Main

To explore the impact of stress on the human immune system, we analyzed blood samples from 29 medical residents working on a tertiary hospital intensive care unit (ICU), a challenging, fast-paced work environment that frequently includes the responsibility of life-or-death decisions. Compared to when off duty, residents working on the ICU reported an increased stress perception, which we assessed with Cohen's Perceived Stress Scale5 (Fig. 1a). Visual analog scales6 documented a higher stress intensity and frequency while working on the ICU (Supplementary Fig. 1a). When comparing samples taken during work to samples taken off duty, we observed an increase in blood leukocytes (Fig. 1b), with higher numbers of neutrophils, monocytes and lymphocytes, after 1 week of intensive care rotation (Supplementary Fig. 1b). The relative frequencies of monocyte subsets did not change (Supplementary Fig. 1c).

Figure 1: Chronic stress increases proliferation of HSPCs in the bone marrow.
figure 1

(a) Cohen's Perceived Stress Scale score in medical ICU residents either off duty or on ICU duty (n = 10 off duty, n = 15 on ICU, Student's t-test). (b) Blood leukocytes in residents (n = 29, Wilcoxon test). (c,d) Leukocytes in mouse blood (c) and bone marrow (d) after 3 weeks of stress or with no stress (n = 9 per group, Student's t-test). Black bars, nonstressed control mice; gray bars, stressed mice. NS, not significant. (e) Gating for LSKs and HSCs isolated from stressed and nonstressed mice. Numbers next to gates represent population frequencies (%). Cell cycle parameters of HSCs, as assessed by Ki-67 and DAPI staining, were quantitated after 3 weeks of stress (n = 5 per group, Mann-Whitney U test). (f) Bone marrow (BM) colony-forming unit (CFU) assay after 1 week of stress (n = 5 per group, Mann-Whitney U test). (g) Bone marrow HSC and LSK numbers after 3 weeks of stress (n = 10 per group, Student's t-test). (h) Long-term competitive repopulation assay using limiting dilutions of whole bone marrow from stressed or nonstressed (control, Ctrl) mice (Poisson statistics for LT-HSC frequencies, P = 0.2, two-tailed Student's t-test). Initially, n = 10 per group, with 4 groups with different cell dilutions for each condition, stressed versus nonstressed. Precise mouse numbers are given in Supplementary Table 2. Open circles show fraction of nonresponders in the test population (stress, red; control, black) for every given concentration. Inverted triangles indicate mice with blood chimerism above the cutoff value. Straight lines represent fit for estimation of population frequencies, and dashed lines show 95% confidence intervals). (i) Percentage donor engraftment after competitive reconstitution with 2 × 106 bone marrow cells from stressed or nonstressed controls co-injected with equal numbers of naive competitor cells, as determined by assessment of blood chimerism (n = 10 mice per group, one-way analysis of variance (ANOVA)). Mean ± s.e.m., *P < 0.05.

Source data

Full size image

To test the hypothesis that stress-induced leukocytosis results from increased leukocyte production, we exposed wild-type mice to chronic variable stressors validated by behavioral neuroscience studies (Supplementary Table 1)7,8,9. Compared to nonstressed controls, stressed mice had increased numbers of leukocytes, neutrophils and monocytes in blood (Fig. 1c), which is consistent with our observations in humans. These cells were also more numerous in the bone marrow (Fig. 1d). We next investigated the influence of chronic stress on blood cell production in the bone marrow and detected increased cycling of Lin−Sca-1+c-Kit+CD150+CD48− hematopoietic stem cells10 (HSCs, Fig. 1e), which also incorporated increased amounts of BrdU (Supplementary Fig. 2a). To assess HSC quiescence, we performed a pulse-chase BrdU label retention experiment. BrdU exposure in drinking water for 2 weeks led to >90% labeling of HSCs, as previously reported11. After completion of the labeling phase, we stressed mice for 3 weeks, which accelerated HSC BrdU washout when compared to nonstressed controls (Supplementary Fig. 2b). Bone marrow harvested from stressed mice had augmented colony-forming capacity, indicative of increased progenitor cell proliferation (Fig. 1f). Enhanced proliferation resulted in higher bone marrow numbers of HSCs, Lin−Sca-1+c-Kit+ progenitors (LSKs, Fig. 1g), granulocyte macrophage progenitors, macrophage dendritic cell progenitors (Supplementary Fig. 3a) and common lymphoid progenitors (Supplementary Fig. 3b).

