On the disparity in osmolarity-induced vascular reactivity

Abstract

Conventional Peritoneal dialysis solutions (PDS) are vasoactive. This study was conducted to identify vasoactive components of PDS and to quantitatively describe such vasoactivity. Anesthetized nonheparinized rats were continuously monitored for hemodynamics, while the microvasculature of the jejunum was studied with in vivo intravital microscopy. In separate experiments, vascular reactivity of rat endothelium-intact and –denuded aortic rings (2mm) was studied ex vivo in a standard tissue bath. In both studies, suffusion of the vessels was performed with filter-sterilized isotonic and hypertonic solutions containing glucose or mannitol as osmotic agents. PDS served as a control (Delflex^® 2.25%). Hypertonic glucose and mannitol solutions produced a significant vascular reactivity in aortic rings, and instantaneous and sustained vascular relaxation at all levels of the intestinal microvasculature. Similarly, lactate dissolved in a low pH isotonic physiologic salt solution produced significant force generation in aortic rings. While isotonic glucose and mannitol solutions had no vasoactivity in aortic rings, isotonic glucose produced a selective, insidious and time-dependent vasodilation in the intestinal premucosal arterioles (18±0.2% of baseline), which was not observed in the larger inflow arterioles (100 μm). This isotonic glucose-mediated vascular relaxation can be attenuated by ∼50% with combined adenosine A_2a, A_2b receptor antagonists, and completely abolished by adenosine A₁ receptor inhibition. By using two different experimental techniques, this study demonstrates that hyperosmolality and lactate are the major vasoactive components of clinical peritoneal dialysis solutions. The pattern and magnitude of such reactivity is dependent on vessel size and on the solutes' metabolic activity. Low pH of conventional peritoneal dialysis solutions is not a vasoactive component by itself; but renders lactate vasoactive. Energy-dependent transport of glucose into cells mediates vasodilation of small visceral arterioles by an adenosine-receptor mediated mechanism, and constitutes a significant fraction of peritoneal dialysis solution-mediated vascular reactivity in the visceral microvasculature.

Introduction

Conventional peritoneal dialysis solutions dilate visceral and parietal microvasculature by mechanisms possibly related to hyperosmolality, low pH, and the buffer anion system of these solutions ^1;2. Studies with hyperosmolar sodium solutions perfused into the intestinal lymph, produced vasodilation of submucosal arterioles through a mechanism partially mediated by a hyperosmolality-induced nitric oxide (NO) release ³. Similarly, intravenous infusion of hypertonic galactose or mannitol solutions in pigs, increased the baseline hepatic blood flow by 37%, presumably by a mechanism attributed to an osmotic stress ⁴. Massett and colleagues have found that in vitro infusion of isolated, cannulated, and pressurized skeletal muscle arterioles with hypertonic solutions of glucose, sucrose, or mannitol at an osmolality of 330 mosmol/kgH₂O equally dilates these arterioles. This vascular reactivity appears to be an endothelium-dependent response, which is independent of the NO or the cyclooxygenase pathways, but can be nearly abolished by glibenclamide, an ATP-sensitive potassium channel (K_ATP inhibitor ⁵. Similar results were obtained with coronary arterioles perfused ex vivo with either hypertonic glucose or sucrose solutions. In these vessels only glibenclamide caused attenuation of the hypertonic solutions-mediated vascular relaxation. In addition, inhibition of inward rectifier potassium channels, or calcium activated potassium channels had no effect on hyperosmolality-induced vascular reactivity ^6;7. Our recent intravital videomicroscopy of the terminal ileum in rats, have shown that exposure of the ileum to a conventional peritoneal dialysis solution, produces an instantaneous and sustained vascular relaxation at all levels of the microvasculature, which is essentially associated with doubling of blood flow in the inflow arterioles (100 μm in diameter) ⁸. In the rat mesentery, conventional lactate-buffered peritoneal dialysis solutions preferentially dilate mesenteric arteries by >20%, passively double mesenteric arteriolar blood flow without change in diameter, and increase the number of perfused mesenteric capillaries by >20%. These vasoactive properties were independent of solutions pH, but occurred only transiently when the mesentery was exposed to conventional peritoneal dialysis solutions with low glucose degradation products. Furthermore, bicarbonate-buffered peritoneal dialysis solutions with low glucose degradation products were entirely nonvasoactive in the rat mesentery. Based on these findings the authors suggested that the vasoactive effects of clinical peritoneal dialysis solutions is exclusively due to their contents of glucose degradation products and lactate contents, and not due to hyperosmolality ⁹.

In a recent study we attempted to investigate the molecular mechanism of peritoneal dialysis solutions-induced vascular relaxation. In this study vascular rings from the aorta and superior mesenteric arteries were exposed to a conventional peritoneal dialysis solution under controlled conditions. The dialysis solution contracted these arteries by an endothelium-independent mechanism that involved a vascular smooth muscle-derived prostanoid pathway ¹⁰. We therefore hypothesize that the pattern and magnitude of vascular reactivity produced with conventional peritoneal dialysis and other hypertonic solutions, is determined primarily by the hyperosmolality and the solutes metabolic activity. We further hypothesize that the role of other components of the dialysis solution is rather subordinate. The aim of the present study is to identify the major vasoactive components of conventional peritoneal dialysis solutions under controlled experimental conditions. This is required prior to investigation of the molecular mechanisms and signal transduction pathways involved in such vasoactivity.

