Macrophages Regulate the Systemic Response to DNA Damage by a Cell Nonautonomous Mechanism

The DNA damage response (DDR) is a comprehensive and complex network of phosphorylation-mediated signaling pathways that originates endogenously from the DNA lesion and activates intrinsic DNA repair mechanisms. Here we describe a macrophage-dependentmechanism that regulates the response to DNA damage. We demonstrate that humanmonocytes, by releasing macrophage-derived HB-EGF, enhance DDR in neighboring cells suffering from DNA damage. Consequently, HB-EGF–treated cells exhibit higher double-strand break (DSB) rejoining and display lower levels of residual DSBs. Diethylnitrosamine (DEN) injection induce DSBs along with elevation in the number of macrophages and HB-EGF expression. Significantly, macrophage depletion or blocking HB-EGF activity results in higher levels of nonrepairable DSBs, suggesting that macrophages play a role in the resolution of DNA damage via HB-EGF. This study establishes that macrophages, acting through the activation of the EGFR cascade, constitute an important cell nonautonomous physiologic component of the DDR and points to a unique role played by immune cells in maintaining genome integrity. Cancer Res; 75(13); 1–11. 2015 AACR.


Introduction
Cells are continuously subjected to a barrage of DNA damage from both endogenous and external sources.Oxidative stress caused by extracellular and intracellular production of reactive oxygen species (ROS) represent a fundamental mechanism that contributes to DNA damage.ROS overproduction that exceeds defense mechanisms may damage intracellular macromolecules, including nucleic acids, with the formation of DNA adducts and DNA strand breaks that can result in mutations (1)(2)(3)(4).It is estimated that each cell's genomic DNA undergoes thousands of oxidative hits per day, and even more under conditions of stress.Consequently, persistent oxidative stress due to ROS is widely considered to be important in the pathology of a range of human diseases and has been specifically associated with carcinogenesis and cancer progression (5,6).
Among the various forms of DNA damage, double-strand breaks (DSB) are the most lethal for cells because failure to repair them can lead to loss of genetic information and chromosome rearrangements.Nevertheless, cells have evolved multiple signaling pathways and biochemical mechanisms that counter-balance DNA damage and trigger DNA repair.Generation of DNA breaks elicits the activation of sophisticated surveillance mechanisms, collectively called the DNA damage response (DDR; ref. 7).The cellular response to DNA breaks is orchestrated by signal transduction pathways, where the protein kinases ataxia telangiectasia mutated (ATM) and ataxia telangiectasia and Rad3-related (ATR), act as master regulators.ATM is activated by monomerization and autophosphorylation that in turn phosphorylates downstream proteins including the histone H2AX (known as gH2AX) enabling the recruitment of an array of proteins involved in DNA repair (8,9).H2AX is phosphorylated over a large region surrounding the DSB and forms readily visualized foci (10).
Many decisions affect the fate of DNA-damaged cell, for example, how to repair the damage or whether to repair it at all.These decisions mainly rely on the extent of initial damage.Relatively minor DNA damage induces transient cell-cycle arrest and then the cell recovers to normal growth after the damage is effectively repaired.At high damage levels, the damaged cells enter permanent cycle arrest and senescence and in the event of irreparable damage die by apoptosis.These decisions are critical as failure of this process that enables cells to progress through the cell cycle even in the face of incomplete repair of the damaged DNA results in genome instability, chromosomal aberrations that are expressed by the cell and its progeny thus increasing the risk of cancer, neurodegeneration, and other pathologies (11,12).
The immune system is well known for its function in defending the host against pathogen invasion.However, in recent years, it has become clear that in addition to its defensive function, the immune system also serves to maintain homeostasis in the face of constant assaults by endogenous and environmental agents (13), and it is called into action by alarm signals released by tissues undergoing stress, damage, or abnormal necrotic death (13,14).In the course of our search for signals that alert the immune system to stress, we found that cells exposed to oxidative stress release "find-me" signals that selectively attract human monocytes and mouse macrophages (15).Importantly, this response was mediated by a cooperation between class A scavenger receptors and signaling through MyD88.Surprisingly, although monocyte migration was MyD88-dependent it was not accompanied by inflammatory cytokine secretion (15).This latter finding left us with the question of what is the role of migrating macrophages in the response to stress.Because, as mentioned above, oxidizing conditions can induce a range of DNA lesions, including DSBs (1,16,17), in the current study we asked whether macrophages could regulate the DDR and DNA repair mechanisms thereby conferring additive protection from injury.
Here we demonstrate that macrophages via HB-EGF-EGFRmediated pathway enhance preexisting DDR network and accelerate DSB rejoining.These findings unveil a novel, extrinsic, physiologic (immune cell-mediated) system that, in addition to the cell-autonomous pathway, regulates the cellular response to DNA damage and may contribute to increased resistance of cells to DNA damage and to the restoration of homeostasis at the tissue level.

Animal model
Experimental protocol was approved by the Hebrew University Institutional Animal Care and Ethical Committee.
For the induction of DNA damage in vivo, 20-day-old C57BL/6 male mice (Harlan) were injected intraperitoneally with DEN (10 mg/kg body wt; Sigma-Aldrich).For macrophage depletion, clodronate-containing liposomes or control PBS-containing liposomes (ClodronateLiposomes) were injected at 150 mL per mouse 2 days before DEN injection and continued every other day.HB-EGF function was blocked by injection of 75 mg cross-reacting material 197 (CRM197, Sigma-Aldrich) per mouse.Macrophages from mouse livers were isolated as previously described (18).More than 90% of isolated cells were macrophages as assessed by F4/80 positivity.

