Comparative prebiotic potential of galacto- and fructo-oligosaccharides, native inulin, and acacia gum in Kenyan infant gut microbiota during iron supplementation

Abstract Iron fortification to prevent anemia in African infants increases colonic iron levels, favoring the growth of enteropathogens. The use of prebiotics may be an effective strategy to reduce these detrimental effects. Using the African infant PolyFermS gut model, we compared the effect of the prebiotics short-chain galacto- with long-chain fructo-oligosaccharides (scGOS/lcFOS) and native inulin, and the emerging prebiotic acacia gum, a branched-polysaccharide–protein complex consisting of arabinose and galactose, during iron supplementation on four Kenyan infant gut microbiota. Iron supplementation did not alter the microbiota but promoted Clostridioides difficile in one microbiota. The prebiotic effect of scGOS/lcFOS and inulin was confirmed during iron supplementation in all investigated Kenyan infant gut microbiota, leading to higher abundance of bifidobacteria, increased production of acetate, propionate, and butyrate, and a significant shift in microbiota composition compared to non-supplemented microbiota. The abundance of the pathogens Clostridium difficile and Clostridium perfringens was also inhibited upon addition of the prebiotic fibers. Acacia gum had no effect on any of the microbiota. In conclusion, scGOS/lcFOS and inulin, but not acacia gum, showed a donor-independent strong prebiotic potential in Kenyan infant gut microbiota. This study demonstrates the relevance of comparing fibers in vitro prior to clinical studies.


Introduction
Iron-deficiency anemia (IDA) is a severe condition that can lead to impaired cognitive and motor development [1,2].An estimated 20%-25% of children worldwide suffer from IDA, with the highest prevalence reported in low-and middle-income countries of South Asia and Africa [3].Kenya has a particularly high IDA prevalence, with reported estimates between 62% and 73% among infants aged 8-12 months [4,5].Supplementing complementary food with iron-containing micronutrient powders is a common strategy to reduce IDA in African infants during weaning [6,7].However, such iron fortification sharply increases the colonic iron levels, as typical absorption in the small intestine is 10%-20% [8].Unabsorbed iron remains available for gut microbes and may potentially affect the gut microbiome homeostasis [9,10].
Indeed, studies in Kenyan infants aged 6 months and 6.5-9.5 months reported adverse effects with dietary iron fortification (12.5 and 5 mg iron/day) on the gut microbiota, with decreased abundance of infant health-associated bifidobacteria and lactobacilli concomitant with increased Enterobacteriaceae and pathogenic Escherichia coli [11][12][13].Similarly, supplementation with 12.5 mg iron/day in the form of micronutrient powder resulted in increasing abundance of Escherichia-Shigella and decreased Bifidobacterium in 6-month-old Pakistani infants [14].Although the exact mechanism of the observed iron-induced dysbiosis and adverse gut events are not fully characterized, strategies to improve the safety of iron fortification are currently being investigated.
Iron is essential for fundamental processes in both mammalian and microbial cells, and its homeostasis is tightly regulated [15,16].For example, iron is a cofactor in iron-dependent enzymes involved in the butyrate production pathways and its supplementation promoted butyrate gut levels in rats [17,18] and butyrate production by Roseburia intestinalis in vitro, while pronounced iron deficiency inhibited in vitro butyrate production by the gut microbiota of children [19].Most enteropathogenic bacteria, such as Salmonella, Shigella, and pathogenic E. coli, rely on iron acquisition for gut colonization and for the expression of virulence genes [20].Therefore, excess iron in the gut lumen may stimulate the colonization by and virulence of various pathogenic bacteria [8,21].In contrast, beneficial commensal gut bacteria such as lactobacilli were shown to require iron only under particular environmental conditions [22], and bifidobacteria isolated from Kenyan infant feces efficiently bound iron via siderophores, potentially limiting the iron availability to pathogens [23].Thus, higher colonic iron levels resulting from fortification may shift the colonic microbiota equilibrium to favor the growth and virulence of pathogenic strains over healthy barrier-protecting strains in infants living in rural areas of low-and middle-income countries.
Applying prebiotics is a potent strategy to stimulate the growth and colonization of specific bacteria in the gut to confer a health benefit [24], and to promote production of short-chain fatty acids (SCFAs).Such acids decrease colonic pH, which can lead to inhibition of enteropathogens [25] and improved iron absorption [26].Adding prebiotic galacto-oligosaccharides (GOS) to a micronutrient powder containing a lower dose of highly bioavailable iron (5 mg/day) partially prevented enteropathogen stimulation by iron supplementation in Kenyan infants; treatment with GOS resulted in higher abundances of Bifidobacteriaceae and Lactobacillaceae and lower abundances of pathogenic virulence and toxin genes [4,12].The prebiotic capacity of short-chain GOS combined with long-chain fructo-oligosaccharides (FOS) (scGOS/lcFOS) [27,28] and of inulin [29] was demonstrated in trials involving Western infants; however, this same effect has not yet been investigated in rural African infant cohorts, which have different gut microbial compositions including different enteropathogens [12,30].Compared to GOS, acacia gum, an arabinogalactan-protein complex, is a more affordable emerging prebiotic that can be fermented by human adult microbiota [31][32][33] and has a demonstrated bifidogenic effect in human adults [34,35].Therefore, it is also a potent prebiotic candidate for its ability to promote barrier-protective bacterial taxa and its functions during iron fortification in the African infant gut microbiota.
In vitro gut microbiota models are effective tools for studying the human gut microbiota function and composition in a controlled environment independent of the host [36].We recently adapted the in vitro continuous gut fermentation model PolyFermS to mimic conditions of infant gut microbiota from rural Kenya through protected transport of fresh fecal samples and adaptation of in vitro cultivation medium and fermentation conditions to the Kenyan infant diet [37,38].
Therefore, the aim of this study was to investigate the effect of prebiotics (scGOS + lcFOS, inulin) and the emerging prebiotic acacia gum during iron supplementation at a dose recommended by the World Health Organization (12.5 mg iron/day to reduce IDA in 6-to 23-month-old infants) on the gut microbiota of African infants at weaning age using the African infant PolyFermS model.We hypothesized that the selected fibers promote growth and metabolism of health-associated infant gut bacteria and inhibit the growth of opportunistic enteropathogens.