Although CD150+CD48− SLAM staining phenotypically quantitates HSCs, only a fraction of the cells in this gate are functional long-term HSCs (LT-HSCs)10. A competitive repopulation assay12 comparing limiting bone marrow dilutions obtained from stressed and nonstressed donors indicated that the frequency of LT-HSCs did not significantly change in stressed mice (Fig. 1h and Supplementary Table 2). When viewed together with the increased bone marrow cellularity in stressed mice (1.64 × 107 versus 2.46 × 107 per femur, P < 0.0001, n = 14 control mice, n = 18 stressed mice), our results suggest that stress neither increases nor exhausts LT-HSCs. An unchanged number of LT-HSCs in stressed mice is further supported by comparable blood chimerism 16 weeks after transfer of 2 × 106 bone marrow cells from either stressed or nonstressed donors with equal numbers of naive competitor cells into lethally irradiated recipients (Fig. 1i). In contrast, infection and interferons increase HSC proliferation while exhausting LT-HSCs13 or impairing their engraftment14, possibly because these stimuli are more severe than chronic stress. Of note, interferon protein levels were unchanged in the bone marrow of stressed mice (Supplementary Fig. 4a).

To test whether leukocyte production or redistribution was the source of stress-induced leukocytosis, we performed a 5-fluorouracil (5-FU) challenge, a treatment that kills actively cycling progenitor cells. 5-FU completely abolished stress-induced leukocytosis, suggesting that the observed leukocytosis is caused by progenitor proliferation. Notably, stressed mice had an enhanced leukocyte rebound on day 14 after 5-FU injection, which was probably caused by increased cycling of hematopoietic progenitors (Supplementary Fig. 4b). Serial intravital microscopy15 in the calvarium of mice that had undergone adoptive transfer of 25,000 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate (DiD)-labeled LSKs detected accelerated dilution of the membrane dye in mice exposed to 7 d of stress (Fig. 2a), indicating accentuated cell proliferation. Flow cytometry confirmed the accelerated membrane dye dilution after stress exposure (Fig. 2b). Taken together, these data indicate that chronic stress activates HSCs, which increase proliferation and differentiate into downstream progenitors.

Figure 2: Stress leads to increased bone marrow hematopoietic progenitor cell proliferation.
figure 2

(a) Intravital microscopy of the mouse calvarium after adoptive transfer of DiD-labeled LSKs (white arrows) before (day 0) and 7 d after stress (n = 5 mice per group, Mann-Whitney U test). Dotted lines outline bone. Scale bar, 50 μm. Each dot in the graphs represents the target-to-background ratio of DiD+ cells either before (top) or after stress (bottom). (b) DiD fluorescence after adoptive transfer of DiD+GFP+ LSKs in nonstressed control (Ctrl) or stressed mice (n = 5 per group). The bar graph quantitates DiD fluorescence in GFP+ LSKs (Mann-Whitney U test). Mean ± s.e.m., *P < 0.05. NS, not significant.

Source data

Full size image

Noradrenaline is a prototypical stress hormone that also regulates circadian progenitor cell migration16 and proliferation17,18. We wondered whether heightened hematopoietic system activity during stress could be related to this catecholamine. Indeed, noradrenaline levels increased in the bone marrow of stressed mice compared to nonstressed controls (Fig. 3a). Immunoreactive staining for tyrosine hydroxylase, a rate-limiting enzyme for noradrenaline synthesis19, rose in sympathetic nerve fibers surrounding blood vessels in the bone marrow (Fig. 3b). This rise was associated with a sharp decrease in CXCL12 mRNA and protein within whole bone marrow (Fig. 3c,d), in line with the role of noradrenaline in regulating CXCL12 synthesis16. Conditional deletion of tyrosine hydroxylase–containing cells in cross-bred iDTR TH-Cre mice18 preserved CXCL12 levels after stress exposure and blunted the bone marrow stress response when compared to stressed wild-type mice (Supplementary Fig. 5).

Figure 3: Stress-induced sympathetic nervous system signaling regulates the proliferation of bone marrow HSCs via CXCL12.
figure 3

(a) Noradrenaline levels, as assessed by ELISA, after 3 weeks of stress (n = 8 per group, Student's t-test). (b) Immunoreactive staining for tyrosine hydroxylase (TH) in bone marrow. Scale bar, 10 μm. The bar graph shows the percentage of TH-positive area per field of view (n = 5 mice per group, Mann-Whitney U test). (c) CXCL12 mRNA in bone marrow of nonstressed, stressed and stressed mice treated with a β3-selective receptor blocker (n = 10 per group, one-way ANOVA). (d) CXCL12 protein in bone marrow (n = 7 per group, one-way ANOVA). (e) Gating of LSKs (left) and HSCs (right). (f,g) Quantitation of BrdU incorporation and absolute cell numbers (f) and effects of a β3-adrenergic receptor blocker on the indicated populations of blood leukocytes (g) (n = 5 per group, Mann-Whitney U test). Mean ± s.e.m., *P < 0.05. NS, not significant.

