Sarah
Blanchet
ab,
Mégane
Bostoën
a,
Veronique
Romé
a,
Isabelle Le
Huërou-Luron
a,
Yves
Le Loir
b,
Sergine
Even‡
*b and
Sophie
Blat‡
*a
aInstitut NuMeCan, INRAE, INSERM, Univ Rennes, F-35000 Rennes, France
bUMR STLO, INRAE, Institut Agro Rennes Angers, F-35000 Rennes, France. E-mail: sergine.even@inrae.fr
First published on 20th June 2025
Human milk (HM) is a complex food that meets nutritional newborn needs. The role of its bioactive components, particularly metabolites, in neonatal development remains poorly understood. This study focused on evaluating the effects of HM short chain fatty acids (SCFA), polyamines, tryptophan derivatives, gamma-aminobutyric acid (GABA), serotonin and lactate on several neonatal gut functions. The effects of these metabolites, at HM concentration, were analyzed individually or in mixture (MTB mix), on an in vitro multicellular model of intestinal epithelium, including Caco-2 (enterocytes), HT29-MTX (goblet cells), NCI-H716 (enteroendocrine cells) and M cells. Transcriptomic semi-screening revealed the impact of these metabolites, especially combined as a mixture, on various intestinal functions. MTB mix upregulated CLDN3 and CLDN4 while downregulating CLDN1 and this was associated with a higher transepithelial resistance, highlighting its potential role in strengthening the intestinal epithelial barrier (IEB). MTB mix also reduced the expression of genes involved in mucus formation (MUC1, TFF3). Besides, MTB mix decreased immune-related gene expression (CXCL8, MYD88, GPX2), suggesting an immunomodulatory effect. Lastly, MTB mix decreased nutrient transporter and enzyme gene expression (SLC2A1, SLC15A1, LCT), suggesting that the mixture modulates digestive function. SCFA, especially butyrate, drove most of these effects, with a contribution from polyamines also, especially on IEB. Individually, GABA had a significant impact on all the examined functions, although these effects were absent with the MTB mix. Overall, this study highlights the ability of HM metabolites to modulate IEB and some genes related to the immune, digestive and endocrine functions in vitro, with some cumulative or attenuated effects when taken altogether vs. individually, emphasizing the importance of studying them as a mixture.
In the colon, SCFA, which include butyrate, acetate and propionate, are mainly derived from the microbial fermentation of non-digestible dietary fibers and absorbed by colonocytes to be used as energy sources.12 Their ability to bind to various receptors such as free fatty acid receptor 2 and 3 (FFAR 2 and 3), as well as their action on histone deacetylase and other transcription factors, such as aryl hydrocarbon receptor (AHR), means that they exert numerous physiological actions in the intestine and they are also major players in the microbiota–gut–brain axis.13,14 They are involved in lipid and carbohydrate metabolisms and help to maintain the intestinal epithelial barrier (IEB) by inducing mucus production and by enhancing the expression of tight junction proteins (TJP).15,16 SCFA also moderate inflammation and promote immune tolerance.17–19 Lactate, an organic acid, which is a product of glycolysis, can be used as an energy source, particularly by the intestinal microbiota, and enables the synthesis of SCFA.20 PA, including spermine, spermidine and putrescine, promote the growth of the gastrointestinal tract21 and enhance IEB by increasing expression of TJP.22,23 They are also involved in the development of the immune system and have antioxidant properties.22–24 TRPd, such as kynurenine and the various metabolites produced by the kynurenine pathway, which is the main tryptophan catabolic pathway (≈90%), have a complex relationship with the immune system and gut microbiota.25 Other TRPd, indole and its derivatives such as ILA, produced by intestinal micro-organisms, enhance IEB by increasing TJP expression in vitro and in vivo and are able to modulate the host immune response.26 GABA, the main inhibitory neurotransmitter in vertebrates, can be used or produced by the microbiota.27 Its receptors are present on enterocytes, enteroendocrine cells and immune cells and it is involved in the regulation of various functions, including intestinal motility, immune and endocrine functions.28 Serotonin, also known as 5-hydroxytryptamine (5-HT), is a neurotransmitter derived from tryptophan. It is involved in numerous gastrointestinal functions, notably motor and sensory functions. Recent studies have established a link between 5-HT, gut microbiota, and immune function.29
The origins of these metabolites within HM are not well understood, but it is likely that some are derived from maternal metabolism, particularly in the mammary gland.30 It is also possible that some of these metabolites originate from the maternal intestinal microbiota via the bloodstream, or are produced by the HM resident microbiota.31 A correlation was previously reported between putrescine concentration and Gammaproteobacteria abundance in the HM.32 HM putrescine and butyrate were suggested to be of microbial origin, while most of the other metabolites targeted in the present study would be of endogenous maternal origin.30,33
While the role of the above-mentioned metabolites has been relatively well documented when they are produced in the gut or supplied by the diet in adults, their role in HM has been little explored. Only butyrate and PA have been studied through supplementation in infant formulas at HM concentrations. Butyrate strengthened the IEB, suppressed inflammatory reactions and could prevent food allergies, as evidenced in a mouse preclinical model and an in vitro human cell line of enterocytes (Caco-2).17,19 PA supplementation during lactation in murine and porcine models promoted intestinal development, immune system and modulated the composition of the intestinal microbiota.22,23,34,35 These interesting effects led us to explore further the role of HM metabolites. We hypothesized that, despite their low concentrations, HM metabolites could modulate several gut functions, either individually or when supplied together, as they are in HM.
