ATP-induced vasodilation and purinergic receptors in the human leg: roles of nitric oxide, prostaglandins, and adenosine

Stefan P. Mortensen, José González-Alonso, Laurids T. Bune, Bengt Saltin, Henriette Pilegaard, Ylva Hellsten


Plasma ATP is thought to contribute to the local regulation of skeletal muscle blood flow. Intravascular ATP infusion can induce profound limb muscle vasodilatation, but the purinergic receptors and downstream signals involved in this response remain unclear. This study investigated: 1) the role of nitric oxide (NO), prostaglandins, and adenosine as mediators of ATP-induced limb vasodilation and 2) the expression and distribution of purinergic P2 receptors in human skeletal muscle. Systemic and leg hemodynamics were measured before and during 5–7 min of femoral intra-arterial infusion of ATP [0.45–2.45 μmol/min] in 19 healthy male subjects with and without coinfusion of NG-monomethyl-l-arginine (l-NMMA; NO formation inhibitor; 12.3 ± 0.3 (SE) mg/min), indomethacin (INDO; prostaglandin formation blocker; 613 ± 12 μg/min), and/or theophylline (adenosine receptor blocker; 400 ± 26 mg). During control conditions, ATP infusion increased leg blood flow (LBF) from baseline conditions by 1.82 ± 0.14 l/min. When ATP was coinfused with either l-NMMA, INDO, or l-NMMA + INDO combined, the increase in LBF was reduced by 14 ± 6, 15 ± 9, and 39 ± 8%, respectively (all P < 0.05), and was associated with a parallel lowering in leg vascular conductance and cardiac output and a compensatory increase in leg O2 extraction. Infusion of theophylline did not alter the ATP-induced leg hyperemia or systemic variables. Real-time PCR analysis of the mRNA content from the vastus lateralis muscle of eight subjects showed the highest expression of P2Y2 receptors of the 10 investigated P2 receptor subtypes. Immunohistochemistry showed that P2Y2 receptors were located in the endothelium of microvessels and smooth muscle cells, whereas P2X1 receptors were located in the endothelium and the sacrolemma. Collectively, these results indicate that NO and prostaglandins, but not adenosine, play a role in ATP-induced vasodilation in human skeletal muscle. The expression and localization of the nucleotide selective P2Y2 and P2X1 receptors suggest that these receptors may mediate ATP-induced vasodilation in skeletal muscle.

  • skeletal muscle
  • adenosine 5′-triphosphate
  • purinergic receptors
  • nitric oxide
  • prostaglandins

skeletal muscle blood flow is closely regulated to match O2 delivery to the metabolic demand of contracting myocytes during a wide array of exercise modalities (2, 15, 29, 42). The precise regulation of muscle blood flow is believed to mainly be the result of the competing influences of locally released vasodilator signals and sympathetic neural vasoconstrictor activity (9). Previous reports employing double blockade of nitric oxide synthase (NOS) and cyclooxygenase (COX) by which nitric oxide (NO) and prostaglindins are derived, indicate that these two endothelial-derived substances are important mediators in skeletal muscle blood flow control during exercise (4, 30, 43). However, the stimulus for their formation remains unknown.

ATP is released from erythrocytes upon exposure to low oxygen tension and mechanical deformation and has been proposed to contribute to local skeletal muscle blood flow control (13, 47). In addition, ATP is released from endothelial cells in response to shear stress and hypoxia (5). ATP is a potent vasoactive substance, and intra-arterial infusion of ATP in the human leg can induce vasodilation close to that observed during maximal exercise (17, 40). Intravascular ATP is believed to exert its circulatory effects via activation of P2Y receptors located on the cell surface of the endothelium, which subsequently initiate vasodilation by releasing endothelial vasoactive substances and by offsetting sympathetic vasoconstriction (1, 23, 40, 41). Indeed, in vitro reports from various tissues have suggested that ATP-induced vasodilation is endothelium dependent (12) and that ATP induces vasodilation by increasing the formation of NO (10, 27) and prostaglandins (20). In the human forearm, inhibiting NO or prostaglandin formation does not alter the response to ATP infusion (44, 50). However, there may be differences in the mechanisms underlying blood flow regulation in arm and leg (32). Also, the magnitude of blood flow in the forearm is low because of the small muscle mass; thus, flow changes resulting from pharmacological interventions could be difficult to detect.

