Production of L. fermentum K73 biomass was performed following the protocol described by Aragón-Rojas et al. (2018). Briefly, the biomass of bioreactors was produced by inoculating MRS broth at 10% (v/v) with L. fermentum K73 and incubating at 37 °C for 12 h. After that, a bioreactor of 1 L (Bioflo 110, New Brunswick Scientific Co, Inc., United States) with culture medium prepared with 8% whey and 0.22% yeast extract, adjusted to pH 5.5 (sterilized at 121 °C for 15 min) was inoculated with the previously prepared inoculum and fermented during 10 h at 37 °C and 100 rpm.

The viability of the probiotic was measured by plate counting. 1 g of microcapsules were decapsulated, plated in duplicate on MRS agar, and incubated at 37 °C for 24 h. Colonies were then counted and expressed as log CFU/mL (Aragón-Rojas et al., 2019).

The following formula was used to calculate the CFU/mL:

In order to express the results of microbial viability in a more concise and manageable way, the values obtained in colony-forming units per milliliter (CFU/mL) were transformed to their logarithm in base 10 (log CFU/mL). The result of CFU/mL was then calculated as the logarithm in base 10 to obtain the result in log CFU/mL.

The viability of the encapsulated microorganisms through in vitro digestion was performed following the standardized INFOGEST protocol proposed by Minekus et al. (2014). The three phases of digestion were performed for all encapsulates: oral, gastric, and intestinal. In the oral phase, 1 g of each powder was mixed with simulated salivary fluid (SSF) in a 1:1 ratio (w/v) without adding α-amylase, since the microcapsules did not contain starch, and kept under stirring for 2 min at 37 °C. At the end of the oral phase, it was mixed with simulated gastric fluid (SGF) and adjusted to pH 3 with HCl; in this phase, porcine pepsin was added to reach 2000 U/mg (Sigma, P6887-5G), and the gastric phase was stirred for 2 h at 37 °C. Finally, the intestinal phase was obtained by mixing the solution obtained in the gastric phase with simulated intestinal fluid (SIF) containing bile extract (Sigma, B8631-100G), trypsin (Sigma, P7545-25G), and lipase (L3126-100G), to reach final concentrations of 10 mM, 100 U/mL, and 2000 U/mL, respectively. The pH was adjusted to 7 with 1 M NaOH and stirred at 37 °C for 2 h.

After each phase, 100 μL aliquots were collected to measure the viability of the probiotic. The methodology for this measurement is described in section 2.3.

The evaluation of the hypocholesterolemic activity of probiotic L. fermentum K73 was carried out at 4 different points of the process: (i) after 24 h of growth in MRS broth, (ii) after 10 h of fermentation in the bioreactor with the medium composed of whey; (iii) after the three encapsulation processes (C1, C2 and C3); (iv) the C1 microcapsule was subjected to in vitro digestion.

The following points were used to determine the molecular identification of the presence of the BSH enzyme: (i) after 24 h of growth in MRS broth, (ii) after 10 h of fermentation in the bioreactor with the medium composed of whey; (iii) Subsequent to the encapsulation process for C1, (iv) the C1 microcapsule was subjected to in vitro digestion.

RNA extraction of L. fermentum K73 in all the points mentioned above and to two more standard strains as controls (L. fermentum ATCC 9338) was carried out using TRIzol reagent, followed by cDNA transcription using the One Script Plus cDNA Synthesis Kit (commercial kit from abm), following their protocol, utilizing Random primers. Once the cDNA was obtained, primers specific for Limosilactobacillus fermentum BHS enzyme were employed using the conventional PCR technique, with the following conditions: One Tag 2x (BioLabs), forward primer 0.2 mM, reverse primer 0.2 mM, cDNA 5 ng/μL. The thermocycler conditions were set to an initial denaturation at 95 °C for 3 min, denaturation at 95 °C for 30 s, annealing at 60 °C for 45 s, and extension at 72 °C for 30 s. This was followed by a final extension at 72 °C for 3 min over 35 cycles, resulting in the production of a 628 bp fragment. Subsequently, the quality of amplified products was analyzed in 1.5% agarose gel, with the cDNA bands visualized using the Quantity One software (Bio-Rad) version 4.6.8.

