Journal of Chromatography A, 1361 (2014) 162–168
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Determination of autoinducer-2 in biological samples by high-performance liquid chromatography with fluorescence detection using pre-column derivatization Xiang-Ning Song a , Hai-Bin Qiu a , Xiang Xiao a , Yuan-Yuan Cheng a , Wen-Wei Li a , Guo-Ping Sheng a,∗ , Xiao-Yan Li b , Han-Qing Yu a,∗ a b
Department of Chemistry, University of Science & Technology of China, Hefei 230026, China Environmental Engineering Research Centre, The University of Hong Kong, Hong Kong, China
a r t i c l e
i n f o
Article history: Received 17 June 2014 Received in revised form 29 July 2014 Accepted 31 July 2014 Available online 11 August 2014 Keywords: Autoinducer-2 Biological samples Detection 2,3-Diaminonaphthalene HPLC–FLD
a b s t r a c t Autoinducer-2 (AI-2), as a small-molecular-weight organic molecule secreted and perceived by various bacteria, enables intra- and inter-species communications. Quantitative determination of AI-2 is essential for exploring the bacterial AI-2-related physiological and biochemical processes. However, current strategies for sensitive detection of AI-2 require sophisticated instruments and complicated procedures. In this work, on the basis of the derivatization of AI-2 with 2,3-diaminonaphthalene, a simple, sensitive and cost-effective high-performance liquid chromatography with fluorescence detector (HPLC–FLD) method is developed for the quantitative detection of AI-2. Under the optimized conditions, this method had a broad linear range of 10–14,000 ng/ml (R2 = 0.9999), and a low detection limit of 1.0 ng/ml. Furthermore, the effectiveness of this approach was further validated through measuring the AI-2 concentrations in the cell-free culture supernatants of both Escherichia coli and Vibrio harveyi. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Bacteria secrete small-molecular-weight molecules, i.e., autoinducers, to communicate with each other, termed as quorum sensing (QS) [1]. QS has been found to regulate biofilm formation, virulence, production of antibiotics, formation of fruiting body, and gene transfer [2,3]. In the past two decades, QS has been extensively studied because of its important roles in the fields of health and environment. Among the autoinducers, autoinducer-2 (AI-2), as one unique autoinducer for enabling both intra- and inter-species communication, is found to be secreted by a variety of bacteria [4]. Many different bacterial species are reported to secrete 4,5-dihydroxy-2,3-pentanedione (DPD), the metabolic precursor of AI-2 [5]. DPD undergoes further rearrangements spontaneously and exists as a mixture of relevant compounds, some of which are recognized by various bacteria [6]. This mechanism enables bacteria
∗ Corresponding authors at: University of Science and Technology of China, Department of Chemistry, Hefei 230026, China. Tel.: +86 551 63607592/+86 551 63607453; fax: +86 551 63601592. E-mail addresses: [email protected] (G.-P. Sheng), [email protected] (H.-Q. Yu). http://dx.doi.org/10.1016/j.chroma.2014.07.103 0021-9673/© 2014 Elsevier B.V. All rights reserved.
to perceive AI-2 derived from DPD secreted by themselves or other bacterial species, and ensures interspecies communication [7]. Because of the crucial roles of AI-2 in QS-related research fields, sensitive and selective detection and quantification of AI-2 in biological samples are greatly desired. However, its various forms and low concentrations found in biological samples make its quantitative analysis difficult. Several methods have been developed to detect and quantify AI-2: e.g., Vibrio harveyi luminescence bioassay [8,9], biosensor based on fluorescence yield change when binding AI-2 to receptor proteins [10], high-performance liquid chromatography with UV detector (HPLC–UV), high-performance liquid chromatography with tandem mass spectrometric detector (HPLC–MS/MS) or gas chromatography–mass spectrometry (GC–MS) using 1,2-phenylenediamine or its derivatives to react with DPD to form the corresponding quinoxaline [11–13]. The V. harveyi bioassay method using the induction of luminescence of BB170 bioreporters is the most common way to detect AI-2 because of its sensitivity over several orders of magnitude. However, this bioassay is a qualitative method and not suitable for quantitative analysis [8,14]. Signal inhibition caused by high-concentration AI-2 is also observed [15] and some assay conditions such as pH, glucose levels and borate concentration interfere with this assay [8,16]. The biosensors based on AI-2 receptor proteins can response to only the fraction of DPD, which has been converted
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
163
to borate, making this assay sensitive to the borate concentration [10]. Besides, its linear ranges is very narrow (∼1–20 M), and the procedures to purify and label the proteins are complicated, time-consuming and expensive. The derivatization method, which employs 1,2-phenylenediamine as the derivatization reagents and detects the corresponding quinoxaline by HPLC-UV and Nuclear Magnetic Resonance (NMR), fails to give any quantitative results [11,17]. The GC–MS method, which also initially undergoes 1,2phenylenediamine derivatization and is followed by the treatment with silylation reagents, has a detection limit (LOD) of 0.7 ng/ml and can meet the AI-2 detection requirements in many cases [13]. However, very complicated procedures, such as two-step derivatization, solid phase extraction and concentration, make it difficult to perform. Recently, an HPLC–MS/MS method based on derivatization with derivatives of 1,2-phenylenediamine has been proposed to detect trace AI-2 in human saliva [12]. However, the derivatization reagents with complicated structures are required and usually need painstaking synthetic procedures, which significantly limit its wide application. In addition, expensive facilities, i.e., HPLC–MS/MS, are required for the determination. In this study, we try to establish a simple, sensitive and costeffective method for the quantitative analysis of AI-2. To this end, 2,3-diaminonaphthalene (DAN), a simple and easily available reagent, was used to react with DPD to form a new fluorescent substance, which has characteristic excitation and emission wavelengths. Thus, HPLC with fluorescence detector (HPLC–FLD) with an excellent selectivity and sensitivity could be developed to detect the AI-2 concentration in complex matrix like the supernatants of Escherichia coli and V. harveyi.
30% solvent B; t = 27 min, 70% solvent A, 30% solvent B. The excitation and emission wavelengths of the fluorescence detector were set at 271 and 503 nm, respectively. The mass spectral of the derivative products was also recorded by LTQ XL orbitrap high resolution mass spectrometry (Thermo Fisher Sci. Inc., USA) fitted with an electrospray source operated in a positive mode to further confirm its structure. The spray voltage was set to 4 kV, capillary temperature to 275 ◦ C. The sheath and auxiliary gas flow rate (both nitrogen) were optimized at 20 and 5 arbitrary units (a.u.). Full MS scans were acquired in the orbitrap analyzer with the resolution set to a value of 60,000.
2. Experimental
2.5. Biological sample preparation
2.1. Chemicals and preparation of standard solutions
E. coli strains MG1655 (wide-type) and DH5␣ (luxS-null) were grown at 37 ◦ C in LB medium supplemented with 0.25% glucose [19]. V. harveyi BB120 (AI-2 positive strain) was cultured in AB medium at 30 ◦ C. During the incubation period, aliquots were withdrawn at regular intervals and OD600 was recorded to determine the cell density. Then, the culture was centrifuged at 13,000 rpm for 5 min and filtered through 0.22 m membranes to remove the cells. The cell-free supernatants were quickly frozen and stored at −80 ◦ C until use or underwent derivatization immediately.
