Antimicrobial Peptides for Detection of Bacteria in Biosensor Assays

Ourth, D. D.; Lockey, T. D.; Renis, H. E. Biochem. ...... Lasarte AragonésKim E. SapsfordCarl W. Brown IIIClare E. RowlandJoyce C. BregerIgor L. Medi...
0 downloads 0 Views 148KB Size
Anal. Chem. 2005, 77, 6504-6508

Antimicrobial Peptides for Detection of Bacteria in Biosensor Assays Nadezhda V. Kulagina, Michael E. Lassman, Frances S. Ligler, and Chris Rowe Taitt*

Center for Bio/Molecular Science & Engineering, Code 6900, Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, D.C. 20375

Bacteria, plants, and higher and lower animals have evolved an innate immune system as a first line of defense against microbial invasion. Some of these organisms produce antimicrobial peptides (AMPs) as a part of this chemical immune system. AMPs exert their antimicrobial activity by binding to components of the microbe’s surface and disrupting the membrane. The overall goal of this study was to apply the AMP magainin I as a recognition element for Escherichia coli O157:H7 and Salmonella typhimurium detection on an array-based biosensor. We immobilized magainin I on silanized glass slides using biotin-avidin chemistry, as well as through direct covalent attachment. Cy5-labeled, heat-killed cells were used to demonstrate that the immobilized magainin I can bind Salmonella with detection limits similar to analogous antibody-based assays. Detection limits for E. coli were higher than in analogous antibody-based assays, but it is expected that other AMPs may possess higher affinities for this target. The results showed that both specific and nonspecific binding strongly depend on the method used for peptide immobilization. Direct attachment of magainin to the substrate surface not only decreased nonspecific cell binding but also resulted in improved detection limits for both Salmonella and E. coli.

many organisms and serve as the first line of defense against microbial invasion. Highly stable to adverse conditions,1-6 AMPs bind semiselectively to microbial cell surfaces and exert their antimicrobial activity through membrane disruption.6,7 Given their ability to bind to multiple target microbes, we postulate that an array consisting of multiple AMPs would potentially be capable of detecting a higher number of target species than an array with a corresponding number of antibodies. Furthermore, the predicted stability of the AMPs within these arrays is expected to improve operational and logistical constraints over current antibody-based systems. The proposed AMP-based arrays differ from standard peptide arrays in that all components are naturally occurring (or derivatives of molecules produced in nature) and have defined secondary structures,8-12 unlike combinatorially derived libraries.13-16 Most importantly, as many AMPs have overlapping specificities, the pattern of differences in binding affinities can be used for identification. One class of AMPs is comprised of linear peptides that naturally fold to form two helical domains: a strongly basic helical region and a hydrophobic helix separated by a short hinge region. Magainins and other amphipathic R-helical AMPs are unstructured in solution, but become helical upon interaction with target membranes. Because of its stability and ability to bind to multiple bacterial species,7,8,17,18 magainin I (GIGKFLHSAGKFGKAFVGE-

With the notable exception of glucose sensors, the majority of rapid detection systems use antibodies or nucleic acid probes for recognition, identification, and quantification of target analytes. Antibody-based detection techniques are powerful, versatile tools for various molecular and cellular analyses, environmental monitoring, and clinical diagnostics due to the specificity and sensitivity imparted by these biological elements. However, antibody-based platforms suffer from a number of logistical constraints that limit their potential for field use: stability under extreme environmental conditions, lack of batch-to-batch consistency, high costs of monoclonal development, sensitivity to proteases, nonspecific or cross-reactive binding when multiplexed, and the requirement for at least one binding pair for every target detected. In an effort to overcome many of these limitations, we have initiated studies using alternative recognition molecules, antimicrobial peptides (AMPs), in place of antibodies in an optical biosensor. AMPs are part of the host’s innate immune system in

