Aequorin-Based Homogeneous Cortisol Immunoassay for Analysis of

Oct 18, 2007 - Homogeneous assays are attractive because they are performed in only one phase, namely, the liquid phase, and thus, they do not require...
0 downloads 0 Views 106KB Size
1772

Bioconjugate Chem. 2007, 18, 1772–1777

Aequorin-Based Homogeneous Cortisol Immunoassay for Analysis of Saliva Samples Laura Rowe,† Sapna Deo,‡ Josh Shofner,† Mark Ensor,† and Sylvia Daunert*,† Department of Chemistry, University of Kentucky, Lexington, Kentucky 40506. Received February 2, 2007; Revised Manuscript Received July 24, 2007

Homogeneous assays are attractive because they are performed in only one phase, namely, the liquid phase, and thus, they do not require separation of phases as their heterogeneous counterparts do. As opposed to heterogeneous assays, the signal generation in a homogeneous assay is a direct result of analyte binding, which allows the multiple washing and incubation steps required in an indirect heterogeneous assay format to be eliminated. Moreover, homogeneous assays are usually fast and amenable to miniaturization and automation. In this article, we describe the development of a homogeneous assay for the hormone cortisol using the bioluminescent photoprotein aequorin as a reporter molecule. A cortisol derivative was chemically conjugated to the lysine residues of a genetically modified aequorin in order to prepare an aequorin–cortisol conjugate capable of binding anticortisol antibodies. The binding of anticortisol antibodies to the aequorin–cortisol conjugate resulted in a linear response reflected in the emission of bioluminescence by aequorin. A competitive binding assay was developed by simultaneously incubating the aequorin–cortisol conjugate, the anticortisol antibodies, and the sample containing free cortisol. Dose–response curves were generated relating the intensity of the bioluminescence signal with the concentration of free cortisol in the sample. The optimized homogeneous immunoassay produced a detection limit of 1 × 10-10 M of free cortisol, with a linear dynamic range spanning from 1 × 10-5 to 1 × 10-9 M. Both serum and salivary levels of cortisol fall well within this assay’s linear range (3.0 × 10-7 M to 7.5 × 10-7 M and 1.0 × 10-8 M to 2.5 × 10-8 M, respectively), thereby making this assay attractive for the analysis of this hormone in biological samples. To that end, it was demonstrated that the assay can be reliably used to measure the concentration of free cortisol in saliva without significant pretreatment of the sample.

INTRODUCTION Homogeneous assays offer the distinct advantage of utilizing relatively simple, “mix and measure” techniques. In a homogeneous assay, the binding of an analyte to its respective binder results in a direct and quantifiable change in signal (1). For this reason, homogeneous assays do not require the multiple washing, filtration, and incubation steps that are often required in heterogeneous assays. Therefore, homogeneous assays are more amenable to automation and high-throughput screening adaptation. Traditional enzymatic immunoassays usually do not work in a homogeneous format. Instead, radioisotopic methods, such as scintillation proximity assays, are often employed. However, the use of radioactive substances creates problems in terms of cost, safety, and disposal when dealing with the large reagent volumes necessary for high-throughput screening. In an effort to reduce these problems, there has been a push to increase the availability and practicality of nonisotopic homogeneous assay methods, utilizing techniques such as fluorescence resonance energy transfer (FRET), fluorescence polarization, and fluorescence correlation spectroscopy (2). These homogeneous fluorescence methods have been very successful in reducing the cost and increasing the efficiency of high-throughput drug screening. Problems are encountered, however, as reagent volumes continue to decrease, approaching the sensitivity limit of fluorescence detection. The fluorescent background present in biological fluids can drastically decrease the signal-to-noise ratio when employing fluorescent detection * E-mail address: [email protected]. † University of Kentucky. ‡ Current Address: Department of Chemistry, Indiana University Purdue University Indianapolis, Indiana 46202.

