CLIN.CHEM.37/9, 1618-1625(1991)

1618 CLINICAL CHEMISTRY, Vol. 37, No. 9, 1991

Prostate-Specific Antigen in Serum Occurs Predominantly in Complex witha.1-AntichymotrypsinHans Lilja,’ Anders Chrlstensson,’ Ulrika Dahl#{233}n,2Marja-Terttu Matlkainen,3 OHe Nilsson,2 Kim Pettersson,3and Timo Lovgren4

Immunologic measurements of the serum concentrationof prostate-specific antigen (PSA), an abundant prostatic-secreted senne proteinase, are frequently used to monitorpatients with prostate cancer, though it has not beenascertained whether this immunoreactivity represents aPSA zymogen, the active proteinase, or PSA complexedto extracellular proteinase inhibitors. To characterize thePSA immunoreactivity in serum, we used monoclonalantibodies produced against PSA and a polyclonal rabbitlgG against a1-antichymotrypsin In the design of threenoncompetitive PSA assays: assay T, which detectedPSA both when present as the active proteinase andwhen complexed to a1-antichymotrypsin;assay F, whichrecognized the active proteinase but most poorlydetectedPSA complexed to a1-antichymotrypsin; and assay C,which was specific for PSA complexed to a1-antichymot-rypsin. We used the three assays to measure PSAimmunoreactivity in 64 patients’ sera and in the effluentafter gel chromatography of sera from four patients. Thisidentified an 80- to 90-kDa complex between PSA anda1-antichymotrypsin as the predominant fraction of thePSA immunoreactMty in blood plasma; an immunoreac-tive 25- to 40-kDa compound was the minor fraction.

AdditIonal Keyphrases: monoclonal antibodies . prostate dis-ease chromatography, gel fiftration fluoroimmunoassay

Prostate-specific antigen (PSA), one of the three mostabundant prostatic-secreted proteins in human semen(1), is involved in the dissolution of the seminal gelstructure (formed at the ejaculatory mixing of the secre-tions from the accessory sex glands) that occurs concom-itantly with the limited proteolytic fragmentation byPSA of the predominant proteins secreted by the semi-nal vesicles (2-7). The single-chain 33-kDa glycopro-tein, PSA, was first described by Wang et al. in 1979 (8).

Independently and almost simultaneously, an appar-

ently identical protein was discovered by Hara andKimura (9) and Graves et al. (10). PSA was later shownto be produced as a presumably inactive precursor (11)that is converted into an active serine proteinase (3,12).

‘Department of Clinical Chemistry, University ofLund, MahnoGeneral Hospital, S-214 01 MalmO, Sweden.

2Pja CanAg, S-402 42 Gothenburg, Sweden.

3Pharmacia Wallac Oy, SF-201 01 Turku, Finland.4Department of Biochemistry, University of Turku, SF-205 00

Turku, Finland.5Nonstandard abbreviations: PSA, prostate-specific antigen;

SDS-PAOE, sodium dodecyl sulfate-polyacrylamide gel electropho-reals;and Mab, monoclonalantibody.

Received February 26, 1991; accepted June 28, 1991.

The active proteinase has a 237-amino-acid polypeptidebackbone that manifests extensive similarity with thatof the glandular kallikreins (EC 3.4.21.35) (11-13) al-though, unlike the trypsin-like glandular kallikreins,PSA displays chymotrypsin-like substrate specificity(7, 14, 15).

In 1980, Papsidero et al. (16) were the first to reportthat PSA is regularly detected in the sera of prostaticcancer patients. Measurements of the serum concentra-tion of PSA are now used widely to monitor patientswith prostatic cancer (17, 18), although above-normalserum concentrations of PSA have also been reportedboth in benign prostatic hyperplasia and secondary tosurgical trauma of the prostate (17,18). However, themolecular mass of the immunoreactive form of PSA inserum is uncertain, the data reported hitherto being indisagreement as to the size of the molecules carrying theiminunoreactive sites (16,19). It has not been thor-oughly investigated whether the estimated immuno-reactivity represents the PSA zymogen, the active pro-teinase, or PSA inactivated by complex formation withextracellular inhibitors of serine proteinases. However,release of the active proteinase in intercellular fluids orblood plasma would be expected to result in rapidinactivation of the enzymatic activity by extracellularinhibitors of serine proteinases. We recently investi-gated the ability of PSA to react with extracellularserine proteinase inhibitors and found that purified PSAformed stable complexes with isolated a1-antichymo-trypsin, a2-macroglobulin, and the cr2-macroglobulin-analog pregnancy zone protein (15). This also occurredwhen purified PSA was added to blood plasma in vitro(15). Therefore, for the present investigation, we de-signed sensitive methods for the assay of noncomplexedPSA and of PSA complexed to the serine proteinaseinhibitor, to characterize the molecular forms of PSAthat occur in extracellular fluids in vivo.

