Gastroenterology 2024;166:323–337

GUT MICROBIOTA
High Soluble Fiber Promotes Colorectal Tumorigenesis Through
Modulating Gut Microbiota and Metabolites in Mice
Jia Yang,1,* Hong Wei,2,3,* Yufeng Lin,1 Eagle S. H. Chu,1 Yunfei Zhou,1 Hongyan Gou,1
Shang Guo,1 Harry C. H. Lau,1 Alvin H. K. Cheung,4 Huarong Chen,1 Ka Fei To,4
Joseph J. Y. Sung,1,5 Yong Wang,2 and Jun Yu1

GUT MICROBIOTA
1
Institute of Digestive Disease and Department of Medicine and Therapeutics, State Key Laboratory of Digestive
Disease, Li Ka Shing Institute of Health Sciences, CUHK Shenzhen Research Institute, The Chinese University of
Hong Kong, Hong Kong SAR, China; 2Department of Laboratory Animal Science, College of Basic Medical Sciences,
Army Medical University, Chongqing, China; 3Department of Precision Medicine, The First Affiliated Hospital, Sun Yat-
sen University, Guangzhou, China; 4Department of Anatomical and Cellular Pathology, The Chinese University of
Hong Kong, Hong Kong SAR, China; and 5Lee Kong Chian School of Medicine, Nanyang Technological University,
Singapore

abolished the pro-tumorigenic effect of mixed high-fiber diet,

See editorial on page 250. while transplanting stools from mice fed with mixed high-
fiber diet accelerated tumor growth in AOM-treated germ-
free mice. We therefore characterized the contribution of
BACKGROUND & AIMS: Dietary fibers are mainly fermented soluble and insoluble fiber in CRC separately. Our results
by the gut microbiota, but their roles in colorectal cancer revealed that soluble fiber inulin or guar gum, but not
(CRC) are largely unclear. Here, we investigated the associa- insoluble fiber cellulose, promoted colorectal tumorigenesis
tions of different fibers with colorectal tumorigenesis in mice. in AOM-treated and Apcmin/þ mice. Soluble fiber induced gut
METHODS: Apcmin/þ mice and C57BL/6 mice with azoxy- dysbiosis with Bacteroides uniformis enrichment and Bifido-
methane (AOM) injection were used as CRC mouse models. bacterium pseudolongum depletion, accompanied by
Mice were fed with mixed high-fiber diet (20% soluble fiber increased fecal butyrate and serum bile acids and decreased
and 20% insoluble fiber), high-inulin diet, high-guar gum diet, inosine. We also identified a positive correlation between
high-cellulose diet, or diets with different inulin dose. Germ- inulin dosage and colorectal tumorigenesis. Moreover,
free mice were used for validation. Fecal microbiota and transplanting stools from mice fed with high-inulin diet
metabolites were profiled by shotgun metagenomic increased colonic cell proliferation and oncogene expressions
sequencing and liquid chromatography–mass spectrometry, in germ-free mice. CONCLUSION: High-dose soluble but not
respectively. RESULTS: Mixed high-fiber diet promoted insoluble fiber potentiates colorectal tumorigenesis in a dose-
colorectal tumorigenesis with increased tumor number and dependent manner by dysregulating gut microbiota and me-
tumor load in AOM-treated and Apcmin/þ mice. Antibiotics use tabolites in mice.
 324 Yang et al Gastroenterology Vol. 166, Iss. 2

Keywords: Dietary Nutrient; Microbiota; Gut Metabolism; Colon

WHAT YOU NEED TO KNOW
Cancer.
BACKGROUND AND CONTEXT
Dietary fibers are mainly fermented by the gut microbiota;
C olorectal cancer (CRC) is one of the most common
and deadly cancers worldwide.1 CRC development is
intensively associated with alteration in the gut micro-
however, their roles in colorectal cancer remain largely
elusive.

biota.2,3 Microbiota composition is readily influenced by a NEW FINDINGS

variety of environmental factors, especially daily food intake,4 High soluble fiber (eg, inulin, guar gum) but not insoluble
and our previous study reported the mechanistic association fiber potentiates colorectal tumorigenesis in a dose-
of high-fat diet with colorectal tumorigenesis.5 In general, the dependent manner in mice, through inducing gut
GUT MICROBIOTA

microbial dysbiosis with pathobiont enrichment and

food we eat provides nutrients to support both the growth probiotic depletion, accompanied with metabolite
and diversity of gut microbes. Dietary fiber, as one of the dysregulation.
major food components in vegetables and fruits, plays an
important role in maintaining intestinal health. Dietary fiber LIMITATIONS
is associated with lowered CRC risk with accumulated evi- This is a preclinical animal study investigating the effect of
dence from multiple cohort studies.6 However, controversial soluble fiber on colorectal tumorigenesis. Clinical
results have emerged in recent studies,7,8 which demon- observational investigations are needed to validate the
impact of soluble fiber on colorectal cancer.
strated the potential correlation between dietary fiber and
colorectal or hepatic tumorigenesis. Butyrate, one of the key CLINICAL RESEARCH RELEVANCE
short-chain fatty acids (SCFAs) generated during gut Dietary and gut microbiota modulations could be
microbe-mediated fiber fermentation, has been shown to promising strategies in the prevention and treatment of
promote colorectal tumorigenesis in a transgenic CRC mouse colorectal cancer. Our research raises the hypothesis
model with mutations of the Apc gene and mismatch repair that very high dietary intake of soluble fiber could be
gene Msh2.7 Moreover, soluble fibers, including inulin and deleterious for individuals at higher risk of colorectal
cancer.
pectin, were found to induce hepatic carcinoma via dysre-
gulating the gut microbiota in mice.8 BASIC RESEARCH RELEVANCE
Dietary fibers are classified into soluble and insoluble, in The crosstalk between diet and gut microbiota is a critical
accordance with water solubility.9–11 Most soluble fibers factor in contributing to colorectal tumorigenesis. Dietary
such as inulin and guar gum, can be fermented by bacteria dosage is a considerable factor linking the gut microbiota/
in the colon to generate metabolites including SCFAs. In metabolites with colorectal cancer in preclinical mouse
contrast, insoluble fibers such as cellulose and hemi- models.
cellulose, can only be slowly digested by gut bacteria or
even resistant to microbial fermentation, yet they are crucial Kong. Mice (6–8 weeks old) were randomly divided into groups
for fecal bulking and bowel movement. In this study, we with different dietary fiber supplementation (Supplementary
examined the effects of different types and doses of dietary Table 1) (Specialty Feeds, Glen Forrest, Australia). After 1
fiber on CRC development in multiple mouse models. Our week adaptation to the diet, mice received 6 consecutive doses
results showed that high dietary fibers with mixed types of azoxymethane (AOM, 10 mg/kg). Drinking water was sup-
(soluble and insoluble fiber) promote colorectal tumori- plemented with or without antibiotics cocktail (0.2 g/L of
ampicillin, neomycin, metronidazole, and 0.1 g/L of vancomy-
genesis in carcinogen-induced and spontaneous transgenic
cin) for 2 weeks, and every other 2 weeks until the end of the
CRC mouse models. The use of antibiotics abolished the pro-
experiment, to deplete the gut microbiota.
tumorigenic features of high-fiber diet (HFiD), whereas fecal
Male Apcmin/þ mice at 6 to 8 weeks old were treated with or
microbiota transplantation (FMT) promoted tumor growth without dietary fiber supplementation same as in the AOM
in germ-free mice, indicating the necessity of an intact gut mouse model to mimic spontaneous genetic-associated CRC.
microbiota in HFiD -associated colorectal tumorigenesis. We Mice were harvested at week 22 and 12 for AOM and Apcmin/þ
then discovered that soluble fibers inulin and guar gum, but models, respectively. All procedures adhered to the guidelines
not insoluble fiber, promote CRC development in a dose-
dependent manner. Mechanically, high-dose soluble fiber
induced gut dysbiosis with depletion of probiotic Bifido- * Authors share co-first authorship.
bacterium pseudolongum and its metabolite inosine,
Abbreviations used in this paper: AOM, azoxymethane; CD, control diet;
accompanied with increased fecal butyrate and serum bile CRC, colorectal cancer; EMT, epithelial-to-mesenchymal transition; FMT,
acids, together driving colorectal tumorigenesis in mice. fecal microbiota transplantation; GC-MS, gas chromatography–mass
spectrometry; HCD, high-cellulose diet; HFiD, high-fiber diet; HGGD,
high-guar gum diet; HID, high-inulin diet; PBS, phosphate-buffered saline;
PCR, polymerase chain reaction; SCFAs, short-chain fatty acids.
Materials and Methods
Most current article
Conventional and Transgenic CRC Mouse © 2024 The Author(s). Published by Elsevier Inc. on behalf of the AGA
Models Institute. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Conventional male C57BL/6 mice were bred in the Labo- 0016-5085
ratory Animal Services Centre at the Chinese University of Hong https://doi.org/10.1053/j.gastro.2023.10.012
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 325

approved by the Animal Experimentation Ethics Committee of HFiD developed obvious macroscopic tumors at week
the Chinese University of Hong Kong. 20,whereas mice fed with CD showed no clear sign of tumor
formation in the colon (Figure 1B). At harvest, macroscopic
examination of the colon manifested that 83.33% of HFiD-
Germ-free Mouse Model
fed mice developed colonic tumors (5 of 6), which was
To determine the direct contribution of HFiD-modulated gut
significantly higher than CD-fed mice (1 if 7, 14.29%) (P <
microbiota on colorectal tumorigenesis, male germ-free C57BL/
.05, Figure 1C). In keeping with these findings, HFiD-fed
6 mice at 8 weeks old fed with HFiD were divided into 2 groups
under HFiD and were gavaged once with fecal samples
mice had significantly increased colonic tumor number (P
collected from HFiD-fed AOM-treated mice or phosphate- < .05) and tumor volume (P < .01) compared with CD-fed
buffered saline (PBS). Briefly, 1 g of stool samples were ho- mice (Figure 1C). Histological examination of colon sec-
tions confirmed that HFiD-fed mice developed a greater

