of plasmid curing on antimicrobial resistance and probiotic features. We determined the MIC of different antibiotics and biocides between uncured and cured strains, and the results did not show any significant differences in response between both strains except for clindamycin, which exibited 20 fold increase in the MIC in the uncured L. pentosus MP-10. Thus, plasmids have no role in the suceptibility to the antibiotics and biocides tested, except clindamycin (Table 7).
Adhesion to mucin was measured in both the uncured and the cured L. pentosus MP-10, and the results showed a statistically significant increase in adhesion capacity to mucin in the uncured L. pentosus MP-10 ( Table 7).

Discussion
Olive brine represents a stressful environment for the growth and survival of many bacteria due to the harsh conditions (i.e., high salt concentration, presence of phenolic compounds and low-nutrient availability), which provide selective pressures for the maintenance of LAB. As such, L. plantarum and L. pentosus have the genetic tools to survive and grow in the hostile olive-brine conditions 10 , and these genetic traits are widely distributed on both the chromosome and the plasmids, with several genes having multiple copies to enhance their adaptability and fitness in different ecological niches.
In this study, L. pentosus MP-10, isolated from Aloreña green table olives, harboured five plasmids with an average GC content (39.52-42.50%) slightly lower than the host chromosome (46.32%), this difference was less than 10% as reported by Nishida 25 for the majority of plasmids. pLPE-5 had remarkably the lowest average GC content (39.52%) than the other four plasmids (pLPE-1, pLPE-2, pLPE-3 and pLPE-4), suggesting it is possibly a recent acquisition from another bacterium. In-silico analysis of plasmid sequences revealed the presence of genes involved in mobilization (mobA) and conjugation (traG) distributed in several plasmids, which suggest their role in gene mobilization and secretion using a type-IV secretion mechanism 26 . Furthermore, mobile genetic elements (e.g., transposon, transposase, integrase and invertase) were also found in several plasmids 2 suggesting a frequent genetic diversification among the L. pentosus MP-10. Furthermore, bacteriophage peptidoglycan hydrolases were found in pLPE-4 and pLPE-5 plasmids; these lysozyme-like proteins may play a key role in L. pentosus MP-10 growth, its cell-wall structure, and immunomodulatory properties as reported by Rolain, et al. 27 .
Metabolic profile within L. pentosus MP-10 plasmids include carbohydrate enzymes such as L-lactate dehydrogenase, glucose uptake and metabolism and xylan catabolic enzymes. L-lactate dehydrogenase was codified by two genes (ldh_7 and ldh_8) located on pLPE-5 plasmid; however, six L-lactate dehydrogenase (ldh_1, ldh_2, ldh_3, ldh_4, ldh_5 and ldh_6) and four D-lactate dehydrogenase (XX999_00315, XX999_00955, XX999_02047 and XX999_02719) coding genes were also present on the chromosome. Both enantiomers (L-lactate and D-lactate) are produced by L. pentosus MP-10 being D-and L-lactate dehydrogenases involved in the reversible metabolism of D-and L-lactate, respectively. This finding is of great interest suggesting that the use of L. pentosus MP-10 as a probiotic may help human to metabolise D-lactate obtained from exogenous sources (e.g., diet and the carbohydrate-fermenting bacteria normally present in the gastrointestinal tract) since mammalian cells lack sufficient D-lactate dehydrogenase required to utilise D-lactic acid-leading to chronic fatigue syndrome and D-lactic acidosis or D-lactate encephalopathy associated with short bowel syndrome [28][29][30] . Further, L-lactate dehydrogenase genes present on the plasmids may enhance their metabolic activity during the fermentation process to produce more L-lactate and energy. However, the presence of L-lactate dehydrogenase (ldh_7 and ldh_8) coding genes on pLPE-5 plasmid did not enhance the acidification capacity, as results were similar after 8 and 24 h incubation in both cured and uncured L. pentosus MP-10, suggesting that these genes either have a minor role in lactate production or they are regulated. Further experiments, based on differential relative expression of ldh gene in both the cured and uncured L. pentosus MP-10 strains, revealed low expression level in the cured strain (Fig. S3), thus the low activity of lactate dehydrogenase gene in the cured strain is enough to give rise to a substantial lactate accumulation in the fermentation broth in a manner similar as the uncured strain. Regarding glucose uptake and metabolism, glcU_and gdhIV genes were over-expressed in the uncured L. pentosus MP-10 indicating the role of plasmid in this process (Fig. S3).
