Shear Stress Induces ATP-Independent Transient Nitric Oxide Release From Vascular Endothelial Cells, Measured Directly With a Porphyrinic Microsensor
Abstract Shear stress causes the vascular endothelium to release nitric oxide (NO), which is an important regulator of vascular tone. However, direct measurement of NO release after the imposition of laminar flow has not been previously accomplished because of chemical (oxidative degradation) and physical (diffusion, convection, and washout) complications. Consequently, the mechanism, time course, kinetics, and Ca2+ dependence of NO release due to shear stress remain incompletely understood. In this study, we characterized these parameters by using fura 2 fluorescence and a polymeric porphyrin/Nafion-coated carbon fiber microsensor (detection limit, 5 nmol/L; response time, 1 millisecond) to directly measure changes in [Ca2+]i and NO release due to shear stress or agonist (ATP or brominated Ca2+ ionophore [Br-A23187]) from bovine aortic endothelial cells. The cells were grown to confluence on glass coverslips, loaded with fura 2-AM, and mounted in a parallel-plate flow chamber (volume, 25 μL). The microsensor was positioned ≈100 μm above the cells with its long axis parallel to the direction of flow. Laminar flow of perfusate was maintained from 0.04 to 1.90 mL/min, which produced shear stresses of 0.2 to 10 dyne/cm2. Shear stress caused transient NO release 3 to 5 seconds after the initiation of flow and 1 to 3 seconds after the rise in [Ca2+]i, which reached a plateau after 35 to 70 seconds. Although the amount (peak rate) of NO release increased as a function of the shear stress (0.08 to 3.80 pmol/s), because of the concomitant increase in the flow rate, the peak NO concentration (133±9 nmol/L) remained constant. Maintenance of flow resulted in additional transient NO release, with peak-to-peak intervals of 15.5±2.5 minutes. During this 13- to 18-minute period, when the cells were unresponsive to shear stress, exogenous ATP (10 μmol/L) or Br-A23187 (10 μmol/L) evoked NO release. Prior incubation of the cells with exogenous NO or the removal and EGTA (100 μmol/L) chelation of extracellular Ca2+ blocked shear stress but not ATP-dependent NO release. The kinetics of shear stress–induced NO release (2.23±0.07 nmol/L per second) closely resembled the kinetics of Ca2+ flux but differed markedly from the kinetics of ATP-induced NO release (5.64±0.32 nmol/L per second). These data argue that shear stress causes a Ca2+-mediated ATP-independent transient release of NO, where the peak rate of release but not the peak concentration depends on the level of shear stress. The transient nature of this response may be due to NO-induced inhibition of Ca2+ influx via a mechanism yet to be determined.
NO is a potent vasodilator released from endothelial cells during exposure to laminar flow.1 2 It is well documented that NO or a close derivative3 modulates vascular tone.4 5 NO also inhibits platelet aggregation6 and adhesion7 to endothelial cells. Deficiencies in NO release have been implicated in a number of serious cardiovascular conditions, including essential hypertension,8 9 coronary artery disease,10 11 and chronic heart failure.12
The hemodynamic forces resulting from flow include two components: (1) shear stress, the tangential frictional force produced when blood flows over the endothelial cell surface, and (2) pressure stretch, acting perpendicular to the vascular wall. Recent evidence suggests that the shear stress acting on the endothelium is responsible for flow-induced NO release.13 14 In large arteries, the average wall shear stress is ≈1 to 20 dyne/cm2. At curves and bifurcations, peak wall shear stress may be as high as 100 dyne/cm2. Immediate (millisecond to second) responses to shear stress include increases in ionic conductances15 16 and intracellular levels of Ca2+17 18 and IP3,19 20 the order of occurrence being unclear. Delayed (minute to hour) responses to chronic shear stress include altered gene expression,21 mRNA levels,22 and cell orientation.23
NO is produced from l-arginine via NOS. Endothelial cells produce a constitutive form of this enzyme (cNOS) that is sensitive to Ca2+ as well as an inducible form (iNOS) that is Ca2+ insensitive. Since iNOS expression requires exposure to inducing agents such as cytokines24 or bacterial lipopolysaccharides,25 shear stress–induced NO release is believed to occur via Ca2+-dependent cNOS. In addition, two putative mechanisms have been proposed to account for shear stress elevations in cytosolic Ca2+: (1) activation of mechanosensitive nonselective cation channels that result in Ca2+ influx across the plasmalemma15 and (2) flow-induced extravasation of intracellular ATP that activates surface P2-purinoceptors, resulting in formation of IP3 and release of Ca2+ from IP3-sensitive pools.26 27
It is also controversial whether continuous shear stress causes transient or continuous (basal) NO release. Kuchan and Frangos28 used the Greiss method to measure NO oxidation products NO2− and NO3− released from endothelial cells exposed to laminar flow and reported biphasic NO release. A large initial rise in NO2−/NO3−, corresponding to the onset of flow, was followed by a small sustained increase in NO2−/NO3−. Since shear stress causes transient elevations in [Ca2+]i,17 18 sustained release of NO would have to occur during the intervening periods. Furthermore, since Ca2+-independent iNOS was reportedly absent in these cells and studies with isolated Ca2+-dependent cNOS have demonstrated that resting Ca2+ levels are below the activation threshold for this enzyme,29 cNOS would have to produce NO in a Ca2+-independent fashion. However, since NO is rapidly oxidized to the more stable NO2−/NO3− anions, only the accumulation of NO2−/NO3− was measured. Therefore, this putative basal elevation of NO2−/NO3− could have resulted from the transient release of NO, a determination that can only be made by measuring real-time NO flux.
