bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
1 Enrofloxacin shifts intestinal microbiota and metabolic profiling and hinders recovery
2 from Salmonella enterica subsp. enterica serovar Typhimurium infection in neonatal
3 chickens
4
5 Boheng Ma,a,b,c,& Xueran Mei,a,b,c,& Changwei Lei,a,b,c Cui Li,a,b,c Yufeng Gao,a,b,c Linghan
6 Kong,a,b,c Xiwen Zhai,a,b,c,d Hongning Wanga,b,c,#
7
8 a College of Life Sciences, Sichuan University, Chengdu, People’s Republic of China
9 b Key Laboratory of Bio-Resource and Eco-Environment of Ministry of Education, Chengdu,
10 People’s Republic of China
11 c Animal Disease Prevention and Food Safety Key Laboratory of Sichuan Province, Chengdu,
12 People’s Republic of China
13 d Sichuan Water Conservancy Vocational Collage, Chengdu, People’s Republic of China
14
15 Running head: Effects of enrofloxacin on chickens
16
17 # Address correspondence to Hongning Wang, [email protected]
18 & Both authors are identified as Co-First Author.
19
20 Word count: 211 words (Abstract); 4895 words (main text)
1 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
21 ABSTRACT
22 Enrofloxacin is an important antibiotic used for prevention and treatment of Salmonella infection
23 in poultry in many countries. However, oral administration of enrofloxacin may lead to the
24 alterations in the microbiota and metabolome in the chicks intestine, thereby reducing
25 colonization resistance to the Salmonella infection. To study the effects of enrofloxacin on
26 chicken cecal Salmonella, we used different concentrations of enrofloxacin to feed 1-day-old
27 chickens, followed by oral challenge with Salmonella enterica subsp. enterica serovar
28 Typhimurium (S. Typhimurium). We then explored the distribution patterns of S. Typhimurium in
29 vivo in intestinal contents using quantitative polymerase chain reaction and microbial 16S
30 amplicon sequencing on days 7, 14, and 21. Metabolome sequencing was used to explore the gut
31 metabolome on day 14. Faecalibacterium and Anaerostipes, which are closely related to the
32 chicken intestinal metabolome, were screened using a multi-omics technique. The abundance of
33 S. Typhimurium was significantly higher in the enrofloxacin-treated group than in the untreated
34 group, and S. Typhimurium persisted longer. Moreover, the cecal colony structures of the three
35 groups exhibited different characteristics, with Lactobacillus reaching its highest abundance on
36 day 21. Notably, S. Typhimurium infection is known to affect the fecal metabolome of chickens
37 differently. Thus, our results suggested that enrofloxacin and Salmonella infections completely
38 altered the intestinal metabolism of chickens.
39
40 KEYWORDS gut microbiota, enrofloxacin, metabolome, chicken, Salmonella Typhimurium
41
2 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
42 IMPORTANCE
43 In this study, we examined the effects of S. Typhimurium infection and enrofloxacin treatment on
44 the microbial flora and metabolite synthesis in chicken guts in order to identify target metabolites
45 that may cause S. Typhimurium colonization and severe inflammation and to evaluate the
46 important flora that may be associated with these metabolites. Our findings may facilitate the use
47 of antibiotics to prevent S. Typhimurium infection.
3 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
48 INTRODUCTION
49 Foodborne diseases are associated with high morbidity and mortality rates worldwide and pose
50 major challenges to food safety and economics (1). Salmonella is an important foodborne
51 pathogen of the family Enterobacteriaceae. This pathogen causes millions of infections
52 worldwide each year, both in humans and livestock. Salmonella has been shown to cause
53 intestinal inflammation and barrier dysfunction in chickens, thereby affecting chicken
54 performance (2). Currently, more than 2,600 serotypes of Salmonella are known, with varying
55 pathogenicity and host specificity. Salmonella enterica subsp. enterica serovar Typhimurium (S.
56 Typhimurium) is a typical representative of non-host-specific Salmonella found in poultry.
57 The main route of infection with Salmonella in poultry is the fecal-oral route. For
58 multiserotype infections, bacteria mainly colonize in the ceca of poultry. After colonizing the
59 intestines, these bacteria invade the intestinal epithelial cells and dendritic cells and reach the
60 submucosa to be phagocytosed by macrophages. The bacteria then replicate in and colonize the
61 spleen and liver, causing systemic infections mediated by bacterium/macrophage interactions (3,
62 4). In mice infected with Salmonella, macrophages produce IL-12 and IL-18 and promote IFN-γ
63 secretion by natural killer (NK) cells (5).
64 Owing to the widespread use of antibiotics in agriculture and animal husbandry,
65 Salmonella resistance has rapidly increased. In 2013, China alone consumed approximately
66 927,000 tons of antibiotics (6). Moreover, the consumption of antibiotics as veterinary drugs
67 increased from 46% to 52% between 2007 and 2013. Enrofloxacin, florfenicol, lincomycin, and
68 amoxicillin are among the most commonly consumed veterinary antibiotics in China (7).
69 Enrofloxacin is the main antibiotic used for treating Salmonella infection in poultry in many
4 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
70 countries, and the concentration of enrofloxacin in chicken manure in China was found to be
71 1,421 mg/kg (8).
72 Quinolones, such as enrofloxacin, affect the composition and number of bacterial species
73 in the chicken gut (9), thereby altering resistance to pathogen invasion. Indeed, bacteria residing
74 in the gut can mediate the colonization of S. Typhimurium in the body by producing short-chain
75 fatty acids (SCFAs), such as propionic acid, which inhibits S. Typhimurium growth by
76 influencing the intracellular pH (10). By producing SCFAs, lactic acid bacteria inhibit
77 Salmonella colonization in the colon and jejunum, accelerate Salmonella clearance from stools,
78 and reduce Salmonella transfer from the small intestine to the spleen (11). The concentrations of
79 SCFAs and butyric acid are inversely proportional to the degree of inflammation (12).
80 Some Enterobacteriaceae in the normal gut flora can mediate Salmonella multiplication
81 through competition for oxygen (13). Additionally, antibiotics reduce the number of butyrate-
82 producing Clostridium perfringens in the intestinal flora, shifting the host cells from oxidative
83 metabolism to lactic acid fermentation and increasing the level of lactic acid in the intestine;
84 Salmonella then gain a colonization advantage by using lactic acid (14). Intestinal metabolism is
85 also affected by the gut flora composition. Using antibiotics increases fucose and sialic acid
86 contents in the intestine, thereby affecting Salmonella and C. difficile colonization (15, 16).
87 Similarly, antibiotics also decrease the levels of secondary bile acids in the intestine and inhibit
88 C. difficile colonization (16).
89 Accordingly, in this study, we investigated the effects of enrofloxacin on Salmonella
90 colonization in chickens and the relationship between intestinal flora and Salmonella
91 colonization. We also evaluated changes in microbial metabolites in the ceca of enrofloxacin-
5 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
92 treated chickens infected with Salmonella and used multiomics association analyses to determine
93 the correlations between microbial community structure and microbial metabolites.
94
6 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
95 RESULTS
96 Response of Salmonella to enrofloxacin
97 First, S. Typhimurium abundance in the samples was evaluated. S. Typhimurium was not isolated
98 from C1, C2, or C3 groups. On day 7, the highest abundance of S. Typhimurium was found in the
99 cecum contents of group E1 (6.128 log10CFU/g), followed by that of groups E3 and (5.090 and
100 3.945 log10CFU/g, respectively; P < 0.001). The abundances of S. Typhimurium in the heart,
101 liver, and spleen of group E1 were significantly higher than those in the other two groups (P <
102 0.001). On day 14, the abundance of S. Typhimurium in group E1 (3.371 log10CFU/g) was
103 significantly lower than those in groups E2 (4.829 log10CFU/g) and E3 (4.366 log10CFU/g; P <
104 0.01). Moreover, the abundances of S. Typhimurium were significantly lower in heart, liver, and
105 spleen samples from group E1 than those from groups E2 and E3, except for heart samples from
106 group E3. On day 21, the relative abundances of S. Typhimurium were 3.170, 5.146, and 5.636
107 log10CFU/g in groups E1, E2, and E3, respectively; that in group E1 was significantly lower than
108 those in the other two groups. However, no S. Typhimurium was detected in the heart, liver, or
109 spleen samples of group E1 (Figure 1).
110
111 Expression levels of immune-related genes following enrofloxacin treatment
112 On day 14, MUC2 expression was significantly lower in group E3 than in group E1 (P < 0.05),
113 TLR21 expression was significantly higher in group E3 than in groups E1 and E2 (P < 0.05), IL-
114 1β and LITAF levels were significantly higher in group E3 than in group E1 (P < 0.05), IL-18
115 and OCLM levels were significantly higher in group E1 than in groups E2 and E3 (P < 0.05), and
116 TLR4 expression did not differ significantly among the three groups (Figure 2).
117
7 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
118 Effects of enrofloxacin on the microbial composition and structure
119 By 16S rRNA sequencing of cecum colonies, as an increased number of samples was sequenced,
120 the dilution curve flattened, indicating that the sequencing data were saturated and that the
121 sequencing results could reflect the majority of the samples (Figure S1). On day 7, significant
122 differences were observed in the microbial diversity of the ceca among groups E1, E2, and E3 (P
123 < 0.05). Group E1 had the lowest diversity, whereas group E3 had the highest diversity. On day
124 14, the microbial diversity of all experimental groups was higher than that on day 7 (P < 0.05).
