Neighboring Group-Controlled Hydrolysis: Towards “Designer” Drug

Bioconjugate Chem. 2007, 18, 209−215

209

Neighboring Group-Controlled Hydrolysis: Towards “Designer” Drug Release Biomaterials Colin P. McCoy,* Ryan J. Morrow, Christopher R. Edwards, David S. Jones, and Sean P. Gorman School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, U.K. Received August 1, 2006; Revised Manuscript Received October 17, 2006

To give the first demonstration of neighboring group-controlled drug delivery rates, a series of novel, polymerizable ester drug conjugates was synthesized and fully characterized. The monomers are suitable for copolymerization in biomaterials where control of drug release rate is critical to prophylaxis or obviation of infection. The incorporation of neighboring group moieties differing in nucleophilicity, geometry, and steric bulk in the conjugates allowed the rate of ester hydrolysis, and hence drug liberation, to be rationally and widely controlled. Solutions (2.5 × 10-5 mol dm-3) of ester conjugates of nalidixic acid incorporating pyridyl, amino, and phenyl neighboring groups hydrolyzed according to first-order kinetics, with rate constants between 3.00 ( 0.12 × 10-5 s -1 (fastest) and 4.50 ( 0.31 × 10- 6 s-1 (slowest). The hydrolysis was characterized using UV-visible spectroscopy. When copolymerized with poly(methyl methacrylate), free drug was shown to elute from the resulting materials, with the rate of release being controlled by the nature of the conjugate, as in solution. The controlled molecular architecture demonstrated by this system offers an attractive class of drug conjugate for the delivery of drugs from polymeric biomaterials such as bone cements in terms of both sustained, prolonged drug release and minimization of mechanical compromise as a result of release. We consider these results to be the rationale for the development of “designer” drug release biomaterials, where the rate of required release can be controlled by predetermined molecular architecture.

INTRODUCTION The design of enhanced performance, particularly drug eluting, bone cement biomaterials is an area of active current interest. Clinically used poly(methyl methacrylate) (PMMA)based acrylic bone cements use a strategy of direct incorporation of a particulate anti-infective agent (typically gentamicin) as prophylaxis against postoperative infection. Particulate drug incorporation represents a suboptimal strategy for two main reasons. First, release is initially a burst from near-surface (1) and is not prolonged (2, 3). Second, the voids left by this initial release act as stress foci for the bulk PMMA material, which act as nucleation sites for cracks (4-6). These cracks can subsequently propagate and compromise the mechanical function of the cement (7). Additionally, the development of these cracks exposes small amounts of additional particulate drug for release, which then elute at subtherapeutic levels, potentially leading to the emergence of resistant infective organisms. A rational strategy to improve this situation is therefore to develop labile drug-polymer conjugates, whereby the copolymerization of the appropriate conjugate allows drug to be liberated from a pendant side chain of the main polymer chain. Such release would move toward elimination of the development of stress foci for crack formation, as a particulate drug is no longer dissolved from the release matrix. Additionally, if the rate of liberation of drug from the drug-polymer conjugate can be controlled, materials can be developed which offer sustained release over longer periods than that offered by current technology. Suitable, labile functional groups which undergo hydrolysis at a variety of rates under differing conditions are esters or amides. However, such bonds are frequently very stable, resulting in an extremely slow bond cleavage and subsequent slow release of drug. Attempts to overcome this * To whom correspondence should be addressed. E-mail: [email protected].

problem have included the use of enzymatically sensitive spacers between the drug and polymer which are designed to break in vivo and thus improve the release profile (8). However, such enzyme-dependent systems require specific, predictable biological conditions, which are frequently not found in biological systems. This can lead to unpredictable drug liberation profiles. Consequently, conjugates which do not rely upon biological enzymes to control the drug release rate make attractive target systems. In particular, hydrolysis of an appropriate drug from a polymer backbone to which it is covalently attached via an ester group is appropriate for adaptation to the fields of bone cement and non-blood contact biomaterials, where esterases are absent, to give both improved release profiles and mechanical properties to drug-eluting drug particle-incorporated strategies in current clinical use. By controlling the rate of drug liberation, the development of “designer” biomaterials, where drug release rate can be controlled precisely and as required clinically, will be possible. A technology which can be adapted to provide control of the hydrolysis rate for drug-polymer conjugates is that of neighboring group-assisted hydrolysis. In this established technology, an intramolecular nucleophile can accelerate the rate of hydrolysis of a hydrolytically labile functional group, such as an ester (9) or amide (10). The process involves incorporating a neighboring group (NG) moiety in an appropriate geometry so as to facilitate intramolecular reaction at the carbonyl carbon of the labile group. This process is termed neighboring group participation (NGP). To accelerate NGP-assisted hydrolysis, the NG must contain a nucleophile, such as that found on the lone pair of a nitrogen atom in an amine (11), a sulfur atom in a thiol or thioacetal (12), or a nitrogen-containing heterocycle such as imidazolyl or pyridyl (13, 14). Phenolate NGs can similarly provide an intramolecular nucleophile through the oxygen bearing a negative charge (15). The molecule must also be able to adopt a conformation whereby minimum strain is required to facilitate the attack of the NG at the carbonyl carbon. NGs

