1

Running title: Ruminal epithelial bacterial colonization 1

2

Taxonomic identification of the ruminal epithelial bacterial diversity during rumen development 3

in goats 4

5

Jinzhen Jiao1,2, Jinyu Huang1,2, Chuanshe Zhou1 and Zhiliang Tan1,# 6

7

1Key Laboratory for Agro-Ecological Processes in Subtropical Region, and South-Central 8

Experimental Station of Animal Nutrition and Feed Science in Ministry of Agriculture, Institute 9

of Subtropical Agriculture, The Chinese Academy of Sciences, Changsha, Hunan 410125, P.R. 10

China; 11

2Graduate University of Chinese Academy of Sciences, Beijing 100049, P. R. China; 12

#Corresponding author: Zhiliang Tan. Address: Institute of Subtropical Agriculture, the Chinese 13

Academy of Sciences, Changsha, Hunan 410125, P. R. China. E-mail: zltan@isa.ac.cn; Tel: +86 14

731 4615230; Fax: +86 731 4612685 15

16

Key words: epithelial bacteria, rumen, microbial colonization, diversity. 17

18

AEM Accepted Manuscript Posted Online 13 March 2015Appl. Environ. Microbiol. doi:10.1128/AEM.00203-15Copyright © 2015, American Society for Microbiology. All Rights Reserved.

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

2

Abstract 19

20

Understanding of the colonization process of epithelial bacteria attached to the rumen tissue 21

during rumen development is very limited. Ruminal epithelial bacterial colonization is of great 22

significance to explore the relationship between the microbiota and the host, as well as can 23

influence host early development and health. Miseq sequencing of 16S rRNA genes and 24

quantitative real-time PCR (qPCR) were applied to characterize ruminal epithelial bacterial 25

diversity during rumen development in this study. Seventeen kid goats were selected to reflect 26

the non-rumination (0 and 7 d), transition (28 and 42 d) and rumination (70 d) phases of animal 27

development. Alpha diversity indices (OTU numbers, Chao and Shannon index) increased (P < 28

0.01) with age, and PCoA analysis revealed that the samples clustered together according to their 29

particular age groups. Phylogenetic analysis revealed that Proteobacteria, Firmicutes and 30

Bacteroidetes were detected as the dominant phyla regardless of the age group, whilst 31

Proteobacteria abundance declined quadratically with age (P < 0.001), while Bacteroidetes (P = 32

0.088) and Firmicutes (P = 0.009) increased with age. At the genus level, Escherichia (80.79%) 33

dominated at 0 d, while Prevotella, Butyrivibrio and Campylobacter surged (linearly, P < 0.01) 34

in abundance at 42 and 70 d. qPCR showed that total epithelial bacteria copy number linearly 35

increased (P = 0.013) with age. In addition, the abundances of genera Butyrivibrio, 36

Campylobacter, Desulfobulbus were positively correlated with rumen weight, rumen papilla 37

length, ruminal ammonia and total volatile fatty acid concentrations, and activities of 38

carboxymethylcellulase (CMCase) and xylanase, respectively. Collectively, colonization of 39

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

3

ruminal epithelial bacteria is age-related (achieved at 2 months), as well as they might participate 40

in the anatomic and functional development of the rumen. 41

42

43

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

4

Introduction 44

45

The recent years have witnessed growing interest in the diversity and function of ruminal 46

epithelial bacteria. It has been demonstrated that ruminal epithelial bacteria are involved in 47

oxygen scavenging, tissue recycling and urea digestion (1). Furthermore, in steers, ruminal 48

epithelial bacterial community is different between acidosis-resistant and acidosis-susceptible 49

groups during subacute ruminal acidosis development, and such difference can be recognized by 50

the host TLRs (Toll like receptors) which are associated with the variation of the rumen epithelial 51

tissue barrier function (2). Similarly, in mice colon, epithelial bacterial diversity correlated with 52

TLR2 and TLR4 gene expressions (3). These reveal that as epithelial bacteria are directly 53

attached to the epithelial surface, their end products may directly or indirectly play a role in host 54

immune responses and tissue barrier function. 55

56

The ruminal epithelial bacteria are distinctly different at taxonomic level from those associated 57

with rumen contents (4, 5). Using culture-based techniques, the ruminal epithelial communities 58

have been found to comprise predominantly Gram-positive, facultatively anaerobic flora, with 59

Butyrivibrio sp. (31.1%), Bacteroides sp. (22.4%), Selenomonas ruminantium (9.9%), 60

Succinivibrio dextrinosolvens (8.7%) and Streptococcus bovis (8.1%) predominant (6). Through 61

pyrosequencing, ruminal epithelial bacteria at the class level were mainly composed of 62

Clostridia (67%), Bacteroidia (9%), Deltaproteobacteria (4%) and Erysipelotrichia (3%) (7). 63

64

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

5

Diet is an indispensable determinant influencing the structure and function of microbial diversity 65

both in rumen content and tissue-attached. In the rumen contents of dairy cows, deep amplicon 66

sequencing of the 16S rRNA gene revealed higher abundance of the Fibrobacteraceae family in 67

total mixed ration samples while lower members of the propionate-producing Veillonelaceae 68

compared with pasture samples (8). Moreover, based on the sequence information from the 69

reference marker of PCR-DGGE, the phyla Firmicutes and Proteobacteria abundances in heifer 70

ruminal epithelial bacterial community decreased in response to the rapid dietary transition from 71

a diet containing 97% hay to a diet containing 8% hay (9). 72

73

The bacterial colonization process in the developing rumen is of interest to understand the 74

achievement of rumen functions. It can influence the early development and productive 75

efficiency of the mature animal (10). In cow rumen contents, the bacterial composition 76

underwent dynamic changes during the first two years, with the Proteobacteria phylum 77

decreased with age while Bacteroidetes and Firmicutes increased with age (11). Our earlier study 78

showed that during the three typical phases of rumen development process (non-rumination, 0 to 79

3 weeks; transition, 3 to 8 week; rumination period, from 8 week on) in goats, microbial 80

colonization in the rumen content is achieved at 1 month, functional achievement at 2 months, 81

and anatomic development after 2 months (12). Nonetheless, research into the age-related 82

dynamic colonization of ruminal epithelial bacterial populations is very limited. Therefore, we 83

aimed to characterize the sequential dynamics of ruminal epithelial bacterial colonization and to 84

explore its relationship with previously published rumen anatomic and functional parameters. 85