Material and Methods

General Animal Care and Surgery

Male Sprague-Dawley rats (Harlan, Inc. Indianapolis) were housed in AAALAC-approved facilities and were maintained on standard rat diet and water ad libitum for at least one week before use. All animal care and experimental procedures conformed to “Principles of Laboratory Animal Care” of the National Society for Medical Research and the “Guide for the Care and Use of Laboratory Animals” of the US National Academy of Science as published by the National Institutes of Health (NIH publication # 80-23, revised 1987) and were prior-approved by the Institutional Animal Care and Use Committee of the University of Louisville and the Louisville Veterans Administration. Experiments were performed on rats (200-210 grams) that had fasted overnight. Anesthesia was induced with intraperitoneal pentobarbital (60 mg/kg) and maintained with supplemental subcutaneous injection as needed. Body temperature was maintained at 37±0.5°C with a rectal probe and a servo-controlled heating pad. Surgery was carried out after loss of the blink and withdrawal reflexes. Tracheostomy was performed to reduce airway resistance and the animal was allowed to breathe room air. The right femoral artery was cannulated with a PE-50 catheter to provide continuous monitoring and online recording of arterial blood pressure.

Bathing solutions

I. Microvascular studies: All chemicals were purchased from the Sigma Chemical Company (St. Louis, Missouri, USA). The intestinal segment was continuously suffused during tissue preparation and equilibration with a nonvasoactive modified Krebs' solution that contained 6.92 g/L sodium chloride, 0.44 g/L potassium chloride, 0.37 g/L calcium chloride, and 2.1 g/L sodium bicarbonate at a pH of 7.4 and osmolality of 285 mOsm/L. Isotonic glucose and mannitol solutions were prepared as Tris-buffered physiologic salt solution (PSS). The isotonic glucose solution contained 36.29 mM Tris-HCl, 13.71 mM Tris-base, 11.1 mM glucose, 88.98 mM NaCl, 5.87 mM KCl, and 2.55 CaCl₂·2H₂O. The isotonic mannitol solution had the same components except that mannitol was substituted for glucose. Hypertonic mannitol solutions were prepared either as a 5% mannitol in-Krebs' solution, or as a 10% mannitol in deionized water. All solutions were filter-sterilized and pre-warmed to 37°C before use. A conventional 2.25% dextrose-based dialysis solution (Delflex^®, Fresenius USA, INC. Ogden, UT) that contained 5.67 g/L sodium chloride, 3.92 g/L sodium lactate, 0.257 g/L calcium chloride, 0.152 g/L magnesium chloride at a pH of 5.5 and an initial osmolality of 398 mOsm/L served as control for the hypertonic solutions.

II. Macrovascular studies: Vascular ring studies are typically conducted in a nonvasoactive physiologic salt solution (PSS). PSS composition in millimolars was 118 NaCl, 4.7 KCl, 2.5 CaCl₂, 1.2 KHPO₃, 1.2 MgSO₄, and 11.1 glucose. All aortic rings were maintained at 37 °C, and bubbled with a 95% O₂ and 5% CO₂ gas mixture to yield a pH of 7.4 (gas bubble dispersion surface; Radnotti Glass Technology, Monrovia, CA). Properties of the test solutions are illustrated in table 1. A conventional 2.25% dextrose-based dialysis solution (Delflex^®, Fresenius USA, INC. Ogden, UT) served as control.

Table 1.

Composition of the test solutions

	5% D-Glucose n = 12	5% D-Glucose n = 12	10% D-Glucose n = 12	5% Mannitol n = 12	5% Mannitol n = 12	10% Mannitol n = 12	Sodium Lactate n = 12
Contents	7.5g D-Glucose 150 ml PSS	7.5g D-Glucose 150ml ddl H2O	50g D-Glucose 500 ml ddl H2O	7.5g D-Mannitol 150 ml PSS	7.5g Mannitol 150 ml ddl H2O	50g Mannitol 500 ml ddl H2O	65 μl of 60% Sodium lactate in 1 L PSS
Osmolarity	554 l	308	613	560	292	580
mOsm/l	553	303	607	567	295	571
pH	7.53		7.44	7.65	5.6	6.65	8.08
Adjusted pH	5.18		4.91	5.49	3.24	5.45	5.03

