Combined inhibition of nitric oxide and prostaglandins reduces human skeletal muscle blood flow during exercise

Abstract

The vascular endothelium is an important mediator of tissue vasodilatation, yet the role of the specific substances, nitric oxide (NO) and prostaglandins (PG), in mediating the large increases in muscle perfusion during exercise in humans is unclear. Quadriceps microvascular blood flow was quantified by near infrared spectroscopy and indocyanine green in six healthy humans during dynamic knee extension exercise with and without combined pharmacological inhibition of NO synthase (NOS) and PG by l-NAME and indomethacin, respectively. Microdialysis was applied to determine interstitial release of PG. Compared to control, combined blockade resulted in a 5- to 10-fold lower muscle interstitial PG level. During control incremental knee extension exercise, mean blood flow in the quadriceps muscles rose from 10 ± 0.8 ml (100 ml tissue)⁻¹ min⁻¹ at rest to 124 ± 19, 245 ± 24, 329 ± 24 and 312 ± 25 ml (100 ml tissue)⁻¹ min⁻¹ at 15, 30, 45 and 60 W, respectively. During inhibition of NOS and PG, blood flow was reduced to 8 ± 0.5 ml (100 ml tissue)⁻¹ min⁻¹ at rest, and 100 ± 13, 163 ± 21, 217 ± 23 and 256 ± 28 ml (100 ml tissue)⁻¹ min⁻¹ at 15, 30, 45 and 60 W, respectively (P < 0.05 vs. control). In conclusion, combined inhibition of NOS and PG reduced muscle blood flow during dynamic exercise in humans. These findings demonstrate an important synergistic role of NO and PG for skeletal muscle vasodilatation and hyperaemia during muscular contraction.

Endothelium-derived vasodilator substances such as nitric oxide (NO), prostaglandins (PG) and endothelium-derived hyperpolarization factor (EDHF) are considered important mediators of tissue blood flow (Duffy et al. 1999a; Halcox et al. 2001). However, their role in regulating exercise-induced muscle blood flow in humans is currently debated and remains unresolved (Radegran & Hellsten, 2000).

A functional vasodilatory role of PG in exercise hyperaemia is compatible with findings of increases in both circulating PG (Wilson & Kapoor, 1993) and interstitial PG, determined by microdialysis, in muscle during exercise (Frandsen et al. 2000). However, discrepant findings in healthy individuals regarding the effect of PG blockade on blood flow at rest and during exercise have been demonstrated (Kilbom & Wennmalm, 1976; Cowley et al. 1984; Wilson & Kapoor, 1993; Shoemaker et al. 1996; Duffy et al. 1999b). Furthermore, the role of NO for functional hyperaemia is unclear, as NOS blockade either with N-nitro-l-arginine methyl ester (l-NAME) or N-monomethyl l-arginine (l-NMMA) diminishes limb blood flow at rest and post exercise, but does not influence blood flow during exercise (Shoemaker et al. 1997; Radegran & Saltin, 1999; Frandsen et al. 2001). One explanation for the lack of any clear effect on blood flow during exercise when either NO or PG is blocked separately, is that compensatory responses may result to ensure blood flow matches metabolic demand. In accordance with this view, it has been shown that PGI₂ synthesis is enhanced during NOS blockade (Barker et al. 1996; Osanai et al. 2000) and conversely that bradykinin-induced PGI₂ production is suppressed during infusion of NO donors (Mathews et al. 1995). However, there is no evidence to support a compensatory increase in muscle interstitial PG release during exercise in humans under NOS blockade by l-NMMA (Frandsen et al. 2000).

In light of the uncertain role of these vasodilator substances, the purpose of this study was to determine the influence of combined inhibition of PG and NO synthase on muscle blood flow during dynamic exercise. We hypothesized that PG and NO function synergistically to control blood flow during exercise and that simultaneous inhibition of these substances would attenuate the exercise-induced increase in muscle blood flow.

