Capture of the Circulating Plasmodium falciparum Biomarker HRP2 in

Sep 18, 2014 - QIMR Berghofer Medical Research Institute, Herston, Queensland 4006, ... ARC Centre of Excellence in Convergent Bio-Nano Science and ...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/ac

Capture of the Circulating Plasmodium falciparum Biomarker HRP2 in a Multiplexed Format, via a Wearable Skin Patch Khai Tuck Lee,†,⊥ David A. Muller,†,‡,⊥ Jacob W. Coffey,†,⊥ Kye J. Robinson,†,⊥ James S. McCarthy,‡,§ Mark A. F. Kendall,†,‡,∥,⊥ and Simon R. Corrie*,†,‡,⊥ †

The University of Queensland, Australian Institute for Bioengineering and Nanotechnology, Delivery of Drugs and Genes Group (D2G2), St. Lucia, Queensland 4072, Australia ‡ Australian Infectious Diseases Research Centre, St. Lucia, Queensland 4067, Australia § QIMR Berghofer Medical Research Institute, Herston, Queensland 4006, Australia ∥ The University of Queensland, Faculty of Health Sciences, St. Lucia, Queensland 4072, Australia ⊥ ARC Centre of Excellence in Convergent Bio-Nano Science and Technology, The University of Queensland, St Lucia, Queensland 4072, Australia S Supporting Information *

ABSTRACT: Herein we demonstrate the use of a wearable device that can selectively capture two distinct circulating protein biomarkers (recombinant P. falciparum rPf HRP2 and total IgG) from the intradermal fluid of live mice in situ, for subsequent detection in vitro. The device comprises a microprojection array that, when applied to the skin, penetrates the outer skin layers to interface directly with intradermal fluid. Because of the complexity of the biological fluid being sampled, we investigated the effects of solution conditions on the attachment of capture antibodies, to optimize the assay detection limit both in vitro and on live mice. For detection of the target antigen diluted in 20% serum, immobilization conditions favoring high antibody surface density (low pH, low ionic strength) resulted in 100fold greater sensitivity in comparison to standard conditions, yielding a detection limit equivalent to the plate enzyme-linked immunosorbent assay (ELISA). We also show that blocking the device surface to reduce nonspecific adsorption of target analyte and host proteins does not significantly change sensitivity. After injecting mice with rPf HRP2 via the tail vein, we compared analyte levels in both plasma and skin biopsies (cross-sectional area same as the microprojection array), observing that skin samples contained the equivalent of ∼8 μL of analyte-containing plasma. We then applied the arrays to mice, showing that surfaces coated with a high density of antibodies captured a significant amount of the rPf HRP2 target while the standard surface showed no capture in comparison to the negative control. Next, we applied a triplex device to both control and rPf HRP2-treated mice, simultaneously capturing rPf HRP2 and total IgG (as a positive control for skin penetration) in comparison to a negative control device. We conclude that such devices can be used to capture clinically relevant, circulating protein biomarkers of infectious disease via the skin, with potential applications as a minimally invasive and lab-free biomarker detection platform. form of skin “tattoos” is highly encouraging for laboratory-free biomarker monitoring,6 but this field is in its infancy and is not yet applicable to protein detection. Lateral flow assays are also available to detect a range of protein analytes in extracted blood and have proven useful in remote locations for infectious disease diagnosis.7 However, a key limitation is the lack of multiplexing capability required for differential diagnosis, which is an increasingly important issue.8 Technologies and devices that can noninvasively capture several protein biomarkers from circulating blood, in multiplexed formats, are of interest to move panel-based testing out of the hospital laboratories and

B

lood continues to be the main sample collected for clinical diagnostics1,2 as it is an easily accessible and rich source of disease-specific biomarkers. However, blood samples are invasively collected with needles as bulk suspensions of plasma and cells that need to be processed in a laboratory in order to extract, capture, and detect the biomarkers of interest.2 The venepuncture procedure can be painful for some patients3,4 (especially those requiring frequent sampling) and requires access to trained personnel5 and complex laboratory facilities.2 Bedside analysis systems are commercially available for a limited range of clinically important markers (e.g., i-STAT System−end point analysis of electrolytes, metabolites, and more recently, cardiac troponin, in a single blood sample); however, blood draws are required for each measurement. The emergence of wearable, real-time monitoring devices in the © 2014 American Chemical Society

Received: August 23, 2014 Accepted: September 18, 2014 Published: September 18, 2014 10474

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

penetration or identify penetration failures. We seek to provide some answers to these issues in the current study. In a range of published studies, methods to improve antigen capture efficiency onto antibody-immobilized films on various substrates in vitro have been thoroughly investigated. The “gold standard” methodology in clinical laboratories is still the enzyme-linked immunosorbent assay (ELISA), in which (a) capture antibodies are adsorbed to a high protein binding polymer surface, (b) “blocking” agents (e.g., milk serum) are added to passivate the remaining surface adsorption sites, (c) sample is incubated on the plate to capture specific proteins, and (d) a secondary, reporter-labeled antibody is added that specifically binds to another site on the analyte such that it can be indirectly detected via absorption or fluorescence. In a series of well-known studies, Buijs et al. showed how antibodies physisorb onto charged and hydrophobic surfaces, describing the adsorption/desorption kinetics and the effects of antibody density and orientation on antigen binding capacity.21,22 Caruso et al. showed that antibodies could be covalently immobilized to gold substrates with antigen capture efficiency similar to that for physisorbed antibodies.23 While controlling orientation of antibodies by attachment via the carbohydrate groups in the Fcregion (periodate treatment) was demonstrated,23,24 antigen binding capacity was not significantly improved in comparison to the randomly, or “statistically”, oriented molecules. Immobilization of antibodies in oriented fashion via the Fc region, using preimmobilization of bacteria-derived Fc- binding proteins (e.g., protein A or G), yielded significantly improved antibody orientation and antigen binding in comparison to random orientation, for antibody concentrations >1 μM.25−30 However, it is unclear what are the effects of these variables on the assay limit of detection (LOD; defined as the minimum detectable analyte concentration), which is the key parameter to determine the analytical sensitivity of an assay. For example, while Song et al. showed that the detection limit of an in vitro prostate-specific antigen (PSA) assay was approximately 100fold improved if antibodies were oriented with protein G28 in comparison to random orientation, the solution conditions used for antibody immobilization in this study were likely suboptimal in light of other studies.30,31 Work by Tajima et al. showed that the effect of antibody orientation on antigen binding capacity is also a function of the antigen’s size; smaller proteins are more likely to be affected, whereas larger proteins do not “see” the effect.25 In a recent study, Saha et al. suggested that for nanoparticles coated with antibodies via EDC/NHS chemistry (under conditions similar to those used in this study), the analytical sensitivity for cardiac troponin (∼24 kDa) diluted in blood increased as the immobilized antibody concentration increased.32 Multiplexing in diagnostic assays allows for the simultaneous detection of multiple biomarkers from the same sample. This has key applications ranging from differential diagnosis of infectious diseases given common clinical presentations8,33 through to the analysis of biomarker “panels” for tumor detection, classification, and monitoring response to treatment.34 A very important application of multiplexing is the provision of internal controls or calibrations, providing a basis for comparison between patients or providing semiquantitative analysis for kinetic monitoring applications (e.g., measuring biomarker rise or fall in response to therapy). In the specific case of microneedle arrays, the inclusion of a positive control to confirm successful penetration will be crucial for clinical use, as

