Impaired Expression and PosttranslationalProcessing of Connexin43 and Downregulation of

Gap Junctional Communication in NeoplasticHuman Prostate Cells

Mohammad Z. Hossain,1* Ajit B. Jagdale,1 Peng Ao,1 Cosette LeCiel,2Ruo-Pan Huang,1 and Alton L. Boynton1

1Department of Molecular Medicine, Northwest Hospital, Seattle, Washington2Pacific Northwest Cancer Foundation, Seattle, Washington

BACKGROUND. Gap junctional communication (GJC) has been implicated in the control ofcell proliferation. Numerous cancer cells show a decrease or loss of GJC compared to theirnormal counterparts. Lack of adequate information on the status of gap junctions duringprostate neoplasia prompted us to examine this form of cancer, which comprises about 14%of male cancer deaths in America.METHODS. Cultured normal human prostate epithelial cells and several different humanprostate tumor lines were used in this study. GJC was assayed by dye transfer, whereasWestern blot and immunofluorescence methods were used to examine connexin43 (Cx43)levels and the presence of gap junctions, respectively.RESULTS. Normal human prostate cultures exhibited extensive cell-communication whichwas completely absent in all the examined tumor cells. This disrupted communication wasassociated with a decreased expression and an impaired posttranslational modification ofCx43 in these cells. Abundant immunostaining of gap junctional channels by a Cx43-antibodywas observed in normal prostate cells but not in tumor cells.CONCLUSIONS. Our data provide further support for the hypothesis that loss of junctionalcommunication is a critical step in progression to human prostate neoplasia. Prostate 38:55–59,1999. © 1999 Wiley-Liss, Inc.

KEY WORDS: gap junctions; connexin43; prostate tumor; cell-cell communication

INTRODUCTION

Intercellular communication via gap junctions is be-lieved to be involved in the regulation of cell growth[1]. These membrane channels between contactingcells are composed of connexins (Cx), a highly con-served group of membrane proteins [2], that allow thepassage of small molecules and ions up to 1,000 Dal-tons, some of which are presumed to act as growth-regulatory signals [1]. This function of gap junctions ishighly evident in cancer development, where loss ofcommunication competence is common among tumorcells [3]. Conversely, when junctional communicationis reestablished in tumor cells following transfectionwith connexins, a reversal of tumor phenotypes in

vitro [4–6] as well as decreased tumor incidences invivo [7] are observed. In addition, factors required fortumorigenesis, including oncogenes and tumor pro-moters, inhibit GJC [8,9] while antineoplastic agentsincrease GJC [10]. Based on these observations, a tu-mor-suppressive role was suggested for gap junc-tional communication [2–4].

Grant sponsor: NIH/NCI; Grant number: CA57064.Cosette LeCiel is currently at Zymogenetics, 1201 Eastlake Ave.East, Seattle, WA 98102.*Correspondence to: Mohammad Z. Hossain, Department of Mo-lecular Medicine, Northwest Hospital, 120 Northgate Plaza, Suite230, Seattle, WA 98125. E-mail: mhossain@nwhsea.orgReceived 11 March 1998; Accepted 5 May 1998

The Prostate 38:55–59 (1999)

© 1999 Wiley-Liss, Inc.

To evaluate and extend our understanding of therole of GJC in neoplasia, junctional communication incultured human normal and tumor prostate cells wasexamined. Gap junctions in prostate cancer have notbeen studied in detail, even though prostate cancer ishighly prevalent in men over age 50 and will claimabout 39,200 lives in the United States in 1998 [11]. Ourdata show a dramatic decrease in GJC as well as inexpression and processing of Cx43 in neoplastic pros-tate cells.

MATERIALS AND METHODS

Cell Culture

Primary human prostate epithelial cells were pur-chased from Clonetics (San Diego, CA) and weregrown as per the instructions of the supplier. Well-characterized neoplastic prostate cell lines, includingPC3, LNCaP, ALVA-31, and TSU-Pr1 [12–14], werecultured in RPMI-1640 with 5% fetal calf serum.

