Journal of Chromatography B, 973 (2014) 55–60

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

Magnetic ligand fishing combination with high-performance liquid chromatography–diode array detector–mass spectrometry to screen and characterize cyclooxygenase-2 inhibitors from green tea Xu Deng a , Shuyun Shi b,∗ , Simin Li c , Tianlun Yang a,∗∗ a

Department of Cardiology, Xiangya Hospital, Central South University, Changsha 410008, PR China College of Chemistry and Chemical Engineering, Central South University, Changsha 410083, PR China c Xiangya Stomatological Hospital, Central South University, Changsha 410008, PR China b

a r t i c l e

i n f o

Article history: Received 6 June 2014 Accepted 8 October 2014 Keywords: Cyclooxygenase-2 Magnetic ligand fishing HPLC–DAD–MSn Green tea Catechin

a b s t r a c t Cyclooxygenase-2 (COX-2) inhibitors may be used to efficiently treat inflammation or cancer diseases. In the present study, we established a new screening assay based on magnetic Fe3 O4 @SiO2 –COX-2 ligand fishing combination with high-performance liquid chromatography–diode array detector–mass spectrometry (HPLC–DAD–MSn ) to screen and identify COX-2 inhibitors from green tea. Optimized conditions (pH at 7.4, temperature at 30 ◦ C, and incubation time for 30 min) for fishing out COX-2 inhibitors were achieved by testing positive control, celecoxib, with active and inactive COX-2. Notably, immobilized COX-2 showed high stability (remained 94.7% after ten consecutive cycles), reproducibility (RSD < 10% for batch-to-batch evaluation). Finally, eight catechins with COX-2 binding activity were screened in green tea, and their structures were characterized by ultraviolet (UV), accurate molecular weight, diagnostic fragment ions and nuclear magnetic resonance (NMR). Particularly, the COX-2 inhibitory activities of two rare catechins, [(−)-epigallocatechin-3-(3 -O-methyl)-gallate (3 -O-methyl-EGCG, IC50 = 0.17 ± 0.03 ␮M 0.16 ± 0.01), (−)-epicatechin-3-(3 -O-methyl)-gallate (3 -O-methyl-ECG, IC50 = 0.16 ± 0.02 ␮M)], were reported for the first time. The results indicated that the proposed method was a simple, robust and reproducible approach for the discovery of COX-2 inhibitors from complex matrix. © 2014 Published by Elsevier B.V.

1. Introduction Cyclooxygenase-2 (COX-2), an inducible COX isozyme rapidly in response to cytokines, growth factors and tumor promoters, is dramatically up-regulated during pathological conditions such as inflammation and cancers [1,2]. Its inhibition has been more directly implicated in reducing the extent of polyposis in patients with familial adenomatous polyposis [3], inducing apoptosis in colon, stomach and prostate cancer in human [2,4,5]. Therefore, COX-2 has become an important drug target for the discovery and development of anti-inflammatory or anti-tumor drugs. Up to now, great varieties of components with effective COX-2 inhibitory activity have been designed and synthesized [6–8]. Moreover, investigations on both animal models and human clinical trials have led to the hypothesis that

∗ Corresponding author. Tel.: +86 731 8883 0833; fax: +86 731 8887 9616. ∗∗ Corresponding author. Tel.: +86 731 8432 7250; fax: +86 731 84327232. E-mail addresses: [email protected] (S. Shi), [email protected] (T. Yang). http://dx.doi.org/10.1016/j.jchromb.2014.10.010 1570-0232/© 2014 Published by Elsevier B.V.

synthesized selective COX-2 inhibitors might be endowed with better anti-inflammatory activity with fewer gastrointestinal side effects than nonselective classical nonsteroidal anti-inflammatory drugs [6]. Unfortunately, cardiovascular adverse effects with some selective COX-2 inhibitors (e.g. coxibs) have ultimately prompted them withdrawal from the market [9,10]. Therefore, the development of selective COX-2 inhibitors with improved safety profiles is the need of hour. Newman and Cragg have reported that about 50% of all the marketed-new chemical entities were shown to be of natural origin during 1981–2010 [11]. Up to now, many natural products, such as Krameria pauciflora, Radix aconiti, Curcuma longa, Terminalia bellerica and Camellia sinensis (tea leaves), have been reported to exhibit significant COX-2 inhibition [12–15]. However, most reports focus on the COX-2 inhibitory activity of crude extracts or commercially isolated compounds [16]. For example, catechins, the main compounds in nonfermented green tea, have diverse pharmacological properties (i.e. antioxidative, antiatherosclerotic, anticarcinogenic, hypocholesterolaemic, antihypertensive, antiallergic, antimicrobial, and COX-2 inhibitory properties [17–19]), and