into primary care clinics or into the home as point-of-care devices. Therefore, a key challenge is to develop methods of selectively capturing circulating disease-related biomarkers,2,9 preferably in a multiplexed manner, in a way that is minimally invasive for patients and does not rely on access to laboratory facilities. The dermal layers of the skin are highly vascularized, raising the possibility of noninvasive blood sampling via the skin using portable devices.9,10 Indeed, lancets have been used to extract fingerprick blood samples in the context of hemoglobin monitoring and infectious disease diagnosis. However, the lancet causes pain by penetrating deeply into the skin4 and is not capable of selective protein capture. As the blood proteome is dominated by ∼25 highly abundant proteins, the “needle in a haystack” problem is simply passed downstream, interfering with complex assay/detection chemistry,2 be it part of a lateral flow assay, electrochemical analyzer, or microfluidic device. To our knowledge the only method currently used in the clinical setting for semiselective protein capture from serum-containing fluids in situ is microdialysis,11 linked to a variety of readouts. While powerful, the purity of the sample dialysate is based on the molecular weight cutoff of the dialysis membrane, hence this is not a particularly selective method and still relies on the selectivity of the downstream assay. To date, the most common approach to selectively capture individual proteins from complex media is to attach binding molecules (e.g., antibodies) to a nonfouling surface, capture the proteins, and use a secondary label or label-free signal transduction method (e.g., surface plasmon resonance) for detection. However, to date all of these techniques are based on in vitro analysis, following blood extraction. Microprojection arrays (MPAs) comprise 10s−1000s of micrometer-scale projections per square centimeter. These microprojections can penetrate the outer layers of the skin to interface directly with the rich vasculature of the dermis. Several clinical studies have confirmed significant reduction in pain scores for similar microneedle-based devices in comparison to the traditional hypodermic needle.12 Since the early 2000s, several groups have investigated the utility of hollow microneedle arrays for interstitial fluid (ISF) sampling and glucose monitoring. Mukerjee et al. demonstrated ISF extraction from the skin using a hollow microneedle array followed by detection of glucose in the extracted fluid.13 Others have incorporated carbon fiber and carbon paste electrodes into the microneedle channels for detection of glucose, glutamate, lactate, and pH.14 Detection of other molecules such as electrolytes (K+ and Na+) have also been proposed using both single microneedles15 and microneedle arrays16 for ISF extraction with integrated transducers. Our group has demonstrated that polycarbonate or gold-coated silicon MPAs can be modified with antifouling polyethylene glycol (PEG) chains and analyte-specific probes, in order to capture antibody or protein biomarkers via skin application in live mice with minimal loss of probes into the skin.17−19 A significant challenge for devices that sample biomarkers in vivo is developing robust surface chemistries that have high capture efficiency, in order to extract sufficient quantities of analytes directly from the complex biological fluid. Recently we have reported improvements in capture efficiency via changing MPA geometry and application time17 and extending from antibody capture to dengue NS1 antigen capture.20 However, key questions remaining include how to improve assay sensitivity and how to incorporate internal controls (multiplexing) to account for differences in skin 10475

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

3,3′,5,5′-tetramethybenzidine (TMB, SurModics) for 10 min, quenched with 1 M phosphoric acid, and read at 450 nm (FLOUstar Omega, BMG Labtech, Germany). MPA Preparation and PEG Coating. Gold-coated MPAs with microprojections of 250 μm in length were produced by Deep Reactive Ion Etching (DRIE) and diced to either 4 × 4 mm2 or 4 × 1.3 mm2 (“logs” for multiplexing). MPAs were applied to the dermal layer of mouse flank skin using a spring loaded applicator, using methods described in detail elsewhere.17,20,37 MPA cleaning and coating was carried out as previously described in Barghav et al.18 Briefly, gold coated silicon MPAs (“Au-MPAs”) were thoroughly rinsed in ethanol, acetone, and Milli-Q water followed by cleaning in a base piranha solution (5 mL 25% ammonia solution in 50 mL water with 10 mL 35% w/v hydrogen peroxide) for 15 min at 80 °C and then rinsed thoroughly with water. Heterobifunctional polyethylene glycol (COOH-PEG-HS; MW, 5000 g/mol; Creative PEGWorks) was coated onto Au-MPAs under “cloud-point” conditions. A concentration of 0.2 mg/mL of PEG in 0.6 M K2SO4 was prepared and heated to 60 °C, and then the Au-MPAs were resuspended in the coating solution and incubated at 60 °C for 48 h. PEG-coated MPAs (“PEGMPAs”) were then rinsed with 0.6 M K2SO4 followed by three times to Milli-Q water to remove excess unbound PEG. Antibody Immobilization to PEG-MPAs. PEG-MPAs were activated by incubation in 0.1 M N-morpholinoethanesulfonic acid hydrate buffer, pH 5 (MES buffer; MP Biomedicals) containing 5 mM of both 1-ethyl-3-(3(dimethylamino)propyl)carbodiimide (EDC; Thermo Scientific Pierce) and N-hydroxysuccinimide (NHS; Thermo Scientific Pierce) at room temperature for 90 min.38 PEGMPAs were then incubated in 200 μL of C1-13 antibody solution at 5 μg/mL (under a range of solution conditions; “standard density” conditions comprise 2 h incubation in PBS buffer, pH 7, at room temperature). Goat antimouse Fc-specific IgG (Sigma; for positive controls in multiplex experiments) antibody and antidengue NS1 monoclonal antibodies (donated by Prof Paul Young, UQ; for negative controls in multiplex experiments) were also coated to PEG-MPA logs for multiplex experiments, both under the standard density conditions. After antibody immobilization “Ab-MPAs” were capped with 0.1 M glycine for 30 min. A washing sequence was then initiated to minimize nonspecific antibody attachment: 1 M ethanolamine, pH 8.5 for 15 min, then 0.1% Brij L23 (polyoxyethylene lauryl ether; Sigma)/PBS for 15 min, then 0.1% Brij L23/PBS 12 times and finally, 6 times with PBST. For immobilization of fluorescently labeled C1-13 or rPf HRP2 to MPA surfaces, C113 antibody and rPf HRP2 were labeled with Dylight 633 and 550, respectively, according to the manufacturer’s instructions (Thermo Scientific Pierce). MPAs coated with fluorescent proteins were imaged using a fluorescent scanner (Tecan LS Reloaded) and the mean fluorescent intensity data for each MPA was retrieved using ImageJ software. In cases where AbMPAs were blocked prior to use, blocking consisted of incubation in 1× KPL milk diluent (KPL, Inc., Gaithersburg, MD) or 5% bovine serum albumin (BSA) diluted in PBS, overnight at 4 °C. In Vivo Microprojection Array ELISA (MPA-ELISA). Six− eight week old female BALB/c mice were maintained under pathogen free conditions in the University of Queensland Biological Resources animal facility at the Australian Institute for Biotechnology and Nanotechnology (AIBN). All animal experiments were performed with approval from a University of

