Anal Bioanal Chem (2006) 384: 572–583 DOI 10.1007/s00216-005-3392-0

R EV IE W

Christel Hempen Æ Uwe Karst

Labeling strategies for bioassays

Received: 4 May 2005 / Revised: 9 June 2005 / Accepted: 10 June 2005 / Published online: 6 October 2005
Springer-Verlag 2005

Abstract Different labeling strategies for enzymatic assays and immunoassays are reviewed. Techniques which make use of direct detection of a label, e.g. radioimmunoassays, are discussed, as are techniques in which the label is associated with inherent signal amplification. Examples of the latter, e.g. enzyme-linked immunosorbent assays or nanoparticle-label based assays, are presented. Coupling of the bioassays to chromatographic separations adds selectivity but renders the assays more difficult to apply. The advantages and drawbacks of the different analytical principles, including future perspectives, are discussed and compared. Selected applications from clinical, pharmaceutical, and environmental analysis are provided as examples. Keywords Labeling Æ Immunoassay Æ Enzyme assay Æ Nanoparticle Æ Liquid chromatography

Introduction Bioassays are analytical techniques in which selectivity and/or sensitivity are generated by a biomolecular interaction, e.g. enzymatic amplification in metabolic assays (enzyme assays) or antibody–antigen recognition in affinity assays (immunoassays). They find widespread application in several fields as clinical chemistry, drug analysis, food analysis, environmental chemistry, and even in the analysis of explosives. Although the first bioassays were applied decades ago, this field of analysis is still open for, and applicable to, new and interesting approaches and developments.

C. Hempen Æ U. Karst (&) Chemical Analysis Group and MESA+ Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands E-mail: [email protected] Fax: +31-53-4894645

Immunoassays are based on the highly selective interaction and binding of antibodies to antigens, which can be monitored by use of a label attached either to the antigen or to the antibody. Although techniques for direct and specific monitoring of the interaction between antibody and antigen have become available in recent years (e.g. surface plasmon resonance, SPR [1], or related techniques), most quantitative assays rely on the indirect approaches based on labeling strategies [2]. The term ‘‘labeling’’ describes a chemical reaction between the analyte and a suitable reagent with formation of a product which enables and/or improves detection of a (bio)molecule or a (bio)molecular interaction. This review focuses on different labeling strategies in enzyme and immunoassays. It is, on the one hand, possible to label a biomolecule in such a way that direct detection and/or quantification are possible after binding. This is achieved in radioimmunoassays (RIAs) and related techniques. Labeling with enzymes, on the other hand, requires a subsequent chemical amplification reaction with formation of a product which may be detected by spectroscopic or electrochemical methods. Enzymatic assays are based on the high selectivity of enzymes. They react as biological catalysts and therefore increase the velocity of reactions, typically by a factor between 106 and 1012. Because of this large turnover, a large improvement of the limit of detection based on the corresponding signal amplification is achieved [2]. In general, the advantages of bioassays are, in addition to high sensitivity and selectivity, low operating costs, simple and readily available instrumentation, and low limits of detection. Another advantage is the extremely high sample throughput possible as a result of massive assay automation with robotic systems. This review provides a general overview of different labeling strategies applied in bioassays, including their advantages and disadvantages and their future perspectives. The selected references provide some insight into the broad area of bioassay application. It is evident from thousands of original papers and numerous dedi-

573

cated textbooks [2, 3] and review articles [4, 5] that this scientific area cannot be covered comprehensively. Rather, selected techniques and applications considered by the authors to be of particularly high value will be summarized and discussed.

Direct labeling Radioimmunoassay Radioimmunoassays are based on competition between radiolabeled (e.g. by means of 125I, 3H) and unlabeled antigens for limited antibody binding sites. The unknown amount of the analyte can be measured by increasing the concentration of ‘‘cold’’ (unlabeled) antigen, which causes replacement of ‘‘hot’’ (labeled) antigen and thus a decrease in measurable radioactivity. The displacement is calibrated by means of known amounts of antigen so that unknown analyte concentrations can be determined by use of the resulting anti-proportional calibration plot [2]. This kind of competitive immunoassay is one of the most sensitive methods for determination of antigens (LOD0.5 pg mL 1) and the results are very precise. It therefore finds widespread application in several fields. Rosalyn Yalow and Solomon Berson developed the first radioisotope immunoassay: They determined the concentration of insulin in plasma samples by using 131Ilabeled insulin as reference [6]. This was possible because of the ability of human insulin to react strongly with the insulin-binding antibodies present in guinea-pig antibovine insulin serum so that 131I-labeled insulin could be competitively replaced. This method yielded quantitative and reproducible results, leading to increased interest in this technique. In 1977, Rosalyn Yalow was awarded the Nobel Prize in Physiology or Medicine for the development of RIAs of peptide hormones. The possibilities of RIAs are best demonstrated with two typical examples of applications in the field of clinical chemistry: In 1988, Coates et al. [7] introduced a RIA for salivary cyclosporine with use of 125I-labeled cyclosporine [2]. Cyclosporine belongs to the group of medicines known as immunosuppressive agents. It is used to reduce the body’s natural immunity in patients who receive organ (for example, kidney, liver, or heart) transplants. When a patient receives an organ transplant, the body’s white blood cells try to reject the transplanted organ. Cyclosporine works by hindering the white blood cells from doing this. In this work, cyclosporine was determined in saliva samples from 38 kidney-transplant patients. The limit of quantification was 0.34 lg L 1. In 1996, Ma et al. [8] developed a RIA for determination of the protein hormone leptin in human plasma or serum samples. They performed a competitive assay of the leptin samples (or calibrations) with 125I-labeled leptin for antibody binding sites with final precipitation of the antibody–antigen complex. Subsequent centrifugation and decantation of the supernatant and counting

