Lactobacillus plantarum 299v (Probi AB) was utilized to evaluate method suitability, with inoculation commercial product powder formulation containing lyophilized cells and cryostocks of the isolated strain. Two inoculation methods were employed. First, the full capsule content of a commercial product (approximately 370 mg, claimed to contain at least 10 billion bacteria) was resuspended in 10 ml of peptone water (casein peptone 10 g/L, sodium chloride 5 g/L) and revived at room temperature for 30 min. Second, fresh colonies from de Man-Rogosa-Sharpe (MRS) agar plates were used to inoculate 2 ml of MRS (Millipore) in 15 ml plastic conical tubes. These tubes were capped and incubated overnight at 37°C and rotated at 180 r.p.m.

For the metabolic activity study using acidification kinetics, cytofluorimetric enumeration, and IMC in parallel, two strains, Lacticaseibacillus rhamnosus GG ATCC 53103 and Limosilactobacillus fermentum LF10 DSM 19187, both from the Probiotical SpA collection, were used. The probiotic strains were activated overnight at 37°C in MRS (Difco, BD, MD) broth and then sub-cultured, using at least two passages over the mid-log phase.

A 10-fold dilution series of the bacterial suspensions of L. plantarum in peptone water was prepared. The isothermal microcalorimeter plate, equipped with titanium vials and plastic inserts, was prepared by adding 30 μl of the individual sample dilutions to 270 μl of sterile MRS broth. The vials were sealed and placed into a calScreener™ isothermal microcalorimeter (Symcel AB, Sweden) in accordance with the manufacturer’s guidelines. Kinetic heat flow was monitored using the calView 2.0 software (Symcel AB, 2023), which recorded heat flow curves over a period of 48 h. Various parameters were extracted using the online analysis tool calData.¹

To assess the impact of turbid media viability counts, commercial plant-based milk (pasteurized oat drink, Oatly Barista Edition) was used. For culturing, 1:1 dilutions of MRS/oat drink were inoculated with serially diluted L. plantarum overnight culture in MRS, and 300 μl of the dilutions were added to vials for calorimetric measurements, as described above.

de Man-Rogosa-Sharpe agar plates were used to determine counts of viable colony-forming units (CFUs). Samples were serially diluted in peptone water in 10-fold steps. Of the diluted sample, 50 μl were spread on top of the agar of the plates, and subsequently incubated at 37°C for 48 h. Following incubation the colonies were counted and the viability value (CFU/ml) was calculated.

For strains GG and LF10, the BD Cell Viability Kit (BD Biosciences, Milan, Italy) was utilized to quantify viable cells (AFU/ml) and total cells (TFU/ml). Cell staining procedures were conducted in accordance with ISO 19344: IDF 232 (2015). Briefly, 100 μl of a diluted suspension containing approximately 10⁵–10⁶ cells/ml in buffered peptone water was mixed with 885 μl of PBS. Subsequently, 10 μl of propidium iodide (previously diluted in water at 0.2 mmol/L) and 5 μl of thiazole orange (42 μmol/L) were added to the dilution, followed by vortexing. The CytoFLEX cytometer (Beckman Coulter SRL), equipped with 488 nm laser excitation and CytExpert software, was employed for analysis. Thresholds for side scatter (SSC-H) and forward scatter (FSC-H) were established for microbial cells, which were gated using forward versus side scatter (FSC-H vs. SSC-H). The optimal discrimination between live and dead populations, used for the enumeration of AFU and TFU, was achieved on an FL1 versus FL3 plot. No counting beads were added to the diluted samples as internal standards as the instrument is designed for volumetric counting, whereby the concentration of events (cells) is determined based on the defined sample volume drawn by the needle.

The overnight cultures of GG and LF10 were inoculated in 30 ml of three growth media of different formulations (medium A, medium B, and medium C), with a final concentration of 10⁷ AFU/ml.

For medium A, we employed MRS as the standard reference for the growth of Lactobacilli. Media B and C had two distinct formulations, each intentionally designed with differences in salt composition and primarily focusing on variations in carbon source concentration and nitrogenous source composition. The standardized cultures were placed at 37°C in a water bath for 24 h and pH variation during fermentation was measured at intervals of 4 min with the iCinac pH monitoring system (KPM Analytics, USA). At fixed time points (t = 0, 3, 5, 7, and 24 h) an aliquot of sample was taken and analyzed by CytofFLEX cytometer, as described above.

Simultaneously, microcalorimetric analysis was performed by adding 10⁷ AFU in each vial provided with a plastic insert and 300 μl of the three different media. As described in section “2.2 Microcalorimetric analysis for viability enumeration,” kinetic heat flow was monitored over a period of 48 h and heat flow curves were generated using the calView 2.0 software. Various parameters were extracted using the online analysis tool calData.

Standard curves were generated using both freshly prepared and freeze-dried cells. The initial cell concentration for fresh cells was approximately 3 × 10⁹ CFU/ml. For rehydrated freeze-dried cells the concentration was approximately 10¹⁰ CFU/ml. The time to peak (TTP) values for each sample (liquid and freeze-dried) were calculated as a viability marker, and GraphPad Prism 10.0.2 (232) was employed for further analysis. The regression curves were constructed by plotting TTP against the logarithm of the cell concentration (CFU/ml). To analyze the reproducibility of IMC for viability assessments an unpaired two-tailed t-test was used in the same software.

