Coronary artery wall imaging in mice using osmium tetroxide and micro-computed tomography (micro-CT)


Han Wen, Room B1D416, Bldg. 10, MSC 1061, 9000 Rockville Pike, Bethesda, MD 20892-1061, USA. T: + 1 301 4962694; F: + 1 301 4022389; E:


The high spatial resolution of micro-computed tomography (micro-CT) is ideal for 3D imaging of coronary arteries in intact mouse heart specimens. Previously, micro-CT of mouse heart specimens utilized intravascular contrast agents that hardened within the vessel lumen and allowed a vascular cast to be made. However, for mouse coronary artery disease models, it is highly desirable to image coronary artery walls and highlight plaques. For this purpose, we describe an ex vivo contrast-enhanced micro-CT imaging technique based on tissue staining with osmium tetroxide (OsO4) solution. As a tissue-staining contrast agent, OsO4 is retained in the vessel wall and surrounding tissue during the fixation process and cleared from the vessel lumens. Its high X-ray attenuation makes the artery wall visible in CT. Additionally, since OsO4 preferentially binds to lipids, it highlights lipid deposition in the artery wall. We performed micro-CT of heart specimens of 5- to 25-week-old C57BL/6 wild-type mice and 5- to 13-week-old apolipoprotein E knockout (apoE−/−) mice at 10 μm resolution. The results show that walls of coronary arteries as small as 45 μm in diameter are visible using a table-top micro-CT scanner. Similar image clarity was achieved with 1/2000th the scan time using a synchrotron CT scanner. In 13-week-old apoE mice, lipid-rich plaques are visible in the aorta. Our study shows that the combination of OsO4 and micro-CT permits the visualization of the coronary artery wall in intact mouse hearts.


Coronary artery disease (CAD) is a major health burden on the population today, causing over 400 000 mortalities each year in the United States (Roger et al. 2010). Much of the knowledge about the development of atherosclerotic plaques in the coronary artery comes from genetically modified mouse models such as the well known apolipoprotein E (apoE) knockout models (Plump et al. 1992; Zhang et al. 1992). When it comes to high resolution three-dimensional (3D) imaging of the coronary arteries in intact mouse heart specimens, micro-CT is a valuable technique due to its non-invasive nature and spatial resolution as high as several micrometer over a sufficiently large field of view. Micro-CT was first presented in the early 1980s (Kujoory et al. 1980; Elliott & Dover, 1982; Burstein et al. 1984; Flannery et al. 1987) and has been successfully used in imaging vascular structure in small animals (Avula et al. 1994; Langheinrich et al. 2007; Degenhardt et al. 2010). Over the last three decades, the technology has advanced considerably due to developments in CCD sensors, better reconstruction algorithms and also due to the increasing availability of dedicated scanners for small animal imaging research (Schambach et al. 2010). In current micro-CT, contrast comes from the different levels of X-ray absorption of the object constituents. Typically, structures with high mineral content (such as bones and teeth) have strong X-ray absorption and are highly visible against soft tissue, while different types of soft tissue (such as muscles, blood vessels, etc.) have low contrast relative to each other. Therefore, ex vivo micro-CT studies usually utilize extrinsic contrast agents to allow better differentiation of soft tissue structures.

The purpose of our study is to develop a tissue staining and micro-CT protocol for imaging the walls of the coronary arteries. To date there have been few published micro-CT studies of coronary arteries in intact mouse hearts (Yamashita et al. 2002; Clauss et al. 2006). Clauss et al. (2006) acquired excellent 3D data of casts of mouse coronary systems, which were formed with a radio-opaque polymer blend (Microfil) that hardens within the vessel lumen. Yamashita et al. (2002) performed an elegant study of mouse coronary angiography in vivo and in Langendorff-perfused beating hearts on a synchrotron X-ray beamline, using intra-aortic infusion of iodine contrast agents. Although these protocols make the vessel lumens stand out and can detect sites of lumen narrowing, the ability to image the arterial wall itself can provide a broader range of specific information on atherosclerotic plaques, which is the central topic of CAD.

