Radiation-Induced Macrophage Senescence Impairs Resolution Programs and Drives Cardiovascular Inflammation Free

Inflammation resolution, an active process that tempers proinflammatory factors and promotes tissue repair, is controlled by several endogenous mediators that include specialized proresolving mediators (SPM), such as lipoxins, resolvins, protectins, and maresins (1). SPMs are protective in vivo, act to control leukocyte trafficking, enhance the clearance of dead cells (i.e., efferocytosis), and promote tissue repair in a manner that does not compromise host defense (1). Dysregulated inflammation resolution is associated with several prevalent human diseases, including atherosclerosis, and so understanding processes that may derail these programs are of clinical interest.

Ionizing radiation, which is broadly used as a treatment for some types of cancers, is associated with increased atherosclerosis (2–4). Documented radiation-induced cardiovascular disease extends beyond cancer therapy and is also associated with environmental and occupational exposure (5). Radiation is thought to promote atherosclerosis through direct injury to endothelial and smooth muscle cells within the vasculature (2), yet other cellular players are likely involved. Along these lines, macrophages reside in nearly every tissue, and understanding how macrophages respond to radiation is of interest (6). Macrophages are also critical effectors of inflammation resolution, and how radiation impacts temporal resolution programs and efferocytosis is not known.

Radiation can arrest the proliferation of cancer cells, but an off-target effect is a maladaptive halt in the proliferation of otherwise healthy cells, a process called senescence (7). Radiation-induced senescence provokes the senescence-associated secretory phenotype (SASP), which is associated with increased oxidative stress (OS), aberrant metabolic programs, and the release of proinflammatory factors (8). There are major gaps in our understanding as to why radiation-induced SASP is left unchecked by our bodies, and assessing the mechanisms associated with inflammation and resolution/repair in this context are highly underexplored.

Hematopoietic cells and macrophages are important players in the progression of atherosclerosis (9, 10). Sublethal radiation leads to hematopoietic stem cell (HSC) senescence and mimics several features of aging and atherosclerosis, such as alterations in HSC functions, including reduced clonogenicity and skewed differentiation toward myeloid lineages (11). Moreover, human macrophages acquire a proinflammatory phenotype when exposed to sublethal radiation (6, 12). Therefore, we posit that radiation promotes macrophage senescence/SASP and those senescent (SC) macrophages contribute to prolonged inflammation and atherosclerosis progression.

In this study, we used a model of sublethal radiation and observed a significant delay in temporal inflammation resolution in vivo that was associated with impaired efferocytosis. Treatment with the SPM called resolvin D1 (RvD1) promoted inflammation resolution and efferocytosis in sublethally irradiated (IR) mice. We found that macrophages do, indeed, undergo senescence when exposed to radiation and that radiation-induced senescence led to increased proinflammatory cytokines/chemokines and OS and defective efferocytosis that was reversed with the treatment of RvD1 in vitro. Necrosis, p16^INK4A cells, and decreased lesional collagen compared with non-IR controls and treatment with RvD1 significantly reduced necrosis and increased lesional collagen. Conditional removal of hematopoietic p16^INK4A cells during advanced atherosclerosis reduced plaque necrosis and increased key SPMs in plaques. Together, we found that sublethal radiation promotes macrophage senescence and that RvD1 can be a new therapeutic strategy to combat radiation-induced damage.

Male 8–10-wk-old C57BL/6 mice were purchased from Taconic Biosciences, and 8-wk-old male Ldlr^−/− mice were purchased from The Jackson Laboratory. Mice were housed in the Albany Medical College Animal Research Facility. All animal experiments were conducted in accordance with the Albany Medical College Institutional Animal Care and Use Committee guidelines for animal care and were approved by the Animal Research Facility at Albany Medical College.

Male C57BL/6 mice (Taconic Biosciences) were mock or sublethally γ-IR (7 Gy) and were given a 3-mo recovery period to induce bone marrow myeloid cell senescence (13). After 3 mo, mice were i.p. injected with 200 µg of zymosan A (ZymA) (catalog no. Z4250; Sigma-Aldrich) per mouse. For RvD1 treatment studies, mice were i.p. injected with 300 ng of RvD1 (catalog no. 10012554; Cayman Chemical) and 200 µg of ZymA simultaneously, and peritoneal exudates were collected by lavage 24 h postinjection. For some experiments, control or SC-macrophages (0.8 × 10⁶ cells per mouse) were i.p. injected simultaneously with ZymA, and peritoneal cells were collected as described below. Peritoneal cells were harvested at the indicated time points and were then enumerated with a hemocytometer and trypan blue exclusion. Remaining cells were washed in FACS buffer (PBS containing 5% [vol/vol] of FBS) and labeled with FITC anti–Ly-6G (catalog no. 127607; BioLegend) and allophycocyanin anti-F4/80 (catalog. no. 17-4801-82; eBioscience,) for 30 min at 4°C. Cells were then washed and resuspended in FACS buffer before performing flow cytometric assessment. Flow cytometry was carried out on a FACSCalibur (BD Biosciences), and data were analyzed by FlowJo software. Also, whole bone marrow was flushed from femurs, and mRNA was extracted with a QIAGEN RNeasy Mini Kit (catalog no.74106). cDNA was synthesized using a QuantiTect Reverse Transcription Kit (catalog no. 205313) according to manufacturer’s instructions. Bone marrow mRNA was assessed for p16^INK4A, p19^ARF, and p21 gene expression by quantitative RT-PCR (qRT-PCR) (details below). In parallel, for experiments to determine polymorphonuclear cell (PMN) frequency, whole bone marrow was flushed from femurs and tibias, and after RBC lysis, cell suspensions were plated and stained using the following Abs (BioLegend): PE-Cy7–conjugated CD11b (M1/70, catalog no.101216), allophycocyanin-Cy7–conjugated Ly-6G (A18, catalog no. 127623) and Pacific Blue–conjugated Ly-6C (HK1.4, catalog no. 128013). Surface-stained cells were analyzed on an LSR II (BD Biosciences) with FACSDiva software and analyzed using FlowJo software (Tree Star, Ashland, OR).

