Dynamic Changes in Three-Dimensional Architecture and Vascular Volume of Transmural Coronary Microvasculature Between Diastolic- and Systolic-Arrested Rat Hearts
Background— The phase difference of coronary arterial and venous flows indicates the importance of intramyocardial capacitance vessels in storing diastolic flow and in discharging volume in systole. However, the anatomic and functional characteristics of the capacitance vessels are unclear. We aimed to clarify those characteristics with their transmural difference by 3D visualization of transmural microvessels under diastole and systole.
Methods and Results— We performed complete intracoronary filling of a contrast medium into Langendorff’s Wistar rat hearts under (1) St Thomas–perfused diastolic arrest (D-mode) and (2) BaCl2-induced systolic arrest (S-mode). Precise transmural 3D architectures of capillaries and of pre- and post-capillary microvessels (ie, microvessels larger than capillaries) were visualized clearly with a confocal laser scanning microscope and x-ray microcomputed tomography (microCT), respectively. Vascular volume fraction (VF) and systolic-induced VF reduction rate from D- to S-mode were analyzed. The net capillary VF in D-mode (20.4±0.9%) was 10 times that of larger microvessels and was reduced in S-mode by 32% without capillary collapse. Systolic-induced VF reduction rate was smaller in capillaries than in larger microvessels (48%; P<0.05). The larger microvessel VF in D-mode (2.2±0.2%) was reduced in S-mode, accompanied by complicated 3D deformation.
Conclusions— Capillaries were relatively resistant to the systolic extravascular compression compared with pre- and post-capillary microvessels, conveniently beneficial for the myocardial oxygen delivery throughout a cardiac cycle. Nevertheless, a larger change in the absolute volume of capillaries may function as effective capacitance. On one hand, the pre- and post-capillary microvessels showed a larger phasic change in resistance, which may function to maintain the capillary patency during systole.
Received September 28, 2001; revision received November 9, 2001; accepted November 15, 2001.
In the coronary circulation, arterial blood inflow increases during diastole, whereas venous outflow increases during systole. This indicates that arterial inflow into the myocardium is stored in capacitance vessels during diastole and the venous flow is ejected out from these capacitance vessels during systole.1–3⇓⇓ However, neither anatomic nor functional characteristics of capacitance vessels have been determined,4 mainly because of the lack of information about intramyocardial microvessels in terms of their phasic volume changes between diastole and systole. Earlier, our studies using intravital videomicroscopy5,6⇓ indicated transmural differences in the phasic diameters; however, they could not evaluate the global appearance of transmural volumetric coronary microvascular dynamics during a cardiac cycle. Conventional histological studies on cross-sectional specimens7,8⇓ suggested microvascular area changes during a cardiac cycle; however, these studies examined only 2-dimensional information on the intramyocardial microvasculature.
To clarify the functional and anatomic characteristics of intramyocardial capacitance and resistance vessels, as well as possible transmural differences, we visualized the 3D architecture of the coronary vasculature by intracoronary injection of contrast medium in diastolic- and systolic-arrested rat hearts. Three-dimensional architectures of capillaries and pre- and post-capillary microvessels (microvessels larger than capillaries [LMs]) were visualized transmurally, and vascular volume fractions (VFs) of the capillaries and LMs were compared. Transmural capillaries and LMs in the identical heart were 3-dimensionally visualized using a confocal laser scanning microscope (CLSM) (higher resolution but smaller sample size) and an x-ray microcomputed tomography (microCT) (lower resolution but larger sample size), respectively. Two different microCT systems were used: a commercially available microCT and a synchrotron radiation microCT. The former was used mainly to compile image data. The synchrotron radiation microCT at Japan Synchrotron Radiation Research Institute (SPring-8, Hyogo, Japan), which provides the world’s most intense collimated beam of monochromatic x-ray, was used to validate the image quality and the VF analysis.