In the present study, we addressed the impact of 12 metabolites present in HM, including SCFA, PA, TRPd as well as GABA, 5-HT and lactate, on the barrier, immune, digestive and endocrine gut functions using an in vitro multicellular model of intestinal epithelium. The model included Caco-2 enterocytes, HT29-MTX goblet cells, NCI-H716 enteroendocrine cells and M cells derived from Caco-2 differentiation. This multicellular model was composed of recognized and widely used cell lines of human origin, ensuring easier transferability to humans, and expressed genes present in infant gut such as lactase, ensuring also to a certain extent transferability to infants. This model was suitable for assessing the impact of several combinations of HM metabolites on the main intestinal functions using a semi-screening approach.36,37 Metabolites were used at HM concentrations, either individually, combined by class, or as a mixture of all metabolites. This approach revealed the individual effects of these HM metabolites, as well as their combined effects, for a better assessment of their role in HM.
Caco-2, HT29-MTX and NCI-H716 cells were seeded into inserts with 0.4 μm pore polyester membrane (Corning, Sigma-Aldrich, Saint Quentin Fallavier, France) coated with Matrigel (Corning) at 25 μg cm−2 and placed in a 24-well plate (Corning). A total of 2 × 105 cell per cm2 (i.e. 6.6 × 104 cell per insert) was added in 250 μL of complete DMEM with a ratio of 8 Caco-2 cells for 1 HT29-MTX cell and 1 NCI-H716, in order to reproduce a functionally optimal model in the relative cell proportions of the small intestine,40,41 and 1 mL of complete DMEM was added to the basal compartment. Medium of both compartments were changed every 3 days. After 15 days of co-culture, 1.6 × 105 RAJI cells, corresponding to a RAJI/initial Caco-2 cells ratio of 3:
1, were added into the basal compartment to allow the differentiation of a part of Caco-2 cells into a M cell phenotype. The proportion of Caco-2 differentiated into M cells was evaluated to be ∼7%,39i.e. close to the M cell ratio in human Peyer's patches (<5%).42 Caco-2, HT29-MTX, NCI-H716 and RAJI cells were co-cultured for 6 days with a daily change of half of the DMEM media of the basal compartment after the first 3 days and a complete change of the media of the apical pole.
Class | Metabolites | [HM]a | [HM+]b | No. CAS/references (Sigma-Aldrich) |
---|---|---|---|---|
a [HM]: concentration based on mean and median concentrations of metabolites in human milk (HM). b [HM+]: high concentration in HM, corresponds to 5- to 10-fold the [HM]. c Propionate concentrations are poorly documented in HM and we aligned it with acetate concentration. d The concentrations used correspond to an estimate based on those found in the intestines and feces of newborn, as data from HM were not available when the study began. A recent study measured indole and ILA in HM at concentrations lower that those used in the present study.11 | ||||
Short chain fatty acid (SCFA) | Butyrate | 0.75 mM | 3.75 mM | 107-92-6 /B103500 |
Acetate | 0.25 mM | 2.5 mM | 64-19-7/0070515 | |
Propionatec | 0.25 mM | 2.5 mM | 79-09-4/P1386 | |
Polyamines (PA) | Spermine | 1 μM | 5 μM | 71-44-3/S4264 |
Spermidine | 1 μM | 5 μM | 124-20-9/S0266 | |
Putrescine | 0.5 μM | 2.5 μM | 333-93-7/P5780 | |
Tryptophan derivatives (TRPd) | Indoled | 20 μM | 100 μM | 120-72-9/W259306 |
Indole-3-lactic acid (ILA)d | 20 μM | 100 μM | 832-97-3/I5508 | |
Kynurenine | 200 nM | 500 nM | 2922-83-0/K8625 | |
Non-proteinogenic amino acid | Gamma-aminobutyric acid (GABA) | 120 nM | 600 nM | 56-12-2/A2129 |
Monoamine | Serotonin (5-HT) | 120 nM | 600 nM | 153-98-0/H9523 |
Organic acids | Lactate | 100 μM | 500 μM | 127-09-3/S2889 |
Paracellular permeability was assessed using Lucifer yellow, a small fluorescent marker (452 Da). After 24 h metabolite stimulation, supernatants were discarded and apical and basal compartments were washed with HBSS p/s. Protected from light, 250 μL of HBSS p/s supplemented with Lucifer yellow at a final concentration of 200 μM (Sigma-Aldrich) were added to the apical compartment and 1 mL of HBSS p/s was added to the basal compartment. Culture plates were placed at 37 °C in a 5% CO2 water-saturated atmosphere during 2 h. Basal supernatants were sampled and Lucifer yellow's fluorescence was measured using spectrometer (FLUOstar Omega, BMG LABTECH, Champigny s/Marne, France). Paracellular permeability to Lucifer yellow was assessed by calculating the percentage of basal pole fluorescence over Lucifer yellow's fluorescence at 200 μM, all relative to the control wells not exposed to the metabolites, according to the following calculation: [(LYbasal compartment − blank)/(LY 200 μM − blank) × (100)]/[(LYbasal compartment − blank)/(LY 200 μM − blank) × (100)]CTRL.
The multivariate statistical analyses were performed using R Statistical Software (v4.2.3; R Core Team 2021). Principal component analysis (PCA) was used for separating metabolites based on their profile with FactoMineR package. 33 variables were used: gene expression and TEER measurements.