Immunohistochemistry and immunofluorescence
Paraffin-embedded liver tissue sections were incubated with rat anti-mouse F4/80 monoclonal antibodies (AbD Serotc) or rat anti-mouse Ly6B.2 (AbD Serotc), followed by a secondary antibody anti-rat for mouse tissue (N-Histonefine) labeled with streptavidin-horseradish peroxidase using a DAB staining kit (Invitrogen).An Olympus BX61 microscope (Olympus) was used for image acquisition.To estimate DNA damage, paraffin-embedded liver tissue sections were immunofluorescence stained with anti-phospho-histone H2AX (gH2AX) monoclonal antibody (Millipore) followed by Alexa-488 goat anti mouse IgG (Molecular Probes).The number of F4/80-positive cells was determined by manual counting of at least 10 fields per sample.Positive gH2AX areas in mice livers were calculated using ImageJ software.

HB-EGF gene expression analysis
The relative level of HB-EGF mRNA in mouse livers was determined by real-time PCR using HPRT mRNA for normalization.Primer sequences are available upon request.

Cells
Primary human dermal fibroblasts (HDF) and CD14 þ cells were isolated and incubated as previously described (15).In the in vitro experimental protocol, which is adapted from our previous work (15), HDF were either incubated with hydrogen peroxide in PBS for 1 hour or were mock-treated with PBS only, and were then washed with fresh media.Only at that stage, either monocytes, HB-EGF, or EGF were added to the culture, and this time point is hence denoted as time 0. The cells were then incubated for the indicated time points and were analyzed as described in Supplementary Methods.

Cell lysis and immunoblotting
HDF were either mock-treated (PBS) or peroxide-treated (H 2 O 2 ; Sigma-Aldrich) for 1 hour.The cells were then washed and fresh media was added.Monocytes were placed in the top chamber of Transwell membranes (Corning; 0.4 mm pore size) of lower wells containing HDF and incubated for the indicated time points.In other experiments, either EGF (Sigma-Aldrich) or HB-EGF (R&D Systems) were added at the indicated concentrations.In certain experiments, cells were treated with the following signaling inhibitors: clinical grade gefitinib (100 mmol/L; Iressa/ZD1830 AstraZeneca), PD0325901 (1 mmol/L; Sigma-Aldrich), SC-514 (50 mmol/L; Cayman Chemical).At each time point, Transwell top chamber was removed and whole-cell extracts were prepared from the HDF at the bottom chamber by lysing the cells with X3 SDS-PAGE sample buffer.Lysates were separated by electrophoresis on 15% or 6% SDS-PAGE gels and then transferred to polyvinylidene difluoride membranes (Bio-Rad).The blots were probed with anti-phosphorylated ATM antibody (Ser1981; Epitomics), anti-phosphorylated DNA-PK (Ser2056; Abcam), anti-phosphorylated ATR (Ser428; Cell Signaling Technology) or anti-phosphorylated H2AX (gH2AX, Ser139; Bethyl), anti-phosphorylated P38 (R&D Systems), antiphosphorylated ERK (Sigma-Aldrich), or anti-phosphorylated AKT (Cell Signaling Technology).Following stripping, the membranes were reprobed with anti-actin monoclonal antibody (mAb; MP Biomedicals, LLC).

Flow cytometry
Freshly isolated monocytes were washed twice in PBS, and stained using anti-HB-EGF mAb (R&D Systems) for 30 minutes on ice, followed by a phycoerythrin (PE)-conjugated anti-goat (Jackson ImmunoResearch).Cells were washed twice in PBS and the immunostained cells were analyzed by a FACS Calibur flow cytometer (Becton Dickinson) using the Cell Quest software.

HB-EGF ELISA assays
Conditioned media were collected from cultures containing mock-or peroxide-treated HDF with or without monocytes or monocytes cultured alone.HB-EGF levels in the conditioned media were assayed by ELISA according to the manufacturer's protocol (R&D Systems).

Comet assay
HDFs were cultured as described.After incubation, cells were harvested and processed and analyzed using Comet Assay Kit (Trevigen) according to the manufacturer's protocol.

Statistical analysis
Data are expressed as means AE SEM.Differences were analyzed by Student t test or ANOVA, where appropriate.P 0.05 was considered significant.

