Competitive Upconversion-Linked Immunosorbent Assay for the

0 downloads 0 Views 979KB Size Report
May 11, 2016 - The enzyme-linked immunosorbent assay (ELISA) is a cost efficient tool for the ... been a few reports on the use of homogeneous competitive.
This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Article pubs.acs.org/ac

Competitive Upconversion-Linked Immunosorbent Assay for the Sensitive Detection of Diclofenac Antonín Hlavácě k,†,‡,§ Zdeněk Farka,†,‡ Maria Hübner,∥ Veronika Horňaḱ ová,‡ Daniel Němeček,‡ Reinhard Niessner,∥ Petr Skládal,‡ Dietmar Knopp,∥ and Hans H. Gorris*,† †

Institute of Analytical Chemistry, Chemo- and Biosensors, University of Regensburg, 93040 Regensburg, Germany CEITEC−Central European Institute of Technology, Masaryk University, Brno 625 00, Czech Republic § Institute of Analytical Chemistry AS CR, v. v. i., Brno 602 00, Czech Republic ∥ Chair of Analytical Chemistry and Institute of Hydrochemistry, Technical University of Munich, 81377 Munich, Germany ‡

S Supporting Information *

ABSTRACT: Photon-upconverting nanoparticles (UCNPs) emit light of shorter wavelength under near-infrared excitation and thus avoid optical background interference. We have exploited this unique photophysical feature to establish a sensitive competitive immunoassay for the detection of the pharmaceutical micropollutant diclofenac (DCF) in water. The so-called upconversion-linked immunosorbent assay (ULISA) was critically dependent on the design of the upconversion luminescent detection label. Silicacoated UCNPs (50 nm in diameter) exposing carboxyl groups on the surface were conjugated to a secondary anti-IgG antibody. We investigated the structure and monodispersity of the nanoconjugates in detail. Using a highly affine anti-DCF primary antibody, the optimized ULISA reached a detection limit of 0.05 ng DCF per mL. This performance came close to a conventional enzymelinked immunosorbent assay (ELISA) without the need for an enzyme-mediated signal amplification step. The ULISA was further employed for analyzing drinking and surface water samples. The results were consistent with a conventional ELISA as well as liquid chromatography−mass spectrometry (LC−MS). human chorionic gonadotropin (LOD, 3.8 ng mL−1/200 pM).12 The advantages of UCNPs in lateral flow assays, e.g., for the detection of worm parasite antigens (LOD, 0.01 ng mL−1/0.1 pM)13 have also been well documented. There have been a few reports on the use of homogeneous competitive immunoassays for the detection of small molecules such as estradiol (LOD, ∼0.1 ng mL−1/400 pM)14 and folate (LOD, 0.4 ng mL−1/1000 pM)15 in blood and a bead-based immunoassay for the detection of mycotoxins in food samples (LOD, ∼0.01 ng mL−1/50 pM).16 The development and widespread availability of more sensitive analytical techniques has resulted in an increasing number of pharmaceuticals that can be detected in the environment after medical or veterinary use.17,18 Diclofenac (2-[2-(2,6-dichlorophenyl) aminophenyl] ethanoic acid; DCF) is a widely used nonsteroidal anti-inflammatory drug (NSAID). In the Indian subcontinent, the widespread use of DCF for veterinary treatment of cattle since the 1990s has led to a precipitous decline of the indigenous vulture population because DCF leads to renal failure in vultures that feed on contaminated carcasses.19 In Europe, DCF belongs to the most

T

he enzyme-linked immunosorbent assay (ELISA) is a cost efficient tool for the specific and highly sensitive detection of many toxic analytes in food and environmental samples as well as clinical diagnosis. There are, however, some disadvantages of a classic ELISA such as an inherent instability of enzymes and time-consuming signal development. Consequently, many research efforts have been made to replace the enzymes by using nanoparticles (NPs) as signal amplifiers, e.g., fluorescent dye-doped polymer or silica NPs,1 metal NPs,2 magnetic NPs,3 catalytic NPs,4 or quantum dots.5 Recently, photon-upconverting nanoparticles (UCNPs) have been used as a new generation of luminescent labels for sensitive immunochemical detection. UCNPs are lanthanide-doped nanocrystals that can be excited by near-infrared light and emit light of shorter wavelengths (anti-Stokes emission),6,7 which strongly reduces autofluorescence and light scattering. Further advantages of UCNPs include (A) a very high photostability, (B) large anti-Stokes shifts allowing for an excellent separation of excitation and detection channels, and (C) multiple and narrow emission bands that can be tuned individually for the multiplexed detection of analytes.8−10 These distinct photophysical features of UCNPs have been used for the design of heterogeneous microtiter plate immunoassays, e.g., for the detection of prostate-specific antigen (limit of detection, LOD, 0.15 ng mL−1/6 pM)11 or © 2016 American Chemical Society