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

6

86

Materials and Methods 87

88

Sample collection 89

90

This is a companion study conducted using samples collected from selected 17 supplemental 91

feeding kid goats (offered kids with some concentrate to the forage after 20 d of birth) used in 92

our previous study (12). The use of animals, including welfare, husbandry, experimental 93

procedures, and the collection of rumen samples used for this study, was approved by the Animal 94

Care Committee, Institute of Subtropical Agriculture, the Chinese Academy of Sciences, 95

Changsha, China. Post-birth, the kid goat was immediately placed in an individual pen to avoid 96

direct contact with other adult animals and other kids. The kids were then maintained within an 97

individual pen for the duration of the study. From 0 to 20 d, the kids were twice offered with goat 98

milk (1 L per meal) at 08.00 and 17.00 h. Between 20 and 40 d, feeds were provided individually 99

twice daily, at 08.00 and 17.00 h, at each meal the kids received a bucket of goat milk (0.5 L), 100

and a separate bucket containing a mixture of forage (fresh grass; 0.04 kg DM/meal) and 101

concentrate starter (0.12 kg DM/meal). The kids were weaned at 40 d, and from there until 70 d, 102

they were fed with concentrate starter (0.17 kg DM/meal)) and forage (0.06 kg DM/meal). The 103

concentrate starter (kg−1 DM basis) was composed of 74.1 g whey powder, 211 g corn flour, 320 104

g bean meal, 65 g fish meal, 220 g fat powder, 51 g milk powder, 8.6 g CaCO3, 25.3 g CaHPO4, 105

5 g NaCl and 20 g premix. The concentrate starter and forage were offered ad libitum. Four kids 106

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

7

were slaughtered at 0 and 42 d, and three kids were slaughtered at 7, 28 and 70 d. Immediately 107

after the kids were slaughtered, the rumen tissue were scraped to remove attached feed particles 108

and rinsed three times with sterile phosphate buffered saline (pH = 7.4) to remove non-adherent 109

bacteria. The samples were then transferred into RNA-later solution (Invitrogen, Carlsbad, CA) 110

and stored at -20°C until further molecular analysis. 111

112

DNA extraction and amplicon preparation for high-throughput sequencing 113

114

Total DNA was extracted from the 17 rumen mucosal tissue samples using the bead beating 115

method as suggested by Chen, et al. (9). The quality and quantity of DNA were measured using a 116

NanoDrop ND1000 (NanoDrop Technologies Inc., DE, USA). Total DNA of rumen tissue was 117

then diluted to 50 ng/μl and used in amplicon preparation for high-throughput sequencing. 118

Conventional PCR was performed to amplify the V2 and V3 region of 16S rRNA gene using 119

universal primer 104F (5'-NNNNNN GGCGVACGGGTGAGTAA-3') and 530R 120

(5'-CCGCNGCNGCTGGCAC-3') (13), with the forward primer 104F contained 6 base barcode 121

sequence (italicized poly-N section of primer above) at the 5’ terminus. All 104F primer barcodes 122

used are presented in Table S1. PCR reactions (50 ul) contained 30 ng DNA template, 5μl of 10 × 123

Pfx Amplification Buffer, 2μl of 10 mM dNTP mixture, 2μl of 50 mM MgSO4, 1μl of Platinum® 124

Pfx DNA Polymerase, 0.25 μl of bovine serum albumin (20 mg/mL) and 2μl of each primer (10 125

uM) (Invitrogen, Life Sciences, USA). Reaction conditions consisted of an initial 95oC for 3 min, 126

followed by 35 cycles of 95oC for 30 s, 58oC for 30 s, and 72oC for 45 s, and a final extension of 127

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

8

72oC for 10 min. PCR products were excised from a 1.5% agarose gel and purified using 128

Wizard® SV Gel and PCR Clean-Up System (Promega, Madison, USA). Barcoded V2–V3 129

amplicons were mixed at equal molar rations, submitted to Illumina paired-end library 130

preparation, cluster generation, and 300 bp pair-end sequencing on an Illumina MiSeq PE300 131

instrument. 132

133

Data analysis 134

135

Raw data were filtered through a quality control pipeline using Quantitative Insight into 136

Microbial Ecology (QIIME) tool kit (14) and mothur (15). Specifically, the 300-bp reads were 137

truncated at any site of more than three sequential bases receiving a quality score < Q20, and any 138

read containing ambiguous base calls or barcode/primer errors were discarded, as were reads 139

with < 80% (of total read length) consecutive high-quality base calls. Sequences were 140

deconvoluted into individual samples based on their barcodes. For pair-end reads, only sequences 141

that overlap longer than > 10% and with < 10 % mismatch were assembled according to their 142

overlap sequence using Connecting Overlapped Pair-End (COPE) (16). The assembled 143

sequences were then trimmed of primers, assigned to operational taxonomic units (OTUs) at 144

97% identity threshold using UPASE (17). Chimeric sequences were identified and removed 145

using UCHIME (18). Taxonomy classification was assigned against the latest Greengenes 146

database (Greengenes May 2013 release) (19) using the mothur-based implementation of the 147

RDP Bayesian classifier with a 0.80 confidence threshold. Sequence alignment was performed 148

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

9

by PyNAST and phylogenetic tree was constructed using FastTree (20). The taxonomic 149

identification and comparisons were performed at the phylum and genus level. 150

151

Alpha diversity of ruminal epithelial bacterial communities of kids at different ages (0, 7, 28, 42 152

and 70 d) was obtained using various diversity indices (observed species, Chao1, ACE, Shannon 153

and Simpson index) using the alpha rarefaction pipeline at a sequence depth of 27,000 as the 154

nearest number to the minimum number of all 17 sequences (27,120). PCoA and hierarchical 155

clustering of microbial communities (Jackkinfied beta diversity of 100 times resampling at a 156

depth of 20250 sequences) was performed using Bray Curtis distance. The clustering method of 157

hierarchical clustering was unweighted pair group method with arithmetic mean (UPGMA). 158

Within-group similarity was calculated as the average of the pairwise similarity between each 159

paired sample within each group using the Bray-Curtis metric. Sequencing data in this study 160

were deposited in the MG-RAST server under IDs 4613702.3 to 4613735.3. And they were also 161

deposited in the NCBI sequence read archive (SRA) under the accession numbers SRR1823055, 162

