Effects of temperature on the wall strength and compliance of frog mesenteric microvessels

Abstract

1. In single perfused mesenteric microvessels of pithed frogs, we assessed wall strength from the critical pressure, P_B, which has to be applied within the vessel in order to induce openings in the walls through which fluid and cells can extravasate. P_B was determined in capillaries and venules of tissues at 12-20°C and after cooling to 0-5°C.

2. The P_B (mean ± s.e.m.) in 22 vessels between 12 and 20°C was 52.64 ± 3.86 cmH₂O. In 19 microvessels cooled to less than 5°C, P_B was 92.0 ± 7.40 cmH₂O which was significantly higher than at room temperature (P < 0.001).

3. The compliance of the vessel wall was estimated using both the red cell method and the oil meniscus technique. There was no measurable effect of temperature on wall compliance. The compliance of vessels from which the cells had been removed by previous perfusion with detergent solutions was very similar to that of intact vessels between 12 and 20°C and between 0 and 5°C.

4. The negligible effects of temperature upon compliance suggest that microvessel walls have to be distended to a greater extent in cold tissue before P_B is reached. This, together with their rapid closure, is consistent with the hypothesis that pressure-induced openings in microvascular walls are dependent on an active response of the endothelium rather than being the result of stress failure of the basement membrane.

When the pressure in a closed-off capillary or venule is raised to a critical value, P_B, gaps open in the vessel wall through which fluid and cells can extravasate into the tissues. The sudden development of these openings is usually interpreted as rupture or stress failure of the vessel wall (West et al. 1991). Since the wall strength of microvessels is believed to be determined by the basement membrane and surrounding interstitial tissue (e.g. Swayne et al. 1989), P_B has been used to estimate the tensile strength of the basement membrane (West et al. 1991).

Previously, we have reported that most of the openings which are induced by high pressures in frog mesenteric vessels pass through (rather than between) the endothelial cells (Neal & Michel, 1996a). Furthermore, the openings close when the pressure is lowered so that the permeability and strength of the vessel wall (as indicated by further estimates of P_B) are recovered within a few minutes. The rapid recovery of wall strength suggests that the opening of gaps in a given microvessel at a particular value of P_B might be related more to the properties of the endothelium than to those of its basement membrane. To explore this hypothesis we have exploited the reversibility of the phenomenon to examine how temperature affects the value of P_B in single microvessels. We argued that if P_B is determined primarily by the properties of the endothelium, then it might be expected to be changed conspicuously when tissue temperature is dropped by 10°C or more. If, on the other hand, P_B is determined by the properties of the endothelial basement membrane, one might expect only a small effect of temperature on P_B.

When our results showed that P_B was greatly increased at low tissue temperature, we realised that this effect might be secondary to a reduction in wall compliance. If openings were induced by a critical strain upon the wall, a higher pressure would be necessary to reach this level of strain in a stiffer vessel. To test this hypothesis, we examined the effects of lowering the tissue temperature on the compliance of microvessel walls and the contribution which the endothelial cells make towards compliance.

Preliminary reports of some of the data in this paper have been presented to The Physiological Society (Neal & Michel, 1996b) and the British Microcirculation Society (Neal & Michel, 1996c).

METHODS

Preparation

The experiments were carried out on microvessels in the mesenteries of pithed frogs. The frogs were stunned by a blow to the head before the brain and upper part of the spinal cord were destroyed by pithing. The animal was then laid on a Perspex tray and a loop of intestine was delivered through an incision in the abdominal wall lateral to rectus abdominus. The intestine was carefully arranged around a short Perspex pillar so that the mesentery, which was spread over the polished upper surface of the pillar, could be transilluminated. The upper surface of the mesentery was washed continuously with frog Ringer solution and the temperature of the underside of the mesentery was monitored by a thermistor which was fixed to the pillar. The temperature of the superfusate was regulated by passing it through a glass heat exchanger immediately before it flowed onto the tissue. The temperature of the cooling fluid in the heat exchanger was kept close to room temperature for control measurements but could be reduced to 0°C within 2 min by switching to a precooled ethylene glycol solution. The mesentery was viewed through a Wild M10 stereomicroscope with a Hitachi CCTV camera attached to the camera tube. The output of the camera was displayed on video monitors and recorded.

