Neuronal Differentiation and Growth Control of Neuro-2a Cells After Retroviral Gene Delivery of Connexin43*

Given the roles proposed for gap junctional intercellular communication in neuronal differentiation and growth control, we examined the effects of connexin43 (Cx43) expression in a neuroblastoma cell line. A vesicular stomatitis virus G protein (VSVG)-pseudotyped retrovector was engineered to co-express the green fluorescent protein (GFP) and Cx43 in the communication-deficient neuro-2a (N2a) cell line. The 293 GPG packaging cell line was used to produce VSVG-pseudotyped retrovectors coding for GFP, Cx43, or chimeric Cx43·GFP fusion protein. The titer of viral supernatant, as measured by flow cytometry for GFP fluorescence, was approximately 2.0 × 107 colony form units (CFU)/ml and was free of replication-competent retroviruses. After a 7-day treatment with retinoic acid (20 μm), N2a transformants (N2a-Cx43 and N2a-Cx43·GFP) maintained the expression of Cx43 and Cx43·GFP. Expression of both constructs resulted in functional coupling, as evidenced by electrophysiological and dye-injection analysis. Suppression of cell growth correlated with expression of both Cx43 or Cx43·GFP and retinoic acid treatment. Based on morphology and immunocytochemistry for neurofilament, no difference was observed in the differentiation of N2a cells compared with cells expressing Cx43 constructs. In conclusion, constitutive expression of Cx43 in N2a cells does not alter retinoic acid-induced neuronal differentiation but does enhance growth inhibition.

Gap junctions are intercellular plasma membrane channels which provide direct cytoplasmic continuity between adjacent cells as well as coordination of the function of individual cells (1). These channels mediate direct exchange of ions and small molecules (less than 1.2 kDa) including second messengers and electrical current among adjacent cells (2,3). The fundamental structural unit of the gap junction is the connexin (Cx) 1 sub-unit. To clarify the role of gap junctional intercellular communication (GJIC) in neural development, the temporal and cellular expression of connexins has been studied. Cx43 is present in neurons in vivo (4) and in vitro (5)(6)(7). Several studies have demonstrated that neuronal GJIC decreases during differentiation (8). In addition, we have shown that Cx43-mediated GJIC decreases with neuronal differentiation (7,9,10).
Given the multiple roles proposed for Cx43-mediated GJIC in development and differentiation, many studies have been undertaken to alter Cx43 expression both in vitro through transfection (11)(12)(13) and in vivo through transgenesis (14,15). The major limitations to transfection approaches have been efficiency and time required to obtain stable gene expression. The recent advent of retroviral vectors to efficiently deliver genes for stable expression overcomes these limitations (16,17). Retroviral vectors enable efficient gene transfer and stable gene expression in cells that are not readily susceptible to transfection, such as primary cells, cells in vivo, and neuronal cell lines (18). However, the titer of most current retroviral supernatants is too low (Ͻ10 7 particles/ml) to achieve acceptable clinical results (19). To overcome this deficit, more recently developed retrovectors pseudotyped with the vesicular stomatitis virus G (VSVG) protein have become the gene delivery tool of choice (16). VSVG-pseudotyped retrovectors have advantages over the current viral vectors in terms of high titer, complement resistance, particle stability and tumor specificity (19).
In this study, the VSVG-pseudotyped retrovector enabled efficient stable expression of Cx43 and Cx43⅐GFP. Cell growth was suppressed but neuronal differentiation was unaffected.
Retrovectoral Construction and Synthesis-The AP2 retroviral vector (19) was used in these studies. Two genes including an inserted cDNA and the enhanced green fluorescent protein reporter gene can be expressed from this single bicistronic, nonsplicing murine plasmid ret-* This work was supported by the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed:  1A) (19). The cDNA fragment coding for IRES and EGFP was removed from the plasmid backbone by single endonuclease digestion with NotI, and the resulting fragment was religated together as outlined below. The resulting plasmid was designated NAP2 (Fig. 1B).
