Prussian Blue Nanoparticles as a Catalytic Label in a Sandwich

Jan 9, 2018 - Enzyme immunoassays are widely used for detection of analytes within various samples. However, enzymes as labels suffer several disadvan...
0 downloads 5 Views 2MB Size
Article pubs.acs.org/ac

Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

Prussian Blue Nanoparticles as a Catalytic Label in a Sandwich Nanozyme-Linked Immunosorbent Assay Zdeněk Farka,*,†,∥ Veronika Č underlová,†,‡,∥ Veronika Horácǩ ová,† Matěj Pastucha,†,‡ Zuzana Mikušová,‡ Antonín Hlavácě k,†,§ and Petr Skládal*,†,‡ †

Central European Institute of Technology and ‡Department of Biochemistry, Faculty of Science, Masaryk University, Kamenice 5, 625 00 Brno, Czech Republic § Institute of Analytical Chemistry, Czech Academy of Sciences, 602 00 Brno, Czech Republic S Supporting Information *

ABSTRACT: Enzyme immunoassays are widely used for detection of analytes within various samples. However, enzymes as labels suffer several disadvantages such as high production cost and limited stability. Catalytic nanoparticles (nanozymes) can be used as an alternative label in immunoassays overcoming the inherent disadvantages of enzymes. Prussian blue nanoparticles (PBNPs) are nanozymes composed of the Fe4[Fe(CN)6]3-based coordination polymer. They reveal peroxidase-like activity and are capable of catalyzing the oxidation of colorless 3,3′,5,5′-tetramethylbenzidine in the presence of H2O2 to form intensely blue product. Here, we introduce the method for conjugation of PBNPs with antibodies and their application in nanozyme-linked immunosorbent assay (NLISA). Sandwich NLISA for detection of human serum albumin in urine was developed with limit of detection (LOD) of 1.2 ng·mL−1 and working range up to 1 μg·mL−1. Furthermore, the microbial contamination of Salmonella Typhimurium in powdered milk was detected with LOD of 6 × 103 colony-forming units (cfu)·mL−1 and working range up to 106 cfu·mL−1. In both cases, a critical comparison with the same immunoassay but using native peroxidase as label was realized. The achieved results confirmed the suitability of PBNPs for universal and robust replacement of enzyme labels.

E

nanoparticles (Fe3O4) of different sizes were characterized for their activity, high turnover number, and good stability.10 Later on, peroxidase-like activity was described for CeO2,11 CuO,12 Co3O4,13 bimetallic,14 and MnO2 nanoparticles,15 graphene oxide nanoplates,16 mesoporous silica nanoparticles coated by catalytic metal complexes,7 and hemin block copolymer micelles.17 Coordination polymers18 are solids, which have been studied recently for their exceptional catalytic properties. Prussian blue nanoparticles (PBNPs) 19 are nanozymes composed of coordination polymer with chemical formula Fe4[Fe(CN)6]3 that reveal peroxidase-like activity.20,21 Previously, we reported the biotinylation scheme of PBNPs allowing for competitive nanozyme-linked immunosorbent assay (NLISA) of protein markers.22 However, competitive immunoassays typically possess lower sensitivity than sandwich variants.23 Here, we introduce a method for direct conjugation of PBNPs with antibodies. The nanomaterial is characterized by transmission electron microscopy (TEM), atomic force microscopy (AFM), and dynamic light scattering (DLS). Two sandwich NLISAs were developed, for the detection of human serum albumin

nzyme-linked immunosorbent assay (ELISA) is considered the gold standard for highly sensitive qualitative and quantitative detection of analytes within various samples. In ELISA, an enzyme is conjugated with either antibody or antigen to be used as a signal amplification label, which catalyzes the conversion of substrate molecules to easily detectable product. For example, the conversion of a colorless substrate to a colored or fluorescent product may be applied.1 The most commonly used enzymatic labels for immunochemical detection and imaging are peroxidases, such as horseradish peroxidase (HRP). However, high production cost, short shelf life, and decrease of activity during conjugation reactions of natural enzymes create an increasing demand for cheap and more stable nanozymes.2,3 Nanozymes are nanomaterials mimicking catalytic activity of enzymes, such as peroxidases.4−6 Catalytically active nanozymes with high signal amplification and good stability generally improve the performance of immunoassays7 and have excellent potential for the detection of single molecules in digital (immuno)assays.8 Nanozymes can be commonly synthesized from benign precursors in aqueous dispersants.9 The applicability of nanozymes for practical use is also promoted by thermal and chemical stability of inorganic nanomaterials.10 Nanozymes with high peroxidase-like activity were studied for their potential use in immunochemical assays.3−6 Initially, magnetite © XXXX American Chemical Society

Received: November 24, 2017 Accepted: January 9, 2018 Published: January 9, 2018 A

