Acute stress reduces intraparenchymal lung natural killer cells via beta-adrenergic stimulation

Abstract

There are lines of evidence that natural killer (NK) cells are sensitive to physical and psychological stress. Alterations in the immune system including NK cells are known to differ among tissues and organs. The effect of stress on the lung immune system, however, has not been well documented in spite of the fact that the lungs always confront viral or bacterial attacks as well as tumour cell metastasis. In this study, we intended to investigate the effect of restraint stress on lung lymphocytes including NK cells. C57BL/6 mice were exposed to 2 h restraint stress. The concentration of plasma epinephrine significantly rose immediately after the release from restraint as compared to home-cage control mice. Flow cytometric analysis revealed that the numbers of most lymphocyte subsets including NK cells were decreased in the lungs and blood but not in the spleen, immediately after restraint stress. Immunohistochemical examination revealed that the number of NK cells was decreased in the intraparenchymal region of the lungs, while the number of alveolar macrophages did not change. The decrease in the number of NK cells in the lungs and blood was reversed by the administration of propranolol, a nonselective beta adrenergic antagonist. Taken together, our findings suggest that acute stress reduces the number of intraparenchymal lung NK cells via activation of beta adrenergic receptors.

Introduction

Stress is now considered an important modulator of the immune system, as it has been shown that many parameters of immunity are affected by physical and psychological stressors [1–3]. Furthermore, it has been revealed that such immune modifications under stress are organ or tissue-specific [4], and are dependent on factors like the type and duration of stress [5–8].

As for clinical significance, it is suggested that stressful conditions could make organs susceptible to virus infection [9] or reactivation [10], lead to higher mortality by virus infection [9,11], and facilitate tumour development in animals [12]. In humans, a stressful event is not only a possible risk factor for breast cancer [13], but also leads to increased recurrence and mortality [13,14].

On the other hand, stress has been found to affect NK cells as well as other types of immune cells [15–18]. Moreover, Imai et al. [19] have reported that lower NK cytotoxicity is a risk factor for cancer incidence.

The lungs are known as a target organ of certain types of viruses, bacteria and metastatic tumours, against which the lung immune cells organize a surveillance system [20–22]. Alveolar macrophages, which are involved in the lungs’ immune surveillance system, are relatively well examined in terms of modification under stress [23–26] including exercise [27]. However, the effect of stress on other types of lung immune cells, namely intraparenchymal NK cells as well as the other intraparenchymal lymphocytes, is less well known, although functional NK cells have been found to exist in the pulmonary interstitium [28].

In this study, we intended to reveal the effect of stress on intraparenchymal lung NK cells. As it has already been revealed that changes in the cell population in the blood stream are involved in stress-induced immune modifications [29–33], we focused on the changes in the number of lung NK cells under stress. Since blood lymphocyte redistribution is mainly caused by beta-adrenergic stimulation [34], we hypothesized that the expected changes in the number of lung lymphocytes including NK cells as well as that of blood lymphocytes under stress would be caused by elevated catecholamine levels followed by beta-adrenergic stimulation.

Materials and methods

Animals

Female C57BL/6 mice were purchased from the mouse supply centre of Tohoku University Graduate School of Medicine, and used for the experiments at 6–8 weeks of age. Animals were maintained under specific pathogen-free conditions on a daily 12 h light/dark cycle, and were able to access food and water ad libitum.

Restraint stress procedure

Each experimental female C57BL/6 mouse was placed for two hours in a 50 ml centrifuge tube. The walls of the tube were stripped in part to avoid an acute rise in the body temperature. Food and water were kept away during stress protocol.

Drug administration

Propranolol (Sigma, St. Louis, MO, USA), a nonselective beta adrenergic antagonist, was intraperitoneally administrated to the mouse by 20 mg/kg at the volume of 200 µl per mouse 30 min before restraint stress. Saline was administrated at the same volume for control.

Anaesthesia

Diethyl ether (Wako, Osaka, Japan) or sevoflurane (Abbott Japan, Tokyo, Japan) inhalation was used to anaesthetize animals. Less than 2 ml of anaesthetics were evaporated into a container with a volume of less than 500 ml. Then, the mice were put into the container and anaesthetized within 15 s.

