High-Fructose Diet Alters Intestinal Microbial Profile and Correlates with Early Tumorigenesis in a Mouse Model of Barrett’s Esophagus
Abstract
:1. Introduction
2. Materials and Methods
2.1. Animals
2.2. Tissue Preparation and Disease Evaluation
2.3. Microbiome Analysis
2.4. Metabolic Analysis
2.5. Immune Cell Analysis via Flow Cytometry
2.6. Gene Expression Analysis
2.7. Statistics
3. Results
3.1. HFrD Accelerates Dysplasia by Increasing Stem Cell Expansion While Decreasing Mucus Production
3.2. HFrD Decreases Gut Bacterial Richness with Increased Firmicutes and Decreased Akkermansia Abundance
3.3. HFrD Alters Colonic and Hepatic Subtrate Utilization, Changes Metabolic Profile of the Host, and Decreases Protective SCFA Concentrations in the Gut and Serum
3.4. Neutrophil and γδ T Cell Infiltration at the SCJ Tissue of HFrD-Fed Mice Increased with Age
3.5. HFrD Promotes Changes in Tissue Reconstruction, Immune Network, and Metabolic Pathway
4. Discussion
4.1. Diet-Related Alteration of Microbial Profile and Stability Indicates Impaired Gut Health and Correlates with Inflammation and Disease
4.2. No Weight Gain Was Detected in HFrD Mice despite Increased Colonic Gluconeogenesis and Hepatic De Novo Lipogenesis
4.3. The Metabolic Profile of HFrD Mice Suggests a Diet-Related Alteration of Substrate Utilization by Host and Microbiome, as Well as a Limited Effect on Systemic Inflammation and Moderate Acceleration of the Dysplasia Phenotype
4.4. Levels of Protectively Acting SCFA Are Decreased in HFrD
4.5. HFrD-Fed Mice Present a Moderate Inflammatory Phenotype in Comparison with HFD-Fed Mice
4.6. Tissue Remodeling, Metabolism, and Gut Barrier Protection Gene Sets Are Enriched in HFrD Mice While Pro-Oncogenic Gene Sets Are Enriched in CD Mice
5. Conclusions
Supplementary Materials
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Stepien, M.; Chajes, V.; Romieu, I. The role of diet in cancer: The epidemiologic link. Salud Publica Mex 2016, 58, 261–273. [Google Scholar] [CrossRef] [Green Version]
- Yerlikaya, A.; Dagel, T.; King, C.; Kuwabara, M.; Lanaspa, M.A.; Andres-Hernando, A.; Covic, A.; Manitius, J.; Sag, A.A.; Kanbay, M. Dietary and commercialized fructose: Sweet or sour? Int. Urol. Nephrol. 2017, 49, 1611–1620. [Google Scholar] [CrossRef]
- Quante, M.; Bhagat, G.; Abrams, J.A.; Marache, F.; Good, P.; Lee, M.D.; Lee, Y.; Friedman, R.; Asfaha, S.; Dubeykovskaya, Z.; et al. Bile acid and inflammation activate gastric cardia stem cells in a mouse model of Barrett-like metaplasia. Cancer Cell 2012, 21, 36–51. [Google Scholar] [CrossRef] [Green Version]
- Quante, M.; Abrams, J.A.; Lee, Y.; Wang, T.C. Barrett esophagus: What a mouse model can teach us about human disease. Cell Cycle 2012, 11, 4328–4338. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Gerson, L.B.; Shetler, K.; Triadafilopoulos, G. Prevalence of Barrett’s esophagus in asymptomatic individuals. Gastroenterology 2002, 123, 461–467. [Google Scholar] [CrossRef]
- Falk, G.W. Barrett’s esophagus. Gastroenterology 2002, 122, 1569–1591. [Google Scholar] [CrossRef]
- Winters, C., Jr.; Spurling, T.J.; Chobanian, S.J.; Curtis, D.J.; Esposito, R.L.; Hacker, J.F., III; Johnson, D.A.; Cruess, D.F.; Cotelingam, J.D.; Gurney, M.S.; et al. Barrett’s esophagus: A prevalent, occult complication of gastroesophageal reflux disease. Gastroenterology 1987, 92, 118–124. [Google Scholar] [CrossRef]
- Münch, N.S.; Fang, H.Y.; Ingermann, J.; Maurer, H.C.; Anand, A.; Kellner, V.; Sahm, V.; Wiethaler, M.; Baumeister, T.; Wein, F.; et al. High-Fat Diet Accelerates Carcinogenesis in a Mouse Model of Barrett’s Esophagus via Interleukin 8 and Alterations to the Gut Microbiome. Gastroenterology 2019, 157, 492–506.e492. [Google Scholar] [CrossRef] [Green Version]
- Sun, L.-P.; Yan, L.-B.; Liu, Z.-Z.; Zhao, W.-J.; Zhang, C.-X.; Chen, Y.-M.; Lao, X.Q.; Liu, X. Dietary factors and risk of mortality among patients with esophageal cancer: A systematic review. BMC Cancer 2020, 20, 287. [Google Scholar] [CrossRef] [PubMed]