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Comfort, Eppendorf). Subsequently, the oxidized pAb-F was purified with a Microcon centrifugal unit YM-10 [10K molecular weight cutoff (MWCO); Merck Millipore]. The oxidized antibody was diluted in 10 mM PBS to a concentration of 0.2 mg·mL−1, mixed with 16.5 nM PBNPdBSA conjugate solution at a volume ratio of 1:1, and incubated at room temperature for 2 h, followed by incubation at 4 °C overnight. The formed Schiff base was reduced by addition of 50 μL of 1 M NaBH3CN per 1 mL of solution and the mixture was incubated for 4 h at room temperature. The free reactive groups were blocked by addition of 50 μL of 1 M Tris-HCl per 1 mL of solution, and this was mixed at room temperature for 30 min. The final PBNP-Ab conjugates were stored at 4 °C. The synthesis of PBNP−Ab conjugates is summarized in Figure 1.

(HSA) in urine and the detection of bacterial pathogen Salmonella Typhimurium in milk, and their performance was critically compared with the corresponding classic ELISA format using similar conjugates based on the natural HRP. In comparison with our previous reported competitive NLISA for HSA, the limit of detection was improved by 3 orders of magnitude.22 This demonstrates that antibody-conjugated PBNPs can be used as an enzyme replacement and provide a universal detection system for sandwich immunoassays.



MATERIALS AND METHODS Chemicals and Reagents. Bovine γ-globulin (BGG), bovine serum albumin (BSA), HSA, HRP, iron(III) chloride hexahydrate, poly-L-lysine hydrobromide (30−70 kDa), sodium azide, sodium borohydride, sodium cyanoborohydride, sodium periodate, 3,3′,5,5′- tetramethylbenzidine (TMB), Tris, and Tween 20 were obtained from Sigma−Aldrich. Ethylenediaminetetraacetic acid (EDTA), glucose, and potassium ferrocyanide trihydrate were purchased from Lachema. Poly(vinyl alcohol) (6 kDa) was purchased from Polysciences. All other common chemicals were obtained in the highest quality available from Penta. Phosphate buffer (PB; 50 mM NaH2PO4/Na2HPO4, pH 7.4), phosphate-buffered saline (PBS; 50 mM NaH2PO4/ Na2HPO4, pH 7.4, containing 150 mM NaCl), washing buffer (50 mM NaH2PO4/Na2HPO4, pH 7.4, 0.01% Tween 20, and 0.05% NaN3), and assay buffer (0.2% BSA, 0.5% BGG, 50 mM Tris, 150 mM NaCl, 5 mM EDTA, 0.2% PVA (w/w), 1% glucose, 0.05% NaN3, and 0.01% Tween 20, pH 7.5) were used throughout this work. For the detection of HSA, mouse monoclonal antibody (mAb) clone AL-01 (Exbio) and fluorescein isothiocyanate (FITC)-labeled swine polyclonal antibody (pAb-F; Sevapharma) were used. The detection of Salmonella was based on mouse mAb ab8274 (Abcam) and rabbit pAb 8209-4006 (AbD Serotec). Preparation of Antibody-Modified Prussian Blue Nanoparticles. PBNPs and denatured bovine serum albumin (dBSA) were prepared according to our previous work.22 Briefly, PBNPs were synthesized by mixing 20 mL of aqueous solution of 1 mM K4[Fe(CN)6] and 25 mM citric acid with 20 mL of 1 mM FeCl3 and 25 mM citric acid, which was preheated to 55 °C. The resulting blue PBNP dispersion was stirred for 10 min and cooled to room temperature. The molar concentration of the synthesized PBNPs was 33 nM; this was based on the amounts of Fe2+ and Fe3+ ions used in the synthesis and the average nanoparticle size determined by TEM. For preparation of dBSA, 10 mg of BSA was reduced by addition of 0.7 mg of NaBH4 in 3.1 mL of water and the mixture was shaken for 60 min at room temperature. In order to decompose excess NaBH4, the mixture was heated to 80 °C for 30 min until bubbles of H2 were no longer formed. Then, prepared dBSA was mixed in a ratio of 10:1 with phosphate buffer (0.5 M, pH 8.0) and stored at −20 °C. The dBSAmodified PBNPs (PBNP-dBSA) were prepared by mixing PBNPs (33 nM) and dBSA (0.36 mg·mL−1) at a volume ratio of 1:1 with subsequent heating to 70 °C for 5 min. To prepare antibody-modified PBNPs, pAb-F (or alternatively pAb Serotec in the case of Salmonella assay) was diluted in 10 mM PBS to a concentration of 3 mg·mL−1. The antibody was oxidized by addition of 100 μL of 0.1 M NaIO4 per 1 mL of antibody solution, and the mixture was incubated at room temperature for 30 min under shaking (Thermomixer

Figure 1. Scheme of PBNP−Ab conjugate synthesis. The PBNPs are modified by denatured BSA to introduce amino groups on their surface, and the oxidized antibody is conjugated via the aldehyde groups.

