INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MANY STUDIES (2, 3, 9, 14, 19, 22) have shown that suppression of nitric oxide (NO) formation with L-arginine analogs during in vivo conditions causes substantial vasoconstriction at rest. However, what fraction of the vasoconstriction results from the loss of NO formed in response to flow-mediated mechanisms versus chemical- or environment-mediated mechanisms is unknown. Studies of isolated arterioles (13, 15, 16) demonstrated the potential importance of flow-mediated vasodilation. In those studies, arteriolar responses during flow were compared with the resting diameter at a "no-flow" but pressurized status. From this initial degree of vascular tone, approximately one-half of the basal smooth muscle contractile state, as judged by vasodilation, can be suppressed by increased shear forces within the physiological range. What remains to be determined is the relative role of resting flow-dependent NO formation in the regulation of in vivo microvascular resistance and the physiological significance of the flow-dependent mechanism during reduced and increased blood flows.

In this study, we tested the hypothesis that regulation of large and intermediate-diameter arterioles is highly dependent on flow-mediated mechanisms to stimulate NO production. For this purpose, the microvasculature of the small intestine was chosen. Previous studies by this laboratory (5, 9) demonstrated that the intestinal microvasculature is highly dependent on NO formation. Pharmacological suppression of resting NO formation results in a large increase in vascular resistance by small arteries and large- and intermediate-diameter arterioles (5, 9). Furthermore, when only the metabolic rate of the intestinal mucosa is reduced and villus vascular resistance increases, the larger arterioles outside the influence of the intestinal mucosa also constrict (9). This constriction may reflect decreased NO generation as intestinal blood flow and shear forces are decreased by events initiated in the villus terminal vasculature. To study these mechanisms, measurement of the perivascular and tissue NO concentration ([NO]) was necessary. [NO] was measured using NO-sensitive recessed-tip microelectrodes developed by Buerk and colleagues (10) and previously used in our laboratory to study NO regulation in rats (5, 18). Our current study demonstrates that the production of NO by large- and intermediate-diameter intestinal arterioles is the dominant stimulus for NO formation at rest. Furthermore, flow-dependent mechanisms dominate NO formation over a wide range of physiologically relevant increases and decreases in flow so long as tissue oxygenation is not compromised.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal and tissue preparation. Male Sprague-Dawley rats (Harlan, Indianapolis, IN) in the weight range of 400-500 g were used in these experiments. An initial anesthetic dosage of 200 mg/kg of sodium thiopental (7) was given subcutaneously at four sites over the lower back and both thighs of the rat. One-fourth of the original dose was given intraperitoneally if supplemental anesthesia was needed. The trachea was cannulated with polyethylene tubing to ensure a patent airway, and the left femoral artery was cannulated to measure arterial blood pressure. The small intestine was prepared for observation with an established technique (4). The intestine was bathed at a rate of 4-5 ml/min with heated (37.5 ± 0.5°C) bicarbonate-buffered physiological saline (4) equilibrated with 5% oxygen, 5% carbon dioxide, balance nitrogen. The tissue support device was also heated to 37.5 ± 0.5°C. The rat's body was maintained at 37.5-38°C and heated by a water jacket support held at 34-35°C.

NO measurement. [NO] was measured using the polarographic technique with glass microelectrodes that had a tip diameter of 6-10 µm and a open recessed tip of 20-30 µm, which was preceded with gold plating for a distance of 50-100 µm. The electrode recess was filled by immersion with Nafion (Aldrich Chemical, Milwaukee, WI), which dried to a thin coating <1 µm in thickness. The techniques for producing and calibrating NO-sensitive microelectrodes are based on the procedures developed by Buerk and colleagues (10) and our experience with the microelectrodes (5, 18). The NO-sensitive microelectrodes were polarized at +0.8 V relative to a World Precision Instruments carbon fiber reference electrode (Sarasota, FL). Currents in the picoampere range were measured with a Keithley model 610C electrometer (Cleveland, OH). The typical NO microelectrode had currents in the range of 8-12 pA in nitrogen-equilibrated saline and for a 1,000 nM increase in NO would generate an additional current of 1-2 pA. This translates to a 300- to 600-mV increase in voltage output on the Keithley electrometer per 1,000 nM of NO. A Diamond General gas tonometer system (Ann Arbor, MI) was used to establish [NO] of 0, ~600, and ~1,200 nM, based on the composition of the NO-N₂ precision calibration gases (Matheson, Joliet, IL) in saline at 37.5°C. Each electrode was found to have a linear current-[NO] relationship. The working resolution of the microelectrodes, taking into account noise and current drift, was typically 10 nM at an average resting vessel wall [NO] of 300-325 nM. After each experiment, NO sensitivity was retested in the calibration system and generally found to have changed less than ±10% unless the micropipette tip was fouled with tissue residue or damaged. In general, the NO-sensitive microelectrodes were usable over three to four experiments.

