1972
Biomacromolecules 2010, 11, 1972–1977
Polysaccharide Pharmacokinetics: Amphotericin B Arabinogalactan ConjugatesA Drug Delivery System or a New Pharmaceutical Entity? Anna Elgart,† Shimon Farber,† Abraham J. Domb,† Itzhack Polacheck,§ and Amnon Hoffman*,† Institute for Drug Research, School of Pharmacy, and Department of Clinical Microbiology and Infectious Diseases, The Hebrew University of JerusalemsHadassah Medical Center, P.O. Box 12065, Jerusalem 91120, Israel Received March 18, 2010; Revised Manuscript Received May 30, 2010
Conjugation of poorly soluble drugs to polysaccharides affects their solubility, pharmacokinetics (PK), and pharmacodynamics. The need for amphotericin B (AmB) analog with improved solubility and reduced toxicity is immense. Conjugation of AmB to arabinogalactan (AG) produced a highly soluble AmB-AG conjugate, with high and low molecular weight (H-Mw and L-Mw) fractions. Its similar antifungal activity to AmB poses the question whether AmB-AG is a prodrug of AmB or a novel pharmaceutical entity. We compared the PK of AmB-AG and AmB in rats. Upon AmB-AG administration, no free AmB was released. The half-lives and the volumes of distribution of AmB, H-Mw and L-Mw were 10.9, 8.8, and 1.5 h and 1630, 217, and 133 mL/kg, respectively. We conclude that PK of small molecules conjugated to polysaccharides is mainly dictated by the macromolecular moiety and shows molecular weight dependency. Our findings define AmB-AG as a novel pharmaceutical entity with high clinical potential.
Introduction Bioconjugates consist of biologically active (typically small) molecules covalently coupled to a second moiety, such as polysaccharide, with different but desired physicochemical properties. They are designed to obtain enhanced physicochemical characteristics useful for their specific applications, while maintaining the biological activity. Conjugation of poorly soluble drugs to polysaccharides was suggested before to improve the pharmaceutical properties.1 Such conjugation may increase solubility and stability, reduce toxicity of the parent drug,2 and modify the drug’s PK profile. It may affect the circulation half-life, leading to prolongation of activity, and alter the volume of distribution (Vd) and clearance (CL) in comparison to the parent drug. The physicochemical properties of polysaccharides, including molecular weight, structure, and charge, can significantly affect the PK and pharmacodynamics of the drug conjugate.3 Most of the polymers used clinically are synthetic and nonbiodegradable.4,5 Therefore, only polymers with a molecular weight below the renal filtration threshold (40 kDa) can be used for conjugation. Chemical modifications of biodegradable natural polymers can also lead to the generation of nondegradable adducts.6 Additionally, many are immunogenic and cannot be given repeatedly.7 Larch arabinogalactans (AG) are long, branched polysaccharides of varying molecular weight and high water solubility (70% w/v), approved by the FDA as a source of dietary fiber. The high water solubility, biocompatibility, biodegradability, and ease of drug conjugation make AG an attractive potential drug carrier.8 * To whom correspondence should be addressed. Tel.: +972-2-6757014. Fax: +972-2-6757246. E-mail:
[email protected]. † Institute for Drug Research. § Department of Clinical Microbiology and Infectious Diseases.
