- Email: [email protected]
Capillary Density and Leukocyte Adhesion in Hamsters with Hereditary Cardiomyopathy
Microvascular Research 56, 85–94 (1998) Article No. MR982087
Capillary Density and Leukocyte Adhesion in Hamsters with Hereditary Cardiomyopathy A. Colantuoni, G. Coppini, and S. Bertuglia CNR Institute of Clinical Physiology, University of Pisa, Pisa, Italy Received September 15, 1997
The aim of this study was to characterize microvascular networks in cheek pouch of cardiomyopathic Syrian hamster (CM) (Bio 14.6), which is an interesting model of idiopathic cardiomyopathy and congestive heart failure. Microcirculation was visualized by fluorescence microscopy. Diameter and length of arterioles, classified according to centrifugal ordering scheme, were measured. A computational method was arranged to determine the density of arterioles and capillaries (total vessel length per unit area, cm21), fractal dimension of capillaries, and the associated Voronoi tesselation. Furthermore, leukocyte adhesion to venules and arteriolar reactivity to drugs were studied. Increase in the number of terminal arterioles and capillary rarefication characterized CM microvasculature compared with that of age-matched controls (58 6 7 versus 25 6 5 cm21, and 128 6 15 versus 240 6 10 cm21, respectively). Fractal dimension of capillaries was reduced in CM compared with controls (1.40 6 0.10 versus 1.85 6 0.09) and associated with increased avascular spaces, as shown by Voronoi tesselation results. Leukocyte adhesion to venules increased significantly in CM. In CM responsiveness of arterioles to nitric oxide inhibition and propranolol was slighter but more marked to norepinephrine and angiotensin II compared with that of control hamsters. In conclusion, the different geometry, increased leukocyte adhesion, and altered arterial responsiveness may contribute to flow disturbances in the microcirculation of CM hamsters. © 1998 Academic Press 0026-2862/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Key Words: microcirculation; cardiomyopathy; capillary density; fractal dimension; leukocyte.
INTRODUCTION The cardiomyopathic (CM) Syrian hamsters (strain BIO 14.6) are known to develop a genetically determined cardiomyopathy associated with progressive development of congestive heart failure (Homburger, 1979; Factor et al., 1982; Finkel et al., 1987, 1992; Ikegaya et al., 1992). These animals have a cardiac and skeletal myopathy characterized by progressive intracellular calcium overload, followed by histologic necrosis at 4 – 6 months and death at 10 –12 months of age (Kagiya et al., 1991; Sapp et al., 1994; Pogue et al., 1995). Cardiomyopathy might involve dysfunctions within the tissues with changes in local microvascular and rheological conditions. However, there are no studies documenting microvascular network design in animals with chronic cardiomyopathy or functional changes in the vessels of microcirculation. Namely, network morphology with arteriolar and capillary spatial density determines distribution of blood flow in microvasculature (Groom et al., 1986) and interferes with blood-tissue transports. Capillary density is the simplest indicator of the distribution of vessels and tissue perfusion (Bertuglia et al., 1993; Nolte et al., 1995). Furthermore, many reports evidenced the role of 85
86
leukocytes in the damage occurring in the tissue in ischemic and reperfused conditions (Arnould et al., 1993; Bertuglia et al., 1995). Many studies suggest that accumulation of leukocytes in in vitro incubation of cultured endothelial cells under hypoxia already develops (Bradury et al., 1993). However, it is not elucidated to what extent, these leukocyte abnormalities develop at the level of microcirculation in cardiomyopathy. The first aim of the study was to describe the structure of CM microvascular networks compared with that of control hamsters (C) using the cheek pouch model. We modified a computational method to measure arteriolar and capillary spatial density, the boxcounting dimension of capillary networks, which is an estimate of the fractal dimension, and the associated Voronoi tesselation to quantitate local geometrical relationships among vessels and the embedding space (Baish et al., 1996). This approach permits evaluation of the related region for a given vessel segment, i.e., the region perfused mainly by that vessel. We also evaluated the number of adherent leukocytes in venules and the changes in arteriolar responses to NG-monomethyl-l-arginine (l-NMMA), a nitric oxide (NO) inhibitor, norepinephrine, propranolol, a nonselective b1–b2 adrenergic antagonist, and angiotensin II in CM hamsters.
Colantuoni, Coppini, and Bertuglia
Observations were made with a Leitz Orthoplan microscope (Leitz, Germany), operating in incident light with a Ploemopak filter block, fitted with a long-working distance objective (34, n.a. 0.14; 320, n.a. 0.25) and 310 eyepiece. Epiillumination was provided by a xenon 150 W lamp source in conjunction with the appropriate filters for fluorescein isothiocyanate bound to dextran MW 150,000 (Leitz I2 Ploemopak filter block) and a heat filter (Leitz KG 1). The tracer was injected intravenously (50 mg/ 100 g body wt., as 5% solution) and followed by a COHU 5253 SIT low light level camera. The scenes were monitored by a Sony PVM 122 CE monitor and recorded by a Sony U-Matic VO 5800 PS video recorder. The images of interest were digitized by a computer-assisted technique.
Network Analysis Arteriolar networks were traced by microscopic observation (microscope long-working distance objectives: 34 and 320) following vessels from the input arterioles to capillaries. Vessels were assigned centrifugal orders, beginning with the largest arterioles classified as order 1 (A1). Hamster cheek pouch was observed for a period of 30 – 60 min to record network geometry.
Computational Method
MATERIALS AND METHODS Eight-month-old male cardiomyopathic hamsters (BIO 14.6, biobreeders, Fitchburg, MA) weighing 80 – 100 g (n 5 15) and age-matched male Syrian hamsters (Charles River, Calco, CO, Italy) weighing 80 –100 g were used (n 5 10) to characterize the microvascular network. The cheek pouch was prepared, as previously reported, to visualize the microvascular network (Bertuglia et al., 1993). Briefly, the membrane was gently everted and fixed to a special stage of the microscope: a very thin black blade was inserted through a small incision between the upper and lower layers of the pouch. The membrane was superfused with a 36 6 0.5°C Ringer solution, with 5% CO2 in 95% N2, adjusted to pH 7.35. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Regions of interest were interactively selected from acquired images to include portions of the microvascular network under investigation. The digital pictures were properly resampled to exploit the available information. We adopted a bilinear interpolation algorithm which provided a reasonable compromise between goodness of fit and computational load to achieve subpixel accuracy in the assessment of vessel size. A 4:1 resampling ratio was used, resulting in a pixel size of about 1 mm. Afterwards, the processing procedure included three steps: (1) segmentation of vascular structures; (2) representation of segmented vessel by proper attributed graphs; (3) parameter extraction. Segmentation was based on the assumption that vessels are usually bright, elongated (bar-like) struc-
87
Microvascular Networks in Cardiomyopathic Hamsters
tures embedded in a slow-varying darker background. To remove background and enhance vessels, images were filtered by a difference of Gaussian (DoG) kernel: DoG(x,y) 5
F
x2 1 y2 1 exp 2 2 ps 21 2 s 21
G
F
G
x2 1 y2 1 2 exp 2 , 2 ps 22 2 s 22 which is widely used for several computer vision tasks. The parameter s1 is chosen slightly smaller than the expected vessel size and s2 should have a larger value (a s1 value ranging from 3 to 5 times s2 is an adequate choice). The original picture of a normal vascular network and the related filter output are reported in Figs. 1a and 1b, respectively. Following DoG filtering, the images were segmented using an adaptive thresholding algorithm based on competitive learning. A segmented picture is shown in Fig. 1c. By processing the segmented picture, the second stage provided a compact description of the imaged network by an attributed graph whose nodes are bifurcations and links between nodes are vessel segments. To build a representation, we extracted the skeleton of the network by iterated morphological thinning. The adopted algorithm, modified by Chin et al. (1990), exhibits good noise immunity in comparison with alternative methods. Furthermore, thinning operators allow a local estimate of vascular thickness. For example, in the case of a perfectly straight vessel segment, at each thinning step two pixels (one pixel per side) were removed until a further thinning step resulted in vessel disappearing. Thus, for an n pixel structure, the number nth of needed thinning steps satisfied the equation: 2n # nth # 2n 1 1. The quantity nth was assumed as a measure of local vessel size and named morphological thinning thickness (MTT). Thus a bidimensional map of pixels representing the vascular axis was obtained, each pixel being labeled by related MTT value. Afterwards, vessel segments were extracted by linking connected skeleton points between consecutive branchings. Each vessel segment (graph node) was characterized by length, mean MTT, and position (as given by the coordinates of the first and last point, respectively). In addition, for each node a list of ves-
sel-point coordinates with the local MTT was stored. It is worth noting that recorded vessels can originate (and terminate) at a natural branching or at an obstruction as well. Therefore, the network-representing graph could be disconnected and related to vessel patency rather than to the topology of its connection pattern. Utilizing the network graph, terminal arterioles, capillaries, and venules were divided in three groups according to the mean MTT (Fig. 1d). We computed the related area-densities for the sets of arterioles (da) and capillaries (dc), defined as the total vessel length per unit area (expressed in cm21). The box-counting dimension Df was used for the capillary network as an estimate of the fractal dimension. Finally, the Voronoi tesselation associated with the capillary net was computed. Thus, the tissue space was partitioned into regions by assigning each point to the nearest vessel segment.
