INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

THE INDIVIDUAL SMOOTH MUSCLE CELLS lying in all systemic arterioles and arteries examined to date are coupled to their neighboring cells to form electrical syncytia (14). In most of these vessels, a sympathetic transmitter initiates rapid excitatory junction potentials that, in turn, cause the opening of muscle voltage-dependent Ca²⁺ channels (14). Because release of the transmitter from individual varicosities is intermittent (16), the transmitter is released at widely spaced points where it triggers depolarizations. Because the individual smooth muscle cells are electrically coupled together, depolarization spreads from cell to cell and Ca²⁺ entry can occur in many cells (14).

In a recent paper (11), it was shown that in rat iridial arterioles the depolarizing and contractile responses produced by sympathetic nerve stimulation depend on the activation of -adrenoceptors. When activated, these receptors cause the formation of a second messenger that, in turn, releases Ca²⁺ from intracellular stores (10). The first aim of these experiments was to determine whether smooth muscle cells in vessels, which depend primarily on the release of Ca²⁺ from the intracellular stores to trigger contraction, are coupled together to form electrical syncytia.

Close contacts between neighboring smooth muscle cells, neighboring endothelial cells, and adjacent smooth and endothelial cells have been described in a number of blood vessels (see Refs. 5, 23, 26). Because the organization of smooth muscle and endothelial cells varies from vascular bed to vascular bed (9), the second aim of the study was to examine rat iridial arterioles at the ultrastructural level for the presence of close contacts between arteriolar smooth muscle cells and endothelial cells. This has allowed a correlation between structural and electrophysiological observations to be carried out on the smooth muscle cells of iridial vessels.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Recordings were made from iridial arterioles. Preparations of the iris were pinned in the same way as described previously for experiments in which recordings were made from either the dilator layer of the iris (13) or from iridial arterioles (11). Briefly, the iris was removed from each eye. After sections of the iris containing an arteriolar tree were pinned out in a chamber through which flowed warmed (35°C) physiological saline composed of (in mM) 119.8 NaCl, 5.0 KCl, 25 NaHCO₃, 1.0 NaH₂PO₄, 2.5 CaCl₂, 2.0 MgCl₂, and 22.0 glucose gassed with 95% O₂-5% CO₂, the arterioles were viewed with an inverted compound microscope (magnification ×200 or ×400). The arterioles were impaled with two independent glass microelectrodes (resistance 150-250 M) each filled with 0.5 M KCl. Under these conditions, it was possible to resolve the position of the microelectrode tips such that separations between the microelectrodes could be assessed to the nearest 5 µm. A successful impalement was judged to be one in which the resting membrane potential stayed the same for >10 min. Membrane potentials were measured with a conventional dual-input amplifier (Axoclamp-2, Axon Instruments, Foster City, CA). After impalement, one of the amplifiers was switched to current-passing mode and used to inject current into the arteriole. All membrane potential and current records were low-pass filtered (cutoff frequency 1 kHz), digitized at 500 Hz, and stored on disk for later analysis. Perivascular nerve stimulation was achieved by use of a pair of fine platinum stimulating electrodes (11).

In several experiments, sections of arteriole (length 1.5-2.5 mm) were cut with a fragment of a broken razor blade (see Ref. 15). The equivalent electrical cable properties of these preparations were determined as described previously (15). Briefly, one recording electrode was placed in the center of the isolated segment of the arteriole. The second electrode was placed at one end of the segment of the arteriole. When a current pulse was passed through the central electrode, an electrotonic potential was detected at the end electrode. If the segments of the arteriole behaved like short cables with sealed ends, the shape, time course, and amplitude of that potential would be determined by the membrane time constant, electrical length constant, and axial resistance of the arteriole (15). This treatment was justified in the case of submucosal arterioles (see Ref. 15) and appears appropriate for segments of iridial arterioles (see RESULTS). The SCoP modeling package version 2.61 (Simulation Resources, Durham, NC) was used for the simulations.