Among defense mechanisms found on plasmids, gene encoding the mazF toxin (pLPE-1), Zeta toxins (pLPE-3 and pLPE-4), and also antitoxins such as RelB antitoxin (pLPE-4) and the bifunctional antitoxin/transcriptional repressor RelB (pLPE-5) were detected in L. pentosus MP-10 plasmids. RelBE and MazEF are known as sequence-specific endo-ribonucleases that inhibit the global translations of cellular mRNAs 31 . MazF toxin is a desirable trait for probiotic bacteria, as its antimicrobial property inhibits several pathogens in foods and the gastrointestinal tract 32 . However, L. pentosus MP-10 must protect itself from the mazF toxin, as no MazE antitoxin was detected. Either their protection relies on other mechanisms because mazF is functional being only expressed in the uncured strain (Fig. S3). On the other hand, genes for RelB antitoxins were found both on plasmids and on the chromosome; however, no RelB-toxin genes were detected. So this antitoxin may contribute a greater defense against other bacteria possessing RelB toxins, possibly increasing its competitiveness and survival in  www.nature.com/scientificreports www.nature.com/scientificreports/ several ecological niches including gastrointestinal tract. This feature was mainly linked to plasmid being relB antitoxin gene over-exopressed in the uncured strain (Fig. S3). Zeta toxins, which are kinases that kill bacteria through global inhibition of peptidoglycan synthesis 33 , are detected both on the chromosome and also on plasmid DNA of L. pentosus MP-10, however no antitoxin was detected. Overall, L. pentosus MP-10 harbored in their plasmids incomplete toxin-antitoxin systems unlike what occur naturally in bacterial genomes, since several toxins or antitoxins were detected without self protection.
Data obtained by in-silico analysis suggests that plasmid-borne genes mediate the persistence of lactobacilli under olive fermentation conditions and enhance their probiotic properties; however, this hypothesis requires further studies for confirmation. As such, plasmid curing experiments carried out with L. pentosus MP-10 showed several differences between the uncured and the cured strains regarding metal tolerances, removal and mucin adhesion. However, plasmid curing did not affect their tolerance to gastro-intestinal conditions (e.g., acids and bile salts); neither their ability to grow under determined conditions (i.e., different pH intervals, bile salts or sodium chloride of 6.5%) nor their colony morphology were changed after plasmid curing (data not shown). However, at high concentration of chloride of 8% (commonly added to brines), L. pentosus pLPE-2 plasmid plays a key role in salt tolerance. In this sense, the results suggest that the plasmids did not govern the fermentation of carbohydrates under these conditions, however different results were obtained by Adeyemo and Onilude 34 which showed that plasmid curing had a significant negative effect on growth, physiological characteristics and colony morphology of L. plantarum isolated from fermented cereals. In this study, plasmids in L. pentosus MP-10 may confer a selective advantage, providing other physiological properties in certain environments such as gut and brines and thus allowing metal tolerance and removal, salt tolerance and adherence to mucin and thus their persistence in competitive ecological niches. Mucin adhesion declined in the cured L. pentosus MP-10 since a serine-rich adhesin for platelets precursor gene (sraP, detected in pLPE-2 plasmid) may be involved in mucin adhesion mechanisms similarly as reported by Hevia, et al. 34 for an extracellular serine/threonine-rich protein as a novel aggregation-promoting factor with affinity to mucin in Lactobacillus plantarum NCIMB 8826. The role of L. pentosus MP-10 plasmids in mucin adhesion was confirmed by relative expression gene analysis as reported by Pérez Montoro et al. 35 , since recA and pgm genes considered as potential biomarkers of mucin adhesion were over-expressed in the uncured strain (Fig. S3). However, auto-aggregation and co-aggregation with some pathogens were not changed after plasmid curing of L. pentosus MP-10.