Leukocyte Adhesion To stain leukocytes the animals received an intravenous injection of acridine red (Chroma, Stuttgart, FRG) (1 mg/100 g body wt in 0.2 ml/100 g/h). Appropriate filters for acridine red (Leitz N2 Ploemopak filter block) were used for microscopic observations. Postcapillary venules with diameters of 16 6 8 mm and lengths ,250 mm were selected for the study. Sticking leukocytes were expressed as number per 100 mm of vessel length, during 30 s of observation.
Arteriolar Responses Two subgroups (n 5 5) of C and CM hamsters were treated with l-NMMA (1 mg/100 g body wt in 500 ml of 0.9% saline) (Bachem, Switzerland), norepinephrine (NE) (0.25 mg/100 g body wt) (Sigma, St. Louis, MO), propranolol (PP) (0.2 mg/100 g body wt) (Sigma), angiotensin II (Ang II) (1 mg/100 g body wt) (Sigma), administered intravenously over 3–5 min. The drugs were randomly injected: the vessels were reexposed to vasoactive agents after a 45- to 60-min equilibration period sufficient to regain control values. RBC velocity (VRBC) and vessel diameter were measured by a dual window technique (102 B IPM velocimeter; IPM, San Diego, CA) and by image shearing Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
88
Colantuoni, Coppini, and Bertuglia
TABLE 1 Mean Diameter, Number of Vessels Observed (n), and Lengths of Arterioles (A1, A2, A3, A4) in the Microvascular Networks of Control (C) and Cardiomyopathic (CM) Hamsters Arteriolar order Control A1 A2 A3 A4 CM A1 A2 A3 A4
Mean diameter (mm)
n
Mean length (mm)
75.4 6 6.5 26.6 6 5.2 16.3 6 4.1 8.2 6 1.1
10 10 23 115
2050 6 150 1380 6 126 827 6 105 219 6 48
67.5 6 6.4 26.8 6 5.3 15.2 6 4.6 8.5 6 1.2
10 10 26 234*
2250 6 130 1550 6 250 795 6 124 141 6 45*
* P , 0.01 compared with C group. Each order is significantly different when compared with previous order (P , 0.01).
(Image Shearing Monitor Mod 907; IPM, San Diego, CA), respectively. Mean arterial blood pressure (Viggo-Spectramed P10E2 transducer, Oxnard, CA, connected to a catheter in the femoral artery), respiratory and heart rates were monitored by a Gould Windograf recorder (Mod. 13-6615-10S Gould Inc. OH). Data were recorded and stored in a computer. The Gaussian distribution fit was examined by Kolmogorov–Smirnov test. When data presented normal distribution we used Student’s t test; for nonparametric testing of null hypothesis Mann–Whitney U test was used. To determine which groups were statistically different, ANOVA, Scheffe`’s post hoc, or Kruskal–Wallis tests were employed. To compare Voronoi tesselation results we used chi-square test. The reported values are means 6 SD. Differences were considered significant at P , 0.05.
RESULTS Network Analysis Table 1 lists the geometrical features of the microvessels of both C and CM hamsters. Microvascular networks of control and CM hamsters were characterized by four orders of arterioles (A1–A4) according to centrifugal ordering scheme (Fig. 1). Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
CM showed an increased number of A4 terminal arterioles branching from A3 parent vessels (Table 1). Consequently, the mean distance between the origin of A4 arterioles along A3 vessels was significantly different compared with that detected in controls (75 6 35 mm versus 236 6 57 mm, P , 0.01). Terminal microvasculature was characterized by the presence of direct arteriolar–venular conduits with higher VRBC when compared with capillary VRBC. VRBC in postcapillary venules of CM (n 5 20) was significantly lower than that of venules in C (n 5 20) hamsters (0.25 6 0.10 versus 0.45 6 0.05 mm/s).
Computational Method The arteriolar and capillary densities were 25 6 5 cm21, 240 6 10 cm21 in C (n 5 10) and 58 6 7 cm21, 128 6 15 cm21 in CM (n 5 10) hamsters, respectively. The capillary fractal dimension was significantly different, 1.85 6 0.09 in C and 1.40 6 0.10 in CM hamsters. Two examples of computed capillary networks superimposed on their original pictures are shown in Fig. 2. A control (Df 5 1.78, dc 5 244 cm21) and a CM network (Df 5 1.45, dc 5 111 cm21) are reported in Figs. 2a and 2b, respectively. The space-filling behavior of a network can be further described by the Voronoi tesselation. In Fig. 3 we plot the total fraction of tissue points against their distance to the nearest vessel. In control cases the distance to the nearest vessel was ,100 mm for more than 90% of tissue points. The percentage was significantly reduced to ,80% in CM hamsters (P , 0.01).
Leukocyte Adhesion CM hamsters demonstrated a significant difference in the number of adherent leukocytes in CM (n 5 40) when compared with C (n 5 30) animals (9.5 6 1.7 versus 2.8 6 0.6/100 mm venular length) (P , 0.01) (Table 2).
Arteriolar Responses In CM hamsters the mean arterial blood pressure (MAP) was significantly lower than that measured in C animals (80 6 4 versus 97 6 5 mm Hg).
Microvascular Networks in Cardiomyopathic Hamsters
89
FIG. 1. Processing phases of a control microvasculature. a) original image, b) DoG filtering, c) thresholding, d) morphological thinning and vessel grouping into arterioles (black thick lines), capillaries (thin lines) and venules (gray thick lines). See text for details.
In presence of l-NMMA the mean diameter of A2 and A4 arterioles decreased significantly in C and CM hamsters (Table 3). In CM hamsters, however, there was a slighter vasoconstriction. A moderate rise in MAP was observed.
With NE there was a significant vasoconstriction in C and CM hamsters, but in CM it was significantly higher. With PP there was a slight arteriolar response in CM hamsters which was significantly different compared Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
90
Colantuoni, Coppini, and Bertuglia
FIG. 2. Computed capillary networks overlaid on the original pictures. a) control (Df 5 1.78, dc 5 244 cm21); b) cardiomyopathy (Df 5 1.45, dc 5 111 cm21).
with that of C hamsters. MAP did not change significantly in CM hamsters. Ang II produced a significantly higher contractile response in CM hamsters compared with that observed in
C hamsters and was sustained for at least 3 min. Ang II produced a dose-dependent increase in MAP (98 6 1.2 mm Hg, P , 0.05) that returned to the initial resting value within 4 6 1 min.