SRR1823127, SRR1823131, SRR1823138, SRR1823141, SRR1830869, SRR1830870 and 163

SRR1830872- SRR1830881. 164

165

A transformation-based principal components analysis (tb-PCA) (21) using the Hellinger 166

distance was used to ordinate the 17 samples according to OTU abundance data. Only the top 10 167

OTUs for which variation of abundance contributed most to the clustering obtained from the 168

tb-PCA were selected and represented in the tb-PCA. The OTUs were identified using the largest 169

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

10

contributors, to the first two principal components as selection criteria. 170

171

Quantification of total bacteria and selected bacterial genera 172

173

Absolute quantitative real-time PCR was performed to quantify copy numbers of 16S rRNA 174

genes of total bacteria and two selected bacterial genera (Prevotella and Bacteroides). The 175

previous validated primers (22-24) are listed in Table S2. The qPCR was performed using 176

SYBR® Premix Ex Taq™ (Perfect Real Time) (Takara, Japan) on an ABI 7900HT Fast Real 177

Time PCR system (Applied Biosystems, Foster City, CA, USA). The standard curves for total 178

bacteria and each bacterial genus were prepared using plasmid DNA containing each unique 16S 179

rRNA insert. The copy numbers of 16S rRNA gene/g of fresh tissue were calculated as described 180

by Zhou, et al. (25). The relative abundance of bacterial genera was calculated by dividing the 181

16S rRNA gene copy number of each genus by the 16S rRNA gene copy number of total 182

bacteria. 183

184

Statistical analysis 185

186

For data of apparent relative abundance of communities at the phylum and genus level, alpha 187

diversity indices (Miseq sequencing) as well as qPCR results of total ruminal epithelial bacteria 188

and abundant bacterial genera, they were analyzed as a completely randomized design with the 189

MIXED procedures of SAS (SAS Inst. Inc., Cary, NC). The model included the fixed effect of 190

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

11

age and the random effect of animal within age, with individual animal as the experiment unit. 191

Orthogonal contrasts were used to test for linear and quadratic effects of age. Statistical 192

significance was accepted at P < 0.05 and a trend was considered at P < 0.10. All presented data 193

are expressed as least-squares means. 194

195

Spearman’s rank correlations between ruminal bacterial community (MiSeq relative abundance) 196

and anatomic, functional variables were analyzed using the PROC CORR procedure of SAS. 197

Only bacterial groups that represented > 2% of the total community in at least one sample and 198

that were detected in > 50% of the rumen tissue samples were included in the analysis. 199

200

Results 201

202

Sampling depth, coverage and alpha diversity 203

204

After data filtering, quality control, pair-end reads assembling, primer, chimera, and low 205

confidence singleton removal, a total of 1,064,895 of V2-V3 16S rRNA sequences reads from the 206

17 samples with an average of 62,641 sequences reads for each sample (the minimum value of 207

one sample was 27,120 and the maximum was 130,437) were used in this project. The average 208

length of sequences reads after primer removal was 383bp. The overall number of OTUs 209

detected by the analysis reached 2409 based on 97% nucleotide sequence identity between reads. 210

211

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

12

With subsample of 27,000 reads every sample, the rarefaction curves (Fig.S1) showed that most 212

of our sampling effort provided sufficient OTU coverage to accurately describe the bacterial 213

composition of each group. Alpha diversity measures (Table 1) indicated that age unaffected (P > 214

0.05) coverage and had an increasing trend (P = 0.058) on Ace index. Moreover, number of 215

OTUs, Chao and Shannon index increased linearly with age (P < 0.05) while Simpson index 216

declined linearly (P = 0.001) with age. 217

218

Taxonomic composition of ruminal epithelial bacterial composition across different age groups 219

220

In total, twenty-six phyla were identified within the ruminal epithelial microbiota. The 221

abundance of 15 phyla in all the samples was < 0.5% and they were Acidobacteria, 222

Cyanobacteria, Elusimicrobia, Gemmatimonadetes, Lentisphaerae, NC10, Nitrospirae, OD1, 223

OP11, Planctomycetes, Spirochaetes, Synergistetes, TM6, Tenericutes and Verrucomicrobia. 224

Among the 26 phyla, the Proteobacteria, Firmicutes and Bacteroidetes were detected as the 225

dominant phyla regardless of the age group (Table 2), but their ratio and composition among the 226

groups varied considerately. The phylum Proteobacteria was the most abundant phylum in 227

samples from 0 and 7 d, reaching as high as more than 85%. Afterwards, it declined quadratically 228

(P < 0.001) with age. The Bacteroidetes phylum exhibited low values during the first week (< 229

5%), and increased significantly afterwards (approximately 10%). The Firmicutes increased 230

quadratically (P = 0.009) with age, becoming dominant at 42 and 70 d (54.71% and 40.29%, 231

respectively). The Actinobacteria (P = 0.096) and TM7 tended to increase linearly (P = 0.067) 232

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

13

with age while age had quadratic effects on Chloroflexi (P = 0.014) and Thermi (P = 0.051) 233

abundance. Furthermore, the Fibrobacteres and Fusobacteria peaked at 0.49% relative 234

abundance at 42 d and 1.18% at 0 d, respectively. 235

236

Of the 26 genera observed with over 0.5% of relative abundance, fifteen changed significantly 237

over age (Table 3), others remained relatively stable. Moreover, a great proportion of sequences 238

(ranging from 13.38% to 82.99%) remained unclassified at the genus level. Within the 239

Bacteroidetes phylum, the genus Bacteroides peaked at 2.46% (7 d), Porphyromonas decreased 240

linearly (P = 0.011) with age, while Prevotella increased linearly (P = 0.022) with age. In the 241

Firmicutes phylum, Butyrivibrio genus exhibited low values during the first week (< 1%), 242

becoming the most dominant afterwards, peaked at 42.38% relative abundance at 42 d. 243

Mogibacterium increased linearly (P = 0.003) with age, p-75-a5 tended to increase quadratically 244

(P = 0.077) with age. In contrast, the relative proportions of Oscillospira and Streptococcus 245

declined with age. In the Proteobacteria phylum, genus Escherichia was very dominant at 0 d 246