negative results in isolation could be due to a range of factors including poor skin penetration. In this study, we demonstrate selective extraction of Plasmodium falciparum Histidine-Rich Protein 2 (Pf HRP2) protein from the skin of live mice with a microneedle-based device, in a multiplexed assay simultaneously capturing positive (total IgG) and negative (dengue virus NS1) controls. To our knowledge this is the first example of multiplexed biomarker capture from the skin. Pf HRP2 is a well-established biomarker for P. falciparum, a malaria parasite that causes significant morbidity and mortality in humans.35 Globally, an estimated 216 million people are infected with malaria with an annual mortality of approximately 655 000 people, and rapid diagnostics can help direct appropriate treatment.36 In order to capture the target protein with a high signal-to-noise ratio in vivo, we found that solution conditions that favored high density capture antibody immobilization led to a significant improvement in the in vitro detection limit (∼100-fold), equivalent to that in the plate ELISA format. We then employed the MPAs to selectively capture the target protein from the skin of live mice, following tail-vein injection of our recombinant Pf HRP2 protein (rPf HRP2). We explored the addition of a blocking agent that reduced nonspecific protein adsorption without significantly altering assay sensitivity. Finally we applied multiplex MPA devices to live mice, simultaneously capturing total IgG (a positive control for penetration) and rPf HRP2 target protein in comparison to a negative control surface.

2. MATERIALS AND METHODS Gel Electrophoresis and Immunoblots. Samples were prepared by combining 4 μL of loading dye with 10 μL of rPf HRP2 samples and 1 μL of β-mercaptoethanol. Samples were then heated to 95 °C and then loaded into a 4−12% BisTris SDS-PAGE Gel (Novex, Invitrogen). Gel electrophoresis was carried out at a constant 200 V for 45 min, and then proteins were transferred to nitrocellulose membranes in a Novex transfer module according to the manufacturer’s instructions. Transfer was performed at 30 V for 1 h followed by a blocking step with 5% skim milk powder for 1 h. Membranes were incubated with anti-Pf HRP2 antibody (1:1000 dilution of C1-13 clone generously donated by the National Bioproducts Institute, Pinetown, South Africa) in phosphate buffered saline solution containing 0.05% tween-20 (PBST) for 1 h. Membranes were then incubated with Li-COR800 goat antimouse detection antibody (Li-COR Biosciences) for 1 h at room temperature (protected from light) and subsequently washed 3 times with PBST. Bands on the nitrocellulose film were detected and visualized using infrared fluorescent scanner (ODYSSEY Li-COR Biosciences). For detailed information on the recombinant expression of rPf HRP2, please see the Supporting Information. rPf HRP2 Plate ELISA. rPf HRP2 capture ELISA was carried out as previously described in Muller et al.20 Briefly, plates were coated with 5 μg/mL C1-13 antibody in carbonate buffer (0.05 M carbonate bicarbonate, pH 9.6; Sigma) and incubated overnight at 4 °C. After coating, plates were blocked with 1× KPL milk diluent (KPL, Inc., Gaithersburg, MD) for 2 h at room temperature. Samples were then incubated on coated plates for 1 h at 37 °C and washed 6 times with PBST. Detection antibody (horseradish peroxidase C1-13 mAb conjugate) was then added for 60 min at 37 °C followed by washing 6 times with PBST. Plates were developed using 10476

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

Queensland Animal Ethics Committee in accordance with NHMRC guidelines. rPf HRP2 antigen capture was carried out via application of Ab-MPAs to the flank skin of BALB/c mice. Mice were anesthetized with 60 μL of intraperitoneal injection of xylazine, ketamine, and saline in a 1:1:2 ratio. The mice were then injected with rPf HRP2 intravenously (50 or 200 μg in 100 μL saline or saline control), and 30 min later, Ab-MPAs were applied to flank skin with a spring-loaded applicator with a velocity of 2.1 m/s, with detailed information on the apparatus published elsewhere.17 Ab-MPAs were subsequently washed with PBST in a 96-well plate, incubated with 50 μL of detection antibody (horseradish peroxidase C1-13 mAb conjugate) for 1 h at 37 °C and washed again with PBST. MPAs were transferred to new wells prior to development with TMB for 10 min. The reaction was quenched with 1 M phosphoric acid and the resulting cultured solutions analyzed at 450 nm on a UV spectrophotometer (FLOUstar Omega, BMG Labtech, Germany). Preparation of Skin Biopsies and Serum. Following MPA-ELISA and prior to euthanization, blood samples were collected from all mice either through retro-orbital or cardiac bleeds. Blood samples were left at room temperature for 1−2 h to clot prior to serum separation (centrifugation of blood samples at 10 000g for 5 min). Mice were then euthanized and flank skin samples excised (same perimeter as the 4 × 4 mm2 MPA), weighed, and transferred to a homogenizing tube (FastPrep lysing matrix D) after which 500 μL of signaling lysis buffer (Merck, Millipore) was added. Tubes were then subjected to 9 cycles of 6500 rpm homogenization (Precellys24) with 30-s cooling intervals. Lysates were then centrifuged at 4 °C at 10 000g for 20 min, and the pellet was discarded. The serum and skin tissue lysate samples were then analyzed by rPf HR2 plate ELISA as described above for quantitation. Data Analysis. Experiments were performed in triplicate or with n = 5 for animal studies, with one- or two-way ANOVA used for statistical analysis. In order to quantitatively compare the intensities of A550-labeled rPf HRP2 and A633-labeled C113 antibody, we normalized the slopes of intensity/concentration plots for each dye using titrated solutions in the Tecan scanner. This analysis was applied to the raw MPA data (Figure 2C) such that intensity differences could be compared in a relative manner.