In general, activation of endothelial P2Y (G protein coupled) receptors induces vasodilation, whereas activation of smooth muscle cell P2X (ion channels) receptors induces vasoconstriction (6, 19). P2Y2, P2Y4, and P2Y11 receptors have been identified to be ATP sensitive (38) and are expressed in endothelial cells of the umbilical vein (P2Y2 and P2Y11) and smooth muscle cells of the mammary artery (P2Y2) (51). Yet, P2X1 receptors located on the endothelium have also been shown to mediate ATP-induced vasodilation in the mesenteric arteries (21). Despite the proposed circulatory effects of plasma ATP on skeletal muscle, no direct evidence exists to date as to the presence of P2 receptors in human muscle vascular endothelial and smooth muscle cells and which particular receptor subtypes that potentially are involved.

Another potential mediator of ATP-induced vasodilation might be adenosine because ATP is rapidly degraded into ADP, AMP, and adenosine via the catalytic actions of circulating and membrane-bound ectonucleotidases (19, 52). Adenosine can induce vasodilation via nucleoside-selective P1 receptors located in the endothelium of skeletal muscle (1, 25). ATP could therefore at least in part mediate its vasodilatory action via activation of P1 receptors. Evidence for a role of adenosine and P1 receptors is, however, conflicting. On one hand, P1 receptor blockade in isolated arterioles has been found to abolish the vasodilator response to ATP (12). On the other hand, a study in the human forearm suggested that adenosine is not involved in ATP-induced vasodilation (39). Therefore, the role of adenosine in ATP-induced vasodilation remains unclear.

Accordingly, the purpose of this study was to examine: 1) whether intravascular ATP can induce vasodilation by causing the release of NO and prostaglandins in human skeletal muscle, 2) whether ATP-induced vasodilation is partly mediated by activation of P1 receptors, and 3) whether purinergic P2 receptors are expressed in human skeletal muscle. To accomplish these aims, ATP was infused in the femoral artery of 13 healthy subjects with and without inhibition of NO and/or prostaglandin formation as well as theophylline (TEO)-induced adenosine receptor blockade in 6 healthy subjects. In addition, levels of mRNA and cellular localization of purinergic P2 receptors was determined in eight healthy subjects. We hypothesized that inhibition of the NO and prostaglandin synthesis, but not adenosine receptor blockade, would blunt the vasodilatory response to ATP and that human skeletal muscle contains P2Y receptors capable of mediating ATP-induced vasodilation.



A total of 27 subjects participated in two studies. The subjects were informed of the risks and discomforts associated with the experiments before giving their informed, written consent to participate. The study was approved by the Ethics Committee of Copenhagen and Frederiksberg (KF 01–013/96 and KF 11289201) and conducted in accordance with the guidelines of the Declaration of Helsinki.

Experimental Protocol

In study 1, a total of 19 moderately trained male subjects with a mean (±SD) age of 25 ± 3 yr, body weight of 78 ± 8 kg, height of 184 ± 7 cm, and V̇o2max of 4.1 ± 0.6 l/min participated. Subjects were instructed to abstain from caffeine for 48 h before the experiment. On the experimental day, the subjects arrived at the laboratory 1 h before the experiment after a light breakfast. Catheters were placed in the femoral artery and vein of the experimental leg and femoral artery of the nonexperimental leg under local anesthesia. After 30 min of rest, ATP (0.03 ± 0.00 μmol·min−1·kg leg mass−1; A7699; Sigma) was infused in the femoral artery for 5 min under the following conditions: 1) control (n = 13) or 2) NG-monomethyl-l-arginine (l-NMMA) alone (n = 7), indomethacin (INDO) alone (n = 7), or combined l-NMMA + INDO (double; n = 6) (Fig. 1). During the control and double blockade trial, ATP was also infused for 3 min at a higher infusion rate (high; 0.22 ± 0.03 μmol·min−1·kg leg mass−1). All tests were separated by a 30-min rest period. Because of the long-term effect of INDO and l-NMMA, the control trial was performed first. Seven subjects completed both single inhibition trials on two separate experimental days separated by 2 wk. In all trials, saline (control), INDO (50 μg·min−1·kg leg mass−1; Confortid, Alphapharma, Denmark), and/or l-NMMA (1.0 mg·min−1·kg leg mass−1; Clinalfa, Laufelfingen, Switzerland) was infused in the femoral artery for 5 min before ATP infusion and then coinfused with ATP during the 5–7 min of ATP infusion. The subject's leg mass was 12.3 ± 0.3 kg, and the infused concentrations were therefore 0.45 ± 0.03 (low) and 2.45 ± 0.36 (high) μmol/min for ATP, 613 ± 12 μg/min for INDO, and 12.3 ± 0.3 mg/min for l-NMMA. Blood samples (1–5 ml) were drawn simultaneously from the femoral artery (nonexperimental leg) and vein during the basal infusion period (4 min) and during ATP infusion (1.5 and 4 min).

