© 1999 American Heart Association, Inc.
Original Contribution |
Hetero-Domain Interactions as a Mechanism for the Regulation of Connexin Channels
From the Departments of Pharmacology (K.S., J.L.A., M.M., J.F.E.-V., M.D.) and Microbiology and Immunology (S.M.T.), SUNY Health Science Center, Syracuse, NY.
Correspondence to Mario Delmar, MD, PhD, Department of Pharmacology, SUNY Health Science Center, 766 Irving Ave, Syracuse, NY 13210. E-mail delmarm{at}vax.cs.hscsyr.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
Abstract—Previous studies have shown that chemical regulation of connexin43 (Cx43) depends on the presence of the carboxyl terminal (CT) domain. A particle-receptor (or "ball-and-chain") model has been proposed to explain the mechanism of gating. We tested whether the CT region behaved as a functional domain for other members of the connexin family. The pH sensitivity of wild-type and Ct-truncated connexins was quantified by use of electrophysiological and optical techniques and the Xenopus oocyte system. The CT domain of Cx45 had no role in pH regulation, although a partial role was shown for Cx37 and Cx50. A prominent effect was observed for Cx40 and Cx43. In addition, we found that the CT domain of Cx40 that was expressed as a separate fragment rescued the pH sensitivity of the truncated Cx40 (Cx40tr), which was in agreement with a particle-receptor model. Because Cx40 and Cx43 often colocalize and possibly heteromerize, we tested the pH sensitivity of Cx40tr when coexpressed with the CT domain of Cx43 (hetero-domain interactions). We found that the CT domain of Cx43 enhanced the pH sensitivity of Cx40tr; similarly, the CT domain of Cx40 restored the pH sensitivity of the truncated Cx43. In addition, the CT domain of Cx43 granted insulin sensitivity to the otherwise insulin-insensitive Cx26 or Cx32 channels. These data show that the particle-receptor model is preserved in Cx40 and the regulatory domain of one connexin can specifically interact with a channel formed by another connexin. Hetero-domain interactions could be critical for the regulation of heteromeric channels.
Key Words: connexin • pH, regulation • insulin • Xenopus oocyte • hetero-domain interaction
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
Multicellular organisms communicate from cell to cell by means of intercellular channels called gap junctions. Gap junctions provide a pathway for the intercellular passage of ions, small molecules, and second messengers. They mediate diverse processes such as the propagation of electrical signals through the specialized conduction system1 2 and myocardium,3 the regulation of cell growth,4 and cardiac development.5 6 Current evidence suggest that not only the presence but also the proper regulation of gap junctions are essential for normal tissue function and in some cases for the preservation of life.5 6 7
Gap junctions are formed by integral membrane proteins called connexins. Six individual connexins oligomerize to form a hemichannel (a connexon), and 2 connexons dock to form one functional channel across a narrow extracellular gap. Currently, 14 members of the connexin family have been identified in the rodent.8 9 In the adult mammalian ventricle, the most abundant gap junction protein is connexin43 (Cx43). Our laboratory has been interested in characterizing the molecular bases for the chemical regulation of Cx43 channels. We have demonstrated that truncation of the carboxyl terminal (CT) region of Cx43 significantly impairs the pH or insulin sensitivity of the channel; the function can be restored by separate coexpression of the CT fragment.10 11 These results led us to propose a particle-receptor model for chemical regulation of Cx43 similar to the ball-and-chain model of voltage-dependent inactivation.12 Whether this model is applicable to the chemical regulation of other connexins remains to be determined. It is also unclear how chemical signals modulate the communication between cells that express more than one connexin isotype. Indeed, several isotypes often colocalize to the same gap junction plaque.8 13 Specific cases of heteromerization (ie, more than one connexin isotype in a hemichannel) have been demonstrated biochemically14 15 and functionally.16 17 However, studies on the regulation of heteromeric channels are lacking. Because the sensitivity of different homomeric gap junctions to biochemical modulators varies widely,8 18 the question arises as to the types of interactions that occur between connexins to integrate regulatory signals in heteromeric channels.
In this study, we describe a quantitative characterization of the pH sensitivities of various wild-type and mutant connexins truncated at their CT end (24 to 40 residues after the predicted end of the fourth transmembrane domain). The data show a wide dispersion of pH sensitivities among the connexin isotypes. The dispersion was unrelated to the molecular weight of the connexin or to the length of the CT domain. Moreover, we found that CT truncations in Cx43, Cx40, and to a lesser extent Cx37 and Cx50 result in altered pH sensitivity, although CT truncation of Cx45 does not. We found also that the particle-receptor model of pH regulation was preserved in Cx40. Finally, we showed that the regulatory domain of one connexin interacts with a homomeric channel formed by a different isotype (called hetero-domain interactions). We propose that in heteromeric channels, the regulation of the gap junction results from a complex integration of homo- and hetero-domain interactions.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
cDNA and mRNA Preparation
Cx26, Cx37, Cx40, and Cx45 cDNAs were gifts from Drs Bruce Nicholson (University of Buffalo) and Klaus Willecke (Universität Bonn). Cx46 and Cx50 cDNAs were gifts from Drs Thomas White (Harvard University) and David Paul (Harvard University). Mutagenesis was either site directed or performed by use of a polymerase chain reaction (PCR).19 The position of the stop codon was modeled after a Cx43 mutant that was truncated at amino acid 257 (M257)20 21 (Table 1
). In this report, we use the
term "Cx43tr" for truncated Cx43 rather than M257 to be
consistent with the terminology used for other Cx truncation
mutants. The topologies of all connexins were inferred from hydropathy
plots (Cx26,22 Cx37,23 Cx40,24
Cx43,25 Cx45,26 Cx46,27 and
Cx5028 ) and a stop codon introduced 24 to 40 residues
from the predicted end of the fourth transmembrane domain (Table 1). Cx50 was truncated at residues 290 and 300 because analogous
truncated molecules may occur naturally as a consequence of
posttranslational cleavage in the lens.29 Cx46 was
truncated at residue 266 (Cx46tr266) but failed to form gap junction
channels; thus, one additional truncation was performed at residue 286.
