Prospecting and characterization of potential probiotic lactobacilli from animal gut and food sources

Abstract

Indigenous Lactobacillus species from African environments remain insufficiently characterised, despite their potential to serve as locally adapted, cost-effective probiotic candidates. This study explored plant- and poultry-derived lactobacilli to identify strains with functional traits relevant to gastrointestinal survival, host interaction, and food-related applications. Eighteen isolates recovered from chicken gut, tomatoes, cucumbers, and bananas were identified by 16S rRNA sequencing and assessed in vitro for acid and bile tolerance, bile-salt hydrolase (BSH) activity, exopolysaccharide (EPS) production, cholesterol assimilation, antioxidant capacity, antimicrobial effects, auto-aggregation, hydrophobicity, adhesion to Caco-2 cells, and safety indicators, including haemolysis, DNase activity, and antibiotic susceptibility. The isolates exhibited pronounced strain-specific variation, with several functional traits occurring in coordinated patterns. Strong acid and bile tolerance frequently aligned with BSH activity, while EPS production correlated with enhanced auto-aggregation. Two tomato-derived Lactiplantibacillus plantarum strains (To3a and To3d) consistently demonstrated the most favourable combination of survival, functional, and safety attributes, including high acid/bile tolerance, notable antioxidant and antimicrobial activity, and strong adhesion to Caco-2 cells. All isolates were non-haemolytic and DNase-negative, and most remained susceptible to clinically relevant antibiotics. Locally sourced L. plantarum strains showed multiple complementary probiotic traits, underscoring their potential as regionally adapted candidates for food or microbial intervention applications. However, genomic characterisation and in vivo validation are required to confirm safety, exclude transferable resistance determinants, and substantiate probiotic efficacy.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-47302-2.

Introduction

Lactobacillus is a prominent genus of lactic acid bacteria (LAB), comprising over 290 Gram-positive, microaerophilic, and catalase- and oxidase-negative species, including L. reuteri, L. acidophilus, and L. plantarum^1–3. These bacteria are ubiquitously distributed across the gastrointestinal tract (GIT), plants, and fermented foods, where they contribute to flavour and texture enhancement and are recognised as safe, holding GRAS and QPS status^4–6. As probiotics, lactobacilli confer a range of health benefits, including modulation of gut microbiota, alleviation of gastrointestinal and metabolic disorders, enhancement of immune function, and potential roles in cancer prevention^3,7–11. Consequently, they have attracted considerable research interest due to their broad health-promoting properties¹².

Probiotic strains have been successfully isolated from diverse sources, including plant material, fermented foods, poultry, fish, and various mammals^13–19. Notable examples include L. acidophilus CM1 and L. plantarum strains derived from silage and artisanal products, which exhibit promising functional characteristics^20,21. most commercialized probiotics originate from Europe, North America, and East Asia, while Africa’s microbial resources remain undercharacterized despite rich fermentative traditions and unique plant–microbe ecosystems. This imbalance limits global strain diversity and reduces local adaptability in African food systems, where environmental, dietary, and regulatory contexts differ from high-income regions. Exploring such ecological niches is particularly valuable, as it may reveal indigenous strains with unique techno-functional and health-promoting potential²². Despite these advantages, access to probiotics remains limited in many developing countries²³. In Africa, most commercial probiotic products rely on expensive imported strains, such as L. rhamnosus GG, rather than utilising locally sourced species^7,24,25. Indigenous Lactobacillus strains from African environments remain largely underexplored, despite their potential to serve as cost-effective, locally adapted probiotics. Locally sourced strains, shaped by regional plant produce and poultry gastrointestinal ecosystems, may offer improved functional robustness in indigenous foods and reduce dependence on imported cultures. To address this gap, this study isolated and characterised lactobacilli from animal GITs and plant sources in Johannesburg, South Africa, to identify promising indigenous probiotic candidates for local application.

Materials and methods

Sample collection and isolation

Bananas, tomatoes, cucumbers (n = 5 each), and chicken gut samples (n = 6) were obtained from retail and street vendors in Johannesburg, South Africa. Samples (1 g) were homogenized in 0.1% sterile peptone, serially diluted (10¹–10⁷), and anaerobically cultured on MRS agar supplemented with 1% cycloheximide at 37 °C for 48 h. Eighteen presumptive Lactobacillus isolates were purified and stored at − 80 °C in MRS broth containing 20% glycerol.

Molecular identification and phylogenetics

Genomic DNA was extracted using the ZR Fungal/Bacterial DNA Miniprep™ Kit. The 16S rRNA gene was amplified with universal primers 27F and 1429R (Appendix A.1). Amplicons were sequenced (Inqaba Biotechnology, South Africa), edited using BioEdit, and identified via BLAST. Phylogenetic analyses were conducted with ETE3 using MAFFT alignment and IQ-TREE (TN + G4 model), with branch support assessed by SH-like aLRT with 1000 replicates²⁶.

Preparation of cell suspensions

Isolates were cultured in MRS broth at 37 °C for 18 h, harvested by centrifugation (10,000 rpm, 4 °C, 5 min), washed twice in PBS (pH 7.2), and adjusted to specific optical densities (OD₆₀₀) of 1.0 for adhesion and antioxidant assays or 0.25 for antimicrobial, hydrophobicity, and safety assays (Appendix A.2). Lacticaseibacillus rhamnosus GG served as the reference control.

In vitro gastrointestinal tolerance

Acid and bile tolerance were evaluated as described by Kumar et al.²⁶. Cell suspensions (10⁹ CFU/mL) were inoculated into MRS broth adjusted to pH 2.5 or 3.0 (HCl) or supplemented with 0.3% or 0.5% bile salts (Sigma-Aldrich). Survival rates were determined after 3 h (acid) or 4 h (bile) using anaerobic plate counts and the standard survival formula (Appendix B.1)²⁷. Only isolates exhibiting > 50% survival at pH 3.0 were included in subsequent assays.

