Vascular endothelial cells are coupled by gap junctions that permit cell-to-cell transfer of small molecules, including signals that may be important for vasomotor responses. Connexin37 (Cx37) and connexin40 (Cx40) are the predominant gap-junction proteins present in mouse endothelium. We examined the effect of eliminating Cx37, Cx40, or both, on interendothelial communication in mouse aorta. Intercellular transfer of biocytin and[2-(4-nitro-2,1,3-benzoxadiazol-7-yl)aminoethyl]trimethylammonium (NBD-TMA)was used to assess gap-junction-mediated coupling. Ablation of Cx40 generally had a greater effect on dye-transfer than ablation of Cx37. The effect of Cx40 ablation on dye-transfer was age dependent. There was a 27-fold reduction in biocytin transfer in embryonic Cx40–/– aortic endothelium, a much larger change than in aortas of 6-7-week-old Cx40–/– animals, which showed a 3.5-fold reduction. By contrast, there was no reduction in biocytin transfer in embryonic Cx37–/– endothelium. Embryonic aortas lacking both Cx37 and Cx40 showed a complete loss of endothelial dye-transfer. Surprisingly,elimination of Cx40 resulted in up to a 17-fold drop in endothelial Cx37 on western blots, whereas deletion of Cx37 reduced endothelial Cx40 up to 4.2-fold. By contrast, in the medial layer, both Cx37 and Cx43 increased∼fourfold in Cx40–/– aortas. Declines in non-ablated endothelial connexins were not mediated by changes in connexin mRNA levels, suggesting a post-transcriptional effect. Our results indicate that Cx37 and Cx40 are the only functional connexins expressed in mouse aortic endothelium and are collectively crucial for endothelial communication. Furthermore, Cx37 and Cx40 are codependent on each other for optimal expression in vascular endothelium.

Introduction

Endothelial cells in the vascular wall are connected by gap junctions– structures that allow for the direct intercellular exchange of ions,small metabolites and signaling molecules (<1 kDa). Gap-junction channels typically cluster into plaques in the plasma membrane and are made up of connexin subunits (Goodenough et al.,1996; Kumar and Gilula,1996,). Nineteen connexin family members have been described in rodents to date, each the product of a distinct connexin gene(Willecke et al., 2002). Individual connexins are expressed in a tissue-specific manner and exhibit functional differences when examined as homomeric channels. In the vascular wall, four connexins have been detected: connexin37 (Cx37), connexin40 (Cx40),connexin 43 (Cx43) and connexin45 (Cx45)(Larson et al., 1990; Bruzzone et al., 1993; Reed et al., 1993; Little et al., 1995; Yeh et al., 1997; Gabriels and Paul, 1998; Traub et al., 1998; van Kempen and Jongsma, 1999; Krüger et al., 2000). Although connexin expression is not uniform in all blood vessels and species,most commonly, arterial endothelial cells express Cx37 and Cx40, whereas vascular smooth muscle cells express Cx43 or Cx45. Cx37 or Cx40 have also been reported in vascular smooth muscle of certain blood vessels(Little et al., 1995; Traub et al., 1998; Li and Simard, 1999;

Nakamura et al., 1999; van Kempen and Jongsma, 1999; Cai et al., 2001; Haefliger et al., 2001; Rummery et al., 2002). In addition, Cx43 is found in a subset of endothelial cells in the rat arterial vasculature (Gabriels and Paul,1998).

Vascular gap junctions could serve several physiological roles. Coupling of vascular endothelial cells is thought to facilitate conduction of vasomotor responses along arterioles by allowing for cell-to-cell transfer of electrical signals (Segal and Duling,1989; Emerson and Segal,2000). Indeed, Cx40–/– mice exhibit diminished conduction of arteriolar dilatation in response to acetylcholine and bradykinin, and they are hypertensive(de Wit et al., 2000). In conduit vessels, gap junctions have been implicated in a role in regulating vascular tone (Christ et al.,1996; Chaytor et al.,1998). Junctional communication may also be important in the control of vascular cell proliferation and migration(Larson and Haudenschild,1988; Pepper et al.,1989; Kurjiaka et al.,1998). In this regard, gap junctions could have important roles in the development of the vasculature and in responses to blood vessel wounding. Despite significant knowledge of connexin expression patterns in blood vessels, the contribution of specific connexins to endothelial communication is still not well understood. In particular, the relative contributions of Cx37 and Cx40 to endothelial coupling has not been fully investigated. While this work was in progress, Krüger et al. investigated the effects of a deficiency of Cx40 and concluded that intercellular transfer of injected dyes was altered in Cx40-deficient aortic endothelium and that expression of endothelial Cx37 was upregulated(Krüger et al., 2002). In this report, we used intercellular dye-transfer to assess interendothelial communication in wild-type, Cx37–/–,Cx40–/–,Cx37+/–Cx40–/– and Cx37–/–Cx40–/– aortic segments isolated from mice of different age groups. In addition, we performed western blots and RNA analysis to determine whether connexin deletion alters the levels of non-ablated connexins. We found that Cx37 and Cx40 are collectively crucial for endothelial communication in mouse arteries and are mutually dependent on each other for optimal expression in vascular endothelium.

Materials and Methods

Animals

Cx37–/– and Cx40–/– mice(C57BL/6 strain) are described elsewhere(Simon et al., 1997; Simon et al., 1998). Cx37+/–Cx40–/– and Cx37–/–Cx40–/– mice(C57BL/6-129/Sv strain) were generated by interbreeding Cx37+/–Cx40–/– mice.

Antibodies

Anti-Cx37, anti-Cx40 and anti-Cx43 sera were provided by David Paul(Harvard Medical School, MA) and were affinity purified(Beyer et al., 1987; Gabriels and Paul, 1998). Selected experiments were confirmed with a commercial Cx37 antibody (Cx37 A11-A) from Alpha Diagnostic International (San Antonio, TX). The following antibodies were obtained commercially: anti-VE-cadherin (11D4.1), from BD Biosciences-Pharmingen (San Diego, CA); anti-caveolin-1, from BD Biosciences-Transduction Laboratories (Lexington, KY);anti-platelet-endothelial cell adhesion molecule-1 (PECAM-1) (M-20) for western blots, from Santa Cruz Biotechnology (Santa Cruz, CA); anti-PECAM-1(MEC 13.3) for immunostaining, from BD Biosciences-Pharmingen; and anti-smooth-muscle actin (1A4), from Sigma-Aldrich (St Louis, MO).

Western blotting

Alkaline-extracted endothelial membranes were collected by passing 200μl of lysis solution (20 mM NaOH and protease inhibitors) three times through the lumen of thoracic aortas, using a 22-24 gauge needle. Only the lumen of the aorta was exposed to lysis solution. Lysates collected from six aortas were pooled. Equal lengths of aorta were extracted in each group. Lysates were passed through a 26-gauge needle ten times and a sample was removed for protein determination. Lysates were spun at 100,000 g for 40 minutes to pellet membranes, which were resuspended in 30 μl of SDS sample buffer. Whole aorta samples (five aortas per group)were collected by removing fat and adventitia and mincing the vessels in 5 mM tris-HCl (pH 7.4), 5 mM EDTA, 5 mM EGTA. Samples were Dounce homogenized (40 strokes) and NaOH was added to a final concentration of 20 mM. After 30 minutes on ice, membranes were pelleted and resuspended in 100 μl of SDS sample buffer. Medial-layer-only samples (five aortas per group) were collected by removing fat and adventitia and extracting the endothelium once with SDS sample buffer (100 μl per 5 aortas) followed by lumenal rubbing with a tungsten wire to remove any residual endothelium. Immunostaining control experiments showed that the endothelium was efficiently removed. SDS-extracted aortas were rinsed well with PBS and then processed as described for whole aortas. Samples were boiled for 5 minutes and run on a 12%SDS-polyacrylamide gel. Cx37, Cx40, Cx43, caveolin-1, VE-cadherin, and PECAM were detected on the same nitrocellulose blots. Following primary antibody incubations, membranes were incubated with horse radish peroxidase(HRP)-conjugated secondary antibodies (Pierce Endogen, Rockford, IL) and then processed for chemiluminescence. VE-cadherin blots were incubated with biotinylated secondary antibody and ABC reagent (Vector Laboratories,Burlingame, CA). Bands were quantified from film by densitometry using a BioRad imaging system. Cx37, Cx40, VE-cadherin and caveolin-1 levels were normalized to PECAM signals. Signals obtained from knockout samples were expressed as a percentage of wild-type signals.

Immunohistochemistry

Tissues were frozen unfixed in Tissue-Tek OCT embedding medium (Sakura Finetek, Torrance, CA) and sectioned at 10 μm. Sections were fixed in acetone at –20°C for 5 minutes, blocked in a solution containing PBS, 4% fish skin gelatin, 1% goat serum, 0.25% Triton X-100, and incubated with primary antibodies. Sections were washed, incubated with CY3-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA),washed and viewed with an Olympus BX51 fluorescence microscope. Images were captured with a Photometrics SenSys 1401 CCD camera. For en face immunostaining, aortas were fixed by perfusion with 0.5% paraformaldehyde and cut open before blocking and incubating with connexin antibodies.

