Time-Resolvedfluorometerfor Lanthanidechelates-A New Generationof Nonisotopicimmunoassays Erkkl Solni and Hannu Kojola

CLIN.CHEM.29/1, 65-68(1983)

Time-ResolvedFluorometerfor LanthanideChelates-A New Generationof NonisotopicImmunoassays Erkkl SolnI and Hannu Kojola

Pulsed-light time-resolved fluorometry of lanthanide chelates ing and fluorescence (3-5). Similar systems have also been has proved to be very sensitive for use with nonisotopic used to study delayed fluorescence and are useful for remov- immunoassays. We describe a manually operated fluorome- ing scattered light and therefore for separating normal and ter with a conventional xenon flashtube. Sensitivity for 1-s delayed fluorescence (6). determinations is similar to that of radioisotopic methods. Another application of time-resolving techniques has

been the measurement of the rotation of proteins and the Downloaded from https://academic.oup.com/clinchem/article/29/1/65/5667462 by guest on 27 September 2021 Chelates of europium and terbium are a potential alterna- degree of flexibility of their active sites. This can be achieved tive to radioisotopic compounds as labels in immunoassays. by measuring fluorescence polarization as a function of time The main advantages of using chelates as fluorescent probes after a light pulse of 1 ns. In time-resolved fluorescence are the high quantum yield, exceptionally large Stoke’s spectra all the different components of the emissions that shift, narrow emission peaks, and optimal emission and occur during the counting period can be distinguished by excitation wavelengths for use with biological material. their different decay times from the prompt emission compo- These characteristics make the lanthanide chelates prefera- nent, which occurs during the delay period and which ble to any conventional fluorescent probe for use with consists mainly of scattered light (7-10). Equipment for ordinary fluorometers. In this paper we discuss the design time-resolving fluorescence spectroscopy and decay-time and operation of a suitable fluorometer; the preparation and determination is available from many manufacturers, one of use of the lanthanide chelates will be presented elsewhere. the earliest suppliers being Ortec (11). When used with biological material, conventional fluores- Brown et al. (12) used a variable nanosecond gated cent probes suffer from serious limitations to sensitivity, photomultiplier technique for separating weak, prompt fluo- owing to the natural fluorescence from various compounds rescence from the total fluorescent and phosphorescent in biochemical samples such as blood serum. This interfer- emission. Lytle and Kelsey (13) reported a time-resolved ence can, to some extent, be reduced by the careful selection system for measuring the fluorescence spectrum without of optical filters in the fluorometer, but ordinarily fluoro- interference from scattering and Raman lines, and empha- metric methods are not much more sensitive than radioiso- sized its usefulness with macromolecular samples where topic methods if conventional fluorescent probes are used. scattered radiation is a severe problem. The concept of using With lanthanide chelates, however, it is possible to reduce time-resolving fluorometry in the field of fluoroimmunoas- the background level significantly by selective detection of say was examined in this laboratory during 1974 and 1975, long-decay fluorescence. The fluorescence decay time of and some conference reports resulted (14-16). Parallel stud- lanthanide chelates is often in the order of 10-1000 ps, ies have been carried out by Wieder (17). Time-resolving whereas the decay time of natural fluorescence in a typical fluorometry has also been reviewed by Soini and Heminila biological sample is in the order of 1-20 ns. For this reason (18). pulsed-light source, time-resolved fluorometers used with Principles and Instrumentation lanthanide chelates are potentially several orders of magni- tude more sensitive than conventional fluorometers. We have developed a simple, manually operated fluoro- The use of time-gating to distinguish between fluores- meter for fluoroimmunoassays with lanthanide chelates as cence and Rayleigh and Raman scattering in a sample of fluorescent probes. The particular unit shown in the block biological fluid has been mentioned in several articles. diagram in Figure 1 was designed for research purposes Loudon (1) suggested the possibility of using time-gating to only. reduce the background fluorescence in Raman spectroseopy, The sample compartment is covered by a light-tight lid and Van Duyne et al. (2) designed a practical system and the sample is changed manually. The samples are held involving pulsed laser excitation for temporal resolution in small disposable tubes or cuvettes made of polystyrene, between the short-lived Raman signal and the relatively which has a reasonably low long-decay background fluores- long-lived fluorescence signal. They included an electronic cence. Because the intensity of the single flashes from the time-gate in the photomultiplier tube circuit and synchro- xenon flashtube was not very reproducible, we had to ensure nized it with the pulsed laser source in such a way as to stabilization of the excitation system. An integrator (P1) for ensure that signals caused by Raman photons were prefer- a semiconductor photodiode serves as the stabilizer of the entially recorded. flash lamp. The flash lamp is activated about i0 times at a Many papers have also been published on pulsed-source frequency of 1 kHz. The exact number of flashes (N) is time-resolved phosphorimetry. In this technique, xenon- controlled by the integrator P1 so that the integrated discharge lamps are used as a pulsed-light source and, after intensity of the photon emission is thus fixed. For the a short delay following each pulse, a photomultiplier detec- stabilization detector we used a photodiode (Model UV- tor measures the luminescence. This method has been used 215B; EG & G Inc., Electro-optics Div., 35 Congress St., effectively to eliminate interfering background light scatter- Salem, MA 01970), operated in the photovoltaic mode and connected to the optical system by a fiber light guide. The integrator is made of an operational amplifier, which pro- Wallac Oy, Research and Development Department, P.O. Box 10, vides a control signal for the flashtube circuit. The integrat- SF - 20101 Turku 10, Finland. ed photon emission from the flashtube is stabilized by this Received June 21, 1982; accepted Sept. 8, 1982. method with a precision of ± (1/N) 100% assuming that the

