Article pubs.acs.org/est
A Membrane-Based ELISA Assay and Electrochemical Immunosensor for Microcystin-LR in Water Samples M. Lotierzo,† R. Abuknesha,‡ F. Davis,† and I. E. Tothill*,† †
Cranfield Health, Cranfield University, Cranfield, Bedfordshire, MK43 0AL, England, United Kingdom King’s College London, Strand, London WC2R 2LS, England, United Kingdom
‡
S Supporting Information *
ABSTRACT: We describe within this paper the development of an affinity sensor for the detection of the cyanobacterial toxin microcystin-LR. The first stage of the work included acquiring and testing of the antibodies to this target. Following the investigation, a heterogeneous direct competitive enzyme-linked immunosorbent assay (ELISA) format for microcystin-LR detection was developed, achieving a detection limit, LLD80 = 0.022 μg L−1. The system was then transferred to an affinity membrane sorbentbased ELISA. This was an amenable format for immunoassay incorporation into a disposable amperometric immunosensor device. This membrane-based ELISA achieved a detection limit, LLD80 = 0.06 μg L−1. A three-electrode immunosensor system was fabricated using thick-film screen-printing technology. Amperometric horseradish peroxidase transduction of hydrogen peroxide catalysis, at low reducing potentials, versus Ag/AgCl reference and carbon counter electrodes, was facilitated by hydroquinonemediated electron transfer. A detection limit of 0.5 μg L−1 for microcystin-LR was achieved. Similar levels of detection could be obtained using direct electrochemical sensing of the dye produced using the membrane-based ELISA. These techniques proved to be simple, cost-effective, and suitable for the detection of microcystin-LR in buffer and spiked tap and river water samples.
1. INTRODUCTION As a possible consequence of eutrophication, toxins from algae and other water microorganisms have become widely recognized as a contributory factor to human health problems. One of these families is cyanobacteria, a frequent component of many freshwater and marine ecosystems. They are aquatic photosynthetic bacteria, usually unicellular, though they often grow in suspended colonies. In several countries, the association of cyanobacterial blooms with animal poisoning episodes and human health problems has raised the possibility of toxin production by the common bloom-forming cyanobacteria.1 Cyanobacteria produce a wide variety of chemically unique secondary metabolites that have a harmful effect on other tissues, cells, or organisms including humans. Toxins from cyanobacteria constitute a major source of natural toxins, ‘biotoxins’ that are often found in surface supplies of freshwater. The microcystins are a group of cyclic heptapeptide hepatotoxins produced by a number of cyanobacterial genera, the most notable being the widespread Microcystis, from which the toxins take their name. Over 60 microcystins (one of the major ones being microcystin-LR) have been isolated so far. They consist of a seven amino acid peptide ring, which is made up of five nonprotein amino acids and two protein amino acids (Figure 1). The LD50 value for microcystin-LR (MC-LR) by the intraperitoneal route is 50 μg kg−1 in rats.2 The primary effect © 2012 American Chemical Society
on health is toxicity to liver cells. The World Health Organization (WHO) has proposed a provisional guideline value of 1 μg L−1 MC-LR in drinking water. Other workers3 investigated the chronic oral toxicity of MC-LR on female mice and recommended a value of 0.01 μg L−1 as a maximum acceptable level for microcystins in drinking water. To measure these low concentrations of microcystins in drinking and surface water, a highly sensitive analytical method is required. Both the low concentration and the large number of more than 60 known analogues of microcystins represent a considerable challenge for the analysis of microcystins. The most common methods used for the detection of marine and freshwater toxins include bioassays, chromatographic techniques, chemical assays, and immunoassay techniques. The routine analysis of microcystins is usually carried out by high-performance liquid chromatography (HPLC), which requires sample processing steps prior to analysis.4 Microcystins display a potent and specific inhibition of serine/threonine protein phosphatases, leading to the use of protein phosphatase inhibition assays for their detection and screening.5−7 An immunoassay based on time-resolved fluorometry was Received: Revised: Accepted: Published: 5504
January 27, 2012 April 10, 2012 April 11, 2012 April 11, 2012 dx.doi.org/10.1021/es2041042 | Environ. Sci. Technol. 2012, 46, 5504−5510
Environmental Science & Technology
Article
Figure 1. Structure of microcystin and nodularin toxins. X and Y are variable L-amino acids. D-MeAsp is D-erythro-β-methylaspartic acid, Adda is (2S,3S,8S,9S)-3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6-dienoic acid and Mdha is N-methyldehydroalanine (Dha = dehydroalanine).