RT-PCR

RT-PCR was performed with RNA obtained from three groups each of wild-type mice, Cx37–/– mice and Cx40–/–mice. Each group consisted of six animals (6-7 weeks old). RNA was isolated from thoracic aorta endothelium by passing 200 μl of lysis solution (4 M guanidinium thiocyanate, 25 mM sodium citrate, 0.5% sodium lauryl sarcosinate,0.1 M β-mercaptoethanol) once through the lumen using a 22-gauge needle. Lysates from six aortas were pooled and passed through a 26-gauge needle ten times. Total RNA was isolated according to Chomczynski and Sacchi(Chomczynski and Sacchi, 1987). RNA was isolated from whole aortas by removing fat and adventitia and homogenizing the aortas in lysis solution. RNA was treated with DNAse I and resuspended in 30 μl. RNA, 6 μl, was reverse-transcribed using an oligo dT18 primer. The final volume was adjusted to 100 μl. 5 μl of cDNAs were used for PCR in 100 μl reactions using Advantage 2 polymerase(BD Biosciences-Clontech, Palo Alto, CA). After 3 minutes at 94°C, five cycles of touch-down PCR were performed with a 1°C decrement in annealing temperature after each cycle. The initial cycling parameters were: 94°C denaturation (30 seconds), 69°C annealing (30 seconds), 72°C extension(30 seconds). Cycles 6-33 were performed with the following parameters:94°C denaturation (30 seconds), 64°C annealing (30 seconds), 72°C extension (30 seconds). 20 μl samples were removed at cycles 25, 27, 29, 31 and 33. Cx37, Cx40, Cx43 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH)bands were quantified. Connexin signals were normalized to GAPDH levels. Signals obtained from knockout samples during the linear portion (cycle 29 was used) were expressed as a percentage of the signal obtained from wild-type samples. The primers used for PCR were as follows:

  • Cx37-sense 5′-GGCTGGACCATGGAGCCGGT-3′;

  • Cx37-antisense 5′-TTTCGGCCACCCTGGGGAGC-3′;

  • Cx40-sense 5′-TTTGGCAAGTCACGGCAGGG-3′;

  • Cx40-antisense 5′-TGTCACTATGGTAGCCCTGAG-3′;

  • Cx43-sense 5′-TACCACGCCACCACCGGCCCA-3′;

  • Cx43-antisense 5′GGCATTTTGGCTGTCGTCAGGGAA-3′;

  • GAPDH-sense 5′TCACTCAAGATTGTCAGCAA-3′;

  • GAPDH-antisense 5′AGATCCACGACGGACACATT-3′

  • Smooth-muscle actin-sense 5′-GAATGGGCCAAAAAGACAGCTATG-3′

  • Smooth-muscle actin-antisense 5′-ATGGCATGAGGCAGGGCATA-3′

Dye-transfer experiments

Postnatal thoracic aortas were cut into three segments. Embryonic aortas were maintained as one piece. Segments were pinned out, stained briefly with 25 μM Hoechst 33342 (Molecular Probes, Eugene, OR) to visualize nuclei,then submerged in PBS containing 0.90 mM Ca2+, 0.49 mM Mg2+ and 1% BSA during microinjections. Two types of tracers were used: (1) a mixture of 5% biocytin (Mr 372, neutral charge) and 10 mg/ml dextran-fluorescein (Mr ∼3×103) (both from Molecular Probes); or (2) 5 mM[2-(4-nitro-2,1,3-benzoxadiazol-7-yl)aminoethyl]trimethylammonium(Mr 266, +1 charge), a fluorescent compound referred to as NBD-TMA(provided by Stephen Wright, University of Arizona, AZ)(Bednarczyk et al., 2000). Microelectrodes were positioned at the surface of an endothelial cell in the region of the nucleus. Tracer was injected for 10 seconds using the capacitance overcompensation feature of the amplifier. For NBD-TMA,dye-transfer was for one minute before counting labeled cells. Transfer of biocytin was for 15 minutes before the segments were fixed in 4%paraformaldehyde. Segments were washed and blocked in PBS containing 2% BSA and 0.25% TX100. Vessels were incubated in tetramethylrhodamine(TMR)-NeutrAvidin (Molecular Probes), washed, mounted on slides in PBS, and the number of biocytin-labeled cells counted.

Silver nitrate staining

Silver nitrate staining of aortic endothelial cells was performed essentially as described by McDonald et al.(McDonald, 1994). This technique stains material in the intercellular space, outlining endothelial cells and marking gaps between cells.

Survival study

Sixteen wild-type and 16 Cx37+/–Cx40–/– mice (both 129/Sv-C57BL/6 strain) were housed under identical conditions. Animals were checked daily for 2 years. The percentage of surviving animals was plotted at two-week intervals.

Statistics

Data were compared statistically using ANOVA (dye injections) or t-test(westerns and RT-PCR).

Results

Western blots and immunohistochemistry

Relative levels of Cx37 and Cx40 protein were compared in aortic endothelium of wild-type, Cx37–/– and Cx40–/– mice by western blotting with anti-connexin antibodies (Fig. 1A). To compare connexin levels specifically in the endothelium rather than whole aorta, we prepared endothelial lysates from thoracic aortas by passing a 20 mM NaOH solution through the lumen of the vessels. Extraction of plasma membranes with an alkaline solution partially enriches for gap-junction plaques(Hertzberg, 1984; Koval et al., 1995). Alkaline-resistant membranes were collected by ultracentrifugation and subsequently used for western blotting. To control for differences in endothelial cell content or sample loading, connexin levels were normalized to levels of PECAM, an endothelial-specific marker, measured on the same blots. Blots of wild-type aortic endothelium probed with anti-Cx37 or anti-Cx40 antibodies yielded a broad Cx37 band or a narrow Cx40 band, respectively,which migrated at the predicted molecular weights(Fig. 1A). As expected, the Cx37 band was absent on blots of Cx37–/– endothelium and the Cx40 band was absent on blots of Cx40–/–endothelium. The specificity of the anti-connexin antibodies was confirmed by immunoblotting samples of recombinant glutathione S-transferase (GST)-connexin fusion proteins (Fig. 1B). Cx43 was not detected on western blots of wild-type, Cx37–/–or Cx40–/– aortic endothelium, but was detected on blots of whole aorta or medial-layer-only blots(Fig. 1C, Fig. 2A,B). The absence of a Cx43 signal from endothelial samples indicates little if any contamination of endothelial lysates with medial layer connexins.

Fig. 1.

Deletion of Cx37 or Cx40 reduces the levels of non-ablated connexins in aortic endothelium. (A) Immunoblots of alkaline-extracted endothelial cell membranes isolated from thoracic aortas of 6-7-week-old and 4-6-month-old mice. Pooled membranes from six aortas were loaded in each lane. Densitometric quantification of all blots is presented in Table 1. Connexin levels were normalized to PECAM levels measured on the same blot. Cx37 protein was substantially reduced in aortic endothelium of Cx40–/–mice compared with wild-type levels. The drop in Cx37 was more pronounced in aortas from the younger mice (17-fold) than aortas from 4-6-month-old mice(3.0-fold). Cx40 was reduced in the endothelium of aortas from Cx37–/– mice, but the decrease was not as large(2.6-4.2-fold). VE-cadherin and PECAM levels were not significantly altered,but caveolin-1 levels were slightly reduced in Cx37–/–and Cx40–/– aortic endothelium of 3-7-week-old mice.(B) Anti-Cx37 and anti-Cx40 antibodies reacted only with appropriate GST-connexin fusion proteins. (C) Cx43 was not detected in western blots of alkaline-extracted endothelial cell membranes but was detected in a similar preparation of whole aorta. (D) Chemiluminescent detection of GST-Cx37 fusion protein was linear over a 16-fold dilution range.

Fig. 1.

Deletion of Cx37 or Cx40 reduces the levels of non-ablated connexins in aortic endothelium. (A) Immunoblots of alkaline-extracted endothelial cell membranes isolated from thoracic aortas of 6-7-week-old and 4-6-month-old mice. Pooled membranes from six aortas were loaded in each lane. Densitometric quantification of all blots is presented in Table 1. Connexin levels were normalized to PECAM levels measured on the same blot. Cx37 protein was substantially reduced in aortic endothelium of Cx40–/–mice compared with wild-type levels. The drop in Cx37 was more pronounced in aortas from the younger mice (17-fold) than aortas from 4-6-month-old mice(3.0-fold). Cx40 was reduced in the endothelium of aortas from Cx37–/– mice, but the decrease was not as large(2.6-4.2-fold). VE-cadherin and PECAM levels were not significantly altered,but caveolin-1 levels were slightly reduced in Cx37–/–and Cx40–/– aortic endothelium of 3-7-week-old mice.(B) Anti-Cx37 and anti-Cx40 antibodies reacted only with appropriate GST-connexin fusion proteins. (C) Cx43 was not detected in western blots of alkaline-extracted endothelial cell membranes but was detected in a similar preparation of whole aorta. (D) Chemiluminescent detection of GST-Cx37 fusion protein was linear over a 16-fold dilution range.

Fig. 2.

Cx37 and Cx43 levels are elevated in the medial layer of Cx40–/– aorta. (A) Alkaline-extracted membranes from whole aortas of 6-7-week-old mice were immunoblotted. Membranes from the equivalent of 1.5 aortas are loaded in each lane. Cx37 levels slightly increased (∼1.4-fold) in Cx40–/– aorta, after normalizing to PECAM. Cx43 levels increased ∼fourfold in Cx40–/– aorta. (B) SDS sample buffer lysates of aortic endothelium and alkaline-extracted membranes from medial-layer only preparations were immunoblotted (equivalent of 1.5 aortas per lane). Total endothelial Cx37 declined ∼8.2-fold in Cx40–/–aorta, whereas medial Cx37 increased ∼4.4-fold. Cx43 also increased about fourfold in the medial layer of Cx40–/– aorta. Cx40 was detected only in the endothelial fraction, as was PECAM. Cx43 was detected only in the medial layer preparation.

Fig. 2.

Cx37 and Cx43 levels are elevated in the medial layer of Cx40–/– aorta. (A) Alkaline-extracted membranes from whole aortas of 6-7-week-old mice were immunoblotted. Membranes from the equivalent of 1.5 aortas are loaded in each lane. Cx37 levels slightly increased (∼1.4-fold) in Cx40–/– aorta, after normalizing to PECAM. Cx43 levels increased ∼fourfold in Cx40–/– aorta. (B) SDS sample buffer lysates of aortic endothelium and alkaline-extracted membranes from medial-layer only preparations were immunoblotted (equivalent of 1.5 aortas per lane). Total endothelial Cx37 declined ∼8.2-fold in Cx40–/–aorta, whereas medial Cx37 increased ∼4.4-fold. Cx43 also increased about fourfold in the medial layer of Cx40–/– aorta. Cx40 was detected only in the endothelial fraction, as was PECAM. Cx43 was detected only in the medial layer preparation.

