Letters in Applied Microbiology ISSN 0266-8254

ORIGINAL ARTICLE

Metabolomics analysis reveals large effect of roughage types on rumen microbial metabolic profile in dairy cows S. Zhao1,2, J. Zhao2,3, D. Bu2, P. Sun2, J. Wang2 and Z. Dong1 1 State Key Laboratory of Microbial Resources, Institute of Microbiology, Chinese Academy of Sciences, Beijing, China 2 State Key Laboratory of Animal Nutrition, Institute of Animal Sciences, Chinese Academy of Agricultural Sciences, Beijing, China 3 College of Animal Science and Technology, Inner Mongolia University for the Nationalities, Tongliao, Inner Mongolia, China

Significance and Impact of the Study: The microbial metabolites in the rumen provide nutritional precursors that are critical for general health and milk production in dairy cows. However, studies of the effect of diet on ruminal microbial metabolism are scant. In our current study, we analysed the ruminal microbial metabolite profile of cows fed different types of roughage. We found that the ruminal microbial metabolite profile of cows fed a mixed-roughage diet differed significantly from that of cows fed a single type of roughage. Certain metabolites, such as acetate, hydrocinnamate and methylamine, were closely correlated with specific types of roughage. Our findings provide insight into the effects of different roughages on ruminal microbial fermentation in dairy cows.

Keywords dairy cows, metabolomics, roughages, rumen fluid. Correspondence Zhiyang Dong, No.1 Beichen West Road, Chaoyang District, Beijing 100101, China. E-mail: [email protected]. and Jiaqi Wang, No.2 Yuanmingyuanxilu, Haidian District, Beijing 100193, China. E-mail: [email protected]. S. Zhao and J. Zhao contributed equally to the development of assays. 2013/2266: received 12 November 2013, revised 6 March 2014 and accepted 6 March 2014

Abstract The aim of our study was to determine the effect of diets with different types of roughage on the ruminal microbial metabolite profile in dairy cows. Holstein dairy cows were fed a diet containing either corn stover (CS group) or a mixture of alfalfa hay, Leymus chinensis hay and corn silage (MF group) at 0700 and 1900 h daily. Rumen fluid was sampled from each cow through a ruminal cannula at 0630 and 1030 h, and the mixed ruminal fluid from 3 day in each cow was analysed using nuclear magnetic resonance (NMR) spectroscopy. A multivariate analysis revealed a significant difference between the ruminal metabolome of the CS and MF groups at both time points. The MF group had higher levels of acetate, valerate, hydrocinnamate and methylamine and lower levels of glucose, glycine, propionate and isovalerate than those in the CS group. Our results showed that different types of roughages can significantly influence the ruminal microbial metabolome, especially with regard to organic acids, amines and amino acids.

doi:10.1111/lam.12247

Introduction The bovine rumen is inhabited by a complex community of micro-organisms, including bacteria, archaea, fungi and protozoa. These microbes directly mediate the digestion and fermentation of the contents of the rumen into dietary-fibre constituents, starches, peptides and lipids in a process known as rumen metabolism (Kim et al. 2011). Certain metabolites, such as monose, amino acids and fatty acids, have been shown to be the main metabolites in

ruminal fluid (Saleem et al. 2012, 2013). These metabolites are required for microbial growth. However, they are also critical to the general health of cows and are important for high-quality milk and meat production. Thus, an analysis of the ruminal metabolite profile might reveal how diet can affect feed utilization and animal productivity. Metabolomics is the study of various factors that affect overall metabolite composition of organs or tissues. Analytical chemistry methods can be used to characterize most low molecular weight metabolites and

