|Home | About | Journals | Submit | Contact Us | Français|
The bone marrow contains a variety of blood vessels that have different functions in bone marrow maintainance and hematopoiesis. Arterioles control the flow of blood into bone marrow compartments, and the sinusoids serve as a conduit to the blood stream and as niches for megakaryocyte development. Most studies of bone marrow vasculature, including studies quantifying changes in the marrow vascular by microvascular density, do not differentiate between different types of marrow vessels. Recognizing changes in different types of blood vessels following chemotherapy exposure or during leukemia development has important physiologic implications. We hypothesized that the functional heterogeneity of marrow vasculature could be recognized using functional markers such as Tie-2 expression or uptake of DiI-AcLDL.
When transgenic mice with GFP expressed downstream of the Tie-2 promoter were injected with acetylated low-density lipoprotein (Ac-LDL), Ac-LDL was specifically endocytosed by sinusoids, and Tie-2 expression was more pronounced in the arteries, arterioles, and transitional capillaries. Combining these two functional endothelial markers and using confocal microscopy to obtain three dimensional images, we identified transitional zones where arterioles emptied into the sinusoids. Alternatively, co-injection of lectin with DiI-Ac-LDL has a similar result in normal mice as seen in Tie2/GFP mice, and can be used to differentiate vessel types in non-transgenic mice.
These results demonstrate that bone marrow vasculature is functionally heterogeneous. Methods to study changes in the marrow vasculature using microvascular density or quantifying changes in the vascular niche need to take into account this heterogeneity.
Bone marrow is the major hematopoietic organ in adult mammals. Bone marrow sinusoids are specialized capillary beds that separate the synthetic bone marrow from the blood stream. Sinusoids control cellular traffic in and out of the marrow parenchyma. Sinusoidal endothelial cells have no basement membrane, and the individual cells are fenestrated.1, 2 Because of the critical role that bone marrow sinusoids are thought to play as hematopoietic niches and conduits to the circulation, 3–5 distinguishing sinusoids from other vessels when describing and quantifying changes in the marrow vasculature is critically important.
Recent studies suggesting that the bone marrow sinusoids serve as niches for hematopeoitic stem cells, hematopoietic progenitors and megakaryocyte precursors, have led to studies focused on the effects of various stresses on the bone marrow vasculature. Furthermore, a number of studies have demonstrated changes in bone marrow microvascular density that occur in diseases such as leukemia, multiple myeloma, and myelodysplasia. It is not clear how these stresses induce marrow angiogenesis or what particular types of vessels are increased or decreased in these conditions.
Because of their critical role as a vascular niche, we were interested in developing methods to specifically identify bone marrow sinusoids. We tested a number of previously described immunohistochemical methods to identify the bone marrow sinusoids, and found that most of these antigens were either not expressed or were unable to distinguish sinusoids from other types of marrow vessels.
Studies in the 1980s established that reticuloendothelial sinusoids 6 in the bone marrow, liver and spleen endocytosed Ac-LDL conjugated to the fluorescent dye 1,1'-dioctadecyl -3,3,3',3'-tetramethyl-indocarbocyanine perchlorate (DiI-Ac-LDL),7 and that this in vivo uptake was relatively specific to sinusoidal beds.7
Tyrosine kinase with immunoglobulin and epidermal growth factor homology domains-2 (Tie2) is a receptor tyrosine kinase for the angiopoietin-1. Tie2 is expressed by vascular endothelial cells. 8,9 Using transgenic mice expressing reporter genes downstream from the Tie2 promoter to identify blood vessels in the mesentery, researchers discovered that the Tie2 promoter is much more active in the arteriolar than the venular side of the vascular system.10
We hypothesized that the functional heterogeneity of the marrow vasculature could be recognized using functional markers such as Tie-2 expression or uptake of DiI-AcLDL. We have demonstrated that endocytosis of DiI-Ac-LDL and Tie-2 promoter activity distinguishs sinusoidal endothelium from presinusoidal arterioles and transitional capillaries. Distinguishing marrow sinusoids from other types of marrow vessels is a prerequisite to mechanistic studies related to marrow angiogenesis and the fate of the vascular niche.
