Nonleaching Antibacterial Glass Surfaces via “Grafting Onto”: The

Langmuir 2008, 24, 6785-6795

6785

Nonleaching Antibacterial Glass Surfaces via “Grafting Onto”: The Effect of the Number of Quaternary Ammonium Groups on Biocidal Activity Jinyu Huang,† Richard R. Koepsel,‡ Hironobu Murata,§ Wei Wu,† Sang Beom Lee,§ Tomasz Kowalewski,† Alan J. Russell,*,| and Krzysztof Matyjaszewski*,† Chemistry Department, Carnegie Mellon UniVersity, Pittsburgh, PennsylVania 15213, and Department of Chemical and Petroleum Engineering, Department of Bioengineering, and Department of Surgery and McGowan Institute for RegeneratiVe Medicine, UniVersity of Pittsburgh, Suite 200, 100 Technology DriVe, Pittsburgh, PennsylVania 15219 ReceiVed June 29, 2007. ReVised Manuscript ReceiVed March 29, 2008 Antimicrobial surfaces were prepared using the “grafting onto” technique. Well-defined block copolymers containing poly(2-(dimethylamino)ethyl methacrylate) and poly(3-(trimethoxysilyl)propyl methacrylate) segments (PDMAEMA/ PTMSPMA) and corresponding random copolymers were prepared via atom transfer radical polymerization (ATRP), followed by covalent attachment to a glass surface through reaction of the trimethoxysilyl groups with surface silanol groups. The density of quaternary ammonium (QA) groups available to bind small molecules in solution increased with polymer solution concentration and immobilization time. For the PDMAEMA97-b-PTMSPMAxdiblock copolymers with a fixed length of PDMAEMA segment (degree of polymerization (DP) ) 97) and varied lengths of PTMSPMA segments, maximal available surface charge was observed when the ratio of DPPDMAEMA to DPPTMSPMA was 5:1. The tertiary amino groups in immobilized PDMAEMA segments were reacted with ethyl bromide to form QA groups. Alternatively, block copolymers with prequaternized PDMAEMA segments were attached to surfaces. Biocidal activity of the surfaces with grafted polymers versus Escherichia coli (E. coli) increased with the density of available QA units on the surface. The number of bacteria killed by the surface increased from 0.06 × 105 units/cm2 to 0.6 × 105 units/cm2, when the density of surface QA increased from 1.0 × 1014 unit/cm2 to 6.0 × 1014 unit/cm2. The killing efficiency of QA on all surfaces was similar with ∼1 × 1010 units of QA needed to kill one bacterium. The AFM analysis indicated that grafting onto the surface resulted in small patches of highly concentrated polymer. These patches appear to increase the killing efficiency as compared to surfaces prepared by grafting onto with the same average polymer density but with a uniform distribution.

Introduction Antimicrobial surfaces can be prepared by immobilizing biocidal polymers or incorporating biocides containing antibiotics, phenols, iodide, quaternary ammonium (QA), or active metal compounds onto surfaces.1–9 Depending on the method of preparation, these antimicrobial surfaces are generally separated into two groups: leaching and nonleaching systems. Although leaching systems are useful, their practical applications are limited because they lose their capacity to kill over time and could contaminate the environment. Therefore, development of nonleaching antimicrobial surfaces, where the antimicrobial agent * To whom correspondence should be addressed. E-mail: km3b@ andrew.cmu.edu (K.M.); [email protected] (A.J.R.). † Carnegie Mellon University. ‡ Department of Chemical and Petroleum Engineering, University of Pittsburgh. § Department of Bioengineering, University of Pittsburgh. | Department of Surgery and McGowan Institute for Regenerative Medicine, University of Pittsburgh.

(1) Golubovich, V. N.; Rabotnova, I. L. Mikrobiologiya 1974, 43, 1115–1117. (2) Nohr, R. S.; Macdonald, J. G. J. Biomater. Sci. 1994, 5, 607–619. (3) Shearer, A. E.; Paik, J. S.; Hoover, D. G.; Haynie, S. L.; Kelley, M. J. Biotechnol. Bioeng. 2000, 67, 141–146. (4) Klueh, U.; Wagner, V.; Kelly, S.; Johnson, A.; Bryers, J. D. Biomed. Mater. Res. 2000, 53, 621–631. (5) Ho, C. H.; Tobis, J.; Sprich, C.; Thomann, R.; Tiller, J. C. AdV. Mater. 2004, 16, 957–961. (6) Melaiye, A.; Sun, Z.; Hindi, K.; Milsted, A.; Ely, D.; Reneker, D. H.; Tessier, C. A.; Youngs, W. J. J. Am. Chem. Soc. 2005, 127, 2285–2291. (7) Francolini, I.; Ruggeri, V.; Martinelli, A.; D’Ilario, L.; Piozzi, A. Macromol. Rapid Commun. 2006, 27, 233–237. (8) Luo, J.; Sun, Y. J. Polym. Sci., Part A: Polym. Chem. 2006, 44, 3588–3600. (9) Gabriel, G. J.; Som, A.; Madkour, A. E.; Eren, T.; Tew, G. N. Mater. Sci. Eng., R: Rep. 2007, R57, 28–64.

is permanently fixed to the surface through covalent bonding, is an attractive alternative strategy. There is now significant literature concerning the covalent attachment of poly(quaternary ammonium) (PQA) compounds to a variety of surfaces including glass,10–12 metal,13 and plastics.14–22 In a number of cases, polymers containing QA ion23,24 were employed as the anti(10) Tiller, J. C.; Liao, C.-J.; Lewis, K.; Klibanov, A. M. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 5981–5985. (11) Kugler, R.; Bouloussa, O.; Rondelez, F. Microbiology 2005, 151, 1341– 1348. (12) Milovic, N. M.; Wang, J.; Lewis, K.; Klibanov, A. M. Biotechnol. Bioeng. 2005, 90, 715–722. (13) Ignatova, M.; Voccia, S.; Gilbert, B.; Markova, N.; Mercuri, P. S.; Galleni, M.; Sciannamea, V.; Lenoir, S.; Cossement, D.; Gouttebaron, R.; Jerome, R.; Jerome, C. Langmuir 2004, 20, 10718–10726. (14) Thome, J.; Hollander, A.; Jaeger, W.; Trick, I.; Oehr, C. Surf. Coat. Technol. 2003, 174-175, 584–587. (15) Cen, L.; Neoh, K. G.; Kang, E. T. Langmuir 2003, 19, 10295–10303. (16) Tiller, J. C.; Lee, S. B.; Lewis, K.; Klibanov, A. M. Biotechnol. Bioeng. 2002, 79, 465–471. (17) Lin, J.; Qiu, S.; Lewis, K.; Klibanov, A. M. Biotechnol. Bioeng. 2003, 83, 168–172. (18) Lin, J.; Murthy, S. K.; Olsen, B. D.; Gleason, K. K.; Klibanov, A. M. Biotechnol. Lett. 2003, 25, 1661–1665. (19) Hu, F. X.; Neoh, K. G.; Cen, L.; Kang, E. T. Biotechnol. Bioeng. 2005, 89, 474–484. (20) Chen, K.-S.; Ku, Y.-A.; Lin, H.-R.; Yan, T.-R.; Sheu, D.-C.; Chen, T.-M. J. Appl. Polym. Sci. 2006, 100, 803–809. (21) Zhang, W.; Chu, P. K.; Ji, J.; Zhang, Y.; Liu, X.; Fu, R. K. Y.; Ha, P. C. T.; Yan, Q. Biomaterials 2005, 27, 44–51. (22) Huang, J.; Murata, H.; Koepsel, R. R.; Russell, A. J.; Matyjaszewski, K. Biomacromolecules 2007, 8, 1396–1399. (23) Ikeda, T.; Hirayama, H.; Yamaguchi, H.; Tazuke, S.; Watanabe, M. Antimicrob. Agents Chemother. 1986, 30, 132–136. (24) Ikeda, T.; Yamaguchi, H.; Tazuke, S. Antimicrob. Agents Chemother. 1984, 26, 139–144.