- Miles, F.L.; Chang, S.-C.; Morgenstern, H.; Tashkin, D.; Rao, J.-Y.; Cozen, W.; Mack, T.; Lu, Q.-Y.; Zhang, Z.-F. Association of sugary beverages with survival among patients with cancers of the upper aerodigestive tract. Cancer Causes Control. 2016, 27, 1293–1300. [Google Scholar] [CrossRef]
- Port, A.; Ruth, M.; Istfan, N. Fructose consumption and cancer: Is there a connection? Curr. Opin. Endocrinol. Diabetes Obes. 2012, 19, 367–374. [Google Scholar] [CrossRef]
- Liu, H.; Huang, D.; McArthur, D.L.; Boros, L.G.; Nissen, N.; Heaney, A.P. Fructose induces transketolase flux to promote pancreatic cancer growth. Cancer Res. 2010, 70, 6368–6376. [Google Scholar] [CrossRef] [Green Version]
- Vizán, P.; Alcarraz-Vizán, G.; Díaz-Moralli, S.; Solovjeva, O.N.; Frederiks, W.M.; Cascante, M. Modulation of pentose phosphate pathway during cell cycle progression in human colon adenocarcinoma cell line HT29. Int. J. Cancer 2009, 124, 2789–2796. [Google Scholar] [CrossRef]
- Montrose, D.C.; Nishiguchi, R.; Basu, S.; Staab, H.A.; Zhou, X.K.; Wang, H.; Meng, L.; Johncilla, M.; Cubillos-Ruiz, J.R.; Morales, D.K. Dietary fructose alters the composition, localization, and metabolism of gut microbiota in association with worsening colitis. Cell. Mol. Gastroenterol. Hepatol. 2021, 11, 525–550. [Google Scholar] [CrossRef] [PubMed]
- Lambertz, J.; Weiskirchen, S.; Landert, S.; Weiskirchen, R. Fructose: A Dietary Sugar in Crosstalk with Microbiota Contributing to the Development and Progression of Non-Alcoholic Liver Disease. Front. Immunol. 2017, 8, 1159. [Google Scholar] [CrossRef] [Green Version]
- Klindworth, A.; Pruesse, E.; Schweer, T.; Peplies, J.; Quast, C.; Horn, M.; Glöckner, F.O. Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Res. 2013, 41, e1. [Google Scholar] [CrossRef] [PubMed]
- Lagkouvardos, I.; Joseph, D.; Kapfhammer, M.; Giritli, S.; Horn, M.; Haller, D.; Clavel, T. IMNGS: A comprehensive open resource of processed 16S rRNA microbial profiles for ecology and diversity studies. Sci. Rep. 2016, 6, 33721. [Google Scholar] [CrossRef] [PubMed]
- Edgar, R.C.; Haas, B.J.; Clemente, J.C.; Quince, C.; Knight, R. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 2011, 27, 2194–2200. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Edgar, R.C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 2010, 26, 2460–2461. [Google Scholar] [CrossRef] [Green Version]
- Edgar, R.C. UNOISE2: Improved error-correction for Illumina 16S and ITS amplicon sequencing. bioRxiv 2016, 081257. [Google Scholar] [CrossRef] [Green Version]
- Quast, C.; Pruesse, E.; Yilmaz, P.; Gerken, J.; Schweer, T.; Yarza, P.; Peplies, J.; Glöckner, F.O. The SILVA ribosomal RNA gene database project: Improved data processing and web-based tools. Nucleic Acids Res. 2013, 41, D590–D596. [Google Scholar] [CrossRef]
- Yoon, S.H.; Ha, S.M.; Kwon, S.; Lim, J.; Kim, Y.; Seo, H.; Chun, J. Introducing EzBioCloud: A taxonomically united database of 16S rRNA gene sequences and whole-genome assemblies. Int. J. Syst. Evol. Microbiol. 2017, 67, 1613–1617. [Google Scholar] [CrossRef] [PubMed]
- Douglas, G.M.; Maffei, V.J.; Zaneveld, J.R.; Yurgel, S.N.; Brown, J.R.; Taylor, C.M.; Huttenhower, C.; Langille, M.G.I. PICRUSt2 for prediction of metagenome functions. Nat. Biotechnol. 2020, 38, 685–688. [Google Scholar] [CrossRef] [PubMed]
- Fernandes, A.D.; Macklaim, J.M.; Linn, T.G.; Reid, G.; Gloor, G.B. ANOVA-Like Differential Expression (ALDEx) Analysis for Mixed Population RNA-Seq. PLoS ONE 2013, 8, e67019. [Google Scholar] [CrossRef] [PubMed]
- Gloor, G.B.; Macklaim, J.M.; Fernandes, A.D. Displaying Variation in Large Datasets: Plotting a Visual Summary of Effect Sizes. J. Comput. Graph. Stat. 2016, 25, 971–979. [Google Scholar] [CrossRef]
- Amin, H.M.; Hashem, A.M.; Ashour, M.S.; Hatti-Kaul, R. 1,2 Propanediol utilization by Lactobacillus reuteri DSM 20016, role in bioconversion of glycerol to 1,3 propanediol, 3-hydroxypropionaldehyde and 3-hydroxypropionic acid. J. Genet. Eng. Biotechnol. 2013, 11, 53–59. [Google Scholar] [CrossRef] [Green Version]
