Znt. J. Cancer: 55,590-597 (1993) Publication of the International Union Against Cancer Publication de I'Union lnternationale Contre le Cancer 0 1993 Wiley-Liss, Inc.

EVALUATION OF PAP AND PSA GENE EXPRESSION IN PROSTATIC

ANALYSES, IN SITU HYBRIDIZATION AND IMMUNOHISTOCHEMICAL STAININGS WITH MONOCLONAL AND BISPECIFIC ANTIBODIES

HYPERPLASIA AND PROSTATIC CARCINOMA USING NORTHERN-BLOT

Leena HAKALAHTI~, Pirkko vIHK01x3, Pirkko HENTTU', Helena AUTIO-HARMAINEN~, Ylermi SOIN$ and Reijo VIHKO' 'Biocenter and Department of Clinical Chemistiy, University of Oulu, Kajaanintie 50, FIN-90220 Oulu; and 2Depaament of Pathology, University of Oulu, Kajaanintie 520, FIN-90220 Oulu, Finland.

In this report we have investigated levels of prostatic acid phosphatase (PAP) and prostate-specific antigen (PSA) gene expression in prostatic carcinoma (Ca) and benign prostatic hyperplaria (BPH) specimens. Northern-blot analyses of total prostatic mRNA indicated that there was a tendency towards lower amounts of PAP mRNA and PSA mRNA in the Ca specimens than in the BPH specimens, although, because of the great variation in the expression levels of both mRNAs, these differences were not statistically significant. In situ hybridization analyses clearly showed that both PAP and PSA mRNAs were confined to the columnar epithelial cells and that stromal cells were devoid of these mRNAs. In addition, PAP and PSA mRNAs were more abundant in BPH tissue than in adjacent Ca tissue within the same specimen. The levels of PAP and PSA enzymes were analyzed immunohistochemically using a bispecific anti- body having high affinity for both PAP and PSA, and the results were compared with those obtained using monoclonal anti-PAP and anti-PSA antibodies. All 3 antibodies stained only epithelial cells and BPH tissue consistently gave more intense staining than Ca tissue. Furthermore, the anti-PSA and the bispecific anti-PAP-PSA antibodies stained well or moderately differenti- ated Ca tissues more strongly than poorly differentiated Ca tissues. No PSA staining was detected in 3 and no PAP staining in 5 of the moderately or poorly differentiated carcinomas (grades II or 111). Our results show that, in comparison with BPH tissue, prostatic Ca tissue is associated with significantly lower levels of mRNAs coding for the prostatic marker enzymes PAP and PSA, as well as with lower concentrations of these enzymes. Furthermore, dedifferentiation of prostate Ca is associated with a decrease in the level of intraprostatic PSA. o I993 Wiley-Liss, Znc.

Measurement of serum concentrations of prostatic marker enzymes has gained a firm position in the clinical detection, diagnosis and management of prostate cancer. The 2 most widely used prostatic marker enzymes are prostatic acid phosphatase (PAP) and prostate-specific antigen (PSA), of which PSA is a serine protease acting on seminal coagulates (Lilja, 1985) and PAP an acid phosphatase (Vihko et al., 1978) with an undetermined function. Normally, both enzymes are secreted from the prostatic epithelial cells into the seminal fluid, but in patients with prostate cancer they leak into the circulation, where they can be detected by means of immuno- logical assays (Heller, 1987; Stamey et al., 1987; Vihko et al., 1990). In recent years, serum PSA measurements have super- seded PAP measurements in the clinical management of prostate Ca (Oesterling, 1991). The fact that the amount of PSA in the circulation increases with the volume of carcinoma- tous tissue and with advancing degree of dedifferentiation (Stamey et al., 1987) contributes to the usefulness and informa- tive value of serum PSA measurements. However, although PSA is a sensitive marker for prostate cancer, it is not specific to Ca, and distinguishing BPH from Ca on the basis of serum PSA concentrations is difficult, as the observed ranges of serum PSA concentrations in patients with BPH and prostate Ca overlap (Stamey et al., 1987).

In comparison with the immense amount of literature de- scribing changes in the serum concentrations of PAP and PSA

in patients with prostatic disorders, and the effects of various treatments on these concentrations, relatively little is known about the changes occurring in expression of the correspond- ing genes within the prostate. We have previously investigated the expression of PAP (Solin et al., 1990) and PSA (Henttu et al., 1990) mRNAs in a limited number of prostatic tissue speci- mens and found that, whereas there is a tendency towards lower amounts of PAP mRNA in Ca specimens, no significant differ- ences are detectable in the relative abundance of PSA mRNA in different types of prostatic tissue. We now report our findings on the expression of PAP and PSA genes in a larger number of prostate cancer and BPH tissue specimens, as analyzed by Northern blotting, in situ hybridization, and immunohistochem- istry with monoclonal and bispecific antibodies. Our results show that, in comparison with BPH tissue, prostate Ca tissue contains less PAP and PSA and that these lowered concentra- tions of mature protein are the result of decreased amounts of transcripts originating from the PAP and PSA genes.