DPD solution (0.3 mg/ml, dissolved in ultrapure water) was purchased from Omm Scientific Inc. (Texas, USA). 2,3Diaminonaphthalene (DAN) was obtained from Aladdin Inc. (Shanghai, China). 4,5-Methylenedioxy-1,2-phenylenediamine (MDB), HPLC-grade formic acid and acetonitrile were purchased from Sigma-Aldrich Inc. (St. Louis, USA). 18.2 M Milli-Q ultrapure water was supplied by Millipore Co. (Billerica, USA). Other chemicals or solvents used were of analytical grade. The working standard solutions with a range of 10–14,000 ng/ml were obtained through diluting the stock solution of DPD. The DAN solution was prepared by dissolving 10 mg DAN into 50 ml 0.1 M HCl. 2.2. Derivatization procedures 400 l standard solution or supernatants after pretreatment were transferred to 2 ml autosampler vials (Agilent Inc., USA) containing an equal volume of DAN solution. The two liquids were thoroughly mixed for 2 min. Then, these samples were incubated in a water bath at 90 ◦ C for 40 min. After the samples became cooled down, they were analyzed by HPLC–FLD directly. 2.3. Chromatographic instrumentation and conditions 20 l samples was injected for analysis using 1260 HPLC system equipped with a fluorescence detector (Agilent Inc., USA). Separation was achieved on an Agilent ZORBAX SB-C18 reverse-phase column (250 mm × 4.6 mm, 5 m) set at 30 ◦ C. The mobile phase contained 0.1% formic acid (solvent A) and acetonitrile (solvent B) at a flow rate of 0.8 ml/min. A gradient elution profile was used as follows: t = 0 min, 70% solvent A, 30% solvent B; t = 4 min, 70% solvent A, 30% solvent B; and t = 12 min, 35% solvent A, 65% solvent B; t = 20 min, 35% solvent A, 65% solvent B; t = 24 min, 70% solvent A,
2.4. Method validation This HPLC–FLD method was validated in terms of linearity, LOD, LOQ, accuracy and precision following the International Conference on Harmonization (ICH) guidelines [18]. For linearity validation, nine concentration levels (10–14,000 ng/ml) were tested following the method stated above and calibration curves were constructed by plotting peak area versus concentration. The LOD and LOQ were defined as the concentration that produced a signal-to-noise ratio of 5 (S/N = 5) and a signal-to-noise ratio of 15 (S/N = 15), respectively. The repeatability of this method was evaluated by injecting the standard sample (n = 6, 300 ng/ml) and measuring the relative standard deviations (RSD) of peak area. Precision and accuracy of intra- and inter-day were estimated by analyzing six replicates spiked by three different concentrations (30, 500, and 4000 ng/ml) DPD to LB and Autoinducer Bioassay (AB) media in a single day and six separate days, respectively.
2.6. AI-2 detection by V. harveyi BB170 bioassay The bioassay follows the procedures from Surette et al. [19] with a slight modification. The BB170 reported strain was cultured for 13–16 h at 30 ◦ C, and then diluted by 1:5000 into fresh AB medium. 180 l of diluted reporter strain solution was added to the wells of 96-well plate, which contained 20 l samples for AI-2 activity. The supernatants of BB120 were used for AI-2 detection and sterile AB medium was employed as the negative control. The 96-well plate was incubated in a rotary shaker at 150 rpm at 30 ◦ C. The bioluminescence was measured using a microplate reader (BioTek, Synergy, USA) at given time intervals and the signal intensity of samples relative to that of the sterile AB medium was calculated as fold induction. 3. Results and discussion 3.1. Fluorescence spectra, MS confirmation and derivatization The derivatization scheme is shown in Fig. 1. It was a typical condensation reaction accompanied by the loss of two water molecules. The derivative product, 1-(3-methylbezo[g]quinoxalin-2-yl)-ethane-1,2-diol, a heterocyclic compound with a rigid and planar structure, has its own characteristic
164
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
Fig. 1. Derivatization scheme for DPD with DAN.
Fig. 2. Fluorescence excitation and emission spectra of the derivative products (A) and MS spectra of the derivative products (B).
excitation and emission spectra (Fig. 2A). In our work, 1,2phenylenediamine, the traditional derivatization reagent, was also tested and the corresponding quinoxaline was found to have a low fluorescence efficiency, and thus could not be used for fluorescence detection. When DAN was used, an extra benzene ring increased the conjugated system of the molecule, which enhanced the fluorescence efficiency of the derivative product [20]. Besides, another aromatic diamine, 4,5-methylenedioxy-1,2phenylenediamine (MDB) was also investigated using the method proposed by Yamaguchi et al. [21]. Details of the procedures could be found in the Supplementary information. Similar results were obtained, confirming the feasibility of using this derivatization reagent. The MDB reagent afforded a detection limit of 5 ng/ml and a linear range of 40–15,000 ng/ml. The LOD of this method was at the same level with that based on DAN. However, it was found that the MDB solution prepared without beta-mercaptoethanol or sodium dithionite became red soon even stored at 4 ◦ C, suggesting that this derivative was not very stable. Furthermore, DMB is much more expensive than DAN. Therefore, considering the stability and high price of MDB, DAN should be the better choice. For the sensitive detection of AI-2, the excitation and emission wavelengths of FLD were set at 271 and 503 nm, respectively. The chromatographic peaks with the same retention times were simultaneously confirmed via high resolution mass spectrometry with ESI in a positive mode (Fig. 2B). The formula of the protonated molecules ([M + H]+ ) of the derivative product was C15 H15 O2 N2. The measured and theoretical exact masses of ion (m/z) were 255.1128 and 255.11252, respectively, which exhibited an excellent agreement within 2.5 mmu. A series of parameters, including pH, temperature and reaction time, were tested to optimize the derivatization procedure. The epimerization, isomerization, and rearrangement of carbohydrates from the medium under the alkaline conditions (pH > 7.5) might form DPD analogs via the Lobry de Bruin–van Ekenstein rearrangement, thus leading to the false positive result [13,22]. Therefore, the reactions were conducted under acidic conditions in our work and the pH values before and after derivatization in different media were measured to be in a range of 1.7–3.2 (data
not shown). Under such operating conditions, no signal at the corresponding retention time was observed in the carbohydrate solutions containing glucose, fructose, sucrose and two complex matrixes, i.e., LB and AB [23] media. In addition, reaction time and temperature were also optimized for the derivatization. 1000 ng/ml DPD solution was reacted with DAN solution at four different temperatures (60, 70, 80, and 90 ◦ C) in a time range of 10 to 420 min. Then, the peak area determined by the HPLC–FLD method was plotted against time at different temperatures (Fig. 3). It was found that a long incubation time at a high temperature might lead to the decomposition of benzoquinoxaline, while reaction at a low temperature needed a very long time to complete the reaction. For example, when reaction was set at 90 ◦ C for over 40 min, the peak area decreased gradually. At the reaction time of 420 min, a very small peak area was observed. However, operation at 80 ◦ C and 70 ◦ C required 120 min and 240 min, respectively, to complete
Fig. 3. Peak areas of the derivative reaction at different reaction times and temperatures (60 ◦ C, 70 ◦ C, 80 ◦ C, and 90 ◦ C).