Received: March 18, 2016 Accepted: May 11, 2016 Published: May 11, 2016 6011

DOI: 10.1021/acs.analchem.6b01083 Anal. Chem. 2016, 88, 6011−6017

Article

Analytical Chemistry

reverse osmosis with UV treatment (Milli-RO 5 Plus, MilliQ185 Plus, Eschborn, Germany). Synthesis of Carboxyl-Silica-Coated UCNPs. UCNPs of 42.5 ± 4.9 nm in diameter were synthesized by hightemperature coprecipitation27 as described in the Supporting Information. The mass concentration was determined by gravimetric analysis and a concentration of 1.0 mg mL−1 of UCNPs was estimated to be equivalent to the molar concentration of 9.8 × 10−9 mol L−1 (Supporting Information). Carboxyl-silica-coated UCNPs (COOH-UCNPs) were prepared by a reverse microemulsion method:25 UCNPs (80 mg) were diluted in cyclohexane to a final volume of 23 mL. This dispersion was mixed with 1800 mg of Igepal CO-520 and 100 μL of tetraethyl orthosilicate (TEOS) and stirred intensively for 10 min. A mixture of 55 μL 25% (w/v) of aqueous ammonium hydroxide and 55 μL of water was added to form a microemulsion that was slowly stirred overnight. Then, 25 μL of TEOS were added and the microemulsion was again stirred for 180 min. After adding 50 μL of 25% (w/v) sodium carboxyethylsilanetriol in water, the microemulsion was first sonicated for 15 min and then stirred for 60 min. The COOHUCNPs were extracted with 1000 μL of dimethylformamide and washed four times with 20 mL of propan-2-ol, three times with 5000 μL of water and finally dispersed in water to yield a final concentration of 150 mg mL−1. The COOH-UCNPs in water were stable at 4 °C for several months. Conjugation of COOH-UCNP and Secondary Antibody. COOH-UCNPs were first activated by 1-ethyl-3-(3(dimethylamino)propyl)carbodiimide (EDC) and N-hydroxysulfosuccinimide (sulfo-NHS). In a typical synthesis, 0.5 mg (∼5 pmol) of COOH-UCNPs was dispersed in water to a final volume of 200 μL. The volume of 50 μL of a mixture containing 2.1 μmol of EDC and 5.5 μmol of sulfo-NHS in 100 μL of 100 mM sodium 2-(N-morpholino)ethanesulfonate (MES) buffer, pH 6.1 was added and mixed for 30 min. A dispersion of 100 μL of activated COOH-UCNPs (1 mg mL−1 or ∼10 nmol L−1) were mixed with 100 μL of horse anti-mouse IgG in borate buffer (100 mM sodium borate, pH 9.0). Three IgG concentrations were employed and incubated for 90 min at room temperature: (1) 330 nmol L−1 IgG resulting in a ratio 33 IgG molecules per UCNP (sample IgG-UCNP-33:1), (2) 67 nmol L−1 IgG resulting in a ratio of 7 IgG molecules per UCNP (IgG-UCNP-7:1), and (3) 33 nmol L−1 IgG resulting in a ratio of 3 IgG molecules per UCNP (IgG-UCNP-3:1). The bioconjugates were centrifuged for 10 min at 4 000g, dispersed in UCNP assay buffer (50 mM Tris, 150 mM NaCl, 0.05% NaN3, 0.01% Tween 20, 0.05% bovine γ-globulin (BGG), 0.5% bovine serum albumin (BSA), 0.2% poly(vinyl alcohol) 6000 (PVA), pH 7.75) and sonicated for 5 min. Nanoparticle Characterization. Transmission electron microscopy (TEM) was performed on a Tecnai F20 FEI instrument (Eindhoven, The Netherlands). About 4 μL of UCNPs were deposited on a 400-mesh copper EM grid coated with a continuous carbon layer and negatively stained with 2% (w/v) aqueous solution of uranyl acetate to increase the contrast of the silica shell. The dried grids were then imaged at 50 000× magnification (2.21 Å pixel−1). The size of individual particles in the TEM images was measured by the imaging software ImageJ (http://imagej.nih.gov).28 The hydrodynamic diameter and zeta potential of UCNP suspensions were determined on a Zetasizer Nano SZ from Malvern Instruments (Malvern, U.K.). FT-IR spectra were recorded on an Alpha FTIR spectrometer from Bruker (Billerica, MA).