FIG. 3. The fraction of tissue points (cumulative frequency) plotted against their distance to the nearest vessel for control (C) and cardiomyopathic (CM) hamsters, respectively. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
91
Microvascular Networks in Cardiomyopathic Hamsters
TABLE 2 Mean Arterial Blood Pressure (MAP), Arteriolar and Capillary Spatial Densities (da, dc), Fractal Dimension (Df), Number of Adherent Leukocytes per Vessel Length (L/100 mm), Capillary Red Blood Cell Velocity (VRBC) in Control (C) and Cardiomyopathic (CM) Hamsters Group
MAP (mm/Hg)
da (cm21)
dc (cm21)
Df
L/100 mm
C n CM n
97 6 5 10 80 6 4* 15
25 6 5 10 58 6 7* 10
240 6 10 10 128 6 15* 10
1.85 6 0.09 10 1.40 6 0.10* 10
2.8 6 0.6 30 9.5 6 1.7* 40
n
VRBC (mm/s) 0.22 6 0.04 20 0.20 6 0.03 40 0.35 6 0.05*a 20
a
VRBC in A–V conduits. Note. Values are mean 6 SD. n 5 number of vessels (L/100 mm; VRBC); in the other cases number of animals. * P , 0.01 compared with C hamsters.
DISCUSSION This study demonstrates that abnormalities in microvascular geometry are detectable in hamsters with chronic cardiomyopathy. CM hamsters showed a significant capillary rarefication, partially compensated by the increase in terminal arteriole spatial density, and an increased number of adherent leukocytes to venules. Furthermore, in CM animals we showed an impaired response of arterioles to propranolol, meanwhile endothelial responses appear to be depressed. Vasoconstriction induced by norepinephrine and angiotensin II was more marked. We observed that the total length of terminal arterioles was significantly higher in CM than in C hamsters. Furthermore, an increased number of terminal arterioles was related to a reduced number of capillaries that eventually appeared to be grouped
in small loops. Furthermore, there were arteriolar– venular conduits characterized by increased blood flow velocity. Computer measurements permitted an interesting analysis of the geometrical and functional arrangements of microvessels. Capillary density can be quantitated by current morphological analysis and descriptive statistics without information on the spatial heterogeneity of microvessels. We found a significant correspondence between morphological and computational network analyses about identification of microvessels. Nevertheless, the cheek pouch is a suitable model for two-dimensional network reconstruction. The method of estimating the fractal dimension by covering the structure to be analyzed with boxes gives a measure of the complexity of the system over space. Therefore, the fractal dimension is an estimate of the related space-filling properties, which plays a relevant role for microvascular blood flow distribution.
TABLE 3 Arteriolar Responses (Percentage Decrease of Baseline Diameter) to l-NMMA, Norepinephrine (NE), Propranolol (PP), and Angiotensin II (Ang II) in Control (C) and Cardiomyopathic (CM) Hamsters Group
Control diameter (mm)
l-NMMA
NE
25.8 6 2.3 8.0 6 1.0
221 6 1% 237 6 4%
240 6 3% 250 6 2%
238 6 5% 240 6 3%
250 6 4% 260 6 3%
26.1 6 2.8 8.4 6 1.3
218 6 1%* 228 6 1%*
280 6 10%* 295 6 5%*
218 6 5%* 24 6 0.2%*
275 6 2%* 287 6 2%*
PP
Ang II
C A2 A4 CM A2 A4
Note. Values are mean 6 SD. n 5 7 for each entry. * P , 0.01 compared with C. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
92
CM hamsters exhibited significant capillary rarefication, reduction of fractal dimension associated with increased avascular spaces, as shown by Voronoi tesselation results, indicating a severe impairment of peripheral perfusion. The structure of network and the effects of mean vascular spacing are important functional parameters on oxygen transport (Pittman, 1995). Conversely, the fractal dimension had higher values, relatively close to the topological dimension of the imaged embedding space in normal hamsters. This observation suggests a plane-filling behavior of the capillary network, in accordance with the findings by Baish et al. (1996). The observed effect of capillary rarefication on overall network resistance is similar to the phenomenon known as percolation studied in computer simulations (Hudetz, 1993; Baish et al., 1996). Thus, the observed reduction of capillary spatial density of about 40% may suggest that CM networks are near to (or below) the percolation threshold. Therefore, sudden increase in arteriolar resistance may determine easily impairment of either nutritive flow or blood flow distribution in the tissues. Indeed, in comparison with normal hamsters the CM vascular architecture appears to be altered with larger spaces among capillaries; therefore, flow can be less uniformly distributed than in a normal microvasculature. Furthermore, there are loops of grouped capillaries and particular conduits between arterioles and venules, so that areas with different resistance and blood flow distribution in the networks are present. These conduits may be particular adaptive changes to increased resistance due to rarefication of capillaries and proliferation of terminal arterioles. It is interesting to note that a number of studies have reported a reduction in the density of microvessels in many animal models of hypertension and in human hypertensive subjects. Ultrastructural studies of the cremaster muscle of rats with chronic reduced renal mass have revealed that microvascular rarefication is mediated by atrophy and structural degeneration of microvessels (Hansen-Smith et al., 1990). In this case the reduction in microvessels is mediated by degenerative changes of endothelial and vascular smooth muscle cells but there are no data on changes in tissue geometry. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Colantuoni, Coppini, and Bertuglia
In CM hamsters we have chronic conditions of low arterial pressure that might cause structural changes in the geometry of the network. These structural changes with formation of new arteriolar vessels may be indicative of a compensatory regulation of arteriolar tone and thereby this microvascular arrangement regulates tissue blood flow and vascular resistance. Thus, heart insufficiency correlated with microvascular impairments may induce chronic conditions of tissue hypoxia. Furthermore, there is an increased leukocyte adhesiveness to venular walls, suggesting impaired mechanism of leukocyte activation and/or altered endothelium responsiveness. Despite many reports with evidence of accumulation of leukocytes in ischemia reperfusion and that demonstrate their role in the damage occurring in the tissue (Bradury et al., 1993; Bertuglia et al., 1995) the initial adhesiveness remains unclear. Adherence is a crucial step in the process of the following infiltration and margination of leukocytes through the endothelium. This process is the result of complex interaction between adhesion molecules present on both the endothelium and the leukocytes. It might be hypothesized that probably there is an increased expression of endothelial adhesion molecules during chronic conditions of hypoxia (Sakata, 1993). Regarding the exact mechanism by which this phenomenon occurs we could suggest that hypotension decreases shear rate in vessels, causing reduction in NO formation and increased number of adherent leukocytes. Indeed, we observed many patterns of flow in postcapillary venules and some were characterized by very low flow compared with that of venules in C hamsters. Indeed, venules directly connected to A-V conduits showed a VRBC lower than that observed in C hamsters. These data on venular VRBC and disturbances of flow support the hypothesis that the reduction in shear rate may be involved in the increased leukocyte adhesion on venular walls. Arnould et al. (1995) found that hypoxia induces neutrophil adherence to umbelical vein endothelium. Thus hypoxic tissue may initiate an inflammatory response like the one reported in ischemic organs, which is then further worsened during developing cardiomyopathy. Furthermore, leukocytes accumulate in the tissue causing capillary plugging that may contribute to the
93
Microvascular Networks in Cardiomyopathic Hamsters
progressive reduction of capillary perfusion. However, the capillaries appeared to be perfused under baseline conditions and we did not observe leukocyte– capillary plugging during the periods of observation in CM hamsters. In chronic congestive heart failure most studies on the peripheral circulation have been confined on muscle circulation and skin resistance arteries, with impaired contraction and relaxation (Angus et al., 1993). Namely, congestive heart failure is associated with abnormalities in b-adrenoceptor function rather than in the number of receptors (Kessler et al., 1989; Cai et al., 1993). Urasawa et al. (1996) studied b-adrenergic receptor in the heart of CM hamsters. The expression was significantly enhanced and contractility of the heart decreased, suggesting that intracellular homeostasis against accelerated stimulation by catecholamines was maintained by a feed back mechanism of b-adrenergic receptor phosphorylation. Furthermore, Larosa and Forster (1996) found that coronary vascular tone via b-adrenoceptor stimulation is modulated also by endothelium in congestive heart failure. Our data show that the arterioles are less sensitive to vasodilatory stimuli acting on b-adrenoceptors but the endothelium depression is marked. Previously, it was found that NO has a fundamental role in determining arteriolar responses and capillary perfusion in ischemia (Kurose et al., 1994; Bertuglia et al., 1995). Lambert et al. (1995) found that Ang II receptor (AT1 receptor) is upregulated in the ventricles of CM hamsters, suggesting that this may play a role in the genesis and maintenance of the cardiac hypertrophy in the hamsters. We found an increased tendency to constriction in arterioles to Ang II, strongly suggesting similar abnormalities. A probable explanation is that significant alterations in the production or in arteriolar responsiveness to NO may lead to increased tendency to vasoconstriction in CM hamsters. Therefore, it may be suggested that NO exerts a fundamental role in vasodilatory responses under conditions of chronic hypoxia where the preservation and control of flow are critical. In CM hamsters as a consequence of depressed endothelial response and increased tendency to vasoconstriction of arterioles there is a reduced pressure trans-
mitted downstream to distal arterioles. Therefore, the chronic conditions of low pressure and low shear stress may determine a progressive rarefication in the capillary network. The increase in the number of terminal arterioles, the capillary rarefication, and the increased leukocyte adherence might be suggested as determinant factors involved in the disruption of microvascular homeostasis which occur in CM hamsters. Furthermore, the altered arterial responsiveness may contribute importantly to microvascular flow disturbances and play a potential role in the pathogenesis of cardiac dysfunction and heart failure. In conclusion, we have shown evident changes in the microvascular network in CM hamsters, namely in capillary perfusion, leukocyte adhesiveness, and arteriolar responsiveness that might lead to the impairment of the microcirculation.
