All chemicals: acetic acid (99.9% of purity), propionic acid (99.6%), butyric acid (99.8%), 4-methyl valeric acid (internal standard), acetone (>99.5%, GC grade), metaphosphoric acid, alchol oxidase (from Pichiapastoris), Tris (Trizma grade), D-(+)-galacturonic acid, sodium chloride, sulfamic acid, formaldehyde, 3-methyl-2-ben- zothiazolinone hydrazine (MBTH), ammonium iron sulfate dodecahydrate, sulfuric acid, 3,5 dimethylphenol (DMP) and copper sulfate were purchased from Sigma-Aldrich (Sigma Chemical Co., St Louis, MO). Fructooligosaccharides (FOS) powder was provided by GTC Nutrition (Golden, CO, USA). Two pectins, HMP (high methoxy pectin) and SBP (sugar beet pectin), were kindly obtained from Tic Gums (Belcamp, MD, USA) and Herbstreith & Fox (Elmsford, N.Y., USA), respectively. Soy pectin (SOY) was prepared according to Crandall & McCain [16] .

Galacturonic acid was released after acid hydrolysis of pectin samples. Total galacturonic acid was determined using a modified DMP (3,5dimethylphenol) method [17] with a maximum absorbance of 450 nm. A modified AO/MBTH method [18] was used to determine the degree of methylation of pectin samples. Each pectin sample (1 mg) was dissolved into a mixture that contained 700 μL of 20 mM Tris-HCl buffer (pH 7.5), 100 μL of MBTH (dissolved in water at 3 mg/mL) and 100 μL of 0.5 M NaCl solution. The mixture reacted with alcohol oxidase (1 unit) at 30˚C for 30 min for full digestion. A 200 μL aliquot of acidic iron solution was added after digestion. After 30 min, absorbance was determined using spectrophotometer (DU 520, Beckman Coulter, Brea, Calif., USA) at 620 nm. The molecular weight distribution of the pectin samples was determined by high-per- formance size-exclusion chromatography on a series ShodexOHpak columns (KB-802, and KB-804) with HPLC system (Waters, Miliford, MA) that consisted of a 515 HPLC pump and 2410 refractive index detector [19] .

The study was reviewed and approved by the Institutional Review Board (IRB) at the University of Arkansas. Due to the limitation of soy pectin (SOY) material, four subjects were selected from University of Arkansas and the surrounding Fayetteville area. Dietary and health/medical questionnaires and informed consent forms were obtained from each participant before the sample collection. Fecal samples were collected from 4 male participants (23 to 28 years of age) who had been on a routine diet and not taken any antibiotics or medicine for 6 months prior to the study. Average energy and dietary fiber intakes of participants was processed with Nutritionist Pro (Version 4.4.0, Axxya system). Each fresh fecal sample was collected in in Commode Specimen Collection System (Fisher Scientific, Waltham, MA).

An in vitro incubation was conducted to evaluate the fermentability of four substrates (FOS as a control, HMP, SBP and SOY) by human fecal bacteria. Two grams of fecal samples were added to 22 mL of sterile brain heart infusion (BHI) media. The BHI media (Difco Laboratories, Detroit, MI) was prepared according to Zheng et al. [20] . Tubes were incubated for each sample as well as a blank (BHI + feces). All steps for fermentation were conducted in an anaerobic chamber (Coy Laboratory Products Inc., Grass Lake, MI). The mixture was homogenized by vortexing right before incubation (37˚C). Four milliliters were taken using a sterile syringe from each tube immediately for time 0 and stored at −20˚C. Subsequent aliquots were obtained at 6, 12, 24, and 30 h and stored at −20˚C until analysis.

To make a standard curve, a standard solution containing the three SCFAs were prepared by mixing 100 μL of 4-methyl valeric acid (50 mM) of the internal standard and diluting with acetone. Sample aliquots were thawed in room temperature and centrifuged at 3500 ×g for 15 min. After centrifugation, 0.5 mL of each aliquot and SCFA standards were mixed with 100 μL of a mixture containing 50 mM of 4-methyl-valeric acid (Internal Standard), 5% meta-phosphoric acid, and copper sulfate (1.56 mg/mL). Each tube was centrifuged at 11,000 ×g for 10 min after reacting with the mixture for 10 min and stored at −20˚C until analysis. SCFA contents was analyzed in a gas chromatograph (Shimadzu GC-2010; Shimadzu Scientific Instruments, Columbia, MD) equipped with a flame ionization detector and using a fused silica capillary column (ID-BP21; SGE, L: 30 m, I.D: 0.25 mm, Film: 25 μm, Ringwood, Victoria, Australia). The oven temperature was increased by 4˚C/min from 100˚C (2 min) to 120˚C (1 min), then at 3˚C/min until 150˚C. Hydrogen was used as the carrier gas. The sample injection (1 μL) was performed in the split mode (30:1). The concentrations of acetic acid, propionic acid, and butyric acid were calculated from the calibration curves of each SCFA standards.