Production of VSVG-pseudotyped Retrovirus-To produce pseudotype retroviral vectors, 293 GPG cells were plated at 2 ϫ 10 6 cells per 60-mm dish the night before transfection in 293 media. They were then transiently transfected with AP2, NAP2Cx43, and NAP2Cx43⅐GFP plasmid retrovectors by the LipofectAMINE PLUS TM (Life Technologies, Inc.) procedure according to the manufacturer's instructions. Serum-free DMEM medium with tetracycline (1 g/ml) was used for all of the transfection processes. After a 6-h incubation of cells in the incubator, an additional 2 ml of the 293 medium was added for transfection. The following morning, the medium was replaced with fresh 293 medium. On the next day, the 293 medium was removed from cells and replaced with 3 ml of DMEM with 10% fetal calf serum. On the next day, the supernatants were collected about every 12-24 h for 1 week. All culture supernatants were filtered through 0.45-m syringemounted filters (Gelman Sciences, Ann Arbor, MI). The filtered culture supernatants were used as infectious viral stock for subsequent experiments. Aliquots of 1.0 ml were also frozen at Ϫ80°C for later use.
Titration of Retroviral Supernatant-Flow cytometric analysis was performed to determine the titer of the viral supernatant as measured by GFP fluorescence. In brief, 2 ϫ 10 5 C6 cells/well were plated in 6-well tissue culture plates the night before infecting the cells with the retrovirus. The next day, cells from 3 wells were trypsinized and counted to determine the average number of cells per well at the time of exposure to retrovirus. Serial dilutions (1:10) of the viral supernatant in a final volume of 1 ml of DMEM, 10% fetal bovine serum were prepared and added to each well. The following day, 2.0 ml of medium were added, and cells were incubated for 2 more days. Three days after viral infection, the transduced C6 cells were trypsinized and resuspended in 2 ml of DMEM for flow cytometric analysis assay. Analysis was performed on a FACStar PlusTM (Becton Dickinson). Live C6 cells were gated based on scatter/side scatter profile and analyzed for GFP fluorescence to determine the percentage of GFP-positive cells. Data acquisition and analysis were performed using CELLQuest TM software (Becton Dickinson). The titer (colony-forming units (CFU) per ml) was calculated as: (% GFP-positive cells ϫ cell number at initial viral exposure)/viral volume (ml) applied when transduction was not saturated (19).
Replication Competent Retrovirus (RCR) Assay-C6 cells infected with GFP were passed in culture for 6 weeks (15 passages) to allow for spread of RCR that might be present. 3 ml of culture supernatant from these C6 cells was then used to infect 10 6 naïve C6 cells, and 48 h later, the newly infected naïve cells were evaluated by detecting GFP-fluorescent cells.
Protein Isolation and Western Blot Analysis-Both undifferentiated and differentiated N2a wild-type and transformant cell cultures were washed twice with PBS and scraped off the plates in lysis buffer (0.05 M Tris, pH 6.8, 0.1% SDS) with a rubber policeman. The protein concentration of the cell lysate was determined in each case using the Bio-Rad protein assay kit. Protein samples from an equal number of cells (2.5 ϫ 10 5 each) were subjected to 10% SDS-polyacrylamide gel electrophoresis. The gel was transferred to nitrocellulose membranes FIG. 1. Schematic representations of plasmid retrovectors. A, AP2 plasmid retrovector is a bicistronic murine retroviral vector allowing the insertion of a cDNA sequence in multiple cloning sites upstream of the IRES and the EGFP cassette. B, NAP2 plasmid retrovector resulted from cutting the IRES and EGFP fragment of the AP2 plasmid retrovector. C, the NAP2-Cx43 plasmid retrovector resulted from insertion of Cx43 cDNA into the NotI-ClaI cloning sites of NAP2. D, the NAP2-Cx43⅐GFP retrovector resulted from insertion of Cx43⅐GFP cDNA into the BglII-NotI sites.
(Bio-Rad) at 5 watts overnight. Membranes were stained with Ponseau S to control for protein loading and transfer. The membranes were blocked in 5% (w/v) nonfat dry milk (NFDM) in PBS for 1 h at room temperature, incubated with monoclonal anti-Cx43 primary antibody overnight at 4°C in PBS with 1% NFDM, washed with PBS, and incubated 1 h with horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse secondary antibody in PBS with 1% NFDM at room temperature. Finally, membranes were incubated with Supersignal substrate working solution (Pierce) for 5 min and exposed to autoradiographic film for 2-10 min.
