CLIN.CHEM.38/5, 725-730 (1992)

CLINICALCHEMISTRY,Vol. 38, No. 5, 1992 725

Direct Time-Resolved Fluorescence Immunoassay of Progesterone in SerumInvolving the Biotin-Streptavidin System and the Immobilized-AntibodyApproachS. E. Kakabakos”2 and M. J. Khosravi2’3

We developed a direct competitive-type immunoassay forprogesteronein serum that combines the advantages ofthe biotin-streptavidinsystemwiththe antibody-immobi-lization approach. We synthesized biotinylatedprogester-one derivatives of five different proteins and, after initialevaluation of the conjugates, selected biotinylated bovinelgG-progesterone as the most suitable tracer. Progester-one released from binding proteins with danazol com-petes with the biotinylated tracer conjugate for binding toa limitedamount of a mouse anti-progesterone monoclo-nal antibody in microtitration wells coated with a goatanti-mouse lgG antibody. The binding of the biotinylatedtracer is then monitored by reaction with a streptavidin-based universal detection reagent developed for time-resolved fluorometry. The assay demonstrated typicalperformance characteristics with respect to the dynamicrange, detection limit, and precision. Recovery averaged99.3% (SD 8.3%) and dilution experiments showed goodlinearity. Measurements correlated well with those fromthree commercially available direct immunoassays forprogesterone.

Additlenal Keyphrases:fluoroimmunoassay radioimmunoas-say intermethod comparison

Determination of progesterone in serum, in conjunc-tion with estradiol concentrations, has been used tomonitor ovulation in both health and disease (infertili-ty) states and as an aid in the differential diagnosis ofearly pregnancy (1-3). Conventional methods requirean extraction procedure followed by radiorinmunoassay(RIA) of the extract (4, 5). Because of the inherentdisadvantages of the extraction step, methods for directanalysis for progesterone involving isotopic (5, 6) andnomsotopic (7-9) detection principles have been devel-oped.

The extraordinarily great binding affinity and speci-ficity of the biotin-avidin or biotin-streptavidin [a pro-tein from the Streptomyces avidinii (10)] interaction arethe main reasons a biotin system isused in both immu-nological and noninununological detection methods.The principle and advantages of the technology forspecific applications were summarized in recent reviews(11-16). Competitive immunoassays incorporatingthebiotin-avidin interaction have been based on the anti-gen-immobilization approach with labeled (biotiny-

‘Department of Clinical Biochemistry, University of Toronto,100 College St., Toronto, Ontario, Canada M5G 1L5.

2CyberFluor Inc., 179 John St., Toronto, Ontario, Canada M5T1X4.3Addresscorrespondenceto this author at CyberFluor Inc.Received August 26, 1991; accepted March 4, 1992.

lated) antibodies and avidin or streptavidin as comple-mentary reagents (Figure IA) (17, 18).

We recently (19) described a novel application of thebiotin-streptavidin system for developing the more clas-sical competitive-type (labeled antigen) immunoassay.In this configuration, streptavidin is labeled with theanalyte of interest and used as the tracer conjugate incombination with biotinylated detection probes and sol-id-phase antibodies (Figure 1C). We demonstrated themerits and usefulness of this new approach in a modelassay for cortisol in serum.

Here we report on the application of a different assayconfiguration, proposed in an abstract (20), for compet-itive immunoassay of haptens; this also complementsthe advantages of the biotin-streptavidin system withthe immobilized-antibody approach but allows use of themost common universal detection reagents based onlabeled streptavidin. In this configuration, conjugates ofanalyte-carrier molecule-biotin serve as the tracer(Figure ]B). In the assay, the biotinylated progesteronederivatives compete with progesterone in standards orsamples for binding to a specific mouse monoclonalantibody in microtitration wells coated with a goatanti-mouse antibody. The amount of bound conjugate isthen quantified by reaction with a streptavidin-baseduniversal detection reagent developed for time-resolvedfluorometry. The developmental details and advantagesof this approach are discussed.