In some experiments, recordings were made with electrodes filled with 3% lucifer yellow in a 150 mM LiCl solution. After the cell was impaled and characterized, the dye was injected by passing hyperpolarizing current pulses (500 ms at 0.1-1 nA) at a frequency of 1 Hz for 5-20 min. After fixation in 4% paraformaldehyde in 0.2 M phosphate buffer, pH 7.3, the preparations were mounted in buffered glycerol and examined with a confocal microscope.

Stretched and unstretched rat irises were prepared for transmission electron microscopy by fixation in 4% paraformaldehyde (1:1) with 1% glutaraldehyde in 0.1 M sodium cacodylate buffer at pH 7.35. Regions of arterioles similar to those selected for physiological recordings were sectioned. In one preparation, serial longitudinal sections (90-100 nm) were cut over a distance of 5 µm, and the center section was photographed as a montage at ×2,200 and further enlarged photographically (×4). Each smooth muscle cell profile in that section was numbered. Very small profiles (<1 µm) were ignored. Thirty muscle cells from a total of 112 cells were selected with a computer-based random-number generator and followed throughout the serial series. Appositions between adjacent cells through that area were observed and recorded. Because individual appositions were usually present in more than one consecutive section (especially in the case of long flat appositions), individual associations were followed and recorded as a single association.

In the other two preparations, serial longitudinal sections (90-100 nm) of arterioles were cut over distances of 20 and 22 µm. Two individual smooth muscle cell profiles were selected from different preparations and photographed for their entire length, starting from a centrally selected section in the series. Associations between each of these smooth muscle cells and the endothelia were observed and recorded as above.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

General observations. When preparations of the iris were viewed with an inverted compound microscope, it could be seen that four major arterioles entered the connective tissue stroma overlying the dilator muscle layer. The arterioles had an external diameter of ~40-60 µm, with wall thicknesses of 4-6 µm. Each arteriole gave rise to a variable number of fine side branches with external diameters of 10-20 µm. These values of arteriolar diameters are an overestimate because the vessels were flattened when the iris was stretched during pinning (see Fig. 1, in which a longitudinal section through a flattened arteriole appears to have a diameter of 10-15 µm). It was possible to follow the path of the side branches until they branched to form capillary networks. The capillaries recollected into a series of radially orientated venules. In a proportion of preparations, the arterioles contracted spontaneously, with constrictions occurring at regular intervals of ~10-12 s. We have no explanation for why some preparations were spontaneously active, whereas others, often from the same animal, lacked spontaneous activity.

View larger version (36K): [in this window] [in a new window]   Fig. 1.   Low-power micrograph of longitudinal section of stretched iris arteriole showing general vessel morphology. SMC, smooth muscle cell; EC, endothelial cell; V, venule; AB, axon bundle; D, dilator; RBC, red blood cells; A, anterior; P, posterior. Bar, 25 µm.

Ultrastructural observations. When sections of arterioles were viewed with an electron microscope, the walls of the arterioles were seen to consist of an inner layer of endothelial cells and an outer layer of arteriolar smooth muscle cells. In preparations that had been stretched in a way similar to those used for electrophysiology, the endothelial layer was ~0.2-1.0 µm thick, rising to 2 µm near the nucleus. The muscular wall of the arterioles usually consisted of a single layer of smooth muscle cells with a thickness of 2-4 µm. In some areas close to the ciliary edge of the arterioles, a second layer of smooth muscle cells was evident. Although occasional deposits of elastin were detected, the endothelial and smooth muscle layers were not separated by a continuous elastin sheet (Fig. 1).

When examined with higher magnification, gap junctions between adjacent muscle cells were frequently observed, i.e., areas of close apposition of membranes of adjacent muscle cells with separations of <20 nm (Fig. 2A; see also Ref. 4). These junctions were often associated with membrane thickenings. Regions of very close apposition were also frequently detected between neighboring endothelial cells. At the regions of contact, the separation between membranes was ~20 nm; occasionally, the membranes at these regions of contact were electron dense (Fig. 2B).