Solutions

The Ringer solutions which were used as perfusates and superfusates were made from the same stock solution which had the following composition (mmol l⁻¹): NaCl, 111.1; KCl, 2.4; MgSO₄, 1.0; CaCl₂, 1.1; NaHCO₃, 10; glucose, 5.5. The superfusates were bubbled with 3 % CO₂ in O₂ and the pH was adjusted to 7.3-7.4 by adding more NaHCO₃. Bovine serum albumin (BSA) was added to the Ringer solutions which were used as perfusates at a concentration of 1.0 mg ml⁻¹. Immediately before the perfusates were used, a small number of washed human red cells were added to give a haematocrit of 2–4 %.

Measurement of P_B

Straight mid-capillaries and venules (15-25 μm in diameter) were selected for this study. With the tissue at room temperature, a chosen vessel was cannulated with a sharpened glass micropipette filled with the frog Ringer solution containing BSA and a few washed red cells. The micropipette was connected to an adjustable water and mercury manometer so that varying pressures could be applied to the vessel (Michel et al. 1974). To measure P_B, the vessel was first occluded with a fine glass microrod at a point several hundred micrometres downstream from the cannulation site but before the next branch of the vessel. The pressure applied to the micropipette was then raised to 30–40 cmH₂O and thereafter in a series of steps of 10 mmHg (13.6 cmH₂O), each of which was applied for 10 s until the vessel appeared to rupture. ‘Rupture’ was said to occur when red cells moved suddenly and rapidly to one or more sites along the vessel, passing through the wall into the tissue here. The pressure at which this occurred was P_B (Neal & Michel, 1996a).

The pressure was then lowered, the occluder removed and the vessel was perfused at low pressure (15-20 cmH₂O) for 10 min. Towards the end of this period the temperature of the superfusate was lowered and when the tissue temperature became steady in the range 0-5°C, the experiment was repeated.

Measurement of compliance

We have expressed wall compliance as the change in vessel radius for a given increment in intraluminal pressure (μm cmH₂O⁻¹). When given with the absolute values of vessel radius and the pressure range over which the change of radius occurred, it provides sufficient information to calculate other estimates of wall compliance (e.g. Young's modulus, tangent modulus, etc.) in an unambiguous way. When the compliance of two groups of vessels is compared, numerical differences in compliance may result from different vessels having different radii. To avoid this, we used the circumferential elastic modulus, E, to compare differences in compliance in different vessels. E was calculated from the expression:

where r is the vessel radius, Δr is the change of radius following a change in pressure of ΔP and Θ is the thickness of the vessel wall which was taken to be 0.1 μm (Swayne et al. 1989).

Two methods were used to determine compliance.

Red cell technique

This method evolved from the use of red cells as flow markers to measure the rate of fluid filtration from occluded segments of single microvessels (Michel et al. 1974; Smaje et al. 1980; Kendall & Michel, 1995). A red cell in the perfusate within an occluded microvessel can be considered to mark the end of a column of fluid between itself and the occlusion site. As fluid is filtered from the vessel, the fluid column will shrink and the red cell will move towards the occlusion site. Providing the transmural pressure and diameter of the occluded segment are constant, the fluid filtration rate can be calculated from the velocity of the red cell. This velocity declines exponentially with time as the red cell approaches the occlusion site. If a step increase in pressure is applied to the vessel, the filtration rate increases and the red cell velocity now declines more quickly, the exponential constant being increased in proportion to the pressure (Michel et al. 1974). The step increase of transmural pressure also expands the vessel wall so increasing vessel diameter and this has the effect of shortening the column of fluid between the red cell and the occlusion site. The time constant of the elastic or compliance component to red cell movement is very much shorter than that of the component arising from fluid filtration (Michel et al. 1974; Smaje et al. 1980; Swayne et al. 1989; Kendall & Michel, 1995).

