Rapid biotransformation of STW 5 constituents by human gut microbiome from IBS- and non-IBS donors

ABSTRACT STW 5, a blend of nine medicinal plant extracts, exhibits promising efficacy in treating functional gastrointestinal disorders, notably irritable bowel syndrome (IBS). Nonetheless, its effects on the gastrointestinal microbiome and the role of microbiota on the conversion of its constituents are still largely unexplored. This study employed an experimental ex vivo model to investigate STW 5’s differential effects on fecal microbial communities and metabolite production in samples from individuals with and without IBS. Using 560 fecal microcosms (IBS patients, n = 6; healthy controls, n = 10), we evaluated the influence of pre-digested STW 5 and controls on microbial and metabolite composition at time points 0, 0.5, 4, and 24 h. Our findings demonstrate the potential of this ex vivo platform to analyze herbal medicine turnover within 4 h with minimal microbiome shifts due to abiotic factors. While only minor taxonomic disparities were noted between IBS- and non-IBS samples and upon treatment with STW 5, rapid metabolic turnover of STW 5 components into specific degradation products, such as 18β-glycyrrhetinic acid, davidigenin, herniarin, 3-(3-hydroxyphenyl)propanoic acid, and 3-(2-hydroxy-4-methoxyphenyl)propanoic acid occurred. For davidigenin, 3-(3-hydroxyphenyl)propanoic acid and 18β-glycyrrhetinic acid, anti-inflammatory, cytoprotective, or spasmolytic activities have been previously described. Notably, the microbiome-driven metabolic transformation did not induce a global microbiome shift, and the detected metabolites were minimally linked to specific taxa. Observed biotransformations were independent of IBS diagnosis, suggesting potential benefits for IBS patients from biotransformation products of STW 5. IMPORTANCE STW 5 is an herbal medicinal product with proven clinical efficacy in the treatment of functional gastrointestinal disorders, like functional dyspepsia and irritable bowel syndrome (IBS). The effects of STW 5 on fecal microbial communities and metabolite production effects have been studied in an experimental model with fecal samples from individuals with and without IBS. While only minor taxonomic disparities were noted between IBS- and non-IBS samples and upon treatment with STW 5, rapid metabolic turnover of STW 5 components into specific degradation products with reported anti-inflammatory, cytoprotective, or spasmolytic activities was observed, which may be relevant for the pharmacological activity of STW 5.

The gastrointestinal tract (GIT) is the spot of highest microbial abundance and diversity in the human body.The GIT microbiome fulfills important tasks for the physiology and health homeostasis of its human host (7) and includes mostly bac terial taxa like Bacteroides, Faecalibacterium, Ruminococcaceae sp., but also archaea (e.g., Methanobrevibacter species), viruses and fungi (8).However, microbial commun ity composition is subject to many factors, such as dietary habits, health status, and medication (9,10).This variation also applies to IBS in which some studies have found the GIT microbiome to differ from that of healthy individuals in terms of microbial composition (shown by 16S rRNA gene sequencing), metabolites (assessed by metabo lomics), and metabolic pathways (based on shotgun metagenomics) (11).The degree of alteration is mainly associated with the severity of the disease as highlighted earlier (12).These differences between the gut microbiome profiles of IBS patients and healthy controls were also discussed in a recent systematic review (13).Enterobacteriaceae, Lactobacillaceae, and Bacteroides taxa were increased in IBS patients, whereas Faecalibac terium and Bifidobacterium were reduced compared to controls (13).
Several studies have already shown a positive effect of herbal medicinal products on dysbiotic GIT microbiota observed in different disease states like colorectal cancer (14) and obesity (15).On the one hand, herbal products such as licorice-a compo nent of STW 5-can modulate the composition of the GIT microbiome (16), exerting prebiotic-like effects (17).On the other hand, the GIT microbiota can biotransform herbal medicines to produce new absorbable small molecules that can have pharma cological effects either in the gut or systemically (18,19).Examples of such molecules are urolithins that are formed upon gut microbial metabolism of ellagitannins (20), phenylvalerolactones or phenolic acids deriving from biotransformation of procyani dins or flavonoids (18,21), or deglycosylated metabolites from triterpene glycoside metabolism, such as ginsenoside-derived compound K (22) or glycyrrhizic acid-derived glycyrrhetinic acid (23).The interactions between medicinal plants commonly used for gastrointestinal disorders and the human gastrointestinal microbiome were recently reviewed.Several of these plants have been found to influence microbial community composition, exerting prebiotic-like or antimicrobial effects, or to modulate microbial production of short-chain fatty acids.Also, many of their constituents have been shown to be subject of microbial biotransformation, leading to the production of potentially bioactive metabolites (24).STW 5 is a herbal medicinal product with proven clinical efficacy in the treatment of functional gastrointestinal disorders, like functional dyspepsia and IBS (25,26).The multi-target pharmacological effect of STW 5 has been confirmed in several in vitro and in vivo studies (27).These potential benefits of STW 5 include an anti-inflammatory effect, modulatory properties on ion channels, and a region-specific eukinetic effect on gut motoric activity.Overall, this contributes to an improvement in gastrointestinal symptoms by restoring impaired intestinal permeability and reducing visceral hypersen sitivity (28,29).In a recent study, inflammatory processes and gut microbiome dysbiosis could be largely prevented by STW 5 administration in induced ulcerative colitis mouse models (27).
STW 5 contains a combination of nine herbal extracts.The major constituents of this herbal mixture are triterpene saponins, flavonoid glycosides, cinnamic acid derivatives, and alkaloids (30).Our previous study showed that the majority of STW 5 constituents are not or only partially degraded in an in vitro static digestion model simulating conditions in the oral cavity, stomach, and small intestine.This indicated that the constituents could reach the human colon and interact with the gut microbiota if not absorbed in the small intestine (30); moreover, a recent study has shown that STW 5-II, a sister formulation containing six of the nine herbal extracts present in STW 5, interacts with human fecal microbiota in an in vitro short-term colonic model (31).
In this study, an experimental ex vivo platform was used to (i) determine the modulation and putative differential effects of STW 5 on the fecal microbial community of individuals with and without IBS and (ii) analyze shifts in the complex metabolic profile of STW 5 constituents caused by microbial biotransformation.This possible two-way interaction was investigated by incubating STW 5 with fecal microbial samples of individuals with and without IBS.

