dietary exposure to antibiotic residues facilitates
TRANSCRIPT
Dietary Exposure to Antibiotic Residues FacilitatesMetabolic Disorder by Altering the Gut Microbiotaand Bile Acid CompositionRou-An Chen
Institute of Food Science and Technology, National Taiwan UniversityWei-Kai Wu
Department of Medical Research, National Taiwan University Hospital https://orcid.org/0000-0002-9476-8998Suraphan Panyod
Institute of Food Science and Technology, National Taiwan UniversityPo-Yu Liu
National Taiwan University https://orcid.org/0000-0003-1290-0850Hsiao-Li Chuang
National Laboratory Animal Center, National Applied Research LaboratoriesYi-Hsun Chen
Department of Internal Medicine, College of Medicine, National Taiwan UniversityQiang Lyu
Department of Chemistry, National Taiwan UniversityHsiu-Ching Hsu
School of Pharmacy, College of Medicine, National Taiwan UniversityTzu-Lung Lin
Department of Medical Biotechnology and Laboratory Science, College of Medicine, Chang GungUniversityTing-Chin Shen
Division of Gastroenterology, Perelman School of Medicine, University of PennsylvaniaYu-Tang Yang
Department of Internal Medicine, College of Medicine, National Taiwan UniversityHsin-Bai Zou
Department of Chemistry, National Taiwan UniversityHuai-Syuan Huang
Institute of Food Science and Technology, National Taiwan UniversityYu-En Lin
Institute of Food Science and Technology, National Taiwan UniversityChieh-Chang Chen
Department of Internal Medicine, National Taiwan University Hospital https://orcid.org/0000-0003-2953-210XChi-Tang Ho
Department of Food Science, Rutgers UniversityHsin-Chih Lai
Department of Medical Biotechnology and Laboratory Science, College of Medicine, Chang GungUniversityMing-Shiang Wu
National Taiwan University Hospital https://orcid.org/0000-0002-1940-6428Cheng-Chih Hsu
Department of Chemistry, National Taiwan University https://orcid.org/0000-0002-2892-5326Lee-Yan Sheen ( [email protected] )
National Taiwan University
Letter
Keywords: antibiotic residue, metabolic dysfunction, tylosin, gut microbiota
Posted Date: November 18th, 2021
DOI: https://doi.org/10.21203/rs.3.rs-1066621/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License
1
Dietary Exposure to Antibiotic Residues Facilitates Metabolic Disorder by 1
Altering the Gut Microbiota and Bile Acid Composition 2
Rou-An Chena†, Wei-Kai Wub†, Suraphan Panyoda, Po-Yu Liuc, Hsiao-Li Chuangd, Yi-3
Hsun Chenc, Qiang Lyue, Hsiu-Ching Hsuf, Tzu-Lung Ling, Ting-Chin David Shenh, 4
Yu-Tang Yangc, Hsin-Bai Zoue, Huai-Syuan Huanga, Yu-En Lina, Chieh-Chang Cheni, 5
Chi-Tang Hoj, Hsin-Chih Laig, Ming-Shiang Wuc,i, Cheng-Chih Hsue* and Lee-Yan 6
Sheena,k,l* 7
aInstitute of Food Science and Technology, National Taiwan University, Taipei, 8
Taiwan; bDepartment of Medical Research, National Taiwan University Hospital, 9
Taipei, Taiwan; cDepartment of Internal Medicine, College of Medicine, National 10
Taiwan University, Taipei, Taiwan; dNational Laboratory Animal Center, National 11
Applied Research Laboratories, Taipei, Taiwan; eDepartment of Chemistry, National 12
Taiwan University, Taipei, Taiwan; fThe Metabolomics Core Laboratory, Center of 13
Genomic Medicine, National Taiwan University, Taipei, Taiwan; gDepartment of 14
Medical Biotechnology and Laboratory Science, College of Medicine, Chang Gung 15
University, Taoyuan, Taiwan; hDivision of Gastroenterology, Perelman School of 16
Medicine, University of Pennsylvania, Pennsylvania, USA; iDepartment of Internal 17
Medicine, National Taiwan University Hospital, Taipei, Taiwan; jDepartment of Food 18
Science, Rutgers University, New Brunswick, New Jersey, USA; kCenter for Food and 19
Biomolecules, National Taiwan University, Taipei, Taiwan; lNational Center for Food 20
Safety Education and Research, National Taiwan University, Taipei, Taiwan 21
22
†There authors contributed equally for the study. 23
*Corresponding authors: 24
Lee-Yan Sheen, No. 1, Sec. 4, Roosevelt Rd., Taipei, Taiwan 10617; Tel: 886-2-25
33661572; E-mail: [email protected] 26
Cheng-Chih Hsu, No. 1, Sec. 4, Roosevelt Rd., Taipei, Taiwan 10617; Tel.: 886-27
33661681; E-mail: [email protected] 28
29
2
Dietary Exposure to Antibiotic Residues Facilitates Metabolic Disorder by 30
Altering the Gut Microbiota and Bile Acid Composition 31
Low dose antibiotic residues in food potentially contribute to obesity and metabolic 32
dysfunction1. However, the effect of chronic exposure to very low-dose antibiotic 33
residue (~1000-fold lower than the therapeutic dose) on gut microbiota and host 34
metabolism is poorly understood. Herein the effect of exposure to a residual dose 35
of tylosin—an antibiotic growth promoter—on host metabolism and gut 36
microbiota is explored in a mouse model. Theoretical maximal daily intake 37
(TMDI) dose of tylosin facilitates high-fat diet-induced obesity, induces insulin 38
resistance, and perturbs gut microbiota composition. Moreover, obesity-related 39
phenotypes are transferrable to germ-free recipient mice following fecal 40
microbiota transplantation. Tylosin TMDI exposure restricted to early life is 41
sufficient to induce metabolic complications, alter the abundance of specific 42
bacteria related to host metabolic homeostasis later in life, and modify the 43
composition of short-chain fatty acids and bile acids. Finally, tylosin TMDI 44
exposure induces lasting metabolic consequences via elevating the ratio of primary 45
to secondary bile acids and its downstream FGF15 signaling pathway. Hence, 46
exposure to very low doses of antibiotic residues, whether continuously or in early 47
life, can exert long-lasting effects on host metabolism by altering gut microbiota 48
and their metabolites. 49
Antibiotics used as growth promoters in livestock and animal husbandry can be 50
detected in animal-derived food2-5. Although the maximum residue limits (MRLs) of 51
these veterinary antibiotics has been evaluated, these estimated levels do not consider 52
the effects of antibiotic residues on gut microbiota and microbial metabolites, or the 53
effects of altered gut microbiota on host metabolism6. Recent studies have established 54
3
an association between exposure to veterinary antibiotics and children obesity1,7. 55
Additionally, animal studies have indicated that exposure to sub-therapeutic doses of 56
antibiotics leads to obesity and NAFLD8-10. In this study, we aimed to further elucidate 57
whether chronic exposure to acceptable residual amounts of antibiotic, which doses are 58
hundred-fold lower than the sub-therapeutic doses used in previous studies, could cause 59
obesity complications. Thus, C57BL6J mice were fed an acceptable daily intake (ADI, 60
0.37 mg/kg/day) and theoretical maximum daily intake (TMDI, 0.047 mg/kg/day) dose 61
of tylosin (Fig. 1a). The ADI dose is the maximum daily dose with no adverse effect6, 62
while the TMDI dose is an estimate of maximum dietary intake level obtained for 63
MRLs and the sum of average daily per capita consumption for each food product6, 64
which was used for simulating the ingestion of antibiotic residues through food 65
consumption in the present study. Additionally, a normal-chow diet and high-fat diet 66
were provided to assess the potential synergistic effects of antibiotics and diet on host 67
metabolism. 68
Compared with NCD-CON (normal chow diet/control) mice, NCD-ADI and 69
NCD-TMDI mice showed significantly greater weight gain from weaning to 5 weeks of 70
age. NCD-ADI mice also exhibited higher weight gain from weeks 13 to 15 (Fig. 1b) 71
with no significant changes detected among the NCD groups in relative fat or lean mass 72
