Main

HSCs can be found in proximity to bone marrow sinusoidal vessels1, and bone marrow endothelial cells regulate HSC homeostasis and regeneration2,3,4,5,6,7,8. It has been previously reported that maintenance of the HSC pool in mice is dependent on the expression of stem cell factor by bone marrow endothelial or perivascular cells, demonstrating the important role of endothelial and perivascular cells in maintaining the HSC pool during homeostasis8. We have shown that adult sources of endothelial cells produce soluble growth factors that promote the expansion of human HSCs in vitro2 and support the regeneration of mouse and human HSCs in vitro after radiation exposure2,3,9,10,11. We have also demonstrated that systemic infusion of autologous or allogeneic endothelial cells accelerates bone marrow HSC reconstitution and hematologic recovery in mice after radiation-induced myelosuppression4,12. A previous study demonstrated a requirement for vascular endothelial growth factor receptor 2 (VEGFR2)-expressing sinusoidal endothelial cells for normal hematologic recovery after TBI6. Similarly, systemic delivery of an antibody to vascular endothelial cadherin (VE-cadherin), which inhibits bone marrow vasculogenesis, substantially delays hematologic recovery after myelosuppression4,5. However, the precise mechanisms through which bone marrow endothelial cells regulate hematopoietic regeneration remain unknown.

Here, using a screen for cytokines in the bone marrow serum from Tie2Cre; Bak1−/−; Baxflox/– mice, which lack BAK and BAX in TIE2-expressing endothelial cells13 and show a strong radioprotective phenotype, we identify EGF as a candidate endothelial cell–derived mediator of radioprotection of the hematopoietic system and show that EGFR signaling regulates hematopoietic regeneration in vivo. Notably, systemic administration of EGF accelerated the recovery of long-term HSCs and improved the survival of mice after radiation-induced myelosuppression, whereas pharmacologic inhibition or genetic deficiency of EGFR antagonized hematopoietic regeneration in vivo.

Results

EGF mediates HSC regeneration in vitro

We previously developed a genetic model to delete the genes encoding BAK and BAX, which regulate the intrinsic pathway of apoptosis13, in TIE2-expressing endothelial cells as a means to protect bone marrow endothelial cells from radiation-induced injury14. After high-dose TBI, Tie2Cre; Bak1−/−; Baxflox/– mice showed protection of the bone marrow vascular and HSC compartments as well as marked improvement in survival of the mice compared to Tie2Cre; Bak1−/−; Baxflox/+ mice, which retain one allele of Bax, or compared to wild-type mice14. To identify secreted factors produced by TIE2-expressing bone marrow endothelial cells that might contribute to the radioprotection observed in Tie2Cre; Bak1−/−; Baxflox/– mice, we generated primary bone marrow endothelial cell lines (CD45−, von Willebrand factor (VWF)+, lectin+ and acetylated low-density lipoprotein (AcLDL)+) from Tie2Cre; Bak1−/−; Baxflox/– mice (FL/– endothelial cells) and Tie2Cre; Bak1−/−; Baxflox/+ mice (FL/+ endothelial cells), as previously described2,3,9,15. When we irradiated wild-type bone marrow c-Kit+Sca-1+Lin− (KSL) progenitor cells with 300 cGy in vitro and then plated them in noncontact culture with FL/– endothelial cells, we found significant increases in the recovery of total cells, colony-forming cells (CFCs) and colony-forming units, spleen, day 12 (CFU-S12) at day +7 compared to when we plated them in noncontact FL/+ endothelial cell cultures (Fig. 1a). These results suggest that bone marrow endothelial cells from Tie2Cre; Bak1−/−; Baxflox/– mice produce soluble factors that promote hematopoietic stem and progenitor cell regeneration after radiation injury. In a complementary experiment, the addition of bone marrow serum from irradiated Tie2Cre; Bak1−/−; Baxflox/– mice to cultures of irradiated bone marrow KSL cells promoted the recovery of total cells and CFCs in 7-d culture, whereas bone marrow serum from Tie2Cre; Bak1−/−; Baxflox/+ mice had no beneficial effect (Supplementary Fig. 1a).

Figure 1: TIE2-expressing bone marrow endothelial cells produce EGF, and EGF mediates HSC regeneration after irradiation.
figure 1

(a) Mean numbers (± s.e.m.) of total cells (left), CFCs (middle) and CFU-S12 (right) at day 7 from noncontact cultures of 300 cGy–irradiated bone marrow KSL cells with FL/– bone marrow endothelial cells compared to culture with FL/+ bone marrow endothelial cells or cytokines alone (TSF). Input, population of KSL cells at day 0. *P = 0.003, ^P = 0.04 compared to TSF and FL/+, respectively, for total cells (n = 3–7 experiments per condition); *P < 0.0001, ^P < 0.0001 compared to TSF and FL/+, respectively, for CFCs (n = 3 experiments per condition, t test); *P = 0.04, ^P = 0.02 compared to TSF and FL/+, respectively, for CFU-S12 (n = 3–5 experiments per condition). (b) Left, concentrations of EGF in the bone marrow serum of Tie2Cre; Bak1−/−; Baxflox/– (red line), Tie2Cre; Bak1−/−; Baxflox/+ (blue line) and C57BL/6 mice (black line) before irradiation (Nonirrad) and at 6 h and 7 d after 750-cGy irradiation. *P = 0.02, **P = 0.04, ***P = 0.04 compared to Tie2Cre; Bak1−/−; Baxflox/+ mice determined by one-tailed t test. n = 3 mice per condition; data are shown as the means ± s.e.m. Right, EGF expression in FL/–, FL/+ and C57BL/6 (BL/6) endothelial cells by quantitative RT-PCR. *P = 0.002, ^P = 0.003 compared to BL/6 and FL/+ endothelial cells, respectively, determined by two-tailed t test. Data are shown as the means ± s.e.m. n = 3 experiments per group. (c) Left, representative EGFR surface expression on bone marrow Lin+, Lin− and KSL cells from C57BL/6 mice. Right, representative EGFR surface expression on nonirradiated bone marrow SLAM+ KSL cells and at 4 h after 700-cGy irradiation. The numbers shown indicate the pergentage of EGFR surface expression on the indicated cell population. (d) Left, EGFR surface expression in each bone marrow subset. *P = 0.009, ^P < 0.001 for bone marrow KSL cells compared to bone marrow Lin+ and Lin− cells, respectively. *P = 0.02, ^P = 0.004 for bone marrow SLAM+ KSL cells compared to Lin+ and Lin− cells, respectively (n = 3–9 experiments per group). Data are shown as the means ± s.e.m. Right, EGFR expression in nonirradiated bone marrow SLAM+ KSL (red line), KSL (blue line), Lin− (black line) and Lin+ cells (green line, along x axis) and at 4 h after 300-cGy and 700-cGy TBI. *P = 0.002 for the difference in SLAM+ KSL cells between the 300 cGy–irradiated and nonirradiated groups, *P = 0.001 for the difference between the 700 cGy–irradiated and nonirradiated groups (n = 3–7 experiments per group). Data are shown as the means ± s.e.m. (e) Representative FACS analyses of phosphorylation of EGFR at Tyr1173 (pEGFR) in bone marrow KSL (left) and bone marrow SLAM+ KSL (middle) cells in serum-free culture (blue curve) or in the same conditions plus EGF (red curve). Right, pEGFR in bone marrow KSL cells in mice treated with EGF (red curve) or saline (blue). Isotype-stained cells are shown in black. Percentages of EGFR phosphorylation are shown for each group at right. *P = 0.008 for EGF compared to control cultures of KSL cells, *P < 0.001 for EGF compared to control cultures of bone marrow SLAM+ KSL cells (n = 5 experiments per group). Data are shown as the means ± s.e.m. *P = 0.03 for EGF compared to saline treatment of KSL cells in vivo (n = 4 experiments per group). (f) CFC and CFU-S12 content of cultures of irradiated bone marrow KSL cells with TSF or TSF plus EGF. *P = 0.0002, *P = 0.0003 for CFCs (n = 8 experiments per group) and CFU-S12 (n = 9–12 experiments per group), respectively, compared to TSF alone. Data are shown as the means ± s.e.m. (g) Peripheral blood donor CD45.1+ cell, myeloid (Mac-1/Gr-1), B cell (B220) and T cell (Thy1.2) engraftment at 8 weeks after transplant in CD45.2+ mice after transplantation of irradiated, TSF-cultured CD34−KSL cells or irradiated, TSF- and EGF-cultured CD34−KSL cells (n = 7–9 mice per group). *P = 0.002 for CD45.1+ engraftment, *P = 0.002 for myeloid engraftment. Right, total peripheral blood CD45.1+ cell engraftment over time (red line, TSF plus EGF culture; black line, TSF alone). *P = 0.002. Data are shown as the means ± s.e.m. (h) Numbers of KSL cells and CFU-S12 from cultures of bone marrow KSL cells with TSF or TSF plus EGF. *P = 0.03, *P = 0.004. Data are shown as the means ± s.e.m. n = 4–6 experiments per group. Right, percentage donor CD45.1+ cell engraftment at 12 weeks in mice transplanted with nonirradiated and TSF-cultured CD34− KSL cells or nonirradiated and TSF- and EGF-cultured CD34− KSL cells. *P = 0.04 determined by two-tailed t test. Data are shown as the means ± s.e.m. n = 4–5 mice per group. A Mann-Whitney test was used for all statistical analyses unless otherwise noted.

