Bei-Yao Gao , Yi-Xing Cao , Peng-Fei Fu, Ying Xing, Dan Liang, Shan JiangYu-Xiao Xie , Min Li

Abstract The anterolateral motor cortex of rodents is an important motor auxiliary area, and its function is similar to that of the premotor area in humans. Activation and inhibition of the contralesional anterolateral motor cortex (cALM) have been shown to have direct effects on motor behavior. However, the significance of cALM activation and inhibition in the treatment of stroke remains unclear. This study investigated the role of optogenetic cALM stimulation in a mouse model of cerebral stroke. The results showed that 21-day optogenetic cALM inhibition,but not activation, improved neurological function. In addition, optogenetic cALM stimulation substantially altered dendritic structural reorganization and dendritic spine plasticity, as optogenetic cALM inhibition resulted in increased dendritic length, number of dendritic spines, and number of perforated synapses, whereas optogenetic activation led to an increase in the number of multiple synapse boutons and the number of dendritic intersections. Furthermore, RNA-seq analysis showed that multiple biological processes regulated by the cALM were upregulated immediately after optogenetic cALM inhibition, and that several immediate-early genes (including cFOS, Erg1, and Sema3f)were expressed at higher levels after optogenetic inhibition than after optogenetic activation. These results were confirmed by quantitative reverse transcription-polymerase chain reaction. Finally, immunofluorescence analysis showed that the c-FOS signal in layer V of the primary motor cortex in the ischemic hemisphere was higher after optogenetic cALM activation than it was after optogenetic cALM inhibition. Taken together, these findings suggest that optogenetic cALM stimulation promotes neural reorganization in the primary motor cortex of the ischemic hemisphere, and that optogenetic cALM inhibition and activation have different effects on neural plasticity. The study was approved by the Experimental Animal Ethics Committee of Fudan University (approval No. 201802173S) on March 3, 2018.

Key Words: anterolateral motor cortex; locomotion activity; middle cerebral artery occlusion; motor function; neural plasticity; optogenetics;stroke; synapse

Introduction

?schemic stroke occurs form a decrease in cerebral blood flow,which leads to irreversible neuronal death, abnormal circuit output function, and corresponding behavioral defects. More than half of all stroke survivors are left with varying degrees of dysfunction that last for months to years, with the most common effect being motor system dysfunction (Huo et al.,2021). In rodents and humans who have experienced stroke leading to motor deficits, maximal neuroplasticity occurs around remote areas of the infarct in the brain approximately 1 and 3 months after stroke, respectively (Bernhardt et al.,2017). During this period, endogenous plasticity, which involves changes in dendrite morphology, dendritic spine turnover, and synaptic connectivity, is enhanced, and these are the primary events that lead to recovery (Bernhardt et al.,2017). Therefore, neuroplasticity is crucial for reconstruction of neurological function in areas of infarction during this period (Joy and Carmichael, 2021).

Following stroke recovery, neuronal reorganization is often observed in both the ipsilesional and contralesional hemispheres. Recent studies suggest that some contralesional brain regions help facilitate recovery of motor function following brain injury (approximately 10–25% of corticospinal tract (CST) fibers were uncrossed after injury) (Van Kaam et al., 2018; Hensel et al., 2021). The functions of the premotor cortex (PMC) are diverse and not yet fully understood. Most researchers believe that contralesional PMC fiber integrity is a promising biomarker of upper limb motor impairment in stroke patients (Boccuni et al., 2019). ?n addition, PMC lesions are associated with impaired imitation of arm postures, which suggests that the PMC, and not the parietal cortex, plays an important role in performance deficits (such as deficits in copying novel and meaningless gestures) (Wong et al., 2019).Motor function improvement is known to be associated with ipsilesional CST integrity. More importantly, in a neuroimaging study that used resting-state functional magnetic resonance imaging, task-based functional magnetic resonance imaging,and diffusion tensor imaging, functional connectivity between M1 and contralesional PMC areas was found to be significantly correlated with motor function improvement in chronic stroke patients by multiple linear regression models (Yuan et al.,2020). On the other hand, inhibition of the contralesional PMC via 1-Hz repetitive transcranial magnetic stimulation improved motor function in the affected hand in patients who had experienced subacute stroke (Lüdemann-Podubecká et al., 2016). Thus, we hypothesized that inhibition and activation of the contralesional anterior lateral motor cortex(cALM) would have different effects on the ipsilateral M1(iM1). To test this, we evaluated the effects of cALM activation and inhibition in a mouse model of stroke based on middle cerebral artery occlusion (MCAO). Dendritic morphology,synaptic transmission, and the activity of mature neurons in iM1, as well as motor recovery, were assessed to determine the effects of cALM stimulation on neural plasticity. The different effects of cALM activation and inhibition on motor recovery were also assessed.