Source data

Full size image

In the hematopoietic niche, CXCL12 derives from mesenchymal stem cells, osteoblasts and endothelial cells20,21,22. Its primary functions include inhibiting hematopoietic stem and progenitor cell (HSPC) proliferation and migration; it also retains neutrophils in the bone marrow23. CXCL12-deficient mice24 and mice that lack the chemokine's cognate receptor (CXCR4)25 show increased HSC cycling and progenitor pool expansion and increased neutrophil release into circulation compared to wild-type mice. Linking the autonomic nervous system and leukocyte trafficking, the β3-adrenergic receptor expressed on niche cells regulates CXCL12 release16. Among relevant niche cells, we found that mesenchymal stem cells express the highest level of the β3 receptor (Supplementary Fig. 6). We therefore investigated whether chronic stress acts on hematopoiesis via the β3-adrenergic receptor. Indeed, mice with a genetic lack of the receptor were protected from stress, as CXCL12 expression and HSC cycling were similar in stressed and nonstressed Adrb3−/− mice (Supplementary Fig. 7). Moreover, treatment of stressed wild-type mice with the β3-selective receptor blocker SR 59230A restored CXCL12 mRNA and protein levels (Fig. 3c,d), decreased BrdU incorporation into HSCs and reduced HSPC numbers in the bone marrow (Fig. 3e,f). As a consequence, downstream granulocyte macrophage progenitor and macrophage dendritic cell progenitor numbers fell (Supplementary Fig. 8a), resulting in lower levels of neutrophils and Ly6Chigh monocytes in circulation (Fig. 3g). In contrast, treatment with a β2 receptor blocker failed to protect the bone marrow against stress (Supplementary Fig. 8b,c).

Atherosclerosis is a chronic inflammatory disease driven by hyperlipidemia26,27,28. Atherosclerotic plaques consist of cholesterol deposits and a leukocyte infiltrate that is dominated by innate immune cells29. Inflammatory monocyte and macrophage infiltration may lead to plaque rupture, myocardial infarction and stroke30,31, and higher blood levels of monocytes and neutrophils correlate with increased mortality29,32. Proteases released from inflammatory leukocytes weaken the fibrous cap and favor plaque disruption that permits contact between the plaque's necrotic core and clotting factors in the bloodstream, inciting local thrombosis and thereby jeopardizing oxygen supply to the heart and brain30,33. A number of risk factors contribute to inflammatory complications of atherosclerosis. Chronic stress is well recognized among them4,34,35,36,37. Yet, the mechanisms that link chronic stress to higher cardiovascular event rates are incompletely understood.

We decided to test the hypothesis that chronic stress acts on the bone marrow via sympathetic nervous system activity to increase inflammatory leukocyte supply to atherosclerotic lesions. Exposure to 6 weeks of stress enhanced hematopoietic system activity in atherosclerosis-prone Apoe−/− mice, as indicated by increased BrdU incorporation into HSCs and higher numbers of granulocyte macrophage progenitors and macrophage dendritic cell progenitors in the femur (Supplementary Fig. 9a,b), whereas body weight and lipid levels were unaffected (Supplementary Fig. 9c,d). Plaque protease levels increased, as assessed by fluorescence molecular tomography–computed tomography (FMT-CT) imaging (Fig. 4a). Innate immune cells are largely responsible for protease production in atherosclerotic plaque29,30. Accordingly, in stressed Apoe−/− mice, we detected higher CD11b+ myeloid cell and neutrophil content in plaque by histology (Fig. 4b) and increased numbers of neutrophils, monocytes and macrophages in whole aortas by flow cytometry (Fig. 4c). Neutrophils may aid monocyte entry into plaque38 but can also have inflammatory functions themselves39. The aortic arches of Apoe−/− mice exhibited an inflammatory cytokine expression profile after stress (Fig. 4d), including increased expression of myeloperoxidase, a pro-oxidant enzyme abundant in neutrophils and inflammatory monocytes.

