Microglia are activated following cerebral ischemia and increase their production of the neuro- and immunomodulatory cytokine tumor necrosis factor (TNF). To address the function of TNF from this cellular source in focal cerebral ischemia we used TNF conditional knock out mice (LysMcreTNFfl/fl) in which the TNF gene was deleted in cells of the myeloid lineage, including microglia. The deletion reduced secreted TNF levels in lipopolysaccharide-stimulated cultured primary microglia by ~93%. Furthermore, phosphorylated-ERK/ERK ratios were significantly decreased in naïve LysMcreTNFfl/fl mice demonstrating altered ERK signal transduction. Micro-PET using 18[F]-fluorodeoxyglucose immediately after focal cerebral ischemia showed increased glucose uptake in LysMcreTNFfl/fl mice, representing significant metabolic changes, that translated into increased infarct volumes at 24 hours and 5 days compared to littermates (TNFfl/fl). In naïve LysMcreTNFfl/fl mice cytokine levels were low and comparable to littermates. At 6 hours, TNF producing microglia were reduced by 56% in the ischemic cortex in LysMcreTNFfl/fl mice compared to littermate mice, whereas no TNF+ leukocytes were detected. At 24 hours, pro-inflammatory cytokine (TNF, IL-1β, IL-6, IL-5 and CXCL1) levels were significantly lower in LysMcreTNFfl/fl mice, despite comparable infiltrating leukocyte populations. Our results identify microglial TNF as beneficial and neuroprotective in the acute phase and as a modulator of neuroinflammation at later time points after experimental ischemia, which may contribute to regenerative recovery.
Microglial activation constitutes a major defense mechanism against ischemic brain injury, with dual effects linking microglial activation to both neurotoxic and neuroprotective effects in the brain1,2. When activated, microglial cells become the main producers of proinflammatory cytokines, such as tumor necrosis factor (TNF), which is crucial in detemining neuronal survival and orchestrating neuroinflammation following stroke2,3,4,5. TNF exists as both transmembrane TNF (mTNF) and soluble TNF (solTNF), which is released into the extracellular space after cleavage of mTNF by the metalloproteinase TNFα converting enzyme (TACE/ADAM17)5. Both forms are biologically active but have distinct functions in the CNS5,6,7,8,9.
The effect of TNF is mediated by two cell surface receptors, TNF-R1 and TNF-R2. These receptors may feed into diverse signaling pathways according to differences in their intracellular domains. A death domain in TNF-R1, not present in TNF-R2, can lead to apoptotic cell death, although most cell types are protected due to expression of anti-apoptotic proteins. Alternative pathways of TNF-R1 signaling involve activation of the ERK1/2 MAP kinases and NF-κB activation10,11 serving neuroprotective functions. Similarly, TNFR2 receptor signaling through the NF-κB pathway is primarily associated with neuroprotection and axonal preservation6,9,12,13.
In experimental stroke, genetic ablation of solTNF, with sustained mTNF expression, is associated with reduced infarct volume, improved functional outcome and an altered pro-inflammatory environment5, just as the systemically administered anti-solTNF compound XPro1595 improves functional outcome after experimental stroke3. On the other hand, conventional TNF knock out (TNF−/−) mice display increased infarct volumes and worse functional outcome after experimental stroke4. The neuroprotective effect of TNF is suggested to be mediated through TNF-R1, findings which are in line with studies showing increased infarct volumes in TNF-R1 deficient mice4,14,15 and that ischemic tolerance is mediated through upregulation of TNF-R116.
Given the contribution of microglia/macrophages to the pathophysiology of stroke, in the present study, we investigated the effect of TNF ablation specifically in myeloid cells on neuroprotection and functional outcome after focal cerebral ischemia. We demonstrate that LysMcreTNFfl/fl mice lacking TNF in myeloid cells display phenotypical alterations in microglia, including altered ERK signaling. In addition, FDG uptake in the ischemic brain during the acute phase after ischemia (1–3 hours) is increased in LysMcreTNFfl/fl mice resulting in increased infarct volumes 24 hours and 5 days after focal cerebral ischemia. The relevance of the present study was underlined by investigating glial activation, TNF and TNF receptor expression in human autopsy material from stroke patients. This may have significant implications for the development of future neuroprotective treatment strategies in the acute phase after stroke, as modulating the microglial TNF response acutely can provide a viable option to boost neuroprotection.
Characterization of LysMcreTNFfl/fl mice with conditional ablation of TNF in myeloid cells
In order to assess the effect of genetic ablation of TNF in myeloid cells, we first characterized LysMcreTNFfl/fl mice under naïve conditions. LysMcreTNFfl/fl mice were comparable to TNFfl/fl mice displaying normal cerebrovascular anatomy at the level of the circle of Willis (Fig. 1a), similar cortical volumes (Fig. 1b), as well as body composition (Suppl. Table 1). Furthermore, conditional TNF ablation resulted in no abnormal locomotor function, as previously observed in conventional TNF−/− mice5, when subjected to the rotarod and open field test (Suppl. Table 1). We found LysMcreTNFfl/fl mice to have normal neuromuscular function, balance, motor coordination and spontaneous activity (Suppl. Table 1), just as anxiety-related behaviors (grooming, fecal droppings, latency to rear and center/peri-meter ratio) were not altered compared to TNFfl/fl mice (Suppl. Table 1). When testing for spontanoeus alternation behavior (SAB) and working memory, we found no difference in SAB % but significantly fewer Y-maze entries in LysMcreTNFfl/fl mice compared to controls (Suppl. Table 1).
Since conventional TNF−/− mice are known to display reduced numbers of microglia4, we next quantified microglia by flow cytometry using CD45 levels to distinguish CD45dim microglia from infiltrating CD45high leukocytes (Fig. 1c–e). We found comparable microglial numbers (Fig. 1d) and percentages (Fig. 1e) between TNFfl/fl and LysMcreTNFfl/fl mice. However, mean fluorescence intensity (MFI) revealed higher CD11b (Suppl. Fig. 1a) and lower CD45 (Suppl. Fig. 1b) expression levels on microglia in naïve LysMcreTNFfl/fl compared to TNFfl/fl mice. Analysis of whole blood showed comparable numbers of monocytes and T cells, but fewer Gr1+ granulocytes in LysMcreTNFfl/fl compared to TNFfl/fl mice (Suppl. Fig. 1c,d). In the spleen, monocytes, granulocytes and T cells were comparable between the two genotypes (Suppl. Fig. 1c,d).
We addressed recombination efficiency in microglial cells by crossing MLys1cre mice with Rosa26-tdTomato reporter mice and quantified the number of cortical tdTomato+ cells in naïve conditions by cell counting and flow cytometry (Suppl. Fig. 2). Under naïve conditions, we found 12% (cell counting) to 20% (flow cytometry) of all CD11b+ microglial cells to be tdTomato+. No tdTomato+ cells co-labelled with GFAP or Olig2, but a small proportion of neurons were tdTomato+ (5%). Since microglia increase LysM expression under ischemic and inflammatory conditions17,18,19, we next tested the cre-mediated reporter gene activation under inflammatory conditions using cultured microglia harvested from TNFfl/fl and LysMcreTNFfl/fl mice (Fig. 1f,g). Following LPS-stimulation, we found that microglia from LysMcreTNFfl/fl mice showed a ∼93% decrease in TNF production as compared to microglia from TNFfl/fl mice (Fig. 1g). Astrocytes stimulated with either IL-1β or TNF secreted comparable amounts of TNF, with no difference between TNFfl/fl and LysMcreTNFfl/fl mice (Fig. 1h), collectively demonstrating a successful TNF ablation in activated microglia from LysMcreTNFfl/fl mice.
Altogether, these data demonstrate a successful TNF ablation in activated microglia and support the use of LysMcreTNFfl/fl mice as a suitable model to study microglial/myeloid TNF function in vivo.
Conditional ablation of myeloid TNF increases infarct volume after focal cerebral ischemia
The effect of myeloid TNF ablation in cerebral ischemia was assessed by comparing infarct volumes in TNFfl/fl to LysMcreTNFfl/fl mice after pMCAO (Fig. 2a,b). In LysMcreTNFfl/fl mice infarct volumes were significantly larger than in TNFfl/fl mice at 24 hours (30%) (Fig. 2a) and at 5 days (45%) (Fig. 2b) after pMCAO. Histological analysis showed that infarcts were evenly distributed along the rostrocaudal axis in both genotypes (Fig. 2a,b).