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
165
Table 1 Intra- and inter-day precision and accuracy for determination of AI-2 in blank LB and AB media. Medium Spiked conc (ng/ml) LB
AB
30 500 4000 30 500 4000
Intraday (n = 6)
Interday (n = 6)
Measdconc (ng/ml)
Precision (%)
Accuracy (%)
Measdconc (ng/ml)
Precision (%)
Accuracy (%)
30.52 490.30 4050.91 29.93 518.64 3983.90
4.62 3.46 2.92 3.81 2.76 2.65
101.7 98.1 101.3 99.8 103.7 99.6
29.90 512.18 3975.56 29.75 495.32 4088.85
4.38 2.18 2.33 4.23 3.40 1.78
99.7 102.4 99.4 99.2 99.1 102.2
Fig. 4. Typical chromatograms of LB medium (A); LB medium spiked with DPD (B); E. coli MG1655 supernatant (C); AB medium (D); AB medium spiked with DPD (E); and V. harveyi BB170 supernatant (F). The arrow indicates the corresponding derivative products with a retention time of 9.247 min.
166
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
Fig. 5. Profiles of OD600 (A) and AI-2 concentration (B) of WT E. coli (MG1655) and luxS− strain (DH5␣) over time.
the reaction. Thus, in order to save time, the reaction time and temperature were selected as 40 min and 90 ◦ C, respectively. 3.2. Method validation Under the optimized chromatographic conditions, the calibration curves exhibited an extended linear dynamic range over four orders of magnitude (10–14,000 ng/ml). The correlation coefficient was 0.9999, indicating an excellent linearity. This method afforded an LOD of 1.0 ng/ml and an LOQ of 3.05 ng/ml. This detection limit is sufficiently low to meet the AI-2 detection requirements for most bacteria except some Gram positive bacteria, which are reported to respond to single digit nM concentrations of DPD [24]. Through measuring the peak area of six samples, the repeatability was calculated to be 0.74%. Accuracy and precision were also assessed and are summarized in Table 1. The precision of this method was acceptable because the RSD was less than 6% at the three given concentrations (30, 500, and 4000 ng/ml). The accuracy of this method was evaluated by comparing the observed and spiked concentrations of AI-2. The intra- and inter-day accuracies of the two complex matrixes (AB and LB media) were found to be in a range of 98.1–103.7% and 99.1–102.4%, respectively. All of these data demonstrate the excellent performance of this HPLC-FLD method.
3.3. Determination of AI-2 in biological samples The method proposed here was used to analyze the concentrations of AI-2 secreted by the widely studied E. coli and V. harveyi. Since the cultivation medium components might change during the bacterial growth, additional recovery examination was performed by spiking a known mount of DPD into the supernatants of the two species at different incubation times (Table S1). Good recoveries (97.3–103.4%) were obtained, indicating that the component change of the two media did not interfere with the detection and our method was feasible. The typical chromatograms of LB medium, LB medium spiked with DPD (500 ng/ml) and E. coli MG1655 supernatant are shown in Fig. 4A–C. Meanwhile, Fig. 4D–F, respectively, shows the chromatograms of AB medium, AB medium spiked with DPD (500 ng/ml) and V. harveyi BB170 supernatant. Separation of the derivative product from the background of the complex matrixes (LB medium and AB medium) was completely achieved as no extraneous peaks were observed at the corresponding retention time (9.247 min). The presence of chromatographic peak at 9.247 min in the supernatant of E. coli MG1655 and V. harveyi BB120 confirms the existence of DPD released from these bacterial strains. The DPD concentration determined from the chromatographic peak for E. coli MG1655 and V. harveyi BB120 were 486.07 ng/ml and
Fig. 6. Profiles of OD600 , DPD concentration and fold induction of luminescence of V. harveyi BB120 over time.
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
167
Table 2 Comparison between the HPLC–FLD method and other quantitative methods reported in the literature. LODa (ng/ml)
Chemicals and process
Instruments
Interference
BB170 bioassay [8]
4.7
Bacterial culture is time-consuming and needs sophisticated operation
Very inexpensive
HPLC–MS/MS [12]
0.030b
Very expensive
GC–MS [13]
0.7
Synthetizing derivatization reagents are complicated and time-consuming A series of complex sample pretreatment, including two-step derivatizations, extraction, and sample concentration step are needed Purification and labeling of the proteins are time-consuming and expensive Easy
Complex biological matrix leads to poor reproducibility and it’s not quantitative None
Expensive
None
Very inexpensive
Sensitive to the borate concentration in the sample None
Fluorescent sensor [10]
∼13.2c
This method a b c
1.0
Inexpensive
Depicted using the molecular weight of DPD. Based on a signal-to-noise ratio of 3. Estimated from the literature [10].
4725.61 ng/ml, respectively. It was worth noted that in our method a standard DPD was required to verify peak identity. Besides, an extra MS or UV spectrum was needed to verify the peak under question. The DPD concentrations and growth of E. coli over time were recorded (Fig. 5). The cell growth curves of the E. coli strains MG1655 (wide-type) and DH5␣ (luxS-null) expressed by OD600 had no difference (Fig. 5A), while the secretions of DPD were different (Fig. 5B). The DPD level of MG1655 increased continuously, peaked at 6 h, and then began to drop gradually. Almost all of DPD was consumed by 22 h and the maximum DPD concentration was found to be ∼8000 ng/ml. No DPD was detected in the supernatant of the DH5␣ throughout its growth process, which is consistent with the results reported previously [25]. The DPD profiles of widetype and luxS-null E. coli were in consistency with the previous reports [10,26]. The conventional V. harveyi bioassay, which utilizes a V. harveyi reporter strain (BB170) for bioluminescence induction, was also compared with our HPLC–FLD method. Since the E. coli strains were incubated with glucose, which might interfere with this bioassay [14,16]. BB120 (AI-2 positive strain) was selected for tests. The profiles of OD600 value, DPD concentration and fold induction of luminescence of V. harveyi BB120 over time are shown in Fig. 6. Although BB170 bioassay is not a quantitative method, the fold induction could reflect the DPD level to some extent. The DPD detected by the HPLC–FLD method exhibited a similar tendency with the fold induction change, also confirming the feasibility of this method. Our HPLC–FLD method could avoid the interference from assay conditions such as pH and medium characteristics. Thus, it can be used to detect AI-2 in the supernatant of some bacteria like Streptococcus [13], which could secrete toxins that are hazardous to the BB170 bioreporter. 3.4. Features of this method In Table 2 the four currently used methods for AI-2 analysis are compared with the method developed in this work in terms of LOD, chemicals & process, instruments and interference. Our method has a low LOD of 1.0 ng/ml, which is comparable to that of the GC–MS method [13] and just one order higher than that of the HPLC–MS/MS method [12]. However, no complex derivatization reagents, tedious steps and expensive equipments were needed in our method. Compared with the bioassay and fluorescent sensor methods, our method has a lower LOD and better selectivity [8,10]. In addition, the very complicated procedures of the fluorescent sensor method are not needed. Given the facts above, the HPLC–FLD method proposed here offers a sufficiently low LOD for AI-2 detection and has a high selectivity. Meanwhile, the procedure is simple and has no need for expensive equipments like HPLC–MS/MS and GC–MS.