10.1021/bc060238y CCC: $37.00 © 2007 American Chemical Society Published on Web 12/09/2006

210 Bioconjugate Chem., Vol. 18, No. 1, 2007

which form pseudo five- or six-membered rings during the first mechanistic stage of the hydrolysis reaction have been reported to be particularly successful at accelerating overall hydrolysis rates. As an example, Bruice et al. found that the hydrolysis of 4-(2′-acetoxyphenyl)-imidazole was approximately 103 times faster with the intramolecular assistance of an imidazolyl group in a pseudo six-membered ring position than without assistance (16). To apply this technology to enhanced release profile bone cements and related medical device biomaterials, it is first necessary to fully elucidate the effect of different neighboring groups and the conformations adopted during NGP on hydrolysis rates for discrete molecular species (i.e., in solution). Once these effects are fully understood, they inform their performance as pendant drug-polymer conjugates, where additional factors such as porosity and surface density must be considered together with the NGP effect. The current study therefore presents the first account of neighboring group-assisted drug release and examines the effect in solution of various NG-containing drug conjugates of nalidixic acid, an appropriate model quinoline antibiotic. We also demonstrate the drug elution behavior of conjugateincorporated PMMA biomaterials.

METHODS AND MATERIALS Materials. Pyridine-2-carboxaldehyde, pyridine-3-carboxaldehyde, benzaldehyde, allylmagnesium bromide, N-phenylethanolamine, triethylamine, allyl bromide, ethyl chloroformate, methyl methacrylate, and benzoyl peroxide were obtained from Aldrich Chemical Co., Poole, Dorset, U.K. Anhydrous magnesium sulfate, nalidixic acid, hydrochloric acid, and sodium hydroxide were obtained from BDH Chemicals Ltd., Poole, Dorset, U.K. Ethyl acetate, hexane, benzene, acetone, anhydrous dichloromethane, and anhydrous tetrahydrofuran (THF) were of analytical grade and were obtained from Lab-Scan Limited, Dublin, Ireland. Nuclear magnetic resonance spectra were recorded on a General Electric GN-Ω500 instrument operating at 500 MHz. Chemical shifts are given in parts per million (ppm, δ), downfield of tetramethylsilane (TMS), used as an internal standard. A Nicolet Prote´ge´ 460 Fourier Transform infrared (FTIR) spectrometer interfaced with Omnic E.S.P. software was used to record infrared spectra as potassium bromide discs (Sigma, Poole, Dorset, U.K.) from 4000 cm-1 to 400 cm-1. Mass spectra were recorded on a Finnegan MAT 900 XLT highresolution double focusing mass spectrometer operating at 70 eV. Melting points were recorded on an Electrothermal 9100 melting point apparatus and are uncorrected. Elemental analysis was carried out using a Perkin-Elmer PE-240 automatic CHN analyzer. Synthesis of 1-Pyridin-2-yl-but-3-en-1-ol, 1. Pyridine-2carboxaldehyde (2.00 g, 1.78 mL, 18.7 mmol) was dissolved in dry ether (180 mL) in a three-necked round-bottomed flask. Allylmagnesium bromide (24 mmol, 24.0 mL) was injected through a septum, and the reaction mixture was stirred under a stream of dry nitrogen for 2 h. After this time, the reaction was quenched with water (10 mL). A further 100 mL of water was added to the mixture and the product extracted with ethyl acetate (3 × 50 mL). The combined organic phases were dried over anhydrous magnesium sulfate, filtered, and evaporated to dryness under reduced pressure. The crude product was purified by flash chromatography over silica gel using ethyl acetate/ hexane (2:3 v/v) as the mobile phase to afford 2.17 g (78%) of a yellow oil. IR (KBr, cm-1): 3369 (OH), 3075 (dCsH), 2907 (CsH), 1641 (CdC), 1595, 1571 (Ar CdC), 1064 (CO), 1001, 916. NMR δ 1H (CDCl3): 2.51 (1H, m, CH1CH2CHdCH2); 2.62 (1H, m, CH1CH2CHdCH2); 4.47 (1H, s, OH); 4.82 (1H, t, J ) 4.8 Hz, CH2sCHsOH); 5.08 (2H, m, CHdCH2); 5.82 (1H, m, CHdCH2); 7.18 (1H, m, ArsH); 7.31 (1H, d, J ) 7.5

McCoy et al.