![]() | ||
Fig. 1 Contrasted impact of human milk metabolites used at human milk high concentrations on the barrier, immune, digestive and endocrine functions in a multicellular model of intestinal epithelium. Principal component analysis was performed on 32 variables corresponding to the expression of genes involved in various intestinal functions (refer to ESI Table S2†) and 1 variable corresponding to the transepithelial electrical resistance (TEER), a marker of para- and trans-epithelial permeability. (A) Scatter plot of individuals on dimension 1 (Dim 1) and 2 (Dim 2), with the confidence circles (corresponding to the confidence interval of the estimate of the barycenter's coordinates) and the barycenters of 7 groups defined according to their metabolic class: SCFA (butyrate, acetate, propionate, and SCFA mix), PA (spermine, spermidine, putrescine and PA mix), TRPd (kynurenine, indole, indole-3-lactic acid and TRPd mix), GABA, serotonin, lactate and the mixture of all the metabolites (MTB). (B) Correlation circle for Dim 1 and 2, with the 21 most involved variables represented. 5-HT: serotonin; GABA: gamma-aminobutyric acid; LT: lactate; SCFA: short chain fatty acid; PA: polyamines; TRPd: tryptophan derivatives. |
Metabolite | Function | Protein name/physiological marker | Gene/Codea | CTRLb | [HM]b | [HM+]b | p-Value |
---|---|---|---|---|---|---|---|
a Gene names are in italics; code refers to physiological marker. b Results presented were obtained by RT-qPCR (mean ± SEM). CTRL refers to control cells not exposed to metabolites and [HM] and [HM+] refer to cells treated with metabolites or groups of metabolites indicated at the mean or high concentrations found in human milk, respectively. SCFA mix includes butyrate, acetate and propionate and PA mix includes spermine, spermidine and putrescine. Differences between groups were assessed by two-way ANOVA. a,b,c different letters indicate that groups differ significantly (p-value < 0.05) and #,## symbols indicate that groups tend to differ (p-value < 0.1). 5-HT: serotonin; GABA: gamma-aminobutyric acid; ILA: indole-3-lactid acid; PA: Polyamines; SCFA: short chain fatty acid. | |||||||
SCFA mix | Permeability | Cadherin 1 | CDH1 | 1.03 ± 0.05a | 1.03 ± 0.06a | 1.22 ± 0.04b | 0.010 |
Claudin 1 | CLDN1 | 0.96 ± 0.06a | 0.85 ± 0.05b | 0.55 ± 0.02c | <0.0001 | ||
Tight junction protein 1 | TJP1 | 1.01 ± 0.07a | 1.03 ± 0.08a | 0.84 ± 0.06b | 0.005 | ||
trans-Epithelial electric resistance | TEER | 1.00 ± 0.05a | 1.17 ± 0.05b | 1.19 ± 0.05b | 0.009 | ||
Mucus | Mucin 1 | MUC1 | 1.06 ± 0.19a | 1.19 ± 0.10a | 0.70 ± 0.08b | 0.014 | |
Mucin 5AC | MUC5AC | 1.03 ± 0.11a | 0.92 ± 0.10ab | 0.56 ± 0.05b | 0.007 | ||
Trefoil factor 3 | TFF3 | 1.01 ± 0.09a | 0.80 ± 0.07b | 0.48 ± 0.03c | <0.0001 | ||
Trophism | Antigen Kiel 67 | Ki-67 | 1.05 ± 0.15a | 0.97 ± 0.10ab | 0.77 ± 0.13b | 0.018 | |
Caspase 8 | CASP8 | 1.01 ± 0.06a | 0.85 ± 0.06ab | 0.69 ± 0.07b | 0.033 | ||
Butyrate | Permeability | Claudin 1 | CLDN1 | 1.02 ± 0.08# | 1.05 ± 0.05# | 0.81 ± 0.07## | 0.041 |
Claudin 3 | CLDN3 | 1.05 ± 0.12a | 1.10 ± 0.22ab | 1.54 ± 0.25b | 0.023 | ||
trans-Epithelial electric resistance | TEER | 1.00 ± 0.03a | 0.99 ± 0.02a | 1.18 ± 0.05b | 0.005 | ||
Mucus | Mucin 1 | MUC1 | 1.08 ± 0.23a | 2.15 ± 0.36ab | 2.12 ± 0.33b | 0.017 | |
Trefoil factor 3 | TFF3 | 1.04 ± 0.12a | 1.01 ± 0.15a | 0.55 ± 0.06b | 0.003 | ||
Trophism | Antigen Kiel 67 | Ki-67 | 1.01 ± 0.05a | 0.90 ± 0.09a | 0.55 ± 0.05b | <0.0001 | |
Acetate | Trophism | Antigen Kiel 67 | Ki-67 | 1.01 ± 0.07a | 0.99 ± 0.05a | 0.78 ± 0.06b | 0.002 |
Propionate | Trophism | Antigen Kiel 67 | Ki-67 | 1.01 ± 0.05a | 0.93 ± 0.03ab | 0.84 ± 0.04b | 0.048 |
PA mix | Permeability | Claudin 1 | CLDN1 | 1.00 ± 0.08a | 0.80 ± 0.06b | 0.79 ± 0.04b | 0.016 |
Tight junction protein 1 | TJP1 | 1.01 ± 0.09# | 0.81 ± 0.06## | 0.83 ± 0.07## | 0.044 | ||
Trophism | Caspase 8 | CASP8 | 1.02 ± 0.07a | 0.73 ± 0.06b | 0.70 ± 0.04b | 0.004 | |
Spermine | Permeability | Claudin 4 | CLDN4 | 1.02 ± 0.08a | 1.60 ± 0.10b | 1.42 ± 0.20b | 0.002 |
