Leuconostoc strains were isolated and identified from sugar beet factory juice and biofilm samples as described (Qi et al., 2025). The whole genome sequencing of these strains was previously reported (Joshi and Bruni, 2025). All strains were grown as precultures in MRS medium (De Man et al., 1960) at 28 °C, 250 rpm, unless specified otherwise. The strains used in this study are shown in Table 1.

Biofilms were grown in a Drip-flow six-chamber biofilm reactor (Biosurface Technologies, Bozeman, MT, USA). Each strain was cultured overnight in 50 mL of MRS medium (De Man et al., 1960) in a 250-mL flask at 28 °C, with shaking at 250 rpm. Bacterial cells were normalized to approximately 5 × 10⁸ colony-forming units (CFU) per mL, expressed as CFU/mL in Tryptone Sucrose Yeast (TSY) medium, 120 g/L sucrose, adapted from Haynes et al. (1955), and inoculum cell counts were confirmed by plating for CFU/mL (see next section). 1 mL of normalized bacterial culture was inoculated into 15 mL of TSY medium in each growth chamber. The growth chambers were clamped at the effluent port to retain the growth medium during the batch adherence phase. Then 0.22-μm filter vents were attached to allow for aerobic growth. The inoculated biofilm reactor was incubated at 28 °C for 48 h during the batch phase to facilitate adherence to stainless steel coupons. Samples were then taken for quantification of microbial CFU/cm² recovered from stainless steel coupons and measurement of biofilm dextran (see below).

For flow-through bioreactor experiments, Leuconostoc spp. were processed similarly to the adherence phase experiment for 48 h. After 48 h, continuous flow was initiated at room temperature with a flow rate of 0.9 mL/min using TSY medium (120 g/L sucrose). The bioreactor was attached to a shaker platform with a speed setting of 8 rpm with front-to-back shaking (Fisherbrand, Waltham, MA, USA). The bioreactor was secured to the shaker platform with bungee cords. The continuous flow-through phase was maintained for 72 h. Then, samples were collected for the quantification of microbial CFU/cm² and biofilm biomass.

Samples for both adherence and continuous flow biofilm material were removed from the topside of stainless-steel coupons using a sterile cell scraper into 30 mL sterile ultrapure water. Biofilm biomass was then homogenized using an Ika T10 homogenizer (Ika Works, NC, USA) at 20,000 rpm for 30 s. Biofilm homogenates were serially diluted in sterile ultrapure water and plated onto MRS agar with an Eddy Jet 2 W Automatic Spiral Plater in triplicate in linear 100 mode (Neutec Group, Inc., NY, USA). Plates were incubated at 28 °C for 1–2 days until colonies appeared. Colonies were counted with the SphereFlash Automatic Colony Counter (Neutec Group, Inc., NY, USA) using the SphereFlash Colonies PRO software version 1.0.0.16. to measure CFU/mL.

CFU/mL from triplicate plates were averaged and divided by the area of the stainless-steel coupon, 18.75 cm², resulting in CFU/cm² metrics for each growth chamber. Each strain was run in side-by-side duplicate growth chambers. The resulting CFU/cm² from duplicate growth chambers was averaged for each strain. L. suionicum BSDF2-3 was used as a positive control for each experiment, allowing for additional experimental strains to be run in duplicate during each experiment since the bioreactor has a total of six side-by-side growth chambers. Each experiment was run at least twice.

For both adherence and continuous flow experiments, biofilm biomass was precipitated with three volumes of absolute ethanol at 4 °C overnight. Samples were then centrifuged at 10,000×g at 4 °C for 30 min to pellet the biofilm biomass. The resulting pellet was lyophilized, weighed, and stored in a desiccator under vacuum.

To examine the adherence phase biofilms formed by each isolate, a thin layer of biofilm sample was mounted on aluminum SEM stubs using a cryo-embedding compound (Pelco, Ted Pella, CA, USA). The prepared stubs were placed on a temperature-controlled stage (Coolstage, Deben, UK), which was maintained at −20 °C throughout imaging to minimize dehydration and preserve surface integrity. Then, the samples were analyzed with a Phenom G6 ProX SEM (Nanoscience Instruments, Phoenix, AZ, USA) operated under low-vacuum conditions to accommodate the partially hydrated state of the biofilms. SEM images were acquired at accelerating voltages of 5 kV and 10 kV, with magnifications of 2,000× and 5,000×, respectively.

