1286
Biomacromolecules 2010, 11, 1286–1290
Structure-Activity Relationships of Antibacterial and Biocompatible Copolymers Thomas R. Stratton,† John A. Howarter,† Bradley C. Allison,† Bruce M. Applegate,‡,§ and Jeffrey P. Youngblood*,† School of Materials Engineering and Departments of Food Science and Biological Sciences, Purdue University, West Lafayette, Indiana Received January 24, 2010; Revised Manuscript Received March 30, 2010
The development of polymers that are both bactericidal and biocompatible would have many applications and are currently of research interest. Following the development of strongly bactericidal copolymers of 4-vinylpyridine and poly(ethylene glycol) methyl ether methacrylate, biocompatibility assays have been completed on these materials to measure their potential biocompatibility. In this article, a new methodology for measuring protein interaction was developed for water-soluble polymers by coupling proteins to surfaces and then measuring the adsorption of copolymers onto these surfaces. Ellipsometry was then used to measure the thickness of adsorbed polymers as a measurement of biocompatibility. These results were then compared and correlated with the results of other biocompatibility assays previously conducted on these polymers, affording a greater understanding of the biocompatibility of the copolymers as well as improving the understanding of the effect of hydrophilic and hydrophobic groups that is vital for the development of these materials.
Introduction Antibacterial polymers have been studied for many years as a possible alternative to the common bactericides in use today.1-4 Current strategies for fighting bacteria focus on antibiotic medications, which have faced increasing challenges because of the development of bacterial resistance.3 Alternatively, strongly biocidal chemicals are also in use, but they are highly toxic to humans and most forms of life. Other alternative bactericides are limited by short lifetimes,5,6 bacterial resistance,4 or high production costs.7,8 Polymer approaches offer the benefits of novel methods of action and low production costs. A major class of antibacterial polymers under current research interest includes those whose active element consists of a cationic nitrogen center alongside linear alkyl chains.9 This includes ammonia- and pyridine-based compounds, among others. It has been theorized that these polymers work through the electrostatic attraction of the cation to the negatively charged groups on the cell membrane, which allows for the membrane to be disrupted by the alkyl groups, leading to lysis.10,11 Previous work has shown that poly(4-vinylpyridine) (PVP) that has been quaternized by an alkyl bromine (Q-PVP) like hexyl bromide (PVP-HB) is effective at killing a wide range of bacteria and could be employed in a wide variety of applications.9 However, all of these materials have been limited in their usefulness because of their lack of biocompatibility and tendency to cause irritation to mammalian cells.12 Previous publications have shown that one potential method of producing antibacterial and biocompatible polymers is to copolymerize 4-vinylpyridine (VP) with monomers that form known biocompatible materials like hydroxyethyl methacrylate (HEMA) and poly(ethylene glycol) methacrylate (PEGMA).13-16 However, determining the polymer or copolymer compositions that would create a biocom* To whom correspondence should be addressed. E-mail: jpyoungb@ purdue.edu. † School of Materials Engineering. ‡ Department of Food Science. § Department of Biological Sciences.