(80.79%), decreased quadratically (P < 0.001) with age, and reached stable values (about 1%) 247

after 28 d. On the contrary, Campylobacter was very minor during the first week (< 1%), surged 248

in proportion afterwards, and peaked at 29.99% at 70 d. Age had decreasing effects on genera 249

Arcobacter (linearly, P = 0.038) and Enterobacter (linearly, P = 0.077), while conversely had 250

effects on Desulfobulbus (linearly, P = 0.033) and Paracoccus (quadratically, P = 0.075). Of 251

other minor genera did not belong to these three phyla, Actinomyces, SHD-231, Fibrobacter, 252

Fusobacterium and Thermus peaked at 0.21% (70 d), 0.80% (28 d), 0.49% (42 d), 1.18% (0 d) 253

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

14

and 1.31% (42 d), respectively. 254

255

OTU diversity and similarity analysis 256

257

To minimize the variation created by different sample depth, Jackkinfied subsampling (100 times 258

resampling at a depth of 20250 sequences) was used in our study before performing beta 259

diversity analysis. PCoA analysis using Bray-Curtis similarity metric revealed that the samples 260

clustered together according to their particular age group (Fig.1). The average within-group 261

similarity showed a significant difference between different age groups, with their values at 28, 262

42 and 70 d exceeded those at 0 and 7 d (Fig. 2). 263

264

Ten dominant OTUs were identified that contributed to the separation of ruminal bacterial 265

community at five different ages (Fig. 3). The two dominant OTUs that separated samples at 0 d 266

from others were identified as Escherichia spp. Two OTUs Neisseriaceae and Ruminococcaceae 267

family separated samples at 7 d from others. Likewise, for the separation of 28 d, one OTU was 268

identified (Ruminococcus spp.). The four OTUs contributing to the separation of 42 d and 70 d 269

from others displayed high identity to Butyrivibro spp. and Campylobacter spp. 270

271

Quantification of total ruminal epithelial bacteria and abundant bacterial genera 272

273

As illustrated by qPCR, total ruminal epithelial bacterial copy numbers (P = 0.013) and genus 274

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

15

Prevotella abundance increased linearly (P = 0.049) with age (Table 4). Age did not affect (P > 275

0.05) genus Bacteroides abundance. 276

277

Relationship between bacterial community, anatomic and functional variables 278

279

Bacterial community at the genus level and anatomic, functional variables were regarded as 280

correlated with each other if their correlation coefficients were above 0.55. The relative 281

abundances of genera Bacteroides, Escherichia and Fusobacterium were negatively correlated 282

with live weight and rumen weight, while the relative abundances of Butyrivibrio, 283

Campylobacter, Desulfobulbus, Prevotella and Ruminococcus were positively correlated with 284

live weight and rumen weight (Fig. 4). Rumen papilla length was positively correlated with the 285

relative abundances of Butyrivibrio, Campylobacter, Desulfobulbus and Ruminococcus while 286

negatively correlated with the relative abundances of Escherichia and Fusobacterium. Ammonia 287

concentration was positively correlated with total bacteria copy number, Butyrivibrio, 288

Campylobacter and Ruminococcus abundances, while negatively correlated with Escherichia and 289

Fusobacterium. Total VFA concentration was positively correlated with Butyrivibrio, 290

Campylobacter, Desulfobulbus abundances while negatively correlated with Bacteroides, 291

Escherichia and Fusobacterium abundances. Furthermore, xylanase activity was positively 292

correlated with Butyrivibrio, Campylobacter and Desulfobulbus abundances, while negatively 293

correlated with Acinetobacter, Bacteroides, Escherichia, Fusobacterium and Thermus 294

abundances. CMCase activity was positively correlated with Desulfobulbus abundance, while 295

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

16

negatively correlated with Acinetobacter, Bacteroides, Escherichia, Fusobacterium and Thermus 296

abundances. Amylase activity was negatively correlated with Butyrivibrio abundance. 297

298

Discussion 299

300

Relationships between gastrointestinal microbial communities and their hosts have been shown 301

in recent years to have an important role in host’s well-being and proper function. The objective 302

of this study was to evaluate how bacteria colonized within the rumen epithelium during normal 303

development. Using Miseq sequencing, we obtained 62,641 reads on average for each sample 304

(the least one at 27,120 reads) with good coverage (> 99.6%). Furthermore, our results suggest 305

that each age group has its own distinct epithelial microbiota, as reflected by clustering of 306

samples by age group using PCoA and tb-PCA. Meanwhile, most of the alpha diversity indices 307

(except ACE and Simpson) increased with age, suggesting that microbiota in older age group is 308

more diverse than those in the earlier age group. This is similar to the microbial colonization 309

process in the rumen contents (11, 26). 310

311

Individual variation is an indispensable determinant affecting microbiota taxa. Although the 312

experimental conditions, diet and sampling procedures were similar between bovine rumen 313

samples, a large fraction of the OTUs (50%) occurred in only a small number of samples (0 to 314

30%), implying bacterial taxa varied considerably between individual bovine rumen contents 315

(27). Similarly, within-group similarity in rumen epithelial bacterial community in our study 316

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

17

during the first week is quite low (around 0.2). Despite this, we observed a drastic increase in 317

within-group similarity with age. This suggests that the rumen epithelium in older groups is a 318

more restricted environment, housing a more homogeneous bacterial community compared to the 319

earlier heterogeneous community. 320

321

Host genetics is another factor affecting gastrointestinal microbial adaption and evolution (28). 322

In new born calves, bacterial community of fecal content samples from twin calves revealed 323

higher similarity in PCR single strand conformation polymorphism (PCR-SSCP) profiles for 324

twins compared to their co-residents of the same age, indicating that the individual microbial 325

diversity might be genetically influenced (29). In our study, effects of age on rumen epithelial 326

bacterial diversity were compared using different groups of kids, which also accounted for the 327

variation. Moreover, total epithelial bacterial populations for alfalfa hay and oats-fed dairy calves 328

(30), hay-fed sheep(31), concentrate and fresh grass-fed kid goats (this study) ranged from 4.4 × 329

109 to 4.2 × 1010, 4.4 × 107 to 2.2 × 108, 1.4 × 108 to 4.4 × 107 per g of wet tissue, respectively. 330

Besides the variation in model animal, diet, DNA extraction method, PCR primers and 331

conditions, the host might also be an important determinant in regulating bacterial density. 332

