A LONELY GUY protein of Bordetella pertussis with unique features is related to oxidative stress

The Gram-negative bacterium B. pertussis is the causative agent of whooping cough. This infection is re-emerging and new features related to Bordetella pathogenesis and microbiology could be relevant to defeat it. Therefore, we focused our attention on BP1253, a predicted exported protein from B. pertussis erroneously classified as lysine decarboxylase. We showed that BP1253 shares the highly conserved motif PGGxGTxxE and the key catalytic amino-acid residues with newly structurally characterized “LONELY GUY” (LOG) proteins. Biochemical studies have confirmed that this protein functions as a cytokinin-activating enzyme since it cleaves the N-glycosidic linkage between the base and the ribose, leading to the formation of free bases, which are the active form of plant hormones called cytokinins. Remarkably, BP1253 selectively binds monophosphate nucleotides such as AMP, GMP and CMP, showing a wider variety in binding capacity compared to other LOGs. Cytokinin production studies performed with B. pertussis have revealed 6-O-methylguanine to be the physiological product of BP1253 in agreement with the higher activity of the enzyme towards GMP. 6-O-methylguanine is likely to be responsible for the increased sensitivity of B. pertussis to oxidative stress. Although BP1253 has a primary sequence resembling the hexameric type-II LOGs, the dimeric state and the presence of specific amino-acids suggests that BP1253 can be classified as a novel type-II LOG. The discovery of a LOG along with its product 6-O-methylguanine in the human pathogen B. pertussis may lead to the discovery of unexplored functions of LOGs, broadening their role beyond plants.


Results
BP1253 is a homolog of the plant cytokinin-activating enzyme LOG. BP1253, along with many proteins from different organisms of various kingdoms, like Plantae, Animalia, Bacteria and Fungi, was miss-annotated as lysine decarboxylase (LDC) 17 . A more accurate in silico analysis has revealed that BP1253 shares with LOG proteins not only the highly conserved motif PGGxGTxxE, but also the Arg and Glu residues of the catalytic core (Fig. 1). BP1253 shows a higher sequence homology with the type-II LOGs, which form hexamers, as compared to type-I LOGs, which form dimers 14,18 . In order to characterize BP1253, we expressed it as a His-tagged www.nature.com/scientificreports www.nature.com/scientificreports/ recombinant protein ( Supplementary Fig. S1). The expression of BP1253 in Tohama I and knock-out Tohama I Δ1253 strain is also shown ( Supplementary Fig. S1). The size exclusion chromatography (SEC) of BP1253 showed a main peak of 50 kDa ( Supplementary Fig. S2), while the multi-angle light scattering (MALS) analysis showed a protein with a molecular weight of 49.32 (±5%) kDa with a polydispersity of 1.010 (±7.7%; Supplementary  Fig. S2). Both methods showed that BP1253 is a dimer. The position of BP1253 in the LOG family phylogenetic tree between proteins classified as type-I or type-II suggests that BP1253 could represent a novel type-II group of LOGs (Fig. 2).
Nucleotide binding activity of BP1253. To get insight into the ability of BP1253 to bind AMP, which is the substrate of LOGs commonly used in in vitro assays, we first resorted to a theoretical approach. This involved overlapping a model of BP1253 with a monomer of the type-I LOG protein of M. marinum co-crystallized with AMP (MmLOG, PDB 3SBX) 14,19 . BP1253 was modeled on a monomer of the type-II LOG protein of T. thermophilus TT1465 (TtLOG), which shares 48% of amino-acid identity with BP1253 (personal observation). The overlap between 3SBX and BP1253 produced a Root Mean Square (RMS) value of 1.21 Å, showing a great similarity between the binding pockets ( Supplementary Fig. S3). Most importantly, surface plasmon resonance (SPR) confirmed experimentally the binding of AMP to BP1253 with a K D of 5.7 µM. Interestingly, BP1253 also binds GMP and IMP with a K D of 38.6 and 71.8 µM, respectively (Fig. 3), and TMP and UMP with K D in the millimolar range (2.4 and 6.1 mM, respectively, data not shown). No binding was identified with CMP. Overall, these data show a wide binding capacity for BP1253 to purine and pyrimidine monophosphate nucleotides.
