Selective Removal of Bilirubin from Human Plasma with Bilirubin

The objective of this study is to prepare bilirubin-imprinted polymeric particles for the selective removal of bilirubin from hyperbilirubinemic human...
13 downloads 0 Views 533KB Size
Ind. Eng. Chem. Res. 2007, 46, 2843-2852

2843

Selective Removal of Bilirubin from Human Plasma with Bilirubin-Imprinted Particles Go1 zde Baydemir,† Mu1 ge Andac¸ ,† Nilay Bereli,† Ry´ dvan Say,‡ and Adil Denizli*,† Department of Chemistry, Biochemistry DiVision, Hacettepe UniVersity, Ankara, Turkey, and Department of Chemistry, Anadolu UniVersity, Eskis¸ ehir, Turkey

The objective of this study is to prepare bilirubin-imprinted polymeric particles for the selective removal of bilirubin from hyperbilirubinemic human plasma. N-methacryloyl-(L)-tyrosine methylester (MAT) was chosen as the complexing monomer. In the first step, functional monomer MAT was synthesized by the reaction of L-tyrosine methylester and methacryloyl chloride and characterized by nuclear magnetic resonance (NMR). Bilirubin then was complexed with MAT and the bilirubin-imprinted poly(2-hydroxyethyl methacrylate-Nmethacryloyl-(L)-tyrosine methylester) [MIP] was produced by bulk polymerization. The template molecules (i.e., bilirubin) then were removed using sodium carbonate and sodium hydroxide. MIP particles were characterized by elemental analysis, water uptake tests, Fourier transform infrared (FTIR) spectroscopy, and scanning electron microscopy (SEM). Bilirubin adsorption experiments from human plasma were performed in a batch experimental setup. Cholesterol and testosterone were used as competing molecules in selectivity tests. Obtained results were as follows: the water uptake ratio of MIP and non-imprinted (NIP) particles were 64.7% and 51.3%, respectively, in water. According to the elemental analysis results, the incorporation of MAT was 69.0 µmol/g for MIP particles. SEM micrographs showed the surface roughness and porosity. The specific surface area of the MIP particles was determined to be 27.8 m2/g. The pore diameter of the MIP particles varied over a range of 20-245 Å and the average pore diameter was 25.0 Å. Template molecules (i.e., bilirubin) were removed from the polymer structure in the ratio of 87% of the initial concentration. Bilirubin adsorption increased as the bilirubin concentration increased, up to 0.8 mg/mL. The maximum bilirubin adsorption capacity was 3.4 mg/g of the dry weight of particles. MIP particles were 6.3 and 3.0 times more selective, with respect to the cholesterol and testosterone, respectively. Reusability of the MIP particles was also investigated. MIP particles showed a negligible loss in the bilirubin adsorption capacity after five adsorption-desorption cycles with the same adsorbent. 1. Introduction Bilirubin is a tetrapyrrole dicarboxylic acid that is formed in the normal metabolism of heme proteins in senescent red blood cells, and it is normally conjugated with albumin to form a water-soluble complex. It is transported to the liver as a complex with albumin, where it is normally conjugated and excreted into the bile.1 The free bilirubin is toxic. High concentrations of free bilirubin can evoke hepatic or biliary tract dysfunction and permanent brain damage or death in more-severe cases.2 Neurological dysfunctions as kernicterus or bilirubin encephalophathy may develop if the bilirubin concentration in the plasma increases to >15 mg/dL. Disorders in the metabolism of bilirubin may cause a yellow discoloration of the skin and other tissues. Several methods have been developed for the treatment of hyperbilirubinemia, such as plasma exchange, hemodialysis, phototherapy, and hemoperfusion. However, treatment with plasma exchange requires large volumes of fresh frozen plasma, which is expensive and difficult to obtain. Hemoperfusion (i.e., the circulation of blood through an extracorporeal column containing an adsorbent system for bilirubin) has become the most promising technique.3-13 Sideman et al. suggested the application of hemoperfusion to the removal of the bilirubin from jaundiced newborn babies using albumin-deposited macroreticular resin.4 Idezuki et al. used anion exchange synthetic * Corresponding author (e-mail: [email protected]). † Department of Chemistry, Biochemistry Division, Hacettepe University. ‡ Department of Chemistry, Anadolu University.

fibers and clinically applied this sorbent system in a selective bilirubin separation.5 Brown prepared oligo-peptide functionalized polyacrylamide particles as an affinity sorbent system for bilirubin removal.6 Chandy et al. used polylysine immobilized chitosan particles for selective bilirubin removal.7 Yamazaki et al. developed poly(styrene-divinyl benzene)-based adsorbents, and succesfully applied in the treatment of more than 200 patients with hyperbilirubinemia.8 Morimoto et al. used plasma exchange and plasma adsorption with styrene-divinyl benzene resin and removed bilirubin from hepatectomized patients.9 This plasma adsorption system provided a possibility for an improved supportive therapy for hepatic failure, especially for patients with hepatic coma and hyperbilirubinemia. Avramescu et al. conjugated bovine serum albumin with ethylene vinyl alcohol adsorptive membranes and they reported high bilirubin binding capacity.10 Yu et al. synthesized aminecontaining cross-linked chitosan resins and investigated adsorption behavior of conjugated bilirubin.11 Kuroda et al. studied the selective adsorption of bilirubin by porous poly(glycidyl methacrylate-co-divinylbenzene) particles.12 Ahmad et al. demonstrated the suitability of rat serum albumin-loaded poly(lactide-co-glycolide) biodegradable microspheres in the removal of bilirubin from a systemic circulation of hyperbilirubinemic rats.13 Denizli et al. used in vitro experiments to show that dye-affinity adsorbents efficiently removed bilirubin from human plasma.14-18 Uzun and Denizli used magnetically stabilized fluidized-bed columns that contained human serum albumin-immobilized magnetic poly(hydroxyethyl methacrylate) beads for the selective bilirubin removal from human plasma.19