Elicited peritoneal macrophages from C57BL/6 mice were collected by i.p. injection of ZymA (300 µg/mouse). Mice were sacrificed 48 h after ZymA injection, and peritoneal cells were collected by lavage. Cells were enumerated as above, and macrophages were plated (700 × 10³ cells per well in a 12-well tissue culture plate) in DMEM containing high glucose (4.5 g/l) (10-013-CV; Corning), 10% FBS, 20% l cell conditioned media (vol/vol), and 1% penicillin–streptomycin overnight. Media was refreshed the next day, and the adherent peritoneal macrophages were subjected to 5 Gy of γ-radiation (Gammacell 40 Exactor; Nordion) and cultured for three additional days with the above media. For some experiments, SC-macrophages that had been exposed to γ-radiation for 2 d were then treated with either vehicle (PBS) or 10 nM RvD1 for an additional 24 h.

Human fetal lung IMR-90 fibroblast cells were purchased from Coriell Institute for Medical Research (I90-83; Camden, NJ). Proliferating IMR-90 cells (passage 6–15) were cultured in Eagle’s MEM (10-009-CV; Corning) with 5% FBS and 1% penicillin–streptomycin until ∼80% confluency. IMR-90 cells were detached with trypsin-EDTA (catalog no. 03690; Sigma-Aldrich), enumerated, and plated in a 10-cm cell culture dish (750 × 10³ cells per dish). Cells were cultured overnight (∼16 h) to allow for attachment and were then subjected to 10 Gy of γ-radiation and cultured (37°C, 5% CO₂) for 10 d. Fresh media was replaced on days 2, 5, and 8. On day 10, SC IMR-90 cells were treated with either vehicle (PBS, catalog no. 21-040-CV; Corning) or 10 nM RvD1 in serum-free media for an additional 24 h, and end-point analyses (all of which are described in detail below) were performed on day 11.

The total RNA was extracted from control or SC-macrophages or IMR-90 cells using a QIAGEN RNeasy Mini Kit (catalog no. 74106), and cDNA was synthesized using QuantiTect Reverse Transcription Kit (catalog no. 205313) according to manufacturer’s instruction. Expression of mRNA was assessed with PerfeCTa SYBR Green FastMix (catalog no. 101414-288; QuantaBio) and run on a Bio-Rad CFX Connect Real-Time qRT-PCR machine. Relative expression (ΔCt) was normalized to housekeeping genes, and the (ΔΔCt) method was used. The sequences for human and murine primers are as follows: murine housekeeping gene 18S forward, 5′- ATG CGG CGG CGT TAT TCC-3′ and reverse, 5′-GCT ATC AAT CTG TCA ATC CTG TCC-3′; murine p16^INK4A forward, 5′-AAT CTC CGC GAG GAA AGC-3′ and reverse, 5′-GTC TGC AGC GGA CTC CAT-3′; murine p21 forward, 5′-TTG CCA GCA GAA TAA AAG GTG-3′ and reverse, 5′-TTT GCT CCT GTG CGG AAC-3′; murine p19^ARF forward, 5′-GCC GCA CCG GAA TCCT-3′ and reverse, 5′-TTG AGC AGA AGA GCT GCT ACGT-3′; human housekeeping gene 18S forward, 5′ATG GGC GGC GGA AAA TAGC-3′ and reverse, 5′-TCT TGG TGA GGT CAA TGT CTGC-3′; human p16^INK4A forward, 5′-CTT CGG CTG ACT GGC TGG-3′ and reverse, 5′-TCA TCA TGA CCT GGA TCG GC-3′; human p21 forward, 5′-AGT CAG TTC CTT GTG GAG CC-3′ and reverse, 5′-GAC ATG GCG CCT CTG-3′; human p19^ARF forward, 5′-CCC TCG TGC TGA TGC TAC TG-3′ and reverse, 5′-ACC TGG TCT TCT AGG AAG CGG-3′.

Control or SC-macrophages were prepared as above and plated in an eight-well–chambered coverslip (Lab-Tek). Three days postradiation, macrophages were fixed with 100% methanol for 5 min, washed with 1× PBS, and incubated with 0.4% Triton X-100 for an additional 10 min at room temperature. Fixed cells were blocked with 5% BSA in PBS for 1 h at room temperature and incubated with rabbit anti-mouse Ki67 primary Ab (ab15580; Abcam) at 1:200 dilution overnight at 4°C. The following day, macrophages were then washed with 1× PBS to remove unbound Ki67 Ab and incubated with Alexa Fluor 594 goat anti-rabbit secondary Ab (A-11037; Invitrogen) at 1:250 dilution for 2 h at room temperature. Nuclei were stained with Hoechst for 10 min, and images were acquired immediately on a Leica SPE confocal microscope; six to seven different fields were acquired per well per group. Macrophages whose Ki67 stain colocalized with the nucleus were considered a Ki67-positive cell. The total number of macrophages and Ki67⁺ macrophages were counted and expressed as a percentage of Ki67⁺ cells.

Control or SC-macrophages were stimulated with vehicle, 10nM RvD1, or 10 µM N-acetyl-l-cysteine (NAC; reactive oxygen species [ROS] inhibitor, catalog no. A7250; Sigma-Aldrich) for 24 h. Cells were then stained with 5 µM CellROX Green staining solution for 30 min as per manufacturer’s instructions (catalog no. C10492; Invitrogen), followed by a counterstain with DAPI, and images were immediately acquired on a Leica confocal microscope. Six to seven different visual fields per treatment were analyzed, and the percentage of CellROX-positive cells per total number of cells (nuclear stain) per visual field was enumerated.

