Address correspondence and reprint requests to I. Lizasoain, Departamento de Farmacología, Facultad de Medicina, Universidad Complutense Madrid, 28040 Madrid, Spain. E-mail: email@example.com
It has been demonstrated that a short ischemic event (ischemic preconditioning, IPC) results in a subsequent resistance to severe ischemia (ischemic tolerance, IT). We have recently demonstrated the role of innate immunity and in particular of toll-like receptor (TLR) 4 in brain ischemia. Several evidences suggest that TLR4 might also be involved in IT. Therefore, we have now used an in vivo model of IPC to investigate whether TLR4 is involved in IT. A 6-min temporary bilateral common carotid arteries occlusion was used for focal IPC and it was performed on TLR4-deficient mice (C57BL/10ScNJ) and animals that express TLR4 normally (C57BL/10ScSn). To assess the ability of IPC to induce IT, permanent middle cerebral artery occlusion was performed 48 h after IPC. Stroke outcome was evaluated by determination of infarct volume and assessment of neurological scores. IPC caused neuroprotection as shown by a reduction in infarct volume and better outcome in mice expressing TLR4 normally. TLR4-deficient mice showed less IPC-induced neuroprotection than wild-type animals. Western blot analysis of tumor necrosis factor alpha (TNF-α), inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2) showed an up-regulation in the expression of these proteins in both substrains of mice measured 18, 24 and 48 h after IPC, being higher in mice with TLR4. Similarly, nuclear factor-kappa B (NF-κB) activation was observed 18, 24 and 48 h after IPC, being more intense in TLR4-expressing mice. These data demonstrate that TLR4 signalling is involved in brain tolerance as shown by the difference in the percentage of neuroprotection produced by IPC between ScSn and ScNJ (60% vs. 18%). The higher expression of TNF-α, iNOS and cyclooxygenase-2 and NF-κB activation in mice expressing TLR4 is likely to participate in this endogenous neuroprotective effect.
Brain ischemic tolerance (IT) describes the adaptive response of brain cells that is initiated by a short ischemic event (ischemic preconditioning: IPC) and the associated period during which their subsequent resistance to severe ischemia is increased (for review see Barone et al. 1998; Kirino 2002; Obrenovitch 2008). Ischemic tolerance has been also demonstrated in human clinical practice: less severe strokes have been described in patients with prior ipsilateral transient ischemics attacks within a short period of time (Weih et al. 1999; Moncayo et al. 2000; Castillo et al. 2003). This topic is attracting much attention as it helps to understand how the brain protects itself and also because its modulation with pharmacological strategies could have important therapeutic applications.
Several mechanisms of induction of tolerance have been described (for review see Barone et al. 1998; Kirino 2002; Dirnagl et al. 2003; Obrenovitch 2008), including some of inflammatory and excitotoxic nature which are also involved in brain damage, such as tumor necrosis factor alpha (TNF-α), inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2). In this context, results from our own group have demonstrated that TNF-alpha converting enzyme (TACE)/TNF-α/NF-κB pro-inflammatory pathway is implicated in both in vitro and in vivo models of ischemic preconditioning (Cárdenas et al. 2002; Hurtado et al. 2002; Romera et al. 2004; Pradillo et al. 2005). We also demonstrated that IPC decreases excitotoxicity by inducing the up-regulation of two glutamate transporters, excitatory amino acid transporter (EAAT) 3 and EAAT2; whereas EAAT3 up-regulation was partly mediated by TACE/TNF-α pathway (Romera et al. 2004); EAAT2 appeared as a novel peroxisome proliferator-activated receptor gamma (PPARγ) target gene (Romera et al. 2007).
