Bone marrow sinusoids maintain homeostasis between developing hematopoietic cells and the circulation, and they provide niches for hematopoietic progenitors. Sinusoids are damaged by chemotherapy and radiation. Hematopoietic stem cells (HSCs) have been shown to produce endothelial progenitor cells that contribute to the repair of damaged blood vessels. Because HSCs home to the marrow during bone marrow transplant, these cells may play a role in repair of marrow sinusoids. Here, we explore the role of donor HSCs in the repair of damaged sinusoids following hematopoietic stem cell transplant. We used three methods to test this role: (a) expression of platelet endothelial cell adhesion molecule to identify endothelial progenitors and the presence of the Y chromosome to identify male donor cells in female recipients; (b) presence of the Y chromosome to identify male donor cells in female recipients, and expression of the panendothelial marker mouse endothelial cell antigen-32 to identify sinusoidal endothelium; and (c) use of Tie-2/green fluorescent protein mice as donors or recipients and presence of Dil-Ac-LDL to identify sinusoids. We found that sinusoids were predominantly host-derived posttransplant. Donor cells spread along the marrow vasculature early post-transplant in a pattern that matched stromal-derived factor-1 expression. Furthermore, these engrafting progenitors were positioned to provide physical support, as well as growth and survival signals in the form of vascular-endothelial growth factor-A. Occasionally, donor cells provide cellular “patches” in the damaged sinusoids, although this occurred at a low level compared with hematopoietic engraftment. Donor support for the repair of the marrow vascular niche may be a critical first step of hematopoietic engraftment.
Disclosure of potential conflicts of interest is found at the end of this article.
Bone marrow failure is a serious medical problem that has been attributed to quantitative and qualitative deficiencies of hematopoietic stem cells. Relatively little is known about the role of stroma in the development of bone marrow failure. Bone marrow transplantation can be a curative treatment for patients with marrow failure and malignancy. However, transplant can be complicated by slow or failed engraftment. Stem cell homing and engraftment require the movement of stem and progenitor cells into specialized microenvironments. Whereas host stromal cells provide migrational cues and growth signals to engrafting cells, relatively little is known about how donor cells support the survival and repair of host stroma. Reciprocal signaling between donor and host cells may be critical to hematopoiesis and engraftment. Abnormalities in these relationships may lead to marrow failure, delayed or failed engraftment, and poor stem cell mobilization. Modulating these interactions may lead to improved engraftment and treatments for patients with marrow failure.
Bone marrow sinusoids are highly specialized capillaries that collect maturing blood cells and deliver them to the central circulation . These stromal cells have pores spanning the cytoplasm called fenestrations that allow hematopoietic cells to enter into the circulation. The sinusoidal endothelial cells support hematopoiesis in critical ways. For instance, sinusoidal endothelial cells produce factors that enhance hematopoietic differentiation and may also provide niches for hematopoietic stem cells [2, 3]. The sinusoidal vascular niche is particularly important to megakaryocyte development, where endothelial cells provide maturational signals to megakaryocyte progenitors . Sinusoidal endothelium also supports maturation of other hematopoietic lineages . However, sinusoidal endothelium and hematopoietic cells have a codependent relationship. Unlike capillaries in other organs, marrow sinusoids lack a basal lamina. As a result of their delicate structure, the sinusoidal capillaries rely on close association with hematopoietic tissue for structural stability. Because of their central role in megakaryopoiesis and stem and progenitor cell homeostasis, integrity of the sinusoids is critical to hematopoietic recovery.