The following primers were used (Genosys Biotechnologies): Cx40tr
sense, 5'-GAC AAG CAC TAG CTG CCT GGC-3'; Cx45tr sense, 5'-GGT GCT TAA
AAT TAT CCT TTC-3'; Cx46tr266 antisense, 5'-GCT GGG AGG TCA AGA GTT
GGC-3'; Cx46tr286 sense, 5'-GTC CCA CAT AAC AGG GAA AGG-3'; Cx50tr290
sense, 5'-CCT TTG ACA TAG GTT GGA ATG-3'; and Cx50tr300 sense, 5'-GTG
GAG ACC TGA CCT CTT TCG-3'. Mutant cDNAs were subcloned into pcDNA3
(InVitrogen) or pBluescript SKII+ (Stratagene). Mutagenesis of Cx37 was
performed by use of PCR.10 The sense primers were designed
to provide a T7 promoter before the beginning of the coding region
(shown in bold) so that the cDNA that resulted could be used directly
as a template for in vitro transcription.30 The antisense
primer had the following premature stop codons: Cx37tr sense,
5'-GCT AAT ACG ACT CAC TAT AGG GAT CCT GGA AAC ATG GGC GAC
TG-3'; and antisense, 5'-CAT GAA TTC GGT CTG AGG CAC TGC CCT AGG
CCG-3'.
|
cDNA for the CT fragments of rat Cx43 (rCx43; residues 258 to 382),10 mouse Cx40 (mCx40; residues 249 to 358), and mouse Cx45 (mCx45; residues 276 to 396) were amplified by PCR with the following primers: Cx40- hemagglutinin (HA; forward primer), 5'-CCG GAT CCA TGT ACC CAT ACG ACG TCC CAG ACT ACG CTC TGC CTG GCC CTC CCA CCA GC; Cx40 (forward primer), 5'-CGG ATC CAT GCT GCC TGG CCC TCC CAC CAG C; Cx40 (reverse primer), 5'-CAT GAA TTC AAA GGA GGA TCA CAC TGA CAG; Cx45 (forward primer) 5'-CCG GAT CCA TGA ATT ATC CTT TCA CTT GGA AC; and Cx45 (reverse primer), 5'-CAT GAA TTC CAA CCA AGA TTA AAT CCA GAC. PCR products were subcloned into pBluescript SK+ (Stratagene) and transcribed. The Cx40 CT region was amplified in 2 forms: a form with 5' epitope tag HA and another without (shown in bold above). Epitope tagging was used because antibodies to the CT region of Cx40 were not commercially available. A chimeric construct, Cx40tr-43CT, of mCx40tr (residues 1 to 244) and rCx43 CT domain (residues 255 to 382) was produced by use of PCR. The following primers produced a unique BamH1 restriction site that consisted of the glycine residue at 244 of Cx40 and the serine residue at 255 of Cx43: Cx40 (forward primer), 5'-CGA AAG CTT GGC AAG ATG GGT GAC TGG AGC-3'; Cx40 (reverse primer), 5'-CGA GGA TCC CTG CCG TGA CTT GCC AAA G-3'; Cx43 (forward primer), 5'-CGA GGA TCC CCA TCA AAA GAC TGC GGA TC-3'; and Cx43 (reverse primer) 5'-GCA AGC TTG AAT TCC AAG CCG GTT TAA ATC TCC-3'. Termination of Cx40 occurred at residue 244 rather than at 249 and the CT domain of Cx43 began at residue 255 rather than at 258,10 so that no additional amino acid residues were added. The resulting 2 PCR-derived cDNAs were sequenced to eliminate the possibility of polymerase errors before cloning. Wild-type and mutant cDNAs were transcribed in vitro with the appropriate in vitro transcription kit (mMessage mMachine, Ambion, Inc).
Oocyte Preparation and Electrophysiology
The procedures for oocyte preparation and the recording
of junctional conductance (Gj) between
oocyte pairs have been previously described.10
Gj was measured with a conventional dual
2-electrode voltage clamp.10 31 All experiments were
performed at room temperature (23°C to 24°C).
Gj was normalized to its maximum for pH
experiments and to control Gj for insulin
experiments.10 11 All stimulation and data
acquisition protocols were performed by use of pClamp software (version
6.0.2, Axon Instruments).
Intracellular pH Measurements and Acidification
The details of this technique have been described
previously.10 Excitation of the fluorophore and
recording of the emitted light was performed with a customized
Spex Fluorolog II system coupled to a Nikon diaphot microscope equipped
for epifluorescence. The method for fluorophore calibration in
Xenopus oocytes was detailed
previously.10 32 Dextran-SNARF (70 000 MW; 17
µmol/L intracellular; Molecular Probes) was excited at 534 nm.
Emission spectra were obtained for calibration purposes; emission
ratios (calculated as 590 to 640 nm) were obtained during the
acidification protocol. This system can accurately measure actual
values of intracellular pH to 1/100 of a pH unit.10 32
Oocyte pairs were superfused continuously with a bicarbonate-buffered
solution that contained the following (in mmol/L): NaCl 88, KCl 1,
MgSO4 0.82, CaCl2 0.74, and
NaHCO3 18. Intracellular acidification was
achieved by bubbling the superfusate with a predetermined
mixture of 95% O2/5% CO2
and 100% CO2; the outlets of the gas sources
were connected to a programmable valve. The proportion of
CO2 was progressively increased, which yielded a
slow ramp of acidification that lasted 
50 minutes.10
For some experiments, it was necessary to alkalinize the cells to
record the entire range of the pH-sensitivity curve. This was
accomplished by incubating the cells in the bicarbonate solution
without the addition of any gases.
Insulin Experiments
Experiments with insulin were performed as previously
described.11 Oocytes were kept in culture medium (0.5x
L-15) before electrophysiological
impalement. Approximately 80 to 120 minutes before data collection,
L-15 was replaced with a sodium acetate Barth's solution (pH 7.4).
