Vol. 3, 2295-2300, Dece,nber 1997 Clinical Cancer Research 2295
Loss of Uteroglobin Expression in Prostate Cancer: Relationship to Advancing Grade1
Ashani T. Weeraratna, Jaime A. Cajigas, INTRODUCTION Arnold Schwartz, Erik G. Enquist, Adenocarcinoma of the prostate is the most commonly diag- Michael J. Manyak, and Steven R. Patierno2 nosed form of cancer among men in the United States today. The number of cases diagnosed in 1996 exceeded 260,000, and that Departments of Pharmacology [A T. W., S. R. P.], Pathology [A. S.], figure is expected to climb to around 334,500 in 1997 (1). It is and Urology [J. A. C., E. G. E., M. J. M,], George Washington projected that about 41 ,800 of these diagnosed cancers will result in University Medical Center. Washington, DC 20037 death due to metastatic progression. Genetic alterations such as the loss of heterozygosity at several loci, including 6p, 7q, 8q, lOq, ABSTRACT 1 lp, l3q, 16q, 17q, and l8q (2-8), the loss of putative metastasis
We have shown previously that the secretory protein suppressor genes such as KAI-l (9) and CD44 (10), as well as uteroglobin (UG) is highly expressed in normal human pros- changes in cell adhesion mechanisms such as E-cadherinla-catenin tate tissue but this expression is either lost or altered in ( 1 1), are thought to contribute to this progression. However, despite the recent strides made in this area of study. it is generally accepted human prostate cancer cell lines. Treatment of these cell that the molecular alterations related to the malignant development lines with recombinant human UG inhibits their ability to of PC require better characterization. invade human reconstituted basement membrane by up to One such alteration may involve a protein known as UG. 90%, implying that the loss of normal UG expression may be UG is a low molecular weight, homodimeric secretory protein, related to the invasive potential of prostate cancer. Because also known as Clara cell 10 kDa (12) protein, which is found in invasion represents a critical step in metastasis, the expres- abundance in human uterus, lung. and prostate (13-15). The UG sion patterns of UG could provide a unique and relevant locus has been localized to a single-copy gene on chromosome indicator of cancer progression. In this study, we present the 1 lql2.3-l3.l (16). This chromosomal region is of interest in immunohistochemical analyses of fresh frozen prostate tis- several hormonally regulated cancers ( 1 7) and has recently been sues from surgical specimens taken from 50 patients. identified as an additional site of loss of heterozygosity in PC Eight slides per patient were analyzed for UG staining. (18). We have reported previously that UG expression is mark- Slides from 26 patients showed evidence of prostate cancer, edly decreased or absent in several PC cell lines and that whereas slides from the remaining 24 patients showed only recombinant human UG modulates prostate tumor cell invasive- benign glands. The results demonstrate UG immunoreactiv- ness, possibly by a mechanism related to the inhibition of the ity in normal prostate, benign prostatic hyperplasia, and actions of PLA2 in the production of eicosanoids (19, 20). prostatic atrophy; low but clearly positive expression in This study sought to determine the immunohistochemical prostatic intraepithelial neoplasia; positive expression in expression of UG in normal human prostate. BPH, PIN, and PC cancerous glands of Gleason’s pattern 2; and complete loss tissue and to assess the possibility that alterations in UG ex- of UG immunoreactivity in cancerous glands of Gleason’s pression may be associated with prostate cancer. pattern 3 or greater. In addition, in the one case of meta- static prostate cancer that we examined, the prostate cancer MATERIALS AND METHODS cells within the lymph node lacked UG expression. These Tissue Source. Informed consent was obtained to study findings suggest that the loss of UG expression may be an tissue from 50 patients who underwent radical prostatectomy for indicator of prostate cancer progression and possibly a com- PC, cystoprostatectomy for transitional cell carcinoma, or open ponent of the molecular natural history of prostate cancer, prostatectomy for BPH. Because a random sampling of cancer which may have prognostic value. patients was desired, there was no inclusion or exclusion criteria before acquisition of these tissues. It is noteworthy. however, that none of these patients had prior endocrine treatment. Ap- proximately two longitudinal specimens, of 1-cm each, extend- ing from base to apex were excised parallel to the urethra and flash frozen in liquid N, and stored at -70#{176}C. Lymphatic tissue Received 6/4/97: revised 8/25/97; accepted 9/17/97. The costs of publication of this article were defrayed in part by the containing PC was obtained from one laparoscopic pelvic lymph payment of page charges. This article must therefore be hereby marked node dissection prior to a planned radical prostatectomy. Of 46 advertisement in accordance with 18 USC. Section 1734 solely to specimens resected for prostatic carcinoma, 26 contained carci- indicate this fact. This work was supported by an award from the Elaine A. Snyder Cancer Research Trust, 2To whom requests for reprints should be addressed. at Department of Pharmacology. George Washington University Medical Center. 2300 I 3The abbreviations used are: PC, prostate cancer: UG. uteroglobin: Street NW.. Washington. DC 20037. Phone: (202) 994-3286: Fax: PLA,, phospholipase A,; BPH, benign prostatic hyperplasia; PIN, pros- (202) 994-2870: E-mail: [email protected]. tatic intraepithelial neoplasia.
Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 1997 American Association for Cancer Research. 2296 Uteroglobin Expression in Prostatic Cancer
noma within the tissue samples obtained intraoperatively for UG fer in 10 mM 3-(cyclohexylamino)ethanesulfonic acid buffer determination. Tissue from the remaining 24 specimens con- (pH 1 1 .0) with 20% methanol for 10 mm at 350 mA. The blot tamed benign prostatic tissue, prostatic atrophy, or fibromuscu- was then blocked overnight at 4#{176}Cin 1 % BSA. Rabbit anti- lar tissue. The 26 cases with sampled carcinoma in the intraop- human UG at a 1 : 100 dilution in 1% BSA in 1 X TBS-T (0.02
erative sample had a similar histopathological tumor pattern to M Tris base, 0. 137 M NaCI, and 0.1 % Tween 20, pH 7.6) was that present in the resected prostate. All 46 prostatectomy cases used to primary stain the blot for 1 h at room temperature. The with preoperative diagnosis of PC contained carcinoma in the blot was then washed four times in 1 X TBS-T for 15 mm each. final surgical pathological review of the resected organ. It was then stained with secondary antibody. peroxidase-conju- Immunohistochemical Procedures. Two hundred serial gated donkey anti-rabbit IgG (Amersham Corp., Arlington sections of 5 p.m each were taken from each tissue and placed on Heights, IL) for 1 h at room temperature, and washed again as silanated slides (two sections per slide). Every tenth slide was above. The presence of UG was detected using the ECL system stained with hematoxylin and diagnosed by a pathologist to deter- (Dupont NEN, Boston MA). This method of detection was mine which serial sections exhibited BPH, PIN, or cancer. This was modified by using a 1:2 dilution (in 1 X TBS-T) of the reagents to determine which slides contained both normal prostatic glands to minimize background staining. The blot was then exposed to (as a positive control) as well as adjacent cancerous glands. The autoradiography film for 20 s. remaining slides were stored at -20#{176}Cuntil diagnosis was com- plete. Selected slides were immunostained as follows. Sections were fixed with 4% formaldehyde for 10 s and then blocked with RESULTS a 1:100 dilution of mouse serum (SIGMA Immunochemicals, St. Patient age in the 28 Caucasian and 22 African-American Louis, MO) for 30 mm at room temperature. Slides were then males ranged from 35 to 76 years with a median of 60.8 years. incubated overnight at 4#{176}Cwith a 1:100 dilution of rabbit anti- Forty-six patients had a preoperative histopathological diagnosis human UG antibody, and corresponding serial sections used as a of prostate adenocarcinoma (26 Caucasians and 20 African- negative control were incubated with a 1 : 100 dilution of nonim- Americans), whereas 4 patients (2 Caucasians and 2 African- munized rabbit serum (BABCO Laboratories, Inc., Berkeley, CA). Americans) had no malignancy of the prostate and underwent The sections were washed three times for 10 mm each in PBS and surgery either for bladder carcinoma ( I Caucasian and 1 Afri-
then incubated with a dilution of 1:1000 biotin-conjugated mouse can-American) or for BPH (Table 1 ). These diagnoses were anti-rabbit IgG (SIGMA Immunochemicals) for 30 mm at room confirmed in histopathological examination of the surgical spec- temperature. The sections were washed three times in PBS for 10 imens. Of the 26 patients (15 Caucasians and 1 1 African- ruin, then incubated with streptavidin complex (DAKO, Carpinte- Americans) with carcinoma within the sections we examined, rim, CA) for 30 mm, and washed again twice with PBS for 10 mm. two had associated PIN. Two samples contained only fibromus- A 5-s immersion in chromogen (liquid DAB, DAKO) was fol- cular tissue. The remaining 22 specimens tested contained only lowed by counter staining with hematoxylin. BPH or prostatic atrophy in the sample. However, the final Antibody Preparation. A peptide comprised of the 20 surgical pathological diagnosis of these radical prostatectomies amino acid residues 37-56 of human UG was synthesized by was adenocarcinoma with Gleason scores ranging from 4 to 10, Genemed (San Francisco, CA). This peptide was conjugated to with 16 patients determined to have moderately differentiated keyhole limpet hemocyanin and used to raise antibody in rabbits tumors (Gleason score from 4 to 6) and 30 patients with poorly by the Berkeley Antibody Company (Berkeley, CA). The spec- differentiated tumors (Gleason score from 7 to 10). The latter ificity of this polyclonal serum to