Bone marrow sinusoids maintain marrow homeostasis. Relative to hematopoietic progenitors, which are exquisitely sensitive to radiation exposure, vascular endothelium has been described as radioresistant . The idea that marrow endothelium is radioresistant is supported by the fact that after lethal irradiation, the marrow empties of hematopoietic cells, whereas surviving sinusoids are clearly evident in the empty marrow. The process of sinusoidal damage and repair was studied during the early 1990s using electron microscopy [6, , –9]. In spite of sinusoidal structures persisting postirradiation, most studies suggest that the marrow sinusoids suffer damage following exposure to chemotherapy or radiation, such as a sloughing of plasma membranes, dilatation of the rough endoplasmic reticulum, and cytoplasmic swelling [6, 10]. Simplistically, disruption of normal homeostatic barriers leads to “holes in the levy,” resulting in the loss of homeostatic integrity of the bone marrow. These holes lead to hemorrhage of nonreticulated erythrocytes into the hematopoietic compartment of the marrow space [9, 11, 12], the escape of marrow elements to the peripheral circulation , and the circulation of hematopoietic stem and progenitor cells that normally reside in the bone marrow . The spleen temporarily becomes a major site of hematopoiesis until homeostasis is restored in the marrow , and progenitors traffic from the marrow to the spleen . How bone marrow sinusoidal integrity and homeostasis are restored following conditioning for a bone marrow transplant is poorly understood. Scanning electron microscopy studies of epoxy microvascular casts suggest that the restoration of the sinusoids is patchy and incomplete following total body irradiation or 5-fluorouracil treatment in mice . Incomplete repair of the vascular niche may explain how exposure to high-intensity chemotherapy and radiation treatment leads to thrombocytopenia , poor stem cell mobilization , and increased susceptibility to myelosuppressive effects of infection . In such a system, small breaches in the sinusoids could lead to a large amount of hemorrhage, but they could be repaired by a small number of appropriately placed cellular patches.
The nomenclature for progenitors that produce vascular cells is complicated. “Endothelial progenitor cells” (EPCs) were described in the late 1990s as bone marrow-derived cells that circulate in response to injury and incorporate into the blood vessels of damaged tissues [20, , , , –25]. Many studies report that EPCs circulate in response to ischemic injury and incorporate into the blood vessels of damaged tissues, including vessels in the liver , muscle [22, 25], heart [25, –27], skin [28, –30], intestine , and eye [31, 32]. One body of work suggests that EPCs are monocyte-like cells that migrate to sites of sprouting new vessel branches to produce tubular structures during the repair of vasculature [33, 34]. These mobile, phagocytic cells use proteases to burrow into damaged tissue to create new vascular tubes. Recent work suggests that EPCs arise from bone marrow-derived myeloid progenitors, downstream of the hematopoietic stem cell, supporting an ontologic link to macrophages [21, 22, 35]. The lack of evidence of marrow-derived cells incorporating into blood vessels during normal development  and during growth of certain tumors  has led to questions regarding the specific circumstances in which EPCs are active, suggesting that this activity may be specific to certain types of ischemic stress and injury. Furthermore, when human bone marrow-derived EPCs are tested in vitro in conditions that promote endothelial proliferation, these putative endothelial progenitors have limited ability to proliferate compared with progenitors derived from the walls of blood vessels. Other circulating endothelial progenitors that have high proliferative potential are called endothelial colony-forming cells [34, 38]. Yoder et al. have established that the endothelial colony-forming cells do not arise from the hematopoietic stem cells, are far more proliferative than marrow-derived progenitors, and are able to produce vascularized channels in immunodeficient mice, unlike marrow-derived EPCs . This has led to the concept that marrow-derived EPCs provide molecular signals that promote proliferation and survival of endothelial cells but do not provide the endothelial cells themselves .
In contrast, other studies have shown that marrow-derived EPCs incorporate into blood vessels. Our group and others have shown that single adult hematopoietic stem cells (HSCs) reconstitute hematopoiesis and repair blood vessels at distant sites [21, 22], demonstrating that HSCs are at least one source of EPCs. Furthermore, the role of donor- and host-derived endothelial progenitor cells in repairing damaged marrow sinusoids following a hematopoietic stem cell transplant is currently unknown. Studies in humans suggest that up to 30% of sinusoidal endothelial cells are donor-derived 3 to 4 months following bone marrow transplant for chronic myelogenous leukemia . We sought to determine the role of HSC-derived donor cells in the repair of marrow sinusoids and restoration of homeostasis following lethal irradiation as conditioning for a hematopoietic stem cell transplant.