We tested the hypotheses that (1) shear stress evokes NO release via a Ca2+-dependent, but ATP-independent, mechanism and (2) during continuous shear stress, NO is released in a transient rather than a basal manner. To accomplish these ends, it was necessary to measure changes in [Ca2+]i and NO flux under well-defined flow conditions in order to understand the regulation and the kinetics of NO release from endothelial cells. We used fura 2 fluorescence and an NO-selective porphyrinic microsensor30 to provide the first real-time measurements of cytosolic Ca2+ elevations and NO release during laminar flow, determined the magnitude and kinetics of NO release as a function of shear stress, evaluated the effects of mass transfer on the measurement of NO release, and assessed the mechanism of transient NO release during continuous shear stress.
Materials and Methods
BAECs were used exclusively in the study outlined below. Since iNOS is induced by lipopolysaccharides from bacterial cell walls,25 sterile techniques were rigorously observed, and pyrogen-free water was used exclusively. Cells were harvested and cultured as previously described,31 but antimycotic agents were omitted. In brief, freshly excised aortas were collected from a slaughterhouse and cleaned in HBSS. All intercostal arteries and one end of the aorta were occluded, the cavity was filled with 0.1% collagenase (type I, Sigma Chemical Co) in medium 199 (Sigma), and the other end was clamped. After an 8- to 12-minute incubation at 37°C, dissociated cells were collected, washed by centrifugation at 1000g for 10 minutes at 4°C, and resuspended in DMEM containing 10% FBS (HyClone Laboratories) and 1% antibiotic (GIBCO). Cells were grown in collagen-coated flasks or on No. 2 glass coverslips (22×22 mm) without substratum. Nonadhering cells were poured off, and the primary isolates were incubated in DMEM with 10% FBS and 1% antibiotic at 37°C in 10% CO2 (pH 7.4). The medium was changed every 3 days. Cultures demonstrating smooth muscle contamination were discarded.31 Cells reached confluence in 3 to 7 days and were subcultured by trypsin dissociation. Subsequent cultures were grown to confluence under the same conditions as primary isolates. In general, cells from primary through 8 passages were used, although cells have been studied up to 12 passages with no apparent difference in response.
Cell-permeant fura 2-AM (acetoxymethyl ester) was used in these experiments. Fura 2-AM readily crosses the cell membrane and is cleaved by intracellular esterases trapping the dye in the cytosol. To minimize incomplete deesterification of fura 2 and buffering of intracellular Ca2+, cells were loaded for 40 minutes at 22°C with 0.5 μmol/L dye.
We used a ratiometric method of in situ calibration using an ionophore. Fura 2 fluorescence was calibrated by perfusing the cells with media containing 5 μmol/L ionomycin and 2.0 mmol/L Ca2+ to obtain a saturated [Ca2+]i signal or 5 mmol/L EGTA and 2 μmol/L ionomycin to obtain a zero [Ca2+]i fluorescence signal. The background fluorescence was measured on cell-free parts of the coverslips, and the autofluorescence was measured in cells not loaded with fura 2 in control experiments. After subtraction of background fluorescence and autofluorescence, the [Ca2+]i was calculated from the ratio of the 340- and 380-nm signals. To calculate [Ca2+]i, the following equation32 was used, relating the measured fluorescence ratio (R, the ratio of Ca2+-bound fura 2 to Ca2+-free fura 2) to [Ca2+]i: where Kd (224 nmol/L) is the dissociation constant for the fura 2–Ca2+ complex, R is the ratio of relative fluorescence, Rmax is the maximum ratio measured in the presence of saturating free Ca2+ (2 mmol/L), and Rmin is the minimum ratio determined at 0 mmol/L Ca2+. Sf2/Sb2 is the fluorescence ratio measured at an excitation wavelength of 380 nm in 0 versus 2 mmol/L Ca2+.
Fluorescence excitation was provided by a PTI Deltascan system (Photon Technology International) with a 75-W xenon arc lamp, dual monochromators, a rotating optical chopper, electronic shutter, and compatible hardware (Zenith 386 computer) and software (oscar, version 2.1, PTI). The alternating fluorescence illumination (340/380 nm) was transmitted through a randomized bifurcated light guide, reflected off a 400-nm long (wavelength)–pass dichroic mirror, and focused onto the cells through a Nikon 40× fluor/phase oil immersion objective (numerical aperture, 1.30). Fluorescence emission was collected by a PMT (Hamamatsu R928-07) at a bandpass (filter) wavelength of 510±10 nm. PMT signals were digitized, sent to a monitor for real-time imaging, and then stored on hard disk for later retrieval and analysis.
Carbon strands (one to seven fibers, 7-μm diameter each; Amoco, Inc) were inserted into glass capillary tubes beveled 45° on one end (Fig 1⇓). A copper wire was connected to the carbon fiber with silver epoxy at the unbeveled end and sealed with wax at the 45° end. While the wax was molten, the protruding fiber was bent at 45° until the wax solidified. The protruding fiber was then cut to the desired length (4 mm).