human UG was confirmed and included the patient with lymph node metastasis (Gleason 10). titred by ELISA and used in the immunostamning of prostate A total of 400 slides from the 50 specimens was evaluated, tissue specimens and performance of Western blot analysis. and UG expression in each slide was scored by three investiga- Western Blot Analysis. Slices of tissue (-50 mg) adja- tors (A. T. W., J. A. C., and A. S.). The degree of UG expres- cent to the areas of tissue used for immunohistochemistry were sion was graded by assigning a value of 0 to an absence of stain homogenized in protein lysis buffer [50 mrvi Tris Base, 350 mrvi and assigning a value of 3+ to the most positive stain. Slides NaCl, 0.1% volume NP4O, 5 mr i EDTA (pH 7.4), containing 2 that stain with intensity between these two extremes were ac- mM PMSF, 20 xg/ml aprotinin, and 1 m i NaOV] using an corded intermediate values. Glands showing an absence of stain Ultratorrax homogenizer (Tekmar, Cincinnati, OH), three times were uniformly marked as 0 by all three investigators. Interme- for 30 5 each. Cell debris was pelleted out by centrifugation at diate scores were usually identical and never differed by more 8000 rpm for 20 mm. The supernatant was then filtered through than a single value. Two of three scorers were blinded as to the a lOO-kDa molecular weight cut-off Centricon-100 spin filter source of the tissue. Immunohistochemical stain was also eval- (Amicon, Beverly, MA) by centrifugation for 1 h at 2300 rpm at uated and scored according to distribution as focal or diffuse. In 4#{176}C.The lysates were then quantitated using a Bio-Rad protein cases of suspected minimal disease, tissue sections were ob- assay (Bio-Rad, Hercules, CA), and 75 i.g of the proteins below tamed from sites adjacent to the localized disease and did not Mr l00 000 were heated for 3 mm at 95#{176}Cin an equal volume always contain tumor. Of the 50 patient samples examined, we of sample buffer (4% SDS, 100 mM Tris-HCI, 20% glycerol, found we were able to analyze slides with cancer from 26. Two 200 mst DTT, 0.2% w/v bromphenol blue, with 0.5% -mer- of these latter slides contained only cancer, with no evidence of captoethanol) and loaded on to a 15% denaturing SDS-poly- any normal glands, and one contained cancer and PIN with no acrylamide gel and subjected to electrophoresis for 40 mm at 65 evidence of any benign glands. Two samples we examined V. The proteins were transferred onto 0.2-p.m polyvinylidene contained only fibromuscular tissue, which lacks staining for difluoride membrane (Bio-Rad, Hercules, CA) by electrotrans- UG. Therefore, we found 45 samples that contained benign
Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 1997 American Association for Cancer Research. Clinical Cancer Research 2297
Table I Patient data cancerous glands of this Gleason’s pattern. Five samples con- Summary of patient characteristics according to age. race, and final tamed cancerous glands that were graded with a Gleason’s pathological diagnosis as determined by surgical pathological examina- pattern 1 or 2, and these were positive for UG staining. One tion of the resected organ. Also included in this table is a summary of sample contained several glands ofGleason’s pattern 3/4 as well the differing pathologies seen among the various fields of view in different patients at the time of slide analysis. as glands that were Gleason’s pattern 2. The Gleason’s pattern 3/4 lacked staining for UG (Fig. 1E, left), whereas the Gleason’s Detail pattern 2 and the normal glands within the sample stained a = 50 No. of patients positive for UG (Fig. 1E, right). The prostate tumor cells in the Age Range 35-76 yr Median 60.8 yr metastatic lymph node likewise registered no signal (Fig. lF). Race African-American 22 Western analysis was performed on prostate tissue from six Caucasian 28 ofthe patients in this study presenting with cancers of Gleason’s Final pathological BPH 4 grade 8 or 9. The UG protein runs at around Mr I 0,000 on an diagnosis” Bladder carcinoma 2 SDS polyacrylamide gel (21). This protein was present in nor- PIN 5 PC Gleason’s grade 5 4 mal tissue lysates but absent in the lysates taken from cancerous PC Gleason’s grade 6 12 tissue (Fig. 2). PC Gleason’s grade 7 20 PC Gleason’s grade 8 2 PC Gleason’s grade 9 7 DISCUSSION PC Gleason’s grade 10 1 Slide diagnoses” BPH 42 We have previously reported our discovery of abundant Prostatic atrophy 3 expression of UG in the epithelial cells of the human prostate PIN 