Transplantation of human neural stem cells repairs neural circuits and restores neurological function in the stroke-injured brain

Abstract

Exogenous neural stem cell transplantation has become one of the most promising treatment methods for chronic stroke. Recent studies have shown that most ischemia-reperfusion model rats recover spontaneously after injury, which limits the ability to observe long-term behavioral recovery. Here, we used a severe stroke rat model with 150 minutes of ischemia, which produced severe behavioral deficiencies that persisted at 12 weeks, to study the therapeutic effect of neural stem cells on neural restoration in chronic stroke. Our study showed that stroke model rats treated with human neural stem cells had long-term sustained recovery of motor function, reduced infarction volume, long-term human neural stem cell survival, and improved local inflammatory environment and angiogenesis. We also demonstrated that transplanted human neural stem cells differentiated into mature neurons in vivo, formed stable functional synaptic connections with host neurons, and exhibited the electrophysiological properties of functional mature neurons, indicating that they replaced the damaged host neurons. The findings showed that human fetal-derived neural stem cells had long-term effects for neurological recovery in a model of severe stroke, which suggests that human neural stem cells-based therapy may be effective for repairing damaged neural circuits in stroke patients.

Introduction

Stroke is one of the leading causes of long-term neurological disability and death worldwide (Meairs et al., 2006). Eighty percent of stroke cases are ischemic strokes caused by thrombosis. Thrombolysis by the tissue plasminogen activator, which is currently the only Food and Drug Administration-approved drug, has a less than 4.5-hour therapeutic window, increases the risk of hemorrhage due to exacerbated blood–brain barrier damage through leakage, and benefits only about 3% of ischemic stroke patients (Tan Tanny et al., 2013). Therefore, it is essential to develop alternative therapies for ischemic stroke. Stem cell therapy is gradually emerging as a potential treatment for stroke and other diseases of the central nervous system (Zhao et al., 2021; Kim et al., 2024; Sheikh et al., 2024; Wang et al., 2024b). Laboratory studies and initial clinical trials have demonstrated the feasibility of neural stem cell (NSC) transplantation in stroke (Stroemer et al., 2009; Huang et al., 2014; Trounson and McDonald 2015; Zhang et al., 2019; Boese et al., 2020; Wang et al., 2024a). The mechanism of NSCs is not yet entirely clear, but previous studies explored the two main reparative pathways that involve functional neural replacement and bystander effects (Darsalia et al., 2007; Bacigaluppi et al., 2009; Liu et al., 2009; De Gioia et al., 2020). Recent evidence suggests that the implantation of NSCs directly into the brain promotes functional recovery and repairs damaged neuronal networks and connections in the peri-infarct region in animal stroke models (Xiong et al., 2021; Zhao et al., 2022).

The transient middle cerebral artery occlusion with subsequent reperfusion (tMCAO/R) animal model, which is widely used in stroke studies, induces rapid infarct progression and large lesions affecting multiple brain regions (i.e., striatum and cortex) (Huang et al., 2014). An extensive battery of neurobehavioral tests is needed to assess the dysfunction and recovery of stroke animals (Palma-Tortosa et al., 2020). The acute phase of stroke involves a large inflammatory response and massive cell death in the ischemic brain, which can create a harmful microenvironment for transplanted cells, so the timing of the transplant can significantly influence graft survival (Huang et al., 2014). This study aimed to investigate the role of human NSCs (hNSCs) in neurological recovery during the subacute stroke period (7 days after 150 minutes of tMCAO). Although previous studies have shown that inflammation has been reduced during the subacute stroke period (Bacigaluppi et al., 2009; Abematsu et al., 2010; Shin et al., 2018), there are still many unanswered questions about the detailed effects, pathways of action, and long-term safety of NSCs transplantation during this period. Therefore, the primary purpose of this study was to systematically evaluate the long-term effects of hNSCs on neurological recovery by implanting them in mouse and rat tMCAO models.

Methods

Animals

All specific-pathogen-free-grade animals (12-week-old male Sprague-Dawley adult rats [body weight, 250–280 g] and 8-week-old male C57BL/6J adult mice [body weight, 24–26 g]) (Zhong et al., 2023) were purchased from the Vital River Laboratory Animal Technology Corporation (Beijing, China; license No. SYXK (Jing) 2021-0056), maintained at room temperature (23 ± 1°C) on a 12-hour light/dark cycle with free access to food and water, and housed two per cage. The present study conformed to the guidelines of the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All animal experiments were approved by Beijing Viewsolid Biotechnology, Beijing, China (approval No. VS2126A 00737) on January 3, 2023, and were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (8^th ed., National Research Council, 2011). This study followed the principles laid out in the 2016 ISSCR Guidelines for Stem Cell Research and Clinical Translation.

Stroke models

Transient focal cerebral ischemia was induced using intraluminal tMCAO in rats (n = 56) and mice (n = 14), as described previously (Sun et al., 2013; Luo et al., 2024). Of the total animals, 36 rats (mortality, n = 6) were used for behavioral testing and in vivo differentiation experiments, 20 rats (mortality, n = 5) for detecting synaptic connections in host-to-graft experiments, and 14 mice (mortality, n = 7) for ex vivo electrophysiological studies (Figure 1). Briefly, laboratory animals were placed in an anesthesia machine (YuYan Instruments, Shanghai, China, ABM3000) containing isoflurane (RWD Life Science, Shenzhen, China, Cat# R510-22-10) anesthetic box. The oxygen flow rate was adjusted to 1 L/min, and the isoflurane flow rate was 0.5 L/min. After anesthesia, the isoflurane flow rate was adjusted to 0.3 L/min for continuous anesthesia through the nasal cavity. The right common carotid artery was exposed and divided upwards to the internal and external carotid arteries’ bifurcation (ICA, ECA). The ECA was exposed approximately 2 cm, sutures were ligated at the distal end, and the ECA was transected. A nylon monofilament (Cinontech Beijing, China, Cat# 3040) was advanced through the ICA until a slight resistance was applied. Then, 150 minutes and 45 minutes after the start of occlusion for rats and mice, respectively, reperfusion was initiated by removing the nylon monofilament. After surgery, we used Longa’s 5-point scale to select experimental animals with a score of 2–3 within 24 hours. In the sham group, rats underwent a similar surgery, but the filament was removed immediately after insertion (no occlusion and no reperfusion) (Luo et al., 2024).