Flow velocity measurements. Red blood cell velocity was measured with the dual-slit cross-correlation technique using an Instrumentation for Physiology and Medicine model 102 system (San Diego, CA). The system was calibrated for linear velocities by measuring the velocity of dried red blood cells on a clear disk rotated by a 1-rpm synchronous motor. The disk radius at specific points was known and used to calculate circumferential velocity. When the centerline flow velocity was measured in vivo, this velocity was divided by 1.6 to estimate the average red blood cell velocity for microvessels with diameters >10 µm (1). The site of vessel velocity and vessel diameter measurements was ~50 µm upstream from the site for [NO] measurement. Vessel diameter was measured from the digitized freeze-frame video image (CCD Camera 200R, Videoscope International, Washington, DC) using the virtual caliper function of Image 1 image analysis software (Universal Imaging, West Chester, PA). The image analysis system was calibrated in the X and Y directions using the image of a stage micrometer marked in 10- and 100-µm units. Once both the mean velocity and inner diameter were known, shear rate was calculated using 4 × mean velocity/inner radius to yield s¹ which is equivalent to the formal equation of (4 × flow)/(3.14 × radius³).

Protocols. A diagrammatic scheme of the intestinal microvessels studied is shown in Fig. 1. The larger arterioles [first-order arterioles (1A)] ran radially about halfway around the bowel wall. These 1A originated from the marginal arteries along the mesenteric aspect of the bowel wall. The intermediate-sized arterioles [second-order arterioles (2A)] chosen were the largest interconnecting arterioles between adjacent 1A. These 2A were typically the first two branches of the 1A. The larger 2A were chosen because they perfused a great number of terminal arterioles, shown as small spur branches in Fig. 1. The terminal arterioles supply the deep submucosa, villi, and overlying muscle layers. The directions of flow for each vessel type during each protocol and the measurement locations for flow velocity, inner arteriolar diameter, and [NO] are shown in Fig. 1.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Top: the normal pattern of microvascular perfusion for adjacent large arterioles (1A), which are extensively interconnected by the intermediate-diameter arterioles (2A). The directions of blood flow are shown by arrows and the site for nitric oxide (NO) measurements with NO-sensitive microelectrodes by [NO] in all panels. As shown in the middle panel (collateral perfusion), the adjacent 1A was occluded to increase blood flow in a 1A and 2A without altering their own tissue perfusion area. The bottom panel (upstream occlusion) shows that to reduce flow in the same 1A and 2A, the upstream 1A was nontraumatically occluded.

Decreased metabolic rate. As explained in the introduction, isosmotic replacement of sodium chloride with mannitol over the villi lowers the resting intestinal metabolic rate and blood flow over a period of 15-20 min (9). The perfusion pattern did not change during periods of reduced metabolic rate and was identical to the normal pattern illustrated in the top panel of Fig. 1. Furthermore, percent saturation of hemoglobin measurements during periods of reduced blood flow indicated that arteriolar oxygen tensions change very little in the 1A and 2A (9). The outer surface of the bowel wall, where the measurements of 1A and 2A were made, was constantly bathed (4-5 ml/min) with standard physiological solution. However, some of the mannitol solution from the mucosal surface perfusion (1 ml/min) did mix with the physiological saline. This is not a problem. If only the exterior of the bowel wall was perfused with the mannitol-bicarbonate solution, the resting intestinal blood flow was not affected (9).