View larger version (176K): [in this window] [in a new window]   Fig. 2.   Cellular associations in iris arterioles. A and B: regions of close contact (arrowheads) between adjacent smooth muscle cells and between adjacent endothelial cells, respectively. C-F: the 4 types of association observed between smooth muscle and endothelial cells. C: long flat region of association. D: long finger- or mushroomlike projection. E: broad blunt projection. F: short projection. Bars: 0.5 µm in A and B; 5 µm in C; 0.25 µm in D-F.

Gap junctions with areas of membrane specialization were not observed between endothelial cells and smooth muscle cells, although regions of close apposition between endothelial and muscle cell membranes were detected. These could be subdivided into four distinct types. Long flat regions of apposition were observed where the cell membranes were separated by a gap of 60-200 nm filled with a continuous layer of fused basal lamina (Fig. 2C). In the second type of apposition, endothelial cells gave rise to long finger- or mushroomlike projections that approached the nearby muscle cell, forming a gap of ~60-150 nm filled with a single layer of basal lamina (Fig. 2D). The third type of apposition between endothelial and muscle cells was characterized by broad blunt-ended projections arising from either the endothelial cells or the muscle cells. The projections approached the target cell to within 40-100 nm; again, the gap between cells was filled with a layer of basal lamina (Fig. 2E). The final type of apposition was similar to the last type, with short projections coming to within 40-100 nm of the adjacent cell (Fig. 2F). When 30 smooth muscle cells and associated endothelial cells were examined, 335 projections were detected. Of these, 9.3% were long and flat, 4.2% were long finger- or mushroomlike, 15.8% were broad and blunt, and 70.7% were short projections. All projections had caveolae along their surfaces; the mushroomlike and blunt projections contained accumulations of vesicles.

When the associations between individual smooth muscle cells and endothelial cells were followed through serial sections, these proportions varied slightly. A total of 72 projections were detected for two cells, of which 6.9% were long and flat, 19.4% were long finger- or mushroomlike, 8.3% were broad and blunt, and 65.3% were short projections. Because very small profiles of smooth muscle cells were omitted from the random counts, these results suggest that the ends of the smooth muscle cells contain numerous projections, particularly the long finger- or mushroomlike appositions.

Electrophysiological observations. When intracellular recordings were made from the iridial arterioles, again two distinct classes of cells were identified (see Ref. 11). In the first group of cells, identified as smooth muscle cells (see Ref. 11), repetitive perivascular nerve stimulation evoked membrane depolarizations with amplitudes of >25 mV that were followed by a constriction (Fig. 3, A and B). Cells of this type in a proportion of the preparations examined (9 of the 20 preparations examined) also exhibited rhythmical membrane potential changes that were associated with arteriolar constrictions (Fig. 4). When detected, rhythmical depolarizations occurred with a frequency between 0.1 and 0.08 Hz: spontaneous depolarizations had peak amplitudes in the range of 5-20 mV. In the cells that displayed rhythmical activity, the resting membrane potential was taken as the peak negative potential. Surprisingly, the "resting" membrane potentials of cells that displayed myogenic activity did not differ from those in which myogenic activity was not detected (see Ref. 11). The group of cells that responded to transmural nerve stimulation, where recordings were made with conventional microelectrodes, had resting potentials in the range of 45 to 75 mV (mean ± SE 64.6 ± 2.5 mV; n = 12 separate preparations).

View larger version (12K): [in this window] [in a new window]   View larger version (104K): [in this window] [in a new window]   Fig. 3.   Structural organization of smooth muscle cells of rat iridial arterioles. A brief train of sympathetic nerve stimuli, 10 impulses at 10 Hz, produces a contraction (A) and a membrane depolarization (B). C: morphological appearance of cell from which these recordings were obtained after it had been filled with lucifer yellow. Cell is seen as a bright radially orientated structure lying in arteriolar wall.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Spontaneous membrane potential changes recorded from smooth muscle cells from the same segment of rat iridial arteriole. When a hyperpolarizing current pulse of 2-nA magnitude was passed through 1 electrode for 200 ms (A), an electrotonic potential was detected at the 2nd electrode (B) that was ~1 mm distant. Peak negative potentials detected at the 2 electrodes were 63 and 67 mV, respectively. Voltage and time calibration bars apply to each trace.

