- Email: [email protected]
Phosphorylation of the gap junction protein Connexin43 in CIN III lesions and cervical carcinomas
Cancer Letters 235 (2006) 291–297 www.elsevier.com/locate/canlet
Phosphorylation of the gap junction protein Connexin43 in CIN III lesions and cervical carcinomas Inga Steinhoffa, Kerstin Leykaufb, Uwe Bleyla, Matthias Du¨rstc, Angel Alonsob,* a
Pathologisches Institut der Fakulta¨t fu¨r Klinische Medizin Mannheim, Ruprecht-Karls-Universita¨t Heidelberg, Theodor-Kutzer-Ufer 1-3, D-68135 Mannheim, Germany b Deutsches Krebsforschungszentrum, Im Neuenheimer Feld-242, D-69120 Heidelberg, Germany c Frauenklinik der Universita¨t Jena, Bachstrasse 18, D-07743 Jena, Germany Received 3 February 2005; received in revised form 20 April 2005; accepted 24 April 2005
Abstract Connexins are proteins that form the connexons, gap junction structures, which allow cells to communicate. Phosphorylation of connexins has been found to impair this communication. Using an antibody specifically recognizing the S279/S282phosphorylated form of connexin43 (Cx43) for immunohistochemistry, we have analysed Cx43 phosphorylation in normal epithelium, CIN III lesions, and carcinomas of the cervix. We found that in normal epithelium the basal layer was devoid of staining and most of the protein was localized in stratum spinosum and stratum granulosum. In pre-malignant CIN-III lesions Cx43 was strongly phosphorylated, but the basal layer was still negative. In squamous carcinomas, the cells were intensely stained. In these tumours, sites of strong staining were adjacent to less stained regions, suggesting that the tumours are intrinsically heterogeneous. Immunoblotting of proteins extracted from carcinomas with the specific antibody showed the classical pattern of multiple reacting bands, with the appearance of low migrating forms of the protein. Our results suggest that increased S279/S282 phosphorylation of Cx43 is the result of altered tissue structure rather than of cell malignization. q 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Lesions; Cervix; Stratified epithelium; Cell communication
1. Introduction Intercellular communication is mediated by gap junctions, structures composed of two hemi-channels/connexons which connect two neighbouring cells. Hemichannels are tubular structures formed by * Corresponding author. Tel.: C49 6221 423215; fax: C49 6221 424932. E-mail address: [email protected] (A. Alonso).
aggregation of six connexin molecules which, after connecting with the neighbouring hemichannel, form a tubular communication structure and allow transfer of water, ions, minerals, and molecules up to a molecular mass of 2000. The connexins are a family of genes coding for at least 20 proteins with molecular weights varying between 25 and 62 kDa [1]. The amount and type of connexins is cell-type-specific, and normally only connexons containing homo-hemichannels are formed, although hetero-connexons have
0304-3835/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2005.04.031
292
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
also been described. The connexins are membrane proteins with four trans-membrane regions, and are found mostly associated with the plasma membrane, although localization in the internal membranes and nuclei has also been described [2]. The proteins are co-translationally integrated into the endoplasmic reticulum to hemi-connexons and then transported by vesicle budding to the plasma membrane through the Golgi apparatus [2–4]. At the plasma membrane they associate with hemi-channels of the adjacent cells forming groups of gap junctions which appear as punctuate structures under the microscope. Tumour growth has been associated with changes in cell–cell communication. In tumour cells growing in vitro, a loss of dye-transfer, a probe for the failure of cells to communicate, has been repeatedly observed, suggesting that connexins are somehow involved in malignant growth [5]. Indeed, loss of connexin expression leads to increased metatastatic potential of the tumour cells when injected into nude mice, an effect which is reversed by delivering the corresponding connexin gene [6]. This has prompted several authors to propose the connexins as putative tumour suppressor genes. Altered expression of connexins has been found in a large number of different tumours, such as glial tumours, hepatocarcinomas, endometrial tumours or tumours of the cervix [7–11]. Immunofluorescence experiments have shown that human skin epithelium and cervix epithelium contain relatively high amounts of Cx43 [12,13]. In cervical lesions a progressive loss of Cx43 has been described, but information concerning the secondary modification of the protein during malignization is lacking [13]. Connexin43 contains at least 12 amino acids susceptible to phosphorylation, and it has been demonstrated that tyrosines as well as serine/ threonines can be phosphorylated [14]. This phosphorylation plays an important role in maintaining the physiology of the gap junctions. Increased connexin phosphorylation results in strong decrease of the cell–cell communication as measured by dye transfer in several cell types [14]. Nevertheless, in HPV16 E5-transfected cells, impaired dye transfer is associated with decreased phosphorylation of Cx43, thus suggesting that different mechanisms can account for substantial changes in cell communication [15]. Although the mechanisms responsible
for the effect of Cx43 phosphorylation on cell–cell communication have not been elucidated, it seems possible that phosphorylation of the protein results in a collapse of the gap junction, which therefore, becomes impermeable to the pass of low molecular weight substances [16]. We have developed an antibody which specifically recognizes S279/S282-phosphorylated Cx43 [17]. Using this antibody, we performed a series of experiments to analyse whether the phosphorylation of Cx43 is already altered in pre-malignant CIN lesions, and if loss of cell–cell communication can be associated with the malignization process. In this report we show that the phosphorylation pattern of Cx43 is greatly increased in developed carcinomas of the cervix.
