Glucocorticoid-induced eosinopenia results from CXCR4-dependent bone marrow migration

Abstract

Glucocorticoids are considered first-line therapy in a variety of eosinophilic disorders. They lead to a transient, profound decrease in circulating human eosinophils within hours of administration. The phenomenon of glucocorticoid-induced eosinopenia has been the basis for the use of glucocorticoids in eosinophilic disorders, and it has intrigued clinicians for 7 decades, yet its mechanism remains unexplained. To investigate, we first studied the response of circulating eosinophils to in vivo glucocorticoid administration in 3 species and found that the response in rhesus macaques, but not in mice, closely resembled that in humans. We then developed an isolation technique to purify rhesus macaque eosinophils from peripheral blood and performed live tracking of zirconium-89-oxine-labeled eosinophils by serial positron emission tomography/computed tomography imaging, before and after administration of glucocorticoids. Glucocorticoids induced rapid bone marrow homing of eosinophils. The kinetics of glucocorticoid-induced eosinopenia and bone marrow migration were consistent with those of the induction of the glucocorticoid-responsive chemokine receptor CXCR4, and selective blockade of CXCR4 reduced or eliminated the early glucocorticoid-induced reduction in blood eosinophils. Our results indicate that glucocorticoid-induced eosinopenia results from CXCR4-dependent migration of eosinophils to the bone marrow. These findings provide insight into the mechanism of action of glucocorticoids in eosinophilic disorders, with implications for the study of glucocorticoid resistance and the development of more targeted therapies. The human study was registered at ClinicalTrials.gov as #NCT02798523.

Conflict of interest statement

Conflict-of-interest disclosure: P.L.C. and N.S. possess a US patent for the 89Zr-oxine complex cell-labeling method and have filed US and international patent applications for the generation and use of the 89Zr-oxine complex employed in this study. The remaining authors declare no competing financial interests.

Figures

Graphical abstract

Figure 1.

In vivo kinetics of circulating eosinophils and other blood cells after glucocorticoid administration in 3 species. Twenty healthy human volunteers received a single IV dose of glucocorticoid (methylprednisolone, 250 mg). Four adult rhesus macaques received a single IV dose of glucocorticoid (methylprednisolone, 4 mg/kg). Ten adult B6.SJL mice were randomly divided into 2 groups. One group (n = 5) received a single IV dose of glucocorticoid (methylprednisolone, 5 µg/g). Another group (n = 5) received vehicle (normal saline) at the same volume as the glucocorticoid-treated animals. The cell composition of peripheral blood was assayed by automated CBCs in a clinical laboratory. (A) Blood composition among the 3 species, at BL and in response to glucocorticoid (GC). The relative proportion of each color corresponds to the mean percentage of the respective cell type in the species displayed. WBC, white blood cell. (B) Top: relative abundance (expressed as a percentage of total WBCs) of eosinophils in peripheral blood, before and after glucocorticoid administration, in each of the 3 species. Bottom: AEC in peripheral blood, before and after glucocorticoid administration, in each of the species. In human and rhesus macaque cells, values at BL and in response to glucocorticoid (4 hours) are shown. In mice, values in vehicle- or glucocorticoid-treated animals are shown. Error bars represent the mean ± SD. Statistical analyses of the human and rhesus data were performed with a paired Student t test. Statistical analysis of the mouse data was performed with a Student t test for independent samples.

Figure 2.

Purification and 89Zr-oxine labeling of rhesus blood eosinophils. (A) Flow cytometric analysis with cell surface staining for CD16, Siglec-8, VLA4, or CD64 in human and rhesus whole-blood samples. Representative plots are shown for each marker. (B) Representative dot plots show flow cytometric analysis with cell surface staining for VLA4 or CD64 followed by intracellular staining for EPX in rhesus granulocytes. (C) A representative image shows nuclear and cytoplasmic morphology of rhesus eosinophils after purification. (D) Eosinophil purity in the CD64-negative rhesus granulocyte fraction. Shown is the relative abundance of 5 WBC types in the purified cell fraction in 5 independent experiments. Error bars represent mean ± SD. (E-H) Apoptosis, viability, activation, and migration assays measured by flow cytometry in purified rhesus eosinophils, with or without 89Zr-oxine labeling: percentage of cells positive for the early apoptosis marker annexin V (E), in 4 independent experiments; percentage of nonviable cells, defined as cells positive for both annexin V and propidium iodide, in 4 independent experiments; (G) percentage of cells positive for the activation marker CD69, in 3 independent experiments (F); and in vitro migration of purified rhesus eosinophils, with or without 89Zr-oxine labeling (H), toward control (no chemokine) or recombinant rhesus CCL11. The percentage of cells that migrated toward CCL11 on a Transwell assay after 2 hours of incubation, in 4 independent experiments, is shown. Statistical analysis of the data in panels E-H was performed with the paired Student t test.

