Neutral Styrene Divinylbenzene Copolymers for ... - ACS Publications

In acute liver failure, a range of potentially toxic substances accumulate in the systemic circulation of the patient.(1) Metabolites associated with ...
0 downloads 0 Views 5MB Size
1322

Biomacromolecules 2008, 9, 1322–1328

Neutral Styrene Divinylbenzene Copolymers for Adsorption of Toxins in Liver Failure Viktoria Weber,*,† Ingrid Linsberger,† Maria Hauner,† André Leistner,‡ Aniela Leistner,‡ and Dieter Falkenhagen† Center for Biomedical Technology, Danube University Krems, Dr. Karl Dorrek-Straße 30, A-3500 Krems, Austria, and Polymerics GembH, Landsberger Allee 378, D-12681 Berlin, Germany Received December 17, 2007; Revised Manuscript Received February 18, 2008

In artificial extracorporeal liver support systems, albumin-bound toxins such as bilirubin, bile acids, or aromatic amino acids are removed by adsorption to polymer beads. To overcome the potential weaknesses of anion exhange polymers currently used in liver support, namely, binding of heparin and activation of coagulation, we prepared two series of neutral polystyrene divinylbenzene resins with average pore sizes of 5-6 and 8-9 nm, respectively. In in vitro experiments using human plasma spiked with bilirubin, cholic acid, tryptophan, and phenol, we found that only pores larger than 5-6 nm were accessible to strongly albumin-bound substances, such as bilirubin. On the other hand, less strongly albumin-bound substances, such as bile acids, were efficiently bound by polymers of the small pore size range due to a higher accessible surface area. None of the neutral resins bound significant amounts of heparin. To assess the influence of the polymers on activation of coagulation, generation of thrombin-antithrombin complexes (TAT) was measured at different citrate concentrations. While none of the neutral polymers induced TAT generation, TAT levels were significantly elevated after incubation of plasma with an anion exchange polymer that is in clinical use for extracorporeal liver support. Binding characteristics of the neutral resins for the natural anticoagulants protein C and antithrombin showed remarkable differences, with weak binding of antithrombin but strong removal of protein C, not only for the anion exchanger, but also for neutral polymers of the large pore size range. In conclusion, neutral polystyrene divinylbenzene polymers with a pore size larger than 5-6 nm are efficient adsorbents for albumin-bound toxins that do not induce generation of thrombin-antithrombin complexes.

Introduction In acute liver failure, a range of potentially toxic substances accumulate in the systemic circulation of the patient.1 Metabolites associated with liver failure include hydrophobic compounds (e.g., bilirubin, bile acids, hydrophobic amino acids, fatty acids), which are bound to plasma proteins (in most cases albumin) with varying affinity and water-soluble compounds of low molecular weight (e.g., ammonia). While water-soluble metabolites can be removed from the patient’s circulation with dialysis techniques, these techniques fail to remove albuminbound toxins efficiently.2 Therefore, various adsorbent-based liver assist devices have been introduced in the last decades to either sustain liver function long enough to permit the organ’s regeneration and functional recovery or to bridge patients to transplantation.3–5 Adsorbents used in these systems include activated carbon,5 copolymers of styrene and divinylbenzene,6 and anion exhange resins.6,7 The latter bind bilirubin with high efficiency; however, there are indications that positively-charged polymers also remove coagulation factors and thus lead to disturbances of the coagulation system.8 This concerns especially those liver support systems in which the plasma or a plasma fraction containing the albumin-bound toxins and coagulation factors, most of which have relative molecular masses similar to albumin, is in direct contact with the adsorbent polymers. * To whom correspondence should be addressed. Phone: ++43 2732 893 2632. Fax: + +43 2732 893 4600. E-mail: [email protected]. † Danube University Krems. ‡ Polymerics GembH.

