Utilization of Quinolone Drugs as Monomers: Characterization of the

Inherently Antimicrobial Biodegradable Polymers in Tissue Engineering. Emma Watson , Alexander M. Tatara , Dimitrios P. Kontoyiannis , and Antonios G...
0 downloads 0 Views 71KB Size
Download PDF
Biomacromolecules 2001, 2, 134-141

134

Utilization of Quinolone Drugs as Monomers: Characterization of the Synthesis Reaction Products for Poly(norfloxacin diisocyanatododecane polycaprolactone) Meilin Yang and J. Paul Santerre* Department of Diagnostics and Biological Sciences, Faculty of Dentistry, University of Toronto, 124 Edward Street, Toronto, Ontario, Canada M5G 1G6 Received August 30, 2000; Revised Manuscript Received November 20, 2000

A broad spectrum antimicrobial agent, 1-ethyl-6-fluoro-1,4-dihydro-4-oxo-7-(1-piperazinyl)-3-quinolinecarboxylic acid (norfloxacin), has been successfully incorporated as a monomer into a polyurethane backbone structure via a three-step polymerization of norfloxacin, diisocyanatododecane (DDI), and polycaprolactone diol (PCL). The reaction was catalyzed by dibutyltin dilaurate and carried out in dimethyl sulfoxide. The sequential order of monomer feeding had a strong influence on the polymerization behavior and final polymer structure. In the preferred reaction scheme norfloxacin is initially reacted with DDI to form an oligomer. This is followed by a second reaction where PCL is introduced in order to produce a drug polymer chain with higher molecular weight and degradable segments. Cross-linking of urea linkages between the norfloxacin and DDI segments was a particular concern and was minimized by feeding PCL into the reaction system immediately following the completion of the first step. Chain extension by 1,4-butanediol or ethylenediamine was shown to be an effective approach for increasing the molecular weight of the polymers. Introduction The use of medical devices fabricated from polymers has increased dramatically over the past 2 decades.1 In parallel with this, there has been a substantial increase in the number of reported bacterial infections associated with biofilm formation on polymeric materials used in biological enviroments.1-4 One approach to establishing the control of infection associated with the polymers is to incorporate antimicrobial agents within the polymers. There is ample evidence in the literature which shows that polymers can be used as carriers for drug delivery and can improve the body’s distribution of therapeutic agents as well as the manner by which cells can take up the agents.5-7 Traditionally, polymers that carry biochemical function have been mainly developed by grafting or coating drug agents on the surface of polymers by chemical or physical processes8-11 or physically entrapping the drug agent within a polymer matrix or micelle.12,13 In the latter systems, the release mechanism of the biochemical agents is mainly based on drug diffusion from a polymer matrix into the tissue. An important issue is that once being initiated, the diffusion process will proceed, without any on/off control, until all of the drug stored in the polymer matrix is released. Quite often this occurs over a short period of time and may result in the delivery of a local and temporary excess of drug.5,14,15 Polymeric carriers have been developed that contain drug moieties as terminal groups, or as pendent groups on the polymer chain. Polymers used for conjugation with drugs * To whom correspondence should be addressed. Phone: 416-979-4903 ext 4341. Fax: 416-979-4760. E-mail: [email protected].

have included poly(R-amino acids),16 polysaccharides such as dextrans and chitin,17 polyurethanes,18-21 and others.22,23 The reports by Nathan are perhaps most relevant to the work discussed in this paper since they relate to polyurethanes. By copolymerizing amino acid moieties into the backbone of the polymer chains, Nathan18-21 et al. have synthesized polyurethanes having pendent drugs to the amino acid unit. The specific conjugation of penicillin V and cephradine as pendant antibiotics to polyurethanes has been reported on by Nathan.20 In the latter work the investigators showed that hydrolytically labile pendant drugs were cleaved and exhibited antimicrobial activities against S. aureus, E. faecalis, and S. pyogenes. Ghosh et al.24 described vinyl monomers with nalidixic acid, a quinolone antibiotic, coupled in a pendant manner to the active vinyl molecule, which was subsequently polymerized. In in vivo hydrolysis studies they reported a 50% release of drug moieties over the first 100 h. This quinolone drug has been shown to be effective against Gram negative bacteria in the treatment of urinary track infections;25 however chemical modifications of the latter (e.g., ciprofloxacin, norfloxacin, and others) have a wider spectrum of activity.26 More recent work on the conjugation of norfloxacin to mannosylated dextran has been carried out in an effort to increase the drug’s uptake by cells, enabling them to gain faster access to microorganisms.27,28 The studies showed that norfloxacin could be cleaved from the drug/ polymer conjugate by enzyme media, and in in vivo studies, the drug/polymer conjugate was effective against Mycobacterium tuberculosis residing in liver.27 In the later system, norfloxacin was attached pendent to sequences of amino

10.1021/bm000087g CCC: $20.00 © 2001 American Chemical Society Published on Web 12/30/2000