If, after a step change in pressure, the logarithm of a red cell's position within an occluded segment is plotted against time, the points, after a few seconds, are found to fall on a straight line. The slope of this line is determined solely by the filtration rate. Backward extrapolation of the line to the instant when the pressure was increased, yields a value for the position (lμm from the occlusion site) at which the red cell would have been immediately after the pressure had been raised if the expansion of the vessel had been instantaneous (Kendall & Michel, 1995). When this position is compared with the actual position which the red cell occupied immediately before the pressure was increased (l₀μm from the occlusion site), the increase in vessel radius resulting from expansion of the vessel can be calculated, i.e.:

(1)

where r₀ is the vessel radius before the pressure was raised and r is the radius after the pressure was raised. Similar calculations can be made when the pressure is reduced in steps and in this way the compliance of the vessel wall can be determined. Values of r₀ were estimated as the mean of at least four measurements of radius from the video-recordings which were also used to follow the changes in l.

Oil meniscus method

This method, which was introduced by Swayne et al. (1989), uses changes in the length of a column of oil of constant volume within a microvessel with changes in the applied pressure in order to determine the compliance of the vessel wall. The principle is exactly the same as with the red cell method except here there are no losses of the oil through the vessel wall so that the changes in column length are not complicated by filtration. To form the oil column we used the same two immiscible fluids as Swayne et al. (1989): (a) a silicone oil mixture of 75 % silicone fluid DC200/0.65 cs (Dow Corning Ltd, Midland, MI, USA) and 25 % liquid paraffin; and (b) fluorocarbon 80 (FC80) (M-6015, 3M Company, Minneapolis, USA). The vessel was first filled via a micropipette with FC80 and then occluded to retain the oil while the micropipette was replaced by one containing the silicone oil mixture. The occluder was then raised momentarily to allow a small volume of the mixture to flow into the vessel and form a column with an oil-oil meniscus 200-400 μm from the micropipette. Steps in pressure applied to the micropipette containing the silicone oil mixture were transmitted to the oil within the microvessel and from video-recordings of the experiment, changes in the length of its column, which varied reciprocally with the vessel volume, could be followed from changes in the position of the meniscus.

It is worth noting the potential accuracy of these estimates. If, with a step increment in pressure, r increases to (r+Δr) while l is reduced to (l–Δl), then since Δr is small compared with r:

Using the M10 microscope with optimal lighting Δl may be resolved from the video-recordings to ±1.0 μm (as judged from measurements from video-recordings of a Leitz micrometer scale). Taking r = 10μm and l = 400μm (minimum), Δr =±(1/80) μm =±0.0125 μm. Estimates of compliance reported in this paper are based on values of Δl following a 10 cmH₂O change in pressure. Thus the limiting accuracy of an estimate of compliance is ±0.00125 μm cmH₂O⁻¹.

Removal of endothelial cells from microvessels

The endothelial cells lining a vessel were removed by perfusing the vessel via a micropipette with 1 % (w/v) Triton X-100 (octyphenoxypolyethoxyethanol, Sigma) in frog Ringer solution. In later experiments, a mixture of 1 % Triton X-100 and 0.5 % (w/v) sodium deoxycholate in Hepes (5 mmol l⁻¹)-buffered NaCl solution (0.9 %) was used to lyse the endothelial cells. Perfusions with these detergent solutions were continued for 5-10 min. At the end of such a perfusion, the detergent-containing micropipette was removed and the vessel recannulated and filled first with the FC80 and then with the silicone-oil mixture. Estimates of compliance of the vessel wall (stripped of its endothelium and pericytes) were then made.

Electron microscopy

Electron micrographs were prepared of some vessels which had been perfused with Triton X-100 and deoxycholate to demonstrate that perfusion with detergent did remove the endothelial cells. After the determination of their wall compliance, these vessels were fixed in situ, by flooding the upper surface of the mesentery with cold 2.5 % glutaraldehyde in 0.1 M cacodylate buffer. Some microvessels which had been perfused with oil but not detergent were also prepared for electron microscopy in a similar way. The position of the perfused vessel in the fixed tissue was recorded by drawings and video recordings. The mesentery was then dissected away from the animal and immersed in buffered 2.5 % glutaraldehyde for a further 2 h at 4°C, stored in fixative overnight and prepared for embedding in Araldite blocks using standard techniques. The detailed arrangement of blood vessels in the mesentery was clearly visible in the block which was then trimmed around the perfused microvessel. Transverse ultra-thin sections (50 nm) of the vessel were then cut using a Reichert-Jung Ultracut E. The sections were transferred to copper slot grids with Formvar-carbon support films and stained with uranyl acetate and lead citrate before being viewed using a Jeol 100 CX electron microscope.