Coating with hydrophilic polymers, such as poly(2-hydroxyethyl) methacrylate, is an approach to enhance the biocompatibility of anion exchange resins and to reduce nonspecific adsorption of plasma proteins, but the coating also reduces the adsorption rate of bilirubin.7 Pretreatment of positively-charged polymers with unfractionated heparin has been shown to reduce activation of coagulation without affecting bilirubin adsorption.9 Alternatively, neutral polymers might be biocompatible replacements for anion exchange resins in liver support therapy. Therefore, the aim of this study was to assess the effect of pore size distribution on the adsorption characteristics of neutral styrene divinylbenzene copolymers for albumin-bound toxins related to liver failure. Based on the working hypothesis that neutral polymers require a certain minimal pore size to efficiently adsorb strongly albumin-bound toxins such as bilirubin10 and that this minimal pore size has to be higher than the Stokes radius of albumin to allow for the entry of this transport molecule into the pores of the adsorbent, two series of adorbents with average pore sizes in the range of 5 and 9 nm, respectively, were developed and tested for adsorption of bilirubin, cholic acid, tryptophan, and phenol, as well as for activation of coagulation. In addition to the influence of pore size, the influence of particle size on the adsorption characteristics was assessed. We demonstrate that there is indeed a critical pore size for neutral polymers to become effective adsorbents for albumin-bound toxins. Our data suggest that neutral polymers might be efficient alternatives to anion exchange resins in extracorporeal liver support.

10.1021/bm701396n CCC: $40.75  2008 American Chemical Society Published on Web 03/18/2008

Neutral Styrene Divinylbenzene Copolymers

Materials and Methods Chemicals and Plasma. Unfractionated heparin (5000 IU/mL) was from Baxter (Vienna, Austria). Bilirubin, phenol, cholic acid, and L-tryptophan were purchased from Sigma-Aldrich (Vienna, Austria). To obtain human plasma, blood was drawn from healthy volunteers into tubes containing trisodium citrate (Vacuette, Greiner Bio-One, Kremsmünster, Austria). The blood was centrifuged for 10 min at 3600 g at room temperature to remove the blood cells. Preparation of Adsorbents. Adsorbent polymers were prepared by suspension polymerization of divinylbenzene and styrene in the presence of an inert solvent (porogen) and modulators as described.11 Pore size and particle size were controlled by a variation of the porogen and modulator. The resulting rigid adsorbent particles were of spherical shape, with diameters in the range from 1 to 100 µm, as confirmed by optical and scanning electron microscopy. After polymerization, the adsorbent polymers were purified in a chromatographic column by subsequent washing steps with water and methanol. Wash solutions were monitored for residual monomers with reversed-phase HPLC on a Symmetry Shield RP18 column (Waters GmbH, Eschborn, Germany) using isocratic elution with methanol/water (9:1) at a flow rate of 0.8 mL/min. Residual monomers were monitored by photo diode array detection at 236 and 228 nm and the purification was continued until the filtrates were free from residual monomers. The polymers were then dried for 12 h in vacuum at 120–130 °C. Complete evaporation of the solvent was verified by thermogravimetric analysis. Dry adsorbent polymers were classified using a type 100 MZR zigzag classifier (Hosokawa-Alpine AG, Germany) to obtain particle size fractions of average sizes of 6, 15, and 28 µm. Particle size distributions of these fractions were measured in isopropanol using a CILAS 1064 laser granulometer (Cilas, France). Analysis of Specific Surface Area and Pore Size Distribution. Specific surface area and pore size distribution of the adsorbents were determined by recording nitrogen adsorption and desorption isotherms at 77 K (liquid nitrogen) using an ASAP 2010 surface area and porosimetry analyzer (Micromeritics Instrument Corporation, U.S.A.). Data analysis of the isotherms was performed using the BET method (Brunauer, Emmett, and Teller) for specific surface area determination, the H-K method (Horvath–Kawazoe) for micropores (0–2 nm), and the BJH method (Barrett, Joyner, Halenda) for mesopores (2– 50 nm) and macropores (>50 nm). The average pore size diameter davg was calculated using the total pore volume Vtotal according to Gurwitsch’s rule [DIN 66134] and the specific surface area A as davg ) 4 Vtotal/A. Preparation of Spiked Plasma and Assessment of Adsorption Characteristics in Batch Experiments. Human plasma was spiked with bilirubin (300 µm), cholic acid (100 µm), phenol (2 mM), and tryptophan (100 µm). To this end, these substances were dissolved in a small volume of 0.3 M NaOH, and the solution was added to the plasma and stirred for 60 min at room temperature to allow for the binding of the substances to albumin. Finally, an equivalent volume of 0.3 M HCl was added to neutralize the NaOH. Prior to use for adsorption experiments, the adsorbents were washed with ethanol, water, and 0.9% saline solution for 60 min each using a ratio of 1 volume part adsorbent and 4 volume parts liquid for each washing step. After the last washing step, a 50% (vol/vol) suspension of adsorbent in 0.9% saline was prepared and autoclaved for 25 min at 121 °C. Aliquots of 1 mL of adsorbent (wet pellet volume after centrifugation at 3000g) were incubated with 9 mL of spiked plasma for 60 min at 37 °C with constant shaking. Samples for the quantification of bilirubin, cholic acid, tryptophan, and phenol were drawn at 0, 15, 30, and 60 min and stored at –80 °C until further processing. As cholic acid was completely removed under these conditions, additional experiments were performed using only 1% of adsorbent (wet pellet volume) related to plasma. The preparation of the spiked plasma and the subsequent batch experiments were performed in tubes protected from light, since bilirubin is light-sensitive.

