The chronic effect of vascular endothelial growth factor on individually perfused frog mesenteric microvessels

Abstract

1. Hydraulic conductivity (L_p) of the wall of perfused microvessels has previously been shown to be chronically increased 24 h after a 10 min perfusion with vascular endothelial growth factor (VEGF). In order to investigate this further, L_p and the effective oncotic pressure difference (σΔΠ) acting across the vessel walls was measured before exposure to VEGF and 24 h later after the mesentery had been replaced in the abdominal cavity.

2. Acute 10 min perfusion with VEGF did not chronically change σΔΠ despite chronically increasing L_p 6.8 ± 1.2-fold. This suggests that pathways formed 24 h after perfusion with VEGF which increase hydraulic conductivity of the capillary walls have the same reflection coefficient as those present before VEGF.

3. Acute 10 min perfusion with VEGF significantly increased the diameter of vessels after 24 h by 48 ± 13%. To determine whether this was due to changes in the compliance of the vessel wall, the distensibility of microvessels was measured before and 24 h after perfusion with VEGF. The distensibility was increased 45 ± 15% by VEGF but this was not great enough to account for the increase in diameter.

4. The chronic increase in L_p could be attenuated by inhibition of nitric oxide synthase with l-NAME. In addition, the chronic increase in permeability was correlated with the acute response to VEGF (r = 0.71, P < 0.01) suggesting that the acute and chronic changes may be related.

5. These results show that VEGF chronically increases L_p without affecting the oncotic reflection coefficient. This may be due to reduced pore path length, or increased small pore numbers, which are properties of fenestrated capillaries. They also show that VEGF increases microvascular distensibility and diameter.

Vascular endothelial growth factors (VEGFs) are a family of at least six different peptides (VEGF A, B, C and D, and placental growth factors 1 and 2; Joukov et al. 1996), secreted by a variety of different tissues and cell types, which have far-reaching effects on the endothelium. They have been shown to result in angiogenesis, endothelial migration and mitosis (Connolly, 1991), extravasation of dye (Senger et al. 1986), vasodilatation, nitric oxide-mediated hypotension and increased blood flow (Ku et al. 1993), and a variety of other physiological and pathological processes (Dvorak et al. 1995). VEGFs have been shown to be over-expressed in many pathological conditions where capillary permeability is increased such as tumour vasculature (Brown et al. 1993) and psoriasis (Detmar et al. 1994). VEGFs are also expressed in normal adult kidney, and during wound healing, corpus luteum development and uterine wall thickening (Ferrara & Bunting, 1996).

Although VEGF was originally identified by a permeability assay (Senger et al. 1986), the time course, pharmacokinetics, cellular and molecular mechanisms of increased permeability, independent of haemodynamic factors, are still very poorly understood. Previous studies of vascular permeability have shown that VEGF results in an acute transient increase in hydraulic conductivity (L_p) lasting up to 2 min, followed by a chronic, sustained increase in L_p after 24 h (Bates & Curry, 1996). Although the cell signalling events underlying the acute response to VEGF have been investigated in vivo (Bates & Curry, 1997) and in vitro (Ziche et al. 1997), and the resulting changes in cell morphology and ultrastructure (Roberts & Palade, 1995; Feng et al. 1997) have been described, it is not clear how VEGF can chronically increase L_p. There have been no detailed investigations of the signal transduction mechanisms, the effect on microvascular protein permeability, or remodelling of the vascular wall under conditions which specifically address the chronic maintenance of the microvascular barrier to fluid and solute exchange in vivo. This study therefore sets out to investigate the chronic effects of VEGF on individually perfused microvessels in vivo. Parts of this work have been presented previously in abstract form (Bates, 1998a, b).