Fig. 1.

Experimental protocol in study 1. Subjects rested in a semisupine position (45°). ATP was infused in the femoral artery for 5–7 min in the following three different protocols: 1) on two separate experimental days, saline (control) and NG-monomethyl-l-arginine (l-NMMA) or saline and indomethacin were separately infused in the femoral artery for 5 min during basal conditions and during ATP infusion (n = 7), 2) saline (control) and l-NMMA + indomethacin was infused in the femoral artery for 5 min during basal conditions and during ATP infusion (n = 6), and 3) before and after femoral venous theophylline infusion (n = 6).

To test the effect of time on leg hemodynamics, we repeated two control trials separated by 30 min of rest in three subjects and found no significant difference in leg blood flow (LBF; 2.3 ± 0.2 vs. 2.2 ± 0.3 l/min) or mean arterial pressure (MAP; 89 ± 2 vs. 91 ± 2 mmHg) between trials. To test the effectiveness of the NOS inhibition, we infused ACh (160 mg/min) in the femoral artery with and without coinfusion of l-NMMA in three subjects and found that NOS inhibition lowered the ACH-induced increase in LBF by ∼60% (2.7 ± 0.3 vs. 1.0 ± 0.1 l/min). To test if the infused inhibitors had systemic effects, we infused l-NMMA and INDO in the femoral vein of four subjects and found no difference in LBF, cardiac output, heart rate (HR), or MAP between baseline and femoral venous l-NMMA and/or INDO infusion (Table 1).

View this table:
Table 1.

Leg and systemic hemodynamics before and during venous infusion of l-NMMA and/or indomethacin

In six subjects, ATP (0.03 ± 0.00 and 0.17 ± 0.04 μmol·min−1·kg leg mass−1) was infused in the femoral artery for 3 min with and without TEO, a nonspecific adenosine receptor antagonist (49). TEO (5 mg/kg body mass; Sygehus Apotekene, Copenhagen, Denmark) was infused in the femoral vein at a rate of 11 mg/min for 30–38 min (total dose, 400 ± 26 mg) followed by 10 min of rest before the start of ATP infusion. This dose of TEO was used since it effectively inhibited the vasodilatory response to 3 min of femoral artery ADO infusion (0.5 and 1.0 mg·min−1·kg leg mass−1; n = 5; ITEM Development, Stocksund, Sweden) (Fig. 2). Blood samples were drawn during the basal conditions and during ATP infusion (2.5 min).

Fig. 2.

Change in blood flow during adenosine-induced vasodilation with and without P1 receptor blockade. Change in leg blood flow from baseline values during intra-arterial adenosine infusion with and without theophylline-induced adenosine receptor blockade. Data are means ± SE for 5 subjects. *Significantly different from control, P < 0.05.

Resting LBF was measured with ultrasound Doppler (CFM 800; Vingmed) (36), whereas LBF during ATP infusion was measured by the constant-infusion thermodilution method (2, 18). A close correlation (r = 0.974) between these two methods has been demonstrated previously (35). HR was obtained from an electrocardiogram, whereas arterial pressure was monitored with transducers positioned at the level of the heart (Pressure Monitoring Kit; Baxter). Cardiac output was calculated by multiplying HR by stroke volume. Stroke volume was calculated using the Modelflow method (BeatScope version 1.1; Finapress Medical Systems BV, Amsterdam, the Netherlands) (15). The change in LBF was calculated as the difference between baseline blood flow and blood flow during ATP or adenosine infusion.