(Extended Data Fig. 1a, b). In contrast, high-fat diet (HFD)-ADI and HFD-TMDI 73
mice exhibited significantly increased weight gain compared with HFD-CON mice 74
from weaning to 17 weeks of age (Fig. 1c). HFD-ADI and HFD-TMDI mice had 75
increased relative fat mass at weeks 8, 12, and 17, compared with HFD-CON mice (Fig. 76
1d), whereas HFD-TMDI mice had decreased relative lean mass (Fig. 1e). Likewise, 77
HFD-TMDI mice had increased visceral fat, including epididymal and perinephric 78
adipose tissues (Fig. 1f, g and Extended Data Fig. 1 c, d). HFD-ADI and HFD-TMDI 79
4
mice also exhibited adipocyte hypertrophy (Fig. 1h) and elevated adipocyte size (Fig. 80
1j) compared with HFD-CON mice. Hence, tylosin-induced adiposity is evident early in 81
life for NCD and HFD mice; however, the continuous effects of tylosin-induced 82
adiposity were preferentially observed in HFD-fed mice. 83
We next sought to investigate whether exposure to residual tylosin dose is 84
causally associated with metabolic complications, including hepatic abnormalities and 85
glucose homeostasis. Given that tylosin increased the fat mass in HFD-fed mice, not 86
NCD-fed mice, additional investigations were conducted in the HFD-fed mice. 87
Histological examination of the liver revealed that tylosin-treated mice exhibited 88
increased lipid droplet formation (Fig. 1i), more inflammatory foci, and higher fatty 89
liver scores (Fig. 1k), suggesting that the presence of residual tylosin caused more 90
severe NAFLD. Both HFD-ADI and HFD-TMDI mice exhibited a trend toward higher 91
plasma glucose during the oral glucose tolerance test (OGTT) (Extended Data Fig. 2a) 92
and fasting glucose test (Extended Data Fig. 2b), with an increased OGTTAUC found in 93
HFD-TMDI mice (Fig 2l). The fasting insulin level and homeostasis model assessment 94
of insulin resistance (HOMA-IR) index were also elevated in HFD-ADI mice, 95
compared with HFD-CON mice (Fig. 2m, n). Hence, a residual dose of tylosin can 96
exacerbate HFD-induced adverse effects on metabolic complications. Remarkably, 97
although the tylosin ADI dose is defined as the maximum antibiotic dose that can be 98
ingested daily without health risk6, it facilitated obesity-related metabolic disorders in 99
our mice model. As such, the tolerance level of antibiotics may need to be re-evaluated 100
in consideration of chronic metabolic diseases which were emerging in western 101
lifestyle. 102
As sub-therapeutic doses of antibiotic treatments have been shown to disrupt the 103
development and maturation of gut microbiota with metabolic consequences8,11, we next 104
5
asked whether ADI and TMDI doses of tylosin alter gut microbiota composition. To 105
this end, fecal 16S rRNA sequencing was performed in mice at 3 (weaning), 8, and 17 106
weeks of age to elucidate changes in the gut microbiota. Compared with HFD-TMDI 107
and HFD-CON mice, HFD-ADI mice had significantly reduced Shannon indices at 108
week 3, which markedly increased during weeks 3–17 (Fig. 2a). Principal coordinate 109
analysis (PCoA) based on Bray-Curtis dissimilarity further revealed that the 110
microbiomes of NCD and HFD mice clustered separately (Extended Data Fig. 3a), 111
indicating that diet may represent the most critical factor influencing the composition of 112
the gut microbiota, while residual tylosin dose exhibited a secondary effect in shifting 113
the gut microbiota in a dose-dependent manner according to the PCoA2 axis (Extended 114
Data Fig. 3a). 115
Further, we analyzed the subsets of NCD (Extended Data Fig. 3b) and HFD-116
fed mice (Fig. 2b) separately to observe the effect of tylosin on age-related microbiota 117
development. In both groups, ADI dose of tylosin substantially influenced the gut 118
microbiota composition, while TMDI dose of tylosin showed a lesser impact. Notably, 119
the gut microbiota of tylosin-treated mice was obviously changed at the early-life stage 120
(week 3) in both NCD and HFD groups, with gradually shifted toward the control group 121
at weeks 8 and 17, respectively. Bray-Curtis distances from HFD-CON mice to HFD-122
ADI or HFD-TMDI mice were also shortened as the mice aged (Fig. 2c). These 123
findings suggest that gut microbiota has increased susceptibility to the perturbation 124
caused by residual tylosin in early life, yet susceptibility may decrease at later stages, 125
possibly due to the establishment and maturation of gut microbiota. 126
To further investigate differentially abundant bacteria modified by tylosin TMDI 127
dose, we identified significantly-altered bacteria at the amplicon sequence variants 128
(ASV) level (defined as log2-fold change >1, p values < 0.05; Fig. 2d, Extended Data 129
6
Fig. 4). Oscillibacter, Blautia and Parasutterella, which are associated with obesity and 130
inflammatory bowel disease12-14, were significantly enriched in HFD-TMDI mice, 131
whereas Bifidobacterium was significantly reduced (Fig. 2d). TMDI dose of tylosin 132
may decrease beneficial bacteria while enriching pathogenic bacteria. 133
We next assessed the association between obesity-related features of 17-week-134
old HFD-fed mice and their gut microbiota composition by the envfit function in the 135
vegan R package, the result is displayed in PCoA (Fig. 2e). The results showed that 136
vectors indicating both obesity-related biomarkers (the blue arrows) and antibiotic 137
exposure (the red arrow) were arrowed toward the mice treated with tylosin in the same 138
direction, indicating the obesity outcomes were associated with the tylosin-shifted gut 139
microbiome. We further performed Spearman's correlation between PCoA1 and PCoA2 140
of gut microbiota against individual obesity-related parameters (Supplementary Table. 141
1) and found that the PCoA2 was significantly correlated with obesogenic phenotypes 142
(fat gain, fat mass, and relative fat mass), fatty liver score, and insulin resistance 143
parameters (fasting insulin and HOMA-IR). 144
To understand the overall obesity phenotype and its relationship with tylosin 145
altered-microbiome, we performed a dimension reduction of the obesity and metabolic 146
disorders related biomarkers by using the principal component analysis (PCA). The 147
result showed that the obesity phenotypes in tylosin-treated mice were significantly 148
different from control mice (p < 0.0001; Extended data Fig. 5) We subsequently 149
calculated a correlation between PC1obesity biomarkers and the PCoA2microbiota which 150
displaying a positive correlation (r = 0.71, p = 0.003). 151
Given the above association between the metabolic disorder phenotypes and 152
tylosin-altered gut microbiota, we conducted a fecal microbiota transplantation (FMT) 153
study to investigate their causative role. Since a TMDI dose can adequately simulate 154
7
human exposure to antibiotics in food, feces from HFD-TMDI mice were transplanted 155
to germ-free mice. Compared with germ-free mice that received feces from HFD-CON 156
mice (FMT-CON), the HFD-TMDI recipient mice (FMT-TMDI) showed higher body 157
weight at 11, 12, 13, 14 weeks of age (Fig. 2g), increased weight gain 2 weeks post-158
FMT (Fig. 2h), and slightly elevated fat mass at week 20 (Fig. 2i), implying that the 159
microbiome from HFD-TMDI mice increased the adiposity of the recipient mice. 160
Additionally, HFD-TMDI mice exhibited increased plasma glucose levels in OGTTAUC, 161
and a higher HOMA-IR index (Fig. 2j, k). Thus, a TMDI dose of tylosin-altered 162
microbiota induced obesity and insulin resistance in germ-free recipients, indicating that 163
residual tylosin exposure facilitates metabolic disorders through the alternation of gut 164