of the radioactivity in the pellets were performed to determine remaining radioactivity. By means of this accurate and precise technique they discovered that leptin concentrations vary little as a result of short-term fasting, age, or race and that leptin concentrations are gender-specific. Several reviews on RIAs, alone or in comparison with other (bio)analytical techniques, are available. In 1993 for instance, Deridovich and Reunova [4] reviewed the analysis of prostaglandins in different tissues by means of RIAs in comparison with other bioassays and HPLC techniques. In 1996, Leveque and Jehl [9] reviewed the clinical pharmacokinetics of the semisynthetic anticancer drug vinorelbine and compared its determination by means of HPLC methods with that using radioactivelylabeled assays. These examples show that RIAs have already successfully been used for a very long period. Nevertheless, the drawbacks of this type of immunoassay must be considered. One major problem is the radioactivity itself; this results, on the one hand, in a lack of shelf life of the radiolabeled antigens and, on the other hand, in the need for special safety precautions and dedicated laboratory equipment, thus resulting in high infrastructure costs. Therefore, other immunoassay principles, which enable similarly low limits of detection to be achieved are becoming increasingly popular. Fluoroimmunoassay The group of immunoassays based on direct labeling, also includes fluoroimmunoassays. This technique works with antibodies labeled with fluorescent markers. Therefore, the selectivity of antibody-binding interactions is combined with simple, sensitive, inexpensive, and hazard-free fluorescence detection. As early as 1941, Coons et al. [10] discovered the possibility of coupling antibodies to fluorescent dyes (e.g. rhodamines) without changing their specificity. The development of powerful fluorescence immunoassay applications took until the 1980s. Two types of fluoroimmunoassays are distinguished: homogeneous and heterogeneous. In heterogeneous fluoroimmunoassays, antigens or antibodies labeled with a fluorophore are used; this requires separation of bound from free tracer before fluorescence detection. In homogeneous assays quantification is possible without performing separation procedures [2]. Again, an example is provided to demonstrate the use of this type of assay. In 1983, Bailey et al. [11] described the use of lucifer yellow VS as a label for fluoroimmunoassays. Lucifer yellow VS is a highly fluorescent vinyl sulfone dye which binds under mild alkaline conditions to both amino and thiol groups in proteins. The large Stokes’ shift of 110 nm and the emission maximum at 540 nm provide advantages over the more commonly used labels (e.g. fluorescein). The use of the dye as a label has been demonstrated by developing a heterogeneous fluoroimmunoassay for human serum

574

albumin by polyethylene precipitation of bound fluorophore and automated fluorimetry on the supernatant. In later studies, Bailey and colleagues [12] used lucifer yellow VS in homogeneous non-separation fluoroimmunoassays. A homogeneous fluorescence resonance energy transfer (FRET)-based immunoassay for plasma albumin was described in which lucifer yellow VS was used to label albumin, and rhodamine B isothiocyanate was used to label anti-albumin antibodies. Results from the assay correlated well with those from the dye-binding method for albumin, and sensitivity and precision compared favorably with those of similar assays using a fluorescein label. Fluoroimmunoassay measurements combined with flow-injection techniques can often be found in literature [13, 14, 15, 16, 17]. Khokhar et al. [14], for instance, worked with immunoreactors containing immobilized protein G as solid phase. The set-up of the measurements is presented in Fig. 1. Insulin, used as model analyte, was labeled with the pH-resistant and fluorescent rhodamine isothiocyanate. After incubation of the rhodamine–insulin conjugate with anti-insulin antibodies and the respective sample or standard solution, the mixture was injected into the system at pH 8.8. The antibody-bound insulin was retained. On changing to pH 2.5 the protein G-combined conjugates were eluted, so that the fluorescence could be determined. In general, detection limits tend to be high, because of background fluorescence and quenching problems, especially in homogeneous assays. Therefore, the popularity of these fluoroimmunoassays is not as high as for RIAs. Heterogeneous immunoassays, especially in combination with flow-injection analysis, can overcome these problems. Considerable improvements have been achieved using time-resolved fluorescence techniques, which enable highly sensitive measurements with reduced background. Hemmila¨ and Webb [18] and, recently, Steinkamp and Karst [19] have given overviews on this technique. The use of fluorophores excited in the red or the near infrared (NIR) part (600–1000 nm) of the electromagnetic spectrum is another successful approach used to reduce interference from background signals [20, Fig. 1 Flow-injection manifold for heterogeneous fluorescence immunoassays

21, 22, 23]. Boyer et al. [20], for example, studied for the first time the use of NIR fluorescent dyes as quantitative labels for immunoassays. After conjugation of an isocyanate-functionalized dye to goat anti-human immunoglobulins it was used in an immunoassay to detect and to quantify the respective analytes, human immunoglobulins, by means of laser diode-induced fluorescence spectroscopy. Although all of these more recent approaches are helpful in reducing the limits of detection, they are still inferior to RIAs. On the other hand, their easy handling and the fact that only readily available instrumentation is required make these assays attractive for applications in which only moderate limits of detection have to be achieved.

