Fermentation substrates

Starch-entrapped microspheres were made by dissolving 0·2 kg of sodium alginate in 8·8 kg water. One kilogram of waxy maize starch was then added, and the suspension was mixed until homogenous. The slurry was then pumped through an atomiser into 20 l of stirred 4 % (w/v) calcium chloride dihydrate. The microspheres were then harvested by filtration. After extensive washing, the microspheres were mixed with ethanol and filtered again. The ethanol was necessary to prevent the microspheres from sticking together. The microspheres were allowed to dry in air until the odour of ethanol could not be detected (about 4 h) and then moved to a forced draft oven at 45°C to finish drying (about 24 h).

Total starch was determined using a colorimetric method (Total Starch (α-amylase/amyloglucosidase method), Megazyme, Wicklow, Ireland) after disruption of the alginate matrix with a ball mill (Brinkman, Westbury, NY, USA). Moisture content was measured by loss in weight after drying in a forced draft oven at 105°C for 18 h. Short-chain FOS was obtained from Encore Technologies (Plymouth, MN, USA). As per the manufacturer's spec sheet, the product contained 94 % FOS, 0–4 % glucose and 4–8 % sucrose.

Faeces

Fresh stools were collected from six patients with inactive CD and six patients with active UC, who had not received antibiotics within the last 4 weeks. CD and UC patients were selected from the IBD Clinic at Rush University Medical Center (Chicago, IL, USA). The diagnosis of UC and CD diseases was made based on established clinical, endoscopic and histological criteria. Patients with inactive CD had a CD activity index of < 150, and had no diarrhoea, abdominal pain or bleeding. Patients with active UC had rectal bleeding, diarrhoea (liquid stool and stool frequency >3/d) and friable and ulcerated rectosigmoid mucosa on their sigmoidoscopy at the time of stool collection. All the patients were taking daily 5-aminosalicylic acid, but none were taking prednisone at the time of stool collection. The healthy controls consisted of faculty and staff of the university who were not involved in the study. Faeces were collected in gas-tight bags containing Anaerocult C (to maintain anaerobosis; Merck, Darmstadt, Germany), frozen in liquid nitrogen and then stored at − 80°C until used for fermentation experiments (approximately 1 month).

To validate our use of frozen stool in fermentation experiments, we analysed a fresh microbiota and the same microbiota stored at − 80°C. Microbial composition of the fresh and frozen microbiotas was checked periodically over 44 weeks of storage using denaturing gradient electrophoresis as described previously⁽Reference Venema, Van Nuenen and Smeets-Peeters¹³⁾. Activity of the fresh and frozen microbiotas (8 d, − 80°C) was assessed by measuring SCFA production during in vitro fermentation as described below, except the total fermentation time was 196 h, and the microbiota were maintained on the standard ileal efflux medium during the entire experiment rather than switching to the test carbohydrates after 16 h.

Fermentation

Faeces were defrosted in an anaerobic cabinet, and samples from six individuals in each of the three categories were combined to yield a microbiota representative of that group. The amounts used from each individual were not equal due to differing amounts of stool collected (Table 1). Pooling was important to reduce the possibility of anomalous results⁽Reference Christl, Katzenmaier and Hylla¹⁴⁾ due to the highly variable nature of gut microbiota among individuals⁽Reference McBurney and Thompson¹⁵⁾. The faeces were homogenised with nine volumes of diluted standard ileal efflux medium (modified from⁽Reference Gibson, Cummings and Macfarlane¹⁶⁾; Table 2). Ninety millilitres of this homogenate were used to inoculate the Toegepast Natuurwetenschappelijk Onderzoek in vitro model of the large intestine: the Toegepast Natuurwetenschappelijk Onderzoek gastro-intestinal model-2 (TIM-2) system (TNO, Zeist, The Netherlands)⁽Reference Minekus, Smeets-Peeters and Bernalier¹⁷⁾. This system has been validated against sudden death individuals⁽Reference Venema, Van Nuenen and Smeets-Peeters¹³⁾ and recently investigated using molecular RNA technology and assessed as closely simulating the microbiota in the colon⁽Reference Kovatcheva-Datchary, Egert and Maathuis¹⁸⁾.

Table 1 Amount of faecal material used from each donor

CD, Crohn's disease; UC, ulcerative colitis; ID, identification.