The evaluation of probiotic adhesion to cells was performed as described by Archer et al. (2018). The cell lines in question were obtained from the American Type Culture Collection (ATCC) [HTB-38 HT-29, and (CRL-2577)] RKO cells derived from human colorectal adenocarcinoma. For this, the two tumor lines were maintained in Dulbecco’s modified Eagle’s minimal essential medium supplemented with 10% (v/v) of inactivated fetal bovine serum, penicillin (100 μg/mL) and streptomycin (100 μg/mL). Cells were incubated with 5% CO₂ at 37 °C during all the assays. For the adhesion assay, 200 μL of the suspension of each cell line was grown in a 96-well plate to a density of 4 × 10⁴ cells/well and maintained under the same incubation conditions until a monolayer was formed. Then the cells were incubated for 2 h with eight different treatments by quadruplicate. (i) L. fermentum K73 growth in MRS broth for 24 h; (ii) Bioreactor after 10 h of fermentation L. fermentum K73 in a food grade medium; (iii) C1 dissolved; (iv) C2 dissolved; (v) C3 dissolved; (vi) C1 after intestinal phase; (vii) C2 after intestinal phase; (viii) C3 is digested after the intestinal phase.

After the incubation period, the cells were washed three times with PBS, fixed with methanol, and stained with Gram stain for microscopic examination. The count of adherent bacterial cells was performed as described by Chauviere et al. (1992), taking 20 microscopic fields at random from each treatment and counting each bacillus present in the analyzed area. This approach expressed probiotic adherence as the number of bacteria adhering to 100 HT-29 and RKO cells.

The R statistical software (version 4.4.3) was utilized to analyze and visualize the data. A one-way analysis of variance (ANOVA) was conducted, with a 95% confidence interval. Subsequently, Tukey’s post-hoc test for multiple comparisons was applied, considering a significance level of p < 0.05.

The measured properties showed that microencapsulates were reproducible. Briefly, the capsules presented lower values of moisture content (less than 5%), which improves the stability of the product during shelf life (Table 1). In addition, lower water activity values favor the quality of the encapsulates and the viability of the probiotic L. fermentum K73 contained therein. This can be compared with that reported by Aragón-Rojas et al. (2020), (C2) used in this work had moisture of <4% and water activity (aw) of <0.2, on the other hand, according to the authors Vargas et al. (2024) and Agudelo-Chaparro et al. (2022) with another type of encapsulation with maltodextrin, the water activity fluctuated between 0.23 and 0.44 and 0.16 to 0.54, respectively.

It has been demonstrated that a product’s water activity (Aw) and moisture content are critical factors that significantly influence microbial stability and viability. Specifically, aw values below 0.61 have been shown to inhibit the growth of pathogenic microorganisms due to limited water availability (Tapía et al., 2020).

However, in the context of probiotic microcapsules, particularly those of Lactobacillus species obtained by spray drying, the relationship between water activity (Aw) and survival is more nuanced. Optimal probiotic survival during storage at room temperature has been reported over an aw range of 0.11 to 0.76 (Liu et al., 2016). This observation is consistent with the findings of studies indicating a significant loss of viability in products containing Lactobacillus rhamnosus with an aw of 0.938 after 28 days of storage at room temperature (Zavaleta et al., 2022).

On the other hand, C3 has a homogeneous size among the beads, obtaining an average of 1.8 ± 0.3 mm, a size like that reported by Rojas-Muñoz, Santagapita and Quintanilla-Carvajal, (2023), where they obtained beads with a size of 1.6 ± 0.2 mm. Also, it was observed that 86% of the beads had a circular shape without presenting irregular edges.

The production of L. fermentum K73 biomass reached a concentration of 8.41 ± 0,29 log CFU/mL after 10 h of the fermentation process. After this, three encapsulation systems were produced accordingly to Bauer Estrada et al. (2025), Aragón-Rojas et al. (2020), and Rojas-Muñoz et al. (2023). The viability of L. fermentum K73 after the production of each one of the microencapsulates was 6.99 ± 0,55, 6.45 ± 0,21, and 8.25 ± 0,49 log CFU/mL, for C1, C2, and C3, respectively.

Compared to the bioreactor, the C1 showed a 1.42 log CFU/mL reduction in viable cell count, decreasing from 8.41 to 6.99 log CFU/mL. This loss, although present, is notably lower than in other formulations and may be attributed to the protective effect provided by the encapsulation design, particularly the inclusion of HOPO oil. The incorporation of oil likely contributed to a more robust barrier against external stressors during the encapsulation process and subsequent gastrointestinal simulation. This observation aligns with findings reported by Yin et al. (2024), who demonstrated that using short-chain oils in double emulsion systems significantly enhanced probiotic survival rates from 43.23 to 65.16% by providing an additional protective layer during spray drying. The presence of oil in the formulation stabilizes the emulsion. It reinforces the encapsulation matrix, suggesting that oil-containing capsules, such as C1, offer superior protection and are more effective in preserving probiotic viability under challenging conditions.