Enzyme labeling Enzymatic assays and enzyme immunoassays play an important role in clinical chemistry and related fields. Enzymes react as biological catalysts by reducing the activation energy of chemical reactions, resulting in acceleration of the reactions [2]. Many different enzymes are used as labels for these types of assay. The advantages of working with enzymes are their high catalytic activity, selectivity, and the sensitivity obtained in the assay owing to strong signal amplification. Enzyme-catalyzed bioassays Numerous enzyme-catalyzed reactions are known and currently applied in bioassays. They are combined with a variety of detection schemes, for example UV–visible absorbance, fluorescence, chemiluminescence, electrochemical, and, most recently, mass spectrometric detection. Enzymes used in bioassays must fulfil several quality demands - high selectivity, no contamination with, e.g., inhibitors or disturbing substances, high stability, a pH optimum which should fit the assay conditions, and reasonable cost. By means of enzymatic assays it is not only possible to determine the concen-

575 Table 1 Popular enzymes used in bioassays Enzyme

EC No.

Origin

Molecular pH Detection mass (kDa) Optimum

Substrate

Ref.

Alkaline phosphatase

EC 3.1.3.1

Calf intestine

140

5-FSAP

[24, 25]

Fluorescence, EALL UV–visible Electrochemistry Fluorescence

b-D-Galactosidase EC 3.2.1.23

Escherichia coli

b-D-Glucose oxidase

EC 1.1.3.4

Aspergillus niger 160

5.5–6.5

Luciferase Peroxidase

EC 1.13.12.7 Photinus pyralis 100 EC 1.11.1.7 Horseradish 44

7.5–7.8 6.0–7.0

Xanthine oxidase

EC 1.1.3.22

Bovine milk

465

9.8

283

8

8.5–9.0

tration of substrates but also the activity of the enzymes themselves. A list of some selected popular enzymes for bioassays, including frequently used substrates and applied detection schemes, is provided in Table 1. Some examples are also provided in the text - the reaction of horseradish peroxidase (POD) with hydrogen peroxide and o-phenylenediamine (OPD), for instance, has been known for decades [34, 37, 38, 39, 40] and is frequently applied for enzymatic and immunoassays. The concentration of the enzyme, hydrogen peroxide, or a respective precursor (hydrogen peroxide-generating enzyme) is determined using the POD-catalyzed conversion of OPD to an orange–red reaction product, 2,3-diaminophenazine, which is most frequently detected by UV–visible absorbance using microplate readers in the wavelength range between 425 and 450 nm and, after quenching with H2SO4, at a wavelength of approximately 490 nm. Phosphatases are another important group of enzymes with applications in bioassays. Acid phosphatase Fig. 2 Reaction scheme for the aP-catalyzed cleavage of 5-FSAP to 5-FSA with subsequent complexation with Tb(III)/EDTA

4-Nitrophenyl phosphate 1-Naphthyl phosphate 4-Methyl-umbelliferone -b-D-galactoside UV–visible 2-Nitrophenyl-b-D-galactoside Fluorescence 4-(N-Methylhydrazino)-7-nitro2,1,3-benzooxadiazole (MNBDH) UV–visible 4-Aminophenazone Bioluminescence Adenosine triphosphate (ATP) Fluorescence MNBDH Fluorescence p-Hydroxyphenyl propionic acid (pHPPA) UV–visible TMB UV–visible OPD UV–visible 2,2¢-Azino-bis(3-ethylbenzothiazoline -6-sulfonic acid) diammonium salt (ABTS) UV–visible Hypoxanthine

[26] [27] [26] [24] [24, 29] [30] [31] [29] [32] [33] [34] [35]

[36]

(acP), for instance, reacts with p-nitrophenyl phosphate (NPP) to form a dye which can be detected photometrically [41]. Evangelista et al. [25] developed a method for determination of alkaline phosphatase (aP) by use of 5-fluorosalicylic phosphate (5-FSAP), which reacts to give the fluorescent 5-fluorosalicylic acid (5-FSA). In their investigations, a second reaction step was introduced during which 5-FSA formed a complex with Tb(III)/EDTA, so that time-resolved measurements were enabled. The reaction scheme is shown in Fig. 2. These so-called enzyme-amplified lanthanide luminescence (EALL) assays have recently been reviewed [19]. Because of the combination of enzymatic amplification and time-resolved measurements, the limits of detection of EALL assays may be very low under optimized conditions. Mass spectrometric detection is attracting increasing interest for enzymatic assays. The first approach was published in 1989 by Henion et al. [42]. They coupled a

576

closed reaction vessel to an electrospray ionization (ESI) interface of a triple-quadrupole mass spectrometer and monitored on-going reactions in continuous-flow experiments. A few years later the same group introduced an off-line LC–ESI–MS approach for determination of the kinetic data for ribonuclease A and b-galactosidase [43]. Liesener and Karst [44] have recently published a review on mass spectrometric monitoring of enzymatic conversions. The advantage of MS-based detection schemes is that multiplexing reactions can be monitored without interference of one reaction with the other. It is even possible to determine the decrease of the educt and the increase of the product simultaneously by means of on-line or at-line measurements. In photometric detection simultaneous measurements are extremely difficult. Enzymatic assays with fluorescence detection do, in principle, enable a simultaneous measurement of a limited number of fluorophores, if the emission bands of the fluorescent products do not overlap [24]. Mass spectrometric detection is therefore likely to dominate the field of simultaneous multianalyte bioassays in the future. Its major drawbacks, however, are expensive instrumentation and limited sample throughput. Fig. 3 Schematic diagrams of A competitive ELISA and B noncompetitive (sandwich) ELISA