frequently detected pharmaceuticals in the water-cycle because it is not easily degraded when passing through sewage treatment plants. DCF has been detected in low μg L−1 amounts in wastewater effluents and also in ng L−1 amounts in surface waters,20 groundwater, and drinking water.21 Very low amounts of DCF can be detected by liquid chromatography−time-of-flight-mass spectrometry (LC−TOF-MS) or high-resolution mass spectrometers.22,23 These instrumental techniques, however, are expensive, time-consuming, labor intensive and need trained personnel. By contrast, immunoassays are more suitable for on-site testing directly in the field or for the analysis of large numbers of samples in small laboratories.24 Here, we have optimized the preparation of monodisperse and stable upconversion reporters for the sensitive detection of DCF in water samples by a competitive upconversion-linked immunosorbent assay (ULISA). Anti-mouse IgG antibodies were conjugated to silica-coated UCNPs exposing carboxyl function on the surface and the conjugates were characterized by gel electrophoresis.25 The competitive detection of DCF was performed using a monoclonal mouse anti-DCF antibody (Figure 1). This antibody was characterized in detail as

Figure 1. Scheme of the indirect competitive ULISA for the detection of diclofenac (DCF). (A) A microtiter plate is coated with a bovine serum albumin-DCF conjugate (BSA-DCF). (B) Dilution series of DCF are prepared in the microtiter plate followed by the addition of anti-DCF mouse antibody. (C) The attachment of anti-DCF antibody is then detected by an anti-mouse IgG-UCNP secondary antibody conjugate. The upconversion luminescence is recorded under 980 nm laser excitation.

described recently26 and showed about 10% cross-reactivity with DCF metabolites such as 5-OH-DCF, 4′-OH-DCF, and DCF-acyl glucuronide but only less than 1% with other structurally related nonsteroidal anti-inflammatory drugs. The performance of the optimized ULISA was compared with a conventional ELISA as well as LC−MS.



EXPERIMENTAL SECTION Chemicals. All standard chemicals and diclofenac sodium salt (D6899, purity ≥98%) were obtained from Sigma-Aldrich (Steinheim, Germany). Carboxyethylsilanetriol sodium salt; 25% (w/v) in water was obtained from ABCR GmbH (Karlsruhe, Germany). The horseradish peroxidase-labeled horse anti-mouse IgG was from Axxora (Lörrach, Germany) and horse anti-mouse IgG was from Vector Laboratories (Burlingame). The monoclonal anti-diclofenac antibody 12G5 was generated in mice using a DCF-thyroglobulin conjugate as described previously.26 An antibody stock solution of 0.45 mg mL−1 was prepared in 20 mM NaH2PO4, 0.1 M Tris-HCl, 0.02% NaN3, pH 7.4 and stored at 4 °C. Buffers and solutions were prepared with ultrapure water, which was obtained by 6012