Data acquisition started 10 minutes before insulin exposure; 1
µmol/L insulin was perfused continuously in a sodium acetate Barth's
solution (pH 7.4) over the course of the experiment.
Data Analysis and Statistics
The methods for generation of the average pH-sensitivity curves
have been detailed and discussed in previous publications from our
laboratory.10 32 33 34 35 Briefly, junctional currents were
analyzed by use of the pClamp system. Intracellular pH
(pHi) was correlated with
Gj for each experiment, and the data were
fit to a sigmoidal equation of the form used by other authors to
characterize the pH dependence of gap junctions (Origins, version
3.73).31 We refer to the function that correlates
pHi to the normalized
Gj as the pH-sensitivity curve. Convergence
was established by nonlinear least squares. As determined by the
sigmoidal fit, the pH at which normalized
Gj was decreased by 50% (pKa) was used to
compare quantitatively the pH sensitivity of different connexins or
mutants obtained from each group.
The
slope of the curve is described by the Hill coefficient. Whether the
Hill coefficient can be used properly as a measure of cooperativity
under our recording conditions remains to be determined (see
References 10, 32, and 3510 32 35 for additional discussion on the subject).
Data points were binned by averaging the normalized
Gj values for each 0.1 pH units or in some
cases 0.05 pH units as previously described.10 34
This was done to average
pHi-Gj curves from
several experiments, although the actual measurements of
pHi were accurate to 1/100 of a pH unit. All
average data (pKa, control Gj, and Hill
coefficient values) were reported as the mean±SEM of the number of
pairs indicated. The pH sensitivity of every mutant was compared with
that of its own wild-type channel. Only 2 independent variables
were compared for each experimental series: the pKa and the Hill
coefficient. Each experimental series was independent from the others.
Therefore, statistical significance was calculated by Student
t test (unpaired, 2-tailed, P<0.05; Excel,
version 97).
Metabolic Labeling of Cellular Proteins and
Immunoprecipitation of Connexins
Cellular proteins were metabolically labeled and
immunoprecipitated in accordance with previously described
methods.36 37 Oocytes were injected with translabeled
35S (4.5 µCi per oocyte) and cRNA for
the CT region of either Cx40 (HA-Cx40; 40 ng), Cx43 (40 ng), or Cx45
(20 ng) followed by incubation at 19°C for 6 hours. Labeled oocytes
were homogenized on ice in 100 µL per oocyte of buffer
(0.4% SDS; 50 mmol/L Tris, pH 7.4; 100 mmol/L NaCl; 2
mmol/L EDTA; 50 mmol/L NaF; 40 mmol/L 223
ß-glycerophosphate; 1 mmol/L
Na2VO4; 1 mmol/L PMSF;
5 mmol/L DFP; 1 mmol/L DTT; and 1 tablet of complete protease
inhibitor cocktail [Boehringer Mannheim]) with a
25-gauge needle and boiled for 3 minutes. The homogenate
was cleared by centrifugation at 13 000 rpm for 5
minutes. Antibodies to Cx43 (polyclonal; Zymed; 3 µg), Cx45
(polyclonal; Chemicon; 6 µg), and anti-HA (monoclonal;
Boehringer Mannheim; 5 µg) were added to the supernatant and
incubated overnight at 4°C with rotation. The samples were incubated
with protein A–Sepharose CL-4B (Sigma; 125 µL/mL) for 1 hour at
4°C with rotation. Sepharose was pelleted in a microfuge at high
speed for 10 seconds. Pellets were washed twice in TBS (10 mmol/L
Tris-HCl, pH 8; 150 mmol/L NaCl; and 0.02%
NaN3) with 2% Triton X-100 and then twice in
TBS. Pellets were then solubilized in 50 µL of sample
buffer,38 and samples were separated on a 15% SDS-PAGE
gel and visualized by a phosphoimager (model 4455SI, Molecular
Dynamics). Oocytes used in these experiments were derived from the same
frog to decrease expression differences.
Alignments and Predictions
All sequence analyses used programs from the Wisconsin
package, version 9.1 (Genetics Computer Group) unless otherwise
indicated. The cytoplasmic loop (CL) and CT domains were first aligned
with PILEUP and these multiple sequence alignments were further
assessed by use of ClustalX (version 1.64b)39 to achieve
an optimal alignment with the Blosum series protein weight matrix.
Multiple-sequence alignments were used to calculate the percentage of
similarity between pairwise members that used OLDDISTANCES with the
Blosum 62 scoring matrix. Pairwise comparisons were also made (Compare
and Dotplot software programs) to determine whether multiple regions of
similarity existed between a given pair of CT domains. Regions within
the CT domain of Cx43 that were known to be important for pH gating as
determined by mutation analysis40 were compared
with analogous regions in the CT domains of the other connexins tested
(BestFit, Digital Equipment Corp).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
pHi Regulation of Wild-Type Connexins
We tested whether the relationship between pHi and Gj differed among members of the connexin family. Figure 1A
through 1D depicts the
pH-sensitivity curves of various connexins. Measurements of
pHi are presented on the abscissae and
correlated to normalized Gj (ordinates).
Notice that the scale of the pH values on the abscissae varies between
figures. Quantitative information is presented in Table 2. The data show that homomeric gap junctions formed from
different connexins have varying sensitivities to
pHi. For illustration purposes only, the pH
sensitivities of wild-type connexins were grouped in accordance with
one of the primary tissues to which they have been localized. Figure 1A shows the pH sensitivity of 2 lens fiber connexins: Cx46 and
Cx50. Cx50 was the most pH sensitive of all connexins tested, followed
by Cx46. Note that both of these connexins were sensitive to pH in or
near the physiological range. Figure 1B
shows the pH sensitivities of 2 connexins found in the vascular
endothelium, Cx37 and Cx40, whereas Figure 1C
represents connexins found in cardiac
tissue13 41 : Cx43, Cx40, and Cx45. Figure 1D
demonstrates the pH dependence of Cx26 versus Cx32, which
represent 2 hepatocyte connexins. Again, connexins
known to be expressed in the same tissue show different regulatory
properties.
|
), 2 lens-fiber
gap-junction proteins. B, pH dependence of vascular
endothelial gap junction proteins Cx37 (•) and Cx40
(
). D, pH sensitivity of
hepatocyte connexins Cx26 (•) and Cx32 (
|
Is There a Role for the CT Domain in Other Connexins?