To confirm the identity of the cells that responded to perivascular nerve stimulation, recordings were made from 10 such cells in 8 preparations with lucifer yellow-containing microelectrodes. The resting potentials determined with such electrodes were very similar to those determined with conventional electrodes, lying in the range of 57 to 70 mV (66.0 ± 2.0 mV; n = 6 cells). The two values of resting membrane potential, determined with different microelectrodes, failed to show a significant difference (P > 0.1 by Student's t-test). Nine of the ten cells injected with lucifer yellow, usually with a hyperpolarizing current of 0.1 nA applied for 10 min, showed positive staining, with the other cell failing to stain. On each occasion, the stained cell had the morphological characteristics of an arteriolar smooth muscle cell (Fig. 3C). The cells had their long axes wrapped circumferentially around the arteriole (see Refs. 19, 24). None of the cells displayed a morphology like that attributed to endothelial cells (19, 24).

In the second group of cells, trains of perivascular stimuli failed to initiate membrane depolarizations. These cells had resting membrane potentials in the range of 76 to 85 mV, (79.5 ± 1.3 mV; n = 9 cells). These values differed significantly from those found in cells that responded to transmural nerve stimulation (P < 0.001 by Student's t-test). Recordings from these cells were usually obtained after the electrode had clearly passed through the smooth muscle layer. Attempts to identify these cells with lucifer yellow were ambiguous. On four occasions when the cells that failed to respond to perivascular nerve stimulation and had membrane potentials more negative than 70 mV were injected with dye, they failed to stain. In these cells, hyperpolarizing current pulses in excess of 0.1 nA for up to 20 min were used to inject dye. Together, these observations suggest that the dye rapidly escaped from this group of cells. On two other occasions, cells that failed to respond to perivascular nerve stimulation were injected with lucifer yellow. These cells had membrane potentials of 56 and 57 mV as determined with lucifer yellow-filled electrodes. Each of these cells was found to lie immediately under the arteriole.

Electrical coupling between arteriolar smooth muscle cells. Intracellular recordings with a pair of microelectrodes were made from two identified smooth muscle cells in the same arteriolar tree on six occasions. In each case, when a current pulse was passed through one electrode, it produced a potential change at the second electrode (see Figs. 4 and 5). When the preparations displayed myogenic activity, the pattern of activity detected at each electrode was very similar (Fig. 4). When the two electrodes were placed close together, that is, within 200 µm, the current-voltage curves were found to be linear for small membrane potential changes (Fig. 5, A and B), with slope resistances in the range of 2.3-4.8 M (3.7 ± 0.4 M; n = 6 cells). The mean resting membrane potential of the more depolarized cell of each pair of recordings was 59.4 ± 6.9 mV (n = 6 cells); the mean resting membrane potential of the more hyperpolarized cell of each pair of recordings was 65.8 ± 5.6 mV (n = 6 cells). These values failed to show a significant difference (P > 0.1 by Student's t-test). Together, these observations suggest that iridial arteriolar smooth muscle cells have similar resting membrane potentials and are coupled together to form an electrical syncytium.

View larger version (19K): [in this window] [in a new window]   Fig. 5.   Electrical coupling between smooth muscle cells of rat iridial arterioles. Membrane potential changes (A) were detected at a recording electrode in response to passage currents of +1, 1, 2, and 3 nA (B) through a 2nd intracellular electrode; electrodes were separated by 120 µm. Resting membrane potentials recorded at the 2 electrodes were 70 and 71 mV, respectively. When current-injecting electrode was withdrawn, a 3-nA current pulse (D) failed to produce a detectable membrane potential change (C) in arteriolar smooth muscle cell. Time calibration refers to each trace.