10.1021/la8003933 CCC: $40.75  2008 American Chemical Society Published on Web 06/03/2008

6786 Langmuir, Vol. 24, No. 13, 2008

bacterial agents. However, many of these QA polymers were prepared through conventional radical polymerization, which could not control molecular weight (MW), polydispersity, or molecular architecture. The lack of control in conventional radical polymerization makes it difficult to elucidate the mechanism through which the surface kills bacteria. In order to understand how PQA interacts with cells, first we have to control chain length, architecture, chain density, and the structure of polymers attached to a surface. Recently developed controlled/living radical polymerization (CRP) techniques enable the preparation of well-defined polymers with precisely controlled structure, MW, and relatively narrow MW distribution.25–31 For instance, well-defined polymer brushes have been prepared by “grafting from” surfaces using atom transfer radical polymerization (ATRP).32–37 Lee et al.38 reported the preparation of nonleaching antimicrobial surfaces (glass and paper surfaces) via the ATRP “grafting from” technique. PQA was covalently attached to the surface by immobilization of an ATRP initiator to the surface, followed by polymerization of 2-(dimethylamino)ethyl methacrylate (DMAEMA) and quaternization with ethyl bromide. While the “grafting from” technique can provide great control of the grafting density, the access to different polymer architectures is limited. In contrast, the “grafting onto” technique allows one to prepare polymers with a variety of architectures and consequently obtain different polymer structures on the surface. Additionally, the procedure of “grafting onto” can be conducted under mild conditions, whereas most of the “grafting from” approaches require complex surface modification and strict reaction conditions. However, the drawback is that the grafting density of the polymer on a surface prepared by “grafting onto” may be less uniform and it is substantially lower than that by “grafting from”.36,37,39–41 In the present work, we report the preparation of antimicrobial surfaces by combining ATRP with the “grafting onto” technique. The preparation of antimicrobial glass surfaces includes three steps: (i) ATRP synthesis of well-defined block copolymers consisting of a poly(DMAEMA) (PDMAEMA) functional segment and a poly(3-(trimethoxysilyl)propyl methacrylate) (PTMSPMA) anchoring segment and also DMAEMA/TMSPMA statistical copolymers; (ii) immobilization of the copolymers onto the glass slide surface via reaction between the anchoring trimethoxysilyl groups in the copolymers and silanol groups from the surface; (iii) conversion of the pendant amino groups in the surface-linked PDMAEMA to QA in the presence of ethyl (25) Patten, T. E.; Matyjaszewski, K. AdV. Mater. 1998, 10, 901–915. (26) Matyjaszewski, K.; Davis, T. P. Handbook of Radical Polymerization; Wiley-Interscience: Hoboken, NJ, 2002. (27) Davis, K. A.; Matyjaszewski, K. AdV. Polym. Sci. 2002, 159, 1–166. (28) Matyjaszewski, K. Prog. Polym. Sci. 2005, 30, 858–875. (29) Matyjaszewski, K.; Spanswick, J. Mater. Today 2005, 8, 26–33. (30) Tsarevsky, N. V.; Matyjaszewski, K. Chem. ReV. 2007, 107, 2270–2299. (31) Braunecker, W. A.; Matyjaszewski, K. Prog. Polym. Sci. 2007, 32, 93– 146. (32) Beers, K. L.; Gaynor, S. G.; Matyjaszewski, K.; Sheiko, S. S.; Moeller, M. Macromolecules 1998, 31, 9413–9415. (33) Pyun, J.; Matyjaszewski, K. Chem. Mater. 2001, 13, 3436–3448. (34) Matyjaszewski, K.; Xia, J. Chem. ReV. 2001, 101, 2921–2990. (35) Matyjaszewski, K.; Miller, P. J.; Shukla, N.; Immaraporn, B.; Gelman, A.; Luokala, B. B.; Siclovan, T. M.; Kickelbick, G.; Vallant, T.; Hoffmann, H.; Pakula, T. Macromolecules 1999, 32, 8716–8724. (36) Pyun, J.; Kowalewski, T.; Matyjaszewski, K. Macromol. Chem. Phys. 2003, 24, 1043–1059. (37) Zhao, B.; Brittain, W. J. Prog. Polym. Sci. 2000, 25, 677–710. (38) Lee, S. B.; Koepsel, R. R.; Morley, S. W.; Matyjaszewski, K.; Sun, Y.; Russell, A. J. Biomacromolecules 2004, 5, 877–882. (39) Saleh, N.; Phenrat, T.; Sirk, K.; Dufour, B.; Ok, J.; Sarbu, T.; Matyjaszewski, K.; Tilton, R. D.; Lowry, G. V. Nano Lett. 2005, 5, 2489–2494. (40) Matyjaszewski, K.; Dong, H.; Jakubowski, W.; Pietrasik, J.; Kusumo, A. Langmuir 2007, 23, 4528–4531. (41) Huang, J.; Cusick, B.; Pietrasik, J.; Wang, L.; Kowalewski, T.; Lin, Q.; Matyjaszewski, K. Langmuir 2007, 8, 1396–1399.

Huang et al.

bromide. Alternatively, copolymers were quaternized in solution and then attached to surfaces. Efficiency of attachment was evaluated by measuring the concentration of available QA groups on the surface. We describe the dependence of measurable grafting density on a variety of parameters including polymer solution concentration, immobilization time, and polymer structure. The effects of polymer structure, chain length, and the density and distribution of QA on biocidal activity against Escherichia coli (E. coli) are also reported.