Figure 1.

Diagram of experimental design.

hNSCs: Human neural stem cells.

Human neural stem cell culture

Primary human fetal NSCs were initially isolated from the telencephalon at 9–10 weeks of gestation, as previously described, with written informed consent (Svendsen et al., 1998). The procedure was approved by the Ethics Committee of Langfang 4th Hospital (approval No. 202001). Following the removal of meninges, the tissue was minced and repeated pipetting of cells was performed with a 5-mL pipette until a single-cell suspension was achieved. Debris was removed by centrifugation. The cells were plated in 5 × 10⁶ cells/T-75 uncoated flask and cultured as neurospheres in Stemline^TM Neural Stem Cell Expansion Medium (Sigma-Aldrich, Saint Louis, MO, USA, Cat# S3194) supplemented with B-27 supplement (Gibco, Grand Island, NY, USA, Cat# A3582801), N-2 supplement (Gibco, Cat# 17502048), 20 ng/mL of recombinant human epidermal growth factor (Invitrogen, Grand Island, NY, USA, Cat# AF-100-15-100UG), 20 ng/mL of recombinant basic fibroblast growth factor (Invitrogen, Cat# AF-100-18C-100UG), and 10 ng/mL of leukemia inhibitory factor (Chemicon, Darmstadt, Germany, Cat# LIF1010). Neurospheres appeared within 2 days and were grown for 7 days before being collected and dissociated. Cells were then characterized by flow cytometry (CytoFlex, Beckman Coulter, CA, USA, AS51394) using the NSC markers nestin (mouse monoclonal antibody, BD Biosciences, San Jose, CA, USA, Cat# 561230, RRID: AB_10562398), Sox2 (mouse monoclonal antibody, BD Biosciences, Cat# 561610, RRID: AB_10712763), and CD133 (mouse monoclonal antibody, Miltenyi Biotech, Bergish Gladbach, Germany, Cat# 130-113-186, RRID: AB_2726012). After induction of differentiation, the phenotype was determined by immunostaining with astrocyte marker glial fibrillary acidic protein (GFAP; rabbit monoclonal antibody, 1:200, Abcam, Cambridge, UK, Cat# ab33922, RRID: AB_732571), neuronal marker Tuj-1 (rabbit polyclonal antibody, 1:200, Abcam, Cat# ab18207, RRID: AB_444319), and oligodendrocyte marker O4 (mouse monoclonal antibody, 1:200, R&D systems, Minneapolis, MN, USA, Cat# MAB1326, RRID: AB_357617).

Cell transplantation

Intracerebral transplantation of hNSCs was performed stereotaxically at 7 days after tMCAO in rats and mice. For behavioral testing and in vivo differentiation experiments, the rats were injected with 5 µL of hNSCs (1 × 10⁵ cells/µL) or vehicle (1:9 with human albumin injection Grifols, Barcelona, Spain, Cat# 5822) and 5% glucose injection (Baxter, Granville, IL, USA, Cat# A6C0087). Using magnetic resonance imaging (MRI) scans, we injected 2 µL into the ischemic periphery of the right cortex (anteroposterior [AP] −4.0 mm, mediolateral [ML] +2.0 mm, dorsoventral [DV] −1.8 mm), and the other 3 µL into the ipsilateral side of the infarct striatum region (AP −0.4 mm, ML +4.0 mm, DV −5.5 mm) at a rate 1 µL per 3 minutes. Coordinates were relative to bregma (Paxinos and Watson, 1998) after the rats were anesthetized (see the part of “Stroke models”for anesthesia). For the host-to-graft experiment, the rats were grafted with hNSCs transduced with tracing or control lentivirus. The amount and location of cell transplantation were the same as in vivo differentiation experiments. For an ex vivo electrophysiology experiment, mice were grafted with green fluorescent protein (GFP)-transduced hNSCs in the infarct cortex and the ipsilateral striatum at a volume of 1 and 1.5 µL of cells, respectively, at a concentration of 100,000 cells/µL. The coordinates relative to bregma were: cortex: AP 0.5 mm, ML +1.5 mm, DV −1.5 mm; striatum: AP −0.1 mm, ML +2.5 mm, DV −3.2 mm (Paxinos and Franklin, 2013).

All animals were intraperitoneally immunosuppressed with cyclosporine A (10 mg/kg, Novartis, Basel, Switzerland) daily for 3 consecutive days before transplantation and for up to 3 weeks after transplantation, then were administered the same dose (Eaton et al., 2011) for up to 10 weeks.