10.1021/ie0611249 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/31/2007

2844

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007

Highly specific adsorbents underline various affinity-based detection and separation techniques.20 For example, antibodies are routinely utilized as analytical reagents in clinical and research laboratories. For many practical reasons, attempts have been made to replace antibodies with more-stable counterparts. One technique that is being increasingly adopted for the generation of artificial antibodies is molecular imprinting of synthetic polymers.21 Given the advantage of easy preparation and chemical stability, molecularly imprinted polymers (MIPs) possess a high potential for use in a variety of applications, such as chromatographic stationary phases,22 immunoassay-type analyses,23 metal-ion-removal studies,24,25 and sensor development.26 Generally, molecular imprinting is a synthetic strategy that is used to assemble a molecular receptor via template-guided synthesis. To prepare MIPs, a print molecule (template) is used to guide the assembly of functional monomers. Polymerization reaction is then applied to fix the preassembled binding groups around the print molecule. Following the removal of the print molecule, the polymer revealed retains specific binding sites that can selectively rebind the original print molecule. Depending on the interactions between the print molecule and the functional monomers/groups involved at the imprinting and rebinding step, molecular imprinting has two different approaches: noncovalent27 and covalent.28 In the noncovalent approach, various noncovalent interactions (such as hydrogen bond, ionic interactions, and hydrophobic effects) are utilized. Given the fact that noncovalent molecular interactions are prevalent in the biological world, exploitation of these binding forces, as it has turned out, has proven to be the most efficient and preferred method for generating robust, biomimetic binding materials.29 In this work, we have produced MIP particles for the selective removal of bilirubin from hyperbilirubinemic human plasma. Different types of MIP particles are manufactured for bilirubin recognition.30-33 However, there are no studies that use MIP particles for the selective bilirubin removal from human plasma. To show bilirubin specificity of the MIP particles, competitive adsorptions has also been studied with different molecules (cholesterol and testosterone). Finally, repeated use of the MIP particles for the removal of bilirubin molecules from human plasma also has been studied. 2. Experimental Section 2.1. Materials. Bilirubin, cholesterol, testosterone, L-tyrosine methylester, and methacryloyl chloride were supplied by Sigma (St. Louis, MO). Hydroxyethyl methacrylate and ethylene glycol dimethacrylate (EGDMA) were also obtained from Sigma, distilled under reduced pressure in the presence of a hydroquinone inhibitor, and stored at 4 °C until use. Ammonium persulfate (APS) was obtained also from Sigma. Methanol and acetonitrile were high-performance liquid chromatography (HPLC) grade and were supplied by Sigma. All other chemicals were reagent grade and were purchased from Merck AG (Darmstadt, Germany). All water used in the experiments was purified using a Barnstead (Dubuque, IA) ROpure LP reverse osmosis unit with a high flow cellulose acetate membrane (Barnstead D2731), followed by exposure to a Barnstead D3804 NANOpure organic/colloid removal and ion exchange packedbed system. Buffer and sample solutions were prefiltered through a 0.2-µm membrane (Sartorius, Go¨ttingen, Germany). All glassware was extensively washed with dilute nitric acid before use. 2.2. Synthesis of N-methacryloyl-(L)-tyrosinemethylester (MAT). The MAT was selected as the functional monomer for bilirubin imprinting. The following experimental procedure was

applied for the synthesis of MAT. Precisely five grams of L-tyrosine methylester and 0.2 g of hydroquinone were dissolved in 100 mL of dichloromethane solution. This solution was cooled to 0 °C. Triethylamine (12.7 g) was added to the solution. Precisely five milliliters of methacryloyl chloride was poured slowly into this solution, and then it was stirred magnetically at room temperature for 2 h. At the end of the chemical reaction period, hydroquinone and unreacted methacryloyl chloride were extracted with ethyl acetate. The aqueous phase was evaporated in a rotary evaporator. The residue (i.e., MAT) was crystallized in an ether-cyclohexane mixture and then dissolved in ethyl alcohol. 2.3. Preparation of Bilirubin-Imprinted Particles. In the first part, a MAT-bilirubin complex was prepared. Briefly, N-methacryloyl-(L)-tyrosine methylester (MAT) (4 mg) was dissolved in sodium carbonate and sodium hydroxide solutions (Na2CO3, 0.1 M; NaOH, 0.1 M); bilirubin (2 mg) then was added in this solution. The bilirubin-monomer complex then was used in the polymerization procedure. Hydroxyethyl methacrylate (HEMA) and the bilirubin-MAT complex were polymerized in bulk polymerization, using ammonium persulphate (APS) as the initiator. Toluene and ethylene glycol dimethacrylate (EGDMA) was included in the polymerization recipe as the pore former and cross-linker, respectively. N,N,N′,N′-Tetraethyldietyhlene-triamine (TEMED) was used as the activator. Ammonium persulfate (20 mg) and TEMED (100 µL) were dissolved in the mixture of monomers (HEMA, 1.0 mL; bilirubin-MAT complex, 500 µL; EGDMA, 500 µL) and porogenic diluent (toluene, 500 µL). The polymerization mixture was poured in a glass tube and sealed after purging with nitrogen for 2 min. Polymerization reaction was completed in 10 min at room temperature. At the end of the polymerization reaction, soluble components were removed from the polymer by repeated decantation with water and methanol. Bilirubin-imprinted poly(HEMA-MAT) (MIP) bulk polymer was ground in a mill after drying and washed several times with methanol and water to remove any unreacted components completely. Washed particles were ground again and sieved through 100 µm sieves (Retsch Standard Sieves Model AS200, Retsch Gmbh & Co, KG, Haan, Germany). Non-imprinted particles (NIPs, i.e., poly(HEMAMAT) particles) were prepared in the same manner as that previously described, without using bilirubin as template. Note that all bilirubin adsorption studies were performed in darkness, to minimize the transformation of bilirubin to biliverdin. 2.4. Removal of the Template Molecule. To remove unreacted monomers and other ingredients, MIP particles were extensively washed with a methanol:water solution (60:40, v/v) for 24 h at room temperature in darkness. After the cleaning procedure, the template was removed from the polymer particles, using 2 M NaOH and 2 N Na2CO3, including an ethylenediamine tetraacetic acid (EDTA) solution. The molar ratio of NaOH and EDTA was 4:1. The imprinted particles were added into this solution system for 48 h at room temperature in darkness. This procedure was repeated until no bilirubin leakage was observed from the MIP particles into the wash solution. The template free particles were cleaned with ethanol and water in a magnetic stirrer at room temperature for 12 h. 2.5. Characterization of Particles. Nuclear magnetic resonance (1H NMR) spectroscopy was used to determine the structure of MAT. The 1H NMR spectrum of MAT monomer was taken in DMSO-D6 on a Bruker-400 MHz instrument (USA). The residual nondeuterated solvent (DMSO) served as an internal reference. Chemical shifts are reported in units of parts per million (ppm) (δ), downfield relative to the solvent.