SC IMR-90 cells or macrophages were treated with vehicle or RvD1 as described above. Cells were seeded (30,000 cells per well) on a Seahorse XF96 cell culture plate (catalog no. 102601-100; Agilent Technologies) on the day of the assay. For measuring glycolytic ATP production, IMR-90 cells were incubated in Seahorse XF DMEM (pH = 7.4) (catalog no. 103575-100; Agilent Technologies) supplemented with 1 mM pyruvate (catalog no. S8636; Sigma-Aldrich), 10 mM glucose (catalog no. G8769; Sigma-Aldrich), and 2 mM l-glutamine (catalog no. 25-005-Cl; Corning) at 37°C for 1 h. Basal extracellular acidification rate (ECAR) was measured using the Seahorse XFe Real-Time ATP Rate Assay Kit (catalog no. 103592-100; Agilent Technologies) and run in the Seahorse XF96 Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA). After three basal ECAR measurements, the oligomycin A (1.5 μM) and rotenone/antimycin A (0.5μM) were serially injected after every three measurements. The glyco-ATP production rate was calculated according to equations described in the Agilent Seahorse XF Real-Time ATP Rate Assay user guide.

Macrophages were incubated in glucose-free, DMEM (pH = 7.4; catalog no. 102353; Agilent Technologies) at 37°C and supplemented with 143 mM NaCl and 2 mM l-Glutamine for 1 h for measuring glycolytic function. After three basal ECAR measurements, 20 mM Glucose (catalog no. G8769; Sigma-Aldrich) followed by 1 μM oligomycin, (catalog no. 75351; Sigma-Aldrich) and then 80 mM 2-deoxy glucose (catalog no. D6134; Sigma-Aldrich) were injected in a sequential manner. The ECAR after addition of glucose is reported as fold change.

Senescence-associated β-galactosidase (SA-β-gal) activity was detected in IMR-90 cells using a Senescence β-Galactosidase Staining Kit (catalog no. 9860; Cell Signaling Technology). IMR-90 cells (40,000 cells per well in a 12-well plate) were seeded in complete media, senescence was induced and treated with vehicle or RvD1 as described above. The cells were fixed and incubated for 12–14 h at 37°C in the presence of the β-galactosidase staining. Next, cells were imaged on a Zeiss brightfield microscope, and 10–15 different fields were acquired per well per group. SA-β-gal⁺ cells were enumerated based on blue staining.

SA-β-gal activity in control or SC-macrophages was determined using flow cytometry (as described in Ref 14). Briefly, 700 × 10³ macrophages per well were seeded into 12-well plates in DMEM, as described above; senescence was induced and treated with vehicle or RvD1 as described above. The media was removed, and cells were incubated with 100vnM bafilomycin A1 for 1 h in fresh cell culture medium. 5-Dodecanoylaminofluorescein di-β-d-galactopyranoside (C12-FDG; 33 µm) is a fluorescent β-gal substrate and was added for an additional 1 h at 37°C. Macrophages were detached using Cellstripper (catalog no. 23-25-056-CI-PK; Corning), centrifuged at 250 × g for 5 min at 4°C, and resuspended in 0.4 ml of FACS buffer. Samples were acquired with an LSR II flow cytometer and analyzed with FlowJo software.

Control and SC-macrophages were fixed with 4% paraformaldehyde for 10 min and permeabilized with 1× Perm/Wash Buffer (catalog no. 51-2091KZ; BD Biosciences) for 30 min at room temperature. Fixed cells were stained with rabbit anti-mouse COX-2–Alexa 488 at 1:100 dilution in a perm/wash buffer (catalog no. 13596S; Cell Signaling Technology) for 1 h at room temperature. Stained cells were washed with FACS buffer (5% BSA in PBS), and 10,000 events were collected in BD FACSCalibur. Data were analyzed using FlowJo software. Results were expressed as fold change of the mean fluorescence intensity of COX-2.

Supernatants from control or SC-macrophages or IMR90 cells were collected and subjected to PGE₂ analysis by ELISA (Cayman Chemical).

Control or SC-macrophages were stimulated with vehicle, 10 nM RvD1, or 10 µM NAC 24 h prior to performing efferocytosis. On the next day, Jurkats were enumerated and then stained with PKH26 (Sigma-Aldrich) according to the manufacturer’s instructions. Excess dye was removed by washing, and the Jurkats were resuspend in RPMI 1640 containing 10% FBS. To induce apoptosis, Jurkats were then exposed to UV radiation (0.16 A, 115 V, 254-nm wavelength) for 15 min at room temperature and then placed in an incubator (37°C, 5% CO₂) for 3 h (15). Apoptotic Jurkats were either cocultured with macrophages in a 3:1 ratio or with IMR-90 cells in 10:1 ratio for an additional 1 h or 2 h, respectively, in 37°C, 5% CO₂. Excess apoptotic cells were removed by washing approximately three times with PBS. The cells were then immediately fixed with 4% formalin and subjected to fluorescence imaging with a Bio-Rad ZOE Fluorescent Cell Imager. Six to seven different fields were acquired per well per group, and an efferocytic event was considered a macrophage containing red apoptotic cells. Results were expressed as the percentage of efferocytosis per total macrophages.