Drugs

PHE (phenylephrine hydrochloride: 1-(5-oxohexyl)-3,7-dimethylxanthine), ACh (acetylcholine hydrobromide), isoproterenol and sodium nitroprusside were purchased from Sigma Chemical (St. Louis, MO). PHE and ACh were utilized to test the integrity of the vascular endothelium for each aortic ring at the start and at the end of each experiment according to protocol. Isoproterenol and sodium nitroprusside, respectively were used in the tissue bath to retard peristalsis and to assess the maximum dilation capacity of the intestinal microvasculature. All endothelium-intact rings demonstrated more than 60% of ACh-induced relaxation (81%±3.19), and all endothelium-denuded rings demonstrated less than 5% of the ACh-induced relaxation (2.7%±1.66). Three selective adenosine receptor antagonists were used: 8-cyclopentyl-1,3-dipropylxanthine (DPCPX) an adenosine A1 receptor antagonist, 8-(3-chlorostyryl) caffeine (CSC) an adenosine A_2a receptor antagonist, and alloxazine an adenosine A_2breceptor antagonist. The final concentration in the tissue bath for DPCPX, CSC and alloxazine were 200 nM, 200 nM, and 600 μM, respectively. These concentrations represent at least 3 times the 50% effective inhibitory concentration (IC50) for each adenosine receptor antagonist as determined by information provided by the manufacturer.

Study I: Microvascular Reactivity

I. Experimental procedure: The peritoneal cavity was exposed through a midline abdominal incision of 1.5 cm, and 2 to 3 cm segment of jejunum was gently withdrawn from the peritoneal cavity with its neurovascular supply intact. The segment was opened along the antimesenteric border by electrocutery. The enteric contents and mucus were gently removed from the mucosal surface. The animals were positioned on a specially-designed polyurethane board. The opened jejunum was suspended, serosal side up, over a viewing port in a tissue bath with 4-0 silk sutures. The nonvasoactive bathing solution was maintained at 37°C, and bubbled with nitrogen and carbon dioxide to maintain the pH at 7.4. Isoproterenol was added to the bathing solution in a very dilute concentration (0.01 μg/ml) to retard peristalsis. This dose of isoproterenol is below the threshold that alters vascular smooth muscle tone¹¹.

The animal board was positioned on the stage of a trinocular microscope for direct in vivo intravital microscopy. Microvascular images were transmitted through the microscope to a photodiode array in an optical doppler velocimeter (Microcirculation Research Institute, Texas A & M University, College Station, Texas) to measure center-line red blood cell velocity for the calculation of blood flow in the intestinal A1 inflow arteriole. The microvascular image was then transmitted to a digital camera (Hitachi Denshi, Models K-P D51/D50), which provided 30 images per second to a computer. The digitalized microvascular images were stored as streamline video in the computer hard drive for later measurement of microvascular diameters with calipers.

Criteria for an acceptable microvascular preparation during intravital microscopy included a baseline mean arterial pressure > 90 mmHg, a center-line red blood cell velocity in a first-order arteriole > 20 mm/second, and an active vasomotion in the intestinal premucosal A3 arterioles. We used a standard nomenclature for intestinal microvessels, as originally described by Bohlen and Gore¹¹. Briefly First-order arterioles (A1) arise from a mesenteric arcade artery to traverse the mesenteric border of the bowel wall and then penetrate through the muscle layers to the submucosal layer. In the submucosal layer, second-order arterioles (A2) arise from the first-order arterioles to run along the longitudinal axis of the bowel. First and second-order venules parallel the first and second-order arterioles. A2 arterioles give rise to branching second-order arcade vessels as well as to smaller third-order arterioles (A3). The A3 vessels branch at right angle from A2 arterioles to form distal A3 (dA3), which terminates in the mucosa as a central villus arteriole. Along their course, A3 arterioles also give rise to smaller proximal A3 arterioles that supply the seromuscular layers of the bowel wall.

II. Experimental protocol and measurements: The intestinal segment was allowed to equilibrate for 40 min in the tissue bath. During this time, the segment was continuously suffused with the nonvasoactive Krebs solution. Blood pressure, heart rate, rectal and bath temperatures and bath pH were continuously monitored (Digi-Med Signal Analyzers, Louisville, KY) and recorded every five minutes. Microvascular data consisted of A1, pA3 (proximal A3), and dA3 (distal A3) arteriolar diameters; and center-line red cell velocity in the inflow A1 arteriole. Baseline measurements were considered valid when the variability in the measurement is < 0.5%. After baseline measurements, the nonvasoactive Krebs' solution was aspirated from the tissue bath and a test solution (see bathing solutions) was randomly added into the tissue bath. Microvascular data points were measured initially at 2-min after the addition of the test solution, and then at 10-min interval during the subsequent 90 min. At the conclusion of the experiment, one dose of Acetylcholine (ACh, 10⁻⁴M) was topically administered in the tissue bath and microvascular data time points were taken at one-minute intervals over a 10-minute to assess endothelial cell function and endothelial-dependent vasodilation. Finally, a single dose of sodium nitroprusside (10⁻⁴M) was administered in the tissue bath to assess the endothelium-independent maximal dilation capacity.

Study II: Macrovascular Reactivity

The experimental and time-line protocols of these studies were depicted in figure 1. Briefly after induction of anesthesia, the thoracic aorta was excised and submerged in a Petri dish filled with a physiological salt solution (PSS). One half of the thoracic aorta was denuded of endothelium by passing a fine glass rod, about the size of the inner diameter of the aorta to and fro once through the lumen. The other half of the thoracic aorta was regarded as endothelium intact aorta. The presence/absence of viable endothelium in these rings was verified with an endothelium-dependent acetylcholine relaxation according to protocol. The aortic segments closer to the aortic arch and ones closer to the diaphragm behave differently to several agonists than the middle segments (unpublished data). Therefore, only the middle 8 mm of the thoracic aorta was used in the present experiments.

