Astrocyte-neuron lactate shuttle plays a pivotal role in sensory-based neuroprotection in a rat model of permanent middle cerebral artery occlusion

We have previously demonstrated protection from impending cortical stroke is achievable by sensory stimulation of the ischemic area in an adult rat model of permanent middle cerebral artery occlusion (pMCAo). We have further demonstrated that a major underpinning mechanism that is necessary for such protection is the system of collaterals among cerebral arteries that results in reperfusion of the MCA ischemic territory. However, since such collateral flow is weak, it may be necessary but not sufficient for protection and therefore we were seeking other complementary mechanisms that contribute to sensory-based protection. We hypothesized that astrocytes-to-neuron shuttle (ANLS) is another potential underpinning mechanism that could complement collateral flow in the protection process. Supporting our hypothesis, using functional imaging, pharmacological treatments, and postmortem histology, we show that ANLS has a pivotal role in sensory-based protection of cortex and therefor serves as the other supporting mechanism underpinning the protection process.


Introduction
Using permanent middle cerebral artery occlusion (pMCAo) in a rat model of ischemic stroke, we have demonstrated protection from impending ischemic stroke by sensory stimulation delivered within a twohour time window of protection after ischemic onset [1][2][3][4][5][6][7][8] . What are the mechanisms underpinning this sensory-based neuroprotection in this rat model? We have demonstrated that a necessary component for such protection is the pial collateral system (leptomeningeal anastomoses) that provide retrograde blood ow to the occluded MCA from other cortical arteries resulting in reperfusion of the ischemic area 1 .
Applying Doppler optical coherence tomography (DOCT) that enables direct quanti cation of cortical blood ow and ux we found that sensory stimulation following pMCAo indeed enhanced retrograde collateral blood ow and ux into the permanently occluded MCA 9 . However, despite such enhancement, the collateral blood ow and ux remained at a very low level during the initial critical hours for protection following pMCAo 9 . These ndings suggested that, although collateral ow is necessary for neuroprotection of the ischemic cortical tissue, it may not be su cient. We therefore searched for another complementary mechanism that could also participate in sensory-based neuroprotection.
Brain energy management is based on collaborative and dynamic interactions of neuron-glia-vasculature (NGV) involved in energy production, transfer, and utilization [10][11][12][13] . In recent years there is growing supporting evidence for the importance of the astrocyte-to-neuron lactate shuttle (ANLS) that provides lactate as a source for 'on demand' energy needs of activated neurons and preferential energy source for some neurons 11,14−19 . ANLS model postulates that glutamate release in activated cortical synapses in response to neuronal stimulation triggers enhanced production of lactate in astrocytes following signaling cascade through aerobic glycolysis 10,20,21 . As protection depends on sensory stimulation of the ischemic cortical tissue, and since whisker stimulation used for protection is indeed known to enhance glucose uptake in cortical astrocytes 22 , we therefore hypothesized that the ANLS is a potential underpinning complementary mechanism in sensory-based neuroprotection. To address this hypothesis, following pMCAo, we used pharmacological manipulations that blocked the lactate receptors on cortical neurons located within the ischemic territory during sensory stimulation. Our results support the hypothesis that the lactate shuttle is also a necessary complementary underpinning mechanism for sensory-based neuroprotection in the rat model of ischemic stroke and highlights the importance of sensory-activated astrocytes in neuroprotection. These ndings, together with our previous ndings that show how sensory stimulation minimizes the post-pMCAo build-up of infarct-predicting widespread cortical synchrony of spontaneous local eld potentials (LFPs) 23,24 , demonstrate how protection from impending ischemic stroke by sensory stimulation is multidimensional involving all components of NGV from the collateral system to ANLS and spontaneous neuronal activity.

Materials And Methods
All procedures followed NIH were approved by UC Irvine Animal Care and Use Committee (IACUC, protocol #: AUP-21-065). All methods are reported in accordance with the ARRIVE guidelines.