333

In all age groups, Firmicutes, Bacteroidetes and Proteobacteria were the dominant phyla in the 334

epithelial microbiota, with their abundance and genus composition varied remarkably, in line 335

with other previous studies(7, 32). Similarly, at the genus level, although the abundance of some 336

genera changed with age, however, they were present in all groups, representing a core 337

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

18

microbiome in the ruminal epithelial microbiota of goats. When compared to those in the rumen 338

contents (11), after 20 d, the current study showed the epithelial microbiota harbored more 339

genera Butyrivibrio and Campylobacter, indicating these two genera might be unique abundant 340

epithelial bacteria. 341

342

The ruminal epithelial bacterial composition of kid goats undergoes sequential changes during 343

rumen development process. On the day born, Proteobacteria (90.13%) phylum far exceeded 344

other phylum, with the majority belonging to the genus Escherichia. These microbiota might 345

derive from the microbiota of mother’s vagina(33), skin, colostrum(34) and environment. This 346

genus is a facultative anaerobic bacteria, with some commensal strains of E. coli can reduce 347

intestinal epithelial inflammatory signaling in vitro (35), while some pathogenic strains of E.coli 348

can invade host cells (36). As reflected by tb-PCA, two dominant OTUs that separated samples 349

at 0 d from others also displayed high identity to this genus. These facultative anaerobes could 350

create the reduced-environment that is required for anaerobic microbes (37). Further work would 351

be needed to ascertain the rumen functions of this genus. Moreover, this genus declined 352

considerately in abundance at 7 d, and a great many genera belonged to unclassified 353

Proteobacteria. The remarkable changes observed in ruminal epithelial bacterial communities 354

during the first week after birth indicated a rapid change in the rumen environment. 355

356

A noticeable change occurring at 28 d after solid food offered (20 d) was the increase in ruminal 357

epithelial bacterial density, and this was reflected by qPCR (almost three times in copy numbers). 358

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

19

Similar observations have been reported in rumen and gut contents (37). Moreover, similar to the 359

compositional changes in rumen contents (11), in rumen tissues, genus Bacteriodes declined in 360

abundance while genera Prevotella, Butyrivibrio, Campylobacter and Succinivibrio surged in 361

proportion. Collectively, solid food caused disruption in ruminal epithelial microbiota by 362

selecting bacterial taxa that were more specific and adapted to new substrates. The genus 363

Prevotella, numerically predominant ruminants, are capable of utilizing starches, other 364

noncellulosic polysaccharides, and simple sugars as energy sources (24). Butyrivibrio group 365

(including the genera Butyrivibrio and Pseudobutyrivibrio) represented 12.98% of total bacteria 366

the rumen of goats (38) and were involved in the biohydrogenation of unsaturated C18 fatty acid 367

(39). Although some species of Campylobacter such as Campylobacter concisus and other 368

non–Campylobacter jejuni Campylobacter species have been implicated in the initiation of 369

gastrointestinal diseases (40), they represent persistent residents of rumen microbial communities 370

(10). Some species of genus Succinivibrio could ferment maltose and glucose, thus participating 371

in the last stage of starch digestion (41). From 28 to 70 d, the abundance of genera Prevotella, 372

Butyrivibrio and Campylobacter still increased with age, implying they were dominant genera in 373

ruminal epithelial microbiota after solid feed provision. However, what role these genera played 374

and how they interacted with the host have not been exploited and remain to be elucidated. 375

376

Based on the already published data on rumen functional variables in the same kid goats, we 377

explored the relationship between ruminal epithelial microbiota and rumen functions. The 378

abundance of genera Butyrivibrio, Campylobacter, Desulfobulbus and Ruminococcus were 379

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

20

positively correlated with rumen weight, rumen papilla length, ammonia concentration and total 380

volatile fatty acid concentration, suggesting they might be involved in ammonia and volatile fatty 381

acid metabolism as well as anatomic development of the rumen. Similarly, genera Butyrivibrio, 382

Campylobacter, Desulfobulbus might also participate in fibrolytic enzyme secretion and genus 383

Fusobacterium is related to starch degradation. Furthermore, specific commensal microbes have 384

been reported to modulate host immune responses via pro-inflammatory and anti-inflammatory 385

pathways, as well as epithelial cell mediated signals (42, 43). Therefore, further investigation is 386

warranted to better explain the symbiotic relationship between ruminal epithelial microbiota and 387

its host. 388

389

Conclusion 390

391

In conclusion, ruminal epithelial bacterial communities were more diverse in the older groups 392

than those in the earlier groups and each age group had its distinct microbiota. Of the three 393

dominant phyla, Proteobacteria declined with age while Bacteroidetes and Firmicutes increased 394

with age. Genus Escherichia dominated at the day born, while Prevotella, Butyrivibrio and 395

Campylobacter surged in abundance at around two months. Furthermore, the correlations 396

between particular genus of the ruminal epithelial bacteria and anatomic, functional variables 397

might indicate a variety of different types of relationships with the host. This study provides 398

some preliminary information of the potential role of the ruminal epithelial bacteria, and further 399

investigations are required to determine their actual roles and interactions with the host. 400

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

21

401

Acknowledgements 402

403

We acknowledge the financial support received from the National Natural Science Foundation of 404

China (Grant No. 31320103917), "Strategic Priority Research Program - Climate Change: 405

Carbon Budget and Relevant Issues" (Grant No. XDA05020700), “CAS Visiting Professorship 406

for Senior International Scientists (Grant No. 2010T2S13, 2012T1S0009), and Hunan Provincial 407

Creation Development Project (2013TF3006). 408

409

410

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

22

411

References 412

413

1. Cheng K. J., R. P. Mccowan and J. W. Costerton. 1979. Adherent epithelial bacteria in ruminants 414

and their roles in digestive-tract function. Am J Clin Nutr. 32: 139-148. 415

2. Chen Y. H., M. Oba and L. L. Guan. 2012. Variation of bacterial communities and expression of 416

Toll-like receptor genes in the rumen of steers differing in susceptibility to subacute ruminal acidosis. 417