2. Material and methods 2.1. Immunohistochemistry For immunohistochemistry paraffin slices were stained using the DAKO Envision system (Dako, K4010). The Cx43 antibody used in our experiments was from Zymed (Cat. Nr. 71-0700) and recognizes unphosphorylated as well as partially phosphorylated Cx43 protein. Antibodies SA226P and SA227P raised against the peptide H2N-CSSPTAPLpSMpSPPGYKL-CONH2 of Cx43 were produced in rabbits as previously described [17]. In the present study, a total number of 24 CIN III, 17 carcinomas, and 10 controls were analysed. 2.2. Immunoblots Slices of frozen specimen were prepared; the tumour regions were excised and incubated in 1% sodium dodecyl sulfate for 5 min at 95 8C. The extract was then centrifuged through a shredder column (Qiagen, Hilden, Germany) to shear the DNA, and the protein content was measured using the DC-system of Bio-Rad (Munich, Germany). Controls were proteins from normal human cervix epithelium. Proteins were separated by acrylamide gel electrophoresis and blotted onto polyvinilidine difluoride (PVDF) membranes. Incubation of
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
the membranes with the antibody was performed as previously described [17].
3. Results 3.1. Expression and serine-phosphorylation of connexin43 in normal cervical epithelium
293
phosphorylation of the connexin in all layers. Remarkably, mitotic cells showed a strong staining (Fig. 2B, stars and inset), indicating activation of the MAP kinases responsible for the phosphorylation of S279/S282 residues [14]. The same staining pattern was observed when antibody SA227P (raised also in rabbits against the same peptide as antibody SA226P) was used for the histochemical analyses. Further, no
In a first series of experiments, we analysed the distribution of total and phosphorylated connexin43 in cervix epithelium. As shown in Fig. 1A, immunohistochemical staining showed low amounts of connexin43 distributed rather homogeneously in all epithelial layers, with the exception of the basal layer. Staining was mostly localized to the plasma membranes and only faint reaction was observed in the cytoplasm. Staining with an antibody specific for the S279/S282 phosphorylated form of the connexin showed a different pattern (see Fig. 1B). The basal layer was negative, in accordance with total connexin43, but in addition the stratum spinosum was strongly stained. A weak staining was observed in the stratum granulosum, whereas the stratum corneum was devoid of any colour. Remarkably, not all cells of the stratum spinosum were homogeneously labelled, with brightly stained cells (see Fig. 1B, arrows) besides cells showing a weak reaction. 3.2. Connexin43 phosphorylation in CIN lesions A reduced content of Cx43 has been reported in cervical dysplasias in a limited study, but no systematic analysis has yet been reported [13]. Furthermore, an analysis of secondary modification of the protein in malignancies has not been reported. We therefore decided to study whether phosphorylation of Cx43 was modulated in CIN III lesions and in developed carcinomas. In CIN III lesions, loss of tissue architecture was accompanied by increased Cx43 phosphorylation compared with normal epithelium (Fig. 2B). Here also, and like the picture observed in the normal epithelium, strongly stained cells were adjacent to cells which show only faint staining. Furthermore, and as a consequence of the disturbance of the normal epithelial pattern the picture gives now the impression of a general
Fig. 1. Connexin43 distribution and phosphorylation in normal cervix epithelium. Slices from normal human cervix were prepared and immunostained with anti-Cx43 from Zymed (Nr. 71-0700), recognizing phosphorylated and non-phoshorylated Cx43 (A), or antibody SA226P recognizing only S279/S282-phosphorylated Cx43 (B). No staining is observed in stratum basale (sb) with either of the antibodies. A heterogeneous distribution of phosphorylated Cx43 is observed in stratum spinosum (ss) and stratum granulosum (sg, arrows), where a large proportion of the staining shows a cytoplasmic location. The stratum corneum (sc) is devoid of any staining.
294
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
Fig. 2. Phosphorylation of Cx43 in CIN III lesions (A, B) and cervical carcinomas (C, D). A, C: haematoxylin-eosin staining. B, D: immunostaining with antibody SA226P. In CIN III lesions a heterogeneous distribution of the staining is observed, with mitotic cells showing strong staining (panel B). The inset in panel B shows an enlarged picture of the mitotic cells labelled with stars. In carcinomas most of the cells are stained and some regions show increased, heterogeneous decoration (panel D, arrows). The stroma cells are negative. Only faint staining is observed in some capillaries.
staining was observed when pre-immune serum was used, indicating that the staining pattern obtained was specific (results not shown). 3.3. Connexin43 phosphorylation in developed carcinomas The overall pattern observed corresponds to a strong phosphorylation of Cx43 (Fig. 2D). Most of the cells displayed a strong staining, localized in both plasma membranes and cytoplasm. Interestingly, some tumour areas showed a heterogeneous staining, with islets of cells being strongly stained, beside others with only faint colour, suggesting that the tumour cells are intrinsically heterogeneous (Fig. 2D, arrows). To further substantiate these results, we analysed the phosphorylation status of Cx43 by immunoblotting. Protein extracts from six normal human cervixes or from six cervical carcinomas were prepared, separated on acrylamide gels, and blotted.