Figure 1.

Vascular ring preparation and protocol. The aorta was harvested under anesthesia and divided to make 2 pair rings of 2mm length each. In one pair the endothelium was removed (denuded rings). Each ring was suspended in physiologic saline solution (PSS) between a transducer and a lower hook under a certain baseline tension level. The change in vessel tension was recorded with a transducer. After equilibration, the PSS was exchanged for the test solution and the change in tension was recorded over 30 min. A washout of the test solution was performed; and phenylephrine (alpha-1 adrenergic agonist) was added in the bath to induce contraction, followed by acetylcholine to induce endothelium-dependent relaxation to demonstrate presence/absence of functional endothelium. PHE: phenylephrine, ACh: acetylcholine.

The endothelium-intact and endothelium-denuded segments of the aorta were divided to produce two 2 mm rings each. Two wires (stainless steel of 0.012-inner diameter) were passed through the lumen of each ring and closed on them to form two wire triangles. One triangle was attached to a fixed hook and the other triangle was attached via a stainless wire to a force transducer, which was connected to a tissue force analyzer. Each vascular ring preparation was suspended in an individual 20 ml tissue bath, which was filled with PSS. Each vascular ring was stretched to produce an initial passive tension called “preload” of 1.0 g in a bath filled with PSS. Each ring was treated with 1.0 μM acetylcholine and 1.0 μM phenylephrine to saturate “non-receptor binding sites” for the agonists, and then washed with PSS for 20 min. After another 40 min of vascular ring equilibration, the preload of each vascular ring was readjusted to the initial 1.0 g preload level. Individual rings were contracted in a sequential manner with six cumulative phenylephrine doses to give phenylephrine bath concentrations of 0.01 through 3.0 μM (in 3X steps). Each ring was then relaxed with 3.0 μM acetylcholine for 10 minutes to demonstrate acetylcholine-induced relaxation and viability of the endothelial cells. Hyperosmolar solutions (see bathing solution) were randomly added in the tissue bath to replace the PSS, and the resultant vascular tension was recorded. Both endothelium-intact and–denuded vascular rings were studied in a paired fashion design.

Data analysis and Statistics

All data is presented as mean ± SEM unless stated otherwise. Percentage change of the vessel diameter from baseline was assessed with One-way ANOVA, and Dunnett's multiple-range test to evaluate changes from the baseline within the same animal. Two-way ANOVA was used to assess the relationship between vascular reactivity and arteriolar type. The maximal force of contraction (Fmax in grams) to the peritoneal dialysis and other hyperosmolar test solutions, the maximal force of contraction to PHE, and the maximal relaxation to ACh was determined for each ring from a computer-stored digitalized raw data. Differences between groups were assessed with two-way ANOVA and Bonferroni post-test. ACh-induced endothelium-dependent relaxation of more than 60% or less than 5% was used to determine presence/absence, respectively of a viable vascular endothelium. A result was considered to be significant if the probability of a type-one error was p < 0.05.

Results

Microvascular Reactivity

I. Effect of solute's metabolic activity: D-glucose unlike mannitol is a metabolically active solute that can be readily and actively transported into cells. As shown in figure 2a, isotonic D-glucose causes a differential reactivity in the intestinal microcirculation (n = 12). This reactivity is characterized by dilation of the smaller premucosal A3 arterioles (8-15 μm, pA3, dA3 respectively). This is in contrast to the largely transient initial constriction observed in the larger inflow A1 arterioles (100 μm). The maximum vascular response during 90 min exposure of the intestine to the isotonic glucose solution, expressed as percentage from baseline, was A1 (−10.58 ± 1.06% recorded at 10 min), pA3 (17.17 ±1.66% recorded at 80 min), and dA3 (19.09 ± 2.41% at 80 min). There was no significant vascular reactivity when the intestine was exposed to the isotonic mannitol solution (p > 0.05, n = 12). The averaged 90 min vascular reactivity to mannitol was − 0.13 ± 1.26%, − 2.90 ± 1.18%, and − 4.45 ± 1.07% from baseline diameter in the A1, pA3, and dA3 arterioles, respectively. Isotonic glucose-mediated premucosal intestinal arteriolar dilation is an insidious and time-dependent response (upper panel, figure 2b). This vascular reactivity was partially attenuated by − 50% in pA3, and − 45% in dA3, with combined adenosine A_2a and A_2b receptor antagonists (middle panel, figure 2b, p < 0.05, n = 12), and completely abolished when the adenosine A₁ receptor was inhibited (lower panel, figure 2b, P < 0.01, n = 12). There was no effect of the adenosine receptor subtype inhibition on the isotonic glucose-elicited selective constriction of the A₁ inflow arterioles.

Figure 2a.