methods in such small sample volumes. In contrast, homogeneous assays based on chemiluminescence or bioluminescence afford the detection limits needed in small volume analysis and are uniquely advantageous in terms of their sensitivity, lack of background noise in biological fluid, and affordability (3). Herein, we report the development of such a bioluminescence homogeneous assay for the hormone cortisol employing the recombinant photoprotein aequorin. This assay, in addition to the aforementioned advantages, is also capable of directly detecting cortisol in saliva samples (4). Cortisol is a steroid hormone that is important for the regulation of both metabolic rates and growth rates. Abnormal cortisol levels are indicative of a wide variety of diseases, such as Addison’s disease and Cushing’s syndrome or the onset of diabetes (5). In order to diagnose and properly treat such diseases, a physician must periodically assay a patient’s level of free cortisol. However, hormones such as cortisol follow circadian rhythms and require multiple samplings over the course of a day. Noninvasive sampling, such as saliva collection, is therefore preferred over more invasive methods, such as plasma collection. Since the hormone concentration detected in saliva is the nonprotein bound fraction of cortisol, the most important fraction when diagnosing pathological conditions, saliva analysis is as practical as serum analysis for clinical diagnoses (6). Another advantage is that the individual concentration of the corticosteroid-binding globulin does not affect the results in salivary cortisol analysis, as it does in serum cortisol analysis. However, the concentration of hormones in saliva is generally 10–50-fold lower than the concentration of hormones in plasma (7). For this reason, assays that detect salivary levels of hormones must be more sensitive than the corresponding serum-based assays.

10.1021/bc070039u CCC: $37.00  2007 American Chemical Society Published on Web 10/18/2007

Aequorin-Based Cortisol Immunoassay for Analysis of Saliva

Figure 1. Schematic of how the homogeneous response works in the assay. The bioluminescence intensity of aequorin (blue ball) increases when cortisol Ab (red figures) binds to the cortisol molecules (green arrows) that are conjugated to the aequorin protein via its lysine residues.

In the homogeneous immunoassay described herein, we employed the bioluminescent photoprotein aequorin as a label (Figure 1). Since bioluminescence is a rare phenomenon in terrestrial organisms, there is virtually zero background bioluminescence found in human fluid, and this allows for the highly sensitive detection of the signal emitted by aequorin (8). Aequorin is a calcium-activated photoprotein isolated from the jellyfish Aequorea Victoria that is composed of apoaequorin, coelenterazine, and molecular oxygen (9). Coelenterazine, an imdazopyrazinone chromophore, is located in the central, hydrophobic, reaction pocket of apoaequorin. Following the binding of calcium, the protein undergoes a conformational change which leads to the oxidation of coelenterazine to coelenteramide. The relaxation of this excited coelenteramide results in the emission of bioluminescence at 469 nm (9). This calcium-induced light emission from aequorin has led to its use as an in ViVo calcium indicator and as an ultrasensitive label for a variety of immuno- and DNA hybridization assays (10–19). In order to develop a homogeneous assay for cortisol, cortisol3-O-carboxymethyloxime N-hydroxysuccinimidyl ester (Cortisol-3-CMO-NHS) was first conjugated to the lysine residues of a cysteine-free aequorin mutant. The cysteine-free mutant of aequorin, in which all three native cysteine residues have been changed to serine residues, was selected on the basis of previous studies performed in our laboratory that demonstrated the higher bioluminescent activity of this mutant, as compared to wild-type aequorin (20). Several conjugates were prepared, which differed in their N-hydroxysuccinimide cortisol-toaequorin ratio so that the conjugate that exhibited the largest signal change could be selected for further studies. The conjugate that presented optimal signal generation as well as binding ability to the anticortisol antibody was chosen for the development of the homogeneous assay. The assay was then characterized and optimized in terms of reproducibility, precision, and accuracy in untreated saliva samples in order to establish the analytical performance of the assay. Finally, the assay developed was employed in the determination of cortisol in human saliva samples.

EXPERIMENTAL PROCEDURES Reagents. Tris (hydroxymethyl) amino methane (Tris) free base, ethylenediaminetetraacetic acid (EDTA) sodium salt, glucose, sodium chloride, calcium chloride, bovine serum albumin, kanamycin, sodium dodecyl sulfate (SDS), and all other reagents were purchased from Sigma (St. Louis, MO). Luria Bertrani (LB) broth was from Gibco BRL (Gaithersberg, MD). Coelenterazine was purchased from Biotium (Hayward,

