Main

Improved neovascularization is an important therapeutic goal after myocardial infarction or limb ischemia1. In recent years, our understanding of the processes responsible for the formation of new blood vessels after tissue ischemia has changed. Vascularization of ischemic tissue in adults was once thought to be restricted to migration and proliferation of mature endothelial cells, a process termed 'angiogenesis'2. However, increasing evidence suggests that circulating EPCs home to sites of ischemia and contribute to the formation of new blood vessels3,4,5. In analogy to the embryonic development of blood vessels from primitive endothelial progenitors (angioblasts), this process is referred to as 'vasculogenesis'6,7. The importance of circulating progenitor cells is demonstrated by the fact that genetic inhibition of their recruitment inhibits tumor angiogenesis8, whereas intravenous infusion of bone marrow–derived progenitor cells augments neovascularization9. EPCs can be mobilized from the bone marrow into the circulation by VEGF or stromal cell–derived factor-1 (SDF-1)10,11,12. Maintenance and mobilization of stem cells in the bone marrow is determined by the local microenvironment, which consists of fibroblasts and endothelial cells13,14,15. VEGF- and SDF-1-induced progenitor cell mobilization is dependent on local secretion of MMP-9 by the hematopoietic and stromal compartments of the bone marrow, which results in release of soluble Kit ligand (KitL; also known as stem cell factor)11. This permits transfer of endothelial and hematopoietic progenitor cells from the quiescent to the proliferative niche11.

Nos3−/− mice, which are deficient in eNOS, are characterized by impaired ischemia-induced neovascularization16. Mobilization of EPCs from the bone marrow into the circulation is required for postnatal neovascularization4. Here we show that mice lacking eNOS have defective hematopoietic recovery and progenitor cell mobilization, resulting in increased mortality after myelosuppression and reduced VEGF-induced mobilization of EPCs. Intravenous infusion, but not bone marrow transplantation, of wild-type stem and progenitor cells rescued the impaired neovascularization of Nos3−/− mice in a model of hind-limb ischemia.

Results

Mobilization of progenitor cells in Nos3−/− mice

Basal numbers of EPCs did not significantly differ between Nos3−/− and wild-type mice (Fig. 1a). However, whereas the number of peripheral EPCs significantly (P < 0.05) increased after VEGF mobilization in wild-type mice, Nos3−/− mice did not mobilize EPCs in response to VEGF (Fig. 1a). These data were confirmed by directly measuring the number of CD34+Flk-1+ cells by fluorescence-activated cell sorting (FACS) analysis (data not shown). To assess whether the reduced mobilization capacity in Nos3−/− mice consequently affects hematopoietic reconstitution after myelosuppression, Nos3−/− mice were treated with 5-fluorouracil (5-FU), which allows for the study of factors that regulate recruitment of stem cells during hematopoietic recovery11. Nos3−/− mice had significantly (P < 0.02) increased mortality compared with wild-type mice (Fig. 1b). The recovery of the white blood cell count was significantly delayed in Nos3−/− mice (P = 0.01; Fig. 1c and Supplementary Fig. 1 online). In line with these findings, we observed that the increase in cellularity and hematopoietic cell clusters detected in histological bone marrow sections of wild-type mice, 6–10 d after 5-FU treatment, was incomplete in Nos3−/− mice (Fig. 1d). These data indicate that eNOS is essential for mobilization and hematopoietic reconstitution.

Figure 1: In vivo recruitment of hematopoietic stem cells and EPCs into the circulation.
figure 1

(a) Peripheral EPC levels after mobilization with human VEGF (n = 5). WT, wild type. (b) Kaplan-Meier survival curves for Nos3−/− and wild-type mice after treatment with 5-FU (n = 10). (c) Recovery of the white blood cell (WBC) count (n = 5) after 5-FU treatment. Data are given as mean ± s.e.m. (d) H&E staining of femur bone marrow sections 10 d after 5-FU treatment (magnification, ×20; n = 5).

