Functional Characterization of Oculodentodigital Dysplasia-Associated Cx43 Mutants

Abstract

Oculodentodigital dysplasia (ODDD) is associated with at least 28 connexin43 (Cx43) mutations. We characterized four of these mutants; Q49K, L90V, R202H, and V216L. Populations of these GFP-tagged mutants were transported to the cell surface in Cx43-negative HeLa cells and Cx43-positive NRK cells. Dual patch-clamp functional analysis in N2A cells demonstrated that channels formed by each mutant have dramatically reduced conductance. Dye-coupling analysis revealed that each mutant exhibits a dominant-negative effect on wild-type Cx43. Since ODDD patients display skeletal abnormalities, we examined the effect of three other Cx43 mutants previously shown to exert dominant-negative effects on wild-type Cx43 (G21R, G138R, and G60S) in neonatal calvarial osteoblasts. Differentiation was unaltered by expression of these mutants as alkaline phosphatase activity and extent of culture mineralization were unchanged. This suggests that loss-of-function Cx43 mutants are insufficient to deter committed osteoblasts from their normal function in vitro. Thus, we hypothesize that the bone phenotype of ODDD patients may result from disrupted gap junctional intercellular communication earlier in development or during bone remodeling.

Keywords:

• Connexin43
• disease-linked mutants
• ODDD
• osteoblast
• cell coupling
• conductance

INTRODUCTION

Gap junction channels are formed by transmembrane protein subunits encoded by a family of connexin genes and allow direct gap junctional intercellular communication (GJIC) in almost all cells of the human body (1). Each connexin has a unique expression pattern but Cx43 is the most ubiquitously expressed member of the connexin family and, therefore, its expression profile and function directly affects the development, function, and physiological properties of several tissues. Cx43 is assembled into connexons in the trans-Golgi network (2, 3) and transported to the cell surface via a microtubule-dependent pathway (4) where adjacent hemichannels on apposed cells dock and cluster in distinct areas to form gap junction plaques (4). Cx43 has a short half-life of one to three hours (5) that includes its eventual fate of being internalized and targeted for degradation in proteasomes or lysosomes (6, 7, 8).

Missense mutations in the polypeptide backbone of any one of a number of the 21 member connexin family cause abnormal connexin trafficking (9, 10, 11) and/or loss of intercellular channel function (9, 12, 13) resulting in several human diseases. Mutations in Cx43 have been linked to a human disorder known as oculodentodigital dysplasia (ODDD) (14, 15). At least 28 autosomal-dominant missense, base pair duplication or premature stop codon mutations have been found in the Cx43 gene resulting in disease-associated Cx43 protein mutations in multiple Cx43 domains (15, 16, 17, 18, 19, 20). ODDD patients display a variety of phenotypes including syndactyly and camptodactyly (15), cardiac abnormalities, some cases of mild retardation (21) and skeletal abnormalities including congenital craniofacial deformities around the eye orbitals and teeth (14) and limb deformities. Only several hundred cases of ODDD have been reported worldwide (21), but the vast pleiotropy of symptoms may account for its possible under diagnosis. In addition, a mouse model of ODDD has been identified with a single mutation in the Cx43 gene, G60S, that corresponds to a position in the first extracellular loop of the protein (22). This mouse displays phenotypes comparable to those of human patients including syndactyly and skeletal abnormalities and, therefore, is an excellent tool for studying the consequence of a Cx43 mutation on tissue development and differentiation.

