Chemiluminescence in Analytical Chemistry

partly overcome by taking advantage of the intrinsic wide dynamic range of CL as detection principle.

When coupled with an optical microscope, CL imaging is a potent analytical tool for the development of ultrasensitive enzymatic, immunohistochemistry (IHC) and in situ hybridization (ISH) assays, allowing spatial localization and semiquantitative evaluation of the distribution of the labeled probe in tissue sections or single cells to be performed.

In this chapter we report recent analytical applications of CL imaging for the detection of biospecific reactions in macrosamples such as microtiter plates of different format (96 or 384 wells), filter membranes and irregular surfaces represented by specimens related to the cultural heritage, and results obtained when the CCD detector is coupled with optical microscopy for enzyme localization, immunohistochemical reactions, and complementary DNA (cDNA) detection (Table 1).

Table 1 Summary of the Applications of BL/CL Imaging Described in the Present Chapter

2. INSTRUMENTATION

The basic instrumentation for CL imaging includes an ultrasensitive video camera, an optical system, and appropriate software for image analysis [21].

The video camera must be characterized by high sensitivity and very low instrument noise to detect the weak CL emission. In the last few years highperformance CCD cameras have substituted for older devices based on Vidicons. Cooled, back-illuminated CCD cameras provide very high detection sensitivity owing to their high quantum efficiency and low background noise and allow a quantification of emitted light at a single-photon level [22–26]. Moreover, these devices do not require an image intensification step, which could negatively affect the image quality and the signal-to-noise ratio [19, 20, 23, 27]. However, relevant limitations of CCD-based imaging, in particular in quantitative analysis, are the narrow dynamic range (2–3 decades lower) and the lower sensitivity (5–10 times lower) with respect to photomultiplier-based detection [24].

For macrosample analysis, the CCD device is connected to standard or custom camera optics and enclosed in a lighttight box. In this configuration the instrument can be used for reading CL emission in 96or 384-well microtiter plates (luminographs able to read four standard plates simultaneously are commercially available), on target surfaces such as gels, thin-layer chromatography plates, or dot blot membranes [28], and in other kinds of samples. Several commercial instruments can be used not only for detecting CL emission, but also for fluorescence (using additional UV sources) or to perform densitometric measurements using a transilluminator. In this way even fluorescence or colorimetric measurements can take advantage of the high sensitivity of the CL detector and the easy acquisition and quantification of the signal.

In conjunction with an optical microscope, a CCD camera can be used to localize the light emission from tissues or cells, and to obtain semiquantitative information on the localization of the probed species [27, 29–35]. Also in this case, to avoid interference from ambient light, a lighttight box is required. Owing to the very low light intensities involved in CL measurements, the potential loss of light in the optical system should be minimized to achieve the maximum analytical detectability. Therefore, the microscope should have a lens coupling system that is as simple as possible and objectives with the highest numerical aperture compatible with focal aberration and depth of field.

In CL measurements many factors that influence the intensity of the CL signal should be taken into account. The CL signal may depend on the geometry of the sample. Internal refraction and reflection at the air-solution interfaces are important factors in determining the measured CL intensity, and should be taken into account, for example, when a CL cocktail is placed over a sample. The effect of sample geometry can be evaluated using model systems, such as enzymes

chemically immobilized on calibrated nylon net, controlled-porosity glass particles, and macroporous acrylic beads [25, 36]. In general samples and references should be analyzed employing the same experimental conditions to minimize any kind of interference due to the experimental setup.

The CL signal measured in a white microtiter plate is several times higher than that obtained in a black one, owing to the higher reflectivity of the well surfaces in the white plate. In preliminary experiments, we observed that when the CL signal is measured in a 384-well microtiter plate using a luminograph the values obtained for the wells lying in the peripheral zone of the plate are affected by serious underestimation (up to 30%), owing to the system geometry [37]. Since this effect is reproducible, depending only on the position and the volume content of the wells, the use of a correction factor, which multiplies the measured CL signal of each well depending on its position, made it possible to equalize the signal over the entire microtiter plate area.

When calculating the intensity of the light emitted by a sample using the data obtained with an imaging device, the light collection angle must be considered [23]. The light collection solid angle (Ω) of an optical device is related to the aperture of the collection lens by the relationship

Ω 2π(1 cosα)

where α is one-half of the aperture angle (i.e., one-half of the light collection angle). The overall collection efficiency of the lens system (η) is given by the ratio between the light collection solid angle and the full solid angle (4π). Therefore,

η (1 cosα)/2

The maximum light collection efficiency (η 0.5) is achieved when the emitting sample is in close contact with the detector. When the detector is coupled with a lens, the light collection efficiency may be much lower.