With respect to metals, which are considered non-biodegradable and non-thermodegradable and are of high concern in both developing and developed countries because of their impact on the environment and health (water and food), the wild strain L. pentosus MP-10 showed greater tolerance to their increased concentrations (MICs higher than 1 mg/ml, except for cadmium and mercury) of iron, cobalt, copper, aluminium and zinc. This suggests that high contamination of metals in the environment from natural and anthropogenic sources 36 may be tolerable by the bacteria. The self-protective mechanisms displayed by L. pentosus MP-10 as a response to metals is promoted by their architecture (cell wall and membrane) and also by their resistance determinants located on the chromosome and the plasmids. Moreover, several chromosomally encoded cation transporters (e.g., encoded by czcD gene) have a predicted substrate range, including cadmium, cobalt and zinc; although the increased  www.nature.com/scientificreports www.nature.com/scientificreports/ resistance towards different metals are displayed by plasmids (especially the pLPE-5 plasmid). Similar results were obtained by van Kranenburg et al. 22 , which reported that the plasmid-borne (pWCFS103) cadC gene coding for a transcription regulator of the cadmium operon was responsible of the increased resistance to cadmium in L. plantarum WCFS1. Furthermore, the synteny of ars genes in both L. pentosus MP-10 and L. plantarum WCFS1 22 was similar suggesting their evolutionary relatedness. Arsenic and cadmium are among the most toxic elements widely ocurring in the environment, often a threat to food and water supply. Arsenic is known as a group A "known" carcinogen according to the United States Environmental Protection Agency (USEPA) and contributes to a range of other illnesses such as cardiovascular and peripheral vascular diseases, neurological disorders, diabetes mellitus and chronic kidney disease [37][38][39] . Detoxification of this metal was earlier established by bacteria. Thus, tolerance of L. pentosus MP-10 is necessary to prevent damage to their cells.
The ability of L. pentosus MP-10 to bind different metals was demonstrated by SEM and EDX analysis. This is of great importance with regards to their application as an adjunct to improve food safety and quality by bioquenching metals and probiotically reduce metal toxicity among human intestinal microbiota and thus protecting the host 40 . Also, we demonstrated that L. pentosus MP-10 contributed to metal removal, especially mercury and cadmium (81 and 67%, respectively).
Metal-and antibiotic-resistance genes often co-exist on the same plasmid, however in this case, we did not find any genes coding for clindamycin resistance on plasmids, which was the only antibiotic with different susceptibility after plasmid curing. Thus, clindamycin resistance in L. pentosus MP-10 may rely on other plasmid-associated genes that we could not deciphered yet.

conclusions
In-silico analysis of L. pentosus MP-10 plasmids suggests that plasmid-borne genes mediate the persistence of lactobacilli under olive-fermentation conditions and enhance their probiotic properties with genes encoding for carbohydrate metabolism, defense mechanisms, metal tolerance and mobilization increasing subsequently its competitiveness and survival in several ecological niches. Plasmid curing demonstrated the role of plasmids in the increased metal tolerance, and bioremoval of several metals (e.g., iron, aluminium, cobalt, copper, zinc, cadmium and mercury). This probiotic property by L. pentosus MP-10 should be exploited to detoxify metals in intestines; basically they could bioquench the metals in the gut thus reducing their toxic exposure to humans and animals, in the food matix and in environmental bioremediation.

Materials and Methods
Bacteria and growth conditions. Lactobacillus pentosus MP-10 isolated from naturally-fermented       Growth properties. To test whether there is any differences in growth between the uncured and the cured L. pentosus MP-10 strains, MRS broth was inoculated (1% v/v) with overnight cultures of each strain and then incubated at 37 °C for 24 h. Growth rates (OD 600nm ) were measured each hour using Microtiter plate reader (iMark Microplate Absorbance Reader, Bio-Rad instrument). Additionally, we measured pH at different time intervals (following 0, 8 and 24 h of incubation at 37 °C).
To determine the effect of pH on the growth of both strains, MRS broth was adjusted to different pH ranges (1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5 and 7.0) with phosphate buffer, and they were inoculated (1% v/v) overnight cultures of both strains and then incubated at 37 °C for 24 h, as described above. www.nature.com/scientificreports www.nature.com/scientificreports/ To test whether brine conditions had an effect on the growth of the plasmid-cured versus uncured L. pentosus MP-10 strains in MRS broths under the following experimental conditions: unsupplemented vs. those supplemented with either 6.5% (or high concentration of 8%) NaCl or phenolic compounds, or modified MRS broth (without glucose) added with xylan (5 g/l) were inoculated with both strains as described above. Phenolic compounds were obtained from freshly pulverized olive leaves using RETSCH laboratory ball mills (Retsh MM 400). The leaf extracts were resuspended in LSM broth, centrifuged and the resulting supernatant was filtered (0.45 μm) and added at different concentrations (0.780 to 200 mg/ml) to MRS broth. The cultures were incubated at 37 °C for 24 h and the OD 600nm was measured as described above.