developed for microcystin detection.8 The assay was performed in a competitive mode and utilized a monoclonal antibody raised against MC-LR. The detection of the assay was 0.1 μg L−1 but requires a laborious protocol. The first ELISA for these compounds9 was based on a direct competitive ELISA format, involved coating anti-MC-LR antibodies on a microtiter plate and using MC-LR-peroxidase as the enzyme marker, and showed a working range of 0.5 to 10 μg L−1, with a minimum detection level of 0.2 μg L−1 in water. Likewise, anti-MC-LR polyclonal antibodies could be used to develop an ELISA kit with a working range of 0.5−50 μg L−1.10 Monoclonal antibodies against MC-LR have been used within assays which demonstrated activities in the range of μg L−1 and was used to detect microcystins in environmental samples.11,12 A technique based on molecular imprinted polymers,13 which were used first as solid-phase extraction cartridges to preconcentrate the toxin, and then similar polymers used as an active layer on a mass-sensitive sensor proved capable of detecting as little as 0.35 μg L−1. There has been a great deal of research into the development of biosensors for a wide range of analytes. These could potentially allow the development of inexpensive sensors which could be used for in situ analysis of a wide range of species in different matrixes without the requirement for time-consuming and expensive laboratory-based procedures. A biosensor based on surface plasmon resonance (SPR) could be used to examine the effects of three microcystins, -LR, -RR, -YR, on the binding between the enzyme protein phosphatase PP-2A and its substrate phosphorylase-a (PL-a).14 The SPR biosensor provides real-time information on the association and dissociation kinetics of PL-a with immobilized PP-2A in the absence and presence of microcystins. However, in samples responding positively, the accurate toxin determination and identification requires the complementary use of an analytical technique such as LC-UV or LC-MS. More recently, antibodies to MC-LR were immobilized along with single-walled carbon
nanotubes onto paper substrates.15 The resultant electrochemical immunosensor was capable of detecting MC-LR at a range up to 40 μg L−1 with a limit of detection of 0.6 μg L−1. Other methods used for MC-LR detection include a simple dipstick assay where antibodies were immobilized onto colloidal gold.16 The resultant assay had a range of 1−5 μg L−1 and could be applied to fresh and salt water and tissue of MC-LR-fed mussels. A capillary ELISA technique combined with chemiluminescense detection17 allowed MC-LR quantification as levels as low as 0.2 μg L−1. Surface plasmon resonance could also be utilized18 to determine MC-LR levels in the range 0.2−2.0 μg L−1. High cross-reactivity was noted for other members of the microcystin family. Simultaneous detection of MC-LR and trinitrotoluene in water was demonstrated19 using a fiber-optic evanescent wave detector, with detection limits of 0.04 μg L−1 for MC-LR and an assay time of about 10 min. Very recently, an immunosensor based on a cantilever detection system has been shown to be capable of detection of MC-LR at extremely low levels20 which could be used in tap or river water and had a sensing range of 1−100 ng L−1. The aim of this work is to develop affinity sensors for the detection of cyanobacterial toxins in environmental samples that can be simple and easy to apply for field analysis and are cost-effective. Because of the wide variety of toxins with different molecular structure and mechanisms of action, the study was focused on a specific toxin family, microcystins toxins (with particular regard to MC-LR). Antibodies against MC-LR were raised and optimized using a method similar to that described,21−23 or they were acquired and used as the natural receptors in the development of ELISA assays and membranebased ELISA (MELISA) assays and in disposable electrochemical immunosensors. The sensors were fabricated using screen-printed electrodes capable of directly transducing the response from an affinity membrane-bound competitive enzyme immunoassay. Previous work within our group has utilized these methods to formulate a membrane assay for the 5505
dx.doi.org/10.1021/es2041042 | Environ. Sci. Technol. 2012, 46, 5504−5510
Environmental Science & Technology
Article
pesticide isoproturon24 and incorporated it within an electrochemical immunosensor.25
sensor construction has been described previously.25 The membrane disk was placed over the working electrode and then covered by a 15 mm2 piece of polyester 250 μm2 mesh. This assembly was then held in place with 30 μL of electrolytic buffer, which allowed the mesh to adhere to the electrode, thus trapping the membrane. The subsequent electrochemistry applied for the enzyme reaction measurement utilized amperometric analysis. The sensor system was assembled together with 50 μL of electrolytic buffer containing 1 mM hydroquinone mediator. After poising the electrodes at a −300 mV reducing potential, the sensor system was allowed to equilibrate at room temperature for between 60 and 100 s. Then 10 μL of 0.5 mM hydrogen peroxide substrate in electrolytic buffer was added. The final current readings were taken between 300 and 400 s, when a steady-state condition had been obtained. The difference in current was then measured between the two steady-state equilibrations, indicating the substrate concentration. As an alternative to the hydroquinone-mediated immunoassay, a second method that utilized the electrochemical detection of ABTS in the same manner was applied. When the ABTS substrate solution was used, the sensor was held in position with 30 μL of electrolytic buffer, and ABTS substrate solution was added after 50 s. Chronoamperometry at +150 mV vs Ag/AgCl was used to interrogate the system.
