nutrients Article Effect of the Degree of Polymerization of Fructans on Ex Vivo Fermented Human Gut Microbiome Erola Astó 1,2 , Iago Méndez 1 , Maria Rodríguez-Prado 3 , Jordi Cuñé 1,† , Jordi Espadaler 1, * and Andreu Farran-Codina 2 1 AB-Biotics, S.A, ESADE Creapolis, Av. Torre Blanca, 57, E-08172 Sant Cugat del Vallès (Barcelona), Spain; asto@ab-biotics.com (E.A.); mendez@ab-biotics.com (I.M.); jordicune@gmail.com (J.C.) 2 Department of Nutrition, Food Science, and Gastronomy, XaRTA—INSA, Faculty of Pharmacy, University of Barcelona, Campus de l’Alimentació de Torribera, Av. Prat de la Riba, 171, E-08921 Santa Coloma de Gramenet, Spain; afarran@ub.edu 3 Animal Nutrition and Welfare Service (SNIBA), Building V. Office V0-308, Autonomous University of Barcelona, C/ Travessera dels Turons s/n, E-08193 Bellaterra (Barcelona), Spain; maria.rodriguez.prado@uab.cat * Correspondence: espadaler@ab-biotics.com; Tel.: +34-935-800-236 † Present address: Devicare, S.A, Eureka Building, UAB Research Park, Av. Can Domènech s/n, E-8193 Cerdanyola del Vallès (Barcelona), Spain.  Received: 28 April 2019; Accepted: 4 June 2019; Published: 7 June 2019  Abstract: Prebiotic supplements are used to promote gastrointestinal health by stimulating beneficial bacteria. The aim of this study was to compare the potential prebiotic effects of fructans with increasing degrees of polymerization, namely fructooligosaccharides (FOS) and inulins with a low and high polymerization degree (LPDI and HPDI, respectively), using an ex vivo fermentation system to simulate the colonic environment. The system was inoculated with pooled feces from three healthy donors with the same baseline enterotype. Changes in microbiota composition were measured by 16S metagenomic sequencing after 2, 7, and 14 days of fermentation, and acid production was measured throughout the experiment. Alpha-diversity decreased upon inoculation of the ex vivo fermentation under all treatments. Composition changed significantly across both treatments and time (ANOSIM p < 0.005 for both factors). HPDI and LPDI seemed to be similar to each other regarding composition and acidification activity, but different from the control and FOS. FOS differed from the control in terms of composition but not acidification. HDPI restored alpha-diversity on day 14 as compared to the control (Bonferroni p < 0.05). In conclusion, the prebiotic activity of fructans appears to depend on the degree of polymerization, with LPDI and especially HPDI having a greater effect than FOS. Keywords: prebiotic; inulin-type fructan; fructooligosaccharide; polymerization degree; metagenomics; microbiome 1. Introduction It is generally accepted that the bacterial community in the human gastrointestinal tract has a great impact on intestinal functionality and human health.The colon is the most colonized region within the gastrointestinal tract, where 1011 –1012 cells/mL have been detected [1,2]. The microbes in our bodies have been estimated to encode 100-fold more unique genes than our own genome [3,4]. High-quality data from the US Human Microbiome Project (HMP) [5], European Metagenomics of the Human Intestinal Tract (MetaHIT) [6] and several other studies have now demonstrated the beneficial role of the normal gut microbiota in health down to the genetic level [6,7]. Most bacterial species colonizing the human gastrointestinal tract belong to the phyla Firmicutes and Bacteroidetes, while species of the phyla Actinobacteria, Proteobacteria, and Verrucomicrobia exist in lower numbers. The Nutrients 2019, 11, 1293; doi:10.3390/nu11061293 www.mdpi.com/journal/nutrients Nutrients 2019, 11, 1293 2 of 14 number of bacterial species within the human intestinal microbiota has often been estimated to be in the range of 500 to over 1000 species [8]. In a healthy state, they contribute nutrients and energy to the host via the fermentation of nondigestible dietary components in the large intestine, and a balance is maintained with the host’s metabolism and immune system [9,10]. The prebiotic concept was first defined in 1995 as a “non-digestible food ingredient that beneficially affects the host by selectively stimulating the growth and/or activity of one or a limited number of bacteria already resident in the colon” [11]. The International