Full size image

Effect of progenitor cells on neovascularization

We used a hind-limb ischemia model to assess whether impaired mobilization of EPCs, which may contribute to the reduced neovascularization in Nos3−/− mice, can be counteracted by infusion of wild-type stem and progenitor cells. As previously reported16, Nos3−/− mice showed significantly reduced limb perfusion 2 weeks after induction of ischemia, compared with wild-type mice (P < 0.01; Fig. 2a,b). The impaired limb perfusion was rescued by intravenous infusion of wild-type EPCs or isolated wild-type bone marrow cells (BMCs; sca-1+Lin−; Fig. 2a,b). Histological analysis of ischemic thigh muscles indicated that infusion of wild-type EPCs or BMCs substantially increased capillary density (Fig. 2c,d) and number of conductant vessels (Fig. 2e) in Nos3−/− mice. Physical incorporation of intravenously infused wild-type cells into endothelial structures was confirmed by immunostaining for CD31 or mouse pan–endothelial cell antigen-32 (MECA-32) (Fig. 3).

Figure 2: Rescue of impaired angiogenic phenotype in eNOS-deficient mice by intravenous (i.v.) injection of wild-type cells.
figure 2

(a) Laser Doppler–derived relative blood flow in Nos3−/− mice 14 d after induction of hind-limb ischemia and injection of wild-type or Nos3−/− cells (Sca-1+Lin− BMCs or ex vivo–cultured EPCs) (n = 6). Wild-type animals were used as a positive control. (b) Representative laser Doppler scans, with low or no perfusion displayed in dark blue and highest perfusion displayed in red. Arrows indicate ischemic leg. (c,d) Capillary density in wild-type and Nos3−/− mice relative to wild-type control animals (n = 6). Representative overview images are shown in d. (e) Left, numbers of conductant vessels in 4-mm2 muscle sections from wild-type and Nos3−/− mice and wild-type control animals. Right, conductant vessels were identified by size (>20 μm) and α-actin immunostaining. a,c,e, *, P < 0.01 compared with wild-type; **, P < 0.01 compared with untreated Nos3−/− mice.

Full size image
Figure 3: Incorporation of infused CellTracker-labeled BMCs into the vascular structures of limb muscles of Nos3−/− mice on day 7 after induction of ischemia.
figure 3

(a) Infused Nos3−/− BMCs showed a significantly lower incorporation rate compared with wild-type (WT) BMCs (n = 4). (b) Representative overview pictures of muscle sections. Arrows indicate incorporated BMCs (yellow) that are positive for CellTracker CM-Dil (red) and the endothelial marker CD31 (green). (c) High-power magnification of a vessel with incorporated wild-type BMCs that are positive for CellTracker CM-Dil (red) and the endothelial marker MECA-32 (green).

Full size image

To investigate whether impaired stem cell–stromal cell interaction contributes to defective neovascularization in Nos3−/− mice, bone marrow cells depleted of stromal cells by CD45 magnetic bead selection were transplanted into lethally irradiated Nos3−/− mice. After completely reconstituting the bone marrow of Nos3−/− mice with wild-type stem cells, the mice were treated with VEGF for three consecutive days. VEGF-induced mobilization of white blood cells or EPCs was blunted in the Nos3−/− mice transplanted with wild-type bone marrow (Fig. 4a,b). In contrast, VEGF significantly (P = 0.006) increased peripheral white blood cell counts in wild-type mice transplanted with Nos3−/− bone marrow (Fig. 4a). Similarly, VEGF significantly increased c-Kit+Sca-1+ peripheral blood cells in wild-type mice transplanted with wild-type bone marrow (211 ± 43% increase; P = 0.03 compared with unstimulated mice; n = 6 per group), and in wild-type mice engrafted with Nos3−/− bone marrow (398 ± 91% increase; P = 0.01 versus unstimulated mice). In contrast, no significant increase was detected when VEGF was injected into Nos3−/− mice transplanted with wild-type bone marrow (141 ± 26%; P = 0.186).

