Targeted delivery of the probiotic *Saccharomyces boulardii* to the extracellular matrix enhances gut residence time and recovery in murine colitis

Engineered probiotics can modulate the gut microbiome, deliver therapeutic payloads, and interact with host immunity, and are clinically relevant but face delivery challenges, notably short gastrointestinal residence time and poor targeting. After oral administration, many probiotics, including Saccharomyces boulardii (S.b.), are rapidly cleared due to colonization resistance and minimal interaction with host tissue, often necessitating daily dosing or non-translatable strategies (e.g., germ-free or antibiotic co-administration) to increase residence time. The authors hypothesized that equipping S.b. with synthetic adhesins to bind extracellular matrix (ECM) proteins overexpressed in inflamed intestinal lesions (e.g., fibronectin, fibrinogen, collagen IV) would increase local retention at disease sites and enhance therapeutic efficacy in inflammatory bowel disease (IBD). They propose a platform where S.b. displays a modular surface handle for attaching ECM-specific biotinylated antibodies, enabling tunable targeting, prolonged gut residence, and improved pharmacodynamics in murine colitis models.

Prior work shows probiotics can support host metabolism, modulate immunity, and deliver therapeutics, and that the first microbiome-based therapy has been FDA approved, underscoring clinical promise and delivery challenges. Probiotic gut residence is limited by rapid GI transit and colonization resistance; S.b. is typically cleared from conventionally raised mice within 24 h unless antibiotics or germ-free conditions are used, which are not translatable. Adhesins in pathogens and commensals establish spatial niches via specific host binding, suggesting that engineered “synthetic adhesins” could improve microbe retention. In IBD lesions, ECM proteins (collagens, laminin, fibronectin, vitronectin, fibrinogen) are overexpressed due to mucosal sloughing and fibrotic remodeling, creating potential anchoring targets. A pooled analysis of seven human ulcerative colitis transcriptomic datasets (n=181 UC, n=82 healthy) shows increased mucosal expression of fibronectin (Fn1), fibrinogen (Fgb), and collagen IV (Col4a1), supporting clinical relevance of these targets.

Engineering: S.b. was genetically modified to stably display monomeric streptavidin (mSA) via yeast surface display. An integrative plasmid encoded constitutive expression of AGA1 and AGA2 fused to mSA (with HA/FLAG tags), integrated into ura3 auxotrophic S.b. using LiAc/PEG transformation. Auxotrophic selection avoided antibiotic resistance genes. The engineered strain is termed S.b. mSA. Biotinylated antibodies against fibronectin (FN), fibrinogen (FB), or collagen IV (CIV) were attached to mSA to generate targeted strains S.b. FN, S.b. FB, and S.b. CIV. Binding assays: Biotin-coated plate binding quantified concentration-dependent attachment of S.b. mSA vs. wild-type S.b. Flow cytometry measured binding of biotinylated antibodies to S.b. mSA and yielded KD values (∼1.11–1.47 nM). ECM-coated plate assays (FN, FB, CIV) quantified targeted vs. non-specific binding at varying seeding densities with fluorescence imaging and ImageJ counting. Temporal stability: Antibody-labeled S.b. mSA was incubated up to 48 h in supplemented simulated intestinal fluid; measurements included yeast growth, percent antibody and mSA remaining on the surface by flow cytometry, and retained ECM-binding capacity over time. Probiotic phenotype assays: (1) Resistance to GI challenges—viability after 4 h culture at pH 2.5, pH 4.0, and OxGall bile salts (0.3%, 0.6%). (2) Short-chain fatty acid (SCFA) secretion—quantification of acetate, propionate, and butyrate by LC-MS at 6, 12, 18 h; secretion rates calculated from linear ranges. (3) Immune modulation—BMDC co-culture (3×10^5 cells/well BMDC; 1×10^6 yeast/well) for 16–18 h with ELISA quantification of IL-10. (4) NFκB pathway-linked cytokine suppression—co-culture of yeast with HT-29 epithelial cells, followed by TNFα stimulation (20 ng/mL) and IL-8 ELISA. Animal models: Female C57BL/6J mice (6–8 weeks) on standard chow were used. Acute DSS colitis: 2% DSS in drinking water for 5 days, then water for recovery; a single oral gavage of yeast (engineered non-targeted S.b. mSA or targeted S.b. FB, S.b. FN, S.b. CIV) was administered on day 5; feces collected at 12, 24, 48, 72 h post-gavage for CFU; colon tissues harvested at endpoint for tissue CFU, cytokine qPCR (Tnfa, Ifng, Il6, Tgfb, Il10), histology (H&E, blinded scoring), colon length, and immunofluorescence (fibronectin and S.b. localization). Chronic DSS colitis: Three cycles of 2% DSS for 5 days followed by 3-day recovery; starting day 9, mice received oral gavages of S.b. mSA, S.b. FN, or S.b. CIV every 3 days until day 24. Fecal CFU measured 48 and 72 h post-dose; endpoint colon tissue CFU, cytokine expression, histology scoring, and colon length. Dosing volumes and CFU: Methods indicate oral doses of 10^6 CFU in 150 µL sterile saline; chronic model also dosed at 10^6 CFU every 3 days. Quantitative culture: Fecal pellets and colon tissues homogenized, plated on antibiotic-supplemented YPD; CFU normalized to sample weight; limits of detection noted. Molecular analyses: Distal colon mRNA extraction (Qiazol, phenol/chloroform), cDNA synthesis, qPCR normalized to ActB, ΔΔCt analysis. Histology: Colon Swiss rolls, FFPE sections, blinded inflammation scoring (mucosal loss, hyperplasia, inflammation, extent). Statistics: Unpaired two-tailed t-tests or one-way ANOVA with Šídák’s, Tukey’s, or Dunnett’s multiple comparisons as appropriate; alpha=0.05. Human datasets: GEO2R analysis of seven UC datasets (GSE13367, GSE9452, GSE38713, GSE47908, GSE73661, GSE114527, GSE87466) for Fn1, Col4a1, Fgb expression.