MATERIAL AND METHODS Tissue collection and preparation

Fresh prostatic tissue was obtained from patients undergo- ing surgery for prostatic hyperplasia or adenocarcinoma. Tissue specimens that were later used for RNA isolation were cut into small pieces immediately after removal and stored in liquid nitrogen. The histopathological status of these speci- mens was determined from separate but adjacent samples. For in situ hybridization analyses, histological sections of prostatic tissue samples from pathological files were prepared as re- ported by Autio-Harmainen et al. (1991). For immunohisto- chemical analyses, prostatic samples were fixed in 4% forma- lin, embedded in paraffin and subsequently cut into 5-Fm sections. These sections were mounted on glycerol-treated glass slides, deparaffinized and rehydrated before incubation with the antibody. The tissue specimens analyzed by Northern- blot analyses, in situ hybridization and immunostaining were derived from different patients.

Probes The PAP and PSA probes used in the Northern-blot

analyses were as described by Henttu et al. (1992). The glyceraldehyde-3-phosphate dehydrogenase cDNA was as pre- viously described (Fort et al., 1985). The PAP and GADPH probes were labelled by nick translation, using [ C X - ~ ~ P I ~ C T P (Amersham, Aylesbury, UK), and the PSA oligonucleotide was labelled with T4 polynucleotide kinase and [Y-~~PIATP (Amersham), Before use, these probes were purified by chromatography through Nick columns (Pharmacia LKB,

3To whom correspondence and requests for reprints should be addressed.

Received: June 8,1993 and in revised form July 12,1993.

PAP AND PSA GENE EXPRESSION IN PROSTATIC TUMORS 591

Uppsala, Sweden). To prepare the labelled sense and anti- sense RNA probes for in situ hybridization experiments, PAP cDNA PAP-1006A was cloned into a pSP72 vector (Promega, Madison, WI) and PSA cDNA PSA-20 into a pGEM-3Zf( +) vector (Promega). Sense and anti-sense riboprobes were generated from these plasmids and labelled with either [cx-~~SICTP or [(U-~~SIUTP (Amersham), using a commercial transcription kit (Promega).

Isolation and analysis of RhLA For the extraction of total cellular RNA, a piece of each

prostatic tissue specimen was homogenized in guanidinium isothiocyanate solution. Total RNA was then purified by centrifugation through a cesium chloride cushion (Davis et al., 1986). Agarose gel electrophoresis of RNAs was performed as reported previously (Henttu et al., 1992). For Northern-blot analysis, 90 p,g of total RNA isolated from the prostatic tissue specimens were subjected to electrophoresis and transferred to Hybond Nylon membranes (Amersham) by capillary blot- ting in 20 x SSPE (1 x SSPE is 180 mmol/l NaCl, 10 mmol/l sodium phosphate, pH 7.4,l mmol/l Na2 EDTA), according to the manufacturer’s recommendations. The RNAs were cova- lently bound to the membrane by UV irradiation. These filters were hybridized with PAP, PSA and GADPH probes as described (Henttu et al., 1992) and exposed to Kodak X A R J films (Eastman Kodak, Rochester, NY) at -70°C for 1 to 14 days. The X-ray films were analyzed with a laser scanner densitometer (Molecular Dynamics, Sunnyvale, CA), equipped with a computer system. All steady-state mRNA levels were corrected for differences in loading, or transfer of the RNA samples to the filters, by correlating the values obtained for specific mRNAs with the expression of GADPH mRNA in the sample. Before re-probing, the filters were stripped of their previous radioactivity by soaking for 2 hr at 65°C in a solution containing 5 mmol/l Tris-HCI, pH 8.0, 2 mmol/l Naz EDTA, 0.1 x Denhardt’s solution (1 X Denhardt’s solution is 0.02% BSA, 0.02% polyvinylpyrrolidone and 0.02% Ficoll).

In situ hybridization analyses Before hybridization, histological sections of prostatic tissue

specimens were treated as described (Autio-Harmainen et al., 1991). Pre-treated sections were incubated overnight at 50°C with hybridization mixture (10 mM DTT, 10 mM Tris-HCI, pH 8.0, 10 mM NaP04, 5 mM Naz EDTA, 0.3 M NaCI, 1 mg yeast tRNA/ml, 50% (v/v) deionized formamide, 10% (v/v) dextran sulfate and 1 x Denhardt’s solution) containing approxi- mately 2.8 x 106 cpm of heat-denatured RNA probe in 40 ml of hybridization buffer per section. After hybridization, the sections were washed as described (Autio-Harmainen et al., 1991). Autoradiography was performed by dipping the slides into Kodak NTB-3 nuclear track emulsion (Eastman Kodak) diluted 1:1 with 1% glycerol in water. After exposure for 7 to 15 days, the slides were developed in Kodak D19 developer (Eastman Kodak) at room temperature for 2 to 4 min, rinsed in acetic acid and fixed for 5 min at room temperature. The sections were counterstained with hematoxylin and eosin.

The hybridization signal was regarded as positive only if the accumulation of grains in individual cells exceeded the back- ground count. This usually meant 10 or more grains per cell. The signal was unconvincing (-) when indistinguishable from the background, slight (+) when the signal was slightly increased in comparison with the background or when only a few cells contained grains, moderate (+ +) when a moderate number of cells expressed a clearly defined signal and strong (+ + + / + + + +) when the cells were covered with grains or when most of the epithelial cells contained grains. The signal intensity in the hyperplastic regions of carcinoma samples were analyzed separately from the signal intensity in the carcinoma- tous regions.