Biofilm samples harvested from adherence experiments were homogenized in 30 mL of sterile ultrapure water, serially diluted, and analyzed for dextran content using a dextran enzyme-linked immunosorbent assay (ELISA) kit (Beacon Analytical Systems, Saco, ME, USA). The calibration curve and sample data were processed according to the manufacturer’s recommendations. Data were analyzed using 4-parameter logistic (4PL) regression, and dilution factors were taken into consideration for each isolate.

Leuconostoc strains were grown, and the EPS was prepared as previously described (Bruni et al., 2024). Briefly, each strain was grown overnight as a preculture in MRS medium (De Man et al., 1960) and then inoculated into flask cultures at a starting OD of 0.05 in 50 mL TSY medium. Cultures were grown for 24 h at 28 °C with 250 rpm shaking. These flask cultures were used for both EPS and viscosity measurements.

After the 24-h incubation, OD₆₀₀ was measured with an Eppendorf Bio-Photometer (Hamburg, Germany), and flask culture viscosity was measured with a Brookfield model DV-11 + viscometer. Viscosity measurements were conducted using approximately 20 mL sample volume withdrawn from the flask culture, without dilution, using a UL spindle and small sample adapter at room temperature (Qi et al., 2025). Complete results from the viscosity analysis are given in the Supplementary material S1. One viscosity measurement was performed on three replicate culture flasks for each strain. For each sample, where possible, viscosity was measured at three different rotational speed instrument settings in the range of 5 to 50 RPM. Three samples, BSDF (5–1, 2–6, and 2–3), were substantially more viscous than other samples and required measurement at a rotational speed of 0.5 RPM (minimum instrument setting). For all other samples BSDF (48–3, 14–9, 47–1, 25–7, 62–9, 52–11), the recorded viscosity was taken as the measured viscosity at the lowest instrument speed setting, where the measured torque output exceeded a value of 10%; instrument manufacturer recommendation is to only use viscosity readings with associated torque values between 10 and 100%. Although shear thinning phenomena have been observed for some samples of this nature in a previous study (Qi et al., 2025), we assume Newtonian fluid behavior (Evageliou, 2020) here as a simplifying assumption in order to enable practical comparison between results and subsequent correlational analyses.

For EPS measurement, samples were centrifuged to remove bacterial cells, and supernatants containing soluble EPS were precipitated with three volumes of ethanol without DNase or proteinase treatment at 4 °C overnight. EPS was pelleted by centrifugation at 10,000 g for 30 min, and then lyophilized, weighed, and stored in a desiccator under vacuum (Bruni et al., 2024).

Precultures of each isolate were grown in MRS medium (De Man et al., 1960) to minimize EPS production and allow for pipetting non-viscous precultures. Optical densities (OD₆₀₀, hereafter OD) of precultures were measured in an Eppendorf Bio-Photometer (Hamburg, Germany) and normalized to 0.2 in TSY medium (50 g/L sucrose) (Haynes et al., 1955). Normalized precultures were pipetted into a 96-well plate in triplicate. A block design was employed to arrange the microbial isolates within the plates, and each block included triplicate blank wells without inoculum to serve as a reference. Plates were incubated at 28 °C with shaking at 250 rpm for 24 h in a TECAN plate reader (TECAN, Männedorf, Switzerland), which recorded OD values every 10 min using Magellan Standard software, version 7.5.

The Richards growth model was fitted to modified curves of the samples (inoculated wells). First, OD values from blank wells were subtracted from those of inoculated wells to generate adjusted OD values. To ensure accurate modeling of the exponential growth phase, data points occurring after the first instance of 99% of the maximum OD value for each curve were excluded, preventing the Richards growth model from overfitting the stationary phase. All model fitting was performed in R version 4.4.1 (Team R.C., 2016). The Richards growth model was fitted to these modified growth curves using the Richards growth function from the R package growth rates version 0.8.4 (Petzoldt, 2022). Visual inspection of the fits was conducted using the ggplot2 R package (Gómez-Rubio, 2017).

The maximum growth rate (mumax) was extracted from the fitted Richards growth equations (Richards, 1959). This value was then used to calculate the doubling time (t) using the exponential growth equation, N(t) = N(0)e^(rt), where N(t) is the final abundance, N(0) is the initial abundance, and ‘e’ is Euler’s number. The doubling time for each isolate was determined by setting N(t) = 2, N(0) = 1, and r = mumax, and solving for t. The doubling times derived for triplicate samples of each isolate were subsequently averaged.

Statistical analysis was conducted using GraphPad Prism Software (version 10.4.2, GraphPad Software, United States). The data represent at least two independent trials and are shown as the mean ± standard error. To compare multiple groups, a one-way ANOVA test was used, followed by a Tukey post hoc test. Significant differences were considered at p-values less than 0.05. Similarly, correlation analysis was performed using the Pearson R correlation test.