METHODS

Frog preparation

All experiments were carried out on male leopard frogs (Rana temporaria, 20–35 g) supplied by Blades, UK. All chemicals were purchased from Sigma unless otherwise specified. The animals were anaesthetized by immersion in 1 mg ml⁻¹ MS222 (3-aminobenzoic acid ethyl ester) in water, and anaesthesia was maintained by superfusion with 0.1 mg ml⁻¹ MS222 in frog Ringer solution (mm: 111 NaCl, 2.4 KCl, 1 MgSO₄, 1.1 CaCl₂, 0.20 NaHCO₃, 2.63 N-2-hydroxyethylpiperazine-N′-2-ethanesulphonic (Hepes) acid and 2.37 Hepes sodium salt). The animal was then laid supine and the limbs lightly secured. A small incision (8–10 mm) was made in the right lateral skin and muscular body wall. The distal ileum was floated out and carefully draped over a 1 cm diameter transparent Perspex pillar. The ileum was supported with cotton wool soaked in frog Ringer solution if necessary. The pH of this solution was 7.40 ± 0.02 at room temperature. The microvessels in the mesentery could be easily visualized under a Leitz inverted microscope (Leitz DMIL). A video camera (Panasonic WVBP32, 8 mm) was attached to the top of the microscope to allow binocular visualization and simultaneous recording of a 270 μm segment of the vessel (out of a total length of 800–2000 μm). The video was connected through an electronic timer (ForA VT33) to a video cassette recorder (Panasonic AG7350). The upper surface of the mesentery was kept continuously superfused with frog Ringer solution during the entire time that it was exposed. All experiments were done at room temperature (20–22°C).

Measurement of baseline L_p

The L_p of isolated perfused mesenteric microvessels was measured using the Landis micro-occlusion method previously described (Michel et al. 1974) which has been extensively discussed in the literature (Curry et al. 1983) and adapted to measure rapid changes in L_p, and chronic changes in L_p (Bates & Curry, 1996). Baseline L_p was defined as the conductivity during perfusion with 1 or 5% bovine serum albumin (BSA) in frog Ringer solution, adjusted to pH 7.4 with 0.115 mm NaOH. Microvessels were selected which had brisk blood flow, had no white cells adhering or rolling along the wall, and had a length of at least 800 μm with no side branches. Microvessels chosen for L_p measurement were either true capillaries (divergent flow at one end and convergent at the other), or first-order venules (convergent flow from two true capillaries at one end and convergent flow at the other), and had a diameter of 12–30 μm. The vessel was cannulated with a glass micropipette (pulled and ground from glass capillary tubes - o.d. 1.5 mm, Clark ElectroMed, UK) filled with 1 or 5% BSA in frog Ringer solution and rat red blood cells as flow markers. The rat red cells were collected by direct cardiac puncture of halothane-anaesthetized rats (5% halothane), and washed three times in frog Ringer solution before use. Rats were killed by cervical dislocation. The micropipette was clamped in a holder (WPI, Stevenage, UK) with a side-port attachment and connected to a water manometer. The vessel was occluded with a glass rod for 3–7 s, and then allowed to flow freely for at least 7 s before another occlusion was made.

Calculation of L_p

The transcapillary water flow per unit area of capillary wall (J_v/A) was calculated from the initial velocity of the red cells (dl/dt) after occlusion, the capillary radius (r) and the length between the marker cell and the point of occlusion (l), all of which were measured off-line from the videotape.

(1)

The hydraulic conductivity (L_p) was calculated from the Starling equation:

(2)

where ΔP is the effective hydrostatic and oncotic pressure difference between the capillary and the interstitium. For 1% BSA the capillary pressure was set at 30 cmH₂O, so ΔP was 26.4 cmH₂O (1% BSA has an effective oncotic pressure of 3.6 cmH₂O), assuming tissue pressure was negligible, and tissue oncotic pressure was equivalent to that in the superfusate (zero). For perfusion with 5% BSA the filtration rate was plotted against the applied pressure and L_p was calculated from the slope of the resulting pressure-flow relationship.

Measurements of L_p during perfusion with VEGF

After baseline L_p measurement a new pipette was filled with a solution containing 1 nm VEGF (Peprotech, Rocky Hill, NJ, USA), rat red cells and 1% BSA, in frog Ringer solution. The pipette containing the control perfusate was removed and the new pipette inserted into the holder and the pressure dropped to less than 1 cmH₂O. The vessel was then recannulated with the VEGF solution, and the pressure increased to 30 cmH₂O immediately, The vessel was occluded for 3–5 s as soon as possible after restoration of flow to allow L_p measurement. The occlusion was released and L_p measured approximately every 10 s over the next 3–5 min.