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2845

Figure 1. NMR spectrum of MAT monomer.

Figure 2. FTIR spectra of bilirubin, MAT, and the MAT-bilirubin complex.

Water uptake ratios of NIP and MIP particles were determined in distilled water. The experiment was conducted as follows. Initially dry particles were carefully weighed ((0.0001 g) before being placed in a 50-mL vial that contained a water uptake medium. The vial was placed into an isothermal water bath with a fixed temperature (25 ( 0.5 °C) for 2 h. The polymer sample was taken out from the medium, wiped using a filter paper, and weighed. The weight ratio of dry and wet samples was recorded. Each experiment was performed in triplicate for quality control and statistical purposes. For each set of data present, standard statistical methods were used to determine the mean values and standard deviations. Confidence intervals of 95% were calculated for each set of samples, to determine the margin of error. The water uptake ratio was calculated using

the following expression:

water uptake ratio (%) )

(

)

Ws - W 0 × 100 W0

(1)

where W0 and Ws are the weights of particles before and after water uptake, respectively. The surface morphology of the particles was examined using scanning electron microscopy (SEM). The samples were initially dried in air at 25 °C for 7 days before being analyzed. A fragment of the dried particles was mounted on a SEM sample mount and was sputter-coated for 2 min. The sample was then mounted in an SEM microscope (model JEM 1200EX, JEOL,

2846

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007

Figure 3. FTIR spectra of NIP and MIP particles.

Tokyo, Japan). The surface of the sample was then scanned at the desired magnification to study the particles. To evaluate the degree of MAT incorporation to the polymer structure, particles were subjected to elemental analysis using a LECO elemental analyzer (Model CHNS-932). Fourier transform infrared (FTIR) spectra of MAT, bilirubin, the MAT-bilirubin complex, and MIP and NIP particles were obtained using a FTIR spectrophotometer (FTIR 8000 Series, Shimadzu, Japan). The MAT monomer, MAT-bilirubin complex, and MIP and NIP particles were dried in a vacuum oven. The dry sample (∼0.1 g) was thoroughly mixed with KBr (0.1 g, IR Grade, Merck, Germany), and pressed into a pellet form and the FTIR spectrum was then recorded. Porosity of the particles was measured by the nitrogen sorption technique, performed on Flowsorb II (Micromeritics Instrument Corporation, Norcross, GA). The specific surface area of particles in a dry state was determined using a multipoint Brunauer-Emmett-Teller (BET) apparatus (Quantachrome, Nova 2200E). A quantity of 0.5 g of particle was placed in a sample holder and degassed in a N2-gas stream at 150 °C for 1 h. Adsorption of the gas was performed at -210 °C and desorption was performed at room temperature. Values obtained from desorption step was used for the specific surface area calculation. The pore volume and average pore diameter were determined using the Barrett-Joyner-Halenda (BJH) model on adsorption. The average size and size distribution of the particles were determined by screen analysis performed using standard sieves (Model AS200, Retsch Gmb & Co, KG, Haan, Germany). 2.6. Bilirubin Removal from Human Plasma. Bilirubin removal from human plasma was studied in a batch system. Human blood was collected from thoroughly controlled voluntary blood donors. Each unit should be separately controlled and found negative for HBS antigen, HIV types I and II, and hepatitis C antibodies. No preservatives were added to the blood samples. Human blood samples were transferred into EDTAcontaining vacutainers in which red blood cells were separated from plasma by centrifugation at 4000 g for 30 min at room temperature, then filtered through a 0.45-µm syringe filters (Model 245-0045, Nalgen Co., Rochester, NY) and the plasma

was subjected to a deep-freeze process at -20 °C. Before use, the plasma was thawed slowly for >1 h at 37 °C. All adsorption experiments were performed in darkness. In a typical adsorption system, 10 mL of the human plasma was incubated with 125 mg of particles, at 25 °C for 2 h. In kinetic tests, the sample was withdrawn in certain time intervals. The concentration of the bilirubin molecules in the plasma after the desired treatment periods was measured using Roche Hitachi Modular-P with Roche Bilirubin Direct, Indirect Test Kits (Diamond Diagnostics). The experiments were performed in replicates of three, and the samples also were analyzed in replicates of three. For each set of data present, standard statistical methods were used to determine the mean values and standard deviations. Confidence intervals of 95% were calculated for each set of samples, to determine the margin of error. The amount of bilirubin adsorption per unit mass of the particles was evaluated using the following expression:

Q)

(C0 - C)V m

(2)

Here, Q is the amount of bilirubin adsorbed onto a unit mass of the particles (expressed in units of mg/g); C0 and C are the concentrations of the bilirubin in the initial plasma and in the plasma after treatment for a certain period of time, respectively (expressed in units of mg/mL); V is the volume of the solution (given in milliliters); and m is the mass of the particles used (given in grams). 2.7. Selectivity Experiments. To show the bilirubin molecules specificity of MIP particles, the competitive adsorption (i.e., cholesterol (MW ) 386 g/mol) and testosterone (MW ) 288 g/mol)) was also studied. Ten milliliters of fresh human plasma was overloaded with competing substances (i.e., cholesterol and testosterone) by the same procedure. Selectivity experiments were conducted with bilirubin/cholesterol and bilirubin/testosterone binary plasma samples. MIP particles were treated with these competitive molecules. After adsorption equilibrium, the concentrations of cholesterol and testosterone

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2847

Figure 4. Optic photographs of NIP (left) and MIP (right) particles.