Activation of the innate immune system is an important component of the inflammatory response which also takes place in the brain and utilizes toll-like receptors (TLRs; Medzhitov 2001). We recently demonstrated that TLR4 participates in inflammation and in brain damage after stroke and in the worsening of stroke after subacute stress, as shown by smaller infarcts in TLR4-deficient mice when compared with wild type controls (Caso et al. 2007, 2008). In contrast to the detrimental role of TLRs in response to ischemia, stimulation of some of these receptors prior to ischemia provides robust neuroprotection (see Marsh et al. 2008). Indeed, it has been recently reported that TLR9 and TLR2 are involved in the preconditioning-induced protective effect against ischemic brain injury (Hua et al. 2008; Stevens et al. 2008). Interestingly several evidences suggest that TLR4 might also be involved in ischemic tolerance: First, TLR4 signalling activates nuclear factor-kappa B (NF-κB) as well as the expression of mediators such as iNOS or COX-2, which have been described to participate in IT (for review see Barone et al. 1998; Kirino 2002; Dirnagl et al. 2003; Obrenovitch 2008). Second, preconditioning with lipopolysaccharide (LPS), a TLR4 ligand, produces ischemic tolerance through a TNF-α-dependent process (for review see Tasaki et al. 1997; Karikóet al. 2004; Rosenzweig et al. 2007).
These pieces of evidence prompted us to investigate whether TLR4 is involved in ischemic tolerance with loss of function studies using mice which do not express TLR4, exposed to a temporary bilateral common carotid arteries occlusion as in vivo model of IPC.
Material and methods
Adult male C57BL/10ScNJ and C57BL/10ScSn mice weighing 28–30 g were used (Jackson Labs, Bar Harbor, ME, USA). The murine strain C57BL/10ScNJ does not express TLR4 because of a naturally total deletion of the TLR4 gene (Poltorak et al. 1998). C57BL/10ScSn substrain does not express the mutation and it is considered as the control group. All experimental protocols adhered to the guidelines of the Animal Welfare Committee of the Universidad Complutense (following DC 86/609/EU). Mice were housed under standard conditions of temperature and humidity and a 12 h-light/dark cycle (lights on at 08:00) with free access to food and water.
Mice were anesthetised with 5% isofluorane (in 70%N2O, 30%O2) for induction and 1.5% isofluorane for maintenance. The rectal temperature was maintained at 37°C with a heating pad. IPC was induced by a period of 6 min of temporary bilateral common carotid arteries occlusion (BCCAO) as described previously (Wu et al. 2001). This period of 6 min of BCCAO was used because it protects the brain from later ischemia without causing tissue damage by itself. During this period, the cerebral blood flow, which was monitored by laser Doppler flowmetry (Periflux System PF5010 Perimed, Stockholm, Sweden), decreased enough as to produce a non-lethal ischemia (Pradillo et al. 2005). Mice in which bilateral common carotid arteries were exposed but not occluded served as sham-operated animals.
Permanent focal ischemia
To evaluate the ability of IPC (BCCAO) to produce IT, permanent middle cerebral artery occlusion (pMCAO) was performed 48 h after sham or BCCAO surgery. Middle cerebral artery (MCA) was exposed and occluded permanently by electrocoagulation as previously described (Caso et al. 2007, 2008). Briefly, for the MCA occlusion (MCAO), an incision perpendicular to the line connecting the lateral canthus of the left eye and the external auditory canal was made to expose and retract the temporalis muscle. A burr hole was drilled and the MCA was exposed by cutting and retracting the dura. The MCA was elevated and cauterised. Following surgery, subjects were returned to their cages and allowed free access to water and food. The survival rate of the animals until the end of the experiment was 80%.
Four groups were used for determination of iNOS, COX-2, TNFα expression and NF-κB activation: (i) Sham-operated mice from both substrains C57BL/10ScNJ (ScNJ) and C57BL/10ScSn (ScSn), used as control groups (Sham-ScNJ and Sham-ScSn); (ii) BCCAO, used as IPC groups (IPC-ScNJ and IPC-ScSn). Six to ten animals per group were used for the analysis of protein expression levels and NF-κB activation. In addition, to evaluate the ability of IPC to produce IT, pMCAO was performed 48 h after sham or BCCAO surgery and the infarct area was measured 24 h after pMCAO. Therefore, four additional groups were established as: (i) SHAM + pMCAO, ScNJ (n = 8), (ii) IPC + pMCAO, ScNJ (n = 8), (iii) SHAM + pMCAO, ScSn (n = 8), (iv) IPC + pMCAO, ScSn (n = 8).