Fecal sample collection, transport, and processing
Fresh fecal samples from four partially breastfed Kenyan infants donors (D1, D2, D3, and D4) aged between 4.8 and 7.9 months living in rural Kenya (Msambweni County) were collected and processed as previously described [37] and were part of the African infant PolyFermS model setup study [38].None of the infants received antibiotics or dietary supplements prior to sample donation; detailed fecal donor information is presented in supplemental data.The fecal samples were collected in context of the registered clinical trial on clinicaltrials.gov(NCT03866837) and from participants prior to intervention.The Ethics Commission of ETH Zürich, Switzerland (EK 2019-N-04) and the Kenya Medical Research Institute (KEMRI) Scientific and Ethics Review Unit (SERU) (KEMRI/RES/7/3/1 no.656) reviewed and approved this study.Informed consent was obtained from the parents or the legal guardians of the infants.

Experimental design for investigating prebiotic effects
The prebiotic potential of three different fibers: acacia gum (FIBREGUM LI, Nexira, France), short-chain GOS (Vivinal GOS powder, Friesland-Campina, The Netherlands) combined with long-chain FOS (Orafti HP, Beneo, Belgium) (scGOS/lcFOS at a 9:1 ratio), and native inulin (Orafti ST, Beneo, Belgium), alone and in combination with iron, were tested.Seven treatments were tested in parallel in the TRs: iron, acacia gum, acacia gum + iron, inulin, inulin + iron, scGOS/lcFOS, and scGOS/lcFOS + iron, all repeated during two experimental phases (Fig. 1).Each phase consisted of a stabilization period (6-13 days) followed by a treatment period (6-9 days).The fibers were supplemented to mimic a daily consumption dose of 5 g /infant.Considering a retention time of 8 h [39] and an infant colon capacity of 300 ml [40], the daily dosage was calculated to be 3.3 g/day/reactor or 5.5 g/l of medium feeding the reactors.The purity of the fiber powders (acacia gum: 100%, scGOS:lcFOS: 70% and inulin: 95%) was taken into account when calculating medium supplementation rates (Table S2).The iron supplementation mimicked oral supplementation of 12.5 mg Each stabilization and treatment period lasted for 6 to 13 days, which was when the metabolic pseudo-steady state, which was monitored by the day-to-day variation of fermentation metabolites, was reached.Variations in day-to-day metabolite production in PolyFermS models are expected to be lower than 10%.This criterium was used to define functional microbial stability and the start or end of a treatment period.Between each treatment period, TRs were disconnected from IR, washed, sterilized and reconnected to the IR to avoid any effects of the previous applied treatment and re-establish the bacterial composition during the consecutive stabilization period.
Eff luent samples from all PolyFermS reactors were taken daily.Samples were centrifuged (14 000 x g for 10 min at 4 • C), and the supernatant was immediately processed or stored at −20 • C, while the bacterial pellet was stored at −80 • C until further analysis.