Bacterial DNA was extracted using QIAamp DNA stool kit (QIAGEN, Valencia, CA). An aliquot (250 mg) of fecal sample was lysed using Garnet beads (MO BIO, Carlsbad, CA) and the lysed sample was centrifuged at 3000 × g for 1 min. The sample was further vortexed for 10 min and incubated at 95˚C for 6 min. After incubation, the sample was centrifuged at 16,100 ×g for 1 min. Eluted DNA was further processed for PCR procedures.

All DNA was amplified using touchdown PCR. Bacterial DNA amplification and denaturing gradient gel electrophoresis (DGGE) procedure were performed according to Hanning and Ricke [21] . A 233 bp portion of the 16s rRNA gene was amplified by PCR using primers of DGGE-F (5-CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG GCCTAC GGG AGG CAG CAG -3) and DGGE-R (5-ATT ACC GCG GCT GCT GG-3) (Integrated DNA Technologies, Coralville, IA) [22] . The touchdown PCR program was: initial denaturation at 95˚C for 2 min, then 17 cycles of 1) denaturation at 94˚C for 1 min; 2) annealing at 67˚C for 45 s decreasing by −0.5˚C per cycle to a touchdown temperature of 59˚C and 3) annealing at 72˚C for 2 min. The reaction was followed with 12 cycles of 4) denaturation at 94˚C for 1 min; 5) annealing at 58˚C for 45 s with a final elongation step at 72˚C for 7 min. PCR product obtained was confirmed by 0.8% agarose gel electrophoresis with ultraviolet transillumination using the Quantity One software (Bio-Rad Laboratories, Richmond, CA).

The 8% polyacrylamide gels (acrylamide:bisacrylamide = 37:1) consisted of a 35% to 60% of urea-deionized formamide gradient; the 100% denaturing acrylamide was composed of 7 M urea and 40% deionized formamide. The equal concentration of PCR product (2 µg) was mixed with loading buffer [0.05% bromophenol blue (w⁄v); 0.05% xylene cyanol (w⁄v); and 70% glycerol (v⁄v)] and the mixed samples were loaded in each well. Electrophoresis was carried out in a D-Code Universal Mutation Detection System (Bio-Rad Laboratories) in TAE (Tris-acetate-EDTA) buffer (1 mM EDTA, 20 mM acetic acid, 40 mM Tris) at 59˚C for 17 h at 60 V. The gels were stained with SYBR Green (1:50,000 dilution, Cambrex Bioscience, Walkersville, MD) in TAE for 40 min and the result was observed with ultraviolet transillumination for image acquisition.

Bands of interest were excised from the DGGE (Denaturing Gradient Gel Electrophoresis) gel using a previous protocol [23] . The extracted DNA was sequenced in the DNA Resource Center at the University of Arkansas (Fayetteville, AR) using an ABI 3100 capillary analyzing system (Applied Biosystems, Foster City, CA) and the sequences were compared with the GenBank database using the Blast algorithm. For showing phylogenetic analysis, UPGMA’s method was used for the analysis using Bio-rad quantity One (software ver. 4.6.7).

Statistical analysis was carried out by statistical program: SAS 9.4 (SAS Inst., Cary, N.C., USA), using either 1) one-way ANOVA when comparing three or more data sets or 2) a t test when comparing two data sets. All data are presented as mean ± S.E.M (Standard Error of Mean). Differences were deemed significant when P < 0.05.

Table 1 shows the participant information including age, body mass index (BMI), average daily calorie, and dietary fiber intake. The BMI of subject 4 was within the normal range (18.5 - 24.9), in contrast, subject 1 through 3 were overweight (25.0 to 29.9). The US Department of Agriculture recommends men consume 2000 to 2600 calories if they are sedentary, 2200 to 2800 calories if they are moderately active. In this study, all of subjects consumed less than 2000 kcal of energy.