Animals
Thirty-two experimental subjects, 230-450g male Sprague Dawley rats (Charles River Laboratories, Wilmington, MA, USA) were individually housed in enriched cages and placed in a temperature, humidity and light-controlled room following a twelve-hour-cycle (6am-6pm). Each rat was handled daily for twenty minutes for 3-5 days, prior to the experiment.

Reagents
The lactate shuttle between astrocytes and neurons is known to act through monocarboxylate transporters (MCT). MCTs are ubiquitously expressed plasma membrane proteins responsible for protonlinked transfer of molecules with one carboxylate group, across the cell membrane 25

Experimental Design
In this study, we applied a within subject design where a baseline is established and compared to post manipulation result. Thirty-two subjects were randomly assigned to one of four experimental groups by an experimenter blind to the experiment protocols. All the experimental groups are shown in detail in Fig. 1D. Rats in group P1 (n = 8) received pMCAo, MCT inhibitor (4-CIN, 8.6 mM), and immediate post occlusion whisker stimulation (referred as + 0h). Rats in group P2 (n = 8) received pMCAo, vehicle (DMSO, 4.5%) and + 0h whisker stimulation. Rats in group P3 (n = 8) received application of vehicle (DMSO) and + 0h whisker stimulation but no pMCAo. Rats in group P4 (n = 8) received sham treatment for MCA occlusion, MCT-inhibitor (4-CIN) and + 0h stimulation. Groups P2-P4 serve as three different control groups to group P1.

Surgical Preparation
At the beginning of each experiment, subjects were brie y anesthetized with 4% Iso urane and injected intraperitoneally (ip) with sodium pentobarbital bolus (55 mg/kg, body weight (bw)). Supplemental injections (14 mg/kg bw) were given as necessary to maintain loss of withdrawal re ex to toe/tail pinch.
C2 whisker was identi ed, and the remaining whiskers were cut before the start of surgical procedures. 5% dextrose (3mL) and atropine (0.05 mg/kg, bw) were administered at the beginning and end of day 1 of the experiment. Body temperature was continuously measured via a rectal probe and maintained at 37° Celsius by a self-regulating thermal blanket. Heart rate and partial oxygen saturation were also monitored throughout the experiment.
A midline incision was made, and soft tissue was resected to expose a ~ 7 mm x 7 mm 'imaging' area of the skull over the left primary somatosensory cortex (rostro medial corner positioned caudal and lateral from bregma) was thinned to ~ 24-32 µm using a dental drill as shown in Fig. 1B&C. After baseline intrinsic signal optical imaging (ISOI) of whisker C2 whisker functional representation (WFR), skull was lifted and 1-4 slits in dura were made at or around the area C2 WFR (see below).
At the end of the imaging sessions on day 1, analgesic (Flunixin meglumine) was injected subcutaneously (2 mg/kg). The closed wound was covered with topical antibiotics (Antimax, Petco), and rats were monitored while recovering from anesthesia. Rats were returned to their home cage and allowed to recover overnight prior to + 24-hour ISOI. Rats were euthanized, with lethal dose of Euthasol (2 ml, ip) and prepared for histology. The experiment timeline and procedures are shown in Fig. 1.

Skull-dura slits
The procedure of dura slits was adapted and improved from our previously reported procedure 30 . In this procedure full craniotomy of the imaging window was replaced by small, aligned skull slits with dura slits. All steps were performed at 50-120x magni cation and shown in Fig. 2. A small opening was made in the thinned skull with ne forceps and the skull ap was gently lifted and bent as shown in Fig. 2A&B.
Notably, the skull is only lifted but not removed or separated. The tip of 30G needle was bent to grab and lift the dura up, away from the cortex as shown in Fig. 2C. The slit of desired size was made in the dura with the sharp part of the needle while it was lifted and made no contact with the cortex (Fig. 2D). After the slit was cut to the right size the needle was carefully separated from the dura. Using micro scissors, a small slit in the skull was made, aligned to the dura slit(s). The thin skull was then returned to its original position as shown in Fig. 2G. Zoomed in images of dura slit and aligned dura-skull slit are shown in Fig. 2C&G, respectively. By making aligned slits in both thinned skull and dura, this procedure ensures minimal tissue damage, prevents cortical herniation, and reduces cortical movement artifacts during imaging.

