Inflammatory ROS promote and cooperate with the Fanconi anemia mutation for hematopoietic senescence.

The proinflammatory cytokine tumor necrosis factor α (TNFα) inhibits hematopoietic stem cell (HSC) expansion, interferes with HSC self-renewal and compromises the ability of HSC to reconstitute hematopoiesis. We have investigated mechanisms by which TNFα suppresses hematopoiesis using the genomic instability syndrome Fanconi anemia mouse model deficient for the complementation-group-C gene (Fancc). Examination of senescence makers, such as senescence-associated β-galactosidase, HP1-γ, p53 and p16INK4A shows that TNFα induces premature senescence in bone marrow HSCs and progenitor cells as well as other tissues of Fancc–/– mice. TNFα-induced senescence correlates with the accumulation of reactive oxygen species (ROS) and oxidative DNA damage. Neutralization of TNFα or deletion of the TNF receptor in Fancc–/– mice (Fancc–/–;Tnfr1–/–) prevents excessive ROS production and hematopoietic senescence. Pretreatment of TNFα-injected Fancc–/– mice with a ROS scavenger significantly reduces oxidative base damage, DNA strand breaks and senescence. Furthermore, HSCs and progenitor cells from TNFα-treated Fancc–/– mice show increased chromosomal aberrations and have an impaired oxidative DNA-damage repair. These results indicate an intimate link between inflammatory reactive oxygen species and DNA-damage-induced premature senescence in HSCs and progenitor cells, which may play an important role in aging and anemia.


Introduction
The genomic instability syndrome Fanconi anemia (FA) shares the common feature of having defective DNA damage response/repair processes with other premature aging syndromes such as Bloom (Bischof et al., 2001) and Werner (Chen et al., 2003), ataxia telangiectasia (Ito et al., 2004) and BRCA1-deficient breast cancer (Cao et al., 2003). The FA pathway has been shown to functionally interact with the proteins responsible for Bloom syndrome (BLM) (Meetei et al., 2003), ataxia telangiectasia (ATM) ] and breast cancer (BRCA1) (Seal et al., 2003). In fact, one of the FA genes, FANCD1, turned out to be the breast cancer gene BRCA2 (Howlett et al., 2002). This suggests a biological link between the FA disease and organismal aging. The FA disease model is unique in that defects are especially profound in the rapidly proliferating hematopoietic system. Thus far, aging studies have not yet focused on hematopoietic stem cells (HSCs) in FA or these other premature aging syndromes. Nevertheless, HSCs are capable of self-renewing, have a high risk of accumulating mutations and, thus, may require fewer events to become senescent or transformed (Pelicci, 2004). In addition, HSC aging can be assessed in a quantitative fashion with well-established assays. Since senescence can cause stem cell depletion, which can conceivably lead to organismal aging, such studies may yield important insights to the mechanisms of aging.
Functional failure of HSCs can result in the decrease in number and function of HSCs and progenitor cells leading to the development of anemia, neutropenia and thrombocytopenia, and has been implicated in hematologic diseases such as bone marrow (BM)-failure diseases such as FA, myelodysplastic syndromes (MDS) and aplastic anemia (Young, 2002;Bagby, Jr, 2003;Chen, 2005). Indeed, it has been shown that FA HSCs and progenitor cells have high rates of stress-induced apoptosis and reduced repopulating ability (Haneline et al., 1998;Haneline et al., 1999;Haneline et al., 2003). The FA proteins are thus believed to play important roles in the maintenance of hematopoiesis. Consistent with the observations that cells derived from FA patients are intolerant of oxidative stress, it has been reported that FA proteins, particularly the complementation group C (FANCC) protein, play a crucial role in oxidative-stress signaling in a variety of cell types including hematopoietic cells (Kruyt et al., 1998;Cumming et al., 2001;Hadjur et al., 2001;Futaki et al., 2002;Park et al., 2004;Saadatzadeh et al., 2004;Pagano et al., 2005). More recently, cytokine hypersensitivity of FA hematopoietic cells to apoptotic cues has also been proposed as a major factor in the pathogenesis of BM failure in three FA mouse models The proinflammatory cytokine tumor necrosis factor ␣ (TNF␣) inhibits hematopoietic stem cell (HSC) expansion, interferes with HSC self-renewal and compromises the ability of HSC to reconstitute hematopoiesis. We have investigated mechanisms by which TNF␣ suppresses hematopoiesis using the genomic instability syndrome Fanconi anemia mouse model deficient for the complementation-group-C gene (Fancc). Examination of senescence makers, such as senescence-associated ␤galactosidase, HP1-␥, p53 and p16 INK4A shows that TNF␣ induces premature senescence in bone marrow HSCs and progenitor cells as well as other tissues of Fancc -/mice. TNF␣-induced senescence correlates with the accumulation of reactive oxygen species (ROS) and oxidative DNA damage. Neutralization of TNF␣ or deletion of the TNF receptor in Fancc -/mice (Fancc -/-;Tnfr1 -/-) prevents excessive ROS production and hematopoietic senescence. Pretreatment of TNF␣-injected Fancc -/mice with a ROS scavenger significantly reduces oxidative base damage, DNA strand breaks and senescence. Furthermore, HSCs and progenitor cells from TNF␣-treated Fancc -/mice show increased chromosomal aberrations and have an impaired oxidative DNA-damage repair. These results indicate an intimate link between inflammatory reactive oxygen species and DNA-damage-induced premature senescence in HSCs and progenitor cells, which may play an important role in aging and anemia.
It is believed that cellular and organismal senescence occurs as a result of chronic exposure to extrinsic environmental factors, primarily of oxidative stress (Sohal and Weindruch, 1996). The degree of oxidative damage has been found to increase exponentially with senescence and aging in a variety of cells, and tissues of different species (Chen et al., 1995). Endogenously formed reactive oxygen species (ROS) continuously damage cellular constituents including DNA. These challenges, coupled with exogenous exposure to agents that generate ROS, are both associated with normal aging processes and linked to cardiovascular disease, agerelated anemia and cancer (Maccio et al., 2005;Sablina et al., 2005;Wallace, 2005). In the aged and certain disease states, ROS produced by proinflammatory cytokinessuch as tumor necrosis factor-␣ (TNF␣) -at inflammatory sites often cause DNA damage (Goossens et al., 1999;Suematsu et al., 2003;Wheelhouse et al., 2003). However, the mechanism through which ROS mediate their effects on the aging process is unclear.
Overproduction of the proinflammatory cytokine TNF␣ has been implicated in pathological conditions related to anemia and aging that is characteristic of HSC function failure (Oster et al., 1989;Kitagawa et al., 1997;Young, 2000;Dufour et al., 2003). Here, we have employed the disease model of genomic instability syndrome FA to investigate the cellular mechanism by which TNF␣-induced inflammatory ROS influence hematopoiesis. We demonstrate that ROS induce premature senescence in BM cells enriched for HSCs and progenitor cells through prolonged oxidative DNA-damage response and increased genomic instability.