Hz, ArsH); 7.67 (1H, d,d, J1 ) 7.7 Hz, J2 ) 7.8 Hz, ArsH); 8.52 (1H, d, J ) 4.9 Hz, ArsH). Synthesis of 1-Pyridin-3-yl-but-3-ene-1-ol, 2. This was prepared by a method analogous to that used for 1, using pyridine-3-carboxaldehyde (2.00 g, 1.78 mL, 18.7 mmol) in place of pyridine-2-carboxaldehyde. The product was purified by flash chromatography over silica gel using ethyl acetate as the mobile phase to afford 2.11 g (76%) of a pale orange oil. IR (KBr, cm-1): 3213 (OH), 3077 (dCsH), 1580 (Ar CdC), 1427, 1061 (CO), 1028 (CN), 918. NMR δ 1H (CDCl3): 2.52 (2H, m, CH2CHdCH2); 4.03 (1H, s, OH); 4.75 (1H, t, J ) 6.5 Hz, CH2sCHsOH); 5.12 (2H, m, CHdCH2); 5.75 (1H, m, CHdCH2); 7.25 (1H, m, ArsH); 7.70 (1H, d, J ) 7.9 Hz, Ars H); 8.37 (1H, d, J ) 4.8 Hz, ArsH); 8.44 (1H, s, Ar-H). Synthesis of N-Allyl, N-Phenyl-2-aminoethanol, 3. NPhenylethanolamine (1.00 g, 0.92 mL, 7.3 mmol) was dissolved in benzene (20 mL). To this solution, triethylamine (7.3 mmol, 1.01 mL) and allyl bromide (7.3 mmol, 0.62 mL) were added. This solution was stirred for 24 h. Benzene was evaporated under reduced pressure, and the reaction mixture dissolved in acetone (20 mL) and cooled in ice for 1 h. This caused precipitation of triethylammonium bromide which was removed by filtration and the filtrate evaporated to dryness under reduced pressure. The product was purified by flash chromatography over silica gel with a mobile phase of ethyl acetate/hexane (2:3 v/v) to afford 0.57 g (44%) of yellow oil. IR (KBr, cm-1): 3384 (OH), 2879 (CsH), 1598 (Ar CdC), 1505, 1037 (CO), 748 (ArsH). NMR δ 1H (CDCl3): 2.01 (1H, s, OH); 3.49 (2H, t, J ) 5.8 Hz, NCH2CH2OH); 3.77 (2H, t, J ) 5.8 Hz, NCH2CH2OH); 3.09 (2H, d, J ) 5 Hz, NCH2CHdCH2); 5.16 (2H, m, CHdCH2); 5.75 (1H, m, CHdCH2); 6.70 (1H, m, ArsH); 6.75 (2H, d, J ) 8 Hz, ArsH); 7.20 (2H, m, ArsH). Synthesis of 1-Phenyl But-3-en-1-ol, 4. This was prepared by a method analogous to that used for 1, using benzaldehyde (1.98 g, 1.89 mL, 18.7 mmol) in place of pyridine-2-carboxaldehyde. The product was purified by flash chromatography over silica gel using ethyl acetate/hexane (4:1 v/v) as the mobile phase to afford 2.44 g (88%) of pale yellow oil. IR (KBr, cm-1): 3381 (OH), 3075 (dCsH), 3029 (ArsH), 2906 (CsH), 1640 (CdC), 1047 (CO), 1000 (CN), 757 (ArsH). NMR δ 1H (CDCl3): 2.00 (1H, s, OH); 2.46 (2H, m, CH2CHdCH2); 4.68 (1H, t, J ) 7.1 Hz, CH2sCHsOH); 5.15 (2H, m, CHdCH2); 5.79 (1H, m, CHdCH2); 7.36 (5H, m, ArsH). Synthesis of [(1-Ethyl-7-methyl-4-oxo-1,4-dihydro-[1,8]naphthyridine-3-carbonyl)-amino]-acetic Acid Ethyl Ester, 6. 1-Ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic acid (nalidixic acid, 3.65 g, 15.7mmol) was dissolved in anhydrous dichloromethane (100 mL). Triethylamine (3.11 mL, 2.26 g, 22.3 mmol) and ethyl chloroformate (2.03 mL, 2.30 g, 21.2 mmol) were added, and the solution was allowed to stir at ambient temperature for 30 min. The reaction mixture was washed with 0.20 M HCl (2 × 20 mL) and water (1 × 40 mL). The organic phase was dried over anhydrous magnesium sulfate, filtered, and evaporated to dryness under reduced pressure. Recrystallization from acetonitrile gave a white solid (3.36 g, 70.4%). Melting point 122-124 °C. IR (KBr, cm-1): 1014 (CsO stretch), 1174 (OdCsOsC stretch), 1243 (CdO stretch), 1609 (phenyl). NMR δ 1H (CDCl3): 1.31 (3H, t, C15H3); 1.41 (3H, t, C11H3); 2.61 (3H, s, C9H3); 4.37 (2H, q, C14H2); 4.48 (2H, q, C10H2); 7.3 (1H, d, C6H); 8.67 (1H, d, C5H); 8.69 (1H, s, C2H). Synthesis of Drug Conjugates. The appropriate alcohol from 1-4 (8.0 mmol) was dissolved in dry THF (100 mL), and sodium hydride (8.8 mmol, 0.211 g) was added. The solution was refluxed for 1 h and cooled to ambient temperature. 6 (8.0 mmol, 2.43 g) was added over 15 min, and the reaction mixture stirred for a further 10 min. Ethanol (5 mL) was added to the