Figure 4: Bone marrow transplantation of wild-type cells does not rescue the impaired angiogenic phenotype of Nos3−/− mice.
figure 4

Wild-type (WT) and Nos3−/− mice were transplanted with wild-type or Nos3−/− CD45+ BMCs and treated as indicated. (a) VEGF-induced mobilization of white blood cells (WBCs; day 4; n = 6). (b) VEGF-induced mobilization of EPCs (day 4; n = 5). (c) Laser Doppler analysis of relative blood flow in ischemic limbs 2 weeks after induction of ischemia (n = 6). (d) Fluorescence in situ hybridization analysis of incorporation of bone marrow–derived cells into the vasculature of the limb muscles in female mice transplanted with male bone marrow. (e) Representative pictures of wild-type bone marrow–derived Y chromosomes in female wild-type (left) and Nos3−/− (right) recipients. Arrows indicate Y chromosomes (green) in incorporated cells. Endothelial cells were identified by immunostaining for CD31 (red).

Full size image

We used a model of hind-limb ischemia to investigate the functional relevance of this impaired mobilization in Nos3−/− mice. In contrast to the improvement of neovascularization induced by intravenous infusion of wild-type progenitor cells (Fig. 2), bone marrow reconstitution with wild-type cells did not rescue the impaired neovascularization phenotype of Nos3−/− mice (Fig. 4c). Taken together, these data suggest that wild-type stem and progenitor cells are only effective when applied to the circulation, but cannot be mobilized from the bone marrow of Nos3−/− mice. The data also provide evidence that eNOS expression in the bone marrow is essential for mobilization of stem and progenitor cells.

Bypassing the mobilization of BMCs by intravenously infusing Nos3−/− cells instead of wild-type cells is insufficient to compensate for the impaired recovery of Nos3−/− mice after ischemia (Fig. 2a–d). Consistently, a significant (P < 0.05) impairment of neovascularization was observed when Nos3−/− BMCs were transplanted into wild-type mice (Fig. 4c), despite the effective mobilization of white blood cells by VEGF (Fig. 4a). These data suggest that eNOS is not only essential for mobilization of stem and progenitor cells from the bone marrow, but also determines their angiogenic capacity in ischemic tissue. The data also underscore the relative importance of bone marrow–derived cells for the angiogenic response to ischemia.

To obtain further insight into the impaired proangiogenic activity of Nos3−/− cells after intravenous application, we injected labeled wild-type or Nos3−/− cells into Nos3−/− mice and counted the cells that incorporated into the capillaries of the ischemic limb after 7 days. A large number of systemically injected wild-type cells incorporated into the microvessels of the ischemic limb and expressed endothelial marker proteins such as CD31 or MECA-32, whereas incorporation of Nos3−/− cells was significantly reduced (P = 0.027; Fig. 3). In addition, fewer Y-chromosome-positive cells were detected in capillaries when wild-type male BMCs were transplanted into female Nos3−/− recipients (Fig. 4d,e). Because of the limited efficiency of Y-chromosome detection in histological sections (40%; ref. 17), the estimated percentage of capillaries incorporating bone marrow–derived cells after ischemia in our study was ∼5% in wild-type mice previously transplanted with wild-type bone marrow. This figure was confirmed using wild-type mice transplanted with green fluorescent protein–expressing bone marrow. We found that 7 ± 11% of the vessels incorporated bone marrow–derived cells expressing endothelial marker proteins (Supplementary Figs. 2–5 online). Taken together, these data suggest that wild-type cells are efficiently incorporated into vessels and express endothelial marker proteins, whereas eNOS-deficient cells are impeded by a marked reduction in homing and incorporation into microvessels.

