OUP user menu

Pharmacologic activation of the human coronary microcirculation in vitro: endothelium-dependent dilation and differential responses to acetylcholine

Francis J. Miller Jr., Kevin C. Dellsperger, David D. Gutterman
DOI: http://dx.doi.org/10.1016/S0008-6363(98)00035-2 744-750 First published online: 1 June 1998


Objectives: In vivo studies of the human coronary resistance circulation cannot control for indirect effects of myocardial metabolism, compression, and neurohumoral influences. This study directly examined the vasodilator responses of the human coronary microcirculation to both receptor-dependent and -independent agonists. Methods: Atrial arterioles were dissected from human right atrial appendage (103±2 μm diameter, n=185 vessels from 145 patients) obtained at the time of cardiopulmonary bypass and left ventricular vessels from explanted human hearts (148±10 μm diameter, n=57 vessels from 18 patients). After dissection, vessels were mounted onto pipettes in Kreb's buffer under conditions of zero flow and at a constant distending pressure of 60 mmHg. Drugs were applied extraluminally and steady state changes in diameter measured with videomicroscopy. Results: After contraction by endothelin or spontaneous tone, increasing concentrations of adenosine diphosphate (ADP) produced a similar dose-dependent dilation in vessels from atria (maximum 89±4%, n=76) and ventricles (maximum 74±9%, n=10). The dilation to ADP was abolished by mechanical removal of the endothelium. Similar dilator responses were found to bradykinin, substance P, arachidonic acid, and the calcium ionophore A23187 in both atria and ventricle. In contrast, acetylcholine (ACh) constricted all atrial vessels (−58±3%, n=63) regardless of patient age or underlying disease. This constriction was attenuated by denudation, but not affected by inhibition of nitric oxide synthase or cyclo-oxygenase. Microvessels isolated from human ventricle exhibited a heterogeneous response to ACh with dilation being the predominant response. Conclusions: We conclude that isolated human coronary arterioles demonstrate endothelium-dependent dilation. However, the response to acetylcholine is unique with vasoconstriction in atrial vessels and dilation in ventricular arterioles.

  • Coronary circulation
  • Microcirculation
  • Human
  • Adenosine diphosphate
  • Nitric oxide
  • Atria
  • Ventricle
  • Acetylcholine

Time for primary review 29 days.

1 Introduction

Coronary vasomotor tone can be modified by a number of agents, many of which are dependent upon endothelium for release of relaxing or constricting factors [1, 2]. In many disease states including atherosclerosis [3–5], diabetes mellitus [6, 7], hypertension [8, 9], and hypercholesterolemia [3, 10], these vasomotor responses are altered. Our understanding of human coronary vascular biology in health and disease has been limited to in vivo studies which are complicated by the associated changes of myocardial compressive forces, perfusion pressure, blood flow, and of neurohumoral factors, all which can independently change vasomotor tone [11]. In vitro studies have the advantage of excluding these competing factors by examining the isolated blood vessel, but the majority of such studies have been performed in isolated large vessels. Because coronary blood flow is regulated by the microvasculature, with the majority of vascular resistance residing in arterioles less than 200 μm in diameter [12], understanding coronary flow regulation requires the study of arteriolar reactivity. This is critical since large and small vessels in the same bed often exhibit fundamentally different vasomotor properties [13]. This observation demonstrates the importance of directly examining the microcirculation in understanding blood flow regulation.

Although coronary microvascular function has been studied in animal models, we must be cautious when extending these findings to humans. There is marked variability in vascular physiology among animal species. For example, canine coronary arterioles dilate to acetylcholine [13]whereas porcine arterioles constrict [14].

For these reasons, we identified a model in which vascular biology of the human coronary microcirculation could be investigated. Responses to endothelium-dependent vasodilators in human atrial arterioles removed at the time of cardiopulmonary bypass were examined and compared to responses from isolated ventricular arterioles. When our findings indicated that acetylcholine constricts atrial arterioles, despite dilating ventricular vessels, we examined the role of the endothelium, prostaglandins, and nitric oxide in mediating this response.