(1) Casteels, P.; Ampe, C.; Jacobs, F.; Vaeck, M.; Tempst, P. EMBO J. 1989, 8, 2387-2391. (2) Ourth, D. D.; Lockey, T. D.; Renis, H. E. Biochem. Biophys. Res. Commun. 1994, 200, 35-44. (3) Jack, R. W.; Tagg, J. R.; Ray, B. Microbiol. Rev. 1995, 59, 171-200. (4) Na, S.; Jia, S.; Chen, X.; Chen, M.; Huan, L. Wei Sheng Wu Xue Bao 2001, 41, 494-498. (5) Porter, E. A.; Weisblum, B.; Gellman, S. H. J. Am. Chem. Soc. 2002, 124, 7324-7330. (6) Hancock, R. E. W.; Chapple, D. S. Antimicrob. Agents Chemother. 1999, 43, 1317-1323. (7) Zasloff, M. Nature 2002, 415, 389-395. (8) Cruciani, R. A.; Barker, J. L.; Zasloff, M.; Chen, H.-C.; Colamonici, O. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 3792-3796. (9) Bechinger, B.; Zasloff, M.; Opella, S. J. Protein Sci. 1993, 2, 2077-2084. (10) Matsuzaki, K.; Mitani, Y.; Akada, K.; Murase, O.; Yoneyama, S.; Zasloff, M.; Miyajima, K. Biochemistry 1998, 37, 15144-15153. (11) Wenk, M. R.; Seelig, J. Biochemistry 1998, 37, 3909-3916. (12) Matsuzaki, K. Biochim. Biophys. Acta 1999, 1462, 1-10. (13) Khan, A. S.; Thompson, R.; Cao, C.; Valdes, J. J. Biotechnol. Lett. 2003, 25, 1671-1675. (14) Petrenko, V. A.; Vodyanoy, V. J. J. Microbiol. Meth. 2003, 53, 253-262. (15) MacBeath, G.; Koehler, A. N.; Schreiber, S. J. Am. Chem. Soc. 1999, 121, 7967-7968. (16) Zhu, H.; Snyder, M. Curr. Opin. Chem. Biol. 2001, 5, 40-45. (17) Zasloff, M. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 5449-5453.

* To whom correspondence should be addressed. [email protected]. Phone: 202-404-4208. Fax: 202-404-8688.

E-mail:

6504 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

10.1021/ac050639r CCC: $30.25

© 2005 American Chemical Society Published on Web 08/30/2005

Figure 1. Immobilization of AMPs on glass slides for detection of bacterial binding. Direct method: Unmodified recognition molecules were attached to thiol-modified substrate through the heterobifunctional cross-linker, GMBS. The maleimide moiety targets thiol functionalities on the substrate, whereas the N-hydroxysuccinimidyl moiety targets primary amines on the magainin. Indirect immobilization: NeutrAvidin was covalently attached to the slide surface using GMBSmediated coupling, as described for direct attachment. Biotin-labeled peptides were subsequently incubated with the NeutrAvidin-coated surface, yielding peptide immobilized by an avidin-biotin bridge.

IMKS) was chosen as a recognition molecule for incorporation into an array-based sensor for detection of pathogenic bacteria, as a first step in creating a multiplexed AMP-based detection system. EXPERIMENTAL SECTION Immobilization of Peptides and Antibodies. The immobilization methodology utilizes sequential incubations with a thiol-terminated silane, the heterobifunctional cross-linker N-(γmaleimidobutyryloxy)succinimide (NHS) ester (GMBS), and a protein or peptide containing one or more primary amines. Briefly, standard soda lime microscope slides (Daigger, Vernon Hills, IL) cleaned with 10% KOH (w/v) in methanol19 were treated for 1 h, under nitrogen, with a 2% solution of (3-mercaptopropyl)trimethoxysilane in toluene. The slides were then rinsed with toluene, dried under nitrogen, and incubated for 30 min in 1 mM GMBS cross-linker (Pierce, Rockford, IL) in absolute ethanol. Following the cross-linker incubation, magainin I and control antibodies were immobilized on the surface using either (1) indirect attachment via avidin-biotin interactions (covalent attachment of an avidin derivative, followed by incubation with biotinylated magainin or antibodies), or (2) direct covalent attachment chemistry (interaction of succinimidyl ester with primary amines on the antibody or magainin) (Figure 1). For avidin-biotin-mediated attachment of capture molecules, the slides were removed from cross-linker, rinsed briefly in water, and then incubated overnight in 33 µg/mL NeutrAvidin (Pierce) in phosphate-buffered saline, pH 7.4 (PBS). The NeutrAvidintreated slides were rinsed in PBS and stored at 4 °C in PBS until patterned with biotinylated capture species. Patterning of capture species onto NeutrAvidin-coated slides was performed by placing a six-channel poly(dimethylsiloxane) (PDMS) patterning template onto the surface of each slide and filling each channel with an appropriate biotinylated capture molecule in PBS.20 Following (18) Zasloff, M.; Martin, B.; Chen, H. C. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 910-913. (19) Cras, J. J.; Rowe-Taitt, C. A.; Nivens, D. A.; Ligler, F. S. Biosens. Bioelectron. 1999, 14, 683-688.