4. Conclusions In this work we develop a simple and sensitive chemical analytical method for AI-2 detection through the derivatization of AI-2 with 2,3-diaminonaphthalene. This method is effective and reliable as it allows qualitative and quantitative analysis over a wide range of concentrations (10–14,000 ng/ml) with a low detection limit (1.0 ng/ml). In addition, its sensitivity is comparable to the currently used methods. This method is successfully used for analyzing AI-2 in the supernatants of bacterial strains. The simplicity of the procedure, high sensitivity and low cost make this method suitable for a quantitative analysis of AI-2 in biological samples. Acknowledgements The authors thank the Natural Science Foundation of China (51129803) and the Program for Changjiang Scholars and Innovative Research Team in University of Ministry of Education of China for supporting this study. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2014.07.103. References [1] A. Camilli, B.L. Bassler, Bacterial small-molecule signaling pathways, Science 311 (2006) 1113–1116. [2] A.M. Stevens, M. Schuster, K.P. Rumbaugh, Working together for the common good: cell–cell communication in bacteria, J. Bacteriol. 194 (2012) 2131–2141. [3] J.B. Xavier, Social interaction in synthetic and natural microbial communities, Mol. Syst. Biol. 7 (2011) 1–11. [4] E. Karunakaran, J. Mukherjee, B. Ramalingam, C.A. Biggs, “Biofilmology”: a multidisciplinary review of the study of microbial biofilms, Appl. Microbiol. Biotechnol. 90 (2011) 1869–1881. [5] C.S. Pereira, J.A. Thompson, K.B. Xavier, AI-2-mediated signalling in bacteria, FEMS Microbiol. Rev. 37 (2013) 156–181. [6] W.R. Galloway, J.T. Hodgkinson, S.D. Bowden, M. Welch, D.R. Spring, Quorum sensing in Gram-negative bacteria: small-molecule modulation of AHL and AI-2 quorum sensing pathways, Chem. Rev. 111 (2010) 28–67. [7] K.R. Hardie, K. Heurlier, Establishing bacterial communities by ‘word of mouth’: LuxS and autoinducer 2 in biofilm development, Nat. Rev. Microbiol. 6 (2008) 635–643. [8] R. Vilchez, A. Lemme, V. Thiel, S. Schulz, H. Sztajer, I. Wagner-Dobler, Analysing traces of autoinducer-2 requires standardization of the Vibrio harveyi bioassay, Anal. Bioanal. Chem. 387 (2007) 489–496. [9] B.L. Bassler, E.P. Greenberg, A.M. Stevens, Cross-species induction of luminescence in the quorum-sensing bacterium Vibrio harveyi, J. Bacteriol. 179 (1997) 4043–4045. [10] J.G. Zhu, D.H. Pei, A LuxP-based fluorescent sensor for bacterial autoinducer II, ACS Chem. Biol. 3 (2008) 110–119. [11] T. Hauck, Y. Hubner, F. Bruhlmann, W. Schwab, Alternative pathway for the formation of 4,5-dihydroxy-2,3-pentanedione, the proposed precursor of
168
[12]
[13]
[14]
[15]
[16] [17] [18]
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168 4-hydroxy-5-methyl-3(2H)-furanone as well as autoinducer-2, and its detection as natural constituent of tomato fruit, Biochim. Biophys. Acta, Gen. Subj. 1623 (2003) 109–119. S.R. Campagna, J.R. Gooding, A.L. May, Direct quantitation of the quorum sensing signal, autoinducer-2, in clinically relevant samples by liquid chromatography–tandem mass spectrometry, Anal. Chem. 81 (2009) 6374–6381. V. Thiel, R. Vilchez, H. Sztajer, I. Wagner-Döbler, S. Schulz, Identification, quantification, and determination of the absolute configuration of the bacterial quorum-sensing signal autoinducer-2 by gas chromatography–mass spectrometry, ChemBioChem 10 (2009) 479–485. S.C.J. DeKeersmaecker, J. Vanderleyden, Constraints on detection of autoinducer-2 (Al-2) signalling molecules using Vibrio harveyi as a reporter, Microbiology—SGM 149 (2003) 1953–1956. M.M. Meijler, L.G. Hom, G.F. Kaufmann, K.M. McKenzie, C.Z. Sun, J.A. Moss, M. Matsushita, K.D. Janda, Synthesis and biological validation of a ubiquitous quorum-sensing molecule, Angew. Chem. Int. Ed. 43 (2004) 2106–2108. Y. Turovskiy, M.L. Chikindas, Autoinducer-2 bioassay is a qualitative, not quantitative method influenced by glucose, J. Microbiol. Meth. 66 (2006) 497–503. J. Zhu, E. Dizin, X. Hu, A.-S. Wavreille, J. Park, D. Pei, S-ribosylhomocysteinase (LuxS) is a mononuclear iron protein, Biochemistry 42 (2003) 4717–4726. International Conferences on Harmonization (ICH), ICH Harmonized Tripartite Guideline: Validation of Analytical Procedures: Text and Methodology, Q2(R1), International Conferences on Harmonization (ICH), Geneva, Switzerland, 2005 http://www.ich.org/LOB/media/MEDIA417.pdf
[19] M.G. Surette, B.L. Bassler, Quorum sensing in Escherichia coli and Salmonella typhimurium, Proc. Natl. Acad. Sci. 95 (1998) 7046–7050. [20] N. Nijegorodov, W. Downey, The influence of planarity and rigidity on the absorption and fluorescence parameters and intersystem crossing rate constant in aromatic molecules, J. Phys. Chem. 98 (1994) 5639–5643. [21] M. Yamaguchi, S. Hara, M. Nakamura, Determination of methylglyoxal in mouse blood by liquid chromatography with fluorescence detection, Anal. Chim. Acta 221 (1989) 163–166. [22] S.J. Angyal, The Lobry de Bruyn–Alberda van Ekenstein transformation and related reactions, in: A.E. Stütz (Ed.), Glycoscience, Springer-Verlag, Berlin, Heidelberg, 2001, pp. 1–14. [23] E.P. Greenberg, J.W. Hastings, S. Ulitzur, Induction of luciferase synthesis in Beneckea harveyi by other marine-bacteria, Arch. Microbiol. 120 (1979) 87–91. [24] A.H. Rickard, R.J. Palmer, D.S. Blehert, S.R. Campagna, M.F. Semmelhack, P.G. Egland, B.L. Bassler, P.E. Kolenbrander, Autoinducer 2: a concentrationdependent signal for mutualistic bacterial biofilm growth, Mol. Microbiol. 60 (2006) 1446–1456. [25] K.B. Xavier, B.L. Bassler, Regulation of uptake and processing of the quorum-sensing autoinducer AI-2 in Escherichia coli, J. Bacteriol. 187 (2005) 238–248. [26] J.R. Gooding, A.L. May, K.R. Hilliard, S.R. Campagna, Establishing a quantitative definition of quorum sensing provides insight into the information content of the autoinducer signals in Vibrio harveyi and Escherichia coli, Biochemistry 49 (2010) 5621–5623.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Determination of autoinducer-2 in biological samples by high-performance liquid chromatography with fluorescence detection using pre-column derivatization Xiang-Ning Song a , Hai-Bin Qiu a , Xiang Xiao a , Yuan-Yuan Cheng a , Wen-Wei Li a , Guo-Ping Sheng a,∗ , Xiao-Yan Li b , Han-Qing Yu a,∗ a b
Department of Chemistry, University of Science & Technology of China, Hefei 230026, China Environmental Engineering Research Centre, The University of Hong Kong, Hong Kong, China
a r t i c l e
i n f o
Article history: Received 17 June 2014 Received in revised form 29 July 2014 Accepted 31 July 2014 Available online 11 August 2014 Keywords: Autoinducer-2 Biological samples Detection 2,3-Diaminonaphthalene HPLC–FLD
