key: cord-0000493-rq08bhu9 authors: Miller, Scott A.; Hiatt, Leslie A.; Keil, Robert G.; Wright, David W.; Cliffel, David E. title: Multifunctional nanoparticles as simulants for a gravimetric immunoassay date: 2010-11-26 journal: Anal Bioanal Chem DOI: 10.1007/s00216-010-4419-8 sha: 53b9ed3c0545c27d946916518be816d6e2da90c5 doc_id: 493 cord_uid: rq08bhu9 Immunoassays are important tools for the rapid detection and identification of pathogens, both clinically and in the research laboratory. An immunoassay with the potential for the detection of influenza was developed and tested using hemagglutinin (HA), a commonly studied glycoprotein found on the surface of influenza virions. Gold nanoparticles were synthesized, which present multiple peptide epitopes, including the HA epitope, in order to increase the gravimetric response achieved with the use of a QCM immunosensor for influenza. Specifically, epitopes associated with HA and FLAG peptides were affixed to gold nanoparticles by a six-mer PEG spacer between the epitope and the terminal cysteine. The PEG spacer was shown to enhance the probability for interaction with antibodies by increasing the distance the epitope extends from the gold surface. These nanoparticles were characterized using thermogravimetric analysis, transmission electron microscopy, matrix-assisted laser desorption/ionization-time of flight, and (1)H nuclear magnetic resonance analysis. Anti-FLAG and anti-HA antibodies were adhered to the surface of a QCM, and the response of each antibody upon exposure to HA, FLAG, and dual functionalized nanoparticles was compared with binding of Au–tiopronin nanoparticles and H5 HA proteins from influenza virus (H5N1). Results demonstrate that the immunoassay was capable of differentiating between nanoparticles presenting orthogonal epitopes in real-time with minimal nonspecific binding. The detection of H5 HA protein demonstrates the logical extension of using these nanoparticle mimics as a safe positive control in the detection of influenza, making this a vital step in improving influenza detection methodology. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1007/s00216-010-4419-8) contains supplementary material, which is available to authorized users. Viruses are the smallest form of life on earth with the ability to replicate and spread within living cells [1] . As they pass from cell to cell, they adapt to evade host immunity and spread disease, creating some of the worst pandemics in history [2] . Improving diagnostics for viruses, such as influenza, would help slow the spread of infection in the event of an emerging virus. It is known that even with reassortment, a common viral defense mechanism, the majority of anti-hemagglutinin (anti-HA) antibodies recognize a specific nine amino acid sequence within the epitope, AYDPVDYPY, which has been the focus of many assays to improve the detection of influenza [3] . Using this immunodominant sequence for the HA epitope, the influenza epitope can be mimicked with a functionalized nanoparticle yielding a comparable affinity to the linear peptide [4] . Another virus, Ebola, has also been effectively mimicked with a monolayer-protected cluster (MPC) through functionalization of the MPC with the antigenic determinant of the Ebola glycoprotein [5] . Integrating biology and materials chemistry using biomimicry in this way has allowed materials chemistry the opportunity to improve current diagnostic treatments, techniques, and limits of detection, but further improvements are still yet to be made [6] . The utilization of immuno-molecular recognition in the assembly of nanoscale sensors has applications in medical diagnosis, treatment, and the understanding of diseases [7] . Enzyme-linked immunosorbent assay (ELISA) is widely utilized in clinics and hospitals as an initial screening for several infectious diseases. While ELISA can be effectively employed in laboratory settings for common infectious agents, several obstacles inhibit the adaptation of this standard clinical assay to portable or select agent detection schemes. Unfortunately, pathogenic agent detection requires calibration with irradiated or otherwise attenuated samples of the organism. This requirement limits the widespread use of immunosensor diagnostics because of the scarcity of these agents and the logistic difficulty in safe transportation to remote locations. Recent cases like the development