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

79

S. Zhao et al.

Roughages cause shifts in rumen metabolic profile

may provide more accurate information regarding the physiological state of the microbiome or the host organism (Nicholson and Lindon 2008; Sundekilde et al. 2011). Nuclear magnetic resonance (NMR) spectroscopy is a powerful analytical technique for the identification of metabolic biomarkers (Wishart 2008; Viant et al. 2009; Chylla et al. 2011). Numerous NMR-based metabolomic studies have analysed urine (Wei et al. 2012; Cho et al. 2013), blood (Serkova et al. 2011; Verwaest et al. 2011) and plasma (Lanza et al. 2008) from humans or mice. Previous metabolomic studies of ruminants have primarily focused on the analysis of urine and milk from dairy cows (Lv and Li 2010; Bertram et al. 2011) or urine from sheep (Nyberg et al. 2010). Only few studies of the influence of diet transitions on the ruminal metabolome and the relationship between the ruminal metabolic profile and the incidence of metabolic disorders in dairy cows have been performed (Ametaj et al. 2010; Saleem et al. 2012, 2013). These studies showed that high-grain diets elevated the concentrations of certain toxic and inflammatory compounds in the ruminal fluid. Different types of roughage are used as forage for dairy cows in different countries based on locally available resources. The effects of forage types on microbial populations in the rumen and animal productivity have been well documented (Huws et al. 2010; Petri et al. 2012). One study showed that corn stover reduced milk production, compared with that of cows fed alfalfa hay (Zhu et al. 2013). However, the effects of forage types on the ruminal metabolome have not been examined in detail. In our current study, we examined the effects of roughage types on the ruminal metabolite profile in dairy cows using NMR spectroscopy.

(a) 3

Results and discussion The NMR spectral matrix was analysed using a PLS-DA statistics, and the PLS-DA score plots showed significant differences between the MF and CS groups at both two time points (Fig. 1a). In the loadings plot, most of the metabolites clustered together, but certain metabolites, including those at d 1
94, d 0
9, d 1
06 and d 3
42 ppm, underwent significant temporary changes in clustering (Fig. 1b). The hierarchical analysis and heat map also showed differences in the clustering of ruminal metabolites between the sampling time points (Fig. 2). Such differences indicated that the sampling times should be taken into account when comparative studies or analysis related to rumen fermentation were performed in vivo. The representative 1H NMR spectra of the ruminal fluid from the two roughage groups are shown in Fig. 3. Most of the metabolites appeared between d 0
8 and d 4
0 ppm, which is consistent with the findings of Saleem et al. (2012). However, Saleem et al. (2012) also reported metabolites between d 7
0 and d 8
5, which were absent in the spectra from our analysis. This inconsistency might have been caused by differences in the diets or experimental conditions used. The PLS-DA analysis showed that significant temporary differences occurred in the number of ruminal metabolites in the CS and MF groups. However, as shown in Table 1, only a limited number of the metabolites affected by diet could be identified using the currently available metabolite databases. The cows in MF had higher levels of acetate (d 1
94 and d 1
9 ppm), valerate (d 0
86 and d 1
3 ppm), hydrocinnamate (d 2
9 ppm), ethanol/organic acid (d 1
18 ppm) and

1·94

(b)

0·9 1·06

0·20

2·06

2 0·10

w*c[2]

t[2]

1 0

0·00

–1 –0·10 –2

2·14 4·42 2·38 2·34 2·42 1·46 0·94 4·54 4·5 4·14 4·46 4·1 1·38 0·86 4·3 4·34 4·38 7·38 1·5 6·7 4·18 7·3 2·78 1·42 4·22 3·34 8·221·58 8·18 4·26 6·78 7·18 2·82 3·06 6·66 0·82 1·740·98 6·74 0·78 7·26 7·34 6·62 4·02 2·66 0·74 7·9 2·02 2·98 0·7 4·06 8·94 6·82 0·66 1·02 7·14 8·62 7·541·54 7·1 0·62 0·46 2·7 0·58 0·54 7·86 2·1 0·42 0·5 7·42 2·5 6·86 0·38 8·26 1·78 0·3 7·06 1·86 9·34 0·34 8·02 9·26 0·26 1·7 8·14 9·02 1·9 0·22 8·74 8·46 9·1 8·06 9·3 9·38 8·98 8·78 8·66 8·7 8·5 8·42 9·14 8·54 8·58 8·38 9·18 8·34 7·5 7·98 8·1 8·3 6·54 3·14 7·62 6·58 3·98 7·02 7·7 6·9 7·66 7·58 7·74 2·26 6·98 6·26 1·66 6·94 7·78 6·22 6·42 6·3 6·5 1·82 6·34 6·38 2·54 3·02 6·46 7·22 7·46 1·22 2·58 7·82 1·1 1·26 7·94 1·62 1·98 2·46 1·3 3·1 3·18 1·34 3·94 3·54 2·3 3·38 3·22 2·94 4·58 3·3 2·74 3·58 3·82 2·86 1·18 1·14 2·22 3·66 3·62 3·9 3·78 3·86 2·9 2·18 3·26 3·46 2·62 3·5 3·74 3·42 3·7