TNF␣ suppresses HSC/progenitor cell activity
To examine the effect of TNF␣ on HSC/progenitor cell activity, we injected intraperitoneally (i.p.) wild-type (WT) or Fancc -/mice with mouse recombinant TNF␣ in PBS or PBS alone. Seven days later, mice were sacrificed, and peripheral blood (PB) and BM were analyzed for hematopoietic functions.
TNF␣-treated Fancc -/mice exhibited mild cytopenia, as evidenced by a decrease in red cell counts, hemoglobin and hematocrit values (see supplementary material Fig. S1). Consistent with this, BM analysis of TNF␣-treated Fancc -/mice revealed a decrease in the number of cells compared with PBS-injected Fancc -/mice (see supplementary material Table  S1). In another set of experiments we analyzed hematopoietic recovery in mice injected with TNF␣. We found that Fancc -/mice recovered slowly from TNF␣-induced hemo-suppression compared with WT mice (see supplementary material Fig. S2). Furthermore, Fancc -/mice showed reduced rates of multilineage recovery (see supplementary material Table S2).
The frequencies of BM HSC/progenitor [lineage-negative, Sca-1-positive, c-Kit-positive (Lin --Sca-1 + -c-Kit + )] cells (LSK cells) were reduced by more than twofold in TNF␣-treated Fancc -/mice compared with PBS controls (Fig. 1A). We thus asked whether TNF␣ suppressed clonogenic progenitor activity. As shown in Fig. 1B, the total number of colonies formed by LSK cells from TNF␣-treated Fancc -/mice was  Wild-type (WT) or Fancc -/mice were injected intraperitoneally (i.p.) with mouse recombinant TNF␣ in PBS at a dose of 0.1 mg/kg per day for two consecutive days. The mice were then sacrificed 24 hours later and BM mononuclear cells were prepared and stained with antibodies against lineage markers (Lin) and Sca-1-PE and c-Kit-PE-Cy7. Cells were then analyzed by flow cytometry to obtain fractions representing HSCs. The frequencies of LSK cells as a percentage of total BM mononuclear cells are indicated. (B) The effect of TNF␣ on colony-forming activity of WT and Fancc -/-BM progenitor cells was evaluated using LSK cells isolated from control (PBS-injected) or TNF␣-injected mice. Data shown represent the mean ± s.d. of colony numbers from three independent experiments; *P<0.05. (C) TNF␣ inhibited BMprogenitor proliferation. LSK cells isolated from TNF␣-injected mice were analyzed for series-plating efficiency of hematopoietic progenitor cells. Data represents the mean ± s.d. of three experiments. (D) TNF␣ inhibited HSC renewal. 2ϫ10 6 BM cells isolated from control (PBS-injected) or TNF␣injected mice (CD45.2 + ) were transplanted together with 1ϫ10 6 competitor cells from B6.BoyJ mice (CD45.1 + ) into lethally irradiated recipient (B6.BoyJ) mice and long-term engraftment was evaluated 16 weeks after transplantation. Shown are representative flow cytometric data of two independent experiments (n=6). Numbers in the corners indicate percent of events in that quadrant. more than threefold lower than that of WT mice. We also observed significantly decreased series-plating efficiency with BM LSK cells of TNF␣-treated Fancc -/mice compared with WT cells (Fig. 1C). To evaluate the in vivo effect of TNF␣ on HSC/progenitor cell activity, we performed BM-stem-cell transplantation. Long-term engraftment analysis shows that Fancc -/-BM HSCs were decreased in their ability to repopulate compared with WT cells (Fig. 1D), which is consistent with previous reports (Haneline et al., 1998;Haneline et al., 1999). Significantly, BM HSCs from TNF␣-Journal of Cell Science 120 (9) treated Fancc -/mice constituted less than 10% of the peripheral blood cells at 16 weeks after transplantation compared with more than 40% reconstitution by those from TNF␣-treated WT mice (Fig. 1D), indicating that TNF␣ impaired long-term hematopoietic reconstitution of Fancc -/-HSCs. Collectively, these results suggest that TNF␣ suppresses hematopoietic functions by inhibiting the self-renewal or proliferative potential of HSCs and progenitor cells.
TNF␣ induces premature senescence in HSC/progenitor cells Given that TNF␣ suppressed the proliferative potential of HSCs and progenitor cells, we wondered whether TNF␣ affected differentiation or apoptosis of these primitive cells. Fig. 2A shows that BM HSCs from TNF␣-injected WT or Fancc -/donors had the ability of multilineage reconstitution, suggesting that TNF␣ does not affect the differentiation of long-term repopulating HSCs. Whereas TNF␣ induced apoptosis in BM LSK cells of treated mice, no statistically significant difference was observed between WT and Fancc -/mice (Fig. 2B). Thus, apoptosis was not the major consequence of TNF␣-mediated inhibitory effect on the self-renewal or proliferative potential of Fancc -/-HSCs and progenitor cells. However, BM LSK cells from TNF␣-injected Fancc -/mice clearly exhibited a prolonged G2-M accumulation with decreased cells in S phase (Fig. 2C), suggesting that the G2 checkpoint is compromised in these BM HSC/progenitor cells. This result reminisces of DNA-damageinduced G2-M arrest, a hallmark of FA (Bagby, Jr, 2003;Kennedy and D'Andrea, 2005).