To assess functional outcome, we subjected the mice to a battery of behavioral tests, but found no locomotor or sensory-motor differences between TNFfl/fl and LysMcreTNFfl/fl mice 2 and 5 days after pMCAO (Suppl. Table 2). Rung walk analysis of both front and hinds paws (total slips) on day 2 after pMCAO showed clear asymmetry between the left (L) and right (R) side (Fig. 2c), as did grip strength of the front paws 5 days after pMCAO (Fig. 2d) in both TNFfl/fl and LysMcreTNFfl/fl mice. Both groups of mice performed similarly on the rotarod 5 days after pMCAO (Fig. 2e). In the open field test 2 days after pMCAO, we found no difference between TNFfl/fl and LysMcreTNFfl/fl mice in total distance traveled, speed, zone changes and number of rearings (Suppl. Table 2). Also no change in anxiety-related behaviors, such as center/perimeter ratio, latency to rear and number of droppings were found, except for grooming behavior, which was significantly increased in TNFfl/fl mice after pMCAO (Suppl. Table 2). The lack of differences in functional outcome most likely reflects the large cortical infarcts in both genotypes.
FDG uptake is increased in the acute phase after pMCAO in LysMcreTNFfl/fl compared to TNFfl/fl mice
To assess ischemia-dependent changes in glucose metabolism immediately after pMCAO (1–3 hours), we performed 18[F]-FDG-PET imaging (Fig. 3a). Dynamic scans revealed an early (<10 min) reduction in FDG uptake in the ipsilateral cortex compared to the contralateral in both genotypes, most likely as a result of ceased blood flow to the lesion site (Fig. 3b). Ten to 15 min after injection, FDG uptake switched and was increased in the ipsilateral side compared to the contralateral side in both TNFfl/fl and LysMcreTNFfl/fl mice. However, beginning approximately 35 min post-injection, ipsilateral FDG uptake was significantly higher in the ipsilateral hemisphere in LysMcreTNFfl/fl compared to TNFfl/fl controls (Fig. 3b), demonstrating an increased glucose demand in the ipsilateral hemisphere after pMCAO. This is consistent with the presence of larger infarcts in LysMcreTNFfl/fl mice where cells increased glucose uptake after stroke onset.
VOI-based analysis of 18[F]-FDG-PET imaging comparing the normalized rate constants (Ki, influx rate; K1, rate blood to tissue; k2, rate tissue to blood; k3, binding rate) revealed significant differences between TNFfl/fl and LysMcreTNFfl/fl mice (Table 1). The increase in Ki corresponded to a 37% elevation in glucose uptake in LysMcreTNFfl/fl compared to TNFfl/fl mice. A second scan 24 hours later revealed a lower ipsilateral FDG uptake in both groups (Fig. 3c) due to loss of cells in the ischemic region. However, there was no significant difference in the uptake pattern 24 hours after pMCAO, as demonstrated by previous findings of a fully mature infarct at this time point20.
Ablation of myeloid TNF alters MAPK protein expression in naïve conditions but not after pMCAO
Since TNF-dependent MAPK signaling has been associated with neuroprotection under ischemic conditions21, we evaluated whether the increase of infarct volumes in LysMcreTNFfl/fl mice could be associated with alterations in MAPK signaling. Interestingly, we found pERK1/ERK1 and pERK2/ERK2 ratios to be significantly decreased in LysMcreTNFfl/fl mice compared to TNFfl/fl mice under naïve conditions (60% and 75%, respectively) (Fig. 4a), suggesting that ERK signaling is intrinsically downregulated in LysMcreTNFfl/fl mice, which could impair ERK-dependent neuroprotective signaling acutely (within the first 6 hours) after pMCAO. Despite tendencies, pERK1/ERK1 and pERK2/Erk2 ratios were not different between TNFfl/fl and LysMcreTNFfl/fl mice 6 hours after pMCAO (Fig. 4a). No differences were found in p-p54/p54, p-p46/p46 (SAPK/JNK) (Fig. 4b), or p-p38/p38 (Fig. 4c) ratios in naïve and ischemic conditions.
Ablation of myeloid TNF reduces the ischemic-induced expression of pro-inflammatory cytokines and chemokines 24 hours after pMCAO
To assess changes in the inflammatory response between TNFfl/fl and LysMcreTNFfl/fl mice, we assessed cytokine, chemokine, and TNFR protein expression in naïve conditions and after pMCAO by multiplex analysis (Fig. 5a–c and Suppl. Table 3). No differences in protein expression were found in naïve conditions except for TNFR1, which was significantly increased in LysMcreTNFfl/fl compared to TNFfl/fl mice (Fig. 5b). Similarly, 6 hours and 5 days after pMCAO no differences in protein expression were detected (Suppl. Table 3). At 24 hours after pMCAO, however, TNF, IL-1β, IL-6, IL-5 and CXCL1 were significantly reduced in LysMcreTNFfl/fl mice compared to TNFfl/fl controls, while IL-2, IL-4, IL-10, IFNγ or IL-12p70 did not change (Fig. 5a). Cytokine levels were also measured in the plasma in naïve and ischemic conditions but no changes were found at any time point between TNFfl/fl and LysMcreTNFfl/fl mice (Suppl. Fig. 3).
Ablation of myeloid TNF does not result in changes in peripheral myeloid cell numbers or infiltrating immune cells after pMCAO
Based on our previous findings of increased leukocyte infiltration in conventional TNF−/− mice4 and decreased macrophage infiltration in mice only expressing the transmembrane form of TNF (mTNFΔ/Δ)5 one day after pMCAO as well as altered microglial activation following anti-TNF treatment3, we investigated possible changes in microglia and infiltrating leukocytes in the cortex of TNFfl/fl and LysMcreTNFfl/fl mice 6 and 24 hours after pMCAO by the use of flow cytometry (Fig. 6a–d and Suppl. Fig. 4a,b). At 6 hours, we found the total number of CD11b+CD45lowLy6C+Ly6G− microglia to be significantly increased in LysMcreTNFfl/fl mice compared to TNFfl/fl mice (Fig. 6b), most likely a result of increased CD11b expression in LysMcreTNFfl/fl mice under naïve conditions as measured by MFI (Suppl. Fig. 1a). As expected, at 24 hours after pMCAO, the time of maximal cytokine production, in TNFfl/fl and LysMcreTNFfl/fl mice the percentage and number of CD11b+CD45dim microglia were decreased in the ipsilateral cortex as compared to the contralateral (Suppl. Fig. 4a,b). However, no differences in percentage and cell number were observed between the two genotypes (Fig. 6b and Suppl. Fig. 4a,b). At 6 hours after pMCAO, the numbers of infiltrating CD11b+CD45highLy6ChighLy6G- macrophages (Fig. 6c) and CD11b+CD45highLy6ChighLy6G+ granulocytes (Fig. 6d) were low and comparable between the two genotypes. By 24 hours, the percentage and number of infiltrating macrophages and granulocytes were significantly increased in the ipsilateral cortex of LysMcreTNFfl/fl and TNFfl/fl mice compared to the contralateral cortex, however with no difference between the two groups (Fig. 6c,d and Suppl. Fig. 4b). The number and percentage of infiltrating CD45highCD3+ T cells were minimal and did not change in the ipsilateral cortex as compared to the contralateral in either TNFfl/fl or LysMcreTNFfl/fl mice, nor between the two groups (Suppl. Fig. 4b).
At 6 hours, we found that the total number of TNF+ microglia was significantly decreased in LysMcreTNFfl/fl mice as compared to TNFfl/fl mice (Fig. 6e) and demonstrated an approximately 56% ablation of TNF in microglia. At 6 hours, none of the few infiltrating macrophages were found to express TNF (Fig. 6f). At 24 hours, the number of TNF+ macrophages was significantly decreased in LysMcreTNFfl/fl compared to TNFfl/fl mice (Fig. 6f) where only 0.5% of all macrophages in LysMcreTNFfl/fl mice were TNF+ compared to 3% in TNFfl/fl mice. At no time point did granulocytes express TNF. This suggests that leukocyte recruitment into the infarct area is not affected by TNF ablation in myeloid cells 6 and 24 hours after pMCAO.
By 24 hours after pMCAO the percentage of all leukocyte populations in the blood and spleen were comparable between LysMcreTNFfl/fl and TNFfl/fl mice (Suppl. Fig. 4d,e). In the bone marrow, CD11b+CD45+Gr1− and CD45+CD3+ populations were significantly increased in LysMcreTNFfl/fl mice compared to TNFfl/fl mice, with a higher number and percentage of all CD45+ cells in LysMcreTNFfl/fl mice compared to controls (Suppl. Fig. 4e).