–0·20

–3 –4

–3

–2

–1

0

t[1]

1

2

3

4

–0·60 –0·50 –0·40 –0·30 –0·20 –0·10 0·00 0·10 0·20 0·30

w*c[1]

Figure 1 Score plots (a) and loading plots (b) of the results of PLS-DA evaluation of the 1H NMR spectra of ruminal fluid from cows fed with corn stover (CS) at 0630 h (●) and 1030 h (○) or mixed forage (MF) at 0630 h (■) and 1030 h (□). Score plot (a) demonstrates the discriminate power of PLS-DA (for first two-component R2X (cumulative) = 60
6%, R2Y (cumulative) = 58
9%, Q2 (cumulative) = 51
4%). The metabolites are indicated as symbol (▲) and chemical shift in plot (b).

80

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

S. Zhao et al.

Roughages cause shifts in rumen metabolic profile

4

2

0

Figure 2 Hierarchical clustering analysis for the ruminal metabolites from dairy cows fed corn stover (CS) or mixed forage (MF). Each row represents one metabolite. Each column represents one sample. Cells are coloured based on the signal intensity measured from NMR. Dark brown represents high rumen levels, blue shows low signal intensity, and grey cells shows the intermediate level (see colour scale on the right of heat map). Treat: ( ) CS and ( ) MF; Time: ( ) 0630 h and ( ) 1030 h.

MF1030_11

MF1030_10

MF1030_9

MF1030_8

MF1030_7

MF1030_6

CS1030_5

CS1030_4

CS1030_3

CS1030_2

CS1030_1

MF0630_11

MF0630_10

MF0630_9

MF0630_8

MF0630_7

MF0630_6

CS0630_5

CS0630_4

CS0630_3

CS0630_2

CS0630_1

–2

MF1030 15

10

14

13 14

5

MF0630 8

9

12

9

7

CS1030 17 16 17

11

CS0630

1

4

6

2

4 3

8

7

6 4

3

2

1

0

Chemical shift (ppm) Figure 3 Representative 1H NMR spectra of ruminal fluid from dairy cows fed corn stover (CS) or mixed forage (MF). The spectra were constructed using TSP at d 0
00 ppm as a reference. Spectral key: 1, acetate; 2, butyrate; 3, butyrate/alanine (overlapped); 4, butyrate/propionate (overlapped); 5, ethanol; 6, glucose (large carbohydrate region); 7, glutamine/glutamate (overlapped); 8, glycine; 9, hydrocinnamate; 10, hydrocinnamate/dimethyl glycine (overlapped); 11, isovalerate; 12, lysine; 13, methylamine; 14, propionate; 15, trimethylamine; 16, valine/isoleucine (overlapped); and 17, valerate. Unidentified metabolites are not shown. CS0630 and MF0630 indicate samples from CS and MF cows collected at 0630 h, respectively, and CS1030 and MF1030 indicate samples from CS and MF cows collected at 1030 h, respectively.

methylamine (d 2
62 ppm) than those in the CS group, whereas the levels of butyrate (d 1
54 and d 1
58 ppm), glucose (d 3 to d 4 ppm), glycine (d 3
54 ppm), propio-

nate (d 1
06 ppm) and isovalerate (d 2
06 ppm) were lower in the MF cows. The glucose level in the CS group was higher than that in the MF group, which might have