Incubation with antibody SA226P revealed three different bands, two of which were common for normal and malignized epithelium. In addition, a third band of lower molecular weight was present in all tumours, but absent in the normal epithelium (Fig. 3). This band was present in all 12 carcinomas analysed by immunoblotting, indicating that it was not the result of an artefact. The nature of this band is unknown, but interestingly a similar band has also been found in breast tumours and premalignant breast lesions (unpublished observations). Stripping of the membrane and re-incubation with antibody 71-0700 revealed one main reacting band. This band comigrates with the low-migrating band identified in the blots with antibody SA226P. After long time exposure of the film, two further bands were visible in normal tissues, but not in tumours (not observable in Fig. 3). These bands migrate with the known P0/P1 forms of connexin43, and confirm results describing a clear decrease of both isoforms in malignized tissues [9,13].
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
cervical tumours
54 04 54 N 32 54 N 28 54 N 26 54 N 02 54 N 24 47 N 49 50 T 07 37 T 27 39 T 86 38 T 42 51 T 83 T
normal tissue
75 50
37 SA226P 75 50
37 71-0700 Fig. 3. Immunoblotting of proteins from normal epithelium or from cervical tumours. Total cellular proteins from six cervical epithelia or from six squamous cell carcinomas were prepared, blotted, and incubated with antibody SA226P. In normal epithelia two reacting bands were observed with the phosphorylation-specific antibody. In tumours, a third reacting band is detectable, absent in all extracts from normal epithelia. Incubation of the blots with antibody 710700 reveals a main reacting band, common for normal and malignized epithelia. Molecular weights of marker proteins are shown on the left of the blot.
4. Discussion The experiments reported here demonstrate a clear expression of Cx43 in the stratified epithelium of the human cervix. As shown in Fig. 1, this expression is observed in stratum spinosum and stratum granulosum, whereas the basal cell layer is devoid of Cx43. An analogous pattern has been described for human neonatal epithelium and for the epithelium of the bulge region of the hair follicle indicating a similar expression pattern for all stratified epithelia [18]. The heterogeneous staining pattern of phosphorylated connexin43 in the suprabasal layers compared to the rather homogeneous staining with an antibody recognizing total Cx43, suggest physiological differences between the cells localized in a layer. The reason for these differences is unknown, but it may be speculated that the phosphorylation pattern correlates with the differentiation stage of the cells inside one epithelial layer. Remarkable is the fact that in strongly
295
staining cells most of the phosphorylated Cx43 is localized in the cytoplasm, whereas in weakly staining cells Cx43 is mainly found in the plasma membrane. It is generally accepted that malignization is accompanied by a reduced cell–cell communication. This reduction results in an isolation of the cells from their neighbours and loss of homeostatic control mechanisms. In addition, cell culture experiments have shown that decreased gap junction communication frequently correlates with an altered connexin phosphorylation [14]. Our results demonstrate that cervix carcinomas display an enhanced phosphorylation of Cx43 but with a somewhat heterogeneous character. Although a study of Cx43 phosphorylation has not been reported, it has been shown that its expression is reduced in most tumours so far analysed, and that this reduction sometimes adopts a heterogeneous pattern [8,9,19–21]. The staining pattern observed in normal epithelium suggests that phosphorylation of Cx43 is associated with an ordered epithelial structure. Thus it can be speculated that increased Cx43 phosphorylation in CIN III lesions and carcinomas is not associated with the malignization process, but rather with disturbance of the epithelial structure and perhaps with the differentiation stage of the cells. This is further supported by the fact that most of the CIN III lesions regress before reaching the stage of a carcinoma in situ, thus strongly suggesting that the loss of cell–cell communication (as a result of Cx43 phosphorylation) is not the cause of the malignization process. This apparently contrasts with experiments showing that transfection of connexin genes may reverse malignization in cell culture and animals [6,22]. Nevertheless, since no experimental data are available regarding the phosphorylation behaviour of the transfected connexins, it is currently not possible to correlate connexin amount, its phosphorylation and malignization as events of the same mechanism. Furthermore, the difficulties in quantifying histochemical results and the fact that the antibodies used to demonstrate Cx43 amount and phosphorylation have different affinities renders it impractical to correlate amount of connexins and phosphorylation in the histological specimens investigated. In this context it should be noted that cervical cell lines expressing human papillomaviruses are unable to communicate through gap junctions,
296
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
although this effect is probably independent of the presence of viral sequences [23]. An interesting observation is the presence of retarded bands in immunoblots. This retardation suggests that secondary protein modifications have occurred in tumours. This is most interesting, since we have identified similar retarded bands in premalignant and malignant lesions of the human breast. In addition, we have not been able to detect such retarded bands in proteins extracted from cell cultures. The reasons for this difference are unknown, but it may be speculated that loss of the tissue architecture is responsible for the changed phosphorylation pattern, although there are no experimental results available supporting this point. On the other hand, alternative explanations such as a changed kinase spectrum in cell cultures cannot be dismissed. The exact nature of the new band exclusively appearing in extracts of the tumours is currently unknown. This band migrates slightly faster than the unphosphorylated form of the connexin, suggesting that it is a proteolytic cleavage form. If this is true, one can speculate that this proteolysis is phosphorylation-specific, since the band is never observed upon incubation with the 71-0700 antibodies. Taken together, our results demonstrate an increase in Cx43 phosphorylation in CIN III lesions and cervical carcinomas. Nevertheless, this increased phosphorylation seems not to be the cause of the malignization process, but the result of an altered structural organization of the tissue.
Acknowledgements This work was been supported by a grant of the Deutsche Forchungsgemeinschaft to MD and AA.