CLINICALCHEMISTRY, Vol. 29, No. 1, 1983 65 >- I- zU) 0 ‘U I- u) 2 w ‘U 0 0 2 2 U.’ 4 0 U) 0 ‘U

U) 0

4 -J U- Downloaded from https://academic.oup.com/clinchem/article/29/1/65/5667462 by guest on 27 September 2021

WAVELENGTH (nm) excitation and emission -.-.- absorption- Fig.2.Typicalabsorption,excitation,andemissionspectra ofeuropium chelates Fig. 1. The functional block diagram flash trigger pulse to the start of scaling was typically t1 = deviation of the intensity of single flashes is not greater 400 .ts and the fast scaler gate open-cycle was 500 ps. In this than ±50%. arrangement, single-photon counting occurred for one-half This stabilization method has many advantages. First of of the total measuring time and the fraction (g2) of the all, the system is simple; the flashtube and its power supply photon emission decay spectrum registered for each flash can be made without any stabilization circuit and less- was expensive flashtubes with lower stability can be used. The 2 _O.693t temperature dependence of the system can be minimized by g2= ef7dt (1) a single compensator element. The flashtube is operated only during a measurement, thus ensuring a long practical 1=01,2 t1 I life. The eventual fatigue of the flashtube will be automati- where decay time T1,2 = 560 s foreuropium and excitation cally compensated by the integrator. cycle time = 1 ms. The value of g2 = 0.40 includes second- The pulsed-light source used in this fluorometer was an and third-order decay fractions. As a consequence the maxi- FX-198 bulb-type xenon flashtube with a 1.5-mm arc cap mum number of pulses registered in the scaler was 5#{149}106 (EG & G Inc.). An EG & G Lite-Pac Trigger Module counts during a 1-s measurement. produced the high-voltage trigger pulses required to operate the flashtube. We operated the flashtube system at + 600 V Sensitivity Evaluations and a flash duration of 0.5 jzs. The rate of emitted photons (specific activity) obtained To provide optimal excitation and emission bands, we used interference band-pass filters (Ferroperm AS, Copen- from a fluorescent sample of Eu (13-NTA)3T0P02.3 is hagen, Denmark) mounted inside the sample compartment nE = (2) for easy and quick replacement. where

The detector is a side-window photomultiplier tube (Mod- = molar absorptivity = 150 000 L mol cm el R928; Hamamatsu TV Co. Ltd., 1126 Ichino-cho, Hama- d = path length of cuvette = 1 cm matsu, Japan) operated with negative-bias voltage, thus n0 = photon intensity of the flash in the position of the obtaining a direct analog signal between the anode and sample cuvette = 7.12W 1011 photons/flash ground. We found this to be a practical arrangement for f1 = repetition rate of the flash = 1000 flashes/s monitoring the total amount of fluorescence and obtaining g1 = factor dependent on the transmission and reflections an indication of counter saturation. of the cuvette =0.74. The photomultiplier tube, operated in the single-photon Q = quantum yield of the fluorescent probe = 0.06 (at the mode, is connected to a fast preamplifier and discriminator 613 nm emission peak of europium) and to a fast scaler having a digital display of seven decades. k = concentration of the sample (molIL) The counting speed of random events is limited to 40 MHz by the preamplifier and single-photon discriminator. The numerical value of nE is 1.1 1019. k photons . s