It should be pointed out that the quality of a CL image (and therefore the precision of a quantitative measurement or the sharpness of a CL probe localization) depends not only on the absolute value of the CL signal but also on its signal-to-noise ratio. Since most of the low-light imaging devices are integrating devices, the signal-to-noise ratio can easily be improved by increasing the exposure time, which means accumulating the CL signal for a longer period [23]. In the ideal case, an increase in the exposure time by a factor N determines an increase of the signal-to-noise ratio of a factor √N.

In our studies the detection and analysis of chemiluminescent signals were performed using two different high-performance, low-light-level imaging apparatuses, the Luminographs LB 980 and LB 981 (EG&G Berthold, Bad Wildbad, Germany), which permit emitted light measurement at the single-photon level.

The instrument setup and CL imaging processing are quite similar for the two devices, which differ as concerns the video systems. The LB 980 is provided with a 1-in. Saticon, high-dynamic-range pickup tube (which is a Vidicon-type tube with Se-As-Tl light target photoconductor) linked to an image intensifier, by high-transmission lenses, and also to a videoamplifier. The LB 981 luminograph, based on a back-illuminated, slow-scan, cooled CCD without intensification stages, was used when higher detectability was required. In both cases the video system is connected to a PC for quantitative image analysis, and a sample dark box is provided to prevent contact with external light. The videocamera can be connected to a Model BH-2 Optical Microscope (Olympus Optical, Tokyo, Japan) also enclosed in a dark box. The system operates in the following consecutive steps: (1) samples are recorded as transmitted light; (2) the luminescent signal is measured with an optimized photon accumulation lasting 1 min; and (3) after a computer elaboration of the luminescent signal with pseudocolors corresponding to the light intensity, an overlay of the images on the screen provided by the transmitted light and by the luminescent signal allows the spatial distribution of the target analytes to be localized and evaluated. The light emission from the sample is quantified by defining a fixed area and counting the number of photon fluxes from within that area.

3. CHEMILUMINESCENCE REAGENTS IN IMAGING

Labeling with enzymes is generally preferred to labeling with CL substances because of the possibility to obtain a higher sensitivity owing to amplification of the CL signal in the presence of an excess of the CL substrate. Moreover, with enzyme labels glow-type light emission kinetics can be usually obtained (e.g., the CL emission reaches a steady-state intensity), which permits both easy handling and standardization of the experimental conditions and quantitation of the labeled probe under investigation, because the steady-state light intensity is directly related to the enzyme activity.

AP and HRP are probably the most widely used CL labels, owing to the availability of many different CL substrates that originate glow-type emission kinetics and permit the detection of very small amounts (in the order of 10 18–10 21 mol) of CL label. The most sensitive and widely investigated CL substrates for AP are based on adamantyl-1,2-dioxetane aryl phosphate derivatives [5, 13], while commercial substrates for HRP are mainly luminol-based reagents (Fig. 1). Adamantyl 1,2-dioxetane aryl phosphates are dephosphorylated by AP to give an unstable intermediate; decomposition of this intermediate produces an excited-state aryl ester that emits light. Luminol and its derivatives are oxidized by peroxides (usually hydrogen peroxide) giving light emission. This

Figure 1 BL and CL reactions commonly used in bioanalytical imaging.

CL reaction can be catalyzed by enzymes other than HRP (e.g., microperoxidase and catalase) and by other substances [hemoglobin, cytochrome c, Fe(III), and other metal complexes]. The presence of suitable molecules such as phenols (p- iodophenol), naphthols (1-bromo-2-naphthol), or amines (p-anisidine) increases the light production deriving from the HRP-catalyzed oxidation of luminol and produces glow-type kinetics [6, 7]. The use of other enzymes, such as glucose- 6-phosphate dehydrogenase [38–41], β-galactosidase [42], and xanthine oxidase [43–46], as CL labels has been reported.

Bioluminescent reactions are also employed for imaging purposes, in particular the firefly and the bacterial luciferin/luciferase ones (Fig. 1). The firefly luciferin/luciferase reaction requires ATP, magnesium ions, and oxygen. Many different luciferins and mutant luciferases have been investigated to optimize the

reaction performance. Bioluminescent bacterial systems involve bacterial luciferases that catalyze the oxidation of FMNH2 and bacterial luciferin (a long-chain aldehyde) in the presence of oxygen. In vivo, the luciferase is coupled to an oxidoreductase that catalyzes the oxidation of NAD(P)H leading to the formation of FMNH2. In vitro, FMNH2 can also be obtained by various chemical means.