- Engineering and targeting: S.b. mSA binds biotinylated antibodies with high affinity (KD ~1.11–1.47 nM) and enables specific binding to ECM proteins (fibronectin, fibrinogen, collagen IV). Initial ECM binding increased up to ~350-fold vs. non-targeted S.b., and remained ≥15–20-fold higher after 48 h in simulated intestinal fluid. mSA expression on the surface was stable over time, while antibody signal decreased with cell division. - Probiotic phenotype preserved: Engineered strains (S.b. mSA and antibody-labeled) retained resistance to low pH and bile salts, SCFA secretion comparable to wild type (differences ≤1.5-fold), stimulated IL-10 secretion from BMDCs, and reduced TNFα-induced IL-8 in HT-29 cells similarly to wild-type S.b. - Acute DSS model (single dose): ECM targeting extended gut residence by 24–48 h and increased colon retention. S.b. FN was detectable in feces for ≥72 h post-gavage in all mice and yielded ~100-fold higher viable S.b. in colon tissue vs. other groups at 72 h. Pharmacodynamics improved: colon lengths increased toward healthy values, TNFα expression decreased most with S.b. FN (~3-fold lower vs. DSS), IL-10 expression increased dramatically (~1000-fold vs. DSS), and histology scores were lowest with S.b. FN. Immunofluorescence showed S.b. FN co-localized with fibronectin in mucosa. Antibodies given alone had no therapeutic effect on colon length. No long-term engraftment detected beyond 48 h in extended recovery. - Chronic DSS model (repeated cycles): Repeated dosing every 3 days showed that S.b. CIV achieved the highest fecal CFU at later timepoints, the greatest AUC exposure, and up to ~1000-fold higher colon tissue CFU at endpoint vs. other groups. S.b. CIV improved body weight recovery after first dose, restored colon length comparable to healthy mice, reduced Tnfa, Ifng, Il6, and increased Il10 and Tgfb, and significantly lowered histological inflammation scores. Chronic DSS elevated ECM transcripts with highest Col4a1 expression, consistent with superior retention and efficacy of S.b. CIV in this model.

Targeting S.b. to ECM components abundant in inflamed colonic lesions provides a tunable strategy to enhance microbial pharmacokinetics (local retention and exposure) and pharmacodynamics (anti-inflammatory efficacy). The mSA-based surface display serves as a modular handle to rapidly swap biotinylated ligands, enabling disease- and lesion-specific targeting aligned with dynamic ECM remodeling. In acute colitis, fibronectin-targeted S.b. (S.b. FN) maintained pharmacologically relevant colon levels through the 72-h recovery, resulting in improved colon length, cytokine profiles, and histology. In chronic, relapsing colitis, collagen IV-targeted S.b. (S.b. CIV) exhibited superior retention and efficacy, paralleling higher Col4a1 expression at endpoint, underscoring the importance of temporal ECM changes in determining optimal targets. The engineered S.b. maintained probiotic mechanisms, and targeting reduced necessary dosing frequency from daily to once every three days while improving outcomes. This platform supports personalization by selecting targeting ligands guided by patient-specific ECM expression. Further mechanistic studies are warranted to delineate immune pathways, microbiome interactions, and the spatiotemporal ECM landscape that govern probiotic retention and therapeutic response.

The study introduces a modular ECM-targeting platform for Saccharomyces boulardii that yields a 24–48 h extension in gut residence, ~100-fold higher colon retention in acute colitis, and robust anti-inflammatory benefits in both acute and chronic DSS models. By stably displaying monomeric streptavidin and attaching biotinylated ECM-specific antibodies, S.b. can be redirected to inflamed lesions, enhancing local exposure and efficacy without compromising probiotic phenotype. Target selection can be tuned to disease stage (e.g., fibronectin in acute, collagen IV in chronic). Future directions include: (1) stably expressing ECM-specific ligands to avoid dilution of surface antibodies and further reduce dosing frequency; (2) testing in immune-driven IBD models (e.g., Il10−/−, T cell transfer, FMT-based) to validate translatability; (3) integrating therapeutic payload secretion for lesion-targeted drug delivery; (4) patient-specific targeting based on biopsy-derived ECM profiles; and (5) deeper evaluation of immune mechanisms and microbiome dynamics during treatment.

- Disease models: DSS-induced colitis relies on chemical injury and may not fully recapitulate human IBD pathophysiology; validation in immune-driven models (Il10−/−, T-cell transfer) and FMT-based systems is needed. - Mechanistic understanding: The precise host immune mechanisms by which S.b. exerts anti-inflammatory effects remain incompletely defined. - Microbiome context: Comprehensive effects on microbiome composition and diversity across disease progression and treatment were not assessed. - Target dynamics: ECM expression is dynamic; optimal targeting may vary over time and among lesions, suggesting the need for lesion- and patient-specific strategies. - Persistence: No permanent engraftment was observed; targeting extends but does not eliminate clearance, and surface antibody dilution over time may limit duration of targeting.