For the first time we have demonstrated that delayed postconditioning performed a few hours after reperfusion of focal ischemia provided long-term protection. The protective effect of delayed postconditioning could be achieved by occluding the ipsilateral CCA, which is clinically relevant, for the ipsilateral CCA is accessible. In addition, delayed postconditioning also improved glucose uptake, inhibited edema and mitigated BBB leakage in the penumbra, and lastly, attenuated the exacerbating effect of t-PA.
We found that the optimal parameters for delayed postconditioning, performed a few hours later, differ from that for rapid postconditioning. We and others have previously reported that the protective effects of rapid postconditioning depend on the degree of ischemia 
, the onset time of postconditioning 
, and the cycle number of occluding and releasing the blood vessels 
. Among these parameters, we found that rapid postconditioning with 10 cycles of 10 sec occlusion/10 sec release of the bilateral CCA offers the strongest protection 
. However, delayed postconditioning with 10 cycles of 10 sec occlusion/10 sec release performed at 3h did not offer protection, while 3 cycles of 30 sec/30 sec performed at the same time after stroke markedly attenuated brain injury.
We further found that the protective effect of delayed postconditioning can be achieved by 6 cycles of occlusion and release of the ipsilateral CCA, initiated as late as 6h post-stroke. MR imaging confirmed that delayed postconditioning might block infarction evolution, as infarction was developed at 24 h in the ischemic rat receiving sham postconditioning, whereas there was no clear difference in the imaging at 5 h post-stroke between both animals. Therefore, this experiment implies that delayed postconditioning was initiated at a time point when the ischemic brain tissue might still be alive. Nevertheless, there is a limitation in our study that T2 weighted sequence does not exclude infarct development post stroke. A further study using diffusion weighted imaging (DWI) is needed to clarify this issue.
The delayed postconditioning conducted using the ipsilateral CCA may have some advantages over that using the bilateral CCA, as we have shown that the bilateral CCA occlusion causes severe, though short, reduction in CBF
; postconditioning with such additional repetitive ischemia after a major stroke, if not well controlled, may further endanger the ischemic brain. Nevertheless, postconditioning conducted by occluding the ipsilateral CCA to the ischemic hemisphere only mildly reduced CBF during each occlusion (data not shown), thus minimizing the potential endangering effect of occluding the bilateral CCA during postconditioning.
The ischemic model and the method of performing postconditioning in our study are clinically important. First, the ischemic model generally mimics frequent clinical cases in which partial reperfusion occur. Most spontaneous recanalization after stroke results in partial reperfusion 
, and further, t-PA treatment leads to partial reperfusion in most stroke patients 
. In addition, patients with carotid artery stenosis after stroke often receive carotid artery endarterectomy, or angioplasty and stenting for revascularization 
. If high grade stenosis of the carotid artery is resolved after stroke, partial reperfusion will occur in the ischemic region. In our study, ischemia is generated by bilateral CCA occlusion combined with distal MCA occlusion, as described originally by Chen and colleagues 
. Two CCAs were released after 30 min of occlusion while the distal MCA remained occluded; the CCA release allowed partial reperfusion and reperfusion was not detected in the ischemic core after CCA release. Therefore, the model used in our study generally mimics partial reperfusion after stroke that frequently occurs in patients; moreover, the damaged region of the cortex is the one most likely detected in human stroke patients. Second, our ischemic model is highly reproducible and reliable 
. The reliability of an ischemic model is essential to characterize the efficacy as well as the protective mechanisms of a neuroprotectant such as postconditioning. Third, and most importantly, the methods used to generate delayed postconditioning in our study are highly clinically relevant, since the cervical carotid arteries are accessible. In fact, physicians often occlude the carotid artery during the insertion of guiding catheters 
, or for temporary balloon occlusion for the treatment of intracranial aneurysms and tumors 
. Carotid endarterectomy is actually widely used for stroke prevention; therefore, briefly occluding the carotid artery has been proven safe 
The clinical importance of our study is further strengthened by the fact that delayed postconditioning attenuated the worsening effect of t-PA. For ischemic stroke patients, reperfusion is mainly achieved by t-PA application, which is the only FDA approved pharmacological agent that dissolves blood clots for acute stroke treatment 
. However, the use of t-PA is limited by a 3 h therapeutic time window after stroke, and is complicated by its side effects of increasing symptomatic intracerebral hemorrhage 
. It has been well-established in the laboratory that t-PA is neurotoxic, and its application increases ischemic injury under certain conditions 
. Thus, t-PA therapy combined with neuroprotectants has been explored aiming at reducing its neurotoxicity 
. In our current study, we found that t-PA treatment increased infarct size, but such devastating effect was attenuated by delayed postconditioning, suggesting that delayed postconditioning may be applicable in combination with t-PA treatment for ischemic stroke patients.
The protection of delayed postconditioning may be achieved by its ability to improve metabolism after stroke. Compelling evidence in this study showed that delayed postconditioning with ipsilateral CCA occlusion improved FDG uptake as detected by PET imaging. However, it is unknown why delayed postconditioning is able to improve metabolism; nevertheless, postconditioning is performed directly on blood vessels, and it has been reported to improve endothelial function 
. In addition, we have previously shown that rapid postconditioning improves CBF recovery after reperfusion 
. Therefore, delayed postconditioning may also improve metabolisms by improving endothelial function and CBF.
The inhibiting effect of delayed postconditioning on BBB leakage may not play critical roles for its neuroprotection. Although delayed postconditioning mitigated BBB permeability at 48h, it had no effect at 24h, and BBB was open as early as 6h before the induction of postconditioning. Therefore, inhibition of BBB leakage at 48 h by delayed postconditioning may merely reflect the result of postconditioning's protection, rather than the cause by which postconditioning reduces infarction.
The molecular mechanisms of delayed postconditioning are unknown. We have shown that rapid postconditioning blunts production of reactive oxygen species or free radicals, and it inhibits apoptosis in the penumbra 
. More recently, we and others have shown that Akt activity contributes to the protection offered by rapid postconditioning 
. The protective mechanisms of rapid postconditioning are also associated with the MAPK and PKC pathways 
. Whether delayed postconditioning has protective mechanisms in common with rapid postconditioning needs further study.
Whether isoflurane protects against cerebral ischemia remains controversial 
. Some studies have demonstrated that isoflurane applied during ischemia protects the ischemic brain; its protective effects dependent on the severity of the ischemia 
. In addition, isoflurane preconditioning 
and isoflurane postconditioning 
have also been shown to reduce ischemic injury. However, many reports disagree, which have shown that isoflurane provides little protection or even worsens ischemic damage
. We did not find any protection of pre-treatment with isoflurane, either 
. Even though isoflurane was used to anesthetize the rats in our current study, it is essential to make sure that the protective effect of delayed postconditioning was not caused by isoflurane. Thus, the rats subjected to control ischemia without delayed postconditioning were treated with 1.5% isoflurane for the same period of 3 h that was received by rats with delayed postconditioning. Therefore, reduction in infarct size in our study was due to the delayed postconditioning rather than the isoflurane.
In conclusion, delayed postconditioning protected against focal ischemia in rats. It appeared to provide long-term protection and improved neurological function, and partially reversed the detrimental effect of t-PA. This novel protective model offers an alternative avenue for studying therapeutic strategies against stroke.