Untargeted Metabolomic Profiling of Amphenicol-Resistant

The gradient conditions were optimized as follows: 0–1 min, 2–5%B; 1–8 min, ...... 2013, 76, 691– 693 .... S. A novel fexA variant from a cani...
0 downloads 0 Views 4MB Size
Download PDF
Article pubs.acs.org/jpr

Untargeted Metabolomic Profiling of Amphenicol-Resistant Campylobacter jejuni by Ultra-High-Performance Liquid Chromatography−Mass Spectrometry Hui Li,† Xi Xia,† Xiaowei Li, Gaowa Naren, Qin Fu, Yang Wang, Congming Wu, Shuangyang Ding, Suxia Zhang, Haiyang Jiang, Jiancheng Li, and Jianzhong Shen* Department of Veterinary Pharmacology and Toxicology, College of Veterinary Medicine, China Agricultural University, Beijing 100193, P. R. China S Supporting Information *

ABSTRACT: Campylobacter jejuni, an important foodborne microorganism, poses severe and emergent threats to human health as antibiotic resistance becomes increasingly prevalent. The mechanisms of drug resistance are hard to decipher, and little is known at the metabolic level. Here we apply metabolomic profiling to discover metabolic changes associated with amphenicol (chloramphenicol and florfenicol) resistance mutations of Campylobacter jejuni. An optimized sample preparation method was combined with ultra-high-performance liquid chromatography−time-of-flight mass spectrometry (UHPLC−TOF/MS) and pattern recognition for the analysis of small-molecule biomarkers of drug resistance. UHPLC−triple quadrupole MS operated in multiple reaction monitoring mode was used for quantitative analysis of metabolic features from UHPLC−TOF/ MS profiling. Up to 41 differential metabolites involved in glycerophospholipid metabolism, sphingolipid metabolism, and fatty acid metabolism were observed in a chloramphenicol-resistant mutant strain of Campylobacter jejuni. A panel of 40 features was identified in florfenicol-resistant mutants, demonstrating changes in glycerophospholipid metabolism, sphingolipid metabolism, and tryptophan metabolism. This study shows that the UHPLC−MS-based metabolomics platform is a promising and valuable tool to generate new insights into the drug-resistant mechanism of Campylobacter jejuni. KEYWORDS: metabolomic, Campylobacter jejuni, antibiotic resistance, UHPLC−MS, glycerophospholipid metabolism

■

Campylobacter was 37.5% in Brazil.10 Facing such serious rates of antibiotic resistance, it is urgent and necessary to investigate the resistance mechanism so as to circumvent this problem. Previous studies reported that the resistance mechanism of bacteria to amphenicols included acetyltransferases inactivation by drug acetylation;11 specific efflux pumps such as cmlA,12 f loR,13 f loRv,14 fexA,15 fexAv,16 and fexB;17 non-specific multidrug efflux pump systems; and methylases synthesized by the cfr gene that affect the binding of amphencol to the transpeptidase core of the ribosome.18 In our laboratory, C. jejuni strains highly resistant to CAP and FFC were obtained by in vitro selection, and a point mutation (G2073A) in the 23S rRNA from amphenicol-resistant mutants was identified.19 Although gene characterization has been profiled in the mutant and parent strains, little is known about the global metabolite alterations associated with resistant mutation. Determining the metabolome changes in mutant strains of C. jejuni may lead to further understanding of the antibiotic resistance mechanisms. Methodology for both the quenching of metabolism and extraction of metabolites is still a controversial area and requires

INTRODUCTION Campylobacter jejuni is a major cause of bacterial gastroenteritis in human.1 In 2012, the number of confirmed cases of campylobacteriosis in the European Union amounted to 214 268.2 In the United States, it is estimated that 2.4 million cases of Campylobacter infections occur each year.3,4 Although most C. jejuni infections are clinically mild, self-limiting, and usually resolved in a few days without antibiotic treatment, severe or prolonged post-infection complications including the peripheral neuropathies Guillain−Barre and Miller−Fisher syndromes can also occur. 5,6 In these circumstances, therapeutic intervention is usually warranted.7 However, C. jejuni is increasingly resistant to clinically important antibiotics, and this rising resistance is a concern for public health. The amphenicol family of antibiotics includes chloramphenicol (CAP), florfenicol (FFC), and thiamphenicol. Amphenicols are promising broad-spectrum antibiotics that are highly effective against a wide variety of Gram-positive and Gramnegative bacteria. CAP and FFC bind to the bacterial ribosome and inhibit peptide bond formation by interaction with the ribosomal peptidyl transferase center.8 In China, an epidemiological survey showed that the CAP and FFC resistance rates of C. jejuni from chicken reached up to 24.5 and 61.7%, respectively,9 and the detection rate of CAP-resistant © 2014 American Chemical Society

Received: October 12, 2014 Published: December 10, 2014 1060

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

with 25 mL of quenching solvent 60% MeOH. The quenched samples were shaken gently and centrifuged for 10 min at 12 298g, −10 °C (Eppendorf 5810R, Hamburg, Germany). The supernatants were discarded, and the cell pellet was washed three times each with 10 mL of 0.85% NaCl (0 °C). Two quenched samples were merged as one prepared sample and centrifuged for 5 min at 12 298g, −10 °C. The prepared samples were stored at −80 °C until further analysis. The temperature of the samples was kept below −10 °C during the whole procedure. For sampling with cold liquid nitrogen quenching, 20 mL of cell suspension was added to a precooled centrifuge tube (−80 °C) and centrifuged at 12 298g, −10 °C for 10 min. The cell pellet was washed three times with 10 mL of 0.85% NaCl (0 °C) and centrifuged at 12 298g, −10 °C for 5 min. Then, the sample was frozen in liquid nitrogen to quench cell metabolism.

significant optimization for individual organisms to yield reproducible data.20 Several methods have been proposed for sampling microbial cultures,21−23 with the most important consideration for this process being the ability to instantaneously quench cell metabolism. However, very few have been focused on the extraction effectiveness to see biologically relevant metabolite changes within a nontargeted metabolomic analysis. In this study, we performed comprehensive cellular metabolomic analysis of mutant and parent strains of C. jejuni to investigate the metabolic consequences of amphenicolresistant mutations by in vitro selection. Different sampling, quenching, and extraction methods were optimized for metabolomic profiling of C. jejuni. Our findings reveal important metabolic regulatory events associated with amphenicol resistance mutations in C. jejuni.