Figure 3.

Glucocorticoids induce bone marrow migration of eosinophils. Circulating rhesus eosinophils were purified, labeled with 89Zr-oxine, and reinjected. PET/CT imaging was performed to track the location of the cells over time, with or without glucocorticoid administration. (A) Experimental design. Day 0 corresponds to the day of cell reinjection. Imaging on day 0 was performed immediately after reinjection (indicated by the 10-minute time point, which corresponds to the midpoint of the 20-minute imaging session), and 1 hour after reinjection. On day 1, animals in the treatment arm each received a single intravenous dose of glucocorticoid (GC, methylprednisolone 4 mg/kg), administered 30 minutes after BL PET imaging. Subsequent images were acquired hourly (counting from the time of BL imaging) for 4 hours. Control (Ctrl) animals were imaged serially at the same time points, without GC administration. Three independent experiments were performed per group, each on a different day in an unrelated adult rhesus macaque. The abbreviations and colors for each time point are preserved in the other panels of the figure to facilitate interpretation. (B) Live tracking of 89Zr-oxine–labeled eosinophils in the first hour after cell reinjection on day 0. Representative MIP PET/CT images at 10 minutes and 1 hour on day 0 are shown (top, SUV: standardized uptake value). The quantification of 89Zr distribution expressed as % ID with decay correction, is plotted for each organ at each imaging time point (bottom). These values indicate the percentage of the total 89Zr activity in a given organ’s volume, relative to the injected dose. (C) Live tracking of 89Zr-oxine labeled eosinophils in the absence (top) or presence (bottom) of glucocorticoid administration on day 1. Representative MIP PET/CT images at BL and 4 hours. (D) Sagittal and axial images of the spine in glucocorticoid-treated animals, before (BL) and after (4 hours) glucocorticoid injection. The red lines indicate the locations of the axial slice planes shown at the bottom. PET/CT images of 1 representative animal are shown. (E) Localization of 89Zr-oxine labeled eosinophils over time, with or without glucocorticoid administration. The x-axis is the time (in hours) after BL imaging. The y-axis displays 89Zr-oxine labeled eosinophil distribution, expressed as a decay-corrected percentage of the injected dose. Data are displayed separately for each of 4 organs: bone marrow, liver, lung, or spleen. Linear regression was performed separately for the 3 animals in the glucocorticoid arm and the 3 animals in the control arm. In both regressions, the response is organ 89Zr distribution (from the 89Zr-oxine labeled-eosinophils) expressed as a percentage of the injected dose; the regressors are the type of organ, the observation time point, and their interaction. The slope, or rate of change in each organ, is shown along with the P value from a Wald test indicative of whether the slope is significantly different from 0.

Figure 4.

Glucocorticoid-induced eosinopenia is CXCR4 dependent. (A) The CXCR4 locus in the human and rhesus genomes, highlighting the location of the GREs relative to the known transcript isoforms in each species. (B) Peripheral blood AECs and eosinophil surface expression of CXCR4 over time after a single dose of methylprednisolone (4 mg/kg IV). Results of 3 independent experiments, each in 1 unrelated adult rhesus macaque. CXCR4 surface levels are expressed as ΔMFI (MFI CXCR4 staining minus MFI isotype control). (C) Experimental design for testing the effect of CXCR4 blockade on glucocorticoid-induced eosinopenia. (D) Glucocorticoid-induced eosinopenia in the presence or absence of CXCR4 blockade. Eosinophil counts are expressed as a percentage of the BL count for each condition. Results of 3 independent experiments, each in 1 unrelated adult rhesus macaque. GC, glucocorticoid; MP, methylprednisolone; RM, rhesus macaque.