Characteristics of Contrast Medium
Our specially prepared contrast medium consists of (1) 20% v/v of BaSO4 (Baritgen zol, Fushimi Co, Ltd) for microCT imaging, (2) 20% v/v of India ink (Saga Kokyu Nosen, Kaimei Co, Ltd) for CLSM observation at reflection mode, (3) 8% w/v of gelatin (Nakalai Tesque, Inc), and (4) distilled water. The warmed medium (42°C) was administered intracoronarily and solidified by intense cooling within ≈30 s. This enabled complete filling of entire microvasculature under physiologically relevant perfusion pressure by the procedure described below.
Experimental procedures and protocols were conducted according to the institutional guidelines approved by the Animal Research Committee of Kawasaki Medical School (No. 99137, No. 00057). Adult male Wistar rats (n=22; 250 to 300 g; Japan SLC Inc, Hamamatsu, Japan) were anesthetized (diethyl ether inhalation), and a heparin (500 IU, IV) was administered. After median sternotomy, the heart was isolated and connected to a Langendorff perfusion system. A polyethylene tube was inserted into the left ventricle (LV) via the mitral valve to unload LV. Another tube attached with a balloon tip was used to impose LV pressure in additional experiments. Throughout each experiment, adenosine (1 mg/min) was administered, and the perfusion pressure and mean coronary flow rate were continuously monitored (CAMINO, V420-9 Laboratories, Inc, and MFV3200, Nihon Kohden Inc, respectively).
Diastolic-arrested hearts (D-mode; n=8) were prepared by retrograde perfusion of oxygenated St Thomas cardioplegic solution (25°C). Systolic-arrested hearts (S-mode; n=8) were prepared by oxygenated modified Tyrode’s solution (37°C), followed by 2 mmol/L of BaCl2 (25°C; for 90 s). For comparison, S-mode hearts under imposed LV pressure of 85 mm Hg also were prepared (S-mode with load; n=6). Thereafter, the contrast medium (42°C) was administered into the coronary circulation by retrograde from the nearest port of the circuit. Perfusion pressure was initially set at 120 mm Hg for 10 s and then lowered to 85 mm Hg throughout the remainder of the perfusion period (3 minutes). The whole heart was quickly immersed into a cold saline-circulating bath (0°C) and kept there for 3 minutes so that the contrast medium solidified under the perfusion pressure. Required time to complete these procedures was <10 minutes. Finally, the heart was removed and immersed in a mixture of 2.5% glutaraldehyde and 2% paraformaldehyde (4°C) overnight.
Fixed hearts were cut along the vertical long axis, and the extent of complete microvascular filling of the contrast medium was examined with a stereoscopic microscope. For microCT studies, a transmural myocardial column (φ4 mm) was punched out from the LV free wall and sealed into a phosphate buffer solution–filled polyethylene tube. For CLSM studies, the remaining sample was embedded in 12% gelatin block and fixed again overnight. This sample block was sliced into 200-μm thicknesses using a vibrating microtome (VT1000S, Leica, Inc).
The LMs were visualized transmurally with the microCT (ELE-SCAN, Nittetsu ELEX, Inc), the specifications of which are as follows: The x-ray image intensifier has 72×54 mm2. The CCD camera has 640×480 pixels. Geometrically, the pixel size under the maximum magnification (×20.0) is 5.6 μm. Reference width-lines of 14.7 μm in the US Air Force Target Scale were discriminated as the maximal resolution. In the present study, the pixel size of 9.6 μm (magnification ×11.8) was adopted to focus on the signals of the vessels larger than capillaries and to defocus the capillaries. The CT image for each slice (9.6 μm in thickness) was obtained by 600 projections with 8 times accumulation at 30 keV, 100 μA.
To validate the image quality and vascular volume analysis by the microCT, 5 samples were visualized using the synchrotron radiation microCT9 at SPring-8, which provides an intense collimated beam of nearly parallel monochromatic x-ray. X-ray images were detected on the fluorescent screen, lens-coupling CCD camera detector with 6-μm pixel size. Reference width-lines of 8.8 μm in the US Air Force Target Scale were discriminated as the maximal resolution. The CT image was obtained with 360 circular projections in 1024×1024 pixels. X-ray energy was 17 keV.