In eight experiments, intracellular recordings were made from sections of arteriole with a pair of microelectrodes, with one recording electrode placed in the smooth muscle layer and the other in a cell that failed to respond to perivascular nerve stimulation. When a current pulse was passed through one recording electrode, it failed to produce a detectable membrane potential change in the second electrode. Clearly, the cells that fail to respond to sympathetic nerve stimulation are not electrically connected to arteriolar smooth muscle cells.

Passive electrical properties of iridial arterioles. Recordings were subsequently made from arteriolar smooth muscle cells lying in segments of arteriole that had been isolated from the main arterial inflow. This was done by making a cut in the arteriole just after its entry into the stroma overlying the dilator muscle layer. Similarly, the arteriole was cut ~2 mm distal to this point. Unfortunately, we found that when cuts were made through any fine side branches that arose from the main arteriolar trunk, it was difficult to avoid damage to the iris, which impeded successful pinning of the preparations. To minimize this complication, sections of arteriole were selected that gave rise to less than three side branches. Because these side branches were short (<100 µm) compared with the major arteriolar trunk, the inaccuracies introduced as a result of current loss would be small. The segment of arteriole was impaled with two independent electrodes, one in the center of the arteriole and the other within 50 µm of a cut end. The length of the arteriole, the separation between electrodes, and the inner and outer diameters were measured. When a current pulse was passed through the central electrode, an electrotonic potential was detected at the second electrode. The peak potential change occurred after the end of a brief current pulse (Fig. 6, C and E), suggesting that the two electrodes were electrically distant from each other (see Ref. 15). The final decay of the electrotonic potential could be described by a single exponential, suggesting that the preparation was behaving as a short cable (see Ref. 15). To determine the equivalent electrical properties of the arterioles, it was assumed that segments of the arteriole behaved as short cables with sealed ends. The time course of the electrotonic potential produced by a short current pulse (100 ms) was recorded. With the use of an optimizing routine, the time constant, length constant, and axial resistance were determined by successively improved fits of the theoretical function (Eq. 1 in Ref. 15) to the data. An example of an electrotonic potential and its theoretical description is shown in Fig. 7. Finally, to check the assumption that the cut ends of the segments had sealed, long current pulses were passed through the microelectrode and the final steady-state membrane potential change was measured (Fig. 6F). If the ends of the arteriole had sealed, then the measured potential changes should be similar to those predicted for a long pulse using the cable parameters calculated in the short-pulse experiments (see Ref. 15). This was the case (see Table 1).

View larger version (9K): [in this window] [in a new window]   Fig. 6.   Electrotonic potentials produced by short and long current pulses recorded from an arteriolar smooth muscle cell lying in an isolated segment of arteriole. Each cell gave a membrane depolarization in response to transmural nerve stimulation. A and B: membrane potential changes detected when 1-nA current pulses (durations 100 ms and 1 s) were passed through current-passing electrode just after it had been withdrawn from muscle cell layer. C and D: electrotonic potentials detected when these current pulses were passed through a centrally located current-passing electrode impaled in a 2nd muscle cell. Electrode separation was 1 mm. E and F: differences between these traces. It can be seen that electrotonic potentials started after onset of current injection. Resting membrane potentials of the 2 cells were 70 and 72 mV, respectively. Voltage and time calibration bars refer to all traces.

View larger version (6K): [in this window] [in a new window]   Fig. 7.   Calculation of equivalent electrical cable properties of rat iridial arterioles. Solid line, best fit description for an electrotonic potential generated in an arteriole 2.1 mm in length when current is passed through a centrally located electrode and potential is detected at a 2nd electrode positioned 50 µm from 1 end of arteriole. Calculation assumes that cut ends of cable had sealed. Dots, time course of electrotonic potential illustrated in Fig. 6E. This calculation suggested that the segment of arteriole had a membrane time constant of ~100 ms and an electrical length constant of ~1 mm.