REFERENCES
Angus, J. A., Ferrier, C. P., Sudhir, K., Kaye, D. M., and Jennings, G. L. (1993). Impaired contraction and relaxation in skin resistance arteries from patients with congestive heart failure. Cardiovasc. Res. 27, 204 –210. Arnould, T., Michiels, C., and Remacle, J. (1993). Increased PMN adherence in endothelial cells after hypoxia: Involvement of PAF, CD18/CD11b and ICAM-1. Am. J. Physiol. 264, C1102–1110. Arnould, T., Michiels, C., Janssens, D., Delaive, E., and Remacle, J. (1995). Hypoxia induces PMN adherence to umbilical vein endothelium. Cardiovasc. Res. 30, 1009 –1016. Baish, J. W., Gazit, Y., Berk, D. A., Nozue, M., Baxter, L. T., and Jain, R. K. (1996). Role of tumor vascular architecture in nutrient and drug delivery: An invasion percolation based network model. Microvasc. Res. 51, 327–346. Bertuglia, S., Colantuoni, A., and Intaglietta, M. (1993). Effect of leukocyte adhesion and microvascular permeability on capillary perfusion during ischemia reperfusion injury in hamster cheek pouch. Int. J. Microcirc. Clin. Exp. 13, 1–28. Bertuglia, S., Colantuoni, A., and Intaglietta, M. (1995). Treatment with L-arginine, L-NMMA, L-NNA in ischemia reperfusion injury in cheek pouch microvasculature. Microvasc. Res. 50, 162–175. Bradury, A. W., Murie, J. A., and Ruckley, C. V. (1993). Role of the leukocyte in the pathogenesis of vascular disease. Br. J. Surg. 80, 1503–1512. Cai, J. J., Hongwei, Y. U., and Lee, H. C. (1993). Developmental changes of protein kinase C and G in hypertrophic cardiomyopathic hamster hearts. J. Lab. Clin. Med. 122, 533–541. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
94 Chin, R. T., and Jang, B. (1990). Analysis of thinning algorithm using mathematical morphology. IEEE Trans. Pattern Anal. Machine Intell. 12, 541–551. Factor, S. M., Minase, T., Dominitz, R., and Sonnenblick, E. H. (1982). Microvascular spasm in the cardiomyopathic hamster: A preventable cause of focal myocardial necrosis. Circulation 66, 342–354. Finkel, M. S., Marks, E. S., Patterson, R. E., Speir, E. H., Steadman, K. A., and Keiser, H. R. (1987). Correlation of changes in cardiac calcium channels with hemodynamics in Syrian hamster cardiomyopathy and heart failure. Life Sci. 41, 153–159. Finkel, M. S., Romeo, R. C., Oddis, C. V., and Salama, G. (1992). Inotropic effects of calcium antagonists in the cardiomyopathic Syrian hamsters. J. Cardiovasc. Pharmacol. 19, 546 –553. Groom, A. C., Ellis, C. G., Wrigley, S., and Potter, R. F. (1986). Architecture and flow patterns in capillary networks of skeletal muscle in frog and rat. In ‘‘Microvascular Networks: Experimental and Theoretical Studies’’ (A. S. Popel and P. C. Johnson, Eds.), pp. 61–76. Basel, Karger. Hansen-Smith, F. M., Greene, A. S., Cowley, A. W., and Lombard, J. H. (1990). Structural changes during microvascular rarefaction in chronic hypertension. Hypertension 15, 922–928. Homburger, F. (1979). Myopathy of hamster dystrophy: History and morphologic aspects. Ann. N.Y. Acad. Sci. 317, 2–17. Hudetz, A. (1993). Percolation phenomenon: The effect of capillary network rarefication. Microvasc. Res. 45, 1–10. Ikegaya, T., Kobayashi, A., Hong, R. B., and Masuda, H. (1992). Stimulatory guanine nucleotide binding protein and adenylate cyclase activities in Bio 14.6 cardiomyopathic hamsters at the hypertrophic stage. Mol. Cell. Biochem. 110, 83–90. Lambert, C., Massillon, Y., and Meloche, S. (1995). Upregulation of cardiac angiotensin II AT1 receptors in congenital cardiomyopathic hamsters. Circ. Res. 77, 1001–1007.
Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Colantuoni, Coppini, and Bertuglia
Larosa, G., and Forster, C. (1996). Coronary beta-adrenoceptors function is modified by endothelium in heart failure. J. Vasc. Res. 33, 62–70. Kessler, P. D., Cates, A. E., Van Dop, C., and Feldman, A. M. (1989). Decreased bioactivity of the guanine nucleotide binding protein that stimulates adenylate cyclase in hearts from cardiomyopathic Syrian hamsters. J. Clin. Invest. 84, 244 –252. Kurose, I., Wolf, R., Grishman, M. B., and Granger, D. N. (1994). Modulation of ischemia/reperfusion-induced microvascular dysfunction by nitric oxide. Circ. Res. 74, 376 –382. Nolte, D., Zeintl, H., Steinbauer, M., Pickelmann, S., and Messmer, K. (1995). Functional capillary density an indicator of tissue perfusion. Int. J. Microcirc. Clin. Exp. 15, 244 –249. Pittman, R. (1995). Influence of microvascular architecture on oxygen exchange in skeletal muscle. Microcirculation 2, 1–18. Pogue, D. H., Moraves, C. S., Roppelt, C., Disch, C. H., Cressman, M. D., and Bond, M. (1995). Effect of lovastin on cholesterol content of cardiac and red blood cell membranes in normal and cardiomypathic hamsters. J. Pharmacol. Exp. Ther. 273, 863– 869. Sakata, Y. (1993). Tissue factors contributing to cardiac hypertrophy in cardiomyopathic hamsters (Bio 14-6): Involvement of trasforming growth factor beta 1 and tissue renin-angiotensin system in the progression of cardiac hypertrophy. Hokkaido Igaku Zasshi 68, 18 –28. Sapp, J. L., and Howlett, S. E. (1994). Density of ryanodine receptors is increased in sarcoplasmic reticulum from prehypertrophic cardiomyopathic hamster heart. J. Mol. Cell. Cardiol. 26, 325–334. Urasawa, K., Joshida, I., Takagi, C., Onozuka, H., Mikami, T., Kawaguchi, H., and Kitabatake, A. (1996). Enhanced expression of b-adrenergic receptor kinase 1 in the hearts of cardiomyopathic hamsters. Biochem. Biophys. Res. Commun. 219, 26 –30.