Experimental Section Materials. Glass slides were purchased from VWR Co. DMAEMA (Aldrich 99%) was dried over CaH2 and distilled before use. 1,2Bis(bromoisobutyryloxy)ethane (BBrIE) was synthesized according to a previously reported procedure.42 CuCl was obtained from Aldrich and purified by stirring in glacial acetic acid overnight, filtering, and washing with dry ethanol. TMSPMA (98%), ethyl 2-bromoisobutyrate (EBriBu) (98%), CuCl2 (98%), 1,1,4,7,10,10-hexamethyltriethylenetetramine (HMTETA) (97%), N,N,N′,N′′,N′′-pentamethyldiethylenetriamine (PMDETA) (99%), 4,4′-dinonyl-2,2′-bipyridine (dNbpy), and fluorescein (Na salt) were purchased from Aldrich and used without further purification. A Live/Dead BacLight bacterial viability kit was obtained from Invitrogen. Instruments and Measurements. Monomer conversion of DMAEMA and TMSPMA was determined on the basis of 1H NMR (Bruker 300 MHz instrument). Molar masses and molar mass distributions were measured on a gel permeation chromatography (GPC) system consisting of a Waters 510 HPLC pump, three Waters Ultrastyragel columns (500, 103, and 105 Å), and a Waters 410 DRI detector, with an N,N-dimethylformamide (DMF) flow rate of 1.0 mL/min and poly(methyl methacrylate) used as the standard. Synthesis of DMAEMA/TMSPMA Random Copolymers. A typical polymerization procedure was as follows: 4 mL (24 mmol) of DMAEMA, 1.1 mL of TMSPMA (4.8 mmol), 19 µL (0.13 mmol) of EBriBu, 83 µL (0.3 mmol) of HMTETA, and 4 mL of dried anisole were added to a 25 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (69 mg, 0.7 mmol) and CuCl2 (9.5 mg, 0.07 mmol) were added under N2. The reaction was carried out at room temperature. Samples were taken to analyze the monomer conversion by 1H NMR and the MW by GPC at different time intervals during the polymerization. The polymerization was stopped by opening the flask to air when a targeted MW was reached. The mixture was then diluted with 20 mL of dried acetone and passed through a small neutral alumina column to remove the catalyst. The final pure product was obtained after precipitating into dried hexanes. Synthesis of PDMAEMA Macroinitiator. A typical polymerization procedure was as follows: 10 mL (60 mmol) of DMAEMA, 22 µL (0.15 mmol) of EBriBu, 80 µL (0.30 mmol) of HMTETA, and 2.5 mL of dried acetone were added to a 25 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (30 mg, 0.3 mmol) and CuCl2 (8 mg, 0.06 mmol) were added under N2. The reaction was carried out at room temperature. Samples were taken to analyze the monomer conversion by 1H NMR and the MW by GPC at different time intervals during the polymerization. The polymerization was stopped by opening the flask to air when a targeted MW was reached. The mixture was then diluted with 30 mL of dried acetone and passed through a small neutral alumina column to remove the catalyst. The final pure product was obtained after precipitating into dried hexanes. Synthesis of PDMAEMA-PTMSPMA Block Copolymer. A typical polymerization procedure was as follows: 0.5 g (0.016 mmol) of PDMAEMA macroinitiator, 1.8 mL (7.4 mmol) of TMSPMA, 7.7 µL (0.037 mmol) of PMDETA, and 3.6 mL of dried anisole were added to a dried 10 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (3.7 mg, 0.037 mmol) was added under N2. The flask was placed in a temperature-controlled oil bath at 60 °C. Samples (42) Karanam, S.; Goossens, H.; Klumperman, B.; Lemstra, P. Macromolecules 2003, 36, 3051–3060.

Antibacterial Surfaces Via “Grafting Onto” Method

Langmuir, Vol. 24, No. 13, 2008 6787

Scheme 1. Schematic Illustration of the Idealized Conformations of the Copolymer after Immobilization on a Glass Surface

were taken to analyze the monomer conversion by 1H NMR and the MW by GPC at different time intervals during the polymerization. The polymerization was stopped by cooling to room temperature and opening the flask to air when a designated MW was reached. The mixture was then diluted with 10 mL dried toluene and passed through a small neutral alumina column. The final pure product was obtained after precipitating into dried hexanes and was stored as solution in dried toluene. Synthesis of PDMAEMA-PTMSPMA-PDMAEMA Triblock Copolymer. A typical polymerization procedure was as follows: 0.3 mL (1.3 mmol) of TMSPMA, 7.2 mg (0.02 mmol) of difunctional initiator (BBriE), 32 mg (0.08 mmol) of dNbpy, and 1 mL of anisole were added to a dried 10 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (4 mg, 0.04 mmol) were added under N2. The flask was placed in a thermostatted oil bath at 60 °C. Samples were taken to analyze the monomer conversion by 1H NMR and the MW by GPC at different time intervals during the polymerization. After 22 h, the monomer conversion reached 90%. Four mL (24 mmol) of deoxygenized DMAEMA and CuCl (4 mg, 0.04 mmol) and 11 µL (0.04 mmol) of HMETEA in 0.1 mL of anisole was then added to the reaction medium. At certain monomer conversion, the polymerization was stopped by cooling to room temperature and opening the flask to air. The mixture was then diluted with 5 mL dried toluene and passed through a small neutral alumina column. The final pure product was obtained after precipitating into dried hexanes and was stored as solution in dried toluene. Synthesis of PDMAEMA Difunctional Macroinitiator. A typical polymerization procedure was as follows: 6.7 mL (44 mmol) of DMAEMA, 0.15 g (0.44 mmol) of BBriE (Scheme 2b), 147 µL (0.53 mmol) of HMTETA, and 2.5 mL of acetone were added to a 25 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (44 mg, 0.44 mmol) and CuCl2 (12 mg, 0.09 mmol) were added under N2. The reaction was carried out at room temperature. Samples were taken to analyze the monomer conversion by 1H NMR spectroscopy and the MW by GPC at different time intervals during the polymerization. The polymerization was stopped by opening the flask to air when a designated MW was reached. The mixture was then diluted with 20 mL of dried acetone and passed through a small neutral alumina column. The final pure product was obtained after precipitating into dried hexanes. Synthesis of PTMSPMA-PDMAEMA-PTMSPMA Triblock Copolymer. In a typical procedure, 0.88 mL (3.6 mmol) of TMSPMA, 0.5 g (0.06 mmol) of difunctional PDMAEMA macroinitiator, 16 µL (0.06 mmol) of PMDETA, and 2 mL of dried toluene were added to a dried 10 mL Schlenk flask. After three freeze-pump-thaw cycles, CuCl (6 mg, 0.06 mmol) was added under N2. The flask was placed in a temperature-controlled oil bath at 60 °C. Samples were taken to analyze the monomer conversion by 1H NMR during the polymerization. After the targeted MW was reached, the polymerization was stopped by cooling to room temperature and opening the flask to air. The mixture was then diluted with 5 mL dried toluene and passed through a small neutral alumina column. The final pure product was obtained after precipitating into dried hexanes and was stored as solution in dried toluene.