References [1] G. So¨hl, K. Willecke, An update of connexin genes and their nomenclature in mouse and man, Cell Commun. Adhes. 10 (2003) 173–180. [2] M. Yeager, V.M. Unger, M.M. Falk, Synthesis, assembly and structure of gap junction intercellular channels, Curr. Opin. Struct. Biol. 8 (1998) 517–524.
[3] M.M. Falk, N.B. Gilula, Connexin membrane protein biosynthesis is influenced by polypeptide positioning within the translocon and signal peptide access, J. Biol. Chem. 273 (1998) 7856–7864. [4] M.M. Falk, N.M. Kumar, N.B. Gilula, Membrane insertion of gap junction connexins: polytopic channel forming membrane proteins, J. Cell Biol. 127 (1994) 343–355. [5] K. Cesen-Cummings, M.J. Fernstrom, A.M. Malkinson, R.J. Ruch, Frequent reduction of gap junctional intercellular communication and connexin43 expression in human and mouse lung carcinoma cells, Carcinogenesis 19 (1998) 61–67. [6] H. Qin, Q. Shao, H. Curtis, J. Galipeau, D.J. Belliveau, T. Wang, et al., Retroviral delivery of connexin genes to human breast cells inhibits in vivo tumour growth by a mechanism that is independent of significant gap junctional intercellular communication, J. Biol. Chem. 277 (2002) 29132–29138. [7] I.V. Budunova, S. Carbajal, T.J. Slaga, The expression of gap junctional proteins during different stages of mouse skin carcinogenesis, Carcinogenesis 16 (1995) 2717–2724. [8] J. Gee, M. Tanaka, H.B. Grossmann, Connexin26 is abnormally expressed in bladder cancer, J. Urol. 169 (2003) 1135–1137. [9] H. Habermann, V. Ray, W. Habermann, G.S. Prins, Alterations in gap junction protein expression in human benign prostatic hyperplasias and prostate cancer, J. Urol. 166 (2002) 2267–2272. [10] D.W. Laird, P. Fistouris, G. Batist, L. Alpert, H.T. Huynh, G.D. Carystinos, M.A. Alaoui-Jamali, Deficiency of connexin43 gap junctions is an independent marker for breast tumors, Cancer Res. 59 (1999) 4104–4110. [11] T. Saito, M. Nishimura, R. Kudo, H. Yamasaki, Suppressed gap junctional intercellular communication in carcinogenesis of endometrium, Int. J. Cancer 93 (2001) 317–323. [12] W.L. Di, E.L. Rugg, I.M. Leigh, D.P. Kelsell, Multiple epidermal connexins are expressed in different keratinocyte subpopulations including connexin 31, J. Invest. Dermatol. 117 (2001) 958–964. [13] T.J. King, L.H. Fukushima, A.D. Hieber, K.A. Shimabukuro, W.A. Sakr, J.S. Bertram, Reduced levels of connexin43 in cervical dysplasia: inducible expression in a cervical carcinoma cell line decreases neoplastic potential with implications for tumour progression, Carcinogenesis 21 (2000) 1097–1109. [14] P.D. Lampe, A.F. Lau, Regulation of gap junctions by phosphorylation of connexins, Arch. Biochem. Biophys. 384 (2000) 205–215. [15] I. Oelze, J. Kartenbeck, K. Crusius, A. Alonso, Human papillomavirus type 16 E5 protein affects cell–cell communication in an epithelial cell line, J. Virol. 69 (1995) 4489–4494. [16] B.J. Warn-Cramer, G.T. Cottrell, J.M. Burt, A.F. Lau, Regulation of connexin-43 gap junctional intercellular communication by mitogen-activated protein kinase, J. Biol. Chem. 273 (1998) 9188–9196.
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297 [17] K. Leykauf, M. Durst, A. Alonso, Phosphorylation and subcellular distribution of connexin43 in normal and stressed cells, Cell Tissue Res. 311 (2003) 23–30. [18] M. Matic, W.H. Evans, P.R. Brink, M. Simon, Epidermal stem cells do not communicate through gap junctions, J. Invest. Dermatol. 118 (2002) 110–116. [19] Y. Jinn, M. Ichioka, F. Marumo, Expresion of connexin32 and connexin43 gap junction proteins and E-cadherin in human lung cancer, Cancer Lett. 127 (1998) 161–169. [20] B. Schneider, M. Teschner, T. Sundermann, B. Pikula, J. Lautermann, Expression of gap junction proteins (connexin 26, 30, 32, 43) in normal mucosa, hyperkeratosis and carcinoma of the human larynx, ORL J. Otorhinolaryngol. Relat. Spec. 64 (2002) 324–329.
297
[21] S. Umhauer, R.J. Ruch, J. Fanning, Gap junction intercellular communication and connexin 43 expression in ovarian carcinoma, Am. J. Obstet. Gynecol. 182 (2000) 999–1000. [22] S.C. Chen, D.B. Pelletier, A.L. Boynton, Connexin43 reverses the phenotype of transformed cells and alters their expression of cyclin/cyclin-dependent kinases, Cell Growth Differ. 6 (1995) 681–690. [23] T. Aasen, M.B. Hodgins, M. Edward, S.V. Graham, The relationship between connexins, gap junctions, tissue architecture and tumour invasion, as studied in a novel in vitro model of HPV-16-associated cervical cancer progression, Oncogene 22 (2003) 7969–7980.