Bioconjugate Chem., Vol. 18, No. 6, 2007 1773

CA). Cortisol-3-CMO-NHS and monoclonal anticortisol antibody were from United States Biological (Swampscott, MA). 96-well polystyrene microtiter plates were purchased from Nunc (Rochester, NY). Bradford assay kits were obtained from BioRad Laboratories (Hercules, CA). The buffers were composed as follows: buffer A, 30 mM Tris, 4 mM EDTA, pH 7.5; buffer B, buffer A with 0.1% Tween-20; buffer C, 30 mM Tris, 100 mM CaCl2, pH:7.5; buffer D, 100 mM NaHCO3, 4 mM EDTA, pH 8.2. Deionized water purified with a Milli-Q System (Waters, Milford, MA) was used to prepare all solutions. Apparatus. Cell cultures were grown in a Forma Scientific orbital shaker (Marietta, OH), and the optical density was determined using a Milton Roy Spectronic 21D spectrophotometer (Ivyland, PA). Cell cultures were centrifuged using a Beckman J2-M1 centrifuge (Palo Alto, CA). Fractions were lyophilized with a Christ Alpha 2-4 lyophilizer (Osterode am Harz, DE). Triplicate bioluminescence measurements were made on a Dynex MLX Microtiter Plate Luminometer (Vienna, VA). All values reported are the mean of three replicates, which have been background-corrected. Methods. Expression and Purification of Apoaequorin. The plasmid pSD110 was initially constructed in our laboratory to contain the gene for a cysteine-free mutant of apoaequorin in which all three cysteine residues were mutated to serine residues (19). This plasmid was transformed into competent Bacillus subtilis cells. The cells containing the aequorin gene were then grown to an OD of 0.6, and 1 mM IPTG was added in order to induce the expression of the apoaequorin. The supernatant from the cells was harvested, acid precipitated with glacial acetic acid to a pH of 4.2, and centrifuged for 30 min at 9500 g, 4 °C, in order to pelletize the precipitated protein. The protein pellet was resuspended and filtered through a 0.2 µm filter; the pH was then adjusted to 7.5. The resuspended protein was then purified with a HQ ion exchange chromatography column (18). Purity of the protein was verified with SDS-PAGE, and concentration was determined with the Bradford Assay. Conjugation of Aequorin to Cortisol. The apoprotein was dialyzed against buffer D. A solution of cortisol-3-CMO-NHS dissolved in anhydrous dimethyl sulfoxide was then added dropwise to the apoprotein, which was stirring in a 5 mL conical vial, at 4 °C. The reaction was allowed to proceed overnight at 4 °C with slow stirring. The stock solution of the cortisol-3CMO-NHS in anhydrous dimethyl sulfoxide was prepared immediately prior to use and added to the protein in the following stoichiometric ratios of aequorin/cortisol: 1:1500, 1:2000; 1:2500. Bioluminescence Emission of the Aequorin–Cortisol Conjugates. The microtiter plate wells for this, and all subsequent experiments, were blocked for nonspecific binding by allowing 300 µL of TE buffer containing 4 mg/mL bovine serum albumin to incubate overnight in the wells at 4 °C. All glass test tubes employed for initial mixing during this, and all subsequent, experiments were also blocked with BSA prior to use in an analogous manner. The wells were then emptied and allowed to air-dry prior to use. The apoaequorin–cortisol conjugates were dialyzed against buffer A and then charged with 100 µg/mL native coelenteraine overnight at 4 °C. Serial dilutions of the conjugates (10-8 to 10-13 M) were prepared using buffer A. 50 µL of the conjugate dilutions were added to the blocked wells, and signal was collected for 5 s on the luminometer, following the injection of 100 µL of buffer C. Bioluminescence Half-Life. The conjugates were diluted until total emitted light was less than 5000 RLUs within a 4 s integration time. Ten microliter aliquots of each conjugate were then injected with 100 µL of buffer C, and light measurements were collected every 0.1 s for 4 s on an MGM Optocomp 1 test tube luminometer (Hamden, CT). The linear portion of the decay

