CLIN. CHEM. 36/3, 509-514 (1990)

CLINICAL CHEMISTRY, Vol. 36, No. 3, 1990 509

Rapid Solid-Phase Immunoassay for 6-Keto Prostaglandin Fia Ofl MicroplatesWilifried Schramm,1 RIchard H. SmIth,’ ThomasM. Jackson,2PaulA. Craig,1 Harry E. Grates,’ and Laura L. Mlnton’

We describe, for the measurement of 6-keto prostaglandinF, a in biological media, a solid-phase immunoassay withimmobilized antibodies that requires a total processing timeof <2 h with hands-on time <30 mm for 40 samples. Themethod combines the convenience of the microplate formatwith the sensitivity of radiolabeled prostaglandin derivativesas tracers in a competitive immunoassay. The intra- andinterassay variations at 50% displacement of the radiola-beled prostaglandin derivative as tracer were 9.0% and11.8%, respectively. At 50% displacement of the radiola-beled tracer, the sensitivity is about 20 pg per well. Optimalincubation time is between 60 and 90 mm. Nonspecificbinding was <1% if about 8 pg of tracer (-25 000 counts/mmper well) was used. Inhibition curves of samples in differentdilutions were parallel to standard curves. The variation ofbound radiolabeled prostaglandin derivative within the wellsof one microplate (n = 96) was <3%. Human plasmasamples and medium from tissue culture assayed for 6-ketoprostaglandin F1a correlated well with results obtained with asolid-phase assay based on use of magnetic particles (r =

0.99, n = 24 for culture-medium samples; r = 0.99; n = 26for plasma samples).

AddItional Keyphrases:radioimmunoassay assay of tissueculture medium eicosanoids

Despite much progress in immunoassay methodology inrecent years, quantitative determination of prostaglandinsin biological media is still an elaborate, time-consumingprocedure. This is partly due to the necessity to extracteicosanoids from a large number of potentially interferingsubstances with similar structure (e.g., lipoproteins) inbiological samples. However, with the introduction andrefinement of extraction cartridges, separation of classes ofeicosanoids has become very convenient (1, 2). Anotherreason for long processing times for prostaglandins is thelack of reliable solid-phase immunoassays that can beperformed within a reasonable time period (not exceeding 2h). A solid-phase assay based on the use of magneticparticles and a lmIIabeled derivative of the prostaglandinas tracer is commercially available, but still requires over-night incubation (Amersham Corp., Arlington Heights, IL).Although more convenient than conventional charcoal-based separation methods involving tritiated prostaglan-dins with a lower quantum yield in signal generation astracer, the overall processing time still takes about 18 h. Tomeet the existing need for a fast, reliable procedure, wehave developed solid-phase immunoassays for 6-keto pros-taglandin Fia (6kPGFia), thromboxane B2, prostaglandinF2a, and prostaglandin Fia, using ‘251-labeled prostaglan-

1 BioQuant of Ann Arbor, Inc., 1919 Green Rd., Ann Arbor, MI48105.

2Michigan Diabetes Research and Training Center, The Uni-versity of Michigan Medical Center, 1331 E. Ann St., Box 580, AnnArbor, MI 48109.

Received October 3, 1989; accepted December 21, 1989.

din derivatives as tracers.3 Here we describe as an examplethe immunoassay for 6kPGFia and compare its resultswith those from the solid-phase, magnetic particle assayfrom Amersham.

Materials and Methods

Materials

Reagents. We purchased the following preparations andchemicals from the Sigma Chemical Co. (St. Louis, MO):mouse IgG (nonspecific), polyethylene glycol (PEG; Mr3350), ammonium sulfate, tyrosine methyl ester, Chlo-ramine T, and Freund’s complete adjuvant. From the Ald-rich Chemical Co., Inc. (Milwaukee, WI) we obtained chlo-roform, acetonitrile,methanol, ethanol, dioxane, anddicyclo-hexyl carbodiimide. Donkey anti-sheep IgG was fromAntibody Development, Inc., Ann Arbor, MI; 6kPGFia wasfrom Cayman Chemical, Arm Arbor, MI; keyhole limpethemocyanin (KLH) was from Bebring Diagnostics, LaJolla, CA; diethylaminoethyl (DEAE)-Sephadex A-50 wasfrom Pharmacia Chemical Inc., Piscataway, NJ; pepsinwas from Worthington Biochemical Corp., Freehold, NJ;and from NEN Research Products, DuPont Co., Wilming-ton, DE, we obtained carrier-free Na’251 and [3Hlthrom-boxane B2 (containing an average of eight tritium atomsper molecule, -200 kCilmol).