Study design
To gain preliminary insights into the bilateral interaction between the herbal prepara tion STW 5 and the human gut microbiome, human fecal suspensions (HFSs) from 10 subjects without IBS (healthy, non-IBS) and 6 subjects with IBS (predominant diarrhea type according to ROME IV criteria) were incubated ex vivo with in vitro pre-digested STW 5 or a control (VEH) sample for up to 24 h.VEH contained all reagents of in vitro pre-digestion, but no STW 5. Pre-digestion was performed to mimic the passage of STW 5 constituents through the upper digestive tract (see Materials and Methods for more details).
In total, 560 samples were collected at time point 0 (c 0Mic , before addition of pre-digested STW 5 or VEH; n = 80), after 30 min (tp 0.5 ; n = 160), 4 h (tp 4 ; n = 160), and 24 h (tp 24 ; n = 160) of incubation (for detailed information on the sample size see Fig. S1).Potential changes in microbial composition due to STW 5 at all time points were assessed by amplicon sequencing of the 16S rRNA gene (V4 region).In addition, the microbial load was quantified by qPCR and in order to assess microbial metabolism of STW 5 constituents, metabolite profile changes in the incubates were detected by untargeted ultra high-performance liquid chromatography-mass spectrometry (UHPLC-HRMS) analysis (Fig. 1).

Quality control of the amplicon data of the HFS samples
An analytical sample size flowchart of the study is provided in Fig. S1.Overall, 560 samples from 16 donors were processed within this study.For quality control, reads obtained from the negative/process controls were subtracted, and samples that did not meet the specified quality criteria (e.g., with respect to number of reads) were excluded.SRS (Scaling with Ranked Subsampling) was used to normalize the data for all further analysis.In this step, additional three samples did not meet the minimal sampling depth of 10,000 and were excluded.The final data set included 552 samples with 4,022 RSVs and 5,520,000 reads for further analyses (for detailed information on the filtering, see Materials and Methods).

Baseline microbial signatures do not differ between fecal samples from non-IBS and IBS donors
To initially assess the original microbial community composition and potential differences between samples from healthy and IBS donors, c 0Mic samples (HFSs without any treatment) were analyzed.
In general, the summarized replicates per healthy subjects did not differ significantly in alpha diversity compared to IBS patients; however, the IBS group seemed to be more even (Fig. 2B; Table S1A).The five technical replicates of each participant formed a distinct cluster, indicating low intra-sample variability and high stability of the applied experimental settings.However, no separate clustering was observed based on health status (IBS, non-IBS) in the PCoA plot (Fig. 2C).
Discriminatory analysis (MaAsLin2_SR) between IBS and non-IBS in c 0Mic samples revealed some taxa with higher prevalence (P value < 0.05) in one or the other health condition.Taxa belonging to the phylum Firmicutes, in particular to the family Lachno spiraceae, were predominantly found in HFS from IBS patients (Table S1D).Two genera, namely Lachnoclostridium (P = 0.02) and Roseburia (P = 0.005), were among the 15 most abundant taxa found in IBS samples (Fig. 2A).Putative biomarkers for non-IBS HFS included some taxa of the families Ruminococcaceae or Rikenellaceae.Nevertheless, none of the mentioned taxa remained significant after correction (q value) (Table S1D).

The experimental setup allows an unbiased 4-h window for ex vivo analysis of effects of STW 5 on fecal samples
Although the relative abundance of the microbial taxa was quite stable during incuba tion, the absolute abundance determined by qPCR showed significant differences over time (Fig. 3A 1 and A 2 ).The copy number of the 16S rRNA gene per gram of stool increased slightly but not significantly after 4 h of incubation, while the microbial load tended to decrease in the vehicle-treated samples.However, after 24 h of incubation, a significant decrease in the microbial load was observed both in STW 5-and vehicle-trea ted samples (Fig. 3A 2 , for P values, see Table S1B and C).Furthermore, the microbial community of t 24 clustered separately from c 0Mic , tp 0.5 , and tp 4 of the replicates of the same participant in the PCoA plot (Fig. 3B 1 and B 2 ).These differences in microbial load and community composition could be explained by the depletion of nutrients or other dynamics occurring in the static batch cultures after an extended period of time.
It shall be noted that the described trends were observable in all groups (non-IBS, IBS, and combined).Therefore, in the following, to be maximally conservative, we focused on further analysis of the microbiome at tp 0.5 and tp 4 to avoid bias in interpretation.