■

Metabolite Extraction

MATERIALS AND METHODS

The cell pellet was resuspended in 500 μL of three different extractants (100% cold MeOH, MeOH/CHL (2:1, v/v), and ACN/H2O (50:50 v/v, containing 0.1% FA, −40 °C), respectively. After vigorously mixing, the suspension was flash frozen in liquid nitrogen and thawed on ice. This freezethawing operation was repeated three times. Then., the sample was centrifuged at 16 099g, −20 °C for 10 min. The supernatant was subsequently transferred to a new Eppendorf tube, and the pellet was extracted again following the procedure previously mentioned. Mixed standard solution (sulfadiazine, difloxacin, monensin, tylosin, and thiamphenicol) at a final concentration of 200 ng/mL was spiked into the extracts for quality control (QC) purposes. The extract was dried by a vacuum rotation evaporator (SpeedVac, Thermo Fisher, Japan) and dissolved in 500 μL of ACN/H2O (50:50, v/v, containing 0.1% FA) by vortexing for 2 min. The solution was centrifuged at 18 317g, 4 °C for 15 min; 10 μL of supernatant was injected into the UHPLC−MS system.

Chemicals and Reagents

HPLC-grade acetonitrile (ACN) and methanol (MeOH) were purchased from Fisher Chemicals (Pittsburgh, PA). HPLCgrade formic acid (FA) and ammonium acetate were purchased from Dikma Technologies (Lake Forest, CA). HPLC-grade ethanol (EtOH) was supplied by J & K Chemical (Petbasheerabad, Andhra Pradesh, India). HPLC water was obtained using a Milli-Q Plus water purification system (Millipore, Bedford, MA). Chloroform (CHL) and sodium chloride were purchased from Beijing Chemical (Beijing, China). Leucine encephalin was supplied by Sigma-Aldrich (St. Louis, MO). HEPES was provided by AMRESCO (Solon, OH). Cell Culture

Antibiotic-resistant mutants of C. jejuni were generated by independent stepwise in vitro selection experiments with CAP or FFC. The minimum inhibition concentrations (MICs) of generated CAP- and FFC-resistant mutants were 256 and 512 μg/mL, respectively. Genetic analysis of 23S rRNA and ribosomal mutations was performed by polymerase chain reaction (PCR) amplification and resequencing of the corresponding genes according to our previously described method.19 The target mutations identified in CAP- and FFCresistant mutants are shown in Table 1. Liquid cultures of

Untargeted Metabolomic Analysis

Samples were analyzed using a Waters Acquity ultra-highperformance liquid chromatography (UHPLC) apparatus coupled to a hybrid TOF−MS SYNAPT HDMS (G1) (Waters, Manchester, U.K.). Metabolites were separated by a Waters BEH Shield RP C18 column (50 mm × 2.1 mm, 1.7 μm) at 30 °C. Mobile phases were composed of phase A (0.1% FA in water) and phase B (0.1% FA in ACN). The gradient conditions were optimized as follows: 0−1 min, 2−5%B; 1−8 min, 5−90%B; 8−10 min, 90−100%B; 10−12 min, 100−2%B. The flow rate was 0.3 mL/min. The MS system was operated in ESI+ and ESI− modes, respectively. The mass range was set at m/z 100−1000 Da in the full-scan mode. The optimized ESI parameters were as follows: capillary voltage, 3.0 kV; cone voltage, 35 V; source temperature, 100 °C; desolvation temperature, 350 °C; desolvation gas flow, 600 L/h. For accurate mass measurement, leucine enkephalin was used as the lock spray standard ([M+H]+ = 556.2771; [M−H]− = 554.2615) at a concentration of 100 ng/mL under a flow rate of 50 μL/min. UHPLC−TOF/MS provides mass accuracy better than 5 ppm (with lock mass calibration) and mass spectral resolution of 10 000 fwhm in V mode.

Table 1. MIC and Target Mutations Identified in CAP- and FFC-Resistant Mutants MIC (μg/mL) strains

CAP

FFC

mutation in the 23S rRNA gene

change in protein L4

ATCC 33560 CAP-resistant mutant FFC-resistant mutant

4 256 256

2 128 512

no mutation G2073A G2073A

no mutation G74D no mutation

parent and resistant strains were grown in brain heart infusion (BHI) broth (Beijing Land Bridge Technology, Beijing, China) at 42 °C under microaerobic conditions (5% O2, 5% CO2, 90% N2) with agitation at 150 rpm for 24 h to a calculated OD600 of 0.5.