Materials and Methods
We purchased 6-week-old C57/BL6 and transgenic C57BL/6-TgN(ACTBEGFP)1Osb (green fluorescent protein [GFP]) donor mice  and Tg(TIE2GFP)287Sato/J (Tie-2/GFP) mice on the sensitivity to Friend leukemia virus B background  (The Jackson Laboratory, Bar Harbor, ME, http://www.jax.org). These studies were approved by the University of Florida Institutional Animal Care and Use Committee.
Hematopoietic Stem Cell Enrichment
Marrow from 6 to 10 male GFP animals was hemolyzed using ammonium chloride. Cells were incubated with a cocktail of rat anti-mouse antibodies to mature hematopoietic lineages (anti-lineage antibodies) as previously described . We washed the cells and incubated them with goat anti-rat magnetic beads (Dynal Biotech, Carlsbad, CA http://www.invitrogen.com/dynal). Lineage-positive cells were removed with a magnet. Lineage negative (Lin)neg cells that express Sca-1 and c-kit (SKL) were incubated with an allophycocyanine conjugate of c-kit, as well as a phycoerythrin conjugate of stem cell antigen-1. SKL cells were isolated using a fluorescence-activated cell sorter. Purity was confirmed as >95% by reanalysis.
Eight- to 10-week-old female recipient animals were lethally irradiated with 950 cGy from a cesium-137 source. Mice were injected within 2 hours of the radiation dose. We injected 2 × 106 nucleated bone marrow cells or 2 × 103 SKL cells from male mice into the retro-orbital sinus.
We identified single-cell-engrafted animals by the presence of circulating GFPpos hematopoietic cells or by the presence of the Y chromosome in female recipients by fluorescence in situ hybridization (FISH) analysis. The animals were phlebotomized and then euthanized, and bones and bone marrow were isolated as described above for histological and fluorescence-activated cell sorting analysis.
The method of tissue fixation depended on how the tissues were going to be analyzed. For immunohistochemistry and FISH analysis, whole tibiae, femora, humeri, and vertebrae were immersed in neutral buffered formalin (NBF) fixative for 20 hours, decalcified in hydrochloric or formic acid, and embedded in paraffin. For mouse endothelial cell antigen (MECA)-32 staining, bones were fixed in 4% paraformaldehyde, decalcified in 5% formic acid, and embedded in paraffin.
We used the following monoclonal antibodies for our studies: primary antibodies included rabbit anti-GFP (Abcam, Cambridge, MA, http://www.abcam.com), rabbit anti-stromal-derived factor-1 (anti-SDF-1) (R&D Systems Inc., Minneapolis, http://www.rndsystems.com), rabbit anti-von Willebrand Factor (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com), goat anti-platelet-endothelial cell adhesion molecule (anti-PECAM) (CD31; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), and rat anti-MECA-32 (BD Pharmingen, San Diego, http://www.bdbiosciences.com). Fluorescent antibody detection used Alexa Fluor secondary antibodies (Molecular Probes Inc., Eugene, OR, http://www.probes.invitrogen.com) with wavelengths of 488 and 594. Slides were mounted with Vectashield containing 4,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). Staining for MECA-32 was performed using antibody at a concentration of 1.6 μg/ml applied overnight at 4°C. DAKO (Glostrup, Denmark, http://www.dako.com) targeted retrieval reagent was used on paraffin-fixed tissue at 95°C for 20 minutes followed by a 20-minute cool down. We used diaminobenzidine to detect our antibody (Vector Laboratories).
Fluorescence In Situ Hybridization
Fluorescence in situ hybridization was performed using STAR FISH whole-chromosome paint probes (Cambio, Dry Drayton, U.K., http://www.cambio.co.uk). Deparaffinized sections were sequentially retrieved and digested before overnight hybridization in a Hybrite oven (Vysis, Downers Grove, IL, http://www.vysis.com). For dual FISH/immunohistochemistry, ABC Elite kits were used following the manufacturer's instructions, with diaminobenzidine used as the chromagen (Vector Laboratories).
