Background
Stroke elicits a strong inflammatory response that critically contributes not only to tissue demise but also to repair and regeneration and also is thought to underlie the number one secondary complication found in stroke patients: infection [
1]. Specifically, the inflammatory response after stroke comprises changes in the expression of inflammatory mediators and immunocompetent cell populations not only locally, in the brain, but also at the blood-brain interface and in peripheral systems, including the spleen. For example, upregulations of interleukin (IL)-1β and CXCL-1 in the brain have been consistently found in experimental models of stroke [
2,
3]. Leukocytes infiltrate the brain parenchyma within hours and days after stroke onset, through a compromised blood-brain barrier (BBB) or via active recruitment [
2,
4‐
6]. Pro-inflammatory cascades are rapidly activated following stroke and are paralleled or followed by an immunosuppressant signaling. Within the first days after stroke, while a decrease in pro-inflammatory pathways may occur in the brain [
7,
8], immunosuppression is more evident in the periphery. Apoptotic loss of natural killer (NK), B and T lymphocytes in the blood and spleen, a decreased capacity of interferon (IFN)-γ production by blood-derived leukocytes along with atrophy of the spleen and thymus have been described both in stroke patients and/or rodents subjected to stroke and implicated in an increased susceptibility to infection [
7‐
12]. In agreement with the view that limiting post-stroke inflammation will decrease neuronal death or promote neurological recovery, IFN-β has been put forward as a candidate drug for the treatment of stroke (National Institute of Neurological Disorders and Stroke-sponsored phase I clinical trial).
IFN-β is a type I IFN that binds to the IFN-α/β receptor (IFNAR). Systemic administration of recombinant (r)IFN-β is currently used in the treatment of multiple sclerosis (MS), although the exact mechanisms by which IFN-β administration is beneficial in MS are not fully understood [
13]. IFN-β may promote a shift from a T helper (T
h)1 to T
h2 response by blood-derived cells and reduce the proliferation of T lymphocytes. It also may diminish the entry of inflammatory and immune cells into the central nervous system (CNS) by downregulating adhesion molecules and matrix metalloproteinase expression by leukocytes, as well as by stabilizing endothelial tight junctions at the BBB [
13]. In a mouse model of MS, endogenous IFN-β signaling was sufficient to promote a better disease outcome, which was associated with a decrease in CNS inflammation but not with changes in autoreactive T cells [
14,
15]. In support, exogenous IFN-β also reduced autoreactive T cell proliferation by inhibiting the antigen-presenting capacity of microglia and astrocytes, the CNS-specific antigen-presenting cells (APCs) [
16,
17].
Studies addressing the effects of IFN-β administration in stroke are scarce and divergent. Systemic administration of human rIFN-β before and up to 4 h after induction of stroke in rabbits resulted in neuroprotection [
18]. Similarly, rats showed smaller infarct volumes when treated with rat rIFN-β even up to 6 h after stroke [
19]. In mice, intracerebroventricular administration of mouse rIFN-β immediately before or after stroke resulted in a decrease in infarct volume at 24 h [
20]. However, systemic administration of wild-type (WT) IFN-β or pegylated IFN-β for 3 or 7 days after transient middle cerebral artery occlusion (tMCAo) had no effect on behavioral performance or infarct volume, but aggravated weight loss in rats [
21]. In a recent study, IFN-β administration following experimentally induced subarachnoid hemorrhage, in rats, also provided no effect on outcome measures [
22].
We investigated the action of endogenous IFN-β signaling on inflammation and on the development of sensorimotor deficits and infarct volume. For that purpose, we used IFN-β and IFNAR knockout mice (IFN-βKO and IFNAR-KO, respectively) and two clinically relevant models of ischemic stroke, tMCAo and permanent middle cerebral artery occlusion (pMCAo). We focused our study on the first 8 days after stroke onset, a time period critical for recovery of neurological function and infarct core formation [
23].
