Anesthetics are commonly used in humans undergoing neurosurgery and experimental animals subjected to various kinds of ischemic stroke. Over the years there has been a growing appreciation that anesthetics have widespread effects on neuronal and vascular function. For example, inhaled anesthetics such as isoflurane can augment GABA, glutamate and nicotinic neurotransmission, mitochondrial function, apoptosis, inflammation and cortical blood flow (Eger, 1981; McPherson et al., 1994; Flood et al., 1997; De Sousa et al., 2000; Yamakura and Harris, 2000; Wu et al., 2012; Zhang et al., 2012; Hofacer et al., 2013; Kotani and Akaike, 2013; Altay et al., 2014). This is a potential concern for pre-clinical studies of cerebral neuroprotection since many potential treatments target the same signaling pathways as those affected by anesthesia. The effects of anesthesia on brain function can last for hours, even after the animal regains consciousness (Kapinya et al., 2002b; Saab et al., 2010), which raises the possibility that anesthesia could affect the outcome of a neuroprotection study.

In order to address this question, one needs to compare potential therapies in a model of stroke that can be induced without anesthesia. There are a few published studies that describe procedures for inducing stroke in awake animals using intracortical endothelin injection or intra-arterial infusion of microspheres (Demura et al., 1993; Gelb et al., 2002). While useful, these methods require invasive procedures such as the implantation of a brain cannula which can augment cerebral blood flow and induce inflammation. In the present study we developed a protocol for inducing focal ischemic stroke in fully awake mice using the photothrombotic method (Watson et al., 1985). An advantage of this approach is that ischemic stroke can be targeted to a specific brain region or artery (Nishimura et al., 2006; Schaffer et al., 2006; Sigler et al., 2008) without extensive surgical training or invasive procedures. Here we show that not only can isoflurane anesthesia reduce the extent of ischemic damage after stroke, but it can mask the protective effects of a stroke therapy.

In our first set of experiments, we used the photothrombotic method to regionally occlude cortical surface vessels over the primary forelimb somatosensory cortex (Figure 1A). Mice were randomly assigned to receive either sham stroke (laser exposure without dye injection or vice versa) or stroke while awake (n = 7) or anesthetized with 1.5% isoflurane for 20 (n = 9) or 45 min. Consistent with previous studies (Brown et al., 2009; Grutzendler et al., 2011), thinning the skull or exposure to the laser in sham operates did not induce reactive gliosis or cell death near the cortical surface (n = 3, Figure 1B). As shown in surface angiograms (Figure 1C) and Fluoro-Jade C stained brain sections (Figure 1D), mice that were awake during stroke had a significantly larger infarction (Figure 1E) than those anesthetized for 20 [t₍₇₎ = 2.04, p < 0.05] or 45 min [t₍₉₎ = 2.72, p < 0.05]. Control experiments imaging the spread of light from the fiber optic tip ruled out the possibility that larger infarcts in awake ambulatory mice was caused by subtle movement of the fiber optic attachment and hence a greater area of photoactivation. Furthermore by measuring Rose Bengal fluorescence in the blood plasma 20 min after injection, we determined that anesthesia did not significantly affect its absorption into the bloodstream [t₍₁₅₎ = 0.31, p = 0.38]. These results demonstrate that even brief exposure to isoflurane during stroke induction reduces the extent of ischemic damage.

