Report

Functional Innervation of Hepatic iNKT Cells Is Immunosuppressive Following Stroke

See allHide authors and affiliations

Science  07 Oct 2011:
Vol. 334, Issue 6052, pp. 101-105
DOI: 10.1126/science.1210301

Abstract

Systemic immunosuppression has been associated with stroke for many years, but the underlying mechanisms are poorly understood. In this study, we demonstrated that stroke induced profound behavioral changes in hepatic invariant NKT (iNKT) cells in mice. Unexpectedly, these effects were mediated by a noradrenergic neurotransmitter rather than a CD1d ligand or other well-characterized danger signals. Blockade of this innervation was protective in wild-type mice after stroke but had no effect in mice deficient in iNKT cells. Selective immunomodulation of iNKT cells with a specific activator (α-galactosylceramide) promoted proinflammatory cytokine production and prevented infections after stroke. Our results therefore identify a molecular mechanism that leads to immunosuppression after stroke and suggest an attractive potential therapeutic alternative to antibiotics, namely, immunomodulation of iNKT cells to prevent stroke-associated infections.

A major cause of death resulting from stroke is infection (1, 2). Immunosuppression, perhaps due to a systemic shift from T helper cell (TH) 1–type to TH2-type cytokine production, has been proposed as a compensatory response to protect the post-ischemic brain from overwhelming inflammation (3). This excessive activation of inhibitory pathways increases the susceptibility to infections (46), although the underlying mechanism has remained elusive to date. Invariant natural killer T (iNKT) cells have a highly restricted repertoire of T cell receptors (TCRs) that recognize lipid antigens presented by CD1d (7, 8). These antigens include various bacterial glycolipids but also endogenous moieties that could function as alarmins and alert the immune system to danger. Because iNKT cells reside in the vasculature of organs like the liver, where circulating antigens can be captured and presented, we proposed that iNKT cells in this tissue are also able to respond to remote sites of injury, such as the brain, and modulate systemic immune responses.

iNKT cells primarily reside in the liver and spleen (9). iNKT cells patrol the hepatic microvasculature and can be tracked in Cxcr6gfp/+ mice (10). When activated with either CD1d ligands or exogenous administration of interleukin (IL)–12 and IL-18, iNKT cells showed altered behavior, including cessation of intravascular crawling associated with activation and release of key cytokines (11). We hypothesized that, on the basis of their intravascular localization, iNKT cells are well positioned to detect distant tissue injury and participate in systemic immunomodulation. To investigate this, we examined liver iNKT cell behavior in response to transient midcerebral artery occlusion (MCAO)–induced brain injury, a rodent model of stroke. Using intravital spinning-disk confocal microscopy, we observed that iNKT cells crawl randomly within liver sinusoids under control conditions (Fig. 1A; fig. S1, A and B; and movie S1) (12). The crawling velocities of iNKT cells in control and sham-operated animals did not differ (fig. S1C). In contrast, we observed markedly restricted crawling of liver iNKT cells after MCAO as reperfusion progressed (Fig. 1, B and C, and fig. S1D). There was a significant decrease in the number of crawling iNKT cells and an increasing number of stationary iNKT cells at 4, 8, and 24 hours after MCAO (Fig. 1D and movie S2). Some of the arrested cells continued to send out pseudopods, “pirouetting” or scanning in a circular pattern (Fig. 1, D and E, and movie S3). These behaviors were particular to the ischemia-reperfusion in the brain, as ischemia-reperfusion injury of the hindlimb had no effect on the behavior of liver iNKT cells (fig. S1E).