Utilization of Quinolone Drugs as Monomers

acids which permitted its cleavage by the lysosomal enzyme, cathepsin B. An alternate approach is to synthesize drug-containing polymers that incorporate the drug units as monomers coupled within the chain of the macromolecule and adjacent to monomers which are sensitive to specific hydrolytic processes within the biological environment. Furthermore, if the cleavage of the bonds between the drug unit and its neighboring units is triggered by a pathological process, such as inflammation, then the release of the biochemical monomers could be timely. Polyurethanes provide an ideal model for the synthesis of copolymeric-type drug polymers such as those described above. These polymers have been widely reported on for their excellent physical properties and interactions with blood components.29,30 Furthermore, polyurethanes have been shown to be degradable31,32 by inflammatory enzymes and their specific susceptibility to enzyme-catalyzed degradation has been shown to be controlled by the structural and chemical arrangements of the units making up the segmented blocks of the polyurethane.33 It is therefore hypothesized that the release of monomers from a polyurethane synthesized with a pharmaceutical agent as a monomer unit could be controlled by the rate of enzyme catalyzed hydrolysis. More importantly, the system could be designed in such a manner that it is specifically sensitive to attack by enzymes that are only expressed under certain clinical conditions.34 For example, when the human body is infected and challenged by bacteria, the inflammatory response is activated35,36 and this upregulates the activity of macrophages.31,36-40 These specific cells have the ability to synthesize an enzyme known as cholesterol esterase41 which has been shown to have a high specificity for urethane and ester linkages in polyurethanes.32,33 Such polymeric materials have recently been reported on for the release of quinolone antibiotics, specifically ciprofloxacin.34 While the latter investigation has confirmed that free ciprofloxacin could be released and that the released compound provided an effective antimicrobial function against P. aeruginosa, when the material was degraded by cholesterol esterase, there was only minimal information provided on the polymerization kinetics and the characterization of the final chemical structures that are produced by their synthesis. This paper reports on the synthesis conditions for polyurethanes having an antimicrobial quinolone drug, specifically norfloxacin, in the backbone, and the manner by which these conditions influence the chemical structure of the polymer and chain growth kinetics of the polymerization. The selection of the drug for the synthesis of the polyurethane in this paper was based on the desire to produce a linear polymer chain rather than a cross-linked matrix; hence a difunctional drug was desired that could react with diisocyanates. A second criterion was that the biochemical molecule should have pharmaceutical activity toward bacteria. Norfloxacin is one of several broad-spectrum fluoroquinolones.26 In this manner, the system could possibly take advantage of the clinical presence of enzymes such as cholesterol esterase in order to deliver a pharmaceutical effect at sites of infection, as examplified in previous work.34

Biomacromolecules, Vol. 2, No. 1, 2001 135

The components of the polymer that will be studied in this paper include norfloxacin (NF), 1,12-diisocyanatododecane (DDI), and polycaprolactone diol (PCL). The structure of NF is given below. It is a broad-spectrum antimicrobial drug with a secondary amine group and a carboxyl group,42 both of which can react with isocyanates to form potentially hydrolyzable linkages. DDI was selected for the synthesis on the basis of the relatively low toxicity of its analogue breakdown product (1,12-diaminododecane) that could be formed during the biodegradation process. To adjust the drug content in the polymer backbone as well as the physical properties, PCL was also used as a monomer. The latter has been previously used to incorporate biodegradability into polyurethanes.32 As well, its breakdown product, caproic acid, is considered to have relatively nontoxicological effects.43

Experimental Section Materials. The reagents used in this study were all purchased from the Sigma-Aldrich Chemical Co., Milwaukee, WI, unless otherwise specified. 1,12-Diisocyanatododecane (DDI) (purity 97%) was vacuum distilled over CaH2 and then was sealed in ampules under nitrogen and stored at room temperature. 1-Ethyl-6-fluoro-1,4-dihydro-4-oxo-7(1-piperazinyl)-3-quinolinecarboxylic acid (norfloxacin (NF), purity 98%) was dried at room temperature for 1 week in a vacuum oven that contained anhydrous calcium sulfate to absorb water. Polycaprolactone diol (PCL) (average molecular weight 2000 g/mol) was dried at 55 °C overnight in a vacuum oven. Dimethyl sulfoxide (DMSO) (purity 99.9%, HPLC grade) was vacuum distilled over CaH2 just prior to use in the polymerization. 1,4-Butanediol (BDO) (purity 99%) was dried by the same method used for drying NF. Dibutylamine (DBA) (purity 99+%) was distilled over CaH2. Ethylenediamine (EDA) (purity 99+%), dibutyltin dilaurate (DBTDL) (purity 95%), N,N-dimethylformamide anhydrous (DMF) (purity 99.8%), N,N-dimethylacetamide (DMAC) (purity 99%), pyridine anhydrous (purity 99.8%), methanol (HPLC grade), diethyl ether, acetic anhydride (purity 97.0%, Anachemia Co., Toronto, ON), acetonitrile (HPLC grade, Fisher Scientific, Springfield, NJ), phosphoric acid (85%, HPLC grade, Fisher Scientific), and tetrabutylammonium hydroxide (40% solution in water) were directly used as received. Monomer Analysis. The isocyanate content of DDI was determined by reacting the compound with excess DBA in DMF, followed by back-titration with hydrochloric acid using bromphenol blue as an indicator.44 The hydroxyl content of PCL was determined by reacting the oligomeric diol with excess acetic anhydride in pyridine, followed by backtitration with potassium hydroxide base using phenolphthalein as an indicator.44