Male Ldlr^−/− mice (8 wk old) were subjected to mock or 7 Gy of γ-radiation and immediately fed a Western diet (WD; TD.88137; Envigo) for 12 wk. Mice were socially housed in standard cages at 22°C under a 12-h light and 12-h dark cycle. During weeks 12–15, WD-fed Ldlr^−/− mice were randomly assigned to receive vehicle (i.e., 500 μl of sterile PBS) or RvD1 (100 ng/mouse) for an additional 3 wk while still on the WD. Mice were sacrificed at the end of 15 wk. Lesion and necrotic area analysis were carried out on H&E-stained lesional cross sections, and lesional collagen was determined by picrosirius red staining as per the manufacturer’s instructions (catalog no. 24901; Polysciences). Images were acquired using an Olympus camera and Olympus DP2-BSW software, as previously described (16). Briefly, frozen specimens were immersed in OCT, cryosectioned, and 10-μm sections were placed on glass slides. The atherosclerotic lesion area, defined as the region from the internal elastic lamina to the lumen, was quantified by taking the average of six sections spaced ∼24 μm apart, beginning at the base of the aortic root.

γ-Radiation–induced murine atherosclerosis experiments were performed as described above. Frozen sections were fixed with ice-cold 100% methanol for 15 min and washed with 1× PBS. Fixed sections were incubated with 0.3% Triton X-100 for 10 min and then blocked with blocking buffer (1% BSA in PBS plus 0.3% Triton X-100) for 1 h at room temperature. Sections were incubated with rabbit anti-mouse CDKN2A/p16^INK4A primary Ab (ab211542; Abcam) at 1:100 dilution overnight at 4°C. On the following day, sections were then washed with 1× PBS and incubated with Alexa Fluor 647 goat anti-rabbit secondary Ab (21246; Invitrogen) at a 1:500 dilution for 2 h at room temperature. Nuclei were stained with DAPI for 10 min, and images were acquired immediately on a Leica SPE confocal microscope. Five to six different fields were acquired per mouse section, and p16^INK4A-positive cells were counted and expressed as a percentage of total lesional cells.

Human coronary artery specimens with atherosclerotic lesions were selected from individuals enrolled in the CVPath Institute Registry with atherosclerosis, a history of being diagnosed with cancer, or a history of being diagnosed with cancer and treated with radiation therapy. Briefly, the artery segments were fixed in formalin, and 2- to 3-mm segments were embedded in paraffin. Cross sections of 5-µm thick were cut from each of the segments and mounted on slides. Slides were stained with CDKN2A/p16^INK4A (ab54210; Abcam) primary Ab at a 1:8000 dilution for immunohistochemical analyses. Plaque classifications were determined according to our previously published criteria (17). Slides were scanned on an Axio Scan.Z1 slide scanner (Carl Zeiss, Oberkochen, Germany), and image panels were prepared on the HALO image analysis platform version 3.0 (Indica Labs, Corrales, NM).

p16-trimodality reporter (3MR) mice were from UNITY Biotechnology. Ldlr^−/− mice were lethally γ-IR for complete ablation of bone marrow. Radiation was given in two doses, each dose being 4.75 Gy, with 4 h between doses. After the second dose of radiation, we i.v. injected bone marrow cells from the femurs and pelvis of either p16-3MR mice or C57BL/6 wild-type control mice into IR Ldlr^−/− mice. The mice were given antibiotic water (SMZ and TMP; catalog no. NDC 65862-496-47; Aurobindo Pharma) and a 6-wk recovery period. After 6 wk, mice were fed WD for an additional 10 wk to allow for the development of atherosclerosis. At the end of 10 wk, mice were randomly assigned to receive i.p. injections of either vehicle (PBS) or 5mg/kg/mouse ganciclovir (GCV; catalog no. G2536; Sigma-Aldrich) for the next 3 wk while still on WD. Mice were sacrificed at the end of 13 wk, and a necrotic core analysis was carried out on H&E-stained lesional cross sections as above.

To identify and quantify the lipid mediators in atherosclerotic aortas from vehicle- or GCV-treated p16-3MR transplanted Ldlr^−/− mice, a targeted liquid chromatography–tandem mass spectrometry (LC–MS/MS)–based analysis was performed as described previously (18). Aortas were isolated and immediately placed in ice-cold methanol. Deuterium-labeled synthetic standards d8-5S-HETE, d5-LXA₄, d5-RvD2, d4-LTB₄, and d4-PGE₂ (Cayman Chemical) were then added to each sample, and the tissue was minced on ice. Supernatants were collected after centrifugation (13,000 rpm, 10 min, 4°C) and acidified to pH 3.5. The samples were then subjected to solid-phase extraction using ISOLUTE C18 columns (Biotage). Lipid mediators were eluted from the column following addition of methyl formate and were then concentrated under N₂ gas and resuspended in methanol:water (50:50). The samples were analyzed by LC–MS/MS using a Poroshell Reversed-Phase C18 Column (100 mm × 4.6 mm × 2.7 μm; Agilent Technologies)–equipped HPLC system (Shimadzu) coupled to a QTRAP 5500 mass spectrometer (SCIEX) operating in negative ionization mode and using scheduled multiple reaction monitoring. Lipid mediators were identified in the experimental samples by matching retention times to synthetic standards run in parallel. The abundance of each mediator was determined using standard curves generated with synthetic standards for each individual mediator and by accounting for the recovery of the internal deuterium-labeled standards, followed by normalization to total protein content.

Elicited peritoneal macrophages were plated in a 6-well plate (2 × 10⁶ cells per well) and then subjected to the senescence protocol as described above. Supernatants were collected, concentrated 2-fold with 3-kDa Ultra Centrifugal Filters (Amicon) and subjected to a Proteome Profiler Mouse Cytokine Array according to the manufacturer’s instructions (catalog no. ARY006; R&D Systems). Immunoblots were analyzed with ImageJ, and data were expressed as protein levels relative to control.

WD-diet fed Ldlr^−/− mice were sacrificed, and blood was collected in 10% EDTA by retro-orbital bleeding procedure. Immediately after collection, blood was centrifuged at full speed for 30 min at 4°C, and a cholesterol assay (catalog no. 999-02601; Wako Diagnostics) was performed according to manufacturer’s instruction.