Full size image

To identify paracrine factors in the bone marrow of Tie2Cre; Bak1−/−; Baxflox/– mice that might contribute to the radioprotection of hematopoietic stem and progenitor cells in vivo, we assessed cytokine levels using a cytokine array on the bone marrow serum from Tie2Cre; Bak1−/−; Baxflox/–, Tie2Cre; Bak1−/−; Baxflox/+ and wild-type C57BL/6 mice before and after 750-cGy TBI. Within the list of cytokines that we described to be significantly enriched in bone marrow serum from Tie2Cre; Bak1−/−; Baxflox/– compared to Tie2Cre; Bak1−/−; Baxflox/+ mice14, EGF was approximately 18-fold increased in concentration in the bone marrow of irradiated Tie2Cre; Bak1−/−; Baxflox/– mice compared to irradiated Tie2Cre; Bak1−/−; Baxflox/+ mice (Fig. 1b). EGF was also expressed at threefold higher levels in Bak1−/−; Baxflox/– compared to Bak1−/−; Baxflox/+ endothelial cells (Fig. 1b). Furthermore, we did not detect EGF by ELISA of supernatants of bone marrow KSL cells from Tie2Cre; Bak1−/−; Baxflox/– mice (data not shown), suggesting that EGF is not produced in an autocrine manner by bone marrow stem and progenitor cells. Notably, Tie2Cre; Bak1−/−; Baxflox/– mice had an increased density of mouse endothelial cell antigen (MECA)+ vessels in the bone marrow compared to Tie2Cre; Bak1−/−; Baxflox/+ or C57BL/6 mice (Supplementary Fig. 1b). Therefore, the increased concentrations of EGF in the bone marrow serum of Tie2Cre; Bak1−/−; Baxflox/– mice compared to the controls may be due, in part, to the increased density of EGF-secreting bone marrow endothelial cells in Tie2Cre; Bak1−/−; Baxflox/– mice.

We next performed fluorescence-activated cell sorting (FACS) analysis to determine whether EGFR is expressed by wild-type bone marrow HSCs and progenitor cells. Less than 2% of bone marrow Lin+ and Lin− cells expressed EGFR, but 7.9% of bone marrow KSL cells and 9.2% of bone marrow signaling lymphocytic activation molecule family (SLAM)+ KSL cells, which are enriched for long-term HSCs1, expressed EGFR (Fig. 1c). EGFR surface expression increased fourfold and sixfold in bone marrow KSL and SLAM+ KSL cells, respectively, at 4 h after 300-cGy irradiation (Fig. 1c,d), indicating that EGFR expression is induced in HSCs after irradiation. After EGF treatment, EGFR phosphorylation levels were increased in bone marrow KSL cells and bone marrow SLAM+ KSL cells in vitro and in bone marrow KSL cells in vivo (Fig. 1e), demonstrating that EGFR is functional in bone marrow hematopoietic stem and progenitor cells.

We next performed gain-of-function studies to determine whether treatment of irradiated bone marrow KSL cells with EGF could support bone marrow stem and progenitor cell regeneration in vitro. When we added 20 ng ml−1 EGF to irradiated (300 cGy) bone marrow KSL cells cultured with cytokines (thrombopoietin, stem cell factor, Flt-3 ligand (TSF)), we found significant increases in the recovery of CFCs and CFU-S12 compared to cultures with cytokines alone (Fig. 1f). The addition of EGF to noncontact FL/+ endothelial cell cultures of irradiated bone marrow KSL cells also increased CFC and CFU-S12 recovery compared to culture with FL/+ endothelial cells alone, whereas the addition of EGF-specific antibody to FL/– endothelial cell cultures caused a substantial decrease in the recovery of CFC and CFU-S12 compared to FL/– endothelial cell cultures alone (Supplementary Fig. 2). These data suggest that FL/– endothelial cell–mediated regeneration of irradiated bone marrow progenitor cells is dependent on EGF. Using competitive repopulation assays, we also found that mice transplanted with the product of of irradiated, EGF-treated bone marrow CD34−KSL cells showed threefold and fivefold higher donor hematopoietic cell engraftment at 8 and 12 weeks after transplant, respectively, compared to mice transplanted with the progeny of cytokine cultures alone (Fig. 1g).

Treatment of nonirradiated bone marrow KSL cells with EGF in vitro also caused a significant expansion of bone marrow KSL cells and CFU-S12 compared to the progeny of cytokine cultures alone (Fig. 1h). Mice competitively transplanted with the progeny of bone marrow CD34− KSL cells cultured with cytokines plus EGF showed more than tenfold increased donor hematopoietic cell repopulation at 12 weeks after transplant compared to mice transplanted with the progeny of bone marrow CD34− KSL cells cultured with cytokines alone (Fig. 1h). These results suggest that EGF also promotes the maintenance of nonirradiated HSCs in culture.

EGF treatment promotes HSC regeneration in vivo

To determine whether EGF treatment can promote HSC regeneration in vivo, we measured hematopoietic reconstitution in C57BL/6 mice after 700-cGy TBI and subsequent intraperitoneal treatment with EGF or saline (control) beginning at 2 h after irradiation and then daily for 7 d (Fig. 2a). At day 7 after TBI, the EGF-treated mice had increased bone marrow cellularity compared to controls (Fig. 2b), as well as significantly increased numbers of bone marrow cells, bone marrow KSL cells, CFCs and CFU-S12 (Fig. 2b–d).