Materials and Methods

Animals

Stroke damage increases with aging in female mice, whereas male mice exhibit less damage after MCAO (Liu et al., 2009);therefore, only male mice were included in this study to eliminate gender as a confounding factor. Thirty-six wild-type adult male C57BL/6J mice (age 10–12 weeks, weight > 23 g)were obtained from the Shanghai Lab Animal Research Center(Shanghai, China; license No. SCXK (Hu) 2018-0006). All experiments were approved by the Experimental Animal Ethics Committee of Fudan University (approval No. 201802173S) on March 3, 2018. All experiments were designed and reported according to the Animal Research: Reporting ofIn VivoExperiments (ARRIVE) guidelines.

Middle cerebral artery occlusion model establishment

The stroke model was established by performing MCAO surgery, as described previously (Zhang et al., 2019a). Briefly,mice were anesthetized by inhalation of 1.5% isoflurane (China National Pharmaceutical Group Co., Beijing, China) in a 30%O2/70% N2O gas mixture. Body temperature was maintained at around 37.0°C by a heat lamp and was monitored by rectal thermometer (Le Shuo Co., Shenzhen, China). A 7–0 nylon monofilament (RWD Life Science Co., Beijing, China) was inserted into the left external carotid artery, and the angle of advancement was then adjusted in order to insert it into the left internal carotid artery, thereby occluding the left middle cerebral artery. After 60 minutes, the filament was withdrawn to establish reperfusion. Cerebral blood flow was measured by laser-Doppler flowmetry (RWD Life Science Co.), and any mice with less than 75% cerebral blood flow reduction were excluded. After the operation, lidocaine (China National Pharmaceutical Group Co.) was injected subdermally to reduce pain, and broad-spectrum antibiotics (amoxicillin-clavulanate,China National Pharmaceutical Group Co., 900 mg/kg per day)were administered temporarily to prevent infection. On day 2 after the MCAO surgery, primary assessment of neurological deficits was carried out by placing the mouse on the ground to observe its walking function, so as to determine whether the stroke model had been established successfully (Schaar et al., 2010) (Additional Video 1).

The MCAO mice were randomly divided into three groups: the sham-treatment group (Sham group), the selective neuronal inhibition with cALM group (SOI group), and the selective neuronal activation with cALM group (SOA group) (n= 6/group). A schematic of the experimental timeline is shown in

Figure 1.

Stereotaxic surgery

After anesthetization by inhalation of 1.5% isoflurane in a 30% O2/70% N2O gas mixture, each mouse was fixed in a stereotactic frame (RWD Life Science Co.), the scalp was cut to expose the skull, and then the holes were drilled to access the cALM in right hemisphere (anteroposterior = +2.5 mm,mediolateral = –1.5 mm, and dorsoventral = –1.0 mm;Figure1C) using a special skull drill after stereo positioning. Then,rAAV-hsyn-eNPHR3.0-mCherry-WPRE-PA (5 × 1012vg/mL,200 nL; SOI group), rAAV-hsyn-hChR2-mCherry-WPRE-PA (5 ×1012vg/mL, 200 nL; SOA group), or nonfunctional opsin (rAAVhsyn-mCherry-WPRE-pA; 5 × 1012vg/mL, 200 nL; Sham group)purchased from OBiO Technology Corp., Ltd. (Shanghai, China)was injected at a rate of 50 nL/min (Cheng et al., 2014). After injection, the electrode filled with the virus was left in place for at least 10 minutes, an optic fiber cannula (Newdoon Technology Co., Ltd., Hangzhou, China) was stereotaxically implanted into the cALM, and the cannula was secured to the skull using dental cement (Figure 1D).

