Loss of Coupling Distinguishes GJB1 Mutations Associated with CNS Manifestations of CMT1X from Those Without CNS Manifestations

CMT1X, an X-linked inherited neuropathy, is caused by mutations in GJB1, which codes for Cx32, a gap junction protein expressed by Schwann cells and oligodendrocytes. Many GJB1 mutations cause central nervous system (CNS) abnormality in males, including stable subclinical signs and, less often, short-duration episodes characterized by motor difficulties and altered consciousness. However, some mutations have no apparent CNS effects. What distinguishes mutations with and without CNS manifestations has been unclear. Here we studied a total of 14 Cx32 mutations, 10 of which are associated with florid episodic CNS clinical syndromes in addition to peripheral neuropathy. The other 4 mutations exhibit neuropathy without clinical or subclinical CNS abnormalities. These “PNS-only” mutations (Y151C, V181M, R183C and L239I) form gap junction plaques and produce levels of junctional coupling similar to those for wild-type Cx32. In contrast, mutants with CNS manifestations (F51L, E102del, V139M, R142Q, R142W, R164W T55I, R164Q and C168Y) either form no morphological gap junction plaques or, if they do, produce little or no detectable junctional coupling. Thus, PNS and CNS abnormalities may involve different aspects of connexin function.

Scientific RepoRts | 7:40166 | DOI: 10.1038/srep40166 of identified point mutations, specifically L239I, Y151C, V181M, and R183C (ref. 31 and G. Nicholson, personal communication), do not have these subclinical CNS abnormalities. We refer to these here as PNS-only mutations and provide evidence that they likely do not disrupt O-O or O-A coupling significantly. For clarity the PNS-only mutations are italicized throughout the manuscript.
Until now only one of the clinically characterized mutations associated with florid CNS symptoms (R75W) has been examined electrophysiologically in mammalian expression systems, where the most reliable characterization can be made using dual whole-cell voltage clamp 32 . However, no systematic examination of the effects of clinically characterized mutations on their ability to form functional gap junctions has been undertaken. For example, the PNS-only mutations mentioned above (i.e. those reported as not being associated with any CNS findings) have not previously been studied with respect to their gap-junction forming ability. In this communication we report on our findings in HeLa and Neuro2a cells regarding the ability of 14 mutant forms of Cx32 to form functional gap junctions; 4 mutants are PNS-only and 10 mutants are associated with episodic and florid CNS clinical syndromes as well. The 10 PNS + CNS mutants studied here, with the R75W mutant previously studied, represent 11 of the 12 mutations identified as causing florid CNS symptoms at the time these studies were conceived; to the best of our knowledge, these still represent about 80% of such mutations identified in the literature. In this study we compare the findings for those ten mutations to what we see for the four PNS-only mutations.

Results
The relevant clinical findings in the patients carrying the PNS + CNS mutations examined here and the corresponding literature references are given in Table 1. One other CNS mutant (R75W) previously studied by us 32 is also included. In addition, we evaluated four mutants associated with typical CMT1X but lacking in CNS abnormalities; these patients had normal visual and brainstem auditory evoked potentials and no known episodes of acute, florid syndrome. Figure 1 shows the positions of the 14 Cx32 mutations studied. For the CNS + PNS mutants there are two in the first extracellular loop, four in the second extracellular loop and one in the second and three in the third transmembrane domains. Three of the PNS only mutants are located in the second extracellular loop while one is in the cytoplasmic domain.
Plaque formation -Immunofluorescence Studies. To determine whether mutants can form plaques on the plasma membrane at regions of cell-cell contact, we examined the subcellular localization of all 15 Cx32 mutants. HeLa cells were transiently transfected to express either the WT Cx32 or one of the mutations. All mutants showed expression, indicating that none of the mutations affected immunoreactivity with our Cx32 antibody. Cells transfected with vector alone showed no immunoreactivity.
In every transfected cell, regardless of mutation, connexin expression could be confirmed by a substantial presence of immunoreactive protein in subcellular compartments within the cytoplasm, presumed to be the endoplasmic reticulum and Golgi apparatus, reflecting the normal pathway of Cx32 expression and oligomerization 33 . In addition to the intracellular immunoreactivity some cells exhibited plasma membrane staining. This typically shows as discontinuous fluorescent puncta along the membrane. Puncta are typically only seen where two cells are in direct contact and these have been shown to correspond to the gap junction plaques observed by electron microscope 34 . The possibility that functional gap junction coupling could be present in the absence of immunofluorescent puncta (i.e. without the channel clustering that produces bright immunofluorescent staining) is not supported by evidence from experiments in which EGFP-tagged connexin was used to correlate plaque size with the level of electrophysiologically measured coupling in live cells 35,36 . Therefore, only cells with punctate staining at the plasma membrane were deemed gap-junction competent.