Six to eight week old C57/BL6 and FVB mice and transgenic C57BL/6-TgN(ACTBEGFP)1Osb (GFP) donor mice11 and Tg(TIE2GFP)287Sato/J (Tie-2/GFP) mice on the FVB background were purchased from Jackson Laboratories (Bar Harbor, Maine). Both male and female mice were used in these studies. These studies were approved by the University of Florida Institutional Animal Care and Use Committee.
Marrow cells were flushed from the marrow cavity with PBS. Erythrocytes were lysed with ammonium chloride (0.15M of NH4Cl, 10mM KHCO3, 0.1mM Na2EDTA pH 7.4). Cells were incubated with 10ug/ml of DiI-Ac-LDL in DMEM medium at 37°C for 4 hours. The cells were washed twice with PBS and labeled with phycoerythrin conjugated antibodies for CD11b, GR1, B220, c-Kit and CD31 (BD Biosciences).
Healthy or irradiated C57BL/6, GFP or Tie2/GFP mice were injected in the retro-orbital plexus with 1ug per gram body weight of DiI-Ac-LDL (Biomedical Technologies). Four hours after injection, animals were sacrificed. Marrow cells were flushed and labeled as in the in vitro experiments.
Four hours after retro-orbital injection with DiI Ac-LDL, animals were sacrificed. Bones were fixed in freshly prepared 4% paraformaldehyde in PBS for 24 hours. Whole bones were sectioned using the Cryojane Tape Transfer System (Instrumedics Inc. NJ). Alternatively, the marrow core was removed by opening the marrow chamber with scissors under the dissecting microscope and carefully lifting the marrow out of the bone. The marrow core was visualized directly under the fluorescent microscope.
DiI-Ac-LDL labeling was performed in the same way as described above. 5 minutes before sacrificing the animal, 5ug per gram body weight of FITC conjugated lectin from Lycopersicon esculentum (tomato) (Vector Lab), or from wheat germ (MP Biomedicals) was injected into the retro-orbital plexus. Bone marrow was processed the same way as for DiI-Ac-LDL labeled Tie2/GFP marrow.
DiI-Ac-LDL labeled bone marrow sections were incubated with anti-CD31 or anti-VWF antibody at 4°C overnight. Donkey anti-rat secondary antibody conjugated to Alexa-Fluor 488 (Molecular Probes) was used for fluorescent detection.
Paraffin embedded sections were treated with DAKO Target Retrieval Solution at 95°C for 20 minutes. Rat anti-mouse MECA-32 antibody 1:10 (Pharmingen) was added to the sample and incubated at 4°C overnight. Staining was detected using an ABC Elite Kit with diaminobenzidine as the chromagen (Vector Labs). Slides were counter-stained with hemotoxylin.
Light microscopy and fluorescent images were captured on an Olympus BX 51 microscope equipped with Optronics Magnafire digital camera system. Confocal imaging and lambda scanning were performed on a Leica TCS SP2 AOBS spectral confocal microscope.
DiI-Ac-LDL has been used as an endothelial marker for over twenty years.12 However, DiI-Ac-LDL uptake is not specific when used in vitro, because other cells such as macrophages that are equipped with scavenger receptors also endocytose the compound. When flushed bone marrow cells are exposed to DiI-Ac-LDL, a large proportion of the cells endocytosed DiI-Ac-LDL as demonstrated by flow cytometric analysis (Figure 1A). There were two DiI-Ac-LDL positive populations identified, one having a higher Ac-LDL intensity (Ac-LDL high) than the other one (Ac-LDL low). Within the DiI-Ac-LDL positive population, 40% expressed integrin alpha M (CD11b or MAC-1). The majority of these cells were in the DiI-Ac-LDL high population. Only 12% of the cells within the CD11b positive population endocytosed DiI-Ac-LDL. Cells expressing GPI linked protein Ly6A (Gr-1), a marker for granulocytes, had a similar staining pattern, with 13% uptaking DiI-Ac-LDL. Interestingly, cells expressing the CD45 isoform and B-cell marker CD45R (B220) also endocytosed DiI-Ac-LDL, but these cells were in the DiI-Ac-LDL low population. There were two well separated B220 positive populations. The B220high population included mature B cells and the B220low population included pro-B, pre-B and immature B cells. 13 The B220low population endocytosed DiI-Ac-LDL but not the B220high population. Platelet endothelial cell adhesion molecule-1 (PECAM-1, CD31) expressing cells endocytosed Ac-LDL. However, the CD31 positive, DiI-Ac-LDL positive cells were not endothelial cells. This point will be illustrated later in this manuscript. CD31 is also expressed on many leukocyte subsets including monocytes and erythrocyte progenitors.14 The stem cell marker, c-kit (CD117), is known to be associated with hematopoietic stem and progenitor cells. Some c-kit positive cells also endocytosed low levels of DiI-Ac-LDL. This low level of DiI-Ac-LDL uptake appeared as a well separated group from the DiI-Ac-LDL high and the DiI-Ac-LDL negative population. The reason behind the low level DiI-Ac-LDL endocytosis in vitro by various types of hematopoietic cells is under investigation. Never-the-less, these data demonstrate that in vitro exposure to DiI-Ac-LDL leads to a large variety of other cell types besides endothelial cells and macrophages endocytosed DiI-Ac-LDL.