136

Biomacromolecules, Vol. 2, No. 1, 2001

Polymerization. The drug polymer synthesis was carried out under dry nitrogen in a glovebox by using two different sequential orders (labeled as methods PDN and NDP, denoting the first initial of reactants, in the order in which they were introduced into the reaction) in a three-step solution polymerization. For the PDN method, the first step involved the reaction of PCL with DDI; then in a second step the reaction with NF was carried out, which was followed by a third step that consisted of chain extension using BDO. The stoichiometry of the polymerizations was 2.6:1.8:0.6:0.2 of DDI/PCL/NF/BDO. A typical procedure was as follows. In the first step of the reaction, PCL and DDI were placed into 15 mL of DMSO, the mixture was heated to 60 °C, and then 0.1 mmol of catalyst (DBTDL) was added. The reaction mixture was stirred for 3 h, and the system remained clear and colorless throughout the reaction. Following the addition of NF in the second step, the solution immediately became pale yellow but remained clear. The reaction was allowed to proceed at 60 °C for 22 h, and was followed by a chain extension using BDO at 60 °C, and terminated with methanol. The resultant polymer from this method will be referred to as PDNB which conveniently refers to the method and the chain extender type by the letter “B” for butanediol. In the second sequential order (method NDP), the first step was to react NF with DDI; then in a second step the reaction with PCL was carried out. This was followed by the chain extension using BDO or EDA. The stoichiometry of the polymerizations was 2.6:0.6:1.8:0.2 for DDI/NF/PCL/ BDO, respectively, and 2.6:0.6:1.6:0.4 for DDI/NF/PCL/ EDA, respectively. A typical procedure for the polymerization using method NDP is described as follows. In the first step of the reaction, NF and DDI were placed into 10 mL of DMSO contained within a glass vessel. The mixture was stirred using a magnetic bar and heated to 60 °C for 10 min. All NF was dissolved, and the system became a clear and pale yellow color. Then 0.1 mmol of DBTDL was added, and this mixture was stirred at this temperature for 1 h. The second step of the reaction began with the addition of PCL and 5 mL of DMSO into the system. The reaction proceeded at 60 °C for 24 h or overnight. Following this period, the system was clear, pale yellow, and viscous. The third step consisted of a chain extension. BDO was added to the reaction solution at 60 °C, and the reaction was allowed to continue for 10 h, with vigorous stirring. When EDA was used as the chain extender, the system was first cooled to room temperature, and then EDA was added and the reaction continued for 1 h, with vigorous stirring. The polymerizations were terminated with methanol. For the purpose of this study, two polymers were produced using this method and these included NDPB and NDPE, which were chain extended using BDO and EDA, respectively. For the purpose of comparison, a polymer synthesized with DDI and PCL in the absence of NF, i.e., poly(diisocyanatododecane polycaprolactone) [P(DDI-PCL)], was also synthesized by a one-step polymerization with a stoichiometry of 1:1 for DDI/PCL. All other conditions for this latter polymerization were the same as with the first step in method PDN. The polymerization was terminated with methanol.

Yang and Santerre

Conversion of Isocyanates. The extent of polymerization was defined as the percentage of reacted isocyanate groups (pi) and was determined by titrating the remaining isocyanate in samples withdrawn from the system at selected time intervals. The details of the titration were similar to that described for the DDI monomer analysis, outlined above. Polymer Purification. At the completion of the reaction period, the solution was diluted with 15 mL of DMSO. The polymer was then precipitated in a bilayer system of distilled water and ether. The top ether layer was necessary for removing residual DBTDL. The solid was washed using distilled cold water in a Soxhlet extractor for 2 days, followed by stirring in a beaker for 1 day to remove unreacted drug. The washing solutions retrieved from this last step were analyzed for unreacted free drug (see HPLC method below). The resulting polymer was dried in a vacuum oven at 4550 °C overnight. Characterization. Size-exclusion chromatography (SEC) analysis was used to measure molecular weights and was carried out using a Waters instrument (Waters 510 pump, Rheodyne 7125 injector with a 200 µL loop, M490 UV detector set at λ ) 280 nm, and 410 RI detector set at 40 °C, Mississauga, ON, Canada) and using three Waters Styragel columns, HR4, HT3, and HT2, assembled in series. The elution solvent was DMF containing 0.05 mol/L LiBr with a flow rate of 1 mL/min. LiBr was used in the mobile phase to minimize intermolecular aggregation.45,46 The columns were housed in a heating unit, set at 80 °C, and were calibrated by narrow-disperse polystyrene standards (Tosoh Corp., Tokyo, Japan). All molecular weight (MW) data are reported as polystyrene equivalents. The RI signal provides a measure of all polymer components while the UV signal is specifically a measure of the NF component. Thus the combined data of the RI and UV should yield the composition distribution of drug in the polymer. The drug content in the final polymer was calculated from the analysis of elemental fluorine. This was carried out by Galbraith Laboratories Inc. (Knoxville, TN). The sample was mixed with 40-50 mg of sucrose in a Schoniger flask. The mixture was combusted in an oxygen-rich atmosphere and absorbed in 40 mL of total ionic strength adjustor buffer (made up of glacial acetic acid, sodium chloride, and 1,2cyclohexylenedinitriloteraacetic acid, pH 5.10-5.25) for 30 min. The solution was transferred to a polyethylene beaker for reading the fluorine content directly by using a Fisher Acumet Specific Ion Meter MP825 with an Orion Fluoride Electrode (90-01) which was calibrated by standard NaF. A NIST SRM 2143 p-fluorobenzoic acid (13.54%) solution was used as the independent control standard.47 The drug content was also determined directly by UV analysis of polymer solutions in DMAC at 280 nm using an LKB Biochrom Ultrospec II spectrophotometer (model 4050) with deuterium lamp. The instrument was calibrated using NF standard solutions in DMAC. A high-performance liquid chromatography (HPLC) system (Waters, Mississauga, ON, Canada) was used to analyze the residual free drug in polymer washing solutions. The instrument included a Waters U6K injector (Milford, MA), a 4.6 × 250 mm Waters µBondapak C18 steel cartridge

