Establishment of the optically pumped chemiluminescence technique

Chem. , 1993, 65 (4), pp 403–408. DOI: 10.1021/ac00052a016. Publication Date: February 1993. ACS Legacy Archive. Cite this:Anal. Chem. 65, 4, 403-40...
2 downloads 0 Views 1MB Size
Anal. Chem. 1993, 65,403-400

403

Establishment of the Optically Pumped Chemiluminescence Technique for Diagnostics M. Motsenbocker,’*+T. Sugawara, M. Shintani, H. Masuya, Y. Ichimori, and K. Kondo Takeda Chemical Industries, Ltd., Pharmaceutical Research Division, 17-85 Juso-honmachi 2-chome, Yodogawa-ku, Osaka 532, Japan

A new chemliumlnescence procedure was established that dhnlnatesthe n e dto use a separate oxidant such as hydrogen peroxidefor chemliumlnescencedetection reactions. I n the procedurecalled optically pumpedchemhnin-e (OPC), red light energy Is used by a photosendtlve dye to oxidize luminoland generate blue light. By pulsingthe excitationlight and measuring chemllumlnescenceduring the otf the, light scatterhg backgroundwas ehlnated and 8ensWty enhanced 10-fold. The OPC reactionmechanim was determinedto be type I (direct reaction of excited dye with substrate), and the dye catalyst is recycled in the reaction. Of the popular photosensitive dyes tested, methylene blue showed the best detection limit in the OPC system. A methylene blue dye derivative was synthesized and covalentlyattachedto proteln. Antlbody proteinswere labeledwith upto 16 dyes per antibody. An a-fetoprotein immunoassay was developed having a detection limit of 17 pg. This array had senrltlvlty of 1.5 ng of alphafetoproteWmLof plasma and showed goodcorrelatbn ( R = 0.98). The OPC catalysts are more stable and krs temperature crenrltlvethan enzymecatalysts,and the substrate solutlon needed Is very stable. Thw, OPC may be valuable in applications where convenience and reagent stability are Important.

INTRODUCTION Because of the analytical sensitivity obtainable with photon counting,chemiluminescence detection systemshave become popular in diagnostics. The first reactions used as sensitive detection systems were microperoxidase-catalyzed luminol chemiluminescence,horseradish peroxidase catalyzed lumino1 chemiluminescence, and acridinium ester chemiluminescence.1-3 These systems require the preparation, storage, and use of hydrogen peroxide as the oxidant. The requirement for a strong oxidant cosubstrate such as hydrogen peroxide in an assay system is frequently inconvenient and causes a high light emission background. Accordingly,one trend in the developmentof chemiluminescence is away from the use of hydrogen peroxide. An example is the stable dioxetane system pioneered by Bronstein and by Schaap which eliminates the need for a high-energy cosubstrate by using a substrate that contains a very stable dioxetane.4~5 Another alternative is to generatehydrogen peroxide during the reaction itself. This technique is employed by Arakawa et al. who used the enzyme glucose oxidase with glucose and ~~~

~

Present address: Law School, Condon Hall,University of Washington, Seattle, WA 98195. (1)Schroeder,H. R.; Boguslawski,R. C.; Carrico, R. J.; Buckler, R. T. Methods Enzymol. 1978,57, 424-459. (2) Whitehead, T. P.; Thorpe, G. H. G.; Carter, T. J. N.; Groucutt, C.; Kricka, L. J. Nature 1983,305, 158-159. (3) Weeks, I.; Sturgeas, M.; Brown, R. C.; Woodhead, J. S.Methods Enzymol. 1986,133, 366-387. (4) Bronstein, I.; Edwards, B.; Voyta,J. C. J . Biolumin. Chemilurnin. 1989,4,99-111. (5) Schaap, A. P.; Handley, R. S.; Giri, B. P. Tetrahedron Lett. 1987, 28,935-938. +

