Mast cell amines and inosineinduced vasoconstriction in the rat hind limb

Under certain circumstances injected inosine causes a net vasoconstrictive effect on the arterioles, which has been attributed to 5-hydroxytryptamine (5HT) released in response to adenosine type 3 (A3) receptor stimulation of mast cells residing in the adventitia. We have sought further evidence for this hypothesis using blood vessels of the rat hind limb perfused in vitro at constant rate with a gelatin-containing physiological salt solution. Injection of inosine (2.7 mg) caused a rise in perfusion pressure, which was only slightly increased by inclusion of N-nitro-L-arginine methyl ester (100 μM) in the perfusate. Inclusion in the perfusate of cyproheptadine (1 μM), compound 48 80 (1 μg ml), 8-phenyltheophylline (1 μM) or 8-cyclopentyl-1,3 dipropylxanthine (0.1 μM) greatly reduced the pressor response to inosine. The pressor effect of injected 5HT (400 μg) was abolished by pre-treatment with cyproheptadine, but not by pre-treatment with compound 48 80. These results suggest that the net pressor response to injected inosine was mainly the result of an A1 receptor-mediated release of 5HT, most probably from mast cells. No evidence was found for an involvement of A3 receptor stimulation.


Introduction
Despite the fact that adenosine generally acts as a vasodilator in mammalian vascular beds, 1 both adenosine and its main metabolite inosine have been shown to cause vasoconstriction in the rat perfused hind limb 2 and in hamster cheek pouch arterioles. 3,4 Adenosine can also contract strips isolated from rat tail arteries. 5 Such responses have been shown to involve the release of mast cell amines, 2 5 and may result from stimulation of the adenosine type 3 (A 3 ) receptors, 6 that have been demonstrated to occur in mast cells. 7 In the present work we have attempted to discover the type of adenosine receptor which is responsible for inosineinduced vasoconstriction in the rat perfused hind limb. A gelatin-containing physiological salt solution (GPSS) was used as perfusate, rather than blood, in order to eliminate the possible effects of 5-hydroxytryptamine (5HT) which may be released from rat platelets. 8 Inosine was used in preference to adenosine since previous workers have shown that arterioles constrict in response to inonsine over a wide range of concentrations, whereas adenosine constricts at low concentrations but at higher concentrations may dilate blood vessels. 3 The types of receptor involved have been explored with the aid of various adenosine and 5HT receptor agonists and antagonists.

Methods
Female rats weighing approximately 250 g were killed by inhalation of chloroform vapour. The abdomen and thorax were opened using a ventero-lateral incision after which the vasculature was heparinized via an intra-cardiac injection. Subsequently, the left iliac artery, both renal arteries and the aorta proximal to the renal arteries were all ligated. A stainless steel cannula (1.07 mm OD), which offered no measurable resistance to ow at 10 ml min, was then inserted into the right iliac artery via an incision in the aorta distal to the renal arteries, and was tied in place. Perfusate was then pumped into the right iliac artery at a rate of 10 ml min, and allowed to drain away through a hole cut in the inferior vena cava. Perfusion pressures were recorded from a Condon manometer connected to the perfusion line and via a strain gauge pressure transducer connected to a Multitrace 2 recorder (Lectromed, UK). A paper speed of 25 mm min was used for the 1 min period immediately following bolus injections of various adenosine receptor agonists or of 5HT. These were delivered into the stream of perfusate using a back-to-back syringe system to minimize injection artefacts. Compounds to be tested as possible antagonists were added to the GPSS at the start of the experiment. All preparations were perfused for an initial stabilizing period of 10 min. Then, after recording baseline perfusion pressure for 3 min, 2.7 mg inosine in 0.2 ml normal saline (NS, 0.9% NaCl) was injected. Perfusion pressures thereafter were recorded continuously for 5 min. After perfusing for a further 5 min, to allow time for the effects of inosine to cease, the baseline pressure was again recorded and a control injection of 0.2 ml NS was given. Finally, this procedure was repeated with a bolus injection of 5HT (400 mg in 0.2 ml NS). Pilot experiments indicated that these doses of inosine and 5HT were necessary to produce suitable pressor responses. Preliminary experiments showed, as have those of previous workers, 2 that the pressor response to inosine vanishes after approximately three injections have been made into the same preparation. Therefore, only one inosine, one NS and one 5HT response (where relevant) were recorded from each animal.
Experiments were also performed using the rat perfused tail vascular bed. For this, male rats weighing about 350 g were killed and heparinized as above. Both renal arteries and both iliac arteries were ligated in this case. After ligating the aorta proximally, the steel cannula was inserted into the tail artery via the distal aorta. Experiments then followed the protocol described above.
Compounds added to the GPSS were N-nitro-L-arginine methyl ester (L-NAME), a nitric oxide (NO) synthase inhibitor; 9 cyproheptadine, a mixed histamine and 5HT receptor blocker; 10 compound 48 80, which degranulates mast cells; 11 8-cyclopentyl-1,3 dipropylxanthine (DPCPX), a relatively speci c A 1 receptor antagonist; 12 or 8-phenyltheophylline (8PT), a mixed A 1 A 2 receptor antagonist. 13 In addition to inosine two other putative adenosine receptor agonists were tested in the hind limb. There were N 6 -cyclopentyladenosine (CPA), which acts primarily via A 1 receptors, 14 and iodobenzyl-5-N-methyl carboxamidoadenosine (IB-MECA), which shows a much greater af nity for A 3 than for either A 1 or A 2 receptors. 15 These two compounds were administered in dimethyl sulphoxide (DMSO) diluted 1 19 with NS. Control injections in such preparations, therefore, were with DMSO NS (1 19) rather than with NS alone.
All pressure recordings were made at the same ampli cation and paper speed, so they could be compared directly with each other. In order to compensate for any residual injection artefact, reponses to each agonist were expressed (in arbitrary units) as the difference between the trace areas (  hind limb perfused with GPSS L-NAME (100 mg). Recordings were made on 1 mm squared paper. The pressor response to inosine in NS was calculatd as the difference between the number of squares in the shaded area above the baseline minus the number of squares in the shaded area below the baseline. The equivalent numbers of squares (above and below the baseline) after an injection of NS (0.2 ml) was calculated from the same preparation. The response to NS alone was then subtracted from the response to inosine in NS. Final results were expressed in arbitrary units ( Fig. 2 and Table 1). The maximum amplitude of the response (in mm) was determined at A.
Inosine and 5HT were dissolved in NS. L-NAME, compound 48 80 and cyproheptadine were added to GPSS as aqueous solutions. DPCPX, 8PT, CPA and IB-MECA were each initially dissolved in DMSO. The nal concentration of DMSO in a perfusate was always , 0 05%.