Previous experiments have shown that the CT domain of Cx43 was the
primary regulator of channel function in response to
acidification.10 21 On the other hand, removal of the CT
domain did not alter the pH sensitivity of Cx32.42
Therefore, we had to determine whether the CT domain plays a role in pH
gating with other members of the connexin family.
To establish whether a role for the CT domain exists in pH gating,
constructs that expressed truncated connexins were produced for 6
isotypes (Table 1
). Robust functional expression of the
truncated connexins was achieved in all connexins (compare control
Gj values to wild type, Tables 2 and 3), except for the truncation mutants of Cx46 (Table 3).
In this case, Gj was not significantly
higher than that observed in antisense-injected oocyte pairs (Table 3).43
|
Figure 2 shows the pH sensitivity of
wild-type Cx40 compared with its truncation mutant Cx40tr. As in Cx43,
a considerable decrease in pH sensitivity was observed when the CT
region was removed. A pKa value could not be determined for Cx40tr
becausefitting of the Hill equation required maximum and minimum
asymptotic values. A less dramatic effect was observed after truncation
of the CT domains of Cx37 and Cx50. Figure 3A shows a comparison of wild-type
Cx37 to its truncation mutant Cx37tr. A small but statistically
significant change in the pKa of Cx37 resulted from truncation of the
CT domain. Interestingly, the Hill coefficient (slope) changed from
wild-type to truncated Cx37 (Figure 3A). The truncation mutants
of Cx50 at positions 290 and 300 showed a differential effect (Figure 3B and 3C); Cx50tr290 showed a statistically significant
shift in the pH sensitivity versus wild-type (Figure 3B).
However, the shift observed for Cx50tr300 was not statistically
significant (Figure 3C). As is shown in Figure 4, the pH sensitivity of Cx45 was
unaffected by CT truncation. Values of pKa and Hill coefficients for
the wild-type and mutant constructs tested are presented on
Tables 2 and 3.
|
|
|
Comparison of C-Terminal and Cytoplasmic Loop Domains
Previous experiments from our laboratory have shown that residues
260 to 300 and 374 to 382 were important for the pH regulation of
Cx43.40 To determine whether these regions of Cx43
represent conserved structural features common to all connexin
isotypes, the primary sequence of selected connexins were compared. A
multiple sequence alignment of the CT domains of 6 members of the
connexin gene family is shown in Figure 5A. Clearly, the CT domains are quite
divergent except for 2 regions of relative sequence conservation
(Figure 5A). A separate pairwise comparison of the proline-rich
region of Cx43 to the CT domains of the other connexins revealed the
greatest similarity between Cx40 and Cx43 (residues 254 to 264 and 274
to 284, respectively). These areas are in bold on Figure 5A.
Interestingly, these 2 isotypes also exhibit the greatest similarity in
the role of their CT domain in regulating pH sensitivity.
|
A possible role for the cytoplasmic loop of Cx32 and Cx43 on pH gating
has been proposed.44 45 Figure 5B shows a multiple
sequence alignment of the cytoplasmic loop domains of the connexins
tested. In general, the cytoplasmic loops were more similar than the CT
domains. Relevant to pH gating, a histidine at position 95 was
conserved among all of the connexins shown here, with the exception of
Cx45. Shaded areas denote selected regions of high similarity
demonstrated by the software. For several connexins, a histidine was
also conserved at position 98. Alternatively, a tyrosine was
present.
pH Regulation of Cx40 Follows a Particle-Receptor Model
Our current findings show a role for the CT domain of Cx40 on pH
gating (Figure 2). Because this CT domain appears to be
structurally similar to that of Cx43 (Figure 5A), we explored
whether the particle-receptor model was preserved as well. Figure 6 shows the pH sensitivity of wild-type
and Cx40tr channels. Gray triangles show data from Cx40tr channels
coexpressed with the Cx40 CT fragment (as a separate protein). Clearly,
the CT domain was able to interact directly or indirectly with the
pore-forming region of Cx40 to restore pH sensitivity. The pKa and Hill
coefficient values of Cx40tr + Cx40CT were not statistically different
from those obtained from wild-type Cx40 (Table 4). These results
are consistent with a particle-receptor (or ball-and-chain)
model for the pH gating of this connexin.
|
). cRNA for
the CT of Cx40 (38 ng) was coexpressed with that of Cx40tr (0.8 ng).
The absolute amounts of cRNA for the channel and CT domain were
titrated to achieve expression of the channel. pKa values for wild-type
Cx40 and coexpression of Cx40tr and the CT of Cx40 were not
statistically different (Table 4
|
Hetero-Domain Interactions Between Cx40 and Cx43
We explored whether the regulatory domain of one connexin (the CT
domain) could regulate the channel formed by a different connexin
isotype (ie, hetero-domain interactions). The data on Figure 7A and 7B demonstrate
hetero-domain interactions between Cx40 and Cx43, 2 connexins that are
coexpressed in cardiac tissue. Figure 7A demonstrates the pH
sensitivity of gap junctions in oocytes that expressed wild-type Cx43,
Cx43tr, or coexpressed Cx43tr and the CT fragment of Cx40. This average
curve (Cx43tr + CT Cx40) combines 2 separate data sets that result from
the injection of 2 different CT Cx40 constructs: one that contains an
HA epitope tag at its amino end (n=5; see Materials and Methods) and
another one without the tag (n=5). Combined data are presented
(Figure 7A) because the results were not different when
analyzed separately (Table 4). In Figure 7B, we
show the converse experiment of that illustrated in Figure 7A;
the pH sensitivities of wild-type Cx40 and Cx40tr are compared with
that of Cx40tr when coexpressed with the Cx43 CT domain. The data show
that the pH dependence of Cx40tr was not only restored but enhanced
(beyond that of wild-type Cx40) after coexpression of the CT domain of
Cx43 (see Table 4, pKa values). The difference in the pKa values
was statistically significant.
|
Hetero-Domain Interactions: A More Effective pH Gating
Mechanism?