In addition to the expected plaques observed in cells transfected with WT Cx32 we also found plaques in 7 of the 14 mutant Cx32-expressing cells -Y151C, V181M, R183C, L239I, R142Q, V139M, and R164W (Fig. 2). In the remaining 7 mutations (F51L, T155I, C168Y, R142W, E102del, R142Q, V177A and R164Q), we saw no evidence of plaque formation despite abundant intracellular fluorescence. Thus, the plaque-forming group is constituted by cells transfected with all four of the PNS-only mutations as well as three of the PNS + CNS mutations. All others appear to not traffic to the membrane to any significant extent and/or fail to dock with hemichannels on neighboring cells.

Cell-cell coupling -Electrophysiological Studies.
To further characterize differences between PNS-only and PNS + CNS mutants, we evaluated junctional conductance electrophysiologically in the 7 plaque-forming Cx32 mutants as well as WT and vector-only transfected cells. To evaluate the functional properties of each mutant, we transiently transfected them in Neuro2a cells and examined the conductance induced between pairs of these cells in response to voltage steps. This cell type is well suited for the dual whole-cell patch-clamp approach due its more spherical shape and compact size, which are attributes that enable effective voltage clamp throughout the cell. More importantly, Neuro2a cells are well-known for having virtually no contaminating endogenous connexin expression. Because the transfection vector expressed EGFP, transfected cells were selected for recording based on EGFP fluorescence. In this system, the presence of EGFP fluorescence provides a good marker for successful transfection 37 , as shown by the fact that presence or absence of fluorescence correlates with presence or absence, respectively, of gap junctional conductance when examining cells expressing Cx32WT, other WT connexins, or any number of mutants that form functional cell-cell channels. Therefore, using EGFP fluorescence to identify connexin-expressing cells, only pairs of cells in which both cells expressed EGFP were targeted for recording. Note that in these recordings both cells of a pair were expressing the same Cx32 variant (i.e. a homotypic configuration of cell-cell channel pairing). Table 2 summarizes our findings on these 7 mutants. Briefly, all 4 PNS-only mutants (Y151C, V181M, R183C, L239I) exhibit a junctional conductance that is statistically indistinguishable from that recorded in cells transfected with WT Cx32. Figure 3 shows that all 4 PNS-only mutants also have current-voltage relations that are Scientific RepoRts | 7:40166 | DOI: 10.1038/srep40166 indistinguishable from that of WT Cx32. As shown in Table 3, this is also reflected in the similarity of the parameters for the Boltzmann fits to the data of Fig. 3 for Cx32 and the four PNS-only mutants. The similarity of the four mutants to Cx32WT is further demonstrated by heterotypic pairing of each mutant with Cx32WT (Fig. 4). In gap junctions, cell-cell channels are formed of two apposed hemichannels that dock together into a single cell-cell channel. Homotypic gap junction channels are composed of two identical apposed hemichannels. This creates a symmetric cell-cell channel and typically a symmetric conductance-voltage (G j − V j ) relation like those shown in Fig. 3, where in the simplest model each limb of the curve is due to gating of one of the two hemichannels. Heterotypic coupling, on the other hand, can produce symmetric or asymmetric G j − V j relations, depending on the similarity of properties of each of the two hemichannels being paired. The relatively symmetric Gj − Vj relations shown in Fig. 4 add further support to the notion that the four PNS-only mutants have normal functional properties.
In contrast to the PNS-only mutants, all three plaque-forming PNS + CNS mutants (R142Q, V139M and R164W) exhibited no junctional conductance in the homotypic configuration; i.e. results were statistically indistinguishable from what is observed in cells transfected with vector expressing EGFP only (Table 2). To confirm that the mutant protein of the 3 PNS + CNS mutants was indeed present at the plasma membrane, as suggested by the immunofluorescent puncta, we paired each of the 3 mutants with Cx32WT (heterotypic configuration).