Phosphate buffers consisted of sodium phosphate, 0.05mol/L, pH 7.4; phosphate-buffered saline (PBS) contained 9g of NaCl per liter of phosphate buffer, and gel-PBSconsisted of PBS containing gelatin, 1 g/L.

Prostaglandin-KLFI conjugate. To induce immune re-sponse in sheep, we prepared a conjugate Of 6kPGFia withKLH. We activated the prostaglandin with equimolaramounts of dicyclohexyl carbodiimide in dioxane for 30 mmat room temperature. This reaction mixture was added to asolution of KLH in water/dioxane (equivolume mixture),with the activated prostaglandin in a 1000-fold molarexcess (based on an average molecular mass of 4 MDa forKLH). We removed precipitated dicyclohexyl urea by fil-tration, and excess prostaglandin derivative was extractedwith chloroform. Remaining traces of chloroform in theaqueous phase were evaporated by bubbling heliumthrough the solution.

Preparation and radiolabeling of 6-keto prostaglandinFia-tyrosine methyl ester. We synthesized 6-keto-pros-taglandin Fia-tyiosiiie methyl ester (6kPGFiaTME) byacylation of the tyrosine methyl ester with 6kPGFia indioxane; we used dicyclohexyl carbodiimide to activate theprostaglandin. The product was isolated by extraction intochloroform; the chloroform was removed by evaporationwith nitrogen, and the residue was stored at -20 #{176}Cinacetonitrile.

We iodinated 6k-PGF10-TME (700 ng) by incubation

3Nonstandard abbreviations: KLH, keyhole limpet hemocya-nm; DEAE, diethylaminoethyl; 6k-PGF1,,, 6-keto prostaglandinF1,,; PBS, 0.05 mol/L phosphate buffer, pH 7.4, containing NaCl, 9g/L; PEG, polyethylene glycol; and TME, tyrosine methyl ester.

510 CLINICAL CHEMISTRY, Vol. 36, No. 3, 1990

with 2 mL of Na’251 (2175 kCi/mol) and Chioramine T (20g) in 0.05 mL of NaHKPO4 (0.5 mollL, pH 7.4) for 1 mm.The reaction was stopped by mixing with 20 g of sodiummeta-bisulfite (3) for 1 mm. The mono-iodinated productwas separated from unreacted 125I and the di-iodinatedcompound by means of a dual-pump HPLC system (RaininInstrument Co., Inc., Woburn, MA) equipped with an Ul-trasphere IP column, 250 mm x 4.6 mm (i.d.), 5-ianparticle size (Beckman Instruments Inc., San Ramon, CA),which we pre-equilibrated with acetonitrile/potassium for-mate buffer (50 mmoL1L, pH 4.0), 20/80 (by vol), at ambienttemperature. Derivatives were separated by elution with alinear gradient of increasing acetonitrile concentration(increasing by 10 mL/L acetonitrile content per minute for60 mm). Elution of the labeled compounds was detectedwith a Model 170 radioisotope detector (Beckman Instru-ments, Inc., Irvine, CA), and radioactive fractions wereautomatically collected with a Model FC203 fraction col-lector (Gilson Medical Electronics, Inc., Middleton, WI).Only the fraction containing the monoi25Iiodinated 6k-PGFiaTME was used for the immunoassays. This fractionwas neutralized with 0.2 mL of NaKHPO4 (0.5 molJL, pH7.4), diluted to 20 j.CiJmL in absolute ethanol, and stored at4 #{176}C.For solid-phase assays, working solutions were pre-pared by diluting the ethanol solution to about 3 xio counts/mm in 1 mL of gel/PBS.

Antisera. We raised antisera against 6kPGF la by immu-nizing sheep with the prostaglandin conjugated to KLH.Each immunization involved 5-10 intramuscular injec-tions in each flank. For the primary (day 0) and secondary(day 31) immunizations, 6 mg of the KLH-conjugate (con-taining 267 pg of 6kPGFia) was emulsified in completeFreund’s adjuvant supplemented with additional Mycobac-terium tuberculosis (2.8 g/L of adjuvant; Difco Laboratories,Detroit, MI). We administered booster injections of theconjugate (in incomplete Freund’s adjuvant) at appropriateintervals, judged by the titer of the antiserum (below). Inthe animal whose serum we used for the data reported inthis study, the titer of the antiserum was stable for at leastsix months after the fifth immunization.