In study 2, eight male subjects with a mean (±SD) age of 26 ± 3 yr, body weight of 80 ± 8 kg, and height of 181 ± 5 cm participated. The day before the experimental trial, the subjects refrained from exercise. On the experimental day, the subjects arrived at the laboratory 2 h after consuming a standardized breakfast. A muscle biopsy was obtained from the middle portion of the vastus lateralis muscle using the percutaneous needle biopsy technique (3) with suction. Muscle biopsies were quickly frozen in liquid nitrogen (<15 s) and stored at −80°C until analyzed.

Analytical Procedures

Blood samples.

Blood gases, hemoglobin, and glucose and lactate concentrations were measured using an ABL725 analyzer (Radiometer, Copenhagen, Denmark) and were corrected for temperature obtained in the femoral vein. Leg mass was calculated from whole body dual-energy X-ray absorptiometry scanning (Prodigy, General Electrics Medical Systems, WI). Plasma norepinephrine concentrations were determined with a radioimmunoassay (LDN, Nordhorn, Germany).

RNA isolation, reverse transcription, cDNA content, and real-time PCR.

Total RNA was isolated from ∼20 mg of muscle tissue by a modified guanidinium thiocyanate-phenol-chloroform extraction method adapted from Chomczynski and Sacchi (8) as previously described (33). The Superscript II RNase H system (Invitrogen) and oligo(dT) were used to reverse transcribe 2 μg total RNA to cDNA as previously described (33). The cDNA samples were diluted in nuclease-free H2O to a total volume of 150 μl.

The mRNA content of a given gene was determined by real-time PCR using the fluorogenic 5′-nuclease assay with TaqMan probes (ABI PRISM 7900 Sequence Detection System, Applied Biosystems), as previously described (24). The sequences of forward and reverse primers as well as TaqMan probes have been published previously (51) and were optimized for the PCR reaction, as previously described (34). The amount of single-stranded DNA was determined in each cDNA sample using OliGreen reagent (Molecular Probes), as previously described (24), and, for each sample, the amount of a given target mRNA was normalized to the total cDNA content of the RT sample. The threshold cycle (CT) is the number of PCR cycles a sample has to be amplified to reach a specific fluorescence level (threshold), and thus reflects the initial copy number of the mRNA in the sample.

In addition, to examine potential gDNA contamination of the RNA samples and thus potential contributing fluorescence signal from gDNA being amplified in the cDNA samples, RNA samples were diluted (relatively) as the cDNA samples. For each purinergic receptor, PCR was run with both the cDNA and the diluted RNA samples from which the gDNA contamination of the cDNA samples was evaluated and subtracted from the level in the cDNA samples.


The cellular localizations of P2Y2 purinergic receptors were determined on 8-μm transverse sections of frozen skeletal muscle samples from five subjects. The sections were fixed in 2% formaldehyde for 2 min at room temperature and −20°C acetone for 30 s. The sections were rinsed with PBS containing 1% BSA and thereafter blocked with 1% PBS containing BSA. The sections were then incubated for 1 h at room temperature with a rabbit polyclonal α-P2Y2 antibody at 8 μg/ml (APR-010; Alomon Laboratories). Endothelial cells were detected with a primary monoclonal antibody (mouse α-CD31 at 50 μg/ml; Clone JC70A; DAKO) or the lectin Ulex Europeaeus Agglutinin-1 at 10 μg/ml (UEA-1; Sigma Aldrich). All sections, with the exception of those incubated with UEA-1, were rinsed and thereafter incubated with either a biotin-coupled goat α-rabbit (E0432; DAKO) or a biotin-coupled rabbit α-mouse antibody (E0354; DAKO). UEA-1 was directly coupled to TRITC. Staining of purinergic receptors and UEA-1 were, for purposes of identification of cellular location of the receptors, performed on the same section and superimposed. Antibody binding was visualized with either streptavidin coupled to fluorescein isothiocyanate (F0422; Glostrup) or avidin-biotin complex (ABC) with alkaline phosphatase (ABC/Complex/AP, DAKO, Denmark). Negative controls were achieved with staining with incubation of the antibody with the corresponding peptide before incubation of slide or without the primary antibody. Immunoreactive cells were examined and photographed in a Zeiss Axioplan Microscope.