STW 5 does not affect overall alpha and beta diversity, but reveals potential effects on specific microbial taxa
To evaluate the changes in microbial composition induced by in vitro pre-digested STW 5 compared to in vitro pre-digested vehicle (VEH; details are given in Materials and Methods) in samples from IBS patients and healthy controls, the relative microbial abundance in tp 0.5 and tp 4 samples was subjected to further statistical analysis.
In comparison to the previous analyses at baseline c 0Mic (Fig. 2A), the predominant phyla and genera remained the same over time (tp 0.5 and tp 4 ) independently of the used treatment (STW 5 and VEH), indicating that STW 5 did not massively influence microbiome composition (Fig. 4A).Moreover, the different treatments had no influence on any of the microbial alpha diversity indices at tp 0.5 and tp 4 in both groups (Fig. 4B; for P-values see Table S1A).In addition, the overall structure of the microbial community was not significantly altered by STW 5 compared to VEH addition at any time point (Fig. 4C).
In general, taxonomic differences (P < 0.01) were detectable between non-IBS and IBS samples based on the two treatment options.A full overview is provided in Fig. S2A.Ruminococcus 1 as well as Actinobacteria and Anaerotruncus distinguished the VEH-treated HFS-samples of non-IBS and IBS donors at tp 0.5 and tp 4 , respectively.STW 5 addition led to higher amounts of Roseburia in IBS and Anaerotruncus signatures in non-IBS samples at tp 4 .
Ruminococcus 1 and Roseburia were found to be decreased (P < 0.01) in non-IBS and IBS samples, respectively, under the influence of STW 5 (Fig. S2A, for P values, see Table S1D).It shall be noted, that most of these variations did not withstand P value corrections, so they have to be considered with caution.
Since the health status did not significantly (q < 0.05) affect microbial community members at c 0Mic or other time points (tp 0.5 , tp 4 ) (Fig. S2A, for P values, see Table S1D), further analyses of microbial biomarkers for STW 5 and VEH, were performed using the combined data set (non-IBS + IBS samples).
Interestingly, the significant associations (P < 0.01) observed in HFS at phylum level (a higher relative abundance of Proteobacteria and Bacteroidetes in vehicle-treated samples and of Firmicutes in STW 5-treated samples) at tp 0.5 changed to the opposite after 4 h of incubation, indicating a potential benefit for Proteobacteria and Bacteroi detes by STW 5 over the course of time (Fig. S2B, for P values, see Table S1D).At genus level, just one taxon, namely unknown Rhodospirillales (tp 4 ) was found to be associated with STW 5, whereas VEH-biomarkers included more taxa at both time points such as Sutterella or Victivallis at tp 0.5 and Ruminococcaceae UCG-013 or Ruminiclostridium 5 at tp 4 (Fig. S2B, for P values, see Table S1D).This could indicate a potential inhibition of those taxa by STW 5.
Analysis over time revealed many time-dependent taxonomic associations in STW 5 and VEH-treated HFS.However, similar trends were observed for both additive groups at the selected time points (e.g., variations in Agathobacter, Faecalibacterium and Rosebu ria, Akkermansia, Alistipes, and Lachnoclostridium).Nevertheless, specific taxa (such as Methanobrevibacter, Bifidobacterium for VEH or Sellimonas and Anaerostipes for STW 5) were associated with just one additive group (Fig. S2B, for P values, see Table S1D).
Since our results indicated enhanced levels of phylum Bacteroidetes und RSV Bacteroides in STW 5-treated samples compared to VEH-treated samples at tp 4 , we checked potential biomarkers belonging to the phylum Bacteroidetes.Analysis based on RSV level revealed three Bacteroidetes RSVs (P < 0.01) to be associated with VEH at tp 0.5 and three with STW 5 at tp 4 .Additionally, more RSVs were defined as biomarkers in STW 5 at tp 4 compared to tp 0.5 and vice versa in VEH.In general, the significantly different RSVs found in the comparisons belonged to similar taxa including Bacteroides, Parasutterella, or Alistipes (Fig. S2C, for P values, see Table S1D).
For completeness, the separated data sets (non-IBS and IBS) showed similar results at phylum and genus levels compared to the combined one.(Fig. S3A and B, for P values, see Table S1D).Again, the significant associations (q < 0.05) observed in non-IBS HFS at phylum level at tp 0.5 changed to the opposite after 4 h of incubation.However, in IBS HFS, the difference was not as prominent.Just one characteristic taxon for VEH, namely the phylum Proteobacteria (q = 0.004), was found at tp 0.5 .No further significant differences could be observed at tp 0.5 nor tp 4 (Fig. S3A and B for P values, see Table S1D).S1B and   C).(C) The additives did not shift the microbial community as a whole in one or the other direction-neither after (C 1 ) 30 min nor (C 2 ) 4 h of incubation.
Mainly, trends (P < 0.01) were detectable at genus level based on the treatments in both IBS-and non-IBS-donor samples.Several taxa were characteristic of VEH-treated samples, but fewer microbial signatures were associated with STW 5. Except for unknown Rhodospirillales (VEH biomarker at tp 0.5 ), no genus was significantly (q < 0.05) associated with one or the other additive (Fig. S3A and B, for P values, see Table S1D).