Several ODDD-linked Cx43 mutations have been characterized as to their effect on Cx43 transport, assembly and ability to assemble functional channels (23, 24, 25). We have previously demonstrated that both the G21R mutation, found at the start of the first transmembrane domain, and the G138R mutation, found in the cytoplasmic loop, are loss-of-function mutations that exhibit dominant-negative properties on wild-type Cx43 (23). Similarly, mutations in the N-terminus (Y17S), first transmembrane region (A40V), second transmembrane domain (L90V) and cytoplasmic loop (I130T, K134E) have been shown to yield loss-of-function mutants that form dysfunctional channels (25). Mutations in the extracellular loops (F52dup, R202H) have been reported to incapacitate the formation of gap junction plaques (25). To assess further the functional status and consequences of cells harboring Cx43 mutants, our laboratory has extended these studies by examining the ODDD-linked Cx43 mutants Q49K, located in the first extracellular loop, L90V, located in the second transmembrane domain, R202H, found in the second extracellular loop, and V216L, found in the fourth transmembrane domain. It is important to note that ODDD patients have one mutant and one wild-type allele coding for Cx43. If equal expression of both proteins is assumed, we speculate that these mutants may have an effect on coexpressed wild-type connexins. To that end, we investigated whether these ODDD-linked Cx43 mutations are loss- or gain-of-function mutations and whether they exert a dominant effect on coexpressed wild-type Cx43. Importantly, since ODDD patients often have craniofacial bone defects, we investigated the role of Cx43 mutants on differentiation of osteoblasts.

MATERIALS AND METHODS

Cell Lines and Reagents

All cell lines were purchased from the American Type Culture Collection (Manassas, VA) and all culture media and reagents were obtained from Invitrogen (Burlington, ON, Canada). Normal rat kidney (NRK), human cervical carcinoma (HeLa) and mouse neuroblastoma (N2A) cells were cultured in regular high glucose DMEM supplemented with 10% fetal bovine serum, 2 mM L-glutamine (except N2A cells), 100 units/ml penicillin and 100 μ g/ml streptomycin. Cells were maintained at 37°C in a moist environment of 95% air and 5% CO₂ and subcultured as required using 0.25% trypsin-EDTA solution.

DNA Constructs and Transfection

Cx43-mutants were made by PCR using the QuikChange® Site Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Human Cx43 in pBluescript vector (a gift from Dr. GI Fishman, NYU Medical Center) was used as a template to make the following nucleotide substitutions using the following primers (Biocorp Inc., Montreal, QC, Canada):

• Q49K: C to A at position 201

• Forward: 5′-GGGAGATGAGAAGTCTGCCTTTC GTTG-3′

• Reverse: 5′-CAACGAAAGGCAGACTTCTCATC TCCC-3′

• L90V: C to G at position 475

• Forward: 5′-GTGTCTGTACCCACAGTCTTGTAC CTGGCTCATGTG-3′

• Reverse: 5′-CACATGAGCCAGGTACAAGACTG TGGGTACAGACAC-3′

• R202H: G to A at position 812

• Forward: 5′-CTGTTTCCTCTCTCACCCCACGG AGAAAACC-3′

• Reverse: 5′-GGTTTTCTCCGTGGGGTGAGAGA GGAAACAG-3′

• V216L: G to C at position 853

• Forward: 5′-CTTCATGCTGGTGCTGTCCTTGGT GTCCCTG-3′

• Reverse: 5′-CAGGGACACCAAGGACAGCACCA GCATGAAG-3′

Following sequence verification, the Cx43 mutant constructs, as well as wild-type human Cx43, were cloned into the BamH1 and HindIII restriction sites of the pcDNA 3.1 vector (Invitrogen). The PCR primers used to engineer GFP-tagged constructs consisted of forward 5′-CGGGGTACCAACATG GGTGACTGGAGC-3′ and reverse 5′-CGCGGA TCCTTGATCTCCAGGTCATCAG-3′. The green fluorescent protein tag was fused to the carboxyl-terminal of wild-type human Cx43 and Cx43-mutants using the BamH1 and KpnI restriction sites of the pEGFP-N1 vector (BD Biosciences, Clontech, La Jolla, CA) after removal of the stop codon. The GFP-tagged G60S Cx43 mutant was previously described (22).

Transfections were performed as previously described (23) using Lipofectamine 2000 (Invitrogen) with 1–3 μ g of purified plasmid DNA in Opti-MEM medium (Invitrogen) at 37°C. Transfection efficiency for GFP-tagged mutants was determined 24–48 h later by visualizing live or fixed cells under a fluorescence microscope (Leica, Richmond Hill, ON) or by immunolabeling previously Cx43-deficient cells for exogenous Cx43.