Biomacromolecules, Vol. 9, No. 4, 2008

1323

Quantification of Albumin, Bilirubin, Cholic Acid, Tryptophan, and Phenol. Albumin, total bilirubin, and cholic acid were quantified on an automated analyzer (Hitachi 902). Albumin and bilirubin were determined with reagent sets from Roche (Penzberg, Germany). Albumin was quantified with the ALBplus reagent set, which is based on the bromcresol green method. The bilirubin test is based on coupling of total bilirubin with a diazonium ion in a strongly acidic medium in the presence of a solubilizing agent. The intensity of the color of the azobilirubin is proportional to the total bilirubin concentration and is quantified photometrically. Cholic acid was quantified with an enzymatic method using a reagent set from Trinity Biotech (Wicklow, Irland). Prior to quantification of tryptophan and phenol, plasma proteins were precipitated by the addition of a 10-fold excess of methanol to the plasma samples and incubation for 20 min at –70 °C. Precipitated protein was removed by centrifugation, and tryptophan and phenol were quantified in the supernatant by absorption at 280 nm after separation by reversed-phase HPLC on a Nucleosil100-5C18 column (150 × 4.6 mm; Varian, Darmstadt, Germany) using isocratic elution with methanol/ water (1:1) at a flow rate of 1 mL/min. Adsorption of Antithrombin and Protein C. Aliquots of 300 µL of adsorbents (wet pellet volume) were incubated with 2.7 mL of plasma for 60 min at 37 °C with constant shaking. Samples were drawn after 15 and 60 min and stored at –80 °C until analysis. Antithrombin was quantified with a test kit from Coachrom, Vienna, Austria (Coamatic Antithrombin). The test was performed according to the instructions of the supplier. Briefly, the method is based on the addition of factor Xa (FXa) in excess to a plasma sample. FXa binds to antithrombin (AT) in the sample, and residual FXa leads to hydrolysis of a chromogenic substrate, which is quantified photometrically. Antithrombin levels are reported as activity (%), with a range for normal plasma of 106 ( 18%. Protein C was determined using the Coamatic Protein C (Coachrom, Vienna, Austria). In this test, protein C is enzymatically cleaved by the addition of a specific enzyme to the plasma sample and the resulting activated protein C is quantified via cleavage of a chromogenic substrate. Protein C levels are given as activity (%), with a range for normal plasma of 70-149%. Adsorption of Heparin. Adsorbents were incubated with human plasma containing 10 IU/mL of unfractionated heparin at 37 °C with constant shaking. The ratio of adsorbent to plasma was 1:9 (vol/vol). Samples were drawn at 0, 15, 30, and 60 min, and heparin was quantified with the Coamatic Heparin test (Coachrom, Vienna, Austria) according to the instructions of the supplier. This test is based on the addition of excess Factor Xa and a chromogenic substrate to the sample, whereby two competing reactions occur simultaneously: inhibition of FXa by the antithrombin-heparin complex and hydrolysis of the substrate by excess FXa. The concentration of heparin is inversely proportional to the increase in absorbance caused by hydrolysis of the substrate. Generation of Thrombin-Antithrombin Complexes. Because citrate suppresses the formation of thrombin-antithrombin complexes, blood used to obtain the plasma for this series of experiments was drawn into tubes containing a lower citrate concentration as compared to commercially available tubes; the target citrate concentration in the blood was 4 mM. Aliquots of 300 µL of adsorbent (wet pellet volume) were incubated with 2.7 mL of plasma for 60 min at 37 °C with constant shaking. Samples were taken after 15 and 60 min, and thrombinantithrombin complexes (TAT) were quantified by ELISA (Enzygnost TATmicro, Dade Behring, Vienna, Austria). Statistical Analysis. Experiments were performed in triplicate. Normal distribution of the data was confirmed using the KolmogrovSmirnov test. Normally distributed data were analyzed with the Student′s t-test. All statistics were performed using Sigma Stat (Jandel, San Rafael, CA). Significance was accepted at P < 0.05.

