Decreased Intercellular Communication and Connexin Expression in Mouse Aortic Endothelium during Lipopolysaccharide-Induced InflammationSimon A.M.a · McWhorter A.R.a · Chen H.a · Jackson C.L.a · Ouellette Y.b
aDepartment of Physiology, University of Arizona, Tucson, Ariz., and bDepartment of Pediatrics, Mayo Clinic College of Medicine, Rochester, Minn., USA Corresponding Author
The role of gap junctional intercellular communication during inflammatory processes is not well understood. In particular, changes in the expression and function of vascular endothelial connexins (gap junction proteins) in response to inflammatory agents has not been fully investigated. In this study, we used intercellular dye transfer methods to assess interendothelial communication in aortic segments isolated from mice treated with or without intraperitoneal lipopolysaccharide (LPS), a potent inflammatory mediator. LPS treatment resulted in a 49% decrease in endothelial dye coupling 18 h after injection. Western blots indicated that LPS treatment also caused a reduction in endothelial connexin40 (Cx40) levels to 33% of control levels. Connexin37 (Cx37) levels decreased only slightly after LPS treatment to 79% of control levels. We also examined endothelial communication in aortic segments isolated from Cx37–/– and Cx40–/– mice. LPS treatment caused a significantly greater decrease in dye transfer in endothelium isolated from Cx37–/– animals compared with endothelium from Cx40–/– animals (71 vs. 26% decrease). LPS injection caused a reduction in Cx40 levels in Cx37–/– endothelium, whereas LPS actually increased Cx37 levels in Cx40–/– endothelium. These results suggest that LPS mediates changes in endothelial gap junction-mediated communication, at least in part, through modulation of Cx40 and Cx37 levels.
Copyright © 2004 S. Karger AG, Basel
In the vascular wall, intercellular communication via gap junction channels is important in the propagation of signals that coordinate vasomotor responses in arterioles and arteries [1, 2, 3, 4]. Gap junctions are structures that allow the passage of ions and small molecules (<1 kD) between adjacent cells [5, 6]. The gap junction channel is comprised of two hemichannels, one in each of the plasma membranes of the contacting cells. Each hemichannel is an hexameric complex made up of connexin subunits, which are expressed in a tissue-specific manner . In vascular endothelium, connexin37 (Cx37), connexin40 (Cx40) and connexin43 (Cx43) expression has been documented and the proteins detected at cell-cell appositions [8, 9, 10, 11, 12, 13, 14, 15]. Expression of vascular connexins, however, varies with vessel type and species. Mouse aortic endothelial cells, for example, most commonly contain only Cx37 and Cx40 and these connexins alone may be responsible for interendothelial communication in mouse aorta [16, 17].
Endotoxin, also known as lipopolysaccharide (LPS), is a potent inflammatory mediator produced by gram-negative bacteria and contributes to a generalized inflammatory response known as septic shock [18, 19, 20]. Septic shock is characterized by disseminated intravascular coagulation, systemic vascular collapse, and development of vascular leak syndrome, leading to multiple organ failure [21, 22, 23, 24, 25, 26]. LPS may exert its effects on the vascular endothelium by both direct and indirect mechanisms. Following LPS injection, endothelial cells are stimulated to produce proinflammatory cytokines, tissue factor, and inflammation-associated adhesion molecules (E-selectin, ICAM-1, VCAM-1) . LPS also induces endothelial apoptosis, a feature which likely contributes to the pathogenesis of sepsis . The pathological changes seen in the circulation of different organs during endotoxemia depend upon alterations that upset the balance inherent in endothelial control mechanisms. For example, control of vasomotor responses is heterogeneous during sepsis and regional maldistribution of blood flow occurs [27, 28]. The abnormal vasomotor responses result in a drop in total peripheral resistance, a major contributor to sepsis-induced hypotension.