Osmium tetroxide (OsO4) is an X-ray attenuating compound that binds to soft tissue covalently and recently has been used to enhance the visibility of soft tissue structures in ex vivo CT studies (Johnson et al. 2006; Litzlbauer et al. 2006; Zhu et al. 2007; Faraj et al. 2009). Of particular interest to coronary artery imaging is the high affinity of OsO4 to lipids in tissue, which therefore can highlight plaques in arterial walls with the appropriate level of staining. OsO4 has been used to detect plaques in pig coronary artery specimens by Zhu et al. (2007) and in dissected aortic specimens of apoE knockout mice by Goel et al. (2008). However, we are not aware of such studies in whole mouse hearts up to date.

The ability to image arterial walls in intact mouse hearts should provide systematic information on the development of plaques and provide guidance to subsequent sectioning of the specimen for detailed pathology studies by light or electron microscopy. In this study, we optimized an OsO4 perfusion protocol in the mouse heart under physiological pressure, which enabled vessel wall imaging in the main coronary arteries and distal branches down to 45 μm diameter in whole mouse heart specimens. We performed micro-CT studies in both wild-type and apoE knockout mice on a table-top commercial micro-CT scanner and a custom micro-CT scanner on a synchrotron beamline. We also performed light microscopy of histology sections to identify the underlying structure that gave rise to the appearance of the artery walls in the micro-CT images.

Additionally, the general anatomy of the coronary arteries can be extracted from osmium-stained micro-CT images. However, automated tracking of the arteries in 3D space is less reliable than manual tracing. We observed two variants of the general disposition of the coronary arteries in both wild-type and apoE knockout mice.



Male and female C57BL/6 (n = 45) wild-type mice of ages between 5 and 25 weeks were procured from Harlan Laboratories (Frederick, MD, USA). Mice were housed and maintained at our facility and utilized in the study in accordance with an Animal Care and Use Committee approved protocol. All mice were fed a standard chow diet. Heart specimens of 39 wild-type mice were scanned by micro-CT, using both a table-top scanner and a synchrotron scanner, as described below.

Five-week-old male and female apolipoprotein E knockout mice (apoE−/− Strain: B6.129P2-apoEtm1Unc/J, Stock: 002052, n = 45) were provided by The Jackson Laboratory and raised at our institution on either a protein diet (16% Protein diet T.2016.15; Harlan Laboratories) or a western diet (42% fat diet, TD.88137; Harlan Laboratories). Heart specimens of 21 apoE mice aged between 6 and 13 weeks were scanned by micro-CT.

Prior to undergoing surgery, each mouse was weighed and given an intraperitoneal (i.p.) injection of 0.5 mL of heparin at a dosage of 50 IU per 25 g body weight. The heparin was allowed to circulate throughout the mouse for 5 min. After 5 min, the mouse was given a lethal i.p. injection of 200 mg kg−1 body weight of pentobarbital.


All materials and supplies for the surgical procedures were provided by the Laboratory of Animal Medicine and Surgery (LAMS) of the NIH. These supplies included heparin, pentobarbital, saline, saline-based 10% formalin solution, and 24-gauge bulb-tipped catheters.

A 0.2 m NaPO4 buffer was prepared by mixing 20 mL of 0.2 m monobasic sodium phosphate with approximately 80 mL of 0.2 m dibasic sodium phosphate to obtain a pH of 7.4. OsO4 1% was prepared in a fume hood by combing 2.0 mL of 4% OsO4 (Electron Microscopy Sciences, Hatfield, PA, USA) with 2.0 mL of deionized H2O and 4.0 mL of 0.2 m NaPO4 buffer.

Osmium tetroxide staining protocol

Once the mouse was euthanized, an incision was made from the pelvic region up to the throat. The chest cavity was opened to expose the heart. The descending aorta was located and the fat and tissue surrounding the aorta were dissected away. A suture was placed on the posterior end of the vessel. A 24-gauge, bulb-tipped catheter was then inserted into the descending aorta, retrograde toward the heart. The catheter was secured in place with a suture. The vena cave was then cut open to allow excess fluid to run out of the heart. Using a gravity-fed i.v. set-up, a heparinized saline solution was perfused through the catheter for 20 min from a height of 2 ft, equivalent to 45 mmHg pressure. Then, using the same i.v. setup, a saline-based 10% formalin solution was perfused through the heart for 20 min.