of meningitis and tularemia infections in researchers who were working with the causative agents of these diseases alert scientists to the hazards accompanying work with live calibrants [8] . The possibility of exposure and high cost are enough to warrant investigation into a safer positive control for these disease detection assays and devices. Without a positive control, the operational status of the sensor cannot be determined. Thus, the development of a nanoparticle mimic would be a safer alternative to current methodology and could be extended to address the needs of other assays that incorporate well-defined epitopes. Many traditional clinical assays lack the ability for electronic adaptation and timely results. The need exists for rapid, sensitive, and inexpensive methods that could be utilized in clinical settings [9] . Therefore, the rapid realtime quantification of a QCM has been combined with the selectivity of monoclonal antibodies to create an immunosensor for evaluation of the multifunctional nanoparticles. The use of a QCM for immunological detection has been demonstrated previously, where it has been shown to detect Staphylococcus epidermidis in clinical samples using nanoparticle amplification [10] , SARS virus in sputum samples [11] , plant pathogens [12] , an antigenic mimic of Ebola [5] , and influenza A and B from nasal washes [13] . In QCM assays, nanoparticles have enabled simultaneous parallel detection and amplification for gravimetry, thus lowering the limit of detection [6, [14] [15] [16] [17] . A QCM-based sensor with a nanoparticle control would allow for the simultaneous rapid and accurate analysis of multiple viral or biological hazards, without posing a safety or health threat. The work reported here builds on this previous work and addresses the specificity of detection of antibody-antigen binding at the QCM using polyepitope-functionalized gold nanoparticles. Gold nanoparticles have favorable characteristics for their use as the basis for multifunctional microorganism simulants [18] . Nanoparticle size, shape, and capacity for surface modifications, based on chemical characteristics and environments, make the use of nanoparticles advantageous for detection, discovery, and diagnosis [7, 19] . Previous studies have shown that nanoparticles are capable of accepting a wide array of functional molecules via the Au-thiol bonds at the interface of the ligand and particle [20] . Since the physical and chemical properties of nanoparticles are dependent upon size [6] , the gold nanoparticles can be customized to simulate the variability in pathogen size. MPCs in this work had an average diameter of 2.6± 0.6 nm. Influenza virions vary in size, normally around 100 nm in diameter, but smaller nanoparticles were chosen in this study since they have a larger surface area to volume ratio and therefore increase the ratio of possible antigen presentation to gold core [21] . These nanoparticles were polyfunctionalized to increase the presentation of the nanoparticle epitope to the antibody. Previous studies have found multivalent ligand attachment to gold nanoparticles enhances the affinity measured in binding studies [22, 23] . The selectivity of antibody sensors for functionalized MPCs has been examined using two orthogonal epitopes: FLAG and HA. The FLAG epitope is a biological peptide sequence commonly used for identification of proteins in biological samples [24] . HA-and FLAG-functionalized MPCs can be used as a synthetic simulant and negative control for the HA epitope of the influenza virus [25, 26] . The binding of these nanoparticles was also compared with binding of negative-control Au-tiopronin nanoparticles and an authentic sample of H5 HA proteins from influenza virus (H5N1). The simulants work as safe controls whose application could be extended to address various pathological threats, in which using attenuated controls is problematic. Chemicals Gold shot was purchased from precious metal vendors (Canadian Maple Leaf, 99.99%) and was initially converted to HAuCl 4 ·3 H 2 O by boiling Au 0 in HCl/HNO 3 solution [27] . N-(2-Mercaptopropionyl)-glycine (tiopronin, reagent grade), bovine serum albumin (BSA, fraction V, 96%), and sodium phosphate (monobasic, reagent grade) were purchased from Sigma. Other chemicals were obtained as follows: protein G was purchased from Southern Biotech, Fmoc-dPEG-COOH from Quanta Biodesign, anti-FLAG