Results The signal strength of the fluorometer is given by the This time-resolved fluorometer was designed primarily formula for use with fluorescent chelates of europium. Typical excitation and emission spectra of europium chelate is shown in n = Figure 2. The photomultiplier sensitivity for the emission where peaks of this chelate depends on the cathode material used. g2 = fraction of the photon emission decay spectrum regis- For the tube used in this device the photocathode quantum tered for each flash = 0.40 efficiency q for europium was 8% at 613 nm. The emission g3 = fraction of pulses passed through the single-photon filters used were chosen for the longer-wavelength peak. pulse discriminator = 0.20 The time-resolving operation was controlled by pulses g4 = transmission of the sample solution, cuvette, and from the clock-pulse generator. The delay time from the optical parts at 613 nm, including reflections = 0.67

66 CLINICAL CHEMISTRY, Vol. 29, No. 1, 1983 g5 = geometrical photon-collection efficiency = 0.28 means that the sensitivity of the present device is sufficient g6 = transmission of the secondary filter = 0.25 for ordinary immunoassays. However, it is not the specific q = quantum efficiency of the photomultiplier tube = 0.08 activity of the label alone that determines the lowest level of at 613 nm detection of an immunoassay; the immunological system is also an important factor. The main advantages of the

This equation gives a numerical value for n = fluorometric assay are, of couse, the short counting time and