It has been suggested that HSC senescence plays an important role in the pathophysiology of hematologic diseases, such as aplastic anemia and MDS (Maciejewski and Risitano, 2003;Zhang et al., 2005a;Zhang et al., 2005b). To determine whether HSC senescence is a defining feature of TNF␣mediated hematopoietic suppression, we examined two established senescence markers: senescence-associated ␤galactosidase (SA-␤-gal) (Dimri et al., 1995) and senescenceassociated heterochromatin foci (Narita et al., 2003). We found that approximately 10% of LSK cells from TNF␣-injected WT mice and more than 20% of those from TNF␣-injected Fancc -/- Fig. 2. Analysis of differentiation, apoptosis and cell cycle of HSCs and progenitor cells from TNF␣-treated mice. (A) TNF␣ did not affect the differentiation of long-term repopulating HSCs. Donorderived (CD45.2 + ) leukocytes in peripheral blood of recipient mice transplanted with the indicated BM cells were stained with antibodies that recognize Gr-1 and Mac-1 or B220 and CD3 or Ter119. Data are expressed as mean ± s.d. of three independent experiments, each with three recipients (nine mice per group). (B) TNF␣ induced apoptosis in BM LSK cells of mice. BM cells from PBS-or TNF␣-treated WT and Fancc -/mice were stained with lineage maker antibodies together with Sca-1 and c-Kit antibodies, and then with annexin V. Percentages of apoptosis in the LSK population were analyzed by flow cytometry. Numbers in the quadrants indicate percent of cells labeled for 7-AAD low annexin V + or 7-AAD high annexin V + . (C) BM LSK cells from TNF␣-injected Fancc -/mice show increased G2-M arrest. BM cells from PBS-or TNF␣-treated WT and Fancc -/mice were gated for LSK cells and stained with propidium iodide (PI) followed by analysis for cell cycle distribution. Shown are representative flow cytometric presentations of three independent experiments. Numbers in plots indicate percent of cells in G2-M phases.
animals stained positive for SA-␤-gal activity (Fig. 3A). LSK cells from TNF␣-treated Fancc -/mice also stained strongly positive for HP1-␥, indicating the formation of senescenceassociated heterochromatin foci ( Fig. 3B) (Narita et al., 2003). In addition, analysis of these two senescence markers in BM sections from TNF␣-injected Fancc -/mice revealed abundant positive staining in those cells (Fig. 3C) accompanied with a weak proliferation index, as identified by the proliferation marker Ki-67 (Fig. 3D). We also examined two established effectors of stress-induced senescence, p53 and p16 INK4A (Randle et al., 2001;Serrano and Blasco, 2001). We observed that the BM LSK cells of TNF␣-injected Fancc -/mice stained WT or Fancc -/mice were injected i.p. with two doses of TNF␣ (100 g/kg per day) for 2 consecutive days. The mice were then sacrificed 24 hours later and BM LSK cells were isolated and stained for SA-␤-gal. The bar graph shows the percentages of the cells stained positive for SA-␤-gal; cells were quantified by counting >100 cells in random fields on a slide for each of three independent experiments. The data represent the mean ± s.d. of three independent experiments; *P<.05. (B) Senescence HP1-␥ staining. BM LSK cells isolated from PBS-or TNF␣-treated WT and Fancc -/mice were stained for HP1-␥. DNA was then labeled with DAPI. (C) Frozen (SA-␤-gal; magnification, 20ϫ) or paraffin-embedded (HP1-␥; magnification, 40ϫ) BM sections of PBS-or TNF␣-treated WT and Fancc -/mice were subjected to SA-␤-gal (left) or HP1-␥ (right) staining. (D) Paraffin-embedded spleen sections of PBS-or TNF␣-treated WT and Fancc -/mice, stained with antibody against the proliferation marker Ki-67 (magnification, 20ϫ). (E) BM LSK cells of TNF␣-treated Fancc -/mice showed strong immunostaining for p53 and p16 Ink4a . The BM LSK cells isolated from PBS-or TNF␣-treated WT and Fancc -/mice were stained with the antibodies against p53 and p16 Ink4a and then counterstained with DAPI. The bar graph shows the percentages of the cells stained positive for p53 and p16 Ink4a ; cells were quantified by counting >100 cells in random fields on a slide for each of three independent experiments; *P<0.05. positive for p53 and p16 INK4A , whereas those from untreated mice stained essentially negative (Fig. 3E). These results suggest that senescence is a defining feature of TNF␣mediated hematopoietic suppression in Fancc -/mice.

