| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Department of Chemical Engineering and 2Department of Mechanical Engineering, City College of the City University of New York, New York, New York; 3Institute of Biomedical Science of the Academia Sinica, Taipei, Taiwan, Republic of China; and 4Department of Medicine, College of Physicians and Surgeons of Columbia University, New York, New York
Submitted 27 December 2006 ; accepted in final form 23 January 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
150 nm), sparse layer immediately beneath the endothelium where the HRP concentration was much higher than that in the matrix below it. Nievelstein-Post et al.'s (Nievelstein-Post P, Mottino G, Fogelman A, Frank J. Arterioscler Thromb 14: 1151–1161, 1994) ultrarapid freezing/rotary shadow etching of the normal rabbit valve's subendothelial space supports the existence of this very thin, very sparse "valvular subendothelial intima," in analogy to the vascular subendothelial intima.
valvular subendothelial intima; focal horseradish peroxidase spots; convective transport
|
Because of the above-mentioned variability in the magnitudes and shapes of the measured transvalvular concentration profiles, the parameters fitted for the different profiles reflected this by exhibiting very large variations. Such variation seemed to signify large mass transport variations both between different leaflets of the same aortic valve and even between different regions of the same valvular leaflet. However, the scale of this variability is troubling. For example, the mass transfer coefficients varied for the monkey's aortic endothelium from 5.1 x 10–9 to 73 x 10–9 cm/s and for its ventricular aspect from 4 x 10–13 to 3 x 10–3 cm/s, ten orders of magnitude. If we exclude the largest and smallest values, this parameter still varied by a factor of 20. It is therefore likely that this 1D model is too simple and is meant only to quantify, rather than to account for, these large variations. This motivates us to question the three assumptions that underlie this 1D model. Since such assumptions are commonly made for transport in vessels and valve leaflets of any species, it is valid to question these assumptions and to experimentally investigate their validity, in particular, in the rat valve model, which presents fewer ethical issues than work on monkeys.
Let us review the transport in the more well-studied artery wall. The endothelium, a monolayer of endothelial cells (ECs), is a barrier that blocks macromolecules in the blood from penetrating into the tissue. Weinbaum et al.'s (31) mathematical model proposed that macromolecules cross the arterial endothelium through the intercellular junctions around rare ECs that temporarily leak while in turnover, rather than being ferried through ECs in vesicles. Theoretically, these wide-junction leaks could transport enough macromolecules across the endothelium that even a few leaky junctions could dramatically increase the total endothelial permeability to macromolecules (23), even though the en face area of these leaky sites could be as little as 10–6 of the entire endothelial surface (31). After short-time LDL circulation, Stemerman et al. (23) found isolated local sites of elevated permeability, i.e., leaks, rather than uniform leakage in rabbit aorta, but they did not elucidate the nature of those spots. Lin et al. found, by scanning the entire rat aorta, that nearly 99% of all mitotic ECs identified by hematoxylin staining leaked Evans blue-albumin (EBA) (14) and 80% of all mitotic ECs leaked the much larger Lucifer yellow-LDL (LY-LDL) (15). However, mitotic ECs accounted for only 23% (14) of the total number of leakage sites for EBA and 45% (15) for LY-LDL. Stigmatic, dying ECs and cells in the process of sloughing off dying ECs accounted for more leakage sites (3). Truskey et al. (27) used 125I-labeled LDL autoradiography to show that only 25% of leaks were associated with mitosis. Although all of this work focused on the arterial endothelium, it is tantalizing to suspect that focal leakage may be responsible for Tompkins et al.'s (26) observed variability, since, even though the valve has a very different structure and function from the artery wall, the valvular endothelium is contiguous with the aortic endothelium, the endothelium being one of the largest "organs" in the body (12). Chuang et al. (3) studied the growth of horseradish peroxidase (HRP) leaks in the rat aortic endothelium as a function of HRP circulation time before death. They found rapid spot growth that, in hindsight, was not consistent with a diffusion-only model for any reasonable diffusivity and hinted at convection's role. Frank and Fogelman's (6) ultrarapid freezing/rotary shadow etchings revealed that the aortic intima was far sparser than the adjacent media, and Huang et al. (10) showed that this meant that it presented far less resistance to flow and tracer advection than the media. Their convection-diffusion theory needed to utilize this structural feature to explain Chuang et al.'s (3) data and showed that, indeed, convection dominated the macromolecular transport in the aortic wall.