Amphotericin B (AmB) is a broad spectrum antimycotic. The need for AmB is continuously growing due to increasing incidences of life-threatening fungal infections, particularly in immunocompromised and AIDS patients.9 However, its clinical use is hampered by dose-related side effects, especially nephrotoxicity and poor water solubility.10,11 Numerous attempts have been made to develop improved AmB formulations, such AmB solubilized with sodium deoxycholate (Fungizone), colloidal dispersion, lipid complex, and liposomal formulation (AmBisome). These formulations resulted in altered PK parameters, efficacy, and toxicity.12-14 Yet, each has specific drawbacks. Liposomal preparations are significantly superior to AmB emulsions or colloidal formulations in terms of bioavailability and side effects. However, these advantages do not always overweigh their high costs.15 The drawback of lipid complexes is that, due to their colloidal characters, they are quickly removed from the circulation by cells of the mononuclear phagocyte system, for example, Kupffer cells, enhancing the risk of hepatic disorders.9 Thus, the search for an optimal AmB formulation is still of great importance. We previously developed an AmB-AG conjugate by attaching AG (20-30 kDa) to AmB. The obtained conjugate is highly water-soluble and contains 20% AmB (w/w). The antifungal activity of AmB-AG against C. albicans and C. neoformans in vitro is similar to that of Fungizone. The efficacy against systemic murine candidiasis and cryptococcosis in vivo was at least as high as Fungizone and AmBisome at equivalent dosages. In vitro and in vivo studies demonstrated a reduced toxicity compared to that of Fungizone.16,17 Previously, attempts were made to determine AmB-AG PK in mice.18 However, that work assessed only free AmB PK, disregarding the fact that the conjugate might represent an independent entity rather than an AmB delivery system. Moreover, the analytical procedure that was utilized was not appropriate for macromolecule quantification. In fact, the obtained results indicate free AmB and L-Mw
10.1021/bm100298r 2010 American Chemical Society Published on Web 07/15/2010
Biodistribution of Polysaccharide Conjugate
AmB-AG conjugate concentrations rather than H-Mw AmB-AG conjugate. Currently, the purification techniques of the synthesis process improved, the free AmB impurities are efficiently removed, hence, enabling a direct assessment of the conjugate PK by superior quantification techniques. Thus, present work upgrades the understandings regarding the disposition of the AmB-AG conjugate. The main purpose of the current study was to assess the PK implications of polysaccharide conjugation to a small molecule. In our case, the PK of AmB-AG conjugate was compared to the PK of free AmB. To accomplish the PK studies, a preliminary aim was set to develop an analytical method for detection and quantification of AmB-AG conjugate in plasma and urine. During the work process, we noticed two peaks on the AmBAG conjugate chromatogram: the first was a H-Mw fraction AmB-AG conjugate and the second had similar retention time to free AmB under the same chromatographic conditions (approximately 85% and 15% of total peak area, respectively). This peak may indicate either free AmB that the purification procedures are unable to remove or AmB conjugated to short AG chains resulting in L-Mw conjugates. Another possibility is that the presence of free AmB as a result of cleavage of AmB from the conjugate. Thus, an additional objective of this work was to establish whether AmB-AG is a novel pharmaceutical entity with unique PK and PD properties or a delivery system for AmB.
Experimental Section Chemicals. AmB was purchased from Alpharma, Copenhagen, Denmark. AmB deoxycholate (Fungizone Bristol-Mayers Squibb, France) was purchased from Hadassah Medical Center. AG was purchased from Larex, MN. All other chemicals were of analytical reagent grade and solvents were HPLC grade. Synthesis of Amphotericin B-Arabinogalactan. A total of 125 mg of AmB (950 U/mg) was added to a solution of oxidized AG (500 mg, 3.87 mmol/g aldehyde groups) in 50 mL of 0.1 M borate buffer (pH 11 ( 0.1), resulting in a final concentration of 2.5 mg/mL of AmB in the solution. The pH of the obtained mixture was maintained at 11 ( 0.1 during the reaction. The mixture was gently stirred in a lightprotected container at room temperature for 48 h and dialyzed against DDW (4 × 5 L) at 4 °C, applying 12000 cutoff cellulose tubing, followed by lyophilization to obtain a deep yellow-orange imine-based conjugate in 85% overall yield. The degree of conjugation was found to be ∼20% w/w (based on the starting feed), as determined by spectrophotometric measurement of AmB (UV, λ ) 390 nm). The amine-based conjugates were obtained after reducing the imines conjugates with excess sodium borohydride (189 mg, 5 mmol) at room temperature for 20 h. The resulting light-yellow solution was dialyzed as described above followed by freeze-drying to obtain a yellowish reduced amine-based conjugate in 80% w/w overall yield and stored at 4 °C. The degree of conjugation was found to be 20 ( 2% w/w (based on the starting feed), as determined by spectrophotometric measurement of AmB (UV, λ ) 390 nm). 