Surprisingly, a significant decrease in non-ablated connexins was observed in aortic endothelium of Cx37–/– and Cx40–/– mice (Fig. 1A) (Table 1). The biggest decline occurred in the levels of Cx37 present in Cx40–/– aorta, where a 17-fold drop was observed on blots of aortas from 3-7-week-old Cx40–/– mice. Likewise, a 4.2-fold decrease in Cx40 was shown in aortic endothelium from Cx37–/– mice of the same age. In contrast to the drop in connexin levels, VE-cadherin levels were not reduced on the same blots of Cx37–/– and Cx40–/– aorta, nor was there a change in PECAM levels (Fig. 1A) (Table 1). Thus, the low levels of non-ablated connexins in the knockout mice were not due to a generalized decrease in endothelial membrane proteins. We also probed the blots with an antibody against caveolin-1, a protein that is highly expressed by vascular endothelial cells and which was recently shown to interact with some connexins (Schubert et al., 2002) (Fig. 1A) (Table 1). Caveolin-1 levels were slightly decreased in both Cx37–/– and Cx40–/– samples from 3-7-week-old mice. To determine whether changes in connexin levels were persistent over time, we performed western blots on aortic endothelium from 4-8-month-old mice (Fig. 1A)(Table 1). In these animals,Cx37 protein was 3.0-fold lower in Cx40–/– aortic endothelium compared with wild-type controls, and Cx40 was 2.6-fold lower in Cx37–/– endothelium. VE-cadherin levels were not significantly altered in the knockout samples, although there was considerable variability in the VE cadherin measurements from aortas of older animals. Caveolin-1 and PECAM were not changed in aortic endothelium from 4-8-month-old Cx37–/– and Cx40–/– mice. These results indicate that alterations in the levels of non-ablated connexins in Cx37–/– and Cx40–/– aortic endothelium are long lasting, although in the case of Cx37, the reduction was more pronounced in younger animals than older animals.

Table 1.

Expression of Cx37, Cx40, VE cadherin, caveolin-1 and PECAM in wild-type(WT), Cx37–/– and Cx40–/– aortic endothelium (as a percentage of wildtype)

Mice aged 3-7 weeks
Mice aged 4-8 months
ProteinWT (n=6)Cx37-/- (n=6)Cx40-/- (n=6)WT (n=4)Cx37-/- (n=3)Cx40-/- (n=4)
Cx37 100 0* 5.8±0.8* 100 0* 33±10* 
Cx40 100 24±6* 0* 100 39±17* 0* 
VE-cadherin 100 106±8 83±12 100 175±54 288±133 
Caveolin-1 100 69±7* 59±9* 100 93±21 105±33 
PECAM 100 89±4 102±11 100 99±11 87±7 
Mice aged 3-7 weeks
Mice aged 4-8 months
ProteinWT (n=6)Cx37-/- (n=6)Cx40-/- (n=6)WT (n=4)Cx37-/- (n=3)Cx40-/- (n=4)
Cx37 100 0* 5.8±0.8* 100 0* 33±10* 
Cx40 100 24±6* 0* 100 39±17* 0* 
VE-cadherin 100 106±8 83±12 100 175±54 288±133 
Caveolin-1 100 69±7* 59±9* 100 93±21 105±33 
PECAM 100 89±4 102±11 100 99±11 87±7 

Numbers represent mean percent of wild-type protein level±s.e.m. Immunoblotted samples consisted of pooled, alkaline-extracted membrane preparations obtained from six thoracic aortas. Experiments were performed three to six times. Autoradiographs were quantified by densitometry. Values listed for Cx37, Cx40, VE-cadherin and caveolin-1 are normalized to PECAM levels measured on the same blot. Values listed for PECAM are the raw percentages obtained (not normalized).

*

P<0.05 versus respective wild-type values.

Western blots were also done to determine whether there were changes in Cx37 in whole aorta, or medial-layer-only fractions of aorta, from 6-7-week-old Cx40–/– mice(Fig. 2A,B). Instead of a decrease in Cx37, a blot of alkaline-extracted membranes from whole aorta showed a slight increase (∼1.4-fold) in Cx37 in Cx40–/– aorta, after normalizing to PECAM(Fig. 2A). On the same blot,Cx43 was elevated ∼fourfold in Cx40–/– aorta(Fig. 2A). The fact that endothelial Cx37 was diminished in aortic endothelium but not in whole aorta from Cx40–/– animals suggested that Cx37 might be elevated in the medial layer of Cx40–/– aortas. To test this idea, aortic endothelium was extracted with SDS sample buffer and the endothelium-free medial layer was homogenized. Alkaline-treated membranes were collected from the medial layer homogenate and analyzed by western blotting(Fig. 2B). Cx37, but not Cx40,was detected in the medial layer samples and the Cx37 signal increased 4.4-fold in the Cx40–/– versus wild-type media. The increase in medial Cx37 paralleled a 4.0-fold increase in Cx43 that was observed with the same samples. The SDS sample buffer lysates of endothelium were also analyzed and Cx37 was found to be reduced by 8.2-fold in Cx40–/– endothelium(Fig. 2B). Similar results for both endothelial and medial layer Cx37 were obtained using the alternative Alpha Diagnostics Cx37 antibody (not shown). Blots for Cx40, Cx43 and PECAM showed little, if any, contamination of endothelial and medial-layer fractions(Fig. 2B).

Immunofluorescent staining was performed to confirm the changes in connexin levels observed by western blotting. Frozen sections of thoracic aorta from 7-week-old wild-type, Cx37–/– and Cx40–/– mice were incubated with anti-connexin antibodies (Fig. 3A-P, Fig. 4A-D). Cx37 was predominantly detected in the endothelium of wild-type aorta. At high magnification, however, faint punctate signals in the medial layer could also be observed following Cx37 immunostaining(Fig. 4A,E). Medial layer staining was likewise detected on sections of 4-month-old wild-type aorta. Cx37 immunostaining has not previously been reported in the medial layer of mouse aorta, perhaps because of the relatively low signal in the media compared with the strong signal in endothelium. Therefore, to determine whether the faint fluorescent signals in the medial layer were genuinely due to the presence of Cx37 and not due to antibody cross reactivity, we compared Cx37 immunostaining in wild-type and Cx37–/– aorta(Fig. 4A,B). Both the strong endothelial staining and the weak medial staining were absent in Cx37–/– aorta sections, indicating that the medial layer staining was specific for Cx37. Next, we immunostained wild-type and knockout aorta sections to look for changes in non-ablated connexins(Fig. 3A-P, Fig. 4). Consistent with the western blot results, we observed a prominent drop in the endothelial signal of Cx37 in sections of Cx40–/– aorta. Only occasional punctae were observed in the endothelial cell layer of Cx40–/– aorta. By contrast, Cx37 appeared elevated in the medial layer of Cx40–/– aorta, as the faint Cx37 signal in this layer was easier to detect in the Cx40–/– sections than in wild-type sections(Fig. 3K, Fig. 4C). Aortas from 4-month-old Cx40–/– animals had particularly significant medial layer Cx37 immunostaining(Fig. 4F). Markedly reduced endothelial Cx37 in Cx40–/– aorta was confirmed by en face immunostaining of thoracic aorta from 7-week-old wild-type and Cx40–/– mice (Fig. 3Q-T). En face immunostaining of wild-type aorta resulted in fields where individual endothelial cells were partly or completely outlined by Cx37-containing punctae. By contrast, en face immunostaining of Cx40–/– aorta resulted in a very weak, patchy Cx37 signal. Overexposure of photographic images(Fig. 3R, inset), confirmed residual levels of Cx37 present in Cx40–/– endothelium. En face staining results for Cx37 were confirmed with the alternative Alpha Diagnostics Cx37 antibody (not shown). The presence of Cx40 and Cx43 in cryosections was also examined. Cx40 was detected exclusively in the endothelium of both wild-type and Cx37–/– aortic sections (Fig. 3E,G,M,O). Although Cx40 immunostaining persisted in the Cx37–/–endothelium, there was a significant decrease in the intensity of the signal,consistent with the reduction in Cx40 levels observed on western blots. Cx43 was not detected in the endothelial layer when aortic sections were immunostained with anti-Cx43 antibodies, but was detected in the media (not shown). Finally, we immunostained sections of abdominal aorta, iliac artery and coronary artery with anti-Cx37 and anti-Cx40 antibodies and obtained similar results (not shown). As in thoracic aorta, endothelial Cx37 signals were significantly reduced in all of the Cx40–/–arteries examined. Thus, the dependence of endothelial Cx37 levels on the presence of Cx40 may be a general feature of the arterial system.

Fig. 3.

Endothelial Cx37 immunostaining is markedly reduced in Cx40–/– aorta. Sections of thoracic aortas from E18.5 embryos (A-H) or 7-week-old mice (I-P) were immunostained for Cx37 or Cx40. Differential interference contrast images are shown to the right of each panel. In sections of aorta from E18.5 Cx40–/– embryos(C,D), endothelial Cx37 immunostaining was virtually absent, whereas endothelial Cx37 staining was readily detected in wild-type sections (A,B). Cx40 immunostaining in sections of E18.5 Cx37–/– aorta(G, H) was similar to that observed with sections of E18.5 wild-type aorta(E,F). Sections of aorta from a 7-week-old Cx40–/–animal showed a substantial reduction in endothelial Cx37 immunostaining (K,L)compared with wild-type (I,J). Only rarely were punctate signals observed in the endothelium. Cx40 immunostaining in sections of 7-week-old Cx37–/– aorta was slightly reduced (O,P) compared with wild-type signals (M,N). En face immunostaining of segments of 7-week-old aorta confirmed that endothelial Cx37 immunostaining was greatly reduced in Cx40–/– aortic segments (R) compared with wild-type(Q). For the inset in panel R, the exposure level was adjusted to bring out faint Cx37 immunostaining in the Cx40–/– endothelium. En face Cx40 staining was slightly reduced in Cx37–/–aortic segments (T) compared with wild-type (S). Bars, 20 μm.

Fig. 3.