The C2 formulation exhibited a viability loss of 1.96 log CFU/mL, decreasing from 8.41 to 6.45 log CFU/mL compared to the bioreactor, which can be compared to what is described by Bhagwat et al. (2020), as they proved that encapsulation with maltodextrin and whey as wall materials leads to loss of the viability of probiotics when they raise the outlet temperature of the spray dryer higher than 60 °C, causing cellular damage such as denaturation of genetic material, ribosomal damage and dehydration of cytoplasmic membranes of probiotics (Bhagwat et al., 2020). Similarly, Liu and collaborators explain that the use of maltodextrin can influence survival, as they observed that encapsulations with this material have a lower protective capacity for probiotics, due to its high solubility, generating a faster exposure of the microorganism to environmental and process conditions (Liu et al., 2019).

On the other hand, the encapsulation using sodium alginate and sweet whey to produce the C3 formulation demonstrated improved probiotic survival, with only a minimal viability loss of 0.16 log CFU/mL, decreasing from 8.41 to 8.25 log CFU/mL compared to the bioreactor. This claim can be substantiated by the encapsulation efficiency data presented by Vega-Carranza et al. (2021), which demonstrated the viability of approximately 99.97% ± 0.22% after encapsulation of B. licheniformis. In addition, this good encapsulation efficiency can be compared with the ones obtained by Raise et al. (2020), which presents high survival rates of the microorganism after in vitro digestion, showing that the technique favors the microorganism’s viability and gives it the necessary protection during passage through the gastrointestinal tract. It was observed that there was no contamination in any process.

Among the three encapsulation types evaluated, statistically significant differences in the viability of L. fermentum K73 were observed only in the C2 formulation when compared to the bioreactor. Tukey’s post hoc test revealed a reduction of 1.96 log CFU/mL in C2 relative to the bioreactor, with a statistically significant adjusted p-value of 0.0172. In contrast, the differences observed for C1 (−1.42 log CFU/mL, p = 0.0611) and C3 (−0.16 log CFU/mL, p = 0.9908) were not statistically significant.

After the encapsulation process, an in vitro digestion was performed to observe the behavior of the encapsulates in the gastrointestinal tract. Here, it was possible to observe that the three encapsulates did not lose logarithmic cycles when in contact with the oral phase (Table 2).

On the other hand, when passing to the gastric phase with low pH, the C2 and C3 capsules lost only one cycle; on the contrary, the C1 encapsulated with palm oil remained stable. This phenomenon has been demonstrated in the study of Eratte et al. (2017), which revealed that the survival of L. casei increased in SGF when co-encapsulated with omega-3 oil, exhibiting a mere 1.5 log CFU/ml decrease. Likewise, although the three encapsulations use whey and the protein is denatured, flocculation was generated, which reduces the area of interaction between the protein and pepsin, thus reducing the protein digestion of the three encapsulations, in addition to the hydrophobic interaction generated between the aromatic amino acids, as presented by Aragón-Rojas et al. (2020). As stated in the work of Zhu et al. (2021) observations were made indicating that soy lecithin also favors protein flocculation. This is corroborated by what was found in this project since the release of the probiotic is done in the intestinal phase and not in the gastric phase, and C1 performs better than the other types of encapsulates (Table 2).

As for the intestinal phase, the three encapsulations showed reductions in viability of 5.08, 2.91, and 5.24 log CFU/mL for capsules C1, C2, and C3, respectively. This finding indicates that the encapsulation processes enable the probiotic to reach the intestine at a specific concentration. However, factors such as the stress induced by the pH shift between the phases and the activity of enzymes present in the intestinal phase can influence the probiotic’s survival, as previously reported by (Masco et al., 2007; Ranadheera et al., 2012).

To assess the cholesterol absorption capacity of Limosilactobacillus fermentum K73, a calibration curve was established using cholesterol concentrations ranging from 10 to 70 μg/mL (Figure 2). Based on prior findings indicating superior performance under simulated gastrointestinal digestion, only the C1 encapsulate was selected for further evaluation in the post-digestion stage, including the assessment of cholesterol removal and the presence of the bsh gene.