Localization of TNF, TNF-R1 and TNF-R2 in human stroke brains
To address TNF signaling in the context of human stroke pathophysiology, we analyzed TNF, TNF-R1, and TNF-R2 expression in the post-mortem brain of two stroke cases (Suppl. Table 4 and Fig. 7a–o, shown for 1 post-mortem brain only). The infarcted area, identified by HE staining (Fig. 7a,b), was characterized by extensive tissue disruption and presence of cell infiltrates immunopositive for CD45 and CD68 (Fig. 7d,e,g,h). Iba1+ cells, mostly with macrophage morphology, were also found in high numbers within the infarct (Fig. 7j,k). GFAP+ astrocytes formed a demarcated glial scar and were found to be almost absent within the infarct (Fig. 7m,n). TNF+ cells were located primarily within the infarct and at the border between infarct and intact tissue (Fig. 7c,f). On the contrary, TNF-R1+ (Fig. 7i) and TNF-R2+ (Fig. 7l,o) cells were found in the border zone only, with TNFR2+ cells accumulating in a dense layer surrounding the infarct. TNFR1+ cells were located at the border zone and further away in the peri-infarct tissue. We also assessed TNF expression in mouse tissue 24 hours after pMCAO and found high numbers of TNF+CD11b+ cells within the infarct and peri-infarct area (Fig. 7p–r), confirming previous findings in mice4. No TNF+ astrocytes were detected (not shown).
Using LysMcreTNFfl/fl conditional KO mice with specific ablation of TNF in myeloid cells, including microglia, we demonstrate that microglial-derived TNF contributes to TNF-mediated protective effects in experimental stroke. We and others previously demonstrated that in the CNS acutely after focal cerebral ischemia TNF is produced by microglia, not neurons or astrocytes3,4,22,23. Using chimeric mice, we also showed that microglial-derived TNF is neuroprotective whereas macrophage-derived TNF plays no role in acute infarct maturation4. Therefore we can conclude that worsening of the stroke pathology in LysMcreTNFfl/fl mice is dependent on the lack of microglial-derived TNF.
Our findings on expression of TNF and its receptors in brain specimens from stroke patients are in line with previous reports (reviewed in2) and underscore this study’s translational relevance of TNF signaling in human stroke pathology, not only in experimental models.
We and others have shown that TNF−/− mice exhibit a behavioral phenotype associated with increased locomotor activity and altered anxiety-related behavior, in addition to impaired spatial learning and memory24,25, while mTNFΔ/Δ mice, expressing only mTNF, display a normal behavioral phenotype5. In this study we show that conditional ablation of myeloid TNF alone did not impair locomotor or neuromuscular function. This may depend on the relatively limited TNF ablation in microglia from LysMcreTNFfl/fl mice in naïve conditions, which may not be sufficient to induce a measurable phenotype, suggesting that constitutively expressed TNF in the brain of naïve LysMcreTNFfl/fl mice is sufficient for maintenance of cognitive functions under physiological conditions. Alternatively, it may be that other sources of TNF in naïve conditions are supporting normal cognitive and locomotor function.
Eventhough in LysMcreTNFfl/fl mutant mice the deletion efficiency is around 85–98% in mature macrophages and near 100% in granulocytes26, we observed that in naïve CD11b+ microglia deletion efficiency was only around 20% and that recombination also occured in approximately 5% of neurons. Neuronal expression was very low, and we consistently never observed increased TNF expression in neurons in this model of experimental stroke22,23,27,28. Furthermore, microglia increase their expression of LysM in ischemic and inflammatory conditions, leading to more efficient gene ablation in LysMcre expressing mice17,18,19. Indeed, we first verified ablation efficiency in LPS-stimulated microglia cultures from LysMcreTNFfl/fl and TNFfl/fl mice by showing a marked reduction in TNF release (93%) and then by flow cytometry 6 hours after pMCAO demonstating a 56% reduction in TNF+ microglia in the ischemic brain. On the contrary, release of TNF was not affected in vitro in astrocytes, further validating efficient and selective ablation of TNF from microglial cells.
When we tested the effect of myeloid TNF deletion in experimental stroke, we found LysMcreTNFfl/fl mice to have significantly increased FDG uptake acutely (>3 hours) and larger infarct volumes at 24 hours and 5 days after pMCAO, supporting our previous findings that constitutively expressed or early induced, microglial-derived TNF is neuroprotective in experimental stroke4. When evaluating the spatiotemporal dynamics of FDG uptake during acute pMCAO in TNFfl/fl and LysMcreTNFfl/fl mice, we found early microglial-derived TNF to affect FDG uptake in the ischemic tissue after pMCAO. Though assumptions of a steady state system is hampered in the acute phase after pMCAO and care should be taken not to use the kinetic model data as absolute true numbers, the findings of hyper-FDG in the ispsilateral cortex compared with the contralateral cortex within the first 2 hours after pMCAO in both strains of mice are consistent with previous findings of increased accumulation of glucose around the ischemic border29,30. This may reflect increased anaerobic glycolysis during acute cerebral ischemia to compensate the loss of ATP29,30,31 or alteratively that lack of microglial-derived TNF alters ongoing signaling cascades in the ischemic brain10,11. The finding of increased FDG uptake in the ipsilateral hemisphere of LysMcreTNFfl/fl mice compared to TNFfl/fl mice during acute ischemia supports the findings of increased infarct volumes 24 hours and 5 days after pMCAO and demonstrate that conditional ablation of TNF in microglia modifies neuronal sensitivity to ischemia. The fact that we in the present study were not able to differentiate between FDG uptake in infarct and peri-infarct areas is due to the limited resolution (1 mm) of the PET scanner as imaging spatial resolution is essential to delineate these areas30.
In some studies JNK/SAPK and p38 MAPK were shown to promote cell death, whereas ERK1/2 opposed cell death32,33. In the present study we found SAPK/JNK and p38 ratios not to be changed both under naïve conditions and 6 hours after pMCAO, whereas ERK1/2 ratios were significantly decreased in naïve LysMcreTNFfl/fl mice, with a tendency still present 6 hours after pMCAO. These findings suggest that ERK signal transduction is altered in LysMcreTNFfl/fl mice. TNF is known to activate all three MAPKs in neurons, with ERK1/2 activation mediated via MEK1/234 and ERK1/2 activation is known to have a dominant-negative effect on the apoptotic signaling of death receptors, including TNF-R133. However, ERK1/2 activation has been shown to primarily mediate cell protection through TNF-R1-mediated NF-κB activation33. We and others have previously shown that the neuroprotective effect of TNF is mediated at least in part via TNF-R1 (reviewed in2). Even though NF-κB activation and translocation are comparable in TNF−/− and WT mice after pMCAO4, it is possible that ablation of microglial TNF leads to a decrease in ERK1/2 ratios and altered NF-κB activation, which ultimately results in lack of NF-κB-induced neuroprotection in LysMcreTNFfl/fl mice. In support of this, Rousseau et al.35 showed that in vitro LPS stimulation of macrophages results in sequential activation of the TPL2-MKK1/MKK2-ERK1/2 pathway followed by secretion of TNF and that inhibition of this pathway resulted in loss of TNF production. Alternatively, we speculate that conditional ablation of TNF in microglia may have affected microglial expression of P2X7 receptors, as previously shown36. ATP, released from damaged cells as a consequence of ischemia, has been shown to act on P2X7 receptors on microglia and thereby activate MAPK (ERK1/2, JNK and p38), which results in the release of neuroprotective TNF37. Thus the balance between ERK and the JNK and p38 MAPK pathways may play an important role in neuronal cell survival or death in response to focal cerebral ischemia.
As a consequence of ischemia, microglial numbers decrease in the ipsilateral hemisphere after experimental stroke whereas leukocyte infiltration increases at the same time4,38. Stroke is known to stimulate bone marrow production of myeloid cells, including monocytes and granulocytes, which are subsequently recruited to the brain38,39. Also T cell numbers in the bone marrow have been shown to increase after experimetal stroke39. We therefore investigated changes in myeloid cell populations in the bone marrow from LysMcreTNFfl/fl and TNFfl/fl mice after pMCAO. We found that numbers and percentages of CD11b+CD45+Gr1− monocytes and CD45+CD3+ T cells were significantly increased in LysMcreTNFfl/fl mice 24 hours after pMCAO suggesting that ablation of mesodermally-derived TNF results in an increase in the number of newly produced, bone marrow-derived monocytes and T cells. The signals that trigger leukocyte mobilization from the bone marrow after stroke are unknown and may represent potential targets for future therapies40,41. The bone marrow indirectly receives central autonomic innervation involving forebrain areas such as the insular and piriform cortex42, areas which often are affected by ischemia. This may account for the increased mobilization of monocytes and T cells in LysMcreTNFfl/fl mice, which also display significantly increased infarcts. We and others have shown that following cerebral ischemia cellularity decreases in the spleen3,43,44, which can be attributed to increased apoptotic cell death of lymphocyte populations43. Therefore, it is possible that the increase in CD3+ T cells in LysMcreTNFfl/fl mice compared to TNFfl/fl mice can be due to decreased ischemia-induced apoptosis of spleen lymphocytes. On the other hand, early splenic responses after stroke involve increased production of pro-inflammatory mediators45. It is therefore possible that the pro-inflammatory enviroment is altered in such a way that less T cells are deployed into the blood.