Endothelial Cx37 immunostaining is markedly reduced in Cx40–/– aorta. Sections of thoracic aortas from E18.5 embryos (A-H) or 7-week-old mice (I-P) were immunostained for Cx37 or Cx40. Differential interference contrast images are shown to the right of each panel. In sections of aorta from E18.5 Cx40–/– embryos(C,D), endothelial Cx37 immunostaining was virtually absent, whereas endothelial Cx37 staining was readily detected in wild-type sections (A,B). Cx40 immunostaining in sections of E18.5 Cx37–/– aorta(G, H) was similar to that observed with sections of E18.5 wild-type aorta(E,F). Sections of aorta from a 7-week-old Cx40–/–animal showed a substantial reduction in endothelial Cx37 immunostaining (K,L)compared with wild-type (I,J). Only rarely were punctate signals observed in the endothelium. Cx40 immunostaining in sections of 7-week-old Cx37–/– aorta was slightly reduced (O,P) compared with wild-type signals (M,N). En face immunostaining of segments of 7-week-old aorta confirmed that endothelial Cx37 immunostaining was greatly reduced in Cx40–/– aortic segments (R) compared with wild-type(Q). For the inset in panel R, the exposure level was adjusted to bring out faint Cx37 immunostaining in the Cx40–/– endothelium. En face Cx40 staining was slightly reduced in Cx37–/–aortic segments (T) compared with wild-type (S). Bars, 20 μm.

Fig. 4.

Cx37 immunostaining is detectable in the medial layer of mouse aorta as well as in endothelium. Sections of thoracic aortas from 6-7-week-old wild-type (A), Cx37–/– (B) and Cx40–/– (C,D) mice were immunostained with anti-Cx37 antibody (A,B,C) or with secondary antibody only (D) and photographed at high magnification. The lumen of the aorta is on the right in each panel. (A) In wild-type aorta, very weak punctate staining was observed in the medial layer,as well as strong endothelial staining. (B) The specificity of the medial layer signal was confirmed by the absence of medial layer staining in Cx37–/– aortic sections. (C) Medial layer Cx37 staining was elevated in Cx40–/– sections. (D) Secondary antibody-only control showed no signal in the medial layer of Cx40–/– aorta. (E,F) Sections of wild-type and Cx40–/– aorta from 4-month-old mice were immunostained for Cx37. In addition to endothelial signal, faint punctate Cx37 staining was observed in the medial layer of wild-type sections. Medial layer Cx37 staining was particularly evident in aortas from 4-month-old Cx40–/– mice. Bar, 20 μm.

Fig. 4.

Cx37 immunostaining is detectable in the medial layer of mouse aorta as well as in endothelium. Sections of thoracic aortas from 6-7-week-old wild-type (A), Cx37–/– (B) and Cx40–/– (C,D) mice were immunostained with anti-Cx37 antibody (A,B,C) or with secondary antibody only (D) and photographed at high magnification. The lumen of the aorta is on the right in each panel. (A) In wild-type aorta, very weak punctate staining was observed in the medial layer,as well as strong endothelial staining. (B) The specificity of the medial layer signal was confirmed by the absence of medial layer staining in Cx37–/– aortic sections. (C) Medial layer Cx37 staining was elevated in Cx40–/– sections. (D) Secondary antibody-only control showed no signal in the medial layer of Cx40–/– aorta. (E,F) Sections of wild-type and Cx40–/– aorta from 4-month-old mice were immunostained for Cx37. In addition to endothelial signal, faint punctate Cx37 staining was observed in the medial layer of wild-type sections. Medial layer Cx37 staining was particularly evident in aortas from 4-month-old Cx40–/– mice. Bar, 20 μm.

Connexin immunostaining was also done with cryosections of E18.5 aortas(Fig. 3A-H). We hypothesized that Cx37 levels might be particularly low in embryonic Cx40–/– aortas, given that western blots had suggested that the effect of eliminating Cx40 on Cx37 levels in aortic endothelium was age dependent, with younger postnatal animals showing greater declines in Cx37. Cx37 was in fact very difficult to detect in E18.5 Cx40–/– aortic endothelium, although it was readily detected in E18.5 wild-type endothelium. By contrast, Cx40 immunostaining looked identical in Cx37–/– and wild-type endothelium at this stage.

RT-PCR analysis

RT-PCR was performed to determine whether decreases in non-ablated connexin levels were due to a decline in connexin mRNA levels(Fig. 5). A lysate was collected from thoracic aorta endothelium of 6-7-week-old mice by briefly passing a lysis solution through the vessel lumen, and total RNA was purified. After reverse transcription, specific primers were used to amplify segments of Cx37, Cx40, Cx43 and GAPDH cDNAs. Amplicons were of the expected size and depended on reverse transcription (Fig. 5A). Cx37 and Cx40 amplicons were absent from Cx37–/– and Cx40–/– samples,respectively, confirming the specificity of the PCR amplifications(Fig. 5A). Semi-quantitative RT-PCR was performed to compare the levels of non-ablated connexin transcripts in wild-type, Cx37–/– and Cx40–/– aortic endothelium(Fig. 5B-D). Cx37 mRNA levels were not significantly different in Cx40–/– versus wild-type aortic endothelium, nor was there a change in Cx40 mRNA in Cx37–/– versus wild-type aorta(Fig. 5D). Amplification of Cx43 signal from endothelial samples was weaker than for Cx37 and Cx40 and more variable (Fig. 5B,D). Differences in Cx43 mRNA levels (in Cx37–/–endothelium, for example) did not reach statistical significance. Because Cx43 protein was not detected in aortic endothelium by western blotting or immunostaining, Cx43 signals in the RT-PCR analysis may be the result of unintended extraction of leukocytes during the RNA isolation procedure. We determined that there was minimal contamination of our endothelial RNA preparations with medial layer RNA by performing RT-PCR with primers for smooth-muscle actin (SMA), comparing endothelial RNA fractions with whole aorta RNA. Although SMA RNA was detected in the endothelial samples, the level of SMA RNA in the endothelial samples was ∼500-fold lower than in an equivalent length of whole aorta. Thus, contamination of endothelial RNA with medial layer RNA was only ∼0.2%.

Fig. 5.

Deletion of Cx37 or Cx40 does not alter mRNA levels of non-ablated connexins in aortic endothelium. (A) RT-PCR for Cx37, Cx40, Cx43 and GAPDH mRNA was performed with RNA isolated from aortic endothelium. Amplicons were of the predicted size and depended on reverse transcription (RT). Cx37 and Cx40 amplicons were absent in samples derived from Cx37–/– or Cx40–/– mice,respectively. (B) Semiquantitative RT-PCR was performed. Connexin signals were normalized to GAPDH signals. (C) Quantification of Cx37 signals obtained from groups of wild-type and Cx40–/– animals yielded similar amplification curves. (D) Mean Cx37, Cx40 and Cx43 mRNA levels (relative to wild-type levels) are plotted. Wild-type signals for each connexin mRNA were assigned the value of 1.0. Error bars represent s.e.m. There were no significant changes in the levels of non-ablated connexin mRNAs present in Cx37–/– and Cx40–/– aortic endothelium (P>0.05). (E) Endothelial or whole aorta RNA preparations (equivalent lengths of aorta) from wild-type mice were analyzed by RT-PCR for smooth-muscle actin mRNA to test for contamination of endothelial RNA fractions with medial layer RNA. A 1/500 dilution of whole aorta cDNA yielded a similar amplification curve to that of undiluted endothelial cDNA. Contamination was therefore approximately only 0.2%. Cx40 RNA levels were the same in each preparation, whereas Cx37 RNA levels were slightly higher in the whole aorta sample.

Fig. 5.

Deletion of Cx37 or Cx40 does not alter mRNA levels of non-ablated connexins in aortic endothelium. (A) RT-PCR for Cx37, Cx40, Cx43 and GAPDH mRNA was performed with RNA isolated from aortic endothelium. Amplicons were of the predicted size and depended on reverse transcription (RT). Cx37 and Cx40 amplicons were absent in samples derived from Cx37–/– or Cx40–/– mice,respectively. (B) Semiquantitative RT-PCR was performed. Connexin signals were normalized to GAPDH signals. (C) Quantification of Cx37 signals obtained from groups of wild-type and Cx40–/– animals yielded similar amplification curves. (D) Mean Cx37, Cx40 and Cx43 mRNA levels (relative to wild-type levels) are plotted. Wild-type signals for each connexin mRNA were assigned the value of 1.0. Error bars represent s.e.m. There were no significant changes in the levels of non-ablated connexin mRNAs present in Cx37–/– and Cx40–/– aortic endothelium (P>0.05). (E) Endothelial or whole aorta RNA preparations (equivalent lengths of aorta) from wild-type mice were analyzed by RT-PCR for smooth-muscle actin mRNA to test for contamination of endothelial RNA fractions with medial layer RNA. A 1/500 dilution of whole aorta cDNA yielded a similar amplification curve to that of undiluted endothelial cDNA. Contamination was therefore approximately only 0.2%. Cx40 RNA levels were the same in each preparation, whereas Cx37 RNA levels were slightly higher in the whole aorta sample.

Dye-transfer experiments

Dye injections were performed to examine the effects of connexin ablation on interendothelial communication mediated by gap junctions. Segments of thoracic aorta from wild-type, Cx37–/–,Cx40–/– and Cx37+/–Cx40–/– mice were removed, and endothelial cells were injected with a mixture of a gap-junction-permeable tracer (Biocytin or NBD-TMA) and a fluorescently labeled dextran. The fluorescent dextran, which is impermeable through gap junctions, was used to mark the injected cell and as a control for transfer that was not mediated by gap junctions. Biocytin transfer to neighboring cells was detected with TMR-Neutravidin (Fig. 6),whereas NBD-TMA could be detected directly because it is a fluorescent compound (Fig. 7). After the transfer period, the number of endothelial cells containing tracer was counted and compared between genotypes. Biocytin transfer experiments were done with aortas from mice aged 3 weeks, 6-7 weeks or ≥8-weeks to determine whether there were age-related differences in the effect of connexin ablation on intercellular communication. Transfer of biocytin occurred extensively in wild-type endothelium (Fig. 6A). In aortas from 6-7-week-old mice, for example, the mean number of labeled endothelial cells was 427±70 cells(Fig. 10A)(Table 2). The absence of Cx37 resulted in a small reduction (36%) in biocytin transfer at 3 weeks, a greater reduction (78%) at 6-7 weeks and no significant reduction at ≥8 weeks(Fig. 6C, Fig. 10A)(Table 2). Ablation of Cx40 resulted in a generally greater reduction in biocytin transfer compared with Cx37 elimination, with the exception of aortas from 6-7-week-old mice, where Cx40–/– and Cx37–/– aortas exhibited about the same amount of transfer. Biocytin transfer was reduced by 73%, 72% and 62% in aortas from mice aged 3 weeks, 6-7 weeks and ≥8 weeks,respectively, compared with wild-type mice(Fig. 6E, Fig. 10A)(Table 2). Aortas from Cx37+/–Cx40–/– mice showed a further reduction in biocytin transfer, which was reduced by 85%, 88% and 83% in aortas from Cx37+/–Cx40–/– mice aged 3 weeks, 6-7 weeks and ≥8-weeks mice, respectively, compared with wild-type. NBD-TMA transfer was determined in aortas from animals that were ≥8 weeks old (Fig. 7). No significant change in NBD-TMA transfer was observed in Cx37–/–versus wild-type aortas at this age (Fig. 7) (Table 2). The absence of Cx40, however, reduced NBD-TMA transfer by 51%.