1774 Bioconjugate Chem., Vol. 18, No. 6, 2007

kinetics curve was then plotted with time (s) versus ln (intensity). The half-life was then determined using y ) mx + b, m ) -k, t1/2 ) .697/k. Binder Dilution CurVes. Anticortisol antibody (330 µg/mL) was serially diluted in buffer A in order to obtain the following dilutions: 33 µg/mL, 3.3 µg/mL, 3.3 × 10-1 µg/mL, 3.3 × 10-2 µg/mL, 3.3 × 10-3 µg/mL, 3.3 × 10-4 µg/mL, 3.3 × 10-5 µg/mL, and 3.3 × 10-6 µg/mL. A volume of 50 µL of the cortisol antibody was incubated with a 50 µL volume of 1 × 10-11 M conjugate (1:1500, 1:2000, 1:2500) for 30 min at room temperature with shaking at 250 rpm. Luminescence measurements were then collected over a 5 s time period following the injection of 100 µL of buffer C. Dose–Response CurVes. A stock solution of 10-2 M cortisol was prepared in methanol, and 10-3 to 10-11 M cortisol dilutions were subsequently made from this stock solution using buffer A. From the binder dilution study, a concentration of 3.3 µg/mL cortisol antibody was selected. A volume of 50 µL of various concentrations of cortisol and a volume of 50 µL cortisol antibody were incubated with 50 µL of the 10-11 M 1: 2500 conjugate in blocked microtiter plate wells. The plate was incubated for 30 min at room temperature with shaking at 250 rpm. The luminescence measurements were then collected in a manner identical to the protocol discussed in the binder dilution study. An initial dose–response curve was produced, and the LOD and dynamic range were calculated from this first curve. Subsequent dose–response curves were produced over a period of several days, and these curves were used during the analysis of the saliva samples. Analysis of SaliVa Samples. Saliva samples were collected from six healthy subjects using the following protocol. Subjects rinsed their mouths out with water after they awoke in the morning and collected their saliva 15 min later in two plastic 1.7 mL tubes. Saliva samples were stored at -80 °C until analysis. The samples were thawed and centrifuged at 3000 rpm, 4 °C, just prior to analysis. The supernatant was used for the study. For this study, the buffer used for all dilutions was buffer B. A volume of 50 µL of the 3.3 µg/mL cortisol antibody, 50 µL of the saliva supernatant, and 50 µL of 10-11 M 1:2500 conjugate was combined in preblocked wells, incubated at room temperature, with shaking at 250 rpm, for 30 min. Signal was collected from each well for 5 s on the Dynex luminometer following the injection of 100 µL of buffer C. This process was repeated three times, on three separate days, in order to determine the intraassay precision of the salivary cortisol measurements. RecoVery Studies in SaliVa. For this study, 1.25 µL of serial dilutions of cortisol standard were added to a pool of saliva (pool 6). Pool 6 refers to the fact that each subject had two separate Eppendorf tubes containing their saliva. The two samples were pooled together to make pool 1 through pool 6. Pool 1 is the combined saliva from subject 1, pool 2 is the combined saliva from subject 2, and so forth. 1.25 µL of 1 × 10-3 to 1 × 10-6 M cortisol standards were added to 250 µL of saliva in order to spike the saliva with 5, 50, 500, and 5000 nmol/L of cortisol. Analysis of the samples was identical to the method detailed in the Analysis of Saliva Samples section.

RESULTS A cysteine-free mutant of aequorin, in which the cysteines at positions 145, 152, and 180 were all mutated to serine residues, was employed as the label for this assay. This cysteinefree aequorin has been shown to be more active than the wildtype protein. A plasmid was previously constructed in our laboratory that encoded for this apoaequorin mutant (21). This plasmid was transformed into Bacillus subtilis cells, and the recombinant apoaequorin was then expressed in the cells and

Rowe et al.

Figure 2. Bioluminescence curve generated by the 1:2500 aequorin– cortisol conjugate. The bioluminescence intensity decreases linearly as the concentration of the conjugate decreases. Data are the mean of three replicates, which have been blank subtracted, (1 standard deviation.