GUT MICROBIOTA
mogenized in 5 mL of PBS. Recipient mice were then trans-
planted with 200 mL of the suspension by gastric gavage. These proportion of high-grade dysplasia compared with CD-fed
germ-free mice also received AOM injection with the same mice (P < .05; Figure 1D). Moreover, colon sections of
regimen as in conventional mice. Mice were killed at week 30 HFiD-fed mice exhibited significantly more Ki-67–positive
following gavage. To determine the effect of inosine, male germ- cells, indicating increased cell proliferation in HFiD-fed mice
free C57BL/6 mice at 8 weeks old were gavaged with PBS and (P < .001; Figure 1E).
fecal samples collected from high inulin diet–fed mice. Mice To verify these findings, we established another CRC
were then gavaged with PBS or inosine (300 mg/kg body mouse model with transgenic Apcmin/þ mice (Figure 1F).
weight) every other day and killed 28 days after the first Consistent with the results in AOM-induced CRC mouse
gavage. model, HFiD significantly increased tumor incidence (P <
To determine the pure effect of different fibers on the gut .05), tumor number (P < .0001), and tumor volume (P <
microbiota of healthy colonic mucosa, male germ-free C57BL/6 .01, Figure 1G) in the colon. Histological examination of
mice at 8 weeks old fed with control diet (CD) were divided colon sections confirmed that HFiD-fed mice developed a
into 3 groups and were gavaged once with fecal samples greater proportion of high-grade dysplasia compared with
collected from high-inulin-fed, high-cellulose-fed, or CD-fed CD-fed mice (P < .01; Figure 1H). HFiD-fed Apcmin/þ mice
mice. Mice were killed at 3 months following gavage. Germ- also exhibited significantly more Ki-67 positive colon cells
free mice were bred at the Department of Laboratory Animal
compared with CD-fed Apcmin/þ mice (P < .01; Figure 1I).
Science, Army Medical University, Chongqing, China. All pro-
Although decreased tumor number (P < .05) and tumor
cedures were approved by the Animal Experimentation Ethics
volume (P < .01) were observed in the small intestine of
Committee of Army Medical University.
HFiD-fed Apcmin/þ mice (Supplementary Figure 1), our
consistent findings between 2 CRC mouse models (AOM-
Statistical Analysis induced and Apcmin/þ) suggested that HFiD could promote
All results were expressed as mean ± standard deviation tumorigenesis in the colon.
unless otherwise indicated. Numerical variables between 2
groups were compared using unpaired Student t test or Mann-
HFiD Induces Gut Microbiota Dysbiosis
Whitney U test where appropriate. Comparisons of categorical
To explore the role of gut microbiota in HFiD-associated
variables between 2 groups were performed using chi-square
test or Fisher’s exact test. Difference in cell growth rate was
colorectal tumorigenesis, we performed microbiota profiling
determined by 2-way analysis of variance, and Dunnett’s test on stool samples from HFiD and CD-fed mice in AOM model.
was used for multiple comparison analysis. Wilcoxon signed- Microbiota compositional discriminations between HFiD-fed
rank test was used for differential analyses of metagenomic and CD-fed mice were demonstrated by principal co-
and metabolomic data. All statistical analyses were conducted ordinates analysis (Figure 2A). Lower bacterial a-diversity
using GraphPad Prism, version 7.0 (GraphPad, La Jolla, CA) or R (Shannon index) was also observed in HFiD-fed mice
package. Differences were considered significant with P values (Figure 2A). Twelve differentially enriched and 13 depleted
less than .05. bacterial species were identified in HFiD-fed mice compared
Additional methods are provided in the Supplementary with CD-fed mice, of which the beneficial probiotic Bifido-
Methods. bacterium pseudolongum was depleted in HFiD-fed mice (P
< .05, Figure 2B).

Results Depleted Microbiota by Antibiotics Attenuates

HFiD Promotes Colorectal Tumorigenesis in Both HFiD-associated Colorectal Tumorigenesis
AOM-treated Mice and Apcmin/þ Mice To further explore whether gut microbial dysbiosis
To examine the role of dietary fiber in colorectal contributed to HFiD-associated CRC development, we
tumorigenesis, we fed AOM-treated conventional C57BL/6 treated HFiD-fed AOM-induced CRC mice with or without
mice with either HFiD (20% soluble and 20% insoluble fi- antibiotics cocktail (Figure 2C). Colonoscopy was performed
ber) or CD, respectively (Figure 1A). Details of diet and we observed that mice receiving antibiotics cocktails
composition are shown in Supplementary Table 1. Colo- showed no obvious sign of tumor formation in the colon at
noscopy was performed to monitor tumor development in week 20 (Figure 2D). At harvest, no significant difference in
mice. Colonoscopy surveillance showed that mice fed with body weight between mice with and without antibiotics
 326 Yang et al Gastroenterology Vol. 166, Iss. 2
GUT MICROBIOTA

Figure 1. HFiD promotes colorectal tumorigenesis in both AOM-treated mice and Apcmin/þ mice. (A) Experimental design for
dietary fiber treatment in AOM-induced CRC mouse model. (B) Representative images of colonoscopy surveillance at week 20.
(C) Representative colon images when mice were killed. Tumor incidence, tumor number, and volume in CD-fed (n ¼ 7) and
HFiD-fed (n ¼ 6) mice. (D) Hematoxylin-eosin (H&E) staining for pathological diagnosis of colon sections. Quantitative analysis
of pathological score was calculated according to the following criteria: 0, normal; and 2, high-grade dysplasia (HGD). (E) Ki-67
staining of colon sections with quantitative analysis. (F) Experimental design for dietary fiber treatment in Apcmin/þ CRC mouse
model. (G) Representative colon images at the time mice were killed. Tumor incidence, tumor number, and volume in CD-fed
(n ¼ 15) and HFiD-fed (n ¼ 15) mice. (H) H&E staining for pathological diagnosis of colon sections. Quantitative analysis of
pathological score was calculated according to the following criteria: 0, normal; 1, low-grade dysplasia (LGD); and 2, HGD. (I)
Ki-67 staining of mice colon sections with quantitative analysis.

treatment was observed (Supplementary Figure 2A). antibiotics-naïve mice (5 of 6, 83.33%) (P < .05, Figure 2E).
Macroscopic examination of colon manifested that only 1 Antibiotics treatment also markedly reduced colonic tumor
mouse developed colonic tumors among antibiotics-treated number and tumor volume (both P < .01, Figure 2E) in
HFiD-fed mice (1 of 7, 14.29%), significantly lower than antibiotics-treated HFiD-fed mice. Histology assessment
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 327

GUT MICROBIOTA
Figure 2. HFiD promotes colorectal tumorigenesis through inducing gut microbiota dysbiosis. (A) Principal coordinates
analysis (PCoA) (left) and Shannon diversity (right) of the gut microbiota in AOM-treated CD-fed (n ¼ 5) and HFiD-fed (n ¼ 5)
mice. (B) Volcano plot of differential bacteria species in stool samples of CD-fed and HFiD-fed mice. Relative abundance of
B. pseudolongum in AOM-treated HFiD-fed and CD-fed mice. (C) Experimental design for antibiotics treatment in AOM-
induced CRC mouse model fed with HFiD. (D) Representative images of colonoscopy surveillance at week 20. (E) Repre-
sentative colon images at time mice were killed. Tumor incidence, tumor number, and volume in HFiD-fed mice treated with
(n ¼ 7) or without antibiotics (n ¼ 6). (F) Hematoxylin-eosin (H&E) staining for pathological diagnosis and Ki-67 staining of colon
sections with quantitative analysis. Quantitative analysis of pathological score was calculated according to the following
criteria: 0, normal; and 2, high-grade dysplasia (HGD). (G) Experimental design for FMT in AOM-treated germ-free mouse
model fed with HFiD. (H) Representative colon images at time mice were killed. Tumor incidence, tumor number, and volume in
HFiD-FMT germ-free mice (n ¼ 16) and control mice (n ¼ 15). Abx, antibiotics.

demonstrated that HFiD-fed mice treated with antibiotics HFiD-fed mice (P < .05; Figure 2F). These findings demon-
harbored a lower proportion of high-grade dysplasia (1 of 7, strated that using antibiotics to deplete the gut microbiota
14.29%), compared with antibiotics-naïve mice (5 of 6, could alleviate HFiD-associated CRC, thus implicating that
83.33%) (P < .01; Figure 2F). We also observed decreased the gut microbiota may play an essential role in mediating
epithelial cell proliferation in the colon of antibiotics-treated HFiD-associated colorectal tumorigenesis.
 328 Yang et al Gastroenterology Vol. 166, Iss. 2