2 Methods

2.1 Tissue acquisition and microvascular preparation

All protocols were approved by the University of Iowa Human Use Committee and conform to the principles outlined in the Declaration of Helsinki. Human right atrial appendage, routinely removed for cannulation during cardiopulmonary bypass, was obtained during elective cardiac surgery. The tissue was immediately placed in oxygenated cold (4°C) Kreb's buffer consisting of (in mmol/l) 118.0 NaCl, 4.7 KCl, 2.5 CaCl2, 1.2 MgSO4, 25.0 NaHCO3, 1.2 KH2PO4, 5.5 glucose, 0.0025 EDTA and transported to the laboratory. Arterioles (50–200 μm in diameter and 1.0–2.5 mm in length) were dissected free along the length of pectinate muscle from the endocardial surface and placed into an organ chamber containing cold Kreb's buffer. For species comparison, right atrial appendage was obtained from mongrel dogs (n=10) and monkeys (n=2).

Fresh human ventricular tissue was obtained at the time of congenital defect repair or from hearts harvested at the time of orthotopic transplantation. Vessels (57–349 μm in diameter) were dissected from the epicardium or the endocardium and transferred to an organ chamber bath containing cold buffer.

In the organ chamber bath, one end of the vessel was cannulated with a glass micropipette filled with Kreb's buffer and secured with 10-0 nylon Ethilon monofilament suture. With low perfusion pressure, any remaining blood products were flushed from the lumen and the other end of the vessel cannulated with a second micropipette and secured with suture. Vessels were not used if they contained side branches. Vessels were studied under conditions of zero flow and a distending pressure of 60 mm Hg. This pressure was chosen since it corresponds to the estimated physiologic pressure of these size of vessels in the beating heart [12], and myogenic responsiveness of these vessels is greatest at 60 mmHg [15].

The preparation was transferred to the stage of an inverted microscope (200× magnification) attached to a video camera, a video monitor, and a calibrated video measuring device. Internal diameters were measured by manually adjusting the video micrometer. The organ chamber was continuously perfused with Kreb's buffer solution maintained at 37°C and oxygenated with a mixture of 5% CO2, 20% O2, and 75% N2 providing a stable pH of 7.40±0.05 and pO2 of 140±10 mmHg.

2.2 Protocols

2.2.1 Pharmacological activation

Vessels were allowed to equilibrate for 30 min. Agonists were applied to arterioles that were in one of three loading conditions: (1) resting state with no tone: (2) spontaneous tone (10–50% constriction), or (3) pharmacologic-induced tone with endothelin-1 (10−10 to 10−9 mol/l; 25–60% constriction). In this way, potential effects of pharmacological vs. physiological constriction could be examined on vasodilator function.

When a stable vessel diameter was achieved, agonists were added to the circulating buffer and changes in steady state diameter recorded. Cumulative concentration responses were obtained for the following: acetylcholine (ACh 10−9–10−4 mol/l), adenosine diphosphate (ADP 10−9–10−4 mol/l), substance P (SP 10−12–10−7 mol/l), bradykinin (BK 10−12–10−6 mol/l), calcium ionophore A23187 (10−12–10−7 mol/l), arachidonic acid (AA 10−9–10−4 mol/l), and nitroprusside (SNP 10−9–10−4 M).

The cumulative administration of drugs increased the volume of the circulating bath by less than 5%. At the end of a concentration–response, the vessel was washed for 20–30 min in fresh buffer and a subsequent dose–response curve generated. At the completion of the experiment, vessels were maximally relaxed with the endothelium-independent vasodilator sodium nitroprusside (SNP 10−4 mol/l).

2.2.2 Endothelial denudation

In some vessels, the endothelium was mechanically removed prior to agonist administration using a technique similar to that described previously [16]. After one end of the isolated arteriole was secured to a pipette, a separate glass micropipette was advanced into the other end and withdrawn under gentle suction. Functional denudation without damage to the vascular smooth muscle was demonstrated by failure to relax to an endothelium-dependent vasodilator (ADP) despite preserved relaxation to an endothelium-independent vasodilator (SNP).