Cx40tr channels were more susceptible to uncoupling if a
heterologous CT domain was expressed (ie, Cx43 CT domain) than if the
homologous CT domain was present (Figure 7B). To determine
whether this was due to an overexpression of the CT domain relative to
the channel (ie, an issue of stoichiometry) or to an enhanced
interaction of Cx40tr with the CT domain of Cx43, a chimera that linked
the Cx40tr channel with the CT domain of Cx43 was constructed (called
Cx40tr-43CT). We tested the pH sensitivity of Cx40tr-43CT (Figure 8) and compared it to that of wild-type
Cx40 and that of Cx40tr plus the CT region of Cx43. The pKa value of
the chimera Cx40tr-43CT was not statistically different from that
obtained from separate expression of the fragments (Cx40tr + CT Cx43;
Table 4, pKa values). However, the chimera was clearly more pH
sensitive than wild-type Cx40 (Figure 8) and Cx43 (Figure 7A). These data suggest that the hetero-domain interaction is
more effective at inducing pH gating than the homologous
counterpart.
|
Specificity of the Cx40–Cx43 Hetero-Domain Interactions
The pH sensitivity of the truncated Cx43 channel was restored by
coexpression of the CT fragment of Cx40; the corresponding interaction
was observed between the truncated Cx40 and the CT domain of the Cx43
isotype (Figure 7A and 7B, respectively). To demonstrate
the specificity of the reaction, we tested whether Cx43tr channels
could interact with the CT region of Cx45. Figure 4 showed that
the CT domain of Cx45 was not involved in the pH-gating reaction of
this channel. Consistent with this observation, coexpresssion
of the Cx45 CT fragment failed to restore pH sensitivity to the
truncated Cx43 (Figure 9A). No
pH-dependent reduction in Gj was observed.
These data show that the CT domains of Cx40 and Cx43 were acting
specifically to modify channel function. Expression of the CT domain of
Cx45 was verified by immunoprecipitation studies (Figure 9B, lane 2). In addition, the data show that all 3 CT domains (Cx45, Cx40,
and Cx43) were expressed at comparable levels (lanes 2, 5, and 8).
Labeled bands of the same molecular weights were not present when
extracts from noninjected oocytes were used (lanes 1, 4, and 7) or when
no antibody was added to the reaction (lanes 3, 6, and 9). These
results confirm that the inability of Cx45 CT domain to regulate Cx43tr
was not consequent to failure of protein expression.
|
Hetero-Domain Interactions Can Mediate
Insulin-Induced Uncoupling
In additional experiments, we demonstrated hetero-domain
interactions between the CT fragment of Cx43 and wild-type Cx32 or Cx26
channels. For these studies, we applied a different experimental model
of connexin regulation. This model is based on the observation that the
extent of intercellular communication between Cx43-expressing oocytes
is significantly reduced when the cells are exposed to extracellular
insulin.11 As in the case of acidification-induced
uncoupling,10 truncation of the CT end of Cx43 prevents
the insulin effect; coexpression of the CT fragment with the truncated
channel restores function.11 We have now found that
wild-type Cx32 and Cx26 gap junctions do not uncouple when exposed to
insulin (Figure 10A and 10B).
However, in both cases, coexpression of the CT fragment of Cx43 as a
separate protein led to a reduction in the degree of cell-to-cell
communication on insulin exposure (Figure 10A and 10B).
Whether the reduction in electrical coupling results from a form of
chemical gating of the channels or from interference with other aspects
of the connexin life cycle (translation-assembly-degradation) remains
to be determined.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
These data show that the differences in pH regulation of connexins are related to the diversity of their primary sequences. Moreover, a role for the CT domain in pH regulation is preserved for some (though not all) members of the connexin family. Cx40 shows the most dramatic decrease in function without its CT domain, which is quite similar to the effect seen in truncated Cx43.10 No role is shown for the CT domain of Cx45, and intermediate effects are observed consequent to CT truncation of Cx37 and Cx50. It is possible that the regulatory behavior of connexins is tailored to the requirements for intercellular communication (extent and nature of the message) of a specific tissue. Though we use pHi as a tool to study regulation and uncoupling, other physiological modulators of coupling (eg, kinases) also show isotype-dependent effects.8 The interactions among connexin isotypes with distinct regulatory mechanisms could influence channel gating properties and lead to the establishment of multiple and unexpected regulatory functions in heteromers. The results also demonstrate that the particle-receptor model of chemical regulation applies to the pH gating of Cx40 and that hetero-domain interactions (those between the CT domain of one connexin isotype and the pore-forming region of another) occur. In fact, a chemically insensitive or less sensitive channel can be regulated by the cytoplasmic fragment of a different connexin isotype. These data allow us to speculate that heteromeric channels could be regulated by hetero-domain interactions among subunits present within the same connexon.
Connexin Diversity and pH Regulation: Structural Bases?
All connexins studied are susceptible to acidification-induced
uncoupling. This would suggest that a pH-dependent mechanism of channel
closure is structurally preserved among connexins. One possibility
(though by no means the only one) is that the preservation of pH
sensitivity is related at least in part to the conservation of specific
primary sequences. A region of homology of particular interest is the
one at the interface between the end of the second transmembrane domain
and the beginning of the cytoplasmic loop. We have previously shown
that mutations of His95 as well as substitutions in amino acid residues
96 and 97 of Cx43 can modify pH sensitivity.44 The
sequence H
H/Y (residues 95 to 98; refers to hydrophobic
residues) is highly conserved among connexins. Whether mutations of
His95, residues 96 to 98, or other regions of the CL could alter the pH
sensitivity of connexins other than Cx43 remains to be determined.