Gavage of HFiD-modulated Stool Samples were observed in HID-fed mice compared with CD-fed or
Promotes Colorectal Tumorigenesis in AOM- HCD-fed Apcmin/þ mice (Figure 3E). Histological examina-
treated Germ-free Mice tion of colon sections confirmed that HID-fed Apcmin/þ mice
To confirm whether HFiD-modulated gut microbiota is developed a greater proportion of adenocarcinoma
the key factor in HFiD-associated CRC, we performed FMT compared with CD-fed mice and HCD-fed mice (both P <
by gavaging stools from either HFiD-fed mice (HFiD-FMT) or .01; Supplementary Figure 4C). Our consistent findings be-
PBS (control) into AOM-treated germ-free mice fed with tween 2 CRC mouse models confirmed that soluble fiber
HFiD (Figure 2G). At harvest, no significant difference in inulin but not insoluble fiber cellulose is responsible for
body weight between HFiD-FMT and controls was observed HFiD-associated colorectal tumorigenesis in mice.
(Supplementary Figure 2B). Macroscopic examination of
Soluble Fiber Guar Gum Promotes Colorectal
GUT MICROBIOTA

colon manifested that HFiD-FMT mice (14 of 16, 87.5%) had

significantly higher tumor incidence than controls (7 of 15, Tumorigenesis in AOM-treated Mice
46.7%) (P < .05, Figure 2H). The transplantation of stools To evaluate whether other soluble fibers have similar
from HFiD-fed mice also greatly increased colonic tumor pro-tumorigenic roles as observed in inulin, we estab-
number (P < .001) and tumor volume (P < .0001, lished another batch of AOM-induced CRC mouse model
Figure 2H). These findings collectively implied that the gut and supplemented with high-guar gum diet (another kind
microbiota could promote HFiD-associated colorectal of soluble fiber, HGGD) or CD (Figure 3F and
tumorigenesis. Supplementary Table 1). At harvest, macroscopic exami-
nation of colon manifested that all HGGD-fed mice devel-
Soluble Fiber Inulin But Not Insoluble Fiber oped colonic tumors (4 of 4, 100%), which was
Cellulose Promotes Colorectal Tumorigenesis in significantly higher than CD-fed mice (1 of 5, 20%) (P <
.05, Figure 3G and H). HGGD also markedly increased
Both AOM-treated Mice and Apcmin/þ Mice
colonic tumor number (P < .05) and tumor volume
Our HFiD contains both insoluble and soluble fibers
compared with CD-fed mice (P < .05, Figure 3H), inferring
(Supplementary Table 1), of which soluble fiber could be
that soluble fiber guar gum also promoted colorectal
fermented by gut microbes, whereas insoluble fiber could
tumorigenesis. Considering the consistent results from
not. Because HFiD promoted colorectal tumorigenesis
different CRC mouse models with various fiber treatments,
dependent on the gut microbiota, we hypothesized that
we confirmed that soluble fibers including inulin and guar
soluble fiber may contribute to colorectal tumorigenesis. We
gum, instead of insoluble fiber cellulose, could promote
therefore fed AOM-treated conventional male C57BL/6 mice
colorectal tumorigenesis in mice.
with different diets including CD, high-cellulose diet (HCD,
insoluble fiber) or high-inulin diet (HID, soluble fiber)
(Figure 3A and Supplementary Table 1). At harvest, con- Soluble Fiber Alters Gut Microbiota and Induces
sumption of HCD or HID diet significantly reduced mouse Depletion of Probiotics
body weight compared with CD-fed mice (both P < .0001, To explore the association of altered gut microbiota with
Supplementary Figure 3). As shown in Figure 3B, all HID-fed soluble fiber–associated colorectal tumorigenesis, we per-
mice harbored colonic tumors (10 of 10, 100%) with tumor formed metagenomic sequencing on stool samples from
incidence significantly higher than CD-fed mice (4 of 10, AOM-treated mice fed with soluble fiber (inulin or guar
40%) (P < .05) and HCD-fed mice (5 of 13, 38.5%) (P < gum), insoluble fiber (cellulose), or CD. Microbiota compo-
.01). HID also markedly increased tumor number and tumor sitional discrimination (principal coordinates analysis) was
volume (both P < .0001; Figure 3B). Histological examina- observed between mice fed with soluble fiber (HID or
tion of colon sections confirmed that HID-fed mice devel- HGGD) and insoluble fiber cellulose (HCD) or CD
oped greater proportions of high-grade dysplasia and low- (Figure 4A). Lower bacterial a-diversity was observed in
grade dysplasia compared with CD-fed mice or HCD-fed samples from soluble fiber HID-fed or HGGD-fed mice,
mice (both P < .01; Supplementary Figure 4A). Moreover, compared with insoluble fiber HCD-fed and CD-fed mice
colon sections of HID-fed mice exhibited significantly more (Figure 4B). Bacteria with differential abundance was then
Ki-67-positive cells, indicating increased cell proliferation in identified in mice fed with soluble fiber (Figure 4C). The
HID-fed mice (P < .01 vs CD; P < .05 vs HCD; enrichment of Bacteroides vulgatus, Bacteroides uniformis,
Supplementary Figure 4B). and Bacteroides sp. A1C1 was observed in HID-fed mice,
For verification, we applied the transgenic Apcmin/þ compared with CD-fed or HCD-fed mice. In contrast, the
mouse model for validation with supplementation of CD, abundances of probiotics and potential beneficial bacteria
HCD, or HID (Figure 3C). Colonoscopy was performed and including Roseburia intestinalis, Lactobacillus lactis, Lacto-
we observed clear traits of colonic tumor formation in HID- bacillus reuteri, Akkermansia muciniphila,12 and Bifido-
fed mice at week 11 (Figure 3D). At harvest, mice fed with bacterium pseudolongum,13 were significantly decreased in
HID developed obvious macroscopic tumors, whereas mice mice fed with soluble fiber. B. pseudolongum was one of the
fed with CD or HCD showed less sign of tumor formation in top depleted bacteria in mice fed with soluble fiber, and we
colon (Figure 3E). Consistent with the results in the AOM- performed in vitro experiments using human CRC cell lines
induced CRC mouse model, increased colonic tumor num- to assess the anti-CRC effect of B. pseudolongum. As shown
ber (both P < .0001) and tumor volume (P < .001 vs CD) in Figure 4D, B. pseudolongum culture medium significantly
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 329

GUT MICROBIOTA

Figure 3. Soluble fiber but not insoluble fiber promotes colorectal tumorigenesis in both AOM-treated mice and Apcmin/þ mice.
(A) Experimental design for soluble fiber and insoluble fiber treatment in AOM-induced CRC mouse model. (B) Representative
colon images at time mice were killed. Tumor incidence, tumor number, and volume in CD-fed (n ¼ 10), HCD-fed (n ¼ 13), and
HID-fed (n ¼ 10) mice. (C) Experimental design for soluble fiber and insoluble fiber treatment in Apcmin/þ CRC mouse model.
(D) Representative images of colonoscopy surveillance at week 11. (E) Representative colon images at time mice were killed.
Tumor number and volume in CD-fed (n ¼ 14), HCD-fed (n ¼ 15) and HID-fed (n ¼ 12) mice. (F) Experimental design for soluble
fiber guar gum treatment in AOM-induced CRC mouse model. (G) Representative colon images at time mice were killed. (H)
Tumor incidence, tumor number, tumor volume, and pathological diagnosis of colons in CD-fed (n ¼ 5) and HGGD-fed (n ¼ 4)
mice. Quantitative analysis of pathological score was calculated according to the following criteria: 0, normal; and 1, low-grade
dysplasia (LGD).
 330 Yang et al Gastroenterology Vol. 166, Iss. 2
GUT MICROBIOTA

Figure 4. Soluble fiber alters gut microbiota and induces depletion of probiotics. (A) Principal coordinates analysis (PCoA) and
(B) Shannon diversity of the gut microbiota in AOM-treated CD-fed, HCD-fed, HID-fed, and HGGD-fed mice. (C) Heatmap for
differential bacteria among stool samples of AOM-treated CD-fed, HCD-fed, HID-fed, and HGGD-fed mice. (D) Coincubation
with B. pseudolongum culture medium inhibited proliferation of HT29 and Caco-2 CRC cells but not NCM460 cells. (E)
Representative images and quantitative analysis of colony formation of HT29 and Caco-2 CRC cells treated with blank control,
Escherichia coli culture medium (as negative control) and B. pseudolongum culture medium, respectively. (F) Heatmap for
functional profiling of metabolism pathways using HUMAnN2 in stool samples of AOM-treated CD-fed, HCD-fed, HID-fed, and
HGGD-fed mice. *P < .05, **P < .01; ***P < .001; ****P < .0001. BHI, brain heart infusion; Bp. CM, B. pseudolongum culture
medium.

inhibited the growth of 2 CRC cell lines HT29 (P < .0001) that soluble fiber could alter both gut microbiota composi-
and Caco-2 (P < .0001), but not the normal colon epithelial tion and its associated functions.
cell line NCM460. Moreover, B. pseudolongum culture me-
dium also markedly lowered colony formation in HT29 (P <
.01) and Caco-2 (P < .001) (Figure 4E). Soluble Fiber Induces Depletion of Probiotics-
We further categorized the gut metabolic pathways produced Metabolite Inosine and Enrichment of
altered by soluble fiber. Our analysis identified that stearate Fecal Butyrate and Serum Bile Acids
biosynthesis II, fatty acid elongation, oleate biosynthesis IV, To reveal metabolic changes in the gut microbiota, we
and superpathway of L-serine and glycine biosynthesis I performed metabolomic profiling on stool samples from
were enriched in mice fed with soluble fiber inulin and guar AOM-treated mice fed with soluble fiber (inulin or guar
gum, compared with CD and HCD (Figure 4F). Meanwhile, gum), insoluble fiber (cellulose), or CD. Gut metabolites
glycogen biosynthesis I was significantly downregulated in were significantly different among these groups of mice
mice fed with soluble fiber. Our results therefore indicated (Supplementary Table 2). Among the differential
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 331