Assessment of plasma catecholamine concentrations

Blood samples were drawn from retro-orbital plexus of sevoflurane-anaesthetized control and experimental mice immediately, 1, and 2 h after the release from 2 h restraint and collected into heparin-coated tubes. After centrifugation at 1700 g for 10 min at 4°C, plasma was collected and catecholamines including epinephrine, norepinephrine, and dopamine were measured by column-switching high-performance liquid chromatography (HPLC) with fluorometric detection as described previously with slight modifications [35,36]. Briefly, after protein precipitation using 5% perchloric acid, plasma samples were injected into a reversed-phase gel column (TSK precolumn CA1, TOSOH, Tokyo, Japan) to remove water-soluble components, and into a cation exchange gel column (TSK precolumn CA2, TOSOH) to remove anionic components. Then, catecholamines were separated using a Wakosil-II 5C18 RS column (Wako), and converted into the fluorescent compounds by the reaction with diphenylethylenediamine (DPE). Components were detected by fluorescence at an excitation wavelength of 347 nm and an emission wavelength of 483 nm.

Preparation of intraparenchymal pulmonary mononuclear cells, peripheral blood mononuclear cells, and splenocytes

Intraparenchymal pulmonary mononuclear cells were isolated using the method described previously with slight modification [37,38]. Briefly, after retro-orbital blood collection under anaesthetization using diethyl ether as described, the lung vascular bed was perfused with 5 ml saline injected into the right ventricle. Then, the lungs were excised immediately, minced into 2–3 mm sections with scissors, and resuspended in RPMI 1640 medium (Sigma) supplemented with 10% heat-inactivated fetal bovine serum (FBS, Sigma) and 150 U/ml of collagenase type IV (Sigma). Following incubation for 1 h at 37°C with vigorous shaking, the lung particles were teased through a metal mesh. After centrifugation, the cells were resuspended in 5 ml of 45% (v/v) Percoll (Pharmacia Biotech, Uppsala, Sweden), layered on 2 ml of 67·5% Percoll, and centrifuged at 600 g for 20 min at 20°C. Then, the interface containing mononuclear cells was collected, and cells were washed twice and counted with a haemocytometer. Peripheral blood mononuclear cells were also separated using Percoll density gradient. Spleen was teased through a metal mesh, and splenocytes were purified by lysing red blood cells with 0·83% NH₄Cl Tris buffer for 5 min

Flow cytometric analysis

Mononuclear cells from the lungs, blood, and spleen of control or restraint mice were subjected to flow cytometric analysis. The following antibodies (Abs) were used to identify each subset of lymphocytes: PE-conjugated anti-NK1·1 (clone PK136), FITC-conjugated anti-CD3 (clone 145–2C11), PE-conjugated anti-CD4 (clone RM4-5), PerCP-conjugated anti-CD8 (clone 53–6·7), and PE-conjugated anti-CD19 (clone 1D3) (all purchased from BD PharMingen, San Diego, CA). Briefly, the cells were stained with each monoclonal antibody on ice in RPMI1640 after preincubation with mouse serum (Sigma) for blocking nonspecific binding to Fc receptors, washed with Ca²⁺, Mg²⁺-free phosphate-buffered saline (– PBS, Nissui, Tokyo, Japan) containing 2% FBS, and subjected to flow cytometric analysis. Data were collected from 10000 individual cells using the parameters of forward scatter and side scatter to set a gate on the lymphocyte population. The absolute number of each lymphocyte subset was calculated as the product of the total cell count, fraction of lymphocytes, fraction of each lymphocyte subset, and fraction of live cells not stained with propidium iodide (PI, Sigma).

Immunohistochemistry of lung tissues

For immunohistochemical analysis, mice were anaesthetized with diethyl ether as described, and exsanguinated via the femoral artery. The lung airways were filled with 1 ml of a 50% solution of optimum cutting temperature compound (O.C.T, Sakura, Tokyo, Japan) in physiological saline supplemented with 5% (w/v) bovine serum albumin (BSA, Sigma). After removal, the lungs were embedded in O.C.T, frozen in liquid nitrogen, and cut into sections 5 µm thick using a Cryostat. The lung sections were then fixed in cold acetone (Wako), stained with anti asialo-GM1 antibody (Wako) [39–41] using a Dako Envision System, Peroxidase (Dako, Carpinteria, CA, USA) and the dextran polymer conjugate two-step visualization method [42,43], and counterstained with haematoxylin (Wako). Small-sized asialo-GM1+ cells in the intraparenchymal pulmonary area were considered NK cells, and large-sized asialo-GM1+ cells in alveolus, alveolar macrophages. Photographs were taken at × 400 magnification in 25 randomly selected regions from 5 mice (5 regions per mouse) in each group, and the number of each type of cell stained with anti asialo-GM1 antibody was counted.