a b s t r a c t Autoinducer-2 (AI-2), as a small-molecular-weight organic molecule secreted and perceived by various bacteria, enables intra- and inter-species communications. Quantitative determination of AI-2 is essential for exploring the bacterial AI-2-related physiological and biochemical processes. However, current strategies for sensitive detection of AI-2 require sophisticated instruments and complicated procedures. In this work, on the basis of the derivatization of AI-2 with 2,3-diaminonaphthalene, a simple, sensitive and cost-effective high-performance liquid chromatography with fluorescence detector (HPLC–FLD) method is developed for the quantitative detection of AI-2. Under the optimized conditions, this method had a broad linear range of 10–14,000 ng/ml (R2 = 0.9999), and a low detection limit of 1.0 ng/ml. Furthermore, the effectiveness of this approach was further validated through measuring the AI-2 concentrations in the cell-free culture supernatants of both Escherichia coli and Vibrio harveyi. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Bacteria secrete small-molecular-weight molecules, i.e., autoinducers, to communicate with each other, termed as quorum sensing (QS) [1]. QS has been found to regulate biofilm formation, virulence, production of antibiotics, formation of fruiting body, and gene transfer [2,3]. In the past two decades, QS has been extensively studied because of its important roles in the fields of health and environment. Among the autoinducers, autoinducer-2 (AI-2), as one unique autoinducer for enabling both intra- and inter-species communication, is found to be secreted by a variety of bacteria [4]. Many different bacterial species are reported to secrete 4,5-dihydroxy-2,3-pentanedione (DPD), the metabolic precursor of AI-2 [5]. DPD undergoes further rearrangements spontaneously and exists as a mixture of relevant compounds, some of which are recognized by various bacteria [6]. This mechanism enables bacteria
∗ Corresponding authors at: University of Science and Technology of China, Department of Chemistry, Hefei 230026, China. Tel.: +86 551 63607592/+86 551 63607453; fax: +86 551 63601592. E-mail addresses: [email protected] (G.-P. Sheng), [email protected] (H.-Q. Yu). http://dx.doi.org/10.1016/j.chroma.2014.07.103 0021-9673/© 2014 Elsevier B.V. All rights reserved.
to perceive AI-2 derived from DPD secreted by themselves or other bacterial species, and ensures interspecies communication [7]. Because of the crucial roles of AI-2 in QS-related research fields, sensitive and selective detection and quantification of AI-2 in biological samples are greatly desired. However, its various forms and low concentrations found in biological samples make its quantitative analysis difficult. Several methods have been developed to detect and quantify AI-2: e.g., Vibrio harveyi luminescence bioassay [8,9], biosensor based on fluorescence yield change when binding AI-2 to receptor proteins [10], high-performance liquid chromatography with UV detector (HPLC–UV), high-performance liquid chromatography with tandem mass spectrometric detector (HPLC–MS/MS) or gas chromatography–mass spectrometry (GC–MS) using 1,2-phenylenediamine or its derivatives to react with DPD to form the corresponding quinoxaline [11–13]. The V. harveyi bioassay method using the induction of luminescence of BB170 bioreporters is the most common way to detect AI-2 because of its sensitivity over several orders of magnitude. However, this bioassay is a qualitative method and not suitable for quantitative analysis [8,14]. Signal inhibition caused by high-concentration AI-2 is also observed [15] and some assay conditions such as pH, glucose levels and borate concentration interfere with this assay [8,16]. The biosensors based on AI-2 receptor proteins can response to only the fraction of DPD, which has been converted
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
163
to borate, making this assay sensitive to the borate concentration [10]. Besides, its linear ranges is very narrow (∼1–20 M), and the procedures to purify and label the proteins are complicated, time-consuming and expensive. The derivatization method, which employs 1,2-phenylenediamine as the derivatization reagents and detects the corresponding quinoxaline by HPLC-UV and Nuclear Magnetic Resonance (NMR), fails to give any quantitative results [11,17]. The GC–MS method, which also initially undergoes 1,2phenylenediamine derivatization and is followed by the treatment with silylation reagents, has a detection limit (LOD) of 0.7 ng/ml and can meet the AI-2 detection requirements in many cases [13]. However, very complicated procedures, such as two-step derivatization, solid phase extraction and concentration, make it difficult to perform. Recently, an HPLC–MS/MS method based on derivatization with derivatives of 1,2-phenylenediamine has been proposed to detect trace AI-2 in human saliva [12]. However, the derivatization reagents with complicated structures are required and usually need painstaking synthetic procedures, which significantly limit its wide application. In addition, expensive facilities, i.e., HPLC–MS/MS, are required for the determination. In this study, we try to establish a simple, sensitive and costeffective method for the quantitative analysis of AI-2. To this end, 2,3-diaminonaphthalene (DAN), a simple and easily available reagent, was used to react with DPD to form a new fluorescent substance, which has characteristic excitation and emission wavelengths. Thus, HPLC with fluorescence detector (HPLC–FLD) with an excellent selectivity and sensitivity could be developed to detect the AI-2 concentration in complex matrix like the supernatants of Escherichia coli and V. harveyi.
30% solvent B; t = 27 min, 70% solvent A, 30% solvent B. The excitation and emission wavelengths of the fluorescence detector were set at 271 and 503 nm, respectively. The mass spectral of the derivative products was also recorded by LTQ XL orbitrap high resolution mass spectrometry (Thermo Fisher Sci. Inc., USA) fitted with an electrospray source operated in a positive mode to further confirm its structure. The spray voltage was set to 4 kV, capillary temperature to 275 ◦ C. The sheath and auxiliary gas flow rate (both nitrogen) were optimized at 20 and 5 arbitrary units (a.u.). Full MS scans were acquired in the orbitrap analyzer with the resolution set to a value of 60,000.
2. Experimental
2.5. Biological sample preparation
2.1. Chemicals and preparation of standard solutions
E. coli strains MG1655 (wide-type) and DH5␣ (luxS-null) were grown at 37 ◦ C in LB medium supplemented with 0.25% glucose [19]. V. harveyi BB120 (AI-2 positive strain) was cultured in AB medium at 30 ◦ C. During the incubation period, aliquots were withdrawn at regular intervals and OD600 was recorded to determine the cell density. Then, the culture was centrifuged at 13,000 rpm for 5 min and filtered through 0.22 m membranes to remove the cells. The cell-free supernatants were quickly frozen and stored at −80 ◦ C until use or underwent derivatization immediately.