125 However, on day 14, no differences were observed in the Good’s coverage index, ACE index,
126 observed species index, or Chao index between the groups (P > 0.05). However, Shannon’s and
127 Simpson’s diversity indices were significantly different (P < 0.05). On day 21, microbial
128 diversity was higher in group E2 than in groups E1 and E3 (P < 0.05; Figure S2).
129 On day 7, the microbial community structures of cecum samples from groups E1, E2, and
130 E3 differed significantly in the hierarchical clustering tree. The differences in the microbial
131 community structure of the samples collected on days 14 and 21 gradually decreased, and the
132 samples were mixed together in the clustering tree (Figure 3). PCA showed that on day 7, the
133 structure of the cecum flora was isolated in groups E1, E2, and E3 owing to the use of different
134 concentrations of antibiotics. However, on days 14 and 21, the antibiotic concentrations had no
135 significant effect on the structure of the cecum flora (Figure S3).
136 An UpSetR chart shows the common and unique OTUs of different categories of chicken
137 cecum microbes. On day 7, 193 OTUs from the three groups were found to be common, whereas
138 22, 24, and 55 OTUs from groups E1, E2, and E3, respectively, were found to be unique. On day
139 14, 391 OTUs from the three groups were found to be shared, whereas 10, 34, and 22 OTUs
140 from groups E1, E2, and E3, respectively, were found to be unique. On day 21, 396 OTUs from
8 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
141 the three groups were found to be shared, whereas 15, 44, and 18 OTUs from groups E1, E2, and
142 E3, respectively, were found to be unique. This showed that as the chicken intestinal flora
143 matured, the OTUs shared among different groups increased, unique OTUs decreased, and the
144 microbial community structures tended to be similar. In total, 200 OTUs were found to be shared
145 by group E1 at the three sampling points. On days 7, 14, and 21, the numbers of unique OTUs
146 were 19, 56, and 57, respectively. In total, 218 OTUs were found to be shared by group E2 at the
147 three sampling points. On days 7, 14, and 21, the numbers of unique OTUs were 9, 50, and 80,
148 respectively. In total, 229 OTUs were found to be shared by group E3 at the three sampling
149 points. On days 7, 14, and 21, the numbers of unique OTUs were 18, 40, and 74, respectively.
150 Thus, with maturation of intestinal microbes, the microbial OTUs in the cecum became more and
151 more abundant (Figure 4).
152 We also measured the relative abundance of the cecal flora at the phylum (Figure 5A)
153 and genus levels (Figure 5B). At the phylum level, three dominant bacterial groups with a
154 relative abundance of more than 1% were identified in the experimental groups (i.e., Firmicutes,
155 Bacteroidetes, and Proteobacteria). Among these, the abundance of Firmicutes was highest in
156 groups E1 and E2 on day 7 and in group E3 on day 14. On day 14, the abundance of
157 Bacteroidetes was higher in group E2 than in the other two groups. On day 21, the abundance of
158 Bacteroidetes reached its highest level in the three groups. Moreover, the abundance of
159 Proteobacteria was highest in group E1 on day 14 and in the other two groups on day 7.
160 At the genus level, the top 10 most abundant microbes in each of the three groups
161 included Coprobacillus, Coprococcus, Blautia, Lactobacillus, Anaerotruncus, Proteus, Dorea,
162 Butyricicoccus, Escherichia, Clostridium, Ruminococcus, Oscillospira, Bacteroides, and
163 Faecalibacterium. On day 7, the relative abundances of Dorea and Anaerotruncus were
9 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
164 significantly higher in group E1 than in the other two groups, and the relative abundances of
165 Clostridium and Coprococcus were significantly higher in group E3 than in groups E1 and E2.
166 On day 14, no significant differences were observed in the relative abundances of the 10
167 dominant bacterial groups. However, the relative abundances of Ruminococcus, Clostridium, and
168 Bacillus were highest in group E1, and those of Oscillospira, Clostridium, Butyricicoccus,
169 Escherichia, Lactobacillus, and Brucella were lowest in the three groups. In group E3, the
170 relative abundances of Ruminococcus and Coprobacillus were lowest among the three groups,
171 whereas the relative abundances of Oscillospira, Dorea, Butyricicoccus, Escherichia,
172 Anaerotruncus, and Lactobacillus were highest. On day 21, the relative abundance of
173 Ruminococcus was significantly higher in group E1 than in the other two groups. On day 21, the
174 relative abundances of Clostridium, Dorea, and Brucella were significantly higher in group E2
175 than in the other groups. Additionally, the relative abundances of Oscillospira, Clostridium,
176 Dorea, Escherichia, and Anaerotruncus were highest in group E1 on day 7, and those of
177 Coprococcus and Lactobacillus were highest in group E1 on day 21. In group E2, the relative
178 abundances of Ruminococcus, Oscillospira, Clostridium, Dorea, Butyricicoccus, and
179 Escherichia were highest on day 7, and those of Coprococcus, Lactobacillus, and
180 Faecalibacterium were highest on day 21. In group E3, the relative abundances of
181 Ruminococcus, Oscillospira, Clostridium, Butyricicoccus, and Coprobacillus were highest on
182 day 7, that of Dorea was highest on day 14, and that of Lactobacillus was highest on day 21
183 (Figure 6). On day 21, the relative abundances of the three groups of lactobacilli peaked,
184 potentially because the intestinal tracts of infected chickens were conducive to the growth of
185 lactic acid bacteria.
10 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
186 LefSe analysis of the bacterial species composition in groups E1, E2, and E3 at the three
187 sampling time points revealed that on day 7, Coprococcus and Clostridium were the most
188 important bacteria differentiating group E1 from the other two groups, with the characteristic
189 colony in group E2 being Faecalibacterium. Salmonella, Streptococcus, and Anaerotruncus
190 showed the most obvious differences between the E3 group and the other two groups.
191 Salmonella could easily distinguish group E3 from the other groups. On day 14,
192 Ruminococcaceae, Clostridia, and Clostridiales in group E1 showed the most significant
193 differences between the three groups, and Faecalibacterium, Aerococcaceae, Bacteroidetes,
194 Bacteroidia, Bacteroidales, Bacteroides, and Bacteroidaceae in group E2 showed the most
195 significant differences among the three groups. In group E3. Firmicutes, Oscillospira,
196 Dehalobacteriaceae. Dehalobacterium, Anaerostipes. Erysipelotrichales, Erysipelotrichaceae, and
197 Erysipelotrichi showed the most significant differences among the groups. On day 21,
198 Microbacteriaceae, Leucobacter, and Ruminococcus in group E1 showed the most significant
199 differences among the three groups. Moreover, Defluviitalea, Coprobacillus, Eubacteriaceae,
200 Anaerofustis, Leuconostocaceae, Weissella, cc_115, Erysipelotrichales, Erysipelotrichi,
201 Erysipelptrichaceae, Dehalobacteriaceae, Dehalobacterium, Eubacterium, Clostridium, Blautia,
202 Dorea, Coprococcus, Oscillospira, Epulopiscium, Lachnospiraceae, Ruminococcaceae,
203 Clostridia, and Clostridiales showed the most significant differences in group E2. In group E3,
204 Staphylococcus, Campylobacterales, Helicobacteraceae, Akkermansia, Verrucomicrobiaceae,
205 Verrucomicrobiales, Bacillus, Lactobacillales, Lactobacillus, and Lactobacillaceae showed the
206 most significant differences among the three groups (Figure 7).
207
208 Metabolomic profiling of the cecum
11 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
209 Next, we evaluated the effects of Salmonella infection on the cecum metabolism of chickens on
210 day 14 after administering enrofloxacin using PLS-DA (Figure S4). The screening conditions for
211 the differential metabolites were as follows: (1) wariable importance in projection (VIP) ≥ 1 of
212 the first two principal components of the PLS-DA model; (2) FC ≥ 1.2 or FC ≤ 0.83; and (3) q <
213 0.05. As shown in Table 1, in the positive ion mode, 486 differential metabolites were detected
214 in groups E1 and E2 (219 increased, 267 decreased). Moreover, 1,577 differential metabolites
215 were detected in groups E1 and E3 (605 increased, 972 decreased). In groups E2 and E2, 554
216 differential metabolites were detected (231 increased, 323 decreased). In the negative ion mode,
217 271 differential metabolites were detected in groups E1 and E2 (149 increased, 122 decreased).
218 In total, 656 differential metabolites were detected in groups E1 and E3 (252 increased, 404
219 decreased). In total, 328 differential metabolites were detected in groups E2 and E3 (102
220 increased, 226 decreased; Figure 8).
221 Venn diagram analysis showed that in the positive ion mode, four differential metabolites
222 were common among the three groups, 18 differential metabolites were specific to group E2
223 versus E1, 437 differential metabolites were specific to group E3 versus E1, and 52 differential
224 metabolites were specific to group E3 versus E2. In the negative ion mode, one differential
225 metabolite was common among the three groups, 16 differential metabolites were specific to
226 group E2 versus E1, 139 differential metabolites were specific to group E3 versus E1, and 37
227 differential metabolites were specific to group E3 versus E2. These results suggested that the
228 enrofloxacin concentration affected the metabolic groups of chicken manure contents (Figure 9).