Figure 2: Systemic administration of EGF promotes HSC regeneration in vivo.
figure 2

(a) Schematic diagram of intraperitoneal treatment of mice after TBI with either EGF or normal saline (NS) for 7 d and subsequent analysis of bone marrow (BM) progenitor cell content and competitive HSC repopulation assays. (b) Representative H&E-stained femurs from EGF- and saline-treated mice. Scale bar, 250 μm. Right, bone marrow cell counts. *P = 0.003. Data are shown as the means ± s.e.m. n = 6 experiments per group. (c) Representative FACS analysis of bone marrow c-Kit+Sca-1+ cells within the Lin− gate (KSL) from nonirradiated (Nonirrad) mice and at day 7 from irradiated mice treated with either saline or EGF. The numbers shown indicate the pergentage of c-Kit+Sca-1+ cells within the Lin− population. (d) Bone marrow KSL cells, CFCs and CFU-S12 at day 7 in irradiated mice treated with either saline or EGF. *P = 0.008 for KSL cells (means ± s.e.m., n = 6 experiments per group); *P < 0.0001, *P = 0.03 for CFCs and CFU-S12 (means ± s.e.m., n = 3–5 experiments per group). (e) Percentage total donor CD45.2+ cell, myeloid (Mac-1/Gr-1), B cell (B220) and T cell (Thy1.2) engraftment in the bone marrow of CD45.1+ mice at 12 weeks after transplantation of 5 × 105 bone marrow cells from nonirradiated donor (NI), irradiated and saline-treated or irradiated and EGF-treated donor mice. *P = 0.0006 for EGF compared to saline; ^P = 0.0001 for EGF compared to NI (CD45.2+ engraftment); *P = 0.002 for EGF compared to saline (myeloid engraftment) (n = 7–10 mice per group). Horizontal lines represent mean percentage of donor cell engraftment. Percentages of donor CD45.2+ cells within the bone marrow KSL population are also shown. *P = 0.001, ^P = 0.0001 for EGF compared to the saline and NI groups, respectively (n = 7–10 mice per group, means ± s.e.m.). (f) Representative FACS plots of total donor CD45.2+ cells and CD45.2+ cell chimerism within bone marrow KSL cells at 12 weeks in mice transplanted with bone marrow from NI mice or irradiated mice treated with either saline or EGF. The numbers shown indicate the percentage of each annotated cell population within the total bone marrow. (g) Percentage donor CD45.2+ cell, myeloid, B cell and T cell engraftment at 12 weeks in the bone marrow of secondary transplant recipient mice (CD45.1+) receiving cells from mice in the NI donor group, the irradiated and saline-treated donor group and the irradiated and EGF-treated donor group. *P = 0.03, *P = 0.04, *P = 0.02, *P = 0.001 for differences in total CD45.2+ cell engraftment, myeloid cell, B cell and T cell engraftment, respectively, between the EGF- and saline-treated donor groups (n = 5–8 mice per group, means ± s.e.m.). The mean percentages of donor CD45.2+ cells within the bone marrow KSL population are also shown. *P = 0.009 for the EGF- compared to saline-treated donor group (n = 5–8 mice per group, means ± s.e.m.). (h) Representative FACS plots of CD45.2+ cell chimerism within bone marrow KSL cells in secondary transplanted mice receiving cells from mice in the NI, saline- and EGF-treated donor groups at 12 weeks after transplantation. The numbers in the bottom row indicate the percentage of donor CD45.2+ cells within the bone marrow c-Kit+Sca-1+Lin− cell population. A Mann-Whitney test was used for all statistical analyses.

Full size image

To determine whether treatment with EGF can rescue the functional bone marrow HSC pool in irradiated mice, we competitively transplanted lethally irradiated (CD45.1+) recipient mice with bone marrow cells from nonirradiated donor mice or from donor mice irradiated with 700 cGy and treated with either EGF or saline (CD45.2+). At 12 weeks after transplant, mice transplanted with 5 × 105 bone marrow cells from nonirradiated donors showed high donor-cell engraftment in the bone marrow (mean, 74.9%; Fig. 2e). Conversely, mice transplanted with the same dose of bone marrow cells from irradiated and saline-treated donors demonstrated much lower donor-cell engraftment (mean, 0.4%). Notably, mice transplanted with an identical dose of bone marrow cells from irradiated and EGF-treated mice showed significantly increased donor-cell engraftment (mean, 8.5%) compared to mice transplanted with bone marrow from irradiated and saline-treated donors (Fig. 2e). This difference in total engraftment was a function primarily of increased donor myeloid cell recovery in the EGF treatment group, suggesting that EGF treatment may have specifically augmented the recovery of short-term HSCs with myeloid reconstituting potential (Fig. 2e). We also found a close correspondence between total donor-cell engraftment and donor-cell chimerism within bone marrow KSL cells in each group of mice (Fig. 2e,f)16. Donor-cell engraftment within bone marrow KSL cells in recipients of bone marrow cells from irradiated and EGF-treated mice was sixfold higher than that in recipients of bone marrow cells from irradiated and saline-treated mice but was approximately 11% of the engraftment observed in mice transplanted with nonirradiated bone marrow cells. Therefore, although EGF treatment significantly increased the recovery of bone marrow HSCs after TBI, it did not fully restore HSC function compared to nonirradiated donor mice.

We also performed competitive secondary transplants in which we transplanted bone marrow cells from primary recipient mice into lethally irradiated syngeneic mice (Fig. 2a). At 12 weeks after transplant, secondary recipient mice receiving cells from the nonirradiated donor group showed high donor-cell engraftment (mean, 53.4%; Fig. 2g). Conversely, mice receiving cells from the irradiated and saline-treated donor group had very low donor-cell engraftment (mean, 0.5%). Secondary recipients receiving cells from the irradiated and EGF-treated donor group showed significantly higher total and multilineage donor engraftment (mean, 24%; Fig. 2g) compared to mice receiving cells from the irradiated and saline-treated donor group. Secondary recipient mice receiving cells from the irradiated and EGF-treated donor group showed a corresponding increase in donor-cell chimerism within bone marrow KSL cells compared to mice receiving cells from the irradiated and saline-treated donor group (Fig. 2h). These results suggest that EGF treatment promotes the recovery of long-term HSCs in mice after TBI.

EGFR inhibition impairs HSC regeneration in vivo

To determine whether EGFR inhibition can suppress HSC regeneration in vivo, we irradiated mice with 700 cGy and then treated them with erlotinib, an EGFR antagonist, or water (control) by oral gavage from day 0 to day 14 (Fig. 3a). At day 7, both the erlotinib-treated and control mice showed depletion of bone marrow stem and progenitor cells (data not shown). At day 14, irradiated control mice showed recovery of bone marrow cellularity, CFCs and CFU-S12, whereas erlotinib-treated mice showed depletion of bone marrow CFCs and CFU-S12 (Fig. 3b,c). The erlotinib-treated mice also had a deficit in short-term and longer-term HSCs after TBI as compared to irradiated control mice as measured by competitive repopulation assay (Fig. 3d). These results suggest that EGFR inhibition impairs hematopoietic stem and progenitor cell regeneration after TBI.

Figure 3: Erlotinib treatment inhibits HSC regeneration in vivo after TBI.
figure 3