Bile salt hydrolase (BSH) activity

BSH activity was assessed qualitatively using the direct plate method²⁸. Isolates (0.1 mL) were inoculated into wells on MRS agar containing 0.37 g/L CaCl₂ and 0.5% sodium taurocholate (TDCA). Activity was indicated by bile acid precipitation zones after 48 h of anaerobic incubation.

Antimicrobial and antioxidant activities

Antimicrobial activity against selected foodborne pathogens (Appendix C.1) was determined via the agar well diffusion method. Cell-free supernatants (CFS) were neutralized to pH 6.5 and treated with catalase to eliminate the effects of organic acids and hydrogen peroxide. Antioxidant potential was evaluated using the DPPH radical scavenging assay following Dehghani et al.²⁷. Adjusted cell suspensions (OD₆₀₀ = 1.0) were incubated with 0.2 mM DPPH solution in the dark for 30 min, and scavenging activity (%) was calculated based on absorbance at 517 nm (Appendix D.1).

Cell surface properties, adhesion, and functional traits

Auto-aggregation and cell surface hydrophobicity (xylene) were assessed over 2–4 h following Kumar et al.²⁶, using standard surface property calculations (Appendix D.2). Auto-aggregation reflects the ability of cells to adhere to one another, influencing intestinal colonization and pathogen exclusion. Hydrophobicity depends on surface proteins, EPS, lipoteichoic acids, and fimbriae. EPS production was assessed by culturing isolates on MRS agar supplemented with 10% sucrose and observing mucoid colony morphology. EPS contributes to auto-aggregation, biofilm formation, and stress resistance, enhancing host colonization and persistence^29–31. Adhesion was evaluated using the Caco-2 human intestinal cell line³². Post-confluent monolayers were incubated with bacterial suspensions (10⁸ CFU/mL) for 2 h. Non-adherent cells were removed via PBS washing; adherent bacteria were detached using Triton X-100 and quantified by plate counts. Cholesterol assimilation was measured in MRS broth supplemented with 0.3% oxgall and 100 µg/mL water-soluble cholesterol. After 24 h incubation at 37 °C, residual cholesterol in the supernatant was quantified using the o-phthalaldehyde method. The percentage of cholesterol removal was calculated relative to uninoculated controls^33,34. Cholesterol-lowering potential is relevant to probiotic metabolic function and host health benefits.

Safety assessment: haemolysis, DNase, and antibiotic susceptibility

Haemolytic activity was screened on blood agar (5% v/v sheep blood), and DNase activity was tested on DNase test agar³⁵. Antibiotic susceptibility was determined using the disc diffusion method for eight clinically relevant antibiotics (Appendix C.2)³⁶. Inhibition zones were measured and interpreted as susceptible, intermediate, or resistant according to EFSA and CLSI guidelines.

Statistical analysis

Experiments used technical replicates from single purified colonies established as stock cultures from independent samples. Data were assessed for normality and variance homogeneity. Parametric or non-parametric tests were applied as appropriate, with pairwise comparisons summarized via Compact Letter Display (α = 0.05). Statistical analyses were performed in R v4.5.2 (see Appendix E).

Results and discussions

Identification of lactobacilli

Eighteen Lactobacillus isolates were identified across diverse sources (Tables 1 and 2): L. johnsonii (38.9%), L. curvatus (22.2%), L. plantarum (22.2%), L. reuteri (11.1%), and L. sakei (5.6%). L. johnsonii dominated chicken gut samples, while L. plantarum and L. curvatus predominated in vegetables and bananas, respectively, reflecting niche-specific adaptation³⁷. Both L. reuteri and L. johnsonii are common poultry GIT residents^38,39, with prevalence influenced by breed, diet, and environment^40–43. Their strong presence underscores their potential to support poultry gut balance⁴⁴. L. curvatus, the second most prevalent species, was isolated from chicken, cucumber, and bananas. Known for bio-preservation and microbiota modulation^45–47, its presence in bananas likely reflects its ability to metabolize plant starches^48,49. Similarly, L. plantarum predominated in vegetable samples, where its metabolic versatility and robust stress-response systems facilitate adaptation to plant matrices^50,51. Through antimicrobial production and biofilm formation, L. plantarum remains a primary candidate for functional food applications^52,53.

Table 1.

Lactobacilli identified from chicken gut, tomatoes, cucumber and banana.

Food source	Frequency of lactobacilli identified	Frequency
Chicken gut	Lactobacillus johnsonii	(n = 7; 38.9%)
	Limosilactobacillus reuteri	(n = 2; 11.1%)
	Latilactobacillus curvatus	(n = 2; 11.1%)
	Latilactobacillus sakei	(n = 1; 5.6%)
Tomato	Lactiplantibacillus plantarum	(n = 3; 16.6%)
Cucumber	Lactiplantibacillus plantarum	(n = 1; 5.6%)
	Latilactobacillus curvatus	(n = 1; 5.6%)
Banana	Latilactobacillus curvatus	(n = 1; 5.6%)

Table 2.

Identification of lactobacilli isolated from chicken gut, tomato, cucumber, and banana using 16S rRNA Gene Sequencing.