Statistical Analysis

A two-way repeated-measures ANOVA was performed to test significance within and between trials. Following a significant F-test, pairwise differences were identified using Tukey's honestly significant difference post hoc procedure. Differences between the single (l-NMMA or INDO) and double-blockade trial (l-NMMA or INDO) were assessed with a t-test. There was no difference in any variable between 1.5 and 4 min during ATP infusion, and presented data are therefore means of the two measurements. The significance level was set at P < 0.05, and data are means ± SE. unless otherwise indicated.


Effect of INDO and/or l-NMMA on Leg Hemodynamics Before and During ATP Infusion

LBF during baseline conditions was 0.38 ± 0.03 l/min but was lower with l-NMMA (0.32 ± 0.03 l/min; P < 0.05) and combined l-NMMA + INDO (0.26 ± 0.04 l/min; P < 0.05), whereas infusion of INDO (0.39 ± 0.08 l/min) had no effect on baseline flow. ATP infusion increased LBF to 2.45 ± 0.29 l/min, but LBF was lower when ATP was coinfused with l-NMMA (1.79 ± 0.17 l/min), INDO (1.83 ± 0.17 l/min), and both inhibitors (1.39 ± 0.03 l/min) (Fig. 3). At the high ATP infusion rate, LBF was 4.62 ± 0.39 l/min and was lower (3.18 ± 0.53 l/min; P < 0.05) when it was coinfused with both inhibitors combined. ATP infusion reduced MAP during control conditions (93 ± 2 vs. 88 ± 2 and 86 ± 2 mmHg at basal conditions vs. low and high ATP infusion, respectively, P < 0.05), whereas ATP infusion did not change MAP when coinfused with either l-NMMA (95 ± 3 vs. 92 ± 3 mmHg), INDO (99 ± 4 vs. 97 ± 4 mmHg), or both inhibitors (98 ± 2 vs. 97 ± 2 and 96 ± 4 mmHg). Leg vascular conductance was lower with l-NMMA (4 ± 0 ml·min−1·mmHg−1; P < 0.05) and l-NMMA + INDO (3 ± 0 ml·min−1·mmHg−1; P < 0.05) compared with control (5 ± 1 ml·min−1·mmHg−1) during baseline conditions, whereas there was no difference with INDO (4 ± 1 ml·min−1·mmHg−1). During ATP infusion, leg vascular conductance was lower with l-NMMA (22 ± 3 ml·min−1·mmHg−1), INDO (20 ± 2 ml·mmHg−1·min−1), and l-NMMA + INDO combined (13 ± 1 and 48 ± 2 ml·min−1·mmHg−1, low and high, respectively) compared with control (26 ± 2 and 28 ± 3 ml·min−1·mmHg−1, low and high, respectively) (P < 0.05). Accounting for the change in LBF during basal conditions with l-NMMA and INDO + l-NMMA, the increase in LBF was reduced by 14 ± 6, 15 ± 9, 39 ± 8, and 36 ± 6% during l-NMMA, INDO, and INDO + l-NMMA (low and high), respectively. Leg vascular conductance was reduced by 13 ± 4, 27 ± 10, 45 ± 7, and 45 ± 6%, respectively.

Fig. 3.

Change in blood flow and vascular conductance during ATP-induced vasodilation with and without inhibition of nitric oxide (NO) and prostaglandin formation. Change in leg blood flow, mean arterial pressure (MAP), and vascular conductance during intra-arterial ATP infusion with and without coinfusion of l-NMMA or indomethacin. Data are means ± SE for 6 and 7 subjects. *Significantly different from control, P < 0.05.

Leg V̇o2 remained unchanged in all conditions, since the lower O2 delivery with l-NMMA during baseline conditions and l-NMMA and INDO during ATP infusion was paralleled by an increase in arteriovenous O2 difference (P < 0.05) (Table 2).

View this table:
Table 2.