Enzyme-linked immunosorbent assays (ELISAs) At the beginning of the 1970s, enzymes were introduced as alternatives to radioisotopes as labels in immunoassays [45, 46]. Since that time they have evolved to become the most versatile and popular class of labels for immunoassays. Their application will probably further increase because no radioactive substances are necessary, and the sensitivity of enzyme immunoassays (EIAs) is, nowadays, comparable with that of RIAs and (timeresolved) fluorescence immunoassays. Prerequisites for enzyme-based immunoassays are the possibility of coupling an enzyme to an antibody/antigen without loss of enzyme activity and with no or only limited change in specificity of the immunological component. In the last reaction step of an ELISA the enzyme label catalyzes the conversion of a substrate into a colored or fluorescent product. Although electrochemical detection schemes [46] have also been used for this purpose, the two earlier techniques are predominantly used in routine analysis. Two different types of heterogeneous enzyme immunoassay, the competitive and non-competitive (sandwich) ELISA, are depicted schematically in Figs. 3A

577

and 3B, respectively. In a competitive assay, unlabeled (analyte) and labeled antigens compete for the binding sites of an antibody which is bound to a solid phase. A high signal in the subsequent detection scheme thus indicates the presence of a low analyte concentration, and vice versa. In the sandwich ELISA, the bound antibody reacts first with the analyte of unknown concentration. An enzyme-labeled antibody is then added. This couples to a second binding site of the analyte. The subsequent enzyme-catalyzed reaction leads to a proportional relationship between signal and analyte concentration. In both techniques addition of a blocking reagent is important for reduction of non-specific binding sites at the (microplate) surfaces. Frequent washing steps between the different immunological reactions are required for both methods, however, and lead to long and laborious assay procedures. Automation has, on the other hand, reduced the level of manual work to a minimum and is a major reason for the predominance of ELISAs in the clinical laboratory. Two examples demonstrate the broad applicability of ELISAs in environmental and biomedical analysis. In 1997, Winklmair et al. [48] developed a highly sensitive competitive immunoassay for determination of triazine herbicides in water samples of different origin using a monoclonal antibody. The peroxidase-labeled triazine tracer reacted in the last step with 3,3¢,5,5¢-tetramethylbenzidine (TMB), one of the most commonly used chromogenic substrates, thus enabling photometric detection. Detection limits were between 0.003 and

0.006 lg mL 1 for terbuthylazine, atrazine, and propazine, 0.01 lg mL 1 for simazine, and 0.05 lg mL 1 for deethylterbuthylazine. The results obtained correlated well with additionally performed comparative gas chromatographic measurements. Ikemoto et al. [49] have developed a sandwichimmunoassay system for measurement of human livertype arginase, which could be useful for diagnosis of a variety of hepatic disorders and for follow-up of the post-operative condition of patients. The serum samples and calibration solutions, respectively, reacted first with the absorbed anti-human liver type arginase antibody and then with the POD-labeled second antibody. Finally, UV–visible detection was performed after the POD-catalyzed oxidation of OPD. They found that the arginase concentration in serum increases markedly and temporally at the time of surgical operation or later injury to the liver. Enzyme immunoassays based on other detection methods, for example fluorescence, chemiluminescence, or electrochemistry, have also been developed. Evangelista et al. [25] combined an immunoassay for rat IgG with time-resolved fluorescence detection for aP based on Tb(III) as described above. In 1997, Dou et al. [50] developed an enzyme immunoassay with surface-enhanced Raman scattering (SERS) detection of the reaction product. The performance of this sandwich assay is based on linkage between an anti-mouse IgG, a mouse immunoglobulin G (IgG) (anti-insulin), and an anti-mouse IgG labeled with POD. Finally, the OPD

Fig. 4 The principle of chemiluminescence assay of aP using lucigenin

578

reaction was performed; an aliquot of the reaction mixture was taken, added to a silver colloid solution, and placed in a cylindrical cell. A SERS spectrum of the reaction product was measured by excitation with an Ar ion laser at a wavelength of 514.5 nm. A chemiluminescence enzyme immunoassay with aP as marker, using cortisol-21 phosphate as substrate and lucigenin as chemiluminescence reagent, was presented by Kokado et al. [51] in 1997. By means of this method, human chorionic gonadotropin (hCG), human growth hormone (hGH), a-fetoprotein (AFP), and estradiol (E2) could by determined in human serum or urine samples. The respective antibodies for each analyte were used. A schematic diagram of the aP-catalyzed reaction and the chemiluminescence detection is presented in Fig. 4. In the first step, cortisol is generated, which reduces lucigenin to the chemiluminescent product in the subsequent reaction step. Valentini et al. [52] developed an ELISA with electrochemical detection for screening of the pesticide 1,1,1trichloro-2,2-bis(p-chlorophenyl)ethane (p,p¢-DTT) and its metabolites 1,1-dichloro-2,2-bis(p-chlorophenyl)ethylene (p,p¢-DDE) and 1,1-dichloro-2,2-bis-(4-chlorophenyl)ethane (p,p¢-DDD) in waste waters. The activity of the label enzyme POD was determined using TMB as substrate. Detection was performed by injecting the reaction mixture into a flow-injection analysis system with a thinlayer transducer cell (glassy carbon working electrode, Ag/AgCl reference, stainless steel auxiliary electrode). To summarize, it can be concluded that interest in enzyme immunoassays is very high and likely to stay at this level. This is indicated by the large number of current scientific publications and overview articles [5, 26, 53] on this topic. Major advantages are a combination of very low limits of detection and a very high degree of automation, thus enabling an extremely high sample throughput. Another advantage is applicability in several different fields of research, for example clinical, environmental, or food chemistry.