Collateral perfusion. The goal of this protocol was to increase resting blood flow through the arterioles being studied as they become collateral vessels to perfuse vessels in adjacent areas of the bowel wall. This was accomplished by nontraumatically occluding the adjacent 1A (see Fig. 1, middle). In most cases, the NO microelectrode was left in place during the occlusion of these very distal 1A. The 1A and 2A providing collateral perfusion were in an area of tissue that was always normally perfused. Furthermore, venous drainage from the flow-deprived tissue was through its normal venous outflow channel beside the occluded 1A. Therefore, the primary stimuli to the 1A and 2A providing collateral flow should be some aspect of flow-mediated vasodilation rather than vasoactive materials in venous blood from the flow-deficient areas of tissue. The blood flow increased in the vessels studied, and the parenchymal tissues they naturally perfused received normal blood flow. Therefore, the oxygen tension within and around these vessels was not compromised by this protocol.

Upstream occlusion. To reduce the flow in the 1A being studied, the upstream small artery was occluded for several minutes with a blunt glass micropipette. The NO microelectrode was kept in place during the brief occlusion. This pattern of perfusion is shown in the bottom panel of Fig. 1. For the 2A, flow direction reversed in the vessels near the mesenteric wall and provided some flow to the 1A. The distal 2A along the occluded 1A had variable flow directions from moment to moment, but in general, flow was in the normal direction.

Statistical analysis. All data are expressed as means ± SE, and statistical analysis was carried out using Sigma Stat (Jandel Scientific, San Rafael, CA). To evaluate differences in each variable, comparisons were made with repeated-measures one-way analysis of variance relative to the initial control condition.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Microelectrode evaluation. Although the NO-sensitive microelectrodes demonstrated a linear relationship between [NO] and current generated in a gas tonometer, whether the tissue physical and chemical environment offsets the relationship or induces artifacts has not been established. The oxygen in living tissue presented a formidable problem to calibration of a microelectrode using externally supplied NO during both in vitro and in vivo conditions. To circumvent this problem, the blood flow was stopped, the tissue was frozen in the tissue support device to stop all metabolism, and then the tissue was rewarmed to body temperature. This approach was used in the classic studies of oxygen-sensitive microelectrodes by Whalen et al. (25) to provide a "tissue" calibration of polarographic microelectrodes. This same procedure allowed us to work in an oxygen-free environment. Furthermore, in an oxygen-free environment, any chemicals in the freeze-damaged cells could interact with the microelectrode to cause an electrical current error signal. However, as shown in Fig. 2, the currents measured by the same microelectrode in the nitrogen-equilibrated bath and in the tissue at a depth of 20-25 µm were identical for practical purposes. An arteriole in the submucosal tissue was used as a tissue target to provide consistent-depth penetrations. Starting from a nitrogen-equilibrated state, ~400 parts per million NO gas in nitrogen was added in stages to the bathing medium reservoir to progressively increase [NO]. The calculated maximum concentration achievable would be ~600 nM at 37.5°C. Measurements of bath and tissue were taken once the bath [NO] stabilized for at least 3-5 min after addition of NO to the reservoir to allow equilibration of the tissue to the new [NO]. Figure 2 lists the range of lowest (0 nM NO during nitrogen equilibration) to highest [NO] achieved for each microelectrode based on the independent calibration of each microelectrode in a precision gas tonometer. Over the range of [NO] tested, the microelectrodes recorded almost the same current in the tissue as in the bath for a given [NO], as demonstrated by tissue-bath current slopes >0.9 and correlation coefficients >0.9. This is good evidence that the physical environment of the tissue and any biochemicals that survived freezing had little effect on the microelectrode's ability to faithfully monitor the [NO] in the physical environment of the tissue. However, although the microelectrodes were stable during careful tissue penetration, obvious bending of the microelectrode tip caused large current offsets.

View larger version (28K): [in this window] [in a new window]   Fig. 2.   The current generated by the NO-sensitive microelectrodes was measured ~10 µm above the surface of the tissue and on a small arteriole in the submucosa used as a depth marker. The tissue was dead after being frozen and rewarmed to body temperature. The bathing medium was initially equilibrated with N2, and then NO gas was added in stages to generate the various data points shown. Paired bath and tissue measurements by the same microelectrode were made for each stage of NO equilibration. The range of NO concentration ([NO]) predicted from the independent calibration of each microelectrode is given, as are the slope and regression coefficient (r2) of the tissue-bath current relationship.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Regression analysis of recessed-tip gold-plated microelectrode (current in pA) to World Precision Instruments (WPI) macroelectrode ([NO] in µM). A calibrated macroelectrode was used to verify [NO] as current generated by a microelectrode was measured. The 2 calibrations (run 1, run 2) were separated by 1 h to demonstrate that NO-sensitive electrodes are stable over time.