Permanent dorsal middle cerebral artery occlusion (pMCAo)
This procedure was rst demonstrated by Davis et al 31 . Brie y, the base of the left middle cerebral artery is permanently occluded at the M1 segment blocking ow to all MCA cortical branches. To do this, the skull and dura are carefully removed from a 1.5 x 1.5 mm 'surgical window' just anterior and lateral to the imaging window (over the M1 segment of MCA, just distal to MCA's lenticulostriate branches and proximal to any cortical branching) as shown in Fig. 1B&E. A small needle is threaded with 8-0 silk thread and passed at two locations through the pial layer of the meninges, below MCA as shown in Fig. 1F. Both threads are then tightly tied around MCA (Fig. 1G). Care is taken to avoid damaging the artery, and experiments are terminated if MCA is hemorrhaged. For the sham surgery group, the same procedure is used but without the nal tying.

Drug Administration
A petroleum jelly (Vaseline) well was made around the imaging window and lled with the drug/vehicle as shown in Fig. 1H. The well was then covered with a cover slip and the drug was allowed to diffuse through skull/dura slits, for one hour before pMCAo and for two hours after pMCAo as two hours after pMCAo is the critical time window for sensory-based neuro-protection 6 .

Intrinsic signal optical imaging (ISOI)
A detailed description of ISOI data acquisition and analysis can be found elsewhere [32][33][34] . Brie y, a charge coupled device (CCD) camera (16-bit Cascade 512F, Photometrics, Tucson, AZ, USA) equipped with an inverted 50 mm AF Nikon lens (1:1:8, Melville, NY, USA) combined with an extender (model PK-13, Nikon, Melville, NY, USA) is used for imaging and controlled by V + + Precision Digital Imaging System software (Digital Optics, Auckland, NZ). Data is acquired in 100 ms frames that are summed to 500 ms frames to improve signal-to-noise ratio. The cortex is illuminated with a single red-light emitting diode (635 ± 15 nm wavelength).
Two types of whisker stimulation protocols, sparse and condensed, are applied during imaging at baseline and 24 hours. Sparse protocol is used as it produces the same temporal sequence of functional response phases as also imaged by BOLD-fMRI, a protocol that was consistently applied in all our previous work [1][2][3][4][5][6][7][8] . A condensed protocol is used because it mimics the naturalistic pattern of whisking in awake animals 35 .

Sparse protocol
During each 15 s trial of sparse protocol, 1.5 s of pre-stimulus data, 1 s of during stimulus and 13.5 s of post-stimulus data was collected, with a 6 ± 5 sec random inter-trial interval.

Condensed protocol
During each 4.5 s trial of condensed protocol, 1.5 s of pre-stimulus data, 1 s of during stimulus and 2 s of post-stimulus data was collected, with a 1 sec constant inter-trial interval.
For both protocols, stimulus consisted of a single whisker being de ected by 9° in the rostral-caudal direction at a rate of 5 Hz for a total stimulus duration of 1 second in each trial. Data is collected in blocks of 64 stimulation trials for sparse and in blocks of 40 for condensed whisking protocol. All post pMCAo (+ 0-hr stimulations) consisted of four blocks of 64 sparse whisking stimulation which have previously been reported as neuroprotective in this model [1][2][3][4][5][6][7][8] . In all animals, 256 trials of whisker stimulation immediately after pMCAo are referred as + 0hr stimulation 1,2 . The parameters of both sparse and condensed protocols are shown in Fig. 1I-J, respectively.

Imaging analysis
From raw images, ratio images were created from calculating fractional change (FC) values by dividing each 500 ms frame of post-stimulus signal activity by the 500 ms frame of pre-stimulus intrinsic signal activity collected immediately before stimulus onset. WFR for sparse protocol shows three distinct phases following stimulation. In this study we analyzed only the rst two phases of WFR: the initial dip (typically black when applying a gray scale) and the following overshoot (typically white when applying a gray scale) 36 . The initial dip represents evoked neuronal activity, and the overshoot represents oxygenated blood ow into the evoked area following evoked activity. The condensed protocol only shows a single phase of growing initial dip.