abling linkage to peptides/proteins. Other coupling principles, however, for example the cyanogen bromide (CNBr) method for covalent particle–biomolecule linkage, are also well-known [54]. Different kinds of nanoparticle can be applied - semiconductor quantum dots (QDs), which contain hundreds or thousands of atoms, are often used [55, 56]. The radius of these particles, generated from group II–VI or group III–V elements (e.g. CdSe-ZnS), is usually between 1 and 10 nm. Their properties are determined by the size - the larger the size of the QD, the higher the maximum fluorescence emission wavelength. Another type of nanoparticle is based on embedding a large number of conventional fluorophores in a protecting polymer [57, 58], or silica [59]. The nanoparticles encapsulate thousands of fluorescent dye molecules (fluorescein, rhodamine), providing a highly amplified and reproducible signal for fluorescence-based bioanalysis. He et al. [60] presented a method for cell recognition in system lupus erythematosus patients that uses photostable luminescent nanoparticles as biological labels. They worked with silica particles, which they doped with the luminescent tris(2,2¢-bipyridyl)dichlororuthenium(II)hexahydrate (Ru(II)(bpy)2+ 3 ). These particles were covalently immobilized with goat anti-human IgG by the CNBr method. The reaction was based on a recognition principle of a surface membrane IgG (SmIgG) from the circulating blood of the test persons and the luminescent nanoparticle-labeled IgG. Comparative measurements with organic dye (fluorescein isothiocyanate)-labeled IgGs resulted in a rapid and significant decrease of fluorescence intensity with time whereas the intensity of the nanoparticles was almost constant. Two years later, the same group published a similar approach with fluorescein isothiocyanate-doped silica particles for recognition of HepG liver cancer cells [60]. A rapid bioconjugated nanoparticle-based bioassay for quantification of pathogenic bacteria (e.g. Escherichia

Nanoparticles as labels In the last few years, interest in the bioanalytical use of nanoparticles has increased enormously. Fluorescent particles of nanometric dimensions are usually applied because of important advantages in comparison with molecular fluorescent dyes. With polymer nanoparticles, for example, the signals obtained are much larger than with individual molecules, because of the presence of a large number of fluorescent centers in one nanoparticle. Depending on the nature of the nanoparticle, there may also be advantages in respect of narrow-banded emission characteristics, which may enable improved simultaneous multianalyte determination. Meanwhile, nanoparticles with anchor groups for linkage to biomolecules are commercially available. These particles are often modified in such a way that functional groups (–COOH, –NH2) are attached to the surface of the particles, en-

Fig. 5 Schematic diagram of a ZnS-capped CdSe quantum dot covalently bound to a protein by mercaptoacetic acid

579

coli) in spiked ground beef samples was presented by Zhao et al. in 2004 [62]. For coupling of the Ru(II)(bpy)2+ particles to the antibody against E. coli, the par3 ticle surface was functionalized with amino groups. The spectrofluorimetric results from the immunoassay applying the nanoparticles correlated well with results from flow-cytometry measurements of the antibodyconjugated nanoparticles bound to single bacterial cells. The possibility of performing the immunoassay on microtiter plates enabled the high-throughput detection of multiple samples. Chan and Warren [63] have developed luminescent semiconductor CdSe–ZnS quantum dot bioconjugates for ultrasensitive non-isotopic detection. These nanoconjugates are biocompatible and are suitable for use in cell biology and immunoassays. The solubility of the QDs in aqueous media was achieved by using mercaptoacetic acid, which bound to the Zn ions of the particle. The polar carboxylic group rendered the QDs water-soluble and enabled protein (here transferrin) coupling. The schematic structure of these QD bioconjugates is shown in Fig. 5. The biocompatibility could be demonstrated in vitro and for living cells by means of fluorescence images. Another approach used for coupling luminescent semiconductor CdSe–ZnS QDs to biomolecules was presented by Goldman et al. in 2002 [64]. In this work, the surface of the QDs was modified with dihydrolipoic acid. The highly positively charged protein avidin was used as a bridge between the bioinorganic nanoparticles and biotinylated antibodies. These particles were successfully used in sandwich-fluoroimmunoassays on microtiter plates for determination of protein toxins (e.g. cholera toxin). First results of the simultaneous analysis on the determination of two protein toxins were also presented. Taylor et al. [65] reported a class of luminescent latex nanobeads covalently linked to DNA-binding proteins that enabled study of specific sequences on single DNA molecules. The latex particles, which contained approximately 100–200 mol of an embedded dye that was protected from the outside environment, conjugated to proteins by amide bond formation. It was demonstrated that the site-specific restriction enzyme EcoRI

can be conjugated to 20-nm fluorescent nanoparticles and that the resulting nanoconjugates have the DNAbinding and cleavage activity of the native enzyme. The feasibility of mapping sequence-specific sites on simple genomic DNA molecules was shown. The increasing number of reviews on the application of nanoparticles in bioassays shows the growing importance and the possibilities of these approaches [19, 66, 67, 68, 69]. Compared with conventional immunoassays, the number of fluorophore molecules which can be attached to an antibody has been significantly increased by embedding many dye molecules in the nanoparticles. This amplification, the reduced photobleaching effects, the safe handling in comparison with radioisotopes, the reduction of one reaction step in comparison with ELISAs, and the possibility of performing multiplexing assays, because of the narrow emission bands of the fluorescent nanoparticles, are advantages which will be used increasingly in the future.

