| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Surgery, University of Wisconsin-Madison, Madison 53792-7375; and The William S. Middleton Memorial Veterans Hospital, Madison, Wisconsin 53705-2286
Submitted 14 March 2003 ; accepted in final form 3 September 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
2,000 alveoli. Dispersion indexes revealed significantly more particle clustering (inhomogeneous distribution) in vasoconstricted lungs than in normal flow controls or in controls in which flow was reduced by either lowering pulmonary arterial pressure or raising left atrial pressure. These results suggest that vasoconstriction occurred in the microvessels themselves, which are much smaller vessels than those previously thought to be capable of vasoconstriction. alveolar perfusion; fluorescent microspheres; vasoconstriction; dispersion index analysis; perfusion heterogeneity
30–40 µm, because these are the smallest vessels around which smooth muscle cells have been detected (9, 12, 15). Our laboratory estimated that 750 alveoli would be served by a 30-µmdiameter arteriole (4). If control does not exist in vessels smaller than this, then 750 alveoli is the smallest functional control unit of the pulmonary circulation. The absence of detectable smooth muscle in vessels smaller than 30 µm leads to this unavoidable conclusion.
In this paper, we present data suggesting that control of perfusion may exist in much smaller vessels, perhaps as small as interalveolar corner vessels. Corner vessels are the second smallest vessel generation in the lung, larger only than the alveolar septal vessels (6, 11). Our method was to measure trapping patterns of 4-µm-diameter latex particles infused into the pulmonary circulation of isolated rat lungs vasoconstricted with the thromboxane receptor analog U-46619. Our laboratory has shown previously that particles of this diameter become trapped mainly in corner vessels (4, 6). We found that particles in vasoconstricted lungs were significantly more clustered than in control lungs and that particles were completely absent from some lung microregions. We believe that such patterns would not be possible if vasoconstriction was restricted to vessels with diameters 
30 µm.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
We anesthetized retired male breeder rats (447 ± 59 g; n = 16) with intramuscular ketamine (40 mg/kg), xylazine (6 mg/kg), and acepromazine (1 mg/kg) and tied them supine. We placed a polyethylene cannula into a femoral vein and infused heparin (750 U/kg). After 10 min to allow the heparin to circulate, we cut the femoral artery and allowed the animal to exsanguinate. We then placed a polyethylene cannual into the trachea, inflated the lungs with air to 5 cmH2O, and opened the chest widely with a sternum-splitting incision. After placing fluid-filled cannulas into the pulmonary artery and left atrium, we removed the heart and lungs from the thorax and placed them into a perfusion chamber, where the arterial cannula was connected to a reservoir containing 3% bovine serum albumin (Sigma Chemical) in phosphate-buffered saline. The tracheal cannula was connected to a piston pump, which we used to ventilate the lungs with air (25 breaths/min) at inflation and deflation pressures of 15 and 5 cmH2O. We began perfusion by raising the arterial reservoir 15 cm above the base of the lung. The outlet of the left atrial cannula was level with the base. Atrial effluent drained into a reservoir beneath the cannula, and this was returned to the arterial reservoir by using a pump. Flow was measured by counting the effluent drop rate. The liquid volume of the perfusion system, excluding the pulmonary circulation, was 27 ml.
We prepared a stock solution of U-46619 (Sigma Chemical) in 20 ml of ethanol, and aliquots of this were added to the lung perfusate reservoir until perfusate flow fell to a plateau equal to 20% of baseline. Aliquot volumes needed to achieve this averaged 330 ± 70 µl, which produced average U-46619 perfusate concentrations of 1.6 x 10-6 M. Flows stabilized at plateau values after 25–30 min. The arterial reservoir remained 15 cm above the base of the lung so that pulmonary artery pressure (Ppa) was unchanged.