Materials and Methods

Reagents

Molecular mass markers for sodium dodecyl sulfate-polyacrylanude gel electrophoresis (SDS-PAGE), Sepha-cryl S-300, CNBr-activated Sepharose, and ProteinA-Sepharose were from Pharmacia LKB Biotechnology(Bromma, Sweden). The Arcus 1230 fluorometer, theeuropium chelate of isothiocyanatobenzyldiethylenetri-aminetetraacetic acid, Delfia#{174}assay buffer, and Delfiaenhancement solution were from Pharmacia Wallac(Thrku, Finland), and 8 x 12 microtiter strip wells werefrom EF-Lab (Helsinki, Finland). The ProtoBlot#{174}sys-

CLINICALCHEMISTRY,Vol.37, No.9, 1991 1619

tem with alkaline phosphatase-conjugated goat anti-rabbit IgG and goat anti-mouse IgG was from PromegaBiotec (Madison, WI). The goat anti-mouse immune-globulin (G,A,M; heavy + light), the peroxidase-labeledisotype-specific rabbit anti-mouse IgGi, IgG2a, IgG2b,IgG3, IgA, and 1gM were from Zymed Labs., Inc. (SanFrancisco, CA). KC-2000 cell culture medium was fromHazleton Biologics, Inc. (Lenexa, KS), and HAT mediasupplement H-0262 (hypoxanthine, aminopterin, andthymidine) was from Sigma Chemical Co. (St. Louis,MO). A Biosil TSK 250 HPLC column and nitrocellulosemembranes (Trans-Blot#{174}transfer medium, O.45-mpore size) were from Bio-Rad Labs., Richmond, CA.Affinity-purified polyclonal rabbit IgG against PSA wasproduced at our laboratory as previously described (1);polyclonal rabbit IgG against a1-antichymotrypsin andrabbit anti-mouse IgG were from Dakopatts (Copenha-gen, Denmark). Purified a2-macroglobulin and a1-anti-chymotrypsin were obtained as previously described(15). Purified PSA, consisting of -65% active proteinaseand 35% modified inactive PSA, was obtained from

human semen as previously described (15). The modifiedinactive PSA, which contains an internal peptide bondcleavage, was purified from incubation mixtures withpurified PSA and purified a1-antichymotrypsin, andisolated as previously described (15). Previously de-scribed procedures were also used to obtain stable com-plexes between purified PSA and purified a1-antichy-motrypsin, and for two-step chromatographic isolationof complexes between PSA and a1-antichymotrypsin(15). AffInity-purified antibodies against a2-macroglob-nun (obtained from a polyclonal rabbit antiserum byafilnity-chromatographic adsorption to purified a2-mac-roglobulin immobilized to Sepharose 4B) were immobi-lized to CNBr-activated Sepharose 4B with standardprocedures.

The serum samples were frozen blood sera (n = 64)sent to our laboratory for routine clinical chemical anal-ysis of prostatic acid phosphatase activity. Seminalplasma was obtained from healthy men as describedpreviously (4) and stored at -20 #{176}Cuntil use.

Procedures

a2-Macroglobulin-depleted serum. Human serum, ob-tained from healthy women volunteers, was chromato-graphed on a column with immobilized rabbit IgGagainst a2-macroglobulin. Fractions eluting in the voidvolume were pooled and found by electroimmunoassay(20) to contain <1% of the initial concentration ofa2-macroglobulin. This pool was stored at -20 #{176}Cuntiluse.

Hydroxylamine-treated serum. The proteinase inhibi-tory activity of the a2-macroglobulin fraction in serumwas irreversibly inactivated as a result of hydrolysis ofits internal thioester bonds by incubating human serumfrom healthy women overnight at 37#{176}Cwith hydroxyl-amine at a final concentration of 25 mmol/L, as de-scribed by Chen et al. (21).

General procedures for fluorometric assays. Using apreviously described procedure (22), we coated microti-

ter strip wells with 1-2 pg of IgG in 200 pL of buffer.The coated plate was stored at 4#{176}Cand the coating wasstable for more than six months. Purified PSA, mono-clonal antibodies against PSA (anti-PSA Mabs), andpolyclonal rabbit IgG against a1-antichymotrypsin werelabeled to a specific activity of 2-5 Eu ions per moleculewith the Eu chelate of isothiocyanatobenzyldiethylene-triaminetetraacetic acid as described by HemmilA et al.(23).

MonoclonalAntibodiesBalb/c mice were immunized by intraperitoneal injec-

tion with 70 pg of PSA emulsified with Freund’s com-plete adjuvant. Booster doses of 50 pg were given atthree- to four-week intervals and a final 40-pg boosterafter another three-week interval. Four days after thefinal booster dose, the mice were killed and their splemclymphoid cells fused with plasmacytoma cells NS-1 at a1:1 ratio (24,25). The fused cells were harvested inmicrotiter strip wells in KC-2000 medium containingfetal calf serum, 200 milL, and HAT supplement (dilut-ed 50-fold).

Anti-PSA-specific antibody production was assayedwith microtiter strip wells coated with rabbit anti-mouse immunoglobulin (22). The coated strips wereincubated overnight at 4#{176}Cwith hybridoma supernates,or with a Mab against PSA (0812; Hybritech Inc., SanDiego, CA) used as standard and diluted in Delfia assaybuffer (22); washed six times; and incubated with Eu-labeled PSA (50 ng/well) for 1 h at room temperature.The amount of bound Eu-labeled tracer was determinedwith an Arcus fluorometer after another six washingsand the addition of Delfia enhancement solution (26).

Master cell lines were cloned by limited dilution (27).The desired cell lines were expanded intraperitoneallyin Balb/c mice and the ascitic fluid was collected within10 days. The Mabs were purified by chromatography onProtein A-Sepharose according to the protocol recom-mended by the manufacturer.