Gap Junctional Communication Assay

Gap junctional communication was assessed bytransfer of the fluorescent dye Lucifer yellow aftersingle-cell microinjection, performed as described pre-viously [10]. Briefly, glass micropipettes were pre-pared using a Flaming/Brown micropipette puller(Sutter Instrument Co., Baltimore, MD) and backfilledwith a 10% solution of Lucifer yellow in 0.33 M LiCl.Injection of cells in confluent cultures viewed by fluo-rescence illumination was achieved with pressure de-livery by a microinjector. After 10 min, the number ofneighboring cells exhibiting dye labeling was re-corded as an index of GJC. For each cell type, a mini-mum of 20–30 cells in different areas of cultures wasmicroinjected.

Western Blot Analysis of Cx43

Western blot analyses for Cx43 were performed asdescribed previously [15,16]. Briefly, confluent cellswere harvested in cold phosphate-buffered saline(PBS) containing 1 mM Na-orthovanadate, 10 mMNaF, and 1 mM phenylmethyl-sulfonyl fluoride(PMSF). The cell pellets were lysed in 1 mM NaHCO3containing 1 mM Na-orthovanadate, 10 mM NaF, and1 mM PMSF. Cell lysates were resolved on polyacryl-amide gels and transferred to polyvinyl difluoridemembranes. After blocking for 16 hr at 4°C in a solu-tion containing 5% nonfat dry milk in Tris-bufferedsaline and 0.2% Tween-20 (TBS-T), membranes wereincubated with an antibody recognizing the C-terminal region of Cx43. After several washes in TBS-

T, membranes were then incubated with horseradishperoxidase-conjugated donkey anti-rabbit IgG andprocessed with an enhanced chemiluminescence kit(Kirkegaard and Perry Laboratories, Gaithersburg,MD).

Immunofluorescence Detection of Cx43

For the immunodetection of Cx43 gap junctionalplaques, cells were grown on glass coverslips in 8-wellcell-culture trays [16]. Confluent cultures were rinsedtwice with PBS and then fixed in 3% formaldehyde for20–30 min at room temperature. They were then per-meabilized with 2% Triton X-100 and incubated infreshly prepared 1% sodium borohydride for 15 min.After washing with PBS, coverslips were incubated inanti-Cx43 antibody for 2 hr and then washed withPBS. After blocking for 30 min in normal goat serum,cultures were incubated with fluorescein isothiocya-nate FITC-conjugated goat anti-rabbit IgG antibody.Coverslips were washed three times in PBS andmounted onto glass slides with antifade medium, andslides were visually evaluated using a Zeiss epifluo-rescence microscope (Carl Zeiss Inc., Thornwood, NY)and photographed.

RESULTS AND DISCUSSION

Normal prostate epithelial cells showed high levelsof junctional communication (Fig. 1a–d). Immediatelyafter reaching confluence (5–6 days), they exhibitedmoderate levels of dye transfer (Fig. 1a,b). Dye trans-fer was observed among 95% of the 30–40 injectedcells, and on average, 8–14 adjacent cells became dye-positive (mean, 12.4 ± 1.3). Several days later (day 10),these cells exhibited variable compactness, with asmaller cell size and a cuboid morphology (Fig. 1c). Atthis stage, a dramatic rise in GJC was observed (Fig.1d). The mean number of communicating cells in-creased to 85.5 ± 8.2. In contrast, dye transfer wascompletely absent in neoplastic LNCaP cells (Fig. 1e,f).We have examined several other established prostatetumor lines, including PC3, ALVA-31, and TSU-Pr1[12–14], and transfer of the microinjected fluorescentdye could not be detected in any of these cell lines(data not shown), except in TSU-Pr1, which showedsome degree of cell communication (mean, 4.7 ± 1.2).

Previous studies showed Cx43 as a constituent ofprostate gap junctions [17,18]. The alteration of Cx43expression in prostate tumors is less well-understood.Decreased Cx43 immunostaining was reported inprostate tumors [16], while unchanged or increasedCx43 levels were also observed [18,19]. To determinewhether the observed poor GJC in prostate tumor cellsis associated with alterations in Cx43 levels, we exam-ined the expression of Cx43 in both normal and tumor