56

X. Deng et al. / J. Chromatogr. B 973 (2014) 55–60

the beneficial effects of catechins have increasingly stimulated the usage of green tea as food and cosmetic additives. However, no biochemical profile of catechins in green tea for COX-2 inhibition has been previously reported, and only commercial catechins have been investigated. Traditional bioassay-guided fractionation of natural products is a time-consuming, labor intensive and expensive strategy, which sometimes leads to the loss of activity because of the dilution or decomposition during the isolation and preparation process [20]. Therefore, the key step in natural product research is to develop a selective screening assay to reduce time, cost and the incidence of false positives/negatives. Affinity based screening assay, depending on the macromolecular target–ligands binding, has been considered as the most convenient and efficient method to fish out potential ligands from complex matrix. Powerful approaches are elimination of nonbinders from macromolecular target–ligands using centrifugation [21], ultrafiltration [22,23], equilibrium dialysis [24], microdialysis [25], magnetic beads [26–28]. Among which, magnetic ligand fishing has advantages of stable immobilized macromolecular targets and easy magnetic isolation [29], which have been successfully used to fish out ligands for HSA/BSA [26,30,31], acetylcholinesterase [32], acetylcholine [33], ␣-amylase [34], xanthine oxidase [35], COX-1 [36], maltase [37], SIRT6 [38,39], etc. To the best of our knowledge, no screening assay based on magnetic ligand fishing for COX-2 inhibitors has been previously reported. In view of in continuation of an ongoing efforts aiming at rapidly and efficiently finding bioactive components from complex natural products [26,35,36], we report proof of principle for the first time of integration of magnetic ligand fishing and HPLC–DAD–MSn for facile, specific screening and elucidation of COX-2 inhibitors from green tea. Six common catechins [(−)gallocatechin (GC), (−)-epigallocatechin (EGC), (−)-epicatechin (EC), (−)-epigallocatechin-3-gallate (EGCG), (−)-gallocatechin3-gallate (GCG), (−)-epicatechin-3-gallate (ECG)] and two rare catechins [(−)-epigallocatechin-3-(3 -O-methyl)-gallate  (−)-epicatechin-3-(3 -O-methyl)-gallate (3 -O-methyl-EGCG), (3 -O-methyl-ECG)] were screened and identified, and their COX-2 inhibitory activities were evaluated. 2. Experimental 2.1. Chemicals and reagents Human recombinant COX-2, arachidonic acid, prostaglandin E2 (PGE2 ), d4 -PGE2 , 3-aminopropyltriethoxysilane (APTES), tetraethyl orthosilicate (TEOS), and glutaraldehyde (25% (w/v) aqueous solution) were purchased from Sigma (St. Louis, MO, USA). Acetonitrile, methanol and acetic acid (HPLC grade) were bought from Tedia Company Inc. (OH, USA). Ultrapure water (18.2 M) was prepared with a Milli-Q water purification system from Millipore (Bedford, MA, USA). Phosphate buffer solution (10 mM, pH 7.4) was selected. Reference standards (purities > 99%), GC, EGC, EC, EGCG, GCG, ECG, and positive control sample, celecoxib, were purchased from the National Institute for the Control of Pharmaceutical and Biological Products (Beijing, China). Yunwu green tea (C. sinensis) was obtained from Tea Research Institute of Hunan province, China, and authenticated by Prof. Mijun Peng, Key Laboratory of Hunan Forest Products and Chemical Industry Engineering, Jishou University, Zhangjiajie, China. 2.2. Instrumental Transmission electron microscopy (TEM) (JEM-2100F, JEOL, Japan) was used to observe morphology of microspheres. The FT-IR spectra (4000–400 cm−1 ) were obtained via a Nicolet 6700