BP1253 is a LOG with phosphoribohydrolase activity. The high similarity in amino acid sequence and the SPR analysis led us to hypothesize that BP1253 functions as a LOG protein. However, since BP1253 was previously annotated as LDC, we first performed a specific assay for LDC activity 20 . No LDC activity was detected (data not shown) showing mis-annotation of this protein. We then quantified PRH, an activity only recently demonstrated in human pathogens 15,16 , which leads to the formation of free and active cytokinins after cleavage of the N-glycosidic linkage between the N 6 -modified adenine and ribose 5'-monophosphate (Fig. 4). The cleavage of AMP was dose-and time-dependent (Fig. 4). SPR results suggested carrying out the enzymatic assay also with GMP as substrate. As shown in Fig. 4, a ten-fold less concentrated BP1253 cleaved GMP almost completely after 30 minutes of incubation, indicating GMP as a more efficient substrate than AMP. A high PRH activity was recorded also with CMP (Fig. 4). This result is in apparent contrast with the undetectable binding of CMP to BP1253 by SPR and could be due to the rapid cleavage of CMP by BP1253. Subsequently, the involvement of R120 and E143 in enzyme catalysis and of K121 in substrate binding was confirmed by using BP1253 mutants ( Supplementary Fig. S4). Substitutions of R120 and E143 with conservative residues resulted in a total absence of activity (Fig. 5), while the replacement of K with R at the residue 121 (K121R) produced a minimal hydrolysis both with CMP and GMP, not comparable to the wild type (Fig. 5). The results obtained from activity and mutagenesis studies strongly support BP1253 as being a B. pertussis ortholog of plant LOG and, henceforth, we will refer to it as BpLOG.
Lc-MS analysis of B. pertussis citokinins. Since BpLOG is a member of the LOG family, in analogy with its homolog proteins it should be able to produce cytokinins. Cytokinins are plant hormones, which are produced in small quantities and released in the extracellular space. In order to verify the biological activity of BpLOG within the context of live bacteria, we generated a knock-out Tohama I Δ1253 strain and, using the LC-MS/MS approach, analyzed the cytokinins in the culture supernatants and lysates of either the wild-type or knock-out strain. We first investigated the presence of canonical cytokinins and identified the presence of isopentenyl adenine and kinetin in the bacterial lysates of both wild-type Tohama I and Δ1253 knock-out strain (Fig. 6). Further analyses revealed a peak with a retention time of 5 minutes in the supernatant of wild-type Tohama I, but not of Δ1253 mutant strain (Fig. 7). Identical result was also obtained with bacterial lysates (data not shown). This  www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ peak, with an apparent molecular formula of C 6 H 7 N 5 O, corresponded to a compound with a molecular weight of 165 Da, which appeared in the chromatogram with a mass of 166 Da due to the positive ion mode method used for detection. In the data bank the best fitting compound with this formula was 6-O-methylguanine. Indeed, the retention time of the synthetic 6-O-methylguanine was identical to the one isolated in Bp (Fig. 7). In addition, the fragmentation mass spectra of the isolated physiological product and of the synthetic 6-O-methylguanine were identical, showing that BpLOG synthesizes the 6-O-methylguanine as a physiological product (Fig. 7). Notably, we identified 6-O-methylguanine in bacterial lysates and in supernatants of the clinical Bp isolates B3629 and B3621 6 (data not shown).
BpLoG is negatively related to oxidative stress. In M. tuberculosis the excess of cytokinin breakdown products increased sensitivity to nitric oxide (NO) 15,[21][22][23] . Since BpLOG is a homologous of MtLOG, we tested the susceptibility of Bp to oxidative stress conditions. Wild-type Bp bacteria were more sensitive to oxidative stress conditions as compared to BpLOG-deficient mutant bacteria Δ1253 as shown by the reduced number of viable bacteria after treatment with 100 mM or 300 mM hydrogen peroxide (H 2 O 2 , Fig. 8A). These results suggested that 6-O-methylguanine produced by BpLOG sensitizes Bp to the oxidative stress due to H 2 O 2 . To test this hypothesis we incubated the two Bp strains with 100 mM H 2 O 2 with exogenously added 10 µM 6-O-methylguanine. As shown in Fig. 8B, addition of 6-O-methylguanine together with H 2 O 2 lowered the viability of Δ1253 knock-out strain. Importantly, addition of 6-O-methylguanine alone to either wild-type or Δ1253 mutant strain had no effect on Bp viability (Fig. 8B). Overall, these results indicated a role of BpLOG in sensitizing Bp to oxidative stress.