Scientific Association for Probiotics and Prebiotics (ISAPP) recently published a consensus statement to update the definition and scope of prebiotics. The statement defines a prebiotic as “a substrate that is selectively utilized by host microorganisms conferring a health benefit” [12]. The scope of this updated definition is therefore wider, as it implies that the modulatory effect of prebiotics is not necessarily limited to gastrointestinal microbiota (i.e., it can be administered to microbiota-colonized sites other than the gut). Furthermore, the ISAPP consensus statement emphasizes that the definition of prebiotics should not be limited to carbohydrate-based compounds, but should apply to any compound that can be used by the microbiota to confer a health benefit. Some examples are polyphenols and polyunsaturated fatty acids. The document also addresses the difficulties involved in providing definitive proof of causality when assessing the beneficial effects of a prebiotic. Still, it states that the health benefits of a prebiotic must be confirmed in the target host for its intended use and should always be mediated by the microbiota. Inulin-type fructans (ITF) are the best-documented oligosaccharides for their effect on intestinal Bifidobacteria and are considered to be important prebiotic substrates [13]. Fructans can also be described by their polymerization degree (PD). Inulin-type fructans consisting of DP ≥10 units are considered long chain, while inulin-type fructans with a DP of <10 are considered short chain [14]. Nutritional studies have recommended a combination of inulins with different chain lengths in order to maximize their fermentative and prebiotic effects [13]. Beneficial effects of inulin and/or oligofructose on the intestinal microbiota, have been demonstrated in adults, such as increasing Bifidobacteria and Lactobacilli, as well as increasing butyrate production [15–17]. Moreover, the prebiotic activity of inulin and oligofructose seems to translate into beneficial health effects such as improved glycemic control, reduced plasma triglycerides, improved barrier function, and decreased oxidative stress [14]. However, not all fructans seem to produce the same effects on the gut microbiota [18–20]. Ex vivo fermentation models are considered reliable tools to assess how microbial populations are altered by gastrointestinal environments such as prebiotics [21]. The common purpose of ex vivo gut fermentation models is to cultivate a complex intestinal microbiota under controlled environmental conditions with the aim of carrying out microbial modulation and metabolism studies. Although there are several well-studied and validated complex in vitro models of human microbiota (e.g., the Simulator of the Human Intestinal Microbial Ecosystem (SHIME) [22,23], the TNO Gastrointestinal Model (TIM-2) [24] and the Three-Stage Compound Continuous Culture System [25], we were interested in performing a simpler continuous culture system with one single stage. Several studies have classified the human intestinal microbiota into three enterotypes in order to reduce the complexity and facilitate the identification of microbiota compositional patterns [26,27]: enterotype 1, which has Bacteroides as its best indicator; enterotype 2, which is driven by Prevotella, a genus whose abundance is inversely correlated with Bacteroides; and enterotype 3, which is distinguished by a significant presence of Firmicutes especially of the Ruminococcus genus [26–28]. However, this classification should be treated with caution, as some authors have suggested that microbiota could be divided into two enterotypes, where Bacteroides enterotype was fused with the less well distinguished Firmicutes enterotype while other authors have proposed four enterotypes, two of theme resembling previously the described Bacteroides and Prevotella enterotypes, while the other two are a more complex mixture with lower levels of Bacteroides, one dominated by a variety of populations affiliated within the Firmicutes phylum and the other combining higher levels of Faecalibacterium and Ruminococcaceae (both belonging to the Firmicutes phylum) with Alistipes (which belongs to the Bacteroidetes phylum) [28,29]. Nutrients 2019, 11, 1293 3 of 14 The objective of the study was to evaluate how the polymerization degree influences the prebiotic effect of various fructans (FOS and two inulins with differing degrees of polymerization) in an ex vivo fermented human microbiome, under conditions that simulated the transverse colon, taken from the Macfarlane model [25,30]. 