Study of the metabolism of spiramycin in pig liver

Journal of Chromatography B, 704 (1997) 197–205 Study of the metabolism of spiramycin in pig liver *P. Mourier , A. Brun ˆRhone Poulenc Rorer, Centre de Recherche de Vitry Alfortville, 13 quai Jules Guesde, F-94403 Vitry sur Seine, France Received 23 June 1997; received in revised form 23 September 1997; accepted 23 September 1997 Abstract A major metabolic pathway of spiramycins in pig liver is described. This biochemical reaction involves L-cysteine – a common amino acid present in most animal tissues – which reacts with the aldehyde function of the antibiotic forming a thiazolidine ring. This transformation of spiramycin derivatives drastically increased their polarity. A preliminary HPLC method enabling the quantitation of each metabolite in the range 0.5 mg/g of liver tissue is proposed. Spiramycin S is used as an internal standard while extraction procedures take into account the physico-chemical properties of the thiazolidine moieties. By comparison, previous HPLC methods underestimated the exact amount of antibiotic residues because these metabolites were not extracted from the studied tissues.  1997 Elsevier Science B.V. Keywords: Spiramycin 1. Introduction procedures. Unfortunately, the structure of this me- tabolite remained unknown. Later on, many studies Spiramycin is a mixture of three macrolide anti- were devoted to the pharmacokinetics of spiramycin biotics mainly constituted of spiramycin I (over I and neospiramycin I [2–8] or to the assay of 85%) while spiramycins II and III are minor deriva- spiramycin I [9–14] but no progress occurred on tives (percentages respectively lower than 5 and identification of spiramycin metabolites. The main 10%) (Fig. 1). problem with these studies was their extraction Due to its absence of toxicity at therapeutic doses procedure using chlorinated solvents. Even though it both in man and animals, spiramycin has been has been shown to be valid for spiramycin I and extensively used to overcome various bacterial in- neospiramycin I, it was inefficient to extract the fections (mainly respiratory disorders). For many polar metabolites. years, the characterization and quantitation of metab- The unstability of spiramycin I in liver or kidney olites remained a subject of discussion. First in 1968, has been mentioned [2,3], but additives such as Terlain and Jolles [1] reported the presence of a CoCl [2–4] avoiding the degradation of spiked2 polar conjugate of neospiramycin revealed thanks to spiramycin seemed sufficient to prevent this bio- the wide range of solvents used during extraction chemical phenomenon. In the present work, a study of the stability of spiramycin in pig liver was conducted. It clearly *Corresponding author. revealed the biotransformation of spiramycins into 0378-4347/97/$17.00  1997 Published by Elsevier Science B.V. All rights reserved. PII S0378-4347( 97 )00477-5 198 P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 Fig. 1. Structures of spiramycins, neospiramycins, spiramycins U and S and cysteyl spiramycin. polar cysteyl derivatives. These metabolites were and neospiramycin I in a cystein rich tissue such as found to constitute about 80% of the residues. As a liver underestimated the effective quantity of consequence, a simple quantitation of spiramycin I spiramycin residues by a factor between 10 and 5. P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 199 2. Experimental residues: a 5 g aliquot of ground pig liver was added to 5 ml of water and magnetically stirred for 5 min, 2.1. Animals resulting in a very fine dispersion of the sample. Next, 40 ml of acetonitrile, mixed with 5 ml of a 25 The study of metabolism was conducted on the mg/ l of spiramycin S in acetonitrile, used here as liver of a treated piglet provided for 7 days ad internal standard, were gradually added to the aque- libitum with a feed containing 400 mg/g of ous suspension of the pig liver with continuous spiramycin embonate. It was slaughtered 12 h after magnetic stirring. After 15 min, the mixture was its last feed. centrifuged at 3500 rpm. Five to 20 ml of the supernatant phase were submitted for 10 min under 2.2. Materials and reagents 200 mm Hg bench vacuum to evaporate the acetoni- trile. A 1.5 ml volume of methanol was then added All reagents were of analytical grade. Acetonitrile, to the aqueous residue, and the total volume adjusted methanol and chloroform were obtained from to 5 ml by water addition. The mixture was Prolabo (Paris, France) and were of HPLC-quality homogenized in an ultrasonic bath and filtered grade. Water was purified through a Milli-Q system through 0.2 mm nylon filters. A 2 ml aliquot of the (Millipore, Bedford, MA, USA). filtered mixture was injected into the precolumn of Neospiramycin I, the demycarosyl residue of the HPLC system. spiramycin I was easily obtained from spiramycin I in acidic media (pH,2). 