microbiota. 165
Growing evidence indicates that antibiotic exposure during infancy, which has 166
been considered the critical window of gut microbiota development8, is associated with 167
increased risk of being overweight and obese15,16. More specifically, antibiotic exposure 168
during early life, even at sub-therapeutic levels, disturbs the colonization and maturation 169
of the intestinal microbiota, leading to lasting effects on the metabolism of the host8,17. 170
Accordingly, an early-exposure experiment was conducted to investigate the influence 171
of early-life exposure to a TMDI dose of tylosin on obesity-related phenotypes and the 172
gut microbiota (Fig. 3a). 173
Cont-TMDI mice, which were continuously exposed to TMDI dose of tylosin 174
throughout the experimental period, displayed continuously elevated body weight, 175
relative fat mass, visceral fat mass, and OGTTAUC, compared with HFD-CON mice 176
(Fig. 3b-e). Interestingly, Early-TMDI mice, which were exposed to tylosin TMDI 177
during pregnancy and the nursing period, also exhibited consistently elevated body 178
weight, relative fat mass, fasting insulin, and HOMA-IR index values after cessation of 179
8
tylosin exposure (Fig. 3b-g). These findings suggest that exposure to antibiotic residue 180
in food products during early life is sufficient to induce long-lasting effects on 181
metabolism and lead to obesity. Compared with previous studies showing that antibiotic 182
exposure limited to early life induces elevated adiposity later in life8, our findings 183
further demonstrate that metabolic complications can emerge later in life even with a 184
food-grade low dose of antibiotic residue. 185
The overall gut microbiota composition based on PCoA of Bray-Curtis distances 186
showed that tylosin influenced the gut microbiota composition at weeks 5 and 20 (p < 187
0.05; Fig. 3h). Compared with CON mice, Cont-TMDI mice showed a greater 188
difference than Early-TMDI mice, suggesting that continuous tylosin exposure has a 189
more significant effect compared to exposure restricted to early life (Fig. 3h). 190
Furthermore, the Early-TMDI and CON mice exhibited greater differences at week 5 191
than week 20, possibly due to establishment of the gut microbiota (Fig. 3h). 192
Interestingly, despite Early-TMDI mice exhibited a more similarly resembled microbial 193
community to that of CON mice at week 20 than at week 5 (Fig. 3h), their fat mass at 194
week 20 was more significantly increased then at week 5 (Fig. 3c). This finding 195
suggests that the impaired metabolic phenotype can persist despite resilience of the 196
microbiota. That is, minor disruption of the microbiota in early life caused by antibiotic 197
residue appears to be sufficient for inducing significant adiposity18,19. 198
Studies have indicated that microbial colonization can be perturbed by exposure 199
to antibiotic early in life, which, in turn, contributes to metabolic disorders in 200
adulthood20,21. Thus, we proposed that although the overall gut microbiota composition 201
can restore in later life, the abundance of specific bacteria associated with metabolic 202
homeostasis of the host are persistently depleted. To investigate this hypothesis, we 203
identified 32 bacterial genera significantly altered by early or continuous exposure to 204
9
TMDI dose of tylosin (Fig. 3i), then examined the association between these bacterial 205
genera and obesity-related phenotypes (Fig. 3j). Bacterial genera increased in both 206
Early-TMDI and Cont-TMDI mice, including Anaerofustis, and demonstrated a 207
significant positive correlation with obesity-related phenotypes (Fig. 3i). In contrast, 208
genera that were depleted in Early-TMDI and Cont-TMDI mice, including bacteria 209
belonging to the Lachnospiraceae and Ruminococcaceae families, exhibited a 210
significantly negative correlation with obesity-related phenotypes (Fig. 3i). Previous 211
studies have also reported that these bacteria are related to obesity. Anaerofustis, the 212
tylosin-enriched bacterium is increased in obese humans22, while tylosin-depleted 213
Ruminococcaceae and Lachnospiraceae are associated with lower long-term weight 214
gain23. Notably, despite the discontinuation of tylosin exposure in Early-TMDI mice at 215
3 weeks of age, several bacterial genera that are negatively correlated with obesity-216
related phenotypes remained diminished at week 20. Consistent with our hypothesis, 217
although early exposure to residual dose of antibiotics may not cause a significant 218
impact on the overall microbial composition in later life, early-depletion of certain 219
bacteria, that continued to be lifelong lost, may contribute to metabolic dysfunctions in 220
later life. 221
Given that antibiotics reportedly alter short-chain fatty acid (SCFA), bile acids 222
composition and their signaling pathways, thereby leading to metabolic 223
consequences24,25, we next investigated the effect of TMDI dose of tylosin on these two 224
major metabolites. SCFA analysis showed that propionic acid and butyric acid, two 225
main SCFAs associated with enhanced intestinal barrier function and insulin 226
sensitivity26,27, exhibited decreasing trends in Cont-TMDI mice (Fig. 4a). Isovaleric 227
acid, a branched SCFA reported to improve insulin-stimulated glucose uptake and 228
enhance insulin sensitivity28, was significantly decreased in Cont-TMDI mice (Fig. 4a). 229
10
Reduction of SCFAs could be attributed to the reduction of Lachnospiraceae and 230
Ruminococcaceae in Early-TMDI and Cont-TMDI mice (Fig. 3i)29,30. 231
The total amount of bile acids and the ratio of conjugated to unconjugated bile 232
acids were not significantly altered by tylosin exposure (Extended Data Fig. 6a, b). 233
Notably, the ratio of primary bile acids (PBA) to secondary bile acids (SBA) was 234
significantly increased in Cont-TMDI and Early-TMDI mice (Fig. 4b). Indeed, a 235
decreased PBA/SBA ratio in NAFLD patients who have undergone bariatric surgery is 236
associated with improved insulin sensitivity, indicating that an increased PBA/SBA 237
ratio tends to induce metabolic disorder31,32. To further investigate the effect of tylosin 238
on the conversion of PBA to SBA, the detected bile acids were classified into non-12-239
OH bile acids (Fig. 4c), muricholic acids (MCA; Fig. 4d), and 12-OH bile acids (Fig. 240
4e) based on their metabolic pathway33. The levels of PBAs, namely, chenodeoxycholic 241
acid, α-MCA, 𝛽-MCA, and cholic acid (Fig. 4c-e) were significantly increased, whereas 242
those of the SBAs, including ursodeoxycholic acid (UDCA) and ω-MCA (Fig. 4c, d), 243
were decreased in tylosin-treated mice, possibly due to inhibition of bacteria associated 244
with epimerization and dehydroxylation of PBAs, such as Clostridia, 245
Peptostreptococcus34 (Extended Data Fig. 6c, d). 246
Antibiotics have also been found to increase the PBA/SBA ratio with a 247
subsequent decrease in plasma fibroblast growth factor 19 (FGF19; human orthologue 248
of FGF15). FGF19 is an insulin-like hormone secreted by intestinal epithelium to 249
regulate hepatic lipid and glucose metabolism by binding to FGFR4 in the liver, thereby 250
decreasing peripheral insulin sensitivity24,35,36. Moreover, Parasutterella, the genus 251
enriched in tylosin-treated mice (Fig. 2e and Extended Data Fig. 4a), reportedly 252
increases 𝛽-MCA and decreases ileal Fgf15 gene expression13. Hence, the 253
FGF15/FGFR4 signaling pathway was further explored. Tylosin reduced ileal FGF15 254
11
expression (Fig. 4g) and portal vein FGF15 levels (Fig. 4f) with subsequent reduction in 255