2. Materials and Methods 2.1. Fructans Analyzed Three commercial fructans with differing average fructosyl chain lengths (i.e., degree of polymerization) were chosen for this study. 1. Oligofructose (FOS): Orafti® P95 (BENEO-Orafti S.A., Tienen, Belgium), a short-chain inulin obtained through partial enzymatic hydrolysis and consisting of fructose units joined by β(2-1) linkages, with an end-standing glucose unit and an average polymerization degree of two to eight units. 2. Low-polymerization-degree inulin (LPDI): Orafti® GR (BENEO-Orafti S.A., Tienen, Belgium), a granulated powder extracted primarily from chicory and consisting of fructose units joined by β(2-1) linkages, with an end-standing glucose unit and an average polymerization degree of ≥10. 3. High-polymerization-degree inulin (HPDI): Orafti® HPX (BENEO-Orafti S.A., Tienen, Belgium), a high-performance inulin derived from chicory and consisting of fructose units joined by β(2-1) linkages, with an end-standing glucose unit and an average polymerization degree of ≥23. The concentration of each prebiotic in each fermentation vessel was 12 g/L. 2.2. Donor Information Inclusion criteria of volunteers were as follows: healthy young adults aged 18–30 years, a BMI of between 19 to 29.9 kg/m2 , and type-3 or type-4 stools over the previous 72 h, as per the Bristol stool scale [31]. Exclusion criteria were as follows: infectious diseases in the three months prior to fecal donation, antibiotic consumption in the three months prior to donation, blood in the donation, and alcohol intake 72 h prior to donation. Participants were asked to fill out a 72 h dietary recall form and a food frequency questionnaire (Supplementary Table S1). Donors were required to consume a varied diet rich in fruit and vegetables (with a minimum of three portions of vegetables and two portions of fruit per day). All subjects gave their informed consent for inclusion before they participated in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of the University of Barcelona (IRB00003099). 2.3. Fecal Sample Collection and Inoculum Preparation Fecal samples were collected (a minimum of 150 g from each donor) and kept in anaerobic conditions with Oxoid AnaeroGen (Thermo Scientific). The samples were homogenized by dissolution in glycerol/PBS (10% w/v) directly after collection to maximize preservation for the subsequent stages. The samples were frozen at −20 ◦ C and stored in ready-to-use aliquots in the ex vivo fermentation [32,33]. Prior to fermentation, one aliquot from each donor was incubated for 1 h at 37 ◦ C. 2.4. Fermentation Fermentation was carried out in a single-flow in vitro digestion system that simulated the transverse colon conditions defined previously [25,30]. The system consisted of eight independent units (500 mL amber glass jars), in which the fermentation conditions were monitored and controlled. The pH of the culture medium was maintained at 6.2 ± 0.2 by means of a pH meter (model PH 28, Crison Instruments S.A., Barcelona, Spain) connected to two peristaltic minipumps (8 RPM, 100 series pump; Williamson, Southwick, UK) to regulate the addition of HCl 2N or NaOH 3N. Temperature Nutrients 2019, 11, 1293 4 of 14 was monitored with a probe connected to a digital temperature controller (Autonics, Busan, South Korea) and maintained at 37 ◦ C ± 0.5 with a circulating water bath (PolyScience, Niles, IL, USA). Anaerobic fermentation conditions were achieved by constant flushing of O2 -free N2 gas. Liquid retention time was 48 h (2.083%/h dilution rate) by means of two precision peristaltic pumps, one for the culture medium input and one for the fermentation medium output (Masterflex Digital Console Drive; Cole-Parmer, Vernon Hills, IL, USA). All equipment and consumables used in the fermentation process were sterilized before the experiment and sampling started. The artificial colon simulator was fed a complex mixture of nutrients defined by Macfarlane and widely used in the literature [25,34,35]. This mixture was used as our basal medium, either supplemented with the fructans or alone as a control. The potentially interesting fructans were tested after being added to the Macfarlane composition, while the basal medium alone was used as a control treatment. An aliquot of 60 mL of the prepared fecal inoculum was added to each fermenter, together with 340 mL of the basal medium. The experiments were carried out in parallel in eight anaerobic jars (two replicas for each treatment with the basal medium or the basal medium plus fructans). 