FT-IR spectrometer (Thermo Nicolet Co., Waltham, MA, USA). The magnetic property was measured at room temperature using vibration sample magnetometer (VSM7407, Lakeshore, USA). 1 H NMR experiments were performed on a VARIAN INOVA-400 (Varian Corporation, USA) NMR spectrometer. Chromatographic separation was performed on an analytical SunFireTM C18 column (250 mm × 4.6 mm i.d., 5 ␮m, Waters, Milford, MA) in tandem with a Phenomenex C18 guard cartridge (4.0 mm × 3.0 mm, Phenomenex, Torrance, CA). The eluent was delivered from an Agilent 1260 HPLC system and diode array detector system. The mobile phase for green tea extract analysis was consisted of A (0.1% acetic acid in water) and B (0.1% acetic acid in acetonitrile) with a 60 min linear gradient elution from 4% to 25% B at a flow rate of 1.0 ml/min at 25 ◦ C, while the chromatograms were monitored at 275 nm and collected by Agilent ChemStation software. Isocratic mobile phase consisted of water–methanol (15:85, v/v) with a flow rate of 0.8 ml/min was prepared for celecoxib analysis at 25 ◦ C, and the chromatograms were acquired at 254 nm. All the mobile phases were prepared daily. The high-resolution MS and MSn experiments were performed on a Thermo Scientific Exactive orbitrap mass spectrometer with negative ion electrospray interfaced to Thermo Accela LC system (San Jose, CA, USA). HPLC conditions were the same as above mentioned. Nitrogen sheath gas was set at 50 arbitrary units and auxiliary gas at 10 arbitrary units. The instrument was operated in full-scan mode from m/z 100 to 800 with a resolution of 50,000. The ion source parameters included heated capillary temperature at 300 ◦ C, capillary voltage and spray voltage at −52 V and −3.5 kV. An additional collision cell for high energy collision induced decomposition (HCD) experiments was incorporated, while the collection of HCD data used a fixed energy setting at 30 V. 2.3. COX-2 inhibition assay COX-2 inhibitory assays were performed according to a previously described PGE2 (a stable oxidation product resulting from COX-2 oxidation of arachidonic acid) detection method [40]. Briefly, hematin (100 ␮M, 2 ␮l) mixed with l-epinephrine (40 mM, 10 ␮l) and made up with buffer solution to a final volume of 150 ␮l. Then COX-2 solution (20 ␮l, 0.2 ␮g) was added and incubated at 30 ◦ C for 2 min. After that, six different concentrations of COX-2 ligands (20 ␮l) from 0 ␮M to 100 ␮M were added and preincubated at 30 ◦ C for 30 min. The COX-2 inhibition reaction was initiated by adding arachidonic acid (50 ␮M, 20 ␮l) and terminated by adding HCl (2.0 M, 20 ␮l). The concentration of product PGE2 was detected by HPLC–MS/MS method by using [d4 ]-PGE2 as surrogate standard. The inhibitory activity was determined by comparing the amount of PGE2 produced with that of a negative control (buffer solution). The COX-2 inhibitory activity was expressed as the half maximal inhibitory concentration (IC50 ). 2.4. Preparation of Fe3 O4 @SiO2 –COX-2 nanoparticles COX-2 was immobilized onto Fe3 O4 @SiO2 solid support by a typical glutaraldehyde activation procedure as shown in Supplementary Scheme S1 [26]. Typically, FeCl3 ·6H2 O (1.35 g), sodium acetate (3.60 g) and polyethylene glycol (1.00 g) were dissolved in ethylene glycol (40 ml), and the mixtures were stirred at room temperature for 60 min and heated at 200 ◦ C in a teflon lined stainless steel autoclave for 8 h to prepare Fe3 O4 nanoparticles (0.37 g). Then, Fe3 O4 nanoparticles (120 mg) were dispersed in the mixture of ethanol (476 ml), water (139 ml) and ammonia aqueous solution (28 wt.%, 15 ml), followed by the addition of TEOS (6 ml). After stirring at room temperature for 8 h, Fe3 O4 @SiO2 nanoparticles (140 mg) were achieved after being washed with water three times and dried in a vacuum oven at 50 ◦ C. Fe3 O4 @SiO2 nanoparticles

X. Deng et al. / J. Chromatogr. B 973 (2014) 55–60

(100 mg) were dispersed in ethanol (200 ml) followed by addition of APTES (2 ml), and then amino-functionalized Fe3 O4 @SiO2 nanoparticles were synthesized after stirring at room temperature for 6 h. After that, amino-functionalized Fe3 O4 @SiO2 nanoparticles (5.0 mg) were suspended in glutaraldehyde solution (5% w/v, 5 ml) for 1 h. After incubation, the aldehyde-modified Fe3 O4 @SiO2 nanoparticles were mixed with COX-2 solution (2.0 ml, containing 1000 ␮g of COX-2) and shaken for 2 h, after which the magnetic Fe3 O4 @SiO2 –COX-2 microspheres were collected magnetically and rinsed three times with buffer solution, and then dispersed in buffer solution at 4 ◦ C for further experiments. The supernatant and rising solution after COX-2 immobilization process were combined to determine the amount of un-immobilized COX-2 by measuring the concentration of PGE2 resulting from COX-2 oxidation of arachidonic acid [40], and then immobilized COX-2 was calculated (mimmobilized COX-2 = morigina COX-2 − mun-immobilized COX-2 ). As a control, the identical manner was applied for the preparation of Fe3 O4 @SiO2 –inactivated COX-2, while inactivated COX-2 was prepared by boiling it in water for 10 min. 2.5. Magnetic solid phase fishing Green tea was pulverized to a homogeneous size, and then 10.0 g of powdered green tea was extracted by 500 ml of 75% (v/v) ethanol three times, each for 3 h. The filtrates were combined and concentrated under vacuum at 40 ◦ C to give dried residue (2.1 g). A stock solution (4 mg/ml, buffer solution) of the residue was then stored at 4 ◦ C for use. Fe3 O4 @SiO2 –COX-2 microspheres (5.0 mg) were suspended in green tea extract (2 ml). After shaking and incubating at 30 ◦ C for 30 min, Fe3 O4 @SiO2 –COX-2–ligand complexes were separated magnetically, and then washed with 2 ml aliquots of buffer solution three times to discard non-specific bound ligands. The washed Fe3 O4 @SiO2 –COX-2–ligand complexes were then incubated with 2 ml of methanol for 1 h to dissociate ligands. After magnetic separation, the supernatant was collected and then analyzed by HPLC–DAD or HPLC–DAD–MSn directly. In the control experiment, green tea extract was incubated with Fe3 O4 @SiO2 –inactivated COX-2. 3. Results and discussion 3.1. Synthesis and characterization of Fe3 O4 @SiO2 –COX-2 microparticles Immobilized enzymes have been suggested for an alternative way to increase their stability and reusability compared to free enzyme [29]. Notably, magnetic microparticles possess significant advantages including, but not being limited to, high surface-to-volume ratio and high supermagnetism. Then, enzyme immobilized on magnetic microparticles can be separated easily and rapidly, which will avoid two-step centrifugation procedures in ultrafiltration assay [35]. Silica was then selected to encapsulate Fe3 O4 microspheres because it could not only form hydrophilic surface to prevent Fe3 O4 microspheres from oxidization and leaking in acidic environment, but also provide them with a silica like surface easily modified with various groups through covalent reaction. Furthermore, our previous research indicated that the encapsulation of Fe3 O4 with SiO2 will improve their dispersion in water, reduce agglomeration phenomenon in repetitious magnetic separation, and then increase their reusability [35]. Therefore, aldehyde-modified Fe3 O4 @SiO2 was selected to immobilize COX-2 through a typical glutaraldehyde activation procedure. Supplementary Figs. S1–S3 showed that COX-2 was successfully immobilized onto the surface of Fe3 O4 @SiO2 microspheres [36],