Microbiota metabolite analysis
Concentrations of SCFA (acetate, butyrate, propionate), intermediate fermentation metabolites (formate, succinate, lactate) and branched-chain fatty acids (isobutyrate, isovalerate, valerate) were determined in PolyFermS eff luent supernatant and fecal samples with a LaChrom high-performance liquid chromatography-system (Merck-Hitachi, Tokyo, Japan) was used with a Security Guard Cartridge Carbo-H (4 × 3.0 mm; Phenomenex Inc., Torrance, CA, United States) connected to a Rezex ROA-organic acid H+ column (300 x 7.8 mm; Phenomenex Inc., Torrance, CA, USA) and an Accela RI detector (Thermo Fisher Scientific Inc., Waltham, MA, USA).Sample volumes of 20 μl were analyzed with a mobile phase of 10 mM H 2 SO 4 at a f low rate of 0.6 ml/min and a column temperature of 40 • C. The data were processed using EZChrom software (Agilent, Santa Clara, CA, USA).

Microbiota composition analysis
The DNA from feces and eff luent pellets was extracted using the Fast DNA TM Spin Kit for Soil (MP Biomedicals, Illkirch-Graffenstaden, France) following the supplier's protocol.The DNA concentration was measured with Nanodrop (ND-1000, Witec AG, Sursee, Switzerland), and samples were diluted to 20 ng DNA/μl using DES water (Fast DNA TM Spin Kit, MP Biomedicals) and stored at −20 • C until further processing.
Selected bacterial targets were further quantified with qPCR in feces and PolyFermS eff luent from the last three days of stabilization and treatment periods with primers listed in in Table S3.Bacterial communities in the fecal and PolyFermS samples were analyzed with V4 16S rRNA gene amplicon sequencing with an Illumina MiSeq platform (Genetic Diversity Centre, ETH Zurich, Switzerland) and described in supplementary methods.

Statistical analysis
Statistical analyses were performed in R version 4.2.0 with RStudio 2022.2 and GraphPad 9.2.0 and details are given in supplementary methods.Each dataset was first tested for normality using the Shapiro-Wilk test.Because data were not normally distributed, the nonparametric Wilcoxon rank sum test with false-discovery rate correction was used.P values <0.05 were considered statistically significant.

Individual donor fecal microbiota profiles were reproduced and maintained in vitro
To assess the effect of the selected fiber mixtures on the target gut microbiota, we used the validated in vitro continuous PolyFermS fermentation model for African infant gut microbiota [38].We performed four independent PolyFermS experiments with IRs containing immobilized Kenyan infant fecal microbiota.Basal data for untreated IR were previously described [38].
The fermentation metabolite profile for D1, D3, and D4 microbiota was characterized as propiogenic (propionate as main end metabolite after acetate) and for D2 microbiota as butyrogenic (butyrate as main end metabolite after acetate) (Fig. S1C).Over the course of the different experimental periods, the four different IR microbiota had stable fermentation metabolite production and stable abundant genera composition (Figs S2 and S3).Overall, our study consisted of in vitro microbiota that differed on alphadiversity, pathogens load and SCFA and thus allows us to test the prebiotic effects of the three fibers and possible donor specificity.