Blood variables before and during ATP infusion alone and in combination with l-NMMA and/or indomethacin

Effect of INDO and l-NMMA on Systemic Variables During ATP Infusion

At the systemic circulation level, neither l-NMMA nor INDO had any effect on cardiac output (5.9 ± 0.2, 6.1 ± 0.2, 6.1 ± 0.2, and 5.5 ± 0.1 l/min during control, l-NMMA, INDO, and l-NMMA + INDO, respectively), HR (65 ± 4, 67 ± 5, 59 ± 4, and 55 ± 4 beats/min, respectively), and stroke volume (92 ± 6, 95 ± 7, 104 ± 6, and 109 ± 5 l/min, respectively) during basal conditions. During ATP infusion, cardiac output and HR increased in all conditions, but HR rate was lower with l-NMMA compared with ATP infusion alone (P < 0.05) (Fig. 4). Stroke volume increased during ATP infusion with and without coinfusion of l-NMMA and l-NMMA + INDO but was lower when ATP was coinfused with INDO and l-NMMA + INDO compared with control (P < 0.05).

Fig. 4.

Change is systemic hemodynamic responses during ATP-induced vasodilation with and without inhibition of NO and/or prostaglandin formation. Change in estimated cardiac output, heart rate (HR), and stroke volume during 5–7 min of ATP infusion with and without coinfusion of l-NMMA and/or indomethacin. Data are means ± SE for 6–8 subjects. P < 0.05, significantly different from baseline value (§) and significantly different from control (*).

Effect of TEO on Leg Hemodynamics During ATP Infusion

TEO increased LBF (0.3 ± 0.0 and 0.5 ± 0.1 l/min in control and TEO, respectively), MAP (94 ± 3 and 98 ± 3 mmHg, respectively), and leg vascular conductance (3 ± 0 and 5 ± 1 ml·min−1·mmHg−1, respectively) during baseline conditions (all P < 0.05). During ATP infusion, there was no difference in the ATP-induced change in LBF, MAP, and leg vascular conductance between control and TEO (Fig. 5).

Fig. 5.

Change in leg hemodynamics during ATP infusion with and without adenosine receptor blockade. Change in leg blood flow, MAP, and vascular conductance during intra-arterial ATP infusion before and after infusion of the adenosine receptor blocker theophylline. Data are means ± SE for 6 subjects.

Analysis of mRNA Levels for Purinergic Receptors and Immunohistochemistry

CT values for the gene expression of the 10 investigated P2 purinergic receptors indicated that P2Y2 mRNA level was the highest of the investigated receptors in skeletal muscles. With the exception of P2Y2, the remaining investigated subtypes could all be detected, but the levels, as indicated by the CT values, were very low (Table 3). Localization of P2Y2 showed a distinct staining in the endothelium of capillaries and in the endothelium and smooth muscle cells of microvessels, whereas P2X1 receptors were present in the vascular endothelium and in skeletal muscle sarcolemma (Fig. 6).

Fig. 6.

Immunohistochemical localization of purinergic P2Y2 and P2X1 receptors in human skeletal muscle. A: positive staining for P2Y2 is shown (APR-010; Alomone Laboratories). B: positive staining for endothelium [Ulex Europeaeus Agglutinin-1 (UEA-1)] on the same section as in A is shown. C: two stainings superimposed. Positive staining for the P2Y2 purinergic receptor was evident in endothelial cells of capillaries and microvessels (white arrow) and in vascular smooth muscle cells (white arrowhead). D: immunopositive staining for the P2X1 receptor (APR-001; Alomone Laboratories). E: staining for endothelium (UEA-1) on the same section as in D. F: two stainings superimposed showing the presence of the P2X1 receptor in vascular endothelium (filled arrow) and in skeletal muscle sarcolemma (open arrow). The purinergic receptors were visualized by coupling to fluorescein isothiocyanate, and UEA-1 was visualized by direct coupling to TRITC.

View this table:
Table 3.

CT values for the mRNA level of the 10 investigated P2 purinergic receptors


This study investigated the role of NO, prostaglandins, and adenosine in ATP-induced vasodilation as well as purinergic P2 receptor expression in human skeletal muscle. The main findings were that: 1) infusion of either l-NMMA or INDO attenuated the vasodilatory response to ATP, whereas infusion of TEO had no effect, 2) combined blockade of NO and prostaglandins with coinfusion of l-NMMA and INDO resulted in the greatest attenuation in the ATP-mediated vasodilatory response, 3) the mRNA expression of the ATP-selective receptor P2Y2 is in human skeletal muscle markedly higher than any of the other nine P2 receptors investigated, 4) the P2Y2 receptors are present on the vascular endothelium of capillaries and other microvessels perfusing the leg muscles and in vascular smooth muscle cells, and 5) the P2X1 receptors are present in the vascular endothelium and skeletal muscle sacrolemma. These results indicate that NO and prostaglandins are important mediators of ATP-induced vasodilation in the human skeletal muscle vasculature, since they account for at least 40% of ATP-mediated leg vasodilation. Furthermore, the expression and the endothelial localization of the nucleotide-selective P2Y2 and P2X1 receptors suggest that these receptors may mediate ATP-induced vasodilation in skeletal muscle.