Characterization of Nanoparticles. Samples of PBNPs and their conjugates were diluted 10-fold with PBS, and hydrodynamic diameters were estimated from dynamic light scattering utilizing a Zetasizer Nano ZS instrument (Malvern). For characterization of PBNPs by AFM, mica (grade V-1 muscovite, SPI Supplies) was cleaned with adhesive tape and impurities on the surface were removed by compressed air. A droplet of poly-L-lysine (20 μL of 10 μg·mL−1 solution in 0.1 M PB, pH 6.9) was deposited on mica for 20 min at room temperature, and then the mica was washed with distilled water and dried. The sample of PBNPs (15 μL) was incubated on the mica surface for 20 min, and the mica was washed and scanned in dry state on a Dimension Fast Scan AFM with Fast Scan-A probe (Bruker); the images were evaluated by use of Gwyddion software. For TEM characterization, the dispersion of nanomaterial (∼4 μL) was deposited onto 400-mesh copper EM grids coated with a continuous carbon layer. Dried grids were imaged by TEM (Tecnai F20, FEI). The dimensions of individual particles were analyzed by use of ImageJ software (National Institutes of Health). Detection of Human Serum Albumin. A sandwich assay based on antibody-modified PBNPs as a label was developed for detection of human serum albumin (HSA) in urine (Figure 2). Due to the presence of HSA in urine from healthy patients,24,25 the urine was first deproteinized by use of the Microcon centrifugal unit YM-10 and then spiked with known B

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

incubated overnight at 4 °C. All following steps were carried out at room temperature; after each step, the plate was washed four times with 200 μL of washing buffer. The plate was blocked with 200 μL of 5% powdered milk (Laktino) in PBS for 60 min. Then, either standard dilutions of HSA in assay buffer or spiked and 10× diluted urine samples (1 ng·mL−1 to 100 μg·mL−1 HSA) were added (100 μL/well), and the plates were incubated for 60 min. Subsequently, PBNP−pAb-F conjugates in the assay buffer were added (750 pM, 100 μL/ well) and the plates were incubated for 60 min. Finally, the color was developed by addition of freshly prepared substrate solution (0.5 mM TMB and 125 mM H2O2 in 200 mM sodium acetate, pH 3.5, 100 μL/well), and absorbance at 652 nm (A652) was measured after 120 min on a Synergy 2 reader (BioTek). Alternatively, the reaction was stopped by addition of 100 μL of 1 M H2SO4 and the absorbance was read at 450 nm (A450). Furthermore, the classic sandwich ELISA was developed and optimized to provide a comparison of NLISA with the standard approach. The conjugation of detection pAb-F with HRP was done according to our previous work.26 The assay protocol remained the same as in the case of NLISA except for the detection conjugate binding and biocatalytic step. HRP−pAb-F conjugate (100 μL; concentration equivalent to 2.5 μg·mL−1 antibody) was allowed to bind for 60 min, followed by addition of commercial TMB substrate solution (Enzymo Plus). The color change was read after 30 min as A652; end-point measurement at 450 nm was carried out as well after addition of 100 μL of 1 M H2SO4. Detection of Salmonella Typhimurium. Salmonella enterica subsp. enterica serovar Typhimurium (ATCC 14028) was obtained from the Czech Collection of Microorganisms (CCM). The previously optimized sample preparation protocol that yields Salmonella with high affinity to the used antibodies was employed.27 The stock solution of Salmonella (100 μL) was

Figure 2. Scheme of PBNP-based sandwich NLISA. The microtiter plate is coated by capture antibody (Ab1), followed by specific binding of antigen (Ag) and formation of sandwich with the detection PBNP− Ab2 label. The colorimetric readout is based on PBNP-catalyzed oxidation of TMB in the presence of H2O2.

HSA concentrations. Nunc MaxiSorp 96-well microtiter plates (Thermo Fisher Scientific) were coated with monoclonal antiHSA antibody AL-01 (4 μg·mL−1 in PBS, 100 μL/well) and

Figure 3. (A, B) AFM images of (A) PBNPs (average diameter 22.1 ± 2.9 nm) and (B) PBNP-dBSA−Ab (116 ± 29 nm) conjugates. (C) TEM image of PBNPs (15.3 ± 3.4 nm). (D) DLS of PBNPs (hydrodynamic diameter ⌀HD 24 nm); PBNP-dBSA (⌀HD 92 nm), and PBNP-dBSA−Ab (⌀HD 128 nm). C

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 4. Optimization of NLISA for HSA detection. (A) Optimization of coating concentration of primary anti-HSA mAb. (B) Different concentrations of PBNP−Ab conjugate are used to generate the signal for a plate coated by 1 μg·mL−1 capture mAb. (C) Optimization of substrate concentration. Open triangles represent LOD values.