TMHPPNi (monomeric porphyrin) was synthesized according to a procedure previously described.33 TMHPPNi was dissolved in 0.1N NaOH and deposited, as a polymeric film, on the carbon fiber by using a multiple potential scanning cyclic voltameter (−0.2 to +1.0 V, model 264A polarographic analyzer, EG&G, Princeton Applied Research). Polymeric TMHPPNi catalyzes the oxidation of NO to NO+. After the porphyrin dried (2 to 3 hours.), the cation exchanger Nafion (Sigma Chemical Co) was applied by dipping the electrode into a 1% solution made up in ethanol. The negative charges of the SO3− functional group of Nafion keeps NO2− (nitrite) and NO3− (nitrate) anions from gaining access to the active porphyrin surface, preventing overestimation of the NO response. The Nafion-coated electrodes were characterized by differential pulse voltametry to determine the redox potential of the oxidation of NO to NO+. Amperometry, performed at a constant potential 50 mV more positive than the redox potential, was used to determine the NO concentration. High purity (>99.99%) NO standards (see below) were prepared to accurately calibrate the electrodes. The currents generated by the oxidation of NO at the porphyrinic interface (0.5 to 1.5 nA/cm2 for 1 μmol/L NO under static conditions) were amplified and converted to voltages (model 264-A Polarographic analyzer) and displayed on a four-channel chart recorder (Gould series 2400, Gould Inc), with fura 2 fluorescence tracings on a monitor for simultaneous real-time display and stored on hard disk for later retrieval and analysis.
To accurately calibrate porphyrinic NO microsensors, high-purity NO standards were prepared daily. NO standards are important for the initial calibration of microsensors and for verification of their continued accuracy. Commercially available NO gas lacks the purity necessary for accurate calibration (>99.99%). Pure standards were made by reacting NaNO2 (sodium nitrite) with 6 mol/L H2SO4. The evolved gaseous NO was passed through 4 mol/L NaOH to remove any higher oxides of nitrogen and then bubbled through an airtight degassed sample of the solution in which biological NO measurements were made (phosphate-buffered Ringer’s solution). The concentration of saturated solutions prepared in this fashion contained 2.0 mmol/L NO at 25°C, as determined by chronoamperometry.34 Calibration curves under static conditions were generated by adding aliquots (0.1 to 4 μL), withdrawn with a Hamilton syringe from a septum sealed bottle containing the NO standard, to 5 mL of Ringer’s solution housing a microsensor. Alternatively, NO stock dilutions (40 to 1600 nmol/L) were prepared to calibrate a microsensor under laminar flow conditions (0.04 to 1.90 mL/min). The diluted NO solutions, made up in degassed Ringer’s buffer, were passed through the parallel-plate chamber with a porphyrinic electrode in place but without endothelial cells.
NO Measurements and Flow Chamber
Cultured cells grown to confluence on No. 2 glass coverslips (22×22 mm) were placed in the bottom half of a parallel-plate flow chamber (Duke University Surgical Instrument Shop) (Fig 2⇓). The coverslip bearing the cells formed the floor of a 0.0125×1×2-cm3 channel (height×width×length) whose dimensions were set by a polytetrafluoroethylene (Teflon) spacer and whose volume was 25 μL. Two ports on the top half of the chamber allowed access for the working electrode (0.025-mm diameter×4-mm-long porphyrinic microsensor) and a counter electrode (0.1-mm diameter×4-mm-long platinum wire). After assembly, the NO microsensor lay in the center of the chamber flush against the upper surface with its long axis aligned parallel to the direction of flow. The counter electrode was recessed into the top half of the chamber, parallel and in close proximity to the working electrode. A third port for injecting drugs was located upstream from the electrodes. For simultaneous NO and [Ca2+]i measurements, cells were loaded with fura 2-AM before mounting in the chamber. To prevent mechanically induced Ca2+ transients and concomitant NO release, coverslips were carefully handled during placement in the chamber. The chamber halves were clamped together and placed on the modified temperature-regulated stage of an inverted microscope (Nikon Diaphot); experiments were conducted at 22°C.
The accurate calculation of the wall shear stress requires that the flow in the channel be one dimensional and fully developed. One-dimensional flow can be guaranteed by choosing an appropriate width-to-height ratio. Based on the solution for two-dimensional flow in a rectangular channel,35 a width-to-height ratio of 20:1 produces an average wall shear stress within 1% of the value for one-dimensional flow. In the present study, the width-to-height ratio was 80:1.
For the entrance region, flow developed from the inlet profile to one-dimensional flow. For a rectangular channel, the entrance length equals the inlet profile of one-dimensional flow, or 0.08 Re36 [Re=2 Q/v(w+h), where Q is flow, v is the kinematic viscosity, w is the width, and h is the channel height]. For low Re, the entrance length does not approach zero but rather an asymptotic value of ≈2.6 h.37 For the conditions used in the present study, the Re values range from 0.074 to 1.86. The corresponding entrance lengths range from 0.0326 to 0.0355 cm. These entrance lengths are <2% of the length of the channel. Based on this analysis, flow is essentially one dimensional and fully developed. Perfusate was pumped (Gilson Minipuls 2) in one end and drained from the other without recirculation. The perfusate contained (mmol/L) NaCl 150, KCl 5.4, MgCl2 1, CaCl2 1.8, glucose 10, and phosphate 10 (pH 7.4).
The cells were exposed to laminar shear stress (τ) of 0.2 to 10 dyne/cm2 calculated by the following formula38 : where μ is the media viscosity (0.0085 g/cm per second), w is the channel width (1.0 cm), h is the channel height (0.0125 cm), and Q is the volumetric flow rate (0.0006 to 0.0316 cm3/s).
Each of the protocols described in this report was carried out in at least six sets of experiments. The results are expressed as mean±SEM. Significant differences between treatment groups were determined by the Student-Newman-Keuls test when making multiple comparisons or Dunnett’s test when making comparisons with controls.