Case number Diagnosis

Relative Relative abundance of abundance of PAP mRNA PSA mRNA

1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38

BPH BPH BPH

BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH BPH

mean * standard deviation Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade 111 mean t standard deviation

305 111 244

56 102

1091 540 329 381

302

217 346 44 1 467 400 374 329

439 896

1090

826 777

457 * 298 424 487 328 188 353 107 100 408

6 106 605 383 113

1044 91

498 327 * 263

67 38 87

141 32

126 91 44 71 71 86 50

161 138 in0 123 129 124 85 64 61 66

89 ? 36 106 54 95

106 111

9

86 4 9

73

ion

85 50

115 13 29

65 t 41

Maintenance of stock cultures All cell lines were cultured in DMEM supplemented with

10% FCS, 100 IU/ml penicillin, 100 mg/ml streptomycin and 2 mM L-glutamine. Cell culture reagents were purchased from GIBCO (Paisley, UK), except for FCS which was obtained from Flow (Irvine, UK). Preparation of bispec@c antibodies

A mouse hybridoma cell line producing MAbs to PAP (clone T411E15) was generated according to a protocol described previously (Hoyhtya et al., 1987), except that the myeloma cell line X63-Ag 8.653 was used in fusion. T411E15 cells were adapted to growing in increasing concentrations of 8-azagua- nine (from 1 to 30 mg/ml). A subclone of the original line, T411E15 B2, was obtained by limiting dilution and found to be sensitive to hypoxanthine-aminopteridine-thymidine (HAT). A hybridoma cell line producing MAbs to PSA, 7E7 (Vihko et al., 1990), was tested for its sensitivity to actinomycin D, and fewer than 1% of the 7E7 cells were found to survive in DMEM medium containing 1 mgirnl actinomycin D. Prior to fusion, 7E7 cells were treated with this concentration of the drug for 1 hr 15 min at 37°C and washed with DMEM; 2 X lo7 7E7 cells were fused with an equal number of T411E15 B2 cells

592

1200

1100

1000

900

800

700

6 0 0

500

400

300

200

100

0

T

PAP m R N A

0 . 0 . . . 0 .

. 0

ooooo . . A-

C A BPH

200

150

100

5 0

0

HAKALAHTI ETAL.

PSA m R N A A-

T

V V T

< V T V V

V V

T T V

V v v V I C A BPH

FIGURE 1 - Distribution of relative amounts of PAP and PSA mRNAs in BPH and prostate Ca specimens. The levels of both PAP and PSA mRNAs in specimen 29 were assigned a value of 100 and the amounts in the other specimens were correlated to this. However, values for the relative abundance of PAP and PSA mRNAs are not comparable. The arrowheads point to the mean value for each category.

TABLE I1 - SEMI-QUANTITATIVE ANALYSIS OF IN SITU HYBRIDIZATION SIGNALS IN PROSTATIC TISSUE SAMPLES

Abundance Abundance ot PSA of PAP

of BPHICA of BPHJCA mRNA in areas mRNA in area? Diagnosi4 Case

number

1 2 3 4 5 6 7 8 9

10 11 12 13

BPH ++++ BPH ++++ BPH +++ BPH ++++ BPH ++

Adenocarcinoma grade I + + + + / + + + Adenocarcinoma grade I + + + + / + + + Adenocarcinoma grade I + + + / + + Adenocarcinoma grade I + + + + / + + Adenocarcinoma grade I + + + + / - Adenocarcinoma grade I1 + + +I + + Adenocarcinoma grade I11 + + + / + + + Adenocarcinoma grade I11 + + + + / + + +

14 Adenocarcinoma grade I11 + + + / + + + 15 Adenocarcinoma grade 111 + + + / + +

+++ +++ +++ ++++ ++

+++I++ ++/++ ++/++ ++I+

+++I+ ++/+

++I++ ++I++

+/+ ++I+

Abundance of specific mRNAs: - not detectable, + slight, + + moderate, + + + strong and + + + + very strong.

in 1 ml of 45% PEG, 10% DMSO and cloned on microtiter plates; finally, fused cells were selected for in HAT-containing medium. Cells demonstrating activity against both PAP and PSA, as tested with specific radioimmunoassays (Vihko et al., 1980; 1990), were further cloned with dilution and assayed for bispecific activity with solid-phase radioimmunoassay. Briefly, 0.2 mg of PSA was coated onto microtiter wells, then wells were washed with PBS and subsequently incubated with PBS containing 1% BSA for 2 hr. The wells were again washed with

FIGURE 2 -Localization of PAP and PSA mRNAs in BPH tissue. BPH sections from adenocarcinoma cases (a, case 15; b, case 1; and c, case 10 in Table 11) were hybridized with the PSA-probe (a, bar = 82 pm) and with the PAP probe (b, bar = 200 pm; c, bar = 50 km) as described in the text. The arrows point to the location of unlabelled myoepithelial cells beneath labelled columnar epithelial cells.

PBS, and growth medium from fused cells was added and incubated overnight at room temperature. Following washes with PBS, 1251-labelled PAP (Vihko et al., 1978) was added and incubated overnight. The amount of bound radioactivity was measured. Four clones out of 384 exhibited continuous produc- tion of antibodies against both PAP and PSA. From these clones, 16 stable cell lines were cloned.