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

81

S. Zhao et al.

Roughages cause shifts in rumen metabolic profile

Table 1 The identified metabolites related to types of roughage and feeding Compared with Chemical shift (ppm) 0
86, 2
18 1
18 3–4 1
90, 2
06 0
90, 1
06 1
54, 2
50, 2
62 3
54 2
38, 1
50 2
94 2
86

1
30

1
94 2
14 1
58 2
90, 7
30, 7
38

2
42

3
02 0
94, 0
98, 1
02 2
22

CS0630*

CS1030*

CS0630

MF0630

Identified metabolites

MF0630*

MF1030*

CS1030

MF1030

Valerate Butyrate/propionate Ethanol Glucose Acetate Isovalerate Butyrate/propionate Propionate Butyrate Hydrocinnamate Methylamine Glycine Glutamine/glutamate Butyrate/alanine Trimethylamine Hydrocinnamate/ dimethyl glycine Lysine Valine/isoleucine Propionate

↑† ↓† ↑ –† ↑ ↓ ↓ ↓ ↓ ↓ – ↓ – – – –

↑ ↓ ↑ ↓ ↑ – ↓ ↓ ↓ ↑ – – – ↓ ↑ ↑

↑ ↑ ↑ ↑ ↓ ↓ ↓ ↓ ↓ ↓ ↑ ↑ – – – ↓

↑ ↑ ↑ ↑ ↓ ↓ ↓ ↓ ↓ ↑ ↑ ↓ ↓ ↓ ↑ ↑

– – –

– ↓ –

– – –

↑ – ↑

*CS0630 and MF0630 indicate samples from CS and MF cows collected at 0630 h, respectively, and CS1030 and MF1030 indicate samples from CS and MF cows collected at 1030 h, respectively. †↑/↓ indicates relative increase/decrease significantly (P < 0
05), – indicates no significant change (P > 0
05).

been caused by a higher level of starch or nonfibre carbohydrate in the CS diet. Ametaj et al. (2010) found that glucose increased in the rumen of cows fed large amounts of barley grain (Ametaj et al. 2010) with abundant starch. Glucose is the major monosaccharide liberated during the degradation of starch, and it can be converted into a number of different polyols and amino acids via the glycolytic pathway and its branches (Michaud and Denlinger 2007). The levels of the ruminal fluid metabolites also changed after the feeding. The levels of valerate, ethanol/organic acid, glucose and methylamine were greater in both groups at 1030 h than at 0630 h, whereas the levels of acetate, propionate and butyrate were lower in both groups at 1030 h. The results from gas chromatography also showed that the concentrations of acetate, propionate and butyrate decreased (P < 0
05) by 10
18, 7
01 and 16
87% after feeding in MF group, respectively (Fig. S1). And the concentration of acetate, propionate and butyrate decreased (P < 0
05) by 9
40, 8
29 and 14
54% after feeding in CS group, respectively. The lower level of acetate, propionate and butyrate after feeding may have been due to ad libitum feed intake by the cows before the morning feed. In addition, the low pH (5
6) of rumen fluid at 82

1030 h may have resulted in subacute ruminal acidosis and disorder of ruminal microbial fermentation, resulting in less SCFA production. The level of hydrocinnamate was higher at 0630 h than at 1030 h in the MF group, whereas the level of glycine was lower at 0630 h in the MF group. Hydrocinnamate is involved in the metabolism of butanoate (Saleem et al. 2012), and glycine is involved in energy production and the synthesis of nucleotides, phospholipids and collagen (Cao et al. 2012). How roughage feeding affects the production of hydrocinnamate and glycine is unclear at present. Although the level of Methylamine did not differ significantly between the two groups, it increased after feeding significantly. This might have been caused by the degradation of glycine, tyrosine, phenylalanine, lipids, cholines or other dietary components (Hill and Mangan 1964). However, the level of trimethylamine was higher in the MF group at 1030 h than in the CS group, whereas no significant difference in the level of trimethylamine was observed between the two groups at 0630 h. Methylamine in the rumen can be converted to toxic metabolites, such as hydrogen peroxide and formaldehyde (Chalmers et al. 2006; Yu et al. 2006; Ametaj et al. 2010; Saleem et al. 2012). However,