56 Hossain et al.

cells by Western blot analysis. In normal prostate cells,Cx43 exists in three molecular forms (Fig. 2, lanes 6–8):the 41-kD nonphosphorylated form (Cx43-NP), and a43–45-kD doublet (Cx43-P1 and Cx43-P2), represent-ing the phosphorylated forms of Cx43 observed inother tissues and cell types [15,16]. The phosphoryla-tion of Cx43 has been reported to be critical for theestablishment of functional gap junctions [20]. In mod-erately communicating 6-day-old cultures of nonneo-plastic human prostate cells (Fig. 2, lane 6), two mo-lecular forms of Cx43 (Cx43-NP and -P1) were pre-dominant; in 10-day-old cultures (Fig. 2, lanes 7 and8), we observed a several-fold increase in total Cx43level as well as the presence of all three forms, whichcorrelated with the high GJC in these cells (Fig. 1c,d).In another nonneoplastic human cell line, WI38 [21],all three forms of Cx43 were also seen (Fig. 2, lane 1)and correlated with its high cell communication abili-

ties (data not shown). In all the prostate tumor cells weexamined (Fig. 2, lanes 2–5), Cx43 expression wasdrastically reduced, especially in PC3 and LNCaPcells, where Cx43 expression was virtually undetect-able. Another major difference between the nonneo-plastic and the neoplastic prostate cells was the com-plete absence of any phosphorylated forms of Cx43 inthe latter.

Immunostaining of normal and tumor prostate cellswith anti-Cx43 antibody revealed a profile similar tothat of the Cx43 Western blots. In 6-day-old cultures ofnonneoplastic cells (Fig. 3a), Cx43 immunoreactivitywas mainly intracellular. In about 20% of cells, a punc-tate Cx43 staining was localized at cell-cell bound-aries, indicating the presence of gap junctionalplaques [16], although in most of the cells, spotty Cx43staining was present in a scattered fashion. In 10-day-old cultures of nonneoplastic cells, an overall increasein Cx43 staining was observed (Fig. 3b). In these cells,in addition to decreased intracellular staining, an in-crease in the size of Cx43-positive aggregates at cell-cell boundaries was seen, which is in agreement withthe high communication efficiency of these cells. Incontrast, the prostate tumor cell line LNCaP was de-void of any positive staining at cell-cell boundariesand only exhibited a faint intracellular staining (Fig.3c).

Lack of junctional communication in tumor cellshas been linked with three types of connexin abnor-mality: 1) downregulation of connexin expression[18,22]; 2) tyrosine-phosphorylation of connexin [8];and 3) impairment in the posttranslational processingof connexin [20]. Our results show a severe impair-ment in Cx43 expression and its posttranslational pro-cessing, resulting in a complete absence of gap junc-tions in neoplastic prostate cells, whereas normalprostate epithelial cells are communication-efficient.

Fig. 1. Gap junctional communication in normal and tumorprostate cells. Single cells in normal (a–d) or tumor (e,f) prostateculture were microinjected with Lucifer yellow and 5 min later,neighboring fluorescent cells were scored and photographed. a,b:Six-day-old normal prostate epithelial cells. c,d: Ten-day-old nor-mal prostate epithelial cells. e,f: Prostate tumor cell line LNCaP. a,c, and e are phase-contrast photographs corresponding to thefluorescent images b, d and f, respectively. Bar, 25 µm.

Fig. 2. Cx43 protein levels in normal and tumor prostate cells.Equal amounts (10 µg) of cellular lysates were separated on SDS-PAGE gels. The proteins were transferred to PVDF membrane,and the membrane was probed with Cx43-antibody followed byenhanced chemiluminescence detection. Different forms of Cx43were labeled. Lane 1, WI38; lane 2, PC3; lane 3, LNCaP; lane4, TSU-Pr1; lane 5, ALVA-31; lane 6, normal prostate epithelialcells (6 days old); lanes 7 and 8, normal prostate epithelial cells(10 days old). After ECL reaction, membranes were exposed for1–2 min to visualize Cx43 expression in tumor cells. Lane 8 wasobtained after a short (15–20 sec) exposure of lane 7 to identifydifferent forms of Cx43 in normal prostate cells.