* Wet weight.

Table 2 Composition of standard ileal efflux medium (SIEM) and diluted SIEM per litre

The TIM-2 system was allowed to equilibrate for 16 h after inoculation, during which time the microbiota (representing healthy, inactive CD or active UC) were fed a standard ileal efflux medium (Table 2) at a rate of 2 ml/h (containing 5 g of carbohydrate per 24 h). After equilibration, the feeding line was shut off, and the microbiotas were starved for 2 h. The test period (t = 0 h) was then begun by introducing the test carbohydrates (starch-entrapped microspheres or FOS) into the TIM-2 units. Carbohydrates were administered with a standard ileal efflux medium prepared without the carbohydrates listed in Table 2. FOS was dissolved in the medium, and starch-entrapped microspheres, due to their insoluble nature, were administered through a second syringe that was kept dry. Test carbohydrates were given at a rate of 10 g of fermentable carbohydrate per 24 h for 72 h.

The bacteria were constantly supplied with carbohydrate; therefore, this model more closely resembles the proximal rather than the distal colon. To approximate the distal colon more closely, during the last 4 h of fermentation (t = 68–72 h), the feeding line was shut off, and the pH was monitored to indicate whether the saccharolytic fermentation continued, or if the bacteria switched to putrefactive fermentation due to lack of carbohydrate. Samples were collected every 24 h for SCFA quantification, and microbial analyses were performed at the initial (t = 0 h) and final (t = 72 h) times. Fig. 1 shows a timeline of these events.

Fig. 1 Timeline for Toegepast Natuurwetenschappelijk Onderzoek gastro-intestinal model-2 (TIM-2) experiments; numbers represent hours; SIEM, standard ileal efflux medium; SM, starch-entrapped microspheres; FOS, fructo-oligosaccharides.

SCFA determination

For SCFA, lumen or dialysate samples were centrifuged (12 000 rpm, 5 min). A mixture of formic acid (20 %), methanol and 2-ethyl butyric acid (internal standard, 2 mg/ml in methanol) was added to the supernatant. A 0·5 μl sample was injected onto a GC column (Stabilwax-DA, length 15 m, inner diametre 0·53 mm, film thickness 0·1 mm; Varian Chrompack, Bergen op Zoom, The Netherlands) in a Chrompack CP9001 GC using an automatic sampler (Chrompack liquid sampler CP9050; Varian Chrompack)⁽Reference Jouany¹⁹⁾.

Microbial analysis

Analysis of the microbiota with the intestinal-chip was performed at TNO (Zeist, The Netherlands), described by Schuren et al. ⁽Reference Maathuis, Hoffman and Evans²⁰⁾. DNA was isolated from luminal samples at the start (t = 0 h) and at the end of the TIM-2 experiments (t = 72 h). Signals were quantified by calculating the mean of all pixel values of each spot and calculating the local background about each spot. Signal intensity was dependent on the position of the probe within the 16S rRNA gene; for that reason, concentrations of individual bacterial classes could not be assessed without setting up standard curves using quantitative PCR for each probe. However, signal intensities could be compared within probe to determine relative changes in abundance of the bacterial class represented by that particular probe⁽Reference Maathuis, Hoffman and Evans²⁰⁾. Because the primary aim was to determine the propensity of each test carbohydrate to alter individual classes of bacteria, fold differences between the signals at the initial and final times were useful for this determination. Standard deviations were relatively large among runs; therefore, each replicate was analysed twice.

Data analysis

Data were analysed with SAS software (version 9.1; SAS Institute, Cary, NC, USA) using a mixed model ANOVA (PROC MIXED). Main effects were microbiota source (1, healthy; 2, inactive CD; 3, active UC) and carbohydrate source (1, starch-entrapped microspheres; 2, FOS). For those analyses that included a time element (i.e., comparison between the initial and final times), a time effect (0 and 72 h) was also included. Fisher's least significant different test was used to determine significant differences among treatment combinations. All the treatment combinations were performed in quadruplicate, except when the active UC microbiota were fermented with FOS, which were performed in triplicate.

Ethical approval

The present study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subject/patients were approved by the Institutional Review Board of Rush University Medical Center. Written informed consent was obtained from all the subjects/patients.