Following the formalin perfusion, the carotid arteries were tied off with a single suture. The thymus and surrounding tissue were dissected away from the heart. The heart, lungs and descending aorta, along with the inserted catheter, were then dissected from the mouse as a single unit and placed in a formalin-filled 20-mL scintillation vial for 120 min. The purpose of retaining the lungs in the circulation loop is that in the subsequent retrograde perfusion of various solutions into the coronary arteries, the solutions were infused into the aorta through the catheter, then flowed retrograde through the left heart chambers, the lungs, and into the right heart chambers and exited out of the vena cave. The lungs provided the flow resistance for maintaining pressure upstream at the root of the aorta, which helped push the solutions into the coronary arteries.

Following the 120-min formalin fixation, the heart was flushed with a manual injection of 3.0 mL of 0.1 m sodium phosphate (NaPO4) buffer (pH 7.2–7.4) through the catheter. The heart was allowed to soak in a 20-mL scintillation vial with 15 mL buffer for 10 min. Subsequently, 2.0 mL of 1% OsO4 was manually injected through the catheter in 0.5 mL increments every 3 min. After the last 0.5-mL injection, the heart was left to soak in another vial of 15 mL of 0.1 m buffer for 60 min. Following the 60-min soak, the heart was flushed two times with 3.0 mL of buffer injected by hand using a 10 cm3 syringe. The extra soaking and flush were done to remove any remaining free OsO4 in the heart chambers and blood vessels. The heart was then stored in a tightly sealed 20-mL scintillation vial of 0.1 m buffer in a refrigerator until it was scanned.

Micro-CT imaging

Two groups of heart specimens underwent micro-CT scanning. The first was the wild-type group of 39 specimens and the second was the apoE knockout group of 21 specimens. Each specimen was scanned twice, once on a table-top scanner and once on a synchrotron CT scanner.

Prior to being scanned, each heart was flushed with a manual injection through the catheter of 3.0 mL of 0.1 m NaPO4 buffer injected over approximately 3 min. The catheter was removed and the heart was allowed to dry (approximately 5 min) on absorbent padding. The heart was then placed in a segment of Kapton tubing (9.5 mm diameter, 127 μm wall thickness, 19 mm length), sealed with Parafilm and secured on the sample stage of the CT scanner. The table-top scanner was a Skyscan 1172 micro-CT system of the NIH Mouse Imaging Facility. It was equipped with a Hamamatsu 100/250 X-ray source and a Hamamatsu 10 megapixel camera (camera pixel size 11.54 μm). Projection images were acquired as 3240 × 2096-pixel frames with an 1178-ms exposure every 0.2° over 180° rotation with camera binning of 1 × 1. Additional imaging parameters were pixel size of 4.97 μm, averaged over 12 frames to minimize noise, random-movement setting of 4, 60 kVp per 10 W source setting, source-sample and sample-detector distances of 95.0 and 220.5 mm, respectively. The run time for each scan was approximately 335 min.

Raw projection images were reconstructed using the nrecon software program provided by Skyscan. Images were corrected using three settings: misalignment compensation, ring artifact reduction, and beam hardening. The reconstructed volumetric data were then converted to DICOM format and inspected using either Vitrea (Vital Images, Inc, Minnetonka, MN, USA) or Inveon Preclinical Research Software (Siemens Preclinical Solutions, Knoxville, TN, USA). The final spatial resolution of the reconstructed volumes was approximately 10 μm.

In addition to the table-top scanner, all specimens were also scanned on a micro-CT system of a synchrotron X-ray beamline (Sector 2-BM, Advanced Photon Source, Argonne National Laboratory). Due to the high intensity of the synchrotron source, a complete CT scan took 10 s and included 1200 projections. Data processing was performed on the multiprocessor cluster of the beamline and used a back-projection reconstruction program developed by the synchrotron facility. The spatial resolution of the reconstructed volumes was also 10 μm.