- Segata, N.; Izard, J.; Waldron, L.; Gevers, D.; Miropolsky, L.; Garrett, W.S.; Huttenhower, C. Metagenomic biomarker discovery and explanation. Genome Biol. 2011, 12, R60. [Google Scholar] [CrossRef] [Green Version]
- Han, J.; Lin, K.; Sequeira, C.; Borchers, C.H. An isotope-labeled chemical derivatization method for the quantitation of short-chain fatty acids in human feces by liquid chromatography–tandem mass spectrometry. Anal. Chim. Acta 2015, 854, 86–94. [Google Scholar] [CrossRef] [PubMed]
- Smyth, G.K. Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. 2004, 3. [Google Scholar] [CrossRef]
- Wishart, D.S.; Tzur, D.; Knox, C.; Eisner, R.; Guo, A.C.; Young, N.; Cheng, D.; Jewell, K.; Arndt, D.; Sawhney, S.; et al. HMDB: The Human Metabolome Database. Nucleic Acids Res. 2007, 35, D521–D526. [Google Scholar] [CrossRef]
- Tsugawa, H.; Cajka, T.; Kind, T.; Ma, Y.; Higgins, B.; Ikeda, K.; Kanazawa, M.; VanderGheynst, J.; Fiehn, O.; Arita, M. MS-DIAL: Data-independent MS/MS deconvolution for comprehensive metabolome analysis. Nat. Methods 2015, 12, 523–526. [Google Scholar] [CrossRef] [PubMed]
- Lai, Z.; Tsugawa, H.; Wohlgemuth, G.; Mehta, S.; Mueller, M.; Zheng, Y.; Ogiwara, A.; Meissen, J.; Showalter, M.; Takeuchi, K.; et al. Identifying metabolites by integrating metabolome databases with mass spectrometry cheminformatics. Nat. Methods 2018, 15, 53–56. [Google Scholar] [CrossRef] [PubMed]
- Dührkop, K.; Fleischauer, M.; Ludwig, M.; Aksenov, A.A.; Melnik, A.V.; Meusel, M.; Dorrestein, P.C.; Rousu, J.; Böcker, S. SIRIUS 4: A rapid tool for turning tandem mass spectra into metabolite structure information. Nat. Methods 2019, 16, 299–302. [Google Scholar] [CrossRef] [Green Version]
- Pelaseyed, T.; Bergström, J.H.; Gustafsson, J.K.; Ermund, A.; Birchenough, G.M.H.; Schütte, A.; van der Post, S.; Svensson, F.; Rodríguez-Piñeiro, A.M.; Nyström, E.E.L.; et al. The mucus and mucins of the goblet cells and enterocytes provide the first defense line of the gastrointestinal tract and interact with the immune system. Immunol. Rev. 2014, 260, 8–20. [Google Scholar] [CrossRef] [Green Version]
- Hooper, L.V. Chapter 3—Epithelial Cell Contributions to Intestinal Immunity. In Advances in Immunology; Alt, F.W., Ed.; Academic Press: Cambridge, MA, USA, 2015; Volume 126, pp. 129–172. [Google Scholar]
- Schellnegger, R.; Quante, A.; Rospleszcz, S.; Schernhammer, M.; Hohl, B.; Tobiasch, M.; Pastula, A.; Brandtner, A.; Abrams, J.A.; Strauch, K.; et al. Goblet Cell Ratio in Combination with Differentiation and Stem Cell Markers in Barrett Esophagus Allow Distinction of Patients with and without Esophageal Adenocarcinoma. Cancer Prev. Res. 2017, 10, 55–66. [Google Scholar] [CrossRef] [Green Version]
- Du, L.; Heaney, A.P. Regulation of adipose differentiation by fructose and GluT5. Mol. Endocrinol. 2012, 26, 1773–1782. [Google Scholar] [CrossRef]
- Chen, W.L.; Wang, Y.Y.; Zhao, A.; Xia, L.; Xie, G.; Su, M.; Zhao, L.; Liu, J.; Qu, C.; Wei, R.; et al. Enhanced Fructose Utilization Mediated by SLC2A5 Is a Unique Metabolic Feature of Acute Myeloid Leukemia with Therapeutic Potential. Cancer Cell 2016, 30, 779–791. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Wang, Q.; Li, F.; Liang, B.; Liang, Y.; Chen, S.; Mo, X.; Ju, Y.; Zhao, H.; Jia, H.; Spector, T.D.; et al. A metagenome-wide association study of gut microbiota in asthma in UK adults. BMC Microbiol. 2018, 18, 114. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Merino, B.; Fernández-Díaz, C.M.; Cózar-Castellano, I.; Perdomo, G. Intestinal Fructose and Glucose Metabolism in Health and Disease. Nutrients 2019, 12, 94. [Google Scholar] [CrossRef] [Green Version]
- Tap, J.; Furet, J.P.; Bensaada, M.; Philippe, C.; Roth, H.; Rabot, S.; Lakhdari, O.; Lombard, V.; Henrissat, B.; Corthier, G.; et al. Gut microbiota richness promotes its stability upon increased dietary fibre intake in healthy adults. Environ. Microbiol. 2015, 17, 4954–4964. [Google Scholar] [CrossRef]