Behavioral testing

Of the total 36 adult male rats, 10 were randomly selected for the sham group, and the other rats underwent 150-minute tMCAO stroke surgery and were evaluated for behavioral deficits 3 days before transplantation. Six animals died 1 day after tMCAO surgery. None of the animals died after cell grafting. According to the modified Neurological Severity Score (mNSS) scores and MRI results of the injured rats, the remaining 20 tMCAO/R rats were divided into a vehicle-treated group (n = 10) or hNSC-treated group (n = 10), with the mean and standard deviation of the mNSS scores and MRI results (lesion volume was evaluated as a percentage of the ipsilateral hemisphere using MRI) of the two groups being as similar as possible. The animals were assessed using a battery of tests at baseline before transplantation and at 1, 2, 4, 8, and 12 weeks after transplantation for their sensorimotor skills using the cylinder, adhesive removal, foot-fault, and mNSS tests.

Briefly, the mNSS is a comprehensive assessment of motor, sensory, balance, and reflex function with a scale of 0–18 (0, normal score; 18, maximal deficit score) in rodents, particularly in stroke models. The scoring system consists of a variety of test items, each with a specific score range. During the assessment, the animal’s response to each test was recorded. The cylinder test used transparent plexiglass cylinders, 18 cm in diameter and 30 cm in height, to assess each rat’s forelimb use. Over a period of 5 minutes, the number of times the rat’s left, right, or both forelimbs touched the cylinder’s sides were recorded. Following each trial, 75% alcohol was used to sanitize and eliminate odors. Only clear and definable actions were included in the data analysis. The analysis focused on assessing preferences, asymmetry, and deviations in forelimb usage, calculated using the formula: Predilection for forelimb use (%) = (number of right forelimb contacts – number of left forelimb contacts)/(number of left forelimb contacts + number of right forelimb contacts + number of double forelimb contacts) × 100. The adhesive removal test required that tape was uniformly placed on the contralateral forepaw of the rat with the same pressure. Subsequently, the time to perceive the tape (judged by whisker, mouth touch, or forelimb swinging) and the time to remove the tape were recorded. If the rat took more than 180 seconds to remove the tape, the time was recorded as 180 seconds. The test was repeated three times, with an interval of 10 minutes between each test. If the rat did not take any action during the experiment, resulting in a perception time of 180 seconds, the data were excluded from the analysis. The foot-fault test was performed using a 150 cm × 140 cm wire mesh with 2.3 cm × 2.3 cm gaps to assess the walking ability and coordination of the forelimbs of rats. During the experiment, three researchers each recorded the number of times the rat’s forelimbs slipped through the mesh and the total number of steps taken while on the grid. The experiment lasted for 2 minutes, and if the rat took fewer than 20 steps within this time, the data were excluded. A ratio was calculated using the formula: (missteps of contralateral forelimb – missteps of ipsilateral forelimb)/ total steps. A positive ratio indicated a deficiency in the function of the contralateral side due to brain lesion, and a negative ratio indicated a deficiency in the function of the ipsilateral side due to the lesion (Liu et al., 2018, 2019; Ruan and Yao, 2020).

Immunohistochemistry

NSC-grafted rats were perfused with 4% paraformaldehyde (G-clone, Beijing, China, Cat# PN4204) 12 weeks after grafting (n = 5). Their brains were fixed in 2% paraformaldehyde with 3% saccharose for 30 minutes, dehydrated with 30% saccharose overnight, embedded in cryomatrix, and cut into 15-µm coronal slices using a microtome (CM-1950, Leica, Heidelberg, Germany). To identify whether the transplanted hNSCs survived after 12 weeks and to determine their differentiation status within the brain, the brain slices (every eighth sections) were immunolabeled using the following primary antibodies: Stem121 (mouse monoclonal antibody, 1:500, Takara, Kyoto, Japan, Cat# Y40410, RRID: AB_2801314) for human cell labeling to assess grafted human cell bodies and double-labeling for neural cell markers, including Tuj-1 (rabbit polyclonal antibody, 1:200, Abcam, Cat# ab18207, RRID: AB_444319) for immature and mature neurons; microtubule associated protein 2 (MAP2; chicken polyclonal antibody, 1:200, Abcam, Cat# ab5392, RRID: AB_2138153) for mature neurons; NG2 (rabbit polyclonal antibody, 1:300, Abcam, Cat# ab83178, RRID: AB_10672215) to label oligodendrocytes, and GFAP (rabbit monoclonal antibody, 1:200, Abcam, Cat# ab33922, RRID: AB_732571) to label astrocytes. Choline acetyltransferase (ChAT; rabbit monoclonal antibody, 1:200, Abcam, Cat# ab181023, RRID: AB_2687983) was used for staining mature excitatory motor neurons, and gamma-aminobutyric acid (GABA; rabbit polyclonal antibody, 1:200, Sigma-Aldrich, Cat# A2052, RRID: AB_477652) for staining GABAergic inhibitory neurons. Microglia were labeled using Iba-1 (microglial marker; rabbit polyclonal antibody, 1:500, Wako Osaka, Japan, Cat# 010-19741, RRID: AB_839504), ED1 (M1 activated microglial marker; mouse monoclonal antibody, 1:500, Millipore, Billerica, MA, USA, Cat# MAB1435, RRID: AB_177576), and Arg1 (M2 microglial marker; goat polyclonal antibody, 1:200, Santa Cruz Biotechnology, Santa Cruz, CA, USA, Cat# sc-271430, RRID: AB_10648473). To evaluate angiogenesis in the infarct regions, we used von Willebrand factor (vWF; mouse monoclonal antibody, 1:100, Santa Cruz Biotechnology, Cat# sc-365712, RRID: AB_10842026) and Ki67 rabbit monoclonal antibody (1:500, Abcam, Cat# ab16667, RRID: AB_302459) to assess proliferating cells. The brain slices were incubated overnight at 4°C for primary antibodies and then incubated with the appropriate secondary antibodies: Alexa Fluor 488- (goat polyclonal antibody, 1:1500, Cat# 111-545-045, RRID: AB_2338049), 594- (goat polyclonal antibody, 1:1000, Cat# 115-585-166, RRID: AB_2338883), or 647- (goat polyclonal antibody, 1:1000, Cat# 111-605-003, RRID: AB_2338072) labeled IgG (all from Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 1 hour at 25 °C in the dark. Nuclei were stained with VECTASHIELD® Antifade Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA, USA, Cat# H-1500-10). The stained sections and cells were visualized under a fluorescence microscope (DM18, Leica, Heidelberg, Germany). To quantify the colocalization of total Stem121 or Iba to total (nuclear-labeled) cells with different markers, at least six randomly chosen areas per section were counted using ImageJ software (version 1.43, National Institutes of Health, Bethesda, MD, USA; Schneider et al., 2012).