View larger version (37K): [in this window] [in a new window]   Fig. 4.   A: effects of increased blood flow on arteriolar [NO] during the collateral flow protocol. The data are plotted at 3-s intervals for 2,500 s (>30-min period). The artifactual spikes are caused by tissue movement bending the tip of the microelectrode. The straight solid lines represent the mean [NO] (artifacts were not calculated into the mean) during each phase of an individual protocol. At 1 point, the bowel moved the microelectrode from the arteriolar wall and the microelectrode was repositioned to the original periarteriolar location. The arteriolar diameter and flow velocity are listed for resting conditions and collateral perfusion, as defined in Fig. 1. Note that the [NO] was elevated for ~1,300 s (>20 min), during collateral perfusion. After release of occlusion, [NO] spontaneously returned to the control concentration range. B: NO current tracing beside a small arteriole whose distal region was occluded to lower the blood flow. The decrease in flow velocity was associated with a sustained decrease in [NO] and ~10% constriction. On release of the occlusion, the vessel resumed a normal [NO]. The beginning and end of the current trace demonstrate the submucosal tissue [NO] as the microelectrode tip was moved through the bowel wall in route to or from the vessel wall.

Vessel characteristics. The resting inner diameter, flow velocity, shear rate, and periarteriolar [NO] are presented in Table 1. In each animal studied, the 2A was a branch of the 1A studied. As mentioned in METHODS, the 2A selected were the first or second 2A branch of a given 1A. These arterioles were typically larger in diameter than the more distal 2A branches and were the primary collateral connectors between adjacent 1A (Fig. 1). Despite major differences in vessel diameters and flow-related variables between 1A and 2A, the average resting [NO] for all 1A was 337 ± 20 (74 vessels, 29 rats) and 318 ± 21 (35 vessels, 21 rats) nM for the 2A.

Reduced metabolic rate. Isosmotic replacement of luminal sodium chloride with mannitol-bicarbonate solution is known to both reduce the intestinal metabolic rate and increase the resistance of the terminal and larger arterioles (9). As shown in Fig. 5, blood flow was reduced to 60-65% of control, which was consistent with prior studies in which reduced intestinal metabolism resulted in significantly increased resistance of small arteries and both 1A and 2A (9). The reductions in shear rate and flow velocity in the 1A and 2A were associated with ~30% and ~20% decrease in periarteriolar [NO], respectively. The 20-30% reduction in periarteriolar [NO], as well as arteriolar constriction, required ~15 min to reach steady state. The time delay was caused by the intentionally slow changes in mucosal luminal fluid performed to avoid dislodging micropipettes and stretching the tissue. Once steady-state reductions in hemodynamic parameters and [NO] were achieved, they persisted for at least an hour, which was the longest time used for testing. Restoration of sodium chloride in the luminal media over the villi resulted in complete recovery of all hemodynamic and [NO] parameters within ~15 min.

View larger version (47K): [in this window] [in a new window]   Fig. 5.   The metabolic rate of the bowel was decreased by limiting the absorption-secretion cycle of sodium chloride in the villi. Sodium chloride in the bath over just the mucosal surface was replaced isosmotically with mannitol. This substitution lowers the oxidative metabolism by ~40-50% (8). As shown in the figure, blood flow is decreased to 60-65% of control. The pattern of flow direction is the same as in the normal perfusion shown in Fig. 1. The microvascular responses caused a decreased flow velocity, shear rate, [NO], and inner diameter for 1A and 2A. The data set is based on 10 vessels for 1A and 6 vessels for 2A in the same 6 rats.

Collateral perfusion studies. The data shown in Figs. 6 and 7 are for arterioles that provided collateral blood flow to an adjacent tissue area whose inflow 1A had been occluded. This pattern of perfusion is shown in the middle panel of Fig. 1. Blood flow increased in the 1A and 2A because they must perfuse their own tissue regions plus a large amount of an adjacent 1A-2A perfusion field (Fig. 1).