- Hou, Y.; Wei, W.; Guan, X.; Liu, Y.; Bian, G.; He, D.; Fan, Q.; Cai, X.; Zhang, Y.; Wang, G.; et al. A diet-microbial metabolism feedforward loop modulates intestinal stem cell renewal in the stressed gut. Nat. Commun. 2021, 12, 271. [Google Scholar] [CrossRef]
- Rinninella, E.; Raoul, P.; Cintoni, M.; Franceschi, F.; Miggiano, G.A.D.; Gasbarrini, A.; Mele, M.C. What is the Healthy Gut Microbiota Composition? A Changing Ecosystem across Age, Environment, Diet, and Diseases. Microorganisms 2019, 7, 14. [Google Scholar] [CrossRef] [Green Version]
- Wahlström, A.; Sayin, S.I.; Marschall, H.-U.; Bäckhed, F. Intestinal crosstalk between bile acids and microbiota and its impact on host metabolism. Cell Metab. 2016, 24, 41–50. [Google Scholar] [CrossRef] [Green Version]
- Wlodarska, M.; Willing, B.P.; Bravo, D.M.; Finlay, B.B. Phytonutrient diet supplementation promotes beneficial Clostridia species and intestinal mucus secretion resulting in protection against enteric infection. Sci. Rep. 2015, 5, 9253. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Hudcovic, T.; Kolinska, J.; Klepetar, J.; Stepankova, R.; Rezanka, T.; Srutkova, D.; Schwarzer, M.; Erban, V.; Du, Z.; Wells, J.M.; et al. Protective effect of Clostridium tyrobutyricum in acute dextran sodium sulphate-induced colitis: Differential regulation of tumour necrosis factor-α and interleukin-18 in BALB/c and severe combined immunodeficiency mice. Clin. Exp. Immunol. 2012, 167, 356–365. [Google Scholar] [CrossRef] [PubMed]
- O’Callaghan, J.; O’Toole, P.W. Lactobacillus: Host-microbe relationships. Curr. Top. Microbiol. Immunol. 2013, 358, 119–154. [Google Scholar] [CrossRef] [PubMed]
- Heeney, D.D.; Gareau, M.G.; Marco, M.L. Intestinal Lactobacillus in health and disease, a driver or just along for the ride? Curr. Opin. Biotechnol. 2018, 49, 140–147. [Google Scholar] [CrossRef] [Green Version]
- Borges-Canha, M.; Portela-Cidade, J.P.; Dinis-Ribeiro, M.; Leite-Moreira, A.F.; Pimentel-Nunes, P. Role of colonic microbiota in colorectal carcinogenesis: A systematic review. Rev. Esp. Enferm. Dig. 2015, 107, 659–671. [Google Scholar] [CrossRef] [Green Version]
- Hieken, T.J.; Chen, J.; Hoskin, T.L.; Walther-Antonio, M.; Johnson, S.; Ramaker, S.; Xiao, J.; Radisky, D.C.; Knutson, K.L.; Kalari, K.R.; et al. The Microbiome of Aseptically Collected Human Breast Tissue in Benign and Malignant Disease. Sci. Rep. 2016, 6, 30751. [Google Scholar] [CrossRef]
- Zhou, K. Strategies to promote abundance of Akkermansia muciniphila, an emerging probiotics in the gut, evidence from dietary intervention studies. J. Funct. Foods 2017, 33, 194–201. [Google Scholar] [CrossRef] [PubMed]
- Parada Venegas, D.; De la Fuente, M.K.; Landskron, G.; González, M.J.; Quera, R.; Dijkstra, G.; Harmsen, H.J.; Faber, K.N.; Hermoso, M.A. Short chain fatty acids (SCFAs)-mediated gut epithelial and immune regulation and its relevance for inflammatory bowel diseases. Front. Immunol. 2019, 10, 277. [Google Scholar] [CrossRef] [Green Version]
- Bates, C.J. NIACIN*. In Encyclopedia of Human Nutrition, 2nd ed.; Caballero, B., Ed.; Elsevier: Oxford, UK, 1998; pp. 253–259. [Google Scholar]
- Thien, K.R.; Blair, J.A.; Leeming, R.J.; Cooke, W.T.; Melikian, V. Serum folates in man. J. Clin. Pathol. 1977, 30, 438–448. [Google Scholar] [CrossRef] [Green Version]
- Lucock, M. Folic acid: Nutritional biochemistry, molecular biology, and role in disease processes. Mol. Genet. Metab 2000, 71, 121–138. [Google Scholar] [CrossRef] [PubMed]
- Dam, H. Hæmorrhages in Chicks Reared on Artificial Diets: A New Deficiency Disease. Nature 1934, 133, 909–910. [Google Scholar] [CrossRef]
- Ferland, G. The Discovery of Vitamin K and Its Clinical Applications. Ann. Nutr. Metab. 2012, 61, 213–218. [Google Scholar] [CrossRef]
- Xv, F.; Chen, J.; Duan, L.; Li, S. Research progress on the anticancer effects of vitamin K2. Oncol. Lett. 2018, 15, 8926–8934. [Google Scholar] [CrossRef] [Green Version]