(a) Schematic diagram of TBI and erlotinib (Erlot) treatment of C57BL/6 mice with evaluation of bone marrow (BM) progenitor cells and HSC repopulating capacity at day 14. (b) Bone marrow cellularity at day 14 in the erlotinib and water treatment groups. Scale bar, 250 μm. (c) Bone marrow CFCs (per 2 × 104 cells) and bone marrow CFU-S12 in erlotinib- and water-treated mice at day 14. *P = 0.008, *P = 0.04 for CFCs and CFU-S12, respectively (means ± s.e.m., n = 3 experiments per group). (d) Percentage donor CD45.2+ cell, Mac-1/Gr-1, B220 and Thy1.2 cell engraftment in the peripheral blood of CD45.1+ mice at 12 weeks after competitive transplantation of 5 × 105 bone marrow cells from irradiated mice treated with either erlotinib or water (controls) (n = 3–4 mice per group, means ± s.e.m.). *P = 0.007 for total CD45.2+ cell engraftment; *P = 0.04, *P = 0.003 for myeloid cell and T cell differences, respectively. Right, peripheral blood donor cell engraftment over time in mice transplanted with bone marrow from irradiated donors treated with erlotinib (red line) or water (black line). *P = 0.002, *P = 0.0002, *P = 0.007 at 4, 8 and 12 weeks, respectively (n = 3–4 mice per group). (e) Schematic diagram of irradiation and treatment of Tie2Cre; Bak1−/−; Baxflox/– mice with erlotinib or water. (f) Bone marrow KSL cells and CFU-S12 content at 2 h after TBI in erlotinib-treated and water-treated mice. *P = 0.02, *P = 0.046 for bone marrow KSL and CFU-S12, respectively (n = 2–5 experiments, means ± s.e.m.). (g) The percentage of bone marrow KSL cells with EGFR phosphorylation (pEGFR) in erlotinib-treated and water-treated mice (n = 2 experiments, means ± s.e.m.). (h) Percentage donor CD45.2+ cell, Mac-1/Gr-1, B220 and Thy1.2 cell engraftment in CD45.1+ mice at 12 weeks after competitive transplant of 3 × 105 bone marrow cells from erlotinib-treated or water-treated mice. *P = 0.03, *P = 0.02 for CD45.2+ cell and T cell engraftment, respectively (means ± s.e.m., n = 4–6 mice per group). Donor CD45.2+ cell engraftment is also shown over time (right) in mice transplanted with bone marrow cells from erlotinib-treated (red line) or water-treated (black line) mice. *P = 0.009, *P < 0.0001, *P = 0.03 for differences at weeks 4, 8 and 12, respectively. A two-tailed t test was used for all statistical analyses.

Full size image

To determine whether EGFR signaling is involved in mediating the radioprotection we observed in Tie2Cre; Bak1−/−; Baxflox/– mice14, we administered erlotinib or water (control) to Tie2Cre; Bak1−/−; Baxflox/– mice beginning 3 d before 300-cGy TBI and evaluated bone marrow HSC and progenitor cell content at 2 h after TBI (Fig. 3e). Erlotinib-treated Tie2Cre; Bak1−/−; Baxflox/– mice had significantly decreased numbers of bone marrow KSL cells and CFU-S12 after TBI compared to control irradiated Tie2Cre; Bak1−/−; Baxflox/– mice (Fig. 3f). These decreases corresponded with decreased EGFR phosphorylation in bone marrow KSL cells in the erlotinib-treated Tie2Cre; Bak1−/−; Baxflox/– mice (Fig. 3g). Irradiated Tie2Cre; Bak1−/−; Baxflox/– mice showed relative protection of bone marrow HSCs with multilineage repopulating capacity after TBI (Fig. 3h). Conversely, erlotinib-treated irradiated Tie2Cre; Bak1−/−; Baxflox/– mice had a marked deficit in bone marrow HSCs capable of multilineage reconstitution in vivo (Fig. 3h). These results suggest that EGFR signaling is important for the radioprotection of bone marrow stem and progenitor cells in Tie2Cre; Bak1−/−; Baxflox/– mice.

EGFR deficiency inhibits hematopoietic recovery after TBI

Erlotinib has been shown to inhibit kinases other than EGFR, including JAK2 and Src kinases17,18. To determine whether the effects of erlotinib on HSCs were specific to EGFR or were through off-target effects, we generated VavCre; Egfrflox/flox (Egfrflox/flox) and VavCre; Egfr+/+ (Egfr+/+) mice and verified the lack of EGFR mRNA expression in bone marrow lineage-negative (Lin−) cells (Fig. 4a). We cultured bone marrow Lin− cells from Egfrflox/flox or Egfr+/+ mice in cytokine medium for 72 h and found decreased total cell growth and CFC production in the EGFR-deficient cells (Fig. 4b). Treatment of Egfrflox/flox bone marrow Lin− cells with erlotinib caused no significant effect on total cell expansion or CFC production compared to Egfrflox/flox bone marrow Lin− cells cultured with cytokines alone (Fig. 4c). Conversely, erlotinib treatment of Egfr+/+ bone marrow Lin− cells decreased total cell expansion and CFC production compared to Egfr+/+ bone marrow Lin− cells cultured with cytokines alone or compared to Egfrflox/flox bone marrow Lin− cells cultured with erlotinib. These data suggest that erlotinib acts specifically on EGFR in bone marrow progenitor cells.

Figure 4: Deficiency of EGFR inhibits hematopoietic progenitor cell regeneration.
figure 4

(a) EGFR mRNA expression in bone marrow Lin− cells from VavCre; Egfrflox/flox (Egfrflox/flox) mice and VavCre; Egfr+/+ (Egfr+/+) mice. *P = 0.008 (means ± s.e.m., n = 5 experiments per group). (b) Numbers of total cells and CFCs at 72 h of cytokine culture of bone marrow Lin− cells from VavCre; Egfrflox/flox and VavCre; Egfr+/+ mice. *P = 0.03, *P = 0.002 for total cells and CFCs, respectively (means ± s.e.m., n = 5–6 experiments per group). (c) Numbers of total cells and CFCs at 72 h of culture of bone marrow KSL cells from VavCre; Egfrflox/flox and VavCre; Egfr+/+ mice with erlotinib. *P = 0.004, *P = 0.04 for total cells and CFCs, respectively (means ± s.e.m.; n = 8 experiments per group). (d) EGFR expression in bone marrow Lin− cells from VavCre; Egfrflox/+ (Egfrflox/+) and VavCre; Egfr+/+ mice. *P = 0.008 (means ± s.e.m., n = 5 experiments per group). (e,f) Complete blood counts (n = 10–15 mice per group) (e) and bone marrow CFCs (f) in VavCre; Egfr+/+ and VavCre; Egfrflox/+ mice (n = 6 experiments per group); means ± s.e.m. (g) Numbers of bone marrow CFCs and bone marrow SLAM+ KSL cells in VavCre; Egfrflox/+ and VavCre; Egfr+/+ mice at day 7 after 500-cGy TBI. *P = 0.002 (n = 6 experiments per group), *P = 0.004 (n = 6–11 experiments per group) for CFCs and SLAM+ KSL cells, respectively; means ± s.e.m. A Mann-Whitney test was used for all statistical analyses.

Full size image

We also compared the in vivo recovery of bone marrow hematopoietic stem and progenitor cells in Egfr+/+ and VavCre; Egfrflox/+ (Egfrflox/+) mice after 500-cGy TBI. At baseline, Egfrflox/+ mice showed decreased EGFR expression in bone marrow Lin− cells relative to Egfr+/+ mice (Fig. 4d) and had no differences in complete blood counts or bone marrow CFCs compared to Egfr+/+ mice (Fig. 4e,f). However, at day 7 after TBI, Egfrflox/+ mice had fivefold decreased bone marrow CFC content and 30-fold decreased numbers of bone marrow SLAM+ KSL cells compared to Egfr+/+ mice (Fig. 4g). These data suggest that EGFR may be necessary for normal bone marrow stem and progenitor cell regeneration after TBI.

EGF induces HSC cycling after irradiation

Activation of EGFR can promote both cell proliferation and survival after injury19,20. We therefore tested whether EGF treatment modulates HSC cycle status after irradiation. We irradiated bone marrow KSL cells with 300 cGy and then treated them in vitro with cytokines alone or cytokines plus EGF. At baseline, more than 90% of the bone marrow KSL cells were in G0/G1 (Fig. 5a). At 72 h after irradiation, EGF treatment of irradiated bone marrow KSL cells caused a significant increase in the number of cells in G2/S/M and a corresponding decrease in the number of cells in G0 compared to cytokine treatment alone (Fig. 5a). EGF treatment had comparable proliferative effects on irradiated bone marrow KSL cells from Tie2Cre; Bak1−/−; Baxflox/– mice (Supplementary Fig. 3a). EGF treatment of mice for 7 d after 700-cGy TBI also significantly increased BrdU incorporation in bone marrow KSL cells compared to saline-treated irradiated mice (Fig. 5b).