2.2.3 Pharmacologic inhibition

The effect of inhibiting cyclo-oxygenase or nitric oxide production on vasoreactivity was assessed in separate experiments. After a concentration response was obtained to an agonist, vessels were washed for 20 min then incubated with the arginine analog NG-monomethyl-l-arginine (l-NMMA, 10−4 mol/l) or indomethacin (10−5 mol/l) for 30 min. These doses inhibit nitric oxide synthase and cyclo-oxygenase, respectively, in other vascular preparations. Vessels were again constricted with endothelin (similar magnitude as in the control response) and administration of the selected agonist was repeated. The order of control response and l-NMMA-treated response was altered. In the absence of an inhibitor, there were no differences in vessel responsiveness to sequential administration of an agonist.

2.2.4 Solutions and drugs

All drugs were obtained from Sigma Chemical Company except for endothelin-1 (Peninsula Laboratories) and atropine sulfate (Fujisawa USA, Inc.). Endothelin was dissolved in 1% bovine serum albumin and stored in frozen aliquots. Indomethacin was prepared in 50 mmol/l Na2HCO3 and buffered to pH 7.4. Stock solutions of A23187 were dissolved in 95% ethanol. After addition to the circulating bath, ethanol concentrations were less than 0.01%. All other solutions were prepared daily in ultra-distilled water. Stock solutions were diluted serially with Kreb's buffer. All reported concentrations are final molar concentrations in the organ chamber bath.

2.2.5 Data analysis

Internal diameters were recorded at steady state conditions. For contraction, the percent change from preconstricted diameter was calculated at each dose of agonist. For dilation, the percent change from preconstricted diameter to the maximal diameter (10−4 mol/l SNP) was determined. Summary data are reported as means±s.e.m. and ED50s were calculated graphically as that dose producing 50% of the maximal observed response. Responses from vessels in the same group from the same patient are averaged. Comparison of response before and after an inhibitor is analyzed with a paired t-test (two tailed). Differences are considered significant when P<0.05. Reported vessel size represents the passive (i.e. maximal) diameter. Intact atrial vessels were included for data analysis only if they dilated greater than 30% to ADP (10−4 mol/l), and 70% to SNP (10−4 mol/l). Demographic information and diagnoses were obtained for each patient by review of hospital computer coded data.

3 Results

3.1 Patient characteristics

Right atrial appendage was obtained from 145 patients at the time of cardiopulmonary bypass. From this tissue 185 atrial vessels were studied ranging in size from 46 to 197 μm in diameter (103±2 μm).

Left ventricular tissue was obtained from 18 patients, ranging in age from 1 to 67 years (42±5 years), 6 were female. Fifty-seven vessels ranging in diameter from 57 to 349 μm (148±10 μm) were dissected from 14 explanted hearts (12 recipients, 2 non-cardiac donors) or from ventricular tissue from 4 patients having congenital defect repair. Thirty-five of the vessels were from the epicardial surface, and 22 from the endocardium. Patient demographics including diagnoses are shown in Table 1.

View this table:
Table 1

Demographic data

Atrial tissue
Number of patients: 145109 CAD99 Coronary artery bypass
Age: 0 days to 93 years40 DM16 Valve replacement
Mean age: 52±2 years66 HTN11 ASD/VSD repair
Gender:37 CHF9 Other congenital defect repair
 101 male23 None of above2 Cardiac transplant
 44 female6 Unknown8 Unknown
Ventricular tissue
Number of patients: 189 CAD14 Cardiectomy
Age: 1–67 years3 DM 12 Transplant recipient
Mean age: 42±5 years2 HTN 2 Non-cardiac donor
Gender:12 Cardiomyopathy
 12 male 8 Ischemic4 Congenital defect repair
 6 female 4 Dilated
4 Congenital defect
  • CAD, coronary artery disease; DM, diabetes mellitus; HTN, hypertension; CHF, congestive heart failure; ASD, atrial septal defect; VSD, ventricular septal defect.

3.2 Effects of adenosine diphosphate

In 76 atrial vessels from 60 patients (106±6 μm diameter) ADP applied extraluminally produced a concentration-dependent relaxation (Fig. 1). There was no difference in relaxation of arterioles regardless of whether preconstriction was spontaneous or pharmacologically induced with endothelin. Ventricular vessels demonstrated similar relaxation to ADP (Fig. 1).