MaterIals and MethodsInstrumentation

Time-resolved fluorescence measurements were per-formed with a Model 615 Immunoanalyzer (CyberFluor

A + + s

“d PIIflI IS,IYIM. aI.p,avI“*0’

B A- #{247}-

...rI ptI.,.flb9.afl.*Od7 9.00..MIb.dy 000.aflt.

c + A- (<or5 +

t0.4 pAas #{149}If9.V#{149}diS. Its, 54119,I9..d,.00d st09so 5,05009 9,05.

ladSOdy Cnh.SoI.

Fig.1. BasIc configuration of competitive immunoassay invoMng mebiotin-streptavidinsystemIn A, immobilizedantigen used with biotinylatedantibodyand streptavidin-based detection probe; in B, immobilizedantibody used with blotinylatedconjugate;in C, immobilizedantibodyused with labeledstreptavidinas tracerand biotinylated detection probe

c__ ritfo

726 CLINICALCHEMISTRY,Vol.38, No. 5, 1992

Inc., Toronto, Canada). The instrument has automaticdata reduction. Radioactivity counting was performedwith an LKB Model 1275 Minigamma counter (LKBWallac, Turku, Finland). Absorbance was measuredwith an HP Model 8450A diode array spectrophotome-ter (Hewlett-Packard Canada, Mississauga, Ontario).The final measurement of the dissociation-enhancedlanthanide fluoroimmunoassay (DELFIA#{174})progesteroneassay was performed with the LKB 1230 Arcus fluorom-eter.4

Materials

Chemicals. Progesterone, progesterone-3-carboxyme-thyloxime (P3CMO), danazol, 1,1’-carbonyldiimidazole(CDI), protease (proteinase K, EC 3.4.21.14; type XX-VIII), bovine serum albumin (BSA; RIA grade; CohnFraction V), bovine IgG, horse spleen apoferritin, bovinethyroglobulin, and ovalbumin were from Sigma Chem-ical Co. (St. Louis, MO 63178). Sulfosuccinimidyl 6-(bi-otinamido)hexanoate (NHS-LC-biotin) was from PierceChemical Co. (Rockford, IL 61105). Europium(ffl) chlo-ride hexahydrate was from Aldrich Chemical Co. (Mil-waukee, WI 53233). White microtitration strips (Micro-lite#{174};12 wells per strip) are a product of Dynatech Labs.Inc. (Alexandria, VA 22314). Goat anti-mouse IgG an-tibody was purchased from Jackson ImniunosearchLabs. Inc. (West Grove, PA 19390). Sources of otherchemicals were previously described (19,21).

Buffers and standards. The assay buffer was 0.1 mol/Lphosphate buffer, pH 7.4, containing 2 mg of danazol, 9g of NaC1, 0.5 g of sodium azide, 5 g of BSA, and 0.5 g ofbovine globulin per liter. The conjugate buffer was 50mmol/L Tris buffer, pH 7.4, containing 9 g of NaC1, 0.5g of sodium aside, and 10 g of BSA per liter. The coating,blocking, and streptavidin-thyroglobulin-4,7-bis(chlo-rosulfophenyl)-1,10-phenanthroline-2,9-dicarboxylic acid(BCPDA) dilution buffers and wash solution were as de-scribed previously (19). Progesterone standards were pre-pared in progesterone-free serum (Scantibodies Lab., San-tee, CA 92071).

Monoclonal antibody. The anti-progesterone mousemonoclonal antibody was supplied by CyberFluor Inc.The antibody was raised against progesterone-li-hemisuccinate-BSA. The cross-reactivities of the anti-body in the assay with 17-hydroxyprogesterone, 5a-pregnan-3,20-dione, 5J3-pregnan-3,20-dione, deoxycorti-costerone, and androsterone were 0.9%, 8.3%, 7.6%,0.3%, and 0.03% respectively. With estradiol, estrone,estriol, testosterone, cortisol, and etiocholanolone, thecross-reactivity was <0.01%.

Comparison methods. For comparison we used twodirect (nonextraction) progesterone radioimmunoassaykits: the Coat-A-Count (Diagnostic Products Corp., LosAngeles, CA) and the Amerlex-M (Amershain Canada

4Nonstandard abbreviations: DELFIA, dissociation-enhancedlanthanide fluoroimmunoassay; P3CMO, progesterone-3-car-boxymethyloxime; CDI, 1,1’-carbonyldiimidazole; BSA, bovine se-rum albumin; NHS-LC-biotin, sulfosuccinixnidyl 6-(biotinamido)-hexanoate; and BCPDA, 4,7-bis(chlorosulfophenyl)-1,10-phenan-throline-2, 9-dicarboxylic acid.