Aliquots of sheep antisera were tested for binding to‘251-labeled 6kPGFia in a double-antibody radioimmuno-

assay. We mixed 200 jL of antiserum in different dilutionswith 10 fmol of 125I labeled 6kPGFiaTME in 0.8 mL ofgel-PBS, and incubated the tubes overnight at 4#{176}C.Wethen warmed the tubes to room temperature (25 #{176}C)andadded 200 L of an appropriate dilution of donkey anti-sheep IgG prepared in gel-PBS containing PEG, 100 gfL.After 60 mm of mixing, we centrifuged the immune pre-cipitates, decanted the nonbound tracer, and counted theradioactivity of the precipitate in a y-counter (Searle 1290;TM Analytic, Elk Grove Village, IL). The serum used inthese studies bound about 30% of radiolabeled tracer at adilution of 1:30 000 to 1:50 000.

Standards. 6kPGFia was desiccated under reduced pres-sure in the presence of silica gel. Stock solutions (2 g/L) of6kPGFia in distilled ethanol were stored in sealed am-pules under argon at -70 #{176}C.The stock solution wasdiluted in gel-PBS by means of a Hamilton glass syringe(Hamilton Co. Inc., Whittier, CA) to give a 6kPGFiaconcentration of 800 p.g/L and subsequently used to prepareeight standards by serial dilutions: 0.0, 13,25,50, 100,200,400, and 800 ng/L (final concentrations). These standardswere dispensed into 1-mL polypropylene tubes (ResearchProducts International Corp., Mount Prospect, IL) ar-

ranged so as to match the format of the microplates andmultiple pipettors.

Samples

We measured the concentration Of 6kPGFia in two typesof samples: culture medium from cultured rat macrophagesand human blood plasma. We investigated the secretion of6kPGFia by rat alveolar macrophages after treating thecells with reactive oxygen metabolites (4), as follows. Alve-olar and peritoneal macrophages, collected by lavage fromrats, were incubated overnight in medium M199 and ex-posed to hydrogen peroxide, Ca2 ionophore (A23 187; Cal-biochem Corp., San Diego, CA), Zymosan, or exogenousarachidonic acid. We then centrifuged the cells, collectedthe supernatant medium, and stored it frozen at -20 #{176}Cuntil analyzed for 6kPGFia.

Plasma samples were obtained from subjects in a studyon the effects of undergoing strenuous exercise both beforeand after treatment with aspirin. Blood was collected inlavender-stoppered Monoject tubes (Sherwood Medical, St.Louis, MO) containing EDTA (7.2 mg per tube). We exam-ined the in vitro production of 6kPGFia in paired plasmasamples, one of which was collected as described above andthen incubated at 37 #{176}Cfor 1 h, the other one beingcollected into a chilled tube containing acetylsalicylic acidas prostaglandin inhibitor (0.21 mg per tube). These pairedsamples provided plasma containing low (with inhibitor)and high (no inhibitor) concentrations of 6kPGFia. Allblood samples were then centrifuged at 1800 x g and theplasma was removed and stored frozen at -20 #{176}Cuntilextraction.

Procedures

Purification of antibodies. We precipitated immunoglob-ulins by adding at 4 #{176}Can equal volume of 90% saturatedammonium sulfate, pH 7.4, to the serum. Recovered immu-noglobulin was taken up in phosphate buffer and precip-itated twice again. The protein was further purified onDEAE-Sephadex gel. We equilibrated the gel in Ths buffer(10 mmolJL, pH 8.0). To a column containing 20 mL ofgel-bed volume we applied 1 mL of a 5 g/L solution of partlypurified immunoglobulin. We eluted protein with phos-phate buffer containing various concentrations of sodiumchloride: 50, 100, 200, and 300 mmollL. The IgG fractionwas eluted with the buffer containing 200 mmolfL. Wedetermined the protein concentration by spectropho-tometry at 280 nm.