Trophism | Caspase 8 | CASP8 | 1.05 ± 0.17a | 0.84 ± 0.06b | 0.83 ± 0.05b | 0.011 | |
Spermidine | Permeability | Claudin 4 | CLDN4 | 1.01 ± 0.06a | 1.15 ± 0.10ab | 1.25 ± 0.10b | 0.045 |
Trophism | Caspase 8 | CASP8 | 1.05 ± 0.15a | 0.86 ± 0.09b | 0.81 ± 0.06b | 0.013 | |
Putrescine | Permeability | Claudin 4 | CLDN4 | 1.02 ± 0.08a | 1.44 ± 0.14b | 1.26 ± 0.07b | 0.002 |
Trophism | Caspase 8 | CASP8 | 1.02 ± 0.10a | 0.91 ± 0.08ab | 0.76 ± 0.07b | 0.050 | |
GABA | Permeability | Claudin 7 | CLDN7 | 1.00 ± 0.04a | 0.87 ± 0.06b | 0.87 ± 0.03b | 0.024 |
Tight junction protein 1 | TJP1 | 1.01 ± 0.07a | 1.19 ± 0.07a | 1.31 ± 0.08b | 0.048 | ||
Trophism | Caspase 3 | CASP3 | 1.02 ± 0.10a | 1.38 ± 0.09b | 1.47 ± 0.09b | 0.002 | |
5-HT | Permeability | Claudin 4 | CLDN4 | 1.02 ± 0.09a | 0.82 ± 0.13ab | 0.76 ± 0.06b | 0.030 |
Kynurenine | Permeability | Claudin 1 | CLDN1 | 1.01 ± 0.05a | 1.10 ± 0.10ab | 1.25 ± 0.09b | 0.027 |
Occludin | OCLN | 1.01 ± 0.04a | 1.06 ± 0.06ab | 1.19 ± 0.06b | 0.008 | ||
ILA | Permeability | Tight junction protein 1 | TJP1 | 1.00 ± 0.06a | 1.13 ± 0.06ab | 1.20 ± 0.06b | 0.030 |
Trophism | Caspase 8 | CASP8 | 1.00 ± 0.03a | 1.09 ± 0.10ab | 1.26 ± 0.07b | 0.038 |
Butyrate and SCFA mix significantly altered expression of genes that mediate tight and adherent junctions. Relative to control cells, expression of claudin 3 (CLDN3) and E-cadherin (CDH1) were greater with butyrate (P < 0.05) and SCFA mix (P < 0.01) respectively, and expression levels of CLDN1 and TJP1 were lower in butyrate (P < 0.05 and <0.1 respectively) and SCFA mix (P < 0.0001 and <0.01 respectively) (Table 2 and ESI Table S3†). Most changes in tight and adherent junction protein expressions were observed at the [HM+] with, sometimes, intermediate or significant effects at [HM] (for CLDN1 and CLDN3 for instance). SCFA also impacted the expression of genes encoding mucins. Butyrate increased the expression of MUC1 (P < 0.05), but the SCFA mix decreased the expression of both MUC1 (P < 0.05) and MUC5AC (P < 0.01). Butyrate and SCFA mix decreased TFF3 expression, as early as [HM] for the SCFA mix (P < 0.01 and <0.0001 respectively). Regarding the trophic factors, SCFA decreased the expression of the Ki-67 proliferation marker both individually (acetate, propionate and butyrate, P < 0.0001, <0.01 and <0.05 respectively) and in combination (SCFA mix, P < 0.05) at [HM+]. The SCFA mix also induced a decrease of caspase 8 (CASP8, P < 0.05) (Table 2).
Some changes in the expression of genes related to the barrier function were also observed with the other metabolites, such as PA that modulated the expression of TJP genes. Indeed, the individual PA spermine, spermidine and putrescine increased expression of claudin 4 (CLDN4, P < 0.01, <0.05 and <0.05 respectively), observable from the [HM]. This was not the case with the PA mix, which conversely decreased the expression of other TJP genes such as CLDN1 and TJP1, from the [HM] (P < 0.05) (Table 2). In addition, PA individually and in combination, decreased the expression of CASP8 (P < 0.01). Unlike with SCFA and PA mix, GABA and ILA increased the expression of TJP1 at [HM+] (P < 0.05). GABA decreased claudin 7 (CLDN7) expression at the two concentrations tested (P < 0.05). Besides, GABA and ILA increased the expression of CASP3 and CASP8 (P < 0.05) respectively, from [HM] for GABA (Table 2). Finally, at [HM+], 5-HT decreased CLDN4 expression (P < 0.05) and kynurenine increased CLDN1 (P < 0.05) and OCLN (P < 0.01) expression (Table 2).
MTB mix recapitulated the effects of the individual metabolites or classes of metabolites, such as the increase in TEER at [HM+] (P < 0.01) observed with SCFA, the decrease in CLDN1 (P < 0.01) observed with SCFA and PA mix, the decrease in MUC1 (P < 0.05) and MUC5AC (P < 0.05) observed with SCFA, and the increase (P < 0.05) in CLDN3 and CLDN4 induced by butyrate and the individual PA respectively (Fig. 3A). MTB mix had no significant impact on proliferation (Ki-67) and apoptosis genes (CASP3 and 8, ESI Table S3†).