In contrast to in vitro exposure, when mice were injected with DiI-Ac-LDL 4 hours before sacrifice, very few bone marrow cells in the prepared single cell suspension had endocytosed DiI-Ac-LDL (Figure 1B). In particular, there were very few CD11b or GR-1 positive granulocytes that had endocytosed DiI-Ac-LDL. We interpreted this result as suggesting that DiI-Ac-LDL remained intravascular, and exposure of macrophages and granulocytes was prevented by the endothelial barrier.
The lack of DiI-Ac-LDL uptake in the flow cytometric analysis was not because the endothelial cells failed to endocytose DiI-Ac-LDL, as can be seen in Figure 2. When marrow cells were obtained by flushing the bone marrow cavity, endothelial cells that had endocytosed DiI-Ac-LDL were either destroyed by the shear forces of flushing, or were still connected to each other and discarded as cell clumps. The result was that there were no endothelial cells labeled by DiI-Ac-LDL present in the single cell suspensions. Attempts to disaggregate the bone marrow sinusoidal endothelial cells with enzymatic digestions such as collegenase and dispase led to similar results.
These results demonstrate that DiI-Ac-LDL uptake is not a specific marker of endothelial cells in cultured murine bone marrow cells. Furthermore, it demonstrates that flow cytometry can not be easily used to analyze bone marrow endothelial cells, as the endothelial cells are not present in single cell suspensions.
The analysis of whole bone marrow cores from DiI-Ac-LDL injected mice was in striking contrast to the single cell suspension of the flow cytometric results. The marrow core contained abundant DiI-Ac-LDL labeled sinusoids forming a network of irregular tubules evenly distributed throughout the bone marrow (Figure 2A). When imaged by confocal microscopy, the detailed three dimensional structure of the sinusoidal system was evident (Figure 2B and supplementary Figure 1). The specificity of DiI-Ac-LDL labeling of sinusoidal endothelial cells (red) can seen in a section of bone marrow from a GFP transgenic mouse with the enhanced GFP cDNA expressed under the control of a chicken beta-actin promoter and cytomegalovirus enhancer (Figure 2C). Endocytosed DiI-Ac-LDL was compartmentalized within lysosomes in endothelial cells,15 producing a granular intracellular signal throughout the sinusoidal wall. Figure 2D shows a bone marrow section from an animal 4 days after receiving 950 cGy of irradiation. DiI-Ac-LDL lined the dilated sinusoids. Because radiation resulted in leakage from the intravascular space into the extravascular space, cells with scavenger receptors residing in the extra vascular space now endocytosed the DiI-Ac-LDL and fluoresced red (yellow arrows). Arterioles, on the other hand, failed to uptake DiI-Ac-LDL (Figure 2 E, pink arrows). After exposure to lethal doses of irradiation, sinusoidal endothelial cells continued to uptake DiI-Ac-LDL demonstrating that they remained functional (Figure 2D and 2E). Endothelial cells lining the venules that emptied into the central venous sinuses and the central venous sinuses themselves, also endocytosed DiI-Ac-LDL (Figure 2F). Injection of DiI-Ac-LDL reproducibly labeled sinusoids regardless of age, gender or mouse strain. The DiI-Ac-LDL remained in the mouse sinusoidal system for up to 24 hours with some reduction in the intensity but loss of the specificity. We did not analyze the durability of DiI-Ac-LDL signal beyond 24 hours.
