Signaling across Myoendothelial Gap Junctions—Fact or fiction?

Abstract

Gap junctions interconnect vascular cells homocellularly, thereby allowing the spread of signals along the vessel wall, which serve to coordinate vessel behavior. In addition, gap junctions provide heterocellular coupling between endothelial and vascular smooth muscle cells, creating so-called myoendothelial gap junctions (MEGJs). Endothelial cells control vascular tone by the release of factors that relax vascular smooth muscle. Endothelial factors include nitric oxide, prostaglandins, and an additional dilator principle, which acts by smooth muscle hyperpolarization and is therefore named endothelium-derived hyperpolarizing factor (EDHF). Whether this principle indeed relies on a factor or on intact MEGJs, which allow direct current transfer from endothelial to smooth muscle cells, has recently been questioned. Careful studies revealed the presence of vascular cell projections that make contact through the internal elastic lamina, exhibit the typical GJ morphology, and express connexins in many vessels. The functional study of the physiological role of MEGJs is confined by the difficulty of selectively blocking these channels. However, in different vessels studied in vitro, the dilation related to EDHF was sensitive to experimental interventions that block MEGJs more or less specifically. Additionally, bidirectional electrical coupling between endothelial and smooth muscle cells was demonstrated in isolated small vessels. In marked contrast, similar approaches used in conjunction with intravital microscopy, which allows examination of vascular behavior in the intact animal, did not verify electrical or dye-coupling in different models investigated. The discrepancy between in vitro and in vivo investigations may be due to size and origin of the vessels studied using these distinct experimental approaches. Additionally, MEGJ coupling is possibly tightly controlled in vivo by yet unknown mechanisms that prevent unrestricted direct signaling between endothelial and smooth muscle cells.

• arterioles
• connexin
• endothelium
• gap junctions
• myoendothelial coupling

GAP JUNCTIONS IN THE VESSEL WALL

In various organs connexins supply important communication channels between adjacent cells that allow transfer of signaling molecules or current in order to coordinate cellular activity. A similar role for connexin proteins has been demonstrated in the vasculature where gap junctions interconnect endothelial cells (ECs) and vascular smooth muscle cells (VSMCs). By coupling neighboring ECs to a functional syncitium, connexins provide the molecular basis of ascending dilations (also termed conducted dilations) that are required for substantial increases of blood flow, e.g., during exercise (reviewed in Segal 2005; de Wit 2004). Likewise, VSMCs are interconnected through connexins, which is a prerequisite for spontaneous rhythmic diameter changes in arteries (vasomotion) (Haddock and Hill 2005; Matchkov et al. 2007). However, experiments and simulations using computational approaches suggest that signals travel less far along the smooth muscle syncitium as opposed to the endothelial cell layer (Diep et al. 2005; Yamamoto et al. 2001).

In contrast to homocellular coupling, the role of heterocellular coupling between ECs and VSMCs is less well accepted and a matter of controversy. However, signaling through gap junctions from ECs to VSMCs may add another powerful endothelial mechanism to control VSM contraction status and thereby vascular tone in addition to the release of autacoids such as nitric oxide and prostaglandins (Dudzinski et al. 2006; Pohl et al. 1993; Wadsworth et al. 2006). In this context, the existence of the most recently suggested endothelial factor, which acts through a hyperpolarization of the smooth muscle and is therefore termed endothelium-derived hyperpolarizing factor (EDHF), has been questioned. It was proposed that the hyperpolarization may likewise be transmitted by direct current transfer through gap junctions from ECs to VSMCs without the need for a chemical factor (Griffith 2004; Sandow 2004). Consequently, the hyperpolarization of VSMCs upon endothelial stimulation has also been termed endothelium dependent hyperpolarization (EDH). It is important to note that this hyperpolarization is independent of nitric oxide (NO) and prostaglandins because both factors themselve may also initiate a hyperpolarization of the VSMC. In this review, we will discuss the significance of heterocelluar coupling through myoendothelial gap junctions (MEGJs) in the vessel wall. We will outline experimental evidence arguing in favor or against a physiological role of signals passing through MEGJs in the control of vessel tone. More extensive reviews on different EDHFs are provided elsewhere (Ahluwalia and Hobbs 2005; Busse et al. 2002; Campbell and Falck 2007; Feletou and Vanhoutte 2006; Fleming 2004; Griffith 2004; Sandow 2004; de Wit et al. 2006; de Wit and Wolfle 2007).