Fig. 6.

Biocytin transfer is reduced in aortic endothelium of connexin-deficient mice. Biocytin and FITC-dextran were injected into endothelial cells of wild-type, Cx37–/–, Cx40–/– or Cx37+/–Cx40–/– thoracic aorta isolated from 6-7-week-old animals. Biocytin transfer was reduced in Cx37–/– (C) and Cx40–/– (E)endothelium compared with wild-type endothelium (A). Biocytin-containing cells in panel A extended beyond the photographic field of view. Transfer was further reduced in Cx37+/–Cx40–/–endothelium (G). FITC-dextran marked the injected cell, but did not transfer to adjacent cells. There was more autofluorescence in panels (B,F) than (D,H)because the FITC-dextran was photographed after fixing the tissue. Quantification is presented in Fig. 10 and Table 2. Bar, 50 μm.

Fig. 6.

Biocytin transfer is reduced in aortic endothelium of connexin-deficient mice. Biocytin and FITC-dextran were injected into endothelial cells of wild-type, Cx37–/–, Cx40–/– or Cx37+/–Cx40–/– thoracic aorta isolated from 6-7-week-old animals. Biocytin transfer was reduced in Cx37–/– (C) and Cx40–/– (E)endothelium compared with wild-type endothelium (A). Biocytin-containing cells in panel A extended beyond the photographic field of view. Transfer was further reduced in Cx37+/–Cx40–/–endothelium (G). FITC-dextran marked the injected cell, but did not transfer to adjacent cells. There was more autofluorescence in panels (B,F) than (D,H)because the FITC-dextran was photographed after fixing the tissue. Quantification is presented in Fig. 10 and Table 2. Bar, 50 μm.

Fig. 7.

NBD-TMA transfer is reduced in aortic endothelium of connexin-deficient mice. The fluorescent tracer NBD-TMA was injected into endothelial cells of wild-type (A), Cx37–/– (B),Cx40–/– (C) or Cx37+/–Cx40–/– (D) thoracic aorta from 8-12-week-old animals. NBD-TMA transfer was reduced in Cx40–/– endothelium compared with wild-type, but not in Cx37–/– endothelium. NBD-TMA transfer was further reduced in Cx37+/–Cx40–/– endothelium. Injected cells are marked with an asterisk. Quantification is presented in Table 2. Bar, 50 μm.

Fig. 7.

NBD-TMA transfer is reduced in aortic endothelium of connexin-deficient mice. The fluorescent tracer NBD-TMA was injected into endothelial cells of wild-type (A), Cx37–/– (B),Cx40–/– (C) or Cx37+/–Cx40–/– (D) thoracic aorta from 8-12-week-old animals. NBD-TMA transfer was reduced in Cx40–/– endothelium compared with wild-type, but not in Cx37–/– endothelium. NBD-TMA transfer was further reduced in Cx37+/–Cx40–/– endothelium. Injected cells are marked with an asterisk. Quantification is presented in Table 2. Bar, 50 μm.

Fig. 10.

Quantification of biocytin transfer in postnatal and E18.5 aortic endothelium. The mean number of biocytin-labeled cells is plotted for postnatal animals (A) or E18.5 embryos (B). In A, the bar for >8-week-old wild-type mice represents pooled data from C57BL/6 and C57BL/6-129/Sv strains,which were not significantly different. Error bars represent s.e.m.*P<0.001 and #P<0.05 versus respective wild-type values.

Fig. 10.

Quantification of biocytin transfer in postnatal and E18.5 aortic endothelium. The mean number of biocytin-labeled cells is plotted for postnatal animals (A) or E18.5 embryos (B). In A, the bar for >8-week-old wild-type mice represents pooled data from C57BL/6 and C57BL/6-129/Sv strains,which were not significantly different. Error bars represent s.e.m.*P<0.001 and #P<0.05 versus respective wild-type values.

Table 2.

Amount of dye-transfer in aortic endothelium of wild-type,Cx37–/–, Cx40–/–,Cx37+/–Cx40–/– and Cx37–/–Cx40–/– mice of various ages (in mean number of labeled cells)

Age groupTracerWT (C57BL/129Sv)WT (C57BL)Cx37-/- (C57BL)Cx40-/- (C57BL)Cx37+/-Cx40-/- (C57BL/129Sv)Cx37-/-Cx40-/- (C57BL/129Sv)
E18.5 Biocytin 275±62 (n=9) 157±35 (n=8) 177±72 (n=7) 5.8±1.0* (n=20) 5.4±1.6* (n=11) 1.1±0.1* (n=7) 
3 weeks Biocytin nd 233±44 (n=16) 150±17** (n=20) 62±8.2* (n=16) 36±4.0* (n=20) na 
6-7 weeks Biocytin nd 427±70 (n=15) 94±8.3* (n=32) 121±18* (n=13) 50±6.2* (n=20) na 
≥8 weeks (mean age 18 weeks) Biocytin 347±73 (n=8) 345±79 (n=10) 259±33 (n=20) 132±18* (n=14) 60±6.5* (n=20) na 
≥8 weeks (mean age 8.3 weeks) NBD-TMA nd 88±7.1 (n=21) 82±4.5 (n=38) 43±3.0* (n=40) 35±1.8* (n=26) na 
Age groupTracerWT (C57BL/129Sv)WT (C57BL)Cx37-/- (C57BL)Cx40-/- (C57BL)Cx37+/-Cx40-/- (C57BL/129Sv)Cx37-/-Cx40-/- (C57BL/129Sv)
E18.5 Biocytin 275±62 (n=9) 157±35 (n=8) 177±72 (n=7) 5.8±1.0* (n=20) 5.4±1.6* (n=11) 1.1±0.1* (n=7) 
3 weeks Biocytin nd 233±44 (n=16) 150±17** (n=20) 62±8.2* (n=16) 36±4.0* (n=20) na 
6-7 weeks Biocytin nd 427±70 (n=15) 94±8.3* (n=32) 121±18* (n=13) 50±6.2* (n=20) na 
≥8 weeks (mean age 18 weeks) Biocytin 347±73 (n=8) 345±79 (n=10) 259±33 (n=20) 132±18* (n=14) 60±6.5* (n=20) na 
≥8 weeks (mean age 8.3 weeks) NBD-TMA nd 88±7.1 (n=21) 82±4.5 (n=38) 43±3.0* (n=40) 35±1.8* (n=26) na 

The table lists the mean number of labeled endothelial cells±s.e.m. following injection of a single cell with biocytin or NBD-TMA and a transfer period. The number of injections done for each group is listed in parentheses. For E18.5 experiments, each injection represents a different aorta. For postnatal injections, 4-8 aortas were used for each experimental group. Mouse strain backgrounds are listed in parentheses below the genotype.*P<0.001 and **P<0.05 versus respective wild-type values. nd, not determined. na, not applicable, because animals do not survive past the first postnatal day.

Dye-transfer measurements were also performed with aortic endothelium that lacked both Cx37 and Cx40. Cx37–/–Cx40–/– mice die perinatally, so dye injections were done on aortas from E18.5 embryos(Simon and McWhorter, 2002). We confirmed that Cx37–/–Cx40–/–embryos lacked both Cx37 and Cx40 in aortic endothelium(Fig. 8A-D). Compensatory changes in Cx43 expression were not observed in the double-knockout aortas(Fig. 8E,F). Wild-type aortas(C57BL6/129Sv strain) exhibited biocytin transfer to 275±62 cells at this developmental stage (Fig. 9A, Fig. 10B)(Table 2). Interendothelial transfer of biocytin was eliminated, however, in aortas from Cx37–/–Cx40–/– E18.5 embryos of the same strain background (Fig. 9D). Thus, only the injected cell was labeled with biocytin in Cx37–/–Cx40–/– aortas. The absence of dye-transfer was not due to gross defects in aortic development, as Cx37–/–Cx40–/– aortas showed normal expression of both endothelial and smooth-muscle-cell markers(Fig. 8G-J). In addition,silver nitrate staining of Cx37–/–Cx40–/– E18.5 aortas revealed that endothelial cell morphology and cell-cell contacts were normal(Fig. 8K,L). Finally, we examined the effects of ablating only one of the endothelial connexins from E18.5 aorta (Fig. 9, Fig. 10B)(Table 2). The absence of only Cx37 did not have any effect on biocytin transfer at this embryonic stage(Fig. 9B). Eliminating only Cx40, however, did have a striking effect on the transfer of biocytin(Fig. 9C). The number of labeled cells was reduced by 96% and 98% in Cx40–/– and Cx37+/–Cx40–/– E18.5 aortas,respectively. The effect of Cx40 ablation on dye-transfer was therefore much more pronounced in the embryonic aorta than in the postnatal aorta.

Fig. 8.

Cx37–/–Cx40–/– E18.5 aortas do not show compensatory expression of Cx43 and exhibit normal expression of blood vessel cell markers. Sections of E18.5 aorta from wild-type (A,C,E) and Cx37–/–Cx40–/– (B,D,F) aorta were immunostained for Cx37, Cx40 and Cx43. (B,D) Cx37 and Cx40 were not detected in aortas with the Cx37–/–Cx40–/– genotype. (F)Cx37–/–Cx40–/– aortas did not show compensatory expression of Cx43 in endothelium. Cx37–/–Cx40–/– aortas exhibited normal expression of the endothelial marker, PECAM-1 (G,H) and smooth-muscle actin (SMA) (I,J). (K,L) Silver nitrate staining, which was performed to examine endothelial cell morphology and cell-cell contacts, was similar in Cx37–/–Cx40–/– (L) and wild-type(K) aorta. Bar, 50 μm for A-J, 20 μm for K,L.