DPD solution (0.3 mg/ml, dissolved in ultrapure water) was purchased from Omm Scientific Inc. (Texas, USA). 2,3Diaminonaphthalene (DAN) was obtained from Aladdin Inc. (Shanghai, China). 4,5-Methylenedioxy-1,2-phenylenediamine (MDB), HPLC-grade formic acid and acetonitrile were purchased from Sigma-Aldrich Inc. (St. Louis, USA). 18.2 M Milli-Q ultrapure water was supplied by Millipore Co. (Billerica, USA). Other chemicals or solvents used were of analytical grade. The working standard solutions with a range of 10–14,000 ng/ml were obtained through diluting the stock solution of DPD. The DAN solution was prepared by dissolving 10 mg DAN into 50 ml 0.1 M HCl. 2.2. Derivatization procedures 400 l standard solution or supernatants after pretreatment were transferred to 2 ml autosampler vials (Agilent Inc., USA) containing an equal volume of DAN solution. The two liquids were thoroughly mixed for 2 min. Then, these samples were incubated in a water bath at 90 ◦ C for 40 min. After the samples became cooled down, they were analyzed by HPLC–FLD directly. 2.3. Chromatographic instrumentation and conditions 20 l samples was injected for analysis using 1260 HPLC system equipped with a fluorescence detector (Agilent Inc., USA). Separation was achieved on an Agilent ZORBAX SB-C18 reverse-phase column (250 mm × 4.6 mm, 5 m) set at 30 ◦ C. The mobile phase contained 0.1% formic acid (solvent A) and acetonitrile (solvent B) at a flow rate of 0.8 ml/min. A gradient elution profile was used as follows: t = 0 min, 70% solvent A, 30% solvent B; t = 4 min, 70% solvent A, 30% solvent B; and t = 12 min, 35% solvent A, 65% solvent B; t = 20 min, 35% solvent A, 65% solvent B; t = 24 min, 70% solvent A,
2.4. Method validation This HPLC–FLD method was validated in terms of linearity, LOD, LOQ, accuracy and precision following the International Conference on Harmonization (ICH) guidelines [18]. For linearity validation, nine concentration levels (10–14,000 ng/ml) were tested following the method stated above and calibration curves were constructed by plotting peak area versus concentration. The LOD and LOQ were defined as the concentration that produced a signal-to-noise ratio of 5 (S/N = 5) and a signal-to-noise ratio of 15 (S/N = 15), respectively. The repeatability of this method was evaluated by injecting the standard sample (n = 6, 300 ng/ml) and measuring the relative standard deviations (RSD) of peak area. Precision and accuracy of intra- and inter-day were estimated by analyzing six replicates spiked by three different concentrations (30, 500, and 4000 ng/ml) DPD to LB and Autoinducer Bioassay (AB) media in a single day and six separate days, respectively.
2.6. AI-2 detection by V. harveyi BB170 bioassay The bioassay follows the procedures from Surette et al. [19] with a slight modification. The BB170 reported strain was cultured for 13–16 h at 30 ◦ C, and then diluted by 1:5000 into fresh AB medium. 180 l of diluted reporter strain solution was added to the wells of 96-well plate, which contained 20 l samples for AI-2 activity. The supernatants of BB120 were used for AI-2 detection and sterile AB medium was employed as the negative control. The 96-well plate was incubated in a rotary shaker at 150 rpm at 30 ◦ C. The bioluminescence was measured using a microplate reader (BioTek, Synergy, USA) at given time intervals and the signal intensity of samples relative to that of the sterile AB medium was calculated as fold induction. 3. Results and discussion 3.1. Fluorescence spectra, MS confirmation and derivatization The derivatization scheme is shown in Fig. 1. It was a typical condensation reaction accompanied by the loss of two water molecules. The derivative product, 1-(3-methylbezo[g]quinoxalin-2-yl)-ethane-1,2-diol, a heterocyclic compound with a rigid and planar structure, has its own characteristic
164
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
Fig. 1. Derivatization scheme for DPD with DAN.
Fig. 2. Fluorescence excitation and emission spectra of the derivative products (A) and MS spectra of the derivative products (B).
excitation and emission spectra (Fig. 2A). In our work, 1,2phenylenediamine, the traditional derivatization reagent, was also tested and the corresponding quinoxaline was found to have a low fluorescence efficiency, and thus could not be used for fluorescence detection. When DAN was used, an extra benzene ring increased the conjugated system of the molecule, which enhanced the fluorescence efficiency of the derivative product [20]. Besides, another aromatic diamine, 4,5-methylenedioxy-1,2phenylenediamine (MDB) was also investigated using the method proposed by Yamaguchi et al. [21]. Details of the procedures could be found in the Supplementary information. Similar results were obtained, confirming the feasibility of using this derivatization reagent. The MDB reagent afforded a detection limit of 5 ng/ml and a linear range of 40–15,000 ng/ml. The LOD of this method was at the same level with that based on DAN. However, it was found that the MDB solution prepared without beta-mercaptoethanol or sodium dithionite became red soon even stored at 4 ◦ C, suggesting that this derivative was not very stable. Furthermore, DMB is much more expensive than DAN. Therefore, considering the stability and high price of MDB, DAN should be the better choice. For the sensitive detection of AI-2, the excitation and emission wavelengths of FLD were set at 271 and 503 nm, respectively. The chromatographic peaks with the same retention times were simultaneously confirmed via high resolution mass spectrometry with ESI in a positive mode (Fig. 2B). The formula of the protonated molecules ([M + H]+ ) of the derivative product was C15 H15 O2 N2. The measured and theoretical exact masses of ion (m/z) were 255.1128 and 255.11252, respectively, which exhibited an excellent agreement within 2.5 mmu. A series of parameters, including pH, temperature and reaction time, were tested to optimize the derivatization procedure. The epimerization, isomerization, and rearrangement of carbohydrates from the medium under the alkaline conditions (pH > 7.5) might form DPD analogs via the Lobry de Bruin–van Ekenstein rearrangement, thus leading to the false positive result [13,22]. Therefore, the reactions were conducted under acidic conditions in our work and the pH values before and after derivatization in different media were measured to be in a range of 1.7–3.2 (data
not shown). Under such operating conditions, no signal at the corresponding retention time was observed in the carbohydrate solutions containing glucose, fructose, sucrose and two complex matrixes, i.e., LB and AB [23] media. In addition, reaction time and temperature were also optimized for the derivatization. 1000 ng/ml DPD solution was reacted with DAN solution at four different temperatures (60, 70, 80, and 90 ◦ C) in a time range of 10 to 420 min. Then, the peak area determined by the HPLC–FLD method was plotted against time at different temperatures (Fig. 3). It was found that a long incubation time at a high temperature might lead to the decomposition of benzoquinoxaline, while reaction at a low temperature needed a very long time to complete the reaction. For example, when reaction was set at 90 ◦ C for over 40 min, the peak area decreased gradually. At the reaction time of 420 min, a very small peak area was observed. However, operation at 80 ◦ C and 70 ◦ C required 120 min and 240 min, respectively, to complete
Fig. 3. Peak areas of the derivative reaction at different reaction times and temperatures (60 ◦ C, 70 ◦ C, 80 ◦ C, and 90 ◦ C).