Once perfusate flows were stable, we infused a solution containing 4-µm-diameter rhodamine-labeled latex particles into the pulmonary arterial cannula. Diameters of these particles typically vary by 1–2% (5). The particle solution was prepared by adding 1 x 107 particles (20 µl of stock particle suspension; Molecular Probes) to 5.0 ml of 3% albumin solution. We then added salts to this (NaCl, NaH2PO4, Na2HPO4) to equal their concentrations in the perfusate. The salts were added after the particles so that the albumin would first coat the particles and prevent them from clumping during addition of the salts. We also added enough U-46619 to equal the concentration in the perfusate. We found that if we omitted this step, perfusate flow would rise during particle infusion because of dilution of the perfusate U-46619 concentration by the infusate. Finally, we added enough Evans blue to produce an Evans blue-to-albumin molar ratio of 1:4 in the lung perfusate after infusion. We did this so that the albumin would fluoresce during subsequent microscopic examination of the lungs so that we could compare the distribution of perfusate liquid, as judged by albumin fluorescence, with that of the particles. The final solution was infused into the pulmonary arterial cannula over 5 min by using a syringe pump.
Once the infusion was complete, both vascular cannulas were simultaneously clamped, then severed, and fitted with stoppers to prevent perfusate redistribution. The tracheal cannula was also clamped to maintain inflation. The lung was then removed from the perfusion chamber, and the tracheal cannula was connected to a source of compressed air with tracheal pressure set to 20–25 cmH2O. The lung was maintained at this pressure for 2 days to allow the lungs to dehydrate.
Control lungs were prepared and infused similarly except that U-46619 was omitted. Three sets of controls were prepared. The first was perfused at the same vascular pressures as the U-46619-perfused lungs (normal Ppa; n = 4). The second set was perfused at reduced Ppa to simulate arteriolar constriction (reduced Ppa; n = 4), and the third set was perfused at raised left atrial pressure to simulate venular constriction (raised Pla; n = 4). Vascular pressures in the reduced Ppa and raised Pla groups were adjusted to produce perfusate flows that approximated those in the U-46619-perfused lungs. We also prepared lungs in which U-46619 was added after particle infusion to determine whether this agent could cause particle redistribution after the particles were trapped within the microcirculation. We carried out these studies at both constant flow and constant pressure. In the constantflow studies (n = 2), the particles were infused with Ppa set at 15 cmH2O, then U-46619 was added to the perfusate, and Ppa was steadily raised to maintain flow at constant values. Ppa stabilized at plateau values of 23.7 ± 1.2 cmH2O after 25–30 min. The constanpressure studies (n = 2) were conducted with Ppa maintained at 15 cmH2O during particle infusion and subsequent U-46619 infusion.
We used a specimen knife (Pathco) to cut each air-dried lung into three to four sections (2–3 mm each) in the apical-basal plane. Each section was placed onto the stage of a confocal fluorescence microscope (Bio-Rad MRC 1024ES) where confocal images were obtained of latex particle fluorescence. Each image was composed of 10 optical sections (3,360 x 3,360 x 10 µm) that were merged to form a single image that encompassed 3,360 x 3,360 x 100 µm. A total of five merged images were obtained from each lung. Co-incident images of Evans blue albumin fluorescence were obtained simultaneously so that albumin distribution could be compared with latex particle distribution. Examples of particle and albumin confocal images are shown in Figs. 1, 2, 3, 4.
|
|
|
|
|
Particle distribution was quantified by using methods developed by our laboratory (4). Briefly, images consisting only of particle fluorescence were subdivided, first into a 2 x 2 array (4 subdivisions), then 4 x 4 (16 subdivisions), and so on, stepwise, up to 512 x 512 (9 steps) (Fig. 3). At each step, the number of particles in each subdivision was counted, and the mean (µ) and variance (
2) were calculated for all subdivisions within that step. Values of log (2/µ) were plotted against the log of the step subdivision volume. The subdivision volume is the volume of tissue represented in the confocal image. For example, in the 2 x 2 array, each subdivision represented a tissue volume of 3,360/2 x 3,360/2 x 100 µm, or 2.8 x 108 µm3 (Fig. 3). The resulting plot of log (2/µ) vs. log tissue volume for all nine steps is referred to as a dispersion index (DI) plot (Fig. 5). In a DI plot, random particle distribution is denoted by values of log (2/µ) = 0. Values of log (2/µ) > 0 denote particle clustering. Values of log (2/µ) < 0 denote a lattice-like distribution as in the distribution of squares on a checker board.