3.38’ iO’5 . k counts which is in good agreement with the use of nonradioactive material. The statistical precision the test results with the fluorometer. can easily be improved by increasing the excitation intensi- A simplified theoretical comparison between 1251 and ty 10-fold, which increases both the signal and background lanthanide chelates may illustrate the potential of the latter rate by the same factor, but the statistical effect is a minor as a very sensitive labeling material. Given insulin (Mr = one. The excitation level can be increased significantly 5800) as a model compound, let us suppose that the insulin without any risk of saturating the probe, because in the is labeled with europium chelate in the molar ratio of 1:1, present system only about 0.01 of the fluorescent probe is which has been proved possible without any analytical side excited. The active volume of the sample is about 200 L, effects. The typical sample volume would be 1 mL; an which represents only 20% of the sample volume. By insulin sample of 1 pg/mL corresponds to 1.72 i0’ mol/L decreasing and the sample volume from 1 mL to 0.2 mL Downloaded from https://academic.oup.com/clinchem/article/29/1/65/5667462 by guest on 27 September 2021 concentration of both insulin and fluorescent probe. The optimizing the optical conditions for smaller samples, the number of emitted photons (specific activity) is then E = sensitivity of the fluorometer, in terms of picomoles per 1.9’ 106 photons/s per picogram. The counting rate obtained milliliter, could thus be improved fivefold. with the fluorometer would be n = 580 cps for the 1 pg/mL Lanthanide chelates and time-resolved fluorescence de- insulin sample. The background rate of the fluorometer, tection provide potentially very high detection sensitivity mainly caused by the afterglow of the xenon flash tube, is for the following reasons: 100 cps for distilled water and isabout the same for a serum #{149}Signallphoton emission (“specific activity”) can be in- blank. The count-to-background ratio in this model example creased by a stronger excitation would thus be 5.8. (For the precision achieved at various #{149}Background fluorescence from sample “blank” can be counting times, see Table 1.) discriminated by using time-resolved detection A typical maximum specific activity of ‘I label is in the #{149}Lanthanide concentrations in biological samples are nor- order of 50 Ci/g (1.85 MBq/j.tg) in the case of insulin. This mally negligible specific activity corresponds to 0.1-0.2 active labels per #{149}Lanthanide labels are biochemically inert (no interaction molecule and gives a count rate of 80 cpmlpg (1 pg 108 with the sample) molecules) at a gamma counting efficiency of 70%. A typical background rate in gamma counting is 40 cpm. Thus the count-to-background ratio is 1.94, which is of the same order We are grateful to Mr. T. Oikariand Mr. R. Hadu forvaluable as when using the fluorometric method but, to attain the advice and laboratory experiments. same precision as the latter, the counting time would be 100 times longer. Table 1 details differences in the performance of fluorometry and gamma counting. References 1. Loudon R. The Raman effect in crystals. Adv Phvs 13, 423-428 Discussion (1964). Jeanmaire The test results and the theoretical example above show 2. Van Duyne RP, DL, Shriver DF. Mode-locked laser Raman spectroscopy-a new technique for the rejection of interfer- clearly the great potential of fluorometry for shortening ing background luminescence signals. Anal Chem 46, 213-222 counting time and for increasing sensitivity. The sensitivity (1974). of the fluorometer and fluorescent probe in this work is 3. Fisher RP, Winefordner JD. Pulsed source time resolved phos- similar to that obtained by using 1251 as a label, which phorimetry, Anal Chem 44, 948-956 (1972). 4. SkleznevAG. Stroboscopicapparatusforluminescencestudies, Prib Tekhn Eksperim No 2,212-216 (1971). 5. Winefordner JD. Time-resolved phosphorimetry. Accounts Chem Table 1. Performance of Time-Resolved Res 2, 361-569 (1969). Fluorometry and Gamma Counting for 1 mL of 6. Kikuchi K, Kokubun H, Koizumi M. Studies on delayed fluorescence by means of a flash technique. Bull Chem Soc J 41, 1545- Labeled Insulin 1551 (1968). Symbol Fluorometry Gamma counting 7. Schuyler R, Isenberg I. A monophoton fluorometer with energy 1.85.106 Typical specific n 1.89.1012 discrimination. Rev Sci Instr 42, 813-817 (1971). activity photons/s perg dps perg 8. Loken MR, Gohlke JR, Brand L. Nanosecond time-resolved Count rate (net), n 580 1.3 fluorescence spectroscopy in molecular biology. In Fluorescence cps/pg Techniques in Cell Biology, Thaer AA, and Sernetz M, Eds., Background rate. B 100 0.67 Springer-Verlag, Berlin, pp 339-344 (1973). cps 9. Yguerabide J. Nanosecond fluorescencespectroscopy of macro- Ratio n/B 5.8 1.94 molecules. Methods in Enzymol 26, 498-578 (1972). Precision obtained 10. Ware WR. Transient luminescence measurements. In Creation for 1 pg/mL and Detection of the Excited State 1A, AA Lamola, Ed.,Marcel concn NY, 1971,pp 213-302. 1-scounting s 4.5% Dekker, New York, time 11. Loken MR. Hayes JW, Gohlke JR, Brand L. Excited-state 1-mm counting s 0.6% 14% protontransferas a biologicalprobe.Determinationof ratecon- time stants by means of nanosecond fluorometry. Biochem 11,4779- 4786,(1972). 12. Brown RE, Legg KD, Wolf MW, SingerLA. Gated nanosecond time resolved emission spectroscopy separation of mixed emissions n from carbonyl compounds. Anal Chem 46, 1690-1694 (1974). 13. Lytle FE, Kelsey MS. Cavity-dumped argon-ion laser as an

CLINICAL CHEMISTRY,Vol. 29, No. 1, 1983 67 excitation source in time-resolved fluorimetry. Anal Chem 46, 855- 16. Soini E. New analytical methods with fluorescent probes in 860 (1974). cytology. Paper given at Cell and Tissue Culture Symposium, 14. Soini E. The use of fluorescing antibodies in immunological organized by the Turku University Medical Faculty, Turku, Oct. measurements. Paper given at Kemian Pflivht, the congress ar- 24, 1974. In Finnish. ranged by the Association of Finnish Chemical Societies, Helsinki, 17. Wieder I. Background rejection in fluorescence immunoassay. Nov. 21, 1975. In Finnish. In Immunofluorescence and Related Staining Techniques, (W 15. Soini E. New technological developments in quantitative fluo- Knapp, K Holubar, G Wick, Eds., Elsevier/North-Holland Biomedi- rescence microscopy. Paper given at Symposium in Genetics, orga- cal Press, Amsterdam, 67-80 (1978). nized by Anatomics Fenmcae and The Finnish Histochemical 18. Soini E, Hemmil#{228}I. Fluoroimmunoassay: Present status and Society,Turku, Sept.5, 1975.In Finnish. key problems. Clin Chem 25,353-361 (1979). Downloaded from https://academic.oup.com/clinchem/article/29/1/65/5667462 by guest on 27 September 2021

68 CLINICAL CHEMISTRY, Vol. 29, No. 1, 1983