GUT MICROBIOTA
Figure 5. Soluble fiber induces depletion of B. pseudolongum-derived inosine and enrichment of fecal butyrate. (A) Heatmap
of differential metabolites among stool samples of AOM-treated CD-fed, HCD-fed, HID-fed, and HGGD-fed mice. (B) The
effect of inosine on cell growth of HT29, Caco-2, and NCM460. (C) Experimental design for inosine administration in germ-free
mice receiving HID-FMT. (D) Protein expression of the cell-proliferating indicator PCNA in colon tissues of PBS, HID-FMTþ
PBS and HID-FMTþ inosine-treated mice. (E) Expressions of 10 transcripts were downregulated in colon tissues of HID-
FMTþ inosine mice compared with HID-FMTþ PBS mice, identified by the Mouse Cancer Pathway Finder PCR array. (F)
Ki-67 staining with quantitative analysis of PBS (n ¼ 7), HID-FMTþ PBS (n ¼ 8), and HID-FMTþ inosine (n ¼ 8) mice colon
sections. (G) Quantitative measurement of butyrate, acetate, propionic acid, hexanoic acid, and valeric acid levels in stool
samples of CD-fed, HCD-fed, and HID-fed mice by targeted metabolomics. (H) Experimental design for butyrate treatment in a
rectal orthotopic mouse model. (I) Representative tumor images at time mice were killed. Tumor weight in PBS-treated (n ¼ 7)
and butyrate-treated (n ¼ 9) mice. (J) Hematoxylin-eosin (H&E) staining for pathological diagnosis of tumor sections. Ki-67
staining for tumor sections with quantitative analysis.

metabolites, 4-hydroxybenzaldehyde and inosine were FMT (P < .01, Figure 5C and D). Cancer pathway polymerase
consistently depleted in mice fed with soluble fiber inulin chain reaction (PCR) array revealed the downregulation of
and guar gum, compared with CD-fed and HCD-fed mice (all oncogenic factors in mouse colon after inosine treatment
P < .05; Figure 5A). Inosine is one of the beneficial metab- (Figure 5E). Genes downregulated by inosine were mainly
olites secreted by B. pseudolongum.14 For validation, we related to DNA damage and repair (Xrcc4, Gadd45g, Ercc5),
treated human CRC cells with inosine and showed that metabolism (Acly, Lpl), apoptosis (Birc3), hypoxia (Arnt),
inosine markedly inhibited the growth of 2 CRC cell lines and cell cycle (Mki67) (Supplementary Table 3). Moreover,
HT29 (P < .0001) and Caco-2 (P < .0001) but not the colon sections of inosine-treated mice had significantly
normal colon epithelial cell NCM460 (Figure 5B and fewer Ki-67 positive cells, indicating decreased cell prolif-
Supplementary Figure 5A). Inosine administration also eration after inosine administration (P < .01 vs HID-FMT;
significantly lowered colonic proliferating cell nuclear anti- Figure 5F). These findings suggested that soluble fiber
gen (PCNA) expression in germ-free mice treated with HID- causes depletion of probiotics including B. pseudolongum as
 332 Yang et al Gastroenterology Vol. 166, Iss. 2

well as their associated metabolites such as inosine, which positive correlation between tumor growth and inulin
could at least in part contribute to CRC development. dosage, of which mice fed with 20% inulin diet had the
Apart from anti-tumorigenic metabolites, we also eval- highest tumor incidence (CD, 41.7% [5of 12]; 5% inulin,
uated the levels of oncometabolites in serum and fecal 8.3% [1 of 12]; 10% inulin, 46.2% [6 of 13]; 15% inulin,
samples of CD-fed, HCD-fed, and HID-fed mice. Our analysis 78.6% [11 of 14]) (Figure 6C). Consistently, 15% or 20%
revealed the significant upregulation of serum bile acids in inulin diet significantly increased tumor number and tumor
HID-fed mice compared with CD-fed or HCD-fed mice, volume in mice (all P < .01; Figure 6C). Histological exam-
including the major primary bile acids chenodeoxycholic ination of colon sections confirmed that mice fed with 15%
acid, as well as 2 secondary bile acids taurochenodeox- or 20% inulin diet developed greater proportions of high-
ycholic acid and tauroursodeoxycholic acid (Supplementary grade dysplasia and low-grade dysplasia compared with
Figure 5B). Intriguingly, all these bile acids were signifi- CD-fed mice (both P < .01; Figure 6D).
GUT MICROBIOTA

cantly downregulated in stools of HID-fed mice For verification, we applied the transgenic Apcmin/þ
(Supplementary Figure 5C), suggesting that their reabsorp- mouse model for validation with supplementation of diet
tion was promoted by the presence of high level of inulin in with different doses of inulin (Figure 6E). Colonoscopy
the gut. surveillance showed that mice fed with 15% or 20% inulin
It is well-known that soluble fiber can be fermented by diet had clear traits of colonic tumor formation at week 9
gut microbes to produce SCFAs. We therefore examined (Figure 6F). At harvest, mice fed with 15% and 20% inulin
SCFA levels in fecal samples of CD-fed, HCD-fed, and HID-fed diet developed obvious macroscopic tumors, whereas mice
mice by targeted gas chromatography–mass spectrometry. fed with CD showed less tumor formation in the colon
The result showed that butyrate, the major product of fiber (Figure 6G). Consistent with the results in an AOM-induced
fermentation, was significantly elevated in HID-fed mice CRC mouse model, increased colonic tumor number (both P
compared with CD-fed mice (P < .05; Figure 5G). The fecal < .0001) and tumor volume (both P < .001) were observed
concentrations of acetate (P < .0001), propionic acid (P < in Apcmin/þ mice fed with 15% or 20% inulin diet, compared
.05), and hexanoic acid (P < .01) were higher in HID-fed with CD-fed mice (Figure 6G). Histological examination of
mice than in HCD-fed mice. In contrast, valeric acid was colon sections confirmed that 15% or 20% inulin diet–fed
significantly decreased in HID-fed mice compared with CD- Apcmin/þ mice developed greater proportions of adenocar-
fed mice (P < .01). For validation, we treated human CRC cinoma and high-grade dysplasia compared with CD-fed
cells with butyrate (0.05 mM) and showed that butyrate mice (both P < .001; Supplementary Figure 6A and 6B).
could promote the growth of 2 CRC cell lines HCT116 (P < Moreover, increased tumor number and tumor volume were
.0001) and Caco-2 (P < .0001), but not the normal colon observed in the small intestine of Apcmin/þ mice fed with
epithelial cell NCM460 (Supplementary Figure 5D). We also 10%, 15%, or 20% inulin diet, compared with CD-fed mice
established an orthotopic mouse model by directly injecting (Supplementary Figure 6C). Our consistent findings be-
murine CRC cell line MC38 into the rectal mucosa to tween 2 CRC mouse models confirmed that inulin promotes
investigate the in vivo effect of butyrate (Figure 5H). Buty- CRC development in mice in a dose-dependent manner.
rate supplementation by enema significantly increased tu-
mor weight compared with PBS control (P < .01; Figure 5I),
along with increased Ki-67-positive cells (P < .01; Inulin-associated Gut Microbiota and Metabolites
Figure 5J), suggesting that direct butyrate administration Directly Promote Colon Cell Proliferation in
into the colonic lumen could contribute to CRC development Germ-free Mice
in mice. Collectively, our untargeted and targeted metab- To further evaluate the direct effect of soluble fiber on
olomic analyses indicated that the increase of serum bile the gut microbiota and metabolites, we performed FMT by
acids and fecal butyrate together with the depletion of gavaging stools from soluble fiber inulin-fed mice (HID-
inosine, may contribute to colorectal tumorigenesis in mice FMT), insoluble fiber cellulose-fed mice (HCD-FMT), or CD-
fed with a high soluble fiber diet. fed mice (CD-FMT) into germ-free mice fed with normal
chow during the whole experiment (Figure 7A). Trans-
plantation of stools from inulin-fed mice significantly pro-
Inulin Promotes Colorectal Tumorigenesis in a moted colon cell proliferation in the recipient germ-free
Dose-dependent Manner in Both AOM-treated mice, as evidenced by increased Ki-67-positive cells (P <
Mice and Apcmin/þ Mice .01; Figure 7B) and PCNA expression (P < .05; Figure 7C).
We acknowledged that the dosage of inulin used in our To mechanically understand how inulin-modulated gut
findings is rather high and could not reflect human physi- microbiota/metabolites promoted oncogenic trans-
ological conditions. The effect of inulin dosage on colorectal formation, colon tissues of germ-free mice were examined
tumorigenesis was therefore examined by feeding AOM- using Mouse Cancer Pathway Finder PCR array. We revealed
treated conventional mice with different doses of inulin the upregulation of oncogenic factors in HID-FMT mice
(5%, 10%, 15%, 20%) or CD, respectively (Figure 6A and compared with CD-FMT mice (Figure 7D and
Supplementary Table 4). Colonoscopy was performed and Supplementary Table 5) with confirmation by quantitative-
we observed obvious signs of colonic tumor formation in PCR (P < .05; Figure 7E). The upregulated genes in HID-
mice fed with 10%, 15%, or 20% inulin diet at week 19 FMT mice were mainly related to DNA damage and repair
(Figure 6B). When mice were killed, our result illustrated a (Lig4, Ddb2, Ercc3), metabolism (Uqcrfs1, G6pdx, Cpt2, Pfk1,
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 333

