Restoration of Cell-to-Cell Communication in Thyroid Cell Lines by Transfection with and Stable Expression of the Connexin-32 Gene

Normal thyroid epithelial cells coexpress connexin-32 and connexin-43, which form distinct gap junctions. In primary culture, connexin-43 is expressed by thyrocytes in monolayers or reorganized into follicles, whereas the expression of connexin-32 is dependent upon the reconstitution of follicles. To study the functional impact of connexin-32 gap junctions in thyroid cells, we transfected connexin-32 cDNA in two thyroid-derived communication-deficient cell lines, FRT and FRTL-5. The selected clones, which stably expressed connexin-32 at high levels and exhibited high gap junction-mediated dye-coupling, presented a reduced proliferation rate as compared with that of the corresponding wild-type FRT and FRTL-5 cells; the mean population doubling time was increased by ∼35%. The proliferation of connexin-32-transfected FRTL-5 cells remained thyrotropin-dependent; the range of thyrotropin concentrations that stimulated growth was the same in transfected and control cells. The expression of connexin-32 led to an increase of thyroglobulin gene expression in FRTL-5 cells. The expression of two other tissue-specific proteins, thyroid transcription factor-1 and Pax-8, was unchanged. These findings provide evidence that connexin-32 gap junction-mediated cell-to-cell communication participates in the control of growth and differentiation of thyroid cells.

Gap junctions (GJ) are ubiquitous intercellular junctions allowing the cell-to-cell exchange of small cytoplasmic molecules. These exchanges have long been thought to play a role in the regulation of cell growth and cell differentiation in a number of tissues (1,2). GJ proteins, connexins (Cx), 1 belong to a multigenic family, each member presenting its own tissuespecific distribution. Despite strong overall homologies between Cx, functional properties (including permeability properties) differ from one Cx to the other. In most tissues, the determination of the role of a given Cx is complicated by the coexpression of several other different Cx (for reviews, see Refs. [3][4][5][6][7]. This applies to endocrine glands (8,9) including the thyroid. Previous work from our laboratory has established that polarized thyroid epithelial cells or thyrocytes present an unusual Cx expression pattern; they coexpress Cx32 and Cx43. Cx32 GJ are scattered along the lateral domain of the plasma membrane, and Cx43 GJ are precisely located within the tight junction network (10). Using pig thyrocytes in primary culture, which, under defined culture conditions, either spread in monolayers or reconstitute follicles, we have shown that the Cx43 gene is always expressed. By contrast, the Cx32 gene is expressed only in thyrocytes that reorganize into follicles (11). Cx32 GJ and Cx43 GJ that reconstitute in these in vitro systems are functional, and thyrotropin (TSH) controls both Cx synthesis and the level of junctional coupling between thyrocytes (11,12). The respective roles of Cx32 GJ and Cx43 GJ in thyroid gland functioning are not known. Cx32 GJ could play a prominent role since Cx32 expression correlates with the expression of thyroid histiotypic differentiation, i.e. follicle formation. To try to document the functional impact of cell-to-cell communication via Cx32 GJ, we have chosen to transfect two rat thyroid-derived cell lines (FRT and FRTL-5) that are communication-deficient with the rat Cx32 cDNA. Several clones that stably expressed high levels of Cx32 have been isolated from each cell line. All of them presented a high level of GJmediated cell-to-cell communication and exhibited a decreased growth rate as compared with the corresponding parental cells. Interestingly, stable expression of Cx32 by differentiated FRTL-5 cells led to up-regulation of the thyroglobulin gene.

EXPERIMENTAL PROCEDURES
Expression Vector-The 1.5-kb pCx32 cDNA (containing the entire coding region) was isolated from the pGEM-3 plasmid (kindly provided by Dr. D. L. Paul) (13) by EcoRI site digestion. Cx32 cDNA was inserted into the pSVK3 plasmid (Pharmacia Biotech Inc.) after linearization and dephosphorylation at its unique EcoRI site, yielding pSVK3-Cx32. The proper orientation of the cDNA insert was controlled by electrophoretic analysis of KpnI and PstI restriction enzyme digestion fragments and nucleotide sequencing of the appropriate region.