Figure 5. SEM micrographs of the (A) MIP and (B) NIP particles.

in the remaining solution were measured by Roche Hitachi Modular-P, with a Roche Direct Bilirubin, Indirect Bilirubin Test Kit. Distribution and selectivity coefficients of cholesterol and testosterone, with respect to bilirubin, were calculated as explained by the following equation:34

Kd )

(

)

Ci - Cf V Cf m

(3)

Here, Kd represents the distribution coefficient; Ci and Cf are the initial and final concentrations of the biomolecules,

respectively. V is the volume of the solution (expressed in milliliters), and m is the mass of particles used (given in grams). The selectivity coefficient for the binding of bilirubin in the presence of other biomolecules can be obtained from equilibrium binding data, according to eq 4:

k) )

[M2]solution[M1]sorbent [M1]solution [M2]sorbent

Kd(bilirubin) Kd(X)

(4)

2848

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007

where k is the selectivity coefficient and X represents the cholesterol and testosterone. A comparison of the k-values of the MIP particles with those steroids allows an estimation of the effect of imprinting on selectivity. The relative selectivity coefficient, k′ (eq 5), can be defined as shown in eq 5:

k′ )

kimprinted kcontrol

(5)

Table 1. Physical Properties of NIP and MIP Particles polymer

surface areaa (m/g)

total pore volumeb (mL/g)

average pore diameterc (Å)

NIP MIP

14.5 27.8

0.032 0.046

23.4 25.0

a Determined using the multipoint BET method. b BJH cumulative desorption pore volume of pores 20-245 Å in diameter. c BJH desorption average pore diameter of pores 20-245 Å in diameter.

2.8. Desorption and Repeated Use. The desorption of bilirubin molecules was studied with two different desorption agents: 2 M NaOH (with EDTA, in a molar ratio of NaOH: EDTA ) 4:1), and 2 N Na2CO3 solution, respectively. MIP particles were placed in this desorption medium and stirred continuously (at a stirring rate of 400 rpm) for 2 h at room temperature. The final bilirubin concentration in the desorption medium was measured by Roche Hitachi Modu¨lar-P, using Roche Bilirubin Direct, Indirect Test Kits. The desorption ratio was calculated from the amount of bilirubin adsorbed on the particles and the final bilirubin concentration in the desorption medium, using the following expression:

desportion ratio (%) ) amount of bilirubin desorbed onto the elution medium × amount of bilirubin adsorbed onto the particles 100 (6) To test the reusability of the MIP particles, the bilirubin adsorption-desorption procedure was repeated five times using the same MIP adsorbent. To regenerate and sterilize, after desorption, the particles were washed with 50 mM NaOH solution. 3. Results and Discussion 3.1. Characterization of MIP Particles. The functional monomer, N-methacryloyl-(L)-tyrosine methyl ester (MAT), was synthesized from the reaction of (L)-tyrosine and methacryloyl chloride. The 1H NMR spectrum of MAT is shown in Figure 1. The characteristic peaks from the groups in MAT monomer of related protons are marked on the spectrum. These characteristic peaks are as follows: 2.05 ppm 3H singlet (-CdCCH3, vinyl methyl (peak a)), 5.4-5.7 ppm 2H (-CdCH2) (peak b), 3.65 ppm 2H (-CH2-C-) (peak c), 4.5 ppm 1H (-CHC-) (peak d), 8.2 ppm 1H (amine protons) (peak f), 6.6-7.1 ppm aromatic protons (peak g). The presence of the carboxyl proton was observed at 9.0-10.0 ppm as a large peak (peak e). DMSO-D6 was used as the solvent. FTIR spectra of the MAT monomer, bilirubin, and MATbilirubin complex are shown in Figure 2. FTIR spectrum of pure bilirubin has two important characteristic peaks, which include N-H amine at 3406 cm-1 and pyrrole at 755 cm-1. FTIR spectrum of MAT has the characteristic stretching vibration amide I and amide II absorption bands at 1653 cm-1 and 1516 cm-1, a carbonyl band at 1733 cm-1, and an aromatic C-H band at 809 cm-1. The functional monomer MAT is expected to interact with bilirubin through hydrogen bonds and hydrophobic interactions through an aromatic ring. In the MAT-bilirubin monomer complex, pyrrole at 755 cm-1, which is the characteristic peak of bilirubin, is apparent. Furthermore, the aromatic peak at 809 cm-1 of MAT monomer shifts upfield to 813 cm-1, because of hydrophobic interactions. MIP and NIP particles give similar peaks, as shown in Figure 3. They were synthesized under the same conditions and they

Figure 6. Time-dependent adsorption of bilirubin on the MIP particles; Vtotal ) 10 mL, 0.8 mg/mL solution, 125 mg polymer, and T ) 25 °C.

have similar functional groups. The characteristic bilirubin peak, pyrrole at 755 cm-1 which is the only evidence of imprinting of bilirubin, seemed to be too noisy and could not be identified in the fingerprint region of MIP particles. Note that the peak intensities were decreased in the resulting polymer. Moreover, Figure 4 represents optical photographs of the NIP (left) and MIP (right) particles. The MIP particles clearly have a characteristic yellow-green bilirubin color. The incorporations of the MAT for MIP and NIP particles were determined to be 69 and 27 µmol/(g polymer), respectively, using nitrogen stoichiometry. Note that HEMA and other polymerization ingredients do not contain nitrogen. This nitrogen amount, determined by elemental analysis, comes from only incorporated MAT groups into the polymeric structure. The MIP particles are cross-linked hydrophilic matrices. The water uptake ratios of the NIP and MIP particles used in this study are 51.3% and 64.7%, respectively. Compared to that of the NIP particles, the water uptake ratio of the MIP particles increases. Several explanations can be offered. First, the formation of molecular cavities in the polymer structure introduces more hydrodynamic volume into the polymer chain, which can adsorb more water molecules into the polymer matrices. Second, the reaction of the bilirubin-MAT complex with HEMA effectively increased the length of the polymer chains. Therefore, water molecules penetrate into the polymer chains more easily, resulting in an improvement of the polymer water uptake in aqueous solutions. Note that these particles are quite rigid, and strong enough, because of their cross-linked structure. Therefore, these imprinted particles are suitable for packed-bed column applications. The surface morphology and internal structure of MIP and NIP particles are exemplified by the SEM micrographs in Figure 5. Figure 5A shows the MIP particles. They are composed of small and interconnected globules that form a porous structure. The size of the globules was determined to be in the range of ∼0.5-2 µm, according to the enlarged SEM micrograph; note that this size is ∼5-fold less than those of conventional porous particles packed in chromatographic devices. The bilirubin