In humans, focal ischemic stroke typically involves a thrombo-embolic occlusion of a specific segment of the MCA rather than simultaneous generation of thrombotic clots in many penetrating arterioles and micro-vessels over a cortical region (Carmichael, 2005; Jackman et al., 2011). Therefore to create a more realistic model of ischemic stroke that could be used for neuroprotection studies, we restricted photoactivation to a single segment of the distal MCA by first thinning a 0.4 mm diameter region of skull directly over the MCA and then masking adjacent areas of the skull (Figure 2A). Consistent with previous studies that have varied occlusion duration (Morancho et al., 2012), our pilot experiments indicated that 60 min of photoactivation, but not 15–30 min (n = 6 mice) was required to reliably induce a permanent MCA occlusion. Longer periods of photoactivation were made possible by the fact that dye circulated in the blood for at least 45 min after injection (Figure 2B). Laser speckle imaging confirmed that blood flow was progressively reduced in branches downstream of the MCA after the start of photoactivation (Figure 2C). In order to illustrate the extent of ischemic cortical tissue in the right hemisphere, Evans blue dye was injected 24 h after stroke (Figure 2D). Dye injection was useful for verifying the presence of a permanent occlusion which was evident by the absence of circulating dye near the site of MCA photoactivation. Our success rate for inducing targeted MCA stroke in awake vs. anesthetized animals was 91.5 and 80.0% respectively (χ2 = 1.18, p > 0.05). The mortality rate was 5.6% for awake mice and 2.2% for anesthetized mice (χ2 = 4.68, p < 0.05). It should be noted that mice subjected to stroke while awake did not show any overt signs of pain or distress such as vocalizations, head scratching or aggressiveness. Interestingly, all mice with a confirmed permanent MCA occlusion (verified post-mortem) showed a reduction in ambulatory behavior at the time of photoactivation. This behavioral pattern was useful to provide real time information about the success of the stroke procedure.

Since distal MCA occlusion is often used in neuroprotection studies, we asked whether the efficacy of a stroke therapy depends on whether the stroke was induced in an awake or anesthetized animal. Accordingly, mice that were awake or anesthetized with isoflurane (1.5% in air) for 20 or 60 min were given either a single intraperitoneal injection of a GluN2B receptor antagonist Ro 25-6981 (6 mg/kg), an α4β2 nicotinic acetylcholine receptor (nAChR) antagonist DHβE (3 mg/kg) or vehicle 90 min after the start of the MCA occlusion (Figure 3A). We reasoned that since ischemia stimulates glutamate and acetylcholine release (Dirnagl et al., 1999; Kiewert et al., 2010), and that isoflurane can alter glutamate receptor signaling and block α4β2 nAChRs (Flood et al., 1997; Yamakura and Harris, 2000), it is possible that the neuroprotective effects of these drugs could have been missed in previous studies. Consistent with our first set of experiments, mice that were awake during stroke induction and treated with vehicle had significantly larger infarcts than vehicle treated mice anesthetized for 20 [t₍₁₇₎ = 3.65, p = 0.001] or 60 min [t₍₁₉₎ = 2.52, p = 0.01; Figures 3B,C see Table 1 for infarct details]. The extent of ischemic damage did not expand beyond the first 24 h, given that there was no difference between infarct volumes measured at 24 vs. 72 h in anesthetized mice [1 day = 13.4 ± 1.1 mm³ vs. 3 day infarct volume = 12.8 ± 1.2 mm³; t₍₁₉₎ = 0.31, p = 0.38]. With respect to stroke treatments, none had any impact on infarct volume in mice that were anesthetized for either 20 [F_{(2, 25)} = 1.01, p = 0.38] or 60 min [F_{(2, 26)} = 0.31, p = 0.73] during the induction of stroke (Figure 3B). The absence of a neuroprotective effect with Ro 25-6981 treatment (in anesthetized mice) is consistent with previous studies (Liu et al., 2007; Sun et al., 2008). However there was a significant effect of drug treatment if the stroke was induced while awake [F_{(2, 26)} = 3.82, p < 0.05]. Specifically, DHβE (but not Ro 25-6981) treatment 90 min after stroke was highly effective in reducing the extent of ischemic damage relative to awake mice treated with saline [t₍₁₇₎ = 2.83, p < 0.01; Figures 3B,C]. There was a trend toward smaller infarcts in an additional group of awake mice treated with DHβE 180 min after occlusion [t₍₂₀₎ = 1.66, p = 0.05; Figure 3B]. These data show that just 20 min of exposure to isoflurane during stroke induction can mask the benefits of a neuroprotection therapy.