0003-2700/93/0365-0403$04.00/0

oxygen to make hydrogen peroxide. The hydrogen peroxide then reacts with an oxalate and a fluor to generate light in the detection reaction steps6 Additional means exist to generate chemiluminescence without adding a special oxidant. For example, photoactivated dyes have been shown to react with fluorogenic and chemilumigenic substrates and can generate light this way.’ Some of these dyes produce much singlet oxygen and are being wed in cancer therapy.8 In the type I photosensitized oxidationreactions of these dyes, the exciteddye triplet reacts directly with substrate by electron or hydrogen transfer (Figure lh9 The type I1 reactions of photosensitive dyes require molecular oxygen and form singlet oxygen which then reacts with and oxidizes a substrate (Figure 1). The predominance of either reaction type is influenced by the concentrations of dye, oxygen, and reactive substrate.9 The use of photoactive dye to generate chemiluminescence would allow simpler detection reaction solutions in immunoassays and also eliminate the enzyme catalyst with its attendant limitations of stability and temperature sensitivity. Such a system would have some of the advantages of both fluorescence and chemiluminescence. One advantage of fluorescence is that excitation energy is supplied by an outside source and not from a strong oxidizer. Photoactive dye sensitized chemiluminescence is similar to fluorescence because light from outside the system is used as an energy source. An advantage of chemiluminescenceis the absence of fluorescence or light-scattering background which reduces sensitivity. OPC is similar to chemiluminescence in that background fluorescence and light scattering are not limiting. This results from using long wavelength light for excitation and by measuring shorter wavelength light emission. Thus, stokes shift (short wavelength to longer wavelength light) light emission is eliminated. Also, since the excitation wavelength is far from the chemiluminescence emission wavelength, light-scattering background is greatly reduced. The purpose of the present study was to explore the use of photoactive dyes as chemiluminescence catalysts for diagnosticassays. In thia technique called “opticallypumped chemiluminescence”(OPC), a high pH solution of luminol is irradiated with 670-nmlight, and the presence of photoactive dye catalyst is deteded by 425nm light emission from luminol chemiluminescence. An instrument was developed for this purpose. The instrument was modified to remove lightscattering interference by pulsing the light source and measuring the chemiluminescencesignal during the off time. In this way the ability to remove light scattering interference without excessive optical fiitering was explored. The chemical reaction was optimized for best sensitivity, and a methylene blue photoactive dye derivative was synthesized and covalently coupled to protein. Finally, an immunoassay was developed to measure a-fetoprotein from blood samples. (6) Arakawa, H.; Maeda, M.; Tsuji, A. Clin. Chem. 1985,31,430-434. (7) Matheson, I. B. C.; Lee, J. Photochem. Photobiol. 1976,605-607. (8)Andersson-Engels,S.;Johanason, J.; Svanberg,S. Anal. Chem. 1989, 61,1367A-1373A. (9) Foote, C. S. Science 1968,162,963-970. 0 1993 American Chemlcal Society

404

ANALYTICAL CHEMISTRY, VOL. 65. NO. 4, FEBRUARY 15, 1993

Photosystem I I

Reagents. MethylenebluewasobtainedfromSigmaChemical Co. (St. Louis, MO!. Luminol was obtained from Aldrich Chemical Co. (Milwaukee,W1) and purified by recrystallization from I M NaOH. The NaOH was 99.997 pure semiconductor grade from Aldrich. Plastic test tubes 12-mm outside diameter by 75-mm height were used. Block ace was from Snow Brand Milk Products rSapporo, Japan). Latex (3.2-pm diameter, was from Sigma Chemical Co. (St. Louis, MOL Solid-phase and conjugate antibi,dies were from Wako Pure Chemicals (Types A-295 and A-4, respectively),and AFP (a-fetoprotein! was from Scripps Laboratories Inc. (Catalog No. A0724, San IXego, CA!. Methylenebluesuccinimidodyederivative(typeIV1 isdescribed elsewhere." All water used was distilled and deionized (Mi1li.Q. \Vaters Instruments. Tokyo,Japan). Phosphate buffered saline (PBS) consisted of 0.05 M sodium phosphate and O.R% sodium chloride pH 7.0. Porphyrindves were from Porphyrin Producrrt. Logan, LIT. Other dyes, horseradish peroxidase (HHP). and chemicals were ordered from Wako Pure Chemical Ltd. (Osaka, Japan) or from Sigma Chemical ('0. and used without further purification Instrumentation. Fi~re2shnwstheschematicconstruction of the OPC instrument. Red (670 nm) light from diode laser A (TOLD9211. Toshiba, is pulsed on and off at a 500.H~square wave r I mson 1 muoff) frequency,entersonesmallarmoffiber