Effect of solute's metabolic activity on intestinal microvascular reactivity. Isotonic glucose but not mannitol produced a differential vascular reactivity in intestinal microcirculation characterized by a significant selective vasodilation of the smaller premucosal arterioles (8-15 μm), and a vasoconstriction of the larger inflow A1 arterioles (100 μm). * p < 0.01 by ANOVA and Bonferroni post-test versus isotonic mannitol. ^§ p < 0.01 by ANOVA and Bonferroni post-test versus A3 premucosal arterioles.

Figure 2b.

Mechanism of isotonic glucose-induced microvascular reactivity. BL = baseline arteriolar diameter; A1 = intestinal inflow arteriole; pA3 and dA3 = intestinal proximal (p) and distal (d) A3 premucosal arterioles. * p < 0.01 by ANOVA and Bonferroni post-test versus BL.

2. Effects of osmolarity perturbation: The effect of osmolarity perturbation on the intestinal microvasculature is depicted in figure 3. Osmolarity was enhanced with either the addition of a 5% mannitol to a nonvasoactive Krebs' solution to obtain a final osmolality of 560 mosml/l (n = 6), or a 10% mannitol to deionized water to obtain a final osmolality of 575 mosml/l (n = 6). A conventional 2.5% dextrose-based peritoneal dialysis solution (Delflex^®, Fresenius USA, INC. Ogden, UT) with osmolality of 398 mosmol/l served as control (n = 6). Enhancement of osmolarity caused a generalized vasodilation at all levels of the intestinal microvasculature. Although the osmolarity of the hypertonic mannitol solutions were significantly higher than that of the dialysis solution, the dialysis solution induced a significantly greater vasodilation in the premucosal arterioles (36.85 ± 3.84%, 44.38 ±6.63%), compared to the 5% mannitol (30.6 ± 4.7, 22.4 ± 3.5), and the 10% mannitol (33.7 ± 3.1, 27.5 ± 2.7%) in the premucosal pA3 and dA3 arterioles, respectively. In addition, all hyperosmolar solutions tested in this series equally doubled the blood flow of intestinal inflow A1 arteriole.

Figure 3.

Effect of mannitol-enhanced osmolality on intestinal microvascular reactivity. PDS = conventional peritoneal dialysis solution; Mn = mannitol, 5% in Krebs or 10% in deionized water, A1 = intestinal inflow arteriole; pA3 and dA3 = intestinal proximal (p) and distal (d) A3 premucosal arterioles. * p < 0.05 by ANOVA and Bonferroni post-test versus PDS-induced vascular reactivity.

Macrovascular Reactivity

1. Effects of osmolarity perturbation: Seven solutions were tested in this series (table 1, n = 12 each). Of these solutions, only the solutions made hypertonic with either D-glucose or Dmannitol produced a significant vascular reactivity in the aorta as measured by force generation (figure 4). The magnitude of this aortic vascular reactivity was largest with the conventional peritoneal dialysis solution, and quantitatively higher in hyperosmolar solutions containing ions, compared to identical hyperosmolar solutions that lack ionic contents (figure 4), suggesting a significant role of the ionic contents of the solution in determining the magnitude of the prevailing vascular response. There was no vascular effect of isotonic glucose or mannitol solutions on the aorta (data not shown).

Figure 4.

Box and whiskers plot of altered osmolality-induced macrovascular reactivity in endothelium (En) intact (+) and denuded (−) aortic rings. PDS = conventional peritoneal dialysis solution; Glu = glucose; Mn = mannitol; 5% = osmotic solute concentration in Krebs or 10% in deionized water; * p < 0.05 versus other hyperosmotic solutions by ANOVA and Bonferroni post-test.

2. Effects of vascular endothelium: In all the hypertonic solutions tested in the present study, endothelium removal significantly attenuated the magnitude of the solution-mediated aortic contraction by −32% (range −20% to −54%, p < 0.05). Such attenuation was maximally seen in endothelium denuded aortic rings exposed to the 5% glucose (−37%), and the 5% mannitol (−54%) in physiologic salt solutions.

3. Effects of solution's pH: The role of pH in hyperosmolality-induced aortic reactivity appears to be influenced by the aortic endothelium and the ionic composition of the solution (figure 6). Of all the hyperosmolar solutions tested in this series, the dialysis solution produced the highest aortic ring contraction, regardless of the endothelium (figure 6). In the aortic denuded rings, the conventional peritoneal dialysis solution produced a −23% less contraction force than rings with intact endothelium. This pattern was also preserved in denuded aortic rings suffused with 5% mannitol −19%, 5% glucose −33%, 10% mannitol −27% and 10% glucose −15%, when the pH of these hyperosmolar solutions was adjusted to < 6 pH units (n = 7 for each solution). Low pH accounted for 4.5% (p < 0.05%) of the total variation in the vascular reactivity produced by the 5% (glucose or mannitol), which contained other ions (figure 6, upper panel). In comparison, in the 10% osmotic solute (glucose or mannitol) in deionized water (figure 6, lower panel), low pH accounted for only 0.3% (p > 0.5) of the total variation, which is exclusively accounted for by the vascular reactivity 15.3% (p < 0.001). In the presence of ions, low pH appears to attenuate vascular reactivity by 30% in denuded rings and by 10% in endothelium-intact rings regardless of the main osmotic solute (figure 6, upper panel). In contrast, in low pH deionized solutions, hyperosmolality due to glucose enhanced aortic reactivity in endothelium intact rings by 16% and attenuated that of endothelium-denuded rings by 14%, whereas in hyperosmolality due to mannitol, aortic reactivity in endothelium-intact rings was enhanced by 8%, and attenuated by 37% in endothelium-denuded rings (figure 6, lower panel).

