Fecal Microbiota and Associated Volatile Organic Compounds Distinguishing No Adenoma From High-Risk Colon Adenoma Adults

1. Introduction

Colorectal cancer (CRC) is the third most common and the second deadliest cancer worldwide with 1.9 million new cases and 0.9 million recorded deaths in 2020 [1]. The multifactorial nature of CRC involves many risk factors, some of which have a clear environmental component and are thus modifiable, such as, lifestyle, obesity, diet, alcohol intake, tobacco use and biological aging, while others have a clear genetic component and are relatively fixed, such as, sex, ancestry, identifiable inherited mutations, and family history of proneness to cancer [2]. Two biomedically quantifiable and modifiable factors are the billions of microbes residing in the intestine and the thousands of metabolites they generate. These affect epithelial homeostasis and the host immune system and in turn tissue regeneration and predisposition to cancer [3,4]. Some of the gut microbes can ferment plant derived dietary fibers and animal protein-derived amino acids facilitating host metabolism and a balanced intestinal biochemistry [5]. An emerging risk factor for CRC development is intestinal dysbacteriosis which results from the presence of certain bacteria, diets, lifestyles and clinical pathologies. [2]. Accordingly, the intestinal bacteriome and metabolome provide the potential to identify novel non-invasive biomarkers for colonic inflammatory disorders and CRC. Fusobacterium nucleatum, colibactin positive (pks+) Escherichia coli and enterotoxigenic Bacteroides fragilis (ETBF) have been causally linked to CRC, while the link to CRC of Clostridium symbiosum, Enterococcus fecalis, Streptococcus bovis, Peptostreptococcus anaerobius, Parvimonas micra and Porphyromonas species remains to be established [6,7,8,9,10,11]. Higher levels of some of these species and strains may distinguish High-Risk Adenoma (HRA) and early-stage CRC patients from healthy No Adenoma (NA) individuals. Moreover, microbial biomarker discovery may be improved, when combined with the characterization of the intestinal metabolome. For example, combining Ultra High Performance Liquid Chromatography Mass Spectrometry (UHPLC-MS) with metagenomics data allowed the link of cholesteryl esters and sphingolipids as well as of Fusobacterium, Parvimonas and Staphylococcus with CRC and provided combinatorial microbiome-metabolome analysis towards early disease diagnosis [7]. Intriguingly, breath, urine, and fecal volatile organic compounds (VOCs) provide an alternative and promising clinical approach to intestinal inflammation and early CRC diagnosis, despite the inadequate strength of evidence and differing analytical platforms [12,13].

Here, we link fecal bacteria at different taxonomic levels and fecal volatile compounds to HRA status by sampling and analyzing a Cypriot population. We use 16S rRNA sequencing (16S-Seq) and HeadSpace Solid Phase Micro-Extraction Gas Chromatography Mass Spectrometry (HS-SPME-GC-MS) to identify differences in fecal bacteria abundance and VOC concentrations between NA and HRA individuals. Moreover, we performed Spearman’s rank order correlations between the fecal bacteriome and VOCs indicating intestinal health versus pre-cancerous dysbiosis, followed by binomial logistic regression modelling.

2. Materials & Methods

2.1. Sample collection

Fecal samples in this study were collected under the Cyprus Intestinal Health Study (MoCo Project EXCELLENCE/1216/0523) funded by the Research and Innovation Foundation of Cyprus. Bioethical approval was obtained from the Cyprus National Bioethics Committee (Protocol numbers: EEBK/ΕΠ/2015/38 and EEBK/ΕΠ/2019/23). 117 participants provided fecal samples ≥15 days after conventional colonoscopy per established assessments [14] which were stored at −80 °C until analysis. 100 of these were assigned an NA status due to the absence of tumor detection during colonoscopy. The remaining 17 individuals were assigned an HRA status according to colonoscopy and histopathological reports recording ≥3 adenomas/serrated polyps, or ≥1 adenoma/serrated polyp ≥1 cm, or ≥1 villus or tubulovillus adenoma, or high-grade dysplasia per established criteria [15,16]. 55 and 62 individuals were males and females, respectively, 50–70 years old, undergoing routine colonoscopy (Table 1). No individual had a history of CRC or inflammatory bowel disease (IBD) or received antibiotics treatment or suffered from gastroenteritis during the month before colonoscopy or sample collection.