Vet. Microbiol. 159: 451-459. 418

3. Wang Y., S. Devkota, M. W. Musch, B. Jabri, C. Nagler, D. A. Antonopoulos, A. Chervonsky 419

and E. B. Chang. 2010. Regional mucosa-associated microbiota determine physiological expression 420

of TLR2 and TLR4 in murine colon. PloS one. 5: e13607. 421

4. Pei C. X., S. Y. Mao, Y. F. Cheng and W. Y. Zhu. 2010. Diversity, abundance and novel 16S rRNA 422

gene sequences of methanogens in rumen liquid, solid and epithelium fractions of Jinnan cattle. 423

Animal. 4: 20-29. 424

5. Cho S. J., K. M. Cho, E. C. Shin, W. J. Lim, S. Y. Hong, B. R. Choi, J. M. Kang, S. M. Lee, Y. H. 425

Kim, H. Kim and H. D. Yun. 2006. 16S rDNA analysis of bacterial diversity in three fractions of 426

cow rumen. J. Microbiol. Biotechnol. 16: 92-101. 427

6. Mead L. J. and G. A. Jones. 1981. Isolation and presumptive identification of adherent epithelial 428

bacteria ("epimural" bacteria) from the ovine rumen wall. Appl Environ Microbiol. 41: 1020-1028. 429

7. Petri R. M., T. Schwaiger, G. B. Penner, K. A. Beauchemin, R. J. Forster, J. J. McKinnon and T. 430

A. McAllister. 2013. Changes in the rumen epimural bacterial diversity of beef cattle as affected by 431

diet and induced ruminal acidosis. Appl. Environ. Microbiol. 79: 3744-3755. 432

8. de Menezes A. B., E. Lewis, M. O'Donovan, B. F. O'Neill, N. Clipson and E. M. Doyle. 2011. 433

Microbiome analysis of dairy cows fed pasture or total mixed ration diets. FEMS Microbiol. Ecol. 78: 434

256-265. 435

9. Chen Y., G. B. Penner, M. Li and M. Oba. 2011. Changes in bacterial diversity associated with 436

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

23

epithelial tissue in the beef cow rumen during the transition to a high-grain diet. Appl. Environ. 437

Microbiol. 77: 5770-5781. 438

10. Li R. W., E. E. Connor, C. Li, R. L. Baldwin and M. E. Sparks. 2012. Characterization of the 439

rumen microbiota of pre-ruminant calves using metagenomic tools. Environ Microbiol. 14: 129-139. 440

11. Jami E., A. Israel, A. Kotser and I. Mizrahi. 2013. Exploring the bovine rumen bacterial 441

community from birth to adulthood. Isme J. 7: 1069-1079. 442

12. Jiao J., X. Li, K. A. Beauchemin, Z. Tan, S. Tang and C. Zhou. 2015. Rumen development 443

process in goats as affected by supplemental feeding versus grazing: age-related anatomic 444

development, functional achievement and microbial colonization. Brit J Nutr.Accepted. 445

13. Hristov A. N., T. R. Callaway, C. Lee and S. E. Dowd. 2012. Rumen bacterial, archaeal, and fungal 446

diversity of dairy cows in response to ingestion of lauric or myristic acid. J Anim Sci. 90: 4449-4457. 447

14. Caporaso J. G., J. Kuczynski, J. Stombaugh, K. Bittinger, F. D. Bushman, E. K. Costello, N. 448

Fierer, A. G. Pena, J. K. Goodrich, J. I. Gordon, G. A. Huttley, S. T. Kelley, D. Knights, J. E. 449

Koenig, R. E. Ley, C. A. Lozupone, D. McDonald, B. D. Muegge, M. Pirrung, J. Reeder, J. R. 450

Sevinsky, P. J. Turnbaugh, W. A. Walters, J. Widmann, T. Yatsunenko, J. Zaneveld and R. 451

Knight. 2010. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 452

7: 335-336. 453

15. Schloss P. D., S. L. Westcott, T. Ryabin, J. R. Hall, M. Hartmann, E. B. Hollister, R. A. 454

Lesniewski, B. B. Oakley, D. H. Parks and C. J. Robinson. 2009. Introducing mothur: 455

open-source, platform-independent, community-supported software for describing and comparing 456

microbial communities. Appl. Environ. Microbiol. 75: 7537-7541. 457

16. Liu B., J. Yuan, S. M. Yiu, Z. Li, Y. Xie, Y. Chen, Y. Shi, H. Zhang, Y. Li, T. W. Lam and R. Luo. 458

2012. COPE: an accurate k-mer-based pair-end reads connection tool to facilitate genome assembly. 459

Bioinformatics. 28: 2870-2874. 460

17. Edgar R. C. 2013. UPARSE: highly accurate OTU sequences from microbial amplicon reads. Nat 461

Methods. 10: 996-998. 462

18. Edgar R. C., B. J. Haas, J. C. Clemente, C. Quince and R. Knight. 2011. UCHIME improves 463

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

24

sensitivity and speed of chimera detection. Bioinformatics. 27: 2194-2200. 464

19. DeSantis T. Z., P. Hugenholtz, N. Larsen, M. Rojas, E. L. Brodie, K. Keller, T. Huber, D. Dalevi, 465

P. Hu and G. L. Andersen. 2006. Greengenes, a chimera-checked 16S rRNA gene database and 466

workbench compatible with ARB. Appl Environ Microbiol. 72: 5069-5072. 467

20. Price M. N., P. S. Dehal and A. P. Arkin. 2009. FastTree: computing large minimum evolution trees 468

with profiles instead of a distance matrix. Mol. Biol. Evol. 26: 1641-1650. 469

21. Legendre P. and E. Gallagher. 2001. Ecologically meaningful transformations for ordination of 470

species data. Oecologia. 129: 271-280. 471

22. Denman S. E. and C. S. McSweeney. 2006. Development of a real‐time PCR assay for monitoring 472

anaerobic fungal and cellulolytic bacterial populations within the rumen. FEMS Microbiol. Ecol. 58: 473

572-582. 474

23. Layton A., L. McKay, D. Williams, V. Garrett, R. Gentry and G. Sayler. 2006. Development of 475

Bacteroides 16S rRNA gene TaqMan-based real-time PCR assays for estimation of total, human, and 476

bovine fecal pollution in water. Appl Environ Microbiol. 72: 4214-4224. 477

24. Stevenson D. M. and P. J. Weimer. 2007. Dominance of Prevotella and low abundance of classical 478

ruminal bacterial species in the bovine rumen revealed by relative quantification real-time PCR. Appl. 479