Measurement of oncotic reflection coefficient

The effective oncotic pressure difference (σΔΠ) and reflection coefficient (σ) were calculated from the filtration rates measured as described above at a variety of different pressures when the vessel was perfused with 5% BSA (Michel, 1980). Briefly, the vessel was perfused with 5% BSA in frog Ringer solution with rat red cells. The pressure was set to 30 cmH₂O, and the vessel occluded for long enough to allow subsequent measurement of filtration rate. The occluder was then removed, and the process repeated twice. This procedure was repeated for 40, 50, 45 and 35 cmH₂O. Measurements were not made for all vessels at all pressures due to the variation of filtration rate. Control measurements were made on vessels treated as above, but perfused with 1% BSA instead of 1 nm VEGF on day 1. The filtration rates were plotted against pressure, and a least-squares regression line fitted to the data. σΔΠ was calculated from the x-intercept of this line. The oncotic reflection coefficient was calculated as described by Michel & Phillips (1987):

(3)

Π_c (the oncotic pressure of the perfusate) was calculated using the relationship:

where C is the concentration of albumin in g (100 ml)⁻¹.

Measurement of distensibility

The distensibility of the vessel wall was measured by determining the distance moved by a marker red cell during a reduction in pressure from 30 to 20 cmH₂O. This is a slight variation of the previously published technique (Smaje et al. 1980). While the vessel was being perfused with 5% BSA the vessel was occluded with the pressure at 30 cmH₂O. After approximately 5 s, the perfusion line was switched to a manometer set at 20 cmH₂O by turning a three-way stopcock. The pressure was switched back about 3 s later and the process repeated. Each set of three measurements was repeated to give six measurements for each vessel. Distensibility (D) is defined as the change in radius (Δr) per unit change in pressure (ΔP):

(4)

The change in radius was calculated according to the relationship:

(5)

where l is the length of the column between the block site and the red cell, and x is the distance moved by the red cell during the pressure step. The subscripts 0 and 1 denote values at the higher and lower pressures, respectively. The change in radius was calculated assuming that the fluid in the vessel was incompressible, and that fluid filtration during the time of the pressure drop (4–8 ms) was negligible (Kendall & Michel, 1995).

Measurement of L_p, σ and distensibility on day 2

After the final occlusion, a map of the mesentery was made. All the visible microvessels in the connective tissue panel and the location of the arteries and veins crossing the mesentery were drawn. The gut was then replaced in the body cavity of the animal and the skin and body wall sutured. The animal was then untied, and placed on a plastic support in a tank with water partially covering the frog. It was allowed to recover, the water changed and the frog kept at room temperature. Twenty-four hours later (day 2) the frog was anaesthetized as above and the mesentery exposed. The same vessel as cannulated before was found from the map. The L_p and σ were measured as above for baseline solutions (5% BSA). The criteria for acceptable measurement of L_p on day 2 were essentially the same as those on day 1: the vessel must have freely moving red cells in it and no white cells sticking to the vessel. The differences between criteria for day 1 and day 2 were: there was no maximum L_p, there must have been at least 400 μm between the previous cannulation site and the last block site, and microvessels with sluggish or intermittent flow on day 2 were accepted. At the end of the experiment frogs were killed by destruction of the cranium (crushing), before recovery to consciousness.

Effect of arginine analogues

In vessels that had been exposed to VEGF on day 1 L_p was measured 24 h later during perfusion with the arginine analogues N^ω-nitro-L-arginine methyl ester (l-NAME) and N^ω-nitro-D-arginine methyl ester (D-NAME). In this case, baseline L_p was measured with 1% BSA on day 2, and then the vessel was recannulated and perfused with 100 μml-NAME or D-NAME and permeability measured over 5 min. Values given are measurements taken after at least 2 min perfusion with NAME.