Single inhibition of either NO or prostaglandin formation reduced leg vascular conductance during ATP infusion by ∼12 and ∼26%, respectively, whereas combined inhibition of the two systems reduced the vascular conductance by ∼42%. Infusion of l-NMMA alone or in combination with INDO altered baseline flow, which could contribute to the reduced LBF during ATP administration, since an absolute amount of vasodilator (ATP) was applied to baseline conditions. However, with the reduced baseline flow accounted for, the reduction in vascular conductance during ATP infusion was still ∼13 and ∼45% during NOS and combined NOS and COX inhibition, respectively (Fig. 3). Thus the alteration in baseline values did not appear to significantly affect the vascular response. This observation is similar to what has been observed for the vasodilatory response to adenosine during NOS blockade in humans, where baseline correction with the NO donor sodium nitroprusside was found not to alter the blood flow response (46). Furthermore, it has repeatedly been shown that, although NOS blockade lowers vascular conductance at rest, it does not significantly alter the level during exercise, suggesting that the degree of arteriolar constriction at baseline does not greatly affect the hyperemic response (37, 45). Taken together, the current findings therefore suggest that NO and prostaglandins are released in response to ATP-induced vasodilation, which is in agreement with in vitro studies demonstrating that single inhibition of either NO (l-NAME) or prostaglandin (INDO) synthesis blunts the response to intraluminal ATP administration (20, 27). In contrast to the present findings, NO and prostaglandin inhibition did not change the response to ATP infusion in the human forearm (50). However, the target mass and thus the magnitude of blood flow in this model is relatively low, and small changes due to pharmacological interventions may therefore be undetectable. Furthermore, there could be limb differences in the response to vasodilators (32). Interestingly, the reduction in leg vascular conductance during combined inhibition of NO and prostaglandin synthesis was similar to the combined effect inhibiting these two systems alone, indicating that there does not seem to be any significant redundancy between the two systems during ATP administration. The present data therefore suggest that, in the human leg, NO and prostaglandins independently play an important role in mediating ATP-induced vasodilation.

In agreement with the reduced flow, the hyperemic conditions during ATP infusion were closely matched by an increase in cardiac output in all conditions, which remained lower during inhibition with l-NMMA compared with control conditions (Fig. 4). It is unlikely that the reduced LBF was caused by systemic effects of the infused drugs, since infusion of INDO and/or l-NMMA in the femoral vein did not reduce cardiac output or MAP. Similarly, we have previously found that ATP infusion in the femoral vein does not affect systemic or leg hemodynamics (16). These observations therefore suggest that altered leg vasodilation is the cause for the parallel changes in cardiac output during ATP infusion as the central circulation adapts to the reduced flow.

The current observation that the mRNA expression of most of the subtypes of purinergic P2 receptors is very low in skeletal muscle, and that only the P2Y2 receptor seems to be relatively highly expressed at the mRNA level, suggests that this isoform is dominant in skeletal muscle. This suggestion is supported by the immunohistochemical analysis revealing that the P2Y2 protein was present in endothelium of capillaries and other microvessels and in vascular smooth muscle. In partial agreement with the present findings, P2Y2 receptors have previously been shown to be most abundant in mammary artery smooth muscle cells, whereas P2Y1 and P2Y11 were found to be most abundant in umbilical vein endothelial cells (51). Interestingly, P2Y2 receptors, but not P2X1 receptors, were present in smooth muscle cells, indicating that they also are involved in mediating smooth muscle contraction (48). Although further studies employing specific antagonists are warranted to determine which P2 receptor subtypes are important for ATP-induced vasodilation, our data suggest that, of the known ATP-sensitive P2Y receptors (P2Y2, P2Y4, and P2Y11), the vasodilatatory effect of plasma ATP in the skeletal muscle vasculature is most likely to be mediated by P2Y2 receptors because the mRNA expression of P2Y4 and P2Y11 receptors was very low or undetectable. However, the presence of ATP-sensitive P2X1 receptors in the endothelium also supports a role of these receptors in mediating vasodilation (21).