2 gland (15). In that preliminary description, there was no analysis PC Gleason’s pattern’ 3 20 of the comparative expression of uteroglobin in tissue sections PC Gleason’s pattern 2 6 exhibiting different pathological conditions. This study exam- Fibromuscular tissue only 2 ines UG expression in fresh frozen cellular fields of BPH, PIN, (1 The two patients with bladder carcinoma had a final prostatic and prostate cancer. We report that UG was absent from pros- diagnosis of BPH. PIN was associated with PC in all five cases; thus n = PC + BPH = 50. tatic stromal tissue but was clearly expressed in normal glandu- h Patterns of glands and cells in the field of view at the time of slide lar epithelial cells, in glandular areas exhibiting BPH and pros- analysis. The numbers here represent ii (patients) of a total of 50 in tatic atrophy. in acinar glands identified as PIN, and in which these pathologies were observed. cancerous glands of Gleason’s pattern 2 or lower. In contrast, ‘ See footnote 4 in text. UG expression in glands of Gleason’s pattern 3 or higher was absent. In addition, in Western analysis of six patient samples of Gleason’s grade 8 or higher, the expression of the UG protein is tissue, of which 42 contained benign glandular hyperplasia and undetectable as compared to normal tissue. This observation three had prostatic glandular atrophy. Because of the difficulty links the loss of UG expression with the pathological pattern of of both finding and diagnosing PIN on frozen sections, the prostatic neoplasia. number of specimens in which PIN was detected remained low. The loss of normal UG expression in fresh frozen tissue UG expression in benign tissue was predominantly diffuse samples of PC shown here corroborates our observations of and moderate in most cases. In 42 cases of histologically con- human prostatic tumor cell lines derived from metastases ( I 9, firmed BPH, there was one that stained 3+, 25 that stained 2+, 20). The cell lines DU14S. TSUprl, and PC-3 were analyzed for and 19 that stained 1 + for UG. In three cases of prostatic UG mRNA expression by Northern blotting and compared with atrophy, two stained 2+ and one stained 1 +. All were clearly normal tissue. Normal prostatic cells expressed an abundant positive (1 + or greater). The staining pattern was cytoplasmic 600-bp mRNA transcript, whereas UG expression in the cell and tended to be more pronounced at the luminal surface of the lines DU145 and PC-3 was absent. In TSUpr1 cells, an over- acinar cells (Fig. 1 . A and B). Immunoreactivity was absent in expressed but aberrantly processed (> lO-kilobase pair) iran- sections of the prostate that lacked epithelial elements, such as script was detected (20). fibromuscular tissue (Fig. 1A). The two cases of PIN examined The absence of UG expression in prostatic tumor cells is for the expression of UG were both clearly positive and were also consistent with reports on the expression of the Clara Cell scored as 1+ (Table 2, Fig. IC’). 10 kDa protein (CC/U) gene. which is believed to be UG (21), Evaluation of slides containing PC contrasted sharply with in lung tissue. Broers et al. (14) and Linnoila et al. (22) showed normal, BPH, or PIN samples. Twenty of 26 specimens con- that CC/U mRNA was abundantly expressed in normal bron- taming cancer did not stain for UG. Each of these cancerous chial tissue but was detectable in only I of 19 non-small cell glands was diagnosed as Gleason’s pattern4 of 3 or above (Fig. lung carcinoma specimens. A large percentage of non-small cell 1, C and D). This absence of stain was consistent among all lung carcinoma (primarily adenocarcinomas) exhibit ultrastruc- tural and morphological characteristics indicative of peripheral airway cell origin (22). Because CC/U was a strong positive marker for normal peripheral airway cells, it is likely that the 4 The Gleason’s number referred to here is that of the particular glands low percentage of CC/U-positive tumors was due to the loss of within the field of view at the time of slide diagnosis and refers to the pattern of the individual glands as opposed to the overall Gleason’s CCIO expression during tumorigenesis. Sandmoller et a!. (23) grade of the patient. also supported this observation by fusing the 5’ flanking region
Downloaded from clincancerres.aacrjournals.org on September 25, 2021. © 1997 American Association for Cancer Research. 2298 Uteroglobin Expression in Prostatic Cancer
B . p
:‘ #{149}t t.. .‘ . 0 ,‘ #{149} C -- _57 , ..-. ‘
, \t. .