Biomacromolecules, Vol. 2, No. 1, 2001 137

Utilization of Quinolone Drugs as Monomers

Figure 1. Comparison between the SEC of drug polymer PDNB, NDPB, and no-drug polymer P(DDI-PCL). Table 1. Polymer Drug Content drug content in the final polymer (wt %) sample

theoretical

elemental analysis

PDNB NDPB NDPE

4.30 4.42 4.81

3.7 ( 0.4 5.3 ( 0.4 3.2 ( 0.7

column, Waters 996 photodiode array detector, and a Waters 600 pump and controller. The mobile phase consisted of 15% acetonitrile and 85% 0.025 mol/L phosphoric acid adjusted to pH 2.5 with tetrabutylammonium hydroxide. This quaternary ammonium salt acted as an ion-pairing reagent, since NF contains a basic amine group and exhibits severe peak tailings in the absence of such additives. The mobile phase was degassed using a helium sparge. The standards were prepared by dissolving NF in 2 mL of mobile phase, to yield concentrations between 1 and 10 mg/L. The samples were prepared by freeze-drying 40 mL of the polymer washing solution in a polyethylene centrifuge tube. Five milliliters of HPLC mobile phase was then added to redissolve the sample. After the sample was filtered, 5 µL of the sample solution or NF standard solution was injected with a mobile phase flow rate of 1.0 mL/min. Data were recorded and processed on a PC with Waters Millenium software. Results and Discussion Molecular Weight and Drug Content Analysis. Figure 1 shows the dual-detector SEC results for three polymers, PDNB synthesized using method PDN, NDPB using method NDP, and P(DDI-PCL) using method PDN without NF. Since NF absorbs very strongly at 280 nm while DDA (diaminododecane, amine analogue of DDI) and PCL show no absorbance at this wavelength, it is possible to specifically identify the distribution of drug using the UV detector and relating it to the distribution of molecular weight for the polymer by the RI detector. The RI chromatograms for all three polymers exhibit a long tail, indicating the presence of low molecular weight oligomers. The sharp peak, found between 28 and 30 min, was consistently present and was associated with water impurities combined with LiBr in the DMF mobile phase.46 The relative molecular weight values

UV spectrophotometer (at 280 nm)

unreacted free drug based on HPLC analysis (µg/g of final polymer)

4.48 ( 0.19 4.40 ( 0.14 4.64 ( 0.17

0.44 0.42 0.87

were calculated based on the RI peaks found prior to the 28 min retention time. It is observed in Figure 1, that P(DDI-PCL) shows no UV absorbance (UV chromatogram) associated with the polymer peak (RI chromatogram), whereas PDNB and NDPB both show a very strong peak in the UV chromatograms which resulted from the drug contained in the polymers. This confirms the insertion of the drug into the polymer. Fluorine elemental analysis of the polymers and UV spectrophotometer readings (at 280 nm) of polymer solutions in DMAC indicated that the drug content of the final polymers was consistent with the amount of drug fed into the polymerization reactions (see Table 1). From the HPLC analysis of the polymer washing solutions, the values listed in the last column of Table 1 are an estimation of the relative amount of residual free drug in the final polymer and indicated that the free drug content following the reaction was lower than 1 ppm in the final polymer. The long tail in the UV chromatogram for PDNB (Figure 1) indicates that there is a large amount of oligomeric drug containing polymer chains in the final polymer prepared by the PDN method. In contrast, the UV chromatogram for NDPB shows much less drug containing oligomer in the polymer made by method NDP since the UV absorption curve is completely overlapped by the polymer distribution curve given by the RI signal. Figure 2 contains RI chromatograms for NDPE synthesized using the NDP method, at various stages of the reaction. The SEC chromatogram for the product of the first reaction step (i.e., prior to PCL addition) indicates that the molecular weight is very low and is overlapping with the salt impurities peak near the retention time of 30 min. Within approximately 3 h after the addition of PCL into the polymer (i.e., at 4.2 h total reaction time) the second reaction step is effectively complete since there was relatively little change in the

