Luminescentoxygenchannelingassay(LOCITM): sensitive,broadlyapplicablehomogeneous immunoassaymethod

Clinical Chemistry 1518-1526

42:9

k-

(1996)

Luminescent oxygen channeling assay (LOCITM): sensitive, broadly applicable homogeneous immunoassay method EDwIN

MARY

F.

ULLMAN,*

ERIcsoN,

SHARAT

SINGH,

HiIR

KIRAKOSSIAN,

A. WARTCHOW,

CHARLES

RAJENDRA

CARL

SINGH,

SKOLD,

MARCEL

RAJESH

NURITH

C.

ARTHUR

PATEL,

KURN,

Luminescent oxygen channeling assay (LOCITM) is a homogeneous immunoassay method capable of rapid, quantitative determination of a wide range of analytes-including high and very low concentrations of large and small molecules, free (unbound) drugs, DNA, and specific 1gM. Assays have been carried out in serum and in lysed blood. Reliable detection of 1.25 tUfL thyrotropin (TSH) and 5 ngfL hepatitis B surface antigen (HBsAg) corresponds to detection limits -3-- and -20-fold lower, respectively, than those of the best commercially available assays. An assay of chorionic gonadotropin is capable of quantification over a 106-fold range of concentrations without a biphasic response. Latex particle pairs are formed in the assay through specific binding interactions by sequentially combining the sample and two reagents. One particle contains a photosensitizer, the other a chemiluminescer. Irradiation causes photosensitized formation of singlet oxygen, which migrates to a bound particle and activates the chemiluminescer, thereby initiating a delayed luminescence emission. Assay times range from 1 to 25 mm. TERMS: latexagglutination,singletoxygen #{149} chemiluminescence . photosensitizers . biotin-streptavidin interaction #{149} digoxin #{149} cyclosporine#{149} thyrotropin #{149} hepatitis #{149} chorionic gonadotropin INDEXING

Development of a sensitive homogeneous immunoassay methodology has been the goal of numerous researchefforts [1, 2]. A method thatwould eliminatethe need to separatebound from free labeled analyte(and the associatedwash steps)would be

Research Department, Syva Business Unit, Behi-ing Diagnostics Inc., P.O. Box 49013, SanJose, CA 95161-9013. *Author for correspondence. Fax 408-239-2707; e-mail edwin.ullman#{174} bdi.hcc.co,n. Received May 3, 1996; accepted June 20, 1996.

1518

JACQUELINE

SWITCHENKO,

PIRI0, ALAN

JOHN

PEASE,

DAFFORN,

and DANIEL

B.

ISHKANIAN,

BENJAMIN

DARIUSH

R.

IRVIN,

DAVALIAN,

WAGNER

particularly useful if it were suitable for both haptens and proteins and sufficiently sensitive to permit quantification of all common analytes. Chemiluminescence-based homogeneous immunoassays provide the required sensitivity [3, 4], but the complex chemistry of chemical excitation and emission renders them highly susceptible to matrix effects. Latex agglutination techniques [5], in which detection is based on low-angle light scattering, also have the theoretical ability to provide exceptionally high sensitivity, but the measurements require rigorous exclusion of adventitious particles [6-8]. Enzyme channeling immunoassays are applicable to both small and large molecules but have limited sensitivity because of their susceptibility to matrix effects. In these assays, an immune reaction brings two enzymes into proximity on a surface; at that surface, one enzyme produces a product that serves as a chromogenic substrate for the second enzyme [9]. The product reaches higher concentrations near the surface than in the bulk solution, which leads to an increase in the rate of reaction with the proximate second enzyme. We report here on the development of an unusually robust and sensitive homogeneous chemiluminescent immunoassay that is based on a combination of the principles of latex agglutination and channeling. The luminescent oxygen channeling assay (LOCIm’M) principle, which has been described previously [10], uses two different ligand- or receptor-coated polystyrene particles at much lower concentrations than ordinarily used in latex agglutination.’ Typically, particle pairs are formed rather than large aggregates, and photochemically triggered chemiluminescence replaces detection by nephelometry or turbidity. One particle has dissolved in it a photosensitizer that

Nonstandard abbreviations: }lBsAg, hepatitisB surfaceantigen;biotin-NIIS ester, biotinyl-6-aminohexanoic acid N-hvdroxvsuccinimide ester; I-lAy, hepatitis A virus;TSH, thyroid-stimulating hormone (thvrotropin); and hCG, human chorionicgonadotropin.

Clinical Chemissiy

produces singlet oxygen when exposed to light. The second particle has dissolved in it an olefin that, upon reaction with singlet oxygen, produces chemiluminescent emission. The reaction with singlet oxygen with the subsequent emission is strongly dependent on the formation of particle pairs. Both types of particles have a hydrogel coating that protects the particles from nonspecific interactions with matrix components and provides a functionalized surface to which antibodies and analytes can be covalently attached.