Leuconostoc strains isolated from juice and biofilm samples in sugar beet factories were examined for the ability to adhere, proliferate on abiotic surfaces, and form biofilms. This was assessed through quantification of viable cell counts and biofilm biomass after 48 h of the batch phase (Figures 1A,B and Supplementary material S2). All nine strains exhibited the capacity to adhere and proliferate on the abiotic surface, but with significant variation in CFU density and biomass accumulation across isolates. In Figure 1A, average CFU/cm² measurements indicated that strains L. suionicum BSDF25-7 and BSDF48-3 exhibited significantly higher adhesion and proliferation, each exceeding 5 × 10⁸ CFU/cm², while strains such as L. suionicum BSDF2-3, BSDF2-6, L. mesenteroides BSDF62-9, and L. citreum BSDF5-1 showed markedly lower CFU levels at 10⁷ CFU/cm². However, biofilm biomass quantification (Figure 1B) revealed that BSDF2-3 formed significantly denser biofilm matrices compared to other strains. Notably, strains such as L. suionicum BSDF25-7 and BSDF48-3 displayed comparatively low biomass, indicating potential differences in extracellular matrix production relative to cell density. Statistical analysis using one-way ANOVA followed by Tukey’s post hoc test confirmed significant differences among strains in both CFU and biomass measures. These findings highlight substantial strain-specific heterogeneity in biofilm-forming phenotypes, suggesting possible differences in cell surface properties, EPS production, or regulatory mechanisms underlying biofilm development in Leuconostoc isolates.

The ability of Leuconostoc strains to proliferate and form biofilms in a continuous flow bioreactor was evaluated in a flow-through bioreactor system for 72 h. Quantification of surface-associated viable cells (CFU/cm²) and total biofilm biomass showed clear strain-level differences in biofilm formation (Figures 2A,B).

L. citreum BSDF5-1 had the highest 1.14 × 10⁹ CFU/cm² in contrast to L. mesenteroides BSDF62-9, which was the lowest biofilm-forming strain with 1.33 × 10⁶ CFU/cm² (p value <0.0001), as shown in Figure 2A. Among L. suionicum, BSDF2-3 showed the highest count of 3.73 × 10⁸ CFU/cm², followed by BSDF2-6 with 2.55 × 10⁸ CFU/cm². These data suggest bacterial proliferation leading to increased CFU counts in 72 h. Biofilm biomass accumulation during the continuous flow-through phase showed trends similar to the batch phase biofilm biomass phenotypes. The highest average biofilm biomass was produced by L. suionicum BSDF2-3 compared to all other strains. However, L. suionicum BSDF47-1 maintained low CFU/cm² levels but produced moderate biomass. The minimal biofilm biomass produced by L. mesenteroides BSDF62-9 also corresponded with the low CFU/cm² values (Figure 2B).

SEM images revealed significant topological diversity in 48-h cultivated biofilm structures (Figures 3A,B and Supplementary material S3). At 2,000× magnification (Figure 3A), L. suionicum strains BSDF2-3, BSDF2-6, and BSDF47-1 formed dense, mat-like biofilms with relatively uniform surface coverage. In contrast, L. suionicum BSDF48-3 and L. citreum BSDF5-1 formed distinct, silo- or dome-like protrusions emerging from a sparsely colonized background. BSDF62-9 showed weak biofilm formation, presenting a loose, watery structure. Notably, BSDF48-3 exhibited a large, centrally aggregated pillar structure, suggesting vertical stratification and localized microcolony development. These elevated morphologies resemble the “mushroom” or “silo” structures commonly seen in mature biofilms of Pseudomonas and Staphylococcus, which are linked to nutrient gradients and protective EPS layers (Coraça-Huber et al., 2020; Wang et al., 2023).

At higher magnification (5,000×, Figure 3B), finer structural details became evident. Leuconostoc strains, such as BSDF62-9 and BSDF14-9, showed coarse, corrugated, net-like surfaces with loosely aggregated cells. In contrast, strains BSDF2-6 and BSDF52-11 exhibited sparse cell clusters interspersed with voids, indicating limited matrix formation and weaker cohesion. The strain BSDF48-3 showed a compact, elevated central mass, composed of tightly packed cells arranged radially, possibly indicating increased localized EPS deposition. Notably, strains L. citreum BSDF5-1 and L. suionicum BSDF2-3 demonstrated a mucinous, spiky monolayer-like conformation with vertical growth, suggesting a biofilm adapted for surface adhesion rather than volume expansion.

