1.

Introduction

Stroke is the second leading cause of death and disability worldwide.¹ Despite ongoing research and the enormous medical and economic burden on society (2% to 4% of total health-care costs are attributed to stroke patients),² limited therapeutic options are available for only a small portion of stroke patients in the acute phase.³ A large number of experimental therapies have shown beneficial effects in the laboratory, but few have reached the clinic.⁴ There are only a limited number in active development such as NA-1, the first drug in a class of neuroprotectants—PSD95 inhibitors—which successfully showed a reduction of stroke damage.⁵ These shortcomings emphasize the limitations of modeling human stroke in laboratory animals, where invasive surgical manipulations are part of most procedures. Despite the advances in developing experimental stroke models, confounding factors such as anesthetics used to ensure animal welfare during the induction of stroke have yet to be overcome. Common anesthetics, such as isoflurane and barbiturates, have been found to interfere with a series of signaling pathways, including GABAergic and glutamatergic, all of which are involved in the potential treatment of stroke.^6,7 Moreover, isoflurane and other inhaled anesthetics can produce abnormal increases in cerebral blood flow.⁸

A recent study clearly shows that the presence of isoflurane anesthesia during stroke induction influences the effectiveness of neuroprotective drugs,⁹ highlighting the need for more refined methodologies when evaluating therapeutics.

Given the recent development of elegant in vivo preparations that provide optical access to the cortical surface in awake, head-fixed mice,^10,11 we sought to validate a method for inducing targeted photothrombosis through the intact skull in awake mice. Taking advantage of the semitransparent nature of the mouse skull, transcranial windows are reliable preparations for longitudinal imaging and optical manipulation, providing a large, bihemispheric access to the cortex and reduced invasiveness.¹⁰ We recently described a chronic transcranial window preparation, previously used for imaging of both awake and anesthetized mice, which provides sufficient physical stability over a large field of view and can be used for longitudinal mesoscopic mapping of cortical function in awake mice.^11,12 In this study, we make use of the optical access offered by this preparation and evaluate a protocol to optically induce focal ischemia to target-specific cortical areas¹³ using photothrombosis.¹⁴ Our protocol does not require any use of anesthetics either during head fixation or stroke induction. In addition, our approach combines laser speckle contrast imaging¹⁵ and wide-field calcium imaging to reveal brain dynamics and spreading peri-infarct depolarizations after stroke over a spatial scale encompassing both hemispheres, indicating that our preparation will facilitate functional cortical mesoscopic mapping during stroke and recovery.

2.

Material and Methods

All procedures were approved by the University of British Columbia Animal Care Committee and conformed to the Canadian Council on Animal Care and Use guidelines and reported according to the animal research: reporting of in vivo experiments guidelines.

GCaMP3 mice were used to investigate spreading waves of activity after stroke. Data from negative GCaMP6 and ChR2 mice were pooled together for the laser speckle imaging, behavioral assessment, and histological quantification. Transgenic GCaMP3 mice of either sex ($n=16$), 4 to 5 months of age were produced by crossing homozygous B6.129S2-Emx1tm1 (cre)Krj/J strain (Jax No. 005628) with hemizygous B6; 129S-Gt(ROSA)26Sortm38(CAG-GCaMP3)Hze/J strain (Jax No. 014538, Ai38).¹² Negative GCaMP6 and ChR2 mice were used for speckle imaging and behavior assays. The absence of one allele yielded the absence of expression of the light sensitive indicator, opsin. Transgenic negative GCaMP6 mice of either sex ($n=12$), 4 to 5 months of age were produced by crossing Emx1-cre B6.129S2-Emx1tm1(cre)Krj/J, (Jax No. 005628), CaMK2-tTA B6.Cg-Tg (Camk2a-tTA) 1Mmay/DboJ, (Jax No. 007004) and either TITL-GCaMP6s Ai94; B6.Cg-Igs7tm94.1(tetO-GCaMP6s)Hze/J, (Jax No. 024104), or TITL-GCaMP6f Ai93; B6.Cg Igs7tm93.1(tetO-GCaMP6f)Hze/J, (Jax No. 024103) strains.¹⁶ PCR confirmed the absence of either TITL-GCaMP6, CaMK2-tTA, or Emx1-cre sequence.