Figure 1. Characterization of recombinant rPf HRP2 protein. (A) SDS-PAGE gel and (B) Western blot (note lane Ni-NTA = rPf HRP2 after Ni-NTA purification, SECC = rPf HRP2 after Ni-NTA and SECC purification); (C) rPf HRP2 plate ELISA showing the effect of antigen purity on detection limit.

3. RESULTS Recombinant Expression of Pf HRP2. Recombinant Pf HRP2 (rPf HRP2) was produced as a model antigen for surface chemistry investigations and mouse experiments (see the Supporting Information for expression methods). Figure 1 shows the appearance of a dominant band at ∼60 kDa in both SDS-PAGE gel and Western Blot, confirming that this protein is reactive against the widely used C-13 monoclonal antibody clone. Size-exclusion column chromatography (SECC) purification was performed after initial purification (Ni-NTA) to remove truncation and multimer products which contributed to an overall reduction in assay limit of detection (LOD) in a traditional plate capture ELISA assay, from ∼3−60 ng/mL (equivalent to other published studies39). We found for this protein that the expression yield was ∼5−10 mg/L purified protein and that the protein was very stable, even after several freeze/thaw cycles (Figure S1 in the Supporting Information). Selection of Antibody Immobilization Conditions for Increased Assay LOD. In translating capture ELISA assays from a plate surface to MPAs, we have previously observed

poorer assay sensitivity and detection limits when target antigen was diluted in complex biological fluids such as blood plasma or serum.20 We hypothesized that by choosing antibody attachment conditions that lead to high target capture efficiency at a single analyte concentration, we could improve the LOD for both in vitro and in vivo application. Using the EDC/NHS activation reaction conditions elegantly optimized by Sam et al.38 and an optimized washing procedure to minimize the fluorescent signal arising from nonspecifically bound antibodies (Figure S2 in the Supporting Information), we attached fluorescently labeled C1-13 monoclonal antibodies to PEGMPAs, varying the attachment temperature, buffer strength, and pH. Previous studies have confirmed the importance of these parameters in terms of capture antibody density and orientation.30,31 Figure 2A shows that immobilized antibody density is significantly higher at low pH and also significantly higher in the low ionic strength buffer, at a given pH. In the MES buffer, it also appears that higher temperature is associated with higher density. However, the immobilization time (2 h vs 16 h) had little observable effect (data not shown). 10477

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

Figure 2. Line and symbol graphs showing results from a matrix experiment designed to explore the effects of solution temperature, pH, and buffer/ strength on (A) antibody density; (B) subsequent target analyte capture efficiency on surfaces with corresponding antibody densities from part A; (C) normalized binding ratio as a function of antibody density; (D) ELISA detection of rPf HRP2 target analyte on Ab-MPA surfaces as a function of antibody immobilization conditions and matrix complexity. Note that for part D, “High” designates the high density conditions (pH 5, MES buffer, 37 °C) and “Standard” designates the standard density conditions as described in the Materials and Methods section.

improved the assay detection limit for rPf HRP2 detection in 20% serum (∼1000 ng/mL to ∼10 ng/mL), in comparison to the standard density condition (PBS buffer, pH 7). This result appears to suggest that, for rPf HRP2 protein, high density antibody surfaces lead to a significant improvement in LOD (Figure 2D), similar to that for the equivalent plate ELISA assay (Figure 1C). Reduction of Nonspecific Binding and Selection of Negative Controls. After identifying a reduced LOD in vitro based on the solution conditions for antibody immobilization onto PEG-MPAs, we moved on to investigating possible strategies to include “negative control” Ab-MPA surfaces into multiplexed assays. These negative controls should provide a low signal based on negligible binding of (a) target protein, (b) nonspecific protein, or (c) secondary antibody specific to the target protein, to the Ab-MPA surface. For initial experiments involving a range of antibodies unrelated to Pf HRP2, we observed significant nonspecific binding to the surface, either in

While this approach does not measure the absolute antibody density, measurement of the fluorescence intensity from the surface-immobilized antibody allows relative quantitation to assess the effects of varying conditions. Interestingly, when these Ab-MPAs were used to capture the fluorescently labeled rPf HRP2 protein diluted in buffer only (i.e., no additional protein background), antigen capture did not appear to be simply dependent on antibody density (Figure 2B). Once again, there appeared to be higher capture efficiency at lower pH and higher temperatures. Combining the data from Figure 2A,B and taking into account the effect of the different dyes used, Figure 2C shows the normalized ratio of antigen to antibody as a function of the antibody immobilization conditions, showing that there appears to be a significantly higher capture efficiency at the higher pH and higher ionic strength conditions. In capture ELISA assays on Ab-MPA surfaces, we found that the higher density antibody immobilization conditions (MES buffer, pH 5) significantly 10478

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

that blocking the Ab-MPA surface with BSA resulted in only a slight decrease in signal across all six surfaces. However, surfaces blocked with milk serum diluent resulted in significantly lower nonspecific signal when the capture antibody was either absent (iii) or unrelated to the target (vi). Furthermore, given that nonspecific signal resulting from adsorption of the detection antibody (v) alone was at background levels regardless of blocking conditions, it suggests that the nonspecific signals for surfaces (iii) and (iv) are both due to adsorption of the target protein. Regardless of the blocking strategy employed, the PEG-MPAs coated with high density C1-13 antibody (ii) yielded consistently higher positive signals in comparative to the low density case (i), in agreement with Figure 2D. Finally, we repeated the serial titration experiment, incorporating a milk serum blocking step, showing that the same LOD could be reached with samples containing 20% mouse serum as described in Figure 2. Capture and Detection of rPf HRP2 in the Skin of Live Mice. After optimizing the in vitro response of Ab-MPAs to complex solutions containing rPf HRP2 protein, we next sought to apply Ab-MPAs to the flank skin of live mice following tail vein injection of saline solutions containing rPf HRP2 (see Figure 4). We confirmed that rPf HRP2 protein introduced

the absence or presence of negative control capture antibody (i.e., on PEG-MPA surfaces alone; data not shown). Therefore, even though the presence of a PEG layer reduced the nonspecific adsorption of rPf HRP2 to the Ab-MPA surface, a significant level of nonspecific signal remained. This is not surprising given that Pf HRP2 is rich in histidine residues, so much so that it can be efficiently captured onto a nickel affinity column without a purification tag. Furthermore, others have reported that histidine tagged proteins adsorb strongly to gold.40 We therefore decided to investigate the effect of using common blocking agents to see if further reduction of nonspecific signal was possible. Figure 3 shows the HRP-labeled detection antibody absorbance readings measured on six different MPA surfaces either unblocked or blocked with BSA or milk serum diluent (KPL) agent. First, from a broad perspective, Figure 3A shows