PAP AND PSA GENE EXPRESSION IN PROSTATIC TUMORS 593

FIGURE 3 - In situ hybridization analysis of well-differentiated, grade-I Ca (a, case 13; b, case 7 in Table 11). The tissue sections were hybridized with the PSA-probe (a) or with the PAP-probe (b). Bars = 200 prn.

Purification of bispecijic antibodies Three fusion cell lines, G1E8, G1F8 and D9H1, of the

original 16 lines were selected for purification of antibodies, because their growth media contained the highest titers of antibodies recognizing both PAP and PSA. Their growth media were collected, then cells were removed and medium was concentrated 30-fold by ultrafiltration. The concentrate was adjusted to pH 8 and applied to a column of Protein-A- Sepharose (Pharmacia). MAbs bound to the column were eluted with 0.1 M sodium citrate buffer, pH 6. The eluate containing the bispecific antibodies was applied subsequently to a column containing PAP bound to CNBr-activated AH- Sepharose 4B-matrix (Pharmacia). The column was eluted step-wise with 0.02 M PBS, pH 7.4; 0.1 M sodium acetate buffer, pH 4.5, and 0.05 M HCI, pH 1-2.

The fraction containing antibodies recognizing PAP was further purified on a PSA-Sepharose 4B column, and the bispecific antibodies were eluted as above. SDS-PAGE analy- sis confirmed that the purified antibodies were homogeneous. The dissociation constant (K,) of the bispecific antibodies for PAP varied between M (GlE8) and 10-l' M (D9H1), while the KDs for PSAwere M (GIE8, GlF8) and M (D9H1). All the antibodies were found to be of subclass IgGI. The bispecific antibodies were incubated with leukocyte, liver, kidney and pancreas extracts, but none of them reacted with antigens present in these tissues.

FIGURE 4 -In situ hybridization of poorly differentiated, grade- I11 Ca (case 15 in Table 11). Tissue sections were hybridized with the PSA probe (a) or with the PAP probe (b). Arrows point to occasional cells exhibiting PAP or PSA mRNAs. Bars = 82 pm.

Immunohistochemistry Prostatic tissue sections were incubated in a solution contain-

ing 1 mg/ml of the antibodies. Murine immunoglobulin was used as a non-specific control antibody. In the immunostain- ing, the avidin-biotin complex technique was used (reagents from Vector, Burlingame, CA). The color was developed with diaminobenzidine, whereafter the tissue sections were lightly counterstained with hematoxylin and mounted. The propor- tion of stained tumor cells within each specimen was deter- mined and the specimens were assigned to 5 groups: 0% of cells staining, -; less than 25% of cells staining, +; 25-50% of cells staining, ++; 50-75% of cells staining, +++ and 75-100% of cells staining, ++++. The intensity of staining was evaluated visually and each specimen was assigned to one of following categories: no staining -; low intensity +; moder- ate intensity + +; high intensity + + + and very high intensity ++++. Statistical analyses

The statistical significances of differences of means calcu- lated for the abundance of mRNAs in hyperplastic and carcinomatous prostatic tissue specimens were analyzed by ANOVA. To analyze the correlation between the levels of PAP and PSA mRNAs in the tissue specimens, Pearson's correlation coefficients were calculated. In the immunohisto- chemical analyses, the Mann-Whitney U-test was used to determine differences between unpaired samples and the Wilcoxon signed rank test was used for paired samples.

594 HAKALAHTI ETAL.

TABLE 111 - IMMUNOSTAINING OF HUMAN PROSTATIC CANCER SPECIMENS WITH ANTI-PSA, ANTI-PAP AND ANTI-PAP-PSA ANTIBODIES

Staining with Staining with Staining with

staining intensihi staining intensity staining intensity

anti-PSA antibodies. anti-PAP antibodies. anti-PAP-PSA antibodies Diagnosis Stained proportion/ Stained proportion/ Stained proportion/

Case number

1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinorna grade I Adenocarcinoma grade I Adenocarcinoma grade I Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinorna grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinoma grade I1 Adenocarcinorna grade I1 Adenocarcinoma grade I1 Adenocarcinorna grade I1 Adenocarcinorna grade I11 Adenocarcinoma grade I11 Adenocarcinoma grade 111 Adenocarcinoma grade I11 Adenocarcinoma grade I11 Adenocarcinoma grade 111

++++/+++ ++++/+++ +++/+++

++++/+++ +++/+++

++/++ ++++/+++ ++++/++++ ++++/++++ ++++/+++ +++/+++ ++++I+++ ++++/+++

- / - ++++/+++ ++++/++++ ++++/++++ ++++/+++ ++++/++ ++/++ ++++/+++ ++++/+++ ++++/++++

+I+ - / - - / - ++++/+++

++++/++ ++++/++ +++/++

++++/++ +/+ +/+

++/+ ++++/++ +++/++ ++/+

++++/++++ - / - ++ /+

+++/+ ++++/+++ ++++/+++ ++++/++ ++++I+++ ++++/+++

- / - ++++/++ ++++/+++ ++++/++ -/- - / - - I - ++++/++

++++/+++ ++++/+++ +++/+++ ++++/++++ +++/++ ++/++ ++++/++++

++++/++++ +++/++ ++++/+++ +++/+++ ++++/+++ ++/++

+++/++ ++++/++ ++++/+++ ++++/+++ ++++/++++ ++++/+++ +++/+++ ++++/++++

+++/++ ++++/++

+ /+ - / - - / - ++++/+++

Stained proportion: -, all cells negative; +, < 25% of cells positive; + +, 25-50% of cells positive; +++, 50-75% of cells positive; +++ +, 75-100% of cells positive. Staining intensity: -, .no staining; +, low intensity; ++, moderate intensity; +++, high intensity; ++++, very high intensity.