optic B (FLP-ZPZC, Welch Allen Inc., Skaneateles Falls, NY), and exits at the large arm to irradiate a luminol solution in test tube C in sample holder D. Chemiluminescencelight (425 nm) and scattered light (during the laser on period) from the sample enter the large arm of the fiber optic, pass through optic filter E (425-nm handpass) and into photon counting head F (H346004, Hammamatsu Ltd.). The photon pulses are converted to TTL signal levels and then counted two ways. The raw (unprocessed) pulses including those from both scattered light and chemiluminescencelight are counted by universal counter G (total counts). The raw pulses are also processed by circuit H which ignores pulses that occur when the red light is on. The (redlightofftime)pulsesfromcircuitHarecounted byuniversal counter I. Universal counter I therefore only counts pulses that occur duringthe laser off time and does not includelight scattering or fluorescencehackground counts. Signal generator J makes a square wave signal that controls laser source A and timing for circuit H. Computer K is interfaced to both frequency counters hyaGPIB interface. Thecomputercontrolspower tolight source A, the photomultiplier voltage (loo0V), and the beginning and ending count times for the universal counters. Circuit H also includes manual control switches and the power supply to the light source. Detection limits for experiments performed using this machine were defined as the concentration of detectable substmcethatgaveasignalequaltotwice theaverage background (0 concentration) signal. Comparisonof Dyes. For comparingdifferent dyesaphoton counting fluorometer (Hitachi 650-40) was used. The machine was adjusted for 2-s measurements, high-voltagegain, and 425nm emission wavelength. The excitation wavelengthwas varied between 500 and 700 nm as needed to match the optimum excitation wavelength of each dye. Additionally, glass filters were placed in both the excitation and emission beams to filter out unwanted light. Each of the dyes studied was diluted into 2 mL of 1mM luminol, pH 12.3 solution, and the 425-nm emitted signals were compared to those from methylene blue. Due to a high lighhcatteringbackgroundin thismachine, detection limits weredefined as theconcentrationofdyewhich gave alightoutput equal to the background signal plus three times the standard deviation of the background signal. Experiments on the OPC Method. For the pulsed light technique experiment, the laser was pulsed on and off with a 500-H~ square wave from signal generator J. Counts recorded on universal counter I are referred to as "signal" counts because they don't includelighhcattering and fluorescencebackgrounds. Counts recorded on universal counter G included high lighb scattering counts in addition to the chemiluminescencesignal. ForinvestigationofOPCreactiontimefromOto2s,methylene blue was diluted to 10 ng in a 0.1-mL total volume of 1 mM luminol and 0.05 M NaOH, and a laser pulse frequency of 2000 Hzwasused. Thedissolvedoxygeninsomesamples wasdepleted by exposing the sample to vacuum and then filling the test tube with nitrogen gas before OPC measurement. For inhibitor studies, ethanolic solutionsof 15mM 1,3-diphenylisobenzofuran (singlet oxygen quencher) or 15 mM 3-tert-butyl-4-hydroxy-5methylphenyl sulfide (free radical quencher) were diluted loo0fold into 1-mLsolutions of the above." For each measurement, 0.1 mLof afreshlypreparedsolution of (10 nglO.1mL) methylene blue was pipetted into the OPC machine. The laser was turned on, and 80-ms light measurements (totalof 1.7 s) were made for each sample For investigation of longer time period reactions light was measured from 60 to 120 s after turning on the laser circuit, and a laser pulse frequency of 500 Ha was used For the pH, temperature, and luminol concentration experiments, 1-mLsolutions of 1mM luminol and 0.05 M NaOH were used except as mentioned below. Light integrations were made for60sfollowinginsertionofeachsampleintotheOPCmachine. Solutions of varying pH were prepared by diluting sodium hydroxidein 1mMluminolandmeasuringtheresultantpH with apH meter. Different temperatures wereobtained by incubating the samples in a water bath for at least 5 min before adding the catalyst. For the temperature studies 10 pL of 100 nglmL

(101 Masuya, H.; Shmaru. H.: Miyawaki. T.; Motsenbwker. M. JP 96930 91.

