- Email: [email protected]
Liquid chromatography–mass spectrometry characterization of FK506 biosynthetic intermediates in Streptomyces clavuligerus KCTC 10561BP
Analytical Biochemistry 393 (2009) 1–7
Contents lists available at ScienceDirect
Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio
Liquid chromatography–mass spectrometry characterization of FK506 biosynthetic intermediates in Streptomyces clavuligerus KCTC 10561BP Je Won Park, Sang-Joon Mo, Sung Ryeol Park, Yeon-Hee Ban, Young Ji Yoo, Yeo Joon Yoon * Department of Chemistry and Nano Science, Ewha Womans University, Seoul 120-750, Republic of Korea
a r t i c l e
i n f o
Article history: Received 7 January 2009 Available online 17 June 2009 Keywords: Liquid chromatography–mass spectrometry (LC–MS) Tacrolimus FK506 profiling
a b s t r a c t The development of an efficient analytical method for the reliable detection and identification of the biosynthetic intermediates found in microbial cultures, which usually produce complex intermediates of the metabolites of interest, is essential for further biosynthetic investigations. This study developed a simple and highly selective method for detecting the biosynthetic intermediates involved in the FK506 pathway of Streptomyces clavuligerus KCTC 10561BP involving a cleanup procedure using a solid-phase extraction technique to provide reliable extraction of FK506-related compounds from a cell culture broth and liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) to separate and detect the FK506-related intermediates at concentrations as low as 0.2 lg/L in the broth. This method enabled the analytical profiling of the intermediates formed during the biosynthesis of FK506 in this S. clavuligerus strain, which produced FK506 as a main product. Eight FK506 intermediates—FK520, 37,38-dihydroFK506, prolylFK506, 9-decarbonyl-9-hydroxylFK506, 9-deoxoFK506, desmethylFK520, prolylFK520, and 9-deoxoFK520—were identified. This is the first report of the LC–ESI–MS/MS characterization of a wide range of FK506 analogs from a bacterial fermentation broth. The protocol employed in this study may be useful for estimating the structure of the metabolites without the need for a time-consuming isolation process and nuclear magnetic resonance (NMR) spectroscopy. Ó 2009 Elsevier Inc. All rights reserved.
FK506 (1, also called tacrolimus) is a 23-membered macrolide polyketide that was discovered in 1987 from the fermentation broth of soil bacteria Streptomyces tsukubaensis and later found to be produced by other streptomycete strains such as Streptomyces kanamyceticus and Streptomyces clavuligerus KCTC 10561BP [1,2]. FK520 (2, also known as ascomycin) is an active FK506 analog in which the C21 allyl group is replaced by an ethyl group, is produced mainly by the fermentation of Streptomyces hygroscopicus var. ascomyceticus, and is found as a by-product of S. tsukubaensis [3,4]. Preceded by the discovery of rapamycin in 1975 [5], these natural products were among the first macrolides discovered with immunosuppressive activities. Since receiving U.S. Food and Drug Administration approval in 1994 for use in liver transplantation, FK506 has been licensed as an immunosuppressive drug after solid organ transplantation, and several promising therapeutic applications have been reported [6]. Over the past decade, the advances made in biotechnology in the characterization and manipulation of biosynthetic gene clusters for natural products, such as the polyketides found in microbes, have encouraged the exploration of a wide range of macrolide biosynthetic pathways. Studies of biosynthetic gene
* Corresponding author. Fax: +82 2 3277 3419. E-mail address: [email protected] (Y.J. Yoon). 0003-2697/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2009.06.021
clusters for immunosuppressants, including FK506, FK520, and rapamycin [7–10], have shown that these clusters encode a modular polyketide synthase (PKS),1 a nonribosomal peptide synthetaselike protein, and other biosynthetic genes that are believed to act as post-PKS modification steps. Unfortunately, the biosynthetic pathway for FK506 is not completely understood. For the functional identification and manipulation of some putative or unknown genes for FK506 production, the development of an analytical method for the reliable and efficient detection and identification of FK506-related intermediates in a microbial culture broth is essential. A number of liquid chromatography (LC) methods for quantifying macrolide immunosuppressants with ultraviolet (UV) or electrospray ionization–tandem mass spectrometry (ESI–MS/MS) detection have been reported [11–13]. Most focused on the therapeutic drug monitoring of the residual immunosuppressive macrolides in a variety of biological fluids from clinical trials and pharmacokinetic studies of the metabolites from oral administration of macrolide drugs. These methods are attractive for the simultaneous determination of a set of immunosuppressants but are unsuitable for the selective profiling of the FK506-related bio1 Abbreviations used: PKS, polyketide synthase; LC, liquid chromatography; UV, ultraviolet; ESI–MS/MS, electrospray ionization–tandem mass spectrometry; NMR, nuclear magnetic resonance; SPE, solid-phase extraction; RSD, relative standard deviation; MRM, multiple reaction monitoring; S/N, signal/noise ratio.
2
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
synthetic intermediates in microbial fermentation due to both the structural similarity of FK506-related macrolides and the possible presence of their isomers during chromatographic resolution, making the profiling of the FK506 analogs a challenging task. So far, no attempt has been made to analyze the LC profiles of the types of intermediates implicated in FK506 biosynthesis from a producing strain. The ideal method for detecting the FK506-related biosynthetic intermediates found in a culture broth would be MS because it provides definitive chemical evidence diagnostic for molecular masses and distinct fragmentation patterns. Therefore, LC coupled with MS/MS is the most desirable analytical method for the profiling of FK506 biosynthetic intermediates. This article reports an efficient LC–ESI–MS/MS method for the separation and identification of the different FK506 biosynthetic intermediates found in S. clavuligerus KCTC 10561BP fermentations based on their characteristic MS/MS fragmentation patterns. The practicability of LC–ESI–MS/MS analysis to identify the trace amounts of microbial FK506-related macrolides, which are difficult to isolate in amounts sufficient for their structural elucidation using nuclear magnetic resonance (NMR) spectrometry, is one of the primary advantages of this study. The method for creating the FK506 biosynthetic intermediate profiles developed in this study can be applied to other FK506-producing wild-type and genetically engineered streptomycete strains, and the profiles would be useful in discerning the FK506 and/or FK520 biosynthetic pathways in combination with genetic approaches.
Materials and methods
and SPE cleanup using an Oasis HLB cartridge (Waters). The whole cultures were extracted and partitioned twice using equal volumes of ethylacetate in a polypropylene tube, and the organic extracts were combined and concentrated under vacuum. The crude residues were reconstituted in 2 ml of methanol, after which they were diluted with 10 volumes of H2O and loaded immediately onto an Oasis HLB column that had been previously conditioned with 3 ml of methanol followed by 3 ml of H2O. The column was washed with 6 ml of a 15% (v/v) methanol solution (in water titrated to pH 8.0 using concentrated NH4OH) and then air-dried under reduced pressure for approximately 1 min. The bound macrolides of interest were eluted twice with 1 ml of methanol, evaporated to dryness at room temperature by vacuum centrifugation, and then kept in a freezer. For analyses, the desired residue was reconstituted to 100 ll with methanol, and 10 ll of this solution corresponding to a culture volume of 10 ml (concentration factor of 0.1) was subjected to LC–ESI–MS/MS analysis. The recoveries of both FK506 and FK520 standards spiked into a blank fermentation medium were determined to check both the performance of the SPE cleanup and the matrix effect of the medium on the isolation of the macrolides during the extraction and cleanup steps. They were extracted as described above and analyzed further by LC–ESI–MS/ MS. From a comparison of the chromatographic peak areas obtained from both spiked blank samples with those obtained from the standards, the percentage recoveries were calculated by using at least three replicates. To validate the SPE cleanup procedure more precisely, the intra- and interday precision and accuracy were assessed by analyzing standard 1 spiked (2 and 10 lg/L) into a blank fermentation medium with five replicates on three different days and are presented as relative standard deviations (RSDs).
