Coordinated behavior within arterioles is necessary for large resistance changes to occur and is reflected as a conduction of dilations/constrictions along arterioles. These responses arise from locally initiated hyper- or depolarizations that propagate via transmembrane channels formed by connexins (gap junctions). Mounting evidence indicates that gap-junctional communication contributes to the control of vascular tone.
Communication between cells within an organ is necessary to coordinate the behavior of individual cells and to ensure proper organ function. An important communication pathway is direct signaling through the transmembrane channels known as gap junctions. They are formed by individual structural units known as connexins and allow transfer of ions and second messengers. By this means signaling between individual cells is established, and additionally a large number of cells can be connected to form a functional unit. Such a signaling pathway may also exist in the vascular system, because endothelial and vascular smooth muscle cells express connexin proteins. In principle, gap junctions can provide coupling between endothelial and smooth muscle cells (heterocellular) or among each cell type separately (homocellular). Functionally, two main aspects of signal transfer can be envisioned: a transversal or short-distance and a longitudinal or long-distance communication pathway. Transversal signaling involves communication between endothelium and smooth muscle and is therefore dependent on heterocellular coupling. The endothelium could affect smooth muscle tone by transfer of autacoids, second messengers, or current and thus control vascular diameter via such a pathway. In contrast, longitudinal signal transfer serves to coordinate the cellular behavior within a certain vessel segment. For this purpose, signaling within each cell layer is additionally required and must therefore also involve homocellular coupling. Indeed, experimental evidence exists for gap-junctional involvement in both of these signaling pathways and is the focus of this brief review.
Information travels along the arteriolar wall
Within the microcirculation the arterioles form a complicated network where the arteriolar tree spans a long distance of several millimeters. Due to their length and concomitant small diameters, a large resistance resides within these arterioles, leading to large pressure drops along them. Although this is appropriate at rest, conditions with high metabolic demand (e.g., exercising skeletal muscle) require that resistance decreases substantially by adequate arteriolar dilation to allow high blood flow to the capillaries. This dilation can be achieved in vessels close to the exercising tissue by the diffusion of metabolites released from skeletal muscle fibers, but it is not necessarily the case for vessels located further upstream, because diffusion distances increase considerably and an upstream dilation cannot be accomplished solely by metabolic dilation. Consequently, resistance residing within these upstream vessels becomes proportionally larger with higher flow loads and ultimately limits flow increases. To overcome such a mismatch of resistance distribution along the distance spanned by the arteriolar tree, a coordination is necessary to promote upstream dilation along with distal dilation.
The observation that peripheral metabolic changes within skeletal muscle induce an ascending, remote dilation has been known for decades. It has been attributed to axon reflexes, nerves associated with microvessels, transport of metabolites via venous drainage from capillaries to upstream vessels, or, more recently, flow-induced dilation. However, in the past few years it has become clear that a conducting mechanism residing within the vascular wall, as originally proposed by Hilton (8), exists and may be involved. Such a longitudinal conduction mechanism has been shown by the application of endogenous neurotransmitters at a small, strictly confined area onto isolated microvessels with glass micropipettes. In such experiments, acetylcholine or noradrenaline not only induced a local vasomotor response but also a response of a similar type (dilation, constriction) at distant, remote sites. In a similar fashion, remote responses can be elicited in the microcirculation in vivo (Fig. 1⇓) (1, 4). The remote responses, also termed “conducted responses,” were not due to changes in pressure or flow. The latter stimulus was excluded because the dilation occurred before any increases in wall shear rate were measured (9, 16). Moreover, flow-induced dilation takes ~10–30 s to develop fully; however, the delay at which conducted dilations appear is <1 s. Signaling via nerves could achieve such transmission speeds, but blockade of nerve conduction with tetrodotoxin did not attenuate conducted responses. Therefore, these initial experiments demonstrate that the mechanisms evoking conducted responses are not related to known pathways within the vascular wall.