Figure 4: Chronic stress increases inflammation in mouse atherosclerotic plaques.
figure 4

(a) Protease activity in aortic roots of Apoe−/− mice measured by FMT-CT after 6 weeks of stress. Dashed circles indicate the aortic root (n = 5 per group, Mann-Whitney U test). (b) Immunoreactive staining of aortic roots for CD11b and Ly6G. Bar graphs show percentage of positive area per root (n = 9 control, n = 10 stressed mice, Student's t-test). Scale bar, 200 μm. (c) Gating and quantification of aortic myeloid cells. Top plots, neutrophils (top gates), monocytes and macrophages (bottom gates). Bottom plots, macrophages (top left quadrants), Ly6Chigh monocytes (bottom right quadrants) (n = 10 per group, Student's t-test). (d) Quantitative real-time PCR for inflammatory gene mRNAs in aortas (n = 10 control, n = 9 stressed mice, Student's t-test). MPO, myeloperoxidase; IL, interleukin; MMP, metalloproteinase. (e) Masson trichrome staining (n = 9 per group, Student's t-test). Scale bars, 50 μm; arrows point at the fibrous cap covering the necrotic core. Bar graphs show fibrous cap thickness and necrotic core area. Mean ± s.e.m., *P < 0.05. NS, not significant.

Source data

Full size image

Whereas aortic root plaque size did not change in stressed versus nonstressed Apoe−/− mice (Supplementary Fig. 10), the inflammatory milieu in the stressed mice led to thinner fibrous caps and larger necrotic plaque cores (Fig. 4e), hallmarks of rupture-prone lesions in patients with acute myocardial infarction or stroke29,33. Heightened recruitment of leukocytes into plaques may also result from enhanced action of adhesion molecules; however, the mRNA and protein levels of the adhesion molecules, with the exception of E-selectin, were unchanged in stressed Apoe−/− mice (Supplementary Fig. 11a,b). When we adoptively transferred equal numbers of GFP+ myeloid cells into stressed and nonstressed Apoe−/− mice, we detected similar recruitment of these cells to plaques (Supplementary Fig. 11c), indicating that stress acts through increased systemic leukocyte supply by the hematopoietic system rather than locally increased cell recruitment. Hypertension, a known risk factor for atherosclerosis, occurred only during stress exposure (Supplementary Fig. 12a) and therefore probably did not play a dominant role in the observed disease progression. Blood pressure, cholesterol and corticosterone levels were unchanged in stressed Apoe−/− mice that received a β3-adrenergic receptor blocker (Supplementary Fig. 12b,c); however, this treatment reduced the number of neutrophils, inflammatory monocytes and macrophages in plaque, providing a direct link between hematopoietic progenitor activity and atherosclerotic plaque inflammation (Supplementary Fig. 13).

In summary, we report how chronic stress interferes with hematopoiesis and describe interactions between the central nervous system, immunity and atherosclerosis. In mice exposed to stress, increased sympathetic nervous system activity decreased CXCL12 expression in the hematopoietic stem cell niche, accelerated HSC proliferation and enhanced neutrophil and monocyte production. These events caused extensive release of inflammatory leukocytes into the circulation and promoted plaque inflammation. Administration of a β3-adrenergic receptor blocker limited disease progression, supporting the notion that sympathetic nervous system signaling via this receptor and targeting of the CXCL12-CXCR4 interaction in the bone marrow should be explored as potential therapeutic avenues.

The data obtained in mice parallel our observations in ICU residents; however, we interpret this association with care, as the nature and the timing of the stress differed. Further, we were unable to investigate HSC activity in stressed humans. Taken together, these data provide further evidence of the hematopoietic system's role in cardiovascular disease29,40 and elucidate a direct biological link between chronic variable stress and chronic inflammation, a general concept with implications beyond atherosclerosis.

Methods

Clinical study.

The clinical study titled 'Effects of Socioenvironmental Stress on the Human Hematopoietic System' was an open, monocenter, single-arm study that enrolled medical residents working on the intensive care unit at University Hospital, Freiburg, Germany. This study was registered with the German Registry for Clinical Studies (DRKS00004821) and was approved by the Ethics Committee of Albert-Ludwigs-University Freiburg, Germany (No. 52/13). All residents working on the ICU were considered eligible to participate in the study. Exclusion criteria were smoking, any acute or chronic illness, regular intake of medication or failure to consent. Twenty-nine volunteers (23 male, 6 female, mean age 33.7 ± 0.8 years) were enrolled after signing the informed consent form. Residents gave two blood samples (baseline and stress). The off-duty sample (baseline) was collected after 10 ± 0.9 consecutive days off duty. The on-duty sample (stress) was collected after 7 ± 0.3 consecutive days of ICU duty. A subcohort of participants completed the Perceived Stress Scale 10-item inventory5 before starting to work on the ICU (baseline), as well as after several weeks on duty (stress). Short-term perception for stress frequency and intensity was measured with visual analog scales (scale 0–10)6, which each participant completed at the time of the blood sampling. The mean circadian time difference between the baseline and the stress sample was 20 ± 15.9 min. Blood samples were analyzed in a blinded fashion at the routine clinical laboratory of the University Hospital, Freiburg, Germany.