Spatial Analysis
The ratio images containing the areal extent for each intrinsic signal phase were selected and Gaussian ltered (half width = 5). The areal extent was quanti ed at a threshold level of 1.75 x 10 − 4 for initial dip and at 2.5 x 10 − 4 threshold for the overshoot, away from zero. Peak amplitude was quanti ed in fractional change units for the pixel with peak activity within the areal extent. Notably, for comparing changes between baseline and 24 hours in all experimental groups, the pixel of peak intensity was selected within the area of the skull-dura slit (drug diffusion).

Temporal Analysis
As spatial parameters for initial dip and overshoot are based on single 500 ms frame, temporal pro les were acquired from all the frames of a single selected peak value pixel within the area of the skull-dura slits. The timing parameters had previously been reported for the WFR and temporal pro le of perfusion and are now used for ISOI-WFR analysis 34,37 . The detailed analysis and results of all timings parameters are reported in supplementary materials ( Figure S1-S2).
Histology (staining for infarct) The brain was sectioned into 2 mm coronal slices, and incubated in 2% TTC solution at 37°C for 20 min in the dark 38 . TTC is enzymatically reduced, producing formazan (a bright red byproduct), by dehydrogenases in active mitochondria. Red stain intensity correlates with the number and functional activity of mitochondria, unstained (white) areas are indicative of infarct 39 .

TTC analysis
The TTC-stained sections were photographed with a digital camera, and images were analyzed using ImageJ software (National Institute of Health). The total infarct volume was determined by multiplying the infarct area of each slice by the thickness of that slice and nal volumes were then corrected for edema. An observer blind to the experimental groups performed the volume calculation. Small damage at the pMCAo surgical site was readily distinguished from the large ischemic infarct and was excluded from infarct analysis. The images with infarct in group P1, were superimposed on the images obtained from the rat brain atlas (George and Pixanos) 40 . The atlas images were selected based on landmarks observed in the slices.

Statistical analysis
For ISOI-WFR imaging data, repeated measures analysis of variance (RM-ANOVA) was run to analyze potential differences between baseline and 24 hours after pMCAo in all experimental groups. Repeated measures were performed for one between subject's variable (experiment group, P1-P4) and one within subjects (time, baseline vs 24-hours) followed by post hoc contrasts to identify which groups differed from baseline at 24-hours. Alpha level was set to 0.05, and Bonferroni corrections were applied to account for multiple contrasts. Infarct volume comparisons were performed by employing two-sample ttests. One-way ANOVA was performed with post hoc Bonferroni corrections to ensure that no statistical difference exists between baseline values of all groups (P1-P4). All statistics and plotting were performed using PRISM (GraphPad version 9). Results are expressed as means and standard errors.

Neuroprotection through ANLS post pMCAo
A strong dependence on ANLS post pMCAo veri es the role of lactate in neuroprotection.
Representative results for experiment groups: P1-P4 For each experimental group P1 through P4, a representative case of a whisker functional representation (WFR) is shown in Fig. 3 where whisker stimulation followed the sparse protocol at baseline and at 24-hr. For experimental group P1 no activity was recovered 24 hours after pMCAo. Conversely, for the P2-P4 control groups the WFR remains intact 24 hours after the intervention.
For each experimental group P1 through P4, single representative result at baseline and 24-hr, for condensed protocol is shown in Fig. 4. Again, only experimental group P1 shows absence of activity after 24 hours. In all the other control groups (P2-P4) the activity remains intact at 24 hours.
Blocking ANLS leads to complete elimination of spatial response of WFR during treatment time.
Blocking ANLS through MCTs affects the WFR also during treatment time, an effect imaged with MCT inhibitors applied in groups P1 and P4. Figure 5 show a representative example of the ISOI-WFR in the presence of MCT inhibitor where the distinct phases of activity are absent. This effect was only present for the duration of MCT inhibition. However, the difference between experimental groups P1 and P4 is seen at 24-hours after the 4-CIN treatment. Only group P1 shows absence of WFR whereas group P4 shows full recovery of WFR. These results were consistent for both condensed and sparse whisking protocols. We further veri ed the complete elimination at 24h by absence of imaged WFR even at the lowest quanti cation threshold.