Statistics

Measurements of L_p from control vessels have previously been shown to be non-normally distributed (Hargrave et al. 1995), therefore non-parametric statistics were used to compare and contrast actual L_p values (Wilcoxon paired test for difference and Spearman Rank for correlation coefficient). Compliance, reflection coefficient, radius and ratio data are given as means ± standard error of the mean (s.e.m.). A probability value P < 0.05 was accepted as significant.

RESULTS

Chronic effect of 1 nm VEGF on L_p

Measurements were made on twenty-two vessels perfused with 1 nm VEGF. Measurements during the first few minutes of VEGF perfusion were not made in one vessel for technical reasons, so peak L_p was only recorded on twenty-one vessels. VEGF resulted in an immediate and transient increase in L_p, as previously described. The L_p increased to a median of 9.1 (± 6.1) × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹ (n = 21, a mean increase of 7.1 ± 2.0-fold). Continued perfusion resulted in the L_p returning to baseline within 2 min. The vessels were perfused for 10 min with 1 nm VEGF. There was a significant increase in baseline L_p 24 h after (day 2) this 10 min perfusion with VEGF (on day 1) from 2.3 (± 0.9) × 10⁻⁷ to 9.3 (± 5.0) × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹ (n = 22, a mean increase of 6.8 ± 1.2-fold, see Fig. 1). The L_p did not increase in control vessels (Fig. 1, Table 1). The L_p on day 2 was significantly correlated with the peak increase on day 1 (r = 0.71, P < 0.01, n = 21, Fig. 2). The regression line was described by the relationship:

Figure 1. Acute and chronic effect of VEGF on L_p.

Hydraulic conductivity of 12 control (▪) and 22 c were first perfused with BSA (day 1, ), then with 1 nm VEGF (day 1, , are the peak values recorded during perfusion with VEGF), then 24 h later perfused again with BSA (day 2, ). **P < 0.01 compared with baseline on day 1.

Table 1.

Changes in capillary permeability, diameter and compliance brought about by VEGF

Parameter		Day 1	Day 2	Change
Baseline Lp (10−7 cm s−1 cmH2O−1)	VEGF	2.3 ± 0.9 (22)	9.3 ± 5.0 (22)**	7.3 ± 5.5
	Control	3.4 ± 1.7 (12)	4.1 ± 6.1 (12)	0.2 ± 3.5
σΔΦ (cmH2O)	VEGF	19.5 ± 1.4 (7)	21.0 ± 1.6 (7)	1.11 ± 0.14-fold
	Control	20.6 ± 1.1 (6)	18.9 ± 1.5 (6)	0.92 ± 0.06-fold
σ (no units)	VEGF	0.84 ± 0.03 (7)	0.87 ± 0.03 (7)	1.04 ± 0.06-fold
	Control	0.87 ± 0.02 (6)	0.83 ± 0.03 (6)	0.96 ± 0.03-fold
Diameter (μm)	VEGF	23.3 ± 1.3 (22)	32.9 ± 2.5 (22)**	48 ± 13%
	Control	23.6 ± 2.9 (12)	27.6 ± 2.4 (12)	28 ± 11%
Distensibility (nm cmH2O−1)	VEGF	28.2 ± 5.1 (7)	39.3 ± 7.1 (7)*	45 ± 15%
	Control	21.1 ± 3.8 (6)	24.8 ± 4.7 (6)	18 ± 7%

Median ± interquartile range (n) values for L_p, and mean ±s.e.m. (n) values for other parameters.

^*P < 0.05

^**P < 0.01 compared with day 1.

Figure 2. Relationship between the acute and the chronic increase in L_p.

The peak of the initial transient increase in L_p during perfusion with 1 nm VEGF (Peak L_p day 1), is plotted against the subsequent chronic increase in baseline L_p 24 h later during perfusion with 1% BSA. The relationship is described as: Day 2 L_p = 0.5219 × Peak day 1 + 5.929 (r = 0.71, P < 0.01, n = 21).