Confocal Laser Scanning Microscope Observation
The capillary was visualized by laser reflection mode of CLSM (TCS-NT, Leica, Inc). Microscopic images were obtained at a magnification of ×40 and scanned for 50 to 60 μm in depth divided by 1 μm for 1 slice thickness. Microscopic observations were performed at the 3 layers; epi- (EPI), mid- (MID), and endo- (ENDO) myocardial layers.
Three-Dimensional Visualization and Vascular Volume Analysis
The CLSM images were automatically rendered (TCS-NT, Leica, Inc) and displayed as 3D images. The microCT images were processed to optimize for vascular enhancement by referring to the signal histogram, and then were used for 3D volume rendering (Voxblast2.2). Digital images were converted into the stack files (NIHimage1.62) for the volumetric analysis. For the vascular volume analysis of the microCT images, the region of interest (ROI) was limited to the middle third of the sample column, which almost exclusively consists of myocardium and microvessels, because the superficial third includes epicardial large conduit vessels and extracardiac space, and the deeper third may include trabeculae. The vascular size of LMs for VF analysis ranged from 300 μm to 9.6 μm in diameter. For the CLSM volumetric analysis, ROI was manually focused only on capillaries, excluding LMs.
Percent vascular area out of total area was analyzed in each slice; then, this process was scanned along depth direction. The VF (percentage) was analyzed as the voxel size ratio of the vessels (LMs and/or capillaries) out of the total mass (MATLAB5.2). The systolic VF reduction rate was calculated as [(VF in D-mode)−(VF in S-mode)]/(VF in D-mode)×100 (%). The sample number used for LM VF in D-mode, S-mode, and S-mode with load was 8, 8, and 6, respectively, because only 1 sample was taken from each heart. The sample number for capillary VF in D-mode, S-mode, and S-mode with load was 44, 49, and 42 from 8, 8, and 6 hearts, respectively.
All data were expressed as mean±SEM. Comparisons between the 3 conditions were performed by 1-way ANOVA followed by Fisher’s pairwise least-significant difference test for multiple comparisons using StatView5.0PPC. Comparisons between vascular size under 3 cardiac conditions were performed by 2-way fractional ANOVA. Statistical criterion was defined as P<0.05.
Representative stereoscopic images revealed well-dilated LV in D-mode, whereas they showed contracted LV with extremely narrowed intra-LV cavity in S-mode (Figure 1). LV configuration (LV diameter and wall thickness) in S-mode with load was between that observed for S-mode and D-mode. Because of the vivid black color of the contrast medium, we could confirm under magnification that the contrast medium had filled completely, without any visible filling defects, in all specimens.
Three-Dimensional Architecture and Vascular Volume Fraction of Large Microvessels
The microCT images clearly demonstrated the 3D architecture of the transmural LMs. No obvious signals indicating capillaries were observed in any of the CT slice images (images not shown). The representative rotational microCT images in D-mode, S-mode, and S-mode with load (Movies I, II, and III, respectively) showed that LMs ran spirally through the myocardial wall. In S-mode and S-mode with load, the size of vascularity seemed to be decreased. The ginger root–shaped venules (GRVs), characterized by large, irregular, planiform shape, were prominent in D-mode, whereas they were likely to be compressed to the point of being nearly flat (obvious in Movie II), with a nonuniform manner, in S-mode.
The LM VF under the 3 conditions is shown in Figure 2A. VF was reduced significantly (by 48%) from D-mode (2.2±0.2%; n=8) to S-mode (1.1±0.2%; n=8) (P=0.0017) and by 54% from D-mode to S-mode with load (1.0±0.2%; n=6) (P=0.0024). Although there was no significant difference between S-mode and S-mode with load (P=0.68), a tendency toward lower VF was observed in the latter.
To validate the above results, the representative images observed by synchrotron radiation microCT are shown in Figure 3. Finer images were obtained, but the principal findings were the same as those in the Movies. The VF in the ROI was 2.8% in D-mode, 1.2% in S-mode, and 0.5% in S-mode with load. These values were in agreement with the data from the Movies.
