WO1988007670A2 - Time-resolved fluorescence apparatus and immunoassay - Google Patents

Abstract

A method and apparatus for time-resolved fluorescence spectroscopy is described in which laser light from a single pulse is used to excite fluorescent photons in a sample (24), which fluorescence is detected by a PMT (14) optimized for linearity and response time to produce photoelectrons which generate a current at the PMT anode. This current is discharged through an R/C network to produce a voltage amplitude waveform which is converted to an optical image (26), intensified (28), stored (36) and digitized (34). The digitized version of the optical image is processed in a data processor (38) to calculate the true fluorescence impulse response. Fluorescence immunoassay for determining single or multiple analytes based upon a single measurement of fluorescence are described.

Description

TIME-RESOLVED FLUORESCENCE APPARATUS AND IMMUNOASSAY

Description

Technical Field The invention relates to fluorescence spectroscopy and instrumentation and methods employed therein.

Background Art

Pulse fluoro etry is one of the theoretically possible approaches for performing time-resolved fluorescence measurements. An overview of this field is presented in "Time-Resolved Fluorescence Spectroscopy" by S.M. Fernandez in Fast Methods in Physical Biochemistry and Cell Biology, R.I. Sha'afi and S.M. Fernandez (Eds.) Elsevier Science Publishers 1983.

A general technigue for studying the dynamic behavior of a physical system is to determine the response of the system to an "instantaneous perturbation"; i.e., its impulse response. The impulse response of a fluorescent system to an instantaneous light pulse consists of the intensity of the fluorescence emission as a function of time after the disappearance of the perturbing impulse. For a homogeneous fluorescent system, i.e., one made up of identical fluorescent molecules, the impulse"^* response takes the form of an exponential decay: f (t) = Aexp (-k/t) (1)

where f(t) is the intensity of the fluorescence at time t and k is the rate constant for emission, i.e., the reciprocal of the fluorescence lifetime T. The amplitude, A, is a pre-exponential factor that corresponds to f(o) , the fluorescence intensity at time zero. The fluorescence lifetime is a characteristic of the molecules under observation. Under suitable conditions, the amplitude, A, is proportional to the concentration of the fluorescent molecules.

When the fluorescent system under observation consists of a heterogeneous population of emitting species, then, in principle, the impulse response takes the form of a sum of exponentials:

f(t) = ∑ A.exp(-k.t) (2) j ^{J J}

where the k. are the rate constants for emission by each species and a. correspond to the relative concentrations of each species. There are cases where fluorescence decay curves can take a form other than those represented by

Equations 1 and 2.

In practice, it is not possible to generate

"instantaneous" excitation light pulses. Instead,' light pulses of short duration relative to the decay being observed are employed. Since fluorescent lifeti es typically are in the range of nanoseconds, light pulses of a nanosecond, or less, duration are generally required. The finite duration of these exciting pulses distorts the measured impulse response of the system with the result that f(t) is no longer directly observable. Instead, one obtains the distorted response F(t) given by: t F(t) = fΕ(t-t')f(t')dt' (3)

"o where E(t) is the time profile of the exciting light pulse as measured by the detection system and t' is a point in time prior to t. In practice, therefore, the true impulse response f(t) is obtained from the measured E(t) and F(t) by mathematical deconvolution of Equation 3 (using a computer) . The above discussion presents the basic principles behind the technique of pulse fluorometry. The fluorescence decay curves F(t) obtained with this method provide useful information in the study of a wide variety of physical, chemical, and biological phenomena, and therefore, pulse fluorometry has become a well-established analytical tool.

In general, a pulse fluorometer consists of a pulsed light source for excitation of the sample; photodetector(s) for detecting the resultant fluorescence photons and generating an electrical signal; a signal processor for processing the information contained in the electrical signal; and a computer for data analysis. Within this general framework, a number of different methods and specific instrument configurations have been employed. The fluorescence detection methods in common use fall into four categories: analog methods using a photomultiplier tube (PMT) and an oscilloscope; transient digitizers; time-correlated photon counting methods employing a time-to-amplitude converter (TAC) and a multichannel analyzer (MCA) ; ■ and streak camera detection. Each of these approaches is briefly discussed below: 1. Analog Methods

In analog systems, the fluorescence emission is detected with a photomultiplier tube (PMT) . The time-dependence of the emission is obtained by displaying the photocurrent output of the PMT on a cathode ray tube (CRT) oscilloscope. The time resolution of this approach is limited by the analog response time of the PMT and by the bandwidth of the CRT oscilloscope. The data, being in analog form, are not amenable to numerical computer analysis for deconvolution. This approach, therefore, is not suitable for analyses that require high precision and accuracy, such as when the decay is short relative to the width of the exciting pulse, or when multiexponential analysis, i.e., analysis of heterogeneous emissions is required. An additional drawback is that continuous irradiation of the sample with light pulses at a high repetition rate is necessary to obtain a relatively steady display of the fluorescence waveform.

Pulse sampling is another method that has been used for obtaining fluorescence decay curves. Specific devices that have been used for this purpose include the "box-car integrator" (Badea and Georghiou, Rev. Sci. Instr. , Vol. 47, 1976) and sampling oscilloscopes. In essence, with pulse sampling, the intensity of the fluorescence emission (output from the PMT) is sampled at a series of sequential time intervals which are delayed with respect to an excitation trigger. Only one point per waveform is sampled, therefore, the fluorescence decay is reconstructed, indirectly, from sampling a large number of events. With these methods, it is also necessary to irradiate the sample, repeti¬ tively, at a high rate. As one illuminates a fluorescent material repeatedly, the fluorescence characteristics may change due to bleaching or other photophysical effects. Useful fluorescence probes are often photolabile and degrade upon prolonged exposure to light.

2. Transient Digitizers A different approach is that of devices that sample many points on the same waveform (see, for example, Crosby and MacAdam, Rev. Sci. Instr., Vol. 52, 1981) . Some of these devices perform an analog average over repeated signals for every sampled point. Digital instruments of this kind, known as transient digitizers, are also available. Devices of this type result in faster data acquisition, as compared to "single-point-per- waveform" apparatus. Conventional transient digitizers, based on fast analog-to-digital converters, are limited in resolution to about one point for every 2 nanoseconds and, thus, are not suitable to measure short-lifetime decays, nor can they be used for accurate multiexponential analysis in the nanosecond time scale. 3. Time-correlated Single Photon

Counting Method

This is a method that circumvents the analog response of the PMT. Instead, the PMT is used to count and time the arrival of individual photons. The occurrence of the exciting light pulse starts a fast clock; detection of the first single photon at the PMT stops the clock. The duration of the time interval elapsed between start and stop is stored. The cycle is repeated a large number of times until a histogram of the distribution of the time intervals measured can be constructed with the desired precision. This histogram then represents the time-dependence of the fluorescence decay.

In practice, the fast clock takes the form of a device called a time-to-amplitude converter, and the data are stored in a multichannel analyzer (for further details, see the Fernandez reference, cited above, or "Time Decay Fluorometry by Photon Counting" by I. Isenberg in Biochemical Fluorescence, Chen & Edelhoch (Eds.), Marcel Dekker, 1975) . In order that the histogram of time intervals, obtained by this method, be an accurate representation of the fluorescence decay, statistical considerations dictate that the probability of more than one photon arriving at the detector during each measurement cycle, be very small. This requires that the photon counting rate be not more than 1% of the light source repetition rate. An advantage of this method is that the data is obtained in digital form and, thus, can be easily manipulated and analyzed by computer. A drawback is that the sample must be repetitively excited a large number of times before data of sufficient precision is acquired. This is a serious limitation when the nature of the application requires data of high precision but rapid collection time, or when the sample degrades upon prolonged exposure to exciting radiation. For example, with a lamp repetition rate of 10 KHz, the maximum allowed photon counting rate is of the order of 100 photons per second (1%) . For a multiexponential decay curve consisting of 256 data points, a total of 10 million counts may be required to be collected to achieve sufficient precision. At the stated count rate, it would take approximately 28 hours to collect the data. Even if one million counts were sufficient, data collection would still be of the order of three hours. There are certain applications of pulse fluorometry where it is necessary to collect several decay curves, from a sample, in a relatively short time. For example, in studies of the time-dependence of the fluorescence emission anisotropy, two decay curves of different polarization must be collected. This would double the data collection time. In studies of time-resolved emission spectra, it is necessary to collect a family of curves at different wavelengths. With the time-correlated single-photon counting approach, the data collection time may become unacceptably long, especially in the case of weakly emitting samples. 4. Streak Camera Detection

When the fluorescence pulse is very intense, a streak camera can be used to detect the decay (see, for example, "Topics" in Applied Physics, S.L. Shapiro, (Ed.), Vol. 18, Springer-Verlag, 1977) . Generally, this method is useful when picosecond resolution is required. Streak cameras, however, are extremely expensive.

FLUORESCENCE IMMUNOASSAY

Since immunoassay was first described by Berson and Yalow in the late 1960's, a number of immun- oassays with different assay formats (e.g. com¬ petitive vs. noncompetitive) , different separation techniques (e.g., secondary antibody separation vs. solid phase separation), and different labels (e.g., isotopic vs. non-isotopic) , have been developed for the detection of a wide spectrum of substances including molecules as small as digoxin and as large as a virus. Until recently, all these assays employed only one signal-generating label at a time which resulted in the limitation of detecting one substance (analyte) in one measurement.