138

Biomacromolecules, Vol. 2, No. 1, 2001

Yang and Santerre

Table 2. Results of Polymerizations with Different Sequential Order of Reaction Steps second step 10-4

first step a

run

r1 (%)

PDNB NDPB NDPE

68.6 23.6 23.6

r2b

(%)

23.6 67.9 61.2

Mn × (g/mol) 1.0 1.3 2.0

final polymer 10-4

Mw × (g/mol) 1.8 2.2 3.8

third step CEX BDO BDO EDA

10-4

Mn × (g/mol) 1.6 1.8 2.5

Mw × 10-4 (g/mol) 3.5 4.1 7.0

a r denotes [PCL]/[DDI] for method PDN and [NF]/[DDI] for method NDP representing the percent of reactive groups for a compound relative to the 1 total NCO groups. b r2 denotes [NF]/[DDI] for method PDN and [PCL]/[DDI] for method NDP representing the percent of reactive groups for a compound relative to the total NCO groups.

Figure 3. Extent of isocyanate group conversion pi vs reaction time in the first and second steps of the polymerizations. Figure 2. SEC RI profiles of polymerization NDPE at different reaction times.

distribution of the polymer batch between 4.2 and 25.5 h. Following the addition of EDA, there is a further increase in molecular weight and broadening of the distribution. NCO Reactivity for the NDP and PDN Methods. Data in Table 2 show comparisons of molecular weight values for the two methods and the different chain extenders with method NDP. It can be seen that the synthesis of the final polymer using the PDN method resulted in a slightly lower molecular weight product than did the polymerization using the NDP method with the same chain extender. This was observed both before and after chain extension with BDO. A comparison of the reaction stoichiometry (r1 + r2, defined in Table 2) from Table 2 with the reacted NCO values (pi) at the end of the second reaction step (25 h, Figure 3) shows that when using the NDP method the (r1 + r2) value was 91.5% and the pi value was near 90.4%, meaning that the reaction was carried out to almost full completion. However in a similar comparison for PDNB, about 10% of the NCO groups had not reacted at the completion of step 2 when using method PDN, i.e., (r1 + r2) ) 92.2% (Table 2) versus pi ) 82% (Figure 3). Figure 3 shows that the first reaction step in the PDN method was fast during the first hour and then slowed considerably. The pi value at 3 h for PDNB was 64.8% and was equivalent to the reaction of approximately 94.5% of the hydroxyl groups in PCL. A further prolongation of the

first reaction step to allow the further reaction of the remaining hydroxyl groups could have led to branched and cross-linked products by side reactions because the system contained excess -NCO groups and the reaction temperature was high (i.e., 60 °C). Hence, further reaction was not pursued. Following the addition of NF to the PDN reaction at 3 h, the second reaction step proceeded at a constant rate (R2 for the slope is 0.9882) until 8.4 h (Figure 3). Since the reactivities of the chemical groups in the monomers toward the -NCO group are anticipated to decrease in the order amine > alcohol > carboxyl,44 it can be assumed that the amines in the NF molecule reacted first and dominated this phase of the reaction. After this the predominant reactant species for NCO would be the carboxyl groups of NF and the residual hydroxyls of PCL. The lower reactivity of these species in combination with the higher molecular weight of the reactants, which led to a more viscous medium and thus limiting diffusion, is believed to have rendered the reaction unfavorable. As a result, the reaction of NCO groups only achieved about 82% completion (Figure 3). The polystyrene equivalent molecular weight data revealed that the average molecular weight values of PDNB after the first reaction step were Mn )1.1 × 104 and Mw ) 1.7 × 104 g/mol. It is also observed from Table 2 that the latter values show virtually no change following the second reaction step. This observation suggests that the drug units are mainly linked to the ends of the polymer chain produced from the first reaction step for PDNB and very little extension of the