Several studies have examined the effects of LPS and cytokines on endothelial gap junction-mediated communication and connexin expression. Lidington et al. [29, 30] demonstrated that LPS increases intercellular resistance between rat microvascular endothelial cells. In addition, in vivo studies with mouse cremaster arterioles showed that LPS reduces arteriolar-conducted responses, suggesting diminished vascular wall coupling [31, 32]. Interleukin-1α (IL-1α) suppresses gap junction-mediated communication in cultures of human umbilical vein endothelial cells (HUVECs) . In another study, LPS, tumor necrosis factor-α (TNF-α), and IL-1β were shown to selectively inhibit human myoendothelial gap junction-mediated communication . TNF-α was also shown to reduce expression of Cx37 and Cx40 in HUVEC cells resulting in a decrease in dye coupling . Despite significant knowledge gained from these studies, the expression and function of specific endothelial connexins during inflammation remains poorly understood. In particular, the relative expression of Cx37 and Cx40 and their role in endothelial cell communication during inflammation have not been fully investigated. In this report, we used intercellular dye transfer to assess interendothelial communication in wild-type, Cx37–/– and Cx40–/– aortic segments isolated from mice treated with or without intraperitoneal LPS. In addition, we performed Western blots to compare the pattern of endothelial connexin expression in animals treated with or without LPS. We demonstrate a decrease in both intercellular communication and expression of Cx40 and Cx37 in wild-type mouse aortic endothelium during LPS-induced inflammation.
Cx37–/– and Cx40–/– mice and PCR genotyping protocols have been previously described [36, 37, 38]. Cx37–/– and Cx40–/– mice were originally generated on a mixed 129/Sv-C57BL/6 strain background. Subsequently, the Cx37–/– and Cx40–/– lines were backcrossed with C57BL/6 mice for 6 generations to place them on a predominantly C57BL/6 strain background. Wild-type mice were also C57BL/6 strain.
LPS (Escherichia coli serotype 055:B5, Sigma-Aldrich) was dissolved in sterile saline (1 mg/ml) and injected intraperitoneally into mice at 5 mg/kg body weight. The mean volume injected was 117 μl. In additional animals, sterile saline was substituted for LPS during the injection as a control. Animals exhibited mild lethargy within 1 h after injection with LPS. Injection of LPS into various animal models affects fluid balance by altering the barrier function of the vascular endothelium . We measured body weights before and after injection of the animals to confirm that LPS was having a physiological effect. Injections with LPS resulted in a decrease in body weight 18 h postinjection (wild-type: –7.4 ± 0.5%, n = 20; Cx37–/–: –5.1 ± 1.0%, n = 5; Cx40–/–: –6.6 ± 0.7%, n = 10). Statistical analysis (ANOVA) indicated no significant difference in weight loss between genotypes. Animals injected with sterile saline alone typically increased in weight slightly after 18 h (mean weight increase for all genotypes was 2.6%) and there was no significant difference in the weight gains between genotypes. Aortas were harvested 18 h after injection and processed for dye transfer experiments or Western blotting. Control saline injections gave dye transfer results that were similar to results obtained previously with noninjected animals .
The methods used for dye transfer experiments have been previously described in detail . Briefly, thoracic aortas were cut into three equally sized segments and pinned out. Segments were stained briefly with 25 μM Hoechst 33342 (Molecular Probes) to visualize nuclei, then submerged in PBS containing 0.90 mM Ca2+, 0.49 mM Mg2+, and 1% BSA during microinjections. A mixture of 5% biocytin (Mr 372, neutral charge) and 10 mg/ml dextran-fluorescein (Mr ∼3 × 103; both from Molecular Probes) was injected into an endothelial cell for 10 s using the capacitance overcompensation feature of the amplifier. Biocytin is permeable through gap junction channels, whereas dextran-fluorescein is too large to pass through the channels. Each aortic segment was microinjected at up to three well-separated sites. We previously found no significant difference in dye transfer results obtained from proximal, middle, or distal thoracic aorta segments, so the data obtained from these aortic segments were pooled. Transfer of biocytin was allowed to proceed for 15 min before the aorta segments were fixed in 4% paraformaldehyde. The fixed tissue was washed and blocked in PBS containing 2% BSA and 0.25% Triton X-100. Vessels were incubated in tetramethylrhodamine (TMR)-NeutrAvidin (Molecular Probes), washed, mounted on slides in PBS, and the number of biocytin-labeled cells was counted using a fluorescence microscope. Cells were scored as receiving transferred biocytin if the fluorescence signal was above the background autofluorescence. Each injection field was counted 4 times and the mean cell count was used for statistical analysis. Dye transfer in this assay is attributed to gap junctional communication by three criteria: (1) biocytin transfers to adjacent cells but dextran-fluorescein does not transfer ; (2) dye transfer was blocked by heptanol, a gap junction blocker (unpubl. data), and (3) dye transfer is eliminated in aortic endothelium lacking both Cx37 and Cx40 . The filters used for fluorescence detection were Olympus U-MWU2 for Hoechst 33342 dye (U excitation, wide band pass, BP 330–385 nm, BA 420 nm, DM 400 nm), Olympus U-MNG2 for TMR (G excitation, narrow band pass, BP 530–550 nm, BA 590 nm, DM 570 nm), and Olympus U-MNB2 for FITC (B excitation, narrow band pass, BP 470–490 nm, BA 520 nm IF, DM 500 nm). Fluorescence overlap was not detected between Hoechst 33342, TMR, and FITC signals.
