INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE STRONG AFFINITY of carbon monoxide (CO) for Hb has afforded researchers the opportunity to measure Hb mass and blood volume (5), examine O₂ transport (12, 18), and most recently to manipulate blood Hb levels while independently manipulating arterial PO₂ to study the control of skeletal muscle blood flow (9). Additionally, mild CO poisoning is an unavoidable consequence of cigarette smoking (36). The usefulness of CO as a tool is enhanced by the fact that unlike hypoxia (H), mild CO exposure does not stimulate the carotid body and does not result in an increase in cardiac output or ventilation (33). However, the validity of using CO in vascular research and the clinical consequences of mild CO poisoning are highly dependent on the magnitude of the diffusive movement of CO from the vascular to intracellular space and the subsequent intracellular consequences of any such movement.

In favor of limited CO movement and consistent with the normal physiological function of CO disposal (as a catalytic byproduct of heme), the ligand affinities for CO are the reverse of those for O₂. Hence, Mb has a higher CO affinity than cytochrome oxidase and Hb has a higher affinity than myoglobin (Mb). Additionally, on the basis of the high affinity Hb has for CO, the major driving force for CO movement (CO partial pressure in the blood) should theoretically remain low with a mild level of Hb-CO. Despite this concept, both researchers and clinicians concern's regarding the exposure of muscle to CO have revolved around the effect of CO on O₂ transport by interfering with Mb-facilitated O₂ diffusion and the metabolic effects of CO binding with cytochrome oxidase (7, 12, 18). In fact, there are data that do indeed suggest that some CO, even at low levels (2-2.5% Hb-CO), moves from Hb to Mb during exercise (6). However, currently there is limited understanding of the extent to which CO moves into skeletal muscle and the intracellular effects of this potential CO movement.

Consequently, the purpose of this study was to combine invasive vascular measures of arterial and venous blood and muscle blood flow (Q) with noninvasive magnetic resonance spectroscopy (MRS) of both deoxy-Mb and high-energy phosphates to determine both the intravascular and intracellular effects of mild CO poisoning (20% Hb-CO) in humans during muscular work. Specifically, beyond the expected increase in Q with decreased Hb-O₂, we tested the following hypotheses. 1) A 20% Hb-CO level will not reduce the intramuscular concentration of Mb ([Mb]) available for O₂ binding. 2) Submaximally, muscle oxygen consumption (O₂) will be unaffected by CO, whereas maximal oxygen consumption (O_2 max) will be attenuated by the effect CO has on O₂ availability. 3) Reduced O₂ availability will reduce intracellular PO₂ in the presence of CO. 4) The reduction in O₂ availability with CO will cause an early perturbation of muscle bioenergetics, as measured by intramuscular levels of phosphocreatine (PCr), ATP, and pH.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. Seven recreationally active males (age mean ± SD): 26 ± 1 yr; weight: 77 ± 4 kg; and height: 178.2 ± 4 cm, volunteered to participate in this study. All but one of the subjects were nonsmokers. The single smoker was a "recreational" smoker only (0-4 cigarettes/day). Informed written consent was obtained according to the Code of Ethics of the World Medical Association (Declaration of Helsinki), the Ethics Committee of Copenhagen and Frederiksberg communities, and the University of Pennsylvania Human Subjects Committee requirements. All subjects were successfully studied during the catheter-based investigations and five of these subjects completed the MRS portion of the work.