Tie2 is a receptor tyrosine kinase that is expressed by vascular endothelial cells,16–20 hematopoietic stem cells,21–25 and vasculogenic monocytes.26 The Tie2 promoter is more active in arteriolar than venular endothelium in the mesenteric and diaphragmatic vasculature.10 In healthy Tie2/GFP mice, the structures expressing GFP most actively were arteries and arterioles. They formed branched structures in the bone marrow to deliver blood from the nutrient arteries to the smaller arterioles and capillaries (Figure 3A). The expression of GFP was particularly strong in arteries (Figure 3B) recognized by the very large lumenal diameter and by multiple layers of smooth muscle cells surrounding them (pink arrows). The presence of smooth muscle could be inferred by the pattern of regularly spaced nuclei of the smooth muscle cells surrounding the arterioles, as can be recognized in hematoxylin and eosin staining and immunostaining for smooth muscle actin (supplementary Figure 2). GFP was also intensely expressed by arterioles (Figure 3C) (red arrows) with smooth muscle cell nuclei forming an evenly spaced pattern along the vessel (pink arrows). Aother type of vessels that were 10–15 microns in diameter, and had no surrounding muscular sheath also expressed Tie2/GFP (orange arrow in Figure 3C and green vessel Figure 3F). The blue arrows in Figure 3C indicate some strong GFP expressing capillaries with 4–6 micron diameters.
The sinusoids had a much larger diameter, and avidly endocytosed DiI-Ac-LDL. Sinusoidal endothelial cells expressed GFP in the perinuclear region but had much less cytoplasmic GFP expression than arterioles or capillaries (yellow arrows in Figure 3D and E). Low cytoplasmic GFP may be due to the thinner walls of the sinusoids compared with arterioles. However, their cytoplasm was intensely labeled by the DiI-Ac-LDL relative to the arterioles, which had negligible DiI-Ac-LDL uptake. GFP signals on sinusoidal endothelial cells from frozen sectioned marrow core were lost during sample processing (yellow arrows in Figure 3B and F).
By using a confocal lambda scanning technique, we demonstrated that the weak green signals from the sinusoidal endothelial cells were true GFP and not due to autofluorescence (Supplementary Figure 3).
Using the combination of DiI-Ac-LDL uptake, Tie2 promoter expression and three dimensional imaging, we observed transitional zones between arterioles and sinusoids. An example of a transitional zone is shown in Figure 4 (also see Supplementary Figure 4). The cytoplasm of endothelium from capillaries that were less than 5 microns in diameter (small white arrows) expressed GFP strongly. These small diameter capillaries emptied into tubes with larger diameters, perinuclear GFP expression and weaker cytoplasmic GFP expression (larger white arrows in the higher magnification panels). These vessels in turn emptied into sinusoids that were characterized by intense DiI-Ac-LDL uptake and perinuclear GFP expression (pink arrows). The dramatic shift from the very narrow capillaries that strongly expressed GFP to the wider diameter sinusoids that weakly expressed GFP and uptake DiI-Ac-LDL avidly occurred through short transitional vessels. These transitional vessels had some features of the vessels they connected: the relatively strong Tie2/GFP expression and lack of uptake Ac-LDL of the straight capillaries, and the perinuclear GFP expression and larger tube diameter of the sinusoids. These features of transitional vessels have not been previously described. The different characteristics of the three different types of endothelium in this transitional zone demonstrate functional differences between the endothelial cells that line these different types of vessels.
In the diaphysis, the arteriole to sinusoid transition points tended to cluster within small areas of the bone marrow in contrast to the sinusoids which were distributed throughout. A low power view of a core with Tie2 and DiI-Ac-LDL staining in Supplementary Figure 5 gives a better understanding of the distribution of the arterioles, the transition zones and the sinusoidal area in the diaphysis. In this area, the major transition zones were not located on the endosteal surface. The majority of vessels that were located close to or on top of the endosteum surface were thin straight capillaries and sinusoids.