Anti-Cx37 (raised against amino acids 229–333 of rat Cx37), anti-Cx40 (raised against amino acids 231–331 of rat Cx40), and anti-Cx43 (raised against amino acids 252–271 of rat Cx43) sera were provided by David Paul (Harvard Medical School) and were affinity purified before use . The anti-Cx37 antibody used for immunostaining was obtained from Alpha Diagnostic International (Cx37 A11-A). The following antibodies were also obtained commercially: anti-caveolin-1 (Transduction Laboratories) and anti- PECAM-1 (Santa Cruz Biotech, M-20).
For wild-type animal experiments, Western blots were performed with aortas collected from eight independent LPS injection experiments. Immunoblotting of aortic endothelial membranes was done as previously described . Alkaline-extracted membranes were collected by passing 200 μl of lysis solution (20 mM NaOH containing the following protease inhibitors: 16 μg/ml benzamidine, 10 μg/ml phenanthroline, 10 μg/ml aprotinin, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, 1 mM phenylmethanesulfonyl fluoride) 3 times through the lumen of thoracic aortas, using a 22- to 24-gauge needle. In each experiment, lysates collected from aortas of 5–6 animals (∼6–7 weeks old) were pooled. Equal lengths of aorta were extracted in each group. Lysates were passed through a 26-gauge needle 10 times and a sample was removed for protein determination. Lysates were spun at 100,000 g for 40 min to pellet membranes, which were resuspended in 30 μl of SDS sample buffer. Samples were boiled for 5 min, run on a 12% SDS-polyacrylamide gel and transferred to nitrocellulose. For membranes probed with multiple antibodies, the top portion of the membrane (above ∼56 kD) was cut off and blotted separately for PECAM-1. The bottom portion of the membrane (below ∼56 kD) was blotted sequentially for Cx40, Cx37, and caveolin-1, with stripping of the membrane between blots. Stripping was achieved by incubating the membrane in 62.5 mM Tris-HCl, pH 6.7, 100 mM 2-mercaptoethanol, 2% SDS for 45 min at 50°C. Cx40–/– samples were blotted sequentially for Cx37, Cx40, and caveolin-1. Anti-Cx37 and anti-Cx40 (∼0.5 mg/ml) were diluted 1/2,500. Anti-PECAM-1 (0.5 mg/ml) and anti-caveolin-1 (0.25 mg/ml) were diluted 1/1,000 and 1/10,000, respectively. Following primary antibody incubations for 2 h at room temperature, membranes were incubated with HRP-conjugated secondary antibodies (Pierce) for 1 h at room temperature (1/80,000 dilution) and then processed for chemiluminescence with Supersignal West Dura substrate (Pierce) or Supersignal West Femto substrate (Pierce). Bands were quantified from film by densitometry using a BioRad Gel Doc system or with a cooled CCD imaging system (Kodak IS2000R). Cx37, Cx40, and caveolin-1 levels were normalized to PECAM-1 signals by determining the ratio of the immunodetected signal to the PECAM-1 signal detected on the same blot. PECAM-1 was considered an appropriate normalization standard because it is expressed by vascular endothelium but not by vascular smooth muscle and because PECAM-1 signals were not significantly different in LPS-injected versus control-injected samples (for an equal length of aorta). Signals obtained from LPS-injected animals were also expressed as a percentage of the signal obtained from control-injected animals. For statistical analysis of differences, samples from four groups of LPS-injected and control-injected animals were analyzed on a single blot so that the variability in control signals as well as LPS-injected signals could be included. For Cx40 blots, the predominant band at 40 kD was quantified; a fainter, faster migrating band is likely a degradation product. Cx37 blots produced a broad signal at ∼37 kD that may consist of several bands with differing mobilities. Cx37 was found previously to be phosphorylated after transfection into cultured cells, which could explain the existence of multiple bands . Since the diffuse Cx37 bands did not resolve well, we quantified all forms of Cx37 in this region together. In some experiments, very faint Cx37-related bands were detected at ∼66 and ∼29 kD; these were considered the result of aggregation and degradation, respectively. Two alternative isoforms of caveolin-1 were detected on the blots, a very strong caveolin-1 band at 24 kD (alpha isoform) and a much weaker, slightly faster migrating caveolin-1 band at 21 kD (beta isoform) . The predominant alpha isoform of caveolin-1 was quantified. As controls, recombinant glutathione S-transferase (GST)-connexin fusion proteins were prepared for Cx37, Cx40, and Cx43 using vectors kindly supplied by David Paul (Harvard Medical School) and Alan Lau (University of Hawaii) [9, 40].
