Low-dose irradiation promotes tissue revascularization through VEGF release from mast cells and MMP-9–mediated progenitor cell mobilization

CEPs contribute to vessel formation at the ischemic site (15). We initially asked whether IR releases hematopoietic progenitors and/or CEP into circulation in an MMP-dependent manner. Both MMP-9–deficient and WT mice tolerated sublethal IR at doses of 1.0, 2.0, 6.5, and 7.5 Gy (unpublished data), and showed no difference in the survival of a lethal IR dose (9.5 Gy). BM hematopoietic recovery after a sublethal dose of 6.5 Gy, as determined by BM cellularity (Fig. 1 A), the hematopoietic colony-forming unit cells (CFU-Cs) (Fig. 1 B), CFU endothelial cell (CFU-EC) recovery (Fig. 1 C), and peripheral blood cell count recovery (Fig. 1 D) was delayed in MMP-9^−/− mice compared with controls. These data are in accordance with our previous study in which we showed that MMP-9–mediated KitL release is essential for BM cell recovery, because the delayed BM recovery of MMP-9^−/− mice after 5-fluorouracil could be restored by soluble KitL (14).

When WT mice were irradiated with a sublethal dose of 6.5 Gy as total body IR (TBI), MMP-9 expression in supernatants of BM cells increased starting from day 7 until day 21 after IR (Fig. 1 E), while TIMP-1 expression did not change (not depicted). Supernatants of BM cells of unirradiated animals had only small amounts of pro– and active forms of MMP-9. Plasma levels of pro–MMP-9 also increased after IR (Fig. S1 A). Immunoreactive MMP-9 was low in unirradiated mice (Fig. 1 F a) but was highly expressed 6 and 12 d after TBI in close contact with osteoblasts (Fig. 1 F, c and d). By day 30 after IR, MMP-9 staining was found in the central areas of the BM (Fig. 1 F, d). By contrast, MMP-2 expression showed no major change as determined by zymography (Fig. S1 B).

What is the role of increased MMP-9 activity in the BM endothelium after IR? Under steady-state conditions, the number of capillaries, as a direct measure of BM endothelium, was not different between MMP-9^+/+ and MMP-9^−/− mice. BM vascular recovery following IR was impaired in irradiated MMP-9^−/− mice, as estimated by determining the number of capillaries (Fig. S1, C and D). The VEGFR1⁺ (Fig. S1 E) and VEGFR2⁺ cell fractions (Fig. S1 F) were lower in MMP-9^−/− mice, and their recovery was delayed.

We observed enhanced mobilization of CFU-Cs (Fig. 1 G) and CFU-ECs (Fig. 1 H) into circulation in irradiated WT, but not in MMP-9^−/− mice. Which factors govern progenitor mobilization? We recently reported that plasma levels of stromal cell–derived factor 1 (SDF-1 or CXCL12) and KitL are elevated following 5-fluorouracil treatment (stress-hematopoiesis) (4, 14). Here, we show that IR increases stem cell–active mobilization factors, including plasma KitL (Fig. 1 I), VEGF (Fig. 1 J), SDF-1 (Fig. S1 G), and PlGF (Fig. S1 H) in a partly MMP-9–dependent manner.

IR at high doses might have adverse effects. If IR could ultimately be used to improve ischemia-related diseases, low-dose IR could have efficacy in progenitor recovery and growth factor production. Hematopoietic recovery following low-dose IR was delayed in MMP-9^−/− as compared with MMP-9^+/+ mice, as assayed by the number of BM cells per femur (Fig. 2 A). We observed that post-IR–induced leukopenia was attenuated after an IR dose of 2 Gy (Fig. 2 B) in MMP-9^−/− mice. Significantly, low-dose IR induced MMP-9 expression in BM cells (Fig. 2 C) and augmented plasma KitL (Fig. 2 D), PlGF (Fig. S2 A), and VEGF levels (Fig. 2 E) in an MMP-9–dependent manner. Moreover, circulating CFU-C and CFU-EC release (Fig. S2 B and C) was increased after low-dose IR in MMP-9^+/+ but not MMP-9^−/− mice.