Fig. 1

Stroke alters the behavior of hepatic iNKT cells in vivo. The tracks of green fluorescent protein–positive (GFP+) cells within the liver during 10 min of recording in sham-operated (A) and post-ischemic Cxcr6gfp/+ mice at 8 hours (B) and 24 hours (C). Paths are normalized for their origins, and the dotted circle denotes 10 μm radius from origin. N ≥ 4 individual mice per group. (D) The percentage of crawling, stationary, and pirouetting GFP+ cells in the liver of sham-operated and post-ischemic Cxcr6gfp/+ mice at 4, 8, and 24 hours after MCAO. Data are expressed as the percentage of GFP+ cells per field of view (FOV). N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, *P < 0.05 versus sham by t test. (E) A representative pirouetting GFP+ cell showing cell surface ruffling and pseudopod protrusion during 8 min of recording. The “x” denotes stationary cell body, and the yellow arrow denotes the direction of the cell’s pseudopod. Scale bar, 25 μm. (F) CD69 expression in the hepatic iNKT cells of control (gray tint), positive control α-GalCer–treated (blue line), or post-ischemic Cxcr6gfp/+ mice at 24 hours after MCAO (red line). (G) The percentage of iNKT cells with CD69 expression in indicated organs or peripheral blood (PBL) of sham-operated and post-MCAO mice at 24 hours was determined by flow cytometry (LN, six individual lymph nodes collected from the periphery). N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, *P < 0.05 by t test. (H) Intracellular hepatic iNKT cell production of IL-10, IFN-γ, and IL-4 from sham-operated and post-MCAO mice at 8 hours reperfusion was examined by flow cytometry. N ≥ 4 individual mice per group; error bars, mean ± SEM; *P < 0.05 by t test.

Stroke induces major immune changes, including severe lymphopenia in the peripheral blood, thymus, and spleen (13, 14). Interestingly, the number of iNKT cells did not decrease in the peripheral blood, liver, spleen, thymus, and lymph node of post-ischemic mice (fig. S2). However, increased expression of CD69 in iNKT cells was observed in the peripheral blood and liver (Fig. 1, F and G), which suggests regional iNKT cell activation after stroke (Fig. 1G). Taken together, these data demonstrate that brain injury has far-reaching effects, including the capacity to induce profound behavioral changes in hepatic iNKT cells.

Activated iNKT cells produce cytokines and chemokines (15). After MCAO, systemic TH1-type cytokines such as interferon-γ (IFN-γ) and IL-12p70 decreased in wild-type mice (fig. S3, Ai and Bi), reaching significance at 8 hours reperfusion. By contrast, TH2-type cytokines, including IL-10 and IL-5, were increased at 4 hours after MCAO (fig. S3, Ci and Di). We did not detect IL-4 at any time. In the liver, iNKT cells produced significantly increased amounts of IL-10, but not IFN-γ or IL-4, at 8 hours after MCAO (Fig. 1H). The increased ratio of TH2-type over TH1-type cytokines in post-ischemic wild-type mice highlights a general switch in systemic immunity from TH1- to TH2-type in the early stages of reperfusion after stroke (fig. S3E).

Consistent with the view that stroke triggers an immunomodulatory response that decreases the antimicrobial drive of the immune system in humans (5), all of the wild-type mice developed infection 24 hours after MCAO, detectable in blood, lung, liver, and spleen (Fig. 2A and fig. S4). These mice also displayed a significant increase in neutrophilic infiltration into lungs, as measured by myeloperoxidase (MPO) levels, and pulmonary edema (Fig. 2, B and C), both hallmark features of pneumonia, the most common infection in humans after stroke. Wild-type mice always demonstrated some mortality after MCAO throughout the study (Fig. 2D), nearly identical to human mortality data of 12 to 14% (16, 17).

Fig. 2

iNKT cells are critical in the defense against stroke-associated infections. (A) Bacteriological analysis was carried out to investigate the bacterial load in the lungs of sham-operated and post-ischemic wild-type and Cd1d–/– mice at 4, 8, and 24 hours of reperfusion. ND, not detectable; †, mice did not survive for analysis. Values represent the number of colony-forming units (CFU) per mg of tissue. N ≥ 4 individual mice per group; error bars, mean ± SEM; *P < 0.05 by t test. The lungs of sham-operated and post-ischemic wild-type and Cd1d–/– mice at 4, 8, and 24 hours of reperfusion were removed and assayed for neutrophil infiltration, as measured by MPO activity (B) or analyzed for lung edema formation (C). †, mice did not survive for analysis. N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, **P < 0.01 Cd1d–/– versus corresponding wild-type counterparts; ###P < 0.001, ##P < 0.01 post-ischemic wild-type versus sham-operated wild-type; &&&P < 0.001, &&P < 0.01, &P < 0.05 post-ischemic Cd1d–/– versus sham-operated Cd1d–/–, all by t test. (D) Survival rate of post-ischemic wild-type and Cd1d–/– mice treated with or without antibiotics. N ≥ 15 mice per untreated group; N ≥ 5 mice per antibiotics-treated group. (E) Infarct size of post-ischemic wild-type and Cd1d–/– mice assessed at 8 and 24 hours after MCAO. ND, not detectable. N ≥ 4 individual mice per group; error bars, mean ± SEM. The percentage of CD3+ (F), CD4+ (G), or CD8+ (H) T cells with CD69 expression in the indicated organs from sham-operated and post-MCAO wild-type and Cd1d–/– mice at 24 hours reperfusion was determined by flow cytometry (LN, six individual nodes collected from the periphery). NS, not statistically significant. N ≥ 3 individual mice per group; error bars, mean ± SEM; **P < 0.01, *P < 0.05 by t test.