Figure 4. Schematic describing the experimental design behind the animal study: (A) BALB/c mice (n = 5) were injected with rPf HRP2 protein (50 or 200 μg) via the tail vein; (B) Ab-MPAs were applied to the flank skin (either as a triplex or singleplex application), blood collected, and skin biopsies homogenized; (C) Ab-MPA, skin, and blood serum samples were analyzed by ELISA.

intravenously was detectable in both serum and skin (Table 1) using sera and tissue ELISAs. The maximum skin rPf HRP2 protein concentration was reached approximately 30−45 min after injection (Figure S3 in the Supporting Information). We found that the skin biopsy, also 16 mm2 in cross-sectional area, contained the equivalent of ∼7−9 μL of analyte-containing plasma. Square MPAs (16 mm2 footprint area) coated with anti-Pf HRP2 capture antibody under high or low density conditions were applied to the flank skin of the mice injected with rPf HRP2 (Figure 5A). Following MPA ELISA, 2-way ANOVA analysis confirmed that both the rPf HRP2 concentration (p-value = 0.0004) and the antibody attachment conditions (p-value < 0.0001) significantly affected the absorbance readings and also that there was significant interaction between the rPf HRP2 concentration and attachment method (p-value = 0.011; expected, as the negative control should yield similar signals in each case). Importantly,

Figure 3. Effect of blocking the Ab-MPA surface on specific and nonspecific binding; (A) direct ELISA assays performed using two common blocking reagents; (B) titration of rPf HRP2 in either PBST or PBS containing 20% mouse serum (MS) for Ab-MPAs coated on standard (“std”) or “high” density, and blocked with KPL prior to assay. Note that in the schematic for part A, (i) refers to antibodies immobilized using the standard (PBS, pH 7) solution conditions while (ii) refers to high density conditions (MES, pH 5). 10479

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

Table 1. Quantification of PfHRP2 in Skin and Serum Samples Following Tail Vein Injection at Two Concentrations amount rPf HRP2 injecteda

sample material

200 μg 50 μg 200 μg 50 μg

serum serum skin homogenate skin homogenate

concnb (ng/mL)

tissue weightc (mg)

tissue concn (ng/mg)d

equivalent plasma volume (μL)e

± ± ± ±

NA NA 8.1 ± 1.6 7.6 ± 0.9

NA NA 11.0 ± 3.2 4.5 ± 1.8

NA NA 7.5 ± 1.8 8.7 ± 4.3

11830 3840 178 67

1245 1100 38 27

a

n = 5 mice per group. bQuantified by ELISA standard curves (see Supplementary Figure S4 in the Supporting Information for raw data). cBiopsy of skin taken with same cross-sectional area as MPA. dCalculated based on the fraction of rPf HRP2 (ng) detected in a biopsy of known tissue weight (mg). eCalculated by dividing rPf HRP2 (ng) detected in a biopsy by the serum concentration (ng/mL).

penetration into skin and hence reduce the number of potential false negative results. To facilitate multiplexed analysis in our study, we simply diced the square MPA devices into three 4 × 1.3 mm2 “logs”, attaching different capture antibodies for the IgG positive control (anti-IgG Fc specific), negative control (antidengue NS1), and rPf HRP2 (C1-13 antibody). Minimal cross-reactivity was confirmed by incubating all three samples in 1% serum spiked with rPf HRP2 (Figure S5 in the Supporting Information). We then combined these logs to give the same skin footprint area as a square MPA and applied the multiplex device to the opposite flank of the same mice. The results, presented in Figure 5B, show that the rPf HRP2 signal increases significantly from the negative saline control through to the mice injected with rPf HRP2. The IgG positive control remains significantly high above the background, while the negative control remains low although with some evidence of nonspecific adsorption. Importantly, it appears that cross-binding did not feature significantly between the three logs even in close proximity. The same 2-way analysis of variance (ANOVA) with multiple comparisons test was performed to analyze the rPf HRP2 and negative control data as outlined for Figure 5A. The 2-way ANOVA analysis confirmed that both the rPf HRP2 concentration (p-value < 0.0001) and the capture antibody used (pvalue < 0.0001) significantly affected the absorbance readings and also that there was significant interaction between the rPf HRP2 concentration and attachment method (p-value = 0.038). The multiple comparisons test showed furthermore that the rPf HRP2 signals for mice injected with the protein were significantly different from Ab-MPAs applied to mice injected with a saline negative control solution (p-value < 0.01 or < 0.0001 for increasing amount of rPf HRP2 skin concentration) whereas signals on the positive IgG and negative control logs were not significantly different from signals detected on negative control mice (p-values > 0.05). This suggests that triplex Ab-MPA application resulted in the simultaneous detection of biomarker-specific signals on each log.

Figure 5. In vivo Ab-MPA application to skin of live mice injected with rPf HRP2 or saline controls using (A) single MPAs coated with antiPf HRP2 antibodies coated under “high” or “standard” solution conditions (each MPA 16 mm2 area footprint); (B) triplex MPA application (each MPA 4 × 1.3 mm2 area footprint) allowing simultaneous detection of rPf HRP2, total IgG in comparison to a negative control surface. The inset of part B shows the potential utility of the total IgG control when used as an internal calibration for antigen levels.



DISCUSSION The increase in antibody density based on changing solution conditions is in alignment with theory and other published work. In terms of pH, maximum antibody adsorption to a surface tends to occur at or around the antibody’s pI, which is typically ∼5. Furthermore, on carboxylic acid-functionalized surfaces, electrostatic interactions are favored in the range between the pI of the antibody and the pKa of the surface (∼3− 441), where the surface maintains an overall negative charge and the antibody is positively charged. This is true even for NHSactivated surfaces, as the efficiency of activation to surfacebound acid groups is relatively low. Under high pH conditions, it has been previously shown that IgG molecules adsorb to charged surfaces via the Fc region, which adsorbs faster in

multiple comparisons analysis also showed that only the high density attachment method yielded significant levels above the negative control saline injection (p-value < 0.05 for the low density case and p-value < 0.0001 for the high density case). This result suggests that the high density attachment conditions for the anti-Pf HRP2 capture antibody result in significantly higher capture efficiency in vivo. For MPA devices to be useful in practice, it is likely that multiplexed analysis is required to incorporate appropriate controls to allow comparison between test subjects. Furthermore, the IgG positive control (introduced in a previous study20) is used to confirm microneedle 10480