Methods
Ethical considerations
We conducted animal experiments in accordance with protocols approved by the Malmö/Lund Ethical Committee for Animal Research (M332-09, M243-07) and Danish Animal Health Care Committee (2011/561-1950), as well as to the ARRIVE guidelines. Experiments were carried out in a blinded and randomized fashion.
Mouse strains and housing conditions
We used 10 to 40-week-old IFN-βKO [
24], as well as 8 to 10-week-old IFNAR-KO [
14] male mice backcrossed to C57BL/6 for over 20 generations; WT controls (C57BL/6, age- and gender-matched) were purchased from Taconic (Ry, Denmark). Mice were housed in climate-controlled rooms under diurnal conditions, with ad libitum access to water and food.
Transient middle cerebral artery occlusion
We modeled stroke in mice by tMCAo as described previously [
3]. Mice were placed in an incubator at 35 °C during the first 2 h post-surgery and in an incubator at 33 °C overnight. We injected 0.3 mL of 5 % glucose in saline (sterile) subcutaneously, 30 min, 24 h, and thereafter every 12 h up to 4 days after surgery; weight loss ceases typically between days 3 and 4 after tMCAo. We measured body temperature 1 h, 2 h, and daily up to 7 days after surgery. In addition, we assessed body weight before and after surgery on a daily basis.
Inclusion criteria
An immediate reduction in regional cerebral blood flow and metabolism (rCBF) upon occlusion of the MCA and reperfusion were set as the primary inclusion criteria (2 of 78 mice were excluded). Moreover, mice that exhibited signs of pain, weakness, and/or distress were euthanatized and included in the mortality rates (one of 76 mice). One IFNAR-KO and one WT control were excluded from the infarct volume analysis due to bad quality of the tissue. Characteristic ipsilateral subcortical and cortical infarcts were verified in all mice [
3].
Permanent middle cerebral artery occlusion
We modeled stroke in mice by pMCAo as reported before [
5]. After surgery, mice were allowed to recover from anesthesia in a 28 °C controlled environment and given a subcutaneous injection of 0.05 mg/Kg (body weight) buprenorphine in saline (Temgesic, Schering-Plough, Ballerup, Denmark) three times, every 8 h, starting immediately after surgery.
Inclusion criteria
All mice exhibiting a neocortical infarct, assessed using 2,3,5-triphenyltetrazolium chloride (TTC) staining, were included (one mouse of a total of 20 mice was excluded).
Behavioral tests
Composite neuroscore
We evaluated gross sensorimotor deficits 24 h after tMCAo by behavioral tests adapted from [
25‐
27]. Mice were scored with respect to (1) spontaneous rotation (from 0, rotation on the body axis, to 5, no rotation), (2) resistance to lateral force applied to the left, (3) left forelimb flexion on suspension by the tail, (4) left hindlimb flexion when only the hind limbs are lifted from the surface, and (5) forelimb impairment (from 0, paralysis, to 5, normal). For tests 2 to 4, a scoring system of 0 (no resistance or flexion) to 4 (normal) was used; resistance to lateral force applied to the right and right limb flexion also were evaluated. Sham-operated animals scored 32.
Rotating pole test
The rotating pole test was performed essentially as described before [
3], with a few modifications. Mice were trained to cross the pole at 0, 3, and 10 rotations per minute (rpm), to the left and right, 2 days and 1 day prior to surgery (tMCAo and sham). With respect to the number of foot slips, only the contralateral forelimb was taken into consideration. Scores 2 and 3 were attributed additionally to animals that clearly met the criteria of scores 3 and 4, but fell from the pole, respectively. We video recorded the sessions for a final analysis. All mice were able to cross the pole prior to surgery (scores 5–6). Sham-operated animals scored 5 or 6 at all the modalities, and we did not observe differences between genotypes.
Grip strength test
In experiments involving IFN-βKO mice, we used the grip strength test (Bioseb, Vitrolles, France) as reported previously [
3]. We present the maximum grip strength value of five trials as percentage of baseline (collected the day before surgery). The grip strength of IFNAR-KO mice and WT counterparts was evaluated as reported previously [
28], and we present the maximum grip strength value of five trials (for each paw).