Next, in order to determine whether bacterial proliferation rate may impact biofilm formation, the doubling time of each Leuconostoc strain was determined under standard growth conditions to establish if biofilm formation phenotypes may be attributable to inherent growth rates (Figure 4). Substantial variation in growth rates was observed across all Leuconostoc strains, with doubling times ranging from approximately 24–53 min. L. suionicum strains BSDF52-11 and BSDF48-3 exhibited the fastest growth rates, with doubling times under 30 min, while L. suionicum strains BSDF2-3 and BSDF2-6 showed slower growth, with doubling times exceeding 50 min.

To evaluate the EPS production capacity among Leuconostoc strains, dextran levels were quantified from batch-phase biofilm samples using a dextran ELISA kit and analyzed using a 4-parameter logistic regression model (R² = 0.99; Figure 5). Substantial differences in dextran concentrations were observed among the nine Leuconostoc strains, with values ranging from approximately 31 to over 548 ppm. In particular, L. suionicum strains BSDF25-7 and BSDF48-3 exhibited the highest dextran production, consistent with high CFU/cm² densities and biomass accumulation observed in the batch-phase experiments. In contrast, strains such as L. suionicum BSDF2-6 and L. mesenteroides BSDF62-9 produced significantly lower levels of dextran. These results highlight extensive heterogeneity in dextran synthesis among Leuconostoc strains and underscore the potential contribution of dextran production to the robustness of biofilms.

EPS and viscosity were measured from 24-h Leuconostoc flask cultures to gain a comprehensive understanding of Leuconostoc strain EPS production capacity under planktonic conditions (Figure 6 and Table 2). Substantial variation was observed in EPS yield dry weight extracted from the flask cultures among the strains, with values ranging from less than 0.5 g to over 4 g per culture. Strains BSDF25-7 and BSDF48-3 produced the highest amounts of EPS, indicating strong extracellular matrix production under planktonic batch growth conditions. In contrast, strains BSDF2-3 and BSDF2-6 exhibited markedly lower EPS levels, which may be attributed to variations in EPS solubility when preparing EPS samples from cultures. Notably, others have reported that some Leuconostoc isolates produce insoluble EPS fractions that remain cell-associated (Padmanabhan and Kim, 2002; Côté and Leathers, 2009). This possibility and the impact of EPS structure on solubility warrant further investigation. The differences in EPS yield broadly align with trends observed in the dextran ELISA results and biofilm biomass accumulation, supporting the notion that EPS production is a key factor in strain-specific biofilm phenotypes.

Furthermore, viscosity measurements demonstrated that strains exhibited varying levels of viscosity in the respective flask culture broths (Table 2). Since measured viscosities span orders of magnitude, values are reported both as—is and as log₁₀—transformed. Due to the wide variation in culture viscosity, culture samples were read at different rpm in order to obtain viscosity measurements with a torque greater than 10. L. suionicum BSDF2-3 and BSDF2-6 strains produced the highest average viscosity, measured at 3.04 ± 0.01 log₁₀ and 3.08 ± 0.01 log₁₀ cP, respectively. L. citreum BSDF5-1 exhibited the next highest viscosity of 2.79 ± 0.07 log₁₀ cP (Table 2). The remaining Leuconostoc isolates had lower viscosities ranging between 0.48 and 1 cP. Taken together, the results indicate that Leuconostoc strains possess diverse capabilities in modifying the viscosity of the growth environment through the production of various EPS.

Finally, the functional contribution of the EPS production capacity of the respective Leuconostoc strains was assessed. Pearson correlation was performed between the dextran levels measured in batch phase biofilms and the extracted EPS dry weight of flask cultures (Figure 7A) since the biofilm biomass from the adherence phase was very low for some strains. A significant positive correlation (R² = 0.85, p = 0.00037) was observed between the EPS weight and the dextran concentration, indicating that strains with greater EPS yields also produced higher quantities of dextran (Figure 7A).

Next, Pearson correlation was performed for the viscosity of the 24-h flask cultures and OD₆₀₀ (Figure 7B). Similarly, a significant positive correlation was found (R² = 0.94, p = 0.000013), suggesting that OD₆₀₀ contributes to the rheological properties of the culture medium (Figure 7B). Interestingly, the observed viscosities showed poor correlation with EPS yield (Figure 6), suggesting additional variables such as OD₆₀₀, EPS structure, and solubility may also impact viscosity. These findings highlight the central role of strain-specific EPS biosynthesis in modulating both biochemical (dextran content) and biophysical (viscosity) characteristics of Leuconostoc cultures.