Reagents FK506-related macrolide LC–ESI–MS/MS analysis Acetonitrile, methanol, glacial acetic acid, ethyl acetate, and water were LC grade and supplied by J.T. Baker (Philipsburg, NJ, USA). MS-grade ammonium acetate was obtained from Fluka Chemie (Steinheim, Germany), and the solid-phase extraction (SPE) cartridge (Oasis HLB, 3 cc/60 mg) and vacuum manifold were obtained from Waters (Milford, MA, USA). FK506 (1) and FK520 (2) were purchased from LC Laboratories (Woburn, MA, USA) and Sigma (St. Louis, MO, USA), respectively. Their stock solutions (1 mg/ ml in methanol) were kept in an amber polypropylene microtube at –20 °C until used to examine the cleanup protocol efficiency using an Oasis HLB SPE cartridge to isolate the FK506 biosynthetic intermediates and tune the optimum ESI–MS/MS parameter for the detection of FK506-related macrolides. All other chemicals were of the highest purity available. Bacterial strain and medium S. clavuligerus KCTC 10561BP was obtained from the Korean Collection for Type Cultures (Daejun, Republic of Korea) [2]. This strain was routinely subcultivated at 28 °C on agar medium containing 1% soluble starch, 0.4% yeast extract, 1% malt extract, and 1.5% agar. The fermentation medium (100 ll) was placed in a 1-L baffled Erlenmeyer flask and sterilized at 121 °C for 20 min, as described previously [2]. The medium was then inoculated with mycelium from S. clavuligerus grown on the agar medium and shake-cultured at 25 °C for 1 week. The whole culture broths were subjected to analysis of the FK506-related macrolides, and all experiments were performed independently in triplicate. FK506-related macrolide extraction and cleanup The extraction and cleanup procedures for the FK506 biosynthetic intermediates in the S. clavuligerus cultures before LC–ESI– MS/MS analysis included organic partitioning via ethylacetate
Analyses of the FK506-related biosynthetic intermediates were performed using a Waters/Micromass Quattro micro MS interface consisting of a Waters 2695 separation module connected directly to a Micromass Quattro micro MS. Separation was performed on a 50 2.1-mm XTerra MS C18 reversed-phase column (3.5 lm, Waters) maintained at 50 °C, and the analytes were eluted at a flow rate of 100 ll/min with a gradient of 5 mM (w/v) ammonium acetate, 0.05% (v/v) acetic acid in H2O (A), and 80% (v/v) acetonitrile with the same additive concentration (B) at 55% B to 70% B for 35 min. The effluent was directed to the ESI–MS/MS operated in positive ion mode. The instrument was calibrated by the direct infusion of a stock solution of authentic FK506 into the ion source at a rate of 40 ll/min. The optimization parameters of the ESI–MS/ MS system were based on the maximum generation, first of the ammoniated molecular ion [M+NH4+] (precursor) and then of the corresponding product ion. The following calibration parameters were used: 0.8 kV capillary voltage, 35 V cone voltage, 120 °C source temperature, 250 °C desolvation temperature, 500 L/h desolvation gas flow rate, and 50 L/h cone gas flow rate. The collision energy in MS/MS mode, which concurred with the full argoninduced fragmentation of the parent ions, was found to be 8 V (pressure reading 3.15 10 6 mbar). The molecular structures and fragmentation patterns of each FK506-related macrolide produced during the cultivation of S. clavuligerus that were obtained under these conditions are shown in Fig. 1. To evaluate the feasibility of this analytical profiling technique for the detection of the FK506 biosynthetic intermediates compared with that of other drug monitoring methods, both the 1 and 2 standards spiked into a blank fermentation medium were quantified using MS/MS in multiple reaction monitoring (MRM) mode. This was carried out by choosing two mass ions (listed in Fig. 1) set to detect a transition of the ammonium adduct precursor to the product ions specific to the selected analytes (1, m/z 822 > 576; 2, m/z 810 > 564).
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
3
Fig. 1. Chemical structure of FK506 and its biosynthetic intermediates in the S. clavuligerus culture broth and ESI–MS/MS fragmentation patterns for identification. FAB, a large fragment ion from the cleavage of both the A and B sites; Fbc, a small ion from cleavage of both the B and C sites; –, no chemical element.
Results and discussion Analytical performance of SPE cleanup In therapeutic drug monitoring of FK506 and its metabolites in human blood and plasma, sample preparation usually involves protein precipitation and solvent dissolution using zinc sulfate and acetonitrile, respectively [13]. However, there are several problems with this protocol when being applied to the analytical profiling of FK506 biosynthetic intermediates from bacterial cultures. For example, the use of water-miscible solvents (i.e., acetonitrile and methanol) to extract the macrolides from matrix-rich bacterial cultures allows retention in the extracts of hydrophilic compounds that would interfere with the subsequent analysis. It has been reported that metal ions, including zinc that remained in the sample ready for LC–MS injection, not only cause the low recovery of FK506 from whole blood but also suppress ionization during MS detection [12]. In particular, streptomycete cultivation medium commonly contains several metal ions such as iron, cobalt, magnesium, and manganese. In addition, even with the use of water-immiscible ethylacetate for extractions, the residual aqueous solution in the extracts can allow the tautomeric behavior of 1 during LC analysis, and this can result in incomplete chromatographic separation and erroneous quantification of FK506-related macrolides, as reported previously [14]. Therefore, it is important to incorporate an additional SPE cleanup to minimize these negative effects. An examination of the analytes recovered from the blank fermentation medium and diluted stock solution spiked with 1 followed by ethylacetate extraction and LC–ESI–MS/MS analysis showed significant differences in the recoveries of 1 (41 ± 8% from a medium and 94 ± 4% from a stock), supporting concerns regarding such matrix effects. There are a few methods for monitoring
immunosuppressant drugs that report an on-line SPE cartridge prior to chromatographic elution [15,16]. In this study, several types of off-line SPE cartridges from different vendors were examined, and an Oasis HLB column was selected due to its ability to enhance the recovery of 1 (>72% from a blank medium) and a cleaner resulting extract that will secure the ESI– MS/MS apparatus. Detailed protocols for cleanup using the cartridge were investigated. It was apparent that the implementation of a slightly alkaline washing step reinforced the recoveries of 1 and 2 to 89 ± 3 and 85 ± 4%, respectively, ensuring the efficient removal of the matrix that may impact chromatographic separation and the MS/MS response. The intra- and interday precision and accuracy using the blank fermentation medium spiked with standard 1 at two different levels were investigated. The precision (RSD) was less than 5%, and the intra- and interday precision and accuracy ranged from 92 to 94% and 92 to 93%, respectively, indicating that SPE cleanup appears to be accurate and precise enough to analyze FK506 and its intermediates produced in the culture broth. Separation of FK506-related macrolide using LC–ESI–MS/MS As generally recognized, 1 and 2 show relatively low affinities to protons during the positive ESI mode but have a high binding capability for ammonium and sodium ions to acquire their adduct ion forms as cationized molecules [11,15]. Although sodium adduct ions tend to be generated in the ion source, they were unacceptable for MS/MS detection, especially for analytical profiling, because of their inadequate fragmentation patterns. In this study, ammonium adduct ions were chosen as precursor ions for MS/ MS, and any contact with potential sodium sources, such as glassware, should be prevented. All experiments performed during the analyses were glassware free.
4
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
The MS/MS spectra of authentic 1 and 2 are presented in Fig. 2A. FK506 had its ammonium adduct precursor m/z peak at 822, which was accompanied by several product ions, including m/z 786, 768, 576 (FAB, a large fragment ion generated by cleavage of both the A and B sites, annotated in Fig. 1), 562, 503 (FBC), 465 (FAD), 336 (Fbc, a small fragment ion generated by cleavage of both the B and C sites), and 265 (Fab), whereas FK520 had a precursor at 810 with fragment ions at m/z 774, 756, 564, 562, 491, 453, 336, and 265. The existence of some identical fragment ions, such as m/z 562, 336, and 265, shows that their structural difference occurred in the polyketide backbone and not in the pipecolate and cyclohexane moieties. However, other fragment ions with a difference of 12 Da could be used as diagnostic ions for a detailed structural analysis of 1 and 2. Therefore, MS/MS detection of both macrolides was performed by monitoring the 822 to 576 mass transition for FK506 and the 810 to 564 mass transition for FK520. From previous reports on the structural elucidation of FK506 and FK520 analogs isolated from the streptomycete fermentation, the FK506 biosynthetic intermediates possess molecular skeletons quite similar to the final product, FK506 (Fig. 1) [17]. Because it is possible that some precursor and product ions may have identical m/z values during LC–ESI–MS/MS analysis, selective chromatographic separation with fully resolved peaks is essential for the reliable profiling of the intermediates. Chromatographic elution of authentic 1 and 2
spiked separately into a blank fermentation medium, followed by the described procedures, was investigated using 5 mM ammonium acetate with 0.05% acetic acid in the mobile phase. The width of the peaks corresponding to FK506 and FK520 decreased with increasing acetic acid concentrations while the elution of early impurities from fermentation was accelerated. Unfortunately, ionization suppression occurred with more than 1% acetic acid, and this limited the utility of this effect. Authentic 1 and 2 had retention times of 29.9 and 27.2 min, respectively, under the above conditions for chromatographic separation of the standard analytes (Fig. 2B). FK506 and FK520 in the medium could be quantified at concentrations as low as 20 ng (with signal/noise ratio [S/N] at 200:1), as determined by the extraction of 100 ml of medium and the injection of 15 ll of the extract after cleanup. Subsequently, analytical profiling of the FK506-related macrolides in the S. clavuligerus culture was performed using the validated cleanup protocol and optimized elution system. Analytical profiling of FK506 biosynthetic intermediates produced in the S. clavuligerus KCTC 10561BP strain In addition to the major final product FK506, eight types of related macrolides were detected in the extract of the S. clavuligerus fermentation broth (Figs. 1 and 3). The identity of each one was as-
Fig. 2. Typical ESI–MS/MS spectra of authentic FK506 and FK520 spiked into the blank medium (A) and their LC–ESI–MS/MS chromatograms traced with selective MRM mode (B).