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2849

Figure 7. Effect of equilibrium bilirubin concentration on the adsorption of bilirubin on MIP particles; Vtotal ) 10 mL, 125 mg polymer, time ) 60 min, T ) 25 °C. Table 2. Comparison of the Adsorption Capacities for Bilirubin of Various Adsorbents material

ligand/interaction type

adsorption capacity (mg/g)

reference(s)

polyacrylamide beads polyacrylamide beads polyacrylamide beads macroreticular resin chitosan particles poly(ethylene vinyl alcohol) poly(glycidyl methacrylate-divinylbenzene copolymer poly(hydroxyethyl methacrylate) particles poly(hydroxyethyl methacrylate) particles poly(hydroxyethyl methacrylate) particles polyamide hollow fiber poly(hydroxyethyl methacrylate) magnetic particles, MSFB chitosan coupled nylon membrane anion exchange resin poly(tetrafluoroethylene) membrane polyamide resin IONEX polypropylene fiber polyacrylonitrile membrane polybutadiene- hydroxyethyl methacrylate gels partially aminated polyacrylamide poly(hydroxyethyl methacrylate) magnetic particles/batch studies cellulose acetate fiber poly(MAA-EGDMA)/MIP poly(MAA-EGDMA)/MIP poly(HEMA-MAT)/MIP

poly-L-lysine poly-D-lysine poly-L-ornithine albumin poly-L-lysine bovine serum albumin albumin Cibacron Blue F3GA Alkali Blue 6B Congo Red Cibacron Blue F3GA human serum albumin Cibacron Blue F3GA ion exchange Cibaron Blue F3GA aminoacid tertiary amine hepatoycte receptor bovine serum albumin β-cyclodextrin human serum albumin Cibaron Blue F3GA molecular recognition molecular recognition molecular recognition

0.2-75 0.2-75 0.2-75 2-24 1.5 25.0 30 6.8-32.5 6.8-32.5 6.8-32.5 48.9 88.3 64.7 4.0-80 76.2 5-80 7.7 2.8 3.1 42.2 64.7 4.0 1.04 0.24-0.85 3.41

3 3 3 4 7 10 12 14-17 14-17 14-17 18 19 35 38 39 40 41 42 43 44 45 46 32 33 this study

Table 3. Langmuir and Freundlich Adsorption Isotherm Constantsa Experimental MIP a

Qexp (mg/g) 3.41

Langmuir Model Qmax (mg/g) 3.95

Kd 8.37

R2 0.9946

Freundlich Model KF 4.45

n 0.52

R2 0.9504

Vtotal ) 10 mL, polymer amount, 125 mg; t ) 60 min; T ) 25 °C.

imprinted particles clearly have a porous structure that is similar to that of the corresponding NIP particles (Figure 5 B). This similarity is important in the quality of competitive studies. MIP particles were synthesized by bulk polymerization. The grounded polymers were sieved and the fraction that fall in the size range of 71-100 µm were used throughout the study. The specific surface area, total pore volume, and average pore diameter of the NIP and MIP particles are presented in Table 1. The specific surface area of the MIP particles was determined to be 27.8 m2/(g polymer), using the multipoint BET measurements. According to the BJH method, the pore diameter of the MIP particles varies over a range of 20-245 Å and the average pore diameter is 25.0 Å. This indicated that the MIP particles contained mainly mesopores.

Bilirubin was extracted from MIP particles extensively with 2 M NaOH and 2 N Na2CO3, including EDTA solution, for 48 h at room temperature in darkness before the bilirubin adsorption studies from human plasma. This procedure was repeated until no bilirubin leakage was detected from the MIP particles. The ratio of bilirubin extracted was 87%, which corresponds to a value of 1.48 mg bilirubin per gram of polymer. Leakage of bilirubin molecules into the interior of the particles is difficult for the diffusion limitations; much greater incubation times are required. 3.2. Adsorption of Bilirubin from Human Plasma. 3.2.1. Effect of Time. Figure 6 shows the time dependence of the adsorption values of bilirubin on MIP particles. The adsorption rate was relatively fast; the time required to reach equilibrium conditions was ∼60 min. The adsorption capacity for bilirubin was 3.41 mg per gram dry weight of particles. This fast adsorption equilibrium is most probably due to high complexation and geometric affinity between bilirubin molecules and bilirubin cavities in the particle structure. The adsorption capacity for NIP particles was 0.96 mg/(g of particles).

2850

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007

Table 4. First- and Second-Order Kinetic Constants for the MIP Particlesa Experimental

First-Order Kinetic

Second-Order Kinetic

initial concentration (mg/mL)

Qeq (mg/g)

slope, k1 (1/min)

intercept, qeq (mg/g)

R2

intercept, k2 (g/(mg min))

slope, qeq (mg/g)

R2

0.8

3.41

0.0414

3.88

0.915

5.87 10-3

4.63

0.984

a

Vtotal ) 10 mL; Cbilirubin ) 0.8 mg/mL; polymer amount, 125 mg; T ) 25 °C Table 5. Kd, k, and k′ Values of Cholesterol and Testosterone, with Respect to Bilirubin NIP Particles

Figure 8. Chemical structures of competitive molecules.

For the extracorporeal removal of bilirubin by various adsorbents, a wide range of adsorption rates have been reported in the literature. For example, Chandy and Sharma considered 2 h to be an equilibrium adsorption time in their bilirubin removal studies, in which they used polylysine-immobilized chitosan beads.7 Morimoto et al. have studied bilirubin removal in patients with post-operative hepatic failure on an anionexchange resin column composed of styrene-divinyl benzene, and each session lasted a mean of 3 h.9 The total bilirubin level was drastically reduced to 40% of the perfusion level. Avramescu et al. have studied bilirubin adsorption on a column that contained ethylene vinyl alcohol polymeric membranes that had bovine serum albumin and the adsorption process is completed within 6 h.10 Syu et al. prepared bilirubin imprinted poly(MAAEGDMA) particles, and their adsorption results were determined within ∼2 h, which is the time required to achieve equilibrium adsorption.32 Xia et al. studied bilirubin removal using a Cibacron Blue F3GA attached microporous nylon membrane and reported an adsorption time of 10 h.35 Annesini et al. have considered 5 h to be a very fast adsorption time in their bilirubin removal studies, in which they used an anion-exchange styrenedivinyl benzene resin.36 A patient diagnosed as having fulminant hepatitis must be treated using commercial Plasmasorb BR 350 anion exchange column around 1 h effective treatment time37 for the removal of bilirubin. The flow rate in the aqueous phase, structural properties of adsorbent (e.g., porosity, surface area), amount of adsorbent, adsorbate properties (e.g., molecular dimensions and solubility), initial concentration of bilirubin determine the adsorption rate. In this study, the hyperbilirubinemia patient plasma was incubated MIP particles for 2 h. It can be concluded that the removal rates obtained with the bilirubin-imprinted particles used in this study seem to be quite promising. 3.2.2. Effect of Equilibrium Concentration of Bilirubin. Figure 7 shows the dependence of the equilibrium concentration of bilirubin molecules of the adsorbed amount of the bilirubin onto the MIP particles. The adsorption values increased as the concentration of bilirubin molecules increased, and a saturation