The use of two radioisotopes, i.e., iodine 125 and cobalt 57, in a radioimmunoassay for detecting two analytes simultaneously in a single measurement has been demonstrated in a Vitamin B 12/Folate test

(U.S. Patent No. 4,146,602, Gutcho et al. ) a TSH/ Free T4 test (Simultrac TM FT4/TΞH, Becton Dickinson

Immunodiagnostics) , and the LH/FSH test (ComboStat TM LH/FSH, Micromedic Systems, ICN) . In these assays, the discrimination between radioisotopes in the measurement is based on differences in their energy spectra.

The use of two fluorophores, e.g., fluorescein and rhodamine, in a qualitative immunoassay for detecting two substances simultaneously in a single sample but not in a single measurement has been demonstrated n the cell sorting techniques (FACS TM

Cell Sorting System, Becton Dickinson) . Recently, the use of two fluorophores, i.e. fluorescein and phycoerythrin, in a quantitative immunoassay for detecting human IgG and IgM in a single sample but not in a single measurement, was reported (R.

Houghton, Abstract, Clin. C e . Vol. 32, p. 1067, 1986) . In these fluorescence-based immunoassays, the resolution of the fluorophores in the measure¬ ment is based on the differences in their excitation and emission spectra.

A method of time-resolved immunoassay is described in U.S. patents 4,058,732 and 4,374,120. In this method a fluorescent probe is employed that has a fluorescence decay (lifetime) that substan¬ tially exceeds the duration of the exciting pulse and the duration of the background non-specific fluorescence. A time-gating is used to reduce the background fluorescence, i.e., the measurement of the fluorescence is delayed until a certain time has elapsed from the moment of excitation. The delay time is sufficiently long for the background fluorescence to have ceased. When the fluorescence signal is measured (after the delay) the measurement is an integrated measurement, i.e. all the light arriving at the detector during the measuring period is measured without regard to the time of arrival. The purpose of this delayed measurement is to ensure that only one fluorescence signal reaches the detector during measurement.

Disclosure of the Invention

The present invention circumvents the need to repetitively excite a sample over a period of time and is capable of generating digital data of high precision in an extremely short time, i.e., less than one second per sample. The cost of the apparatus, which is assembled from commercially available components, is low, compared to alternative approaches.

The apparatus comprises a stable pulsed laser light source for emitting substantially monochromatic light to excite fluorescence in a sample contained in a sample holder. Optical filter elements are employed to isolate a narrow optical band of the emitted fluorescence from the sample. A photomultiplier tube (PMT) , optimized for linearity and response time, is used to detect the fluorescence transient, i.e., the photons emitted over the decay time period produced by a single excitation pulse from the laser. The PMT produces photoelectrons at the PMT photocathode in response to the fluorescence photons. The photoelectrons are accelerated through a series of dynodes in a vacuum to generate cascades of secondary electrons. These secondary electrons form a small current pulse at the PMT anode. This current pulse is discharged through a resistance/capacitive (RC) network to produce a time varying voltage amplitude signal. This electrical voltage waveform is then converted to an optical image, intensified and stored for digitization. Conversion to an optical waveform may be conveniently accomplished by coupling the PMT voltage signal to the vertical sweep of a CRT while driving the horizontal input with a fast sweep generator; thereby displaying the voltage amplitude signal on the CRT face. The waveform image on the CRT is intensified by a microchannel plate (MCP) image intensifier. The image intensifier is optically coupled to a solid state image sensor, such as a charge coupled device (CCD) , which stores the image. This stored image is decoded into a digitized data set by known means and sent to buffer memory.

The impulse response Eft) of the detector system to the laser pulse is also detected and digitized and stored in memory, as above. The buffered digitized waveforms are then processed, as follows, in a computer to determine the true fluorescence impulse response. The true fluorescence impulse response f(t) to a single excitation pulse has been masked, or distorted, by the finite width of the exciting pulse and by the ^" impulse response of the detector circuit. Thus, the signal stored in memory is not the true fluorescence impulse response f(t) , but a distorted version thereof, F(t) . The computer, by an iterative data fitting process determines the parameters of the true fluorescence impulse response f(t) , by convoluting E(t) with a trial function corresponding to an estimate of f(t) . The trial function is chosen on the basis of physico-chemical considera- tions. Parameters of the trial function are adjusted to generate a calculated function F(t) .. , which is compared point-by-point with the actually measured fluorescent decay curve F(t) .

The variance between the two curves, i.e., the measured versus the calculated curve, is minimized by a "least square" method, to obtain parameters that best fit the experimental data. To do this, the parameters of the trial function curve are adjusted, iteratively, until the "chi-square" function " " , that describes the variance between the experimentally measured fluorescent decay curve and the calculated decay curve, is minimized. At each iteration step, the minimization criteria lead to improved values of the parameters until successive values of the chi-square function differ by less than some arbitrarily chosen convergence criteria.

Unlike the prior art time-correlated single photon counting technique, the method of the present invention processes substantially all the photons excited by a single excitation pulse arriving at the photomultiplier. Furthermore, fluorescence decay curves of sufficient accuracy to perform multiexponential analysis by deconvolution are collected from one single excitation pulse. Thus, it is not necessary to repetitively excite the sample. These and other features and advantages of the invention will now be described in detail in connection with the following drawings. This invention also pertains to a fluorescence immunoassay for the detection of one, two or more analytes in a single sample which requires only a single measurement of the signal generated by the fluorescently labeled specie(s) . In assays for multiple analytes, different fluorescent probes having distinct fluorescence lifetimes are used to selectively detect and quantify the different analytes.

The fluorescence immunochemical assays of this invention can be performed in the heterogeneous or homogeneous, competitive or noncompe itive modes. In each assay mode, a fluorescently labeled analyte or antibody is employed and the fluorescence emitted is proportional (either directly or inversely) to the amount of analyte. The amount of fluorescence is determined by the amplitude of the fluorescence decay curve for the fluorescent species. This amplitude parameter is directly proportional to the amount of fluorescent species and accordingly to the analyte.

In general spectroscopic measurement of fluo¬ rescence is accomplished by: a. exciting the fluorophore with a pulse of light; b. detecting and storing an image of the excitation pulse and an image of all the fluorescence (the fluorescent transient) induced by the excitation pulse; c. digitizing the image; e. calculating the true fluorescent transient from the digitized data; f. determining the amplitude of the fluorescent transient as an indication of the amount of fluorescent species. In assays for multiple analytes where two or more fluorescent species contribute to the overall fluorescence, the fluorescent transient for each component is determined based upon the known fluorescence lifetime of each fluorescent specie.

According to the method, substantially all of the fluorescence emitted by the fluorescent specie(s) reaching the detector^'as a function of time from the instant of excitation is measured. As a consequence, the signal being detected is a superimposition of several component signals (for example, background and one analyte specific signal; or signals from different labeled analytes in the case of a multiple analyte assay, etc.) . As men¬ tioned, the individual contributions to the overall fluorescence reaching the detector are distinguished based on the different fluorescence decay rates (lifetimes) of signal components. In order to quantitate the magnitude of each contribution, the detected signal data is processed to obtain the amplitude of each component. The amplitude of each component signal is proportional to the concentra- tion of the fluorescent specie(s) .

The advantages of the time-resolved fluores¬ cence immunoassay of this invention include no handling of hazardous radioactive material, and the ability to determine more than one analyte in a single sample in one measurement which is cost effective. Furthermore, the present invention minimizes the problem of background noise which is commonplace in the fluorescence immunoassay (espe¬ cially in the homogenous mode) . This is because the decay lifetime of background fluorescence is usually significantly shorter than the decay lifetime of the fluorescence of fluorophores used in the immuno¬ assay.

Brief Description of the Drawings Fig. 1 is a block diagram of a time resolved fluorometer, in accordance with the invention.

Fig. 2 is a flow chart of the process for determining parameters of the digitized waveforms.

Fig. 3 is a flow chart of the steps for obtaining the concentration parameter of a fluorescent sample of known T.

Figure 4 shows an inhibition curve for a competitive binding assay for bovine serum albumin (BSA) where the amount of fluorescently labeled BSA bound is determined by measuring fluorescence intensity.

Figure 5 shows an inhibition curve for a competitive binding assay for human serum albumin (HSA) where the amount of fluorescently labeled HSA bourid is determined by measuring fluorescence intensity.

Figure 6 shows an inhibition cuve for a competitive binding assay for BSA where the amount of fluorescently labeled BSA bound is determined by the amplitude of the fluorescence decay curve. Figure 7 shows an inhibition curve for a competitive binding assay for HSA where the amount of fluorescently labeled HSA bound is determined by the amplitude of the fluorescence decay curve. Figure 8 shows an inhibition curve for a competitive binding assay for simultaneous determination of BSA and HSA, wherein the amount of labeled BSA and HSA- is determined by the amplitude of each component of the fluorescence decay curve.