Carbon black perfusion
The cerebral vasculature was visualized using a protocol described previously [
3].
Isolation of immune and inflammatory cells
Mice were anesthetized using 1.8–2.5 % isoflurane (IsobaVet, Schering-Plough Animal Health, Milton Keynes, UK) in O2:N2O (30:70). From each mouse, 400 μL of blood was extracted by intracardiac puncture, using heparinized syringes. Blood was gently homogenized and maintained at 4 °C. Thereafter, mice were perfused transcardially with 10 mL of saline (5 mL/min). Brains were quickly dissected, freed from meninges, and kept in Hank’s Balanced Salt Solution (HBSS, Invitrogen, Paisley, UK) supplemented with 0.2 % bovine serum albumin (BSA, Sigma-Aldrich, Deisenhofen, Germany) and 0.01 % ethylenediaminetetraacetic acid (EDTA, Invitrogen) at 4 °C. Spleens were weighed and transferred to 2 % fetal bovine serum (FBS, Invitrogen) in Dulbecco’s modified Eagle medium (DMEM, Invitrogen) at 4 °C. Blood plasma was obtained following standard centrifugation (1300×g for 10 min, 4 °C) of blood; plasma was stored at −80 °C.
Brain
Ispilateral or contralateral hemispheres of three to four mice of the same genotype (IFN-βKO or WT) were pooled together. Tissue was mechanically dissociated in HBSS supplemented with 0.2 % BSA and 0.01 % EDTA using a Dounce homogenizer and passed through a 40-μm nylon cell strainer (BD Biosciences, Stockholm, Sweden). The cell suspension was centrifuged at 400×g for 10 min at room temperature. The resulting pellet was resuspended in 30 % Percoll (GE Healthcare, Sweden) in HBSS and gently overlaid on a 37–70 % Percoll gradient. Following centrifugation at 500×g for 20 min at room temperature, cells were collected at the 37–70 % interface and rinsed with 10 % FBS in HBSS. After a last centrifugation at 400×g for 10 min at room temperature, the cell pellet was resuspended in 2 % FBS in phosphate-buffered saline (PBS, Invitrogen).
Blood and spleen
We added red blood cell (RBC) lysis buffer (eBioscience, San Diego, CA, USA) to 200 μL of blood. We dissociated whole spleens in RBC lysis buffer, using a 40 μm nylon cell strainer (BD Biosciences). We stopped RBC lysis by adding 2 % FBS in PBS. After centrifugation (300×g for 5 min at 4 °C), cells were resuspended in 2 % FBS in PBS. Splenocytes were stained with trypan blue and total numbers estimated using a Bürker chamber.
Flow cytometry
Isolated immune and inflammatory cells were incubated with primary antibodies for 20 min at 4 °C. After rising with 2 % FBS in PBS, cells were incubated with the secondary antibody Streptavidin PErCP (1:200, BD Biosciences) for 20 min at 4 °C (only for detection of CD122). Subsequent to rinsing, cells were exposed to Cytofix for 20 min at 4 °C and rinsed with BD Perm/Wash buffer (both reagents were purchased from BD Biosciences). Finally, we resuspended the cells using 2 % FBS in PBS. Flow cytometry was carried out using a FACSCalibur flow cytometer (BD Biosciences), with analysis being performed using CellQuest (BD Biosciences) for acquisition after exclusion of duplets and FlowJo 8.8.6 (Tree Star, Ashland, OR, USA). For each sample, we analyzed a total of 100,000 events (cells). We present the results as percentage of total cells analyzed, unless otherwise indicated.
We used the following primary antibodies (purchased from BD Biosciences, unless otherwise indicated), each at a dilution of 1:200: B220-FITC, MCHII-PE, CD4-APC, CD25-FITC, CD8-PE, CD122-biotin, CD11b-APC, CD45.2-FITC (BioLegend, USA), and NK1.1-PE.
