Selective Mobilization of Cytotoxic Leukocytes by Epinephrine

Acute stress in humans is well-known to cause a rapid and transient increase in WBC counts, that is, leukocytosis, presumably reflecting the demargination of these cells from the marginal pool (1–4). Phenotypic analysis revealed that this mobilization of leukocytes is selective and primarily affects, among others, effector CD8⁺ T cells and NK cells (3, 5). Immediate demargination of leukocytes from vascular endothelial cells is promoted by sympathetic nervous system (SNS) catecholaminergic activity stimulating PBMCs that express predominantly high-affinity β2-adrenoceptors (1, 3, 6–10). Interestingly, expression of β-adrenoceptors increases with immune cell activation and differentiation (10–14). Thus, β2-adrenoceptor-mediated mobilization of immune cells with cytotoxic effector potential has been proposed as a basic mechanism that allows cell redistribution to sites of injury during stress (3, 15, 16).

Aside from β2-adrenoceptor expression, the selective recruitment of leukocytes is determined by the specific profile of adhesion molecule expression on the cells. Stress as well as diurnal increases in SNS activity mobilize cells with low expression of L-selectin (CD62L) and high expression of integrin LFA-1 (CD11a) such as effector memory (EM) and effector CD8⁺ T cells, CD16⁺CD56^dim cytotoxic NK cells, and CD14^dimCD16⁺ proinflammatory monocytes, whereas their CD62L⁺CD11a^dim counterparts (e.g., naive T cells, CD16⁻CD56^bright immunomodulatory NK cells, and CD14⁺CD16⁻ conventional monocytes) are distinctly less sensitive to changes in catecholaminergic activity (10, 16–25). Along with the CD62L⁻CD11a^bright phenotype, effector immune cells are generally characterized by high expression of inflammatory chemokine receptors, such as CCR5, CXCR1, CXCR3, and CX3CR1 (26–28). Despite all these findings and that chemokines are main regulators of leukocyte traffic, the role of chemokine receptors in SNS-induced demargination of leukocytes has attracted surprisingly little attention (22, 24, 29–31).

We hypothesized that the mobilizing effect of catecholamines is linked to a specific profile of adhesion molecule and chemokine receptor expression. To identify such a profile in a comprehensive study, we examined in humans the effects of i.v. epinephrine infusion at concentrations within the physiological normal range on 14 different T cell, NK cell, and monocyte subsets. Epinephrine was chosen because it is well-known, via binding with high affinity to the β2-adrenoceptor of PBMCs, to play a key role in mediating acute stress-induced leukocytosis (1–3, 32–34). We show that epinephrine selectively mobilizes cells with cytotoxic effector functions, namely, CCR7⁻CD45RA⁺CD8⁺ effector T cells (EFF), CD4⁻CD8⁻ γ/δ T cells, CD3⁺CD56⁺ NKT-like cells, CD16⁺CD56^dim cytotoxic NK cells, and CD14^dimCD16⁺ proinflammatory monocytes. Analysis of adhesion molecule and chemokine receptor expression reveals a specific profile in these cells, characterized by high expression of CD11a and CX3CR1 (fractalkine receptor). The same subsets of cytotoxic effector cells show preferential adhesion to activated endothelium in vitro, which can be blocked by epinephrine. Our data suggest cytotoxic effector cells highly expressing CD11a and CX3CR1 to define a first line of immunological defense mobilized during acute stress.

Subjects were eight healthy nonsmoking men (mean age 26 y; range, 22–35 y) presenting with a normal sleep/wake pattern and not taking any medication at the time of the experiments. None had a medical history of any relevant chronic disease or mental disorder. Acute illness was excluded by physical examination and routine laboratory investigation, including a chemistry panel, C-reactive protein concentration <6 mg/l, and a WBC count <9000/μl. The study was approved by the Ethics Committee of the University of Lübeck. All the men gave written informed consent prior to participating in accordance with the Declaration of Helsinki.