MORPHOLOGICAL EVIDENCE FOR THE PRESENCE OF GAP JUNCTIONS BETWEEN ECs AND VSMCs

Gap junctions (GJs) are composed of connexin proteins, of which six molecules assemble to form a hexameric hemichannel. Two hexamers of adjacent cells link together by end-to-end docking and form a cytoplasmic bridge, which is tightly sealed against the surrounding space to exclude exchange of substances with the extracellular milieu (Unger et al. 1999; Yeager et al. 1998). The dodecameric channel allows the transfer of small molecules (<1 kDa) and ions, thereby permitting current transfer between the interconnected cells. Hundreds of such channels are located at GJs, which can be visualized by electron microscopy as an area of close contact between adjoining plasma membranes separated only 2 nm apart from each other (Barr et al. 1965; Goodenough et al. 1996; Saez et al. 2003). Thus, GJs appear at high magnifications as an electron-dense pentalaminar area and can be identified easily between adjacent ECs. The effortless detection of GJs at this location is mostly due to the large abundance of GJs and the large size of the plaques that interconnect ECs, e.g., in large arteries like aorta, pulmonal artery, coronary and mesenteric arteries (Dora et al. 2003; Ko et al. 1999; Rhodin 1980; Yeh et al. 1997; Yeh et al. 1998), as well as in small arteries and arterioles (Sandow et al. 2003b; Sandow and Hill 2000). Although GJs are found to a lesser extent between VSMCs (Beny and Connat 1992; Sosa-Melgarejo et al. 1991), possibly due to the smaller plaque size and the limited number of GJs, they have also been unequivocally demonstrated here by electron microscopy in large and small arteries as well as arterioles (Sandow et al. 2002, 2003b). However, channel density may be too low to be morphologically detected as a plaque in spite of intact functional coupling. Expression of connexins was also demonstrated by immunohistochemistry in ECs and VSMCs in differently sized arteries (Bastide et al. 1993; Bruzzone et al. 1993) and arterioles (Little et al. 1995a; Sandow et al. 2003b). However, immunohistochemistry does not allow to unambigiously distinguish heterocellar from homocellular coupling.

Endothelial and smooth muscle cells are separated by the internal elastic lamina, which, however, possesses holes, demonstrated in large arteries (Campbell and Roach 1981; Osborne-Pellegrin 1978; Osborne-Pellegrin 1985; Song and Roach 1984), that would allow close cellular contact. Already 40 years ago, Rhodin demonstrated projections of vascular cells (either from ECs, VSMCs, or both) through the fenestrated basal and elastic lamina that establish close contact between the membranes of ECs and VSMCs (Rhodin 1967, 1980). Although Rhodin named these areas myoendothelial junctions, a doubtless demonstration of GJs is missing (Rhodin 1980) and therefore they should probably be called myoendothelial bridges or myoendothelial contacts (Kristek and Gerova 1992; Sosa-Melgarejo and Berry 1992) to distinguish such areas of close contact without the detection of GJs from a true myoendothelial gap junction (MEGJ) comprising GJ channels. However, MEGJs exist in carotid arteries (Spagnoli et al. 1982) and more recent work by Sandow and colleaques has profoundly widened our knowledge about the presence of MEGJs in different vessels using serial-section electron microscopy (Dora et al. 2003; Emerson and Segal 2000; Sandow et al. 2002, 2003a, b; Sandow and Hill 2000).

ECs send projections through the fenestrated internal elastic lamina to make contact with VSMCs. At these sites of contact, the typical pentalaminar structures of GJs can be visualized in rat and mouse mesenteric artery (Dora et al. 2003; Sandow and Hill 2000), rat caudal artery (Sandow et al. 2003a), small skeletal muscle arteries, and arterioles (Emerson and Segal 2000; McSherry et al. 2006; Sandow et al. 2003b) as well as brain vessels of humans (Aydin et al. 1991) and rats (Sokoya et al. 2007). MEGJs not only are formed at projections arising from ECs, but also when VSMCs send projections that make contact with ECs. Furthermore, MEGJs are located within the internal elastic lamina where projections from ECs and VSMCs meet (Sandow et al. 2003a, b). The source of the projections varies between different vessel types, e.g., in small skeletal and mesenteric arteries nearly all MEGJs are found on EC projections (Sandow et al. 2003b), whereas in other large arteries, the majority of MEGJs are at projections arising from both cell types, ECs and VSMCs (Sandow et al. 2003a). Most of these GJs consist of only one plaque and these plaques are additionally small in size (<100 nm) in comparison to the considerably larger EC-EC plaques (Dora et al. 2003; Little et al. 1995a; Sandow and Hill 2000). The rare occasion of MEGJs in conjunction with their small size explains the difficulty in finding and convincingly identifying them. Moreover, not all projections of vascular cells sent through the elastic lamina do make contact with another vascular cell (Sandow et al. 2003b). Interestingly, the number of MEGJs increases with smaller vessel size as documented in the rat mesentery (Sandow and Hill 2000) and varies also dramatically between vascular beds in vessels of similar size (Sandow et al. 2003b). Despite the aforementioned evidence of the existence of MEGJs in mesenteric, carotid, caudal, brain, and small skeletal arteries, MEGJs have not been found in other vessels, e.g., femoral artery, even if studied by serial-section electron microscopy (Sandow et al. 2002), or in mouse pulmonary (Tolsa et al. 2006) or human coronary and mammary arteries (Conejo et al. 2007). In all of these studies, homocellular GJs between ECs and/or VSMCs were readily identified. Having this in mind, it cannot be assumed that in all vessels MEGJs are formed and indeed present. Even if the vascular cells send projections towards or through the internal elastic lamina, it does not imply that the morphological structure of a MEGJ is indeed existent. Morphological evidence of MEGJs in small resistance-sized arterioles in the microcirculation (with a diameter below 50 µm) is scarce (Gustafsson et al. 2003), possibly because these vessels have not been studied using serial-section electron microscopy. However, in guinea pig mesenteric arterioles, MEGJs have been demonstrated (Yamamoto et al. 2001).