GUT MICROBIOTA

Figure 6. Inulin promotes colorectal tumorigenesis in a dose-dependent manner in both AOM-treated mice and Apcmin/þ mice. (A)
Experimental design for treatment of different inulin doses in AOM-induced CRC mouse model. (B) Representative images of colo-
noscopy surveillance at week 19. (C) Representative colon images at time mice were killed. Tumor incidence, tumor number, and
volume in CD-fed (n ¼ 12), 5% inulin diet–fed (n ¼ 12), 10% inulin diet–fed (n ¼ 13), 15% inulin diet–fed (n ¼ 14), and 20% inulin diet–
fed (n ¼ 14) mice. (D) Representative images of hematoxylin-eosin (H&E) staining and pathological diagnosis of colons in CD-fed, 5%
inulin diet–fed, 10% inulin diet–fed, 15% inulin diet–fed, and 20% inulin diet–fed mice. Quantitative analysis of pathological score was
calculated according to the following criteria: 0, normal; 1, low-grade dysplasia (LGD), 2, high-grade dysplasia (HGD). (E) Experimental
design for various doses of inulin treatment in spontaneous Apcmin/þ CRC mouse model. (F) Representative images of colonoscopy
surveillance at week 9. (G) Representative colon images at time mice were killed. Tumor number and volume in CD-fed (n ¼ 12), 5%
inulin diet–fed (n ¼ 13), 10% inulin diet–fed (n ¼ 14), 15% inulin diet–fed (n ¼ 13), and 20% inulin diet–fed (n ¼ 15) Apcmin/þ mice.
 334 Yang et al Gastroenterology Vol. 166, Iss. 2
GUT MICROBIOTA

Figure 7. Inulin-associated gut microbiota and metabolites directly promote colon cell proliferation in germ-free mice. (A)
Experimental design for stool transplantation from CD-fed, HID-fed, and HCD-fed mice to germ-free mice (CD-FMT, HCD-
FMT, and HID-FMT) fed with control diet. (B) Ki-67 staining with quantitative analysis of colon sections in CD-FMT (n ¼ 8),
HID-FMT (n ¼ 8), and HCD-FMT (n ¼ 8) mice. (C) Protein expression of the cell-proliferating indicator PCNA in colon tissues of
CD-FMT, HID-FMT, and HCD-FMT mice. (D) Expressions of 25 transcripts were upregulated in colon tissues of HID-FMT mice
compared with CD-FMT mice, identified by the Mouse Cancer Pathway Finder PCR array. (E) Quantitative-PCR of the top
upregulated gene Snail3. (F) Systematic diagram of major oncogenic pathways implicated by upregulated genes. (G) Summary
diagram of the study. Soluble fibers drive colorectal tumorigenesis through inducing gut dysbiosis with pathobionts enrich-
ment and probiotics depletion, accompanied with increase of serum bile acids and fecal butyrate, and decrease of inosine.

Gpd2, Atp5a1), epithelial-to-mesenchymal transition (EMT) evaluate the contribution of gut microbiota in HFiD-
(Snail3, Ocln), angiogenesis (Kdr), and cellular senescence associated CRC development. Depleting the gut microbiota
(Tbx2, Sirt1, Ing1, Igfbp3, Map2k1, Serpinb2) (Figure 7F). by antibiotics significantly attenuated HFiD-associated tu-
mor formation compared with mice with intact microbiota
(Figure 2), suggesting that microbial dysbiosis induced by
Discussion HFiD contributes to HFiD-associated colorectal tumorigen-
In this study, we provided evidence that HFiD plays a esis. Moreover, the fecal microbiota of HFiD-fed mice
pro-tumorigenic role in CRC using multiple mouse models. significantly promoted colorectal tumorigenesis in AOM-
To our knowledge, we are the first to report that high level treated germ-free mice, hence confirming that the gut
of mixed dietary fibers (20% soluble and 20% insoluble microbiota could mediate HFiD-associated CRC
fiber) could promote colorectal tumorigenesis in both development.
carcinogen AOM-induced and spontaneous transgenic Apc- Dietary fibers can be divided into soluble and insoluble
min/þ
CRC mouse models (Figure 1). The composition of gut fibers.10 To explore the role of different fibers in colorectal
microbiota is readily influenced by various dietary nutri- tumorigenesis, soluble fiber (inulin or guar gum) or insol-
ents,4 whereas alteration in the microbial community is uble fiber (cellulose) was supplemented to CRC mouse
involved in CRC development. For instance, our previous models. We revealed that high soluble fibers inulin and guar
study demonstrated that high-fat diet, which is considered gum, but not insoluble fiber cellulose, promote colorectal
as a common diet habit in developed countries, promotes tumorigenesis in both AOM-treated mice and Apcmin/þ mice
colorectal tumorigenesis through modulating the gut (Figure 3). In keeping with our findings, several studies
microbiota.5 Here, we used different mouse models to have demonstrated that soluble fiber promotes the
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 335

development of CRC7 and hepatocellular carcinoma8 in microbial dysbiosis with enrichment of pathobionts and
mice. By contrast, another study observed the anti- depletion of probiotics to contribute to colorectal
tumorigenic effect of inulin on CRC.15 This discrepancy in tumorigenesis.
the effect of fibers on tumorigenesis may be explained by Apart from gut microbes, we also unveiled the contri-
the butyrate paradox,16 which argues that the ability of bution of gut metabolites to colorectal tumorigenesis by
butyrate to promote or suppress cell proliferation is performing untargeted liquid chromatography–mass spec-
dependent on cell type, time, and butyrate concentration. trometry and targeted gas chromatography–mass spec-
Indeed, our analysis identified the significant elevation of trometry metabolomic profiling. We showed that the fecal
SCFAs including butyrate and acetate in HID-fed mice metabolome of mice fed with soluble fiber is greatly altered
(Figure 5). Our functional experiments also revealed the compared with CD-fed mice or HCD-fed mice (Figure 5).
potential pro-tumorigenic role of butyrate, as butyrate (0.05 Inosine, a metabolite mainly produced by B. pseudolongum

GUT MICROBIOTA
mM) could promote the proliferation of CRC cells in vitro, and A. muciniphila, was significantly depleted in mice with
and direct administration of butyrate (0.5 mM) by enema soluble fiber feeding, while it also inhibited CRC cell growth
enhanced rectal tumor development in an orthotopic mouse in vitro and suppressed colonic epithelial cell proliferation
model. These findings were supported by previous studies in vivo. These results align with our in silico analysis, as
showing that butyrate concentration at 0.05 mM is sufficient both B. pseudolongum and A. muciniphila were significantly
to promote hepatic cell proliferation, thereby contributing depleted in mice fed with soluble fiber. Consistently, a
to soluble fiber–induced hepatocellular carcinoma.8 Buty- recent study revealed that B. pseudolongum and its derived
rate could also exhibit pro-tumorigenic effect depending on inosine could enhance the function of antitumor T cells to
host genetic background and cell type. For instance, buty- boost immunotherapy efficacy in different mouse cancer
rate could fuel the hyperproliferation of MSH2/ colon models including intestinal cancer,14 hence indicating the
epithelial cells, whereas antibiotics treatment or a low- anti-tumorigenic role of inosine. Meanwhile, we identified
carbohydrate diet to reduce intestinal butyrate, inhibited the increase of serum bile acids including chenodeoxycholic
polyp formation in Apcmin/þ Msh2/ mice.7 Altogether, acid, taurochenodeoxycholic acid, and tauroursodeoxycholic
these findings therefore suggest that butyrate could be an acid, yet these bile acids were decreased in stools of HID-fed
oncometabolite in some circumstances. mice. Intriguingly, a previous study reported a similar
We demonstrated that soluble fiber induces gut dys- observation that microbiota-derived bile acids are enriched
biosis, and such microbial compositional change plays an in serum but depleted in stools after supplementation of
essential role in fiber-associated CRC development. 26% inulin diet, suggesting that high level of inulin may
Decreased a- and ß-diversities were observed in the gut stimulate bile acid reabsorption in the gut.24 It is known
microbiota of soluble fiber–fed mice (Figure 4), consistent that fecal bile acids are positively correlated with the risk of
with findings in patients with CRC with loss of microbial CRC,25–27 while there are fewer studies on the role of serum
richness and diversity.2,17 Dramatical depletion in probiotics bile acids in colorectal tumorigenesis. Bile acids can be
including B. pseudolongum, R. intestinalis, L. lactis, L. reuteri, transported from the gut to the bloodstream through in-
and A. muciniphila was identified in mice with soluble fiber testinal epithelial absorption, enterohepatic circulation, and
feeding, of which these probiotics, especially liver absorption and transformation.28,29 Recently, a pro-
B. pseudolongum, could significantly inhibit CRC cell growth spective study demonstrated that serum bile acids are
in vitro. B. pseudolongum is a probiotic species with well- strongly associated with increased risk of CRC in females,30
studied anti-tumorigenic role. Notably, although inulin is suggesting the critical involvement of serum bile acids in
known to promote the growth of Bifidobacterium, most CRC. Altogether, our findings indicated that the increase of
studies reported its enrichment after inulin consumption at serum bile acids together with fecal butyrate elevation and
a physiologically reasonable dose.18 In comparison, here we inosine depletion, may contribute to colorectal tumorigen-
fed mice with a diet containing 20% inulin. Given that the esis in mice fed with high soluble fiber diet.
bifidogenic effect of inulin is dependent on its concentration, We further examined the effect of different doses of
such high inulin dose may confer the occurrence of gut inulin in CRC. Our results revealed that higher dosage such
dysbiosis with the depletion of Bifidobacterium. On the as 15% and 20% inulin diet promotes colorectal tumori-
other hand, potential pathogenic bacteria B. uniformis was genesis in both AOM-treated mice and Apcmin/þ mice. We
significantly enriched in mice fed with soluble fiber. also observed that 5% or 10% inulin diet could not increase
Although our co-culture experiment has not shown the pro- tumor number (Figure 6). These solidify the fact that excess
proliferative effect of B. uniformis in CRC cell lines (data not inulin intake could accelerate colonic tumor formation. To
shown), B. uniformis was significantly enriched in patients our knowledge, we are the first to report the pro-
with Lynch syndrome who have higher predisposition to tumorigenic role of high-dose inulin in colorectal tumori-
CRC compared with healthy individuals.19 In general, gut genesis in mice. Several studies have reported the associa-
dysbiosis is closely associated with colorectal tumorigen- tion of inulin with disease progression, as inulin could
esis.20 Although pathogenic bacteria like Peptostreptococcus promote hepatocellular carcinoma in TLR5-deficient mice8
anaerobius drives CRC via stimulating oncogenic signal- and exacerbate inflammatory bowel disease with gut dys-
ings,21,22 probiotic supplements could inhibit the growth of biosis and surfeit colonic butyrate.31 Another recent study
pathobionts to suppress tumor formation.23 Collectively, our also showed that 26% inulin fiber promotes type 2
results demonstrated that soluble fiber induces gut inflammation at barrier surfaces.24 However, the dose effect
 336 Yang et al Gastroenterology Vol. 166, Iss. 2