Regions of conservation were also found in the CT domain, particularly
for the last 16 to 20 residues (Figure 5A). However, it is
unlikely that this region is a fundamental determinant of the pH
sensitivity of all connexins because its deletion in Cx45 and
Cx3242 leaves pH gating unaffected. We have found that
amino acid residues 271 to 287 of Cx43 are important for pH-dependent
regulation.33 40 Because a similar region is present
in the CT of Cx40 (residues 254 to 264), future experiments must
address whether this region is responsible for the similarity in
behavior between Cx40 and Cx43.
|
Connexin Regulation by pHi and Tissue Function |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
pH Sensitivity of Cx46 and Cx50 and Regulation of Lens Gap Junctions by pHi
Our data show that Cx46 and Cx50 are highly sensitive to pHi. At first glance, this suggests pHi-mediated regulation of intercellular communication in the native lens. Yet previous studies show that although peripheral fibers uncouple in response to low pHi,47 48 inner fibers do not. This apparent decrease in the pH sensitivity of gap junctions could be an adaptive response to the drop in cytoplasmic pH of the normal lens from the outer cortex (pHi 7.02) to the inner cortex (pHi 6.81).47 Recently, Lin and colleagues29 found that the CT domain of ovine lens Cx50 is posttranslationally cleaved. In additional experiments, Lin and colleagues50 demonstrated a reduced susceptibility of CO2-induced uncoupling of the cleaved connexin when expressed in oocytes. The authors concluded that truncation of Cx50 could be responsible for the preservation of intercellular communication in the lens in spite of the acidic environment. However, we found an impaired pH sensitivity in Cx50 after CT truncation (Figure 3B
); the difference presented here is not as dramatic as
that reported by Lin and colleagues.50 Technical differences may account for the discrepancy in pH sensitivity. The pHi measurements in our experiments were based on ratiometric determinations of the light emission of a pH-sensitive fluorophore (SNARF), whereas Lin and colleagues50 relied on the use of pH-sensitive microelectrodes. As initially noted by Peracchia,51 pH-sensitive microelectrodes are not reliable when intracellular acidification is induced by CO2 exposure because CO2 can modify electrode calibration and seriously skew the results.52 Moreover, the apparent dependence of gap junction channels on pHi could be affected by the rate at which intracellular acidification is achieved.10 32 If the truncation delays the velocity at which the channels are closed, Gj could trail behind pHi values and leave the impression of a lack of pH sensitivity. We have overcome this limitation with the use of slow acidification ramps that minimize a time dephasing of the Gj and pHi variables. Our data show that the truncated forms of Cx50 are still capable of pH gating within the ranges reported for the lens.47 Thus, other cellular factors that are independent of the calpain-mediated truncation of Cx50 must be involved in the preservation of cell-to-cell communication. It is appealing to speculate that the pH sensitivity of a heteromeric channel formed by Cx46 and truncated Cx50 is actually significantly less than that of each individual homomer, because hetero-oligomerization of Cx46 and Cx50 has been demonstrated.14 Modification of channel function as a result of heteromerization is suggested by data presented herein.
Cardiac Connexins and pH Dependence
In certain cardiac pathophysiological
states, including myocardial ischemia and infarction,
hypertrophy, and atrial fibrillation, the loss or abnormal
distribution of gap junction proteins has been postulated as a cause of
slow, nonuniform conduction and thought to be a component of the
arrhythmogenic substrate.53 In addition, the differences
in connexin expression patterns observed in cardiac
tissue13 41 and the differential sensitivity of wild-type
cardiac connexins to acidification (Figure 1C) probably
contribute to the diverse conduction disturbances and
arrhythmias seen in myocardial ischemia. In particular,
the higher pH sensitivity of Cx45 compared with Cx43 appears
consistent with the observation that the conduction system is
more susceptible than the myocardium to
ischemia-induced propagation block.54 55
Role of the CT Domain in Connexin pH Gating
Previous results from our laboratory suggested the CT-domain
length as a structural basis to explain the differences in pH
sensitivity between homomeric Cx43, M257 (a truncated form of Cx43),
and Cx32 channels.10 In our experiments, we found that the
CT domain does participate in modulating pH regulation for other
connexins (Cx37, Cx40, and Cx50). However, a direct relationship does
not exist between the level of pH sensitivity and CT length. This lack
of correlation is emphasized by the results from the truncation of Cx45
as well as the wild-type Cx26 because these 2 connexins can be
regulated by pHi despite the absence of a long CT
domain. These data underscore the fact that the intrinsic mechanisms of
pH regulation may vary among connexins, although the phenomenon of pH
gating is preserved.
Other Connexins and the Particle-Receptor Model
We have previously proposed that pH gating of Cx43 follows a
particle-receptor model,10 which is similar to the
ball-and-chain hypothesis of voltage-dependent
inactivation.12 The data presented show that this
model does not universally apply to all connexins. Yet chemical gating
of Cx40 (and perhaps to a lesser extent that of Cx37 and Cx50) is
consistent with the principle of a particle-receptor
interaction. It is also possible that the receptor structure is
preserved. Indeed, connexins have been classified as 
or ß,
depending on the similarity of their primary sequences.46
Some authors further classify Cx45 and Cx36 as separate
subgroups.9 56 57 Note that those connexins that rely (at
least partly) on the CT domain for pH gating are evolutionarily closer
(-connexins) than those in which pH gating is unaffected by the
absence of the CT domain (ß- and 
-connexins).9 46 A
particle-receptor mechanism may still be operative in other connexins,
although the particle may not be located in the CT domain. We found
that in the CT domain of Cx43, an -connexin was able to regulate
Cx26 or Cx32, both of which areß-connexins. These results show that
hetero-domain interactions occur not only between the CT domain and a
truncated mutant but with a wild-type channel as well. In addition, the
restoration of function by the CT domain of a heterologous channel was
not unique to acidification-induced uncoupling but could also be
applied to other forms of chemically induced regulation. In the context
of the particle-receptor model, our data indicate that the functional
domain that acts as a receptor is well conserved across connexins that
are thought to be far apart in the phylogeny.46 56 57 58
This information can set the stage for the elucidation of the
structural regions that act as receptors during the chemical gating
process.