Regional microvascular blood flow in the quadriceps muscles was quantified using near infrared spectroscopy (NIRS) and indocyanine green (Boushel et al. 2000). Local interstitial changes in PG with and without pharmacological blockade were measured by microdialysis positioned adjacent to the NIRS probes.

Methods

Six young, healthy individuals participated in the study after informed written consent as approved by the Ethical Committee of Copenhagen and Frederiksberg (01-210/99) and in accordance with the Declaration of Helsinki. Under local anaesthesia, a catheter was inserted in the right femoral artery using the Seldinger technique, and another catheter similarly placed retrogradely in the femoral vein of the same leg. Cardiac output was measured by indocyanine green (Cardio-Green; Passel & Lorel GmbH & Co., Hanau, Germany) dye dilution. Dye (4-5 mg) was injected rapidly into the femoral vein from a calibrated syringe followed by a 5 ml flush of isotonic saline. Blood from the femoral artery was drawn with a pump (Harvard, 2202A) at 20 ml min⁻¹ through a linear photodensitometer (Waters CO-10, Rochester, MN, USA) for measurement of the arterial dye concentration. The dye curves were displayed on a chart recorder (Gould 8000) and collected in a data acquisition system (PowerLab, AD Instruments). Withdrawn arterial blood was reinfused into the femoral vein in a closed loop system. Cardiac output was computed as the ratio of dye injected to the average arterial indocyanine green (ICG) concentration over the time interval of the curve and expressed as [ICG] per minute. Following each experiment an ICG calibration curve was derived from measuring the voltage deflection from three separate samples of blood from each subject with known concentrations of ICG added.

Blood pressure was determined from pressure transducers (T100209A, Baxter, Unterschleissheim, Germany) positioned at the level of the heart, interfaced with a monitor (Dialogue 2000, Danica, Copenhagen) and connected to a data acquisition system (PowerLab, AD Instruments). Mean arterial pressure was determined as the integrated pressure wave curve over time. Total peripheral resistance was determined as the ratio of mean arterial pressure to blood flow.

Microvascular blood flow in the quadriceps muscles was measured with a NIRO 300 (Hamamatsu Photonics, Germany) spectrophotometer with dual-channel laser diodes during bolus injections of indocyanine green (ICG) as previously described in detail (Boushel et al. 2000). This method has been validated in humans against the ¹³³Xe and dye dilution methods. One set of emitting and receiving optodes was placed in the vertical plane over the vastus lateralis midway between the greater trochanter and patella and another over the vastus medialis 80 mm superior to the patella. The optode separation distance for both muscles was 50 mm, corresponding to a penetration depth of ≈25 mm. Light attenuation in muscle was measured at 6 Hz immediately after venous bolus injection of ICG. Changes in tissue [ICG] were determined by measuring light attenuation at 775, 813, 850 and 913 nm wavelengths, analysed with an algorithm incorporating the Modified Beer-Lambert Law.

The ICG contribution to the absorption signal was isolated using a matrix operation (MATLAB) incorporating pathlength-specific extinction coefficients for each of the chromophores at each wavelength employed by the NIRS device (Hamamatsu photonics). Blood flow was calculated as the ratio of tissue ICG accumulation measured by NIRS, to the arterial [ICG] determined by photodensitometry over the same time intervals (Boushel et al. 2000).

In addition to quantifying blood flow, separate analyses were made to evaluate patterns of muscle ICG accumulation. The magnitude of peak muscle [ICG] was determined for each curve, as was the dispersion rate, which represents the time for the tissue to receive half of the ICG bolus (from 10–90 % of the first half of the curve).