Immobilization of Polymers onto Glass Surfaces. The glass slides (2.5 × 7.5 cm2) were activated by immersing them into a “piranha” solution (H2O2/H2SO4 ) 3:7 (v/v)) for 1 h, followed by washing with deionized water, tetrahydrofuran (THF), and acetone. The immobilization of polymer was carried out by immersing the treated glass slides into a toluene solution of copolymers (0.5-20 g/L) at 70 °C. After targeted time, the glass slides were washed with THF, methanol, and acetone. Quaternization of PDMAEMA on the Surface. The polymer grafted slides were immersed in the mixture of ethyl bromide and acetonitrile (1/2, v/v) at 40 °C. After reaction for 12 h, the slides were washed with methanol, water, and acetone. According to 1H NMR, quaternization in solution was quantitative under similar conditions. Immobilization of PQA. A 125 mg portion of copolymer was dissolved in a glass container with 40 mL of dried isopropanol, followed by the addition of 10 mL of ethyl bromide. The quaternization reaction was carried out at 40 °C for 4 h, and then the activated glass slides (2.5 × 7.5 cm2) were immersed in the reaction solution and the temperature was increased to 70 °C. After the immobilization reaction, the glass slides were washed with THF and methanol. Microcontact Printing. The well-defined polymer pattern with a spacing between the parallel polymer domains of about 25 µm was prepared via the contact printing technique using a cross-linked poly(dimethylsiloxane) (PDMS) stamp, on which a polymer film (PDMAEMA97-PTMSPMA40) was pasted from 10 g/L solution in toluene and then dried briefly. These coatings were further cured by heating at 60 °C for 1 h and then quaternized, according to the procedure described above. Any physically adsorbed excess polymer on the surface was readily removed by washing with methanol, whereas chemically attached block copolymer monolayers were intact. Determination of Density of QA and Chain Grafting Density on the Surface. The surface density of QA on the various glass surfaces was measured by a colorimetric method based on fluorescent complexation and UV-vis spectroscopy, as described by Tiller et al.10 QA attached glass slides (1 × 2.5 cm2) were dipped in a 1 wt % solution of fluorescein (Na salt) in distilled water for 10 min, rinsed with distilled water, placed in 3 mL of 0.1 wt % cetyltrimethylammonium chloride in distilled water, and shaken for 20 min to desorb the dye. The absorbance of the resultant aqueous solution was measured at 501 nm after adding 10 vol % of 100 mM aqueous phosphate buffer, pH 8.0. The concentration of fluorescein dye was calculated taking a value of 69 mM-1 cm-1as an extinction coefficient that was independently determined. The polymer chain density on the surface was derived via dividing the density of the QA by the degree of polymerization (DP) of PDMAEMA, assuming that one QA complexes with one dye molecule. Fluorescent Labeling and Microscopy. To test the ability of the surface to kill cells in situ, overnight cultures of E. coli were grown and diluted to ∼108 colony forming units (CFU)/mL. A 100 µL portion of cells was dropped onto the glass slide and covered with a glass coverslip. After 20 min incubation, the slide was rinsed briefly in distilled water. Live/dead staining of E. coli was performed with the BacLight kit (Invitrogen). A solution of the stains was

6788 Langmuir, Vol. 24, No. 13, 2008

Huang et al.

Scheme 2. Synthesis of DMAEMA/TMSPMA Block Copolymers via ATRP: (a) PDMAEMA-PTMSPMA; (b) PDMAEMA-PTMSPMA-PDMAEMA; (c) PTMSPMA-PDMAEMA-PTMSPMA

prepared according to the manufacturers instructions, and 100 µL was dropped onto the slide and covered with a coverslip. After 15 min incubation in the dark, the slide was rinsed with water and allowed to dry. The slide was observed using a Leica inverted microscope (Leica, Wetzlar, Germany) equipped with a multiple fluorescent filter turret. Atomic Force Microscopy (AFM). AFM studies were carried out with the aid of a Nanoscope III (Digital Instruments, Santa Barbara, CA) equipped with phase extender module and vertical

Figure 1. (a) GPC traces in ATRP of DMAEMA, [EBriBu]0/[DMAEMA] 0/[CuCl]0/[CuCl2]0/[HMTETA]0 ) 1:400:2:0.4:2.4, [DMAEMA]0 ) 4.8 M, VM/Vacetone ) 4:1, T ) 25 °C, and (b) chain extension of PDMAEMA with TMSPMA, [PDMAEMA-Cl]0/[TMSPMA]0/[CuCl]0/[PMDETA]0 ) 1:460:2.3:2.3, [PDMAEMA-Cl]0 ) 2.9 mmol/L, VM/Vanisole ) 1:2, T ) 60 °C.

engage J scanner. The images were acquired in tapping mode (light tapping mode near resonance frequency) under ambient conditions with standard silicon cantilevers with a nominal spring constant of 50 N/m and a resonance frequency around 300 kHz. Static Contact Angle Measurements. The water contact angle on the surface was measured using a VCA Optima system where a drop size of 1.50 µL of deionized water was used. Five measurements at different spots were taken with each substrate, and the average of these values was determined. Antimicrobial Activity Determination. Antimicrobial testing was performed using a modified ASTM standard: E2149-01 Standard Test Method for Determining the Antimicrobial ActiVity of Immobilized Antimicrobial Agents Under Dynamic Contact Conditions. 43 A colony of E. coli (K12) grown on a Luria agar (L-agar) plate was used to inoculate 5 mL of Luria broth in a sterile 50 mL conical tube. The culture was incubated at 37 °C while being shaken at 300 rpm (G24 Environmental Incubator Shaker, New Brunswick Scientific) for 18-20 h. The cells were diluted with Sorensen’s Phosphate Buffer (pH 6.8, 0.3 mM KH2PO4) to the desired concentration. The actual number of cells used for a given experiment was determined by standard serial dilution. PQA attached glass slides (1 × 2.5 cm) were immersed in 5 mL of cell suspension (2.9 × 105 in a 50 mL conical tube (Falcon) at 37 °C and 300 rpm. Blank glass slides were used as a control. A certain volume of bacterium suspension was taken after 1 h, diluted appropriately, and plated on L-agar plates. Each viable bacterium developed into a bacterial colony that was identified using magnifying glasses and counted. The biocidal activity of a modified surface was determined by using eq 1. Ncontrol and Nsample correspond to the colonies on the L-agar plates of the control and the sample, respectively, while Fcontrol and Fexperiment represent the dilution factor of the control and experiment, respectively.

Antibacterial Surfaces Via “Grafting Onto” Method

Langmuir, Vol. 24, No. 13, 2008 6789

Figure 2. 1H NMR spectrum of the PDMAEMA97-PTMSPMA5 block copolymer in CDCl3.