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

S. Zhao et al.

trimethylamine and other methylamines are substrates for methylotrophic methane production and can be rapidly converted to methane. Thus, these results are consistent with the findings of previous studies which reported increased methane production after feeding (Hill and Mangan 1964). In summary, our 1H NMR spectroscopy and multivariate analysis identified changes in the ruminal metabolite profile that correlated with feeding upon different types of roughage. Our findings provide valuable insight into the influence of roughage types on ruminal microbial fermentation. Future studies of diet-related changes in ruminal metabolites would benefit from the expansion of metabolite databases. Material and methods Twelve rumen-cannulated, mid-lactating primiparous Holstein dairy cows (mean  SD; days in milk = 100  10 day; body weight = 550  50 kg) were randomly assigned to the corn stover (CS) or mixed forage (MF) groups. The CS group (n = 6) was fed corn stover only, and the MF group (n = 6) was fed roughage consisting of 48% alfalfa hay, 46% corn silage and 6% Leymus chinensis hay. Each type of roughage was mixed with corn and soybean concentrate to make a total mixed ration (TMR) that was adequate for the nutritional requirements of a 550 kg lactating cow, according to the Nutrient Requirements of Dairy Cattle (2001) guidelines. The dried matter (DM) in the two TMR diets contained the same proportion of crude protein (17
4  1
9%), but contained different proportions of neutral detergent fibre (MF, 39
6  2
1%; CS, 35
6  1
8%), nonfibre carbohydrate (MF, 19
1  1
2%; CS, 22
3  1
5%) and starch (MF, 13
0  0
9%, CS, 15
2  1
1%). The cows were fed at 0700 and 1900 h for ad libitum intake and milked at 0500 and 1700 h daily. The cows were cared following the standard procedures approved by Animal Ethics Committee in Institute of Animal Science (Permit Number: RNL1008). Ruminal fluid was sampled from the 11 rumencannulated cows at 0630 and 1030 h on days 81, 82 and 83. The ruminal fluid of one cow in the CS group was not sampled because of mastitis. The ruminal fluid from the same cow in 3 days was strained through four layers of cheesecloth and mixed. The ruminal fluid samples were first centrifuged at 13 000 g for 15 min to remove particulate matter and sterilized by passing through a 0
22 lm syringe filter. A 800-ll aliquot of each sample was used for detection of short-chain fatty acids (SCFAs) concentration by an Agilent gas chromatography model 7890A (Agilent Technologies, Palo Alto, CA). Another 300-ll aliquot of each sample was com-

Roughages cause shifts in rumen metabolic profile

bined with 50 ll of deuterated water (99
9% purity, Cambridge Isotope Laboratories, Tewksbury, MA) containing 1 mg ml 1 sodium trimethylsilyl-[2,2,3,3-2H4] -1-propionate (TSP; Merck, Whitehouse Station, NJ) as an internal chemical shift reference. The mixture was added to 300 ll of phosphate buffer (200 mmol l 1, pH 7
4) and centrifuged at 13 000 g for 10 min. The supernatant was transferred to a 5 mm NMR tube, and the spectra were measured at 600 MHz using a Varian INOVA 1H NMR spectrometer (Agilent Technologies). A 1D nuclear Overhauser enhancement pulse sequence was used for each sample, and solvent signal suppression was achieved by presaturation during the relaxation and mixing steps of the procedure. The spectra were collected for each sample using 32 K data points with a sweep width of 8000 Hz, a mixing time of 0
1 s, a relaxation delay of 2
0 s and 64 transients per increment. The free induction decay was zero-filled to 128 K, and an exponential line-broadening function of 0
5 Hz was applied before the spectra were Fourier transformed. The proton NMR spectra were phased, and the baseline was corrected manually. The area for each segmented region and the integral values were used to calculate the intensity distribution of the entire spectrum. The spectral matrix was further reduced in size by integrating the 0
04-ppm-wide bins in the range of d 0
4–d 9
4 ppm, with the exclusion of residual water and urea signals between d 4
6 and d 6
2 ppm. All of the imported data were normalized to a constant sum. The data were Pareto scaled and tested by orthogonal signal correction, before conversion to the EXCEL format (Microsoft, Redmond, WA). The data sets were a, imported into the SIMCA-P program (ver. 10.04, Ume Sweden) for the multivariate statistical analysis. A partial least-squares discriminant analysis (PLS-DA) was used to maximize the separation between the data for the roughage groups and sampling time points. The t-test was used for statistical comparisons of metabolites between groups. A hierarchical clustering analysis was performed using the METABOANALYST (Xia et al. 2012) program to explore the clustering patterns among the various ruminal metabolites. The concentrations of SCFAs determined by gas chromatography were analysed using the PROC MIXED procedures of the SAS system (ver. 9.0; SAS Institute Inc., Cary, NC). P < 0
05 was considered statistically significant. Acknowledgements This work was supported by the National Basic Research Program of China (2011CB100804) and China Postdoctoral Science Foundation (2013M540152).