We have previously shown that the number of infiltrating leukocytes is significantly increased in TNF−/− compared to WT mice 24 hours after stroke4. Since CD11b expression levels, known to be involved in inflammation by regulating leukocyte adhesion and migration46, were increased in naïve microglia, we investigated whether myeloid-derived TNF affected leukocyte infiltration into the ischemic brain. We found that infiltration of macrophages, granulocytes and T cells was not affected in LysMcreTNFfl/fl mice at this time point, nor at 6 hours after pMCAO. Anti-TNF therapies decrease the number of granulocytes infiltrating the ischemic infarct 24 hours after pMCAO3, an effect suggested to be mediated via the acute phase response in the liver47. Since TNF is not deleted from liver cells in LysMcreTNFfl/fl mice26, it is possible that signals mediating granulocyte influx into the brain are not altered in these mice. Suprisingly, macrophage infiltration was not affected in LysMcreTNFfl/fl mice, given that we previously demonstrated reduced infiltration of macrophages into the brain of mTNFΔ/Δ mice 24 hours after pMCAO5. As monocytes/macrophages have been shown to infiltrate the ischemic brain relatively late after pMCAO48, it is possible that the time point investigated in the present study is not optimal for addressing the influx of macrophages. Nevertheless, we have previously shown, that only microglial-derived TNF, and not macrophage-derived TNF, plays a role in the neuroprotective effects of TNF after pMCAO4, which was also supported by this study where we found no TNF+ leukocytes 6 hours after pMCAO.
All together, the present study supports microglial-derived TNF as beneficial and neuroprotective in the acute phase after experimental stroke and as a strong modulator of the neuroinflammatory response at later time points after experimental ischemia.
Materials and Methods
MLys1cre/x TNFflox/flox (LysMcreTNFfl/fl) breeding couples, with a specific deletion of TNF in myeloid cells26, were transferred from the Russian Academy of Sciences, Moscow to the Biomedical Laboratory at the University of Southern Denmark where they were established as a colony. The extent of TNF gene deletion in macrophages and neutrophils is almost complete (<98%), with no deletion in liver and thymus26. Littermate TNFflox/flox (TNFfl/fl) mice, with normal TNF epxression26, were used as controls. In addition, MLys1cre/mice49 were purchased from the Jackson Laboratory and transferred to the Miami Project to Cure Paralysis where they were crossed with Rosa26-loxp-Stop-loxp-tdTomato reporter mice (established colony)50 to generate LysMtdTomato mice51. All mice used in the present study were adult (7–10 weeks) males. All experimental protocols were approved by The Danish Veterinary and Food Administration (J. No. 2011/561–1950 and 2013-15-2934-00924) and according to the University of Miami IACUC and NIH guidelines. All efforts were made to minimize pain and distress and the methods were carried out in accordance with the relevant guidelines.
TNFfl/fl mice were genotyped by a two-oligonucleotide primer PCR assay and Cre-mediated recombination and completion of deletion were established using a three-oligonucleotide primer PCR assay26. Primer sequences were: KO41 (5′-TGAGTCTGTCTTAACTAACC) and KO42 (5′-CCCTTCATTCTCAAGGCACA) for TNF and a combination of primers Cre8 (CCCAGAAATGCCAGATTACG), Mlys1 (CTTGGGCTGCCAGAATTTCTC) and Mlys2 (TTACAGTCGGCCAGGCTGAC) for Cre.
The circle of Willis and its branches were examined in LysMcreTNFfl/fl and TNFfl/fl mice essentially as previously described4 using a Toluidine blue (TB) solution. Vessels were photographed under a dissecting microscope and the left an right posterior communicating (PCOM) arteries scored as follows: 0, absent; 1, present, but poorly developed (hypoplastic); 2, well formed. A single PCOM score was obtained by averaging left and right scores.
Body Composition Measurements
Total tissue mass (g), fat mass (g), fat-%, lean tissue mass (g), lean-%, bone area (cm2), bone mineral content (BMC, g/cm) and bone mineral density (BMD, g/cm2) were measured in naïve LysMcreTNFfl/fl and TNFfl/fl using dual-energy X-ray absorptiometry (DXA) using a PIXImus2 (Version 1.44; Lunar Corporation, Madision, WI, USA) as previously described5,52.
Induction of focal cerebral ischemia
The distal part of the left middle cerebral artery was permanently occluded (pMCAO) under Hypnorm/Stesolid anesthesia [Fentanyl Citrate (0.315 mg/mL; Jansen-Cilag) and Fluanisone (10 mg/mL; Jansen-Cilag); and Diazepamum (5 mg/mL; Dumex)]22,28. After surgery, mice were supplied with 0.9% saline and maintained in a 28 °C controlled environment. Mice with 5 days survival were housed in the behavioral room in the conventional animal facility after 24 hours. For post-surgical analgesia, mice were treated with Temgesic (0.001 mg/20 g body weight buprenorphium; Reckitt and Colman) three times at 8 hours intervals starting prior to surgery.
Positron emission tomography (PET) analysis
Micro PET imaging was performed on a Siemens Inveon PET scanner. The animals were scanned twice. The first scan was performed in the acute phase, 1–2 hours after surgery, and the second scan 24 hours later. Animals were placed in prone position on a heated scanner bed and injected with 18[F]-FDG via a tail vein catheter. The animals were anesthetized throughout the entire imaging procedure using a mixture of 1.5–2% isoflurane and 100% oxygen.
At day 0 (1–2 hours after pMCAo) 10.9 ± 2.4 MBq 18[F]-FDG was injected and data was acquired for 90 min. Data was binned in 20 frames (4 × 30 s + 10 × 1 min + 1 × 3 min + 5 × 15 min), beginning with short frames to capture the fast vascular phase moving to a slower frame rate appropriate for the dynamics beyond 15 min. At day 1 (24 hours after pMCAo) 10.7 ± 1.2 MBq 18[F]-FDG was injected. After a 30 min delay a 15 min static image was acquired.
All images were reconstructed using OSEM3D/MAP matrix 128 × 128, 4 OSEM3D iterations and 18 MAP iterations with requested resolution of 0.8 mm. A zoom factor of 2 was applied yielding a voxel size of approximately 0.4 mm × 0.4 mm in the XY-plane and a slice thickness of 0.8 mm. The reconstruction was carried out with the build in algorithm of Siemens Inveon Acquisition Workplace software version 1.5.
A computer tomography (CT) image of a single animal was acquired and used as an anatomical reference. All images were analyzed and the kinetic modeling done within the Siemens Inveon Research Workplace software version 4.2. The reference CT and the PET images were manually co-registered. The images acquired at day 0 were analyzed using a 2-tissue-compartment irreversible FDG model. Image derived blood input functions53 were used to calculate the 3 rate constants K1 (transport rate constant from blood to brain tissue), k2 (rate constant from brain tissue back to blood), k3 (binding in the cells (phosphorylation))31. The FDG uptake (influx) rate constant Ki was calculated as Ki = K1k3/(k2 + k3). The glucose uptake (metabolism) is proportional to Ki, glucose uptake = Ki (blood glucose/LC), where LC is the lumped constant. The FDG uptake is a very sensitive measure and is influenced by many factors including dietary condition (insulin level), ambient temperature and mode of anesthesia54. In order to overcome some of these limitations, we used the animals as their own reference and calculated the rate constants from the ipsilateral volume of interest (VOI) normalized to contralateral reference rate constants. Curves are presented as standardized uptake values (SUV = tissue activity concentration (Bq/mL) x Animal weight (g)/injected activity (Bq)).
Volumes of interest were semi-automatically made by first drawing a volume covering the increased FDG uptake in the ipsilateral region. The final VOI was calculated by setting a threshold of 70% of maximum within the last image of the dynamic series. This VOI was mirrored onto the contralateral brain hemisphere as a reference VOI for the unaffected brain tissue. The blood input VOI was manually created in the first or second frame where the vena cava was clearly visualized. The blood input VOI was manually drawn as the part of vena cava with highest activity.