Fig. 3. Representative LC–ESI–MS/MS chromatograms of FK520 (A), 9-decarbonyl-9-hydroxyFK506 and 37,38-dihydroFK506 (B), prolylFK506 and 9-deoxoFK506 (C), and prolylFK520, 9-deoxoFK520, and desmethylFK520 (D) using their specific MRM mode. 1*, FK506 isotope; 2, FK520; 3, 9-decarbonyl-9-hydroxyFK506; 4, 37,38-dihydroFK506; 5, prolylFK506; 6, 9-deoxoFK506; 7, prolylFK520; 8, 9-deoxoFK520; 9, desmethylFK520.
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
signed based on a comparison of the LC–ESI–MS/MS fragmentation patterns with those of authentic 1 and 2. The appearance of a peak with a 27.2-min retention time (by monitoring m/z 810 > 564, Fig. 3A) showed that S. clavuligerus had produced both FK506 and FK520 even though several publications have reported that these two macrolides are produced separately by different actinomycete strains [2–4]. Two peaks with 22.9- and 42.1-min retention times that were captured by the mass transition from m/z 824 to 578 were presumed to be reduced FK506 biosynthetic intermediates (Fig. 3B). The identical fragmentation patterns, with a precursor ion (m/z 824) and characteristic ions of m/z 578 (FAB) and 467 (FAD), revealed the addition of 2 Da to the corresponding ions of 1 (Figs. 4A and B), suggesting that there was no structural change in the region spanning the pipecolate and cyclohexane moieties (C1–C6 and C24–C34 positions) within the molecules. A peak with a 22.9-min retention time was found to be 9-decarbonyl-9-hydroxylFK506 (3) because the fragment ion at m/z 503 (FBC) was identical to that of FK506 and the ions at m/z 564 (FAC) and 338 (Fbc) were different by 2 Da from the corresponding ions of FK506, showing that a reduction process had taken place within the region between the C8 and C14 positions. Although there has been no report of this proposed compound, it appears to be a biosynthetic intermediate during oxidation from 9-deoxoFK506 (as described below) to FK506 that occurred as a post-PKS step [10]. A mutant with a specific monooxygenase encoding gene rapJ disrupted could produce deoxorapamycin from a recent biotechnological invention that generated a set of rapamycin analogs from the recombinant strains of S. hygroscopicus [18]. This suggests the consecutive oxidation from deoxo-, via hydroxyl-, subsequently to carbonyl function within rapamycin at the C9 position, which is structurally similar to FK506, and the possible existence of 3 in the culture of FK506-producing S. clavuligerus. In the case of the latter peak, the diagnostic fragment ions at m/z 562 (FAC) and 336 (Fbc) were iden-
5
tical to those of FK506, and an ion at m/z 505 (FBC) was 2 Da different from the corresponding ion of FK506. This suggests that reduction had occurred within the region covering the polyketide backbone between the C15 and C23 positions. Therefore, the structure of a compound eluted at 42.1 min was assigned to be 37,38dihydroFK506 (4), which was isolated previously from the S. tsukubaensis culture as an FK506 derivative and named by dihydroFK506 [17]. The peak with a 29.9-min retention time detected by the same mass transition was most likely due to the M+2 isotope peak of 1 (1*, Fig. 3B) that potentially underwent the 824 to 578 MRM transition, which is a transition identical to the MRM transition noted for the two macrolides 3 and 4. This was also observed when analyzing authentic standard 1 by monitoring 824 to 578 (data not shown). Two minor peaks at retention times of 17.7 and 25.8 min that showed a specific m/z 808 to 562 mass transition (Fig. 3C) were also found to be FK506 derivatives. Both macrolides showed identical fragmentation patterns, with a precursor ion of m/z 808 and characteristic ions at m/z 562 (FAB), 548 (FAC), and 322 (Fbc) (Figs. 4C and D). The loss of 14 Da from the corresponding ions of compound 1 suggested the loss of a methylene or carbonyl functional group. The presence of a fragment ion at m/z 465 (FAD) from the 25.8-min peak was identical to one from FK506, indicating that a structural change had occurred in the pipecolate moiety, whereas the fragment ion of m/z 451 (FAD) was unique to the other 17.7min peak. This indicates that a structural modification had occurred within the region spanning the C8 to C14 positions. Therefore, the macrolides with 17.7- and 25.8-min retention times were identified as prolylFK506 (5) and tentatively as 9-deoxoFK506 (6), respectively. ProlylFK506 (5, formerly known as FK525) was isolated from S. tsukubaensis fermentation [19]. During the biosynthesis of rapamycin, which is structurally similar to FK506, a pipecolate-incorporating enzyme encoded by the rapP gene is believed to accept the amino acid l-proline and some of its analogs
Fig. 4. Representative ESI–MS/MS spectra of 9-decarbonyl-9-hydroxyFK506 (A), 37,38-dihydroFK506 (B), prolylFK506 (C), 9-deoxoFK506 (D), prolylFK520 (E), 9-deoxoFK520 (F), and desmethylFK520 (G) from S. clavuligerus fermentation.
6
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
as substrates, resulting in the production of prolylrapamycin [20]. From these results, 5 appeared to be a congener of the final product 1, depending on the substrate flexibility of a RapP-like FK506 biosynthetic enzyme (e.g., FkbP) [8,9]. On the other hand, 9-deoxoFK506 (6) appeared to be a FK506 precursor for the sequential post-PKS oxidation step, that is, from 6, via 3, and finally to 1 (as described above). From previous functional analysis studies of biosynthetic gene clusters for rapamycin and FK506, the specific genes whose products are categorized as cytochrome P450 monooxygenase are believed to catalyze the above oxidation process: rapJ/rapN for rapamycin and fkbD/fkbO for FK506 [10]. The detection of the above two minor FK506 biosynthetic intermediates 3 and 6 provided valuable information on a full understanding of the biosynthetic pathway of FK506. Therefore, further purification of these compounds will provide indispensable substrates for functional characterization of the putative biosynthetic gene products. Three additional minor peaks with retention times of 15.9, 19.8, and 23.2 min that were captured by a mass transition from m/z 796 to 550 (Fig. 3D) were assumed to be FK520-related biosynthetic intermediates. The mass spectrum of the 15.9- and 23.2-min peaks gave rise to a prominent m/z 491 ion (FBC) identical to that of FK520 but showed the characteristic m/z 550 (FAB), 548 (FAC), and 322 (Fbc) ions (Figs. 4E and F). Again, the loss of 14 Da from the corresponding ions of 2 suggests the loss of a methylene or carbonyl functional group from 2. The main difference in the fragment patterns obtained from both macrolides was the presence of the diagnostic fragment ions m/z 453 (FAD) and 439 (FAD) from the peaks with retention times of 23.2 and 15.9 min, respectively. As observed for other macrolides, such as 5 and 6, this suggests that the substitution of pipecolate by a prolyl moiety and the removal of a 9-carbonyl group had occurred in the structure of 2, resulting in prolylFK520 (7) and 9-deoxoFK520 (8), respectively. Moreover, there was a similarity in the relative peak intensity and shifts in the retention time of prolyl- and 9-deoxo analogs of 1 and 2 compared with those of the corresponding products 1 and 2, supporting the identities of both macrolides 7 and 8. The FK520 analog 7 would be useful for characterizing the substrate specificity of the RapP-like FK506 biosynthetic enzyme or for detecting the existence of other RapP-like enzymes involved in FK520 biosynthesis, whereas the analog 8 would also be valuable for determining whether the above-mentioned oxidation enzymes (probably derived from fkbD/fkbO gene products) involved in FK506 biosynthesis are responsible for similar steps in FK520 biosynthesis or whether other similar enzymes specific to FK520-related intermediates exist in S. clavuligerus. The mass spectrum of the remaining 19.8-min peak monitored by the mass transition m/ z 796 to 550 showed the ion fragments of both m/z 562 (FAC) and 336 (Fbc), which were identical to the corresponding fragments of 2, and the diagnostic fragments of m/z 550 (FAB), 477 (FBC), and 439 (FAD), which were 14 Da different from those obtained from FK520, suggesting demethylation on the polyketide backbone. Based on this information and a comparison of the mass spectrum with the spectrum for 2, a compound with a retention time of 19.8 min was found to be desmethylFK520 (9). This FK520 analog 9 was isolated from FK520-producing S. hygroscopicus var. ascomyceticus fermentation and was formerly described as desmethylascomycin or FK523 [21]. A qualitative comparison of the abundance of the FK506 biosynthetic intermediates detected in the FK506-producing S. clavuligerus culture was carried out by LC–ESI–MS/MS using MRM mode. The estimate was based on the relative intensity of a specific product ion at m/z 265, which is common to the FK506 related macrolides, compared with that of the authentic 1 ([intensity of relevant mass peak] / [intensity of reference mass peak at m/z 265 produced from the standard FK506]), FK506 (1, 85%, 117 mg/L), FK520 (2, 8%, 11 mg/L), 37,38-dihydroFK506 (4, 3%), prolylFK506 (5, >1%), 9decarbonyl-9-hydroxylFK506 (3, 1%), 9-deoxoFK506 (6, <1%), desmethylFK520 (9, >0.5%), prolylFK520 (7, >0.1%), and 9-deoxoFK520
(8, <0.1%). With the exception of 3, all macrolides described in this study were characterized previously and their ESI–MS/MS fragmentation patterns provide sufficient structural data to support their identity. An examination of the analytical profiles from the genetically engineered FK506-producing strains would be helpful for understanding the FK506 and/or FK520 biosynthetic pathways and for characterizing the function of the previously unassigned biosynthetic genes included in the FK506 and/or FK520 biosynthetic gene clusters.