molecule

Kd (mL/g)

bilirubin cholesterol testosterone

1.36 0.53 0.23

MIP Particles k

Kd

k

k′

2.56 5.91

5.00 0.31 0.28

16.18 17.86

6.32 3.02

value is achieved at a bilirubin concentration of 0.8 mg/mL, which represents saturation of the active binding cavities on the MIP particles. The maximum adsorption capacity was 3.41 mg/(g dry polymer). Recently, research interest focused on the preparation of bilirubin adsorbents with reactive ligands, including different peptide sequences and dye ligands. A comparison of the adsorption capacity of MIP particles with those of some other affinity adsorbents reported in literature is given in Table 2. The bilirubin adsorption capacity of MIP particles is 25 times smaller than that in Uzun and Denizli,19 and 19 times smaller than that in Xia et al.35 Also note that, in molecularly imprinted polymer (MIP) studies, the selectivity behavior is important rather than adsorption capacity. However, the adsorption capacity of MIP particles was also good, when compared to other MIP adsorbents.30-33 Differences of bilirubin adsorption capacity are due to the properties of each adsorbent, such as structure, functional groups, ligand loading, and accessible surface area. 3.2.3. Langmuir Adsorption Model. Two important physicochemical aspects for evaluation of the adsorption process as a unit operation are the kinetics and the equilibria of adsorption. Modeling of the equilibrium data has been done using the Langmuir and Freundlich isotherms. The Langmuir and Freundlich isotherms are represented as follows in eqs 7 and 8, respectively:

qmaxbCe 1 + bCe

(7)

q ) KFCe1/n

(8)

q)

where b is the Langmuir isotherm constant, KF the Freundlich constant, and n the Freundlich exponent. The term 1/n is a measure of the surface heterogeneity. The ratio of qe gives the theoretical monolayer saturation capacity of particles. According to the correlation coefficients of isotherms, Langmuir adsorption model is favorable. Table 3 shows the Freundlich adsorption isotherm constants, n and KF and the correlation coefficients. The magnitude of the KF and n values of the Freundlich model showed the easy uptake of bilirubin with a high adsorption capacity of the MIP particles. 3.2.4. Adsorption Kinetics. To quantify the extent of uptake in adsorption kinetics, the kinetic models (pseudo-first- and second-order equations) can be used in this case, assuming that the measured concentrations are equal to adsorbent surface concentrations.47 The first-order rate equation of Lagergren is

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007 2851

Figure 9. Adsorbed template and competitive molecules both in MIP and NIP particles; 0.8 mg/mL, 10 mL solution, 125 mg polymer, T ) 25 °C.

one of the most widely used for the adsorption of solute from a liquid solution. It may be represented as follows:

log(qe- qt) ) log(q1cal) -

k1t 2.303

(9)

where qe is the experimental amount of bilirubin adsorbed at equilibrium (expressed in units of mg/g), qt the amount of bilirubin adsorbed at time t (again, expressed in units of mg/g); k1 the equilibrium rate constant of first order adsorption (1/ min), and q1cal the adsorption capacity calculated by the pseudofirst-order model (expressed in units of mg/g). The rate constant for the second-order adsorption could be obtained from the following equation:

1 1 t ) + t qt k2q2cal2 q2cal

(10)

where k2 is the equilibrium rate constant of pseudo-second-order adsorption (expressed in units of g/(mg min)) and q2cal is the adsorption capacity calculated by the pseudo-second-order kinetic model (expressed in units of mg/g). Table 4 shows the results, which are for both the first-order and second-order kinetic models. The results show that the second-order mechanism is applicable (R2 values are the highest). These results suggest that the pseudo-second-order mechanism is predominant and that chemisorption might be the rate-limiting step that controls the adsorption process. The ratecontrolling mechanism may vary during the course of the adsorption process three possible mechanisms may be occurring.48 There is an external surface mass transfer or film diffusion process that controls the early stages of the adsorption process. This may be followed by a reaction or constant rate stage and finally by a diffusion stage where the adsorption process slows down considerably.49 3.2.5. Selectivity Experiments. Molecular recognition selectivity is the most important parameters in characterizing MIPs, because molecular recognition is the essential character of MIPs. Competitive adsorption of bilirubin and cholesterol and bilirubin and testosterone from their mixtures were also studied in a batch system. Cholesterol and testosterone were chosen as competitive molecules. Figure 8 shows the chemical structures of bilirubin, testosterone, and cholesterol. The molecular weight of bilirubin is 584 g/mol, whereas that of cholesterol is 386 g/mol and that of testosterone is 288 g/mol. Their molecular weights indicate that these molecules are of a similar size. Cholesterol and testosterone were chosen for the

Figure 10. Adsorption-desorption cycle of MIP particles. Vtotal ) 10 mL, 125 mg polymer, time ) 30 min, T ) 25 °C.

comparison because they, together with bilirubin, are always found in the serum. Hence, having knowledge of how they interfere with the binding of bilirubin by the MIP is essential for the in vitro measurement of serum samples. A comparison of the Kd values for the MIP samples with the control samples shows an increase in Kd for bilirubin, whereas Kd decreases for cholesterol and testosterone. The relative selectivity coefficient is an indicator that expresses the adsorption affinity of recognition sites to the bilirubin-imprinted molecules. These results show that the relative selectivity coefficients of imprinted particles for bilirubin/cholesterol and bilirubin/testosterone were 6.32 and 3.02 times greater than that of the non-imprinted matrix, respectively. Cholesterol and testosterone were less adsorbed by the MIP, because of their fewer chances to form hydrogen bonds, compared to bilirubin. Nevertheless, the binding specificity for bilirubin from MIP was sufficient for the recognition of bilirubin from other compounds. Table 5 summarizes the Kd, k, and k′ values in the selectivity studies. Figure 9 shows the adsorbed template and competitive molecules both in MIP and NIP particles (in terms of mg/(g polymer)). 3.3. Desorption and Repeated Use. To show the stability and reusability of the MIP particles, the adsorption-desorption cycle was repeated six times, using the same polymeric particles in a batch experimental setup. For sterilization after one adsorption-desorption cycle, the particles were washed with 50 mM NaOH solution for 30 min. After this procedure, particles were washed with distilled water for 30 min. At the end of five adsorption-desorption cycles, there was no remarkable decrease in the adsorption capacity (Figure 10). As observed here that the polymer beads are very stable, and maintain their adsorption capacity at almost constant value of 87%.

