Carboxy terminus and pore-forming domain properties specific to Cx37 are necessary for Cx37-mediated suppression of insulinoma cell proliferation

Abstract

Connexin 37 (Cx37) suppresses cell proliferation when expressed in rat insulinoma (Rin) cells, an effect also manifest in vivo during vascular development and in response to tissue injury. Mutant forms of Cx37 with nonfunctional channels but normally localized, wild-type carboxy termini are not growth suppressive. Here we determined whether the carboxy-terminal (CT) domain is required for Cx37-mediated growth suppression and whether the Cx37 pore-forming domain can be replaced with the Cx43 pore-forming domain and still retain growth-suppressive properties. We show that despite forming functional gap junction channels and hemichannels, Cx37 with residues subsequent to 273 replaced with a V5-epitope tag (Cx37-273tr*V5) had no effect on the proliferation of Rin cells, did not facilitate G1-cell cycle arrest with serum deprivation, and did not prolong cell cycle time comparably to the wild-type protein. The chimera Cx43*CT37, comprising the pore-forming domain of Cx43 and CT of Cx37, also did not suppress proliferation, despite forming functional gap junctions with a permselective profile similar to wild-type Cx37. Differences in channel behavior of both Cx37-273tr*V5 and Cx43*CT37 relative to their wild-type counterparts and failure of the Cx37-CT to interact as the Cx43-CT does with the Cx43 cytoplasmic loop suggest that the Cx37-CT and pore-forming domains are both essential to growth suppression by Cx37.

Keywords: connexin; endothelium; gap junction channel; growth suppression; protein interactome.

Fig. 1.

Cx37–273tr*V5 and Cx43*CT37 are inducibly expressed by iRin cells. A: schematic of Cx37, Cx37–273tr*V5 and Cx43*CT37 sequences. Sequence shown in black in Cx37 (left) was replaced with sequence encoding V5 to make Cx37–273tr*V5 (middle: black residues from V5 sequence). Cx43*CT37 (right) included the first 236 amino acids of Cx43 (black) and the last 97 amino acids of Cx37 (gray). B: 2 clonal populations of iRin37tr cells, 1B3 and 2E5, express a V5 positive protein of the correct predicted mass at apparently low (1B3) or high levels (2E5). C: multiple clones of iRin43*CT37 express Cx43*CT37 at apparently different levels, with apparently high expression in the 3C2 clone. D: iRin37tr-2E5 clone does not express Cx37–273tr*V5 in the absence of doxycycline induction (Dox⁻ lane), but near maximal expression was induced by 2 μg/ml doxycycline at both 24 and 48 h. E: 3C2 clone of iRin43*CT37 cells was also maximally induced by 2 μg/ml doxycycline, but this clone had greater expression at 48 than 24 h. D and E: both proteins were detected in the triton insoluble fraction (TX lane). Blots were probed with V5 or Cx37 antibody; 30 or 50 μg of total cell protein (for iRin37tr and iRin43*CT37, respectively) were loaded in all lanes except the TX lane where all the triton insoluble protein from an equivalent aliquot of total cell protein was loaded; M_r indicates lane loaded with molecular mass markers; arrowheads indicate the protein of interest and “ns” marks a nonspecific band sometimes detected in total protein from Rin cells.

Fig. 2.

Cx37–273tr*V5 and Cx43*CT37 expression by Rin cells. Cx37–273tr*V5 (red) is not detected in iRin37tr cells in the absence of doxycycline induction (A) and is detected perinuclearly (B) as well as at points of cell-cell contact (C) following doxycycline induction. Blue marks the nuclei. D: Cx43*CT37 is expressed perinuclearly as well as in appositional membranes in iRin43*CT37 cells. Scale bar = 10 μm in all frames.

Fig. 3.

Cx37–273tr*V5 and Cx43*CT37 expression levels are comparable. A: Western blot of Positope and total protein isolated from the 1B3 and 2E5 clones of iRin37tr cells. Different amounts of Positope were loaded in each lane as indicated (in fmol), and 20 μg of total protein were loaded for each clonal cell line. Mass of M_r markers is indicated at left. Positope runs at 53 kDa whereas Cx37–273tr*V5 runs at ∼30 kDa. B: optical density of bands (from A) plotted as a function of the amount of loaded Positope for 25–200 fmol, the range encompassing the band densities for 20 μg of total protein isolated from the iRin37tr clones. Standard curve was fit by linear regression, R² = 0.994, intercept −2.08 and slope 1.799 AU/fmol. C: Western blot of Cx37-GST fusion protein and total protein isolated from the 3C2 clone of iRin43*CT37 cells. Different quantities of the fusion protein were loaded in each lane as indicated (in pmol); 50 μg of total protein were loaded for the 3C2 clone. Mass of M_r markers is indicated on the left. Full-length Cx37-GST fusion protein runs at ∼37 kDa (an internal cut-site within GST-portion of the protein resulted in the lower mass bands). D: optical density of Cx37-GST (all bands in a lane) and molar quantity of the loaded fusion protein were linearly related (R² = 0.99; intercept −4; slope 700 AU/pmol). The 3C2 clone expressed within the linear range of the standard curve.

Fig. 4.