Lactobacillus isolate	Source	Molecular identity	Similarity index
Mcya	Chicken gut	Lactobacillus johnsonii	100%
Mcyb	Chicken gut	Lactobacillus johnsonii	100%
Mcyc	Chicken gut	Limosilactobacillus reuteri	100%
Mcxb	Chicken gut	Lactobacillus johnsonii	100%
Mcxc	Chicken gut	Lactobacillus johnsonii	100%
Chcg	Chicken gut	Latilactobacillus curvatus	100%
Chcb	Chicken gut	Latilactobacillus curvatus	100%
Chcx2	Chicken gut	Lactobacillus johnsonii	99.74%
Chcx3a	Chicken gut	Limosilactobacillus reuteri	99.47%
Chx3b	Chicken gut	Lactobacillus johnsonii	100%
Chcx4a	Chicken gut	Lactobacillus johnsonii	100%
Mcg	Chicken gut	Latilactobacillus sakei	100%
To3a	Tomato	Lactiplantibacillus plantarum	99.54%
To3b	Tomato	Lactiplantibacillus plantarum	100%
To3d	Tomato	Lactiplantibacillus plantarum	100%
Cu2f	Cucumber	Latilactobacillus curvatus	99.20%
Cu3e	Cucumber	Lactiplantibacillus plantarum	99.73%

Phylogenetic analysis of lactobacilli isolates

Phylogenetic reconstruction (Fig. 1) revealed four well-defined lineages with strong statistical support (≥ 95% SH-aLRT). The first clade comprised a tightly clustered Lactiplantibacillus plantarum micro-clade (To3a, To3b, To3d, Cu3e) with 100% support, indicating high strain-level similarity⁵⁴. The remaining lineages included a Pediococcus-proximal grouping (97.5%), chicken gut-derived Lactobacillus johnsonii isolates (Mcya, Mcyb, Mcxc) with minimal sequence divergence (99.8%), and a distinct Lactobacillus reuteri sister group (98.3%). Conversely, L. rhamnosus GG and specific L. johnsonii isolates (Chx3b, Chx4a, Chx2) appeared as outliers with longer branch lengths, reflecting greater genetic divergence⁵⁵. These patterns demonstrate clear species-specific clustering with minor intra-species variation across ecological sources, highlighting conserved evolutionary relationships alongside the strain-level diversity and habitat adaptability typical of Lactobacillaceae⁵⁶.

Fig. 1.

Phylogenetic analysis of Lactobacillus strains isolated from chicken gut, tomato, cucumber, and banana. The phylogenetic tree of Lactobacillaceae strains was constructed using 16S rRNA gene sequences. Multiple sequence alignment was performed with MAFFT (v6.861b, default settings), and a maximum-likelihood tree was reconstructed in IQ-TREE (v1.5.5) using the TN + G4 substitution model selected by ModelFinder. Branch support was assessed with SH-like aLRT (1,000 replicates), and branch lengths represent substitutions per site. Reference strains are indicated with GenBank accession numbers, and isolates from this study are shown in bold.

Acid tolerance

Probiotics must withstand approximately three hours of gastric transit at pH 2.5–3.5 to successfully reach the intestines^32,57. In this study, L. plantarum (To3b, To3d), L. reuteri (Mcyc), L. curvatus (Pb2), and L. johnsonii (Chx3b) demonstrated acid tolerance equal to or exceeding that of the reference strain L. rhamnosus GG (Table 3). Survival under acidic conditions is mediated by mechanisms such as F₀F₁-ATPase proton pumps, extracellular polysaccharide (EPS) production, biofilm formation, and metabolic adaptation^58–61. Acid resilience varies depending on strain origin, membrane composition, and cell envelope architecture^62–65, with gut-derived (L. reuteri Mcyc) and tomato-derived (L. plantarum) isolates exhibiting particularly strong pH tolerance. These results are consistent with previous reports showing > 80% survival at pH 3.0 for various L. plantarum, L. reuteri, and L. johnsonii strains^20,65–68. In contrast, low-tolerance strains, such as L. plantarum LN-3-1, fail to survive gastric passage⁶⁹. High acid tolerance is critical for effective intestinal colonization and the realization of health benefits^70,71, with ≥ 50% survival at pH 3.0 serving as a key benchmark for probiotic efficacy^20,67.

Table 3.

Acid tolerance levels of lactobacilli isolates at pH 2.5 and pH 3.0 after 3 h of incubation.

Isolate code	Cell survival rate (%) at pH 2.5	Cell survival rate (%) at pH 3.0
To3a	78.39 ± 0.69 a	97.20 ± 5.79 a
To3b	67.82 ± 6.03 a	80.74 ± 7.81 a
To3d	73.85 ± 0.84 b	84.00 ± 1.57 b
Chx3a	83.83 ± 6.06 a	89.87 ± 1.88 b
Chx3b	64.46 ± 0.92 c	71.17 ± 7.24 a
Cu2f	44.99 ± 1.55 a	53.79 ± 2.19 d
Cu3e	68.59 ± 3.56 e	89.90 ± 3.39 a
Pb2	72.39 ± 6.25 a	89.57 ± 7.47 a
Mcyc	89.94 ± 6.68 a	98.10 ± 1.32 a
Mcxb	53.35 ± 1.75 d	65.65 ± 2.48 c
Mcxc	63.84 ± 6.12 a	75.71 ± 0.75 c
GG	67.65 ± 1.14 c	93.03 ± 5.37 a

Data are presented as mean ± SD of technical triplicates, each derived from a single purified colony from the stock culture of the respective isolate. Differences in survival rate were analyzed using Welch’s ANOVA (p = 4.44 × 10⁻⁸ at pH 2.5; p = 1.42 × 10⁻⁷ at pH 3.0), followed by Games–Howell post hoc tests. Isolates sharing the same Compact Letter Display (CLD) are not significantly different (p > 0.05). Effect sizes: pH 3.0: F(11, 9.26) = 72.56, η² = 0.92 (95% CI [0.92, 0.98]), ω² = 0.88 (95% CI [0.88, 0.97]); pH 2.5: F(11, 9.34) = 91.41, η² = 0.92 (95% CI [0.91, 0.98]), ω² = 0.88 (95% CI [0.87, 0.97]).