Preparation of F(ab)2 fragments. To obtain F(ab)2 frag-ments, we enzymatically cleaved the Fc region from IgG(10 g/L) with pepsin (0.2 g/L) in acetate buffer (0.01 molJL,pH 4.5) for 3 h at 37 #{176}C(5). After adjusting the acidity to pH7.5 with Tris buffer (2 mol/L), we removed the remainingintact antibody by fractionation over immobilized ProteinA (Pierce Chemical Co., Rockford, IL), according to themanufacturer’s specifications. The preparation was puri-fied by size-exclusion chromatography on polyacrylamidegel (P-60; Bio-Rad, Richmond, CA). We equilibrated the gelin borate buffer (0.1 molJL, pH 8.5) or phosphate buffer (pH6.0), depending on further use (see below) and eluted thefragments with the same buffer.

Preparation of donkey anti-sheep F(ab)2 fragments. Toobtain donkey IgG specifically binding the Fc region ofsheep IgG, we prepared a solid matrix with immobilizedsheep F(ab)2 fragments for affinity purification. Nonspecificsheep IgG (Sigma) was digested with pepsin as described

CLINICALCHEMISTRY, Vol. 36, No. 3, 1990 511

above. The F(ab)2 fragments were immobilized on 6%crosslinked agarose activated with carbonyldiimidazole(Pierce) in borate buffer. This gel [F(ab)2-gel} was used foraffinity purification of donkey anti-sheep IgG.

We purified donkey anti-sheep IgG by ammonium sulfateprecipitation and subsequent DEAE fractionation as de-scribed above. For further affinity purification on the ma-trix prepared as described above, we applied a 10 g/Lsolution of donkey IgG to a column containing 10 mL ofF(ab)2-gel in phosphate buffer. The first 4 mL of phosphatebuffer applied to the column was discarded and the next 10mL was collected for recovering antibodies specificallybinding to the Fc region of sheep IgG.

Affinity-purified donkey IgG was fragmented enzymi-cally with pepsin as described above, and the F(ab)2 frag-ments obtained were further reduced to Fab fragmentswith 2-mercaptoethylamine in phosphate buffer containing5 mmol of EDTA per liter, as described previously (6).

Immobilization of antibodies. We used a modification of apreviously described method (7) to immobilize the antibod-ies to 6kPGFia. We filled microwells consisting of 12 stripswith eight break-apart wells each (Costar, Cambridge, MA)with 200 pL of purified anti-sheep Fab fragments (0.1 mg/Lin phosphate buffer, 50 mmol/L, pH 7.4) and incubatedovernight at room temperature. After discarding the con-tents, we washed the wells three times with 300 L ofde-ionized water. Treated plates were dried in a vacuumdesiccator overnight in the presence of silica gel. Platescould be stored at 4#{176}Cfor as long as three months. Weincubated overnight at room temperature nonpurified an-tiserum to 6kPGFia, diluted 1:10 000 in PBS containing 1g of PEG per liter, 200 pL per well. Thereafter, the contentsof each well were discarded and the wells were washedthree times with 300 tL of de-ionized water. The plateswere dried as mentioned above and stored at 4#{176}Cuntilused.

Extraction of prostaglandins. We extracted prostaglan-dins from plasma and tissue-culture samples by means ofC18 Sep-Pak columns (Waters Associates, Milford, MA),using a modification of a method previously described (2). Avacuum manifold, adjusted to provide a flow rate of 10 mLper minute, was used to expedite batch processing ofsamples.

We pre-washed the columns with 5 mL of methanolfollowed by 5 mL of de-ionized water. To estimate recovery,we added [3H]thromboxane B2 (0.005 pCi, -10 pg) to 1 mLof each sample. This prostaglandin has the same elutionproffle as 6k-PGF1 but does not substantially cross-reactwith the antibody used in this assay. After applying thismixture to the column, we washed the column with 5 mL ofmethanol/de-ionized water (20/80 by vol), then with 5 mL ofde-ionized water. Prostaglandins were eluted with twosequential washes of 1 mL of methanollde-ionized water(80/20 by vol). We evaporated the solvent in the eluent fromthe last 2 mL, using a stream of air at ambient tempera-ture, and stored the residue at -20 #{176}Cuntil further analy-sis. Just before assay, the residue was reconstituted with 1mL of gel-PBS. To determine recovery of the prostaglan-dins, we removed 100 pL of reconstituted sample, dilutedthis with 1 mL of PBS and 4 mL of scintillation cocktail(“Ecolume”; ICN Radiochemicals, Irvine, CA), and countedfor beta emission.