Fig. 8.

Cx37–/–Cx40–/– E18.5 aortas do not show compensatory expression of Cx43 and exhibit normal expression of blood vessel cell markers. Sections of E18.5 aorta from wild-type (A,C,E) and Cx37–/–Cx40–/– (B,D,F) aorta were immunostained for Cx37, Cx40 and Cx43. (B,D) Cx37 and Cx40 were not detected in aortas with the Cx37–/–Cx40–/– genotype. (F)Cx37–/–Cx40–/– aortas did not show compensatory expression of Cx43 in endothelium. Cx37–/–Cx40–/– aortas exhibited normal expression of the endothelial marker, PECAM-1 (G,H) and smooth-muscle actin (SMA) (I,J). (K,L) Silver nitrate staining, which was performed to examine endothelial cell morphology and cell-cell contacts, was similar in Cx37–/–Cx40–/– (L) and wild-type(K) aorta. Bar, 50 μm for A-J, 20 μm for K,L.

Fig. 9.

Biocytin transfer is eliminated in Cx37–/–Cx40–/– aortic endothelium from E18.5 embryos and is sharply reduced in E18.5 Cx40–/– endothelium. Biocytin was injected into endothelial cells of thoracic aortas from wild-type (A),Cx37–/– (B), Cx40–/– (C) or Cx37–/–Cx40–/– (D) E18.5 embryos. Biocytin transfer occurred equally well in Cx37–/– and wild-type aortic endothelium. Dye-transfer was reduced by 96% in Cx40–/– aortas and was eliminated in Cx37–/–Cx40–/– endothelium. Quantification is presented in Fig. 10 and Table 2. Bar, 50 μm.

Fig. 9.

Biocytin transfer is eliminated in Cx37–/–Cx40–/– aortic endothelium from E18.5 embryos and is sharply reduced in E18.5 Cx40–/– endothelium. Biocytin was injected into endothelial cells of thoracic aortas from wild-type (A),Cx37–/– (B), Cx40–/– (C) or Cx37–/–Cx40–/– (D) E18.5 embryos. Biocytin transfer occurred equally well in Cx37–/– and wild-type aortic endothelium. Dye-transfer was reduced by 96% in Cx40–/– aortas and was eliminated in Cx37–/–Cx40–/– endothelium. Quantification is presented in Fig. 10 and Table 2. Bar, 50 μm.

Long-term survival of Cx37+/–Cx40–/– versus wild-type mice

The long-term effects of a deficiency in endothelial communication have not been previously investigated. To address this issue, the survivability of Cx37+/–Cx40–/– versus wild-type mice was compared over a two-year period (Fig. 11). We chose to examine Cx37+/–Cx40–/– mice because aortas from these animals exhibit lower levels of interendothelial dye-transfer than either Cx37–/– or Cx40–/–aortas. Furthermore, dye-transfer is persistently reduced in Cx37+/–Cx40–/– animals, even in older ones. Cohorts of Cx37+/–Cx40–/– and wild-type animals were housed under identical conditions for up to two years. Although Cx37+/–Cx40–/– animals did not show a sharp drop-off in viability, they began dying earlier than the wild-type mice (Fig. 11). Furthermore, after two years, only 13% of the Cx37+/–Cx40–/– animals were alive,compared with 42% of the wild-type animals. These data suggest that a deficiency in endothelial communication may have long-term consequences on vascular health and longevity.

Fig. 11.

Survival curves for Cx37+/–Cx40–/–and wild-type mice. Cx37+/–Cx40–/– and wild-type mice were housed for two years and the percentage of surviving animals was plotted every two weeks. Cx37+/–Cx40–/– animals began dying earlier than wild-type mice and showed a decrease in survival after two years(13% versus 42%).

Fig. 11.

Survival curves for Cx37+/–Cx40–/–and wild-type mice. Cx37+/–Cx40–/– and wild-type mice were housed for two years and the percentage of surviving animals was plotted every two weeks. Cx37+/–Cx40–/– animals began dying earlier than wild-type mice and showed a decrease in survival after two years(13% versus 42%).

Discussion

To gain insight into the role of gap junctions in the vascular wall, we examined the effects of ablating specific vascular connexins on interendothelial communication. In addition, we determined how deletion of a specific connexin altered the levels of non-ablated connexin proteins. Our results indicate that deletion of either Cx37 or Cx40 reduces but does not eliminate interendothelial dye-transfer in the aorta, whereas elimination of both connexins abolishes transfer. Thus, Cx37 and Cx40 are likely to be the only connexins that are functionally expressed in mouse aortic endothelium. Consistent with these findings, we did not detect expression of Cx43 protein in endothelium of wild-type mouse aorta, nor did we see compensatory expression of Cx43 in aortas from connexin knockout mice. Krüger et al. also did not detect Cx43 in adult mouse aortic endothelium(Krüger et al., 2002). The absence of Cx43 from mouse arterial endothelium differs substantially from results obtained from rat and bovine studies(Yeh et al., 1997; Gabriels and Paul, 1998; Yeh et al., 1998; van Kempen and Jongsma, 1999). Recent studies have shown that Cx45 is also not expressed significantly in aortic endothelium of postnatal mice, although it is expressed by vascular smooth-muscle cells (Krüger et al.,2000). Taken together, these results indicate that Cx37 and Cx40 are the crucial players involved in interendothelial communication in mouse aorta.

The consequences of a deficiency in Cx37 versus Cx40 were substantially different with regard to interendothelial dye-transfer. Ablation of Cx40 generally had a greater effect on dye-transfer than ablation of Cx37, except in the 6-7 week-old age group, where the decline in transfer was about equivalent in Cx37–/– and Cx40–/– endothelium. In the other age groups,Cx40–/– aortic endothelium exhibited significantly less dye-transfer than Cx37–/– endothelium. Possibly, Cx40 is more abundant than Cx37 in the endothelium, but we did not attempt to compare the relative abundance of Cx37 and Cx40. Surprisingly, the effect of Cx40 deficiency on dye-transfer was age-dependent. In postnatal Cx40–/– animals, dye-transfer was reduced to a greater extent in younger versus older animals. Continuing this trend, in Cx40–/– aortas from E18.5 embryos, dye-transfer was drastically reduced in the absence of Cx40, with transfer occurring to only a few cells. By contrast, the elimination of Cx37 did not have any effect on biocytin dye-transfer in the embryonic endothelium. Thus, the effects on interendothelial communication of a deficiency in either Cx37 or Cx40 are not identical, and vary with developmental stage. Compensatory mechanisms may come into play in older postnatal animals to partially restore communication to more normal levels.

An important finding of this study is that targeted ablation of a specific endothelial connexin not only eliminates the targeted connexin, it also results in a substantial decrease in the levels of the non-ablated connexin. Thus, Cx37 and Cx40 are mutually dependent on each other for optimal expression in vascular endothelium. In particular, ablation of Cx40 resulted in up to a 17-fold drop in Cx37 protein in aortic endothelium as determined by western blotting. The drop in Cx37 was observed both with alkaline-extracted membrane fractions, which enrich for gap-junction plaques, and with SDS sample buffer lysates of endothelium. Thus, both total endothelial Cx37 and plaque-associated Cx37 are affected by the absence of Cx40. In the reciprocal case, ablation of Cx37 resulted in a decline in Cx40, but to a lesser extent;elimination of Cx37 resulted in up to a 4.2-fold drop in Cx40 protein in aortic endothelium. The dependency of Cx37 levels on the presence of Cx40 was age-dependent, with greater reductions in Cx37 occurring in aortas from younger versus older animals. This finding also suggests that age-dependent compensatory mechanisms may be a factor in partially restoring communication in older animals. Interestingly, age-dependent changes in endothelial connexin expression and colocalization have been reported for rat aorta(Yeh et al., 2000). Expression of endothelial Cx37, for example, showed a downward trend with age, and the extent of colocalization of Cx37 and Cx40 declined with age(Yeh et al., 2000). Therefore,age-dependent declines in non-ablated connexins could be related, in part, to changes in connexin expression in wild-type endothelium that occur with age. Finally, we also observed a small decrease in endothelial caveolin-1 levels in younger connexion-knockout mice. Recent evidence indicates that caveolin-1 can interact with Cx43, as well as with Cx26, Cx32 and Cx46 in cultured cells(Schubert et al., 2002). Possibly, caveolin-1 interacts with Cx37 and Cx40 in aortic endothelial cells and its synthesis, targeting or stability could be affected by connexin ablation.

Immunohistochemistry confirmed that decreases in non-ablated connexin levels occurred specifically in the endothelium of Cx37–/– or Cx40–/– aorta. Cx37 immunostaining was markedly reduced in the aortic endothelium of 7-week-old Cx40–/– mice compared with wild-type controls. In addition, we observed a dramatic reduction in Cx37 immunostaining in aortas from E18.5 Cx40–/– mice. This last result provides a molecular explanation for the sharp reduction in interendothelial dye-transfer that occurs in E18.5 endothelium in the absence of Cx40, as Cx37 levels were quite low in embryonic Cx40–/– endothelium. In postnatal Cx40–/– animals, especially older animals,endothelial Cx37 is reduced to a lesser extent than in E18.5 animals, relative to wild-type, and consequently, the effect on interendothelial communication is not as great.

Dependency of connexin levels on the presence of another connexin family member is not unique to endothelial cells. Targeted ablation of connexin32 in mice significantly reduced the levels of connexin26 in the liver, where they are normally co-expressed in hepatocytes, and also reduced total gap junctional plaque area in the liver by 1000-fold(Nelles et al., 1996). In another recent example, deletion of connexin46 (Cx46) eliminated the expression of connexin50 (Cx50) from the lens nucleus, where Cx46 and Cx50 are normally co-expressed by lens fibers (Rong et al., 2002). Interestingly, ablation of Cx50 did not eliminate Cx46 in the lens nucleus. This result is similar to what we observed with Cx37 and Cx40 in endothelial cells, with Cx37 being much more dependent on the presence of Cx40 than vice versa. Finally, in Cx43-deficient hearts, Cx45 immunostaining at cardiac gap junctions was markedly reduced, although total levels of Cx45 were unchanged (Johnson et al., 2002). Thus, in responses to connexin deletions, decreases in co-expressed, non-ablated connexins seem to be a more common occurrence than compensatory upregulation. These findings may be of some importance with regard to understanding the pathology of connexin mutations in humans,especially in tissues that express multiple connexins.