Phosphorylation of the gap junction protein Connexin43 in CIN III lesions and cervical carcinomas Inga Steinhoffa, Kerstin Leykaufb, Uwe Bleyla, Matthias Du¨rstc, Angel Alonsob,* a
Pathologisches Institut der Fakulta¨t fu¨r Klinische Medizin Mannheim, Ruprecht-Karls-Universita¨t Heidelberg, Theodor-Kutzer-Ufer 1-3, D-68135 Mannheim, Germany b Deutsches Krebsforschungszentrum, Im Neuenheimer Feld-242, D-69120 Heidelberg, Germany c Frauenklinik der Universita¨t Jena, Bachstrasse 18, D-07743 Jena, Germany Received 3 February 2005; received in revised form 20 April 2005; accepted 24 April 2005
Abstract Connexins are proteins that form the connexons, gap junction structures, which allow cells to communicate. Phosphorylation of connexins has been found to impair this communication. Using an antibody specifically recognizing the S279/S282phosphorylated form of connexin43 (Cx43) for immunohistochemistry, we have analysed Cx43 phosphorylation in normal epithelium, CIN III lesions, and carcinomas of the cervix. We found that in normal epithelium the basal layer was devoid of staining and most of the protein was localized in stratum spinosum and stratum granulosum. In pre-malignant CIN-III lesions Cx43 was strongly phosphorylated, but the basal layer was still negative. In squamous carcinomas, the cells were intensely stained. In these tumours, sites of strong staining were adjacent to less stained regions, suggesting that the tumours are intrinsically heterogeneous. Immunoblotting of proteins extracted from carcinomas with the specific antibody showed the classical pattern of multiple reacting bands, with the appearance of low migrating forms of the protein. Our results suggest that increased S279/S282 phosphorylation of Cx43 is the result of altered tissue structure rather than of cell malignization. q 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Lesions; Cervix; Stratified epithelium; Cell communication
1. Introduction Intercellular communication is mediated by gap junctions, structures composed of two hemi-channels/connexons which connect two neighbouring cells. Hemichannels are tubular structures formed by * Corresponding author. Tel.: C49 6221 423215; fax: C49 6221 424932. E-mail address: [email protected] (A. Alonso).
aggregation of six connexin molecules which, after connecting with the neighbouring hemichannel, form a tubular communication structure and allow transfer of water, ions, minerals, and molecules up to a molecular mass of 2000. The connexins are a family of genes coding for at least 20 proteins with molecular weights varying between 25 and 62 kDa [1]. The amount and type of connexins is cell-type-specific, and normally only connexons containing homo-hemichannels are formed, although hetero-connexons have
0304-3835/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2005.04.031
292
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
also been described. The connexins are membrane proteins with four trans-membrane regions, and are found mostly associated with the plasma membrane, although localization in the internal membranes and nuclei has also been described [2]. The proteins are co-translationally integrated into the endoplasmic reticulum to hemi-connexons and then transported by vesicle budding to the plasma membrane through the Golgi apparatus [2–4]. At the plasma membrane they associate with hemi-channels of the adjacent cells forming groups of gap junctions which appear as punctuate structures under the microscope. Tumour growth has been associated with changes in cell–cell communication. In tumour cells growing in vitro, a loss of dye-transfer, a probe for the failure of cells to communicate, has been repeatedly observed, suggesting that connexins are somehow involved in malignant growth [5]. Indeed, loss of connexin expression leads to increased metatastatic potential of the tumour cells when injected into nude mice, an effect which is reversed by delivering the corresponding connexin gene [6]. This has prompted several authors to propose the connexins as putative tumour suppressor genes. Altered expression of connexins has been found in a large number of different tumours, such as glial tumours, hepatocarcinomas, endometrial tumours or tumours of the cervix [7–11]. Immunofluorescence experiments have shown that human skin epithelium and cervix epithelium contain relatively high amounts of Cx43 [12,13]. In cervical lesions a progressive loss of Cx43 has been described, but information concerning the secondary modification of the protein during malignization is lacking [13]. Connexin43 contains at least 12 amino acids susceptible to phosphorylation, and it has been demonstrated that tyrosines as well as serine/ threonines can be phosphorylated [14]. This phosphorylation plays an important role in maintaining the physiology of the gap junctions. Increased connexin phosphorylation results in strong decrease of the cell–cell communication as measured by dye transfer in several cell types [14]. Nevertheless, in HPV16 E5-transfected cells, impaired dye transfer is associated with decreased phosphorylation of Cx43, thus suggesting that different mechanisms can account for substantial changes in cell communication [15]. Although the mechanisms responsible
for the effect of Cx43 phosphorylation on cell–cell communication have not been elucidated, it seems possible that phosphorylation of the protein results in a collapse of the gap junction, which therefore, becomes impermeable to the pass of low molecular weight substances [16]. We have developed an antibody which specifically recognizes S279/S282-phosphorylated Cx43 [17]. Using this antibody, we performed a series of experiments to analyse whether the phosphorylation of Cx43 is already altered in pre-malignant CIN lesions, and if loss of cell–cell communication can be associated with the malignization process. In this report we show that the phosphorylation pattern of Cx43 is greatly increased in developed carcinomas of the cervix.