Premature senescence of HSC/progenitor cells is linked to TNF␣-induced ROS production
Telomere shorting has been proposed to be a cause of HSC senescence (Lansdorp, 2005). The telomere length of BM cells Journal of Cell Science 120 (9) from TNF␣-injected mice was comparable to that of untreated mice, regardless of genotype (see supplementary material  The ROS scavenger NAC mitigated TNF␣-induced senescence in HSCs and progenitor cells. WT and Fancc -/mice were injected i.p. with TNF␣ (100 g/kg per day) for 2 consecutive days. NAC (1 mg/mouse per day) was administered 30 minutes before and after each TNF␣ injection. The isolated BM LSK cells (top and middle groups of panels) or paraffin-embedded BM sections (bottom group of panels; magnification, 40ϫ) were stained with the antibody against the senescence marker HP1-␥. The LSK cell slides were also counterstained with DAPI. (B) BM LSK cells from WT and Fancc -/mice were cultured in the absence or presence of H 2 O 2 (100 M) for 45 minutes, washed and further cultured in fresh medium for 24 hours. Cells were then stained with the antibody against the senescence marker HP1-␥. Shown is quantification of HP1-␥-positive cells by scoring >100 cells in random fields on a slide for each of three independent experiments; *P<0.05. (C) BM LSK cells from WT and Fancc -/mice were cultured in the absence or presence of H 2 O 2 (100 M) for 45 minutes, washed and further cultured in fresh medium for 24 hours. 1000 cells were mixed with 1ϫ10 6 competitor cells were transplanted into lethally irradiated recipient mice and long-term engraftment was evaluated 16 weeks after transplantation. Data are expressed as mean ± s.d. of two independent experiments, each with six recipients (12 mice per group). P<0.05 between untreated and treated Fancc -/samples. (D) ROS production. BM cells from PBS-or TNF␣-treated WT and Fancc -/mice were labeled with CM-H2DCFDA followed by flow cytometry. (E) ROS production was mediated by TNF␣. WT, Fancc -/mice or their littermates lacking the Tnfr1 gene were injected with PBS or TNF␣ at 0.1 mg/kg per day for 2 consecutive days. 24 hours after TNF␣ injection, BM cells from individual animals were isolated and labeled with CM-H2DCFDA followed by flow cytometry. Data represent the mean ± s.d. of two independent experiments, each with six mice (total 12 mice per group). (F) WT, Fancc -/mice or their littermates lacking the Tnfr1 gene were injected with TNF␣ (0.1 mg/kg per day) for 2 consecutive days. TNF␣-treated mice were injected with 20 g of a TNF␣neutralizing antibody 30 minutes after each TNF␣ injection. At 24 hours after TNF␣ injection, frozen spleen sections were prepared and subjected to SA-␤-gal staining.
implicated as a mechanism of TNF␣-induced cell death (Sakon et al., 2003;Ventura et al., 2004). To address whether ROS might be relevant for TNF␣-induced hematopoietic senescence, we pretreated the TNF␣-injected mice with the ROS scavenger N-acetyl-L-cysteine (NAC). Remarkably, NAC almost completely abrogated the induction of senescenceassociated heterochromatin foci (HP1-␥) by TNF␣ in both BM LSK cells and BM sections of Fancc -/mice (Fig. 4A), indicating that TNF␣-induced senescence was mediated by oxidative stress. Examination of the effect of exogenous peroxide (H 2 O 2 ; a potent ROS producer) on BM LSK cells demonstrated that, indeed, Fancc -/-HSC/progenitor cells were vulnerable to oxidative-stress-induced senescence (Fig. 4B) and inhibition of hematopoietic reconstitution (Fig. 4C). Analysis of ROS production by flow cytometry revealed that TNF␣ did not cause substantial ROS production in the BM of WT mice (Fig. 4D). By contrast, there was a significant ROS accumulation in the Fancc -/-BM. In vitro culture of isolated BM LSK cells in the presence of TNF␣ further demonstrated increased ROS production in Fancc -/-HSC/progenitor cells (see supplementary material Fig. S4).
To ascertain that it was TNF␣ that induced ROS production in hematopoietic cells, we took two approaches: pretreatment of mice with neutralizing anti-TNF␣ antibody and inactivation of TNF␣ signaling in Fancc -/mice. For the latter, Fancc -/mice were crossed with mice carrying a null mutation in the type 1 TNF receptor (Tnfr1 -/mice). Nearly complete prevention of TNF␣-induced ROS accumulation (Fig. 4E) and senescence induction (Fig. 4F) was obtained with TNF␣injected mice deficient for Tnfr1. Whereas treatment of mice with anti-TNF␣ antibody 30 minutes after each TNF␣ injection did not completely neutralize serum TNF␣ [mean ELISA values for anti-TNF␣ and vehicle-injected animals were 78.6±12.4 pg/ml and 245±21.5 pg/ml, respectively, in TNF␣treated WT mice (n=6) compared with 63.9±17.2 pg/ml and 227±23.8 pg/ml, respectively, in TNF␣-treated Fancc -/mice (n=6); the basal level of serum TNF␣ in either WT or Fancc -/mice was approximately 20 pg/ml], administration of the anti-TNF monoclonal antibody resulted in significant reduction of senescence in the spleen (Fig. 4F). In vitro assays further demonstrated that Fancc -/-;Tnfr1 -/-BM HSC/progenitor cells were resistant to TNF␣-induced ROS production and senescence (see supplementary material Fig. S5). Thus, oxidative stress is responsible for TNF␣-induced hematopoietic senescence in Fancc-deficient mice.
To specifically assess TNF␣-induced oxidative damage in hematopoietic organs, we immunostained for 4-hydroxy-2nonenal (HNE), an established marker of ROS-induced tissue damage and an aldehyde product of polyunsaturated fatty acid oxidation (Cauwels et al., 2003). HNE staining was evident in the spleen of WT mice treated with TNF␣ (see supplementary material Fig. S6), in which we did not observe significant induction of senescence (Figs 3, 4). We found further increase in HNE immunoreactivity in the treated Fancc -/mice (see supplementary material Fig. S6). This increase in HNE immunoreactivity was largely prevented by anti-TNF␣ antibody or NAC treatment (see supplementary material Fig.  S6). Thus, oxidative organ damage induced by TNF␣, as detected by fatty acid oxidation, appears to be proportional to the level of ROS production but not the extent of induction of senescence.