Conclusion This article has described a simple and selective technique for profiling the FK506-related metabolites from a FK506-producing S. clavuligerus strain using a combination of a cleanup protocol employing an Oasis HLB cartridge and optimized LC–ESI–MS/MS conditions, and the technique was found to provide adequate chromatographic separation and detection of FK506/FK520 biosynthetic intermediates at concentrations as low as 20 ng per component from 100 ml of the culture broth. A detailed inspection of the mass spectra of the intermediates allowed differentiation of the structural modifications that occurred during fermentation, generating a profile in which eight intermediates were detected in addition to FK506: FK520, 37,38-dihydroFK506, prolylFK506, 9-decarbonyl-9-hydroxylFK506, 9-deoxoFK506, desmethylFK520, prolylFK520, and 9-deoxoFK520. This approach offers an attractive alternative for overcoming the time-consuming separation process as well as NMR spectroscopy normally needed to determine the structure of these FK506 biosynthetic intermediates or derivatives. Acknowledgments This study was supported by a Korea Science and Engineering Foundation (KOSEF) NRL Program grant funded by the Korean government (MEST, R0A-2008-000-20030-0); a KOSEF grant funded by the Korean government (MEST, R11-2005-008-00000-0, M10749000201-08N4900-20110); the 21C Frontier Microbial Genomics and Applications Center Program, Ministry of Education, Science, and Technology, Republic of Korea; and the Marine and Extreme Genome Research Center Program of the Ministry of Land, Transportation, and Maritime Affairs, Republic of Korea. References [1] H. Muramatsu, S.I. Mokhtar, M. Katsuoka, M. Ezaki, Phylogenetic analysis of immunosuppressant FK506-producing streptomycete strains, Actinomycetologica 19 (2005) 33–39. [2] T.W. Kang, B.T. Choi, H.S. Kim, S.S. Ryu, Tacrolimus-producing microorganism and its large-scale production method, Korean patent 10/0485877, 2005. [3] T. Kino, H. Hatanaka, M. Hashimoto, M. Nishiyama, T. Goto, M. Okuhara, M. Kohasaka, H. Aoki, H. Imanaka, FK506, a novel immunosuppressant isolated from a Streptomyces: I. Fermentation, isolation, and physico-chemical and biological characteristics, J. Antibiot. 40 (1987) 1249–1255. [4] R. Regentin, L. Cadapan, S. Ou, S. Zavala, P. Licari, Production of a novel FK520 analog in Streptomyces hygroscopicus: improving titer while minimizing impurities, J. Industr. Microb. Biotechnol. 28 (2002) 12–16. [5] C. Vézina, A. Kudelski, S.N. Sehgal, Rapamycin (AY-22, 989), a new antifungal antibiotic: I. Taxonomy of the producing streptomycete and isolation of the active principle, J. Antibiot. 28 (1975) 721–726. [6] G. Sierra-Paredes, G. Sierra-Marcuño, Ascomycin and FK506: pharmacology and therapeutic potential as anticonvulsants and neuroprotectants, CNS Neurosci. Ther. 14 (2008) 36–46. [7] T. Schwecke, J.F. Aparicio, I. Molnár, A. König, L.E. Khaw, S.F. Haydock, M. Oliynyk, P. Caffrey, J. Cortés, J.B. Lester, G.A. Böhm, J. Staunton, P.F. Leadlay, The biosynthetic gene cluster for the polyketide immunosuppressant rapamycin, Proc. Natl. Acad. Sci. USA 92 (1995) 7839–7843. [8] H. Motamedi, A. Shafiee, The biosynthetic gene cluster for the macrolactone ring of the immunosuppressant FK506, Eur. J. Biochem. 256 (1998) 528–534. [9] K. Wu, L. Chung, W.P. Revill, L. Katz, C.D. Reeves, The FK520 gene cluster of Streptomyces hygroscopicus var. ascomyceticus (ATCC 14891) contains genes for biosynthesis of unusual polyketide extender units, Gene 251 (2000) 81–90.
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7 [10] H. Motamedi, A. Shafiee, S.J. Cai, S.L. Streicher, B.H. Arison, R.R. Miller, Characterization of methyltransferase and hydroxylase genes involved in the biosynthesis of the immunosuppressants FK506 and FK520, J. Bacteriol. 178 (1996) 5243–5248. [11] Y. Dai, M.F. Hebert, N. Isoherranen, C.L. Davis, C. Marsh, D.D. Shen, K.E. Thummel, Effect of CYP3A5 polymorphism on tacrolimus metabolic clearance in vitro, Drug Metab. Dispos. 34 (2006) 836–847. [12] M.J. Bogusz, E.A. Enazi, H. Hassan, J. Abdel-Jawaad, J.A. Ruwaily, M.A. Tufail, Simultaneous LC–MS–MS determination of cyclosporine A, tacrolimus, and sirolimus in whole blood as well as mycophenolic acid in plasma using common pretretment procedure, J. Chromatogr. B 850 (2007) 471–480. [13] P. Salm, P.J. Taylor, F.A. Rooney, High-performance liquid chromatography– mass spectrometry method using a novel atmospheric pressure chemical ionization approach for the rapid simultaneous measurement of tacrolimus and cyclosporin in whole blood, Ther. Drug Monit. 30 (2008) 292–300. [14] T. Akashi, T. Nefuji, M. Yoshida, J. Hosoda, Quantitative determination of tautomeric FK506 by reversed-phase liquid chromatography, J. Pharm. Biomed. Anal. 14 (1996) 339–346. [15] T. Koal, M. Deters, B. Casetta, V. Kaever, Simultaneous determination of four immunosuppressants by means of high speed and robust on-line solid phase
[16]
[17]
[18] [19]
[20]
[21]
7
extraction–high performance liquid chromatography–tandem mass spectrometry, J. Chromatogr. B 805 (2004) 215–222. K. Hoogtanders, J. van der Heijden, M. Christiaans, P. Edelbroek, J.P. van Hooff, L.M. Stolk, Therapeutic drug monitoring of tacrolimus with the dried spot method, J. Pharm. Biomed. Anal. 44 (2007) 658–664. H. Tanaka, A. Kuroda, H. Marusawa, H. Hatanaka, T. Kino, T. Goto, M. Hashimoto, Structure of FK506: a novel immunosuppressant isolated from Streptomyces, J. Am. Chem. Soc. 109 (1987) 5031–5033. M.A. Gregory, S. Gaisser, S.J. Moss, H. Petkovic, Production of polyketides and other natural products, WO 04/007709, 2004. H. Hatanaka, T. Kino, M. Asano, T. Goto, H. Tanaka, M. Okuhara, FK-506 related compounds produced by Streptomyces tsukubaensis No. 9993, J. Antibiot. 42 (1989) 620–622. L.E. Khaw, G.A. Böhm, S. Metcalfe, J. Staunton, P.F. Leadlay, Mutational biosynthesis of novel rapamycins by a strain of Streptomyces hygroscopicus NRRL 5491 disrupted in rapL, encoding a putative lysine cyclodeaminase, J. Bacteriol. 180 (1998) 809–814. H. Hatanaka, T. Kino, S. Miyata, N. Inamura, A. Kuroda, T. Goto, H. Tanaka, M. Okuhara, FR-900520 and FR-900523, novel immunosuppressants isolated from a Streptomyces: II. Fermentation, isolation and physico-chemical and biological characteristics, J. Antibiot. 41 (1988) 1592–1601.