Conducted responses have been observed in skeletal muscle of hamsters and mice (cremaster muscle, cheek pouch, retractor muscle) and also in kidney, cerebral, and coronary arterioles. In addition to acetylcholine, other endogenous substances such as bradykinin and adenosine also initiate conducted dilations (4, 15). However, other vasodilators (such as NO donors, e.g., sodium nitroprusside, 3-morpholinosydnonimine) only induce a dilation constrained to the site of application (9). This demonstrates that a local dilation per se is not sufficient to elicit a concomitant remote response and raises the question of the nature of the initiating mechanism to elicit a conducted vasomotor response. Considering the high speed of propagation of the diameter change, which is in the range of several millimeters per second, it seems likely that an electrical signal mediates the response. Accordingly, those substances known to induce a conducted dilation also induce a hyperpolarization of smooth muscle and/or endothelial cells. We have shown previously (9) that in the hamster cremaster microcirculation the dilation on acetylcholine application is partially elicited by an endothelium-derived hyperpolarizing factor (EDHF), which induces a hyperpolarization via the activation of Ca2+-dependent K+ channels. To determine which site of activation of these channels is necessary, K+ channel blockers (charybdotoxin or iberiotoxin) were applied either at the site of acetylcholine stimulation or at remote sites. When administered at the site of acetylcholine stimulation, the local and the remote responses were attenuated. In contrast, similar application at remote sites did not affect the remote or local dilation (9). This demonstrates that activation of Ca2+-dependent K+ channels, presumably by EDHF, is necessary to initiate a conducted response but is not required at remote sites to mediate the dilation. Similarly, the blockade of the synthesis of NO or prostaglandins did not affect this remote dilation. Analogous observations were made in other tissues: blockade of NO synthase had only a marginal effect on conducted dilations in hamster cheek pouch, but the responses were severely attenuated after blockade of EDHF. These findings suggest that a locally induced change in membrane potential, elicited by the local action of EDHF, spreads along the vascular wall, inducing a dilation at distant sites itself. In fact, a hyperpolarization of endothelial and vascular smooth muscle cells on confined stimulation with acetylcholine was not only measured at the local but also at remote sites in isolated arterioles as well as in vivo (6, 20). Thus convincing evidence has been presented that EDHF (or a hyperpolarization by other means) is crucial to initiate a conducted response (Fig. 2⇓). In a similar manner constrictions can be elicited by a depolarization, which also propagates along the vascular wall. The initial depolarization can be experimentally induced by the application of high-K+ solution or by addition of vasoconstrictors, e.g., norepinephrine. If sufficient in amplitude, changes in membrane potential give rise to vasomotor responses at local and remote sites: depolarization resulting in constriction and hyperpolarization leading to vasodilation. Although these signals are opposite in polarity, they both reflect the conduction along the vascular wall that is possible because the cells are interconnected by low-resistance channels, i.e., gap junctions (Fig. 2⇓).
An important issue in this context is how far the signals are transmitted. An electrical signal can spread in a purely electrotonic fashion or might be regenerated by an amplifier. A signal that spreads only electrotonically dissipates with distance, and a length constant can be calculated, usually given as the distance at which the signal has decreased to 37% of its initial value. This length constant is dependent on intercellular resistance (i.e., gap-junctional conductance) and on the loss of charge via the cell membrane (short-circuit current). It will be higher when the intercellular resistance is lower and the cell membrane resistance is higher. From data obtained in vitro, a length constant of 0.7 mm was calculated for changes in membrane potential. However, potential and diameter changes in vivo were found to be only slightly attenuated at distances exceeding 1 mm, especially in the case of hyperpolarizations and concomitant dilations (6). This suggests that a regenerative, amplifying mechanism is involved in vivo that is activated by the initial signal. It may consist of voltage-gated or inward-rectifier K+ channels, which can amplify the initial small hyperpolarization and enhance the local change in membrane potential and thus the concomitant diameter change. In coronary arterioles of the pig, blockade of inward-rectifier K+ channels with barium attenuated remote dilations on adenosine application, suggesting that the distance encompassed by the signal was related to signal amplification by these K+ channels (15).