General disposition of the coronary arteries

The coronary arteries were visible in virtual dissections of the 3D volume. Figure 1 shows the left and right coronary ostia in the aortic valvar complex of a 25-week C57BL/6 mouse. Both left and right coronary arteries arise at the base of the aortic valvar sinus. Figure 2 shows two options by which one can obtain the disposition of the coronary arteries in 3D space from the volume CT data. The first is automated segmentation of the artery lumens using straightforward intensity thresholding of the raw data (product software of the table-top CT scanner). This method reliably delineated the proximal trunks of the main arteries but frequently failed to extend to more distal segments. The second option is manual tracking of the coronary arteries in serial virtual dissections until they become invisible. This option reliably traced the coronary arteries into distal segments and branches, although it was time-consuming.

Figure 1.

 Illustration of the origins of the coronary arteries in the aortic valvar complex of a 25-week-old C57BL/6 mouse. Data were from the table-top CT scanner. (A) In this longitudinal cross-section along the axis of the aorta, the ostium of the left coronary artery is indicated by the arrow, and the aorta lumen is indicated by the • symbol. The initial segment of the left coronary artery is also visible. ‘LV’ and ‘RV’ mark the left and right ventricles. (B) In this longitudinal cross-section the ostium of the right coronary artery (arrow) and the initial segment of the artery can be seen.

Figure 2.

 Illustration of the procedure to obtain the three-dimensional disposition of the coronary arteries. Data were from the table-top CT scanner of an 11-week-old C57BL/6 mouse. The top panel shows a virtual dissection through the CT volume that reveals the left coronary artery and its first branching point. From such volume data, automated segmentation of the lumens of the arteries by intensity thresholding can reliably identify the proximal trunks of the main coronary arteries, as shown in the lower left panel (right coronary artery in pink, left coronary artery in blue, septal branch of the LCA in yellow). Manual tracing in serial cross-sections is needed to identify the positions of the distal portions of the arteries. The manual inputs in many cross-sections are combined in 3D space to yield the trajectories of the arteries and branches. An example is shown in the lower right panel (RCA, right coronary artery; LCA, left coronary artery; LSA, septal branch of the left coronary artery).

By the above procedures we observed two variants of the general anatomy of the coronary arteries in both wild-type C57BL/6 mice and apoE knockout mice. The first is that there are two septal arteries originating from the left and right main arteries, respectively, and the septal artery from the left main is dominant. In the second variant only one septal artery is present, originating from the right coronary artery. These two variants are a subset of the anatomical variants that Hu et al. (2005) observed in their histology study of 60-week-old C57BL/6 wild-type and apoE knockout mice. In both wild-type and apoE knockout mice, Hu et al. observed two additional variants, including a single septal artery arising from the left coronary artery and a single septal artery arising directly from the aorta. In C57BL/6 wild-type mice, Icardo & Colvee (2001) in a vascular corrosion cast study and Clauss et al. (2006) in a micro-CT study observed one of the two variants we observed, namely a single septal artery arising from the right coronary artery. in Also in two strains of C57BL/6 mice, Fernandez et al. (2008) observed the same variants we observed and an additional variant in which a single septal artery originated directly from the aorta.

To show a longitudinal cross-section of a curved coronary artery, we used a ‘manual montage’ method where we recorded images of planar dissections in a series of points along the length of the artery. Each dissection is along the axis of the vessel at that point, and adjacent dissection points were chosen to be close enough such that the vessel segments revealed in the dissections have some overlap. The images of the longitudinal dissections were then cropped to retain only the coronary artery segments, and these segments were assembled manually in the ‘montage’ to give effectively a curve-following longitudinal cross-section of the length of the artery.