2. MATERIALS AND METHODS 2.1. Antibody Generation, Conjugate Formation, and ELISA optimization. Details of the chemicals and immunoassay reagents are in Supporting Information. The binding activity of the anti-MC-LR polyclonal antibody raised in sheep was assessed using an indirect solid-phase enzyme immunoassay as described in Supporting Information. Optimization of the gelatin−MR-LR conjugate used and of the competitive ELISA assay is described in Supporting Information. 2.2. MELISA Based on Antigen Coating. Gelatin−MCLR conjugate was immobilized on UltraBind membranes, at different dilutions (12, 0.6, 0.12 g L−1) in a coating buffer. A 5− 10 μL aliquot of the conjugate was applied for 15 min to each side of a 15 mm diameter membrane disk. The membrane sorbent selected was an UltraBind US-800, 0.8 μm (Pall Gelman Science, Portsmouth, U.K.) covalent attachment membrane. Membrane discs of 6 mm diameter (28.3 mm2 planar area) were cut and handled only with fine membrane forceps (BDH, Gillingham, U.K.). This polyethersulfone preactivated membrane contains active aldehyde groups, which covalently attach to the free amino groups of a coating analyte. The membrane was left to dry completely at room temperature before proceeding further. After the immobilization procedure, the membrane discs were washed in phosphate-buffered saline (PBS) containing 0.05% Tween 20 (PBST). The remaining active sites on the membrane were blocked with 5% gelatin in PBS buffer at pH 7.2 for 45 min (50 μL). The primary antibody anti-MC-LR (50 μL of a 1/1500 dilution) was then added to the membrane and incubated for 1 h at 37 °C with gentle shaking. After washing with buffer (three times, 200 μL per each membrane), the secondary antibody, labeled with HRP (50 μL of 1/20 000 dilution), was added and incubated for 1 h at 37 °C. The immobilized peroxidase activity was measured at 450 nm, after incubation with ABTS (2,2′azinobis(3-ethylbenzothiazoline-6-sulfonic acid)) substrate (150 μL) solution for 20 min. A volume of 100 μL of the colored solution was dispensed in empty wells, and the optical density of the colorimetric reaction was measured at 405 nm. Blocking procedures were investigated using several blocking agents at different concentrations. The membranes were coated with free gelatin or MC-LR−gelatin conjugate (0.6 g L−1) and then blocked with different protein blockers, 0.5%, 2%, 5% gelatin, 2% human albumin, 1% casein, or 1% ovalbumin. A control was prepared, whereby no blocking agent was used. Blocking was carried out for 30 min at 37 °C and the assay continued, as described previously. 2.3. Construction and Interrogation of an Electrochemical Immunosensor. The screen-printed electrodes produced in the context of this work were manufactured using an automated screen-printing machine DEK 248 (DEK Printing Machines Ltd., Weymouth, England) and have been described extensively in previous work.25 They consisted of carbon working and counter electrodes along with an Ag/AgCl reference electrode. The immunosensor development culminated in the combination of the optimized MELISA and amperometric detection systems. This brought the membrane immunosorbent into intimate contact with the SPE such that the assay response could be directly transduced amperometrically. The immuno-
3. RESULTS AND DISCUSSION 3.1. Development of the ELISA. The antibody against the toxin MC-LR and the gelatin−MC-LR conjugate were sequentially incorporated into an enzyme immunoassay (EIA) format using different concentrations, allowing the reagents properties and interactions to be investigated. First, a microtiter plate enzyme-linked immunosorbent assay (ELISA) was developed and optimized which then allowed the assay to be transferred to a membrane-based ELISA (MELISA) and then to an electrochemical immunosensor application. The specific binding of anti-MC-LR immunoglobin G (IgG) to the analyte conjugate (MC-LR−gelatin) was initially investigated using a double dilution checkerboard titration assay. The most appropriate dilution of antibody for the immunoassay development was found to be approximately 1/ 1500 used in association with a 1/400 dilution of MC-LR− gelatin conjugate. The sensitivity of anti-MC-LR IgG against gelatin−MC-LR antigen was further investigated using the ELISA assay format and the affinity constant KD calculated (from the reciprocal Hughes−Klotz plot) to be 5 × 10−11 M, showing a strong affinity as well as indicating that binding was specific to the MC-LR unit and that nonspecific binding to the gelatin did not occur. High binding capacity microtiter plates were then used as sorbent microtiter plates using 0.1 M sodium carbonate− bicarbonate buffer, pH 9.6, as the coating buffer and 0.5% gelatin as the blocking agent. The effect of different combinations of coating and blocking incubation times at room temperature, without shaking, were investigated. The preferable working conditions for this competitive ELISA assay development were 1 h coating and 45 min blocking. Under these optimized conditions, a competitive ELISA for microcystin-LR detection was conducted (Figure 2A). The blank values (A0, no toxin added) were generally about 1−1.2. A typical sigmoidal response for MC-LR concentrations from 0 to 10 μg L−1 was shown. The detection range considered was between 80% and 20% of the curve optical absorbance value of 5506