Previous work indicates that there is little or no contamination of our endothelial membrane preparations with medial layer- or advential layer-derived connexins . Cx43, a connexin expressed only in the medial layer of mouse aorta, was detected in whole aorta membrane preparations but not in the endothelial membrane preparations. In addition, immunostaining of aortic cross sections after our luminal extraction procedure showed that endothelial Cx37 the Cx40 staining was efficiently removed, whereas medial layer immunostaining for Cx43 and Cx37 was unaffected.
Aortas were frozen unfixed in Tissue-Tek OCT embedding medium and sectioned at 10 μm. Sections were fixed in acetone at –20°C for 5 min, blocked in a solution containing PBS, 4% fish skin gelatin, 1% goat serum, 0.25% Triton X-100, and incubated with primary antibodies for 2 h at room temperature. Anti-Cx37 (Cx37 A11-A, 0.5 mg/ml), anti-Cx40 (∼0.5 mg/ml) and anti-Cx43 (∼0.5 mg/ml) were diluted 1/250, 1/300 and 1/300, respectively. Sections were washed and then incubated with CY3-conjugated secondary antibodies (Jackson ImmunoResearch) diluted 1/400 for 1 h at room temperature. After washing, sections were mounted in Mowiol 40-88 (Aldrich) containing 1,4-diazobicyclo-(2,2,2)-octane and viewed with an Olympus BX51 fluorescence microscope. For en face immunostaining, aortas were fixed by perfusion with 0.5% paraformaldehyde and cut open longitudinally before blocking and incubating with connexin antibodies. En face stained segments were mounted in PBS. Images were captured with a SensSys 1401 CCD camera (Photometrics) and ImagePro software (Media Cybernetics). For comparisons, the same exposure settings were used to obtain the digital images of control and LPS-treated aortas.
Endothelial cell density in the thoracic aorta was determined after en face immunostaining for Cx40, which outlines individual endothelial cells. Fields of immunostained endothelial cells were photographed with fluorescence optics using a 40× objective and the number of endothelial cells outlined in each field was divided by the area of the field (0.063 mm2). The mean cell density ± SEM of four fields was determined for each treatment group.
Dye transfer data and Western blot data were compared statistically using an unpaired t test.
Dye transfer experiments were performed to examine the effects of intraperitoneal LPS injection on endothelial cell-to-cell communication. We measured biocytin dye coupling between endothelial cells in segments of thoracic aorta from mice 18 h following intraperitoneal injection of LPS or sterile saline (control). To illustrate the dye transfer method, examples of biocytin dye diffusion in wild-type endothelium are shown in figure 1. Under control conditions, the mean number of stained cells after a 15-min transfer time was 337 cells (table 1). Treatment with LPS reduced the number of stained cells significantly to 51% of control levels (p < 0.01). Thus, intraperitoneal injection of LPS resulted in a significant reduction in aortic endothelial dye transfer.
Fig. 1. Example of biocytin dye transfer method to assess gap junction-mediated coupling in the aortic endothelium. A Low-power image showing the extent of biocytin transfer in wild-type endothelium. Transfer was detected to a few hundred neighboring cells. B A higher magnification image of a different wild-type dye transfer experiment is shown. Injected cells are marked with an asterisk. Biocytin was detected with tetramethylrhodamine (TMR)-NeutrAvidin. Scale bars: 50 μm.