Immunoassays. We measured the concentration of 6k-PGF1a in extracts from both human plasma and cell-culture medium with reagents as described above. Pre-

dispensed radioactive tracer and buffer were added with aneight-channel multipipettor (Flow Laboratories, Inc.,McLean, VA). One set of calibrators was dispensed in onestep with the eight-channel pipettor into the microplates.Each well contained a total volume of 200 L: eithercalibrator plus radiolabeled tracer, or 100 L of reconsti-tuted sample plus 100 pL of radioactive tracer (25 000counts/mm; -4 pg of the prostaglandin equivalent). Theplates were sealed with an adhesive cover and incubatedfor 60 min under agitation. Thereafter, the contents of thewells were discarded, the wells were washed three timeswith a stream of de-ionized water, broken apart, anddropped in 12 x 75 mm test tubes to be counted forradioactivity.

We compared the response curve for dilutions of mediaextracts with the response curve obtained from standardsolutions of 6k-PGF1, (parallelism). To determine the ac-curacy of the assay, we measured the prostaglandin con-centration after the addition of 6kPGFja (at 100, 200, and400 pg/mL) to extracts of cell-culture medium. Finally, wecorrelated the results of this new assay procedure (Bio-Quant) with results obtained with a commercial kit thatinvolves use of overnight incubation and a second antibodycoated onto magnetic polymer particles as matrix for sep-arating free from bound antigens (Amersham Corp., Ar-lington Heights, IL) as directed by the manufacturer. Theseexperiments were performed with culture medium andwith human plasma samples (correlation studies).

To analyze the data, we used the four-parameter logisticcurve-fit option of an MS-DOS immunoassay program(SASPRO; Arbor Immunalysis Services, Ann Arbor, MI) ona Z-158 PC computer (Zenith Data Systems, St. Joseph,MI). An estimate of the intra-assay precision over therange of each assay, i.e., the precision proffle, was devel-oped from the precision of replicate determinations inseveral batches of samples (8).

Results

Radiolabeling of the prostaglandin derivative. The tyro-sine methyl ester of 6k-PGF1 provides two derivativesupon iodination: the mono- and the di-iodinated com-pounds. Under the applied HPLC elution conditions, theseare easily separated from each other (Figure 1). Mono-iodmnated prostaglandin can be distinguished from thedi-iodinated product by using different concentrations of1251 as starting material: increasing the amount of iodineshifts the ratio of the two peaks in favor of the secondpeak.Therefore, we conclude that this peak constitutes the di-iodinated product. For this preparation, we used a 1.4-foldmolar excess of 6k-PGF1-TME over Na125I. The non-

1251

16.9ni

125

2.47_m

Fig. 1. Purification of iodinated 6k-PGF1,,-TMEderivative on HPLCand on-linedetectionof 1251

The peak eluted at 16.9 mm represents mono-iodinated, the peak at 18.8 mmdi-iodinatedcompound.Onlymono-iodinatedderivativewas usedfor immu-noassays

Fig. 2. Three-dimensional representation of the binding of 1251

labeled 6k-PGF1,.to antibody immobilized on the surface of micro-titerwellsCountsper minute of the bound radlolabeled tracer is plottedIn differentprections of arraysof 8 x 12(x- vs y.axls) polystyrenemicrowells.Bindingofantigenas an expressionof quantityof immunoglobulin perwell is shownonthe z-axis (counts/mm per well). The almostplanardistributionof boundradlolabeled tracerover the whole plate demonstratesthe high precisioninimmobilizingthe antibody throughoutthe arrayof96 wells

30.

0. -

40

.30_

.2O

10

.02.5

-0- Bo,T, %-ui- pg/well at 50% inhibition

0 j 1.5 2

Crossreactlvfty, %

1002.10.050.030.020.030.050.068.67.90.070.0018

512 CLINICAL CHEMISTRY, Vol. 36, No. 3, 1990

iodinated starting material (not shown) elutes about 2 mmbefore the mono-iodinated 6kPGFiaTME. Therefore, ahighly purified tracer can be isolated that contains essen-tially one iodine atom per molecule.

Crossreactivity. The crossreactivity of eicosanoids withthe antibody is shown in Table 1. We expressed the cross-reactivity as the percentage of the amount of eicosanoidthat displaced 50% of the radiolabeled 6k-PGF1,. Arachi-donic acid has only a very low crossreactivity with theantibody and therefore does not interfere with the quanti-fication of 6kPGFia. Actually, we had to use a saturatedsolution of arachidonic acid to determine its crossreac-tivity.