Characterization of purified Mabs against PSA. Theisotype of each Mab was determined in microtiter stripwells coated with goat anti-mouse immunoglobulin (G,A, M; heavy + light) and identified with peroxidase-labeled isotype-specific rabbit anti-mouse immunoglob-ulin. The ability of each antibody to recognize PSA indifferent PSA-containing samples was determined withsample proteins blotted to poly(vinylidene fluoride)membranes after agarose gel electrophoresis or SDS-PAGE. A previously described procedure was used to blotproteins from agarose gels (28) after electrophoresis in75 mmol/L barbital buffer containing Ca2, 2 mmol/L,at pH 8.6 (29). SDS-PAGE was run in gradients of70-120 or 90-170 g/L, as described by Laemmli (30), butwith the buffer system of Blobel and Dobberstein (31).Western blotting, performed essentially as described byBurnette (32), was done in an apparatus for semidryelectroblotting. Immunoreacted antibodies were de-tected either with alkaline phosphatase-conjugated goatanti-rabbit IgG or goat anti-mouse IgG.

Design of noncompetitive assays. Three different as-

B

kD.A B #{149}C D

43.0_ . .1

30.0_ “‘‘ #{149}“-

123 123 123 123FIg. 1. Immunodetectlonof proteins blotted to poly(vinylidenefluoride) membranesafter agarose gel electrophoresis(left) and after

SDS-PAGE(rht)Immunodetectionwas performed with (A) polyclonalrabbit anti-PSAgO, ( Mab2E9. (C) Mab2H1 1, and (C)Mab 5A10. Lane 1,1 pg of purified PSA; lane 2,I pg of purified PSA IncubatedwIth 6 pg ofpurified a,-antlchymotrypsln for 30 mmat 37 ‘C; lane 3,6 pg of purified a,-antichymotiypsin. Left: * Indicates site ofapplication; right all samples were reduced (10 mmol/L dfthiothreltol) and carboxymethylated before analysis with SDS-pAc3EIn 100 WI gels

1620 CLINICALCHEMISTRY,Vol.37, No.9, 1991

says were designed: assay T (total PSA), with 2E9anti-PSA Mab as the capture antibody and Eu-labeled2H11 anti-PSA Mab (both described below) as the de-tection antibody; assay F (free PSA), with the samedetection antibody as assay T but with the 5A10 anti-PSA Mab (described below) as the capture antibody; andassay C (complexed PSA), with the same capture anti-body as assay T but Eu-labeled polyclonal rabbit IgGagainst a1-antichymotrypsin as the detection antibody.AssaysT and F were standardized with purified PSA(the concentration of PSA being determined by quanti-tative amino acid analysis), whereas assay C was stan-dardized with an isolated complex between PSA anda1-antichymotrypsin. Samples were analyzed as follows:25 pL of sample and 100 pL of Delfia assay buffer wereincubated for 1 h in the microtiter strip wells withconstant shaking, after which the wells were washedand then incubated with Eu-labeled detecting antibody(100 ng/well) for 1 h, and the amount of bound Eu-labeled antibody was determined.

Recovery of PSA added to human serum. PSA-contain-ing samples (10-200 pg of PSA) were added in aliquots of<40 pL to 1-mL aliquots of human serum (untreatedserum from healthy women volunteers, a2-macroglobu-lin-depleted serum, or hydroxy1smine-treated serum)and incubated for 0-20 h at room temperature. After theincubation, the recovery of the added amount of PSA wasmeasured with assay T.

Analysis for PSA in patients’ sera. PSA immunoreac-tivity in samples from 64 patients was measured withthe three PSA assays (T, F, and C), and with theTandem-R PSA kit (Hybritech Inc.) according to theprocedure recommended by the manufacturer. Calcula-tions of intra-assay precision and the detection limits ofassays T, F, and C were made from 10 replicate analysesof the standards of each assay. The detection limit wasdetermined with the following formula: {2 SD counts/s(std. 0)/[mean counts/s (std.1) - mean counts/s (std. 0)]}x concu (std. 1). The “clinically useful” detection limitwas defined as the mean +2 SD of the detection limit of10 separate analyses.

Gel-filtration chromatography. Patients’ sera withvery high concentrations of PSA were diluted threefoldwith 20 mmol/L sodium phosphate buffer (pH 6.8) con-taining 0.15 mol of sodium chloride per liter, and sub-

+ C D

*

1 23 1 23 123 123

jected to gel filtration to estimate the molecular size ofthe PSA-immunoreactive material. We subjected 50 pLof each diluted sample to gel filtration on a 300 x 7.5mm Biosil TSK 250 HPLC column (exclusion limit 200kDa), equilibrated and eluted with the dilution buffer,at a flow rate of 0.5 mlJmin. We collected the effluent infive-drop fractions and used Blue Dextran (2.0 MDa),bovine serum albumin (67 kDa), ovalbumin (43 kDa),and chymotrypsinogen (25 kDa) to calibrate the column.The concentration of PSA imniunoreactivity in eachfraction was determined with assays T, F, and C asdescribed above.

Results

Characterizationof the EpitopesDefined by Three Mabsagainst PSA

The three Mabs (2E9, 2H11, and 5A10) were found tobe of the IgG1 isotype. Polyclonal anti-PSA IgG and thethree Mabs were used to probe proteins blotted to thepoly(vinylidene fluoride) membranes after agarose gelelectrophoresis or SDS-PAGE (Figure 1). On blots fromthe agarose gel, purified PSA (lane 1, Figure 1, left) andthe modified inactive PSA (not shown) were identifiedby the polyclonal antiserum and the three Mabs; PSAcomplexed to a1-antichymotrypsin was also identifiedby the polyclonal antiserum and by two of the Mabs butnot by Mab 5A10 (lane 2, Figure 1, left).