(11)Weishaupt, K. R.; Gomer, C.J.: Dougherty, T.J. Can. Res. 1976, 23262329.

photon

slate)

blue photon

Figure 1. Type I and type I1 photosensnlzed oxidallon reactions.

T'

HOLDER

L (

j

MULTIPLIER

SIGNAL GENERATOR

LASER

FILTER

CIRCUIT

USER

OFF

COUNTING

COMPUTER

Fbura 2. OPC instrument wlth pulsegated detection. The laser 1s pulsed (square wave) at a frequency of 500 Hz.

EXPERIMENTAL SECTION

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

methylene blue were added after temperature equilibration and a 2-s light measurement was made after 30 8. For the methylene blue detection limit study 0.1-mL solutions of methylene blue in 0.1 M NaOH and 1 mM luminol were placed into plastic test tubes, and light measured from 60 to 120 s after exposure to red light in the OPC machine. In order to determine the number of luminols oxidized per dye molecule, the OPC machine was calibrated to measure luminol by oxidizing 100% of trace quantities of luminol and generating a standard curve. This experiment was performed by mixing 50 pL of 2 mM hydrogen peroxide with 50 pL of 0.2 pM microperoxidase (MP-11, Sigma) and a trace amount of luminol in 1 mM NaOH in the OPC reaction tube during light measurement.12 mol of methylene blue was added to a reaction Then, 2.2 X mixture of 0.1 mL of 0.1 M NaOH and 1 mM luminol, and the number of moles of luminol used up over a 2-h time period were measured. The number of luminolmoleculesoxidized was divided by the number of methylene blue molecules used to determine how many times on average each methylene blue molecule was used. Synthesis of photoactive dye conjugates. Methylene blue dye derivative (Figure 3) was coupled to antibody by mixing 2.0 mg/ mL of dye derivative in 0.1 M sodium phosphate buffer at a molar ratio of 50:l with 0.5 mg of IgG at a final volume of 1mL. This was left at 6 "C overnight, and the unreacted dye was removed by Sephadex G-50gel filtration. Absorbanceof proteinbound dye was estimated by completelyreacting a small amount of dye derivative with an excess of polylysine and measuring the change in absorbance (approximately 2-fold) at 663 nm. BSA was used as a carrier protein in an effort to link more dye moleculesto an antibody binding site. In this experiment 0.5 mg of BSA was incubated with 0.5 mg of dye in a final volume of 0.1 M sodium phosphate buffer pH 7.0 overnight, and the unreacted dye was removed by Sephadex G-50 gel filtration. BSA was also used as a carrier protein after denaturation with Tween-20 detergent (partial unwinding) and after complete denaturation with sodium dodecyl sulfate and reducing agent (completeunwinding). In the former experiment 1.5 mg of BSA was dissolved in 4 mL of 0.5% Tween-20 and 25 mM sodium phosphate pH 7.0. This was heated to 90 "C for 10 min and cooled to room temperature. Then 2.0 mg of methylene blue dye derivative was added, and the mixture was stored at 6 "C overnight. Derivatized BSA was separated from unreacted dye and from detergent by Sephadex G-50gel filtration. In the latter experiment, 50 mg of BSA were dissolved in 25 mL of 25 mM sodium phosphate buffer that contained 0.5 % sodium dodecylsulfate and 0.25 % 8-mercaptoethanol. This was heated to 90 "C for 10min. The mixture was cooled and protein separated from reducing agent by Sephadex G-50 gel filtration. A half milliliter of 1.0mg of denatured BSA and 0.5 mL of 1.0mg of dye derivative in 0.1 M sodium phosphate buffer pH 7.0 were mixed and incubated overnight at 6 "C. The protein was separated from unreacted dye by Sephadex G-50 gel filtration. The BSA-dye and denatured BSA-dye conjugates were coupled to antibody fragment (Fab) using a dimaleimide procedure to couple to sulfhydryl residues of each. One milligram of Fab prepared from anti-AFP antibody was reduced with 1.0 mg of dithiothreitol by incubation in 1.0 mL of 2.5 mM EDTA and 0.1 M sodium phosphate buffer pH 6.0 at 37 "C for 90 min. The excess dithiothreitol was removed by gel filtration in the same buffer. The prepared protein was immediately reacted with 2 mg of N,"-bis(3-maleimidopropionyl)-2-hydroxy-1,2propanediamine at 37 "C for 30 min. The excess dimaleimide was removed by gel filtration leaving each Fab with one free maleimide conjugated to its sulfhydryl residue. The protein recovered from gel filtration was mixed with 1 mg of BSA-dye conjugate in about 5 mL and incubated overnight at 6 "C. The prepared Fab-BSA-dye derivative was separated from Fab and BSA by gel filtration. a-Fetoprotein Immunoassay. This was a single incubation sandwich method in which one anti-AFP antibody was immobilized to 3.2-pm latex solid-phaseparticles. The second antibody (complexed with BSA-dye) and AFP were incubated together. All wash/separationsteps to remove free Fab-BSA-dye conjugate (12) Schroeder,H. R.; Yeager, F. M. Anal. Chern. 1978,50,1114-1118.