CONNEXIN PROTEINS FORMING GAP JUNCTIONS BETWEEN ECs AND VSMCs

An even more interesting question that is yet unresolved relates to the connexin subtypes that build MEGJs. In the cardiovascular system, Cx40, Cx37, Cx43, and Cx45 are expressed (Saez et al. 2003). The expression pattern of different Cx is not specific for ECs and VSMCs, but Cx40 and Cx37 prevail in the endothelium throughout the vascular tree and in some vessels Cx43 is also present (Gustafsson et al. 2003; Haddock et al. 2006; Ko et al. 1999; Rummery and Hill 2004; Sandow et al. 2003b; Yeh et al. 1998; de Wit et al. 2000). In most VSMCs, Cx45 and Cx43 are expressed but the expression pattern of Cx in VSMCs seems to be heterogenous (Hill et al. 2002; Rummery et al. 2002; Wolfle et al. 2007). For example, Cx43 is lacking in VSMCs in some vascular beds and Cx37 was identified instead (Rummery et al. 2002). In contrast to Cx37, there are only a few observations suggesting that Cx40 is also expressed in VSMCs, e.g., in renal vessels (Arensbak et al. 2001), which is possibly related to the special role of Cx40 in renin-producing cells (Kurtz et al. 2007; Wagner et al. 2007). Even more confusing, in some vascular beds (mesenteric resistance arteries and arterioles), Cx40, Cx37, or Cx43 were not detected in VSMCs and Cx45 was not investigated, raising the possibility that some vascular beds lack Cx in VSMCs (Gustafsson et al. 2003). Summarizing these results, MEGJs could be composed of Cx40, Cx37, and Cx43 provided by ECs and Cx43, Cx45, and Cx37 provided by VSMCs. However, this has not been studied frequently in vessels yet. Interestingly, using a coculture system, it was demonstrated that MEGJs are composed of Cx40 and Cx43 provided by both cell types, ECs and VSMCs. This occured despite the expression of Cx37 in ECs and VSMCs, which was excluded from MEGJs in this in vitro coculture model (Isakson and Duling 2005). Only one study investigated the connexin composition of MEGJs in vessels until now. Using serial section and immunoelectron microscopy, the authors clearly demonstrate MEGJs in the rat basilar artery and were able to locate Cx40 and Cx37 at these GJs (Haddock et al. 2006).

EXPERIMENTAL APPROACHES TO STUDY MYOENDOTHELIAL COUPLING

Myoendothelial coupling can be assessed by studying dye- or electrical-coupling. Additionally, the effect of manipulating gap junctional coupling on diameter changes can be investigated. This latter approach provides also insight about functional implications of MEGJ signaling in vessels, but it inheres the difficulty in distinguishing whether homocellular or indeed heterocellular coupling has been affected by the experimental manipulation. As outlined above, an endothelial hyperpolarization initiates a hyperpolarization of the VSMC by a factor (EDHF) or simply by the transfer of current through MEGJs, which is possibly the most important physiological function of MEGJs (Griffith 2004; Sandow 2004; de Wit and Wolfle 2007). In addition, VSMCs may signal a feedback to ECs, which alters endothelial NO formation and thereby the vessel response upon stimulation (Dora et al. 1997; Isakson et al. 2007). These distinct functions may relate to different indices of MEGJ permeability, i.e., unitary conductance, charge selectivity, and permeability to larger signaling molecules (Goldberg et al. 2004). It has become evident that these indices do not correlate well with each other and thus need to be considered separately (Moreno 2004; Veenstra et al. 1995). Permeability to larger cytoplasmic molecules is best evaluated by studying dye-coupling and most likely reflects signaling through MEGJs by the transfer of second messengers (e.g., inositol triphosphate [IP₃] or Ca^2 +). In order to evaluate conductance and flux of ions, electrical coupling has to be analyzed and thus such experiments rely on the measurement of membrane potential and its changes. However, the most often used approach is the blockade of GJs, which does not allow to distinguish between effects on permeability to larger molecules or electrical conductance. Additionally, nonspecific effects of the substances used may cloud the experimental findings or mimic responses attributed to the blockade of GJs.