ISOI-WFR spatial analysis
Experimental results at baseline and 24hr for areal extent and peak amplitude as quanti ed for the initial dip and overshoot phases during application of the sparse protocol are shown in Fig. 6. Only group P1 shows a signi cant decrease in both the areal extent (denoted as 'area') and peak value of initial dip and overshoot are evident. All the other groups (P2 through P4) do not show any signi cant difference.
Results from groups P2 and P3 show no signi cant effect for vehicle (DMSO) application after pMCAo (P2) or during normal conditions (P3). Lactate transport inhibition in the sham pMCAo group (P4) also did not show any signi cant change in the area size and peak value. Figure 6 also shows the spatial parameters of the initial dip during application of the condensed whisking protocol and show a similar trend as in Fig. 6, where the loss of spatial parameters is only evident in group P1, and all the other groups (P2 through P4) show no signi cant effect at 24 hours after the treatment.
Blocking ANLS leads to complete absence of temporal pro le of WFR.
Blocking ANLS in 4-CIN treated area results in elimination of temporal ISOI-WFR parameters at 24 hours as shown by temporal quanti cation of WFR.

ISOI-WFR temporal analysis
Experimental results for temporal parameters of the timing analysis are shown in supplementary gure S2. As with the spatial analysis, only in group P1 there was a signi cant decrease in the temporal parameters, half time ratio and peak time ratio of initial dip and overshoot phases after 24hr. All the other groups did not show any signi cant difference, indicating no signi cant effect of the vehicle (DMSO) after pMCAo (P2) and during normal conditions (P3). Blocking lactate transport in sham rats also did not show any signi cant change at 24 hours in the temporal parameters of WFR, but only a temporary inhibitory effect on the WFR during drug administration (Fig. 5) that disappears at 24 hours, as shown in our results at 24-hr in group P4.
The timing parameters measured for all the groups P1-P4 also show signi cant change only in the group where lactate shuttle is blocked after pMCAo (P1). The non-signi cant change in all the parameters for group P2-P4 provided further proof of perseverance of functional response in presence of vehicle and in Sham animals.
Brain slices from a representative selected animal for each of the group P1-P4 are shown in Fig. 7. The superimposition of Paxinos and Watson's rat brain atlas according to the slices' landmarks show the resulting infarct following pMCAo in a case where the lactate transport is blocked over large cortical area (largest skull-dura-slits). The superimposition of TTC slices with the rat brain atlas shows infarct covering parts of primary motor cortex, primary somatosensory cortex, auditory cortex all the way up to parts of primary visual cortex, spanning almost the entire MCA territory 41 .
Infarct volume is directly proportional to the size of slits.
The location of infarct is also always found directly under the slit as shown in Fig. 8A-F. One skull-dura slit produced infarct directly under it (Fig. 8A-C) and two different locations around the area of evoked activity show infarct at two locations around the area of functional activity, preserving function and structure of cortex not covered by dura slit (Fig. 8D-F). Therefore, functional, and structural preservation is directly correlated with the size of the slit created for drug diffusion to block the MCTs for lactate transport as shown by our results. For all the rats in group P1, our results show a strong linear correlation between the size of slit to the volume of infarct as shown in Fig. 8G.