DOI: 10.1021/acs.analchem.6b01083 Anal. Chem. 2016, 88, 6011−6017

Article

Analytical Chemistry

laser spot of ∼0.8 mm was focused on the bottom of the microtiter wells. Each well was scanned 100 times in a raster with the step size of 0.4 mm and 500 ms signal integration time. The truncated mean was calculated for each well after discarding the 10 highest and 10 lowest measurements of the luminescence intensity to account for local irregularities on the microtiter well surface that result in signal outliers. Enzyme-Linked Immunosorbent Assay. The ELISA was performed as described earlier.26 A transparent 96-well microtiter plate with high protein binding capacity (Greiner Bio-one, Frickenhausen, Germany) was coated with 0.5 μg mL−1 of ovalbumin-DCF conjugate in coating buffer (50 mM NaHCO3/Na2CO3, 0.05% NaN3, pH 9.6; 200 μL per well) at 4 °C overnight. All subsequent steps were carried out at room temperature. The plate was automatically washed with a plate washer (ELx405 Select, Bio-Tek Instruments, Bad Friedrichshall, Germany) four times with washing buffer (50 mM KH2PO4/K2HPO4, 146 mM NaCl, 0.05% Tween 20, pH 7.6; PBST). The free binding sites in each well were blocked with 300 μL of 1% BSA in PBST for 1 h. The plate was washed four times with washing buffer. First, standard dilutions of DCF in double distilled water or environmental samples (100 μL per well) were added, immediately followed by the anti-DCF monoclonal mouse antibody (12G5, 0.5 μg mL−1 in PBS; 100 μL per well) and incubated for 30 min. After four washing steps, the secondary horseradish peroxidase-labeled antibody was added (0.2 mg mL−1 in PBS; 200 μL per well) and incubated for 1 h. After final washing, the substrate solution (200 μL per well) was added and the plates were shaken for about 15 min for color development. The substrate solution consisted of 25 mL substrate buffer (prepared with 46.0 mL of potassium dihydrogen citrate and 0.1 g of potassium sorbate in 1 L of water, pH 3.8), 500 μL of 3,3′,5,5′-tetramethylbenzidine stock solution (375 mg in 30 mL of dimethyl sulfoxide), and 100 μL of 1% hydrogen peroxide. The enzyme reaction was stopped by adding 100 μL of 5% sulfuric acid per well. The absorbance was read at 450 nm by a microplate reader (Synergy HT, Bio-Tek Instruments). Data Analysis. A four-parameter logistic function (eq 1) was used for a regression analysis of the calibration curves:

Agarose Gel Electrophoresis. Following our previous work,25 COOH-UCNPs and their bioconjugates were characterized by agarose gel electrophoresis (0.5% w/v agarose, 45 mM Tris, 45 mM H3BO3 with pH 8.6, 15 min at 100 V). Samples were mixed in a ratio of 10:1 with 50% w/w glycerol and 8 μL aliquots were loaded onto the gel. A custom-built upconversion reader (Chameleon multilabel microplate reader, Hidex, Turku, Finland) equipped with a continuous 980 nm laser (4 W) was used to scan agarose gels with a spatial resolution of 0.5 mm as described earlier.29 Conjugation of Diclofenac to BSA. BSA-DCF conjugates were prepared using either 1.5 μmol or 7.5 μmol of DCF and 9.7 μmol of sulfo-NHS added to a mixture of 400 μL of MES buffer and 100 μL of dimethylformamide. DCF was activated by the addition of 47 μmol of EDC and incubation at room temperature for 30 min. After adding 500 μL of 0.15 μmol BSA in water and 250 μL of 50 mM aqueous Na2CO3, the mixture was incubated at room temperature for 4 h and then dialyzed (cellulose membrane, MWcut-off: 14 kDa, D9277, SigmaAldrich) three times against 150 mL of 50 mM Na2CO3. BSA-DCF was adjusted to a concentration of 1.6 mg mL−1 by adding 50 mM of Na2CO3 and stored at 4 °C in the presence of 0.05% NaN3. The conjugate was analyzed by matrix-assisted laser desorption ionization (MALDI)-TOF-MS (Bruker, Ultraflex TOF/TOF, N2-laser, 337 nm, positive mode). Water Samples. Munich tap water and two surface water samples were collected in Southern Bavaria from Lake Wörthsee and the Würm River. The fresh water samples were filtrated over a glass microfiber filter (GF/C, Whatman catalog no. 1822 047) and stored at 4 °C. The concentrations of Ca2+, Mg2+, and total organic content (TOC) as well as the conductivity and pH were determined (Supporting Information, Table S1). ELISA and ULISA were performed with undiluted and spiked samples. For LC−MS, the samples were subjected to generic solid phase extraction (SPE) and analyzed by an Orbitrap-based Exactive benchtop mass spectrometer (Thermo Scientific, Dreieich, Germany) as described earlier.26 Upconversion-Linked Immunosorbent Assay. A transparent 96-well microtiter plate with high protein binding capacity (Corning, Wiesbaden, Germany) was coated with BSA-DCF in coating buffer (optimal concentration, 1 μg mL−1 BSA-DCF in 50 mM NaHCO3/Na2CO3, 0.05% NaN3, pH 9.6; 200 μL per well) at 4 °C overnight. All subsequent steps were carried out at room temperature. The plate was washed manually four times with 250 μL of washing buffer (50 mM NaH2PO4/Na2HPO4, 0.01% Tween 20, 0.05% NaN3, pH 7.4). The free binding sites in each well were blocked with 250 μL of 1% BSA in 50 mM NaH2PO4/Na2HPO4, 0.05% NaN3, pH 7.4 for 1 h. The plate was washed four times with washing buffer. Either standard dilutions of DCF in double distilled water or environmental samples (100 μL per well) were added, immediately followed by the anti-DCF monoclonal mouse antibody (12G5; optimal concentration, 0.225 μg mL−1 in 100 mM NaH2PO4/Na2HPO4, 300 mM NaCl, 100 μL per well) and incubated for 1 h. After four washing steps, the microtiter plate was incubated for 1 h with 100 μL of the IgG-UCNP conjugate (optimal concentration, 10 μg mL−1 in 50 mM Tris, 150 mM NaCl, 0.05% NaN3, 0.01% Tween 20, 0.05% BGG, 0.5% BSA, 0.2% PVA, pH 7.75). After four washing steps, the upconversion luminescence was read out from empty wells using a custom-built upconversion microplate reader (Chameleon multilabel microplate reader, Hidex, Turku, Finland) equipped with a continuous 980 nm laser (4 W). A collimated