Blood was sampled anaerobically from the femoral artery and vein simultaneously for measurement of blood gases and potassium, and pH (Oxylite, Radiometer, Copenhagen). Microdialysis was performed in principle as previously described (Langberg et al. 1999). Under ultrasound guidance, microdialysis catheters (CMA 60; CMA/Microdialysis AB; 20 kDa molecular cut-off, 0.5 mm outer diameter; length 30 mm) were inserted parallel to the muscle fibres in m. vastus lateralis and m. vastus medialis, directly adjacent to the NIRS probes. Two microdialysis catheters were positioned in each muscle with an inter-probe distance of 10 mm and perfused, via a high-precision syringe pump (CMA 100; Carnegie Medicine, Solna, Sweden), at a rate of 5 μl min⁻¹ with a Ringer acetate solution containing 3 mm glucose and 1 mm lactate. The in vivo recovery of prostaglandin E₂ (PGE₂) was determined by adding 5 nm [15-³H(N)]-PGE₂ (specific activity: 3.7 GBq mmol⁻¹; NEN, Boston, USA) as previously described (Langberg et al. 1999). The samples were immediately frozen to −70 °C until analyses were done 1–2 weeks later.

PGE₂ was analysed using a commercially available PGE₂ radioimmunoassay kit (NEK-020, Du Pont, Boston, MA, USA). Samples or standards, together with ¹²⁵I-PGE₂ as the tracer, were incubated with rabbit anti-PGE₂ antibodies overnight at 4 °C. The samples were precipitated by polyethylene glycol, centrifuged, decanted and the radioactivity in the pellet was determined in a gamma counter.

Pharmacological interventions

Cyclooxygenase inhibition was accomplished by oral administration of 100 mg indomethacin (Confortid, Alpharma, Denmark) 16 h prior to the experiment. NO synthase inhibition was performed by systemic administration of l-NAME (Clinalfa, Laufelfingen, Switzerland). l-NAME (4 mg kg⁻¹) was dissolved in 30 ml of sterile saline and infused by automated pump over 30 min as previously described (Sander et al. 1999). Reversal of NO synthase inhibition was done by intravenous l-arginine (200 mg kg⁻¹ over 10 min) (Clinalfa) after completion of all measurements in the l-NAME protocol.

Protocol

After a resting period of 20 min in the seated position, baseline resting measurements were obtained for heart rate and blood pressure, followed by a bolus injection of ICG for simultaneous measurement of cardiac output and muscle blood flow. Femoral arterial blood was withdrawn by a Harvard pump through the photodensitometer, while 5 mg of ICG was rapidly injected into the femoral vein followed by a 10 ml flush of saline. The arterial ICG dye curve was recorded after circulation to the femoral artery and with a short time delay, the NIRS-ICG signals in quadriceps muscles were recorded. The measurements of cardiac output and NIRS-ICG blood flow were completed within ≈30 s. Femoral arterial and venous blood samples were then taken for blood gases and metabolites. Subjects then began dynamic knee extension exercise at 15 W for a period of 10 min. Heart rate and blood pressure were recorded continuously on-line over the exercise interval. Cardiac output and NIRS-ICG blood flow were determined after 4 min of exercise, followed by arterial and venous blood sampling and the same procedures were repeated at 9 min. Subjects then rested for 10 min and the same measurement sequence was followed for exercise at 30, 45 and 60 W.

Data analysis

Values are presented as means ± s.e.m. Differences in haemodynamic parameters and blood gases with exercise and between control and combined blockade of PG and NOS were analysed by the Friedman test and if found significant such differences were located by Wilcoxon's test. Differences were considered significant if P < 0.05.