Signaling pathways in the vessel wall
The spread of de- or hyperpolarizations along the vascular wall requires longitudinally interconnected cells. Because gap junctions are found among endothelial as well as smooth muscle cells, each cell layer may provide a conduction pathway. In experiments in hamster vessels, two independent pathways have been proposed, as elegantly demonstrated by destroying the smooth muscle and/or the endothelial cell layer in a confined region. In these in vivo experiments, destruction of the smooth muscle cell layer prevented the transmission of constrictions through this area. In contrast, conducted dilations were still propagating freely, demonstrating that the endothelial layer sufficed as a signal pathway for dilations. Only additional abolition of the endothelial cell layer prevented the propagation of dilation beyond the injured part of the vessel (1). On the other hand, the sole destruction of the endothelial cell layer did not prevent remote dilations. These experiments show that both the endothelial and smooth muscle cell layer itself can transmit the hyperpolarization initiated by acetylcholine (Fig. 2⇑). Slightly different conclusions were obtained in vessels residing outside the skeletal muscle where, in vitro, the endothelial cell layer was crucial to transmit dilations to conducted sites. In contrast to the in vivo preparation, constrictions did not propagate, suggesting that smooth muscle cells are coupled poorly in feed arteries (6). The reasons for this varying behavior in vessels with different functions may be due to different innervation patterns or differential expression of connexins in the vascular wall (see below). Moreover, the varying behavior might be achieved by regulatory processes interacting with gap-junctional conductivity, which can be affected by the type of preparation studied (e.g., in vivo vs. in vitro or arterioles vs. feed arteries). However, these questions must be addressed in further studies. One of the first tasks to resolve is the expression level of connexins in different vessel types. If differentially expressed, the challenging question will be what governs their expression level. However, it has to be kept in mind that function of gap junctions is important to allow signal propagation and that channel function might vary even with similar expression levels.
Connexins are involved in information transfer
Gap junctional communication requires the assembly of intercellular channels that span the intercellular gap and two plasma membranes (17). Six of the individual structural units known as connexin proteins oligomerize into a connexon, delineating a central pore and building an hemichannel within the membrane of a single cell. It docks with a counterpart in an adjacent cell to establish a functional pore (Fig. 3⇓). Connexins consist of a large family of different members, which are distinguished by their molecular weight and named accordingly. In vascular cells four connexins have been identified, namely Cx43, Cx40, Cx37, and Cx45. A connexon may contain either a single type of connexin (homomeric) or multiple connexins (heteromeric). Additionally, adjacent cells can contribute identically or differentially composed connexins, thus forming homotypic or heterotypic intercellular channels. Although only four different connexins are expressed in the vascular wall, the number of structurally and physiologically distinct channel types is thus large. However, the significance of this molecular diversity is not yet clear. In large conduit arteries, connexin proteins are differentially expressed in different cell types of the vascular wall. Whereas Cx43 is the most abundant within the vascular smooth muscle, Cx40 is mainly found in the endothelium. Cx37 is also mainly expressed in endothelial cells, although quantitatively in a smaller amount than Cx40 (21). Cx45 is present in vascular smooth muscle but is expressed only in small quantities. Whereas the expression of different connexins is relatively easily studied in large arteries, this task is more difficult in the microcirculation. Certainly gap junctions exist in the microcirculation; however, only a few studies have addressed the distribution of different connexins in these vessels. From these studies, it is clear that Cx43, Cx40, and Cx37 are found in the microcirculation (12), although it is unresolved at which locations or in which cells. Moreover, functional data from our laboratory show that specific connexins serve specific functions (4). The lack of Cx40 resulted in an attenuation of the conduction of dilations initiated by endothelium-dependent dilators (acetylcholine, bradykinin). In contrast, the conduction of constrictions induced by K+ depolarization remained unaffected. Recently, an attenuated propagation of a dilation (but not of a constriction) induced by electrical stimulation was confirmed in Cx40-deficient animals by other investigators (7). This divergent effect of the lack of Cx40 and the differential distances that constrictions or dilations propagated along the vascular wall (Fig. 1⇑) suggests that different cell layers serve as distinct conduction pathways. The preferential expression of Cx40 in these cremaster arterioles within the endothelial layer, as demonstrated by a colocalization of Cx40 and an endothelial cell marker in immunohistochemistry studies (4), further corroborates the hypothesis that Cx40 is mainly involved in signaling along the endothelial cell layer (Fig. 2⇑). Despite the fact that other members of the connexin family may be upregulated, as was demonstrated in the aorta for Cx37 in Cx40-deficient animals (10), the function of Cx40 cannot be completely rescued. However, in Cx40-deficient mice a remaining propagation of dilations in response to acetylcholine was also observed. If this remaining response is transmitted via the endothelial cell layer (e.g., by the upregulation of other connexins) or if this reflects the spread along the smooth muscle cell layer remains to be elucidated.
