It has been reported that bone marrow cells (BMCs) differentiate into endothelial cells of blood vessels, and that granulocyte colony-stimulating factor (G-CSF) mobilizes progenitors in the BMCs to the peripheral blood, while macrophage colony-stimulating factor (M-CSF) augments the production of monocytes. We examined whether M-CSF augments the differentiation of BMCs into endothelial cells of blood vessels using a hindlimb-ischemic model. Either G-CSF or M-CSF, or both, was administered to the hindlimb-ischemic mice for 3 days. Both M-CSF and G-CSF augmented the differentiation of BMCs into endothelial cells of blood vessels through vascular endothelial cell growth factor (VEGF), resulting in early recovery of blood flow in the ischemic limbs.
It has been reported that endothelial progenitor cells (EPCs) exist in the peripheral blood, and that they can differentiate into the endothelial cells of blood vessels . Bone marrow cells (BMCs) have also been reported to differentiate into the endothelial cells of blood vessels .
Several cytokines, such as granulocyte colony-stimulating factor (G-CSF), macrophage colony-stimulating factor (M-CSF), and granulocyte macrophage colony-stimulating factor (GM-CSF) are known to augment the numbers and functions of granulocyte-monocyte lineage cells [3, 4, 5, 6]. G-CSF and GM-CSF have been reported to mobilize peripheral blood stem cells (PBSCs) [3, 7] and EPCs [8, 9]. However, it remains unclear whether M-CSF contributes to neovascularization and mobilizes EPCs from the bone marrow. Here, we report that M-CSF mobilizes EPCs from the bone marrow and that both M-CSF as well as G-CSF augment neovascularization of ischemic limbs.
Materials and Methods
C57BL/6 (B6, H-2b) and C3H HeN (C3H, H-2k) mice were purchased from SLC (Shizuoka, Japan; http://22.214.171.124/SLC/SLC.htm). Enhanced green fluorescent protein (EGFP)–transgenic mice (green mice), which were based on B6 mice, were kindly donated by Dr. H. Okabe (Osaka University, Osaka, Japan) .
Preparation of Hindlimb Ischemia
On day 0, unilateral hindlimb ischemia was induced by resecting the right femoral artery and vein of B6 mice or bone marrow–transplanted B6 mice. Three days after surgery, the skeletal muscles (bilateral gastrocnemius muscles) were isolated, embedded in optimal cutting temperature (OCT) compound (Sakura, Elkhart, IN), and snap-frozen in liquid nitrogen.
Administration of G-CSF and M-CSF
Hindlimb ischemia was induced in mice on day 0. Either recombinant human G-CSF (250 μg/kg) or recombinant human M-CSF (250 μg/kg), or both—kindly donated by Chugai Pharmaceutical Co. (Tokyo, Japan; http://www.chugai-pharm.co.jp/english/) and Kyowa Hakko Kogyo (Tokyo, Japan; http://www.kyowa.co.jp/en/), respectively—was administered to the mice i.p. for 3 days (day 0, day 1, and day 2) [5, 7]. The dosages of the cytokines were determined according to the method described in our previous paper . On day 3, the mice were sacrificed to study neovascularization. Bilateral gastrocnemius muscles were isolated and snap-frozen in liquid nitrogen.
To study the regeneration of blood vessels from BMCs, we performed two experiments. One was chimeric mice from B6-based transgenic mice expressing EGFP  into B6 mice. The other was chimeric mice from C3H into B6 mice. For bone marrow transplantation (BMT), we transplanted donor BMCs (2 × 107) into the bone marrow cavity of lethally irradiated recipients, as previously described . One month after BMT, we confirmed that more than 90% of the peripheral blood in the mice showed donor-type major histocompatability complex (MHC) (H-2k) molecules. On day 0, hindlimb ischemia was induced, and either G-CSF or M-CSF, or both, was administered to the mice for a continuous 3 days (day 0–2). On day 3, the mice were sacrificed, and the gastrocnemius muscles were collected for microscopical observation.
Histological analyses were performed as previously described . Briefly, 2-μm sections of frozen muscles were stained with fluorescein isothiocyanate (FITC)–labeled anti-CD31 Ab (Caltag Laboratories, Burlingame, CA; http://www.caltag.com/) to detect blood vessels. To detect donor-derived blood vessels, the sections obtained from the chimeric mice (EGFP→B6) were stained with phycoerythrin (PE)–labeled anti-CD31 Ab (Caltag Laboratories). To examine whether or not newly developed vessels are due to fusion, the sections obtained from chimeric mice (C3H→B6) were stained with PE-labeled anti-CD31 Ab, FITC-labeled anti-H-2Kb Ab, and biotin-labeled anti-H-2Kk Ab, followed by Alexa-labeled avidin. They were then observed using a confocal laser microscope (Olympus, Tokyo, Japan; http://www.olympus.co.jp/en/).