For all in vivo studies, mice were randomly assigned to their respective groups. Results are represented as mean ± SEM. Prism (GraphPad, La Jolla, CA) software was used for statistical analysis, and statistical differences were determined using the two-tailed Student t test, one-way ANOVA, or two-way ANOVA with Tukey or Sidak multiple comparison post hoc analysis. Details regarding statistical tests can be found in the figure legends.

To test the impact of radiation on resolution, we first mock or sublethally IR male C57/BL/6 mice with 7 Gy of ionizing radiation. The mice were given a 3-mo recovery period to induce hematopoietic cell senescence (11, 13). Senescence is mediated by changes in the expression of cell cycle regulators, such as p16^INK4A, p19^ARF, or p21 as examples. Consistent with the literature, we observed that bone marrow from IR mice had significantly higher expression of the senescence markers p16^INK4A (Supplemental Fig. 1A), p19^ARF (Supplemental Fig. 1B), and p21 (Supplemental Fig. 1C) compared with controls. Accordingly, we also observed other features of senescence and aging, including a significant increase in long-term HSCs, a decrease in short-term HSC, a strong trend toward increased MMP2s (i.e., erythroid progenitors) (Supplemental Fig. 1D), and a significant decrease in lymphoid-biased MMP4s (Supplemental Fig. 1E). To directly test the role of sublethal radiation-induced senescence on the resolution response, we used the widely known in vivo model of ZymA-induced sterile, self-limited inflammation (19). Control or IR mice were injected with 200 µg of ZymA per mouse, and peritoneal exudates were collected by lavage 4, 24, and 48 h postinjection. Leukocytes were enumerated and PMN were assessed by flow cytometry. The peak PMN response (4 h) is a measure of inflammation, and the time from the peak to when the PMN reach half maximal (e.g., ∼24 h) is a quantitative measure of resolution called the resolution interval or R_i (19). We found that the R_i for the mock (control) mice was 23 h, whereas the R_i for the IR mice was 35 h (Fig. 1A), which suggests an overall delay in resolution by 12 h in the IR mice. We observed that the PMN numbers were significantly higher in the IR group at 24 h (Fig. 1B). Because the timely loss of PMN in tissues is associated with efficient resolution, we next quantified the percent retention of PMN from 4 to 24 h in the IR versus control mice. The IR mice had significantly higher PMN from 4 to 24 h (Fig. 1C), again suggesting an overall delay in tissue resolution in mice that received a sublethal dose of radiation. The modestly reduced influx of neutrophils to the peritoneum at early time points was not due to a loss of bone marrow neutrophils (Supplemental Fig. 1F). Indeed, bone marrow neutrophils in control mice had a significantly higher frequency 24 h after ZymA, followed by a significant decline by 48 h, whereas bone marrow neutrophil frequency in the IR mice did not change throughout the time course (Supplemental Fig. 1F). Moreover, peritoneal macrophage numbers were significantly higher in the control mice at 24 h after ZymA injection compared with IR mice (Supplemental Fig. 1G). To rule out nonspecific effects of IR, we performed a bone marrow transfer (BMT) in which mice were given a lethal dose of γ-radiation followed by replenishment of otherwise healthy bone marrow cells from C57BL/6 mice. Lethal doses of γ-radiation also deplete peritoneal macrophages, which are replenished by bone marrow precursors cells upon BMT (20). We found no significant differences in the resolution interval between BMT and non-IR control mice (Supplemental Fig. 1H). These results suggest that BMT and thus a replenishment of healthy bone marrow did not delay resolution.

We next questioned whether administration of a key proresolving ligand, RvD1, to IR mice would improve resolution end points. Indeed, i.p. injection of RvD1 (300 ng/mouse) given with ZymA significantly decreased PMN numbers at 24 h after ZymA (Fig. 1D). Mechanistically, the loss of PMN from the peritoneum during the resolution phase is, in part, through efferocytosis (21). To determine efferocytosis in vivo, we used flow cytometry to assess free PMN versus those associated with macrophages. We observed a significant increase in the free/associated PMN ratio in the IR mice compared with controls (Fig. 1E), which suggests a defect in efferocytosis. Together, these results provide evidence that sublethal radiation impairs temporal inflammation resolution and deranges efferocytosis in vivo.

Certain peritoneal and elicited macrophages have a proliferative capacity, so we next questioned whether sublethal radiation maladaptively halted their proliferation to promote senescence/SASP. For these experiments, we harvested zymosan-elicited peritoneal macrophages 48 h after ZymA injection. These macrophages were plated, then IR (5 Gy) and cultured in the presence of L cell conditioned media for an additional 3 d. Both control and IR macrophages excluded trypan blue equally (not shown). SC cells are known to acquire a flattened shape in vitro, and IR macrophages exhibited an enlarged and flattened shape compared with control macrophages (Fig. 2A). Importantly, IR macrophages also exhibited several features of cellular senescence, including a significant reduction in proliferation, as determined by Ki67 staining (Fig. 2B, Supplemental Fig. 2A), and a significant increase in p16^INK4A (Fig. 2C) and p21 expression (Fig. 2D). The expression of p19^ARF was not increased in IR macrophages compared with non-IR controls (Fig. 2E). SA-β-gal activity is another known marker of senescence, and we used a quantitative flow cytometric method in which control or IR macrophages were incubated for 1 h with C12-FDG, a fluorescent substrate for β-galactosidase enzyme (14). Using this approach, we observed that the IR macrophages also had a significant increase in SA-β-gal activity compared with control macrophages (Fig. 2F). We also observed that IR macrophages exhibited other features of senescence and SASP, including increased COX-2 levels as determined by intracellular staining by flow cytometry (Fig. 2G). Using metabolic flux analysis, we observed that IR macrophages had significantly higher ECAR after the addition of glucose, consistent with an increase in aerobic glycolysis capacity (Fig. 2H). A representative ECAR tracing is shown on the left, and the quantification after addition of glucose is shown on the right (Fig. 2H). Irradiated macrophages also had enhanced OS, as determined by CellROX staining (Fig. 2I, Supplemental Fig. 2B), and increased PGE₂ levels as determined by ELISA (Fig. 2J). Further, we assessed numerous cytokine/chemokines using a protein array and found soluble ICAM-1 (sICAM-1), CXCL1, CXCL2, and TNF-α were significantly increased in IR macrophages compared with control macrophages (Fig. 2K). A representative immunoblot array as well as quantification of other cytokines/chemokines that did not significantly change between IR and control macrophages are shown in Supplemental Fig. 2C. Together, these results strongly suggest that IR macrophages undergo senescence accompanied with a proinflammatory SASP, and we will refer to these cells as SC-macrophages.