Figure 5: EGF promotes HSC cycling and survival after irradiation.
figure 5

(a) Representative FACS analysis of cell-cycle status of bone marrow KSL cells at day 0 and 72 h after irradiation in cells cultured with either TSF or TSF plus EGF. G0, white bar; G1, light gray bar; G2/S/M, gray bar. *P = 0.002, ^P = 0.002 compared to TSF for G0 and G2/S/M, respectively (means ± s.e.m., n = 3–5 experiments per group). The numbers shown indicate the percentage of each annotated population within bone marrow KSL cells. (b) Representative BrdU incorporation in bone marrow KSL cells in vivo at day 7 after 700-cGy TBI and treatment with EGF or saline (left). Right, mean BrdU incorporation. *P = 0.02 (means ± s.e.m., n = 3 experiments per group, two-tailed t test). The numbers shown indicate the percentage of BrdU+ cells within the total bone marrow KSL population. (c) Left, percentage phosphorylated AKT (pAKT) in bone marrow KSL cells after 300-cGy irradiation and the culture conditions shown. *P = 0.03 for TSF compared to EGF, ^P = 0.0006 for EGF compared to EGF plus Ly294002 (Ly29) (means ± s.e.m., n = 7–8 experiments per group). Middle, CFCs at 72 h after 300-cGy irradiation in the culture conditions shown. *P = 0.002 for TSF compared to EGF, ^P = 0.002 for EGF compared to EGF plus Ly29 (means ± s.e.m., n = 6 experiments per group). Right, cell-cycle status of KSL cells at 72 h after 300-cGy irradiation in the culture conditions shown. G0, white bar; G1, light gray bar; G2/S/M, gray bar. *P < 0.0001 for G0 EGF compared to TSF, ^P = 0.004 for G2/S/M EGF compared to TSF; *P < 0.0001 for G0 EGF plus Ly29 compared to EGF, ^P = 0.0001 for G2/S/M EGF plus Ly29 compared to EGF (n = 4–5 experiments per group, means ± s.e.m., two-tailed t test). (d) Left, percentage of annexin+ bone marrow KSL cells at 72 h of culture with TSF or TSF plus EGF and after 300-cGy irradiation and the culture conditions shown. *P = 0.002 for EGF compared to TSF; *P = 0.004 for medium compared to TSF (means ± s.e.m., n = 4–8 experiments per group). Right, percentage of annexin+ CD45+MECA− cells in the bone marrow at day 7 after 700-cGy TBI and treatment with saline or EGF. *P = 0.03 (means ± s.e.m., n = 4–5 experiments per group). (e) Left, PUMA mRNA expression in bone marrow KSL cells from Trp53+/+ and Trp53−/− mice at 6 h in the conditions shown. NI, nonirradiated control. *P = 0.01 for NI compared to irradiated TSF; *P = 0.03 for EGF compared to TSF; *P = 0.03 for Trp53+/+ TSF compared to Trp53−/− TSF (n = 4–6 experiments per group, means ± s.e.m.). Middle, representative FACS analyses of PUMA protein expression in irradiated and TSF-cultured KSL cells (blue line), irradiated and TSF- and EGF-cultured KSL cells (red line) and nonirradiated bone marrow KSL cells (black line) from Trp53+/+ and Trp53−/− mice. Right, mean percentages of PUMA protein expression in KSL cells from Trp53+/+ and Trp53−/− mice in the culture conditions described above. *P < 0.001 for NI compared to irradiated TSF; *P = 0.01 for EGF compared to TSF; *P < 0.001 for Trp53+/+ TSF compared to Trp53−/− TSF (n = 3 experiments per group, means ± s.e.m., two-tailed t test). (f) Representative FACS plots of the percentage of annexin+ bone marrow KSL cells from PUMA-expressing (+PUMA) and PUMA-deficient (PUMA-def) mice at 72 h in irradiated TSF cultures (blue lines), irradiated TSF plus EGF cultures (red line) and nonirradiated TSF cultures (black line). (g) Mean percentage of annexin+ bone marrow KSL cells from PUMA-expressing and PUMA-deficient mice at 72 h in the groups shown. *P = 0.002 for NI compared to irradiated TSF (PUMA expressing); *P = 0.02 for EGF compared to TSF (PUMA expressing); *P = 0.003 for NI compared to irradiated TSF (PUMA deficient); P = 0.002 for PUMA-expressing TSF compared to PUMA-deficient TSF (n = 4–9 experiments per group, means ± s.e.m.). (h) CFCs from bone marrow KSL cells from PUMA-expressing and PUMA-deficient mice in irradiated TSF culture and irradiated TSF plus EGF culture groups. *P = 0.006 for NI compared to irradiated TSF (PUMA expressing); *P = 0.001 for EGF compared to TSF (PUMA expressing); P = 0.02 for PUMA-expressing TSF compared to PUMA-deficient TSF (n = 6 experiments per group, means ± s.e.m.). A Mann-Whitney test was used for all statistical analyses.

Full size image

EGFR can mediate cell proliferation through activation of the mitogen-activated protein kinase (MAPK) and phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)-AKT pathways21,22. In irradiated bone marrow KSL cells, EGF treatment did not alter MAPK phosphorylation (data not shown) but did increase AKT phosphorylation (Fig. 5c). This induction of AKT in response to EGF treatment corresponded with increased CFC recovery in EGF-treated bone marrow KSL cells after irradiation (Fig. 5c). Treatment of irradiated bone marrow KSL cells with Ly294002, a PI3K inhibitor, blocked EGF-mediated AKT phosphorylation and inhibited bone marrow progenitor cell recovery in response to EGF (Fig. 5c). Irradiated bone marrow KSL cells treated with EGF and Ly294002 also showed significantly decreased cell cycling compared to KSL cells treated with EGF alone (Fig. 5c). These data suggest that EGF-driven effects on HSC cycling and progenitor cell recovery are mediated, at least in part, by activation of the PI3K-AKT pathway.

EGF inhibits HSC apoptosis by repression of PUMA

We performed annexin and 7-aminoactinomycin D (7-AAD) staining to assess the effects of EGF on HSC survival after irradiation. At 72 h after 300-cGy irradiation, EGF-treated cultures contained twofold decreased numbers of annexin-positive KSL cells compared to cytokine cultures (Fig. 5d). C57BL/6 mice irradiated with 700 cGy and then treated with EGF for 7 d also contained fourfold decreased numbers of annexin-positive bone marrow hematopoietic cells compared to saline-treated controls (Fig. 5d). EGF treatment also decreased the numbers of annexin-positive cells in cultures of irradiated bone marrow KSL cells from Tie2Cre; Bak1−/−; Baxflox/– mice (Supplementary Fig. 3b). This result suggests that EGF may mediate HSC survival after irradiation through additional mechanisms, such as DNA repair, that are not directly related to inhibition of apoptosis23.

PUMA is an essential mediator of radiation-induced hematopoietic toxicity24,25. Because EGF promoted HSC survival after irradiation, we examined whether EGF causes such effects through inhibition of PUMA. PUMA mRNA expression increased significantly in bone marrow HSCs from Trp53+/+ mice after 300-cGy irradiation, whereas PUMA mRNA expression did not change in irradiated HSCs from Trp53−/− mice, indicating that PUMA induction in HSCs is p53 dependent (Fig. 5e). EGF treatment repressed radiation-induced PUMA expression in bone marrow KSL cells from Trp53+/+ mice but had no effect on those from Trp53−/− mice (Fig. 5e). PUMA protein expression in bone marrow KSL cells correspondingly increased in response to radiation in a p53-dependent manner, and EGF treatment significantly decreased PUMA protein expression in these cells after irradiation (Fig. 5e).