229 According to KEGG database analysis, in the negative ion mode, the differential
230 metabolites of group E2 versus E1 were enriched in four metabolic pathways: vascular smooth
231 muscle contraction (one differential metabolite), metabolic pathways (five differential
12 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
232 metabolites), vitamin B6 metabolism (one differential metabolite), and lysine degradation (one
233 differential metabolite; Figure 10A). In the positive ion mode, the differential metabolites of
234 group E2 versus E1 were enriched in seven metabolic pathways: phenylalanine metabolism
235 (three differential metabolites), metabolic pathways (12 differential metabolites), linoleic acid
236 metabolism (two differential metabolites), ⍺-linolenic acid metabolism (two differential
237 metabolites), peroxisome proliferator-activated receptor (PPAR) signaling pathway (one
238 differential metabolite), vascular smooth muscle contraction (one differential metabolite), and
239 biosynthesis of amino acids (two differential metabolites; Figure 10B).
240 In the negative ion mode, the differential metabolites of group E3 versus E1 were
241 enriched in four metabolic pathways: ⍺-linolenic acid metabolism (two differential metabolites),
242 metabolic pathways (nine differential metabolites), cysteine and methionine metabolism (two
243 differential metabolites), and vascular smooth muscle contraction (one differential metabolite;
244 Figure 10C). In the positive ion mode, the differential metabolites of group E3 versus E1 were
245 enriched in 12 metabolic pathways: metabolic pathways (35 differential metabolites); linoleic
246 acid metabolism (five differential metabolites); alanine, aspartate, and glutamate metabolism
247 (three differential metabolites); phenylalanine, tyrosine, and tryptophan biosynthesis (three
248 differential metabolites); phenylalanine metabolism (four differential metabolites); unsaturated
249 fatty acids biosynthesis (four differential metabolites); arachidonic acid metabolism (four
250 differential metabolites); tryptophan metabolism (four differential metabolites); ⍺-linolenic acid
251 metabolism (three differential metabolites); biosynthesis of amino acids (four differential
252 metabolites), tyrosine metabolism (three differential metabolites), and PPAR signaling pathway
253 (one differential metabolite; Figure 10D).
13 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
254 In the negative ion mode, the differential metabolites of group E3 versus E2 were
255 enriched in four metabolic pathways: alanine, aspartate, and glutamate metabolism (one
256 differential metabolite); ⍺-linolenic acid metabolism (one differential metabolite); lysine
257 degradation (one differential metabolite); and cysteine and methionine metabolism (one
258 differential metabolite; Figure 10E). In the positive ion mode, the differential metabolites of
259 group E3 versus E2 were enriched in eight metabolic pathways: metabolic pathways (19
260 differential metabolites); alanine, aspartate, and glutamate metabolism (two differential
261 metabolites); purine metabolism (three differential metabolites); ⍺-linolenic acid metabolism
262 (two differential metabolites); galactose metabolism (two differential metabolites); lysine
263 degradation (two differential metabolites); phenylalanine metabolism (two differential
264 metabolites); and tyrosine metabolism (two differential metabolites; Figure 10F).
265
266 Correlation analysis of the microbiome and metabolome
267 Next, we identified relationships among microbes and metabolites using multiomics.
268 Bacteroidetes were negatively correlated with phosphonate and phosphinate metabolism.
269 Verrucomicrobia were positively correlated with the intestinal immune network for IgA
270 production, fatty acid biosynthesis, and fatty acid elongation. Firmicutes were inversely
271 correlated with phototransduction. Cyanobacteria were positively correlated with cysteine and
272 methionine metabolism. Tenericutes were positively correlated with glycolysis/gluconeogenesis,
273 fatty acid metabolism, terpenoid backbone biosynthesis, fructose and mannose metabolism, and
274 purine metabolism and negatively correlated with primary bile acid biosynthesis. Proteobacteria
275 were positively correlated with galactose metabolism, pentose and glucuronate interconversion,
276 arginine biosynthesis, and primary bile acid biosynthesis. Actinobacteria were negatively
14 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
277 correlated with vitamin B6 metabolism and glycolysis/gluconeogenesis and positively correlated
278 with primary bile acid biosynthesis and drug metabolism (Figure 11).
279 At the genus level, rank correlation analysis was performed on differential metabolites
280 and microbial groups. In group E2 versus E1, Faecalibacterium were positively correlated with
281 three metabolites and negatively correlated with nine other metabolites, Bacteroides were
282 positively correlated with three metabolites and negatively correlated with one metabolite,
283 Escherichia were negatively correlated with one metabolite, and Anaerostipes were negatively
284 correlated with three metabolites. In group E3 versus E1, Dehalobacterium were positively
285 correlated with four metabolites and negatively correlated with one metabolite, whereas
286 Anaerostipes were negatively correlated with 13 metabolites and positively correlated with one
287 metabolite. In group E3 versus E2, Bacteroides were negatively correlated with two metabolites,
288 Dehalobacterium were positively correlated with three metabolites, and Faecalibacterium were
289 negatively correlated with 14 metabolites and positively correlated with one metabolite (Figure
290 12).
291
15 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
292 DISCUSSION
293 The gut microbiome plays vital roles in resistance to exogenous microorganisms via phage
294 deployment, secretion of antibacterial substances and competing nutrients, and intestinal barrier
295 function. Using antibiotics destroys the gut microbiome, making the conditions suitable for
296 pathogenic microorganisms to colonize the intestine, and affects the metabolomics. Here, we
297 investigated the effects of enrofloxacin on the colonization of S. Typhimurium in the intestinal
298 tracts of chickens. Our results showed that on days 7 and 14, S. Typhimurium was highly
299 abundant in the cecum contents of mice treated with antibiotics, similar to previous reports (9,
300 15, 17). Thus, our findings showed that different concentrations of enrofloxacin altered the
301 ability of S. Typhimurium to colonize the gut in chickens.
302 In this study, early enrofloxacin treatment suppressed S. Typhimurium colonization.
303 However, as the gut microbial community matured, the abundance of S. Typhimurium in
304 untreated chickens was gradually decreased, whereas that in enrofloxacin-treated chickens
305 gradually increased and persisted. These results were similar to those of a previous study (15)
306 and suggested that enrofloxacin may cause severe S. Typhimurium infection.
307 On day 7, the highest diversity was observed in the high-dose group, and the lowest
308 diversity was observed in the low-dose group. On day 21, the low-dose group showed the highest
309 diversity, whereas the high-dose group showed the lowest diversity. These results are different
310 from previous findings, potentially because of the use of antibiotics to challenge Salmonella in
311 this study (9, 18, 19). Generally, the alpha diversity of the cecum flora increases with age. A
312 higher flora diversity is thought to be a marker of mature intestinal flora that is less sensitive to
313 environmental factors and less susceptible to infection (20). Previous studies have shown that
16 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
314 mice with lower intestinal flora complexity are less resistant to Streptococcus enterocolitis
315 colonization and more susceptible to enterocolitis than normal mice (21).
316 Our findings showed that using antibiotics significantly altered the microbial community
317 structure, with clear differences according to the antibiotic concentration. The main reasons for
318 the changes in microbial structure were decreased Proteobacteria and Firmicutes and increased
319 Bacteroidetes at the three sampling points. Indeed, the early microbial community mainly
320 comprised Firmicutes and Proteobacteria, accounting for more than 90% of the total sequence,
321 whereas the late microbial community was dominated by Bacteroidetes and Firmicutes. Studies
322 have shown that Bacteroides strains can produce propionate to reduce the colonization of S.
323 Typhimurium (10).
324 Firmicutes are primarily anaerobic bacteria. The anatomic physiology and feeding habits
325 of chickens may explain the high abundance of Firmicutes. Turnbaugh et al. found that the
326 appendices of obese mice contain higher concentrations of acetic acid and butyric acid and that
327 the ratio of Firmicutes to Bacteroidetes is increased in comparison with normal mice (22).
328 Changes in the ratio of Firmicutes to Bacteroidetes in the experimental group may affect the
329 accumulation of chicken fat and the production of SCFAs, which may explain the higher
330 abundance of S. Typhimurium in group E2 on day 14. Additionally, on day 14, the abundances of
331 Enterobacteriaceae in groups E2 and E3 were lower than that in group E1, although the
332 difference was not significant.