Stromal and progenitor cell functions in vitro

Because our data indicate an impaired interaction between stem cells and stromal cells in the bone marrow of Nos3−/− mice, we analyzed the expression pattern of eNOS in different cellular components of bone marrow from wild-type mice. Whereas stromal cells strongly express eNOS, selected CD45+ bone marrow leukocytes showed only very faint eNOS expression (Fig. 5a). Additional subfractionation of cultivated stromal cells revealed that eNOS is predominantly expressed in MECA-32+ vascular cells (Fig. 5a and Supplementary Fig. 6 online). The effect of eNOS on stem and progenitor cell function was further investigated by determining the proliferative and colony-forming activity of bone marrow stem cells. In the bone marrow of Nos3 deficient mice, the numbers of proliferating bromodeoxyuridine (BrdU)-positive Lin−Sca-1+ cells (Fig. 5b,c) and colony-forming units (Fig. 5d) were significantly reduced (P < 0.05). Because the bone marrow microenvironment created by stromal cells may influence the growth of stem and progenitor cells in the bone marrow, we isolated stromal cells from Nos3−/− or wild-type mice to generate feeder cells for functional long-term in vitro stem cell assays. We then cultured total bone marrow from wild-type mice on Nos3−/− or wild-type feeder cells to grow stem cells with repopulating potential and colony-forming capacity. When bone marrow mononuclear cells were grown on Nos3−/− feeders, the number of colony-forming units was significantly reduced (P < 0.01; Fig. 5e), indicating that stromal cell–derived NO is important for the culture of long-term stem cells. Although eNOS was only weakly expressed in CD45+ bone marrow leukocytes, we observed a significantly (P < 0.01) reduced number of colony-forming units when Nos3−/− bone marrow cells were grown on wild-type stromal feeders (Fig. 5a,e). This impairment of Nos3-deficient bone marrow cells was not related to increased apoptosis in vitro (data not shown). These results suggest that stem and progenitor cells derived from Nos3−/− mice are already partially impaired when grown ex vivo on wild-type feeder layers. These data are in line with our in vivo observation that intravenous infusion of Nos3−/− BMCs into Nos3−/− mice did not rescue the impaired neovascularization.

Figure 5: In vitro characterization of hematopoietic stem cells from Nos3−/− and wild-type (WT) bone marrow.
figure 5

(a) Western blot analysis of eNOS in isolated stromal cells, total bone marrow (BM), stromal cell–free (CD45+) bone marrow, and MECA-32+ vascular and MECA-32− stromal cells. A representative blot is shown (n = 4). (b) FACS analysis of Lin− cells (circled) of the lymphocyte-monocyte bone marrow fraction after pulsing with BrdU and staining with BrdU-FITC, Sca-1-PE (phycoerythrin) and a biotinylated lineage cocktail, followed by streptavidin-allophycocyanin (APC). (c) Statistical analysis of proliferation (BrdU+) of Sca-1+Lin− hematopoietic stem cells (n = 6). (d) Colony-forming capacity of Nos3−/− and wild-type total bone marrow cells, as shown by scoring CFUs (n = 9). (e) Cultivation of long-term stem cells in an ex vivo microenvironment produced by stromal cell feeders (n = 6).

Full size image

Pro-MMP-9 levels and activity in Nos3−/− mice

Finally, we examined the potential mechanisms underlying the impaired functional activity and mobilization of stem and progenitor cells in Nos3−/− mice. The expression of classical stem cell regulatory factors such as the β1 integrin CD29 (ref. 18), adhesion molecules such as CD44 (ref. 19) and the chemokine receptor CXCR4 (ref. 20) were not altered in bone marrow cells of Nos3−/− mice (Fig. 6a). MMP-9, which is essential for VEGF- and 5-FU-induced stem cell mobilization11,21, is activated by NO22. Indeed, pro-MMP-9 ] was significantly (P < 0.01) reduced in bone marrow plasma from Nos3−/− mice (Fig. 6b). In addition, the bone marrow of Nos3−/− mice showed not only profoundly reduced basal activity of pro-MMP-9 (45.5 ± 6.7% compared with wild-type; P = 0.003; n = 4), but also a blunted increase in pro-MMP-9 activity in response to 5-FU treatment (Fig. 6c). Consistent with this observation, incubation of the NO donor S-nitrosopenicillamine increased expression of pro-MMP-9 in cultivated bone marrow stromal cells (Supplementary Fig. 7 online).

Figure 6: Molecules involved in stem and progenitor cell mobilization.
figure 6

(a) Percentage of BMCs in the lymphocyte-monocyte fraction that stained for the integrin CD29, the adhesion molecule CD44 and the chemokine receptor CXCR4, as measured by FACS analysis (n = 3–5). (b) Expression of pro-MMP-9 in bone marrow plasma of Nos3−/− and wild-type (WT) mice, as determined by ELISA (n = 9). (c) Gelatin zymography of bone marrow plasma before and 3 d after a single 5-FU injection. Densitometric quantification of representative zymography is shown. (d) Soluble KitL in bone marrow plasma was measured on day 6 after 5-FU treatment (n = 6–8 per group).