Immunofluorescent Labeling and Confocal Microscopy

Cells grown on glass coverslips were fixed in cold 80% methanol and 20% acetone solution for 15 min at 4°C then rinsed with phosphate buffer saline (PBS) and prepared for immunolabeling as previously described (26). Primary antibodies included P4G9 hybridoma monoclonal anti-Cx43 antibody (Fred Hutchinson Cancer Research Center Antibody Development Group, Seattle, WA, 1:40 dilution), mouse anti-trans-Golgi network38 (TGN38) antibody (BD Transduction Laboratories, Mississauga, ON, 1:200 dilution) and mouse anti-protein disulfide isomerase (PDI) antibody (Stressgen, Victoria, BC, 1:500 dilution). Appropriate anti-rabbit or anti-mouse secondary antibodies (1:200 dilutions) conjugated to Texas red were obtained from Jackson Immunoresearch Laboratories (Westgrove, PA). Cells were imaged on a Zeiss LSM 510 META confocal microscope (Thornwood, NY) as previously described (4).

Microinjection

Microinjection was performed as described previously (23). Briefly, one cell within a cluster of transiently transfected NRK cells expressing the GFP-tagged protein was pressure microinjected with Lucifer yellow (Molecular Probes, Eugene, OR). The percent of microinjected cells that transferred the dye to at least one contacting cell within two min was determined from images collected on a Leica DM IRE2 inverted epifluorescent microscope.

Patch Clamp Electrical Physiology

Functional gap junction coupling between paired N2A cells expressing Cx43-GFP or GFP-tagged Cx43 mutants was assessed using the dual whole-cell voltage clamp technique (23). N2A cells were chosen because they do not endogenously express connexins or exhibit GJIC and, therefore, have no base-line conductance. Furthermore, they are of ideal shape for patch-clamping studies. Individual N2A cell pairs with obvious green fluorescent plaques were selected for recording and the data were acquired using pClamp9 software (Axon Instruments Inc., Downingtown, PA).

Osteoblast Isolation and Retroviral Infection

Primary osteoblasts were isolated from neonatal rat calvaria. Use of rats for this study was approved by the University of Western Ontario Animal Care and Use Committee. The calvaria from 10–12 pups were cleaned of periosteum by scraping, pooled, cut into small pieces and digested with five changes of 525 units/ml of collagenase A (Sigma, Oakville, ON) in buffer (10 mM NaCl, 500 μ M KCl, 0.5 mg/ml K₂HPO₄, 100 μ M CaCl₂, 2 mM HEPES, 3 mM D-mannitol, 2 mg/ml glucose, 2 mg/ml BSA, 1% antibiotics, pH 7.3) at 37°C, shaking at 100 RPM. The first three incubations lasted 10 min while the last two were of 20-min duration. Cells released from the last two digestions were pooled, pelleted, washed, resuspended, and cultured in α MEM media (Invitrogen) supplemented with 10% fetal bovine serum (Cansera, Rexdale, ON), 100 units/ml penicillin and 100 μ g/ml streptomycin. To permit osteoblast differentiation, the culture media was supplemented with 10 mM β -glycerophosphate and 25 μ M ascorbic acid, with media replacement every 48 h, beginning the day following plating for up to 20 days.

cDNA constructs encoding GFP-tagged wild-type Cx43, G21R, G138R (23) and G60S (22) were cloned into the AP2 retroviral vector (27) as previously described (28). The recombinant vectors were then transfected into 293GPG packaging cells (29) to produce replication-defective viral particle-containing supernatants. Primary cells isolated from rat calvaria were incubated for two consecutive 24-h rounds with the infectious supernatant, as described in Mao et al. (30), beginning on the day of isolation. The cells were then subcultured and distributed equally to 35 mm dishes and differentiation was induced the following day. Cx43 and mutant expression efficiency was determined throughout the experiments by visualizing live or fixed cells under a fluorescent microscope (Leica).