Rabies virus monosynaptic tracing

Retrovirus and lentivirus were used to observe retrograde transsynaptic tracing from connections with host and transplanted neurons (Grealish et al., 2015; Tornero et al., 2017; Palma-Tortosa et al., 2020; Xiong et al., 2021). The tracing (LV-EF1α-TVA-P2A-oRVG-P2A-EGFP-CMV, BrainVTA, Wuhan, China, Cat# LV-0478) and control (rLV-EF1α-EGFP-CMV, BrainVTA, Cat# LV-0215) viruses were used at a titer of ≥ 1.00 × 10⁹ TU/mL. Both viruses were generated to express a nucleus-localized GFP. The dissociated single 1 × 10⁶ hNSCs were plated onto a 24-well plate and infected with GFP-containing lentivirus at a multiplicity of infection of 10 for 6 hours under standard culture conditions (37°C, 5% CO₂). At 7 days after cell harvesting, the transduction efficiency according to GFP-expressing cells was evaluated using a fluorescence microscope and flow cytometry (CytoFlex, AS51394, Beckman Coulter, Brea, CA, USA). Finally, hNSCs were stably transfected into the lentiviral tracing vector, which expressed a histone-tagged GFP, the TVA receptor necessary for selective infection of the modified rabies virus (∆G-rabies) and rabies glycoprotein (GP) to allow for transsynaptic spread. The control vector was used for unspecific labeling, which contained GFP and TVA, but lacked GP. At 7, 11, 15, and 23 weeks (n = 3 per time point) after cell transplantation, 2 µL ∆G-rabies virus (RV-ENVA-∆G-mcherry, titer ≥ 2.0 × 10⁸ infectious units/mL, BrainVTA, Cat# R01004) was injected at the same sites as the coordinates of cell transplantation (Palma-Tortosa et al., 2020; Xiong et al., 2021). One week after the injection of ∆G-rabies virus, rats were transcardially perfused with 4% paraformaldehyde. The ∆G-rabies with the foreign coat protein EnvA, in which the gene mCherry replaced the gene for GP in the viral genome, only infected cells with the TVA receptor. The grafted hNSCs expressing nuclear GFP as targeted cells were visualized by expression of GFP and mCherry after infection by ∆G-rabies. In contrast, the presynaptic neurons were identified by their expression of mCherry without GFP. Further propagation of ∆G-rabies was not possible as GP expression was restricted to the starter neurons, so only monosynaptic tracing occurred. All coronal sections with 40-μm thickness (1:5 series) without staining were imaged by a 20× objective with a fluorescence microscope (DM18, Leica, Heidelberg, Germany).

Ex vivo electrophysiology

The GFP-transduced hNSCs were transplanted into mice cortex and striatum 7 days after tMCAO. Four months later, the whole brain slices (200 μm) were prepared for electrophysiological analysis. The detailed protocol for electrophysiological recording was described previously (Palma-Tortosa et al., 2020; Xiong et al., 2021; Xiao et al., 2023). Briefly, a single slice was transferred to a recording chamber and submerged in a continuously flowing artificial cerebrospinal fluid (in mM: 135 NaCl, 5 KCl, 1.3 MgCl₂, 1.2 NaH₂PO₄, 2.6 CaCl₂, 10 glucose, and 26 NaHCO₃; pH 7.4) at 35°C. The recording electrodes were pulled (resistance of 3–5 MΩ) and filled with intracellular solution (in mM): 5 EGTA, 10 HEPES, 0.5 CaCl₂, 0.5 Na₂GTP, 5 MgATP, 12 phosphocreatine (pH 7.25, osmolality = 290) for current-clamp recording. Resting membrane potential was recorded immediately upon acquisition of whole-cell configuration. Neuronal membrane input resistance and evoked excitability were measured in current-clamp mode by 500-ms step-current injections from −30 pA to +200 pA with a step of 10 pA. Cells were held at −60 mV for spontaneous excitatory postsynaptic current recording and 0 mV for spontaneous inhibitory postsynaptic current recording. Recordings were performed from grafted corticostriatal neurons of the preinfarct zone. The first visible neuron was defined when the objective of the microscope was moved from the center of the ischemic region toward the periphery. Lastly, membrane current, current-induced action potentials, and spontaneous action potentials were monitored using a HEKA double patch-clamp EPC10 amplifier (HEKA, Stuttgart, Germany), and PatchMaster (HEKA) was used for data acquisition. Neurons exhibiting unstable abnormal spiking patterns or resting membrane potential in current-clamp mode were not included in the data analysis.