Subjects were i.v. infused for 30 min with sodium chloride (placebo) or the endogenous β-adrenoceptor ligand epinephrine (0.005 μg/kg/min, Suprarenin; Sanofi-Aventis, Bridgewater, NJ) on two different occasions, according to a double-blind within-subject design. The order of substance administration was balanced across subjects. Subjects were prepared for blood sampling 1 h before infusions started and remained in a supine position throughout the session. Infusions started in the evening at 9 pm when endogenous epinephrine concentration is low. Blood for determining hormones and immune parameters was sampled via a second catheter (inserted into the vein of the other arm) before (baseline at 9 pm), during (15 and 30 min), and after (60 and 90 min) the start of the infusion. Heart rate and electrocardiogram were continuously monitored to exclude any adverse effects.

Whole blood was collected in plasma or serum tubes, kept on ice, and spun down within 10 min after collection. Plasma or serum were stored at −80°C until assay. Epinephrine and norepinephrine were measured in plasma by standard HPLC with subsequent electrochemical detection (Chromsystems, Munich, Germany) (35). Cortisol was measured in serum using a commercial assay (Immulite, DPC-Biermann GmbH, Bad Nauheim, Germany). The sensitivity and intra- and interassay coefficients of variation were as follows: cortisol, 0.2 μg/dl, <10%; epinephrine, 2.0 pg/ml, <5.6%; and norepinephrine, 5.0 pg/ml, <6.1%.

The following fluorochrome-conjugated Abs were used: CCR5 (Clone 2D7), CCR7 (3D12), CXCR1 (5A12), CXCR3 (1C6), CD3 (SK7), CD4 (SK3), CD8 (RPA-T8), CD11a (G43-25B), CD16 (3G8), CD45 (2D1), CD45RA (HI100), CD56 (B159), CD62L (SK11), HLA-DR (L243) all from BD Biosciences (San Jose, CA), CX3CR1 (2A9-1) from MBL International (Woburn, MA), CD11b (M1/70.15.11.5), CD49d (MZ18-24A9) from Miltenyi Biotec (Bergisch Gladbach, Germany), and CD14 (My4) from Beckman Coulter (Fullerton, CA). All isotype control Abs were purchased from BD Biosciences.

Based on previous reports, our analyses of leukocyte subsets in the in vivo administration study included naive, central memory (CM), EM and effector CD4⁺ and CD8⁺ T cells, CD4⁻CD8⁻ γ/δ T cells, NKT-like cells, immunomodulatory and cytotoxic NK cells, and conventional and proinflammatory monocytes. Because CD4⁻CD8⁻ T cells represent to the greatest extent γ/δ T cells, we refer to this population as CD4⁻CD8⁻ γ/δ T cells (36, 37).

Absolute counts of CD3⁺ (total T cells), NKT-like cells (CD3⁺CD56⁺), immunomodulatory NK cells (CD16⁻CD56^bright), cytotoxic NK cells (CD16⁺CD56^dim), conventional monocytes (CD14⁺CD16⁻), and proinflammatory monocytes (CD14^dimCD16⁺) were determined by a “lyse no-wash” flow cytometry procedure. To discriminate better CD3⁻CD16⁺ cells into either CD16⁺ NK cells or CD16⁺ monocytes, we used HLA-DR Ag [NK cells are negative for HLA-DR, whereas monocytes express this marker (23)]. An undiluted blood sample (50 μl) was immunostained with CD3, CD14, CD16, CD45, CD56, and HLA-DR Abs in Trucount tubes (BD Biosciences). After a 15-min incubation at room temperature, 0.45 ml FACS lysing solution (BD Biosciences) was added, followed by incubation for 15 min. No washing was performed to avoid cell loss. Finally, samples were mixed gently and at least 100,000 CD45⁺ leukocytes were acquired on a FACSCanto II using FACSDiva Software (BD Biosciences). The absolute number of the cells per microliter of blood was calculated using the following formula: Cells per microliter = (acquired cell events in the respective gate) × (number of beads per tube)/(acquired bead events) × (sample volume [microliters]).