To test the potential therapeutic pro-angiogenic effect of low-dose IR and its regulation by MMP-9, we chose a mouse model of hind limb (HL) ischemia. Operated MMP-9^+/+ mice receiving an IR dose of 2 Gy did not show pad necrosis (unpublished data), whereas unirradiated controls showed necrotic changes. In muscle tissue of MMP-9^+/+ mice, larger vessels stained positive for MMP-9 (Fig. 2 F) after IR (2 Gy). Kidney, lung, and spleen tissue showed no difference in MMP-9 staining before or after IR (unpublished data). HL ischemia increased mobilization of CFU-Cs (Fig. 2 G) and CFU-ECs (Fig. 2 H). However, HL ischemia, in combination with IR, resulted in a profound increase in CFU-Cs (up to 30-fold compared with steady-state values) and CFU-ECs, which coincided with peaks in plasma levels of VEGF (Fig. 2 I), KitL (Fig. S2 D), and PlGF (Fig. S2 E) in MMP-9^+/+, but not in MMP-9^−/− mice. Plasma VEGF, KitL, and PlGF levels rose further (two to threefold) after IR in these mice (Fig. 2 I; and Fig. S2, D and E). Soluble factors known to be involved in ischemic regeneration (16, 17)—such as monocyte chemotactic protein 1 (MCP-1; Fig. S2 F), transforming growth factor β (TGF-β; Fig. S2 G), and basic fibroblast growth factor (unpublished data)—showed slight increases following HL ischemia induction, but were identical in MMP-9^+/+ and MMP-9^−/− mice. Importantly, IR did not further increase the plasma levels of these factors in either group of mice.

To understand which cell types increase after IR, we analyzed peripheral blood mononuclear cells (PBMCs) via FACS (Becton Dickinson) using mAbs against c-kit and VEGFR-2 after IR. No differences were found in the number of VEGFR⁻/c-kit⁺ cells following HL ischemia or after IR. HL ischemia mobilized VEGFR-2⁺/c-kit⁻ cells, most likely comprised of ECs (18) in MMP-9^+/+ mice, but to a lower extent in MMP-9^−/− mice (Fig. 2 J). IR further augmented the release of VEGFR-2⁺/c-kit⁻ cells into circulation in a MMP-9–dependent manner. Importantly, IR increased the VEGFR-2⁺/c-kit⁺ cell population in MMP-9^+/+ but not MMP-9^−/− mice. Toluidine blue O⁺ cells were the major constituent of the VEGFR-2⁺/c-kit⁺ fraction (Fig. 2 J).

Revascularization of the ischemic limb was macroscopically delayed in MMP-9^−/− mice (unpublished data), with fewer von Willebrand factor (vWF)-positive capillaries, resulting in significantly more necrosis compared with MMP-9^+/+ mice (Fig. 3, A–C). In addition, the number of vessels covered with smooth muscle cells (SMCs) was higher in HL ischemia–induced MMP-9^+/+ mice compared with MMP-9^−/− animals (Fig. 3 A).

After IR, HL ischemia–induced MMP-9^+/+ mice showed few signs of necrosis compared with unirradiated controls (Fig. 3, A and B). Muscle tissue from HL ischemia–induced MMP-9^−/− mice showed vast areas of necrotic tissue. IR only slightly decreased the necrotic area in these animals. IR under ischemic conditions increased the number of capillaries in MMP-9^+/+ mice but only slightly in MMP-9^−/− mice (Fig. 3 C). Restoration of vascular function as measured by thermography (Fig. 3, A and D) and laser Doppler (Fig. 3, A and E) paralleled the vascular regeneration. Ten days after induction of HL ischemia, laser Doppler analysis revealed decreased blood flow in HL ischemia–induced MMP-9^−/− compared with MMP-9^+/+ mice (Fig. 3 E). IR of HL ischemia–induced MMP-9^+/+ mice increased both mean temperature (Fig. 3, A and D) and blood flow (Fig. 3, A and E) compared with nonirradiated controls. In contrast, IR of HL ischemia–induced MMP-9^−/− mice only slightly improved mean temperature (Fig. 3, A and D) and blood flow (Fig. 3, A and E). Upon histological examination, necrotic area, density of vWF⁺ vessels, and perfusion of irradiated animals without induction of ischemia was not different in irradiated compared with unirradiated muscle tissue in MMP-9^+/+ or MMP-9^−/− mice (Fig. 3, B–E). These data suggest that low-dose IR alone does not cause visible changes in the muscle microenvironment, but applied under ischemic conditions stimulates angiogenesis.