333 Enterobacteriaceae can inhibit the colonization of Salmonella in the intestine by
334 competing for oxygen (13). Notably, the abundances of Anaerotruncus, Butyricicoccus,
335 Ruminococcus, Dorea, and Clostridium were gradually decreased. In addition, higher
17 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
336 concentrations of enrofloxacin resulted in lower abundances of Ruminococcus and
337 Anaerotruncus on days 7 and 14, similar to a previous study of florfenicol (23).
338 Many anaerobic bacteria can produce SCFAs via microbial fermentation of
339 carbohydrates, which have immunomodulatory and anti-inflammatory effects. Butyric acid can
340 mediate the inhibition of Salmonella by downregulating the expression of pathogenic island I
341 and can also reduce Salmonella-induced pro-inflammatory responses to intestinal cells in vitro
342 (24, 25). These aforementioned bacteria compete with Salmonella for oxygen and produce high
343 concentrations of SCFAs to limit their colonization (26). SCFAs exert inflammatory effects by
344 regulating the production of cytokines and prostaglandins (27) and can enhance host immune
345 function. Sunkara et al. found that SCFAs reduce the Salmonella load in the intestinal contents
346 and enhance the host’s defense ability against Salmonella (28). Butyrate can promote the β-
347 oxidation of colonic epithelial cells, reduce the oxygen partial pressure in the intestinal cavity,
348 and inhibit pathogenic bacteria (29). Any reduction in SCFA production could explain the high
349 abundance of Salmonella in the cecum of antibiotic-treated chickens. In this study, on day 14,
350 Lactobacillus and Oscillospira numbers were higher in groups E2 and E3 than in group E1,
351 consistent with previous studies. This may be due to the microaerophilic growth of Lactobacillus
352 and the production of reactive oxygen species by granulocytes in the infected chicken cecum to
353 penetrate the site of inflammation and provide facorable growth conditions for lactic acid
354 bacteria (26, 30, 31). Notably, accumulation of lactic acid may damage the intestinal defense
355 barrier and increase the osmotic load in the intestinal cavity, and Salmonella can oxidize lactic
356 acid to enhance their intestinal adaptability (14, 32). Thus, our results indicated that enrofloxacin
357 decreased the colonization resistance of the original S. Typhimurium in the intestine, allowing S.
358 Typhimurium to more easily colonize the intestine.
18 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
359 Tumor necrosis factor-α (TNF-α) and MUC2 genes were upregulated in the cecum of
360 chickens infected with S. Typhimurium following enrofloxacin treatment. These genes are
361 important indicators for evaluating the inflammatory response in birds and can induce an
362 inflammatory response following pathogen infection. MUC2 is the most important mucin (33).
363 Salmonella can use type I flagellates to bind to mannose residues distributed on the surface of the
364 mucin layer to attach to the mucin surface and colonize the intestine (34). In this study, IL-18
365 expression was inhibited in the enrofloxacin group. IL-18 can induce Th1-type cells and NK
366 cells to produce IFN-γ, thereby promoting neutrophil and macrophage killing and clearance of
367 intracellular bacteria. In addition, downregulation of IL-18 increases bacterial load and decreases
368 survival in the liver and spleen (35, 36).
369 Occludin (OCLN) is a major tight junction protein in the intestine. In this study, gene
370 expression was significantly downregulated in group E3, suggesting that high concentrations of
371 enrofloxacin may diffuse more effectively throughout the body by reducing tight junction
372 activity between the intestinal tissue and intestinal mucosal epithelial cells. Additionally, the
373 expression of immune-related genes was significantly altered in chickens infected with S.
374 Typhimurium after enrofloxacin administration, with increased pro-inflammatory gene
375 expression and decreased anti-inflammatory gene expression in comparison with the untreated
376 group. These findings suggested that enhanced S. Typhimurium colonization may cause a more
377 severe inflammatory response in the intestinal tract of chickens.
378 Metabolomic analysis showed that the differential metabolic pathways shared by groups
379 E2 versus E1 and E3 versus E1 included linoleic acid metabolism and ⍺-linolenic acid
380 metabolism. According to several reports, polyunsaturated fatty acids, such as linoleic acid, α-
381 linolenic acid, and γ-linoleic acid, are metabolized by macrophages into fatty acids with a ketone
19 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
382 structure. Notably, fatty acids containing an enone structure have the most effective anti-
383 inflammatory activity (37).
384 Linolenic acid can kill Helicobacter pylori and reduce the bacterial load in the stomachs
385 of mice. In vivo, treatment with linolenic acid reduces the levels of pro-inflammatory cytokines
386 (e.g., IL-1β, IL-6, and TNF-α) (38). Interestingly, addition of flaxseed oil to the diet of pigs
387 recovered the levels of α-linolenic acid, eicosapentaenoic acid, and total n-3 polyunsaturated
388 fatty acids in the intestinal tract and improved intestinal morphology. Flaxseed oil can also
389 increase jejunal lactase activity and claudin-1 protein expression; downregulate the mRNA
390 expression of intestinal necrosis signals; downregulate intestinal TLR4 mRNA, its downstream
391 signals (MyD88, NF-κB, NOD1, and NOD2), and their adaptor molecules; and decrease the
392 functionality of RIPK2 expression (39).
393 Notably, the changes observed in the PPAR signaling pathway and amino acid
394 biosynthesis were due to the metabolic effects of S. Typhimurium colonization in the chicken
395 cecum. The abundance of α-linolenic acid was lower in groups E2 and E3 than in group E1 and
396 lower in group E3 than in group E2. This decrease in the level of α-linolenic acid may be the
397 reason for the high abundance of S. Typhimurium in groups E2 and E3. However, further studies
398 are required to elucidate the effects of administering α-linolenic acid on S. Typhimurium
399 colonization.
400 In this study, we found that IgA production by the Verrucomicrobia intestinal immune
401 network was positively related to fatty acid biosynthesis. The genus Verrucomicrobia contains
402 only one member, i.e., Akkermansia muciniphila, which is a member of the human intestinal
403 flora and a common bacterium that degrade mucins in the gut (40, 41). Colonization of
404 Aspergillus slime has been reported to be protective against diet-induced obesity (42, 43) and to
20 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
405 promote mucosal wound healing (44) and antitumor responses during anti-programmed death-1
406 immunotherapy (45). Therefore, when enrofloxacin is administered, changes in Verrucomicrobia
407 may affect S. Typhimurium colonization.
408 In this study, we found that Faecalibacterium was associated with the most metabolites
409 in the E2 versus E1 and E3 versus E2 groups and that Anaerostipes was associated with the most
410 metabolites in the E3 versus E1 group. Several studies have reported that the metabolites
411 secreted by Faecalibacterium prausnitzii attenuate inflammation in cells and 2,4,6-
412 trinitrobenzenesulphonic acid-induced colitis models by blocking nuclear factor (NF)-κB
413 activation and IL-8 production (46). Notably, F. prausnitzii secretes a protein with a molecular
414 weight of 15 kDa, which mediates resistance to colitis in animal models by inhibiting the NF-κB
415 pathway in intestinal epithelial cells (47). Moreover, F. prausnitzii can adapt to the mucus layer
416 and ferment to produce butyrate. The absence of F. prausnitzii and other microorganisms with
417 similar functions is often accompanied by inflammatory bowel disease and obesity (48). These
418 bacteria may be the major bacteria affecting the intestinal metabolome following enrofloxacin
419 administration.
420 Our results indicated that infection with S. Typhimurium after enrofloxacin administration
421 affected the colony structure, metabolite composition, and cecum gene expression in the chicken
422 cecum, thereby altering S. Typhimurium colonization in the gut. The effects varied according to
423 the concentration of enrofloxacin used. The colonies and metabolites identified in this study may
424 provide important insights into the prevention and control of S. Typhimurium infection. The
425 observed increase in S. Typhimurium infection induced by enrofloxacin also facilitate the more
426 careful and rational use of antibiotics in poultry.
427
21 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
428 MATERIALS AND METHODS
429 Bacterial strain
430 S. Typhimurium ATCC14028 (stored in our laboratory) was incubated in lysogeny broth at 37°C
431 for 16 h. For bacterial counting, the liquid was diluted and inoculated on XLT4 agar plates at
432 37°C for 24 h.
433
434 Enrofloxacin exposure and infection in chickens
435 The experimental protocol involving animals in this study was approved by the Animal Ethics
436 Committee of Sichuan University. One-day-old specific-pathogen-free chickens (n = 180) were
437 randomly divided into six groups (n = 30 each). Each group was housed in a separate isolator
438 containing drug-free feed and water. From 1 to 5 days of age, chickens were fed different
439 concentrations for enrofloxacin (C1 and E1 [control group], 200 μL distilled water; C2 and E2,
440 200 μL of 10 mg/kg enrofloxacin; C3 and E3, 200 μL of 100 mg/kg enrofloxacin). At 6 days of
441 age, chickens in groups E1, E2, and E3 were challenged with oral administration of 1 × 108
442 colony-forming units (CFU) S. Typhimurium, whereas chickens in groups C1, C2, and C3 were
443 given equal amounts of Luria-Bertani broth. All chickens were euthanized on days 7, 14, or 21 (n
444 = 10/group at each time), and cecum contents, heart, spleen, liver, and ceca were collected. All
445 cecum contents were collected from the same part of the cecum. The ileum and colon were
446 rapidly clamped to avoid overflow of gastrointestinal digestive fluids, which can contaminate
447 other parts of the intestine (Figure 13).
448
449 Total RNA extraction
22 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
450 Total RNA from chicken cecum tissues previously collected on day 14 was extracted using
451 TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions.
452 Approximately 60 mg tissue was ground into a powder in the presence of liquid nitrogen in a 2-
453 mL tube, followed by homogenization for 2 min and then resting in a horizontal position for
454 5 min. The mixture was centrifuged (12,000 × g) at 4°C for 5 min, and the supernatant was then
455 taken and placed in a new Eppendorf tube. A mixture of 0.3 mL chloroform/isoamyl alcohol
456 (24:1) was injected, shaken vigorously for 15 s, and then centrifuged at 4°C for 10 min at
457 12,000 × g. The remaining upper aqueous phase RNA was then transferred to a new tube, and an
458 equal volume of isopropanol alcohol was added, followed by centrifugation at 4°C for 20 min at
459 12,000 × g. After discarding the supernatant, the RNA pellet was washed twice with 1 mL of
460 75% ethanol, and the mixture was centrifuged at 12,000 × g for 3 min at 4°C to collect any
461 residual ethanol. The pellet was air dried for 5–10 min in a biosafety cabinet, and RNA was
462 dissolved with 25–100 μL diethylpyrocarbonate-treated water. Subsequently, a
463 spectrophotometer and bioanalyzer were used for total RNA characterization and quantification.