RESULTS AND DISCUSSION Characterization of Prussian Blue Nanoparticles and Their Conjugates. For the first time, we describe the direct covalent conjugation of PBNPs with antibodies. The employed synthetic method provided PBNPs with cubic shape; the nanoparticles were characterized by TEM, AFM, and DLS. The average size estimated by AFM was 22.1 ± 2.9 nm (Figure 3A), the size determined by TEM was 15.3 ± 3.4 nm (Figure 3C), and DLS revealed hydrodynamic diameter of 24 nm with polydispersity index 0.213 (Figure 3D). The surface of PBNPs was modified with reductively denatured bovine serum albumin and as we reported previously; the attachment of dBSA was accompanied by slight aggregation of PBNPs.22 The size of these aggregates estimated by AFM was 107 ± 20 nm; DLS provided a hydrodynamic diameter of 92 nm with polydispersity index of 0.222. Then prepared PBNP−dBSA was conjugated with antibodies by a reductive amination method,28 resulting in the PBNP-Ab tracer. The size of the final PBNP− Ab conjugate provided by AFM was 116 ± 29 nm (Figure 3B); the hydrodynamic diameter was 128 nm with polydispersity index 0.232. Design of Nanozyme-Linked Immunosorbent Assay. Synthesized conjugates of PBNPs with anti-HSA antibody were employed in the sandwich immunoassay for HSA. Due to the simple conjugation and shorter analysis time (fewer incubation steps), PBNPs conjugated with detection antibody were used as tracer directly instead of the universal biotin-labeling protocol, which is also possible but prolongs the assay.22 The assay procedure was based on two anti-HSA antibodies: monoclonal AL-01 and polyclonal pAb-F. As expected, the most efficient binding was observed when the mAb was used for analyte capture and pAb as the label conjugated on the PBNPs (Figure S1). This arrangement allows oriented binding of the analyte on the surface and reduces probability that the label could not bind to the analyte due to inaccessible binding sites. Furthermore, binding of HSA via the capture antibody enhanced its accessibility for the PBNP−Ab tracer, which resulted in higher signals in sandwich compared to the direct assay with HSA coated on the microtiter plate surface. The coating concentration and blocking conditions were optimized in order to achieve high sensitivity and wide working range. Unlike the competitive assays where the coating concentration needs to be fine-tuned to achieve a compromise between efficient competition and measured signal intensity,29 the sandwich format generally provides higher responses for higher coating concentrations. In agreement with theory, the measured signals increased with increasing coating concen-

inoculated into 25 mL of low-salt Lysogeny broth (LB) broth (Duchefa Biochemie) in an Erlenmeyer flask, which was incubated aerobically overnight (37 °C, mild shaking). The bacterial suspension was centrifuged twice for 10 min at 6800g and resuspended in PBS. The microorganism samples were treated by heat at 80 °C for 30 min with mild shaking (900 rpm) in a Thermomixer Comfort instrument. Selected samples were also homogenized by use of a Q700 sonicator (Qsonica). The 1.6 mm Microtip probe was placed into a microtube with 1 mL of Salmonella sample, and amplitude of 70% was maintained for approximately 10 min until the total energy consumed by the sample reached 1 kJ. The concentrations of thus-treated bacteria are expressed as colony-forming units per milliliter (cfu·mL−1) corresponding to the initial amount of viable cells. The real samples were prepared by dissolving powdered milk in assay buffer to a concentration of 1% and spiking it with varying Salmonella concentrations. For Salmonella detection, the surface of the microtiter plate was coated by monoclonal anti-Salmonella antibody Abcam ab8274 (2 μg·mL−1), and the detection label consisted of PBNPs modified by polyclonal antibody Serotec 8209-4006. Other assay steps remained the same as in the case of HSA detection. The reference ELISA experiments were performed utilizing the same experimental conditions; the only difference was use of the conjugate of HRP with polyclonal antibody Serotec at a concentration of 2.5 μg·mL−1. Data Analysis. For each concentration, the mean and standard deviation were calculated from three replicate wells. A four-parameter logistic function was used for regression analysis of the calibration curves: y=

ymax − ybg 1+

s

( ) c EC50

+ ybg

where c is the concentration of either HSA or Salmonella and y is the measured absorbance at 652 nm (A652). The equation yields as parameters the maximum (ymax) and background (ybg) signals, the concentration that reduced (ymax − ybg) by 50% (EC50), and the slope at the inflection point (s). The limit of detection (LOD) was calculated from the regression curve as the concentration corresponding to the yLOD value: yLOD = ybg + 3s0

where s0 is the standard deviation of blank measurement (no analyte present in the sample). D

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 5. (A) Calibration curve of sandwich NLISA for detection of HSA in buffer (black; LOD 1.3 ng·mL−1, EC50 57 ng·mL−1) and spiked urine (red; LOD 1.2 ng·mL−1, EC50 43 ng·mL−1) under optimized conditions (4 μg·mL−1 coating, 750 pM PBNP−Ab). (B) Calibration curve of standard ELISA for HSA detection in buffer (black; LOD 5.3 ng·mL−1, EC50 142 ng·mL−1) and spiked urine (red; LOD 3.7 ng·mL−1, EC50 152 ng·mL−1). Open triangles represent LOD values.