Human gut microbiota rapidly biotransform STW 5 constituents into lower molecular-weight metabolites
The microbial turnover of the previously identified major STW 5 constituents (30) and the formation of metabolites were analyzed by UHPLC-HRMS.
These compounds were consistently detectable at baseline, i.e., in the c 0Met samples (Annotation, see Table S2).Upon incubation with HFS, the levels of these compounds were found to consistently decrease over time, while they remained unchanged in the microbiome-free control samples.The levels of the phenolic compounds rosmarinic acid, 2-glucosyloxy-4-methoxycinnamic acid isomer, and liquiritin pentoside decreased very fast, with median ratios compared to c 0Met below 0.5 already after 0.5 h of incubation.Liquiritigenin was initially contained in STW 5, but obviously also intermittently formed by deglycosylation of liquiritin pentoside (see Fig. 5B; for P values, see Table S1E).Therefore, its median ratio compared to c 0Met decreased below 0.5 only after 4 h of incubation.The triterpene glycoside glycyrrhizic acid was degraded slower than the phenolic compounds, reaching a median ratio below 0.5 compared to that present in c 0Met only after 4 h of incubation.
As far as the impact of health status on the turnover of STW 5 constituents is concerned, in general, similar trends were observed for metabolism of STW 5 constitu ents in IBS and non-IBS samples (Fig. 5B 1 ; for P values, see Table S1E).In case of HFS from non-IBS subjects, the constituents glycyrrhizic acid, 2-glucosyloxy-4-methoxy-cinnamic acid isomer, and rosmarinic acid significantly decreased from c 0Met to tp 0.5 , whereas glycyrrhizic acid was significantly decreased only after tp 4 in IBS-HFS (Fig. 5B 1 ; for P values, see Table S1E).Nevertheless, only the turnover ratio (tp 0.5 /c 0Met ) of liquiritin pentoside isomer at tp 0.5 was found to be significantly different between the non-IBS and IBS samples (Fig. 5B 2 ; for P values, see Table S1E).Apart from that, no statistically significant differences of any turnover product could be observed based on health status (Fig. 5B; for P values, see Table S1E).

The composition of the microbiome reflects the turnover of STW 5 com pounds only to a small extent
In this study, the turnover of five major STW 5 constituents over time was observed.This metabolism happened after administration to HFS samples, indicating an active role of the microbes present in stool in the turnover.To identify the microbes that are potential candidates for metabolism, correlation analysis was performed based on our collected metabolome and microbiome data.Conversion of liquiritigenin to davidigenin has been described for Eubacterium ramulus (38).(A 3 ) 2-glucosyloxy-4-methoxycinnamic acid isomer and herniarin.
(Continued on next page) When considering the dominant taxa, just two correlations remained significant after multiple testing correction (q < 0.05).Agathobacter and Christensenellaceae-R7 group were correlated with one STW 5 turnover product-2-(2-hydroxy-4-methoxyphenyl) propionic acid isomer at tp 4 and davidigenin at tp 0.5 , respectively (Fig. 6A).Both taxa could not be identified as taxonomic biomarkers for one or the other additive by discriminative analysis (see above).However, Agathobacter tended to be more abundant in tp 0.5 HFS compared to tp 4 (Fig. S2 and S3).
A possible contribution to the metabolic turnover of the STW 5 constituents was also investigated based on the identified biomarkers for STW 5 HFS (Fig. 6B).The only taxon that was linked to STW 5 administration did not correlate significantly with any of the observed metabolites in the HFS.In the case of VEH biomarkers, just Ruminiclostridium 5 was found to be negatively associated with davidigenin (Fig. 6B).

