Abstract Dye tracers were chosen, based on net charge, chemical structure, and reactive groups, to test for the existence of and to provide novel insight into channel selectivities of junctional pathways connecting smooth muscle and endothelial cells of the arteriolar wall. Dyes were injected into individual smooth muscle or endothelial cells of hamster cheek pouch arterioles using microiontophoresis. Coupling, independent of tracer net charge, was seen both within and between cell layers. Endothelial cells were well coupled by all of the tested dyes. Smooth muscle junctions appeared less effective in dye transfer than endothelial junctions. Lucifer yellow was confirmed to be a poor tracer of smooth muscle gap junctions, and remarkably this dye and other related sulfate-containing molecules interfered with dye movement through smooth muscle but not endothelial junctions. Myoendothelial junctions showed a striking polarity of dye movement, with dye transfer from endothelial to smooth muscle cells but little or no transfer in the reverse direction. Because the dyes have size and charge characteristics similar to those of known cellular second messengers, these findings have important implications for cell-cell signaling in the vessel wall.
Communication of vascular signals is thought to occur between endothelial and vascular smooth muscle cells via both direct chemical signaling1 2 3 and ion flow through gap junctional channels.4 5 6 Intercellular coupling may provide a means of synchronizing vasoactive7 or growth responses8 and may provide a pathway for second-messenger signaling between adjacent cells.9 10 We have previously shown, through use of immunofluorescence techniques, that at least two different gap junctional proteins can be identified within both arteriolar smooth muscle and endothelial cells.11 The presence of junctional proteins within both cell types potentially allows for three distinct intercellular communication pathways to exist: between adjacent smooth muscle cells, adjacent endothelial cells, and adjacent smooth muscle and endothelial cells. Previous dye studies using Lucifer yellow revealed endothelial cell coupling within arterioles but failed to show either smooth muscle or smooth muscle–endothelial coupling.12 However, electrical coupling has been reported to exist between smooth muscle cells of arterioles5 6 and between endothelial and smooth muscle cells of coronary arteries13 in the absence of detectable dye coupling.
The lack of consistency between electrical and dye-coupling findings has, to date, confounded the understanding of cell-cell communication patterns within arterioles. Thus, it is critical to note that although the anionic dye Lucifer yellow has been commonly used as a standard tracer to assess junctional coupling, in a number of tissues Lucifer yellow fails to diffuse between cells shown to be coupled by other low-molecular-weight dyes.14 15 16 Thus, the possibility exists that previous detection of intercellular coupling patterns within arterioles was limited by the choice of dye tracer used. Furthermore, the observation that certain gap junctions display charge selectivity17 18 19 led us to consider that nonselective junctional coupling might be present between endothelial cells, whereas smooth muscle junctions might be charge selective.
We therefore employed a strategy utilizing diverse tracers (Table 1⇓), including biocytin, a molecule shown to be superior to Lucifer yellow in tracing gap junctions,14 15 that allowed us to test for the presence of the three hypothesized pathways between cells of the intact arteriole as well as to examine charge-selective features of each of the identified pathways.
Materials and Methods
All procedures and protocols in this study were approved by the University of Virginia Animal Care and Use Committee. Male Golden hamsters were anesthetized with pentobarbital sodium (50 mg/kg IP), and the cheek pouches were excised. Individual arterioles, together with a thin sheet of surrounding connective tissue, were dissected and pinned flat. The preparation was superfused with bicarbonate buffer (pH 7.4, 37°C). Probenecid (2 mmol/L; Sigma) was added to the superfusate during injections of negatively charged dyes to inhibit the nonspecific leak of the dyes via anion exchange. Heptanol, when used to uncouple intercellular communication, was added to the superfusate at a final concentration of 1 or 3 mmol/L. We found that 1 mmol/L heptanol blocked dye transfer, and this concentration was used subsequently.
Dyes (Molecular Probes), aniline-2-sulfonic acid (Aldrich Co), and potassium sulfate (Sigma) were stored as frozen stock solutions, which were diluted each day (see Table 1⇑) in 0.1 mol/L KCl (0.1 mol/L LiCl for Lucifer yellow), 0.05 mol/L Tris, pH 7.8. Injection of biocytin by negative current was facilitated by the high pH of the electrode solution, which served to increase the percent of negatively charged molecules. High-molecular-weight anionic FITC- or TRITC-dextran (3000 Da) was coinjected with the low-molecular-weight dyes, providing a marker of the injected cell.