Consistent bifidogenic and SCFA-promoting effects of scGOS/lcFOS and inulin supplementation across all donors
Weighted Jaccard PCoA indicated that supplementation with scGOS/lcFOS and inulin, irrespective of iron addition, induced a stronger significant shift in overall composition in D1, D2, and D3 microbiota compared to acacia gum, which was driven by the genus Bifidobacterium (Fig. 3A).We also observed that changes in microbiota composition upon treatment were very consistent between the two treatment periods within each modelled microbiota (Fig. 3B).Iron supplementation did not induce an overall shift in microbiota composition except in D3 (Fig. 3B) (PERMANOVA, P = 0.043 and P = 0.0082 in periods 1 and 2, respectively).No significant treatment-induced shift in overall microbiota composition was detected for D4 microbiota.Neither iron nor any of the fiber supplementations resulted in significantly changes in alpha diversity-based ASV richness and Pielou's evenness when compared to the distance between treatment and stabilization phases of the non-treated IR (Fig. S4).
The bifidogenic effect from scGOS/lcFOS and/or inulin when supplemented alone or when combined with iron was confirmed by qPCR in all four in vitro donor microbiota, resulting in increased Bifidobacterium concentrations after fiber supplementation compared to the stabilization period (Fig. 4A).The concentration of Bifidobacterium increased the most in D2 microbiota after fiber supplementation (e.g.+0.72 ± 0.05 log for inulin and + 1.28 ± 0.03 log for inulin with iron).Acacia gum only impacted Bifidobacterium growth in D3 microbiota.Supplementation with iron alone did not result in an overall decrease in Bifidobacterium, though a decrease was observed in one replicate treatment in each of the D2 and D4 microbiota.We observed that Bifidobacterium ASV0003 (closest assigned to Bifidobacterium breve, Bifidobacterium scaligerum, and Bifidobacterium longum subsp.infantis) was dominant in three donors (D2, D3, D4), while ASV0008 (closest to Bifidobacterium adolescentis) and ASV0013 (closest to Bifidobacterium callitrichidarum, Bifidobacterium catenulatum, Bifidobacterium gallicum, and Bifidobacterium pseudocatenulatum) dominated in D1 (Fig. 4B and Table S4).Bifidogenic effects of both scGOS/lcFOS and inulin induced an increase of the most abundant Bifidobacterium ASVs in D1, D2, and D3 compared to the levels observed before supplementation.In contrast, when scGOS/lcFOS was supplemented during iron treatment in D1, only the relative abundance of ASV0013 increased in both periods, which suggests an iron-induced change in bifidogenic effect in D1 microbiota.
The prebiotics scGOS/lcFOS as well as inulin, both with and without iron, showed an evident increase in production of total fermentation metabolites upon initial treatment when compared to non-treated IR and the treatments iron and acacia gum (Fig. 5A and Fig. S5).Iron treatment did not result in a significant or consistent alteration in acetate levels in all four microbiota in both treatment periods.In contrast, supplementation with scGOS/lcFOS and inulin resulted in a strong acetogenic response, with increased acetate up to +40 mM for D1, D3, D4 and up to +15 mM in D2 in vitro microbiota (Fig. 5B).Butyrate production was consistently increased upon supplementation with scGOS/lcFOS and inulin, independent of iron supplementation, in D2 (+9 to +13 mM), D3 (+2 to +3 mM) and D4 (+0.8 to +4.5 mM) microbiota (Fig. 5B).Increased propionate production was observed for all treatments across the different donors, with the largest effect for scGOS/lcFOS in D1 microbiota (e.g.+13.0 ± 1.7 mM for scGOS/lcFOS and + 8.3 ± 3.7 mM for scGOS/lcFOS with iron in period 1) (Fig. 5B).The intermediate metabolites formate or lactate increased upon scGOS/lcFOS and inulin supplementation with and without iron in D1, D3 and D4 in vitro microbiota (up to +11 mM for formate with scGOS/lcFOS in D3 and up to +15 mM for lactate with inulin in D1) (Fig. S6).

Donor-and prebiotic-specific changes in opportunistic enteropathogen concentration
Supplementation with scGOS/lcFOS resulted in a consistent decrease of C. perfringens in D1 and D4 microbiota under both iron-supplemented and iron-limited conditions when compared to IR and iron treatment (Fig. 6A).With acacia gum and inulin, the detected decreases were low (D1) or were not replicated in both treatment periods (D4).Iron supplementation resulted in a significant increase in C. difficile concentrations in D2 microbiota (+ 1.0 ± 0.3 log), which was not the case when fibers were cosupplemented (Fig. 6B).In D1 and D4 microbiota, supplementation with inulin and particularly scGOS/lcFOS resulted in lower C. difficile concentrations.The treatment effect on EPEC concentrations was variable across fiber supplementations; only a consistent increase upon supplementation with acacia gum combined with iron was observed in Donor 4 microbiota (Fig. 6C).