Alkaline Phosphatase Enzyme Assay

The alkaline phosphatase enzyme assay was performed as described in Chung et al. (31) with modifications. Primary osteoblasts were solubilized in 50 mM Tris-HCl, pH 9.8, 100 mM glycine, 0.1% Triton X-100, and protein concentration was determined using a BCA protein assay reagent kit (Pierce, Rockford, IL). Five micrograms of protein were diluted into 100 μ l of 50 mM Tris-HCL, pH 9.8/100 mM glycine buffer in a 96-well plate. Fifty microlitres of an 8 mg/ml p-nitrophenyl phosphate solution were added to the wells and incubated at 37°C for 30 min before stopping the reaction with 50 μ l of 3 M NaOH. The optical density of p-nitrophenol, resulting from the activity of alkaline phosphatase, was then measured by spectrophotometry (Bio-Tek Instruments Inc., Winooski, VT) at a wavelength of 412 nm.

Alkaline Phosphatase and Von Kossa Staining

Histological staining for alkaline phosphatase activity and mineralization was performed as described in Bonnelye et al. (32) with slight modifications. Briefly, on designated days, culture dishes of primary osteoblasts were fixed with 10% cold neutral formalin buffer for 15 min and washed in distilled H₂O. Fixed cells were incubated in freshly prepared substrate of 100 μ g/ml Naphthol AS MX PO₄ (Sigma), 600 μ g/ml Red Violet LB salt (Sigma), 0.1 M Tris-HCl, pH 8.3 for 45 min at room temperature and rinsed in H₂O several times before incubation at 37°C in 2.5% silver nitrate solution. Cells were then rinsed in H₂O several times and imaged with an HP scanjet 5470C precision scanner (Mississauga, ON).

Statistics

Data for the patch-clamping and the alkaline phosphatase assay are presented as mean values ± standard error (SE). For patch-clamping, data is presented for n cells and unpaired Student's t-tests were used to determine whether the conductance was statistically significant (p-value < 0.0001) from that in cells expressing wild-type Cx43.

RESULTS

Cx43 Mutants Localize to the Cell Surface of HeLa and NRK Cells

To determine the localization pattern of the Cx43 mutants, untagged and GFP-tagged constructs were transfected into Cx43-negative and GJIC-deficient HeLa cells or Cx43-positive and GJIC-competent NRK cells. Localization patterns were similar for all untagged and tagged mutants. Populations of all mutants were able to traffic to the cell membrane and form structures reminiscent of gap junction plaques in HeLa cells but subpopulations of both the R202H mutants and the V216L mutants were also retained intracellularly (Figure 1, left column). To assess the localization patterns of Cx43 mutants under conditions more similar to that found in ODDD patients where both wild-type and mutant alleles would be expressed, Cx43-positive NRK cells were used. A similar localization pattern to that observed in HeLa cells was found for all the mutants indicating that intracellular subpopulations of R202H or V216L mutants were not notably rescued to the cell surface by the coexpression of endogenous Cx43. The Q49K and L90V mutants were primarily transported to the cell surface (Figure 1E, Figure 1F, Figure 1H, Figure 1I) and resembled that of wild-type Cx43-GFP (Figure 1B, Figure 1C). However, while a subpopulation of the R202H (Figure 1K, Figure 1L) and V216L (Figure 1N, Figure 1O) mutants localized to the plasma membrane, a significant population of R202H and V216L also appeared to be localized to the endoplasmic reticulum and Golgi apparatus as assessed by colocalization patterns to resident proteins of these organelles. Consequently, the R202H and V216L mutations may impede normal Cx43 trafficking to the plasma membrane as has been shown previously for R202H (25).

Figure 1. Cx43 Mutants Form Channels with Reduced Electrical Conductance. Dual-patch clamp functional analysis of Cx43 mutants expressed in GJIC-deficient N2A cells revealed that Q49K, L90V, and R202H form channels with only minimal conductance while V216L did not assemble into functional channels. Asterisks denote significance difference from Cx43-GFP channels at a p-value < 0.0001. Inset: an example of a representative trace illustrating the difference in conductance between Cx43-GFP channels and a channel formed by the L90V-GFP mutant.