To examine the suitability of IMC for viability assessment and enumeration of probiotic products, we utilized a 48-channel isothermal microcalorimeter, the calScreener™, to test the reproducibility of the method.

In this assessment, a freeze-dried commercial probiotic product containing L. plantarum was diluted in fresh MRS medium and evenly distributed into 32 titanium vials. This plate was then introduced into the microcalorimeter, where over 24 h the instrument continuously monitored the heat flow from the samples and generated heat flow curves (thermograms). Notably, the readings from the individual sample vials gave consistent results, as indicated by the minimum SD (range 2.906–0.03 μW for the collected data points) of the thermograms. The same experiment was repeated on a different day, using a different instrument and a new batch of freeze-dried bacteria from the same product (Figure 2). An unpaired two-tailed t-test revealed no statistically significant difference [t(582) = 0.0258, p = 0.9794] between the two groups. This indicates that the observed difference in the means of the two thermograms is likely due to random chance and not a true underlying effect.

In our efforts to quantify the viability of L. plantarum, we conducted a series of experiments involving serial dilutions from overnight cultures, which were then inoculated into fresh MRS medium for isothermal calorimetric measurements at 37°C. The resulting thermograms consistently showed the thermal fingerprint of L. plantarum emerging at predicted intervals, confirming the method’s suitability. There was a linear relationship between the initial inoculum size and the appearance of the thermograms (Figure 3A). We chose to analyze the data using the time to metabolic peak parameter to illustrate the principle. Time to signal detection, another calorimetric parameter extracted from the thermograms, exhibited a linear relationship with an initial number of microbial cells. Determining this parameter involved applying a threshold value set at a certain μW value to the data.

In parallel, the initial cell number was determined by plate count assay and a calibration curve was plotted. To create a standard curve for inoculum estimation, we applied a linear regression model to the data.

This approach allowed us to obtain a standard curve specific to strain under specified conditions, such as medium (here MRS), volume, and temperature (Figure 3B).

To further challenge the method’s robustness, we investigated whether the linear relationship between inoculum size and TTP held true under different pre-culture and incubation conditions. We first used a dilution series of L. plantarum in a one-to-one MRS medium-oat drink mixture. The thermogram profiles remained consistent under these defined conditions, exhibiting the expected rightward shift with lower initial bacterial loads (Figure 4). Importantly, the strong linear relationship between inoculum size and TTP was maintained (Table 1).

Secondly, we explored the impact of pre-culture conditions by using three independently processed, freeze-dried samples. The standard curves generated from these samples closely resembled those obtained with the original liquid culture (Figure 5). As expected, the linear regression models for all three independent experiments again demonstrated a strong linear relationship between inoculum size and TTP (Table 1). To further solidify this finding, an F-test revealed no statistically significant difference between the slopes of the five linear regressions (F = 1.569, DFn = 4, DFd = 27, P = 0.2111), reinforcing the method’s robustness across culture conditions.

Generalizing this standard curve across a broader range of strains and growth conditions may require further validation.

Empirical data involving various media formulations has shown that combining biomass quantification using time point measurements and metabolic activity measurements provide a comprehensive view of microbial activity and viability under diverse conditions. Especially during the substrate or media optimization process, it is relevant to consider various parameters.

Here, we executed three different experiments – IMC, flow cytometry for viability enumeration, and acidification measurements – in parallel on two representative probiotic strains, L. rhamnosus and L. fermentum (Figure 6). We demonstrated how information obtained by isothermal calorimetry, such as metabolic activity (heat flow over time) and peak metabolic activity gives an indication of viability, with information on growth, metabolic dynamics, and biomass formation during medium fermentation.

Unlike IMC, flow cytometry does not permit kinetic measurements and samples are taken at different time points. For the measurements, cells were sampled at t = 0, 3, 5, 7, and 24 h and viable counts expressed as active fluorescent units (AFU) (Figures 6B, D). When compared to thermograms, there was a correlation between metabolic activity profiles and biomass data obtained by flow cytometry (Figure 6). In particular, it is evident how the formulation of medium B, properly designed for strain activation, proves to be more active in stimulating and supporting growth compared to the standard medium for lactobacilli (MRS) and the third formulation, medium C, which is less effective in reaching high cell counts. For L. fermentum, we see a first, smaller metabolic peak at around 3–4 h when cultivated in medium B or C. This indicates a metabolic adaptation phase with a metabolic shift as indicated with arrows in Figure 6C. No such information was revealed in acidification curves, nor from the time point biomass measurements.

Additionally, we see that low biomass formation, early acidification, and higher acidification are associated with lower overall metabolic activity and a reduced metabolic peak. Figures 6C, D illustrate that IMC accurately reflects growth dynamics. For instance, the delayed and lower biomass formation measured by flow cytometry of L. rhamnosus in medium C is mirrored in the delayed metabolic activity signal, a nuance not captured by the acidification curves.