RESULTS Analysis of PAP and PSA mRNA amounts by Northern-blot hybridization

Northern-blot hybridization was utilized to analyze the expression of PAP and PSA mRNAs in 22 tissue specimens of prostatic hyperplasia and 16 tissue specimens of prostatic carcinoma, of which 6 were grade I, 9 grade I1 and one grade 111. All RNA samples were first hybridized with the GADPH probe and only those demonstrating a detectable signal for GADPH mRNA were included in this study. The level of GADPH mRNA was used to correct for uneven loading or transfer of RNA samples. The abundance of PAP and PSA transcripts was assigned a value of 100 in specimen 29 (Table I). This sample was included in all filters and the expression levels of PAP and PSA mRNAs in the other samples were compared with the expression level in specimen 29.

As shown in Table I and Figure 1, the expression of PAP mRNA ranged from 56 to 1,091 in the BPH samples and from 6 to 1,044 in the Ca samples. However, the difference between the means of PAP expression in these groups, 327 f 263 (SD) for the Ca samples and 457 f 298 for the BPH samples, was not statistically significant ( p = 0.172). The amounts of PAP mRNA in the well-differentiated (grade I) and moderately differentiated carcinomas (grade 11) did not differ, being 315 5 143 and 317 & 336 respectively for grade-I and grade-I1 carcinomas. The expression levels of PSA mRNA in the BPH and CA samples did not differ significantly (p = 0.058), although there was a tendency towards higher values in the BPH group (mean f SD, 89 ? 37) than in the Ca group (mean 65 f 41). As in the case of PAP mRNA, the level of PSA expression varied greatly in both groups, from 32 to 161 in the BPH specimens and from 4 to 115 in the Ca specimens. There was a tendency towards less PSA mRNA in the moderately

differentiated grade-I1 carcinomas (mean 5 SD, 59 * 42) than in the well-differentiated grade-I carcinomas (mean f SD 81 ? 41), although the difference was not significant because of the great variation in both subgroups. Interestingly, the level of PSA mRNA expression in the grade-I carcinomas was closer to the expression level observed in the samples containing BPH (mean f SD, 89 +- 36). There was a slight correlation (Pearson’s correlation coefficient 0 . 5 2 1 , ~ < 0.05) between the expression of PAP and PSA mRNAs in the Ca specimens, suggesting similarities in the regulation of these genes.

Analysis of PAP and PSA mRNA amounts by in situ hybridization

In situ hybridization analyses were performed using PAP and PSA probes with 5 BPH specimens and 10 Ca specimens of which 5 were well-differentiated carcinomas (grade I) and 5 moderately or poorly differentiated carcinomas (grades I1 or 111). The carcinoma samples also contained some areas of benign hyperplasia that were evaluated separately for expression of the PAP and PSA transcripts. The results of the in situ hybrid- ization analyses are given in Table 11. In the BPH specimens, both PAP and PSA mRNAs were found only in the glandular and ductal epithelial cells, as shown in Figure 2. Myoepithelial cells lining the columnar epithelial layer were devoid of both signals (Fig 2a, c). Where present, the signals for both mRNAs were strong and no quantitative differences were detected.

PSA mRNA was clearly detectable in all of the Ca speci- mens, except for one grade-I specimen, and no difference was detected in the expression level between well and poorly differentiated carcinomas. However, in the areas of BPH, there was significantly more PSA mRNA than in the areas of Ca of the same sample ( p < 0.01). PAP mRNA was detectable in all Ca specimens, but the intensity of the signal was lower

PAP AND PSA GENE EXPRESSION IN PROSTATIC TUMORS

than that of PSA mRNA. Areas of BPH contained more PAP mRNA than areas of Ca within the same specimen (p < 0.05), although the difference was less pronounced than in the case of PSA mRNA. Figure 3 depicts results from in situ hybridiza- tion analysis of a well-differentiated carcinoma specimen and Figure 4 shows results from a poorly differentiated specimen.

Analyses of PAP and PSA enzyme amounts using immunohistochemical staining with monoclonal and bispecijic antibodies

Twenty-seven formalin-fixed, paraffin-embedded prostatic Ca tissue specimens, of which 6 were grade-I, 15 grade-I1 and 6 grade-111 carcinomas, were treated and sectioned as described above and stained with anti-PAP (T411E15), anti-PSA (7E7) and bispecific anti-PAP-PSA antibodies (GlE8). The results of these experiments are given in Table 111.