Ltd., Toronto, Canada). We also compared the resultswith those obtained by the time-resolved DELFIA (Phar-macia/Wallac, Turku, Finland).

Human serum samples were obtained from TorontoWestern Hospital, Toronto, Canada M5T 2S8. Ly-phocheck#{174}immunoassay control sara (human) 1,11, andIII were from Bio-Rad (Clinical Division, Richmond, CA94801).

Procedures

Coating of microtitration wells. We coated microtitra-tion wells with the goat anti-mouse IgG antibody asdescribed previously (19), Before use, the wells werewashed twice with the wash solution.

Synthesis of the biotinylated P3CMO-carrier proteinconjugate. We prepared 126 mL of a 16 mmol/L solutionof P3CMO in dimethyl sulfoxide in a dry screw-cap tube.To activate the steroid derivative, we added 0.8 mL of a30 mmol/L solution of CDI in dimethyl sulfoxide, whichcorrespondsto 1.2 mol of CDI per mole of progesterone.The tube was purged with N2 and wrapped in foil. Afterthe reaction had proceeded for 15 mm at room temper-ature, we added the required amounts of the CDI-activated progesterone in a final volume of 0.5 mL ofdimethyl sulfoxide to 2 mg of various proteins that weredissolved in 2 mL of 0.2 mol/L Na2CO3 buffer, pH 9.1.The concentrations of the CDI-activated progesteroneused for labeling BSA, bovine IgG, bovine thyroglobu-lin, apoferritin, and ovalbumin were 4.9,3.0, 1.13, 2.02,and 2.22 pmol per tube, respectively. The labeling ratioused corresponded to 2.5 CDI-activated progesteronemolecules per amino group on each protein (Table 1).We allowed the labeling reaction to proceed for 2 h atroom temperature, then dialyzed the reaction mixtureextensively against 0.05 mol/L Na2CO3 buffer, pH 8.After dialysis, we filtered the contents of the dialysisbag through a 0.45-pm (pore size) filter (Gelman Sci-ences, Ann Arbor, MI 48106). We adjusted the volume ofeach conjugate to 3 mL by adding water and then added1 mL of a 1 mol/L Na2CO3 buffer, pH 9.1.

For biotinylation we added to the above preparationsthe required amounts of NHS-LC-biotin in 50 pL ofdimethyl sulfoxide to achieve a labeling ratio of 12NHS-LC-biotin molecules per amino group on each

Table 1. CharacterizatiOn of BlotlnylatedProgesterone

Recovery,Carrier protein miss, lcD. Pssrone BiOtlnb %C

Bovine IgG 160 1951d 0.75 44 86.3Ovalbumin 45(20) 0.20 11 96.7Thyroglobulln 660(150] 0.68 86 84.9BSA 66(59) 0.40 25 93.4Apoferrltin 480(194] 0.85 78 88.5

Molar ratio; determinedby RIA.b Molarratio;assessedby 2-(4’-hydroxyazobenzene)benzolc acid method

(22).C Assessed by the Bradford method for protein determination(23).#{176}No.of NH2groups in brackets.

700000

200000

100000

CLINICAL CHEMISTRY, Vol.38, No. 5, 1992 727

protein. The reaction was allowed to proceed for 2 h atroom temperature, followed by extensive dialysis of themixture against 0.05 molIL NaHCO3 buffer, pH 8,containing 9 g of sodium chloride and 0.5 g of sodiumazide per liter. After dialysis, we removed the contentsof the dialysis bag and stored the conjugate stock solu-tions at 4#{176}Cuntil use. Under this condition the conju-gates are stable for �12 months.

Characterization of the conjugate. Protein quantifica-tion was performed by Bradford’s method (23). We usedthe difference in protein concentration before and afterlabeling and dialysis to calculate percent recovery. Theextent of progesterone labeling was determined by RIA(Coat-A-Count) after complete digestion of a sample ofthe various conjugates with proteinase K (9).