2.4. Structural analysis Authentic cysteyl derivatives were prepared by addition to spiramycin of an excess of L-cysteine NMR experiments were performed in DMSO D ,6 base in a 0.02 M Na HPO –CH OH (70–30 v/v) using a 400 MHz AM400 Bruker Spectrometer.2 4 3 medium. Assignment of all the resonances was obtained Spiramycin U is a natural impurity of spiramycin. thanks to 2D homo- and heteronuclear experiments Spiramycin S was synthesized by reduction of (COSY 45 and HMQC). Mass spectrometry was spiramycin U by NaBH in methanol, and used as performed using a Sciex API III spectrometer in the4 internal standard. electrospray mode. 2.3. Extraction procedure 2.5. Instrumentation and chromatographic method Extraction protocol used at the beginning of our The HPLC system consisted of 3 Gilson Model study of spiramycin metabolism, so called ‘alkaline 305 pumps (Villiers-le-Bel, France). Column tem- methanol’ extraction method, was a simplified ver- perature was controlled by circulating water through sion of an extraction method used for microbiologi- a jacket using a thermo electric heating /cooling cal assays. A 3 g aliquot was sampled and added to accessory. The column switching system was ob- 30 ml of methanol–0.1 M sodium phosphate buffer tained from Touzart and Matignon (Les Ulis, pH 9, (70:30, v /v). The mixture was then stirred for France). Detection was carried out with a Hewlett 30 min and then submitted to 10 min of centrifuga- Packard (Palo Alto, CA, USA) 1040 diode array tion (3500 rpm). The supernatant was separated and detector (DAD). diluted twice with water before HPLC injection. This A basic HPLC system (HPLC Method 1), used at method was assumed to be quantitative for the the beginning of our study, was operated isocratical- extraction of spiramycin residues, but it also ex- ly at room temperature using a reversed-phase tracted other impurities from pig liver, thus hindering Nucleosil C column (25034.6 mm I.D.; 5 mm8 the UV detection, especially when trace level was particle size) (Touzart and Matignon, Les Ulis, reached. France). Mobile phase was a mixture of acetonitrile– This is why another method was developed in 0.05 M sodium phosphate buffer pH 2.2, (65:35, order to enhance the selectivity toward spiramycin v/v) to which 6 g/ l of NaClO were added. The4 200 P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 flow-rate was 1 ml /min. Twenty to 50 ml were Once the precolumn was conditioned (about 10 injected through a conventional Rheodyne injection min), up to 2 ml of extract were injected. The loop. Multi-wavelength detection at 200 nm, 232 nm maximum percentage of methanol in the sample was and 280 nm was achieved with a 1040 DAD. 40% (v/v) in order to trap all spiramycin derivatives Later, the HPLC system had to be adapted to the and especially the most polar one, i.e. those trans- analysis of the transformation products of spiramycin formed by cysteine. (HPLC Method 2, Fig. 2). A preconcentration step Two min after the injection, the precolumn was through a precolumn (Fig. 2) was added to reach switched in series with the column for 2 min. The trace levels present in animal tissues. The HPLC mobile phase used for the separation was a mixture system was operated isocratically at 608C to avoid of 0.05 M sodium phosphate buffer pH 2.3–CH CN3 peak splitting due to the interconvertion of (67:33, v /v); NaClO 6 g/ l. The flow-rate was 1.14 thiazolidine isomers [21]. This system included 2 ml /min. columns: a reversed-phase Nucleosil C column It was not necessary to leave the precolumn in8 (3034.6 mm I.D.; 5 mm particle size) (Touzart and series with the analytical one longer than 2 min. In Matignon, Les Ulis, France) was used for sample that way, the most tightly retained hydrophobic clean up and preconcentration step; a Kromasil C solutes were not able to elute, thus contaminating8 column (25034.6 mm I.D; 5 mm particle size) subsequent analyses due to their late elution time. (Touzart and Matignon, Les Ulis, France) was used So, while separation was occurring, the precolumn for analytical separations. was switched off 2 min after injection and acetoni- The precolumn was initially out of the column trile was circulated through it for 5 min using Pump circuit, and equilibrated with a relatively non-eluting 3. The precolumn was then reconditioned with the solvent through Pump 2. This ‘concentration mobile concentration solvent (Pump 2) while the analysis phase’ was a 0.05 M sodium phosphate buffer pH still went on (Fig. 2). 