mouse monoclonal antibody from Stratagene, and anti-HA mouse monoclonal antibody from the Vanderbilt Molecular Recognition Core facility. The following reagent was obtained through the NIH Biodefense and Emerging Infections Research Resources Repository, NIAID, NIH: H5 HA protein from influenza virus, A/Hong Kong/156/97 (H5N1), recombinant from baculovirus, and NR-652 (56 kDa) [28] . Analytical grade solvents for nuclear magnetic resonance (NMR) were obtained from Cambridge Isotope Laboratories, and water was purified using a Modulab Water Systems unit (∼18 MΩ/cm). Buffers were prepared according to standard laboratory procedure. Other chemicals were reagent grade and used as received. MPC synthesis and characterization Gold-tiopronin-protected MPCs were synthesized as previously described [29] [30] [31] . Briefly, tiopronin-protected gold nanoparticles were synthesized by dissolving tiopronin and HAuCl 4 ·3H 2 O (3:1) into a MeOH/acetic acid solution (6:1). The reaction was stirred for 30 min and then cooled in an ice bath. NaBH 4 was then added in 10× molar excess of gold. Reaction product was stirred for 1 h. The solvent was removed under vacuum, and the pH was lowered to 1 with concentrated HCl. Reaction product purification by dialysis with cellulose ester membranes (Spectra/Por CE, MWCO= 10,000) removed any excess tiopronin. Average particle diameter was determined by transmission electron microscopy (TEM). TEM images were taken on a Phillips CM20 instrument after applying aqueous MPC samples to Formvar-coated 200-mesh copper grids (Ted Pella). The microscope operated at 200 keV with magnification in the range ×150-750,000. Thermogravimetric analysis (TGA) was performed with a TGA 1000 (Instrument Specialists, Inc.) to calculate Au-tiopronin stoichiometry. From TEM and TGA data, the nanoparticle size and Au-tiopronin stoichiometry were obtained. MPCs used in the following experiments had an average diameter of 2.6±0.6 nm and a base composition of Au 544 Tiop 204 (140.4 kDa), calculated from TEM and TGA data (at 650°C), respectively (Electronic Supplementary Material Figs. S1, S2, and S4) [32] [33] [34] . Peptide synthesis and characterization The FLAG and HA epitopes were synthesized with standard 9fluorenylmethoxycarbonyl (Fmoc protocols on a solid resin support) [35, 36] . Epitope sequences were modified with a linker region comprised of a discrete polyethylene glycol (PEG, purchased as Fmoc-dPEG™ from Quanta Biodesign) and a C-terminal cysteine. PEG was added to attenuate nonspecific binding, while the C-terminal cysteine provided a thiol linkage to the gold surface. PEG formed the link between the cysteine and the rest of the peptide, HA-PEG-C for example. After initial MPC, studies performed without the PEG linker were found to have no binding; all future studies incorporated this linkage. The notations HA-Au, HA-FLAG-Au, and FLAG-Au are assumed to include the incorporation of this linkage. Epitope antigens HA (AYDPVDYPY-(PEG) 6 -C, 1562.6 Da) and FLAG (KDDDDKYD-(PEG) 6 -C, 1451.5 Da) were synthesized on an Apex 396 (Advanced Chemtech) equipped with a 96-well reaction block capable of vortex mixing. 2-Chlorotrityl resin was swollen in dichloromethane (Fisher) prior to synthesis [37] . Fmoc amino acids (Synpep) and Fmoc-dPEG 6 acid (Quanta Biodesign) were coupled using O-benzotriazole-N,N,N′,N′-tetramethyl-uronium-hexafluoro-phosphate (HBTU, 5 eq with respect to resin, Synpep), 1-hydroxybenzatriazole (Hobt, 5 eq, Synpep), and N,N-diisopropylethylamine (DIEA, 10 eq, Advanced Chemtech) in N,N,-dimethylformamide (DMF, Fisher). Peptides were cleaved with 90% (v/v) trifluoroacetic acid, 5% anisole (Sigma), 3% thioanisole (Sigma), and 2% ethanedithiol (Sigma). Final purification was performed on a Waters C18 semi-prep RP HPLC column using a water (0.05% trifluoroacetic acid and acetonitrile) gradient. Peptide identity was confirmed via matrix-assisted laser desorption/ionization-time of flight analysis. Place exchange Place exchange reactions between tiopronin-MPCs and thiolate-containing peptides, HA and FLAG, followed a method similar to Hostetler et al. [18, 38, 39] and assumes that epitope-tiopronin S N 2 place exchange is one for one. While it has been shown by 1 H NMR that the interchange of ligands is 1:1, the final ratio of the original ligand to exchanged ligand depends on ligand length, concentration of entering and exiting ligands, as well