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

83

Roughages cause shifts in rumen metabolic profile

Conflict of Interest No conflict of interest declared. References Ametaj, B.N., Zebeli, Q., Saleem, F., Psychogios, N., Lewis, M.J., Dunn, S.M., Xia, J.G. and Wishart, D.S. (2010) Metabolomics reveals unhealthy alterations in rumen metabolism with increased proportion of cereal grain in the diet of dairy cows. Metabolomics 6, 583–594. Bertram, H.C., Yde, C.C., Zhang, X.M. and Kristensen, N.B. (2011) Effect of dietary nitrogen content on the urine metabolite profile of dairy cows assessed by nuclear magnetic resonance (NMR)-based metabolomics. J Agric Food Chem 59, 12499–12505. Cao, M., Zhao, L.C., Chen, H.G., Xue, W. and Lin, D.H. (2012) NMR-based metabolomic analysis of human bladder cancer. Anal Sci 28, 451–456. Chalmers, R.A., Bain, M.D., Michelakakis, H., Zschocke, J. and Iles, R.A. (2006) Diagnosis and management of trimethylaminuria (FMO3 deficiency) in children. J Inherit Metab Dis 29, 162–172. Cho, K., Cho, S., Chung, J., Yoon, S., Jang, I. and Cho, J. (2013) Global metabolomics profiling of human urine reveals change in endogenous metabolites after metformin and pioglitazone administration. Clin Pharmacol Ther 93, S15. Chylla, R.A., Hu, K.F., Effinger, J.J. and Markley, J.L. (2011) Deconvolution of two-dimensional nmr spectra by fast maximum likelihood reconstruction: application to quantitative metabolomics. Anal Chem 83, 4871–4880. Hill, K.J. and Mangan, J.L. (1964) The formation and distribution of methylamine in the ruminant digestive tract. Biochem J 93, 39–45. Huws, S.A., Lee, M.R.F., Muetzel, S.M., Scott, M.B., Wallace, R.J. and Scollan, N.D. (2010) Forage type and fish oil cause shifts in rumen bacterial diversity. FEMS Microbiol Ecol 73, 396–407. Kim, M., Morrison, M. and Yu, Z.T. (2011) Status of the phylogenetic diversity census of ruminal microbiomes. FEMS Microbiol Ecol 76, 49–63. Lanza, I.R., Zhang, S., Karakelides, H., Raftery, D. and Nair, S. (2008) Quantitative H-1 NMR-based metabolomics of plasma and urine in type 1 diabetic humans: the effects of insulin deprivation. Diabetes 57, A698–A699. Lv, Y. and Li, Q.Z. (2010) Early stage diagnosis of mastitis of dairy cows using H-1 NMR-based metabolomics. J Dairy Sci 93, 525. Michaud, M.R. and Denlinger, D.L. (2007) Shifts in the carbohydrate, polyol, and amino acid pools during rapid cold-hardening and diapause-associated coldhardening in flesh flies (Sarcophaga crassipalpis): a metabolomic comparison. J Comp Physiol B 177, 753– 763.