Utilization of Quinolone Drugs as Monomers

polymer chains takes place. Since the chains that are capped with -COOH and >NH endgroups cannot react with BDO in the third chain extension step, there will be a significant number of oligomeric chains in the final product. This is reflected by the long tail in the UV chromatogram of the SEC for the final polymer (see Figure 1, PDNB). Given the poor reactivity for the PDNB system, it should be considered that some side reactions, such as branching and cross-linking resulting from -NCO groups reacting with the -NH- groups of the urethane and urea linkages of the polymer chains, may be taking place during the second reaction step in the PDN method. Since the concentrations of the -OH and -COOH groups at this stage of the polymerization were decreasing while the concentration of -NH- segments on the polymer chain was increasing, as well as the fact that the excess -NCO groups were maintained at a relatively high temperature (60 °C) for more than 20 h, the possibility for branching to occur was increased. Although the PDN method was attempted several times by changing various parameters (i.e., catalyst, concentration, temperature, stoichiometry, drug feeding time, etc.), the final polymers always contained some insoluble substances, which were suspected of being cross-linked reaction products. The group having the lowest activity toward the isocyanate in the polymer synthesis was the carboxylic group44 of NF. The poor reactivity of this group appears to be less of a problem in the NDP method than in the PDN method discussed above. The relatively lower molecular weight of the products in the first reaction step of the NDP method (leading to lower viscosities and better diffusion of reactant moieties) along with the presence of high excess levels of -NCO group at the time of adding the -COOH groups into the reaction vessels are thought to have favored the reaction between isocyanate and carboxylic acid groups. Figure 3 shows that the reaction between NF and DDI (i.e., the first reaction step) in the NDP method (the solid line, with a slope 24.8) was much faster than that of the drug reaction in the second step for the PDN method (the doted line with a slope 2.8). The pi (23.6%, Figure 3) and r1 (23.6%, Table 2) values for the first step of the NDP method indicate that both the >NH and -COOH groups of NF underwent also complete reaction within 1 h. The relatively short reaction time in the first step leads to a reduced chance of self-addition between the excess isocyanate groups.35 A comparison of the estimated initial reaction rates (i.e., within the first 3 h, Figure 3) for the second step of both systems shows that the NDP method was much faster than the initial reaction period of the PDN method (i.e., a slope of 18.5 versus 2.8, respectively). The fast rate for the NDP method is supported by the observation of a rapid increase in molecular weight for this step in Figure 2. In addition the pi continued to increase linearly in the latter period of step 2 for the NDP reaction whereas this was not observed for the PDN method (Figure 3). These results indicate that the products of the first step reacted relatively well with PCL during the second reaction step, unlike the scenario discussed above for the PDN method. The molecular weight of the NDP product, before chain extension was between 1.5 and

Biomacromolecules, Vol. 2, No. 1, 2001 139

Figure 4. Extent of isocyanate group conversion pi vs reaction time in the first step of method NDP, comparing different initial ratios of norfloxacin to diisocyanate.

2 times that of the PDN product at the same stage of the synthesis (Table 2). Chain Extension. Chain extension appears to be an effective manner of increasing the molecular weight of the drug polymers. Data in Table 2 show that the Mw of the polymer is increased by about 75-100% over that of step 2. This was observed for either method, PDN or NDP, and was consistent for either of the chain extenders, EDA or BDO. The relatively small effect of chain extension on Mn indicates that a significant number of high molecular weight chains were involved in the chain extension. The SEC chromatograms in Figure 2 also depict the shift to higher molecular weight chains following chain extension. The use of EDA as a chain extender permitted the reaction to be completed within 1 h. However, with the latter chain extender the temperature of the system must be cooled and maintained at room temperature in order to avoid branching and crosslinking reactions of -NCO end groups with the amines of the urethane and urea groups. When BDO was used, the chain extension was slow (20 h) and required a higher temperature (60 °C) because of the low reactivity between -OH and -NCO groups. Reaction between DDI and NF. A further analysis of the reaction for step 1 of the NDP method was carried out in order to assess the possibility of optimizing the favored reaction and minimizing side reactions. Figure 4 shows the pi values of four reactions with different initial ratios of [NF]/ [DDI] in DMSO. These reactions were carried out at 60 °C, and no PCL was added. It is obseved that the reaction rate increased with an increasing ratio of [NF]/[DDI]. The four systems were all homogeneous at the beginning of the reaction; however, precipitates gradually appeared and produced cloudiness at different times for the different runs. The precipitation occurred earlier for the reactions having the higher [NF]/[DDI] ratios. At the [NF]/[DDI] ratio of 0.200 and 0.236, there was no precipitation observed within the first hour reaction period, which was equivalent to step

140

Biomacromolecules, Vol. 2, No. 1, 2001

1 of method NDP. On the basis of the data in Figure 4, it was observed that the pi value at which the solution became cloudy within the first hour corresponded closely to the molar ratio of the reactants. Since the reaction of -NCO groups in systems with [NF]/[DDI] values of 0.243 and 0.387 continue to progress beyond the available amount of >NH and -COOH groups on NF, it is concluded that alternate reactions must be occurring. They are likely cross-linking or branching reactions with the amines in the urethane or urea groups of the polymer chain, which produce macromolecules that are prone to precipitation. While polymerizations having [NF]/[DDI] ratios of 0.200 and 0.236 did not exhibit an -NCO group consumption greater than their respective ratios within the first hour at 60 °C, they did show additional -NCO consumption and polymer precipitation after subsequent reaction periods of 20-25 h (respectively) at room temperature. Hence, to avoid cross-linking reactions in the NDP method, it is important that PCL be fed immediately into the system following the completion of the first reaction step. The critical importance of this will be dependent on the actual [NF]/[DDI] ratio being used. Conclusion Norfloxacin was successfully synthesized into the backbone of a polyurethane by a three-step polymerization catalyzed by DBTDL in DMSO. Although there could be concerns over the use of DBTDL and DMSO for the synthesis, in terms toxicity, it has been generally found in the biomedical polyurethane literature29 that if appropriately extracted, the use of these agents in the synthesis does not translate into a toxicity issue in the final material. In previously reported work with ciprofloxacin polyurethanes, residual levels of tin were not found in the washed polymers34 and more recently cytotoxicity studies with mouse fibroblast cells cultured directly on the ciprofloxacin drug polymer showed no sign of abnormal cell behavior, relative to cells cultured on tissue culture polystyrene.48 The sequential order of monomer feeding influenced the polymerization process and the final polymer structure. A polymerization sequence allowing for the reaction of norfloxacin and DDI in the initial reaction step followed by a second reaction with PCL produced higher molecular weight drug polymers. To avoid cross-linking, PCL must be fed into the system as soon as the -NCO groups have saturated the functional groups on the NF molecule. Chain extension by BDO or EDA can successfully increase the molecular weight of the final polymer. Work is now in progress on the study of the rates of polymer degradation in the presence of enzymes and human macrophage cells for a norfloxacin polymer, synthesized using the favored NDP method. It should however be noted that other work with poly(ciprofloxacin diisocyanatododecane polycaprolactone), synthesized using a protocol similar to that of the PDN method reported in this paper, has examined issues related to the toxicity of released degradation products, efficacy of the drug release and degradation by human neutrophils.48 These studies have shown that mouse fibroblast cultures exposed to incubation