2.5. Metagenomics Analysis 2.5.1. DNA Extraction Bacterial DNA was extracted from 0.2 g of each sample using the PowerFecal™ DNA isolation kit (MO BIO, Carlsbad, CA, USA), in accordance with manufacturer’s instructions. DNA samples (100 µL) were stored at −20 ◦ C until further processing. To assess for contamination from the laboratory or reagents, blank samples were processed at the same time. Samples obtained from the fermenters were processed for DNA extraction upon collection. 2.5.2. PCR Amplification and Massive Sequencing The V4 region of the 16S rRNA gene was amplified with the widely used primer pair F515 (50 - GTGYCAGCMGCCGCGGTAA-30 ) and R806 (50 -GGACTACNVGGGTWTCTAAT-30 ). Both primers included sequencing adaptors at the 50 end and forward primers were tagged with different barcodes to pool samples in the same sequencing reaction. The PCR mixture (25 µL) contained 2 µL of DNA template (5 ng), 5 µL of 5× Phusion® High-Fidelity Buffer, 2.5 µL of dNTPs (2 mM), 0.2 µM of each primer and 0.5 U of Phusion® Hot Start II Taq Polymerase (Thermo Fisher, Waltham, MA, USA). The PCR thermal profile consisted of an initial denaturation at 98 ◦ C for 30 s, followed by 30 cycles at 98 ◦ C for 15 s, 55 ◦ C for 15 s, 72 ◦ C at 20 s and a final step at 72 ◦ C for 7 min. To assess possible reagent contamination, each PCR reaction included a no template control (NTC) sample. For each amplicon, the quality and quantity were assessed with an Agilent 2100 Bioanalyzer and a QubitTM fluorometer, respectively. Each sequencing pool included 40 barcoded samples that were sequenced on an Ion Torrent Personal Genome Machine (PGM) with the Ion 318 Chip Kit v2 and the Ion PGM™ Sequencing 400 Kit (Life Technologies, Carlsbad, CA, USA), in accordance with manufacturer’s instructions. 2.5.3. Quality Control of the Sequences and OTU Picking Raw sequencing reads were demultiplexed and quality filtered using QIIME 1.9.1 [36]. The reads included presented: a length greater than 225 bp; a mean quality score above 25 in a sliding window of 50 nucleotides; no mismatches on the primer; and default values for other quality parameters. The quality-filtered reads were subsequently processed with VSEARCH v1.1 pipeline [37]. An initial dereplication step was applied, then the reads were clustered into operational taxonomic units (OTUs) at 97% similarity with a de novo approach and, finally, de novo chimera checking was performed with the Uchime function. The raw OTU table was transferred into QIIME 1.9.1 and the taxonomic assignment of representative OTUs was performed using the Ribosomal Database Project (RDP) Classifier [38] against the Greengenes v13.8 database [39]. The sequences were aligned with PyNAST [40]. We sequentially Nutrients 2019, 11, 1293 5 of 14 applied some extra filtering steps in the aligned and taxonomy-assigned OTU table to filter out sequences belonging to chloroplasts and those belonging to the Shewanellaceae and Halomonadaceae families, which were represented in the NTC and considered as contamination due to the reagents. 2.5.4. Data Analysis Data obtained from relative abundance and reads for each sample were organized at the phylum, class, order, family, and genus levels. The alpha-diversity (within a given sample) was calculated at the genus level using two indexes: the exponential Shannon index [41] (i.e., the number of effective OTUs, or Hill number of order 1) and the inverse Simpson index [42] (i.e., the number of highly effective OTUs, or Hill number of order 2). Two-way analysis of variance (ANOVA) was used to assess the effect of treatment and time on alpha diversity, community stabilization (i.e., beta-diversity compared to the previous day) and acidification activity, with the