57

mAU 30

Celecoxib

CF3 N N O S O NH2

20 10

a b c

0 0

2

4

6

min

Fig. 1. HPLC chromatograms of celecoxib (0.8 ␮g/ml) (a) and ligand fishing assay eluent by Fe3 O4 @SiO2 –active COX-2 (b) and Fe3 O4 @SiO2 –inactive COX-2 (c).

and the microspheres had high magnetic responsivity (Supplementary Fig. S4). The amount of immobilized COX-2 on Fe3 O4 @SiO2 surface was about 193 ␮g/mg. However, the activity of immobilized COX-2 was 84.1% of that of free COX-2 in solution, which was probably the unavailable and improperly oriented active site of immobilized COX-2. 3.2. Assay verification When specific binding to COX-2 occurred, the HPLC peak of ligand was obvious, while that incubation with Fe3 O4 @SiO2 –inactive COX-2 was not detected or largely lower. Known COX-2 selective inhibitor, celecoxib, was selected for the development of magnetic ligand fishing. Working pH and temperature affected the activity of COX-2, incubation time affected the binding efficiency of bioactive components to COX-2, washing step eliminate the nonspecific binding, and dissociation solvent was indispensible for the dissociation of specific bound component. Therefore, some essential incubation conditions including working pH (from 5 to 9), incubation temperature (from 25 ◦ C to 65 ◦ C), and incubation time (from 0 min to 60 min) were optimized during the ligand fishing process, while washing times and dissociation solvent (methanol/acetonitrile solution with different amount of water) were investigated. The results revealed that the highest COX-2 activity and binding efficiency of celecoxib could be achieved in the following conditions, pH at 7.4, temperature at 30 ◦ C, and time for 30 min. Working buffer was used as washing solution for three times and methanol was selected to dissociate the specific bound components. As shown in magnetic ligand fishing–HPLC chromatograms in Fig. 1, celecoxib was detected after binding, washing and dissociation from Fe3 O4 @SiO2 –COX-2 but not to Fe3 O4 @SiO2 –inactive COX-2, which indicated that immobilized COX-2 was enzymatically active, no non-specific binding of celecoxib to inactive COX-2 occurred. Meanwhile, celecoxib solution was incubated with Fe3 O4 @SiO2 particles, no obvious celecoxib peak was observed in the HPLC chromatogram of ligand fishing assay (data not shown), which also indicated that non-specific binding to Fe3 O4 @SiO2 could also be negligible. Therefore, the results demonstrate the specificity of this magnetic ligand fishing for the screening of COX-2 ligands. 3.3. Reusability and reproducibility of immobilized COX-2 Seldom changed activity of immobilized COX-2 is a crucial prerequisite to obtain highly reproducible and accurate data by capturing exactly the same amount of ligands in a series of binding/dissociation cycles. The reusability of Fe3 O4 @SiO2 –COX-2 was evaluated by calculating the binding efficiency of celecoxib to COX2 for great cost benefit on extending its applications, and the

58

X. Deng et al. / J. Chromatogr. B 973 (2014) 55–60

mAU 400

169.0136 Galloyl group

100

OH OH HO

4

200

7

100

1

2

3

6

5

8 a b

0 0

10

20

30

OH

80

40

50

min

Fig. 2. HPLC chromatograms of green tea extract (a) and ligand fishing assay eluent by Fe3 O4 @SiO2 –active COX-2 (b).