Three-Dimensional Architecture and Vascular Volume Fraction of Capillaries
Three-dimensional architecture of the capillary network was visualized clearly by CLSM. The capillaries also were compressed from D- to the S-modes, without any visible collapse or interrupted findings transmurally, even in ENDO in S-mode and S-mode with load (Figure 4). Significant VF reduction due to myocardial contraction was observed, as shown in Figure 2B. The capillary VF was reduced significantly (by 32%) between D-mode (20.4±0.9%; n=44) and S-mode (13.9±0.6%; n=49) (P<0.0001) and by 39% between D-mode and S-mode with load (12.4±0.6%; n=42) (P<0.0001). There was no significant difference between S-mode and S-mode with load (P=0.149). Importantly, the capillary VF (20.4±0.9%) was nearly 10-fold greater than that in LMs (2.2±0.2%) in D-mode. This indicates that capillaries contribute importantly to intramyocardial capacitance, although the VF reduction from D- to S-mode was significantly smaller in the capillaries (32%) than in LMs (48%, P<0.05).
Layer-Dependent Difference in Capillary Vascular Volume Fraction
Differences in capillary VF due to the cardiac conditions (ie, D-mode, S-mode, and S-mode with load) were significant, but VF differences between the layers were not significant (P<0.0001 and P=0.40, respectively). The VF in ENDO was reduced by 37% from D-mode (21.8±1.9%) to S-mode (13.7±0.9%) and by 47% from D-mode to S-mode with load (11.6±0.7%) (both P<0.0001, Figure 5). The VF in MID was reduced by 34% from D-mode (21.1±1.4%) to S-mode (13.9±1.1%) and by 41% from D-mode to S-mode with load (12.5±1.4%) (P<0.0001 and P<0.0002, respectively). The VF change in EPI from D-mode (17.7±2.3%) to S-mode (14.4±0.8%) was not significant (19% reduction, NS), but the change from D-mode to S-mode with load (13.2±0.7%; 25% reduction; P<0.05) was significant. In each layer, VF in S-mode with load tended to be lower than that in S-mode, although the difference was not statistically significant.
We evaluated the cardiac contraction–induced changes in 3D architecture by focusing on the VF of capillaries and LMs in the rat heart with the use of imaging systems with different spatial resolutions (CLSM and microCT) to elucidate the differences in vascular mechanical properties and in the functional roles between capillaries and LMs during systole and diastole.
Historically, capillaries were considered to be rigid tubes because the diameter changes in capillaries from the mesentery were negligible even when blood pressure was changed over a range of 100 cm H2O. Fung10 explained the capillary rigidity by the “tunnel in gel” concept. He also predicted, however, that when surrounding tissues were minimized, capillary rigidity would lessen.11 Fung also speculated that, because cardiac myocytes and capillaries are coupled tightly, systolic tissue pressure could compress the capillary lumen.10 Caulfield et al12 speculated that capillary patency may be maintained by tethering struts attached to the capillaries, which corroborates other reports in which collapse of capillaries was not observed in sections from systolic-arrested hearts.7
In the present study, we confirmed volume reduction but no visible collapse of capillaries throughout the myocardial wall during systole. Systolic reduction in capillary VF (32%) was smaller than that of LM VF (48%). These findings may indicate that the capillaries are relatively resistant to the systolic extravascular compression compared with LMs, as we speculated earlier.5 We considered that the capillary tolerance to systolic extravascular compression may be advantageous for myocardial oxygen delivery throughout a cardiac cycle of an in vivo beating heart.
Capillary VF (20% in D-mode) was demonstrated to occupy ≈10 times more than LM VF (2% in D-mode). Systolic reduction in capillary VF was reduced by 32% in S-mode, which is comparable to the systolic reduction of 38% reported previously using 123I-labeled albumin.8 The widely spread capillary network and larger absolute volume change in the capillaries may play a major role as effective capacitance to accommodate diastolic myocardial inflow and to eject it out to venous outflow in systole. Although our model was static systole, it is possible to extrapolate the present data to normally ejecting heart. Maximal ejection of intramyocardial blood pooled during diastole occurs during the early half of systole, because both venous forward outflow13 and arterial reverse flow14 are observed predominantly in early systole. Accordingly, overestimation of the systolic microvascular volume reduction because of static systole (long systole) may be small. Alternatively, the value of S-mode with load by Ba2+ contraction may have been smaller than that of a beating heart, and this may cause some underestimation of systolic volume reduction. These factors may counter each other. Collectively, we speculate capillary volume change of an in vivo beating heart is ≈40%. Finally, this capacitance predominantly resides in the deeper layers because larger reduction of VF in ENDO (by 37%) and MID (by 34%) than in EPI (19%) was observed.