Both PAP and PSA were found only in prostatic epithelial cells. Twenty-two, 24 and 25 of the 27 Ca specimens gave positive immunohistochemical results with the anti-PAP, anti- PSA and anti-PAP-PSA antibodies, respectively. There were no significant differences either in the proportion of cells staining with any of the antibodies, or in the intensity of the staining between grade-I and grade-I1 tumors. In contrast, the grade-I11 tumors demonstrated significantly lower intensities of staining with the anti-PSA (p < 0.05) and the anti-PAP- PSA (p < 0.01) antibodies compared with grade 1-11 tumors. Figure 5 depicts the results of immunostaining with the bispecific antibodies (GlE8) of tissue sections containing well-differentiated and moderately differentiated Ca tissue from the same patient (number 21 in Table 111). The staining intensity was greater in the well-differentiated regions than in the moderately differentiated regions. All the Ca specimens studied contained some areas of BPH tissue. Regardless of the antibodies used, these BPH areas showed consistently higher staining intensities and contained more stained cells than the adjacent carcinomatous areas, as demonstrated in Figure 6.

The results obtained with the anti-PSA and anti-PAP-PSA antibodies were similar, both in respect to the proportion of cells staining and the intensity of the staining. The bispecific anti-PAP-PSA antibodies stained more cells (p < 0.05) and the intensity of staining was higher (p < 0.01) than with the anti-PAP antibody. The proportion of immunostained cells was the same with anti-PAP and anti-PSA antibodies. The bispecific anti-PAP-PSA antibodies stained all the specimens that were positive for either one of the antigens. However, 2 poorly differentiated carcinomas (specimens 25 and 26 in Table 111) that were not stained with the monospecific anti- bodies were also negative when analysed with the bispecific antibodies.

DISCUSSION

In this report, expression of the genes coding for 2 prostatic marker enzymes of clinical importance was studied with 3 different techniques in BPH and prostatic Ca tissue specimens. Northern-blot analysis of total prostatic mRNA indicated that there was a tendency towards lower amounts of PAP and PSA mRNAs in the Ca specimens than in BPH specimens, although there was great variation in the expression levels of both mRNAs. However, in this technique, the epithelial or stromal origin of mRNAs cannot be selected for and the presence of different proportions of these tissue components in different specimens hampers the interpretation of the data. In contrast, in situ hybridization allows detection of the specific cells expressing the mRNA of interest. Our results from in situ hybridization analyses demonstrated that both PAP and PSA mRNAs were more abundant in BPH tissue than in Ca tissue within the same specimen. Furthermore, the amounts of PAP

595

FIGURE 5 - Immunohistochemical staining of histological sec- tions of a prostatic Ca specimen. Histological sections of a Ca specimen (case 21 in Table 111) were stained with the bispecific anti-PAP-PSA antibody (GlE8). (a) A region containing well- differentiated Ca; (b) a region containing moderately differenti- ated carcinoma in the same specimen. (b) A strongly staining benign prostatic duct is shown on the left. Bars = 100 pm.

and PSA enzymes were analysed by immunohistochemistry, using monoclonal anti-PAP and anti-PSA antibodies and a bispecific antibody having high affinity for both antigens. All 3 antibodies showed more pronounced staining of BPH tissue than of Ca tissue. Furthermore, the anti-PSA antibodies and the bispecific anti-PAP-PSA antibodies stained well-differenti- ated Ca tissues more strongly than moderately or poorly differentiated Ca tissues.

The increased accumulation of both PAP and PSA in the circulation of patients with prostate Ca is well documented (Vihko et al., 1980, 1990; Heller, 1987; Stamey et al., 1987; Oesterling, 1991), but the changes occurring within the pros- tate appear to be opposite and more subtle. Our results, showing decreased expression of prostatic marker enzymes in Ca compared with BPH, are in agreement with those of Qiu et al. (1990), who reported that BPH tissue contains more PSA mRNA than Ca tissue, and that poorly differentiated carcino- mas contain less PSA mRNA and protein than well-differenti- ated ones. Similarly, Gallee et al. (1990) found that PSA immunoreactivity was lower in areas of Ca than in adjacent sections of BPH in the same specimen. In their analysis of intraprostatic PSA concentrations, Pretlow etal. (1991) reached similar conclusions. In contrast, others have failed to detect a direct or uniform connection between PSA staining and tumor differentiation (Nadji et al., 1981; Sinha et aL, 1986). Previ- ously, Mori and Wakasugi (1985) have reported that neoplastic

596 HAKALAHTI ETAL.

FIGURE 6 - Difference in staining intensity between benign and malignant epithelia stained with the bispecific anti-PAP-PSA antibody (GlE8). Histological sections of a tissue specimen containing well-differentiated Ca adjacent to BPH tissue (case 7 in Table 111) were stained. More intensivelv stained. dilated benign glands are shown on the left and well-differentiated carcinomatous glands on the right. Bar = 200 pm.

- -

prostatic tissue contains less PAP than does BPH tissue, and that the secretory mechanisms behind PAP appear to be different in well and poorly differentiated carcinomas. Thus, the increased routing of prostatic secretory proteins towards the circulatory system is most likely the consequence either of the loss of polarity in cancer cells or of deformed prostatic acinar-ductal structures, or both. Increased tissue concentra- tions of PAP and PSA do not account for the accumulation of these prostatic secretory enzymes in the serum.