Use of frozen microbiota

Fig. 2 shows that after 44 weeks of storage at − 80°C, the composition of the microbiota in terms of viable cells was not substantially affected, as evidenced by similar banding patterns on denaturing gradient electrophoresis gels. Furthermore, the stored, frozen samples still contained the microbial diversity of the fresh faecal sample, indicating that freezing had minimal influence on the microbiota composition.

Fig. 2 Denaturing gradient gel electrophoresis profiles of a fresh faecal sample (0) and the same faecal sample after frozen storage ( − 80°C) for 4, 20, 24, 28 and 44 weeks. M, marker.

The production of SCFA during 196 h of fermentation was also measured using a fresh microbiota and the same microbiota that had been frozen at − 80°C for 8 d. On comparison, the fresh and frozen microbiotas showed similar SCFA production over at least 150 h of fermentation (Fig. 3). Most notably, during the first 72 h, which is the time period used in the present study, SCFA production was nearly identical. Moreover, the ratios of SCFA produced during fermentation (acetate:propionate:butyrate) were not significantly different for fresh and frozen microbiotas (data not shown).

Fig. 3 Production of SCFA (acetate+propionate+butyrate) from a fresh microbiota (○) and the same microbiota after frozen storage (8 d, − 80°C; ●). Error bars show standard deviation; some error bars are too small to see; n>3.

Previously, others have reported using frozen microbiotas from animals⁽Reference Stanco, Di Meo and Piccolo^21, Reference Hervas, Frutos and Giraldez²²⁾. For example, Hervas et al. ⁽Reference Hervas, Frutos and Giraldez²²⁾ discovered that storage of rumen fluid at − 18°C for 24 h did not affect fermentation parameters when starch and cellulose were fermented; although average fermentation rate and the extent of polysaccharide degradation were affected, when barley straw neutral-detergent fibre and alfalfa hay were fermented. Stanco et al. ⁽Reference Stanco, Di Meo and Piccolo²¹⁾ showed that it was possible to use frozen inoculum from the caecum of rabbits (stored at − 18°C) for at least 3 months without affecting gas production, degradation of organic matter and SCFA production by the microbiota during in vitro fermentation. In a review aimed at examining factors that influence microbial activity of the rumen microbiota, Mould et al. ⁽Reference Mould, Klien and Morgan²³⁾ concluded that during storage and sample transport, exposure to air should be excluded. For this reason, we maintained anaerobosis during storage and transport using Anaerocult C. These reports, together with the data provided in Figs. 2 and 3, provided sound evidence that the use of a frozen microbiota provides results that are comparable to a fresh microbiota for in vitro fermentation studies.

SCFA production

SCFA may be of particular importance to colonic health. They reduce colonic pH, thereby inhibiting the growth of some opportunistic, pathogenic bacteria⁽Reference Cummings, Christie and Cole⁵⁾, decreasing the activity of co-carcinogenic enzymes such as glucuronidases, glycosidases and 7α-hydroxylases⁽Reference Goldin and Gorbach^24, Reference Thornton²⁵⁾. Butyrate, one of the SCFA produced during fermentation, may be especially important in IBD. In vitro studies have revealed that butyrate suppresses the immuno-inflammatory response by inhibiting activation of the transcription factor NF-κB⁽Reference Inan, Rasoulpour and Yin²⁶⁾ and its downstream products like pro-inflammatory cytokines TNF and IL-12 as well as up-regulating the production of the anti-inflammatory cytokine IL-10⁽Reference Saemann, Bohmig and Osterreicher⁸⁾. Patients treated with butyrate enemas have shown decreased inflammation due to a reduction in macrophages positive for activated NF-κB⁽Reference Luhrs, Gerke and Muller²⁷⁾.

Patients with IBD have not only shown decreased SCFA production⁽Reference Packey and Sartor²⁸⁾, they may also have impaired ability to harvest their beneficial effects. Higher concentrations of sulphate-reducing bacteria have been reported in patients with IBD compared with controls⁽Reference Loubinoux, Bronowicki and Pereira²⁹⁾. An increase in sulphate-reducing bacteria increases hydrogen sulphide production, which blocks colonocyte oxidation of butyrate⁽Reference Roediger, Duncan and Kapaniris³⁰⁾. Therefore, dietary fibres that can increase butyrate production by enhancing growth of butyrate producing bacteria, or fibres that can improve butyrate utilisation by inhibiting the growth of sulphate-reducing bacteria may be helpful in alleviating colonic inflammation.