Best Mode of Carrying Out the Invention

Generally, the apparatus of the invention consists of a laser module 10, a sample compartment 12, an image/storage/digitizer system 40, and a computer 38. Within the laser module,, a pulsed laser 26, preferably a nitrogen dye laser, is contained in an RFI/EMI shielded compartment 11 electrically isolated from sample compartment 12. Suitable power supply and control logic circuitry is also enclosed, but not shown, within module 10. A single excitation pulse (dotted line) from laser 26 is used to excite a sample 24 disposed on, or within, a holder 18 in compartment 12. Preferably, laser 26 provides an energy output of at least 15 icrojoules per pulse. This is equivalent to approximately 3 x 10 13 photons per pulse. The full width at half-maximum of each pulse is preferably about 3, or less, nanoseconds. This pulse energy should be substantially stable and repeatable over long periods of time. The spectral bandwidth of the laser pulse is preferably less than 1 nanometers, so that the light is essentially monochromatic. With a N_/DYE laser output wavelengths are tunable within a range of 360-900 nanometers.

The sample 24 may be in liquid form and contained in a quartz cuvette of conventional design. Alternatively, the sample may be in solid form and held in a suitable sample holder. The laser pulse is passed through a neutral density filter F and impinges on a beam splitter, or beam sampler, 16. About 10% of the light is reflected to a reference diode detector D. The other 90% passes through a rotatable graduated neutral density filter F_ to the sample 24 to produce fluorescence photons which are focused by lens L, passed through a narrow passband interference filter F_ and impinge on the sensitive area of PMT 14, producing photoelectrons at the cathode C thereof. These electrons are accelerated by successive dynode stages D -D^. producing secondary electrons and, ultimately, a small current at anode A. This current is passed through R/C load circuit to ground producing a voltage amplitude waveform V, which is coupled via a preamplifier 23 to the vertical input, Vert, of the CRT 26. The horizontal input, Hor'z, of the CRT is coupled to a fast sweep generator 22. In this manner, substantially all the fluorescent light photons reaching the PMT 14 from the excited sample are converted to photoelectrons to produce the amplitude varying voltage waveform V displayed on the face F of CRT 26.

The portion of the fluorescence photon emission split at beam sampler 16, detected by diode detector D, used, via filter control circuit, 20, to control the rotational position of a gradient density filter F disposed in the main path of the photon beam B. Filter F is mounted on a rotatable shaft S coupled to m'otor M. Thus, by providing an error signal V , a suitable sensor control signal from circuit 20 can be generated to control the position of motor M to maintain a constant incident light intensity or to attenuate it appropriately, i.e., to prevent photodetector saturation.

Beam stopper 25 is a light trap, which is used to prevent undesirable back scatter by absorbing specular reflection. Filter F is a narrow bandpass interference filter which minimizes the effects of Raman scattering, which often occurs at wavelengths between that of the exciting beam and those of the fluorescent emissions.

It is important, from the standpoint of instrument sensitivity, that substantially all the photons reaching the PMT detector surface from a single excitation pulse contribute to the useful output voltage signal. This means that these photons must be detected and converted to photoelectrons and collected at the anode in about 1/2 microsecond or less. Therefore, the PMT should be optimized for speed, range and linearity rather than gain as is customary in other applications.

The analog time variant voltage amplitude signal corresponding to the fluorescence decay curve of the excited sample is displayed on the face F of CRT 26 for a time period that depends upon the persistence of the materials used on the CRT face. Because of the high sweep rate, the CRT trace is of low intensity and the waveform must be intensified and stored so that it can be digitized via scanning analog to digital (A/D) circuitry. This is accomplished via a microchannel plate image intensifier 28, optically coupled to CRT face F, and employed to intensify the CRT display image. The intensified image is then optically coupled to a solid state image sensor 30, a charged coupled device (CCD) which stores the image in the form of charges of varying magnitude on an array of CCD's. The stored image is then raster scanned by a raster scan circuit 32, digitized in digitizer 34 and stored in frame buffer 36 for analysis by computer 38. The impulse response of the system to the exciting laser pulse is also processed, as above, and stored in memory by replacing the sample 24 with a non-fluorescent light scattering material, such as latex microspheres.

The digitized data transmitted to the computer 38 from the buffer 36 consists of sets of numbers which represents points on a plot of the variation of the intensity versus time of the excitation light pulse and of the emitted fluorescence respectively. As previously noted, the intensity-time profile thus obtained is not an accurate representation of the true time profile of the fluorescence emitted. Distortions, due to the instrument response characteristics, affect the time profiles measured. These distortions can be grouped into two types: Those arising from the time response of the photomultiplier tube (PMT) 14 and associated circuitry and those -arising from the finite duration of the excitation light pulse from the laser.

A. Distortions Arising from the Finite Width of the Exciting Pulse

The laser pulse used to excite the fluorescent sample is not a true impulse function, since its time profile e(t) has a finite width. Consequently, different molecules of the fluorescent sample are excited and begin emitting light at different times.

This produces a distortion, which is most significant at the beginning of the fluorescence decay and with fluorophores having a short decay time. In any case, it complicates the process of establishing the origin of the waveform, which must be accomplished accurately if subsequently determined concentration values are to be valid.

The time profile F' (t) of the fluorescence emission, as distorted by the finite duration of the incident exciting pulse, is given by the convolution integral of the exciting pulse shape e(t) and the true fluorescence impulse response function f(t) :

t

Wherein, F' (t) represents the time profile of the fluorescence signal incident on the photomultiplier, and f(t) is the waveform of interest and t* is a point in time prior to t.

B. Distortions Arising from the PMT and Associated Circuitry The response of the PMT 14 and its associated circuitry is neither instantaneous nor linear over the full dynamic range. Thus, the PMT response to a time-varying incident light signal is not an exact representation of the incident time profile, but is distorted by the PMT characteristics. This fact alters the shape (time profile) of the observed data curves of both the excitation pulse e(t) and the fluorescence decay f(t) curves.

As previously stated, the detector is first set up to directly observe the exciting pulse by replacing the fluorescent sample, 24, in the sample holder 24a with a non-fluorescent scatterer and allowing the scattered incident light to reach the PMT 14. The output of the PMT will not, in practice, be an accurate representation of the incident pulse time-profile, due to the PMT impulse response characteristics. If the true time profile of the exciting light pulse is designated e(t) , then the measured time profile will be a distorted function E(t) . E(t) is related to e(t) by a convo¬ lution similar to that of Equation 1. Namely, E(t) is the convolution integral of the true light pulse shape e(t) and the impulse response of the PMT and its associated circuitry, I(t) :

t E(t) = i(t^l)e(t-t^l)dt' (5)

"o

When a fluorescence decay curve is to be observed, the appropriate fluorescent sample 24 is placed in the sample holder 24, and a suitable narrow bandpass optical filter 16 is placed in front of the PMT to isolate the spectral range of the fluorescence and to reject scattered incident light. As discussed above, the time profile of the fluorescence signal incident on the PMT is F' (t) and is given by Equation 4. This signal F' (t) will also be distorted by the impulse response of the PMT and associated circuitry, so that the output of the PMT will be a distorted function F(t), where the incident signal F' (t) , and the observed signal F(t) are related through a convolution integral with the PMT impulse response I(t) , similar to Equation 5: t F(t) = j I(t' )F' (t-t')dt' (6) o C. The True Fluorescence Impulse Response As shown above, the experimentally determined waveforms E(t) and F(t) are distorted versions of the true exciting light pulse profile e(t) and the fluorescence signal incident on the PMT, F'(t) . Furthermore, E(t) represents the light pulse distorted by the detector impulse response, and F(t) is a representation of the true fluorescence impulse response f(t) that has been distorted twice. The first distortion is caused by the finite width of the exciting pulse (Equation 4) and the second, by the impulse response of the detector I(t) .

The waveform of interest is f(t) , the true fluorescence impulse response, and it is not directly observable. It can be shown that f(t) is related to the experimentally determined waveforms E(t) and F(t) by the following convolution integral:

t

F (t) = Ε (t ' ) f (t-t' ) dt ' ( 7 )

The waveform f(t) can be obtained from Equation 7 by computer analysis, as described below in connection with the flow chart of Fig. 2.

D. Computer Analysis

Two sets of data are obtained, as above. One set represents E(t) , the amplitude versus time waveform, or curve, of the observed exciting light pulse, and and the other represents F(t) , the observed fluorescence decay curve. From these two sets of data, f(t) is calculated by computer analy¬ sis from Equation 7. E(t) and F(t) are described by sets of N numbers which are stored in the computer for numerical analysis. Each member of the number sets represents the digitized value of a point on one of the curves. These values are read and decoded from the data stored in frame buffer 34 (See Step 1, Fig. 2) . N such data points for each curve are selected to represent each of the waveforms E(t) and F(t) (See Step 2, Fig. 2) . Typically, 100 data points (N = 100) are sufficient to represent each curve with the desired degree of precision.

The definition of the nomenclature employed for the various waveforms is repeated below: e(t) = true shape of the exciting light pulse as emitted by the laser E(t) = observed shape of the exciting light pulse as distorted by detector f(t) = true fluorescence impulse response function F* (t) = fluorescent signal incident on the PMT distorted by shape of exciting light pulse F(t) = observed fluorescent signal distorted by both the light pulse and the detector

In order to obtain the parameters that best characterize f(t) , an analytical expression for f(t) is assumed a priori. This analytical expression is called the "trial function", and is based on the physical nature of the problem. For example, in the case of a single non-reactive fluorescent species, it is known by physico-chemical considerations that the fluorescence decay is essentially exponential in nature, therefore, a trial function which is a single exponential decay, is chosen for this case (See step 3 Fig. 2), i.e.:

f(t) = A exp(-t/T) (8)

In this example, the two parameters, A and T, fully characterize f(t) . A is the amplitude of the fluorescence waveform, and T is the fluorescence lifetime. The determination of these two parameters is the ultimate goal of the measurement process.