Transient Time Course and NO Inhibition of Shear Stress–Induced NO Release
Fig 3⇓ shows that endothelial cells release NO in a transient fashion during continuous exposure to laminar shear stress. Confluent BAECs were placed in the parallel-plate flow chamber, and a shear stress of 0.1 dyne/cm2, which did not cause NO release, was maintained for 5 minutes. This ensured that assembly of the chamber did not disrupt cells, releasing intracellular agonists (such as ATP, acetylcholine, and bradykinin) that might cause NO release from intact cells. Shear stress was then increased to 3 dyne/cm2 and maintained as indicated by the bar. Within 5 seconds, the upstroke of NO release was detected, continued at a rate of 2.26±0.07 nmol/L/per second for 59.5±2.8 seconds (resulting in a maximum NO concentration of 134±3 nmol/L) (n=6), and then returned to baseline at a rate that was dependent on the flow. For 3 dyne/cm2 (0.57 mL/min), the time for the downstroke to decrease to 50% of peak height was 49.6±0.3 seconds at a concentration rate of 1.35±0.04 nmol/L per second (n=6). If flow was stopped, the lack of washout slowed the declining phase to 0.72 nmol/L per second. When measurements were made in a 200 nmol/L static NO solution, this rate of decline resulted in a half-time of 139±2 seconds (n=8; data not shown). This slow decline in the absence of flow was due principally to oxidative degradation of NO to NO2− (nitrite) and NO3− (nitrate), which are more stable than NO but are not detected by the Nafion-coated microsensor.30 Comparison of the half-time for NO oxidation (139±2 seconds) and the residence time of the solution in the chamber (2.64 seconds) suggests that the downstroke of the NO transient during washout is principally due to decreasing NO production by the endothelial cells and not oxidative breakdown or washout of NO.
In the presence of continuous shear stress, the transient release of NO was followed by an unresponsive interval (15.5±2.5 minutes, n=6) and then a second transient release of NO similar in time course and magnitude to the first release. This interval was independent of the magnitude of the shear stress. Incubation of the cells with exogenous NO (10 μmol/L for 2 minutes) before placement in the chamber blocked the initial NO release after imposition of shear stress (n=6; data not shown).
Effects of Increased Shear Stress on the Rate and Magnitude of NO Release
To characterize the relation between the level of shear stress and the magnitude of NO release, cells were exposed to a range of shear stresses (0.2 to 5 dyne/cm2, Fig 4⇓). Because of the 13- to 18-minute unresponsive period that follows the transient release of NO, each NO tracing depicted in Fig 4⇓ was recorded from a different coverslip of cells. A shear stress of 0.1 dyne/cm2, which did not elicit NO release, was in place for at least 5 minutes before the start of each experiment. At each arrow, the indicated level of shear stress was applied. It is readily apparent from Fig 4⇓ that the peak amplitude (current) increases as a function of the level of the shear stress. However, a comparison of these currents with currents obtained during calibration of the microsensor under flow conditions (in the absence of endothelial cells) indicated that there was only a small change of NO concentration with increasing shear stress. This consistency in NO concentration is evident from the 100-nmol/L calibration bar to the right of each amperogram.
In the flow system, the porphyrinic microsensor can operate as an amperometric, quasiamperometric, or coulometric detector. In the amperometric determination (electrochemical conversion efficiency close to 0%), current output is dependent on the cube root of the velocity.39 Conversely, the current response for coulometric detection (conversion efficiency close to 100%) is directly proportional to the velocity of the flowing solution. For the geometry of the flow cell and area of the porphyrinic microsensor in the present study, detection of NO appears to be quasiamperometric, with current being proportional to the square root of the flow.
Fig 5⇓ is a plot of changes in the amount or peak rate of release of NO (○) and the peak concentration of NO (▪) at various levels of shear stress (n=6 at each level of shear stress). The peak rate of NO release (in picomoles per second) was calculated by multiplying the peak NO concentration (in nanomoles per milliliter) times the flow rate (in milliliters per second). The peak release rate increased linearly with the shear stress. Exposure of the endothelial cells to laminar flow at the relatively low shear stress of 0.2 dyne/cm2 resulted in a very rapid production of NO, which reached a maximum concentration of ≈142±7 nmol/L. Under continuous flow, this concentration decreased only slightly, with a significant increase in the shear stress (up to 10 dyne/cm2).
Because the electrode is physically separated from the cells by ≈100 μm, the measured rate of NO release is affected by mass transfer (diffusion and convection) and oxidative degradation in the fluid as depicted in Fig 6⇓. One way to lessen these influences would be to decrease the chamber height. Unfortunately, a height of <100 μm has proved impractical. Therefore, a simplified mathematical model was developed in order to evaluate the effects of mass transport and oxidation and estimate the intrinsic rate of NO release after the onset of flow.
Model Analysis of NO Transport and Oxidation
The mathematical model is shown schematically in Fig 6⇑ and developed in the “Appendix.” A conservation relation for NO includes diffusion, convection, and oxidation in the fluid. Endothelial cells were assumed to release NO at a constant rate. The microsensor response was assumed to be proportional to the concentration of NO at the upper surface. The equations were solved by using the finite-element software package fidap (version 7.06, Fluid Dynamics International).
The NO concentration changes as a function of position along the flow chamber (Fig 7⇓). The NO concentration drops from 90% to 60% from the endothelial cell surface to the upper surface of the flow chamber. At the upstream portion of the electrode, NO concentrations are the lowest. Further downstream, more NO reaches the electrode by diffusion across the chamber and convection from upstream. Oxidation of NO in the fluid reduces the NO concentration slightly (Fig 7⇓).
Model results were used to calculate the flux of NO to the electrode. The intrinsic release rate was calculated by using the measured NO concentrations and the definition of the dimensionless concentration (“Appendix”). This approach was used to calculate the intrinsic release rate from the data in Fig 5⇑, and the results are shown in Fig 8⇓. Between 0.2 and 10 dyne/cm2, the intrinsic relative rate of NO release increased linearly with the shear stress. The model results are in good agreement with values estimated from data (Fig 5⇑). Including NO oxidation in the model increases the release rate by only 10%. In order for the electrode to record an equivalent NO concentration, the release rate is greater in the presence of NO oxidation because of the loss of NO as it is transported across the chamber. This small effect arises because the half-time of NO oxidation is much longer than the residence time of NO in the flow chamber.