Figure 6.

Effect of pH perturbation on hyperosmolality-induced macrovascular reactivity. * p < 0.01 versus endothelium intact aortic rings. ^§ p < 0.05 versus controlled pH by ANOVA and Bonferroni post-test.

4. Effects of the buffer anion: In this series, sodium lactate was dissolved in an isotonic physiologic salt solution (PSS) to match the lactate concentration of peritoneal dialysis solutions. Addition of sodium lactate (0.392 g/l) to the non-vasoactive PSS yielded a pH > 8 pH units. At this high pH, which was maintained in the tissue bath during the experiment, the maximum contraction force in aortic rings with intact endothelium was 0.01 ± 0.02 g (n = 12, p > 0.1) versus − 0.02 ± 0.02 g (n = 12, p > 0.1), in endothelium denuded rings. When tissue bath pH was adjusted to 5.03, there was a significant contraction force in both endothelium intact aortic rings 0.33 ° 0.11 g (n = 12, p < 0.05), and 0.45 ± 0.16 g (n = 12, p < 0.01) in endothelium denuded aortic rings. This pH-dependent lactate-induced aortic contraction was significantly greater in the endothelium-denuded rings (p < 0.05).

Discussion

The salient findings of the present studies are that: 1) conventional peritoneal dialysis solutions produce an instantaneous and sustained vasodilation at all levels of the intestinal (visceral) microvasculature; 2) the pattern and magnitude of such dilation is dependent on vessel size and on the osmotic solute's metabolic activity; 3) energy-dependent transport of glucose into cells mediates an insidious vasodilation preferentially on small visceral arterioles by an adenosine receptor-mediated mechanism, which constitutes a significant fraction of a glucose-based peritoneal dialysis solution-mediated reactivity in the visceral microvasculature; 4) hyperosmolality is the major vasoactive component of the conventional peritoneal dialysis solution whereas lactate the buffer anion system of this solution is vasoactive only at low pH; 5) low pH and other ionic contents of conventional peritoneal dialysis solutions, modifies the magnitude of vascular reactivity instigated by these solutions.

Technique

Our present studies were performed with a standard tissue bath procedure under well controlled experimental conditions. The tissue under investigation, a small segment of the intestine (2-4 cm), or aortic rings (2 mm), were positioned in a relatively large tissue bath in which temperature, pH, PO₂, PCO₂ and osmolality were monitored and controlled, while we simultaneously made direct observations of the intestinal microcirculation, or continuously record the change in force of individual aortic rings. During the equilibration period, the small intestinal segment and the aortic rings were continuously suffused with a nonvasoactive physiologic salt solution, which results in variability of less than 0.5% in vessel diameter and force generation measurements. This small variability in the measurements indicates that changes in vessel reactivity in our present studies can only be attributed to a specific experimental intervention, rather than to a baseline change in vascular reactivity. The experimental design of our present studies does not allow for the simultaneous determination of solute and water transport across the blood peritoneal barrier. Therefore, other issues such as the effect of bioincompatibility and dialysis solution composition on transperitoneal exchange are not addressed in the present study.

Vascular reactivity

In our present studies, two standard techniques were utilized to identify the vasoactive components of conventional peritoneal dialysis solutions and quantitatively describe their vasoactivity. All hyperosmolar solutions, including our control peritoneal dialysis solution, produced significant vascular reactivity in aortic rings and intestinal microvasculature. In addition, sodium lactate dissolved in isotonic physiologic salt solution produced a significant vascular reactivity in aortic rings only when H⁺ concentration in the solution was increased. Results of our studies were consistent with literature data which supported the concept that hyperosmolality and lactate are the major vasoactive components of peritoneal dialysis and other hyperosmolar solutions^1;2;8;12-15.