Microbiol. Biot. 75: 165-174. 480

25. Zhou M., E. Hernandez-Sanabria and L. L. Guan. 2009. Assessment of the microbial ecology of 481

ruminal methanogens in cattle with different feed efficiencies. Appl Environ Microbiol. 75: 482

6524-6533. 483

26. Rey M., F. Enjalbert, S. Combes, L. Cauquil, O. Bouchez and V. Monteils. 2014. Establishment 484

of ruminal bacterial community in dairy calves from birth to weaning is sequential. J. Appl. 485

Microbiol. 116: 245-257. 486

27. Jami E. and I. Mizrahi. 2012. Composition and similarity of bovine rumen microbiota across 487

individual animals. Plos One. 7:e33306. 488

28. Ley R. E., M. Hamady, C. Lozupone, P. J. Turnbaugh, R. R. Ramey, J. S. Bircher, M. L. 489

Schlegel, T. A. Tucker, M. D. Schrenzel, R. Knight and J. I. Gordon. 2008. Evolution of 490

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

25

mammals and their gut microbes. Science. 320: 1647-1651. 491

29. Mayer M., A. Abenthum, J. Matthes, D. Kleeberger, M. Ege, C. Hölzel, J. Bauer and K. 492

Schwaiger. 2012. Development and genetic influence of the rectal bacterial flora of newborn calves. 493

Vet Microbiol. 161: 179-185. 494

30. Malmuthuge N., M. J. Li, Y. H. Chen, P. Fries, P. J. Griebel, B. Baurhoo, X. Zhao and L. L. 495

Guan. 2012. Distinct commensal bacteria associated with ingesta and mucosal epithelium in the 496

gastrointestinal tracts of calves and chickens. FEMS Microbiol. Ecol. 79: 337-347. 497

31. Wallace R. J., K. J. Cheng, D. Dinsdale and E. R. Orskov. 1979. Independent microbial-flora of 498

the epithelium and its role in the eco-microibiology og the rumen. Nature. 279: 424-426. 499

32. Sadet S., C. Martin, B. Meunier and D. P. Morgavi. 2007. PCR-DGGE analysis reveals a distinct 500

diversity in the bacterial population attached to the rumen epithelium. Animal. 1: 939-944. 501

33. Mändar R. and M. Mikelsaar. 1996. Transmission of mother’s microflora to the newborn at birth. 502

Neonatology. 69: 30-35. 503

34. Wise G. H. and G. W. Anderson. 1939. Factors affecting the passage of liquids into the rumen of the 504

dairy calf. I. Method of administering liquids: drinking from open pail versus sucking through a 505

rubber nipple. J Dairy Sci. 22: 697 - 705. 506

35. Collier-Hyams L. S., V. Sloane, B. C. Batten and A. S. Neish. 2005. Cutting edge: bacterial 507

modulation of epithelial signaling via changes in neddylation of cullin-1. J. Immunol. 175: 508

4194-4198. 509

36. Kalita A., J. Hu and A. G. Torres. 2014. Recent advances in adherence and invasion of pathogenic 510

Escherichia coli. Curr Opin Infect Dis. 27: 459-464. 511

37. Jost T., C. Lacroix, C. P. Braegger and C. Chassard. 2012. New insights in gut microbiota 512

establishment in healthy breast fed neonates. PloS one. 7: e44595. 513

38. Zhu Z., S. Hang, S. Mao and W. Zhu. 2014. Diversity of Butyrivibrio group bacteria in the rumen 514

of goats and its response to the supplementation of garlic oil. Asian-Australas J Anim Sci. 27: 515

179-186. 516

39. Boeckaert C., B. Vlaeminck, V. Fievez, L. Maignien, J. Dijkstra and N. Boon. 2008. 517

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

26

Accumulation of trans C-18:1 fatty acids in the rumen after dietary algal supplementation is 518

associated with changes in the Butyrivibrio community. Appl. Environ. Microbiol. 74: 6923-6930. 519

40. Man S. M., N. O. Kaakoush, S. T. Leach, L. Nahidi, H. K. Lu, J. Norman, A. S. Day, L. Zhang 520

and H. M. Mitchell. 2010. Host attachment, invasion, and stimulation of proinflammatory cytokines 521

by Campylobacter concisus and other non-Campylobacter jejuni Campylobacter species. J Infect Dis. 522

202: 1855-1865. 523

41. Oherrin S. M. and W. R. Kenealy. 1993. Glucose and carbon-dioxide metabolism by 524

Succinivibrio-Dextrinosolvens. Appl. Environ. Microbiol. 59: 748-755. 525

42. Klaenhammer T. R., M. Kleerebezem, M. V. Kopp and M. Rescigno. 2012. The impact of 526

probiotics and prebiotics on the immune system. Nat. Rev. Immunol. 12: 728-734. 527

43. Thomas L. V. and T. Ockhuizen. 2012. New insights into the impact of the intestinal microbiota on 528

health and disease: a symposium report. Brit J Nutr. 107 Suppl 1: S1-13. 529

530

531

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

1

Table 1. Alpha diversity measures of the ruminal epithelial bacterial communities at different ages in kids (27000 reads)

Alpha indices Age (d)

SEM P value

0 7 28 42 70 L Q OTUs 272.3 279.7 430.3 527.5 453.0 73.24 0.032 0.123 Chao 437.1 466.9 566.9 726.9 640.2 75.51 0.025 0.166 Ace 581.1 609.3 565.2 790.7 661.2 69.81 0.210 0.436 Shannon 1.61 1.37 2.91 2.96 3.22 0.383 0.002 0.158 Simpson 0.37 0.56 0.20 0.17 0.10 0.065 0.001 0.326 Coverage 0.995 0.995 0.995 0.993 0.994 0.001 0.151 0.456

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

2

Table 2. Phylum level taxonomy composition (% of sequences) of the ruminal epithelial bacterial communities at different ages in kids (top 9 phyla)

Phylum Age (d)