For detection of naive (CCR7⁺CD45RA⁺), CM (CCR7⁺CD45RA⁻), EM (CCR7⁻CD45RA⁻), effector (CCR7⁻CD45RA⁺), and γ/δ (CD3⁺CD4⁻CD8⁻) T cells, whole blood was incubated with CCR7, CD3, CD4, CD8, and CD45RA Abs for 15 min. Cells were then lysed and washed twice before measuring on a FACSCanto II (BD Biosciences). The absolute counts were calculated based on percentage of the respective subpopulation and CD3⁺ absolute counts obtained by the lyse no-wash procedure.

The adhesion molecule profile on PBMC subpopulations was measured in morning blood collected from six subjects. Cells were processed and analyzed in the same way as in the main experiments. Results were expressed as percentage from the respective cell population or as mean fluorescence intensity (MFI). Isotype controls were used to set the markers determining positive and negative populations.

Vital HUVECs were purchased from Provitro (Berlin, Germany) and cultured in gelatin-coated 25 cm² tissue culture flasks (Sarstedt, Nümbrecht, Germany) in a humidified 37°C, 5% CO₂ environment using MCDB 131 culture medium (Invitrogen, Karlsruhe, Germany), supplemented with 8% FCS (Biochrom, Berlin, Germany), 2% heat-inactivated human serum, 100 μg/ml heparin, 2 mM l-glutamine, 100 U/ml penicillin, 0.1 mg/ml streptomycin, 2.5 μg/ml amphotericin B (all from Sigma-Aldrich, Seelze, Germany), and endothelial cell growth supplement (Promocell, Heidelberg, Germany). Tissue culture plates were coated with gelatin for 1 h at room temperature using autoclaved 1% gelatin solution made from gelatin powder (Serva, Heidelberg, Germany). HUVECs were used for experiments in the second to fifth passage.

PBMCs were isolated from heparinized whole blood obtained from seven healthy men. Blood was mixed with an equal volume of PBS and separated using 50-ml Leucosep tubes prefilled with Ficoll (Greiner Bio-One, Frickenhausen, Germany). PBMC fraction was rinsed twice with PBS containing 0.9 M Ca²⁺, 0.5 M Mg²⁺, and 0.5% BSA.

For the adhesion assay, HUVECs were detached using 0.05% trypsin and 0.02% EDTA, plated in 24-well culture plates precoated with gelatin using ∼4 × 10⁴ cells/well, and allowed to grow to confluence for 2 d. On the day of the binding assay, HUVECs were stimulated for 4 h with recombinant TNF-α and IFN-γ (at final concentrations of 25 ng/ml and 20 ng/ml, respectively) (Provitro) to enhance adhesion molecule and chemokine expression. At the end of the incubation, the HUVECs were rinsed twice with PBS containing 0.9 M Ca²⁺, 0.5 M Mg²⁺, and 0.5% BSA before adding 1 ml PBMCs (5 × 10⁵ cells/well). Adhesion was manipulated by adding epinephrine at a final concentration of 10⁻⁸ M [1832 pg/ml, i.e., a dose that in foregoing studies proved maximal efficacy (2, 38)] and all tests were performed in triplicate. Cells were allowed to adhere for 30 min (37°C, 5% CO₂). Nonadherent cells were carefully removed, and the plate was washed three times, twice with PBS containing Ca²⁺ and Mg²⁺, and one time with PBS without supplements. After the final wash, 250 μl trypsin/EDTA solution was added to obtain a single-cell suspension containing HUVECs and PBMCs. After 1 min of incubation (37°C, 5% CO₂), plates were shaken vigorously for an additional 1 min and 50 μl FCS was added, followed by 2 ml PBS containing 0.1% NaN₃. The cells were then separated equally into two tubes, centrifuged at 500g to throw away the supernatant, and stained with the same combinations of monoclonal Abs as used in the main study. Samples were acquired on a FACSCanto II using FACSDiva Software (BD Biosciences) until all events in the tubes were collected. As previously described (39), light scatter properties were used to distinguish PBMCs from HUVECs (HUVECs are larger and more granular than PBMCs). Events and percentages of the same T cell, NK cell, and monocyte subpopulations analyzed in the main study were counted and calculated.