In the present study, we focused exclusively on the rat model and asked which of these features carries over to the valve leaflet. We asked whether HRP can penetrate the rat's valvular endothelium uniformly, or just focally. In the latter case, larger molecules such as LDL, would, a fortiori, also only cross focally. We used HRP as tracer to interrogate details of the rat valve leaflet's structure and the related nature and magnitude of the quasi-steady (that is, averaged over the cardiac cycle) water convection patterns into and within the leaflet's matrix. It should be noted that, as in any tracer diagnostic test of structure or of steady flows (e.g., tracer studies to detect perturbed blood flow around faulty heart valves), such conclusions are not limited by the length of the tracer experiment, or the specific (passive) tracer. The experiments reported below directly address the three assumptions of Tompkins et al.'s model in the rat leaflet only. In the second paper in this series (Ref. 34; Part II, this issue)—not here—we shall guess that the monkey's valve leaflets have similarities to the rat's and attempt to justify this extrapolation by constructing a theoretical model that attempts to rationally explain, among others, Tompkins et al.'s studies (26), and their great variability, quantitatively with a single, unique set of parameters.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
We anesthetized healthy male Sprague-Dawley rats, weighing 300–400 g on a normal diet, intraperitoneally with 1% pentobarbital sodium solution (Sigma; 7 mg/100 g body wt) and cannulated both of the rat's femoral arteries and its left femoral vein with PE-50 polyethylene tubing. HRP (type II, Sigma; dissolved in 0.5 ml of normal saline and at a dosage level of 2 mg/100 g body wt) was injected through the femoral vein and allowed to circulate for 30 s in seven rats, 60 s in four rats, 120 s in four rats, and 240 s in five rats. One rat (control) received no HRP. Ten seconds before the termination of each circulation period, we injected 0.5 ml of heparin (Elkins-Sinn; 5,000 USP units/ml) through the femoral vein to prevent blood coagulation, followed by 2 ml of an overdose of pentobarbital to stop the heart. The chest was opened immediately. The heart was perfused through the left ventricle, punctured by a needle catheter with heparinized saline from a pressure reservoir set at the physiological pressure of 110 mmHg until clear fluid emerged from both femoral arteries, which served as egress sites for the perfusate. We then switched the perfusate to 30 ml of 2% glutaraldehyde. We harvested the heart and carefully dissected out the aortic valve's leaflets under a dissection microscope. The valvular leaflets were then processed either for en face study under light microscope (LM) or for an ultrastructural study of the subendothelial matrix under the transmission electron microscope (TEM) (3, 9).
A further control followed the above procedure through the 30-s HRP circulation. We perfused this rat with PBS pressurized to 100 mmHg in the reservoir through the right femoral vein to wash out the blood. When clear fluid replaced blood exiting through the femoral arteries (10 min), we cannulated the descending aorta, tied off the aortic bifurcations between the cannulation and the heart, and used PBS, pressurized at 100 mmHg, to close the valve and create a pressure gradient (HRP washout conditions) across it, for 1 h. We harvested the valve, fixed it, carried out the 3,3'-diaminobenzidine (DAB) reaction, and prepared slides as above.
En face study of valve leaflets under LM.
We incubated the valve leaflets that had been subjected to HRP circulation at 37°C for 1 h in a mixture of 0.05 M Tris·HCl buffer (pH 7.0) containing 45 mg of DAB tetrahydrochloride (Sigma) and 20 µl of 30% H2O2. In some experiments the valve leaflets were stained with Harris' hematoxylin (Sigma) for 45 s to mark the EC nuclei. We then mounted the valve leaflets onto a glass slide, covered them with coverslips, and examined them en face under LM (Olympus BX51). We calculated the effective radius R of an HRP spot by outlining the brown spot with the cursor, calculating the area A outlined, and setting R = (A/
)1/2.
Ultrastructural study of subendothelial matrix under TEM. After 30 s (1 rat) or 4 min (1 rat) of HRP circulation, we processed the valve leaflets for TEM examination. They were further fixed in 2% glutaraldehyde for 30 min and then in a mixture of 1% tannic acid and 1.5% potassium ferrocyanate for 1 h to obtain enhanced contrast with a modified Karnovsky method (11). After rinsing was completed, we postfixed the leaflets in 1% osmium tetroxide for 90 min. Subsequently, the leaflets were washed with distilled water, stained with 2% aqueous solution of uranyl acetate at 60°C for 15 h, dehydrated in a graded series of ethanol, infiltrated with propylene oxide, and embedded in Epon 812. The leaflets were properly oriented during embedding so as to be cut perpendicularly to the valvular endothelial surface. A diamond knife on an ultramicrotome (Reichert-Jung, Vienna, Austria) cut ribbons of ultrathin sections with a silver or pale gold reflection color. We collected them on Formvar-coated single-slot grids without poststaining and examined them in a JEM 1200EX electron microscope (JEOL, Tokyo, Japan).