hepatic FGFR4 levels (Fig. 4h). Collectively, tylosin TMDI-treated mice showed an 256
increased PBA/SBA ratio, as well as lower FGF15 levels in the ileum and portal vein, 257
and decreased expression of hepatic FGFR4, which may cause metabolic disorders by 258
affecting metabolism-related signaling pathways in the liver35-37. 259
This study has certain limitations. First, we only investigated the in vivo effects 260
of one antibiotic (tylosin), whereas exposure to other antibiotics may lead to different 261
outcomes owing to the specific antimicrobial action and spectrum of each antibiotic. 262
Second, the human dietary pattern is dynamic, with daily exposure to multiple types of 263
antibiotics, and even pesticides. Future research is warranted to investigate the effects of 264
other antibiotics at residual amounts, as well as the combination of different antibiotics 265
to better reflect real-life conditions. 266
In conclusion (Fig. 4i), tylosin at ADI and TMDI doses, which are generally 267
regarded as harmless, facilitated HFD-induced metabolic complications and perturbated 268
the gut microbiota composition. Moreover, the altered gut microbiota was critical for 269
tylosin TMDI-induced metabolic consequences, suggesting that continuous exposure to 270
very low doses of antibiotic residues in food can affect human metabolism by altering 271
gut microbiota. Since the gut microbial community is dynamic and susceptible to 272
environmental shifts in early life, early exposure to residual doses of tylosin was 273
sufficient to persistently deplete specific bacteria involved in the metabolic homeostasis 274
of the host, which may induce lasting metabolic consequences. Finally, we 275
demonstrated that residual tylosin altered the conversion of bile acids with downstream 276
effects on the FGF15 signaling pathway, a possible mechanism that participates in the 277
lasting metabolic disorders. Taken together, these findings indicate that permissible 278
12
exposure levels of antibiotic residues should be re-established while considering its 279
impact on the gut microbiota, for which this study provides valuable insights. 280
281
13
Materials and Methods 282
Antibiotic selection and dose calculation 283
Tylosin was selected as a model antibiotic growth promoter due to its high annual 284
consumption as a veterinary antibiotic (Bureau of Animal and Plant Health Inspection 285
and Quarantine, 2014) and its presence in animal-derived food38-40. The ADI and TMDI 286
doses were obtained from the World Health Organization Technical Report Series41. 287
Experimental design of the animal studies 288
Animal experiments were performed with permission from the Institutional Animal 289
Care and Use Committee of National Taiwan University (approval number: NTU-106-290
EL-051and NALC 107-0-006-R2). All mice were purchased from the National 291
Laboratory Animal Center (Taipei City, Taiwan). For animal studies Ⅰ and Ⅲ, each 292
experimental group comprised 3 C57BL/6J mother mice. The number of mice in each 293
experimental group varied according to the number of offspring of the mother mice. 294
Ⅰ. Antibiotic residue exposure model 295
C57BL/6J mother mice received tylosin at ADI (0.37 mg/kg) and TMDI (0.047 mg/kg) 296
doses 10 days before giving birth, and their offspring were continuously administered 297
the antibiotic. Tylosin was administered through drinking water. Control mice (CON) 298
did not receive antibiotics. To investigate the syngenetic effect of tylosin and diet, the 299
offspring were randomly divided into NCD (MFG, Oriental Yeast Co., Ltd., Tokyo, 300
Japan) and HFD (60% kcal from fat) groups (D12492, Research Diets, New Brunswick, 301
NJ, USA). Body composition was measured at 8, 12, and 17 weeks. The metabolic 302
measurements and OGTT were performed at 20 weeks prior to sacrifice. After 303
euthanasia by carbon dioxide inhalation, blood and tissue samples were collected and 304
stored at −80 ℃. 305
Ⅱ. Fecal microbial transplantation study 306
14
Eight-week-old C57BL/6 germ-free mice were randomly divided into fecal microbiota 307
transplantation-control (FMT-CON) and FMT-TMDI groups, which were transplanted 308
with fecal microbiota from HFD-CON and HFD-TMDI mice, respectively. After FMT, 309
the recipient mice were housed in two independent isolators and fed with irradiated 310
HFD until 20 weeks of age. Before the recipient mice were euthanized, the body 311
composition analysis and OGTT were performed. 312
Ⅲ. Early-life exposure model 313
The mice were divided into three groups: CON, feeding conditions were the same as 314
those of HFD-CON; Early-TMDI, exposure duration was limited to the gestation and 315
lactation period; Cont-TMDI, feeding conditions were the same as those of HFD-TMDI. 316
TMDI administration of tylosin started on day 10 of gestation. Before the mice were 317
weaned, both Early-TMDI and Cont-TMDI mice were exposed to a TMDI dose of 318
tylosin. After weaning, only the Cont-TMDI mice were continuously exposed to tylosin 319
through drinking water. Body composition was measured at 5, 10, 15, and 20 weeks. 320
The OGTT was performed before euthanizing the mice at 20 weeks. 321
Body composition analysis 322
Body composition was determined using Minispec LF50 TD-NMR Body Composition 323
Analyzer (Bruker, Billerica, MA, USA), which provides the measurement of body 324
weight and lean and fat mass. The relative fat mass was calculated as the fat mass 325
(g)/body weight (g) ratio. 326
Glucose and insulin sensitivity 327
For the OGTT, mice were deprived of food for 5 h. Blood was collected from the 328
submandibular vein, and the glucose levels were measured using a glucometer (Roche, 329
Basel, Switzerland) at 0, 15, 30, 60, 90, and 120 min after oral administration with 2 330
g/kg glucose. The fasting insulin levels were detected by an enzyme-linked 331
15
immunosorbent assay (ELISA) kit (Mercodia, Uppsala, Sweden). The HOMA-IR index 332
was calculated using the formula: fasting glucose (nmol/L) × fasting insulin 333
(µU/mL)/22.5 42. 334
Histopathological analysis of liver and adipose tissue 335
Liver and adipose tissue sections were collected and fixed in 10% formalin solution. 336
Histopathological analysis was performed by the formalin-fixed, paraffin-embedded, 337
and hematoxylin and eosin (H&E)-stained slide method. The fatty liver score was 338
estimated by a pathologist as previously described43. The fatty liver score included the 339
evaluation of steatosis (macrovesicular, microvesicular, and hypertrophy) and 340
inflammation (number of inflammatory foci). The visceral adipocyte quantification was 341
performed by HCImage Live software (HCImage, Sewickley, PA, USA). 342
Quantification of plasma lipopolysaccharides 343
Plasma samples were centrifuged and added to HEK-Blue™ mTLR4 cell lines in the 344
HEK-Blue™ Detection medium (InvivoGen, San Diego, CA, USA). After 24 h of 345
incubation, the color of secreted embryonic alkaline phosphatase (SEAP) released by 346
the reporter cells was measured by a spectrophotometer at 620 nm44. 347
Metabolic measurements 348
Metabolic measurements (food intake, locomotor activity, VO2 consumption, and VCO2 349
production) were made using the Promethion metabolic phenotyping system (Sable 350
Systems, Las Vegas, NV, USA). Monitoring was performed for 24–48 h with ad libitum 351
access to food and water after mice had been acclimatized to cages for 6–12 h. 352
Fecal microbiota extraction and 16S rRNA gene sequencing 353
The mice in each group were randomly selected for fecal microbiota analysis. Mice 354
