The experiments described in this paper extend our observation that the PK of perfused frog mesenteric microvessels increases with the velocity at which they are perfused, by demonstrating that this correlation is lost when the vessel is perfused with agents that are known to increase intracellular levels of cAMP. Perhaps the most significant conclusion to be drawn from this is that, since it can be inhibited by specific pharmacological agents, the relation between PK and U is a biological phenomenon and is not an artefact of measurement. This leads us to consider the possible mechanisms responsible for it. Changes in flow velocity are presumably detected by receptors which are sensitive to shear stress or shear rate and located either within or at the surface of the microvascular endothelium. Stimulation of these receptors may then activate a cascade of reactions which ultimately give rise to the increase in PK. While our experiments reveal nothing about the receptors and little about the consequences of stimulating them, they do appear to eliminate the involvement of both NO and prostacyclin as important members of the signalling cascade. Furthermore, although increased intracellular levels of cAMP might act in several ways to prevent microvascular permeability from increasing, the values of PK which we have estimated at high perfusion velocites raise important questions about the pathways which are responsible for raising potassium permeability under these conditions.
Lack of effects of L-NMMA and indomethacin on the PK-U relation
The failure of L-NMMA to influence the relation between PK and U contrasts with the report of Yuan et al. (1992) who found that L-NMMA abolished the flow dependence of albumin permeability in venules from pig heart. We, ourselves, were surprised by our finding. Since NO release from the endothelium appears to be an important step both in flow-induced vasodilatation of large vessels (Griffith, 1994) and in mediator-induced increases in permeability of small vessels (including those of the frog mesentery), it seemed likely that NO would be a component of the signalling cascade of flow-dependent permeability. When we found that L-NMMA failed to affect the changes in PK with U, we examined the effects of indomethacin, as Koller & Kaley (1990) have shown that, in the microcirculation of the rat cremaster muscle, prostacyclin rather than NO is responsible for flow-induced vasodilatation. Indomethacin also failed to influence the relation between PK and U.
We now believe that the failure of L-NMMA to influence the flow dependence of PK reflects a difference in the signalling cascades between frog and mammalian endothelia. We have recently shown that PK varies directly with flow in rat mesenteric venules but here the flow dependence is abolished with NOS inhibitors (Kajimura & Michel, 1998B).
Microvascular permeability and raised [cAMP]i
Our finding that increases in PK with U become insignificant when intracellular levels of cAMP are raised is consistent with many reports of cAMP tightening the barrier properties of endothelium. In single frog mesenteric microvessels, He & Curry (1993) reported that the increased hydraulic permeability induced by either ionomycin or ATP could be inhibited by perfusing the vessel with 8-bromo-cAMP for 15 min prior to applying these mediators. This inhibitory effect of cAMP occurred in spite of the increase in intracellular calcium concentration which followed the application of ionomycin. More recently, Adamson et al. (1998) have reported that, in frog mesenteric microvessels, the reduction in baseline hydraulic permeability, induced by perfusion with forskolin and rolipram, is associated with a small but significant increase in the number of tight junctional strands in the intercellular clefts.
If raising the [cAMP]i within microvascular endothelium increases the number of junctional strands in the intercellular clefts, it is reasonable to propose that the increments of PK with increasing U are the result of a reduction in the number of junctional strands, thus raising the fraction of the intercellular cleft available for exchange. While changes of this kind could be involved, there are quantitative difficulties in accepting them as being entirely responsible for the increase in PK with U. These difficulties arise from the absolute values of PK. If the intercellular clefts were the only pathways which open to raise PK as U increases, then PK should reach a maximum when there are no longer any junctional strands impeding the diffusion of K+ through the intercellular clefts. This maximum value of PK can be calculated from the diffusion coefficient (D) of potassium in free solution and the dimensions of the intercellular clefts in frog mesenteric microvessels, i.e.
where L is the length of open cleft per unit area of the endothelium, w is the width of the cleft, and Δx is the distance through the cleft from the lumen of the vessel to the endothelial basement membrane. Detailed ultrastructural studies of the intercellular clefts of frog mesenteric capillaries (Mason et al. 1979; Bundgaard & Frøkjær-Jensen, 1982; Clough & Michel, 1988) suggest that L is 2000 cm cm−2, w is 20 nm and Δx is 0.4 μm. Taking these values together with a value of 2 × 10−5 cm2 s−1 for D leads to the calculated maximum permeability of the clefts to K+ being 20 × 10−4 cm s−1. This figure is regularly exceeded by our estimates of PK at high perfusion velocities (e.g. see Fig. 1). While it is possible that, by incorporating too large a factor (α) into our calculation of PK to account for the rise in pericapillary concentration of K+, we may have overestimated the true value of PK, this does not resolve the dilemma as values of PK greater than 20 × 10−4 cm s−1 are calculated even when this correction is ignored (i.e. when α= 0). This leads us to conclude that the increase in PK with increasing U cannot be accounted for entirely by a reduction in the length of the junctional strands increasing the fraction of the intercellular cleft available for exchange.
PK could be increased if Δx, the distance through the cleft, was reduced as perfusion velocity increased. Clough & Michel (1988) found that some of the variation in the hydraulic permeability between individual frog mesenteric microvessels could be accounted for by a relation between hydraulic permeability and mean estimates of (1/Δx). If Δx were reduced to a minimum value of 0.1 μm, the maximum value of PK could increase to 80 × 10−4 cm s−1, providing, of course, that the junctional strands were reduced to mere points of contact between adjacent cells. The hypothesis could be tested by comparing the ultrastructure of the interendothelial clefts in vessels perfused at high perfusion velocities prior to fixation with those perfused at low perfusion velocities.
If it emerges that the increase in PK with U cannot be accounted for by reductions in Δx, alternative pathways through the endothelium will have to be considered. These would include the formation of fenestrae, the opening of transendothelial channels through the vesiculo-vacuolar organelles (Dvorak et al. 1996) or the opening of intercellular or transcellular gaps (Neal & Michel, 1995) in the endothelium.
The opening of gaps in the endothelium, whether they be intercellular or transcellular, is usually associated with an increased leakage of macromolecules and Yuan et al. (1992) reported increased permeability to serum albumin in response to increases in the perfusion rate of isolated coronary venules. In our experiments, however, there was no observable increase in the leakage of Evans Blue albumin (which was present in all normal-K+ perfusates) when U was raised. Since this would have been expected if albumin permeability increased with increasing flow it would seem that, if the permeability to macromolecules does increase in frog mesenteric vessels at high perfusion velocities, the increase is small. At this stage we are left to speculate on the nature of the pathways for K+ which are opened as flow increases.