Uniformity of immobilization. Binding of radiolabeledantigen to the wells of one plate with immobilized antibodyresulted typically in a variation of between 2% and 3% of1251-labeled 6k-PGF1,,-TME, a value comparable with thatof liquid-phase immunoassays. The between-wells varia-tion of one plate is shown in Figure 2. Plate-to-platevariation was 5.6% within one batch for single strips takenfrom random (selected without conscious bias) positions ofeach of six plates. Provided the plates were prepared atconstant temperature and evaporation of liquid during theincubation periods was prevented, we observed no system-atic inconsistencies in the amount of antibodies bound tothe surface in relation to the position of the well in theplate, as has been described for other systems (9).

Kinetic studies. To determine the optimal time for incu-bating tracer with sample or calibrators, respectively, weterminated the incubation of standard curves after varioustimes (Figure 3). The percentage of tracer bound vs thetotal amount of tracer added to the wells over time in-creased asymptotically over time to reach equilibrium after60 mm. Likewise, the sensitivity at 50% inhibition ofradiolabeled 6k-PGF1,,-TME derivative increased overtime, almost in a mirror image of the previous curve. Forthis experiment we used 2 pmol of immobilized IgG perwell. The sensitivity (measured as the amount of unlabeledprostaglandin required to displace 50% of radiolabeledtracer) increases with decreasing amount of immobilizedantibody per well (data not shown); however, the samekinetic relationship as shown in Figure 3 applies.

Standard curves obtained after different incubationtimes were almost parallel. Figure 4 shows plots of twostandard curves obtained after 15 mm and 90 mm ofincubation. As mentioned above, the sensitivity increasesafter longer incubations but the difference in results be-

Table 1. Crossreactlvlty of Elcosanoids with anAntiserum Specifically Binding 6k-PGF,

Compound

6-Keto-PGF1PGF211-$-PGF2,13,1 4-Dihydro-15-keto-PGF213,1 4-Dihydro-6,15-diketo-PGF2,,PGD2PGE2PGE16-Keto-PGE1PGFI,.Thromboxane B2Arachidonicacid

IncubaLion time [h]Fig. 3. Bindingof radiolabeled tracerand sensitivityas a functionofincubationtimeThe binding of l-Iabeled prostaglandinderivative increasesfrom about 10%(binding B5/T; left ordinate)after 15-mm incubationtoabout30% bInding after90-mm incubation. Similarly, the sensitivity, expressed as 50% dIsplacementof radiolabeled tracerby unlabeled prostaglandin (right ordinate), Increasesfrom45 pg/well at 15 mm to 18 pg/well after 90 mm. This study wasperformedwith 2 pmol of ImmobilizedlgG perwell

tween 60 and 90 mm is negligible. Therefore, we selectedan incubation time of 60 mm for all subsequent assays.

Parallelism. Samples from tissue-culture medium M199assayed in four different dilutions provided inhibitioncurves parallel to the standard curve (Figure 5).

Accuracy. We added known amounts of 6k-PGF1,. toculture medium and assayed the samples to determine theaccuracy of the analytical system. Between 80% and 99% ofprostaglandin added in different concentrations was de-tected (Table 2); on the average, this was 91.7% of theknown added concentration (CV = 8.8%, n = 12).

Recovery. The analytical recovery of tritiated thrombox-ane added to culture-medium and plasma samples was 70(SD 6.4)%.

B

/Bo

100

80

60

40

20

010 100 10001

80

040

20

01

ng/LFig. 6. Intra-assaycoefficientsof variationat differentconcentra-tions of 6k-PGF1,. for the BioQuant assay (heavy line) and theAmershamassay (light line)

1*) 100 1000

%8oB

/Bo 40

0

0.01

relative dilution factorFig. 5. Inhibition’curves of four samplesassayed in differentdilu-tions (differentsymbols)were parallelto the standardcurveof theassay (solid line, no symbols)

0.1 1

I

200

0 200 400 600 800

0 200 400 600 800

Amersham [ng/L]Fig. 7. Correlation of concentrations in the AmershamandBioQuantassays for 6k-PGF1,, in culture-medium samples (top; y = -17.4 +0.882x, r = 0.99; n = 24) and in plasma samples (bottom; y =