- Nimptsch, K.; Rohrmann, S.; Kaaks, R.; Linseisen, J. Dietary vitamin K intake in relation to cancer incidence and mortality: Results from the Heidelberg cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC-Heidelberg). Am. J. Clin. Nutr. 2010, 91, 1348–1358. [Google Scholar] [CrossRef] [Green Version]
- Khedkar, G.D.; Prakash, B.; Khedkar, C.D.; Chopade, B.A. Nucleic Acids. In Encyclopedia of Food and Health; Caballero, B., Finglas, P.M., Toldrá, F., Eds.; Academic Press: Oxford, UK, 2016; pp. 84–92. [Google Scholar]
- Chen, W.-H.; Lu, G.; Bork, P.; Hu, S.; Lercher, M.J. Energy efficiency trade-offs drive nucleotide usage in transcribed regions. Nat. Commun. 2016, 7, 11334. [Google Scholar] [CrossRef] [Green Version]
- Vernocchi, P.; Del Chierico, F.; Quagliariello, A.; Ercolini, D.; Lucidi, V.; Putignani, L. A Metagenomic and in Silico Functional Prediction of Gut Microbiota Profiles May Concur in Discovering New Cystic Fibrosis Patient-Targeted Probiotics. Nutrients 2017, 9, 1342. [Google Scholar] [CrossRef] [Green Version]
- Silbernagel, G.; Lütjohann, D.; Machann, J.; Meichsner, S.; Kantartzis, K.; Schick, F.; Häring, H.-U.; Stefan, N.; Fritsche, A. Cholesterol synthesis is associated with hepatic lipid content and dependent on fructose/glucose intake in healthy humans. Exp. Diabetes Res. 2012, 2012, 361863. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Zubiría, M.G.; Alzamendi, A.; Moreno, G.; Rey, M.A.; Spinedi, E.; Giovambattista, A. Long-Term Fructose Intake Increases Adipogenic Potential: Evidence of Direct Effects of Fructose on Adipocyte Precursor Cells. Nutrients 2016, 8, 198. [Google Scholar] [CrossRef] [Green Version]
- Millard, P.; Pérochon, J.; Létisse, F. Functional Analysis of Deoxyhexose Sugar Utilization in Escherichia coli Reveals Fermentative Metabolism under Aerobic Conditions. Appl. Environ. Microbiol. 2021, 87, e0071921. [Google Scholar] [CrossRef] [PubMed]
- White, N.J.; Ward, K.R.; Pati, S.; Strandenes, G.; Cap, A.P. Hemorrhagic blood failure: Oxygen debt, coagulopathy, and endothelial damage. J. Trauma Acute Care Surg. 2017, 82, S41–S49. [Google Scholar] [CrossRef] [PubMed]
- MacIntyre, N.R. Tissue hypoxia: Implications for the respiratory clinician. Respir. Care 2014, 59, 1590–1596. [Google Scholar] [CrossRef]
- Klekner, V.; Löbl, V.; Šímová, E.; Novák, M. Conversion of disaccharides to 3-ketodisaccharides by nongrowing and immobilized cells ofAgrobacterium tumefaciens. Folia Microbiol. 1989, 34, 286–293. [Google Scholar] [CrossRef]
- Flint, H.J.; Scott, K.P.; Duncan, S.H.; Louis, P.; Forano, E. Microbial degradation of complex carbohydrates in the gut. Gut Microbes 2012, 3, 289–306. [Google Scholar] [CrossRef] [Green Version]
- Gerschenson, L.N.; Rojas, A.M.; Fissore, E.N. Chapter 3—Carbohydrates. In Nutraceutical and Functional Food Components; Galanakis, C.M., Ed.; Academic Press: Cambridge, MA, USA, 2017; pp. 39–101. [Google Scholar]
- Gray, G.M. Starch digestion and absorption in nonruminants. J. Nutr. 1992, 122, 172–177. [Google Scholar] [CrossRef]
- Hay, N. Reprogramming glucose metabolism in cancer: Can it be exploited for cancer therapy? Nat. Rev. Cancer 2016, 16, 635–649. [Google Scholar] [CrossRef] [Green Version]
- Damerow, S.; Hoppe, C.; Bandini, G.; Zarnovican, P.; Buettner, F.F.R.; Lüder, C.G.K.; Ferguson, M.A.J.; Routier, F.H. Depletion of UDP-Glucose and UDP-Galactose Using a Degron System Leads to Growth Cessation of Leishmania major. PLoS Negl. Trop. Dis. 2015, 9, e0004205. [Google Scholar] [CrossRef] [Green Version]
- Vogt, J.A.; Pencharz, P.B.; Wolever, T.M.S. l-Rhamnose increases serum propionate in humans. Am. J. Clin. Nutr. 2004, 80, 89–94. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Vogt, J.A.; Ishii-Schrade, K.B.; Pencharz, P.B.; Wolever, T.M. L-Rhamnose increases serum propionate after long-term supplementation, but lactulose does not raise serum acetate. Am. J. Clin. Nutr. 2004, 80, 1254–1261. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Baldi, S.; Menicatti, M.; Nannini, G.; Niccolai, E.; Russo, E.; Ricci, F.; Pallecchi, M.; Romano, F.; Pedone, M.; Poli, G.; et al. Free Fatty Acids Signature in Human Intestinal Disorders: Significant Association between Butyric Acid and Celiac Disease. Nutrients 2021, 13, 742. [Google Scholar] [CrossRef]