DISCUSSION
This study investigated the possible interactions between pre-digested STW 5 and human gut microbiota in an experimental ex vivo platform.The main findings can be summarized as follows: 1.The established ex vivo analysis platform showed a high applicability to analyze the turnover of herbal medicines within 4 h without major shifts in the micro biome caused by abiotic factors.The microbial load of HFS remained stable up to 4 h after incubation; however, a drastic decrease was detectable in both treatment groups after 24 h of incubation, most likely due to nutrient deprivation.2. Minor differences were observed between the microbial composition of HFS treated with pre-digested VEH or STW 5. 3. When incubated with HFS, very fast metabolic turnover of STW 5 constituents into their respective metabolites was observed.4. The metabolic turnover was not well mirrored by an extensive, statistically significant shift of the overall microbiome or microbial taxa. 5. Non-IBS and IBS microbiomes, in our study, were not significantly different from each other and did not reveal significant differences in the capability for metabolic transformation of STW 5 constituents (except for liquiritin-pentoside isomers).
In the applied experimental ex vivo platform, viable/intact microbial biomass (PMA-treated) was maintained up to 4 h after inoculation without additional nutrient supply.However, a significant decline in the microbial load was observed after 24 h of incubation both in STW 5-and vehicle-treated samples.This drastic drop in the microbial load may be attributed to the depletion of nutrient levels in the static batch culture, or other unknown factors (32,33).Therefore, incubation for up to 24 h may not be meaningful, and in future experiments, intermediate samples should be taken for better clarity.Due to the significant drop in absolute microbial load at tp 24 , this time point was not considered in further analyses.
The use of a substrate-limited batch culture can be regarded as a limitation of this study, since the colon can be essentially regarded as a continuous culture.
A second limitation is that this study was performed as a pilot investigation with a relatively small number of donor samples.Due to the strictness of the exclusion criteria, it  42) can be induced by the listed bacterial taxa (caffeic acid and dihydrocaffeic acid intermediates have not been detected in this study) (B) Peak areas of STW 5 constituents and their turnover products over time in HFS.(B 1 ) Peak areas compared over time in non-IBS and IBS, respectively.(B 2 ) Ratios (tp 0.5 /c 0Met and tp 4 /c0 Met ) were used to compare metabolite peak areas based on health status.Horizontal lines in the boxes represent median peak areas.Average peak area per metabolite for each participant was calculated to minimize the bias by using replicates (statistical tests: paired samples (over time): Wilcoxon and paired t test; unpaired samples (health status): Mann-Whitney U test or unpaired t test; for P values, see Table S1E).was only possible to recruit six IBS patients during the study period, while for the non-IBS group, 10 donors were recruited.
Moreover, the experimental setup of the simulated upper intestinal tract digestion as well as of the gut microbial fermentation did not allow to consider absorption processes such as the uptake by intestinal transporters which may be relevant to certain STW 5 constituents under in vivo conditions (34,35).Also, potential interactions of STW 5 with microbiota occurring in the small intestine, which may on the one hand be a relevant factor in the pathogenesis of IBS and may on the other hand play a role for the biotransformation of STW 5 constituents (36) remained unconsidered in the applied experimental setup.In vivo studies are necessary to consider these aspects in the future.
While some research reports indicate that the gut microbiome may play a role in irritable bowel syndrome, evidence remains inconclusive.Several studies have shown variations in microbial abundance and composition based on health status, with fecal samples from IBS patients generally displaying lower alpha diversity and spe cific microbial markers compared to those of healthy individuals (11)(12)(13)43).Putative microbial changes appear to be influenced by the subtype and severity of IBS (11,12).In contrast to those studies, other research observed a similar or only modestly altered microbial community of healthy and IBS subjects and concluded that poor self-rated health or reduction of the data complexity by bioinformatic methods or by selecting only the most important parameters had led to observation of differences in microbial richness and taxa between the groups (38,43).Consistent with the latter findings, our study was unable to distinguish the HFS of IBS and non-IBS subjects based on microbiome composition.Just a trend toward lower alpha diversity in IBS stool samples was visible, which was not statistically significant.Since this study did not find any significant differences in the gut microbiome between healthy individuals and those with IBS, it suggests that any potential treatment could be equally effective for both groups.Furthermore, this enabled us to combine the data sets of both groups for further analyses.
Major compound classes present in STW 5 include phenolic compounds such as flavonoid glycosides, dihydroxycinnamic acid derivatives, and triterpene glycosides (30).These compound classes are known to have a modulatory effect on the composition and function of the human gut microbiome: the chemical structures of polyphenols suggest antimicrobial effects which may shape gut microbial community composition, but more importantly, several polyphenol classes have been shown to possess prebioticlike effects in vitro, in preclinical and in clinical studies.These effects may be beneficial in diseases associated with gut microbiome dysbiosis (39).Recent findings suggest that also triterpene glycosides possess prebiotic-like effects (40).Moreover, certain microbial taxa are able to use the sugar moiety of various plant glycosides as energy source, leading on the one hand to the cleavage of these glycosides, and on the other hand to a preferential growth of these taxa (41,42).These data suggest a potential impact of the studied preparation on the gut microbial composition.
However, in contrast to two animal studies suggesting that STW 5 affects gut microbiome composition (27,44), only minor differences in microbial signatures were observed between STW 5-and vehicle-treated HFS in this study.At the phylum and genus levels, a small number of microbial signatures were identified as potential biomarkers for one of the additives.Interestingly, more taxa were associated with the VEH additive (10 genera) compared to STW 5 (one genus) and also the changes over time were similar between STW 5-and VEH-treated samples.This suggests that there may be an inhibitory effect of STW 5 on the decreased taxa, possibly due to antimicrobial effects of the contained polyphenols, but a comparable effect on the microbiome of both health groups.