Cells were grown to confluence on glass, fixed in 4% paraformaldehyde (w/v) for 10 min at 4°C, washed with PBS, pH 7.4, permeabilized with 0.1% Triton X-100 for 10 min at room temperature, blocked in PBS with 10% (v/v) normal goat serum (Dimension Laboratories) and 1.0% (w/v) bovine serum albumin for 30 min, and then incubated for 1 h with primary antibody(s) in PBS with 1% (w/v) bovine serum albumin. The coverslips were washed with PBS and then incubated for 1 h with the secondary antibody(s) in PBS with 1% (w/v) bovine serum albumin. Coverslips were again washed and were mounted on slides in PBS containing 50% (v/v) glycerol and 1% (w/v) p-phenylenediamine (Fisher Scientific) to minimize quenching of the fluorescent signal. Fluorescence was visualized on a Zeiss Axiophot photomicroscope (Zeiss, North York, ON). Samples incubated without primary antibody served as negative controls.
Current Injection-Monolayers of undifferentiated and differentiated N2a, N2a-Cx43, and N2a-Cx43⅐GFP cells grown on coverslips were placed in a bath chamber containing culture medium at room temperature (ϳ22°C) and viewed with a Leica DMIRB inverted microscope with a ϫ 20 phase-contrast objective. Glass microelectrodes were prepared from microcapillary tubing (World Precision Instruments (WPI), Kwik-Fill, no. 1B100F-3) with a Narishige vertical microelectrode puller. After backfilling with 1 M KCl, the electrodes had tip resistances of 40 -80 megohm. They were mounted on mechanical micromanipulators (Leica) and connected by chlorinated silver wires to electrometers (WPI, Intra 767). The bath was grounded using a copper wire connected to the bath chamber. Two independent electrodes were used, one to pass current into a cell and the other to record changes in membrane potentials at various distances, i.e. 25, 100, and 400 m respectively. Electrotonic potentials were generated by passing hyperpolarizing pulses of current (50 nA; 100 ms) down one electrode from a stimulus-isolating instrument (WPI, A310 Accupulser). Membrane and electrotonic potentials were recorded on a Gould RS 3200 two channel recorder.
In cell coupling experiments, each measurement of the electrotonic voltage at a given interelectrode distance required that both the electrodes obtain and maintain the full cell membrane potential. Upon insertion of a microelectrode into a cell, a sharp drop in the electrometer reading (identified as the resting membrane potential) was noted. Current pulsing was initiated shortly after gaining a stable membrane potential at the injecting electrode; after 4 -5 pulses, the electrode was removed. Upon removal of the electrode from the cell, the measured potential returned to zero: recordings that did not return to within 10% of the baseline were assumed to be ruptured cells, and these recordings were discarded. Three interelectrode distances were used: 25 m (the distance to an adjacent cell), 100 m, and 400 m.
Dye Injection-Clusters of N2a wild-type and N2a transformants that had been infected with retrovector encoding Cx43 or Cx43⅐GFP were pressure injected with 10 mM carboxyfluorescein (Molecular Probes, Eugene, OR) in distilled water using an Eppendorf microinjection system (Eppendorf, Hamburg, Germany). In all cases, it took up less than one minute for carboxyfluorescein to pass to adjacent cells after 0.5-s pressure injection. To be regarded as coupled, at least one adjacent cell received the dye from the injected donor cell.
In Vitro Growth Analysis-N2a wild-type and transformants from confluent cultures were seeded at 2.0 ϫ 10 5 cells per 6-well culture plate. The growth analysis was carried out in the absence and present of retinoic acid (RA) (20 M). The cell viability was determined by trypan blue staining. At day 0, 3, 5, and 12 after plating, triplicate plates were dissociated with 0.25% trypsin (Life Technologies, Inc.) and 1 mM EDTA (BDH) in PBS, pH 7.4 and counted in a COULTER COUN-TER® Z1 Series Particle Counter (Coulter Electronics, Burlington, ON).
Statistical Analysis-Statistical analysis involving testing for significance of differences for quantitative observations was performed. First, all sets of data were subjected to a one-way analysis of variance to determine significant difference between any two sets. Subsequently, the sets of data were subjected to Student's t test to determine whether significant differences existed between the individual samples. The data was analyzed with Prism (Graphpad Inc. Software, San Diego, CA).
