In Situ Hybridization Analysis of Genes CodingCollagen IV a1 Chain, Laminin b1 Chain, and

S-Laminin in Prostate Tissue and ProstateCancer: Increased Basement Membrane Gene

Expression in High-Grade and Metastatic Lesions

Claudia Pfohler,1 Thomas Fixemer,1 Volker Jung,2 Steven Dooley,2Klaus Remberger,1 and Helmut Bonkhoff1*

1Department of Pathology, University of the Saarland, Homburg/Saar, Germany2Department of Human Genetics, University of the Saarland, Homburg/Saar, Germany

BACKGROUND. Recent immunohistochemical data have shown that invasive prostate can-cer cells are separated from the host tissue by basement membranes (BM), and expressassociated adhesive molecules that bind to these de novo synthesized extracellular matrices.METHODS. In the present study, we used in situ hybridization techniques to determinesteady-state levels of genes coding BM components (a1 chain of collagen IV, laminin b1 chain,and S-laminin) in prostate tissue obtained from 15 radical prostatectomy specimens and 5lymph node metastases of common prostatic adenocarcinomas.RESULTS. In benign prostate tissue, transcripts of these genes were detected predominantlyin the basal cell layer, indicating that components of epithelial BMs are synthesized by basalcells and not bey stromal cells. The cancerous lesions investigated revealed increasing colla-gen IV, laminin b1 chain, and S-laminin mRNA levels when compared with benign prostatetissue. The highest steady-state levels were found in high grade (primary Gleason grade 4 and5) carcinoma and lymph node metastases, and were predominantly localized in epithelialcompartments of the cancerous tissue.CONCLUSIONS. These findings indicate that neoplastic BM in prostatic adenocarcinomaderive from tumor cells and not from the host tissue. Increasing transcriptional activities ofgenes coding BM components detected in poorly differentiated and metastatic lesions mayaccelerate the BM-forming process, which probably contributes to the ability of tumor cells topenetrate the extracellular matrix during the process of stromal invasion and metastasis.Prostate 36:143–150, 1998. © 1998 Wiley-Liss, Inc.

KEY WORDS: basement membrane; gene expression; prostate cancer

INTRODUCTION

Basement membranes (BM) are highly specializedextracellular matrix components that regulate cell-stroma interactions in such various biological pro-cesses as embryogenesis, tissue repair, tumor inva-sion, and metastasis. Over the last few years, consid-erable experimental evidence has been amasseddocumenting the role of BM components in normaland abnormal prostatic growth [1–6]. Benign prostaticepithelial cells require BM constituents for mainte-

nance of the differentiated phenotype and secretoryfunction in vitro [5,6]. Extracellular matrix compo-nents promote cancer-cell differentiation, and altergene expression and tumorigenic potential of prostatecancer-cell lines [1,2,6]. Although these data have con-

Contract grant sponsor: Deutsche Forschungsgemeinschaft; Con-tract grant number: B0101812-2.*Correspondence to: Helmut Bonkhoff, Institute of Pathology, Uni-versity of the Saarland, 66424 Homburg/Saar, Germany.Received 7 October 1997; Accepted 4 February 1998

The Prostate 36:143–150 (1998)

© 1998 Wiley-Liss, Inc.

vincingly demonstrated the regulatory impact of ex-tracellular macromolecules on prostatic growth invitro, its significance for the human prostate is lesswell established. Like other epithelial tissue, prostatictubuloalveolar structures are separated from thestroma by BM which mainly contain collagen IV andVII, laminin, heparansulfate proteoglycan, and entac-tin [7,8]. The basal cell layer of the prostatic epitheliumexpresses a panel of adhesive molecules that mediateattachment to the vaious components of epithelial BM,and thus maintain tissue integrity and orderly, regu-lated prostatic growth [9–12].