405

Table I. Activities of Photosensitive Dyes Compared relative detection limit dye methylene blue 11 Pg nile blue lo00 Pg thionine lo00 Pg Cu chlorophyllin no activity eosin Y 90 Pg hemin 600 pg chlorin E6 760 Pg rose bengal 50 Pg m-tetrakis(4-sulfonatono activity pheny1)porphine

Flgure 9. Structure of methylene blue dye derlvatlve.

utilized brief (304 low-speed centrifugation in a desktop centrifuge. The immobilizedantibody solid phase was prepared by washing 0.5 mL of 10% latex twice with 5 mL of water. The latex was then washed with 5 mL of 0.1 M carbonate buffer pH 9.5 and resuspended in 5 mL of 0.1 M carbonate buffer pH 9.5. After incubation at 56 OC for 20 min, 1.0 mg of AFP antibody was added and the suspension incubated another 40 min at 56 "C and then for 2 h a t room temperature. This was washed with 5 mL of 0.1 M phosphate buffer pH 7.0 that contained 0.2% Tween-20and then with 25 5% Block ace in PBS and finally stored in PBS that contained 25% block ace, 0.5% BSA, and 0.02% merthiolate. In the ELISA procedure 50 pL of 2 % latex were incubated with 5.0 p L of dye conjugate prepared above in a total volume of 250 pL in PBS that contained 10% Block ace. Thirty microliters of plasma were added and the incubation continued for 30 min at 37 OC. Each sample was then washed twice with 0.5 mL of PBS that contained 0.1 % Tween-20 and then with PBS. Then 0.4 mL of a 1.0 mM luminol and 0.1 M NaOH solution was added and the suspension transferred to a 12-mm-diameterplastic test tube for measurement by the OPC procedure. Detection limit was defined as the amount of AFP which gave an OPC measurement equal to twice the mean background signal. Plasma a-Fetoprotein Correlation Studies. These used the sandwich method described above. Plasma samples were prepared by collection of adult male blood in syringes that contained EDTA followed by centrifugationat 6"C. After adding AFP to various concentrations, each sample was stored frozen at -40 "C before use. In the percent recovery study AFP was added to 100 ng/mL to three blood plasmas, and the amount of AFP present was assayed. For interassay variability studies the assay was performed on 5 successive days (replicates of two for each day) and for intraassay studies the assay was performed five times. Detection limit was defined as the concentration of AFP which gave a signal in the assay that corresponded to the mean background signal plus two standard deviations of the zero measurement.

RESULTS AND DISCUSSION Of the photoactive dyes studied, methylene blue had the most activity in the OPC system (Table I). Methylene blue has an absorbance maximum (663 nm) which closely matches the emission of a popular low cost laser diode (670 nm). Thus, a 670-nm laser was chosen for the OPC instrument, and methylene blue or methylene blue dye derivative was used for all succeeding experiments. The methylene blue succinimide ester shown in Figure 3 was used for coupling to protein. This dye derivative was easier to work with than was methylene blue because it did not associate as strongly to the

406

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993 2

4

I n

0 7

X

B

n

0

I

0

0

I

I

I

50

100

150

200

Methylene Blue (pg/ml) Fbur 4. Photons counted during laser off time. 0% laser off countlng time results.