TNF␣ induced senescence of HSC/progenitor cells as a result of accumulation of oxidative DNA damage and increased genomic instability Oxidative stress can induce DNA damage that causes growth arrest and cellular senescence (Ames et al., 1993;Chen, 2000). To determine whether TNF␣-generated ROS induced DNA damage in HSC/progenitor cells that appeared to have (A) The ROS scavenger NAC or deletion of Tnfr1 gene rescued progenitor growth. WT, Fancc -/mice or their littermates lacking Tnfr1 were injected with PBS or TNF␣ (0.1 mg/kg per day) for 2 consecutive days. Mice were injected with NAC (1 mg/mouse per day) 30 minutes before and after TNF␣ injection. Twenty-four hours later, BM cells were isolated and subjected to clonogenic assay. Data shown represent the mean ± s.d. of the total number of colonies from three independent experiments; *P<0.05. (B) Anti-oxidant NAC or deletion of the Tnfr1 gene restored HSC self-renewal ability. 2ϫ10 6 BM mononuclear cells isolated from the mice described in A were transplanted together with 1ϫ10 6 competitor cells from B6.BoyJ mice (CD45.1 + ) into lethally irradiated recipient mice; long-term engraftment was evaluated 16 weeks after transplantation. Data represent the mean ± s.d. of three independent experiments with three recipients per group for each experiment. *P<0.05. undergone ROS-dependent senescence (Fig. 4), we employed the comet assay (Fairbairn et al., 1995) using a Fpg-FLARE (fragment length analysis using repair enzymes) assay kit. This assay measures specifically oxidative DNA damage including single-and double-strand DNA breaks. There was significant accumulation of DNA damage in BM LSK cells freshly isolated from TNF␣-treated WT mice and, to a much greater degree, in TNF␣-treated Fancc -/mice (Fig. 6A). Moreover, there was a significant difference between WT and Fancc -/cells in terms of kinetics of DNA damage repair, as measured by the remaining amounts of DNA strand breaks over a period of 8 hours of TNF␣ treatment (see supplementary material Fig.  S7). The levels of TNF␣-induced DNA strand breaks remained high in Fancc -/cells at each time point compared with those in WT cells. A similar increase in 8-oxo-deoxyguanosin (8-oxodG), an established marker of oxidative DNA damage, was also demonstrated in BM LSK cells from TNF␣-injected Fancc -/mice compared with treated WT mice (Fig. 6B). Treatment of TNF␣-injected Fancc -/mice with neutralizing anti-TNF␣ antibody or NAC reduced the accumulation of both DNA strand breaks and 8-oxodG (Fig. 6A,B). Thus, TNF␣induced senescence in HSC/progenitor cells appears to involve oxidative DNA damage.
Genomic instability is the cellular hallmark of FA (Bagby, Jr, 2003;Kennedy and D'Andrea, 2005), and unrepaired TNF␣-induced oxidative DNA damage may increase genomic instability leading to premature senescence of HSC/progenitor cells. To test this possibility, we conducted karyotype analysis of BM Lin -(enriched for HSC/progenitor) cells from TNF␣injected WT and Fancc -/mice. Chromosomal aberrations were rare in cells from treated WT mice, with fewer than 10% of the cells showing one aberration and almost no cells containing two or more aberrations (Table 1). By contrast, BM HSC/progenitor cells from TNF␣-injected Fancc -/mice showed significant increase in chromosomal aberrations (56% of the cells contained at least one aberration, 35% had two or more aberrations, and 23% had three or more aberrations; Table 1). FISH analysis of the chromosomal damages indicated that the TNF␣-induced aberrations in Fancc -/cells consisted mostly of chromatid breaks, gaps, chromosomal breaks, dicentric chromosomes, double minutes, chromosome fragments and fusions ( Fig. 6C; Table 2), suggesting that the efficiency and accuracy of DNA strand break/repair may have been compromised in these cells.
The persistent high levels of oxidative DNA damage observed in HSC/progenitor cells from TNF␣-injected Fancc -/mice suggest that a deficiency in the FA pathway renders chromosomal DNA susceptible to ROS attack, thereby increasing oxidative DNA damage. Alternatively, a defect in the FA function might compromise the damage response/repair process. To distinguish between these possibilities, we treated BM cells from WT and Fancc -/mice with H 2 O 2 , and conducted a time-course study to assess DNA-repair kinetics by examining the levels of p53 Ser20 and ␥H2AX. Compared with WT cells, Fancc -/cells consistently showed a significant delay in the kinetics of DNA-damage response/repair as evidenced by the much slower clearance of the markers of oxidative DNA damage (p53 Ser20 ) and DNA strand break WT or Fancc -/mice were injected i.p. with two doses of TNF␣ (100 g/kg per day) for 2 consecutive days. NAC (1 mg/mouse per day) was administered 30 minutes before and after each TNF␣ injection. The mice were then sacrificed 24 hours later and BM cells from individual mice were analyzed for DNA strand breaks in a the comet assay. Larger tails represents higher levels of DNA damage. For each treatment, at least 100 cells were scored from random sampling. Data are expressed as the mean ± s.d. of two independent experiments, each with three mice (six mice per group). (B) WT or Fancc -/mice were injected i.p. with two doses of TNF␣ (100 g/kg per day) for 2 consecutive days. NAC (1 mg/mouse/day) was administered 30 minutes before and after each TNF␣ injection. The mice were then sacrificed 24 hours later and BM LSK cells were isolated and stained for the oxidative DNA damage marker 8-oxo-deoxyguanosin (8-oxodG). The bar graph shows the percentages of the cells stained positive for 8-oxodG and quantified by counting >100 cells in random fields on a slide for each of two experiments with total 6 mice. (C) Examples of metaphase chromosomes prepared from TNF␣-treated WT and Fancc -/-BM cells. Arrows indicate aberrant chromosomes. (D) WT or Fancc -/mice were injected i.p. with two doses of TNF␣ (100 g/kg per day) for 2 consecutive days. NAC (1 mg/mouse/day) was administered 30 minutes before and after each TNF␣ injection. Mice were sacrificed 24 hours later and BM LSK cells were then isolated and stained for p53 Ser20 and ␥H2AX. Bar graphs show the percentages of the cells stained positive for p53 Ser20 and ␥H2AX. Cells were quantified by counting >100 cells in random fields on a slide for each of two experiments with total six mice. *P<0.05. (E) Kinetics of DNA repair of oxidative DNA damage and DNA strand breaks. BM cells from WT and Fancc -/mice were treated with or without H 2 O 2 (100 M) for the indicated time periods, and protein extracts were prepared and analyzed by immunoblotting with antibody against phosphorylated p53 Ser20 (p53 Ser20 ) and anti-␥H2AX and anti-actin antibodies. Extracts were also prepared from cells 2 hours (2-) after H 2 O 2 withdrawal after the cells had been treated with H 2 O 2 for 4 hours. (␥H2AX) (Fig. 6E). This suggests that Fancc -/cells accumulated high levels of oxidative DNA damage due to impairment of DNA-damage response/repair rather than to an increase in the susceptibility of their DNA to oxidative damage.