Baseline on day 2 = 0.52 × Peak on day 1 + 5.9294.

Chronic effect of 1 nm VEGF on oncotic reflection coefficient

In order to determine the nature of the transvascular pathways opened by acute perfusion with VEGF, the oncotic reflection coefficient of microvessels was measured before, and 24 h after, perfusion with 1 nm VEGF. Control experiments (e.g. Fig. 3A) showed that anaesthesia, exteriorization, cannulation and perfusion with control solutions did not change either the measured mean effective oncotic pressure difference (σΔΠ) (20.6 ± 1.1 cmH₂O on day 1, 18.9 ± 1.5 cmH₂O on day 2, n = 6) or the mean oncotic reflection coefficient (σ) calculated according to eqn (3) (0.87 ± 0.02 before perfusion, 0.83 ± 0.03 after perfusion, σΔΠ =σ²Π_c, see Discussion).

Figure 3. Measurement of reflection coefficient in a control vessel (A) and before and after VEGF (B).

A, pressure-filtration rate plot of a single control vessel on day 1 (▴) and day 2 (□). Neither L_p nor σΔΠ were changed (day 1: σΔΠ = 17.6 cmH₂O, L_p = 3.5 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹; day 2: σΔΠ = 16.6 cmH₂O, L_p = 2.1 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹). B, pressure-filtration rate plot of a single vessel before perfusion with VEGF (day 1, ▴) and 24 h after perfusion of VEGF (day 2, □). L_p was increased 12.8-fold, but σΔΠ was unaltered (day 1: σΔΠ = 21.2 cmH₂O, L_p = 1.2 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹; day 2: σΔΠ = 21.0 cmH₂O, L_p = 15.4 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹).

In seven vessels which were perfused with 1 nm VEGF, the mean effective oncotic pressure difference (σΔΠ) did not change 24 h after perfusion (21.0 ± 1.6 cmH₂O) compared with before perfusion (19.5 ± 1.4 cmH₂O, n = 7, P > 0.05). The mean reflection coefficient did not change either (σ = 0.84 ± 0.03 before perfusion, 0.87 ± 0.03 after perfusion). This was despite an increase in L_p from a median of 6.2 (± 5.7) × 10⁻⁷ to 26.5 (± 78.2) × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹. This is shown graphically in Fig. 3B, where a single typical experiment is shown. In this vessel, the L_p increased 12.8-fold from 1.2 × 10⁻⁷ to 15.4 × 10⁻⁷ cm s⁻¹ cmH₂O⁻¹, yet the effective oncotic pressure difference was almost identical on the second day (21.0 cmH₂O) and the first day (21.2 cmH₂O).

Effect of 1 nm VEGF on microvascular distensibility

Microvascular diameters were measured in twenty-two vessels with the pressure at 30 cmH₂O before, and 24 h after, perfusion with 1 nm VEGF, and twelve vessels not perfused at all with VEGF. There was no significant increase in the diameter of vessels not exposed to VEGF (see Fig. 4). Vessels perfused with 1 nm VEGF increased their diameter from 23.3 ± 1.3 to 32.9 ± 2.5 μm, (P < 0.01, n = 22). In order to determine whether the increase in diameter was due to an increase in distensibility of the microvascular wall, the distensibility was measured in six control vessels (perfused with 1% BSA instead of VEGF), and seven experimental vessels perfused with 1 nm VEGF on day 1. There was no significant increase in the distensibility of vessels not exposed to VEGF (see Fig. 4). Vessels perfused with VEGF on day 1 had a significantly higher distensibility on day 2. Distensibility increased from 28.2 ± 5.1 nm cmH₂O⁻¹ on day 1 to 39.3 ± 7.1 nm cmH₂O⁻¹ (n = 7, P < 0.05) on day 2. There was no correlation between the increase in distensibility and the increase in diameter (r = 0.56, P > 0.05).

Figure 4. Chronic effect of 10 min perfusion of 1 nm VEGF on distensibility and diameter.