Fig. 1

Response of human coronary microvessels to adenosine diphosphate. Increasing concentrations of ADP dilate ventricular (▾, n=10 vessels, 10 patients, 108±7 μm in diameter) and atrial (●, n=76 vessels, 60 patients, 106±6 μm) microvessels similarly. Data represent mean±s.e.m. The percent change from preconstricted to maximal diameter is shown on the y-axis.

Endothelial dependence of dilation to ADP was examined in mechanically denuded arterioles. In 11 atrial vessels, relaxation to ADP (10−4 mol/l) was essentially abolished after denudation (81±4% with endothelium vs. 9±4% after denudation, P<0.05), while relaxation to SNP (10−4 mol/l) was preserved (92±4% with endothelium vs. 80±7% after denudation, P=n.s.). This verifies that relaxation of human atrial microvessels to ADP is dependent upon an intact endothelium.

3.3 Effects of substance P, A23187, bradykinin, arachidonic acid and sodium nitroprusside

Human atrial arterioles dilated to increasing concentrations of substance P, calcium ionophore A23187, bradykinin, arachidonic acid, and sodium nitroprusside (Fig. 2, Table 2). Each agonist dilated precontracted vessels to greater than 75% of passive diameter in dose-dependent fashion. Although fewer ventricular arterioles were available for study, they relaxed similarly to these agonists (n=3 for SP, A23187, and AA; n=4 for BK and SNP, data not shown).

Fig. 2

Response of human atrial vessels to exogenous dilators. The percent change from preconstricted to maximal diameter is shown for BK (●, n=17 vessels, 10 patients), A23187 (■, n=13 vessels, 8 patients), substance P (▾, n=5 vessels, 3 patients), and SNP (n=12 vessels, 10 patients). Data represent mean±s.e.m.

View this table:
Table 2

Maximal response of atrial arterioles to vasodilators

AgonistLog Mn% Change
Adenosine diphosphate10−46081±4
Substance P10−7392±4
Arachidonic acid10−4482±5
Sodium nitroprusside10−41091±3
  • n=number of patients. % Change represents the percent change from preconstricted diameter. Data are mean±s.e.m.

3.4 Effects of acetylcholine: human atrial and ventricular microcirculation

The effect of ACh on atrial arteriolar tone was examined in 63 isolated vessels (passive diameter 105±5 μm; range 52–185 μm) from 41 patients. At each dose tested, acetylcholine constricted atrial vessels (Fig. 3). Constriction of atrial arterioles to ACh occurred regardless of the magnitude of preconstriction (0–70%) and independent of the preconstrictor (endothelin vs. spontaneous). These same vessels demonstrated the ability to dilate to endothelium-dependent and independent agents as indicated above (Figs. 1 and 2). In the presence of atropine (10−4 mol/l), constriction to ACh was prevented (0.3±4% constriction, n=5, P<0.05 vs. vehicle).

Fig. 3

Response of human and canine atrial arterioles to acetylcholine. Human atrial vessels constrict with increasing doses of ACh (n=63 vessels, 41 patients, 105±5 μm in diameter) whereas canine atrial vessels dilate (n=12 vessels, 10 dogs, 127±11 μm). For human vessels, the y-axis indicates the percent change from preconstricted diameter. For canine vessels, the y-axis indicates the percent change from preconstricted to maximal diameter. Data are mean±s.e.m.

In contrast to human arterioles, vessels isolated from the right atrial appendage of mongrel dogs dilated to increasing doses of ACh (Fig. 3). Two vessels from monkey atria also dilated to ACh (data not shown). This observation suggests that the constrictor response of human atrial microvessels is species specific and not the result of a general difference in vascular physiology of the atria compared to the ventricle.

To determine whether constriction to ACh was an indication of endothelial dysfunction, the vasomotor response of patients with no clinical evidence of atherosclerosis was compared to those with documented coronary atherosclerosis. Atrial arterioles from patients with (n=42, age 42±4 years, 106±5 μm diameter) and without coronary atherosclerosis (n=18, age 18±5 years, 113±7 μm) constricted similarly to increasing doses of ACh (maximal response −63±4% without CAD vs −58±3% with CAD, P=n.s.). Likewise, the ED50 to ACh was similar between the two groups.