464
465 16S rRNA amplicon sequencing
466 Samples were collected from 10 chickens from the same group at each sampling time. According
467 to the manufacturer’s instructions, a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany)
468 was used to test 200 mg aliquots and 30 mg tissue from each chicken manure sample. Total
469 genomic DNA was extracted from the samples, and the concentration and quality of the
470 extracted DNA were measured using a NanoDrop spectrophotometer and agarose, followed by
471 confirmation using gel electrophoresis. The extracted DNA was stored at −20°C until further
472 analysis.
23 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
473 Using the isolated genomic DNA as the template, polymerase chain reaction (PCR)
474 amplification was performed on the V3-V4 hypervariable region of the bacterial 16S rRNA gene.
475 The primers used were 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-
476 GGACTACHVGGGTWTCTAAT-3′). Sample-specific 7-bp barcodes were assigned to
477 multiplex sequencing primers. The 25-μL PCR system and PCR reaction parameters for PCR
478 amplification were set up using Agencourt. AMPure XP magnetic beads were used for
479 purification of PCR amplification products, and products were dissolved in elution buffer. The
480 libraries were labeled, and the fragment range and concentration of the libraries were checked
481 using an Agilent 2100 Bioanalyzer. The qualified libraries were sequenced on a HiSeq platform
482 according to the inserted fragment size.
483
484 16S rRNA sequencing data analysis
485 Both off-machine data filtering and the remaining high-quality clean data are used for analyses.
486 Overlaps between reads were used to stitch the reads into tags, cluster the tags into operational
487 taxonomic units (OTUs), and compare them with the database. The reads were also used for
488 species annotation. Sample species complexity analysis and analysis of species differences
489 between groups were performed according to the OTU and annotation results as well as
490 correlation analysis and model prediction. The original sequencing data were processed as
491 follows to obtain clean data. For sequence splicing, Fast Length Adjustment of SHort reads
492 software (FLASH, v.1.2.11) (49) was used to assemble the paired reads obtained by paired-end
493 sequencing into a sequence using overlapping relationships to obtain tags in hypervariable
494 regions.
24 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
495 Generally, OTUs are a unified flag set for taxonomic units (strain, genus, species,
496 grouping, etc.), facilitating analyses in phylogenetic or population genetic studies. To elucidate
497 the bacterial number, genus, and other information in the sequencing results of a sample,
498 clustering operations were performed. Using the classification operation, sequences were
499 classified into many groups according to their similarities, with each group being considered an
500 OTU. Tags with a similarity above 97% were clustered into an OTU.
501 After obtaining the OTU representative sequence, RDP Classifier software (v.2.2) (50)
502 was used to compare this sequence with the database for species annotation (confidence
503 threshold: 0.8). To obtain the species classification information corresponding to each OTU, the
504 RDP Classifier’s Bayesian algorithm was used to conduct a taxonomic analysis on the OTU
505 representative sequences and for statistical analysis of the phylum, class, order, family, and
506 genus level composition for each sample.
507
508 Bioinformatics and statistical analysis
509 Alpha diversity is an analysis of the diversity of species in a single sample. The richer the species
510 in the sample, the larger the observed species index, Chao index, ACE index, Shannon index,
511 and Simpson index and the smaller the Good coverage index (51). In addition, higher Good
512 coverage values indicated lower the probability that the sequence in the sample was not detected.
513 Beta-diversity analyses (51-53) were used to compare differences between pairs of samples in
514 terms of species diversity. In this analysis, the content of each species in the sample was first
515 analyzed, and the beta-diversity value was then calculated between samples. Beta-diversity
516 analysis was performed using QIIME (54) (v.1.80). Linear discriminant analysis (LDA) effect
517 size (LefSe) (55) was used to identify high-dimensional biomarkers and reveal genomic features
25 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
518 based on genes, metabolism, and classifications for distinguishing between two or more
519 biological taxa.
520
521 Untargeted metabolomics for chicken cecal content
522 After thawing samples slowly at 4°C, 25 mg was measured and added to a 1.5-mL Eppendorf
523 tube. Then, 800 μL extract (methanol/acetonitrile/water, 2:2:1, v/v/v, precooled at −20°C), 10 μL
524 internal standard 1, and 10 μL internal standard 2 were added. Two small steel balls were then
525 added, and samples were ground in a tissue grinder (50 Hz, 5 min), sonicated at 4°C for 10 min,
526 and placed in a freezer at −20°C for 1 h. After centrifugation, 600 μL supernatant was removed,
527 placed in a refrigerated vacuum concentrator, and dried. Next, 200 μL of a double solution
528 (methanol/H2O, 1:9, v/v) was added for recombination, and the samples were vortexed for 1 min
529 and sonicated for 10 min at 4°C. The supernatants were then centrifuged at 25,000 × g for 15 min
530 at 4°C and placed in a sample vial. Then, 20 μL supernatant from each sample was mixed with
531 quality control samples to assess the repeatability and stability of the liquid chromatography-
532 mass spectrometry (LC-MS) analysis. A BEH C18 column (1.7 μm × 2.1 mm × 100 mm; Waters,
533 Milford, MA, USA) was used for metabolite isolation. The mobile phases used in the positive
534 ion mode were an aqueous solution containing 0.1% formic acid (liquid A) and a solution of
535 100% methanol containing 0.1% formic acid (liquid B). The mobile phases in the negative ion
536 mode were an aqueous solution containing 10 mM ammonium formate (liquid A) and a solution
537 of 10 mM ammonium formate and 95% methanol (liquid B). A Q Exactive mass spectrometer
538 (Thermo Fisher Scientific) was used for primary and secondary mass spectrometry data
539 acquisition. All LC-MS/MS mass spectrometry raw data (original files) were sent to Compound
540 Discoverer (v.3.0; Thermo Fisher Scientific) for data processing. Identification of metabolites
26 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
541 was performed using several databases, such as the BGI Library, mzCloud, and ChemSpider
542 (HMDB, KEGG, and LipidMAPS). Results from Compound Discoverer v.3.0 were imported
543 into metaX (56) for data preprocessing. All the identified metabolites were classified and
544 annotated with reference to the Kyoto Encyclopedia of Genes and Genomes (KEGG) and
545 HMDB databases to understand the classification of the metabolites. Multivariate statistical
546 analysis (principal component analysis [PCA] and partial least squares discriminant analysis
547 [PLS-DA]) and univariate analysis (fold change [FC] and Student’s t-tests) were used to screen
548 for between-group differential metabolites. Here, PCA and PLS-DA (57, 58) were used to model
549 the relationship between metabolite expression and sample category, which allowed the
550 prediction of the sample category, combined with the multiplier of difference and t-test, to finally
551 identify the differences in metabolites between groups. Data were analyzed using clustering
552 analysis, log2 conversion, and z-score normalization (zero mean normalization). Hierarchical
553 clustering was used in the clustering algorithm, and the Euclidean distance was used to calculate
554 the metabolic pathway enrichment of differential metabolites in the KEGG database.
555
556 Bacterial enumeration
557 First, 200 mg organ tissue and cecum content from each animal was added to 10 mL sterile saline
558 and shaken well. Then, 1 mL bacterial solution was added to 9 mL sterile saline for
559 homogenization; samples were serially diluted and spread on XLT4 agar plates. The medium
560 was then incubated for 24 h under aerobic conditions at 37°C, and S. Typhimurium colonies were
561 counted. The results are expressed as log10 CFU/g. All analyses were performed in triplicate.
562
563 Real-time quantitative reverse transcription PCR (qPCR)
27 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
564 Total RNA was extracted as previously described. The quality and concentration of RNA were
565 determined using a Nanodrop 2000 spectrophotometer. Briefly, 1 μg total RNA from each
566 sample was reverse-transcribed into cDNA using SuperScript II (Life Technologies, Invitrogen).
567 The primers used for reverse transcription were oligonucleotide (dT) primers and random
568 hexamers. Quantitative PCR was performed using 2 × T5 Fast qPCR Mix (SYBR Green I) with a
569 Bio-Rad CFX Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, USA)
570 according to the manufacturer’s protocol. Table S1 shows the primers used in this study. The
571 relative mRNA expression levels of each target gene were calculated using the 2−ΔΔCT method
572 with β-actin as the endogene. In all cases, all samples were reacted in triplicate in parallel.
573
574 Correlation analysis between the microbiome and metabolome
575 Pretreatment of the collected positive and negative ions in the metabolome yielded 3,417
576 metabolites, which were downscaled using gene co-expression network analysis, yielding 42 co-
577 expression clusters. In total, 424 of the 3,417 metabolites were annotated into 95 pathways, and
578 pathway abundance was calculated on the basis of metabolite intensity. Correlations between
579 metabolite intensity and microbial abundance were assessed using Spearman’s correlation
580 analysis.