Figure 3 shows a montage of coronary arteries in a 5-week-old C57BL/6 wild-type mouse. Although the LCA, its septal branch and the RCA arteries are clearly visible, there is a minute septal branch of the RCA that is difficult to see. As the mice got older, their coronary arteries became more visible in the CT images. Figure 4 is a montage of the coronary arteries of a 25-week-old wild-type mouse. Here the septal branch of the RCA is visible in addition to the three arteries seen in the 5-week-old mouse.

Figure 3.

 Montages of longitudinal sections showing the coronary arteries of a 5-week-old C57BL/6 mouse (from left to right: right coronary artery, septal-conal artery from the left main artery and the left coronary artery). The vessels (*) originate from the aortic valvar complex (•). Note that the hyperintense circular region within the aorta is the catheter tip. The individual cross-sections are maximal intensity projections (MIP) over 15-μm-thick sections. The montage is constructed manually (see text for details). Scale bar indicates 100 μm. Data were from the table-top CT scanner.

Figure 4.

 Montages of longitudinal sections showing the coronary arteries of a 25-week-old C57BL/6 mouse (from left to right: right coronary artery, septal branch of the RCA, septal branch of the LCA and the LCA). The aorta and coronary artery lumens are indicated by • and *, respectively. Scale bar: 100 μm. Data were from the table-top CT scanner.

Sizes of coronary arteries visible under micro-CT

To illustrate the size of the coronary arteries that can be seen in this micro-CT study, the vessel diameters in the left and right coronary systems of a 25-week C57BL/6 mouse are shown in Figs 5 and 6. Figure 5(i,ii) shows the measurements in the left coronary artery (LCA) and the septal-conal branch of the left main artery, respectively. In Fig. 5(i), the diameters of the visible branches of the LCA ranges between 280 and 40 μm. The artery wall appears distinct from the surrounding tissue down to the third branch and can be seen in segments of more distal branches (Fig. 5i,G). We observed several branches of relatively large take-off angles from their main trunks, including one at 127° (Fig. 5i,C) and one at 101° (Fig. 5i,F).

Figure 5.

 The left coronary artery system of a 25-week-old C57BL/6 mouse as seen in the volume data from the table-top CT scanner. (i) The left coronary artery and cross-sections/magnified views at various locations (A–H) along the LCA. There is high level of contrast staining at the apex of the bifurcation in panel (A) (20-μm-sized region indicated by arrow). Dotted arrows in figures (B, D and H) indicate small-caliber branches directly connected to arteries of substantially larger diameters. (ii) Septal artery arising from the left main artery and cross-sections at three different locations (A–C) along its length. In this mouse, this is the main septal artery. Cross-section views were acquired using 50-μm average intensity projections. Scale bars: 100 μm. Main and branch vessel lumen are indicated by * and •, respectively.

Figure 6.

 The right coronary artery system of a 25-week-old C57BL/6 mouse as seen in the volume data from the table-top CT scanner. (i) The right coronary artery and cross-sections/magnified views at various locations (A–F) along the RCA. (ii) Septal artery arising from the RCA and cross-sections at three different locations (A–C) along its length. Main and branch vessel lumen are indicated by * and •, respectively. Scale bars: 100 μm.

Figure 5(ii) shows details of the septal-conal branch arising from the left main artery. The diameters of the visible branches of this artery ranged between 161 and 55 μm. The artery wall is visible up to a distal bifurcation point.

Figure 6 shows a montage of the right coronary artery system of a 25-week-old C57BL/6 mouse. Figure 6 shows the main RCA and the septal branch, respectively. Figure 6(i) shows that the visible portions of the main RCA and its branches (excluding the septal branch) have diameters as small as 43 μm. The artery wall is visible over the proximal 1.5 mm length of the main RCA (Fig. 6i,A–C). The septal branch of the right coronary artery is visible down to a diameter of 55 μm (Fig. 6ii,A). The vessel wall is visible over the first 0.6 mm of its length.

Visualization of coronary artery wall

An advantage of the OsO4 contrast agent is that it is permanently retained in tissue, which provides stable enhancement of soft tissue structures such as the coronary artery wall. Figures 3–6 show that in the left and right main coronary arteries as well as in their septal branches, the artery wall is visible in both longitudinal and transverse cross-sections, starting from the proximal end down to distal locations where the vessel diameters decrease to below 100 μm. The vessel wall remains visible in some of the more distal branches which have higher levels of contrast staining.