2. Material and methods 2.1. Immunohistochemistry For immunohistochemistry paraffin slices were stained using the DAKO Envision system (Dako, K4010). The Cx43 antibody used in our experiments was from Zymed (Cat. Nr. 71-0700) and recognizes unphosphorylated as well as partially phosphorylated Cx43 protein. Antibodies SA226P and SA227P raised against the peptide H2N-CSSPTAPLpSMpSPPGYKL-CONH2 of Cx43 were produced in rabbits as previously described [17]. In the present study, a total number of 24 CIN III, 17 carcinomas, and 10 controls were analysed. 2.2. Immunoblots Slices of frozen specimen were prepared; the tumour regions were excised and incubated in 1% sodium dodecyl sulfate for 5 min at 95 8C. The extract was then centrifuged through a shredder column (Qiagen, Hilden, Germany) to shear the DNA, and the protein content was measured using the DC-system of Bio-Rad (Munich, Germany). Controls were proteins from normal human cervix epithelium. Proteins were separated by acrylamide gel electrophoresis and blotted onto polyvinilidine difluoride (PVDF) membranes. Incubation of
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
the membranes with the antibody was performed as previously described [17].
3. Results 3.1. Expression and serine-phosphorylation of connexin43 in normal cervical epithelium
293
phosphorylation of the connexin in all layers. Remarkably, mitotic cells showed a strong staining (Fig. 2B, stars and inset), indicating activation of the MAP kinases responsible for the phosphorylation of S279/S282 residues [14]. The same staining pattern was observed when antibody SA227P (raised also in rabbits against the same peptide as antibody SA226P) was used for the histochemical analyses. Further, no
In a first series of experiments, we analysed the distribution of total and phosphorylated connexin43 in cervix epithelium. As shown in Fig. 1A, immunohistochemical staining showed low amounts of connexin43 distributed rather homogeneously in all epithelial layers, with the exception of the basal layer. Staining was mostly localized to the plasma membranes and only faint reaction was observed in the cytoplasm. Staining with an antibody specific for the S279/S282 phosphorylated form of the connexin showed a different pattern (see Fig. 1B). The basal layer was negative, in accordance with total connexin43, but in addition the stratum spinosum was strongly stained. A weak staining was observed in the stratum granulosum, whereas the stratum corneum was devoid of any colour. Remarkably, not all cells of the stratum spinosum were homogeneously labelled, with brightly stained cells (see Fig. 1B, arrows) besides cells showing a weak reaction. 3.2. Connexin43 phosphorylation in CIN lesions A reduced content of Cx43 has been reported in cervical dysplasias in a limited study, but no systematic analysis has yet been reported [13]. Furthermore, an analysis of secondary modification of the protein in malignancies has not been reported. We therefore decided to study whether phosphorylation of Cx43 was modulated in CIN III lesions and in developed carcinomas. In CIN III lesions, loss of tissue architecture was accompanied by increased Cx43 phosphorylation compared with normal epithelium (Fig. 2B). Here also, and like the picture observed in the normal epithelium, strongly stained cells were adjacent to cells which show only faint staining. Furthermore, and as a consequence of the disturbance of the normal epithelial pattern the picture gives now the impression of a general
Fig. 1. Connexin43 distribution and phosphorylation in normal cervix epithelium. Slices from normal human cervix were prepared and immunostained with anti-Cx43 from Zymed (Nr. 71-0700), recognizing phosphorylated and non-phoshorylated Cx43 (A), or antibody SA226P recognizing only S279/S282-phosphorylated Cx43 (B). No staining is observed in stratum basale (sb) with either of the antibodies. A heterogeneous distribution of phosphorylated Cx43 is observed in stratum spinosum (ss) and stratum granulosum (sg, arrows), where a large proportion of the staining shows a cytoplasmic location. The stratum corneum (sc) is devoid of any staining.
294
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
Fig. 2. Phosphorylation of Cx43 in CIN III lesions (A, B) and cervical carcinomas (C, D). A, C: haematoxylin-eosin staining. B, D: immunostaining with antibody SA226P. In CIN III lesions a heterogeneous distribution of the staining is observed, with mitotic cells showing strong staining (panel B). The inset in panel B shows an enlarged picture of the mitotic cells labelled with stars. In carcinomas most of the cells are stained and some regions show increased, heterogeneous decoration (panel D, arrows). The stroma cells are negative. Only faint staining is observed in some capillaries.
staining was observed when pre-immune serum was used, indicating that the staining pattern obtained was specific (results not shown). 3.3. Connexin43 phosphorylation in developed carcinomas The overall pattern observed corresponds to a strong phosphorylation of Cx43 (Fig. 2D). Most of the cells displayed a strong staining, localized in both plasma membranes and cytoplasm. Interestingly, some tumour areas showed a heterogeneous staining, with islets of cells being strongly stained, beside others with only faint colour, suggesting that the tumour cells are intrinsically heterogeneous (Fig. 2D, arrows). To further substantiate these results, we analysed the phosphorylation status of Cx43 by immunoblotting. Protein extracts from six normal human cervixes or from six cervical carcinomas were prepared, separated on acrylamide gels, and blotted.