1H NMR (D2O): 0.5-2.5 ppm (bm, AmB), 3.73 ppm (bm, AG hydrogens), 5.22 ppm (m, 1H, anomeric hydrogen of AG), 6.47 ppm (m, 7H, polyene hydrogens of AmB). FT-IR (KBr): 1070 (C-O-C), 1417 (CHR1)CHR2 cis), 1572 (polyene CdC), 1645 (conj. CdC), 1792 (CdO), 2116 (intramolecular bonded OH), 3390 (OH), 2933 (anomeric C-H) cm-1. Characterization of the High and Low Molecular Weight Fractions of the AmB-AG Conjugate. High and low molecular weight AmB-AG were separated and fractionized by gel permeation chromatography (GPC-Spectra Physics instrument, Darmstadt, Germany), using size exclusion column Shodex SB-803-HQ (Phenomenex, Japan) and under the following conditions: mobile phase of acetonitrile (35%);
Biomacromolecules, Vol. 11, No. 8, 2010
1973
DDW (65%); flow rate 1 mL/min; using UV detector at 406 nm. Fractions were collected and combined in one single solution. The ACN excess was evaporated and the aqueous solution was lyophilized. After lyophilization, a yellow powder was received which was used for further physico-chemical analysis (solubility tests, AmB content using UV method, 1H NMR and MALDI-MS). Collected fractions constituted from 93% of high Mw fraction and 7% of L-Mw fraction. 1 H NMR spectrum of the H-Mw fraction was similar to the spectrum of the nonfractionized conjugate. 1H NMR spectrum of the L-Mw fraction (DMSO 30%/D2O 70%): 0.5-2.5 ppm (bm, AmB), 3.58 ppm (bm, arabinogalactan, small peaks), 6.293 ppm (m, 7H, polyene part of AmB). Animal Studies. All surgical and experimental procedures were reviewed and approved by the Animal Experimentation Ethics Committee of The Hebrew University Hadassah Medical School, Jerusalem. Male Wistar rats (Harlan, Israel) weighing 300-350 g were kept under a 12 h light/dark cycle with free access to water and food (standard rat chow) prior to investigation. The rats underwent cannulation of the right jugular vein one day prior to the drug administration. Animals were anesthetized for the period of surgery by intraperitoneal injection of 1 mL/kg of ketamine-xylazine solution (90/10%, respectively). An indwelling cannula was placed in the right jugular vein of each animal for systemic blood sampling. The cannula was tunneled beneath the skin and exteriorized at the dorsal part of the neck. After completion of the surgical procedure, the animals were transferred to metabolic cages to recover overnight (12-18 h). During this recovery period and throughout the experiment, food, but not water, was deprived. Animals were randomly assigned to the different experimental groups. The animals were euthanized by overexposure to CO2 after the last sample collection. For AmB experiments, 50 mg AmB (Fungizone vial content) was dissolved in 10 mL of water for injection to obtain an AmB concentration of 5 mg/mL. This solution was further diluted with 5% dextrose solution to obtain an AmB concentration of 125 µg/mL. The final dose for IV infusion administration was 1 mg AmB/kg body weight (∼2.4 mL dosing solution). The solution was prepared immediately prior to administration and administered via the jugular vein cannula over 2 h. Heparinized saline (0.5 mL) was administered immediately following the drug solution to ensure no drug is left inside the cannulation tubing. After dosing the animals were housed in metabolic cages. Blood samples of 400 µL were obtained 0, 2, 4, 6, 8, 10, 22, 30, and 48 h post-infusion. For AmB-AG experiments, AmB-AG solution (3.5 mg/mL) was prepared in 5% w/v dextrose solution for injection. The final dose for IV administration was 20 mg AmB-AG/kg body weight (∼2 mL dosing solution) equivalent to 4 mg/kg body weight AmB. The solution was freshly prepared prior to administration. The drug solution was administered via the jugular vein cannula in a slow IV bolus over 1 min. Heparinized saline solution (0.5 mL) was administered immediately following the drug solution to ensure no drug was left inside the cannulation tubing. After dosing, the animals were housed in metabolic cages. Blood samples of 400 µL were obtained 0, 5, 15, 45 min and 1, 2, 4, 8, 12, 24, 30, and 48 h post-dosing. Urine was collected over 24 h prior to experiment, 0-24 h post-dosing, and 24-48 h postdosing. Urine volume was measured and aliquots were kept at -20 °C, pending analysis. Heparinized saline (450 µL) was administered after each sampling. Blood was collected into heparinized plastic eppendorf tubes and immediately centrifuged at 4000 rpm and 6 °C for 7 min. Each plasma sample was divided into two eppendorf tubes to be analyzed for AmB and AmB-AG contents separately and stored at -20 °C, pending analysis. Analytical Procedures. The concentrations of H-Mw AmB-AG in plasma were measured directly using a size exclusion chromatographic (HPLC) system (Waters 2695 Separation Module) with a photodiode array UV detector (Waters 2996) and fluorescence multi λ detector
1974
Biomacromolecules, Vol. 11, No. 8, 2010
Elgart et al.