To determine whether RvD1 could reduce senescence/SASP in these cells, we stimulated SC-macrophages with RvD1 (10 nM) 2 d postradiation and cultured them for an additional 24 h. Treatment with RvD1 moderately increased the percentage of Ki67⁺ macrophages (Fig. 2B, Supplemental Fig. 2A) and slightly decreased p16^INK4A expression, which did not reach significance (Fig. 2C). Also, RvD1 did not modulate the expression of p21 (Fig. 2D) or p19^ARF (Fig. 2E), which suggests that RvD1 did not significantly modulate the cell cycle and therefore did not reverse senescence once the macrophages were already committed to this program. We next questioned whether RvD1 modulated the phenotype of the SC cells. Indeed, RvD1 significantly decreased SA-β-gal activity (Fig. 2F), COX-2 levels (Fig. 2G), aerobic glycolysis (Fig. 2H), OS (Fig. 2I, Supplemental Fig. 2B), and PGE₂ (Fig. 2J). Indeed, RvD1 significantly reduced OS as potently as the ROS inhibitor NAC, and the cotreatment of NAC and RvD1 did not further limit OS (Fig. 2I, Supplemental Fig. 2B). Both NAC and RvD1 decreased PGE₂ (Fig. 2J), which suggest that OS may play a role in elevated PGE₂ levels in SC-macrophages. Moreover, RvD1 significantly decreased levels of sICAM-1, CXCL1, M-CSF, MCP-1, CXCL2, and TNF-α in SC-macrophages (Fig. 2K, Supplemental Fig. 2C). Together, these results suggest RvD1 limits the SASP in macrophages.

A key function of macrophages is their ability to clear dead cells, a process called efferocytosis. Previous studies have shown that increased OS and TNF-α limits efferocytosis (22) and that intact ICAM-1 on macrophages promotes efferocytosis (23). Because SC-macrophages have elevated levels of intracellular ROS, increased TNF-α, and elevated sICAM-1, we next questioned whether SC-macrophages have impaired efferocytosis. Indeed, we observed that SC-macrophages have significantly less efferocytosis compared with healthy proliferating controls (Fig. 3A, 3B). Representative images of efferocytosis are shown in (Fig. 3A. Importantly, RvD1 or NAC rescued defective efferocytosis (Fig. 3A, 3B), which suggests that elevated OS in SC-macrophages limits efficient efferocytosis. These results also suggest that RvD1 rescues efferocytosis through limiting OS in SC-macrophages.

We next questioned whether SC-macrophages exhibited a proinflammatory and anti-resolution phenotype in vivo. For these experiments we adoptively transferred either control or SC-macrophages simultaneously with ZymA. Exudates were collected after 4 h, and PMN were enumerated. We observed that the transfer of SC-macrophages had significantly higher PMN numbers at 4 h after ZymA injection compared with controls (Fig. 3C). These results suggest that SC-macrophages drive prolonged inflammation in vivo.

IMR-90 fibroblasts are a well-accepted model to study senescence in vitro (7). Therefore, we also wanted to determine whether this commonly accepted senescence paradigm exhibited similar results as the SC-macrophages. We subjected IMR-90 cells to 10 Gy of radiation and then monitored SASP and senescence markers 10 d after radiation (7). First, we found that, similar to IR SC-macrophages, SC IMR-90 fibroblasts also had a significant decrease in efferocytosis compared with controls (Fig. 4A). Consistent with our macrophage results, we observed that SC IMR-90 cells had significantly higher SA-β-gal activity, which is almost completely abrogated by RvD1 (Fig. 4B). Representative images of SA-β-gal staining (blue color) in IMR-90 cells are shown in (Fig. 4B and reveal that SC cells have increased blue staining compared with controls and that RvD1 significantly limited the blue staining (Fig. 4B). Additionally, RvD1 dramatically reduced PGE₂ levels, almost to that of control cells (Fig. 4C), and significantly decreased aerobic glycolysis based on ECAR values (Fig. 4D) and ATP resulting from glycolysis (Fig. 4E). Lastly, we observed that RvD1 moderately reduced the expression of the senescence genes p16^INK4A (Fig. 4F), p19^ARF (Fig. 4G), and p21 (Fig. 4H). Together, these results suggest that RvD1 limits key components of the SASP in IMR-90 cells.