Changes in the levels of non-ablated connexins in the endothelium could be due either to alterations in the efficiency of translation of connexin mRNA or connexin processing, or to changes in connexin stability. Our RT-PCR experiments indicated no significant changes in Cx37 mRNA levels in Cx40–/– aortic endothelium, nor were there changes in Cx40 mRNA in Cx37–/– endothelium. Therefore, it is unlikely that the reduction in non-ablated connexins is due to decreased transcription of the connexin genes. Possibly, Cx37 and Cx40 contribute to heteromeric channels of mixed connexin content, which might be more stable than homomeric channels. According to this model, eliminating one connexin would permit the assembly of only less-stable homomeric channels, and therefore levels of the non-ablated connexin would fall. Heteromeric channels containing two connexins have been shown for several connexins by both biochemical and electrophysiological methods(Stauffer, 1995; Jiang and Goodenough, 1996; Brink et al., 1997; He et al., 1999). In addition,Cx37 and Cx40 have been shown to colocalize to the same gap junction plaques in endothelial cells (Yeh et al.,1998; Ko et al.,1999). Finally, the presence of both Cx37 and Cx40 could be important for normal clustering of gap-junction channels into plaques in the endothelial plasma membrane. Inefficient clustering in the absence of Cx40,for example, might reduce the stability of Cx37-containing channels. Recently,it was suggested that Cx40 could contribute to plaque structure by interacting with proteins linked to the cytoskeleton(Krüger et al.,2002).

Our results are substantially different from those published recently by Krüger et al., who investigated an independently generated Cx40–/– mouse line(Krüger et al., 2002). Those investigators concluded that there was upregulation of Cx37 in the aortic endothelium of Cx40–/– mice rather than the downregulation of endothelial Cx37 we describe here. They presented western blot data that documented a ∼fourfold increase in Cx37 protein in Cx40–/– thoracic aorta, as well as a ∼twofold increase in Cx43. The western blot data obtained by Krüger et al.,however, might not accurately represent Cx37 levels in the endothelium, as they collected protein from the entire aortic segment, including the medial layer. In this study, we analyzed separately connexins from whole aorta,endothelium-only fractions or medial layer-only preparations on western blots. This methodological difference is relevant because we detected some Cx37 in the medial layer by western blotting and by immunostaining. Cx37 has previously been reported in the media of rat aorta, and most recently in the caudal artery of the rat, where it was found to be more abundant than Cx43 or Cx45 (Nakamura et al., 1999; Rummery et al., 2002). The amount of Cx37 reported in the medial layer varies with different arteries,however, and appears to be lower in rat thoracic aorta than caudal artery(Rummery et al., 2002). Final confirmation of the presence of Cx37 at medial smooth-muscle gap junctions awaits careful analysis by immunogold electron microscopy. In this study, we found that medial layer Cx37 signal was significantly elevated in Cx40–/– aortas as detected by both western blotting and immunostaining. In this regard, Cx37 in the medial layer behaves similarly to Cx43, which we find is also elevated in Cx40–/– media,in agreement with Krüger et al.(Krüger et al., 2002). Possibly, medial Cx37 and Cx43 both increase in response to hypertensive changes described in Cx40–/– animals(de Wit et al., 2000). In some experimental models of hypertension in the rat, expression of aortic Cx43 was shown to increase by ∼twofold, similar to the ∼fourfold increase in medial Cx43 and medial Cx37 we observed in Cx40–/–aorta (Haefliger et al.,1997). The specificity of the medial layer Cx37 immunosignal was confirmed by immunostaining Cx37–/– aorta, which showed no medial layer staining. Therefore, the western blot results of Krüger et al. were possibly affected by changes in the levels of Cx37 present in the medial layer as well as in the endothelial layer. Indeed, when whole aorta preparations were analyzed by western blotting, we did not see the decline in Cx37 in Cx40–/– aortas that was observed with endothelium-only preparations.

Our immunostaining results also differ from those presented by Krüger et al., who reported that Cx37 immunosignals were distributed more homogeneously in Cx40-deficient versus wild-type endothelium, so that areas without Cx37 staining were less frequently observed in the Cx40–/– aorta(Krüger et al., 2002). By contrast, we observed a very substantial decrease in Cx37 immunostaining in Cx40–/– endothelium, both in cross sections and by en face staining. These results are fully consistent with our western blot data. The reasons for the different immunostaining data obtained in the study by Krüger et al. and in this study are not clear. Differences in mouse strains and ages or antibodies used in the studies might account for some differences. In addition to the western blot and immunofluorescence data, our RT-PCR data do not support the idea that there is an increase in endothelial Cx37 gene expression when Cx40 is ablated. Finally, Krüger et al. found very few gap junction plaques in the intima of Cx40–/–mice when aortic sections were analyzed by electron microscopy(Krüger et al., 2002). They suggested that aggregates of Cx37 might be too small to be seen in the electron micrographs, but could be detected by immunofluorescence. Alternatively, the scarcity of morphologically identifiable gap junctions in the Cx40–/– endothelium could be taken to support our findings of a substantial reduction in endothelial Cx37 levels in the absence of Cx40.

One might expect Cx40–/– endothelium to exhibit diminished coupling, especially since ablation of Cx40 also reduces endothelial Cx37. In the present study, we found that biocytin transfer was substantially lower in Cx40–/– endothelium compared with wild-type endothelium. Krüger et al. also compared dye-transfer levels in Cx40–/– versus wild-type aortic endothelium(Krüger et al., 2002). They reported extensive transfer of neurobiotin but not of Lucifer Yellow in Cx40–/– endothelium, whereas transfer of both tracers occurred efficiently in wild-type endothelium. On the basis of these results,they suggested that Cx37 channels have more restricted gating properties than Cx40 channels. Previous studies have indicated that negatively charged tracers diffuse poorly through Cx37 channels in some cell types, perhaps explaining weak Lucifer Yellow transfer in Cx40–/– aorta, since Lucifer Yellow has a charge of –2(Veenstra et al., 1994). Potentially, differences in the physical properties of neurobiotin versus biocytin could account for the different dye-transfer results obtained in this study, given that biocytin is structurally similar, but not identical to neurobiotin. Neurobiotin has a molecular mass of 287 Da and carries a positive charge at physiological pH, whereas biocytin is 372 Da and is electrically neutral. However, we also observed a decrease in transfer of NBD-TMA, which has a molecular mass of 266 Da and carries a positive charge: properties that are similar to neurobiotin. Thus, it is unclear why Cx40–/– endothelium would exhibit decreased transfer of biocytin and NBD-TMA, but not of neurobiotin.

What are the consequences of deficient endothelial communication on vascular health? Mice that completely lack both Cx37 and Cx40 are not viable beyond the first postnatal day and exhibit severe vascular abnormalities(Simon and McWhorter, 2002). Cx37–/–Cx40–/– mice have localized hemorrhages in skin, testis, gastrointestinal tissues and lungs, as well as blood vessel dilatation and congestion. Moreover, vascular dysmorphogenesis is evident in testis and intestine of Cx37–/–Cx40–/– animals. Thus,endothelial communication is required for the normal development and/or functional maintenance of portions of the mouse vasculature. The survival data presented here suggest that even with connexin-deficient animals that retain low levels of endothelial coupling, there may be a reduction in long-term survival. After two years, there was a substantial difference in the percentage of surviving Cx37+/–Cx40–/–animals versus wild-type animals. Cx37+/–Cx40–/– animals, which have∼80% reduction in aortic endothelial dye-transfer, may therefore serve as a good animal model for studying the long-term as well as short-term physiological effects of compromised endothelial communication.

Acknowledgements

We are grateful to Jan Burt for advice about dye injections and for the use of injection equipment. We thank Steve Wright for the gift of NBD-TMA, David Paul for connexin antisera, Ann Baldwin for advice about silver nitrate staining, and Julie Dones for excellent technical assistance. This work was supported by the Arizona Disease Control Research Commission (10018 to A.M.S.)and by the National Institutes of Health (HL64232 to A.M.S.).