as the placement of the ligand (vertice, edge, or terrace) on the surface of the cluster [18, 38] . It should also be noted that the reported ratio will be an average exchanged ratio calculated from 1 H NMR, as the dispersity of MPC size will cause variations in the exact number of exchanged ligands on each cluster [18] . With the X-ray diffraction identification [40] of staple motifs, which are composed of one Au(I) and two thiolates, as part of the capping structure on gold-thiolate nanoparticles, place exchange must insert the new thiolate into the existing staples at the same time as the previous thiolate is lost in the S N 2 mechanism. Tiopronin MPCs were co-dissolved in DI water with free thiolated epitope (1:10). If necessary to aid the solvation of the epitopes into water, the epitopes were first dissolved into ethanol. The place exchange reaction took only one night for the exchange of a single epitope, while requiring 3 days for the exchange of two epitopes simultaneously. Solutions were dialyzed at room temperature for approximately 3 days as previously described and then dried under air [29] . The epitope composition of the MPCs was determined by TGA and 1 H NMR analysis (Electronic Supplementary Material Figs. S2 and S3). Nuclear magnetic resonance 1 H NMR experiments were run at 300 MHz on a Bruker DPX-300 instrument with 5 s relaxation times. Samples were dissolved in D 2 O. The extent of place substitution of tiopronin by epitope on gold nanoparticles was determined by 1 H NMR, using methods previously described [18, 31] . Enzyme-linked immunosorbent assay Initial tests to determine whether the functionalized nanoparticles were suitable as both a viral simulant and a synthetic calibrant were performed using ELISAs. Positive activity was established by adhering gold nanoparticles (250-500 ng/per well) with a variety of functionalities (Au-Tiop, HA-Au, FLAG-Au, and HA-FLAG-Au) to a 96-well Immulon 2HB plate. Buffer and free peptides (25 μg/well) were plated and tested in a similar fashion. Blocking was achieved with BSA (1 mg/mL), followed by exposure to either anti-FLAG or anti-HA primary antibodies (at recommended dilutions, monoclonal mouse). Horseradish peroxidase linked antimouse antibodies (1:5000 dilution) were then added, followed by exposure to tetramethylbenzidine (TMB, Sigma) substrate solution. The reaction was halted with 2 M H 2 SO 4 , and the plates were read at 450 nm (BioTek Synergy HT plate reader). Immunosensor assembly The immunosensor was assembled on a 5-MHz AT-cut quartz crystal. Before assembly, the gold electrode was triple cleaned in piranha, and then received a final ethanol or acetone rinse and dried in a stream of N 2 (grams). The quartz crystal was then mounted in a flow cell holder, rinsed with phosphate buffer (PB, 50 mM phosphate, pH 7.2), and brought to resonant frequency at room temperature. For the work described herein, a Stanford Research Systems quartz crystal microbalance model 200, which measures both frequency and resistance, was used. Solutions were passed through the cell at a flow rate of 28 μL/min controlled by a Masterflex peristaltic pump. The sequence for biosensor assembly was protein G (20 μg/mL, 10 min), BSA (1 mg/mL, 5 min), antibody (20 μg/mL, 16 min), and viral simulant (500 μg/mL, 11 min) all in PB. The nanoparticles were filtered prior to use with a 0.22-μm filter from Millipore. The crystal was washed for a minimum of 10 min with PB between each step. Deposition of protein G allowed for the immobilization of the Fc region of anti-HA and anti-FLAG antibodies. The binding of non-functionalized Autiop nanoparticles (500 μg/mL, 11 min) with anti-HA antibodies was used as a negative control and as a test for nonspecific binding. Positive control was established utilizing anti-HA's recognition of recombinant H5 HA protein (5 μg/mL, 10 min). The adsorbance of each of these molecules resulted in changes in both frequency and resistance. The bound analyte mass is proportional to changes in the oscillation frequency of the quartz crystal as described by the Sauerbrey equation (Eq. 1), where Δf is the change in frequency, C f is the known sensitivity factor of a 5-MHz crystal (56.6 Hz cm 2 μg −1 ), and Δm is the change in mass [41] : In addition to mass, the frequency of the crystal is also dependent on the density and viscosity of the contact medium, a consequence of the solution's resistance to