Open field test
Locomotor activity was investigated using the open field test in a group of naïve LysMcreTNFfl/fl and TNFfl/fl mice as previously described52 in order to detect potential phenotypic differences in locomotor- and anxiety-related activities. In addition, locomotor activity was investigated 2 days after pMCAO in order to test for changes in behavior induced by the ischemic brain lesion. The total distance traveled (cm), speed (V(mean), cm/sec), center/perimeter ratio, and entries into the three zones (wall, inter peri and center of the box) were recorded automatically for 10 min. Rearing, time to first rear, grooming and droppings were recorded manually and are given as number of events (n).
Y-maze spontaneous alternation test
In order to test spatial leaning and memory in naïve LysMcreTNFfl/fl and TNFfl/fl the spontaneous alternation test was performed using a Y-maze as previously described52. Each mouse was placed in the arm designated (A) of the Y-maze and allowed to freely explore the maze for 8 min. The number of entries, except for the first two, and the number of triads were recorded and the percentage of alternation was calculated.
In order to evaluate motor coordination/performance and balance52, the rotarod test (LE8200, Panlab Harvard Apparatus) was performed. The test comprised a pre-training part prior to surgery (30 sec at 4 rotations per min (rpm)) and a trial part consisting of 4 trials (T1 – T4, with at least 15 min in-between) 5 days after surgery. Mice were placed on the rotarod set in accelerating mode. Speed of the rotor was accelerated from 4 to 40 rpm over 5 min. Total time spent on the rotarod for each mouse was measured.
Rung walk test
In order to test stepping, inter-limb coordination and balance, the rung walk test was performed 2 days after pMCAO as previously described6,55. The total number of mistakes on each front and hind paw was plotted for analysis of asymmetry. Prior to surgery, mice were pre-trained in the rung walk test and no asymmetry was observed under baseline conditions (data not shown).
Grip strength test
The grip strength measuring device (BIO-GT-3, BIOSEB) was used to study neuromuscular function in mice subjected to pMCAO. The peak amount of force was recorded in 5 sequential trials and the highest grip value was recorded as the score4. We analyzed the grip strength in individual (left and right) front paws prior to (baseline) and 3 and 5 days after pMCAO. The unit of force measured is presented as grams (g). Asymmetry between paws in individual mice following pMCAO was calculated and is presented as delta (Δ) grip strength5.
Fresh frozen tissue
Mice were killed by cervical dislocation, and brains quickly removed, frozen in CO2 and cut coronally in six parallel series of 30 μm and stored at −80 °C until further processing5. Blood samples were collected in EDTA-coated Eppendorf tubes, spun 2 × 10 min at 3,000 g, 4 °C, and plasma stored at −80 °C until further processing.
Paraformaldehyde (PFA) fixed tissue
LysMtdTomato mice were anesthetized with Ketamine/Xylazine (100 mg/10 mg/kg) and transcardially perfused with 4% PFA. For cell counting, tissue was processed as previously described51, and brains were cut into 20 μm-thick, sagittal cryostat sections and stored at −20 °C until further processing. For cortical volume estimations, LysMcreTNFfl/fl and TNFfl/fl mice were overdosed using Pentobarbital (200 mg/ml)/Lidocainhydrochlorid (20 mg/ml) and transcardially perfused with 4% PFA. Brains were fixed in 4% PFA overnight, changed to 1% PFA and finally 0.1% PFA before they were horizontally vibratome-cut into six 60 μm-thick parallel series of free floating sections and cryoprotected in de Olmo’s solution5. For immunofluorescent staining, brains were cryoprotected in 20% sucrose overnight, frozen in CO2 snow and cut into 10 series of parallel 20 μm-thick coronal sections on a cryostat. Sections were stored at −20 °C until further processing.
Every sixth section (1 series) from mice with 24 hours and 5 days survival after pMCAO was stained with TB for estimation of infarct volume using the Cavalieri principle28,56. The rostrocaudal distribution of the infarct was analyzed as previously described5 using the anterior commisure as an antomical landmark.
The volume of the neocortex was estimated in naïve TNFfl/fl and LysMcreTNFfl/fl mice using Cavalieri’s principle as previously decribed in detail5.
In fresh-frozen sections, TNF was detected using a rabbit anti-TNF antibody (clone R4-6A2, 1:200, Pierce Antibody Products) and AP-conjugated anti-rabbit antibody as previously described27. Primary microglia and primary astrocytes were stained for microglial CD11b (clone M1/70.15, 1:600, Abdserotec) and astroglial GFAP (1:4,000, Dako) in order to evaluate the purity of the cultures. Substitution of the primary antibody with rabbit IgG (DakoCytomation) or IgG1 gave no signal.
Immunofluorescent staining for TNF co-locatization was performed essentially as described in Clausen et al.57 using rabbit anti-TNF, rat anti-CD11b and mouse anti-human GFAP-Alexa Fluor-488 (clone 131-17719, 1:400, Life Technologies) antibodies. Donkey anti-rabbit Alexa Fluor-594 and goat anti-rat Alexa Fluor-488 antibodies (1:200, Invitrogen) were used as secondary antibodies. Rabbit IgG was used as a serum control, rat IgG2b (IG-851125, Biosite) and mouse IgG-488 (Invitrogen) were used as isotype controls. Substitution of the primary antibody with isotype control or rabbit IgG gave no signal. Digital images were captured using an Olympus BX51 microscope connected to a PC containing DP manager software (version 184.108.40.206). Figures were composed using Photoshop CS5 software.
LysMcre co-localization in the adult brain was investigated using rat anti-CD11b (clone M1/70.15, 1:500, Invitrogen), rat anti-GFAP (clone 2.2B10, 1:2,000, Invitrogen), mouse anti-NeuN (clone A60, 1:200, Millipore), rabbit anti-red fluorescent protein (RFP) (1:4,000, Rockland), and rabbit anti-Olig2 (1:200, Millipore). Donkey anti-mouse Alexa Fluor-488, donkey anti-rat Alexa Fluor-488, donkey anti-rabbit Alexa Fluor-488, and goat anti-rabbit Alexa Fluor-568 (1:500) were used as secondary antibodies. DAPI (4’,6-diamidino-2-phenylindole) was used as a nuclear marker. Sections were coverslipped using Vectashield (Vectorlabs). Digital images of the cortex were captured using a Nikon Eclipse Ti epifluorescent microscope with NIS Elements software.
Estimation of cell numbers
Quantification of LysM+ cells was based on co-localization of the tdTomato reporter signal, which was enhanced using a RFP antibody with the various cell identification markers (NeuN, CD11b, and GFAP). Co-localization with oligodendroglial Olig2 was performed with the endogenous tdTomato signal. In total, 151–171 LysM+ (tdTomato) cells spread across at least three non-consecutive tissue sections were counted per animal (n = 3). All cells had a clearly identifiable DAPI+ nucleus.
Primary microglia and astrocyte cultures
For preparation of primary microglia and astrocyte cultures, 0–2 day old LysMcreTNFfl/fl and TNFfl/fl pups were decapitated, the brains isolated, and the meninges removed. Tail biopsies were obtained for genotyping. Brains were then placed in pre-warmed DMEM-GlutaMAX culture media (Gibco) containing 20% FCS and 1% penicillin/streptomycin (Gibco), mechanically homogenized by pipetting and filtered through a 70 μm cell strainer (Falcon). Cells corresponding to the number obtained from 1 brain were seeded into poly L-ornithine pre-coated 24-well culture dishes (Costar) and grown at 36 °C (20% O2 and 5% CO2) for up to 21 days. Fifty percent of the culture medium was changed every 3–4 days. After 7 days, the medium was changed to culture medium containing 10% FCS and 1% penicillin/streptomycin and cells grown for an additional 7 days. After 14 days in vitro, microglia were removed from the layer of confluent astrocytes by gently shaking at 100 rpm for 2 hours at room temperatur. Microglia were harvested from the same dish twice with 7 days intervals. Microglia and astrocytes were separated and centrifuged separately at 1,000× g for 5 min at room temperature. The supernatant was removed and cells resuspended in medium before plated into 24-well plates (20,000 cells/well). Microglia were then stimulated with 10 ng/mL lipopolysaccharide (LPS) (from Escherichia coli 0111:B4, Sigma) and astrocytes with either 10 ng/ml recombinant TNF (Sigma) or 10 ng/ml recombinant interleukin (IL)-1β (Sigma) for 24 hours. Unstimulated cultures were used as controls. After 24 hours, the media was removed and stored at −80 °C until further analysis. Cells were then prepared for immunohistochemistry by 3 x rinses in 0.15 M Sorensen Buffer (SB), fixation for 20 min i 4% PFA, 3 x rinses in 0.15 M SB and stored in 0.15 M SB for evaluation of purity (described above), which was found to be >95% (data not shown).