Distensibility (left-hand axis and bars) and diameter (right-hand axis and bars) of control vessels (perfused with 1% BSA) on day 1 (□) and day 2 (▪), and vessels before perfusion with 1 nm VEGF () and 24 h after perfusion with VEGF (). *P < 0.05 and **P < 0.01 compared with day 1.

Effect of l-NAME on the chronic increase in L_p brought about by VEGF

In order to investigate the mechanism by which VEGF causes the chronic increase in permeability, vessels were perfused which had previously been exposed to 1 nm VEGF with the nitric oxide synthase (NOS) inhibitor l-NAME. An example of a typical experiment is shown in Fig. 5. Perfusion of five vessels with VEGF resulted in a 15.5 ± 5.1-fold chronic increase in L_p after 24 h (Fig. 6). Subsequent perfusion for 2 min with 100 μml-NAME on day 2 resulted in an attenuation of this increased L_p by 66 ± 5%, to 3.1 ± 0.7-fold compared with baseline on day 1, over a period of 2–3 min. In order to determine whether this effect was specific, experiments were carried out as above, but using the inactive enantiomer, D-NAME. This had no effect on the L_p (2.0 ± 10% change compared with baseline on day 2) (see Fig. 6).

Figure 5. Effect of 100 μm l-NAME on chronically increased permeability.

L_p measurements on a single microvessel before and during perfusion with 100 μm l-NAME, 24 h after perfusion with VEGF. The baseline on day 1 is shown. Perfusion of the vessel with l-NAME was started at time = 0 s.

Figure 6. Effect of arginine analogues on chronically increased permeability.

The effect of arginine analogues on permeability of experimental (□) or control (▪) microvessels. Hydraulic conductivities of the walls of vessels perfused with 1% BSA (Base, Day 1), and 1 nm VEGF (Peak, Day 1). Twenty-four hours later the vessels were again perfused with 1% BSA (Base, Day 2), and then perfused with 100 μm l-NAME (experimental group, □, n = 6) or D-NAME (control group, ▪, n = 5). * Significantly lower than baseline on day 2, P < 0.05.

DISCUSSION

The chronic increase in L_p and its relationship to the initial peak increase

Although vascular endothelial growth factor was first identified as a permeability factor in 1989 the first quantitative measurements of the increase in permeability brought about by exposure to VEGF in vivo was not described until 1996 (Bates & Curry, 1996). This was due to the widespread use of the Miles assay which does not distinguish between haemodynamic and permeability changes. The acute, transient increase was described in both single microvessels of frog mesentery perfused in situ (Bates & Curry, 1996) and isolated rat coronary venules (Wu et al. 1996). The chronic increase in permeability was described in frog mesenteric microvessels. Qualitative indications of a VEGF-induced chronic increase in microvascular permeability have been described in rat mesentery (Roberts & Palade, 1997), but the nature of the relationship between the acute and chronic increases has not been clearly understood. The data presented here suggest a relationship between the acute and chronic increase, since they are correlated. Although a direct mechanistic explanation for this relationship has not been described, there is circumstantial evidence that such a mechanism may exist. This evidence implicates the role of chronic upregulation of NOS 24 h after exposure to VEGF (Hood et al. 1998).

Chronic effect of 1 nm VEGF on the oncotic reflection coefficient

Michel & Phillips (1987) have shown that Starling's equation can be expressed as:

(6)

where Pe (the Péclet number) is the ratio of convective to diffusive flux. If Pe is sufficiently high then this reduces to:

(7)

Since J_v/A =L_p(ΔP -σΔΠ), then, if this assumption holds true, σΔΠ =σ²Π_c (Michel & Phillips, 1987). The validity of this relationship is therefore dependent on a sufficiently high Péclet number. The Péclet number can be calculated according to the equation Pe =J_v(1 - σ)/P_aA, where P_a is the diffusive permeability to albumin. Pe can therefore be calculated from the filtration rates measured in the experiments, and previously measured diffusive permeabilities for albumin (P_a = 2.3 × 10⁻⁷ cm s⁻¹) (Curry et al. 1990). If, in the worst case, a reflection coefficient of 0.95 is assumed, then Pe = 2.1. At this Peclet number the ratio of the reflection coefficient calculated according to the approximation (eqn (7)), to that calculated according to the correct interpretation (eqn (6)) reaches 0.99. The assumption that eqn (6) approximates to eqn (7) is therefore 99% valid.