Figures 5 and 6 show that it is easier to delineate the coronary artery wall in the left and right coronary arteries than in the septal branches. In the distal segments of the coronary arteries, the wall structure is more clearly seen in the LCA and its septal branch than in those of the right side circulation. In the LCA of a 25-week-old wild-type mouse, the intimal-medial layer was measured to be approximately 20 μm thick from the origin to the third branching point (Fig. 5i,A–E), and as much as 36 μm thick in a distal branch (Fig. 5i,G). In the septal-conal artery from the left main artery, the measured wall thickness ranged between 33 μm proximally (Fig. 5ii,A) to 16 μm (Fig. 5ii,B) just before it bifurcated into two equal-sized branches. The observed wall thickness of the RCA was approximately 37 μm at the proximal end and decreased to approximately 20 μm at 1.5 mm downstream (Fig. 6i,A–C) before it became indistinguishable from the surrounding tissue. The intimal-medial layer of the septal branch of the RCA is visible in the proximal 0.6-mm segment (Fig. 6ii,A,B). Its thickness ranged between 37 and 20 μm.

In Figures 3–6, the coronary vessel wall frequently appears to be ‘separated’ from the surrounding tissue: the wall is identifiable as a bright band surrounding the lumen, and a gap of low brightness separates the wall from the outer tissue of high image intensity. To identify the underlying structure that gives rise to this appearance, we performed microscopy of hematoxylin and eosin-stained 5-μm-thick paraffin-embedded histology sections of two mouse hearts. Figure 7 shows micro-CT cross-sections with different degrees of separation between the bright signals of the wall and the surrounding tissue, and their corresponding histology sections. In the histology micrographs the osmium contrast agent darkens the tissue, and cell nuclei are stained blue. The micrographs show that the artery wall contains several layers of cells which are stained with osmium. The wall is surrounded by myocytes that are also osmium-stained. In some areas the wall is in continuous contact with the myocytes without any gap, whereas in other areas a segment of the wall is detached from the myocytes with a gap in-between. The gap is devoid of tissue. In yet other areas, the vessel wall is almost completely detached from the myocytes. Thus the different degrees of separation in the micro-CT images result from the extent of detachment of the vessel wall from the surrounding myocytes.

Figure 7.

 Micro-CT cross-sections of coronary artery wall showing different degrees of separation between the wall and the surrounding tissue (left column), and their corresponding histology micrographs (right column). The histology sections are hematoxylin and eosin-stained. Nuclei are stained blue and osmium-loaded tissue appears darker. (A) The artery wall is in contact with the surrounding tissue without any space in-between. (B) Histology shows that parts of the artery wall are physically detached from the surrounding myocytes, resulting in a dark band between the wall and the surrounding tissue in the micro-CT image. (C) In this case the vessel wall is almost completely detached from the myocytes and a similar morphology is seen in the micro-CT image. Data were from the table-top CT scanner.

Figure 8 shows a montage view of longitudinal cross-sections from the synchrotron scan of the LCA of a 6-week-old female apoE knockout mouse, together with transverse cross-sections at various levels of the LCA tree. The clarity of the coronary vessels is comparable or better than the clarity obtained with the table-top scanner (Figs 5 and 6), even though the scan time was 1/2000th of the table-top system. In this mouse, blood clotting occurred in distal branches of the LCA which lead to hyperintense volumes in the lumens of the branches. These can be seen at and below the level of cross-section ‘d’.

Figure 8.

 Montage of longitudinal cross-sections of the LCA of a 6-week-old female apoE−/− mouse, and transverse cross-sections along the main trunk and branches. Data were acquired on a synchrotron CT scanner in 10 s, or approximately 1/2000th the time it took on the table-top scanner. Note that in the branches downstream from cross-section (D), blood clots occurred during the fixation process which created the highly stained bright volumes in the lumen. Scale bar: 100 μm.