values were observed (Figure S5), which confirms the high reliability of both conjugate preparation and the assay itself; the accuracy of results is further ensured by measuring the calibration curve as a part of each microtiter plate with unknown samples. The ability to analyze real samples was demonstrated by detection of HSA in human urine. Because of the high sensitivity of the developed NLISA and trace amounts of HSA in urine from healthy patients, HSA was first removed from the urine sample by filtration through a 10 kDa centrifugal microfilter, followed by spiking of known HSA concentrations. The calibration curve comparing HSA detection in buffer and urine is shown in Figure 5A. Possible interference from various ions present in urine, which might potentially catalyze substrate conversion, is suppressed by the washing step; therefore, no large difference in the response was observed when HSA was analyzed in buffer and in the urine samples. To compare the capabilities of PBNP- and enzyme-based labels, ELISA based on the same combination of antibodies was developed, optimized (Figure S6), and employed for analysis of HSA in spiked urine (Figure 5B). The LOD of NLISA (1.2 ng· mL−1) was slightly better than in the case of ELISA (3.7 ng· mL−1). The similar results suggest that the assay is limited mainly by the affinity of the selected antibodies for the target analyte and not by the detection step. However, the catalytic nanoparticles provide several practical advantages compared to the enzymatic label, such as simple and cheap preparation, higher stability, and possibility to function at much higher substrate concentrations. The PBNPs alone provide practically unlimited shelf life (several years at 4 °C); the stability-limiting element is the antibody as the biorecognition molecule. Further enhancements in label stability could be achieved by functionalizing the PBNPs with artificial aptamers or molecularly imprinted polymers. A detailed comparison of NLISA and ELISA with respect to properties and preparation of catalytic label and assay performance is shown in Table S1. The overall analysis time was 4 h starting from the coated and blocked plate. The concentration range typical for microalbuminuria, which is a diagnostic hallmark for diabetic nephropathy, is from 20 to 200 μg·mL−1 in the 24 h specimens. The optimized assay provided LOD of 1.2 ng·mL−1 for HSA in urine, with a working range up to 1 μg·mL−1. Therefore, the parameters of developed NLISA are sufficient to discriminate positive microalbuminuria samples.30 The achieved detection limit is better than values reported for competitive ELISA (LOD 89 ng·mL−1),31 competitive fluoroimmunoassays (LOD

tration (Figure 4A). However, the background also increased proportionally, which led to a constant signal-to-background ratio of 3−4:1 that did not depend significantly on the coating level. The coating from the higher part of the tested scale (4 μg· mL−1) was selected since the higher signals in general allow readout less prone to instrumental and pipetting errors. This leads to smaller standard deviations and hence better LOD values. The effects of different blocking conditions on assay performance are shown in Figure S2A. Compared to blocking with pure BSA (10 mg·mL−1 in PBS), the more complex solutions of milk (50% in PBS) and powdered milk (5% in PBS) provided lower degrees of nonspecific binding. Powdered milk was selected due to slightly lower background and expected higher reproducibility. During preparation, centrifugation was used to conveniently separate PBNP−Ab conjugates from the unbound antibody that might potentially compete with the conjugate for binding on HSA. The comparison of centrifuged and uncentrifuged conjugates of two PBNP−Ab samples (based on two different antibody oxidation times) is shown in Figure S2B. For both conjugates, significant signal decrease was observed after centrifugation. Even though high centrifugation speed (14000g) and long centrifugation time (30 min) were used, a substantial amount of PBNPs remained in the supernatant, while the sedimented particles were rather difficult to resuspend. On the other hand, no significant interference from the unseparated free antibody was observed in the case of uncentrifuged particles; therefore, this sample was used for further immunoassays. The concentration of detection PBNP− Ab conjugate substantially influenced the overall measured signals (Figure 4B); however, it did not affect the signal-tobackground ratio, as there were more particles that could bind nonspecifically to the surface. Unlike enzymes, the nanoparticle labels catalyze conversions of high concentrations of substrate for longer times; such conditions usually cause denaturation of natural biocatalysts. The optimization of substrate concentration is shown in Figure 4C, and the time dependence of product formation is shown in Figure S3. The readout after 120 min was selected as a compromise between the overall measured signal and assay time requirements. Stopping the reaction by H2SO4 provided higher absorbance values; however, it did not affect the LOD values as the errors were increased proportionally (Figure S4). To evaluate reproducibility, the assay was performed multiple times with different batches of PBNP−Ab conjugate. Only small variations in the LOD (1.3−4.0 ng·mL−1) and EC50 (28.3−77.3 ng·mL−1) E

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

Figure 6. (A) Calibration curve for NLISA-based detection of Salmonella Typhimurium in buffer (black; LOD 2 × 103 cfu·mL−1, EC50 4 × 104 cfu· mL−1) and spiked powdered milk (red; LOD 6 × 103 cfu·mL−1, EC50 5 × 104 cfu·mL−1). (B) Calibration curve for ELISA-based detection of Salmonella Typhimurium in buffer (black; LOD 3 × 103 cfu·mL−1, EC50 9 × 104 cfu·mL−1) and spiked powdered milk (red; LOD 3 × 103 cfu·mL−1, EC50 1 × 105 cfu·mL−1). Open triangles represent LOD values.