Macrophages play a role in the resolution of DNA damage in DEN-treated mouse livers
To test our hypothesis that macrophages play a role in DDR, we first used a mouse model where we have induced oxidative DNA damage in mouse livers by a single injection of DEN.DEN is a DNA alkylating agent that leads to the formation of mutagenic DNA adducts.In addition, DEN metabolic activation by cytochrome P450 can generate ROS (19), which consequently Macrophages play a role in the resolution of DNA damage.Mice were injected with DEN and sacrificed at the indicated time points and their liver tissues were fixed.A, representative images of gH2AX staining (4Â objective) of paraffinembedded liver sections used to evaluate the extent of DSBs.B, a scheme depicting the spatial distribution of gH2AX foci in liver tissue sections shown in A. Green dots, gH2AX-positive cells; small red, blue, and light blue dots, portal triad; large light blue dot, central vein.C, the number of F4/80-positive cells in these liver tissue sections (average AE SEM) at the indicated times after DEN treatment.An average of 3 to 4 mice in each group is shown.ÃÃÃ , P < 0.001 by ANOVA.D-F, mice were either left untreated or injected with DEN and treated with either PBS-or clodronate-loaded liposomes.Mice were sacrificed two and 6 days after DEN injection.Representative images of F4/80 after 6 days (D) and gH2AX staining in liver tissue sections (E and F; two images for each treatment for 2 and 6 days after treatment, respectively) are shown (4Â objective).The two images at the bottom of panel F are at higher magnification (10Â objective; green, gH2AX staining; blue, DAPI).Graphs show F4/80 cell counts (D, right; average AE SEM) and gH2AX-staining areas (F, right).An average of 4 mice in each group is shown.ÃÃÃÃ , P < 0.0001.damage proteins, lipids, and DNA (20)(21)(22).Phosphorylated histone H2AX (gH2AX) foci formation is an accepted and reliable quantitative marker of DSBs, providing a sensitive assay to monitor DNA damage (23,24).We therefore performed immunofluorescence staining to detect nuclear foci containing gH2AX (Supplementary Fig. S1A) in liver tissue sections at various time points after DEN injection.gH2AX foci were clearly evident around the portal triads 24 hours after DEN injection (Fig. 1A  and B) and persisted up to 3 days postinjection.The number of cells harboring gH2AX foci then declined significantly at 6 and 9 days, leaving only few cells with detectable foci (Fig. 1A and Supplementary Fig. S1B).This time-dependent recovery is likely due to the activity of various DNA repair mechanisms (24,25).The increase in the number of gH2AX foci in DEN-injected livers was accompanied by infiltration of F4/80 þ cells (Fig. 1C), but not of neutrophils (data not shown).In our previous study (15), we have clearly demonstrated that malondialdehyde (MDA), a byproduct of lipid peroxidation, is sensed by macrophages through scavenger receptors and induce their attraction in a MyD88-dependent mechanism.Similarly, in DEN-treated livers, early increase in MDA concentration is detected (26).Addition of N-acetylcysteine, an antioxidant able to minimize the detrimental effects of oxidative stress, to the drinking water significantly reduced macrophage recruitment following DEN injection (Supplementary Fig. S1C).To test specifically for a role of macrophages in the DDR, we depleted macrophages using clodronate liposomes (Fig. 1D).We specifically measured how macrophage depletion affects the amount of residual gH2AX foci that is observed 2 and 6 days after DEN injection as a measure of the overall DSB repair capacity.While there was no significant difference in the extent of initial damage in the livers two days after DEN injection (Fig. 1E), in the absence of macrophages, both the extent and the intensity of residual DSBs 6 days after DEN injection were significantly higher as compared with control mice that were treated with PBS-containing liposomes (Fig. 1F), suggesting that macrophages play a role in the resolution of DNA damage in the liver.