Results

At rest under control conditions, muscle blood flow was 10.5 ± 0.8 ml (100 ml)⁻¹ min⁻¹ in the vastus lateralis (VL) and 10.1 ± 1 ml (100 ml)⁻¹ min⁻¹ in the vastus medialis (VM) (Fig. 1). During knee extension exercise, blood flow in the VL and VM rose similarly to, respectively, 120 ± 16 and 127 ± 23 ml (100 ml)⁻¹ min⁻¹ at 15 W, 238 ± 18 and 252 ± 31 ml (100 ml)⁻¹ min⁻¹ at 30 W, 325 ± 14 and 334 ± 35 ml (100 ml)⁻¹ min⁻¹ at 45 W, and 300 ± 13 and 323 ± 37 ml (100 ml)⁻¹ min⁻¹ at 60 W. During combined inhibition of NOS and PG, values for blood flow in the VL and VM were, respectively, 9 ± 0.3 and 8 ± 0.6 ml (100 ml)⁻¹ min⁻¹ at rest. During exercise with blockade, blood flow in the VL and VM rose to, respectively, 94 ± 7 and 107 ± 19 ml (100 ml)⁻¹ min⁻¹ at 15 W, 150 ± 14 and 175 ± 29 ml (100 ml)⁻¹ min⁻¹ at 30 W, 212 ± 14 and 221 ± 32 ml (100 ml)⁻¹ min⁻¹ at 45 W and 242 ± 22 and 270 ± 38 ml (100 ml)⁻¹ min⁻¹ at 60 W. In both VL and VM, blood flow was significantly reduced at 30–60 W by NOS and PG inhibition (P < 0.05).

Figure 1. Quadriceps microvascular blood flow during exercise.

Regional microvascular blood flow in the quadriceps muscles during incremental dynamic knee extension (upper panel). Blood flow in the vastus lateralis (middle panel) and vastus medialis (lower panel) in control knee extension (open symbols) and during combined NOS + PG blockade (filled symbols). Asterisks indicate difference between control and blockade conditions (P < 0.05).

A representative tracing of ICG accumulation and dispersion during exercise (30 W) in control and blockade trials is shown in Fig. 2. At rest in the control trial, dispersion times in the VL and VM were, respectively, 6 ± 0.5 and 6.2 ± 0.9 s and fell to 2.6 ± 0.2 and 2.3 ± 0.2 s at the peak load of 60 W. In the blockade trial, dispersion times were significantly longer in both muscle regions at rest and during all exercise loads (P < 0.05). At rest, the durations were 10 ± 0.9 and 9.8 ± 0.8 s in the VL and VM, respectively, and decreased to 3.28 ± 0.2 and 3.14 ± 0.1 s at 60 W. Peak [ICG] accumulation in the control condition at rest in the VL and VM were, respectively, 0.3 ± 0.05 and 0.28 ± 0.05 μg (100 ml)⁻¹. During exercise at 15, 30, 45 and 60 W, peak [ICG] rose to, respectively, 2.96 ± 0.5 and 3.25 ± 0.5, 5.4 ± 0.6 and 5.64 ± 0.7, 6.93 ± 0.7 and 6.96 ± 0.7, and 5.3 ± 0.7 and 5.73 ± 0.5 μg (100 ml)⁻¹. In the blockade trial, peak [ICG] accumulation in the VL and VM at rest were, respectively, 0.62 ± 0.14 and 0.46 ± 0.07 μg (100 ml)⁻¹, and at 15, 30, 45 and 60 W, rose to, respectively, 2.8 ± 0.5 and 3.8 ± 0.6, 4 ± 0.5 and 4.5 ± 0.8, 5.2 ± 0.5 and 5.4 ± 0.6, and 5.16 ± 0.4 and 6.06 ± 0.6 μg (100 ml)⁻¹. Peak ICG accumulation was reduced during the blockade trial at 30 and 45 W (P < 0.05).

Figure 2. Muscle ICG accumulation.

A representative tracing of ICG accumulation patterns during exercise at 30 W in control conditions (continuous line), and during NOS + PG blockade (interrupted line).