Exercise modality and protocol. Single leg knee-extensor exercise, designed to limit work to the quadriceps muscles of the left leg (24), was employed. Before the exercise protocol, all subjects were familiarized with the testing environment and ergometer. Briefly, subjects were semirecumbent in a chair placed in the laboratory for the catheter-based studies or within a 2.0-Tesla Oxford imaging magnet for the MRS studies. A special ankle boot placed on their left leg connected them by a bar to the ergometer (24). Contractions of the quadriceps muscles caused the lower part of the leg to extend from an angle of 90° to 170°. Therefore, the lower leg traveled with an arc-shaped trajectory of ~80°. The momentum of the ergometer passively returned the relaxed leg to the start position and, as a result, the quadriceps muscle was functionally isolated (23). The graded exercise tests required subjects to maintain each work rate for 180 s, after which the work rate was incremented. The subjects continued until they were unable to maintain a cadence of 60 rpm for the entire 3 min. During the catheter-based studies, heart rate and arterial blood pressure were recorded continuously and muscle Q was measured (thermodilution) at minutes 2 and 3 of exercise at each work level. Arterial and venous blood samples were withdrawn after Q measurements. During magnet-based studies, data were collected continuously, with the ³¹P and ¹H data being interleaved every 4 s. The exercise bouts involving CO were always performed last, due to the relatively long elimination period required to return to low Hb-CO levels; within this logistical restraint, the H and normoxia and normoxia + CO (CO_norm) and hyperoxia + CO (CO_hyper) treatments were performed in a balanced order across subjects. One hour before the exercise tests in the magnet, total ischemia in the left leg was induced by a suprasystolic cuff to allow the assessment and calibration of the Mb signal. The assessment of the cuffed Mb signal in the presence of CO was performed with the subject on the rebreathing system 1 h before the CO_norm exercise bout.

H and CO treatments. Before and during exercise in both the H and CO trials, subjects breathed in a closed-circuit system (previously illustrated in Ref. 9). This rebreathing system consisted of a 4-liter custom-made chamber containing CO₂ absorber (Sofnolime, Molecular Products, Essex, UK), a 10-liter reservoir, a two-way breathing valve (Hans-Rudolph, Kansas City, MO), and two hoses connecting the chamber and the breathing valve. The inspired gas in the reservoir was manually adjusted using gas regulators connected to two tanks containing 11% O₂ in N₂ and 100% O₂, respectively, while O₂ and CO₂ were monitored online (Beckman OM11 and LB2, Fullerton, CA). During the rest period after the hypoxic trial, CO (95% purity) was administered through an extension line connected to the two-way valve, using plastic syringes. A series of CO boluses estimated to increase Hb-CO by 10-15% was initially administered and venous Hb-CO was measured spectrophotometrically (OSM-3 Hemoximeter, Radiometer, Copenhagen, Denmark) after 20 min via a venous blood sample. A second CO dose aimed to increase Hb-CO to 20% was then administered (total CO administered 312 ± 25 ml).

MRS: [Mb], intracellular PO₂, and ³¹P. Spectra were collected from the muscle region below the 7-cm diameter surface coil double-tuned to proton (85.45 MHz) and phosphorus (34.59 MHz) placed over the rectus femoris portion of the quadriceps group (35) ~20-25 cm proximal to the knee. For these studies, this "sensitive region" was <100 cm³ of muscle, which isolated signal detection predominantly to the rectus femoris (1). Details of the theory behind oxygen-sensitive Mb signals have been published previously (3, 24). Briefly, the heme iron exhibits oxygen-dependent spin states that in turn influence nearby protons. The N- proton on proximal histidine F8, one of the ligands coordinated to the iron, is particularly sensitive to these changes. When oxygen is bound to the active site, the resonance of this proton is hidden beneath the dominant water signal. However, when Mb becomes deoxygenated, changes in the iron spin state shift this peak to a temperature-dependent position that is clearly distinct from all other resonances. Fractional deoxy-Mb (^fdeoxy-Mb) during exercise was determined by normalizing the signal areas (integration under the curve) to the average signal areas obtained between the 10th and 14th min of cuff ischemia at suprasystolic pressure (270 mmHg). Intramuscular O₂ depletes within ~8 min of occlusion (39). Therefore, the plateaued signals obtained during the last 2 min of cuff occlusion represent complete deoxygenation of Mb that are used to estimate total Mb content within the muscle. Conversion to PO₂ values were then calculated from the oxygen binding curve for Mb: PO₂ = ^fMbO₂ · Mb P₅₀/^fdeoxy-Mb, where ^fMbO₂ is the fraction of Mb that is oxygenated and P₅₀ is the O₂ pressure where 50% of the Mb binding sites are bound with O₂. The temperature-dependent Mb half saturation (P₅₀) of 3.2 mmHg was used (32).