Materials and Methods Reagents. The phthalocvanine, I (see Fig. I), was prepared by modifications of a published method [11]. Streptavidin was from Aaston (Wellesley, MA). Antibodies, antigens, and haptens were from Behring Diagnostics (San Jose, CA), except for monoclonal antibodies to hepatitis B surface antigen (HBsAg), which were from the British Technology Group (London, UK). All other compounds were from Pierce (Rockford, IL), Aldrich (Milwaukee, WI), or Sigma (St. Louis, MO), except as indicated. Calibrators were obtained from Behring Diagnostics. Buffers. The buffer used in LOCI hapten

assays

as

diluent and

the chemiluminescer and sensitizer was 0.1 molJL Tris containing detergents and proteins to minimize nonspecific binding interactions. media

for

1519

on-board detection unit. The detection unit was designed to expose the sample alternately to a light source and then to a photomultiplier tube. The light source, an electronically modulated 680-nm laser, was used with a Hamamatsu R4632 (Bridgewater, NJ) photomultiplier tube equipped with an electromechanical shutter and a 580-620-nm bandpass filter. Typically, each chemiluminescence determination involved 3 to 10 sequential periods of 1 s of irradiation and 1 s of photon counting. The signal was recorded as the total counts detected. ASSAYS

MATERIALS

storage

42, No. 9, 1996

as

particles designed

Biotinylated conjugates. Biotinylated antibodies were prepared by coupling with biotinyl-6-aminohexanoic acid N-hydroxysuccinimide ester (biotin-NHS ester), followed by dialysis. Biotinylated haptens were prepared by reaction of the biotin-NHS ester with an amine-substituted hapten or by preparing an active ester of a carboxy-substituted hapten and coupling to an aminoalkylamide of biotir,. Latex particles. Polystyrene latex particles (205 nm diameter) with about one carboxyl group per 0.5 nm2 of surface were from Seradyn (Indianapolis, IN). Typical procedures used for incorporating dyes into the particles and coupling proteins and DNA were described previously [10]. Streptavidin was coated onto the sensitizer particles at surface densities, determined by the binding of [3Hjbiotin, typically in the range of 2 000-4000 streptavidin molecules per particle. Antibodies were bound to chemiluminescer particles, and their surface densities were determined by measuring the binding of labeled antigen or the depletion of antigen from solution after separation of the particles by centrifugation. Typical antibody/particle ratios were 3-15 for hapten assays and 300-700 for protein assays. Biotinylated haptens were bound to the streptavidin-coated sensitizer particles by simply mixing the two components; no separation or washing was necessary. INSTRUMENTATION

LOCI assays were performed by a modified robotic sample processor (Tecan US, Research Triangle Park, NC) equipped with one probe for fluid-handling steps and a second probe for transferring reaction cuvettes from an incubation rack to an

Hapten assays. Haptens sample (2-50 jiL) at 37 chemiluminescer particles a suspension containing cles (-600 jiL). After

were determined by incubating the #{176}C with 0.1-10 jig of antibody-coated in --400 jiL of buffer and then adding excess hapten-labeled sensitizer partian additional incubation period, the chemiluminescence signal was measured. Each incubation period was 0.5-10 mm. Total assay times ranged from 1 to 14 mm. Antigen

antigens were determined by first incucontaining the sample (2-66% of the total reaction volume) and a reagent containing 0.1-10 jig of antibody-coated chemiluminescer particles and a biotinylated bating

assays. Protein

at 37 #{176}C solutions

noncompetitive second antibody. These reagents were simultaneously combined with the sample but were frequently stored separately. Later experiments showed that they could be stored and used as a single reagent. A suspension containing 1-20 jig of streptavidin-coated sensitizer particles was then added, and the chemiluminescence signal was measured after an additional incubation period. Incubation times ranged from 2 to 12 mm. The amount of biotinylated antibody used in the assay was always less than the binding capacity of the streptavidin-coated particles. Specific anti-HA V 1gM assay. The procedure was similar to that used for antigen assays except that the sample was prediluted 1:400. Hepatitis A virus (I-IAV) antigen-coated chemiluminescer particles were prepared by simply combining anti-FIAV antibody-coated chemiluminescer particles with I-IAV antigen (Meddiagnost, Tubingen, Germany). A mixture of 50 jiL of the diluted sample and 250 jiL of buffer containing 2.5 jig (3.5 pmol/L) of these particles and 0.2 jig (4.4 nmol/L) of biotinylated goat anti-human 1gM was incubated for 7 mm. The mixture was then diluted with 600 jiL of a suspension of 20 jig of streptavidin-labeled sensitizer particles (final concentration 9.3 pmollL) and incubated another 7 mm before the chemiluminescence was read. Nucleic acid assay. A 60-base oligodeoxynucleotide having a (T)24 3’-end and a (TACT)6 sequence at the 5’-end was used as a model analyte. LOCI was carried out by combining this oligonucleotide, in 100 jiL of buffer, with 2.5 jig (10 pmol/L) of sensitizer particles labeled with 5600 (A)24 oligodeoxynucleotides and 5.0 jig (21 pmol/L) of chemiluminescer particles labeled with 3700 oligodeoxynucleotides having the sequence (AGTA)7. The mixture was heated to 95 #{176}C to assure full dissociation of