Evidence has been firmly established that consuming recommended dietary fiber (25 - 35 g/day) exerts a beneficial influence on human colonic health and provides many health benefits such as preventing coronary heart disease, stroke, hypertension, diabetes and obesity [24] [25] . In addition, long-term dietary fiber intake is

Average values are expressed as mean ± SEM.

associated with the number and composition of gut microbiota as well as reducing gut apoptosis [26] [27] . Subject 1 and 3 consumed close to 30 g daily, while subject 2 and 4 (18.9 g and 17.7 g, respectively) consumed less than the recommended daily intake of dietary fiber. It is presumed that the gut microbiota from subject 1 and 3 might have greater availability to use dietary fiber than the individuals who consumed the lower level of dietary fiber (subject 2 and 4). Lesser availability of dietary fiber might be one of the fundamental reasons for lower production of SCFAs.

The average degree of polymerization (DP) of pectin samples was determined using an HPLC system; SOY- 1510, SBP-3729, HMP-492 (Figure 1). Also, the degree of pectin methylation (DM, %) of each sample was determined as 22.8% (SOY), 21.2% (SBP), and 76.0% (HMP), respectively. SOY and SBP were determined as low-methoxypectins (LMP, DM < 50%). As shown in Figure 1, the HPLC profile of SBP and SOY exhibited a similar pattern, while the HPLC profile of HMP responded differently.

Intrinsic factors such as DP, DM, and chain length are key parameters affecting the behavior of pectin [28] . There are only few studies that have compared bioavailability or health effects between pectin with different chemical characteristics (DP and DM), source, and type. Citrus pectin sources (high DM and high MW) exerted greater LDL cholesterol-lowering properties than low DM and low MW of citrus pectin and high DM of orange pulp fiber [29] . However, Judd and Truswell [30] reported that LMP has similar cholesterol-lowering effects compared to HMP. Wet and dry weights of feces were higher on HMP, while gut transit time and fecal water were similar in both groups (HMP and LMP). With even limited evidence from previous studies, differences with DP and DM of pectin samples might have different fermentability by human gut microbiota and bioavailability for human health.

Production of SCFAs can be stimulated with dietary fiber. Titgemeyer et al. [31] found that fermentation with citrus and apple pectin increased the SCFAs more than fiber sources from sugar beet and oat. In the present study, total SCFAs production of subjects with samples is shown in Table 2. These values were calculated for the sum of SCFAs: acetate, propionate, and butyrate. Each pectin sample had different patterns for production of SCFAs. Specifically, fermentation with SOY showed significantly higher propionate production (P < 0.05). FOS showed the highest butyrate production during 30 hrs.

Values are expressed as mean + SEM. Superscripts not sharing a common letter within the same row are significantly different among groups at P < 0.05. FOS: fructooligosaccharides; HMP: high methoxy pectin; SBP: sugar beet pectin; SOY: soy pectin.

All pectin samples (HMP, SBP, and SOY) showed higher acetate production than fermentation without substrate (Figure 2(A)) (P < 0.05). Acetate, the principal SCFA in the colon, is readily absorbed and transported to the liver, and therefore less metabolized in the colon [32] . In addition, acetate can be used as a source for butyrate. Diez-Gonzalez et al. [33] reported two different mechanisms (acetate utilization and lactate fermentation) and enzymes (butyrylCoA:acetate CoA transferase and butyrate kinase) that are important factors for production of butyrate in the gut. Also, Coprococcus sp., Roseburia sp., R. intestinals, and Facalibacteriumprausnitzii are known as bacteria that can convert butyrate, using acetate [34] . Acetate is often used to monitor colonic events because it is the primary SCFA. Also, increasing the concentration of acetate resulted in lowering pH; this reaction might be related to the beneficial influence on the composition of gut microbiota and preventing the proliferation of harmful species and growth of pathogenic bacteria.

The production of propionic acid during in vitro fermentation (30 h) is illustrated in Figure 2(B). Fermentation with SOY demonstrated significantly higher propionate production than other pectin samples (HMP and SBP) and FOS during 30 hrs (P < 0.05). Propionate has been shown to lower glucose-induced insulin secretion in isolated pancreatic islet cells of rats [35] . It is also reported that propionate has anti-proliferative effect towards colon cancer [36] [37] and is related to weight control and feeding behavior [38] [39] .

Butyrate is the preferred energy source of colonocytes; approximately 70% to 90% of butyrate is metabolized by the colonocytes. Various studies have found that butyrate has a positive effect in the prevention of colon cancer [3] [36] [40] . Diez-Gonzalez et al. [33] reported that butyrate is formed from two molecules of acetyl coenzyme A that yields acetoacetyl-CoA, which is converted into butyryl-CoA. Consequently, butyryl-CoA might yield butyrate through butyrate kinase by some butyrate-producing strains such as Butyrivibriofibrisolvens or via butyryl-CoA: acetate-CoA transferase. Compared to other SCFAs, concentration of butyrate increased at later time point than other SCFAs (Figure 2(C)). According to Duncan’s study [41] , butyrate can be converted, as gut microbiota consumes acetate. Thus, current result implies production of SCFAs also might be affected by other SCFAs and SCFAs can be changed interdependently in anaerobic fermentation.