Yang and Santerre

solutions, containing enzyme generated polymer hydrolysis products, were not toxic, and the released drug from macrophage associated enzymes and human neutrophils showed an effective kill of P. aeruginosa and E. coli bacteria over that of controls. Furthermore, the isolated antimicrobial activity was confirmed to be that of free ciprofloxacin. The ability of the fluoroquinolone to be incorporated into the backbone of a polymer chain, subsequently cleaved by enzymes synthesized within macrophages and then exhibit antimicrobial activity further extends the work of Rosseeuw et al.27 on drug polymers with pendent norfloxacin. Since the drug used in the above polymer48 has virtually the same auxopharmacore as that of the norfloxacin polymers used in the current study, it is anticipated that biological testing results with the latter will convey a similar outcome. Future work still remains to be carried out in terms of assessing the drug release kinetics for this class of fluoroquinolone polymers and further confirming its biological activity in vivo. On the basis of the findings of the current paper, these studies will be preferably carried out on materials that are synthesized using the NDP method, since the latter allows for an optimal incorporation of drug into the polymer chains and contributes to fewer side chain reactions. Acknowledgment. This research was supported by Bayer Healthcare, Inc., and the Medical Research Council of Canada. References and Notes (1) Mittelman, M. W. In The Molecular and Ecological DiVersity of Bacterial Adhesion; Flectcher, M., Ed.; Wiley-Liss, Inc.: New York, 1996; p 89. (2) Gristina, A. Science 1987, 237, 1588-1597. (3) Christensen, G. D.; Baddour, L. M.; Hasty, D. L.; Lawrance, G. H.; Simpson, W. A. In Infections Associated with Indwelling Medical DeVices; Bison, A. L., Waldvogel, F. A., Eds.; American Society for Microbiology: Washington, DC, 1989; p 59. (4) Nickel, J. C.; Costerton, J. W. Can. J. Infect. Dis. 1992, 3, 261267. (5) Duncan, R.; Kopecek, J. AdV. Polym. Sci. 1984, 57, 53-101. (6) Drobnik, J. AdV. Drug DeliVery ReV. 1989, 3, 229-245. (7) Yokoyama, M.; Inoue, S.; Kataoka, K.; Yui, N.; Okano, T.; Sakurai, Y. Makromol. Chem. 1989, 190, 2041-2054. (8) Trooskin, S. Z.; Dontetz, A. P.; Harvey, R. A.; Greco, R. S. Surgery 1985, 97, 547-551. (9) Modak S. M.; Sampath, L.; Fox, C. L.; Benvenisty A.; Nowygrod, R.; Reemstmau, K. Surg., Gynecol. Obstert. 1987, 164, 143-147. (10) Bach, A.; Schmidt, H.; Bo¨ttiger, B.; Schreiber B.; Bo¨hrer, H.; Motsch, J.; Martin, E.; Sonntag, H. G. J. Antimicrob. Chemother. 1996, 37, 315-322. (11) Phaneuf, M. D.; Ozaki, C. K.; Bide, M. J.; Quist, W. C.; Alessi, J. M.; Tannenbaum, G. A.; LoGerfo, F. W. J. Biomed. Mater. Res. 1993, 27, 233-237. (12) Schierholz, J. M.; Rump, A.; Pulverer, G. Arzneim.-Forsch. 1997, 47, 70-74. (13) Lowman, A. M.; Peppas, N. A. Macromolecules 1997, 30, 49594965. (14) Bach, A.; Bo¨hrer, H.; Motsch, J.; Martin, E.; Geiss, H. K.; Sonntag, H. G. J. Antimicrob. Chemother. 1994, 33, 969-978. (15) Jansen, B.; Jansen, S.; Peters, G.; Pulverer, G. J. Hosp. Infect. 1992, 22, 93-107. (16) Li, X.; Bennnett, D. B.; Adams, N.; Kim, S. W. In Polymeric Drugs and Drug DeliVery Systems; Dunn, R. L., Ottenbrite, R. M., Eds; American Chemical Society: Washington, DC, 1991; pp 100-116. (17) Ohya, Y.; Masunaga T.; Baba, T.; Ouchi, T. J. Biomater. Sci., Polym. Ed. 1996, 7, 1085-1096. (18) Nathan, A.; Zalipsky, S.; Kohn, J. Polym. Prepr. (Am. Chem. Soc., DiV. Polym. Chem.) 1990, 31, 213-214. (19) Nathan, A.; Bolikal, D.; Vyavahare, N.; Zalipsky, S.; Kohn. J. Macromolecules 1992, 25, 4476-4484.