Functional implications of coupling in the vascular wall
The fact that conducted responses can be elicited by certain stimuli does not imply that they are also involved in physiologically important regulatory mechanisms, such as flow increases during high metabolic demand (active hyperemia). However, capillaries could serve as sensors for metabolic needs, because they are located in close contact with the tissue and might transmit this information through cell coupling to upstream vessels. In fact, capillary endothelial cells respond to pharmacological stimulation (as explained above) with changes in membrane potential that lead to changes of tone in upstream arterioles (14). This hypothesis was tested further in the hamster cremaster, in which a group of skeletal muscle fibers can be stimulated electrically and diameter changes of the supplying arterioles along the vascular tree can be measured during and after such an exercise bout. This contraction of muscle fibers induced a graded, frequency-dependent dilatory response in upstream arterioles. Interestingly, the dilations were reduced by local application of inhibitors of gap junctions at a site along the arteriole, such as halothane or a high-molar sucrose solution (3). During direct electrical stimulation of muscle fibers acetylcholine is not released at the neuromuscular junction, and this may have an additional effect. However, it can be released by direct stimulation of the motor nerve. Experiments conducted in this way show a dilation that ascended up to the feed artery independent of skeletal muscle contraction (blocked by a nicotinic receptor antagonist). The dilation was abrogated by the additional blockade of endothelial muscarinic receptors and amplified by the inhibition of acetylcholine breakdown. In addition, a contribution of acetylcholine acting on endothelial receptors to the hyperemic response was confirmed by an attenuation of the dilatory response during unhindered skeletal muscle contraction in the presence of a muscarinic receptor blocker. These experiments suggest that spillover of the neurotransmitter acetylcholine contributes to an ascending dilation (19). In further experiments, the authors could show that the endothelial cell layer transmits the signal to induce feed artery dilation. Thus gap-junctional coupling seems to contribute to hyperemic responses. First, acetylcholine spillover induces ascending dilations, and second, metabolically released mediators might change the membrane potential of capillary endothelial cells. This signal may then propagate upstream and be integrated at the arteriolar level.
Transversal signaling in the vessel wall
A different aspect of gap-junctional communication involves local signal transfer from endothelial cells to their adjacent neighbors the smooth muscle cells or vice versa. Two key questions arise: are the two cell layers indeed functionally coupled, and what is transferred through these gap junctions? In morphological studies, heterocellular junctions between endothelial and smooth muscle cells have been demonstrated and were typically found on projections arising from endothelial cells. Compared with gap junctions within the endothelium, the myoendothelial junctions are very small and not numerous, which explains the difficulty in finding them. They have been identified in selected vascular beds (e.g., mesenteric arteries) but not in others (e.g., femoral arteries). This variability also exists between different vessel sizes and necessitates the need for more functional data.
One functional aspect of heterocellular coupling is charge transfer, which can be assessed by the measurement of membrane potential. In isolated arteries of the hamster, hyperpolarizations induced by acetylcholine were recorded in endothelial and smooth muscle cells, and these recordings were indistinguishable from each other (6). Similarly, in mesenteric arterioles in vitro, current injected into an endothelial cell changed the membrane potential of smooth muscle cells and vice versa, which shows bidirectional current transfer (20). In contrast, in in vivo experiments changes in membrane potential in response to acetylcholine as well as spontaneous fluctuations in membrane potential varied between the two cell layers in the microcirculation, suggesting that potentials are not completely transferred. We have made similar observations in mice in vivo (unpublished observations). Whereas the measurement of membrane potential tests for gap junction conductivity, diffusion of dye after injection into a single cell examines permeability. However, data obtained by this approach are in accordance with potential measurements, because dye coupling between endothelial and smooth muscle cells has been shown in arterioles in isolated tissues (13) but not in vivo. An intact dye coupling would suggest that second messengers or small molecules may diffuse through these heterocellular junctions from endothelium to smooth muscle. This may be specifically important for endothelial cell signals, because the endothelium releases different autacoids such as NO that affect the constriction level of smooth muscle and hence vascular tone. Although NO diffuses easily through membrane barriers, other autacoids such as EDHF may not. It is beyond the scope of this review to discuss the nature of EDHF; however, there are studies showing that gap-junctional communication between endothelium and smooth muscle is crucial to achieve the full dilator potency of this NO- and prostaglandin-independent dilator mechanism. In isolated vessels, the portion of the acetylcholine-induced dilation that is attributed to EDHF was severely reduced by inhibition of gap-junctional communication (2). This suggests that “something” is transferred from the endothelium to smooth muscle through heterocellular gap junction channels, which could be an endothelial dilator substance or pure charge. Together these observations propose that heterocellular coupling varies between vessel types (and is related to function) or that differences are related to the type of preparation studied, i.e., in vivo vs. in vitro. If the latter is true, it is interesting to speculate what modulator of gap-junctional communication prevents heterocellular coupling in vivo.