Bile tolerance

Bile salts (0.3–0.5%) in the human GIT can disrupt membranes and induce DNA damage, making bile tolerance essential for survival in the small intestine^40,50,72,73. In this study, L. plantarum To3b and To3d and L. johnsonii Mcxb exhibited significantly higher tolerance than the reference strain L. rhamnosus GG at 0.5% bile (p ≤ 0.05), maintaining survival rates above 94% (Table 4). In contrast, survival decreased markedly in Cu2f, Cu3e, Mcxc, and Chx3a, highlighting strain-dependent variability. Bile tolerance is supported by mechanisms such as BSH activity, active efflux, and alterations in membrane fatty acid composition^7,39,57. In L. plantarum, proteins including GshR1, Bsh1, and OpuA, along with structural components like S-layer proteins and peptidoglycan, enhance stability under bile stress³³. While some vegetable- or chicken-derived isolates fail to tolerate even 0.2–0.3% bile³³, the robust resistance shown by the top-performing isolates in this study underscores their potential for functional activity in the small intestine^12,58,74.

Table 4.

Bile tolerance levels of lactobacilli isolates at 0.3% and 0.5% bile after one hour of incubation.

Isolate code	0.3% bile tolerance (%)	0.5% bile tolerance (%)
To3a	98.72 ± 1.27 a	89.00 ± 0.75 b
To3b	98.74 ± 0.78 a	97.42 ± 1.63 a
To3d	98.22 ± 1.63 a	94.57 ± 4.03 a
Chx3a	89.66 ± 5.99 c	73.64 ± 1.86 de
Chx3b	81.12 ± 0.86 d	62.05 ± 1.46 f
Cu2f	99.08 ± 1.79 a	81.88 ± 1.50 cd
Cu3e	95.37 ± 3.78 ab	83.20 ± 6.33 cd
Pb2	90.87 ± 1.58 bc	86.70 ± 1.30 bc
Mcyc	98.37 ± 0.86 a	85.25 ± 0.67 bc
Mcxb	99.31 ± 0.93 a	95.43 ± 2.23 a
Mcxc	93.38 ± 5.72 bc	69.64 ± 5.95 e
GG	92.92 ± 1.00 bc	88.77 ± 3.11 b

Data are mean ± SD of technical triplicates, each derived from a single purified colony from the stock culture of the respective isolate. Normality and variance homogeneity were confirmed with Shapiro–Wilk and Levene’s tests (p > 0.05). One-way ANOVA showed significant isolate effects: 0.3% bile, F(11, 24) = 11.23, p = 5.65 × 10⁻⁷; 0.5% bile, F(11, 24) = 35.07, p = 4.34 × 10⁻¹². Tukey’s HSD post hoc test was applied; isolates not sharing superscript letters differ significantly at α = 0.05. Effect sizes: 0.3% bile: η² = 0.837 (95% CI [0.828, 0.968]), ω² = 0.758 (95% CI [0.744, 0.952]); 0.5% bile: η² = 0.941 (95% CI [0.939, 0.988]), ω² = 0.912 (95% CI [0.909, 0.982]).

Bile salt hydrolase activity

L. plantarum To3a and Cu3e, along with L. reuteri Chx3a, exhibited the strongest BSH activity, producing inhibition zones of 10–15 mm, significantly exceeding the reference strain, while L. johnsonii Mcxc showed no detectable activity (Table 5). BSH activity reduces bile toxicity and modulates cholesterol metabolism through the hydrolysis of conjugated bile salts^61,70. In L. plantarum, this activity is primarily associated with the bsh1 and bsh3 genes⁵⁹, with cell surface components further supporting bile tolerance^39,57. Strains derived from the gastrointestinal tract (GIT), naturally exposed to high bile concentrations, often display robust BSH activity³⁴. These findings are consistent with reports of significant BSH activity in various L. plantarum and L. reuteri strains^74–77, though activity is strain-dependent and influenced by gene presence and substrate specificity^13,61,62,78. Strong BSH activity enhances both intestinal colonization and cholesterol-lowering potential²⁸.

Table 5.

BSH activity, EPS production ability, and cholesterol-lowering ability of lactobacilli isolates.

Isolate code	BSH activity1	EPS production2	Cholesterol-lowering ability (%)
To3a	++	+	49.9 ± 1.3 b
To3b	+	+	40.6 ± 1.2 c
To3d	+	+	54.0 ± 0.2 a
Chx3a	++	+	57.6 ± 1.4 a
Chx3b	+	–	38.2 ± 0.2 c
Cu2f	+	+	18.1 ± 1.6 e
Cu3e	++	+	46.5 ± 1.7 b
Pb2	+	–	31.0 ± 1.0 d
Mcyc	+	–	23.8 ± 1.0 e
Mcxb	+	+	30.8 ± 1.6 c
Mcxc	–	–	13.1 ± 1.2 f
GG	+	+	41.8 ± 1.4 c

¹BSH activity based on precipitation zone diameter: no zone (–), ≤ 10 mm (+), 10–15 mm (++), > 15 mm (+++).

²EPS production: (+) positive growth; (–) no EPS production.

Cholesterol-lowering data are mean ± SD of technical duplicates (n = 2) from a single purified colony per isolate. Due to unequal variances and small sample size, Welch’s ANOVA was applied: F(11, 4.70) = 94.69, p = 7.63 × 10⁻⁵, followed by Games–Howell post hoc tests. Superscript letters (CLD) indicate isolates not significantly different (p > 0.05). Effect sizes were extremely large: η² = 0.992 (95% CI [0.993, 0.999]); ω² = 0.983 (95% CI [0.986, 0.998]).