UHPLC−Triple Quadrupole MS Quantitative Analysis

Quenching of C. jejuni Cell Cultures

To verify the reliability of UHPLC−TOF/MS analysis, relative quantitative analysis was further carried out to investigate the fold change of the metabolite candidates using UHPLC−triple

For sampling with cold centrifugation quenching, 10 mL of cell suspension was sucked into precooled centrifuge tubes filled 1061

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

Table 2. Putative Identities of Differential Metabolites in CAP-Resistant Mutants fold change (n = 5) ESI+

ID

RT (min)

upregulated

1

5.72

downregulated

[M+H]+ (Da) 281.1540 b

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

5.52 6.44 6.78 8.99 8.89 8.91 9.02 9.48 9.59 9.92 3.11 4.42 5.83 4.47 8.07 8.63

356.3512 438.2970 577.1320b 622.4400 660.4605b 666.3631 686.4681b 688.4921 714.5117b 732.5556 265.2512b 274.2710 311.2720 318.2990b 337.1654b 339.1957

18 19 20

8.89 9.01 8.63

353.1960b 353.2108b 380.2201b

21 22

6.27 9.01

393.2497b 394.2352b

23 24

9.01 4.50

401.3510b 406.3521b

25 26 27 28 29 30 31 32 33

5.30 6.15 6.41 6.66 6.26 6.04 6.45 6.92 6.50

424.2523b 466.2920b 480.3140b 494.3291b 496.3404b 520.3405b 522.3581b 524.3731b 530.3257b

[M−H]− (Da)

ESI

ID

RT (min)

upregulated

34

8.12

339.2177b

downregulated

35

8.89

395.1853b

36 37 38 39 40 41

3.27 7.83 5.93 5.80 6.25 9.10

396.0501 431.1376 438.2615 450.2592b 480.3088 646.4413

−

a

SIMCA-P

MRM transitions

putative identification

mean

CVa

value

parent ion

(4E)-7-(4-hydroxyphenyl)-1-phenyl-4-hepten-3one arachidoyl ethanolamide GPEtn(P-16:0/0:0) proanthocyanidin A2 GPEtn(14:0/13:0)[U] GPEtn(18:2(9Z,12Z)/12:0) amphibine B GPEtn(14:0/18:3(9Z,12Z,15Z)) GPEtn(16:1(9Z)/16:1(9Z))[U] GPEtn(16:0/18:3(9Z,12Z,15Z)) PC(16:0/16:1(9Z)) 9,12-octadecadienal hexadecasphinganine 13(S)-HODE methyl ester phytosphingosine 1-(2-butoxy-2-oxoethyl)-2-butyl phthalate 3,6-dimethoxy-19-norpregna-1,3,5,7,9-pentaen20-one 9,10-dichloro-octadecanoic acid apo-8′-bixinal 3-{(E)-2-[4-(benzyloxy)-3-methoxyphenyl] vinyl}-5-pentyl-4,5-dihydro-1,2-oxazole CPA(16:0/0:0) 4-ethyl-3-(4-methoxyphenyl)-2,2-dimethyl-8-(1pyrrolidinylmethyl)-2H-chromen-7-ol L-palmitoylcarnitine (±)N-(2-fluoro-ethyl)-2,16-dimethyl5Z,8Z,11Z,14Z-docosatetraenoyl amine GPEtn(14:1(9Z)/0:0) PC(14:1(9Z)/0:0) GPEtn(18:1(9Z)/0:0) PC(16:1(9Z)/0:0) PC(16:0/0:0)[U] PC(18:2/0:0) PC(18:1(9Z)/0:0) PC(18:0/0:0) GPEtn(22:4/0:0)

1.70

0.34

0.9909

280.9

106.7, 134.8

20

1.88 1.72 85.22 1.75 2.61 1.94 4.33 3.30 5.47 3.68 3.96 1.86 2.07 1.85 1.72 2.70

0.28 0.08 0.04 0.33 0.31 0.37 0.07 0.07 0.18 0.12 0.17 0.28 0.04 0.26 0.41 0.20

0.9952 −0.9682 −0.9998 0.9690 −0.9891 −0.9824 −0.9921 −0.9204 −0.9372 −0.9646 0.9705 0.9969 0.9994 0.9884 0.9818 0.9847

356.4 438.1 577.1 622.2 660.5 666.1 686.0 688.5 714.5 732.2 264.7 274.3 311.1 318.3 336.9 339.3

88.1, 265.5, 148.9, 271.0, 519.4, 565.2, 430.8, 430.6, 573.4, 183.8, 205.6, 106.1, 69.2, 102.1, 156.6, 303.3,

338.3 283.7 192.5 510.4 621.3 620.7 544.9 547.5 658.6 591.1 247.0 256.3 83.1 256.3 170.3 321.3

40 25 30 25 22 28 20 20 25 25 25 35 20 35 25 10

1.73 1.86 2.09

0.39 0.13 0.24

0.9641 0.9316 0.9963

353.1 353.2 380.4

135.1, 279.3 146.3, 253,1 278.7, 339.1

25 25 15

2.61 2.62

0.29 0.21

0.9890 0.9847

392.5 394.4

154.6, 239.1 184.7, 352.6

20 25

1.62 1.75

0.04 0.10

−0.9902 0.9921

401.4 406.4

81.0, 95.1 256.3, 300.3

30 33

423.9 105.4, 282.6 466.3 325.3, 448.3 480.3 308.3, 339.3 494.3 353.3, 476.3 496.3 184.1, 478.3 520.3 184.1, 481.2 522.3 104.1, 184.1 524.4 184.1, 319.1 530.3 389.3, 487.3 MRM transitions

10 20 32 18 28 25 30 30 20

putative identification 3-{(4a′S,5′S,8a′S)-8a′-[(methoxymethoxy) methyl]-5′-methyloctahydro-5′methyloctahydro-2′H-spiro[1,3-dioxolane2,1′-naphthalen]-5′-yl}propanal alpha,4,2′-trihydroxy-4-Ogeranyldihydrochalcone alanyl-poly(glycerolphosphate) medicarpin 3-O-glucoside GPEtn(15:0/0:0) GPEtn(16:1(9Z)/0:0) GPEtn(18:0/0:0) GPEtn(15:0/14:1(9Z))