patible material has proven to be challenging because of the large number of variables involved in the synthesis of such a polymer, including the monomer composition, the structure of the cationic nitrogen group (amine-based versus pyridine-based), the bonding of the cationic nitrogen (primary, secondary, tertiary, or quaternary), and the length and structure of the alkyl groups. As of yet, no straightforward model exists to predict the biocompatibility of a specific polymer material reliably. This challenge is further increased by the complicated nature of biocompatibility, which can be measured by many different assays, each of which has its own advantages and disadvantages. In vitro tests, in particular, tend to suffer from significant drawbacks, even when completed on human cells. For example, in the common red blood cell (RBC) hemolysis assay, the erythrocytes lack the full components of most human cells (in particular, they lack a defined nucleus) and are also removed from the protective blood proteins that are involved in foreign body interactions. Previous work has shown that the results of RBC hemolysis and other in vitro assays are not always identical.16,17 Additional assays are therefore useful to better understand the biocompatibility of a material. Previous publications have explored the biocompatibility of VP copolymers14,16 but have not yet established a comprehensive understanding of what properties render these and other antibacterial polymers biocompatible. To this end, we have completed an additional measurement important for understanding biocompatibility by measuring the adhesion of our copolymers to a surface that has been treated to present a common blood plasma protein, bovine serum albumin (BSA). Plasma proteins like BSA are responsible for many tasks within the blood, helping to protect RBCs from physical stresses, maintaining osmotic pressure within the blood, and controlling host response to a foreign body by adsorbing to surfaces and materials in the body.18 This adsorption of plasma proteins occurs rapidly after initial exposure, and the response of the host is dictated by the amount and nature of the protein adsorbed on the material, making this interaction important in evaluating
10.1021/bm1000839 2010 American Chemical Society Published on Web 04/09/2010
Antibacterial and Biocompatible Copolymers
biocompatibility. Albumin is present in plasma in the highest concentration of all blood proteins and has the highest diffusion coefficient of all plasma proteins,19 allowing it to dominate initial adsorption. Albumin is not known to have any cell receptors; therefore, albumin adsorbed to a surface creates a passivated layer inhibiting other proteins from adsorbing to the surface.20 Protein adhesion is thus highly undesirable for most biomaterial applications. Hydrophilic polymers are commonly found in biomaterial applications for this reason because they impede the hydrophobic interactions necessary for protein adsorption;21 therefore, we predict that we will see favorable adhesion results from our copolymers, which feature hydrophilic poly(ethylene glycol) groups that have been shown in previous experiments to reduce protein adsorption and cell adhesion.22 Previous experiments have also shown that amphiphilic copolymers with a vaguely similar structure to our copolymers (featuring PEG and hydrophobic sytrenic groups in a copolymer) greatly decreased protein adsorption,22 which served as a major experimental inspiration for these assays. However, our experimental procedure differs slightly from this and other tests of protein adsorption: instead of attaching polymer to a surface and measuring the differential adsorption of protein, we have chosen to attach protein to a surface and measure the differential adsorption of polymer to this attached protein, in accordance with previous work,23 for primarily two reasons: first, we found the protein attachment technique to be convenient, consistent, and measurable. Second, our VP and PEGMA copolymers are designed to be water-soluble and are likely to be dispersed in aqueous solutions when they come into human contact. Through our testing method, we are able to simulate more accurately the environmental condition (dissolution in aqueous solution) that proteins will see when they encounter these polymers in vivo. We compare these results to previously completed RBC hemolysis and in vitro cell viability assays14,16 and combine these results with the results of bactericidal assays to improve our understanding of the relationship between bactericidal efficacy and biocompatibility for these polymers.
Experimental Section Materials. 2,2′-Azobisisobutyronitrile (AIBN), 1-bromohexane (hexyl bromide), VP, poly(ethylene glycol) methyl ether methacrylate of Mn 300, 475, and 1100, methanol, chloroform, dimethyl sulfoxide, hydrogen peroxide, and sulfuric acid were purchased from SigmaAldrich (St. Louis, MO). 3-Isocyanatopropyltriethoxysilane (ICPTES) was purchased from Gelest (Morrisville, PA). BSA was purchased from GE Healthcare (Piscataway, NJ). The monomers were purified to remove inhibitor. VP was purified by trap-to-trap distillation. PEGMAbased monomers were purified by column chromatography eluted in chloroform with 70-240 µm silica mesh acting as the stationary phase. All other materials were used as received. Polymer Synthesis. The synthesis and quaternization of our polymers are identical to a previous publication,13 but a brief summary is given here. Random copolymers of VP and PEGMA with a total molecular weight (Mn) of 300, 475, and 1100 g/mol (PEGMA 300, PEGMA 475, and PEGMA 1100, respectively, PEGMA n when referring to all copolymers as a group) were synthesized through free radical polymerization in chloroform using AIBN as initiator, as shown in Figure 1. Following polymerization, precipitation, and characterization, the polymers were redissolved in solvent and mixed with 1-bromohexane. This mixture was heated, precipitated, and dried to form poly(vinyl pyridine-co-poly(ethylene glycol) methacrylate)-hexylbromide (P(VP-co-PEGMA n)-HB). 1-Bromohexane was chosen as the quaternizing agent because previous work has shown that the ideal chain length for the quaternized homopolymer of PVP is six carbon
Biomacromolecules, Vol. 11, No. 5, 2010
1287
Figure 1. Synthetic scheme of P(VP-co-PEGMA n)-HB.