Cell Culture and Establishment of Stable Transfectants-FRT and FRTL-5 cells were kindly provided by Prof. L. Nitsch (Dipartimento di Biologia e Patologia Cellulare e Moleculare "L. Califano," Universita Degli Studi di Napoli Federico II, (Naples, Italy). These two cell lines, derived from normal thyroids of Fischer rats, have been previously characterized (14,15). FRTL-5 cells were grown in complete culture medium composed of Coon's modified Ham's F-12 medium (Seromed, Berlin, Germany) containing 100 units/ml penicillin and 100 g/ml streptomycin and supplemented with 5% calf serum and a five-hormone mixture (10 g/ml insulin, 10 nM hydrocortisone, 5 g/ml transferrin, 10 ng/ml glycyl-L-histidyl-L-lysine acetate, and 1 milliunits/ml TSH) as described (15). All hormones were from Sigma. FRT cells were cultured in the same medium as FRTL-5 cells, except that TSH was omitted and calf serum was replaced by fetal calf serum (Sigma). Cell cultures were maintained at 37°C under a 95% air and 5% CO 2 humidified atmosphere and were routinely subcultured by trypsinization with a change of medium twice weekly.
FRT and FRTL-5 cells (ϳ3-5 ϫ 10 5 cells/100-mm Petri dish) were cotransfected with two plasmids (pSVK3-Cx32 and pCMV-neo) using the calcium phosphate precipitation procedure followed by a glycerol shock. The optimal conditions of transfection were slightly different for the two cell lines. FRT cells were incubated for 3-4 h with calcium phosphate in the presence of 20 -30 g of pSVK3-Cx32 and 2-3 g of pCMV-neo and then subjected to a 1-min glycerol shock. For FRTL-5 cells, the incubation in the presence of calcium phosphate was reduced to 1 h, and then cells were subjected to a 3-min glycerol shock. For both cell types, 48 h after the outset of transfection, the neomycin analogue, G418 (Life Technologies, Inc.), was added to the medium at a concentration of 0.4 mg/ml. The medium containing G418 was changed every 3 days. After 3 weeks of selective pressure in the presence of G418, neomycin-resistant FRT colonies were picked by trypsinization in cloning cylinders and grown separately under selective conditions. In the case of FRTL-5 cells, only neomycin-resistant colonies judged to be well coupled by microinjection of lucifer yellow (see "Measurement of Intercellular Communication") were picked up and grown for subsequent analyses. For controls, cells were cotransfected under the conditions described above, but with pCMV-neo and the pSVK3 vector lacking Cx32 cDNA. FRT and FRTL-5 transfectants were maintained in the appropriate media containing G418 at a concentration of 0.05 mg/ml.
Northern Blot Analysis-cDNAs encoding rat Cx26 (clone Cx26-1; 1.1 kbp), rat Cx32 (1.5 kbp), and rat Cx43 (clone G1; 2.5 kbp) originate from Drs. B. J. Nicholson (16), D. L. Paul (13), and E. C. Beyer (17), respectively. cDNA clone G1 contains 92% of the coding region for Cx43; the other two cDNAs contain the complete coding regions of Cx26 and Cx32. G1 Cx43 cDNA was extracted from the Bluescript plasmid using EcoRI and then digested with StuI (Promega); this EcoRI-StuI fragment (1.03 kbp; positions 293-1323) was used for hybridization experiments. Cx32 cDNA as well as Cx26 cDNA were extracted from the pGEM-3 plasmid by EcoRI site digestion. The thyroglobulin cDNA probe corresponds to the human thyroglobulin 0.7-kbp M1 fragment (18). The rat Pax-8 cDNA fragment (0.3 kbp), corresponding to the paired domain fragment (19), was extracted from the C27 B2xx22 plasmid by EcoRI and HindIII restriction site digestion. The rat TTF-1 cDNA fragment (0.6 kbp), corresponding to the 3Ј-untranslated region, was extracted from the Bluescript THA plasmid (20). cDNA probes, prepared from DNA fragments isolated by electrophoresis and purified using the QIAEX gel extraction kit (QIAGEN Inc.), were labeled with [␣-32 P]dCTP by random hexanucleotide primed synthesis. Total RNA was isolated from wild-type FRT and FRTL-5 cells, stable transfected clones, and control rat tissues (liver, cervix, and heart) using the guanidinium isothiocyanate/acid phenol extraction method of Chomczynski and Sacchi (21). Northern blot analysis of total RNA (20 g/sample) was performed as described (22). Hybridized membranes were exposed to Kodak X-Omat AR film (Eastman Kodak Co.) at Ϫ80°C with intensifying screens. The efficiency of transfer and the integrity of RNA were checked by hybridization with a ␤-actin probe.