References

Bednarczyk, D., Mash, E. A., Aavula, B. R. and Wright, S. H.(
2000
). NBD-TMA: a novel fluorescent substrate of the peritubular organic cation transporter of renal proximal tubules.
Pflugers Arch.
440
,
184
-192.
Beyer, E. C., Paul, D. L. and Goodenough, D. A.(
1987
). Connexin43: a protein from rat heart homologous to a gap junction protein from liver.
J. Cell Biol.
105
,
2621
-2629.
Brink, P. R., Cronin, K., Banach, K., Peterson, E., Westphale,E. M., Seul, K. H., Ramanan, S. V. and Beyer, E. C. (
1997
). Evidence for heteromeric gap junction channels formed from rat connexin43 and human connexin37.
Am. J. Physiol. Cell Physiol.
42
,
C1386
-C1396.
Bruzzone, R., Haefliger, J. A., Gimlich, R. L. and Paul, D. L. (
1993
). Connexin40, a component of gap junctions in vascular endothelium, is restricted in its ability to interact with other connexins.
Mol. Biol. Cell
4
,
7
-20.
Cai, W. J., Koltai, S., Kocsis, E., Scholz, D., Schaper, W. and Schaper, J. (
2001
). Connexin37, not Cx40 and Cx43, is induced in vascular smooth muscle cells during coronary arteriogenesis.
J. Mol. Cell. Cardiol.
33
,
957
-967.
Chaytor, A. T., Evans, W. H. and Griffith, T. M.(
1998
). Central role of heterocellular gap junctional communication in endothelium-dependent relaxations of rabbit arteries.
J. Physiol.
508
,
561
-573.
Chomczynski, P. and Sacchi, N. (
1987
). Single-step method of RNA isolation by acid guanidium thiocyanate-phenol-cholorform extraction.
Anal. Biochem.
162
,
156
-159.
Christ, G. J., Spray, D. C., el-Sabban, M., Moore, L. K. and Brink, P. R. (
1996
). Gap junctions in vascular tissues. Evaluating the role of intercellular communication in the modulation of vasomotor tone.
Circ. Res.
79
,
631
-646.
de Wit, C., Roos, F., Bolz, S. S., Kirchhoff, S., Krüger,O., Willecke, K. and Pohl, U. (
2000
). Impaired conduction of vasodilation along arterioles in connexin40-deficient mice.
Circ. Res.
86
,
649
-655.
Emerson, G. G. and Segal, S. S. (
2000
). Endothelial cell pathway for conduction of hyperpolarization and vasodilation along hamster feed artery.
Circ. Res.
86
,
94
-100.
Gabriels, J. E. and Paul, D. L. (
1998
). Connexin43 is highly localized to sites of disturbed flow in rat aortic endothelium but connexin37 and connexin40 are more uniformly distributed.
Circ. Res.
83
,
636
-643.
Goodenough, D. A., Goliger, J. A. and Paul, D. L.(
1996
). Connexins, connexons, and intercellular communication.
Ann. Rev. Biochem.
65
,
475
-502.
Haefliger, J. A., Castillo, E., Waeber, G., Bergonzelli, G. E.,Aubert, J. F., Sutter, E., Nicod, P., Waeber, B. and Meda, P.(
1997
). Hypertension increases connexin43 in a tissue-specific manner.
Circulation
95
,
1007
-1014.
Haefliger, J. A., Demotz, S., Braissant, O., Suter, E., Waeber,B., Nicod, P. and Meda, P. (
2001
). Connexins 40 and 43 are differentially regulated within the kidneys of rats with renovascular hypertension.
Kidney Int.
60
,
190
-201.
He, D. S., Jiang, J. X., Taffet, S. M. and Burt, J. M.(
1999
). Formation of heteromeric gap junction channels by connexins 40 and 43 in vascular smooth muscle cells.
Proc. Natl. Acad. Sci. USA
96
,
6495
-6500.
Hertzberg, E. L. (
1984
). A detergent-independent procedure for the isolation of gap junctions from rat liver.
J. Biol. Chem.
259
,
9936
-9943.
Jiang, J. X. and Goodenough, D. A. (
1996
). Heteromeric connexons in lens gap junction channels.
Proc. Natl. Acad. Sci. USA
93
,
1287
-1291.
Johnson, C. M., Kanter, E. M., Green, K. G., Laing, J. G.,Betsuyaku, T., Beyer, E. C., Steinberg, T. H., Saffitz, J. E. and Yamada, K. A. (
2002
). Redistribution of connexin45 in gap junctions of connexin43-deficient hearts.
Cardiovasc. Res.
53
,
921
-935.
Ko, Y. S., Yeh, H. I., Rothery, S., Dupont, E., Coppen, S. R. and Severs, N. J. (
1999
). Connexin make-up of endothelial gap junctions in the rat pulmonary artery as revealed by immunoconfocal microscopy and triple-label immunogold electron microscopy.
J. Histochem. Cytochem.
47
,
683
-692.
Koval, M., Geist, S. T., Westphale, E. M., Kemendy, A. E.,Civitelli, R., Beyer, E. C. and Steinberg, T. H. (
1995
). Transfected connexin45 alters gap junction permeability in cells expressing endogenous connexin43.
J. Cell Biol.
130
,
987
-995.
Krüger, O., Plum, A., Kim, J., Winterhager, E., Maxeiner,S., Hallas, G., Kirchoff, S., Traub, O., Lamers, W. H. and Willecke, K.(
2000
). Defective vascular development in connexin 45-deficient mice.
Development
127
,
4179
-4193.
Krüger, O., Beny, J. L., Chabaud, F., Traub, O., Theis, M.,Brix, K., Kirchhoff, S. and Willecke, K. (
2002
). Altered dye diffusion and upregulation of connexin37 in mouse aortic endothelium deficient in connexin40.
J. Vasc. Res.
39
,
160
-172.
Kumar, N. M. and Gilula, N. B. (
1996
). The gap junction communication channel.
Cell
84
,
381
-388.
Kurjiaka, D. T., Steele, T. D., Olsen, M. V. and Burt, J. M.(
1998
). Gap junction permeability is diminished in proliferating vascular smooth muscle cells.
Am. J. Physiol.
275
,
C1674
-C1682.
Larson, D. M. and Haudenschild, C. C. (
1988
). Junctional transfer in wounded cultures of bovine aortic endothelial cells.
Lab. Invest.
59
,
373
-379.
Larson, D. M., Haudenschild, C. C. and Beyer, E. C.(
1990
). Gap junction messenger RNA expression by vascular wall cells.
Circ. Res.
66
,
1074
-1080.
Li, X. and Simard, J. M. (
1999
). Multiple connexins form gap junction channels in rat basilar artery smooth muscle cells.
Circ. Res.
84
,
1277
-1284.
Little, T. L., Beyer, E. C. and Duling, B. R.(
1995
). Connexin 43 and connexin 40 gap junctional proteins are present in arteriolar smooth muscle and endothelium in vivo.
Am. J. Physiol.
268
,
H729
-H739.
McDonald, D. M. (
1994
). Endothelial gaps and permeability of venules in rat tracheas exposed to inflammatory stimuli.
Am. J. Physiol.
266
,
L61
-L83.
Nakamura, K., Inai, T. and Shibata, Y. (
1999
). Distribution of gap junction protein connexin 37 in smooth muscle cells of the rat trachea and pulmonary artery.
Arch. Histol. Cytol.
62
,
27
-37.
Nelles, E., Butzler, C., Jung, D., Temme, A., Gabriel, H. D.,Dahl, U., Traub, O., Stumpel, F., Jungermann, K., Zielasek, J. et al.(
1996
). Defective propagation of signals generated by sympathetic nerve stimulation in the liver of connexin32-deficient mice.
Proc. Natl. Acad. Sci. USA
93
,
9565
-9570.
Pepper, M. S., Spray, D. C., Chanson, Montesano, R., Orci, L. and Meda, P. (
1989
). Junctional communication is induced in migrating capillary endothelial cells.
J. Cell Biol.
109
,
3027
-3038.
Reed, K. E., Westphale, E. M., Larson, D. M., Wang, H.-Z.,Veenstra, R. D. and Beyer, E. C. (
1993
). Molecular cloning and functional expression of human connexin37, an endothelial cell gap junction protein.
J. Clin. Invest.
91
,
997
-1004.
Rong, P., Wang, X., Niesman, I., Wu, Y., Benedetti, L. E.,Dunia, I., Levy, E. and Gong, X. (
2002
). Disruption of Gja8(alpha8 connexin) in mice leads to microphthalmia associated with retardation of lens growth and lens fiber maturation.
Development
129
,
167
-174.
Rummery, N. M., Hickey, H., McGurk, G. and Hill, C. E.(
2002
). Connexin37 is the major connexin expressed in the media of caudal artery.
Arterioscler. Thromb. Vasc. Biol.
22
,
1427
-1432.
Schubert, A. L., Schubert, W., Spray, D. C. and Lisanti, M. P. (
2002
). Connexin family members target to lipid raft domains and interact with caveolin-1.
Biochemistry
41
,
5754
-5764.
Segal, S. S. and Duling, B. R. (
1989
). Conduction of vasomotor responses in arterioles: a role for cell-to-cell coupling?
Am. J. Physiol.
256
,
H838
-H845.
Simon, A. M. and McWhorter, A. R. (
2002
). Vascular abnormalities in mice lacking the endothelial gap junction proteins connexin37 and connexin40.
Dev. Biol.
251
,
206
-220.
Simon, A. M., Goodenough, D. A., Li, E. and Paul, D. L.(
1997
). Female infertility in mice lacking connexin 37.
Nature
385
,
525
-529.
Simon, A. M., Goodenough, D. A. and Paul, D. L.(
1998
). Mice lacking connexin40 have cardiac conduction abnormalities characteristic of atrioventricular block and bundle branch block.
Curr. Biol.
8
,
295
-298.
Stauffer, K. A. (
1995
). The gap junction proteins β-1-connexin (connexin-32) and β-2-connexin (connexin-26)can form heteromeric hemichannels.
J. Biol. Chem.
270
,
6768
-6772.
Traub, O., Hertlein, B., Kasper, M., Eckert, R., Krisciukaitis,A., Hulser, D. and Willecke, K. (
1998
). Characterization of the gap junction protein connexin37 in murine endothelium, respiratory epithelium, and after transfection in human HeLa cells.
Eur. J. Cell Biol.
77
,
313
-322.
van Kempen, M. J. and Jongsma, H. J. (
1999
). Distribution of connexin37, connexin40 and connexin43 in the aorta and coronary artery of several mammals.
Histochem. Cell Biol.
112
,
479
-486.
Veenstra, R. D., Wang, H. Z., Beyer, E. C., Ramanan, S. V. and Brink, P. R. (
1994
). Connexin37 forms high conductance gap junction channels with subconductance state activity and selective dye and ionic permeabilities.
Biophys. J.
66
,
1915
-1928.
Willecke, K., Eiberger, J., Degen, J., Eckardt, D., Romualdi,A., Guldenagel, M., Deutsch, U. and Sohl, G. (
2002
). Structural and functional diversity of connexin genes in the mouse and human genome.
Biol. Chem.
383
,
725
-737.
Yeh, H. I., Dupont, E., Coppen, S., Rothery, S. and Severs, N. J. (
1997
). Gap junction localization and connexin expression in cytochemically identified endothelial cells of arterial tissue.
J. Histochem. Cytochem.
45
,
539
-550.
Yeh, H. I., Rothery, S., Dupont, E., Coppen, S. R. and Severs,N. J. (
1998
). Individual gap junction plaques contain multiple connexins in arterial endothelium.
Circ. Res.
83
,
1248
-1263.
Yeh, H. I., Chang, H. M., Lu, W. W., Lee, Y. N., Ko, Y. S.,Severs, N. J. and Tsai, C. H. (
2000
). Age-related alteration of gap junction distribution and connexin expression in rat aortic endothelium.
J. Histochem. Cytochem.
48
,
1377
-1389.