Clough et al. (1988) have shown that the intercept of the relationship between σΔΠ and reciprocal of the relative increase in L_p, (L_p0/L_p, being defined as L_p before the increase relative to L_p after the increase) can be used to determine the σΔΠ, and hence σ, for the pathways which have opened up after perfusion with VEGF. When the values for L_p0/L_p were plotted against σΔΠ (Fig. 7) there was no significant correlation. Hence the reflection coefficient of the gaps opened up by perfusion with VEGF was not significantly different from the reflection coefficient on day 1, i.e. 0.84. The ultrastructural pathways that are stimulated to form by VEGF therefore have a significant reflection coefficient to albumin, like those present in the control state.

Figure 7. Relationship between effective oncotic pressure difference and the increase in hydraulic conductivity.

The regression line is shown (continuous line) σΔΠ = 0.8L_p0/L_p+ 19.7 cmH₂O. Dashed line shows the predicted regression line if the reflection coefficient of the new pores was zero, i.e. large unrestricting gaps.

Previous studies on the reflection coefficient of pathways opened up when various specific treatments are applied to vessels have shown two types of interaction. In rat mesenteric microvessels exposed to histamine, there is an acute increase in L_p, but a significant reflection coefficient to albumin, whereas exposure to serotonin resulted in a reduction in reflection coefficient consistent with the opening up of a pathway which was not reflective to albumin (Michel & Kendall, 1997). There are therefore precedents for the maintenance of a high oncotic reflection coefficient with a high L_p. There are a number of theoretical bases for an increase in L_p without changing reflection coefficient. These include a reduction in pore path length, without a change in either pore number or pore area (e.g. shortened intercellular cleft length, fenestrations, or transcellular gaps full of glycocalyx or fibre matrix), and an increase in the ratio of small to large pores.

This is the first quantitative description of the sieving properties of the capillary wall after perfusion with VEGF in vivo. Ultrastructural investigation of the microvascular walls during or after exposure to VEGF has previously been described, however. VEGF has been shown to result in the formation of fenestrated capillaries in vivo (Roberts & Palade, 1995) and to cause fenestrations in endothelial monolayers in vitro (Esser et al. 1998). Fenestrated vesicles, transcellular gaps (Neal & Michel, 1997), or unfenestrated, vesiculo-vacuolar organelles (Feng et al. 1997) have also been reported. The quantitative sieving properties of these ultrastructural changes during VEGF perfusion are unknown, but the reflection coefficient to albumin of fenestrated capillaries is 0.78 in both the knee joint (Knight et al. 1988) and the ascending vasa recta (Pallone, 1992), and the sieving properties of fenestrated endothelium of the glomerulus remain intact with a high reflection coefficient to albumin (Savin et al. 1992). These previously published findings are therefore consistent with the results described here. These data therefore provide quantitative information about the nature of the pores opened up by VEGF. They indicate that either the path length through the endothelial barrier has been reduced, or VEGF has induced the opening up of small pores, which have a high reflection coefficient to albumin. Possible ultrastructural correlates of this functional change include formation of fenestrations or fenestrated vesicles, reduction of the intercellular cleft length, and opening of transcellular gaps with reflective material plugging them.