If the sample contains two non-reacting fluorescent species, then physico-chemical considerations dictate that the trial function be a sum of two exponentials, i.e.:

f(t) = A_ιeχp(-t/T₂) + A_exp(-t/T₂) (9)

where now four parameters, two amplitudes, Al and A2, and two lifetimes, Tl and T2, are necessary to fully characterize the fluorescence decay f(t) of the two species.

Once a trial function is chosen on the basis of physico-chemical considerations, such as known fluorescence decay kinetics, the parameters that best characterize it, based on the measured E(t) and F(t) must be found. These parameters, once determined, constitute the "result" of the measurement. Therefore, in the next step (Step 4, Fig. 2) , a set of trial parameter values for the trial function are chosen; again, based on prior experience as to anticipated ranges of values. Determination of the best parameters is accomplished by an iterative data fitting process, whereby the trial function, with adjustable trial parameters, e.g., A and T, is convoluted with E(t) to generate a calculated F(t) , according to Equation

7 (Step 5, Fig. 2) . Let this calculated version of

F(t) be called F(t) ca.c. Now F(t) ca,c is compared point-by-point with the measured F(t) to calculate the variance between the two curves (Step 6, Fig. 2).

A least squares method (see, for example, "Least Squares Method of Analysis", B.K. Selinger, et al., in Time Resolved Fluorescence Spectroscopy in Biochemistry and Biology, R.B. Cundall and R.E. Dale, Eds., Plenum Press, New York, 1983) is used to obtain the parameters that best fit the experimental data. In essence, this is accomplished by selecting a new "best guess" trial function if an adequate fit between the first trial function and the experimental data is not achieved (Step 7, Fig. 2) . The "best guess" trial function is optimized by adjusting its parameters iteratively (Step 8, Fig.

2) , until the chi-square function, 2 that describes the variance between F(t) and F(t) cal,c is minimized. The chi-square function is given by: N * . 1 -^Fcalc^^ (10)

N-α 1=1 F(I)

wherein: N = the number of data points

I = an integral index number which identifies the data points and ranges from 1=1 to N

F(I) = the absolute magnitude of F at point I

The process of searching for the best parameters is optimized by simultaneously varying the set of fitting parameters according to the algorithm of

Marquardt (Marquardt, D.W. , 1963, "An Algorithm for Least-Squares Estimation of Non-Linear Parameters", J. Soc. Industr. Appl. Math., 11, 431-441) . At each iteration step, the minimization criteria lead to improved values of the parameters until successive values of the chi-square function differ by less than some arbitrarily chosen convergence criterion.

E. Application to Time-Resolved Immunoassay

In the preceding description, the problem of data handling and the analysis of fluorescent decay spectra, was described generally, without reference to any specific application. Specific examples of the use of the method and apparatus of the invention for the performance of a fluorescent immunoassay is now described in connection with Fig. 3. In this example, the fluorescence from an immunoassay sample is analyzed. In the simplest form, such a sample may comprise a fluorescent species, namely, a fluorescently labelled probe employed in the immunoassay.

For example, in a standard competitive type immunoassay, an analyte competes with a fixed amount of fluorescently labelled analyte (probe) for binding to antibody against the analyte. The concentration of labelled analyte bound to the antibody is indirectly proportional to the amount of analyte present in a test sample. In a noncompetitive type immunoassay, a fluorescently labelled antibody is used as a probe for analyte complexed with a first (capture) antibody (present in excess) against the analyte. The label associated with the complex is directly proportional to the amount of analyte in a sample.

From physico-chemical considerations, the fluorescence decay curve f(t) , for the fluorescent label, is represented by a single exponential:

f(t) = Aexp(-t/T) (11)

The lifetime T is known a priori, since a well-characterized probe of known fluorescent lifetime is chosen as the fluorescent label for the reagent (e.g., labelled analyte or labelled antibody, depending on the format of the immunoassay) . The amplitude A is the only unknown and is the parameter sought by the analysis. A is proportional to the concentration of the fluorescent species (la^'belled analyte or labelled antibody) , and, therefore, to the concentration of analyte. To perform the analysis in this example, a trial function of the form of Equation 11 is chosen in Step 3, Fig. 2, and the known lifetime parameter. T of the probe is also fixed to its known value in Step 4, Fig. 2. The function of the next steps in the analysis is simply to find the best value of A.

The value of A, thus obtained, is proportional to the concentration of the analyte. In order to determine the absolute concentration of the analyte from A, the value of A is compared to a calibration curve which has been previously obtained and stored in the memory of computer 38. In other words, prior to analyzing the unknown sample,- a series of standard samples of known concentration are analyzed, so that a calibration table can be made of values of A versus concentration.

This addition step (Step 10) is shown in the Fig. 3 flow chart of the computer program.

The procedure for a double analyte immunoassay is similar. In a competitive type assay, a fixed amount of each of two labelled analytes is employed with antibody against each analyte. In the noncompetitive format two different capture antibodies, each specific for one of the analytes, are used and two fluorescently labelled antibody probes, each having different fluorescence lifetimes, are used to determine the amount of each analyte-capture antibody complex.

In this case, the trial function is a sum of two exponentials:

f(t) = A_ιeχp(-t/ ₂) + A₂exp(-t/T₂) (12)

The two lifetimes, T and T , are known and the computer fits the two amplitudes, Al and A2 , that are proportional to -the concentrations of each analyte. The calibration curve for the computer memory and comparison Step 10, is generated from a series of standard solutions which contain both analytes in appropriate known concentrations.

Immunoassay Formats

The fluorescence immunoassays of this invention can be performed in competitive or noncompetitive format; and they may be performed in the hetero- genous or homogenous mode.

In a heterogeneous assay, the bound analyte is separated from unbound (free) analyte and the amount of bound or free is measured. Heterogeneous assays based upon the use of a solid phase to effectuate separation of bound from free analyte are described in detail below. In an homogeneous mode, no sepa¬ ration step is needed. The bound form of analyte (or antibody) and the free form are distinguished by a detectable change in the fluorescence lifetime of the labeled species which is induced by binding. The competitive format is based upon the ability of unlabeled analyte to compete with fluo¬ rescently labeled analyte and to inhibit the binding of labeled analyte to a limited amount of antibody against the analyte. As a result of the competitive inhibition the proportion of bound labeled analyte decreases as the concentration of unlabeled analyte increases. The concentration of unlabeled analyte in the sample is determined by referring to a standard inhibition -curve.

In a competitive assay for two or more ana¬ lytes, antibody for each analyte to be determined is incubated with the sample to.be tested and with fixed amounts of the analytes, each analyte being labeled with a distinct fluorophore. As a conjugate species, each labeled analyte has a different, known fluorescence lifetime (see ^'below for a detailed discussion for the properties of the fluorophores and fluorophore conjugates) . Based on this dif- ference, the contribution of each fluorophore to the total fluorescence emitted upon excitation can be distinguished. The amount of each labeled analyte bound by antibody is determined by the amplitude of the fluorescence decay curve for each analyte. The amplitude of the fluorescence decay curve of each component is directly proportional to the amount of conjugate, which is inversely proportional to the amount of unlabeled analyte in the test sample.

Non-competitive immunoassays are based on the reaction of analyte with an excess amount of antibody (capture antibody) . The complexed analyte is then measured and this amount of complexed analyte is directly proportional to the amount of analyte in a sample. The immunocomplex between the capture antibody and analyte is detected with a fluorophore-labeled antibody conjugate against either component of the complex (i.e. the analyte or the capture antibody) . The antibody conjugate is reacted with the complex of capture antibody-analyte and the amount of fluorescent label associated with the complex is directly proportional to the amount of analyte in the sample. The amount of fluorescence associated with the complex is determined by measuring the amplitude of the fluorescence decay curve for the conjugate.

In an immunometric assay for multiple analytes a mixture of capture antibodies containing antibody specific for each analyte is employed. Fluores¬ cently labeled antibody conjugates specific for each capture antibody-analyte complex - each conjugate having a different fluorescence lifetime - are used to determine the amount of analyte complexed with its respective capture antibody. The fluorescent conjugate can comprise, for example, a fluorophore linked to a second antibody specific for an analyte. The fluorophore for each analyte-specific antibody has fluorescence characteristics which results in conjugates having different fluorescence lifetimes. The preferred assay format for simultaneous determination of two or more analytes is a solid phase immunometric assay. Especially preferred is the "sandwich" type assay. The preferred mode is the forward mode but reverse and simultaneous modes may also be used. In assays of this format, an immunoadsorbent is provided which comprises a solid phase to which is affixed capture antibody specific for each of the analytes to be determined. The immunoadsorbent is incubated with the sample to be tested under conditions and for a period of time sufficient for analytes in the sample to complex with their respective capture antibodies on the solid phase. After the incubation, the immuno¬ adsorbent and the sample are separated. Normally, the solid phase is then washed to remove unbound and/or nonspecifically adsorbed substances.

The amount of each analyte bound to the immuno¬ adsorbent is determined by means of fluorescent conjugates specific for each analyte which have distinct fluorescence lifetimes. As explained, the fluorescence lifetime provides a means of distin¬ guishing between the conjugates and thus, deter¬ mining the amount of each analyte.