We determined the number of biotin molecules cou-pled to each protein molecule by the 2-(4-hydroxy-azobenzene) benzoic acid method (22), using the Gibco(Burlington, Canada) BRL protein biotinylation kit.

Preparation of the detection reagent (streptavidin-thyroglobu1in-BcPDA-Eu). The synthesis and charac-terization of the streptavidin-based macromolecularcomplex labeled with the europium chelator BCPDA weredetailed elsewhere (24,25). The detection reagent wasdiluted 50-fold into the streptavidin-thyroglobulin-BCPDA dilution buffer to yield a working solution of 4.4nmol/L of streptavidin.

Assay Protocol

Conjugate-binding assay. We added 50-pL portions ofthe mouse anti-progesterone antibody (40 pg/L, 2 ng/wellin conjugate buffer) and 50 pL of increasing concentra-tions of each of the conjugates (0.1-4 mg/L in conjugatebuffer) into microtiter wells coated with the goat anti-

mouse IgG antibody. After shaking the wells for 1 h atroom temperature, we washed them four times with thewash solution, added 100 pL of the diluted detectionreagent per well, and shook the wells for 30 mm. Wewashed the wells again and dried them as described (19).The fluorescence of the solid-phase complex (goat anti-mouse IgG antibody/mouse monoclonal anti-progester-one antibody/progesterone-carrier molecule-biotin/strep-tavidin-thyroglobulin-BcPDA-Eu31 was then mea-sured with the CyberFluor Immunoanalyzer. In theseexperiments the amount of the anti-progesterone anti-body added per well was substantially below the maxi-mum binding capacity of the solid-phase second anti-body (-100 ng/well). The latter was determined fromthe antibody titration experiments in the presence of aconstant amount of conjugate (biotinylated bovine IgG-.progesterone) under conditions similar to those above.

Progesterone assay. We added 10 pL of standards orserum samples, 50 pL of the biotinylated bovine IgG-progesterone conjugate working solution (140 pg/L, 7ng/well in assay buffer), and 50 pL of the mouse anti-progesterone monoclonal antibody (80 pg/L, 4 ng/well inassay buffer) into the wells coated with goat anti-mouseIgG, in duplicate. After incubating the wells with stir-ring at room temperature for 1 h, we washed them fourtimes with the wash solution, added 100 pL of the

diluted detection reagent per well, and incubated thewells for an additional 30 mm. The fluorescence of thefinal complex on the dried solid phase was then mea-sured as above. The same procedure was used to evalu-ate the performance characteristics of the remainingconjugates.

ResultsSynthesisand Characterizationof BiotinylatedProgesteroneDerivatives

We prepared biotinylated conjugates of five differentproteins with progesterone under identical conditionsand, after determining the approximate conjugationratios (Table 1), tested increasing amounts of eachconjugate for maximum binding to a limitedamount ofa monoclonal anti-progesterone antibody. As shown inFigure 2, all conjugates were capable of effective bind-ing to both the anti-progesterone antibody and thestreptavidin-based detection reagent. In all cases, therelative rate of change in binding signal followed aninitial rapid increase before reaching a plateau.

We then tested the conjugates for competition withprogesterone in a competitive binding assay involving amouse monoclonal antibody and a solid-phase-boundgoat anti-mouse IgG antibody.As shown in Figure 3, allconjugates competed well with progesterone, and theresulting calibration curves were substantially parallelthroughout the assay range. Under optimal assay con-ditions developed for the various conjugates, analysis ofseveral serum samples showed close agreement betweenthe expected and measured values (data not shown). Weselected the biotinylated bovine IgG-progesterone de-rivative for furtherinvestigation.

