Ninety-three healthy adults were contacted prior to a scheduled colonoscopy in Fort Collins CO. Forty adult males and females provided written informed consent to collect a stool sample and an ascending and descending colon tissue biopsy. Twenty-four individuals (colon and stool) were assessed for non-targeted metabolomics. Eligible participants were provided a stool kit and study instructions. The gastroenterology clinical nurses and research staff confirmed study code number assignments and ensured the completed de-identification at the site of colonoscopy procedure. Colorado State University study personnel were contacted by clinic staff for sample retrieval immediately following procedure. Three study groups were BMI 20-24.9 for normal weight (n = 9), BMI 25-29.9 for overweight (n = 9) and BMI 30+ for obese (n = 6) adults. One normal weight female (BMI 24) had the right and left colon and stool sample applied for metataxonomic analysis (16S rRNA gene sequencing). Participant’s inclusion criteria for this study were at least 18 years of age, a scheduled routine colonoscopy, no prior history of colorectal cancer diagnosis, non-smoker, and not having taken antibiotics for at least one month prior to the standard of care, routine screening colonoscopy.

The colon tissue collected for this study was visually determined by the gastroenterologist performing the procedure to be normal, healthy tissue without polyps. Each participant had an about 5 mm biopsy of ascending (right) and descending (left) colon tissue and a self-collected stool sample prior to bowel preparation. Colonoscopy was completed by Centers for Gastroenterology-Fort Collins and University of Colorado-Health North Gastroenterology Clinic (Fort Collins, CO, United States). Samples were de-identified for personal information and study ID coded before storage and metabolite processing at Colorado State University. The number of polyps removed by the doctor with the respective location was provided following the procedure. This study received IRB approval, and include protocol number; Colorado State University IRB No. 15-6051, and University of Colorado Health IRB No. 0010144. Participants in this study had no history of diseases related to the liver or biliary tract and they did not have previous procedures such as cholecystectomy or ileal resections. This study did not collect information regarding the family history of colorectal cancer and did not perform hereditary genetic or epi-genetic screening history on the patients. Table 1 shows the study participant characteristics.

Stool samples were self-collected by participants in a pre-labeled study coded container and frozen at -80 °C. Approximately 5 mm of normal healthy colon tissue were stored immediately at -80 °C following collection. Samples were shipped on dry ice to Metabolon, Inc. (Durham, NC, United States) and a single participant sample underwent DNA extraction for metataxonomics.

Tissue and stool metabolite extraction was completed using 80% methanol as previously described[7], prior to ultrahigh performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS) as completed by Metabolon, Inc. Positive and negative ion modes were chosen to provide broad, non-targeted detection of metabolites.

Samples were extracted using the automated MicroLab STAR^® system from Hamilton Company. A set of recovery standards were added prior to the first step in the extraction process for quality control purposes. To remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for 2 min (Glen Mills GenoGrinder 2000) followed by centrifugation. The resulting extract was divided into five fractions: two for analysis by two separate reverse phase (RP)/UPLC-MS/MS methods with positive ion mode electrospray ionization (ESI), one for analysis by RP/UPLC-MS/MS with negative ion mode ESI, one for analysis by HILIC/UPLC-MS/MS with negative ion mode ESI, and one sample was reserved for backup. Samples were placed briefly on a TurboVap^® (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis.

The UPLC-MS/MS portion of the platform was based on a Waters ACQUITY ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-Exactive high resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35000 mass resolution. The protocol has been previously described by our lab[7].

Raw data were extracted, peak-identified, and processed using Metabolon’s hardware and software. Compounds were identified by comparison to library entries of purified standards or recurrent unknown entities as previously described[7].