Disruption of normal epithelial-stromal relations isa hallmark of invasive carcinomas. There is increasingevidence to suggest that the regulatory impact of BMon differentiation and epithelial-stromal interactionsdecreases during the process of tumor invasion [13–17]. Epithelial neoplasias tend to lose BM, and down-regulate associated adhesive molecules when com-pared with their benign counterparts [13–17]. The pro-gressive loss of adhesive elements has been associatedwith tumor progression, metastasis, and poor progno-sis in a variety of human carcinomas. This conceptmay be true for most commonly diagnosed humanmalignancies, including lung [18,19], colorectal [20–22], and breast [23–25] carcinomas, but not for prostatecancer. In the latter tumor, invasive neoplastic cells areseparated from the host tissue by distinct BM, andexpress associated integrin receptors that mediate ad-hesion to these BM-like matrices [7,9,26]. In addition,this particular feature of prostate cancer seems to bemaintained through the various stages of the disease[7,9,26], although former immunohistochemical stud-ies have reported markedly reduced levels of type IVcollagen and laminin in poorly differentiated andmetastatic lesions [26,27].

Type IV collagen represents the major structuralcomponent of BM, while laminins confer a number ofbiological properties, including cell adhesion, prolif-eration, differentiation, and spreading [28,29]. All ofthe seven known laminins are composed of one heavychain (a) and two light chains (b and g) arranged in acruciform structure [30,31]. Currently, nine geneticallydistinct laminin subchains have been identified [31].The b1 subchain is contained in laminin 1 (a1b1g1)(Engelbreth-Holm-Swarm (EHS) or classical laminin),laminin 2 (a2b1g1), and laminin 6 (a3b1g1) [31]. Theb2 chain (or S-laminin) is present in laminin 3(a1b2g1), laminin 4 (a2b2g1), and laminin 7 (a3b2g1)[31]. Laminin 1 predominantly occurs in epithelial andendothelial BM, whereas S-laminin is restricted to BMof neuromuscular junctions, glomerula, and perineu-rium [32]. In the present study, we evaluated the tran-scriptional activities of genes coding BM components(a1 chain of collagen IV, S-laminin, and laminin b1

chain) in benign prostate tissue, premalignant lesions(high grade prostatic intraepithelial neoplasia), andprimary and metastatic cancer using in situ hybridiza-tion (ISH). We chose S-laminin for ISH evaluation be-cause recent data have reported considerable amountsof surface-bound S-laminin in the highly tumorigenicDU 145 prostate cancer cell line [33].

Results of the present ISH study demonstrate highsteady-state levels of genes coding the a1 chain ofcollagen IV, the laminin b1 chain, and the S-laminin inpoorly differentiated and metastatic lesions, and fur-ther document the pathogenic significance of BM com-ponents for the invasive and metastatic process in hu-man prostate cancer.

MATERIALS AND METHODS

Tissue Selection and Processing

Human prostate tissue was obtained from 15 menbetween age 57–82 years (mean, 68 years) who under-went radical prostatectomy and lymphadenomectomyfor prostate cancer. Pathological stages included pT3c(n = 13), pT2c (n = 2), and pN1 (n = 5). From each case,several representative tissue samples were collectedfrom cancerous lesions, and from benign prostate tis-sue of the peripheral and the transition zone.

Tissue sections were quick-frozen in liquid isopen-tane at −80°C until sectioning. Subsequent cryostatsections of 6 mm were cut at −20°C, mounted on ami-noalkysilane-treated glass slides, air-dried for 10 min,fixed in 4% paraformaldehyde, and stained with he-matoxylin and eosin for histological examination. Thecancerous lesions selected for ISH analysis includedprimary Gleason grades 3 (n = 12), 4 (n = 15), and 5(n = 9). Histological evaluation further revealed fivecases of fibroglandular hyperplasia, normal prostatetissue from the peripheral and the transition zone,and high grade prostatic intraepithelial neoplasias(HGPIN) from 12 patients.

The material submitted for study was air-dried at40°C on a hotplate for 10 min and fixed in freshlyprepared 4% paraformaldehyde in phosphate-buffered saline (PBS). Subsequent sections werewashed for 5 min in PBS, incubated for 30 min in 2 ×standard saline citrate (SSC) at 70°C, dehydrated ingraded alcohols, air-dried, and submitted for the ISHprocess.

Labeling of Oligonucleotides

Sense and antisense probes for the a1 chain of col-lagen IV, the laminin b1 chain, and S-laminin used inthe present study were provided by MWG-Biotech(Ebersberg, Germany).