The increase in Toluidine blue–positive mast cell–like cells in circulation following IR (Fig. 2 J) led us to hypothesize that IR improves mast cell migration into ischemic tissue. A higher number of mast cells was found in ischemic muscle tissue from MMP-9^+/+ compared with MMP-9^−/− mice (Fig. 4 A, a–d). Mast cells were identified as Toluidine blue– and (unpublished data) mast cell tryptase–positive cells. Using adjacent tissue sections, we detected VEGF mRNA in mast cells via in situ hybridization (Fig. 4 A, e–h), implying that these cells are the major source of VEGF in the tissues after ischemia and IR. Staining of adjacent sections with VEGFR-2 mAb revealed that both mast cells and ECs stained positive for VEGFR-2 (Fig. 4 A, i–l). IR further augmented the number of mast cells in the muscle tissue of MMP-9^+/+ mice but not MMP-9^−/− mice (Fig. 4 A, a–d, m).

Because mice had been exposed to TBI, we determined the number of mast cells in other tissues. The number of Toluidine blue–positive mast cells was higher 14 d after low-dose IR in the skin (Fig. 4 B, a and b) of MMP-9^+/+ mice, but not MMP-9^−/− mice, compared with untreated controls. As a measure of mast cell activation, we determined histamine levels in irradiated and unirradiated WT mice. IR increased plasma histamine levels threefold as compared with unirradiated controls (Fig. S2 H).

KitL and VEGF promote mast cell migration (19, 20), and mast cells show baseline MMP-9 activity. We therefore examined whether the decreased number of mast cells in MMP-9^−/− mouse tissue was due to a migratory defect of mast cells toward KitL and/or VEGF. In a transwell migration assay, KitL and VEGF induced migration of MMP-9^+/+ mast cells (Fig. 4 C) but not MMP-9^−/− mast cells. VEGF-mediated migration of MMP-9^+/+ mast cells was partially blocked by neutralizing antibodies against VEGFR-1 or VEGFR-2 but was completely blocked by addition of an MMP inhibitor (MPI CGS 27023A).

PT18 cells (mast cell line) and MMP-9^+/+ mast cells showed baseline VEGF secretion (Fig. 4 D), which was significantly augmented by 2 Gy IR. Addition of KitL only marginally elevated VEGF levels in supernatant of mast cell cultures. VEGF release depended on the presence of MMP-9, because baseline VEGF secretion from MMP-9^−/− mast cells was low, and IR only slightly enhanced VEGF release (Fig. S2 I). The observed differential release of VEGF from mast cells with and without IR was not due to changes in mast cell numbers (unpublished data). The stromal cell line MS-5, on the other hand, produced little VEGF, and VEGF release did not change with IR (Fig. 4 D). Using a more complex BM-derived stromal cell population from MMP-9^+/+ animals, VEGF secretion did not change before (1961 ± 143 pg/mL) or after IR (2107 ± 150 pg/mL).

Which cell type produces KitL? We show that IR increased KitL release by MS-5 cells in vitro, which could be blocked by a synthetic metalloproteinase inhibitor (Fig. 4 E). In contrast, mast cells only showed baseline secretion of KitL and no change in KitL production after IR. Thus our results indicate cooperation between two cell types in mobilizing mast cells: stromal cells produce KitL, while mast cells produce VEGF in response to IR, and both depend on MMP-9.

To test if IR-induced vasculogenesis/angiogenesis is dependent on mast cells in vivo, mast cell–deficient Sl/Sl^d mice and WBB6F1^+/+ controls received 2 Gy IR after induction of HL ischemia. Necrosis was macroscopically detectable on day 7 in Sl/Sl^d, but not as profound in the WT mice (Fig. 4 F). By day 14, blood flow recovered in WBB6F1^+/+ littermates, whereas Sl/Sl^d mice showed complete limb amputation (Fig. 4, F and G). IR accelerated the ischemic recovery in WBB6F1^+/+ mice, but not in Sl/Sl^d mice (Fig. 4 G). When Sl/Sl^d mice with HL ischemia were treated with IR, all mice died by day 14. These data support the hypothesis that improved angiogenesis after IR is driven by mast cells.

rKitL protects against lethal IR and increases the number of progenitors (21). Mobilization of hematopoietic (Fig. 5 A) and endothelial progenitors (Fig. 5 B) following HL ischemia was similar in WBB6F1^+/+ and Sl/Sl^d mice; however, IR-induced progenitor mobilization was completely blocked in Sl/Sl^d but not WBB6F1^+/+ mice, indicating that IR-induced progenitor mobilization depends on the presence of KitL. In confirmation of our data that mast cells are the source of VEGF following HL ischemia and IR, we detected lower VEGF plasma levels in Sl/Sl^d mice after HL ischemia as compared with WBB6F1^+/+ controls (Fig. 5 C); in addition, IR enhanced VEGF plasma levels in WBB6F1^+/+ but not Sl/Sl^d mice.