Impaired resolution programs, such as defective efferocytosis and limited repair/remodeling, are associated with atherosclerosis progression. A consequence of radiation therapy in humans is an increased risk for atherosclerosis (24), and murine models also reveal that radiation exacerbates atherosclerosis (25, 26). In agreement with the literature (25, 26), we also found that sublethal radiation significantly increased the percentage of plaque necrosis per lesion area in Ldlr^−/− mice compared with mock controls (Fig. 5A, 5B). Lesion area, which was quantified using H&E sections as well as Oil Red O staining, were not significantly different between the groups (Supplemental Fig. 3A, 3B). We also found that sublethal radiation significantly decreased the percentage of plaque collagen per lesion area in Ldlr^−/− mice compared with mock controls (Fig. 5C, 5D), which overall suggests that sublethal radiation promotes less “stable” plaques. We next questioned whether treatment with RvD1 could limit plaque necrosis and promote repair in the context of sublethal radiation. For these studies, Ldlr^−/− mice were mock or sublethally IR and then immediately placed on a WD for 12 wk. After 12 wk on WD, mice were randomly assigned to receive vehicle or RvD1 (100 ng/mouse) three times per week for an additional 3 wk while still on the WD. Mice were then sacrificed, and aortic root lesions were interrogated for plaque necrosis and collagen content. Treatment with RvD1 significantly decreased the percentage of lesional necrosis per total lesion area compared with IR mice (Fig. 5A, 5B) and significantly increased plaque collagen content (Fig. 5C, 5D). Representative H&E images of lesional necrosis and fibrous cap thickness are shown in (Fig. 5A and (5C, respectively. There were no differences in lesion area between vehicle- versus RvD1-treated mice that received sublethal radiation (Supplemental Fig. 3A, 3B). Body weight (33.75 ± 0.90 g for IR, 31.45 ± 0.87 g for IR plus RvD1) and plasma cholesterol levels (1009 ± 121.3 mg/dl for IR versus 897.1 ± 119.5 mg/dl for IR plus RvD1) were not different between treatment groups.

We next questioned whether sublethal radiation increased intraplaque p16^INK4A cells. We performed immunofluorescence staining on murine plaques and found that there were significantly more p16^INK4A cells in the IR plaques compared with controls (Fig. 5E, 5F). Moreover, RvD1 treatment reduced the percentage of lesional p16^INK4A cells (Fig. 5E, 5F). Representative immunofluorescence images of lesional p16^INK4A cells are shown in (Fig. 5E and clearly depict reduced p16^INK4A staining (shown as pink) in mice treated with RvD1. These results, together with our in vitro findings in (Figs. 2 and (4, suggest that RvD1 may limit initiation of new p16^INK4A cells or promote their removal in tissues. Together, these results suggest that RvD1 mitigates sublethal radiation-induced atherosclerosis.

Next, we assessed whether radiation leads to increased p16^INK4A cells in human plaques. We obtained human coronary artery specimens with atherosclerotic lesions from individuals enrolled in the CVPath Institute Registry and assessed plaques from patients with atherosclerosis and no history of cancer, plaques from patients who had cancer but no radiation, and plaques from patients who had cancer and radiation treatment. These human plaque sections were stained with p16^INK4A, and representative lesional p16^INK4A images are shown in (Fig. 5G. First, we found immunohistochemical analysis revealed that human atherosclerotic plaques exhibited p16^INK4A cells (Fig. 5G, 5H), which is consistent with the literature (27) and suggests that atherosclerosis and aging promotes the accumulation of SC cells in plaques. Importantly, patients who had received radiation therapy had significantly more p16^INK4A cells in the plaques (Fig. 5G, 5H). Therefore, these results suggest that radiation further promotes the accumulation of SC cells in human plaques. Collectively, these results suggest that SC cells accumulate in advanced plaques and are elevated in the context of sublethal irradiation.

To determine the impact of hematopoietic cell senescence and SPM synthesis in plaques, we took advantage of the p16^INK4A-driven 3MR fusion protein, which contains functional domains of a synthetic Renilla luciferase (LUC), monomeric red fluorescent protein (mRFP), and truncated HSV-1 thymidine kinase (HSV-TK) (Fig. 6A) (28). The HSV-TK allows killing of p16^INK4A-positive cells by GCV, a nucleoside analog that has a high affinity for HSV-TK but low affinity for the cellular TK. We performed a BMT from p16-3MR mice into Ldlr^−/− mice, and after 6 wk of recovery, mice were placed on WD for 10 wk. After 10 wk, mice were randomly assigned to receive vehicle or GCV (5 mg/kg, three times a week) i.p. injections for an additional 3 wk while still on WD (Fig. 6A). Representative H&E images of aortic root cross sections display less necrosis in the GCV-treated plaques, as outlined by dashed lines, compared with vehicle controls (Fig. 6B, left panel). We also found that there was a significant decrease in lesion area (vehicle: 48,693 ± 1700 μm²; GCV: 25,838 ± 3700 μm²) as well as necrosis:lesion area (vehicle: 0.4102 ± 0.02; GCV;: 0.2931 ± 0.04) in GCV-treated mice compared with vehicle. There was no significant change in body weight (vehicle: 35.8 ± 0.75 g; GCV: 34.39 ± 1.01 g), plasma cholesterol (vehicle: 1187 ± 110.8 mg/dl; GCV: 960.8 ± 77.62 mg/dl), or blood glucose levels (vehicle: 183.1 ± 9.80 mg/dl; GCV: 169.9 ± 5.81 mg/dl) between the groups. Importantly, we transplanted C57BL/6 marrow in Ldlr^−/− mice and carried out experiments as in (Fig. 6A. We found that GCV treatment did not alter lesion area (vehicle: 54,152 ± 2229 μm²; GCV: 55,128.16 ± 3467 μm²) or necrotic area (vehicle: 19,125 ± 714.2 μm²; GCV: 19,134 ± 284.8 μm²). A representative H&E image is shown in Supplemental Fig. 3C. Body weight and plasma cholesterol levels were nonsignificantly different between the groups (data not shown). These results suggest that the GCV treatment did not exert any actions on plaque from mice that were transplanted with C57BL/6 bone marrow. Together, these results suggest that hematopoietic SC cells drive atheroprogression.