Discussion
Astrocytic glycogen, a precursor of lactate, is the only endogenous fuel reserve for energy management during metabolically intensive or stressed conditions [42][43][44] . Astrocytic glycogen reserve is dynamically maintained through glycogenesis (via glycogen synthase) and glycogenolysis (via glycogen phosphorylase) 45,46 and is further metabolized during a metabolic demand from active neurons to generate lactate and therefore supplement such active neurons with additional energy to maintain functionality 15,27,28 . This process starts when evoked synaptic activity causes glutamate release at cortical synapses that in turn triggers astrocytic lactate release 47,48 . During bursts of increased neuronal ring the glycogenolytic ux also increases and the astrocytic glycolytic pathway provides rapid lactate release to match intensive energy requirement 29 49 . The lactate produced via glycogenolysis is released to the extracellular space through monocarboxylate transporters (MCTs) to be consumed by an oxidative site, primarily neurons, constituting the astrocyte-neuron lactate shuttle (ANLS) 50 . During ischemic conditions and injury in general, astrocytes change their morphology and function to become reactive astrocytes 51,52 . The role of lactate produced by reactive astrocytes in response to sensory stimulation within the critical 2h time window of protection has not been tested yet. We therefore hypothesized that lactate released from reactive astrocytes in response to sensory stimulation of neurons in the ischemic area could be the complementary mechanism to collateral blood ow in the process of protection by whisker stimulation in rat model of pMCAo. To test our hypothesis, we used quanti cation of ISOI-WFR over time following two stimulation protocols, in pharmacologically treated groups of rats to study in real time the role of astrocytic lactate in vivo.
Solving two key issues was instrumental for the success of the pharmacological applications in this study: 1) the novel development of the aligned skull-dura slits, and 2) the choice of the appropriate concentrations of the vehicle DMSO and blocker 4-Cin.
1) For drug administration ( Fig. 1 & Fig. 2 ) we created aligned skull-dura slits at or around the site of activity, avoiding major pial vascular network. This preparation has several advantages: a) Controlled drug diffusion over time only at the slits region, b) Being superior to an intracerebral injection by avoiding damage to cortex, c) No mechanical disturbances and drug pocket in the cortical milieu due to pressure injection, d) This procedure does not limit the drug volume as is the case in injections, e) keeping the thinned skull preparation with small sized slits, as small as 1 mm, eliminates cortical herniation and motion artifacts produced by heartbeat and respiration, and nally f) the size of the slits are linearly correlated with the cortical volume that is affected by the pharmacological intervention.
2) The use of DMSO has been reported to be potentially toxic above a certain dosage, therefore for this study it was important to use a concentration which was not toxic to the cortical milieu 53,54 as veri ed by the ISOI-WFR. Indeed, our results in control groups P2 (DMSO + pMCAo) and P3 (DMSO without pMCAo) demonstrated that no spatial or temporal changes were detected for the WFR, verifying that the concentration of DMSO used for our study had no toxic effects. The concentration of 4-Cin used in this study has been in the middle range of what has been previously used for MCT inhibition [55][56][57] . This concentration was chosen by considering that the dynamic diffusion through the cortical extracellular space does not minimize the e cacy of the drug, and conversely that it does not exceed a concentration that could have any spatial or temporal effect on the WFR during normal condition, as shown in control group P3 (DMSO without pMCAo).
We used ISOI in combination with pharmacological manipulation of lactate transporters to reveal the role of ANLS in sensory-based protection following pMCAo by quanti cation of the WFR over time. Previously, we have shown that despite pMCAo, rats show cortical structural and functional protection following whisker stimulation within 2h window after pMCAo 1,2,4 . By pharmacologically inhibiting MCTs using 4-CIN, lactate transport is blocked between neurons and astrocytes resulting in both immediate and 24-hr abolishment of the WFR in addition of causing an infarct as seen in postmortem histology. These functional and structural results highlight the pivotal role of lactate shuttle in the neuroprotection process following pMCAo in rats. Further proof of ANLS support in neuroprotection of our pMCAo model is evident by our nding that the volume of infarct is directly proportional to the size of the dura slits or the volume of cortical region that had ANLS inhibited, as shown in Fig. 8. In addition, we have demonstrated that blocking the ANLS transport in normal (sham, group P4) conditions during the application of 4-Cin had also strong obliterating effects on the imaged WFR (Fig. 5). Corroborating these results, previous studies reported that blocking lactate transport by downregulation of MCT transporters (MCT2 or MCT4) abolishes cortical evoked BOLD response to whisker stimulation (equivalent the ISOI overshoot phase) as measured by fMRI and by nuclear magnetic resonance spectroscopy 22,58−60 .