On Western blots from SDS-PAGE of nonreduced sam-ples, purified PSA and the modified inactive PSA wereidentified by the polyclonal antiserum and the threeMabs. PSA complexed to a1-antichymotrypsin was iden-tified by the polyclonal antiserum and two of the Mabs,whereas the complex was poorly detected by the 5A10Mab (not shown). On Western blots from SDS-PAGE ofreduced samples, purified PSA (lane 1, Figure 1, right)and PSA complexed to a1-antichymotrypsin (lane 2,Figure 1, right) were identified by the 2E9 Mab and thepolyclonal antiserum; the 2H11 anti-PSA Mab produceda weak, barely discernible reaction with the purifiedPSA (but not with the PSA complexed to a1-antichymo-trypsin) (lanes 1 and 2, Figure 1C, right); the 5A10anti-PSA Mab produced a similarly weak, barely dis-cernible reaction with both the purified PSA and thePSA complexed to a1-antichymotrypsin (lanes 1 and 2,Figure ID, right). In this system, the glycosylated heavy

Purified PSAPurified PSAand200 pg of purified

a1-antlchymotrypslnPooledseminal plasmaPooledseminal plasmaand 200 pg of

purified a1-antlchymotrypsln1:1 molar ratio complex between purified

PSAand a1-antichymotrypsln

100 100 293 50 50

97 95 298 35 70

97 8 100

a samples contaIning25 pg of PSA were incubated for 2 h after dilutionwithphosphate-buffered saline (20 mmolof sodiumphosphate and 150 mmol ofNaCIper liter, pH 7.2), then were further diluted 200-fold with Ins-bufferedsaline (per liter, 50 mmol of TrIs, 150 mmol of NaCI,0.5 got NaN3,and 75 gof bovine serum albumin, pH 7.75) before assay.

bin assays land F, the response magnitudeofeachsamplewas expressedIn percent of purified PSA, a relatIve response of 100% being assigned to thepurified PSA. In an Identical procedure, the response magnitude of eachsample In assay C was expressed In percentof purifiedcomplex betweenPSAand a1-antlchymotrypsln,this complex havingbeen assigned to give a relativeresponse of 100% In assay C.

CLINICAL CHEMISTRY, Vol. 37, No.9, 1991 1621

chain (residues 1-145) in the modified inactive PSA wasalso identified by 2E9 Mab (not shown).

NoncompetitiveAssays of PSABecause assay T was designed to measure the total

amount of PSA in a sample, both when present in itsnoncomplexed form and when inactivated by complexformation with a1-antichymotrypsin, we used two Mabs(2E9 and 2H11) that identified both purified PSA andPSA complexed to a1-antichymotrypsin in the blottingprocedure from agarose gels. In assay F, designed specif-ically to measure the noncomplexed PSA, we used acapture antibody (5A10) that identified purified PSA butnot the PSA complexed to a1-antichymotrypsin in blotsfrom agarose gels, and the same Eu-labeled detectionantibody (2H11) as in assay T. The capture antibody inassay C (2E9), recognizing both PSA and PSA complexedto a1-antichymotrypsin in the blots from agarose gels,was combined with the Eu-labeled polyclonal rabbit IgGagainst a1-antichymotrypsin to render the assay specificfor PSA complexed to a1-antichymotrypsin.

Analysis for purified PSA and the PSA in the pooledseminal plasma gave identical dose responses in assayT. Although the dose-response of purified PSA wasslightly greater than that of the purified PSA incubatedwith a1-antichymotrypsin, the dose-response curve forthe PSA in pooled seminal plasma was identical withthat for the PSA in pooled seminal plasma incubatedwith a1-antichymotrypsin in this assay (Table 1). More-over, purified PSA, the PSA in pooled seminal plasmaincubated with a1-antichymotrypsin, and the isolatedcomplex between PSA and a1-antichymotrypsin all dis-played identical, parallel, and linear dose-response re-lationships in the assay (Figure 2, left).

With assay F, purified PSA and the PSA in the pooledseminal plasma gave almost identical dose-responsecurves, whereas only 30-50% of the purified PSA incu-bated with a1-antichymotrypsin or of the PSA in pooled

Table 1. RelatIve Responses of the Three PSA AssaysRe4lvs iws-ons., %b

Assay T Assay F Assay C

seminal plasma incubated with a1-antichymotrypsinwas detected (Table 1). The detected fraction in thepurified PSA incubated with a1-antichymotrypsin andthe PSA in seminal plasma incubated with a1-antichy-motrypsin is presumably the 30-40% of the PSA in theseminal plasma that contains an internal peptide bondcleavage that renders the protein unreactive with a1-antichymotrypsin (15). Thus, the PSA complexed toa1-antichymotrypsin appears to be poorly detected bythis assay. Consistent with this interpretation, the iso-lated PSA-a1-antichymotrypsin complex was verypoorly detected by the assay (Table 1) and did notdisplay any parallel dose-response relationship as com-pared with purified PSA (Figure 2, middle).

Assay C (with Eu-labeled polyclonal rabbit IgGagainst a1-antichymotrypsin as the detecting antibody)was specifically designed to detect PSA in complex witha1-antichymotrypsin; thus, it failed to detect eitherpurified PSA or PSA in seminal plasma (Table 1; Figure2, right). Therefore, assay C had to be standardized withthe isolated PSA-a1-antichymotrypsin complex, theconcentration of PSA in the complex having been deter-mined with assay T. Only 50-70% of the purified PSAincubated in a1-antichymotrypsin or the PSA in semi-nal plasma incubated with a1-antichymotlyp8in wasdetected by the assay. This agrees with the analysis ofthese samples with assay F and with our previousfinding that 30-40% of the PSA in these samples wasunreactive with a1-antichymotrypsin (15).