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

comparison to the Fab components.21 Thus, our results are in agreement with previous studies, in that the antigen/antibody ratio was higher at high pH, while antibody density was higher at the antibody pI and lower ionic strength.21,27,30,31 The dramatic improvement in detection limit (∼100-fold) on MPA surfaces in the presence of background serum, due to increased capture antibody density, appears to be quite a novel finding from this study. Indeed, our results may have broader implications for in vitro diagnostics, as we have identified a regime in which changes in antibody immobilization conditions result in significant improvements in assay LOD, for samples diluted in serum, linked to surface density rather than orientation. We identified no difference in assay LOD for antigen titrations in PBST alone (even though slightly higher signals were observed for any given analyte concentration), suggesting that the effect was limited to cases where nonspecific protein was present. Very few studies have investigated this effect, especially for surfaces coated with solution conditions favoring high antibody density as we have. Our results are broadly in agreement with a very recent study by Saha et. al, where it was also identified that increasing antibody surface density resulted in an improved assay LOD (prostate specific antigen) using EDC/NHS chemistry.32 Of clear relevance for in vivo devices, we also showed an improvement in signal when MPAs were applied to live mice. We kept the capture antibody concentration in the immobilization solutions constant at 5 μg/mL in our study, which is standard for ELISA assays (as per Figure 1C) but different to previous studies involving EDC/NHS-mediated antibody immobilization. ELISA surfaces are typically designed to be “high protein binding”, and the coating efficiency for capture antibodies is high. The capture efficiency would be expected to be significantly lower on PEG-coated substrates, given the antifouling nature of the surface. Indeed, most studies published using covalent antibody attachment chemistry onto gold or silicon substrates require concentrations >100 μg/mL for sufficient antigen capture.28−30,32 Interestingly, Wang et al. demonstrated that at antibody concentrations below 100 μg/ mL, the antibody density and subsequent antigen capture efficiency were similar or even slightly increased for EDC/ NHS-immobilized antibodies in comparison to oriented immobilization via protein A.29 Our results are in agreement, suggesting that at low antibody concentrations, antibody density can be optimized with significant effect on the assay LOD, without requiring complicated orientation approaches. However, this also raises the possibility that if the antibody concentrations are increased, assay LOD may be further improved. Indeed, ongoing work in our group is focused on improvements in assay sensitivity based on surface chemistry, projection geometry, and application time. In any case, we therefore suggest, in agreement with other studies,30,31 that solution conditions chosen for antibody immobilization can have significant effects on the resultant antibody density and orientation and that optimal orientation (i.e., Fc-down) alone may not always yield the surface with highest sensitivity or LOD. For regulatory approval of MPAs as implantable medical devices, it would be preferable to avoid coating projection surfaces with agents that are poorly defined or with uncharacterized toxicity in vivo. However, it is unlikely to be able to conclusively establish that no deposition could occur because of limits in the sensitivity of currently available technology. Therefore, it is more prudent to incorporate

biocompatible and nontoxic materials that are considered safe upon low-level deposition. The use of antibody-based probes represents a logical approach because therapeutic antibodies are widely accepted and delivered intravenously at significantly higher doses. Humanizing or reformatting (e.g., from whole antibody to a fragment that lacks Fc-function) of antibodies is a widely applied approach to overcome undesired immune responses. In the case of the blocking agents used in our study, our results suggest that they may impart significant benefits to warrant their inclusion, due to the observed reduction in nonspecific protein adsorption. BSA and milk serum diluents are very commonly used in ELISA assays; however, in this study the latter performed significantly better. In our case, rPf HRP2 protein itself adsorbs strongly to gold, in comparison to IgG and other high abundance proteins (data not shown), due to the high percentage of histidine residues, which are known to adsorb to gold surfaces.40 However, given the poorly defined nature of these particular blocking agents, we suggest that incorporation of synthetic, biocompatible blocking agents into stable surface chemistries may prove useful for inclusion into future microneedle devices. Following intravenous injection of our model analyte, rPf HRP2, we found that when the levels in skin reached a peak, a skin biopsy (same footprint area as an MPA) contained the same analyte concentration as 7−9 μL of plasma. This combination of results from both skin and blood tissue supports previously published studies, from our group and others, that skin contains significant amounts of circulating protein analytes for diagnostic application.9,42 The detection limit of in vitro ELISA assays is reported to be in the range of 1−10 ng/mL, consistent with our own ELISA and MPA surfaces; however, we have not determined the in vivo LOD for the MPA device. While not a focus of the current study, we did notice a significant difference in kinetics between serum levels and skin levels of rPf HRP2. A lag of between 2 and 45 min has also been observed between glucose levels in interstitial fluid and venous blood.43 Because of the lower permeability of capillaries to proteins, this delay may take hours.44 However, the kinetics may have been artificially altered in our study by increased intravascular pressure following intravenous injection. Given the dynamics of glucose in the blood, this is a significant issue in terms of the level of insulin to be injected, due to the narrow concentration range required for glucose. However, in the case of detecting infectious disease proteins, this is unlikely to be a problem, as the dynamics are significantly slower for the disease time course (e.g., days) than for serum/skin equilibration (minutes−hours). Further studies on the kinetics of biomarker levels in skin vs blood are currently underway in the context of relevant disease models, as analysis of such phenomena in “artificial” models may not be representative of disease-specific pathophysiology. The multiplexing experiment presented in Figure 5 confirmed that MPA devices can be combined and applied at the same time with minimal cross-reactivity between controland analyte-specific antibodies. It is possible that a PEG-MPA itself could be used as a negative control; however, we chose to keep the surface chemistry as similar as possible to the other MPAs, assuming inherent differences in adsorption might be observed between PEG-MPAs and Ab-MPAs. While the negative control signal did not change significantly with analyte concentration over the range investigated, there may be an upward trend indicative of further nonspecific binding of this particular analyte. In terms of a positive control for penetration, 10481

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry



it is unclear if IgG levels are affected by the presence of the specific analyte; if so, this may affect the interpretation of a ratiometric measurement. In the past we also used albumin as a penetration control;45 however, a dedicated study is required to investigate this aspect further and is beyond the scope of the current study. In terms of a ratiometric analysis, if it is assumed that the positive control and analyte Ab-MPA are equally affected by nonspecific binding, then the ratio rPf HRP2/IgG could be a useful indicator of antigen load in an individual. Indeed, using this method, two levels of quantitation are envisaged: (1) binary “present/absent” indication based on a lower “cut-off” based on uninfected individuals and (2) relative quantitation, where there is some discrepancy between high/ low analyte levels which could provide some insight as to the disease time course.