We were able to generalize our ndings using two very different types of whisker stimulation protocols. A sparse protocol, which was used in all our previous protection studies, was applied in this study with the advantage of being able to quantify two out of the three phases (initial dip and overshoot) that also characterize high-powered BOLD-fMRI response following sensory stimulation. The condensed protocol shows only one phase, but its advantage is in its similarity to whisker stimulation parameters in naturally whisking rats. Our results show that the spatial and temporal WFR is preserved in all the different conditions of interventions and drugs used (groups P2-P4) for both whisker stimulation protocols and consequently demonstrate that ANLS-based protection is successful regardless of these very different stimulation protocols.
Neuronal activity and astrocytic support to sustain neuronal activity requires both glucose and glycogen consumption and is energy intensive even for a healthy brain 15 . The energy consumption during ischemia coupled with lactate release following whisker stimulation could serve in part to support the transition of astrocytes to their protective reactive state that in turn could potentially continue supporting the cortex even beyond the 2h time protection window. With blocked ANLS, neurons could still take up glucose through glucose receptors from the collateral blood ow potentially providing glucose for energy consumption. Using DOCT, a technique that allows for quanti cation of blood ow and ux, the restoration of collateral blood ow has been reported, albeit weak, only about 10% of the baseline levels in the same model of pMCAo in rats 9 . Our results show that neuroprotection was not due to neurons taking up glucose from collateral blood ow but through reactive astrocytes supplementing energy through ANLS. We therefore posit that the major contribution for neuronal survival by the ANLS following pMCAo is due to astrogliosis and the process of turning the astrocytes into protective reactive astrocytes during ischemia 51,61,62 , together responding to the needed increase in glycogen production. Some researchers have also suggested that glycogen content during basal condition in the healthy rat brain is in a very minimal amount (10-12 micromole/g), which can only help in sustaining the tissue for only few minutes after the ischemic insult 63 . However, our previous and current results do not support the case of 'few minutes of support' following pMCAo as we have previously shown that following pMCAo the cortex is still protected for at least 2h even without any sensory stimulation 1,2 . The current study shows different results from blocking of the ANLS in experimental groups P1(pMCAo + 4-CIN) and P4 (sham pMCAo + 4-Cin) suggesting that the ANLS is supporting neurons beyond few minutes, otherwise its inhibition for 2-hr protective time-window should not have blocked any neuroprotective effects later, contrary to our results in group P1. The elimination of WFR and presence of infarct in group P1 are clear indications of cortical dependence on ANLS and not just glucose, after pMCAo. Notably, glycogen, in addition to being a precursor of lactate, is also a precursor of the cortical neurotransmitter glutamate required for neuronal stimulation 45 . As neuronal activation is increasing due to sensory stimulation, more glutamate is released, and consequently more glycogen is used to support the active neurons.
Another potential source of astrocytes reactivation could be related to changes in subthreshold neuronal cortical activity within the ischemic cortex following pMCAo. Our previous results using microelectrode arrays in the same rat model have demonstrated that during spontaneous activity, within few minutes following pMCAo, there is a remarkable widespread buildup of tight spatiotemporal synchronization of local eld potentials (LFPs) over the entire cortical depth and the entire spatial extent of the occluded MCA territory, without any change in sensory evoked LFPs and spikes as compared to pre-pMCAo baseline. Such LFP synchronization following pMCAo is the result of underlying synchronous bursts of low frequencies oscillations 23 . The continuous buildup of such synchronization over time results in an infarct, unless whisker stimulation is delivered during the 2h protective window resulting in desynchronization of the LFPs and protection from impending infarct as veri ed by postmortem histology 24 . The widespread synchrony of low frequency bursts underlying the LFP synchrony buildup following pMCAo could also lead to high energy demand to support this synchronous burst activation, which in turn could therefore also result in increase of astrogliosis and reactive astrocytes. These ndings, together with the contributions of the collaterals support system and the ANLS support system demonstrate that protection by sensory stimulation following pMCAo is a multi-dimensional integrated activity that involves neurons, astrocytes and blood vessels, all members of the NGV-unit 64 .
A major nding of this study is that ANLS is a pivotal neuroprotective component along with collateral blood ow and that neuronal stimulation is fundamental in maintaining this ANLS within the critical time of protection. Most of the relevant preclinical and clinical work has been focused on administrating exogenous lactate after ischemia, which enhances the neuroprotection of ischemic region by reducing the infarct lesion 19,65−73 . Our ndings show that sensory-based stimulation of neurons and astrocytes in ischemic area allows for protection via ANLS without the need of external lactate administration, demonstrating the ability of the cortex to sustain itself following pMCAo, if neuronal stimulation is delivered within the critical time window for protection.