Y=

Ymax − Ybg 1+

(

[DCF] IC50

S

)

+ Ybg (1)

where [DCF] is the concentration of diclofenac, and Y is either the upconversion luminescence or the absorbance at 450 nm. Equation 1 yields the maximum (Ymax) and background (Ybg) signal, the DCF concentration that reduces (Ymax − Ybg) by 50% (IC50) and the slope at the inflection point (s). All measurements were made at least in triplicate. The concentration of DCF in real water samples was determined by utilizing an inverse function of eq 1 and the limit of detection (LOD) was defined as before:26



Y (LOD) = 0.85 × (Ymax − Ybg) + Ybg

(2)

RESULTS AND DISCUSSION Surface Modification and Characterization of UCNPs. The development of a competitive upconversion immunoassay (ULISA) for the detection of DCF (Figure 1) critically depends on the design of the luminescent reporter that replaces the conventional enzyme amplification steps.30 Oleic acid-capped

6013

DOI: 10.1021/acs.analchem.6b01083 Anal. Chem. 2016, 88, 6011−6017

Article

Analytical Chemistry

Figure 2. Preparation and characterization of IgG-UCNP conjugates. (A) TEM image of silica-coated UCNPs exposing carboxyl groups on the surface (COOH-UCNPs, Supporting Information Figure S3). (B) The carboxyl groups are activated by EDC/sulfo-NHS and conjugated to antimouse IgG. (C) The conjugates are prepared by using different ratios of anti-mouse IgG and COOH-UCNPs (I/II, 33 to 1; III/IV, 7 to 1; V/VI, 3 to 1; VII/VIII, no IgG). Each sample is centrifuged either with 4 000g (I, III, V, VII) or with 10 000g (II, IV, VI, VIII) and characterized by agarose gel electrophoresis. The migration distance (Δ) is indicated with red lines. (D) The relative electrophoretic mobility (Δratio[x]/Δno IgG) of the conjugates is linearly dependent on the ratio of IgG molecules per UCNP.

observed when the concentration of COOH-UCNPs in the reaction mixture was 1 mg mL−1. The conjugation of the secondary antibody reduced the negative surface potential of the COOH-UCNPs as shown by zeta potential measurements and led to a stronger retardation in the agarose gel.36 The shift of the electrophoretic mobility was linearly dependent on the ratio of IgG molecules per UCNP and indicated the degree of surface modification (Figure 2D). Additionally, larger aggregates of nanoparticles remained in the gel pockets and could not enter the agarose matrix. Sample IgG-UCNP-33:1 shows a main fraction of monodisperse bioconjugates separated as a distinct band and a smaller fraction of slowly moving components, which are probably partially aggregated and cross-linked bioconjugates. This result is consistent with a bimodal particle distribution and a higher polydispersity index observed in the DLS measurement. The IgG-UCNP conjugates were purified from an excess of unbound secondary anti-mouse IgG and components of the reaction mixture by differential centrifugation. At first, the bioconjugates were centrifuged at 10 000g, which however, led to strong nanoparticle aggregation (Figure 2C). When the centrifugal speed was reduced to 4 000g followed by short sonication, purified, and monodisperse IgG-UCNP were obtained. Further lowering of the centrifugal field was not efficient since the sedimentation of IgG-UCNP was too slow. The small retardation coefficient of UCNPs prepared with the lowest amount of IgG (IgG-UCNP-3:1, Figure 2C, lane V) indicated an insufficient surface modification. Therefore, this bioconjugate was not used for the following ULISA experiments. Design of Upconversion-Linked Immunosorbent Assay. In a competitive immunoassay, a low concentration of coating antigen ensures that the free analyte can compete efficiently for the binding sites of the detection antibodies. On the other hand, the signal generation has to be strong enough for a reliable readout. Here, we prepared two different coating conjugates consisting of BSA-DCF. The conjugates were analyzed by MALDI-TOF mass spectrometry, which showed