- Al-Lahham, S.a.H.; Peppelenbosch, M.P.; Roelofsen, H.; Vonk, R.J.; Venema, K. Biological effects of propionic acid in humans; metabolism, potential applications and underlying mechanisms. Biochim. Biophys. Acta Mol. Cell Biol. Lipids 2010, 1801, 1175–1183. [Google Scholar] [CrossRef] [PubMed]
- Van Immerseel, F.; Ducatelle, R.; De Vos, M.; Boon, N.; Van De Wiele, T.; Verbeke, K.; Rutgeerts, P.; Sas, B.; Louis, P.; Flint, H.J. Butyric acid-producing anaerobic bacteria as a novel probiotic treatment approach for inflammatory bowel disease. J. Med. Microbiol. 2010, 59, 141–143. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Niccolai, E.; Baldi, S.; Ricci, F.; Russo, E.; Nannini, G.; Menicatti, M.; Poli, G.; Taddei, A.; Bartolucci, G.; Calabrò, A.S.; et al. Evaluation and comparison of short chain fatty acids composition in gut diseases. World J. Gastroenterol. 2019, 25, 5543–5558. [Google Scholar] [CrossRef]
- Heimann, E.; Nyman, M.; Pålbrink, A.-K.; Lindkvist-Petersson, K.; Degerman, E. Branched short-chain fatty acids modulate glucose and lipid metabolism in primary adipocytes. Adipocyte 2016, 5, 359–368. [Google Scholar] [CrossRef] [PubMed]
- Nilsson, A.; Johansson, E.; Ekström, L.; Björck, I. Effects of a brown beans evening meal on metabolic risk markers and appetite regulating hormones at a subsequent standardized breakfast: A randomized cross-over study. PLoS ONE 2013, 8, e59985. [Google Scholar] [CrossRef] [Green Version]
- Gio-Batta, M.; Sjöberg, F.; Jonsson, K.; Barman, M.; Lundell, A.-C.; Adlerberth, I.; Hesselmar, B.; Sandberg, A.-S.; Wold, A.E. Fecal short chain fatty acids in children living on farms and a link between valeric acid and protection from eczema. Sci. Rep. 2020, 10, 22449. [Google Scholar] [CrossRef] [PubMed]
- Kondratova, M.; Czerwinska, U.; Sompairac, N.; Amigorena, S.D.; Soumelis, V.; Barillot, E.; Zinovyev, A.; Kuperstein, I. A multiscale signalling network map of innate immune response in cancer reveals cell heterogeneity signatures. Nat. Commun. 2019, 10, 4808. [Google Scholar] [CrossRef] [Green Version]
- Mirzaei, H.R.; Mirzaei, H.; Lee, S.Y.; Hadjati, J.; Till, B.G. Prospects for chimeric antigen receptor (CAR) γδ T cells: A potential game changer for adoptive T cell cancer immunotherapy. Cancer Lett. 2016, 380, 413–423. [Google Scholar] [CrossRef] [Green Version]
- Porta, C.; Paglino, C.; Mosca, A. Targeting PI3K/Akt/mTOR Signaling in Cancer. Front. Oncol. 2014, 4, 64. [Google Scholar] [CrossRef] [Green Version]
- Kunze, B.; Wein, F.; Fang, H.Y.; Anand, A.; Baumeister, T.; Strangmann, J.; Gerland, S.; Ingermann, J.; Munch, N.S.; Wiethaler, M.; et al. Notch Signaling Mediates Differentiation in Barrett’s Esophagus and Promotes Progression to Adenocarcinoma. Gastroenterology 2020. [Google Scholar] [CrossRef]
- Harris, S.L.; Levine, A.J. The p53 pathway: Positive and negative feedback loops. Oncogene 2005, 24, 2899–2908. [Google Scholar] [CrossRef] [Green Version]
- Saxton, R.A.; Sabatini, D.M. mTOR Signaling in Growth, Metabolism, and Disease. Cell 2017, 168, 960–976. [Google Scholar] [CrossRef] [Green Version]
- Kalluri, R.; Weinberg, R.A. The basics of epithelial-mesenchymal transition. J. Clin. Investig. 2009, 119, 1420–1428. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Xu, F.; Na, L.; Li, Y.; Chen, L. Roles of the PI3K/AKT/mTOR signalling pathways in neurodegenerative diseases and tumours. Cell Biosci. 2020, 10, 54. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Winkler, J.; Abisoye-Ogunniyan, A.; Metcalf, K.J.; Werb, Z. Concepts of extracellular matrix remodelling in tumour progression and metastasis. Nat. Commun. 2020, 11, 5120. [Google Scholar] [CrossRef] [PubMed]
- Gupta, S.; Takebe, N.; Lorusso, P. Targeting the Hedgehog pathway in cancer. Ther. Adv. Med. Oncol. 2010, 2, 237–250. [Google Scholar] [CrossRef] [Green Version]