ELISA for TNF
The amount of TNF present in the media from LPS-stimulated and unstimulated microglia and TNF- or IL-1β-stimulated or unstimulated astrocyte cultures was examined using the Mouse TNF ELISA MAXTM Deluxe Sets (Biolegend) according to the manufacturer’s instructions. The obtained data was processed and analyzed by the interface software program SoftMax®Pro. Each standard and test was determined in duplicate.
Cre expression in the cortex of LysMtdTomato mice
For flow cytometry, cells from LysMtdTomato mice were isolated as previously described58, with minor modifications. Briefly, after transcardial perfusion with PBS, cortices were dissected out and placed in cold Hanks’ Balanced Salt Solution (HBSS). Tissues were enzymatically dissociated using the neural tissue dissociation kit containing papain (Miltenyi Biotec) and washed in cold HBSS. Suspensions were cleared from myelin by incubation with Myelin Removal Beads II followed by negative selection with LS columns (Miltenyi Biotech), according to manufacturer’s instructions. Cells were washed in FACS buffer (eBioscience), blocked with anti-CD16/32 (FcR block, eBioscience) to prevent non-specific staining, and stained with FITC-anti-CD45 (1:200, eBioscience) and PE-Cy7-anti-CD11b (1:200, eBioscience). Cell suspensions were fixed in 1% PFA and resuspended in FACS buffer. Samples were analyzed with an LSRII flow cytometer (BD Biosciences) equipped with FACS-Diva 6.0 software (BD Biosciences).
Cell population analysis in TNFfl/fl and LysMcreTNFfl/fl mice
Flow cytometry was performed as previously described using the FACSVerse (BD Biosciences) and data analyzed using the FACSuite software3,5. Microglia [CD11b+CD45dim or CD11b+CD45dimLy6ClowLy6G−], macrophages [CD11b+CD45highGr1− or CD11b+CD45highLy6ChighLy6G−], granulocytes [CD11b+CD45highGr1+ or CD11b+CD45highLy6C+Ly6G+], T cells [CD45highCD3+], TNF+ microglia and TNF+ macrophages were identified as previously described3,5.
Prior to fixation, cells were stained for live/dead cells using a Fixable Viability Dye eFluoro 506 (eBioscience) diluted in PBS5. For TNF expression analysis, cell suspensions were incubated for 4 to 6 hours in vitro with the protein transport inhibitor GolgiPlug (BD Biosciences) and processed as previously described3,4. A total of 1,000,000 events were collected using FSC and SSC and analysis of the live/dead gate revealed comparable numbers of dead cells in all the samples. Furthermore, blood and spleen samples, in addition to bone marrow at 24 hours after pMCAO, were collected and analyzed for CD45, CD11b, Gr1, and CD3 markers.
Positive staining was determined based on fluorescence levels of the respective isotype and fluorescence minus one (FMO) controls. Antibodies were directly conjugated with fluorochromes: PerCP Cy5.5 anti-CD45 (BD Biosciences, clone 30-F11), PE anti-CD11b (BD Biosciences, clone M1/70), PE-Cy7 anti-Ly6G/Ly6C (Gr1) (Biolegend, clone RB6-8C5), PE-Cy7 anti-Ly6C (BD Biosciences, clone AL-21), BV421 anti-Ly6G (BD Biosciences, clone 1A8), APC anti-TNF (Biolegend, clone MP6-XT22), and APC anti-CD3 (BD Biosciences, clone 145-2C11). Isotype controls used were hamster IgG1κ (BD Biosciences, clone A19-3), rat IgG2b (BD Biosciences, clone A95-1 or Biolegend, clone RTK4530), rat IgG1κ (Biolegend, clone RTK2071), rat IgMκ (BD Biosciences, clone R4-22), Lewis IgG2aκ (BD Biosciences, clone R35-95), and mouse IgG1 (BD Biosciences, clone MOPC-21). The mean fluorescence intensity (MFI) was calculated as the geometric mean of each population in the TNF, CD45 and CD11b positive gates, respectively3.
Multiplex and Western blotting analyses
One series of brain tissue was lysed in 1x Tris Lysis buffer (150 nM NaCl, 20 mM Tris, 1 mM EDTA, 1 mM EGTA, 1% Triton-x100, pH 7.5) containing phosphatase inhibitor (Sigma) and cOmplete mini, EDTA-free proteinase inhibitor cocktail (Roche). Protein estimations were performed using the Bradford method.
Plasma and brain samples were measured in the MSD Mouse Proinflammatory V-Plex Plus Kit (IFNγ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, CXCL1, TNF; K15012C, Mesoscale) and in brain samples the mouse TNF-RI and TNF-RII Ultra-Sensitive Kits (Mesoscale) using a MSD QuickPlex (SQ120) Plate Reader (Mesoscale) according to the manufacturer’s instructions5. Samples were run in duplex and diluted 2- or 4-fold in Diluent 41 prior to measurement. Data was analyzed using MSD Discovery Workbench software. The lower limit of detection was a calculated concentration based on a signal 2.5 SD above the blank (zero) calibrator.
Western blot analysis
Western blot analysis for TNF (1:2,000, Abcam) was performed using 20 μg protein extract separated on 4–12% SDS-PAGE gels (Nupage, Invitrogen) essentially as previously described3,4. As a positive control, 0.5 ng 17 kDa murine recombinant TNF (Sigma Aldrich) was included. Western blotting analysis for SAPK/JNK (Cell Signaling, 1:1,000), phosphorylated (p)-SAPK/JNK (Thr183/Tyr185) (Cell Signaling, 1:1,000), p44/p42 MAPK (Cell Signaling, 1:1,000), p-p44/p-p42 MAPK (ERK1/2)(Cell Signaling, 1:1,000), p-38 (Cell Signaling, 1:1,000), p-p38 MAPK (Tyr180/Tyr182)(Cell Signaling, 1:1,000), and Iba1 (Wako, 1:500) was performed by resolving equal amounts of protein lysates by SDS-PAGE on NuPage Bis 4–12% gels, using MOPS SDS (Invitrogen) containing 0.25% antioxidant (Invitrogen) essentially as previously described6,59. TFIIB (1:1,000, Cell Signaling), GAPDH (1:2,500, Abcam) and α-actin (1:8,000, Millipore) were used as loading controls. SeeBlue Plus2 prestained standard (Invitrogen) was used as a molecular weight marker. Bands were quantified with Quantity One software (Biorad). Analysis was performed on 2–4 independent gels with n = 2–3/group and data were normalized to TFIIB, GAPDH or α-actin and represented as percentages relative to naïve TNFfl/fl mice or as ratios (for MAPK).
Human stroke tissue
The study was performed on post mortem brain tissue from two stroke cases obtained from the Department of Pathology, Odense University Hospital and the use of human brains was approved by the Danish Biomedical Research Ethical committee for the Region of Southern Denmark (permission number S-20080042). Sex, age, infarcted brain area and age of infarct are given in Suppl. Table 4. Specimens from these brains have been part of a previous study on surfactant protein-D in human ischemic brain tissue60.
Immunohistochemistry in human stroke tissue
Tissue blocks containing infarcted human brain tissue were imbedded in paraffin and 2 μm thick, serial sections were cut on a microtome, deparaffinized, blocked in 1.5% H2O2 in Tris-buffered saline (TBS) and demasked in T-EG buffer (10 mM Tris + 0.5 mM EGTA, pH 9.0), whereafter they were loaded onto a Dako autostainer (Dako, Denmark)61. Sections were stained using the following primary antibodies: rabbit anti-Iba1 (ionized calcium binding adaptor molecule 1, 1:1,000, Wako), mouse anti-CD45 (1:200, Dako), mouse anti-CD68 (1:100, Dako), rabbit anti-GFAP (glial fibrillary acidic protein, 1:2,000, Dako), rabbit anti-TNF (1:100, Endogen)4,23,27, rabbit anti-TNFR1 (1:50 (H-271), Santa Cruz), and rabbit anti-TNFR2 (1:50, Sigma-Aldrich). Secondary antibodies included EnVision + System horseradish peroxidase (HRP)-labelled Polymer (Dako) for CD45, CD68 and GFAP, Advance HRP (Dako) for TNF and TNFR1, and PowerVision + Poly-HRP IHC detection system (AH Diagnostics) for TNFR2. Omission of primary antibody and comparable concentrations of rabbit IgG were used to control for unspecific binding in the immunohistochemical protocols for TNF, TNFR1 and TNFR2 and these sections were devoid of staining. In addition, parallel sections were stained for hematoxylin and eosin (HE) using standard protocols at the Department of Pathology, Odense University Hospital.