Bone marrow KSL cells from PUMA-deficient mice had a lower percentage of apoptotic cells and increased CFC content at 72 h after 300-cGy irradiation compared to PUMA-expressing bone marrow KSL cells (Fig. 5f–h). EGF treatment increased HSC survival and CFC regeneration in PUMA-expressing KSL cell cultures after irradiation but had no effect on HSC survival or CFC production in PUMA-deficient KSL cell cultures after irradiation (Fig. 5f–h). These results suggest that EGF-mediated inhibition of radiation-induced HSC apoptosis is dependent on inhibition of PUMA.

EGF treatment improves the survival of lethally irradiated mice

To test whether EGFR signaling affects the survival of mice after TBI, we treated C57BL/6 mice with 10 μg per g body weight erlotinib or water (control) from day –3 to day +14 after 700-cGy TBI. Fifty-three percent (8 of 15) of control irradiated mice remained alive through day +30. In contrast, none (0 of 15) of the erlotinib-treated mice survived beyond day +27 (Fig. 6). We irradiated an additional group of C57BL/6 mice with 700 cGy and treated them with either 0.5 μg per g body weight EGF or saline intravenously for 7 d beginning 2 h after TBI. Fifty-seven percent of saline-treated mice (8 of 14) survived through day +30 (Fig. 6). Conversely, 93% of EGF-treated mice (14 of 15) remained alive through day +30. These results suggest that pharmacologic modulation of EGFR signaling alters survival after TBI.

Figure 6: Pharmacological modulation of EGFR signaling alters mice survival after TBI.
figure 6

Survival curves of C57BL/6 mice that were irradiated with 700-cGy TBI followed by daily EGF (red line) or saline (blue line) treatments for 7 d. *P = 0.02 for survival in the EGF compared to saline treatment groups (14 of 15 mice alive at day 30 in the EGF treatment group; 8 of 14 mice alive at day 30 in the saline treatment group). The survival curves of C57BL/6 mice irradiated with 700-cGy TBI and treated with 10 μg per g body weight erlotinib (green line) or water (black line) from day –3 to +14 are also shown. *P = 0.003 for survival in the erlotinib compared to water treatment groups (8 of 15 mice alive at day 30 in the water treatment control group; 0 of 15 mice alive at day 30 in the erlotinib treatment group). A log-rank test was used for all statistical analyses.

Full size image

Discussion

Recent studies suggest that hematopoietic regeneration in vivo is regulated by bone marrow endothelial cells4,5,6. However, the mechanisms through which these cells regulate hematopoietic regeneration remain largely unknown. Identification of the mechanisms that govern hematopoietic regeneration could have broad implications for the treatment of patients receiving myelosuppressive chemotherapy or radiotherapy or those undergoing stem cell transplantation26. Here we demonstrate that EGF, which we identified using a screen of bone marrow serum from radioprotected mice bearing deletion of Bak and Bax in TIE2-expressing endothelial cells, mitigates radiation injury to HSCs, and systemic administration of EGF can improve the survival of irradiated mice. Of note, EGF is a mitogen for nonhematopoietic cells that are affected by radiation injury (for example, intestinal epithelium), and such effects may have contributed to the EGF-mediated improvement in mouse survival. Nonetheless, our results suggest that systemic administration of EGF could have therapeutic potential to accelerate hematopoietic recovery in patients who have received TBI, as well as in victims of acute radiation sickness27,28. In that regard, it is worth noting that EGF treatment caused a significant enhancement in myeloid reconstitution in primary transplanted mice, suggesting a specific effect on short-term HSCs with myeloid repopulating potential. Furthermore, whereas EGF treatment was associated with improved multilineage hematopoietic reconstitution in secondary transplanted mice, stronger myeloid recovery was observed, and T cell reconstitution was relatively low. This skewing toward myeloid recovery at the expense of T lymphoid recovery in recipients of bone marrow cells from irradiated donors mimics some aspects of the HSC aging phenotype and may reflect TBI-mediated DNA damage to the HSC pool29,30.

As erlotinib has recently been shown to mediate cellular effects through inhibition of targets other than EGFR17,18, we used VavCre; Egfrflox/flox mice to determine the specific role of EGFR in regulating the hematopoietic response to radiation. In vitro studies demonstrated that erlotinib treatment had no effect on EGFR-deficient hematopoietic cells in culture, whereas erlotinib treatment of Egfr+/+ hematopoietic cells significantly inhibited both cell expansion and CFC production in culture. Therefore, erlotinib acts specifically through EGFR inhibition to diminish hematopoietic progenitor cell recovery in our model. Notably, VavCre; Egfrflox/+ mice, which are heterozygous for EGFR expression, had significantly decreased bone marrow HSC and progenitor cell recovery early after TBI compared to VavCre; Egfr+/+ mice, which retain both EGFR alleles. These results suggest that EGFR signaling has an important role in regulating hematopoietic regeneration after radiation injury.

Gamma radiation causes direct and indirect DNA damage that can result in cell-cycle arrest or apoptosis of hematopoietic progenitor cells31. In the presence of radiation-induced DNA damage, cell-cycle arrest can occur through p53-dependent or -independent mechanisms32,33. Cell-cycle arrest of hematopoietic cells can be overridden by treatment with cytokines34, and cytokine-mediated induction of hematopoietic progenitor cell proliferation early after radiation exposure may have beneficial effects toward promoting short-term hematopoietic recovery and improved near-term survival10,35,36. Although the mechanism behind these effects is not clear, cytokine treatment may induce synchronous entry of hematopoietic stem and progenitor cells into late S phase, a more radioresistant phase of the cell cycle36,37. Cytokine-driven differentiation of HSCs into myeloerythroid progenitor cells could also provide short-term benefit, as systemic infusion of myeloerythroid progenitor cells alone can completely radioprotect mice after lethal-dose TBI38. In nonhematopoietic tissues, EGFR signaling can regulate cell proliferation in a context-dependent manner, producing both cell-cycle arrest and augmented proliferation22,39. Here we show that EGF treatment induces early HSC cycling after radiation exposure and that this effect is mediated through activation of the PI3K-AKT pathway. Although EGF can modulate radiation responses through additional mechanisms such as induction of DNA repair mechanisms23, our data suggest that EGF-mediated induction of HSC proliferation contributes to the early recovery of the hematopoietic progenitor pool after irradiation.

Deletion of PUMA has been shown to protect bone marrow hematopoietic stem and progenitor cells from radiation-induced death and confer a survival advantage in mice after TBI24,25. We have shown that EGF treatment represses radiation-induced expression of PUMA in HSCs. Furthermore the effects of EGF are dependent largely on repression of PUMA. This result is consistent with previous studies that have shown that cytokines such as interkeukin-3 (IL-3) can inhibit PUMA expression in hematopoietic cells and that cytokine withdrawal mediates hematopoietic cell death in a PUMA-dependent manner40,41. Of note, AKT, which is induced by EGF treatment of HSCs, was reported to suppress p53-dependent induction of PUMA and promote PUMA protein destabilization in leukemia cells42. Although our results suggest that EGF-mediated inhibition of HSC apoptosis is dependent on repression of PUMA, it remains possible that changes in the expression of other cell-cycle regulators (for example, induction of p21) in PUMA-deficient mice43 may contribute to the effects of PUMA deficiency on the regenerative effects of EGF.

EGF has been implicated in regulating cell survival and stem cell fate outside the hematopoietic system. EGF is an established mitogen and prosurvival factor for epithelial cells and endothelial cells, and mutations in EGFR can be tumorigenic44,45,46,47,48. EGF also regulates stem cell functions in nonhematopoietic tissues such as the brain and liver49,50,51. Previous studies have suggested that EGFR is not expressed in hematopoietic stem cells52,53,54, although these studies did not use multiparametric flow cytometry to isolate bone marrow HSCs as we performed here. Although EGFR was recently shown to be expressed in bone marrow c-Kit+Lin− progenitor cells55,56, EGF was not previously known to regulate HSC self renewal or regeneration. In a previous study, the addition of EGF to stromal cell cocultures was shown to inhibit hematopoietic progenitor cell growth in vitro, but these effects could be attributed to indirect effects on stromal cells57. It was also recently reported that EGFR inhibition facilitates granulocyte colony-stimulating factor (GCSF)-mediated mobilization of hematopoietic progenitor cells in mice55. In this study, no effects were demonstrated in the absence of G-CSF and no effects on HSC content, proliferation or function were described55.