Assay OptimizationWe optimized the assay for the amount of antibody

and conjugate.For a constantvolume of sample (10 pL

C

I-

0

UCa,U

a,0

Conjugate (mgJL)Fig. 2. AntIbody saturation curvesThe mean fluorescence signal of triplicate measurements are plotted vsconjugate concentration: U, bovine lgG; 0, ovalbumin; S, thyroglobulin; A,BSA; 0, apoferritln

106

141 10 100 1000

Dilution factor

728 CLINICALCHEMISTRY,Vol.38, No.5, 1992

Progesterone (nmol/L)

FIg.3. CompetItive binding characteristics of the various biotinytatedprotein-progesterone conjugatesThe optimumamounts of conjugate used were 0.93,1.91.3.6,1.77, and 7.08nmol/L forbovine lgG, BSA, ovalbumin, thyroglobulin, and apoferritin,respec-tIvely; symbols as in Fig. 2

per well) and antibody (80 .ig/L), increasing the conju-gate concentration from 20 to 700 pg/L caused a propor-tional increase in the B0 value (fluorescence of the zerostandard), followed by a plateau at higher conjugateconcentrations. Similarly, for a constant amount ofconjugate (140 1zg/L), the B0 value increased in responseto increasing antibody concentrations from 20 to 400pg/L and then reached a plateau for concentrations asgreat as 2000 pg/L. The standard curve characteristics(detection limit, slope, and dynamic range) were optimalat conjugate and antibody concentrations of 40-160 and40-80 gfL, respectively. The nonspecific binding of thedetection reagent without either the antibody or theconjugate was <0.3% of the B0 value.

We also investigated the effect of incubation time onthe immunoreaction step and selected a 60-mn dura-tion. By this time the maximum fluorescence signalgenerated had reached a plateau.

Analytical Variables

Calibration curve and detection limit. A typical stan-dard curve for the assay with the biotinylated bovineIgG-progesterone as the tracer is shown in Figure 3.The detection limit, defined as the analyte concentra-tion corresponding to the mean fluorescence reading ofthe zero standard - 2 SDs, was 0.2 nmolJL. The preci-sion profile of the assay derived from 12 replicate mea-surements of each standard point gave CVs between2.1% and 7.2% over a concentration range of 1.6-127.2nmol/L (data not shown).

Precision. We determined the within-run precision byreplicate analysis of three control serum samples in asingle assay and between-run precision by duplicatemeasurement of the control samples in 20 differentruns. At mean progesterone concentrations of 2.7-65nmol/L, the within-run CV was between 3.3% and 9.6%

and the between-run CV was between 4.6% and 8.6%.Dilution test. We evaluated dilution linearity of the

assay by assaying samples serially diluted with proges-terone-free human serum. There was good agreementbetween the measured and expected values (Table 2),which were derived from the initial concentrations ofprogesterone in undiluted samples.

Analytical recoveiy. The recovery of known concentra-tions of exogenous progesterone (3.2-64 nmol/L) addedto portions of three serum samples ranged from 82% to110% (mean 99.3%). The recovery was assessed byanalyzing the samples before and after adding proges-terone and then subtracting the estimated concentra-tions of endogenous progesterone.

Correlation with other methods. Comparison of resultsobtained by the present method and by three coxnmer-cially available RIA and FIA methods showed goodagreement. The linear-regression equations are summa-rized in Table 3.

Discussion

The assay presented here combines the advantages ofthe biotin-streptavidin system with the immobilized-antibody approach. In this configuration, biotinylatedprogesterone derivatives serve as both the tracer andthe bridge linking captured conjugates to the streptavi-din-based detection probe. The use of a high-molecular-mass carriermolecule is advantageous because, as inlabeled (biotinylated) antibody assays (17, 18), amplifi-cation can be introduced by multiple-site biotinylationof the conjugate.

The basic design of the competitive immunoassaymediated by the biotin-atreptavidin interaction isshown in Figure 1. Although the conventional design(Figure IA) has the advantages of developmental andoperational simplicity, assay optimization is entirelydependent on the amount of antibody because that is theonly variable component. In addition, with directlyimmobilized antigens, it may be difficult to reproduciblycontrol coating consistency, so the precision and sensi-tivity of the assay may be somewhat compromised.