Donor-microbiota dependent taxonomic response to fiber treatments
We next assessed the differential response of gut microbiota by 16S RNA sequencing by comparing the different treatments to non-treated IR microbiota with DESeq2 analysis.Only the genera with both significant and consistent log2 fold changes (FC) between treatment and stabilization in both periods are presented (Fig. 7, Table S5).
Overall, supplementation with iron had a limited effect on the genus relative abundance, with an increase in low relative abundant Propionibacterium (log 2 FC +4.8 ± 1, to 0.10%) and a decrease in Morganella (log 2 FC -1.3 ± 0.2, to 0.47%) in D1 microbiota, and a decrease in low relative abundant Peptoniphilus (log 2 FC -1.6 ± 1.4, to 0.39%) in D2 microbiota.No change in Enterobacteriaceae genera was observed in any of the investigated donor microbiota.In line with qPCR results in D2 microbiota, an increased relative abundance of Clostridioides (log 2 FC +3.6 ± 1.3, from 0.02% to 0.33%) was observed after iron supplementation (Fig. 7, Table S5).
We also observed genus increases upon fiber supplementation when no iron was co-supplemented such as upon scGOS/lcFOS supplementation in in vitro microbiota of D1 (Streptococcus, from 22.4% to 26.3%), D2 (Megasphaera, from 7.4% to 15.3%) and D3 (Flavonifractor, from 5.1% to 6.9% and Collinsella, from 16.3% to 24.7%), or upon inulin supplementation in D3 microbiota (Lactococcus, from 3.2% to 14.8%).On the other hand, iron supplementation in combination with scGOS/lcFOS enhanced Dialister in D2, or iron with inulin enhanced Levilactobacillus in D1 microbiota; these increases were not observed when the fibers were added alone.In line with the beta diversity data, no consistent significant taxonomic shifts were detected in D4 microbiota in the two treatment periods.

Dose-response effects for inulin
Inulin was selected to determine whether a 50% lower dose still stimulated SCFA production and growth of Bifidobacterium in D3 and D4 in vitro microbiota.Both the basal and 50% lower inulin dose resulted in a reproducible acetogenic effect in both tested microbiota (Fig. 8A).The mean increase in acetate production upon the lower inulin dose (+ 15 mM for D3 and + 22 mM for D4) was half as high compared to the basal inulin dose (+ 31 mM for D3 and + 45 mM for D4), showing a clear dose-response for  inulin on acetate production.Butyrate production was enhanced upon the 50% inulin dose in both microbiota that were tested in Period 2, but to a lesser extent than the 100% dose.The 50% lower inulin dose resulted in a bifidogenic effect, but only in period 2 in both microbiota and thus was less effective in promoting Bifidobacterium growth compared to the basal inulin dose (Fig. 8B).The latter was ref lected in the overall microbiota response, as a reproducible change in weighted Jaccard distance from the nontreated microbiota was only observed for the basal and not for the lower inulin dose (Fig. S7).