Analysis of effects of i.v. epinephrine infusion on leukocyte subsets was based on ANOVA, including a substance and time factor, followed by paired t tests. Paired t test was also used to analyze differences in leukocyte adhesion to activated endothelium after epinephrine or placebo treatment. For correlation analyses, Pearson’s coefficients were calculated. Means across subjects were calculated for the 14 leukocyte subsets of interest for the “increase in vivo” (cell count after epinephrine/cell count after placebo at 9:30 pm) and correlated with the averaged CD11a, CXCR1, CX3CR1, and CD62L expression, respectively. CX3CR1 MFI was normalized by log transformation. The p values < 0.05 were considered significant. Data are presented as means ± SEM.

We i.v. infused eight healthy men in a balanced order with epinephrine (i.e., the major endogenous ligand of β2-adrenoceptors on PBMCs) and placebo (sodium chloride solution). Infusion started at 9 pm and was discontinued after 30 min. Low doses of epinephrine (0.005 μg/kg/min) were chosen to induce increases in plasma concentrations of the hormone comparable with levels observed during mild stress (16) and near to the effective dose generating 50% of maximal NK cell detachment in vitro [(38); Fig. 1]. Plasma epinephrine concentration reached peak values of 50.71 pg/ml (2.8 × 10⁻¹⁰ M) at the end of infusion and, thereafter, immediately decreased to recover baseline values 30 min later, reflecting the short half life of this hormone in blood (∼2 min). Epinephrine infusion did not affect plasma norepinephrine or cortisol concentration (p > 0.5; data not shown).

Within the CD4⁺ and CD8⁺ T cell subsets, infusion of epinephrine, compared with placebo, induced a selective and distinct increase in numbers of effector CD8⁺ T cells (p < 0.01). The increase was short lived and focused on the end of the infusion when epinephrine levels were maximal (Fig. 2H). All other CD4⁺ and CD8⁺ T cell subsets remained unchanged (p > 0.5). The increase in effector CD8⁺ T cell numbers was paralleled by significant increases in CD4⁻CD8⁻ γ/δ T cell and NKT-like cell counts during infusion of epinephrine (p < 0.05, Fig. 2I, 2J). Again, cell counts peaked at the end of the infusion and recovered normal values shortly afterward. As expected, most pronounced increases were revealed for cytotoxic NK cell numbers that were increased by 200% at the end of the epinephrine infusion period, and then rapidly returned to baseline values within 30 min (p < 0.01, Fig. 2L). Although to a smaller extent, epinephrine infusion also enhanced counts of proinflammatory monocytes (p < 0.05, Fig. 2N). In contrast, immunomodulatory NK cells and conventional monocytes remained unchanged by epinephrine (p > 0.3). The pattern of epinephrine-induced increases in the different subsets of T cells, NK cells, and monocytes was confirmed by analyses based on percentages of cells expressed with reference to the respective parent population (data not shown).

Adhesion molecules and chemokine receptors control leukocyte migration and thus represent candidate mechanisms for the differential demargination of leukocyte subsets by epinephrine. We defined adhesion molecule and chemokine receptor profiles in all leukocyte subpopulations of interest in samples from healthy donors collected in the morning (results are summarized in Fig. 3). Cell surface expression of L-selectin (CD62L), integrins (CD11a, CD11b, and CD49d), and so-called inflammatory chemokine receptors (CCR5, CXCR1, CXCR3, and CX3CR1) were examined. As to the adhesion molecules, all cytotoxic effector cell subsets were well-characterized by low expression of CD62L and high levels of CD11a. Integrins CD11b and CD49d did not show any preferential expression on the cells of interest (Fig. 3B).