Statistics
Signi cant differences (p , 0 05) were determined using Bonferroni's test for comparing several treatment groups with one control group.

Responses to inosine and 5HT
In experiments using rat hind limb vessels perfused in vitro at constant rate with GPSS, a bolus of inosine (2.7 mg) caused a pressor response that began to wane after the rst minute. There was no evidence that inosine produced any vasodilatation. A transient fall in perfusion pressure, however, occurred after each injection of NS. Hence this residual injection artefact needed to be compensated for in the responses to inosine and 5HT. Consequently the magnitude of the NS response was routinely deducted from that recorded after giving inosine or 5HT in the same animal. The pressor effect of 5HT (400 mg), unlike that due to inosine, continued to increase slowly for about 5 min. Because of these differing time courses, we decided to present the results obtained using area units. However, the maximum pressor amplitudes attained during the rst minute after injection of either inosine or 5HT were closely related to the corrected areas of the pressor responses, but are not separately presented. The pressor effects of inosine and 5HT measured over the rst minute post-injection were 54 1 15 3 and 141 8 32 9 area units respectively. Similar results were obtained with the rat tail vessels perfused with GPSS, where the pressor responses to inosine and to 5HT were 17 0 13 9 and 231 6 45 8 area units respectively. Mean starting perfusion pressures with GPSS owing at 10 ml min in the hind limb and tail vascular beds were approximately 34 and 70 mmHg respectively. When L-NAME (100 mM) was present in the GPSS perfusing the hind limb, the pressor responses to inosine and to 5HT were increased slightly to values of 61 3 9 6 and 192 0 30 7 area units respec-tively. This suggests that when NO-synthase was still active a slight dilatory effect from endogenous NO, 16 may have blunted the net vasoconstriction that was produced by both inosine and 5HT. In all subsequent experiments L-NAME was added to the GPSS.

Effects of cyproheptadine and compound 48 80
Inclusion of cyproheptadine (1 mM) or compound 48 80 (1 mg ml) in an L-NAME-containing GPSS signi cantly reduced the pressor effects of both injected inosine (Fig. 2) and 5HT ( Table 1), suggesting that the pressor effects of inosine were attributable to 5HT released from a source within or adjacent to the vasculature.

Effects of speci® c adenosine receptor antagonists and agonists
When 8PT (1 mM) or DPCPX (1 mM) was included in an L-NAME-containing GPSS the response to inosine in the hind limb vessels was signi cantly reduced (Fig. 2). The speci c A 1

Mediators of In¯ammation´Vol 6´1997
receptor agonist CPA (500 mg) caused virtually no change in perfusion pressure ( 0 5 26 0 area units, n 6). This was converted to a small depressor effects ( 9 0 14 3 area units, n 3) after pre-treatment with compound 48 80 (1 mg ml), but the difference from CPA responses without compound 48 80 pre-treatment was not statistically signi cant. The highly potent and speci c A 3 agonist IB-MECA (5 mg), exerted a statistically insigni cant haemodynamic effect using this protocol ( 16 7 41 8 area units, n 3).