Contents lists available at ScienceDirect
Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio
Liquid chromatography–mass spectrometry characterization of FK506 biosynthetic intermediates in Streptomyces clavuligerus KCTC 10561BP Je Won Park, Sang-Joon Mo, Sung Ryeol Park, Yeon-Hee Ban, Young Ji Yoo, Yeo Joon Yoon * Department of Chemistry and Nano Science, Ewha Womans University, Seoul 120-750, Republic of Korea
a r t i c l e
i n f o
Article history: Received 7 January 2009 Available online 17 June 2009 Keywords: Liquid chromatography–mass spectrometry (LC–MS) Tacrolimus FK506 profiling
a b s t r a c t The development of an efficient analytical method for the reliable detection and identification of the biosynthetic intermediates found in microbial cultures, which usually produce complex intermediates of the metabolites of interest, is essential for further biosynthetic investigations. This study developed a simple and highly selective method for detecting the biosynthetic intermediates involved in the FK506 pathway of Streptomyces clavuligerus KCTC 10561BP involving a cleanup procedure using a solid-phase extraction technique to provide reliable extraction of FK506-related compounds from a cell culture broth and liquid chromatography–electrospray ionization–tandem mass spectrometry (LC–ESI–MS/MS) to separate and detect the FK506-related intermediates at concentrations as low as 0.2 lg/L in the broth. This method enabled the analytical profiling of the intermediates formed during the biosynthesis of FK506 in this S. clavuligerus strain, which produced FK506 as a main product. Eight FK506 intermediates—FK520, 37,38-dihydroFK506, prolylFK506, 9-decarbonyl-9-hydroxylFK506, 9-deoxoFK506, desmethylFK520, prolylFK520, and 9-deoxoFK520—were identified. This is the first report of the LC–ESI–MS/MS characterization of a wide range of FK506 analogs from a bacterial fermentation broth. The protocol employed in this study may be useful for estimating the structure of the metabolites without the need for a time-consuming isolation process and nuclear magnetic resonance (NMR) spectroscopy. Ó 2009 Elsevier Inc. All rights reserved.
FK506 (1, also called tacrolimus) is a 23-membered macrolide polyketide that was discovered in 1987 from the fermentation broth of soil bacteria Streptomyces tsukubaensis and later found to be produced by other streptomycete strains such as Streptomyces kanamyceticus and Streptomyces clavuligerus KCTC 10561BP [1,2]. FK520 (2, also known as ascomycin) is an active FK506 analog in which the C21 allyl group is replaced by an ethyl group, is produced mainly by the fermentation of Streptomyces hygroscopicus var. ascomyceticus, and is found as a by-product of S. tsukubaensis [3,4]. Preceded by the discovery of rapamycin in 1975 [5], these natural products were among the first macrolides discovered with immunosuppressive activities. Since receiving U.S. Food and Drug Administration approval in 1994 for use in liver transplantation, FK506 has been licensed as an immunosuppressive drug after solid organ transplantation, and several promising therapeutic applications have been reported [6]. Over the past decade, the advances made in biotechnology in the characterization and manipulation of biosynthetic gene clusters for natural products, such as the polyketides found in microbes, have encouraged the exploration of a wide range of macrolide biosynthetic pathways. Studies of biosynthetic gene
* Corresponding author. Fax: +82 2 3277 3419. E-mail address: [email protected] (Y.J. Yoon). 0003-2697/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2009.06.021
clusters for immunosuppressants, including FK506, FK520, and rapamycin [7–10], have shown that these clusters encode a modular polyketide synthase (PKS),1 a nonribosomal peptide synthetaselike protein, and other biosynthetic genes that are believed to act as post-PKS modification steps. Unfortunately, the biosynthetic pathway for FK506 is not completely understood. For the functional identification and manipulation of some putative or unknown genes for FK506 production, the development of an analytical method for the reliable and efficient detection and identification of FK506-related intermediates in a microbial culture broth is essential. A number of liquid chromatography (LC) methods for quantifying macrolide immunosuppressants with ultraviolet (UV) or electrospray ionization–tandem mass spectrometry (ESI–MS/MS) detection have been reported [11–13]. Most focused on the therapeutic drug monitoring of the residual immunosuppressive macrolides in a variety of biological fluids from clinical trials and pharmacokinetic studies of the metabolites from oral administration of macrolide drugs. These methods are attractive for the simultaneous determination of a set of immunosuppressants but are unsuitable for the selective profiling of the FK506-related bio1 Abbreviations used: PKS, polyketide synthase; LC, liquid chromatography; UV, ultraviolet; ESI–MS/MS, electrospray ionization–tandem mass spectrometry; NMR, nuclear magnetic resonance; SPE, solid-phase extraction; RSD, relative standard deviation; MRM, multiple reaction monitoring; S/N, signal/noise ratio.
2
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
synthetic intermediates in microbial fermentation due to both the structural similarity of FK506-related macrolides and the possible presence of their isomers during chromatographic resolution, making the profiling of the FK506 analogs a challenging task. So far, no attempt has been made to analyze the LC profiles of the types of intermediates implicated in FK506 biosynthesis from a producing strain. The ideal method for detecting the FK506-related biosynthetic intermediates found in a culture broth would be MS because it provides definitive chemical evidence diagnostic for molecular masses and distinct fragmentation patterns. Therefore, LC coupled with MS/MS is the most desirable analytical method for the profiling of FK506 biosynthetic intermediates. This article reports an efficient LC–ESI–MS/MS method for the separation and identification of the different FK506 biosynthetic intermediates found in S. clavuligerus KCTC 10561BP fermentations based on their characteristic MS/MS fragmentation patterns. The practicability of LC–ESI–MS/MS analysis to identify the trace amounts of microbial FK506-related macrolides, which are difficult to isolate in amounts sufficient for their structural elucidation using nuclear magnetic resonance (NMR) spectrometry, is one of the primary advantages of this study. The method for creating the FK506 biosynthetic intermediate profiles developed in this study can be applied to other FK506-producing wild-type and genetically engineered streptomycete strains, and the profiles would be useful in discerning the FK506 and/or FK520 biosynthetic pathways in combination with genetic approaches.
Materials and methods
and SPE cleanup using an Oasis HLB cartridge (Waters). The whole cultures were extracted and partitioned twice using equal volumes of ethylacetate in a polypropylene tube, and the organic extracts were combined and concentrated under vacuum. The crude residues were reconstituted in 2 ml of methanol, after which they were diluted with 10 volumes of H2O and loaded immediately onto an Oasis HLB column that had been previously conditioned with 3 ml of methanol followed by 3 ml of H2O. The column was washed with 6 ml of a 15% (v/v) methanol solution (in water titrated to pH 8.0 using concentrated NH4OH) and then air-dried under reduced pressure for approximately 1 min. The bound macrolides of interest were eluted twice with 1 ml of methanol, evaporated to dryness at room temperature by vacuum centrifugation, and then kept in a freezer. For analyses, the desired residue was reconstituted to 100 ll with methanol, and 10 ll of this solution corresponding to a culture volume of 10 ml (concentration factor of 0.1) was subjected to LC–ESI–MS/MS analysis. The recoveries of both FK506 and FK520 standards spiked into a blank fermentation medium were determined to check both the performance of the SPE cleanup and the matrix effect of the medium on the isolation of the macrolides during the extraction and cleanup steps. They were extracted as described above and analyzed further by LC–ESI–MS/ MS. From a comparison of the chromatographic peak areas obtained from both spiked blank samples with those obtained from the standards, the percentage recoveries were calculated by using at least three replicates. To validate the SPE cleanup procedure more precisely, the intra- and interday precision and accuracy were assessed by analyzing standard 1 spiked (2 and 10 lg/L) into a blank fermentation medium with five replicates on three different days and are presented as relative standard deviations (RSDs).