feces were collected and frozen immediately at −80 °C. Fecal DNA extraction and 355
library preparation and sequencing were performed according to the protocol provided 356
16
by QIAGEN (QIAGEN, MD, USA). Briefly, fecal microbial DNA was isolated by 357
using QIAamp 96 PowerFecal QIAcube HT Kit (QIAGEN, Germantown, MD, USA). 358
The 16S V4 region was amplified by the forward primer (F515, 5ʹ-GTGCCAGCMG 359
CCGCGGTAA-3ʹ) and the reverse primer (R806, 5ʹ-GGACTACHVGGGTWTCTAAT-360
3ʹ). The reaction conditions were as follows: 3 min at 95 °C, followed by 25 cycles of 361
95 °C for 30 s, 55 °C for 30 s, 72 °C for 30 s, and 5 min at 72 °C for a final extension. 362
For library construction, the amplified PCR product was attached with Illumina 363
sequencing adapters by Nextera XT Index Kit then purified using AMPure XP beads. 364
The library quantification was conducted using the DNA 1000 kit and 2100 Bioanalyzer 365
Instrument (Agilent Technologies, Santa Clara, CA, USA). The sequencing (paired-end 366
reads, 2 × 150 bp) was performed by Illumina NextSeq (Illumina, San Diego, CA, 367
USA). 368
Bioinformatic analysis for microbial taxonomic profiling 369
The amplicon sequences were processed by a QIIME 2 pipeline (version 2019.10, 370
https://qiime2.org). The primer sequences of raw reads were trimmed using the cutadapt 371
plugin. The trimmed single-end (forward) sequences were subsequently denoised with 372
the DADA2 plugin of QIIME245. To obtain qualified data, we truncated reads to 130 bp 373
from the 3ʹ end based on the quality score. The DADA2 outputs of high confident ASVs 374
went through quality filtering, denoising, and chimeric-read removing. The classify-375
consensus-vsearch plugin was applied for taxonomy assignment by aligning against 376
SILVA 132 99% 16S rDNA sequences, and identity cutoff was set at ≥80% sequence 377
similarity by default. The vegan package in R was used to calculate Shannon index and 378
perform principal coordinate analysis (PCoA) based on the Bray–Curtis distance46. The 379
permutation multivariate analysis of variance (ANOVA) using distance matrixes 380
(adonis) was performed to calculate the significance of microbiota composition among 381
17
groups. The PCoA plot of gut microbiota with an association between obesity-related 382
features of 17-week-old HFD-fed mice was performed by the envfit function in the 383
vegan R package, displaying as the vectors (p < 0.05). 384
Correlation analysis 385
The Spearman’s correlation analysis between PCoA1 and PCoA2 of gut microbiota 386
against individual obesity-related parameters was performed with FDR-adjusted p-387
value. The correlation coefficient and p-value of Spearman’s correlation between 388
PC1obesity biomarkers (representing the overall obesity phenotype) and PCoA2microbiota 389
(representing the gut microbiota composition) was calculated. 390
Principal component analysis (PCA) of obesity biomarkers 391
PCA was used for dimension reduction of obesity biomarkers10, including body weight 392
at 17 weeks of age, fat mass at 17 weeks of age, relative fat mass at 17 weeks of age, 393
weight gain between 3 to 17 weeks of age, fat mass gain between 8 to 17 weeks of age, 394
fasting glucose, OGTTAUC, fasting insulin, HOMA-IR, weight of eWAT, and fatty liver 395
score. 396
Fecal short-chain fatty acid analysis 397
The SCFAs we analyzed included acetic acid (C2), propionic acid (C3), butyric acid 398
(C4), isobutyric acid (C4), valeric acid (C5), and isovaleric acid (C5). Briefly, 20 mg of 399
the raw feces was dissolved in 500 μL of a 0.5% H3PO4 aqueous solution and 400
homogenized with Geno/Grinder® at 1,000 rpm for 2 min. The sample solutions were 401
then centrifuged at 18,000 × g for 10 min at 4 °C to separate depositions. Afterwards, 402
285 µL of the supernatant was collected in a clean centrifuge tube, and 15 µL of 403
acetate-d3 was added. For the liquid–liquid extraction of SCFAs, 300 μL of butanol and 404
the final solution were mixed and centrifuged. Last, 20 μL of internal standard 405
propionate-d5 was added to 180 μL of the upper organic layer. GC–MS analysis was 406
18
performed using an Agilent 7890A gas chromatograph (Agilent Technologies) coupled 407
with a Pegasus 4D GC × GC–TOF–MS system (Leco Corporation, St. Joseph, MI, 408
USA) using an VF-WAXms capillary column47. 409
Fecal bile acid analysis 410
The bile acid quantification included four primary bile acids (𝛼-MCA, 𝛽-MCA, CA, 411
and UDCA), four secondary bile acids (𝜔-MCA, CDCA, DCA, and LCA), and seven 412
conjugated bile acids (T𝛽-MCA, TUDCA, TCA, GCA, TCDCA, TDCA, TLCA). 413
Briefly, 20 mg of raw feces was dissolved in 200 μL of 70% d3-cholic acid aqueous 414
solution containing 2-ppm d4-cholic acid as internal standard, and it was homogenized 415
using an ultrasonicator for 30 min. The sample solutions were then centrifuged at 416
18,000 × g for 5 min. LC–MS analysis was performed using UltiMate 3000 liquid 417
chromatography system (Thermo Fisher Scientific, Dreieich, Germany) coupled with a 418
high-resolution Q Exactive Plus instrument equipped with the ESI source (Thermo 419
Fisher Scientific), and an Acquity HSS T3 (2.1 × 100 mm, 1.7 µm) column (Waters, 420
Milford, MA, USA). 421
Biochemical analysis of the FGF15–FGFR4 pathway 422
For western blotting, frozen tissues were lysed in lysis buffer containing 7-M urea, 2-M 423
thiourea, 2% CHAPS, 0.002% bromophenol blue, 60-mM DTT, and a protease and 424
phosphatase inhibitor cocktail. The cell lysates were sonicated for 5 min and centrifuged 425
at 17,500 × g for 30 min at 4 °C. Total protein content was measured based on the 426
Bradford assay with a Bradford reagent (Bioshop Canada Inc., Burlington, Canada) and 427
an ELISA reader. For the loading buffer, 62.5-mM Tris-HCl, 10% glycerol, 2% SDS, 428
and 0.01% bromophenol blue were mixed with the protein samples then heated at 95 °C 429
for 10 min for protein denaturation. Proteins were separated by 10% or 12% 430
SDS−polyacrylamide gel electrophoresis, they were then transferred onto 431
19
polyvinylidene difluoride membranes (Millipore, Burlington, MA, USA). After 432
blocking with 5% bovine serum albumin in tris-buffered saline with Tween 20, the 433
membranes were incubated with rabbit monoclonal anti–GAPDH antibody (1:5000; 434
Catalog No. 5174; Cell Signaling, Danvers, MA, USA), or rabbit polyclonal anti–435
FGF15 antibody (1:500; Catalog No. ab229630; Abcam, Cambridge, UK), or rabbit 436
polyclonal anti–FGFR4 antibody (1:500; Catalog No. ab119378; Abcam) at 4 °C for 16 437
h and subsequently incubated with HRP-linked anti-rabbit IgG antibody (1:3000; 438
Catalog No. 7074; Cell Signaling). The protein expression signal was captured and 439
quantified using the BioSpectrum AC imaging system (UVP, Upland, CA, USA) and 440
ImageJ (version 1.53; National Institutes of Health, Bethesda, MD, USA), 441
respectively48. Portal FGF15 levels were quantified using an ELISA kit (Mercodia, 442
Uppsala, Sweden). 443
Statistical analysis 444
Data were represented as mean ± standard deviation (SD) or mean ± standard error of 445
the mean (SEM). One-way ANOVA and Tukey’s range test or Student’s t-test were 446
applied for intergroup comparisons. Statistical assessment of the gut microbiome, 447
SCFAs, and bile acids was performed using Kruskal–Wallis with/without false 448
discovery rate or one-way ANOVA with Tukey’s range test. All statistical data were 449
analyzed using GraphPad Prism software (version 9.2.0; GraphPad Software, San 450
Diego, CA, USA) or RStudio (version 1.2.5001, RStudio, Boston, MA, USA). 451
452
20
Acknowledgments: This study was funded by the Ministry of Science and Technology 453