Capillary Density and Leukocyte Adhesion in Hamsters with Hereditary Cardiomyopathy A. Colantuoni, G. Coppini, and S. Bertuglia CNR Institute of Clinical Physiology, University of Pisa, Pisa, Italy Received September 15, 1997
The aim of this study was to characterize microvascular networks in cheek pouch of cardiomyopathic Syrian hamster (CM) (Bio 14.6), which is an interesting model of idiopathic cardiomyopathy and congestive heart failure. Microcirculation was visualized by fluorescence microscopy. Diameter and length of arterioles, classified according to centrifugal ordering scheme, were measured. A computational method was arranged to determine the density of arterioles and capillaries (total vessel length per unit area, cm21), fractal dimension of capillaries, and the associated Voronoi tesselation. Furthermore, leukocyte adhesion to venules and arteriolar reactivity to drugs were studied. Increase in the number of terminal arterioles and capillary rarefication characterized CM microvasculature compared with that of age-matched controls (58 6 7 versus 25 6 5 cm21, and 128 6 15 versus 240 6 10 cm21, respectively). Fractal dimension of capillaries was reduced in CM compared with controls (1.40 6 0.10 versus 1.85 6 0.09) and associated with increased avascular spaces, as shown by Voronoi tesselation results. Leukocyte adhesion to venules increased significantly in CM. In CM responsiveness of arterioles to nitric oxide inhibition and propranolol was slighter but more marked to norepinephrine and angiotensin II compared with that of control hamsters. In conclusion, the different geometry, increased leukocyte adhesion, and altered arterial responsiveness may contribute to flow disturbances in the microcirculation of CM hamsters. © 1998 Academic Press 0026-2862/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Key Words: microcirculation; cardiomyopathy; capillary density; fractal dimension; leukocyte.
INTRODUCTION The cardiomyopathic (CM) Syrian hamsters (strain BIO 14.6) are known to develop a genetically determined cardiomyopathy associated with progressive development of congestive heart failure (Homburger, 1979; Factor et al., 1982; Finkel et al., 1987, 1992; Ikegaya et al., 1992). These animals have a cardiac and skeletal myopathy characterized by progressive intracellular calcium overload, followed by histologic necrosis at 4 – 6 months and death at 10 –12 months of age (Kagiya et al., 1991; Sapp et al., 1994; Pogue et al., 1995). Cardiomyopathy might involve dysfunctions within the tissues with changes in local microvascular and rheological conditions. However, there are no studies documenting microvascular network design in animals with chronic cardiomyopathy or functional changes in the vessels of microcirculation. Namely, network morphology with arteriolar and capillary spatial density determines distribution of blood flow in microvasculature (Groom et al., 1986) and interferes with blood-tissue transports. Capillary density is the simplest indicator of the distribution of vessels and tissue perfusion (Bertuglia et al., 1993; Nolte et al., 1995). Furthermore, many reports evidenced the role of 85
86
leukocytes in the damage occurring in the tissue in ischemic and reperfused conditions (Arnould et al., 1993; Bertuglia et al., 1995). Many studies suggest that accumulation of leukocytes in in vitro incubation of cultured endothelial cells under hypoxia already develops (Bradury et al., 1993). However, it is not elucidated to what extent, these leukocyte abnormalities develop at the level of microcirculation in cardiomyopathy. The first aim of the study was to describe the structure of CM microvascular networks compared with that of control hamsters (C) using the cheek pouch model. We modified a computational method to measure arteriolar and capillary spatial density, the boxcounting dimension of capillary networks, which is an estimate of the fractal dimension, and the associated Voronoi tesselation to quantitate local geometrical relationships among vessels and the embedding space (Baish et al., 1996). This approach permits evaluation of the related region for a given vessel segment, i.e., the region perfused mainly by that vessel. We also evaluated the number of adherent leukocytes in venules and the changes in arteriolar responses to NG-monomethyl-l-arginine (l-NMMA), a nitric oxide (NO) inhibitor, norepinephrine, propranolol, a nonselective b1–b2 adrenergic antagonist, and angiotensin II in CM hamsters.
Colantuoni, Coppini, and Bertuglia
Observations were made with a Leitz Orthoplan microscope (Leitz, Germany), operating in incident light with a Ploemopak filter block, fitted with a long-working distance objective (34, n.a. 0.14; 320, n.a. 0.25) and 310 eyepiece. Epiillumination was provided by a xenon 150 W lamp source in conjunction with the appropriate filters for fluorescein isothiocyanate bound to dextran MW 150,000 (Leitz I2 Ploemopak filter block) and a heat filter (Leitz KG 1). The tracer was injected intravenously (50 mg/ 100 g body wt., as 5% solution) and followed by a COHU 5253 SIT low light level camera. The scenes were monitored by a Sony PVM 122 CE monitor and recorded by a Sony U-Matic VO 5800 PS video recorder. The images of interest were digitized by a computer-assisted technique.
Network Analysis Arteriolar networks were traced by microscopic observation (microscope long-working distance objectives: 34 and 320) following vessels from the input arterioles to capillaries. Vessels were assigned centrifugal orders, beginning with the largest arterioles classified as order 1 (A1). Hamster cheek pouch was observed for a period of 30 – 60 min to record network geometry.
Computational Method
MATERIALS AND METHODS Eight-month-old male cardiomyopathic hamsters (BIO 14.6, biobreeders, Fitchburg, MA) weighing 80 – 100 g (n 5 15) and age-matched male Syrian hamsters (Charles River, Calco, CO, Italy) weighing 80 –100 g were used (n 5 10) to characterize the microvascular network. The cheek pouch was prepared, as previously reported, to visualize the microvascular network (Bertuglia et al., 1993). Briefly, the membrane was gently everted and fixed to a special stage of the microscope: a very thin black blade was inserted through a small incision between the upper and lower layers of the pouch. The membrane was superfused with a 36 6 0.5°C Ringer solution, with 5% CO2 in 95% N2, adjusted to pH 7.35. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Regions of interest were interactively selected from acquired images to include portions of the microvascular network under investigation. The digital pictures were properly resampled to exploit the available information. We adopted a bilinear interpolation algorithm which provided a reasonable compromise between goodness of fit and computational load to achieve subpixel accuracy in the assessment of vessel size. A 4:1 resampling ratio was used, resulting in a pixel size of about 1 mm. Afterwards, the processing procedure included three steps: (1) segmentation of vascular structures; (2) representation of segmented vessel by proper attributed graphs; (3) parameter extraction. Segmentation was based on the assumption that vessels are usually bright, elongated (bar-like) struc-
87
Microvascular Networks in Cardiomyopathic Hamsters
tures embedded in a slow-varying darker background. To remove background and enhance vessels, images were filtered by a difference of Gaussian (DoG) kernel: DoG(x,y) 5
F
x2 1 y2 1 exp 2 2 ps 21 2 s 21
G
F
G
x2 1 y2 1 2 exp 2 , 2 ps 22 2 s 22 which is widely used for several computer vision tasks. The parameter s1 is chosen slightly smaller than the expected vessel size and s2 should have a larger value (a s1 value ranging from 3 to 5 times s2 is an adequate choice). The original picture of a normal vascular network and the related filter output are reported in Figs. 1a and 1b, respectively. Following DoG filtering, the images were segmented using an adaptive thresholding algorithm based on competitive learning. A segmented picture is shown in Fig. 1c. By processing the segmented picture, the second stage provided a compact description of the imaged network by an attributed graph whose nodes are bifurcations and links between nodes are vessel segments. To build a representation, we extracted the skeleton of the network by iterated morphological thinning. The adopted algorithm, modified by Chin et al. (1990), exhibits good noise immunity in comparison with alternative methods. Furthermore, thinning operators allow a local estimate of vascular thickness. For example, in the case of a perfectly straight vessel segment, at each thinning step two pixels (one pixel per side) were removed until a further thinning step resulted in vessel disappearing. Thus, for an n pixel structure, the number nth of needed thinning steps satisfied the equation: 2n # nth # 2n 1 1. The quantity nth was assumed as a measure of local vessel size and named morphological thinning thickness (MTT). Thus a bidimensional map of pixels representing the vascular axis was obtained, each pixel being labeled by related MTT value. Afterwards, vessel segments were extracted by linking connected skeleton points between consecutive branchings. Each vessel segment (graph node) was characterized by length, mean MTT, and position (as given by the coordinates of the first and last point, respectively). In addition, for each node a list of ves-
sel-point coordinates with the local MTT was stored. It is worth noting that recorded vessels can originate (and terminate) at a natural branching or at an obstruction as well. Therefore, the network-representing graph could be disconnected and related to vessel patency rather than to the topology of its connection pattern. Utilizing the network graph, terminal arterioles, capillaries, and venules were divided in three groups according to the mean MTT (Fig. 1d). We computed the related area-densities for the sets of arterioles (da) and capillaries (dc), defined as the total vessel length per unit area (expressed in cm21). The box-counting dimension Df was used for the capillary network as an estimate of the fractal dimension. Finally, the Voronoi tesselation associated with the capillary net was computed. Thus, the tissue space was partitioned into regions by assigning each point to the nearest vessel segment.