2.2. Fecal bacteria DNA isolation and 16S gene amplicon sequencing

Fecal bacteria DNA isolation and purification was performed using the PureLink™ Microbiome DNA Purification Kit (Invitrogen™) using 0.18–0.2 g from the initial fecal sample, and 90 μl of eluted DNA was stored at −80 °C. The 16S rRNA gene hypervariable regions V3 and V4 were sequenced using the Nextera XT Library Preparation kit (Illumina™, Inc., San Diego, CA, United States) [17] and the following primers containing overhang adapter sequences:

Forward: TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG

Reverse: GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC

16S Gene Amplicon Sequencing for the taxonomic classification was performed using a 300 bp paired-end run on an Illumina MiSeq™ platform, following the standard Illumina protocols.

2.3. Bioinformatics and statistical analyses for metagenomics

Metagenomics Operational Taxonomic Units (OTU) analysis was performed, determining the relative abundance of each bacterial taxon from phylum to species. Bioinformatic OTU analysis was performed from FASTQ files with paired-end reads utilizing Ribosomal Database Project Classifier against the RefSeq RDP 16S ver4.3 database [18]. Quality control was applied by requiring the detection of each OTU in at least 30% of samples and a minimum average relative abundance of 1% in at least one of the two groups (NA and HRA). Average bacterial abundance at the phylum, family, genus, and species taxonomic level was compared between the NA and HRA groups using a Mann-Whitney U test.

To display bacterial taxa percentiles (raw relative abundance) and hierarchy based on identified reads for the NA and the HRA groups we used Krona visualization with classification imported in an Excel template detailing lineage and magnitude [19]. Inverse Simpson index was used to calculate alpha diversity and UniFrac index to compute beta diversity, both by using the VEGAN package [20].

2.4. HS-SPME-GC-MS headspace analysis of fecal VOCs

18 NA and 10 HRA individuals were randomly selected for fecal volatolomics. Approximately 0.6 g of each frozen sample added in a 20 ml headspace glass vial (Agilent; Part#: 8010-0413) was thawed for 24 hrs at room temperature. Then, 5μL of internal standard solution of chlorobenzene-d5 (Sigma-Aldrich; Product#: 48086, CAS#: 3114-55-4) with a final concentration of 25 ppb, was injected in the headspace vial the sample left to equilibrate in the closed crimp seal vial for 24 h at room temperature and it was then incubated in a water bath at 60 °C for 1 hr. Consecutively, the 75μm CAR/PDMS SPME fiber was exposed to the headspace phase of the vial for 30 min, so as to achieve the extraction of the small volatiles contained in the headspace phase. VOCs were thermally desorbed from the SPME fiber in an Agilent Single-Quadrupole GC-MS Instrument (GC-7890B, MSD-5977B, Agilent Technologies, USA) (Supplementary Information in SI1).

2.5. Volatolomic and combinatorial omic statistical analyses

For normalization, raw VOCs values were divided by the value of the internal standard (chlorobenzene-d5, 25 ppb). Quality control was applied by minimal detection of at least 3 values of VOCs in both groups. To VOCs levels between NA and HRA groups the Shapiro Wilk normality test was performed prior to the Mann-Whitney U test (non-parametric, if normality fails), or independent sample t-test (parametric, for normally distributed variables). Principal Component Analysis (PCA) was done with the FACTOEXTRA package in R and the production of a heatmap to show the values of VOCs across each sample was completed with the GPLOTS package in R [21,22]. Bacterial families and genera correlations with VOCs, shown in Figures 5 and 6 and Tables 3 and 4, were measured in terms of strength and direction via Spearman’s rank order correlation. The probability of certain interactions among an individual status (NA or HRA), VOCs and relative abundance of bacteria taxa, shown in Figures 6 and 8, was tested via binomial logistic regression analysis, as a tool for predictive modeling of CRC proneness. The a-priori set threshold for statistical significance in all tests was p-value ≤ 0.05.