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
165
Table 1 Intra- and inter-day precision and accuracy for determination of AI-2 in blank LB and AB media. Medium Spiked conc (ng/ml) LB
AB
30 500 4000 30 500 4000
Intraday (n = 6)
Interday (n = 6)
Measdconc (ng/ml)
Precision (%)
Accuracy (%)
Measdconc (ng/ml)
Precision (%)
Accuracy (%)
30.52 490.30 4050.91 29.93 518.64 3983.90
4.62 3.46 2.92 3.81 2.76 2.65
101.7 98.1 101.3 99.8 103.7 99.6
29.90 512.18 3975.56 29.75 495.32 4088.85
4.38 2.18 2.33 4.23 3.40 1.78
99.7 102.4 99.4 99.2 99.1 102.2
Fig. 4. Typical chromatograms of LB medium (A); LB medium spiked with DPD (B); E. coli MG1655 supernatant (C); AB medium (D); AB medium spiked with DPD (E); and V. harveyi BB170 supernatant (F). The arrow indicates the corresponding derivative products with a retention time of 9.247 min.
166
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
Fig. 5. Profiles of OD600 (A) and AI-2 concentration (B) of WT E. coli (MG1655) and luxS− strain (DH5␣) over time.
the reaction. Thus, in order to save time, the reaction time and temperature were selected as 40 min and 90 ◦ C, respectively. 3.2. Method validation Under the optimized chromatographic conditions, the calibration curves exhibited an extended linear dynamic range over four orders of magnitude (10–14,000 ng/ml). The correlation coefficient was 0.9999, indicating an excellent linearity. This method afforded an LOD of 1.0 ng/ml and an LOQ of 3.05 ng/ml. This detection limit is sufficiently low to meet the AI-2 detection requirements for most bacteria except some Gram positive bacteria, which are reported to respond to single digit nM concentrations of DPD [24]. Through measuring the peak area of six samples, the repeatability was calculated to be 0.74%. Accuracy and precision were also assessed and are summarized in Table 1. The precision of this method was acceptable because the RSD was less than 6% at the three given concentrations (30, 500, and 4000 ng/ml). The accuracy of this method was evaluated by comparing the observed and spiked concentrations of AI-2. The intra- and inter-day accuracies of the two complex matrixes (AB and LB media) were found to be in a range of 98.1–103.7% and 99.1–102.4%, respectively. All of these data demonstrate the excellent performance of this HPLC-FLD method.
3.3. Determination of AI-2 in biological samples The method proposed here was used to analyze the concentrations of AI-2 secreted by the widely studied E. coli and V. harveyi. Since the cultivation medium components might change during the bacterial growth, additional recovery examination was performed by spiking a known mount of DPD into the supernatants of the two species at different incubation times (Table S1). Good recoveries (97.3–103.4%) were obtained, indicating that the component change of the two media did not interfere with the detection and our method was feasible. The typical chromatograms of LB medium, LB medium spiked with DPD (500 ng/ml) and E. coli MG1655 supernatant are shown in Fig. 4A–C. Meanwhile, Fig. 4D–F, respectively, shows the chromatograms of AB medium, AB medium spiked with DPD (500 ng/ml) and V. harveyi BB170 supernatant. Separation of the derivative product from the background of the complex matrixes (LB medium and AB medium) was completely achieved as no extraneous peaks were observed at the corresponding retention time (9.247 min). The presence of chromatographic peak at 9.247 min in the supernatant of E. coli MG1655 and V. harveyi BB120 confirms the existence of DPD released from these bacterial strains. The DPD concentration determined from the chromatographic peak for E. coli MG1655 and V. harveyi BB120 were 486.07 ng/ml and
Fig. 6. Profiles of OD600 , DPD concentration and fold induction of luminescence of V. harveyi BB120 over time.
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168
167
Table 2 Comparison between the HPLC–FLD method and other quantitative methods reported in the literature. LODa (ng/ml)
Chemicals and process
Instruments
Interference
BB170 bioassay [8]
4.7
Bacterial culture is time-consuming and needs sophisticated operation
Very inexpensive
HPLC–MS/MS [12]
0.030b
Very expensive
GC–MS [13]
0.7
Synthetizing derivatization reagents are complicated and time-consuming A series of complex sample pretreatment, including two-step derivatizations, extraction, and sample concentration step are needed Purification and labeling of the proteins are time-consuming and expensive Easy
Complex biological matrix leads to poor reproducibility and it’s not quantitative None
Expensive
None
Very inexpensive
Sensitive to the borate concentration in the sample None
Fluorescent sensor [10]
∼13.2c
This method a b c
1.0
Inexpensive
Depicted using the molecular weight of DPD. Based on a signal-to-noise ratio of 3. Estimated from the literature [10].
4725.61 ng/ml, respectively. It was worth noted that in our method a standard DPD was required to verify peak identity. Besides, an extra MS or UV spectrum was needed to verify the peak under question. The DPD concentrations and growth of E. coli over time were recorded (Fig. 5). The cell growth curves of the E. coli strains MG1655 (wide-type) and DH5␣ (luxS-null) expressed by OD600 had no difference (Fig. 5A), while the secretions of DPD were different (Fig. 5B). The DPD level of MG1655 increased continuously, peaked at 6 h, and then began to drop gradually. Almost all of DPD was consumed by 22 h and the maximum DPD concentration was found to be ∼8000 ng/ml. No DPD was detected in the supernatant of the DH5␣ throughout its growth process, which is consistent with the results reported previously [25]. The DPD profiles of widetype and luxS-null E. coli were in consistency with the previous reports [10,26]. The conventional V. harveyi bioassay, which utilizes a V. harveyi reporter strain (BB170) for bioluminescence induction, was also compared with our HPLC–FLD method. Since the E. coli strains were incubated with glucose, which might interfere with this bioassay [14,16]. BB120 (AI-2 positive strain) was selected for tests. The profiles of OD600 value, DPD concentration and fold induction of luminescence of V. harveyi BB120 over time are shown in Fig. 6. Although BB170 bioassay is not a quantitative method, the fold induction could reflect the DPD level to some extent. The DPD detected by the HPLC–FLD method exhibited a similar tendency with the fold induction change, also confirming the feasibility of this method. Our HPLC–FLD method could avoid the interference from assay conditions such as pH and medium characteristics. Thus, it can be used to detect AI-2 in the supernatant of some bacteria like Streptococcus [13], which could secrete toxins that are hazardous to the BB170 bioreporter. 3.4. Features of this method In Table 2 the four currently used methods for AI-2 analysis are compared with the method developed in this work in terms of LOD, chemicals & process, instruments and interference. Our method has a low LOD of 1.0 ng/ml, which is comparable to that of the GC–MS method [13] and just one order higher than that of the HPLC–MS/MS method [12]. However, no complex derivatization reagents, tedious steps and expensive equipments were needed in our method. Compared with the bioassay and fluorescent sensor methods, our method has a lower LOD and better selectivity [8,10]. In addition, the very complicated procedures of the fluorescent sensor method are not needed. Given the facts above, the HPLC–FLD method proposed here offers a sufficiently low LOD for AI-2 detection and has a high selectivity. Meanwhile, the procedure is simple and has no need for expensive equipments like HPLC–MS/MS and GC–MS.