1 μg·mL−1),32 competitive NLISA (LOD 0.27 μg·mL−1),22 method based on specific fluorescence probe (LOD 0.4 μg· mL−1),33 immunonephelometry (LOD 2 μg·mL−1),34 immunoturbidimetry (LOD 6 μg·mL−1),34 and the semiquantitative dipstick method (LOD 10 μg·mL−1).35 The developed sandwich NLISA provided LOD comparable to radioimmunoassay (LOD 16 ng·mL−1)34 and sandwich ELISA (LOD 6.25 ng·mL−1),36 as summarized in Table S2. Detection of Salmonella Typhimurium. To demonstrate the universal applicability of PBNPs as labels in immunoassays, Salmonella Typhimurium was detected as a representative example of a large microbial antigen. The assay was optimized on the basis of findings obtained during the detection of HSA; the optimization of coating concentration is shown in Figure S7A. Also in this case, the coating concentration influenced the total measured signals but did not have an effect on the signalto-background ratio. It was previously shown that the selected antibodies exhibit higher affinity toward heat-treated samples compared to native Salmonella,27 therefore, heat treatment and sonication were tested for sample preparation (Figure S7B). The sonicated Salmonella provided a wider working range compared to the heat-treated sample; however, similar LOD values were achieved. Heat treatment is preferable for sample preparation due to lower requirements for equipment and time. Detection in powdered milk (diluted by assay buffer) spiked with the microbe was carried out to demonstrate the ability of NLISA to detect Salmonella in real samples (Figure 6A). The presence of milk matrix during the Salmonella binding step served to further block the surface and allowed us to decrease the level of nonspecific binding with minimal effect on the LOD. The optimized assay provided LOD of 6 × 103 cfu·mL−1 in powdered milk, with a working range up to 106 cfu·mL−1. The obtained results are better than values for traditional ELISA that provide typical LODs in the range between 104 and 106 cfu·mL−1.37−39 The sandwich ELISA developed here, utilizing the same combination of capture and detection antibodies, provided LOD of 3 × 103 cfu·mL−1 in powdered milk (Figure 6B). This is in agreement with the results for the detection of HSA, which suggest that the affinity of antibodies has a more significant impact on the LOD than the label used. Another assay scheme based on direct ELISA was also tested and compared (Figure S8); however, the LOD did not improve compared to the sandwich arrangement. The LOD can be further improved by a shortened cultivation-based pre-enrichment; however, such assays have higher demands on analysis time (∼1 day instead of several

hours).40 A faster way to concentrate the analyte and separate it from a complex matrix is immunomagnetic separation. This approach is often used in biosensor-based detection.41 It is not regularly used in immunoassays, though microfluidics-based ELISA utilizing magnetic beads was recently reported for the detection of C-reactive protein.42 Various nanoparticle-based approaches were recently reported to improve the properties of traditional ELISA for Salmonella detection.43 One of the most popular ways is to improve the catalytic properties of the label. For example, conjugation of detection antibody and HRP on single-walled carbon nanotubes44 allowed researchers to reach LOD of 103 and 104 cfu·mL−1 for direct and sandwich assay, respectively. This method provided higher catalytic activities than simple conjugates of HRP with antibody; however, it was still limited by the instability of the enzyme label. PBNPs, on the other hand, not only provide comparable LOD but also can be used to catalyze higher substrate concentrations (allowing higher measured signals) and are significantly more stable (which provides longer shelf life of the label). The comparison of immunoassays for Salmonella detection without pre-enrichment steps is shown in Table S3.