Metabolite | Function | Protein name | Gene | CTRLa | [HM]a | [HM+]a | p-Value |
---|---|---|---|---|---|---|---|
a Results presented were obtained by RT-qPCR (mean ± SEM). CTRL refers to control cells not exposed to metabolites and [HM] and [HM+] refer to cells treated with metabolites or groups of metabolites indicated at the mean or high concentrations found in human milk, respectively. SCFA mix includes butyrate, acetate and propionate. Differences between groups were assessed by two-way ANOVA. a,b different letters indicates that groups differ significantly (p-value < 0.05). GABA: gamma-aminobutyric acid; ILA: indole-3-lactid acid; SCFA: short chain fatty acid. | |||||||
SCFA mix | Anti-inflammatory | Transforming growth factor beta 1 | TGFB1 | 1.00 ± 0.05a | 0.98 ± 0.04ab | 0.87 ± 0.04b | 0.050 |
Pro-inflammatory | Chemokine C–X–C motif ligand 8 | CXCL8 | 1.02 ± 0.10a | 1.12 ± 0.08a | 0.70 ± 0.06b | 0.007 | |
Receptor/cellular signaling | Myeloid differentiation primary response gene | MYD88 | 1.02 ± 0.09a | 0.79 ± 0.09b | 0.62 ± 0.04b | 0.003 | |
Anti-oxidant | Glutathione peroxidase 2 | GPX2 | 1.00 ± 0.07a | 1.02 ± 0.06a | 0.56 ± 0.03b | <0.0001 | |
Defense | Intestinal alkaline phosphatase | ALPI | 1.08 ± 0.15a | 1.48 ± 0.16a | 2.94 ± 0.20b | <0.0001 | |
Butyrate | Receptor/cellular signaling | Myeloid differentiation primary response gene | MYD88 | 1.03 ± 0.09a | 0.92 ± 0.08ab | 0.73 ± 0.05b | 0.034 |
Aryl hydrocarbon receptor | AHR | 1.02 ± 0.09a | 1.17 ± 0.09ab | 1.35 ± 0.06b | 0.003 | ||
Defense | Intestinal alkaline phosphatase | ALPI | 1.04 ± 0.12a | 0.82 ± 0.08a | 2.44 ± 0.29b | <0.0001 | |
Acetate | Pro-inflammatory | Chemokine C–X–C motif ligand 8 | CXCL8 | 1.04 ± 0.11a | 0.94 ± 0.19ab | 0.61 ± 0.07b | 0.044 |
Anti-oxidant | Glutathione peroxidase 1 | GPX1 | 1.14 ± 0.23a | 0.69 ± 0.08b | 0.99 ± 0.08ab | 0.013 | |
Propionate | Receptor/cellular signaling | Aryl hydrocarbon receptor | AHR | 1.02 ± 0.09a | 1.09 ± 0.05a | 1.27 ± 0.04b | 0.011 |
Spermine | Anti-oxidant | Glutathione peroxidase 1 | GPX1 | 1.10 ± 0.20a | 0.70 ± 0.09ab | 0.63 ± 0.08b | 0.026 |
GABA | Anti-inflammatory | Transforming growth factor beta 1 | TGFB1 | 1.01 ± 0.05a | 0.88 ± 0.06b | 0.87 ± 0.05b | 0.020 |
Pro-inflammatory | Chemokine C–X–C motif ligand 8 | CXCL8 | 1.02 ± 0.07a | 1.10 ± 0.16ab | 1.32 ± 0.11b | 0.029 | |
Receptor/cellular signaling | Aryl hydrocarbon receptor | AHR | 1.00 ± 0.07a | 1.32 ± 0.05b | 1.31 ± 0.09b | 0.015 | |
Kynurenine | Receptor/cellular signaling | Aryl hydrocarbon receptor | AHR | 1.02 ± 0.07a | 1.21 ± 0.15ab | 1.34 ± 0.12b | 0.004 |
ILA | Pro-inflammatory | Cyclooxygenase-2 | COX-2 | 1.00 ± 0.03a | 1.08 ± 0.03a | 0.86 ± 0.04b | 0.001 |
Receptor/cellular signaling | Aryl hydrocarbon receptor | AHR | 1.00 ± 0.06a | 1.25 ± 0.10b | 1.18 ± 0.07ab | 0.050 |
The other metabolites tested hardly affected the expression of immune markers, especially PA, whose impact was limited to a higher GPX1 expression with spermine. TRPd, such as ILA and kynurenine increased AHR expression at [HM] and [HM+] respectively, and ILA decreased cyclooxygenase-2 (COX-2) expression (Table 3).
Most of the regulations observed previously were also found after exposure to the MTB mix, especially those induced by SCFA mix, including lower expression of CXCL8 (P < 0.001), MYD88 (P < 0.05), GPX2 (P < 0.0001) and higher expression of ALPI (P < 0.05) (Fig. 3B).