Stimulation paradigm in contralesional side

Mice received photostimulation beginning day 7 after the MCAO surgery, as described previously (Zhang et al., 2019b).Each mouse was placed in its own cage with a laser cable(Beijing Stelle Laser Technology Co., Ltd., Beijing, China)connected to the fiber cannula. All mice received daily stimulations for 21 days. In the SOA group, a 473-nm blue laser was controlled by a driver (RWD Life Science Co.), and mice were stimulated with a laser set to 10-Hz, 20-ms light pulses. In the SOI group, a 594-nm yellow laser was controlled by the same driver but stimulated with direct current for the same duration as for the SOA group. Sham stimulation was conducted without turning on the laser for the Sham group (Figure 1D). One session of stimulation involved five, 1-minute stimulations, and each 1-minute stimulation period was followed by a 5 minutes of no stimulation (Figure1E). Stimulations were performed at 12:00 p.m. daily. The light intensity was set at 5 mW to avoid direct activation of neurons, as described previously (Magno et al., 2019). After 21 days of one session of stimulation per day, mice were inhalation anesthetized by inhalation of 1.5% isoflurane in a 30% O2/70% N2O gas mixture and then sacrificed. Accurate injection of the virus into the cALM region was confirmed under a fluorescence microscope.

Figure 1｜Study design and optogenetic stimulation in the contralesional anterolateral motor cortex (cALM).

Behavioral test

Behavior was evaluated on day 1 (the day before the MCAO surgery was performed) and days 6, 14, and 28 after MCAO(Figure 2A).

Revised neurological severity score

Revised neurological severity scores (NSS-R) were obtained for six mice per group, as described previously (Yarnell et al., 2016). The score was calculated based on the following aspects: motor tests, lifting the mouse by its tail, placing the mouse on the floor, sensory tests, beam balance tests,absent reflexes, and abnormal movements. The maximum score was 20: a score of 15–20 indicated severe injury, 8–14 indicated moderate injury, and 1–7 indicated mild injury. The rater and the data analysis statisticians were all blinded to the experimental group information.

Open field test

Mice (n= 6 per group) were subjected to the open field test using a maze (Noldus Co., Ltd., Wageningen, the Netherlands)that contained one mouse at a time, and the speed and total distance of travel within the inner zones of the maze were analyzed using an EthoVision XT system (Noldus Co.,Ltd.), as described previously (Abbasi et al., 2020). Real-time recordings were obtained throughout all testing procedures.On the day that the open field test was performed, the SOI and SOA stimulation protocols were not performed, and instead optical stimulation was performed during the test.

Golgi staining

On the 28thday after MCAO, mice (n= 4 per group) were euthanized by inhalation of 1.5% isoflurane in a 30% O2/70%N2O gas mixture. The brain was removed and rinsed with double distilled H2O, immersed in a mixture of mercuric chloride, potassium dichromate, and potassium chromate,stored in the dark at 22–25°C for 14 days, and finally immersed in histological cassettes (Rotilabo? embedding cassettes; K114.1, Carl Roth GmbH, Karlsruhe, Germany)for 3 days. The embedded brains were then cut into 100-μm thick slices using a vibrating tissue slicer (CM1950 Leica Biosystems Co. Wetzlar, Germany), after which the slices were mounted on gelatin-coated glass slides, and gradient alcohol dehydration, xylene transparency, and gum xylene mounting were performed as described previously (Gao et al., 2020a).Golgi staining was performed using a RapidGolgiStainTMkit (FD NeuroTechnologies, Ellicott City, MD, USA). Three brain slices were observed for each mouse, and five fields of view were assessed for each brain slice under an optical microscope(Nikon E200 model; Nikon Co., Minato, Japan). ImageJ (Java 1.8, National Institutes of Health, Bethesda, MD, USA) was used to analyze dendritic length, dendritic complexity, and length and numbers of dendritic spines along apical and basal dendrites in layer V of iM1.

Electron microscopy

On day 28 after MCAO, mice (n= 4 per group) were euthanized as described above. Each mouse was then subjected to intracardiac perfusion fixation, and the brain was removed quickly. Next, a 1-mm3cubic tissue sample was retrieved from layer V of the iM1 (Gao et al., 2020b). Four synapses from each mouse were randomly selected for observation using a transmission electron microscope (JEOL, Ltd., Tokyo, Japan).ImageJ was used to analyze the presynaptic membrane and calculate the number of the synapses as described below.

Two special forms of synapses were observed. One is the perforated synapse, a synaptic subtype with perforated synapses have discrete postsynaptic density, abundant α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid(AMPA) receptors, and larger overall postsynaptic density regions, indicating increased functional efficacy and maturity.?t is characterized by the discontinuous electron-dense plate of the postsynaptic membrane (Kim et al., 2018). The other is the multi-synaptic boutons (MSBs), defined by multiple dendritic spines (postsynaptic partners) contacting the same axon terminal (presynaptic boutons) independently,or boutons forming synaptic contacts with no less than two postsynaptic element (Kim et al., 2018).