In the metaphysic, there were fewer arterioles. Some sinusoids were longitudinally orientated towards the ends of the bone (Figure 5A). There were some transitional vessels in this area (Figure 5B white arrow) but it lacked a large transition zone as seen in the diaphysis. More examples of the vessel distribution in diaphysis and metaphysis in sectioned bone marrow are shown in Supplementary Figure 6.
The methods to differentially identify arterioles, capillaries and sinusoids described above require Tie2/GFP transgenic mice. A method to differentiate vessels in non-transgenic mice would be more generally applicable. In pursuing this goal, we first tried to use Tie2 antibodies to detect expression of Tie2 protein on bone marrow blood vessels, but the staining was weak and the background was high. Since DiI-Ac-LDL specifically labels the sinusoids, a second method was necessary to label other vessel types. We co-injected FITC conjugated lectin with DiL-Ac-LDL. As seen in Figure 6A, arteries and arterioles fluoresced green and did not uptake Ac-LDL. In contrast, the sinusoids were stained by lectin and endocytosed DiI-Ac-LDL, leading to both green and red fluorescence (yellow). In Figure 6B, we can see a thin capillary (green) connected to wider caliber transition vessel (yellow vessel) that resembles the transition vessel in Figure 4. Co-injection of lectin with DiI-Ac-LDL serves as a second method that can be used differentiate vessels in all types of mice. However, because of the smaller size of the lectin molecule, some lectin leaked out of the vessels and stained surrounding cells (Pink arrows in Figure 6 A and B). Therefore, staining with lectin resulted in a ‘dirtier’ image than using Tie-2/GFP mice. Lectin also stained vessels in other organs such as heart and liver at levels rivaling the GFP expression in Tie2/GFP mice. Supplementary Figure 7 shows examples of the vascular staining patterns in other organs labeled by co-injection of FITC conjugated wheat germ lectin with DiI-Ac-LDL.
Recent interest in the role of the bone marrow sinusoids as niches for hematopoietic progenitors and stem cells has led to efforts to understand the cellular and molecular mechanisms by which the sinusoids develop and are maintained. An important prerequisite to this is the ability to distinguish the sinusoids from other types of vascular structures in the bone marrow.
Transgenic mice expressing GFP downstream of the Tie2 promoter provide a useful tool to visualize vessels in the murine bone marrow. However, Tie2 promoter activity is not equivalent in all blood vessel types. Specifically, Tie2 reporter activity is more pronounced in arteries and arterioles.10 Anghelina reported a “dead end” distribution of the Tie2/LacZ vessels in the mouse diaphragm. Similarly, we observed pronounced Tie2 reporter activity in the arteries, arterioles, and capillaries, with faint perinuclear expression of the GFP reporter in the venular sinusoids. The perinuclear location of GFP expression in the sinusoidal endothelial cells has not been previous reported. However, lambda scanning confirmed that this signal was due to Tie2 promoter activity and not auto-fluorescence.
Tie2 expression by hematopoietic stem cells 21, 27 and monocytes26 has been reported in several articles. However, the Tie2/GFP mice used in our experiments showed little GFP expression in these cell types. Employing the Tie2 antibody used by Suda’s group 27 to identify Tie2 positive hematopoietic stem cells and monocytes, we found very few Tie2 positive stem cells and monocytes. The majority of these Tie2 positive cells identified by antibody staining were GFP negative in our Tie2/GFP mice. (See Supplemental Figure 8 for flow cytometric comparison of Tie2/GFP expression vs. Tie2 antibody in our Tie2/GFP mice).
Injection of DiI-Ac-LDL led to the fluorescence of the entire venular side of the marrow vasculature, including the sinusoids, venules and the central sinus, but not the arteriolar side of vessels. The failure of Ac-LDL to label arterial vessels and capillaries in other tissue has also been noticed.6 In culture, endothelial cells isolated from the arterial vessels were able to endocytose Ac-LDL. It was found that the effect of shearing force of blood flow decreased the efficiency of Ac-LDL endocytosis by aortic endothelial cells. Shear flow inhibited scavenger receptor-mediated endocytosis and resulted in decreased uptake of Ac-LDL.28 This lack of efficient uptake of Ac-LDL on the arterial side of vessels provided us with a tool to distinguish between histologically different types of blood vessels.