crystal oscillation [42] . The crystal frequency and resistance were recorded during the QCM experiments to allow for corrections to be made for solution resistance, following the work of Kanazawa and Martin [43] [44] [45] , which modifies the standard Sauerbrey equation to: Frequency change for solution resistance is nominally −2.464 Hz/Ω given an active crystal surface area equal to 0.40 cm 2 per sucrose calibrations similar to work reported previously [32] . The resulting change in mass is therefore a combination of changes in resistance and frequency, while taking into account solution resistance. The QCM results corroborated results from the ELISAs. In the development of a synthetic positive control for this influenza immunoassay, functionalized gold MPCs were synthesized and characterized. TEM analysis of the tiopronin-gold nanoparticles (Au-tiop) resulted in an average cluster size of 2.6±0.6 nm, with a range from 1.5 to 4 nm (Electronic Supplementary Material Fig. S1 ). The TGA tiopronin MPC mass loss equaled 37.9% and yielded a 2.67 Au:Tiop stoichiometric ratio for the nanoparticles assuming the loss of the Au(I)-thiolate staples (Electronic Supplementary Material Figs. S2 and S4 ). The assumption of the loss of the Au(I)-thiolate staples is justified by the observation of gold-thiolate compounds upon thermal decomposition in a mass spectrometer [46] . The average number of gold atoms calculated per cluster equaled 544 atoms. Combining the TEM and TGA yields a calculated empirical formula Au 544 Tiop 204 for the 2.6-nm diameter particle. Following place exchange of epitope(s) for tiopronin, the epitope loading was determined through 1 H NMR (Electronic Supplementary Material Fig. S3 ). The epitopes HA and FLAG each contain an exclusive amino acid whose proton signal occurs at a unique location in the 1 H NMR spectrum. Specifically, valine (V) is unique to the FLAG epitope and lysine (K) to the HA epitope. Tyrosine (Y) is common to both epitopes, but occurs in different stoichiometric ratios. The integrated resonance for each functionalized nanoparticle can be compared with that for nanoparticles protected solely with tiopronin to determine the average epitope stoichiometry of the nanoparticles. Peaks used for quantification consisted of signals at 1.2 ppm (V, 2 CH 3 ), 1.45 ppm (tiopronin, CH 3 ), 2.90 ppm (K, 2 ε-CH 2 ), 6.75 ppm (Y, δ-CH), and 7.05 ppm (Y, ε-CH Material Fig. S4) . The ELISA results confirmed specificity of the functionalized gold nanoparticles to the anti-HA and anti-FLAG antibodies (Fig. 1 ). Evidence of cross reactivity between the gold nanoparticles presenting both peptides and both antibodies corroborates with QCM data. HA-FLAG-Au and FLAG-Au were detected with the anti-FLAG antibodies, while HA peptides and HA-Au were not. Correspondingly, HA peptides, HA-Au, and HA-FLAG-Au were recognized by anti-HA, while FLAG-Au was not. The LOD was calculated to be 0.146 OD using the average absorbance of wells incorporating buffer only in the initial step, which served as a negative control. The gold particles displaying only tiopronin (Au-tiopronin) appeared to nonspecifically bind with anti-HA. Nonspecific binding could be reduced in ELISAs with the use of higher concentrations of BSA or with the use of more stringent rinsing protocols, such as the use of Tween 20, a nonionic surfactant, between steps, which might also remove bound particles and lower the sensitivity of these assays [47] . MPCs functionalized without the PEG linker between the epitope and the cysteine linkage showed no binding with their respective antibodies in either ELISA or QCM experiments, but once the PEG link was incorporated into the mimic's a b designs, the antibodies were able to bind with the epitopefunctionalized nanoparticles. Calculations using Spartan computational software show the spatial projection of tiopronin molecules to be about 7 Å from the gold surface, and the peptides, due to the PEG addition, extend 20 Å, from the gold surface [48] . The addition of the six-mer PEG link between the epitope and the terminal cysteine enhanced the probability for cluster-epitope interaction with antibody by distancing the epitope from the gold surface. This enhanced binding was demonstrated with both ELISA and QCM studies. Several observations can be made by inspection of the QCM responses (Fig. 2 , recorded data and schematic of adsorption) during sensor formation. Both