Impaired GJC in Neoplastic Prostate Cells 57

The loss of junctional competence in all examinedprostate tumor cell lines indicates that it is perhaps arequired step in prostate neoplasia. Although Mehtaet al. [19] observed Cx43 expression in neoplastic butnot in normal prostate cells, our findings are in accordwith two reports showing the presence of Cx43 in nor-mal human prostate [17,18], which decreased in high-grade prostatic tumors [17]. Moreover, the communi-cation abilities of both normal and tumor prostate cellsare in complete agreement with their correspondinglevels of Cx43 expression observed in this study. Themechanism of the downregulation of connexin expres-sion is presently unknown. Transcriptional regulationof eukaryotic genes is generally modulated by protein-DNA interaction at specific nucleotide sequences atthe promoter region; characterization of Cx responseelements as well as identification of tumor-specificregulatory proteins will be required for solving this

puzzle. Another reason for the communication defi-ciency in tumor cells appears to be obstructed post-translational steps of Cx43 that generate Cx43-phosphoforms which are necessary for the formationof functional gap junctions [20] and are abundantlyfound in communicating tissues and cells [15,16]. Con-stitutive expression of cell adhesion molecules [20] orinhibition of glycosylation [23] has been shown toovercome this obstruction. It will be interesting to ex-amine whether similar approaches are useful in cor-recting abnormal Cx43 processing in neoplastic pros-tate cells, which are usually devoid of E-cadherin ex-pression [24].

Blockade of GJC in numerous neoplastic cells butnot in carcinogen-initiated cells [10] indicates a criticalrole of GJC in the development of the neoplastic phe-notype. It has been postulated that by shutting downthe communication channels, a tumor cell can com-pletely isolate itself from the control of growth-regulatory signals produced by surrounding normalcells and can thus undergo aberrant cell proliferationto eventually yield a tumor [25]. Although the identityof such growth-regulatory molecules traversing gapjunctions is yet to be determined, reversion of tumor-igenic phenotypes following restoration of junctionalcommunication [4–6] supports this hypothesis. Sincethe reversion of the neoplastic phenotype has beensuggested to be dependent upon connexin and tissuetypes [6], it remains to be seen whether the neoplasticphenotype of prostate cells can be reverted by an in-creased expression and/or normalization of posttrans-lational processing of Cx43.

ACKNOWLEDGMENTS

The authors thank Dr. Sai L. Su (Northwest Bio-therapeutics, Seattle, WA) for providing the TSU-Pr1cell line, and M. Bates for photography.

REFERENCES

1. Loewenstein WR, Rose B. The cell-cell channel in the control ofgrowth. Semin Cell Biol 1992;3:59–79.

2. Bruzzone R, White TW, Paul DL. Connection with connexins:The molecular basis of direct intercellular signaling. Eur J Bio-chem 1996;238:1–27.

3. Yamasaki H, Naus CCG. Role of connexin genes in growth con-trol. Carcinogenesis 1996;17:1199–1213.

4. Chen SC, Pelletier DB, Ao P, Boynton AL. Connexin43 reversesthe phenotype of transformed cells and alters their expression ofcyclin/cyclin-dependent kinases. Cell Growth Differ 1995;6:681–690.

5. Zhu D, Caveney S, Kidder GM, Naus CCG. Transfection of C6glioma cells with connexin43 cDNA: Analysis of expression,intercellular coupling, and cell proliferation. Proc Natl Acad SciUSA 1991;88:1883–1887.

6. Mesnil M, Krutovskikh V, Piccoli C, Elfgang C, Traub O, Wil-

Fig. 3. Immunofluorescent staining of Cx43 in fixed cultures.Cells grown on glass coverslips were fixed in 3% formaldehydeand, following Triton permeabilization, were incubated with anti-Cx43 antibody. Binding of Cx43 antibody was visualized by FITC-conjugated goat anti-rabbit IgG antibody. a: Six-day-old normalprostate culture. b: Ten-day-old normal prostate culture. c: Pros-tate tumor cell line LNCaP. Arrowheads in a and b indicate posi-tive staining of gap junctional plaques at the cell-cell boundary. Bar,12 µm.

58 Hossain et al.

lecke K, Yamasaki Y. Negative growth control of HeLa cells byconnexin genes: Connexin species specificity. Cancer Res 1995;55:629–639.

7. Naus CCG, Elisevitch K, Zhu D, Belliveau DJ, Del Maestro RF.In vivo growth of C6 glioma cells transfected with connexin43cDNA. Cancer Res 1992;52:4208–4213.