As shown in Table 3, L. fermentum K73 demonstrated cholesterol absorption capacities of 63.9% in 24-h MRS broth culture and 62.3% in the bioreactor, corresponding to 44.67 and 43.56 μg/mL of absorbed cholesterol, respectively. These values are slightly lower than those Aragón Rojas (2012) reported, who documented an absorption of 53.06 μg/mL (75.8%) by the same strain. Among the three encapsulation strategies tested, C1 exhibited the most favorable outcome, achieving an absorption of 23.74 μg/mL (33.91%) in its dissolved state, outperforming C2 and C3.

These results suggest that the encapsulation process, while critical for ensuring survival during transit, may impose a trade-off on probiotic functionality, possibly due to incomplete release of viable cells during in vitro decapsulation Zhou et al. (2024). Despite this, the C1 formulation demonstrated remarkable recovery post-digestion. Following the simulated intestinal phase, C1 exhibited the highest cholesterol uptake across all experimental conditions, reaching 45.84 μg/mL, equivalent to 65.49% absorption.

This improvement may be attributed to the complete release of viable probiotic cells facilitated by enzymatic activity and matrix disintegration in the intestinal phase. According to Melchior et al. (2022), digestion-mediated release mechanisms, particularly those involving protein clots formed by encapsulation matrices such as whey proteins, can enhance probiotic bioactivity. Furthermore, Padhmavathi et al. (2023) and Asgari et al. (2020) emphasize that the compatibility of wall materials with gastrointestinal conditions is crucial in safeguarding probiotic integrity while enabling functional performance.

Additionally, the cholesterol-lowering effect observed may be mechanistically linked to BSH-mediated bile salt deconjugation. Öner et al. (2014) reported that strains such as Bifidobacterium longum and Lactobacillus plantarum could precipitate cholesterol in bile salt-enriched media due to their deconjugation of taurocholic acid (TCA), leading to reduced cholesterol solubility and enhanced removal. The current study suggests a similar mechanism may be at play in L. fermentum K73, which showed greater affinity for TCA and TDCA. These findings reinforce the hypothesis that the enhanced cholesterol absorption observed post-digestion in C1 may be driven by effective deconjugation of tauroconjugated bile salts and subsequent cholesterol precipitation.

The presence of the bsh gene, with an expected amplicon size of approximately 638 bp, was successfully confirmed through conventional PCR in both control and experimental samples, corroborating the results obtained in the semi-qualitative bile salt deconjugation assay and the cholesterol absorption analysis. The amplification reactions yielded clean and distinct bands without evidence of nonspecific products or primer dimers, confirming the specificity of the primers and the integrity of the cDNA used (Figure 3).

Significantly, the gene was consistently amplified from L. fermentum K73 across different stages of processing, including growth in MRS broth (Figure 3, Lane 2), fermentation in a bioreactor (Lane 3), and encapsulation via spray drying in capsule formulation C1 (Lane 4). These findings suggest that the strain retains the bsh gene in its genome regardless of the applied biotechnological processes and over time, indicating genetic stability and functional conservation of this cholesterol-lowering mechanism.

In line with these observations, Jayashree et al. (2014) reported the presence of two bsh-related genes, bsh1 and bsh2, in another L. fermentum strain, with larger amplicon sizes of 927 bp and 978 bp, respectively. Similarly, Liang et al. (2018) identified homologous bsh sequences across seven complete L. fermentum genomes through in silico analysis. Complementary to these findings, Rani et al. (2017) described a 975 bp bsh gene in Lactobacillus gasseri FR4, which includes a conserved catalytic cysteine residue (Cys1) crucial for enzymatic activity in the presence of bile salts. These studies support the diversity of bsh gene variants among Lactobacillus species and highlight their role in bile salt metabolism and microbial adaptation to the intestinal environment.

However, it is notable that in Figure 3, Lane 5—corresponding to C1 post-digestion—no amplification product was detected, indicating an absence of detectable bsh gene expression following simulated gastrointestinal digestion. This phenomenon may be attributed to the physiological stress encountered during digestion, such as drastic pH fluctuations, exposure to bile salts, and enzymatic activity, which may impair cellular function or reduce transcriptional activity as part of a survival strategy (Foster, 2005). Han et al. (2021) similarly emphasized that gastrointestinal transit conditions can compromise microbial cell integrity and DNA stability, potentially leading to reduced gene expression or damage to genetic material.