Leukocyte Adhesion To stain leukocytes the animals received an intravenous injection of acridine red (Chroma, Stuttgart, FRG) (1 mg/100 g body wt in 0.2 ml/100 g/h). Appropriate filters for acridine red (Leitz N2 Ploemopak filter block) were used for microscopic observations. Postcapillary venules with diameters of 16 6 8 mm and lengths ,250 mm were selected for the study. Sticking leukocytes were expressed as number per 100 mm of vessel length, during 30 s of observation.
Arteriolar Responses Two subgroups (n 5 5) of C and CM hamsters were treated with l-NMMA (1 mg/100 g body wt in 500 ml of 0.9% saline) (Bachem, Switzerland), norepinephrine (NE) (0.25 mg/100 g body wt) (Sigma, St. Louis, MO), propranolol (PP) (0.2 mg/100 g body wt) (Sigma), angiotensin II (Ang II) (1 mg/100 g body wt) (Sigma), administered intravenously over 3–5 min. The drugs were randomly injected: the vessels were reexposed to vasoactive agents after a 45- to 60-min equilibration period sufficient to regain control values. RBC velocity (VRBC) and vessel diameter were measured by a dual window technique (102 B IPM velocimeter; IPM, San Diego, CA) and by image shearing Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
88
Colantuoni, Coppini, and Bertuglia
TABLE 1 Mean Diameter, Number of Vessels Observed (n), and Lengths of Arterioles (A1, A2, A3, A4) in the Microvascular Networks of Control (C) and Cardiomyopathic (CM) Hamsters Arteriolar order Control A1 A2 A3 A4 CM A1 A2 A3 A4
Mean diameter (mm)
n
Mean length (mm)
75.4 6 6.5 26.6 6 5.2 16.3 6 4.1 8.2 6 1.1
10 10 23 115
2050 6 150 1380 6 126 827 6 105 219 6 48
67.5 6 6.4 26.8 6 5.3 15.2 6 4.6 8.5 6 1.2
10 10 26 234*
2250 6 130 1550 6 250 795 6 124 141 6 45*
* P , 0.01 compared with C group. Each order is significantly different when compared with previous order (P , 0.01).
(Image Shearing Monitor Mod 907; IPM, San Diego, CA), respectively. Mean arterial blood pressure (Viggo-Spectramed P10E2 transducer, Oxnard, CA, connected to a catheter in the femoral artery), respiratory and heart rates were monitored by a Gould Windograf recorder (Mod. 13-6615-10S Gould Inc. OH). Data were recorded and stored in a computer. The Gaussian distribution fit was examined by Kolmogorov–Smirnov test. When data presented normal distribution we used Student’s t test; for nonparametric testing of null hypothesis Mann–Whitney U test was used. To determine which groups were statistically different, ANOVA, Scheffe`’s post hoc, or Kruskal–Wallis tests were employed. To compare Voronoi tesselation results we used chi-square test. The reported values are means 6 SD. Differences were considered significant at P , 0.05.
RESULTS Network Analysis Table 1 lists the geometrical features of the microvessels of both C and CM hamsters. Microvascular networks of control and CM hamsters were characterized by four orders of arterioles (A1–A4) according to centrifugal ordering scheme (Fig. 1). Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
CM showed an increased number of A4 terminal arterioles branching from A3 parent vessels (Table 1). Consequently, the mean distance between the origin of A4 arterioles along A3 vessels was significantly different compared with that detected in controls (75 6 35 mm versus 236 6 57 mm, P , 0.01). Terminal microvasculature was characterized by the presence of direct arteriolar–venular conduits with higher VRBC when compared with capillary VRBC. VRBC in postcapillary venules of CM (n 5 20) was significantly lower than that of venules in C (n 5 20) hamsters (0.25 6 0.10 versus 0.45 6 0.05 mm/s).
Computational Method The arteriolar and capillary densities were 25 6 5 cm21, 240 6 10 cm21 in C (n 5 10) and 58 6 7 cm21, 128 6 15 cm21 in CM (n 5 10) hamsters, respectively. The capillary fractal dimension was significantly different, 1.85 6 0.09 in C and 1.40 6 0.10 in CM hamsters. Two examples of computed capillary networks superimposed on their original pictures are shown in Fig. 2. A control (Df 5 1.78, dc 5 244 cm21) and a CM network (Df 5 1.45, dc 5 111 cm21) are reported in Figs. 2a and 2b, respectively. The space-filling behavior of a network can be further described by the Voronoi tesselation. In Fig. 3 we plot the total fraction of tissue points against their distance to the nearest vessel. In control cases the distance to the nearest vessel was ,100 mm for more than 90% of tissue points. The percentage was significantly reduced to ,80% in CM hamsters (P , 0.01).
Leukocyte Adhesion CM hamsters demonstrated a significant difference in the number of adherent leukocytes in CM (n 5 40) when compared with C (n 5 30) animals (9.5 6 1.7 versus 2.8 6 0.6/100 mm venular length) (P , 0.01) (Table 2).
Arteriolar Responses In CM hamsters the mean arterial blood pressure (MAP) was significantly lower than that measured in C animals (80 6 4 versus 97 6 5 mm Hg).
Microvascular Networks in Cardiomyopathic Hamsters
89
FIG. 1. Processing phases of a control microvasculature. a) original image, b) DoG filtering, c) thresholding, d) morphological thinning and vessel grouping into arterioles (black thick lines), capillaries (thin lines) and venules (gray thick lines). See text for details.
In presence of l-NMMA the mean diameter of A2 and A4 arterioles decreased significantly in C and CM hamsters (Table 3). In CM hamsters, however, there was a slighter vasoconstriction. A moderate rise in MAP was observed.
With NE there was a significant vasoconstriction in C and CM hamsters, but in CM it was significantly higher. With PP there was a slight arteriolar response in CM hamsters which was significantly different compared Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
90
Colantuoni, Coppini, and Bertuglia
FIG. 2. Computed capillary networks overlaid on the original pictures. a) control (Df 5 1.78, dc 5 244 cm21); b) cardiomyopathy (Df 5 1.45, dc 5 111 cm21).
with that of C hamsters. MAP did not change significantly in CM hamsters. Ang II produced a significantly higher contractile response in CM hamsters compared with that observed in
C hamsters and was sustained for at least 3 min. Ang II produced a dose-dependent increase in MAP (98 6 1.2 mm Hg, P , 0.05) that returned to the initial resting value within 4 6 1 min.