The architecture of LMs, running spirally through the myocardial wall, seems to be compressed in a complicated manner by myocardial contraction. Although we did not differentiate arterioles and venules in the microCT images, except for the GRVs, the overall dynamics of LMs (namely, resistance vessels15) are quite important for evaluating the mechanism of the phasic coronary flow dynamics. A smaller amount (2%) but a greater decrease (by 48%) in VF by cardiac contraction may indicate their predominance as the phasic effective resistance change. According to our direct observation of in vivo beating porcine5 or canine6 hearts, 10% to 20% decreases in diameter of both arterioles and venules in subendocardium occur during systole. Undoubtedly, the increase in systolic arteriolar and venular resistances impedes the retrograde flow to arteries and forward flow to veins from the capillary bed, respectively, keeping the capillary patent during systole. Because the extravascular compressive force decreases almost linearly from subendocardium to subepicardium, the capillary VF changes may decrease from deeper to superficial layers.
Again, 3D evaluation is important for the elucidation of the vascular dynamics, especially for GRV dynamics, the VF change of which has been difficult to observe with 2D image analysis. Neither biomechanical properties nor functional roles of the GRV for coronary flow dynamics have been known. The GRV is characterized as running along the cleavage planes and inpoured by multiple capillaries and smaller venules (Figure 4), giving an appearance which is also called “human hand–like” or “turnip root–like”.16,17⇓ Such a planiform shape with lower intravascular pressure of venous system may indicate higher collapsibility. Ellipsoidal thin-walled tubes are simulated as the most collapsible by extravascular compression,18,19⇓ and indeed we observed systolic compression of the GRVs in the present study. We speculate that the collapsibility of GRVs may function as systolic local flow interruptor to keep the capillary open, in addition to its capacitor function to accommodate blood during diastole.
Critique of Methods and Study Limitations
Our preparations were suitable for microvascular visualization using microCT and CLSM. Overall, the complete filling with the physiologically relevant perfusion pressure was most important for the appropriate vascular volumetric analysis. The possible reasons for the complete filling are as follows: (1) complete washout of the whole blood cells, (2) effectively lower viscosity of the warmed contrast medium, (3) contrast medium injection without chemical fixatives, and (4) our selected gelatin product, which was essential.
Our preliminary studies indicated that 1 cm3 of solidified contrast medium decreased ≈8% in volume after fixation overnight and that the contrast medium–filled rat heart decreased by ≈13%. Despite this shrinkage, we considered it fair to compare the relative VF differences among the 3 different myocardial conditions because all were fixed identically.
Some possibilities for the change in vasomotor tone and myocardial metabolic contraction of the sample heart should be discussed. The vasomotor tone was nearly uniform as maximally dilated, because excess dose of adenosine was administrated and the time elapsed from adenosine administration until the beginning of the rapid solidification of the contrast medium was only ≈10 s. Because the total time to complete each preparation was short enough (≈10 minutes) while oxygenized perfusates with strictly controlled temperature had been administrated, the myocardial metabolic contraction may be minimal. Although the warm contrast medium was administered in a short period of time (≈10 s), intense refrigeration (0°C) followed without delay. Therefore, we assure that our methodology with the rapidly solidified contrast medium is justified for accurate microvascular morphological studies.
In the present study, we evaluated the mechanical effects of myocardial contraction on the coronary microvasculature under unifying perfusion pressure in Langendorff’s mode using quasistatic models, as reported earlier.7,20⇓ Because the blood redistribution from superficial to deeper layers in diastolic arrest21 and opposite directional redistribution in systolic arrest22 were reported in these models, the effects of systole and diastole on the blood volume change may be enhanced, especially in deeper layers.