Discussion
The resurgence of the pertussis infection is a serious public health problem. Waning immunity induced by the aP vaccine and lack of herd immunity can lead to Bp strains diffusion. Therefore, it is necessary to increase our basic knowledge on Bp microbiology in order to find new therapeutic/vaccine strategies and/or to improve the existing ones. In this study, combining in silico analyses and in vitro experiments, we demonstrate that BP1253 is a novel member of the LOG family since it contains the highly conserved LOG motif PGGxGTxxE, the catalytic amino acids Arg and Glu, and possesses PRH activity. Through this activity, LOG proteins synthesize cytokinins. Although BpLOG has a primary sequence similar to hexameric type-II LOGs, the dimeric state and the presence www.nature.com/scientificreports www.nature.com/scientificreports/ of specific amino-acids strongly support the hypothesis that BpLOG could represent a novel class of type-II LOGs. Noticeable differences are the presence of Asp at position 142 occupied by Glu in the majority of type-I LOGs, and Phe at position 117. Another noteworthy difference is at position 118, which is occupied by Tyr in BpLOG and by Phe and His in the other LOGs. The presence of different residues all involved in the formation of the prenyl-group binding site could explain the capability of this protein to bind and hydrolyze more effectively molecules like GMP and CMP. These peculiarities that make BpLOG unique, are mirrored by the production of the physiological product. In fact, BpLOG produces 6-O-methylguanine, a guanine-derivative with a methyl-group at O 6 , while the cytokinins described so far are adenine-derivatives modified at N 6 . Indeed, synthetic O 6 -substituted guanine derivatives have been described as acting like cytokinins, inducing cell division in plants 24 . Remarkably, the synthesis of the 6-O-methylguanine is in agreement with the higher PRH activity of BpLOG with GMP. The flexibility of BpLOG towards other nucleotides suggests that BpLOG could synthesize also other non-canonical cytokinins containing guanine or cytosine as nitrogenous bases. The production of modified adenine cytokinins by Bp is probably due to the presence of BP0547, which shares an amino-acid identity of 54% with the characterized type-I P. aeruginosa LOG 18 and 39% with MtLOG, which are both able to synthesize adenine-derivate cytokinins 15 , but only 23% with BpLOG. While prediction studies indicate that BP0547 could be localized in the cytoplasm, BpLOG could be located in the periplasmic space or on the external membrane due to the presence of a putative trans-membrane helix (Fig. S5). The different cellular localization of the two enzymes could partially explain the low amount of 6-O-methylguanine found in bacterial lysates (not shown), while the inability to detect www.nature.com/scientificreports www.nature.com/scientificreports/ adenine-derivate cytokinins in the supernatant of wild-type Bp was probably due to the low amount secreted. Our results show that BpLOG is responsible for the synthesis of 6-O-methylguanine while they do not exclude that BpLOG is involved in the biochemical pathway of canonical cytokinins. This discovery has a strong impact, since 6-O-methylguanine could represent a novel class of molecules with cytokinin activity.
The capability of BpLOG to recognize and metabolize several nucleotides prompts the hypothesis that BpLOG could protect Bp by degrading non-canonical nucleotides present in the external environment, which can interfere with the physiology of the bacterium. Indeed, in S. cerevisiae the over-expression of the analog LOG1 (YJL055W) gene conferred resistance to the chemotherapeutic drugs 6-N-hydroxylaminopurine (HAP) and 5-fluorouracil (5-FU), due to the detoxification properties of the LOG protein encoded by this gene 25 .
The function of cytokinins in Bp and in other human pathogens has just started to emerge. In plants, cytokinins are hormones that act locally or distant from the cell at very low concentrations 26 , modulating plant metabolism and morphogenesis in response to environmental factors 27 . 6-O-methylguanine could be a signaling molecule for inter-bacteria communication and/or a way to modulate the expression of specific genes helping the bacterium to establish a successful infection. Recently, it was reported that in M. tuberculosis isopentenyl adenine regulates the expression of genes associated with the bacterial envelope and virulence 28 . The enhancing effects of 6-O-methylguanine in combination with ROS, produced by lung epithelial cells 29 or by resident microbes 30 , could produce compounds toxic for epithelial cells to enhance colonization and/or favor Bp towards other bacterial species competing for the same niche. Indeed, Bp could be protected from the bacterial clearance associated with an increased level of ROS 29 by the cytokinin kinetin produced by BP0547, which reduces generation of ROS and induces a raise in the activity of the antioxidant protection systems [31][32][33][34] . The latter confirms a relation of LOGs and their products with oxidative stress in agreement with M. tuberculosis, where the over expression of MtLOG led to an accumulation of cytokinin degradation products which, interacting with NO, are toxic for the bacteria 15 . However, further investigation is required to clarify the in vivo role of BpLOG.