UCNPs were coated with a silica shell exposing carboxylic acid functional groups on the surface. The carboxyl groups improve the dispersibility in water and serve as attachment sites for subsequent conjugation steps. We previously described a onestep water-in-oil microemulsion protocol for coating the surface of small UCNPs (∼12 nm in diameter) with a carboxylated silica shell that showed only a weak upconversion luminescence.25 For the immunoassay, we have synthesized larger UCNPs of 42.5 ± 4.9 nm in diameter (Supporting Information Figure S1) that are much brighter because they are less affected by surface quenching effects.31 The one step silica-coating protocol, however, resulted in aggregation when directly applied to bigger nanoparticles. Therefore, we developed a two-step protocol to prepare a thicker, compact, and more stable silica shell on the surface of UCNPs.32 First, TEOS was added to the microemulsion to generate a thin layer of bare silica (2.4 ± 0.4 nm, Supporting Information Figure S2). This step alone was not sufficient to prevent aggregation. Therefore, TEOS was added for a second time, which changed the thickness of the silica shell only slightly. The second carboxylation step ensured an excellent dispersibility of COOH-UCNP in water.33 The total diameter of COOHUCNPs was consistent as determined by transmission electron microscopy (TEM, 46.9 ± 5.0 nm; Supporting Information Figure S3 and Figure 2A) and atomic force microscopy (AFM, 45.4 ± 7.6 nm; Supporting Information Figure S4). Dynamic light scattering measurements confirmed an increase of the hydrodynamic diameter from 55 to 65 nm after silica coating (Supporting Information Figure S5). The optimized COOH-UCNPs were then conjugated to a secondary anti-IgG antibody via standard EDC/sulfo-NHS chemistry (Figure 2B).34,35 A low concentration of COOHUCNP was utilized to prevent that one antibody molecule binds to several UCNPs, which would lead to aggregation. The conjugates were characterized by agarose gel electrophoresis (Figure 2C), dynamic light scattering (DLS), zeta potential measurements, and FT-IR spectroscopy (Supporting Information Figures S6−S8). The lowest degree of aggregation was 6014

DOI: 10.1021/acs.analchem.6b01083 Anal. Chem. 2016, 88, 6011−6017

Article

Analytical Chemistry

Figure 3. ULISA optimization. (A) Microtiter plates are coated with 1 μg mL−1 of BSA carrying either 10 (red □) or 5.7 (○) DCF residues. (B) The upconversion luminescent (UCL) signal is generated by using 10 μg mL−1 of IgG-UCNP-33:1 (○) or IgG-UCNP-7:1 (red □), respectively. (C) The detection of DCF is optimized by using the monoclonal anti-DCF antibody in concentrations of 0.5 μg mL−1 (○) (IC50, 0.68 ng mL−1), 0.25 μg mL−1 (red □) (IC50, 0.23 ng mL−1), 0.1 μg mL−1 (blue △) (IC50, 0.13 ng mL−1) or 0.02 μg mL−1 (green ◇) (IC50, 0.08 ng mL−1). Error bars represent standard deviations in upconversion signals from three replicate wells.