2.3–CH CN (94:6, v /v) NaClO 6 g/ l. The flow- Approximate retention times obtained are listed in3 4 rate was 1.3 ml /min. Table 1. The close time found for cysteyl spiramycin I and neospiramycin I could explain why these compounds were previously confused. Multi-wave- length detection was identical to that used in HPLC Method 1. The molar response coefficient of all spiramycin derivatives were taken equal to that of spiramycin base, which is rational since the only UV absorbing function of spiramycin, i.e. the two conjugated double bonds of the macrolide, was not implied in Table 1 HPLC retention times of spiramycin derivatives using the pre- column concentration system at 608C (HPLC Method 2) Solute Retention time (min) Cysteyl neospiramycin I 4.34 Cysteyl spiramycin I 5.2 Neospiramycin I 5.7Fig. 2. Three-pump chromatographic system used for spiramycin Spiramycin I 7.75 analysis in pig liver. Valve positions are the following (arrows Cysteyl spiramycin U 7.8indicate the flow of the eluents): t50 min: injection; Valve 1 Cysteyl spiramycin III 10.1(- - - -), Valve 2 (———); t52 min: Valve 1 (———), Valve 2 Spiramycin S 13.2(———); t54 min: Valve 1 (- - - -), Valve 2 (———); t59 min: Spiramycin III 15.5precolumn rinse, Valve 1 (- - - -), Valve 2 (- - - -); t520 min: Spiramycin U 16.8precolumn conditioning, Valve 1 (- - - -), Valve 2 (———). P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 201 the transformation. The quantity of each metabolite The only compounds that remained unchanged was first expressed in quantity equivalent of were the derivatives of spiramycin with a reduced spiramycin S, the internal standard. The response aldehyde group (either CH OH or CH ) such as2 3 coefficient of spiramycin S being taken equal to that spiramycin S. This demonstrated the implication of of spiramycin I, the content of all metabolites were the aldehyde function of the macrocycle in the expressed in equivalent quantity of spiramycin I. reaction. It was also observed that the reaction was partially reversible (between 5 and 10%) in acidic media or when samples were kept a long time in the 3. Results and discussion refrigerator. When the derivatives of spiramycin were added at 3.1. Transformation of the spiramycin derivatives 1000 mg/g in the liver, the transformed compound was present in sufficiently high quantity to be in part The metabolism of the antibiotic by the liver was extracted by chloroform enabling their structural studied extensively. Spiramycin derivatives were determination using mass spectrometry. The ob- added to pig liver, to determine which reactions served molecular weight corresponded to that of the occurred. Several derivatives of spiramycin, such as initial derivatives, increased by 103 Da. If the spiramycin I and III, neospiramycin I or spiramycin hypothesis of the aldehyde reaction was admitted, U were quickly transformed. This transformation can with the loss of H O, it was consistent to consider2 be followed on the chromatograms of the spiked the addition of a molecule of mass 1031185121, extracts by the formation of peaks having identical which considerably reduced the possibilities. In fact, UV spectrum but eluting at lower retention times, L-cysteine was a very likely candidate. which is characteristic of the formation of derivatives As the reaction of spiramycin with L-cysteine can with higher polarity (Fig. 3). be easily obtained in low alkaline (pH.7) media, a Fig. 3. Chromatograms of ‘alkaline methanol’ liver extracts with a starting adduct of 200 mg/g of spiramycin I, of spiramycin III, and spiramycin U. Detection: 232 nm; HPLC Method 1. 202 P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 complete structural determination of the synthesized strength, presence of buffer and temperature. Par- derivative led to the structure shown in Fig. 1. Fig. 4 ticularly, the pH and the cysteine concentration are shows the obtained mass spectrum. of utmost importance. Similar trends are observed 2D NMR experiments (both homo and heteronu- with spiramycin. The reaction is slow in acidic clear) were performed at 400 MHz in DMSO D . media but is faster at increasing pH (about 6).6 The thiazolidine resonances of the 2 isomers (50/50) Complete structural analysis was not conducted on 1 were assigned at: H, d54.8 and 4.6 ppm (dd, the spiramycin adducts obtained after incubation thioaminal protons), 3.9 and 3.6 ppm (dd, a protons) with liver. Nevertheless, the mass spectra of these and 2.8 and 3.1 ppm (dd, b protons); while the spiramycin metabolites were identical to those ob- 13 respective C resonances were determined at d569 tained from synthetic work. In addition, all the and 70 ppm (thioaminal carbons), 65 and 67 ppm (a chromatographic profiles of these compounds were methine carbons) and 30 and 31 (b methylene). The respectively superimposable. chemical shifts pertaining to the other protons were The acidity constants of cysteyl derivatives re- similar to those of the parent compound in the range mained unchanged concerning the two N(CH )3 2 of 60.15 ppm. showing a pK close to 7.7 while the pK of thea a The reaction of aldehydes with L-cysteine is well carboxylic function is 1.5. The polar character of the known and has been extensively described for form- molecule makes it difficult to be extracted in chlori- aldehyde [15–19]. It involves several consecutive nated solvents (dichloromethane, chloroform). This chemical equilibria whose kinetics depend on could be explained by the presence of the