Table 1. Amount of dye transfer in aortic endothelium of wild-type and connexin-deficient mice following intraperitoneal injection of sterile saline (control) or LPS
We tested the relative involvement of Cx37 and Cx40 in the LPS-mediated reduction in interendothelial communication by performing dye transfer experiments with Cx37–/– and Cx40–/– mouse aorta (table 1). Since Cx43 is not detected in wild-type or knockout mouse aortic endothelium, the knockout mice allow a test of the effect of LPS when only Cx37 or Cx40 is present [16, 17]. Consistent with our previously published data, Cx37–/– and Cx40–/– saline-injected aortas showed less dye transfer than wild-type control-injected aortas, presumably reflecting decreased total gap junction channel content in the knockout endothelium . Surprisingly, LPS treatment had a significantly greater effect on dye transfer in endothelium isolated from Cx37–/– animals compared with endothelium from Cx40–/– animals (table 1). LPS treatment reduced dye transfer in aortas isolated from Cx40–/– mice to 74% of control values (table 1). In contrast, with aortas isolated from Cx37–/– animals, there was a substantially greater reduction in dye transfer (to 29% of control value) following LPS treatment. Thus, LPS had a greater effect on endothelial coupling when only Cx40 was present (Cx37–/–) than when only Cx37 was present (Cx40–/–). These results indicate that LPS-induced inflammation significantly affects endothelial coupling in mouse aorta and suggested that Cx40 expression or function might be more strongly affected by LPS treatment than that of Cx37.
Western blotting was done to compare changes in connexin levels between saline and LPS-treated groups. Endothelial-specific lysates were prepared from thoracic aortas and alkaline-resistant membranes collected by ultracentrifugation. Extraction of plasma membranes with an alkaline solution partially enriches for gap junction plaques [41, 42]. The specificity of the antibodies used to detect Cx37 and Cx40 by Western blotting is demonstrated in figure 2. The Cx37 antibody predominantly detected a broad band (or complex of bands) at ∼37 kD on blots of wild-type endothelial membrane preps and this signal was absent from Cx37–/– samples (fig. 2A). Likewise, the Cx40 antibody predominantly detected a protein of 40 kD that was present in wild-type samples but absent from Cx40–/– samples (fig. 2A). In addition, the specificity of the anticonnexin antibodies was confirmed by immunoblotting samples of recombinant GST-connexin fusion proteins (fig. 2B) [9, 40].
Fig. 2. The specificity of the antibodies used to detect Cx37 and Cx40. A Endothelial membrane preparations were collected from wild-type, Cx37–/–, and Cx40–/– thoracic aortas and immunoblotted with Cx37 or Cx40 antibodies. The Cx37 antibody detected a broad signal at ∼37 kD in the wild-type sample which was absent in the Cx37–/– sample. The Cx40 antibody detected a band at 40 kD in the wild-type sample which was absent in the Cx40–/– sample. B The Cx37 and Cx40 antibodies recognized only the appropriate GST-connexin fusion proteins, confirming the specificity of the antibodies. Positions of molecular weight markers are indicated in kilodaltons.
To examine the role of connexin expression changes in the LPS-mediated decrease in endothelial cell-to-cell communication, relative levels of Cx37 and Cx40 protein were compared in aortic endothelium of wild-type animals treated with or without LPS (fig. 3). To control for differences in endothelial cell content, connexin levels were normalized to levels of PECAM-1 measured on the same blots. The results of four separate LPS injections analyzed on a single blot are shown in figure 3. Quantification of this blot revealed that the amount of Cx40 protein in LPS-treated animals significantly decreased to 33% of control-treated animals (p < 0.05) (table 2). Cx37 protein levels decreased only slightly after LPS treatment to 79% of control-treated animals (p < 0.05) (table 2). We also probed the blot with an antibody against caveolin-1, a out data because of low sample size.protein that is highly expressed by vascular endothelial cells and which was recently shown to interact with some connexins . Caveolin-1 increased slightly (16%) after LPS treatment (p < 0.05) (table 2). Quantification of four additional LPS Western blot experiments, analyzed on separate blots, also showed reductions in Cx40 and Cx37 levels after LPS treatment (not shown).
Fig. 3. Intraperitoneal injection of LPS results in a decrease in endothelial Cx40 and Cx37 levels. Membrane fractions collected from five wild-type thoracic aortas were pooled in each lane and analyzed by Western blotting. Results from four independent LPS injection experiments were analyzed on the same blot. Cx40, Cx37, caveolin-1 and PECAM-1 were detected on the membrane and signal intensities normalized to PECAM-1 levels. The amount of Cx40 declined significantly following LPS treatment compared to sterile saline control (Con.). Cx37 levels also decreased following LPS treatment, but to a lesser extent (see table 2 for quantification). Positions of molecular weight markers are indicated in kilodaltons.