Physiological parameters (rectal temperature, mean arterial pressure, blood glucose levels) were not significantly different between all studied groups (data not shown). No spontaneous mortality was found after MCAO with this model, and this was unaffected by the different experimental treatments.
Brains were removed 24 h after pMCAO, and cut into seven of 1-mm coronal brain slices (Brain Matrix, WPI, UK), which were stained in 1% TTC (2,3,5-triphenyl-tetrazolium chloride, Merck, Madrid, Spain) in 0.1 mol/L phosphate buffer, and infarct size was determined as described (Mackensen et al. 2001). Infarct volumes were measured by sampling stained sections with a digital camera (Nikon Coolpix 990, Nikon Corporation, Tokyo, Japan), and the image of each section was analyzed using ImageJ 1.33u (NIH, Bethesda, MD, USA). The digitalised image was displayed on a video monitor. With the observer masked to the experimental conditions, the contralateral hemisphere perimeter was overlapped with the ipsilateral hemisphere in order to exclude edema, and infarct borders were delineated using an operator-controlled cursor. The area of infarct, which was not stained, was determined by counting the number of pixels contained within the outlined regions of interest and expressed in square millimetres. Infarct volumes (in mm3) were integrated from the infarct areas over the extent of the infarct calculated as an orthogonal projection. All the animals used presented infarctions after the occlusion, which include cortex, subcortex and striatum, depending on the intensity of the lesion.
Prior to kill, neurological deficits were measured as previously described (Hunter et al. 2000) according to the following graded scoring system: 0 = no deficit; 1 = flexion of contralateral torso and forelimb upon lifting of the whole animal by the tail; 2 = circling to the contralateral side, when held by tail with feet on floor; 3 = spontaneous circling to contra-lateral side; 4 = no spontaneous motor activity. Each animal was scored for each of the above behaviours for approximately 1 min. A total of four trials were performed in order to optimize the consistency of the test. The results were expressed as the average value obtained for each animal.
Protein expression in brain homogenates and in cytosolic and nuclear extracts
Brain cortical tissue was collected from the preconditioned and surrounding areas. For determination of iNOS, COX-2 and TNFα protein expression levels, mice were killed 2, 18, 24 and 48 h after IPC. Brain areas corresponding to the preconditioned and surrounding area were collected and homogenised as previously described (Pereira et al. 2005). NF-κB was determined in cytosolic and nuclear extracts which were prepared as described (Pereira et al. 2005) and obtained from brains of mice killed 2, 18, 24 and 48 h after IPC.
Western blot analysis
Samples containing 40 μg of protein were loaded and the proteins size-separated in 7–10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (90 V). Proteins were blotted onto a polyvinylidene difluoride membrane (HybondTM-P, Amersham Biosciences Europe GmbH, Friburg, Germany) and incubated with specific primary antibodies against iNOS (Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1 : 1000), COX-2 (Santa Cruz, 1 : 2000), pro-TNFα and TNFα (Preprotech, Rocky Hill, NJ, USA 1 : 10 000), p65 (Santa Cruz, 1 : 500) and Inhibitory kappa B alpha (IκBα) (Santa Cruz, 1 : 500). Proteins recognised by the antibody were revealed by ECLTM-kit following manufacturers instructions (Amersham Biosciences Europe GmbH, Friburg, Germany). β-actin and Sp1 levels were used as loading controls for total-cytosolic and nuclear protein expression respectively.
Materials and statistical analysis
Reagents were from Sigma (Madrid, Spain) or as indicated in the text. Results are expressed as mean ± SEM of the indicated number of experiments; statistical analysis involved one way analysis of variance (anova or the Kruskal–Wallis test when the data were not normally distributed) followed by individual comparisons of means (Student–Newman–Keuls or Dunn’s method when the data were not normally distributed). Comparisons between the groups of mice under two different factors (presence or not of pMCAO and gene deficiency) were performed with two-way anova with post hoc Newman–Keuls test (intergroup analysis). p < 0.05 was considered statistically significant.