- Liu, R.; Li, X.; Hylemon, P.B.; Zhou, H. Conjugated Bile Acids Promote Invasive Growth of Esophageal Adenocarcinoma Cells and Cancer Stem Cell Expansion via Sphingosine 1-Phosphate Receptor 2-Mediated Yes-Associated Protein Activation. Am. J. Pathol. 2018, 188, 2042–2058. [Google Scholar] [CrossRef]
- Kurashima, Y.; Kiyono, H. Mucosal Ecological Network of Epithelium and Immune Cells for Gut Homeostasis and Tissue Healing. Annu. Rev. Immunol. 2017, 35, 119–147. [Google Scholar] [CrossRef]
- Van den Steen, P.; Rudd, P.M.; Dwek, R.A.; Opdenakker, G. Concepts and principles of O-linked glycosylation. Crit. Rev. Biochem. Mol. Biol. 1998, 33, 151–208. [Google Scholar] [CrossRef]
- Myles, I.A. Fast food fever: Reviewing the impacts of the Western diet on immunity. Nutr. J. 2014, 13, 61. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Thorburn, A.N.; Macia, L.; Mackay, C.R. Diet, metabolites, and "western-lifestyle" inflammatory diseases. Immunity 2014, 40, 833–842. [Google Scholar] [CrossRef] [Green Version]
- Prinz, P. The role of dietary sugars in health: Molecular composition or just calories? Eur. J. Clin. Nutr. 2019, 73, 1216–1223. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Lv, J.; Guo, L.; Liu, J.-J.; Zhao, H.-P.; Zhang, J.; Wang, J.-H. Alteration of the esophageal microbiota in Barrett’s esophagus and esophageal adenocarcinoma. World J. Gastroenterol. 2019, 25, 2149–2161. [Google Scholar] [CrossRef]
- Lopetuso, L.R.; Severgnini, M.; Pecere, S.; Ponziani, F.R.; Boskoski, I.; Larghi, A.; Quaranta, G.; Masucci, L.; Ianiro, G.; Camboni, T.; et al. Esophageal microbiome signature in patients with Barrett’s esophagus and esophageal adenocarcinoma. PLoS ONE 2020, 15, e0231789. [Google Scholar] [CrossRef]
- Deshpande, N.P.; Riordan, S.M.; Castaño-Rodríguez, N.; Wilkins, M.R.; Kaakoush, N.O. Signatures within the esophageal microbiome are associated with host genetics, age, and disease. Microbiome 2018, 6, 227. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Noble, E.E.; Olson, C.A.; Davis, E.; Tsan, L.; Chen, Y.-W.; Schade, R.; Liu, C.; Suarez, A.; Jones, R.B.; de La Serre, C.; et al. Gut microbial taxa elevated by dietary sugar disrupt memory function. Transl. Psychiatry 2021, 11, 194. [Google Scholar] [CrossRef] [PubMed]
- Strober, J.W.; Brady, M.J. Dietary Fructose Consumption and Triple-Negative Breast Cancer Incidence. Front. Endocrinol. 2019, 10, 367. [Google Scholar] [CrossRef]
- Charrez, B.; Qiao, L.; Hebbard, L. The role of fructose in metabolism and cancer. Horm. Mol. Biol. Clin. Investig. 2015, 22, 79–89. [Google Scholar] [CrossRef]
- Kubo, A.; Corley, D.A.; Jensen, C.D.; Kaur, R. Dietary factors and the risks of oesophageal adenocarcinoma and Barrett’s oesophagus. Nutr. Res. Rev. 2010, 23, 230–246. [Google Scholar] [CrossRef] [Green Version]
- Lagkouvardos, I.; Fischer, S.; Kumar, N.; Clavel, T. Rhea: A transparent and modular R pipeline for microbial profiling based on 16S rRNA gene amplicons. PeerJ 2017, 5, e2836. [Google Scholar] [CrossRef] [Green Version]
- RStudio Team. RStudio: Integrated Development Environment for R; RStudio, Inc.: Boston, MA, USA, 2016. [Google Scholar]
- R Core Development Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2018. [Google Scholar]
- Chambers, M.C.; Maclean, B.; Burke, R.; Amodei, D.; Ruderman, D.L.; Neumann, S.; Gatto, L.; Fischer, B.; Pratt, B.; Egertson, J. A cross-platform toolkit for mass spectrometry and proteomics. Nat. Biotechnol. 2012, 30, 918–920. [Google Scholar] [CrossRef]
- Smith, C.A.; Want, E.J.; O’Maille, G.; Abagyan, R.; Siuzdak, G. XCMS: Processing mass spectrometry data for metabolite profiling using nonlinear peak alignment, matching, and identification. Anal. Chem. 2006, 78, 779–787. [Google Scholar] [CrossRef] [PubMed]
- Tautenhahn, R.; Böttcher, C.; Neumann, S. Highly sensitive feature detection for high resolution LC/MS. BMC Bioinform. 2008, 9, 504. [Google Scholar] [CrossRef] [Green Version]