Mice.

We used C57BL/6, CD45.1 (B6.SJL-Ptprca Pepcb/BoyJ), UBC-GFP (C57BL/6-Tg(UBC-GFP)30Scha/J), Apoe−/− (B6.129P2-Apoetm1Unc/J), TH-Cre (B6.Cg-Tg(Th-Cre)1Tmd/J) and iDTR (C57BL/6-Gt(ROSA)26Sortm1(HBEGF)Awai/J) mice, all female and 10–12 weeks of age (Jackson Laboratories, Bar Harbor, ME). Adrb3−/− mice16 were donated by P. Frenette (Albert Einstein College of Medicine, New York, NY, USA) and B. Lowell (Beth Israel Deaconess Medical Center, Boston, MA, USA). Nestin-GFP mice41 were a gift from G. Enikolopov (Cold Spring Harbor Laboratory, NY). All procedures were approved by the Subcommittee on Animal Research Care at Massachusetts General Hospital. For each experiment, age-matched female littermates were randomly allocated to study groups. Animal studies were performed without blinding of the investigator.

Stress procedures.

Mice were exposed to socioenvironmental stressors7,8,9 for one or three weeks in C57BL/6 mice or six weeks in Apoe−/− mice. Stress procedures were performed between 7 a.m. and 6 p.m. The following stressors were applied. For cage tilt, the cage was tilted at a 45° angle and kept in this position for six hours. For isolation, mice were individually housed in an area one-quarter of the original cage size (12 cm × 8 cm) for four hours, followed by crowding, during which 10 animals were housed in one cage for two hours. Mice were monitored during the crowding procedure, and 'fighters' were separated. For damp bedding, water was added to the cage to moisten the bedding without generating large pools. Mice were kept for six hours with damp bedding. For rapid light-dark changes, using an automatic timer, the light was switched with an interval of seven minutes for two hours. For overnight illumination, mice were housed in a separate room with illumination from 7 p.m. to 7 a.m. All stressors were randomly shuffled in consecutive weeks. Efficacy of the chronic stress procedures was confirmed by measurement of blood corticosterone levels (Supplementary Fig. 12c).

Lethal irradiation.

Mice were irradiated using a split dose of 2 × 600 cGy with an interval of 3 h between doses. Animals were irradiated 12 h before bone marrow reconstitution.

Bone marrow reconstitution assays.

For competitive bone marrow repopulation assays42, we co-transferred 2 × 106 whole bone marrow cells from CD45.1 mice after three weeks of stress or from nonstressed controls together with equal cell numbers of CD45.2 competitor cells from nonstressed wild-type mice into lethally irradiated UBC-GFP CD45.2 mice. Engraftment was assessed by comparing blood leukocyte chimerism for CD45.1 cells between groups after 2, 3 and 4 months. For limiting dilution experiments42, donor doses of 1.5 × 104, 6 × 104, 12.5 × 104 or 5 × 105 whole bone marrow cells from CD45.1 mice after three weeks of stress or from nonstressed controls were co-transferred with 5 × 105 CD45.2 competitor cells into lethally irradiated CD45.2 recipients. Engraftment was assessed after four months as at least >0.1% multilineage blood chimerism for B lymphocytes, T lymphocytes and myeloid lineage cells derived from donor bone marrow. Poisson's statistic was calculated using L-calc software (Stemcell Technologies) and ELDA software43. Bone marrow of two mice was pooled for each cell population.

Treatment with adrenergic receptor antagonists.

To inhibit β3-adrenergic signaling, a specific antagonist for the β3-adrenergic receptor (SR 59230A, Sigma-Aldrich) was injected at 5 mg/kg body weight i.p. twice per day44. For inhibition of β2-adrenergic signaling, ICI118,551 hydrochloride (Sigma-Aldrich) was injected daily at a dose of 1 mg/kg body weight i.p. (ref. 18) for three weeks. The control groups received saline injections.

Depletion of sympathetic nerve fibers.

TH-Cre mice were cross-bred with iDTR mice. 10–12 week old female TH-iDTR mice were intraperitoneally injected with 0.1 μg/kg body weight diphtheria toxin (DT) on day 0 and day 3 after initiation of stress procedures18. Age-matched littermates (TH-Cre, iDTR or WT) that were also stressed and injected with DT served as controls.

5-Fluorouracil challenge.

Nonstressed mice and mice that had been stressed for three weeks were injected intravenously with 150 mg/kg body weight 5-FU (Sigma)45 on day 0. Mice were then followed over the course of 21 days, and the absolute number of blood leukocytes was measured after 7, 14 and 21 days. Stress exposure continued for the remaining 3 weeks after 5-FU exposure.