2852

Ind. Eng. Chem. Res., Vol. 46, No. 9, 2007

Literature Cited (1) Lee, K. H.; Wendon, J.; Lee, M.; Da Costa, M.; Lim, S. G.; Tan, K. C. Predicting the decrease of conjugated bilirubin with extracorporeal albumin dialysis MARS using the predialysis molar ratio of conjugated bilirubin to albumin. LiVer Transplant. 2002, 8, 591. (2) Tiribelli, C.; Ostrow, J. D. The molecular basis of bilirubin encephalopathy and toxicity: Report of an EASL Single Topic Conference, Trieste, Italy, 1-2 October, 2004. J. Hepatol. 2005, 43, 156. (3) Zhu, X. X.; Brown, G. R.; St-Pierre, L. E. Adsorption of bilirubin with polypeptide-coated resins. Biomater. Artif. Cells, Artif. Organs 1990, 18, 75. (4) Sideman, S.; Mor, L.; Mordohovich, D.; Mihich, M.; Zinder, O.; Brandes, J. M. In vivo hemoperfusion studies of unconjugated bilirubin removal by ion exchange resin. Trans. Am. Soc. Artif. Intern. Organs 1981, 27, 434. (5) Idezuki, Y.; Hamaguchi, M.; Hamabe, S.; Moriya, H.; Nagashima, T.; Watanabe, H.; Sonoda, T.; Teramoto, K.; Kikuchi, T.; Tanzawa, H. Removal of bilirubin and bile acid with a new anion exchange resin: experimental background and clinical experiences. Trans. Am. Soc. Artif. Intern. Organs 1981, 27, 428. (6) Brown, G. R. Int. J. Bio-Chromatogr. 1994, 1, 73. (7) Chandy, T.; Sharma, C. P. Polylysine-immobilized chitosan beads as adsorbents for bilirubin. Artif. Organs 1992, 16, 568. (8) Yamazaki, Z.; Inoue, N.; Wada, T.; Oda, T.; Atsumi, K.; Kataoka, K.; Fujisaki, Y. Use of an AR-1 resin column to reduce bilirubin-level in modified ascitic fluid. Trans. Am. Soc. Artif. Intern. Organs 1979, 25, 480. (9) Morimoto, T.; Matsushima, M.; Sowa, N.; Ide, K.; Sawanishi, S. Plasma adsorption using bilirubin-adsorbent materials as a treatment for patients with hepatic failure. Artif. Organs 1989, 13, 447. (10) Avramescu, M. E.; Sager, W. F. C.; Borneman, Z.; Wessling, M. J. Adsorptive membranes for bilirubin removal. J. Chromatogr., B 2004, 803, 215. (11) Yu, Y.; He, B.; Gu, H. Adsorption of bilirubin by amine-containing crosslinked chitosan resins. Artif. Cells Blood Substitutes, Immobilization Biotechnol. 2000, 28, 307. (12) Kuroda, H.; Ranaka, T.; Osawa, Z. Selective adsorption of bilirubin by macroporous poly(glycidyl methacrylate-co-divinylbenzene) beads. Die Angew. Makromol. Chem. 1996, 237, 143. (13) Ahmad, N.; Arif, K.; Faisal, S. M.; Neyaz, M. K.; Tayyab, S.; Owais, M. PLGA-microsphere mediated clearance of bilirubin in temporarily hyperbilirubinemic rats: An alternate strategy for the treatment of experimental jaundice. Biochim. Biophys. Acta 2006, 1760, 227. (14) Kocakulak, M.; Denizli, A.; Rad, A. Y.; Piskin, E. New sorbent for bilirubin removal from human plasma: Cibacron blue F3GA-immobilized poly(EGDMA-HEMA) microbeads. J. Chromatogr., B 1997, 693, 271. (15) Denizli, A.; Kocakulak, M.; Piskin, E. Alkali blue 6B-derivatized poly(egdma/hema) microbeads for bilirubin removal from human plasma. J. Macromol. Sci., Pure Appl. Chem. 1998, A35, 137. (16) Denizli, A.; Kocakulak, M.; Piskin, E. Specific sorbents for bilirubin removal from human plasma: Congo red-modified poly (EGDMA/HEMA) microbeads. J. Appl. Polym. Sci. 1998, 683, 73. (17) Denizli, A.; Kocakulak, M.; Piskin, E. Bilirubin removal from human plasma in a packed-bed column system with dye-affinity microbeads. J. Chromatogr., B 1998, 707, 25. (18) Senel, S.; Denizli, F.; Yavuz, H.; Denizli, A. Bilirubin removal from human plasma by dye affinity microporous hollow fibers. Sep. Sci. Technol. 2002, 37, 1989. (19) Uzun, L.; Denizli, A. Bilirubin removal performance of immobilized albumin in a magnetically stabilized fluidized bed. J. Biomater. Sci.,Polym. Ed. 2006, 17, 791. (20) Mosbach, K. Molecular Imprinting. Trends Biochem. Sci. 1994, 19, 9. (21) Sellergren, B. Molecularly Imprinted Polymers: Man-made Mimics of Antibodies and Their Application in Analytical Chemistry; Elsevier: Amsterdam, 2001. (22) Wulff, G.; Knorr, K. Stoichiometric noncovalent interaction in molecular imprinting. Bioseparation 2002, 10, 257. (23) Whitcombe, M.; Rodriguez, M. E.; Villar, P.; Vulfon, E. N. A new method for the introduction of recognition site functionality into polymers prepared by molecular imprinting: Synthesis and characterization of polymeric receptors for cholesterol. J. Am. Chem. Soc. 1995, 117, 7105. (24) Andac, M.; Say, R.; Denizli, A. Molecular recognition based cadmium removal from human plasma. J. Chromatogr., B 2004, 811, 119. (25) Andac, M.; Ozyapi, E.; Senel, S.; Say, R.; Denizli, A. Ion-selective imprinted beads for aluminum removal from aqueous solutions. Ind. Eng. Chem. Res. 2006, 45, 1780.