Table 2. Expression of Cx40, Cx37,caveolin-1, and PECAM protein in aorticendothelium of LPS-injected wild-typemice versus control-injected mice
Relative levels of Cx37 and Cx40 protein were also compared in aortic endothelium of connexin knockout animals treated with or without LPS (fig. 4). These Western blots allowed us to examine the effect of LPS on connexin levels when only Cx37 or Cx40 is present. We previously showed that nonablated endothelial connexin levels are reduced in Cx40–/– and Cx37–/– aorta, relative to wild-type levels . We therefore used a more sensitive chemiluminescence substrate for detection of the connexin immunosignals from the knockout animals. Cx40 levels in LPS-treated Cx37–/– animals decreased to 41% of control-treated Cx37–/– animals, similar to the change observed in wild-type animals (fig. 4, table 2). Surprisingly, Cx37 levels in LPS-treated Cx40–/– animals did not decrease, but rather increased by 69% (fig. 4, table 2). Thus, LPS had a similar effect on Cx40 levels whether or not Cx37 was present, but the effect of LPS on relative Cx37 levels differed depending on the presence or absence of Cx40.
Fig. 4. Effect of LPS treatment on connexin levels in Cx37–/– and Cx40–/– aortic endothelium. Membrane fractions collected from five Cx37–/– or Cx40–/– thoracic aortas were pooled in each lane and analyzed by Western blotting. Cx40, Cx37, caveolin-1 and PECAM-1 were detected on the same membrane. Signal intensities were normalized to PECAM-1 levels. With Cx37–/– aortas, LPS treatment resulted in a decrease in Cx40 levels compared with sterile saline control (Con.). Surprisingly, LPS treatment resulted in an increase in Cx37 levels compared to control treatment in Cx40–/– aortas. Positions of molecular weight markers are indicated in kilodaltons.
We investigated the localization of Cx37 and Cx40 in wild-type aorta after treatment with or without intraperitoneal LPS. Cryosections showed that both Cx37 and Cx40 were localized predominately in the aortic endothelium in control (fig. 5A, B, E, F) and LPS-treated animals (fig. 5C, D, G, H). Cx37 is also expressed at lower levels in the medial layer of mouse aorta . In order to further analyze the distribution of Cx37 and Cx40 in the endothelium, we used en face whole-mount immunocytochemistry of aortas from control-treated (fig. 5I, K) and LPS-treated (fig. 5J, L) animals. The distribution of Cx40 and Cx37 in aortic endothelial cell membranes was similar in LPS and control-injected animals. The intensity of the Cx40 immunosignal typically appeared slightly reduced following LPS treatment. We also immunostained aortic cryosections with antibodies against Cx43 to test if LPS treatment affected Cx43 expression. Cx43 was not detected in the endothelial layer of LPS or control-injected animals (not shown). Cx43 was detected in the medial layer of both treatment groups, but did not appear to be upregulated with LPS treatment.
Fig. 5. Cx37 and Cx40 immunostaining in wild-type aortic endothelium following intraperitoneal injection of LPS. A–H Cryosections (cross sections) of aortas from LPS-injected (C, D, G, H) or sterile saline-injected (A, B, E, F) control (Con.) animals were immunostained with anti-Cx40 (A–D) or anti-Cx37 (E–H). En face preparations of aortas from LPS-injected (J, L) or control (I, K) animals were immunostained for Cx40 (I, J) or Cx37 (K, L). The distribution of Cx40 and Cx37 in aortic endothelial cells was similar in LPS and control-injected animals. The intensity of the Cx40 immunosignal typically appeared slightly reduced following LPS treatment. Scale bars: 20 μm.
To test if changes in endothelial cell density or morphology might have affected our dye transfer experiments, we used en face whole-mount immunocytochemistry to determine endothelial cell density in LPS versus control-injected animals. Cx40 immunostaining, which outlines the individual endothelial cells, was used to identify cells for counting (see fig. 5 for example). No significant difference was found in endothelial cell density in LPS-treated animals (2,470 ± 60 cells/mm2, 626 cells counted) versus control-treated animals (2,486 ± 37 cells/mm2, 630 cells counted). In addition, we did not detect changes in endothelial cell morphology with LPS treatment. These results indicate that the LPS-induced changes in dye transfer reflect gap junction-mediated communication rather than changes in cell shape or size. Furthermore, these data are consistent with the observation that PECAM-1 levels in LPS and control-treated aortas (measured from equal lengths of aorta) were not significantly altered by LPS (table 2).