Monocytes enhance DDR via HB-EGF release
To further assess DDR signaling pathways, HDFs were exposed to peroxide, which like DEN inflicts double-strand breaks, and incubated for variable time periods after peroxide removal.At the end of the incubation periods, HDFs were lysed and the phosphorylation of H2AX, and of the sensor kinases ATM, ATR, and DNA-PK (27,28), were determined.Progressive accumulation of gH2AX and pATM was observed in peroxide-treated, but not in mock-treated HDFs.Phosphorylation of these two markers was observed 30 minutes after exposure to peroxide and continued to accumulate even after peroxide removal (Supplementary Fig. S2A).Phosphorylation of DNA-PK and of ATR that are predominantly related to single-strand breaks, was not detected (not shown).
To assess whether monocytes influence DDR signaling pathways, we have turned to an in vitro system used in our previous study that is based on oxidative stress of primary culture of HDFs (15).HDFs were exposed to peroxide for 1 hour as above, washed and incubated for varying time periods with or without monocytes.To follow signaling in HDFs, monocytes were placed in an overlying 0.4 mm Transwell and at the end of the indicated time, monocytes in the Transwell were discarded and the HDFs were lysed.As can be seen in Fig. 2A, the addition of monocytes greatly enhanced phosphorylation of both H2AX and ATM in peroxidetreated HDFs, but had no effect on mock-treated cells (data not shown).
Our finding that monocytes affect the DDR in HDF cells while being separated from these cells by a Transwell membrane, points to a soluble factor mediating this activity.Gene array analysis revealed that heparin-binding EGF-like growth factor (HB-EGF) was one of the genes that were upregulated in monocytes in response to their interaction with cells suffering from oxidative stress (Fig. 2B).HB-EGF is a member of the EGF family of proteins, which binds EGF receptors with a higher affinity than EGF itself and is a far more potent mitogen.Importantly, HB-EGF transcripts are highly abundant in human monocytes and HB-EGF is expressed on the cell surface of monocytes/macrophages (Supplementary Fig. S1C; refs.29,30).Following its shedding from the surface membrane, HB-EGF is involved in macrophage-mediated cellular proliferation of fibroblasts and smooth muscle cells (29,30) The gene array data was further corroborated by ELISA.This data documented the release of HB-EGF from monocytes cocultured with HDFs that were treated with either mild (50 mmol/L) or severe (500 mmol/L) peroxide conditions (Fig. 2B and Supplementary Fig. S2D).HB-EGF secretion in response to stressed fibroblasts was detected as early as 3 hours (data not shown) and peaked at 24 hours (Fig. 2B).We have previously demonstrated that oxidatively stressed cells selectively attracts monocytes in a scavenger receptors (SR)-dependent manner (15).Like monocyte migration toward stressed cells, HB-EGF release was abrogated by treatment with the competitive SR-specific ligand, dextran sulfate (DxSO 4 ), but not with its noncompetitive counterpart, chondroitin sulfate (ChSO 4 ; Supplementary Fig. S2E), suggesting that SRs contribute to HB-EGF shedding.
Several reported findings suggested a role for HB-EGF and EGFR signaling in the cellular response to DNA damage.HB-EGF was found to protect cells from chemotherapy-induced cell death.(31) Moreover, EGFR inhibition was shown to induce cellular senescence in response to radiation and this was attributed to an increase in the levels of nonrepairable DSBs (32), suggesting a link between the EGF:EGFR pathway and the process of DNA damage repair.Finally, a previous study has demonstrated that in tumor cells exposed to ionizing radiation, activation of EGFR by EGF enhances the DSB repair capacity, whereas inhibition of EGFR attenuates DSB rejoining (33)(34)(35).
To investigate whether HB-EGF released by monocytes is biologically active and plays a causal role in DDR signaling, we performed blocking antibody experiments.As seen in Fig. 2A, the addition of neutralizing antibodies against HB-EGF or EGF receptor significantly reduced the effect exerted by monocytes on both H2AX and ATM phosphorylation in peroxide-treated fibroblasts.Interestingly, blocking antibodies reduced H2AX phosphorylation to a level lower than that observed peroxide-treated HDFs in the absence of monocytes, suggesting a role for HB-EGF and EGFR signaling in the DDR that is augmented in the presence of monocytes.Moreover, the addition of recombinant human HB-EGF (2 ng/mL) increased H2AX and ATM phosphorylation in peroxide-treated fibroblasts (Fig. 2C).HB-EGF had a similar effect on H2AX phosphorylation when lower H 2 O 2 concentrations were used (50 mmol/L instead of 500 mmol/L; Supplementary Fig. S2F), but had no effect on PBS-treated (mock) cells (Supplementary Fig. S2G), suggesting that HB-EGF enhances the DDR, but does not activate this signaling cascade by itself.Similar results were obtained when EGF was used instead of HB-EGF (Fig. 2C).However, as expected from the relative lower affinity of EGF to the EGFR as compared with HB-EGF, EGF was not active at 2 ng/mL (data not shown) and significantly higher concentrations of EGF were required to obtain a response similar to HB-EGF (i.e., 20-30 ng/mL EGF vs. 2 ng/mL HB-EGF).Given that in general there is a direct relationship between the number of microscopically visualized gH2AX foci and the extent of DNA DSBs in an individual cell, we further evaluated gH2AX foci formation in HDFs following exposure to peroxide.At the end of the peroxide treatment (time 0 hour) there was a dramatic induction of foci in all cells, which increased over time, reaching a peak at around 60 minutes after treatment.These foci were relatively small and indistinct, probably because of their large number (Fig. 2D and Supplementary Fig. S3A).In the presence of HB-EGF, there was an increase in the intensity of gH2AX staining observed in peroxide-but not in mock-treated HDFs (Fig. 2D and data not shown).However, owing to the large number of gH2AX foci, it was not possible to determine whether the increase observed was due to a higher overall number of foci or that each foci became brighter and larger.