EPS production ability

Most isolates, including L. plantarum (To3a, To3b, To3d, Cu3e), L. reuteri (Chx3a), L. curvatus (Cu2f), and L. johnsonii (Mcxb), produced exopolysaccharides (EPS), whereas others, such as Chx3b, Mcxc, Pb2, and Mcyc, did not (Table 5). EPS production facilitates stress tolerance, gut colonization, and auto-aggregation^28,68 by forming protective extracellular structures^79–81. This ability is strain-dependent, involving specific enzymes and eps gene clusters^80,82–84. Structural variations in EPS influence both its biological and technological functions⁸⁵, with producing strains serving as natural bio-thickeners and stabilizers in food products^71,86. Conversely, non-producing strains may have reduced functional utility^58,60, likely due to the absence of key biosynthetic operons^72,84.

Cholesterol-lowering ability of Lactobacillus isolates

Cholesterol reduction among isolates ranged from 13.1% to 57.6% (Table 5). L. plantarum (To3a, To3d), L. reuteri (Chx3a), and L. plantarum (Cu3e) significantly outperformed L. rhamnosus GG (41.8%; p ≤ 0.05). Notably, To3d and Chx3a exceeded 50% reduction, indicating potential for hypercholesterolemia management⁸⁷. Mechanisms underlying cholesterol removal include BSH-mediated deconjugation⁸⁷, direct membrane assimilation⁵⁹, surface adsorption⁸⁸, short-chain fatty acid (SCFA)-enhanced conversion²⁸, and EPS-mediated binding of bile acids⁸⁹. These results are consistent with previous studies reporting 58–85% cholesterol removal by various lactobacilli^12,88–91. The high activity of To3d and Chx3a highlights their suitability for reducing cardiovascular risk⁹².

DPPH scavenging activity of lactobacilli isolates

Scavenging activity of the isolates ranged from 6.8% to 24.5% (Table 6). L. plantarum To3d (24.5%) significantly exceeded L. rhamnosus GG (23.2%; p ≤ 0.05). Probiotic lactobacilli mitigate oxidative stress, linked to numerous chronic diseases⁹², through enzymatic antioxidants, such as catalase and superoxide dismutase (SOD), as well as non-enzymatic molecules^93,94. Catalase, EPS, and organic acids further enhance radical neutralization^95,96. This antioxidant capacity is highly strain-specific, influenced by metabolic pathways and redox-related genes^71,79,97. Our findings reflect the wide variability reported in the literature, where some strains significantly outperform GG while others show minimal potential^98–100.

Table 6.

DPPH scavenging activity (%) of lactobacilli isolates.

Antioxidant assay	DPPH scavenging activity (%)
To3a	19.7 ± 2.3 a
To3b	17.9 ± 1.1 b
To3d €	24.5 ± 1.2 a
Chx3a	18.3 ± 0.6 b
Chx3b	17.1 ± 1.7 a
Cu2f	6.8 ± 1.6 c
Cu3e	18.1 ± 2.9 a
Pb2	11.1 ± 2.3 b
Mcyc #	20.9 ± 1.6 a
Mcxb	7.9 ± 1.4 c
Mcxc	13.3 ± 1.7 b
GG	23.2 ± 1.4 a
Ascorbic acid (10 mg/mL)	35.6 ± 2.2 d

Data are mean ± SD of technical triplicates (n = 3) from a single purified colony per isolate. Means with different superscript letters differ significantly (p < 0.05). € indicates isolates with significantly higher DPPH scavenging than the reference strain GG (p ≤ 0.05). # indicates isolates not significantly different from the reference strain GG (p ≤ 0.05). Due to unequal variances, Welch’s ANOVA was applied: F(11, 9.33) = 25.03, p = 1.55 × 10⁻⁵, followed by Games–Howell post hoc tests. Effect sizes were very large: η² = 0.931 (95% CI [0.930, 0.976]); ω² = 0.897 (95% CI [0.894, 0.964]). CLD superscript letters indicate isolates not significantly different (p > 0.05).

Antimicrobial activity of Lactobacillus isolates

The Lactobacillus isolates demonstrated varying antimicrobial activity against selected pathogens (Table 7). Lactiplantibacillus plantarum isolates from tomato (To3a, To3b, To3d) showed strong antimicrobial effects comparable to Lacticaseibacillus rhamnosus GG, producing inhibition zones of 10–15 mm against Escherichia coli, Staphylococcus aureus, Bacillus cereus, and Listeria monocytogenes¹⁵. These inhibitory effects are mainly attributable to metabolites such as bacteriocins, hydrogen peroxide (H₂O₂), and organic acids, which disrupt pathogen cell membranes and reduce environmental pH^38,101–103. Additional mechanisms include competitive exclusion, interference with macromolecular synthesis, osmotic destabilization, and physical barrier formation through EPS production or auto-aggregation^104,105. Surface proteins, such as SlpA in Lactobacillus acidophilus, may also inhibit pathogen adhesion and enzyme function¹⁰⁶. Several isolates—including Limosilactobacillus reuteri (Chx3a, Mcyc), Lactobacillus johnsonii (Chx3b, Mcxb, Mcxc), Latilactobacillus curvatus (Cu2f, Cu3e, Pb2), and L. plantarum Cu3e—exhibited lower antimicrobial activity (< 10–15 mm). Reduced inhibition may reflect limited metabolite production, absence of key bacteriocin genes, or environmental factors such as neutral or elevated pH, which reduce organic acid efficacy^107–109.

Table 7.

Antimicrobial activity of CFS of lactobacilli against pathogenic organisms.