When comparing among faecal microbiotas grown on starch-entrapped microspheres, the healthy microbiota produced more butyrate than microbiotas from IBD stools (Table 3), demonstrating impaired butyrate production by faecal microbiota in IBD patients. This was most pronounced in the inactive CD microbiota. In faecal microbiota grown on FOS, only the microbiota from patients with inactive CD showed significant inhibition. Of note is that starch-entrapped microspheres resulted in significantly higher butyrate production than FOS when fermented with the healthy and active UC faecal microbiotas.

Table 3 Butyrate and total SCFA production from healthy, inactive Crohn's disease (CD) and active ulcerative colitis (UC) microbiotas when administered starch-entrapped microspheres (SM) or fructo-oligosaccharides (FOS); values reported as mmol produced per amount of carbohydrate administered

^a,b,c,d Mean values within a column with unlike superscript letters were significantly different (P < 0·05).

Total SCFA production was similar when comparing between carbohydrate sources, except when the faecal microbiota from active UC were used (Table 3). In this case, starch-entrapped microspheres produced significantly more SCFA. Moreover, starch-entrapped microspheres were able to induce a total SCFA production in this diseased microbiota that were similar to the healthy faecal microbiota. The same was not true for FOS.

During the adaptation period (t = − 16–0 h), the faecal microbiota were fed a standard ileal efflux medium, which contained a mixture of carbohydrates that approximates the range of substrates that colonic bacteria may be exposed to under normal conditions⁽Reference Minekus, Smeets-Peeters and Bernalier¹⁷⁾. Therefore, the molar ratios of SCFA produced during this time period could be compared to the ratios at the end of the test period (t = 72 h) to determine whether there was a shift in the molar ratios of SCFA produced. In particular, an increase in the molar proportion of butyrate produced would suggest an increase in butyrate producing bacteria, which may be beneficial due to the anti-inflammatory nature of butyrate.

In the healthy faecal microbiota, a significant increase in the proportion of butyrate produced was only experienced when the faecal microbiota were exposed to starch-entrapped microspheres (Fig. 4; with a concomitant decrease in acetate production, data not shown). The microbiota from patients with inactive CD showed a shift toward butyrate production with both starch-entrapped microspheres and FOS (with concomitant decrease in acetate production, data not shown), while that of active UC showed no significant shift toward butyrate production.

Fig. 4 Molar percentage of butyrate (of total SCFA) produced from healthy (■), inactive Crohn's disease () and active ulcerative colitis (□) microbiotas when administered a standard ileal efflux medium (SIEM, Table 2) during the adaptation period (t = − 16–0 h, Fig. 1) compared with starch-entrapped microspheres (SM) or fructo-oligosaccharides (FOS) during the test period (t = 0–72 h, Fig. 1); bars marked with the different capital letters are significantly different (P < 0·05).

Thus, in light of the trophic effects of SCFA, and in particular butyrate, on the colon, our data suggest that starch-entrapped microspheres may be more beneficial than FOS to promote gut health in healthy subjects and to promote anti-inflammatory milieu in the colon of patients with active UC.

Fermentation rate and pH changes

As mentioned, the TIM-2 system more closely approximates the proximal rather than the distal colon because the microbiotas are constantly supplied with test carbohydrates during fermentation. To approximate conditions that may occur in more distal regions of the colon, the feeding line of the TIM-2 system was shut off during the last 4 h of fermentation. Fig. 5 shows the changes in pH during the last 4 h of fermentation when starch-entrapped microspheres and FOS were inoculated with the faecal microbiota from healthy individuals or patients with IBD. SCFA from the saccharolytic fermentation cause a decrease in pH that is detected by the automatic titrator that prevents the pH from dropping below 5·8 by addition of sodium hydroxide. Therefore, a jagged line indicates the saccharolytic fermentation, and the more jagged the line (crests closer together), the more rapid the fermentation rate.