144 Pfohler et al.

The following oligonucleotides were chosen frompublished c-DNA sequences and are numbered ac-cording to Genebank format: 58 aggctctgaaattgaaac-gggcag 38 (5165–5188, sense) and 58 agtgcccgctactttcct-catag 38 (5436–5459, antisense) for the S-laminin c-DNA, 58 cgatcagcgagttagagaggaatg 38 (3464–3487,sense) and 58 tgaaaatggagatgcctagcaccc 38 (3059–3082,antisense) for the laminin b1 chain c-DNA, and 58gaaaaggagatcaagggatagcg 38 (3464–3487, sense) and 58ccaggaattcctggatccaaagga 38 (3059–3082, antisense) forthe a1 chain of collagen c-DNA. These probes werechosen because of their C-G content of 50% to ensurean optimal hybridization sensitivity. The Tm valuescalculated were identical for all probes.

One hundred picomoles of each oligonucleotidewere labeled with digoxigenin (DIG) 11-dUTP usingthe Oligonucleotide Tailing Kit (Boehringer Mann-heim, Mannheim, Germany). One microliter of DIG11-dUTP, 1 ml dATP, 4 ml cobalt chloride solution, and4 ml of the terminal transferase buffer of the tailing kitwere mixed with 1 ml terminal transferase, and incu-bated for 10 min at 37°C. The labeling reaction wasstopped by adding 1 ml of the glycogen solution (20mg/ml) and 200 ml 0.2 M EDTA, pH 8.0. The oligo-nucleotides were precipitated with 2.5 ml 4 M LiCl and75 ml 100% ethanol (−20°C), and stored at −70°C for 30min.

After centrifugation at 14,000 rpm at 4°C for 10 min,the supernatant was removed. The procedure was re-peated with 70% ethanol. The pellet was dried anddissolved in 20 ml aqua dest. To prove the labelingefficiency, the probes were serially diluted in 10 ml 6 ×SSC. One microliter of each dilution was applied to anitrocellulose membrane and cross-linked. Detectionwas achieved with the detection kit of BoehringerMannheim, using nitrobluetetrazolim/5-bromo,4-chloro-indolylphosphate (NBT/BCIP) (BoehringerMannheim) as color reagents. The labeling efficiencyof each probe was compared with a standard prela-beled oligonucleotide of the tailing kit (BoehringerMannheim). Probes that revealed the same signal in-tensity as the standard probe on dot-blot analysis wereused for the hybridization process to ensure an iden-tical labeling efficiency of each probe.

Hybridization Procedure

The tissue sections are predenatured at 90°C on ahotplate for 20 min and immediately placed on ice.The hybridization mixture [50% formamide (Sigma-Aldrich Chemie GmbH, Deisenhofen, Germany), 2 ×SSC, 10% dextrane sulfate (Sigma-Aldrich ChemieGmbH), 1 ml yeast t-RNA (Boehringer Mannheim), 1ml sheared salmon sperm DNA (Sigma-Aldrich Che-mie GmbH), and 4 ml of dissolved oligonucleotide

probe] was denatured for 5 min at 95°C in a water bathand immediately placed on ice. Twenty-five microli-ters of hybridization mixture were applied to eachslide, covered with coverslips, and sealed with rubbercement. After denaturation for 30 min at 80°C in ahumidified chamber and overnight hybridization at37°C, the slides were washed twice in 2 × SSC for 5min at room temperature, twice in 2 × SSC at 42°C for5 min each, and finally three times in 0.1 × SSC at 42°Cfor 10 min each. After incubation for 10 min in TN-buffer (0.1 M Tris, pH 7.6, 0.15 M NaCl), the sectionswere blocked for 30 min with the blocking solution(Blocking Reagent, Boehringer Mannheim). The slideswere incubated with anti-DIG antibodies conjugatedwith alkaline phosphatase (1:50) and the blocking so-lution for 60 min at 37°C. Unbound antibodies wereremoved by washing twice in Tween-20 (Sigma-Aldrich Chemie GmbH) in TN buffer. Detection ofspecific bound probes was achieved with NBT/BCIP.The tissue sections were sightly counterstained withnuclear fast red (Merck Darmstadt, Darmstadt, Ger-many).