Unlike the homotypic configuration, which models the reflexive gap junctions between the layers of myelin sheath within a single myelinating Schwann cell (as well as the gap junctions formed between oligodendrocytes), the heterotypic configuration is likely to obtain in vivo only between oligodendrocytes in the CNS of female carriers, where a Lyonized pattern of expression can occur. In this experiment we used it primarily as a way of detecting the presence of the mutant connexin on one side of the gap junction (i.e. its presence at the plaque) by its effect on the conductance of the WT channels on the other cell. For two of the PNS + CNS mutants (R142Q and R164W), we were able to demonstrate that the mutant was present at the gap junction by virtue of its ability to generate an asymmetrical conductance-voltage (G j − V j ) relation (Fig. 5). This type of asymmetrical Gj − Vj relation is never observed in Cx32WT-expressing cells paired with cells transfected with vector-EGFP (or vector-IRES DsRed), a situation in which no conductance is seen at all. However, to capitalize on heterotypic coupling to confirm the presence of the mutant at the membrane the mutant needs to exhibit an open state at some transjunctional voltage, irrespective of how non-physiological this voltage may be. One also relies on the ability of mutant hemichannels in one cell to dock with the WT hemichannels in the other to form a cell-cell channel.
For the third mutant (V139M), all we could detect was a small amount of very asymmetric junctional current in one of ten cell pairs. A likely explanation is that V139M hemichannels are able to dock between cells but have an extremely low probability of exhibiting the open state. (In addition to using EGFP as a marker for transfection, Cx expression in N2a cells was confirmed for this mutant using immunocytochemistry; not shown.) This is consistent with work from our laboratory 38 showing that in oocytes the V139M mutant formed functional heterotypic channels that showed a very low open probability at relevant voltages. Together, our results make it highly unlikely that any of the PNS + CNS mutants can form functional gap junctions in vivo even when exhibiting gap junction plaques on immunofluorescent studies.

Discussion
Over 400 mutations in Cx32 have been identified in patients with CMT1X (http://hihg.med.miami.edu/code/ http/cmt/public_html/index.html#/); for most, no information is available on whether they could lead to acute florid CNS dysfunction. Many CMT1X patients have stable subclinical abnormalities in their central nervous system as evidenced by abnormal (but asymptomatic) visual or auditory evoked responses. For this reason, we decided to restrict our definition of "PNS-only" mutants to only those which, in addition to having had no history of episodic CNS symptoms, were documented to have shown no evidence of overt or even subclinical CNS involvement, including documentation of normal visual and auditory evoked responses. With this definition as the basis for establishing a potential correlation between symptomatology and the nature of connexin dysfunction we first examined the subcellular localization of ten CMT1X mutations of Cx32 associated with acute transient neurologic CNS dysfunction as well as the only four mutants satisfying our PNS-only definition. There is strong justification for using the presence of punctate staining as a necessary (although not sufficient) condition for a mutant's ability to form functional gap junctions. This is because whenever WT connexins are  These are digital images of HeLa cells that have been transiently transfected to express one of the indicated mutations as well as empty vector (vector) and wild type Cx32 (WT). The cells were immuno-stained for Cx32 (red) and counterstained with DAPI (blue) to label nuclei. Some of the PNS + CNS mutants-F51L, T55I, E102G, R142W, R164W, C168Y and V177A -never displayed gap-junction plaques, while some -V139M, R142Q, and R164W -had gap junction plaques (arrows), as did all of the PNS-only mutants (Y151C, V181M, R183C and L239I) and WT Cx32. Scale bar = 10 microns.  (e) and L239I (g) and the corresponding steady state G j − V j relations for homotypically paired cells (b,d,f,h, respectively). Mutant G j − V j relations are similar to those seen for Cx32WT though G min differs in some mutants. In b, d, f, and h, data are shown as mean + /− SEM. Solid lines are Boltzmann fits to the data shown by the filled squares. See Table 3 for fit parameters.

G max ( + V) G min ( + V) V 0 ( + V) K ( + V) G max (−V) G min (−V) V 0 (−V) K (−V)
Homotypic 32WT  examined, there is a one-to-one correlation between punctate staining and functional gap junctions 35,[39][40][41] . In the few cases in which a connexin does not form GJs (e.g. Cx29 42 ) punctate staining is also not found 43 . When studying subcellular localization of connexins, the ability of connexin mutants to exhibit plaques in HeLa cells has uniformly been a good predictor of plaque formation ability in vivo, both in terms of failure to form plaques (T55I 44,45 , R75W 45,46 , R142W 47 (and this report) and in terms of successful plaque-formation (280 G 44,48 ). Therefore, in this study we used plaque formation in HeLa cells as a screen for the GJ-forming potential of each mutant examined. In the second part of the study we used Neuro2a cells for exogenous expression of all 7 CMT1X mutants that passed the plaque formation screen. Neuro2a cells are a useful model for studying the junctional properties of connexins, both because of their extremely low levels of endogenous coupling and because they are generally well suited for patch clamp electrophysiological recording. While the number of clinically-characterized PNS-only mutations is not as high as one would need to make more definitive statements, several trends do emerge from our data which support a distinction between PNS-only versus PNS + CNS based on parameters related to junctional conductance. First, all four PNS-only mutants produce gap junction plaques as we show in HeLa cells. In addition, all four PNS-only mutants show steady state conductance-voltage relations and magnitudes of junctional coupling that differ only subtly from that of WT Cx32. A number of other mutant forms of Cx32 have similarly been demonstrated to show relatively  Table 3 for fit parameters.