validation parameters for celecoxib was shown in Supplementary Table S1. In avoid of particle loss for cycling, Fe3 O4 @SiO2 –COX-2 was separated by prolonged exposure to the magnetic field. After ten consecutive binding–dissociation cycles under same conditions, the binding efficiency was still as high as 94.7% of the first cycle. The results manifested that Fe3 O4 @SiO2 –COX-2 were stable and ten cycles had little influence on the activity of immobilized COX-2. The stability of the immobilized COX-2 was tested through inter-day assays. No significant loss of celecoxib binding efficiency (RSD < 8%, n = 5) appeared after 5 days storage. The stable immobilized COX-2 can decrease the test costs and enhance the experimental efficiency. Excellent reproducibility was also attained among three batches of Fe3 O4 @SiO2 –COX-2 (RSD < 10%). 3.4. Optimization of HPLC–DAD–MSn analysis Major bioactive components in green tea were phenolic compounds and purine alkaloids [41], then acid was added into the mobile phase for higher resolution. Moreover, the acid as additive to mobile phase could affect ESI ionization [42]. Different mobile phase compositions (methanol–water, acetonitrile–water with different concentrations of acetic acid/formic acid) and gradient program were investigated to obtain chromatograms with good resolution within an acceptable analysis time. The results indicated that 0.1% acetic acid in water as solvent A and 0.1% acetic acid in acetonitrile as solvent B had higher elution efficiency. The chromatogram was acquired at about maximum absorption wavelength, 275 nm. Under the optimum chromatographic conditions, best resolution, shortest analysis time and lowest pressure variations were achieved, and the HPLC spectrum of green tea extract were shown in Fig. 2A. Catechins were tested to contain COX-2 inhibitory activity [19]. EGCG, the main catechin in green tea extract [41], was then selected and infused for optimization of initial ESI-MS parameters and investigation of diagnostic fragment ions for structural identification. Negative ion mode was selected, and other parameters were optimized automatically. EGCG had a characteristic absorption at 241 and 276 nm with deprotonated molecule [M−H]− with m/z 457.0778 (C22 H17 O11 , 1.5 ppm error) (Fig. 3). The high intensity fragment ion with m/z 169.0136 (C7 H5 O5 ) corresponded to the galloyl group, while the ion with m/z 305.0664 [M−H−C7 H4 O4 ]− , corresponding to GC or EGC, arose from the loss of galloyl group as a neutral species. The ion with m/z 287.0559 [M−H−C7 H6 O5 ]− was a neutral loss of H2 O from fragment ion with m/z 305.0664. The presence of the fragment ion with m/z 125.0241 (C6 H5 O3 ) corresponded to unsubstituted Aring. In addition, a fragment ion with m/z 331 [M−H−C6 H6 O3 ]− was observed probably for the elimination of B-ring followed by cyclization [41]. The results coincided with the published data [43].

Relative Intensity/%

300

O O OH

OH

O

60

OH OH

20 A ring 125.0241 0 100

[M-H]457.0778

[M-H-C7H4O4]305.0664 [M-H-C H O ]-

40

7

6

5

287.0559 331.0456 193.0138

150

200

250

300 m/z

350

400

450

500

Fig. 3. HCD spectrum of EGCG with collision energy at 30 V.

3.5. COX-2 inhibitors in green tea extract Green tea extract showed potent COX-2 inhibitory activity with IC50 value at 437.3 ␮g/ml, which indicated that green tea extract was rich of COX-2 inhibitors. Direct MS detection without chromatographic separation is a rapid and convenient method, which sometimes encounters significant problems of ion suppression due to matrix effects, therefore, components, especially low-abundant components, cannot be identified. Here, magnetic Fe3 O4 @SiO2 –COX-2 ligand fishing combined with HPLC–DAD–MSn may offer a convenient and effective approach to tackle this problem. Fig. 2b showed HPLC chromatogram COX-2 binders in green tea after Fe3 O4 @SiO2 –COX-2 solid phase fishing, which indicated that about eight compounds (1–8) (Fig. 4) were fished out with the ability to bind with COX-2. Compounds 1–8 all typically showed the characteristic absorption maximum of catechins at about 245 and 275 nm. Compounds OH OH

2''

O O 2

HO

R2

3 R1 OH

O

1''

OR3 3'' OH

OH Galloyl group (GA)

Configuration Compounds

R1

R2

R3

1 2S, 3R

(–)-gallocatechin (GC)

OH OH /

2 2R, 3R

(–)-epigallocatechin (EGC)

OH OH /

3 2R, 3R

(–)-epicatechin (EC)

OH H

4 2R, 3R

(–)-epigallocatechin-3-gallate (EGCG)

GA OH H

5 2S, 3R

(–)-gallocatechin-3-gallate (GCG)

GA OH H

6 2R, 3R

(–)-epigallocatechin-3-(3″-O-methyl)-gallate GA OH CH3

/

(3″-O-methyl-EGCG) 7 2R, 3R

(–)-epicatechin-3-gallate (ECG)

GA H

H

8 2R, 3R

(–)-epicatechin-3-(3″-O-methyl)-gallate

GA H

CH3

(3″-O-methyl-ECG) Fig. 4. Chemical structures of eight investigated catechins.