Overall, we characterized BP1253 as a LOG of the human-exclusive pathogen Bp with peculiar features. The discovery of this new LOG in Bp suggests that LOGs may play an important role in the pathogenesis of human bacterial diseases. Furthermore, a better understanding of the role of LOGs in bacterial infections could lead to the rational design of novel therapeutic strategies that are urgently needed due to the emerging concerns about antibiotic resistance.

Methods
Materials. Isopropyl-1-thio-β-D-galactopyranoside was obtained from Calbiochem; the BCA reagent for protein quantification and the ECL immunoblotting detection system were from Bio-Rad, the protein inhibitor cocktail from Roche, and standard Bovine Serum Albumin (BSA) was obtained from Pierce. SimplyBlue SafeStain, Nu-PAGE gels and iBlot membranes were purchased from Invitrogen. All other reagents used in this study were from Sigma Aldrich. Bioinformatics analysis. The BLAST algorithm was used to investigate the sequence homologies between BP1253 and LOGs, while Clustal omega was used to realize amino acid sequence alignments with specific type-I and type-II LOGs. The structural analysis was performed by means of PDB Viewer and to create the figure and represent the sequence alignment ESPript 3.0 software was used. The graphic software PyMOL was used to visualize the overlaps, while the phylogenetic tree was generated with MAFFT and visualized through the software tool Archaeopteryx.js. www.nature.com/scientificreports www.nature.com/scientificreports/ Bacterial strains and growth conditions. The following B. pertussis strains were used in this study: Tohama I-derivative BP536 35 , the clinical isolates BP3629 and BP3621 6 . Bacteria were stored at −80 °C and recovered by plating on Bordet-Gengou (BG) agar plates, supplemented with 15% (v/v) sheep blood for 3 days at 37 °C. The bacteria were then inoculated at an initial 600 nm optical density (A 600 ) of 0.05-0.1 in Stainer-Scholte medium supplemented with 0.4% (w/v) L-cysteine monohydrochloride, 0.1% (w/v) FeSO 4 , 0.2% (w/v) ascorbic acid, 0.04% (w/v) nicotinic acid, and 1% (w/v) reduced glutathione. Cultures were grown in rotary shakers at 37 °C.
Generation of E. coli expressing recombinant BP1253 protein and purification procedure. The synthetic gene1253 from Bp pertussis was synthesized and assembled by GeneArt (Thermofisher) starting from synthetic oligonucleotides and/or PCR products. The fragment was inserted into pET24b(+)_D455 as a fusion construct with a carboxyl-terminal 6x histidine tag separated from the last amino acid of the protein by a linker of 2 amino-acid residues. The sequence congruence verified by sequencing of the final construct within the insertion sites was 100%. E. coli strain BL21(DE3) cells (England BioLabs) were transformed with the above constructs and used for protein expression. Cells were grown using BioSilta medium (Enpresso B Animal-free growth systems), at 30 °C for 8 h with gentle shaking (160 rpm). The expression of Bp1253 gene was induced to an A 600 of 0.5-0.6 by the addition of 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) 24 h at 17 °C. Cells were harvested by centrifugation and re-suspended in lysis buffer containing 50 mM NaH 2 PO 4 , 300 mM NaCl, pH 7.4 and EDTA-free protease inhibitor (Boheringer Mannheim) at a ratio of 10 ml of lysis buffer per 1 g of bacterial pellet. After lysis, performed via sonication (Qsonica Q700), cell lysates were clarified by centrifugation at 15000 x g for 50 min at 4 °C, and the supernatant, containing the expressed protein, was filtered using 0.22 μm membrane filters (EMD Millipore filters) before starting the first chromatography step. BP1253 was purified by Co 2+ -affinity chromatography (5 mL HiTrap TALON crude, GE Healthcare) at 25 °C using an AKTA purifier 100 system (GE Healthcare). The column was equilibrated with buffer A (50 mM Tris pH 8, 300 mM NaCl). After loading the crude extract, the column was washed with 10 bed volumes of buffer A. Bound proteins were eluted with buffer A containing 500 mM imidazole. The content of lipopolysaccharide (LPS) on the purified protein was checked using the Endosafe nexgen-PTS system (Charles River). When the content of LPS was out of the range, it was removed using the EndoTrap Red columns (Hyglos). The purity of the protein was checked using 4-12% SDS-PAGE gradient gels in MES buffer, after identification of the fractions containing BP1253 samples were pooled and stored at −20 °C for subsequent analysis.