-0.498 + 0.973x, r = 0.99; n = 26)

CLINICALCHEMISTRY, Vol. 36, No. 3, 1990 513

pg/wellFig. 4. Standard curves for 6k-PGF1,.in a solid-phaseassay after15- and 90-mm incubationsThe slopes ofthe logit-logtransformationsare-2.49 and -2.32, respectively.Standards were assayed in duplicates. Standard curves at 60- and 90-mmincubatlonswerealmost congruent.lmmunoglobulmnwas Immobilized at 2pmol per well

Table 2. Accuracy of the Solid-Phase Assay for6k-PGF1a

6k-PGF1,, added, ng/L 6k-PGF1,.assayed,% of added100 99.3200 83.5400 92.2

n = 12each.

Precision. 1) Detection limit. With the solid-phase assaywith microwells, we could detect at least 2.1 pg of 6k-PGF1,.per well, estimated as the amount that exceeds 3 SD of thecounts from radiolabeled 6k-PGF1,. in the absence of unla-beled prostaglandin.

2) Intra-assay variation. The intra-assay CV was 9% forsingle determinations at 20 pg per well of prostaglandin(190 degrees of freedom from 72 duplicate determinationsand 63 triplicates). This compares very closely with that forthe Amersham assay (Figure 6). If the same quality-controlsample was measured repeatedly (at about 20 pg/well),intra-assay variability was 5.7% (n = 16, single determi-nations).

3) Interassay variation. Calculated from 10 assays per-formed on different days by the same person, the totalvariation was 11.8% at 50% displacement of the radiola-beled tracer. At 20% binding of radiolabeled derivative, wemeasured 80.9 (SD 10.2) pg/well, and at 80% binding 6.8(SD 1.5) pg/well (n = 10).

4) Nonspecific binding. Nonspecific binding of the radio-labeled prostaglandin derivative was <1% for -25 000counts/mn used in 200 L total volume per well (141 ± 25counts/mn, mean ± SD, n = 10).

Correlation. Measured concentrations of 6k-PGF1,. inculture-medium samples (range 50 to 700 ngIL) werehighly correlated in the BioQuant and Amersham assay(Figure 7), although values estimated with the BioQuantassay were somewhat lower over the whole range. Forsamples of human plasma, however, the concentrationslargely agreed within standard errors of the assay.

DiscussionWe have developed a solid-phase assay of 6kPGFja that

combines the advantages of the microplate format with thesensitivity of ‘251-labeled derivatives as tracers in immu-

514 CLINICALCHEMISTRY,Vol.36, No. 3, 1990

noassays. Although progress in solid-phase technology hasprovided fast, reliable immunoglobulin-based analyticalsystem for many antigens, the commercially availableimmunoassays for prostaglandins still require incubationand handling times exceeding an 8-h working day. Theassay we described can be performed with an incubationtime of 60 to 90 mm. In addition, because multiple pipettorscan be used and washing the plates consists of just pouringoff supernatant fluids, hands-on time is substantially re-duced so that 400 samples can be comfortably processed induplicate during an 8-h period.

Immunoassays involving 1251-labeledderivatives in com-bination with the solid-phase technology are comparativelynew for the measurement of prostaglandins. Therefore,extensive validations by procedures based on independentprinciples (e.g., gas chromatography/mass spectrometry)are not yet available. Although both methods presented inthis publication are based on immunochemical detection,they substantially differ with regard to several aspects: theantibodies’ species of origin, the technique for the iminobi-lization of immunoglobulins, the crossreactivities of theantibodies, the separation techniques, the structure of thesolid phase, and incubation times. The value of this com-parison study, therefore, is our finding that the results ofthe simpler method are very similar to those obtained witha previously available assay.

The use of a highly purified, mono-iodinated 6kPGFiaderivative as radiolabeled tracer contributes to the sensi-tivity and the simplicity of the analytical system. Only oneincubation step is required, without elaborate preparationof reagents. The small volume of incubation medium (200zL per well) and the small size of the assay container

(microwell) reduces the amount of radioactive waste mate-rial substantially.