Simultaneous NO and [Ca2+]i Measurements: Ca2+ Dependence of NO Release
Shear stress–induced NO release follows elevations of [Ca2+]i, as demonstrated in Fig 9⇓. At the bar, shear stress was increased from 0.1 dyne/cm2, which does not elevate cytosolic Ca2+, to 5 dyne/cm2. The initial rate of rise in [Ca2+]i was 2.4±0.2 nmol/L per second, which reached a semiplateau after 5.8±0.9 seconds. This semiplateau was followed by a more rapid second rate of rise of 16.3±0.8 nmol/L per second, which reached a final plateau after 17.3±1.3 seconds (n=4). The rise in [Ca2+]i preceded the release of NO by 2 to 3 seconds. The increase in concentration of NO was 2.5±0.2 nmol/L per second for 51.5±1.3 seconds, resulting in a 127.6±11.9 nmol/L peak NO release (n=4). The return to baseline of the cytosolic Ca2+ transient, which was not influenced by the flow rate, preceded the cessation of NO release. In other experiments, removal and EGTA (100 μmol/L) chelation of extracellular Ca2+ blocked the shear stress–induced rise in [Ca2+]i and the concomitant release of NO (n=4; data not shown).
Transient NO Release During Continuous Shear Stress
In the experiment shown in Fig 10B⇓, the periodicity of transient NO release seen in the experiment in Fig 10A⇓ was interrupted by the exogenous application of ATP (10 μmol/L) and Br-A23187 (10 μmol/L). Both experiments used a sustained shear stress of 3 dyne/cm2, the duration of which is indicated by the bars. The data argue that depletion of intracellular Ca2+ stores or substrate (l-arginine) and cofactors (calmodulin, tetrahydrobiopterin, NADPH, flavin adenine dinucleotide, or flavin mononucleotide) of eNOS do not account for the absence of a continual release of NO during constant shear stress. The time course and magnitude of NO release due to shear stress in Fig 10B⇓ was 2.4 nmol/L per second for 56 seconds, resulting in a 134 nmol/L peak NO release, similar to the shear stress–induced transients in Fig 10A⇓. Also similar were the time course and magnitude of NO release due to the 10 μmol/L bolus injection of the Ca2+ ionophore Br-A23187 (2.2 nmol/L per second for 65 seconds, resulting in a 143 nmol/L peak NO release). Transient releases of NO due to bolus injections of 10 μmol/L ATP were different (6.4 nmol/L per second for 30 seconds, resulting in a 192 nmol/L peak NO release for each injection). In addition to a more rapid rate of rise, ATP-evoked NO transients also returned to baseline more rapidly (3.8 nmol/L per second at 3 dynes/cm2), concomitant with cytosolic Ca2+ transients of shorter duration (data not shown).
The release of NO from the vascular endothelium, due to shear stress and endogenous agonists such as ATP, plays a major role in regulating vascular tone. This has been clearly demonstrated in experiments where removal or impairment of the endothelium has resulted in major increases in systemic blood pressure.40 41 However, the mechanism, time course, and kinetics of NO release due to shear stress and the effect of ATP on flow-dependent NO release are poorly understood. Although Milner et al26 and Dull et al27 reported that shear stress indirectly evoked NO release through ATP-mediated elevation of cytosolic Ca2+, Kuchan and Frangos28 argued that shear stress directly evoked biphasic NO release. The first phase of NO release was reportedly transient and Ca2+ dependent; the second phase, continuous or basal and Ca2+ independent. However, what our results show is that shear stress causes Ca2+-dependent, but ATP-independent, transient NO release. Furthermore, that release of NO may inhibit Ca2+ influx, resulting in the periodic release of NO during continuous shear stress. These novel findings were obtainable through the first direct measurements of changes in [Ca2+]i and NO release from BAECs in laminar flow.
NO electrodes can be made small enough to insert into a single cell.30 However, this approach was not used, because the presence of the electrode in the plasma membrane or near its surface could disturb the flow and alter the forces acting on the cell. The electrode (diameter, 0.025 mm; length, 4 mm) was placed at the top of the chamber ≈100 μm above the cells. When changes in [Ca2+]i were measured, cell fluorescence was acquired from an area (diameter, ≤0.025 mm) located in the focal plane of the cells, beneath the NO electrode. When flow was initiated in the parallel-plate chamber, shear stress was applied uniformly over the cell population, and the cells underwent elevations in cytosolic Ca2+ that were followed by the synthesis and release of NO in unison. Therefore, alignment of the area imaged for fluorescence with the NO electrode was not critical. When shear stress was applied (0.2 to 10 dyne/cm2), the rise in cytosolic Ca2+ always preceded the measured release of NO, irrespective of the location of the area imaged for fluorescence. However, when ATP or Br-A23187 was perfused into the chamber at 0.1 to 10 dyne/cm2 (at 0.1 dyne/cm2, drug could be added at a flow rate that did not evoke NO release in the absence of drug), agonist would be expected to initially activate cells at the inlet, and changes in cytosolic Ca2+ and NO release would be expected to parallel the time course of response to agonist. In the present study, the temporal relation between the occurrence of Ca2+ and NO fluxes depended on the relative locations of the recording sites.