Fig. 5 Changes in pH during the last 4 h of fermentation (when the feeding line had been shut off) of faecal microbiotas from healthy individuals (a), patients with inactive Crohn's disease (b) or active ulcerative colitis (c) that had been previously fed starch-entrapped microspheres (—) or fructo-oligosaccharides ().

During the first hour after the feeding line was shut off (68–69 h), all the samples showed saccharolytic fermentation. The starch-entrapped microspheres clearly showed a less jagged line than FOS, indicating slower fermentation. The faecal microbiota from healthy subjects showed continued the saccharolytic fermentation after the feeding line was shut off (t = 69–72 h) when grown on starch-entrapped microspheres. Conversely, FOS showed a gradual increase in pH during this time period, which is indicative of a shift toward putrefactive fermentation, which results in NH₃ production. This supports our previous study showing slow digestion and fermentation of the starch-entrapped microspheres⁽Reference Rose, Keshavarzian and Patterson¹²⁾ and suggests this material could provide beneficial fermentable carbohydrates in the distal colon, whereas FOS would not.

As with the healthy microbiota, the microbiota from IBD patients showed continuous fermentation when fermented with starch-entrapped microspheres (Fig. 5). The large peaks for the active UC microbiota when fed starch-entrapped microspheres were due to inadequate mixing. For an unknown reason, when starch-entrapped microspheres were fermented with the active UC microbiota, the slurry became very thick and the mixing laboured. Although this situation was not ideal, this curve still did not show a gradual upward trend (as seen with FOS), indicating the occurrence of saccharolytic fermentation with the starch-entrapped microspheres.

In contrast to starch-entrapped microspheres, the microbiota from patients with IBD when incubated with FOS, showed more pronounced increases in pH during the final 4 h of fermentation compared with the stool from healthy individuals. This suggests that slowly fermentable dietary fibres, such as starch-entrapped microspheres, may be of particular importance in IBD due to their ability to maintain a low pH during prolonged fermentation.

Fermentation rate is important for the maintenance of gut health. As carbohydrates are fermented and their amount diminishes, putrefactive fermentation becomes significant, leading to an increase in pH due to lack of SCFA and production of NH₃ ⁽Reference Macfarlane, Gibson and Cummings¹⁰⁾. The higher pH may allow the proliferation of undesirable bacteria⁽Reference Cummings, Christie and Cole⁵⁾. The distal colon also contains higher concentrations of toxic metabolites⁽Reference Macfarlane, Gibson and Cummings¹⁰⁾ due to fermentation of protein instead of carbohydrate. These undesirable conditions in the distal colon may exacerbate inflammation and explain the high rate of colonic neoplasia in left side of the colon. Thus, improving colonic milieu in the distal colon by delivering slow fermentable carbohydrate could be a beneficial intervention for promotion of gut health and prevention of colonic diseases like colonic polyps and cancer in healthy subjects as well as controlling colonic inflammation in inflammatory disorders of colon like UC. Our data provide compelling evidence that starch-entrapped microspheres are the ideal substrate to achieve this goal, although further in vivo, long-term studies in healthy subjects and patients with IBD are required to confirm these gut-health promoting effects of starch-entrapped microspheres.

Initial composition of the microbiotas from healthy individuals and patients with inactive Crohn's disease and active ulcerative colitis

Table 4 shows the initial (t = 0 h) fold differences among the diseased and healthy faecal microbiotas of some of the major genera of bacteria found in the large intestine⁽Reference Macfarlane, McBain, Gibson and Roberfroid^31, Reference Conway, Mackie, White and Isaacson³²⁾. Therefore, changes in these classes of bacteria would likely represent a change in the gut metabolism as a whole. The microbiota from patients with inactive CD displayed far weaker signals for Bacteroides, Clostridium and Eubacterium compared with the healthy microbiota. In contrast, the faecal microbiota from patients with active UC displayed only slightly different signals for these three genera of bacteria compared with the faecal microbiota from healthy individuals. This suggests that microbiota from patients with active UC more closely resembled the healthy microbiota than did the inactive CD microbiota, as these three genera of bacteria represent the major proportion of the gut microbiota as a whole⁽Reference Vernazza, Rabiu, Gibson, Gibson and Rastall³³⁾. Additionally, these are the most important butyrate producers in the gut. The fact that these genera were so much higher in the healthy and active UC microbiota compared with inactive CD at least partially explains why butyrate production was so low in the inactive CD microbiota (Table 3).