Prostaglandins are typically extracted from samples by asimple chromatographic step on commercially availablereversed-phase liquid-chromatography cartridges. Becausebiological media of different origin will always containphospholipids and a large variety of prostanoids, the ex-traction decreases potential interference from crossreac-tivity and matrix effects. However, an extraction stepmakes desirable a method for estimating recovery. Manyinvestigators determine the recovery by summation ofseparate studies, whereas others prefer to determine therecovery for each sample separately, e.g., by adding a smallamount of tritiated prostaglandin to the sample and mea-suring the beta-radiation in an aliquot after extraction. Inhighly sensitive assays as described here, this might causeproblems. To obtain a reasonable signal for tritiated pros-taglandmn, an equivalent of 10 pg needs to be added to 1 mLof sample. We added tritiated thromboxane, which hassimilar adsorption and elution properties to 6k-PGF1,, butdoes not cross-react substantially with the antibody (Table

The kinetic studies summarized in Figure 3 allow us todraw some conclusions as to the relative binding affinity ofthe native 6kPGFia and the radiolabeled derivative, ifthey differ. As we have shown in studies with monoclonal

antibodies in solid-phase assays (9, 10), standard curvesobtained after different incubation times show inflectionpoints of the sigmoidal plot at different concentrations ofantigen per well (Figure 4). This is a reflection of the abilityof the antibody to recognize, in part, the bridge groupbetween the prostaglandin and the tyrosine methyl estermoiety in the radiolabeled derivative. Because the bridgegroup was the same for the hapten-carrier protein that wasprepared for the immunization of the sheep, apparently apopulation of immunoglobulins in this polyclonal serumalso partly recognizes the bridge group. A relatively largeramount of native 6kPGFia is required at shorter incuba-tion times to displace the same amount of radiolabeledprostaglandin. Therefore, the on-rate constant for bindingthe 6kPGFiaTME is greater than that for 6kPGFia. Inthese circumstances, the specificity of the assay systemmay vary as a function of the incubation time (11). Inpractical terms, this means that, if several plates areprocessed, care should be taken to keep the incubation timeof the plates approximately equal. If not, each plate shouldcontain a separate standard curve against which sampleson that plate are to be estimated.

We are indebted to Dr. M. Peters-Golden,University of Michi-gan, for the medium from the cultured rat macrophages,and Drs.K. Winter and A. Jimenez,Harvard Medical School,for the humanplasma.

References1. Powell WS. Rapid extraction of arachidonic acid metabolitesfrom biological samples by using octadecylsilyl silica. MethodsEnzymol 1982;86:467-76.2. Westcott JY, Chang 5, Balazy M, et al. Analysis of 6-ketoPGF1,,,5-HETE and LTC4 in rat lung: comparison of GCIMS, RIAand EIA. Prostaglandins 1986;32:857-73.3. Hunter WM, GreenwoodFC. Preparation of iodine-131 labelledhuman growth hormoneof high specific activity. Nature (London)1962;194:495-6.4. Spron PH, Peters-GoldenM, Simon RH. Hydrogen-peroxide-induced arachidonicacid metabolism in the rat alveolar macro-phage.Am Rev Respir Dis 1988;137:49-56.5. Ishikawa E, Hashida S. Kohno T, Tanaka K. Methods forenzyme-labelingof antigens, antibodies and their fragments. In:Ngo TP, ed.Nonisotopic immunoassays. New York: Plenum Press,1988:27-55.6. Ishikawa E, Imagawa M, Hashida 5, Yoshitake 5, HamaguchiY, Ueno T. Enzyme-labeling of antibodies and their fragments forenzyme immunoas8ay and inimunohiatochemical staining. J Im-munoassay 1983;4:209-327.7. Sankolli GM, Stott RA, Kricka U. Improvement in the anti-body binding characteristics of microtitre wells by pretreatmentwith anti-IgG Fc immunoglobulin. J Immunol Methods 1987;104:191-4.8. Edwards PR, Ekins RP. Mass action modelbased microproces-sor program for RIA data processing.In: Hunter WM, Come JET,eds. Immunoassays for clinical chemistry, 2nd ed. Edinburgh:Churchill Livingstone, 1983:640-52.9. Schranun W, Yang T, Midgley AR. Surface modificationwithProtein A for uniform binding of monoclonal antibodies. ClinChem 1987;33:1338-42.10. Schramm W, Yang T, Midgley AR. Monoclonal antibodiesused in solid-phase assays and liquid-phase assays, as exemplifiedby progesterone assay. Clin Chem 1987;33:1331-7.11. Vimng RF, Compton AP, McGinley AR. Steroid immunoas-say-effect of shortened incubationtime on specificity.Clin Chem1981;27:910-.3.