of inulin was not investigated because a single dose of inulin 2. Yu J, Feng Q, Wong SH, et al. Metagenomic analysis of
was applied in these studies. Nevertheless, most of our faecal microbiome as a tool towards targeted non-
findings were based on mouse models, showing that only invasive biomarkers for colorectal cancer. Gut 2017;
extremely high dose of inulin could lead to colorectal 66:70–78.
tumorigenesis. In daily life, cereals are an important source 3. Nakatsu G, Li XC, Zhou HK, et al. Gut mucosal micro-
of dietary fiber, with a concentration of soluble fiber at only biome across stages of colorectal carcinogenesis. Nat
1.5% to 3%.32 Hence, 15% or 20% inulin dose used in this Commun 2015;6:8727.
study is much higher than daily intake, suggesting that these 4. Yang J, Yu J. The association of diet, gut microbiota and
results may not reflect physiological conditions in humans. colorectal cancer: what we eat may imply what we get.
Finally, we explored the effects of gut microbiota/me- Protein Cell 2018;9:474–487.
tabolites modulated by soluble fiber inulin on healthy 5. Yang J, Wei H, Zhou Y, et al. High-fat diet promotes
GUT MICROBIOTA

colonic mucosa of germ-free mice fed with normal chow colorectal tumorigenesis through modulating gut micro-
(Figure 7). Germ-free mice receiving stools from HID-fed biota and metabolites. Gastroenterology 2022;
mice showed significantly increased cell proliferation, sug- 162:135–149.e2.
gesting the direct effect of inulin on the gut microbiota and 6. Hansen L, Skeie G, Landberg R, et al. Intake of dietary
metabolites to promote cancer development. Mechanisti- fiber, especially from cereal foods, is associated with
cally, the gut microbiota/metabolites modulated by inulin lower incidence of colon cancer in the HELGA cohort. Int
J Cancer 2012;131:469–478.
elevated expressions of oncogenic factors mainly func-
tioning in DNA damage and repair (Lig4, Ddb2, Ercc3), 7. Belcheva A, Irrazabal T, Robertson SJ, et al. Gut micro-
bial metabolism drives transformation of MSH2-deficient
metabolism (Uqcrfs1, Cpt2), cell senescence (Tbx2, Sirt1),
colon epithelial cells. Cell 2014;158:288–299.
and EMT (Snail3). Snail3, the key transcriptional repressor
8. Singh V, Yeoh BS, Chassaing B, et al. Dysregulated
of E-cadherin, could induce EMT to increase migration and
microbial fermentation of soluble fiber induces chole-
invasion in colon cancer33 and prostate cancer.34 DNA ligase
static liver cancer. Cell 2018;175:679–694.e22.
IV, the protein product of Lig4, is essential for the mainte-
9. Makki K, Deehan EC, Walter J, et al. The impact of di-
nance of genome integrity but its abnormal variation could
etary fiber on gut microbiota in host health and disease.
lead to hypersensitivity to DNA damage, thus its over-
Cell Host Microbe 2018;23:705–715.
expression has been linked to poor prognosis in prostate
10. Deehan EC, Duar RM, Armet AM, et al. Modulation of the
cancer35 and glioblastoma.36 Interestingly, we found that
gastrointestinal microbiome with nondigestible ferment-
tumor formation is completely abolished in wild-type HFiD- able carbohydrates to improve human health. Microbiol
fed mice without AOM injection (data not shown), consis- Spectr 2017;5(5).
tent with a previous study in mice with the supplementation 11. Gill SK, Rossi M, Bajka B, et al. Dietary fibre in gastro-
of a diet containing 7.5% inulin.8 Our findings therefore intestinal health and disease. Nat Rev Gastroenterol
suggested that soluble fiber modulates host-microbe in- Hepatol 2021;18:101–116.
teractions to contribute to colorectal tumorigenesis in mice 12. Yan J, Sheng L, Li H. Akkermansia muciniphila: is it the
only under the presence of carcinogen or CRC-associated Holy Grail for ameliorating metabolic diseases? Gut Mi-
oncogene mutation. crobes 2021;13:1984104.
In conclusion, we demonstrated that soluble fiber such 13. Cizeikiene D, Jagelaviciute J. Investigation of antibacte-
as inulin and guar gum, instead of insoluble fiber cellulose, rial activity and probiotic properties of strains belonging
promotes colorectal tumorigenesis in mice in a dose- to Lactobacillus and Bifidobacterium genera for their
dependent manner. High intake of soluble fiber could potential application in functional food and feed prod-
induce gut dysbiosis with enrichment of potential patho- ucts. Probiotics Antimicrob Proteins 2021;
bionts and depletion of commensal probiotics, thereby 13:1387–1403.
contributing to colorectal tumorigenesis. On the other hand, 14. Mager LF, Burkhard R, Pett N, et al. Microbiome-derived
soluble fiber causes dysregulation of gut metabolism, inosine modulates response to checkpoint inhibitor
including increase of fecal butyrate and serum bile acids and immunotherapy. Science 2020;369:1481–1489.
decrease of fecal inosine, which also plays a pivotal role in 15. Pool-Zobel BL. Inulin-type fructans and reduction in
soluble fiber-associated CRC development (Figure 7G). colon cancer risk: review of experimental and human
data. Br J Nutr 2005;93(Suppl 1):S73–S90.
16. Bultman SJ, Jobin C. Microbial-derived butyrate: an
Supplementary Material oncometabolite or tumor-suppressive metabolite? Cell
Note: To access the supplementary material accompanying Host Microbe 2014;16:143–145.
this article, visit the online version of Gastroenterology at 17. Dai ZW, Coker OO, Nakatsu G, et al. Multi-cohort
www.gastrojournal.org, and at https://doi.org/10.1053/ analysis of colorectal cancer metagenome identified
j.gastro.2023.10.012. altered bacteria across populations and universal
bacterial markers. Gastroenterology 2018;154:
S1047–S1048.
References 18. Vandeputte D, Falony G, Vieira-Silva S, et al. Prebiotic
1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. inulin-type fructans induce specific changes in the hu-
CA Cancer J Clin 2020;70:7–30. man gut microbiota. Gut 2017;66:1968–1974.
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 337

19. Mori G, Orena BS, Cultrera I, et al. Gut microbiota 34. Osorio LA, Farfan NM, Castellon EA, et al. SNAIL tran-
analysis in postoperative Lynch syndrome patients. scription factor increases the motility and invasive ca-
Front Microbiol 2019;10:1746. pacity of prostate cancer cells. Mol Med Rep 2016;
20. Wong SH, Yu J. Gut microbiota in colorectal cancer: 13:778–786.
mechanisms of action and clinical applications. Nat Rev 35. Grupp K, Roettger L, Kluth M, et al. Expression of DNA
Gastroenterol Hepatol 2019;16:690–704. ligase IV is linked to poor prognosis and characterizes a
21. Long X, Wong CC, Tong L, et al. Peptostreptococcus subset of prostate cancers harboring TMPRSS2:ERG
anaerobius promotes colorectal carcinogenesis and fusion and PTEN deletion. Oncol Rep 2015;
modulates tumour immunity. Nat Microbiol 2019; 34:1211–1220.
4:2319–2330. 36. Kondo N, Takahashi A, Mori E, et al. DNA ligase IV as a
22. Tsoi H, Chu ESH, Zhang X, et al. Peptostreptococcus new molecular target for temozolomide. Biochem Bio-