Catheter-based vascular measurements across the working quadriceps muscle. Two catheters were placed using the Seldinger technique, one in the femoral vein and the other in the femoral artery of the right leg. Both catheters were positioned in close proximity to the inguinal ligament and femoral arterial and femoral venous blood samples were collected from these sites. To facilitate the thermodilution technique for determining Q (2), both venous and infusate temperatures were measured continuously during saline infusion (15-20 s, 110 ml/min; Harvard pump 44, Harvard Apparatus, Millis, MA) via thermistors (Edslab, TD probe 94-030-2.5F) connected to a custom-made electronic box, which was interfaced to a Power Macintosh computer using a MacLab/16 s data-acquisition system (ADInstruments, Sydney, Australia). The venous thermistor was positioned ~10 cm beyond the tip of the 8-cm catheter. Muscle Q was calculated using a heat balance equation.

Muscle O₂ transport conductance and mean capillary PO₂ calculations. With the use of the measured intracellular PO₂ values, Hill n, and P₅₀ values muscle O₂ transport conductance (DO₂) and mean capillary PO₂ (PcapO₂) were calculated as described previously (38) at maximum work rate. Briefly, a numerical integration procedure is used to determine that value of DO₂, assumed constant along the capillary, which produces the measured femoral venous PO₂, given the measured arterial PO₂. As these calculations used the measured Mb-associated PO₂, they were no longer burdened by the assumption of a low intracellular PO₂ at maximal work rate (38). An additional, and at this time unavoidable, assumption of this calculation is that the only explanation of O₂ remaining in the femoral venous blood is diffusion limitation of O₂ efflux from the muscle microcirculation. Perfusion/O₂ heterogeneity and perfusional or diffusional shunt are considered negligible. To the extent that these phenomena do not contribute O₂ to femoral venous blood, the parameter DO₂ is a conductance coefficient that expresses the diffusing capacity that would be required to achieve the measured O_2 max, assuming only diffusion limitation. This assumption cannot be avoided currently for the lack of specific means for detecting perfusion/O₂ heterogeneity and shunt in this model. Mean capillary PO₂ is the numerical average of all PO₂ values computed, equally spaced in time, along the capillary from the arterial to the venous end.

Statistical analyses. Multivariate ANOVA was used to determine differences within and between treatments across multiple work rates, with Bonferroni adjusted t-tests used to identify the differences. One-way repeated-measures ANOVA was performed to determine differences at a single work intensity (e.g., maximal effort), followed by Bonferroni adjusted t-tests used to identify the differences. When appropriate, significant differences were also identified using paired t-tests. Statistics were performed on commercially available software (SPSS, Chicago, IL). Variables were considered significantly different when the P value was 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

H and CO levels. The level of H was maintained at 11-12% throughout the H trials. The mean levels of Hb-CO in the CO_norm and CO_hyper conditions were 20.1 ± 0.4 and 20.0 ± 0.7%, respectively. Immediately before suprasystolic cuff inflation and the subsequent measurement of the maximal deoxy-Mb signal in normoxia, the mean Hb-CO level was 2.0 ± 0.1%. All subjects tolerated these procedures well, neither reporting nor showing signs of discomfort.

Deoxy-Mb signal. At rest, the intramuscular oxygen stores depleted within 6-8 min, as measured by the appearance of the deoxy-Mb signal (4). Thus the signals from this point until the blood pressure cuff release represent a steady-state deoxygenation as illustrated in Fig. 1. The developing deoxy-Mb signal intensity over time during the cuff occlusion is indicative of oxygen uptake and was not different with or without the CO load (P = 0.3; Fig. 1). The maximum deoxy-Mb signal intensity achieved during the suprasystolic cuff occlusion was also not different with and without the CO load (P = 0.2)

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Similar development of a deoxymyoglobin (Mb) signal over time with and without a 20% carboxyhemoglobin (Hb-CO) load during suprasystolic cuff occlusion (270 mmHg).