Fluorescence Microscopy and Image Analysis-Cells were evaluated with a Zeiss Axioskop microscope equipped with a filter set for fluorescein (exciter filter BP 485, banner filter 515-565 nm) and for Cy3 (exciter filter BP 546, banner filter BP 590) using ϫ 40 or ϫ 63 Plan-Neufluor objectives. The images were captured using Sensicontrol 4.02 (The Optikon Corporation Ltd., Kitchener, ON). Once all of the digital images were captured, they were imported and compiled into final figures using Adobe Photoshop 5.0 (Adobe Systems Incorporated, San Jose, CA) and Corel Draw 8.0 (Corel Co., Ottawa, ON).

Effectiveness of Retroviral Infection of C6 Glioma Cells-C6
cells were chosen for titering retrovector supernatant as C6 cells divide rapidly and are easily infected. Using GFP as a reporter protein allowed the rapid titration of retroviral supernatant. The supernatant containing AP2 retrovector was sequentially collected, filtered, and serially diluted in a final volume of 1 ml and added to 3.28 ϫ 10 5 C6 cells/well in a 6-well plate. Three days after a single application of retrovector, the infected C6 cells were analyzed by flow cytometric analysis to determine the percentage of cells expressing the GFP reporter protein (Fig. 2). The titer extrapolated from these experiments was about 2.0 ϫ 10 7 CFU/ml, calculated as described under "Experimental Procedures." This titer is at least 1,000 times higher than that of NIH 3T3-based retroviral packaging cell lines (22).
Stability of Retroviral Gene Delivery and Replication Competent Retrovirus Assay-C6 cells infected with viral supernatant containing AP2 retrovectors demonstrated strong GFP fluorescence for at least 15 passages over a 6-week period (Fig.  3). Naturally or because of recombination events, viruses that contain all of the cis-acting viral elements and genes coding for necessary viral structural proteins are "replication-competent" (18). To test for RCR, 3 ml of culture supernatant from these C6 cells was then used to infect 10 6 naïve C6 cells. 48 h later the newly infected naïve cells were evaluated by detecting GFPfluorescent cells. Retroviruses produced from 293 GPG packaging cells were free of detectable RCR upon long term cultivation as demonstrated by the failure of GFP fluorescence to be transferred from virally transduced C6 cells to naïve C6 cells (data not shown).
Expression of Cx43 and Cx43⅐GFP in Differentiated N2a Cells-Communication-deficient N2a cells were infected with retrovector encoding for Cx43 and Cx43⅐GFP to determine their ability to express full-length Cx43 and Cx43⅐GFP chimeric protein after differentiation. Wild-type N2a and N2a-GFP cells were used as controls. The synthesis of Cx43 and Cx43⅐GFP protein was investigated by Western blot analysis of total cell lysates from wild-type N2a cells, N2a-GFP, N2a-Cx43, and N2a-Cx43⅐GFP transformants under differentiated conditions. Differentiated N2a-Cx43 and N2a-Cx43⅐GFP cells expressed Cx43 protein (42-44 kDa) and Cx43⅐GFP protein (72-74 kDa) (Fig. 4). As expected Cx43 was not detected in wild-type N2a cells or N2a-GFP expressing cells. The C6 -13 cells which were transfected with Cx43 were used as a positive control, and contain multiple bands of immunoreactivity, which represents phosphorylated and unphosphorylated species of Cx43 (23). The low amount of phosphorylated species in N2a cells infected with wild-type Cx43 or Cx43⅐GFP cDNAs may reflect poor post-translational processing of Cx43 in this cell line.

Localization of Cx43 and Cx43⅐GFP in Undifferentiated and
Differentiated N2a Cells-To determine the localization of Cx43 in N2a-Cx43, N2a-Cx43⅐GFP, N2a-GFP, or wild-type N2a cells, differentiated or undifferentiated cells were either immunolabeled with anti-Cx43 antibodies or directly examined for GFP fluorescence (Fig. 5).