RNA sequencing

RNA extraction and mRNA isolation from the iM1 were performed in the same way for RNA sequencing and for quantitative polymerase chain reaction (PCR) analysis. Six independent tissues were harvested from each mouse (n= 3 mice from the SOI and SOA groups) to obtain RNA samples.We generate six RNA sequencing libraries using a TruSeq RNA Library Preparation Kit v2 (?llumina, ?nc., San Diego, CA,USA) according to the manufacturer’s instruction (Shanghai Majorbio Bio-pharm Technology Co., Ltd., Shanghai, China).After the sequencing libraries were established, all samples were sequenced by Illumina HiSeq X10 (Illumina, Inc.), as described previously (Gao et al., 2020a). The data were analyzed using the Majorbio Cloud Platform (www.majorbio.com) according to Mus_musculus and GRCm38.p6 (http://asia.ensembl.org/Mus_musculus/Info/Index). Differences in gene expression levels among groups were analyzed using the Bioconductor package edgeR (Version 3.24.3, https://bioconductor.org/packages/release/bioc/html/edgeR.html)(Robinson et al., 2010). The screening threshold was set at|log2FC| ≥ 1 andP< 0.05. We also performed sequencingrelated quality assessment of the original sequencing data from each sample using fastx_toolkit_0.0.14 (Version 0.0.14,https://github.com/agordon/fastx_toolkit). Genes/transcripts with similar expression patterns are usually functionally related, so we performed hierarchical clustering of genes/transcripts or samples using an unsupervised learning algorithm, followed by application of an iterative method to classify each gene. A heat map was used to cluster the genes/transcript expression patterns within selected gene sets(Haarman et al., 2015). We then used the GO database (http://www.geneontology.org/), a comprehensive database that standardizes biological terms for genes and gene products from different databases and uniformly defines and describes gene and protein functions (The Gene Ontology Consortium,2017). to classify and annotate genes according to biological process (BP), molecular function (MF), and cellular component(CC). We used Goatools (https://pypi.org/project/goatools) to perform GO enrichment analysis of the genes in each gene set based on Fisher’s exact test analysis (Klopfenstein et al., 2018).When the correctedPvalue was < 0.05, the GO function was considered to be significantly enriched. The enrichment factor was calculated as follows: (number of differential genes that appear in a term/ number of all differential genes)/(all genes included in the term/total number of genes in the database).

Real-time quantitative reverse transcription PCR

At 28 days post-stroke, mice (n= 3 per group) were sacrificed as described above and perfused with cold, sterile 1×phosphate-buffered saline (PBS). Brain samples (iM1, three samples per mouse) were removed and stored on ice in 1× PBS.RNA was extracted using a Qiagen RNeasy Plus kit (QIAGEN,Manchester, UK). Real-time quantitative reverse transcription PCR reaction mixtures were prepared with Taq polymerase and TaqMan primers (Takara Biomedical Technology Co., Ltd.,Beijing, China).Table 1shows the forward and reverse primer sequences. Quantitative PCR was performed using a Bio-Rad real-time PCR detection system (Bio-Rad Laboratories, Inc.,Hercules, CA, USA). Each sample was analyzed in triplicate.Quantitative PCR data were analyzed using the 2–??CTmethod(Livak and Schmittgen, 2001).

Table 1 ｜Sequences of the forward and reverse primers used for real-time quantitative reverse transcription polymerase chain reaction analysis

Fluorescent immunohistochemistry

The mice (n= 4 per group) were anesthetized quickly as described above, and samples were harvested within 1 hour after the final light stimulation. Fluorescent immunohistochemistry staining was performed as described previously (Zhou et al., 2020). Briefly, the brain samples were fixed in the paraformaldehyde and immersed in a sucrose solution, followed by sectioning on a cryostat (China National Pharmaceutical Group Co.). Next, the sections were washed with PBS-10% Triton X containing 3% horse serum(Beyotime Biotechnology Co., Shanghai, China), followed by staining with the primary antibody (rabbit anti-c-Fos antibody, RRID: AB_2106777, Cat# PA1-830, Thermo Fisher Scientific, Waltham, MA, USA; 1:1000 in PBS-10% Triton X containing 3% horse serum) for 3 days at 4°C, then staining with the secondary antibody (donkey anti-rabbit IgG (H+L),Cat# P0179, Beyotime Biotechnology; 1:1000 in PBS-10%Triton X containing 5% bovine serum albumin) for 1 hour at room temperature, and finally incubation in 4′,6-diamidino-2-phenylindole (1:1000) in PBS. Five fields of view within layer V of iM1 were analyzed for each brain slice under an optical microscope (#eclipse e100, Nikon, Tokyo, Japan). The average fluorescent intensity ratio of 4′,6-diamidino-2-phenylindole to c-FOS was quantified by ?mageJ.