antibodies are found to bind well to protein G. Observed spikes during assembly occur when reagent flowing through the peristaltic pump is momentarily interrupted for reactant exchange. A time delay (1-2 min) between spike and sensor response was due to the requisite transport time for the analyte to move through the peristaltic pump to the QCM sensor. binding of the nanoparticles and the HA protein to the HAantibody (Fig. 3a) and to the FLAG-antibody (Fig. 3b) . Detection of the HA protein confirms the ability of this assay to detect HA specific to the H5 HA subtype, and thus the use of this HA-functionalized nanoparticle as a viral simulant. This change in mass is based on the relationship that both frequency and resistance changes have on mass load. Originally, the Sauerbrey equation was applied in air or in a vacuum, and the resulting equation was only valid for thin solid layers deposited on the resonator [49]. Since then, extensive work has been done to establish the use of the QCM to probe interactions in a liquid environment, involving suitable oscillator circuits, fluid modeling in viscous and lossy fluids, as well as determination of the relationship between motional resistance and mass load [12, 45, 50] . The sensing layers utilized in this study should yield at most a layer 30-nm thick, using liberal estimates, where the Sauerbrey equation can be applicable to thin films less than 250 nm [51] [52] [53] [54] . Previous work, which uses a sucrose calibration, modifies the Sauerbrey equation to account for the changes that do occur in part from motional resistance, and therefore allows the ideology behind the Sauerbrey equation to apply in environments where energy is dissipated in the non-rigid liquid environment [30] . The nanoparticles increase the resistance of the crystals, and thus the rigidity, as well as decreased the frequency, resulting in a detectable increase of mass adsorbed on the surface of the crystal. Binding to the HA-antibody (positive Δm) occurs if the gold nanoparticle is functionalized with either HA or both HA and FLAG epitopes, but does not occur if the cluster is only FLAG functionalized (Fig. 3a) . The nonspecific binding observed with ELISA of Autiopronin to anti-HA was not observed with the QCM. QCM naturally prohibits nonspecific binding through the acceleration of adhered particles. This acceleration is generated by the oscillation of the quartz and can help remove weakly bound or nonspecifically bound molecules [55, 56] . Similar binding of anti-FLAG to FLAG-Au and HA-Au-FLAG but not HA-Au was measured with the QCM (Fig. 3b) . Thus, orthogonal and normalized binding at the QCM is observed consistent with immunological results obtained from ELISA, but is observable in a much shorter time interval (10 min compared to the 4 h it took to complete the ELISAs) when using the developed immunoassay and the QCM. The demonstrated binding between bi-functionalized nanoparticles and their respective antibodies makes evident the practicality of their use as a simulant for microorganisms, while lacking the difficulties associated with use of an attenuated or killed pathogen. When a change in Fig. 4 Calibration of HA protein binding to anti-HA. a The average change in mass of H5N1 HA protein binding was determined at varying concentrations. Briefly, (black line) 40 μg/mL (n=2), (red line) 20 μg/mL (n=5), (blue line) 10 μg/mL (n=3), (teal line) 5 μg/mL (n=3), and (pink line) 1 μg/mL (n=4). Also shown for comparison is the binding of HA-Au (green line) and HA-FLAG-Au (orange line). b A linear representation of the Langmuir isotherm produced by this average binding is shown with a linear relationship of y ¼ 0:035 AE 0:004 ð Þ x þ 4:90 AE 15:5 ð Þ Â 10 À10 and R 2 =0.96. c Calibration curve for an assay time of 1.5 min where the black squares are HA proteins, and the red and blue dots are HA-FLAG-Au and HA-FLAG, respectively b mass is observed upon introduction of the functionalized MPC to the immunosensor, the immunological response is shown to be above the limit of detection (∼3 ng, calculated by three times the average noise). Results show the average change in mass for protein G to be 128±48 ng, for antibodies to be 276±75 ng, and for the nanoparticle simulants to be 37±8 ng (Table 1 ). While the time was held constant for each step, the binding varied slightly, which could be due to slight variations in surface roughness and