Immunofluorescence Labeling-Cells attached to Petri dishes fixed with 4% paraformaldehyde in phosphate-buffered saline containing 0.25% Triton X-100 for 30 min at 20°C were incubated with anti-Cx32 antibodies diluted 1:1000 in phosphate-buffered saline containing 1 mg/ml bovine serum albumin for 1 h at 20°C. Anti-Cx32 antibodies were generated in rabbit against a synthetic peptide corresponding to the sequence of the cytoplasmic loop of rat Cx32 (residues 98 -124) (10). A goat anti-rabbit Ig F(abЈ) 2 fragment conjugated to fluorescein (Sigma) was used as secondary antibody. Immunofluorescent images were taken using a SIT camera (LHESA Electronique, Cercy Pontoise, France) installed on an Axiophot microscope (Zeiss, Oberkochen, Germany) coupled to an image-processing system (Sapphire, Quantel, Montigny-le-Bretonneux, France). Photomicrographs were prepared using a video printer (UP 5000 P, Sony, Tokyo).
Measurement of Intercellular Communication-Cell-to-cell communication was analyzed using the low M r fluorescent probe lucifer yellow CH (Sigma), originally described by Stewart (23). The probe was microinjected into one cell, and after diffusion to adjacent cells, the number of lucifer yellow-labeled cells was counted as described previously (12).
Growth Rate Determination-Cells were seeded at low density in multiwell culture dishes. After different culture times, cells were dissociated by mild trypsinization, and the number of cells/dish was counted under an inverted microscope equipped with phase-contrast optics using an hemocytometer.

RESULTS
FRT and FRTL-5 cells do not express the Cx expressed in normal thyrocytes. No Cx32 transcript was detected in total RNA from FRT and FRTL-5 cells by Northern blotting (Fig. 1A, lanes 2 and 7, respectively); likewise, Cx43 and Cx26 transcripts could not be found (data not shown). The same negative results were also obtained at the protein level. Neither Cx32 (Fig. 1B, lanes 2 and 7) nor Cx43 (data not shown) was detected by Western blot analyses of FRT or FRTL-5 detergent-resistant membrane extracts. Immunofluorescence labeling with anti-Cx32 (Fig. 2, A and E) or anti-Cx43 (data not shown) antibodies was also negative. The inability to detect any of the three Cx (Cx32, Cx43, and Cx26) was in keeping with the absence of or the very low level of junctional coupling analyzed by lucifer yellow microinjection. In FRTL-5 cells, lucifer yellow was never transmitted from the injected cells to surrounding cells as illustrated Fig. 3B (panel b). In FRT cells, the dye was sometimes detected in some cells (two to five cells) adjacent to the injected cell several minutes after microinjection (Fig. 3A,  panel b). The low level of coupling observed in FRT cells could be either related to the presence of a low number of GJ channels composed of Cx32, Cx43, or Cx26 expressed in minute amounts or dependent on channels composed of another Cx than those normally expressed in thyroid cells.
Cx32 Expression in FRT and FRTL-5 Cells Transfected with the Rat Cx32 cDNA-Cx32-transfected FRT cells were identi- fied by the presence of the 1.6-kb Cx32 mRNA. Among ϳ70 clones, the three clones expressing the highest levels of Cx32 mRNA (clones A, B, and C) (Fig. 1A) were selected and used for subsequent studies. The Cx32 mRNA content of clones A and C (Fig. 1A, lanes 4 and 6) was higher than that of clone B (lane 5).