Table 4 Fold differences in signal for select bacterial classes in the initial microbiota from patients with inactive Crohn's disease (CD) or active ulcerative colitis (UC) compared with the healthy microbiota; values reported as fold difference from healthy faecal microbiota, with negative numbers indicating a lower signal in the diseased stool and a positive number signifying a higher signal in the diseased stool

Mean values were significantly different from those of healthy microbiota: *P < 0·05, **P < 0·01, ***P < 0·001.

For the probiotic genera Lactobacillus and Bifidobacterium, no substantial differences were found in the concentrations of Lactobacillus among microbiotas, but the healthy microbiota had a far greater quantity of Bifidobacterium than the microbiota from active UC and inactive CD patients. These genera of bacteria are reported to be beneficial to gut health⁽Reference Gibson and Roberfroid⁴⁾; and their lack may be related to the diseased condition.

When comparing the general bacterial signals, the microbiota from patients with inactive CD showed a 2-fold lower concentration than that of healthy individuals. However, signals from a number of the genus probes shown in Table 4 were substantially lower for the inactive CD microbiota compared with the healthy stool. In particular, the major genera, Bacteriodes, Clostridium and Eubacterium, were far more than 2-fold lower. This suggests that many of the bacteria in the inactive CD microbiota belonged to genera not represented in Table 4. This supports the findings of Seksik et al. ⁽Reference Seksik, Rigottier-Gois and Gramet³⁴⁾, who demonstrated that a high percentage (30 %) of the dominant microbiota in CD belongs to unusual phylogenetic groups.

Changes in the faecal microbiota of healthy subjects during fermentation

After 72 h of fermentation, FOS induced a major increase in the concentration of Eubacterium in faecal microbiota from healthy subjects, whereas starch-entrapped microspheres did not (Table 5). While Eubacteria are important butyrate producers⁽Reference Duncan, Belenguer and Holtrop³⁵⁾, the significantly higher butyrate production from the healthy microbiota when administered starch-entrapped microspheres compared with FOS (Table 3) suggests that starch-entrapped microspheres must have increased the growth of other butyrate producing bacteria more than FOS increased the growth of Eubacterium.

Table 5 Fold changes in the signals of selected bacterial classes in the healthy, inactive Crohn's disease (CD) and active ulcerative colitis (UC) microbiotas after 72 h of fermentation with starch-entrapped microspheres (SM) or fructo-oligosaccharides (FOS); values reported as fold difference from the initial respective faecal microbiota, with negative numbers indicating a lower signal after fermentation, and positive numbers signifying a higher signal after fermentation

Mean values were significantly different from those of initial microbiota: *P < 0·05, **P < 0·01, ***P < 0·001.

Concentrations of Streptococci were decreased with starch-entrapped microspheres, but left largely unchanged with FOS (Table 5). Prior studies reported that Streptococcus species exhibit mixed effects on the colonic environment. While some species, such as Streptococcus agalactiae and Streptococcus bovis, display negative effects on colonic melieu⁽Reference Reynolds, Silva and McCormack^36–Reference Santos-Sierra, Golenbock and Henneke³⁸⁾, other species such as S. salivarius, show positive characteristics⁽Reference Sartor^39, Reference de Palencia, Lopez and Corbi⁴⁰⁾. In the present study, of these three species, S. bovis decreased slightly and S. salivarius decreased substantially after fermentation with starch-entrapped microspheres. This suggests that FOS has greater benefit than starch-entrapped microspheres with respect to Streptococcus growth in the healthy faecal microbiota.