Intracellular PO₂. At the lower work intensities (<60% of normoxic work rate), a quantitative assessment of the group average intracelluar PO₂ was limited to two to four subjects in each condition. This was a consequence of relatively high intracellular PO₂ values and the average signal-to-noise of the deoxy-Mb signal achieved in this study (5:1). Hence, relatively small deoxy-Mb signals (15-20% of cuff) were lost within the noise of the measurement. Intracellular PO₂ fell somewhat variably between subjects with the continued increases in work rate from 40 to 60% of normoxic maximum (Fig. 2). Beyond 60% of normoxic maximum, all subjects produced a clear and measurable deoxy-Mb signal, and the fall in intracellular PO₂ ceased in all conditions and a plateau in intracellular PO₂ of varying lengths, dependent on work rate achieved, was clearly apparent (Fig. 2). During the hypoxic condition, intracellular PO₂ fell to a significantly lower level (3.3 ± 0.7 mmHg) (P = 0.03) at its nadir than the three other exercise trials (normoxia = 4.9 ± 0.9 mmHg; CO_norm = 5.6 ± 0.9 mmHg; CO_hyper = 5.0 ± 0.7 mmHg).

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Relationship between intracellular PO2 and single leg knee-exercise work rate in conditions of normoxia, hypoxia (H), and 20% Hb-CO and hyperoxia. For improved clarity, the 20% Hb-CO and normoxic data were omitted, as these data did not differ from the Hb-CO and hyperoxia data. *H significantly different from the 3 other conditions.

Bioenergetics. In normoxia, PCr depleted from initial resting levels linearly with increasing work rate. In H, CO_norm, and CO_hyper conditions PCr fell with a similar slope to normoxia until 40% of the maximum normoxic work rate; beyond this point the fall in PCr significantly increased such that the same level of PCr as normoxia was achieved at only 86-87% of the normoxic work rate (Fig. 3A). A similar pattern was apparent in terms of intracellular pH, with a greater rate of pH fall evident in H, CO_norm, and CO_hyper than during the normoxic exercise culminating in a similar minimum pH for normoxia, CO_norm, and CO_hyper. In contrast, during hypoxic exercise, the ultimate pH achieved was significantly lower than all other conditions (Fig. 3B) (P = 0.02). ATP remained relatively constant in all conditions until 40% of the maximum normoxic work rate, at which point ATP levels fell more rapidly in the H, CO_norm, and CO_hyper conditions compared with the normoxic ATP fall (Fig. 3C). The normoxic ATP fall at maximum exercise was equaled at 86-87% of this maximum in the H, CO_norm, and CO_hyper conditions. It is interesting to note that in one subject, the classic concept of a relatively constant ATP level, even during high-intensity exercise, was supported by moderate ATP fall (6%) even at maximal exercise. However, in the majority of cases ATP levels fell significantly to average 70% of the resting values, indicating the high metabolic demand attainable with this single leg knee-extensor exercise model. The water resonance did not move by >0.1 ppm or increase in width >2 Hz for any of the subjects, indicating that motion was limited by the leg yoke designed for the ergometer and did not degrade the signal during the study.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Relationship between phosphocreatine (PCr; A), pH (B), and ATP (C) and single leg knee-exercise work rate in conditions of normoxia, 20% Hb-CO and normoxia, 20% Hb-CO and hyperoxia, and H. *Three other conditions significantly different from normoxia, #three other conditions significantly different from H.

Q, a-v O₂ difference, and O₂ across the exercising muscle. During submaximal work rates, the a-v O₂ difference was reduced from the normoxic value (13.4 ± 0.2 ml/dl) to 11.3 ± 0.3, 10.7 ± 0.3, and 10 ± 0.2 ml/dl in H, CO_hyper, and CO_norm, respectively (P = 0.01). At maximal exercise a-v O₂ difference was reduced to an even greater extent from the normoxic value (16.2 ± 0.5 ml/dl) to 13.0 ± 0.4, 12.2 ± 0.5, and 11.4 ± 0.3 ml/dl in H, CO_hyper, and CO_norm, respectively (P = 0.005). Muscle Q was significantly elevated for a given work rate in H, CO_hyper, and CO_norm (Fig. 4B) such that the relationship between O₂ and work rate was unaltered in the four trials, but was truncated in all but the normoxic exercise (Fig. 4B). At maximal effort, muscle Q was significantly elevated above the other trials in the CO_norm condition (Fig. 4B) (P = 0.01). Mean arterial pressure increased, as expected, across work rates, but was unaltered across the four trials; thus leg vascular conductance was elevated in the H, CO_hyper, and CO_norm trials. Heart rate also increased across work rates and was significantly elevated above the normoxic values more in H (10-15 beats/min) than in the CO trials (5-10 beats/min) (P = 0.03). At maximal exercise, this difference across trials was no longer significant.