Healthy colon tissue or stool metabolite profiles were semi-quantified in terms of relative abundance and median scaled to 1. Fold differences were calculated for normal weight, overweight and obese (colon tissue and stool) and for colon tissue between ascending and descending sites. A matched-pairs 2-way ANOVA was completed using the scaled relative abundance of each metabolite, experimental groups in ArrayStudio on log transformed data, were used for normal weight, overweight and obese. Metabolite profile distinctions between ascending and descending colon tissue were evaluated using P < 0.05 for statistical significance with matched pair t-test. An estimate of the false discovery rate (Q value) was calculated to account the multiple comparisons across metabolites that are typical of metabolomics-based studies with a Q value ≥ 0.01. A linear regression analyses for colon metabolites were preformed to compare the groups with polyp removal to no polyp removal, after adjusting for the effect of weight category of the subjects.

Principal component analysis and hierarchical clustering were applied to understand the similarities and differences between samples and/or groups of samples in a complex dataset. Unsupervised clustering was performed using the ward D2 method[20]. Random forest (RF) analysis, a supervised classification technique, was applied for identifying candidate biomarkers. To determine which variables (biochemicals) make the largest contribution to the classification of BMI, a “variable importance” measure was computed. We used the “Mean Decrease Accuracy” as this metric prediction accuracy[21].

DNA was extracted from colon tissue and stool with the MoBio PowerSoil Kit according to manufacturer protocols. Amplification of the V4 region of the 16S rRNA gene and amplicon sequencing followed the standards outlined by the Earth Microbiome Project. Raw FASTQ-formatted forward reads were imported into the Quantitative Insights Into Microbial Ecology 2 (QIIME 2) platform[22]. A feature table comprised of amplicon sequence variants (ASVs) was inferred from reads using the DADA2 algorithm[23]. Taxonomy was assigned to each representative ASV sequence using Naïve Bayes classifiers trained against 99% OTU reference collections from Greengenes 13_8 or SILVA 132. The raw feature table, representative sequences, and taxonomy tables were exported from QIIME 2 for further processing using R[24]. Following import, a master table comprised of ASV IDs with corresponding representative sequences, full and truncated Green-genes and SILVA taxonomic lineages, and absolute abundances for all ASVs within each sample was constructed. This master table served as the entry point for all downstream processing and analysis. Comparisons of microbiota composition proceeded from the compositional data analysis paradigm with count zero multiplicative replacement prior to applying the centred log-ratio (clr) transformation[25]. Taxon abundance are depicted as proportions (i.e., relative abundances). Supplemental Methods for additional details regarding amplification conditions, library preparation, sequencing, and a comprehensive account of analytical approaches.

Metataxonomics sequence data supporting the conclusions of this manuscript are available via NCBI SRA BioProject Accession PRJNA594611 and on this project’s GitHub repository located at github.com/kdprkr/ConjurersBrew, along with each of the materials needed to reproduce the analysis.

The stool metabolome of healthy adults classified according to BMI as normal weight, overweight or obese had a total of 842 named compounds (Supplemental Table 1). The 842 stool metabolites consisted of 175 amino acids, 26 peptides, 33 carbohydrates, 11 energy, 345 lipids, 46 nucleotides, 47 cofactors and vitamins, and 159 were classified as exogenous and referred to as xenobiotics. There were 98 stool metabolites that significantly differ according to BMI were 22 amino acids, 1 peptide, 8 nucleotides, 2 cofactors and vitamins, 9 xenobiotics and 56 lipids. Principal component analyses (PCA) for the stool metabolome (Supplemental Figure 1A) and hierarchal clustering (Supplemental Figure 1B) did not clearly separate participants by BMI groups. The 504 colon metabolites with known identity and 44 unnamed/unknown compounds from 24 male and female participants are provided in Supplemental Table 2. The colon metabolome contained 93 amino acids, 13 peptides, 35 carbohydrates, 10 metabolites were classified under TCA cycle and oxidative phosphorylation and there were 20 cofactors and vitamins. The largest portion of the colon tissue metabolome were lipids (about 50%, 262 lipid metabolites) that span 40 sub-metabolic pathways. Other notable small molecules from colon were 38 nucleotides and 32 exogenous, xenobiotic metabolites. Fourteen out of twenty-four healthy participants had 1-14 polyps removed during screening colonoscopy. A regression analysis for colon metabolites adjusting for the effect of weight was done, revealing 17 colon metabolites that had lower expression correlated with polyps removed (Supplemental Table 3). Figure 1A, shows 2-arachidonoylglycerol, 3-phosphoglycerate, and 6-phosphogluconate had a lower relative abundance in ascending and descending tissue from participants with polyps removed. Figure 1B-C, shows 1-dihomo-linolenylglycerol, aspartate, and glycerophosphorycholine (GPC) with lower expression in ascending colon tissue, while glutarate (C5-DC), and 2-hydroxyarachidate had lower expression in descending tissue from participants with polyps removed (Figures 1B and C).