Transient Versus Continuous Basal NO Release
The data in the present study clearly demonstrate the transient release of NO from cultured endothelial cells exposed to continuous shear stress. The advantage in using cultured cells is that the measured response is free from possible NO contributions from vascular smooth muscle and neuronal cells. Previous reports describing continuous or basal release of NO from endothelial cells exposed to shear stress were based on two categories of experiments, neither of which directly measures NO. What these experiments measured were (1) the time-averaged accumulation of the oxidation products of NO during continuous shear stress28 and (2) the time-averaged increase in soluble guanylyl cyclase and cGMP-inducing activities of effluent from blood vessel segments.42 43
The first category of experiments used cultured endothelial cells mounted in a closed-loop parallel-plate flow chamber.28 In those studies, measurement of the accumulation of the oxidation products of NO, NO2−/NO3−, was interpreted as basal NO release. However, through direct real-time measurements of NO release, we are able to report that these putative continuous elevations in NO2−/NO3− are, in fact, due to transient NO release. The basal release theory is a result of the oxidation of NO to the more stable NO2−/NO3− anions, which equilibrate in the recirculating perfusate and appear to increase gradually when measured intermittently over time.
In the second category of experiments, guanylyl cyclase and cGMP activities were used to estimate NO production. However, this estimated value could just as readily result from intermittent NO release as it could from continuous or basal NO release. Furthermore, since these experiments are typically carried out in freshly isolated preparations or in vivo, the possibility of basal NO release due to expressed iNOS cannot be ruled out. We also cannot rule out the possibility that the endothelium in freshly excised blood vessels expresses cNOS that exhibits basal activity and that this activity is lost in cultured cells. Under static conditions, cultured endothelial cells may release a trace amount of NO. However, the concentration never exceeded 5×10−9 mol/L NO, and this release was neither due to nor enhanced by shear stress.
Relation Between Shear Stress and NO Release
We report for the first time that exposure of endothelial cells to increasing laminar shear stress stimulates a linear increase in the amount or peak rate of NO release while the peak concentration of NO in the flowing medium remains essentially constant. This is a result of corresponding increases in the cytosolic [Ca2+] as shown in simultaneous experiments where Ca2+ and NO fluxes closely resembled one another (Fig 9⇑). Therefore, the cytosolic [Ca2+] was linearly proportional to the amount of NO produced but not to the actual concentrations of NO. This was in agreement with the work of other laboratories that measured only Ca2+ transients and observed that they increased as a function of shear stress.17 18 What was also novel in our findings was that shear stress induced intermittent Ca2+ transients and concomitant NO release.
Controversy has existed for some time regarding the shear stress inducibility of Ca2+ transients in endothelial cells. Whereas some groups reported multiple transients,17 18 others could evoke them only irregularly or not at all.44 45 46 One explanation may be differences in cytosolic Ca2+ buffering. When our endothelial cells were loaded with a commonly reported concentration of fura 2-AM (2 μmol/L for 20 minutes at 22°C), shear stress evoked multiple Ca2+ transients, but NO was not released. When the dye concentration was decreased (0.5 μmol/L for 40 minutes at 22°C), shear stress caused intermittent Ca2+ transients and NO release. Therefore, differences in the concentration of the dye or in its uptake by the cell may account for this disparity. Prior incubation of cells with exogenous NO (10 μmol/L for 2 minutes) prevented the cells from releasing NO for a period of time (13 to 18 minutes), similar to that observed between shear stress–induced transients.
It has been previously reported that NO may modulate its own release by means of shear stress through feedback inhibition of cNOS.47 In the present study, BAECs would respond to ATP (10 μmol/L) or Br-A23187 (10 μmol/L) during the period when they would not respond to shear stress. Thus, NO-induced inhibition was not occurring at the level of cNOS, its substrate, or its cofactors. Furthermore, the levels of NO necessary to partially inhibit cNOS48 are much higher than the levels of NO released by our cells. Therefore, the lack of continual NO release during constant shear stress probably involves an intermediate step in the mechanotransduction pathway. The fact that Ca2+ influx (Br-A23187) caused NO release when the cells would not respond to shear stress and that removal and EGTA (100 μmol/L) chelation of extracellular Ca2+ blocked the shear stress response strongly suggests that Ca2+ influx was being blocked. This blockade occurred whether or not there were any shear stress–evoked NO releases before the removal and EGTA (100 μmol/L) chelation of extracellular Ca2+. The fact that the exogenous application of NO initiated this unresponsive interval attributes this inhibition to NO. Recent findings have demonstrated that NO decreases Ca2+ currents in neonatal49 and adult50 ventricular myocytes as well as cardiac sinoatrial nodal cells.51 However, whether this putative blockade in endothelial cells involves the stretch-activated Ca2+ channels that were first described by Lansman et al,15 surface mechanoreceptors, or a second-messenger step(s) in the mechanotransduction pathway cannot be discerned by the present study.
It has been proposed that exogenous ATP is necessary for shear stress–induced NO release.26 27 According to this theory, endothelial cells release ATP that can bind to surface P2Y-purinoceptors, leading to IP3 formation, Ca2+ release from internal stores, and influx across the plasmalemma and NO formation. During low flow, ectoenzymes on the cell deplete the ATP concentration near the surface. As a result, cytosolic Ca2+ is not elevated. As flow increases, ATP is presented to the purinoceptors faster than the ectoenzymes can degrade it, resulting in Ca2+ transients and NO release. Although we do not discount the possibility that ATP may be released and that the cells respond accordingly, our data argue that shear stress–induced and ATP-evoked responses are separate entities. The fact that ATP evoked NO release during the period when cells were unresponsive to flow and that the kinetics of ATP-evoked and shear stress–induced NO release were different is strong evidence that different pathways are involved (see Fig 10⇑). However, during the period when cells were unresponsive to shear stress, NO release due to ATP (or A23187) reset the 13- to 18-minute interval between shear stress responses.