Tissue processing.

Flushed bone marrow was passed through a 40-μm cell strainer and collected in PBS containing 0.5% BSA and 1% FBS (FACS buffer). Aortas were excised, minced and digested in collagenase I (450 U/ml), collagenase XI (125 U/ml), DNase I (60 U/ml) and hyaluronidase (60 U/ml) (all Sigma-Aldrich) at 37 °C at 750 r.p.m. for 1 h. For sorting niche cells, bones were harvested from nestin-GFP mice. Bone marrow endothelial cells (ECs) and mesenchymal stem cells (MSCs) were obtained by flushing out bone marrow, which was then digested in 10 mg/ml collagenase type IV (Worthington) and 20 U/ml DNase I (Sigma)46. For obtaining bone osteoblastic lineage cells, we crushed bones, washed off residual bone marrow cells three times and then digested and incubated the bone fragments47,48.

Flow cytometry.

For myeloid cells, cells were first stained with mouse hematopoietic lineage markers (1:600 dilution for all antibodies) including phycoerythrin (PE) anti-mouse antibodies directed against B220 (BD Bioscience, clone RA3-6B2), CD90 (BD Bioscience, clone 53-2.1), CD49b (BD Bioscience, clone DX5), NK1.1 (BD Bioscience, clone PK136) and Ter-119 (BD Bioscience, clone TER-119). This was followed by a second staining for CD45.2 (BD Bioscience, clone 104, 1:300), CD11b (BD Bioscience, clone M1/70, 1:600), CD115 (eBioscience, clone M1/70, 1:600), Ly6G (BD Bioscience, clone 1A8, 1:600), CD11c (eBioscience, clone HL3, 1:600), F4/80 (Biolegend, clone BM8, 1:600) and Ly6C (BD Bioscience, clone AL-21, 1:600). Neutrophils were identified as (CD90/B220/CD49b/NK1.1/Ter119)low(CD45.2/CD11b)highCD115lowLy6Ghigh. Monocytes were identified as (CD90/B220/CD49b/NK1.1/Ter119)lowCD11bhigh(F4/80/CD11c)lowLy-6Chigh/low or (CD45.2/CD11b)highLy6GlowCD115highLy-6Chigh/low. Macrophages were identified as (CD90/B220/CD49b/NK1.1/Ter119)lowCD11bhighLy6Clow/intLy6GlowF4/80high. For hematopoietic progenitor staining, we first incubated cells with biotin-conjugated anti-mouse antibodies (1:600 dilution for all antibodies) directed against B220 (eBioscience, clone RA3-6B2), CD11b (eBioscience, clone M1/70), CD11c (eBioscience, clone N418), NK1.1 (eBioscience, clone PK136), TER-119 (eBioscience, clone TER-119), Gr-1 (eBioscience, clone RB6-8C5), CD8a (eBioscience, clone 53-6.7), CD4 (eBioscience, clone GK1.5) and IL7Rα (eBioscience, clone A7R34) followed by pacific orange–conjugated streptavidin anti-biotin antibody. Then cells were stained with antibodies directed against c-Kit (BD Bioscience, clone 2B8, 1:600), Sca-1 (eBioscience, clone D7, 1:600), SLAM markers10 CD48 (eBioscience, clone HM48-1, 1:300) and CD150 (Biolegend, clone TC15-12F12.2, 1:300), CD34 (BD Bioscience, clone RAM34, 1:100), CD16/32 (BD Bioscience, clone 2.4G2, 1:600) and CD115 (eBioscience, clone AFS98, 1:600). LSKs were identified as (B220 CD11b CD11c NK1.1 Ter-119 Ly6G CD8a CD4 IL7Rα)lowc-KithighSca-1high. HSCs were identified as (B220 CD11b CD11c NK1.1 Ter-119 Ly6G CD8a CD4 IL7Rα)lowc-KithighSca-1highCD48lowCD150high. Granulocyte macrophage progenitors were defined as (B220 CD11b CD11c NK1.1 Ter-119 Ly6G CD8a CD4 IL7Rα)lowc-KithighSca-1low(CD34/CD16/32)highCD115int/low. Macrophage dendritic cell progenitors were defined as (B220 CD11b CD11c NK1.1 Ter-119 Ly6G CD8a CD4 IL7Rα)lowc-Kitint/highSca-1low(CD34/CD16/32)highCD115high. Common lymphoid progenitors were identified as (B220 CD11b CD11c NK1.1 Ter-119 Ly6G CD8a CD4)lowc-KitintSca-1intIL7Rαhigh. For staining endothelial cells, we used ICAM-1 (Biolegend, clone Yn1/1.7.4, 1:300), ICAM-2 (Biolegend, clone 3C4, 1:300), VCAM-1 (Biolegend, clone 429, 1:300), E-selectin (CD62E) (BD Bioscience, clone 10E9.6, 1:100), P-selectin (CD62P) (BD Bioscience, clone RB40.34, 1:100), CD31 (Biolegend, clone 390, 1:600), CD107a (LAMP-1) (Biolegend, clone 1D4B, 1:600) and CD45.2 (Biolegend, clone 104, 1:300). Streptavidin–pacific orange was used to label biotinylated antibodies. Endothelial cells were identified as CD45.2low, CD31high and CD107aintermed/high. For analysis of human monocyte subsets, cells were stained for HLA-DR (Biolegend, clone L243, 1:600), CD16 (Biolegend, clone 3G8, 1:600) and CD14 (Biolegend, clone HCD14, 1:600) after red blood cell lysis (RBC Lysis buffer, Biolegend). Monocytes were identified using forward and side scatter as well as HLA-DR. Within this population, frequencies of monocyte subsets CD14high, CD16high and CD14high/CD16high were quantified.