Neighboring Group-Controlled Drug Delivery

solution to deactivate any residual sodium hydride. The reaction mixture was evaporated to dryness under reduced pressure, ethyl acetate (70 mL) was added to the oily residue, and any solid material was removed by filtration to precipitate and remove any nalidixic acid that was present in the reaction mixture. The product was purified by flash chromatography over silica gel using ethyl acetate as the mobile phase in each case to give conjugates 7-10 from alcohols 1-4, respectively. Conjugate 7: 1-Ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic Acid 1-Pyridin-2-yl-but-3-enyl Ester. Yield 34.6%. mp 118-120 °C. IR (KBr, cm-1): 1727 (OCd O), 1619 (CdO), 1439, 1590 Ar CdC), 1082 (CsO), 806 (Ars H). NMR δ 1H (CDCl3): 1.50 (3H, t, J ) 7.33 NCH2CH3); 2.66 (3H, s, ArsCH3); 2.89 (2H, m, CH2CHdCH2); 4.47 (2H, m, NsCH2CH3); 5.09 (2H, d,d, J1 ) 10.2 Hz, J2 ) 15.2 Hz, CHdCH2); 5.89 (1H, m, CHdCH2); 6.17 (1H, t, J ) 7.2 Hz, CO2sCH-pyridyl); 7.18 (1H, m, ArsH); 7.24 (1H, d, J ) 8.1 Hz, ArsH); 7.7 (2H, m, ArsH); 8.57 (1H, d, J ) 4.5 Hz, Ars H); 8.65 (2H, m, ArsH and CdCH). Anal. Calcd for C21H21N3O3 (%): C, 69.41; H, 5.82; N, 11.56. Found: C, 69.81; H, 5.80; N, 11.37. Conjugate 8: 1-Ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic Acid 1-Pyridin-3-yl-but-3-enyl Ester. Yield 38.2%. mp 132-134 °C. IR (KBr, cm-1): 1713 (OCd O), 1624 (CdO), 1442, 1084 (CsO), 805 (ArsH). NMR δ 1H (CDCl3): 1.47 (3H, t, J ) 7.2 Hz, NCH2CH3); 2.66 (3H, s,ArsCH3); 2.74 (1H, m, CH1CH2CHdCH2); 2.86 (1H, m, CH1CH2CHdCH2); 4.47 (2H, m, NsCH2CH3); 5.09 (2H, m, CHdCH2); 5.82 (1H, m, CHdCH2); 6.10 (1H, t, J ) 6.7 Hz, CO2sCH-pyridyl); 7.24 (1H, d, J ) 7.5 Hz, ArsH); 7.30 (1H, m, ArsH); 7.93 (1H, d, J ) 7.9 Hz, ArsH); 8.52 (1H, d, J ) 4.8 Hz, ArsH); 8.58 (1H, s, ArsH); 8.66 (1H, d, J ) 8.1 Hz, ArsH); 8.69 (1H, s, CdCH). Anal. Calcd for C21H21N3O3 (%): C, 69.41; H, 5.82; N, 11.56. Found: C, 69.76; H, 5.94; N, 11.46. Conjugate 9: 1-Ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic Acid 2-(Allylphenylamino)-ethyl Ester. Yield 29.6%. mp 93-94 °C. IR (KBr, cm-1): 1719 (OCdO), 1631 (CdO), 1611 (CdC), 1593 and 1505 (Ar CdC), 1442, 1086, 805. NMR δ 1H (CDCl3): 1.44 (3H, t, J ) 7.2 Hz, NCH2CH3); 2.66 (3H, s, ArsCH3); 3.75 (2H, t, J ) 6.6 Hz, CO2CH2CH2); 4.04 (2H, d, J ) 4.8 Hz, CH2CHdCH2); 4.38 (2H, q, J ) 7.2 Hz, NsCH2CH3); 4.49 (2H, t, J ) 6.6 Hz, CO2CH2CH2N); 5.16 (2H, m, CHdCH2); 5.85 (1H, m, CHd CH2); 6.67 (1H, t, J ) 6.4 Hz, ArsH); 6.82 (2H, d, J ) 8.8 Hz, ArsH); 7.23 (3H, m, ArsH); 8.49 (1H, s, CdCH); 8.67 (1H, d, J ) 8.1 Hz, ArsH). Anal. Calcd for C23H25N3O3 (%): C, 70.57; H, 6.44; N, 10.73. Found: C, 70.75; H, 6.45; N, 10.64. Conjugate 10: 1-Ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic Acid 1-Phenyl -but-3-enyl Ester. Yield 32.0%. mp 124-126 °C. IR (KBr, cm-1): 1724 (OCdO), 1617 (CdO), 1442, 1082 (CsO), 806 (ArsH). NMR δ 1H (CDCl3): 1.45 (3H, t, J ) 7.2 Hz, NCH2CH3); 2.64 (3H, s, ArsCH3); 2.65-2.84 (2H, m, CH2CHdCH2); 4.45 (2H, m, NsCH2CH3); 5.09 (2H, m, CHdCH2); 5.85 (1H, m, CHdCH2); 6.07 (1H, t, J ) 7.2 Hz, CO2sCH-Ph); 7.23-7.28 (2H, m, ArsH); 7.35 (2H, t, J ) 7.8 Hz, ArsH); 7.5 (2H, d, J ) 7.6 Hz, ArsH); 8.57 (1H, s, CdCH); 8.66 (1H, d, J ) 8.1 Hz, ArsH). Anal. Calcd for C23H25N3O3 (%): C, 70.57; H, 6.43; N, 10.73. Found: C, 70.84; H, 6.65; N, 10.79. Synthesis of PMMA-Drug Conjugate Copolymers. Methyl methacrylate (MMA) (20 mL) was washed with 0.10 M sodium hydroxide (3 × 20 mL) followed by deionized water (3 × 20 mL) to remove monomethyl ether hydroquinone inhibitor. The MMA layer was isolated, dried over anhydrous magnesium sulfate, and filtered. To a stirred solution of this MMA (12.0 g), the appropriate conjugate from 7-10 (4% w/w),