In both undifferentiated and differentiated N2a-Cx43 and N2a-Cx43⅐GFP cells, Cx43 was localized to areas of cell-cell contact as well as in punctate structures within the cytoplasm (Fig. 5, A, B, E, and F). As control, GFP fluorescence was located throughout the cytoplasm in N2a-GFP cells (Fig. 5, C and G) and no fluorescence was detected in wild-type N2a cells immunolabeled for Cx43 (Fig. 5, D and H). In all cases, the undifferentiated N2a and N2a transformants were not neurofilament positive (data not shown).
An induction of neurofilament protein (200 kDa, phosphorylated and non-phosphorylated) was apparent in processes of wild-type N2a cells and all three N2a transformants (N2a-GFP, N2a-Cx43 and N2a-Cx43⅐GFP) after 20 M RA treatment for 7 days (Fig. 5, I, J, K, and L). On average, 90% of the cells were neurofilament positive after exposure to 20 M RA for 7 days. There was no apparent difference in neurofilament staining between wild-type N2a, N2a-Cx43 and N2a-Cx43⅐GFP cells.
Gap Junctional Intercellular Electrical Coupling Up-regulation in Cx43-and Cx43⅐GFP-expressing N2a Cells-To determine whether Cx43 and Cx43⅐GFP formed functional gap junc-tion channels in N2a cells before and after differentiation, communication-deficient N2a cells and N2a-Cx43 and N2a-Cx43⅐GFP transformants were examined for electrical coupling. For undifferentiated N2a-Cx43 cells, electrotonic spread of injected current, resulting in a membrane potential change at the second electrode, could be detected 400 m from the injecting electrode (n ϭ 8). For undifferentiated N2a-Cx43⅐GFP cells, potential deflections were detected at 25 m (n ϭ 8) but only twice in six attempts at 100 m and never at 400 m (n ϭ 8). Conversely, undifferentiated N2a cells displayed no electrical coupling at any distance tested (n ϭ 9). In differentiated N2a-Cx43 cells, potential defections were detected at 400 m (n ϭ 4). Also, potential deflections were detectable in differentiated N2a-Cx43⅐GFP cells at 25 m (n ϭ 3) but not at 100 m (n ϭ 2) or 400 m (n ϭ 4). Thus, both N2a-Cx43 and N2a-Cx43⅐GFP cells formed gap junctional channels that were electrically coupled. In addition, the N2a-Cx43⅐GFP cells appeared less adherent (i.e. they were easily pulled off the coverslip by the microeletrodes, which may account for the differences in coupling between Cx43 and Cx43⅐GFP cells. Gap Junctional Intercellular Dye Coupling Is Up-regulated When Cx43 or Cx43⅐GFP Is Expressed in N2a Cells-To examine the dye-coupling capacity of Cx43 or Cx43⅐GFP-mediated GJIC, both undifferentiated and differentiated N2a wild-type and transformants were microinjected with 10 mM carboxyfluorescein (CF). Undifferentiated and differentiated N2a-Cx43⅐GFP cells (Fig. 6, A and B, asterisk), N2a-Cx43 cells (Fig.   FIG. 2. Titration of retroviral  6, E and F, asterisk) and N2a wild type (Fig. 6, I and J, asterisk) were injected with CF. CF spreads rapidly to adjacent cells in both differentiated and undifferentiated N2a-Cx43⅐GFP and N2a-Cx43 cells within a minute (Fig. 6, A, B, E, and F). The gap junctional plaques formed by Cx43⅐GFP were seen as punctate fluorescent spots between adjacent cells (Fig. 6, A and B, arrows). N2a-Cx43 and N2a-Cx43⅐GFP cells effectively transferred CF to neighboring cells in more than 70% of cases under undifferentiated conditions and in greater than 50% after neuronal differentiation (Table I).