N ) FcontrolNcontrol - FsampleNsample

(1)

Results and Discussion Synthesis of the DMAEMA/TMSPMA Copolymers. Block and random copolymers composed of DMAEMA and TMSPMA were prepared via ATRP. PTMSPMA acts as an anchoring segment to immobilize the functional PDMAEMA on the surface. Depending on the polymer architecture, ideally, different polymer conformations (L-, V-, Ω-, and M-type) were expected on the surface after immobilization (Scheme 1). The scheme, of course, does not imply that we can predict the exact structures of polymers on the surface, especially for Ω and M types. Rather, we use this scheme to stress that we synthesized families of polymers that differ in structure (sequence of bioactive and surface active groups), MW, and number of surface reaction group per chain. The immobilization of these polymers on the surface could allow us to investigate the effects of polymer chain properties on biocidal activity. PDMAEMA-PTMSPMA Diblock Copolymer. PDMAEMA-PTMSPMA diblock copolymers were prepared in two steps as shown in Scheme 2a. The PDMAEMA-Cl macroinitiator was prepared in the presence of EBriBu/CuCl/CuCl2/HMTETA.44–48 Halogen exchange was used to increase initiation efficiency49–51 and also to reduce the probability of quaternization of alkyl halide at the chain end with pendant amino groups.52,53 (43) E 2149-01: Standard test method for determining the antimicrobial activity of immobilized antimicrobial agents under dynamic contact conditions In Annual Book of ASTM Standards 2002; ASTM International: West Conshohocken, PA, 2002; Vol. 11.05, pp 1597-1600. (44) Lee, S. B.; Russell, A. J.; Matyjaszewski, K. Biomacromolecules 2003, 4, 1386–1393. (45) Pietrasik, J.; Sumerlin, B. S.; Lee, R. Y.; Matyjaszewski, K. Macromol. Chem. Phys. 2007, 208, 30–36. (46) Zhang, X.; Xia, J.; Matyjaszewski, K. Macromolecules 1998, 31, 5167– 5169. (47) Zhang, X.; Matyjaszewski, K. Macromolecules 1999, 32, 1763–1766. (48) Pintauer, T.; Matyjaszewski, K. Coord, Chem, ReV 2005, 249, 1155– 1184. (49) Matyjaszewski, K.; Shipp, D. A.; Wang, J.-L.; Grimaud, T.; Patten, T. E. Macromolecules 1998, 31, 6836–6840. (50) Matyjaszewski, K.; Shipp, D. A.; McMurtry, G. P.; Gaynor, S. G.; Pakula, T. J. Polym. Sci.,Part A: Polym. Chem. 2000, 38, 2023–2031. (51) Shipp, D. A.; Wang, J.-L.; Matyjaszewski, K. Macromolecules 1998, 31, 8005–8008. (52) Coessens, V.; Matyjaszewski, K. J. Macromol. Sci., Pure Appl. Chem. 1999, A36, 653–666. (53) Coessens, V.; Pintauer, T.; Matyjaszewski, K. Prog. Polym. Sci. 2001, 26, 337–377.

Figure 3. GPC traces in ATRP of TMSPMA ([BBrIE]0/[TMSPMA]0/ [CuCl]0/[dNbpy]0 ) 1:65:2:4, [DMAEMA]0 ) 4.8 M, VM/Vanisole ) 1: 3, T ) 60 °C) and GPC trace of PDMAEMA-PTMSPMA-PDMAEMA ([Cl-PTMSPMA-Cl]/[DMAEMA]0/[CuCl]0b/[HMTETA]0 ) 1:1200: 2:2, T ) 25 °C. bMore CuCl was added for the chain extension reaction.

The polymerization was well-controlled. GPC traces (Figure 1a) show a progressive increase of the MW with a narrow MW distribution during the polymerization. PDMAEMA-Cl was then used as a macroinitiator to synthesize the block copolymer by chain extension with TMSPMA in the presence of CuCl/ PMDETA. The GPC trace of the macroinitiator shifted to the high MW side after chain extension (Figure 1b), indicating the formation of block copolymer. The small shoulder of the GPC trace of the block copolymer at the low MW side could indicate some loss of the initiating sites in the macroinitiator. Figure 2 shows a representative1H NMR spectrum of the resultant PDMAEMA-PTMSPMA diblock copolymer. Polymer composition was determined by comparing the peak areas of the protons derived from two blocks (e.g., -CH2N-; b, 2.6 ppm) in the PDMAEMA segment and -Si(OCH3)3 (i, 3.6 ppm) in the PTMSPMA segment). The DP of PDMAEMA was determined by either GPC or monomer conversion, while DP of PTMSPMA was obtained based on the peak area ratio of two monomer units in 1H NMR and DP of PDMAEMA. However if a small fraction of PDMAEMA macroinitiators could not be extended to form

6790 Langmuir, Vol. 24, No. 13, 2008

Huang et al.

Scheme 3. Grafting of PQA on Glass Surfaces: (a) Immobilization of PDMAEMA/PTMSPMA Copolymers onto the Surface, Followed by Quaternization Using Ethyl Bromide; (b) Quaternization of PDMAEMA/PTMSPMA Copolymers, Followed by Immobilization

block copolymer, the actual length of the anchoring segment in the final block could be underestimated. Synthesis of DMAEMA/TMSPMA Triblock Copolymers. To prepare the PDMAEMA-PTMSPMA-PDMAEMA triblock copolymers, a difunctional PTMSPMA macroinitiator was first synthesized in the presence of difunctional ATRP initiator and CuCl/ dNbpy (Scheme 2b). Polymerization of TMSPMA was well controlled, as demonstrated by a relatively narrow MW distribution and a progressive increase of MW over time (Figure 3). In order to avoid potential cross-linking of PTMSPMA induced by moisture during the purification process of PTMSPMA, chain extension was carried out by adding DMAEMA to the reaction medium of TMSPMA at high monomer conversion (90%) without isolating PTMSPMA macroinitiator. After chain extension, GPC curves exhibited a visible shift from the macroinitiator to the block copolymer, indicating the formation of well-defined block copolymers (Figure 3). A high MW shoulder could be due to a small proportion of radical coupling. The final polymer compositions were determined using 1H NMR. PTMSPMAPDMAEMA-PTMSPMA triblock copolymers were synthesized using a procedure similar to that described in the preparation of PDMAEMA-PTMSPMA-PDMAEMA, although the PDMAEMA difunctional macroinitiator was prepared first and then the chain was extended with TMSPMA to form the block copolymer (Scheme 2c). Synthesis of DMAEMA/TMSPMA Random Copolymers. Copolymers were obtained by ATRP copolymerization of TMSPMA and DMAEMA in the presence of CuCl/HMTETA. Because TMSPMA and DMAEMA had similar reactivity, the compositions of the copolymers should be similar to those in the feeding monomer mixture. 54 Grafting of Polymers onto Surfaces. As shown in Scheme 3, immobilization was achieved by either attachment of PDMAEMA/PTMSPMA copolymers onto the surface, followed by quaternization using ethyl bromide or quaternization of PDMAEMA/PTMSPMA copolymers, followed by immobilization. Immobilization of DMAEMA/TMSPMA Block Copolymers Followed by Quaternization. Scheme 3a represents the process of immobilization of PDMAEMA-PTMSPMA block copolymer and quaternization. (54) Teoh, R. L.; Guice, K. B.; Loo, Y.-L. Macromolecules 2006, 39, 8609– 8615. (55) Murata, H.; Koepsel, R. R.; Matyjaszewski, K.; Russell, A. J. Biomaterials 2007, 28, 4870–4879.