Discussion
We used the disease model of the genomic instability syndrome FA to investigate the role of TNF␣-mediated inflammation in HSC senescence. We demonstrated that TNF␣ induced premature senescence in BM HSCs and progenitor cells as well as other tissues of mice deleted for the Fancc gene, which is required for the maintenance of genetic integrity (Bagby, Jr, 2003;Kennedy and D'Andrea, 2005). Importantly, TNF␣induced senescence of HSCs and progenitor cells correlated with the accumulation of inflammatory ROS, and oxidative DNA damage. ROS are involved in the induction of a senescent phenotype characterized by irreversible growth arrest (Chen et al., 1995;Sohal and Weindruch, 1996). The question naturally arises as to whether the inhibition of ROS can rescue HSCs and progenitor cells from TNF␣-mediated senescence. Our findings revealed that pretreatment of TNF␣-injected Fanccdeficient mice with antioxidant abrogated the exacerbated inflammatory phenotype, and prevented oxidative DNA damage and hematopoietic senescence. It is clear that the persistent inflammatory response, as reflected by prolonged production of inflammatory molecules, is primarily responsible for the generation of ROS and senescence, because mice deficient for both FA function and TNF␣ signaling (Fancc -/-;Tnfr1 -/-) were resistant to TNF␣-induced ROS production and senescence. Moreover, HSCs and progenitor cells from TNF␣-treated Fancc-deficient mice showed increased genomic instability and an impaired repair of oxidative DNA damage. Our study thus provides a link between inflammatory ROS and HSC senescence. Compelling evidence indicates that ROS are widely implicated in the inflammatory process (Lavrovsky et al., 2000). However, the mechanism by which ROS contribute to the proinflammatory states of the aging process is not well defined. Our results demonstrated that TNF␣-induced ROS contributed to the increase in oxidative DNA damage in Fanccdeficient HSCs and progenitor cells. Inflammatory ROS have been associated with the initiation or aggravation of diverse pathological states including aging and cancers (Cutler, 2005;Wallace, 2005;Yoshida et al., 2005). We believe that in cells with compromised DNA repair capacity, the ability of inflammatory ROS to damage DNA is the mechanism through which ROS mediate their effects on the inflammatory and, thus, aging process. Indeed, studies have shown that the production of ROS by TNF␣ at inflammatory sites causes DNA damage (Goossens et al., 1999;Suematsu et al., 2003;Wheelhouse et al., 2003). The role of ROS in TNF␣-induced inflammation and HSC senescence was validated by the use of the ROS scavenger NAC, which showed that inhibition of ROS accumulation reduced oxidative DNA damage and cellular senescence. In addition, mice treated with TNF␣ accumulated high levels of ROS, and administration of NAC significantly reduced the amounts of secreted pro-inflammatory cytokines (Y.Q. and Q.P., unpublished results), indicating that the inflammatory response is, at least in part, mediated by ROS. Thus, these results suggest that ROS serves as a link between inflammation and senescence.
Journal of Cell Science 120 (9) By investigating four well-established senescence makers (SA-␤-gal, HP1-␥, p53 and p16 INK4A ), we provided evidence that TNF␣ induced premature senescence in BM HSCs and progenitor cells and also other tissues of Fancc -/mice. Recently, another senescence marker protein, SMP30, has been shown to play an important role in senescence and aging, and SPM-deficiency renders mice highly susceptible to oxidative stress (Sato et al., 2006). It would be interesting to examine the effect of TNF␣ on the expression pattern of this new senescence marker in HSCs and progenitor cells of Fancc -/mice.
The persistent DNA damage response and increased genomic instability in senescent HSCs, and progenitor cells are two important features of our study. BM progenitor cells from TNF␣-treated Fancc -/mice contained high levels of DNA strand breakage and oxidative DNA damage. Moreover, studies of repair kinetics revealed much slower clearance of the oxidative DNA damage and DNA strand break markers in Fancc -/-HSCs and progenitor cells than in WT cells. Consistent with this, DNA damage response was persisted in these Fancc -/cells. This suggests that premature senescence observed in Fancc -/-HSCs and progenitor cells may result from the prolonged activation of the oxidative DNA damage and DNA double-strand-break checkpoints. However, this prolonged checkpoint activation did not facilitate damage repair. Instead, we found dramatically increased genomic instability in Fancc -/cells. Thus, we conclude that it is unrepaired DNA damage that results in the persistent DNA damage response in these cells.
Our study raises an important question: can inflammatory response be a link between aging and HSC function? It is long known that general inflammatory stress tends to increase with age (Chung et al., 2001). Under inflammatory conditions, HSCs must be able to produce a large number of leukocytes, which are then activated to fight against invaders or stressing agents. This, ultimately, may lead to premature exhaustion of the HSC pool. In the meantime, the HSCs become targets of the toxicity of inflammatory ROS. The ensuing consequence will depend upon the capacity of HSCs to repair ROS-induced damage, particularly DNA damage. In aged and certain disease states, inefficient repair of the oxidative damage may lead to the decrease of the HSC quality (self-renewal capacity). Hence, the toll of inflammatory stress consists in premature senescence of HSCs.
Chronic inflammation and oxidative stress are important features in the pathogenesis of BM diseases such as FA (Lavrovsky et al., 2000;Bagby, Jr, 2003;Chen, 2005). The increased oxidative stress in FA patients may be the result of an increased burden of endogenously produced oxidants as well as increased amounts of ROS generated by various inflammatory cytokines -as suggested in our study. Therefore, understanding the relationship between ROS and inflammation in the context of HSC senescence and aging in these disease states provides a unique opportunity to mechanistically comprehend, and potentially intervene in these physiologically important processes. In addition, our results suggest that, antioxidant compounds may be of therapeutic value in monitoring disease progression, and antioxidant therapy could be used to stop the initiation and propagation of inflammatory response in these diseases.