GUT MICROBIOTA
anaerobius induces intracellular cholesterol biosynthesis phys Res Commun 2009;387:656–660.
in colon cells to induce proliferation and causes
dysplasia in mice. Gastroenterology 2017;
152:1419–1433.e5. Author names in bold designate shared co-first authorship.
23. Li Q, Hu W, Liu WX, et al. Streptococcus thermophilus in- Received March 15, 2022. Accepted October 2, 2023.
hibits colorectal tumorigenesis through secreting beta-
galactosidase. Gastroenterology 2021;160:1179–1193.e14. Correspondence
Address correspondence to Jun Yu, MD, PhD, Department of Medicine and
24. Arifuzzaman M, Won TH, Li TT, et al. Inulin fibre pro- Therapeutics, Faculty of Medicine, Prince of Wales Hospital, the Chinese
motes microbiota-derived bile acids and type 2 inflam- University of Hong Kong, Shatin, NT 00000, Hong Kong. e-mail:
[email protected]; or Yong Wang, PhD, Department of Laboratory Animal
mation. Nature 2022;611:578–584. Science, Army Medical University, Chongqing, China. e-mail:
25. Wirbel J, Pyl PT, Kartal E, et al. Meta-analysis of fecal [email protected].
metagenomes reveals global microbial signatures that Acknowledgments
are specific for colorectal cancer. Nat Med 2019; The authors thank Chuangen Li, Lihong Liu, and Minnie Y.Y. Go for their
25:679–689. support to the experiments. Patients and/or the public were not involved in
the design, conduct, reporting, or dissemination plans of this research.
26. Ou J, Carbonero F, Zoetendal EG, et al. Diet, microbiota,
and microbial metabolites in colon cancer risk in rural CRediT Authorship Contributions
Jia Yang, PhD (Formal analysis: Lead; Investigation: Lead; Methodology: Lead;
Africans and African Americans. Am J Clin Nutr 2013; Writing – original draft: Lead)
98:111–120. Hong Wei, PhD (Resources: Supporting)
27. Jia W, Xie G, Jia W. Bile acid-microbiota crosstalk in Yufeng Lin, Mphil (Formal analysis: Supporting; Methodology: Supporting)
Eagle SH Chu, MPhil (Investigation: Equal)
gastrointestinal inflammation and carcinogenesis. Nat Yunfei Zhou, Mphil (Investigation: Supporting; Methodology: Supporting)
Rev Gastroenterol Hepatol 2018;15:111–128. Hongyan Gou, PhD (Investigation: Supporting)
Shang Guo, Mphil (Investigation: Supporting)
28. Tremaroli V, Backhed F. Functional interactions between Harry CH Lau, PhD (Writing – review & editing: Supporting)
the gut microbiota and host metabolism. Nature 2012; Alvin HK Cheung, MBChB (Investigation: Supporting)
489:242–249. Huarong Chen, PhD (Investigation: Supporting)
Ka Fei To, MBChB (Investigation: Supporting)
29. Dodd D, Spitzer MH, Van Treuren W, et al. A gut bacterial Joseph JY Sung, MD, PhD (Commented on study: Lead)
pathway metabolizes aromatic amino acids into nine Yong Wang, PhD (Investigation: Supporting; Project administration: Lead;
Resources: Lead)
circulating metabolites. Nature 2017;551:648–652. Jun Yu, MD, PhD (Conceptualization: Lead; Funding acquisition: Lead;
30. Loftfield E, Falk RT, Sampson JN, et al. Prospective Project administration: Lead; Supervision: Lead; Writing – review & editing:
Lead)
associations of circulating bile acids and short-chain
fatty acids with incident colorectal cancer. JNCI Cancer Conflicts of interest
Spectr 2022;6:pkac027. The authors disclose no conflicts.

31. Singh V, Yeoh BS, Walker RE, et al. Microbiota Funding

fermentation-NLRP3 axis shapes the impact of dietary This project was supported by National Key R&D Program of China (No.
2020YFA0509200/2020YFA0509203), Shenzhen-Hong Kong-Macao Science
fibres on intestinal inflammation. Gut 2019;68:1801–1812. and Technology Program (Category C) Shenzhen
32. Nirmala Prasadi VP, Joye IJ. Dietary fibre from whole (SGDX20210823103535016), RGC Research Impact Fund Hong Kong
grains and their benefits on metabolic health. Nutrients (R4632–21F), RGC Theme-based Res Scheme Hong Kong (T21–705/20-N),
National Natural Science Foundation of China (NSFC; 81972576), RGC-CRF
2020;12:3045. Hong Kong (C4039–19GF and C7065–18GF), RGC-GRF Hong Kong
33. Roy HK, Smyrk TC, Koetsier J, et al. The transcriptional (14110819, 14111621), Vice-Chancellor’s Discretionary Fund CUHK.
repressor SNAIL is overexpressed in human colon can- Data Availability
cer. Dig Dis Sci 2005;50:42–46. Data are available within the article and its Supplementary Materials.
337.e1 Yang et al Gastroenterology Vol. 166, Iss. 2

Supplementary Methods chromatograph system coupled with an Agilent 5977B mass

spectrometer. The system used a DB-FFAP capillary column.
The energy was 70 eV in electron impact mode. Data were
acquired in Scan/SIM mode with a m/z range of 33 to 150
Orthotopic Mouse Model after a solvent delay of 2.5 minutes. SCFAs were identified
Murine CRC cell line MC38 (5  105 cells in 10 mL by comparing both MS spectra and retention times with
Matrigel [354248, Corning, Corning, NY] per mouse) was those of standard compounds. The peak area of each
orthotopically implanted into the rectum of male C57BL/6 derivatized SCFA was calculated.
mice at 8 weeks old. Three days after implantation, mice
were treated with sodium butyrate (0.5 mM [303410,
Immunohistochemistry Staining
Sigma-Aldrich, St Louis, MO]) by enema every other day.
Paraffin-embedded colon sections were used for
Mice were harvested at day 28.
analyzing the expression of Ki-67 with anti-Ki-67 primary
antibodies (1:100 dilution; ab16667, Abcam, Cambridge,
Shotgun Metagenomic Sequencing and MA). The proportion of Ki-67–positive cells was determined
Taxonomic Annotation by counting at least 1000 cells in 5 random microscopic
Mice fecal DNA was extracted using DNeasy PowerSoil Kit fields for each sample.
(QIAGEN, Valencia, CA). Shotgun metagenomic sequencing of
mice fecal DNA was performed on Illumina HiSeq 2000 Cell Culture
platform (Illumina, San Diego, CA) and quality controlled as
Human CRC cell lines Caco-2, HT29, and HCT116 were
previously described.e1 Taxonomy was assigned to meta-
obtained from American Type Culture Collection (Manassas,
genomic reads using k-mer–based algorithms implemented in
VA). Human normal colon epithelial cell line NCM460 was
Kraken taxonomic annotation pipeline. A standard database
obtained from INCELL Corporation (San Antonio, TX). All
comprising of 13,844 bacterial genomes from NCBI (http://
cells were grown in Dulbecco’s modified Eagle’s medium
www.ncbi.nlm.nih.gov) was built using Jellyfish program by
(DMEM) (Invitrogen, Carlsbad, CA) supplemented with 10%
counting distinct 31-mer in the reference libraries. Each k-
fetal bovine serum (Sigma-Aldrich) and 1% penicillin/
mer in a read was mapped to the lowest common ancestor of
streptomycin in a humidified atmosphere containing 5%
all reference genomes with the exact k-mer matches. Each
carbon dioxide.
query was thereafter classified to a taxon with the highest
total hits of k-mer matched by pruning the general taxonomic
Metabolites Treatment
trees affiliated with mapped genomes. The final metagenomic
Inosine was obtained from Bidepharm (Shanghai,
read counts were normalized by cumulative sum scaling
China). Cells were seeded (1000 cells/well) into a 96-well
method using metagenomeSeq R/Bioconductor package.
plate, and maintained in medium with or without supple-
mentation of inosine for up to 4 days.
Metabolomic Profiling Butyrate was obtained from Merck (Sigma-Aldrich).
A total of 50 mg of fecal sample was collected from each Cells were seeded (1000 cells/well) into a 96-well plate,
mouse for metabolomic profiling. For untargeted metab- and were treated with or without supplementation of
olomics, samples were first treated with 80% cold meth- butyrate with no fetal bovine serum for up to 4 days.
anol. After centrifugation at 21,500g for 15 min at 4 C, the
supernatant was collected for liquid chromatography–mass
Colony Formation
spectrometry (LC-MS) analysis. LC-MS was performed on
Cells (1000 cells/well) were seeded into a 6-well plate,
Agilent 1290 Infinity UPLC system (Agilent Technologies,
followed by treatment with Bifidobacterium pseudolongum
Santa Clara, CA) coupled to Sciex TripleTOF 6600 mass
conditioned medium (20% in DMEM). Escherichia coli
spectrometer (Q-TOF, AB Sciex, Toronto, Canada). Chro-
conditioned medium (20% in DMEM) was used as control.
matographic separation was achieved on Waters BEH
Treatment medium was changed every 3 days. After
Amide column (2.1 mm  100 mm, 1.7 mm). The mass
culturing for 14 days, cells were fixed by 70% ethanol and
spectrometer was operated in an information-dependent
stained with 0.5% crystal violet solution. Colonies with
acquisition mode. Data pretreatment including peak detec-
more than 50 cells were counted. The experiment was
tion and retention time correction was achieved by XCMS
conducted in triplicate per group.
and CAMERA packages implemented in R language.
Metabolite identification was based on an in-house MS2
database, Human Metabolome Database (www.hmdb.ca), Cell Proliferation
and METLIN metabolite database (metlin.scripps.edu). Cell viability was evaluated by 3-(4,5-dimethylthiazol- 2-
For targeted SCFA metabolomics, samples were first yl)-2,5-diphenyltetrazolium (MTT) assay. The amount of
treated with double-distilled water. After ultrasonic ho- MTT formazan product was determined by measuring
mogenization and centrifugation at 12,000 rpm for 15 min absorbance at a wavelength of 570 nm (OD570) with a
at 4 C, the supernatant was collected for gas microplate reader (Multiskan GO Microplate Spectropho-
chromatography–mass spectrometry (GC-MS) analysis. tometer, Thermo Scientific, Vantaa, Finland). The experi-
GC-MS was performed using an Agilent 7890B gas ment was conducted in 6 replicates per group.
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 337.e2

Western Blot by spectrophotometer and was reversely transcribed us-

Total protein was isolated and separated by sodium ing a complementary DNA conversion kit (Takara,
dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- Kusatsu, Japan). PCR array was performed using RT2
PAGE; 6%–12%). Proteins in SDS-PAGE was transferred Profiler PCR Array Mouse Cancer Pathway Finder (PAMM-
onto polyvinylidene difluoride membranes (EMD Millipore, 033Z, QIAGEN, Valencia, CA) in combination with SYBR
Billerica, MA) for 1 to 2 hours, which was then blocked with Green qPCR Master Mix (Takara) in ABI QuantStudio 7
10% bovine serum albumin in 0.05% Tris-based saline- Flex Real-Time PCR System (Invitrogen). Cycle threshold
Tween 20 for 2 hours at room temperature. The membrane values were recorded and normalized based on a panel of
was incubated with primary antibodies overnight at 4 C, reference genes. Fold changes were calculated using the
followed by secondary antibodies at room temperature for 1 delta-delta cycle threshold method. Genes with fold
hour. Protein band intensities were detected by ECL Plus changes more than 1.5 were considered as biologically
Western Blotting Detection Reagents (GE Healthcare, Wau- significant.
kesha, WI). Glyceraldehyde-3-phosphate dehydrogenase
(GAPDH) was used as housekeeping control.