Effect of Polymer Solution Concentration on Available Surface Charge Density. The effect of polymer concentration on grafting density was evaluated by a colorimetric method based on fluorescein complexation and UV-vis spectroscopy. Fluorescein dye binds electrostatically to available surface QA groups. The concentration of a dye remaining on the surface, after rinsing with a large excess of cetyltrimethylammonium chloride, was obtained by UV-vis spectroscopy. The conversion of the dye concentration to surface charge density was made by assuming that one surface QA complexes with one dye molecule. The polymer chain density on the surface was then derived via dividing the density of the QA by the DP of PDMAEMA. Grafting density was found to increase with the solution concentration of polymer. This dependence was observed for both block copolymers and random copolymers (Figure 4). Higher polymer concentration in solution accelerated the immobilization and resulted in higher number of polymer chains attached. A more significant increase was observed at lower polymer concentrations. Effect of Immobilization Time on Grafting Density. AFM images of samples grafted for different amounts of time revealed that the surface was covered with two types of clusters (Figure 5). The presence of these clusters allowed us to test whether surface charge was more biocidal when delivered in a cluster or when evenly spread on a surface (as obtained by a “grafting from” procedure).55 In principle, the clusters could have resulted from three distinct causes or a combination of them: (1) crosslinking of polymer chains in solution (e.g., due to a trace amount of water), followed by immobilization to the surface; (2) “nucleation” mechanism during “grafting onto”, in which the

Figure 4. Plot of grafting density vs polymer solution concentration; immobilization conditions: 70 °C, t ) 24 h in toluene.

Antibacterial Surfaces Via “Grafting Onto” Method

Langmuir, Vol. 24, No. 13, 2008 6791

Figure 5. AFM images of glass surfaces after being immersed in the solution of PDMAEMA97-PTMSPMA60 in toluene (C ) 2.5 g/L) for (a) 2.5 h, (b) 7.0 h, (c) 19 h, (d) 70 h at 70 °C.

Figure 6. Quantitative analysis of AFM images. (a) Patch height distributions, and (b) total area of patches within the image for increasing grafting times of PDMAEMA97-PTMSPMA60 onto glass.

polymer chains preferentially attached to already existing grafted sites as a result of the presence of a relatively large amount of reactive trimethoxysilyl groups in these regions; (3) uneven distribution of the reactive surface groups (this scenario is is less likely given the well-know fact that the concentration of surface hydroxyl groups on glass is on the order of 2-4 groups/nm2).56 Further insight into the grafting mechanism were obtained through quantitative analysis of AFM height maps in which the glass baseline was shifted to zero. The analysis was carried

out using custom AFM image analysis procedures written in MATLAB 7.0 and allowed for the determination of patch height distributions (Figure 6a, using the maximum height of each cluster) and, through volume integration, estimation of the total volume of grafted polymer chains (Figure 6b).

(56) Tsujii, Y.; Ohno, K.; Yamamoto, S.; Goto, A.; Fukuda, T. AdV. Polym. Sci. 2006, 197, 1–45.

6792 Langmuir, Vol. 24, No. 13, 2008

Huang et al.

Figure 7. Proposed explanation of the origin of patch height effects. (a) In newly formed patches, because of the small lateral extent of the patch, the polymer chains partially collapse onto the surface. (b) Upon lateral growth of the patches, polymer chains adopt an extended, brush-like conformation, leading to an increase in patch height. Red “pedestals” indicate PTMSPMA blocks facilitating attachment to the surface.

Figure 8. Plot of grafting density vs immobilization time. Immobilization conditions: toluene, T ) 70 °C. (π: the grafting density was determined by AFM; others by the colorimetric method).

The height distribution obtained for the sample grafted for 2.5 h (Figure 6a, top) had a single mode at ∼13 nm. For all samples grafted for longer times, the distributions were distinctly bimodal, with the second (dominant) mode at ∼25 nm (close to the expected value of contour length of a fully extended PDMAEMA97 block). The distinctly bimodal character of height distributions, and the fact that they did not change significantly with grafting time beyond early grafting, appears to rule out mechanism 1, since progressive cross-linking of copolymer chains in solution should lead to the continuous increase of the mean cluster size with time. The amount of dry polymer on the surface was calculated based on the volumes of the polymer clusters, assuming that the bulk density of the attached polymer is ∼1 g/cm3. The grafting density was then derived based on the amount of polymer, the values of Mn determined by GPC, and the surface area. The total grafting density increased with grafting time and eventually leveled off, leading to a grafting density of 0.04 chains/nm2 (Figure 8). As argued below and shown schematically in Figure 7, this result appears to point to the “nucleation mechanism” (2). In this scenario, the clusters corresponding to the first mode of height distribution would correspond to “primary” grafting sites, where, because of the small lateral extent of the grafted patch, the polymer chains collapsed onto the surface (Figure 7, top). Upon further attachment of polymer chains to these primary patches, steric constraints driven by the tendency to maximize the local grafting density, would then force the polymer chains to adopt the more extended, brush-like conformation, leading to an increase in patch height and the appearance of the ∼25 nm mode in height distributions. At the same time, the “primary” patches would continue to appear on the surface, maintaining the bimodal character of the distributions (Figure 7, bottom). The overall grafting densities were also calculated on the basis of the colorimetric method (Figure 8). The grafting densities determined by AFM were higher than those obtained by the

Figure 9. Plot of grafting density vs DP of PTMSPMA for PDMAEMA-PTMSPMA diblock copolymers: DPDMAEMA ) 97 (b), DPDMAEMA ) 193 (9), immobilization condition: C ) 2.5 g/L of PDMAEMA-PTMSPMA in toluene, t ) 24 h, 70 °C (immobilization followed by quaternization); DPQA ) 97 ([), immobilization condition: C ) 2.5 g/L of PQA-PTMSPMA in isopropanol, t ) 24 h, 70 °C (quaternization followed by immobilization).

colorimetric method. This deviation is plausibly due to the fact that the colorimetric method may underestimate the grafting densities, since it is less probable that every QA was complexed with the dye molecule or that every tertiary amine on the surface was quaternized to form QA. Effect of Polymer Chain Size on Grafting Density. In the first series of experiments with diblock copolymers, the length of the PDMAEMA block was kept constant (DPDMAEMA ) 97), while the length of the anchoring segment was varied. Grafting density increased from 0.015 to 0.045 chain/nm2 as the DP of the anchoring segment increased from 5 to 20. Interestingly, a further increase in the length of the anchoring segment led to a decrease in grafting density (Figure 9). Maximum grafting density (0.045 chain/nm2) was obtained at a 5:1 ratio of DPDMAEMA to DPTMSPMA. For copolymers with a shorter anchoring segment (DP ) 5), the probability of anchoring to surface silanol groups was smaller, resulting in lower immobilization efficiency. Conceivably, when the length of the anchoring segment exceeded DP ) 20, each individual anchoring segment took up a larger surface area, resulting in reduced grafting density. Grafting density was similarly dependent on the length of the anchoring segment, as observed for polymers with higher MW (DPDMAEMA ) 193). Comparison of the grafting density of these two polymer series (DPDMAEMA ) 193 and 97) suggests that increasing the chain length or the MW of the copolymer resulted in lower grafting density. This behavior is often observed in anchoring of polymer chains onto surfaces and has been attributed to a balance between adsorption energy and repulsive forces resulting from entropic and steric effects.57,58 Although grafting density was inversely (57) Siqueira, D. F.; Breiner, U.; Stadler, R.; Stamm, M. Langmuir 1995, 11, 1680–1687.