If VEGF signaling following IR promotes angiogenesis under ischemic conditions, administration of neutralizing antibodies against VEGFR-1 and VEGFR-2 should prevent IR-induced angiogenesis. MMP-9^+/+ mice received vessel ligation followed by TBI of 2 Gy (Fig. 5 D). Macroscopic examination at day 7 revealed that 2 Gy of IR completely prevented signs of limb necrosis, whereas amputation occurred following HL ischemia (Fig. 5 D, b). The effect of IR in preventing tissue necrosis/amputation was blocked by anti–VEGFR-2 treatment or a combination of anti–VEGFR-1 and anti–VEGFR-2 treatment by day 10, but was only partially inhibited in mice receiving anti–VEGFR-1 by day 14 (Fig. 5 D). Single injections of anti–VEGFR-1 resulted in necrosis but not complete limb amputation.

Administration of rVEGF intraperitoneally caused up-regulation of MMP-9 in BM (Fig. 5 E, a) and increased plasma levels of KitL in treated mice (Fig. 5 E, b). If VEGF promotes mast cell migration into ischemic tissue, blockade of VEGF signaling using VEGFR-1 or VEGFR-2 mAbs should decrease the number of mast cells in the muscle tissue following HL ischemia and/or IR. Indeed, HL ischemia–induced mice treated with VEGFR-2 mAbs or a combination of VEGFR-1 and VEGFR-2 mAbs showed a decreased number of mast cells in the muscle tissue compared with untreated controls (Fig. S2 J). Administration of VEGFR-1 diminished the number of mast cells partially. These data underscore the importance of VEGF signaling in regulating the number of resident mast cells under “stress conditions.”

Local low-dose IR rather than TBI would be a more relevant treatment for future clinical applications by reducing systemic side effects. Therefore, C57BL/6 mice received unilateral HL ischemia. Groups of mice were then administered local IR (2 Gy) of the ischemic limb, IR of the contralateral nonischemic limb, or no IR. IR of the contralateral, nonischemic limb controlled for effects due to BM IR. Muscle tissue regeneration was faster in mice when the ischemic limb was irradiated compared with nonirradiated controls (Fig. 5 F). Local IR of the contralateral, nonischemic limb did not result in faster tissue regeneration. Thus, activation of the BM alone was not sufficient to promote angiogenesis. These data implicate that IR of the ischemic tissue, but not of nonischemic tissue, “conditions” the ischemic tissue microenvironment and is critical for IR-induced angiogenesis.

To understand if progenitors mobilized after IR alone could improve vasculogenesis/angiogenesis, PBMCs isolated from irradiated and unirradiated donor mice were transplanted into groups of C57BL/6 recipient mice after HL ischemia surgery. We observed HL ischemic recovery in mice transplanted with PBMCs from unirradiated mice, but not in mice transplanted with PBMCs isolated from 2 Gy irradiated donors (Fig. 5 G). These data implicated that circulating cells induced by IR alone were not sufficient to duplicate the previously observed angiogenesis-promoting effects of IR.

Do BM-derived cells play any role at all in IR-induced angiogenesis under ischemic conditions? When C57BL/6 mice reconstituted with donor GFP-expressing BM cells received HL surgery and 2 Gy IR, BM-derived GFP⁺ cells contributed to vessel formation in muscle tissue of mice that had undergone HL ischemic surgery (Fig. 5 I). These BM-derived cells were covered by smooth muscle actin (SMA)⁺ cells, indicating their incorporation into mature vessels. The highest density of donor-derived GFP⁺ vessels was found in animals that had received HL ischemic surgery and 2 Gy IR (82%) (Fig. 5 I). Donor-derived GFP⁺ cells were undetectable in unirradiated mice without HL ischemia (0%). These data imply that IR applied under ischemic conditions promotes incorporation of BM-derived cells into regenerating vasculature.