overnight incubation at 4 °C, each channel was emptied and rinsed with PBS. Slides were then blocked for 30 min in 10 mg/mL gelatin, rinsed with Nanopure water, and dried under nitrogen. The following capture species were patterned on NeutrAvidincoated slides: (1) 1 mg/mL (0.4 mM unless otherwise mentioned) custom-synthesized magainin I possessing a C-terminal biotin (SynPep, Dublin, CA, 99% pure); (2) 1 mg/mL (0.4 mM) magainin I (AnaSpec, San Jose, CA) labeled with an amine-specific biotin derivative; and (3) 10 µg/mL (66 nM) biotinylated control antibodies. For direct covalent attachment of recognition molecules, slides were removed from cross-linker, rinsed briefly in water, dried, and then placed in contact with the PDMS patterning templates. Unlabeled antibodies (10 µg/mL in PBS) and magainin I (1 mg/ mL in PBS) were injected into appropriate channels and incubated overnight at 4 °C. The channels were then emptied and rinsed with PBS. Patterned slides were blocked with 10 mg/mL gelatin as above, dried, and stored at 4 °C for up to 2 weeks. Biotinylation of Capture Molecules. Rabbit anti-Escherichia coli O157:H7 (KPL, Gaithersburg, MD), anti-Salmonella typhimurium (Biodesign, Saco, ME), and anti-Listeria (Biodesign) were biotinylated with a 5-fold molar excess of the long-chain derivative of biotin N-hydroxysuccinimidyl ester (EZ-Link NHS-LC-biotin, Pierce) as described in detail elsewhere.11 Unlabeled magainin I was incubated with EZ-Link NHS-LC-biotin at a 1:1 (biotin/ peptide) molar ratio in PBS for 24 h at room temperature; the biotin was first dissolved in a small volume of dimethyl sulfoxide (DMSO) prior to adding to the labeling mix. After 24-h incubation, samples were loaded into dialysis tubing (1000 MWCO) and dialyzed against PBS over 3 days, with six changes of buffer. The biotinylated magainin I was characterized by electrospray mass spectrometry using a QSTAR pulsar I (Applied Biosystems, Foster City, CA) with nanoflow direct infusion. A custom-synthesized magainin possessing a C-terminal biotin was also used in this study. Preparation of Fluorescent Cells. Heat-killed S. typhimurium and E. coli O157:H7 cells (KPL) were rehydrated in PBS as recommended by the manufacturer. Approximately 108 cells/mL were incubated for 30 min in 50 mM sodium borate, pH 8.5, with one packet of Cy5 bisfunctional N-hydroxysuccinimidyl ester (Amersham, Arlington Heights, IL) dissolved in 25 µL of anhydrous DMSO immediately before use. Labeled cells and unincorporated dye were loaded into dialysis tubing (1000 MWCO) and dialyzed overnight at 4 °C against three changes of PBS. Labeled cells were then removed from the bag and stored in the dark at 4 °C until use. Assay Protocol. Patterned slides were placed in contact with PDMS assay templates molded to contain six channels oriented orthogonal to the channels in the patterning templates. The slides with the attached assay templates were connected to a multichannel peristaltic pump at one end of each flow channel via syringe needles (outlet). The opposite end of each flow channel was connected to a 1-mL syringe barrel used as reservoir. To rehydrate the slide, each channel was washed with 1 mL of PBS containing 1 mg/mL bovine serum albumin and 0.05% Tween-20 (PBSTB) at 0.8 mL/min. Samples (0.1 mL of Cy5-labeled cells in PBSTB) (20) Feldstein, M. J.; Golden, J. P.; Rowe, C. A.; MacCraith, B. D.; Ligler, F. S. J. Biomed. Microdevices 1999, 2, 139-153.