FRTL-5 cells expressing the Cx32 gene were identified by their high junctional coupling. Three clones (clones I, P, and Q) were selected; they expressed similar levels of Cx32 transcripts (Fig.  1A, lanes 9 -11). Cx32 mRNA was absent in FRT and FRTL-5 cells cotransfected with the neomycin gene and the empty pSVK3 vector (Fig. 1A, lanes 3 and 8). The expression of the Cx32 gene in FRT and FRTL-5 cells was further analyzed by Western blotting and immunofluorescence labeling. The Cx32 protein of the expected size (27)(28) was immunodetected in all the Cx32 mRNA-positive clones (Fig. 1B, lanes 4 -6 for FRT cell-derived clones and lanes 9 -11 for FRTL-5 cell-derived clones). As expected from mRNA data, the amount of Cx32 protein in clone B of FRT cells was lower than that detected in both clones A and C (Fig. 1B, compare lane 5 with lanes 4 and  6). In transfected FRTL-5 cells, the amount of Cx32 protein was similar in the three clones (Fig. 1B, lanes 9 -11). Fig. 2 illustrates the cellular distribution of Cx32 in transfected FRT and FRTL-5 cells. The immunofluorescence labeling of Cx32-transfected FRT cells was qualitatively similar for the three clones (Fig. 2, B-D); it appeared as discontinuous lines and/or dots delineating the region of cell-cell contacts. Bright dots were also visible over the cells. In Cx32-transfected FRTL-5 cells, the labeling profile was different. In the three clones (Fig. 2, F-H), the labeling appeared as rather large and round spots more randomly distributed over the cells. As FRTL-5 cells did not spread very well on the Petri dish, but rather remained tightly grouped, the location of the labeled spots was more difficult to establish; fluorescent spots were found in the regions of cell-cell contacts, but also in other regions of the plasma membrane or possibly inside the cells. In both cell types, Cx32-transfected cells were intensely labeled.
The level of Cx32 expression (assessed by mRNA and protein measurements) remained stable beyond 30 and 20 passages in transfected FRT and FRTL-5 cells, respectively. These results show that the Cx32-transfected cells we selected stably overexpressed the exogenous rat Cx32 gene, the major part of the protein being addressed to the plasma membrane.
Cell-to-Cell Communication in FRT and FRTL-5 Cells That Stably Express Cx32-Cx32-transfected cells exhibited a high level of GJ-mediated intercellular communication. When microinjected into one cell, lucifer yellow was detected within seconds in numerous cells in the vicinity of the microinjected one; this is illustrated for FRT cell-derived and FRTL-5 cellderived clones in Fig. 3 (A and B, respectively). The level of cell-to-cell communication or junctional coupling of each clone was quantified; the results are presented in the form of histograms of frequency (Fig. 4) that allow the collection and comparison of data from multiple microinjection tests on the different clones. Among the Cx32-transfected FRT cells, clones A and C, which presented the highest level of Cx32 expression, showed the highest dye-coupling capacity. Indeed, Ͼ50% of the cells from clones A and B and only 15% from clone C communicated with Ͼ20 neighbor cells. The three clones of Cx32transfected FRTL-5 cells exhibited similar junctional coupling levels. These data unequivocally establish that Cx32 protein synthesized by FRT and FRTL-5 cells from the exogenous Cx32 cDNA is competent to form functional gap junction channels. In this study, FRT and FRTL-5 cells that stably express Cx32 are designated Cx32-FRT and Cx32-FRTL-5, respectively.
Growth Rate of Cx32-FRT and Cx32-FRTL-5 Cells-The proliferation rate of Cx32-transfected cells was compared with that of control cells (wild-type cells and cells expressing the neomycin resistance gene: FRT-neo or FRTL-5-neo). The results of Fig. 5A show that the proliferation rate of the three clones of Cx32-FRT cells was significantly lower than that of control cells. Interestingly, cells from clone B (the clone among  the three that expressed the lowest level of Cx32 and intercellular communication) had an intermediate growth rate between that of control cells and that of cells from the two other clones (clones A and C). The growth rate of the three clones of Cx32-FRTL-5 cells was also markedly reduced as compared with that of control cells. After a 10-day culture period, the population of Cx32-FRTL-5 cells represented less than half of that of wild-type FRTL-5 cells. Table I gives the average population doubling time calculated for each clone in different experiments. One can observe that the mean doubling time was increased by ϳ35% in two of the three clones of Cx32-FRT cells and in the three clones of Cx32-FRTL-5 cells (as compared with that of the corresponding parental wild-type cells).