Presentation of data

The data for P_B were subjected to Rankit analysis to determine whether it was reasonable to treat them as being distributed normally (Wardlaw, 1985). Their values are presented as means ±s.e.m. unless otherwise stated. Student's paired and unpaired t tests were used to assess the significance of differences between mean values of P_B, the level of significance being set at P < 0.05. Analysis of variance was used to assess the homogeneity of estimates of compliance in different vessels over different ranges of pressures. Differences between mean values of compliance and elastic modulus were assessed using paired and unpaired t tests.

RESULTS

Effect of reduced tissue temperature on P_B

In an initial series of experiments, estimates of P_B were made on microvessels at 0-5°C and compared with values obtained in similar vessels at room temperature (16-20°C). In seven vessels at 0-5°C, P_B was found to be 105 ± 16.1 cmH₂O. This was significantly greater than the value for ten different vessels (from 6 frogs) at room temperature where P_B was 54.6 ± 5.1 cmH₂O (P < 0.05, unpaired t test). The results of these experiments are summarised in Fig. 1A.

Figure 1. Values for P_B, the pressure necessary to induce openings in the walls of single microvessels at 12-20°C (▪) and between 0 and 5°C (□).

A, estimates (means ±s.e.m.) of P_B made on ten vessels at room temperature and seven different vessels in cooled tissues (*P < 0.05, Student's unpaired t test). B, estimates (means ±s.e.m.) of P_B on 12 vessels between 12 and 20°C and on the same vessels between 0 and 5°C (**P < 0.0005, Student's paired t test).

Since the openings in microvascular walls, induced by high pressures, close rapidly when the pressure in the vessels is lowered (Neal & Michel, 1996a), we carried out a series of measurements in which P_B was estimated at 12-20°C and also at 0-5°C on the same vessel. In 12 vessels (from 5 frogs) where such paired comparisons could be made, P_B had a mean value of 51.0 ± 5.8 cmH₂O at 12-20°C and 84.4 ± 6.9 cmH₂O at 0-5°C (Fig. 1B). A paired t test revealed that this increase of P_B on cooling the tissues was highly significant (P < 0.0005). When the data from these paired measurements of P_B are added to the unpaired measurements, the mean value of P_B at room temperature is 52.64 ± 3.86 cmH₂O and that in cooled tissue is 92.0 ± 7.40 cmH₂O. Ignoring the paired nature of some of these data and comparing all the measurements at 12-20°C with those at 0-5°C, the difference between the mean values of P_B is highly significant (for 39 degrees of freedom, d = 4.69 and P << 0.001).

The changes in P_B with temperature were independent of whether the temperature was raised or lowered. In some experiments, after a second estimate of P_B had been made on a vessel, the tissue temperature was changed yet again and further estimates of P_B were made. Figure 2 illustrates the results from one of these experiments where successive estimates of P_B were made at high and low temperature. It demonstrates how reversible both the effects of pressure and the effects of temperature might be on a single microvessel.

Figure 2. Repeated estimates of P_B made on the same vessel in warm and cool tissues.

Values of P_B are plotted against time. After an initial estimate of P_B at 19.7°C, perfusion pressure was lowered to allow the vessel to recover. The tissue was then cooled and P_B remeasured. Estimates of P_B were repeated after warming and cooling the tissue. Horizontal bars indicate periods over which the tissue temperature was held constant for the estimation of P_B. The temperature of the tissue was changing during the intervals between these periods.

The effects of temperature on compliance of microvascular walls

If the vessel wall is stiffer at low temperatures, the greater value of P_B might represent the greater stress required to extend the vessel to a point at which openings develop in its walls.

To test this hypothesis, we first estimated the compliance of the microvascular wall from the movements of the red cells with step increments in pressure in recordings from a subset of six of the paired experiments in which P_B had been determined at room temperature and after cooling the tissue. In four of the experiments P_B had been determined first at room temperature and then after the tissue had been cooled to below 5°C. In the other two experiments, P_B had been determined first in the cold and then after warming to 13.7-14.6°C. In these and all subsequent experiments, r₀ was re-measured after the tissue temperature had been changed and this value was used in the calculation of Δr at that temperature.