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
Figure 2A shows how the radius of HRP leakage spots in the valve leaflet increases with HRP circulation time. Two separate experimenters independently did these experiments, and the results are quite consistent. The spots grow rapidly in the first minute and gradually in the following 3 min. Finally, after 4 min of HRP circulation, the spots fuse together, forming wide areas of diffuse HRP staining region. The striking feature of this HRP leakage spot growth curve for the aortic valve is how similar it is in shape and magnitude to (only 25% slower than) that in the thoracic aorta (3). Since the valve leaflet experiences a pressure gradient that can potentially drive a convective flow across it only when the valve is closed, and the valve is closed for roughly 70% of a cardiac cycle, Fig. 2B also replots the leaflet spot growth curve as a function of only the HRP circulation time during which the valve is closed. It should be noted that in the first minute the valve and aorta curves are extremely close, and then the former grows slightly slower. Since the tracer is injected into the femoral vein, there is a time lag before it reaches the blood surrounding the valve. Huang et al. (10) estimated this time as 25 s for the aorta by extending the theoretical spot size curves that agree with the experimental points back to a nondimensional radius of 1. Clearly, the delay for the valve should be almost identical, and one would expect a valve theory to similarly extend backward to radius 1 for a similar time.
The number of rats, about 20, in the present investigation was not large enough for detailed quantitative and statistical assessment of the spatial distribution and frequency of HRP spots. In general, we found no preference for HRP spots among the three leaflets of the same aortic valve in a normal rat. However, in a single leaflet, most of the spots tended to localize in the pressure-bearing part of the leaflet and near the line of coaptation where the ECs are more easily damaged or killed because of the recurring coaptation at the cardiac frequency. Figure 1B shows a typical distribution of spots in a leaflet. In addition, one could easily find incomplete spots along the sample's cutting edge, i.e., the line of attachment between the leaflet and the sinus of Valsalva. Of 47 leaflets examined, we found 81 HRP leaky spots and there were 0–3 HRP leaky spots per leaflet, including both aspects; since our observation was through a leaflet under LM, we did not distinguish on which aspect a leak appeared. Since human valve lesions occur almost exclusively on the pressure-bearing aortic face (29), Nievelstein-Post et al. (18) looked for lipid deposits mainly there. If convection, as we shall see, is responsible for the large spots, then one expects all of the larger spots to emanate from the aortic aspect, since convection drives tracer into aortic aspect leaks and out of the leaflet for ventricularis leaks. Any (rare) ventricularis leaks would tend to lower the average spot size at each circulation time relative to only aortic face leaks.
Figure 3A shows the TEM examination of the traverse sections of the valve leaflet after 4 min of HRP circulation, serially sliced until we encountered regions associated with ECs whose junctions were leaky to HRP. The lumen is on the upper left, and an EC is aligned diagonally. The intense black color, representing HRP, appears particularly strong in a very thin region (roughly 150 nm thick) directly under ECs in Fig. 3A (but not in the control, Fig. 3B), followed by an abrupt change to a much lighter color, indicating that HRP concentration is much higher in this subendothelial thin layer. We have many similar figures from this rat and others from the rat exposed to only 30-s HRP circulation. The sections from the HRP-exposed rats that were not near localized leaks resemble those of the control.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Is molecular diffusion the only important mass transport mechanism in the rat valve leaflet?