Coupling to separation techniques Enzymatic assays Combination of enzymatic assays with chromatographic separation techniques is found frequently in literature. On the one hand, enzymes are separated on a HPLC column and subsequently reacted with an enzymatic substrate. On the other hand, substrates may be separated and react by post-column derivatization with the enzyme to the respective product. In both cases, the post-column reaction enables detection so that either the substrate or, in the latter case, the enzyme act as label. Wide-pore LC columns are normally used for enzyme separation, because of the large dimensions of the proteins and strong adsorption of the enzymes on the stationary phase. For enzymes or analogues with lower molecular masses, however, for example microperoxidases (MPs), separation on standard columns used for small-molecule separation is possible. The following examples show the applicability of enzymatic assays in

Fig. 6 HPLC set-up for analysis of the explosives TATP and HMTD, including photoreactor and post-column apparatus for enzymatic conversion

580

Fig. 7 Schematic diagram of the on-line continuous-flow system for the determination of fungus samples

separation techniques and in several different fields of application. Heinmo¨ller et al. [70] presented a sensitive HPLC method for separation of a series of n-alkyl hydroperoxides and detection by means of a post-column reaction with POD and p-hydroxyphenylacetic acid (pHPAA), with subsequent fluorescence detection. It was possible to separate seven n-alkyl hydroperoxides from C4 to C18 on an ODS-Hypersil column and to perform subsequent derivatization with only minimally reduced efficiency at greater alkyl chain lengths. Application of POD in post-column reactions for trace analysis of peroxide-based explosives was described by Schulte-Ladbeck et al. [71]. An RP-HPLC method with post-column UV irradiation and fluorescence detection was developed for analysis of triacetonetriperoxide (TATP) and hexamethylenetriperoxide diamine (HMTD). After separation, the analytes were photochemically degraded to H2O2, which was subsequently oxidized in a POD-catalyzed reaction with p-hydroxyphenylacetic acid. The resulting fluorescent dimer of pHPAA was then detected. The set-up of this method is illustrated schematically in Fig. 6. The method enables analysis of post-explosion sites and tracing of residues of either HMTD or TATP. Irth et al. [72] developed an approach for on-line coupling of HPLC with a continuous-flow enzymatic assay with subsequent ESI–MS detection. By means of the arrangement presented in Fig. 7, natural products, e.g. an extracted fungus sample, were separated and reacted with added enzyme in a knitted reaction coil. As long as no bioactive compounds were eluted from the RP-18 column, the enzyme continuously converted added substrate into products (Fig. 7, Eq. 2). Bioactive compounds eluting from the column reacted as enzyme inhibitors (Fig. 7, Eq. 1), which resulted in a decrease of product formation (Fig. 7, Eq.3). Chromatographic immunoassays Several reviews give an overview of the broad topic of chromatographic immunoassays [73, 74, 75]. Shahdeo

and Karnes [74] describe the possibilities of combination of pre- and post-column capillary immunoassay with chromatographic and electrophoretic separation techniques. Hage and Nelson [75] describe the variety of formats and labels and the sensitivity of chromatographic immunoassays. The examples presented here, and others which can be found in the literature, show various approaches in this field. Different measurement techniques have been developed for chromatographic immunoassays. It is possible to perform the assays on-line or off-line, pre- and postseparation. The on-line post-separation technique is often favored, because of the possibility of highthroughput measurements, which is especially important in routine analysis. In 1988, Stone and Soldin [76] presented a HPLC method coupled with off-line immunoassay measurements for determination of digoxin in serum. By means of this method, they improved the accuracy of digoxin determination, which usually suffers from interference from metabolites and endogenous digoxin-like factors, so that overestimation of the digoxin content frequently occurs. The serum samples were extracted, injected into the HPLC system, separated on a RP18 column, and the different fractions were collected separately from each other and finally quantified by a fluorescence polarization immunoassay. Calibration was performed by means of an internal standard. The results from 49 samples from different patients taking digoxin were compared with those from the commonly applied fluorescence polarization immunoassay. The study indicated that for approximately 20% of a general patient population serum digoxin concentration was significantly overestimated by the old method without LC separation. Irth et al. [77], as one example, have published an overview of strategies for on-line coupling of immunoassays to HPLC using either labeled antibodies or labeled antigens (Fig. 8). In the first of these, they used fluorescein-labeled antibodies to monitor the presence of antigenic analytes, for example digoxin and metabolites, in human plasma samples, which were separated by LC. In the first assay step, post-column reaction between the effluent and added labeled antibodies occurred. A sep-