Although the experiments had not been designed to estimate wall compliance, the movements of the red cells with step increments of pressure between 15 and 30 cmH₂O allowed wall compliance to be estimated over this range in all six vessels both at room temperature and in the cooled tissue. The mean initial vessel radius (r₀) was 8.92 ± 0.90 μm and at room temperature the compliance, expressed as the mean increase in radius per unit increase in pressure (Δr/ΔP), was 0.023 ± 0.004 μm cmH₂O⁻¹. In cooled tissue, the compliance over the same 15 cmH₂O increment in pressure was 0.018 ± 0.002 μm cmH₂O⁻¹. A paired t test indicated the effects of temperature on compliance were not significant.

Attempts to measure compliance at higher pressures were frustrated by short-lived increases in fluid filtration which we have previously suggested may represent the opening and closing of pressure-induced pores in the endothelium (Neal & Michel, 1996a). These transient phenomena magnify the error of estimates of filtration rate, increasing the uncertainty of calculated changes of vessel volume following step increases in pressure. In three of the six vessels, however, the transients were less frequent than in the other three and here we were able to estimate compliance. It was found that compliance decreased as pressure was raised, falling from 0.012 to 0.004 μm cmH₂O⁻¹ as pressure was raised from 45 to 100 cmH₂O. Figure 3 shows results of one of these experiments. This not only indicates the similarity of compliance at room temperature and in cooled tissue but also demonstrates that under warmer conditions, P_B occurred at a lower vessel radius than when the tissue was cooled.

Figure 3. The effect of cooling on the compliance of a mesenteric capillary.

The vessel radius has been plotted against the intravascular pressure as the latter was raised to P_B first with the tissues at a temperature of 17.6°C (•) and then after the tissue temperature had been lowered to 4°C (○). The values of P_B are indicated by the vertical arrows. After the initial determination of P_B, the vessel was perfused at low pressure for 11 min to allow its wall to recover before the tissue temperature was lowered. Changes in vessel radius were estimated using the red cell technique.

These findings were supplemented by using the red cell technique to estimate compliance in four additional vessels as pressure was increased and decreased over the ranges 15-25 cmH₂O and 25-35 cmH₂O both at room temperature and after the tissue had been cooled. Two to three estimates of compliance were made in each vessel at room temperature and below 5°C. The results of these experiments are summarised in Table 1. Here it is seen that compliance is lower between 25 and 35 cmH₂O than it is between 15 and 25 cmH₂O. Cooling, however, did not decrease the compliance of the vessel significantly when pressure was being raised. When pressure was lowered from 35 to 25 cmH₂O the reduction in the vessel radius was significantly less in the cooled tissue than at room temperature (paired t test, P < 0.01). This was not true when pressure was reduced from 25 to 15 cmH₂O. In Table 1, the changes in vessel radius with increments and decrements of pressure indicate that the vessel radius was not restored immediately to its initial value when pressure was lowered. Thus the vessel wall did not behave like a perfectly elastic structure but showed hysteresis of its stress-strain relation (Swayne et al. 1989; Baldwin & Gore, 1989).

Table 1.

The effect of tissue temperature on wall compliance.

	Δr/ΔP
ΔP (cmH2O)	At 14–18 °C (μm cmH2O−1)	At 0–5°C (μm cmH2O−1)
15 → 25	0·317 ± 0·0265	0·3708 ± 0·0437
25 → 35	0·173 ± 0·0432	0·2069 ± 0·0207
35 → 25	0·177 ± 0·0507	0·1232 ± 0·015
25 → 15	0·228 ± 0·0431	0·2844 ± 0·0527

ΔP, pressure change; Δr, diameter change. Δr/Δp, compliance. Values are given as means ±s.e.m.

Comparisons of estimates of compliance with red cell and oil meniscus techniques

To establish whether the compliance of microvascular walls at pressures above 15 cmH₂O was determined by the basement membrane and the surrounding connective tissue rather than the cells, we investigated the compliance of vessels in which the endothelium had been destroyed and largely removed by perfusion with detergent solutions leaving the basement membrane intact. Compliance cannot be estimated by the red cell technique in such vessels because the hydraulic permeability of their walls is so high. It can, however, be estimated by the oil meniscus technique since the interfacial tension between the oil and the aqueous interstitial fluid should prevent the oil from passing through pores of 2 μm in diameter at pressures of less than several hundred centimetres of water.