SEM P value

0 7 28 42 70 L Q Actinobacteria 0.24 0.22 1.86 1.18 1.28 0.51 0.096 0.139 Bacteroidetes 1.57 3.31 12.80 8.61 10.57 3.77 0.088 0.227 Chloroflexi 0.00 0.01 0.80 0.22 0.42 0.11 0.013 0.014 Fibrobacteres 0.00 0.01 0.06 0.49 0.18 0.24 0.344 0.453 Firmicutes 1.94 7.06 16.99 54.71 40.29 4.84 <.0001 0.009 Fusobacteria 1.18 0.85 0.04 0.00 0.00 0.61 0.151 0.363 Proteobacteria 90.13 86.29 60.65 28.64 45.42 4.74 <.0001 0.0003 TM7 0.03 0.01 0.01 0.29 0.17 0.08 0.067 0.562 Thermi 0.23 1.21 1.13 1.31 0.00 0.49 0.511 0.051 Unclassified 4.63 0.81 0.74 4.06 1.12 1.73 0.565 0.891 Others (< 0.5%) 0.06 0.23 4.94 0.50 0.54 0.15 0.025 0.600

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

3

Table 3. Genus level taxonomy composition (% of sequences) of the ruminal epithelial bacterial communities at different ages in kids

Phylum Genus Age (d)

SEM P value

0 7 28 42 70 L Q Bacteroidetes Bacteroides 0.12 2.46 0.80 0.03 0.01 0.591 0.107 0.759 Porphyromonas 0.70 0.44 0.02 0.00 0.00 0.178 0.011 0.083 Prevotella 0.47 0.20 4.86 5.76 5.18 1.718 0.022 0.149 Firmicutes Bulleidia 0.00 0.01 0.24 0.05 0.07 0.093 0.576 0.240 Butyrivibrio 0.10 0.56 8.90 41.38 30.13 5.606 0.000 0.086 Lysinibacillus 0.00 0.00 0.55 0.00 0.00 0.213 0.947 0.197 Mogibacterium 0.00 0.01 0.42 0.17 0.51 0.102 0.003 0.648 Oscillospira 0.02 0.91 0.12 0.04 0.04 0.095 0.005 0.839 Ruminococcus 0.02 0.05 0.82 0.52 0.30 0.351 0.441 0.158 Streptococcus 0.38 0.12 0.35 0.00 0.00 0.117 0.056 0.964 Succiniclasticum 0.00 0.01 0.28 0.04 0.12 0.107 0.468 0.353 p-75-a5 0.00 0.00 0.02 0.11 0.82 0.182 0.006 0.077 Proteobacteria Acidovorax 0.21 0.16 0.14 0.30 0.00 0.106 0.343 0.286 Acinetobacter 0.08 0.05 0.11 0.72 0.01 0.322 0.752 0.255 Actinobacillus 0.49 0.01 0.01 0.00 0.00 0.219 0.259 0.332 Arcobacter 0.37 0.75 0.08 0.02 0.00 0.206 0.038 0.406 Burkholderia 0.00 0.01 0.26 0.00 0.00 0.084 0.901 0.129 Campylobacter 0.02 0.50 4.27 20.33 29.99 4.467 <.0001 0.560 Desulfobulbus 0.00 0.01 0.00 0.40 3.78 1.154 0.033 0.160 Enterobacter 0.27 0.01 0.02 0.02 0.00 0.071 0.077 0.178 Escherichia 80.79 7.48 1.65 0.23 0.06 3.578 <.0001 <.0001 Herbaspirillum 0.00 0.06 1.09 0.00 0.00 0.376 0.896 0.156 Janthinobacterium 0.00 0.01 0.33 0.00 0.00 0.110 0.902 0.141 Mannheimia 0.31 0.01 0.01 0.00 0.00 0.141 0.269 0.366 Paracoccus 0.00 0.00 0.13 0.21 0.00 0.089 0.631 0.075 Succinivibrio 0.00 0.04 3.27 0.01 0.00 1.142 0.938 0.156 Variovorax 0.00 0.01 0.26 0.00 0.00 0.087 0.915 0.151 Actinobacteria Actinomyces 0.03 0.01 0.03 0.00 0.21 0.078 0.116 0.184 Chloroflexi SHD-231 0.00 0.01 0.80 0.21 0.42 0.111 0.013 0.014 Fibrobacteres Fibrobacter 0.00 0.01 0.06 0.49 0.18 0.237 0.344 0.453 Fusobacteria Fusobacterium 1.18 0.85 0.04 0.00 0.00 0.612 0.151 0.363 Thermi Thermus 0.23 1.21 1.13 1.31 0.00 0.490 0.510 0.051 Unclassified 13.38 82.99 66.02 26.00 27.13 7.535 0.055 0.014 Others(<0.5%) 0.83 1.05 2.91 1.65 1.00 0.812 0.845 0.107

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

4

Table 4. qPCR results of total bacteria and two genera of the ruminal epithelial bacterial communities at different ages in kids

Genus Age (d) SEM P value

0 7 28 42 70 L Q Bacteriaa 1.37 1.43 3.72 3.09 4.40 0.838 0.013 0.510 Prevotella b 2.36 4.29 3.75 8.25 7.26 1.820 0.049 0.556 Bacteroides b 2.39 3.23 1.13 2.91 4.71 1.396 0.314 0.259

a, Total bacteria was expressed as ×108 copy numbers per gram rumen tissue. b, Prevotella and Bacteroides relative abundances (%) were expressed as by dividing the 16S rRNA gene copy number of each genus by the 16S rRNA gene copy number of total bacteria copy number.

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

5

Figure legends

Figure 1. Principal co-ordinate analysis (PCoA) of the ruminal epithelial bacterial OTUs at

different ages in kids.

Figure 2. Within-group similarity, calculated as the average of the pairwise similarity between

each paired sample within each group using the Bray-Curtis metric. The closer the similarity

is to 1, the higher the average similarity within a group. Letters above the bars indicate

significance between groups; groups with different letters have significant different similarity

values at P < 0.05.

Figure 3. Transform-based PCA of ruminal epithelial bacterial OTUs at different ages in kids.

The top ten OTUs contributing to the separation of samples are displayed on the figure (OTU

number).

OTU2, Neisseriaceae family. OTU108, Ruminococcaceae family. OTU5, Escherichia genus.

OTU3977, Escherichia genus. OTU120, Ruminococcus genus. OTU6, Butyrivibro genus.

OTU7, Campylobacter genus. OTU12, Butyrivibro genus. OTU113, Campylobacter genus.

OTU1340, Butyrivibro genus.

Figure 4. Correlation coefficients between relative abundances of the ruminal epithelial

bacterial genera, anatomic and functional variables.

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from

on June 29, 2018 by guesthttp://aem

.asm.org/

Dow

nloaded from