W 0 7

X

E

n

0

Methylene Blue (pglml) Flguro 5. Photons counted at all times. OPC results with continuous

photon countlng.

surfacesof containers used, presumably because of its negative charge. In order to use a pulse technique to remove the lightscattering signal from measurements the OPC light emission must persist after the excitation light is turned off. Therefore the OPC reaction half-life must be much larger than the laser off pulse time. Because the decay half-life of OPC light emission following a flash of excitation light from the laser was found to be more than 10 ms, a laser pulse frequency of 500 Hz (1ms laser on and 1ms laser off periods) was chosen. In this way the laser light pumps energy into the reaction solution for 1ms during the laser off time and reaction kinetics (having more than 10 ms half-life) allows chemiluminescence light to remain steady. The data in Figures 4 and 5 show light measured during the laser off time and total light measured (photons counted a t all times) for a methylene blue standard curve determination. By counting photons only during the laser off time, the light-scattering background was eliminated and the detection limit lowered from 220 to 18pg in this experiment. Using the pulse technique to remove the light scattering signal allowed us to employ a very simple optic filter, and measurements could be made even from highly reflective samples such as wells of a white microtiter plate. Some optic filtering was required because otherwise high levels of reflected light entering the photomultiplier tube during the laser on time

10

40

I 3

Time (seconds) Flgurr 6. OPC reaction time kinetlcs. The 670-nm light was turned on at 12 s and turned off at 50 s.

overloaded it and raised the dark count background. An additional adjustment was made by placing a 100-ps delay between turning the laser off and starting the laser off counting interval and a 50-ps delay between ending the laser off counting interval and turning the laser back on. These delay times were necessary to remove the effects of hysteresis in the PMT and in the circuitry that caused a background. In the final OPC instrument configuration there was no background from light scattering. The background signal produced in the absence of luminol originated only from the PMT dark counts and was 300 cpm. The background produced in the presence of luminol (from spontaneous luminol chemiluminescence) was about 5000 cpm. Parameters of the OPC Reaction. OPC light emission kinetics showing an increase in OPC emission upon turning on the laser circuit and a decrease in OPC emission after turning off the laser circuit are shown in Figure 6. In this experiment the laser circuit was turned on at 12 s and turned off again at 50 s for a total time of 38 s of pumping red-light energy into the OPC reaction solution. Upon turning on the laser circuit OPC emission increases gradually for about 1 min. Upon turning off the laser circuit, OPC emission decays with a half-life of about 1s. The slow increase in OPC emission compared to the fast decrease upon removal of red excitation light may indicate that concentration of an inhibitor (e.g oxygen) is slowly decreasing during the first minute of red light activation. The reaction time course was somewhat sensitive to the intensity of red excitation light and increased faster for stronger light sources. Figure 7 shows chemiluminescence light produced during the first 2 s after turning on the red laser circuit. Figure 7 also shows the light produced during this time from a sample that had been degassed to remove most (but not all) of the oxygen. The total level of light produced from the degassed sample is lower than that produced from the nondegassed sample partly because the degas technique caused some loss of photoactive dye. A comparison of these last two figures indicates oxygen may quench OPC chemiluminesce especially during the first few seconds of reaction. Thus, type I photosensitized oxidation may be the dominant reaction under the conditions used because in this reaction type oxygen at high concentrations would be expected to compete with the direct reaction between dye and substrate. It should be stressed that some oxygen is needed for the type I reaction (Figure 1)and that oxygen has important roles to play in both reaction types. Which reaction type is dominant for a particular condition is often determined by the relative concentrations of photosensitive dye, other substrate(&,and ~ x y g e n .We ~ found that addition

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

407

(D

0 F

x

x

1-

B

n

0

I

04 0

1

2

5

Reaction Time (seconds) Flgwo 7. OPC reaction time. The first 2 s after twnlng on the laser: (13)chemiluminescence from 0.1 mL reaction solution; (+) chemiluminescence from 0.1 mL of reaction solution after degasslng to