Uptake of Acetylated Low-Density Lipoprotein by Donor-Derived Endothelial Cells
Transplant recipients received 20 μg of 1,1′-dioctadecyl 1,3,3,3′,3′tetramethylindocarbocyamine-labeled acetylated low-density lipoprotein (Ac-LDL) (Biomedical Technologies, Stoughton, MA, http://www.btiinc.com). Four hours after retro-orbital injection, bones were placed in NBF for 20 hours and then washed in phosphate-buffered saline (PBS). Marrow was then carefully removed using a dissecting microscope and saturated with 8% sucrose in PBS. Marrow was then embedded in optimal cutting temperature medium, frozen, sectioned, and visualized with a fluorescent microscope.
Alternatively, squash preps were performed to better visualize GFP expression and the three-dimensional structure of the sinusoids. Marrow was fixed in either NBF or 4% paraformaldehyde, the bone was carefully broken, and the marrow was lifted onto a glass slide. We found that this method was the best way to preserve and image GFP expression in the sinusoids, as sectioning led to leaching of GFP out of the tissues. A drop of Vectashield mounting medium containing DAPI (Vector Laboratories) was placed on the section, and a coverslip was placed on top. Sections were visualized using conventional or confocal fluorescent microscopy.
Fluorescent Microscopy and Confocal Analysis
Images were captured on an Olympus BX 51 microscope (Tokyo, http://www.olympus-global.com) equipped with an Optronics Magnafire digital camera system (Optronics, Goleta, CA, http://www.optronics.com). Confocal imaging was performed on a Leica TCS SP2 AOBS Spectral Confocal Microscope (Leica, Heerbrugg, Switzerland, http://www.leica.com). Spectral analysis of single cells was performed at the time of image collection and was compared with the known spectrum of GFP, which has a major emission fluorescent peak at 525 nm.
Measuring Protein Expression
We measured protein expression of SDF-1 and vascular endothelial growth factor-A (VEGF-A) from a single femur flushed with 1 ml of PBS plus a protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). Nonidet P40 (1%) was added to the cell suspension. Cells were lysed by freeze/thaw, and cell debris was centrifuged and discarded. We measured the concentration of SDF-1 and VEGF-A using enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems) and an ELISA plate reader. The average concentration and standard deviation for three animals per time point were calculated.
Multicolored images were merged by using Photoshop (Adobe Systems Inc., San Jose, CA, http://www.adobe.com). Alternatively, merged images from confocal z-stacks were made using LCS Lite imaging software (Leica). Movies of the stacks were produced using ImageJ (NIH).
Conditioning Radiation Damages Marrow Sinusoids
To determine the effect of radiation, a common conditioning regimen for bone marrow transplant, on marrow sinusoidal blood vessels, we irradiated mice with 950 cGy from a cesium source and transplanted 2 × 106 nucleated bone marrow cells. We used H&E-stained sections to determine the effect of radiation on marrow sinusoids. Nonirradiated bone marrow was highly vascularized and contained a variety of blood vessels, including arterioles, capillaries, venules, and veins. The marrow sinusoids formed a network of thin-walled vessels throughout the healthy bone marrow that were regularly distributed (Fig. 1A). The sinusoids could be identified based on thin walls, irregular shape, and the presence of intraluminal erythrocytes (Fig. 1B). Endothelial cells lined the sinusoids and were identified by their elongated, fusiform nuclei; continuity with the sinusoidal endothelial membrane; and location immediately adjacent to the sinusoidal lumen (Fig. 1C) . Seven days following radiation, homeostasis between the extravascular and intravascular space was lost (Fig. 1D). Sinusoidal lumens increased in diameter and adopted unusual shapes (Fig. 1E). In many areas, anucleate erythrocytes were present in the extravascular space, resulting either from hemorrhage or from the process of engraftment (Fig. 1F). One month post-transplant, homeostasis between the intra- and extravascular spaces was restored (Fig. 1G). Numerous round fat droplets appeared in close association with some sinusoids, and sinusoidal capillaries decreased in diameter (Fig. 1H, 1I). This demonstrates that the marrow sinusoids are damaged by ionizing irradiation, leading to increased sinusoidal diameter and hemorrhage.