BrdU experiments.

For BrdU pulse experiments, we used APC/FITC BrdU flow kits (BD Bioscience). One mg BrdU was injected i.p. 24 h before organ harvest. BrdU staining was performed according to the manufacturer's protocol. For BrdU application over 7 days, osmotic micropumps (Alzet) filled with 18mg BrdU were implanted. For the BrdU label-retaining pulse chase assay, BrdU was added to drinking water (1 mg/ml) for 17 days11.

Cell cycle analysis.

After surface staining, intracellular staining was performed according to eBioscience's protocol: cells were fixed and permeabilized using the Foxp3/Transcription Factor Staining Buffer Set (eBioscience) and then stained for the nuclear antigen Ki-67 (eBioscience, clone SolA15). Cell cycle status was determined using 4,6-diamidino-2-phenylindole (DAPI, FxCycle Violet Stain, Life Technologies).

Cell sorting.

To isolate HSPCs, we used MACS depletion columns (Miltenyi) after incubation with a cocktail of biotin-labeled antibodies (as described in the flow cytometry section) followed by incubation with streptavidin-coated microbeads (Miltenyi). Next, cells were stained with c-Kit and Sca-1, and LSKs were FACS-sorted using a FACSAria II cell sorter (BD Biosystems). To purify niche cells from hematopoietic cells, we used MACS depletion columns after incubation with a cocktail of biotin-labeled antibodies as above followed by incubation with streptavidin-coated microbeads. Cells were then stained with CD45.2, Sca-1, CD31 and CD51 (Biolegend, clone RMV-7, 1:100). Endothelial cells were identified as LinlowCD45lowSca-1highCD31high. Bone marrow MSCs were identified as LinlowCD45lowCD31lowSca-1high/intermediate and GFP+. Osteoblasts were LinlowCD45lowSca-1lowCD31lowCD51high. For adoptive transfer of GFP+ neutrophils and Ly6Chigh monocytes, bone marrow cells were collected from UBC-GFP mice for purification of neutrophils and monocytes using MACS depletion columns after incubation with a cocktail of PE–labeled antibodies including B220, CD90, CD49b, NK1.1 and Ter-119 followed by an incubation with PE-coated microbeads. Aortic endothelial cells were identified as CD45.2lowCD31highCD107aint/high and FACS-sorted using a FACSAria II cell sorter.

Adoptive transfer.

We injected 2 × 106 neutrophils together with 2 × 106 Ly6Chigh monocytes intravenously into nonstressed and stressed Apoe−/− mice (the mice were stressed for 6 weeks, and the cells were injected 2 days before the end of the 6 weeks). Aortas were harvested 48 h later. The number of CD11bhighGFP+ cells within the aorta was quantified using flow cytometry.

Histology.

Aortic roots were harvested and embedded to produce 6-μm sections that were stained using an anti-CD11b (BD Biosciences, clone M1/70, 1:15 dilution) or anti-Ly6G (Biolegend, clone 1A8, 1:25 dilution) antibody followed with a biotinylated secondary antibody. For color development, we used the VECTA STAIN ABC kit (Vector Laboratories, Inc.) and AEC substrate (DakoCytomation). Necrotic core and fibrous cap thickness were assessed using Masson trichrome (Sigma) staining. Necrotic core was evaluated by measuring the total acellular area within each plaque. For fibrous cap thickness, three to five measurements representing the thinnest part of the fibrous cap were averaged for each plaque as previously described49. For tyrosine hydroxylase staining, femurs were harvested and fixed in 4% paraformaldehyde for 3 h and then decalcified in 0.375 M EDTA in PBS for 10 days before paraffin embedding. Sections were cut and stained with anti–tyrosine hydroxylase antibody (Millipore, AB152, dilution 1:100) after deparaffinization and rehydration. Sections were scanned with NanoZoomer 2.0-RS (Hamamatsu) at 40× magnification and analyzed using IPLab (Scanalytics).