HB-EGF facilitates DNA breaks rejoining
To further evaluate the effect of HB-EGF on DNA DSBs repair following oxidative damage, we have used the comet assay, which reflects the degree of DNA fragmentation in individual cells (36,37).HDFs that were exposed to peroxide and then incubated in the presence or absence of HB-EGF were subjected to the neutral comet assay to detect DSBs (Fig. 2E).Significant amount of DSB were evident in cells immediately after peroxide exposure and this damage increased following incubation.Surprisingly, in the presence of HB-EGF, there was no significant difference in the extent of DNA fragmentation (Fig. 2E), suggesting that the early increase in the global gH2AX and pATM levels (Fig. 2C and D) in the presence of HB-EGF is not due to an increase in the number of DNA breaks but most likely a sign of enhanced DDR signaling induced by HB-EGF.
The phosphorylation of H2AX is considered an important step in the signaling and initiation of the DNA DSB repair process, and early gH2AX foci represent either a single or a cluster of DSB that are actively being repaired.(38) Hence, the early increased response observed within the first few hours in the presence of HB-EGF (that is not coupled to enhanced DNA fragmentation) likely represents sites of enhanced DDR signaling and ongoing DSB repair.On the other hand, a decline in the number of gH2AX foci over time is believed to reflect the kinetics of DNA DSB repair per se.(8) Therefore, we next evaluated the number of residual gH2AX foci 24 hours after treatment.The high numbers of gH2AX foci observed following peroxide treatment (Fig. 2D and data not shown) significantly declined after 24 hours, leaving only few distinct foci in each cell (Fig. 3A and Supplementary Fig. S3A).These distinct nuclear foci also displayed costaining with 53BP1, an established player in the cellular response to DNA damage that is also recruited to damage sites (Supplementary Fig. S3B; ref. 39).As expected, the average number of residual foci per cell was significantly higher in cells treated with 500 mmol/L H 2 O 2 than in cells treated with 50 mmol/L H 2 O 2 (Fig. 3B).Importantly, in the presence of either HB-EGF or EGF, the average number of residual gH2AX foci per cell was significantly lower compared with control peroxide-treated HDFs at both peroxide concentrations (Fig. 3A and B).
To correlate the number of remaining microscopically visualized gH2AX foci 24 hours after peroxide treatment and the extent of DNA fragmentation, we further evaluated the effect of HB-EGF on DNA DSB repair following oxidative damage, using the neutral comet assay (Fig. 3C).A significant amount of DSBs were evident in cells immediately following exposure to peroxide (at both 50 and 500 mmol/L), before the addition of HB-EGF or EGF.Under the severe stress condition of 500 mmol/L peroxide, the extent of DSBs was considerably high even after 24 hours, in agreement with the high number of residual gH2AX foci in these cells (Fig. 3B), whereas under mild peroxide concentration (50 mmol/L), the majority of DSBs disappeared after 24 hours (Fig. 3C).A small but significant decrease in the extent of DNA fragmentation was seen in HB-EGF-and EGF-treated cells as compared with their untreated counterparts, implying that DSB rejoining was higher in these cells.As expected, under the severe stress condition (500 mmol/L peroxide), the extent of DSBs was still considerably high even in the presence of either HB-EGF or EGF.Given that DSBs were apparent before the addition of HB-EGF, we conclude that cells treated with HB-EGF repaired DSBs more efficiently.
To test whether the lower number of residual gH2AX foci per cell and the corresponding DSBs (detected by the comet assay) represented more efficient repair in HB-EGF-treated cells, we asked whether HB-EGF could promote homologous recombination (HR)-mediated DSB repair.To test this, we used an in vivo inducible on/off switch GFP reporter system that was previously described by Shahar and colleagues (40).The reporter used is designed to measure error-free HR of a site-specific break formed by the rare-cutting endonuclease I-SceI.In this reporter system, an HR repair of the I-Sce1-induced DSB restores GFP gene.To that end, we treated U2OS cells carrying the EGFP-based HR substrate with dexamethasone to activate the endonuclease I-Sce1, and then incubated the cells for 72 hours with or without either HB-EGF or EGF.HR activity was examined by flow cytometry to determine the frequency of GFP-positive cells, indicating HR-mediated repair events.In the absence of dexamethasone, the percentage of GFP þ cells remained low and was the same in the absence or presence of HB-EGF or EGF in the culture.Upon activation of the HR system by dexamethasone, we detected the expression of GFP in a small subset of cells that had undergone the relatively rare process of HR.Notably, in the presence of HB-EGF or EGF, the number of cells that expressed GFP, hence had undergone HR repair, was significantly increased compared with that in controluntreated cells (Fig. 3D).
We next followed EGFR signaling events in cells suffering from DNA damage.Interestingly, following peroxide treatment we observed an induction of signals associated with the EGFR pathways including phosphorylation of AKT, ERK, and P38 in HDF cells (Fig. 4A).Notably, while EGFR triggering by HB-EGF caused a decrease in the number of residual DSBs 24 hours after peroxide treatment (Fig. 3A and B), blocking EGFR signaling by gefitinib (that inhibits the EGFR tyrosine kinase domain) as well as blocking MEK1/2 activity with PD0325901 or attenuating the downstream NF-kB-induced gene expression by the IKKb-selective inhibitor SC514 increased the number of gH2AX foci and reversed HB-EGF effect (Fig. 4B).Taken together, these data demonstrate a link between EGFR signaling and the DSB repair process, and suggest that an additive triggering of the EGFR by macrophage-derived HB-EGF may further facilitate DSBs repair.

HB-EGF rescue oxidized cells to resume proliferation
We next asked whether the observed beneficial effect of HB-EGF on DNA repair also affected cell-cycle arrest.Indeed, we show here that at least under mild (i.e., 50 mmol/L peroxide) oxidative conditions cell-cycle-arrested HDF can resume proliferation in the presence of HB-EGF (Supplementary Fig. S3E-S3G).Interestingly, as revealed by carboxyfluorescein diacetate succinimidyl ester (CFSE) dilution experiments following peroxide treatment all HDF completed one cell division at the first 24 hours, and then stopped proliferating (see ref. 15; Supplementary Fig. S3H).Moreover, in the 500 mmol/L H 2 O 2 -treated HDF, the addition of monocytes or HB-EGF did not restore proliferation (Supplementary Fig. S3H).Since, under the same conditions, significant increase in DSBs rejoining is observed (Fig. 3A and B), the data suggest that HB-EGF effect on DNA repair cannot be solely explained by HB-EGF mitogenic activity that promotes cell-cycle entry, during which HR is mainly utilized.