Quantification of Th1/Th2 cytokines protein levels in the blood plasma
We determined the protein concentrations of IFN-γ, IL-1β, IL-10, IL-12, IL-2, IL-4, IL-5, tumor necrosis factor (TNF)-α, and mouse keratinocyte-derived factor (mK or GRO/CXCL-1) in blood plasma by a sandwich immunoassay (Mouse Th1/Th2 9-Plex Ultra-Sensitive Kit, Meso Scale Discovery, Gaithersburg, MD, USA). We performed the assay according to the manufacturer’s instructions (Ultra-Sensitive Kit). We used 35 μL of undiluted plasma per well (96-well plates). Plates were read using a SECTOR Imager 6000 (Meso Scale Discovery).
Immunohistochemistry
For a subset of mice, perfusion with saline was followed by perfusion with 45 mL of 4 % formaldehyde in PBS (5 mL/min). Brains were kept in formaldehyde overnight and thereafter in 40 % sucrose in PBS, always at 4 °C. Free-floating, 30-μm-thick coronal brain slices were rinsed three times with PBS and kept in a 5 % blocking solution (5 % normal serum, Jackson ImmunoResearch, Suffolk, UK, and 0.25 % Triton X-100 in PBS) for 1 h, at room temperature. Following blocking, we incubated slices with primary antibody(ies) in 2 % blocking solution overnight at 4 °C. After rinsing, slices were incubated with secondary antibody(ies) in 2 % blocking solution for 2 h at room temperature.
Primary antibodies and respective dilutions used in this study are as follows: rabbit anti-claudin-5 (1:100, Bioworld Technology); rat anti-CD45 (1:500, MCA1031G, AbD Serotec, Europe); rat anti-galectin-3, Gal-3 (biotinylated, 1:500; Acris Antibodies, Herford, Germany); rat anti-Gal-3 (1:1000; kindly provided by Professor Hakon Leffler, Lund University, Sweden); rabbit anti-ionized calcium-binding adapter molecule 1, Iba-1 (1:1000; Wako, Osaka, Japan); and mouse anti-neuronal nuclei, NeuN (1:800; Sigma-Aldrich). The two antibodies targeting Gal-3 yielded equivalent results. Primary antibodies were detected using streptavidin-Alexa488 (Molecular Probes, Invitrogen); anti-rat Cy3; anti-rabbit Cy3; and anti-mouse biotinylated or anti-mouse Cy2; all reagents were diluted 1:400 (antibodies were purchased from Jackson ImmunoResearch). The biotinylated secondary antibody was detected by avidin-HRP (Vector Laboratories, Peterborough, UK). With respect to BBB integrity assays, to detect mouse IgG in the brain parenchyma, we used a specific HRP-conjugated reagent by cell signaling (catalog number 8125), following the instructions provided by the manufacturer. HRP was, in all cases, detected through 3,3′-diaminobenzidine (DAB) precipitates. In some cases, nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich).
Assessment of CD45+ cells, in situ
To estimate the number of CD45+ cells in the brain, in situ, we performed CD45 immunolabeling, as described above. From each brain, we used four slices with the following stereotaxic coordinates: approximately 0.86, 0.26, −0.34, and −1.34 mm (anterior-posterior, AP) in relation to the bregma. We counted CD45+ cells populating one cortical and/or one striatal infarct core region, each corresponding to a field of vision of about 1 mm2 per section. Regions were sampled equivalently across mice. Cells were counted under a fluorescent microscope (Nikon Eclipse 80i, Nikon, Solna, Sweden) by an experimenter blinded to the genotype. In addition, CD45+ cells were imaged using Nikon A1 Confocal on a Ti-E microscope and data processed using the NIS-Elements software (Nikon).