As to the chemokine receptors, CCR5 was absent in naive T cells, but was expressed preferentially on EM CD4⁺ and CD8⁺ T cells, CD4⁻CD8⁻ γ/δ T cells, NKT-like cells, and immunomodulatory NK cells. CXCR3 expression was not selective for any T cell subset, but it was expressed preferentially on immunomodulatory NK cells. In contrast, CX3CR1 (fractalkine receptor) expression clearly characterized those cytotoxic effector immune cells that were mobilized by epinephrine infusion (black bars in Fig. 3B). Representative flow cytometry data demonstrating the selective expression of CX3CR1 in cytotoxic effector cells are shown in Fig. 4. Expression of CXCR1 (IL-8 receptor) showed a pattern comparable to that of CX3CR1, except that it was absent in proinflammatory monocytes (Fig. 3B).

To further explore the functional significance of these markers for epinephrine-induced mobilization, we correlated the expression of adhesion molecules and chemokine receptors in the 14 leukocyte subsets of interest with the increase in respective cell counts after epinephrine administration. Pronounced positive correlations were revealed for CD11a (r = 0.86, p = 0.00015) and CX3CR1 expression (r = 0.78, p = 0.0015, Fig. 3C, 3D), suggesting a dominant contribution of these two molecules in mediating the epinephrine-induced demargination of leukocytes. Correlations were in negative direction for CD62L (r = −0.66, p < 0.05) and only approached significance for CXCR1 (r = 0.53, p = 0.064), (which does not exclude that CXCR1 adds to the exaggerated epinephrine-induced mobilization in vivo of cytotoxic NK cells). CD11a and CX3CR1 showed a high degree of coexpression (r = 0.84, p < 0.001).

As a further test of the importance of CX3CR1 for mobilizing leukocytes, we studied the effect of epinephrine infusion specifically on total T cells expressing this chemokine receptor. CX3CR1⁺ T cells were substantially mobilized after epinephrine administration (p < 0.01, Fig. 5D). In contrast, CX3CR1⁻ as well as CCR5⁺ and CXCR3⁺ T cell counts were not influenced (p > 0.3), and CXCR1⁺ T cells were only marginally increased (p < 0.1). Virtually the same differential effect of epinephrine was revealed when NK cells and monocytes were categorized based on CX3CR1 expression, because the CX3CR1 receptor was present on practically all cytotoxic NK cells and proinflammatory monocytes (results not shown).

We characterized in vitro adhesion of the cell subsets of interest to activated HUVECs after incubating the endothelial cells with PBMCs for 30 min. The proportion of cells of a subset (with reference to total PBMCs) was compared in the fraction prior to incubation and in the adherent fraction after incubation. For the main total populations (T cells, NK cells, and monocytes), we found distinct enrichment of NK cells (from 12 ± 2% to 31.5 ± 3.1%) and monocytes (from 9.4 ± 1.1% to 24.2 ± 1.6%) in the adherent fraction (p < 0.001, Table I). This enrichment was primarily due to cytotoxic, but not immunomodulatory NK cells, and was more pronounced for proinflammatory than conventional monocytes. In contrast, percentage of T cells decreased from 63.3 ± 2.6% prior to adhesion to 31.1 ± 2.5% in the adherent fraction (p < 0.001). This decline was due to a large extent to a drop of naive CD4⁺ and naive CD8⁺ T cells (Table I). Unlike naive T cells, proportions of more differentiated T cells with cytotoxic function, namely, EM and effector CD8⁺ T cells, CD4⁻CD8⁻ γ/δ T cells and NKT-like cells, remained unchanged or were even increased in the adherent fraction (Table I).