X. Deng et al. / J. Chromatogr. B 973 (2014) 55–60

59

Table 1 Retention time, UV and MS characteristics of six COX-2 binders fished out from green tea. Peak

tR (min)

max (nm)

[M−H]− ( ppm)

Molecular formula (neutral form)

Structure

1 2 3 4 5 6 7 8

11.94 22.23 30.83 32.17 35.50 42.34 44.91 55.37

246, 273 246, 273 246, 278 241, 276 241, 276 241, 276 241, 277 241, 277

305.0657 (−1.3) 305.0659 (−0.7) 289.0716 (1.4) 457.0778 (1.5) 457.0777 (1.3) 471.0934 (1.5) 441.0820 (−0.5) 455.0983 (1.1)

C15 H14 O7 C15 H14 O7 C15 H14 O6 C22 H18 O11 C22 H18 O11 C23 H20 O11 C22 H18 O10 C23 H20 O10

GC EGC EC EGCG GCG 3 -O-methyl-EGCG ECG 3 -O-methyl-ECG

1 and 2 gave the same deprotonated molecule ([M−H]− with m/z 305, C15 H13 O7 , consistent to GC or EGC) and fragment ions (m/z at 125, C6 H5 O3 , corresponded to unsubstituted A-ring; m/z at 179, C9 H7 O4 , elimination of B-ring [44]), which indicated that they are stereoisomers and could not be distinguished by MS spectra. Similarly, compounds 4 and 5 were isomeric compounds with the same molecular ion and fragment ions with those of EGCG. The m/z of the [M−H]− ion of 3 was 16 Da less than that of 1, which suggested the difference between 1 and 3 was one hydroxyl group. Compound 7 showed an [M−H]− ion with m/z 441.0820 (C22 H17 O10 , −0.5 ppm error), which gave characteristic fragment ions at m/z 289.0714 (C15 H13 O6 , [M−H−C7 H4 O4 ]− , the same with 3) and 169.0139 (C7 H5 O5 the existence of galloyl group). Therefore, compound 7 contained one more galloyl group than 3. Finally, compounds 1–5 and 7 were unequivocally identified as GC, EGC, EC, EGCG, GCG, and ECG, based on comparison of their retention time, UV, accurate molecular weight and MS fragmentation pattern with reference compounds (Table 1). Compounds 6 and 8 exhibited [M−H]− ion at m/z 471.0934 (C23 H19 O11 , 14 Da greater than that of EGCG) and 455.0983 (C23 H19 O10 , 14 Da greater than that of ECG), respectively, indicating that compounds 6 and 8 were methylated EGCG and methylated ECG. The characteristic fragment ion with m/z 183.0291 (C8 H7 O5 ) displayed that methyl group was situated at the galloyl group, however, the position of methyl group was unclear. Further, the structural verification of compounds 6 and 8 were confirmed by 1 H NMR data after repetitive chromatography on open silica gel columns eluting with chloroform–methanol system. 1 H NMR data of 6 and 8 were very similar to that of EGCG and ECG except for two proton signals (ı 7.04 for H-2 , 7.17 for H-2 ) for galloyl group and a methoxy signal (ı 3.79). Therefore, compounds 6 and 8 were characterized as (−)-epigallocatechin-3(3 -O-methyl)-gallate (3 -O-methyl-EGCG) and (−)-epicatechin-3(3 -O-methyl)-gallate (3 -O-methyl-ECG), respectively [45].

4. Conclusion A new assay based on magnetic Fe3 O4 @SiO2 –COX-2 ligand fishing combination with HPLC–DAD–MSn analysis were developed to rapidly screen and characterize COX-2 inhibitors from complex matrix. The proposed method was verified by COX-2 selective inhibitor, celecoxib, which indicated the method could successfully fish out COX-2 ligands. Another particularly noteworthy advantage of the developed method was that immobilized COX-2 was stable, which could continuously perform multiple assays with relatively low cost and high throughput screening. Eight catechins with COX2 binding activities were screened and characterized from green tea, which were consistent with the in vitro enzyme assay. This newly developed method could definitely accelerate the discovery of active compounds from natural products. In addition, the high efficiency and specificity of the proposed method and the wide range of immobilization macromolecular targets (enzyme, protein, receptor, etc.) provided convenient conditions for screening of a broad range of bioactive components from complex natural products. Acknowledgements Funding for this project was supported by National Natural Science Foundation of China (81241009), and Hunan Provincial Innovation Foundation for Postgraduate, China. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jchromb. 2014.10.010. References