In conclusion, our novel findings argue that shear stress causes a Ca2+-mediated ATP-independent transient release of NO during which the amount or peak rate of release, but not the peak concentration, depends on the level of shear stress. The transient nature of this response may be due to NO-induced inhibition of Ca2+ influx via a mechanism yet to be determined. Thus, under this paradigm, if NO-inducing agents (such as acetylcholine, ATP, and bradykinin) are absent from a given region of the vasculature for >15 to 18 minutes, shear stress would evoke a transient release of NO. Although the flow rates would be expected to vary, the peak concentration of NO released would remain the same.
Selected Abbreviations and Acronyms
|BAECs||=||bovine aortic endothelial cells|
|Br-A23187||=||brominated Ca2+ ionophore|
|cNOS||=||constitutive NO synthase|
|eNOS||=||endothelial NO synthase|
|FBS||=||fetal bovine serum|
|iNOS||=||inducible NO synthase|
|TMHPPNi||=||nickel (II) tetrakis(3-methoxy-4-hydroxyphenyl) porphyrin|
Development of a Numerical Model of NO Release and Transport
To evaluate the effects of NO transport and oxidation in the fluid on the measured rate of release, a mathematical model of shear stress–induced NO release from BAECs was developed. As shown in Fig 6⇑, the local concentration in the fluid represents a balance among diffusion around the cells, convection in the direction of flow, and oxidation of NO in the fluid. For dilute solutions, Fick’s law of diffusion is applicable. Applying the conservation of mass to a differential volume yields the following38 : (A1) ∂C/∂t+vx(∂C/∂x)=D(∂2C/∂y2+∂2C/∂x2)−ROX
where vx is the local fluid velocity (in centimeters per second) and ROX is the rate of NO oxidation. Flow is laminar and fully developed. The velocity profile is as follows: (A2) vx=6<vx>(y/h−y2/h2)
where <vx>is the average channel velocity, y=0 is the base of the channel where endothelial cells are present, and h is the channel height.
A number of additional simplifications were made. Diffusion in the x direction was neglected for the following reason: Average fluid velocities range from 0.05 to 1.26 cm/s. The Peclet number (<vx>L/D, where L is the length of the chamber and D is the diffusion coefficient of NO, 3.33×10−5 cm2/s)52 represents the ratio of diffusion to convection. When the Peclet number is large, convection dominates over diffusion. Conversely, when the Peclet number is much smaller than one, diffusion is the dominant mechanism in the direction of flow. For this problem, the Peclet number ranges from 3000 to 75 700. Thus, in the x direction, diffusion is negligible, relative to convection.
Although the release of NO is time dependent, a steady state analysis was used to evaluate the transport steps. This is appropriate because we are interested in how flow affects the measured release rate and are not interested in modeling the detailed kinetics of release.
The kinetics of NO oxidation were assumed to be second order in the presence of NO and first order in the presence of O253 : (A3) ROX=k′2CO2C2NO
where the rate constant k′2 equals 6.3×106 L2/mole2 per second.53
Since the dissolved oxygen concentration in aqueous solution (1.2×10−3 mol/L) was considerably larger than the NO concentration (≤10−6 mol/L), the oxygen concentration was essentially constant. Thus, the reaction was considered pseudo–second order: (A4) ROX=k2C2NO
The published52 value of k′2 yields k2=k′2CO2=7560/M-s. From measurements of NO oxidation under static conditions, we found that k2=35 971±518/M-s. The latter value was used in all subsequent calculations.
By use of these simplifications, the conservation relation reduces to the following: (A5) 6<vx>(y/h−y2/h2)(∂C/∂x)=D(∂2C/∂y2)−k2C2
Boundary conditions for concentration need to be specified at x=0, y=0, and y=h. At the inlet x=0, the NO concentration is zero. Endothelial cells at y=0 release NO at a rate RO (moles NO/cm2-s). The response of the microsensor was assumed to be rapid relative to the rate of NO delivery to the microsensor. The microsensor was assumed not to consume significant amounts of NO. Thus, the flux at y=h was set to 0.
To simplify the solution and presentation of results, Equation A5 and boundary conditions were placed in dimensionless form: (A6) 6(y*−y*2)(∂C*/∂x*)=(∂2C*/∂y*2)−φ2C*2
where C*=CD/hRO, y*=y/h, x*=xD/vx>h2, and φ2=k2(hRO/D)h2/D. Based on measured values of the rate constants and the dissolved oxygen concentration, φ is ≤1.00. The initial boundary conditions were as follows: (A7a) x*=0 with C*=0 (A7b) y*=0 with ∂C*/∂y*=−1 (A7c) y*=1 with ∂C*/∂y*=0
The equations were solved by using the finite-element software package fidap (version 7.06, Fluid Dynamics International). The Galerkin form of the method of weighted residuals was used.54 The number of nodes used was between 1000 and 10 000, depending on the value of x*.
The release rate was calculated by determining the mixing cup concentration and the definition of the dimensionless concentration.
This study was supported by National Research Award 1F32-HL-09072-01A1 from the National Heart, Lung, and Blood Institute, Grand-in-Aid NC-95-GB-08 from the American Heart Association, North Carolina affiliate (Dr Kanai), grants HL-19216 and HL-45132 from the National Heart, Lung, and Blood Institute (Dr Strauss), Grant-in-Aid 93012390 from the American Heart Association (Dr Truskey), and a grant from the Biotechnology Research Program of Oakland University (Dr Malinski). The authors would like to thank Donald Pearce of the Duke University Surgical Instrument Shop for machining the parallel-plate flow chamber.
Previously presented as preliminary results in abstract form (Circulation. 1994;90[suppl I]:I-139).