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Relationship between muscle oxygen consumption (A), muscle blood flow (B), and net CO uptake from the blood (C) and single leg knee exercise work rate in conditions of normoxia, 20% Hb-CO and normoxia, 20% Hb-CO and hyperoxia, and H. #Normoxia significantly different form the 3 other conditions, +Normoxia and hypoxia significantly different from the CO conditions.

Maximal work rate. As illustrated in Figs. 2-4, maximal work rate was significantly and equally compromised in all studies during the H, CO_hyper, and CO_norm trials compared with the normoxic exercise. Specifically, in the intravascular studies (which provided absolute work rates), the maximum work rate fell from 73 ± 6 W (100%) in the normoxia to 63 ± 5 W (87%) in H, 63 ± 5 W (86%) in CO_norm, and 63 ± 5 W (86%) in CO_hyper (P = 0.008; Fig. 4).

Net CO uptake from the vascular space. The measurement of arterial and venous Hb-CO across the leg revealed a significant CO uptake in the normoxic and hypoxic conditions that increased with increasing work intensity of the quadriceps muscles (Fig. 4C). This was in contrast to the almost zero net CO uptake in the CO_norm and CO_hyper conditions (Fig. 4C).

Muscle O₂ transport conductance and PcapO₂. The normoxic and CO_hyper had similar PcapO₂ values, both of which were higher than the other two conditions. The PcapO₂ for CO_norm was lower than normoxia and CO_hyper but was greater than the PcapO₂ in H (Fig. 5, top). As the Mb-associated PO₂ was invariant across the normoxic, CO_hyper, and CO_norm conditions, the gradient from blood to cell was smallest in the CO_norm condition due to its having the lowest PcapO₂. Interestingly, although the H condition had an even lower PcapO₂ than CO_norm, the driving PO₂ gradient from blood to cell was maintained by the reduced intracellular PO₂ recorded in H (Fig. 5, top). The analysis of PcapO₂ in relation to changes in O_2 max results in the assessment of muscle DO₂ as illustrated in Fig. 5, bottom. All conditions, except CO_hyper, produced similar maximal DO₂ values and therefore fell on a line passing through the origin.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Top: intracellular PO2 appeared to be invariant with changes in calculated mean capillary PO2 until what may be speculated to be a critical intravascular PO2 where intracellular PO2 has to fall to maintain an adequate O2 grandient. This was only achieved in H. *H lower than 3 other conditions; +normoxia + CO lower than 3 other conditions; #normoxia and hyperoxia + CO higher than 2 other conditions. Middle: altered relationship between maximal O2 consumption (O2 max) and O2 delivery (shift from solid to dashed line) illustrating the attenuated O2 extraction in the presence of CO. *Maximal O2 delivery in hypoxia significantly lower than 3 other conditions. Bottom: relationship between O2 max and mean capillary PO2 during knee-extensor exercise in normoxia, H, 20% Hb-CO normoxia, and 20% Hb-CO and hyperoxia. *Significantly reduced O2 conductance coefficient.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Directly testing the four original hypotheses, this study determined the following. 1) Hb-CO levels of 20% do not influence Mb-O₂ binding as indicated by the unaltered deoxy-Mb signal during total cuff occlusion. 2) Resting skeletal muscle metabolic rate, as measured by the rate of appearance of the deoxy-Mb signal during cuffing, and muscle O₂ during submaximal exercise was unaffected by the presence of 20% Hb-CO. However, O_2 max was significantly attenuated in the presence of CO, as seen in H. 3) Unlike H, which resulted in a similar fall in CaO₂, the CO load did not result in a significant decrement in intracellular PO₂ when compared with normoxia. 4) The 20% Hb-CO altered exercising muscle bioenergetics, as measured by pH, PCr, and ATP levels, in a very similar fashion to H, but with a better defense of intracellular pH at maximal exercise. In addition to the original hypotheses, this study documented that there was less net uptake of CO from the blood to tissue during the 20% Hb-CO trials than during those with 1-2% Hb-CO levels in normoxia and H. Thus it can be concluded that a mild CO load does reduce maximal work rate and muscle O_2 max; however, on the basis of the current measurements, this level of CO appears to have benign intracellular effects. These effects are most likely secondary to limited maximal O₂ availability, similar to those associated with H, and not due to direct metabolic CO poisoning per se. Therefore, the mild CO loads associated with smoking appear to have no direct cellular effects, and CO as a vascular research tool is well suited for use in human biomedical studies.