Discussion
In this in vitro study, we compared the prebiotic potential of three different fibers alone and in combination with iron in four different gut microbiota of Kenyan infants living in low hygiene environments and at high risk of developing anemia during weaning.
Our first objective was to induce an alteration of the gut microbiome with iron supplementation.However, on our in vitro model, iron supplementation at 12.5 mg iron/day did not impact the Enterobacteriaceae abundance or the overall gut microbiota profile, which contradicts previous in vivo findings [11][12][13].Similar results were seen in children populations with low enteropathogen levels (50 mg iron/day, [41]) and in in vitro studies with child microbiota (30.4 mg iron/day, [42]).In healthy adult human fecal microbiota, 3.4 mg/day of iron led to minimal, donorspecific microbiota changes without increased Enterobacteriaceae levels [43].Our study differs from previous in vitro studies, as we investigated African infant gut microbiota containing several naturally-occurring opportunistic enteropathogens.Our data suggest that the in vivo observed iron-induced microbiota dysbiosis may be due to an iron-host-microbiota interaction instead of a direct iron-microbiota interaction.
We observed that iron supplementation in vitro resulted in increased concentration and relative abundance of Clostridioides difficile in one microbiota.Such an increase was also observed in previous trials with Kenyan infants aged 6-10 months after iron supplementation with [44] or without [11] antibiotics.In line with this, a recent study demonstrated that iron (Fe3+, which is also  S5. the form expected in our in vitro model at pH 5.8) stimulated the growth of C. difficile in a dose-dependent manner and consequently led to higher levels of Toxin A, necessary for pathogenesis, after iron supplementation [45].
In our model, scGOS/lcFOS and inulin inhibited growth of C. difficile and C. perfringens in two of the four in vitro microbiota both during and without iron supplementation, which is consistent with studies in European infants supplemented with the same prebiotics [46][47][48].Decreased levels of enteropathogens, concomitant with a pronounced increase in SCFA production, were measured upon supplementation with scGOS/lcFOS and inulin with iron.We anticipate that this effect could be stronger in vivo because the resulting lower pH may further inhibit the growth of enteropathogens.Additionally, SCFA also inhibit the expression of their virulence genes [49].A lower fecal pH was indeed detected in European infant studies involving formula supplemented with scGOS/lcFOS or inulin [27,48,50].
Supplementation with scGOS/lcFOS and native inulin strongly enhanced production of acetate (up to 2-fold) and the growth of Bifidobacterium during iron supplementation in all studied donor microbiota, despite their baseline compositional and functional differences.This is consistent with previous studies in European infants who received the same scGOS/lcFOS (9:1 ratio) [27,47,51] or with in vitro batch incubations with European fecal infant microbiota supplemented with native inulin [29,52].We observed that a 50% lower dose of native inulin equally promoted acetate production and Bifidobacterium growth, but its overall effect on the microbial community composition was lower compared to the 100% dose, which suggests that higher doses may be needed to induce community changes.The fiber or prebiotic dose-response effect on the gut microbiome is not well studied in infants.In adults, trials with distinct dietary fibers showed that the dominant effects were dose-dependent and plateaued at 35 g/day [53].
While Bifidobacterium exhibited convergent response across donors, other taxa had variable responses to fiber treatments during and without iron supplementation.This emphasizes the importance of considering individual responses when assessing the impact of prebiotics on gut microbiota as previously observed for adult microbiota [53][54][55].In our study, scGOS/lcFOS demonstrated a butyrogenic effect alongside a reduced acetogenic response in the D2 microbiota, suggesting potential cross-feeding interactions between different bacterial groups, possibly involving acetate-to butyrate-producing taxa or lactate-to butyrateproducing taxa [56].Megasphaera, containing butyrate-producing species, was notably stimulated by scGOS/lcFOS supplementation in D2 microbiota.Previously, cross-feeding between FOS-induced lactate and butyrate production by Megasphaera elsdenii was demonstrated in an animal model [57].Also, other taxa from the detected genera Eubacterium, Hungatella, Flavonifractor, and Lachnoclostridium could have contributed to the increased butyrate production in D2 microbiota.Next, Eggerthella, an acetate-, lactate-, and succinate-producing gut commensal, increased with scGOS/lcFOS and inulin in D1 microbiota and could explain the observed increase in lactate in this system.Increased relative abundance of Eubacterium in microbiota supplemented with scGOS/lcFOS and acacia gum was only observed in D3 microbiota, and was previously detected in rats upon GOS supplementation [58].Inulin supplementation led to lower Bacteroides relative abundance in D2 microbiota, potentially due to the competitive advantage of inulin-degrading Bifidobacterium taxa.Our in vitro finding supports the negative association detected between Bacteroides fragilis and two Bifidobacterium species in fecal metagenomes from the same Kenyan infant population [59].
The emerging prebiotic candidate acacia gum induced little to no effect on the model African infant gut microbiota in contrast to the established prebiotics scGOS/lcFOS and inulin.The chemical structure of acacia gum is more complex and its degradation requires different bacterial glycoside hydrolases (GH) from different carbohydrate-active enzyme families including GH16: endo-β-1,3-galactanase, GH43: exoβ-1,3galactanase, GH154/GH105: β-glucuronidase, GH145/PL27: α-L-rhamnosidase/L-rhamnose-α-1,4-D-glucuronate lyase, and GH27/GH97: β-L-arabinopyranosidase/ α-galactosidase [60].Moreover, B. longum subsp.longum was recently identified as a specialist degrader of acacia gum, because of its enzyme 3-O-α-D-galactosyl-α-L-arabinofuranosidase (GH39) [61].Acacia gum-degrading taxa could be absent or at low prevalence in the gut microbiota of partially breast-fed Kenyan infants aged 6-10 months.This was indeed confirmed in a recently published metagenomic study of fecal samples of the same Kenyan infant population, which detected genes for only three of the five bacterial enzymes for acacia gum degradation (GH43, GH16, and GH27) at low gene prevalence and abundance [59].Previously published bifidogenic and SCFA-promoting effects of acacia gum were mainly reported in studies with human adult feces [32,34,60]; in vitro studies involving American infant feces also did not detect acacia gum fermentation [62,63].Therefore, our study highlights the importance of evaluating established prebiotics and emerging prebiotic candidates in the target host microbiota.In this case, there is a marked difference in functional capacity and phylogenetic composition between Western and rural African infants, which are still insufficiently characterized [64].
We conclude that both scGOS/lcFOS and native inulin, but not acacia gum, have strong prebiotic potential in Kenyan infant gut microbiota during iron supplementation by promoting growth of beneficial gut bacteria including bifidobacteria and inhibiting enteropathogen growth by enhanced SCFA production.Both fibers promoted butyrate or propionate production in a microbiotaspecific way.Acacia gum had little impact on fermentation and microbiota composition and therefore has low prebiotic potential in Kenyan infants at weaning age.This study demonstrates the importance of comparing prebiotics and other fibers in the target population in vitro prior to in vivo clinical studies.Future clinical trials with iron supplementation in combination with scGOS/l-cFOS or native inulin in African infants are needed to assess their prebiotic effect and impact on iron absorption.