Individual smooth muscle or endothelial cells were loaded with dye using microiontophoresis. Electrode resistance ranged from 300 to 350 MΩ for dye electrodes (80 to 100 MΩ when filled with 3 mol/L KCl). Square wave current pulses (1 Hz, 500 milliseconds) were used (−5 nA for negatively charged dyes, +0.3 nA for positively charged dyes) to inject the dyes. Injection times ranged from 5 to 20 minutes. Times from injection to observation ranged from 5 to 90 minutes. Multiple injections were performed in each arteriole, with each dye tested in a minimum of 3 individual arterioles. Following injections, most arterioles were fixed with 4% paraformaldehyde. Biocytin-labeled arterioles were permeabilized with 0.02% Triton-X-100 and stained overnight with a 1:100 dilution of Neutralite avidin–Texas red (Molecular Probes).
Arterioles were visualized using confocal fluorescence microscopy. The MRC 600 laser scanning confocal microscope (Biorad, Inc) was used with filter sets T1 and T2, allowing simultaneous measurement of fluorescein/Lucifer yellow–wavelength and rhodamine/Texas red–wavelength dyes. Arterioles were oriented horizontally on the microscope slide and optically sectioned. Nikon Fluor ×40 (NA 1.3) and Nikon ×100 (NA 1.3) oil immersion objectives were used, providing optical section thicknesses of 0.70 μm and 0.35 μm, respectively. Previous studies have shown that these conditions allow separate imaging of smooth muscle and endothelial cell layers in whole-mount arterioles.11 Silver nitrate staining of arterioles has shown that endothelial cells, which are oriented parallel with the long axis of the arteriole, can be readily distinguished from smooth muscle cells, which encircle the arteriole, perpendicular to the long vessel axis.20 Images of stained arterioles were processed using Image 1 (Universal Imaging), Picture Publisher, and Designer (Micrografx). Standard epifluorescence video microscopy was used to record diffusion of nonfixable dyes (carboxyfluorescein). A Nikon ×40 (NA 0.55) water immersion objective and fluorescein- and rhodamine-wavelength filter packages were used. A DAGE-MTI SIT 66 intensified camera was used to obtain video images, which were recorded on VHS cassette and analyzed using Image 1.
Charge Selectivity and Junctional Coupling
Adjacent cells in the endothelial cell layer appeared well coupled in each case (100%, n=7 injections) following intracellular injection of biocytin, a zwitterionic molecule that is 98% neutral at pH 7.4 (Fig 1A⇓). The dye diffused extensively both transversely and longitudinally, distributing homogeneously within the cytoplasm of each endothelial cell. Furthermore, in all endothelial injections of biocytin, dye diffused homogeneously into numerous overlying smooth muscle cells (Fig 1B⇓). Smooth muscle cell staining was seen both immediately above and also several hundred micrometers distant from the injected endothelial cell. This movement of dye is in agreement with the hypothesized presence of a direct intercellular pathway linking endothelial cells and smooth muscle cells through myoendothelial junctions and represents the first such observation in an intact arteriole.
Biocytin injected into a single smooth muscle cell diffused from the injected cell into multiple adjacent smooth muscle cells (Fig 1C⇑) in 95% of injections (20 of 21 cells). Dye coupling between smooth muscle cells appeared more heterogeneous than in the endothelium, implying a lower overall level of coupling and suggesting that not all smooth muscle cells are equally well coupled. In contrast to the observations of endothelial cells, biocytin was confined to the smooth muscle layer, with no detectable dye movement into underlying endothelial cells.