Deoxy-Mb signal and CO. When CO binds to Mb, the deoxy-Mb signal is attenuated and the reciprocal signal for Mb-CO increases (8). Hence, although the Mb-CO signal was not measured in the present study, an increased Mb-CO would have been reflected by a reduced deoxy-Mb signal. This did not occur despite the scenario of total occlusion for 12 min in which the intracellular PO₂ had fallen to an extremely low level, perhaps even zero (26, 39). This would be considered a perfect scenario in which to invoke CO movement into the cell, if it were going to occur, as CO moves much more readily toward hemoproteins in a reduced state. During exercise, where Mb typically desaturates to only 50% of the level attained during the cuff procedure (25, 26, 28), there is even less likelihood of CO binding Mb. This is also supported by the similarity of the deoxy-Mb signal attained during exercise in the normoxic, CO_norm, and the CO_hyper conditions (converted to PO₂ in Fig. 2).

CO and O₂ transport into the muscle cell. The CO load resulted in a clear reduction O₂ extraction (CaO₂  CvO₂) (Fig. 5, middle); this is in agreement with previous studies using CO that have proposed that the mechanism may involve Mb blockade when CO is present (12, 18). The present study affords the assessment of intravascular and intracellular processes and clearly refutes this mechanism by the observed reduction in O₂ extraction with CO present, but has no effect on functional [Mb].

Intracellular PO₂ and CO. The calculation of intracellular PO₂ or, more specifically, the ultimate plateau in intracellular PO₂ attained during high-intensity exercise provides an interesting window from which to view the effect of CO on intracellular O₂ levels (Fig. 2). Prior to 60% of maximal normoxic work rate, the intracellular PO₂ values across conditions were statistically indiscernible. However, consistent with previous reports (24, 27), beyond 60%, the hypoxic exercise bout resulted in significantly lower intracellular PO₂ values than the normoxic exercise. Somewhat surprisingly, intracellular PO₂ in the CO_norm condition did not fall to the level of hypoxic exercise, but was maintained at the normoxic level. The fall in intracellular PO₂ during CO_norm was anticipated based on the assumption of a lowered mean capillary PO₂, the expected result of the left-shifted Hb-O₂ dissociation curve in the presence of CO. It is possible that the increased O₂ delivery recorded in both CO trials was enough to offset this potential fall in intracellular PO₂. However, as illustrated in Fig. 5, top, it can be implied that although CO reduced the PcapO₂, the effect was not great enough in CO_norm to reach what may be a critical intravascular PO₂ achieved in H. Consistent with this concept of a reduced driving force of O₂ from blood to cell in the presence of CO is the matching of PcapO₂ in CO_hyper and normoxia (Fig. 5, top). As a result, CO_hyper failed to elevate intracellular PO₂ as previously reported in simple hyperoxia in similar subjects (25). Independent intracellular pH measurements, collected simultaneously, support these intracellular PO₂ observations because of the previously recognized concept that a decreased intracellular PO₂ will be accompanied by a decreased intracellular pH (28). Maximal exercise during H was the only condition in which pH was significantly reduced, suggestive of an increased lactate efflux due, perhaps in part, to an altered intracellular oxygenation state (28). Similar to previous work with H, the measurement of muscle bioenergetics during CO_norm, CO_hyper, and H revealed a more rapid fall in PCr, ATP, and pH when compared with normoxia (14). However, pH and the deoxy-Mb signal in the hypoxic exercise bout were the only MRS variables that did not reach the same ultimate end-exercise level in all conditions (Figs. 2 and 3).