(MOST), Taiwan (MOST 106-3114-B-002-003, MOST 107-2321-B-002-039, MOST 454
108-2321-B-002-051, MOST 109-2327-B-002-005, and MOST 109-2314-B-002-064-455
MY3). We would like to acknowledge the technical support provided by the following 456
institutes: Fecal microbiota transplantation, isolator service and metabolic measurement 457
were conducted by National Laboratory Animal Center, National Applied Research 458
Laboratories, Taiwan. Analysis of bile acids was supported by the National Taiwan 459
University Consortia of Key Technologies. Analysis of short-chain fatty acid was 460
supported by the ‘Center of Precision Medicine’ from The Featured Areas Research 461
Center Program by the MOST. Body composition analysis was performed by Taiwan 462
Mouse Clinic, Academia Sinica and Taiwan Animal Consortium. We thank 463
BIOTOOLS for bioinformatics technique support. 464
Author contributions: R.A.C. designed and performed the animal study, performed 465
metabolomic, bioinformatics, and statistical analysis, and drafted the manuscript; 466
W.K.W. proposed, designed, and instructed the study; S.P. assisted the experiments and 467
bioinformatics analysis; P.Y.L. instructed the bioinformatics analysis; H.L.C. performed 468
the germ-free mice study and assisted the metabolic measurement; Y.H.C. assisted the 469
germ-free animal study and performed the LPS analysis; Q.L. instructed the bile acid 470
analysis; H.C.H. and H.B.Z. performed SCFA analysis and assisted with mass 471
spectrometry analysis; T.L.L. instructed the LPS analysis; Y.T.Y. conducted the PCR 472
and library preparation for 16S rRNA sequencing; H.S.H. and Y.E.L. instructed the 473
western blotting; S.P., T.C.D.S., W.K.W., P.Y.L. and L.Y.S. critically revised the 474
manuscript; W.K.W., L.Y.S., C.C.H., M.S.W., H.C.L., C.C.C. and C.T.H., provided 475
professional insights, techniques, and relevant resources for the study. 476
Data availability: The raw 16s rRNA sequencing data are accessible at the National 477
Center for Biotechnology Information Short Read Archive (BioProject: 478
PRJNA715326). 479
Competing interests: The authors declare that the research was conducted in the 480
absence of any commercial or financial relationships that could be construed as a 481
potential conflict of interest. 482
483
21
References 484
1 Wang, H. et al. Antibiotics detected in urines and adipogenesis in school 485
children. Environ Int 89-90, 204–211 (2016). 486
2 Brussow, H. Growth promotion and gut microbiota: insights from antibiotic use. 487
Environ Microbiol 17, 2216–2227 (2015). 488
3 Ji, K. et al. Influence of water and food consumption on inadvertent antibiotics 489
intake among general population. Environ Res 110, 641–649 (2010). 490
4 Li, N., Ho, K. W. K., Ying, G. G. & Deng, W. J. Veterinary antibiotics in food, 491
drinking water, and the urine of preschool children in Hong Kong. Environ Int 492
108, 246–252 (2017). 493
5 Wang, H. et al. Antibiotic residues in meat, milk and aquatic products in 494
Shanghai and human exposure assessment. Food Control 80, 217–225 (2017). 495
6 WHO. Principles and methods for the risk assessment of chemicals in food. 496
Environmental Health Criteria 70. 240 (2009). 497
7 Riley, L. W., Raphael, E. & Faerstein, E. Obesity in the United States - 498
dysbiosis from exposure to low-dose antibiotics? Front Public Health 1, 69 499
(2013). 500
8 Cox, L. M. et al. Altering the intestinal microbiota during a critical 501
developmental window has lasting metabolic consequences. Cell 158, 705–721 502
(2014). 503
9 Cho, I. et al. Antibiotics in early life alter the murine colonic microbiome and 504
adiposity. Nature 488, 621–626 (2012). 505
10 Mahana, D. et al. Antibiotic perturbation of the murine gut microbiome 506
enhances the adiposity, insulin resistance, and liver disease associated with high-507
fat diet. Genome Med 8, 48 (2016). 508
11 Livanos, A. E. et al. Antibiotic-mediated gut microbiome perturbation 509
accelerates development of type 1 diabetes in mice. Nat Microbiol 1, 16140 510
(2016). 511
12 Egshatyan, L. et al. Gut microbiota and diet in patients with different glucose 512
tolerance. Endocr Connect 5, 1–9 (2016). 513
13 Ju, T., Kong, J. Y., Stothard, P. & Willing, B. P. Defining the role of 514
Parasutterella, a previously uncharacterized member of the core gut microbiota. 515
ISME J 13, 1520–1534 (2019). 516
14 Lam, Y. Y. et al. Increased gut permeability and microbiota change associate 517
with mesenteric fat inflammation and metabolic dysfunction in diet-induced 518
obese mice. PLoS One 7, e34233 (2012). 519
15 Chen, L. W. et al. Implication of gut microbiota in the association between 520
infant antibiotic exposure and childhood obesity and adiposity accumulation. Int 521
J Obes (Lond) 44, 1508–1520 (2020). 522
16 Mbakwa, C. A. et al. Early Life Antibiotic Exposure and Weight Development 523
in Children. J Pediatr 176, 105–113 e102 (2016). 524
22
17 Schulfer, A. F. et al. The impact of early-life sub-therapeutic antibiotic treatment 525
(STAT) on excessive weight is robust despite transfer of intestinal microbes. 526
ISME J 13, 1280–1292 (2019). 527
18 Cox, L. M. & Blaser, M. J. Antibiotics in early life and obesity. Nat Rev 528
Endocrinol 11, 182–190 (2015). 529
19 Nobel, Y. R. et al. Metabolic and metagenomic outcomes from early-life pulsed 530
antibiotic treatment. Nat Commun 6, 7486 (2015). 531
20 Matamoros, S., Gras-Leguen, C., Le Vacon, F., Potel, G. & de La Cochetiere, 532
M. F. Development of intestinal microbiota in infants and its impact on health. 533
Trends Microbiol 21, 167–173 (2013). 534
21 Zeissig, S. & Blumberg, R. S. Life at the beginning: perturbation of the 535
microbiota by antibiotics in early life and its role in health and disease. Nat 536
Immunol 15, 307–310 (2014). 537
22 Insenser, M. et al. Gut Microbiota and the Polycystic Ovary Syndrome: 538
Influence of Sex, Sex Hormones, and Obesity. J Clin Endocrinol Metab 103, 539
2552–2562 (2018). 540
23 Menni, C. et al. Gut microbiome diversity and high-fibre intake are related to 541
lower long-term weight gain. Int J Obes (Lond) 41, 1099–1105 (2017). 542
24 Vrieze, A. et al. Impact of oral vancomycin on gut microbiota, bile acid 543
metabolism, and insulin sensitivity. J Hepatol 60, 824–831 (2014). 544
25 Reijnders, D. et al. Effects of Gut Microbiota Manipulation by Antibiotics on 545
Host Metabolism in Obese Humans: A Randomized Double-Blind Placebo-546
Controlled Trial. Cell Metab 24, 341 (2016). 547
26 Canfora, E. E., Jocken, J. W. & Blaak, E. E. Short-chain fatty acids in control of 548
body weight and insulin sensitivity. Nat Rev Endocrinol 11, 577–591 (2015). 549
27 Cani, P. D. et al. Microbial regulation of organismal energy homeostasis. Nat 550
Metab 1, 34–46 (2019). 551
28 Heimann, E., Nyman, M., Palbrink, A. K., Lindkvist-Petersson, K. & Degerman, 552
E. Branched short-chain fatty acids modulate glucose and lipid metabolism in 553
primary adipocytes. Adipocyte 5, 359–368 (2016). 554
29 Louis, P. & Flint, H. J. Formation of propionate and butyrate by the human 555
colonic microbiota. Environ Microbiol 19, 29–41 (2017). 556
30 Vacca, M. et al. The Controversial Role of Human Gut Lachnospiraceae. 557
Microorganisms 8 (2020). 558
31 Belgaumkar, A. P. et al. Changes in Bile Acid Profile After Laparoscopic Sleeve 559
Gastrectomy are Associated with Improvements in Metabolic Profile and Fatty 560
Liver Disease. Obes Surg 26, 1195–1202 (2016). 561
32 Talavera-Urquijo, E., Beisani, M., Balibrea, J. M. & Alverdy, J. C. Is bariatric 562
surgery resolving NAFLD via microbiota-mediated bile acid ratio reversal? A 563