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

6505

were then injected into appropriate channels and allowed to incubate for 1 h at room temperature in the dark. Each channel was then washed with 1 mL of PBSTB at 0.3 mL/min. After removing the PDMS templates, the slides were washed with deionized water, dried under nitrogen, and imaged using the array biosensor. Fluorescence Imaging, Data Acquisition, and Analysis. Optical components of the Naval Research Laboratory’s (NRL) array biosensor have been described in detail elsewhere.20,21 Briefly, it consists of a 635-nm, 12-mW diode laser for evanescent excitation of surface-bound fluorophores, a waveguide support, a GRIN lens array, several emission filters, and a Peltier-cooled charge-coupled device (CCD) imaging array. Digital images of the pattern of fluorescent spots were captured by the CCD and saved in Flexible Image Transport System (FTS) format. A custom data analysis software program22 was used to extract data from the FTS file, calculate the mean fluorescence intensity within each array element, and subtract out localized background, resulting in a mean net fluorescence value for each array element. Limits of detection (LODs) were defined as the lowest concentration tested for which the mean net fluorescence values (n > 3) are greater than three standard deviations above both negative control values and localized background values. Dose-response curves were fitted to a three-parameter exponential rise to maximum function using Sigma Plot (version 8.0; SPSS Inc., Chicago, IL). RESULTS AND DISCUSSION As a first step in creating a multiplexed AMP-based detection system, magainin I was used as a capture molecule. For proof of principle, direct detection assays for fluorescent cells were developed using NRL’s array biosensor, an optical detection system capable of performing multitarget analyses in parallel.23-25 Two methods, avidin-biotin coupling and direct covalent attachment of recognition molecules through a cross-linking agent, were used to immobilize magainin I and positive control antibodies onto sensor substrates (Figure 1); both have been successfully used in this22,23 and other rapid biosensor systems.26-30 Preparation of avidin-derivatized surfaces and attachment of recognition species via avidin-biotin interactions has proven to be convenient for preparing highly stable “generic” sensor substrates that can be patterned with biotinylated “capture” molecules at will to suit the user’s need.32 Direct covalent attachment may also provide stable (21) Golden, J. P.; Ligler, F. S. Biosens. Bioelectron. 2002, 17, 719-725. (22) Sapsford, K. E.; Liron, Z.; Shubin, Y. S.; Ligler, F. S. Anal. Chem. 2001, 73, 5518-5524. (23) Rowe, C. A.; Scruggs, S. B.; Feldstein, M. J.; Golden, J. P.; Ligler, F. S. Anal. Chem. 1999, 71, 433-439. (24) Rowe, C. A.; Tender, L. M.; Feldstein, M. J.; Golden, J. P.; Scruggs, S. B.; MacCraith, B. D.; Cras, J. J.; Ligler, F. S. Anal. Chem. 1999, 71, 38463852. (25) Rowe-Taitt, C. A.; Hazzard, J. W.; Hoffman, K. E.; Cras, J. J.; Golden, J. P.; Ligler, F. S. Biosens. Bioelectron. 2000, 15, 579-589. (26) Ligler, F. S.; Anderson, G. P.; Davidson, P. T.; Foch, F. J.; Ives, J. T.; King, K. D.; Page, G.; Stenger, D. A.; Whelan, J. P. Environ. Sci. Technol. 1998, 32, 2461-2466. (27) Schuderer, J.; Akkoyun, A.; Brandenburg, A.; Bilitewski, U.; Wagner, E. Anal. Chem. 2000, 72, 3942-3948. (28) Abel, A. P.; Weller, M. G.; Duveneck, G. L.; Ehrat, M.; Widmer, H. M. Anal. Chem. 1996, 68, 2905-2912. (29) Jung, A.; Stemmler, I.; Brecht, A.; Gauglitz, G. Fresenius J. Anal. Chem. 2001, 371, 128-136. (30) Bierkert, O.; Haake, H.-M.; Schutz, A.; Mack, J.; Brecht, A.; Jung, G.; Gauglitz, G. Anal. Biochem. 2000, 282, 200-208.