atoms;10 it is possible that for these copolymers, there would exist better chain lengths for the antibacterial and biocompatible properties; however, in the interest of minimizing the number of experimental variables to control, we have quaternized all of our copolymers with the same six carbon atom alkyl chain. Homopolymers of quaternized VP (PVP-HB) and PEGMA n (PPEGMA n) were also synthesized through this method. The homopolymer PPEGMA n (0 wt % VP) is not supposed to be bactericidal: it contains only PEGMA monomers and no VP monomers and is therefore not quaternized because of the lack of any nitrogen atoms for quaternization. Composition and quaternization were confirmed by 1H nuclear magnetic resonance spectroscopy24 and gel permeation chromatography results, which indicated Mn around 15 000 g/mol for all polymers.13 Previously completed work showed that essentially all VP groups have been quaternized.24 Copolymers were synthesized with a wide variety of monomer compositions ranging from 0 (i.e., the hompolymer of PEGMA n) through 100 mol % VP, representing a spectrum of copolymer compositions. Previously completed work on the reactivity ratios of these polymers showed R1R2 values ranging between 0.54 and 0.9 (Kelen-Tudos method), indicating a generally random structure.14 All VP and PEGMA n quaternized copolymers were soluble in chloroform. A wide range of PEGMA n quaternized copolymers were also soluble in water; these results are displayed in Table 1. Protein Adhesion Testing. To determine the interaction of synthesized copolymers with protein, we devised a test where water-soluble copolymers were adsorbed onto BSA-bound surfaces, following related work in this field.22,23 Wafers of silicon were cut into 1 cm2 slides and cleaned through immersion in a 2:1 solution of sulfuric acid and hydrogen peroxide (piranha solution) for 30 min. Slides were immersed for 1 h in a 1% v/v solution of ICPTES in anhydrous toluene, as following the reaction kinetics of the formation of alkyltrichlorosilane self-assembled monolayers.25 The ethoxysilane groups in the ICPTES molecule are known to react with silica forming a covalent bond with the surface,26 and the resulting isocyanate surface is reactive toward the exposed alcohol and amine groups in the protein, forming a covalently bonded monolayer of protein. Surface thickness was characterized by ellipsometry measurements on a Gaertner L116S variable angle Stokes ellipsometer at a wavelength of 544 nm and incident angle of 70°. Measurements were performed on three to five different positions for each specimen; measurements were taken after the silanization procedure and once again following the attachment of the protein layer. After protein treatment, the slides were rinsed and then immersed in solutions of synthesized polymer at concentrations of 10 mg/mL for 1 h. Slides were then rinsed again and dried under vacuum. The adsorption thickness was then measured again by ellipsometry; the thickness of the silane and protein layers are subtracted
1288
Biomacromolecules, Vol. 11, No. 5, 2010
Stratton et al.
Table 1. Solubility of P(VP-co-PEGMA n)-HB Copolymers in Watera comonomer PEGMA 300 PEGMA 475 PEGMA 1100 a
0% VP
10% VP
25% VP
50% VP
75% VP
90% VP
100% VP
O O O
O O O
O O O
O O O
O O O
X X O
X X X
O indicates that a given copolymer is water-soluble; X indicates that it is not. All percentages listed are mol %.