In one embodiment of the assay, a solution of fluorescent antibody conjugates is incubated with the immunoadsorbent under conditions and for a period of time sufficient for the conjugates^" to complex analytes associated with the solid phase. After the incubation period, the immunoadsorbent and the solution are separated and the fluorescence associated with the solid phase is analyzed by time resolved spectroscopy as described herein. The fluorescence associated with solid phase may be analyzed while the fluorescence conjugate is affixed to the solid phase or the conjugate may be removed from the solid phase (e.g. by eluting the conjugate from the solid phase under alkaline conditions) and analyzed in suspension or in solution.

In another embodiment, the amount of analyte can be analyzed by employing a second antibody to bind the analyte associated with the immunoadsorbent and then employing a fluorescent antibody conjugate against the second antibody to determine the amount of analyte. In assays for multiple analytes, the second antibodies against each analyte are antigenically distinguishable. The conjugates comprise fluorescently labeled antibodies against the second antibodies. Each conjugate has a different fluorescence lifetime, which provides the basis for distinguishing the contributoion of each to the overall fluorescence signal generated upon excitation.

The second antibodies used in assays of this type can be derived from different animal species. In an assay for two analytes, for example, the second antibody against one analyte can be a murine antibody and the conjugate can be a labeled anti- murine antibody. The second antibody against a second analyte can be a rabbit antibody and the fluorescent conjugate can be a fluorescently labeled anti-rabbit antibody. The antibodies used in the assays of this invention can be monoclonal antibodies, polyclonal antibodies, or both. In the embodiment where a second antibody is employed, the preferred kinds of antibody are as follows: the capture (solid phase) antibody is a monoclonoal antibody, the second antibody against the analyte is polyclonal and the fluorescent conjugate contains a polyclonal antibody against the second antibody. The use of a poly- clonal antibody as second antibody and as the antibody component of the conjugates results in an amplification of fluorescence signal because of the multi-epitopic binding of polyclonal antibodies.

G. Fluorescent Conjugates The fluorescent conjugates comprise conjugates of a fluorophore and either an analyte or antibody (depending on the assay format) . The conjugates can be prepared by standard techniques for conjugation of fluorophores to proteinaceous and nonproteina- ceous analytes and/or to antibodies. For example, fluorophores can be conjugated to proteins via functional groups such as amine, carboxyl or sulf- hydryl groups. Fluorophores can also be linked through various carbohydrate moieties to carbo- hydrates or analytes which contain carbohydrate moieties such as glycoproteins. A preferred tech¬ nique for conjugation is the isothiocyanate tech¬ nique. The isothiocyanate group is very reactive to primary amino groups of proteins. Preferred fluorophores have appropriate groups such for linkage to protein either directly or via a linker.

Before use in the assays the conjugates are characterized as to spectral characteristics in- eluding optimal excitation and emission wavelength and fluorescence lifetime. All of these properties of the conjugate can be determined by standard techniques. The fluorescence lifetime of the conjugate may vary dependent upon the ratio of fluorophore to analyte/antibody in the conjugate. For proteins this ratio is referred to as the dye/protein (D/P) ratio. In general, conjugates having a high D/P ratio are preferred because this enhances the sensitivity of the assay. However, D/P ratios which are too high may lead to self-quenching or loss of immunoreactivity of a labeled antigen or antibody. The optimal D/P ratio for any specific conjugate can be determined empirically.

H. Fluorophores For the purposes of the invention a fluorophore can be a substance which itself fluoresces or can be made to fluoresce or it can be a fluorescent ana¬ logue of an analyte.

In principle, any fluorophore can be used in the assays of this invention. Preferred fluoro¬ phores, however, have the following characteristics: a. A fluorescence lifetime of greater than about 15 nsec; b. An excitation wavelength of greater than about 350 nm; c. A Stoke's shift (a shift to lower wave¬ length of the emission relative to absorp- tion) of greater than about 20 nm; d. For homogeneous assays, fluorescence lifetime should vary with binding status; and e. The absorptivity and quantum yield of the fluorophore should be high.

The longer lifetime is advantageous because it is easier to measure and more easily distinguishable from the Raleigh scattering (background) . Exci¬ tation wavelengths greater than 350 nm reduce background interference because most fluorescent substances responsible for background fluorescence in biological samples are excited below 350 nm. A greater Stoke¹s shift also allows for less back¬ ground interference. The fluorophore should have a functional group available for conjugation either directly or in¬ directly to analyte/antibody. An additional cri¬ terion in selecting the fluorophore is the stability of the fluorophore: it should not be photophys- ically unstable, and it should be relatively insen¬ sitive to the assay conditions, e.g., pH, polarity, temperature and ionic strength.

Preferably (though not necessarily) , fluo¬ rophores for use in heterogenous assays are rela- tively insensitive to binding status. In contrast, fluorophores for use in homogeneous assay must be sensitive to binding status, i.e., the fluorescence lifetime must be alterable by binding so that bound and free forms can be distinguished.

In choosing two or more fluorophores for use in assays for multiple analytes, the following criteria pertain: a. Ideally, the fluorophores should have substantial overlap of absorption bands so that they may be efficiently excited at a single wavelength. b. The emission wavelengths should have substantial overlap of emission bands so that the fluorescence contribution of each probe can be effectively monitored at a single wavelength. c. The difference in fluorescence lifetime between fluorophores should be at least about 5 nsec.

Examples of fluorophores useful in the inven¬ tion are naphthalene derivatives (e.g. dansyl chloride) , anthracene derivatives (e.g. N-hydroxy- succini ide ester of anthracene propionate) , pyrene derivatives (e.g. N-hydroxysuccinimide ester of pyrene butyrate) , fluorescein derivatives (e.g. fluorescein isothiocyanate) and rhodamine derivative (e.g. rhodamine isothiocyanate) . I. Other Parameters of Assays

For each incubation step in the various formats of the assays, the time and conditions of incubation are selected to ensure maximal binding of analyte to the immobilized antibody (the immunoadsorbent) and to the fluorescent conjugate. Optimal conditions of pH, temperature, incubation time, and ionic strength for each incubation can be determined empirically. In the heterogeneous solid phase assay of this invention, the immunoadsorbent is separated- from incubation mixtures containing the liquid test sample or the fluorescent probe. Separation can be accomplished by any conventional separation tech¬ nique such as sedimentation or centrifugation. Preferably, (though not necessarily) , the immunoad¬ sorbent is washed prior to contacting it, when required, with a second incubation medium and prior to measuring the amount of fluorescent conjugate associated with the immunoadsorbent. The washing removes nonspecific interfering substances or excess labeled antibody which may affect the accuracy and sensitivity of the assay.

The immunoassays of this invention are used to detect and quantify analytes in a liquid sample or histological specimen. Liquid samples include essentially all biological fluids such as blood, or components of blood such as plasma or serum and urine, lymph, etc. Also, the liquid sample may be a sample of a liquid medium in which lymphocytes or other mammalian cells have been cultured. They may also be extracts or supernatants of microbial cultures or environmental samples such as sewage effluents, etc. Histological specimens include tissue slices or cell samples.

Many types of solid phases can be employed in the preferred assays of this invention. These include beads formed from glass, polystyrene, polypropylene, dextran, and other materials; the microwells of a microwell plate or tubes formed from or coated with such materials, etc. The antibody can be either covalently or noncovalently bound to the solid-phase by techniques such as covalent bonding via an amide or ester linkage, biotin-avidin bridges or adsorption. Those skilled in the art will know many other suitable solid-phases and methods for immobilizing antibodies thereon, or will be able to ascertain such using no more than routine experimentation.

To determine the amount of analyte in a liquid sample, either the amount of fluorescent conjugate associated with the immunoadsorbent or the amount of unbound conjugate (i.e. that which remains in soluble form) , is measured. Generally, it is preferable to measure the conjugate bound to the immunoadsorbent because at very low concentrations of antigen, only small amounts of labeled antibody bind the immunoadsorbent. Thus, for accuracy the fluorescent label associated with the immunoad¬ sorbent should be measured.

The reagents for performance of the assays in their various formats may be assembled in assay kits. For instance, a kit for performing a solid phase immunometric assay for multiple analytes may comprise: a) a solid phase immunoadsorbent containing capture antibody specific for each analyte to be measured; and b) fluorescent conjugates comprising antibody specific for each analyte and a fluoro¬ phore, each conjugate having a different fluorescence lifetime.

A kit for performing a solid phase immunometric assay for multiple analytes where a secondary antibody is employed can comprise: a) a solid phase immunoadsorbent containing capture antibody specific for each analyte to be measured; b) second antibodies specific for each analyte, each second antibody being antigenically distinct; and c) fluorescent conjugates comprising antibody against the second antibody and a fluorophore, each conjugate having a distinct fluorescence lifetime. As mentioned, in a preferred configuration, the capture antibody can be monoclonal and the second antibody and antibody conjugate can be polyclonal.