CO uptake across the muscle. In addition to the lack of an effect of CO on the deoxy-Mb signal the vascular data (Fig. 4C) offer more compelling evidence that CO at the level that binds 20% of the available Hb exhibits no net movement out of the vascular space. This is in contrast to the normoxic and hypoxic conditions where Hb-CO levels were 2% due to environmental CO levels (Fig. 4), where there was measurable net CO uptake that increased with increasing exercise intensity. These latter observations are in agreement with the previous results of Clark and Coburn (6) who, aiming not to compromise O_2 max but wishing to use CO uptake and its assumed binding to Mb to determine intracellular PO₂, invoked only 2-2.5% levels of Hb-CO in human subjects. Under these conditions, CO moved from the vascular space to a greater extent both with increasing work rate and with the reduction in the fraction of inspired O₂. It is clear from the current lowest Hb-CO conditions (1-2% Hb-CO), Clark and Coburn's work (6), and the most recent MRS work of Glabe et al. (8) that, under the appropriate conditions, CO can move into muscle tissue and potentially bind Mb to some extent. However, the present study provides no evidence of net CO movement from blood to skeletal muscle or Mb binding with Hb-CO levels of 20%. The explanation for this is likely related to the interplay between several key factors: the PCO₂ in the blood, the degree to which the Hb-O₂ dissociation curve is shifted to the left, and the level to which Mb is deoxygenated. In the CO conditions, intracellular PO₂ during high-intensity work was not reduced compared with normoxic levels; the PCO₂ in the blood would be very low due to the Hb sink and the leftward shift in the O₂ dissociation curve not only increases the affinity of Hb for O₂ but also increases the affinity of Hb for CO (diminishing off-loading at the tissue). Hence, it is suggested that the combination of these variables acted to protect against CO movement from blood to tissue (at the 20% Hb-CO level), while elevated Q facilitated an unaltered oxygen flux and consumption for a given work rate (Fig. 4A).

CO and blood volume calculations. The use of CO to assess Hb mass and blood volume is not universally accepted because of potential CO loss from blood to cell. The currently employed practice of elevating Hb-CO levels by 5-15% appears appropriate; however, the present data positively support the use of the higher Hb-CO levels, as this appears to minimize CO loss into the tissue. Consequently, with the proviso that moderate levels of CO are used, the current data indicate and support prior predictions that it is not necessary to correct Hb and blood volume measurements for loss of CO to Mb (5).

ATP fall at maximal exercise. As many studies have failed to clearly perturb the level of ATP during exercise (20), it may concern some to see a significant reduction in ATP in the current data. In fact, there are a considerable number of studies that have documented a significant fall in ATP in the range of 25% as a consequence of high-intensity muscular exercise (11, 15, 16). However, even more recently, a similar ATP fall in type I muscle fibers was reported in an elegant study by Karatzaferi et al. (17), but this study also documented an ATP fall of 80% for the type II muscle fiber population. This has significant implications for the current study, in which we report an average fall in ATP as great as 38% in the CO_norm condition because there is no doubt that the muscles being studied are made up of a roughly equal number of slow- and fast-twitch fibers (31). It is important to recognize that, as in other studies, the attainment of ATP fall at maximal exercise was quite varied: subject A, ATP 98% of rest, PCr 11.3% of rest, and pH 6.68; subject B, ATP 48% of rest, PCr 1.3% of rest, and pH 6.50. Not only do these data illustrate the heterogeneity of the data, but they also refute the issue of a large signal loss (that would be evident across the complete spectrum) due to potential movement during high-intensity exercise. In subject A, it is clear that PCr demonstrated a significant fall that was not accompanied by the fall in ATP that would occur were this fall due simply to signal loss. It should also be recognized that in the current study using knee-extensor exercise, the intracellular pH falls to very low levels, which is not the case in previous studies that have revealed invariant ATP levels, even at maximal exercise (20).

Experimental limitations. Experimental limitations associated with this study include a small sample size (n = 5) that was the result of the complex, dual-site investigation, and invasiveness of the catheter-based assessments. However, it should be noted that previously both interesting and statistically significant results have been reported with a similar combination of exercise, MRS, and vascular methodologies and small sample sizes (n = 5) (25) and exercise and MRS alone (n = 4) (20).