Statistical analysis

GraphPad Prism Version 6.0 (GraphPad Software, LLC. San Diego, CA, USA) was used to blind the samples for data analysis. NSS-R scores, total distance traveled in the inner maze zones, and speed of movement were analyzed by twoway analysis of variance with Tukey’s multiple comparisonspost hoctest. Shore analysis was used to describe the morphological characteristics of neurons (dendritic length,dendritic complexity, and numbers of dendritic spines along apical and basal dendrites), as described previously (Patel et al., 2017). Dendritic complexity was defined as the number of intersections per shell as a function of distance from the soma (Patel et al., 2017). The number of perforated synapses,number of MSBs, dendrite length, and number of dendritic spines were analyzed by one-way analysis of variance with Tukey’s multiple comparisonspost hoctest, and dendritic complexity was analyzed by two-way analysis of variance with Tukey’s multiple comparisonspost hoctest. RT-qPCR and immunofluorescence data were analyzed by two-tailed unpairedt-test. All data are shown as the mean ± standard error of mean (SEM).P< 0.05 was considered statistically significant.

Results

Effect of optogenetic cALM stimulation on neurobehavioral changes in MCAO mice

There was no difference in NSS-R scores among the three groups on days 6 and 14 after MCAO (P> 0.05). The NSS-R scores for the mice in the SOI and SOA groups were lower than those for the mice in the Sham group on day 28 after MCAO (Shamvs. SOI:P< 0.001; Shamvs. SOA:P= 0.002;n=6/group).

In the open field test, the speed and total distance traveled within the inner zones of the maze were analyzed. No statistically significant differences in speed or total distance were found among the three groups on days 6 and 14 after MCAO (P> 0.05). However, on day 28, speed (P< 0.001,vs.Sham;P< 0.001,vs. SOA) and total distance (P< 0.001,vs.Sham;P< 0.001,vs. SOA) were higher in the SOI group than in the SOA and Sham groups (Figure 2BandC). The walking tracks for mice from all three groups at different time points after MCAO are shown inFigure 2D.

Figure 2｜Effects of optogenetic cALM stimulation on behavior in MCAO mice.

Effect of optogenetic cALM stimulation on dendritic morphology in the iM1 of MCAO mice

The results from Golgi staining of layer V of the iM1 at 28 days after MCAO are shown inFigure 3A. The dendritic length was significantly longer in the SO? group than in the Sham group(P= 0.0156;Figure 3B). There was no significant difference in dendritic complexity at 20, 80, or 100 μm from the soma among the three groups (P> 0.05). The dendritic complexity at 40 and 60 μm was increased in the SOA group compared with the Sham group (40 μm:P< 0.05, 60 μm:P< 0.05;Figure3C).

Net, dendritic spine length and numbers were analyzed, as dendritic spines represent an early adaptive response after stroke (Brown et al., 2008). Pyramidal neurons have both basal and apical dendrites. The spine length of apical dendrites in layer V of the iM1 area was significantly greater in the SOI group than in the SOA group (P= 0.0117), and the SOI group also exhibited more apical dendrite spines than did the Sham group (P= 0.0335;Figure 3D–F). However, there were no significant differences in basal dendritic spine numbers among the three groups (P> 0.05).

Figure 3｜Effects of optogenetic cALM stimulation on dendritic morphology in the ipsilateral M1 in MCAO mice.

Effect of optogenetic cALM stimulation on synaptic connectivity in MCAO mice

Perforated synapses exhibit discrete postsynaptic density,abundant AMPA receptors, and larger overall postsynaptic density regions, indicating increased functional efficacy and maturity. They are characterized by a discontinuous electrondense plate at the postsynaptic membrane (Kim et al., 2018)(Figure 4A–D). Quantitative analysis showed that the SOI group had more perforated synapses than the Sham (P=0.0004) and SOA groups (P= 0.032;Figure 4E).MSBs were defined as multiple dendritic spines (postsynaptic partners) contacting the same axon terminal (presynaptic boutons independently, or boutons forming synaptic contacts with no less than two postsynaptic elements) (Kim et al., 2018)(Figure 5A–D). The SOA group exhibited more MSBs than the Sham (P= 0.0016) and SOI groups (P= 0.0163;Figure 5E).

Figure 4｜Effects of optogenetic cALM stimulation on synaptic connections in MCAO mice.