CONCLUSIONS Here we describe the first multiplexed wearable skin device capable of capturing circulating proteins from live mice. We found that controlling the surface chemistry of antibody attachment, using EDC/NHS chemistry, had a direct and significant effect on assay LOD in serum-containing solutions. Surfaces coated with a high density of antibodies were able to capture a significant quantity of rPf HRP2 target analyte, whereas the lower density surfaces, with higher antigen/ antibody ratios, did not capture target with respect to negative controls. Finally, we demonstrated for the first time that MPA devices can be applied to live mice to extract two different biomarkers from the skin in a practical multiplexed manner, paving the way for biomarker panel analysis in the future, with in-built controls for internal calibration purposes. ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.



REFERENCES

(1) Anderson, N. L. Clin. Chem. 2010, 56, 177−185. Anderson, N. L.; Anderson, N. G. Mol. Cell. Proteomics 2002, 1, 845−867. (2) Labuz, M.; Takayama, S. Lab Chip 2014, 14, 3165−3171, DOI: 10.1039/C4LC00125G. (3) Nir, Y.; Paz, A.; Sabo, E.; Potasman, I. Am. J. Trop. Med. Hyg. 2003, 68, 341−344. Fruhstorfer, H.; Muller, T.; Scheer, E. Pract. Diabetes Int. 1995, 12, 184−185. (4) Fruhstorfer, H. Eur. J. Pain-London 2000, 4, 301−305. Fruhstorfer, H.; Schmelzeisen-Redeker, G.; Weiss, T. Eur. J. PainLondon 1999, 3, 283−286. (5) Hopkins, H.; Oyibo, W.; Luchavez, J.; Mationg, M. L.; Asiimwe, C.; Albertini, A.; Gonzalez, I. J.; Gatton, M. L.; Bell, D. Malaria J. 2011, 10, 1−9. Luchavez, J.; Lintag, M. E.; Coll-Black, M.; Baik, F.; Bell, D. Malaria J. 2007, 6, 149. (6) Jia, W. Z.; Bandodkar, A. J.; Valdes-Ramirez, G.; Windmiller, J. R.; Yang, Z. J.; Ramirez, J.; Chan, G.; Wang, J. Anal. Chem. 2013, 85, 6553−6560. Windmiller, J. R.; Wang, J. Electroanalysis 2013, 25, 29− 46. (7) Ochola, L. B.; Vounatsou, P.; Smith, T.; Mabaso, M. L. H.; Newton, C. Lancet Infect. Dis. 2006, 6, 582−588. (8) D’Acremont, V.; Kilowoko, M.; Kyungu, E.; P, S.; Sangu, W.; Kahama-Maro, J.; Lengeler, C.; Cherpillod, P.; Kaiser, L.; Genton, B. N. Engl. J. Med. 2014, 370, 809−817. (9) Paliwal, S.; Hwang, B. H.; Tsai, K. Y.; Mitragotri, S. Eur. J. Pharm. Sci. 2013, 50, 546−556. (10) Wang, C. Y.; Maibach, H. I. Cutaneous Ocul. Toxicol. 2011, 30, 1−6. (11) Sjogren, F.; Davidsson, K.; Sjostrom, M.; Anderson, C. D. AAPS J. 2012, 14, 187−195. Sjogren, F.; Anderson, C. D. AAPS J. 2010, 12, 741−749. Varol, A. L.; Anderson, C. D. Arch. Dermatol. Res. 2010, 302, 383−393. Sjogren, F.; Anderson, C. D. Acta Derm.-Venereol. 2009, 89, 459−465. (12) Haq, M. I.; Smith, E.; John, D. N.; Kalavala, M.; Edwards, C.; Anstey, A.; Morrissey, A.; Birchall, J. C. Biomed. Microdevices 2009, 11, 35−47. Gill, H. S.; Denson, D. D.; Burris, B. A.; Prausnitz, M. R. Clin. J. Pain 2008, 24, 585−594. Kaushik, S.; Hord, A. H.; Denson, D. D.; McAllister, D. V.; Smitra, S.; Allen, M. G.; Prausnitz, M. R. Anesth. Analg. (Hagerstown, MD, U. S.) 2001, 92, 502−504. (13) Mukerjee, E.; Collins, S. D.; Isseroff, R. R.; Smith, R. L. Sens. Actuators, A: Phys. 2004, 114, 267−275. (14) Miller, P. R.; Gittard, S. D.; Edwards, T. L.; Lopez, D. M.; Xiao, X.; Wheeler, D. R.; Monteiro-Riviere, N. A.; Brozik, S. M.; Polsky, R.; Narayan, R. J. Biomicrofluidics 2011, 5, No. 013415. Windmiller, J. R.; Valdes-Ramirez, G.; Zhou, N.; Zhou, M.; Miller, P. R.; Jin, C.; Brozik, S. M.; Polsky, R.; Katz, E.; Narayan, R.; Wang, J. Electroanalysis 2011, 23, 2302−2309. (15) Miller, P. R.; Skoog, S. A.; Edwards, T. L.; Lopez, D. M.; Wheeler, D. R.; Arango, D. C.; Xiao, X.; Brozik, S. M.; Wang, J.; Polsky, R.; Narayan, R. J. Talanta 2012, 88, 739−742. (16) Gardeniers, H. J. G. E.; Luttge, R.; Berenschot, E. J. W.; de Boer, M. J.; Yeshurun, S. Y.; Hefetz, M.; van’t Oever, R.; ven den Berg, A. J. Microelectromech. Syst. 2003, 12, 855−862. (17) Coffey, J. W.; Corrie, S. R.; Kendall, M. A. F. Biomaterials 2013, 34, 9572−9583. (18) Bhargav, A.; Muller, D. A.; Kendall, M. A. F.; Corrie, S. R. ACS Appl. Mater. Interfaces 2012, 4, 2483−2489. (19) Yeow, B.; Coffey, J. W.; Muller, D. A.; Grondahl, L.; Kendall, M. A. F.; Corrie, S. R. Anal. Chem. 2013, 85, 10196−10204. (20) Muller, D. A.; Corrie, S. R.; Coffey, J.; Young, P. R.; Kendall, M. A. F. Anal. Chem. 2012, 84, 3262−3268. (21) Buijs, J.; White, D. D.; Norde, W. Colloids Surf., B: Biointerfaces 1997, 8, 239−249. Buijs, J.; vandenBerg, P. A. W.; Lichtenbelt, J. W. T.; Norde, W.; Lyklema, J. J. Colloid Interface Sci. 1996, 178, 594−605. (22) Buijs, J.; Norde, W.; Lichtenbelt, J. W. T. Langmuir 1996, 12, 1605−1613. Hlady, V.; Buijs, J. Curr. Opin. Biotechnol. 1996, 7, 72−77. Buijs, J.; Lichtenbelt, J. W. T.; Norde, W.; Lyklema, J. Colloids Surf., B: Biointerfaces 1995, 5, 11−23.