Effect of ischemic preconditioning on infarct outcome in TLR4-deficient mice after pMCAO
In mice which express TLR4 normally (C57Bl/10ScSn), ischemic preconditioning produced a reduction in infarct volume and areas and a better functional outcome measured 24 h after pMCAO (IPC + MCAO, 13.9 ± 2 mm3; 40% of SHAM + MCAO value, n = 8) when compared with the no preconditioned animals (SHAM + MCAO, 35.4 ± 3 mm3, 100%; n = 8; p < 0.05; Fig. 1 and Table 1). In TLR4-deficient mice (C57Bl/10ScNJ), IPC also produced a protective effect (SHAM + MCAO, 28.2 ± 2 vs. IPC + MCAO, 23.1±2 mm3; 18% reduction, p < 0.05; Fig. 1 and Table 1). However, the protective effect found in TLR4-deficient mice was smaller when compared with mice expressing TLR4 as demonstrated when percentages of protection were compared (60% in ScSn vs. 18% in ScNJ; p < 0.05; Fig. 1).
Table 1. Effect of ischemic preconditioning (IPC) on neurological status in TLR4-deficient mice (C57Bl/ScNJ) and mice that express TLR4 normally (C57Bl/ScSn) after pMCAO. Data are shown as the number of animals that showed each neurological score versus total number of animals per group
Neurological score (points)
SHAM + pMCAO C57BL/ScSn (n = 8)
IPC + pMCAO C57BL/ScSn (n = 8)
SHAM + pMCAO C57BL/ScNJ (n = 8)
IPC + pMCAO C57BL/ScNJ (n = 8)
*p < 0.05 vs. SHAM + pMCAO, #p < 0.05 vs. IPC + pMCAO ScSn.
No deficit (0)
0 of 8
0 of 8
0 of 8
0 of 8
2 of 8
8 of 8
4 of 8
7 of 8
4 of 8
0 of 8
4 of 8
1 of 8
Spontaneous circling (3)
2 of 8
0 of 8
0 of 8
0 of 8
No activity (4)
0 of 8
0 of 8
0 of 8
0 of 8
Expression of TNF-α, iNOS and COX-2 in TLR4-deficient mice after ischemic preconditioning
The protein expression of TNF-α, iNOS and COX-2 were studied, as mediators which are known to mediate tolerance after ischemic preconditioning (Nawashiro et al. 1997; Dirnagl et al. 2003; rev in Obrenovitch 2008). IPC increased the expression of the pro-form and the active forms of TNF-α in brains from both substrains studied, as shown by their levels 18 and 24 h after the transitory occlusion (Fig. 2a). Maximal levels were observed at 18 h, decreasing to normal values 48 h after occlusion. TLR4 deficient mice had significantly lower expression of both forms of TNF-α at all times examined when compared with mice that express TLR4 normally (Fig. 2a).
Ischemic preconditioning also resulted in a potent induction of iNOS and COX-2 in both substrain studied. Levels of both enzymes were increased at 18, 24 and 48 h after occlusion. TLR4 deficient mice had significantly lower expression of both enzymes at the times examined when compared with mice that express TLR4 normally (Fig. 2b and c).
Activation of NF-κB in TLR4-deficient mice after ischemic preconditioning
As a sign of NF-κB activation, the nuclear levels of its subunit p65 and the cytosolic levels of IκBα were determined 2, 18, 24 and 48 h after IPC. IPC caused activation of NF-κB in both substrains studied as revealed by the increased levels of the NF-κB subunit p65 in the nuclear fraction and a decreased expression of cytosolic levels of IκBα when measured 18 h after the IPC (Fig. 3a and b). NF-κB subunit p65 levels remained high at later times. In contrast, cytosolic levels of IκBα only decreased at 18 h, increasing at 24 h. These results can be explained because the inhibitory protein IκBα is a target gene of NF-κB which is rapidly induced after NF-κB activation (Le Bail et al. 1993; Sun et al. 1993). TLR4 deficient mice had significantly reduced activation of NF-κB at the time examined when compared with control mice (Fig. 3).