To understand whether IR conditions the local ischemic microenvironment to promote mast cell incorporation, we induced bilateral HL ischemia in WBB6F1^+/+ mice. IR of one leg resulted in faster recovery of blood flow compared with the unirradiated ischemic limb (Fig. 5 H). Increased numbers of mast cells were found in muscle tissue of the irradiated but not the nonirradiated limb (Fig. S2 K, a and b). Plasma levels for VEGF and KitL after bilateral HL ischemia further increased following loco-regional IR (Fig. S2, K, c).

We report that low-dose IR promotes neovascularization in the ischemic limb. We provide in vitro and in vivo evidence that mast cells act as a key cellular mediator orchestrating the IR-induced formation of a new vascular network (Fig. 6). IR mediates neovascularization by promoting the release of VEGF from mast cells in a partly MMP-9–dependent manner. VEGF can further up-regulate MMP-9 in the ischemic microenvironment, causing improved mast cell migration. On the other hand, IR-induced KitL increases the number of mast cells in the ischemic tissue and might promote recruitment, incorporation, and differentiation of progenitors into regenerating tissue. The critical importance for mast cells and VEGF in IR-mediated neovascularization under ischemic conditions was underscored by blocking studies using mast cell–deficient Sl/Sl^d mice and blocking Abs against VEGFR-1 and VEGFR-2 to inactivate VEGF signaling.

These data introduce a new mechanism, how IR under ischemic conditions alters the cellular and chemical composition of the ischemic microenvironment. IR promotes tissue regeneration by inducing the expression of genes in mast and stroma cells to further promote incorporation of BM-derived cells, mast cells, pericytes/SMCs, and possibly progenitors to the ischemic side, providing vascular stability and further release of angiogenic factors through the control of proteases. IR has been shown to up-regulate MMP-9 in ECs (22) and to increase the production of nitric oxide (NO) and endothelial nitric oxide synthase (eNOS) (6). NO activates MMP-9 via nitrosylation (23) and improves endothelial and mural cell migration (24). Both NO and eNOS play a critical role in VEGF-induced angiogenesis (25). Mice lacking eNOS show an MMP-9–dependent defect in progenitor cell mobilization and angiogenesis (26). We speculate that the observed IR-induced up-regulation of MMP-9 might involve the NO pathway, but further studies are needed.

BM-derived cells might directly contribute to newly forming vessels or act as bystander cells of the stromal compartment, where they produce cytokines/chemokines (15, 27, 28). Hematopoietic progenitors are mobilized after low-dose IR (29). We show that IR-induced progenitor mobilization is dependent on KitL/mast cells, because no progenitor mobilization was observed in Sl/Sl^d mice. BM-derived GFP⁺ cells contributed to vessel formation in ischemic tissues, and IR improved BM cell incorporation in irradiated ischemic tissues. We demonstrated that low-dose IR through VEGF released from mast cells up-regulates MMP-9 in BM stroma and ECs. These data are in accordance with the idea that IR conditions the ischemic microenvironment (mast or stromal cells) to release soluble factors (VEGF or KitL), which can attract circulating cells and promote their tissue incorporation.

Another hint that IR activates the ischemic microenvironment came from the finding that elevated levels of plasma SDF-1, a chemokine produced by stroma cells, was detectable (30). Progenitor mobilization following HL ischemia has been attributed to increased plasma SDF-1 levels (31). Further blocking studies are needed to determine where SDF-1 fits into our pathway of IR-induced angiogenesis.

This report is the first to demonstrate that IR can mobilize CEPs, and that this mobilization is dependent on MMP-9 and KitL, because it did not take place in MMP-9^−/− and Sl/Sl^d mice. We showed reduced angiogenic revascularization in MMP-9^−/− mice on a CD1 background, confirming data by others using MMP-9^−/− mice in C57BL/6 and 129Sv backgrounds (32, 33). Faster ischemic recovery in MMP-9^+/+ irradiated mice was associated with improved vessel formation. In contrast, newly formed microvessels in MMP-9^−/− mice showed decreased pericyte coverage and slower ischemic recovery after IR. These data are in accordance with studies showing that VEGF pretreatment of VEGFR-1⁺ SMCs activates MMP-9 (34) and that MMP-9^−/− SMCs have decreased cell migration compared with MMP-9^+/+ cells in vitro (35).