Addition of epinephrine to the adhesion assay resulted in clearly reduced percentages of effector CD8⁺ T cells, CD4⁻CD8⁻ γ/δ T cells, NKT-like cells, and cytotoxic NK cells and an increased percentage of conventional monocytes in the adherent fraction. This diminished adhesion of effector CD8⁺ T cells, CD4⁻CD8⁻ γ/δ T cells, NKT-like cells, and cytotoxic NK cells in the presence of epinephrine was confirmed both for direct comparisons of absolute numbers of subpopulations in the adherent fraction with and without epinephrine (Fig. 6) and for the coefficients of adherence (indicating the ratio of cell numbers in the attached fraction to the cell numbers prior to adhesion, measured in percentage; see Table I, right two columns, p < 0.05 for all four subpopulations). Adhesion of the other T cell subpopulations, immunomodulatory NK cells, and both monocyte subpopulations was not significantly affected.

We aimed at a comprehensive characterization of the cell subsets and mechanisms involved in the epinephrine induced mobilization of immune cells in humans, a phenomenon well-known to accompany the organism's response to acute stress. Cell surface markers were used to perform phenotypical analyses of 10 T cells, 2 NK cells, and 2 monocytes subsets that greatly differ in their function. We show that stresslike increases in epinephrine concentrations invoke a specific increase in immune subpopulations with cytotoxic effector potential, namely, effector CD8⁺ T cells (CD3⁺CCR7⁻CD45RA⁺), γ/δ T cells (CD3⁺CD4⁻CD8⁻), NKT-like cells (CD3⁺CD56⁺), cytotoxic NK cells (CD16⁺CD56^dim), and proinflammatory monocytes (CD14^dimCD16⁺). We show that these leukocyte subpopulations are characterized by a common adhesion molecule and chemokine receptor profile with a CD62L⁻CD11a^brightCX3CR1^bright phenotype. CD11a and CX3CR1 expression correlates with the degree of adrenergic mobilization.

It has been repeatedly demonstrated that the number of immune cells increases after physical exercise or mental stress, with the most pronounced increases occurring in NK cell and CD8⁺ cytotoxic T cell counts (3, 5, 16, 34), whereas both populations are decreased when endogenous catecholamines are suppressed by stellate ganglion block (40). These effects are caused by adrenal medullary epinephrine binding to β2-adrenoceptors on PBMCs, with high and intermediate levels of β2-adrenoceptor expression on NK cells and cytotoxic T cells, respectively (1, 3, 7, 8, 19, 33, 41). Our findings substantially expand these previous findings in showing that the epinephrine-induced increase in circulating immune cells is highly specific to cytotoxic effector subtypes, namely, effector CD8⁺ T cells, γ/δ T cells, NKT-like cells, cytotoxic NK cells, and proinflammatory monocytes, whereas no change is seen for cells lacking any cytotoxic potential, namely, CD4⁺ T cells, CD8⁺ T cells at their early stage of differentiation, immunomodulatory NK cells, and conventional monocytes. Consistent with previous observations (3, 16, 25, 42), cytotoxic NK cells show the most pronounced increase after epinephrine.

The epinephrine-induced mobilization of effector CD8⁺ T cells is consistent with previous studies of increased effector CCR7⁻/CD62L⁻CD45RA⁺ CD8⁺ T cell numbers after stress or catecholamine administration (19, 22, 25). These effector CD8⁺ T cells represent a stage of T cell differentiation occurring late in the immune response, and are hence called terminally differentiated cytotoxic T cells. They do not proliferate in culture and express markers such as the killer cell lectin-like receptor G1 and CD57, defining their replicative senescence. Recent studies indicated a mobilization of CD8⁺ T cells expressing the killer cell lectin-like receptor G1 and CD57 after physical exercise (25, 43, 44). Effector CD8⁺ T cells can also express the NK cell marker CD56 that seems to define those cells with high direct cytolytic capacity (42, 45). Hence, the epinephrine-induced increase in CD3⁺CD56⁺ NKT-like cells observed in the current as well as in previous studies (22, 42) might to some extent overlap with a mobilization of effector CD8⁺ T cells.