Statistics. Results are expressed as means ± SD. We used a repeated-measures analysis of variance to compare flow data among treatment groups. Differences were considered to be significant at P 
0.05. Analyses were performed with Statview statistical software (SAS Institute, Cary NC).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
DI analysis of latex particle distributions in confocal images showed that particle distributions in U-46619-perfused lungs were significantly more clustered than in lungs perfused without U-46619 (P = 0.0001). This is shown by the significantly higher log DI values for the U-46619-perfused lungs, especially at the largest subdivision volumes, which correspond to 1 x 103 alveolar volumes (Fig. 5). Clustering differences can also be seen visually by comparing latex particle images from U-46619-perfused with those of control lungs (Figs. 3 and 4). Particle clustering in U-46619 perfused lungs is marked by areas of both high and low particle densities, whereas in control lungs, particle clustering is more uniform. Log DI values for the three control groups were not significantly different from each other.
It is worthwhile to compare latex particle distributions with Evans blue albumin distributions in U-46619-perfused lungs. This can be seen in Figs. 1 and 2 where the particles are red and the fluorescent albumin is green. Note the albumin fluorescence distribution appears to be quite uniform compared with that of the particles. Lung areas that lack particles do not appear to lack green fluorescence. Our interpretation of this is that the liquid (albumin) fraction of the perfusate was able to flow freely throughout all lung microregions, whereas the particle fraction was not. To confirm this appearance, we statistically compared the particle distribution relative to the albumin.
We did this by conducting DI analysis of the particles only within that portion of each lung image occupied by Evans blue albumin fluorescence. In the DI analysis described above (Fig. 5), the particle distribution was measured throughout the entire frame of the image area. In this analysis, the particle distribution was measured only in that portion of each image frame occupied by albumin fluorescence. This allowed us to measure particle distribution relative to that of the albumin.
We collected two coincident confocal images of each lung field: one showing only particle fluorescence and the other showing only Evans blue albumin fluorescence. The pixels in the albumin image were used as the space in which the particle DI analysis was conducted. The mathematical basis for this approach has been described previously (4). We completed this analysis on 10 image pairs in two U-46619-perfused lungs (5 pairs per lung) obtained by using the x4 objective of the microscope. The results (Fig. 6) show that the log DI values averaged 1.20 ± 0.42 at the largest tissue volume (2.8 x 108 µm3). If the particles had been distributed randomly with respect to the albumin, the average log DI value would have been zero, which is the value associated with a random distribution. The fact that the maximum log DI value was significantly greater than zero shows that the particle distribution was significantly clustered compared with that of the albumin. This suggests that during U-46619 perfusion, the liquid (albumin) fraction of the perfusate was able to flow much more freely throughout lung microregions than the particles. If the albumin and particle distributions had been impeded similarly, we would have expected the average log DI value for this analysis to be near zero.
|
We conducted several studies to confirm that U-46619 had no direct effect on the latex particles that might have accounted for their clustered appearance within the lung independent of the drug's vasoactive properties. First, we conducted DI analysis on smears of particles on microscope slides. The smears contained either particles in albumin solution or particles in albumin solution to which U-46619 had been added (1.6 x 10-6 mol/l). DI values were equal in both sets of smears, which demonstrates that U-46619 had no effect on the particles themselves.
We also conducted DI analysis on lungs into which U-46619 was infused after particles had been infused to determine whether U-46619 could cause particle redistribution after the particles were trapped within the lung. We did this in lungs perfused at constant pressure or constant flow. In lungs perfused at constant pressure, the DI plot (not shown) was nearly identical to that of the normal Ppa plot in Fig. 5, and had a maximum log DI value of 1.02 ± 0.26 (not significant). The normal Ppa plot in Fig. 5 has a maximum log DI value of 0.97 ± 0.57. This demonstrates that adding U-46619 to the perfusate after particles had been infused had no measurable effect on particle distribution. In lungs perfused at constant flow, the DI plot showed slight but insignificantly more clustering than the normal Ppa plot in Fig. 5 (not signficant), but significantly less clustering than that of the U-46619 plot in that figure (P = 0.025). The maximum log DI value for the constant-flow plot was 1.27 ± 0.30. By comparison, the maximum log DI value for the U-46619 plot in Fig. 5 is 1.41 ± 0.43. Our interpretation is that raising the perfusion pressure to maintain constant flow after U-46619 infusion in these lungs (maximal Ppa, 23.7 ± 1.2 cmH2O) may have caused slight but insignificant particle redistribution.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
30 µm, and the percentage of such vessels surrounded by smooth muscle is most likely not more than 2%. This is based on the data of Meyrick and Reid (12), who measured frequencies of muscularized arteries in the lungs of rats exposed to hypoxia for up to 7 wk. After 2 days of exposure, they found that only 1 artery in 55 (2%), with diameters of 15–85 µm, was surrounded by smooth muscle (12). This suggests that in normoxia, where arteriolar muscularization is less than that in hypoxia, the majority of vessels surrounded by smooth muscle must have diameters >85 µm. Nonetheless, if we assume that 30 µm is the smallest resistance vessel diameter, it is difficult to hypothesize how the particle trapping patterns we observed could be explained by constriction of such vessels. One possibility is that vasoconstriction with U-46619 caused narrowing but not collapse of alveolar corner vessels, which have diameters of 4–5 µm (6). This would explain why the latex particles became clustered without any apparent change in the albumin distribution. Nonhomogenous corner vessel narrowing would impede the particles in some areas and produce a complete absence of particles in others. Such narrowing would not restrict the distribution of vascular liquid. This is the pattern shown by our confocal images. If this hypothesis is correct, it suggests that control of perfusion may exist in much smaller vessels than previously thought.