In a recent study, Mortier and colleagues have found that in the rat mesentery, conventional and new bicarbonate-buffered re-sterilized peritoneal dialysis solutions that contain high glucose degradation products (GDPs) dilate mesenteric arteries (250 to 350 μm), whereas new bicarbonate-buffered solutions that contain low GDPs were nearly non-vasoactive ⁹. Although none of the measured GDPs was singled out as a possible potent vasoactive agent, the authors postulated that a possible combination of the measured GDPs is likely the cause of the dilation response. In their study, multiple solutions with or without nitroglycerin (10⁻⁴ M) were randomly tested in the same animal. In addition, the vascular reactivity to the test solution was observed in mesenteric arteries while a change in blood flow in a smaller network arteriole was interpreted to mirror a change in the diameter of the mesenteric artery. Our experience with intravital microscopy of the intestinal microcirculation suggests that testing multiple solutions or pharmacologic manipulations of the tissue bath drastically affect baseline microvascular hemodynamics. In particular endothelium-independent nitric oxide donors such as sodium nitroprusside and nitroglycerin causes a transient maximum dilation followed by a rapid drop in the arterial blood pressure and the local blood flow due to systemic absorption of the drug. The vascular reactivity (∼20% of baseline) of mesenteric arteries to conventional peritoneal dialysis solutions appears to be of the same magnitude seen in the intestinal first order arterioles in our present (A1, figure 3) and previous study ⁸. However, the interpretation of the results as to what component of the dialysis solution is vasoactive differs between our study and that by Mortier and colleagues. There are several potential explanations that could account for the difference in interpretation of the results between the two studies. It is well established that vessels of similar size use different mechanisms for endothelium-dependent regulation of vascular tone depending on vascular bed ¹⁶. Similarly, the relative contribution of agonist-stimulated NO and endothelium-dependent hyperpolarizing factor (EDHF) to endothelium-dependent relaxation appears to differ between genders ¹⁷, arteriolar size within the same vascular bed ¹⁸, and between arterioles from different vascular beds ^16;19;20. Consistent with these experimentally validated observations is the differential microvascular response within the same vascular bed to specific events such as hemorrhage and sepsis ^21;22, or exposure to agonists such as serotonin, angiotensin and activated complement ^23-25. In addition, our present experiments were performed in a well defined vascular bed that has a unique microvascular architect, which is identical in all rats ¹¹. In contrast, the mesentery is relatively void of cells and possesses a poorly defined microvascular network, which accounts for much of the variations in local blood flow in different segments, and for the variety of effects of vasoactive agents on the dynamics of the mesenteric microcirculation ^26;27.

Data of the present study indicates that an isotonic glucose solution causes an insidious and preferential vasodilation in the smaller premucosal intestinal arterioles, and a slight but significant constriction of the larger inflow arterioles. This unique vasoactive property of the isotonic glucose is mediated by an adenosine receptor mechanism. In contrast, the intestinal microvessels remained at baseline during exposure to an isotonic mannitol solution. However, hypertonic mannitol solutions either as 5% in Kreb's or 10% in deionized water caused an instantaneous and sustained dilation similar in magnitude to the Delflex^®-induced dilation at all levels of the intestinal microvasculature. Although mannitol is an efficient osmotic agent, it is a metabolically inert solute that is totally excluded from cells. This data clearly demonstrates that hyperosmolality is a major vasoactive component by itself, but that the mechanisms of this dilation effect may differ depending on the specific metabolic activity of the osmotic solute. Isolated cannulated and pressurized skeletal muscle arterioles, elicits a dilation response proportional to increasing concentrations of glucose added to a superfusion solution⁵. Similar changes in arteriolar diameter were obtained in response to superfusion with sucrose or mannitol⁵. These data suggest that in in vitro models, arteriolar vasodilation occurs in proportion to the degree of hyperosmolality. This contrasts with the aortic vascular reactivity in our current study in which the least aortic vascular reactivity was obtained with solution's osmolality > 550 mosml/L. Therefore, translation of the in vitro data to the in vivo situation should be tentatively approached. This does not refute the fact that such in vitro vascular models are the gold standard for assessing vascular control mechanisms and the signal transduction pathways of these mechanisms. Our current and previous data does not support a linear relationship between hyperosmolality and vascular reactivity. However, it appears that the magnitude of such reactivity is modified by the H⁺ concentration, ionic contents and other vasoactive components of the solution such as lactate, as well as by the specific solute metabolic activity. Indeed earlier intravital videomicroscopy studies of the rat's cremaster muscle, have shown that the magnitude of arteriolar dilation evoked by hyperosmolar solutions of dextrose, sucrose or sodium chloride was similar but that the dilation rate constant differs between the three hyperosmolar solutions ². Other perfusion studies of the dog's forelimb ²⁸, and cats ileum ²⁹ have found that both the magnitude and the time course of the dilatory effects of hyperosmolar dextrose and sodium chloride solutions differed significantly.

Our current data on the intestinal vasoactivity suggests that hypertonic glucose-based solutions dilate the intestinal microvasculature by at least two mechanisms, one is an instantaneous microvascular vasodilation related to the osmotic stress, and the second is a more insidious, time-dependent vasodilation, stimulated by an energy-dependent transport of glucose into cells. This pathway is exclusively accounted for by an adenosine receptor mediated mechanism as demonstrated in our present study. For the vasodilation related to the osmotic stress, we suggest that during crystalloid-induced osmosis, the osmotic water flux through the transendothelial water exclusive channels (Aquaporin-1, AQP-1) is the primary mechanism whereby the endothelium is being stimulated to instigate vasodilation effects. Initial cell shrinkage caused by osmotic-driven water flow, results in a relative increase in cellular ionic contents especially, Ca²⁺ and K⁺, which are known to stimulate endothelium-dependent dilation pathways. Simultaneously, such initial cell volume decrease triggers a regulatory volume increase characterized by net water and ionic uptakes as well as stimulation of organic osmolyte transporters in order to restore the original cell volume ^30;31. It has been show that during conditions of osmotic stress there is activation of Ca²⁺-activated K⁺ channels (K_Ca) and ATP-sensitive K⁺ channels (K_ATP). Such activation results in a dilation response that can be nearly completely abolished by specific inhibition of these channels ⁵. Large arteries and other inflow arterioles minimally express AQP-1 ³², and possess insufficient adenosine receptor subtypes ³³. This explains the subordinate magnitude of dilation seen in theses macrovessels compared to the marked dilation observed in the smaller premucosal intestinal arterioles in the present studies.