All of the calculations from this experiment and three similar experiments are shown in first four examples shown in Table 1. Also shown in Table 1 are data obtained from experiments in which recordings were made from spontaneously active preparations. Although the electrotonic potentials were considerably distorted by the myogenic depolarizations such that the cable fits were poor, it is clear that smooth muscle cells in these preparations were also electrically coupled to each other to form a syncytium.

In summary, isolated segments of iridial arteriole behave in a way that is adequately described by considering them to be short one-dimensional cables that have sealed ends (see Refs. 15, 16). The smooth muscle layer of the arterioles form electrical syncytia that have electrical length constants of ~900 µm and membrane time constants of ~80 ms.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The vessels examined in this study had many structural features found in other arterioles. Their intima was made up of a single layer of endothelial cells, and the media comprised essentially a single layer of smooth muscle cells (see Refs. 5, 6, 23). The internal elastic lamina was far less pronounced in iridial arterioles than in many other arterioles (20). Interdigitation of adjacent endothelial cells or adjacent smooth muscle cells characteristically described in other blood vessels was also observed (see Refs. 3, 17, 22). When intracellular recordings combined with dye injections were made from iridial arterioles, our experiments suggest that we were only able to impale cells in the smooth muscle layer. This is in clear contrast to a previous study (24) on arterioles of the hamster cheek pouch where endothelial cells were readily labeled. When injected with lucifer yellow, individual iridial smooth muscle cells were visualized; as described previously (19), lucifer yellow failed to pass from cell to cell.

The electrical studies show that individual smooth muscle cells were invariably electrically coupled to neighboring cells. Thus the resting membrane potentials of any two cells in a given preparation were very similar. When myogenic activity was detected in one arteriolar smooth muscle cell, similar potentials were detected in nearby muscle cells. Sympathetic nerve stimulation evoked depolarizations sensitive to blockade by -adrenoceptor antagonists (11) in each arteriolar smooth muscle cell. Current pulses passed through one intracellular electrode invariably produced membrane potential changes in neighboring smooth muscle cells. These observations indicate that iridial arteriolar smooth muscle cells, similar to all other vascular tissues examined to date (14), are coupled together to form an electrical syncytium. When sections of an iridial arteriole were isolated, both the membrane time constant and electrical length constant of the arteriole could be determined. The vessels had membrane time constants of ~80 ms. This suggests that the specific membrane resistance of these cells is lower than in most other vascular tissues (14, 15). The electrical length constants of rat iridial arterioles were ~900 µm, again a value less than that determined in guinea pig arterioles [~1.6 mm (15)]. However, the calculated values of axial resistance for guinea pig and rat arterioles are very similar (compare Table 1 with Table 1 in Ref. 16). Thus the lower value of length constants determined for iridial arterioles reflects the ease with which current escapes across the wall of the iridial arterioles rather than poorer conduction between adjacent cells.

The axial resistance is dominated by the resistance of the electrical pathways between adjacent iridial smooth muscle cells (for a discussion, see Ref. 16). Because the measures of axial resistance are similar in the two distinct arteriolar beds, it seems likely that electrical connections between adjacent muscle cells have similar values in both sets of arterioles. Presumably, electrical coupling between muscle cells occurs at the areas of close contact that were readily detected in the ultrastructural studies. The contacts between smooth muscle cells found in iridial arterioles were similar to those previously described in other vascular beds (for example, Refs. 23, 25), with the adjacent membranes at the sites of these communicating junctions showing electron-dense thickenings.

In iridial arterioles, numerous junctions were found between adjacent endothelial cells (Fig. 2). Although such junctions, unlike those between adjacent smooth muscle cells, frequently lacked membrane specializations, they presumably represent the sites where electrical coupling between endothelial cells occurs. Coupling between nearby endothelial cells has been demonstrated in pig coronary arteries and hamster cheek pouch arterioles with dye injections (19, 24) and implied by the demonstration of gap junction proteins in the endothelial layer of several different arterioles (18). Unfortunately, we were unable to directly confirm coupling between adjacent endothelial cells because we were unable to make recordings from this layer.