FUNCTIONAL EVIDENCE FOR MYOENDOTHELIAL COUPLING IN ISOLATED VESSELS

Gap Junction Blockers and Vascular Responses

Numerous studies have addressed the role of myoendothelial coupling in EDH-type dilation in isolated vessels using inhibitors of GJs such as carbenoxolone, derivatives of glycyrrethinic acid, heptanol, palmitoleic acid, and others (Dora et al. 2003; Harris et al. 2000; Taylor et al. 1998; Ungvari et al. 2002). Some of these blockers additionally alter nonjunctional membrane conductances by activation of not clearly identified cationic currents in concentrations that do not affect gap junctional communication (Coleman et al. 2001b; Matchkov et al. 2004; Tare et al. 2002). Therefore, the attenuation of EDH-type dilation by these substances observed in isolated vessels has to be cautiously interpreted and possibly reconsidered. More specific blockers are small peptides with homology to the extracellular loop of the pore forming connexins, so-called connexin-mimetic or gap peptides (Evans and Boitano 2001; Warner et al. 1995). Experiments in segments of small mesenteric arteries suggested that these peptides exerted an inhibitory action on gap junctions (i.e., enhanced intercellular resistance) without affecting membrane conductances in individual smooth muscle cells obtained from these vessels, suggesting that they do not exhibit major nonjunctional effects (Matchkov et al. 2006). However, recently the specificity has been questioned using the Xenopus oocyte expression system in which currents through pannexin (a novel group of proteins that can also form intercellular channels; for review, see Barbe et al. 2006) or connexin channels have been measured. Herein, gap peptides inhibited currents through pannexin channels but not currents through channels formed by connexins from which the sequence was derived (Wang et al. 2007). Therefore it is premature to regard gap peptides as highly specific. However, application of these peptides blocked EDH-type dilations in arteries of different organs from rat, rabbit, guinea pig, and pig that were induced by different endothelial agonists (Chaytor et al. 1998, 2001; De Vriese et al. 2002; Dora et al. 1999; Hutcheson et al. 1999; Sandow et al. 2003a; Sokoya et al. 2006; Ujiie et al. 2003). Although these data are convincing, gap peptides will interact in a similar manner with GJs that interconnect endothelial or smooth muscle cells homocellularly. Such a homocellular coupling may be decisive to provide an efficient source of a hyperpolarizing current in ECs that spreads to VSMCs. In addition, a functionally coupled smooth muscle unit may also be required to initiate a full dilation, especially in larger vessels, which contain multiple VSMC layers.

Membrane Potential Recordings in Large Arteries

Because the dilation is a consequence of VSMC hyperpolarization, measurement of the membrane potential itself provides further insight. Moreover, comparing potential changes simultaneously measured in ECs and VSMCs yield direct evidence of electrical coupling through MEGJs. Such measurements can be done using sharp electrodes and are often performed in conjunction with the assessment of dye-coupling because the identification of the impaled cell requires its staining by dye delivered out of the electrode. About 15 years ago, such approaches were applied by Beny and coworkers to study coupling through MEGJs in the isolated pig coronary artery. Although endothelial stimulation using bradykinin evoked hyperpolarizations in ECs and VSMCs, dye (Lucifer yellow) or even current injected in ECs was not detected in VSMCs (Beny 1990). Conversely, membrane potential changes initiated in VSMCs by isoproterenol or electrical field stimulation hyper- or depolarized ECs (Beny 1997; Beny and Pacicca 1994). These results suggest electrical coupling through MEGJs with strong rectification, only allowing charge transfer from VSMCs to ECs. However, the observed unidirectionality may be related to an insufficiency of ECs to provide a large current source that is able to hyperpolarize many layers of VSMCs (Beny 1999). This seems to be emphasized if EC hyperpolarization is initiated by nonphysiological stimuli (i.e., current). Because the VSMC hyperpolarization upon bradykinin administration could not be blocked using a nonspecific GJ blocker, which blocked the reverse potential transfer (i.e., from VSMCs to ECs), the authors suggested that in the coronary artery of the pig, an EDHF hyperpolarizes VSMCs. Thus, EDHF may exist in spite of the presence of MEGJs and EDHF ensures VSMC hyperpolarization upon EC stimulation. In a different vessel (ciliary artery), dye transfer from ECs to VSMCs (Lucifer yellow, propidium iodide) and a strong endothelial hyperpolarization upon bradykinin administration was found. However, the hyperpolarization of VSMCs was small and only detected in the subintimal part (Beny et al. 1997). These examples suggests that in some vessels ECs are unable to sufficiently hyperpolarize VSMCs electrotonically regardless of the presence of MEGJs and their potential to transfer current and/or dye. It should be noted that such an incompetence may also be related to a lack of augmentation mechanisms residing in VSMCs or due to an alteration of the spread of hyperpolarization within the smooth muscle layer.