Article

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +617 3346 4209. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript Notes

The authors declare the following competing financial interest(s): Authors M.A.K. and S.R.C. are inventors on a patent that relates to this work. Related patents underpinning some of this work have been licensed to Vaxxas Pty Ltd. M.A.K. is the cofounder of Vaxxas and Chief Technical Officer.



ACKNOWLEDGMENTS The authors acknowledge funding from the Australian Research Council (ARC DECRA Fellowship (S.R.C.); Future Fellowship (M.A.F.K.)) and the Australian Infectious Diseases Research Centre. We also acknowledge the facilities and the scientific and technical assistance of Prof. Paul Young’s research group, the Australian National Fabrication Facility (ANFF Queensland), and the Australian Microscopy & Microanalysis Research Facility (AMMRF) at the Centre for Microscopy and Microanalysis, The University of Queensland. 10482

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483

Analytical Chemistry

Article

(23) Caruso, F.; Rodda, E.; Furlong, D. N. J. Colloid Interface Sci. 1996, 178, 104−115. (24) Derwinska, K.; Sauer, U.; Preininger, C. Talanta 2008, 77, 652− 658. (25) Tajima, N.; Takai, M.; Ishihara, K. Anal. Chem. 2011, 83, 1969. (26) Vashist, S. K.; Dixit, C. K.; MacCraith, B. D.; O’Kennedy, R. Analyst 2011, 136, 4431−4436. Kausaite-Minkstimiene, A.; Ramanaviciene, A.; Kirlyte, J.; Ramanavicius, A. Anal. Chem. 2010, 82, 6401−6408. (27) Yuan, X.; Fabregat, D.; Yoshimoto, K.; Nagasaki, Y. Colloids Surf., B 2012, 99, 45−52. (28) Song, H. Y.; Zhou, X.; Hobley, J.; Su, X. Langmuir 2011, 28, 997−1004. (29) Wang, D.-S.; Chang, C.-C.; Shih, S.-C.; Lin, C.-W. Biomed. Eng.: Appl., Basis Commun. 2009, 21, 303−310. (30) Zhao, X.; Pan, F.; Garcia-Gancedo, L.; Flewitt, A. J.; Ashley, G. M.; Luo, J.; Lu, J. R. J. R. Soc., Interface 2012, 9, 2457−2467. (31) Pei, Z.; Anderson, H.; Myrskog, A.; Duner, G.; Ingermasson, B.; Aastrup, T. Anal. Biochem. 2010, 398, 161−168. (32) Saha, B.; Evers, T. H.; Prins, M. W. J. Anal. Chem. 2014, 86, 8158−8166. (33) Reyburn, H.; Mbakilwa, H.; Mwangi, R.; Mwerinde, M.; Olomi, R.; Drakeley, C.; Whitty, C. J. M. Br. Med. J. 2007, 334, 403−406A. (34) Hartwell, L.; Mankoff, D.; Paulovich, A.; Ramsey, S.; Swisher, E. Nat. Biotechnol. 2006, 24, 905−908. Etzioni, R.; Urban, N.; Ramsey, S.; McIntosh, M.; Schwartz, S.; Reid, B.; Radich, J.; Anderson, G.; Hartwell, L. Nat. Rev. Cancer 2003, 3, 243−252. (35) Mabey, D.; Peeling, R. W.; Ustianowski, A.; Perkins, M. D. Nat. Rev. Microbiol. 2004, 2, 231−240. (36) World Malaria Report, 2011; World Health Organization, 2011. www.who.int/malaria/world_malaria_report_2011/en/. (37) Jenkins, D.; Corrie, S.; Flaim, C.; Kendall, M. A. F. RSC Adv. 2012, 2, 3490−3495. (38) Sam, S.; Touahir, L.; Andresa, J. S.; Allongue, P.; Chazalviel, J. N.; Gouget-Laemmel, A. C.; de Villeneuve, C. H.; Moraillon, A.; Ozanam, F.; Gabouze, N.; Djebbar, S. Langmuir 2010, 26, 809−814. (39) Kifude, C. M.; Rajasekariah, H. G.; Sullivan, D. J.; Stewart, V. A.; Angov, E.; Martin, S. K.; Diggs, C. L.; Waitumbi, J. N. Clin. Vaccine Immunol. 2008, 15, 1012−1018. Lee, N.; Baker, J.; Andrews, K. T.; Gatton, M. L.; Bell, D.; Cheng, Q.; McCarthy, J. J. Clin. Microbiol. 2006, 44, 2773−2778. (40) Aldeek, F.; Safi, M.; Zhan, N. Q.; Palui, G.; Mattoussi, H. ACS Nano 2013, 7, 10197−10210. Zhang, Y.; Hu, Q.; Paau, M. C.; Xie, S. P.; Gao, P. F.; Chan, W.; Choi, M. M. F. J. Phys. Chem. C 2013, 117, 18697−18708. (41) Chan, Y. H. M.; Schweiss, R.; Werner, C.; Grunze, M. Langmuir 2003, 19, 7380−7385. Dicke, C.; Hahner, G. J. Am. Chem. Soc. 2002, 124, 12619−12625. (42) Kool, J.; Reubsaet, L.; Wesseldijk, F.; Maravilha, R. T.; Pinkse, M. W.; D’Santos, C. S.; van Hilten, J. J.; Zijlstra, F. J.; Heck, A. J. R. Proteomics 2007, 7, 3638−3650. Macdonald, N.; Cumberbatch, M.; Singh, M.; Moggs, J. G.; Orphanides, G.; Dearman, R. J.; Griffiths, C. E. M.; Kimber, I. Clin. Exp. Dermatol. 2006, 31, 445−448. (43) Boyne, M. S.; Silver, D. M.; Kaplan, J.; Saudek, C. D. Diabetes 2003, 52, 2790−2794. (44) Staberg, B.; Groth, S.; Rossing, N. Clin. Physiol. 1983, 3, 375− 380. (45) Corrie, S. R.; Fernando, G. J. P.; Crichton, M. L.; Brunck, M. E. G.; Anderson, C. D.; Kendall, M. A. F. Lab Chip 2010, 10, 2655.

10483

dx.doi.org/10.1021/ac5031682 | Anal. Chem. 2014, 86, 10474−10483