The aim of this study was to test the hypothesis that LPS-induced inflammation attenuates cell-to-cell communication in mouse aortic endothelial cells and to assess changes in the expression of Cx37 and Cx40 during the LPS-mediated response. Wild-type animals treated with LPS exhibited a 49% decrease in dye coupling between aortic endothelial cells. For the dye transfer studies, we used biocytin, an uncharged tracer of low molecular weight which transfers well in the aortic endothelium. The available data indicate that biocytin transfer is a good indicator of overall coupling. The use of other tracers, with connexin-specific differences in permeability, might be a way to further differentiate changes in coupling seen following LPS treatment.
The decrease in cell coupling with LPS was more pronounced in Cx37–/– animals, where only Cx40 is present in aortic endothelium, and was associated with a significant decrease in endothelial Cx40 levels in both wild-type and Cx37–/– animals. These results suggest that LPS mediates changes in endothelial cell-to-cell communication, at least in part, through modulation of Cx40 levels. Our studies do not allow conclusions about the mechanism of LPS-mediated Cx40 modulation, which could result from transcriptional, posttranscriptional, or posttranslational regulation or a combination of mechanisms. In addition, determination of the signaling cascade(s) activated by LPS will be required to resolve whether LPS directly or indirectly results in diminished connexin expression and reduced cellular coupling.
To a lesser extent, endothelial Cx37 levels were also reduced after LPS-treatment of wild-type animals. Surprisingly, however, LPS caused an increase in endothelial Cx37 levels in Cx40–/– animals rather than a decrease. This apparent paradox could be explained if LPS exerts its effects on Cx37 in two distinct ways, one of which is unmasked in the absence of Cx40. First, the LPS-induced drop in Cx40 might indirectly influence Cx37 levels in a downward direction because Cx37 is dependent on Cx40 for normal expression levels. We have previously shown that Cx37 levels are substantially reduced in Cx40–/– endothelium . The decrease in Cx37 in the absence of Cx40 occurs via a posttranscriptional mechanism, as mRNA levels are unaffected. We speculate that Cx37 stability is affected by a reduction in Cx40 levels, perhaps because they normally interact in heteromeric channels. Second, LPS might also cause an increase in de novo synthesis of Cx37. In the LPS-treated Cx40–/– mice, the indirect effect mediated by changes in Cx40 levels is eliminated as a contributing factor and a relative increase in Cx37 after LPS treatment is revealed. In wild-type aortas, however, both of these mechanisms: (1) decreased Cx37 stability due to reduced Cx40 levels and (2) increased de novo Cx37 synthesis would be operating, and the net effect of LPS treatment is a small decrease in Cx37 levels.
At present, it is unclear why dye transfer in Cx40–/– endothelium decreased by 26% with LPS treatment when Western blots indicate an increase in Cx37 levels. In general, junctional transfer is not only affected by channel numbers but also by single channel conductance and open time probability. The effect of LPS treatment on Cx37 could therefore be complex, affecting not only Cx37 protein levels but also the functional activity of Cx37-containing channels, possibly via posttranslational modification . If this is the case in mouse aorta, some of the Cx37 expressed in LPS-treated Cx40–/– endothelium may be less than fully active.
In our experiments, we used a model of LPS-induced inflammation. LPS is a key factor in the development of gram-negative sepsis. The release of LPS from invading bacteria into the circulation results in the activation of leukocytes and macrophages, which subsequently release cytokines and NO as inflammatory mediators . Thus, LPS initiates a complex inflammatory cascade and the direct effects of LPS on vascular cell function (and specifically, vascular cell communication) are not completely understood. However, it is becoming increasingly clear that complex signaling pathways are involved in mediating cellular responses to LPS [18, 20, 21, 45, 46, 47, 48]. Recent studies have shown that LPS binds to membrane and soluble forms of CD14, LPS-binding protein (LBP), and Toll-like receptor 4 (TLR-4) . TLR-4 is present on macrophages and endothelial cells and there is strong evidence that this transmembrane receptor is a true LPS receptor [49, 50]. Future studies will require the use of specific signaling agonists and antagonists, as well as cytokines, to elucidate the mechanism of LPS-induced changes in endothelial coupling and connexin expression.