Indeed, a significant role of homocellular coupling of VSMCs in large conduit vessels was verified by the use of Cx-mimetic peptides. In pig coronary arteries, such a peptide directed against various connexins substantially decreased VSMC hyperpolarizations if the cells were impaled from the adventitial side, but not if impalements were made from the intimal side (Edwards et al. 2000). A further advancement is possible by construction of different peptides exhibiting a small alteration of their amino acid sequences to block specifically different types of connexins. Such an approach was tested in a coculture model of ECs and VSMCs and revealed that Cx43 significantly contributes to myoendothelial dye-transfer (Martin et al. 2005). Such a specific blockade was exploited in rabbit iliac artery to separate between myoendothelial charge transfer and homocellular signaling within the VSMC layer. Measurements of VSMC hyperpolarizations in the subadventitial and subintimal area suggested that Cx40 and Cx37 are required to transmit EC hyperpolarizations towards subintimally located VSMCs, whereas GJs composed of Cx43 transfer the membrane potential change within the VSMC layer (Chaytor et al. 2005). Both GJ pathways are possibly subject to regulation (Edwards et al. 2007; Griffith et al. 2005). Taken together and in spite of the complicating fact of the necessity for the spread of the hyperpolarization throughout the VSMC layer in conduit vessels, the evidence acquired by the use of Cx-mimetic peptides suggest a role for electrical signaling through MEGJs in endothelium-dependent responses in isolated large arteries.

Recordings of the Potential Obtained in Smaller Arteries and Arterioles

Coupling through MEGJs should be more effective and prominent in smaller vessels because VSM consists of only one or two layers and MEGJs are more frequent in certain vascular beds as described above. Small vessels typically studied are harvested from the mesentery or the skeletal muscle due to their easy accessability. In such vessels (hamster cheek pouch), dye injected into ECs was transferred to adjacent ECs and VSMCs, but not or only weakly from VSMCs to ECs. Moreover, heterocellular coupling was not observed using Lucifer yellow, which also prevented the transfer of other dyes if coadministered (Little et al. 1995b). The supposedly GJ blocking properties of Lucifer yellow may have to be considered for correct interpretation of experiments on MEGJ coupling. Electrophysiological recordings in these vessels revealed alike shapes of membrane potential changes in ECs and SMCs initiated by phenylephrine (Xia et al. 1995). In feed arteries supplying a skeletal muscle (retractor muscle) from the same species, propidium iodide did not diffuse from ECs to VSMCs but only to adjacent ECs. In contrast, electrical bidirectional coupling was observed as evidenced by simultaneous measurements of membrane potential in ECs and VSMCs. De- and hyperpolarizing currents injected in ECs (or VSMCs) were retrieved in VSMCs (or ECs) and accompanied by vasoconstriction and -dilation. Moreover, spontaneous fluctuations of the membrane potential or the responses upon acetylcholine administration were indistinguishable between recordings obtained in ECs or VSMCs (Emerson et al. 2002; Emerson and Segal 2000). These experiments provide conclusive support of instantaneous and bidirectional electrical coupling between ECs and VSMCs in spite of a lack of dye transfer. Surprisingly, resistance of MEGJs as calculated from injected current and membrane potential changes was not different from homocellular coupling resistance between ECs or VSMCs. This is possibly due to the use of microelectrodes, which prevents exact calculations of these parameters.

Experiments on small mesenteric arterioles from guinea pigs (50 to 100 µm) by other groups yielded similar conclusions. Spontaneous action potentials, which were initiated by barium in VSMCs (as verified by blocking GJs), were recorded identically in ECs (Yamamoto et al. 1998). Conversely, EC hyperpolarizations initiated by acetylcholine were registered alike in VSM and were sensitive to uncoupling of GJs (Yamamoto et al. 1999). In further studies, these authors used two patch-clamp electrodes simultaneously, allowing the initiation of electrotonic potentials and a more accurate calculation of resistances between cells. MEGJs allowed bidirectional current flow without rectification, but behaved as an ohmic resistor, which was about 10 times higher (approximately 0.9 GΩ considering a single VSMC) than resistances between homocellularly coupled VSMCs (Yamamoto et al. 2001). These experiments suggest that bidirectional coupling without rectification allows vascular cells to modify their membrane potentials mutually but the resistance of MEGJs is appreciable. However, taking into account that ECs are extremely well coupled and potentially able to regenerate a hyperpolarizing signal, they should affect VSMC potential substantially. Indeed, only a slight attenuation of the membrane potential in VSMCs was recorded if acetylcholine is used to initiate the hyperpolarization in ECs (Yamamoto et al. 2001). Coleman and coworkers came to very similar conclusions in even smaller arterioles (20 to 50 µm) obtained from the same vascular bed (Coleman et al. 2001a, b, 2002). Dual–patch-clamp recordings in isolated pial arterioles (diameter of 20 µm) revealed electrical coupling also in these vessels consisting only of an EC and a single VSMC layer. These authors considered the spatial arrangement of the cells in the vascular wall for the calculation of resistances between two VSMCs (0.33 GΩ) and between ECs and VSMCs (1.68 GΩ) (Yamazaki and Kitamura 2003), which corresponds fairly well to the values obtained in mesenteric arterioles (Yamamoto et al. 2001). Moreover, the analysis of the conductance/voltage dependency revealed symmetry between adjacent VSMCs but asymmetry between ECs and VSMCs (Yamazaki and Kitamura 2003). The asymmetric behavior of MEGJs suggests that different connexins form these pores whereas GJ between adjacent VSMCs are built by a single Cx protein subtype. However, the single channel conductance measured between adjacent VSMcs (200 to 230 pS) is not readily attributable to one of the Cx subtypes expressed in vessels.