Magnetic resonance imaging

MRI scans were performed 7 days after the stroke and 12 weeks after cell transplantation in rats, as previously described (Tian et al., 2019; Donners et al., 2023). MRI (Biospec 94/30 preclinical system, Bruker, Karlsruhe Germany) data were acquired with the following parameters: field of view = 20 × 9 mm², repetition time/echo time = 3000/35 ms, matrix = 128 × 128 pixels, slice thickness = 0.16 mm, average = 6. The whole brain and lesion volume were measured using a manual segmentation method (Li et al., 2021; Lo et al., 2023).

Statistical analysis

All data were expressed as the mean ± standard error of the mean and analyzed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA, www.graphpad.com). Comparisons between multiple groups were analyzed by one-way analysis of variance followed by Tukey’s post hoc test. An unpaired two-tailed Student’s t-test was used for two-group comparisons. P < 0.05 was considered statistically significant.

Results

Growth and fate of human fetal neural stem cells expanded in spheres in vitro

Human fetal NSCs formed clusters of small, round cells that grew into floating spheres (Figure 2A). Flow cytometry analysis of the single cells dissociated from the spheres showed the cells were positive for CD133, nestin, and Sox2 (Figure 2B, C, and G). Immunofluorescence staining that was performed after 10 days of induced differentiation showed that the hNSCs generated cells of all three lineages: astrocytes (GFAP: 34.9% ± 2.1%), neurons (Tuj-1: 89.8% ± 4.6%), and oligodendrocytes (O4: 0.9% ± 0.5%; Figure 2D–F and 2H). Thus, human fetal NSCs showed multilineage differentiation potential in our culture system.

Figure 2.

hNSCs characterization in vitro.

(A) Morphology of hNSCs (the four ball structures with clear edges are the neurospheres aggregated by hNSCs). Photomicrograph was taken from culture at 7 days after neurospheres formed (scale bar: 100 µm). (B, C) Flow cytometry showed high levels of NSC markers (CD133, 91.4%; nestin, 98.7%; and Sox2, 98.83%). (D–F) Immunofluorescence images showed that the induced hNSCs were positive for GFAP (green), Tuj-1 (red) and O4 (red). Nuclear staining (DAPI, blue) is included in merged panel. Scale bars: 50 µm. (G) Bar graph showing the average percentage of cells positive for neural stem cell markers CD133, nestin, and Sox2 in flow cytometry plots (B, C). (H) Bar graph showing the average percentage of hNSCs expressing the three lineages in the immunofluorescence images (D–F): astrocytes (GFAP), neurons (Tuj-1), and oligodendrocytes (O4). Data are presented as the mean ± standard error of the mean (SEM). DAPI: 4’,6-Diamidino-2-phenylindole; GFAP: glial fibrillary acidic protein; hNSCs: human neural stem cells.

Transplanted human neural stem cells ameliorate the long-term neurological dysfunction and decrease the lesion volume in a stroke model

The timeline of the behavioral animal experiments was shown in Figure 3A. The animals were divided into three groups: sham, vehicle-treated and hNSCs-treated. Two weeks after the graft, when compared with the vehicle-treated rats, the hNSCs-treated rats already had improved mNSS function and were significantly better than the vehicle group over 3 months (Figure 3B; P < 0.05 at 2 and 4 weeks; P < 0.01 at 8 and 12 weeks). The adhesive removal test was used to assess motor function more precisely. The mean removal time for hNSCs-treated animals was significantly shorter than that for the vehicle-treated animals (Figure 3C; P < 0.05 at 2 and 8 weeks; P < 0.01 at 4 and 12 weeks), suggesting that the recovery of functional deficit was accelerated following hNSCs transplantation. We observed similar trends in both the cylinders test (Figure 3D; P < 0.05 at 2, 4, 8, and 12 weeks) and foot-fault test (Figure 3E; P < 0.05 at 2, 4, and 8 weeks; P < 0.001 at 12 weeks). Thus, the hNSCs-treated group demonstrated sustained recovery in a range of sensorimotor activities during a long-term period from 2 weeks to 12 weeks.

Figure 3.

Long-term behavioral functional recovery after hNSCs treatment in rats after tMCAO.

(A) The behavioral test pretraining was performed 3 days before tMCAO, and hNSCs were transplanted into the cortex and striatum regions of the rats’ brains at 7 days after stroke. Outcomes were assessed at the indicated intervals. (B–E) Functional recovery was evaluated using modified neurological severity score (mNSS) (B), adhesive removal test (C), cylinder test (D), and foot-fault test (E, at 2, 4, 8, and 12 weeks) at 3 days before tMCAO (pretraining), 0 days (as a baseline), and 1, 2, 4, 8, and 12 weeks after transplantation. *P < 0.05, **P < 0.01, ***P < 0.001, vs. hNSCs-treated group. Data are presented as the mean ± SEM; n = 10/group. hNSCs: Human neural stem cells; mNSS: modified neurological severity score; tMCAO: transient middle cerebral artery occlusion.

The infarct variation was tracked by MRI at 3 months before and after cell transplantation (Figure 4A–D). Quantitative measurement of lesion size with MRI showed that hNSCs treatment reduced lesion volume significantly compared with that in the vehicle group (Figure 4E; P = 0.0458). NSCs-treated animals demonstrated a significant reduction over 3 months in the infarct volume in comparison with vehicle animals. In contrast, the lesion volume of vehicle animals was increased by 7.253% ± 3.035% (Figure 4F; P = 0.0092), demonstrating that the vehicle animals worsened without drug treatment. Thus, our findings supported the results of previous studies that hNSCs treatment improved behavioral outcomes and reduced infarct volume in an animal model of cerebral ischemia.

Figure 4.