Bioconjugate Chem., Vol. 18, No. 1, 2007 211

a loading equivalent to that used in commercial bone cement biomaterials, and benzoyl peroxide (1% w/w) were added. The mixture was injected into a mold comprising two glass plates separated with 1 mm medical grade silicone tubing lined with release liner. The molds were incubated at 60 °C for 2 h. The resulting polymer sheets were washed with deionized water for 12 h to remove residual MMA. Characterization of Hydrolysis of Conjugates 7-10. A solution of each of the conjugates 7-10 (10 mL, 2.5 × 10-5 mol dm-3) was prepared in acetonitrile, and aqueous sodium hydroxide solution (0.10 mol dm-3, 1.0 mL) was added. Hydrolysis was characterized by recording UV-visible absorption spectra between 200-400 nm using a Perkin-Elmer LAMBDA 650 UV-visible spectrophotometer at various time intervals for each solution in a 1 cm path length quartz cuvette. Subsequently, to characterize the products of hydrolysis, the solvent was removed under reduced pressure, yielding a colorless oil. Separation of the products by preparative-scale thin layer chromatography using ethyl acetate as the eluant allowed the identification of the products by means of 1H NMR, infrared spectroscopy, and mass spectrometry. In Vitro Characterization of Drug Release from Copolymers. Copolymers were cut into 1 cm × 1 cm square samples and suspended using a needle in phosphate buffered saline (pH 7.4, 10.0 mL, 0.254 M ionic strength). Samples were incubated in an orbital incubator at 37.4 ( 0.1 °C for defined periods and the release of nalidixic acid quantified using UV-visible spectroscopy. At each sampling, the copolymers were transferred to fresh, prewarmed buffer to maintain sink conditions. For each copolymer, only nalidixic acid could be detected in the release medium. The identity of nalidixic acid was confirmed by 1H NMR and mass spectrometry following preparative-scale thin layer chromatography following release. Statistical Analysis. Differences in the hydrolysis rates, amount released from copolymers, and percentage hydrolysis at 60 min were statistically evaluated using the Kruskal-Wallis test. Individual differences between conjugates were identified using Dunn’s test. In all cases, at least three replicates were analyzed and p e 0.05 denoted significance.