Our findings of substantial mobilization of γ/δ T cells after epinephrine likewise agrees with the view of a first line of immunological defense promoted through the catecholamine (46). CD3⁺CD4⁻CD8⁻ γ/δ T cells constitute a unique subset of T cells that exhibit spontaneous non-MHC restricted cytotoxicity. Although their function is not fully understood, they likely serve as an immediate defense against foreign pathogens (47). Moreover, the observed increase in cells after epinephrine administration was restricted to cytotoxic NK cells and proinflammatory monocytes that complements previous studies (16, 20, 21, 24, 25, 42), whereas no change was seen for immunomodulatory NK cells or conventional monocytes. Like effector CD8⁺ T cells, γ/δ T cells, and NKT-like cells, both cytotoxic NK cells and proinflammatory monocytes are important cytotoxic effector cells in the early immune defense (28, 48–52). Of note, all five subsets with high effector activity produce IFN-γ and TNF-α, show high tissue-migrating potential, and, except for proinflammatoty monocytes, share phenotype signs of senescence (CD27⁻, CD28⁻) and high cytotoxicity (granzyme, perforin). In combination, these subsets thus build up an acute defense system whose mobilization on epinephrine release allows for efficient surveillance of tissues and rapid accumulation at sites of injury and infection (22, 25–28, 42, 45, 48, 52–54). In this context, it would be of interest to examine the effect of epinephrine also on other cytotoxic effector leukocytes, for example, CD8⁺ type 1 NKT cells showing strong antitumor cytotoxicity (54, 55).

The fast mobilization [within 1 min (20)] and recovery of cytotoxic cell numbers in blood that has been observed in this study and in previous studies after stress, exercise, and infusion of catecholamines, reflects the mobilization of the cells from a quite dynamic compartment, namely, the marginal pool that is well-known to house a variety of T cells, NK cells, and monocyte subpopulations (3, 4). Consistent with previous examinations of stress (16), in the current study, peak epinephrine levels of 10⁻¹⁰ M induced a 3-fold increase in NK cell numbers. In vitro experiments revealed a clear dose dependency of this NK cell detachment after epinephrine (38), and it becomes likewise evident from a comparison across different in vivo experiments (1, 33, 42). Acute increases in PBMCs to stress and exercise have been consistently revealed to be mediated via epinephrine binding to β2-adrenoceptors, whereas noradrenergic influences and changes in blood flow seem to play minor roles (1–3, 33, 34, 38, 56).

What is the basis for the selectivity of the epinephrine induced increase specifically in cytotoxic cells? We hypothesized that a unique adhesion molecule and chemokine receptor profile make cytotoxic effector cells residents of the marginal pool, and that changes of these molecules via stimulation of β2-adrenergic receptors induce their demargination. In line with previous findings (4, 16, 17, 20, 21, 25, 53, 57, 58), in this study, epinephrine-sensitive cytotoxic cells showed a lack of CD62L and a high density of CD11a. Importantly, however, we revealed that cytotoxic effector cells also selectively express CXCR1 and especially CX3CR1 (26–28, 31, 48, 53, 58). The role of these inflammatory chemokine receptors in adrenergic leukocytosis is currently obscure although particularly CX3CR1 is a very likely candidate mediating demargination of cytotoxic effector cells by catecholamines: CX3CR1 replaces function of selectin as adhesion molecule on CD62L⁻ cells (27, 30, 53), like all chemokine receptors its stimulation enhances the affinity of integrins such as CD11a (27, 59) and, like the β2-adrenoceptor, it is a G protein–coupled receptor that enables immediate influences of catecholamines on CX3CR1 signaling (60, 61). In fact, β2-adrenoceptors are generally expressed at high levels in cytotoxic effector cell populations like NK cells and effector CD8⁺ T cells [characterized by a CD62L⁻CD11a^brightCX3CR1^bright phenotype (8, 10)]. Our data add to this view: Epinephrine in vivo mobilized only CX3CR1⁺ T cells, whereas their CX3CR1⁻ counterparts remained unaffected. Moreover, CX3CR1 as well as CD11a expression was strongly correlated with the degree of epinephrine-induced mobilization. In summary, cytotoxic effector cell populations show a high expression of β2-adrenoceptors, CD11a and CX3CR1 that correspond to a great capability of these cells to be mobilized by epinephrine. A number of experiments indicates that β2-adrenoceptor-mediated leukocytosis is associated with an increase in intracellular cAMP (3, 38, 41, 62, 63). This second messenger was shown to attenuate chemokine triggered integrin affinity within seconds (60, 61). We are thus tempted to speculate that PBMCs are attached to the endothelium via CX3CR1 triggered adhesive CD11a signaling and become immediately released into the circulation on epinephrine binding.