purified with ion exchange chromatography. Following purification, the cysteine-free aequorin was chemically conjugated to a cortisol-3-CMO-NHS through the lysine residues of aequorin. Conjugates with aequorin–cortisol derivative initial reaction molar ratios of 1:1500, 1:2000, and 1:2500 were prepared and characterized in terms of their bioluminescent activity and their ability to bind anticortisol antibodies. The bioluminescence of the aequorin–cortisol conjugates was found to be stable for over three months when stored at -80 °C. In order to study the bioluminescent activity of the conjugates, light was collected for 5 s following the addition of a calcium-containing bioluminescence triggering buffer. All three conjugates exhibited acceptable activity, which progressively decreased as the concentration of the cortisol derivative used during conjugation increased. For example, the 1:1500 conjugate exhibited higher activity than the 1:2000 conjugate, and the 1:2000 conjugate was more active than the 1:2500 conjugate. The conjugates were characterized in terms of their half-life to determine whether there were any changes in decay profiles caused by the conjugation. Half-lives were obtained by collecting light for 4 s, at 0.1 s intervals, following the addition of a calcium-containing buffer. The half-life data were analyzed in the same manner as previously described (12). The 1:1500, 1:2000, and 1:2500 conjugates gave successively increasing half-lives of 1.48, 1.51, and 1.96 s, respectively. The aequorin–cortisol conjugates were then characterized in terms of their bioluminescence properties. For that, a curve was generated that related the concentration of the conjugate to the intensity of the bioluminescence signal. From this curve, the lowest possible concentration of the aequorin–cortisol conjugate that could be employed in the development of the assay was determined, i.e., the lowest concentration of conjugate that still exhibited significant bioluminescence emission when compared to a blank control. For this study, the conjugates were first serially diluted with buffer A, and their bioluminescence intensities were then measured over a 5 s time period. From this curve, it was determined that a concentration of 1 × 10-11 M conjugate should be employed in all further studies (Figure 2). In order to determine the antibody concentration that would produce an optimal change in signal, a binder dilution study was performed employing a 1 × 10-11 M concentration of the aequorin–cortisol conjugate. For this, 50 µL of the conjugate was incubated with 50 µL of different concentrations of anticortisol antibody ranging from 33 to 3.3 × 10-6 µg/mL. Following a 30 min incubation at room temperature with shaking, the bioluminescence intensity of the conjugate–antibody solution was measured over a period of 5 s in a microtiter plate luminometer. Analysis of the binder dilution curve resulted in the selection of the 3.3 µg/mL dilution of anticortisol antibody

Aequorin-Based Cortisol Immunoassay for Analysis of Saliva

Bioconjugate Chem., Vol. 18, No. 6, 2007 1775 Table 1. Intra-Assay Precisiona saliva pool pool pool pool pool pool pool

Figure 3. Binder dilution curve for 1:2500 aequorin–cortisol conjugate. The greater the cortisol antibody concentration, the larger the bioluminescent signal from aequorin. Data are the mean of three replicates, which have been blank subtracted, (1 standard deviation.

1 2 3 4 5 6

observed value 1 (nmol/L)

observed value 2 (nmol/L)

observed value 3 (nmol/L)

mean observed value ( SD (nmol/L)

CV (%)

15.6 5.6 8.0 5.3 7.9 12.4

15.0 6.0 7.1 5.2 7.8 11.7

15.1 6.0 7.7 5.2 7.8 11.7

15.2 ( 0.3 5.9 ( 0.2 7.6 ( 0.5 5.3 ( 0.1 7.8 ( 0.1 12.0 ( 0.4

2.0 3.7 6.2 1.6 0.9 3.5

a The saliva of six subjects was analyzed on three subsequent days in order to determine the precision of the assay. Data are the means of three replicates, which have been blank subtracted, (1 standard deviation.

Table 2. Assay Accuracya added (nmol/L)

observed value (nmol/L)

expected value (nmol/L)

mean recovery (%)

bias (%)

5.0 50.0 500.0 5000.0 mean

18.3 ( 0.2 62.0 ( 0.5 539.4 ( 3.2 5230.5 ( 42.3

18.2 63.2 513.2 5013.2

100.8 ( 0.9 98.1 ( 0.9 105.1 ( 0.6 104.3 ( 0.8 102.1 ( 0.8

+ 0.8 - 1.9 + 5.1 + 4.3 + 2.1

a A saliva pool was spiked with serial dilutions of cortisol standards in order to determine the accuracy of the assay. Data are the means of three replicates, which have been blank subtracted, (1 standard deviation.

Figure 4. Dose–response curve for cortisol generated using 1:2500 aequorin–cortisol conjugate. Data are the means of three replicates, which have been blank subtracted, (1 standard deviation.