Isolate code	E. coli	S. aureus	B. cereus	L. monocytogenes
To3a ∞	++	+++	++	++
To3b ∞	++	+++	++	++
To3d ∞	++	++	++	++
Chx3a	–	–	–	++
Chx3b	–	–	++	–
Cu2f	+	++	++	++
Cu3e	–	–	–	–
Pb2	–	–	++	+
Mcyc	–	–	–	–
Mcxb	–	–	++	++
Mcxc	–	–	–	–
GG	++	++	++	++

Results are from independent experiments (n = 3). Zone of inhibition (mm) key:– = not detected; = < 10 mm; ++ = 10–15 mm; +++ = > 15 mm. ∞ indicates isolates with antimicrobial activity similar to the reference strain GG (inhibition ≥ 10 mm for at least one tested pathogen). Technical triplicates (n = 3) were performed for each isolate, derived from a single purified colony from the stock culture.

Cell auto-aggregation

The Lactobacillus isolates exhibited varying auto-aggregation abilities that increased with incubation time (Table 8). After 2 h, Chx3b (40.0%), Cu2f (37.3%), and Mcyc (33.3%) showed the highest aggregation. By 4 h, Mcyc reached 52.0%, followed by Chx3b (48.7%) and Cu2f (48.0%), with most isolates significantly outperforming the reference strain L. rhamnosus GG (p ≤ 0.05). Auto-aggregation—the ability of cells to adhere to one another—is a key probiotic trait associated with enhanced survival, intestinal colonization, and pathogen exclusion^57,110. This strain-specific property is influenced by surface-associated factors such as sortase-dependent proteins and cell surface hydrophobicity, as well as environmental conditions including pH and temperature^111–113. Overall, isolates with strong auto-aggregation potential may exhibit improved adhesion and persistence within the host GIT¹¹⁴.

Table 8.

Auto-aggregation ability of lactobacilli isolates after 2 h and 4 h incubation.

Isolate code	2 h Auto-aggregation (%)	4 h Auto-aggregation (%)
To3a µ	24.0 ± 4.0 a	33.0 ± 4.6 a
To3b	22.6 ± 6.1 a	29.3 ± 6.1 a
To3d µ	26.7 ± 4.6 a	32.0 ± 4.0 a
Chcx3a µ	22.7 ± 8.3 a	31.3 ± 5.0 a
Chx3b µ	40.0 ± 3.5 a	48.7 ± 3.1 a
Cu2f µ	37.3 ± 4.6 a	48.0 ± 4.0 a
Cu3e	18.7 ± 6.1 a	24.0 ± 6.9 a
Pb2	12.7 ± 3.1 b	18.0 ± 2.0 b
Mcyc µ	33.3 ± 8.3 a	52.0 ± 6.9 a
Mcxb	18.7 ± 2.3 a	21.3 ± 6.1 b
Mcxc	28.0 ± 4.0 a	30.7 ± 2.3 a
GG	20.7 ± 1.2 a	24.7 ± 3.1 b

Data represent mean ± SD of technical triplicates (n = 3), derived from a single purified colony from the stock culture of each isolate. Means with different superscript letters (a–d) differ significantly (p < 0.05). µ indicates isolates with significantly higher auto-aggregation than the reference strain at the corresponding time point(s) and a significant increase from 2 to 4 h (p ≤ 0.05). Auto-aggregation was analyzed using Welch’s ANOVA followed by Games–Howell post hoc tests; isolates sharing a Compact Letter Display (CLD) letter do not differ (p > 0.05). Effect sizes: 2 h: F(11, 8.67) = 4.44, p = 0.018; η² = 0.547 (95% CI [0.454, 0.920]), ω² = 0.333 (95% CI [0.207, 0.880]); 4 h: F(11, 9.38) = 7.26, p = 0.0026; η² = 0.693 (95% CI [0.635, 0.943]), ω² = 0.545 (95% CI [0.468, 0.914]).

Cell surface hydrophobicity

The Lactobacillus isolates showed substantial variation in cell surface hydrophobicity (Table 9). L. curvatus Cu2f exhibited the highest hydrophobicity (80–88%), significantly exceeding the reference strain L. rhamnosus GG (48–60%; p ≤ 0.05). Most other isolates showed intermediate levels (16–60%), while Pb2 and Chx3a demonstrated very low hydrophobicity (0–12%). High hydrophobicity in strains such as Cu2f, Cu3e, and Mcyc is likely due to abundant surface-associated molecules, including S-layer proteins, mucus-binding proteins (e.g., MapA), and hydrophobic EPS, which enhance host cell interactions and biofilm formation^29–31. This strain-specific variation is influenced by surface protein expression, growth phase, and structural components such as lipoteichoic acids and fimbriae^27,97. Conversely, low hydrophobicity in isolates such as To3a–To3d, Chx3a, Pb2, and Mcxb/Mcxc may result from the absence of S-layer proteins or EPS layers that mask underlying adhesins^33,97. Similar low hydrophobicity profiles have been reported in other L. reuteri, L. johnsonii, and L. plantarum strains^33,103,115. Strains with low hydrophobicity may exhibit reduced epithelial adhesion and limited colonization potential, potentially diminishing their probiotic effectiveness⁴⁵.

Table 9.

Cell surface hydrophobicity (%) of lactobacilli isolates after 1 h incubation.

Isolate code	Median cell surface hydrophobicity (%)	Range (%)
To3a	16 a	12–20
To3b	20 a	16–20
To3d	28 a	24–28
Chcx3a	4 a	4–12
Chx3b	44 a	40–60
Cu2f	84 b	80–88
Cu3e	72 a	68–76
Pb2	0 c	0–0
Mcyc	80 a	76–84
Mcxb	16 a	16–20
Mcxc	24 a	20–28
GG	52 a	48–60

Data represent mean ± SD of technical triplicates (n = 3), derived from a single purified colony from the stock culture of each isolate. Cell surface hydrophobicity was analyzed using the Kruskal–Wallis test with Dunn–Holm post hoc comparisons; isolates sharing the same Compact Letter Display (CLD) letter do not differ significantly (α = 0.05). Kruskal–Wallis revealed a significant isolate effect: H(11) = 33.98, p = 3.64 × 10⁻⁴, with a very large effect size (ε² = 0.958, 95% CI [0.956, 0.985]).