(26) Zimmerman, S. C.; Lemcoff, N. G. Synthetic hosts via molecular imprintingsAre universal synthetic antibodies realistically possible? Chem. Commun. 2004, 1, 5. (27) Fu, Q.; Sanbe, H.; Kagawa, C.; Kunimoto, K. K.; Haginaka, J. Uniformly sized molecularly imprinted polymer for (S)-nilvadipine. Comparison of chiral recognition ability with HPLC chiral stationary phases based on a protein. J. Anal. Chem. 2003, 75, 191. (28) Bu¨yu¨ktiryaki, S.; Say, R.; Erso¨z, A.; Birlik, E.; Denizli, A. Selective preconcentration of thorium in the presence of UO22+, Ce3+ and La3+ using Th(IV)-imprinted polymer. Talanta 2005, 67, 640. (29) Shi, H.; Tsai, W. B.; Garrison, M. D.; Ferrari, S.; Ratner, B. D. Template-imprinted nanostructured surfaces for protein recognition. Nature 1999, 398, 593. (30) Syu, M. J.; Deng, J. H.; Nian, Y. M. Towards bilirubin imprinted poly(methacrylic acid-co-ethylene glycol dimethylacrylate) for the specific binding of R-bilirubin. Anal. Chim. Acta 2004, 504, 167. (31) Syu, M. J.; Nian, Y. M. An allosteric model for the binding of bilirubin to the bilirubin imprinted poly(methacrylic acid-co-ethylene glycol dimethylacrylate). Anal. Chim. Acta 2005, 53, 997. (32) Syu, M. J.; Deng, J. H.; Nian, Y. M.; Chiu, T. C.; Wu, A. H. Binding specificity of R-bilirubin-imprinted poly(methacrylic acid-coethylene glycol dimethylacrylate) toward R-bilirubin. Biomaterials 2005, 26, 4684. (33) Syu, M. J.; Nian, Y. M.; Chang, Y. S.; Lin, X. Z.; Shiesh, S. C.; Chou, T. C. Ionic effect on the binding of bilirubin to the imprinted poly(methacrylic acid-co-ethylene glycol dimethylacrylate). J. Chromatogr., A 2006, 1122, 54. (34) Dai, S.; Burleig, M. C.; Shin, Y.; Morrow, C. C.; Barnes, C. E. Imprint coating: A novel synthesis of selective functionalized ordered mesoporous sorbents. Angew. Chem., Int. Ed. Engl. 1999, 38, 1235. (35) Xia, B.; Zhang, G.; Zhang, F. Bilirubin removal by Cibacron Blue F3GA attached nylon-based hydrophilic affinity membrane. J. Membr. Sci. 2003, 226, 9. (36) Annesini, M. C.; Di Paola, L.; Marrelli, L.; Piemonte, V.; Turchetti, L. Bilirubin removal from albumin-containing solution by adsorption on polymer resin. Int. J. Artif. Organs 2005, 28, 686. (37) Ishida, H.; Sato, S.; Sannomiya, A.; Tsuji, K. Coagulopathy after bilirubin adsorption in a living-related partial liver transplant recipient. Transplant. Proc. 2000, 32, 2241. (38) Davies, C. R.; Malchesky, P. S.; Saide, G. M. Temperature and albumin effects on adsorption of bilirubin from standard solution using anion-exchange resin. Artif. Organs 1990, 14, 14. (39) Zhang, L.; Jin, G. Bilirubin removal from human plasma by Cibacron Blue F3GA using immobilized microporous affinity membranous capillary method. J. Chromatogr., B 2005, 821, 112. (40) Henning, D. S.; Brown, G. R.; St-Pierre, L. E. Polymer resins with amino acid containing pendants for sorption of bilirubin. II. Polyamide resins with various basic amino acids. Int. J. Artif. Organs 1986, 9, 33. (41) Kanai, F.; Takahama, T.; Lizuka, I.; Hiraishi, M.; Yamazaki, Z.; Fujimori, Y.; Maruyama, Y.; Wada, T.; Asano, K.; Sonoda, T. Artif. Organs 1985, 9, 75. (42) Chernova, I. A.; Gurevich, K. Y. Influence of the polymeric morphology of sorbents on their properties in affinity chromatography. J. Membr. Sci. 1996, 113, 161. (43) Alvarez, C.; Strumia, M.; Bertorello, H. Influence of the polymeric morphology of sorbents on their properties in affinity chromatography. J. Biochem. Biophys. Methods 2001, 49, 649. (44) Wang, H.; Ma, J.; Zhang, Y.; He, B. Adsorption of bilirubin on the polymeric β-cyclodextrin supported by partially aminated polyacrylamide gel. React. Funct. Polym. 1997, 32, 1. (45) Rad, A. Y.; Yavuz, H.; Kocakulak, M.; Denizli, A. Bilirubin removal from human plasma with albumin immobilised magnetic poly(2hydroxyethyl methacrylate) beads. Macromol. Biosci. 2003, 3, 471. (46) Ma, Z.; Kotaki, M.; Ramakrishna, S. Electrospun cellulose nanofiber as affinity membrane. J. Membr. Sci. 2005, 265, 115. (47) Ho, Y. S.; McKay, G. Pseudo-second order model for sorption processes. Proc. Biochem. 1999, 34, 451. (48) Oncel, S.; Uzun, L.; Garipcan, B.; Denizli, A. Synthesis of phenylalanine-containing hydrophobic beads for lysozyme adsorption. Ind. Eng. Chem. Res. 2005, 44, 7049. (49) Allen, S. J.; Koumanova, B.; Kircheva, Z.; Nenkova, S. Adsorption of 2-nitrophenol by technical hydrolysis lignin: Kinetics, mass transfer, and equilibrium studies. Ind. Eng. Chem. Res. 2005, 44, 2281.

ReceiVed for reView August 25, 2006 ReVised manuscript receiVed January 31, 2007 Accepted February 24, 2007 IE0611249