Figure 5｜Effects of optogenetic cALM stimulation on synaptic connections in MCAO mice.

Optogenetic cALM inhibition and activation promote the expression of different genes in the iM1 after MCAO

Transcriptome analysis yielded a total of 47.01 Gb of clean data (at least 6.8 Gb for each sample), and the percentage of Q30 bases was greater than 93.63%. ?n total, 4142 expressed genes were detected, including 3737 known genes and 405 new genes, as well as 16,008 expressed transcripts,including 9102 known transcripts and 6906 new transcripts.Approximately 95.4% to 96.96% of these transcripts mapped to the reference genome. A heat map of differential gene expression between the SOI and SOA groups is shown inFigure 6A. Four of the differentially expressed genes––FOS,ARC,Sema3f, andEgr1––had similar expression patterns,participated in the same metabolic process, and clustered together in the heatmap. GO analysis was performed to determine the top functions of the most significantly enriched genes (Figure 6B). For example, the top transcripts were highly enriched in transcripts corresponding to the GO term“positive regulation of biological processes”. This GO term also correlated with 163 genes that exhibited the highest number of enriched transcripts. The original gene expression data matrix is shown inAdditional file 1. The GO term “positive regulation of biological process” was associated with 163 related genes, includingFOS,ARC,Sema3f,Egr1, which were more highly expressed in the SOI group than in the SOA group,as determined by real-time quantitative reverse transcription PCR (FOS:P= 0.0003;Egr1:P< 0.001;ARC:P< 0.001;Sema3f:P< 0.001;Figure 6C). The relative gene expression data are shown inAdditional file 1andAdditional Table 1. FOS expression was also verified by immunofluorescence: FOS immunopositivity was higher in the SO? group than in the SOA group (P= 0.0072;Figure 7A–C).

Figure 6｜Differentially expressed genes in the brain of MCAO mice after optogenetic cALM stimulation.

Figure 7｜Effects of optogenetic cALM stimulation on c-Fos expression in the ipsilateral M1 of MCAO mice.

Discussion

In the present study, we sought to determine the effects of optogenetic simulation of cALM during stroke recovery.We demonstrated that cALM stimulation decreased NSS-R and improved motor function in MCAO mice. Furthermore,we found that optogenetic activation and inhibition had different effects on neural plasticity (including dendritic structural reorganization, dendritic spine plasticity as assessed by Golgi staining, and synaptic reconnection as assessed by transmission electron microscopy), and that cALM inhibition upregulated genes associated with the GO term“positive regulation of biological process” as assessed by RNA-sequencing. These findings further our understanding of circuit reorganization after stroke and highlight the contribution of cALM to recovery.

The MCAO mouse model of is one of the most accepted animal models for replicating human cerebral ischemic stroke(Kaiser and West, 2020). Our data show that optogenetic cALM inhibition promoted functional recovery after MCAO compared with the control group. We assessed NSS-R to detect long-term effects and performed an open field test to detect short-term effects. NSS-R scores were recorded and compared at 14 and 28 days post-MCAO, given that neuromodulation in functional recovery occurs over time.During the open field test, we carried out simultaneous photostimulation to provide direct evidence that the stimulus promotes motor function. We made real-time records of the motion trajectories to analyze the direct impact of stimulation on motor function. None of the groups showed improvement on day 14 after MCAO, and ontogenetic stimulation (whether inhibition or activation) did not improve NSS-R score or locomotion. ?mprovement in behavior test performance was only seen on day 28, which may be due to a confounding effect of the MCAO procedure on motivation. MCAO in rodent models induces motor deficits but also post-stroke depression,or loss of incentive motivation/interest (Pang et al., 2015). ?n future studies, depression after MCAO could be investigated by other behavioral tests such as the forced swimming test and the sucrose preference test, which are commonly used to test the effects of antidepressants.

Recent studies have highlighted the role of ALM neurons in the planning and execution of motor output (Li et al., 2015).The most commonly used behavior test for “preparatory activity” in mice is auditory and tactile delayed-response tasks(Inagaki et al., 2018). In normal mice (without any ischemic injury), inactivation of ALM neurons altered tactile decisionmaking behaviors, suggesting that neuronal activity in the ALM is related to preparatory activity and peri-movement and can be modulated by optogenetic silencing (Guo et al., 2014). Interestingly, ALM neurons in each hemisphere have almost equivalent involvement in both contra- and ipsilateral movements. To fully understand this phenomenon,scientists performed two-photon imaging together with electrophysiology and identified two major classes of ALM neurons: intratelencephalic and pyramidal-tract neurons.Intratelencephalic neurons have mixed function for both ipsi- and contralateral movements, whereas pyramidaltract neurons appear to be involved only in contralateral movements (Guo et al., 2014). A more recent study concluded that ALM neurons preferentially control contralateral movements, and that this preference is driven by pyramidaltract neurons, while the activity of intratelencephalic neurons is transferred to pyramidal-tract neurons to affect motor output (Mahrach et al., 2020).