This manuscript was sent to John T. Shepherd, Consulting Editor, for review by expert referees, editorial decision, and final disposition.
- Received February 27, 1995.
- Accepted April 14, 1995.
- © 1995 American Heart Association, Inc.
Buga GM, Gold NE, Fukuto JM, Ignarro LJ. Shear stress–induced release of nitric oxide from endothelial cells grown on glass beads. Hypertension. 1991;17:187-193.
Cook JP, Rossitch E, Andon NA, Loscalzo L, Dazu VJ. Flow activates an endothelial potassium channel to release an endogenous nitrovasodilator. J Clin Invest. 1991;88:1663-1671.
Myers RR, Minor RL, Guerra R, Bates JN, Harrison DG. Vasorelaxant properties of the endothelium-derived relaxing factor more closely resembles S-nitrosocysteine than nitric oxide. Nature. 1990;365:161-163.
Linder L, Kiowski W, Bühler FR, Lüscher TF. Indirect evidence for release of endothelium-derived relaxing factor in human forearm circulation in vivo: blunted response in essential hypertension. Circulation. 1990;81:1762-1767.
Ontkean M, Gay R, Greenberg B. Diminished endothelium-derived relaxing factor activity in an experimental model of chronic heart failure. Circ Res. 1991;69:1088-1096.
Shen J, Luscinskas FW, Connolly A, Dewey CF, Gimbrone MA. Fluid shear stress modulates cytosolic free calcium in vascular endothelial cells. Am J Physiol. 1992;262:C384-C390.
Malek AM, Gibbons GH, Dzau VJ, Izumo S. Fluid shear stress differentially modulates expression of genes encoding basic fibroblast growth factor and platelet-derived growth factor B chain in vascular endothelium. J Clin Invest. 1993;92:2013-2021.
Davies PF, Tripathi SC. Mechanical stress mechanisms and the cell: an endothelial paradigm. Circ Res. 1994;72:239-245.
DePaola N, Gimbrone MA, Davies PF, Dewey CF. Vascular endothelium responds to fluid shear stress gradient. Arterioscler Thromb. 1992;12:1254-1257.
Stuehr DJ, Marletta MA. Mammalian nitrate biosynthesis: mouse macrophages produce nitrite and nitrate in response to Escherichia coli lipopolysaccharide. Proc Natl Acad Sci U S A. 1985;82:7738-7742.
Kuchan MJ, Frangos JA. Role of calcium and calmodulin in flow-induced nitric oxide production in endothelial cells. Am J Physiol. 1994;266:C628-C636.
Himmel HM, Rasmusson RL, Strauss HC. Agonist-induced changes of [Ca2+]i and membrane currents in single bovine aortic endothelial cells. Am J Physiol. 1994;267:C1338-C1350.
Grynkiewixz G, Poenie M, Tsien RY. A new generation of Ca2+ indicators with greatly improved fluorescence properties. J Biol Chem. 1985;260:3440-3450.
Mesaros S, Rievaj M, Bustin D. Absolute chronoamperometric determination with cylindrical semimicroelectrode. Collect Czech Chem Commum. 1993;58:281-290.
Batchelor GK. An introduction to fluid dynamics. Cambridge, UK: Cambridge University Press; 1967:179-182.
Schlichting H. Boundary-Layer Theory. 7th ed. New York, NY: McGraw Hill Publishing Co; 1979:185-187.
Fung YC. Biodynamics: Circulation. New York, NY: Springer-Verlag; 1984:141.
Bird RB, Stewart WE, Lightfoot EN. Transportation Phenomena. New York, NY: John Wiley & Sons Inc; 1960:47.
Probstein RF. Physiochemical Hydrodynamics. Stoneham, Mass: Butterworth Publishers; 1989:53-81.
Pohl U, Holtz J, Busse R, Bassenge E. Crucial role of endothelium in the vasodilator response to increased flow in vivo. Hypertension. 1986;8:37-44.
Hecker M, Mülsch A, Bassenge E, Busse R. Vasoconstriction and increased flow: two principles of shear stress-dependent endothelial autacoid release. Am J Physiol. 1993;34:H828-H833.
Busse R, Mülsch A, Fleming I, Hecker M. Mechanisms of nitric oxide release from the vascular endothelium. Circulation. 1993;87(suppl V):V-18-V-25.
Mo M, Eskin SG, Schilling WP. Flow-induced changes in Ca2+ signaling of vascular endothelial cells: effect of shear stress and ATP. Am J Physiol. 1991;29:H1698-H1707.
Dull RO, Davies PF. Flow modulation of agonist (ATP)-response (Ca2+) coupling in vascular endothelial cells. Am J Physiol. 1991;30:H149-H154.
Buga GM, Griscavage JM, Rogers NE, Ignarro LJ. Negative feedback regulation of endothelial cell function by nitric oxide. Circ Res. 1993;73:808-812.
Balligand JL, Kelly RA, Marsden PA, Smith TW. Control of cardiac muscle cell function by an endogenous nitric oxide signaling system. Proc Natl Acad Sci U S A. 1993;90:347-351.
Méry PF, Pavoine C, Belhassen L, Pecker F, Fischmeister R. Nitric oxide regulates cardiac Ca2+ currents. J Biol Chem. 1993;35:26286-26295.
Han X, Shimoni Y, Giles W. An obligatory role for nitric oxide in autonomic control of mammalian heart rate. J Physiol (Lond). 1994;476.2:309-314.
Kharitonov VG, Sundquist AR, Sharma VS. Kinetics of nitric oxide autoxidation in aqueous solution. J Biol Chem. 1994;269:5881-5883.
Finlayson BA. Nonlinear Analysis in Chemical Engineering. New York, NY: McGraw-Hill; 1980:126-136.