In a very recent functional study, Cx40 was blocked by loading antibodies selectively into ECs of isolated small mesenteric arteries, which abrogated EDH-type dilation without altering the endothelial Ca^2 + increase upon stimulation with acetylcholine. Antibodies directed against other connexins (Cx37 or Cx43) were without effect. These functional observations were supported by the localisation of Cx40 to MEGJs using immunoelectron microscopy and demonstrate a central role of MEGJs in EDH-type dilation in vitro and highlight a primary importance of Cx40 therein (Mather et al. 2005).

In summary, in vitro experiments provide convincing evidence that signaling through MEGJs contributes significantly to the regulation of vascular tone and provides a pathway that can account for an EDH-type dilation. The most convincing data have been gathered in vessels from the mesenteric vascular bed and therein most prominently in small arteries and arterioles. Some studies also suggest that signaling through MEGJs is effective in arteries from vascular beds supplying skeletal muscles. However, in these vessels, convincing evidence that MEGJs account for EDH-type dilation is still lacking. If the mechanisms of GJ blockade by Cx-mimetic peptides is more clearly understood and proves to be specific, MEGJs can be accounted for an EDH-type dilation in additional vessels. The evidence that other GJ blockers alter membrane conductances is accumulating and these substances should therefore not be used to verify the contribution of MEGJs to physiological signaling in vessels.

IN VIVO EXPERIMENTS QUESTION THE IMPORTANCE OF SIGNALING THROUGH MEGJs

Gap Junction Blockers In Vivo

There are many studies using nonspecific GJ blockers (derivatives of glycyrrethinic acid or heptanol) in vivo; however, their results will not be discussed here, because they do not provide compelling evidence for signaling through MEGJs. To the best of our knowledge, the effect of the more specific Cx-mimetic peptides was investigated only in a single study, in which these peptides were infused into the renal artery (De Vriese et al. 2002). The application of a peptide directed against Cx40 abolished, and against Cx43 partially inhibited, the EDH-type dilation upon acetylcholine administration, whereas responses upon administration of endothelium-independent dilators remained unaffected. Both peptides enhanced basal vessel tone. Although these data suggest a role for Cx40 and Cx43 in EDH-type dilation, it is unclear if this inhibitory effect indeed reflects the blockade of MEGJs (see above). Studying the physiological relevance or the existence of MEGJs in vivo without the use of such substances is substantially more demanding.

Membrane Potential Recordings of Vascular Cells In Vivo

Only some groups have measured the membrane potential in vascular cells directly in vivo, which is certainly a difficult task. Already in 1992 Segal and Beny published membrane potential recordings using sharp electrodes from vascular cells in arterioles of the hamster cheek pouch, which is a very common preparation to study arterioles in the microcirculation in vivo. The dye Lucifer yellow used to identify the cells diffused into adjacent ECs if an EC was impaled. However, ECs were not dye-coupled to VSMCs in spite of the presence of myoendothelial bridges. Dye introduced into VSMCs did also not diffuse to other VSMCs or ECs (Segal and Beny 1992). The failure to demonstrate dye-coupling may be due to the dye chosen, which was reported earlier even to prevent coupling of other dyes (Little et al. 1995b). Resting membrane potential in a very limited number of observations (each n=3) were −52±8 and −44±2 mV in ECs and VSMCs, respectively. Both cell types hyperpolarized upon acetylcholine adminsitration. However, recordings obtained did not provide information about signaling through MEGJs and due to an absent dye-coupling, it was concluded that VSMC hyperpolarization and the EDH-type dilation is not related to direct current transfer from ECs to VSMCs (Segal and Beny 1992). In the same preparation, phenylephrine or norepinephrine induced VSMC depolarization (and arteriolar constriction), but the membrane potential of ECs remained unaltered, although ECs were in principle able to depolarize (in response to high potassium solution). The lack of simultaneous membrane potential changes in ECs and VSMCs was not related to the polarity of the membrane potential change because a hyperpolarization initiated by acetylcholine was also dissimilar in ECs and VSMCs. In VSMCs, this hyperpolarization lasted longer and was followed by a brief depolarization. In contrast to MEGJ coupling, both layers (EC and VSMC) proved to be coupled homocellularly because the membrane potential changes initiated at local application sites conducted to distant locations, verifying that impalement of the cell did not result in an electrical isolation of the impaled cell. Moreover, the distinct shape of the hyperpolarization initiated at the local application site was retrieved in the respective cell layer at distant sites. This further suggests that MEGJs do not adjust or modify the hyperpolarisation, when it propagates along the vessel wall. Nevertheless, the resting membrane potential of ECs and VSMCs were in a comparable range, in this study amounting to about −35 mV (Welsh and Segal 1998). A different stimulus to alter the membrane potential of VSMCs is the change of the ambient oxygen pressure (pO₂). Elevating the pO₂ in the superfusion fluid from∼20 to 150 mm Hg initiates a strong arteriolar constriction in the microcirculation that was preceded by a depolarization of VSMCs (from −37±3 to −15±1 mV). However, the membrane potential of ECs remained unchanged in response to this stimulus in spite of a similar constriction (Welsh et al. 1998). Taken together, the data acquired by Segal and coworkers in this prepation (hamster cheek pouch arterioles) clearly demonstrate that membrane potential changes are not transferred from VSMCs to ECs in vivo. These studies also do not provide any clue that current is transferred from ECs to VSMCs.