J. Analytes

Virtually all types of analytes can be deter¬ mined by the method of the invention. These include hormones, vitamins, therapeutic drugs, drugs of abuse, tumor markers, neonatal markers, microbes, viruses and antibodies. Examples of some specific protein hormones include thyroid stimulating hormone (TSH) , Luteinizing hormone (LH) , Follicle stimu¬ lating hormone (FSH) , growth hormone (GH) , human chorionic gonadotropin (hCG) and adrenocortico- trophic hormone (ACTH) . Steroid hormones include androgens, progestins, estrogens, corticosteroids and aldosterone. Therapeutic drugs include theophylline, digoxin, dilantin and phenobarbital. Viruses such as hepatitis virus, human immunodeficiency virus and herpes viruses, viral components, or antibodies against these viruses (as an indication of infection) can also be detected. In the design of assays for two or more ana¬ lytes, clinically related analytes can be combined. Examples of such combinations are TSH/Free T. , LH/FSH, Vitamin B12/Folate, B-HCG/AFP, CEA/PAP, CK/CK-MB, LDH₁/LDH₂, HIV/HBAgs, HBAgs/HBAgc, Herpes/ Chlamydia and TSH/T /TBG.

K. Fluorescence Spectroscopy

According to the method of this invention, all the fluorescence reaching the detector as a function of time from the instant of excitation is measured. Thus, the detected signal is a superposition of several signals (for example, background and one analyte-specific signal; or signals from different labeled analytes in the case of a multiple analyte assay, etc.). The individual contributions to the overall fluorescence reaching the detector are distinguished based on the different fluorescence decay rate (lifetime) of signal. The amplitude of a component of the signal is proportional to the specie responsible for the signal component.

In the preferred mode, the amplitude of the fluorescence signal is determined by:

(a) exciting the fluorescent specie(s) with a single pulse of light energy to induce fluorescence;

(b) separately detecting in a detector and generating:

(i) an electrical signal corresponding to the fluorescence transient waveform

F(t) induced by the single pulse, as distorted by said detector and the pulse, and (ii) an electrical signal corresponding to the waveform E(t) of the single pulse, as distorted by the impulse response of the detector,

(c) separately displaying an image of the waveforms F(t) and E(t) ; (d) digitizing a predetermined number of data points on each such image, as digital numbers representing points on the waveforms; (e) storing the numbers in memory as data points of T(t) and F(t) ; and (f) calculating the true impulse response fluorescence waveform F(t) from the stored data point numbers by convoluting E(t) with a predetermined trial function F(t)cai,c having adjustable parameters and comparing the data points of the convo¬ luted F(t) , with data points cor- responding to F(t) data points. Where a single fluorescent species is employed (e.g. in assays for a single analyte) , the trial function curve is an exponential curve in the form of Aexp- (-t/T) wherein the parameter A is the amplitude and T is the fluorescence decay lifetime. Where a plural number of fluorescent species are employed (e.g. assays for multiple analytes) the trial function is the sum of a plural number, ith, of exponential curves A,exp(-t/T ) + A_exp(-t/T_)

A.. , exp(-t/T... ) . T, the fluorescence lifetime, is known in advance (as it is a property of the fluorescent conjugate (or fluorescent analogue) employed in the assay) and the parameter A is proportional to the concentration of the fluorescent specie. With two fluorescent species, the lifetimes of which are known in advance, the amplitude parameter is proportional to the concentration of each fluorescent specie. The absolute concentration of the fluorescent specie is determined by comparing the calculated amplitude parameter with a standard curve which has been obtained in advance which provides the relationship between concentration and amplitude of the fluorescence transient waveform. The invention is illustrated further by the following exemplification.

Exemplification

Preparation of dansylated BSA (bovine serum albumin) BSA (6.9 mg) was dissolved in 1.0 ml of sodium borate buffer (0.1 M, pH 9.0) and the resulting solution was stirred. To this solution, dansyl chloride (2.8 mg) in 50 ul dimethyl formamide (DMF) was added dropwise over a period of 3 minutes at room temperature. The resulting solution was stirred for an additional 18 hours at room tem¬ perature. The reaction mixture was centrifuged to remove any particulate, and the resulting super¬ natant was applied onto a Sephadex G-25 gel fil¬ tration column (1.6 cm x 45 cm) which was previously equilibrated in phosphate buffered saline (PBS; 10 mM sodium phosphate buffer, pH 7.4, 0.15 M NaCl) and the same buffer was used for elution. The protein elution profile was monitored by a 280 nm UV de¬ tector, the protein conjugate fraction was collected (the conjugate was in the first eluted peak) and the dansyl-BSA conjugate (stock solution; 564 ug/ l) was stored at 4°C until use.

Preparation of human serum albumin (HSA) pyrene butyrate conjugate

HSA (6.9 mg) was dissolved in 1 ml of 0.1 M sodium borate buffer, pH 9.0, and the solution was kept stirred at room temperature. To this solution, pyrene butyrate succinimide ester (3.2 mg) in 50 ul of DMF was added dropwise over a period of 3 min¬ utes. The resulting solution was stirred for an additional 18 hours at room temperature, and the reaction mixture was centrifuged to remove any par¬ ticulate. The supernatant was applied onto a Seph¬ adex G-25 gel filtration column chromatography (same column size and elution buffer as above) to remove any unconjugated pyrene butyrate ester. The protein elution profile was monitored using a 280 nm UV detector and the protein-pyrene butyrate conjugate (stock solution 290 ug/ l) was collected and stored at 4°C until use.

Characterization of dansyl-BSA and pyrene butyrate- HSA conjugates.

Protein concentration of the conjugates was determined by the Lowry technique. The concentra¬ tions of the fluorophores was determined by spec- trophotometric methods (See Chen, R. , Anal. Biochem. Vol. 25, pp 412-416, 1968 and Rawitch, A.B., et al., J. Biol. Chem. Vol. 244, pp 6543-6547, 1969), and the dye/protein (D/P) ratios were calculated. The D/P ratios were between 3.5 and 52.5 for the dansyl- BSA conjugate and between 3 and 71 for the pyrene butyrate-HSA conjugate. The fluorescence excitation and emission spectra of the conjugates were measured. The excitation maximum was at 347 nm and the emission maximum was at 515 nm for the dansyl-BSA conjugate regardless of the D/P ratios. For the pyrene butyrate-HSA conjugate, the exci¬ tation maximum was at 347 nm with a secondary peak at 337 nm, and the emission maximum was at 376 nm or at 465 nm with a secondary peak at 395 nm depending on the D/P ratios. The lifetime of the fluorescence decay of the conjugates was also measured. The lifetimes were between 15 ns and 20 ns for the dansyl-BSA conjugates and between 70 and 100 ns for the pyrene butyrate-HSA conjugates depending on the D/P ratios of the conjugates. The larger the D/P ratio was, the shorter was the life time. Dansyl- BSA conjugates chosen had a D/P of 48 and a lifetime of 16 ns; pyrene butyrate-HSA conjugate had a D/P of 41 and a lifetime of 84 ns. Antibody binding activity was measured on Ouchterlony radial immunodiffusion plate (0.6% agarose in PBS containing 0.01% sodium azide) in which the conjugate was loaded into one well and its corres¬ ponding antibody into another well. The plate was incubated at room temperature for 4 hours. Fluores- cence immuno-precipitin bands were observed under a short wave UV hand lamp. The anti-BSA dansyl-BSA immunoprecipitin band was bluish and the anti-HSA pyrene butyrate HSA-immunoprecipitin band was greenish. Im unoassay procedures

Single analyte fluorescence immunoassay

Dansyl-BSA conjugate (100 ul of a forty-fold dilution of the stock solution) or pyrene butyrate- HSA conjugate (100 ul of four hundred-fold dilution of the stock solution) was incubated with 100 ul of its respective antibody (produced in rabbit) at (1/125 dilution anti-HSA; 1/75 dilution anti-BSA, Cooper Biomedical Co. PA.) in the absence of un- labeled BSA or HSA, or in the presence of varying amount of unlabeled BSA (0.1, 0.25, 0.5, 0.75, 1.0, 2.5, and 5.0 ug in 100 ul) or unlabeled HSA (0.1, 0.25, 0.5, 0.75, and 1.0 ug in 100 ul) , respec¬ tively. To the reaction mixture, 100 ul normal rabbit serum (4% in PBS) was added. The incubation was performed in PBS containing 0.1% gelatin and 0.1% sodium azide at 37°C for 60 minutes in a shaking water bath. 100 ul each of goat anti- rabbit IgG (20% in PBS) and polyethylene glycol (12.5% in water) were added and incubated at 37"C for an additional 30 minutes in a shaking water bath. At the end of the incubation, PBS (2 ml) containing 0.1% sodium azide was added, then the incubation mixture was centrifuged for ten minutes and the pellet collected. To the pellet, 200 ul each of 0.IN NaOH, 0.1 N HC1 and PBS were added in sequence to resuspend and stabilize the conjugate. For conventional fluorescence immunoassay, the fluorescence intensity of the suspension was mea- sured by spectrofluorometer (SLM 8000 C) and the dose response curve was constructed by plotting the concentration of the unlabeled BSA or unlabeled HSA against the fluorescence intensity (Figures 4 and

5) . For time-resolved fluorescence immunoassay, deconvolution analysis of the fluorescence decay curve of the suspension was performed and the dose response curve was constructed by plotting the concentration of unlabeled BSA or unlabeled HSA against the amplitude from the convolution curve (Figures 6 and 7) .