The radiosensitivity of white spotting and Sl/Sl^d mice has been attributed to a role for KitL in the proliferation of BM-derived mast cells (36). If IR results in the release of KitL, more c-kit⁺ mast cells should be found in irradiated tissue. Indeed, in the recovery phase after IR, concomitant with increased KitL plasma levels, the number of mast cells was higher in muscle tissue of MMP-9^+/+ mice compared with MMP-9^−/− mice.

Mast cells migrate into tissues via blood circulation as immature cells and undergo complete maturation in the tissues (37). Murine mast cells migrate toward VEGF (20) and express VEGFR-2 (38). We showed that circulating mast cell–like cells express c-kit and VEGFR-2 and that VEGF-induced mast cell migration was partially blocked using VEGFR mAbs. We demonstrated that MMP-9^−/− mast cells failed to migrate toward KitL and VEGF. These data indicate that VEGF improves mast cell migration by further activating MMPs, because a synthetic MPI blocked this migration. Interestingly, a study by Tanaka et al. showed that MMP-9 promoted the migration of mast cell precursors into tissues (5).

Mast cells contain extracellular matrix–degrading MMPs and constitutively secrete VEGF (39, 40). We demonstrate that VEGF production partly depends on the presence of MMP-9, because baseline VEGF release was low in MMP-9^−/− mast cells in vitro. Irradiated MMP-9^+/+ mast cells secreted more VEGF than unirradiated mast cells, consistent with a study showing that VEGF gene expression was up-regulated in irradiated versus unirradiated cells (7). Blocking VEGF signaling using neutralizing mAbs against VEGFR-1 and VEGFR-2 prevented mast cell migration and IR-induced vasculogenesis/angiogenesis. Interestingly, we detected increased levels of histamine in the plasma of IR animals as early as 6 h, indicating that mast cell activation had taken place.

Our data show that mast cells are critical in ischemic recovery and IR-induced angiogenesis: ischemic regeneration was delayed, the number of IR-induced circulating progenitors reduced, and the VEGF plasma levels following IR were significantly lower in mast cell–deficient Sl/Sl^d mice than in unirradiated WT mice. Even though the release of VEGF and KitL depends in part on MMP-9, MMP-9^−/− mice do not completely lack these factors and show growth factor elevation to some degree after ischemia and IR. The leakiness in the system might explain why improved tissue vascularization was detectable in MMP-9^−/− mice, albeit at lower levels. It is also conceivable that MMP-9–independent angiogenesis-modulating factors are released from mast cells under ischemic conditions, but future studies will be necessary. These data support our current model that VEGF and mast cells, partly under the control of MMP-9, are critical in the angiogenesis-promoting effect of IR. It is conceivable that growth factor–independent IR can modulate the vascular tonus of the distal vasculature (41).

The results of the present study have important implications for the use of radiotherapy in nonmalignant and malignant diseases. We demonstrated the clinical potential of low-dose IR to induce neovascularization in ischemia-associated diseases. However, potential adverse effects due to the release of KitL and/or activation of mast cells might occur, including mast cell degranulation with increased pruritus, increased melanization of the epidermis, and allergic-like reactions such as urticaria, with or without respiratory symptoms (42). Likewise, long-term studies are needed to rule out negative side effects resulting from local IR. We propose that IR modulates the ischemic and possibly tumor microenvironment via MMP-9–mediated cytokine and angiogenic factor-driven mast cell and progenitor recruitment. We identified mast cells as a major source of VEGF release, which is known to promote tumor-angiogenesis. Tumor growth is controlled by angiogenesis. In cancer patients, the cytotoxic effect of low-dose IR on tumor cells might outweigh the IR-induced proangiogenic effects, resulting in a net tumor mass reduction. On the other hand, our model might shed light into clinical settings in which a tumor returned despite initial killing of tumor cells. Radio-resistant mast cells, whose number could be further increased by IR-induced release of KitL might guide those processes (21).

MMP-9^+/+ and MMP-9^−/− mice were used after >10 back crosses to CD1 mice (14). KitL WT (WBB6F1^+/+) and KitL-deficient (Sl/Sl^d) and C57BL/6 mice were purchased from SLC. Mice expressing GFP under a β-actin promoter were provided by A. Urabe (Juntendo University, Tokyo, Japan). Animal studies had been approved by the Animal Review Board of Juntendo University, Tokyo.