We applied an in vitro assay with HUVECs to further examine these mechanisms of adrenergic leukocytosis. Earlier studies using this assay revealed distinctly reduced cell adhesion of NK or T cell populations after administration of β2-adrenoceptor agonists (2, 41). Using the long-acting β2-adrenoceptor antagonist, GR81706, it was shown that this phenomenon is mediated via receptors on the leukocyte itself and not on the endothelium (62). Our in vitro experiments indicate that cytotoxic effector leukocytes display the highest adherence to activated endothelium, strongly suggesting that these cell populations form the predominant part of cells in the marginal pool also in the in vivo condition. We further showed that epinephrine selectively inhibits adhesion of effector CD8⁺ T cells, γ/δ T cells, NKT-like cells, and cytotoxic NK cells to endothelium. Of the cytotoxic cells that are mobilized in vivo by epinephrine, proinflammatory monocytes were the only subset not affected by epinephrine in vitro. This discrepancy may reflect the specific conditions in vitro: 1) the activation of HUVECs and hence higher expression of, for example, intercellular cell adhesion molecule-1 and fractalkine; 2) the isolation of PBMCs in the in vitro assay that possibly leads to a further activation of monocytes and expression of α-adrenoceptors on these cells; 3) the epinephrine concentration that was >30-fold higher than in vivo, and thus possibly costimulated α-adrenoceptors with presumed opposite effects on leukocyte adhesiveness; and 4) the use of a static assay that does not accurately reflect the conditions of flow in vivo (7, 27, 28, 39, 64–66). These differences highlight that conclusions from the in vitro data have to be drawn with caution, as well as the need for additonal studies using different doses of epinephrine and selective α- and β-adrenergic agonists and antagonists.

In summary, we demonstrate that epinephrine in vivo selectively increases numbers of circulating cytotoxic leukocytes, including effector CD8⁺ T cells, γ/δ T cells, NKT-like cells, cytotoxic NK cells, and proinflammatory monocytes. Except for proinflammatory monocytes, all these cytotoxic cell populations show reduced adhesion to endothelium after epinephrine administration in vitro. Importantly, our in vivo studies used small and short-lasting increases in epinephrine concentrations to establish conditions closely comparable with those observed during acute stress. They cannot be used to infer effects of chronic β-adrenoceptor stimulation as seen, for example, in patients with heart failure where catecholamines seem to exert effects in opposite direction with selectively reduced numbers of circulating NK cells and CD8⁺ T cells (67, 68). On acute stress, increased adrenal medullary epinephrine release likely serves to selectively recruit CD62L⁻CD11a^brightCX3CR1^bright cytotoxic effector cells as the first line of defense against pathogens.

We are grateful to C. Benedict, D. Heutling, and A. Tschulakow for technical assistance.

Disclosures The authors have no competing financial interests.