Cx43 Mutant Channels Have Reduced or No Electrical Conductance in N2A Cells

In order to determine if the Cx43 mutants that localize to the cell surface are forming functional gap junction channels, we expressed GFP-tagged wild-type Cx43 and Cx43 mutants in GJIC-deficient N2A cells. We used a dual-patch clamp technique to test the level of conductance between two neighboring cells displaying fluorescent plaque formation. It is assumed the GFP-tag does not interfere with conductance measurements of Cx43 mutants as was found to be the case for wild-type Cx43 (33). N2A cells transfected with Cx43-GFP showed a mean gap junctional conductance of 42.1 nS. When N2A cells expressed Q49K-GFP, the amount of gap junctional conductance was only 2.6 nS. Cells expressing L90V-GFP also had a low conductance of only 4.2 nS and R202H produced channels with a conductance of only 5.6 nS (Figure 2). The gap junctional conductance status of cells expressing V216L-GFP was negligible with a conductance of only 0.5 nS (Figure 2). These results suggest that these ODDD-associated Cx43 mutants were either incapable or inefficient in forming functional gap junction channels.

Figure 2. Cx43 Mutants Localize to the Cell Surface. Populations of all the mutants traffic to the cell membrane in connexin-deficient HeLa cells (left column, large image = GFP-tagged; insert = immunolabeled untagged wild-type and Cx43 mutants). Populations of GFP-tagged Cx43 and Cx43 mutants were also transported to the plasma membrane in Cx43-positive NRK cells. No substantial population of wild-type Cx43 or the Q49K and L90V mutants were localized to the endoplasmic reticulum (as defined by the PDI localization pattern, label-red) (B, E, H) or to the Golgi apparatus (as defined by the TGN38 localization pattern, label-red) (C, F, I). However, a significant subpopulation of the R202H and V216L mutants localized to both the endoplasmic reticulum (K, N) and the Golgi apparatus (L, O). Cell nuclei were stained with Hoechst 33342 and appear blue. Bars = 10 μ m.

Cx43 Mutants Exert Dominant-Negative Effects on Endogenous Cx43

In order to assess whether Cx43 mutants exert any inhibitory effect on coexpressed functional wild-type Cx43, dye coupling assays were performed in NRK cells. NRK cells expressing Cx43-GFP, as a control, had a similar percentage of dye coupled cells as wild-type, which was notably higher than any of the mutant-expressing NRK cells. Only 65% of microinjected NRK cells expressing Q49K-GFP exhibited dye transfer as compared to 97% coupling in wild-type NRK cells implying a dominant inhibition of wild-type Cx43 functional gap junction channels (Figure 3). In addition, the L90V-GFP, R202H-GFP and V216L-GFP mutants exerted a robust dominant-negative effect and reduced coupling in NRK cells to 18%, 12.5% and 14.6%, respectively (Figure 3). In addition to the human Cx43 mutants, we also investigated the Cx43 mutation, G60S, found in the Gja1^Jrt/+ mouse that displays an ODDD-like phenotype (22). NRK cells transfected with the G60S-GFP construct demonstrated only 14.6% dye coupling (Figure 3) confirming the strong dominant-negative effect on endogenous wild-type Cx43 previously shown in mouse granulosa cells (23).

Figure 3. Cx43 Mutants Exert Dominant-Negative Effects on Endogenous Cx43. GFP-tagged wild-type Cx43 and Cx43 mutants were expressed in Cx43-positive NRK cells and the incidence of Lucifer yellow dye-coupling was assessed. The Q49K mutant reduced the incidence of overall dye coupling by over 30% while the L90V, R202H, V216L, and G60S mutants inhibited the incidence of dye coupling by over 80%. Inset shows an example of a representative image series of the Q49K-GFP mutant that forms plaques that do not allow dye transfer. Asterisks = microinjected cell; Bars = 10 μ m.