Metabolite | Function | Protein name | Gene | CTRLa | [HM]a | [HM+]a | p-Value |
---|---|---|---|---|---|---|---|
a Results presented were obtained by RT-qPCR (mean ± SEM). CTRL refers to control cells not exposed to metabolites and [HM] and [HM+] refer to cells treated with metabolites or groups of metabolites indicated at the mean or high concentrations found in human milk, respectively. SCFA mix includes butyrate, acetate and propionate and PA mix includes spermine, spermidine and putrescine. Differences between groups were assessed by two-way ANOVA. a,b different letters indicate that groups differ significantly (p-value < 0.05). GABA: gamma-aminobutyric acid; PA: polyamines; SCFA: short chain fatty acid. | |||||||
SCFA mix | Nutrient transporter | Glucose transporter type 1 | SLC2A1 | 1.01 ± 0.06a | 1.03 ± 0.07a | 0.81 ± 0.03b | 0.011 |
Monocarboxylate Transporter 1 | SLC16A1 | 1.05 ± 0.18a | 1.42 ± 0.43a | 2.60 ± 0.39b | 0.017 | ||
Peptide transporter | SLC15A1 | 1.01 ± 0.06a | 0.94 ± 0.19a | 0.85 ± 0.08b | 0.005 | ||
Digestive enzyme | Alanyl aminopeptidase | ANPEP | 1.01 ± 0.09a | 1.41 ± 0.08b | 1.34 ± 0.12b | 0.010 | |
Lactase | LCT | 1.03 ± 0.11a | 0.98 ± 0.06a | 0.54 ± 0.03b | <0.0001 | ||
Endocrine | Proglucagon | GCG | 1.15 ± 0.29a | 0.56 ± 0.13b | 0.51 ± 0.10b | 0.019 | |
Butyrate | Nutrient transporter | Monocarboxylate transporter 1 | SLC16A1 | 1.09 ± 0.18a | 1.20 ± 0.15a | 1.66 ± 0.11b | 0.013 |
Digestive enzyme | Alanyl aminopeptidase | ANPEP | 1.10 ± 0.19a | 1.10 ± 0.17a | 1.64 ± 0.18b | 0.003 | |
PA mix | Nutrient transporter | Peptide transporter 1 | SLC15A1 | 1.01 ± 0.08a | 1.05 ± 0.07ab | 1.34 ± 0.10b | 0.029 |
Putrescine | Nutrient transporter | Peptide transporter 1 | SLC15A1 | 1.02 ± 0.13a | 1.10 ± 0.11ab | 1.55 ± 0.18b | 0.043 |
GABA | Nutrient transporter | Glucose transporter type 1 | SLC2A1 | 1.01 ± 0.04a | 0.88 ± 0.07b | 0.89 ± 0.05b | 0.026 |
Peptide transporter 1 | SLC15A1 | 1.00 ± 0.04a | 1.10 ± 0.06ab | 1.28 ± 0.07b | 0.004 | ||
Digestive enzyme | Sucrase isomaltase | SI | 1.04 ± 0.12a | 1.48 ± 0.14ab | 1.69 ± 0.18b | 0.047 | |
5-HT | Nutrient transporter | Glucose transporter type 1 | SLC2A1 | 1.01 ± 0.06a | 0.73 ± 0.05b | 0.84 ± 0.06b | <0.0001 |
Digestive enzyme | Sucrase isomaltase | SI | 0.95 ± 0.11a | 1.42 ± 0.15ab | 1.52 ± 0.14b | 0.025 | |
Lactate | Nutrient transporter | Glucose transporter type 1 | SLC2A1 | 1.01 ± 0.06a | 0.84 ± 0.04b | 0.90 ± 0.06b | 0.002 |
Kynurenine | Digestive enzyme | Alaniyl aminopeptidase | ANPEP | 1.01 ± 0.06a | 1.14 ± 0.08ab | 1.27 ± 0.08b | 0.049 |
MTB mix decreased the expression of SCL2A1 (P < 0.05) similar to the response to SCFA mix, GABA, 5-HT and lactate, and decreased (P < 0.01) the expression of SLC15A1 and LCT as observed with the SCFA mix as well. Surprisingly, MTB mix decreased SI expression at [HM], while it was higher with GABA and 5-HT at [HM+] (Fig. 3C).
Finally, some genes related to the endocrine function were affected by HM metabolites, including a lower expression of the proglucagon encoding gene (GCG) with SCFA mix at both concentrations (P < 0.05) and an increase in the endocrine marker chromogranin A (CHGA) with MTB mix at the [HM+] (P < 0.05) (Table 4 and Fig. 3C).
Our study revealed that several HM metabolites affected the expression of genes associated with barrier, immune, digestive and endocrine functions in a multicellular model of intestinal epithelium. The impacts of the HM metabolites on these gene expressions indicate that HM metabolites could potentially be bioactive in the newborn during breastfeeding, and could therefore modulate and influence the different gut functions, which are under maturation in infancy. In general, the metabolites had an effect at the [HM+], although some statistical effects were also observed at [HM]. Although [HM+] corresponds to metabolite concentrations 5 or 10-fold higher than their mean\median HM concentrations, it remains within a concentration range that can be found in HM (ESI Table S1†) and could therefore be physiologically relevant in newborns. Multivariate PCA analysis, a non-supervised analysis, highlighted contrasting effect of metabolites, depending on their class, with notably a separation between SCFA, PA and GABA/TRPd effects. Interestingly, despite different and possibly opposite effects of the different classes of metabolites, the PCA analysis highlighted that the MTB mix displayed the most contrasting effect compared to all the other metabolites, either individually or in mixtures by classes. Notably, the effects of the MTB mix were close to those of the SCFA mix, suggesting that the effects of the MTB mix were mainly driven by SCFA.