In the control trial, the mixed femoral venous O₂ saturation (S_v,O2) was 62 ± 3 % at rest, 46 ± 4 % at 15 W, 44 ± 3 % at 30 W, 44 ± 2 % at 45 W, and 45 ± 3 % at 60 W (Fig. 3). During the blockade trial, S_v,O2 was lower than in the control trial at rest (52 ± 3 %), and during exercise at 15 W (33 ± 2 %), 30 W (31 ± 2 %), 45 W (32 ± 1 %) and 60 W (35 ± 0.5 %) (P < 0.05). Similarly, venous P_O2 and O₂ content were lower in the blockade trial compared to the control trial (P < 0.05) as shown in Fig. 3. Consistent with the blood gas data indicating a lower leg blood flow at rest and during exercise, femoral venous pH was also lower (Fig. 3) and venous potassium was higher in the combined blockade trial compared to control (P < 0.05).

Figure 3. Mixed femoral venous blood parameters.

Femoral venous oxygen saturation (S_v,O2; upper left), oxygen pressure (P_v,O2; upper right), oxygen content (C_v,O2; middle left), potassium (K⁺; middle right), carbon dioxide pressure (P_v,CO2; lower left), and pH (lower right) at rest and during dynamic knee extension. Open symbols are control exercise and filled symbols are combined NOS + PG blockade. Asterisks indicate difference between control and blockade conditions (P < 0.05).

In the control trial, cardiac output (CO) was 5.2 ± 0.3 l min⁻¹ at rest and increased to 7.2 ± 0.6, 10 ± 0.6, 11.2 ± 0.5 and 12 ± 0.43 l min⁻¹ during exercise at 15, 30, 45 and 60 W, respectively (Fig. 4). In the NOS and PG inhibition trial, CO was lowered at rest and at all exercise loads (P < 0.05). At rest and at 15, 30, 45 and 60 W, respectively, CO was 3.32 ± 0.3, 4.8 ± 0.5, 6.7 ± 0.5, 8.1 ± 0.5 and 10.7 ± 0.7 l min⁻¹. In the control trial, mean arterial pressure (MAP) at rest was 89 ± 2 mmHg, and during exercise at 15, 30, 45 and 60 W, it rose to 96 ± 3, 98 ± 2, 100 ± 3, and 105 ± 3 mmHg, respectively. In the blockade trial, MAP was 106 ± 2 mmHg at rest and rose to 113 ± 2, 114 ± 4, 117 ± 5 and 121 ± 4 mmHg, at 15, 30, 45 and 60 W, respectively, and all values were higher than in the control trial (P < 0.05). Total peripheral resistance (TPR) in the control trial was 17 ± 1 mmHg l⁻¹ min, and during exercise at 15, 30, 45 and 60 W it fell to 14 ± 1, 9.4 ± 0.6, 8.8 ± 0.4 and 9.3 ± 0.5 mmHg l⁻¹ min, respectively. In the blockade trial, TPR was 32 ± 2 mmHg l⁻¹ min at rest and fell to 25 ± 4, 17 ± 0.6, 15 ± 0.7 and 11 ± 0.5 mmHg l⁻¹ min at 15, 30, 45 and 60 W, respectively, and all values were higher than in the control trial (P < 0.05).

Figure 4. Systemic haemodynamics.

Cardiac output (upper panel), mean arterial blood pressure (MAP; middle panel), and total peripheral resistance (TPR; lower panel), at rest and during incremental knee extension exercise. Open symbols are control exercise and filled symbols are combined NOS + PG blockade. Asterisks indicate difference between control and blockade conditions (P < 0.05).

The muscle interstitial PG content collected by microdialysis from the quadriceps 60 min after probe insertion in the control trial was 6.8 ± 3 ng ml⁻¹ in the vastus medialis and 3.9 ± 1 ng ml⁻¹ in the vastus lateralis. After 90 min they fell to 4.3 ± 1.7 and 3.2 ± 1 ng ml⁻¹, respectively. At the first exercise load of 15 W, PG values were 2.1 ± 0.2 ng ml⁻¹ in the vastus medialis and 1.6 ± 0.7 ng ml⁻¹ in the vastus lateralis. They were unchanged at 30 W, but increased after the heavier loads of 45 and 60 W to, respectively, 4.4 ± 1 and 3.6 ± 0.7 ng ml⁻¹ in the vastus medialis and 3.2 ± 1 and 3 ± 0.6 ng ml⁻¹ in the vastus lateralis (P < 0.05). During the blockade trial, interstitial PG content measured 60 min after probe insertion was 0.4 ± 0.04 ng ml⁻¹ in the vastus medialis and 0.42 ± 0.1 ng ml⁻¹ in the vastus lateralis. The was no change from baseline level in the blockade trial, and [PG] remained 5- to 10-fold lower throughout the exercise and recovery periods.