3. Results

3.1. Fecal bacteria prevalence and diversity in NA and HRA individuals

Fecal samples from 117 female and male adults between the age of 50 and 70 (Table 1), divided into 100 NA and 17 HRA in accordance with established criteria [15] were analyzed via 16S-Seq generating 38,6 million quality-filtered reads, 87% of which were identified. Krona plots revealed raw relative abundances as percentiles of total identified bacteria sequence reads per taxonomic level for the NA and the HRA group (Figure 1 and Figure 2). Firmicutes was the most prevalent phylum in both groups, covering 68% in NA and 41% in HRA individuals. Lachnospiraceae and Ruminococcaceae, the most prominent Firmicutes families in the NA group were tentatively less prevalent in the HRA group. The Blautia, Roseburia and Fusicatenibacter genera collectively covered 48% of the Lachnospiraceae family sequence reads in ΝA, and 19% in HRA individuals. Bacteroidetes and Actinobacteria phyla were comparable in the two groups: Bacteroidetes covered 20% of the sequence reads in NA, and 14% in HRA individuals, while Actinobacteria covered 9% in NA and 10% in HRA. Interestingly, Proteobacteria covered 23% and 1% of the sequence reads in HRA and NA individuals, respectively. Accordingly, Enterobacteriaceae, a prominent Proteobacteria family, was prominently abundant in HRA individuals.

Inverse Simpson index was applied to assess alpha-diversity within the groups of 100 NA and 17 HRA individuals at the phylum, family, genus and species level. The higher the value of this index the greater the diversity within the group. As expected, the alpha-diversity increased for each group from family to species level, but the index of the NA and HRA groups at a given taxonomic level was in all cases comparable (Supplementary Figure S1). To determine potential dissimilarities in the microbial communities between the NA and the HRA group, the phylogenetic distance between sets of phyla, families, genera, and species unique to either the NA or the HRA group we used the UniFrac phylogenetic method. The Multidimensional Scaling (MDS) representation of beta-diversity measurement showed no significant differences at any taxonomic level between the NA and HRA groups (Supplementary Figure S2).

3.2. Significant fecal bacteria differences between NA and HRA individuals

We used a Mann-Whitney non-parametric test to pinpoint statistically significant differences between NA and HRA individuals in the relative abundance of bacteria at the phylum, family, genus, and species level (Table 2, Figure 3). Differences were accepted at p-value ≤ 0.05 and normalized mean relative abundance ≥ 1% in at least one of the two groups. Accordingly, the phylum of Proteobacteria was ≈2 times more abundant in HRA versus NA individuals. Members of the Lachnospiraceae family, namely, the Roseburia and Fusicatenibacter genera, were ≈2 times more abundant in NA individuals. At the species level, Roseburia faecis, Blautia luti, Fusicatenibacter saccharovorans and Eubacterium rectale belonging to the Lachnospiraceae family, as well as Phascolarctobacterium faecium belonging to the Acidaminococcaceae family were more abundant in NA individuals. To the contrary, Parabacteroides distasonis a bacterial species with a potential pathogenic role belonging to the Tannerellaceae family was ≈3 times more abundant in HRA individuals.

3.3. VOCs abundance in NA and HRA individuals

HS-SPME-GC-MS analysis was used to evaluate the profile of volatile organic compounds in the feces of a subset of the initial individuals: 18 NA and 10 HRA individuals. Out of over 250 detected volatiles 71 were present in the samples of at least 3 individuals in each group. Principal Component Analysis (PCA) for the 71 essential volatiles emitted (Supplementary Figure S3) indicated dispersed distributions of samples, that is, significant sample to sample variations in VOCs. Moreover, there was a lack of distinct sample clustering indicating similarity between the VOCs of the NA and the HRA group. This may be partially due to differences in the dietary habits of the sampled individuals. However, branched chain fatty acids (BCFAs), methyl propionate, methyl butyrate and isobutyric acid, and the ketones, methyl butyl ketone (2-Hexanone) and ethyl acetone (2-Pentanone) were significantly more abundant in HRA versus NA individuals (Figure 4, Supplementary Table S1).

3.4. Associations between bacterial families and VOCs in NA and HRA individuals

We examined correlations between fecal bacterial families (FBFs) and fecal VOCs using 28 fecal samples from 18 NA and 10 HRA individuals. We identified 68 pairwise associations between 16 FBFs and 30 VOCs, as shown in the Chord diagram of Figure 5 and the Supplementary Table S2. Of all FBFs, Bacteroidaceae and Eubacteriaceae exhibited the most associations with VOCs, 10 and 9, respectively. Of all VOCs, acetaldehyde and propanal exhibited the most associations with FBFs, 7 of them each.