In order to determine whether the neuroprotective effects of DHβE in awake mice could be explained by increased blood flow in the cortical penumbra, we transcranially recorded blood flow in awake mice using laser doppler flowmetry (Figures 4A,B). Our analysis indicated that blood flow in the cortical penumbra was reduced by 44 ± 17% in vehicle and 50 ± 10% in DHβE treated mice 90 min after initiating photothrombosis (Figures 4B,C). Treatment with DHβE did not significantly alter blood flow in the penumbra relative to pre-injection levels [Figure 4C, t₍₈₎ = 0.10, p = 0.46]. Furthermore, there were no significant differences between saline and DHβE treated mice in heart rate, respiration, blood gas/electrolyte concentrations or blood pH (Table 2; n = 3 per group, p > 0.05 for all comparisons). Therefore the neuroprotective effects of DHβE in awake mice are not explained by differences in penumbral blood flow or global changes in several cardiovascular parameters.

The goal of the present study was to address two key questions: (1) does isoflurane anesthesia affect the extent of ischemic stroke damage and (2) does the presence of isoflurane anesthesia during stroke induction influence how well a neuroprotective therapy appears to work? In order to investigate the first question, we needed to develop a method for inducing stroke that eliminated the need for anesthetics. Photothrombosis was a logical choice for this given that the photosensitive dye could be administered through an intraperioneal injection, while excitation light could be delivered in a relatively non-invasive manner through a fiber optic assembly. Our stroke protocol provides certain advantages over other commonly used experimental models of focal ischemia (Longa et al., 1989). Most obvious is the elimination of anesthesia which necessarily complicates the study of how stroke or a neuroprotective treatment affects brain function, circulation, and viability. Using un-anaesthetized animals also opens up the possibility for examining how certain behavioral states (sleep-wakefulness, satiety, exercise) can affect ischemic damage. Another advantage is that no craniectomy is required (Sugimori et al., 2004). This reduces the likelihood of craniectomy induced hemorrhage and infection which can occur using endothelin, thermal or mechanical occlusion of the MCA. More practical advantages include the speed and ease with which the stroke can be performed. In our lab, 3 different experimenters were able to become proficient in this method of stroke in just a few practice attempts. It should also be noted that the success rate for inducing stroke was quite high, particularly in awake mice (~92%). Moreover, the co-efficient of variation for infarct volumes in awake mice was 20.5%, suggesting that ischemic damage was highly reproducible. Given some of the advantages outlined above and the fact that targeted photothrombotic occlusion of the MCA induces a penumbral region of reduced cortical blood flow (see Figures 4B,C) (Yao et al., 2003), the present model could be very useful for future neuroprotection studies.

As is the case with any model of stroke, there are limitations and potential caveats with our approach. For one, our stroke protocol still requires 15–20 min of exposure to anesthesia to thin the skull and attach the headcap. Since 3 h of isoflurane exposure can affect stroke damage when induced 24 h later (Kapinya et al., 2002a), there is a possibility that this exposure could affect our results. However, this scenario seems unlikely since all stroke groups (awake and anesthetized) were briefly exposed to isoflurane and we were still able to detect a robust difference in infarct volumes between awake and anesthetized mice. Second the mice need to be tethered to the light weight fiber optic for 60 min to ensure a permanent occlusion of the MCA. Therefore it is possible that the fiber optic attachment might induce a stress response or suppress some natural behaviors. While a potential concern, we noticed in mice with a sham stroke or those few instances where we were unsuccessful inducing the stroke, mice showed normal exploratory and grooming behaviors during the 60 min photoactivation period. It should also be noted that no signs of cell death were observed in these mice, therefore any heat associated with the laser was not sufficient to induce damage which is consistent with previous studies using photothrombosis (Sugimori et al., 2004; Sweetnam et al., 2012). Future iterations of the present model could incorporate a 568 nm laser beam (Sugimori et al., 2004) which more efficiently excites Rose Bengal dye thereby reducing the time needed for photoactivation. Another potential drawback is that photo-activation of Rose Bengal dye generates singlet oxygen which directly damages the vascular endothelium. This can lead to a rapidly progressing infarct and vasogenic edema (Watson et al., 1985; Witte, 2010). In humans these events are usually delayed a few hours, therefore how closely photochemically induced ischemic damage models the human condition is unclear (Witte, 2010). Furthermore, since photothrombosis leads to a permanent irreversible occlusion, the ability for many neuroprotectant drugs to promote re-perfusion into the ischemic core may be limited.