Compared to that, STW 5-II, a formulation containing six of the nine herbs making up STW 5, has shown more pronounced effects on microbiome composition of healthy donors in recently performed in vitro study in a short-term colon model.This may be due to differences in experimental setup as well as in the applied formulation (31).
Consistently with their similar microbial profiles, IBS and non-IBS samples were also functionally similar in terms of biotransformation of STW 5 constituents, albeit with slightly but in most cases not significantly different biotransformation kinetics.In both groups, levels of the genuine constituents of pre-digested STW 5 were significantly decreased over incubation time in all donor samples, and this decrease paralleled an increase in the levels of their respective metabolites.Similar metabolic reactions have also been observed when STW 5-II, containing six of the nine herbs contained in STW 5, was incubated with microbiota from healthy human subjects in a short-term colonic in vitro model (31).
When STW 5 was incubated with the incubation medium in the absence of fecal samples, its constituents remained stable.This suggests that the fecal microbiota were responsible for the observed biotransformations.
The O-deglycosylation of flavonoids, such as the conversion of liquiritin pentosides to liquiritin observed herein (Fig. 5A 2 ), is generally accomplished by a wide range of microbial taxa, such as Bacteroides sp., Parabacteroides sp., Lactobacillus sp., Bifidobacte rium sp., Enterococcus sp., and Eubacterium sp.(37).The further biotransformation of liquiritin to the dihydrochalcone davidigenin constitutes a reductive cleavage of the ether bond in ring C.This conversion has been described to be accomplished by an oxygen-sensitive NADH-dependent flavanone-and flavanonol-cleaving reductase from the human intestinal anaerobe Eubacterium ramulus (54).Metabolism of the 2-glucosy loxy-4-methoxycinnamic acid isomer starts with deglycosylation, followed on one hand by formation of a δ-lactone ring, leading to intermittently increased herniarin, and on the other hand by reduction of the exocyclic double bond, leading to 2-(2-hydroxy-4methoxyphenyl) propionic acid that increased over time and may additionally result from further conversion of herniarin (Fig. 5A 3 ).While the lactonization of (Z)−2-glucosy loxy-4-methoxycinnamic acid to herniarin has been described in plants (55), it is hitherto unknown to be accomplished by gut microorganisms.2-(2-Hydroxy-4-methoxyphenyl) propionic has been observed as gut microbial metabolite of herniarin in rats (56).However, the microorganisms and enzymes involved in its formation which includes hydrolysis of the lactone ring and hydrogenation of the exocyclic double bond have not been discovered to date.
The only detectable metabolite potentially derived from the microbial biotransfor mation of rosmarinic acid was 3-(3-hydroxyphenyl)propanoic acid (Fig. 5A 4 ).Esterases capable to hydrolyze rosmarinic acid (57) and other hydroxycinnamic acid esters (58,59) are known from various intestinal microorganisms, including Lactobacillus sp., Bifido bacterium sp., and Escherichia sp.Subsequent hydrogenation of the exocyclic double bond and dehydroxylation in the aromatic ring lead to the formation of the observed 3-(3-hydroxyphenyl)propanoic acid (47).In an early study, Peptostreptococcus spp.and Clostridium perfringens were found capable to hydrogenate caffeic acid, and E. coli and Steptococcus faecalis var.liquefaciens were able to dehydroxylate dihydrocaffeic acid (60).Studies on involved enzymes are not available to date.
In the present study, we did not observe any significant changes in the relative abundance of any of these taxa or identify any of them as characteristics microbial signatures for STW 5-treated HFS.Moreover, contrary to our presumptions, the taxa characteristic for pre-digested VEH via MaAsLin analysis were not negatively correlated with the STW 5 constituents or their metabolites.Hence, even though we could detect the turnover of the STW 5 constituents into their major microbiome-induced metabo lites, we were not able to identify the responsible taxa.This may have several reasons: first, in the substrate-limited batch culture where no nutrients had been added, with just the remaining nutrients from the fecal suspension available, STW 5 constituents might not have been sufficient to allow survival and/or replication of the present microbes, thus leading to only minor and insignificant microbial shifts in the incubates; second, possibly the taxa responsible for the observed biotransformation reactions did not increase in their relative abundance, e.g., because the biotransformation was accom plished by microbial exoenzymes; and third, the biotransformations possibly have been accomplished by a panel of several microbial taxa with the same functional capabilities, which are therefore difficult to trace back by means of correlation analysis.Future studies should therefore include functional analyses, preferably on RNA level, to delineate the responsible microbial enzymes and taxa, upregulation of which would be expected.
Overall, our results indicate that despite STW 5 addition did not significantly affect the fecal microbiome composition, major constituents of STW 5 that are possibly able to reach the colon were extensively biotransformed by fecal bacteria, regardless of the donor's health status, to form breakdown products.The metabolites originating from the constituents of STW 5 may subsequently have the potential to elicit pharmacological effects.For example, davidigenin, the metabolite resulting from biotransformation of liquiritin pentosides, has been shown to possess anti-inflammatory and spasmolytic activity in vitro and ex vivo (61,62).Also, for 3-(3-hydroxyphenyl)propanoic acid and 18β-glycyrrhetinic acid, the metabolites resulting from glycyrrhizic acid and rosmar inic acid biotransformation, anti-inflammatory and cytoprotective effects have been described (63)(64)(65).These bioactive metabolites may contribute to the beneficial effects observed for STW 5 in IBS.
In this study, we were able to show that microbial biotransformation of STW 5 constituents into potentially bioactive metabolites occurs fast, but independent from large microbial community shifts and independent from the health status of the fecal sample donors.The fact that also gut microbiota from IBS donors are able to produce these metabolites indicates that IBS patients may profit from their potential health-bene ficial effects.