1324

Biomacromolecules, Vol. 9, No. 4, 2008

Weber et al.

Table 1. Pore Size Characteristics and Specific Surface Areas for the Neutral Resins Used in this Study mesopore volume (BJH desorption) [cm3/g]

micropore volume (H-K) [cm3/g]

sample (avg size)

BET surface area [m2/g]

2–50 nm

2–300 nm

MPD [nm]

0.5–2 nm

avg pore diameter [nm]

L3 (28 µm) L2 (15 µm) L1 (6 µm) S3 (28 µm) S2 (15 µm) S1 (6 µm)

683 678 660 792 736 890

1.17 1.30 1.34 0.82 0.84 0.96

1.18 1.60 1.43 0.89 0.89 0.98

59.4 70.2 88.6 57.7 86.6 58.5

0.28 0.28 0.27 0.32 0.30 0.36

7.69 8.75 9.38 5.34 5.54 5.19

Results Characterization of the Styrene-Divinylbenzene Copolymers. Two series of polymers with average pore sizes in the range of 8-9 nm and 5-6 nm, respectively, were produced. For both groups, particles with average diameters of 28, 15, and 6 µm were prepared. Average particle sizes, pore size characteristics, and specific surface areas for the polymers are summarized in Table 1. Figure 1 shows typical raster electron micrographs of the polymers (large pore size series). Figure 2 shows the pore size distribution (combination of micro and mesopores) for polymer L-1 (upper panel) and S-1 (lower panel). Adsorption of Bilirubin, Bile Acids, Tryptophan, and Phenol. Adsorption of toxins related to liver failure was assessed with in vitro batch experiments using human plasma spiked with

Figure 1. Raster electron micrographs of adsorbent particles from the large pore size series.

bilirubin (300 µm), cholic acid (100 µm), L-tryptophan (100 µm), and phenol (2 mM). These spike concentrations of bilirubin, tryprophan, and cholic acid are in the range of toxin levels usually found in patients with liver failure.12 The phenol concentration used in this study was significantly higher than reported from patients with liver failure13 and was chosen as a marker for weakly bound aromatic compounds. The adsorption of bilirubin, cholic acid, and tryptophan is shown in Figure 3a-c. Polymers L1-L3 efficiently adsorbed bilirubin, which is strongly albumin-bound,14 with an association constant of 9.5 × 107 M-1, whereas S1-S3 exhibited comparatively poor bilirubin binding. The difference in bilirubin binding was statistically significant for all three particle size groups (P < 0.005 at 60 min). This indicates that pores smaller than 5-6 nm are not accessible to strongly albumin-bound substances. For both pore size ranges, bilirubin adsorption correlated inversely with particle size, but this effect was more obvious for adsorbents with small pores, pointing to the strong effect of the outer surface in case of unaccessible pores. Cholic acid (association constant with albumin: 0.33 × 104 -1 M ) was completely removed by adsorbents of both pore size

Figure 2. Pore size distribution for polymer L-1 (upper panel) and S-1 (lower panel).

Neutral Styrene Divinylbenzene Copolymers

Biomacromolecules, Vol. 9, No. 4, 2008

1325

Table 2. Amount of Bilirubin Bound to the Neutral Resins in Batch Experiments after 60 Mina bilirubin bound sample (particle size)

(µmol/mL adsorbent)

L3 (28 µm) L2 (15 µm) L1 (6 µm) S3 (28 µm) S2 (15 µm) S1 (6 µm)

1.45 1.88 2.30 0.32 0.51 0.70

cholic acid bound

(nmol/m2)

(µmol/mL adsorbent)

(nmol/m2)

7.3 9.6 11.9 1.4 2.4 2.7

9.9 8.3 4.0 13.7 13.5 13.4

62.7 53.0 26.2 59.6 63.2 51.9

a The BET surface area (Table 1) was used to calculate the amount of toxins bound per m2.

Figure 4. Adsorption of unfractionated heparin (incubation time in the batch test, 60 min). Heparin binding for the neutral polymers L1-L3 and S1-S3 was compared to a positively-charged polymer (CP) that is used clinically in extracorporeal liver support. Note the different particle size for the charged polymer (200-350 µm) .