Random forest analysis of colon tissue comparing normal weight vs overweight/obese yielded 30 metabolites with a predictive accuracy of 56% for the overweight/obese phenotype (Supplemental Figure 2A). BMI associated colon metabolites were bile acids and cofactors/vitamins (e.g., biliverdin, alpha-tocopherol, and pyridoxate). The predictive accuracy for metabolites in ascending versus descending colon was 73%, and 24 of the top 30 metabolites were classified as lipids (Supplemental Figure 2B). Table 2 further shows the 12 metabolites with statistically significant fold difference identified in right/left colon tissue and in stool by BMI comparisons. Figure 2 shows lipids that are significantly different between weight groups for colon tissue and stool (P < 0.05). The phospholipid, trimethylamine N-oxide (TMAO), was 2.80-fold difference from ascending in obese adults, and 6.23-fold difference from stool in overweight adults when compared to normal weight adults (Figure 2A). Endocannabinoids, linoleoylethanolamide (2.11-fold) and oleoylethanolamide (1.60-fold) difference from ascending in normal weight adults and decrease in stool of obese adults (Figure 2B). Furthermore, the median scaled relative abundance of primary and secondary bile acids (chenodeoxycholate 2.89-fold and 0.41-fold, cholate 0.55-fold, and taurodeoxycholate 1.49-fold) had significant differences by colon location and stool (Figure 2C). The primary bile acid, chenodeoxycholate is 2.89-fold difference from ascending in normal weight and 0.41-fold difference from descending in overweight adults, while 21.80-fold difference in stool of overweight adults when compared to normal weight. Cholate is 0.55-fold difference from obese descending colon and 61.06-fold difference from overweight stool when compared to normal weight (Figure 2C). The secondary bile acid, taurodeoxycholate, had 1.49-fold difference from ascending colon in overweight adults and 9.32-fold difference in stool of overweight adults compared to normal weight (Table 2 and Figure 2C).

We next utilized metabolomics to distinguish ascending and descending colon tissue. There were 87 metabolites with statistically significant fold differences between ascending vs descending colon by BMI. These included, 46 metabolites in normal weight, 37 metabolites in overweight, and 28 metabolites in obese. Notably, the number of metabolites distinguishing ascending and descending colon decreased as BMI increased. We found that 62% of the metabolites distinguishing ascending and descending colon were lipids. Table 3 shows 54 colon tissue lipids with statistically significant fold difference between ascending and descending colon. In normal weight adults there were 29 colon lipids, 24 of which were fatty acids and lysophospholipid (1.48-fold – 2.16-fold difference) from ascending, and 5 that were from descending colon. Overweight adults had 24 significantly different colon lipids; 11 metabolites from ascending, (8 derived from monoacylglycerols 1.67-fold – 2.61-fold), and 13 metabolites with higher abundance from descending tissue (Table 3). Obese adults had the fewest significant differences in colon lipids between ascending and descending tissue (15 identified metabolites). Table 3 shows 4 metabolites increased in ascending tissue and 11 increased in descending tissue that were primarily fatty acids. Figure 3 shows median scaled relative abundance of right and left colon tissue lipids, including those that are food derived long chain fatty acids and microbiome-products. The long chain fatty acids; palmitate 1.35-fold, arachidate 1.35-fold, and stearate 1.39-fold from ascending tissue in normal weight, and palmitate 0.73-fold, arachidate 0.69-fold, and stearate 0.68-fold from descending tissue in obese adults (Figure 3A). Figure 3B shows microbiome-influenced metabolites that show significant fold difference in colon; 15-HETE from ascending tissue is 1.69-fold in normal weight and 1.93-fold in obese while, 3-hydroxybutyrate is 0.48-fold from descending in normal weight, and 2-archidonoylglycerol is 1.68-fold from ascending tissue in overweight adults (Figure 3B). This study suggests as BMI increases lipid diversity decrease in the colon and primarily in the ascending colon.