Blocking HB-EGF activity attenuates DSBs repair in vivo
Finally, we tested whether HB-EGF plays a role in the DDR in vivo, using DEN-treated mice.After DEN injection, HB-EGF mRNA levels in mouse livers increased, peaking at 2 to 3 days after injection and declining to basal levels by days 6 to 9 (Fig. 5A).This kinetics of HB-EGF expression corresponds to the number of F4/80 þ cells in DEN-injected livers (Fig. 1C).Macrophages purified from mouse livers before and after DEN injection contained significantly higher levels of HB-EGF transcripts when compared with the corresponding whole liver tissues (Fig. 5B, left).Moreover, total HB-EGF expression in DEN-treated livers was significantly reduced after depletion of macrophages using clodronate liposomes, and declined to levels lower than those found in untreated animals (Fig. 5B, right).Collectively, these data demonstrate that HB-EGF is predominantly expressed by resident kupffer cells and by macrophages recruited to DEN-treated livers, and are consistent with previous data showing that HB-EGF mRNA levels increased primarily in kuppfer cells after partial liver hepatectomy but not in hepatocytes (41).Cross-reacting material 197 (CRM197) is a diphtheria toxin mutant that binds directly to the EGF-like domain of HB-EGF and represses its activity (42,43).Notably, mice treated with CRM197 exhibited significant increase in residual DSB 6 days after DEN injection as compared with control mice (Fig. 5C).While this result was also achieved by macrophage depletion (Fig. 1E), it is important to note that CRM197 treatment had no effect on the number of F4/80 þ cells in DEN-injected livers (Fig. 5D).
In summary, our study provides a novel insight on how the macrophages via the EGFR signaling pathway maintains genome integrity by enhancing DNA damage signaling and facilitating repair in neighboring cells.

Discussion
In this study, we have uncovered a new and as yet unknown function for macrophages that act both in mice and humans.The data presented here show that macrophages participate in vivo in the regulation of DDR.Moreover, the release of HB-EGF by macrophages allows these cells to coordinate the DDR in animals experiencing DSBs.Our results show that shortly after the induction of DSBs in mouse livers the numbers of macrophages increased and the levels of HB-EGF transcripts were elevated.Both subside by day 6 after DEN injection in correlation with DSBs resolution.Significant for this study, the resolution of DSBs was significantly impaired in the absence of macrophages or when HB-EGF activity is selectively blocked as determined by the number of remaining gH2AX foci.Thus, we propose that in addition to the autonomous mechanisms of DDR and repair that transpire inside the cell, the recruitment of hematopoietic cells, constitute an additional, cell-extrinsic, or nonautonomous regulatory circuit for maintaining genome integrity at the tissue level.In accord with this proposal (Supplementary Fig. S4), in response to oxidative DNA damage (15), macrophages are recruited to the damaged site and in turn invoke a signaling pathway that is based on the release of HB-EGF and triggering of the EGFR pathway that interacts with the cell DDR.As a consequence, DDR is enhanced, leading to higher DSBs rejoining.
While in this study we suggest that macrophage-derived HB-EGF plays a central role in the DDR it is important to note that HB-EGF expression is not limited to macrophages as previous reports have demonstrated the expression of HB-EGF in various other cell types.Indeed, HB-EGF transcripts were readily detected in HDFs and the protein was detected in HDFconditioned media, albeit at significantly lower levels as compared with monocytes.Moreover, EGFR signaling is clearly activated in peroxide-treated HDFs and blocking EGFR signaling pathway by specific inhibitors attenuates DSB rejoining.Nevertheless, we suggest that this process is further regulated systemically by excess release of HB-EGF by monocytes and macrophages that in turn augment DDR signaling and facilitate damage repair.
The exact mechanism downstream to EGFR signaling and the way it interacts with the DDR remains to be established.Several previous studies have demonstrated that ionizing radiation activated the DNA repair genes the X-ray cross-complementing group 1 protein (XRCC1) and the nucleotide excision repair gene (ERCC1) in an ERK1/2-dependent fashion, and that activation of EGFR by EGF administration markedly increased the up-regulation of phosphorylated-ERK, ERCC1 and XRCC1 After 1 hour, the cells were washed and fresh culture media was added.At the indicated time points, HDFs were immediately lysed.Cell extracts were subjected to SDS-PAGE followed by immunoblotting using antiphosphorylated AKT, anti-phosphorylated ERK, and anti-phosphorylated P38.Anti-b-actin immunoblotting revealed relative amounts of protein in each lane.B, HDFs were plated on a cover slide in a 6-well dish and were treated as in A in the presence or absence of the indicated inhibitors.After 1 hour, the cells were washed and fresh culture media was added and the cells were cultured for 24 hours in the presence or absence of HB-EGF (2 ng/mL).Then, HDFs were fixed and immunostained using anti-gH2AX mAb.Graph shows the number (median AE SEM) of gH2AX foci per cell (at least 100 nuclei counted for each treatment).One representative of three independent experiments is shown.ÃÃÃ , P < 0.001; ÃÃÃÃ , P < 0.0001.(44)(45)(46)(47)(48).These observations suggest a role for EGFR-induced MAPK signaling in the recovery from DNA damage following exposure to ionizing radiation.Moreover, a study by Dittmann and colleagues (49) suggested that EGFR translocates to the nucleus in response to radiation-induced DNA damage, and more importantly, the nuclear EGFR shuttling was shown to be involved in the DNA damage repair.Nonetheless, our findings support these and other reports that demonstrated that activation of EGFR enhances overall DSB repair (33)(34)(35).
The relationship between the immune system and cancer is complex and has been extensively studied.However, fundamental questions remain unanswered, such as: When are cancer cells recognized by immune cells?What are the consequences of this initial recognition event?Historically, the experimental approaches to these questions have revolved around the contro-versial notion of "tumor immune surveillance," and have been investigated through the lens of immune-mediated cytotoxicity and tolerance induction in established tumors or premalignant lesions (50).This study introduces a more general perspective for the way the immune system operates and for the notion of tumor "immune surveillance".
The current study offers an additional angle for considering tumor immune surveillance, by positing a unique new role for immune cells in the earliest events of carcinogenesis.While immune cells play decisive role at different stages of tumor development and promotion, our current concept of immune: tumor interactions is based on assessments of already established tumors.This study extends the spectrum of immune effects on the pathogenic continuum of cancer to the earliest event of carcinogenesis, i.e., the initial DNA damage that results in the accumulation of tumor-promoting mutations.Hence, this line of investigation provides fundamental insights into the role of immune cells in cancer initiation.Macrophages may promote cellular repair and provide a window of opportunity for the damaged cell to correct itself.But, this represents a double-edge sword.While macrophages may prevent tumor initiation by assisting the damaged tissue to overcome a damaging DNA hit, under certain circumstances, they may contribute to cancer development by enabling cells to progress through the cell cycle even in the face of incomplete repair of the damaged DNA.Regardless of which edge of the "doubleedged" sword turns out to be dominant, this line of investigation should provide fundamental insights into the role of immune cells in cancer initiation.
In addition, in already established tumors, macrophages may help tumor cells acquire chemo-and radioresistance by facilitating DNA damage resolution.In this regard, a recent study by Houthuijzen and colleagues (51) has demonstrated that macrophages residing in the spleen are activated by circulating fatty acids to secrete secondary lysophospholipids, which confer the acquisition of chemotherapy resistance in tumors by altering the DDR.
In summary, our findings move the prevailing view of the DDR from a cell-autonomous model to a dynamic two-compartment model, in which the immune system represents a regulatory network that plays a role in maintaining tissue internal environment.This novel physiologic link between a classical immune cell (macrophage) and a common cellular event (DNA damage) has implications for a broad range of cells in diverse human diseases, such as cancer, neurodegeneration, and other pathologies.The elucidation of these novel pathways holds promise for revealing new therapeutic targets.