These results underline the necessity of incorporating gene expression quantification techniques, such as real-time quantitative PCR (qPCR), in future studies. Such approaches would enable not only confirmation of gene presence but also assessment of transcriptional dynamics under different physiological conditions, providing a more complete understanding of the role of bsh in probiotic function.

The adhesion capacity of Limosilactobacillus fermentum K73 to intestinal epithelial cells was evaluated using two cell lines: HT-29 and RKO. The adhered bacteria were quantified according to the protocol described by Chauviere et al. (1992). The results obtained for the HT-29 cell line are shown in Figure 4.

The highest adhesion rate to HT-29 cells was recorded in the 24-h broth culture, with 2,488 ± 24.11 adhered to the bacteria. This value exceeds those reported for Lactiplantibacillus plantarum strains, which exhibited adhesion levels of approximately 800 (Muryany et al., 2018). It is also more consistent with results from Zawistowska-Rojek et al. (2022), who observed values ranging from 1800 to 2,200. These findings confirm the strong adhesion potential of L. fermentum K73 in its unencapsulated form.

Following encapsulation and subsequent processing, a significant reduction in adhesion was observed across treatments. ANOVA confirmed statistically significant differences between treatments (p < 0.001), and Tukey’s post hoc test indicated that adhesion in the 24 h broth was significantly higher than all other groups (p < 0.001). The bioreactor treatment retained moderate adhesion (195 ± 8.10), while C1, C2, and C3 dissolved capsules showed reduced values of 21 ± 0.72, 5 ± 0.25, and 33 ± 0.81, respectively. After in vitro digestion, the encapsulated formulations showed partial recovery of adhesive potential, with C1 digest: 41 ± 0.79, C2 digest: 61 ± 1.04, and C3 digest: 45 ± 0.90. These values were significantly higher than their respective dissolved forms (p < 0.01), highlighting the protective effect of encapsulation and the release of viable adhesive cells upon digestion.

This reduction in adhesion may be attributed to interactions between the probiotic and matrix proteins, particularly whey proteins. According to Zhu et al. (2021), these interactions may lead to flocculation, reducing surface protein exposure and impairing adhesion. Similarly, Hsein et al. (2015) observed that matrix components may modulate covalent interactions between bacterial adhesins and epithelial receptors. Additionally, peptides released during the digestion of whey proteins may exert antimicrobial effects, inhibiting pathogen adhesion while supporting probiotic colonization (Morrissey et al., 2010).

Burgain et al. (2013) demonstrated that protein structure and microcapsule architecture influence specific interactions between Lactobacillus rhamnosus GG and dairy proteins. They proposed that bacteria tend to localize at the fat–protein interface, particularly when displaced during network formation, further affecting adhesion efficiency. The observations in the current study reinforce the hypothesis that encapsulation matrix composition and digestion processes influence bacterial adhesion properties in vitro.

The RKO cell line, representing early-stage colorectal carcinoma, was also employed to broaden the evaluation. This line is recognized for its genetic plasticity and suitability in mechanistic studies involving gene expression and cellular response. As shown in Figure 5, L. fermentum K73 exhibited higher adhesion in RKO cells than HT-29 for equivalent treatments. The 24 h broth recorded 2,806 ± 28.71 adhered cells, followed by bioreactor (952 ± 11.44), and C1, C2, and C3 dissolved (44 ± 1.21, 22 ± 0.53, and 201 ± 4.80, respectively). ANOVA confirmed significant differences between treatments (p < 0.001), and Tukey’s post hoc analysis revealed that all pairwise comparisons between treatments were statistically significant (p < 0.05), except between C1 and C2 dissolved forms.

The greater adhesion observed in RKO cells may be due to differences in surface receptor expression and epithelial maturity compared to HT-29. However, adhesion assessments post-digestion were not feasible in RKO due to enzymatic degradation of the cell monolayer, which led to detachment from the 96-well plates. This limitation is likely related to RKO cells’ immature and fragile differentiation status.

Nonetheless, RKO cells are widely used in adhesion studies due to their altered expression of fibronectin and other adhesive ligands (Tarapatzi et al., 2022). Optimizing enzyme inactivation or protective layering strategies in future studies could preserve epithelial integrity, thereby allowing for comprehensive evaluation of post-digestion adhesion under simulated gastrointestinal conditions.