The expression of PAP and PSA genes is under complex control and, as shown in the prostatic carcinoma cell line LNCaP, androgens (Henttu et al., 1992), growth factors (Henttu and Vihko, 1993) and cellular kinases (data not shown) all affect the amount of mRNA originating from the genes as well as the amount of secreted protein. Although the mechanisms behind the development of prostate Ca and BPH remain largely uncharacterized, these processes are involved in changes in local production of and responsivity to growth factors, such as acidic and basic fibroblast growth factors, epidermal growth factor and transforming growth factors a, PI, pz and P3 (reviewed in Davies and Eaton, 1991). Moreover, expression of the mRNA coding for the androgen receptor, the cellular mediator of androgen action, is decreased in prostate Ca tissue compared with BPH tissue (data not shown). As these growth factors (Henttu and Vihko, 1993) and androgens (Henttu etal., 1992) modify the expression of PAP and PSA genes in LNCaP cells, they are also likely to control the amounts of PAP and PSA gene expression inside prostatic tissue. Thus alterations in the amounts of androgen receptor, growth factors and their receptors can contribute to the observed decrease in the expression of these prostatic marker enzymes. Despite some similarities in the regulation of PAP and PSA genes, there also

appear to be some differences. This was suggested by the fact that some of the specimens (eg. those from patients 28,32 and 37 in Table I, patient 10 in Table I1 and patients 14 and 20 in Table 111) expressed only a very low level of one enzyme or its mRNA, or none at all, whereas the other enzyme or its mRNA was expressed at a relatively high level.

The mechanisms behind dedifferentiation of prostatic can- cer cells remain largely unresolved. Decreased amounts of immunoreactive PSA in poorly differentiated carcinomas sug- gest that mechanisms which regulate PSA gene expression may also be involved in the dedifferentiation process. Furthermore, the fact that the expression level of the PAP gene did not differ between well and poorly differentiated carcinomas suggests that some fundamental difference exists in the regulatory mechanisms of these 2 genes. For example, these differences may reflect varying contributions of stromal-epithelial interac- tions to the expression of PAP and PSA genes.

We developed bispecific antibodies that recognize both PAP and PSA with high affinity by fusing 2 hybridoma cell lines producing monoclonal anti-PAP and anti-PSA antibodies. The performance of the bispecific antibodies in immunohistochemi- cal analysis of prostatic tissue specimens was determined and compared with that of the parent antibodies. The staining pattern obtained with the bispecific antibody (GlE7) was similar to that of the parent PSA antibodies as regards the proportion of stained cells and the intensity of staining. This may reflect the relatively high amount of PSA synthesized by prostatic cancer cells, as previously shown with the human prostatic carcinoma cell line LNCaP, which secretes about 7 times more PSA than PAP (Henttu and Vihko, 1993), the serum concentrations of mice bearing the prostatic tumor PC-82 (Perala-Heape et al., 1991), as well as comparison of

PAP AND PSA GENE EXPRESSION IN PROSTATIC TUMOKS 597

immunohistochemical staining with anti-PAP and anti-PSA antibodies (Ellis et aZ., 1984). Compared with the parent antibodies, the bispecific antibodies were able to stain the carcinomas (specimens 12 and 14 in Table 111) that were negative for PAP or PSA, thus increasing the sensitivity of staining. However, 2 of thc 27 Ca specimens were negative for both PAP and PSA.

In summary, our results show that PAP and PSA genes are expressed at higher levels in the epithelial cells of BPH tissue than in those of Ca tissue. Furthermore, lower levels of PSA enzyme were detected in poorly differentiated Cas than in well or moderately differentiated Cas. Thus, the increased serum accumulation of PAP and PSA enzymes cannot b e explained by concomitant changes in their intraprostatic con- centrations, and other factors, such as altered polarization of

REFER

AUTIO-HARMAISEN. H., SANDBERG, M., PIHLAJANIEMI, T. and VUORIO, E., Synthesis of laminin and type IV collagen by tro hoblastic and fibroblastic stromal cells in the early human placenta. fab. InLesr., 64, 483-491 (1991). DAVIES, P. and EATOS, C.L., Regulation of prostate growth. J. Endocn'nol., 131,5-I7 (1991). DAVIS, L.G., DIDSIX, M.D. and BATLEY, J.F., Basic methods in molecular biology, pp. 130-135, Elsevier, Amsterdam (1986). ELLIS, D.W., LEFFERS, S., DAVIS, J.S. and NG, A., Multiple immu- noperoxidase markers in benign hyper lasia and adenocarcinoma of the prostate.Amer. J. clin. Pafhol., 81,279-284 (1984). FORT, P., MARTY, L., PIECHACZYK, M., EL SABROUTY, S., DANI, C., JEANTEL'R, P. and BLANCHARD, J.M., Various rat adult tissues express only one major mRNA species from the glyceraldehyde-3-phosphate deh drogenase multigene family. Nucl. Acids Rex, 13, 1431-1442 (19J5). GALLEE, M.P., VISSEH DE JONG, E., VAN DER KoKPUI, J.A., VAN DER KWAST, T.H., TEN -YE, F.J., SCHROED~K, F.H. and TKAPMAS, J.. Variation of prostate-specific antigen cxpression in different tumour growth patterns present in prostatectomy specimens. Urol. Res., 18,