In order to demonstrate that the two methods yield similar values for compliance, we compared estimates of compliance made by the oil meniscus technique and the red cell technique in the same four vessels from four different frogs. In each vessel the changes in vessel radius were estimated with 10 cmH₂O step increments in pressure from 15 to 45 cmH₂O, first using the red cell technique and then using the oil meniscus method. All measurements were made at room temperature (15-20°C). The results are summarised in Fig. 4, where the compliance (Δr/ΔP) estimated by the red cell technique has been plotted against the corresponding estimate made by the oil meniscus technique in the same vessel for the same increment in pressure. Although there is some scatter, there is a good correlation between the estimates (r = 0.8909, P < 0.001). The regression line through the points has a slope of unity and its small positive intercept does not differ significantly from zero.

Figure 4. Comparison of two methods for measuring wall compliance in four single microvessels.

In each vessel compliance was estimated by the red cell technique and the oil meniscus technique using the same steps in pressure. Values for compliance determined by the oil meniscus method over a given pressure increment have been plotted on the abscissa against the corresponding value obtained by the red cell (RBC) technique over the same pressure range in the same microvessel. The continuous line is the regression line (r = 0.88, P < 0.005); the dashed line is the line of identity. Different symbols indicate pressure range over which compliance was estimated: 15-25 (○), 25-35 (□) and 35-45 cmH₂O (▵).

Compliance of microvessels after removal of endothelium

The oil meniscus technique was then used to estimate the compliance of nine vessels from which the endothelium had been removed by preliminary perfusion with detergents (Triton X-100 and deoxycholate). In four vessels (mean radius 11.75 ± 1.175 μm) compliance was estimated both at room temperature and between 0 and 5°C for 10 cmH₂O changes in pressure between 20 and 50 cmH₂O. In five vessels (mean radius 14.30 ± 1.520 μm), compliance was estimated at room temperature and between 0 and 5°C with 10 cmH₂O changes in pressure between 15 and 45 cmH₂O. Similar findings were obtained from the two groups of experiments and the results of the second group are shown in Fig. 5. In neither group was compliance significantly changed by cooling the tissue and the mean compliance over the range 25-35 cmH₂O in one group (0.026 ± 0.008 μm cmH₂O⁻¹) was almost identical to that estimated from the other group over the range 30-40 cmH₂O (0.0249 ± 0.008 μm cmH₂O⁻¹). These values are similar to estimates of compliance obtained in intact vessels. They are slightly greater than the values given in Table 1 but this is a consequence of their larger radii. When the circumferential elastic modulus is calculated for each of the vessels shown in Fig. 5, these have mean values of 4.13 (± 1.35) × 10⁶ N m⁻² over the range 15-25 cmH₂O and 7.99 (± 1.85) × 10⁶ N m⁻² over the range 25-35 cmH₂O. Mean values of E for the vessels shown in Table 1 are 4.04 (± 0.98) × 10⁶ N m² over the range 15-25 cmH₂O and 11.13 (± 3.15) × 10⁶ N m⁻² over the range 25-35 cmH₂O. Analysis of variance indicated that the variations in E over a given pressure range were consistent with a single normal distribution. Unpaired t tests revealed no significant differences between the vessels with and without endothelium (over the range 15-25 cmH₂O, t = 0.0795 and for 25-35 cmH₂O, t = 1.34).

Figure 5. The compliance of microvessels from which the endothelium had been removed.

Mean values (±s.e.m.) for radius (relative to its value at 15 cmH₂O) from five microvessels are plotted on the ordinate against values of pressure applied to the lumen of the vessel via the micropipette. Determinations on each vessel were made using the oil meniscus technique at temperatures of 14-17°C (•) and 3-4°C (○). Endothelium had been removed by perfusion with 1 % Triton X-100 prior to estimation of compliance.

In one experiment, compliance was estimated by the oil meniscus technique in a single microvessel first with endothelium present and then later after the endothelium had been removed by perfusion with detergent. The results are shown in Fig. 6 where it is seen that treatment with detergent (with removal of the endothelium) had little effect upon the compliance of the vessel wall.

Figure 6. The compliance of a single microvessel before and after its endothelium had been removed by perfusion with detergent.