Most of the commercially available CL substrates can, in principle, be used for imaging purposes. However, in general neither the substrates nor the experimental conditions are optimized for imaging. A crucial factor in determining the applicability of a CL substrate in imaging, particularly as concerns their use in the imaging of microsamples through an optical microscope [47], is the diffusion of the excited species responsible for the CL emission. In fact, to achieve a sharp localization of the labeled probe, the light emission should occur as near as possible to the site of the primary biospecific recognition. The use of CL substrates that produce excited species with quite long half-life results in poor spatial resolution in localization of the labeled probe because the excited species diffuse in the solution before light emission occurs. CL substrates suitable for imaging should therefore be characterized by very short-lived excited species to make diffusion of the emitting species negligible. Resolution can also be improved by increasing the viscosity of the chemiluminescent cocktail solution by adding substances like gelatin, glycerol, or polyvinylpyrrolidone at suitable concentrations.

The problem related to the diffusion of the species involved in the CL reaction is even greater when CL derives from a chain of enzymatic reactions, which is a rather common situation in enzyme activity determinations. In this case several diffusion processes, involving both the emitting species and the intermediate reactive species (or enzymes), could contribute to the decrease of spatial resolution in localization of the labeled probe. Therefore, careful optimization of the experimental conditions and of the concentrations of all reagents is required. The 3α-hydroxysteroid dehydrogenase (3α-HSD) enzyme, immobilized on a nylon net, was localized by means of coupled enzymatic reactions involving the bacterial enzymes FMN-oxidoreductase and luciferase. Localization of 3α-HSD was very poor with the two bioluminescent enzymes free in solution (Fig. 2a). It improved using a nylon net on which FMN-oxidoreductase was also immobilized (Fig. 2b), and became quite good with all the enzymes coimmobilized (Fig. 2c) [25, 48]. The image quality can be improved by increasing the viscosity of the solution and the concentration of the indicator bioluminescent enzymes.

The analytical detectability applying a CL method should, in principle, be comparable to that obtained using radioactive labels, without all the disadvantages related to the use of isotopic labeling. In fact, assuming reasonable values for the quantum efficiency of the chemiluminescent reaction (ΦCL 0.01), for the overall photon collection efficiency of the optical system-CCD camera assembly (η 0.01%), and for the intensity of the lowest detectable CL signal (about

Figure 2 Effect of enzyme immobilization on luminescent image spatial resolution evaluated using coupled enzymatic reactions on nylon net as a model system. (a) Immobilized 3α-hydroxysteroid dehydrogenase; (b) immobilized 3α-hydroxysteroid dehydrogenase and FMN-NADH oxidoreductase; (c) immobilized 3α-hydroxysteroid dehydrogenase, FMN-NADH oxidoreductase, and bacterial luciferase. (From Ref. 47. Copyright John Wiley & Sons Ltd. Reproduced with permission.)

1 photon/s/pixel), the detection limit using a CL label may be of the order of 10 18 mol. If an enzyme is used as a label instead of a CL substrate, amplification factors of 103–104 can easily be obtained; thus the detection limit can be as low as 10 21 mol of enzyme. In practice, the detection limit depends not only on the absolute intensity of the CL signal, but also on the signal-to-noise ratio. Therefore, among different CL systems the maximum sensitivity is not necessarily obtained using the CL system that gives the maximum emission intensity, because even if it is desirable to have an intense absolute signal it is more important to have a high signal-to-noise ratio to improve the image quality and to allow sensitive and specific quantification of the analyte.

4. CHEMILUMINESCENT ANALYSIS OF MACROSAMPLES

4.1 Filter Membranes

As stated earlier, the main advantage of CL imaging with respect to other detection systems is the possibility to perform a direct and rapid quantitative evaluation of the signal over a relatively wide dynamic range [34]. The analytical performance of CL imaging is then comparable or superior to that of systems using different principles to detect immunological or genetic reactions, such as radioisotopes or color-producing substrates [11, 49], or even of CL with photographic detection. Moreover, CL images can be permanently recorded for archiving or further elaboration and easily exchanged with other laboratories.

Nucleic acid hybridization techniques are able to detect viral genomes directly in clinical samples, allowing rapid and sensitive diagnosis of viral infections, especially for viruses that do not grow in cell cultures or have a long replication cycle. In CL dot blot hybridization reactions the specimens are dotted on a membrane, hybridized with a specific gene probe labeled with a hapten, and then the hybrid is detected using an enzyme-conjugated antihapten antibody and a suitable CL substrate [50–52]. The analytical performances of several CL substrates for HRP or AP were compared in dot blot hybridization assays for the detection of B19 parvovirus DNA [32, 53]. The assays used digoxigenin-labeled DNA probes that were immunoenzymatically revealed using antidigoxigenin Fabs (antigen-binding fragments) conjugated with HRP or AP. The detection limits were between 0.5 and 2 pg of target homologous DNA using HRP as label, and between 10 and 50 fg of DNA using AP as label. Since the detection limits for colorimetric methods are about 5 pg and 100 fg when HRP and AP are used as labels, respectively, the chemiluminescent method was superior to colorimetry.