In the alternative strategy (Figure 1B), antibody canbe directly immobilized to the solid phase or used inconjunction with a solid phase pretreated with an anti-

Table 2. DIlution Linearity of Samples with LargeProgesterone Concentration (nmol/L)

SerumSample UndIluted 2 3 4 5 16

No. 1

No. 2

ExpectedObserved

-

72.336.233.4

24.123.6

18.117.3

9.09.1

4.54.8

ExpectedObserved

-

129.064.560.2

43.044.8

32.232.0

16.115.3

8.17.4

ExpectedObserved

-

82.041.037.3

27.326.9

20.520.1

10.311.2

5.15.5

No. 3

References1. Abdulla U, Dever MJ, Hipkin LI, Davis JC. Plasma progester-one levelsas an index of ovulation. Br J Obetet Gynaecol1983;90:543-8.2. Landgren BM, Unden AL, Diczfalusy E. Hormonalprofileof thecycle in 68 normally menstruating women. Acta Endocrinol1980;94:89-98.3. Radwanska E, Frankenberg J, Allen El. Plasma progesteronelevels in normal and abnormal early human pregnancy. FertilSteril 1978;30:398-402.4. Cameron EDH, Scarisbrick JJ. Radioimxnunoassay of plasmaprogesterone. Clin Chem 1973;19:1403-8.5. Rateliife WA, Corner JET, Daiziel AR, Macpherson JS. Direct‘I-radioligand assay for serum progesterone compared with as-says involving extraction of serum. Clin Chem 1982;28:1314-8.6. Blight LF, White GH. lmIlabeled radioimmunoassay kits forprogesterone evaluated foruse in an in vitro fertilization program.Clin Chem 1983;29.1024-7.7. Sjoblom P, Wildand M, Hahlin M, Nilsson L, Lindblom B.Evaluation of a time-resolved fluoroimmunoassay for analysis ofsex steroids in serum. Hum Reprod 1990;5:396-401.8. Miller SA, Morton MS, Trukes A. Chemiluminescence immu-noassay for progesterone in plasmaincorporating acridinium esterlabeled antigen. Ann Clin Biochem 1988;25:27-34.9. Elder PA, Yeo KHJ, Lewis JG, CliffordJK. An enzyme-linkedimmunosorbent assay (EUSA) for plasma progesterone: immobi-lized antigen approach. Clin Chim Acta 1987;162:199-206.10. Tijssen P. Non-immunologic molecular recognition systemsused in immunoassays. In: Burdon RH, van Knippenberg PH, eds.Practice and theory of enzyme immunoassays. Amsterdam: Else-vier Science Publishers, 1985:21-4.11. Bayer EA, Wilchek M. The use of the avidin-biotin complexasa tool in molecular biology [Review). Methods Biochem Anal1980;26:1-45.12. Guesdon JL, Ternynch T, Avrameas S. The use of avidin-biotin interaction in immunoenzymatic techniques. J HistochemCytochem 1979;27:131-9.13. Wilchek M, Bayer EA. The avidin-biotin complex in bioana-lytical applications [Review]. Anal Biochem 1988;171:1-32.14. Wilchek M, Bayer EA, eds. Methodsof enzymology, Vol. 184.New York:Academic Press, 1990.15. Hantowich DJ, Virzi F, Rusekowski M. Investigations of

CLINICALCHEMISTRY,Vol.38, No. 5, 1992 729

Table 3. LInear-RegressIon Correlation with Resultsfrom Comparison Assays

ComparatIve aasaya Slope Intercept r n

Coat-A-Count, AlA 1.01 -0.50 0.98 159Amerlex-M, AlA 1.14 -0.28 0.96 70DELFIA,FIA 1.03 -0.27 0.92 70

a Comparative assay = y = present time-resolvedFIA.

species second antibody as described here. The latterhas the advantage of improved reproducibility (26,27)and allows assay optimization by changing the concen-tration of both the tracer and the antibody. Further-more, a solid phase coated with an appropriate secondantibody may be used as a common binder for a varietyof antigen-specific first antibodies.

Here we demonstrated that biotinylated progesteronederivatives of a variety of proteins (Table 1) can be usedas tracer. The relative fluorescence signal obtained forthe various conjugates was bovine IgG > ovalbumin>thyroglobulin > BSA> apoferritin (Figure 2). At thesaturation point, the corresponding maximum bindingsignals were approximately 595, 307, 298, 245, and 159kilo-fluorescence units, respectively. Because we did notobserve any obvious relation between signal and vari-ables such as molecular mass and the extent of biotiny-lation, we postulate that steric factors affecting theproportional binding of the various conjugates to theantibody and (or) the detection reagent may be respon-sible for the wide variation in signal.