Full size image

Stimulation of MMP-9 by 5-FU leads to cleavage of the membrane-bound Kit ligand (KitL) and subsequent release of soluble KitL11. We therefore measured soluble KitL in bone marrow plasma, and found that Nos3−/− mice showed a profound impairment of 5-FU-induced release of soluble KitL (Fig. 6d). To further delineate the causal contribution of impaired soluble KitL generation to the delayed hematopoietic recovery of Nos3−/− mice after 5-FU treatment, we supplemented Nos3−/− mice with soluble KitL. For this purpose, Nos3−/− mice received twice-daily subcutaneous injections of recombinant mouse soluble KitL, from days 3 to 11 after initial treatment with a single intravenous injection of 5-FU according to a previously published protocol11. Soluble KitL treatment rescued the recovery of white blood cell counts and augmented the number of EPCs in Nos3−/− mice after 5-FU treatment. The delayed recovery of white blood cells in Nos3−/− mice after 14 d of 5-FU treatment (42% inhibition; Fig. 1c) was thereby abolished (white blood cell count was 15,144 ± 5,313 cells/μl in Nos3−/− mice treated with 5-FU and soluble KitL; n = 6; P < 0.05 compared with Nos3−/− mice treated with 5-FU alone). Notably, white blood cell counts after combined treatment with soluble KitL and 5-FU were higher than the baseline levels. A similar response was previously described in MMP-9-deficient mice11.

Discussion

Data from the present study show that eNOS is essential for the functional activity of hematopoietic stem and progenitor cells. Defective mobilization of stem and progenitor cells contributes to the impairment of ischemia-induced neovascularization in Nos3−/− mice. These findings are in line with previous studies showing that blockade of endothelial and hematopoietic precursor cell recruitment is sufficient to reduce angiogenesis8. The number of bone marrow–derived cells that incorporate into capillary-like structures and express endothelial marker proteins varies in the literature3,8,23. In the present study, we found that ∼5–7% of capillaries show incorporation of bone marrow–derived cells. Given the large change in capillary density, potential paracrine effects of the infiltrating bone marrow–derived cells are likely to contribute to improved neovascularization, in addition to the cells being incorporated into the endothelial linings. The concept that bone marrow–derived cells can promote neovascularization without being physically incorporated into the endothelial lining has been proposed by a recent report that activation of a suicide gene in bone marrow–derived cells can reduce tumor angiogenesis, although none of the infiltrating cells showed coexpression of endothelial markers in that experimental model23.

Deficiency in eNOS also reduced hematopoietic recovery and subsequently increased mortality after myelosuppression. A similar phenotype in response to 5-FU and VEGF was recently reported for MMP-9-deficient animals, which show reduced transfer of endothelial and hematopoietic stem cells from quiescent to proliferative niches in the bone marrow11. Mice deficient in Nos3 show not only reduced basal expression and activity of MMP-9, but also a significantly blunted increase in MMP-9 activity in response to 5-FU. Thus, a reduction in MMP-9 may indeed underlie the mobilization defect in Nos3−/− mice described in the present study. This hypothesis is supported by recent data identifying MMP-9 as a major target for NO, which activates MMP-9 by S-nitrosylation22. In addition to regulating MMPs, NO also modulates proliferation, differentiation and apoptosis in various cell types24,25,26. Although our study excludes an effect on stem and progenitor cell apoptosis in vitro, we cannot rule out the possibility that endothelium-derived NO interferes with cell cycle progression and/or differentiation of stem cells in the bone marrow. Previous ex vivo studies have shown that low concentrations of NO donors, equivalent to those produced by eNOS in vivo, can specifically increase the colony-forming activity of white blood cell precursors from CD34+ cells27.