Recent studies indicate that LPS can alter endothelial communication via a mechanism that involves posttranslational modification of Cx43. LPS increases intercellular resistance in rat microvascular endothelial cells without changes in connexin expression and this effect is blocked by tyrosine kinase inhibitors . Subsequent studies demonstrated that Cx43 in rat microvascular cells is tyrosine phosphorylated following exposure to LPS, suggesting that tyrosine phosphorylation of Cx43 might mediate the LPS-induced increase in intercellular resistance . In this model, cytokines typically associated with sepsis, IL-1, IL-6, TNF-α, and interferon-=γ did not affect cellular coupling, suggesting a direct LPS effect [31, 51]. Finally, in vivo studies by Tyml and colleagues [31, 32] showed that LPS reduces arteriolar conducted responses induced by local electrical stimulation and that tyrosine phosphorylation may be involved.
Several reports suggest that proinflammatory mediators are involved in the regulation of gap junctional intercellular communication [33, 34, 52, 53]. Exposure of HUVECs to TNF-α decreases expression of Cx37 and Cx40 without changes in the expression of Cx43, resulting in a decrease in dye coupling of 40% . Alterations in gap junction expression have been noted in a number of different disease states associated with an underlying inflammatory process. In the liver, downregulation of Cx32 mRNA and protein is observed after LPS-induced inflammation [54, 55]. Ischemia, which is associated with upregulation of cytokines such as IL-1β and TNF-α in many tissues, has also been associated with downregulation of gap junctions composed of Cx32 in the liver  and Cx43 in the heart . TNF-α plays an important role in mediating the downregulation of heart Cx43 after administration of LPS, and this response involves transcriptional regulation of the Cx43 gene . However, injection of LPS into kidney or lung resulted in increased expression of Cx43 . IL-1β has been shown to downregulate gap junction communication as well as Cx43 mRNA and protein in human astrocytes in vitro . In addition, IL-1β increased functional expression of Cx43 and potentiated intercellular communication in articular chondrocytes . Finally, activation of human polymorphonuclear cells with LPS or TNF-α, in the presence of endothelial cell-conditioned medium, induced formation of functional gap junctions and expression of Cx40 and Cx43 . These observations support the conclusion that gap junction expression is altered at sites of inflammation in vivo and could thus contribute to pathophysiological responses.
The functional consequences of decreased aortic endothelial cell coupling following LPS-induced inflammation are not entirely clear. Gap junctions enable extensive coupling of endothelial cells in the vascular wall and therefore gap junction channels may serve to coordinate the normal functions of the endothelium. Thus, decreased endothelial cell coupling during inflammation likely contributes to endothelial dysfunction and altered vasomotor responses. The decrease in gap junctional communication in the aortic endothelium during inflammation may therefore be part of a generalized vascular response to injury.
In summary, our results indicate that aortic endothelial cell communication is decreased during LPS-induced inflammation. This reduction in coupling is associated with decreased expression of Cx40 and, to a lesser degree, Cx37. We hypothesize that downregulation of Cx40 in particular plays a major role in the decrease in cell communication following LPS treatment. Previous studies have shown that LPS can reduce endothelial coupling through a mechanism that likely involves posttranslational modification of Cx43 [29, 30]. Our study indicates that a second mechanism by which LPS can reduce endothelial coupling is through a reduction in endothelial Cx40 and Cx37 content.
We thank David Paul for Cx37 and Cx40 antisera. We are grateful to Jan Burt for advice about dye injections and for the use of injection equipment. This work was supported by grants from the Arizona Disease Control Research Commission (10018 to A.M.S.), National Institutes of Health (HL64232 to A.M.S.), and the Heart and Stroke Foundation of Canada (Y.O.).
Dr. A.M. Simon
Department of Physiology
University of Arizona
Arizona, AZ 85724-5051 (USA)
Tel. +1 520 621 9778, Fax +1 520 626 2383, E-Mail email@example.com
Received: August 12, 2003
Accepted after revision: May 19, 2004
Published online: July 7, 2004
Number of Print Pages : 11
Number of Figures : 5, Number of Tables : 2, Number of References : 61
Journal of Vascular Research (Incorporating International Journal of Microcirculation)
Founded 1964 as Angiologica by M. Comèl and L. Laszt (1964–1973) continued as Blood Vessels by J.A. Bevan (1974–1991)
Official Journal of the European Society for Microcirculation
Vol. 41, No. 4, Year 2004 (Cover Date: July-August 2004)
Journal Editor: U. Pohl, Munich
ISSN: 1018–1172 (print), 1423–0135 (Online)
For additional information: http://www.karger.com/jvr