Isoflurane was chosen in the present study because of its widespread use in animal studies of stroke. However, isoflurane and other inhaled anesthetics are well known to have multiple effects on brain function. For example, isoflurane dose dependently enhances GABAergic inhibitory post-synaptic currents, while inhibiting glutamate release and AMPA mediated receptor currents (Flood et al., 1997; Larsen and Langmoen, 1998; De Sousa et al., 2000; Saab et al., 2010; Kotani and Akaike, 2013). Isoflurane has also been shown to dilate cerebral vessels and increase cortical blood flow (Eger, 1981; McPherson et al., 1994; Iida et al., 1998). Given these diverse effects, it is plausible that isoflurane could have intrinsic neuroprotective properties. However, there is some debate in the literature on this issue. For example, Nehls et al. (1987) reported that monkeys anesthetized with isoflurane were more likely to suffer a stroke and resultant hemiplegia than those treated with barbiturates. Contrasting with these results, a retrospective study of patients undergoing cardiac surgery revealed that those anesthetized with isoflurane had a reduced risk of ischemic damage compared to those exposed to enflurane or halothane (Michenfelder et al., 1987). Experimental models of cerebral ischemia have found more consistent results in that inhaled anesthetics such as isoflurane and sevoflurane attenuate stroke damage and reduce the severity of neurological deficits (Warner et al., 1993; Sakai et al., 2007; Traystman, 2010; Zhou et al., 2010). One potential limitation of past studies was that the effects of isoflurane were not compared to an animal that was completely free of anesthetic on the day stroke was induced. Using both regional and targeted occlusion of cerebral vessels in awake and anesthetized mice, our data support the hypothesis that isoflurane can protect the brain from ischemic damage. While we cannot extrapolate our findings to other anesthetics, future studies utilizing the awake stroke model could help with this determination.

With respect to our second objective, our data clearly indicate that the efficacy of a drug therapy is in some cases dependent on whether the stroke was induced with or without a general anesthetic. To our knowledge, this is the first study to show such an effect. Moreover, no previous study has shown that blocking α4β2 nicotinic receptors in vivo can protect the brain from ischemic damage. Given that isoflurane is a potent inhibitor of α4β2 nAChRs (Yamakura and Harris, 2000), and that most preclinical studies induce stroke with an anesthetic (often using isoflurane), it is possible that the neuroprotective effects of an α4β2 nAChR antagonist could have been missed in past studies. Indeed, we too did not find any beneficial effects of DHβE if the treatment was initiated in mice exposed to isoflurane anesthesia during stroke induction. The mechanism through which the α4β2 nAChR antagonist confers neuroprotection in awake mice remains speculative but it was not a consequence of increased collateral blood flow or changes in physiological variables such as heart rate, respiration, blood gas or electrolyte concentrations. Furthermore, a recent study demonstrated that systemic injection of DHβE (at 3 mg/kg) in adult C57BL6 mice has minimal effect on core body temperature, thus ruling out hypothermia as a cause (Rezvani et al., 2012). While beyond the scope of the present study, future experiments directed at understanding how α4β2 nAChRs modulate glutamate release/excitotoxicity (Lambe et al., 2003), spreading depression (Sheardown, 1997; Douglas et al., 2011), blood brain barrier permeability (Hawkins et al., 2005) or inflammatory responses to ischemia should help resolve this issue.