Growth of Cx32-FRTL-5 Cells Remains TSH-dependent-FRTL-5 cells are characterized by their TSH-dependent growth (15). As illustrated in Fig. 6A, the proliferation of FRTL-5 cells is arrested upon TSH withdrawal. When TSH was omitted from the culture medium, the growth of Cx32-FRTL-5 cells was also totally blocked, indicating that the proliferation of the communication-competent cells remained TSH-dependent. To determine whether the reduction in the rate of proliferation of Cx32-FRTL-5 cells was related to a change in their TSH responsiveness, we compared the growth rate of Cx32-transfected and wild-type FRTL-5 cells in response to increasing concentrations of TSH. The concentration-response curves shown in Fig. 6B indicate that the concentration of TSH required to obtain 50% of the maximum growth rate was very similar for transfected and nontransfected cells (ϳ50 microunits/ml).
Proliferation of FRTL-5 cells is known to be activated by insulin. In this study, insulin (10 g/ml) was systematically added to the FRTL-5 culture medium. The difference in growth between transfected and nontransfected cells was similar in the presence and absence of insulin (Fig. 6C). By varying the serum concentration from 0.5 to 5%, the difference in proliferation between wild-type and Cx32-FRTL-5 cells was maintained (Fig. 6D).
Expression of the Thyroglobulin Gene Is Increased in Cx32-FRTL-5 Cells-FRTL-5 cells expressed several thyroid-specific genes, including the genes encoding thyroglobulin, the thyroid prohormone, and two transcription factors (TTF-1 and Pax-8), as shown in Fig. 7 (lane 1 in A-C, respectively). TTF-1 and Pax-8 mRNA levels were similar in Cx32-FRTL-5, wild-type FRTL-5, and control FRTL-5-neo cells; but a huge increase in the amount of thyroglobulin mRNA was reproducibly observed in the three clones of Cx32-FRTL-5 cells (Fig. 7A, compare  lanes 3-5 with lanes 1 and 2). The increase of thyroglobulin gene expression in Cx32-transfected cells was confirmed at the protein level. The Western blot analysis results reported in Fig. 7E show that thyroglobulin was present in higher amounts in Cx32-FRTL-5 cells (lanes 3-5) than in control cells (lanes 1 and 2).
FRT cells expressed Pax-8, but neither TTF-1 nor thyroglobulin. The level of Pax-8 mRNA was similar in Cx32-FRT and control cells (data not shown). DISCUSSION This study clearly demonstrates that, as a consequence of transfection with and expression of the exogenous Cx32 gene that led to the restoration of cell-to-cell communication, the rate of proliferation of FRT and FRTL-5 cells was significantly reduced. In all the selected clones, the levels of Cx32 mRNA and Cx32 protein as well as the level of junctional coupling, on one hand, and the reduction of growth, on the other, remained stable beyond 20 or 30 passages (in Cx32-FRT and Cx32-FRTL-5 cells, respectively), indicating a stable insertion of the cDNA into the genome of host cells. Insertion of the pCMV-neo cDNA did not modify any of the properties of parental cells.