View larger version (27K): [in this window] [in a new window]   Fig. 6.   The changes in [NO] (A) and arteriolar diameter (B) in response to increased blood flow caused by the collateral flow protocol were measured in the same large arterioles before and after the arteriole was embolized with an intravascular oxygen bubble. The data, based on 5 vessels from 3 rats, demonstrate that increased flow caused both increased [NO] and diameter, both of which were negated after embolization. Furthermore, embolization decreased the resting [NO] by ~40%.

View larger version (30K): [in this window] [in a new window]   Fig. 7.   As shown in the collateral perfusion panel of Fig. 1, the left-side 1A and 2A provided collateral blood flow to the dependent tissues no longer perfused by the occluded 1A. Blood flow was increased in the collateral 1A and 2A as downstream vasodilation occurred in the underperfused tissues. Consequently, flow velocity increased, which caused a concomitant increase in local shear rate. This in turn increased NO formation and caused vasodilation. The data set is based on 19 1A from 14 rats and 13 2A from 9 of the 14 rats in which 1A were studied.

Upstream occlusion studies. The purpose of these studies was to both reduce flow velocity in the 1A and force a decrease in oxygen availability to the vessel and nearby tissue (Fig. 8). We used these studies to demonstrate that [NO] will increase despite a reduction in flow velocity and shear rate if oxygen availability is limited. After initial control measurements were made, the origin of the 1A was occluded just outside the intestinal wall. The occlusion was ~6,000-7,000 µm from the 1A observation area. This perfusion pattern is shown in the bottom panel of Fig. 1. The retrograde perfusion of 2A provided ~50% of the normal flow in the 1A whose origin was occluded. Although we did not measure the oxygen tension in the 1A, the color of blood darkened because of oxygen desaturation as it flowed retrograde through the 2A to the 1A. Despite the large decrease in flow velocity and shear rate, [NO] in the 1A increased to 127 ± 5% of control. In addition, the 1A was able to maintain its resting diameter despite a large reduction in perfusion. In our prior experience with occlusion of 1A, microvascular pressure was typically reduced by about one-half (17).

View larger version (28K): [in this window] [in a new window]   Fig. 8.   The inflow of the 1A studied for Fig. 7 was occluded as shown in the upstream occlusion panel of Fig. 1. The direction of blood flow was reversed in the left-side 2A of Fig. 1. The flow direction in 1A was normal and primarily derived from retrograde flow through 2A near the origin of the 1A. Despite the reduction in intravascular pressure caused by the decline in flow, the 1A increased their resting diameters. [NO] increased in 1A, although flow velocity and shear rate were decreased. The data set is based on 16 of the 19 1A studied in Fig. 7.

Velocity and shear rate effects on [NO]. The relationship of [NO] to percentage of control red blood cell velocity and shear rate in all of the 1A and 2A is shown in Fig. 9. These data are based on the results obtained during the collateral and metabolic protocols (Figs. 5 and 7) for each of the vessels studied. Normalization of the data was used because, in reviewing responses by individual vessels, the [NO] changed in approximate proportion to the flow velocity and shear rate. However, as one might expect, there was a wide range of velocities and shear rates even within the same vessel as daughter branches were given off and the vessel gradually tapered to smaller diameters over distance. The slopes of the relationships for velocity and shear rate of 1A [NO] were 0.85 and 0.75, respectively, and for 2A [NO] were 0.59 and 0.52, respectively, for velocity and shear rate. These data predict that NO formation in larger arterioles is more dependent on flow-related mechanisms than that in intermediate diameter arterioles.