8. Crow DS, Beyer EC, Paul DL, Kobe SS, Lau AF. Phosphorylationof Cx43 gap junction protein in uninfected and Rous sarcomavirus-transformed mammalian fibroblasts. Mol Cell Biol 1990;10:1754–1763.

9. Oh SY, Grupen GS, Murray AW. Phorbol ester induces phos-phorylation and down-regulation of connexin43 in WB cells.Biochim Biophys Acta 1991;1094:243–248.

10. Hossain MZ, Wilkens LR, Mehta PP, Loewenstein WR, BertramJS. Enhancement of gap junctional communication by retinoidscorrelates with their ability to inhibit neoplastic transformation.Carcinogenesis 1989;10:1743–1748.

11. Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics,1998. CA 1998;48:6–29.

12. Loop SM, Rozanski TA, Ostenson RC. Human prostate tumorline, ALVA-31: A new model for studying the hormonal regu-lation of prostate tumor cell growth. Prostate 1993;22:93–108.

13. Webber MM, Bello D, Quader S. Immortalized and tumorigenicadult human prostatic epithelial cell lines: Characteristics andapplications. Part 2. Tumorigenic cell lines. Prostate 1997;30:58–64.

14. Iizumi T, Yazaki T, Kanoh S, Kondo I, Koiso K. Establishment ofa new prostatic carcinoma cell line (TSU-Pr1). J Urol 1987;137:1304–1306.

15. Hossain MZ, Murphy LJ, Hertzberg EL, Nagy JI. Phosphorylat-ed forms of connexin43 predominate in rat brain: Demonstra-tion by rapid inactivation of brain metabolism. J Neurochem1994;62:2394–2403.

16. Hossain MZ, Ao P, Boynton AL. Rapid disruption of gap junc-tional communication and phosphorylation of connexin43 by

platelet-derived growth factor in T51B rat liver epithelial cellsexpressing PDGF receptor. J Cell Physiol 1998;174:66–77.

17. Tsai H, Werber J, Davia MO, Edelman M, Tanaka KE, MelmanA, Christ GJ, Gelieber J. Reduced connexin43 expression in highgrade, human prostatic adenocarcinoma cells. Biochem BiophysRes Commun 1996;227:64–69.

18. Wilgenbus KK, Kirkpatrick CJ, Knuechel R, Willecke K, TraubO. Expression of Cx26, Cx32 and Cx43 gap junction proteins innormal and neoplastic human tissues. Int J Cancer 1992;51:522–529.

19. Mehta PP, Lokeshwar BL, Schiller PC, Bendix MV, OstensonRC, Howard GA, Roos BA. Gap junctional communication innormal and neoplastic prostate epithelial cells and its regulationby cAMP. Mol Carcinog 1996;15:18–32.

20. Musil LS, Cunningham BA, Edelman GM, Goodenough DA.Differential phosphorylation of the gap junction protein Cx43 injunctional communication-competent and -deficient cell lines. JCell Biol 1990;111:2077–2088.

21. Tupper JT, Smith JW. Growth factor regulation of membranetransport in human fibroblasts and its relationship to stimula-tion of DNA synthesis. J Cell Physiol 1985;125:443–448.

22. Lee SW, Tomasetto C, Paul D, Keyomarsi K, Sager R. Transcrip-tional downregulation of gap junctional proteins blocks junc-tional communication in human mammary tumor cell lines. JCell Biol 1992;118:1213–1221.

23. Wang Y, Mehta PP, Rose B. Inhibition of glycosylation inducesopen connexin43 cell-cell channels and phosphorylation andTriton X-100 insolubility of connexin43. J Biol Chem 1995;270:26581–26585.24. Umbas R, Schalken JA, Aalders TW, Carter BS,Karthaus HFM, Schaafsma HE, Debruyne FMJ, Isaaca WB. Ex-pression of cell adhesion molecule is reduced or absent in high-grade prostate cancer. Cancer Res 1992;52:5104–5109.

25. Mehta PP, Bertram JS, Loewenstein WR. Growth inhibition oftransformed cells correlates with their junctional communica-tion with normal cells. Cell 1986;44:187–196.

Impaired GJC in Neoplastic Prostate Cells 59