The current study further emphasized the importance of poststroke neural plasticity in iM1 that can be modulated by the cALM. For example, optogenetic inhibition of cALM promoted changes in dendritic morphology, as well as synaptic connectivity, after ischemic injury. At present, the function of the ALM after brain injury, especially after large-scale cerebral infarction caused by middle cerebral artery embolism, is unclear. When this type of injury occurs, substantial brain reorganization occurs; for example, the cortex undergoes function remapping and axonal protrusion in key areas during recovery after stroke (Clarkson et al., 2013). Our findings helps help explain how dendrites and synapses in M1 are reorganized after ischemic injury. The mechanism by which optogenetics induces ALM neural reorganization and recovery of motor function may be based on certain anatomical and physiological features of the ALM, specifically: 1) ALM is anatomically close to M1, which allows the PMC to receive input and project directly to M1; 2) intratelencephalic neurons in the ALM project to the CST and then out of the cortex to regulate movement; and 3) ALM signaling through motor thalamus activity ultimately triggers movement.

?n human and primates, preparatory activity can be detected in the PMC, similar to that seen in the rodent ALM (Fried et al., 2011). Interestingly, previous research on humans and primates has shown PMC reorganization and novel connections following M1 cortical lesion (Hoyer and Celnik,2011). Ipsilateral PMC facilitation was much more common and had a much more powerful effect on intrinsic hand and forearm muscle movement than contralateral PMC,but conditioning of the contralateral PMC had both elicit facilitatory and inhibitory effects on M1 output (Quessy et al., 2016). In addition, another group reported that M1 facilitation by repetitive transcranial magnetic stimulation induced dorsal PMC activity in the contralesional hemisphere(Takeuchi et al., 2005). Several lines of argument suggest that the contralesional PMC plays a role in post-stroke recovery.First, PMC is anatomically located between the posterior M1 and the anterolateral dorsal prefrontal cortex, which allows the PMC to receive direct input, process the information, and project the output to M1 to induce functional movement.Second, PMC projection can take over regulation of the motor function of M1 by the CST through input to the spinal cord or the input of the remaining part of iM1 after injury.Third, auxiliary input from the contralateral PMC to the iM1 can assist in movement recovery through the uncrossed pyramidal tract fibers, as well as through the transcallosal pathways (Kantak et al., 2012). Finally, a clinical study that used transcranial magnetic stimulation as a noninvasive technique to stimulate contralesional PMC area showed that PMS involvement significantly correlated with motor function improvement in chronic stroke patients (Sankarasubramanian et al., 2017).

Further studies in stroke patients are needed to clarify the clinical relevance of these neuronal networks. Synaptic rearrangements have been found in the cortex, brainstem,and spinal cord that may be related to recovery of motor function after injury (Filli and Schwab, 2015). This structural plasticity was originally observed in the corticospinal system after spinal cord injury or stroke (Filli and Schwab,2015). In the human brain, changes in synaptic density and morphology have been observed in neonatal brains as well as in adult brains. Mature synapses have separate presynaptic projections, while immature synapses exhibit an irregular presynaptic dense band (Huttenlocher, 1979). Preservation of synapses, in particular glutamatergic synapses, in the frontal cortex in stroke patients is associated with better clinical outcomes, and increased levels of synaptosomal-associated protein 25 and syntaxin in the brain, as determined by postmortem assessment, correlate with recovery (Honer et al.,2012). In this study, we chose to analyze perforated synapses and MSBs, two synaptic morphologies that reflect synaptic connectivity. On the one hand, MSBs in the peri-infarct M1 reflected competitive influences, and it is possible that projections from the cALM were the source of at least some of the competing synapses. On the other hand, we found more perforated synapses, which contribute to dendritic and somatic depolarization, in the SOI group (Toni et al.,2001). Non-perforated synapses, however, were frequently“silent” (Nicholson and Geinisman, 2009). In addition, the RNA-sequencing results suggested that pathways associated with the GO term “positive regulation of biological process”were upregulated in the SOI group compared with the SOA group (we chose this term because it was associated with the largest number of enriched genes, and the other GO terms inFigure 6, such as “cerebellar granular layer development”and “olfactory bulb mitral cell layer development” have only rarely been studied in the context of neural activity after cerebral infarction). Therefore, we investigated positive regulation of biological process in iM1, and confirmed upregulation of c-Fos expression by immunofluorescence.Neuronal gene expression changes dynamically, rapidly, and selectively in response to neuronal activity. ?n particular, the expression of immediate-early genes, for example, egr-1,c-fos, and Arc, is upregulated in subsets of neurons associated with synaptic plasticity. c-Fos (also known as Fos), which is induced by a broad range of stimuli, is a reliable marker for neural activity. The induction of c-Fos expression by chronic sensorimotor cortex stimulation demonstrated its functional activation. An earlier study showed increased c-Fos expression in cortical neurons in response to transcranial magnetic stimulation (Fujiki et al., 2020). c-Fos protein expression patterns in the human brain, including in the cerebral cortex,hippocampus, striatum, thalamus, and cerebellum, are very under different physiological and nonphysiological conditions(Zhang et al., 1992). Our data showing modulation of c-Fos expression suggest optogenetic-induced neural plasticity.More importantly, the long-term structural and functional information generated by the present study provide a rich data set for the mechanistic analysis of stroke recovery, which involves both structural and physiological plasticity.