Hill Coefficients Are Preserved After Truncation and Rescued by a
Separate CT Domain
Hill coefficients were segregated on the basis of the
channel isotype. In wild-type Cx43, the Hill coefficient was 5, and
in Cx40, it was 3 (Table 2). All Hill coefficients collected
from the studies on Cx40tr channels were similar to that of the
wild-type channel, regardless of whether pH gating was restored by the
CT domain of Cx43 or of Cx40 or whether the Cx43CT was expressed as a
fragment or as a chimera (Table 4). Similarly, the Hill
coefficient of Cx43tr in the presence of Cx40CT was close to that of
wild-type Cx43. We have always been cautious about the interpretation
of Hill coefficients.10 32 Yet this is a very
consistent observation that allows us to conclude that the
structures involved in the determination of the Hill coefficient are
contained within the truncated channel and act independently from the
nature of the CT domain. This suggests that if cooperative steps are
involved in the pH gating reaction, they are not due to the presence of
the gating particle. Rather, cooperativity may be consequent to the
effector steps that follow the particle-receptor interaction and lead
to channel closure.
Hetero-Domain Interactions as a Basis for Regulation of
Heteromeric Channels
Functional diversity of connexins can be attributed to the
presence of many connexin gene products, as well as to the
possibility of heteromerization among connexin subunits. To study
functional heteromerization, other investigators have used either
coexpression in exogenous systems or concatenated constructs.
Concatenation of entire subunits, although successfully used for other
channels,59 60 has proven difficult for connexins because
the functions of the individual elements are not
preserved.61 Even linkage of domains has led to some
unexpected results.62 63 The Cx40tr-43CT chimera used in
this study expressed functionally and gated in a manner similar to that
observed when the domains were separately expressed. The apparent
conservation of function of this construct when compared with
others62 may be related to the site at which the
constructs were linked. Experiments to assess this chimera for other
properties such as single-channel conductance, voltage dependence, or
susceptibility to insulin-induced uncoupling will be conducted.
We favored the approach of characterizing hetero-domain interactions as an initial step toward understanding the regulation of heteromeric channels. By expressing different fragments of the 2 separate connexins, we were able to attribute a specific regulatory role of one CT domain on a homologous channel pore formed by a different connexin isotype. This approach yields easily interpretable results because of the simplistic design of the study. However, we realize that these interactions may be different in the physical proximity of a truly heteromeric structure. Whether the particular connexins assessed heteromerize in native tissues and behave as predicted by the hetero-domain interactions remains to be tested.
Hetero-Domain Interactions and Synergism of pH Regulation
The data presented in this study suggest that
promiscuity of interactions between specific regulatory domains may
provide plasticity of function to the integrated, heteromeric gap
junction. For example, the combination of the CT domain of Cx43 with
Cx40tr showed that this hetero-domain interaction is more effective at
closing the channel in response to acidification than the homo-domain
interaction present in the wild-type channel. The pKa recorded
from the Cx43tr+Cx40 CT experiment illustrated in Figure 7A is
also more alkaline that the one recorded in our laboratory when we
studied the interaction between Cx43tr and its homologous CT
domain.10 These data indicate that enhanced pH sensitivity
can be demonstrated in both hetero-domain configurations. It is
tempting to speculate that because of these heterologous interactions,
the pH sensitivity of a heteromeric channel could be higher than that
of either homomer from which it forms.
In summary, our data show that a segregation of function is present in the connexin family. Although some connexins use their CT domain for the purposes of regulation by pHi, others do not. We have demonstrated that the particle-receptor mechanism is preserved in Cx40. In addition, we show that hetero-domain interactions can regulate the gap junction channel for both acidification- and insulin-induced uncoupling. These hetero-domain interactions may be the first level of integration in a complex process that controls the intercellular passage of molecular signals between neighboring cells.
|
Acknowledgments |
|---|
This work was supported by grant PO1-HL39707 (M.D., S.M.T.) National Heart, Lung, and Blood Institute, NIH, and by a Grant-in-Aid from the American Heart Association, New York State Affiliate (J.F.E.V.). The work was performed during M.D.'s tenure as an established investigator of the American Heart Association. We thank Dr José Jalife for his advice and Dr Karen Vikstrom for her advice and critical reading of the manuscript. We also thank Wanda Coombs, Christine Burrer, and Laura Hofmann for their excellent technical support.
Received December 1, 1998; accepted March 12, 1999.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConnexin Regulation by pHi...References |
|---|
- Simon AM, Goodenough DA, Paul DL. Mice lacking connexin40 have cardiac conduction abnormalities characteristic of atrioventricular block and bundle branch block. Curr Biol. 1998;8:295–298.[Medline] [Order article via Infotrieve]
- Kirchhoff S, Nelles E, Hagendorff A, Kruger O, Traub O, Willecke K. Reduced cardiac conduction velocity and predisposition to arrhythmias in connexin40-deficient mice. Curr Biol. 1998;8:299–302.[Medline] [Order article via Infotrieve]
-
Delmar M, Michaels DC, Johnson T, Jalife J. Effects of
increasing intercellular resistance on transverse and longitudinal
propagation in sheep epicardial muscle. Circ Res. 1987;60:780–785.
[Abstract/Free Full Text] - Loewenstein WR, Rose B. The cell-cell channel in the control of growth. Semin Cell Biol. 1992;3:59–79.[Medline] [Order article via Infotrieve]
-
Reaume AG, de Sousa PA, Kulkarni S, Langille BL, Zhu
D, Davies TC, Juneja SC, Kidder GM, Rossant J. Cardiac
malformation in neonatal mice lacking connexin43. Science. 1995;267:1831–1834.
[Abstract/Free Full Text] - Lo CW, Wessels A. Cx43 gap junctions in cardiac development. Trends Cardiovasc Med. 1998;8:264–269.[Medline] [Order article via Infotrieve]
-
Gabriel HD, Jung D, Butzler C, Temme A, Traub O,
Winterhager E, Willecke K. Transplacental uptake of glucose is
decreased in embryonic lethal connexin26-deficient mice. J
Cell Biol. 1998;140:1453–1461.