Bonferroni post-test for pairwise analyses at each time point. The ANOVA was performed and the graphs generated with GraphPad Prism 5 (GraphPad Software Inc, San Diego, CA, USA). Beta-diversity (among samples, either from the same or a different treatment) was studied using two types of multivariate analysis at the genus level: non-metric multidimensional scaling (NMDS) for visual grouping, and analysis of similarities (ANOSIM) for statistical significance. More precisely, the significance of the effect of the treatment and time factors was evaluated with two-way ANOSIM, while pairwise differences between treatments on days 7–14 were evaluated with one-way ANOSIM. This procedure was performed with both the Sorensen [43] and Bray–Curtis [44] beta-diversity indexes, using the PAST 3.0 program [45]. 3. Results 3.1. Relative Abundance The samples from the three donors (two females and one male aged between 25 and 27) belonged to the Firmicutes-predominant enterotype according to the MetaHit classifier, and a 1:1:1 mixture was therefore prepared as an inoculum for all ex vivo experiments. As expected, the bacterial communities were dominated by the classes Bacilli, Clostridia, and Erysipelotrichia (all three of which belong to the Firmicutes phylum), Bacteroidia (the Bacteroidetes phylum), Actinobacteria, and Coriobacteriia (the Actinobacteria phylum) and Betaproteobacteria, Deltaproteobacteria, and Gammaproteobacteria (Proteobacteria phylum) (Figure 1); Clostridia and Bacteroidia were the two most abundant classes in the inoculum (Figure 1A). Nutrients 2019, 11, 1293 6 of 14 Nutrients 2019, 11, x FOR PEER REVIEW 6 of 14 Figure Figure1.1.Representation Representationofofrelative relativeabundance abundanceto tothe theclass classlevel level from from the the initial initial time, time, day day 22 (D2), (D2), day day 7 7 (D7), (D7), andand day day 14 14 Assays (D14). (D14). were: Assays were: donors; (A) Sample (A) Sample donors; (B) Controls (B)(C)Controls (Cnt.); (Cnt.); (C) Fructooligosaccharides Fructooligosaccharides (FOS); (D) Low-polymerization-degree (FOS); (D) Low-polymerization-degree inulin (LPDI); and (E) High- inulin (LPDI); and (E) High-polymerization-degree inulin polymerization-degree inulin (HPDI). Each (HPDI). Each assay was performed in duplicate. assay was performed in duplicate. Under Underallallexperimental experimentalconditions, conditions,Gammaproteobacteria Gammaproteobacteriawere wereincreased, increased,Clostridia Clostridiawere were moderately reduced,andand moderately reduced, Betaproteobacteria Betaproteobacteria were reduced. were markedly markedlyCoriobacteria, reduced. Erysipelotrichia, Coriobacteria, Erysipelotrichia, Bacilli, and Deltaproteobacteria Bacilli, and Deltaproteobacteria had minor fluctuations hadoverminor time,fluctuations over time,that except Erysipelotrichia, except were Erysipelotrichia, that with moderately increased wereHPDI. moderately increased Actinobacteria andwith HPDI. continued Bacteroidia Actinobacteria and Bacteroidia to be present throughout continued to be present the fermentation throughout time, but presentedtheconsiderable fermentationfluctuations time, but presented even amongconsiderable replicates.fluctuations The control even among(Macfarlane treatment replicates. medium The control treatment without (Macfarlane prebiotic) resultedmedium without in the most prebiotic) consistent growth resulted in the of Bacteroidia, most whileconsistent growth of supplementation Bacteroidia, with whileinsupplementation HPDI resulted with HPDI the most stable abundance ofresulted in the Clostridia, most albeit stable somewhat abundance of Clostridia, reduced compared to thealbeit somewhat donors’ reduced compared original microbiotas to the donors’ original microbiotas (Figure 1). (Figure 1). 3.2. Alpha-Diversity 3.2. Alpha-Diversity Two indexes to determine diversity were calculated: the exponential Shannon index and the inverse TwoSimpson indexes index, which correspond to determine to Hillcalculated: diversity were numbers 1theandexponential 2, respectively. Broadly Shannon similar index andtrends the were observed inverse Simpsoninindex, both indexes (Figure 2). Diversity which correspond was markedly to Hill numbers 1 and 2,reduced on dayBroadly respectively. 