Antibacterial Surfaces Via “Grafting Onto” Method

Langmuir, Vol. 24, No. 13, 2008 6793

Figure 10. Water contact angle of (a) pristine glass surface, (b) the surface of the glass slide grafted with PDMAEMA97-PTMSPMA20, and (c) the surface of the glass slide after quaternization of PDMAEMA. The water contact angle on the surface was measured using VCA Optima system where a drop size of 1.50 µL of deionized water was used.

proportional to polymer chain size, the total mass of the polymer attached to the surface was fairly constant when the block copolymers had similar compositions, and was independent of the size of the polymer chain. Quaternization of DMAEMA/TMSPMA Block Copolymers Followed by Immobilization. An alternative immobilization approach was investigated, where PDMAEMA97-PTMSPMAx was first quaternized, followed by immobilization (Scheme 3b). According to 1H NMR, quaternization in solution was quantitative. The resulting grafting densities were similar to those with the same concentration of polymer in solution, using the approach described in the previous section (Figure 9). Charged QA groups on the polymer chain could have two opposite effects on the polymer immobilization via intramolecular and intermolecular electrostatic repulsion.59 On one hand, the intramolecular electrostatic repulsion could have led to stretched polymer chains that occupied less space on the surface after immobilization, thereby increasing chain density. On the other hand, because of increased interchain repulsive forces, it should have been more difficult for the free polymer to approach the surface and undergo immobilization, leading to a decrease in the grafting density. Similar correlation of the grafting density with the chain length of the anchoring segment was observed for both techniques. The grafting density increased when the DP of PTMSPMA in PDMAEMA97-PTMSPMAx diblock increased from 5 to 20 and then decreased as the length of the anchoring segment continued to increase (Figure 9). Static Contact Angle Measurements. One surface property, wettability, was measured using a static contact angle technique. Figure 10 shows that a pristine glass slide was relatively hydrophilic and had a water contact angle of 27° due to the hydrophilic silanol groups on the surface. After immobilization of PDMAEMA97-PTMSPMA20, the contact angle increased to 60° because of the disappearance of silanol groups and introduction of a less hydrophilic polymer. After quaternization, the formation of more hydrophilic QA groups led to a decrease of contact angle (45°). These changes in contact angle confirm the presence of the polymer at the surface and that the subsequent copolymer quaternized. Interaction between Surface QA Groups and E. coli We have recently reported the biocidal activity of surface immobilized quaternized PDMAEMA prepared by the “grafting from” technique.55 We concluded that the surface-reactive QA groups destabilize the membrane by exchange with the divalent cations, causing cell death. Both Gram-positive and Gram-negative bacteria are killed by surface-bound cationic polymers10,37 This (58) Huang, H.; Penn, L. S. Macromolecules 2005, 38, 4837–4843. (59) Biesalski, M.; Ruehe, J. Macromolecules 1999, 32, 2309–2316.

observation suggests a common mechanism for both groups. On the basis of our recent quantification of the activity of QA groups against E. coli,54 we decided to use that organism as a test subject in the current study. In order to determine whether “grafted onto” polymer was acting in a manner similar to the previously described “grafted from” system,55 we prepared a patterned surface. Contact printing on a glass slide was used to construct a patterned surface that had 50 µm wide QA stripes spaced 25 µm apart. The pattern was achieved by using a cross-linked PDMS stamp, on which a polymer film (PDMAEMA97-PTMSPMA40) was pasted from a 10 g/L solution in toluene. The pattern was stamped onto the glass slide, and the dried slide was exposed to E. coli. A Live/Dead two-color fluorescence method (Molecular Probes) was used in a bacterial viability assay with a mixture of SYTO 9 green fluorescent nucleic acid stain and the red fluorescent nucleic acid stain, propidium iodide. SYTO 9 is membrane-permeable and labels all bacteria; propidium iodide, on the other hand, only penetrates bacteria with damaged membranes. The red propidium iodide stain cancels the green fluorescence of SYTO 9 by displacing it from complexes with nucleic acids when both dyes are present. Consequently, bacteria with intact membranes show a green color, whereas those with ruptured membranes have a red color. The ability to discriminate live cells from dead cells was combined with the patterned surface to determine whether the QA groups grafted onto the surface were able to kill cells and whether the material could kill cells at a distance by a potential leaching process. For this experiment, 100 µL of cell suspension (∼107 E. coli) was dropped onto the patterned glass slide and covered with a glass coverslip. After 20 min incubation, the cover slide was removed, and the patterned slide was rinsed briefly in distilled water, stained with fluorescent dyes, and viewed with a fluorescence microscope. Figure 11 shows that most of the bacteria remaining on the slide were bound to polymer domains. This bioattractive behavior of PQA can be ascribed to the electrostatic attraction between the negatively charged E. coli surface and the positively charged polymer surface. Some live cells (green) were present on the unmodified glass surface, while almost all cells attached to the PQA-coated surface were dead (red). The polymer stripes are densely covered with dead cells and some live cells. The live cells on the polymer stripes are most likely on top of the dead cells and blocked from exposure to the PQA or not yet killed due to the short duration of the exposure. In either case, the concentration of cells on the stripes and the localization of dead cells to the stripes clearly demonstrates that QA groups were responsible for killing E. coli and that this was accomplished through direct contact. Also, it is clear that

6794 Langmuir, Vol. 24, No. 13, 2008

Figure 11. Fluorescence microscopy image of E. coli on a PQA-patterned glass slide. The image is a result of the superposition of an image with a green band-pass filter showing intact bacteria, and an image taken by using a red band-pass filter showing bacteria with damaged cell membranes. Bar size is 50 µm.

Figure 12. The biocidal activity of the surfaces vs density of QA units on surfaces (2.9 × 105 bacteria in control, surface area: 5 cm2).

the biocidal QA groups did not leach. The overwhelming predominance of live cells between and distal to the polymer domains further confirm that direct contact between the bacteria and QA is a prerequisite for antibacterial action. In addition, the observation of the red color indicates that the surface-deposited QA groups killed bacteria by rupturing their cell membranes. The destruction of bacterial membranes by QA has also been demonstrated using transmission electron microscopy to characterize the structure of the cell membrane of E. coli after exposure to a QA surface.60 Effect of Modification Chemistry on the Efficiency of E. coli Cell Kill Capacity. The effects of the density of QA, polymer chain length, and polymer structure on the biocidal activity were systematically investigated. Biocidal activity was defined by the absolute number of the bacteria killed by antibacterial agents at the measured area of the surface (1 × 1 cm). Surfaces varying in the density of QA were obtained by controlling either immobilization time or polymer solution concentration. A linear correlation between the density of QA and biocidal activity was observed (Figure 12). Surfaces with higher grafting density possessed higher biocidal activity. For instance, for surfaces (60) Lenoir, S.; Pagnoulle, C.; Galleni, M.; Compere, P.; Jerome, R.; Detrembleur, C. Biomacromolecules 2006, 7, 2291–2296.