Compliance was estimated by the oil meniscus technique before (•) and after (○) perfusion with detergent solution to remove the endothelium. The mean value of the radius at 15 cmH₂O (r₀) was 14.5 ± 0.2 μm and was based on measurements of diameter at ten different sites along the length of the vessel before and after perfusion with detergent. Values for the radius at other pressures (r) are based on these initial values and three to four estimates of the changes in radius (calculated from the changes in length of the oil column with changes in pressure).

In further experiments we extended the range over which compliance was measured in vessels after removal of their endothelium. Figure 7 shows the relations between the extension of the circumference and the intravascular pressure for three vessels from which the endothelium had been removed. In two of these experiments the intravascular pressure was raised to 260 cmH₂O and in the other two to 160 cmH₂O without any evidence of leakage of oil into the extravascular space. Two experiments were conducted at low temperature but the third was carried out at room temperature. It is seen that in all three vessels, the compliance of the vessel wall, which here consists largely of basement membrane, falls as the pressure is raised from initial values in the range of 0.03 to 0.003 μm cmH₂O⁻¹ as pressure rises above 100 cmH₂O. There is no evidence for a sudden increase in compliance in the range of 50 to 150 cmH₂O that would be expected to encompass the range of P_B in intact vessels.

Figure 7. The compliance of the basement membranes of three microvessels over a wide range of pressure.

The relative change of radius of the vessels, which was estimated by the oil meniscus technique, has been plotted against the intraluminal pressure. The mean initial radius was 16.7 μm. Two of the experiments were carried out with tissue temperature at 4°C (○) and the third at 14°C (•).

Electron microscopy of detergent-perfused microvessels

No endothelial cells were seen in electron micrographs of sections of vessels that had been perfused with oil for measurement of compliance after perfusion with Triton X-100. In some areas from these vessels, however, there was debris covering the luminal aspect of the basement membrane. It seemed possible that this may have represented remnants of the endothelial cytoskeleton but we were unable to identify any specific structures here (Fig. 8A).

Figure 8. Ultrastructure of microvessels after perfusion with detergent solution and the measurement of compliance.

A, electron micrograph of part of a transverse section of a frog mesenteric capillary after perfusion with 1 % Triton X-100 for 7 min followed by perfusion with FC80 and the silicone oil mixture. No intact cells can be seen. Collagen fibres (c) of the mesenteric interstitium are clearly visible and the basement membrane (bm) can be seen at the interface between the interstitium (i) and the vessel lumen (l). B, electron micrograph to show the normal appearance of part of a transverse section of a frog mesenteric capillary. Here the microvascular endothelium (e) clearly defines the boundary between the interstitium and the vessel lumen. Scale bars, 500 nm.

Transverse sections of vessels which had been perfused with detergent but not subsequently filled with oil revealed the occasional ghostly remnant of an endothelial cell. Because they were never seen in the oil-perfused vessels, it is possible that perfusion with oil detached these ghosts sweeping them downstream. In support of this conjecture was the observation that when a vessel was filled with oil after detergent perfusion, the aqueous side of the oil-water meniscus acquired debris (probably cell fragments) as the first oil column was advanced through the vessel.

DISCUSSION

Our results show that when the temperature of frog mesenteric microvessels is lowered from room temperature to less than 5°C, the critical pressure (P_B) which is necessary to induce openings in the vessel is raised by 60-100 %. Although we have not investigated the effects on P_B of lowering temperature in a series of small steps, our data indicate that temperature has to fall below 12°C before a significant increase in P_B can be demonstrated.

By contrast, the compliance of the walls of these microvessels appears to be affected very little by changes of temperature between 0 and 20°C. Our estimates of compliance are very similar to those reported by Swayne et al. (1989) and Baldwin & Gore (1989). Swayne et al. (1989) summarised their data as Young's modulus, E, at low and high strain. Assuming a wall thickness (basement membrane) of 0.1 μm, they reported that at low strain (intraluminal pressure, 13.6 cmH₂O), E was between 2 × 10⁶ and 4 × 10⁶ N m⁻²; at high strain (intraluminal pressure, 54.4 cmH₂O) E was between 19 × 10⁶ and 30 × 10⁶ N m⁻². Carrying out similar calculations on the data of Baldwin & Gore (1989) leads to values for E in the range 0.7 × 10⁶ to 10 × 10⁶ N m⁻². For our data, changes in pressure between 25 and 45 cmH₂O fit comfortably in the range 5 × 10⁶ to 15 × 10⁶ N m⁻² both for intact vessels and for the vessels which were stripped of their endothelium. For the few experiments where pressure has been raised above 100 cmH₂O, E rises progressively as strain increases, rising well above 100 × 10⁶ N m⁻². Thus our data yield values for E which are comparable with those given by Swayne et al. (1989) and also with the data of Baldwin & Gore (1989).