In most vessels, it seems likely that smooth muscle and endothelial cell layers are electrically coupled together (1). The idea that the endothelial and smooth muscle layers might be coupled arose from structural studies: numerous projections from the endothelial layer formed organized junctions with nearby smooth muscles (23). Direct evidence for electrical continuity comes from studies made on the pig coronary artery (27) and hamster cheek pouch arterioles (28). In the latter vessels, the resting membrane potentials of the endothelial cells were similar to those of the smooth muscle cells, and responses to nerve stimulation were detected in cells subsequently found by use of dye injections to be endothelial cells. When functional connections between cell layers were examined with dye movements, it became apparent that the selection of dye influenced the result. Thus, although Segal and Beny (24) were able to demonstrate the movement of lucifer yellow from endothelial cell to endothelial cell in hamster cheek pouch arterioles, they were unable to demonstrate dye movement between neighboring smooth muscle cells or between the endothelial and muscle layers. In a more extensive study with several dyes, Little et al. (19) demonstrated that only a limited number of dyes moved between cells of cheek pouch arterioles. With the appropriate dyes, movements from endothelial cells to smooth muscle cells but surprisingly not in the reverse direction were detected.

Our structural studies suggest that the junctions normally found between the smooth muscle cells and the endothelial cells in many arterioles are absent in iridial arterioles. Thus myoendothelial junctions, where bulbous projections of endothelial cells or smooth muscle cells penetrate the internal elastic lamina and basal lamina to become enclosed within the adjacent target cell (see Ref. 23), were not observed. A close association between the smooth muscle cells and the endothelial cells without an intervening basal lamina was only observed on one occasion. However, the other types of projections and contacts between smooth muscle cells and endothelial cells described in other blood vessels (23) were detected in iridial arterioles. It could be argued that the lack of myoendothelial junctions in iridial arterioles reflects the fact that the preparations were stretched before fixation. This was done initially to make it possible to obtain electrical recordings. For consistency, tissues for the structural studies were prepared in a similar way. It is possible that the form of myoendothelial junctions may have been disrupted by this procedure. This seems unlikely because adhesive junctions have been shown to be resistant to physical damage (12) and tissue stretch (26). Moreover, myoendothelial junctions would have to have been selectively disrupted because cell-to-cell coupling was maintained in both the smooth muscle and endothelial cell layers.

Our finding that close myoendothelial junctions are not present in iridial arterioles raises the possibility that coupling between the muscle and endothelial layers might not occur in this vessel. If this were the case, then endothelial-derived factors such as nitric oxide (21) and endothelium-derived hyperpolarizing factor (7) would have to diffuse through the extracellular space to influence the excitability of the smooth muscle cell layer. Perhaps the vesicle-containing regions of apposition detected between smooth muscle cells and endothelial cells serve as sites for the release of endothelium-derived factors.

In other arteries and arterioles, the electrical connections between neighboring smooth muscle cells permit current flow between cells. Each cell is depolarized during sympathetic nerve activity, thereby allowing the activation of voltage-dependent Ca²⁺ channels in their membranes (14). In iridial arterioles, it seems likely that constriction is triggered after the formation of a second messenger substance, probably inositol 1,4,5-trisphosphate that, in turn, triggers the release of Ca²⁺ from internal stores (10). The responses to nerve stimulation are little affected by organic Ca²⁺-channel antagonists, suggesting that depolarization and Ca²⁺ entry play only a minor role in the initiation of contraction. Our observations raise the possibility that the electrical connections between iridial arteriolar smooth muscle cells serve as diffusional pathways to allow the passage of a messenger substance such as inositol 1,4,5-trisphosphate or Ca²⁺ between cells so that each cell may contribute to the vasoconstriction. Such intercellular diffusion of second messengers has been shown in cultured smooth muscle and epithelial cells (2, 8).