To gain insight regarding distinctions between the NA and HRA groups, we explored all combinations and managed to associate 4 of the VOCs found significantly more abundant in HRA individuals, butyl methyl ketone, isobutyric acid, methyl butyrate and ethyl acetone (Figure 4), via Spearman’s correlation analysis with 6 FBFs, Eubacteriaceae, Lactobacillaceae, Bacteroidaceae, Erysipelotrichaceae, Acidaminococcaceae, and Peptostreptococcaceae (Table 3 and Figure 5). Moreover, we associated Lachnospiraceae a FBF significantly less abundant in HRA individuals (Figure 3), with 4 VOCs, propanal, methacrolein, methyl 4-methylvalerate, dimethyl trisulfide (Table 3 and Figure 5). Strikingly, the direction of each FBF-VOC interaction was opposite in most cases between the NA and the HRA group (Table 3). Similarly, binomial logistic regression analysis of the 18 NA and the 10 HRA samples regarding the key FBF-VOC associations described in Table 3, revealed five FBF-VOC co-abundance correlations and distinct trends for the NA and the HRA groups, as potential predictive models of CRC proneness (Figure 6).

3.5. Associations between bacterial genera and VOCs in NA and HRA individuals

We also examined correlations between fecal bacterial genera (FBGs) with fecal VOCs using the 28 fecal samples from 18 NA and 10 HRA individuals. We identified 96 different associations between 27 fecal bacterial genera and 41 VOCs, as shown in the Chord diagram of Figure 7 and Supplementary Table S3. Of all FBGs, Bacteroides and Eubacterium exhibited the most associations with VOCs, 10 and 9, respectively. Of all VOCs propanal exhibited the most (13) associations with FBGs, while acetaldehyde, isobutyraldehyde, isovaleraldehyde and methyl cyclohexanecarboxylate with 5 FBGs each.

To gain distinct insight regarding the NA and HRA groups, we explored all combinations and managed to associate four VOCs found significantly enriched in HRA individuals, butyl methyl ketone, isobutyric acid, methyl butyrate, ethyl acetone (Figure 4), via Spearman’s correlation analysis with 8 FBGs, Ruminococcus, Lachnospiraceae incertae sedis, Collinsella, Bacteroides, Coprococcus, Bacteroides, Holdemanella and Eubacterium. Moreover, we associated FBGs significantly less abundant in HRA individuals, Roseburia and Fusicatenibacter (Figure 3), with 3 VOCs, propanal, p-Cresol and indole (Table 4, Figure 7). Strikingly, the direction of each FBG-VOC interaction was opposite in most cases between the NA and the HRA group (Table 4). Similarly, binomial logistic regression analysis of the 18 NA and the 10 HRA samples regarding the key FBG-VOC associations described in Table 4, revealed 10 FBF-VOC co-abundance correlations and distinct trends for the NA and the HRA groups, as potential predictive models of CRC proneness (Figure 8).

4. Disscusion

Countless interactions take place within the large intestine between volatile compounds and the intestinal microbiota influenced by host diet, age, metabolism, inflammation-related processes, medications and other environmental factors. VOCs are generated, modified or degraded by bacteria residing the intestine and the host itself. Some VOCs may benefit, while others may destroy the health promoting microbial composition balance [31]. During the last decade, microbiome changes and specific bacterial species have been linked to cancer development and progression [32,33]. Nevertheless, it is still early to name microbial and biochemical signatures of predictive value for CRC risk. Through our ongoing clinical study, we find significant changes in the microbial composition and fecal VOCs between NA and HRA individuals residing in the island of Cyprus. Beyond our independent microbiome and volatilome analysis, we indicate a level of complexity revealed by assessing fecal bacteria-VOC associations and another one by assessing differences in these associations between NA and HRA individuals.

4.1. Fecal bacteriome analysis

To determine differential microbial abundances between the NA and HRA individuals, we compared microbial composition at the phylum, family, genus and species level utilizing Mann-Whitney non-parametric statistical analysis. Proteobacteria and Parabacteroides distasonis, a member of the Tannerellaceae family, were more abundant in HRA individuals, while members of the Lachnospiraceae family, Roseburia faecis, Blautia luti, Fusicatenibacter saccharovorans and Eubacterium rectale, as well as Phascolarctobacterium faecium, member of the Acidaminococcaceae family, were more abundant in NA individuals.