MRI of lesion volume after NSC treatment in rats after tMCAO.

(A–D) Representative schematic illustration of MRI coronal serial scans (1–6) for vehicle- and hNSCs-treated groups before transplantation (A, C) and 12 weeks post-transplantation (B, D) (red dashed lines show the border of the infarct area). (E) Changes in infarct volume before and 12 weeks post-transplantation in vehicle and hNSCs groups. Infarct volume was significantly different (P = 0.0458) between the NSC treatment and vehicle treatment groups. Relative infarct = infarct volume/right hemisphere volume. (F) Differences of the relative infarct volume before and 12 weeks post-transplantation in ischemic rats. Infarct size of the hNSCs group was significantly reduced compared with that of vehicle animals (P = 0.0092). Differences of the relative infarct = infarct volume (12 weeks post-transplantation − pre-transplantation) / right hemisphere volume. Data are presented as the mean ± SEM; n = 10/group; *P < 0.05, **P < 0.01. hNSCs: Human neural stem cells; MRI: magnetic resonance imaging; tMCAO: transient middle cerebral artery occlusion.

Survival and differentiation of human neural stem cells in the ischemic host brain

Immunohistochemical analysis was performed in the boundary zone of the ischemic core. Human NSCs transplanted into the rat striatum and cortex regions 7 days post-stroke showed robust survival (labeled with human-specific antibody Stem121, red) and extensive migration toward the ischemic lesions at 12 weeks post-transplantation (Figure 5A and B). The abundant hNSCs survival within the striatum and cortex lesions reflected a large graft volume (Figure 5A). Thus, the results indicated that many hNSCs were able to survive close to the lesion as long as they were transplanted into nonischemic tissue.

Figure 5.

Cell survival and differentiation of hNSCs in the ischemic brain in rats.

(A) A representative coronal section image showed good survival and migration of hNSCs (identified with human-specific antibody STEM121; red) in the peri-infarct region at 12 weeks post-transplantation (scale bar: 1000 µm). (B) Representative photomicrograph of the region of interest from (A) (scale bar: 100 µm). (C–H) Representative images of the different differentiation markers (MAP2, Tuj-1, NG2, GFAP) (green) and functional neurons (ChAT and GABA) (green) of transplanted hNSCs expressing the human-specific cytoplasmic marker STEM121 (red). Nuclear staining (DAPI, blue) is included in merged panel (scale bars: 100 µm). (I) Bar graph shows the differentiation rate of grafted hNSCs in ischemic rat brain. ChAT: Choline acetyltransferase; DAPI: 4′,6-diamidino-2-phenylindole; GABA: gama-aminobutyric acid; GFAP: glial fibrillary acidic protein; hNSCs: human neural stem cells; MAP2: microtubule associated protein 2; tMCAO: transient middle cerebral artery occlusion.

To assess the differentiation pattern of transplanted hNSCs, we used lineage-specific phenotypic markers and Stem121 double staining. The results showed that grafted NSCs (recognized with the human-specific marker Stem121) in the peri-infarct area were differentiated into MAP2-positive mature neurons, Tuj-1-positive neurons (including both immature and mature neurons), and NG2-positive oligodendrocytes, and residual cells were GFAP-positive glial cells (Figure 5C–F and I).

To examine the specificity of the functional neurotransmitter, double immunohistochemistry of the grafted cells with either ChAT or GABA was performed. The vast majority of grafted neurons were excitatory, expressing ChAT, and only a few individual cells exhibited the GABAergic inhibitory phenotype at the 12-week time point (Figure 5G–I).

Transplantation of human neural stem cells alters the M1/M2 phenotype of microglia in ischemic brain

Double staining of CD68 (a classical proinflammatory marker for M1 microglia) or Arg1 (an alternative anti-inflammatory marker for M2 microglia) and Iba1 was performed 12 weeks after transplantation to examine the possible effects of hNSC transplantation on the phenotypic transition of M1/M2 microglia. The results indicated that the expression of Arg1⁺/Iba1⁺ was significantly increased (P < 0.0001; Figure 6B and D), and the expression of ED1⁺/Iba1⁺ was significantly decreased (P < 0.0001) in the hNSCs-treated group compared with the vehicle group (Figure 6A and C). This suggested that NSC transplantation shifted the M1/M2 phenotype of microglia and enhanced neuroprotective function.

Figure 6.

hNSCs treatment changes the ratio of M1/M2 phenotype of microglia in injured brain.

(A, B) Representative fluorescence images of ED1 (red)/Iba1 (green)-positive cells (A) and of Arg1 (red) and Iba1 (green) staining (B) at 12 weeks after grafting compared with the vehicle group in the ischemic lesion. Nuclei were stained blue. Scale bars: 100 μm. (C) Percentage of ED1⁺/Iba1⁺ cells (n = 5/group). (D) Percentage of Arg1⁺/Iba1⁺ cells (n = 5/group). Data are presented as the mean ± SEM; ***P < 0.0001. DAPI: 4′,6-Diamidino-2-phenylindole (blue); hNSCs: human neural stem cells; Iba1: ionized calcium-binding adapter molecule 1.

