Bioengineered intestinal tubules as a tool to test intestinal biological efficacy of lettuce species

The intestine is a primary site where food exerts biological effects, yet traditional safety and efficacy testing often relies on animal models that are costly, slow, ethically debated, and low-throughput. Microphysiological systems, such as organ-on-chip models, more closely recapitulate the organ microenvironment than conventional in vitro systems. The authors previously developed a bioengineered intestinal tubule containing enterocytes, goblet, Paneth, enteroendocrine, and stem cells with 3D tube-like architecture and villi, suitable for transepithelial transport and validated for nutrition safety and efficacy assessments using markers such as ZO-1 for barrier integrity, alkaline phosphatase for brush border function, and immune markers (NO, IL-6, IL-8). Lettuce (Lactuca sativa) is a major leafy vegetable; breeding emphasizes yield and disease resistance rather than consumer health impacts. Wild relatives (L. serriola, L. saligna, L. virosa) form gene pools with varying genetic relatedness to L. sativa and harbor distinct metabolite profiles, including bitter sesquiterpene lactones. The study asks whether the bioengineered intestinal tubule can discriminate intestinal biological efficacy and potential toxicity among extracts from cultivated lettuce and its wild relatives, and whether these effects align with domestication level.

The introduction and discussion synthesize prior work on organ-on-chip advantages over traditional in vitro and animal models, and prior validation of the bioengineered intestinal tubule for nutrient safety/efficacy testing. Lettuce breeding has historically targeted agronomic traits, with wild Lactuca species providing disease resistance and stress tolerance. Metabolomic studies identified extensive chemical diversity across the Lactuca gene pool, with wild species enriched in bitterness-associated sesquiterpene lactones such as jacquinelin and lactucopicrin. Domestication in crops, including lettuce, often reduces anti-nutritional and bitter compounds and alters metabolite profiles (e.g., quinate, chlorogenic acid, SLs). Potential protective metabolites in cultivated lettuce include fumarate and myo-inositol derivatives, which have anti-inflammatory and barrier-protective properties, whereas certain sesquiterpene lactones can exert antimicrobial or pro-apoptotic effects in intestinal cells. These prior findings frame expectations that domestication status may correlate with intestinal effects of lettuce extracts.

Design: Developed and used a bioengineered Caco-2-based intestinal tubule to test crude plant extracts for intestinal biological efficacy. Endpoints: epithelial barrier integrity (inulin-FITC permeability; ZO-1 immunostaining quantification), cell viability (mitochondrial activity via PrestoBlue), cell attachment (nuclei count per area via DAPI), brush border enzyme activity (alkaline phosphatase, pNPP assay), and immune markers (IL-6, IL-8 ELISAs; nitric oxide via Griess). A clustering analysis integrated endpoints across samples. Model construction: A hollow fiber membrane (HFM) was mounted in a custom 3D-printed PLA chamber with perfusion ports. After sterilization, HFM was L-DOPA treated and coated with human collagen IV. Caco-2 cells were seeded at 1.0×10^5 cells/fiber and cultured 21 days; last 7 days under gentle flow (0.006 dyne/cm²) on a rocker to promote differentiation and 3D morphology. Plant materials: Validation set from supermarket: butterhead, lollo rosso (red leaf), red iceberg (crisphead), and stalk lettuce. Laboratory-grown set: L. sativa cv. Salinas (CGN25281, GP1), L. sativa cv. Olof (CGN05786, GP1), L. serriola US96UC23 (GP1; CGN25282), L. saligna CGN05271 (GP2), L. virosa CGN04683 (GP3). Plants grown 2.5–3 weeks under 16 h/8 h light, ~200 µmol s−1 m−2 LED white light, 21/19 °C day/night, 70% RH. Leaves (or stalk lettuce stems) were snap-frozen, ground, stored at −80 °C. Exposure: Unfiltered crude extracts prepared at 0.5 g plant material per mL culture medium. Tubules exposed for 24 h. Supernatants collected for immune marker assays; tubules washed and processed for permeability, staining, viability, and alkaline phosphatase assays. Assays: Inulin-FITC perfusion (0.1 mg/mL, 0.1 mL/min, 10 min); fluorescence normalized to unseeded HFM as 100% permeability. Immunostaining for ZO-1 and MUC2; DAPI for nuclei. Image analysis in Fiji/ImageJ using maximum intensity projections, line-intersection quantification for ZO-1, and nuclei counting per area. PrestoBlue viability measured fluorometrically, normalized to tubule length and to medium control. Alkaline phosphatase activity via pNPP colorimetry at 405 nm, normalized per mm tubule. IL-6 and IL-8 quantified by ELISA; cross-reactivity assessed in plant extracts alone. NO measured by Griess at 490 nm, also in plant extracts alone. Statistics and clustering: Outliers identified by ROUT (Q=1%). Significance tested by t-tests and one-way ANOVA with P<0.05. For comprehensive analysis, per-run means per line and parameter were min-max scaled (0–1) and clustered by K-means (k=1–11); optimal k evaluated by SSE knee and silhouette coefficient. Hierarchical clustering visualization also performed. Replicates: validation set typically n=7 independent experiments; lab-grown set n=6.