Antibiotic susceptibility

The Lactobacillus isolates exhibited diverse antibiotic susceptibility profiles (Table 10). All strains were susceptible to penicillin G, chloramphenicol, erythromycin, and ampicillin, suggesting the absence of resistance genes (e.g., mecA, ermB) and minimal risk of resistance transfer^36,71. Conversely, all isolates showed intrinsic resistance to aminoglycosides (kanamycin, gentamicin, streptomycin) due to low membrane permeability and the lack of cytochrome-mediated drug uptake^51,116. This intrinsic resistance allows Lactobacillus species to persist during therapy, potentially aiding gut microbiota restoration⁵⁰. Six isolates (Limosilactobacillus reuteri Chx3a and Mcyc, Lactobacillus johnsonii Chx3b, Mcxb, and Mcxc, and Latilactobacillus curvatus Pb2) were resistant to tetracycline, likely reflecting its common use in poultry farming. This resistance, potentially mediated by transferable tet(M) or tet(W) genes, poses a risk of horizontal gene transfer^36,117. Additionally, 66.7% of isolates displayed intrinsic vancomycin resistance linked to the D-Ala-D-Lac target^71,118, which may confer survival advantages during antibiotic treatment.

Table 10.

Antibiotic susceptibility profile of lactobacilli isolates using the disk diffusion method.

Antibiotic (μg)	To3a	To3b	To3d	Chx3a	Chx3b	Cu2f	Cu3e	Pb2	Mcyc	Mcxb	Mcxc	GG
Penicillin G (10)	S	S	S	S	S	S	S	S	S	S	S	S
Chloramphenicol (30)	S	S	S	S	S	S	S	S	S	S	S	S
Kanamycin (30)	R	R	R	R	R	R	R	R	R	R	R	R
Gentamicin (10)	R	R	R	R	R	R	R	R	R	R	R	R
Streptomycin (10)	R	R	R	R	R	R	R	R	R	R	R	R
Tetracycline (30)	I	S	S	R	R	S	S	R	R	R	R	S
Erythromycin (15)	S	S	S	S	S	S	S	S	S	S	S	S
Ampicillin (10)	S	S	S	S	S	S	S	S	S	S	S	S
Vancomycin (30)	R	R	R	I	S	R	R	R	R	S	S	R

Results are presented as: S = Susceptible (≥ 21 mm), I = Intermediate (16–20 mm), R = Resistant (≤ 15 mm). Technical triplicates (n = 3) were performed for each isolate, each derived from a single purified colony from the stock culture.

Haemolysis and DNase activity

All tested Lactobacillus isolates and the reference strain L. rhamnosus GG exhibited uniform γ-haemolysis (non-haemolytic) and were negative for DNase activity (Table 11), indicating the absence of haemolytic and DNase enzyme production and reflecting favourable safety profiles regarding virulence-associated markers³⁵. The lack of haemolysis suggests these isolates do not possess functional haemolysin-encoding genes¹⁰¹, consistent with previous reports that Lactobacillus species are generally non-haemolytic^12,14 and satisfy key safety criteria for probiotic use⁷⁷. Similarly, no isolates showed DNase activity, indicating a lack of potential for nucleic acid degradation, a common pathogenic trait^35,119. This aligns with prior studies reporting no DNase activity in lactobacilli^12,67, further supporting their safety for food and probiotic applications.

Table 11.

Blood haemolysis and DNase activity of lactobacilli isolates.

Isolate code	αHaemolysis	βDNase activity
To3a	ϒ	–ve
To3b	ϒ	–ve
To3d	ϒ	–ve
Chx3a	ϒ	–ve
Chx3b	ϒ	–ve
Cu2f	ϒ	–ve
Cu3e	ϒ	–ve
Pb2	ϒ	–ve
Mcya	ϒ	–ve
Mcxb	ϒ	–ve
Mcxc	ϒ	–ve
GG	ϒ	–ve

α = possible outcomes: ϒ = no haemolysis, α = partial haemolysis, β = complete haemolysis.

β = possible outcomes: –ve = no DNase activity, + ve = DNase activity. Technical triplicates (n = 3) were performed for each isolate, each derived from a single purified colony from the stock culture.

Caco-2 cells adhesion capacity

The Lactobacillus isolates displayed variable adhesion to Caco-2 cells (Table 12), which model mature enterocytes³². L. johnsonii Chx3b (15.3%) and L. plantarum To3d (12.9%) showed the highest adhesion, surpassing the reference strain L. rhamnosus GG (11.4%), whereas L. curvatus Pb2 and L. plantarum Cu3e exhibited minimal adhesion (2.6–3.6%). Adhesion is a crucial probiotic trait for gastrointestinal persistence and pathogen exclusion^20,114. The strong binding in Chx3b and To3d likely reflects cell wall features such as fimbriae, pili, and adhesins, including EF-Tu and MapA^{45,106,120,121}. Adhesion generally correlates with auto-aggregation and hydrophobicity^64,71, with EPS and lipoteichoic acids further supporting biofilm formation and persistence^103,109. Weak adhesion in Pb2 and Cu3e may result from masked S-layer proteins, EPS interference, or low hydrophobicity^58,103,120. Overall, these strain-specific differences reflect origin and surface molecule expression^46,122, with high-adhesion strains offering superior colonization and immunomodulatory potential¹⁰⁷.