2.62 0.06 2.67 0.25 2.54 0.05 7.87 0.10 2.37 0.11 2.69 0.12 2.20 0.05 1.85 0.07 2.07 0.43 fold change (n = 5)

0.9958 0.9941 0.9896 0.9954 0.9987 0.9969 0.9961 0.996 0.9936 SIMCA-P

daughter ions

collision energy (V)

mean

CV

value

parent ion

daughter ions

collision energy (V)

25540.4

0.01

−0.9987

338.9

162.9, 182.7

32

4.12

0.12

0.9811

395.3

84.8, 116.6

25

2.30 2.08 7.25 2.50 2.88 2.49

0.43 0.40 0.04 0.06 0.36 0.17

0.9735 0.9342 0.9829 0.9989 −0.9867 0.9903

395.9 430.9 438.1 450.0 480.1 645.8

a

171.6, 162.8, 227.2, 195.5, 195.7, 198.9,

351.7 451.0 240.8 252.7 283.2 267.0

20 25 20 20 22 25

Coefficient of variation. bDifferential metabolites detected in both CAP- and FFC-resistant mutants.

tandem mass spectrometry was operated in ESI+ and ESI− modes, respectively. ESI parameters were as follows: capillary voltage, 3.0 kV; source temperature, 100 °C; desolvation temperature, 300 °C; cone gas flow, 30 L/h; desolvation gas

quadrupole MS (UHPLC−MS/MS) in multiple reaction monitoring (MRM) mode. The UHPLC conditions were the same as the untargeted metabolomic analysis method previously described. A Waters Quattro LC triple quadrupole 1062

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

Table 3. Putative Identities of Differential Metabolites in FFC-Resistant Mutants fold change (n = 5) [M+H]+ (Da)

ESI+

ID

RT (min)

upregulated

1 2 3 4 5

5.65 3.97 4.50 5.52 4.53

133.0646 219.1722 318.2990b 356.3510b 406.3520b

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

4.96 6.15 6.78 9.07 8.88 8.99 8.76 8.79 8.79 9.59 3.11 10.28 8.07 5.61 8.89 9.01 8.62

302.3040 466.2920b 577.1322b 338.3404 660.4604b 686.4774b 608.4260 634.4455 656.4280 714.5117b 265.0950b 225.0893 337.1655b 263.1261 353.1960b 353.2114b 380.2206b

23 24

6.27 9.01

393.2500b 394.2353b

25 26 27 28 29 30 31 32 33 34 35 36

9.00 5.30 6.40 6.76 6.24 6.02 6.39 6.92 6.50 5.19 6.24 6.03

401.3513b 424.2523b 480.3140b 494.3284b 496.3400b 520.3399b 522.3554b 524.3739b 530.3268b 387.1780 454.2910 544.3406

downregulated

ESI−

ID

RT (min)

[M−H]− (Da) b

upregulated

37

8.12

339.2177

downregulated

38

8.89

395.1853b

39 40

5.80 5.63

450.2592b 424.2453

a

putative identification L-asparagine

6,8,10,12-pentadecatetraenal phytosphingosine arachidoyl ethanolamide (±)N-(2-fluoro-ethyl)-2,16-dimethyl5Z,8Z,11Z,14Z-docosatetraenoyl amine sphinganine PC(14:1(9Z)/0:0) proanthocyanidin A2 13Z-docosenamide GPEtn(18:2(9Z,12Z)/12:0) GPEtn(14:0/18:3(9Z,12Z,15Z)) GPEtn(12:0/14:0) GPEtn(16:1(9Z)/12:0) GPEtn(18:4(6Z,9Z,12Z,15Z)/12:0) GPEtn(16:0/18:3(9Z,12Z,15Z)) 9,12-octadecadienal 3-hydroxy-L-kynurenine 1-(2-butoxy-2-oxoethyl)-2-butyl phthalate artemisin 9,10-dichloro-octadecanoic acid apo-8′-bixinal 3-{(E)-2-[4-(benzyloxy)-3-methoxyphenyl] vinyl}-5-pentyl-4,5-dihydro-1,2-oxazole CPA(16:0/0:0) 4-ethyl-3-(4-methoxyphenyl)-2,2-dimethyl-8(1-pyrrolidinylmethyl)-2H-chromen-7-ol L-palmitoylcarnitine GPEtn(14:1(9Z)/0:0) GPEtn(18:1(9Z)/0:0) PC(16:1(9Z)/0:0) PC(16:0/0:0)[U] PC(18:2/0:0) PC(18:1(9Z)/0:0) PC(18:0/0:0) GPEtn(22:4/0:0) Lys Cys His GPEtn(16:0/0:1) PC(20:4(5Z,8Z,11Z,14Z)/0:0)

putative identification

mean

CVa

SIMCA-P

1.64 1.84 2.38 2.41 1.72

0.33 0.39 0.12 0.05 0.49

0.9979 −0.9992 0.9973 0.9981 0.9988

132.8 219.1 318.3 356.4 406.4

104.7, 132.9, 102.1, 88.1, 256.3,

122.6 202.8 256.3 338.3 300.3

18 25 35 40 33

1.71 2.17 64.39 1.75 2.59 4.35 1.87 1.82 1.62 5.03 1.46 2.45 2.47 1.81 2.78 1.85 2.30

0.16 0.31 0.09 0.22 0.31 0.27 0.29 0.09 0.23 0.39 0.08 0.42 0.10 0.07 0.27 0.30 0.20