Our data also indicate that NO in the bone marrow acts primarily in a paracrine manner, as it is generated mainly by vascular cells of the bone marrow stroma. The stromal cell population also contains endothelial cells, and ∼30% of the stromal cells express VEGF receptor-2 (Flk-1). Hypoxia-induced upregulation of VEGF activates eNOS predominantly through Flk-1-mediated, Akt-dependent eNOS phosphorylation28,29. It is thus possible that VEGF-stimulated NO synthesis in bone marrow stromal cells regulates MMP-9, which in turn induces mobilization of hematopoietic and endothelial progenitor cells30. However, increased expression of inducible NOS may also partially compensate for the lack of eNOS, and may thus be responsible for the moderate phenotype observed in Nos3−/− mice. Further studies are required to elucidate whether double knockout of the genes encoding eNOS and inducible NOS might create an even more pronounced phenotype.

In addition to the essential role of eNOS in mobilization of stem and progenitor cells from the bone marrow, eNOS is also required for angiogenesis in ischemic tissue. This is indicated by our finding that intravenous infusion of Nos3−/− cells or bone marrow engraftment of Nos3−/− cells into wild-type animals does not rescue the impaired neovascularization. These data are in accordance with previous studies showing that eNOS is essential for the survival, migration and angiogenic response of mature endothelial cells24,25,26. The definitive mechanism underlying the impaired capacity of Nos3−/− to rescue neovascularization is not clear. In our study, Nos3−/− cells showed significantly reduced incorporation into ischemic tissue. Nos3−/− cells might be more sensitive to apoptosis in vivo, leading to a reduction in cell numbers, or might have reduced migratory capacity, resulting in impaired transmigration into the ischemic tissue. However, we did not detect an increase in apoptosis when the Nos3−/− bone marrow cells used in the infusion studies were kept ex vivo for 24 h, ruling out the possibility that eNOS deficiency directly affects progenitor cell survival.

The results of the present study may have important implications for cardiovascular disease progression. It is well established that endothelial NO bioavailability, which determines the development of coronary artery disease31, is not only impaired in atherosclerotic coronary arteries, but is also reduced systemically in patients at risk for coronary artery disease32. Consequently, endothelial NO synthesis may also be reduced in the bone marrow of heart patients, resulting in reduced mobilization of progenitor cells. Indeed, patients at risk for developing coronary artery disease have significantly reduced numbers of circulating EPCs33,34. In addition, reduced numbers of circulating EPCs were recently shown to be associated with endothelial dysfunction in healthy adults35. Thus, the defective mobilization of progenitor cells caused by reduced NO bioavailability may impair endogenous vascular repair mechanisms, thereby contributing to the acceleration of vascular aging as well as to reduced neovascularization and cardiac regeneration in patients with ischemic heart disease.

Methods

Animal experiments.

Nos3−/− mice and their age- and sex-matched wild-type littermates (C57BL/6J) were purchased from Jackson Laboratories. Human VEGF165 (500 μg per kg per d) or saline were subcutaneously administered once a day for a period of 3 d, along with 5-FU in a single dose of 250 mg/kg. Recombinant mouse soluble KitL was injected twice daily (150 μg/kg).

Bone marrow transplantation.

CD45+ bone marrow cells were isolated using CD45-bound microbeads and an automated separation device (Miltenyi Biotec). Recipient mice were lethally irradiated with 9.5 Gy and received an intravenous injection of 106 donor bone marrow cells 24 h after irradiation. As a control for transplantation efficiency, bone marrow from CD45.1 mice was transplanted into Nos3−/− and wild-type mice (both CD45.2; Supplementary Fig. 8 online)36. Hind-limb ischemia was induced 6 weeks after bone marrow transplantation.

Hind-limb ischemia.

The proximal femoral artery, including the superficial and deep branches, and the distal portion of the saphenous artery were ligated in the right hind limb37. Two weeks later, we measured blood flow ratios in the ischemic (right) versus normal (left) limbs, using a laser Doppler blood flow imager (Moor Instruments; see Supplementary Fig. 9 online for time course). Capillary density in the semimembraneous and adductor muscles was determined in 5-μm cryosections using FITC-conjugated monoclonal antibody to CD31 (BD PharMingen). The endothelial phenotype was confirmed by immunostaining for the pan-endothelial marker MECA-32 (BD PharMingen). Injected cells were labeled with CellTracker CM-Dil (Molecular Probes). Bone marrow–derived cells were identified by fluorescence in situ hybridization38 (probe from Cambio).