Cx43 Mutants Do Not Inhibit Rat Calvarial Osteoblast Differentiation

One of the most obvious phenotypes displayed by ODDD patients are skeletal abnormalities. The craniofacial involvement includes a long, thin nose with hypoplastic alae nasi and a prominent nasal bridge, a short palpebral fissure, mandibular overgrowth and cleft palate and cranial hyperostosis (15, 34, 35). In the appendicular skeleton, hypoplasia and aplasia of the phalanges are observed (15). In order to examine the role of Cx43 mutants in bone development, we infected primary cells isolated from neonatal rat calvaria with replication-defective retrovirus encoding ODDD-associated Cx43 mutants. For this osteoblast differentiation study, we chose human G21R and G138R mutants (23) as well as the G60S mutant (22), all previously shown to be functionally incompetent and having a dominant-negative effect. After growing the cells in differentiation-permissive media, the cultures were assayed for markers of osteoblast differentiation, alkaline phosphatase (ALP) activity, and mineralization at discrete time points. Consistent with the literature, osteoblasts expressed Cx43 and exhibit a time-dependent increase in alkaline phosphatase activity (Figure 4). Infection with the empty vector did not alter endogenous Cx43 expression, its distribution, or the levels of alkaline phosphatase activity (Figure 4A, Figure 4B). For analysis purposes, staining for alkaline phosphatase in mutant-expressing cells was compared to cells infected with an empty retroviral vector. Wild type and empty vector control osteoblasts revealed a time-dependent increase in alkaline phosphatase activity and when cultured in differentiation-permissive media, began to mineralize by day 20 (Figure 4A, Figure 4B). Osteoblasts showed no significant difference in ALP activity with the over-expression of wild-type Cx43-GFP or the over-expression of G21R-GFP, G138R-GFP, or G60S-GFP mutants (Figure 4A, Figure 4B). Furthermore, expression of any of these mutants failed to alter the extent of mineralization after 20 days of differentiation (Figure 4B). These results suggest that impairment of Cx43-mediated GJIC does not inhibit the in vitro differentiation of committed rat osteoblasts present at birth. Consequently, ODDD-associated mutants may contribute to abnormalities in craniofacial bone development by affecting patterning and earlier stages of bone development or other cells involved with bone remodeling such as osteoclasts.

Figure 4. Cx43 Mutants Do Not Inhibit Rat Osteoblast Differentiation in vitro. Expression of ODDD-linked Cx43 mutants in rat osteoblasts did not significantly alter alkaline phosphatase (ALP) activity or mineralization as compared to cells infected with empty vector. (A) Increased ALP activity is an enzymatic measure of osteoblast maturation. Standard error is shown for all samples (n = 3–4) except the G60S mutant which is the average of only two experiments. One unit of activity is defined as the amount of enzyme required to hydrolyze 1 mmol of p-nitrophenyl phosphate in 1 min at 37°C. (B) As a measure of properties associated with late stage osteoblast differentiation, Von Kossa staining detects regions of mineralization (black areas) while ALP activity is denoted by the pink staining. GFP-tagged Cx43 mutant expression was assessed by fluorescent microscopy. Bar = 20 μ m. Controls for both (A) and (B) are from cell cultures after five days of growth in the absence of ascorbate and β -glycerophosphate.

DISCUSSION

In this study, we have characterized the localization and function of four ODDD-associated Cx43 mutants. The Q49K, L90V, R202H, and V216L mutants transiently transfected into GJIC-deficient HeLa cells and Cx43-positive, GJIC-competent NRK cells were variably competent in being transported to the cell surface and assembled into gap junction plaque-like structures. However, all mutants exhibited poor channel-forming characteristics as assessed by electrical conductance measurements. Furthermore, each mutant displays modest or robust dominant-negative properties on the function of coexpressed wild-type Cx43. Surprisingly, the human ODDD-associated G21R-GFP, G138R-GFP mutants and the mouse G60S-GFP mutant had no effect on rat osteoblast differentiation in vitro.