Huang et al.

Figure 13. The biocidal activity of the surfaces modified with various di- and triblock copolymers with 3.2 × 1014 units/cm2 QA units (2.9 × 105 bacteria in control, total surface area ) 5 cm2).

modified with PDMAEMA97-PTMSPMA20 diblock copolymers, the number of killed bacteria per square centimeter on the surface increased from 0.06 × 105 to >0.56 × 105, while the density of surface QA increased from 1.0 × 1014 units/cm2 to 6.0 × 1014 units/cm2. The number of bacteria (challenge) was 2.9 × 105, and the total surface area was 5 cm2 (two-sided 1 × 2.5 cm2 glass slide). This means that only about 10% of bacteria were killed by the surface with lower density of surface QA (1.0 × 1014 units/cm2) while around 97% of the bacteria were killed by the surface with higher density of QA (6.0 × 1014 units/cm2). We observed a linear correlation between the amount of surface QA and biocidal activity. This implies that although surfaces varying in the amount of QA exhibited different biocidal activity, the killing efficiency of QA on all the surfaces was similar (∼1 × 1010 QA units or 0.015 pmol of QA is needed to kill one bacterium based on the slope in Figure 12). In contrast to surface density, changes in the polymer chain length and structure were not correlated to changes in biocidal activity. As shown in Figure 13, surfaces were immobilized with three copolymers PDMAEMAx-PTMSPMA40,varying in the length of PDMAEMA from DP ) 53 to 197, while the density of QA present on the surface was kept constant (3.2 × 1014 units/cm2). These surfaces exhibited similar biocidal activity (the number of killed bacteria per square centimeter ≈ 0.3 × 105). We also compared the biocidal activity of the surface-immobilized diblock copolymers with those modified with triblocks and random copolymer. Again, when the density of surface QA was kept constant (3.2 × 1014 units/cm2), all surfaces possessed similar biocidal activity. We varied polymer chain lengths to perceive the effect of the chain length on biocidal action. There was no obvious dependence between polymer chain length and biocidal activity. The surfaces showed similar biocidal activity once they contained a similar amount of QA (Figure 13). It is interesting to note that the surface immobilized with short QA chains (DP ) 97, i.e., ∼24 nm) also exhibited biocidal activity. Since the length of this QA was smaller than the thickness of the cell envelop (∼46 nm, cytoplasmic membrane, periplasmic space, peptidoglycan, and outer membrane from inside to outside),61,62 it indicates that QA groups could not kill bacteria by penetrating the cell membrane. This implies that recent demonstrations that (61) Matias, V.; Beveridge, T. Mol. Microbiol. 2005, 56, 240–251. (62) Matias, V. R.; Al-Amoudi, A.; Dubochet, J.; Beveridge, T. J. J. Bacteriol. 2003, 185, 6112–6118.

Antibacterial Surfaces Via “Grafting Onto” Method

Figure 14. Dependence of the biocidal activity of the surfaces prepared by “grafting onto” and by “grafting from”55 on the average density of QA groups.

high-density QA brushes kill cells via ion-exchange rather than penetration might also hold for low-density brushes.11,55,60 Because bacteria, especially Gram-negative bacteria such as E. coli, have a highly negatively charged surface, they behave essentially as large, two-dimensional polyelectrolytes. The outer membrane of E. coli is a lipid/glycolipid structure. The entire structure is stabilized by divalent metal cations such as Mg2+ and Ca2+. Kuglar11 hypothesized that the major function of these divalent ions is to neutralize charges (especially those on the phosphate groups of the lipopolysaccharides that make up the outer membrane). This hypothesis suggests that when a bacterium approaches a cationic solid substrate, positive charges on the substrate can replace the cations in the membrane. This in turn leads to a release of these mobile cations. This cation release finally leads to destabilization of the cell’s outer membrane and an increase in permeability, causing cell death. Surfaces with higher densities of QA possess higher biocidal activities because they cause a faster release of counterions. It is likely, therefore, that surfaces with regions or patches of high polymer density could be even more effective at cell killing than surfaces with equal average grafting density where the polymer was equally spaced across the surface. Further support to the notion of superactive patches is given below. Comparison of Biocidal Activities of Surfaces Prepared by “Grafting Onto” with Those Prepared by “Grafting From” Figure 14 illustrates how concentration of QA groups affects biocidal activity of surfaces prepared by the “grafting onto” process, presented herein, in comparison with the earlier results obtained by the “grafting from” process.54 The first important observation is that the “grafting from” provides much higher density of QA groups than the “grafting onto” (∼1 × 1016 vs

Langmuir, Vol. 24, No. 13, 2008 6795

6 × 1014 QA/cm2). This is in agreement with many earlier observations showing that the concurrent and controlled growth of all chains initiated from the flat surface enables unusually high density of grafting up to ∼0.5 chain/nm2.56 The biocidal activity for “grafting from” levels off at ∼105 bacteria/cm2, which corresponds to essentially all bacteria killed for this challenge. The second observation is that the “grafting onto” provides a lower density of QA, and consequently lower overall biocidal activity. However, the surfaces prepared by “grafting onto” have higher biocidal activity than surfaces prepared by “grafting from” at comparable QA densities. We propose that this intriguing phenomenon can be explained by the uneven distribution of QA groups on surfaces prepared by “grafting onto”. Figures 5 and 6 clearly showed that grafting onto results in the “patches” illustrated schematically in Figure 7. This fact may indicate that biocidal activity depends not only on the average QA density but also on the local density of biocidal centers. Apparently, surfaces with patches of the same overall QA density act more efficiently than surfaces with the same overall amounts of QA but with their uniform distribution. This points to the lowest threshold of concentration of QA needed for biocidal activity and may agree with the minimal local surface charge needed for the exchange of divalent cations required for the disruption of the cell membrane.

Conclusions Antimicrobial glass surfaces have been prepared by the “grafting onto” technique with well-defined DMAEMA/ TMSPMA copolymers prepared by ATRP. The copolymers were immobilized onto surfaces and quaternized to form QA groups. Alternatively, block copolymers with quaternized PDMAEMA were attached to surfaces. The grafting density increased with polymer concentration and immobilization time. For diblock copolymers, the highest grafting density was obtained for PTMSPMA with DP ∼ 20 and the ratio of DP of PDMAEMA to DP of PTMSPMA of ca. 5. “Grafting onto” resulted in the nonuniform distribution of QA on the surface. Biocidal activity increased with concentration of QA, but was not strongly affected by polymer architecture. At the same density of QA, the biocidal activity of surfaces prepared by “grafting onto” was higher than for surfaces prepared by “grafting from”. This could be explained by the nonuniform coverage of the former surfaces and biocidal activity affected by localized patches with high concentration of QA. Acknowledgment. Financial support from the industrial members of the CRP Consortium at Carnegie Mellon University and DARPA (HR0011-05-C-0002) are gratefully acknowledged. LA8003933