CONCLUSIONS The straightforward method for conjugation of Prussian blue nanoparticles with antibodies was developed and applied in nanozyme-linked immunosorbent assay (NLISA). The assay protocol is based on the oxidation of TMB forming blue product; therefore a common colorimetric reader can be used without special requirements for instrumentation. The universal applicability of PBNPs as labels is demonstrated by the development of two sandwich immunoassays. Human serum albumin was detected in urine for the diagnosis of albuminuria. The assay provided LOD of 1.2 ng·mL−1 and a working range of 3 orders of magnitude, up to 1 μg·mL−1. NLISA was also employed for the detection of bacterial pathogen Salmonella Typhimurirum in powdered milk with LOD of 6 × 103 cfu· mL−1 and a working range up to 106 cfu·mL−1. For comparison, the same antibodies were also used in standard sandwich ELISA formats with natural HRP enzyme. Comparable assay parameters were achieved, which suggests that the assay is limited mainly by the affinity of the antibodies for the target analyte and not by the detection step. However, PBNPs provide several practical advantages compared to the enzymatic label, such as easy synthesis from cheap precursors and higher stability. The results achieved in the NLISA are also highly F

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

(13) Dong, J. L.; Song, L. N.; Yin, J. J.; He, W. W.; Wu, Y. H.; Gu, N.; Zhang, Y. ACS Appl. Mater. Interfaces 2014, 6 (3), 1959−1970. (14) He, W. W.; Liu, Y.; Yuan, J. S.; Yin, J. J.; Wu, X. C.; Hu, X. N.; Zhang, K.; Liu, J. B.; Chen, C. Y.; Ji, Y. L.; Guo, Y. T. Biomaterials 2011, 32 (4), 1139−1147. (15) Liu, X.; Wang, Q.; Zhao, H. H.; Zhang, L. C.; Su, Y. Y.; Lv, Y. Analyst 2012, 137 (19), 4552−4558. (16) Song, Y. J.; Qu, K. G.; Zhao, C.; Ren, J. S.; Qu, X. G. Adv. Mater. 2010, 22 (19), 2206−2210. (17) Qu, R.; Shen, L. L.; Chai, Z. H.; Jing, C.; Zhang, Y. F.; An, Y. L.; Shi, L. Q. ACS Appl. Mater. Interfaces 2014, 6 (21), 19207−19216. (18) Kitagawa, S.; Kitaura, R.; Noro, S. Angew. Chem., Int. Ed. 2004, 43 (18), 2334−2375. (19) Herren, F.; Fischer, P.; Ludi, A.; Halg, W. Inorg. Chem. 1980, 19 (4), 956−959. (20) Zhang, X. Q.; Gong, S. W.; Zhang, Y.; Yang, T.; Wang, C. Y.; Gu, N. J. Mater. Chem. 2010, 20 (24), 5110−5116. (21) Zhang, W. M.; Ma, D.; Du, J. X. Talanta 2014, 120, 362−367. (22) Č underlová, V.; Hlavácě k, A.; Horňaḱ ová, V.; Peterek, M.; Němeček, D.; Hampl, A.; Eyer, L.; Skládal, P. Microchim. Acta 2016, 183 (2), 651−658. (23) Jackson, T. M.; Ekins, R. P. J. Immunol. Methods 1986, 87 (1), 13−20. (24) Datta, P.; Dasgupta, A. J. Clin. Lab. Anal. 2009, 23 (5), 314− 318. (25) Tsai, J. Z.; Chen, C. J.; Settu, K.; Lin, Y. F.; Chen, C. L.; Liu, J. T. Biosens. Bioelectron. 2016, 77, 1175−1182. (26) Farka, Z.; Juřík, T.; Pastucha, M.; Skládal, P. Anal. Chem. 2016, 88 (23), 11830−11836. (27) Farka, Z.; Juřík, T.; Pastucha, M.; Kovár,̌ D.; Lacina, K.; Skládal, P. Electroanalysis 2016, 28 (8), 1803−1809. (28) Algar, W. R.; Prasuhn, D. E.; Stewart, M. H.; Jennings, T. L.; Blanco-Canosa, J. B.; Dawson, P. E.; Medintz, I. L. Bioconjugate Chem. 2011, 22 (5), 825−858. (29) Hlavácě k, A.; Farka, Z.; Hübner, M.; Horňaḱ ová, V.; Němeček, D.; Niessner, R.; Skládal, P.; Knopp, D.; Gorris, H. H. Anal. Chem. 2016, 88 (11), 6011−6017. (30) Doumas, B. T.; Peters, T. Clin. Chim. Acta 1997, 258 (1), 3−20. (31) Zhao, L. X.; Lin, J. M.; Li, Z. J. Anal. Chim. Acta 2005, 541 (1− 2), 197−205. (32) Hlavácě k, A.; Bouchal, P.; Skládal, P. Microchim. Acta 2012, 176 (3−4), 287−293. (33) Kessler, M. A.; Meinitzer, A.; Petek, W.; Wolfbeis, O. S. Clin. Chem. 1997, 43 (6), 996−1002. (34) Busby, D. E.; Bakris, G. L. J. Clin. Hypertens. 2004, 6, 8−12. (35) Marshall, S. M.; Shearing, P. A.; Alberti, K. Clin. Chem. 1992, 38 (4), 588−591. (36) Watts, G. F.; Bennett, J. E.; Rowe, D. J.; Morris, R. W.; Gatling, W.; Shaw, K. M.; Polak, A. Clin. Chem. 1986, 32 (8), 1544−1548. (37) Magliulo, M.; Simoni, P.; Guardigli, M.; Michelini, E.; Luciani, M.; Lelli, R.; Roda, A. J. Agric. Food Chem. 2007, 55 (13), 4933−4939. (38) Roda, A.; Mirasoli, M.; Roda, B.; Bonvicini, F.; Colliva, C.; Reschiglian, P. Microchim. Acta 2012, 178 (1−2), 7−28. (39) Wang, W. B.; Liu, L. Q.; Song, S. S.; Tang, L. J.; Kuang, H.; Xu, C. L. Sensors 2015, 15 (3), 5281−5292. (40) Wu, X. L.; Wang, W. B.; Liu, L. Q.; Kuang, H.; Xu, C. L. Anal. Methods 2015, 7 (21), 9047−9053. (41) Liebana, S.; Brandao, D.; Alegret, S.; Pividori, M. I. Anal. Methods 2014, 6 (22), 8858−8873. (42) Liu, D.; Li, X. R.; Zhou, J. K.; Liu, S. B.; Tian, T.; Song, Y. L.; Zhu, Z.; Zhou, L. J.; Ji, T. H.; Yang, C. Y. Biosens. Bioelectron. 2017, 96, 332−338. (43) Farka, Z.; Juřík, T.; Kovár,̌ D.; Trnková, L.; Skládal, P. Chem. Rev. 2017, 117 (15), 9973−10042. (44) Chunglok, W.; Wuragil, D. K.; Oaew, S.; Somasundrum, M.; Surareungchai, W. Biosens. Bioelectron. 2011, 26 (8), 3584−3589.