Statistic analysis

The data from flow cytometric studies and the measurement of catecholamine concentrations were analysed using anova with Fisher's PLSD post hoc test. The data from immunohistochemical studies were analysed using Student's t-test. P < 0·05 was considered significant.

Results

Effect of stress on plasma catecholamines concentrations

In order to investigate neuroendocrine modulations under restraint stress, we measured plasma concentrations of catecholamines (epinephrine, norepinephrine, and dopamine). The plasma epinephrine concentration was significantly elevated immediately after restraint stress compared with the control (Fig. 1, P < 0·001), and returned to baseline within 1 h after the release from restraint. Plasma concentrations of norepinephrine and dopamine were also elevated immediately after restraint stress and returned to baseline within 1 and 2 h, respectively, after the release from restraint, although their increase was not statistically significant by anova (Fig. 1).

Fig. 1.

Effect of restraint stress on plasma concentrations of (a) epinephrine, (b) norepinephrine and (c) dopamine. Blood samples were drawn from retro-orbital plexus of anaesthetized control mice (Con, n = 6) and of mice anaesthesized immediately, 1, and 2 h after the release from 2 h restraint stress (RS0, n = 6; RS1, n = 6; RS2, n = 5, respectively). Plasma catecholamines (epinephrine, norepinephrine, and dopamine) were measured by HPLC with fluorometric detection. The data represent the means ± SE. The asterisks indicate a statistically significant difference; ***P < 0·001; anova with Fisher's PLSD post hoc test.

Effect of restraint stress on the number of lymphocytes

The numbers of total lymphocytes and all the subsets including NK cells, NKT cells, CD4+ T cells, CD8+ T cells, and B cells in the lungs and blood were decreased immediately after restraint stress (Figs 2–4), whereas the numbers of total splenocytes (Fig. 2) and all their subsets (data not shown) did not change. The number of total lung and blood lymphocytes substantially recovered 4 h after the release from restraint (Fig. 2). However, recovery period of each lymphocyte subset number were different, especially in the lungs. The number of lung T lymphocytes hardly recovered even 4 h poststress, although the number of lung B lymphocytes effectively recovered at the time point (Fig. 3). The percentage of each lymphocyte subset in the lungs and blood at the baseline (control mice) are shown in Table 1. The percentages of most of lymphocyte subsets in the lungs were different from those observed in the blood.

Fig. 2.

Effect of restraint stress on the number of lymphocytes in the (a) lungs, (b) blood and (c) spleens. Lungs and blood samples were collected from mice at 0 (RS0), 2 (RS2), and 4 h (RS4) after 2 h restraint stress (n = 6 each) and from untreated control (Con; n = 6) mice, and spleens from mice at 0 (RS0), 1 (RS1), and 2 h (RS2) after restraint (n = 6 each) and from control (Con; n = 6 each) mice; and mononuclear cells were separated for each time group. The absolute number of lymphocytes from each organ was calculated as the product of total cell count, fraction of lymphocytes gated using flow cytometric parameters of forward and side scatter, and fraction of live cells not stained with PI. The results for the lungs and spleen represent the numbers of cells per organ; for the blood samples, the numbers of cells per 100 µl. The data represent the means ± SE. The asterisks indicate a statistically significant difference as compared to the control; ***P < 0·001; anova with Fisher's PLSD post hoc test.

Fig. 4.