ChR2 mice of either sex ($n=12$), 4 to 5 months of age from the Jackson Laboratory, line 18, stock 007612, strain B6.Cg-Tg (Thy1-COP4/EYFP) 18Gfng/J were used as well.

2.3.

Setup and Procedure to Induce Stroke

Mice were head-fixed using a slightly modified version of an apparatus previously described for imaging awake mice.¹¹ A locking ball and socket mount (Thorlabs, Product: TRB1) was mounted on the platform where the stroke was induced using two optical posts and a 90-deg turn. The head-fixing screw (4/40 threads) was attached to the ball and socket mount (8/32 threads) using a thread adapter (AP84E, Thorlabs). A clear Plexiglas tube (28-mm diameter) was attached on the same platform to provide a chamber for the mouse. The setscrew was fitted with a nut and threaded into the ER025 post by rotating the ball within the socket. The nut was subsequently tightened upward against the ER205 post to immobilize the head without applying pressure to the cranial window. After a period of habituation, when mice are introduced to the apparatus and head-fixed daily for a few minutes, mice can be gently persuaded into the tube. Once habituated, mice show minimal signs of struggling and can show normal behaviors, such as whisking or grooming.

A target area between sensory and motor cortex—stereotactic coordinates (1.5; 0.5) mm from bregma—was previously identified based on response during intrinsic optical signal imaging.¹⁸

To induce photothrombotic occlusion,¹⁴ we used a custom-made compact laser system¹³ (Fig. 1) consisting of a 40-mW diode pump solid-state 532-nm laser MGM-20 (Beta Electronics, Columbus, Ohio) attenuated to 11 mW at the window surface through a polarizer. Beam diameter is 1.2 mm and it was measured at full width at half max amplitude. The system includes a fan and heat sink as well as a transistor-transistor logic-triggered power supply. While wearing safety goggles, the laser was turned on and targeted to the preselected area within the cranial window using bregma as a landmark. Once the laser was correctly targeted, the laser was turned off, and all the clamps were tightened to prevent movement. Previous reports showed that this amount and duration of the green laser irradiation did not produce tissue damage or impact the targeted vessels in any way unless combined with Rose Bengal (RB) administration.^20,21

Fig. 1

Hardware for awake head-fixation and stroke induction. (a) A set screw was surgically implanted behind the cranial window to stably head fix the mouse. (b) Blow up of the chronic window preparation where surface vessels are visible through the intact skull, scale bar of 1 mm. (c) Mouse brain atlas image (from the Allen Institute Atlas)¹⁹ showing cortical areas that may be targeted with this method. M1 and M2, primary and secondary motor areas, respectively; BC, barrel field; FL, forelimb; and HL, hindlimb regions of the primary somatosensory area, respectively (S1); S2, supplemental or secondary somatosensory area; V1, primary visual area; AU, primary auditory area; and RS, retrosplenial area. (d) Mice are persuaded into a chamber mounted on the system for the stroke induction, and head-fixed while awake.

Using an insulin syringe, mice were injected intraperitoneally (IP) (0.1 cc per 10 g of body weight) with a photosensitive dye solution—RB (R3877-5G, Sigma-Aldrich, St. Louis, Missouri). The solution was freshly prepared by dissolving 0.01 g of RB in 1 ml of saline solution. The mix was vortexed (Fischer Vortex, Fischer Scientific, Hampton, New Hampshire) and sonicated (Branson, Danbury, Connecticut). Exposure to light was avoided, and the solution was filtered with a $0.2\text{-}\mu \mathrm{m}$ filter (PALL Life Science, PN4612). RB was injected IP by gently holding the tail of the mouse and lifting it up to get a clear view of the abdomen. After the injection, we allow 2 min for the dye to reach the brain and then the laser illumination was turned on for 13 min at the targeted area to induce focal ischemia (Fig. 2). During stroke induction, the room lights were dimmed to increase comfort and reduce stimulation for the mouse. Sham surgery controls were treated in the same way with the omission of the laser illumination. In this case, the surgery did not produce any sign of damage. Due to the minimally invasive nature of the procedure, awake mice were imaged immediately after surgery.