To explore associations of taxonomic groups with ascending colon, descending colon, and stool in a healthy weight adult female, we constructed a compositional PCA biplot from centred log-ratio transformed relative abundances. We observed marked separation across all three sample types at the phylum-, family-, and genus-levels. (Figure 4, Supplementary Table 4). Differences between ascending and descending colon were driven by increased abundance for several taxa in the Firmicutes phylum, including Anaerostipes, Blautia, Dorea, and Fusicatenibacter (all members of the Lachnospiraceae family), as well as Streptococcus and Romboutsia (members of the Streptococcaceae and Peptostreptococcaceae families, respectively). Stool samples were also differentiated by the genus Bifidobacterium (a member of the phylum Actinobacteria). Comparisons of the composition between colon samples (ascending versus descending) indicated enrichments for Bacteroides, Ruminiclostridium 9, Ruminococcus gnavus, and Tyzzerella in the ascending colon, while the descending colon harbored more Barnesiella, Faecalibacterium, Parabacteroides, Parasutterella, and Roseburia (Figure 4C and Supplemental Table 4).

Obesity is a risk factor for colorectal cancer, yet metabolic distinctions between healthy right and left colon tissue, before cancer is diagnosed, remains largely unknown.

Colon cancer of the ascending colon has a poorer prognosis and survival when compared to colon cancer on the descending colon. Stool metabolite composition does not accurately reflect proximal/ascending/right sided colon. Development of healthy colon tissue small molecule signatures for ascending and descending colon will aid in our understanding of how to improve gut metabolism and may help prevent or mitigate colorectal cancer risk.

This study compared right-ascending and left-descending colon tissue metabolomes and sought to identify differences from the stool metabolome in normal weight, overweight, and obese adults.

Global, non-targeted metabolomics was applied to assess right-ascending and left-descending colon tissue collected from healthy adults undergoing screening colonoscopies to test the hypothesis that body mass index (BMI) differentially impacts colon tissue metabolite profiles. The colon tissue and stool metabolome of healthy adults was analyzed for metabolite signatures and metabolic pathway networks implicated in progression and prevention of colorectal cancer.

This is the first report using metabolomics to compare the right-ascending vs left--descending colon tissue of healthy adults. Our findings show that BMI was associated with metabolite profile differences between the ascending and descending colon. Disturbances in multiple metabolic pathways of the right and left colon from being overweight/obese may have important implications for increasing colorectal cancer risk.

There were metabolite profile differences between right-ascending and left-descending colon tissue in healthy adults receiving routine, screening colonoscopies. BMI impacts the number, type and magnitude of metabolite differences between the ascending and descending colon. Colon lipids and other metabolites in obese and overweight adults were distinguished from normal weight participants and associated with gut inflammation, nutrient absorption, and products of microbiota metabolism.

Right and left colon tissue metabolites that differ in relative abundance between normal weight, overweight, obese adults may be sensitive biomarkers for colon cancer risk. Diet and lifestyle influence right and left sided colon tissue metabolite composition that shape inflammation and cancer risk in overweight and obese adults. Development of healthy colon tissue small molecule signatures for ascending and descending colon will aid in our understanding of how to improve gut metabolism and may help prevent or mitigate colorectal cancer risk.

Metataxonomics sequence data supporting the conclusions of this manuscript are available via NCBI SRA BioProject Accession PRJNA594611 and on this project’s GitHub repository located at github.com/kdprkr/ConjurersBrew, along with each of the materials needed to reproduce the analysis.

Metataxonomics sequence data are available via NCBI SRA BioProject Accession PRJNA594611 and on this project’s GitHub repository github.com/kdprkr/ConjurersBrew.