In conclusion, elucidation of the differences in biomechanical properties between capillaries and pre- and post-capillary microvessels in response to myocardial contraction and relaxation indicated the existence of a longitudinal hierarchy for the rational myocardial perfusion in a cardiac cycle. The capillaries were relatively resistant to systolic extravascular compression because of the increase in pre- and post-capillary resistances together with deformation of the GRVs. This may be beneficial for myocardial oxygen delivery during the cardiac cycle. Nevertheless, the capillaries may function as the effective capacitance to accommodate blood during diastole and eject it out to the veins during systole.
This study was supported by a Grant-in-Aid for Scientific Research (No. B12480269) and a Grant-in-Aid for Encouragement of Young Scientists (No. 12780657) from the Japanese Ministry of Education, Science, Sport and Culture. We thank Prof Masao Fukunaga and Dr Teruki Sone (Department of Nuclear Medicine, Kawasaki Medical School) for supporting the use of the microCT, Chikako Tokuda and Hiroyuki Tachibana (from our laboratory) for technical assistance, and Prof William M. Chilian (Department of Physiology, Medical College of Wisconsin) for excellent academic suggestions.
Movies I, II, and III are available in an online only Data Supplement at http://www.circulationaha.org
- ↵Kajiya F, Tsujioka K, Goto M, et al. Functional characteristics of intramyocardial capacitance vessels during diastole in the dog. Circ Res. 1986; 58: 476–485.
- ↵Hoffman JI, Spaan JA. Pressure-flow relations in coronary circulation. Physiol Rev. 1990; 70: 331–390.
- ↵Spaan JA, Cornelissen AJ, Chan C, et al. Dynamics of flow, resistance, and intramural vascular volume in canine coronary circulation. Am J Physiol. 2000; 278: H383–H403.
- ↵Yada T, Hiramatsu O, Kimura A, et al. In vivo observation of subendocardial microvessels of the beating porcine heart using a needle-probe videomicroscope with a CCD camera. Circ Res. 1993; 72: 939–946.
- ↵Goto M, Flynn AE, Doucette JW, et al. Cardiac contraction affects deep myocardial vessels predominantly. Am J Physiol. 1991; 261: H1417–H1429.
- ↵Judd RM, Levy BI. Effects of barium-induced cardiac contraction on large- and small-vessel intramyocardial blood volume. Circ Res. 1991; 68: 217–225.
- ↵Umetani K, Yagi N, Suzuki Y, et al. Observation and analysis of microcirculation using high-spatial-resolution imaging detectors and synchrotron radiation. Proceeding of SPIE. 2000; 3977: 522–533.
- ↵Fung YC. Coronary blood flow.In: Biomechanics: Circulation. 2nd ed. New York: Springer-Verlag; 1997: 446–513.
- ↵Fung YC, Sobin SS. Elasticity of the pulmonary alveolar sheet. Circ Res. 1972; 30: 451–468.
- ↵Caulfield JB, Borg TK, Abel FL. The effects of systole on left ventricular blood flow.In: Chazou E, Saks V, Rona G, eds. Advances in Myocardiology. New York: Plenum; 1983; 4: 379–393.
- ↵Chilian WM, Marcus ML. Effects of coronary and extravascular pressure on intramyocardial and epicardial blood velocity. Am J Physiol. 1985; 248: H170–H178.
- ↵Chilian WM, Eastham CL, Marcus ML. Microvascular distribution of coronary vascular resistance in beating left ventricle. Am J Physiol. 1986; 251: H779–H788.
- ↵Brown RE. The pattern of the microcirculatory bed in the ventricular myocardium of domestic mammals. Am J Anat. 1965; 116: 335–374.
- ↵Munch DF, Comer HT, Downey JM. Barium contracture: a model for systole. Am J Physiol. 1980; 239: H438–H442.
- ↵Flynn AE, Coggins DL, Goto M, et al. Does systolic subepicardial perfusion come from retrograde subendocardial flow? Am J Physiol. 1992; 262: H1759–H1769.
- ↵Iwanaga S, Ewing SG, Husseini WK, et al. Changes in contractility and afterload have only slight effects on subendocardial systolic flow impediment. Am J Physiol. 1995; 269: H1202–H1212.