By combining levels of Tie2 promoter activity and uptake of Ac-LDL, the transition between arteriolar to venular vessels became apparent. The differences in morphology, level of Tie2 promoter activity and efficiency of uptake Ac-LDL allowed these transition vessels to be neatly separated from the arterioles and sinusoids that are connected by these vessels. In the book entitled “Blood Supply of Bone” by Brookes and Revell,29 the authors described “transition points” between the arterioles and sinusoids called the “arterio-venous junctions”. A funnel-shaped junction was found where arterioles and venules met.30 However, in other studies using India ink injected into the femoral artery or vein to visualize the marrow vasculature, the junctions were found to be more gradual widening from the arteriole to sinusoids occurring over two or three microscopic fields. 31 With the functional labeling method reported in this manuscript, the two types of vessels can be clearly differentiated, and the transitional vessels linking arterioles to the sinusoids can be unambiguously identified.
The transition zones contain a high concentration of capillaries. Compared to the evenly distributed sinusoids, the uneven distribution of the capillaries could have an effect on the calculation of vascular density, particularly when the staining methods do not differentiate between these two vessel types. When sectioning through the transitional zone, more vessel cross sections would be present compared to the sections outside the zone. Also, because of the small diameter of the capillaries and their presence in high concentrations in specific areas, presence of these vessels could be falsely identified as newly “sprouted” vessels or “neovessels” as described in some papers.
The significance of these different vascular beds in the bone marrow relative to human disease is currently poorly understood. For instance, could autoimmune vasculitis or other vascular abnormalities within the bone marrow cause bone marrow failure? The effect of chemotherapy and radiation on the marrow sinusoids leads to intramedullary hemorrhage, but to what degree does damage to these structures causes bone marrow failure in heavily pretreated cancer patients? Furthermore, what type of vessels are increased leukemic and myelodysplastic states where there is increased microvascular density? We are developing staining methods that can be used in human samples to differentiate various types of marrow vasculature that might lead to answers to these questions.
Bone marrow sinusoids are highly specialized capillaries and do not always express antigens that are expressed in other capillary beds. Furthermore, the mouse bone marrow sinusoids differ from human sinusoids in that not all antigens expressed in human marrow vasculature, such as CD31, CD34 and vWF, are expressed by blood vessels in the murine bone marrow. This is true in spite of the fact that some of these proteins are expressed abundantly in vessels of other murine organs such as the liver. Examples of staining patterns of other widely used agents in the murine bone marrow are given in Supplementary Figure 9. One important drawback to our methods is that DiI-Ac-LDL and lectin staining require blood flow through a vessel in order for exposure to the agent, and could be misleading in states that cause impaired marrow blood flow.
A summary of the characteristics and the distributions of different vessels illustrated in this manuscript are shown in Figure 7. This work demonstrates that studies of bone marrow vasculature require careful choice of markers depending on the vessels of interest.
Changes in the bone marrow microvasculature have been demonstrated to result from the presence of certain forms of leukemia in the bone marrow. Increases in the number of vascular cross sections within a given section of the bone marrow have been the primary method of demonstrating increases in marrow vascular density. Furthermore, antiangiogenic treatments have been used to treat hematologic malignancies such as multiple myeloma. However, understanding the biology of this extra vascular tissue is imperative to developing better targeted therapy. In this manuscript, a mouse model is used to demonstrate that the arteriolar and venous vasculature are physiologically different. Expression of the receptor Tie-2 (tyrosine kinase with immunoglobulin and EGF homology domains) was predominantly seen in the arteriolar vessels. Furthermore, endocytosis of the functional endothelial marker acetylated low density lipoprotein was more pronounced in the marrow sinusoids and veins but not the arterioles. These results suggest that the bone marrow vasculature is functionally heterogeneous. Recognizing this heterogeneity, studies should be undertaken to determine what types of vessels increase in disease states such as multiple myeloma. Understanding this heterogeneity and how disease effects specific types of blood vessels could allow for the development of better targeted therapy to treat marrow disorders.
Funding Sources: With support from the Bryce Buchanan Young Investigator Award from the Farb Climb for Cancer Foundation, the Biomedical Research Support Program for Medical School’s award to the University of Florida College of Medicine by the Howard Hughes Medical Institute (to W.B.S.), and The University of Florida American Cancer Society Institutional Research Grant IRG-01-188-01.