Figure 1 .
Figure 1.Macrophages play a role in the resolution of DNA damage.Mice were injected with DEN and sacrificed at the indicated time points and their liver tissues were fixed.A, representative images of gH2AX staining (4Â objective) of paraffinembedded liver sections used to evaluate the extent of DSBs.B, a scheme depicting the spatial distribution of gH2AX foci in liver tissue sections shown in A. Green dots, gH2AX-positive cells; small red, blue, and light blue dots, portal triad; large light blue dot, central vein.C, the number of F4/80-positive cells in these liver tissue sections (average AE SEM) at the indicated times after DEN treatment.An average of 3 to 4 mice in each group is shown.ÃÃÃ , P < 0.001 by ANOVA.D-F, mice were either left untreated or injected with DEN and treated with either PBS-or clodronate-loaded liposomes.Mice were sacrificed two and 6 days after DEN injection.Representative images of F4/80 after 6 days (D) and gH2AX staining in liver tissue sections (E and F; two images for each treatment for 2 and 6 days after treatment, respectively) are shown (4Â objective).The two images at the bottom of panel F are at higher magnification (10Â objective; green, gH2AX staining; blue, DAPI).Graphs show F4/80 cell counts (D, right; average AE SEM) and gH2AX-staining areas (F, right).An average of 4 mice in each group is shown.ÃÃÃÃ , P < 0.0001.

Figure 2 .
Figure 2.Monocytes enhance DDR signaling in neighboring cells via HB-EGF release.A, HDFs were treated with 500 mmol/L peroxide and were then cultured with or without either anti-HB-EGF blocking antibodies or anti-EGFR-neutralizing antibodies; the cells were then cultured in the presence or absence of monocytes placed in a 0.4 mm Transwell insert.At the indicated time points, the Transwells were discarded and the HDFs were immediately lysed.Cell extracts were subjected to SDS-PAGE and anti-phosphorylated H2AX (gH2AX) and anti-phosphorylated ATM immunoblotting.Anti-b-actin immunoblotting revealed relative amounts of protein in each lane.B, monocytes were either cultured alone or cocultured with mock-or peroxide-treated HDF (50 and 500 mmol/L).After 24 hours (black bars), 48 hours (gray bars), and 72 hours (white bars), conditioned media were collected and the levels of HB-EGF secreted into the media were determined using ELISA.The data represent the mean values of triplicate samples and SDs.Data represent one of three independent experiments.C, HDFs were treated, as above, and then cultured in the absence or presence of either HB-EGF (2 ng/mL) or EGF (30 ng/mL).At the indicated time points, HDFs were immediately lysed.Cell extracts were subjected to SDS-PAGE followed by anti-phosphorylated H2AX (gH2AX) and anti-phosphorylated ATM mAbs.Anti-b-actin immunoblotting revealed relative amounts of protein in each lane.Similar results were obtained in three independent experiments.D, HDFs were plated on a cover slide and were treated with 50 mmol/L peroxide, and cultured in the absence or presence of HB-EGF (2 ng/mL).HDFs were then fixed and immunostained using anti-gH2AX mAb.Representative immunofluorescence images showing gH2AX foci (green) with DAPI counterstain (blue).E, HDFs were peroxide-treated, as above, and then were cultured in the absence (black bars) or presence of HB-EGF (white bars; 2 ng/mL).At the indicated times, cells were collected and were subjected to the neutral comet assay.Time 0 represents cells that were harvested immediately at the end of the peroxide treatment.The comet moment value that represents the extent of genomic DNA fragmentation in individual cells is shown.At least 60 cells were scored for each treatment.Data are expressed as median AE SEM (in arbitrary unit).One representative of two independent experiments is shown.ns, nonsignificant.