Reactionsafter Adding PSA to Serum

If purified PSA or pooled seminal plasma was added toserum (experiments 1 and 2, Table 2), the recovery ofthe added amount of PSA decreased within 1 h to60-70% of the original value with assay T. The poorrecovery of PSA in these experiments might have beendue to inaccessibility of the epitopes of PSA after itscomplexing with a2-macroglobulin. This complex forma-tion is somewhat faster than that between PSA anda1-antichymotrypsin after in vitro addition of PSA toblood serum (15). To check this finding, we added theisolated PSA-a1-antichymotrypsin complex to serum(experiment 3, Table 2). In another experiment, the PSAin seminal plasma was allowed to form a stable complexwith a1-antichymotrypsin by incubating the seminalplasma with a1-antichymotrypsin for 2 h before theseminal plasma reacted with a1-antichymotrypsin wasadded to serum (experiment 4, Table 2). Both experi-ments resulted in 90-95% recovery of the added amountof PSA. The interpretation that the poor recovery ofPSA is due to its complex formation with a2-macroglo-bulin in blood plasma is supported by the high recovery(89-98% of the original value) of PSA when the seminalplasma was added to a2-macroglobulin-depleted serum(experiment 5, Table 2). The same high recovery of PSAwas obtained when seminal plasma was added to serumcontaining hydroxylamine-inactivated a2-macroglobu-un (experiment 6, Table 2).

l0

io6

I0U,a.U

U, I0a.U

I I#{149}0 laough

10 00 000 I 0 00

ug/L US/I.

000

Table 2. AnalytIcal Recovery of Different Forms of PSAAdded to Human Serum

PSA, Mg/L (and recovery, %)

ASSaYI Assay F Assay C

1622 CLINICALCHEMISTRY,Vol.37, No.9, 1991

FIg. 2. Dose-response relationshipsofthe three PSA assaysAsaayT (Mab2E9 solid-phase capture antibody andMab 2H1 I Eu-labeledtracer), assay F (5A10 solid-phase capture antibodyand 2H1 1 Eu-labeledtracer), andassay C (2E9 solid-phase capture antibodyand polyclonalantl-a,-antichymotrypsinEu-labeledtracer). PSA standardsconsistedof x, purified PSA; A, 1:1 molarratio complex between purified PSA and a,-antlchymotiypeln; and c, pooledseminalplasmaIncubated with purified a1-antichymotrypsln(1:4 molar ratio of PSAto a1-antlchymotiypsln). The standards were diluted to 1,5, 10, 100,and 500 pg of PSA per liter with Iris-buffered saline (per liter, 50 mrnolof Iris, pH 7.75,0.15mel of NaCI, 0.5g of NaN3,and 759 of bovine serum albumin) and then analyzed as described In Materials and Methods

Exps,Imsnta Oh I h 20h

1 109 (101)27 (98)

6417

(59)(63)

not determinednotdetermined

2 145 (96)71 (96)

101

53(67)

(72)notdeterminednotdetermined

3 215 (104)102 (104)10.7 (94)

197

9110.3

(95)

(93)(90)

190 (91)

94 (96)9.8 (86)

4 182 (97)96 (99)11.4(101)

19510410.5

(104)(107)(93)

179 (95)

90 (93)10.7 (95)

5 162 (101)81 (97)8.6 (98)

154787.8

(96)(94)(89)

143 (89)85 (102)7.9 (89)

6 161 (100)77 (95)8.6 (102)

149748.2

(92)(91)

(98)

150 (93)69 (85)8.6(102)

aSamples of purified PSA or seminal plasma (10-200 pg of PSA) wereadded to serum In the following forms: exp. 1, purIfied P5A; exp. 2, pooledseminal plasma; exp. 3, Isolated complex with a,-antlchymotiypeln; exp. 4,seminal plasma (complexed to a1-antlchymotzypelnby Incubationof PSA Inpooled seminalplasma with a fourfoldmolarexcess of purified a1-antichymo-trypeln at 37 ‘C for 2 h); exp. 5, pooled seminal plasma (the serum beinga2-macroglobullndepleted); and exp. 6, pooledsemInal plasma (the serumhaving been hydroxylamine treated). PSA was measured at various timesafter mixing.

Analysisfor PSA in Serum

We measured the concentration of PSA immunoreac-tivity in serum samples from 64 patients, using thethree assays (T, F, and C). Estimates of the intra-assaycoefficients of variation were 7.2-3.4% over the range ofconcentrations of the standard curve with assay T,9.4-4.0% with assay F, and 13.5-5.2% with assay C.The detection limit, calculated as described earlier, was0.12 ugfL in assay T, 0.15 p.gfL in assay F, and 0.56 pg/Lin assay C. Regression analysis of the results obtainedwith assays T and C gave y = 0.89x + 6.55 (r = 0.97).This is consistent with the regression analysis betweenassays T and F: y = 0.lOx + 9.56 (r = 0.82). Further-

more, the analysis with assays T and C indicated that>50% of the PSA immunoreactivity in serum (medianvalue 86%) was complexed to a1-antichymotrypsin in 55of the 64 patients’ samples. This agrees with the findingthat 7-50% of the PSA imniunoreactivity in serum(median value 22%) was not in complex with a1-antichy-motrypsin in 56 of the patients’ sera analyzed withassays T and F. In eight of the sera, assays C and Tindicated that <50% of the PSA immunoreactivity wascomplexed with a1-antichymotrypsin. This result alsoagreed with those of assays F and T, which indicatedthat >50% of the PSA immunoreactivity was not com-plexed with a1-antichymotrypsin in these sera. Regres-sion analysis of the results obtained with assay T (x) andthe Tandem-R PSA (y) gave y = 1.lOx - 0.90 (r = 0.98,n = 58).