Negative Controls

To prove the specificity of the hybridization pro-cess, the following negative controls were performedin each case. The slides were hybridized with the cor-responding sense probe. Secondly, the hybridizationprocedure was performed by omitting the sense andantisense probe. In addition, slides pretreated withRNase (Boehringer Mannheim) were hybridized asdescribed above. By choosing the same hybridizationkinetics (C-G content of the probes, optimal hybrid-ization temperature, and the same hybridization,washing, and development conditions) for all probeswe were enable to compare the ISH results obtainedwith collagen IV, S-laminin, and b1 chain on consecu-tive sections.

Quantitative Analysis of Signal Intensity

The material submitted for study was hybridized induplicate with paired sense and antisense probes. Hy-bridization signal intensities obtained with the anti-sense probe were classified in seven categories, rang-ing from 0 to +6. Quantitative evaluation of collagenIV, b1 chain, and S-laminin mRNA levels was as-sessed relative to the signal intensities obtained in thebasalcell layer of benign prostate tissue by the collagenIV antisense probe and was arbitrarily designated +3.The strongest hybridization signals observed in tissuesections investigated (i.e., primary Gleason grade 5lesions and lymph node metastases) were arbitrarilydesignated +6. Collagen IV, b1 chain, and S-laminin

BM Gene Expression in Prostate Tissue 145

ISH analysis was performed on subsequent tissue sec-tions. For each case and grade, the average signal in-tensity score was evaluated in 10 low-power fields(×40) and reported in Figure 5.

Statistical Analysis

All correlation and variance analyses were per-formed using SAS software. For the Kruskal-WallisH-test and the Wilcoxon and Mann-Whitney U-testswe used SAS software and SPSS software (SPSSGmbH, Munich, Germany). P < 0.05 was regarded asstatistically significant, and ‘‘m’’ denotes the slope.

RESULTS

ISH analysis revealed collagen IV, S-laminin, andb1 chain mRNA expression in epithelial and stromalcompartments of the human prostate. (Figs. 1–4). Al-though hybridization signals were predominantly de-tected in the perinuclear region of labeled cells, in-creasing levels of mRNA frequently resulted in a morenuclear localization of the colored end product. Nev-ertheless, no hybridization signals were obtained withthe corresponding sense probe or following RNAsepretreatment. (Figs. 1b, 4b). In epithelial compart-ments of benign prostate tissue, hybridization signalsfor collagen IV and the b1 chain were generally moreintense than S-laminin mRNA levels. The basal celllayer of normal or hyperplastic glands and prostaticducts uniformly exhibited mRNA expression (Fig. 1).Semiquantitative evaluation of the ISH results in basalcells revealed average signal intensities of +3 for col-lagen IV (standard deviation, s = 0.351), +2.5 for lam-inin b1 chain (standard deviation, s = 0.653), and +1.8

for S-laminin (standard deviation, s = 0.374). Signifi-cant differences between these mean values were ap-proved by the Kruskal-Wallis H-test (P < 0.05). De-creased levels of mRNA were detected in secretoryluminal cells, ranging from 0 to +1 (Fig. 1). The pros-tatic stroma expressed low-to-moderate levels of col-lagen IV, b1 chain, and S-laminin mRNA in fibro-blasts, endothelial cells, and outer root sheet cells.Most stromal cells adjacent to epithelial BM of tubu-loalveolar structures did not reveal mRNA expressionof the BM components investigated (Fig. 1).

In HGPIN, the basal cell layer showed collagen IV,b1 chain, and S-laminin expression with the same in-tensity as detected in benign prostate glands (P < 0.05).In contrast to the latter, increasing mRNA levels weredetected in secretory luminal cell types of HGPIN,

Fig. 1. Prostatic duct. a: a1 chain of collagen IV mRNA expression determined by in situ hybridization. b: The absence of backgroundhybridization with the a1 chain of collagen IV sense probe documents the specificity of the hybridization process. Collagen IV transcriptsare predominantly expressed in the basal cell layer when compared to secretory luminal cells. The prostatic stroma shows collagen IVmRNA expression in some stromal and endothelial cells. Stromal cells adjacent to epithelial basement membranes lack detectable mRNAexpression (×100).