Clostridium concentrations were substantially affected by the test carbohydrates to about the same degree for the two treatments (Table 5). Clostridium is sometimes recognised as a harmful or undesirable genus of the gut microbiota⁽Reference Gibson and Roberfroid⁴⁾. For example, Clostridium difficile produces a toxin that causes colitis-like inflammation⁽Reference Freeman⁴¹⁾, and infection with C. difficile exacerbates inflammation in patients with IBD⁽Reference Sartor, Rath and Sellon⁴²⁾. Unfortunately, we did not have a probe for C. difficile and were unable to determine how the test carbohydrates affected this bacterium. Two Clostridia that we did have probes, for Clostridium perfringens and Clostridium butyricum, were decreased by starch-entrapped microspheres and FOS by roughly the same degree. C. perfringens may have detrimental effects, suggested by significantly more C. perfringens found in faecal samples from patients with IBD compared with controls⁽Reference Mangin, Bonnet and Seksik⁴³⁾. In contrast, C. butyricum may be beneficial through its high SCFA (including butyrate) production and ability to alleviate dextran sodium sulphate-induced colitis in rats⁽Reference Araki, Andoh and Takizawa⁴⁴⁾. In fact, many clostridia are major contributors to the colonic butyrate pool⁽Reference Packey and Sartor²⁸⁾, which indicate that they may have an anti-inflammatory effect.

Escherichia concentrations were decreased during fermentation on starch-entrapped microspheres and FOS (Table 5). Escherichia, like clostridia, have mixed effects on the colonic environment, with some species exhibiting benefits and others exhibiting detrimental effects⁽Reference Gibson and Roberfroid⁴⁾. While a virulent strain of Escherichia coli (adherent-invasive E. coli) has been isolated from the ileal mucosa of rats with CD⁽Reference Darfeuille-Michaud, Boudeau and Bulois⁴⁵⁾, and high concentrations of E. coli have been associated with colitis in rats⁽Reference Onderdonk, Richardson and Hammer⁴⁶⁾, other strains of E. coli (e.g. Nissle 1917) are established probiotics⁽Reference Schultz⁴⁷⁾.

In the healthy microbiota, the concentrations of Bifidobacterium and Lactobacillus were unaffected during fermentation, independent of carbohydrate source (Table 5). It is surprising that there was not an increase in Bifidobacterium on FOS, as FOS is known to be bifidogenic⁽Reference De, Vanhoutte and Huys⁴⁸⁾. A likely explanation is that the concentration of Bifidobacterium in the faecal microbiota community in healthy subjects, as suggested by Table 4, was initially quite high. If such was the case, previous research has demonstrated that it is almost impossible to then increase Bifidobacterium even further⁽Reference Roberfroid, Van Loo and Gibson⁴⁹⁾.

Changes in the microbiota from patients with inactive Crohn's disease during fermentation

The faecal microbiota from patients with inactive CD showed a substantial increase in Bacteroides concentration during fermentation (Table 5). Because faecal microbiota from patients with inactive CD had an extremely low concentration of Bacteroides at the initial time compared with the healthy microbiota (Table 4), an increase would represent a shift toward the healthy microbiota; thus suggesting a beneficial shift. The fact that FOS was able to increase Bacteroides more than starch-entrapped microspheres suggests that FOS was more desirable than starch-entrapped microspheres in this regard; however, some species of Bacteroides are undesirable. For instance, Bacteroides fragilis is found in much larger numbers in the mucosal biofilm in patients with CD compared with healthy individuals, and thus may play a role in the pathogenesis of this disease⁽Reference Swidsinski, Weber and Loening-Baucke⁵⁰⁾. Also, Bacteroides vulgatus expresses a membrane protein that induces inflammation⁽Reference Bamba, Matsuda and Endo⁵¹⁾. There was an increase in B. fragilis during fermentation with FOS, whereas no change was observed with starch-entrapped microspheres (Table 5). B. vulgatus were increased with both carbohydrate treatments, but only FOS showed a significant increase. Therefore, starch-entrapped microspheres may be superior to FOS at preventing the growth of potentially harmful Bacteroides species.

Eubacterium and Veillonella concentrations were also affected differently by starch-entrapped microspheres compared with FOS. Eubacterium concentration was increased during fermentation with FOS, while no change was observed with starch-entrapped microspheres (Table 5). As mentioned, the major benefit of Eubacterium is high butyrate production⁽Reference Duncan, Belenguer and Holtrop³⁵⁾, yet the production of this metabolite was not significantly different among starch-entrapped microspheres and FOS when fermented with the microbiota from inactive CD patients (Table 3). This indicates that the increase in the concentration of Eubacterium seen with FOS must not have exceeded the increase in other butyrate producers induced by starch-entrapped microspheres. Veillonella was increased with FOS, whereas starch-entrapped microspheres had little effect. It is noteworthy that Veillonella is a harmful gut genus⁽Reference Gibson and Roberfroid⁴⁾, because it is not saccharolytic and produces NH₃ and hydrogen sulphide⁽Reference Rogosa⁵²⁾. Our finding suggests that starch-entrapped microspheres may be preferred over FOS by limiting growth of Veillonella.