84

S. Zhao et al.

National Research Council. (2001) Nutrient Requirements of Dairy Cattle: Seventh Revised Edition. Washington, DC: The National Academies Press. Nicholson, J.K. and Lindon, J.C. (2008) Systems biology metabonomics. Nature 455, 1054–1056. Nyberg, N.T., Nielsen, M.O. and Jaroszewski, J.W. (2010) Metabolic trajectories based on H-1 NMR spectra of urines from sheep exposed to nutritional challenges during prenatal and early postnatal life. Metabolomics 6, 489–496. Petri, R.M., Forster, R.J., Yang, W., McKinnon, J.J. and McAllister, T.A. (2012) Characterization of rumen bacterial diversity and fermentation parameters in concentrate fed cattle with and without forage. J App Microbiol 112, 1152–1162. Saleem, F., Ametaj, B.N., Bouatra, S., Mandal, R., Zebeli, Q., Dunn, S.M. and Wishart, D.S. (2012) A metabolomics approach to uncover the effects of grain diets on rumen health in dairy cows. J Dairy Sci 95, 6606–6623. Saleem, F., Bouatra, S., Guo, A.C., Psychogios, N., Mandal, R., Dunn, S.M., Ametaj, B.N. and Wishart, D.S. (2013) The bovine ruminal fluid metabolome. Metabolomics 9, 360– 378. Serkova, N.J., Standiford, T.J. and Stringer, K.A. (2011) The emerging field of quantitative blood metabolomics for biomarker discovery in critical illnesses. Am J Resp Crit Care 184, 647–655. Sundekilde, U.K., Frederiksen, P.D., Clausen, M.R., Larsen, L.B. and Bertram, H.C. (2011) Relationship between the metabolite profile and technological properties of bovine milk from two dairy breeds elucidated by NMR-based metabolomics. J Agric Food Chem 59, 7360–7367. Verwaest, K.A., Vu, T.N., Laukens, K., Clemens, L.E., Nguyen, H.P., Van Gasse, B., Martins, J.C., Van der Linden, A. et al. (2011) H-1 NMR based metabolomics of CSF and blood serum: a metabolic profile for a transgenic rat model of Huntington disease. Bba-Mol Basis Dis 1812, 1371–1379. Viant, M.R., Bearden, D.W., Bundy, J.G., Burton, I.W., Collette, T.W., Ekman, D.R., Ezernieks, V., Karakach, T.K. et al. (2009) International NMR-based environmental metabolomics intercomparison exercise. Environ Sci Technol 43, 219–225. Wei, H., Pasman, W., Rubingh, C., Wopereis, S., Tienstra, M., Schroen, J., Wang, M., Verheij, E. et al. (2012) Urine metabolomics combined with the personalized diagnosis guided by Chinese medicine reveals subtypes of prediabetes. Mol BioSyst 8, 1482–1491. Wishart, D.S. (2008) Metabolomics: applications to food science and nutrition research. Trends Food Sci Technol 19, 482–493. Xia, J.G., Mandal, R., Sinelnikov, I.V., Broadhurst, D. and Wishart, D.S. (2012) MetaboAnalyst 2.0-a comprehensive server for metabolomic data analysis. Nucleic Acids Res 40, W127–W133. Yu, P.H., Lu, L.X., Fan, H., Kazachkov, M., Jiang, Z.J., Jalkanen, S. and Stolen, C. (2006) Involvement of

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

S. Zhao et al.

semicarbazide-sensitive amine oxidase-mediated deamination in lipopolysaccharide-induced pulmonary inflammation. Am J Pathol 168, 718–726. Zhu, W., Fu, Y., Wang, B., Wang, C., Ye, J.A., Wu, Y.M. and Liu, J.X. (2013) Effects of dietary forage sources on rumen microbial protein synthesis and milk performance in early lactating dairy cows. J Dairy Sci 96, 1727–1734.

Roughages cause shifts in rumen metabolic profile

Supporting Information Additional Supporting Information may be found in the online version of this article: Figure S1 Concentration of rumen fluid SCFAs measured by gas chromatography.

Letters in Applied Microbiology 59, 79--85 © 2014 The Society for Applied Microbiology

85