Alternative Approaches to Verify MEGJs

A different approach to verify signaling through MEGJs is the evaluation of the channels that open and produce the hyperpolarization in response to acetylcholine. The endothelial hyperpolarization is the first step of the EDH-type dilation and is achieved by the opening of calcium-dependent K⁺ channels (K_Ca) (Griffith 2004; Sandow 2004; de Wit and Wolfle 2007). Different subtypes of K_Ca channels exist, which are divided into three major groups, namely small- (SK_Ca), intermediate- (IK_Ca), and large- (BK_Ca) conductance channels. In ECs, SK_Ca and IK_Ca channels are expressed and their activation is required to initiate an EDH-type dilation (Griffith 2004; Si et al. 2006). Most interestingly, hyperpolarizations of ECs and VSMCs in response to acetylcholine in the cremaster muscle microcirculation of mice measured in vivo was differently affected by blockers of K_Ca channels. VSMC hyperpolarization was dependent on BK_Ca activation and sensitive to a specific blocker of this channel (iberiotoxin), whereas EC hyperpolarization remained unaltered by this toxin, which strongly argues against direct current transfer from ECs to VSMCs. In contrast, EC hyperpolarization was largely dependent on the opening of SK_Ca (as demonstrated by its sensitivity to apamin) and apamin had only a slight effect on VSMC hyperpolarization. If indeed MEGJs significantly contributed to VSMC hyperpolarization, the sensitivity of EC and VSMC hyperpolarization towards specific K_Ca channel blockers should be alike and similar, which, however, was not the case. Moreover, the amplitude and duration of acetylcholine-induced hyperpolarizations in ECs and VSMCs were distinct, as was the resting membrane potential of these cells, further arguing against a strong myoendothelial coupling in vivo (Siegl et al. 2005).

In addition to transversal heterocellular communication, electrical signals propagate also longitudinally along the vascular wall due to homocellular coupling. This communication form is reviewed in detail elsewhere (Segal 2005; de Wit 2004) but may help to clarify if heterocellular signaling exists. Homocellular coupling is most often studied by assessing the mechanical response of the vessel at certain distances away from a site of local stimulation (Hoepfl et al. 2002). These remote, distant responses rely on the propagation of an electrical signal along homocellularly coupled ECs and/or VSMCs. Destruction of these signaling pathways by a light-dye technique (applied in hamster cheek pouch arterioles) reveals the cell layer that is crucial for transmitting the signal. If MEGJs were present and would allow a significant current transfer from ECs to VSMCs (or vice versa), an electrical signal propagating along the vessel wall should traverse a site at which a single layer is destroyed. Further, two sites with a selectively destroyed cell layer in succession along the vessel wall should likewise be traversed. If such single layer damages are applied in alternative order along the vessel, such a procedure reveals the preferential current transfer (from ECs to VSMCs or VSMCs to ECs), depending on the cell layer damage that brings the propagating signal to an halt. However, neither of these possibilities proved to be true. Instead, the propagation stopped always at the second site of damage and this was independent of the damaged cell layer. This implies that both cell layers, EC and VSMC, act as separate signaling pathways ('conduction cables') unconnected to each other. A signal that is travelling along the wall cannot jump between the layers via MEGJs (Budel et al. 2003).

Connexin-Deficient Mice

As outlined above, specific blockade of GJs is difficult to achieve in vivo. However, mechanical responses elicited by acetylcholine in the microcirculation of Cx40-deficient mice have been evaluated. This is of special interest because Cx40 was implicated from in vitro experiments to be critical for EDH-type dilation in mesenteric small arteries (Mather et al. 2005). In vivo, injection of acetylcholine into the arterial bed of Cx40-deficient mice elicited similar drops in arterial pressure that were not different from wild-type littermates independent of the presence of NO (de Wit et al. 2003). This suggests that the dilation upon acetylcholine administration in resistant vessels in vivo does not require Cx40 (in MEGJs). This view is supported by the direct observation of arteriolar dilations in skeletal muscle arterioles (cremaster muscle) of these mice, which was only very mildly attenuated in Cx40-deficient animals (de Wit et al. 2000). Although in that study the contribution of NO was not evaluated, the dilation upon acetylcholine administration is in this preparation almost independent of the NO pathway (Koeppen et al. 2004; Wolfle and de Wit 2005). It needs to be highlighted that these experiments do not disprove a role for MEGJs, but they clearly demonstrate that Cx40 is not a critical connexin for intact signaling through MEGJs in vivo, as was shown in mesenteric arteries in vitro. However, it cannot be excluded that in animals chronically deficient for Cx40 other Cx compensate for its loss. Experiments to eliminate Cx40-dependent gap junctional signaling acutely have not been performed in the microcirculation yet.