BrdU proliferation assay.

Bone marrow was collected aseptically by flushing tibias and femurs, and was pulsed with 10 μM BrdU (BrdU flow kit, BD PharMingen) for 1 h before immunostaining for surface antigens. Cells were incubated with phycoerythrin-conjugated rat antibody to mouse Sca-1, and a cocktail of biotinylated antibodies to the lineage markers CD3, B220, CD11b, Gr-1 and TER119 (BD PharMingen) for 20 min at 4 °C, followed by streptavidin-allophycocyanin (BD PharMingen). Cells were fixed and permeabilized with Cytofix/Cytoperm buffer (BD PharMingen) according to the manufacturer's instructions, stained with FITC-conjugated antibody to BrdU (BrdU flow kit, BD PharMingen) and fixed in 1% formaldehyde and PBS. The analysis was done using a FACSCalibur with CellQuest software (BD PharMingen).

Colony-forming-unit assay.

Total bone marrow cells (105 per well) were plated in 1.5 ml of 1% methylcellulose (Methocult, StemCell Technologies) containing 15% FBS, 1% BSA, 10 μg/ml human insulin, 200 μg/ml human transferrin, 10−4 M 2-mercaptoethanol, 2 mM L-glutamine, 50 ng/ml mouse stem cell factor, 50 ng/ml human VEGF, 10 ng/ml mouse interleukin-3, 10 ng/ml human interleukin-6 and 3 U/ml human erythropoietin. Colonies (>50 cells) were scored on day 7.

Stromal cell culture.

Adherent cells from total bone marrow were cultured in IMDM with 10% FCS, 50 μM 2-mercaptoethanol and antibiotics for 6 weeks, until the formation of 80% confluent monolayers. MECA-32-positive cells were selected using magnetic beads (antibody from BD PharMingen).

Long-term stem cell assay.

Bone marrow stromal layers were irradiated (20 Gy), overlaid with 106 wild-type or Nos3−/− total bone marrow cells and cultured in IMDM supplemented with 10% FCS, 10% horse serum, 10−5 M hydrocortisone (Sigma) and antibiotics in 12-well plates. Cells were fed twice a week with IMDM. After 35 d, nonadherent cells were flushed from the stromal feeder layers, 105 cells were plated in the methylcellulose clonal assay and colonies were scored after 7 d.

EPC assay.

For EPC culture, mononuclear cells (4 × 106 cells/well) were isolated from spleen-cell homogenates as previously described39, in a modified protocol using 100 ng/ml VEGF (Peprotech) in the culture medium. The identity of EPCs was confirmed by Dil-labeled acetylated low-density lipoprotein/lectin staining and Flk-1 expression.

Detection of MMP-9 and soluble KitL.

For gelatin zymography, bone marrow plasma was collected by flushing femurs and tibias with a total of 0.5 ml PBS. After spinning, supernatants were analyzed for metalloproteinase activity by gelatinolytic zymography as described40. Mouse pro-MMP-9 (20 μg protein) and soluble KitL expression were measured by ELISA (R&D Systems).

Western blot analysis.

Cells were lysed and western blot analysis was done as described28, using monoclonal antibody to eNOS (BD Biosciences) followed by mouse antibody to horseradish peroxidase (Santa Cruz Biotechnology) and reprobing for ERK1/2 (Cell Signaling Technology).

Statistical analysis.

Comparisons between groups were analyzed by two-sided t-test or ANOVA for experiments with more than two subgroups. Post hoc range tests and pairwise multiple comparisons were done using the two-sided t-test with Bonferroni adjustment. The Kaplan-Meier method was used to construct survival curves, and the probability of survival between pairs was compared with a log rank test. P < 0.05 was considered statistically significant. All analyses were done with SPSS 11.5 software.

Note: Supplementary information is available on the Nature Medicine website.