Fig. 2

Head fixation and stroke induction. Example of head fixation procedure of an awake mouse following that an area chosen on stereotactic coordinates is targeted for stroke induction. After an IP injection of RB (not shown), the laser is turned on for 13 min (Video 1, MOV, 473 KB [URL: http://dx.doi.org/10.1117/1.NPH.4.3.035001.1]).

3.

Results

Cranial windows were implanted during a surgical procedure at least 7 days prior to stroke induction as in Ref. 11. The procedure was simple to perform, requiring less than 30 min per mouse, which minimizes the exposure of the mouse to anesthesia. It resulted in a stable cranial window [Fig. 1(b)] with a large field of view of the cortex [Fig. 1(c)] with little disturbance to mouse behavior (grooming, feeding, and climbing). During habituation to head-fixing sessions, the setscrew kept the cranial window and entire head of the mouse firmly in place regardless of body movement [Figs. 1(a) and 1(d)].¹¹

This transcranial window preparation permitted the tracking of cerebral blood flow changes using laser speckle imaging [Fig. 3(a)].¹⁵ Prior to stroke induction, baseline measurements were recorded clearly and without apparent optical distortion from the window. The head-fixing setscrew of the cranial window allowed us to perform the photothrombotic stroke in awake mice without major disturbances to the mouse or further need of invasive manipulation [Fig. 1(d)]. Laser exposure in mice injected with RB resulted in the formation of thrombus within 13 min similar to the time required using other existing window preparations in anesthetized animals.²⁸ Mice remained awake during the procedure and did not exhibit signs of pain or discomfort, such as audible vocalizations or writhing. Following the procedure, mice are monitored for the first few hours and daily thereafter; mice do not show overt signs of distress, such as hunching over, shaking, or aggression. Given the lack of aggressive behavior, we continue to group house mice after the awake stroke procedure.

Fig. 3

Laser speckle imaging for changes in blood flow following ischemic stroke. (a) Drawing of the laser speckle imaging setup where the head-fixed mouse is placed while fully awake. The imaging is performed in a darkened room without visual, tactile, or auditory stimuli. (b) Representative pictures of laser speckle contrast imaging of cortical blood flow at baseline and at 5 min, 1, 3, 5, and 7 days following sham or stroke surgery. Dotted lines (drawn by hand) are used to highlight the infarcted area. Scale bar of 1 mm. (c) Quantification of the changes in cortical blood flow expressed in percentage of baseline in mice that underwent sham (white symbols $n=10$) and stroke surgery (black symbols $n=12$). Pairwise two-way ANOVA with the Bonferroni correction for posthoc comparisons. $\text{Mean}\pm \mathrm{S}\mathrm{D}$.

On day 1 (24 h after stroke), blood flow within the targeted area was examined using laser speckle imaging [Figs. 3(a)–3(c)]. Blood flow was significantly lower in stroke mice within the targeted area (ROI included the area between sensory and motor cortex, where the stroke was induced) ($P<0.0001$). The results from the two-way ANOVA on the complete time course, i.e., until day 7, corroborate the significance of the effect of stroke performed in awake mice on local blood flow [Fig. 3(c); $P<0.0001$, $F_{(\mathrm{1,80})}=46.89$]. Stroke was induced in the area between sensory and motor cortex (stereotactic coordinates 1.5-mm lateral and 0.5-mm anterior of bregma) in awake and anesthetized mice, and functional investigation was performed in parallel. Using wide-field calcium imaging within 5 min from the stroke induction [Figs. 4(a)–4(h)], ischemic depolarizing waves were observed in 6 of the 10 awake mice and 2 of the 4 anesthetized mice (chi-squared test, $P=0.732$) (Fig. 5). Observed waves, one for each mouse during the imaging session, had a significantly smaller magnitude of calcium change from resting baseline in awake mice, (maximum $\mathrm{\Delta }F/F$: $97\%\pm 13\%$ versus $223\%\pm 22\%$ for awake and anesthetized mice, respectively, $P=0.002$). Median propagation speed ($5.16\pm 0.48$ versus $3.96\pm 0.84\mathrm{mm}/\min$, $P=0.269$) was not significantly different between awake and anesthetized mice, and no relationship was observed between speed and amplitude [Pearson’s correlation: $r=-0.29$, $P=0.48$, Fig. 4(i)]. Propagation speed was in the same range of previous findings (3 to $5\mathrm{mm}/\min$), where depolarizing waves were investigated in anesthetized mice and rats using electrophysiological recordings and optical intrinsic signal imaging.^29,30 No propagating wave of that order of magnitude was observed in the opposite hemisphere or in mice that underwent the sham procedure (not shown). Mice were subsequently returned to their home cages.