Previous studies have shown that the vascular response to ACh of human ventricular epicardial arteries is dependent on age [17]. However, we found no evidence that the magnitude of atrial constriction to ACh was age dependent (r2=0.04, P=n.s.).

To examine the mechanism of ACh-induced coronary arteriolar constriction, we examined the role of the endothelium. In vessels treated with l-NMMA (10−5 mol/l) basal tone was not affected, and there was a non-significant trend toward less constriction at high doses of ACh compared to control vessels (Fig. 4A). Thus, no evidence for ACh-mediated dilation through formation of nitric oxide was found. Atrial vessels denuded of endothelium constricted less to ACh than non-denuded vessels (Fig. 4B). This finding is unlikely the result of vascular smooth muscle damage since constriction to endothelin was similar in denuded (41±3%) and non-denuded (43±4%) vessels. Preserved smooth muscle function was also evident by relaxation of denuded vessels to SNP (85±4%). Denudation was verified by impaired relaxation to ADP (18±8%). These observations suggest that ACh induce no significant release of nitric oxide in human atrial arterioles. In contrast, release of an endothelial derived constricting factor may contribute to the ACh constriction.

Fig. 4

Denudation of atrial vessels impaired constriction to ACh (n=10), whereas inhibition of nitric oxide synthase (n=22) or cyclo-oxygenase (n=13) had no effect. *P<0.05 vs. control. Data are mean±s.e.m. The y-axis indicates the percent change from the preconstricted diameter.

To determine if a prostanoid vasoconstrictor is involved in the atrial arteriolar constriction to ACh, the effect of cyclo-oxygenase inhibition was examined. Treatment of vessels with indomethacin (10−5 mol/l) did not affect the constriction to ACh (Fig. 4C). The combination of indomethacin and l-NMMA (n=12) also failed to affect ACh-mediated constriction.

In contrast to atrial vessels, human ventricular arterioles and small arteries dilated to ACh (Fig. 5). However, variability of ventricular responses was common. In some vessels ACh produced only dilation, while in other vessels, sometimes from the same heart, no response or constriction was observed.

Fig. 5

Comparison of human atrial and ventricular arteriolar responses to ACh. Isolated ventricular vessels dilate to increasing concentrations of ACh (▾, 57 vessels, 18 patients; 151±17 μm maximal diameter) while atrial vessels constrict (●, 63 vessels, 41 patients; 105±5 μm). Data are expressed as percent change from preconstricted diameter (mean±s.e.m.). Ventricular data includes vessels which constricted to increasing concentrations of ACh.

4 Discussion

The major findings of this study relate to the first direct evaluation of vascular reactivity in the human coronary resistance circulation. In isolated human coronary microvessels, we demonstrate that atrial and ventricular arterioles exhibit similar responses to the receptor-mediated endothelium-dependent dilators ADP, bradykinin, and substance P, to the receptor-independent dilator A23187, and to the endothelium-independent dilator SNP. However, unique to the human heart, atrial arterioles constrict to ACh while ventricular vessels display a heterogeneous response to ACh, exhibiting dilation as a group. The mechanism of atrial arteriolar constriction to ACh includes direct activation of vascular smooth muscle, may include release of endothelial constriction factors, and is independent of prostaglandins.

In other studies, intracoronary infusion of ACh in humans produced dilation, no change, or even constriction of conduit vessels [4, 9, 17]. In normal conduit segments, ACh produces dilation by release of EDRF [18]. Conversely, ACh causes constriction in angiographically atherosclerotic epicardial vessels, although these vessels relax normally to nitroglycerin [4]. The degree of impaired vasodilation may be related to the number of risk factors for atherosclerosis [19]. Isolated human epicardial arteries obtained at autopsy also constrict to ACh while exhibiting endothelium-dependent relaxation to BK, SP, and A23187. Denudation does not enhance the constriction to ACh in these large vessels [20].