Furthermore, CF was extensively transferred to at least the 6th order of neighboring undifferentiated Cx43 or Cx43⅐GFP expressing cells (Fig. 6, A and E), although this was reduced to the 2nd to 3rd order when the same cells were differentiated with retinoic acid (Fig. 6, B and F). As a control, communication-deficient wild-type N2a cells exhibited no dye coupling before or after differentiation (Fig. 6, I and J). N2a Cells Expressing Cx43 or Cx43⅐GFP Display in Low but Significant Reduction in Cell Growth After RA-induced Differentiation-Under the normal culture condition, Cx43 and Cx43⅐GFP exhibited no growth inhibition in N2a cells (data not shown). In the presence of 20 M of RA, however, the growth rate of N2a-Cx43 and N2a-Cx43⅐GFP cells was significantly decreased (Fig. 7) at every time point when compared with wild-type N2a or N2a-GFP cells. There were no significant differences between N2a-Cx43 and N2a-Cx43⅐GFP transformants or between wild-type N2a and N2a-GFP cells (Fig. 7). DISCUSSION By virtue of its intrinsic fluorescence, GFP is readily detected in living cultured cells, and the efficiency of transduction can be rapidly and directly determined. GFP fluorescence provides a visual assessment for rapidly determining the viral titer by flow cytometric analysis, which is consistent with a direct test of viral-mediated transfer of drug resistance to host cells (24). In this study, the titer of the AP2 retrovector determined by flow cytometric analysis of GFP fluorescence was about 2.0 ϫ 10 7 CFU/ml which is similar to other studies (19). The efficient gene transduction of high titer viral supernatant eliminates the need to generate stable transformed cell lines, which requires months of selection and characterization. Therefore, the AP2 retroviral gene delivery system is well suited for quickly examining expression of exogenous gene(s) in vitro.
Many connexins are differentially expressed during neuronal differentiation. During development, neuroblasts are highly coupled, with a loss or reduction of GJIC during termi- nal differentiation (8,25). A similar decrease in the connexin expression level and GJIC has been reported during in vitro differentiation of two neuronal stem cells, i.e. the P19 mouse embryonal carcinoma cell line (9) and the NT2 human teratocarcinoma cell line (6,7).
In our study, N2a neuroblastoma cells provide an ideal in vitro model to examine the putative role of Cx43-mediated GJIC in neuronal differentiation and growth control because they are of neuronal origin, do not express any known connexins (26), and exhibit no detectable gap junctional intercellular coupling. Retroviral gene transduction, Western blot analysis, and immunocytochemistry revealed that N2a cells were able to maintain the expression of Cx43 and Cx43⅐GFP protein after neuronal differentiation. Cx43 and Cx43⅐GFP were localized as punctate staining in the cytoplasm and in the membrane between apposed cells. About 90% of the N2a cells overexpressing Cx43 and Cx43⅐GFP could be differentiated into mature neurons, similar to wild-type cells. These results suggest that Cx43-mediated GJIC does not alter neuronal differentiation of N2a cells. It is possible that in their undifferentiated state, N2a cells have already gone through a possible GJIC-dependent differentiation stage because undifferentiated N2a cells express a certain level of the neuronal marker microtubule-asso-ciated protein 2 (MAP2) (27).
Gap juctions are known to provide a channel for the intercellular flow of essential electrical signals and small molecules (28). The current flow through gap junctions is important in the rapid and synchronous activation of excitable tissues (29). In our studies, all N2a-Cx43 and N2a-Cx43⅐GFP cells were electrically coupled both before and after neuronal differentiation. However, the cells expressing Cx43⅐GFP exhibited lower levels of electrical coupling than cells expressing Cx43. Because the carboxyl-terminal region of Cx43 is presumably involved in the regulation of channel gating (1), fusion of GFP to this portion of the connexin may interfere with channel function. Using dual whole-cell patch clamp recording, Bukauskas et al. (30) have reported some alterations in transjunctional voltage gating for N2A cells transfected with Cx43 or Cx43⅐GFP but offered no explanation for this difference. We have also noted that N2A cells transfected with Cx43 or Cx43⅐GFP are less dye coupled after differentiation. This result is consistent with the in vivo studies, which have indicated that electrotonic junctions are wide spread between mammalian neurons in many areas, including neocortex, hippocampus, inferior olive, locus coeruleus, hypothalamus, striatum, and retina (3, 31, 32).