comprehensive review. Surg Obes Relat Dis 16, 1361–1369 (2020).. 564
33 Jia, W., Wei, M., Rajani, C. & Zheng, X. Targeting the alternative bile acid 565
synthetic pathway for metabolic diseases. Protein Cell (2020). 566
23
34 Jia, W., Xie, G. & Jia, W. Bile acid-microbiota crosstalk in gastrointestinal 567
inflammation and carcinogenesis. Nat Rev Gastroenterol Hepatol 15, 111–128 568
(2018). 569
35 Ge, H. et al. Fibroblast growth factor receptor 4 (FGFR4) deficiency improves 570
insulin resistance and glucose metabolism under diet-induced obesity conditions. 571
J Biol Chem 289, 30470–30480 (2014). 572
36 Potthoff, M. J. et al. FGF15/19 regulates hepatic glucose metabolism by 573
inhibiting the CREB-PGC-1alpha pathway. Cell Metab 13, 729–738 (2011). 574
37 Shin, D. J. & Osborne, T. F. FGF15/FGFR4 integrates growth factor signaling 575
with hepatic bile acid metabolism and insulin action. J Biol Chem 284, 11110–576
11120 (2009). 577
38 Onyinye Stella, O. I., Vivienne Ezenduka, E. & John Anaelom, N. Screening for 578
tylosin and other antimicrobial residues in fresh and fermented (nono) cow milk 579
in Delta state, South-South, Nigeria. Vet World 13, 458–464 (2020). 580
39 USDA. FY2015 Food Safety and Inspection Service (FSIS) National Residue 581
Program Data. (2015). 582
40 USDA. 2016 Food Safety and Inspection Service (FSIS) National Residue 583
Program Data. (2016). 584
41 WHO. Evaluation of certain veterinary drug residues in food. WHO Technical 585
Report Series 954 (2008). 586
42 Matthews, D. R. et al. Homeostasis model assessment: insulin resistance and 587
beta-cell function from fasting plasma glucose and insulin concentrations in 588
man. Diabetologia 28, 412–419 (1985). 589
43 Liang, W. et al. Establishment of a general NAFLD scoring system for rodent 590
models and comparison to human liver pathology. PLoS One 9, e115922 (2014). 591
44 Lai, H. C. et al. Gut microbiota modulates COPD pathogenesis: role of anti-592
inflammatory Parabacteroides goldsteinii lipopolysaccharide. Gut, gutjnl-2020-593
322599 (2021). 594
45 Suzuki, T. A., Martins, F. M. & Nachman, M. W. Altitudinal variation of the gut 595
microbiota in wild house mice. Mol Ecol 28, 2378–2390, 596
doi:10.1111/mec.14905 (2019). 597
46 Oksanen, J. et al. vegan: Community Ecology Package. (2020). 598
47 Hsu, Y. L. et al. Evaluation and Optimization of Sample Handling Methods for 599
Quantification of Short-Chain Fatty Acids in Human Fecal Samples by GC-MS. 600
J Proteome Res 18, 1948–1957 (2019). 601
48 Panyod, S. et al. Diet Supplementation with Allicin Protects against Alcoholic 602
Fatty Liver Disease in Mice by Improving Anti-inflammation and Antioxidative 603
Functions. J Agric Food Chem 64, 7104–7113 (2016). 604
605
24
Figures 606
607
Fig. 1 | Residual dose of tylosin facilitates HFD-induced obesity and metabolic 608
disorders. 609
a, Experimental design of antibiotic residue exposure model. b, Body weight changes in 610
NCD and c, HFD mice (n = 8–12/group). d, Relative fat mass and e, relative lean mass 611
changes in HFD mice (n = 8–12/group). f, Weight of epididymal adipose tissue and g, 612
perinephric adipose tissue (n = 8–12/group). h-i, histological features of H&E stained in 613
epididymal adipose tissue (h) and liver (i) (n = 8–12/group). j, Adipocyte diameter of 614
epididymal adipose tissue (n = 6/group). k, Fatty liver score including steatosis 615
(macrovesicular, microvesicular, and hypertrophy) and inflammation (number of 616
inflammatory foci) (n = 8–12/group). l, Area under the curve (AUC) derived from the 617
OGTT (n = 7–8/group). m, Plasma insulin level after overnight fasting (n = 7–8/group). 618
n, HOMA-IR index represented as an indicator of insulin resistance (n = 7–8/group). 619
Data are expressed as mean ± SD. For b-e, and j, Statistical analyses were performed 620
by one-way ANOVA with Tukey’s range test within diet groups (NCD or HFD) as 621
40
50 HFD-CON
HFD-ADI
HFD-TMDI*
#
20
-30
30
-40
40
-50
50
-60
60
-70
70
-80
80
-90
90
-10
0
10
0-1
10
11
0-1
20
12
0-1
30
13
0-1
40
14
0-1
50
15
0-1
60
16
0-1
70
17
0-1
80
18
0-1
90
19
0-2
00
-10
0
10
20
30
40
Adipocyte size (µm)
Perc
cent of adip
ocyte
s (
%)
##
**
#
#
***
* *
##
**
##
**
#
**##
*
**
***
*
40
50 HFD-CON
HFD-ADI
HFD-TMDI*
#HFD-ConHFD-ADIHFD-TMDI
HFD-CON
HFD-ADI
HFD-TMDI
b c
d e f g h
j k l m n
HFD-CON
HFD-CON
HFD-ADI
HFD-ADI
HFD-TMDI
HFD-TMDI
100 μm
100 μm
100 μm
100 μm
100 μm
100 μm
i
8 12 17
0
10
20
30
40
Age (week)
Rela
tive fat m
ass (
%)
*
#
*
#
*
8 12 17
50
55
60
65
70
75
80
Age (week)
Rela
tive le
an m
ass (
%)
**
#
**
#
*
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
35
40
45
50
Age (week)
Body w
eig
ht (g
)
HFD-CON
HFD-ADI
HFD-TMDI
**
##
***
*****
*****
******* ***
**
*** **
#
####
## ####
## ### #####
# # # #
*#
3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
0
5
10
15
20
25
30
35
40
45
50
Age (week)
Body w
eig
ht (g
)
NCD-CON
NCD-ADI
NCD-TMDI
**
***
#
#####
#*
*#
Normal chow diet (NCD)
or high-fat diet (HFD)NursingPregnancy
NCD-CON
NCD-ADI
NCD-TMDI
HFD-CON
HFD-ADI
HFD-TMDI
-10 d 0 d 3 wks 20 wks8 wks 12 wks 17 wks
NCD group
HFD group
Microbiome analysis
Body composition analysis
OGTT and metabolic measurements
a
CON ADI TMDI
0.0
0.5
1.0
1.5
Perinephric a
dip
ose tis
sue (
g)
HFD
*
CON ADI TMDI
0
1
2
3
Epid
idym
al a
dip
ose tis
sue (
g)
HFD
*
CON ADI TMDI
30000
40000
50000
60000
70000O
GT
TA
UC (
mg/d
L x
min
)
HFD
p=0.05
CON ADI TMDI
0
5000
10000
15000
Insulin
(pg/m
l)HFD
**
CON ADI TMDI
0
50
100
150
200
250
HO
MA
-IR
(m
mol/L
x µ
U/L
)
HFD
**
p=0.051
40
50 HFD-CON
HFD-ADI
HFD-TMDI*
#
Mice
12
12
8
12
12
12
25
follows: CON vs ADI (#p < 0.05; ##p < 0.01); CON vs TMDI (*p < 0.05; **p < 0.01; 622
***p < 0.001). For f-g, and k-n, one-way ANOVA with Tukey’s range test (*p < 0.05; 623
**p < 0.01; ***p < 0.001). Abbreviations: ADI, acceptable daily intake; AUC, area 624
under curve; CON, control; HFD, high-fat diet; HOMA-IR, homeostatic model 625
assessment of insulin resistance; NCD, normal chow diet; OGTT, oral glucose tolerance 626
test; TMDI, theoretical maximum daily intake. 627
628
26
629
Fig. 2 | Residual dose of tylosin remodels the gut microbiota composition, and their 630
microbiota shift verifies its pathogenesis on generating the obesogenic and 631
metabolic phenotype in germ-free mice. 632
a, Change in Shannon diversity index in 3-, 8-, 17-week-old HFD-fed mice (n = 633
6/group). b, Gut microbiota composition as represented by principal coordinate analysis 634
(PCoA) of Bray-Curtis distances for 3-, 8-, and 17-week-old HFD-fed mice (n = 635
6/group). c, Bray-Curtis dissimilarity index comparing distances of tylosin-treated 636
groups to CON at different time points (n = 6/group). d, Volcano plot showing the 637
enriched or decreased bacteria based on log2 fold change >1 and significant difference 638
(p < 0.05) using the Wilcoxon signed-rank test (n = 6/group). e, Bray-Curtis distances-639
based PCoA of 17-week-old HFD-fed mice's gut microbiota and the fitted antibiotic 640
exposure/obesity-related variables which significantly correlated to the shifted 641
microbiome (p < 0.05) by using the envfit package in R (n = 5-6/group). f, Spearman's 642
a c
fe
g h i j k
b
40
50 HFD-CON
HFD-ADI
HFD-TMDI*
#
3 8 17
2.5
3.0
3.5
4.0
Age (week)
Shannon in
dex **
##
0.2
0.4
0.6
0.8****
**
3 8 17 3 8 17
0.0
0.2
0.4
0.6
0.8
1.0
Bra
y-C
urt
is d
ista
nce
CON to ADI
HFD
CON to TMDI
***
****
****
****
**
8 9 10 11 12 13 14 15 16 17 18 19 20
20
30
40
50
60
Age (week)
Body w
eig
ht (g
)
FMT-CON
FMT-TMDI
**
*
*
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●●
● ●
●
●
●
●
●
●
●
●
●●
●
●
●●●
●
●
● ●
●
●
●
0.4
0.2
0.0
0.2
●
●
