This study focused on two strains of L. plantarum, labeled as c19 (a wild strain isolated from Bella di Cerignola table olives and with some functional properties; Bevilacqua et al., 2010) and DSM 1055 (from the German Collection of Microorganisms, Braunschweig, Germany), and on the collection isolates B. animalis subsp. lactis DSM 10140, B. longum subsp. longum DSM 20219, and B. longum subsp. infantis DSM 20088.

The strains of L. plantarum were stored at −20°C in MRS broth (Oxoid, Milan, Italy) added with 33% of sterile glycerol (J.T. Baker, Milan, Italy), while Bifidobacterium spp. were stored at −20°C in MRS broth supplemented with 0.5% cysteine (cMRS) (Sigma-Aldrich, Milan, Italy) added with 33% of sterile glycerol; cysteine creates a reducing environment for Bifidobacteria. Before each assay, the microorganisms were grown under anaerobic conditions either in MRS broth or cMRS broth, incubated at 37°C for 24 h in jars. Then, the cultures were centrifuged at 4,000 × g for 10 min at 4°C; the supernatant was discarded, and the pellet suspended in distilled water.

Bacteria suspension at ∼10⁷ CFU/ml, prepared as reported in section “Microorganisms,” were treated through a VC Vibra Cell Ultrasound equipment, model VC 130 (Sonics and Materials Inc., Newtown, CT, USA; net power, 130W) as follows. 20 ml of bacterial suspension were put in 50 ml sterile tubes; then, ultrasonic probe was placed 1–2 cm below the surface and in the middle of tube. Each strain was US-treated through 6 different combinations, different for power (from 10 to 50% of the net power of equipment, i.e., 130 W), pulse (from 0 to 10 s), and duration of the treatment (from 2 to 10 min). These three variables were combined through a centroid approach, as reported in Table 1. According to the centroid approach each variable was set to 3 different levels, identified by the codes 0 (minimum), 1 (maximum), 0.5 (mean value). Table 1 shows the 6 combinations of the centroid and the control (i.e., an additional combination in which power, treatment time, and pulse were set to 0, that is microorganism not treated through US). US was done in aerobic conditions.

Before each treatment, the ultrasonic probe was washed with sterile distilled water, ethanol at 70% and again with distilled water; immediately after processing, the sample was cooled in ice. After each US treatment, a plate count was performed on MRS agar (Lactobacilli) or cMRS agar (Bifidobacteria), incubated at 37°C for 48 h; plates were incubated under anaerobic conditions (jars and anaerobic kit).

Bacterial suspensions treated through US and control strains were used to inoculate broth containing glass slides (25.4 mm × 76.2 mm), as surface adhesion for biofilm formation. Before each experiment, it is important to perform a preliminary treatment on glass slides to remove any fingerprints, grease and other impurities that might be present on the material. Thus, the slides were washed with acetone for at least 30 min, rinsed in distilled water and immersed in 1 N NaOH for 1 h. After a final rinse in distilled water, the slides were allowed to air dry. Finally, the slides were autoclaved at 121°C for 15 min before use. After the treatment, the slides were placed in 50 ml-tubes (one slide per tube) containing 45 ml of MRS broth (Lactobacilli) or cMRS broth (Bifidobacteria). The final step for sample preparation is the inoculation of tubes with slides and broth with US bacteria, adding 200 μl of the suspensions prepared as reported in section “US—Treatment” (US-treated suspensions and untreated bacteria) to gain an initial inoculum of ∼10⁵ CFU/ml).

After preparation and inoculation, samples were incubated at 37°C in jars; for each microorganism 7 different combinations were tested (6 US combinations and untreated microorganism).

After 1, 5, 7, 9, 12, and 16 days, slides were aseptically removed from the culture medium, rinsed with sterile distilled water to remove unattached cells, and placed in a tube containing 40 ml of sterile saline solution (0.9% NaCl). Then, the samples were sonicated at 20% power for 3 min to promote detachment of cells from the surface (Speranza et al., 2009). Microbiological analyses were done on this saline solution on either MRS agar or cMRS agar, incubated at 37°C for 48 h in jars.

All tests were performed in duplicate on two independent samples for each combination and for each strain. Cell concentration recovered from saline solution (section “Biofilm formation”) was converted to log CFU/cm² through the following formula:

where C₁ is the viable count expressed in log CFU/cm²; C₂ is the viable count expressed as log CFU/ml; VS is the volume during the sonication treatment (40 ml) and SA the area of adhesion tested (39 cm², surface area of the slide considering both sides).

Data of L. plantarum were analyzed through a multifactorial Analysis of Variance, using strain (c19 or DSM 1055), age of biofilm or sampling time (1, 5, 7, 12, and 16 days) and combinations of US treatment (control, or combinations from 1 to 6) as categorical predictors; Fisher LSD test was used as the post-hoc test. Statistic was done through the software Statistica for Windows (Statsoft, Tulsa, OK, United States).

Afterwards, data at day 1 for B. infantis and after 16 days for L. plantarum DSM 1055 were analyzed through a multiple regression approach through the option DoE/mixture design of the software Statistica to assess the significance of the three parameters of US treatment (power, pulse, and duration).

Table 2 shows the viable count of US-treated bacteria (log CFU/ml), compared to the control (CNT, untreated strains), immediately after the treatment; US did not reduce cell count, which was between 7.65 ± 0.068 log CFU/ml (L. plantarum) and 6.86 ± 0.02 log CFU/ml (Bifidobacterium spp.) in the control, against values of 7.71 ± 0.03 log CFU/ml (L. plantarum) and 6.23 ± 0.15 log CFU/ml (Bifidobacterium spp.) recovered after the most drastic treatment (combination 1, 50% of the net power, 2 min, pulses at 0 s), thus suggesting that US did not affect the viability of the test strains and could be used for a positive modulation of biofilm formation.