Figure 2. Ellipsometry results of measured thickness of P(VP-co-PEGMA n)-HB series adsorbed on albumin-isocyanate surfaces as a function of wt % VP. The right Figure displays a detail of the low VP wt % regime.
from these results for each sample, eliminating the potential effect of minor variations in preparation from sample to sample. The refractive index for albumin is estimated to be 1.52.27
Results The results of the ellipsometry assays completed to measure protein adsorption on the isocyanate-modified surfaces for each set of copolymers are presented in Figure 2. In general, the copolymers all show a similar relationship: complex behavior showing a high adsorption peak at a relatively low concentration of VP in the copolymer, which forms the most distinguishing feature in all series, followed by a drop-off and then a slow rise in adsorbed polymer. Notably, this peak appears to change substantially depending on the PEGMA monomer composition. The overall height of the peak appears to be dependent on the PEG chain length, with the shorter PEG-length monomers displaying a higher adsorption thickness at the maximum. The adsorbed thickness ranges from 40 Å on the PEGMA 300 copolymers to 20 Å on the PEGMA 1100 copolymers. Additionally at the high VP extreme, adsorbed thickness is similarly dependent on PEG length with shorter PEG showing higher adsorbed thicknesses. It must also be noted that these changes are not due to changes in the molecular weight of the polymers, which are roughly the same across the copolymer spectrum.13 In Figure 2, wt % was used in place of mol % to separate the effect of PEG length as opposed to overall PEG composition. Because two copolymers with the same wt % of PEGMA in their composition will have comparable amounts of PEG groups, we can directly compare the effect of PEG content (as opposed to PEG length), which would not be the case when mol % is used. The data suggests that at low VP content, peak adsorption placement is correlated to the overall PEG composition and not the length of the PEG chain, unlike the total thickness adsorbed, which is correlated to PEG length. Because the VP content is increased, the adsorbed thickness drops off quickly, and between 10 and 20 wt % VP composition, we observe the lowest adsorption values in all series. Interestingly, the minimum adsorbed thickness in this region may be correlated with PEG length, with short PEGs showing lowest thicknesses. However, these data points are statistically similar because they are within the error. As VP compositions are raised further, a slow and steady increase in adsorbed polymer is observed. As stated
previously, in the high VP extreme, there is a correlation between PEG length and adsorption, with thicker adsorbed polymer layers for shorter PEGs, similar to the peak adsorption. The behavior that these copolymers show is complex and difficult to explain rationally because it is caused by a complex interplay of factors. The shape of the curves is likely dependent on the balance of the hydrophilic and lipophilic groups because the curve shape is only dependent on PEG content and not length. The presence of the positive charges from the VP units will increase protein adsorption, as compared with the slightly negatively charged PEG groups. However, at higher VP compositions, there likely exists structural conformation changes due to changes in the hydrophobicity of the material. This arrangement leads to a globular structure to segregate the hydrophobic regions away from water and into the interior, similar to amphiphilic detergents and lipids. This behavior would lead to the concentrated presence of PEG groups on the surface of the globule, which would shield proteins from the adsorptioninducing positive charges and hydrophobicity of the VP groups, similar to “stealth” particles decorated with PEG.28,29 At even higher VP concentrations, there does not exist enough PEG groups to shield the material, leading to a net increase in protein adsorption, as observed in the experimental results. It is known that protein adsorption on polymer-brush surfaces can be dependent on PEG-length within certain size regimes, and reorganization to bury the hydrophobic VP and expose the hydrophilic PEG may allow for low protein interaction. However, additional experimentation would be needed to support this conjecture.