Figure 3. Adsorption of bilirubin (a), cholic acid (b), and tryptophan (c) from human plasma. Adsorption was tested in vitro using human plasma spiked with bilirubin (300 µm), cholic acid (100 µm), tryptohan (100 µm), and phenol (2 mM) as described in Materials and Methods. The ratio of adsorbent to plasma was 1:9 (vol/vol) for bilirubin and tryptophan and 1:99 (vol/vol) for cholic acid.

series in batch experiments using 10 vol % adsorbent in plasma. Therefore, the amount of adsorbent was reduced to 1 vol % (Figure 3b). The adsorption capacity for cholic acid correlated inversely with pore size; for S1-S3, cholic acid was almost completely eliminated even with 1 vol % of adsorbent. Again, the difference in cholic acid binding between the two pore size ranges was statistically significant (P < 0.001 at 60 min for all three particle size groups). The observation that cholic acid concentrations in the supernatant, after an initial drop, increased over time for polymers with large pores points to a release of initially bound cholic acid from the adsorbent during the course of the experiment, most likely due to competition with bilirubin. The amounts of bilirubin and cholic acid bound to the different polymers are summarized in Table 2. In principle, tryptophan (KA ) 1 × 104 M-1) showed a similar adsorption behavior as cholic acid (preferred binding

to polymers with small pores; Figure 3c; the difference was statistically significant with P < 0.05 at 60 min), but in contrast to cholic acid, tryptophan binding was only moderate. However, tryptophan can be removed to a large extent by dialysis, which is performed during extracorporeal liver support therapy. Phenol, which is only loosely associated with albumin, bound equally well to all polymers tested. For all polymers, irrespective of pore and particle size, phenol concentrations were reduced from 2 ( 0.2 mmol/L (initial spike concentration) to 0.15 ( 0.03 mmol/L after 5 min without any further decrease at later time points, indicating a saturation of the adsorbents by the high phenol concentrations used. Plasma albumin levels remained practically unchanged during the adsorption experiments. In conclusion, a high percentage of micropores favors adsorption of substances with medium or low affinity to albumin, whereas efficient adsorption of strongly albumin-bound substances requires a minimal pore size of at least 5.5 nm. Adsorption of Heparin, Antithrombin-III, and Protein C. Unfractionated heparin is commonly used as an anticoagulant in extracorporeal blood purification. Therefore, heparin binding by adsorbents is an unwanted side effect that should be minimized. Binding of heparin to the neutral polymers of both pore size ranges is shown in Figure 4. None of the polymers of the series S1-S3 showed significant binding of heparin. Polymers L1-L3 removed minor amounts of heparin, while heparin was almost completely removed by an anion exchange polymer (CP), which is in clinical use for extracorporeal liver support (polystyrene divinylbenzene copolymer modified with trimethylamine, particle size 200-350 µm). AT-III is a 58 kDa plasma glycoprotein, which binds to and inactivates thrombin as well as factors Xa and IXa and thus

1326

Biomacromolecules, Vol. 9, No. 4, 2008

Figure 5. Binding of AT-III to the neutral polymers S1-S3 and L1-L3 in comparison to a positively-charged polymer (CP) that is used in extracorporeal liver support.

Weber et al.

Figure 7. Formation of thrombin-antithrombin complexes (TAT) upon incubation of adsorbents in plasma in presence of different citrate concentrations. Note that the citrate concentrations given in the figure legend refer to the value measured in blood; the concentration in plasma was about twice as high. S1-S3 and L1-L3, neutral polymers; CP, anion exchanger (note the different particle size).

place to some extent for the charged polymer even in the presence of citrate concentrations that are clinically used in regional citrate anticoagulation.

Discussion

Figure 6. Binding of protein C to the neutral polymers S1-S3 and L1-L3 in comparison to a positively-charged polymer (CP) that is used in extracorporeal liver support.

acts as a natural anticoagulant. As shown in Figure 5, AT-III levels did not change upon incubation with polymers S1-S3 under the experimental conditions used, while there was a statistically significant reduction for polymers L1-L3. Still, the AT-III levels remained in the normal plasma range after incubation with the adsorbents. There was no significant reduction in AT-III levels for the anion exchanger (CP). Binding of protein C, another natural anticoagulant, to polymers S1-S3 and L1-L3 is shown in Figure 6. As for ATIII, polymers S1-S3 did not bind significant amounts of protein C. Upon incubation with polymers of large pore size (L1-L3), protein C levels were strongly reduced with complete removal of protein C after 60 min for L1 and L2 and a reduction to 20% of the initial amount for L3. Likewise, a strong reduction in protein C levels was observed for the anion exchanger (CP in Figure 6), with almost complete adsorption of protein C after 60 min, despite its much larger particle size. Formation of TAT Complexes. The formation of thrombinantithrombin complexes was assessed to quantify activation of coagulation. Because citrate inhibits activation of coagulation, adsorbents were incubated in plasma that contained reduced amounts of citrate as compared to standard tubes used for blood sampling. The final citrate concentration in blood was 5 mM versus 12.1 mM for the standard tubes. Generation of TAT complexes under these conditions is shown in Figure 7. At neither of the two different citrate concentrations, there was significant generation of thrombin-antithrombin complexes for the neutral polymers. In contrast, TAT levels were considerably elevated for the charged polymer at both citrate concentrations, and TAT generation correlated inversely with citrate concentration. This means that activation of coagulation took

Several extracorporeal liver support systems have been introduced in the past decades. Common to these systems is the use of adsorbent polymers for the binding of hydrophobic, mainly albumin-bound substances that accumulate in the blood during liver failure. Adsorbents used in liver support systems are selective and eliminate groups of substances with common physicochemical properties. While this approach allows for the binding of a range of pathogenic factors, there is the potential shortcoming of also removing valuable substances, which might result in disturbed equilibria in the body. In the context of liver failure, alterations of the coagulation cascade are particularly relevant, as blood coagulation is impaired in patients with liver failure both due to reduced synthesis of coagulation factors in the liver and to enhanced consumption. Because there have been reports in the literature on activation of coagulation by extracorporeal liver support systems8,15 or by adsorbents used in these systems,9 we aim at the development of polymers that efficiently bind toxins related to liver failure without negatively affecting coagulation. Most liver support systems that are in clinical application use combinations of adsorbents. Many of the albumin-bound toxins can be efficiently adsorbed with neutral polymers. In addition, anion exchangers are currently used to bind bilirubin, although there have been a number of approaches to use neutral polymers for bilirubin adsorption as well (summarized in ref 16). Many of the factors of the coagulation cascade and natural anticoagulants such as protein C are negatively-charged at physiological pH and thus are prone to adsorption by anion exchangers. From this point of view, the use of neutral resins seems preferable. In this study, we aimed at the modulation of adsorption characteristics of neutral styrene divinylbenzene copolymers by variation of their pore size distribution. We assumed that polymers have to have a certain critical pore size to allow for the entry of strongly albumin-bound substances such as bilirubin. Normally, bilirubin is conjugated with glucuronic acid in the hepatocyte prior to secretion in the bile. In liver failure, conjugation and excretion of bilirubin is impaired. Unconjugated bilirubin is strongly albumin-bound, with an association constant of 9.5 × 107 M-1 corresponding to an unbound fraction in the equilibrium of less

Neutral Styrene Divinylbenzene Copolymers

than 0.1%. The albumin molecule in solution is an ellipsoid of 4 × 14 nm.17 We found previously that styrene divinylbenzene copolymers with a high percentage of micropores (size range 0.5 to 2 nm) are poor adsorbents for bilirubin, while polymers with pore sizes greater than 7 nm showed high bilirubin adsorption (data not shown). Therefore, we prepared two series of polymers with average pore sizes of 5 (S1-S3) and 9 (L1-L3) nm, respectively. As expected, polymers with small pores were found to be poor bilirubin adsorbents, while large pores were associated with high bilirubin adsorption. Bile acids are less tightly bound to albumin with association constants ranging from 0.2 × 104 to 2 × 105 M-1. The association constant for cholic acid, which was used in this study, is 0.3 × 104 M-1, corresponding to an unbound fraction in the equilibrium of about 16%.18 Cholic acid was efficiently eliminated by polymers of both pore size ranges, with polymers of the small pore size range showing higher adsorption capacity. Likewise, the assumption that tryptophan (KA ) 1 × 104 M-1) would be more efficiently bound by polymers of the small pore size range was confirmed, although the overall binding capacity was moderate. However, unpublished data from our laboratory show that substantial amounts of tryptophan can be removed by dialysis. In summary, our in vitro adsorption data demonstrate that the minimal pore size for binding of strongly albumin-bound substances is above 5.5 nm. One potential limitation of this study is that the pore size values given here are average data, while the shape of the pores (e.g., cylindrical vs cone shaped) might also play a critical role. In contrast to the anion exchange polymer that is in clinical application and was used for comparison in this study, none of the neutral resins bound significant amounts of heparin, the most widely used anticoagulant used in extracorporeal blood purification. To assess if and to which amount the neutral resins developed in this study would influence parameters of blood coagulation, we measured adsorption of naturally occurring anticoagulants (antithrombin and protein C) as well as the generation of thrombin-antithrombin complexes (TAT) upon incubation of plasma with the adsorbents. TAT formation was tested in plasma obtained from blood containing 5 mM citrate; thus, the citrate concentration in the plasma was about 10 mM. This citrate concentration has been found sufficient to suppress complement activation during regional citrate anticoagulation in extracorporeal blood purification,19 which is in agreement with yet unpublished data from our own group. Under these conditions, none of the neutral polymers induced measurable TAT generation, while TAT levels remained elevated in the presence of the anion exchange polymer. As for adsorption of antithrombin and protein C, neither of the two factors was significantly removed by polymers of the small pore size range, while a significant reduction of both factors was seen for the polymers with large pores. However, while AT-III levels were reduced by about 20% but stayed in the range of normal plasma values, protein C was almost completely eliminated. Both antithrombin and protein C have relative molecular masses of about 62 kDa and similar isoelectric points of about 5. Therefore, their different adsorption characteristics and, above all, the strong binding of protein C to the neutral polymers were surprising at first sight because binding of protein C and other factors of the coagulation cascade to anion exchange resins has been attributed to their negativelycharged γ-carboxylated Gla-domains.8 However, in plasma, the carboxyl groups of Gla-domains interact with Ca2+, which

Biomacromolecules, Vol. 9, No. 4, 2008

1327

mediates membrane binding of coagulation factors20 and, consequently, their negative charge is masked. If, as in our study, Ca2+ is complexed by citrate or EDTA, the negatively-charged residues of the Gla-domain are exposed and, most likely, the binding to anion exchange resins is promoted. Supportive evidence for this model comes from a study of Rezaie et al.,21 who showed that, in the absence of Ca2+, the Gla-domain of factor Xa prevents heparin binding due to its highly negativelycharged domain. We suggest that this mechanism might play a general role whenever adsorbents are used in plasma anticoagulated with citrate. In this case, the negative charges of Gladomains, which are found on various coagulation factors, become exposed due to complexation of Ca2+, which promotes the binding of these factors especially to anion exchange polymers but, in a less-pronounced manner, also to neutral resins. The reasons for the interaction of carboxylated compounds with neutral resins remain under investigation, but untypical retention of carboxylated compounds is a phenomenon known from size exclusion chromatography on neutral polystyrene divinylbenzene copolymers (A. Leistner, personal communication) and might also be responsible for the interaction of protein C with the neutral resins in this study. Interestingly, polymer L3 showed the comparatively lowest protein C adsorption. Within the polymers of the large pore size range, L3 is the resin with the smallest pores (7.7 nm as compared to 8.7 and 9.4 nm for L2 and L1, respectively). This might indicate a window of pore sizes between 5.5 and approximately 7 nm, which on the one hand allows for efficient binding of bilirubin, but on the other hand does not lead to the removal of large amounts of coagulation factors. Further studies are underway to test this assumption.

Conclusion According to the findings of this work, the minimal pore size for neutral polymers to be efficient adsorbents for strongly albumin-bound substances such as bilirubin is about 5.5 nm. In contrast to anion exchangers, which are currently used for bilirubin binding in extracorporeal liver support, neutral resins do not adsorb heparin. Still, neutral polymers influenced coagulation parameters, although less strongly than charged polymers, and showed significant binding of the natural anticoagulant protein C. We propose that complexation of Ca2+, as it takes place for instance during citrate anticoagulation, might lead to exposure of Gla-domains with high negative charge density and promote binding of Gla-domain containing factors to adsorbent polymers. From this point of view, and under the condition that the patient’s antithrombin levels are sufficiently high, the use of neutral resins for bilirubin adsorption in combination with anticoagulation using unfractionated heparin seems optimal. Acknowledgment. We thank Ute Fichtinger for excellent technical assistance and Karin Strobl for support with citrate measurements. This work was supported by a grant from the Federal Government of Lower Austria, with cofinancing by the European Commission (article 4 EFRE). The development of polymeric microparticles having various pore size distributions was supported by Grant No. KF150701UL5 from the German Federal Ministry of Economics and Technology.