RESULTS AND DISCUSSION Synthesis and Characterization of Drug Conjugates. A series of novel, polymerizable conjugates of nalidixic acid (1ethyl-7-methyl-4-oxo-1,4-dihydro-1,8-naphthyridine-3-carboxylic acid), 5, was synthesized by a three-step process. First, a series of four alcohols, 1-4, were prepared from commercial materials via Grignard reaction with allylmagnesium bromide as summarized in Scheme 1. Alcohols 1, 2, and 4 were prepared from pyridine-2-carboxaldehyde, pyridine-3-carboxaldehyde, and benzaldehyde, respectively, in high yield (76-88%). Alcohol 3 was prepared from N-phenyl ethanolamine and allyl bromide using triethylamine as a base in dry benzene. Following workup and purification by column chromatography, all alcohols were characterized by FTIR and 1H NMR spectroscopies and were determined to be analytically pure and suitable for use in subsequent syntheses. The second synthetic step involved synthesis of a suitably reactive derivative of nalidixic acid, 5, for coupling with alcohols 1-4. Direct esterification, even under catalytic conditions, of nalidixic acid is unsuccessful due to the relatively strong intramolecular hydrogen bond between the hydrogen of the carboxylic acid group with the carbonyl oxygen of the carbonyl group of the naphthyridine ring. The stability of this intramolecular hydrogen bond can be ascribed to the cyclic pseudo sixmembered ring geometry which the hydrogen-bonded system adopts (17). Additionally, synthesis of a reactive acyl chloride derivative has been hampered by the reactivity of the methyl

212 Bioconjugate Chem., Vol. 18, No. 1, 2007 Scheme 1. Synthesis of Alcohol Intermediates 1-4

Scheme 2. Acid, 5

Synthesis of 6, a Mixed Anhydride of Nalidixic

group directly attached to the naphthyridine ring toward thionyl chloride and phosphorus oxychloride, resulting in chlorination of the methyl group preferentially to production of acyl chloride (17). Attempts at esterification using coupling reagents dicyclohexylcarbodiimide (DCC) and 1-(3-dimethylaminopropyl)3-ethylcarbodiimide (EDCI) also failed, which we ascribe to the inhibition of formation of the required imide intermediate by the intramolecular hydrogen bond. A successful strategy for the synthesis of amides of nalidixic acid has employed a mixed anhydride as a reactive intermediate (18, 19). We have adapted this strategy to the synthesis of esters of nalidixic acid by isolating the mixed anhydride formed by the reaction of nalidixic acid with ethyl chloroformate, as shown in Scheme 2, and subsequently reacting this anhydride with an appropriate deprotonated alcohol. Reaction of nalidixic acid with ethyl chloroformate in dichloromethane at 0 °C in the presence of triethylamine gave mixed anhydride, 6, after workup in 70% yield. The third, convergent step of the synthesis involved coupling of the mixed anhydride of nalidixic acid, 6, with each of the alcohols 1-4 separately, to give conjugates 7-10 as summarized in Scheme 3. To achieve efficient reaction, each alcohol was deprotonated with sodium hydride, and nucleophilic attack of the resulting alkoxide with the reactive carbonyl carbon of 6 yielded the corresponding drug conjugate in good yield following workup and purification by column chromatography. Each conjugate was fully characterized using FTIR and 1H NMR spectroscopies, mass spectrometry, and elemental analysis, and each was determined to be analytically pure. All alcohols and conjugates were synthesized as racemates. Nalidixic acid, a broad-spectrum quinoline antibiotic, was chosen as the model antibiotic to demonstrate the system, as it is active against typical pathogens found in infected, cemented hip arthroplasties (20). The design of synthesized conjugates 7-10 is such that each conjugate incorporates nalidixic acid covalently linked via an ester functionality to a moiety (a vinyl

McCoy et al.