Discussion

The important new finding in this study was that regional microvascular blood flow in the quadriceps femoris muscles was markedly reduced (up to 50 %) during exercise under combined inhibition of PG and NO by indomethacin and l-NAME, respectively. This study is, to our knowledge, the first to report the influence of combined blockade of these vasodilator substances on functional hyperaemia in humans. The present results support the concept that NO and PG play a synergistic role in the control of regional muscle blood flow during dynamic exercise, and are in agreement with findings on a role for NO and PG in regulating coronary blood flow (Duffy et al. 1999a). In addition to attenuated blood flow to specific regions of muscle during exercise, the mixed femoral venous blood data in this study also indicate that blood flow to quadriceps muscles as a whole was significantly reduced. Venous O₂ saturation, O₂ pressure and O₂ content were lowered under combined NOS and PG inhibition, reflecting enhanced O₂ extraction to compensate for the reduction in muscle O₂ delivery. Also reflecting a lower blood flow were the lower mixed venous pH and higher potassium (Fig. 3). The present findings are of interest in the light of previous studies where no clear, independent vasoregulatory role has been observed for PG or NO during muscle contraction.

Previous studies which have used either ultrasound Doppler flowmetry or thermodilution to measure whole limb blood flow, found that NOS blockade alone by either l-NAME or l-NMMA significantly diminished muscle blood flow at rest and post exercise, whereas no effect of blockade was observed during exercise (Shoemaker et al. 1997; Radegran & Saltin, 1999; Frandsen et al. 2001). Others who have used plethysmography to measure forearm blood flow have also shown a decrease in muscle blood flow at rest and have implied that blood flow is reduced also during exercise (Gilligan et al. 1994; Duffy et al. 1999b). These discrepant results may be explained in part by the different techniques used. As plethysmographic measurements of flow are confined to conditions of rest, intervals after contractions and reactive hyperaemia, these findings may actually reflect the early post-exercise blood flow response. As such, they are consistent with thermodilution and ultrasound Doppler flowmetry findings of a role for NO at rest and during recovery from, but not during contraction (Radegran & Hellsten, 2000).

As with NO, a role for PG in muscle vasodilatation in humans has been demonstrated at rest (Duffy et al. 1999b), during exercise and reactive hyperaemia (Kilbom & Wennmalm, 1976; Cowley et al. 1984; Wilson & Kapoor, 1993). Yet, most studies supporting a role for PG in functional hyperaemia in humans are also based on plethysmographic measures which may reflect post-exercise blood flow. Accordingly, it has been shown in healthy individuals that forearm blood flow is unaffected during dynamic forearm exercise with PG blockade measured by ultrasound Doppler flowmetry (Shoemaker et al. 1996). This pattern is in contrast to findings in heart failure patients where leg blood flow measured by thermodilution was reduced during exercise with PG blockade (Lang et al. 1997), which may be explained by a differential metabolic control of blood flow in heart failure patients.