For multiple comparisons, one-way or two-way ANOVA was applied. When appropriate, repeated measures were used. Each test was followed by the relevant post-hoc test. For single comparisons unpaired and paired student’s t-tests were applied. P values ≤ 0.05 were considered statistically significant. Data are presented as mean ± SEM.
How to cite this article: Clausen, B. H. et al. Conditional ablation of myeloid TNF increases lesion volume after experimental stroke in mice, possibly via altered ERK1/2 signaling. Sci. Rep. 6, 29291; doi: 10.1038/srep29291 (2016).
Schwartz, M., Butovsky, O., Bruck, W. & Hanisch, U. K. Microglial phenotype: is the commitment reversible? Trends Neurosci 29, 68–74 (2006).
Lambertsen, K. L., Biber, K. & Finsen, B. Inflammatory cytokines in experimental and human stroke. J Cereb Blood Flow Metab 32, 1677–1698, doi: 10.1038/jcbfm.2012.88 (2012).
Clausen, B. et al. Systemically administered anti-TNF therapy ameliorates functional outcomes after focal cerebral ischemia. J Neuroinflammation 11, 203, doi: 10.1186/PREACCEPT-2982253041347736 (2014).
Lambertsen, K. L. et al. Microglia protect neurons against ischemia by synthesis of tumor necrosis factor. J Neurosci 29, 1319–1330, doi: 10.1523/JNEUROSCI.5505-08.2009 (2009).
Madsen, P. M. et al. Genetic ablation of soluble tumor necrosis factor with preservation of membrane tumor necrosis factor is associated with neuroprotection after focal cerebral ischemia. J Cereb Blood Flow Metab, doi: 10.1177/0271678X15610339 (2015).
Novrup, H. G. et al. Central but not systemic administration of XPro1595 is therapeutic following moderate spinal cord injury in mice. J Neuroinflammation 11, 159, doi: 10.1186/s12974-014-0159-6 (2014).
Probert, L. TNF and its receptors in the CNS: The essential, the desirable and the deleterious effects. Neuroscience, doi: 10.1016/j.neuroscience.2015.06.038 (2015).
Alexopoulou, L. et al. Transmembrane TNF protects mutant mice against intracellular bacterial infections, chronic inflammation and autoimmunity. Eur J Immunol 36, 2768–2780, doi: 10.1002/eji.200635921 (2006).
Brambilla, R. et al. Inhibition of soluble tumour necrosis factor is therapeutic in experimental autoimmune encephalomyelitis and promotes axon preservation and remyelination. Brain 134, 2736–2754, doi: 10.1093/brain/awr199 (2011).
Chen, G. & Goeddel, D. V. TNF-R1 signaling: a beautiful pathway. Science 296, 1634–1635 (2002).
Subramaniam, S. & Unsicker, K. ERK and cell death: ERK1/2 in neuronal death. FEBS J 277, 22–29, doi: 10.1111/j.1742-4658.2009.07367.x (2010).
Marchetti, L., Klein, M., Schlett, K., Pfizenmaier, K. & Eisel, U. L. Tumor necrosis factor (TNF)-mediated neuroprotection against glutamate-induced excitotoxicity is enhanced by N-methyl-D-aspartate receptor activation. Essential role of a TNF receptor 2-mediated phosphatidylinositol 3-kinase-dependent NF-kappa B pathway. J Biol Chem 279, 32869–32881 (2004).
Taoufik, E. et al. Transmembrane tumour necrosis factor is neuroprotective and regulates experimental autoimmune encephalomyelitis via neuronal nuclear factor-kappaB. Brain 134, 2722–2735, doi: 10.1093/brain/awr203 (2011).
Taoufik, E. et al. FLIP(L) protects neurons against in vivo ischemia and in vitro glucose deprivation-induced cell death. J Neurosci 27, 6633–6646 (2007).
Bruce, A. J. et al. Altered neuronal and microglial responses to excitotoxic and ischemic brain injury in mice lacking TNF receptors. Nat Med 2, 788–794 (1996).
Pradillo, J. M. et al. TNFR1 upregulation mediates tolerance after brain ischemic preconditioning. J Cereb Blood Flow Metab 25, 193–203 (2005).
Frank, M. & Wolburg, H. Cellular reactions at the lesion site after crushing of the rat optic nerve. Glia 16, 227–240, doi: 10.1002/(SICI)1098-1136(199603)16:3<227::AID-GLIA5>3.0.CO;2-Z (1996).
Atarashi, R. et al. Abnormal activation of glial cells in the brains of prion protein-deficient mice ectopically expressing prion protein-like protein, PrPLP/Dpl. Mol Med 7, 803–809 (2001).
Hedtjarn, M., Mallard, C. & Hagberg, H. Inflammatory gene profiling in the developing mouse brain after hypoxia-ischemia. J Cereb Blood Flow Metab 24, 1333–1351, doi: 10.1097/01.WCB.0000141559.17620.36 (2004).
Clausen, B. H., Lambertsen, K. L. & Finsen, B. Glyceraldehyde-3-phosphate dehydrogenase versus toluidine blue as a marker for infarct volume estimation following permanent middle cerebral artery occlusion in mice. Exp Brain Res 175, 60–67, doi: 10.1007/s00221-006-0526-3 (2006).
Goel, G. et al. Combined effect of tumor necrosis factor (TNF)-alpha and heat shock protein (HSP)-70 in reducing apoptotic injury in hypoxia: a cell culture study. Neurosci Lett 483, 162–166, doi: 10.1016/j.neulet.2010.07.069 (2010).
Lambertsen, K. L., Meldgaard, M., Ladeby, R. & Finsen, B. A quantitative study of microglial-macrophage synthesis of tumor necrosis factor during acute and late focal cerebral ischemia in mice. J Cereb Blood Flow Metab 25, 119–135, doi: 10.1038/sj.jcbfm.9600014 (2005).
Lambertsen, K. L., Gregersen, R. & Finsen, B. Microglial-macrophage synthesis of tumor necrosis factor after focal cerebral ischemia in mice is strain dependent. J Cereb Blood Flow Metab 22, 785–797. (2002).
Bianchi, M., Sacerdote, P., Ricciardi-Castagnoli, P., Mantegazza, P. & Panerai, A. E. Central effects of tumor necrosis factor alpha and interleukin-1 alpha on nociceptive thresholds and spontaneous locomotor activity. Neurosci Lett 148, 76–80 (1992).
Golan, H., Levav, T., Mendelsohn, A. & Huleihel, M. Involvement of tumor necrosis factor alpha in hippocampal development and function. Cereb Cortex 14, 97–105 (2004).
Grivennikov, S. I. et al. Distinct and nonredundant in vivo functions of TNF produced by t cells and macrophages/neutrophils: protective and deleterious effects. Immunity 22, 93–104, doi: 10.1016/j.immuni.2004.11.016 (2005).
Lambertsen, K. L., Gregersen, R., Drojdahl, N., Owens, T. & Finsen, B. A specific and sensitive method for visualization of tumor necrosis factor in the murine central nervous system. Brain Res Brain Res Protoc 7, 175–191 (2001).
Gregersen, R., Lambertsen, K. & Finsen, B. Microglia and macrophages are the major source of tumor necrosis factor in permanent middle cerebral artery occlusion in mice. J Cereb Blood Flow Metab 20, 53–65 (2000).
Ginsberg, M. D., Reivich, M., Giandomenico, A. & Greenberg, J. H. Local glucose utilization in acute focal cerebral ischemia: local dysmetabolism and diaschisis. Neurology 27, 1042–1048 (1977).
Yuan, H. et al. Spatiotemporal uptake characteristics of f-2-fluoro-2-deoxy-d-glucose in a rat middle cerebral artery occlusion model. Stroke 44, 2292–2299, doi: 10.1161/STROKEAHA.113.000903 (2013).
Schroeter, M. et al. Neuroinflammation extends brain tissue at risk to vital peri-infarct tissue: a double tracer [11C]PK11195- and [18F]FDG-PET study. J Cereb Blood Flow Metab 29, 1216–1225, doi: 10.1038/jcbfm.2009.36 (2009).