Conclusion
In conclusion the results of this study are consistent with previous studies that suggest that lactate shuttle is important for neuronal functional preservation especially in a pathological condition like ischemia and highlights the important component of neuronal activation that has been mostly overlooked in previous studies. Astrocytic lactate shuttle in response to neuronal activation together with collateral blood ow and LFP desynchronization protects the ischemic cortex in our model of pMCAo. In the future, we believe that our pMCAo stroke model coupled with sensory-based neuroprotection in rats, and better understanding why such protection fails in mice 74 could further our understanding and ways to control the ischemic cortex from impending infarct.

Declarations Data availability
The dataset generated from raw images is available on request from corresponding author.

Competing interests (mandatory)
The author declares no competing interests.  Aligned skull and dura slits. A) A small opening is made in the skull to lift part of it. The white streak is the re ection of overhead light. B) Blue arrows show lifted up skull. C) A small opening is made in dura with 30G bent needle hook. D&E) dura slit of desired size is made. Blue arrows show retraced dura after the slit at 120X magni cation. F) The skull-slit halfway toward its original position, blue arrow. G) aligned dura-skull slits at 150X magni cation bounded by blue dotted triangle.

Figure 3
A representative result for each experimental group following the sparse whisker stimulation protocol.
The ISOI results at baseline (before pMCAo and drug diffusion) and at 24 hours (after pMCAo and drug diffusion) are shown for each experimental protocol (P1-P4). The red dotted lines denote the start and end of 1 sec 5hz-whisker stimulation denoted on top as a pulse train. The yellow and green circles show the speci c frames which include the regions of interest used for quanti cation of initial dip and overshoot, respectively. Linear gray scale bar indicates intrinsic signal strength, C and L denotes caudal and lateral respectively. Each frame is ~7 mm x 7 mm. Black and white streaks are large surface blood vessels.  ISOI-WFR of whisker C2 during sparse stimulation protocol. Top raw shows typical ISOI-WFR 3 phases response to stimulation during baseline, whereas the bottom raw shows the absence of the ISOI-WFR in presence of MCT inhibitor.

Figure 6
The spatial quanti cation of ISOI-WFR phases for the two stimulation paradigms. Top row: peak amplitude and area of the initial dip following application of the sparse whisker stimulation protocol as calculated at baseline and 24-hrs for each of the experimental groups P1-P4. Graphs show a signi cant difference only for group P1 at baseline and 24-hrs for both peak and area (****p<0.0001 and *p < 0.05). Middle row: peak amplitude and area of the overshoot calculated at baseline and 24-hrs following sparse protocol is shown for each of the experimental groups P1-P4 A signi cant difference is evident only for the P1 at baseline and at 24-hrs (***p<0.0021). Bottom row: peak amplitude and area of the initial dip following condensed whisker stimulation protocol as calculated at baseline and 24 hours for each of the experimental groups P1-P4. Graph shows a signi cant difference only for the P1 group calculated at baseline and at 24-hrs for both peak and area (***p<0.0021 and **p<0.03). There was no signi cant difference between the baseline values of all groups for all quanti ed parameters (p>0.1).