Incubation with antibody SA226P revealed three different bands, two of which were common for normal and malignized epithelium. In addition, a third band of lower molecular weight was present in all tumours, but absent in the normal epithelium (Fig. 3). This band was present in all 12 carcinomas analysed by immunoblotting, indicating that it was not the result of an artefact. The nature of this band is unknown, but interestingly a similar band has also been found in breast tumours and premalignant breast lesions (unpublished observations). Stripping of the membrane and re-incubation with antibody 71-0700 revealed one main reacting band. This band comigrates with the low-migrating band identified in the blots with antibody SA226P. After long time exposure of the film, two further bands were visible in normal tissues, but not in tumours (not observable in Fig. 3). These bands migrate with the known P0/P1 forms of connexin43, and confirm results describing a clear decrease of both isoforms in malignized tissues [9,13].
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
cervical tumours
54 04 54 N 32 54 N 28 54 N 26 54 N 02 54 N 24 47 N 49 50 T 07 37 T 27 39 T 86 38 T 42 51 T 83 T
normal tissue
75 50
37 SA226P 75 50
37 71-0700 Fig. 3. Immunoblotting of proteins from normal epithelium or from cervical tumours. Total cellular proteins from six cervical epithelia or from six squamous cell carcinomas were prepared, blotted, and incubated with antibody SA226P. In normal epithelia two reacting bands were observed with the phosphorylation-specific antibody. In tumours, a third reacting band is detectable, absent in all extracts from normal epithelia. Incubation of the blots with antibody 710700 reveals a main reacting band, common for normal and malignized epithelia. Molecular weights of marker proteins are shown on the left of the blot.
4. Discussion The experiments reported here demonstrate a clear expression of Cx43 in the stratified epithelium of the human cervix. As shown in Fig. 1, this expression is observed in stratum spinosum and stratum granulosum, whereas the basal cell layer is devoid of Cx43. An analogous pattern has been described for human neonatal epithelium and for the epithelium of the bulge region of the hair follicle indicating a similar expression pattern for all stratified epithelia [18]. The heterogeneous staining pattern of phosphorylated connexin43 in the suprabasal layers compared to the rather homogeneous staining with an antibody recognizing total Cx43, suggest physiological differences between the cells localized in a layer. The reason for these differences is unknown, but it may be speculated that the phosphorylation pattern correlates with the differentiation stage of the cells inside one epithelial layer. Remarkable is the fact that in strongly
295
staining cells most of the phosphorylated Cx43 is localized in the cytoplasm, whereas in weakly staining cells Cx43 is mainly found in the plasma membrane. It is generally accepted that malignization is accompanied by a reduced cell–cell communication. This reduction results in an isolation of the cells from their neighbours and loss of homeostatic control mechanisms. In addition, cell culture experiments have shown that decreased gap junction communication frequently correlates with an altered connexin phosphorylation [14]. Our results demonstrate that cervix carcinomas display an enhanced phosphorylation of Cx43 but with a somewhat heterogeneous character. Although a study of Cx43 phosphorylation has not been reported, it has been shown that its expression is reduced in most tumours so far analysed, and that this reduction sometimes adopts a heterogeneous pattern [8,9,19–21]. The staining pattern observed in normal epithelium suggests that phosphorylation of Cx43 is associated with an ordered epithelial structure. Thus it can be speculated that increased Cx43 phosphorylation in CIN III lesions and carcinomas is not associated with the malignization process, but rather with disturbance of the epithelial structure and perhaps with the differentiation stage of the cells. This is further supported by the fact that most of the CIN III lesions regress before reaching the stage of a carcinoma in situ, thus strongly suggesting that the loss of cell–cell communication (as a result of Cx43 phosphorylation) is not the cause of the malignization process. This apparently contrasts with experiments showing that transfection of connexin genes may reverse malignization in cell culture and animals [6,22]. Nevertheless, since no experimental data are available regarding the phosphorylation behaviour of the transfected connexins, it is currently not possible to correlate connexin amount, its phosphorylation and malignization as events of the same mechanism. Furthermore, the difficulties in quantifying histochemical results and the fact that the antibodies used to demonstrate Cx43 amount and phosphorylation have different affinities renders it impractical to correlate amount of connexins and phosphorylation in the histological specimens investigated. In this context it should be noted that cervical cell lines expressing human papillomaviruses are unable to communicate through gap junctions,
296
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297
although this effect is probably independent of the presence of viral sequences [23]. An interesting observation is the presence of retarded bands in immunoblots. This retardation suggests that secondary protein modifications have occurred in tumours. This is most interesting, since we have identified similar retarded bands in premalignant and malignant lesions of the human breast. In addition, we have not been able to detect such retarded bands in proteins extracted from cell cultures. The reasons for this difference are unknown, but it may be speculated that loss of the tissue architecture is responsible for the changed phosphorylation pattern, although there are no experimental results available supporting this point. On the other hand, alternative explanations such as a changed kinase spectrum in cell cultures cannot be dismissed. The exact nature of the new band exclusively appearing in extracts of the tumours is currently unknown. This band migrates slightly faster than the unphosphorylated form of the connexin, suggesting that it is a proteolytic cleavage form. If this is true, one can speculate that this proteolysis is phosphorylation-specific, since the band is never observed upon incubation with the 71-0700 antibodies. Taken together, our results demonstrate an increase in Cx43 phosphorylation in CIN III lesions and cervical carcinomas. Nevertheless, this increased phosphorylation seems not to be the cause of the malignization process, but the result of an altered structural organization of the tissue.
Acknowledgements This work was been supported by a grant of the Deutsche Forchungsgemeinschaft to MD and AA.