- Benton, H.P.; Want, E.J.; Ebbels, T.M. Correction of mass calibration gaps in liquid chromatography–mass spectrometry metabolomics data. Bioinformatics 2010, 26, 2488–2489. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Kebede, M.; Favaloro, J.; Gunton, J.E.; Laybutt, D.R.; Shaw, M.; Wong, N.; Fam, B.C.; Aston-Mourney, K.; Rantzau, C.; Zulli, A.; et al. Fructose-1,6-bisphosphatase overexpression in pancreatic beta-cells results in reduced insulin secretion: A new mechanism for fat-induced impairment of beta-cell function. Diabetes 2008, 57, 1887–1895. [Google Scholar] [CrossRef] [Green Version]
- Patel, C.; Douard, V.; Yu, S.; Tharabenjasin, P.; Gao, N.; Ferraris, R.P. Fructose-induced increases in expression of intestinal fructolytic and gluconeogenic genes are regulated by GLUT5 and KHK. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2015, 309, R499–R509. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Parekh, S.; Ziegenhain, C.; Vieth, B.; Enard, W.; Hellmann, I. The impact of amplification on differential expression analyses by RNA-seq. Sci. Rep. 2016, 6, 25533. [Google Scholar] [CrossRef] [Green Version]
- Macosko, E.Z.; Basu, A.; Satija, R.; Nemesh, J.; Shekhar, K.; Goldman, M.; Tirosh, I.; Bialas, A.R.; Kamitaki, N.; Martersteck, E.M.; et al. Highly Parallel Genome-wide Expression Profiling of Individual Cells Using Nanoliter Droplets. Cell 2015, 161, 1202–1214. [Google Scholar] [CrossRef] [Green Version]
- Love, M.I.; Huber, W.; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014, 15, 550. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Subramanian, A.; Tamayo, P.; Mootha, V.K.; Mukherjee, S.; Ebert, B.L.; Gillette, M.A.; Paulovich, A.; Pomeroy, S.L.; Golub, T.R.; Lander, E.S.; et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA 2005, 102, 15545. [Google Scholar] [CrossRef] [Green Version]
- Liberzon, A.; Subramanian, A.; Pinchback, R.; Thorvaldsdóttir, H.; Tamayo, P.; Mesirov, J.P. Molecular signatures database (MSigDB) 3.0. Bioinformatics 2011, 27, 1739–1740. [Google Scholar] [CrossRef] [PubMed]
- Liberzon, A.; Birger, C.; Thorvaldsdóttir, H.; Ghandi, M.; Mesirov, J.P.; Tamayo, P. The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst. 2015, 1, 417–425. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Dewdney, B.; Roberts, A.; Qiao, L.; George, J.; Hebbard, L. A Sweet Connection? Fructose’s Role in Hepatocellular Carcinoma. Biomolecules 2020, 10, 496. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Christen, P.; Jaussi, R.; Benoit, R. (Eds.) Glykolyse und Citratzyklus. In Biochemie und Molekularbiologie: Eine Einführung in 40 Lerneinheiten; Springer: Berlin/Heidelberg, Germany, 2016; pp. 161–175. [Google Scholar]
- Lee, H.-J.; Cha, J.-Y. Recent insights into the role of ChREBP in intestinal fructose absorption and metabolism. BMB Rep. 2018, 51, 429–436. [Google Scholar] [CrossRef] [PubMed] [Green Version]
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Proaño-Vasco, A.; Baumeister, T.; Metwaly, A.; Reitmeier, S.; Kleigrewe, K.; Meng, C.; Gigl, M.; Engleitner, T.; Öllinger, R.; Rad, R.; et al. High-Fructose Diet Alters Intestinal Microbial Profile and Correlates with Early Tumorigenesis in a Mouse Model of Barrett’s Esophagus. Microorganisms 2021, 9, 2432. https://doi.org/10.3390/microorganisms9122432
Proaño-Vasco A, Baumeister T, Metwaly A, Reitmeier S, Kleigrewe K, Meng C, Gigl M, Engleitner T, Öllinger R, Rad R, et al. High-Fructose Diet Alters Intestinal Microbial Profile and Correlates with Early Tumorigenesis in a Mouse Model of Barrett’s Esophagus. Microorganisms. 2021; 9(12):2432. https://doi.org/10.3390/microorganisms9122432
Chicago/Turabian StyleProaño-Vasco, Andrea, Theresa Baumeister, Amira Metwaly, Sandra Reitmeier, Karin Kleigrewe, Chen Meng, Michael Gigl, Thomas Engleitner, Rupert Öllinger, Roland Rad, and et al. 2021. "High-Fructose Diet Alters Intestinal Microbial Profile and Correlates with Early Tumorigenesis in a Mouse Model of Barrett’s Esophagus" Microorganisms 9, no. 12: 2432. https://doi.org/10.3390/microorganisms9122432