Size-exclusion chromatographic analysis. The investigation of BP1253 oligomerization was performed using analytical size exclusion chromatography. The chromatographic step was performed using a BEH200 column 4.6 × 300 mm (Waters) at a flow of 0.4 ml/min with a buffer containing 10 mM NaH 2 PO 4 and 400 mM (NH 4 ) 2 SO 4 pH 6.0. Protein samples of 15 μl, 0.62 mg/ml and 3 mg/ml were analyzed. The molecular weights of the different forms of BP1253 were calculated from a calibration curve based on standard proteins.
Generation of the BP1253 Knockout strain. The deletion strain for the genes BP1251-1252-1253 of the B. pertussis BP536 Tohama I-derivative was constructed as follows. The 5′ and 3′ extremities of the locus were amplified by PCR using Bp chromosomal DNA as template and the oligonucleotides FlankingUP-Fw ccgGAATTCCGAAAACCGTAGCGGTCGAA and FlankingUP-rev ggaGGATCCGGACCGATGTCGGCCAATTT, FlankingDOWN-Fw ggaGGATCCCGCGTCTATGTCGACCACG and FlankingDOWN-rev cccAAGCTTCGAACTGCACCTGACCATCC as primers, respectively. The amplicons were then successively introduced as EcoRI-BamHI and BamHI-HindIII fragments into pUC19, together with a BamHI-BamHI fragment encoding the kanamycin resistance cassette for selection. The resulting EcoRI-HindIII fragment was then purified and introduced into the EcoRI-HindIII sites of pSORTP1, a mobilizable suicide plasmid used for conjugation between Bp and E. coli. Conjugation was performed on BG-blood agar plates containing 10 mg/ml MgCl 2 for 5 hours, and co-integrates were selected on BG-blood agar plates containing 10 µg/ml gentamycin and 20 µg/ml nalidixic acid to prevent growth of the E. coli donor. Single crossing overs determine the insertion of the pSORTP1 vector in the chromosome and confer gentamycin resistance and streptomycin sensitivity. Double crossing overs were selected by a successive step on BH-blood plates containing 25 µg/ml kanamycin and 400 µg/ml streptomycin. After 3 to 4 days growth on selective media, isolated kanamycin and streptomycin-resistant colonies were analyzed by PCR to confirm the deletion.
High-Throughput purification of BP1253 mutants. Single point conservative and non-conservative mutations were inserted into the Bp1253 gene to generate six different mutants. Mutated genes were synthesized and assembled by GeneArt (Thermofisher) starting from synthetic oligonucleotides and/or PCR products, using the same plasmid and procedure as described above. The mutants generated were R120A, R120K, K121A, K121R, E143A, and E143D. Purification of mutants was performed under vacuum conditions in a 96-well Vacuum plate. Cells were lysed, using B-PER (Sigma) and applied on a His Multitrap HP 50 ml NiSepharose High Performance 96 wells previously washed with water and equilibration buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , pH 8). After loading the samples, the plate was washed with 80 vol. of washing buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 20 mM imidazole, pH 8) at 25 °C. The His-fusion proteins were eluted in two steps by addition of 6 vol. of elution buffer (300 mM NaCl, 50 mM NaH 2 PO 4 , 500 mM imidazole, pH 8) twice. All the elutions related to the same protein were subsequently pooled. All purification steps were carried out applying a vacuum not exceeding the maximum pressure of 5 mmHg. www.nature.com/scientificreports www.nature.com/scientificreports/ of the mixture onto a 4-12% gradient NuPAGE Bis-Tris gel (Life Technologies). SeeBlue Plus Prestained Standard markers were run on each gel, which was stained with Comassie Blue to visualize the proteins. Separated proteins were electro-transferred onto nitrocellulose membranes with iBlot 2 Dry Blotting System (Life Technologies). The membranes were blocked for 60 min with 0.1% Tween 20 and 10% milk in phosphate buffer solution (PBS), and then incubated for 1 h with specific mouse polyclonal α-BP1253 antibodies (1:500 dilution) in 0.1% Tween 20 and 3% milk buffer in PBS. After three washes with 0.1% Tween 20 in PBS (T-PBS) the membrane was incubated with a secondary rabbit α-mouse horseradish peroxidase conjugated antibody (Jakson Immune research Laboratory, 1:1000 dilution) for 30 min at room temperature. Bound antibodies were visualized, after washing membranes three times with T-PBS, using the ECL immunoblotting detection system (Bio-Rad) according to the manufacturer's instructions. phosphoribohydrolase activity assay. Phosphoribohydrolase activity was assessed by detecting the adenine ring compounds, after separation by thin layer chromatography (TLC) as described 36 . Briefly, enzyme reaction was carried out in a mixture containing, in a final volume of 20 μl, 40 mM Tris-HCl (pH 8), 20 mM monophosphate nucleotide as substrate and the amount of purified recombinant protein as specified in figure legends. After incubation at 30 °C, samples (5 μl) were denatured at 95 °C for 5 min and then 1 μl dotted on PEI-cellulose-F-plastic TLC sheet (Merck Millipore). Since the solubility of guanine is poor in water and to enhance its visibility on TLC, inactivation was performed with 1 M NaOH (v/v). The nitrogenous base ring was separated from the phosphoribose moiety using a TLC method with a mobile phase containing 1 M sodium chloride, except for pyrimidine rings where the mobile phase was acetone/water 30/70 (v/v) with the addition of 0.2 mol/l NaCl. After the run the sheet was completely dried and the separated dots visualized by means of a UV lamp (264 nm).

SDS
Generation of mouse immune sera. BALB/c mice (10 female/group, 6-week old) (Charles River Laboratories International Inc., Wilmington, MA) received three intraperitoneal immunizations, with a 4-week interval, with aluminum hydroxide (2 mg/ml) adjuvanted recombinant BP1253-His (10 μg per dose) at one fifth of a human dose. Sera were collected before immunization and 2 weeks after the third immunization. Control mice immunized with adjuvant only were included in the experiments. All the experiments involving animals were performed in compliance with Italian law, with the approval of the local Animal Welfare Body (AWB 2014/06) followed by authorization of Italian Ministry of Health.

Identification and characterization of cytokinins by LC-MS/MS.
For the analysis of the cytokinins, the supernatant was obtained by spinning down the bacterial growth broth at 9000 x g for 20 min. and filtered and treated as described 37 . Briefly, the supernatant was cleared by molecules with MW higher than 3 kDa using a protein concentrator 3 K (Amicon), following the vendor's instructions. Subsequently, the samples were concentrated down to 1 ml volume through a Speed Vac and loaded on a solid phase extraction (SPE) column (MLX matrix, Oasis). Before loading, samples were previously acidified to pH 3 with 98% formic acid. The column was then washed with 0.5 mL SPE load solvent (1 M formic acid), followed by 1 mL water. The samples of interest were eluted with 0.5 mL of freshly prepared solvent 2 (0.35 M ammonium hydroxide in 70% methanol, to 70 mL methanol there was the addition of 2.5 mL of 25% ammonium hydroxide filled to 100 mL with distilled water), the flow-through collected into a new 2 mL microcentrifuge tube was dried in SpeedVac at 10 mBar and 40 °C. The dried fractions were stored at −20 °C until LC-MS analysis. At the time of starting the analysis in LC-MS the dried samples were dissolved in 100 μL 5% methanol in water and centrifuged at 20000 × g, 4 °C for 20 min. The supernatant was transferred into an autosampler vial and used for the analysis without further modification. LC-MS/ MS analysis was performed using an LTQ-Orbitrap XL coupled with an Ultimate 3000 HPLC system equipped with a reverse phase column Luna C18(2) 100 mm × 2 mm, 3 µm, 100 Å. The mobile phases A and B used for the analysis of the sample were 0.1% formic acid in water and 0.1% formic acid in acetonitrile respectively. The gradient started with 100% of A, maintained for 3 minutes and the phase B increased up to 20% in 12 minutes. Then after two minutes the phase B reached 100% and it was held constant for two minutes. The flow rate was set to 200 µl/min. UV at 268 nm while a positive ion mode with electro spray was used for mass spectrometric detection. The ESI ion source spray voltage was set to 5 kV, the capillary temperature was 300 °C, and tube lens and capillary voltage were 110 and 35 V respectively. The sheath and auxiliary gas (N 2 ) were set to 20 and 5 a.u. The mass spectra and the MS/MS spectra were recorded using the data dependent scan mode at resolution of 30000, while the fragmentation mass spectra were recorded in low resolution with a collision energy of 26 a.u.