In competitive antibody-binding assays, all of theconjugates competed well with native progesterone, andthe resulting competition curves were nearly superim-posable (Figure 3). In all cases, the B0 values weresufficiently high and the standard curves generatedwere typical for progesterone assays (5-9). We could notcompletely explain the substantial variation in theamount of different conjugates required for optimumperformance by the variation in progesterone-conjuga-tion ratio of the various proteins (Table 1). Other fac-tors, e.g., possible differences in the exact number ofantibody accessible progesterone molecules per conju-gate molecule, may be involved.

We selected the biotinylated bovine IgG-progesteronederivative because it gave the highest signal, had anintermediate molecular mass, and required the leastamount of conjugate per well. Using this conjugate astracer, we were able to develop a direct immunoassay ofprogesterone in serum that showed performance char-acteristics similar to or better than those reported forthe currently available assays (5-9). Results by thepresent method correlated well with those obtained withtwo RIAs and one FIA (Table 3).

A recent report described labeling procedures andstructural requirements for directly labeling estradiolwith biotin for effective antibody binding and competi-tion in enzyme iminunoassays (28). We found standard-curve characteristics similar to those described herewith directly biotinylated progesterone molecules, ex-cept that the B0 value of the conjugate was about 8- to

10-fold lower than that of the biotinylated bovine IgG-progesterone derivative (data not shown). An additionaldisadvantage of direct labeling was the requirement forextensive conjugate purification by HPLC. However,conjugate preparation as describedhere had a consis-tently large yield, and after labeling, a simple dialysis ofthe reaction mixture was sufficient to ensure purity.Conjugates carrying as little as 0.1 mol of progesteroneper mole of protein could be used because we couldcompensate for the effect of a lower conjugation ratio byincreasing the amount of the conjugate per well. Thelatter did not adversely affect assay performance be-cause nonspecific binding did not increase by increasingconjugate concentration and was consistently <0.3% ofthe B0 value.

In conclusion, we have described the application of anew assay configuration that combines the advantagesof the biotin-avidin interaction and the immobilizedantibody approach with simple and economical prepar-

ative procedures. We propose that the present design iswell suited for competitive immunoassay of haptens.The conjugates are also suitable for assay designs rely-ing on biotin-.avidin interaction as a separation meansor for indirect coating of haptens on avidin-coat.ed solidsupports.

730 CLINICAL CHEMISTRY, Vol. 38, No. 5, 1992

avidin and biotin for imaging applications. J Nucl Med1987;28:1924-302.16. Diamandis EP, Christopoulos TK. The biotin-(strept)avidinsystem: principlesand applicationsin biotechnology [Review].ClinChem 1991;37:625-36.17. Diaxnandis EP, Bhayana V, Conway K, Reichstein E, Papa-nastasiou-Diamandi A. Time-resolved fluoroimmunoassay of cor-tisol in serum with a europium chelate as label. Clin Biochem1988;21:291-6.18. Papanaatasiou-Diamandi A, Conway K, Diamandis EP.Digoxin immunoassay with monoclonal and polyclonalantibodiesusing time-resolved fluorometry. J Pharm Sci 1989;78:617-21.19. Khosravi MJ, MortonRC. Novel application of streptavidin-haptenderivatives as protein-tracerconjugatein competitive typeimmunoassays involving biotinylated detection probes. Clin Chem1991;37:58-63.20. Papanastasiou-Diamandi A, Morton RC, Khosravi MJ, Dia-mandis EP. New assay configuration for competitive-type time-resolved fluoroimmunoassay of haptens [Abstract]. Clin Biochem1989;22:260-1.21. Khosravi MJ, Morton RC, Diamandis EP. Sensitive, rapidprocedure for time-resolved immunofluorometry of lutropin. ClinChem 1988;34:1640-4.