In all cases, experiments were done in triplicate.
To analyse the removal of metals by cured and uncured L. pentosus MP-10, MRS broth supplemented with ½MIC of each metal was inoculated with 2% (v/v) of an overnight culture of each strain and then incubated 24 h at 37 °C. After incubation, the bacterial cells were removed by centrifugation and kept for the subsequent examination of metal sorption. The resulting supernatants were filter sterilized using a 0.22 μm filter (Millipore, Spain) and then used to check metal removal. MRS broth added either with different metals (with ½MIC) or not were used as positive and negative controls, respectively. The positive controls (MRS broth with individual metal added: Fe at 2 mg/ml; Al, Co and Cu at 1 mg/ml; Zn at 0.5 mg/ml; Cd at 4 μg/ml and 0.5 μg/ml; and Hg at 1 μg/ml and 0.5 μg/ml) were considered "100%" baselines to calculate relative metal removal rates (as a percentage).
Metal concentrations were measured using 7900 ICP-Mass Spectrometer (Agilent, USA) with graphite tube atomizer and autosampler, a superior matrix tolerance and advanced collision/reaction cell (CRC) technology to remove the polyatomic interferences that can affect some of the trace elements. The spectrometer software was Agilent ICP-MS MassHunter Work Station, which provides simple autotuning functions, and a Method Wizard automates the method setup process.
Biosorption of metals by L. pentosus MP-10 was further examined using scanning electron microscope (SEM) coupled with energy dispersive X-ray spectroscopy before and after metal uptake. For this, a drop of the bacterial pellet, which had been previously exposed to a metals (as previously described), were disposed into microporous capsules (ANAME, Spain), dried and then dehydrated in a series of 20, 40, 60, 80, and 100% ethanol solutions (15 min each) before suspension in acetone for 1 h. After this, the capsules were subjected to critical-point drying before examination by SEM (FESEM, MERLIN de Carl Zeiss, Oxford).
Safety and probiotic properties. To determine differences in antimicrobial (antibiotic and biocide) susceptibility of L. pentosus MP-10C versus wild strain, we determined the MIC of several antimicrobials following the method previously described by Casado Muñoz, et al. 42,43 using LSM broth (Oxoid).
To determine if plasmids further play a role in several probiotic peroperties, we analyzed acid-and bile-tolerances, auto-aggregation, co-aggregation with pathogens (L. innocua CECT 910, S. aureus CECT 4468, E. coli CCUG 47553, and S. Enteritidis UJ3449) and mucin adhesion in both L. pentosus strains (MP-10 and MP-10C) according to the methods reported by Pérez Montoro et al. 35 .
Gene expression analysis. To analyse the role of plasmid in several metabolic and probiotic properties, both the uncured and cured L. pentosus strains were subjected to RNA extraction using Direct-zol ™ RNA Miniprep (Zymo Research, California, USA) according to the manufacturer's instructions. RNA quantification and quality assessment were carried out by using a NanoDrop 2000 spectrophotometer (Thermo Scientific). RNAs were adjusted to a concentration of 500 ng/ml and frozen at −80 °C until required for analysis.
The expression of selected genes (Table S1) was determined by quantitative, real-time PCR (qRT-PCR) using SensiFASTTM SYBR & Fluorescein One-Step Kit (BIOLINE) as reported in Pérez Montoro et al. 35 .
Statistical analysis. All analyses were performed in triplicate. Statistical descriptors were calculated using Excel 2007 (Microsoft Corporation, Redmond, Washington, US), e.g., determining averages and standard deviations. Statistical comparison of growth and probiotic properties assays were conducted by analysis of variance (ANOVA) using Statgraphics Centurion XVI software (Statpoint Technologie, Warrenton, Virginia, US). The same software was used to perform Shapiro-Wilk and the Levene tests to check data normality and to perform 2-sided Tukey's multiple contrast to determine the pair-wise differences between strains. Level of significance was set at P < 0.05.