MMP-9^+/+ and MMP-9^−/− mice were irradiated with 1, 2, 6.5, or 9.5 Gy (Caesium Gy source). Sl/Sl^d and WBB6F1^+/+ littermates were irradiated with 2 Gy. For local IR (2 Gy), mice were kept in a radio-dense 2-cm-thick lead cage; only the irradiated limb left the cage.

MMP-9^+/+ mice received rVEGF 300 ng/mouse/d intraperitoneally.

PBMCs (2 × 10⁶/mouse) from irradiated (2 Gy) and unirradiated C57BL/6 mice (from day 0–16) were injected intravenously into unirradiated HL ischemia–induced mice.

Unilateral or bilateral ligation of the femoral artery and vein and cutaneous vessels branching from the caudal femoral artery side branch were performed. Groups of MMP-9^+/+ and MMP-9^−/− mice or Sl/Sl^d and WBB6F1^+/+ littermates were then irradiated with or without 2.0 Gy. Tissue sections from the lower muscles of ischemic limbs were harvested. Thermography of the collateral region was performed using a Lisca PIM II camera (Gambro). A laser Doppler perfusion image analyzer (Moor Instruments) recorded blood flow postoperatively. Blood was collected via retro-orbital bleeding using heparinized capillaries. Plasma samples were stored at −80°C. White blood cell counts were determined. PBMCs were isolated from heparinized blood after centrifugation over Lympholyte-M (Cedarlane).

Unilateral HL ischemia–inducing surgery was performed on MMP-9^+/+, WBB6F1^+/+, or Sl/Sl^d mice receiving 2 Gy or no IR. Mice were coinjected intraperitoneally with 800 μg anti–mouse VEGFR-1 (MF-1), anti–mouse VEGFR-2 (DC101), or IgG at 2-day intervals starting at day 0 (4).

Paraffin-embedded sections were deparaffinized and stained with hematoxylin-eosin. Vascular structures within the BM with EC morphology, which contained erythrocytes, were counted as capillaries in hematoxylin-eosin–stained BM. The mean number of capillaries per square millimeter (capillary density) of vWF stained slides was counted. The mean ischemic area per square millimeter (n = 10) of hematoxylin-eosin–stained slides was determined using an image analyzer (Carl Zeiss Vision, Hallbergmoss). Necrotic muscle fibers were identified via morphology, differential eosin staining, and presence of infiltrating cells near degenerating fibers. Paraffin sections of muscle or skin sections were stained with Toluidine blue O, and mast cells were counted under a light microscope.

Deparaffinized BM sections were stained for MMP-9 (7-11C) or a murine Ab against MMP-9 as described (14). Muscle sections were stained with CD31 (BD Biosciences, MEC13.3), vWF (DakoCytomation) or VEGFR-2 (DC101). vWF stained sections were developed with 3′,3′-diaminobenzidine. ECs were identified using rat mAb against mouse CD31 followed by biotin-conjugated Goat anti–rat IgG and Cy3-conjugated streptavidin (Alexa 594, Molecular Probes) staining. SMCs were identified using mouse anti-human SMA (1A4) followed by biotin-conjugated Goat anti–mouse IgG and FITC-conjugated streptavidin (Alexa 488, Molecular Probes).

VEGF was detected in muscle sections using a murine-specific probe for VEGF as described previously (43).

Murine BM cells (10⁶ cells) from MMP-9^+/+ and MMP-9^−/− isolated before and after TBI were placed in serum-free medium (X-VIVO-15) overnight. Supernatants were analyzed via Western blotting using murine mAb to MMP-9 as described previously (14).

Groups of MMP-9^+/+ and MMP-9^−/− mice were set up in which HL ischemia was followed by 2 Gy IR or no IR. Mice were bled on day 7. PBMCs were labeled with VEGFR-2/Cy2 and c-kit/PE, separated using MoFlo. Cytospins were prepared from the cell fractions, which were stained with Toluidine blue.

PT18 cells (5 × 10⁵ cells/mL) and a confluent layer of MS-5 cells were cultured for 48 h in serum-free medium. Cells were irradiated using a dose of 2 Gy or were not irradiated. Supernatants were stored at −20°C.