Cellular mechanisms by which such narrowing could be mediated are unknown. However, there are several possibilities. Pulmonary microvascular endothelial cells (PMVECs) are known to be amply equipped with intracelluar contractile proteins. PMVECs have been shown to contain filamentous actin at the cell periphery, within the cytoplasm, and around the nucleus. Also, the perinuclear region has been shown to contain nonmuscle myosin (7, 13). Evidence for the contractile power of these proteins was shown by Morel and colleagues (14), who added angiotensin II and bradykinin to PMVECs in culture, which caused the cells to contract enough to deform their silicone rubber substrates. Furthermore, these authors reported that endothelial cells obtained from pulmonary arteries and retinal microvessels did not contract under these conditions. These findings demonstrate that PMVECs have contractile properties that are unique to these cells.
Other contractile cells are known to be present in the vicinity of lung microvessels. Weibel (16) reported that pericytes are present around alveolar capillaries, and he described the appearance of these cells as similar to that of smooth muscle cells in venules. Kapanci and colleagues (10) also described interstitial cells in alveolar septa that contained microfilaments that stained positively for both actin and myosin (10). These investigators found these cells most often in pre- and postcapillary regions, which would place them adjacent to corner vessels. This is the region in which we believe our latex particles became trapped. Although pericytes and interstitial cells both contain filaments, it is not known under what conditions these filaments contract, if at all. Thus their function is unclear.
The known contractile properties of PMVECs, plus the potential contractile properties of pericytes and interstitial cells, are all possible mechanisms by which alveolar mircrovessel caliber might be controlled. Our hypothesis is that these mechanisms may produce subtle changes in PMVEC thickness or microvessel diameter that could affect intraluminal caliber. Small reductions in caliber would inhibit the ability of red blood cells (or latex particles) to flow down the bore of those vessels, without altering the flow of vascular liquid. This might be a mechanism for controlling erythrocyte distribution at the alveolar level.
Additional evidence for microvascular flow control has been suggested by results of studies in hypoxia. Clough and colleagues (3) used microfocal angiography to show that hypoxia reduced lung microvascular volume, which they suggested might have been due to active contraction at the capillary level. They hypothesized that pericytes or interstitial cells were potential sources of such contraction.
U-46619 mimics the effects of thromboxane A2 on smooth muscle, and it has been known for many years that it is a potent pulmonary vasoconstrictor. Its action on pulmonary resistance vessels has been the subject of much study. Barnard and colleagues (2) used the double-occlusion method to identify vascular segments in which U-46619 vasoconstriction occurred in isolated rat lungs. They concluded that it increased resistance only in small arteries and decreased compliance in the middle segment, which, although anatomically undefined, would include the alveolar microvessels, where we found that latex particles were clustered (2). However, in dog lungs, U-46619 was shown to increase venular resistance (1). Possible differences in resistance effects on arterioles and venules were the reason that we prepared low flow controls by either lowering arterial pressure or raising venular pressure. However, neither method was able to reproduce the particle-trapping patterns we saw in U-46619-vasoconstricted lungs. It might be argued that lowering arterial pressure and raising venous pressure do not adequately reproduce the effects of vascoconstriction, and this is especially true if vasoconstriction is nonuniform and does not affect all resistance vessels equally. The patchy particle distribution we observed in U-46619-perfused lungs certainly suggests nonuniformity. However, it occurred at much smaller scales than would be expected if vasoconstriction had been mediated through effects on 30-µmdiameter vessels, each of which serves 750 alveoli (4). We found that variations in particle trapping occurred on much smaller scale, as would be expected if variations in vasoconstriction occurred in much smaller vessels.