FIG. 3. The fraction of tissue points (cumulative frequency) plotted against their distance to the nearest vessel for control (C) and cardiomyopathic (CM) hamsters, respectively. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
91
Microvascular Networks in Cardiomyopathic Hamsters
TABLE 2 Mean Arterial Blood Pressure (MAP), Arteriolar and Capillary Spatial Densities (da, dc), Fractal Dimension (Df), Number of Adherent Leukocytes per Vessel Length (L/100 mm), Capillary Red Blood Cell Velocity (VRBC) in Control (C) and Cardiomyopathic (CM) Hamsters Group
MAP (mm/Hg)
da (cm21)
dc (cm21)
Df
L/100 mm
C n CM n
97 6 5 10 80 6 4* 15
25 6 5 10 58 6 7* 10
240 6 10 10 128 6 15* 10
1.85 6 0.09 10 1.40 6 0.10* 10
2.8 6 0.6 30 9.5 6 1.7* 40
n
VRBC (mm/s) 0.22 6 0.04 20 0.20 6 0.03 40 0.35 6 0.05*a 20
a
VRBC in A–V conduits. Note. Values are mean 6 SD. n 5 number of vessels (L/100 mm; VRBC); in the other cases number of animals. * P , 0.01 compared with C hamsters.
DISCUSSION This study demonstrates that abnormalities in microvascular geometry are detectable in hamsters with chronic cardiomyopathy. CM hamsters showed a significant capillary rarefication, partially compensated by the increase in terminal arteriole spatial density, and an increased number of adherent leukocytes to venules. Furthermore, in CM animals we showed an impaired response of arterioles to propranolol, meanwhile endothelial responses appear to be depressed. Vasoconstriction induced by norepinephrine and angiotensin II was more marked. We observed that the total length of terminal arterioles was significantly higher in CM than in C hamsters. Furthermore, an increased number of terminal arterioles was related to a reduced number of capillaries that eventually appeared to be grouped
in small loops. Furthermore, there were arteriolar– venular conduits characterized by increased blood flow velocity. Computer measurements permitted an interesting analysis of the geometrical and functional arrangements of microvessels. Capillary density can be quantitated by current morphological analysis and descriptive statistics without information on the spatial heterogeneity of microvessels. We found a significant correspondence between morphological and computational network analyses about identification of microvessels. Nevertheless, the cheek pouch is a suitable model for two-dimensional network reconstruction. The method of estimating the fractal dimension by covering the structure to be analyzed with boxes gives a measure of the complexity of the system over space. Therefore, the fractal dimension is an estimate of the related space-filling properties, which plays a relevant role for microvascular blood flow distribution.
TABLE 3 Arteriolar Responses (Percentage Decrease of Baseline Diameter) to l-NMMA, Norepinephrine (NE), Propranolol (PP), and Angiotensin II (Ang II) in Control (C) and Cardiomyopathic (CM) Hamsters Group
Control diameter (mm)
l-NMMA
NE
25.8 6 2.3 8.0 6 1.0
221 6 1% 237 6 4%
240 6 3% 250 6 2%
238 6 5% 240 6 3%
250 6 4% 260 6 3%
26.1 6 2.8 8.4 6 1.3
218 6 1%* 228 6 1%*
280 6 10%* 295 6 5%*
218 6 5%* 24 6 0.2%*
275 6 2%* 287 6 2%*
PP
Ang II
C A2 A4 CM A2 A4
Note. Values are mean 6 SD. n 5 7 for each entry. * P , 0.01 compared with C. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
92
CM hamsters exhibited significant capillary rarefication, reduction of fractal dimension associated with increased avascular spaces, as shown by Voronoi tesselation results, indicating a severe impairment of peripheral perfusion. The structure of network and the effects of mean vascular spacing are important functional parameters on oxygen transport (Pittman, 1995). Conversely, the fractal dimension had higher values, relatively close to the topological dimension of the imaged embedding space in normal hamsters. This observation suggests a plane-filling behavior of the capillary network, in accordance with the findings by Baish et al. (1996). The observed effect of capillary rarefication on overall network resistance is similar to the phenomenon known as percolation studied in computer simulations (Hudetz, 1993; Baish et al., 1996). Thus, the observed reduction of capillary spatial density of about 40% may suggest that CM networks are near to (or below) the percolation threshold. Therefore, sudden increase in arteriolar resistance may determine easily impairment of either nutritive flow or blood flow distribution in the tissues. Indeed, in comparison with normal hamsters the CM vascular architecture appears to be altered with larger spaces among capillaries; therefore, flow can be less uniformly distributed than in a normal microvasculature. Furthermore, there are loops of grouped capillaries and particular conduits between arterioles and venules, so that areas with different resistance and blood flow distribution in the networks are present. These conduits may be particular adaptive changes to increased resistance due to rarefication of capillaries and proliferation of terminal arterioles. It is interesting to note that a number of studies have reported a reduction in the density of microvessels in many animal models of hypertension and in human hypertensive subjects. Ultrastructural studies of the cremaster muscle of rats with chronic reduced renal mass have revealed that microvascular rarefication is mediated by atrophy and structural degeneration of microvessels (Hansen-Smith et al., 1990). In this case the reduction in microvessels is mediated by degenerative changes of endothelial and vascular smooth muscle cells but there are no data on changes in tissue geometry. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Colantuoni, Coppini, and Bertuglia
In CM hamsters we have chronic conditions of low arterial pressure that might cause structural changes in the geometry of the network. These structural changes with formation of new arteriolar vessels may be indicative of a compensatory regulation of arteriolar tone and thereby this microvascular arrangement regulates tissue blood flow and vascular resistance. Thus, heart insufficiency correlated with microvascular impairments may induce chronic conditions of tissue hypoxia. Furthermore, there is an increased leukocyte adhesiveness to venular walls, suggesting impaired mechanism of leukocyte activation and/or altered endothelium responsiveness. Despite many reports with evidence of accumulation of leukocytes in ischemia reperfusion and that demonstrate their role in the damage occurring in the tissue (Bradury et al., 1993; Bertuglia et al., 1995) the initial adhesiveness remains unclear. Adherence is a crucial step in the process of the following infiltration and margination of leukocytes through the endothelium. This process is the result of complex interaction between adhesion molecules present on both the endothelium and the leukocytes. It might be hypothesized that probably there is an increased expression of endothelial adhesion molecules during chronic conditions of hypoxia (Sakata, 1993). Regarding the exact mechanism by which this phenomenon occurs we could suggest that hypotension decreases shear rate in vessels, causing reduction in NO formation and increased number of adherent leukocytes. Indeed, we observed many patterns of flow in postcapillary venules and some were characterized by very low flow compared with that of venules in C hamsters. Indeed, venules directly connected to A-V conduits showed a VRBC lower than that observed in C hamsters. These data on venular VRBC and disturbances of flow support the hypothesis that the reduction in shear rate may be involved in the increased leukocyte adhesion on venular walls. Arnould et al. (1995) found that hypoxia induces neutrophil adherence to umbelical vein endothelium. Thus hypoxic tissue may initiate an inflammatory response like the one reported in ischemic organs, which is then further worsened during developing cardiomyopathy. Furthermore, leukocytes accumulate in the tissue causing capillary plugging that may contribute to the
93
Microvascular Networks in Cardiomyopathic Hamsters
progressive reduction of capillary perfusion. However, the capillaries appeared to be perfused under baseline conditions and we did not observe leukocyte– capillary plugging during the periods of observation in CM hamsters. In chronic congestive heart failure most studies on the peripheral circulation have been confined on muscle circulation and skin resistance arteries, with impaired contraction and relaxation (Angus et al., 1993). Namely, congestive heart failure is associated with abnormalities in b-adrenoceptor function rather than in the number of receptors (Kessler et al., 1989; Cai et al., 1993). Urasawa et al. (1996) studied b-adrenergic receptor in the heart of CM hamsters. The expression was significantly enhanced and contractility of the heart decreased, suggesting that intracellular homeostasis against accelerated stimulation by catecholamines was maintained by a feed back mechanism of b-adrenergic receptor phosphorylation. Furthermore, Larosa and Forster (1996) found that coronary vascular tone via b-adrenoceptor stimulation is modulated also by endothelium in congestive heart failure. Our data show that the arterioles are less sensitive to vasodilatory stimuli acting on b-adrenoceptors but the endothelium depression is marked. Previously, it was found that NO has a fundamental role in determining arteriolar responses and capillary perfusion in ischemia (Kurose et al., 1994; Bertuglia et al., 1995). Lambert et al. (1995) found that Ang II receptor (AT1 receptor) is upregulated in the ventricles of CM hamsters, suggesting that this may play a role in the genesis and maintenance of the cardiac hypertrophy in the hamsters. We found an increased tendency to constriction in arterioles to Ang II, strongly suggesting similar abnormalities. A probable explanation is that significant alterations in the production or in arteriolar responsiveness to NO may lead to increased tendency to vasoconstriction in CM hamsters. Therefore, it may be suggested that NO exerts a fundamental role in vasodilatory responses under conditions of chronic hypoxia where the preservation and control of flow are critical. In CM hamsters as a consequence of depressed endothelial response and increased tendency to vasoconstriction of arterioles there is a reduced pressure trans-
mitted downstream to distal arterioles. Therefore, the chronic conditions of low pressure and low shear stress may determine a progressive rarefication in the capillary network. The increase in the number of terminal arterioles, the capillary rarefication, and the increased leukocyte adherence might be suggested as determinant factors involved in the disruption of microvascular homeostasis which occur in CM hamsters. Furthermore, the altered arterial responsiveness may contribute importantly to microvascular flow disturbances and play a potential role in the pathogenesis of cardiac dysfunction and heart failure. In conclusion, we have shown evident changes in the microvascular network in CM hamsters, namely in capillary perfusion, leukocyte adhesiveness, and arteriolar responsiveness that might lead to the impairment of the microcirculation.