The molecular mass of the PSA immunoreactivity insera of four patients was studied by gel-filtration chro-matography, followed by analysis of the PSA immuno-reactivity in the eluted fractions with each of the threeassay procedures (Figure 3). The recovery of PSAimmunoreactivity from the gel-filtration chromatogra-phy was 82-107%, as measured in the elated fractionswith the three different assays. Fractions analyzed withassay T, which detected noncomplexed PSA and thePSA complexed to a1-antichymotrypsin equally well,contained one predominant peak of PSA immunoreac-tivity that eluted at a volume corresponding to a mass of80-90 kDa. Post-chromatographic analysis of patients’sera with assay C, which detected PSA complexed toa1-antichymotrypsin but not noncomplexed PSA, alsoidentified one predominant peak of immunoreactivitythat eluted at a volume corresponding to a mass of80-90 kDa.

Analysis of chromatographed fractions with assay Talso identified a minor peak of PSA immunoreactivitythat eluted at a volume corresponding to a molecularmass of 25-40 kDa (Figure 3), a mass corresponding tothat previously reported for purified PSA (8,11,15).

The idea that this minor peak of PSA immunoreactivitymay represent noncomplexed PSA derives support fromthe inability of assay C to identify any distinct peak ofimmunoreactivity eluting at this position in the chro-

Sample 2Sample I

I 35 45 55Fraction

Sample 3

O.

Fraction

Fig. 3. Gel chromatographyofsamples

Sample 4

3 45 55

Fraction

PSA immunoreactMtyIn serum

Patients’sera(50fL ofa threefolddIlution)were chromatographedona BlosilTSK 250 HPLC column and the effluent was measuredfor PSA Immunoreac-tlvity with assay T (D), assay F (s), and assay C (x) (see Ag. 2 caption fordescriptionofassays).The elution positions for molecular mass references(Blue Dextran2.0 mDa,bovine serumalbumin67kDa,ovalbumln43 kDa,andchymotlypslnogen 25 kDa) were determined in a separate run. PSA Immuno-reactivityof sample I was 1950 1ig/Lwith assay 1,265 tg/L with assay F, and1894 pg/I. with assay C; that of sample 2 was 1134, 225, and 1225 p.g/L.respectively; that of sample 3 was 705, 85, and 643 pg/L. respectively;andthat of sample4 was 1745, 90, and 1921 ttg/L respectively

CLINICALCHEMISTRY,Vol.37, No.9, 1991 1623

matogram. Also, in analysis of the eluted fractions withassay F (which detected poorly the PSA complexed toa1-antichymotrypsin as compared with its detection ofnoncomplexed PSA), these fractions yielded a predomi-nant peak of PSA immunoreactivity that eluted at avolume corresponding to a molecular mass of 25-40 kDaand only a minor peak of iinmunoreactivity eluting at avolume corresponding to a mass of 80-90 kDa (Figure 3).

DiscussionThe data presented here include a partial character-

ization of the epitope specificity of three Mabs againstPSA. However, to facilitate interpretation of some of thefindings, we emphasize that the purified PSA used inthe experiments was obtained from seminal fluid, wherethe active serine proteinase constitutes 60-70% of thetotal PSA content; 30-40% is inactive because of aninternal peptide bond cleavage (15). One of the Mabsagainst PSA (2E9) identifies an epitope on the PSAmolecule still exposed on the PSA complexed to a-antichymotrypsin. This Mab may be relatively insensi-tive to conformational changes of the PSA moleculebecause, on Western blots from SDS-PAGE, Mab 2E9detected both the free uncomplexed PSA molecule andPSA complexed to a1-antichymotrypsin. We found thisepitope to be located on the heavy chain of the cleavedinactive PSA (residues 1-145).

Mab 2H11 also defines an epitope on the PSA mole-cule that remains accessible when PSA is complexed toa1-antichymotrypsin. In all likeithood this epitope isdifferent from that defined by Mab 2E9, the formerbeing sensitive to conformational changes of the PSA

molecule, as demonstrated by the inability of Mab 2H11to detect PSA blotted to poly(vinylidene fluoride) mem-branes after SDS-PAGE of reduced samples.

Mab 5A10 defines a third epitope on the PSA mole-cule, one that is almost inaccessible when PSA is com-plexed to a1-antichymotrypsin. This epitope isalso sen-sitive to conformational changes of PSA, as demon-strated by the inability of Mab 5A10 to detect PSAblotted to poly(vinylidene fluoride) membranes afterSDS-PAGE of reduced samples.

We used Mabs 2E9 and 2H11 in the design of thenoncompetitive assay (assay T) capable of detecting

both uncomplexed PSA and PSA complexed to ar-anti-chymotrypsin because these antibodies define two dif-ferent PSA epitopes, both of which remain accessiblewhen PSA is complexed to a1-antichymotrypsin. Therequired specificity of the assay was confirmed by themost identical dose-response relationships obtained fornoncomplexed PSA and for PSA complexed to a1-anti-chymotrypsin. Thus, the concentration of PSA in thepurified complex between PSA and a1-antichymotrypsincould also be determined with assay T by using thepurified PSA as standard. We therefore used this com-plex to standardize assay C, which, owing to the designof the assay, did not detect purified PSA.