Starch-entrapped microspheres and FOS induced similar effects on Enterococcus, Escherichia and Bifidobacterium concentrations (Table 5). Enterococcus signals were not increased with either carbohydrate. This is likely beneficial because high concentrations of Enterococcus have been associated with colitis in rats⁽Reference Onderdonk, Richardson and Hammer⁴⁶⁾ and Enterococcus facialis leads to distal colitis in mice⁽Reference Kim, Tonkonogy and Albright⁵³⁾. Escherichia concentrations were decreased with starch-entrapped microspheres and FOS. The mixed effects that this may have on the gut microbiota have been discussed earlier. Shifts in Bifidobacterium were unexpected. Significant decreases in this genus were observed when the microbiota from patients with inactive CD was exposed to starch-entrapped microspheres or FOS. One possible explanation is that stool microbiota community in patients with CD had an initially low concentration of Bifidobacterium; therefore, it is likely that these bacteria were unable to compete with the other bacteria for substrate, and thus were unable to proliferate. Alternatively, this decrease may simply be due to differences among healthy and diseased microbiota, i.e., a carbohydrate that is beneficial for one type of microbiota, may not be beneficial for another.

Changes in the microbiota from patients with active ulcerative colitis during fermentation

Starch-entrapped microspheres and FOS both resulted in similar decreases in Bacteroides signals during fermentation (Table 5). Unlike the microbiota from inactive CD, the microbiota from patients with active UC showed a much higher concentration of Bacteroides than the faecal microbiota in healthy subjects before fermentation; therefore, a decrease in Bacteroides after fermentation may be beneficial. Of the two detrimental Bacteroides that we had probes for, B. fragilis and B. vulgatus, FOS induced far greater decreases than starch-entrapped microspheres. This suggests that FOS may result in more favourable changes in Bacteroides than starch-entrapped microspheres in active UC due to more substantial decreases in these potentially harmful species belonging to this genus.

Clostridium concentrations were affected largely to the same extent regardless of carbohydrate source. Within the clostridial genera, the beneficial bacterium, C. butyricum, and the detrimental bacterium, C. perfringens, were also substantially decreased by both the carbohydrates (Table 5).

The effects of starch-entrapped microspheres and FOS were different with respect to their influence on Enterococcus, Fusobacterium, Streptococcus and Veillonella. Both the carbohydrates decreased Enterococcus concentrations, but FOS only showed a 2·5-fold decrease, while starch-entrapped microspheres showed a 4·5-fold decrease (Table 5). In light of the possible detrimental effects of Enterococcus on gut health⁽Reference Onderdonk, Richardson and Hammer^46, Reference Kim, Tonkonogy and Albright⁵³⁾, these data would suggest a greater benefit of starch-entrapped microspheres compared with FOS. Fusobacterium is another genus of bacteria that is likely detrimental to gut health; Fusobacterium varium has been suggested as a causative agent for UC⁽Reference Ohkusa, Okayasu and Ogihara⁵⁴⁾. Even though we did not have a probe for this specific bacterium, a decrease in its genera may be beneficial. Starch-entrapped microspheres decreased the concentration of this genus, while FOS did not significantly affect Fusobacterium concentrations. Streptococcus increased with FOS and exhibited no substantial change with starch-entrapped microspheres; however, no change in the signals from individual Streptococcus species probes was observed. Veillonella was unchanged by starch-entrapped microspheres, but exhibited an increase with FOS. These data indicate that starch-entrapped microspheres may be better than FOS at preventing the proliferation of this purported harmful bacterial genus⁽Reference Gibson and Roberfroid⁴⁾.

There were unexpected changes in Bifidobacterium. Starch-entrapped microspheres did not substantially affect the Bifidobacterium concentration, but FOS decreased its concentration (Table 5). This supports the conclusions from the inactive CD microbiota that Bifidobacterium were unable to compete with other bacteria for substrate, or that a carbohydrate that is beneficial in one case may not be beneficial in all instances.