●
●
●
●●
●
●
3
3
3
8
8
8
17
17
17
HFD-CON-3wk
HFD-CON-8wk
HFD-CON-17wk
HFD-ADI-3wk
HFD-ADI-8wk
HFD-ADI-17wk
HFD-TMDI-3wk
HFD-TMDI-8wk
HFD-TMDI-17wk
3
8
17
3
8
17
3
8
17
PCoA1 (23.89%)
-0.2 0.0 0.2 0.4
PC
oA
2 (
15
.91
%)
-0.4
-0.2
0.2
0.0
d
CON TMDI
-2
0
2
4
6
8
Body w
eig
ht change
in 2
weeks a
fter
FM
T (
g)
FMT
*
CON TMDI
20
30
40
50
Rela
tive fat m
ass (
%)
FMT
p=0.086
CON TMDI
1500
2000
2500
3000
OG
TT
AU
C (
mg/d
L x
min
)
FMT
****
CON TMDI
0
10
20
30
HO
MA
-IR
(m
mol/L
x µ
U/L
)
FMT
*
FMT
Adonis: p < 0.0001
1.0
1.5
2.0
0.5
0.0
-lo
g1
0 (p
.valu
e)
log2 fold change
Bifidobacterium
Oscillibacter
Blautia
Parasutterella
0 5 10-5-10
Decreased genus
Enriched genus
-0.6 -0.4 -0.2 0.0 0.2 0.4
-0.4
-0.2
0.0
0.2
0.4
0.6
PCoA1 (32.4%)
PC
oA
2 (
25.2
%)
Body weight
Relative fat mass
Fatty liver score
Fasting glucose
OGTTAUC
Antibiotic exposure
Adonis: p < 0.0001
r=−0.71
p=0.003HFD-ADI
HFD-CON
HFD-TMDI
●
●●
●●
●
●●
●
●●
● ● ●
●
●
r=−0.71
p=0.003
-4 -2 0 2
-0.4
0.4
-0.2
0.0
0.2
r = 0.71 p = 0.003
HFD-ADI
HFD-CON
HFD-TMDI
r=−0.71
p=0.003HFD-ADI
HFD-CON
HFD-TMDI
PC1obesity biomarkers (84.4%)
PC
oA
2m
icro
bio
ta (
25
.2%
)
r = 0.71 p = 0.003
27
correlation of PCoA2microbiota and PC1obesity biomarkers (n = 5-6/group). g, Body weight 643
change in mice transplanted with feces of HFD-CON and HFD-TMDI mice (n = 644
10/group). h, Body weight change at 2 weeks after transplantation (ΔBW(W10-W8)) (n = 645
10/group). i, Relative fat mass at 20 weeks of age (n = 10/group). j, AUC derived from 646
the OGTT (n = 9/group). k, HOMA-IR index (n = 9/group). Data are expressed as mean 647
± SD. For a, Statistical analyses were performed by Wilcoxon signed-rank test with 648
FDR-adjust as follow: HFD-CON vs HFD-ADI (##p < 0.01); HFD-CON vs HFD-TMDI 649
(**p < 0.01). For c, One-way ANOVA with Tukey’s range test (**p < 0.01; ***p < 650
0.001; ****p < 0.0001). For e, Wilcoxon signed-rank test with FDR-adjust. For g-k, un-651
paired t-test (*p < 0.05; ****p < 0.0001). Abbreviations: AUC, area under curve; CON, 652
control; FMT, fecal microbiota transplantation; HOMA-IR, homeostatic model 653
assessment of insulin resistance; OGTT, oral glucose tolerance test; TMDI, theoretical 654
maximum daily intake. 655
656
28
657
Fig. 3 | Early-life exposure of tylosin residue sufficiently induces obesity and 658
metabolic disorder disease and modifies gut microbiota composition by enhancing 659
the re-structuring of obesity-related genera with deep-rooted change. 660
a, Experimental design of early-life exposure model. b, Body weight change (n = 8-661
11/group). c, Relative fat mass change (n = 8-11/group). d, Weight of visceral adipose 662
tissue (n = 8-11/group). e, AUC derived from the OGTT (n = 7-8/group). f, Insulin level 663
after overnight fasting (n = 7-8/group). g, HOMA-IR index (n = 7-8/group). h, PCoA 664
based on Bray-Curtis distances of gut microbiota at 5 and 20 weeks of age (n = 665
8/group). i, Heatmap showing 32 bacteria with significant differences (q < 0.05) among 666
groups and j, their Spearman's correlation with obesity-related variables (n = 8/group). 667
Data are expressed as mean ± SD. For b-c, Statistical analyses were performed by one-668
way ANOVA with Tukey’s range test as follow: CON vs Early-TMDI (#p < 0.05 and 669
##p < 0.01), CON vs Cont-TMDI (*p < 0.05; **p < 0.01; ***p < 0.001), or Cont-TMDI 670
CON
Cont-TMDI
Early-TMDI
-10 d 0 d 3 wks 20 wks5 wks 10 wks 15 wks
High-fat diet (HFD)NursingPregnancy
Microbiome analysis
Body composition analysis
SCFAs and BAs analysis
OGTT and metabolic measurements
a b c
d
g
fe i
5 10 15 20
0
10
20
30
40
50
Age (week)
Rela
tive fat m
ass (
%)
CONCont-TMDI
Early-TMDI
*
*
*
##
* #
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
0
10
20
30
40
50
Age (week)
Body w
eig
ht (g
)
CON
Cont-TMDI
Early-TMDI
* #
***
****
*****
** **** *
* * **
** **
#
###
## # # #
##
†
†
CON ContTMDI
EarlyTMDI
0
2
4
6
8
Vecera
l adip
ose tis
sue (
g)
*
CON ContTMDI
EarlyTMDI
1600
2000
2400
2800
OG
TT
AU
C (
mg/d
L x
min
)
*
CON ContTMDI
EarlyTMDI
0
1000
2000
3000
4000
Insulin
(µ
g/L
) *
CON ContTMDI
EarlyTMDI
0
10
20
30
40
HO
MA
-IR
(m
mol/L
x µ
U/L
)
**
p=0.09
●
●
●
●
●
●
●
●
●● ●
●
●●
●
●
● ●
●
●
●
●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
PCoA1 ( 42.85 %)
PC
oA
2 (
19.0
4 %
)
−0.4 −0.2 0.0 0.2 0.4
−0.3
−0.2
−0.1
0.0
0.1
0.2
0.3
●
●
●
●
●
●
5
20
5
20
5
20
CON-5wk
CON-20wk
5
20
5
20
5
20
Cont-TMDI-5wk
Cont-TMDI-20wk
Early-TMDI-5wk
Early-TMDI-20wk
h
Mice
8
7
11
j
Adonis: p < 0.0001
29
vs Early-TMDI (†p < 0.05). For d-g, One-way ANOVA with Tukey’s range test (*p < 671
0.05; **p < 0.01). For h, Adonis was performed to test the difference among groups. 672
For i, Kruskal-Wallis test with FDR-adjusted p-value (q < 0.05). For j, Spearman’s 673
correlation with FDR-adjusted p-value (* q < 0.1; **q < 0.05). Abbreviations: AUC, 674
area under curve; CON, control; HFD, high-fat diet; HOMA-IR, homeostatic model 675
assessment of insulin resistance; OGTT, oral glucose tolerance test; TMDI, theoretical 676
maximum daily intake. 677
30
678
Fig. 4 | The modification of fecal primary-secondary bile acids ratio by gut 679
microbiota and downregulates FGF15 signaling pathway is involved in the 680
consequences of obesogenic and metabolic dysfunctions caused by tylosin residue 681
exposure at TMDI dose. 682
a, Fecal short-chain fatty acids levels (n = 7/group). b, Ratio of fecal primary bile acids 683
to secondary bile acids (n = 7-11/group). c, Levels of non-12-OH bile acids, d, 684
muricholic acids and e, 12-OH bile acids (n = 7-11/group). f, Portal FGF15 levels (n = 685
6-11/group). g, Western blotting of ileal FGF15 expression normalized to GAPDH (n = 686
4/group). h, Western blotting of hepatic FGFR4 level normalized to GAPDH. Data are 687
presented by mean ± SD (n = 4/group). For a-h, Statistical analyses were performed by 688
one-way ANOVA with Tukey’s range test (*p < 0.05; **p < 0.01; ***p < 0.001). 689
Abbreviations: AGP, antibiotic growth promoter; α-MCA, α-muricholic acid; β-MCA, 690
β-muricholic acid; ω-MCA, ω- muricholic acid; CON, control; CDCA, 691
a b c
d
g
fe
h
CON ContTMDI
EarlyTMDI
0.0
0.5
1.0
1.5
2.0
PB
A / S
BA
*
*
T!MCA "MCA !MCA #MCA
0.0
0.2
0.4
0.6
0.8
5
10
15
20
Bile
acid
s (
%)
**
**
******
*
Primary Secondary
TCDCA CDCA TUDCA UDCA LCA TLCA
0
2
4
20
40
60
80
100
Bile
acid
s (
%)
CON
Cont-TMDI
Early-TMDI
Primary Secondary
**
**
* *
Acetic acid
Propionicacid
Butyricacid
Valericacid
Isobutyricacid
Isovalericacid
0
100
200
300
4001000
2000
3000
4000
Short
chain
fatty a
cid
s (
µg/g
)
CON
Cont-TMDI
Early-TMDI
p=0.066
p=0.068
p=0.059** *
CON ContTMDI
EarlyTMDI
0
10
20
30
Port
al F
GF
15 (
pg/m
L) *
TCA CA TDCA DCA
0.0
0.2
0.4
0.6
0.810
20
30
40
Bile
acid
s (
%)
Primary Secondary
*
CON
Cont-TMDI
Early-TMDI
CON
Cont-TMDI
Early-TMDI
iFGF15
GAPDH
FGFR4
GAPDH
CON ContTMDI
EarlyTMDI
0.0
0.4
0.8
1.2
Rela
tive e
xpre
ssio
n le
vel
(FG
FR
4 / G
AP
DH
) ***
*
CON ContTMDI
EarlyTMDI
0.0
0.4
0.8
1.2
Rela
tive e
xpre
ssio
n le
vel
(FG
F15 / G
AP
DH
)
**
**
31
chenodeoxycholic acid; FGF15, fibroblast growth factor 15; FGFR4, fibroblast growth 692
factor receptor 4; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; LCA, 693
lithocholic acid; PBA, primary bile acid; SBA, secondary bile acid; T-β-MCA, tauro-694
beta-muricholic acid; TCDCA, taurochenodeoxycholic acid; TLCA, taurolithocholic 695
acid; TMDI, theoretical maximum daily intake; TUDCA, tauroursodeoxycholic acid; 696
UDCA, ursodeoxycholic acid. 697
Supplementary Files
This is a list of supplementary �les associated with this preprint. Click to download.
ExtendedDataFigures20211109NatMetab.pdf
Supplementarytable20211109NatMetabver1.pdf