View larger version (32K): [in this window] [in a new window]   Fig. 9.   The relationships of percentage of control red blood cell velocity and shear rate to the percentage of control [NO] are shown for vessels studied for Figs. 4 and 6. For about two-thirds of the vessels studied, the resting, increased, and decreased flow and shear rate data were available; for the remainder, only the responses to increased or decreased flow and shear rate were available. Flow velocity and shear rate changes were about equally correlated to changes in [NO]. For the 1A, the relative change in [NO] for a given change in velocity or shear rate was ~1.3 times greater than for the 2A. The equations for the linear regression lines are as follows. Velocity: 1A, 0.85X + 20.8, r2 = 0.53, P < 0.05; 2A, 0.52X + 50.7, r2 = 0.58. Shear rate: 1A, 0.75X +33.6, r2 = 0.53, P < 0.05; 2A, 0.59X + 45.2, r2 = 0.56, P < 0.05.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Resting periarteriolar [NO]. The resting periarteriolar [NO] was of interest in the context of what fraction of maximum activity of vascular smooth guanylate cyclase might be generated. Stone and Marletta (24) reported the in vitro [NO] for half-maximal activation of guanylate cyclase in vascular smooth muscle to be 250 nM [dissociation constant (K_d)] at 10°C. The physiological K_d at 37-38°C should be a somewhat higher concentration, but just how much higher remains to be determined. However, within the limitations of current knowledge, the [NO] we measured at the outer surface of the arteriolar wall should generate approximately half-maximum cGMP. The resting [NO] can be increased by ~100%. Supraphysiological 1 mM L-arginine and 360 mosM sodium chloride (5) are both capable of nearly doubling the [NO], as is also the case during the highest possible physiological increases in flow velocity and shear rate during collateral perfusion events (Fig. 9). Therefore, we believe that the resting [NO] was about one-half the physiological maximum concentration that in vivo intestinal arterioles can generate for sustained periods. We have recorded >200% of control [NO] but only after vessel wall injury by penetration of the endothelial layer or severe hypoxia, and these increases are invariably transient.

Effect of flow-mediated stimuli on [NO]. To confirm that endothelial cells were the source of NO during flow perturbations, [NO] at rest and at elevated blood flow was measured before and after localized oxygen embolization of large arterioles. Nase and Boegehold (20) showed that localized gas embolization virtually eliminated endothelium-dependent vasodilation to locally applied acetylcholine. On the basis of their observations, we tested whether flow-dependent increases in [NO] and associated vasodilation were eliminated by oxygen bubble emobolization. The localized endothelial damage immediately reduced the resting [NO] by ~40%, as shown in Fig. 6. In a prior study of intestinal arterioles (5), we found that topical N^G-nitro-L-arginine applied topically to a single large arteriole in the intestinal wall caused a comparable decrease in resting [NO]. When blood flow was increased after embolization, the vessels did not increase either [NO] or inner diameter. However, the other arterioles in the area continued to respond with appropriate dilation to increased blood flow. These various observations indicated that vasodilation to increased flow velocity was highly dependent on an intact endothelium and was associated with a relatively large increase in [NO].

Effect of hypoxia on NO formation. The results and conclusions discussed thus far are for conditions in which reduced tissue oxygenation was not an issue during variations in blood flow. Our interest in tissue oxygenation and NO formation was stimulated by Busse and colleagues' (11) and Pohl and Busse's (21) finding that hypoxia is a stimulus for NO-mediated vasodilation. The data in Fig. 8 represent conditions when flow into a 1A was reduced by complete upstream occlusion of its origin at the bowel wall-mesenteric border. Thereafter, the perfusion of the 1A was ~50% of normal through retrograde perfusion of the larger 2A near the mesenteric border of the bowel wall. Although we did not measure the oxygen tension in the vessel wall of 1A, the color of the blood darkened, indicating oxygen desaturation. Despite the ~50% reduction of the flow velocity and shear rate, [NO] increased an average of 27% (Fig. 8) and the arterioles dilated ~10%. In contrast, for a similar magnitude reduction in flow velocity and shear rate when the metabolic rate was decreased, both a 20-30% reduction in [NO] and 10% arteriolar constriction occurred (Fig. 5). The net difference in [NO] between the two "low-flow" states was 50-60% of control, which corresponded to a [NO] difference of 175-200 nM. This was a large difference in [NO]. For example, the [NO] increased ~40-50% or ~120-175 nM during the transition from rest to maximal collateral blood flow (Fig. 9). In effect, compromising tissue oxygenation completely changes the regulation of NO formation from a mechanism dominated by flow-dependent forces to a potentially very complex set of mechanisms related to the diverse effects of hypoxia on vascular and parenchymal tissue. As a consequence, understanding the regulation of NO through flow-dependent and non-flow-dependent mechanisms during in vivo studies depends critically on whether tissue oxygenation is compromised by a given perturbation.