carboxylic physico-chemical parameters such as: pH, ionic acid function, fully ionised in alkaline media. The Fig. 4. Electrospray mass spectrum (TIC) of cysteyl spiramycin I (M’51001 is due to traces of cysteyl spiramycin III). P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 203 pK of the NH function of the thiazolidine group was I – were first examined for cysteyl derivatives. Thea much lower than that of amino acids in general, and one set up by Sanders et al. [2–4] consisted of a of cysteine in particular, since it should be about 6.3 chloroformic extraction from various animal tissues [15,17]. such as liver, muscle or kidneys, added with an It is important to point out that the reverse reaction aqueous K HPO solution in presence of CoCl .2 4 2 where cysteyl derivatives are partially transformed After centrifugation, a triphasic system was obtained: back to the starting material may occur in the chloroformic and aqueous phases were present with absence of cysteine and in acidic media. However, an intermediate phase constituted of an emulsion of this reaction is rather slow and minor. Studies [18] liver with chloroform. A sample of blank pig liver on 2-substituted aliphatic thiazolidine-4-carboxylic has been first spiked with a mixture of spiramycin I acid showed that decomposition was mainly obtained and spiramycin S, both at 25 mg/g and then ex- in strong basic solutions (1–3 M NaOH). This tracted following the described procedure. degradation work was not conducted on spiramycin Spiramycin I, cysteyl spiramycin I and spiramycin S due to the opening of the macrocycle in these were determined in each phase. The liver residue conditions. was further consecutively extracted three more times Sanders [2–4] reported that the extensive use of with methanol to check the extraction efficiency. We additional CoCl inhibited ‘spiramycin degradation’ found after summing the contribution of each phase2 in liver. In fact, the complexation of CoCl with that 70–80% of the spiked spiramycin I was trans-2 L-cysteine [22–24] avoided any further reaction of formed into cysteyl spiramycin I. The bottom chloro- the spiked free spiramycin used as standard. On the formic phase contained 10% of the spiramycin other hand the cysteyl spiramycin compounds re- residue: 6% due to the free spiramycin I but only 4% mained unchanged because of the very slow kinetic of cysteyl spiramycin I. The remaining 90% were rate of the back transformation. The extracting present in the top aqueous phase and in the liver solvent used (CHCl ) was too apolar to efficiently emulsion. This assay seemed to show that this type3 extract all the cysteyl derivatives. As a matter of fact, of method extracted about 8% of cysteyl derivatives the partition coefficient K5C /C of the cysteyl and should result in an underestimation of theorg a q derivatives in the system CHCl /buffer pH 9 is residue by a mean factor of 5–10, knowing that the3 between 1.5 and 2. Consequently the given recovery obtained value was not reproducible due to volume rates, calculated on the spiked spiramycin, were not variations of chloroform still emulsified with liver. representative of the overall spiramycin residues As a consequence, extraction with pure methanol effectively present in liver tissue. was considered, because it both extracted spiramycin Reactivity of other compounds chemically close to and its cysteyl metabolites. However, it also ex- cysteine were also tested. It leads to the following tracted other biomolecules interfering with the me- conclusions: when the cysteamine group is present in tabolites. Extractions with pure acetonitrile were the reactant molecule (ethyl or methyl ester of L- unsuccessful, since liver tissues tended to agglomer- cysteine), the reaction occurred. On the other hand, if ate in this medium. Only 40% of the cysteyl conju- the thiol function is protected no reaction could be gates were then extracted. observed (homocysteine, cystine or methionine). The property of water to disperse the liver tissue Glutathione and serine gave no reaction. was used to develop extraction conditions with acetonitrile–water (90:10) (10 ml per g of liver). A 3.2. Extraction methods good dispersion of the liver was then obtained and pollution by polar interfering compounds coextracted The solvent used should lead to the complete from the liver was limited. Attention should be paid extraction of both spiramycin and their cysteyl because evaporation of acetonitrile led to precipi- derivatives. It should thus have a solubilizing capaci- tation while the subsequent addition of methanol was ty over a sufficiently broad polarity range. not sufficient to dissolve all the solid material. That The extracting efficiency of previously published is why the mixture had to be filtered through 0.2 mm methods – valid for spiramycin I and neospiramycin nylon filters. 204 P. Mourier, A. Brun / J. Chromatogr. B 704 (1997) 197 –205 3.3. HPLC method Column temperature appe