Page 31 - Read Online
P. 31

Puhlmann et al. Microbiome Res Rep 2024;3:28  https://dx.doi.org/10.20517/mrr.2024.04  Page 5 of 20

               given in the Supplementary Materials.


               Human gut barrier function model - Ussing chamber
               To explore the possible effects of fermentation products on gut barrier function, we performed an ex vivo
               Ussing chamber experiment using human colonic biopsies according to previously described methods .
                                                                                                       [40]
               Biopsies from four healthy donors (mean age 42 years) were obtained by endoscopy without prior bowel
               cleansing and exposed to fermentation supernatants from in vitro fermentation of dried chicory root cubes.
               The same donor’s feces were used to obtain fermentation supernatants for the respective colonic biopsy to
               study the individual interaction between the donor’s gut epithelium and their gut microbiota-induced fiber
               fermentation. Changes in transepithelial resistance (TER) (measure of overall gut integrity ) and
                                                                                                   [41]
               paracellular permeability (assessed by Fluorescein isothiocyanate-dextran concentration; assessing passage
               between cells) in control biopsies over time were compared at 60 and 90 min to biopsies previously exposed
               (20 min) to the fermentation supernatant and biopsies stressed with sodium deoxycholate (SDC) as well as
               biopsies both exposed to the fermentation supernatant and stressed with SDC. Details on the human
               donors, the fecal in vitro batch fermentation model, and analysis of the gut microbiota composition in
               fermentation pellets, pH and gut microbial metabolites in fermentation supernatants, and gas production
               are given in the Supplementary Materials.

               Statistical analysis
               Data were analyzed using R version 4.2.3 . Normality was checked by inspecting QQ plots. Descriptive
                                                   [42]
                                                         [43]
               statistics were calculated using the rstatix package  and data were expressed as mean with standard error of
               the mean. While the use of statistical hypothesis testing for small sample sizes, as commonly used in in vitro
               triplicate experiments, is debatable, statistical inference was made for the purpose of understandability.
               Differences in metabolites (SCFA, BCFA, lactate, and ammonium) and other outcomes (pH, gas
               production, α-diversity) between products at each time point were tested using robust ANOVA with
                                                                                        [44]
               corresponding post hoc and Benjamini-Hochberg correction from the WRS2 package  and implemented
                                     [45]
               in the ggstatsplot package . Robust ANOVA was chosen as it can handle violations against normality and
               homoscedasticity and sample sizes were too small (triplicates) for reliable non-parametric testing
               implementing Chi-square distributions. Differences in gut integrity and permeability at each time point
               between conditions were tested using ggstatsplot within-subject ANOVA, taking into account the biopsies’
               paired nature. Graphs were made using ggplot2  or Microsoft Office 365 Excel. Gut microbiota outcomes
                                                        [46]
                                                                                  [48]
                                                [27]
                                                               [47]
               were analyzed as described previously  using the mare  and vegan package . Multivariate community
               analysis was done on the genus level by using Principal Coordinate Analysis (PCoA) and constructing a
               principal response curve based on Bray-Curtis dissimilarity (β-diversity), as well as calculating gut bacterial
               richness (α-diversity) based on the number of detected taxa. For univariate analysis at the genus level, taxa
               counts were converted into relative abundances (%) and differential abundance testing with false discovery
                                                                  [47]
               rate (fdr) correction was performed as implemented in mare .

               RESULTS
               We employed a series of complementary in vitro and ex vivo gastrointestinal models primed with dried
               chicory root cube-like particles or milled into powder as to determine the intactness of the plant cell walls,
               its impact on gut microbial fermentation kinetics and composition, and finally the effect on gut barrier
               integrity in human colonic biopsies.

               Intactness of the plant cell wall in dried chicory root as assessed in an upper gastrointestinal in
               vitro digestion model
               First, we set out to assess whether the plant cell matrix was still intact after drying the chicory root particles.
               Using SEM, we indeed observed that the overall plant cell structure was intact in both freshly cut chicory
   26   27   28   29   30   31   32   33   34   35   36