Cx37–273tr*V5 does not suppress proliferation of iRin cells nor confer on these cells sensitivity to growth factors. A: with doxycycline induction, proliferation of iRin37 cells was slowed significantly (closed vs. open circles: *P < 0.01, 3 experiments each in triplicate). In contrast, proliferation of iRin37tr cells (data from 2E5 and 1B3 clones combined; each with 3 experiments done in triplicate) was unaffected by doxycycline induction (n.s.: closed vs. open squares). Proliferation rates (slopes of lines in inset) for the 2 iRin37tr clones induced or not with 2 μg/ml doxycycline (dox) were not different: 1B3, Dox⁻ 0.39 ± 0.05; 1B3, Dox⁺ 0.37 ± 0.042; 2E5, Dox⁻ 0.33 ± 0.15; 2E5, and Dox⁺ 0.34 ± 0.05. Points correspond to the actual data, lines are from the best fit of each data set. B: serum deprivation for 72 h, with doxycycline added for the terminal 24 h, resulted in significant (P < 0.001, n = 10 for serum deprived and nonserum deprived) accumulation iRin37 cells in G₁. In contrast, similar treatment of iRin37tr cells resulted in a significant (P < 0.001; n = 11 for serum deprived and n = 6 for nonserum deprived) decrease in G₁ cells. This decrease in G₁ cells was also observed in similarly treated iRin parental cells (points before x-axis break in C) suggesting this reduction is not specific to expression of Cx37–273tr*V5. C: restoration of 10% serum to iRin37 cells (at time 0 in C) released these cells from G₁ arrest (n = 10) and had no distinct effect on iRin37tr or iRin cells (n = 3 for each cell line). †Significant difference between serum exposed and deprived Cx37wt cells or Cx37tr cells; *Significant difference between cells expressing Cx37wt vs. Cx37tr.

Fig. 5.

Cx37–273tr*V5 forms functional gap junction channels and hemichannels. A: mean junctional conductance [G_j; measured by dual whole cell voltage clamp with transjunctional voltage difference (V_j) = 10 mV] of 30 iRin37 and 8 iRin37tr-2E5 cells was similar (P > 0.35, 3.4 ± 0.7 and 2.5 ± 0.8 nS, respectively). B: hemichannel-mediated dye uptake in iRin37 and iRin37tr-2E5 cells doxycyline induced or not for 24 or 48 h. Total cell no. (n), fields visualized (f), no. of experiments (N) for each data set were as follows: iRin37 (24 h, Dox: n = 1,069, f = 33, N = 11; 24 h, Dox⁺: n = 574, f = 26, N = 10; 48 h, Dox⁻: n = 521, f = 10, N = 3; 48 h, Dox⁺: n = 286, f = 9, N = 3); iRin37tr (48 h, Dox⁻: n = 409, f = 17, N = 6; and 48 h, Dox⁻: n = 295, f = 18, N = 6). *Significantly different from noninduced cells (P ≤ 0.05).

Fig. 6.

Cx37 displays complex channel behavior in Rin cells. A–D: segments of single channel records and corresponding all points histogram from 3 of the 11 studied cell pairs (B and C from same cell pair). Note the presence of multiple stable open states within and between cell pairs. V_j = 25 mV, scale bars correspond to 1 s and 2 pA. Arrow, when present, indicates beginning of pulse. E: compiled all points histogram from the 11 cell pairs (5-min total record time; sampling frequency 2 kHz). Cx37 channels spent the bulk of their time in the closed state (0 pS), with 40- and 250-pS subconductance states also evident; note that the fully open state was not visited sufficiently frequently or for a long enough duration to be evident in the all points histogram. F: relative frequency of 1,235 observed transition amplitudes in the 11 cell pairs.

Fig. 7.

Behavior of Cx37tr channels is less complex than wild-type Cx37 channels. A–C: segments of single channel record and corresponding all points histogram from 2 of 7 cell pairs (A and B from same cell pair). The fully open state is frequently observed in the truncation mutant, and fewer substates were evident than in wild-type Cx37. V_j = 25 mV; scale bars correspond to 1 s and 2 pA. Arrow indicates beginning of pulse. D: compiled all points histogram from the 7 cell pairs (20 min total record time; sampling frequency 2 kHz). Cx37tr channels spent the bulk of their time in a residual state (30 or 50 pS), but the fully open channel is also evident. E: relative frequency of observed transition amplitudes in 6 of 7 cell pairs. To be included in this histogram, transition had to be preceded and followed by current stability of 50 ms or longer. Note the preponderance of transitions in the 300- to 400-pS range, corresponding to transitions from closed to fully open or from the 30- to 50-pS substate to fully open. F: difference plot wherein relative frequency of transitions in each bin for the truncated channel were subtracted from the relative frequency of transitions in each corresponding bin for the wild-type channels. Transitions to and from the fully open state were far more common for the Cx37tr than Cx37 channel.

Fig. 8.

Cx43*CT37 expression has no effect on Rin cell proliferation, possibly due to absence of interaction between Cx37-CT and Cx43-CL. A: exponential growth rate (evident from slopes of the regression lines) for iRin cells was significantly (P < 0.0001) slowed by expression of Cx37 but not Cx43*CT37 or Cx37–273tr*V5 (open symbols, gray solid lines represent no doxycycline; filled symbols, dashed lines represent doxycycline-stimulated cells). Points correspond to the means of corresponding data, lines represent the best fit of each data set. [iRin37: N ≥ 12 (each experiment in triplicate) for each time point; iRin43*CT37: N = 3, each in triplicate; iRin37tr: N = 6 data from 2E5 and 1B3 clones combined]. B: circular dichroism shows Cx37-CT is disordered at both pH 7.4 and 5.8. C: ¹⁵N-HSQC spectra were collected for the ¹⁵N-Cx43-CL in the presence (red) and absence (black) of unlabeled Cx37-CT. The 2 spectra were entirely coincident, indicating no interaction between these proteins.