[Abstract/Free Full Text] - Bruzzone R, White TW, Paul DL. Connections with connexins: the molecular basis of direct intercellular coupling. Eur J Biochem. 1996;238:1–27.[Medline] [Order article via Infotrieve]
- Sohl G, Degen J, Teubner B, Willecke K. The murine gap junction gene connexin36 is highly expressed in mouse retina and regulated during brain development. FEBS Lett. 1998;428:27–31.[Medline] [Order article via Infotrieve]
-
Morley GE, Taffet SM, Delmar M. Intramolecular
interactions mediate pH regulation of connexin43 channels.
Biophys J. 1996;70:1294–1302.
[Abstract/Free Full Text] -
Homma N, Alvarado JL, Coombs W, Stergiopoulos K, Taffet
SM, Lau AF, Delmar M. A particle-receptor model for the insulin-induced
closure of connexin43 channels. Circ Res. 1998;83:27–32.
[Abstract/Free Full Text] -
Hoshi T, Zagotta WN, Aldrich RW. Biophysical and
molecular mechanisms of Shaker potassium channel
inactivation. Science. 1990;250:533–538.
[Abstract/Free Full Text] - Gros D, Jongsma HJ. Connexins in mammalian heart function. Bioessays. 1996;18:719–730.[Medline] [Order article via Infotrieve]
-
Jiang JX, Goodenough DA. Heteromeric connexons in lens
gap junction channels. Proc Natl Acad Sci U S A. 1996;93:1287–1291.
[Abstract/Free Full Text] -
Stauffer KA. The gap junction proteins
ß1-connexin (connexin-32) and
ß2-connexin (connexin-26) can form heteromeric
hemichannels. J Biol Chem. 1995;270:6768–6772.
[Abstract/Free Full Text] - Brink PR, Cronin K, Banach K, Peterson E, Westphale EM, Seul KH, Ramanan SV, Beyer EC. Evidence for heteromeric gap junction channels formed from rat connexin43 and human connexin37. Am J Physiol. 1997;273:C1386–C1389.
- Elenes S, Rubart M, Moreno AP. Multiple gap junciton unitary conductances between isolated canine right atrial myocytes. Circulation. 1998;98(suppl):I–256. Abstract.
- Lau AF, Kurata WE, Kanemitsu MY, Loo LW, Warn-Cramer BJ, Eckhart W, Lampe PD. Regulation of connexin43 function by activated tyrosine protein kinases. J Bioenerg Biomembr. 1996;28:359–368.[Medline] [Order article via Infotrieve]
- Vandeyar MA, Weiner MP, Hutton CJ, Batt CA. A simple and rapid method for the selection of oligodeoxynucleotide-dictated mutants. Gene. 1988;65:129–133.[Medline] [Order article via Infotrieve]
-
Dunham B, Liu S, Taffet S, Trabka-Janik E, Delmar M,
Petryshyn R, Zheng S, Perzova R, Vallano ML. Immunolocalization and
expression of functional and nonfunctional cell-to-cell channels from
wild-type and mutant rat heart connexin43 cDNA. Circ Res. 1992;70:1233–1243.
[Abstract/Free Full Text] -
Liu S, Taffet S, Stoner L, Delmar M, Vallano ML, Jalife
J. A structural basis for the unequal sensitivity of the major cardiac
and liver gap junctions to intracellular acidification: the carboxyl
tail length. Biophys J. 1993;64:1422–1433.
[Abstract/Free Full Text] -
Zhang JT, Nicholson BJ. Sequence and tissue
distribution of a second protein of hepatic gap junctions, Cx26, as
deduced from its cDNA. J Cell Biol. 1989;109:3391–3401.
[Abstract/Free Full Text] -
Willecke K, Heynkes R, Dahl E, Stutenkemper R,
Hennemann H, Jungbluth S, Suchyna T, Nicholson BJ. Mouse connexin37:
cloning and functional expression of a gap junction gene highly
expressed in lung. J Cell Biol. 1991;114:1049–1057.
[Abstract/Free Full Text] - Hennemann H, Suchyna T, Lichtenberg-Frate H, Jungbluth S, Dahl E, Schwarz J, Nicholson BJ, Willecke K. Molecular cloning and functional expression of mouse connexin40, a second gap junction gene preferentially expressed in lung. J Cell Biol. 1992;17:1299–1310.
-
Beyer EC, Paul DL, Goodenough DA. Connexin43: A protein
from rat heart homologous to a gap junction protein from liver.
J Cell Biol. 1987;105:2621–2629.
[Abstract/Free Full Text] - Hennemann H, Schwarz HJ, Willecke K. Characterization of gap junction genes expressed in F9 embryonic carcinoma cells: molecular cloning of mouse connexin31 and -45 cDNAs. Eur J Cell Biol. 1992;57:51–58.[Medline] [Order article via Infotrieve]
-
Paul DL, Ebihara L, Takemoto LJ, Swenson KI, Goodenough
DA. Connexin46, a novel lens gap junction protein, induces
voltage-gated currents in nonjunctional plasma membrane of
Xenopus oocytes. J Cell Biol. 1991;115:1077–1089.
[Abstract/Free Full Text] - White TW, Bruzzone R, Goodenough DA, Paul DL. Mouse Cx50, a functional member of the connexin family of gap junction proteins, is the lens fiber protein MP70. Mol Biol Cell. 1992;3:711–720.[Abstract]
- Lin JS, Fitzgerald S, Dong Y, Knight C, Donaldson P, Kistler J. Processing of the gap junction protein connexin50 in the ocular lens is accomplished by calpain. Eur J Cell Biol. 1997;73:141–149.[Medline] [Order article via Infotrieve]
-
Milligan JF, Groebe DR, Witherell GW, Uhlenbeck OC.
Oligoribonucleotide synthesis using T7 RNA polymerase and
synthetic DNA templates. Nucleic Acids Res. 1987;15:8783–8798.
[Abstract/Free Full Text] - Spray DC, Harris AL, Bennett MVL. Gap junctional conductance is a simple and sensitive function of intracellular pH. Science. 1981;211:712–715.