6506 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

Figure 2. CCD images of Cy5-labeled cells binding to magainin on sensing arrays using the two methods. (A) Magainin immobilized by indirect (avidin-biotin-mediated) attachment. NeutrAvidin-coated slides were patterned with stripes of biotinylated recognition molecules including C-terminal biotinylated magainin (MB), anti-Salmonella antibody (RS), anti-E. coli antibody (RE), and anti-Listeria antibody (RL), indicated above the panel. Various concentrations of Cy5labeled Salmonella were applied to the slide (indicated to the left of the panel). (B) Magainin immobilized by direct attachment. GMBStreated slides were patterned with stripes of unlabeled magainin I (M), anti-Salmonella antibody (RS), and anti-chicken IgY antibody (RCh), indicated above the image. Various concentrations of E. coli and Salmonella (Sal) were applied to the slide (indicated to the right of the image). PBS served as a negative control.

surfaces with specifically oriented recognition molecules, provided there are unique moieties available for modification, but has been found in some cases to result in lower activities of the immobilized biomolecules.26,31 Moreover, this methodology is not as convenient to use, due to the necessity of choosing the recognition species at the same time that the wet chemistry is performed.32,33 Fluorescently labeled, heat-killed S. typhimurium (Figure 2) and E. coli O157:H7 were both detected on arrays where magainin I was immobilized via its C-terminal biotin (panel A) and unmodified magainin I was immobilized by direct covalent attachment (panel B). In general, slides with directly immobilized magainin (panel B) had lower levels of nonspecific binding, lower backgrounds, and higher signals from fluorescent bacteria bound to the peptide spots than avidin-coated slides patterned with Cterminal biotin-magainin (panel A, P < 0.05). Since the calculations for detection limits were based on both specific and nonspecific binding and variability thereof, both E. coli and S. typhimurium could be detected at significantly lower levels where magainin I had been immobilized directly (Figure 3). Detection limits for E. coli and S. typhimurium on covalently immobilized magainin I were 1.6 × 105 and 6.5 × 104 cell/mL, respectively; detection limits on surfaces where magainin I was immobilized via its C-terminal biotin were at least 4-fold highers6.8 × 105 and 5.6 × 105 cell/mL, respectively. Although sensitivity for E. coli was 1 order of magnitude poorer in magainin-based assays than analogous, optimized antibody-based assays (L. Shriver-Lake, personal communication), the LOD for Salmonella was in the same range as determined previously with the antibody used here as a control.34 The difference in detection limits for the two bacteria may be due to differences in affinity of magainin for lipopolysac(31) Narang, U.; Anderson, G. P.; Ligler, F. S.; Burans, J. Biosens. Bioelectron. 1997, 12, 937-945. (32) Taitt, C. R.; Golden, J. P.; Shubin, Y. S.; Shriver-Lake, L. C.; Sapsford, K. E.; Rasooly, A.; Ligler, F. S. Microb. Ecol. 2004, 47, 175-185. (33) Bhatia, S. K.; Hickman, J. J.; Ligler, F. S. J. Am. Chem. Soc. 1992, 114, 4432. (34) Taitt, C. R.; Shubin, Y. S.; Angel, R.; Ligler, F. S. Appl. Environ. Microb. 2004, 70, 152-158.

Figure 4. Effect of surface density of immobilized magainin on binding of Salmonella. Different concentrations of C-terminal biotinylated magainin were patterned onto NeutrAvidin-coated surfaces (abscissa). Binding of Cy5-labeled Salmonella (107 cell/mL) to different densities of immobilized magainin were measured (ordinate). Fluorescent intensities were normalized with respect to signal from binding of 107 cell/mL Salmonella to anti-Salmonella antibodies. Error bars indicate SEM (n g 3).

Figure 3. Concentration-dependence curves for Cy5-labeled Salmonella (A) and E. coli (B) cells binding to magainin immobilized by direct covalent attachment through GMBS (b), and by avidin-biotin chemistry (9). Fluorescent intensities (ordinate) were normalized with respect to signal from binding of Salmonella (107 cell/mL) or E. coli (107 cell/mL) to appropriate antibodies. Error bars indicate SEM (n g 3).