RNA Extraction and PCR Array of Gene Supplementary Reference

Expression e1. Coker OO, Nakatsu G, Dai RZ, et al. Enteric fungal
Total RNA was extracted using TRIzol Reagent microbiota dysbiosis and ecological alterations in
(Thermo Fisher Scientific). RNA quality was determined colorectal cancer. Gut 2019;68:654–662.
337.e3 Yang et al Gastroenterology Vol. 166, Iss. 2

Supplementary Table 1.Diet Composition (Specialty Feeds Inc.)

Control High-fiber High-cellulose High-inulin High-guar

Diet formulas diet (CD) diet (HFiD) diet (HCD) diet (HID) gum diet (HGGD)

Ingredient g/kg g/kg g/kg g/kg g/kg

Casein (Acid) 200 200 200 200 200
Sucrose 100 100 100 100 100
Canola Oil 70 70 70 70 70
Cellulose (insoluble fiber) 50 200 400 50 50
Guar Gum (soluble fiber) 0 200 0 0 350
Inulin (soluble fiber) 0 0 0 350 0
Wheat Starch 404 54 54 54 54
Dextrinised Starch 132 132 132 132 132
L Methionine 3.0 3.0 3.0 3.0 3.0
Calcium Carbonate 13.1 13.1 13.1 13.1 13.1
Sodium Chloride 2.6 2.6 2.6 2.6 2.6
AIN 93 Trace Minerals 1.4 1.4 1.4 1.4 1.4
Potassiun Citrate 2.5 2.5 2.5 2.5 2.5
Potassium Dihydrogen Phosphate 6.9 6.9 6.9 6.9 6.9
Potassium Sulphate 1.6 1.6 1.6 1.6 1.6
Choline Chloride (75%) 2.5 2.5 2.5 2.5 2.5
AIN93 Vitamins 10 10 10 10 10
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 337.e4

Supplementary Table 2.Detailed Metabolites List for Supplementary Table 2. Continued

Heatmap of Differential Metabolites
Among Stool Samples of AOM- LSerine
treated CD-fed, HCD-fed, HID-fed,
Deoxycytidine
and HGGD-fed Mice (From Top to
Bottom) Nerolidol 3O[aLrhamnopyranosyl(1>2)bD
glucopyranos
N4Acetylaminobutanal
LCyclo(alanylglycyl)
3Acetamidobutanal
3Amino2piperidone
4Guanidinobutanoic acid
Indole3propionic acid
2Hydroxyethanesulfonate
NAcetylglucosamine 1phosphate
4Acetylimidazo[4,5c]pyridine
Cytidine monophosphate
LAspartic acid
Adenosine 20 phosphate
Glycylproline
Uridine 50 monophosphate
Taurocholic acid
20 Deoxyguanosine 50 monophosphate
Allocholic acid
Trimethadione
3Hydroxymethylglutaric acid
Glycine
Citreoviridinol A1
gammaGlutamylmethionine
ValylLysine
DAlanylDalanine
SM(d18:1/16:0)
gammaAminobutyric acid
ArginylGammaglutamate
30 AMP
MethionylValine
LCarnitine
AlanylProline
3Methyladenine
LeucylSerine
3Methylguanine
ThreoninylThreonine
Pirbuterol
LysoPC(18:1(11Z))
Cyclovariegatin
PC(20:3(8Z,11Z,14Z)/14:0)
LGlutamic acid
ArginylSerine
2(3,4dihydroxy5methoxyphenyl)3,5,7trihydroxy4H
ThreoninylSerine chrom
LysylValine dIMP
LMethionine Pyroglutamic acid
ArginylPhenylalanine Cincassiol B
Protein serine 1deoxy1(N6lysino)Dfructose
LeucylAspartate DAspartic acid
Taurine Malic acid
Phosphodimethylethanolamine Racemethionine
Methionine sulfoxide Capric acid
gammaGlutamylleucine Cytosine
N2gammaGlutamylglutamine Uridine
Glutamylaspartic acid 10E,12ZOctadecadienoic acid
LalphaAspartylLhydroxyproline Linoleic acid
Aminoadipic acid AlphaLinolenic acid
PI(22:5(4Z,7Z,10Z,13Z,16Z)/16:0) Petroselinic acid
Sarcosine 3Dehydrosphinganine
337.e5 Yang et al Gastroenterology Vol. 166, Iss. 2

Supplementary Table 2. Continued Supplementary Table 3.Complementary DNA Expression

Array Analysis on Colon Tissues of
Benzaldehyde Inosine-treated HID-FMT Mice
Compared With PBS-treated HID-
8Hydroxypinoresinol 4glucoside
FMT Mice
Oleic acid
Gene Fold change Pathway alignment
(13R,14R)8Labdene13,14,15triol
4Dodecylbenzenesulfonic Acid Acly 4.54 Metabolism

2,8DiOmethylellagic acid Birc3 2.56 Apoptosis

Malathion Arnt 2.50 Hypoxia signaling

Pipericine Sirt2 2.32 Telomeres and telomerase

Cortisol Ets2 2.08 Cellular senescence

4Hydroxybenzaldehyde Lpl 1.92 Metabolism

Benzoic acid Mki67 1.92 Cell cycle

Glycylalanylprolylmethionylphenylalanylvalinamide Xrcc4 1.65 DNA damage and repair

LysoPC(22:4(7Z,10Z,13Z,16Z)) Gadd45g 1.50 DNA damage and repair

LysoPC(20:4(5Z,8Z,11Z,14Z)) Ercc5 1.50 DNA damage and repair

Choline
HomoLarginine
LysoPE(18:1(9Z)/0:0)
Inosine
2(3,4Dihydroxybenzoyloxy)4,6dihydroxybenzoate
PC(20:5(5Z,8Z,11Z,14Z,17Z)/20:3(5Z,8Z,11Z))
Saccharin
Isohyodeoxycholic acid
Deoxycholic acid
Sclareol
Guanosine
February 2024 Soluble Fiber, Gut Microbiota, and Colorectal Cancer 337.e6

Supplementary Table 4.Diet Composition (Specialty Feeds Inc.)

Diet Formulas Control diet (CD) 5% inulin diet 10% inulin diet 15% inulin diet 20% inulin diet

Ingredient g/kg g/kg g/kg g/kg g/kg

Casein (Acid) 200 200 200 200 200
Sucrose 100 100 100 100 100
Canola Oil 70 70 70 70 70
Cellulose (insoluble fiber) 50 50 50 50 50
Guar Gum (soluble fiber) 0 0 0 0 0
Inulin (soluble fiber) 0 50 100 150 200
Wheat Starch 404 354 304 254 204
Dextrinised Starch 132 132 132 132 132
L Methionine 3.0 3.0 3.0 3.0 3.0
Calcium Carbonate 13.1 13.1 13.1 13.1 13.1
Sodium Chloride 2.6 2.6 2.6 2.6 2.6
AIN 93 Trace Minerals 1.4 1.4 1.4 1.4 1.4
Potassiun Citrate 2.5 2.5 2.5 2.5 2.5
Potassium Dihydrogen Phosphate 6.9 6.9 6.9 6.9 6.9
Potassium Sulphate 1.6 1.6 1.6 1.6 1.6
Choline Chloride (75%) 2.5 2.5 2.5 2.5 2.5
AIN93 Vitamins 10 10 10 10 10
337.e7 Yang et al Gastroenterology Vol. 166, Iss. 2

Supplementary Table 5.Complementary DNA Expression

Array Analysis on Colon Tissues of
HID-FMT Mice Compared With CD-
FMT Mice

Gene Fold change Pathway alignment

Snai3 3.39 EMT

Wee1 2.69 Cell proliferation
Pfkl 2.39 Metabolism
Ocln 2.27 EMT
Bcl2l11 2.16 Apoptosis
Map2k1 2.09 Cellular senescence
G6pdx 2.05 Metabolism
Lig4 2.04 DNA damage and repair
Tbx2 1.96 Cellular senescence
Ing1 1.95 Cellular senescence
Kdr 1.93 Angiogenesis
Cpt2 1.92 Metabolism
Ercc3 1.91 DNA damage and repair
Casp2 1.87 Apoptosis
Ldha 1.83 Hypoxia
Uqcrfs1 1.81 Metabolism
Casp7 1.75 Apoptosis
Tep1 1.73 Telomeres and telomerase
Ddb2 1.72 DNA damage and repair
Gpd2 1.71 Metabolism
Birc3 1.70 Apoptosis
Igfbp3 1.68 Cellular senescence
Atp5a1 1.57 Metabolism
Serpinb2 1.55 Cellular senescence
Sirt1 1.54 Cellular senescence