3.6. COX-2 inhibitory activity Catechins were well known bioactive components with highest yield in tea leaves and possessed many different biological activities [46]. The COX-2 inhibitory of them was highly depended on the galloylation in the molecule. Seeram and coworkers had investigated COX-2 inhibitory activity of catechin in detail [18], and they concluded that cis-trans isomerization and epimerization did not affect their COX-2 inhibitory activity significantly, but the galloyl derivative showed greater inhibition to COX-2. 3 -O-methyl-EGCG (6) and 3 -O-methylECG (8) were rare catechins, which had been previously isolated from tea leaves [41]. However, to the best of our knowledge, the COX-2 inhibitory activities of them were not reported. Our in vitro assay showed that 3 -O-methyl-EGCG and 3 -O-methyl-ECG exhibited potent COX-2 inhibitory activities with IC50 values at 0.17 ± 0.03 and 0.16 ± 0.02 ␮M, respectively, a little high than that of celecoxib (IC50 value at 0.11 ± 0.02 ␮M).

[1] Y.T. Lin, Y.H. Wu, C.K. Tseng, C.K. Lin, W.C. Chen, Y.C. Hsu, J.C. Lee, PLoS ONE 8 (2003) 1–10. [2] X.H. Liu, S. Yao, A. Kirschenbaum, A.C. Levine, Cancer Res. 58 (1998) 4245–4249. [3] G. Steinbach, P.M. Lynch, R.K.S. Phillips, M.H. Wallace, E. Hawk, G.B. Gordon, N. Wakabayashi, B. Saunders, Y. Shen, T. Fujimura, L.K. Su, B. Levin, L. Godio, S. Patterson, M.A. Rodriguez-Bigas, S.L. Jester, K.L. King, M. Schumacher, J. Abbruzzese, R. DuBois, W.N. Hittelman, S. Zimmerman, J.W. Sherman, G. Kelloff, N. Engl. J. Med. 342 (2000) 1946–1952. [4] A.L. Hsu, T.T. Ching, D.S. Wang, X.Q. Song, V.M. Rangnekar, C.S. Chen, J. Biol. Chem. 275 (2000) 11397–11403. [5] R.A. Gupta, R.N. DuBois, Nat. Rev. Cancer 1 (2001) 1–21. [6] S. Bansal, M. Bala, S.K. Suthar, S. Choudhary, S. Bhattacharya, V. Bhardwaj, S. Singla, A. Joseph, Eur. J. Med. Chem. 80 (2014) 167–174. [7] A.K. Tewari, V.P. Singh, P. Yadav, G. Gupta, A. Singh, R.K. Goel, P. Shinde, C.G. Mohan, Bioorg. Chem. 56 (2014) 8–15. [8] J. Grover, V. Kumar, V. Singh, K. Bairwa, M.E. Sobhia, S.M. Jachak, Eur. J. Med. Chem. 80 (2014) 47–56. [9] J.J. Talley, D.L. Brown, J.S. Carter, M.J. Graneto, C.M. Koboldt, J.L. Masferrer, W.E. Perkins, R.S. Rogers, A.F. Shaffer, Y.Y. Zhang, B.S. Zweifel, K. Seibert, J. Med. Chem. 43 (2000) 775–777. [10] J.M. Dogne, C.T. Supurran, D. Pratico, J. Med. Chem. 48 (2005) 2251–2257. [11] D.J. Newman, G.M. Cragg, J. Nat. Prod. 75 (2012) 311–335. [12] M.A. Ramirea-Cisneros, M.Y. Rios, R. Rios-Gomez, A.B. Aguilar-Guadarrama, Planta Med. 78 (2012) 1942–1948.