Reagents FK506-related macrolide LC–ESI–MS/MS analysis Acetonitrile, methanol, glacial acetic acid, ethyl acetate, and water were LC grade and supplied by J.T. Baker (Philipsburg, NJ, USA). MS-grade ammonium acetate was obtained from Fluka Chemie (Steinheim, Germany), and the solid-phase extraction (SPE) cartridge (Oasis HLB, 3 cc/60 mg) and vacuum manifold were obtained from Waters (Milford, MA, USA). FK506 (1) and FK520 (2) were purchased from LC Laboratories (Woburn, MA, USA) and Sigma (St. Louis, MO, USA), respectively. Their stock solutions (1 mg/ ml in methanol) were kept in an amber polypropylene microtube at –20 °C until used to examine the cleanup protocol efficiency using an Oasis HLB SPE cartridge to isolate the FK506 biosynthetic intermediates and tune the optimum ESI–MS/MS parameter for the detection of FK506-related macrolides. All other chemicals were of the highest purity available. Bacterial strain and medium S. clavuligerus KCTC 10561BP was obtained from the Korean Collection for Type Cultures (Daejun, Republic of Korea) [2]. This strain was routinely subcultivated at 28 °C on agar medium containing 1% soluble starch, 0.4% yeast extract, 1% malt extract, and 1.5% agar. The fermentation medium (100 ll) was placed in a 1-L baffled Erlenmeyer flask and sterilized at 121 °C for 20 min, as described previously [2]. The medium was then inoculated with mycelium from S. clavuligerus grown on the agar medium and shake-cultured at 25 °C for 1 week. The whole culture broths were subjected to analysis of the FK506-related macrolides, and all experiments were performed independently in triplicate. FK506-related macrolide extraction and cleanup The extraction and cleanup procedures for the FK506 biosynthetic intermediates in the S. clavuligerus cultures before LC–ESI– MS/MS analysis included organic partitioning via ethylacetate
Analyses of the FK506-related biosynthetic intermediates were performed using a Waters/Micromass Quattro micro MS interface consisting of a Waters 2695 separation module connected directly to a Micromass Quattro micro MS. Separation was performed on a 50 2.1-mm XTerra MS C18 reversed-phase column (3.5 lm, Waters) maintained at 50 °C, and the analytes were eluted at a flow rate of 100 ll/min with a gradient of 5 mM (w/v) ammonium acetate, 0.05% (v/v) acetic acid in H2O (A), and 80% (v/v) acetonitrile with the same additive concentration (B) at 55% B to 70% B for 35 min. The effluent was directed to the ESI–MS/MS operated in positive ion mode. The instrument was calibrated by the direct infusion of a stock solution of authentic FK506 into the ion source at a rate of 40 ll/min. The optimization parameters of the ESI–MS/ MS system were based on the maximum generation, first of the ammoniated molecular ion [M+NH4+] (precursor) and then of the corresponding product ion. The following calibration parameters were used: 0.8 kV capillary voltage, 35 V cone voltage, 120 °C source temperature, 250 °C desolvation temperature, 500 L/h desolvation gas flow rate, and 50 L/h cone gas flow rate. The collision energy in MS/MS mode, which concurred with the full argoninduced fragmentation of the parent ions, was found to be 8 V (pressure reading 3.15 10 6 mbar). The molecular structures and fragmentation patterns of each FK506-related macrolide produced during the cultivation of S. clavuligerus that were obtained under these conditions are shown in Fig. 1. To evaluate the feasibility of this analytical profiling technique for the detection of the FK506 biosynthetic intermediates compared with that of other drug monitoring methods, both the 1 and 2 standards spiked into a blank fermentation medium were quantified using MS/MS in multiple reaction monitoring (MRM) mode. This was carried out by choosing two mass ions (listed in Fig. 1) set to detect a transition of the ammonium adduct precursor to the product ions specific to the selected analytes (1, m/z 822 > 576; 2, m/z 810 > 564).
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
3
Fig. 1. Chemical structure of FK506 and its biosynthetic intermediates in the S. clavuligerus culture broth and ESI–MS/MS fragmentation patterns for identification. FAB, a large fragment ion from the cleavage of both the A and B sites; Fbc, a small ion from cleavage of both the B and C sites; –, no chemical element.
Results and discussion Analytical performance of SPE cleanup In therapeutic drug monitoring of FK506 and its metabolites in human blood and plasma, sample preparation usually involves protein precipitation and solvent dissolution using zinc sulfate and acetonitrile, respectively [13]. However, there are several problems with this protocol when being applied to the analytical profiling of FK506 biosynthetic intermediates from bacterial cultures. For example, the use of water-miscible solvents (i.e., acetonitrile and methanol) to extract the macrolides from matrix-rich bacterial cultures allows retention in the extracts of hydrophilic compounds that would interfere with the subsequent analysis. It has been reported that metal ions, including zinc that remained in the sample ready for LC–MS injection, not only cause the low recovery of FK506 from whole blood but also suppress ionization during MS detection [12]. In particular, streptomycete cultivation medium commonly contains several metal ions such as iron, cobalt, magnesium, and manganese. In addition, even with the use of water-immiscible ethylacetate for extractions, the residual aqueous solution in the extracts can allow the tautomeric behavior of 1 during LC analysis, and this can result in incomplete chromatographic separation and erroneous quantification of FK506-related macrolides, as reported previously [14]. Therefore, it is important to incorporate an additional SPE cleanup to minimize these negative effects. An examination of the analytes recovered from the blank fermentation medium and diluted stock solution spiked with 1 followed by ethylacetate extraction and LC–ESI–MS/MS analysis showed significant differences in the recoveries of 1 (41 ± 8% from a medium and 94 ± 4% from a stock), supporting concerns regarding such matrix effects. There are a few methods for monitoring
immunosuppressant drugs that report an on-line SPE cartridge prior to chromatographic elution [15,16]. In this study, several types of off-line SPE cartridges from different vendors were examined, and an Oasis HLB column was selected due to its ability to enhance the recovery of 1 (>72% from a blank medium) and a cleaner resulting extract that will secure the ESI– MS/MS apparatus. Detailed protocols for cleanup using the cartridge were investigated. It was apparent that the implementation of a slightly alkaline washing step reinforced the recoveries of 1 and 2 to 89 ± 3 and 85 ± 4%, respectively, ensuring the efficient removal of the matrix that may impact chromatographic separation and the MS/MS response. The intra- and interday precision and accuracy using the blank fermentation medium spiked with standard 1 at two different levels were investigated. The precision (RSD) was less than 5%, and the intra- and interday precision and accuracy ranged from 92 to 94% and 92 to 93%, respectively, indicating that SPE cleanup appears to be accurate and precise enough to analyze FK506 and its intermediates produced in the culture broth. Separation of FK506-related macrolide using LC–ESI–MS/MS As generally recognized, 1 and 2 show relatively low affinities to protons during the positive ESI mode but have a high binding capability for ammonium and sodium ions to acquire their adduct ion forms as cationized molecules [11,15]. Although sodium adduct ions tend to be generated in the ion source, they were unacceptable for MS/MS detection, especially for analytical profiling, because of their inadequate fragmentation patterns. In this study, ammonium adduct ions were chosen as precursor ions for MS/ MS, and any contact with potential sodium sources, such as glassware, should be prevented. All experiments performed during the analyses were glassware free.
4
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
The MS/MS spectra of authentic 1 and 2 are presented in Fig. 2A. FK506 had its ammonium adduct precursor m/z peak at 822, which was accompanied by several product ions, including m/z 786, 768, 576 (FAB, a large fragment ion generated by cleavage of both the A and B sites, annotated in Fig. 1), 562, 503 (FBC), 465 (FAD), 336 (Fbc, a small fragment ion generated by cleavage of both the B and C sites), and 265 (Fab), whereas FK520 had a precursor at 810 with fragment ions at m/z 774, 756, 564, 562, 491, 453, 336, and 265. The existence of some identical fragment ions, such as m/z 562, 336, and 265, shows that their structural difference occurred in the polyketide backbone and not in the pipecolate and cyclohexane moieties. However, other fragment ions with a difference of 12 Da could be used as diagnostic ions for a detailed structural analysis of 1 and 2. Therefore, MS/MS detection of both macrolides was performed by monitoring the 822 to 576 mass transition for FK506 and the 810 to 564 mass transition for FK520. From previous reports on the structural elucidation of FK506 and FK520 analogs isolated from the streptomycete fermentation, the FK506 biosynthetic intermediates possess molecular skeletons quite similar to the final product, FK506 (Fig. 1) [17]. Because it is possible that some precursor and product ions may have identical m/z values during LC–ESI–MS/MS analysis, selective chromatographic separation with fully resolved peaks is essential for the reliable profiling of the intermediates. Chromatographic elution of authentic 1 and 2
spiked separately into a blank fermentation medium, followed by the described procedures, was investigated using 5 mM ammonium acetate with 0.05% acetic acid in the mobile phase. The width of the peaks corresponding to FK506 and FK520 decreased with increasing acetic acid concentrations while the elution of early impurities from fermentation was accelerated. Unfortunately, ionization suppression occurred with more than 1% acetic acid, and this limited the utility of this effect. Authentic 1 and 2 had retention times of 29.9 and 27.2 min, respectively, under the above conditions for chromatographic separation of the standard analytes (Fig. 2B). FK506 and FK520 in the medium could be quantified at concentrations as low as 20 ng (with signal/noise ratio [S/N] at 200:1), as determined by the extraction of 100 ml of medium and the injection of 15 ll of the extract after cleanup. Subsequently, analytical profiling of the FK506-related macrolides in the S. clavuligerus culture was performed using the validated cleanup protocol and optimized elution system. Analytical profiling of FK506 biosynthetic intermediates produced in the S. clavuligerus KCTC 10561BP strain In addition to the major final product FK506, eight types of related macrolides were detected in the extract of the S. clavuligerus fermentation broth (Figs. 1 and 3). The identity of each one was as-
Fig. 2. Typical ESI–MS/MS spectra of authentic FK506 and FK520 spiked into the blank medium (A) and their LC–ESI–MS/MS chromatograms traced with selective MRM mode (B).