Discussion The best way to predict biocompatibility, short of expensive in vivo tests, is to complete a broad battery of in vitro biocompatibility assays, some of which have been completed in previous publications.14,16 The results of our protein adsorption assay can be compared with the results of other biocompatibility assays performed on this group of copolymers.14 Figure 3 shows an overlay of the results of previous RBC hemolysis testing on these materials14 on the results of the protein adhesion assay. The RBC hemolysis tests showed a strong correlation between the VP composition and hemolysis, namely, that
Antibacterial and Biocompatible Copolymers
Biomacromolecules, Vol. 11, No. 5, 2010
1289
monomers in the copolymer. (See Figure 2.) Therefore, copolymers that include PEGMA monomers of a high molecular weight or even pure PEG groups should be considered as most favorable for future water-soluble antibacterial copolymer development. We take this comparison a step further in Figure 4, which additionally includes data from the results of in vitro cell viability assays that were performed as a measure of biocompatibility.16 In these experiments, monolayers of Caco 2 human intestinal epithelial cells were grown on assay plates, exposed to our copolymers for 2 h, and then tested to determine the viability of the cells by a cellular staining assay. The results of these in vitro cell viability experiments further corroborate the results we received from the protein adsorption and RBC assays. The reason for the existence of a biocompatible limit for these materials at the 10-15 wt % region may represent a critical balance of hydrophilic groups and micellar aggregation between VP and PEG groups in the copolymer system. Table 2 combines the results of all of these assays in tabular form for the copolymers of VP and PEGMA 1100. We can take the results of these biocompatibility assays and combine them with previously completed measurements of antibacterial behavior as completed by Stratton et al.15 In this article, the authors exposed 106 colonies of pathogenic E. coli 0157:H7 bacteria to the various copolymers and observed the growth and elimination of the bacteria after exposure to determine a minimum bactericidal concentration. The combination of these assays with the biocompatibility assays are displayed in Table 3 and Figure 5. The Table shows clearly the 10-15 wt % VP composition limit for biocompatibility because the biocompatible polymer exposure concentrations decrease rapidly beyond this point. Although the 10-15 wt % area is not a maximum of antibacterial activity, copolymers with 9 wt % VP have shown antibacterial activity that is similar to that of the nonbiocompatible PVP-HB quaternized homopolymer.15 This antibacterial activity combined with substantial biocompatible properties means that our copolymers feature substantially improved selectivity over the homopolymer. (See Table 3.) The selectivity, defined as the ratio of biocompatible polymer concentration (as determined by a biocompatibility assay) over the minimum bactericidal or inhibitory concentration (MBC or MIC, the concentration of polymer that will kill a specified
Figure 3. Protein adsorption with isocyanate surface modification combined with RBC hemolysis data from Allison et al.14 as a function of wt % VP. Lines added to guide the eye.
Figure 4. Combination of protein adsorption results, RBC hemolysis results from Allison et al.,14 and cell viability results from Stratton et al.16 for P(VP-co-PEGMA 1100)-HB copolymers as a function of wt % VP. Lines added to guide the eye.
increased VP compositions led to an abrupt increase in hemolysis between 15 and 20 wt % VP. Interestingly, the point of least adsorption closely corresponds to the step function that occurs in hemolysis data (the composition point where the copolymers begin to lyse RBCs). Therefore, we can surmise that the optimal biocompatibility lies in the range of 10-15 wt % VP. However, our protein adsorption assay has yielded additional information, showing that an overall decrease in protein adsorption for copolymers included longer chains of PEG, even when normalizing for the wt % of PEGMA
Table 2. Cell Viability Concentration and Hemolysis Concentration Values for the Copolymers of VP and PEGMA 1100 As Found Previously14,16a VP comp. mol %
wt %
LC0
LC50
HC50
0% 10% 50% 75% 90%
0% 1% 9% 21% 47%
>1 mg/mL >7 mg/mL 1 mg/mL 10 µg/mL 10 µg/mL
>1 mg/mL >7 mg/mL >10 mg/mL 100 µg/mL 100 µg/mL
10 mg/mL 10 mg/mL 10 mg/mL 1 µg/mL