The human intestinal microbiota has been known to contain complicated colonies composed of at least several hundred different species of bacteria with approximately 10¹¹ to 10¹² cells per gram of feces [42] [43] .

Based on DGGE results, Figure 3 and Table 3 show patterns of each subject, excised bands number and identifications. In this study, all subjects had very diverse gut microbiota. Even though most of the sequences of bands were assigned to uncultured bacterium or were not identified, Table 3 shows the dominant strains indicated by the banding patterns.

Microbial metabolites, potential growth factors in fecal material, and dietary habit are important factors to affect the composition of gut microbiota. In the present study, Firmicutes species, Faecalibacterium species, Bacteroides species that have been regarded dominant human fecal bacterium were detected in subjects 1 to 4. The band of Firmicutes species was enriched after exposure to dietary FOS, SBP and HMP from subject 1 as well as FOS with subject 3. While Firmicutes species were detected from subject 1 and 3, the band of Firmicutes species was not detected in subjects 2 and 4. DGGE bands for fecalibacterium species were detected in subjects 1, 3, and 4. The band identified as fecalibacterium was present in all samples and the blank from subjects 1, 3, and 4. Bacteroides was present in subject 2 and 3. These three strains: Firmicutes species, Faecalibacterium species, and Bacteroides species were reported to be related to the production of SCFAs and host health. According to Shen’s study, Bacteroides numbers may correlate with the increase in propionate production [44] . The number of dominant fecal bacterium is changed by disease status or body weight [45] [46] . Schwiertz et al. [47] reported

that the ratio of Firmicutes to Bacteroidetes altered in favor of the Firmicutes in overweight and obese subjects.

Interestingly, specific bacteria were also detected from each subject. Lactobacillus rumis is one of the dominant Lactobacillus species that have been known as probiotics. L. rumis showed positive effects on human colonic health, such as inhibition of pathogenic microorganisms and relief of lactose maldigestionsymtoms [48] . In subject 1, the band for L. rumis was found to be most intense with soy pectin sample (SOY). Based on this result, soy pectin might have a positive effect on the growth of L. rumis. The band of Bifidobacterium species was found in subject 3 and 4, fermentation with SBP or without substrates. Pseudobutyrivibrio species were also detected in subject 4. The bands of Pseudobutyrivibrio specises in subject 4 were enriched with fermentation of all pectin samples and FOS, except Blank. This strain has been reported that it is closely related to butyric acid production [49] .

From analysis of SCFAs, all pectin samples showed higher butyrate production. This result might be thought that pectin samples and FOS stimulate growth of Pseudobutyrivibrio specises and subsequently metabolize more butyrate production.

In order to compare composition of microbiota fermented with different samples by each subject, analysis with the phylogenetic trees is shown and illustrated the correlation between samples and subjects (Figure 4). The phylogenetic trees of samples generated by the UPGMA algorithm exhibited 5 different groups in each treatment. Subject 4 can be categorized as a normal weight person, while other subjects are considered as overweight. From phylogenetic analysis with bands from samples fermented without substrates (BL) it appears that subjects 4 and subject 2 (lower dietary fiber intake) are clustered with 70% of relatedness. Also, subject 4 was clustered with subject 3 (65%) and subject 1 (64%), respectively. This result indicates that subject 4 and subject 2 have similar pattern of microbiota, than the other subjects. This pattern was also detected with FOS and SBP samples. However, samples fermented with SOY and HMP exhibited different patterns compared to the blank. In particular, the relatedness was decreased by fermentation with the respective samples. Even if same source was utilized for fermentation, the composition of microbiota became more diverse due to the difference of initial

microbiota composition of subjects. This result implies that composition of microbiota can be altered by substrates, but ecology of microbiota did not change significantly during 30 h of fermentation, because each subject had a different composition of microbiota at the beginning.

The current study investigated the changes in microbiota during in vitro fermentation. Even though no relationships between specific species and pectin samples were determined due to the limitation of the DGGE experiment, the data demonstrated that pectin samples showed different fermentation patterns. Further study could validate this with quantification to determine the specific microbiota related to SCFA production.