to be used for the rest of the assay development studies. This concentration was the lowest antibody concentration that caused a significant change in aequorin’s signal (Figure 3). Moreover, the results of the binder dilution study indicated that the 1:2500 aequorin–cortisol conjugate yielded the most significant change in signal following anticortisol binding. Therefore, only the 1:2500 aequorin–cortisol conjugate was employed for subsequent dose–response curves and salivary analysis. Dose–response curves were produced by simultaneously incubating 50 µL of serially diluted cortisol standard, 50 µL of the 1:2500 aequorin–cortisol conjugate, and 50 µL of the anticortisol antibody in a microtiter plate well for 30 min at room temperature, while shaking. Following incubation, light was collected for 5 s after injecting 100 µL of buffer C. The average of three replicates was then plotted (Figure 4). The assay exhibited a sigmoidal response to changing cortisol concentrations with increasing cortisol concentrations resulting in a lower bioluminescence signal. These results corresponded to the data from the binder–dilution curves in that increasing the concentration of anticortisol antibody available to bind the aequorin–cortisol conjugate resulted in an increase in bioluminescent signal. The detection limit for this assay was 1 × 10-10 M of cortisol. The linear dynamic range of the assays spanned over 5 orders of magnitude, from 1 × 10-5 to 1 × 10-9 M of cortisol. To demonstrate clinical feasibility, we next investigated the ability of this assay to function in untreated biological samples, specifically in human saliva samples. The saliva of six healthy subjects was analyzed, and the concentration of cortisol in the samples fell within, or very close to, the accepted clinical range of morning cortisol saliva levels of 10–25 nmol/L. In order to study the intra-assay precision, the saliva pools of these six subjects were analyzed three times, on subsequent days, and the intra-assay precision was found to be between 0.9% and 6.2% (Table 1). The accuracy of this assay was tested with a percent recovery study. In order to study the performance over a wide range of cortisol concentrations, 5, 50, 500, and 5000

nmol/L samples of cortisol were spiked into the original saliva pool. These spiked samples were then run with the assay, and the results produced an accuracy ranging from 98% to 105% (Table 2).

DISCUSSION In an effort to develop homogeneous assays capable of yielding low detection limits, we selected a bioluminescent photoprotein as the signal-generating label. Other labels commonly employed in these types of assays include enzymes, fluorophores, and radioisotopes. We selected the bioluminescent aequorin over other labeling options because it is nontoxic, very sensitive in biological fluids, does not require long incubation periods, and can be genetically and chemically altered to yield a homogeneous modulation of its bioluminescence signal. The homogeneous, bioluminescence-based assay for cortisol discussed herein represents a distinct advantage over currently available cortisol assays, since it is fast, easy, and has the ability to achieve the very low detection limits that are necessary for salivary analysis. After conjugating cortisol to aequorin via lysine residues, we determined the decay profile of the conjugates. The results from the activity and half-life studies were similar to the way aequorin responded in previous studies that employed aequorin chemically conjugated to an analyte through its lysine residues (3, 10). In these studies, as in ours, as the molar ratio of analyte to aequorin in the conjugate increased, the bioluminescence activity of the conjugate decreased and the decay half-life of the conjugate increased. We then undertook a calibration study in order to determine the minimum conjugate concentration needed for assay development. From the calibration curves, we selected the 1 × 10-11 M conjugate concentration for a binder dilution study (Figure 2). The binder dilution study was undertaken in order to determine what antibody concentration was optimal for the homogeneous response of the 1 × 10-11 M conjugates. From the binder dilution study, it was determined that a 3.3 µg/mL anticortisol antibody concentration was the minimum antibody concentration required to yield a change in the bioluminescent signal (Figure 3). It was also observed from the binder dilution study that the higher the molar ratio of cortisol