581

Fig. 8 a Set-up of the on-line detection system using fluorescein-labeled antibodies (Ab*). b Schematic diagram of on-line coupling of RPHPLC to immunochemical detection using fluorescein-labeled antigens (Ag*). (Ab Antibody, Ag antigen)

aration step was then performed by means of an immobilized antigen support packed in a short column. Free antibodies were trapped by the support, and the fluorescent antibody–antigen complexes passed through and were detected by means of a fluorescence detector. The high selectivity of this method was shown by a limit of detection of 2·10 10 mol L 1 for digoxin and two metabolites. The second possibility, the use of labeled antigens was also presented by Irth et al. because of the lack of commercially available purified labeled antibodies. The first step of this assay is similar to the first step of the previous assay—the eluting analytes reacted with (in this case unlabeled) antibodies. Fluorescein-labeled antigens were then added and reacted with unlabeled antibodies. Before detection of the generated labeled complexes, they were separated from free fluorescein–labeled antigens by means of an inserted column containing a restricted-access chromatographic support. Low-molecular-mass compounds, as labeled antigens, were retained in the pores of the hydrophobic surface, whereas high-molecular-mass compounds, as the labeled antigen-antibody complex, passed through and could be detected by fluorescence spectroscopy. They applied this technique to avidin–biotin as model system using fluorescein–biotin as labeled antigen. Oates et al. [78] developed and optimized an HPLCbased one-site immunometric assay working with postcolumn chemiluminescence detection of the analyte L-thyroxine (T4). Anti-T4 antibody Fab fragments were conjugated with chemiluminescent acridinium ester labels via lysine residues or terminal amine groups, combined in excess with the analyte, incubated, and then loaded onto a column containing T4. Unbound Fab fragments were retained on the column whereas

T4-bound antibodies passed directly through. Detection could be achieved either by post-column derivatization of the unretained T4-bound antibodies or by reaction of the retained and later eluted non-T4-bound antibodies with H2O2 under alkaline conditions. The chemiluminescence of the generated N-methylacridone was measured. As results could be provided 1.5 min after sample injection, many samples could be analyzed between column-elution steps within a short time. These examples demonstrate the wide range of possibilities of obtaining sensitive and selective results by combining immunoassay techniques with chromatographic methods. One drawback, however, especially when using a solid phase to separate antibody bound and free fractions, is the need to regenerate the stationary phase after each injection. This may cause technical problems and expenditure of time, in particular when measuring the calibration standards. It is, therefore, more difficult to apply these methods in a high-throughput environment.

Conclusions Current labeling strategies for metabolic (enzyme) and affinity (antibody-based) bioassays have been summarized and discussed. It is evident that, despite achievements in the fields of label-free bioassays, labeling techniques will continue to play a leading role in field of bioassays. It can be expected that labeling techniques based on radioactive isotopes will significantly decrease in importance, because of stability, safety, and handling considerations. Enzyme labels, on the other hand, are most likely to take over the leading role for the next

582

years, although extremely strong competition arises from fluorescent nanoparticle labels owing to easy handling, robustness, and compatibility with existing (microplate) spectroscopic instrumentation. It can be expected that the current rapid progress will lead to an even faster implementation of fluorescent nanoparticles into new detection schemes. In the long term there might be significant replacement of ELISAs by nanoparticlebased assays. Bioassays based on a combination of liquid-phase separation and labeling are likely to play an important role in solving analytical problems in which selectivity is most important. On the other hand, they are laborious and still require well-trained personnel for routine applications. Acknowledgements Financial support of the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO, The Hague, The Netherlands) is gratefully acknowledged.

References 1. Green RJ, Frazier RA, Shakesheff KM, Davies MC, Roberts CJ, Tendler SJB (2000) Biomaterials 21:1823–1835 2. Price CP, Newman DJ (1997) Principles and practice of immunoassay, Stockton Press, New York 3. Diamandis EP, Christopoulos TK (1996) Immunoassay. Academic Press, San Diego 4. Deridovich II, Reunova OV (1993) Comp Biochem Physiol 104:23–27 5. Wisdom GB (1976) Clin Chem 22:1243–1255 6. Yalow RS, Berson SA (1960) J Clin Invest 39:1157–1175 7. Coates JE, Lam SF, McGaw WT (1988) Clin Chem 34:1545– 1551 8. Ma Z, Gingerich RL, Santiago JV, Klein S, Smith CH, Landt M (1996) Clin Chem 42:942–946 9. Leveque D, Jehl F (1996) Clin Pharmacokinet 31:184–197 10. Coons AB, Creech HJ, Jones RN (1941) Proc Soc Exp Biol Med 47:200–202 11. Bailey MP, Rocks BF, Riley C (1983) Ann Clin Biochem 20:213–216 12. Bailey MP, Rocks BF, Riley C (1984) Ann Clin Biochem 21:59–64 13. Palmer DA, Evans M, Miller JN (1994) Analyst 119:943–947 14. Khokhar MY, Miller JN, Seare NJ (1994) Anal Chim Acta 290:154–158 15. Lim TK, Nakamura N, Matsunaga T (1997) Anal Chim Acta 354:29–34 16. Bereczki A, Horvath V (1999) Anal Chim Acta 391:9–17 17. Yang HH, Zhu QZ, Qu HY, Chen XL, Ding MT, Xu JG (2002) Anal Biochem 308:71–76 18. Hemmila¨ IA, Webb S (1997) DDT 2:373–381 19. Steinkamp T, Karst U (2004) Anal Bioanal Chem 380:24–30 20. Boyer AE, Lipowska M, Zen JM, Patonay G, Tsang VCW (1992) Anal Lett 25:415–428 21. Daneshvar MI, Peralta JM, Casay GA, Narayanan N, Evans L, Patonay G, Strekowski L (1999) J Immunol Meth 226:119– 128 22. Sowell J, Parihar R, Patonay G (2001) J Chromatogr B 752:1–8 23. Zhao X, Shippy SA (2004) Anal Chem 76:1871–1876 24. Hempen C, Karst U (2004) Anal Chim Acta 521:117–122 25. Evangelista RA, Pollak A, Gudgin Templeton EF (1991) Anal Biochem 197:213–224 26. Ishikawa E (1987) Clin Biochem 20:375–385 27. Athey D, Ball M, McNeill CJ (1993) Ann Clin Biochem 30:570–577 28. Suzuki K (1978) Meth Enzymol 50:456–488