To investigate the role of iNKT cells in stroke and the associated systemic bacterial infection and tissue injury, mice deficient in iNKT cells (Cd1d–/–) were also subjected to MCAO. Bacterial cultures from blood and lungs were clearly evident as early as 8 hours after MCAO in Cd1d–/– mice (Fig. 2A and fig. S4A). Cd1d–/– mice developed even greater pulmonary damage as early as 4 hours after MCAO, with more prominent pulmonary neutrophil infiltration (Fig. 2B) and edema (Fig. 2C), suggestive of even earlier pneumonia-like symptoms. The majority of post-ischemic Cd1d–/– mice did not survive past 12 hours of reperfusion (Fig. 2D); this occurred despite the fact that both strains of mice showed similar brain infarct size (Fig. 2E) (18).

We hypothesized that the high mortality rate of post-ischemic Cd1d–/– mice was the result of their increased susceptibility to post-stroke infections. Indeed, prophylactic administration of antibiotics in post-ischemic mice dramatically improved the survival rate of both strains of mice (Fig. 2D). Most striking was the increase in survival of post-ischemic Cd1d–/– mice. The antibiotic treatment did not affect the infarct size of the brain lesion after MCAO but completely prevented the infections in post-ischemic wild-type and Cd1d–/– mice (fig. S5). Moreover, wild-type mice pretreated with recombinant IL-10, a TH2-type cytokine that iNKT cells were shown to produce after MCAO (Fig. 1H), developed increased stroke-induced lung infections (fig. S6). Clearly, stroke-activated iNKT cells continued to function and afforded some protection to the host, whereas a complete absence of iNKT cells rendered the animals even more susceptible to post-stroke infections, consistent with the lower TH1-type cytokine levels observed in these mice, including no detectable IFN-γ (fig. S3, Aii and E).

We next investigated where iNKT cells fit into the previously described peripheral lymphocyte changes in post-stroke mice (14, 19, 20). T cell activation (CD69 expression) was increased in the peripheral blood and liver after stroke in wild-type but not Cd1d–/– mice (Fig. 2F). In fact, Cd1d–/– mice failed to activate CD4+ T cells in the peripheral blood (Fig. 2G) and CD8+ T cells in the liver after MCAO (Fig. 2H), tissues where iNKT cells were observed to be activated after stroke (Fig. 1G). These data suggest that the immune regulation of post-ischemic iNKT cells is upstream of CD4+ and CD8+ T cell activation in the peripheral blood and liver, respectively, and that iNKT cells function as the conductor of immunity, whereby their acute responses modulate and facilitate the adaptive immune response. Although a systematic assessment of numbers of lymphocytes, NK cells, and granulocytes after stroke revealed some additional changes in blood and tissues, these were not affected by the presence or absence of iNKT cells (fig. S7), suggesting that not all changes to leukocyte cell numbers are modulated by iNKT cells.

The manner in which iNKT cells detect tissue damage after stroke could be through endogenous glycolipids presented by CD1d, cytokines like IL-12 and/or IL-18 released from other sentinel cells (e.g., macrophages), or some other as-yet-unknown mechanism. Antibody blockade of CD1d had no effect on iNKT cell arrest in response to MCAO (fig. S8, A to C), whereas it prevented cessation of iNKT cells caused by stimulation with the CD1d ligand α-galactoceramide (α-GalCer), a specific activator of iNKT cells (fig. S8D) (21). Another inhibitor that blocks the presentation of glycolipid ligands in the context of CD1d, isolectin B (iB) 4, also did not alter iNKT cell arrest after MCAO (fig. S8, A to C), ruling out glycolipid presentation by CD1d as the pathway alerting iNKT cells to distal tissue injury in stroke. Furthermore, blockade of IL-12 and IL-18, cytokines known to activate and arrest iNKT cells (fig. S8E) (11), had no effect on iNKT cell arrest in response to MCAO (fig. S8, A to C). Finally, apyrase, an inhibitor of ATP, a well-known “alarmin” in the brain and liver (22), had no effect on iNKT cell responses to MCAO (fig. S8, A to C).