Fig. 3. Representative LC–ESI–MS/MS chromatograms of FK520 (A), 9-decarbonyl-9-hydroxyFK506 and 37,38-dihydroFK506 (B), prolylFK506 and 9-deoxoFK506 (C), and prolylFK520, 9-deoxoFK520, and desmethylFK520 (D) using their specific MRM mode. 1*, FK506 isotope; 2, FK520; 3, 9-decarbonyl-9-hydroxyFK506; 4, 37,38-dihydroFK506; 5, prolylFK506; 6, 9-deoxoFK506; 7, prolylFK520; 8, 9-deoxoFK520; 9, desmethylFK520.
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
signed based on a comparison of the LC–ESI–MS/MS fragmentation patterns with those of authentic 1 and 2. The appearance of a peak with a 27.2-min retention time (by monitoring m/z 810 > 564, Fig. 3A) showed that S. clavuligerus had produced both FK506 and FK520 even though several publications have reported that these two macrolides are produced separately by different actinomycete strains [2–4]. Two peaks with 22.9- and 42.1-min retention times that were captured by the mass transition from m/z 824 to 578 were presumed to be reduced FK506 biosynthetic intermediates (Fig. 3B). The identical fragmentation patterns, with a precursor ion (m/z 824) and characteristic ions of m/z 578 (FAB) and 467 (FAD), revealed the addition of 2 Da to the corresponding ions of 1 (Figs. 4A and B), suggesting that there was no structural change in the region spanning the pipecolate and cyclohexane moieties (C1–C6 and C24–C34 positions) within the molecules. A peak with a 22.9-min retention time was found to be 9-decarbonyl-9-hydroxylFK506 (3) because the fragment ion at m/z 503 (FBC) was identical to that of FK506 and the ions at m/z 564 (FAC) and 338 (Fbc) were different by 2 Da from the corresponding ions of FK506, showing that a reduction process had taken place within the region between the C8 and C14 positions. Although there has been no report of this proposed compound, it appears to be a biosynthetic intermediate during oxidation from 9-deoxoFK506 (as described below) to FK506 that occurred as a post-PKS step [10]. A mutant with a specific monooxygenase encoding gene rapJ disrupted could produce deoxorapamycin from a recent biotechnological invention that generated a set of rapamycin analogs from the recombinant strains of S. hygroscopicus [18]. This suggests the consecutive oxidation from deoxo-, via hydroxyl-, subsequently to carbonyl function within rapamycin at the C9 position, which is structurally similar to FK506, and the possible existence of 3 in the culture of FK506-producing S. clavuligerus. In the case of the latter peak, the diagnostic fragment ions at m/z 562 (FAC) and 336 (Fbc) were iden-
5
tical to those of FK506, and an ion at m/z 505 (FBC) was 2 Da different from the corresponding ion of FK506. This suggests that reduction had occurred within the region covering the polyketide backbone between the C15 and C23 positions. Therefore, the structure of a compound eluted at 42.1 min was assigned to be 37,38dihydroFK506 (4), which was isolated previously from the S. tsukubaensis culture as an FK506 derivative and named by dihydroFK506 [17]. The peak with a 29.9-min retention time detected by the same mass transition was most likely due to the M+2 isotope peak of 1 (1*, Fig. 3B) that potentially underwent the 824 to 578 MRM transition, which is a transition identical to the MRM transition noted for the two macrolides 3 and 4. This was also observed when analyzing authentic standard 1 by monitoring 824 to 578 (data not shown). Two minor peaks at retention times of 17.7 and 25.8 min that showed a specific m/z 808 to 562 mass transition (Fig. 3C) were also found to be FK506 derivatives. Both macrolides showed identical fragmentation patterns, with a precursor ion of m/z 808 and characteristic ions at m/z 562 (FAB), 548 (FAC), and 322 (Fbc) (Figs. 4C and D). The loss of 14 Da from the corresponding ions of compound 1 suggested the loss of a methylene or carbonyl functional group. The presence of a fragment ion at m/z 465 (FAD) from the 25.8-min peak was identical to one from FK506, indicating that a structural change had occurred in the pipecolate moiety, whereas the fragment ion of m/z 451 (FAD) was unique to the other 17.7min peak. This indicates that a structural modification had occurred within the region spanning the C8 to C14 positions. Therefore, the macrolides with 17.7- and 25.8-min retention times were identified as prolylFK506 (5) and tentatively as 9-deoxoFK506 (6), respectively. ProlylFK506 (5, formerly known as FK525) was isolated from S. tsukubaensis fermentation [19]. During the biosynthesis of rapamycin, which is structurally similar to FK506, a pipecolate-incorporating enzyme encoded by the rapP gene is believed to accept the amino acid l-proline and some of its analogs
Fig. 4. Representative ESI–MS/MS spectra of 9-decarbonyl-9-hydroxyFK506 (A), 37,38-dihydroFK506 (B), prolylFK506 (C), 9-deoxoFK506 (D), prolylFK520 (E), 9-deoxoFK520 (F), and desmethylFK520 (G) from S. clavuligerus fermentation.
6
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7
as substrates, resulting in the production of prolylrapamycin [20]. From these results, 5 appeared to be a congener of the final product 1, depending on the substrate flexibility of a RapP-like FK506 biosynthetic enzyme (e.g., FkbP) [8,9]. On the other hand, 9-deoxoFK506 (6) appeared to be a FK506 precursor for the sequential post-PKS oxidation step, that is, from 6, via 3, and finally to 1 (as described above). From previous functional analysis studies of biosynthetic gene clusters for rapamycin and FK506, the specific genes whose products are categorized as cytochrome P450 monooxygenase are believed to catalyze the above oxidation process: rapJ/rapN for rapamycin and fkbD/fkbO for FK506 [10]. The detection of the above two minor FK506 biosynthetic intermediates 3 and 6 provided valuable information on a full understanding of the biosynthetic pathway of FK506. Therefore, further purification of these compounds will provide indispensable substrates for functional characterization of the putative biosynthetic gene products. Three additional minor peaks with retention times of 15.9, 19.8, and 23.2 min that were captured by a mass transition from m/z 796 to 550 (Fig. 3D) were assumed to be FK520-related biosynthetic intermediates. The mass spectrum of the 15.9- and 23.2-min peaks gave rise to a prominent m/z 491 ion (FBC) identical to that of FK520 but showed the characteristic m/z 550 (FAB), 548 (FAC), and 322 (Fbc) ions (Figs. 4E and F). Again, the loss of 14 Da from the corresponding ions of 2 suggests the loss of a methylene or carbonyl functional group from 2. The main difference in the fragment patterns obtained from both macrolides was the presence of the diagnostic fragment ions m/z 453 (FAD) and 439 (FAD) from the peaks with retention times of 23.2 and 15.9 min, respectively. As observed for other macrolides, such as 5 and 6, this suggests that the substitution of pipecolate by a prolyl moiety and the removal of a 9-carbonyl group had occurred in the structure of 2, resulting in prolylFK520 (7) and 9-deoxoFK520 (8), respectively. Moreover, there was a similarity in the relative peak intensity and shifts in the retention time of prolyl- and 9-deoxo analogs of 1 and 2 compared with those of the corresponding products 1 and 2, supporting the identities of both macrolides 7 and 8. The FK520 analog 7 would be useful for characterizing the substrate specificity of the RapP-like FK506 biosynthetic enzyme or for detecting the existence of other RapP-like enzymes involved in FK520 biosynthesis, whereas the analog 8 would also be valuable for determining whether the above-mentioned oxidation enzymes (probably derived from fkbD/fkbO gene products) involved in FK506 biosynthesis are responsible for similar steps in FK520 biosynthesis or whether other similar enzymes specific to FK520-related intermediates exist in S. clavuligerus. The mass spectrum of the remaining 19.8-min peak monitored by the mass transition m/ z 796 to 550 showed the ion fragments of both m/z 562 (FAC) and 336 (Fbc), which were identical to the corresponding fragments of 2, and the diagnostic fragments of m/z 550 (FAB), 477 (FBC), and 439 (FAD), which were 14 Da different from those obtained from FK520, suggesting demethylation on the polyketide backbone. Based on this information and a comparison of the mass spectrum with the spectrum for 2, a compound with a retention time of 19.8 min was found to be desmethylFK520 (9). This FK520 analog 9 was isolated from FK520-producing S. hygroscopicus var. ascomyceticus fermentation and was formerly described as desmethylascomycin or FK523 [21]. A qualitative comparison of the abundance of the FK506 biosynthetic intermediates detected in the FK506-producing S. clavuligerus culture was carried out by LC–ESI–MS/MS using MRM mode. The estimate was based on the relative intensity of a specific product ion at m/z 265, which is common to the FK506 related macrolides, compared with that of the authentic 1 ([intensity of relevant mass peak] / [intensity of reference mass peak at m/z 265 produced from the standard FK506]), FK506 (1, 85%, 117 mg/L), FK520 (2, 8%, 11 mg/L), 37,38-dihydroFK506 (4, 3%), prolylFK506 (5, >1%), 9decarbonyl-9-hydroxylFK506 (3, 1%), 9-deoxoFK506 (6, <1%), desmethylFK520 (9, >0.5%), prolylFK520 (7, >0.1%), and 9-deoxoFK520
(8, <0.1%). With the exception of 3, all macrolides described in this study were characterized previously and their ESI–MS/MS fragmentation patterns provide sufficient structural data to support their identity. An examination of the analytical profiles from the genetically engineered FK506-producing strains would be helpful for understanding the FK506 and/or FK520 biosynthetic pathways and for characterizing the function of the previously unassigned biosynthetic genes included in the FK506 and/or FK520 biosynthetic gene clusters.