The MTB mix affected the barrier function of the multicellular intestinal epithelium model at [HM+] by increasing TEER as well as by modulating TJP and mucin gene expressions despite sometimes different effects of individual metabolites on TJP gene expression. Expression of CLDN3 and 4 was increased with the MTB mix, as with butyrate and individual PA, respectively. On the contrary, CLDN1 was decreased with the MTB mix, as with the SCFA mix and PA mix, despite increased expression with kynurenine. Thus, the action of SCFA and/or PA prevailed on that of kynurenine over CLDN1 expression. SCFA and PA mix decreased TJP1 expression, whereas GABA and ILA increased it. As a result, the MTB mix had no effect. Of note, TEER increased with the MTB mix at [HM+] only, while it increased from [HM] with SCFA, suggesting that different effects of the different metabolites may occur individually or within the MTB mix. The effects of SCFA on intestinal permeability have been widely documented in the literature, and there is a consensus that they strengthen the IEB both in vitro and in vivo,16,19,46 as found in the present study with measurements of TEER and Lucifer yellow permeability. SCFA have been reported to increase the expression of tight and adherent junction,44,46,47 as we found in our study with CLDN3 and CDH1 respectively. However, exposure to SCFA also decreased the expression of other TJP in our study, such as CLDN1 and TJP1, as previously described in in vitro models.47,48 These opposite effects of SCFA on different tight and adherent junction gene expression underline the complexity and important dynamics of tight and adherent junction genes under the influence of SCFA. Furthermore, a study carried out in Caco-2 cells showed that the action of SCFA did not only involve regulation of TJP expression, but also a more efficient recruitment of TJP to the cell junctions, mediated by the activation of AMP-activated protein kinase.16,49 PA modulated TJP expression without impacting TEER, with differential effects when added individually (increased CLDN4) or in mixture (decreased CLDN1 and TJP1). In vitro, PA depletion studies highlighted their role in the expression of TJP such as OCLN and TJP1.50 In a pig model, PA supplementation increased TJP gene expression.22,23 The effects of PA on intestinal permeability are largely dependent on their concentration, with permeabilizing effects at concentrations above 15 mM.51,52 In our study, concentrations were in the μM range and it is difficult to conclude on the functional effect of PA, as TEER was not affected. However, the increase in CLDN4 expression and the decrease in CLDN1 expression, also present with the MTB mix, highlight their potential effects on the IEB. Overall, despite a combination of differential effects on TJP expression, the MTB mix seemed to strengthen IEB in vitro because the TEER was increased in the conditions we tested. The effect seemed driven by SCFA and PA because the effects observed in the MTB mix were also found in the latter.
Mucins, which are transmembrane or secreted glycoproteins, are also essential components of the IEB as they form a protective mucus layer. MUC1, a gene highly expressed in HT29-MTX cells, decreased with the MTB mix only at [HM], while MUC5AC, a secreted mucin, was higher at [HM] than at [HM+], but did not differ from the control. Mucin genes are finely regulated and extremely sensitive to the environment such as nutritional environment, bacterial products or cytokines.15,53 For instance, these modulations could be due to a transient effect of SCFA. Indeed, butyrate increased MUC1 expression, but SCFA mix reduced both MUC1 and MUC5AC. SCFA are known to modulate mucus production in both in vitro and in vivo studies.15,19,54 A previous study has also shown that SCFA mixtures can decrease MUC5AC expression, though at higher acetate and propionate concentrations than the ones we used.15 Finally, TFF3, which encodes a peptide involved in intestinal repair, decreased with the MTB mix, SCFA mix, and butyrate, consistent with previous studies showing butyrate ability to reduce TFF3 expression.55
Regarding trophicity, the MTB mix did not affect genes related to proliferation or apoptosis. However, SCFA mix, PA mix, GABA and ILA modulated some of these markers. SCFA, especially butyrate, has been shown to have an antiproliferative effect on cancer cells by reducing the Ki-67 marker.56,57 Likewise, most of these metabolites reportedly affect caspase expression.58–62 Thus, in the MTB mix, possible interferences between metabolites could occur allowing fine regulation of trophicity.
In terms of intestinal immune markers, the MTB mix overall decreased their expression, with, notably, a decrease in CXCL8 which encodes the pro-inflammatory cytokine interleukin 8 (IL-8), MYD88 involved in toll like receptor (TLR) signal transduction, GPX2 involved in scavenging oxygen free radicals, suggesting downregulation of inflammation by the MTB mix. These effects occurred with the SCFA mix as well, suggesting SCFA as main drivers of the modulations induced by the MTB mix on immune markers, as already mentioned for IEB markers. Notably, certain metabolites had opposite effects on immune marker expression. For example, SCFA decreased CXCL8 expression, while GABA increased it, but the overall effect with the MTB mix was a reduction. The effects of SCFA on immunity are well documented, especially in the colon, where they are known for their inflammation modulation properties and role in immune tolerance.18,19 Hence, SCFA at concentrations similar or lower than those in the colon (2.5 to 0.625 mM) reduced IL-8 gene expression and production in a Caco-2 model stimulated by tumor necrosis factor alpha.17 In our study, under non-inflammatory conditions and at HM concentrations, SCFA also decreased CXCL8 expression, likely due to acetate. Similarly, MYD88 was down-regulated by SCFA, in line with in vivo studies showing that microbiota-derived SCFA inhibit the TLR4/MyD88/NF-kB pathway, with NF-kB being the nuclear factor-kappa B.63 SCFA influence oxidative stress responses differently depending on the model,64,65 and in the present study, they reduced GPX2 expression. SCFA influence many regulators and pathways, such as NF-kB, extracellular signal-regulated kinases (ERK), p38 mitogen-activated protein kinases (MAPK), c-Jun N-terminal kinases (JNK), and tyrosine-protein kinase (SYK), which explains their impact on immune function.17 Furthermore, as mentioned for the effect on IEB, the effect of the MTB mix is not just a sum of individual metabolite effects. For instance, the increased expression of AHR, a transcription factor involved in the immune response, with butyrate, GABA, kynurenine and ILA, was not observed with the MTB mix. As a central regulator, AHR is tightly regulated, with several ligands including kynurenine, ILA and butyrate,66–68 able to modulate its activity as well as its expression through feedback loop.69,70