Another important finding in this study was that combined inhibition of NO and PG production caused a significant increase in MAP and TPR, as well as a reduction in CO at rest and during exercise. It has been demonstrated previously that systemic NOS inhibition with l-NAME induces a significant increase in MAP in humans at rest, and this effect is attributed to an attenuation of basal vasodilatation as well as to a loss of NO modulation of sympathetically mediated vasoconstriction (King-Van Vlack et al. 1998; Sander et al. 1999). To our knowledge, this is the first study to report a reduction in CO during exercise in humans with NO and PG blockade, either separately or in combination. The reduction in CO both at rest and during exercise was associated with a significant lowering of heart rate, which is consistent with previous observations during NOS blockade with l-NAME (Frandsen et al. 2000, 2001). While there is insufficient evidence to suggest that NOS and PG blockade intrinsically reduce CO, the higher TPR and MAP during NOS and PG blockade may have contributed to lower CO due to baroreflex-mediated lowering of heart rate as previously suggested (Frandsen et al. 2001). This response is relevant as it could be argued that muscle blood flow was reduced due to diminished CO as seen during β₁ receptor blockade (Pawelczyk et al. 1992) and in heart failure (Sullivan & Cobb, 1991). Yet, in this study, despite a lower baseline value, CO increased with exercise intensity at a rate similar to that in the control trial. Furthermore, using the same exercise model as in the present study, Frandsen et al. (2001) found that NOS blockade alone with l-NAME had no influence on whole leg blood flow nor O₂ extraction during exercise despite virtually identical effects on heart rate and blood pressure as in the present study. While they did not measure CO, the same lowering of heart rate and elevation of blood pressure during l-NAME administration imply that CO was also reduced, but without influence on leg perfusion or O₂ extraction. Thus, in comparison to the findings of Frandsen et al. (2000, 2001) and others who have examined the influence of NOS blockade alone (Shoemaker et al. 1997; Radegran & Saltin, 1999), the results of this study showing both a decrease in regional muscle blood flow and a persistent increase in whole leg O₂ extraction during exercise, emphasize that the combined actions of NO and PG are important for muscle vasodilatation and blood flow regulation during muscle contraction.

To further examine the dynamics of NOS and PG blockade on regional blood flow, we analysed the patterns of ICG accumulation in muscle. As shown in Fig. 2, muscle blood flow was reduced with NOS and PG inhibition due to a slower rate of ICG accumulation (dispersion) and a lower magnitude of ICG accumulation (see Results). Both components together (rate and magnitude of ICG accumulation) define the level of tissue blood flow for a given [ICG] in the feed artery (femoral). Under blockade conditions where CO was reduced, there was a higher concentration of ICG in the femoral artery. If a similar regional microvascular dilatation in muscle tissue occurred as in control exercise, there would be a comparable or larger magnitude of ICG accumulation in muscle under blockade conditions, which did not occur. These data support the conclusion that combined inhibition of NOS and PG attenuated muscle vasodilatation during exercise.

The effectiveness of the PG blockade was assessed by determination of interstitial PGE₂ from microdialysis probes placed in both the vastus lateralis and medialis muscle regions adjacent to the NIRS probes. The results showed that indomethacin reduced [PGE₂] by ≈10-fold at rest, and there was no increase throughout the exercise periods, which importantly indicates a diminished PG influence at the vascular branching order where blood flow is controlled. In the control trial, a progressive decrease in [PGE₂] occurred over the 95 min resting period following placement of the microdialysis probes, reflecting a gradual recovery from the initial effect of insertion. [PGE₂] levels plateaued during and after the 15 W bout, and remained unchanged after 30 W. However, there was a significant rise in [PGE₂] at 45 and 60 W (P < 0.05), which supports a role of PG for muscle blood flow regulation at heavier work rates. Regarding the effectiveness of NOS blockade, we gave subjects the same dose of l-NAME that produced a 60 % reduction in NOS activity in a previous in vitro study (Frandsen et al. 2001). Together, there is evidence for substantial blockade of both vasodilator substances.