Laser Doppler Perfusion Image
The hind limbs of mice were shaved using a razor. The mice were anesthetized with 160 mg/kg pentobarbital, then fixed supine on the cork plate. We next measured the blood flow of limbs using a laser Doppler perfusion image (LPDI) analyzer (Moor Instruments, Millwey, Devon, England; http://www.moor.co.uk/), as described previously . The LPDI index was expressed as the ratio of the ischemic (right) to normal (left) limb blood flow.
The skin temperature of the lower part of the femur was measured with a thermograph (INFRA-EYE 1200A, Fujitsu, Tokyo, Japan; http://www.fujitsu.com) at a room temperature of 18°–20°C. The difference of the skin temperature between control legs and ischemina-induced legs was calculated (skin temperature of the control leg minus skin temperature of the ischemia-induced leg).
Estimation of Endothelial Progenitor Cells in Peripheral Blood
Three days after injecting the mice with either G-CSF or M-CSF, or both (each at 250 μg/kg per day), the cell numbers in the peripheral blood were counted using an SF-3000 auto-analyzer for the peripheral blood (Sysmex Corporation, Kobe, Japan; http://www.sysmex.com/). To stain surface markers of white blood cells (WBCs) in the peripheral blood, the red blood cells (RBCs) in the peripheral blood were then lysed with PharM Lyse (BD Biosciences-Pharningen, San Diego, CA; http://www.bdbiosciences.com/pharmingen/). Since Sca-1+ cells , Flk-1+/CD45− cells, and Sca-1+/c-kit+/CD45− cells have been reported to be an EPC-enriched fraction , peripheral blood cells (PBCs) were stained with PE-labeled anti-Sca-1 Ab (BD Biosciences-Pharmingen) + TC-labeled CD3 Ab (Caltag Laboratories) + TC-labeled CD19 Ab (Caltag Laboratories); TC-labeled anti-CD45 Ab (Caltag Laboratories) + PE-labeled anti-Sca-1 Ab (Caltag Laboratories) + FITC-labeled anti-c-kit Ab (Caltag Laboratories); or TC-labeled anti-CD45 Ab + PE-labeled anti-Flk-1 (VEGF-R2) Ab (BD Biosciences-Pharmingen). The stained cells were analyzed using a FACScan (BD Biosciences-Pharmingen). The numbers of positive cells for the indicated markers were calculated with the percentage of the cells in WBCs and the number of WBCs in the peripheral blood.
Enzyme-Linked Immunoassay (ELISA) for Vascular Endothelial Growth Factor (VEGF)
Spleen cells and BMCs from B6 mice were adjusted to 2 ×106/ml and cultured with or without either G-CSF or M-CSF, or both (each at 10 ng/ml) for 1 day or 3 days. Supernatants were collected after culture for the indicated days, and ELISA for VEGF was performed using the Quantikine: Mouse VEGF Immunoassay Kit (R&D Systems, Minneapolis, MN; http://www.rndsystems.com/).
Statistical analyses were performed using Student's t-test.
Neovascularization Induced by G-CSF with or without M-CSF in Ischemic Limbs
We first examined whether either G-CSF or M-CSF, or both, is involved in the formation of new blood vessels. Either G-CSF or M-CSF, or both, was administered to mice in which the right hindlimbs were induced with ischemia, as described in Materials and Methods. On day 3, the skeletal muscles of the lower limbs were collected. The specimens were stained with anit-CD31 Ab, and blood vessels were detected by CD31+ cells and their shape (morphology).
As shown in Figure 1, nontreated limbs showed similar numbers of blood vessels in all groups. However, the ischemic hindlimbs of mice treated with either G-CSF or M-CSF, or both, showed a significant increase in the number of blood vessels in the skeletal muscles. We also obtained similar results using anti-CD146 Ab, which is also a specific marker for endothelial cells (data not shown).
There are three possibilities: (1) Either G-CSF or M-CSF, or both, accelerate the extension of existing blood vessels, (2) Either G-CSF or M-CSF, or both, accelerate the generation of new blood vessels from EPCs in the peripheral blood, and (3) both (1) and (2).
Differentiation from BMCs into Blood Vessels
It has been reported that the EPCs in the peripheral blood and BMCs differentiate into endothelial cells of blood vessels [1, 2]. We therefore examined whether BMCs could differentiate into endothelial cells of blood vessels. One month after the BMCs of EGFP-transgenic mice were transplanted into the lethally irradiated B6 mice, unilateral hindlimb ischemia was induced. Either G-CSF or M-CSF, or both, was then injected i.p. for 3 days (day 0 to day 2). On day 3, the mice were sacrificed, and frozen sections of the gastrocnemius muscles were prepared to examine vascularization using a confocal microscopy. In the ischemic limbs of either G-CSF or M-CSF, or both, treated mice, we found cells expressing both EGFP and CD31 (Fig. 2). The numbers of blood vessels in each group were similar to those in the experiment using B6 mice (Fig. 1 and Fig. 2B), suggesting that BMT had no effect on the neovascularization by either G-CSF or M-CSF, or both.