Figure 1 .
Figure 1.Overview of experimental set-up of PolyFermS in vitro model.(A) IR with immobilized fecal Kenyan infant microbiota and parallel second-stage test reactors (TR1-TR7).(B) Assessment of the effect of prebiotic screening of acacia gum, inulin, and scGOS/lcFOS in combination with iron supplementation for both treatment periods.The supplemented dose for all fibers was mimicking 5 g/d and for iron 12.5 mg/d.(C) Dose-response assessment for inulin with low dose 2.5 g/d versus basal dose 5 g/d.STAB: stabilization period; TREAT: treatment period.

Figure 2 .
Figure 2. Differences in four donor in vitro baseline microbiota.Beta diversity represented as PCoA of binary Jaccard distance metrices comparing the 16S rRNA gene composition of feces and in vitro PolyFermS IR (inoculum) microbiota samples of all four donors (D1, D2, D3, and D4).Vectors indicate the top 10 genera that determine the binary Jaccard community structure.Data were derived from [38].

Figure 3 .
Figure 3. Overall microbiota composition response.(A) PCoA of weighted Jaccard distance metric assessing donor-dependent treatment effect upon supplementation of iron and of acacia gum, scGOS/lcFOS and inulin alone and in combination with iron in microbial composition, and main taxonomic drivers of intra-donor compositional differences in D1, D2, D3, and D4.(B) Weighted Jaccard distance metric assessing the intra-treatment induced microbiota shifts for D1, D2, D3, and D4 in period 1 (above) and period 2 (below).The values indicate the distance between the community at the end of stabilization and the community at the end of the treatment within each treatment.Pairwise Wilcoxon rank sum test with multiple comparison correction: * P < 0.05, * * P < 0.01, * * * P < 0.005, * * * * P < 0.001.

Figure 7 .
Figure 7. Donor-microbiota dependent taxonomic response.Differentially abundant genera bar plot visualization of DESeq2 outcome.Significant shifts (treatment vs. stabilization) upon treatments in genera relative abundance within TRs compared to IR microbiota in Period 1 and Period 2. The relative abundance and log2FoldChange is given in TableS5.