Generally, Bifidobacteria and L. plantarum showed different adhesion properties, with a very low stability of biofilm for Bifidobacterium spp.; therefore, the strains were separately analyzed to avoid a confounding effect due to the higher count of sessile cells in L. plantarum.

Ultrasound treatment did not affect the adhesion properties of B. longum subsp. longum and B. animalis subsp. lactis strains. After 24 h, the concentration of sessile cells was 6.3 and 5.85 log CFU/cm² for B. longum subsp. longum and B. animalis, respectively, and US treatment did not change this trend; in addition, the count of sessile cells experienced a strong reduction in all samples and was below the detection limit after 5–7 days (data not shown).

On the other hand, a significant effect was found for B. longum subsp. infantis (Table 3); after 1 day, biofilm produced by US-treated microorganism was at higher levels than the control (5.0 log CFU/cm² vs. 5.80–6.87 log CFU/cm²) and the highest count was found in the combination 1 (treatment at 50% of power, for 2 min, no pulse). Moreover, this combination also resulted in a higher stability of biofilm, with a residual concentration of sessile cells at 3.70 log CFU/cm² after 9 days, while for untreated microorganism or for cells treated with other US combinations sessile cells were below the detection limit.

As reported elsewhere L. plantarum strains showed different trends; moreover, the stability of biofilm was higher than that found for Bifidobacteria. Finally, the two strains behaved in a different way, thus requiring a preliminary standardization of data as biofilm detachment or reduction of biofilm count compared to the day 1; therefore, data for these two strains should be read as reduction of the count of sessile cells throughout time.

Modeling was done through MANOVA; the table of standardized effects points out that all predictors, as individual or interactive terms (combination, strain, age of biofilm, combination × strain, combination × age of biofilm, strain × age of biofilm, combination × strain × age of biofilm), were found to be significant; however, their statistical weights were different. The most important factor, as an individual term, was the age of biofilm, followed by the strain, and finally by the combination of US-treatment; on the other hand, the most important interactive term was “strain × age of biofilm” (Table 4).

The estimation of the quantitative effects of predictors is possible through the decomposition of the statistical hypothesis, which does not show real trends but the mathematical correlation of each predictor vs. the dependent variable.

Figures 1–3 show the decomposition of the statistical hypothesis for the individual effects of predictors (combination of US treatment, strain, and age of biofilm). Regarding the effect of the combination (Figure 1), the reduction of sessile cells (biofilm detachment) was minimum in combination 3 (power, 10%; duration of the treatment, 2 min; pulse, 10 s), while the other combinations did not show significant differences compared to control. The two strains produced a biofilm with a different stability, as generally L. plantarum DSM 1055 experienced a higher biofilm detachment (ca. 2.0 log CFU/cm²) (Figure 2); as expected, biofilm age negatively affected its stability, as the detachment increased over time (Figure 3).

The approach of the decomposition of the statistical hypothesis was also used for the estimation of interactive terms. Figures 4, 5 show two-way interactions, namely for the interactions strain × age of biofilm (Figure 4) and combination of US treatment × strain (Figure 5).

The first interactive term highlights the higher stability of biofilm produced by the strain c19 mainly after 12 and 16 days, while for the different combinations the decomposition of the statistical hypothesis points out a higher stability of c19 biofilm in the control and in the combination 2 and 6.

Finally, Figure 6 shows 3-way interaction, that is the actual values; after 12 days significant differences were observed between the two strains in all combinations, showing a higher detachment of the biofilm for the strain DSM 1055. However, for this strain, a US treatment before biofilm formation is crucial, as biofilm detachment after 16 days was 5.5. log CFU/cm² in the control, 3.2 log CFU/cm² for the microorganism treated with the combination 3 and 4.4 log CFU/cm² when the microorganism had been treated with combinations 1, 4, and 5.

Due to the significant effect of US on the viable count of sessile cells at day 1 for B. longum subsp. infantis and at day 16 for L. plantarum DSM 1055, the data for these sampling points were standardized as difference of sessile cells between US-treated microorganisms and untreated microorganisms and analyzed through a multiple regression procedure to understand how the three parameters of US (power, duration of the treatment, and pulse) could improve biofilm stability.

The first result of this approach is a set of standardized effects (Fisher-test values), which allows us to understand which variables were significant and which, among them, exerted the strongest effect. From a statistical point of view, for B. longum subsp. infantis the most important term was the power (F-test, 5.76), followed by the duration of the treatment (F-test, 3.81), and pulse (F-test, 3.47); the interactive terms (power × time, power × pulse, pulse × time) were not significant. In the case of L. plantarum, the most significant terms were power (F-test, 2.99) and pulse (F-test, 2.61).

However, the standardized effects show only the significance of the variables and do not allow the estimation of their quantitative effect. A better description can be obtained through the triangular plots; for B. longum subsp. infantis the triangular plot points out a higher value of biofilm for power-coded values around 1 (50% of net power), time 0 (treatment for 2 min), pulse 0 (0 s) (Figure 7). For L. plantarum DSM 1055 the highest difference US-treated strain vs. control (that is a higher stability of biofilm after US treatment) was found for the highest values of power or pulse (50% power and pulse at 10 s) or with the combination 0.5 power+0.5 pulse, as coded values, corresponding to real values of 30% of net power and 6 s of pulse, while an increase of the duration of the treatment exerted a detrimental effect on biofilm stability (Figure 8).