4. Conclusions In this work we develop a simple and sensitive chemical analytical method for AI-2 detection through the derivatization of AI-2 with 2,3-diaminonaphthalene. This method is effective and reliable as it allows qualitative and quantitative analysis over a wide range of concentrations (10–14,000 ng/ml) with a low detection limit (1.0 ng/ml). In addition, its sensitivity is comparable to the currently used methods. This method is successfully used for analyzing AI-2 in the supernatants of bacterial strains. The simplicity of the procedure, high sensitivity and low cost make this method suitable for a quantitative analysis of AI-2 in biological samples. Acknowledgements The authors thank the Natural Science Foundation of China (51129803) and the Program for Changjiang Scholars and Innovative Research Team in University of Ministry of Education of China for supporting this study. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2014.07.103. References [1] A. Camilli, B.L. Bassler, Bacterial small-molecule signaling pathways, Science 311 (2006) 1113–1116. [2] A.M. Stevens, M. Schuster, K.P. Rumbaugh, Working together for the common good: cell–cell communication in bacteria, J. Bacteriol. 194 (2012) 2131–2141. [3] J.B. Xavier, Social interaction in synthetic and natural microbial communities, Mol. Syst. Biol. 7 (2011) 1–11. [4] E. Karunakaran, J. Mukherjee, B. Ramalingam, C.A. Biggs, “Biofilmology”: a multidisciplinary review of the study of microbial biofilms, Appl. Microbiol. Biotechnol. 90 (2011) 1869–1881. [5] C.S. Pereira, J.A. Thompson, K.B. Xavier, AI-2-mediated signalling in bacteria, FEMS Microbiol. Rev. 37 (2013) 156–181. [6] W.R. Galloway, J.T. Hodgkinson, S.D. Bowden, M. Welch, D.R. Spring, Quorum sensing in Gram-negative bacteria: small-molecule modulation of AHL and AI-2 quorum sensing pathways, Chem. Rev. 111 (2010) 28–67. [7] K.R. Hardie, K. Heurlier, Establishing bacterial communities by ‘word of mouth’: LuxS and autoinducer 2 in biofilm development, Nat. Rev. Microbiol. 6 (2008) 635–643. [8] R. Vilchez, A. Lemme, V. Thiel, S. Schulz, H. Sztajer, I. Wagner-Dobler, Analysing traces of autoinducer-2 requires standardization of the Vibrio harveyi bioassay, Anal. Bioanal. Chem. 387 (2007) 489–496. [9] B.L. Bassler, E.P. Greenberg, A.M. Stevens, Cross-species induction of luminescence in the quorum-sensing bacterium Vibrio harveyi, J. Bacteriol. 179 (1997) 4043–4045. [10] J.G. Zhu, D.H. Pei, A LuxP-based fluorescent sensor for bacterial autoinducer II, ACS Chem. Biol. 3 (2008) 110–119. [11] T. Hauck, Y. Hubner, F. Bruhlmann, W. Schwab, Alternative pathway for the formation of 4,5-dihydroxy-2,3-pentanedione, the proposed precursor of
168
[12]
[13]
[14]
[15]
[16] [17] [18]
X.-N. Song et al. / J. Chromatogr. A 1361 (2014) 162–168 4-hydroxy-5-methyl-3(2H)-furanone as well as autoinducer-2, and its detection as natural constituent of tomato fruit, Biochim. Biophys. Acta, Gen. Subj. 1623 (2003) 109–119. S.R. Campagna, J.R. Gooding, A.L. May, Direct quantitation of the quorum sensing signal, autoinducer-2, in clinically relevant samples by liquid chromatography–tandem mass spectrometry, Anal. Chem. 81 (2009) 6374–6381. V. Thiel, R. Vilchez, H. Sztajer, I. Wagner-Döbler, S. Schulz, Identification, quantification, and determination of the absolute configuration of the bacterial quorum-sensing signal autoinducer-2 by gas chromatography–mass spectrometry, ChemBioChem 10 (2009) 479–485. S.C.J. DeKeersmaecker, J. Vanderleyden, Constraints on detection of autoinducer-2 (Al-2) signalling molecules using Vibrio harveyi as a reporter, Microbiology—SGM 149 (2003) 1953–1956. M.M. Meijler, L.G. Hom, G.F. Kaufmann, K.M. McKenzie, C.Z. Sun, J.A. Moss, M. Matsushita, K.D. Janda, Synthesis and biological validation of a ubiquitous quorum-sensing molecule, Angew. Chem. Int. Ed. 43 (2004) 2106–2108. Y. Turovskiy, M.L. Chikindas, Autoinducer-2 bioassay is a qualitative, not quantitative method influenced by glucose, J. Microbiol. Meth. 66 (2006) 497–503. J. Zhu, E. Dizin, X. Hu, A.-S. Wavreille, J. Park, D. Pei, S-ribosylhomocysteinase (LuxS) is a mononuclear iron protein, Biochemistry 42 (2003) 4717–4726. International Conferences on Harmonization (ICH), ICH Harmonized Tripartite Guideline: Validation of Analytical Procedures: Text and Methodology, Q2(R1), International Conferences on Harmonization (ICH), Geneva, Switzerland, 2005 http://www.ich.org/LOB/media/MEDIA417.pdf
[19] M.G. Surette, B.L. Bassler, Quorum sensing in Escherichia coli and Salmonella typhimurium, Proc. Natl. Acad. Sci. 95 (1998) 7046–7050. [20] N. Nijegorodov, W. Downey, The influence of planarity and rigidity on the absorption and fluorescence parameters and intersystem crossing rate constant in aromatic molecules, J. Phys. Chem. 98 (1994) 5639–5643. [21] M. Yamaguchi, S. Hara, M. Nakamura, Determination of methylglyoxal in mouse blood by liquid chromatography with fluorescence detection, Anal. Chim. Acta 221 (1989) 163–166. [22] S.J. Angyal, The Lobry de Bruyn–Alberda van Ekenstein transformation and related reactions, in: A.E. Stütz (Ed.), Glycoscience, Springer-Verlag, Berlin, Heidelberg, 2001, pp. 1–14. [23] E.P. Greenberg, J.W. Hastings, S. Ulitzur, Induction of luciferase synthesis in Beneckea harveyi by other marine-bacteria, Arch. Microbiol. 120 (1979) 87–91. [24] A.H. Rickard, R.J. Palmer, D.S. Blehert, S.R. Campagna, M.F. Semmelhack, P.G. Egland, B.L. Bassler, P.E. Kolenbrander, Autoinducer 2: a concentrationdependent signal for mutualistic bacterial biofilm growth, Mol. Microbiol. 60 (2006) 1446–1456. [25] K.B. Xavier, B.L. Bassler, Regulation of uptake and processing of the quorum-sensing autoinducer AI-2 in Escherichia coli, J. Bacteriol. 187 (2005) 238–248. [26] J.R. Gooding, A.L. May, K.R. Hilliard, S.R. Campagna, Establishing a quantitative definition of quorum sensing provides insight into the information content of the autoinducer signals in Vibrio harveyi and Escherichia coli, Biochemistry 49 (2010) 5621–5623.