An unexpected finding of our study was the differential response between atrial and ventricular arterioles to ACh. Isolated atrial microvessels constricted whereas ventricular arterioles from the same hearts dilated to ACh. Atrial arteriolar constriction was atropine sensitive and independent of the method of preconstriction (basal tone vs. pharmacological). Potential explanations of this paradoxical response include: (1) selective atrial arteriolar endothelial damage resulting from underlying disease or trauma from dissection; (2) differential diffusion barrier between atrial and ventricular vessels of extraluminal ACh; (3) competing constriction from activation of muscarinic receptors in the VSM; and (4) insufficient muscarinic receptor density on the atrial vascular endothelium, or a defect in the intracellular signaling process which normally leads to release of EDRF and dilation.

It is unlikely that constriction of atrial arterioles to ACh results from disease-mediated endothelial dysfunction since it was a universal finding in over 200 atrial vessels studied, many of which were from children as young as 1 day old. It is also unlikely that trauma from our dissection is selective for human atrial arterioles since we observe dilation to ACh in human ventricular and dog atrial and ventricular coronary arterioles.

In our studies, all drugs were applied to the extraluminal surface of the vessel and allowed to diffuse to the adjacent endothelium. It is possible that given the distribution of muscarinic receptors in the vascular smooth muscle, intraluminal application of ACh in atrial arterioles would produce vasodilation. However, parasympathetic nerves terminate in the adventitia and media of small coronary vessels [21], therefore the extraluminal application of ACh would be expected to more closely mimic the physiological situation of neuronal release.

The observation that vessels actually constrict less to ACh after denudation is consistent with the release of an endothelial vasoconstrictor substance. There does not appear to be an important role for prostaglandins, since indomethacin did not alter ACh-induced constriction. An alternative explanation for reduced constriction to ACh following denudation is damage to underlying smooth muscle. However, dilation to SNP and constriction to endothelin was not affected, arguing against vascular damage.

The most likely explanation for the absence of dilation of atrial arterioles to ACh is reduced muscarinic receptor density on the endothelium or a defect in the coupling of the receptor to the intracellular signaling system. The ability of atrial arterioles to relax to other endothelium-dependent dilators indicates intact mechanisms for synthesis and release of EDRF.

Angus et al. showed previously in a small number of patients ACh-induced constriction of small arteries from human right atrial appendage despite relaxation to substance P [22]. However, in that study, KCl was used as the preconstrictor. Depolarization of vessels with KCl may inhibit the action of endothelial derived hyperpolarizing factor, or EDHF, an important mediator of vasodilation in microvessels [23]. In our study, ACh constricted human atrial vessels under conditions (spontaneous tone or constricted with endothelin) which would not be expected to suppress activity of EDHF.

The physiologic significance of atrial coronary constriction to ACh is unclear. No other study to our knowledge has compared atrial and ventricular responses to vasoactive agents. In the dog (Fig. 3B) and monkey (data not shown), right atrial arterioles dilate to ACh, indicating that disparate responses of human atrial and ventricular microvessels to ACh are not generalized to all species.

Heterogeneous responses in the normal heart to humoral substances have been described. Du et al. showed that at concentrations less than 10−6 mol/l, ACh was a positive inotrope in ventricular trabeculae, but a negative inotrope in atrial trabeculae obtained from human hearts [24]. Other vasoactive agents show contrasting effects on large and small vessels. Both thrombin and vasopressin dilate conduit, but constrict small canine coronary vessels [13].

In summary, human arterioles isolated from either the atria or ventricle demonstrate relaxation to BK, A23187, substance P, and ADP. In contrast, ACh consistently constricts atrial vessels. This constriction occurs regardless of age of the patient and does not appear to be a function of underlying disease or atherosclerotic risk factors. In microvessels and small arteries from the ventricle, ACh produces a heterogeneous response dilating some vessels while constricting others.


The authors thank Joanne Schwarting and Diann McCoy for their technical assistance and the cardiothoracic surgeons and nurses at the University of Iowa Hospitals and Mercy Hospital, Iowa City for their role in providing tissue specimens. Financial Support: Merit Review Award from the Veterans Administration, R01 HL51308, American Heart Association Grant-In-Aid Iowa-Affiliate, NIH Individual National Research Service Award. D.D.G. and K.C.D. are recipients of an American Heart Association Established Investigator Award.


View Abstract