group) capable of taking part in free radical polymerization with, for example, acrylates, methacrylates (as used in bone cements), and related free-radical polymerization biomaterials (21). Additionally, a further moiety is included in each conjugate which is designed to either accelerate or retard the rate of liberation of nalidixic acid by hydrolysis. This moiety, separated from the ester via an inert spacer and termed the neighboring group (NG), is varied in the conjugates studied to explore how variations in two key factorssintramolecular nucleophilicity (for facilitated hydrolysis) and steric bulk (for hindered hydrolysis)s affect the rate of drug release. A schematic representation of the molecular design is shown in Figure 1. Conjugates 7 and 8 incorporate a pyridyl moiety which is either 2- or 3-substituted with respect to the potential nitrogen nucleophile. Conjugate 10 shares the steric geometry of these conjugates but lacks the intramolecular nucleophile. Conjugate 9 offers a different (amino) intramolecular nucleophile and was thus expected to offer a further, different rate of hydrolysis. The conjugates were thus designed to exemplify how the ability to act as an intramolecular nucleophile and steric bulk can be simultaneously controlled to give a wide control of hydrolysis, and hence drug release, rates. The choice of and geometry of intramolecular nucleophile were informed by related neighboring group participation ester hydrolysis systems incorporating amines, pyridyl (14), and other nucleophilic moieties such as imidazole (16) and related moieties (13). Characterization of Drug Liberation in Solution. To characterize the drug liberation behavior of the conjugates, hydrolysis studies were carried out in acetonitrile solution containing 1% (v/v) water. Overlaid UV-visible spectra for each of the conjugates 7-10 at various times during the course of hydrolysis show very similar behavior for each conjugate. As a representative example, overlaid spectra for conjugate 8 are shown in Figure 2. Prior to hydrolysis, all conjugates show maxima at 231, 257, and 335 nm (334 nm for conjugate 7 and 336 nm for conjugate 10) with a shoulder at 326 nm (325 nm for conjugate 7). As hydrolysis progresses, these bands show a hypsochromic shift, and the longest-wavelength band with a shoulder resolves, with a hypsochromic shift of 3 nm, to a single band. During the hydrolysis, the spectra for each conjugate show an isosbestic point at 344, 345, 345, and 342 nm for conjugates 7-10, respectively. The closely similar spectral trends for conjugates 7-10 indicate that similar processes occur for each conjugate. Such spectral changes, with a clear isosbestic point, indicate that a single reaction is taking place with no side reactions. The longest-wavelength absorption is, for all conjugates, due to the nalidixate chromophore, which has an extended π-system. Large wavelength changes upon hydrolysis are not expected, as nalidixic acid, with the same π-system, is liberated. The observed 3-nm hypsochromic shift, together with a broadening of the band, indicates the loss of a small, stabilizing interaction arising from the relevant NG. This stabilization may arise from the nonbonded nitrogen lone pair in conjugates 7-9 or, more likely, a small degree of overlap with the aromatic π-system for all conjugates, which is lost upon hydrolysis. Monitoring of absorbance changes with time at an appropriate wavelength therefore provides a sensitive probe to the progress of the hydrolysis reaction. Quantification of the first-order rate constant using spectral data at either 257 nm or 334 nm (r2 > 0.99 in all cases) and the percentage hydrolysis completed after 1 h is shown in Table 1. The only significant difference between the conjugates is the time required for the hydrolysis reaction to proceed to completion, which is ascribed to the nature of the neighboring group. At the conclusion of the hydrolysis experiments, chromatographic separation and identification of the reaction products

Bioconjugate Chem., Vol. 18, No. 1, 2007 213

Neighboring Group-Controlled Drug Delivery Scheme 3. Synthesis of Drug Conjugate Monomers 7-10

showed that free nalidixic acid was liberated in all cases. This is in agreement with related NGP hydrolyses reported in the literature and indicates that drug is liberated in the manner expected. Data from Table 1 demonstrate that the nature of the neighboring group has a very significant effect on the drug

Figure 1. Schematic representation of design of neighboring groupcontrolled drug ester conjugates where D ) drug, NG ) neighboring group, and P ) polymerizable group. Interaction of NG with the ester group is indicated by a dashed line.

Figure 2. Overlaid UV-visible spectra of conjugate 8 after various times during hydrolysis in acetonitrile containing 1% (v/v) 0.10 M aqueous sodium hydroxide. The spectral changes are representative of conjugates 7-10. Arrows indicate trends in spectral change with time. Spectra shown were recorded at times of 0, 0.25, 0.5, 1, 2, 3, 4, 5, 6, 9, and 25 h. Table 1. Hydrolysis Rate Constants and Percentage Hydrolysis after 60 Min for Conjugates 7-10 conjugate

hydrolysis rate constant k (s-1)

hydrolysis after 60 min (%)