1520

Ullman

et al.: Luminescent

c’

oxygen

channeling

assay (LOCITM)

c ‘b=O.00002

s

ENERGY TRANSFER

ct

emr

>0.27

Fig. 1. Chemiluminescence

480, 510 nm

any complexes and then annealed at 37 #{176}C for 30 mm before chemiluminescence signal was read.

the

Results PREPARATION

OF ASSAY

COMPONENTS

LOCI assays require the use of two reagents. Phthalocyanine, (Fig. 1), was chosen for use in the sensitizer particles because

I of

its strong absorption at 680 nm (e = 180 000 L mol’ cm) and its efficient photosensitization of oxygen. The use of this wavelength minimized interference from normal serum components and allowed us to take advantage of a commercially available 680-nm solid-state laser. As the concentration of I in 200-nm latex particles was increased, we found that the particles produced increasingly higher yields of singlet oxygen upon irradiation, but the yields plateaued at -100 mmollL. If un-

modified phthalocyanine was used in place of I, the maximum yieldwas lower,probably because of more efficient stacking of the dye when dissolved in the polystyrene. The choice of a chemiluminescent compound was dictated by the need for a compound that was highly soluble in polystyrene and could react rapidly with singlet oxygen. High concentrations and high reactivity of the chemiluminescer are required to efficiently capture the singlet oxygen entering a particle before it can decay back to its ground state or diffuse out of the particle. In addition, the chemilummnescer-singlet oxygen adduct must have a high chemiluminescence quantum yield and, to minimize the time required to reach maximum light emission, the rate of decay of the adduct must be very fast. The thioxene II (X = S) met most of these requirements [10]: It is soluble in latex up to 60 mmollL and reacts very rapidly with singlet oxygen (2 >< i0 L mol’ s’). Previously [10], we found that particles containing 20 mmol/L of the less-reactive dioxene H (X = 0) captured --65% of the singlet oxygen, so thioxene H (X = S) is probably even more efficient. Moreover, the dioxetane ifi, produced by reaction of II (X = S) with singlet oxygen, decayed rapidly (Ic = 0.33 s_I in toluene, -1.3 s’ multiphasic decay in latex), which permitted light emission to peak after an exposure time of only -1 s . Although the ensuing light emission at 390 nm had a very low quantum yield (2 X i0 in toluene, based on a ‘4C standard), this was due to a low fluorescence quantum yield of the diester product IV and not to a low chemi-excitation efficiency. Thus, dissolving the fluorescence

reaction.

See text for details.

energy acceptor 9, 10-bisphenylethynylanthracene along with II (X = S) in the latex particles greatly increased the quantum yield (estimated as >0.27 from solution-phase studies); further increases (estimated as >0.46) were obtained by adding various other energy acceptors. After preparation of the sensitizer and chemiluminescer particles, each particle type was coated with a hydrogel that contained active groups for binding to receptors and ligands. The fully coated particles could be heated at 95 #{176}C for several hours with minimal loss of the dyes or of the bound compounds. Nevertheless, the particular compounds that were bound to their surfaces retained whatever thermal sensitivity the compounds had had before binding, so the particles were usually stored at 4 #{176}C in the presence of antimicrobial reagents. Under these conditions, no settling was detectable over a 2-month period, and calibration curves did not drift. The near-ideal behavior of the particles is attributable to their highly hydrophilic surface, which looked quite smooth when viewed by atomic force microscopy of a water-immersed sample (Fig. 2). Interestingly, if the particles were allowed to dry, the surface coating appeared to shrivel (Fig. 2). PARTICLE

BINDING

CHARACTERISTICS

The LOCI response was found to depend on the duration of incubation, the binding affinities, and the number of ligands and receptors associated with each particle. A model assay was

‘

256rvn

256rwn

Fig. 2. Atomic force microscopy of hydrogel-coated particles scanned under water (left) or in air (right).