Ethidium bromide was used to test for charge selectivity of the junctions. This cationic dye binds with high affinity to DNA, and its fluorescence emission intensity increases greatly upon binding. We reasoned that the high level of localized nuclear fluorescence emitted by ethidium bromide would greatly increase our ability to detect low levels of dye diffusion from smooth muscle into the underlying endothelial cells. Ethidium bromide injected into endothelial cells brightly stained neighboring endothelial cell nuclei, as well as multiple overlying smooth muscle cell nuclei (100%, n=5) (Fig 2A⇓). Ethidium bromide disclosed coupling between smooth muscle cells as effectively as biocytin (91%, 10 of 11 cells) (Fig 2B⇓). Of the 11 smooth muscle cell injections, only 2 showed very weak smooth muscle–endothelial cell coupling. Thus, although ethidium bromide allowed visualization of low levels of coupling from smooth muscle to endothelium, dye movement from endothelium to smooth muscle was much more prominent.
We confirmed previous reports that anionic Lucifer yellow is capable of tracing coupling only between endothelial cells. Following most endothelial cell injections of Lucifer yellow (Fig 3A⇓), dye spread to multiple adjacent cells (75%, 3 of 4 cells) but never diffused into smooth muscle cells. Furthermore, Lucifer yellow injected into smooth muscle cells remained confined to the cell of injection (n=9 cells) regardless of the time provided for injection or diffusion (Fig 3B⇓).
We also injected cells with anionic carboxyfluorescein, which has the same net charge as Lucifer yellow. Endothelial cell injections (100%, n=5) showed coupling between endothelial cells and also from endothelium to smooth muscle (60%, 3 of 5 cells), as seen with biocytin and ethidium bromide. In contrast to results seen with Lucifer yellow, smooth muscle–smooth muscle cell coupling was evident with carboxyfluorescein (94%, 16 of 17 cells) (Fig 3D⇑). Coupling patterns seen with carboxyfluorescein closely resembled those seen with injected biocytin or ethidium bromide. Again, dye was not visible within the underlying endothelium following smooth muscle injection.
Because all dyes except Lucifer yellow demonstrated similar coupling patterns, it appeared likely that in each case cell-cell movement occurred by the same mechanism. Heptanol, a well-established uncoupler of gap junctional communication,21 22 was used to test for involvement of gap junctions in dye spread. Heptanol treatment (1 mmol/L) completely blocked carboxyfluorescein coupling following smooth muscle (n=3) or endothelial (n=3) cell injections (not shown), thus providing strong evidence for involvement of gap junctions in the movement of dye through each of the detected pathways.
Lucifer Yellow and Junctional Coupling
We observed a surprising phenomenon when Lucifer yellow was coinjected into smooth muscle cells with other tracers. Not only did Lucifer yellow fail to diffuse between adjacent smooth muscle cells, it also blocked the diffusion of both biocytin (n=5) and ethidium bromide (n=3) when these dyes were coinjected (Fig 3C⇑). This blocking effect could be minimized by lowering the Lucifer yellow concentration (1% or 0.5%). Lucifer yellow did not, however, block the movement of biocytin to neighboring endothelial cells when the two dyes were coinjected into endothelial cells. Biocytin diffusion into overlying smooth muscle cells was prevented in the area immediately surrounding the injection site. Dye block did not occur with coinjections of tracers other than Lucifer yellow, such as ethidium bromide and carboxyfluorescein (n=4).
Because of the close similarity in molecular weights of all the tracers used, it seems unlikely that the behavior of Lucifer yellow could be explained solely on the basis of molecular size. Charge selectivity can also be discounted, based on our data using anionic carboxyfluorescein. Therefore, we attempted to discover those aspects of the chemical composition of Lucifer yellow that would be sufficient to produce the block of dye transfer between smooth muscle cells.
Brilliant sulfoflavin, a dye with a chemical structure similar to that of Lucifer yellow but lacking the reactive hydrazide group (see Table 1⇑), was used to exclude the simple possibility that intracellular binding of Lucifer yellow had prevented its diffusion. When injected into smooth muscle cells, this dye revealed weak coupling similar to that seen with Lucifer yellow (2 of 9 injections), and in those cases dye spread no further than a single adjacent smooth muscle cell. When brilliant sulfoflavin was coinjected with ethidium bromide (n=3), diffusion of the latter dye was blocked.