Table 12.

Adhesion capacity of lactobacilli isolates to Caco-2 epithelial cells.

Isolate code	Adhesion (%)
To3a	9.9 ± 0.1 d
To3b	6.9 ± 0.8 fg
To3d	12.9 ± 1.1 ab
Chcx3a	7.8 ± 0.7 ef
Chx3b	15.3 ± 0.3 a
Cu2f	6.01 ± 0.8 fg
Cu3e	3.6 ± 0.0 h
Pb2	2.6 ± 0.8 h
Mcyc	10.5 ± 0.3 cd
Mcxb	6.0 ± 0.8 fg
Mcxc	9.1 ± 0.9 de
GG	11.4 ± 0.3 bc

Data represent the mean ± SD of technical triplicates (n = 3), each generated from a single purified colony from the stock culture. Adhesion was analyzed using Welch’s ANOVA followed by Games–Howell post hoc tests; isolates sharing the same Compact Letter Display (CLD) are not significantly different (p > 0.05). Welch’s ANOVA showed a significant isolate effect on adhesion to Caco-2 cells: F(11, 8.89) = 1758.00, p = 1.85 × 10⁻¹³. Effect sizes were extremely large: η² = 0.9769 (95% CI [0.9763, 0.9927]) and ω² = 0.9654 (95% CI [0.9649, 0.9890]).

Conclusions

This study evaluated eighteen Lactobacillus isolates, revealing strain-specific differences in stress tolerance, enzymatic activity, and adhesion. Acid and bile tolerance were often associated with BSH activity, while EPS production enhanced auto-aggregation but sometimes reduced surface hydrophobicity. Among the isolates, Lactiplantibacillus plantarum strains To3a and To3d (from tomatoes) emerged as the most promising, demonstrating robust survival under stress, notable antioxidant capacity, antimicrobial activity, and strong adhesion to Caco-2 cells. However, as the study relied solely on laboratory assays with a single biological replicate per isolate, the findings lack in vivo validation and genomic safety confirmation. Future research should include whole-genome sequencing to exclude transferable resistance genes and animal trials to verify host safety and probiotic efficacy before these strains can be considered for practical application.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

LAB

Lactic acid bacteria

OD

Optical density

CFU/mL

Colony-forming units per milliliter

BSH

Bile salt hydrolase

EPS

Exopolysaccharide

DPPH

2,2-Diphenyl-1-picrylhydrazyl

CFS

Cell-free supernatant

GIT

Gastrointestinal tract

MRS

De Man, Rogosa, and Sharpe

PBS

Phosphate-buffered saline

HCl

Hydrochloric acid

Appendix A: molecular and technical specifications

A.1. PCR primers and cycling conditions

Forward (27F): 5′-AGAGTTTGATCCTGGCTCAG-3′; Reverse (1429R): 5′-GGTTACCTTGTTACGACTT-3′.

Thermal Cycling: 95 °C (3 min); 35 cycles of 95 °C (30 s), 55 °C (55 s), 72 °C (60 s); final extension at 72 °C (10 min).

A.2. Equipment and media

Spectrophotometry: Genesys 10S VIS (Thermo Scientific, USA).

Incubation: Anaerobic jars with GasPak™ systems.

Media: MRS Broth/Agar (Sigma-Aldrich/Merck, SA).

Appendix B: mathematical formulas

B.1. Survival rate calculation²⁷

Appendix C: biological and chemical specifications

C.1. Indicator pathogens

Staphylococcus aureus ATCC 25923, Escherichia coli ATCC 25922, Salmonella enterica Typhi ATCC 14028, Listeria monocytogenes ATCC 19115.

C.2. Antibiotics tested

Penicillin G (10 µg), Ampicillin (10 µg), Vancomycin (30 µg), Chloramphenicol (30 µg), Erythromycin (15 µg), Gentamicin (10 µg), Streptomycin (10 µg), Kanamycin (30 µg), Tetracycline (30 µg).

Appendix D: functional calculations

D.1. Radical scavenging activity

D.2. Surface property calculations

Appendix E: Statistical analysis

To minimize biological variability, experiments used technical replicates derived from a single purified colony of each isolate. Representative colonies were selected from independent samples (bananas, tomatoes, cucumbers, and chicken gut) and established as stock cultures. All in vitro assays were performed using multiple replicates from these stocks to ensure reproducibility and technical consistency.

Data were assessed for normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test). Depending on the data distribution and variance:

Parametric tests:

• When assumptions were met, one-way ANOVA followed by Tukey’s post hoc test was applied.

• If residuals were approximately normal but variances were unequal, Welch’s ANOVA with Games–Howell post hoc test was used.

Non-parametric tests:

• For non-normal data, group differences were evaluated using the Kruskal–Wallis test with Dunn’s post hoc test and Holm adjustment.

Pairwise comparisons were summarized using a Compact Letter Display (CLD) at α = 0.05. Analyses were performed using R version 4.5.2, with graphical data expressed as mean ± SD.

Author contributions

CVA and FTT participated in the conceptualization and design of the study; CVA drafted the original manuscript; CVA collected the sample and data; CVA performed the experiments; CVA and FTT analyzed the data; FTT supervised; CVA and FTT wrote, reviewed & edited; FTT read and approved the final manuscript.

Funding

This work is based on research supported by the National Research Foundation (NRF) of South Africa (Grant number: 129095).

Data availability

The datasets generated and/or analysed during the current study are available in the NCBI Sequence Read Archive (SRA) repository, under BioProject accession number PRJNA1370507 (https://www.ncbi.nlm.nih.gov/sra/PRJNA1370507).

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval and consent to participate

This research protocol was approved by the College of Agriculture and Environmental Science (CAES) Health Research Ethics Committee at the University of South Africa.