From our dye-coupling analysis, undifferentiated N2a-Cx43 FIG. 5. Cx43 and Cx43⅐GFP expression in N2a wild-type and transformants before and after neuronal differentiation. In undifferentiated transformants, the Cx43 and Cx43⅐GFP proteins appeared mostly as punctate staining (A and B, arrows) on the cell surface against a diffuse fluorescent background. In differentiated transformants, Cx43 and Cx43⅐GFP protein are located in the processes, cell body, and cell surface (E and F, arrowheads). No Cx43 or Cx43⅐GFP was detected in N2a wild-type and N2a-GFP cells (C, D, G, and H). As control, GFP fluorescence is located throughout the cytoplasm in N2a-GFP cells (C and G). A and E were immunolabeled with anti-Cx43. B, C, F, and G show only GFP fluorescence. A monoclonal neurofilament 200 antibody was used in panels I-L. In all cases, the undifferentiated cells were neurofilament negative (data not shown). After neuronal differentiation, about 90% of all the RA-treated cells were neurofilament positive (I, J, K, and L). Magnification is ϫ 1,000. and N2a-Cx43⅐GFP cells were found to be coupled on average to 6th order neighbors. After differentiation, however, the coupling decreased on average to 2nd or 3rd order. One possibility is that the differentiated cells bear long processes, and their spatial arrangement differs from undifferentiated cells. Therefore, the dye must pass a similar distance but involves fewer cells. It is also possible that the functional dye-coupling capacity of Cx43 and Cx43⅐GFP does decrease with neuronal differentiation. Our studies demonstrate for the first time highly coupled differentiated neuronal cells, which provide a suitable model to further investigate the electrotonic junction between mature neurons both in vitro and in vivo.
One of the major characteristics of neoplastic cells is their uncontrolled rapid growth. Extensive aspects of tumor growth have been studied, including the implication that loss or lack of gap junctional communication is associated with tumor development (33)(34)(35)(36)(37). The introduction and overexpression of connexin cDNAs in tumor cell lines by transfection has demonstrated a reduced growth rate (38 -40). C6 cells transfected with Cx43 cDNA exhibited a decrease in cell growth both in vitro and in vivo (11)(12)(13)41). In contrast to C6-Cx43 cells which grow slower than the parental C6 cells, the N2a-Cx43 and N2a-Cx43⅐GFP transformants demonstrated no significant decrease in growth rate in vitro when undifferentiated. It appears that the establishment of GJIC in communication-deficient FIG. 6. Dye-coupling assay of N2a wild-type and transformants both before and after differentiation. The undifferentiated N2a-Cx43 and N2a-Cx43⅐GFP cells were coupled on average to 6th order neighbors (A and E). After differentiation, however, the coupling decreased on average to 2nd to 3rd order (B and F). CF spreads rapidly from injected cells (A, B, E, and F; asterisk) to adjacent cells when compare with phase contrast images (C, D, G, and H). The direct visualization of Cx43⅐GFP during microinjection is shown in A and B (arrows). N2a wild type and N2a-GFP exhibited no dye coupling to adjacent cells either before or after differentiation (I and J) when compared with phase contrast images (K and L). Magnification is ϫ 850.  N2a wild-type and derived transformants from confluent cultures were seeded at 2.0 ϫ 10 5 per 6-well culture plate. 3, 5, and 12 days after plating in the presence of RA, triplicate plates were dissociated into individual cell suspensions and counted. The statistical significance (p Ͻ 0.05, 0.01, and 0.001) are represented by 1, 2, or 3 asterisks, respectively. n ϭ 6; bars, mean Ϯ S.E. tumor cells does not always lead to growth inhibition. It has been demonstrated that certain connexin genes exert a growth control effect whereas others do not (38). In addition, gap junctions composed of different connexins demonstrate some molecular transport selectivity (42). In the presence of retinoic acid (20 M), however, the growth of N2a-Cx43 and N2a-Cx43⅐GFP was significantly inhibited. This finding suggests that growth inhibition in N2a transformants may be because of the RA treatment and related transjunctional molecule(s) rather than Cx43 itself.
This study constitutes the first report of an in vitro generation of highly coupled mature neurons obtained using retroviral delivery of connexin constructs. Recent therapeutic applications have employed retroviral delivery of the Herpes simplex virus thymidine kinase gene to brain tumors, with limited success (19). Given the potential tumor suppressor effects of connexins, as well as the role of GJIC in mediating some aspects of the "bystander effect" associated with Herpes simplex virus thymidine kinase expression and combined pro-drug (i.e. ganciclovir) treatment, we propose that retroviral delivery of connexin genes may be therapeutically relevant.