charide O-antigen components of the outer membranes.35 Other AMPs, such as cecropins, defensins, etc., may have higher affinities for one or both targets; therefore, it is expected that optimized assays using additional AMPs may yield significantly lower LODs. The ability to capture target bacteria was strongly dependent on the density of immobilized magainin on the sensor surface (Figure 4). Bacterial binding to control antibodies saturated at ∼60 nM antibody in the patterning solution, independent of whether immobilization was direct or via avidin-biotin interactions. However, the AMP-based assays required significantly higher concentrations of magainin in solution during the immobilization step before saturation was observed (0.4 mM); this effect was observed with both C-terminal biotinylated magainin I and magainin immobilized covalently. This difference in concentrations required to achieve optimal surface density was not explained by the 80-fold difference in molecular size. Although bacterial binding was demonstrated to magainin I immobilized directly and via a C-terminal biotin, no binding of either labeled species was observed to immobilized magainin biotinylated using an amine-specific biotin (data not shown). As the initial interaction of R-helical AMPs with membranes of target bacteria is postulated to occur through binding of positively charged amino acids on the AMP with negatively charged phospholipids in the bacterial membrane,11,36,37 the lack of binding activity observed in our studies may well have been due to (35) Rana, F. R.; Macias, E. A.; Sultany, C. M.; Modzrakowski, M. C.; Blazyk, J. Biochemistry 1991, 30, 5858-5866. (36) Matsuzaki, K.; Murase, O.; Miyajima, K. Biochemistry 1995, 34, 1255312559.

modification of an amine-containing residue critical to this initial process. This postulate was supported by the ability of magainin with a C-terminal biotin to bind cells, albeit at a lower level than magainin immobilized directly. The potential for modification of an essential amine moiety was further exacerbated by modification of multiple residues by the amine-specific biotin. Despite the 1:1 molar ratio (biotin/magainin I) used in the labeling reaction, peptides with molecular weights corresponding to incorporation of one, two, and three biotins were observed through electrospray mass spectrometry. A similar overlabeling phenomenon has also been observed with the polymyxin family of AMPs,38 with consequent loss of microbial binding activity (data not shown). The majority of biotinylated magainin molecules contained two biotin groups; the C-terminal lysine was not modified. Although we did not evaluate the use of ratios of biotin/magainin I lower than 1:1 in this work, our studies with the polymyxin family showed that multiple biotinylation occurred even at 0.5:1 molar ratio (biotin/polymyxin B). This study showed that characteristics of surface chemistry commonly considered as disadvantages in other systems (e.g., lack of diffusion and steric hindrance)39-41 worked to advantage when immobilizing a small peptide for detection of bacterial species. As the majority of amine moieties targeted by the crosslinker reside in the amino-terminal domain of magainin, the domain presumed responsible for the initial interaction with microbial membranes,11,36,37 a decrease in binding activity of the immobilized species (versus free in solution) was not unexpected; furthermore, as others have shown that net charge on magainin greatly affect its activity,36,37,42-44 modification of these charged (37) Matsuzaki, K.; Sugishita, K.-I.; Harada, M.; Fujii, N.; Miyajima, K. Biochim. Biophys. Acta 1997, 1327, 119-130. (38) Lassman, M. E.; Kulagina, N. V.; Taitt, C. R. Rapid Commun. Mass Spectrom. 2004, 18, 1277-1285. (39) Achtnich, U. R.; Tiefenauer, L. X.; Andres, R. Y. Biosens. Bioelectron. 1992, 7, 279-290. (40) Kodadek, T. Chem. Biol. 2001, 8, 105-115. (41) Boozer, C.; Ladd, J.; Chen, S.; Yu, Q.; Homola, J.; Jiang, S. Anal. Chem. 2004, 76, 6967-6972. (42) Hancock, R. E. W.; Falla, T.; Brown, M. H. Adv. Microb. Physiol. 1995, 37, 135-175. (43) Powell, W. A.; Catranis, C. M.; Maynard, C. A. Mol. Plant-Microbe Interact. 1995, 8, 792-794.