0.9964 −0.9857 −0.9920 −0.9587 −0.9891 −0.9969 −0.9884 −0.9951 −0.9940 −0.9585 0.9825 0.9116 0.9972 0.9972 0.9768 0.9543 0.9927

302.3 466.3 338.3 577.1 660.5 686.0 608.3 634.4 656.4 714.5 262.5 224.7 336.9 262.5 353.1 353.2 380.4

106.1, 325.3, 148.9, 121.1, 519.4, 430.8, 182.9, 493.4, 519.4, 573.4, 205.6, 181.9, 156.6, 84.7, 170.7, 135.1, 278.7,

284.3 448.3 192.5 296.3 621.3 544.9 467.0 615.5 613.4 658.6 247.0 207.0 170.3 133.0 184.6 279.3 339.1

35 20 30 33 22 20 20 20 30 25 25 35 25 20 20 25 15

1.99 1.99

0.26 0.44

0.9894 0.9943

392.5 394.4

154.6, 239.1 184.7, 352.6

20 25

9.46 0.29 2.62 0.06 1.65 0.36 2.32 0.19 2.52 0.14 1.63 0.36 1.64 0.28 4.01 0.17 1.91 0.41 1.82 0.34 2.35 0.07 1.86 0.49 fold change (n = 5)

0.9898 0.9958 0.9985 0.9739 0.9977 0.9971 0.9996 0.9854 0.9964 0.9946 0.9982 0.9925

401.4 423.9 480.3 494.3 496.3 520.3 522.3 524.4 530.3 387.2 454.3 543.8

81.0, 95.1 105.4, 282.6 308.3, 339.3 353.3, 476.3 184.1, 478.3 184.1, 481.2 104.1, 184.1 184.1, 319.1 389.3, 487.3 105.1,358.6 313.3, 436.3 104.1, 184.0

30 10 32 18 28 25 30 30 20 17 18 25

daughter ions

collision energy (V)

MRM transitions

daughter ions

collision energy (V)

CVa

SIMCA-P

parent ion

0.25

−0.9987

338.9

162.9, 182.7

32

3.64

0.49

0.9621

395.3

84.8, 116.6

25

2.58 2.93

0.45 0.28

0.9976 0.9898

450.0 423.9

195.5, 252.7 195.6, 226.7

20 20

mean

3-{(4a′S,5′S,8a′S)-8a′-[(methoxymethoxy) methyl]-5′-methyloctahydro-5′methyloctahydro-2′H-spiro[1,3-dioxolane2,1′-naphthalen]-5′-yl}propanal alpha,4,2′-trihydroxy-4-Ogeranyldihydrochalcone GPEtn(16:1(9Z)/0:0) GPEtn(14:0/0:0)

MRM transitions parent ion

1676

Coefficient of variation. bDifferential metabolites detected both in CAP- and FFC-resistant mutants.

flow, 650 L/h. Collision-induced dissociation was performed using argon as the collision gas at the pressure of 2.5 × 10−3 mbar in the collision cell. MS/MS parameters including fragment ions and collision energy of each feature were optimized (Table 2 and Table 3). The quantitative analysis was determined using five biological replicates. Only the metabolite

candidates with relatively large fold change (>1.5) were considered as the potential biomarkers. Data Analysis

Data were processed by MarkerLynx software (Waters). Peak detection was performed across the mass range of m/z 100− 1063

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

1000 with retention time (RT) between 0.3 and 11 min, peak widths automatically detected, an intensity threshold of 100, mass window of 0.02, RT window of 0.1, and noise elimination of 6.0. Multivariate statistical analysis was performed by EZinfo software (Waters) for unsupervised principal component analysis (PCA) to obtain a general overview of the variance of metabolic phenotypes. Then, orthogonal projection to latent structures−discriminant analysis (OPLS−DA) was used to maximize the differences of metabolic profiles between parent and mutant strains. Loading plots were generated from OPLS, showing the impact of variables on the formation of vectors. The combined S plots and variable importance in the projection (VIP) plots from the OPLS−DA analysis were used to select distinct variables as potential biomarkers. For metabolite feature identification, elemental composition analysis was first performed to obtain some parameters such as deviation from calculated mass (mDa or ppm), double bond equivalent (DBE), and i-fit value (the isotopic pattern of the selected ion), which were used to evaluate the accuracy of possible formulas. Then, MassFragment software (Waters) was applied to facilitate the MS/MS fragment ion analysis. Identification of features was based on retention behavior, mass assignment, and online database query. Databases queried included Lipdmaps, Markers, KEGG, Metlin, Chemspider, Human Metabolome Database (HMDB), and MassBank. These databases were searched with a mass tolerance of 0.01 Da. The identities of the specific metabolites were confirmed by comparing their mass spectra and chromatographic retention behavior with those obtained using commercially available reference standards. Metabolic pathway analysis was performed in the KEGG (http://www.genome.jp/kegg/) database to investigate the disturbed metabolic pathways and facilitate biological interpretation. The metabolites and corresponding pathways were imported into Cytoscape software (v.2.8.3) for visualization of the network models. The heat map was constructed with MetaboAnalyst data annotation tools (version 2.0) based on the potential candidates of importance that were extracted with OPLS−DA and UHPLC−MS/MS analysis.