REFERENCES
Angus, J. A., Ferrier, C. P., Sudhir, K., Kaye, D. M., and Jennings, G. L. (1993). Impaired contraction and relaxation in skin resistance arteries from patients with congestive heart failure. Cardiovasc. Res. 27, 204 –210. Arnould, T., Michiels, C., and Remacle, J. (1993). Increased PMN adherence in endothelial cells after hypoxia: Involvement of PAF, CD18/CD11b and ICAM-1. Am. J. Physiol. 264, C1102–1110. Arnould, T., Michiels, C., Janssens, D., Delaive, E., and Remacle, J. (1995). Hypoxia induces PMN adherence to umbilical vein endothelium. Cardiovasc. Res. 30, 1009 –1016. Baish, J. W., Gazit, Y., Berk, D. A., Nozue, M., Baxter, L. T., and Jain, R. K. (1996). Role of tumor vascular architecture in nutrient and drug delivery: An invasion percolation based network model. Microvasc. Res. 51, 327–346. Bertuglia, S., Colantuoni, A., and Intaglietta, M. (1993). Effect of leukocyte adhesion and microvascular permeability on capillary perfusion during ischemia reperfusion injury in hamster cheek pouch. Int. J. Microcirc. Clin. Exp. 13, 1–28. Bertuglia, S., Colantuoni, A., and Intaglietta, M. (1995). Treatment with L-arginine, L-NMMA, L-NNA in ischemia reperfusion injury in cheek pouch microvasculature. Microvasc. Res. 50, 162–175. Bradury, A. W., Murie, J. A., and Ruckley, C. V. (1993). Role of the leukocyte in the pathogenesis of vascular disease. Br. J. Surg. 80, 1503–1512. Cai, J. J., Hongwei, Y. U., and Lee, H. C. (1993). Developmental changes of protein kinase C and G in hypertrophic cardiomyopathic hamster hearts. J. Lab. Clin. Med. 122, 533–541. Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
94 Chin, R. T., and Jang, B. (1990). Analysis of thinning algorithm using mathematical morphology. IEEE Trans. Pattern Anal. Machine Intell. 12, 541–551. Factor, S. M., Minase, T., Dominitz, R., and Sonnenblick, E. H. (1982). Microvascular spasm in the cardiomyopathic hamster: A preventable cause of focal myocardial necrosis. Circulation 66, 342–354. Finkel, M. S., Marks, E. S., Patterson, R. E., Speir, E. H., Steadman, K. A., and Keiser, H. R. (1987). Correlation of changes in cardiac calcium channels with hemodynamics in Syrian hamster cardiomyopathy and heart failure. Life Sci. 41, 153–159. Finkel, M. S., Romeo, R. C., Oddis, C. V., and Salama, G. (1992). Inotropic effects of calcium antagonists in the cardiomyopathic Syrian hamsters. J. Cardiovasc. Pharmacol. 19, 546 –553. Groom, A. C., Ellis, C. G., Wrigley, S., and Potter, R. F. (1986). Architecture and flow patterns in capillary networks of skeletal muscle in frog and rat. In ‘‘Microvascular Networks: Experimental and Theoretical Studies’’ (A. S. Popel and P. C. Johnson, Eds.), pp. 61–76. Basel, Karger. Hansen-Smith, F. M., Greene, A. S., Cowley, A. W., and Lombard, J. H. (1990). Structural changes during microvascular rarefaction in chronic hypertension. Hypertension 15, 922–928. Homburger, F. (1979). Myopathy of hamster dystrophy: History and morphologic aspects. Ann. N.Y. Acad. Sci. 317, 2–17. Hudetz, A. (1993). Percolation phenomenon: The effect of capillary network rarefication. Microvasc. Res. 45, 1–10. Ikegaya, T., Kobayashi, A., Hong, R. B., and Masuda, H. (1992). Stimulatory guanine nucleotide binding protein and adenylate cyclase activities in Bio 14.6 cardiomyopathic hamsters at the hypertrophic stage. Mol. Cell. Biochem. 110, 83–90. Lambert, C., Massillon, Y., and Meloche, S. (1995). Upregulation of cardiac angiotensin II AT1 receptors in congenital cardiomyopathic hamsters. Circ. Res. 77, 1001–1007.
Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.
Colantuoni, Coppini, and Bertuglia
Larosa, G., and Forster, C. (1996). Coronary beta-adrenoceptors function is modified by endothelium in heart failure. J. Vasc. Res. 33, 62–70. Kessler, P. D., Cates, A. E., Van Dop, C., and Feldman, A. M. (1989). Decreased bioactivity of the guanine nucleotide binding protein that stimulates adenylate cyclase in hearts from cardiomyopathic Syrian hamsters. J. Clin. Invest. 84, 244 –252. Kurose, I., Wolf, R., Grishman, M. B., and Granger, D. N. (1994). Modulation of ischemia/reperfusion-induced microvascular dysfunction by nitric oxide. Circ. Res. 74, 376 –382. Nolte, D., Zeintl, H., Steinbauer, M., Pickelmann, S., and Messmer, K. (1995). Functional capillary density an indicator of tissue perfusion. Int. J. Microcirc. Clin. Exp. 15, 244 –249. Pittman, R. (1995). Influence of microvascular architecture on oxygen exchange in skeletal muscle. Microcirculation 2, 1–18. Pogue, D. H., Moraves, C. S., Roppelt, C., Disch, C. H., Cressman, M. D., and Bond, M. (1995). Effect of lovastin on cholesterol content of cardiac and red blood cell membranes in normal and cardiomypathic hamsters. J. Pharmacol. Exp. Ther. 273, 863– 869. Sakata, Y. (1993). Tissue factors contributing to cardiac hypertrophy in cardiomyopathic hamsters (Bio 14-6): Involvement of trasforming growth factor beta 1 and tissue renin-angiotensin system in the progression of cardiac hypertrophy. Hokkaido Igaku Zasshi 68, 18 –28. Sapp, J. L., and Howlett, S. E. (1994). Density of ryanodine receptors is increased in sarcoplasmic reticulum from prehypertrophic cardiomyopathic hamster heart. J. Mol. Cell. Cardiol. 26, 325–334. Urasawa, K., Joshida, I., Takagi, C., Onozuka, H., Mikami, T., Kawaguchi, H., and Kitabatake, A. (1996). Enhanced expression of b-adrenergic receptor kinase 1 in the hearts of cardiomyopathic hamsters. Biochem. Biophys. Res. Commun. 219, 26 –30.