Fig. 2. High grade prostate intraepithelial neoplasia (HGPIN)and adjacent prostate cancer (primary Gleason grade 3). Hybrid-ization with laminin b1 antisense probe shows increasing steady-state levels in secretory luminal cells of HGPIN, reaching signalintensities of adjacent invasive cancerous acini (×40).

146 Pfohler et al.

reaching signal intensities of adjacent carcinoma (col-lagen IV, P = 0.004; b1 chain, P = 0.028, S-laminin, P =0.071) (Figs. 2, 5). When compared with benign pros-tate tissue, increasing collagen IV, b1 chain, and S-laminin mRNA expression was observed in prostatecancer (Figs. 3–5). However, the distribution and den-sity of hybridization signals were more heterogeneousthan in benign prostate tissue (Fig. 5). When comparedto steady-state levels detected in basal cells, the mostsignificant increase of mRNA expression from HGPINto primary Gleason grade 5 carcinomas was recordedfor S-laminin (P = 0.0001, m = 0.799) and the b1 chain(P = 0.0002, m = 0.554), and revealed a significant cor-relation with the primary Gleason grade in linear re-gression analysis. Although collagen IV levels pro-gressively increased from HGPIN to primary Gleasongrade 5 carcinomas, no significant correlation wasfound with the primary Gleason grade (P = 0.1291, m= 0.185). The highest hybridization signals observed inthe present study were obtained with the collagen IVprobe in lymph node metastases (Fig. 4). Slightly de-creasing S-laminin (P = 0.0598, m = −1.293) and b1chain (P = 0.0168, m = −1.285) signals were observed inlymph node metastases as compared with primaryGleason 5 carcinomas. Stromal compartments of inva-sive lesions showed low-to-moderate mRNA levels insubsets of stromal cells. Increasing mRNA expressionadjacent to invasive cancer was not observed (Figs.3, 4).

DISCUSSION

Recent experimental data have convincingly dem-onstrated the regulatory impact of BM constituents onboth normal and abnormal prostate growth, by affect-

ing the differentiation and tumorigenic potential ofprostatic epithelial cells in vitro [1–6]. Using ISH tech-niques, the present study analyzes the differential ex-pression of genes coding for BM proteins (a1 chain ofcollagen IV, laminin b1 chain, and S-laminin) in hu-man prostate tissue, prostate cancer, and metastaticdisease. Results of the current study strongly suggestthe epithelial origin of BM surrounding tubuloalveo-lar structures in the human prostate. The widespreadabsence of detectable transcripts for BM componentsin most stromal cells adjacent to the prostatic epithe-lium virtually excludes the stromal origin of epithelialBM. Conversely, the preferential localization of colla-gen IV, b1 chain, and S-laminin mRNA in the basalcell layer indicates that epithelial BM components de-rive from basal cells. This further highlights the im-portance of basal cell types in normal and abnormalprostatic growth. In fact, the basal cell layer representsthe proliferation compartment of the prostatic epithe-lium, and most probably houses the prostatic stem cellpopulation [34–38]. Since BM constituents are re-quired by prostatic epithelial cells for in vitro growthand differentiation [5,6], it is conceivable that produc-tion of these extracellular macromolecules by basalcells is also crucial for the maintenance of orderlyregulated growth and differentiation processes invivo.

The most significant morphological difference be-tween premalignant and invasive phases of prostaticcancerogenesis is the consistent absence of basal celldifferentiation in invasive lesions [4,39,40]. The pro-gressive loss of basal cells during neoplastic progres-sion may be related to abnormal adhesive interactionsinterfering with differentiation processes within theprostatic epithelial cell system [11,12,37]. Recent im-munohistochemical data suggest that neoplastic pro-gression of HGPIN to early invasive cancer is associ-ated with the loss of hemidesmosome structures andrelated adhesive elements, including collagen typeVII, laminin 5 and a6, b4 integrins [10–12]. Results ofthe current study show that HGPIN reveals increasingBM subunit mRNA expression in secretory luminalcells, reaching steady-state levels of adjacent primaryGleason grade 3 and 4 carcinomas. The aberrant ex-pression of BM-encoding genes in the dysplastic epi-thelium may represent an initial step in neoplasticprogression to early invasive growth, by modifyingthe biochemical nature of the underlying BM. Furtherimmunohistochemical studies evaluating the differen-tial expression of BM components and associated ad-hesive molecules are required to establish the poten-tial implications of BM macromolecules in the processof early stromal invasion.