Grafted human neural stem cell–derived neurons receive direct synaptic inputs from stroke-injured host brain

To analyze whether grafted neurons form connections with host neurons, a retrograde transsynaptic tracing strategy with ∆G-rabies was used (Callaway and Luo, 2015). The hNSCs were stably transduced with the lentiviral tracing vector, which expressed a histone-tagged GFP, the TVA receptor necessary for selective infection with ∆G-rabies, and rabies GP to allow for transsynaptic spread. The ∆G-rabies with the foreign coat protein EnvA, in which the gene mCherry replaced the gene for GP in the viral genome, infected only the cells with the TVA receptor (Figure 7A). Fluorescence imaging (Figure 7B) showed that GFP containing lentiviral tracing vector transduced hNSCs with a transduction efficiency of 99.07% (Figure 7C). After 7 days post-tMCAO/R, rats were implanted with GFP-transduced hNSCs adjacent to the injury in the cortex or striatum (Figure 7D). Grafted hNSCs expressing nuclear GFP were identified as starter neurons (Figure 7E). After infection by ∆G-rabies (Figure 7D), the targeted neurons were visualized by expression of GFP and mCherry (Figure 7F and G). In contrast, presynaptic traced neurons were identified by mCherry expression without GFP (Figure 7G).

Figure 7.

Grafted hNSCs receive direct monosynaptic inputs from host neurons in injured host tissue.

(A) Schematic representation of lentiviral tracing vector construct including histone-tagged GFP, TVA receptor, and rabies GP (left image), and representation of the map for monosynaptic tracing (right image). (B) Fluorescent images showed expression of GFP (green) in the hNSCs neurospheres. Scale bar: 50 µm. (C) Flow cytometry showed that the transduction efficiency of GFP lentiviral tracing vector was 99.07%. (D) Experimental approach for implantation of GFP⁺ hNSCs (green) in the rat cortex and striatum 7 days post-tMCAO/reperfusion, and injection of ∆G-rabies (red) at 7, 11, 15, and 23 weeks. (E–G) Transplanted cells as “starter” cells were detected based on GFP (green) expression in the rat cortex or striatum (E). ∆G-rabies with a clear mCherry⁺ (red) signal (F) selectively infected the starter neurons (G) (yellowish) (scale bars: 100 µm). Because the starter neurons contained GP, the “traced” neurons (mCherry⁺/GFP⁻, red) were easily distinguished from the starter neurons. Only presynaptic tracings were formed. GFP: Green fluorescent protein; GP: glycoprotein; hNSCs: human neural stem cells; tMCAO: transient middle cerebral artery occlusion.

Before tracing host-to-graft connectivity, the time points when grafted neurons were likely to receive functional inputs were initially identified. The spontaneous postsynaptic activity was observed from 6 weeks post-injection, as was previously reported (Grealish et al., 2015). Retrograde tracking evaluation of rat brain tissues was performed at 7, 11, 15, and 23 weeks after cell transplantation, followed by injection of ∆G-rabies 1 week later. Using fluorescence microscopy, we observed a large number of GFP⁺ nuclei that co-expressed mCherry within the core of the transplant while migrating to distant regions over time (Figure 8A–D). The majority of mCherry⁺/GFP⁻ traced neurons were adjacent to the graft core over time, and appeared in distant regions like the contralateral cortex, thalamus, and hippocampus (indicated by white dashed lines, Figure 8C and D, on weeks 16 and 24). These indirectly supported that grafted hNSC-derived neurons established widespread axonal projections to both ischemic lesions and the contralateral areas. The purpose of the control vector was to avoid unspecific labeling. The lack of GP in the control vector prevented it from transmitting across synapses. Thus, hNSCs infected with the control vector showed only mCherry⁺/GFP⁺ signals in vivo, no monosynaptic connections between host and grafted cells were observed (Figure 8E).

Figure 8.

Synaptic inputs from host neurons to grafted hNSCs remain after 24 weeks.

(A–D) Representative coronal section images showed the overview of the location of the graft core (mCherry⁺/GFP⁺) and distribution of mCherry⁺/GFP⁻ (red) traced neurons in a brain at 8 (n = 3), 12 (n = 3), 16 (n = 3) and 24 (n = 3) weeks post-grafting (indicated by green and white dashed lines). The majority of mCherry⁺/GFP⁻ traced neurons were found close to the graft core (indicated by green dashed lines). At 16 weeks after transplantation, traced mCherry⁺/GFP⁻ host neurons were detected in distal structures, such as contralateral cortex, thalamus (thal), and hippocampus (hip) (C, D) (indicated by white dashed lines). (E) The control vector lacked the GP and thus could not form monosynaptic contact. All GFP⁺ nuclei co-expressed mCherry (n = 3). Scale bars: 2000 µm in A–E. (F–H) The bar graphs show quantifications of starter cells (F), traced cells (G) and the ratio of traced to starter cells (H) did not differ across time points (P > 0.05, n = 3 for each time point, mean ± SEM). GFP: Green fluorescent protein; hNSCs: human neural stem cells.

At these time points, analysis of traced cells expressing only mCherry showed that these cells represented 12.44% ± 1.98% of starter cells at 8 weeks post-injection, and that the ratio of traced cells to starter cells remained constant at 12, 16, and 24 weeks post-injection (Figure 8H). Additionally, the number of starter cells (Figure 8F) and traced cells (Figure 8G) did not significantly increase between 8 and 24 weeks after transplantation (P > 0.05). Our finding was almost identical to the previous study that presynaptic contacts from the host brain to corticostriatal grafts with human NSCs were stable over time, as shown with rabies virus tracing (Grealish et al., 2015; Tornero et al., 2017).