Visualization of aortic plaques in apoE−/− mouse hearts

In the oldest apoE mice of our study (13 weeks) we observed plaques in the aortic wall extending up from the valvar complex and in the aortic arch, in agreement with the osmium-stained micro-CT study of Goel et al. (2008). The data from a 13-week-old apoE−/− female mouse are shown in Fig. 9. Figure 9A is a transverse cross-sectional view of the aorta at the valvar complex, showing significant plaque build-up. In the main coronary arteries as well as distal branches of the apoE mice, we did not see hyperintense areas that would indicate the presence of lipid-rich plaques, or any narrowing of the vessel lumens. This was the case in the scans done on both the table-top scanner and the synchrotron beamline.

Figure 9.

 Longitudinal and transverse (inset panel a) cross-section views of the aorta of a 13-week-old female apoE mouse on a Western diet. Data were from the table-top CT scanner. Plaque deposits in the aortic wall are indicated by hollow triangles. The dotted ‘a’ line marks the location of the transverse cross-section shown in inset (A). The solid arrows indicate the wall of the aorta. Hyperintense areas outside the aorta are osmium-stained pericardial fatty tissue. Scale bar: 100 μm.


We demonstrated that micro-CT in combination with OsO4 staining allows visualization of the coronary artery wall in mouse hearts in all the main segments and some distal branches. Starting at the age of 5 weeks, the coronary walls for the left and right coronary arteries are distinguishable down past several branch points. Septal coronary arteries are difficult to delineate at 5 weeks but can be easily seen at 13–14 weeks. Where a septal artery arises from the left main artery, a septal branch from the RCA is also present. The right septal branch is typically shorter than the left septal artery but is visible by 25 weeks. Although in most of our wild-type mice two septal arteries were seen branching from the left and right coronary arteries, respectively, we also observed in several mice a second anatomical variant where a single septal artery originated from the right main coronary artery, and the septal branch of the left main was absent in these mice. It should be noted that for studying the general disposition of the coronary arteries, the radio-opaque vascular cast protocol (Clauss et al. 2006; Langheinrich et al. 2007) is more suitable than our tissue-staining protocol, since the X-ray absorption of the cast material is drastically higher than soft tissue. In micro-CT images of vascular casts the contrast between the cast material and the surrounding tissue is binary-like, allowing automated segmentation of blood vessels down to small-caliber branches by setting a simple intensity threshold in the volume data. In comparison, the osmium staining protocol we used is more suitable for seeing structures in soft tissue.

Our micro-CT images showed that the hyperintense coronary wall is often separated from the hyperintense myocardium by a dark band. Micrographs of histology sections show that it reflects physical detachment of the vessel wall from the adjacent myocytes, which likely occurred during the fixation process. The detachment is also seen in the histology micrographs of mouse coronary arteries by Hu et al. (2005). In contrast, Zhu et al. (2007) in micro-CT of pig coronary arteries saw similar dark areas, which they showed by histology to be perivascular connective tissue. The likely reason for the difference between mouse and pig is that in larger animals such as pigs, the coronary arteries are on the epicardial surface and surrounded by connective tissue. The connective tissue is elastic, so any deformation introduced by the fixation process does not result in detachment of the coronary artery wall from the perivascular tissue. In mice the coronary arteries are intramyocardial and are mostly in direct contact with the myocytes, without a thick connective tissue layer in-between. Under physiological conditions the internal pressure of the coronary arteries is sustained by the systolic and diastolic blood pressure. In the specimens internal pressure was note present, and the artery walls could shrink or collapse and separate from the adjacent myocardium. This notion is consistent with the irregular shape of the coronary artery cross-section in Fig. 7C. We also saw fracturing of the myocardium in some histology slides (Fig. 7B,C). The same was seen in the histology study of Hu et al. (2005) and is likely an artifact of the histology sectioning process.