In contrast, assay F was designed solely to detectpurified PSA (and not purified PSA complexed to a-antichymotrypsin). Thus, purified PSA and PSA inseminal plasma displayed almost identical dose-re-sponse relationships when analyzed with this assay,whereas PSA complexed to a1-antichymotrypsin wasvery poorly recognized. The poor recognition of thiscomplex suggests that Mab 5A10 (the solid-phase cap-tore antibody in assay F) defines an epitope on the PSAmolecule that is poorly accessible on PSA complexed toa1-antichymotrypsin, and that Mab 2H11 defines anepitope that is accessible on PSA complexed to a-antichymotrypsin. This interpretation is also consistentwith the finding (see above) that Mab 5A10 was unableto detect PSA complexed to a1-antichymotrypsin inimmunoblotting experiments.

Using assay T, in which purified PSA and PSAcomplexed to a1-antichymotrypsin displayed identicaldose-response relationships, we obtained a low recoveryof the PSA immunoreactivity when purified PSA or PSAin seminal plasma was added to blood serum in vitro.We believe the poor recovery may be due to complexformation between PSA and proteinase inhibitors otherthan a1-antichymotrypsin. Recently, active PSA hasbeen reported to form SDS-stable complexes with a2-macroglobulin when added to humRn serum in vitro(15). The in vitro rate of complex formation betweenPSA and a2-macroglobulin was higher than that of PSAand a1-antichymotrypsin (15). -Macroglobulin isknown to encapsulate the target enzyme at complexformation (33), which is consistent with the previouslyreported loss in PSA immunoreactivity of the PSA and-macroglobulin complex as analyzed by electroimmu-noassay (15). The ability of a2-macroglobulin to encap-

1624 CLINICALCHEMISTRY,Vol.37, No.9, 1991

sulate a target proteinase is also consistent with ourrecovery experiments, where a loss of PSA immunoreac-tivity in blood serum was found with the in vitroaddition of active PSA. This encapsulation property of-macroglobulin is also supported by the absence of lossin PSA immunoreactivity both when PSA was allowedto form a stable complex with a1-antichymotrypsinbefore being added to the serum, and when active PSAwas added to a2-macroglobulin-depleted serum or toserum with inactivated -macroglobulin. These perfor-mance data may also be interpreted as evidence thatassay T is unable to detect PSA complexed to the-macroglobulin, a conclusion also strongly supportedby the closecorrelation of assays T and C in the analysesof 64 patients’ sera.

The findings from the assay of PSA immunoreactivityinpatients’ sera strongly suggest that PSA complexed toa1-antichymotrypsin constitutes the predominant formof immunoreactive PSA in blood plasma in vivo. Evi-dence of this is based on both the analysis of themolecular mass of the PSA immunoreactivity in sera offour patients determined with assay T (which detectsboth noncomplexed PSA and PSA complexed to a1-antichymotrypsin) and with assay C (which solely de-tects PSA complexed to a1-antichymotrypsin), and theclose correlation between results obtained with thesetwo assays in the analyses of 64 patients’ sera. More-over, the results obtained for patients’ sera with assay Tcorrelated very closely to those obtained with the Tan-dem-R PSA kit, which suggests that the Tandem-R PSAassay also detects PSA complexed to a1-antichymo-trypsin as the predominant form of the immunoreactivePSA in patients’ sera. The reported half-life of PSAimmunoreactivity in serum-two to three days(34,35)-is more compatible with that expected for theinactive complex between PSA and a1-antichymo-trypsin than the short half-life expected for an active33-kDa proteinase in vivo.

The results of analysis for the different forms of PSAimmunoreactivity in patients’ sera also indicate thatfree noncomplexed PSA is a minor fraction in serum.This conclusion is based on estimations of the molecularmass of the PSA immunoreactivity in patients’ sera,where assay T identified only a minor immunoreactive25- to 40-kDa fraction, a fraction constituting <30% ofthe total PSA immunoreactivity in four patients’ sera asalso analyzed with assay T. At an identical position inthe chromatogram, assay F (which detects the freenoncomplexed PSA but poorly detects the PSA corn-plexed to a1-antichymotrypsin) identified the major

fraction of the PSA immunoreactivity in the same sera.That the 25- to 40-kDa free (noncomplexed) PSAimmunoreactivity is a minor fraction of the total anti-PSA immunoreactive proteins present in the 64 pa-tients’ sera is also supported by our comparison of theresults of the analyses performed with assays T and F.The findings provide no indication of whether the non-complexed 25- to 40-kDa PSA immunoreactivity consti-tutes the PSA zymogen, active PSA, or the PSA macti-

vated as a result of an internal proteolytic cleavage suchas that demonstrated in a minor fraction of the PSA inseminal fluid (15). This issue will require further study.

This investigation was supported by the Swedish Medical Re-search Council (grant no. B90-13X-7903-4B), the Faculty of Med-icine at the University of Lund, the Research Fund and the CancerResearch Fund at MahnO General Hospital, the Magnus BergvallFoundation, the Crafoord Foundation, the John and AugustaPersaon Foundation, Syskonen SvenasonsFund, the Alfred Oster-lund Foundation, and the Fundacion Federico S.A.