■

Figure 1. Standard Venn diagram depicting the number of unique and shared features for select CAP- (C) or FFC-resistant (F) mutants and parent strain (3) in positive ion mode using different quenching methods. MC, 60%MeOH quenching and MeOH/CHL (2:1 v/v) extraction; MA, 60%MeOH quenching and ACN/H2O (50:50 v/v, 0.1% FA) extraction; MM, 60%MeOH quenching and 100%MeOH extraction; NC, liquid nitrogen quenching and MeOH/CHL (2:1 v/v) extraction; NA, liquid nitrogen quenching and ACN/H2O (50:50 v/v, 0.1% FA) extraction; NM, liquid nitrogen quenching and 100%MeOH extraction.

extraction solvents (MeOH, MeOH/CHL (2:1, v/v), 60% EtOH (containing 5 mM ammonium acetate), and ACN/H2O (50:50, v/v, containing 0.1% FA)) and two different extraction methods (sonication and freeze−thaw cycles). The extraction efficiency of freeze−thaw method was better, and the sonication method was time-consuming when processing multiple samples. As shown in Figure 1 and Table S3 in the Supporting Information, our results also indicated that it was necessary to adopt different extraction solvents to cover as much meaningful metabolites as possible. Overall, our sample preparation procedure consisted of two quenching methods and three different extraction solvents (Figure S1 of the Supporting Information).

RESULTS AND DISCUSSION

Optimization of Sample Preparation Methods

A variety of quenching techniques have been adopted for bacterial metabolomic analysis, with the most popular methods being quenching in cold organic solvents, followed by centrifugation and extraction or fast filtration. Howlett et al.24 used liquid nitrogen to quench the cell metabolism and direct injection mass spectrometry for metabolomic analysis of C. jejuni for mutant characterization. In this study, we evaluated different quenching methods of cell metabolism according to the number of significant biomarkers retrieved using each method. First, two different quenching methods (cold centrifugation and fast filtration) were compared using 60% MeOH as quenching solution. According to the result of detected metabolite candidates, cold centrifugation was better than fast filtration (data not shown). Next, four different quenching solutions were studied as well as direct quenching with liquid nitrogen. We found that the results of cold centrifugation with 60%MeOH and direct quenching with liquid nitrogen were complementary (Table S1 of the Supporting Information). We next optimized the extraction of the metabolites from the quenched sample. Samples were extracted with four different

Discovery and Identification of Metabolites

For QC purposes, several reference standards were added to verify the reliability of the metabolomics analysis. This included the accuracy of calculated mass and the stability of RT and peak area of reference standards. As shown in Supplementary Table S2 in the Supporting Information, the average error of calculated mass was below 10 ppm, the RT was nearly identical, and the RSD of the peak area in three batches (different extracts) was 1.6−5.1%. After raw data processing and peak detection, 4400 features in the positive mode and nearly 900 features in negative mode were observed. The PCA scores plot effectively clustered biological replicates of different groups, and the antibioticresistant mutants were separated from parent strains (Figure 2A,C). From the corresponding loading plots of OPLS analysis, 1064

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

Figure 2. Metabolomic profile of mutant and parent strains in both positive or negative ionization mode by UHPLC−TOF/MS. (A) Principal component analysis of microbial extracts from mutant and parent strains in ESI+. (B) Representative S plot of parent strain (+1) versus CAPresistant strain (−1) in ESI+. (C) Principal component analysis of microbial extracts from mutant and parent strains in ESI−. (D) Representative S plot of parent strain (+1) versus CAP-resistant strain (−1) in ESI−.

Figure 3. Typical identification of metabolite features. (A) Extract ion chromatogram of m/z 496.3404 in quenched C. jejuni cells using UHPLC− TOF/MS in positive ion mode. (B) Extract ion chromatogram of m/z 496.3404 in commercial reference standard using UHPLC−TOF/MS in positive ion mode. (C) Product ion scan spectrum of m/z 496.3404 in quenched C. jejuni cells. (D) Product ion scan spectrum of reference standard.

the ions far away from the origin were regarded as the differential metabolites (Figure 2B,D). Filtering by VIP plot scores (VIP> 1.5) and the Standard in Multivariate Data Analysis (SIMCA-P) analysis value (P < −0.9 or P > 0.9), more than 200 potential features were obtained from different quenching and extracting process. After that, we determined the fold-change ratios of these features between mutant and parent strains by UHPLC−MS/MS in MRM mode, which is the standard technique for quantitative analysis of small

molecules in biological matrices. Only about one-fifth of the potential features were confirmed by the relative quantitative analysis. The identification of metabolites is a challenge in the metabolomics analysis. The feature at m/z 496.3404 was taken as an example to show the identification process workflow in this study. Among the possible chemical formulas generated by elemental composition analysis, C24H50NO7P was selected because of its low mass error and low i-fit value. Three compounds appeared as candidates from Markers, HMDB, and 1065

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

Figure 5. Metabolomic network visualization of disturbed metabolic pathways in amphenicol-resistant mutants compared with parent strains. Green, disturbed metabolic pathways; gray, differential metabolites with fold change 5. (A) Visualization of disturbed metabolic pathways in CAP-resistant mutants. (B) Visualization of disturbed metabolic pathways in FFCresistant mutants. Figure 4. Heat map visualization constructed based on the differential metabolites of importance for CAP- (A) or FFC- (B) resistant mutants. Rows: samples; Columns: differential metabolites. Color key indicates metabolite expression value: green, lowest; red, highest.

almost not detected in the parent strains but overexpressed in both resistant strains. The metabolite data were visualized using a heat map, and the mutant groups and parent group could be distinguished by the identified features (Figure 4). Metabolic Pathway Analysis