60

X. Deng et al. / J. Chromatogr. B 973 (2014) 55–60

[13] H.B. Zhu, S. Liu, X. Li, F.R. Song, Z.Q. Liu, S.Y. Liu, Anal. Bioanal. Chem. 405 (2013) 7437–7445. [14] W. Maldonado-Rojas, J. Olivero-Verbel, J. Mol. Graph. Model. 30 (2011) 157–166. [15] T.C. Reddy, P. Aparoy, N.K. Babu, K.A. Kumar, S.K. Kalangi, P. Reddanna, Protein Peptide Lett. 17 (2010) 1251–1257. [16] C.H. Hong, S.K. Hur, O.J. Oh, S.S. Kim, K.A. Nam, S.K. Lee, J. Ethnopharmacol. 83 (2002) 153–159. [17] M. Dumarey, I. Smets, Y.V. Heyden, J. Chromatogr. B 878 (2010) 2733–2740. [18] N.P. Seeram, Y.J. Zhang, M.G. Nair, Nutr. Cancer 46 (2003) 101–106. [19] A.B. Sharangi, Food Res. Int. 42 (2009) 529–535. [20] S.Y. Shi, H.H. Zhou, Y.P. Zhang, X.Y. Jiang, X.Q. Chen, K.L. Huang, Trends Anal. Chem. 28 (2009) 865–877. [21] S.L. Su, L. Yu, Y.Q. Hua, J.A. Duan, H.S. Deng, Y.P. Tang, Y. Lu, A.W. Ding, Biomed. Chromatogr. 22 (2008) 1385–1392. [22] S.Y. Shi, M.J. Peng, Y.P. Zhang, S. Peng, Anal. Bioanal. Chem. 405 (2013) 4213–4223. ˜ [23] M.A. Martínez-Gómez, L. Escuder-Gilabert, R.M. Villanueva-Camanas, S. Sagrado, M.J. Medina-Hernández, J. Chromatogr. B 889–890 (2012) 87–94. [24] B.C. Wang, J. Deng, Y.M. Gao, L.C. Zhu, R. He, Y.Q. Xu, Fitoterapia 82 (2011) 1141–1151. [25] Q. Shang, J.F. Xiang, X.F. Zhang, H.X. Sun, L. Li, Y.L. Tang, Talanta 85 (2011) 820–823. [26] L.L. Liu, S.Y. Shi, X.Q. Chen, M.J. Peng, J. Chromatogr. B 932 (2013) 19–25. [27] C.F. Cho, G.A. Amadei, D. Breadner, L.G. Luyt, J.D. Lewis, Nano Lett. 12 (2012) 5957–5965. [28] N.D. Shapiro, S. Soh, K.A. Mirica, G.M. Whitesides, Anal. Chem. 84 (2012) 6166–6172. [29] K. Ashtari, K. Khajeh, J. Fasihi, P. Ashtari, A. Ramazani, H. Vali, Int. J. Biol. Macromol. 50 (2012) 1063–1069.

[30] N. Jonker, A. Kretschmer, J. Kool, A. Fernandez, D. Kloos, J.G. Krabbe, H. Lingeman, H. Irth, Anal. Chem. 81 (2009) 4263–4270. [31] R. Moaddel, M.P. Marszall, F. Bighi, Q. Yang, X. Duan, I.W. Wainer, Anal. Chem. 79 (2007) 5414–5417. [32] K.L. Vanzolini, Z.J. Jiang, X.Q. Zhang, L.C.C. Vieira, A.G. Corrêa, C.L. Cardoso, Q.B. Cass, R. Moaddel, Talanta 116 (2013) 647–652. [33] L. Pochet, F. Heus, N. Jonker, H. Lingeman, A.B. Smit, W.M.A. Niessen, J. Kool, J. Chromatogr. B 879 (2011) 1781–1788. [34] Y.F. Li, Y. Chen, C.Y. Xiao, D. Chen, Y.X. Xiao, Z.N. Mei, J. Chromatogr. B 960 (2014) 166–173. [35] L.L. Liu, S.Y. Shi, H.D. Zhao, J.G. Yu, X.Y. Jiang, X.Q. Chen, J. Chromatogr. B 945–946 (2014) 163–170. [36] Y.P. Zhang, S.Y. Shi, X.Q. Chen, M.J. Peng, J. Chromatogr. B 960 (2014) 126–132. [37] Y. Tao, Z. Chen, Y.F. Zhang, Y. Wang, Y.Y. Cheng, J. Pharm. Biomed. Anal. 78–79 (2013) 190–201. [38] M. Yasuda, D.R. Wilson, S.D. Fugmann, R. Moaddel, Anal. Chem. 83 (2011) 7400–7407. [39] N. Singh, S. Ravichandran, K. Spelman, S.D. Fugmann, R. Moaddel, J. Chromatogr. B 968 (2014) 105–111. [40] H.M. Cao, R. Yu, Y. Tao, D. Nikolic, R.B. van Breemen, J. Pharm. Biomed. Anal. 54 (2011) 230–235. [41] D.M. Wang, J.L. Lu, A.Q. Miao, Z.Y. Xie, D.P. Yang, J. Food Compos. Anal. 21 (2008) 361–369. [42] X. Chen, L.X. Li, S. Chen, Y.T. Xu, Q. Xia, Y. Guo, X. Liu, Y.T. Tang, T.J. Zhang, Y. Chen, C. Yang, W.Q. Shui, Anal. Chem. 85 (2013) 7957–7965. [43] G.Q. Liu, J. Dong, H. Wang, L.R. Wan, Y.S. Duan, S.Z. Chen, J. Chin. Mass Spectrom. Soc. 30 (2009) 287–294. [44] J.P. Sun, F. Liang, Y. Bin, P. Li, C.Q. Duan, Molecules 12 (2007) 679–693. [45] K. Kida, M. Suzuki, N. Matsumoto, F. Nanjo, Y. Hara, J. Agric. Food Chem. 48 (2000) 4151–4155. [46] V. Crespy, G. Williamson, J. Nutr. 134 (2004) 3431S–3440S.