Our previous immunohistochemical data showedthe presence of BM at the interface between epithelial

Fig. 3. S-laminin mRNA expression determined by in situ hy-bridization in poorly differentiated (primary Gleason grade 5) car-cinoma. Markedly increased steady-state levels of S-laminin aredetected in tumor cells, while the tumor stroma shows weakhybridization signals in some endothelial and stromal cells (×40).

BM Gene Expression in Prostate Tissue 147

and stromal compartments in prostate cancer, regard-less of histological grade and pathologic stages [7,26].The present ISH analysis confirms this concept, andstrongly suggests the epithelial origin of these tumor-associated BM. All adenocarcinomas investigated inthe present study expressed collagen IV, b1 chain, andS-laminin mRNA at variable levels, while adjacentstromal cells generally lacked detectable transcripts.This clearly indicates that invasive prostatic cancerouscells produce BM components that form BM-like ma-trices at the site of contact with the stroma.

Among the three BM components investigated, the

S-laminin gene is most consistently overexpressed ininvasive cancer when compared to the low steady-state S-laminin mRNA levels detected in basal cells ofbenign prostate glands. Recent in vitro studies haveshown that S-laminin is differently expressed in thevaious prostate carcinoma cell lines DU 145, LNCaP,and PC-3, differing by their tumorigenic potentials. Infact, the highly tumorigenic DU 145 cell line producesand secretes considerable amounts of S-laminin,which probably contributes to its ability to form tu-mors in mice without exogeneously added BM com-ponents [33]. Taken together, these observations sug-

Fig. 4. Lymph node metastasis. Hybridization with a1 chain collagen IV antisense probe demonstrates high steady-state levels in thismetastatic lesion. a: Lymph node capsule is seen at bottom. b: Absence of background hybridization with the collagen type IV sense probedocuments the specifity of the hybridization process (×40).

Fig. 5. Semiquantitative evaluation of a1chain of collagen IV (C IV), laminin b1 chain(b1), and S-laminin (S) mRNA levels in highgrade prostatic intraepithelial neoplasia(HGPIN), primary Gleason grades 3–5 ad-enocarcinoma, and lymph node metastases(LNM), using a signal-intensity score rang-ing from 0–+6. Bars indicate signal intensi-ties of collagen IV (+3), laminin b1 chain(+2.5), and S-laminin (+1.8) mRNA expres-sion detected in the basal cell layer of be-nign prostatic glands.

148 Pfohler et al.

gest that S-laminin is a functionally important con-stituent of tumor-associated BM in human prostatecancer. This warrants further analysis of the immuno-profile of S-laminin and other laminin subtypes to de-fine the local composition to BMs in the various stagesof the disease.

Perhaps the most striking finding of the presentstudy is the markedly increased transcriptional activ-ity of BM-encoding genes in high grade (primaryGleason grade 4 and 5) and metastatic prostatic carci-nomas. Our previous immunohistochemical datashowed that poorly differentiated and metastatic car-cinomas retain BM, and express a2, b1 collagen-, anda6, b1 laminin-receptors, which are extensively coex-pressed with their corresponding ligands in peritu-moral BM formations [7,9,26]. Accordingly, highmRNA levels detected in these lesions obviously re-sult in an increase in biosynthesis and secretion of BMcomponents into the extracellular host tissue which, inturn, may enhance the tumorigenic potential of can-cerous cells to penetrate the extracellular matrix.

CONCLUSIONS

Results of the present study suggest that prostatecancer cells actively produce BM components anddocument high steady-state levels of BM-coding genesin poorly differentiated and metastatic tumors. Giventhe regulatory functions of BM components docu-mented in vitro, including cell adhesion, proliferation,and spreading, the present data may indicate thatthese extracellular macromolecules are important ele-ments in the process of tumor invasion and metastasesin human prostate cancer.

ACKNOWLEDGMENTS

Consultation in statistical methods and computa-tion was provided by Dr. J. Konig (Institut fur Medizi-nische Biometrie, Universitatskliniken Homburg,Homburg, Germany).