Immunohistochemical analysis of the lung tissues. (a, b) The lung sections stained with anti asialo-GM1 antibody from (a) the control mice and (b) stressed mice. Lungs were removed from the control mice (Con) and the mice immediately after the release from 2 h restraint stress (RS), frozen in liquid nitrogen, and sliced into sections. After fixing in acetone, the lung sections were stained with anti asialo-GM1 antibody, and counterstained with haematoxylin. Asialo-GM1+ cells stained brown. The arrows with the tail represent small-sized asialo-GM1+ cells in the pulmonary intraparenchymal area (NK cells), and the arrowheads represent large-sized asialo-GM1+ cells in the alveolus (alveolar macrophages). Photographs were taken at ×400 magnification. A representative section of each group is presented. (c, d) Effect of restraint stress on the number of asialo-GM1+ cells in the lungs. The small-sized asialo-GM1+ cells in pulmonary intraparenchymal area were considered NK cells (c), and the large-sized asialo-GM1+ cells in the alveolus, alveolar macrophages (d). The number of each type of cell stained with anti asialo-GM1 antibody was counted in 25 randomly selected regions from 5 mice (5 regions per mouse) in control group (Con) and restraint stress group (RS). The data represent the means ± SE per lung region. The asterisks indicate a statistically significant difference; **P < 0·01; Student's t-test.

Fig. 3.

Effect of restraint stress on the numbers of lymphocyte subsets in (a) the lungs and (b) blood. Lungs and blood samples were collected from mice at 0 (RS0), 2 (RS2), and 4 h (RS4) after 2 h restraint stress (n = 6 each) and from untreated control (Con; n = 6) mice, and mononuclear cells were separated for each time group. The absolute number of each lymphocyte subset was calculated as the product of total cell count, fraction of lymphocytes gated using flow cytometric parameters of forward and side scatter, fraction of each subset, and fraction of live cells not stained with PI. NK1·1+ CD3- cells were considered NK cells; NK1·1+ CD3+ cells, NKT cells; CD4+ CD3+ cells, CD4+ T cells; CD8+ CD3+ cells, CD8+ T cells; and CD19+ cells, B cells. The results for the lungs represent the numbers of cells per organ; for the blood samples, the numbers of cells per 100 µl. The data represent the means ± SE. The asterisks indicate a statistically significant difference as compared to the control; *P < 0·05, **P < 0·01, ***P < 0·001; anova with Fisher's PLSD post hoc test.

Table 1.

Flow cytometric analysis of subsets of lung and blood lymphocytes from the control mice. NK1·1+ CD3- cells were considered as NK cells; NK1·1+ CD3+ cells, as NKT cells; CD4+ CD3+ cells, as CD4+ T cells; CD8+ CD3+ cells, as CD8+ T cells; and CD19+ cells, as B cells.

	Percentage (%) of lymphocytes
	Blood	Lung
NK cells	5·74 ± 0·21	22·40 ± 0·90 ***
NKT cells	0·72 ± 0·06	2·04 ± 0·19 ***
B cells	41·69  ± 1·20	41·35  ± 2·67
CD8+ T cells	16·04 ± 0·98	13·20 ± 1·13
CD4+ T cells	33·88 ± 0·73	15·73 ± 0·66 ***

The data represent the means ± SE (n = 6). The asterisks indicate a statistically significant difference as compared to blood results;

^***P < 0·001; Student's t- test

Effect of restraint stress on intraparenchymal pulmonary NK cells detected by immunohistochemical analysis

Small-sized cells stained with anti asialo-GM1 antibody, namely NK cells (Fig. 4a, arrows with tail), were decreased immediately after restraint stress in the intraparenchymal pulmonary area (P < 0·01), although the number of the large-sized cells, namely alveolar macrophages stained with anti asialo-GM1 antibody in the alveolus, did not change (Fig. 4a, arrowheads). Numbers of intraparenchymal pulmonary NK cells and alveolar macrophages per lung region are shown in Fig. 4b.

Effect of the administration of propranolol on the modulation of each lymphocyte subset

In order to investigate the mechanism of reduction in the number of lymphocytes in the lungs and blood under restraint stress, propranolol, a nonselective beta adrenergic receptor antagonist, was administrated to the mice. In response to propranolol administration, the numbers of lung and blood NK cells recovered significantly (Fig. 5). Total lymphocytes and other lymphocyte subsets, however, with the exception of NKT cells, showed hardly any recovery in the number in response to the propranolol treatment (Fig. 6). NKT cells were markedly increased in the blood in response to the propranolol treatment in the stressed condition; at the same time, they significantly decreased in the lungs after the administration of propranolol in the nonstressed mice (Fig. 6).

Fig. 5.