Fig. 4

Calcium signaling imaging in awake and anesthetized mice following ischemic stroke. (a) Montage of the raw fluorescence recorded at 7-s interval in an awake mouse. (b) Fluorescence recorded in four different points indicated in (c). (c) Estimation of the relative time to peak of the propagation wave (color) normalized by the amplitude ($\mathrm{\Delta }F/Fo$, luminance). (d) Montage of the raw fluorescence recorded at 6-s interval in an anesthetized mouse. (e) Fluorescence recorded in four different points indicated in (f). (f) Estimation of the relative time to peak of the propagation wave (color) normalized by the amplitude ($\mathrm{\Delta }F/Fo$, luminance). Black pixels represent those saturated at any point during the recording. (g) Distribution of the wave speeds and amplitude as well as (h) relationship between them (scatter plot) in awake and anesthetized mice. Chi-squared test was used to assess differences between the two groups.

Fig. 5

SD in targeted stroke. Calcium signaling imaging in anesthetized and awake mice following ischemic stroke shows spreading depolarizing waves a few minutes after stroke induction. Speed $45\times$ (Video 2, MP4, 10,105 KB [URL: http://dx.doi.org/10.1117/1.NPH.4.3.035001.2]).

Both stroke and sham mice received an NDS ²⁷ of 0 during baseline assessment. Stroke mice received a score significantly different from baseline ($P<0.0001$) and the sham mice ($P<0.0001$, $F_{(\mathrm{4,88})}=31.62$), which never received an NDS other than 0 [Fig. 6(a)].

Fig. 6

Neurological deficit score and histological quantification following ischemic stroke. (a) NDS assessed in sham (white symbols $n=10$) and stroke (black symbols $n=14$) mice. (b) Representative pictures showing ROIs (cortex) of coronal brain sections stained with FJC for ischemic cell death. Scale bar of 1 mm. (c) Quantitative analysis of infarct volume 7 days after sham ($n=11$) or stroke ($n=12$) surgery. Correlation at day 1 between: (d) lesion volume and changes in blood, (e) lesion volume and NDS, (f) changes in blood flow and NDS one day following stroke induction. (g) Time courses of correlation between lesion volume and changes in blood flow, NDS and lesion volume, and NDS and changes in blood flow. Results from NDSs were evaluated with the Scheirer–Ray–Hare extension of the Kruskal–Wallis test. Infarct volume between groups was evaluated using the Mann–Whitney test, and data are shown as $\text{median}\pm 75/25$ percentile. Pearson’s $r$ correlation coefficient was used to assess correlations.

On day 7, mice were euthanized, and the infarct volume was quantified. The infarct area was outlined with FJC staining [Fig. 6(b)]. This type of damaged tissue was visible in the brain slices from all mice that received the stroke procedure, resulting in a median lesion volume of $0.9\pm 0.811{\mathrm{mm}}^3$, with a minimum volume of $0.2{\mathrm{mm}}^3$ and a maximum of $2.8{\mathrm{mm}}^3$, in contrast to mice that received the sham procedure, which had a lesion volume of $0{\mathrm{mm}}^3$, a statistically significant difference between procedures [$P=0.0005$; Fig. 6(c)].