REFERENCES

1. Chung LWK: Fibroblasts are critical determinants in prostaticgrowth and dissemination. Cancer Metastasis Rev 1991;10:263–274.

2. Thompson TC, Timme TL, Park SHP, Ren C, Baley PA, EasthamJA, Sehgal J, Yang G, Kadmon D: Tissue and cell-cell interac-tions in prostate cancer progression. Cancer 1995;75:1885–1891.

3. Passaniti A, Isaacs JT, Haney JA, Adler SW, Cujdik T, Long PV,Kleinman HK: Stimulation of human prostatic carcinoma tumorgrowth in athymic mice and control of migration in culture byextracellular matrix. Int J Cancer 1992;51:318–324.

4. Ware JL: Prostate cancer progesssion. Implications of histopa-thology. Am J Pathol 1994;145:983–993.

5. Fong C-J, Sherwood ER, Sutkowski DM, Abu-Jawdeh GM, Yo-koo H, Bauer KD, Kozlowski JM, Lee C: Reconstituted basementmembrane promotes morphological and functional differentia-tion of primary human prostatic epithelial cells. Prostate 1991;9:221–235.

6. Fong C-J, Sherwood ER, Braun EJ, Berg LA, Lee C, KozlowskiJM: Regulation of prostatic carcinoma cell proliferation and se-cretory activity by extracellular matrix and stromal secretions.Prostate 1992;19:121–131.

7. Bonkhoff H, Wernert N, Dhom G, Remberger K: Basementmembrane in fetal, adult normal, hyperplastic and neoplastichuman prostate. Virchows Arch [A] 1991;418:375–381.

8. Nagle RB, Knox JD, Wolf C, Bowden GT, Cress AE: Adhesionmolecules, extracellular matrix, and proteases in prostate carci-noma. J Cell Biochem [Supl] 1994;19:232–237.

9. Bonkhoff H, Stein U, Remberger K: Differential expression of a6and a2 very late antigen integrins in the normal, hyperplastic,and neoplastic human prostate. Simultaneous demonstration ofcell surface receptors and their extracellular ligands. HumPathol 1993;24:243–248.

10. Knox JD, Cress AE, Clark V, Manriquez K-SA, Dalkin BL, NagleRB: Differential expression of extracellular matrix moleculesand the a6-integrins in the normal and neoplastic prostate. AmJ Pathol 1994;145:167–174.

11. Nagle RB, Hao J, Knox JD, Dalkin BL, Clark V, Cress AE: Ex-pression of hemidesmosomal and extracellular matrix proteinsby normal and malignant human prostate tissue. Am J Pathol1994;146:1498–1507.

12. Hao J, Yang Y, McDaniel KM, Dalkin BL, Cress AE, Nagle RB:Differential expression of laminin 5 (a3b3g2) by human malig-nant and normal prostate. Am J Pathol 1996;149:1341–1349.

13. Barsky SH, Siegal GP, Jannotta F, Liotta LA: Loss of basementmembrane components by invasive tumors but not by theirbenign counterparts. Lab Invest 1983;49:140–147.

14. Albelda SM: Biology of disease: Role of integrins and other celladhesion molecules in tumor progression and metastasis. LabInvest 1993;68:4–17.

15. Bosman FT: The borderline: Basement membranes and the tran-sition from premalignant to malignant neoplasia. Microsc ResTech 1994;28:216–225.

16. Flug M, Kopf-Maier P: The basement membrane and its involve-ment in carcinoma cell invasion. Acta Anat (Basel) 1995;152:69–84.

17. Pignatelli M, Vessey CJ: Adhesion molecules: Novel moleculartools in tumor pathology. Hum Pathol 1994;25:849–856.

18. Damjanovich L, Albelda SM, Mette SA, Buck CA: The distribu-tion of integrin cell adhesion receptors in normal and malignantlung tissue. Am J Respir Cell Mol Biol 1992;6:197–206.

19. Matsui K, Kitagawa M, Sugiyama S, Yamamoto K: Distributionpattern of the basement membrane components is one of thesignificant prognostic correlates in peripheral lung adenocarci-nomas. Hum Pathol 1995;26:186–194.