10

15

20

25

Methylene Blue (pglrnl) Flgwo 8. Methyleneblue standard curve. One-tenth-mlllter solutbns of 1 mM luminol In 0.05 M NaOH were used.

deplete the oxygen.

of nucleophiles to the reaction mixture often quenched the OPC reaction. When the singlet oxygen quencher (l,&diphenylisobenzofuran)11 was included in the OPC reaction, light output decreased by only 5% but when the soluble free radical quencher ((3-tert-butyl-4-hydroxy-5-methylphenyl)isobenzofuran)was included, light output decreased more than 99 9%. This data suggests that the singlet oxygen (reaction type 11) pathway is not very important under the conditions used but that reaction type I (direct reaction of excited photoactive dye with substrate) is dominant (Figure 1). The optimum pH for OPC light emission was between 12 and 13. OPC light emission dropped more than 25 times from pH 12 to pH 11and slowly dropped by two-thirds from pH 12.7 to pH 13.7. When this experiment was performed with methylene blue derivatized protein, the optimum pH was shifted higher to pH 13. This may have resulted from protein-bound dye becoming more accessible to solvent at high pH while the protein became denatured. An additional observationin support of this explanationwas that the activity of photoactive dye bound to protein gradually increased with time during the OPC reaction a t high pH. From these experiments, a pH of 13 was chosen as optimum. The OPC reaction was fairly insensitive to temperature. There was almost no change in activity between 24 and 44 "C and there was a 20% drop in activity from 24 to 6 OC. This temperature sensitivity is much smaller than that normally seen with enzyme-catalyzedreaction systems which typically increase reaction velocity by 10% per degree increase in temperature. In fact when this temperature study was repeated with enhanced HRP chemiluminescence the temperature sensitivity was a t least 5 times higher. OPC light emission increased with luminol concentration up to 0.25 mM luminol but increased only 20% from 0.25 to 1mM luminol. The concentration of luminol which gave the best detection (signal to noise ratio; S/N) was found to be 1 mM because the background light signalcontinued to increase at higher concentrations of luminol. The recommended detection reaction solution (1 mM luminol, 0.1 M NaOH) was found to be stable for at least 3 months during storage at room temperature and room light conditions. By contrast, the substrate solution used for enhanced HRP chemiluminescence was stable for only about 1h. A methylene blue standard curve is shown in Figure 8. The detection limit from this experiment was 0.2 pg (7.0 X 10-16 mol). A sample of methylene blue (2.2 X 10-13 mol) was

1 .o

1

R I

Q)

0

5

0.8-

2 0.6-

0.4

-

Flgurr 0. Absorbance spectrum of methylene blue derivative before (El) and after (+) coupling to antlbody.

found to catalyze chemiluminescence from 2.8 X 10-10 mol of luminol over a 2-h period. Thus each methylene blue molecule on average was responsible for chemiluminescence from 1273 luminol molecules over this time period. This indicates that methylene blue is recycled in the type I photosensitized oxidation reaction and does not become destroyed upon reaction with luminol. When up to three methylene blue dye derivatives were coupled to antibody protein the protein had activity in an immunoassay but coupling a t higher ratios caused lower activitywhen subsequentlyused in the alphafetoproteinaway. Upon covalent coupling of methylene blue derivative to protein, the absorbance spectrum of the dye changed (Figure 9). The maximum absorbance before coupling was a t 663 nm, and the maximum absorbance after coupling was a t 624 nm. There was a corresponding (about 50%) drop in chemiluminescence activity upon coupling to protein, and the change in absorbance spectrum is likely to be responsible for this. The bovine serum albumin and denatured serum albumin proteins were coupled (1:l)with antibody fragments via the antibody fragment hinge domain sulfhydryl residue. The usefulness of these protein conjugates for a-fetoprotein immunoassay using the OPC method was directly tested, and the results are summarized in Table 11. This table shows that denaturing bovine serum protein prior to coupling with dye catalyst either with detergent alone or with added reducing agent to remove disulfidelinkagea within the protein improved