We found that the optimal way to facilitate flow of the OsO4 agent in the coronary circulation was to utilize a retrograde perfusion technique. By applying a heparin flush before and immediately after euthanasia, we minimized clotting issues and eased the injection of the staining agent. The use of physiological pressures up to fixation should permit us to preserve plaque structures; since the tissue stiffens after fixation, a manual approach is used for subsequent injections of the OsO4 and buffer washes. The presence of formalin ensures that the coronary arteries do not collapse. Since it has been well established in electron microscopy studies that OsO4 penetration is not very deep, a slow injection rate is necessary to ensure that larger regions of the coronary circulation are exposed to the staining agent. Subsequently, allowing the heart to soak in a vial of buffer followed by extra saline flushes helped remove any free OsO4 that might have remained in the heart chambers and the blood vessels.

Although we saw extremely bright lipid-rich plaques in the aorta of the apoE mice at age 13 weeks, we did not see any lesions in the coronary arteries. The likely reason is that our mice were not old enough to develop coronary plaques. Indeed, Hu et al. (2005) performed a comprehensive histology study of 60-week-old apoE mice and found that the aortic plaques extend to the origins of the coronary arteries or a short distance down the main trunks at that age. Saraste et al. (2008) observed 30–97% luminal narrowing in the LCA of 58–72-week-old LDLR/ApoB48 double knockout mice. We have not found reports of coronary lesions in younger mice in the literature. Currently, we are raising apoE−/− mice of older ages for this purpose.

In terms of the time required to scan a heart specimen, the main limitation of our study is the power of the X-ray tube in the table-top micro-CT system, which is limited to 10 W. In comparison, the X-ray tube of a clinical C-arm operates at tens of kilowatts. As a result of the low power of the X-ray source, each heart specimen took 6.5 h to scan on the table-top scanner. Although synchrotron X-ray sources (Yamashita et al. 2002) produce several orders of magnitude brighter X-ray beams, permitting rapid scans of high signal intensity, they are far less available than compact X-ray sources. In contrast, recent developments in table-top laser-driven coherent X-ray sources are approaching synchrotron beam intensity (Kneip et al. 2010) and may soon offer ideal sources for compact micro-CT systems. An additional technical challenge with high resolution volumetric studies is the approximately 10–15 Gb of data of each heart specimen, which makes the entire process of data processing non-trivial. We are currently engaged in developing software on a parallel computing cluster to be able to handle the large data volumes.


As a first study of the coronary artery wall in intact mouse heart specimens, our results show that micro-CT in combination with OsO4 tissue staining is able to visualize the arterial wall. This imaging protocol may potentially be a useful tool for detecting plaques in the coronary arteries of mouse CAD models.


We are grateful to Dr. Mathew P. Daniels of the Electron Microscopy Core Facility, NHLBI, and Randy Clevenger at the Laboratory of Animal Medicine and Surgery, NHLBI, for helpful discussions on the contrast infusion and surgical protocols.

Conflicts of interest and source of funding

This research was funded by the Division of Intramural Research of the National Institutes of Health (Project Number HL006142-01 to H.W.). Use of the Advanced Photon Source at Argonne National Laboratory was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357. No conflicts of interest, financial or otherwise, are declared by the author(s).

Author contributions

V.M.P. designed the experiments, aided in mice dissections, acquired micro-CT data, built the parallel cluster to reconstruct the data, wrote the image processing software and used it to reconstruct the data, segmented the data, partially analyzed the histology slides and wrote the manuscript. M.K. performed the mice dissections and the osmium staining and acquired micro-CT data. DD aided in the micro-CT acquisition and segmentation. E.M. aided in writing up the image processing software and reconstructing the data. X.X. aided in setting up the synchrotron CT scans at Argonne National Laboratory and processing the data. M.Y.C. aided in segmenting the mouse coronary data and helped with the segmentation software. Z.X.Y. did the histology sectioning and hematoxylin and eosin staining of the paraffin-embedded sections. P.C. aided in developing the staining protocol and generating the media for this purpose. K.J. was responsible for the dissections and catheter insertions and isolating the mice hearts. H.W. designed the experiments, aided in synchrotron CT acquisition, aided in interpreting the images, analyzed the histology slides, wrote the manuscript and is responsible for the general quality of the study.