However, this study did have some limitations. First, we did not analyze changes in preparatory activity in MCAO mice. Our study focused on the cALM, which regulates the motor cortex/CST after cerebral infarction, given that motor dysfunction after cerebral infarction primarily manifests as abnormal muscle strength and muscle tone. The behavior patterns that we observed that corresponded to “preparatory activity”may explain the apparent changes in ALM brain activity.Therefore, in future studies we will investigate preparatory activity pathways in rodent models of brain injury. Second,despite the encouraging results highlighting the potential of cALM stimulation for the treatment of stroke in this rodent model, additional research is required before this can be applied to humans, and there are several challenges regarding translation to the clinic. For example, it will be important to determine how to effectively locate the target light-sensitive protein genes (whether by transgene or virus transplantation)without triggering an immune response, to determine how to ensure that the photoreceptor gene is stably and continuously expressed, and to ensure its long-term safety. When it comes to the design of the device, it will be important to determine how to install the light source, how to make the light source as small as possible, how to make it implantable, and how to easily supply power to the device.

Nonetheless, the current findings strongly demonstrate that cALM inhibition rather than activation can contribute to persistent synaptic plasticity gene expression and rescue neurobehavioral defects. These findings suggests that dendritic reconnection and synaptic plasticity within cALM help lead to recovery after stroke. Our study provides a new target for precise and controllable neural regeneration that could be relevant for pharmacological research or neuromodulation technology translational research. Pairing chronic cortical stimulation of this novel target via repetitive transcranial magnetic stimulation and pharmacological therapy (targeting either intratelencephalic or pyramidaltract neurons) could potentially induce neuroplasticity and ultimately improve locomotion. Additionally, our findings lead to the hypothesis that contralateral circuitry is functionally connected with motor signals and could be a useful target for therapeutic strategies designed to improve motor function after stroke. Further research is needed to elucidate the molecular mechanism underlying the predominant effect of long-term stimulation of the contralateral side (Shijo et al.,2008) and to determine how this could be applied to precision treatment and other translational applications.

Author contributions:Study design and conception: BYG, YXC, ML;major experimental implementation: YXC, PFF; behavioral tests and data analysis: DL, YX, SJ; manuscript writing: YXX; BYG. All authors approved the final manuscript before submission for publication.

Conflicts of interest:None declared.

Financial support:None.

Institutional review board statement:This study was approved by the Experimental Animal Ethical Committee of Fudan University (approval No.201802173S) on March 3, 2018.

Copyright license agreement:The Copyright License Agreement has been signed by all authors before publication.

Data sharing statement:Datasets analyzed during the current study are available from the corresponding author on reasonable request.

Plagiarism check:Checked twice by iThenticate.

Peer review:Externally peer reviewed.

Open access statement:This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix,tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms.

Open peer reviewers:Francesca Keefe, Cardiff University, UK; Gyorgy Lur,University of California, USA.

Additional files:

Additional Video 1: Mouse model of middle cerebral artery occlusion.

Additional Table 1: RNA-sequencing raw data for expression analysis in the SOI and SOA groups.

Additional Table 2: Validation of RNA-sequencing data by real-time quantitative reverse transcription polymerase chain reaction.

Additional file 1: Open peer review reports 1 and 2.