Finally, the MTB mix reduced the expression of nutrient transporters and digestive enzymes, suggesting it may affect nutrient absorption. Specifically, the expression of SLC2A1 (encoding GLUT1) and SLC15A1 (encoding PEPT1) decreased. Lower SLC2A1 expression with the MTB mix was also observed with SCFA, GABA, lactate, and 5-HT, possibly due to alternative energy sources being available when glucose is low.15,71 In agreement, a RNA-seq study reported that SCFA treatment altered glucose metabolism, increasing GLUT4 and decreasing GLUT2 and GLUT9,72 albeit at higher concentrations than the ones we used. The expression of SLC15A1 decreased with SCFA mix while it increased with PA mix and putrescine, as already demonstrated in vivo after spermine supplementation.73,74 It was also increased with GABA, although GABA is not transported by the latter.75 So, once again, the effect of SCFA was predominant. Nevertheless, this modulation contrasts with the results of a study showing the activation of SLC15A1 expression by 5 mM butyrate after 24 h of exposure in a Caco-2-BBE model.76 This discrepancy may be due to the different models and the Caco-2 clone used. Finally, expression of LCT, encoding lactase, was decreased with the MTB mix, as it was with SCFA. In a Caco-2 study, LCT expression decreased after 11 days of differentiation, making it an early marker of maturation.77 This decrease might be related to increased ALPI, a maturation marker that also increases after exposure to SCFA. SCFA have been shown to positively affect the differentiation and maturation of intestinal epithelium models,78 which is consistent with our results.
The MTB mix up-regulated expression of CHGA, a marker of endocrine cells, suggesting an effect of the mixture on the NCI-H716 enteroendocrine cells. However, the modulation observed with SCFA, i.e. the reduction of GCG expression, was not observed with the MTB mix. This decrease, although not in line with the consensus that SCFA induce GCG expression, confirms some studies that have highlighted the particular behavior of NCI-H716 after exposure to butyrate.79,80
One of the main objectives of the study was to decipher the role of several HM metabolites on expression of genes related to various functions of the intestinal epithelium despite their low or very low concentrations in HM, in any case lower than in the gut. Here, we have clearly shown that HM metabolites, sometimes alone, but more often in mixture, have an impact on the expression of genes related to all the gut functions we tested in vitro. Within the MTB mix, the impact of SCFA appeared predominant, as most genes affected by the MTB mix were similarly affected by SCFA. Within the SCFA, it was mainly butyrate that drove the modulations, maybe as it was used at a concentration 3-fold higher than acetate and propionate, as found in HM. Nevertheless, the use of mixtures, by classes or with all the metabolites clearly highlighted some cumulative effects of individual metabolites, resulting in amplified or, on the contrary, attenuated effects. This was clearly demonstrated with GABA, which individually had a significant effect on the expression of genes related to all the functions screened, but whose effects in a mixture seemed to be barely visible. In line with this, we could speculate that, depending on the relative concentrations of the different metabolites in HM, different effects would be observed on the intestinal epithelium. Our study was designed based on mean and median concentrations found in HM, but large variability in the concentration of HM metabolites between individuals and studies has indeed been reported.7–11,45 The question then arises as to the impact of these different metabolite profiles on the expression of genes related to the different intestinal functions, with possible consequences for infant health. When considering the possible consequences of different profiles of HM metabolites on neonatal health, it is important to note that although our multicellular model attempts to introduce the cellular diversity present in the intestinal epithelium, it has some limitations. In fact, it is an adult model, using cancer cell lines, but these cell lines do express genes present in infant gut such as that encoding for the enzyme lactase.77 Another limitation of the present study is that we evaluated the effect of HM metabolites on several gut functions mainly through gene expression, which allowed an overview of the HM metabolite effects, and revealed a pleiotropic effect of HM metabolites, but this should be further corroborated by physiological data. While some of them were included in the present study such as the TEER and paracellular permeability with Lucifer Yellow, additional physiological readouts could be monitored, including the production of cytokines. Besides, interferences could occur between the HM metabolites and the intestinal microbiota metabolites, which were not considered in the present study. However, in the proximal parts of the infant intestine, the microbiota is less abundant and so are its metabolites. It is therefore possible that the metabolites of interest, despite their low concentration in the HM, act and/or are absorbed in the small intestine. We can therefore expect direct effects on the proximal part of the intestine, as investigated in vitro in this study, but we cannot exclude more indirect effects on the distal parts of the intestine and on the intestinal microbiota. For this reason, it would be important to explore the HM metabolite effects in in vivo preclinical models, in order to study them in more physiological models considering the influence of the microbiota, and the dynamic of growth and development during the early postnatal/lactation period, such as in some studies carried out on butyrate and PA.19,34 Further studies are also needed to understand the potential matrix effects of HM on intestinal epithelium and host responses. Indeed, we studied the metabolites in a mix as in HM, but they were added to the cellular model in a cell culture medium, which does not reflect the complexity of the HM composition and does not consider possible interference with the matrix.
We thank Charles Le Bras and Nathalie Daniel for sharing their expertise in cell culture and for sharing the protocol on epithelial permeability measurement using Lucifer yellow. We are most grateful to Biogenouest Genomics, ANAEE-Fr and the EcogenO core facility of Rennes (OSUR) for its technical support.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5fo01144b |
‡ Sergine Even and Sophie Blat contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2025 |