408

ANALYTICAL CHEMISTRY, VOL. 65, NO. 4, FEBRUARY 15, 1993

Table I1

conjugate used IgGmethylene blue Fab-BSA-methylene blue Fab-BSA-methylene blue (detergent-treated BSA) Fab-BSA-methylene blue (detergent-and reducing agent-treated BSA)

____~

detection limit 1.56 ng >2.0 ng 0.33 ng 0.081 ng

the a-fetoprotein detection limit 20-fold (from 1.56 to 0.08 ng). The undenatured BSA conjugate had a high (0 ng of AFP) background of 60 208 cpm, but the conjugates made from partly and fully denatured BSA had backgrounds of 30 362 and 24 859 cpm, respectively. Thus, it seems that denaturing the BSA might have had an effect on lowering nonspecific binding of conjugate to the solid phase used in the immunoassay. An estimated 16 dyes per protein were incorporated into the fully denatured BSA. However, when IgG was labeled directly without carrier protein only about three dyes were incorporated into each IgG molecule. At higher dye incorporationratios the antibody binding activity became destroyed and gave poorer performance in the immunoassay. Also, the chemiluminescence activity of incorporated dye was about 40 5% as high compared to free dye in solution. Thus, when using direct labeling of antibody, an upper limit of about three dyes could be profitably incorporated per antibody, and the photoactive dye lost some activity upon binding to protein. The chemiluminescence activity of the denatured BSA-dye complex was higher than the native BSA-dye conjugate even though they had a similar amount of incorporated dye. This was probably because of less quenching of excited dye by aromatic residues of the protein that became disordered by the denaturation treatment. Much of the improved performance of this conjugate resulted from ita higher chemiluminescence activity. The sodium lauryl sulfate denatured Fab’-BSA-dye conjugate was used for development of a blood plasma AFP assay. Figure 10shows a standard curve obtainedwith this conjugate in the presence of 10% plasma. The detection limit for AFP in buffered solution was 17.4 pg, and the detection limit for plasma AFP was 1.5 ng/mL plasma. AFP added to three plasma samplesshowed good recoveries (104%,108% and 102%). The intraaasay CV was 8.2% and 7.9 % for 20 and 100ng/mL AFP plasma samples, respectively. The interassay (day to day) CV was 9.6% and 7.8% for 40 and 137 ng/mL plasma samples, respectively. Adult male plasma AFP concentrations are normally undetectable by most assay methods. In the present study AFP was added to 16 male plasma samples, and the concentrations were measured using the opticallypumped system. The correlation relationship between AFP added and AFP measured (I?)was calculated to be 0.98.

u,

0 T

X

2

a

0

0

100

200

300

AFP (ng/ml plasma) Flgure 10. Plasma a-tetoprotednstandard curve. Plasma was diluted 10-fold and assayed using a double antibody system with 0% detection.

CONCLUSIONS The technique of periodically turning the OPC laser on and off and measuring chemiluminescence light only during the off time allowed removal of the light-scattering background signal and improved the detection limit more than 10-fold. The oxygen removal and chemical quenching studies showed that the OPC reaction occurs via type I photosensitized oxidation reaction and that the dye is recycled. Because the dye is recycled, it may be possible to find accelerators of the reaction to improve light output in the future. The optimum substrate solution was found to be 1 mM luminol in 0.1 mM NaOH although conditions for other dyes were not similarly optimized. The temperature sensitivity and substrate solution instability were considerably less than that of enzyme systems. Thus, OPC may have value in diagnostic applications where convenience, temperature sensitivity, and substrate stability are concerns. Coupling of the methylene blue succinimide dye to protein was easy and the technique of using denatured bovine serum albumin protein improved assay sensitivity 20-fold without adversely affectingthe conjugate. Plasma a-fetoprotein measurements using the OPC method showed good recoveries,variabilities, sensitivity, and correlation. OPC should therefore be useful for the assay of other analytes as well.

ACKNOWLEDGMENT The authors wish to thank Mr. K. Oda for technical assistance and Dr. S. Terau for encouragement.

RECEIVED for review June 18, 1992. Accepted September 30, 1992.