We have elucidated a previously unknown function of EGF in promoting HSC regeneration after radiation-induced myelosuppression. We show that bone marrow HSCs express functional EGFR and that EGF acts directly on HSCs to increase HSC cycling and survival after irradiation. These observed effects of EGF on HSC growth are comparable to those described for fibroblast growth factor 1 (FGF1), which also activates PI3K-AKT signaling, suggesting a possible convergence of action of EGF and FGF1 on crucial signaling pathways in HSCs58,59. Translationally, these results suggest that EGF may have therapeutic potential in patients undergoing hematopoietic cell transplant who receive TBI-based conditioning before transplant. In light of a recent report of a deleterious bystander effect of TBI-induced reactive oxygen species on transplanted donor HSCs, it will be interesting to determine whether EGF administration can ameliorate such effects to augment hematopoietic reconstitution60. EGF also has potential utility for the treatment of acute radiation sickness, which can cause life threatening bone marrow failure and for which few treatments exist.

Methods

Animals.

Ten- to 12-week-old female C57BL/6 (CD45.2+) mice and B6.SJL (CD45.1+) mice were obtained from Jackson Laboratory (Bar Harbor, ME). Tie2Cre; Bak1−/−; Baxflox/– and Tie2Cre; Bak1−/−; Baxflox/+ mice were generated as previously described13. Egfrflox/flox mice61 (Mutant Mouse Regional Resource Centers, Chapel Hill, NC) were bred with VavCre mice (Jackson Laboratory) to generate VavCre; Egfrflox/– mice. In VavCre mice, floxed alleles are excised by Cre in Vav-expressing cells and their progeny62,63. To generate VavCre; Egfrflox/flox mice, we mated VavCre; Egfrflox/+ mice with Egfrflox/flox mice. Mice were genotyped for the cre allele by Transnetyx, Inc. (Cordova, TN) and for the Egfrflox/flox allele as previously described61. Egfr deletion in bone marrow cells was quantified using RT-PCR (Applied Biosystems, Carlsbad, CA). Egfrflox/+ littermate mice were used as controls in some experiments. Trp53+/– and Trp53−/− mice64 were purchased from Jackson Laboratories and genotyped through Transnetyx Inc. Bbc3+/+ (PUMA-expressing) and Bbc3−/− (PUMA-deficient) mice were purchased from Jackson Laboratory65. Trp53 and Bbc3 deletions were quantified with RT-PCR (Applied Biosystems, Carlsbad, CA). All animal studies described here were approved by the Duke University Animal Care and Use Committee. The background strain of the Trp53, Bbc3 and Egfr floxed mice was C57BL/6J.

Hematopoietic progenitor cell assays.

Bone marrow cells were collected into PBS (Cellgro, Manassas, VA) with 10% FBS (Hyclone, Logan, UT) and 1% penicillin and streptomycin (GIBCO, Grand Island, NY). Viable bone marrow cells were quantified using Trypan blue stain (Lonza, Basel, Switzerland) to exclude apoptotic and dead cells. Cells were then incubated with antibodies to c-Kit, Sca-1, a Lin-specific antibody cocktail, CD41 (1% dilution, clone eBio MW Reg 30, eBiosciences, San Jose, CA), CD48 (1% dilution, clone HM48-1, eBiosciences) and CD150 (1% dilution, clone TC15-12F12.2, BioLegend, San Diego, CA; BD) to measure KSL progenitor cells and SLAM KSL cells as previously described1. CFC and CFU-S12 assays were performed to measure functional hematopoietic stem and progenitor cell content. For CFCs, either whole bone marrow or cultured Lin−cells or KSL cells were plated onto methylcellulose (StemCell Technologies, Vancouver, BC, Canada), and colonies were scored on day 14. A total of 1 × 105 bone marrow or 2 × 105 cells were collected from donor mice and injected in the tail vein of recipient C57BL/6 mice that had been given 950-cGy TBI. At day 12 after injection, spleens from recipient mice were harvested and stained with Bouin's fixative solution (Ricca Chemical Company, Arlington, TX), and colonies were counted as previously described66. Complete blood counts were performed on a HemaVet 950 (Drew Scientific, Dallas, TX).

Generation and culture of primary bone marrow endothelial cells from Tie2Cre; Bak1−/−; Baxflox/– mice.

For isolation and generation of primary bone marrow endothelial cells from FL/– and FL/+ mice, whole bone marrow was collected from bilateral femurs and passed through a 70-μm filter. Bone marrow vessel fragments were then plated, rinsed with 10% FBS, washed in PBS and treated with 0.25% trypsin-EDTA . Bone marrow vessel explants were cultured on 10% gelatin-coated wells (Sigma-Aldrich) with EGM-2 (endothelial cell growth medium-2) (Lonza) as previously described2,3,9,10,15. Wells were washed daily for 7–10 d and primary cells were passaged when confluent.

Bone marrow KSL cells from adult C57BL/6 mice were exposed to 300 cGy in vitro and then cultured with TSF (20 ng/ml thrombopoietin, 125 ng/ml stem cell factor and 50 ng/ml Flt-3 ligand; TSF, R&D Systems, Minneapolis, MN) alone or in noncontact culture with FL/– or FL/+ endothelial cells. In some experiments, cultures were supplemented with 20 ng/ml EGF or 1 μg/ml of a blocking antibody to EGF (1% dilution, catalog number AF2028, R&D Systems, Minneapolis, MN). After 7 d in culture, cell progeny were collected and CFC and CFU-S12 assays were performed as previously described2.

Cytokine array and EGF and EGFR expression analyses.

Whole bone marrow was collected from adult, nonirradiated Tie2Cre; Bak1−/−; Baxflox/–mice and Tie2Cre; Bak1−/−; Baxflox/+ and C57BL/6 mice and at 6 h and 7 d after 750-cGy TBI. After centrifugation, bone marrow supernatants were collected into Iscove's modified Dulbecco's medium and analyzed for cytokine concentrations using the Quantibody mouse cytokine array 1000 according to manufacturer's guidelines (RayBiotech, Inc., Norcross, GA). For analyses of protein expression of EGFR by C57BL/6 bone marrow Lin+, Lin−, KSL and SLAM+ KSL cells, bone marrow Lin− cells were isolated using MACS microbeads and LS columns according to the manufacturer's specifications (Miltenyi Biotec, Auburn, CA). Bone marrow Lin− cells were stained with fluorescein isothiocyanate (FITC)–labeled EGFR antibodies (1% dilution, catalog number ab11400, Abcam, Cambridge, MA) and antibodies to SLAM and KSL markers, as noted above. For analysis of EGFR expression after radiation, C57BL/6 mice were exposed to 300-cGy or 700-cGy TBI and then euthanized at 4 h. Bone marrow Lin− cells were isolated and stained with 7-AAD and antibodies to SLAM and KSL markers, as noted above. For analysis of phosphorylation of the Tyr1173 residue of EGFR, bone marrow Lin+ and Lin− cells were isolated using MACS Lin depletion columns as described above. Bone marrow KSL and SLAM+ KSL cells were isolated using FACS. Cells were cultured for 45 min in X-Vivo medium (Lonza) alone or in medium supplemented with 20 ng/ml EGF and then fixed in 4% paraformaldehyde and permeabilized in 0.25% saponin. Cells were stained with rabbit polyclonal antibody to mouse pTyr1173 (1% dilution, catalog number ab5652, Abcam) or isotype control antibody and then incubated with a secondary goat antibody to rabbit Alexa Fluor 488 (0.5% dilution, catalog number A-11070, Life Technologies, Grand Island, NY). For analysis of phosphorylation of EGFR in vivo, mice were injected with 20 μg of EGF or saline by tail vein injection and then euthanized 10 min after injection. Bone marrow cells were stained for KSL, fixed, permeabilized and stained for pEGFR, as noted above.