- Commercial lettuce extracts (butterhead, lollo rosso, red iceberg, stalk lettuce) did not disrupt epithelial barrier integrity (inulin-FITC leakage, ZO-1), did not reduce cell viability or attachment, and did not alter alkaline phosphatase activity after 24 h exposure at 0.5 g/mL, compared to medium controls (n≈7; ANOVA/t-tests, non-significant for primary endpoints). - IL-6, IL-8, and NO were detected in starting plant extracts and in supernatants after exposure; increases in tubule supernatants were comparable to baseline levels in extracts, indicating background interference. Slight IL-6 increase with lollo rosso and decreased NO with some supermarket lettuces were observed but confounded by plant-derived signals. - Lab-grown L. sativa (cv. Salinas, cv. Olof) and the close wild relative L. serriola (GP1) did not significantly affect barrier integrity (inulin-FITC; ZO-1 abundance) versus control. - Extracts from wild relatives L. saligna (GP2, CGN05271) and L. virosa (GP3, CGN04683) significantly reduced epithelial barrier integrity (higher inulin-FITC permeability), with ZO-1 density significantly reduced for L. saligna. Both caused complete loss of mitochondrial activity (cell viability marker) and markedly reduced cell attachment and alkaline phosphatase activity (ANOVA/t-tests, p<0.05 to p<0.0001 across endpoints; n≈6). - Some reduction in cell attachment and alkaline phosphatase activity was noted for L. sativa cv. Salinas and L. serriola, but to a lesser extent than GP2/GP3 lines. - Immune markers (IL-6, IL-8, NO) in lab-grown set again mirrored baseline levels present in plant extracts, limiting interpretability. - Clustering of scaled endpoint means separated samples into three groups consistent with domestication level: (1) medium controls and commercial lettuce (minimal effects), (2) lab-grown GP1 lines (L. sativa cvs and L. serriola; mild or no adverse effects), and (3) GP2/GP3 wild species L. saligna and L. virosa (pronounced barrier disruption and cytotoxicity). Optimal k indicated at 3–4 by SSE knee; silhouette supported k=3.

The study demonstrates that a bioengineered intestinal tubule can distinguish intestinal biological effects of lettuce extracts across domestication gradients. Commercial and domesticated or closely related GP1 lines preserved barrier integrity, viability, and brush border enzyme activity, whereas extracts from more distant wild relatives L. saligna and L. virosa induced significant barrier disruption, loss of viability, reduced cell attachment, and decreased alkaline phosphatase activity. These results address the core question by showing that organ-on-chip intestinal models can functionally assess potential toxicity and efficacy of plant germplasm. The clustering analyses reinforce that intestinal effects align with domestication status. Potential mechanistic contributors include metabolite differences: cultivated lettuce tends to have higher levels of fumarate and myo-inositol derivatives with anti-inflammatory and barrier-supportive properties, while wild species are enriched in sesquiterpene lactones that can exert antimicrobial or pro-apoptotic effects in intestinal cells. However, the immune readouts (IL-6, IL-8, NO) were confounded by plant-derived background and cross-reactivity, limiting conclusions about immunomodulation. The model’s multicellular nature and 3D features may be advantageous, but the necessity of full cellular heterogeneity for these assessments remains to be determined.

Bioengineered intestinal tubules provide a practical organ-on-chip platform to evaluate the intestinal biological efficacy and potential toxicity of lettuce germplasm. Extracts from domesticated L. sativa and the close relative L. serriola preserved epithelial function and viability, whereas extracts from more distant wild species L. saligna and L. virosa compromised barrier integrity, cell viability, and brush border enzyme activity. Integrated clustering analyses partitioned samples by domestication level, supporting the model’s discriminative power. The approach could be applied to screen breeding materials, especially those incorporating wild relatives, for safety and nutritional properties. Future directions include linking metabolomic profiles to intestinal effects, clarifying immunomodulatory outcomes via gene expression and protein analyses that avoid cross-reactivity, testing broader germplasm and growth conditions, optimizing extract preparation and dosing to reflect dietary exposures, and evaluating the contribution of specific intestinal cell types in the model.

- Immune marker assays (IL-6, IL-8 ELISAs; NO via Griess) showed substantial background in plant extracts, likely due to cross-reactivity and endogenous plant NO, limiting their utility as readouts in this context. - Extracts were crude and unfiltered at a relatively high concentration (0.5 g/mL), which may not reflect typical dietary exposure and could include non-physiological components affecting readouts. - Commercial lettuce and lab-grown samples differed in age and growth conditions, which can alter phytochemical profiles; although lab-grown lines were cultivated side-by-side, broader environmental effects cannot be fully excluded for supermarket materials. - The necessity and specific role of the model’s heterogeneous epithelial cell composition in detecting effects were not dissected. - As an in vitro Caco-2-based model, translation to human in vivo responses may be limited without validation. - Quantitative mechanistic links to specific metabolites were hypothesized but not directly measured in this study.