HELLER, J.E., Prostatic acid phosphatase: its current clinical status. J. Urol., 137,1091-1103 (1987). H t N r r ~ , P., LIAO, S. and VIHKO, P., Androgens up-regulate the human prostate-specific antigen messenger ribonucleic acid (mRNA), but down-regulate the prostatic acid phosphatase mRNA in the LNCdP cell line. Endocrinology, 130,766-772 (1992). HENTTU, P., LL'KKARINEN, 0. and VIHKO, P., Expression of the gene coding for human prostate-specific antigen and related hGK-I in benign and malignant tumors of the human prostate. Int. J. Cancer, 45, 654-660 (1990). HENTTU, P. and VIHKO, P., Growth factor regulation of gene expres- sion in the human rostatic carcinoma cell line LNCaP. Cancer Res., 53,1051-1058 (1995. HOYHTYA, M., VIHKO, P., VUOLAS, L., TRYGGVASON, K. and VIHKO, R., High-affinity monoclonal antibodies specific for human prostatic acid phosphatase. Clin. Chem., 33,103-107 (1987). LIUA, H., A kallikrein-like serine protease in prostatic fluid cleaves the redominant seminal vesicle protein. J. din. Invest., 76, 1899-1903 (198;). MORI, K. and WAKASUGI, C., Immunocytochemical demonstration of prostatic acid phosphatase: different secretion kinetics between nor-

181-187 (1990).

cancer cells or impaired integrity of ductal-acinar structures must be responsible for this finding.

ACKNOWLEDGEMENTS

The skillful technical assistance of Ms. H. Auno, Ms. A. Huhtala, Mr. V. Lcinonen, Ms. M. Vahera and Ms. A. Vesala is gratefully acknowledged. This work was supported in part by the Sigrid Juselius Foundation, the Finnish Cancer Founda- tion, the Research Council for Medicine of the Academy of Finland and the Technology Development Center of Finland ( E K E S ) . T h e Department of Clinical Chemistry, University of Oulu, is a W H O Collaborating Center for Research in Reproduction supported by the Ministry of Education, the Ministry of Health and Social Affairs and the Ministry of Foreign Affairs, Finland.

LENCES

mal, hyperplastic and neoplastic prostates. J. Urol., 133, 877-883 (1985). NADJI, M., TAHEI, S.Z., CASTRO, A., CHU, M., MURPHY, G.P., WANG, M.C. and MORALES, A.R., Prostate-specific antigen: an immunohisto- logic marker for prostatic neoplasms. Cancer, 48, 1229-1232 (1981). OESTERLING, J.E., Prostate specific antigen: a critical assessment of the most useful tumor marker for adenocarcinoma of the prostate. 1. Urol., 145,907-923 (1991). PERALA-HEAPE, M., VIHKO, P., LAINE, A., HEiuiLA, J. and VIHKO, R., Effects of tumor mass and circulating antigen on the biodistribution of l'l-Indium-labelled F(ab)z-fragments of human prostatic acid phospha- tase monoclonal antibody in nude mice bearing PC-82 human prostatic tumor xenografts. Europ. J . nucl. Med., 18,339-345 (1991). PRETLOW, T.G., PRETLOW, T.P., YAYG, P., UETZEL, CH.S., DEL- MORO, C.M., KAMIS, S.M., BODNER, D.R., KL'RSH, E., RESNICK, M. and BRADLEY, E.L., JR., Tissue concentrations of prostate specific antigen in prostatic carcinoma and benign prostatic hyperplasia. Inr. J. Cancer, 49,645-649 (1991). Qiu, S.-D., YOUNG, C.Y.-F., BILHARTZ, D.L., PRESCOTT, J.L., FAH- ROW, G.M., HE, W.-W. and TINDALL, D.J., I n situ hybridization of prostate-s ecific antigen mRNA in human prostate. J. Urol., 144, 155O-15%!( 1990). SINHA, A.A., HAGEN, K.A., SIBLEY, R.K., WIISON, M.J., LIMAS, C., REDDY, P.K., BLACKARD, C.E. and GLEASON, D.F., Analysis of fixation effects on immunohistochemical localization of prostatic-specific anti- gen in human pr0state.J. Urol., 136,722-727 (1986). SOI.IN, T., KONTTURI, M., POHLMANN, R. and VIHKO, P., Gene expression and rostate specificity of human prostatic acid hospha- tase (PAP): evafuation by RNA blot analyses. Biochim. biopfys. Acra,

STAMEY, T.A., YANG, N., HAY, A.R., MCNEAL, J.E., FREIHA, F.S. and REDWISE, E., Prostate-s ecific antigen as a serum marker for adenocar- cinoma of the prostate. f i e w Engl. J. Med., 317,909-916 (1987). VIHKO, P., KONTTURI, M. and KORHONES, K., Purification of human prostatic acid phosphatase by affinity chromatography and isoelectric phocusing. Part. I . Clin. Chem., 24,46&470 (1978). VIHKO, P., KOSTAMA, A., JANSE, O., SAJAKTI, E. and VIHKO, R., Rapid radioimmunoassay for prostate-s ecific acid phosphatase in human serum. Clin. Chem., 26,1544-154f(1980). VIHKO, P., KURKELA, R., RAMBERG, J., PELKONES, I. and VIHKO, R., Time-resolved immunofluorometric assay for human prostate-specific antigen. Clin. Chem., 36,9245 (1990).

1048,72-77 (1990).