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

6507

residues was also assumed to adversely affect binding activity. Therefore, it was surprising that magainin I immobilized via its C-terminal biotin (with native +4 charge) did not bind bacterial cells as well as magainin immobilized directly using an aminespecific cross-linker. We believe that steric hindrance encountered during the direct immobilization procedure may have prevented modification of residues essential for target binding, as well as prevented modification of multiple sites. Such overlabeling was observed when magainin I was reacted (in solution) with an aminereactive biotin; a similar phenomenon was observed with other amine-rich AMPs (ref 38 and unpublished observations). Moreover, it is possible that the orientation of the directly immobilized magainin is optimal for target binding. We have not yet determined which amino acid residues are directly linked to the surface. Furthermore, the binding activity of magainin may have been improved by the higher surface density when immobilized directly. Binding of labeled cells was observed when high concentrations of magainin were immobilized onto surfaces through direct covalent attachment or via a biotin moiety on the C-terminal amino acid. The 2-fold higher potential packing density of surfaces with magainin immobilized directly versus magainin immobilized via its C-terminal biotin (assuming helical conformation for magainin and 20-30 Å between biotin binding sites on avidin45) may have endowed these surfaces with sufficient avidity to detect bacterial targets at lower concentrations. In addition, given the high concentrations of magainin (∼0.4 mM) required for optimal binding activity, formation of peptide multilayers was probable for both surfaces. However, it is possible that the conformation or orientation of magainin molecules immobilized directly more effectively promoted formation of peptide multilayers. Peptidepeptide interactions have been postulated to be required for strong target binding and microbicidal activity. 8-10,12 To date, there have been limited reports describing use of AMPs for capture and detection of target analytes. James et al.46 described use of polymyxin B as a capture molecule on a fiber(44) Park, Y.; Lee, D. G.; Jang, S. H.; Woo, E. R.; Jeong, H. G.; Choi, C. H.; Hahm, K. S. Biochim. Biophys. Acta 2003, 1645, 172-182. (45) Pugliese, L.; Coda, A.; Malcovati, M.; Bolognesi, M. J. Mol. Biol. 1993, 231, 698-710. (46) James, E. A.; Schmeltzer, K.; Ligler, F. S. Appl. Biochem. Biotechnol. 1996, 60, 189-202. (47) Laitinen, S.; Kangas, J.; Husman, K.; Susitaival, P. Ann. Agric. Environ. Med. 2001, 8, 213-219. (48) Nikaido, H. In Escherichia coli and Salmonella: Cellular and Molecular Biology, 2nd ed.; Neidhart, F. C., Curtiss, R., Eds.; ASM Press: Washington, DC, 1996; Vol. 2, pp 29-47. (49) Gregory, K.; Mello, C. M. Appl. Environ. Microbiol. 2005, 71, 1130-1134.

6508

Analytical Chemistry, Vol. 77, No. 19, October 1, 2005

optic biosensor for detection and quantification of E. coli lipopolysaccharide (LPS) in 5-min competitive assays. The detection limit in these polymyxin B-based assays, ∼10 ng/mL, calculates to approximately the same number of bacteria per milliliter (3 × 105-1.3 × 106 cells/mL) as observed with magainin I in our studies, assuming LPS monomer molecular weight of LPS between 4000 and 20 000, and 1.2 × 106 LPS molecules/cell;47,48 magainin has also been observed to bind to LPS.12 A report has recently been published that describes use of cecropin P1, another amphipathic R-helical AMP, to immobilize E. coli cells onto microtiter plates.49 As the thrust of this latter study was bacterial enrichment, the total analysis time (2.5 h) and detection limits (∼107 cfu/mL) were significantly different from those obtained in the present study. The current report is the first to show sensitive detection of bacterial cells or cell fragments in a rapid biosensor assay using antimicrobial peptides for target recognition. Furthermore, this study demonstrates proof of concept that these simple 70-min AMP-based assays provided detection limits within 1 order of magnitude of antibody-based assays, but with greater stability at ambient and elevated temperatures (data not shown). These assays have recently been extended to include sandwich format assays, utilizing fluorescently labeled antibodies for detection of unlabeled cells bound to AMPs; the sandwich-type format is preferred when modification or labeling of sample components is inconvenient or not desired. Development of these assays is the first step in creating a multiplexed detection platform that uses the semiselective binding of multiple AMPs to detect large numbers of bacterial species. We have preliminary evidence that the directly immobilized magainin shows semiselective binding characteristics; only trace binding of Campylobacter sp. and Bacillus sp. was observed under analogous conditions (data not shown). ACKNOWLEDGMENT N.V.K. and M.E.L. are recipients of National Research Council and American Society of Engineering Education postdoctoral fellowships, respectively. The authors thank the National Institutes of Health Grant EB and the U.S. Department of Defense for their financial support. Received for review April 14, 2005. Accepted August 1, 2005. AC050639R