Stromal-Derived Factor-1 Recruits Progenitors to Sites of Engraftment, Where They Produce Survival Signals
SDF-1 [46, 47] and VEGF-A [30, 48, 49] recruit endothelial progenitors to sites of vascular damage. EPCs express the receptor for SDF-1 (CXCR-4) [46, 47] and vascular endothelial growth factor receptor 1 and 2 [30, 50]. We used ELISA to compare the protein expression levels of these key angiogenic factors after radiation, with or without a bone marrow transplant. Expression of SDF-1 peaked 3 days post-transplant and peaked again on day 7 (Fig. 2A). Protein levels of SDF-1 did not differ between animals that received bone marrow transplants and irradiated controls. This result suggests that SDF-1 was produced primarily by the host. In contrast, protein levels of VEGF-A were very different between mice that received bone marrow transplants and irradiated controls. In transplanted animals, VEGF-A levels were minimal until 7 days post-transplant and increased as engraftment proceeded. In contrast, irradiated controls had a peak in VEGF-A expression 7 days postirradiation (Fig. 2B). This supports the ideas that VEGF-A is produced by both host and donor cells and that the timing of production of VEGF-A by the host is affected by the presence or absence of donor cells. We then sought to determine the sites of expression of SDF-1 and VEGF-A. Using antibody to SDF-1, we performed immunohistochemical staining at different time points post-transplant. In a representative marrow 7 days post-transplant, SDF-1 expression was predominantly along the endosteum and in spaces between the expanded, surviving sinusoids (Fig. 2C). High-power confocal images revealed the presence of SDF-1 in the extracellular spaces along the endosteum and between the marrow sinusoids (Fig. 2D). These same spaces filled with engrafting progenitors that were found to express the endothelial survival and growth factor VEGF-A 14 days post-transplant (Fig. 2F). VEGF-A expression was less pronounced in poorly engrafted areas, where it could be seen in scattered stromal cells and megakaryocytes (Fig. 2G). In contrast to SDF-1, we saw little expression of VEGF-A along the endosteal surface 14 days post-transplant (Fig. 2H). These results demonstrate that SDF-1 is located appropriately to recruit progenitor cells into the spaces between the damaged sinusoids. Engrafting progenitor cells produce VEGF-A in proximity to damaged vasculature, where it can promote survival and proliferation of damaged endothelium.
Donor-Derived Progenitor Cells Spread Along the Endosteal Niche and Surviving Vascular Structures as a First Step of Engraftment
We used sex-mismatched transplants and tracked donor-derived male cells using a fluorescence in situ hybridization probe for the Y chromosome in bone marrow sections. We followed engraftment patterns weekly from day 8 to day 28 post-transplant. Eight days post-transplant, donor-derived cells marked by a single X and Y chromosome were present along the endosteum and surviving stromal structures between adjacent vessels (Fig. 3A–3D). In numerous locations, cells were present in layers that were 1–4 cells thick (Fig. 3B). This pattern suggested that progenitors spread along surviving host vascular structures (Fig. 3B–3D). By 1 month post-transplant, spaces between the blood vessels were filled with hematopoietic cells (Fig. 3F, 3G). Donor cells surrounded fat droplets that formed large, round, empty spaces in the marrow sections because fat was lost in the marrow processing. These results demonstrate that engraftment occurs through the spreading of progenitors along surviving sinusoidal structures. These engrafting progenitor cells are positioned to circulate, provide structural support and survival signals, or integrate into the walls of the damaged sinusoids.
PECAM Was a Poor Marker for Donor Sinusoids
To identify donor-derived cells that had incorporated into the sinusoidal endothelium, we first attempted to use the widely used endothelial marker PECAM to identify sinusoidal endothelium and the Y chromosome to identify donor cells. In contrast to the portal vein, where a subset of cells that were PECAM-positive and CD45-negative could be identified , all of cells that expressed PECAM also expressed at least low levels of CD45 in marrow disaggregated with collagenase and dispase (Fig. 4A). Immunohistochemical staining of the bone marrow showed donor-derived PECAM-positive cells with the shape of hematopoietic cells 9 days post-transplant (Fig. 4B). At day 28 post-transplant, PECAM-positive donor-derived cells were present in thin layers along stromal structures (Fig. 4C). However, sinusoidal endothelium in healthy, nonirradiated bone marrow did not express PECAM (Fig. 4D). These results suggest that PECAM is expressed ubiquitously in the bone marrow and is not a good marker for identifying sinusoidal endothelium. Therefore, we looked for a more definitive approach to assess the role of the donor in sinusoidal repair.