Effect of propranolol on the reduction of NK cell numbers in (a) the lungs and (b) blood under restraint stress. Propranolol (Prop, 20 mg/kg) or saline was i.p. administrated to control mice (Con) and restraint mice (RS) at the volume of 200 µl per mouse 30 min before restraint (n = 6 each). Then, lungs and blood samples were collected from mice immediately after 2 h restraint stress and from untreated control mice, and mononuclear cells were separated. The absolute number of each lymphocyte subset from each organ was calculated as the product of total cell count, fraction of lymphocytes gated using flow cytometric parameters of forward and side scatter, fraction of each subset, and fraction of live cells not stained with PI. NK1·1+ CD3– cells were considered NK cells. The results for the lungs represent the numbers of cells per organ; for blood samples, the numbers of cells per 100 µl. The data represent the means ± SE. The asterisks indicate a statistically significant difference; *P < 0·05, **P < 0·01, ***P < 0·001; anova with Fisher's PLSD post hoc test.

Fig. 6.

Effect of propranolol on the reduction of lymphocyte and lymphocyte subset numbers in (a) the lungs and (b) blood under restraint stress. Propranolol (Prop, 20 mg/kg) or saline was i.p. administrated to control mice (Con) and restraint mice (RS) at the volume of 200 µl per mouse 30 min before restraint (n = 6 each). Then, lung and blood samples were collected from mice immediately after 2 h restraint stress and from untreated control mice, and mononuclear cells were separated. The absolute number of lymphocytes from each organ was calculated as the product of total cell count, fraction of lymphocytes gated using flow cytometric parameters of forward and side scatter, and fraction of live cells not stained with PI. The absolute number of each lymphocyte subset from each organ was calculated as the product of total lymphocyte number obtained above and fraction of each subset. NK1·1+ CD3+ cells were considered NKT cells; CD4+ CD3+ cells, CD4+ T cells; CD8+ CD3+ cells, CD8+ T cells; and CD19+ cells, B cells. The results for the lungs represent the numbers of cells per organ; for blood samples, the numbers of cells per 100 µl. The data represent the means ± SE. The asterisks indicate a statistically significant difference; *P < 0·05, **P < 0·01, ***P < 0·001; anova with Fisher's PLSD post hoc test.

Discussion

In this study, all the lymphocyte subsets including NK cells were decreased in the lungs and blood immediately after release from restraint stress, and substantially recovered 4 h after release except for lung T lymphocytes. The decrease in the number of NK cells in the lung was localized in the intraparenchymal area. Stress-induced modulation in the numbers of majority of lung lymphocyte subsets seemed to be similar to that observed in the blood. Considering the lung is rich in blood capillaries, the distribution of lung lymphocytes may be under the direct influence of the blood flow. Lymphocytes may continuously come and go from the blood stream into intraparenchymal pulmonary areas, but at reduced concentrations under conditions of stress. However, the percentages of lymphocyte subsets in the lungs were not proportional to those observed in the blood. This result may indicate that lung has a specialized mechanism of regulating immune cell trafficking.

Previous studies have shown that stress induces a redistribution of lymphocytes. However, it seems that acute stress influences humans and animals differently in the redistribution of blood lymphocytes. In humans, it has been reported that the number of blood lymphocytes, especially NK cells, increases immediately after psychological [29–33] or physical stress including exercise [29, 31, 32], and then decreases below the initial level [30]. In animals, stress immediately induces lymphopenia including a marked decrease in the number of NK cells, possibly due to elevated adrenal hormones [44–46].

Catecholamines and glucocorticoids have been considered main factors causing the changes of lymphocyte distribution both in humans and animals. This is supported by the facts that adrenalectomy attenuates lymphopenia induced under stress [44,46], and that corticosterone causes lymphopenia in adrenalectomized rats [45,46]. Glucocorticoids seem to cause lymphopenia both in humans [47–49] and animals [44, 45, 50], whereas catecholamines induce lymphocytosis [34,49] and a marked increase of the number of blood NK cells [34, 49, 51] in humans. It has also been shown that the redistribution of human NK cells is caused via beta2-adrenoceptor stimulation [34], and that the adhesion of human NK cells to cultured endothelium was down-regulated by beta-adrenoceptor stimulation [52]. In contrast to these studies in humans, it was reported that epinephrine induces lymphopenia as well as neutrophilia in rats [50], although the effect of catecholamines on the redistribution of lymphocytes including NK cells has not been well investigated in animals yet.