CLIN.CHEM. 38/5, 730-735 (1992)

22. Green NM. A spectrophotonietric assay for avidinand biotinbased on binding dyes by avidin. Biochem J 1965;94:23c.23. Bradford MM. A rapid and sensitive methodfor the quantita-tion of microgram quantities of protein utilizing the principle ofprotein-dye binding. Anal Biochem 1976;72:248-54.24. Diamandis EP, Morton EC, Reichstein E, Khosravi MJ. Mul-tiple fluorescence labeling with europiumchelators.Applicationtotime-resolved fluoroimmunoassays. Anal Chem 1988;61:48-53.25. Morton RC, Diamandis EP. Streptavidin-based macromolecu-lar complex labeled with a europiumchelatorsuitable for time-resolved fluorescence immuneassay applications. Anal Chem1990;62:1841-5.26. Kakabakos SE, Evangelatce GP, Ithakissios DS. Immunoad-sorption of IgG onto second antibody covalently attached to amino-Dylark beads for radioiminunoassays. Clin Chem 1990;36:497-500.27. Sarkolli GM, Stott RAW, Kricka LI. Improvement in theantibodybindingcharacteristies of microtiter wells by pretreat-ment with anti4gO Fc inununoglobulin. J Immunol Methods1987;104:191-4.28. Tiefenauer LX, Andrea RY. Biotinyl-estradiol derivatives inenzyme immunoassays: structural requirements for optimal anti-body binding. J Steroid Biechem1990;35:633-9.

Quantification of Islet Cell Antibodies by Microscope Photometry and AntibodyTitration ComparedGeorge M. Bright

Islet cell antibodies can predict eventual insulin-depen-dent diabetes mellitus. In the standard method used bythe Immunology of Diabetes Workshop (IDW), predictivepower depends on antibody titer as determined by testingmultiple serum dilutions. Here I report use of a micro-scope-based photometry system (MPS) to test whetherislet fluorescence intensities from undiluted sera canpredict IDW results. MPS testing of 120 undiluted IDWspecimens correctly identified 27 of 31 positive and 84 of89 negative specimens. The relationship between IDWconsensus values and MPS readings of undiluted sera is:IDW = 2.45 MPS - 1.44 (r = 0.67, P <0.001). Thus,corrected islet intensities from undiluted sera correlatedwell with lOW values. However, the ability of MPS topredict high IDW consensus values is limited, which mayreflect the variable antigenic content of pancreas speci-mens; moreover, background correction as currently ap-plied is incomplete: nonspecific immunofluorescence ap-pears to emanate from within the islet.

Addftlonal Keyphrases: autoimmune disease diabetesimmunofluorometric assay

period of time has obscured the initial sequence ofpathogenic events (1-5). For this reason, investigatorshave sought measurable quantities that temporallyrelate the latent endocrinopathy to the affected individ-ual. Thus, the development of autoantibodies to thecytoplasm of islet cells, to a 64-kDa islet cell protein,and to insulin may predate a clinical diagnosis of insu-ha-dependent diabetes mellitus (IDDM) (6-15). Adre-nal cortical antibodies may also predate adrenal insuf-ficiency (16). The ability of these temporal markers todocument individual cases of latent endocrine autoim-munity offers the possibility of a more complete descrip-tion of the molecular and clinical events that lead toendocrine cell destruction. Furthermore, each describedevent in the pathogenic sequence can become a focus forpotential preventive therapy.

IDDM is found in <1% of moat populations (17-23)and, although concordance among first-degree relativesis well known, most new cases arise de novo (19, 23).That IDDM is partly due to immune-mediated damage(1-5) is a testable hypothesis, the validation of whichwill ultimately come from preventive clinical trials

The diagnostic criteria for endocrine autoiminunediseases apparently are fulfilled only after a variable

Nemours Children’s Clinic, 807 Nira St., Jacksonville, FL32207.

Received August 7, 1991; accepted February 21, 1992.

1Nonstandard abbreviations: IDDM, insulin-dependent diabe-tes mellitus; IDW, Immunology of Diabetes Workshop; ICA, islet-cell antibodies; MPS, microscope photometry system; Fl, fluores-cence intensity; JDF, Juvenile DiabetesFoundation; CON, theIDW consensus endpoint dilution results; END, the endpointdilution results from the author’s laboratory; PBS, phosphate-buffered saline; and FITC, fluorescein isothiocyanate.