References

1. Lija H, Abrahamason PA. Three predominant proteins se-creted by the human prostate gland. Prostate 1988;12:29-38.2. Lilja H, Laurell C-B. Liquefaction ofcoagulated human semen.Scand J Cliii Lab Invest 1984;44:447-52.3. Lilja H. A kallikrein-like serine protease in prostatic fluidcleaves the predominant seminal vesicle protein. J Clin Invest1985;76:1899-903.4. Lilja H, OldbrungJ, Rannevik G, Laurell C-B. Seminal vesicle-secreted proteins and their reactions during gelation and liquefac-tion of human semen. J Clin Invest 1987;80:281-5.5. McGee RS, Herr JC. Human seminal vesicle-specific antigenduring semen liquefaction. Biol Reprod 1987;37:431-9.6. McGee RS, Herr JC. Human seminal vesicle-8peciflc antigenisa substrate for prostate-specific antigen (or P-30). Biol Reprod1988;39:499-510.7. Lilja H, Abrahaxnsson PA, Lundwall A. Semenogelin, thepredominant protein in human semen: primary structure andidentificationof closely related proteins in the male accessory sexglands and on the spermatozoa. J Biol Chem 1989;264:1894-900.8. Wang MC, Valenzuela LA, Murphy GP, Chu TM. Purificationof a human prostate specific antigen. Invest Urol 1979;17:159-63.9. Hara M, Kimura H. Two prostate-specific antigens, semino-protein and p-microseminoprotein. J Lab Clin Med 1989;113:541-8.10. Graves HCB, Sensabaugh GF, Blake ET. Postcoital detectionofa male-specific semen protein:applicationto the investigation ofrape. N Engi J Med 1985;312:338-43.11. Lundwall A, Lilja H. Molecular cloning of human prostate-specific antigen cDNA. FEBS Lett 1987214:317-22.12. Watt KWK, Lee P.,JL, Timkulu TM, Chan W-P, Loor R.Human prostate-specific antigen: structural and functional simi-larity with serine proteases. Proc Natl Acad Sd USA 1986;83:3166-70.13. Schaller J, Akiyama K, Tsuda R, Hara M, Marti T, Rickli E.Isolation, characterization and amino-acid sequence of ‘-semino-protein, a glycoprotein from human seminal plasma. Eur J Bio-chem 1987;170:111-20.14. Akiyama K, Nakamura T, Iwanaga 5, Hara M. The chymo-trypsin-like activity of human prostate-specific antigen, y-semino-protein. FEBS Lett 1987;225:168-72.15. Christenason A, Laurell C-B, Lilja H. Enzymatic activity ofthe prostate-specific antigen and its reactions with extracellularserine proteinase inhibitors. Eur J Biochem 1990;194:755-63.16. Papsidero LD, Wang MC, Valenzuela LA, Murphy GP, ChuTM. A prostate antigen in sera of prostatic cancer patients. CancerRes 1980;40:2428-32.17. Duffy MJ. New cancer markers. Ann Clin Biochem 1989;26:379-87.18. Brawer MK, Lange PH. Prostate-specific antigen in manAge-ment of proatatic carcinoma. Urology 1989;5(Suppl):11-6.19. Allthan H, Stenman UI!. Falsely low results obtained with theHybritech Tandem’-R PSA assay [Tech BriefI. Cliii Chem1988;34:2152.20. Laurell C-B, McKay EJ. Electroimmunoassay. Methods En-zymol 1981;73:339-.69.21. Chen BJ, Yuan Al, Wang D, Feinman RD. Effect of methyl-amine on the reaction of a2-macroglobulin with enzymes. Bio-chemistry 1990;29:3361-5.22. Lovgren T, Heminila I, Pettersson K, Eskola JU, Bertoff E.Determination of hormones by time-resolved fluoroimmunoassay.Talanta 1984;31:909-16.23. Hemmilfi I, Dakubu S, Mukkala VM, Siitari H, Lovgren T.

CLINICALCHEMISTRY,Vol.37, No.9, 1991 1625

Europium as a label in time-resolved immunofluorometric assays.Anal Biochem 1984;137:335-.43.24. Matikainen MT, Terho P. Immunochemical analysis of anti-genic determinnta of Chiamydia trachomaas by monoclonal an-tibodies. J Gen Microbiol 1983;129:2343-50.25. Fazekas de St GrothS, Scheidegger D. Production of monoclo-nal antibodies: strategy and tactics. J immunol Methods 1980;35:1-21.26. Lovgren T, Hemmilfi I, Pettersaon K, Halonen P. Time-resolved fluorometry in immunoassay. In: Collins WP, ed. Alter-native immunoassays. New York: John Wiley & Sons, Ltd.,1985:203-17.27. Staszewski R. Cloning by limiting dilution an improvedestimate that an interesting culture is monoclonal. Yale J BiolMed 1984;57:865-8.28. Olafsaon I, Gudmunsson G, Abra1imon M, Jenason 0,Grubb A. The aminoterminal portion of cerebroepinal fluid cysta-tin C in hereditary cystatin C amyloid angiopathy is nottruncated:direct sequence analysis from agarose electropherograms. Scand JClin Lab Invest 1990;50:85-93.29 Jeppason J-O, Laurell C-B, Franz#{233}nB. Agaroee electrophore-

sin. Clin Chem 1979;25:629-38.30. Laemmli UK. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)1970;227:680-5.31. Blobel G, Dobberstein B. Transfer of proteins across mem-branes. J Cell Biol 1975;67:835-51.32. Burnette WN. ‘Western blotting”: electrophoretic transfer ofproteins from sodium dodecyl sulfate-polyacrylamide gels to un-modified nitrocelluloee and radiographic detectionwith antibodyand radioiodinated Protein A. Anal Biochem 1981;112:195-203.33. Sottrup.Jenaen L. a-Macroglobulins: structure, shape, andmechanism of proteunase complex formation. J Biol Chem1989264:11539-42.34. Stamey TA, Yang N, Hay AR, McNeal JE, Freiha PS, Red-wine E. Prostate-specific antigen as a serum marker foradenocar-cinoina of the prostate. N Engl J Med 1987;317:909-16.35. Oesterhng JE, Chan DW, Epstein JI, et al. Prostate-specificantigen in the preoperative and postoperative evaluation of local-ized prostatic cancer treated with radical prostatectomy. J Urol1988;139:766-72.