Scientific RepoRts | 7:40166 | DOI: 10.1038/srep40166 normal conductance voltage relations. and/or normal levels of coupling in oocytes 4,49-51 . However only a few have been evaluated in mammalian cells for both magnitude of coupling and conductance voltage characteristics and none of these mutations have been reported in patients with florid acute syndromes, the focus of this study. Second, all PNS + CNS mutants studied, including the previously studied R75W mutant, are either unable to form gap junction plaques or have such abnormal conductance characteristics as to render them functionally unable to mediate coupling at the range of transjunctional voltages likely to obtain between glial cells. In other words, PNS + CNS mutants are associated with changes predicted to lead to complete loss of function of homotypic Cx32 channels.
Our results, when viewed in the context of the known clinical data, suggest the possibility that retention of normal or near normal cell-cell junctions may protect against florid CNS symptoms and possibly against slowing of central conductions (i.e. abnormal VERs and BAERs). In other words, disruption of this function could be a necessary condition for florid CNS symptoms. This line of reasoning is supported by the recent report of recurrent florid symptoms in a patient carrying a Met1Ile mutation 28 , predicted to lead to complete loss of Cx32 protein 52 . This reasoning also suggests the possibility that complete loss of reflexive coupling in Schwann cells may lead to reduced peripheral nerve conduction velocities in patients with CNS + PNS mutants when compared to those with PNS only mutants, a possibility that could potentially be clinically assessed in the future. At the same time, the episodic nature of florid symptoms also suggests that additional not yet identified genetic and/or environmental factors are necessary for CNS manifestations in CMT1X patients. Moreover, if these factors were to have different effects depending on the mutation expressed this would add further diversity to the pathogenesis of CNS symptoms in CMT1X.
In the CNS, oligodendrocyte-oligodendrocyte (O:O) coupling is mediated by Cx32:Cx32 and Cx47:Cx47 homotypic channels, and oligodendrocyte-astrocyte (O:A) coupling is mediated by Cx32:Cx30 and Cx47:Cx43 heterotypic channels 5 . Thus, it is likely that CNS Cx32 mutations affect both Cx32:Cx32 (O:O) and Cx32:Cx30 (O:A) coupling, However, it appears that disrupted Cx32:Cx32 (O:O) coupling is likely to be the more important for the transient CNS dysfunction because the lesions are in the white matter where, at least in mice, O:O coupling predominates over O:A coupling 6,7 . Our results suggest that in CNS, loss of junctional coupling through Cx32 only becomes an important factor under specific but yet to be defined circumstances. Moreover, in light of emerging evidence that connexins play roles other than their classical gap junction-forming role the less predictable nature of CNS manifestations may reflect other, non-junctional, roles for Cx32 in oligodendrocytes.
In the PNS, our data suggest that the ability of Cx32 to form gap junction channels may not even be a deciding factor in whether a mutation causes disease. Nonetheless, clinical data suggest that loss of function is sufficient to cause PNS disease, since patients with deletion of the entire coding sequence of GJB1 or nonsense mutations near the N terminus have a PNS phenotype very similar to that seen in patients with other mutations 53 . It is possible that, whereas conductance of small electrolytes appears to not be significantly affected in the PNS-only mutants, the flux of some larger biologically important molecule(s) could be affected as shown previously for the S26L 54,55 , Delta111-116 54 and R220stop 54 mutants. Such a critical molecule, if one exists, remains to be identified. However, one candidate is ATP, the release of which appears to be dependent on Cx32 hemichannels and is likely affected by mutations in GJB1 56 . In this regard, this study has shown that the four PNS-only mutations, if they indeed obstruct the flux of larger molecules, could be useful in future studies aimed at determining which small molecules could be relevant to CMT1X pathology in peripheral nerve. At the same time, the notion that the primary role of Cx32 in Schwann cells is to mediate a radial pathway for diffusion through reflexive junctional coupling in non-compact myelin (its only proposed role in Schwann cells) might need to be reconsidered, as previously suggested by ex vivo experiments in which dye tracers demonstrated the presence of a reflexive pathway in the peripheral nerve of both Cx32-expressing and Cx32-lacking mice 57 .