To our knowledge, FRT and FRTL-5 cells are the first cells in which the stable expression of the Cx32 cDNA leads to a reduction of growth in vitro. Other cell lines have already been stably transfected with the Cx32 gene, but in no case have modifications of cell proliferation in vitro been described (24 -26); retardation of growth in vivo in nude mice has been observed in SK hepatoma cells (24) and C6 glioma cells (25). Since FRT and FRTL-5 cells are not tumorigenic, the behavior of Cx32-transfected cells was not investigated in an in vivo context. The other interesting aspect of our results is that transfection with Cx32 cDNA resulted in the same reduction of proliferation in two distinct cell lines (FRT and FRTL-5) that derive from the same tissue. This suggests that the expression of Cx32 and the restoration of Cx32 GJ-mediated cell-to-cell communication generate responses that could be cell type-specific and dependent on special feature(s) of the recipient cells. The peculiarity of FRT and FRTL-5 cells could reside in the fact that both cell lines derive from normal rat thyroid cells that physiologically express Cx32. The functional impact of the reexpression of a Cx in communication-deficient cells could depend on the ability of the cells from which they derive to express this or other Cx. This hypothesis of cell type-specific effects of Cx is supported by previous findings of Mesnil et al. (27) in HeLa cells. Transfection of HeLa cells with the Cx26 gene (the Cx originally expressed in normal cervix tissue, from which HeLa cells derive) resulted in the reduction of in vitro cell proliferation. Such an effect was not observed in HeLa cells transfected with Cx43 or Cx40 cDNA (27). Similar observations have been made in C6 glioma cells, in which reduction of growth was observed after transfection with Cx43 cDNA, which is the Cx expressed in normal glial cells (28), but not after transfection with Cx32 cDNA (25). Restoration of cell-to-cell communication by transfection with Cx43 cDNA in cells deriving from Cx43-expressing cells was found to suppress growth in some (but not all) cell types (27)(28)(29)(30)(31)(32). Altogether, these data support the idea that GJ-mediated cell-to-cell communication controls cell growth in a cell type-and Cx-dependent manner.
The mechanism by which re-expression of Cx and restoration of cell-to-cell communication could control cell proliferation is not known at present. While one cannot totally exclude effects at levels unrelated to GJ function, such as changes in cell-cell adhesion processes, the most widespread hypothesis is that neoformed GJ act by providing channels for the cell-to-cell transfer of intracellular factors involved in the control of cell population growth (33). Up to now, the identity of these factors has not been established. As TSH, the essential growth factor for FRTL-5 cells, acts through the activation of the cyclic AMP cascade, it is tempting to postulate that the reduced rate of the rat Cx32 gene Wild-type FRT cells, FRT cells transfected with the neomycin gene only (clone neo) or cotransfected with the neomycin and Cx32 genes (clones Cx32-A, Cx32-B, and Cx32-C), wild-type FRTL-5 cells, and FRTL-5 cells transfected with the neomycin gene (clone neo) or cotransfected with the neomycin and Cx32 genes (clones Cx32-I, Cx32-P, and Cx32-Q) were cultured in duplicate wells for 2, 4, 6, 8, and 10 days. At each time point, cells were dissociated by mild trypsinization and counted. The population doubling time (PDT) was calculated according to the following formula: PDT t ϭ t⅐ln 2/ln(N t /N i ), where N i is the plating cell number and N t is the cell number at time t. The population doubling time of a given clone in a given experiment was the mean of the PDT t values measured at days 2, 4, 6, 8, and 10

FIG. 6. Effects of TSH, insulin, and serum on the growth of wild-type and Cx32-transfected FRTL-5 cells.
Wild-type FRTL-5 and Cx32-FRTL-5 cells (clone P) were cultured in 24-well culture plates. A, proliferation of cells in the presence or absence of TSH. Cells were seeded at 10 5 cells/well in complete culture medium or in the same medium devoid of TSH. Cells were counted at the indicated times. B, cell proliferation in response to increasing concentrations of TSH. Cells were seeded at 15 ϫ 10 4 cells/well in complete culture medium. Twentyfour h after plating, the medium was changed, and the cells were maintained for 8 days in the TSH-free medium. Then, increasing concentrations of TSH (0 -10 milliunits/ml) were added, and the cells were counted 36 h later. C, cell proliferation in the presence or absence of insulin. Cells were seeded at 5 ϫ 10 4 cells/well in complete culture medium. After 24 h, the medium was changed, and the cells were cultured for up to 10 days in complete culture medium or in the same medium devoid of insulin (Ins). Cells were counted at the indicated times. D, cell proliferation as a function of serum concentration. Cells were seeded at 5 ϫ 10 4 cells/well in complete culture medium. Twentyfour h after plating, the cells were washed and cultured for 7 days in the presence of increasing concentrations of serum (0.5, 2, or 5%). In each panel, the symbols represent the mean of triplicate measurements in a representative experiment. Similar results were obtained with clones I and Q.