We then examined the diffusion of carboxyfluorescein when coinjected with either a nonfluorescent sulfated ring (aniline-2-sulfonic acid) or simply potassium sulfate. Carboxyfluorescein coinjected into endothelial cells with sulfated aniline (n=3) or sulfate (n=3) spread to adjacent endothelial cells, although the dye did not diffuse to smooth muscle cells (Fig 3E⇑). However, in all smooth muscle injections of aniline-2-sulfonic acid (n=5) or sulfate (n=4), the injected cell was brightly stained with carboxyfluorescein, but no movement into adjacent smooth muscle or endothelial cells was seen (Fig 3F⇑). Only with very long diffusion times (1 or 2 hours) could faint movement of carboxyfluorescein into several adjacent smooth muscle cells be detected. Intracellular membrane potential was not altered following injection of either sulfate or aniline-2-sulfonic acid (data not shown).
Our studies provide strong evidence for each of the three hypothesized cell-cell communication pathways. Homocellular coupling exists within the endothelial cell layer, and contrary to previous results reported with Lucifer yellow and in affirmation of electrical recordings, homocellular coupling exists within the smooth muscle cell layer. We have also shown evidence for a novel heterocellular dye coupling between endothelial and smooth muscle cells. This pathway is unique in the sense that it is strongly biased in the direction of dye movement from endothelium to smooth muscle. Finally, we have shown that Lucifer yellow can produce a selective blockade of dye movement either through homocellular or heterocellular smooth muscle junctions.
Coinjections of high-molecular-weight dextran with each tracer dye allowed us to dismiss the possibility that dye transfer occurred as a result of cytoplasmic bridging between cells, because at no time did we see the high-molecular-weight dextran fill cells adjacent to the injected cell. Furthermore, dye transfer was reversibly blocked by heptanol, a putative uncoupler of junctional communication, providing strong evidence for involvement of gap junctions in the movement of dye from cell to cell.
The tracers were selected to reveal any charge selectivities of the detected pathways, since certain gap junctional channels (Cx40, Cx37, Cx45) are known to be cation selective.17 18 19 However, neutral (biocytin), anionic (carboxyfluorescein), and cationic (ethidium bromide) dyes all showed similar extents of coupling in both smooth muscle and endothelial cell layers (see Table 2⇑), arguing against prominent charge selectivities of the channels involved.
Movement of molecules through gap junctions is thought to be directionally symmetrical. Although this was the case within smooth muscle and endothelial cell layers, “polar” coupling from endothelium to smooth muscle was demonstrated by each of the above tracers. These observations raise the provocative idea that small molecules can move from endothelium to smooth muscle, but that diffusion in the opposite direction is limited. A similar degree of polarity existing across gap junctions of two different cell types has been previously reported in a neuronal cell population.15 There are two possible explanations for this phenomenon. The first is that the polarity results from heterotypic gap junction formation within the myoendothelial junctional complex. Cx43 and Cx40 are present in both smooth muscle and endothelium of hamster cheek pouch arterioles,11 and Cx37 has been reported within some vascular endothelial preparations.23 24 25 Localization of particular connexin proteins to smooth muscle versus endothelial sides of the junction could result in the formation of multi-connexin myoendothelial junctions, which results in the nontypical rectifying properties seen, as has been modeled by Loewenstein.26 This model is dependent on the occurrence of a unique site on the “restrictive” (smooth muscle) side of the junction, which interacts with diffusing molecules and creates an asymmetrical free-energy barrier to diffusion of molecules, thus favoring movement from the opposite (endothelial) direction through the channel. The smooth muscle–specific site could be either a component of the junctional pore (a function of the specific connexin amino acid sequence) or a modification of the channel (occurring only within smooth muscle cells).
Second, apparent polarity in dye movement might be explained by differential patterns of dye dilution in endothelium compared with smooth muscle. Although endothelial cells have average lengths of 140 μm, they are very thin and the calculated cell volume is much smaller than that of a smooth muscle cell (length 65 μm).20 The comparatively small volume of endothelial cells may allow dye or second-messenger molecules within the endothelium to diffuse to large numbers of adjacent endothelial cells before the molecules are greatly diluted. Therefore, many endothelial cells could provide a source for dye movement into overlying smooth muscle cells. On the other hand, the larger cell volumes of smooth muscle cells would result in rapid dilution of molecules into only several adjacent smooth muscle cells. The diffusion of molecules from smooth muscle to endothelium would then be highly dependent on the frequency of myoendothelial connections and thus might result in an apparent polarity in the movement of molecules between the two cell layers.