Because Chuang et al.'s observations (3) of the rate of HRP spot growth in the rat aorta led to important insights into the mechanism of mass transport in the rat's artery wall, we performed a similar study, with the same HRP circulation times of 30, 60, 120, and 240 s, on the rat's aortic valve leaflet. En face LM examination of HRP spots in the aortic valve (Fig. 2) showed growth reflecting a continuous influx of HRP across leaky EC junctions and subsequent transport, strikingly similar in rate and extent to that in rat aorta (3), which had a large component parallel to the endothelium in the subendothelial matrix. The valve's spots reached a radius of 75 µm in 4 min, but the valve was closed, experiencing a pressure difference across it of 90 mmHg for only 70% of those 4 min and nearly 0 mmHg for the balance. In contrast, the transmural aortic pressure cycles from 80 to 120 mmHg over the cardiac cycle. [In both the leaflet and the aorta, the transendothelial pressure drop is a fixed fraction of the total drop across the entire tissue. In the aorta, this fraction is 1/2 (25) for these pressures; a value for this fraction in the valve is determined in Part II.] Figure 2B also plots a curve that rescales the time axis for the valve's spot growth to take into account only the time the valve was closed. Its similarity to the aortic spot curve is even more striking, strongly suggesting that HRP transport through the leaky junction and in the subendothelial space in heart valves and in arteries may be similar. Huang et al. (10) showed that only convection in the subendothelial space with a strong component parallel to the endothelium in a subendothelial intima that was much sparser and therefore presented far less flow (and advection) resistance than the media could account for the observed rapid HRP spot growth rates for short circulation times in the rat aorta. Thus the rapid short-time HRP spot growth in the rat valve leaflet is a way to interrogate the quasi-steady pulsatile convection in the leaflet. It indicates that the pressure drop across the leaflet when the valve is closed likely drives a (pulsatile) convective flow through normal junctions and through rare, isolated endothelial leaks across the aortic aspect's endothelium. The mismatch in water and permeability between the leaky and normal junctions then likely causes a subendothelial flow near the leak that is mainly parallel to the endothelium and pointing away from the leak. This parallel flow becomes negligible further from the leak and, augmented by the slower water flow through normal junctions, becomes normal to the endothelium. If a passive tracer is present, this flow advects it through releak and in a way to produce spots that initially grow quickly and then level off, as observed. A rational model for water and macromolecular transport in the rat leaflet must allow for this possibility.
Can the interposed matrix be treated as a single, homogeneous, isotropic layer? In a first attempt to explain HRP spot growth in the artery, Yuan et al. (33) proposed that the internal elastic layer (IEL) functions as the transport barrier that forces fluid crossing the endothelium to flow parallel to the endothelium in the arterial intima, thereby creating large tracer spots, rather than proceeding directly into the media. This assumption alone, under the presumption that the intima and media structures have similar intrinsic flow resistances, led to tracer spots that were far smaller than observation (10, 33). Huang et al. (10) used ultrarapid freezing/rotary shadow etchings (6) that showed an extremely sparse intimal structure in an ab initio theory for the intimal transport parameters. They found that the intimal resistance is far lower than the media's and that these nonuniformities lead to rapid intimal convection parallel to the endothelium, i.e., to tracer spot sizes in agreement with observation. Since the aortic valve leaflet apparently has no IEL, it is natural to ask whether the valve has a layer analogous to the artery's subendothelial intima and, if so, if one needs to account for it. Or can a model that assumes a uniform, isotropic interposed matrix without any internal barriers such as an IEL, as in Tompkins et al.'s 1D model (26), but made 2D and including convection, account for the observed spot growth? One can reason that, in such a model, HRP traversing an endothelial leak would spread at least as fast in the direction normal (the direction of the overall transleaflet pressure drop) to the endothelial surface as parallel to it, as Yuan et al.'s arterial study (33) found. One would thus expect it to yield significantly slower HRP spot growth for the valve than for the artery, in contrast to our observation (Ref. 3 and Fig. 2).
Is it possible that the valve leaflet, like the arterial wall, contains a very thin and sparse intima-like layer directly underneath the endothelium, whose flow resistance is much lower than the balance of the interposed region? If so, could such a region, even in the absence of an IEL, induce a convective flow to proceed preferentially in the direction parallel to the endothelium, and thus explain the observed HRP spot growth in the valve?
Figure 3 provides evidence to support the first of these hypotheses. It shows a TEM of a transverse section of a leaflet through a localized HRP leak after 4-min circulation. There is a very thin, 150-nm, intensely black HRP layer, indicating a very high HRP concentration, immediately beneath the ECs, followed by an abrupt change to a much lighter color. Because a certain minimum level of peroxidase is needed to be detectable, the lighter areas need not be free of HRP or other, naturally occurring peroxidases, but their concentrations probably fall below this threshold. If the valve's subendothelial matrix's HRP-void space were uniform, one would have expected to observe a gradual decrease of HRP reaction product intensity with increasing depth into the valve's matrix. Instead, this abrupt change immediately beneath the endothelium and the dark intensity in this thin region appear to indicate the presence of a thin, sparse arterial subendothelial intima-like layer that, in analogy to the aorta, likely has a much higher void space than the matrix below it. Consequently its permeability and effective macromolecular diffusion coefficient are likely much larger than those of the deeper layers. These conditions would favor lateral convective transport in this region as in the aortic subendothelial intima. Finally, the thickness of this very thin layer, roughly 150 nm, is comparable to that of the arterial subendothelial intima in the normal rat aorta, 100–500 nm (10). Again, this is a structural conclusion that is independent of the HRP circulation time used as the diagnostic.