Donor-Derived Cells Expressing MECA-32 Incorporated into the Sinusoids Following Transplant of HSCs
To further define the role of the donor in post-transplant sinusoidal repair, we used the panendothelial marker MECA-32  and presence of the Y chromosome to identify donor cells to further test the role of the donor in the repair of the vascular niche. The sinusoids were easily visualized by light microscopy using this method (Fig. 5A, 5B). FISH analysis of 189 nuclei from three animals suggested that donor-derived cells expressing MECA-32 were abundant, making up nearly 40% of the endothelial cell nuclei that had FISH signal (Fig. 5C, 5D). We observed abundant donor-derived cells along the abluminal surfaces of MECA-32-expressing endothelial structures, such as the central sinus (Fig. 5E). To generate clearer images of the engraftment patterns of donor-derived cells in relation to cells that express MECA-32, we analyzed the marrow of three engrafted animals by confocal microscopy. In the more thinly engrafted areas of the marrow, MECA-32-expressing sinusoids (Fig. 5F, white) were lined by thin layers of donor cells in close association with adipocytes (Fig. 5F, **) 1 month post-transplant. However, fewer than 1 in 10 cells that appeared to be donor-derived by conventional fluorescence were bona fide donor-derived cells when analyzed by confocal microscopy (Fig. 5I–5N). This discrepancy was due to the fact that by standard fluorescence microscopy, overlying cells could not be distinguished. Thus, although donor cells occasionally incorporated into the walls of sinusoids, this occurred at a low level, roughly in the range of 1 in 100 endothelial cells. More than half of the sinusoidal endothelial cells had no FISH signal, making the donor's role in sinusoidal repair difficult to quantify using this method. We therefore sought additional methods to asses the role of host and donor in post-transplant sinusoidal repair.
Marrow Sinusoids Are Predominantly Host-Derived Following Transplant
The angiopoietin receptor tyrosine kinase, Tie-2, is upregulated in the sinusoidal endothelium of the bone marrow following myelosuppressive doses of the chemotherapeutic agent 5-fluorouracil and sublethal irradiation . We reasoned that Tie-2 expression would identify cells that participated in the repair of marrow vascular sinusoids following hematopoietic stem cell transplant, and GFP expression would provide a marker that identified both donor and endothelium. We transplanted wild-type (WT) animals with whole bone marrow or Sca-1pos c-kitpos Linneg hematopoietic stem and progenitor cells from mice expressing GFP downstream from the Tie-2 promoter (Tie-2/GFP), and in parallel, we transplanted female Tie-2/GFP animals with marrow from male WT donors. Because Tie-2 expression is not specific to endothelium, as it is also expressed by hematopoietic stem cells and subsets of monocytes, we used Ac-LDL to identify marrow sinusoids and used the expression of GFP to identify either donor or recipient cells upregulating Tie-2. We compared the expression of GFP in the sinusoidal cells of these animals with expression in Tie-2/GFP hosts transplanted with wild-type BM cells. In sectioned marrow, we saw little evidence of Tie-2 expression except in straight vessels that appeared to be arterioles. Preservation of signal was improved by imaging whole tissue by confocal microscopy, avoiding the leaching of GFP that occurs with sectioning. Using these methods, we observed GFP expression in some sinusoids in untreated Tie-2/GFP animals (Fig. 6A). Although there were some autofluorescent cells in wild-type control animals (Fig. 6B), this signal could be distinguished from GFP by spectral analysis. Eighteen days to 1 month following transplant with WT whole bone marrow cells, Tie-2/GFP hosts upregulated GFP expression in their sinusoids. Most sinusoidal endothelial cells expressed GFP, both in perinuclear and cytoplasmic areas, as demonstrated by confocal imaging (Fig. 6C; supplemental online Fig. 1A). In contrast, WT animals receiving Tie-2/GFP whole bone marrow or SKL cells had no discernable GFP-positive cells contributing to the sinusoidal vasculature 2 weeks to 1 month post-transplant (Fig. 6D). Autofluorescent cells were abundant in these samples. The spectra of 10 randomly chosen green cells were analyzed and did not match GFP but were similar to the spectrum of autofluorescent cells from WT controls. These results showed that when vascular homeostasis was restored 2 weeks to 1 month post-transplant, sinusoids were composed mainly of host endothelial cells. However, because sinusoidal expression of Tie-2 is directly linked to damage, the lack of GFP signal from WT animals receiving Tie-2/GFP bone marrow cells could not be interpreted as meaning that there was no incorporation into the sinusoids, as the donor cells had not been exposed to radiation.