Next, we counted EGFP+/CD31+ capillaries (Fig. 2C), since EGFP+ cells are thought to be derived from transplanted BMCs. Bone marrow–derived endothelial cells, the number of which increased in ischemic limbs of cytokine-treated mice, accounted for close to half of the augmented blood vessels in the cytokine-treated groups (Fig. 2B, 2C). These results suggest that G-CSF and M-CSF induce not only the proliferation of existing endothelial cells but also the differentiation from BMCs into the endothelial cells.
Since there are several reports suggesting that cell fusion instead of regeneration should be considered , we performed further experiments to assess whether some new blood vessels are truly derived from BMCs. BMT from C3H mice to B6 mice were performed, as described in Materials and Methods. One month after BMT, we confirmed that more than 90% of the peripheral blood showed donor-type MHC (H-2k) molecules, suggesting that host BMCs (H-2b) were replaced by donor BMCs (H-2k). We induced unilateral hindlimb ischemia in the mice, followed by cytokine injection for 3 days, as described above. As shown in Figure 3, the endothelium cells, which wereH-2k+/H-2b−/CD31+, existed in the capillary walls of ischemia-induced muscle of cytokine-treated mice. Moreover, the numbers of H-2k+/H-2b−/CD31+ capillary were similar to those of EGFP+ capillaries (Fig. 2 and data not shown). These results suggest that the expression of donor MHC molecules by endothelial cells was not due to cell fusion but to real differentiation from donor BMCs.
Increases in Number of EPCs
We next examined the number of EPC-enriched populations present in the circulating blood. Several surface markers of EPCs have been reported: Sca-1+/CD3−/CD19− cells are thought of as an EPC-enriched population , and both Flk-1+/CD45− cells and Sca-1+/c-kit+/CD45− cells have been reported to be an EPC-enriched fraction . The cell numbers in these populations increased in the mice treated with cytokines, particularly G-CSF plus M-CSF (Fig. 4). These results suggest that G-CSF and M-CSF accelerate the mobilization of EPCs from the bone marrow.
Functional Analyses of Increased Blood Vessels by G-CSF and M-CSF
We examined the functional effects of G-CSF and M-CSF on improvement in the blood flow in the ischemic limbs using LDPI and thermography. As shown in Figure 5, blood flow recovered rapidly under the condition with either G-CSF or M-CSF (or both). Particularly, G-CSF plus M-CSF showed remarkable recovery, an example of which was the LPDI of ischemic limbs in the mice treated with G-CSF and M-CSF, reaching around 78% on day 7, while that of ischemic limbs in the untreated mice was only around 38%. These results were confirmed using thermography (Fig. 6). All three CSF-treated mice showed better skin temperature of ischemic limbs than did limbs of saline-treated mice.
Increase in VEGF Production in Bone Marrow by G-CSF and M-CSF
To clarify the mechanisms underlying the mobilization of EPCs in the peripheral blood by G-CSF and M-CSF, we performed ELISA for VEGF, since it has been reported that M-CSF stimulates monocytes to produce VEGF . At first, we performed ELISA using the sera of mice treated with either G-CSF or M-CSF, or both (each at 250 μg/kg) for 3 days, as described in Material and Methods. However, we did not detect VEGF in the sera using the ELISA kit (data not shown).
Next, we cultured spleen cells and BMCs (2 × 106/ml) with, or both G-CSF or M-CSF (each at 10 ng/ml), then we measured the VEGF using the ELISA kit. We did not detect VEGF in the supernatants of spleen cells even when cytokines had been added, but we did in the supernatants of BMCs (Fig. 7). The amount of VEGF in the supernatants of BMCs with either G-CSF or M-CSF, or both, increased. These results suggest that G-CSF and M-CSF stimulate BMCs, probably monocyte-lineage cells, to produce VEGF, followed by an increase in the number of EPCs and augmentation of their mobilization.
In the study reported here, we examined whether either M-CSF and G-CSF, or both, augment the formation of new blood vessels in ischemic limbs. We demonstrated that M-CSF and G-CSF stimulates the differentiation of BMCs into endothelial cells of blood vessels in ischemic limbs and increases blood flow.