581
582 Statistical analysis
583 S. Typhimurium count results are expressed in log10 CFU/g. The relative mRNA expression
584 levels of each immune gene were calculated using the 2-ΔΔCT method. One-factor ANOVA was
585 performed on the above results using SPASS 20.0.
586
28 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
587 ACKNOWLEDGMENTS
588 This work was supported by the General Program of National Natural Science Foundation of
589 China (grant nos. 31830098 and 3177131163), the China Agriculture Research System National
590 System for Layer Production Technology (grant no. CARS-40-K14), the Key Research and
591 Development Projects in Sichuan Province (grant no. 2014NZ0002), and the Fundamental
592 Research Funds for the Central Universities (grant no. SCU2019D013).
29 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
593 REFERENCES
594 1. Kirk MD, Pires SM, Black RE, Caipo M, Crump JA, Devleesschauwer B, Dopfer D,
595 Fazil A, Fischer-Walker CL, Hald T, Hall AJ, Keddy KH, Lake RJ, Lanata CF,
596 Torgerson PR, Havelaar AH, Angulo FJ. 2015. World Health Organization estimates
597 of the global and regional disease burden of 22 foodborne bacterial, protozoal, and
598 viral diseases, 2010: a data synthesis. PLoS Med 12:e1001921.
599 2. Wu QJ, Zheng XC, Wang T, Zhang TY. 2018. Effect of dietary oridonin
600 supplementation on growth performance, gut health, and immune response of broilers
601 infected with Salmonella pullorum. Ir Vet J 71:16.
602 3. LaRock DL, Chaudhary A, Miller SI. 2015. Salmonellae interactions with host
603 processes. Nat Rev Microbiol 13:191–205.
604 4. Hallstrom K, McCormick BA. 2011. Salmonella interaction with and passage through
605 the intestinal mucosa: through the lens of the organism. Front Microbiol 2:88.
606 5. de Jong HK, Parry CM, van der Poll T, Wiersinga WJ. 2012. Host-pathogen
607 interaction in invasive Salmonellosis. PLoS Pathog 8:e1002933.
608 6. Zhang QQ, Ying GG, Pan CG, Liu YS, Zhao JL. 2015. Comprehensive evaluation of
609 antibiotics emission and fate in the river basins of China: source analysis, multimedia
610 modeling, and linkage to bacterial resistance. Environ Sci Technol 49:6772–6782.
611 7. Van Boeckel TP, Brower C, Gilbert M, Grenfell BT, Levin SA, Robinson TP,
612 Teillant A, Laxminarayan R. 2015. Global trends in antimicrobial use in food
613 animals. Proc Natl Acad Sci U S A 112:5649–5654.
614 8. Zhao L, Dong YH, Wang H. 2010. Residues of veterinary antibiotics in manures from
615 feedlot livestock in eight provinces of China. Sci Total Environ 408:1069–1075.
30 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
616 9. Li J, Hao H, Cheng G, Liu C, Ahmed S, Shabbir MAB, Hussain HI, Dai M, Yuan Z.
617 2017. Microbial shifts in the intestinal microbiota of Salmonella infected chickens in
618 response to enrofloxacin. Front Microbiol 8:1711.
619 10. Jacobson A, Lam L, Rajendram M, Tamburini F, Honeycutt J, Pham T, Van Treuren
620 W, Pruss K, Stabler SR, Lugo K, Bouley DM, Vilches-Moure JG, Smith M,
621 Sonnenburg JL, Bhatt AS, Huang KC, Monack D. 2018. A gut commensal-produced
622 metabolite mediates colonization resistance to Salmonella infection. Cell Host
623 Microbe 24:296–307.e297.
624 11. He T, Zhu YH, Yu J, Xia B, Liu X, Yang GY, Su JH, Guo L, Wang ML, Wang JF.
625 2019. Lactobacillus johnsonii L531 reduces pathogen load and helps maintain short-
626 chain fatty acid levels in the intestines of pigs challenged with Salmonella enterica
627 Infantis. Vet Microbiol 230:187–194.
628 12. Borton MA, Sabag-Daigle A, Wu J, Solden LM, O'Banion BS, Daly RA, Wolfe RA,
629 Gonzalez JF, Wysocki VH, Ahmer BMM, Wrighton KC. 2017. Chemical and
630 pathogen-induced inflammation disrupt the murine intestinal microbiome.
631 Microbiome 5:47.
632 13. Litvak Y, Mon KKZ, Nguyen H, Chanthavixay G, Liou M, Velazquez EM, Kutter L,
633 Alcantara MA, Byndloss MX, Tiffany CR, Walker GT, Faber F, Zhu Y, Bronner DN,
634 Byndloss AJ, Tsolis RM, Zhou H, Baumler AJ. 2019. Commensal Enterobacteriaceae
635 protect against Salmonella colonization through oxygen competition. Cell Host
636 Microbe 25:128–139 e125.
637 14. Gillis CC, Hughes ER, Spiga L, Winter MG, Zhu W, Furtado de Carvalho T, Chanin
638 RB, Behrendt CL, Hooper LV, Santos RL, Winter SE. 2018. Dysbiosis-associated
31 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
639 change in host metabolism generates lactate to support Salmonella growth. Cell Host
640 Microbe 23:54–64 e56.
641 15. Ng KM, Ferreyra JA, Higginbottom SK, Lynch JB, Kashyap PC, Gopinath S, Naidu
642 N, Choudhury B, Weimer BC, Monack DM, Sonnenburg JL. 2013. Microbiota-
643 liberated host sugars facilitate post-antibiotic expansion of enteric pathogens. Nature
644 502:96–99.
645 16. Suwandi A, Galeev A, Riedel R, Sharma S, Seeger K, Sterzenbach T, Garcia Pastor
646 L, Boyle EC, Gal-Mor O, Hensel M, Casadesus J, Baines JF, Grassl GA. 2019. Std
647 fimbriae-fucose interaction increases Salmonella-induced intestinal inflammation and
648 prolongs colonization. PLoS Pathog 15:e1007915.
649 17. Lichtman JS, Ferreyra JA, Ng KM, Smits SA, Sonnenburg JL, Elias JE. 2016. Host-
650 microbiota interactions in the pathogenesis of antibiotic-associated diseases. Cell Rep
651 14:1049–1061.
652 18. Ahmed N, Qu A, Brulc JM, Wilson MK, Law BF, Theoret JR, Joens LA, Konkel
653 ME, Angly F, Dinsdale EA, Edwards RA, Nelson KE, White BA. 2008. Comparative
654 metagenomics reveals host specific metavirulomes and horizontal gene transfer
655 elements in the chicken cecum microbiome. PLoS One 3.
656 19. Danzeisen JL, Kim HB, Isaacson RE, Tu ZJ, Johnson TJ. 2011. Modulations of the
657 chicken cecal microbiome and metagenome in response to anticoccidial and growth
658 promoter treatment. PLoS One 6:e27949.
659 20. Huang MZ, Wang SY, Wang H, Cui DA, Yang YJ, Liu XW, Kong XJ, Li JY. 2018.
660 Differences in the intestinal microbiota between uninfected piglets and piglets
661 infected with porcine epidemic diarrhea virus. PLoS One 13:e0192992.
32 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
662 21. Stecher B, Chaffron S, Kappeli R, Hapfelmeier S, Freedrich S, Weber TC, Kirundi J,
663 Suar M, McCoy KD, von Mering C, Macpherson AJ, Hardt WD. 2010. Like will to
664 like: abundances of closely related species can predict susceptibility to intestinal
665 colonization by pathogenic and commensal bacteria. PLoS Pathog 6:e1000711.
666 22. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. 2006. An
667 obesity-associated gut microbiome with increased capacity for energy harvest. Nature
668 444:1027–1031.
669 23. Li R, Wang H, Shi Q, Wang N, Zhang Z, Xiong C, Liu J, Chen Y, Jiang L, Jiang Q.
670 2017. Effects of oral florfenicol and azithromycin on gut microbiota and adipogenesis
671 in mice. PLoS One 12:e0181690.
672 24. Bearson SM, Bearson BL, Rasmussen MA. 2006. Identification of Salmonella
673 enterica serovar Typhimurium genes important for survival in the swine gastric
674 environment. Appl Environ Microbiol 72:2829–2836.
675 25. Gantois I, Ducatelle R, Pasmans F, Haesebrouck F, Hautefort I, Thompson A, Hinton
676 JC, Van Immerseel F. 2006. Butyrate specifically down-regulates salmonella
677 pathogenicity island 1 gene expression. Appl Environ Microbiol 72:946–949.
678 26. Arguello H, Estelle J, Leonard FC, Crispie F, Cotter PD, O'Sullivan O, Lynch H,
679 Walia K, Duffy G, Lawlor PG, Gardiner GE,2019. Influence of the intestinal
680 microbiota on colonization resistance to Salmonella and the shedding pattern of
681 naturally exposed pigs. mSystems 4.
682 27. Cox MA, Jackson J, Stanton M, Rojas-Triana A, Bober L, Laverty M, Yang X, Zhu
683 F, Liu J, Wang S, Monsma F, Vassileva G, Maguire M, Gustafson E, Bayne M, Chou
684 CC, Lundell D, Jenh CH. 2009. Short-chain fatty acids act as antiinflammatory
33 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
685 mediators by regulating prostaglandin E (2) and cytokines. World J Gastroenterol
686 15:5549–5557.