An as-yet-unidentified, long-distance pathway was affecting the crawling behavior and activation of iNKT cells in the liver after cerebral ischemia. Previous literature suggested that the nervous system may affect immune cells, including iNKT cells, and alter their function (23, 24), thereby potentially regulating the magnitude of the host response to infection or injury (25, 26). Administration of the nonspecific β-adrenergic receptor blocker, propranolol, reversed the iNKT cell phenotype induced by MCAO (Fig. 3, A to C; fig. S9A; and movie S4). Furthermore, post-ischemic cessation of iNKT cell crawling was completely inhibited by specific chemical depletion of peripheral neuronal terminals containing noradrenaline with 6-hydroxydopamine (6-OHDA), suggesting a neural rather than humoral input (Fig. 3, A, B, and D, and fig. S9A). Despite the phenotypic changes of iNKT cells after systemic administration of propranolol or 6-OHDA, these treatments did not alter the blood flow (fig. S9B) or infarct size in post-ischemic mice (fig. S9C). Mortality at 24 hours of reperfusion was reduced by 50% with 6-OHDA and completely inhibited by propranolol (fig. S9D).

Fig. 3

iNKT cell crawling cessation after MCAO is dependent on sympathetic innervation. The percentage of crawling (A) and stationary (B) GFP+ cells in the liver of sham-operated and post-ischemic Cxcr6gfp/+ mice treated with propranolol (PPL) or 6-hydroxydopamine (6-OHDA) at 24 hours after MCAO. Data are expressed as percentage of GFP+ cells per FOV. N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, **P < 0.01 by t test. The tracks of GFP+ cell within the liver during 10 min of recording in post-ischemic Cxcr6gfp/+ mice treated with propranolol (C) or 6-OHDA (D) at 24 hours after MCAO. Paths are normalized for their origins, and the dotted circle denotes 10 μm radius from the origin. N ≥ 4 individual mice per group. The percentage of crawling (E) and pirouetting (F) GFP+ cells within the liver of sham-operated and Cxcr6gfp/+ mice treated with localized noradrenaline superfusion. Data are expressed as percentage of GFP+ cells per FOV. N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, **P < 0.01 by t test. Representative photographs of isolated iNKT cells for in vitro analysis in untreated (G) and noradrenaline-treated conditions (H). Scale bar, 50 μm. N ≥ 4 individual experiments per group. (I) The percentage of nonpolarized (nonactivated) and polarized (activated) iNKT cells was determined in untreated, noradrenaline-treated, and propranolol-pretreated plus noradrenaline-treated in vitro. Data are expressed as percentage of iNKT cells per FOV. N ≥ 4 individual experiments per group; error bars, mean ± SEM; ***P < 0.001, *P < 0.05 by t test.

Localized noradrenaline administration directly mimicked the behavior of iNKT cells in the liver of post-ischemic Cxcr6gfp/+ mice in vivo. Significantly fewer iNKT cells crawled, and more cells adopted a pirouetting phenotype in the epicenter of noradrenaline administration (Fig. 3, E and F, and fig. S10). By contrast, in an area of liver distant from the localized noradrenaline superfusion, iNKT cells did not alter their crawling behavior (Fig. 3, E and F, and fig. S10). In addition, when noradrenaline was applied to iNKT cells in vitro, these cells acquired a “flattened” and pseudopod protruding phenotype reminiscent of iNKT cell behavior after MCAO in vivo (Fig. 3, G and H). In fact, pretreatment of iNKT cells with propranolol inhibited this behavioral change (Fig. 3I), suggesting that noradrenaline directly induces the biology we observed in vivo.

Next, we examined whether direct immunomodulation of iNKT cells can reverse the stroke-induced immunosuppression and infection. Administration of α-GalCer in a therapeutically relevant manner significantly increased systemic levels of endogenous IFN-γ (Fig. 4A) and reduced stroke-induced neutrophil pulmonary influx, lung edema (Fig. 4, B and C), and infections in post-ischemic mice (Fig. 4, D to G). α-GalCer is an immunostimulant that could potentially have deleterious effects on cerebral ischemia, but we found no notable differences in infarct sizes (fig. S11A). Furthermore, α-GalCer has been documented to induce liver damage, but we found no significant difference in liver enzyme levels within the blood of post-ischemic mice after the single dose of α-GalCer (fig. S11B).