References and Notes (1) Hughes, R. D. Int. J. Artif. Organs 2002, 25, 911–917. (2) Bammens, B.; Evenepoel, P.; Verbeke, K.; Vanrenthergem, Y. Am. J. Kidney Dis. 2004, 44, 278–285.

1328

Biomacromolecules, Vol. 9, No. 4, 2008

(3) Falkenhagen, D.; Stobl, W.; Vogt, G.; Schrefl, A.; Linsberger, I.; Gerner, F. J.; Schoenhofen, M. Artif. Organs 1999, 23, 81–86. (4) Stange, J.; Mitzner, S. R.; Risler, T.; Erley, C. M.; Lauchart, W.; Goehl, H.; Klammt, S.; Peszynski, P.; Freytag, J.; Hickstein, H.; Loehr, M.; Liebe, S.; Schareck, W.; Hopt, U. T.; Schmidt, R. Artif. Organs 1999, 23, 319–330. (5) Ash, S. R.; Sullivan, T. A.; Carr, D. J. Ther. Apheresis Dial. 2006, 10, 145–153. (6) Vienken, J.; Christmann, H. Ther. Apheresis Dial. 2006, 10, 125– 131. (7) Nakayi, S.; Hayashi, N. Ther. Apheresis Dial. 2003, 7, 98–103. (8) Meijers, B. K.; Verhamme, P.; Nevens, F.; Hoylaerts, M. F.; Bammens, B.; Wilmer, A.; Arnout, J.; Vanrenthergem, Y.; Evenepoel, P Am. J. Transplant. 2007, 7, 2195–2199. (9) Weber, V.; Lind, J.; Linsberger, I.; Falkenhagen, D. Int. J. Artif. Organs 2006, 29, 512. abstract. (10) Yamazaki, Z.; Inoue, N.; Wada, T.; Oda, T.; Atsumi, K.; Kataoka, K.; Fujisaki, Y. Trans. Am. Soc. Artif. Intern. Organs 1979, 25, 480– 485. (11) Leistner, A.; Leistner, A. Adsorbing material for blood and plasma cleaning method and for albumin purification; European Patent EP1578526, 2005.

Weber et al. (12) Sauer, I. M.; Goetz, M.; Steffen, I.; Walter, G.; Kehr, D. C.; Schwartlander, R.; Hwang, Y. J.; Pascher, A.; Gerlach, J. C.; Neuhaus, P. Hepatology 2004, 39, 408–414. (13) Wiltfang, J.; Jaugsch, F.; Fuhrmann, H.; Heinze, H. J.; Canzler, H.; Schmidt, F.W.; Brunner, G. In AdVances in ammonia metabolism and hepatic encephalopathy; Soeters, P. B., Wilson, J. H. P., Meijer, A. J., Holm, H. Eds.; Elsevier: New York, 1988; pp 356–666. (14) Brodersen, R. Crit. ReV. Clin. Lab. Sci. 1980, 11, 305–399. (15) Bachli, E. B.; Schuepach, R. A.; Maggiorini, M.; Stocker, R.; Müllhaupt, B.; Renner, E. L. LiVer Int. 2007, 27, 475–484. (16) Denizli, A.; Kocakulak, M.; Piskin, E. J. Chromatogr., B 1988, 707, 25–31. (17) Peters, T. All About Albumin. Biochemistry, Genetics, and Medical Applications; Academic Press: New York, 1996. (18) Roda, A.; Cappelleri, R.; Aldini, R.; Roda, E.; Barbara, L. J. Lipid Res. 1982, 23, 490–495. (19) Bosch, T.; Lennertz, A.; Duhr, C.; Fink, E.; Samtleben, W. Artif. Organs 2000, 24, 696–704. (20) Huang, M.; Rigby, J. C.; Morelli, X.; Grant, M. A.; Huang, G.; Furie, B.; Seaton, B.; Furie, B. C. Nat. Struct. Biol. 2003, 10, 751–756. (21) Rezaie, A. R. J. Biol. Chem. 1998, 273, 16824–16827.

BM701396N