ODDD Loss-of-Function Mutants Behave in a Dominant Negative Manner

Several ODDD-associated Cx43 mutations have been characterized in vitro, and they are all classified as loss-of-function mutations. G21R (23, 25), G138R (23), Y17S, A40V, L90V, I130T, and K134E (25) are all capable of trafficking to the membrane to form gap junction-like plaques, but they demonstrate little or no channel conductance. Our study shows a similar phenomenon for the Q49K mutant and confirms the result for the L90V mutant. Shibayama et al. (25) found that the F52dup and R202H mutants were retained intracellularly in HeLa cells, predominantly in the endoplasmic reticulum, although a population of both mutants could be rescued to the cell surface in the presence of wild-type Cx43. Consistent with the previous findings for the R202H mutant, our studies, revealed that a significant population of the mutant was localized to the endoplasmic reticulum and/or Golgi apparatus, however, a subpopulation did traffic to the cell surface in both HeLa and N2A cells although the channels formed had negligible conductance. These results would suggest that populations of the R202H, as well as V216L, mutants are able to escape quality control mechanisms that may be responsible for retaining them in intracellular compartments. It is possible that lower R202H mutant expression levels may explain why Shibayama et al. (25) did not detect a subpopulation of the R202H mutant at the plasma membrane.

In the mouse, the ablation of Cx43 is lethal while Cx43^+/− mice appear to be normal (36). ODDD patients carry one mutant allele and one wild-type allele likely resulting in the equal expression of both wild-type and mutant Cx43, but this arrangement is not adequate for normal development, suggesting that the mutants may act in a dominant-negative manner. To test this possibility and partially mimic the ODDD condition, Cx43-positive NRK cells were engineered to coexpress GFP-tagged mutants. In this study, the ratio of mutant to wild-type Cx43 was not measured but our previous studies indicate the mutant was likely 4–5 times in excess of the wild-type (23). The L90V, R202H and V216L mutants inhibited wild-type Cx43 function by over 80%, as measured by dye transfer. The Q49K mutant was less inhibitory, only reducing dye transfer by 35%. Consistently, the L90V mutant was previously shown to have dominant-negative effects on wild-type Cx43 junctional conductance (25). Interestingly, in that study, the R202H mutant did not inhibit either wild-type Cx43 single channel or macroscopic conductance even when it was in excess of wild-type Cx43. However, as in our study, the R202H mutant did inhibit dye transfer in heteromeric channels with wild-type Cx43 (25). Based on these results, one would expect that the severity of disease found in ODDD patients is due to both the potency of a mutant to act as a dominant-negative, and possibly as a trans-dominant on other members of the connexin family, as well as compensatory mechanisms provided by other coexpressed connexins.

Loss-of-function Cx43 mutations may alter the subcellular localization or trafficking of the mutant and limit its ability to form gap junctions as seems to be the case for R202H and V216L mutants. In these cases, the end result is a reduction of Cx43 gap junction channels and overall GJIC. Alternatively, the mutant may traffic to the cell surface, but have reduced or no channel function as we have seen for Q49K and L90V. In some cases, the mutations may disrupt normal connexin folding or oligomerization resulting in a connexon with an abnormal structure that is incompetent in forming proper gap junction channels. The L90V and V216L mutations are located in the second and fourth transmembrane domains, respectively. These mutations are conserved amino acid substitutions between two nonpolar, hydrophobic and similar sized residues (valine to leucine), thus it is remarkable that these conserved substitutions would alter connexin or connexon structure sufficiently to perturb functional channel formation. However, there is some evidence that residues within the second transmembrane domain line the pore and even small changes in this region may have major effects on proper Cx43 oligomerization into a hemichannel (reviewed in (37). In Cx32, the amino acid in the analogous position (L89) to L90 was specifically identified by the substituted cysteine accessibility method as lining the pore when the channel is in the open state (38). While the precise function of the fourth transmembrane domain is unknown, it is obvious that even a small change in this region, as with V216L substitution, is crucial to the formation of functional channels.