Figure 3 .
Figure 3.HB-EGF as well as EGF accelerate DNA damage repair.HDFs were plated on a cover slide and were either mock-or peroxide-treated (either 50 or 500 mmol/L H2O2), as above, and then were cultured for 24 hours in the presence or absence of either HB-EGF (2 ng/mL) or EGF (30 ng/mL).Then, HDFs were fixed and immunostained using anti-gH2AX mAb.A, representative confocal immunofluorescence images showing gH2AX foci (green) with DAPI counterstain (blue) in HDFs treated with 50 mmol/L H2O2 and incubated without (none) or with either EGF or HB-EGF.B, graph showing the number (median AE SEM) of gH2AX foci per cell (at least 100 nuclei counted for each treatment) in HDFs treated with 500 mmol/L (black bars) and 50 mmol/L (gray bars) H2O2.One representative of three independent experiments is shown.ÃÃÃ , P < 0.001; ns, nonsignificant.C, HDFs were treated as above with either 500 mmol/L (top) or 50 mmol/L H2O2 (bottom) in the presence or absence of either HB-EGF (2 ng/mL; left) or EGF (30 ng/mL; right).After 24 hours, cells were harvested and were subjected to the neutral comet assay.Time 0 represents cells that were harvested immediately at the end of the peroxide treatment.At least 100 cells were scored for each treatment.Data are expressed as the median value of the comet moment AE SEM (in arbitrary unit) for each treatment after subtracting the median value of a sample of untreated healthy cells used to control for endogenous levels of damage within cells and for damage that may occur during sample preparation.For each graph, one representative of two independent experiments is shown.Ã , P < 0.05; ÃÃ , P < 0.01; ÃÃÃ , P < 0.001.D, U2OS reporter cells were either left untreated or treated with dexamethasone (Dex) that activates I-SceI and induces DSBs in the GFP-based reporter cassettes.The cells were then incubated for 3 days with or without either HB-EGF (2 ng/mL) or EGF (30 ng/mL).The frequency of successful repair by HR was detected by the appearance of GFP þ cells as determined by flow cytometric analysis.One representative experiment of three is shown.No error bars are shown in D, because their normalized percentages of GFP þ cells are set to one.ÃÃ , P < 0.01; ÃÃÃ , P < 0.001.

Figure. 4 .
Figure.4.Blocking EGFR signaling attenuates DSBs repair.A, HDFs were plated in a 24-well dish and treated with the indicated inhibitors (see Materials and Methods) and were either mock-treated (left) or treated with 50 mmol/L peroxide (right).After 1 hour, the cells were washed and fresh culture media was added.At the indicated time points, HDFs were immediately lysed.Cell extracts were subjected to SDS-PAGE followed by immunoblotting using antiphosphorylated AKT, anti-phosphorylated ERK, and anti-phosphorylated P38.Anti-b-actin immunoblotting revealed relative amounts of protein in each lane.B, HDFs were plated on a cover slide in a 6-well dish and were treated as in A in the presence or absence of the indicated inhibitors.After 1 hour, the cells were washed and fresh culture media was added and the cells were cultured for 24 hours in the presence or absence of HB-EGF (2 ng/mL).Then, HDFs were fixed and immunostained using anti-gH2AX mAb.Graph shows the number (median AE SEM) of gH2AX foci per cell (at least 100 nuclei counted for each treatment).One representative of three independent experiments is shown.ÃÃÃ , P < 0.001; ÃÃÃÃ , P < 0.0001.

Figure 5 .
Figure 5.Blocking HB-EGF activity attenuates DSBs repair in vivo.A, qRT-PCR analysis of HB-EGF expression in RNA extracts from frozen mouse liver tissues harvested at various time points following DEN injection.The average of 3 mice in each time point is shown.B, qRT-PCR analysis of HB-EGF expression in RNA extracts from macrophages isolated from liver tissues harvested from untreated or DEN-injected mice.The expression level of HB-EGF in the corresponding whole liver tissue is set to 1 (left).qRT-PCR analysis of HB-EGF expression in RNA extracts from frozen mouse liver tissues harvested from mice treated with either PBS-or clodronate-loaded liposomes, 6 days after DEN injection (right).An average of 3 mice in each group.Ã , P < 0.05; ÃÃ , P < 0.01; ÃÃÃ , P < 0.001.C, mice were injected with either PBS or CRM197 and then either left untreated or injected with DEN.Mice were sacrificed 6 days after DEN injection.Representative images of gH2AX staining in liver sections (two images for each treatment) are shown.Graph shows gH2AX staining areas (right; average AE SEM).An average of 4 mice in each group.ÃÃÃÃ , P < 0.0001.D, the number of F4/80-positive cells in liver tissue sections of mice injected with either PBS or CRM197 and then injected with DEN (average AE SD).An average of 3 to 4 mice in each group is shown.ns, nonsignificant.