Figure 1. Amphotericin B-arabinogalactan conjugate.
(Waters 2475). EtOH (150 µL) was added to 100 µL of plasma. Samples were vortex-mixed for 1 min and further centrifuged at 4000 rpm at 6 °C for 10 min for protein precipitation. The resultant supernatant was transferred to HPLC vials and 90 µL aliquot was injected into the HPLC system. AMB-AG was separated from endogenous material and quantified using Shodex OHpack SB-803 HQ with Shodex SB-LG guard precolumn (Shodex, Kanagawa, Japan). The mobile phase consisted of water/acetonytrile (65:35) and was pumped at a flow rate of 1.0 mL/min. AmB-AG was detected at a wavelength of 406 nm. The column temperature was ambient during the analytic procedure. AmB-AG eluted at ∼6 min and was well separated from the endogenous peaks in plasma. The relationship between the peak area of AmB-AG and the calibration curve was linear (r2 ) 0.990) over the studied range (2-250 µg/mL). The assay had a lower limit of quantification of 2 µg/mL. The concentrations of AmB-AG in urine were measured at the same HPLC system. Ethanol (300 µL) was added to 200 µL urine aliquots. Aqueous fluorescein labeled dextran 4 kDa (FD4) solution (10 µL, 10 µg/mL) was added as an internal standard. Samples were then vortexmixed for one minute and further centrifuged at 10000 rpm for 10 min. The resultant supernatant was then transferred to fresh glass test tubes, and 200 µL of trypsin was added for further protein degradation and precipitation. Samples were then vortex-mixed for 1 min and incubated for 1 h at 37 °C. The samples were transferred to a vacuum evaporator (Vacuum Evaporation System, Labconco, Kansas City, MO) for 1 h at 37 °C and reconstituted with water (130 µL). Aliquots (95 µL) were injected into the HPLC system. AmB-AG was separated from endogenous material and quantified using Shodex OHpack SB-802.5 HQ column. The mobile phase consisted of water and was pumped at a flow rate of 1.0 mL/min. The AmB-AG was detected by a UV detector set on a wavelength of 406 nm at ambient temperature. FD4 was detected by a fluorescence detector: excitation at 490 and emission at 520 nm. The relationship between the ratio of AmB-AG peak area and the FD4 peak height and AmB-AG concentration in urine samples was linear (r2 ) 0.993) over the studied range 25-1000 µg/mL. The assay had a lower limit of quantification of e25 µg/mL. The concentrations of AmB in plasma were measured directly using the same HPLC system. Methanol (200 µL) was added to100 µL rat plasma. Samples were then vortex-mixed for one minute and further centrifuged at 4000 rpm at 6 °C for 10 min and extracted with 3 mL of n-hexane. Following evaporation samples were reconstituted with 130 µL mobile phase and aliquots (80 µL) were injected into the HPLC
system. AmB was separated from endogenous material and quantified using XTerra, RP18, 3.5 lm, 2.1 · 150 mm column (Waters Co.,Milford, MA). The mobile phase consisted of 5 mM EDTA/acetonitrile (57:43) and was pumped at a flow rate of 0.150 mL/min. The AmB was detected at wavelength of 406 nm. The column temperature was set to 35 °C. AmB eluted at ∼5.5 min and was well separated from the endogenous peaks in plasma. The calibration curve was linear (r2 ) 0.997) over the studied range (50-1000 ng/mL). PK Calculations. Plasma concentrations versus time data obtained for AmB and AmB-AG conjugate in individual rats were analyzed by means of both compartmental and noncompartmental analysis model using WinNonlin Professional software version 4.0.1 computer program (Pharsight Company; Mount View, CA). The percentage of the total dose in urine at different times was calculated from the urine concentration (Cu) and urine volume (Vu) by [(CuVu)100]/dose. All data is presented as mean ( SEM, if not stated otherwise.