N. Multi-analyte time-resolved fluorescence immunoassay A mixture of dansyl BSA and pyrene butyrate HSA conjugates (100 ul ) was incubated with 100 ul of a mixture of anti-BSA (1/75) antibody and anti-HSA antibody (1/125) in the absence of unlabeled BSA and unlabeled HSA, or in the presence of 100 ul of a mixture of varying quantity but fixed ratio of unlabeled BSA and unlabeled HSA (0.25, 0.5, 0.75, 1.0, 2.5, and 5.0 ug of each serum albumin) in PBS buffer containing 0.1% gelatin, 0.1% sodium azide and 1% normal rabbit serum in a total volume of 400 ul at 37°C for 60 minutes. Thereafter, 100 ul of goat anti-rabbit IgG (20% in PBS) and 100 ul of polyethylene glycol (12.5% in H O) were added to the reaction mixture and incubated at 37°C for an additional 30 minutes. The reaction mixture was then centrifuged and the pellet was collected. To the pellet, 200 ul each of 1.0 N NaOH, 0.1 N HC1 and PBS were added in sequence to resuspend and sta¬ bilize the suspension. Fluorescence was measured with an Horiba time-resolved spectrofluorometer. A two component deconvolution analysis of the fluores¬ cence decay curve of the suspension was performed.- The dose response curve was constructed by plotting the concentration of unlabeled BSA or unlabeled HSA against the respective amplitude derived from the deconvolution analysis (figure 8) .

0. Equivalents

This completes the description of the preferred embodiments of the invention. Modifications may be made without departing from the scope of the invention, which should not be limited, except as provided in the following claims. For example, while the invention has been described in terms of time varying voltage amplitude signals produced by a PMT detector, other optical detectors, such as photodiodes and multichannel plate devices, may be substituted therefore. Also, rather than displaying the voltage varying waveform from the photodetector as a two-dimensional image on a CRT, the output of the photodetector can be used to modulate the intensity of the electron beam in the CRT to encode the waveform into a one-dimensional linear array detector.

Claims

1. A method for fluorescent spectroscopy of material wherein the material may be naturally fluorescent or tagged with a known fluorescent substance, comprising the steps of: a) exciting the material with a single pulse of light energy to induce fluorescence; b) separately detecting in a detector and generating: (i) an electrical signal corresponding to the fluorescence transient waveform F(t) induced by said single pulse, as distorted by said detector and said pulse, and (ϋ) a electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector, c) separately displaying an image of said waveforms. F(t) and E(t) ; d) digitizing a predetermined number of data points on each such image, as digital numbers representing points on said waveforms; e) storing said numbers in memory as data points of T(t) and F(t) ; and f) calculating the true impulse response fluorescence waveform f(t) from the stored data point numbers.

2. The method of Claim 1 wherein the true impulse response fluorescence waveform is calculated by convoluting E(t) with a predetermined trial function F(t) , having adjustable parameters and comparing the data points of the convoluted

F(t) cai,c with data points corresponding to F(t) data points.

3. The method of Claim 2 wherein the predetermined trial function curve F(t) ca₌,ic is established by physico-chemical considerations of the known characteristics of the constituents in the" material.

4. The method of Claim 2 wherein the parameters of the trial function curve F(t)' cai,c are iteratively adjusted until a best fit is achieved between the trial function curve F(t) Ccl,_LC and the stored curve F(t) .

5. The method of Claim 4 wherein a best fit occurs when the variance between F(t) cai,c and F(t) is minimized when successive values of the chi-square function thereof differ by less than a predetermined amount.

6. The method of Claim 2 wherein the material comprises a single fluorescent species, the trial function curve is an exponential curve in the form of Aexp(-t/T) wherein the parameter A is the amplitude and T is the fluorescent decay lifetime.

7. The method of Claim 2 wherein the material comprises a plural number of fluorescent species and the trial function is the sum of a plural number, ith, of exponential curves A₁exp(-t/T₂) + A₂exp(-t/T₂)

^Aith^{exp("t T}ith⁾ -

8. The method of Claim 6 wherein the material is tagged with a known fluorescent specie and T is therefore known in advance and the parameter A is proportional to the concentration of the fluorescent species in the material.

9. The method of Claim 6 wherein the material is tagged with at least two known fluorescent specie, the lifetimes of which are known in advance and the amplitude parameter is proportional to the concentration of each fluorescent specie in the material.

10. The method of Claims 8 or 9 wherein the absolute concentration of the fluorescent specie is determined by comparing the calculated amplitude parameter with a curve stored in computer memory; said curve having been obtained in advance by analyzing a series of samples of known concentration and generating a table of the amplitude parameter A versus actual concentration.

11. The method of Claim 1 wherein the images displayed in step (c) are intensified and then stored in an image storage device for subsequent digitizing.

12. The method of Claim 11 wherein the detector comprises a photomultiplier tube and the waveform is in the form of an amplitude varying voltage provided at the output of the photomultiplier tube.

13. The method of Claim 12 wherein the voltage is coupled to a cathode ray tube for display as an image on the face thereof and said image is intensified and stored by a charge-coupled device.

14. Apparatus for fluorescent spectroscopy of material wherein the material may be naturally fluorescent or tagged with a known fluorescent substance, comprising: a) a stable source of light pulses for exciting the material with a single pulse of light energy to induce fluorescence; b) detector means for separately detecting and generating: (i) an electrical signal corresponding to the fluorescence transient waveform F(t) induced by said single pulse, as distorted by said detector means and said pulse, and

(ii) an electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector means, c) image display means responsive to said electrical signals for separately displaying an image of said waveforms F(t) and E(t) ; d) digitizing means for producing a predetermined number of data points on each such image, as digital numbers representing points on said waveforms; e) storage means for storing said numbers in memory as data points of E(t) and F(t) ; and f) computer means for calculating the true impulse response fluorescence waveform f(t) from the stored data points.

15. The apparatus of Claim 14 including means for calculating the true impulse by convoluting E(t) with a predetermined trial function F(t) , having adjustable parameters and comparing the data points of the convoluted F(t) cai-,c with data points corresponding to the

F(t) data points.

16. The apparatus of Claim 15 including means for adjusting the parameters of the trial function curve F(t)' cai,c until a best fit is achieved between the trial function curve F(t) , and ^; caic the stored curve F(t) .

17. The apparatus of Claim 15 wherein the material comprises a single fluorescent species, the trial function curve is an exponential curve in the form of Aexp(-t/T) wherein the parameter A is the amplitude and T is the fluorescent decay lifetime.

18. The apparatus of Claim 15 wherein the material comprises a plural number of fluorescent species and the trial function is the sum of a plural number, ith, of exponential curves A₁exp(-t/T₂) + A₂exp(-t/T₂) — A_ithexp(-t/T_ith) .

19. The apparatus of Claim 17 wherein the material is tagged with a known fluorescent specie and T is therefore known in advance and the parameter A is proportional to the concentration of the fluorescent species in the material.

20. The apparatus of Claim 15 wherein the material is tagged with at least two known fluorescent specie, the lifetimes of which are known in advance and the amplitude parameter is proportional to the concentration of each fluorescent specie in the material.

21. The apparatus of Claim 14 wherein the light source comprises a pulsed laser, the detector means comprises a photomultiplier tube, the electrical signal is a voltage amplitude waveform, the image display means comprises a cathode ray tube coupled to said photomultiplier tube for displaying said voltage waveform on the face of the cathode ray tube, and image intensifier means for intensifying the image on said face and an image storage device coupled to said intensifier means for storing said image as charges on an array of charge-coupled devices.

22. The apparatus of Claim 14 including beam splitter means for sampling a portion of the induced fluorescence transient waveform to generate control signal to maintain a constant light intensity input to the detector means.

23. The apparatus of Claim 22 wherein the detector means comprises a photomultiplier tube, the light pulses comprise pulses of laser energy which induce fluorescence photons, said photons being split by said beam splitter into a main beam and a sampling beam, said sampling beam being detected to generate a control signal which is used to control the position of a gradient density filter in the path of the main beam to maintain constant light intensity to the input of said photomultiplier tube.

24. A fluorescence immunoassay for multiple ana¬ lytes, comprising the steps of: a. forming an incubation mixture of: i) antibodies against each analyte; ii) a predetermined amount of a fluores¬ cently labeled analytes wherein each fluorescently labeled analyte has a different fluorescence lifetime; and iϋ) a sample to be tested; b. incubating the mixture under conditions and for a period of time sufficient for the antibody and analytes to complex; c. determining the amount of each fluores¬ cently labeled analyte bound with antibody as an indication of the amount of each corresponding analyte in the sample, by determining the amplitude of each fluores¬ cence .decay curve for the antibody-bound fluorescently labeled analyte as an indication of the amount of each fluores¬ cently bound analyte bound.