BM-derived mast cells were produced by culturing MMP-9^+/+ and MMP-9^−/− BM cells in RPMI 1640 supplemented with 0.075% sodium bicarbonate (Sigma), 1 mM nonessential amino acids (GIBCO BRL), 5.5 × 10⁻⁵ M 2-ME (Wako), 1 mM sodium pyruvate (ICN Biomedicals), 10% FBS (GIBCO BRL), and 10% WEHI-3–derived conditioned medium. Mast cells from 3-wk cultures were cultured over night in serum-free medium (Ex Vivo 15). Supernatants were stored at −20°C. 3-wk-old BM mast cell cultures contained >97% c-kit⁺ cells by FACS and 95% Toluidine blue⁺ cells.

Mast cells derived from MMP-9^+/+ and MMP-9^−/− BM cultures were added to 5-μm pore size transwell inserts (Costar). Murine rKitL (100 ng/mL; PeproTech) or rVEGF (200 ng/mL; PeproTech) were added to the lower chamber with/without neutralizing mAbs against murine VEGFR1 (MF-1), VEGFR1 (DC101), the MPI CGS 27023A (1 nM; Novartis), and neutralizing mAbs against KitL (R&D Systems) that were added to both chambers. Transmigrated cells were expressed as the percentage of transmigrated cells.

PBMCs (10⁵ cells/plate) or BM mononuclear cells (BMNCs) (10⁵ cells/plate) were assayed as described previously (14).

PBMCs or BMNCs before and obtained at various time points after IR (2 Gy) were cultured as described previously (14). To determine the EC origin, adherent cells were stained with anti–mouse VEGFR-2 (Avas-12a1, PharMingen), followed by peroxidase-conjugated streptavidin incubation (DakoCytomation), and visualized using an AEC+ Substrate-Chromosome System (DakoCytomation). Cultures were costained using rat monoclonal anti–mouse CD45 (BD Biosciences) to exclude leukocytes, followed by alkaline phosphatase–labeled streptavidin (Vector), and visualized using Nitro blue Tetrazolium Chloride (DakoCytomation). VEGFR-1⁺ and VEGFR-2⁺ cells were quantified via FACS using Cy2-labeled mAb to VEGFR-1 (MF-1) and VEGFR-2 (DC101).

Supernatants from BM cells from MMP-9 WT mice before and after IR with 6.5 Gy and 2 Gy were cultured overnight in serum-free medium. Supernatants were analyzed as described previously (14).

In plasma/culture supernatants SDF-1, VEGF, PlGF, KitL, MCP-1, and pro–MMP-9 were measured via ELISA (R&D Systems). TGF-β plasma levels were assayed using a TGF-β ELISA kit (Genzyme Boston). Plasma obtained 20 min after IR was analyzed for histamine via ELISA (ImmunoTECH).

Data were analyzed using an unpaired two-tailed Student's t test and are expressed as the mean ± SEM. P values less than 0.05 were considered significant.

Fig. S1 shows MMP-9 but not MMP-2 expression and plasma levels for SDF-1 and PlGF and BM recovery of VEGFR-1⁺ and VEGFR-2⁺ cells following IR with 6.5 Gy. Fig. S2 shows progenitor mobilization after 2 Gy IR and includes data on KitL, PlGF, MCP-1, TGF-β, and histamine release following HL ischemia in combination with 2 Gy IR. Fig. S2 also contains data on VEGF release from mast cells in vitro and data on the number of mast cells in muscle tissues after mice had been treated with neutralizing antibody against VEGFR-1 and VEGFR-2, as well as data on the number of mast cells and the release of VEGF and KitL in mice following bilateral ligation of the femoral artery and vein.

We thank Drs. A. Furuhata, N. Tada, and M. Kuhara for their excellent technical assistance and H. Nakauchi for his helpful discussions.

This work was supported by grants from the JSPS (to B. Heissig), the Leukemia and Lymphoma Foundation (to B. Heissig), and the NIH (AG23218 and ES012801 to Z. Werb); Grants-in-Aid for Scientific Research from MEXT (to K. Hattori, B. Heissig); and in part by grants from the Mochida Memorial Foundation (to K. Hattori), AstraZeneka (to K. Hattori), Naito Memorial Foundation (to K. Hattori), Mitsubishi Pharma Research Foundation (to K. Hattori), Mitsukoshi Health and Welfare Foundation (to K. Hattori), Mitsui Life Social Welfare Foundation (to K. Hattori), and Sagawa Foundation (to K. Hattori). Nihon Kohden Co. Ltd. provided thermography analysis.

The authors have no conflicting financial interests.