MassBank after online database query through MarkerLynx. The structural formula of each candidate was imported into MassFragment to obtain the possible fragmentation pattern, which was compared with MS/MS spectrum of the feature. The result indicated that this metabolite was PC(16:0/0:0)[U], which was finally confirmed by an authentic standard (Figure 3). In this study, 16 compounds were identified using authentic standards. A total of 41 differential metabolites (11 upregulated and 30 down-regulated) were identified and contributed to CAP resistance in C. jejuni (Table 2), and a panel of 40 differential metabolites (17 up-regulated and 23 down-regulated) responded to FFC resistance in C. jejuni (Table 3). Among all of the identified metabolites, 27 features simultaneously appeared in CAP- and FFC-resistant mutants. According to the metabolite rankings from the PCA and OPLS−DA, the compounds considered most important for differentiating the sample groups were features at m/z 577.1320 (ESI+) and m/z 339.2177 (ESI−). The two metabolites were

To facilitate access to the metabolic pathway data, altered metabolites from the merged data set were mapped to KEGG reference pathways, and interaction networks were generated in Cytoscape. As shown in Figure 5, the association network of differentially expressed metabolites was constructed. In this study, glycerophospholipid metabolism, sphingolipid metabolism, and fatty acid metabolism were the most important metabolic pathways related to CAP resistance in C. jejuni. Additionally, the top three metabolic pathways of importance, including glycerophospholipid metabolism, sphingolipid metabolism, and tryptophan metabolism, were found to be disturbed in FFC-resistant mutant strains. Most importantly, glycerophospholipid metabolism was significantly affected owing to the occurrence of amphenicol resistance in C. jejuni. Glycerophospholipids and sphingolipids are important membrane lipids in microorganisms and play a vital role in cellular functions, including the regulation of transport processes, 1066

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

protein function, and signal transduction.25 Additionally, glycerophospholipids are essential components of lipoproteins and influence their function and metabolism.26,27 Several studies demonstrated that drug resistance in Mycobacterium tuberculosis could decrease gene expression and result in downregulation of glycerophospholipid metabolism, and there was a correlation between cellular drug resistance and alterations in glucosylceramides (belonging to lipids) metabolism in multidrug-resistant cancer cells.28,29 In this study, we found that the identified glycerophospholipids were overexpressed in both resistant strains. These results suggested that glycerophospholipids exerted an important role in acquiring or maintaining amphenicol resistance in C. jejuni.

(3) Bae, W.; Kaya, K. N.; Hancock, D. D.; Call, D. R.; Park, Y. H.; Besser, T. E. Prevalence and antimicrobial resistance of thermophilic Campylobacter spp. from cattle farms in Washington State. Appl. Environ. Microb. 2005, 71, 169−174. (4) Gupta, A.; Nelson, J. M.; Barrett, T. J.; Tauxe, R. V.; Rossiter, S. P.; Friedman, C. R.; Joyce, K. W.; Smith, K. E.; Jones, T. F.; Hawkins, M. A.; Shiferaw, B.; Beebe, J. L.; Vugia, D. J.; Rabatsky-Ehr, T.; Benson, J. A.; Root, T. P.; Angulo, F. J. Antimicrobial resistance among Campylobacter strains, United States, 1997−2001. Emerging Infect. Dis. 2004, 10, 1102−1109. (5) Nachamkin, I. Chronic effects of Campylobacter infection. Microbes Infect. 2002, 4, 399−403. (6) Luangtongkum, T.; Jeon, B.; Han, J.; Plummer, P.; Logue, C. M.; Zhang, Q. Antibiotic resistance in Campylobacter: emergence, transmission and persistence. Future Microbiol. 2009, 4, 189−200. (7) Allo, B. M. Campylobacter jejuni infections: update on emerging issues and trends. Clin. Infect. Dis. 2001, 32, 1201−1206. (8) Schwarz, S.; Kehrenberg, C.; Doublet, B.; Cloeckaert, A. Molecular basis of bacterial resistance to chloramphenicol and florfenicol. FEMS Microbiol. Rev. 2004, 28, 519−542. (9) Chen, X.; Naren, G. W.; Wu, C. M.; Wang, Y.; Dai, L.; Xia, L. N.; Luo, P. J.; Zhang, Q.; Shen, J. Z. Prevalence and antimicrobial resistance of Campylobacter isolates in broilers from China. Vet. Microbiol. 2010, 144, 133−139. (10) de Moura, H. M.; Silva, P. R.; da Silva, P. H.; Souza, N. R.; Racanicci, A. M.; Santana, A. P. Antimicrobial resistance of Campylobacter jejuni isolated from chicken carcasses in the Federal District, Brazil. J. Food Prot. 2013, 76, 691−693. (11) Wang, Y.; Taylor, D. E. Chloramphenicol resistance in Campylobacter coli: nucleotide sequence, expression, and cloning vector construction. Gene 1990, 94, 23−28. (12) Bissonnette, L.; Champetier, S.; Buisson, J. P.; Roy, P. H. Characterization of the non-enzymatic chloramphenicol resistance (cmlA) gene of the In4 integron of Tn1696, similarity of the product to transmembrane transport proteins. J. Bacteriol. 1991, 173, 4493− 4502. (13) Arcangioli, M. A.; Leroy-Setrin, S.; Martel, J. L.; Chaslus-Dancla, E. A new chloramphenicol and florfenicol resistance gene linked to an integron structure in Salmonella typhimurium DT104. FEMS Microbiol. Lett. 1999, 174, 327−332. (14) He, T.; Shen, J.; Schwarz, S.; Wu, C.; Wang, Y. Characterization of a genomic island in Stenotrophomonas maltophilia that carries a novel f loR gene variant. J. Antimicrob. Chemother. 2014, DOI: 10.1093/ jac/dku491. (15) Kehrenberg, C.; Schwarz, S. fexA, a novel Staphylococcus lentus gene encoding resistance to florfenicol and chloramphenicol. Antimicrob. Agents Chemother. 2004, 48, 615−618. (16) Gomez-Sanz, E.; Kadlec, K.; Feßler, A. T.; Zarazaga, M.; Torres, C.; Schwarz, S. A novel fexA variant from a canine Staphylococcus pseudintermedius isolate that does not confer florfenicol resistance. Antimicrob. Agents Chemother. 2013, 57, 5763−5766. (17) Liu, H.; Wang, Y.; Wu, C.; Schwarz, S.; Shen, Z.; Jeon, B.; Ding, S.; Zhang, Q.; Shen, J. A novel phenicol exporter gene, fexB, found in enterococci of animal origin. J. Antimicrob. Chemother. 2012, 67, 322− 325. (18) Schwarz, S.; Werckenthin, C.; Kehrenberg, C. Identification of a plasmid-borne chloramphenicol-florfenicol resistance gene in Staphylococcus sciuri. Antimicrob. Agents Chemother. 2000, 44, 2530−2533. (19) Ma, L.; Shen, Z.; Naren, G.; Li, H.; Xia, X.; Wu, C.; Shen, J.; Zhang, Q.; Wang, Y. Identification of a novel G2073A mutation in 23S rRNA in amphenicol-selected mutants of Campylobacter jejuni. PLoS One 2014, 9, e94503. (20) Meyer, H.; Weidmann, H.; Lalk, M. Methodological approaches to help unravel the intracellular metabolome of Bacillus subtilis. Microb. Cell Fact. 2013, 12, 69−81. (21) Bolten, C. J.; Kiefer, P.; Letisse, F.; Portais, J. C.; Wittmann, C. Sampling for metabolome analysis of microorganisms. Anal. Chem. 2007, 79, 3843−3849.