1776 Bioconjugate Chem., Vol. 18, No. 6, 2007

in the conjugate, the more pronounced the homogeneous response following anticortisol antibody binding. In other words, the 1:2000 aequorin–cortisol conjugate gave a better homogeneous response to anticortisol antibodies than the 1:1500 aequorin–cortisol conjugate, and the 1:2500 aequorin–cortisol conjugate exhibited the best homogeneous response, giving the most pronounced change in bioluminescence signal following anticortisol antibody binding. For this reason, the 1:2500 aequorin–cortisol conjugate was selected for the remaining studies, including the dose–response curve and the analysis of saliva samples. Next, a dose response curve was obtained by incubating the aequorin–cortisol conjugate, the anticortisol antibody, and serially diluted cortisol standards in microtiter plate wells for 30 min. Following incubation, the wells were washed, and the bioluminescence intensity was measured using an automated microtiter plate luminometer. Results of the dose–response curve indicated that decreasing concentrations of cortisol, and corresponding increasing concentrations of free antibody, resulted in an increase in the bioluminescent signal from the conjugate (Figure 4). One may initially assume that increasing the amount of antibody available to bind the conjugate would result in sterically hindering aequorin and, thereby, reduce its bioluminescence, as shown in the homogeneous aequorin-based assay for biotin (21). However, a similar homogeneous response to ours, wherein signal increased with increasing antibody concentration, was found by Lewis and colleagues when they studied the response of an aequorin–thyroxine conjugate to increasing concentrations of antithyroxine antibody (22). This type of response, seen both here and in Lewis’s thyroxine study, is more advantageous than the response found in the biotin assay, since the signal increases as the concentration of free cortisol decreases, thereby allowing a lower detection limit The detection limit of this assay was 1 × 10-10 M cortisol, which is over 1 order of magnitude lower than the detection limit obtained for a heterogeneous bioluminescence aequorin-based cortisol immunoassay, namely, 3.0 × 10-9 M (4). The linear range for this assay ranged from 1 × 10-5 M to 1 × 10-9 M, compared to the heterogeneous aequorin-based cortisol assay, whose linear range spanned only 3 orders of magnitude, from 1 × 10-6 to 1 × 10-8 M (22). The assay time was 30 min, which is slightly longer than the 18 min time afforded by the Roche Cobas cortisol assay. However, our assay is more sensitive than the Roche assay (0.1 nmol/L vs 0.5 nmol/L) and requires significantly less saliva volume (50 µL vs 500 µL), making our assay more adaptable to miniaturization (23). Moreover, the Roche assay utilizes electrochemiluminescence detection, which requires a specialized automated instrument, whereas our assay can be performed using standard luminometers and automated platforms such as Tecan, Hamilton, and so forth. The low detection limit of the dose–response curve indicated that saliva analysis was possible, due to the fact that salivary levels of free cortisol ranged between 10 and 25 nmol/L. Moreover, salivary cortisol measurements are primarily used to rule out Cushing’syndrome. Cortisol concentrations are very low, generally less than 10 nmol/L, during the standard latenight sampling times used for these Cushing’s syndrome tests. Therefore, the 0.1 nmol/L detection limit of this aequorin-based assay suggested that it may be suitable for clinical applications. Also, biological samples generally must be extracted prior to assaying, which is a time-consuming pretreatment. For this assay, the only pretreatment of saliva samples was a brief centrifugation to pellet particulate matter. No extraction or complex pretreatment of biological samples was needed, thereby streamlining the assay. The intra-assay precision with saliva samples was found to be between 0.9% and 6.2%, and the

Rowe et al.

accuracy of the assay was found to be between 98% and 105% using spiked saliva samples. This level of accuracy and precision corresponds to the levels previously reported for cortisol assays, and is a realistic performance for a salivary cortisol assay (4). In summary, we have developed a homogeneous assay using the photoprotein aequorin that can detect physiological levels of cortisol in human saliva. This assay has a dynamic range spanning 5 orders of magnitude and a sub-nanomolar detection limit. Since the level of free cortisol in human saliva ranges from 10 to 25 nmol/L, this assay is also well-suited for real samples, and the analysis of unextracted saliva samples gave acceptable precision and accuracy results. The fact that this homogeneous assay can analyze untreated saliva samples without multiple washing and incubation steps makes it superior to most of the currently commercially available assays for cortisol. Additionally, the low detection limit and long linear range of this assay are sufficient for both serum and saliva analysis. The feasibility of this homogeneous, bioluminescent assay in determining salivary cortisol levels has been clearly demonstrated. However, additional validation methods, such as extensive interference tests, interassay precision, and mass spectrometric confirmation, will be necessary before clinical implementation. However, the speed and simplicity of this homogeneous assay suggests that it may offer a simple alternative for testing salivary cortisol levels in the future, and represent an advance over current technologies by being applicable to both high-throughput screening and on-site monitoring technologies.

ACKNOWLEDGMENT This study was supported by the National Institutes of Health (CHE 467917) and the National Aeronautics and Space Administration. We thank the Vice President of Research at the University of Kentucky for a University Research Professorship to S.D. S.D. also acknowledges support from a Gill Eminent Professorship. L.R. acknowledges support from Predoctoral Fellowships from the National Institutes of Health and National Science Foundation-IGERT Program at the University of Kentucky.