Lesion volume measured at day 7 was inversely correlated with changes in blood flow on each day after stroke induction, confirming the association between the decrease in blood flow and the infarcted area [Figs. 6(d) and 6(g)]. Lesion volume was positively correlated with NDS on days 1 ($r=0.685$, $P=0.014$) [Fig. 6(e)] and 3 ($r=0.7272$, $P=0.0074$), but not at day 5 and 7 [Fig. 6(g)]. A negative correlation was also found between NDS and changes in blood flow on days 1 ($r=-0.6158$, $P=0.033$) [Fig. 6(f)] and 3 ($r=-0.5763$, $P=0.0499$) [Fig. 6(g)].

4.

Discussion

We developed a method for stroke induction in awake, head-fixed mice avoiding potential confounds from anesthesia. This method is combined with longitudinal mesoscopic mapping of cortical function through the optical access facilitated by a transcranial window preparation.

To induce focal ischemic stroke in awake mice, we used the photothrombotic model initially proposed by Rosenblum and El-Sabban,³¹ and improved upon by Watson et al.¹⁴ The method is based on the light-dependent production of reactive oxygen species and triggering of clotting cascades after an injected photosensitive dye, such as RB, is irradiated.¹⁴

In a limited number of previous studies, invasive approaches such as the implantation of brain cannulas to deliver endothelin³² were used in awake animals to induce stroke. Only recently, Seto et al.⁹ developed a protocol for inducing focal ischemic stroke in awake mice inducing photothrombosis in a way that significantly reduces invasive manipulation. In this protocol, laser light needed for photothrombosis was focused on a plastic optical fiber, which requires the installation of a headcap on the skull of the mouse. Despite the procedure being less invasive than previously described, the fiber stroke protocol still required 15 to 20 min of prior exposure to anesthesia to thin the skull and attach the headcap.⁹ The thinning skull procedure itself may be a source of variability between mice and can mechanically disrupt the underlying structures with consequent microglia and astrocyte activation.³³ A nonuniform skull thickness may cause spherical aberrations and distortion of fluorescent structures located deep in the cortex.³⁴ Moreover, mice were tethered to the optical fiber for the time of occlusion $\sim 60\min$, which may limit natural behaviors of mice or induce a stress response.⁹ The placement of a fiber cannula also obscures the measurement of functional signals such as blood flow that can be important to define the extent of lesions for later recovery enhancing treatments. With our approach using a head-fixing screw¹¹ (that facilitates the immobilization of mice while preserving optical access), we eliminate anesthesia at stroke induction and assessment. Anesthesia has long lasting effects on brain function and it can reduce the extent of ischemic damage and mask the beneficial outcome of a potential therapy.⁹ Another advantage of a protocol that does not require anesthesia is that it allows chronic daily imaging, which would otherwise increase the exposure to confounding factors in the form of repeated anesthetic administration for either stroke induction or imaging assessments. There is evidence that the effects of anesthetics may have prolonged effects in mice.³⁵ Moreover, most anesthetics depress myocardial contractility, arterial blood pressure, and are often respiratory depressants.^36,37 Daily exposure to anesthetics alters systemic physiology, and its effects outlast the duration of anesthesia. In our protocol, stroke induction is a relatively less invasive procedure and can be easily targeted to a specific cortical region, such as the FL representation as identified during sensory mapping or estimated for stereotaxic coordinates. This is an advantage for the study of specific cortical regions targeted by stroke, as well as regions that are functionally connected to the damaged area and the peri-infarct cortex.³⁸ However, the lesion target was defined based on stereotactic coordinates, without doing any discrimination based on type or dimensions of surface vessels. Variability in vessel distribution of each mouse could explain a range of lesion volumes,²¹ although the photoactivated area is large $\sim 1\mathrm{mm}$ in diameter and should contain all classes of surface vessels. Future refinements to the procedure could target individual identified arterioles for photothrombosis^39,40 or employ magnetic nanoparticles for higher precision.⁴¹ Notably, our analysis suggests that changes in blood flow are a good predictor for variability in lesion volume providing a method to potentially stratify the animals in future work where lower variability within groups is desired.