dx.doi.org/10.1021/es2041042 | Environ. Sci. Technol. 2012, 46, 5504−5510
Environmental Science & Technology
Article
to indicate the presence (and levels) of these toxin families, without specific reference to a particular peptide. Because the field samples will be mainly river and lake water (in addition to extracts of cells), it was necessary to assess the effects of tap water and filtered river water (from Thames River, U.K.) on the binding of anti-MC-LR antibodies. The matrix effects, due to the presence of salts or organic compounds (humic acid in water), were assessed by a calibration assay for MC-LR, under varied conditions where the standard analyte solutions were made up in ELISA buffer containing 0.05% Tween 20 in PBS buffer, in tap water, or in filtered Thames water. The results (Figure 2B) indicated that there were no significant effects from the test matrixes. Tap and filtered river water showed binding similar to that of PBS buffer but higher than that of the ELISA buffer which contains Tween 20 (0.05%). The curves shifted to the right, without any significant changes in the graph slope. This indicates the ability of using the developed assay in river and tap water samples. 3.2. Development of the MELISA. The MC-LR IgG competitive ELISA was then transferred to a membrane to form a bound MELISA. The membrane chosen was UltraBind from Pall Gelman Sciences U.K. The membranes are suitable for solid-phase immunodiagnostic tests and flow-through assays and claim low nonspecific binding and background levels. The high response, fast wicking, double-sided properties of the membrane (without support material) rendered the US-800 membrane suitable for amperometric detection, where a fast diffusion of the reagents to the electrode surface is essential. The MC-LR−gelatin conjugate (0.12−12 g L−1) was applied to 6 mm membrane discs by spotting with 5 μL of the solution and incubating for 15 min. After coating, the US-800 membranes were blocked with 0.5% gelatin for 45 min at room temperature. The membranes were first incubated with MC-LR antibody (50 μL) and then washed with PBS buffer and then HRP-labeled secondary antibody (50 μL, 1/20 000 dilution) was added an incubated for 1 h at 37 °C. Following intensive washing of the membranes, the ABTS substrate solution was added (150 μL) and incubated for 20 min. The result from the titration profile showed high nonspecific binding with high standard deviation (% CV was 8.4, 12, and 18.2 for 12 g L−1, 0.6 g L−1, and 0.12 g L−1 respectively showing a very low coating reproducibility (data not shown). To reduce the nonspecific signal of the enzyme-linked immunoassay reaction on the membrane a variety of different blocking agents were investigated (0.5%, 2% and 5% gelatin, 2% human albumin, 1% casein, and 1% ovalbumin). The ELISA assay was conducted as previously, and the results indicated that the blocking efficiency improved when the gelatin concentration was increased from 0.5% to 5% (signal/background ratio (S/B) from 1.6 to 2.9, respectively) and that other blocking agents did not improve performance better than the 5% gelatin (S/B ∼ 2 for ovalbumin). However, in all cases, assay reproducibility was very poor with mean % CV > 15. Upon further optimization of the incubation time during the blocking step, the primary and secondary antibody concentrations did not lead to an improvement in the performance of the assay. It was therefore concluded that the poor assay performance could be due to the low efficiency of the MC-LR−gelatin binding to the membrane. An enzyme immunoassay with anti-MC-LR antibody immobilized directly onto the UltraBind membrane surface was then investigated and optimized. A commercial reagent of the conjugate MC-LR−HRP (Adgen, Scotland, UK) was employed, which competes with free MC-LR toxin in the
Figure 2. A. Competitive ELISA assay for MC-LR. Assay was coated with gelatin−MC-LR and 0.5% gelatin blocker. MC-LR standards and final dilution of 1/1500 MC-LR IgG followed by 1/20 000 2° IgG− HRP. B. Matrix effect for analyte MC-LR, gelatin conjugate 1/400 coating. 0.5% gelatin blocking. MC-LR standards in different solutions competed with SH IgG 1/15000 for 1 h at 37 °C. 1/20 000, 2° IgG− HRP incubation for 1 h at 37 °C. Optical density read at 405 nm after 20 min of ABTS substrate solution incubation. Error bars = ±SD, n = 4.
0.022−0.8 μg L−1. The responses were highly reproducible, as indicated by the low standard deviations and the mean % coefficient of variation (%CV)