We next investigated SPMs in aortic plaques from vehicle- or GCV-treated p16-3MR–transplanted Ldlr^−/− mice by targeted LC–MS/MS analysis. We found that 17-HDHA [i.e., a biosynthetic pathway biomarker to RvD1 (Fig. 6C) and LXA₄ (Fig. 6D), which is an SPM that binds and signals through the same G protein–coupled receptor as RvD1], was significantly increased, whereas the proinflammatory lipid mediator LTB₄ (Fig. 6E) was significantly decreased in the GCV-treated plaques compared with vehicle controls. A complete list of identified lipid mediators is shown in Supplemental Table I. These results suggest that removal of p16^INK4A-positive cells during advanced atherosclerosis promotes key SPMs. Collectively, these results suggest a maladaptive role of SC hematopoietic cells on inflammation-resolution responses in atherosclerosis progression.

The findings of this study provide a previously unappreciated link between senescence and inflammation-resolution programs that may provide a new framework to approach treatment strategies in contexts in which SC cells accumulate. Our human plaque data suggests atherosclerosis is associated with p16^INK4A cells, which is consistent with the literature (27). Moreover, we also found that radiation further increases p16^INK4A cells in human plaques. A long-term consequence of mediastinal radiation is coronary artery disease. Radiation-induced vascular injury is thought to be a major factor that drives long-term occlusive atherosclerotic disease, but there are likely other aspects as well (29). Mediastinal radiation can impact the bone marrow, which is not surprising, given that sterna and vertebrae are major sites of hematopoiesis in adults (30). Therefore, it is also possible that even focal thoracic radiation may impact the bone marrow to progress atherosclerosis. Nevertheless, more detailed human studies on bone marrow changes, radiation, and atherosclerosis need to be investigated.

Additionally, atherosclerotic plaques have long been associated with the presence of SC cells, namely endothelial and smooth muscle cells (31, 32). Recent data suggest that plaque macrophages may also become SC (33), and our data in this study strongly suggest that SC hematopoietic cells and macrophages contribute to the progression of atherosclerosis in part but not limited to a defect in SPM synthesis. Along these lines, macrophages are known to proliferate in progressing atherosclerotic plaques, but their proliferation appears to be restricted to a few cycles (34, 35). How senescence is involved in the eventual arrest of proliferation of plaque macrophages remains to be explored, but our results provide evidence that removal of SC hematopoietic cells during advanced atherosclerosis dampens atheroprogression. These findings, along with others, suggest bone marrow cells and macrophages are critical effectors of plaque progression (36, 37).

Along these lines, sublethal radiation in mice mimics several features of hematopoietic aging (38). Indeed, a major risk factor for atherosclerosis is age, and how aging impacts the development and progression of atherosclerosis is a critically underexplored arena (39). A previous study showed that BMT from aged mice into young Ldlr^−/− mice produced larger plaques than Ldlr^−/− mice who received young marrow (40). These findings suggest that the bone marrow from aged mice promotes atherosclerosis, and a deeper understanding as to which cell types and cues within the aged bone marrow drives atheroprogression are of interest.

Previous reports suggested that macrophages reveal features of senescence, but context and function remained underdeveloped (33, 41, 42). Macrophages are highly responsive to their local tissue microenvironment and can exhibit proinflammatory, profibrotic, or even proresolving and proregenerative functions. A recent study observed that resident peritoneal macrophages from p16^INK4A-activated mice exhibited several features of senescence and had increased uptake of zymosan particles (42). In the context of progressing atherosclerotic lesions, macrophages also exhibited features of senescence, albeit function was not determined (33). Our work in this study suggests that radiation-induced senescence of macrophages leads to an impairment in efferocytosis that was associated with increased levels of ROS, sICAM-1, and TNF-α. Increased ROS can lead to activation of ADAM17 on macrophages (43), which is an enzyme that facilitates the release of activated TNF-α and sICAM-1 (44). ADAM17 also promotes the cleavage of MerTK (43), which is a critical efferocytosis receptor on macrophages. We previously found that released factors from SC cells cleave MerTK to limit efferocytosis on otherwise heathy macrophages (15). Therefore, it is possible that released factors from radiation-induced SC-macrophages may act in an autocrine manner to limit efferocytosis. Nevertheless, context is critical, and the manner in which a cell is driven toward senescence may drive its ultimate phenotype/function.

Along these lines, in the context of aging and advanced atherosclerosis, the accumulation of SC cells and their resulting SASP is maladaptive (45, 46). However, senescence can also be protective because this is a program that can limit cancer, promote cutaneous wound healing and facilitate patterning during embryogenesis (8, 28). Recent reports suggest that SC cells in arthritic joints can facilitate healing (47). Several questions remain in our understanding as to what drives SC cells toward a tissue-reparative versus a tissue-destructive phenotype, and a deeper understanding of SC cell markers and animal models is needed.

Last, we offer a proof-of-concept that RvD1 limits sublethal radiation-induced accumulation of SC cells in atherosclerosis. Currently, there are limited options to quell the SASP or remove Sc cells from tissues. Senotherapeutics are emerging as intriguing new strategies to limit SC cells in humans (48). Senolytics, for example, inactivate prosurvival mechanisms of SC cells to promote apoptosis and subsequent clearance. Because efferocytosis is impaired in aging and atherosclerosis (33, 49), an increase in apoptotic cells over time may limit the benefit of senolytics, as these apoptotic cells can undergo secondary necrosis. The work presented in this study provides an entirely new strategy to limit the most deleterious aspect of SC cells (i.e., the SASP) and suggests that RvD1 may act as a novel senotherapeutic in the context of age-related pathologies. SPMs in general are not immunosuppressive and act to promote tissue repair and regeneration (1) and may be a promising strategy to limit SC cells in advanced atherosclerosis.

We thank Justin Heinz and Nicholas Rymut for technical expertise.

The authors have no financial conflicts of interest.