Mice and treatments
Generation of Fancc knockout mice has been described by Chen et al. (Chen et al., 1996). Fancc +/mice were intercrossed with C57Bl/6 mice for more than ten generations to develop an inbred strain. Fancc -/mice and their WT littermate controls were generated by interbreeding the heterozygous Fancc +/mice. The mice were maintained on a C57BL/6 (CD45.2 + ) background. Fancc -/-;Tnfr1 -/doubleknockout mice were generated by mating Fancc +/with Tnfr1 -/mice (Jackson Laboratory, Bar Harbor, ME), followed by the mating of F1 heterozygous siblings. All mice were used at approximately 10-14 weeks of age. Mice were injected intraperitoneally (i.p.) with mouse recombinant TNF␣ (Peprotech) in PBS at 0.1 mg/kg per day for 2 consecutive days. N-acetyl-L-cysteine (NAC; Sigma) was injected i.p. at 1 mg per mouse 30 minutes before and after each TNF␣ injection. For anti-TNF␣ antibody treatment, TNF␣-treated mice were injected with 20 g of neutralizing mouse anti-TNF␣ antibody (R&D Systems) 30 minutes after each TNF␣ injection. All experimental procedures conducted in this study were approved by the Institutional Animal Care and Use Committee of Cincinnati Children's Hospital Medical Center.