the surface coverage of prior adsorption steps. Even with a standard deviation of 75 ng for antibody adsorption, a deviation of only 8 ng was measured for the final detection step. The binding of the nanoparticles demonstrates saturation behavior (Fig. 3) . Based on the shape of the QCM curves and the rapid increase in mass with time, it can be assumed that the kinetics would occur quickly and with presumably large equilibrium association constants. To test this theory, a calibration of anti-HA to HA binding was determined, and the binding of the functionalized nanoparticles was compared. The HA protein from H5N1 was exposed to anti-HA antibodies at concentrations ranging from 1 to 40 μg/mL (Fig. 4a, b) . This binding can be compared at any time point to generate a calibration curve (Fig. 4c , example shown is at 1.5 min). Based upon the desired separation of the lower data points, the binding can continue for several more minutes. Also, the amount of nanoparticle exposed to the surface can be lowered to prevent overloading of the sensor. Measuring the maximum binding that occurs, as opposed to lower time points, can be used to determine the equilibrium association constant (K a ) and increase our understanding of the affinity of our sensor. This constant was determined by fitting to a linear rearrangement of the Langmuir adsorption isotherm, where C was plotted versus C/Δm (Eq. 3, Fig. 4b ) [30] : This yielded an equilibrium association constant for the binding between the HA protein and anti-HA of 7.14± 0.26×10 7 . This K a is in the range expected for antibodyantigen interactions, from 10 6 to 10 10 M −1 [57] . In fitting the binding of the nanoparticles to this Langmuir isotherm calibration (at the experimentally used nanoparticle concentration of 500 μg/mL), the sensor response to HA-FLAG-Au would have the same binding as 0.92±0.01 μM HA protein, and HA-Au would generate the same sensor response as 1.50±0.02 μM HA protein. The large response seen is at the maximum of the Langmuir isotherm. This demonstrates that even at lower concentrations, these functionalized nanoparticles can be used as a positive control. The functionalized MPC-antibody binding is not inhibited by the presence of an additional non-interacting epitope (either FLAG or HA) on the polyfunctionalized nanoparticle; therefore, multiple binding interactions can be explored simultaneously. The ability to create multi-epitope-presenting nanoparticles that can orthogonally bind to specific monoclonal antibodies has been demonstrated using both ELISA and immunological QCM. Determination of the extent of antibody-functionalized nanoparticle binding is rapid using the QCM compared to ELISA. Also, like ELISA, the immunological response is specific, with QCM incurring less nonspecific binding. Interaction of the epitope with its antibody was improved through the use of a PEG linkage for epitope attachment to the MPC. Binding studies at the QCM show that polyfunctionalized gold nanoparticles exhibit the expected affinity to both antibodies that a normal immunological response is achieved from matched antibody-antigen couples and that an orthogonal response results otherwise. The results demonstrate that binding of polyfunctionalized gold nanoparticles could be used to determine sensor functionality, without resorting to the use of attenuated or killed microorganisms, or extracted and purified whole proteins. Fundamentals of molecular virology BU grad student develops infection: possible link to lab experiments raises new safety concerns. The Boston Globe Handbook of preparative inorganic chemistry Quartz crystal microbalance characterization of nanostructure assemblies in biosensing Fmoc solid phase peptide synthesis, The practical approach series Chemical approaches to the synthesis of peptides and proteins Applications of the QCM to electrochemistry Theoretical interpretation of the QCM response Spartan '04. Wavefunction, Inc., Irvine 49. Sauerbrey G (1959) Zeitschrift fuer Lehninger principles of biochemistry Acknowledgements The authors acknowledge support from NIH grant R01 GM076479 and DTRA HDTRA1-10-0067. Leslie A. Hiatt acknowledges support from a Department of Education GAANN fellowship (P200A040022). Robert G. Keil acknowledges sabbatical leave support received from the University of Dayton and Vanderbilt University. We thank Brian N. Turner for his assistance in TEM and TGA data collection.