Results and Discussion A structure of the AmB-AG conjugate is presented in Figure 1. A covalent amine bond is created between AmB and AG in the conjugation process. A typical chromatogram of an AGAmB conjugate is presented in Figure 2. It can be seen that as a result of size exclusion quantification technique that was utilized, the conjugate with a higher molecular weight exit the column earlier than the L-Mw fraction. The H-Mw conjugate is actually an ensemble of variety of molecular weights, which results in a wide peak with two subpeaks indicating a prevalence of two molecular weight ranges of the conjugate. In order to ensure that L-Mw AmB-AG is not unconjugated AmB, fractions were separated using the semipreparative chromatographic method. The differences between high and low molecular weight fractions were then determined following separation by solubility, UV, 1H NMR and MS. Solubility testing showed that H-Mw fractions were freely soluble in water (170 mg/mL), while L-Mw fractions poorly dissolve either in water nor in dimethylsulfoxide (DMSO). L-Mw fractions dissolved in 70% water/30% DMSO solution. In addition, a UV study showed that only 15-17% of AmB concentration was present in the H-Mw fractions, whereas the L-Mw fractions contained 45-50% of AmB. 1H NMR of
Biodistribution of Polysaccharide Conjugate
Figure 2. Typical HPLC chromatogram of an AG-AmB conjugate. The first peak is H-Mw AmB-AG (retention time ∼ 6 min) and the second peak is the L-Mw AmB-AG (retention time ∼ 11 min).
Figure 3. Mean ((SE) plasma concentrations vs time profile of H-Mw AmB-AG following 20 mg/kg IV bolus (n ) 6). The continuous line illustrates the concentrations predicted by the two-compartment PK model.
the L-Mw fractions confirmed the MS results, which showed presence of two to five saccharide units in the same fractions. The relatively high dose of AmB-AG (20 mg/kg equivalent to 4 mg/kg AmB), used in the PK study (4-fold larger than the commonly used dose of AmB) was selected to facilitate the assay procedure. This dose of AmB-AG could be used since it is not associated with toxicity. For the PK assessment of AmB the commonly used dose of 1 mg/kg was utilized and it was tolerated by the animals. This difference in dose does not interfere with the PK comparison because it has been shown before that disposition kinetics of AmB is linear both in rats19 and in humans.20 Hence, the PK parameters are constant in this dose range. The selection of infusion mode of administration of AmB is unavoidable due to its acute toxicity. Due to linear PK, the parameters calculated from this mode of administration are the same as could be determined following bolus administration. On the other hand, the significantly reduced toxicity of AmB-AG enables a bolus administration. Following AmB-AG administration no free AmB was detected in the plasma samples. This finding is in agreement with the fact that the bond created between AmB and AG in the conjugate synthesis process is a covalent amine bond which is stable in plasma and is not subjected to degradation by various enzymes. The mean concentration vs time profile of H-Mw AmB-AG in blood after IV administration of 20 mg AmB-AG/kg body weight to six rats is shown in Figure 3. The concentration-time course data were analyzed by both model independent and compartmental techniques to fit the experimental data. Com-
Biomacromolecules, Vol. 11, No. 8, 2010
1975
partmental analysis demonstrated two clearly differentiated phases (Figure 3). Therefore, the two compartment model was selected as the most suitable to describe the time course. The PK parameters obtained by noncompartmental analysis parameters are presented in Table 1. The two compartmental analyses yielded distribution and elimination half-lives of 0.39 ( 0.12 and 7.95 ( 1.9 h, respectively. Figure 4 represents the mean L-Mw AmB-AG concentration obtained in rat plasma samples following IV administration of AmB-AG (20 mg/kg). L-Mw AmB-AG could be detected in plasma samples up to 8 h after administration. The AUC of plasma concentrations was determined as the average value obtained from the individual rat data. The main PK parameters for the model independent analysis of the concentration time course data are summarized in Table 1. The concentration time course data obtained following IV administration of Fungizone (free AmB 1 mg/kg; Figure 5) were analyzed by model independent method and the main PK parameters summarized in Table 1 are in agreement with those previously reported in literature.21,19,22-26 The very different physicochemical properties of the conjugates with respect to these of the parent small molecule dictate dramatic changes in the corresponding PK profiles. It can be seen that PK parameters obtained for L-Mw AmB-AG following AmB-AG administration, especially the elimination half-life, significantly differ from the PK parameters obtained for AmB after Fungizone administration (Table 1). L-Mw AmB-AG conjugate is characterized by significantly reduced Vd and slightly increased CL compared to free AmB. This fact is in agreement with our initial assumption that conjugation of small molecules to water-soluble polysaccharide results in reduced Vd due to higher water solubility and larger molecular volume. The increased CL is probably a result of the small enough size of the L-Mw conjugate to be readily eliminated through the kidney (