To test this possibility, we irradiated with 400 cGy C57BL/6 mice that had received 2 × 106 Tie2/GFP bone marrow nucleated cells 3 months prior and were stably engrafted. Figure 6E shows expression of GFP in most sinusoidal cells that had endocytosed Ac-LDL in a Tie-2/GFP mouse 7 days after irradiation. In contrast, none of the cells that endocytosed Ac-LDL expressed GFP in a stably engrafted Tie-2/GFP transplanted mouse 1 to 2 weeks post-transplant (Fig. 6F). We examined more than 100 cells looking for GFP expression in sinusoidal endothelial cells that had endocytosed Ac-LDL 2 weeks post-transplant, and we found three cells that fluoresced green. Spectral analysis of the green signal from these three cells was consistent with autofluorescence rather than GFP. This indicates that the sinusoidal endothelial cells are mainly host-derived following transplant when 950 cGy is used as conditioning and that the lack of GFP expression in animals receiving Tie-2/GFP bone marrow is due to the low level of contribution of these cells to the sinusoids.
In this study, we demonstrate that conditioning radiation damages the marrow vascular niche. This leads to loss of marrow homeostasis, with hemorrhage of nonreticulated erythrocytes into the extravascular space. Donor-derived progenitor cells are recruited through expression of SDF-1 and spread along the sinusoids, where they produce the angiogenic cytokine VEGF-A. From this position, engrafting cells provide structural support and angiogenic cytokines. Occasionally, donor-derived cells plug holes in the damaged vasculature. We propose that the physical support, production of cytokines, and plugging activity of donor cells promote the repair of the sinusoidal vascular niche and restoration of marrow homeostasis (Fig. 7).
Our study illustrates why stem and progenitor cells circulate during the recovery period from myelosuppressive chemotherapy or radiation . This circulation is possible because stem and progenitor cells spread over the surface of the leaky sinusoidal vessels. Because of this location, engrafting stem and progenitor cells have direct access to the circulation. The use of corticosteroids to prevent graft-versus-host disease could potentially block this inflammatory response to SDF-1 signals and decrease the spread of progenitors along the sinusoids, hindering vascular repair and engraftment.
In studying post-transplant vascular repair, it is difficult to ignore the appearance of numerous lipid droplets that are closely associated with the sinusoids. The appearance of adipocytes in conjunction with angiogenic macrophages has been observed in the development of blood vessels that penetrate subcutaneously implanted Matrigel plugs (BD Biosciences, San Diego, http://www.bdbiosciences.com) . The reciprocal relationship between vascular repair and adipogenesis is well described but poorly understood . We speculate that as part of this process of vascular repair, adipocytes fill the empty marrow space. In addition to providing a matrix upon which damaged vessels can be repaired, growing fat globules physically force engrafting cells toward the vasculature. The presence of these fat globules prevents engrafting progenitor cells from losing contact with the vasculature. In proximity to the vasculature, engrafting cells provide growth and survival signals to the damaged vasculature, receive nutrients from the circulation, and enter the circulation when mature. These fat globules form a matrix upon which the damaged sinusoidal vessels can be repaired, providing structural support similar to the foam popcorn used to package and ship fragile cargo.