2, but tended similarto stabilization trends thereafterinfor were observed most both treatments. indexes (FigureAt2). the end of the Diversity wasexperiment markedly (day 14), HPDI reduced on dayachieved 2, but greater recovery of alpha-diversity, given that it almost reached the initial levels as per tended to stabilization thereafter for most treatments. At the end of the experiment (day 14), HPDIthe inverse achieved greater recovery of alpha-diversity, given that it almost reached the initial levels as per the inverse Simpson index. LPDI and FOS had lower alpha-diversity, and the control treatment (no Nutrients 2019, Nutrients 11, x1293 2019, 11, FOR PEER REVIEW 7 7of of 14 14 prebiotic) resulted in the lowest alpha-diversity on day 14. A two-way ANOVA analysis with Simpson index. Bonferroni LPDIindicated correction and FOSthathadthe lower onlyalpha-diversity, and the significant pairwise controlwas difference treatment (no prebiotic) that of HPDI against the control on day 14 (p < 0.05), while all other pairwise comparisons between groups and Bonferroni resulted in the lowest alpha-diversity on day 14. A two-way ANOVA analysis with days were correction indicated that the only significant pairwise difference was that of HPDI against the control on nonsignificant. day 14 (p < 0.05), while all other pairwise comparisons between groups and days were nonsignificant. A B 15 8 Control FOS 6 LPDI 10 HPDI 4 5 2 D0 D2 D7 D14 D0 D2 D7 D14 Figure 2. Alpha-diversity at the genus level as per (A) the exponential Shannon index (Hill number 1); Figure and (B)2.the Alpha-diversity at the inverse Simpson genus index (Hilllevel as per number 2). (A) the exponential Shannon index (Hill number 1); and (B) the inverse Simpson index (Hill number 2). 3.3. Stabilization of the System 3.3. Stabilization of the System Microbiota changes at the genus level over time and within each treatment were studied with the Sorensen and Bray–Curtis Microbiota changes atbeta-diversity the genus level indexes. over timeIn general, and withinsimilarities with respect each treatment weretostudied the previous with timeSorensen the point increased over time,beta-diversity and Bray–Curtis thus indicating a moreInstable indexes. composition general, (Figure similarities with 3). The lowest respect to the similaritytime previous was point observed between increased overdays 0 and time, thus2, indicating thereby indicating a marked a more stable compositional composition departure (Figure 3). The from the lowest inoculum. similarity wasGreater observedstability between wasdaysobserved 0 andfor 2, the Sorensen thereby index athan indicating for the marked Bray–Curtis compositional index, especially departure from the oninoculum. days 2 and 7, thusstability Greater indicatingwasthat the more observed for abundant the Sorensen genera played index than aformore the significant role Bray–Curtis in the index, differences especially on dayswithin 2 and each treatment 7, thus indicatingoverthat time,theasmore the Bray–Curtis abundant generaindexplayed gives agreater weight to these more significant role genera compared to in the differences the Sorensen within index. The each treatment oversimilarity time, as between days 7 and the Bray–Curtis 14 index was the gives highest greater for all weight totreatments, these genera which wouldtosuggest compared that the the Sorensen microbial index. community The similarity composition between days 7 stabilized, and 14 wasyetthe a plateau highestwas for not observed within all treatments, which thewould experimental suggest timeframe. A two-way that the microbial ANOVA community analysis withstabilized, composition Bonferroniyet correction a plateau indicated was not that treatment observed with FOS within resulted in thetimeframe. the experimental lowest stability A two-on day 2ANOVA way (i.e., the analysis largest departure from thecorrection with Bonferroni inoculum), as per the indicated Sorensen that treatmentindex, withcompared to allin FOS resulted other the treatments lowest (Bonferroni stability < 0.05 on day 2p (i.e., theforlargest the comparison departure fromvs control, HPDI andasLPDI). the inoculum), per theConversely, HPDI Sorensen index, had a lower compared to stability than all other all other treatments treatments (Bonferroni p