Herbal medicinal product information
Lyophilized STW 5 (dry residue 5.71%, batch number 631714) was kindly provided by Steigerwald Arzneimittel GmbH (Bayer Consumer Health).The quality of this batch complies with the quality prerequisites for STW 5 as described by reference (66).The preparation was pre-digested in a static three-phase in vitro digestion model called InfoGest (30).After the intestinal phase of pre-digestion, the final pre-digested STW 5, as well as the pre-digested vehicle, was composed of a mixture of STW 5 constituents (vehicle = 0.625% ethanol for pre-digested vehicle), fluids of all three phases (simulated salivary, stomach, and small intestine fluid), the enzymes pepsin and pancreatin, and bile acids (30).

Preparation of pre-digested study samples
In vitro digestion of lyophilized STW 5 was performed as previously described (30), based on the slightly modified Infogest protocol (67), a standardized static in vitro model to simulate upper intestinal tract digestion.Briefly, a solution of lyophilized STW 5 in 5% (vol/vol) ethanol (6.96 mg/mL) was subsequently subjected to oral phase diges tion with simulated salivary fluid containing no α-amylase, to gastric phase digestion, involving incubation with simulated gastric fluid containing pepsin, and to intestinal phase digestion, i.e., incubation with simulated intestinal fluid containing pancreatin and ox bile.In every digestive phase, the sample was diluted 1:1, resulting in a final STW 5 concentration of 0.87 mg/mL.In parallel, a control sample (VEH) was prepared by subjecting 5% (vol/vol) aqueous ethanol to the same incubation procedure.The incubation experiment was performed in 10 replicates for STW 5 and for VEH, respec tively.The intestinal phase incubates of STW 5 and the control sample (VEH) (ca.80 mL per replicate) were pooled, divided into 40 mL of aliquots, and kept frozen at −80°C prior to being subjected to incubation with fecal suspensions.
For the incubation experiment, 27 mL of 11.11% HFS was mixed with 3 mL pre-diges ted STW 5 or pre-digested vehicle in five replicates under anaerobic conditions at 37°C to achieve a final concentration of 10% HFS.To rule out degradation by the incubation conditions, the stability of pre-digested STW 5 during incubation in anoxic PBS buffer containing no fecal suspension was assessed by incubating 27 mL anoxic PBS buffer with 3 mL pre-digested STW 5 or pre-digested vehicle in five replicates (70).The final concentration 0.087 mg of STW 5 lyophilizate per mL of a 10% HFS was based on the daily dosage of STW 5 (corresponds to 171.3 mg lyophilizate) and an estimated small intestine fluid volume of approximately 200 mL (30).Samples were collected after 30 min (tp 0.5 ), 4 h (tp 4 ), and 24 h (tp 24 ).

Microbiome analysis-DNA extraction, microbial load and composition
For microbiome analysis, 250 µL of the drawn samples was immediately treated with 1.25 µL PMA to evaluate the abundance of viable bacterial cells present at this time point.After incubation without light exposure for 5 min (IKA Rocker 2D digital) and subse quent light treatment (PMA-Lite LED Photolysis Device; Biotium) for 15 min, samples were frozen at −80°C and stored until analysis.HFS microbiome control samples (c 0Mic ) without any additive (vehicle or STW 5) were prepared in the same way.DNA was extracted from all samples containing HFS (in total 560 samples from time points c 0Mic , tp 0.5 , tp 4 , and tp 24 ) using the E.Z.N.A. Stool DNA Kit (Omega bio-tek) according to the manufacturer's extraction protocol.

Absolute abundance
For quantification, Sso Advanced Universal SYBR Green Supermix (Bio-Rad) was used together with the following primer pair, forward primer 331F (5′-TCCTACGGGAGGCAG CAGT-3′) and reverse primer 797R (5′-GGACTACCAGGGTATCTAATCCTGTT-3′) (71).PCR reactions were carried out in a CFX384 Touch Real-time PCR Detection System (Bio-Rad).Each sample was analyzed in triplicates.After an initial denaturation at 95°C for 15 min, the following cycling conditions were repeated 40 times: 94°C for 15 s (denaturation), 54°C for 30 s (annealing), 72°C for 40 s (elongation + plate read).The efficiency of all runs was >90%.The detection limit was defined by the average Cq values of our non-template controls.The quantity of each sample was given in microbial load per gram of stool.