Iba-1 and Gal-3 immunoassays
Additionally, we performed Iba-1 and Gal-3 (double) immunolabeling. Here, we used, from each brain, slices corresponding to the following coordinates: approximately 1.10 mm (level 1), 0.50 mm (level 2), and −1.34 mm (level 3) AP in relation to the bregma. Fluorescent dyes were imaged using a Zeiss LSM 510 inverted confocal microscope. For each brain slice, three (within level 1) or four (within levels 2 and 3) Z-stack images with the dimensions of 640 μm × 640 μm × ~40 μm were collected for analysis; one Z-stack image was acquired from the peri-infarct region (negative control for Gal-3) and the remaining from the infarct core.
The density of Iba-1
+ and Gal3
+ macrophages/microglia within the infarct core 8 days after tMCAo is relatively high, and cellular processes may overlap. Under these conditions, counting cells is suboptimal, particularly when using epifluorescence microscopy. We therefore quantified the fraction of Iba-1
+ signal colocalizing with Gal-3 and vice versa by calculating Mander’s coefficients of the acquired Z-stack images (for simplification, referred to as the percentage of Iba-1
+ cells expressing Gal-3 and vice versa), using ImageJ software (National Institutes of Health, Bethesda, MD, USA), [
29]. Signal intensity thresholds were set manually after applying a denoise filter. Our approach minimized false positives by treating each optical slice of each Z-stack independently, instead of analyzing collapsed Z-stack images. Analysis of Gal-3
+ cell morphology was carried out using Image J. For each brain, we selected 30–40 Gal-3
+ cells within the infarct core and manually plotted their outline. For each cell, area, perimeter, circularity, Feret’s diameter, and aspect ratio values were calculated. Analysis was carried out on projected Z-stacks, and only non-overlapping cells were included (as determined on non-projected Z-stacks in both XY and Z dimensions).
Gal-3 immunoreactivity
We used three additional brain slices per animal (approximately 0.98, 0.38, and −1.06 mm in relation to bregma) for Gal-3 (single) immunolabeling; Gal-3
+ signal was estimated following a procedure previously described in detail [
3].
Assessment of blood-brain barrier integrity through IgG and claudin-5 immunoassays
When investigating the presence of IgG in the parenchyma of WT and IFN-βKO brains, we used brain slices centered at 0 mm AP in relation to the bregma (equivalent slices were used across animals). The IgG-specific signal within the infarct core and peri-infarct region was, for each section, qualitatively and qualitatively assessed. Adjacent brain slices were used for analysis of claudin-5 immunoreactivity, as described before [
30]. Briefly, only claudin-5
+ vessels showing a relatively strong and continuous (claudin-5) signal were analyzed. The area occupied by and average length of claudin-5
+ vessels in a selected region of the peri-infarct cortex (1 mm
2) were quantified, for each section, using the Image J software. All images were acquired using a Nikon Eclipse 80i microscope and the NIS-Elements software.
Infarct volume
IFN-βKO
For each animal, a series of 30-μm-thick coronal brain slices of the entire brain (240 μm between consecutive slices) was used for NeuN immunolabeling; we used equivalent series among animals. Micrographs were acquired with a Nikon Eclipse 80i microscope, using the NIS-Elements software, under standardized conditions. For each slice, the areas corresponding to the contralateral hemisphere (Contra) and to NeuN+ tissue within the ipsilateral hemisphere (IpsiNeuN) were encircled using ImageJ. The infarct area (IA) was estimated accounting for edema or tissue shrinkage: IA = Contra − IpsiNeuN. Infarct volume was obtained by volumetric integration.
IFNAR-KO
Mice subjected to tMCAO were given an overdose of pentobarbital and fixative-perfused. A series of 30-μm-thick coronal brain slices (180 μm between consecutive slices) was stained with toluidine blue, and infarct volumes were estimated as described in [
5]. For mice subjected to pMCAO, brains were sliced into 1-mm-thick coronal slices; slices were stained with TTC as described before [
3]. Infarct volumes were estimated as described above for IFN-βKO mice and WT counterparts.