It is generally conceived that peritoneal dialysis-induced vasodilation occurs only initially during a hypertonic dwell. Carlsson and Rippe have attributed an inflation of the permeability surface area product (PS) of small solutes to an initial vasodilation during a hypertonic dwell ³⁴. However, our data suggest that a significant fraction of the dilation response is sustained preferentially in the small precapillary arterioles for as long as these vessels are exposed to an isotonic solution containing glucose. The influence of this precapillary vasodilation on the mass transfer area coefficient (MTAC, or PS) for small solutes during the late phase of the dwell, where the osmotic gradient has dissipated remains to be determined. It is likely that relaxation of these small vessels in response to exposure to conventional peritoneal dialysis solutions is detrimental in the number of perfused capillaries and in the modulation of the effective capillary surface area available for exchange during peritoneal dialysis. Thus, the instantaneous submaximal vasodilation at all levels of the intestinal micorovasculature including the inflow A1 feed arteriole, which doubles its blood flow in the initial phase of the dwell when osmolality is high, could provide a plausible explanation for the high PS for small solutes during the early phase of the dwell. With dissipation of the osmotic gradient with time due to glucose absorption, a subordinate vasodilation is preferentially maintained in the smaller intestinal premucosal A3 arterioles, whereas the feed A1 arterioles at most restore their baseline diameter and blood flow, which explains the relatively lower PS for small solutes during the late phase of the dwell. In contrast, in peritoneal transport rate studies in man, the addition of a clinical dose of sodium nitroprusside to a dialysis solution, produced no effect on peritoneal fluid kinetics, a slight increase in MTAC for small solutes, but a great increase in macromolecular clearances ³⁵. Similarly, in rabbits, a clinical dose of sodium nitroprusside exclusively enhanced peritoneal macromolecular clearance ³⁶. These clearance data might reflect a change in peritoneal microvascular permeability, rather than a vasodilation-mediated modulation of the functional peritoneal surface area in terms of capillary recruitment. For recruitment of functional capillary surface area by dialysis solution-induced dilation, a contact between the dialysate and peritoneal tissue must be established. We have shown that less than half of the mouse anatomical peritoneum is in contact with a large volume of solution in the peritoneal cavity ³⁷, and that agitation or use of surfactant-supplemented dialysis solution increases the fraction of contact area, resulting in enhanced transperitoneal exchange ^37;38. Although the visceral peritoneum accounts for ∼60% of the anatomic peritoneum, its fractional contribution to the overall PS for small solutes is only of the order of 30%, whereas the major fraction of PS for small solutes is accounted for by the much smaller parietal peritoneum³⁸. This is attributed to the rather complex geometry of the visceral peritoneum that encompass “macro-unstirred” pockets of fluid, which equilibrates faster than the rest of the dialysate, limiting small solute diffusion ³⁸. Furthermore, in single membrane models permeability and surface area are multiplicatively linked to form the lumped parameter PS, which is difficult to separate. Therefore, pharmacologic targeting of this lumped parameter in an attempt to improve adequacy is more likely to simultaneously change permeability and surface area. Thus, efforts to improve dialysis adequacy should be directed toward improving the wetted peritoneal surface area, which is the anatomical peritoneum in contact with the dialysate, and to create favorable mixed conditions at the visceral peritoneum.

In conclusion, using two different experimental techniques, we demonstrated that hyperosmolality and lactate are the major vasoactive components of clinical peritoneal dialysis solutions. The pattern and magnitude of such reactivity is dependent on vessel size and on the solutes' metabolic activity. Low pH of conventional peritoneal dialysis solutions is not a vasoactive component by itself; but renders lactate vasoactive. Energy-dependent transport of glucose into cells mediates vasodilation of small visceral arterioles by an adenosine-receptor mediated mechanism, and constitutes a significant fraction of peritoneal dialysis solution-mediated vascular reactivity in the visceral microvasculature. Further investigation is required to define the signal transduction pathway and the molecular mechanisms of hyperosmolality-induced vascular reactivity.

Figure 5.

Effect of vascular endothelium (En) on Hyperosmolality-induced macrovascular reactivity. PDS = conventional peritoneal dialysis solution; Glu = glucose; Mn = mannitol; 5% = osmotic solute concentration in Krebs or 10% in deionized water; * p < 0.05 versus endothelium intact aortic rings by ANOVA and Bonferroni post-test.

Figure 7.

Lactate-induced macrovascular reactivity. En (+) = endothelium intact; En (−) = endothelium denuded. * P < 0.01 versus high pH and ^§ p < 0.01 versus endothelium-intact aortic rings by ANOVA and Bonferroni post-test.