Conclusion This article has described a simple and selective technique for profiling the FK506-related metabolites from a FK506-producing S. clavuligerus strain using a combination of a cleanup protocol employing an Oasis HLB cartridge and optimized LC–ESI–MS/MS conditions, and the technique was found to provide adequate chromatographic separation and detection of FK506/FK520 biosynthetic intermediates at concentrations as low as 20 ng per component from 100 ml of the culture broth. A detailed inspection of the mass spectra of the intermediates allowed differentiation of the structural modifications that occurred during fermentation, generating a profile in which eight intermediates were detected in addition to FK506: FK520, 37,38-dihydroFK506, prolylFK506, 9-decarbonyl-9-hydroxylFK506, 9-deoxoFK506, desmethylFK520, prolylFK520, and 9-deoxoFK520. This approach offers an attractive alternative for overcoming the time-consuming separation process as well as NMR spectroscopy normally needed to determine the structure of these FK506 biosynthetic intermediates or derivatives. Acknowledgments This study was supported by a Korea Science and Engineering Foundation (KOSEF) NRL Program grant funded by the Korean government (MEST, R0A-2008-000-20030-0); a KOSEF grant funded by the Korean government (MEST, R11-2005-008-00000-0, M10749000201-08N4900-20110); the 21C Frontier Microbial Genomics and Applications Center Program, Ministry of Education, Science, and Technology, Republic of Korea; and the Marine and Extreme Genome Research Center Program of the Ministry of Land, Transportation, and Maritime Affairs, Republic of Korea. References [1] H. Muramatsu, S.I. Mokhtar, M. Katsuoka, M. Ezaki, Phylogenetic analysis of immunosuppressant FK506-producing streptomycete strains, Actinomycetologica 19 (2005) 33–39. [2] T.W. Kang, B.T. Choi, H.S. Kim, S.S. Ryu, Tacrolimus-producing microorganism and its large-scale production method, Korean patent 10/0485877, 2005. [3] T. Kino, H. Hatanaka, M. Hashimoto, M. Nishiyama, T. Goto, M. Okuhara, M. Kohasaka, H. Aoki, H. Imanaka, FK506, a novel immunosuppressant isolated from a Streptomyces: I. Fermentation, isolation, and physico-chemical and biological characteristics, J. Antibiot. 40 (1987) 1249–1255. [4] R. Regentin, L. Cadapan, S. Ou, S. Zavala, P. Licari, Production of a novel FK520 analog in Streptomyces hygroscopicus: improving titer while minimizing impurities, J. Industr. Microb. Biotechnol. 28 (2002) 12–16. [5] C. Vézina, A. Kudelski, S.N. Sehgal, Rapamycin (AY-22, 989), a new antifungal antibiotic: I. Taxonomy of the producing streptomycete and isolation of the active principle, J. Antibiot. 28 (1975) 721–726. [6] G. Sierra-Paredes, G. Sierra-Marcuño, Ascomycin and FK506: pharmacology and therapeutic potential as anticonvulsants and neuroprotectants, CNS Neurosci. Ther. 14 (2008) 36–46. [7] T. Schwecke, J.F. Aparicio, I. Molnár, A. König, L.E. Khaw, S.F. Haydock, M. Oliynyk, P. Caffrey, J. Cortés, J.B. Lester, G.A. Böhm, J. Staunton, P.F. Leadlay, The biosynthetic gene cluster for the polyketide immunosuppressant rapamycin, Proc. Natl. Acad. Sci. USA 92 (1995) 7839–7843. [8] H. Motamedi, A. Shafiee, The biosynthetic gene cluster for the macrolactone ring of the immunosuppressant FK506, Eur. J. Biochem. 256 (1998) 528–534. [9] K. Wu, L. Chung, W.P. Revill, L. Katz, C.D. Reeves, The FK520 gene cluster of Streptomyces hygroscopicus var. ascomyceticus (ATCC 14891) contains genes for biosynthesis of unusual polyketide extender units, Gene 251 (2000) 81–90.
LC–MS characterization of FK506 biosynthetic intermediates / J.W. Park et al. / Anal. Biochem. 393 (2009) 1–7 [10] H. Motamedi, A. Shafiee, S.J. Cai, S.L. Streicher, B.H. Arison, R.R. Miller, Characterization of methyltransferase and hydroxylase genes involved in the biosynthesis of the immunosuppressants FK506 and FK520, J. Bacteriol. 178 (1996) 5243–5248. [11] Y. Dai, M.F. Hebert, N. Isoherranen, C.L. Davis, C. Marsh, D.D. Shen, K.E. Thummel, Effect of CYP3A5 polymorphism on tacrolimus metabolic clearance in vitro, Drug Metab. Dispos. 34 (2006) 836–847. [12] M.J. Bogusz, E.A. Enazi, H. Hassan, J. Abdel-Jawaad, J.A. Ruwaily, M.A. Tufail, Simultaneous LC–MS–MS determination of cyclosporine A, tacrolimus, and sirolimus in whole blood as well as mycophenolic acid in plasma using common pretretment procedure, J. Chromatogr. B 850 (2007) 471–480. [13] P. Salm, P.J. Taylor, F.A. Rooney, High-performance liquid chromatography– mass spectrometry method using a novel atmospheric pressure chemical ionization approach for the rapid simultaneous measurement of tacrolimus and cyclosporin in whole blood, Ther. Drug Monit. 30 (2008) 292–300. [14] T. Akashi, T. Nefuji, M. Yoshida, J. Hosoda, Quantitative determination of tautomeric FK506 by reversed-phase liquid chromatography, J. Pharm. Biomed. Anal. 14 (1996) 339–346. [15] T. Koal, M. Deters, B. Casetta, V. Kaever, Simultaneous determination of four immunosuppressants by means of high speed and robust on-line solid phase
[16]
[17]
[18] [19]
[20]
[21]
7
extraction–high performance liquid chromatography–tandem mass spectrometry, J. Chromatogr. B 805 (2004) 215–222. K. Hoogtanders, J. van der Heijden, M. Christiaans, P. Edelbroek, J.P. van Hooff, L.M. Stolk, Therapeutic drug monitoring of tacrolimus with the dried spot method, J. Pharm. Biomed. Anal. 44 (2007) 658–664. H. Tanaka, A. Kuroda, H. Marusawa, H. Hatanaka, T. Kino, T. Goto, M. Hashimoto, Structure of FK506: a novel immunosuppressant isolated from Streptomyces, J. Am. Chem. Soc. 109 (1987) 5031–5033. M.A. Gregory, S. Gaisser, S.J. Moss, H. Petkovic, Production of polyketides and other natural products, WO 04/007709, 2004. H. Hatanaka, T. Kino, M. Asano, T. Goto, H. Tanaka, M. Okuhara, FK-506 related compounds produced by Streptomyces tsukubaensis No. 9993, J. Antibiot. 42 (1989) 620–622. L.E. Khaw, G.A. Böhm, S. Metcalfe, J. Staunton, P.F. Leadlay, Mutational biosynthesis of novel rapamycins by a strain of Streptomyces hygroscopicus NRRL 5491 disrupted in rapL, encoding a putative lysine cyclodeaminase, J. Bacteriol. 180 (1998) 809–814. H. Hatanaka, T. Kino, S. Miyata, N. Inamura, A. Kuroda, T. Goto, H. Tanaka, M. Okuhara, FR-900520 and FR-900523, novel immunosuppressants isolated from a Streptomyces: II. Fermentation, isolation and physico-chemical and biological characteristics, J. Antibiot. 41 (1988) 1592–1601.