Subramaniam, S. & Unsicker, K. Extracellular signal-regulated kinase as an inducer of non-apoptotic neuronal death. Neuroscience 138, 1055–1065, doi: 10.1016/j.neuroscience.2005.12.013 (2006).
Tran, S. E., Holmstrom, T. H., Ahonen, M., Kahari, V. M. & Eriksson, J. E. MAPK/ERK overrides the apoptotic signaling from Fas, TNF, and TRAIL receptors. J Biol Chem 276, 16484–16490, doi: 10.1074/jbc.M010384200 (2001).
Barbin, G., Roisin, M. P. & Zalc, B. Tumor necrosis factor alpha activates the phosphorylation of ERK, SAPK/JNK, and P38 kinase in primary cultures of neurons. Neurochem Res 26, 107–112 (2001).
Rousseau, S. et al. TPL2-mediated activation of ERK1 and ERK2 regulates the processing of pre-TNF alpha in LPS-stimulated macrophages. J Cell Sci 121, 149–154, doi: 10.1242/jcs.018671 (2008).
Humphreys, B. D. & Dubyak, G. R. Modulation of P2X7 nucleotide receptor expression by pro- and anti-inflammatory stimuli in THP-1 monocytes. J Leukoc Biol 64, 265–273 (1998).
Suzuki, T. et al. Production and release of neuroprotective tumor necrosis factor by P2X7 receptor-activated microglia. J Neurosci 24, 1–7, doi: 10.1523/JNEUROSCI.3792-03.2004 (2004).
Ritzel, R. M. et al. Functional differences between microglia and monocytes after ischemic stroke. J Neuroinflammation 12, 106, doi: 10.1186/s12974-015-0329-1 (2015).
Denes, A. et al. Experimental stroke-induced changes in the bone marrow reveal complex regulation of leukocyte responses. J Cereb Blood Flow Metab 31, 1036–1050, doi: 10.1038/jcbfm.2010.198 (2011).
Brambilla, R., Couch, Y. & Lambertsen, K. L. The effect of stroke on immune function. Mol Cell Neurosci, doi: 10.1016/j.mcn.2012.08.011 (2012).
Clausen, B. H. et al. Cell therapy centered on IL-1Ra is neuroprotective in experimental stroke. Acta Neuropathol, doi: 10.1007/s00401-016-1541-5 (2016).
Denes, A. et al. Central autonomic control of the bone marrow: multisynaptic tract tracing by recombinant pseudorabies virus. Neuroscience 134, 947–963, doi: 10.1016/j.neuroscience.2005.03.060 (2005).
Prass, K. et al. Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke T helper cell type 1-like immunostimulation. J Exp Med 198, 725–736, doi: 10.1084/jem.20021098 (2003).
Offner, H. et al. Splenic atrophy in experimental stroke is accompanied by increased regulatory T cells and circulating macrophages. J Immunol 176, 6523–6531 (2006).
Izci, Y. Splenectomy may be a prophylactic treatment for cerebral ischemia? Med Hypotheses 75, 347–349, doi: 10.1016/j.mehy.2010.03.018 (2010).
Solovjov, D. A., Pluskota, E. & Plow, E. F. Distinct roles for the alpha and beta subunits in the functions of integrin alphaMbeta2. J Biol Chem 280, 1336–1345, doi: 10.1074/jbc.M406968200 (2005).
Campbell, S. J. et al. Immunomodulatory effects of etanercept in a model of brain injury act through attenuation of the acute-phase response. J Neurochem 103, 2245–2255, doi: 10.1111/j.1471-4159.2007.04928.x (2007).
Garcia, J. H. et al. Influx of leukocytes and platelets in an evolving brain infarct (Wistar rat). Am J Pathol 144, 188–199 (1994).
Cho, I. H. et al. Role of microglial IKKbeta in kainic acid-induced hippocampal neuronal cell death. Brain 131, 3019–3033, doi: 10.1093/brain/awn230 (2008).
Arenkiel, B. R. et al. Activity-induced remodeling of olfactory bulb microcircuits revealed by monosynaptic tracing. PLoS One 6, e29423, doi: 10.1371/journal.pone.0029423 (2011).
Zhu, Y. et al. Hematogenous macrophage depletion reduces the fibrotic scar and increases axonal growth after spinal cord injury. Neurobiol Dis 74, 114–125, doi: 10.1016/j.nbd.2014.10.024 (2015).
Lambertsen, K. L. et al. Genetic KCa3.1-deficiency produces locomotor hyperactivity and alterations in cerebral monoamine levels. PLoS One 7, e47744, doi: 10.1371/journal.pone.0047744 (2012).
Lanz, B., Poitry-Yamate, C. & Gruetter, R. Image-Derived Input Function from the Vena Cava for 18F-FDG PET Studies in Rats and Mice. Journal of nuclear medicine : official publication, Society of Nuclear Medicine 55, 1380–1388, doi: 10.2967/jnumed.113.127381 (2014).
Fueger, B. J. et al. Impact of animal handling on the results of 18F-FDG PET studies in mice. Journal of nuclear medicine: official publication, Society of Nuclear Medicine 47, 999–1006 (2006).
Metz, G. A. & Whishaw, I. Q. Cortical and subcortical lesions impair skilled walking in the ladder rung walking test: a new task to evaluate fore- and hindlimb stepping, placing, and co-ordination. J Neurosci Methods 115, 169–179 (2002).
Lin, T. N., He, Y. Y., Wu, G., Khan, M. & Hsu, C. Y. Effect of brain edema on infarct volume in a focal cerebral ischemia model in rats. Stroke 24, 117–121 (1993).
Clausen, B. H. et al. Interleukin-1beta and tumor necrosis factor-alpha are expressed by different subsets of microglia and macrophages after ischemic stroke in mice. J Neuroinflammation 5, 46, doi: 10.1186/1742-2094-5-46 (2008).
Brambilla, R. et al. Astrocytes play a key role in EAE pathophysiology by orchestrating in the CNS the inflammatory response of resident and peripheral immune cells and by suppressing remyelination. Glia 62, 452–467, doi: 10.1002/glia.22616 (2014).
Bracchi-Ricard, V. et al. Inhibition of astroglial NF-kappaB enhances oligodendrogenesis following spinal cord injury. J Neuroinflammation 10, 92, doi: 10.1186/1742-2094-10-92 (2013).
Lambertsen, K. L. et al. No effect of ablation of surfactant protein-D on acute cerebral infarction in mice. J Neuroinflammation 11, 123, doi: 10.1186/1742-2094-11-123 (2014).
Hermansen, S. K., Christensen, K. G., Jensen, S. S. & Kristensen, B. W. Inconsistent immunohistochemical expression patterns of four different CD133 antibody clones in glioblastoma. J Histochem Cytochem 59, 391–407, doi: 10.1369/0022155411400867 (2011).
This work was supported by The Lundbeck Foundation (R54-A5539), The Carlsberg Foundation (2007_01_0176), The Novo Nordisk Foundation (R153-A-12550 and R168-A14120), Fonden til Lægevidenskabens Fremme (KLL), The Lundbeck Foundation (R67-A6383) (BHC), NINDS 1R01NS081040-01 (JKL), and Russian Science Foundation 14-50-00060 (SAN). The authors acknowledge the technical assistance provided by Signe Marie Andersen, Louise Lykkemark, Dorte Lyholmer, Ulla Damgaard Munk, Sussanne Petersen, Helle Wohlleben, Mehran Taherian and Anni Petersen. Professors Poul Flemming Højlund and Daniel Clive Anthony are acknowledged for their input on PET imaging and Nicholas Ditzel for assistance with DXA scanning of mice. We thank Dr. Fan Wang for donating the Rosa26-tdTomato reporter mice.
The authors declare no competing financial interests.
About this article
Cite this article
Clausen, B., Degn, M., Sivasaravanaparan, M. et al. Conditional ablation of myeloid TNF increases lesion volume after experimental stroke in mice, possibly via altered ERK1/2 signaling. Sci Rep 6, 29291 (2016). https://doi.org/10.1038/srep29291
A novel biscoumarin derivative dephosphorylates ERK and alleviates apoptosis induced by mitochondrial oxidative damage in ischemic stroke mice
Life Sciences (2020)
Brain Sciences (2020)
Acta Neuropathologica Communications (2020)
Acute Neurofilament Light Chain Plasma Levels Correlate With Stroke Severity and Clinical Outcome in Ischemic Stroke Patients
Frontiers in Neurology (2020)
Intravenous Immunoglobulin (IVIg) Induce a Protective Phenotype in Microglia Preventing Neuronal Cell Death in Ischaemic Stroke
NeuroMolecular Medicine (2020)