proliferation of Cx32-FRTL-5 cells could be somewhat related to a decrease in the active intracellular cyclic AMP concentration due to the cell-to-cell transfer of the second messenger through the neoformed GJ. In that case, one would expect to normalize growth of communication-competent cells by adding exogenous cyclic AMP. Attempts to verify this hypothesis were unsuccessful; dibutyryl cyclic AMP failed to restore growth of Cx32-FRTL-5 cells. Taking into account that different Cx, expressed in a given cell type, generate distinct responses, one has to postulate that GJ channels composed of different Cx have selective permeability toward different regulatory molecules directly or indirectly involved in the control of cell proliferation. Recent data support this hypothesis; selective permeabilities toward chemical compounds have been found for channels composed of different Cx (34).
What are the possible targets of the regulatory molecules that could pass from cell to cell via GJ? There is some evidence that restoration of GJ-mediated cell-to-cell communication can affect the expression of proteins or factors involved in cell cycle control. Chen et al. (35) have reported changes in the expression of cell cycle regulatory proteins in TRMP cells transfected with Cx43 cDNA. The reduction of cell proliferation was accompanied by a decrease in cyclins A, D1, and D2 and cyclin-dependent kinases CDK5 and CDK6 and by a prolongation of both G 1 and S phases. In another study, Cx43-transfected C6 glioma cells were found to produce a factor capable of reducing the growth rate of nontransfected parental cells (36); this factor could be an insulin-like growth factor-binding protein, IG-FBP-4 (37), that inhibits the mitogenic action of insulin-like growth factors.
In our study, the reduction of Cx32-FRTL-5 cell growth does not seem to be secondary to the production of an inhibitory growth factor or an inhibitor of growth factor(s) since wild-type FRTL-5 cells cultured in conditioned medium of nontransfected or Cx32-transfected cells exhibited a similar growth rate (data not shown). The alteration of Cx32-FRTL-5 cell growth was probably not due to a change in the signaling pathway by which TSH activates cell multiplication because the TSH concentra-tion that induced the half-maximum stimulation of growth was similar in transfected and nontransfected cells. As the difference in growth rate between non-communicating wild-type FRTL-5 cells and highly communicating Cx32-transfected cells was maintained in the presence or absence of insulin and at low or high serum concentration, it is reasonable to think that growth retardation of thyroid cells overexpressing Cx32 could result from changes in the cell cycle machinery rather than alterations of the pathways by which growth factors trigger and control cell proliferation.
The other main finding of this work is that the level of expression of the thyroglobulin gene is increased in Cx32transfected cells as compared with wild-type FRTL-5 cells. This is particularly interesting since thyroglobulin is one of the major parameters of thyroid cell differentiation. We are currently analyzing whether the expression of other thyroid-specific genes such as that of the sodium iodide symporter is also up-regulated in highly communicating cells. Up-regulation of the expression of the thyroglobulin gene could result from changes in the level of expression or the activity of the thyroid transcription factors, TTF-1, TTF-2, and Pax-8 (38), that bind to the thyroglobulin promoter. As a first approach, we looked for possible alterations of TTF-1 and Pax-8 expression. The mRNA levels of the two transcription factors were similar in Cx32-transfected and wild-type cells. We will analyze the TTF-2 mRNA levels as soon as the probe is available. Recent reports from different laboratories indicate that TTF-1, TTF-2, and Pax-8 proteins are subjected to post-translational modifications that modulate their activity to bind to thyroid gene promoters and to regulate transcription (39 -41). Analyses of the DNA binding properties of the thyroid transcription factors from nuclear extracts of Cx32-transfected and wild-type FRTL-5 cells should help in understanding the mechanism by which restoration of cell-to-cell communication leads to a modulation of gene expression. Finally, we have reported an experimental system in which restoration of intercellular communication induces opposite changes in cell growth and expression of differentiation.