Statistics
We present the data as mean ± SD, unless otherwise indicated. To compare two groups, we used Student’s t test (two-tailed), and comparison of three or more groups was done by two-way analysis of variance (ANOVA), with the following exceptions. For the composite neuroscore and rotating pole scores, we used the Mann-Whitney U test. Mortality rates were assessed using Fisher’s exact test. For IFN-βKO and WT mice undergoing tMCAo or sham surgery, differences in temperature and weight loss post-surgery were analyzed using a mixed model ANOVA, with multiple comparisons being performed for representative time points (24 h, 2, 3, and 7 days). Analysis of the weight before surgery was done using one-way ANOVA. For IFNAR-KO and WT mice subjected to tMCAo, temperature and grip strength asymmetry post-surgery were evaluated using repeated measures ANOVA. All multiple comparisons were performed post hoc using the Student’s t test, with Bonferroni correction. Results were considered significant when p < 0.05. Unless otherwise specified, in the figures, we present only the results obtained for pair-wise comparisons, with Bonferroni correction.
Discussion
In rodents, the first week post-stroke is thought to constitute a critical period for the formation of the infarct core and activation of mechanisms that promote or repress recovery, in both of which inflammation plays a major role [
1]. Although the critical impact of inflammation on stroke outcome is well accepted, a unified view on underlying molecular and cellular determinants remains to be established. Given its reported anti-inflammatory capacity and use in the treatment of relapsing-remitting MS, IFN-β has been put forward as a candidate drug for acute stroke treatment. However, the inflammatory response post-stroke is highly dynamic, both spatially and temporally, and the precise modes of action and effects IFN-β in stroke remain largely unexplored. First, literature reporting on the therapeutic potential of IFN-β in stroke is sparse and controversial [
18‐
22]. Secondly, the function of endogenous IFN-β signaling, which may enable a better understanding of underlying pathophysiological mechanisms and thus a potentially higher success in designing the future treatment of stroke, had not been explored.
We provide evidence that endogenous IFN-β regulates the post-stroke inflammatory response, in the brain and periphery, and we reveal its specific actions. Previous studies have consistently shown that CD11b
+ cells found within the brain parenchyma can express relatively high or low levels of CD45, the first including infiltrating leukocytes (notably monocyte-derived macrophages) and the second resident microglia [
2,
5,
32]. In C57BL/6 mice, CD45
highCD11b
+ cells accumulate in the brain parenchyma within the first days after focal cerebral ischemia [
2,
4]. Our results indicate that endogenous IFN-β signaling limits the local accumulation of blood-originating cells that occurs after stroke. Indeed, following tMCAo, IFN-βKO mice showed a higher percentage of CD45
highCD11b
+ cells in ipsilateral hemispheres, and a higher number of CD45
+ cells within the infarct core than their WT counterparts. This finding is in line with previous studies showing that, in rats, administration of rIFN-β consistently reduced the number of leukocytes in the brain up to 7 days after stroke, despite the usage of different administration protocols [
19,
21]. The percentage of CD45
highCD11b
+ cells expressing the activation marker MHCII did not differ significantly between the IFN-βKO and WT groups. Yet, given the larger percentage of CD45
highCD11b
+ cells, these data likely reflect a higher total number of CD45
highCD11b
+MHCII
+ cells in the absence of IFN-β. In contrast to its effects on CD45
highCD11b
+ cells, abrogation of IFN-β did not result in concomitant changes in the percentages of CD45
dimCD11b
+ cells or respective MHCII expression. Albeit the debate surrounding the exact nature and dynamics of the macrophages/microglial response after stroke, the lack of an increase in the relative numbers of microglia and activated microglia 2 days after tMCAo (Fig.
2d) probably reflects a delay in the accumulation and activation of the microglial population at the injury site. Indeed, the dramatic increase in the expression of Iba-1 and Gal-3 observed at 8 days after tMCAo is unlikely to result predominantly from invading macrophages [
4]. This delay might be related, at least in part, to the extent of cerebral damage [
10,
12,
30,
40]. We also show that Iba-1 and Gal-3 immunoreactivities 8 days after stroke induction are unaltered by the abrogation of IFN-β, suggesting that, in stroke, IFN-β rather acts by diminishing a relatively earlier leukocytic central invasion. Indeed, it is likely that the enhanced accumulation of leukocytes observed in the brains of IFN-βKO reflects an increase in the infiltration of these cells into the brain parenchyma. This could occur via a relative decrease in BBB integrity, through an enhanced activity of metalloproteinase-9 [
19]. However, when analyzing IgG and claudin-5 signals, reflecting vessel extravasation and tight junction function, respectively, in the infarct core and/or peri-infract region, we did not find evidence for alterations in post-stroke BBB integrity induced by knocking out the gene encoding IFN-β, at least not 2 days after tMCAo. Yet on the other hand, it has been previously reported that, during EAE, IFN-βKO mice show an increase in the levels of circulating chemokines (namely CCL3 and CCL5), which also could underlie an increased leukocytic infiltration into the brain parenchyma in comparison to their WT counterparts [
41]. Similarly, lack of IFNAR in mice has been shown to relate to an increase in leukocyte numbers and metalloproteinase-9 expression, as well as to a decrease in the chemoattractant CXCL10 in the CNS under inflammatory conditions [
42]. Nevertheless, as indicated by a recent report, local regulatory mechanisms exerted by endogenous IFN-β [
43], such as controlling local proliferation and activation of innate immune cells, cannot be excluded at present.
In addition to attenuating central inflammation, our results indicate that IFN-β is important in regulating peripheral immune cell subsets. Most notably, IFN-βKO mice showed a higher total percentage of circulating B220
+ (immature) B cells, but similar percentages of circulating B220
+MHCII
+ (mature and activated) B cells in relation to WT counterparts, 2 days after tMCAo. This set of data indicates that the ratio of proliferative B cells is lower in WT mice, which is consistent with the general anti-proliferative role of IFN-β. In fact, the anti-proliferative and pro-apoptotic actions of IFN-β are well documented [
15,
43]. In support of a previous report [
15], no major T
h1-T
h2 shift was detected between genotypes. Interestingly, however, IL-1β and IL-5 plasma concentrations decreased in the IFN-βKO group, and these cytokines have been extensively associated with multiple aspects of B cell function, including proliferation [
44,
45]. Although we found similar numbers of splenocytes and splenic B cells in WT and IFN-βKO 2–8 days after tMCAo, higher numbers of circulating B cells also could be due, at least in part, to an increased mobilization of these cell population from splenic or other (notably thymic) pools. It is worth noting that the decrease in circulating B220
+ B cells following tMCAo might be in agreement with previous studies showing a stroke-induced apoptotic loss of B lymphocytes [
7,
11].
Besides its effects on circulating immune cells, we also provide evidence for a role of IFN-β in the regulation of splenic immune cell subsets. Indeed, the IFN-βKO group did not show a stroke-induced decrease in the total number of splenocytes, and multiple observations point toward a differential regulation of CD11b
+, NK, NKT, and T cells in the presence and absence of endogenous IFN-β. First, only IFN-βKO mice showed a significant increase in the percentage of CD11b
+ cells in the spleen 2 days after tMCAo (compared to sham surgery). In line with B cell results, this data possibly reflects a higher proliferative capacity of CD11b
+ cells in IFN-βKO mice. Second, in the IFN-βKO group, stroke did not lead to a transient increase in NK cells (otherwise observed in WT mice) or to significant changes in the percentage of NKT cells. Third, a stroke-induced reduction in the CD3
+CD4
+/CD3
+CD8
+ T cell ratio was absent in IFN-βKO mice. Abrogation of IFN-β also led to a lack of suppression of IFN-γ levels 2 days after tMCAo. The described peripheral changes converge toward an overall deficient immunoregulation in the IFN-βKO group, which could explain the increased brain inflammation. It is worth noting that while sham-operated IFN-βKO mice did have smaller spleens, and that deletion of IFN-β was previously shown to cause structural alterations, along with a decrease in resident macrophages in the spleen [
46], we did not find significant shifts in immune cell populations in the spleen (or blood). Moreover, at least a certain degree of responsiveness of IFN-β splenocytes (particularly CD11b
+ cells) to the post-stroke milieu appears to have been preserved.
In relapsing-remitting EAE, endogenous IFN-β signaling is sufficient to limit inflammation and promote a better disease outcome [
15]. Here, IFN-β KO mice, in contrast to their WT counterparts, showed a significant weight loss, one of the most accepted consequences of neuroinflammatory conditions (such as EAE), and a significant reduction in grip strength 2–3 days after stroke. Using other behavioral approaches, different time points after stroke induction, and the receptor knockout, we could not detect further differences in sensorimotor deficits. Also, we show, not only at the level of the cytokine but also at the level of the receptor and using two models of stroke, tMCAo and pMCAo, that endogenous IFN-β signaling does not influence infarct volume 2 to 8 days after stroke. On the one hand, and albeit having found no difference in body temperature between IFN-βKO mice and WT controls, a potential differential effect of our post-surgical care strategy on infarct volume cannot be ruled out at present. In addition, differences in the extent of sensorimotor deficits or in the functional recovery curve may have been masked by the sensitivity of behavioral tests. Yet on the other hand, these results are consistent with two previous studies showing no effect of IFN-β administration within the first 3–7 days following ischemic stroke [
21], and more recently, subarachnoid hemorrhage [
22]. We hypothesize that this uncoupling between inflammatory changes and, to some extent, outcome is due to the multiple actions of IFN-β. While it may limit the accumulation of immune and inflammatory cells in the brain, it may also, via anti-proliferative and pro-aptoptotic actions contribute to stroke-induced peripheral immunosuppression. In fact, it also is becoming increasingly clear that, within the injured CNS, inflammatory and immune cells may play both pro- and anti-inflammatory roles [
47]. A CNS injury-induced peripheral immunosuppression may serve to diminish a potentially harmful immune response toward the brain, particularly given the exposure of self-antigens [
48], but a certain degree of autoimmunity may be beneficial [
49]. Moreover, peripheral immunosuppression has been associated to an increased susceptibility to infections, which constitutes a common cause of death after cerebral ischemia [
9,
12].
Acknowledgements
This work was supported by grants from: Fundação para a Ciência e a Tecnologia (ARI), Portugal; A.E. Berger, The Crafoord Foundation, G.E. Kock’s Foundation, The Gyllenstiernska Krapperup Foundation, The Royal Physiographic Society in Lund, Swedish National Stroke Foundation, and the Swedish Research Council grant no. 2012–2229 (TD), Sweden; The Danish Independent Research Council, and The Lundbeck Foundation (SI-N), Denmark; Brødrene Hartmann Fond (BHC), Savværksejer Jeppe Juhl og Hustru Ovita Juhls Mindelegat (KLL), and The Lundbeck Foundation (BHC and KLL), Denmark; Agency for Research and Innovation, Council for Health Research (RK supported by a grant to Trevor Owens), Denmark.
We thank Tadeusz Wieloch for the valuable support, particularly with respect to modeling stroke in vivo and subsequent behavioral analysis, Kerstin Beirup for technical assistance, Trevor Owens and Bente Finsen for critical reading of the manuscript, and Hakon Leffler for kindly providing an anti-Gal-3 antibody. We would like to acknowledge Susann Ullén (MultiPark) for important input regarding statistical analyzes.
Competing interests
Competing financial and non-financial interests: the authors declare that no conflict of interest exists.
Authors’ contributions
ARI, YL, KLL, SI-N, and TD designed the experiments, analyzed and interpreted the data. ARI, YL, BC, and TD performed experiments. MS, KK, and TS assisted with immunohistochemistry, imaging, and data analysis. SI-N provided the IFN-β KO mice, and RK provided the IFNAR-KO mice. ARI wrote the manuscript, together with SI-N, TD, and KLL. All authors discussed the results, commented on, and/or edited the manuscript. All authors read and approved the final manuscript.