References [1] G. So¨hl, K. Willecke, An update of connexin genes and their nomenclature in mouse and man, Cell Commun. Adhes. 10 (2003) 173–180. [2] M. Yeager, V.M. Unger, M.M. Falk, Synthesis, assembly and structure of gap junction intercellular channels, Curr. Opin. Struct. Biol. 8 (1998) 517–524.
[3] M.M. Falk, N.B. Gilula, Connexin membrane protein biosynthesis is influenced by polypeptide positioning within the translocon and signal peptide access, J. Biol. Chem. 273 (1998) 7856–7864. [4] M.M. Falk, N.M. Kumar, N.B. Gilula, Membrane insertion of gap junction connexins: polytopic channel forming membrane proteins, J. Cell Biol. 127 (1994) 343–355. [5] K. Cesen-Cummings, M.J. Fernstrom, A.M. Malkinson, R.J. Ruch, Frequent reduction of gap junctional intercellular communication and connexin43 expression in human and mouse lung carcinoma cells, Carcinogenesis 19 (1998) 61–67. [6] H. Qin, Q. Shao, H. Curtis, J. Galipeau, D.J. Belliveau, T. Wang, et al., Retroviral delivery of connexin genes to human breast cells inhibits in vivo tumour growth by a mechanism that is independent of significant gap junctional intercellular communication, J. Biol. Chem. 277 (2002) 29132–29138. [7] I.V. Budunova, S. Carbajal, T.J. Slaga, The expression of gap junctional proteins during different stages of mouse skin carcinogenesis, Carcinogenesis 16 (1995) 2717–2724. [8] J. Gee, M. Tanaka, H.B. Grossmann, Connexin26 is abnormally expressed in bladder cancer, J. Urol. 169 (2003) 1135–1137. [9] H. Habermann, V. Ray, W. Habermann, G.S. Prins, Alterations in gap junction protein expression in human benign prostatic hyperplasias and prostate cancer, J. Urol. 166 (2002) 2267–2272. [10] D.W. Laird, P. Fistouris, G. Batist, L. Alpert, H.T. Huynh, G.D. Carystinos, M.A. Alaoui-Jamali, Deficiency of connexin43 gap junctions is an independent marker for breast tumors, Cancer Res. 59 (1999) 4104–4110. [11] T. Saito, M. Nishimura, R. Kudo, H. Yamasaki, Suppressed gap junctional intercellular communication in carcinogenesis of endometrium, Int. J. Cancer 93 (2001) 317–323. [12] W.L. Di, E.L. Rugg, I.M. Leigh, D.P. Kelsell, Multiple epidermal connexins are expressed in different keratinocyte subpopulations including connexin 31, J. Invest. Dermatol. 117 (2001) 958–964. [13] T.J. King, L.H. Fukushima, A.D. Hieber, K.A. Shimabukuro, W.A. Sakr, J.S. Bertram, Reduced levels of connexin43 in cervical dysplasia: inducible expression in a cervical carcinoma cell line decreases neoplastic potential with implications for tumour progression, Carcinogenesis 21 (2000) 1097–1109. [14] P.D. Lampe, A.F. Lau, Regulation of gap junctions by phosphorylation of connexins, Arch. Biochem. Biophys. 384 (2000) 205–215. [15] I. Oelze, J. Kartenbeck, K. Crusius, A. Alonso, Human papillomavirus type 16 E5 protein affects cell–cell communication in an epithelial cell line, J. Virol. 69 (1995) 4489–4494. [16] B.J. Warn-Cramer, G.T. Cottrell, J.M. Burt, A.F. Lau, Regulation of connexin-43 gap junctional intercellular communication by mitogen-activated protein kinase, J. Biol. Chem. 273 (1998) 9188–9196.
I. Steinhoff et al. / Cancer Letters 235 (2006) 291–297 [17] K. Leykauf, M. Durst, A. Alonso, Phosphorylation and subcellular distribution of connexin43 in normal and stressed cells, Cell Tissue Res. 311 (2003) 23–30. [18] M. Matic, W.H. Evans, P.R. Brink, M. Simon, Epidermal stem cells do not communicate through gap junctions, J. Invest. Dermatol. 118 (2002) 110–116. [19] Y. Jinn, M. Ichioka, F. Marumo, Expresion of connexin32 and connexin43 gap junction proteins and E-cadherin in human lung cancer, Cancer Lett. 127 (1998) 161–169. [20] B. Schneider, M. Teschner, T. Sundermann, B. Pikula, J. Lautermann, Expression of gap junction proteins (connexin 26, 30, 32, 43) in normal mucosa, hyperkeratosis and carcinoma of the human larynx, ORL J. Otorhinolaryngol. Relat. Spec. 64 (2002) 324–329.
297
[21] S. Umhauer, R.J. Ruch, J. Fanning, Gap junction intercellular communication and connexin 43 expression in ovarian carcinoma, Am. J. Obstet. Gynecol. 182 (2000) 999–1000. [22] S.C. Chen, D.B. Pelletier, A.L. Boynton, Connexin43 reverses the phenotype of transformed cells and alters their expression of cyclin/cyclin-dependent kinases, Cell Growth Differ. 6 (1995) 681–690. [23] T. Aasen, M.B. Hodgins, M. Edward, S.V. Graham, The relationship between connexins, gap junctions, tissue architecture and tumour invasion, as studied in a novel in vitro model of HPV-16-associated cervical cancer progression, Oncogene 22 (2003) 7969–7980.