Relative abundance
16S rRNA-targeted Illumina Next-Generation Sequencing was performed for all collected samples.To amplify the 16S rRNA gene V4 hypervariable region, the following primers: F515 (5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGGTGCCAG CMGCCGCGGTAA-3′) and R806 (5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGGAC TACHVGGGTWTCTAAT-3′) (72).The reaction conditions were: 94°C for 3 min (initial denaturation), followed by 32 cycles of 94°C for 45 s (denaturation), 50°C for 60 s (annealing), and 72°C for 90 s (elongation), and 72°C for 10 min (final elongation).Extraction blanks and PCR negative controls were processed in parallel.The amplicons were sequenced by the Illumina MiSeq technique, which was performed in cooperation with the Core Facility Molecular Biology at the Center for Medical Research at the Medical University of Graz, Austria (73).
In short, the resulting fastq files were analyzed using Quantitative Insights Into Microbial Ecology (QIIME2) (74), which includes the Divisive Amplicon Denoising Algorithm (DADA2) (75).Paired end reads were joined and the quality of the produced sequences was checked.Representative sequences were taxonomically classified using SILVA v132 as the reference database (76).The R package decontam (77) was used to deal with potential contaminations (threshold 0.5 and prevalence method).Subsequently, features belonging to mitochondria and chloroplasts were removed together with features with 0 reads.SRS (scaling with ranked subsampling; https://vitorheidrich.shinyapps.io/srsshinyapp/)normalization was used to set a minimum sampling depth of 10,000 reads per sample.

Statistical analysis and data visualization of the microbial information
The final RSV table was used to perform different microbiome analyses using the web-based tool NAMCO (78) (including alpha and beta diversity) and R studio.The packages used for the different analyses and visualizations are as follows: "Maaslin2" (differential abundance analysis; default settings); "ggpubr" and "tidyverse" (Spearman's rho correlation and heatmaps); "ggpubr, " "tidyverse, " "dplyr, " and "scales" (most abundant taxa).Furthermore, an overview of the most abundant taxa was also generated using the web-based tool RAWgraphs (79).
Statistical analyses were performed with IBM SPSS Amos version 27 and the effect size of the PCoA plots was calculated using the R packages "phyloseq, " phytools, " and "vegan."

Ultra high-performance liquid chromatography-mass spectrometry metabo lomics analysis and data processing
For UHPLC-HRMS analysis, the HFS and PBS buffer samples were centrifuged at 13,000 rpm at 4°C for 10 min, filtrated (0.45 µm), aliquoted, and stored at −20°C until utilization.In addition, metabolomic control samples (c 0Met ) were prepared by mixing centrifuged and filtrated HFS with pre-digested STW 5 or vehicle at a 10:1 ratio.
Before UHPLC-HRMS, the samples were thawed, vortexed three times for 10 s, and centrifuged at 13,000 rpm for 10 min at room temperature.In addition, we measured a blank with pure methanol.For UHPLC-HRMS analysis, we used the same method described in (30).
The LCMS data for each donor were processed using Compound Discoverer 2.1, separately in negative and positive mode as described previously (30), with the following modifications: Total intensity threshold was 100,000 for spectrum selection and detection of unknown compounds.RT tolerance for grouping was 0.3 min.The resulting feature table was exported to Microsoft Excel, where mean peak areas of the five technical replicates were calculated and areas were compared between STW 5-and vehicle-treated samples.Features with ratios >150% of the HFSs from at least one of the donors or time points, as well as features with ratios >150% in the PBS buffer samples without HFS were considered to be related to STW 5 or its metabolites and kept for further data evaluation.In order to calculate the metabolization ratios of the genuine STW 5 constituents and the newly formed metabolites at each time point (tp 0.5 and tp 4 ), ratios were set in relation to the c 0Met samples (i.e., microbiota-free control samples in which STW 5 was added after centrifugation and filtration of the fecal suspensions).Significantly different features (P < 0.05; peak areas and metabolization ratios: tp 0.5 /c 0Met or tp 4 /c 0Met ) were determined using ALDEx2 (80).PBS-buffer control incubations for pre-digested STW 5 and its vehicle were also prepared for the same time points (tp 0.5 or tp 4 ).The respective peak areas of the detected STW 5 constituents were compared to the peak areas in the microbiome-free contol samples to differentiate metabolization processes induced by gut microbiota from those induced by other factors, such as incubation temperature and time (70).
Due to the low concentration of STW 5 used in the incubation experiments (0.087 mg/mL) in order to stay in the range of the recommended daily dose of the preparation, and due to the interference of highly abundant matrix constituents derived from ex vivo pre-digestion and from the fecal samples, only the major STW 5 constitu ents were detectable in c 0Met samples and therefore annotated.Annotation of STW 5 constituents and metabolites was either performed by comparison with authentic references or based on reference data in the literature.

FIG 2
FIG 2 Microbial composition of non-IBS and IBS HFS at baseline (c 0Mic ).(A) Pie charts of the 15 most abundant genera found in non-IBS and IBS samples, respectively.(B) Alpha diversity of the gut microbiome is not affected by health status.The means of the participant replicates were used for the analysis to deal with the technical replicates (Health status: unpaired; (Continued on next page)

FIG 6
FIG 6 Spearman's rho correlation between metabolites and top 15 markers (A) and biomarkers for VEH and STW 5 (B), respectively, found in the combined data set.