A volume differential between the two cell types might contribute to the observed polarity, but it does not account for the behavior of Lucifer yellow in smooth muscle as compared with endothelium, strongly suggesting that a difference in the junctions comprising each cell type is the cause of the directional coupling. Although the failure of Lucifer yellow to pass through certain gap junctions has been reported previously in several different cell systems,14 15 16 a mechanistic explanation for the poor dye passage is lacking. Through use of structurally similar dyes, we have shown that a specific structural component of Lucifer yellow interferes with the movement of dye through smooth muscle gap junctions but not through endothelial junctions. The blocking characteristic of Lucifer yellow could be mimicked by the structurally similar fluorescent dye brilliant sulfoflavin, by nonfluorescent aniline-5-sulfonic acid, and by sulfate itself. Because of the apparently specific action of sulfate on smooth muscle–endothelial and smooth muscle–smooth muscle junctions, with no detectable effect on membrane potential, sulfate may present a powerful tool for manipulating the blockade of smooth muscle junctional pathways while leaving the endothelial pathway intact.
The patterns of movement of Lucifer yellow and other sulfate-containing molecules are consistent with the behavior of a molecule that either interacts directly with junctional proteins or causes junctional uncoupling. Although gap junctional channels have been extensively modeled as large nonspecific aqueous pores, recent data from electrophysiological studies suggest that fixed charges within the pore may strongly influence the movement of charged molecules through the channel.18 19 Binding of Lucifer yellow and other sulfated compounds to sites in the channel would explain the low incidence of Lucifer yellow passage through the junctions, as well as the blockade of transfer of coinjected dyes. It is also conceivable that these molecules interact with a cytoplasmic site on the connexin protein, resulting in channel gating to a lower conductance state or perhaps complete channel closure. A third explanation would be a sulfate-specific activation of a regulatory second-messenger cascade within the smooth muscle cells, resulting in channel gating or closure. Clearly, it must also be proposed that endothelial cells lack the same mechanisms. At this time, the available molecular data are insufficient to either refute or support any one model.
The direct applicability of dye movement to that of ion movement remains to be shown, as limits in dye movement may not necessarily represent limits in current flow. However, the dye tracers do mimic, in size and charge characteristics, common second-messenger molecules and thus allow us to make inferences about potential signaling pathways within the arteriole. Many cytoplasmic molecules that have been implicated in signaling pathways (Ca2+, IP3, cAMP) are of a size compatible with gap junctional transfer. Ca2+ (40 Da) and/or IP3 (420 Da) diffusion between cells has been reported in a number of cultured cell systems,9 10 27 potentially mediating and synchronizing a host of vasoactive and growth-related responses involving large numbers of cells.
Lucifer yellow and similar molecules define a clear distinction in the junctions found between smooth muscle cells compared with those found between endothelial cells. Our data provide novel evidence for heterocellular dye coupling between endothelial and smooth muscle cells. The high level of coupling within the endothelial layer, as well as the demonstrated existence of direct intercellular communication from endothelium to smooth muscle, suggests a dominant role for the endothelium in integrating and coordinating vascular, blood-borne signals, which can then be passed to large numbers of overlying smooth muscle cells. The limited capacity for hydrophilic molecules to diffuse from smooth muscle to endothelial cells may imply a hierarchy of signal transduction within the cells of the arteriolar wall.
This work was supported by National Institutes of Health grants HL-19242, HL-12792, and American Heart Association, Virginia affiliate, VA-94-F-24 (J.X.). We thank J.D. Lechleiter and the Markey Center for the use of the Biorad 600 confocal imaging system, and D.N. Damon for excellent technical assistance.
Presented in part at the annual Experimental Biology meeting, Anaheim, Calif, April 24-28, 1994.
- Received November 23, 1994.
- Accepted December 22, 1994.
- © 1995 American Heart Association, Inc.