Utilization of Quinolone Drugs as Monomers (20) Nathan, A.; Zalipsky, S.; Ertel, S. I.; Agarthos, S. N.; Yarmush, M. L.; Kohn, J. Bioconjugate Chem. 1993, 4, 54-62. (21) Nathan, A.; Zalipsky, S.; Kohn. J. J. Bioact. Compat. Polym. 1994, 9, 239-251. (22) Maeda, H. AdV. Drug DeliVery ReV. 1991, 6, 181-202. (23) Hoffman A. S. In Biorelated Functional Polymers: Controlled Release and Applications in Biomedical engineering; Okano, T., Ed.; Academic Press: New York, 1998; pp 231-248. (24) Ghosh, M. In Progress in Biomedical Polymers, Gebelein, C., Dunn, R., Eds.; Plenum Press: New York, 1990; pp 335-345. (25) Norris, S.; Mandell, G. L. In The Quinolones; Andriole, Ed.; Academic Press: New York, 1988; pp 1-21. (26) Remers, W.; Bear S. In Burger’s Medicinal Chemistry and Drug DiscoVery, 5th ed.; Wolff, M. E., Ed.; John Wiley & Sons: New York, 1997; Vol. 4, Therapeutic Agents, pp 267-271. (27) Roseeuw, E.; Coessens V.; Schacht, E.; Vrooman, B.; Domurado, D.; Marchal, G. J. Mater. Sci.: Mater. Med. 1999, 10, 743-746. (28) Coessens V.; Schacht E.; Domurado D. J. Controlled Release 1997, 47, 283-291. (29) Lelah, M. D.; Cooper, S. L. Polyurethanes in Medicine; CRC Press: Boca Raton, FL, 1986. (30) Ito, Y.; Imanishi, Y. Crit. ReV. Biocompact. 1989, 5, 45-104. (31) Santerre, J. P.; Labow, R. S.; Adams, G. A. J. Biomed. Mater. Res. 1993, 27, 97-109. (32) Santerre, J. P.; Labow, R. S.; Duguay, D. G.; Erfle, D.; Adams, G. A. J. Biomed. Mater. Res. 1994, 28, 1187-1199. (33) Santerre, J. P.; Labow, R. S. J. Biomed. Mater. Res. 1997, 36, 223232.

Biomacromolecules, Vol. 2, No. 1, 2001 141 (34) Woo, G. L. Y.; Mittelman, M. W.; Santerre, J. P. Biomaterials 2000, 21, 1235-1246. (35) Talja, M.; Korpela, A.; Jarvi, K. Br. J. Urol. 1990, 66, 652-657. (36) Anderson, J. M. CardioVasc. Pathol. 1993, 2, 33S-41S. (37) Remes, A.; Williams, D. F. Biomaterials 1992, 13, 731-743. (38) Labow, R. S.; Duguay, D. G.; Santerre, J. P. J. Biomater. Sci., Polym. Ed. 1994, 6, 169-179. (39) Salthouse, T. N. J. Biomed. Mater. Res. 1976, 13, 731-743. (40) Labow, R. S.; Erfle, D. J.; Santerre, J. P. Biomaterials. 1995, 16, 51-59. (41) Labow, R. S.; Meek, E.; Santerre, J. P. J. Biomed. Mater. Res. 1998, 36, 469-477. (42) Irikura T. Belg. Pat. 863,429. (43) Pitt, C. G. In Biodegradable Polymers as Drug DeliVery Systems; Chasin, B., Langer, R., Eds.; Dekker: New York, 1990; pp 71120. (44) Wirpsza Z. In Polyurethanes: Chemistry, Technology and Applications, Methods of analysis of polyurethanes; Translated by Skup, A.; Ellis Horwood: Toronto, 1993. (45) Domard, A.; Rinaudo, M.; Rochas, C. J. Polym. Sci., Polym. Phys. Ed. 1979, 17, 673-681. (46) Scheuing, D. R. J. Appl. Polym. Sci., 1984, 29, 2819-2828. (47) Baumann, E. W. Anal. Chim. Acta 1968, 42, 127-132. (48) Woo, G. L. Y.; Yang, M. L.; Yin, H.; Jaffer, F.; Mittelman, M. W.; Santerre, J. P. Submitted for publication in J. Biomed. Mater. Res.

BM000087G