687 28. Sunkara LT, Achanta M, Schreiber NB, Bommineni YR, Dai G, Jiang W, Lamont S,
688 Lillehoj HS, Beker A, Teeter RG, Zhang G. 2011. Butyrate enhances disease
689 resistance of chickens by inducing antimicrobial host defense peptide gene
690 expression. PLoS One 6:e27225.
691 29. Byndloss MX, Olsan EE, Rivera-Chávez F, Tiffany CR, Cevallos SA, Lokken KL,
692 Torres TP, Byndloss AJ, Faber F, Gao Y. 2017. Microbiota-activated PPAR-γ
693 signaling inhibits dysbiotic Enterobacteriaceae expansion. Science (New York, N.Y.)
694 357:570–575.
695 30. Vídeňská P, Sisak F, Havlickova H, Faldynova M, Rychlík I. 2013. Influence of
696 Salmonella enterica serovar Enteritidis infection on the composition of chicken cecal
697 microbiota. BMC Vet Res 9:140.
698 31. Drumo R, Pesciaroli M, Ruggeri J, Tarantino M, Chirullo B, Pistoia C, Petrucci P,
699 Martinelli N, Moscati L, Manuali E, Pavone S, Picciolini M, Ammendola S, Gabai G,
700 Battistoni A, Pezzotti G, Alborali GL, Napolioni V, Pasquali P, Magistrali CF. 2015.
701 Salmonella enterica serovar Typhimurium exploits inflammation to modify swine
702 intestinal microbiota. Front Cell Infect Microbiol 5:106.
703 32. Ling Z, Liu X, Jia X, Cheng Y, Luo Y, Yuan L, Wang Y, Zhao C, Guo S, Li L, Xu X,
704 Xiang C. 2014. Impacts of infection with different toxigenic Clostridium difficile
705 strains on faecal microbiota in children. Sci Rep 4:7485.
706 33. Wu M, Wu Y, Li J, Bao Y, Guo Y, Yang W. 2018. The dynamic changes of gut
707 microbiota in Muc2 deficient mice. Int J Mol Sci 19.
34 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
708 34. Kolenda R, Ugorski M, Grzymajlo K. 2019. Everything you always wanted to know
709 about Salmonella type 1 fimbriae, but were afraid to ask. Front Microbiol 10:1017.
710 35. Miao EA, Leaf IA, Treuting PM, Mao DP, Dors M, Sarkar A, Warren SE, Wewers
711 MD, Aderem A. 2010. Caspase-1-induced pyroptosis is an innate immune effector
712 mechanism against intracellular bacteria. Nat Immunol 11:1136–1142.
713 36. Bao S, Beagley K, Mp, Shen J, Husband A. 2000. Interferon-gamma plays a critical
714 role in intestinal immunity against Salmonella typhimurium infection. Immunology
715 99:464–472.
716 37. Yang HE, Li Y, Nishimura A, Jheng HF, Yuliana A, Kitano-Ohue R, Nomura W,
717 Takahashi N, Kim CS, Yu R, Kitamura N, Park SB, Kishino S, Ogawa J, Kawada T,
718 Goto T. 2017. Synthesized enone fatty acids resembling metabolites from gut
719 microbiota suppress macrophage-mediated inflammation in adipocytes. Mol Nutr
720 Food Res 61.
721 38. Thamphiwatana S, Gao W, Obonyo M, Zhang L. 2014. In vivo treatment of
722 Helicobacter pylori infection with liposomal linolenic acid reduces colonization and
723 ameliorates inflammation. Proc Natl Acad Sci U S A 111:17600–17605.
724 39. Zhu H, Wang H, Wang S, Tu Z, Zhang L, Wang X, Hou Y, Wang C, Chen J, Liu Y.
725 2018. Flaxseed oil attenuates intestinal damage and inflammation by regulating
726 necroptosis and TLR4/NOD signaling pathways following lipopolysaccharide
727 challenge in a piglet model. Mol Nutr Food Res 62:e1700814.
728 40. Derrien M, Vaughan EE, Plugge CM, de Vos WM. 2004. Akkermansia muciniphila
729 gen. nov., sp. nov., a human intestinal mucin-degrading bacterium. Int J Syst Evol
730 Microbiol 54:1469–1476.
35 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
731 41. Derrien M, Collado MC, Ben-Amor K, Salminen S, de Vos WM. 2008. The Mucin
732 degrader Akkermansia muciniphila is an abundant resident of the human intestinal
733 tract. Appl Environ Microbiol 74:1646–1648.
734 42. Everard A, Belzer C, Geurts L, Ouwerkerk JP, Druart C, Bindels LB, Guiot Y,
735 Derrien M, Muccioli GG, Delzenne NM, de Vos WM, Cani PD. 2013. Cross-talk
736 between Akkermansia muciniphila and intestinal epithelium controls diet-induced
737 obesity. Proc Natl Acad Sci U S A 110:9066–9071.
738 43. Greer RL, Dong X, Moraes AC, Zielke RA, Fernandes GR, Peremyslova E, Vasquez-
739 Perez S, Schoenborn AA, Gomes EP, Pereira AC, Ferreira SR, Yao M, Fuss IJ,
740 Strober W, Sikora AE, Taylor GA, Gulati AS, Morgun A, Shulzhenko N. 2016.
741 Akkermansia muciniphila mediates negative effects of IFNgamma on glucose
742 metabolism. Nat Commun 7:13329.
743 44. Alam A, Leoni G, Quiros M, Wu H, Desai C, Nishio H, Jones RM, Nusrat A, Neish
744 AS. 2016. The microenvironment of injured murine gut elicits a local pro-restitutive
745 microbiota. Nat Microbiol 1:15021.
746 45. Routy B, Le Chatelier E, Derosa L, Duong CP, Alou MT, Daillère R, Fluckiger A,
747 Messaoudene M, Rauber C, Roberti MP. 2018. Gut microbiome influences efficacy
748 of PD-1–based immunotherapy against epithelial tumors. Science (New York, N.Y.)
749 359:91–97.
750 46. Sokol H, Pigneur B, Watterlot L, Lakhdari O, Bermúdez-Humarán LG, Gratadoux J-
751 J, Blugeon S, Bridonneau C, Furet J-P, Corthier G. 2008. Faecalibacterium
752 prausnitzii is an anti-inflammatory commensal bacterium identified by gut microbiota
753 analysis of Crohn disease patients. Proc Natl Acad Sci U S A 105:16731–16736.
36 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
754 47. Quevrain E, Maubert MA, Michon C, Chain F, Marquant R, Tailhades J, Miquel S,
755 Carlier L, Bermudez-Humaran LG, Pigneur B, Lequin O, Kharrat P, Thomas G,
756 Rainteau D, Aubry C, Breyner N, Afonso C, Lavielle S, Grill JP, Chassaing G, Chatel
757 JM, Trugnan G, Xavier R, Langella P, Sokol H, Seksik P. 2016. Identification of an
758 anti-inflammatory protein from Faecalibacterium prausnitzii, a commensal bacterium
759 deficient in Crohn's disease. Gut 65:415–425.
760 48. No authors listed. 2015. Your microbes at work: fiber fermenters keep us healthy.
761 Nature 518:S9.
762 49. Fletcher JR, Erwin S, Lanzas C, Theriot CM. 2018. Shifts in the gut metabolome and
763 Clostridium difficile transcriptome throughout colonization and infection in a mouse
764 Modem. mSphere 3.
765 50. Wang Q, Garrity GM, Tiedje JM, Cole JR. 2007. Naive Bayesian classifier for rapid
766 assignment of rRNA sequences into the new bacterial taxonomy. Appl Environ
767 Microbiol 73:5261–5267.
768 51. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, Lesniewski
769 RA, Oakley BB, Parks DH, Robinson CJ, Sahl JW, Stres B, Thallinger GG, Van
770 Horn DJ, Weber CF. 2009. Introducing mothur: open-source, platform-independent,
771 community-supported software for describing and comparing microbial communities.
772 Appl Environ Microbiol 75:7537–7541.
773 52. Lozupone C, Lladser ME, Knights D, Stombaugh J, Knight R. 2011. UniFrac: an
774 effective distance metric for microbial community comparison. ISME J 5:169–172.
37 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
775 53. Lozupone CA, Hamady M, Kelley ST, Knight R. 2007. Quantitative and qualitative
776 beta diversity measures lead to different insights into factors that structure microbial
777 communities. Appl Environ Microbiol 73:1576–1585.
778 54. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK,
779 Fierer N, Peña AG, Goodrich JK, Gordon JI. 2010. QIIME allows analysis of high-
780 throughput community sequencing data. Nat Methods 7:335–336.
781 55. Manuel M. 2013. A new semi-subterranean diving beetle of the Hydroporus
782 normandi-complex from south-eastern France, with notes on other taxa of the
783 complex (Coleoptera: Dytiscidae). Zootaxa 3652:453–474.
784 56. Wen B, Mei Z, Zeng C, Liu S. 2017. metaX: a flexible and comprehensive software
785 for processing metabolomics data. BMC Bioinform 18:183.
786 57. Barker M, Rayens W. 2003. Partial least squares for discrimination. J Chemomet
787 17:166–173.