Busse R, Fleming I, Hecker M. Signal transduction in endothelium-dependent vasodilatation. Eur Heart J. 1993;14(suppl I):2-9.
Beny J-L. Homo- and heterocellular dye and electrical coupling between endothelial and smooth muscle cells in pig coronary artery. In: Rubanyi GM, Vanhouette PM, eds. Endothelium-derived Relaxing Factors. Basel, Switzerland: S Karger; 1990:117-123.
Beny J-L. Endothelial and smooth muscle cells hyperpolarized by bradykinin are not dye coupled. Am J Physiol. 1990;258:H836-H841.
Duling B, Matsuki T, Segal S. Conduction in the resistance-vessel wall: contributions to vasomotor tone and vascular communication. In: Bevan JA, ed. The Resistance Vasculature. Totowa, NJ: Humana Press; 1991:193-215.
Christ GJ, Moreno AP, Parker ME, Gondre CM, Valcic M, Melman A, Spray DC. Intercellular communication through gap junctions: a potential role in pharmacomechanical coupling and syncytial tissue contraction in vascular smooth muscle isolated from the human corpus cavernosum. Life Sci. 1991;49:PL195-PL200.
Boitano S, Dirksen E, Sanderson M. Intercellular propagation of calcium waves mediated by inositol trisphosphate. Science. 1992;258:292-295.
Christ G, Moreno A, Melman A, Spray D. Gap junction-mediated intercellular diffusion of cultured human corporal smooth muscle cells. Am J Physiol. 1992;263:C373-C383.
Little TL, Beyer EC, Duling BR. Connexin43 and connexin40 gap junctional proteins are present in both arteriolar smooth muscle and endothelium in vivo. Am J Physiol. 1995;268:H729-H739.
Segal S, Beny J-L. Intracellular recording and dye transfer in arterioles during blood flow control. Am J Physiol. 1992;263:H1-H7.
Beny J-L, Pacicca C. Bidirectional electrical communication between smooth muscle and endothelial cells in the pig coronary artery. Am J Physiol. 1994;266:H1465-H1472.
Robinson SR, Hampson ECGM, Munro MN, Vaney DI. Unidirectional coupling of gap junctions between neuroglia. Science. 1993;262:1072-1074.
Veenstra RD, Beblo DA, Wang HZ, Westphale EM, Beyer EC. Unique ionic conductance and selectivity of connexin40 channels, a major gap junction protein in conducting tissues. Circulation. 1993;88:175. Abstract.
Veenstra RD, Wang HZ, Beyer EC, Brink PR. Selective dye and ionic permeability of gap junction channels formed by connexin45. Circ Res. 1994;75:483-490.
Little TL, Duling BR. Dimensions favor a preferential role for endothelium over smooth muscle in longitudinal conduction of vasomotor responses. FASEB J. 1994;8(5):M101. Abstract.
Spray D, Burt J. Structure-activity relations of the cardiac gap junction channel. Am J Physiol. 1990;258:195-205.
Moreno AP, Campos de Carvalho AC, Christ G, Melman A, Spray DC. Gap junctional communication between human corpus cavernosum smooth muscle cells: gating properties and unitary conductance. Am J Physiol. 1993;264:C80-C92.
Reed J, Westphale E, Larson D, Wang HZ, Veenstra R, Beyer E. Molecular cloning and functional expression of human connexin37, an endothelial cell gap junction protein. J Clin Invest. 1993;91: 997-1004.
Bruzzone R, Haefliger JA, Gimlich R, Paul D. Connexin40, a component of gap junctions in vascular endothelium, is restricted in its ability to interact with other connexins. Mol Biol Cell. 1993;4:7-20.
Gabriels JE, Paul DE. Characterization of connexin expression in rat aortic endothelial cells during in vitro wounding. Mol Biol Cell. 1993;4:329a. Abstract.
Loewenstein WR. Junctional intercellular communication: the cell-to-cell membrane channel. Physiol Rev. 1981;61:829-913.
Charles A, Naus CC, Zhu D, Kidder G, Dirksen E. Intercellular calcium signaling via gap junctions in glioma cells. J Cell Biol. 1992;118:195-201.