comparable with other recent nanoparticle-based immunoassays, including those based on gold nanoparticles, quantum dots, and carbon nanotubes (Table S3).



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b04883. Eight figures and three tables showing optimization of NLISA assay, reference ELISA measurements, and comparison of assay parameters with literature (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail [email protected] (Z.F.). *E-mail [email protected]; telephone +420 54949 7010 (P.S.). ORCID

Zdeněk Farka: 0000-0002-6842-7081 Petr Skládal: 0000-0002-3868-5725 Author Contributions ∥

Z.F. and V.Č . contributed equally to this work.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS We thank Miroslav Peterek for taking the TEM images. CIISB research infrastructure project LM2015043, funded by MEYS CR, is gratefully acknowledged for financial support of the measurements at CF Cryo-electron Microscopy and Tomography and CF Nanobiotechnology. This research has been financially supported by the Ministry of Education, Youth and Sports of the Czech Republic under the project CEITEC 2020 (LQ1601) and by Grant Agency of the Czech Republic (P20612G014 GACR).



REFERENCES

(1) Wild, D. The Immunoassay Handbook, 4th ed.; Elsevier: Oxford, U.K., 2013. (2) Lavery, C. B.; MacInnis, M. C.; MacDonald, M. J.; Williams, J. B.; Spencer, C. A.; Burke, A. A.; Irwin, D. J. G.; D’Cunha, G. B. J. Agric. Food Chem. 2010, 58 (15), 8471−8476. (3) Wei, H.; Wang, E. K. Chem. Soc. Rev. 2013, 42 (14), 6060−6093. (4) Kuah, E.; Toh, S.; Yee, J.; Ma, Q.; Gao, Z. Q. Chem. - Eur. J. 2016, 22 (25), 8404−8430. (5) Ragg, R.; Tahir, M. N.; Tremel, W. Eur. J. Inorg. Chem. 2016, 2016 (13−14), 1906−1915. (6) Nasir, M.; Nawaz, M. H.; Latif, U.; Yaqub, M.; Hayat, A.; Rahim, A. Microchim. Acta 2017, 184 (2), 323−342. (7) Kumari, S.; Dhar, B. B.; Panda, C.; Meena, A.; Sen Gupta, S. ACS Appl. Mater. Interfaces 2014, 6 (16), 13866−13873. (8) Walt, D. R. Anal. Chem. 2013, 85 (3), 1258−1263. (9) Shokouhimehr, M.; Soehnlen, E. S.; Khitrin, A.; Basu, S.; Huang, S. P. D. Inorg. Chem. Commun. 2010, 13 (1), 58−61. (10) Gao, L. Z.; Zhuang, J.; Nie, L.; Zhang, J. B.; Zhang, Y.; Gu, N.; Wang, T. H.; Feng, J.; Yang, D. L.; Perrett, S.; Yan, X. Nat. Nanotechnol. 2007, 2 (9), 577−583. (11) Xu, C.; Qu, X. G. NPG Asia Mater. 2014, 6, No. e90, DOI: 10.1038/am.2013.88. (12) Chen, W.; Chen, J.; Liu, A. L.; Wang, L. M.; Li, G. W.; Lin, X. H. ChemCatChem 2011, 3 (7), 1151−1154. G

DOI: 10.1021/acs.analchem.7b04883 Anal. Chem. XXXX, XXX, XXX−XXX