788 58. Westerhuis JA, Hoefsloot HCJ, Smit S, Vis DJ, Smilde AK, van Velzen EJJ, van
789 Duijnhoven JPM, van Dorsten FA. 2008. Assessment of PLSDA cross validation.
790 Metabolomics 4:81–89.
38 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
791 FIGURE LEGENDS
792 Figure 1. Salmonella abundance. (A) Cecum contents, (B) heart, (C) liver, and (D) spleen. The
793 x-axis represents log10 copies/g, and the y-axis represents different groupings. Compared with the
794 same sampling point in group E1 group, P < 0.001, indicated by ***; 0.001 ≤ P ≤ 0.01, indicated
795 by **; and 0.01 < P ≤ 0.05, indicated by *. P > 0.05, no mark. Comparing groups E2 and E3, P <
796 0.001, indicated by ###; 0.001 ≤ P ≤ 0.01, indicated by ##; and 0.01 < P ≤ 0.05, indicated by #.
797 P > 0.05, n.s.
798
799 Figure 2. Gene expression. The β-actin gene was used as an endogene, and relative expression
800 was calculated by normalization to this gene using the 2−ΔΔCT method. (A) IL-1β, IL-8L2, and
801 OCLM; (B) TLR4, TLR5, and TLR21; (C) LITAF, NOD, and TJP; (D) IL-18, IL-8L1, and MUC2
802 on the x-axis, with different genes on the y-axis. Compared with group E1 at the same sampling
803 point, P < 0.001, indicated by ***; 0.001 ≤ P ≤ 0.01, indicated by **; and 0.01 < P ≤ 0.05,
804 indicated by *. P > 0.05, no mark. Comparing groups E2 and E3, P < 0.001, indicated by ###;
805 0.001 ≤ P ≤ 0.01, indicated by ##; and 0.01 < P ≤ 0.05, indicated by #. P > 0.05, n. s.
806
807 Figure 3. Beta-diversity analysis of the cecum microbiota after Salmonella infection. According
808 to the statistical results of the variability of each sample, a clustering analysis was performed on
809 the samples in groups E1, E2, and E3 at the three sampling points and the intersample distance
810 was calculated. (A, B, C) Cluster trees for groups E1, E2, and E3 on days 7, 14, and 21. (D, E, F)
811 Cluster trees for groups E1, E2, and E3 at the three sampling points.
812
39 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
813 Figure 4. The left histogram shows the number of OTUs on the x-axis and the name of the
814 comparison group on the y-axis. The upper right histogram shows the intersection of the
815 different comparison groups on the x-axis and the number of OTUs on the y-axis. Each column
816 in the lower right indicates the relationship between the left comparison group and the
817 intersection above. Dark dots indicate that the intersection contains the comparison group, light
818 dots indicate that the intersection does not contain the comparison group, and dark dots are
819 connected by lines.
820
821 Figure 5. Average relative abundance of microbial species in the cecum at the (A) phylum level
822 and (B) genus level.
823
824 Figure 6. Key species difference comparisons, selecting the top 10 species with the highest
825 abundance, demonstrating the mean relative abundance of each group and the significance of the
826 difference test. (A, B, C) Groups E1, E2, and E3 groups with the top 10 species with the highest
827 abundance on days 7, 14, and 21. (D, E, F) Alpha diversity of groups E1, E2, and E3 at the three
828 sampling points. P < 0.001, denoted by **; 0.001 ≤ P ≤ 0.01, denoted by **; and 0.01 < P ≤
829 0.05, denoted by *. P > 0.05, no mark.
830
831 Figure 7. LefSe analysis of the chicken cecum microbial community in groups E1, E2, and E3 at
832 the three sampling points. LefSe plots showing microbial strains with significant differences in
833 groups E1, E2, and E3 (A, C, E). The different groups are represented by different colors, the
834 microbiota that plays an important role in the different groups is represented by nodes of
835 corresponding colors, and the organism markers are indicated by colored circles. The microbiota
40 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
836 that does not play a significant role in the different groups is indicated by yellow nodes. From
837 inside to outside, the circles are ordered by species at the level of phylum, class, order, family,
838 and genus. Linear discriminant analysis (LDA) diagram (B, D, F). The different colors represent
839 microbial groups that play a significant role in groups E1, E2, and E3. Biomarkers with statistical
840 differences are emphasized, with the colors of the histograms representing the respective groups
841 and the lengths representing the LDA score, which is the magnitude of the effects of significantly
842 different species between groups.
843
844 Figure 8. Heat map showing differential metabolites with VIP ≥ 1 for the first two principal
845 components of the PLS-DA model, fold change (FC) ≥ 1.2 or FC ≤ 0.83, and q < 0.05. The green
846 color represents downregulated differential metabolites, and the red color represents upregulated
847 differential metabolites. (A, B) Differential metabolites in the negative or positive ion mode of
848 group E2 versus E1; (C, D) differential metabolites in the negative or positive ion mode of group
849 E3 versus E1; (E, F) differential metabolites in the negative or positive ion mode of group E3
850 versus E2.
851
852 Figure 9. Venn diagrams showing common or unique differences in metabolites between the
853 three groups. (A) Negative ion mode; (B) positive ion mode.
854
855 Figure 10. KEGG pathway analysis of metabolites in the contents of the appendix on day 14.
856 The metabolites in the different comparison groups were analyzed for metabolic pathway
857 enrichment on the basis of the KEGG database. The x-axis (RichFactor) represents the ratio of
858 the number of differential metabolites annotated to the pathway to all the metabolites annotated
41 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
859 to the pathway. The size of the dots represents the number of differential metabolites annotated
860 to the pathway. (A) Upregulated metabolites in group E2 versus E1, (B) downregulated
861 metabolites in group E2 versus E1, (C) upregulated metabolites in group E3 versus E1, (D)
862 downregulated metabolites in group E3 versus E1, (E) upregulated metabolites in group E3
863 versus E2, and (F) downregulated metabolites in group E3 versus E2.
864
865 Figure 11. Rank correlation analysis of metabolic pathways and phylum-level microbial groups.
866 The top 20 relationship pairs with the strongest rank correlation (with the largest correlation
867 coefficient absolute value and a P value of < 0.05) are shown as chord diagrams. The blue line in
868 the middle represents a positive correlation, and the red line represents a negative correlation.
869
870 Figure 12. Rank correlation analysis was performed on differential metabolites and genus-level
871 microbial groups. The top 20 relationship pairs with the strongest rank correlation (with the
872 largest correlation coefficient absolute value and a P value of < 0.05) are shown as chord
873 diagrams. The blue line in the middle represents a positive correlation, and the red line represents
874 a negative correlation. (A) Group E2 versus E1, (B) group E3 versus E1, and (C) group E3
875 versus E2.
876
877 Figure 13. Animal experiment design. In total, 180 specific-pathogen-free chickens were
878 randomly divided into six groups (C1, C2, C3, E1, E2, and E3). Groups E1, E2, and E3 were fed
879 Salmonella Pullorum 1 × 108 colony-forming units (CFUs)/mL per chicken at 6 days of age, and
880 groups E1 and C1 were fed equal amounts of lactose broth. Samples were taken from each
881 chicken at 7, 14, and 21 days of age.
42 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
882
883 Figure S1. Rarefaction curves for chicken-blind flora. (A) ACE index, (B) Chao1 index, (C)
884 Good coverage index, (D) observed species index, (E) Shannon index, and (F) Simpson index.
885
886 Figure S2. Alpha-diversity analysis of the cecum microbiota after Salmonella infection. (A, B,
887 C) Alpha diversity in groups E1, E2, and E3 on days 7, 14, and 21. (D, E, F) Alpha diversity in
888 groups E1, E2, and E3 at the three sampling points.
889
890 Figure S3. Beta-diversity analysis of the cecum microbiota after Salmonella infection. Principal
891 component analysis (PCA), in which the absolute abundance of operational taxonomic units
892 (OTUs) in groups E1, E2, and E3 at the three sampling points was used to calculate the
893 distribution of each sample. (A, B, C) PCA for groups E1, E2, and E3 on days 7, 14, and 21. (D,
894 E, F) Cluster tree for groups E1, E2, and E3 at the three sampling points.
895
896 Figure S4. Partial least squares discriminant analysis (PLS-DA) fractional scatterplots of the
897 identified metabolites with the first principal component on the horizontal axis and the second
898 principal component on the vertical axis. The number in parentheses indicates the score for that
899 principal component and represents the percentage of the overall variance explained by the
900 corresponding principal component. (A, B) For metabolites in the negative or positive ion mode
901 of group E2 versus E1; (C, D) for metabolites in the negative or positive ion mode of group E3
902 versus E1; (E, F) for metabolites in the negative or positive ion mode of group E3 versus E2.
43 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.
903 TABLE
904 Table 1. Differential metabolite statistics.
Analysis mode Significantly different metabolite species (VIP ≥ 1, FC ≥ 1.2 or FC ≤ 0.83,
Comparison group q < 0.05)
Positive ion Negative ion Trend Trend mode mode
E2 versus E1 219↑ 149↑ 486 271 267↓ 122↓
E3 versus E1 605↑ 252↑ 1577 656 972↓ 404↓
E3 versus E2 231↑ 102↑ 554 328 323↓ 226↓
905
44 bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. bioRxiv preprint doi: https://doi.org/10.1101/2020.07.13.201830; this version posted July 14, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.