Existing methods for brain imaging in awake head restraint animals employ strategies to reduce the stress response during imaging. An example of this approach is the use of an air-supported spherical treadmill to allow walking and running.⁴² However, our system uses a relatively simpler approach to both induce stroke and image the brain after the mouse is habituated to the head fixation.

In this context, our preparation uses a chronic window that allows us to perform bihemispheric wide-field imaging in awake mice immediately after receiving stroke. This preparation is quick to perform with little or no risk of damage or inflammation that can easily occur with more invasive techniques such as bone thinning.³³ More relevant to our purposes is the investigation of events that happen over a larger spatial scale after stroke. The wide field of view of the window facilitates the investigation of multiple ROIs and correlation with functional maps. We have shown that with our combined approach we could reveal brain dynamics after stroke at a mesoscopic level. Using wide-field calcium imaging, we could observe ischemic depolarizing waves in awake mice a few minutes after stroke induction. These depolarizing waves have been previously imaged in anesthetized mice;^23,43 however, our data are the first to show that ischemic waves propagate across the cortex in awake mice. The effects of anesthetics on ischemic wave propagation can involve a variety of mechanisms, such as reduced energy depletion as a result of suppression of electrical activity, interference with pathophysiologic processes such as disturbance of ion homeostasis, or accumulation of excitotoxins during ischemia and postischemic free radical formation. All of which could potentially interfere with the wave’s propagation. Among those factors, intracellular communication through gap junctions should be taken into account considering that commonly used anesthetics such as halothane are gap junction blockers.⁴⁴ Furthermore, the ability to visualize mesoscopic cortical activity may prove valuable in understanding critical neurovascular processes at the acute stage after stroke. One limitation of our approach is the inability to perform GCaMP functional imaging at early time points following stroke induction within the peri-infarct cortex. This region becomes excessively bright due to apparently high intracellular calcium masking any smaller functional signal. Moreover, the low number of mice used to compare wave propagation in anesthetized and awake mice does not provide an exhaustive comparison and should be considered preliminary. The aim of our study was to develop a protocol that dispenses with the used anesthetics and allows functional investigations in short time.

Low mortality, the ability to target the infarct within different functional subdivisions of the cortex, and the minimal surgical intervention required are all advantages of this photothrombotic stroke model. The relatively small size of the infarct allows this model to more faithfully mimic the usual cases of human stroke, where most strokes are small in size and associated with some degree of recovery.⁴⁵ Stroke models that require suture occlusion of the middle cerebral artery frequently produce tissue damage in a substantial portion of the ipsilateral hemisphere.⁴⁶ Moreover, the reproducibility of infarction is affected by invasive procedural factors such as suture diameter that have been shown to correlate with the infarct volume.⁴⁷ We suggest that parallel monitoring of regional blood flow may be used to further sort animal groups based on the degree and location of ischemia. It is particularly important to control these metrics in studies seeking to promote circuit level plasticity and recovery. However, it must be considered that this is a regional-relative measure that may not reflect the full loss of blood within the infarcted core.

A stroke model that does not require anesthesia permits both the assessment and alteration of activity at early time points that could potentially influence stroke outcome. In the acute phase of stroke, particular forms of neuronal activity may modulate vascular flow leading to neuronal survival and improved outcome.^48,49 By contrast, other forms of activity may increase metabolic load and exacerbate damage through peri-infarct depolarizations.^23,43,50 The role of activity in the acute stroke setting (first hours to weeks) is critical for not only saving tissue but also defining the time window where activity-based interventions such as rehabilitation can be safely initiated. Recent work in rehabilitation has shown that earlier mobilization after stroke leads to poor outcome.⁵¹ Integrating basic and clinical research results is also necessary to develop knowledge that can lead to the development of translational rehabilitation aids and implement brain–machine interfaces or closed-loop stimulation to restore function in patients.^52,53 We suggest that chronic stroke models that lack the confounding effects of anesthesia offer the ideal scenario to manipulate activity at short-time windows following stroke induction or to best stratify animals into groups with similar lesions where later manipulations can be tested with lower variance.