In the present study, each subset of lymphocytes in the lungs and blood differed in its reactivity to restraint stress. CD8+ T cells and B cells were more sensitive to restraint stress as compared to CD4+ T cells. Previously, it has been reported that each subtype of lymphocytes differs in the density of beta adrenergic receptor expression both in mice [53,54] and humans [55–58], whereas such a difference in the receptor density among lymphocyte subsets was not observed with glucocorticoid receptors [59–61]. In detail, B cells express more beta adrenergic receptors than T cells in mice [53,54], and similar results have been obtained with human blood lymphocytes [56–58]. Moreover, cytotoxic T cells express more beta receptors than helper T cells in humans [55, 57, 58]. In this study, the concentration of each plasma catecholamine fraction was increased under restraint stress. The differences among lymphocyte subsets in reactivity to stress in our study were found to be in accordance with differences in the density of beta-adrenergic receptor expression among lymphocyte subsets, as demonstrated in previous studies.

In order to investigate whether the decrease in each lymphocyte subset after restraint was due to beta-adrenergic stimulation mediated by elevation of catecholamines, we administrated a beta-adrenergic antagonist, propranolol, to the mice. Interestingly, NK cells were the only subset whose number was reversed in the lungs and blood by the administration of propranolol. In contrast, NKT cells showed a rather complex kinetics in the lungs and blood after the administration of propranolol. NKT cells seem to require beta adrenergic stimulation even at the baseline secretion of catecholamine in order to stay in the lungs. This idea leads to the expectation that under stressed condition in which the level of catecholamine is increased, more NKT cells could be retained in the lung and less in the circulation. Our experiment, however, demonstrated less NKT cells in the lung under stress. Therefore the large increase of NKT cells in the blood stream after the proplanolol treatment of the stressed mice may reflect retaining of NKT cells in organs other than lung under elevated catecholamine level. Concurrent involvement of other adrenergic receptor activation may also be complicating the distribution of NKT cells in the stressed mice, since it has been demonstrated that alpha-adrenergic receptor stimulation mediates the decrease of every lymphocyte subset besides NK cells in mouse blood [62]. Further studies are required to elucidate the mechanism by which catecholamines influence the distribution of NKT cells, as well as other lymphocyte subsets. On the other hand, present study demonstrated that the elevated catecholamine levels returned to baseline within a few hours after the release from restraint, prior to the recovery of NK cell number, showing that NK cells probably need some more hours to recover their numbers after the recovery of catecholamine levels.

Contrary to the present study, Kradin et al. [63] reported that epinephrine yields translocation of lymphocytes including NK cells to the lungs in mice. But in their study, a very large dose of epinephrine was administrated to the mice. The blood concentration of epinephrine in their study may far exceed the range of physiological concentration. It is therefore unlikely that such translocation of lymphocytes to the lungs could happen when catecholamine concentration is elevated within physiological levels, as in the case of restraint stress in the present study.

One important question remains to be solved. Dhabhar et al. [44] have shown that lymphopenia is induced not by cell destruction but by cell migration, confirming that the number of lymphocytes is recovered within a few hours, and that there is no plasma lactate dehydrogenase (LDH) elevation, which would reflect cell destruction. Thus, the question is: To where do the cells migrate after leaving the lungs and blood? Interestingly, Dhabhar & McEwen [64] have proposed that the main tissue infiltrated by lymphocytes during stress is the skin; they have shown that the number of leucocytes increases in the skin, and that cutaneous delayed-type hypersensitivity (DTH) is augmented after acute stress.

As for the molecular mechanisms of redistribution of lymphocytes, changes in the expression of adhesion molecules have been suggested. An increase of CD62L^– human CD4+, CD8+ T cells in the blood was caused after acute psychological stress [33], and a decreased density of CD62L on human blood lymphocytes was observed in response to exercise, although the density of CD11a on blood lymphocytes was increased both under psychological and physical stress [31]. Furthermore, decreased expression of CD44 and CD18 on human blood NK cells was observed in response to exercise when catecholamine levels were elevated [65].

Thus, these findings partly support our hypothesis that acute stress causes lung lymphocyte redistribution including NK cells via beta-adrenergic stimulation elicited by elevated catecholamine levels. Further study is required to establish clinical significance of the reduction in the number of lung lymphocytes caused by stress.