Based on previous studies depletion of members of the Clostridia class, such as Lachnospiraceae, Ruminococcaceae and Eubacteriaceae, indicate colonic subclinical inflammation within which high-risk adenomas may form [2,34]. Clostridia can ferment dietary plant fibers producing butyrate and other SCFAs, such as propionate, acetate and valerate. Moreover, the anaerobic Lachnospiraceae and Ruminococcaceae families may play a preventive role in CRC development, since the relative abundances of Lachnospiraceae and its metabolites have been inversely correlated with CRC progression [24].

R. faecis, B. luti, E. rectale and F. saccharovorans species are members of the Lachnospiraceae family and SCFA-producers that preserve gut-homeostasis and protect the intestinal mucosal cells from becoming hyperplastic, dysplastic or malignant by regulating colonic inflammation [35]. Moreover, members of the Fusicatenibacter and Roseburia genera, such as F. saccharivorans, and R. faecis, are significantly reduced in CRC patients compared to healthy individuals [23,30]. Similarly, B. luti is depleted in CRC patients vs. healthy-controls [25,36], consistent with its probiotic properties [27]. P. faecium, another SCFA producer, utilizes succinate, generated in the large intestine by bacteria of the Bacteroides and Parabacteroides genera, sustaining its abundance during aging [37]. The abundance of the five aforementioned commensal bacterial species may thus indicate healthy versus cancer-prone status and may serve as bacterial biomarkers of health, although prospective and experimental studies are required to provide deeper insight into their role.

On the other hand, P. distasonis appears to be a pathobiont in some cases, present in a healthy gut, while enriched in human abscesses, extra-intestinal abdominal infections and in Lynch syndrome patients [38,39,40,41]. While potentially anti-inflammatory and protective against CRC in other cases, it is associated with pre-existing inflammatory bowel disease in both humans and animal models [38]. Therefore, P. distasonis along with Proteobacteria species enriched in HRA individuals, may be indicative of intestinal inflammation and CRC risk.

Raw relative abundance of Gammaproteobacteria class and the Enterobacteriaceae family derived from Krona plots was many-fold higher in HRA individuals in agreement with their clear association with conventional and serrated adenomas [34,42]. Similarly, the low but detectable presence of the opportunistic pathogen and oncobacterium Fusobacterium nucleatum was only recorded in HRA individuals, indicating potential similarities between the microbial ecosystem composition within the gut of HRA individuals and CRC patients [43,44,45,46,47].

4.2. Fecal volatilome analysis

Methyl propionate, a carboxylic ester, and methyl butyrate, a fatty acid ester, were found elevated in the feces of HRA individuals, suggesting that these BCFAs may contribute to pathogenesis [48]. Both of them are low molecular weight volatiles, highly abundant in the human feces, and products of the exogenously esterification of propionate and butyrate derived from dietary fibers and microbial metabolism within the large intestine [5,49].

Isobutyric acid is a branched-chain saturated fatty acid primarily derived from the branched chain amino acid valine via intestinal fermentation mediated by Clostridium and Bacteroides species. Its concentration increases progressively along the proximodistal colon axis and in feces. BCFAs are proposed to affect human health but they are relatively unexplored compared to SCFAs [5]. Interestingly, we found isobutyric acid and Bacteroides and Clostridium genera in higher levels in the feces of HRA individuals. Similarly, we found methyl propionate, methyl butyrate and isobutyric acid in higher levels in HRA individuals, in agreement with previous findings about the greater abundance of isobutyric acid in CRC relative to HRA and healthy control individuals [50]. These BCFAs may thus serve as candidate biomarkers of CRC risk.