What can be learned from connexin-deficient animals?
The deletion of connexin proteins has allowed additional insights into their functions in the vasculature. However, not all connexin-deficient mice are viable. Cx43- and Cx45-deficient mice die due to defects in cardiac development, whereas Cx40- and Cx37-deficient animals survive. In Cx37-deficient mice an obvious vascular phenotype was not detected, whereas Cx40-deficient mice exhibit an altered conduction of dilations on endothelial stimulation. Remote dilations to local stimulation with acetylcholine or bradykinin were attenuated. In contrast to dilations, the propagation of locally induced constrictions remained unaffected (4). Because the stimulus targets endothelial cells in the first case and the smooth muscle in the second, we concluded that Cx40 is of functional importance in the conduction of signals along the endothelial layer (Fig. 2⇑). Interestingly, Cx40-deficient mice were hypertensive without a change in heart rate, as observed in anesthetized as well as conscious animals of both genders (Fig. 4⇓) (5). The arterial hypertension was not due to an alteration of NO efficacy or release. In addition, pressure drops on injection of acetylcholine were unaffected, suggesting that the vasodilator potency of endothelial autacoids was intact. However, we observed irregular vasomotion patterns that consisted of spontaneously appearing confined constrictions that propagated to downstream sites and led eventually to complete arteriolar occlusion and intermittent flow stop (5). Although we do not yet know the cause of the observed hypertension, our results highlight the importance of cell coupling within the vasculature, especially within the endothelial cell layer where Cx40 is mainly expressed.
Aside from Cx40, Cx43 is also expressed within the endothelial cell layer. However, in contrast to Cx40, the specific loss of Cx43 in endothelial cells was reportedly associated with a lowered, rather than increased, arterial pressure in mice. Since the hypotension was associated with elevated levels of NO derivates, Liao et al. (11) concluded that enhanced NO release was causing the lowered pressure. However, in other mice with endothelial deletion of Cx43 generated in a similar manner (Cre recombinase under the control of a TIE2 promoter), we did not find alterations in hemodynamics, i.e., arterial pressure and heart rate in these animals were similar to littermate controls (18). The reasons for these discrepancies remain unresolved; however, it is clear that the divergent effects of the lack of Cx40 as opposed to the loss of endothelial Cx43 point to a specific role of the individual members of the connexin family in vascular control. Although transgenic mice have disadvantages such as secondary developmental changes, mouse models of connexin deficiency have been instrumental in revealing an important function of gap junctions within the vascular system.
Conclusions and perspectives
The architecture of the arteriolar network necessitates a coordination of cellular behavior. A dilation of terminal arterioles results only in limited flow increases, because resistance residing in further-upstream vessels impedes stronger augmentations of flow. Only when dilations encompass the entire length of the network can maximal flow increases be achieved. Signals traveling along the vascular wall can accomplish such a coordination. However, considerable distances have to be overcome and coordination would be slow if it relied on diffusion of second messengers. In contrast, electrotonic spread of local membrane potential changes propagate rapidly along the vascular wall and induce remote diameter changes. A substantial amount of data demonstrates that changes in membrane potential give rise to vasomotor responses at local and remote sites: depolarization resulting in constriction and hyperpolarization resulting in dilation. Although these signals are opposite in polarity, they both spread along the vascular wall due to the interconnection of cells via the gap junctions. In principle, both cell types within the vascular wall (endothelial and smooth muscle cells) are able to act as a communicating syncytium. Initial data indicate that such communication contributes to physiological responses such as active hyperemia. The structural units of gap junctions are connexins, and four members of this family have been found in vascular cells (Cx43, Cx40, Cx37, and Cx45). The different connexins have distinct functions, as reflected by their differential expression in the cells of the vascular wall. Cx40 appears to be important in signaling along the endothelial layer, and its loss leads to arterial hypertension. Besides longitudinal signaling, gap junctions may have a role in transversal signaling, i.e., from endothelial cells to the smooth muscle layer. Although substantial differences between vessel types and in vivo vs. in vitro experiments have been reported, a significant role of gap-junctional communication along the vascular wall is becoming ever more clear.
Our work referred to in this article was supported by the Deutsche Forschungsgemeinschaft (WI 2071/1-1) and the Friedrich-Baur Stiftung.
I apologize to all authors whose work is not acknowledged due to the lack of space.
- © 2004 Int. Union Physiol. Sci./Am.Physiol. Soc.