LITERATURE CITED (1) Diamandis, W., and Christopoulous, T. (1996) Immunoassay (Diamandis, W., Christopoulus, T., Eds.), pp 488–490,Chapter 21, Academic Press, San Diego, CA. (2) Sittampalam, S., Kahl, S., and Janzen, W. (1997) High throughput screening:advances in assay technology. Curr. Opin. Chem. Biol. 1, 384–391. (3) Desai, U., Deo, S., Hyland, K., Poon, M., and Daunert, S. (2002) Determination of prostacyclin in plasma through a bioluminescent immunoassay for 6-keto-prostaglandin F1R: Implications of dosage in patients with primary hypertension. Anal. Chem. 74, 3892–3898. (4) Mirasoli, M., Deo, S., Lewis, J., Roda, A., and Daunert, S. (2002) Bioluminescence immunoassay for cortisol using recombinant aequorin as a label. Anal. Biochem. 306, 204–211. (5) Tortora, G., and Grabowski, S. Principles of Anatomy and Physiology (1996) Harper Collins College Publishers, Menlo Park, CA. (6) Vining, R. F., McGinley, R. A., and Symons, R. G. (1983) Hormones in saliva: Mode of entry and consequent implications for clinical interpretation. Clin. Chem. 29, 1752–1756. (7) Kahn, J. P., Rubinow, D. R., Davis, C. L., Kling, M., and Post, R. M. (1988) Salivary cortisol: A practical method for evaluation of adrenal function. Biol. Psychiatry 23, 335–349. (8) Daunert, S., and Deo, S. Photoproteins in Bioanalysis (2006) Wiley-VCH, Berlin.

Aequorin-Based Cortisol Immunoassay for Analysis of Saliva (9) Head, J., Inouye, S., Teranishi, K., and Shimomura, O. (2000) The crystal structure of the photoprotein aequorin at 2.3 Å resolution. Nature (London) 405, 372–376. (10) Witkowski, A., Ramanthan, S., and Daunert, S. (1994) Bioluminescence binding assays for biotin with attomole detection based on recombinant aequorin. Anal. Chem. 66, 1837–1840. (11) Feltus, A., Ramanthan, S., and Daunert, S. (1997) Interaction of immobilized avidin with and aequorin-biotin conjugate: An aequorin linked assay for biotin. Anal. Biochem. 254, 62–68. (12) Lewis, J. C., Cullen, L. C., and Daunert, S. (2000) Sitespecifically labeled photoprotein-thyroxine conjugates using aequorin mutants containing unique cysteine residues:pplications for binding assays (Part 2). Bioconjugate Chem. 11, 140–145. (13) Hart, R. C., Scheuer, B. L., Rivera, H. N., and Gray, J. (1993) in Bioluminescene and Chemiluminescence: Status Report (Szalay, A. A., Kricka, J. L., and Stanley, P., Eds), pp. 320– 323, Wiley, Chichester. (14) Casadie, J., Powell, M. J., and Kenten, J. H. (1990) Expression and secretion of aequorin as a chimeric antibody by means of a mammalian expression vector. Proc. Natl. Acad. Sci. U.S.A. 87, 2047–2051. (15) Zenno, S., and Inouye, S. (1990) Bioluminescent immunoassay using a fusion protein of protein A and the photoprotein aequroin. Biochem. Biophys. Res. Commun. 171, 169–174. (16) Ramanthan, S., Lewis, J. C., Kindy, M. S., and Daunert, S. (1998) Heterogeneous bioluminescence binding assay for an octapeptide using recombinant aequorin. Anal. Chim. Acta 368, 181–188.

Bioconjugate Chem., Vol. 18, No. 6, 2007 1777 (17) Witkowski, A., Daunert, S., Kindy, M. S., and Bachas, L. (1993) Enzyme-linked immunosorbent assay for an octapeptide based on a genetically engineered fusion protein. Anal. Chem. 65, 1147–1151. (18) Deo, S. K., and Daunert, S. (2001) An immunoassay for Leuenkephalin based on a C-terminal aequorin-peptide fusion. Anal. Chem. 73, 1903–1908. (19) Galvan, B., and Christopoulous, T. (1996) Bioluminescence hybridization assays using recombinant aequorin. Anal. Chem. 68, 3545–3550. (20) Lewis, J. C., Lopes-Moya, J. J., and Daunert, S. (2000) Bioluminescence and secondary structure properties of aequorin mutants produced for site-specific conjugation and immobilization. Bioconjugate Chem. 11, 65–70. (21) Grosevnor, A., Feltus, A., Conover, R., Daunert, S., and Anderson, K. (2000) Development of binding assays in microfabricated picoliter vials: an assay for biotin. Anal. Chem. 72, 2590–2594. (22) Lewis, J., and Daunert, S. (2001) Bioluminescence immunoassay for thyroxine employing genetically engineered recombinant aequorin containing unique cysteine residues. Anal. Chem. 73, 3227–3233. (23) Vogeser, M., Durner, J., Seliger, E., and Auernhammer, C. (2006) Measurement of late-night salivary cortisol with an automated immunoassay system. Clin. Chem. Lab. Med. 44, 1444–1445. BC070039U