7 8 9 10

7.51 ( 0.52 × 10-6 1.79 ( 0.39 × 10-5 3.00 ( 0.12 × 10-5 4.50 ( 0.31 × 10-6

12.5% 30.1% 48.8% 11.1%

liberation characteristics of each conjugate. The order of hydrolysis rate from fastest to slowest is conjugate 9 > 8 > 7 > 10, which is reflected in the percentage of hydrolysis shown in each system after 1 h. After this time, 48.8% of conjugate 9 is hydrolyzed in comparison to 11.1% for conjugate 10. The strong influence of the neighboring group on the hydrolysis rate can be rationalized in terms of the nucleophile strength and relative stability of the required intramolecular intermediate. Conjugate 9, which hydrolyzes most rapidly, has the strongest potential nucleophile of the series. As an aliphatic amine, its NG is significantly more basic than the pyridyl NGs of conjugates 7 and 8. Additionally, the nature of the intermediate can be predicted from the examination of precision space-filling Corey-Pauling-Koltun (CPK) molecular models (Harvard Apparatus, Massachusetts) (22). These models use accurate bond angles according to both the nature and the hybridization of both atoms participating in that bond and also accurately reflect typical van der Waals atomic radii. As such, space-filling CPK models are considered accurate to within 0.3 Å (23). Models show that intramolecular attack of the amine for conjugate 9 is conformationally accessible through an unstrained five-membered intermediate. This conformation is therefore expected to be readily thermally accessible in solution; hence, hydrolysis can proceed relatively rapidly. The trend in hydrolysis rate also implies that the 3-pyridyl NG conjugate, 8, participates more readily as an intramolecular nucleophile than the 2-pyridyl NG conjugate, 7. This is not necessarily expected from a simple examination of the structures; however, a more detailed examination using space-filling molecular models allows a rationale for this behavior to be achieved. For intramolecular nucleophilic reaction of the pyridyl nitrogen with the ester carbonyl carbon, a pseudo five-membered ring must form as an intermediate for conjugate 7, while a pseudo six-membered ring must form for conjugate 8. Sixmembered species are usually considered marginally more stable than five-membered species. Additionally, in the case of conjugate 7, the geometry of the nitrogen lone pair is such that it cannot be in line with the rear side of the carbonyl carbonoxygen σ bond, which is optimal for nucleophilic attack. Conversely, conjugate 8 can achieve this geometry, and the more rapid observed drug release can thus be accounted for in terms of this improved geometry for reaction compared to conjugate 7, given that the nature of the nucleophiles for conjugates 7 and 8 are the same. Conjugate 10, which lacks an intramolecular nucleophile, but has almost identical steric considerations as conjugates 7 and 8, is observed to have the slowest hydrolysis rate. This conjugate demonstrates that each of the conjugates

214 Bioconjugate Chem., Vol. 18, No. 1, 2007

Figure 3. Plot of cumulative release with time for copolymers of MMA with conjugates 7-10. Error bars ((1 standard deviation), in all cases, were less than the indicated point size and have been omitted for clarity.

with an intramolecular nucleophile shows an accelerated hydrolysis rate. Characterization of Drug Liberation from Copolymers. To characterize the in vitro drug liberation behavior of the conjugates in model bone cement biomaterials, conjugates 7-10 were each copolymerized separately with methyl methacrylate with a conjugate loading of 4% (w/w), which is a similar drug loading to commercial drug-eluting PMMA bone cements (14). Release into phosphate buffered saline over 14 days demonstrated that only nalidixic acid was released. The conjugate design is such that the neighboring group remains covalently attached to the copolymer backbone following hydrolysis and is not observed in the release medium, which is important, as only the active drug is released and thus bioavailable. Quantitation of the cumulative relative amount of nalidixic acid released from each copolymer at various times is shown in Figure 3. Importantly, Figure 3 demonstrates that the polymeric matrices give the same order of kinetic release as seen in solution, with the relative release rate decreasing in the order conjugate 9 > 8 > 7 > 10. This reinforces the mechanism of drug liberation and demonstrates that NGP-controlled hydrolysis can applied to polymeric biomaterials. The observed release is primarily from the copolymer surface, as PMMA-based materials are hydrophobic and show only a small degree of subsurface ingress of water. The resulting reduced steric availability of water, and use of physiological pH, leads to a slower release rate than that in solution while retaining the neighboring groupcontrolled trend in release rate. As release is from the material surface only, 0.5-1.5% of the total incorporated drug is released over 14 days. This is in line with the surface release of drug from drug-incorporated PMMA biomaterials (1). Prolonged release from these copolymers introduces several additional considerations, such as the formation of cracks and voids, common in drug-eluting PMMAbased biomaterials (4-6), which provides additional, fresh polymer surface for hydrolytic release, and the consideration of relatively slow drug diffusion from nonsurface, hydrated subsurface areas of the copolymers, which adds an additional phase of release to the hydrolysis-induced surface release demonstrated here (1). Such considerations, particularly of crack and void formation, relate to the important biomaterial considerations of tensile properties (7). The mechanical, porosity, and diffusion aspects are beyond the scope of this paper and will be published elsewhere with data on prolonged release. The system demonstrated here is well-suited to relatively nonporous biomaterials such as PMMA-based bone cement, where surface burst release currently observed would be minimized by the judicious combination of conjugates in a composite biomaterial giving different release rates of the same drug, depending on the nature of the neighboring group, to give

McCoy et al.

a sustained release profile. In conventional, drug-incorporated systems, the rapid (