We hereby report that mice which express TLR4 normally (C57BL/10ScSn) present a significantly greater protection after ischemic preconditioning when compared with mice which lack expression of TLR4 (C57BL/10ScNJ). These data demonstrate that TLR4 signalling is involved in ischemic tolerance.
Indeed, we have found that ischemic preconditioning performed 24 h before a permanent MCAO affords a significantly better stroke outcome in TLR4-expressing mice, as shown by a remarkable reduction in infarct volume and a substantial recovery in neurological deficit, when compared with TLR4-lacking mice.
In the search for mechanisms involved in this effect, we found that IPC-induced TNF-α expression is higher in brains from normal mice that in those from TLR4-deficient mice. In this context, it has been described that TNF-α is protective in the setting of preconditioning (Nawashiro et al. 1997; Ginis et al. 1999, 2002); we also demonstrated that TACE/TNF-α/NF-κB pathway is implicated in both in vitro and in vivo models of ischemic preconditioning (Cárdenas et al. 2002; Hurtado et al. 2002; Romera et al. 2004; Pradillo et al. 2005). Furthermore, preconditioning with LPS, a TLR4 ligand, provides a neuroprotective tolerance effect, through a TNFα-dependent process as neutralization of TNF-α blocks LPS preconditioning (Tasaki et al. 1997; Rosenzweig et al. 2007). Recently, it has been demonstrated that preconditioning with the TLR9 ligand CG dinucleotide motif-rich DNA (CpG-DNA) oligodeoxynucleotide induces neuroprotection against ischemic injury and, more interestingly, through a mechanism that also involves a TNFα-dependent process as mice lacking TNFα are not protected (Marsh et al. 2008, 2009; Stevens et al. 2008). Therefore, the higher protection found in mice which express TLR4 normally might be explained at least in part by the increased expression of TNFα.
We also found that IPC-induced iNOS expression is significantly higher in brains from normal mice that in those from TLR4-deficient mice. TLR4 has been shown to mediate iNOS expression in macrophages (Lee et al. 2005), as well as in brain tissue after ischemia (Caso et al. 2007, 2008). More important, iNOS has been implicated in ischemic tolerance, not only after BCCAO but also after LPS administration as preconditioning stimuli, as the beneficial effects were not observed either in mice lacking iNOS or in mice treated with specific iNOS inhibitors (Cho et al. 2005; Kunz et al. 2007). Therefore, the higher protection found in mice which express TLR4 normally might be also explained by the increased expression of iNOS. Thus, TLR4-induced iNOS expression, in addition to its deleterious effects after brain ischemia, can also be beneficial by promoting ischemic tolerance.
Similarly to iNOS, TLR4 activation may trigger the expression of COX-2 in different settings (Rhee and Hwang 2000; Lee et al. 2001) including brain after ischemia (Caso et al. 2007, 2008). COX-2 induction has been also implicated in ischemic tolerance not only in in vitro models of IPC in which a selective inhibitor of COX-2 abolished IPC-induced neuroprotection, but also in in vivo models of IPC, through different signalling pathways (Gendron et al. 2005; Choi et al. 2006; Horiguchi et al. 2006; Kim et al. 2007, 2008). Our results demonstrating that IPC-induced COX-2 expression is significantly higher in brains from mice expressing TLR4 confirm the involvement of COX-2 in IT and give an additional mechanism to explain IPC-induced neuroprotection by TLR4.
Interestingly, the transcription of all these mediators can be induced through the activation of NF-κB, an important signalling mechanism downstream TLR4 activation (Rhee and Hwang 2000; Medzhitov 2001; Ginis et al. 2002; Cho et al. 2005). Indeed, we have found that IPC caused activation of NF-κB after the transitory occlusion as revealed by the nuclear translocation of the NF-κB subunit p65 at all the times studied and by a decreased expression of cytosolic levels of IκBα 18 h after IPC. Furthermore, TLR4 deficient mice had significantly reduced activation of NF-κB at the time examined when compared with control mice. The involvement of NF-κB in IT has been known for several years (for review see Barone et al. 1998; Kirino 2002; Obrenovitch 2008). Now, our present data strongly support the idea that TLR4 signalling is implicated in ischemic tolerance through the activation of NF-κB thus inducing the expression of TNF-α, iNOS and COX-2.