4.3. Fecal bacteriome to volatilome analysis

Spearman’s rank order correlation analysis of fecal bacteria families and genera with fecal VOCs revealed 11 FBG-VOC associations (Table 3), and 12 FBF-VOC associations (Table 4). Isobutyric acid correlates with the Lacnospiraceae incerta sedis, Bacteroides, Colinsella and Coprococcus genera in different ways between the NA and HRA groups (Table 4). This BCFA was positively correlated via predictive modeling using binomial logistic regression with Lacnospiraceae incerta sedis and Bacteroides in NA individuals, and negatively in HRA individuals (Figure 8). While isobutyric acid predictive modeling correlations with Coprococcus and Colinsella were found in reverse: negative in NA and positive in HRA individuals (Figure 8). BCFAs, such as isobutyric and isovaleric acids, are less abundant than SCFAs in the human large intestine and feces. They are markers of amino acid fermentation, and their intestinal and fecal abundance is related to diet and aging [5]. Previous studies indicate members of the Proteobacteria and Actinobacteria phyla, the Lachnospiraceae family, and Fusobacteria and Bacteroides genera as main producers of BCFAs and markers of colonic protein fermentation, a process that also generates p-cresol, phenol and ammonia [5,51,52]. Moreover, we associated Lachnospiraceae with 4 VOCs, propanal, methacrolein, methyl 4-methylvalerate, dimethyl trisulfide (Table 3). Propanal was also associated with Roseburia (Table 4), but the biological significance of the association of this and other potentially pathogenic VOCs with potentially beneficial bacteria is unclear due to the positive correlation of this volatile in breath (not fecal) samples of CRC patients [53]. Notably, breath sample VOCs are expected to defer from fecal VOCs, despite their potential predictive power [54,55].

Methyl butyrate, a volatile compound involved in SCFA butyrate production, corelates with Bacteroides and Coprococcus genera (Table 4). Moreover, it was positively correlated via predictive modeling with Bacteroides in NA individuals and negatively corelated in HRA individuals (Figure 8). On the other hand, methyl butyrate was negatively correlated with Coprococcus in NA individuals and positively corelated in HRA individuals (Figure 8). These observations need to be further investigated, since Bacteroides and Coprococcus genera are abundant SCFA producers and regulators of protein fermentation and complex oligosaccharides digestion within the gut [51,56].

We also highlighted ethyl acetone (2-hexanone) as highly abundant in the human feces [57], which was positively correlated with Eubacterium in NA individuals, while negatively in HRA individuals (Table 4 and Figure 8). Depletion of Eubacterium is noted in IBD individuals and those adopting a western diet of high intake of animal protein and fat and less intake of plant fiber [58]. Furthermore, ethyl acetone was positively correlated with Holdemanella, an Erysipelotrichaceae family member, in NA individuals, while negatively correlated with HRAs. Both associations need to be further investigated, since Holdemanella biformis has been suggested as an antitumorigenic SCFA-generator able to control CRC cell proliferation and intestinal metabolism [59,60,61].

P-cresol and indole were detected in all fecal samples tested, since they are the main fermentation products of the essential amino acids, phenylalanine and tyrosine (p-cresol) and tryptophan (indole). Both volatiles are also products of bacterial metabolism within the large intestine, and precursors of toxic metabolite compounds, referred to as uremic toxins, with p-cresol having a potential to contribute to genotoxicity and colorectal oncogenesis. Their concentration among fecal samples varies widely, as a result of dietary differences and protein intake from animal versus plant sources [62]. Both p-cresol and indole were positively correlated with Fusicatenibacter abundance in both subgroups of NA and HRA subjects, although p-cresol was overall more abundant than indole. This agrees with previous studies showing members of the Lachnospiraceae, Clostridiaceae, Eubacteriaceae, Peptostreptococcaceae, Enterobacteriaceae, Oscillospiraceae and Sutterellaceae families produce p-cresol and indole in culture [63,64], and associate with p-cresol and indole in human feces [62].

Despite the wealth of bacteriome and volatile correlations, no combinatorial biomarkers of CRC risk have been established [65,66]. Thus, microbial-metabolite signatures need to be further investigated to address the potentially toxic volatiles differentially produced by the dysbiotic microbiome of HRA individuals. One limitation of this and previous studies is that metagenomics and volatolomics analyses used fecal samples, reflecting the relative abundances of bacterial communities and VOCs at the lumen of the distal GI tract, without taking into consideration the other parts of the colon or the mucosa. Different levels of microbial abundances and volatiles emission derived in fecal matter cannot directly reflect the complex host-microbiota interactions taking place within the colonic mucosa of proximal and distal colon. Hence, studies sampling different sites along the colonic mucosa may provide a broader picture of the metagenomic and metabolomic milieu.