■

CONCLUSIONS In the current study, we presented a LC−MS-based metabolomics method to characterize metabolome scale changes in amphenicol-resistant C. jejuni to investigate the mechanism of antibiotic resistance. Different quenching and extraction methods were necessary to cover a wide range of metabolites due to the complexity of their physicochemical properties. UHPLC coupled to quadrupole mass spectrometry was a complementary tool for rapid confirmation of the altered metabolites resulting from UHPLC−TOF/MS analysis. Our findings strongly indicated that glycerophospholipid metabolism was closely related to the amphenicols resistance in C. jejuni.

■

ASSOCIATED CONTENT

S Supporting Information *

Figure S1: Optimized workflow of metabolomic analysis of C. jejuni. Table S1: Optimization of the quenching method of C. jejuni cell metabolism. Table S2: QC results of UHPLC−TOF/ MS analysis. Table S3: Differential metabolites of different quenching and extracting methods using UPLC−TOF/MS in positive ion mode. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Tel: +86-10-62732803. Fax: +86-10-62731032. E-mail: sjz@ cau.edu.cn. Author Contributions †

H.L. and X. X. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by grants from National Basic Research Program of China (973 Program, grant no. 2013CB127200), Program for New Century Excellent Talents in University (grant no. 2014FG045), and Chinese Universities Scientific Fund (grant no. 2013YJ005).

■

REFERENCES

(1) Mead, P. S.; Slutsker, L.; McCaig, L. F.; Bresee, J. S.; Shapiro, C.; Griffin, P. M.; Tauxe, R. V. Food-related illness and death in the United States. Emerging Infect. Dis. 1999, 5, 607−625. (2) EFSA. European Union summary report on trends and sources of zoonoses, zoonotic agents and food-borne outbreaks in 2012. EFSA J. 2014, 12 (3547), 312. 1067

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068

Journal of Proteome Research

Article

(22) Faijes, M.; Mars, A. E.; Smid, E. J. Comparison of quenching and extraction methodologies for metabolome analysis of Lactobacillus plantarum. Microb. Cell Fact. 2007, 6, 27. (23) Van Gulik, W. M.; Canelas, A. B.; Taymaz-Nikerel, H.; Douma, R. D.; de Jonge, L. P.; Heijnen, J. J. Fast sampling of the cellular metabolome. Methods Mol. Biol. 2012, 881, 279−306. (24) Howlett, R. M.; Davey, M. P.; Paul Quick, W.; Kelly, D. J. Metabolomic analysis of the food-borne pathogen Campylobacter jejuni: application of direct injection mass spectrometry for mutant characterisation. Metabolomics 2014, 10, 887−896. (25) Ecker, J.; Liebisch, G. Application of stable isotopes to investigate the metabolism of fatty acids, glycerophospholipid and sphingolipid species. Prog. Lipid Res. 2014, 54, 14−31. (26) Scherer, M.; Bottcher, A.; Liebisch, G. Lipid profiling of lipoproteins by electrospray ionization tandem mass spectrometry. Biochim. Biophys. Acta 2011, 1811, 918−924. (27) Wiesner, P.; Leidl, K.; Boettcher, A.; Schmitz, G.; Liebisch, G. Lipid profiling of FPLC-separated lipoprotein fractions by electrospray ionization tandem mass spectrometry. J. Lipid Res. 2009, 50, 574−585. (28) Chen, L. C.; Yeh, H. Y.; Yeh, C. Y.; Arias, C. R.; Soo, V. W. Identifying co-targets to fight drug resistance based on a random walk model. BMC Syst. Biol. 2012, 6, 5. (29) Lavie, Y.; Cao, H.; Bursten, S. L.; Giuliano, A. E.; Cabot, M. C. Accumulation of glucosylceramides in multidrug-resistant cancer cells. J. Biol. Chem. 1996, 271, 19530−19536.

1068

dx.doi.org/10.1021/pr501061d | J. Proteome Res. 2015, 14, 1060−1068