Effect of 1 nm VEGF on microvascular diameter and distensibility

These results show that the diameter of vessels is increased 24 h after exposure to VEGF. These vessels are considered, according to flow patterns, to be either true capillaries or first-order venules. There was no difference in the increase in diameter between these two categories of vessel, nor was there any relationship between the initial diameter and the increase in the diameter of the vessels. Capillaries are classically considered to be non-contractile, but it is possible that they can in fact dilate or contract, and can do so in response to specific stimuli. That possibility was not investigated. Although diameter changes of capillaries have been reported in response to flow and pressure (Bosman et al. 1995), diameter changes with applications of agonists have not been widely reported in the literature. The diameter may have increased due to increased distensibility of the vessels. Indeed, these experiments showed that there was an increase in distensibility of the vessels 24 h after perfusion with VEGF. However, this was of the order of 11 nm cmH₂O⁻¹, whereas the increase in diameter was 9.6 μm. In order for the diameter to be increased from 23.3 to 32.9 μm solely by increased distensibility, the transmural pressure would need to be increased from 30 to 152 cmH₂O. The increase in distensibility is therefore unlikely to explain the increase in diameter. A third possibility for the increase in diameter would be increased number of endothelial cells forming the vessel wall. This possibility has not been investigated, but is likely, since VEGF is mitogenic for endothelial cells, and the cell cycle time for endothelial cells has been shown to be about 18 h (Ferrara & Bunting, 1996).

The physiological significance of increased vessel diameter is that the resistance of the capillary bed is determined by capillary diameter. According to Poiseuille's law, which states that the resistance depends on the fourth power of the radius, if the radius of the capillary were to change from 11.7 to 16.5 μm (an increase of 41%) the resistance would drop by more than 3.3-fold. This in itself may cause an increase in blood flow, and hence the nutrients available to tissue and stimulation of flow-dependent blood vessel growth.

The increase in distensibility, although not able to account for the increase in diameter, is still a significant finding. There have been no previous studies of individually perfused microvessels showing that distensibility can be increased by perfusion with agonists. The structural elements that determine the distensibility have previously been discussed (Swayne et al. 1989). It has previously been argued that most of the support for the vessels comes from the basement membrane (Swayne et al. 1989). Since VEGF has been proposed to cause secretion of interstitial collagenases and gelatinase A (Lamoreaux et al. 1998), this increase in the distensibility may be due to the breakdown of the extracellular matrix.

Effect of nitric oxide on chronically increased permeability

The mechanism by which L_p is chronically increased by VEGF is still poorly understood. Recent studies have shown that VEGF can increase intracellular calcium concentration in vivo, and that it is this increase in intracellular calcium concentration that results in acutely increased permeability (Bates & Curry, 1997). The increase in calcium has been shown to result in activation of mitogen-activating protein kinase (MAPK) in endothelial cells via NOS activation (Parenti et al. 1998). Endothelial NOS (eNOS) mRNA has been shown to be transcriptionally upregulated by MAPK, and eNOS mRNA is increased in cultured cells 24 h after exposure to VEGF (Hood et al. 1998). Although the role of nitric oxide production in the generation and maintenance of increased permeability is still unclear (He et al. 1997), there is a growing body of evidence to show that increased nitric oxide production in endothelial cells increases permeability (Rumbaut et al. 1995). The possibility that VEGF could chronically increase permeability through upregulating nitric oxide production was therefore investigated. The results described above show that the increase in permeability brought about by VEGF could be significantly attenuated by perfusion of the vessel with NOS inhibitors, but not by the inactive enantiomer of the inhibitor. It has recently been shown that the effect of NOS inhibitors on the permeability of isolated perfused microvessels is dependent on the initial permeability level (Ahmad & Michel, 1998). The permeability of vessels is reduced when the baseline is high, and increased when the baseline is low, showing that nitric oxide can act in at least two different ways to produce opposite results. The data shown above are consistent with the hypothesis that VEGF chronically stimulates nitric oxide production, and that this results in increased permeability.

Conclusions

The data presented in this paper show that 10 min perfusion with 1 nm VEGF chronically increases microvascular L_p without affecting the reflection coefficient to albumin. Vessels exposed to VEGF can therefore still maintain a high osmotic transmural pressure gradient. The chronic increase in permeability correlates with the acute response to VEGF, implying that the chronic effect may be a result of the acute effect. The data also show that the diameter of the vessels is significantly increased, but that this can only be attributed in part to an increase in distensibility. The increased diameter may significantly affect the perfusion of tissue exposed to VEGF. Finally, the data show that the chronic increase in L_p can be attenuated by perfusion with NOS inhibitors.