29. Meyer J, Bu¨ldt A, Vogel M, Karst U (2000) Angew Chem Int Ed 39:1453–1455 30. Raba J, Li SF, Mottola HA (1995) Anal Chim Acta 300:299– 305 31. Maeda M (2003) J Pharm Biomed Anal 30:1725–1734 32. Guilbault GG, Brignac PJ, Juneau M (1968) Anal Chem 40:1256–1263 33. Marquez LA, Dunford HB (1997) Biochem 36:9349–9355 34. Tarcha PJ, Chu VP, Whittern D (1987) Anal Biochem 165: 230–233 35. Ma¨kinen KK, Tenovuo J (1982) Anal Biochem 126:100–108 36. Hernandez B, Luque FJ, Orozco M (1996) J Organ Chem 61:5964–5971 37. Zhang K, Cai R, Chen D, Mao L (2000) Anal Chim Acta 413: 109–113 38. Griess P (1871) J Prakt Chem 3:143–144 39. Knoevenagel E (1914) J Prakt Chem 89:25–75 40. Niu SY, Zhang SS, Ma LB, Jiao K (2004) Bull Korean Chem Soc 25:829–832 41. Bergmeyer HU (1974) Methoden der enzymatischen Analyse. VCH, Weinheim 42. Lee ED, Mu¨ck W, Henion JD, Covey TR (1989) J Am Chem Soc 111:4600–4604 43. Hsieh FYL, Tong X, Wachs T, Ganem B, Henion J (1995) Anal Biochem 229:20–25 44. Liesener A, Karst U (2005) Anal Bioanal Chem 382:1451–1464 45. Van Weeman BK, Schuurs AHWM (1971) FEBS Lett 15:232– 236 46. Engvall E, Perlmann P (1971) Immunochemistry 8:871–874 47. Thompson RQ, Barone GC, Halsall HB, Heineman WR (1991) Anal Biochem 192:90–95 48. Winklmair M, Weller MG, Mangler J, Schlosshauer B, Niessner R (1997) Fresenius J Anal Chem 358:614–622 49. Ikemoto M, Ishida A, Tsunekawa S, Ozawa K, Kasai Y, Totani, Ueda K (1993) Clin Chem 39:794–799 50. Dou X, Takama T, Yamaguchi Y, Yamamoto H (1997) Anal Chem 69:1492–1495 51. Kokado A, Tsuji A, Maeda M (1997) Anal Chim Acta 337:335–340 52. Valentini F, Compagnone D, Giraudi G, Palleschi G (2003) Anal Chim Acta 487:83–90 53. Morozova VS, Levashova AI, Eremin SA (2005) J Anal Chem 60:202–217 54. Schall CA, Wiencenk JM (1997) Biotechnol Bioeng 53:41–48 55. Murphy CJ, Coffer JL (2002) Appl Spec 56:16A–27A 56. Bruchez M, Moronne M (1998) Science 281:2013–2016 57. Sahoo PK, Mohapatra R (2003) Eur Polymer J 39:1839–1846 58. Qhobosheane M, Santra S, Zhang P, Tan W (2001) Analyst 126:1274–1278 59. Charreyre MT, Tcherkasskaya O, Winnik MA (1997) Langmuir 13:3103–3110 60. He X, Wang K, Tan W, Li J, Yang X, Huang S, Li D, Xiao D (2002) J Nanosci Nanotech 2:317–320 61. He X, Duan J, Wang K, Tan W, Lin X, He C (2004) J Nanosci Nanotech 4:585–589 62. Zhao X, Hilliard LR, Mechery SJ, Wang Y, Bagwe RP, Jin S, Tan W (2004) PNAS 101:15027–15032 63. Chan WCW, Nie S (1998) Science 281:2016–2018 64. Goldman ER, Balighian ED, Mattoussi H, Kuno MK, Mauro JM, Tran PT, Anderson GP (2002) J Am Chem Soc 124:6378– 6382 65. Taylor JR, Fang MM, Nie S (2000) Anal Chem 72:1979–1986 66. Penn SG, He L, Natan MJ (2003) Current Op Chem Biol 7:609–615 67. Ozkan M (2004) DDT 9:1065–1071 68. Gupta AK, Gupta M (2005) Biomaterials 26:3995–4021 69. Haes AJ, Stuart DA, Nie S, Van Duyne RP (2004) J Fluorescence 14:355–367 70. Heinmo¨ller P, Kurth HH, Rabong R, Tuerner WV, Kettrup A, Ga¨b S (1998) Anal Chem 70:1437–1439 71. Schulte-Ladbeck R, Kolla P, Karst U (2003) Anal Chem 75:731–735

583 72. De Boer AR, Letzel T, Van Elswijk DA, Lingeman H, Niessen WMA, Irth H (2004) Anal Chem 76:3155–3161 73. Hage DS (1998) J Chromatogr B 715:3–28 74. Shahdeo K, Karnes HT (1998) Mikrochim Acta 129:19–27 75. Hage DS, Nelson MA (2001) Anal Chem 73:198A–205A

76. Stone JA, Soldin SJ (1988) Clin Chem 34:2547–2551 77. Irth H, Oosterkamp AJ, Tjaden UR, Van der Greef J (1995) Trends Anal Chem 14:355–361 78. Oates MR, Clarke W, Zimlich A, Hage DS (2002) Anal Chim Acta 470:37–50