between signal generation and sensitivity for the determination of DCF and was used in all further experiments. Calibration and Sensitivity of ULISA and ELISA. For each type of competitive immunoassay, it is necessary to find the optimal balance between detection sensitivity for an analyte and signal development. It should also be noted that a high affinity and a low cross-reactivity of the primary antibody are the most distinctive features that determine the sensitivity and specificity of the analyte detection. Figure 4 shows calibration

a coupling density of either 5.7 or 10 DCF residues per BSA molecule (Supporting Information Figures S9−S11). When the conjugate with the higher degree of derivatization was used for coating in the immunoassay, the signals were, as expected, about twice as high but also showed stronger signal fluctuation and a hook effect (Figure 3A), which may be the consequence of two binding sites of IgG molecules forming cyclic complexes (Supporting Information Figure S12).37 By contrast, the conjugate exposing 5.7 DCF residues per BSA molecule yielded more stable signals and a slightly lower IC50 value (1.2 ng mL−1 compared to 1.5 ng mL−1) and a lower detection limit for DCF. Consequently, this coating conjugate was used in all further experiments. An optimal signal generation was observed with a coating concentration of 1 μg mL −1 (Supporting Information Figure S13). The competition step including free DCF and anti-DCF detection antibody was performed in analogy to a sensitive conventional ELISA.26 Only the enzyme-mediated color generation was replaced by an IgG-UCNP conjugate as a direct luminescent reporter (Figure 3B). The higher degree of UCNP surface coverage (IgG-UCNP ratio of 33:1 compared to 7:1) increased the maximum signal intensity by a factor of 5 although both conjugates were prepared with a molar excess of IgG molecules per nanoparticle. This difference can be explained because not every surface-conjugated antibody may have the right orientation or be fully functional in order to bind efficiently to the primary antibody. Consequently, a higher degree of derivatization resulted in a proportionally higher number of functional antibodies. On the downside, using IgGUCNP-33:1 resulted in strong signal fluctuations as well as a hook effect, which impedes the reproducible determination of DCF. It should also be noted that the degree of surface substitution did not significantly affect IC50 or the LOD, and a concentration of 10 μg mL−1 IgG-UCNP-7:1 resulted in the most reproducible upconversion signal generation (Supporting Information Figure S14). In contrast to the UCNP-bound secondary antibody, the primary anti-DCF antibody is directly involved in the competition step. Figure 3C shows that both the upconversion signal intensity and the IC50/LOD for DCF strongly depend on the antibody concentration. A higher primary antibody concentration leads to a higher signal intensity because more antibodies can bind to the DCF-BSA coating conjugate, but they also consume a larger amount of free DCF and thus deteriorate the assay sensitivity. A concentration of 0.25 μg mL−1 primary anti-DCF antibody yielded an optimal balance

Figure 4. Normalized calibration curves of ULISA (red □, replotted red curve from Figure 3C; IC50, 0.23 ng mL−1; LOD, 0.05 ng mL−1) and ELISA (○, IC50, 0.05 ng mL−1; LOD, 0.01 ng mL−1). Error bars represent standard deviations of three replicate wells.

curves of ULISA and ELISA recorded under similar conditions and using the same anti-DCF primary antibody. In both cases a signal to background ratio (Ymax/Ybg) of 5:1 was adjusted to achieve the most sensitive detection of DCF but also to obtain a reliable signal generation. The competitive ULISA (LOD, 0.05 ng mL−1/170 pM) has a 5 times higher detection limit than a conventional ELISA (LOD, 0.01 ng mL−1/ 34 pM) but allows for an easier and faster signal generation. As the detection sensitivity is ultimately dependent on the anti-DCF antibody, it can be expected that the ULISA can be further optimized by developing brighter UCNPs reporter conjugates. Competitive immunoassays for small molecules are typically less sensitive than sandwich immunoassays where the signal generation is directly proportional to the analyte concentration. The highest sensitivity was described for the detection of Schistosoma circulating anodic antigen by using micrometersized upconversion particles in a lateral flow assay (LOD, 0.01 ng mL−1/0.1 pM).13 This particular analyte displays repetitive surface epitopes and facilitates binding of several primary antibodies per analyte molecule. The competitive immunoassay for DCF affords a similar sensitivity as a magnetic bead-based competitive immunoassay for the detection of aflatoxin that was 6015

DOI: 10.1021/acs.analchem.6b01083 Anal. Chem. 2016, 88, 6011−6017

Article

Analytical Chemistry Table 1. Detection of DCF in Unspiked and Spiked Real Water Samples spiked (ng mL−1)

ULISA (ng mL−1)

ELISA (ng mL−1)

LC−MS (ng mL−1)

Lake Wörthsee

1 10