Further evidence for this proposition comes from Nievelstein-Post et al.'s (18) clear electron micrograph of an ultrarapid freezing/rotary shadow etching of a normal rabbit's cardiac valve's subendothelial space. Figure 4 shows that the subendothelial extracellular matrix immediately adjacent to the EC membrane, the diagonal middle layer in the figure, is much sparser than the matrix toward the upper right area of the figure that is further into the valve. This micrograph of the atrioventricular valve's immediate subendothelial layer is surprisingly similar to that of the rabbit aortic subendothelial intima in Ref. 6 (see, in particular, Fig. 5 of Ref. 6), indicating the likelihood that these similar structures have similar properties. In particular, the typical spacing between the dominant fibers and collagens is 30–40 nm in Fig. 4, which is consistent with that in the aortic intima (6, 10, 18). [In Ref. 10 we argue that the freeze etchings in Refs. 6 and 18 do not show the glycosaminoglycans (GAGs), but our calculations indicate that the GAGs do not act to hinder transport in the aortic intima. The same is likely true for Fig. 4 as well.] The apparent (because the crack plane's angle is unknown, it is only an upper bound) thickness of this layer is 200 nm, remarkably consistent with that estimated from our TEM examinations of the rat valve (Fig. 3).
|
|
1 h (3, 30)].
The above evidence strongly supports the hypothesis of a sparse, 150- to 200-nm-thick subendothelial intima-like layer in the rat (and rabbit) valve leaflets, with much higher void space, Darcy permeability, and effective tracer diffusion coefficient than the balance of the matrix in the interposed region. In analogy to the artery, we call this layer, composed of a basement membrane plus loose connective tissue, the "valvular subendothelial (to exclude the endothelium) intima" despite the fact that, unlike the arterial intima, this region is not bounded by a continuous IEL. Although the valvular subendothelial intima comprises a tiny fraction of the 15-µm-thick valve leaflet, it may, in analogy to the arterial subendothelial intima, play a critical role in macromolecular transport into and in the concentration distributions within the valve. Our new theoretical models in Part II investigate how its inclusion affects the prediction of short-term tracer spot growth in heart valves as observed, e.g., in the present paper. Figure 5, a diagram of the cross section of the valve leaflet, indicates the structural features that we have determined.
To date, the existence of a valvular intima has been noted by Nievelstein-Post et al. (18), Haberland et al. (7), and Simionescu et al. (21), but no one seems to have yet addressed its structure, and therefore its importance in macromolecular transport in the cardiac valves. The likely reasons why it has escaped study are that 1) unlike large arteries, the heart valve has no IEL to highlight the existence of its intima; 2) there has been little quantitative study of short-time macromolecular transport in heart valves, where the importance of the valvular intima is most obvious; and 3) it is natural to suppose that such a thin layer cannot strongly influence the overall transport processes, especially for biologically interesting long times. As we shall see in Part II, this latter line of reasoning may be too facile.
In summary, we have used short-time tracer diagnostics and TEM to draw conclusions about the detailed structure of the rat valve leaflet and how it affects the quasi-steady pulsatile water convection into and within the rat's valve leaflet. This flow appears strong enough to determine the fate of a passive macromolecular tracer in the rat's valve leaflet. These structural and steady-state conclusions are independent of the duration of the diagnostic, and this work and its conclusions have focused exclusively on the rat. Part II hypothesizes that the squirrel monkey's leaflets share these features with the rat's leaflet and constructs a 2D convection-diffusion model for a leaflet with subendothelial intimae (as in Fig. 5) to try to explain, among other experiments, all of Tompkins et al.'s (26) disparate squirrel monkey transvalvular LDL profiles with a single set of parameters, most of which are measured and the three remaining ones fit once and for all from a single curve.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Present address of Y. Yin: Dept. of Pathology, Albert Einstein College of Medicine, Bronx, NY 10461.
|
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 |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  Z. Zeng, Y. Yin, K.-M. Jan, and D. S. Rumschitzki Macromolecular transport in heart valves. II. Theoretical models Am J Physiol Heart Circ Physiol, June 1, 2007; 292(6): H2671 - H2686. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. Zeng, P. Nievelstein-Post, Y. Yin, K.-M. Jan, J. S. Frank, and D. S. Rumschitzki Macromolecular transport in heart valves. III. Experiment and theory for the size distribution of extracellular liposomes in hyperlipidemic rabbits Am J Physiol Heart Circ Physiol, June 1, 2007; 292(6): H2687 - H2697. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