Another interesting aspect regards to the mechanism by which ischemic preconditioning activates TLR4. TLRs are transmembrane or endosomal receptors that recognise pathogen-associated molecular patterns thereby acting as pathogen sensors; but TLRs also respond to molecules released by damaged tissue (damage-associated molecular patterns; rev. in Medzhitov 2001). After severe ischemia, it is clear that a variety of molecules derived from dead cells or damaged vessels might act as endogenous ligands which activate TLR4. TLR4 activation is also to be expected in the setting of LPS-induced preconditioning as LPS is a TLR4 ligand. However, the putative ligand responsible of TLR activation is less obvious after IPC, a condition that is reputed not to induce damage. In this context, TLR4 could be activated by endogenous ligands such as heat-shock proteins. Indeed, it is known that one of them, HSP70, is up-regulated after ischemic preconditioning (Aoki et al. 1993; Nishi et al. 1993; Chen et al. 1996; rev in Obrenovitch 2008), suggesting that TLR4 might be activated by this or other molecules in this system. These data indicate that tolerance might be regulated not only by TLR4 signal transduction pathways but also by those mechanisms that lead to its activation.
To our knowledge, this is the first report showing a direct implication of TLR4 signalling in brain tolerance. As we have mentioned before, several pieces of evidence suggested that TLR4 might also be involved in IT but it had not been demonstrated so far (Karikóet al. 2004; Stevens and Stenzel-Poore 2006; Obrenovitch 2008). Indeed, data showing that LPS produces ischemic tolerance in several systems including brain (Tasaki et al. 1997; Dawson et al. 1999), apparently through stimulation of TLR4 as LPS is an endogenous ligand of this receptor, or that TLRs have remarkably similar signalling cascades to those implicated on ischemic preconditioning strongly supported TLR4 involvement in IT.
In addition, not only TLR4, but also other TLRs are involved in IT. It has been described that activation of TLR2 by its agonist, lipoteichoic acid, induces tolerance in heart and in a renal model of ischemic injury (Zacharowski et al. 2000; Chatterjee et al. 2002). Recently, the TLR2 specific ligand, Pam3CSK4, induces tolerance on focal cerebral ischemia/reperfusion injury in mice as demonstrated by a reduced brain infarct size and edema and by the preservation of the blood brain barrier function (Hua et al. 2008). Finally, it has been also demonstrated that TLR9 is a new target of ischemic preconditioning in the brain as the TLR9 ligand CpG oligodeoxynucleotide induces neuroprotection against ischemic injury through a TNFα-dependent mechanism and by altering the genomic response to stroke in circulating leukocytes and in the brain (Marsh et al. 2008, 2009; Stevens et al. 2008).
In summary, our data demonstrate the involvement of TLR4 in the development of ischemic tolerance after IPC in mice brain. Our findings strongly support the idea that TLR4 recruits NF-κB activation leading to the expression of TNF-α, iNOS and COX-2 to induce this neuroprotective effect. As ischemic preconditioning activates endogenous signalling pathways that culminate in marked protection against ischemic brain damage and furthermore, as several pieces of evidence support the existence of cross-tolerance between TLRs (e.g. TLR9 and TLR4), the development of the new drugs trying to stimulate TLRs and specifically TLR4 might be useful to protect the brain.
Part of this work was presented in the XVII European Stroke Conference (ESC-Nice 2008) and it was awarded the ESC Junior Investigator′s Award. This work was supported by grants from Spanish Ministry of Health RD06/0026/0005 (IL), S-BIO-0170/2006 MULTIMAG (IL); Ministry of Education and Science SAF2005-05960 (IL) and SAF2006-01753 (MAM) and Fundacio La Caixa BM05-228-2 (MAM). JMP, IGY, DF-L are recipients of fellowships funded by UCM, FIS AND FPU-MEC, respectively.