25. A method of claim 24 wherein the antibodies are immobilized on a solid phase.

26. A method of claim 24 wherein the solid phase is separated from the mixture after incubation.

27. A method of claim 24 wherein a pair of analytes is determined, the pair being selected from the group consisting of TSH/Free T , LH/FSH, Vitamin B12/Folate, B-HCG/AFP, CEA/PAP, CK/CK-MB, LDH-_j^/LD^, HIV/HBAgs, HBAgs/HBAgc, and Herpes/Chlamydia;

28. A method of Claim 24, wherein the amplitude of the fluorescence decay curve for each analyte is determined by:

(a) exciting the antibody-bound fluorescently labeled analyte with a single pulse of light energy to induce fluorescence; (b) separately detecting in a detector and generating:

(i) an electrical signal corresponding to the fluorescence transient waveform F(t) induced by the single pulse, as distorted by said detector and said pulse, and (ii) an electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector,

(c) separately displaying an image of the waveforms F(t) and E(t) ;

(d) digitizing a predetermined number of data points on each such image, as digital numbers representing points on the wave¬ forms;

(e) storing said numbers in memory as data points of T(t) and F(t) ; and (f) calculating the true impulse response fluorescence waveform F(t) from the stored data point numbers by convoluting E(t) with a predetermined trial function

F(t) cai,c having adjustable parameters and comparing the data points of the convo¬ luted F(t) Cαl,C with data points cor- responding to F(t) data points, wherein the trial function is the sum of a plural number, ith, of exponential curves A₁exp(-t/T₂) + A₂exp(-t/T₂) A.. _hexp(-t/T.., ) , where i is the number of analytes, T is the fluorescence lifetime, and the parameter A is proportional to the concentration of each antibody-bound fluorescently labeled analyte.

29. A fluorescence immunometric assay for multiple analytes, comprising the steps of: a. forming an incubation mixture of: i) antibodies against each of the analytes in amount in excess of the analytes; ii) a sample to be tested; and iii) fluorescent antibody conjugates which bind specifically to each of the complexes, each conjugate having a different fluorescence lifetime; b. incubating the mixture under conditions and for a period of time which allows the antibodies and analytes in the sample to form immunocomplexes; and c. determining the amplitude of the fluores¬ cence decay of each conjugate associated with a complex as an indication of the amount of corresponding analyte in the sample.

30. A method of Claim 29, wherein the incubation mixture is formed first of antibodies against each of the analytes and the sample to be tested and then, the fluorescent antibody conjugates which bind specifically to each of the complexes are added.

31. A method of Claim 29, wherein the incubation mixture is formed first of the fluorescent antibody conjugates and the sample to be tested and then the antibody against each of the analytes is added.

32. A method of Claim 29, wherein a pair of analytes is determined, the pair being selected from the group consisting of TSH/Free T , LH/FSH, Vitamin B12/Folate, B-HCG/AFP, CEA/PAP, CK/CK-MB, LDH₁/LDH₂, HIV/HBAgs, HBAgs/HBAgc, and Herpes/Chlamydia;

33. A method of Claim 29, wherein the amplitude of the fluorescende decay curve for each analytes is determined by:

(a) exciting the antibody-bound fluorescently labeled analyte with a single pulse of light energy to induce fluorescence;

(b) separately detecting in a detector and generating:

(i) an electrical signal corresponding to the fluorescence transient waveform

F(t) induced by the single pulse, as distorted by said detector and said pulse, and (ii) an electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector,

(c) separately displaying an image of the waveforms F(t) and E(t) ;

(d) digitizing a predetermined number of data points on -each such image, as digital numbers representing points on the wave¬ forms;

(e) storing said numbers in memory as data points of T(t) and F(t) ; and (f) calculating the true impulse response fluorescence waveform F(t) from the stored data point numbers by convoluting E(t) with a predetermined trial function F(t) -. having adjustable parameters and comparing the data points of the convo¬ luted F(t) , with data points cor¬ responding to F(t) data points, wherein the trial function is the sum of a plural number, ith, of exponential curves A-_jexpf-t/ .) + ₂exp(-t/T₂)

A.. ,exp(-t/T.. , ) , where i is the number of analytes, T is the fluorescence lifetime, and the parameter A is proportional to the concentration of each antibody-bound fluorescently labeled analyte.

34. A fluorescence immunoassay for multiple ana¬ lytes, comprising the steps of: a. providing an immunoadsorbent comprising a solid phase having affixed thereto capture antibodies specific for each analyte to be tested; b. incubating the immunoadsorbent with a sample to be tested for a period of time and under conditions sufficient for analytes in the sample to complex with the capture antibodies; c. separating the immunoadsorbent and the sample; d. incubating the immunoadsorbent and a solution containing fluorescently labeled antibodies specific for each analyte, each labeled antibody having a different fluorescence lifetime; e. separating the immunoadsorbent and the solution; f. determining the amplitude of the fluores¬ cence decay curve of each conjugate associated with the immunoadsorbent as an indication of the amount of each analyte in the sample.

35. An assay of Claim 34, wherein the solid phase is a bead, microwell, or test tube.

36. A method of Claim 34, wherein a pair of ana¬ lytes is determined, the pair being selected from the group consisting of TSH/Free T , LH/FSH, Vitamin B12/Folate, B-HCG/AFP, CEA/PAP, CK/CK-MB, LDH_1Λ HIV/HBAgs, HBAgs/HBAgc, and Herpes/Chlamydia.

37. A method of Claim 34, wherein, the amplitude is determined by:

(a) exciting the fluorophores associated with the immunoadsorbent with a single pulse of light energy to induce fluores¬ cence;

(b) separately detecting in a detector and generating: (i) an electrical signal corresponding to the fluorescence transient waveform F(t) induced by said single pulse, as distorted by said detector and said pulse, and (ϋ) a.n electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector,

(c) separately displaying an image of said waveforms F(t) and E(t) ;

(d) digitizing a predetermined number of data points on each such image, as digital numbers representing points on said waveforms; (e) storing said numbers in memory as data points of T(t) and F(t) ; and

(f) calculating the true impulse response fluorescence waveform F(t) from the stored data point numbers by convoluting E(t) with a predetermined trial function

F(t)_ca,c„ having adjustable parameters and comparing the data points of the convo¬ luted F(t) , with data points cor- responding to F(t) data points, wherein the trial function is the sum of a plural number, ith, of exponential curves

A exp(-t/T_) + A_exp(-t/TH2Y)

A.. , exp(-t/T. ,, ) , where i is the number of analytes, T is the fluorescence lifetime, and the parameter A is proportional to the concentration of each antibody-bound fluorescently labeled analyte.

38. A fluorescence immunoassay for multiple an- alytes, comprising the steps of: a. providing an immunoadsorbent comprising a solid phase having affixed thereto capture antibodies specific for each analyte to be tested; b. incubating the immunoadsorbent with a sample to be tested for a period of time and under conditions sufficient for analytes in the sample to complex with the capture antibodies; c. separating the immunoadsorbent and the sample; d. incubating the immunoadsorbent with a solution containing secondary antibodies against each of the analytes, the incu¬ bation being carried out under conditions and for a period of time which permits the second antibodies to complex with analyte bound to the immunoadsorbent, each of the second antibodies are antigenically distinct; e. separating the immunoadsorbent and the solution; f. incubating the immunoadsorbent with a solution containing fluorescent antibody conjugates against the second antibodies, the conjugates comprising an antibody against each second antibody labelled with a fluorophore, each conjugate having a different fluorescence lifetime; g. separating the immunoadsorbent and the solution; and h. determining the amplitude of the fluores¬ cence decay of each conjugate associated with the immunoadsorbent as an indication of the amount of each corresponding analyte in the sample.

39. A method of Claim 38, wherein the solid phase is a bead, microwell or test tube.

40. A method of Claim 38, wherein the capture antibodies are monoclonal antibodies, the second antibodies are polyclonal antibodies, and the conjugates comprise fluorescently labeled polyclonal antibodies.

41. A method of Claim 38, wherein the conjugates have fluorescent lifetimes which differ from each other by greater than about 20 nanosec.

42. A method of Claim 38 wherein the amplitude is determined by:

(a) exciting fluorescently labeled antibody associated with the immunoadsorbent with a single pulse of light energy to induce fluorescence; (b) separately detecting in a detector and generating:

(i) an electrical signal corresponding to the fluorescence transient waveform F(t) induced by said single pulse, as distorted by said detector and said pulse, and- (ii) an electrical signal corresponding to the waveform E(t) of said single pulse, as distorted by the impulse response of said detector,

(c) separately displaying an image of said waveforms F(t) and E(t) ;

(d) digitizing a predetermined number of data points on each such image, as digital numbers representing points on said waveforms;

(e) storing said numbers in memory as data points of T(t) and F(t) ; and

(f) calculating the true impulse response fluorescence waveform F(t) from the stored data point numbers by convoluting E(t) with a predetermined trial function ^• F(t) , having adjustable parameters and comparing the data points of the convo¬ luted F(t) -i with data points cor¬ responding to F(t) data points, wherein the trial function is the sum of a plural number, ith, of exponential curves

A₁exp(-t/T₂) + A_exp(-t/TH2Y)

A.. , exp(-t/T. , , ) , where i is the number of analytes, T is the fluorescence lifetime, and the parameter A is proportional to the concentration of each antibody-bound fluorescently labeled analyte.

43. A fluorescence immunoassay for a single analyte wherein the amount of fluorescent probe em¬ ployed to quantify the analyte is determined by determining the amplitude of the fluorescence decay curve of the probe, wherein the fluores¬ cence decay curve of the probe is distinguished from that of the background fluorescence on the basis of fluorescence lifetime.

44. A flourescence immunoassay for determination of multiple analytes wherein the amount of each fluorescent probe employed to quantify each analyte is determined by determining the amplitude of the fluorescence decay curve of each probe, wherein the fluorescence decay curve of each probe is distinguished from that of the other probes and the background fluorescence on the basis of fluorescence lifetime.