Discussion
Most adenosine receptor agonists can relax most types of vascular smooth muscle. 1 This effect is exerted, at least partly, via direct stimulation of both the muscular A 1 receptors, which can open both K ATP channels 17,18 and K Ach channels, 18 and via the muscular A 2 receptors, which activate adenylyl cyclase. 19 Vasodilatation occurring during periods of ischaemia is associated with, and partly due to, the increased concentrations of adenosine and its metabolite inosine 20 which occur in the vicinity of blood vessels under these circumstances. 21 24 Exogenous inosine can also dilate the coronary blood vessels. 25 The vasoconstrictor responses to inosine which we report here, therefore, although con rming previous work, 2 4 are rather atypical, but not peculiar to the rat, having been observed also in the hamster. 3,4 The amount of inosine that was required to cause vasoconstriction in the present experiments (2.7 mg) was quite high, but not surprisingly so, since K i values for inosine at A 1 , A 2 and A 3 receptors are reported to be in the 20± 50 mM range. 15 Furthermore, about 1 mg of adenosine is required to constrict blood vessels in the rat hind limb during perfusion with a physiological salt solution, whereas 30 mg is suf cient to cause vasoconstriction during perfusion with blood. 26 These same workers have demonstrated the presence of 5HT in venous ef uent collected during treatment with adenosine, but they did not speculate upon the site of origin of the 5HT. Perivascular mast cells are one obvious possibility, but coronary artery endothelial cells also contain, and may actually secrete, 5HT. 27 Circulating platelets provide another source of 5HT when blood is used as the perfusate. 8 Adenosine itself has been shown to stimulate granule amine release from various types of mast cells. 28 32 Inosine has a similar effect, but only at higher concentrations. 30,32 Moreover, mast cells are normally present in the adventitial layer of various mammalian arterioles. 3,33,34 Since rat mast cell granules contain 5HT 35 adenosine receptor agonists that are delivered via the blood vessel lumen should be able to release 5HT from mast cells in the wall, provided that they can readily penetrate the intimal and medial layers of the wall, or reach the adventitia via the capillaries. Since 5HT is predominantly a vasoconstrictor, 36 it is entirely possible that vasoconstrictor responses to adenosine receptor agonists that are seen in blood vessels of the rat hind limb, are due to released 5HT. Indeed, in the present experiments inosine lost its vasopressor effect after pre-treatment with cyproheptadine or with compound 48 80. Our evidence, therefore, supports a mast cell origin for the 5HT.
If inosine releases 5HT from adventitial mast cells, as the foregoing results would suggest, then it remains to decide which class of adenosine receptor was responsible. The fact that pre-treatment with either 8PT (a mixed A 1 A 2 antagonist), 13 or with DPCPX (a selective A 1 receptor antagonist), 12 was able to prevent the pressor response to inosine suggests that it was A 1 receptor activation which caused the release of 5HT here. Previous work on rat omental mast cells has indicated that degranulation occurs in response to selective A 1 receptor agonists, 32 including CPA. However, in the present experiments CPA produced no net effect on the perfusion pressure. This may have been because of a slower penetration of CPA than of inosine through the intimal and media layers of blood vessel walls on its way to the adventitia. In other words, CPA may not have stimulated adventitial mast cells powerfully enough to overcome the direct vasodilatory action which is likely to have been produced by CPA via A 1 A 2 receptors on the muscle cells (or the endothelial cells) of the wall. Of relevance in this connection is that on rat isolated aortae CPA was found to exert a net relaxant effect which was attributable to activation of endothelial A 2b receptors. 37 IB-MECA is claimed to be a selective A 3 receptor agonist, the K i values at A 1 , A 2 and A 3 receptors being 54 5 nM, 56 8 nM and 1 1 0 3 nM respectively. 15 In contrast, at A 3 receptors inosine shows a K i value approximately 50 000-fold higher than IB-MECA. 15 On this basis, therefore, if inosine had caused its effects in the present experiments via A 3 receptors, one would have expected IB-MECA to have mimicked inosine, but at a 50 000-fold lower concentration. In fact, IB-MECA failed to mimic the effect of inosine at a concentration of 1000-fold less than that of the effective dose of inosine. It seems unlikely, therefore, that A 3 receptors are involved. Similarly, IB-MECA had no degranulating effect on omental mast cells in vitro. 32 Such a lack of response may merely re ect poor tissue penetration. Nevertheless, so far we have failed to obtain any evidence that A 3 receptor stimulation was responsible for the release of mast cell amines in our experiments. This contrasts with the ndings of some previous workers 6,34 under different experimental conditions. These discrepancies are so far unexplained, but it may be relevant that the IB-MECA-induced hypotension which occurs in cats was attributable to the activation of A 1 and A 2 receptors rather than of A 3 receptors. 38 Hopefully, with the advent of more speci c adenosine receptor agonists these divergent observations will be explained.