While the present study demonstrates significant reductions in blood flow during blockade (30 % on average), 50–70 % of the exercise-induced skeletal muscle blood flow was still maintained, and it remains to be determined what vasoactive substances other than PG and NO were released to enable blood flow to increase during exercise. EDHF is considered to play a role in muscle vasodilatation (Honing et al. 2000), and in this regard the finding that production of EDHF is attenuated by increased NO (Bauersachs et al. 1996) may imply that it is elevated under NOS inhibition. Consistent with this premise is the finding that when both NOS and PG are blocked at rest administration of miconazole, which blocks all cytochrome P450 enzymes and thus the formation of EDHF (Halcox et al. 2001), further attenuates the vasodilatory response to bradykinin infusion into the human forearm. Furthermore, preliminary experiments have shown that combined blockade of NOS and cytochrome P450 also reduces leg blood flow during exercise (Y. Hellsten & B. Saltin, unpublished data).

Considering that combined blockade of NO and PG elicits a significant decrease in muscle blood flow during exercise, bradykinin may be an important factor in exercise hyperaemia as it stimulates the release of NO, PG and EDHF. We have recently demonstrated that bradykinin release in the muscle interstitium, measured by microdialysis, increases during exercise and is correlated to exercise-induced increases in muscle blood flow (Langberg et al. 2002). While the vasodilatory response to adenosine is attenuated with NOS blockade (Smits et al. 1995; Woodley & Barclay, 1998), there are no compensatory increases in interstitial adenosine during exercise with l-NAME (Frandsen et al. 2000). It cannot be excluded that adenosine release during exercise is diminished during combined NO and PG blockade.

Several aspects of the study merit consideration for further investigation. First, it is unclear why blood flow was not significantly reduced at the 15 W load under blockade conditions as it was during the higher loads. It has been proposed that at such light work loads, blood flow is governed predominantly by mechanical factors while vasodilator substances become more important for increasing blood flow at higher intensities (Sheriff & Van Bibber, 1998; Radegran & Saltin, 1999). A related finding of interest in this study was that leg oxygen extraction was increased at 15 W under combined blockade while muscle blood flow was unchanged from control. A similar finding of enhanced oxygen extraction and oxygen uptake during exercise with NOS blockade has been reported in dogs (Shen et al. 2000), but this is the first study to show this response in humans. This pattern raises the possibility that muscle oxygen uptake was elevated in the blockade condition, a response that is consistent with the concept that NO inhibits mitochondrial respiration. Confirmation of this response would, however, require simultaneous measurement of local muscle oxygenation together with blood flow, and remains to be investigated.

It was expected that muscle blood flow would be reduced significantly at rest, as observed in other studies with either NOS or PG blockade. Compared to the ≈20 % flow reduction under combined blockade in this study, Radegran & Saltin (1999) found a 50 % reduction in quadriceps blood flow, while Shoemaker et al. (1997) found a 30 % decrease in brachial artery flow, both measured by ultrasound Doppler flowmetry during NOS blockade with l-NMMA. Also, Duffy et al. (1999b) reported a 30 % reduction in resting forearm flow using plethysmography after aspirin administration. One possible explanation for the different findings is that the NIRS technique measures more superficial regions of the muscle microcirculation compared to the other techniques, and there can be significant flow heterogeneity throughout a muscle (R. Boushel, H. Langberg, C. Gemmer, J. Olesen & M. Kjær, unpublished findings). Also, under resting conditions the NIRS photon path may penetrate a microvascular volume as low as 1 ml or less, and with the bolus ICG method we did not control for temporal fluctuations in flow that correspond to vasomotion. Future studies on the effect of isolated blockade of NOS and PG using the same exercise model and techniques would be important to confirm the synergistic role of these substances for regional muscle perfusion during exercise found in the present study.

In conclusion, the results of this study show a reduction in skeletal muscle blood flow during moderate and heavy exercise when NO and PG are inhibited pharmacologically, implying that the combined actions of NO and PG are essential for the normal increase in skeletal muscle blood flow during exercise. The findings also suggest the need to investigate further the potential influence of NO on muscle oxygen consumption during exercise in humans.