20. Havenith MG, Arends JW, Simon R, Volovics A, Wiggers T,Bosman FT: Type IV collagen immunoreactivity in colorectalcancer. Prognostic value of basement membrane deposition.Cancer 1988;62:2207–2211.

21. Pignatelli M, Smith MEF, Bodmer WF: Low expression of col-lagen receptor in moderate and poorly differentiated colorectaladenocarcinomas. Br J Cancer 1990;61:636–638.

22. Koretz K, Schlag P, Boumsell L, Moller P: Expression of VLA-a2,VLA-a6, and VLA-b1 chains in normal mucosa and adenomasof the colon, and in colon carcinomas and their liver metastases.Am J Pathol 1991;138:741–750.

23. Zutter MM, Mazoujian G, Santoro SA: Decreased expression of

BM Gene Expression in Prostate Tissue 149

integrin adhesive protein receptors in adenocarcinoma of thebreast. Am J Pathol 1990;137:863–870.

24. D’Ardenne AJ, Richman PI, Horton MA, Macaulay AE, Jordan S:Co-ordinate expression of the alpha-6 integrin laminin receptorsubunit and laminin in breast cancer. J Pathol 1991;165:213–220.

25. Zutter MM, Krigman HR, Santoro SA: Altered integrin expres-sion in adenocarcinoma of the breast. Am J Pathol 1993;142:1439–1447.

26. Bonkhoff H, Wernert N, Dhom G, Remberger K: Distribution ofbasement membranes in primary and metastatic carcinomas ofthe prostate. Hum Pathol 1992;23:934–;939.

27. Fuchs ME, Brawer MK, Rennels MA, Nagle RG: The relation-ship of basement membrane to histological grade of humanprostatic carcinoma. Mod Pathol 1989;2:105–111.

28. Sinha AA, Gleason DF, DeLeon OF, Wilson MJ, Limas C, ReddyPK, Furcht LT: Localization of type IV collagen in the basementmembranes of human prostate and lymph nodes by immuno-peroxidase and immunoalkaline phosphatase. Prostate 1991;18:93–104.

29. Timpl R, Dziadek M: Structure, development, and molecularpathology of basement membranes. Int Rev Exp Pathol 1987;29:1–112.

30. Timpl R, Brown JC: The laminins. Matrix Biol 1994;14:275–281.31. Burgeson RE, Chiquet M, Deutzmann R, Ekblom P, Engel J,

Kleinman H, Martin GR, Meneguzzi G, Paulsson M, Sanes J:A new nomenclature for the laminins. Matrix Biol 1994;14:209–211.

32. Sanes JR, Hunter DD, Green TL, Merlie JP: S-laminin. ColdSpring Harbor Symp Quant Biol 1990;55:419–430.

33. Rabinovitz I, Cress AE, Nagle RB: Biosynthesis and secretion oflaminin and S-laminin by human prostate carcinoma cell lines.Prostate 1994;25:97–107.

34. Bonkhoff H, Stein U, Remberger K: The proliferative function ofbasal cells in the normal and hyperplastic human prostate. Pros-tate 1994;24:1144–118.

35. Bonkhoff H, Stein U, Remberger K: Multidirectional differentia-tion in the normal, hyperplastic and neoplastic human prostate:Simultaneous demonstration of cell specific epithelial markers.Hum Pathol 1994;25:42–46.

36. Bonkhoff H, Remberger K: Differentiation pathways and histo-genetic aspects of normal and abnormal prostatic growth: Astem cell model. Prostate 1996;28:98–106.

37. Bonkhoff H: Role of the basal cells in premalignant changes ofthe human prostate: A stem cell concept for the development ofprostate cancer. Eur Urol 1996;30:201–205.

38. Montironi R, Bostwick DG, Bonkhoff H, Cockett ATK, HelpapB, Troncosco P, Waters D: Origins of prostate cancer. Cancer1996;78:362–365.

39. Brawer MK, Peehl DM, Stamy TA, Bostwick DG: Keratin im-munoreactivity in the benign and neoplastic human prostate.Cancer Res 1985;45:3663–3667.

40. Bostwick DG: High grade prostatic intraepithelial neoplasia:The most likely precursor of prostate cancer. Cancer 1995;75:1823–1836.

150 Pfohler et al.