HSC cycling and survival assays.

Cell-cycle analyses were performed by flow cytometric analysis modified from previous reports67,68,69. Briefly, cells were fixed and permeabilized with 0.25% saponin (Calbiochem, La Jolla, CA), 2.5% paraformaldehyde and 2% FBS in 1× PBS and then labeled with antibodies to Ki67-FITC (1% dilution, catalog number 556026, Becton Dickinson) and 7-AAD (BD). Cell proliferation was measured in C57BL/6 mice exposed to 700-cGy TBI and administered BrdU (BD) in drinking water from the day of irradiation until day +7. Bone marrow cells were labeled with antibodies to c-Kit–phycoerythrin (PE) (1% dilution, catalog number 553869, BD, Sca-1–PE–Cy7 (1% dilution, catalog number 558162, BD), Lin-allophycocyanin (APC) (1% dilution, catalog number 558074, BD) and BrdU-FITC (1% dilution, catalog number 559619, BD). Incorporation of BrdU was analyzed by flow cytometry according to the manufacturer's staining protocol (BD).

Because 300-cGy irradiation in vitro induces apoptosis in bone marrow hematopoietic progenitor cells, we irradiated 3 × 103 C57BL/6 KSL cells with 300 cGy and then placed them in culture with TSF alone, TSF with 20 ng/ml EGF or TSF, EGF and 1 μM Ly294002 (Cell Signaling Technology, Danvers, MA) for 72 h. Cell apoptosis and necrosis were analyzed by flow cytometry according to manufacturer's protocols with annexin V–FITC and 7-AAD staining (BD, San Jose, CA). For analysis of phosphorylation AKT at Ser473, bone marrow KSL cells were cultured for 15 min with TSF or TSF plus 20 ng/ml EGF or with 20 μM Ly294002. Cells were fixed and permeabilized with Fix Buffer I and Perm Buffer III (BD) and then stained with mouse antibody to pAKT (Ser473)-PE (BD) (1% dilution, catalog number 560378, BD) or isotype control antibody.

p53 and PUMA gene expression analyses were performed on bone marrow KSL cells from C57BL/6, Trp53+/+ and Trp53−/− mice after exposure to 300-cGy irradiation and culture with TSF or TSF plus 20 ng/ml EGF for 6 h, along with nonirradiated controls. Analysis of percentahe PUMA protein expression by flow cytometry was performed 12 h after 300 cGy on C57BL/6 KSL cells in culture with TSF or TSF plus 20 ng/ml EGF, along with nonirradiated controls. Cells were fixed in 4% paraformaldehyde and permeabilized in 0.25% saponin. Cells were stained with a primary antibody to PUMA (1% dilution, catalog number ab9643, Abcam) followed by a secondary goat antibody to rabbit Alexa Fluor 488 (Life Technologies). Functional studies were also performed using bone marrow KSL cells from Bbc3+/+ and Bbc3−/−mice to quantify the percentage of annexin+ cells and CFC recovery in vitro after 300-cGy irradiation. Bone marrow Lin− cells from VavCre; Egfr+/+ or VavCre; Egfrflox/flox mice were cultured with TSF, TSF plus 10 μM erlotinib or TSF after 300-cGy irradiation for 72 h and then collected for total cell counts and CFC analysis.

Competitive repopulation assays and survival studies.

Competitive repopulation assays were performed using bone marrow cells harvested at day +7 from donor C57BL/6 mice (CD45.2+) that had been irradiated with 700-cGy TBI and given daily intraperitoneal injections of 0.5 μg per g body weight EGF (R&D Systems, Minneapolis, MN) or 200 μl PBS starting at 2 h after TBI on day 0 through day +7. Competitive repopulation assays were also performed using bone marrow cells from nonirradiated C57BL/6 mice as positive controls to compare with irradiated and saline-treated and irradiated and EGF-treated donors. Competitive repopulation assays were also performed with bone marrow harvested at day +14 from donor C57BL/6 mice (CD45.2+) that had been irradiated with 700-cGy TBI and gavaged daily with 10 μg per g body weight erlotinib (Genentech, San Francisco, CA) or 150 μl water beginning on day 0 and continued through day +14. Donor bone marrow cells were injected by tail vein into recipient B6.SJL mice (CD45.1+) at a dose of 5 × 105 cells with a competing dose of host 1 × 105 bone marrow mononuclear cells (MNCs). Primary total CD45.2+ donor cell engraftment, multilineage engraftment and donor chimerism within bone marrow KSL cells were measured in recipient mice at 12 weeks after transplant. Secondary competitive repopulation assays into lethally irradiated B6.SJL mice were performed using 75% of the total whole bone marrow cells from a group of primary recipient mice, transplanted as described above, and a competing dose of host 1 × 105 bone marrow MNCs. Measurement of donor cell chimerism within the bone marrow KSL population was performed at 12 weeks after transplantation in primary and secondary transplanted mice, as previously described16.

Tie2Cre; Bak1−/−; Baxflox/– mice were gavaged daily with 10 μg per g body weight erlotinib or water starting day –3 and given 300-cGy TBI on day 0. Erlotinib administration continued until the time point of donor bone marrow cell collection and analysis. Tie2Cre; Bak1−/−; Baxflox/– bone marrow cells were injected by tail vein into recipient CD45.1+ mice at a cell dose of 3 × 105 cells with a competing dose of host 1 × 105 CD45.1+ cells. Multilineage hematopoietic reconstitution was measured in the peripheral blood of recipient mice by flow cytometry at weeks 4, 8 and 12 after transplant. For survival studies with erlotinib administration, C57BL/6 mice were exposed to 700-cGy irradiation and then given 10 μg per g body weight erlotinib or water starting on day –3 and continuing daily through day +14. Age-matched adult C57BL/6 mice were also exposed to 700-cGy TBI and then given tail vein injections with 0.5 μg per g body weight EGF or saline beginning at +2 h after TBI and then daily through day +7. We chose 700 cGy for lethality studies because this radiation dose causes 50% lethality by day +30 in adult C57BL/6 mice at Duke University Medical Center (Cs137 irradiator).

Immunohistochemical analyses.

Femurs were decalcified and embedded in optimal cutting temperature (OCT) medium (Sakura Finetek, Torrance, CA), as previously described, on day 7 or 14 after 700-cGy TBI with daily administration of EGF or erlotinib. Ten-micrometer sections were cut using the CryoJane tape system (Instrumedics Inc, Hackensack, NJ, USA). Femurs were stained with H&E or antibody to MECA-32 (1% dilution, catalog number 550563, BD) as previously described4 to assess bone marrow cellularity and the bone marrow vasculature after irradiation. Images were obtained using an Axiovert 200 microscope (Carl Zeiss Microscopy, Thornwood, NY) or a Leica SP5 confocal microscope (Leica Microsystems Inc, Buffalo Grove, IL). Adobe Photoshop software (version 9.0.2, Adobe Systems, San Jose, CA) was used to quantify a positive signal as a measure of the total number of pixels70,71.

Statistical analyses.

Data are shown as means ± s.e.m. We used the Mann-Whitney test (two-tailed nonparametric analysis) for the majority of comparisons, along with the Student's t test (two-tailed or one-tailed distribution with unequal variance). Comparisons of overall survival were performed using a log-rank test.