The vasodilatory response to ATP was not completely abolished during simultaneous inhibition of NO and prostaglandin formation. There are several possible explanations for this observation. First, ATP is degraded rapidly by membrane-bound and soluble nucleotidases and could therefore induce vasodilation through its breakdown product adenosine (19, 52). To examine this possibility, we blocked P1 receptors and found that it did not reduce the ATP-induced hyperemia. This finding is in agreement with a previous study in the human forearm (39). Second, ATP may in part increase blood flow by blunting the basal sympathetic vasoconstrictor activity (functional sympatholysis). Indeed, in vivo evidence suggests that intra-arterial ATP infusion can override an augmented α-adrenoreceptor stimulation (23, 40), and sympathetic denervation increases limb blood flow at rest by ∼30% (22). Third, ATP could also induce vasodilation via P2Y receptors located on the smooth muscle cells, an effect that may not require the formation of vasodilators. This is supported by some in vitro studies indicating that ATP exerts vasodilation partially via endothelium-independent mechanisms (26, 28). Fourth, ATP-induced vasodilation may be endothelium-derived hyperpolarizing factor (EDHF) dependent. EDHF blockade with, tetraethylammonium chloride (TEA; Ca2+-dependent K+ channel blocker) did not alter the response to ATP-induced vasodilation in the human forearm (50), but TEA may not block all EDHFs (7, 31). Fifth, the infused inhibitors may not have completely abolished NOS and COX activity. When we infused l-NMMA and INDO at a five times higher dose (62.5 ± 3.0 and 3.1 ± 0.1 mg/min, respectively) in three subjects, we found a similar (2.4 ± 0.1 vs. 2.6 ± 0.1 l/min with low and high dose, respectively) reduction in LBF during ATP infusion, indicating that, if NOS and/or COX was not sufficiently inhibited, it was likely caused by a failure of the inhibitors to reach their target rather than an insufficient dose of inhibitors.

Indirect evidence suggests that ATP released from erythrocytes could play an important role in the precise matching of skeletal muscle O2 delivery to the metabolic demand of contracting myocytes (11, 13, 17). We have previously examined the effect of double blockade on exercise hyperemia in conditions that produced similar levels of hyperemia as in the present study (2.4 vs. 2.5 l/min during ATP infusion and exercise, respectively) allowing a comparison without the potentially confounding effects of signal triggered by the increases in muscle metabolism (30). In support of a role for ATP in skeletal muscle blood flow, we found a similar reduction in LBF during one-legged knee-extensor exercise with combined inhibition of NOS and COX than seen here in resting conditions. Nonetheless, a difference between studies is that single inhibition of NOS or COX during leg exercise does not reduce blood flow (14, 30, 37). Thus, if ATP plays an important role in skeletal muscle hyperemia, other factors, possibly related to muscle contractions and the enhanced metabolic demand, must explain the observed redundancy between the NOS and COX systems during muscle contractions.

Perspectives and Significance

The present data add support for a role of ATP as a blood flow regulator by demonstrating that the receptors capable of mediating the vascular effects of ATP are present in human skeletal muscle and that the same second messenger systems as during muscle contractions are mediators of ATP-induced vasodilation. This could indicate an involvement of ATP in exercise hyperemia although conclusive evidence for a role of ATP in skeletal muscle blood flow regulation is lacking and awaits the availability of selective P2Y antagonists. Furthermore, studies proving a link between changes in skeletal muscle blood flow and plasma ATP concentrations as well as knowledge about how ATP is released in plasma are needed to prove the physiological role of ATP.

In conclusion, these results suggest that NO and prostaglandins are important mediators of ATP-induced vasodilation in the human leg, whereas adenosine appears to be less important. The P2Y2 and/or P2X1 receptors located on the vascular endothelium may be important for the vasodilation induced by plasma ATP. Because much of the ATP-induced hyperemia was maintained during combined inhibition of NO and prostaglandin formation, ATP may also mediate vasodilation via other mechanisms.


This study was supported by a grant from the Lundbeck foundation and the Danish Medical Research Council. S. P. Mortensen was supported by a grant from the Copenhagen Hospital system.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract