Background
The proinflammatory cytokines interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) play key roles in the pathogenesis of ischemic stroke [
1‐
3]. IL-1β exerts neurotoxic effects in ischemic stroke and blocking its action has been shown to reduce ischemic brain damage [
4,
5]. In comparison, there is evidence that TNF-α has both neurotoxic [
6,
7] and neuroprotective [
8‐
10] roles after ischemic stroke in rats and in mice. Increasing evidence implicates both cytokines in the early inflammatory response that precedes and accompanies ischemia-induced neuronal damage [
6,
11]. However, detailed knowledge about the contribution of different cell types to the production of IL-1β and TNF-α is still not available.
The relative physiological outcome of increased IL-1β and TNF-α signaling in ischemic stroke may depend on the kinetics and location of cytokine producing cells. There is compelling evidence that IL-1β and TNF-α are primarily synthesized by activated microglia and infiltrating macrophages [
12‐
14], although granulocytes and astrocytes have also been suggested to produce both IL-1β [
15‐
17] and TNF-α [
18,
19]. Precise identification of cell source has, however, been compromised by the lack of microglial and macrophage specific markers, which prevents discrimination of these cell types at the histological level [
12,
14]. Furthermore, it is presently unknown whether IL-1β and TNF-α are expressed to the same extent by the same or different subsets of microglia and macrophages following ischemic stroke.
We have previously shown that IL-1β mRNA and TNF-α mRNA, and TNF-α protein are produced by CD11b
+ microglia and by CD11b
+ macrophages at the edge of and within areas of infarction, and that this production reaches maximum levels of expression between 12 and 24 hours after permanent middle cerebral artery occlusion (pMCAO) in mice [
12,
14,
20]. The objective of the present study was to provide additional insight into the cell types and cell subpopulations that produce IL-1β and TNF-α within the first 24 hours following ischemic stroke in mice [
20]. To distinguish microglia from infiltrating macrophages after pMCAO, we used flow cytometry with CD45 and CD11b as myeloid-lineage specific markers, we used a radiated, bone marrow (BM) chimeric mouse model; and we used intracellular cytokine-staining, and double immunofluorescence staining. In addition, since CD11b is expressed by both macrophages and granulocytes [
21,
22], we also analyzed cytokine production by granulocytes using the granulocyte specific marker Gr1. Our results show that IL-1β and TNF-α are produced by largely segregated subsets of microglia and macrophages, and that very few cells express both cytokines.
Discussion
We have shown for the first time that IL-1β and TNF-α are produced by largely non-overlapping subsets of microglia and macrophages after induction of ischemic stroke in mice. Using flow cytometry and histology, validated by the use of BM-chimeric mice, we showed that microglia and macrophages are the major producers of IL-1β and TNF-α after pMCAO, and that at maximum 1.2% of microglia and 4.5% of infiltrating macrophages co-express IL-1β and TNF-α within the first 24 hours after pMCAO. Granulocytes, which accounted for 50% of the CD11b
+CD45
high cell population and which are known to exacerbate ischemic brain damage [
30], accounted for only 2.4% of IL-1β and 0.8% of TNF-α expression by CD11b
+CD45
high cells after 24 hours of pMCAO, indicating that the contribution of these cells to the production of IL-1β and TNF-α after pMCAO in mice is negligible. A scenario therefore emerges wherein different subsets of microglia and macrophages may have different roles in ischemic stroke, and may thus either improve or reduce the chance of survival of ischemic neurons.
Our observation of a peak in the total number of IL-1β- and TNF-α-expressing cells 24 hours after pMCAO is in line with previous demonstrations of a time-dependent peak in the number of IL-1β mRNA-, TNF-α mRNA-, and TNF-α protein-expressing cells in SJL and C57BL/6 mice, 12–24 hours after pMCAO [
12,
14,
25]. By taking advantage of the ability of flow cytometry to distinguish between CD11b
+CD45
dim microglia and infiltrating CD11b
+CD45
high macrophages, we showed that the number of IL-1β- and TNF-α-expressing microglia by far exceeded the number of cytokine-expressing macrophages 12 hours after pMCAO. This result was expected based on the known steady increase in the number of infiltrating macrophages over the first 24 hours of pMCAO, but this result has not been previously demonstrated.
It was striking to observe that a relatively large proportion of the small number of macrophages present in the cortex from unmanipulated mice express TNF-α, and that the MFI levels of TNF-α expression in these cells are comparable to mean cellular TNF-α expression levels in pMCAO and sham-operated mice. This indicates that TNF-α expression by macrophages is relatively constant no matter whether the cells have infiltrated the cortex prior to or after pMCAO. Interestingly, the proportion of cytokine-expressing macrophages in sham-operated mice was at least as high as that in mice subjected to pMCAO. This likely reflects that sham surgery in itself induces a focal lesion in the cortex [
12,
14]. It is important to note, however, that overall numbers of cytokine-expressing macrophages are not increased versus control mice, since there was no significant recruitment of CD11b
+CD45
high cells to the cortex of sham-operated mice.
Permanent MCAO results in formation of a pan-necrotic infarct, with loss of all cell types including microglial cells. Although neurons and microglia can still be clearly detected after 6 hours of pMCAO [
12,
14], the developing infarct is usually characterised by severe cell loss 12 hours after pMCAO [
12,
14,
20] We were therefore surprised not to observe a reduction in the number of CD11b
+CD45
dim microglia 12 and 24 hours after pMCAO. We wondered whether determination of upregulation of CD45, which is widely used to detect both resting and activated microglia in flow cytometry [
24,
27‐
29,
31] and histology [
32,
33], might lead to increased detection. Indeed, MFI analysis showed that CD45 levels are upregulated in microglia 12–24 hours after pMCAO. However, microglial expression of CD45 was far below that expressed by macrophages/granulocytes, confirming results by others [
34]. Using a bone marrow chimeric approach, we observed that approximately 7% of the CD45
dim microglia were GFP
+ after 24 hours of pMCAO, suggesting that BM-derived microglial precursors could also contribute to the expansion of the microglial population after ischemic stroke in non-chimeric mice. Infiltration of GFP
+ cells was specific to the infarcted cortex, since no increase was observed in the contralateral hemisphere of BM-chimeric mice or in unmanipulated BM-chimeric mice. Other factors contributing to the expansion of the microglial population 12 and 24 hours after pMCAO might be immigration of microglia from regions of the brain not included in the preparation used for flow cytometry, or microglial proliferation. However, microglial proliferation is not prominent at 12 and 24 hours, but is first evident at 48 and 72 hours after induction of focal cerebral ischemia [
34], a finding that is similar to observations in other models of acute neural injury [
32,
36,
37].
The use of a GFP BM-chimeric approach in our study of pMCAO served two purposes: 1) direct visualization of infiltrating GFP
+ cells in tissue sections and 2) validation of the BM origin of the CD11b
+CD45
high macrophages/granulocytes identified by flow cytometry in our model. In addition to the massive infiltration of GFP
+ macrophages/granulocytes 24 hours after pMCAO, we found that a small proportion of microglia in the infarcted cortex could be classified as GFP
+ CD11b
+CD45
dim microglial cells. These observations confirm findings by others [
38‐
42], showing a lesion-induced recruitment of microglial progenitors of bone marrow origin into the infarcted cortex. Recently it has been suggested that these microglial progenitors would not enter into the bloodstream or cross the blood-brain barrier under normal physiological conditions [
43,
44]. However, BM-derived cells have been reported to infiltrate non-irradiated normal brain [
45‐
47], and CD45
high leukocytes are routinely detected in unmanipulated, perfused brains by flow cytometry [
24,
27]. Although focal cerebral ischemia disrupts the blood brain barrier [
33,
39], and irradiation in itself preconditions the brain for cells to infiltrate the neuropil [
43,
44], we in line with earlier findings [
44] observed only sporadic GFP
+ cells in contralateral, non-ischemic cortex and in the brains of unmanipulated BM-chimeric mice. Furthermore, numbers of CD11b
+CD45
high macrophages/granulocytes recruited to ischemic cortex 24 hours after pMCAO were not different between BM-chimeric mice and non-irradiated mice. This suggests that any damage induced by irradiation alone was insufficient to trigger excessive entry of BM-derived cells into the CNS.
Our observation of a significant reduction in microglial population in the cortex of unmanipulated BM-chimeric mice compared to unmanipulated non-chimeric mice indicates that irradiation might impair microglial turnover in normal brain. This is supported by observations by Wirenfeldt et al. [
24], who reported that microglial numbers were reduced by approximately 30% in unmanipulated contralateral hippocampi of perforant pathway-lesioned BM-chimeric mice. That study also reported a lesion-induced impairment of the mitotic capacity of microglia in BM-chimeric mice [
24]. Taken together, we ascribe the reduced microglial numbers in BM-chimeric mice to irradiation damage.
The quantification of BM-derived GFP
+ cells was done by use of an approximated stereological approach, and by use of flow cytometry. The results show that approximately 2/3 of GFP
+ cells are lost during tissue processing procedures prior to flow cytometric analysis. Since GFP
+ cells located in perivascular aggregates were not included in the approximated stereological analysis, cell loss using flow cytometry might be larger than 2/3. However, the consistency in the data shows that flow cytometry is a reliable and robust tool with which to obtain quantitative data on microglia and infiltrating macrophages/granulocytes in brain pathology, as also has been shown in previous studies [
24,
27]. Observations made by others using BM-chimeric mice have shown that BM cells have the capacity to differentiate into microglia and perivascular cells [
24,
39,
48], as well as non-myeloid cell types such as endothelial cells, pericytes, astrocytes and neurons [
38,
42,
47,
49]. However, in our study of mice with 24-hour survival after pMCAO, we observed no evidence of co-expression of GFP with the astroglial marker GFAP, confirming earlier studies [
38,
49]. Similarly, we find no clear evidence of co-expression of GFP and vWF/CD31, although such co-expression has been previously reported at later times of observation (3 days – 1 month) after induction of ischemia [
38,
50].
Acknowledgements
This study received financial support from Hørslev Fonden, Legat til yngre kvindelige forskere på Sundhedsvidenskab, Fonden til Lægevidenskabens Fremme, Familien Hede Nielsens Fond, Kong Christian d. 10 Fond, Frimodt-Heineke Fonden, Direktør Kurt Bønnelycke og hustru Grethe Bønnelyckes Fond, Mogens Svarre Mogensens Fond, Carla Cornelia Storch Møllers legat, Else Poulsens Mindelegat (B. Clausen), Augustinusfonden, Beckett Fonden, Aase og Ejnar Danielsens Fond, Overlægerådets Legatudvalg, Fhv. Dir. Leo Nielsen og Hustru Karen Magrethe Nielsens Legat for Lægevidenskabelig Grundforskning, Fonden til Lægevidenskabens Fremme, Tømrermester Alfred Andersen og Hustru's Fond, Harboefonden and the Danish MRC (K.L. Lambertsen). The technical assistance provided by Lene Jørgensen, Inger-Margrethe Rasmussen, Sussanne Petersen, Inger Nissen and Inger Kathrine Andersen, University of Southern Denmark is greatly appreciated. The assistance with animal care by Aarhus University and the Laboratory of Biomedicine, University of Southern Denmark is also greatly appreciated
Competing interests
The authors of this manuscript declare that there are no actual or potential conflicts of interest. The authors affirm that there are no financial, personal or other relationships with other people or organizations that have inappropriately influenced or biased their research.
Authors' contributions
BHC and KL contributed equally to the experimental and data producing part of this paper. BHC did the data analysis and writing of the manuscript whereas KL assisted in editing the manuscript. AAB was involved in the setup of flow cytometry experiments, interpretation of flow cytometry data and editing of the manuscript, FDH was involved in animal experimentation and TH made the LPS induced macrophage cell cultures. BF contributed to the overall design of experiments and assisted in editing the manuscript.