In summary, in vivo studies do not verify a significant role for signaling through MEGJs in the physiological regulation of vascular tone and vessel responses in EDH-type dilations. Electrophysiological measurements as well as functional experiments do not provide a clue for the hypothesis that signaling through MEGJs is effective under these conditions. In addition, dye-coupling of ECs to VSMCs is lacking in arterioles of the hamster cheek pouch or of the skeletal muscle. Cx-mimetic peptides have not been used routinely in vivo and the only study in which these were applied in such manner demonstrated an inhibitory effect. However, this inhibitory effect cannot be targeted without doubts towards an effect on MEGJ.

POSSIBLE EXPLANATIONS FOR THE DISCREPANCY BETWEEN IN VIVO AND IN VITRO EXPERIMENTS

The reasons for the discrepant results using microelectrodes and other approaches obtained in vitro and in vivo are unclear. Using microelectrodes requires penetration of the cell membrane and thereby inevitable injury to the cell, which may alter coupling itself. However, such injury cannot explain the divergent results because it is used in a similar fashion in vitro demonstrating intact myoendothelial coupling and in vivo failing to do so as outlined above. Additionally, microelectrodes do not likewise interfere with homocellular coupling in vivo (Siegl et al. 2005; Welsh and Segal 1998). If indeed the use of microelectrodes leads to the discrepant findings, the technique must have specifically interfered with MEGJs and only in the in vivo situation, which is a rather unlikely event. More likely, the difference is related to vessel size because arterioles investigated in vitro are usually larger (>80 µm) than those studied in vivo (30 to 80 µm). However, convincing data verifying strong bidirectional coupling between ECs and VSMCs have been obtained in arterioles with a diameter below 50 µm using electrophysiological approaches. Another likely explanation is differences between vascular beds. Indeed, most vessels studied in vitro are harvested from the mesentery because these vessels are easily accessible. In contrast, none of the studies performed in vivo have addressed MEGJ coupling in this vascular bed but were instead done on skeletal muscle arterioles. In fact, connexin expression varies between vascular beds, which may relate to specialized functions in different organs and can possibly explain part of the divergent findings.

Alternatively, MEGJ coupling may be specifically and tightly controlled in the in vivo situation. Factors accountable for such a control are the presence of blood, shear stress, and blood flow with NO release (Rodenwaldt et al. 2007) and sympathetic innervation, to name a few of many physiological variables that are not present in in vitro studies. The most obvious difference is the presence of blood (proteins and cells), which is known to have many effects on the vessel wall itself (e.g., contribution of proteins to microvascular permeability (Michel and Curry 1999)). Another likely stimulus is blood flow and shear stress because isolated vessels are either mounted isometrically on wires or cannulated with pipettes, without, however, perfusing these vessels. These conditions alter the membrane potential of vascular cells, which reflects an alteration of ion channel activity. Obviously, this could likewise affect MEGJs, but this has yet to explored. Recently, oxidized phospholipids have been demonstrated to affect the expression of connexins and their phosphorylation status, especially at the myoendothelial junctions in in vitro coculture system (Isakson et al. 2006). This suggests that the complexity of connexin regulation is only starting to be unraveled and many factors present in vivo but lacking in isolated vessels may contribute to the discrepancy. Alternatively, studies in vivo add other variables that may account for differences. One of these is anesthesia, which has been reported also to affect the release of EDHF in the microcirculation (de Wit et al. 1999). It is conceivable that anesthetics may uncouple MEGJs in vivo. However, anesthetic regimens were distinct in the aforementioned in vivo studies and anesthetics must have altered MEGJs specifically to account for these differences, because in numerous studies on anesthetized animals, homocellular coupling is readily to be seen using electrophysiological methods or studying conducted responses. Finally, other possibilities have to be considered. For example, it is conceivable that unknown factors mediate EDHF-type responses in vivo, which are lacking in isolated vessels. Thus, the role of MEGJs comes to the fore under these circumstances. Other gaps in our knowledge of connexin function in vascular physiology need to be filled as well (Figueroa et al. 2004). New substances that specifically interfere with gap junction or connexin function and are easily delivered as well as genetically engineered mice may help to improve our understanding of this fascinating field.

Acknowledgements

This work was supported by the Deutsche Forschungsgemeinschaft (WI 2071/2-1).