Methods
Analysis of transfected cells-immunofluorescent staining. Mutants were generated using the QuikChange II XL Site-Directed Mutagenesis Kit. (Stratagene, La Jolla, CA), from a human GJB1 cloned into pIRES2-EGFP or pIRESpuro3 and subcloned as needed into pIRES2-EGFP (Clontech, Mountain View, CA) 5 ; mutations were confirmed by sequencing.
Communication-incompetent HeLa cells originally obtained from Prof. Klaus Willecke (Institute for Genetics, University of Bonn, Germany) were grown to 80% confluency. Transient transfection utilized Jetprime (Polypus Transfection, Illkirch, France) and DNA containing either WT Cx32, or the relevant Cx32 mutations according to manufacturer's instructions. 28 hours later, cells were washed in phosphate buffered saline (PBS), fixed in acetone at − 20 °C for 10 min, and blocked with fish gelatin in PBS containing 0.1% Triton for 1 hr. at room temperature (RT). Primary antibodies were added in the same blocking solution and the samples incubated overnight at 4 °C. For Cx32 staining, a rabbit antiserum to the C terminus produced in the Scherer lab 58 was used (diluted 1:2500). After washing in PBS, fluorescein-conjugated donkey cross-affinity purified secondary antibodies (diluted 1:200; Jackson ImmunoResearch, West Grove, PA) were added in the same blocking solution for 1 hr at RT. Cell nuclei were visualized with 4′ ,6′ -diamidino-2-phenylindole (DAPI) (Sigma). Slides were covered with mounting medium (Vectashield, Vector Laboratories, Burlingame, CA) and images were photographed under a Leica epifluorescence microscope with a digital camera.

Analysis of transfected cells-dual whole cell recordings. pIRES2-EGFP vectors containing DNA
encoding human Cx32WT or mutants were transiently transfected into confluent Neuro2a cells. Homotypic cell pairs were prepared by re-plating one day following transfection. For heterotypic cell pairs, transfected cells were washed and cells expressing a pIRES2-EGFP construct expressing mutant Cx32 were mixed with cells expressing a pIRES2-DsRed construct expressing WT Cx32, in a 1:1 ratio. Coupling was assessed by dual whole-cell patch clamping of cell pairs 6-48 hours after re-plating as previously described 32 . Only cells clearly expressing the appropriate fluorescent proteins (thus indicating that they were transfected) were used for recording. Recording solutions used were as follows (in mM): pipette solution, 145 CsCl, 5 EGTA, 1.4 CaCl 2 , and 5.0 HEPES, pH 7.2; bath solution, 150 NaCl, 4 KCl, 1 MgCl 2 , 2 CaCl 2 , 5 dextrose, 2 pyruvate, and 10 HEPES, pH 7.4. Heterotypic pairings between cells are shown as "connexin expressed in cell 2/connexin expressed in cell 1"; normalized junctional conductance-junctional voltage (G j − V j ) relations were determined from isolated pairs by measuring junctional current (I j ) responses in cell 2 following 12.5 second junctional voltage pulses (from − 100 mV to 100 mV in 20 mV increments) applied to cell 1, and applying Ohm's law. For all homotypic and heterotypic pairings with G j − V j plots fit to a double Boltzmann distribution, steady state conductance levels were estimated by fitting current traces to a sum of exponentials and using the conductance level at t = ∞ Baseline (V j = 0) junctional conductances were similarly determined by measuring instantaneous I j responses to voltage steps from V j = 0 to + 40 mV and to − 40 mV and using the average of the two measurements. Cytoplasmic bridges were excluded by demonstrating sensitivity of junctional conductance to application of octanol-containing bath solution. G j 's were not corrected for series resistance. For G j − V j plots, data was imported into Origin (Originlab, Northampton, CA), and fit to a "double Boltzmann" equation of the form: where G ss is the steady state junctional conductance normalized to V j = 0, G max1 and G max2 are the maximal and G min1 and G min2 the minimal normalized conductances for the limb reflecting the normalized residual conductances for the negative or positive limbs, and V 01 and V 02 are the voltage at which the conductance for the negative or positive limb is 1/2 of the difference between G min and G max for that limb. K 1 and K 2 are parameters which reflect the slope of the negative or positive limb of the G j − V j plot and are a measure of voltage sensitivity. See Abrams et al. 49 for further discussion.
Statistics. All statistical tests were performed in GraphPad (San Diego, CA) Prism. For dual whole-cell patch clamp assays junctional conductance values were compared using the Kruskal-Wallis test with Dunn's post-test.