The existence of a common progenitor of both hematopoietic cells and endothelial cells of blood vessels has been reported . It has also been reported that PBCs and BMCs have the capacity to differentiate into endothelial cells [1, 2]. These results suggest that EPCs exist in PBCs and BMCs, and that these EPCs can differentiate into endothelial cells of blood vessels and contribute to neovascularization. Recently, it has been reported that human CD34+ cells mobilized by G-CSF are involved in neovasculogenesis in infarcted hearts of rats . It has also been reported that GM-CSF mobilizes EPCs from BMCs and contributes to neovascularization . These newly formed blood vessels were found to work functionally; transplanted human EPCs prevented ischemic hindlimbs of nude mice from falling off , providing an experimental basis for the ongoing therapies using BMCs to treat arteriosclerosis obliterans .
These results suggest that G-CSF and GM-CSF mobilize not only hemopoietic progenitors but also EPCs and raise the possibility of therapy for ischemic diseases using cytokine-induced neovascularization. M-CSF is also well known as a cytokine for augmenting the proliferation and function of monocytes . We have here demonstrated that M-CSF contributes to neovascularization and that a part of the effect can be attributed to mobilization of EPCs from the bone marrow, resulting in an early recovery of blood flow in the ischemic limb. Although Scholz et al.  stated that monocytes are related to angiogenesis, when arteries are occluded, and Eubank et al.  recently reported that M-CSF induces the differentiation of the monocytes that produce VEGF, we did not detect VEGF in the sera of mice injected with G-CSF or M-CSF for the continuous 3 days. Neither did we detect VEGF in the supernatants of the spleen cells cultured with G-CSF or M-CSF. However, we detected a high concentration of VEGF in supernatants of BMCs cultured with either G-CSF or M-CSF. These results suggest that G-CSF and M-CSF augment EPCs through VEGF production.
Mobilized EPCs could participate in the development of collateral arteries. In our experiments, the cell numbers of EPC-enriched fractions in the peripheral blood mobilized by G-CSF plus M-CSF were significantly higher than those by each single factor. However, the number of blood vessels generated by G-CSF plus M-CSF is comparable with that by each single factor. These results suggest that the trapping of EPCs in the ischemic limbs might be limited even if many more EPCs exist in the peripheral blood.
We used 250 μg/kg of G-CSF or M-CSF in this experiment, as in our previous experiments . Although the dosages of the cytokines (for example, 200–250 μg/kg) seem to be extremely high in comparison with human dosages, both our group and other groups have used these doses for experiments to mobilize hematopoietic progenitor cells [22, 23].
It is well known that inflammation is involved in the development and progression of atherosclerosis . Indeed, several papers describe the possibility of enhancement of atherosclerosis by M-CSF and G-CSF [25, 26]. Therefore, in clinical settings, the short-term use of cytokines is thought to be better than extended use in order to limit the exacerbation of atherosclerosis. Bensinger et al.  have described that, in studies with humans, the absolute number of CD34+ cells per ml in the peripheral blood increases 10-fold over baseline and peaks on approximately day 5 of G-CSF. They have also shown that 3-day administration of G-CSF is sufficient to mobilize CD34+ cells. Therefore, we examined the effects of short-term administration of G-CSF or M-CSF on acute ischemia. Our results also showed that the 3-day administration of G-CSF or M-CSF is effective in augmenting blood vessels in acute ischemia.
Previous experiments also demonstrated that the peak of incorporation of human EPCs that had been administered into nude mice in which ischemic limbs had been induced was within 3–7 days and rescued the ischemic limbs . Flk-1+ cells derived from embryonal stem cells expressed VE-cadherin, one of the markers of endothelial cells, after only 1 day of culture . Therefore, we analyzed the number of capillaries in the ischemic muscles and the temperature of the skin on day 3. In the control limbs, in which no ischemia was induced, there was no difference in neovascularization between saline-injected mice and G-CSF- or M-CSF-treated mice. These results suggest that mobilized EPCs might home in on the sites of ischemia, which is where neovascularization is necessary; no unnecessary angiogenesis therefore occurs. This would be a valuable phenomenon in clinical applications.
The differences in LPDI index and skin temperature between the nontreated limbs and ischemic limbs of G-CSF-and M-CSF-treated mice were significantly less than in the saline-injected mice. These results reflect enhanced blood flow in the ischemic limbs.
It has been reported that M-CSF has no effect on the number of peripheral blood cells or monocytes in the peripheral blood , whereas G-CSF and GM-CSF cause an increase in the number of peripheral blood cells, especially granulocytes [2, 3]. Therefore, M-CSF would be readily applicable to the treatment of various diseases in humans, since it is well known that leukocytosis, particularly granulocytosis, accelerates tissue injury .
We thank professor H. Okabe (Osaka University, Osaka, Japan) for the donation of EGFP-transgenic mice. We also thank Ms. Sachiko Miura, Ms. Mari Murakami-Shinkawa, and Ms. Yoko Tokuyama for their expert technical assistance, and also Mr. Hilary Eastwick-Field and Ms. Keiko Ando for the preparation of this manuscript. Keizo Minamino and Yasushi Adachi contributed equally to this work.