Clonogenic progenitor assays
BM progenitor cells were cultured in a 35-mm tissue culture dish in 4 ml of semisolid medium containing 3 ml of MethoCult M 3134 (Stem Cell Technologies) and the following growth factors: 100 ng/ml stem cell factor (SCF), 10 ng/ml interleukin-3 (IL-3), 100 ng/ml granulocyte colony-stimulating factor (G-CSF) and 4 U/ml erythropoietin (Peprotech). On day 10 after plating, the colonies were counted. Colony growth results are given expressed as the mean (of triplicate plates) ± s.d.

Flow cytometric analysis of HSC and lineage differentiation
Cells were suspended in FACS buffer (0.1% FCS in 0.02% sodium azide) and incubated with the indicated antibodies on ice for 30 minutes, followed by two washes. Data were collected on a FACSCalibur (Becton Dickinson). Labeling was done using the following antibodies: anti-Sca-1, anti-c-Kit, anti-B220, anti-CD3e, anti-CD4, anti-CD8, anti-Gr-1, anti-Mac-1 and anti-Ter119 (all from BD PharMingen, San Diego, CA).

Apoptosis assay and cell cycle analysis
Cells were stained with annexin V and 7-AAD using BD ApoAlert Annexin V kit (BD PharMingen) in accordance with the manufacturer's instructions. Apoptosis was analyzed by quantification of annexin-V-positive cell population by flow cytometry. For cell cycle analysis, cells were permeabilized with 0.3% Nonidet P-40 (NP-40), and then stained with propidium iodide (PI) containing 1 mg/ml RNase A, followed by FACS analysis of the G0-G1, S and G2-M populations.

Determination of ROS production
Cells were incubated with CM-H 2 DCFDA (Molecular Probe) in the dark for 15 minutes at 37°C. After washing, cells were analyzed by flow cytometry using a FACSCalibur (BD Biosciences). Data were analyzed by using the CellQuest program (BD Biosciences).

Immunohistochemistry
During necropsy, organs were removed, preserved in formalin and then embedded in paraffin blocks. Paraffin sections were deparaffinized, rehydrated, incubated in 0.1 mM sodium citrate (pH 6.0), washed and incubated with peroxidase blocking reagent (Vector Laboratories, VectaStain Elite ABC kit). After washing in PBS, slides were incubated with primary antibodies against HNE (11-S; Alpha Diagnostic International, San Antonio, TX) or HP1-␥ (07-332; Upstate Cell Signaling Solutions, Lake Placid, NY). Following three PBS washes, slides were incubated with secondary antibody and then detected with the VectaStain Elite ABC reagents.

Retroviral vectors and infection
The retroviral vector expressing the mTERT was kindly provided by Fuyuki Ishkawa (Kyoto University, Japan). Retroviruses were prepared by the Vector Core at Cincinnati Children's Research Foundation. Retroviral supernatant was collected 36 hours, 48 hours and 60 hours after transfection. Cells were plated onto non-tissue culture 24-well plates coated with Retronectin (Takara-Shuzo, Shiga, Japan) and pre-stimulated for 2 days in Iscove's modified Dulbecco's medium (IMDM) containing 20% FCS, 100 ng/ml SCF, 20 ng/ml IL-6, and 50 ng/ml Flt-3L (Peprotech). Cells were then exposed to the retroviral supernatant for 3 hours at 37°C in the presence of 4 g/ml Polybrene (Sigma). Cells were centrifuged at 600 g for 45 minutes. Infection was repeated twice and infection efficiency was assessed by the detection of green fluorescent protein (GFP)-positive cells by FACS.

Serum levels of cytokines
Serum levels of inflammatory cytokines were measured using enzyme-linked immunoadsorbent assays (ELISA) kits (R&D Systems).

Comet assay
Generation of DNA strand breaks was assessed by the single-cell gel electrophoresis (comet) assay (Fairbairn et al., 1995), using a Fpg-FLARE (fragment-length analysis using repair enzymes) comet assay kit in accordance with the manufacturer's instructions (Trevigen, Gaithersberg, MO). For each experimental point at least three different cultures were analyzed, and 100 cells were evaluated for comet-tail length from each culture.

Cytogenetic analysis
Cells were incubated in 0.1 mg/ml colcemid for 30-60 minutes and then incubated in 75 mM KCl. Chromosomes were subsequently fixed in methanol:acetic acid (3:1) and dropped onto glass slides. Metaphase chromosomes were Giemsa-stained and examined for abnormalities. Chromosome aberrations were defined using the nomenclature rules from the Committee on Standard Genetic Nomenclature for Mice.

Statistics
Data were analyzed statistically using a two-tail Student's t-test. The level of statistical significance stated in the text was based on the P values. P<0.05 was considered statistically significant.