200-nm latex

Clinical Chemistry

42, No. 9, 1996

1521

constructed

in which an average of 3000 streptavidins were bound to each sensitizer particle and -2000 biotins were bound to each chemiluminescer particle. Long incubations of increasing numbers of either of these particles with a fixed concentration of the other particle provided a linear increase in signal of over five orders of magnitude. Deviation from linearity occurred only when the concentration of the added particles approached the concentration of the fixed particles (Fig. 3). The responses were practically the same regardless of which particle was in excess. The lower limit of detection (3 SD above the mean for the negative calibrator, n = 10) was only 1600 added particles per milliliter (2.7 amol/L). Mathematical modeling of the number of one type of particle that can bind directly to a second like-sized particle suggested that rosettes averaging -8 particles are formed about each added particle. Thus, under the conditions of the experiment, the detection limit was -12 800 particle pairs (8 x 1600). Independent measurement of the binding rate (see below) showed that the binding reaction had proceeded nearly to completion. Investigation of the rate of association of the two types of LOCI particles revealed near-diffusion-controlled kinetics, if sufficient binding groups were present on each particle. Fig. 4 illustrates an experiment in which the streptavidin-labeled sensitizer particles were incubated at 37 #{176}C with a small amount of biotin-labeled chemiluminescer particles. Using the previously determined endpoint signal [10] and the initial rate of increase in signal intensity we calculated the second-order rate constant for binding to be 4.4 X 10” L mol ‘ s. However, when the number of receptors was sufficiently decreased, the binding rate was compromised. This is illustrated in Fig. 5, which shows the rate of binding of sensitizer particles labeled with 300 digoxmns to chemiluminescer particles labeled with various amounts of anti-digoxmn monoclonal antibodies. Binding rate constants as low as 4 x l0 L mol s were found when only 3 antibodies

I’,

C cn

0

0 -J

400

600

800

1000

t

time, s Fig. 4. Kinetics of binding at 37 #{176}C of the streptavidiri-labeled sensitizer particles (2.2 pmol/L) with the biotin-labeled chemiluminescer particles described in Fig. 3 (3.33 fmol/L).

per particle were present, although extrapolation to infinite antibodies per particle (not shown) provided a rate constant similar to that observed with streptavidin-biotin binding (2.0 X i0 L mol s_I) LOCI

ASSAYS

Digoxin. Clinically significant serum concentrations of digoxmn cover the range 0.5-5 jig/L. An assay for digoxin therefore was a good model for testing the sensitivity of LOCI for hapten applications. Mouse monoclonal anti-digoxin antibodies were bound to chemiluminescer particles, and a biotin- digoxin conjugate was bound to streptavidin-coated sensitizer particles (-100 molecules per particle). A frequently used 5-mm protocol and the corresponding calibration curve are given in Fig. 6. As expected for any competitive immunoassay, the sensitivity increased and the signal modulation decreased with increasing

io8 8x1 w

V

1o6

Ca

C

6x10 8

C

2

8 0 a)q)

0

4x10

10 C

0 0

102

2xlo:

101

102

io i05 106107 108 Number of particles addedlmL

io9 1010

Fig. 3. Maximum signals produced after long incubation of sensitizer particles (labeled with -3000 streptavidins) and chemiluminescer particles (labeled with -2000 biotins). A fixed concentration (2.1 pmol/Ll of either sensitizer (+) or chemiluminescer particles (LI) was incubated with various amounts of the complementary particles for 3 days at 4 C and then at 37 #{176}C for 6 h, after which the chemiluminescencesignals were read.

0

10 20 30 40 Anti-digoxin lgG molecules/particle

Fig. 5. Particle binding rates at 37 #{176}C in equimolar

50

mixtures (2.8 pmol/L) of sensitizer particles labeled with -300 digoxins per particle and chemiluminescer particles labeled with various numbers of monoclonal anti-digoxin lgG antibody. The absolute second-orderrate constant was determinedfor chemiluminescer particles with 24 antibodies per particle and used to adjust the initial rates measuredover 1.6 mm for the other antibody/particle ratios.

Ullman

1522

et al.: Luminescent

oxygen

concentration of antibody. The effect of increasing the number of particles while keeping the total antibody concentration constant (i.e., decreasing the number of IgG molecules per particle) was less predictable. The data suggest that 6-30 antibodies per particle was about optimal, at least for this analyte. The calibration curve in Fig. 6 was produced by using a particle loading of -10 antibodies and had -80% modulation over the range of 0-5 .tg/L digoxin. Excess digoxin (1 mg/L) produced >97% inhibition of signal. The intraassay CVs of the assay at 0.5, 1.5, and 3.0 g/L digoxin were 5.1%, 2.2%, and 1.4% (n = 15). Sample-to-sample signal CVs of