In the bone marrow, donor cells do not contribute to post-transplant sinusoidal repair by generating new vessels through mechanisms of vasculogenesis or angiogenesis. Rather, the activity of donor cells plugging sites of vascular injury is reminiscent of systems of hemostasis in lower organisms, where nucleated macrophage-like “amoebocytes” or “thrombocytes” fight infection and seal wounds . We speculate that in a state of injury and in the setting of thrombocytopenia, engrafting monocytes fill gaps in the damaged marrow vasculature, similar to the activity of amoebocytes in invertebrate arthropods [54, 55]. Macrophages represent the primitive “natural” immune system and are also the first immune cells to appear following transplant. Macrophages precede the appearance of the more evolutionarily advanced T and B lymphocytes, which comprise the adaptive immune system. It is plausible that an evolutionarily primitive program (thrombocyte activity) precedes the evolutionarily advanced program (platelet activity) during recovery from myeloablation.
We propose the following model regarding the role of the donor in the repair of the marrow vascular niche. Following transplant, self-renewing hematopoietic stem cells migrate to the endosteal niche, where they receive survival signals from osteoblasts [56, , , –60]. As engraftment proceeds, cells migrate along the endosteal surface and toward the center of the marrow space along surviving stromal structures present in narrow channels between massively dilated sinusoids. Unlike other capillary beds that have specialized cells, such as pericytes, that support the endothelium, marrow sinusoidal capillaries rely on hematopoietic tissue in addition to reticular cells to provide physical and growth factor support. In the resting bone marrow, SDF-1 is expressed primarily by reticular cells that reside near the endosteal surface and sinusoidal vasculature . Following lethal irradiation, SDF-1 increases in the marrow space from a combination of increased expression from damaged marrow cells  and increased transport of SDF-1 across damaged sinusoids . Progenitor cells migrate toward SDF-1 gradients present within the spaces between dilated sinusoids. The engrafting progenitors form a sheath around the sinusoids, providing physical support and producing growth factors that support the survival and proliferation of sinusoidal cells (Fig. 6B). Adipocytes aid in the process by providing physical support to the damaged vasculature and filling space to prevent engrafting cells from losing contact with the vasculature. Our study demonstrates that engrafting cells that surround the damaged sinusoids produce VEGF-A. Production of VEGF-A by engrafting progenitors may provide a mechanism whereby vascular remodeling occurs specifically in areas where engraftment is occurring. This would provide a mechanism to use nutrients and oxygen efficiently during the repair process (Fig. 6C). In addition to providing growth factor support, donor cells plug gaps in the damaged endothelium and subsequently incorporate into the vascular wall (Fig. 6C). What is not clear is whether these cells become bona fide endothelial cells (Fig. 6D) or are eventually replaced by endothelium derived from other sources (Fig. 6E). Regardless, by plugging “holes in the dike,” these donor cells may play a critical role in the restoration of homeostasis and successful engraftment following hematopoietic stem cell transplantation.
In summary, our study demonstrates that irradiation leads to intramedullary hemorrhage. During the earliest stages of hematopoietic engraftment, the sinusoidal vasculature is surrounded by donor-derived progenitor cells responding to gradients of SDF-1, and these cells in turn produce vasculogenic cytokines, such as VEGF-A, that promote the survival and proliferation of surviving host endothelial cells. The engrafting cells are positioned to provide structural support and cell-to-cell signaling and to form cellular plugs to damaged sinusoids. This study suggests that repair of the marrow vascular niche is a primary task of engrafting hematopoietic cells during a bone marrow transplant.
Disclosure of Potential Conflicts of Interest
W. Slayton and E. Scott own stock in RegenMed.
Special thanks are due to Dr. Kate Harfe, Chris Cogle, and Robert Fisher for reviews and editorial advice. Douglas Smith provided valuable technical assistance. This work was performed with support from National Institutes of Health grants HL03962 (to W.B.S.), CA72769 and HL70738 (to E.W.S.), 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 American Cancer Society Institutional Research Grant IRG-01-188-01 (to W.B.S.). W.B.S. and X.-M.L. contributed equally to this work.