If vasoconstriction was restricted to vessels with diameters 30 µm, it seems unlikely that such vessels could constrict enough to obstruct the passage of 4-µm-diameter latex particles. Fike and colleagues (8) showed that 1 x 10-6 M concentrations of U-46619 caused a 40% reduction in the diameters of isolated piglet pulmonary arteries 100–400 µm in diameter. This is equivalent to U-46619 concentrations in perfusates of our isolated rat lungs. However, a 40% constriction in a 30-µm-diameter vessel would reduce it to 18 µm, which would be insufficient to exclude the passage of 4-µm-diameter particles. This suggests that the particle clustering pattern we observed was caused by vessels smaller than those thought to be surrounded by smooth muscle. Impediment of a 4-µm-diameter particle could be caused by a 40% diameter reduction in a vessel with an unconstricted diameter of 6–7 µm. However, this is significantly smaller than the smallest vessels thought to be capable of vasoconstriction.
Our studies were conducted in isolated lungs, and it should be noted that the results we obtained in control (nonvasoconstricted) lungs (normal Ppa) differed from those our laboratory reported recently in the lungs of intact, spontaneously breathing rats (4). In our present studies, log DI for controls reached maximum values of 0.5–1.0 (Fig. 5). In intact rats infused with particles of the same number and diameter, our laboratory reported maximum log DI values of 0.1–0.2 (4). These smaller values show that particles were less clustered in the lungs of intact rats than in isolated lungs, which suggests that perfusion heterogeneity in isolated lungs is greater than that in intact lungs. Reasons for this difference are unknown, although cardiac outputs in intact rats are reportedly more than 10-fold greater than those we measured in our normal Ppa controls (17). This might explain the log DI differences we measured, but our results also demonstrate that microvascular perfusion distribution in isolated lungs does not equal that in intact lungs.
In our laboratory's recent publication describing the methods we used in this report, we concluded that the number of latex particles infused affected the resolution of the DI analysis method. We concluded that, in the lungs of intact rats, 2 x 108 4-µm particles provided better DI resolution than 1 x 107 particles (4). However, in preliminary studies, we found that infusion of 2 x 108 particles into U-46619-perfused lungs produced particle densities that were too high for individual particles to be resolved in particle clusters. We therefore chose to infuse 1 x 107 particles for our present studies. Despite this, it was still difficult to discriminate individual particles in clusters, as can be seen in Figs. 1 and 3. Higher magnification revealed that the particles were not clumped in those clusters (Fig. 2), although at low magnification (Fig. 1) they appeared to be. Despite this, the DI method was still able to resolve a statistically significant difference in particle clustering between U-46619-perfused lungs and controls (Fig. 5).
Conclusion. Our results suggest that U-46619-induced vasoconstriction caused a redistribution of 4-µm-diameter latex particles that cannot be explained in terms of conventional theories of the control of pulmonary microvascular perfusion. We believe that our data provide evidence that thromboxanemediated vasoconstriction occurs in much smaller vessels than those previously thought to be capable of such constriction.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
GRANTS
This study was supported by a grant from the Department of Veterans Affairs.
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. T. Lovering, M. K. Stickland, A. J. Kelso, and M. W. Eldridge Direct demonstration of 25- and 50-{micro}m arteriovenous pathways in healthy human and baboon lungs Am J Physiol Heart Circ Physiol, April 1, 2007; 292(4): H1777 - H1781. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
), normal Ppa controls (
), raised left atrial pressure (Pla) controls (
), and reduced Ppa controls (
). Tissue subdivision volumes are shown on the y-axis in units of µm3 and in multiples of alveolar volume (alveolar diameter, 75 µm). Particle distribution is statistically random at a log dispersion index value of 0, and it becomes increasingly clustered as log dispersion index exceeds 0. Data for the U-46619-vasoconstricted lungs have greater log dispersion index values (mean ± SD) than those of controls at all lung volumes >1x105 µm3 (P 