The Q49K and R202H mutations are located in the first and second extracellular loop, respectively, and these domains play an integral role in hemichannel docking (39). Consequently, these mutations may not affect connexon structure or channel permeability but instead simply prevent proper connexon-connexon docking due to a structural change that alters the 3-dimensional placement of the conserved cysteine residues. R202H is a conserved amino acid substitution from a positively charged arginine to a positively charged histidine, however the structures of these residues are quite different. Q49K is a nonconserved amino acid switch, from an uncharged glutamine to a positively-charged lysine. Therefore, as a first approximation, it would appear that the Q49K mutation would be more potent as a dominant-negative when in fact in this study it only modestly affects the function of wild-type Cx43. However, due to the fact that we cannot quantify the expression levels of the Cx43 mutants in individual NRK cells used for dye coupling analysis, it remains to be seen if all Cx43 mutants are equally effective in dominantly inhibiting wild-type Cx43 function.

ODDD Mutants Do Not Disrupt Rat Calvarial Osteoblast Differentiation

Many of the phenotypes associated with ODDD, including congenital craniofacial deformities and limb defects, may be caused by abnormalities in bone development. Cx43 is expressed in osteoblasts (40, 41, 42, 43, 44), osteocytes (41, 45, 46) and osteoclasts (41, 47, 48), all of which are capable of functional coupling. It is unknown specifically how Cx43 gap junctions contribute to bone formation and remodeling but it possibly involves calcium signaling (49, 50). Nonetheless, Cx43 is essential in normal growth and development of bone. The Cx43 null mouse embryonically exhibits delayed ossification in the skull, facial bones, and axial and appendicular skeleton (51). Furthermore, when osteoblasts are isolated from calvaria and long bones of Cx43 null mice, transcription and translation of osteoblast-specific markers are reduced in comparison to cells isolated from the wild-type or heterozygote (51). In addition, the Cx43 null osteoblasts demonstrate an increased rate of drug-induced apoptosis and delayed alkaline phosphatase activity and culture mineralization (51, 52). Indeed, the inhibition of Cx43-GJIC in osteoblast cell lines results in a decrease in transcription of osteoblast specific genes (53) and reduction in alkaline phosphatase activity and mineralization (54). Since the Cx43 null mouse dies at birth, a study using targeted Cx43 ablations in differentiated osteoblasts in vivo to analyze the role of Cx43 in adult bone is underway (55).

In our study, it was surprising that neither alkaline phosphatase activity or mineralization was significantly altered following induction of differentiation of rat calvarial cells overexpressing the dominant-negative ODDD-associated Cx43 mutants. This suggests that rat osteoblasts have reached a threshold level of differentiation at birth and can continue on this differentiation path regardless of full Cx43 function. Indeed, in the case of the Cx43-null mouse, both endochondral and intramembranous ossification are delayed throughout embryonic development but mineralization of most bones reach normal levels by birth except for the calvaria (51). Disruption of Cx43-mediated GJIC may affect the timing of patterning cues thereby altering the time course of differentiation and the sensitivity to Cx43 mutants, in terms of osteoblast differentiation, may also be tissue-specific explaining the differential effects on certain skeletal elements. The contribution of Cx43 to the commitment of mesenchymal or bone marrow precursors to the osteoblast lineage, along with changes to bone remodeling mechanisms, should be further investigated. Furthermore, since only rat osteoblasts were studied here, some attention should be given to the possibility that the expression of ODDD-associated Cx43 mutants may have notable effects on the differentiation of primary mouse and/or human osteoblasts reflecting species variations in the properties of osteoblast differentiation.

While in vitro experiments remain an asset to the investigation of the role of ODDD-associated mutations in bone development, the Gja1^Jrt/+ mouse model of human ODDD recently identified (22) allows us to examine a model that mimics more closely the dosage of mutant to wild-type Cx43 that human ODDD patients maintain. To this end, analysis of cells isolated from this mouse is invaluable for the examination of the mechanism by which Cx43 affects bone development and will compliment our studies of the human ODDD-associated mutants.

ACKNOWLEDGEMENTS

We would like to thank Gregory IL Veitch for assistance with the microinjection studies. This work was funded by grants to DWL, SMB and DB from the Canadian Institutes of Health Research and the Canada Research Chair Program to DB and DWL. EM was funded by the National Sciences and Engineering Research Council.