We have also demonstrated that the cellular components of the microvascular wall make little contribution to its compliance when the intravascular pressure is greater than 15-20 cmH₂O. This finding is similar to that of Welling & Grantham (1972) and Welling & Welling (1978) on the compliance of renal tubules. While our experiments do not eliminate the possible contribution of non-cellular structures outside the basement membrane to the microvascular wall compliance, the calculations of Welling et al. (1995) suggest that the contribution of the basement membrane is dominant.

Since compliance is affected so little by cooling, we conclude that the vessel wall expands to a greater extent before P_B is reached in cold tissues than it does in tissues at room temperature. Although direct evidence for this conclusion has been difficult to obtain, the limited data we have obtained appear to support it (e.g. Fig. 3). There are at least two types of interpretation of this phenomenon. The first is that P_B is a measure of stress failure of the microvascular wall and when the tissues are cooled the tensile strength of the vessel wall is increased. Although we cannot discount this argument, there are reasons for doubting it at this stage. First, since the stresses which the vessel wall is subjected to as the pressure in the vessel is raised must be borne by the basement membrane, the stress failure at P_B should be that of the basement membrane. Within a few minutes of lowering the pressure, however, the strength of the vessel wall (as judged by P_B) is completely restored. It is difficult to see how the basement membrane could be repaired so quickly. Second, if the tensile strength of the basement membrane is increased at temperatures below 5°C, one might expect to see a change in the compliance of the basement membrane. We failed to see a change in compliance at low temperature and while this does not prove that the tensile strength of the basement membrane has not increased, it would seem to make it more likely that it is unchanged. A third reason for doubting that the effect of temperature on P_B is an increase in the tensile strength of the basement membrane is suggested by the measurement of compliance in vessels from which the endothelium had been removed at pressures of over 200 cmH₂O. The basement membranes of these vessels were subjected to stresses greater than those applied to intact vessels when the intravascular pressure is equal to P_B. If P_B is determined by stress failure of the basement membrane then unless these three vessels possess uncharacteristically strong basement membranes, one might have expected to see passage of oil into the tissues or some irregularity in the compliance curve.

An alternative interpretation arises if the development of pressure-induced openings in the microvascular wall represents an active response of the endothelial cells (Neal & Michel, 1996a). Here it can be argued that lowering tissue temperature increases the threshold stimulus necessary to trigger the cellular processes which lead to the formation of openings. Indirect evidence for an active role of the endothelium in the formation of pressure-induced openings in microvascular walls comes from studies by Parker and colleagues (Parker & Ivey, 1997; Parker et al. 1998) who have investigated the increases in permeability induced by increased stretch in positive pressure ventilated rat lungs. Parker & Ivey (1997) reported that isoprenaline attenuated this increased permeability, presumably by raising intracellular levels of cAMP. In a later report (Parker et al. 1998) described how the increased permeability associated with high pressure ventilation in isolated rat lungs could be prevented by infusion of gadolinium chloride. The latter was believed to be acting by blocking stretch-activated channels in the endothelium (e.g. Lansman et al. 1987) which triggered the response. Although we (Neal & Michel, 1998) have found that perfusion of single frog capillaries with gadolinium solutions does increase P_B, we were concerned that Gd³⁺ ions appeared to precipitate in our perfusates. At this stage, therefore, we are not confident that the Gd³⁺ ions were acting solely on stretch-activated channels in our preparation (Neal & Michel, 1998).

In spite of these reservations, the temperature sensitivity of P_B appears to complement the rapid recovery of permeability and wall strength when pressure is lowered, as indirect evidence for the view that openings in microvascular walls, which are induced by high pressures, arise from active response of the endothelial cells. This conclusion is also reinforced by the fact that the majority of the openings appear to be transcellular (Neal & Michel, 1996a). It remains to be seen whether more direct evidence will support this view.