4.2 Microtiter Plate Format

Chemiluminescent imaging can be used to measure CL emission in microtiter plates, even if the performance of imaging devices is not comparable to that of the standard luminometers used for this purpose. In particular, both the sensitivity and the dynamic range of imaging devices are lower than those of photomulti- plier-based microtiter plate readers. However, commercial imaging devices make it possible to measure simultaneously, in a minute or less, up to four microtiter plates (corresponding to up to 4 384 microtiter wells). This time is comparable to or shorter than that required by standard microtiter plate readers, making this technique interesting in the development of high-throughput-screening (HTS) analytical methods based on CL. Such methods could be used, for example, in screening the identity and biological activity of new compounds synthesized using combinatorial chemical synthesis. Another potential application field for im-

aging devices are CL analyses based on the kinetics of the CL emission [54]. In fact, standard luminometers, which read the CL emission well by well or strip by strip, would not be able to follow relatively fast-emission kinetics across an entire microtiter plate. Despite these potential advantages of imaging techniques, as far as we know no data have been reported in the literature on this type of application.

A CL method has been used to evaluate the antioxidant activity of natural compounds, in comparison to that of a reference compound, in a 384-well microtiter plate [37, 55]. In fact, antioxidant compounds are able to temporarily interrupt the light output of a luminol-based CL system because they act as radical scavengers and block the formation of the luminol radicals that drive the CL reaction. Light emission is restored after an interval that is directly proportional to the amount and activity of the antioxidant added to the CL solution. The same analytical format has been used for determination of the activity of acetylcholinesterase (AChE) inhibitors [37, 56]. For this determination, AChE was involved with choline oxidase and HRP in a chain of enzymatic reactions leading to light emission. Under suitable experimental conditions (i.e., excess of both choline oxidase and HRP) the intensity of the CL emission is proportional to the activity of AChE. Therefore, the addition of AChE inhibitors to the system results in a reduction of the light output proportional to the enzyme inhibition. Both methods were proven to be suitable for analysis of the biological activity of a large number of samples (about 200 for each 384-well microtiter plate) in a short period; the results obtained for the AChE inhibitors were comparable to those obtained using conventional color-producing AChE substrates to evaluate the inhibitory activity.

We have developed chemiluminescent immunoenzymatic assays for β-ago- nist drugs in the 96-well-microtiter-plate format. Such competitive assays have been used for determination of clenbuterol and of the overall content of β-agonist drugs in the sample. They matched the standard requirements of precision and accuracy, and were more sensitive compared to the conventional colorimetric methods. Moreover, CL detection was very rapid, making these assays suitable for screening analysis.

4.3 Irregular Surfaces

A rapid, nondestructive method based on determination of the spatial distribution of ATP, as a potential bioindicator of microbial presence and activity on monuments, artworks, and other samples related to the cultural heritage, was developed [57]. After cell lysis, ATP was detected using the bioluminescent firefly luciferinluciferase system and the method was tested on different kinds of surfaces and matrices. Figure 3 reports the localization of biodeteriogen agents on a marble specimen. Sample geometry is a critical point especially when a quantitative analysis has to be performed; however, the developed method showed that with opti-

Figure 3 Localization of biodeteriogen agents on a marble sample by luminescent ATP detection using the firefly luciferin-luciferase system. (Courtesy of Dr. G. Ranalli, University of Molise, Campobasso, Italy.)

mized experimental conditions an accurate evaluation of the spatial distribution can be achieved. Along with rapidity, nondestructivity, and sensitivity, the most interesting feature of such a technique is the potential use as a rapid diagnostic tool for in situ applications.

4.4 Electrodes

CL imaging can also be a tool for evaluating the distribution of a protein (or another biomolecule) immobilized on a given solid support. It can be used, for example, to study the immobilization of a protein on the gold surface of a quartz microbalance electrode via disulfide bond formation with the activated metal surface. The quartz microbalance makes it possible to easily determine the amount of protein on the electrode, but does not give information on its distribution on the electrode surface. By using a suitably labeled protein, CL imaging could allow comparison of different immobilization procedures in terms of both the amount of protein immobilized and its distribution on the solid support. A protein A-HRP conjugate was immobilized on a quartz microbalance electrode (Fig. 4a)