Intravital microscopy.

For intravital microscopy of hematopoietic progenitors in the bone marrow of the calvarium, LSKs were isolated from either wild-type C57BL/6 or C57BL/6-Tg(UBC-GFP)30Scha/J mice and labeled with the lipophilic membrane dye DiD (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate, Invitrogen). 25,000 labeled LSKs were transferred i.v. into nonirradiated C57BL/6 recipient mice. For blood pool contrast, TRITC–dextran (Sigma) was injected immediately before imaging. OsteoSense 750 (PerkinElmer) was injected i.v. 24 h before in vivo imaging to outline bone structures in the calvarium50. In vivo imaging was performed on days 1 and 7 after the adoptive cell transfer using an IV100 confocal microscope (Olympus)15. Three channels were recorded (DiD excitation/emission 644/665 nm, OsteoSense 750 excitation/emission 750/780 nm, TRITC–Dextran excitation/emission 557/576 nm) to generate z stacks of each location at 2-μm steps. Image postprocessing was performed using Image J software. Mean DiD fluorescence intensity was measured for each labeled cell and then normalized to the background by calculating the target to background ratio.

Colony-forming unit assay.

Colony-forming unit (CFU) assays were performed using a semisolid cell culture medium (Methocult M3434, Stem Cell Technology) following the manufacturer's protocol. Bones were flushed with Iscove's Modified Dulbecco's Medium (Lonza) supplemented with 2% FCS. 2 × 104 bone marrow cells were plated on a 35-mm plate in duplicates and incubated for 7 days. Colonies were counted using a low magnification inverted microscope.

Blood pressure and heart rate measurement.

Blood pressure and heart rate were measured using a noninvasive tail-cuff system (Kent Scientific Corporation) according to the manufacturer's instructions. For each value, the mean of three consecutive measurements was used.

Quantitative real-time PCR.

Messenger RNA (mRNA) was extracted from aortic arches or bone marrow using the RNeasy Mini Kit (Qiagen) or from FACS-sorted cells using the Arcturus PicoPure RNA Isolation Kit (Applied Biosystems) according to the manufacturers' protocol. One microgram of mRNA was transcribed to complementary DNA (cDNA) with the high capacity RNA to cDNA kit (Applied Biosystems). We used Taqman primers (Applied Biosystems). Results were expressed by Ct values normalized to the housekeeping gene Gapdh.

Fluorescence molecular tomography–computed tomography.

After six weeks of stress, FMT-CT imaging was performed and compared to nonstressed, age-matched Apoe−/− controls. Pan-cathepsin protease sensor (Prosense-680, PerkinElmer, 5 nmol) was injected intravenously 24 h before the imaging as previously described51.

ELISA.

Blood corticosterone levels were measured by ELISA (Abcam). Serum was collected between 10 a.m. and 12 p.m. For measurements of noradrenaline in the bone marrow, a 2–CAT (A–N) Research ELISA (Labor Diagnostika Nord) was used. One femur was snap-frozen and immediately homogenized in a catecholamine stabilizing solution containing sodium metabisulfite (4 mM), EDTA (1 mM) and hydrochloric acid (0.01 N). Prior to the ELISA, the pH of the sample was adjusted to 7.5 using sodium hydroxide (1 N). ELISAs for CXCL12 (R&D), IFN-α (PBL Biomedical Laboratories) and IFN-γ (R&D) in the bone marrow were performed using one femur and one tibia per mouse14. ELISAs were performed according to the manufacturers' instructions.

Statistical analyses.

Statistical analyses were performed using GraphPad Prism software (GraphPad Software, Inc.). Results are depicted as mean ± standard error of mean if not stated otherwise. For a two-group comparison, a Student's t-test was applied if the pretest for normality (D'Agostino-Pearson normality test) was not rejected at the 0.05 significance level; otherwise, a Mann-Whitney U test for nonparametric data was used. For a comparison of more than two groups, an ANOVA test, followed by a Bonferroni test for multiple comparison, was applied. For analysis of clinical data, a Wilcoxon test for matched pairs was used. P values of <0.05 indicate statistical significance. No statistical method was used to predetermine sample size.