Fig. 4

Selective modulation of iNKT cells decreased stroke-induced lung injury and infectious complications. (A) For determination of IFN-γ production, blood samples were collected and cytokine expression was analyzed as described in the supporting online material. N ≥ 5 individual mice per group; **P < 0.01 by t test. The lung tissues were removed at 24 hours after MCAO and measured for neutrophil infiltration by MPO activity (B) or lung edema (C). N ≥ 4 individual mice per group; error bars, mean ± SEM; ***P < 0.001, **P < 0.01 by t test. Bacteriological analysis was performed to investigate the bacterial culture from peripheral blood (D), lungs (E), livers (F), and spleens (G) in post-ischemic mice treated with α-GalCer, propranolol, or 6-OHDA 24 hours after MCAO. ND, not detectable. Data are presented as number of CFU per ml of blood or mg of tissue. N ≥ 4 individual mice per group; error bars, mean ± SEM; **P < 0.01, *P < 0.05 versus MCAO by one-way analysis of variance with Bonferroni’s post-test.

Interestingly, wild-type mice receiving propranolol or 6-OHDA also demonstrated significantly reduced bacterial infections at 24 hours after MCAO in a manner similar to that observed in α-GalCer–treated mice (Fig. 4, D to G). Furthermore, the effects of propranolol were entirely dependent on iNKT cells, because the addition of propranolol to Cd1d–/– mice provided no protection from infection or mortality (fig. S12). Notably, post-ischemic wild-type mice treated with propranolol reversed the preference for intracellular IL-10 production back to an intracellular IFN-γ dominant production and toward a TH1-dominant phenotype (fig. S13). These data strongly suggest that direct modulation of iNKT cells with α-GalCer or through the blockade of noradrenergic neurotransmitters was sufficient to modulate iNKT cells in a manner that results in reduced infection and associated lung injury after stroke.

iNKT cells are emerging as an important population of cells crucial for the regulation of immunity. We have described an essential role of the sympathetic nervous system and iNKT cells in the defense against infectious complications after stroke. Although aspiration pneumonia is a contributing factor to increased infection in stroke patients, it cannot explain the immunosuppression noted by us and others. Our data suggest that a functional innervation of iNKT cells in the liver contributes to this immunosuppression. Our study also provides insights into the cross-talk that occurs between the central nervous system and the immune system, which is only beginning to be understood, and may be a step toward the development of an effective therapy for the number one killer in stroke patients, namely, infection. Although antibiotics may be a viable option for treatment, with the ever-increasing problem of antibiotic resistance, immunomodulation is an attractive alternative.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1210301/DC1

Materials and Methods

Figs. S1 to S13

References (2732)

Movies S1 to S4

References and Notes

  1. Acknowledgments: We thank D. R. Littman (New York University School of Medicine) for the Cxcr6gfp/+ knock-in mice, and the Live Cell Imaging Facility funded by the Canada Foundation for Innovation and P. Colarusso for training and assistance related to microscopy. We thank C. Badick for excellent technical support. The NIH Tetramer Core Facility provided mouse PBS57-loaded CD1d-tetramer for identification of iNKT cells by flow cytometry. We also thank the University of Calgary Flow Cytometry Facility and L. Kennedy for their assistance with the flow cytometric analysis. The work is supported by the Canadian Association of Gastroenterology (C.H.Y.W), the Canadian Institutes of Health Research (C.H.Y.W., W-Y.L., and P.K.), the Canada Research Chairs Program (P.K.), the Alberta Innovates Health Solutions (C.N.J and P.K.), and the Calvin, Phoebe, and Joan Snyder Chair for Translational Research in Critical Care Medicine (C.L.). The data reported in this paper are tabulated in the supporting online material. The authors declare no competing financial interests. C.H.Y.W designed and did most of the experiments, analyzed the results, and prepared the manuscript; C.N.J. did some flow cytometry experiments; W-Y.L. isolated iNKT cells for the in vitro experiments; C.L. performed the multiplex mouse cytokine/chemokine assay; and P.K. provided overall supervision, helped design all of the experiments, and prepared the manuscript.
View Abstract

Navigate This Article