Grafted human neural stem cell-derived neurons become functionally integrated into host neural circuitry

Finally, to determine whether cells generated from GFP-transfected hNSCs exhibited neuronal function, we examined the electrophysiological activity of cells using whole-cell patch-clamp recordings in acute brain slices at 16 weeks after transplantation into cortex and striatum adjacent to the lesion areas of tMCAO mice. We found that human GFP-positive cells in the cortex (Figure 9A and B) and striatum (Figure 9C and D) grafts generated current-induced action potentials and spontaneous action potentials (Figure 9E). Out of 113 GFP-positive cells identified from seven mice, approximately 72% of patched cells were functionally mature neurons with resting membrane potentials of approximately −57 mV and −73 mV in the cortex and striatum, respectively (Figure 9E and F), and produced spontaneous excitatory postsynaptic currents, suggesting functional maturation of grafted human neurons. Other cells with unstable, abnormal spiking patterns or resting membrane potentials (Figure 9G and H) in current-clamp mode were excluded from the analysis. Interestingly, no spontaneous inhibitory postsynaptic currents were observed.

Figure 9.

Electrophysiological properties of functional neurons in acute brain slices of rat cortex and striatum.

(A–D) Whole-cell patch-clamp recording in the cortex (B) and striatum (D) from GFP-positive cells. Photomicrograph of grafted GFP-positive hNSCs in the cortex (A) and striatum (C). Scale bars: 20 µm. (E) Typical traces of spontaneous action potentials (sAPs) in host endogenous mature neurons or cortical or striatal grafted hNSC-derived neurons 4 months after transplantation. (F) Typical traces of spontaneous excitatory postsynaptic currents (sEPSCs) in host endogenous mature neurons or cortical or striatal grafted human NSC-derived neurons 4 months after transplantation. (G, H) Typical traces of sAPs and sEPSCs of less mature neuronal or nonneuronal phenotypes. GFP: Green fluorescent protein; hNSCs: human neural stem cells; sAPs: spontaneous action potentials; sEPSC:spontaneous excitatory post-synaptic current.

Discussion

Previous studies of cell treatment after tMCAO in animals have shown improved functional recovery, reduced infarct size (Huang et al., 2014; Tian et al., 2019; Chang et al., 2024), transplanted cell survival, enhanced plasticity (Boese et al., 2020), differentiation into neuronal and glial cells, and established axonal projections into the host brain (Callaway and Luo 2015; De Gioia et al., 2020). Our study confirmed and extended the previous findings. Baker et al. (2019) reported that most preclinical rodent models show rapid and spontaneous functional recovery, making it difficult to determine whether the functional improvement 2 or 3 months after transplantation is related to cell integration. Therefore, we constructed a severe tMCAO rat model with 150 minutes of ischemic injury to study the long-term behavioral recovery capacity of NSCs after transplantation. Emerging preclinical studies have focused on 7 days post-ischemic stroke as the optimal time for transplantation; this time point, compared with the acute phase (less than 3 days), better avoids inflammatory responses, decreases tissue damage or even death, and creates a more favorable graft environment (Kelly et al., 2004; Bacigaluppi et al., 2009; Jin et al., 2010). Here, using hNSCs transplantation 7 days after stroke, we showed enhanced long-term functional recovery of adult animals with a large infarct size (approximately 35%) for up to 12 weeks, electrophysiological evidence of functional neuronal differentiation for up to 16 weeks, and synaptic input from host-to-graft neurons over time up to 24 weeks, which may be critical for successful integration of stem cells.

Serial MRI was used to assess the long-term effect of hNSCs treatment, which provided a high spatial resolution to identify and evaluate the damaged region for each group. Our MRI results supported previous reports that hNSCs treatment reduced infarction volume at 12 weeks after stroke (Huang et al., 2014; Tian et al., 2019; Chang et al., 2024), whereas vehicle animals showed increased lesion volume without drug treatment. Moreover, hNSCs-treated rats had a significantly improved functional recovery compared with that of vehicle-treated rats in behavioral tests. These findings align with other studies showing that NSC transplantation promotes structural repair and functional recovery in stroke animal models (McGinley et al., 2018; Shin et al., 2018; Tian et al., 2019; De Gioia et al., 2020).

In the present study, functional improvement was evident 2 weeks after implantation, with earlier behavioral recovery in rats, and sustained improvement was observed for over 3 months. Thus, in the early stage after cell transplantation, motor function of the rats recovered more quickly than that of the control group. This may be due to trophic factors secreted from the transplanted hNSCs, which would suggest that the paracrine function of cells is very important, especially in the initial stage of cell transplantation (Huang and Zhang 2019). To clarify this issue, we further demonstrated that exogenous hNSC transplantation promoted endogenous angiogenesis (Additional Figure 1 ^{(2MB, tif)} ), survival of transplanted cells (Additional Figure 2 ^{(2.2MB, tif)} C) and ameliorated the inflammation microenviroment by downregulating the expression of proinflammatory factor M1 and upregulating the expression of anti-inflammatory factor M2. These findings indicate that the transplanted NSCs play a beneficial role through a paracrine effect (Rajan et al., 2019). A previous study has shown that animal motor and sensory functions may also recover when the lesion volume does not change, which is also related to paracrine effects (Daadi et al., 2010). However, the paracrine effect caused by transplanted hNSCs after cerebral ischemia involves many aspects, such as promoting endogenous neurogenesis and reducing endogenous neuronal apoptosis (Dabrowski et al., 2019). We are investigating this in further studies by knocking down important secretion factors of hNSCs.

The other purpose of exogenous NSC transplantation is to replace dead nerve cells and rebuild new neural circuits, so as to achieve long-term effects in vivo. The unique ability of NSCs is their potential for differentiation into multiple cell types, mature neurons, astrocytes, and oligodendrocytes (Kaminska et al., 2022; Santos et al., 2022), which are the cell types lost after cerebral infarction, resulting in behavioral dysfunction. Therefore, our study investigated whether NSCs can differentiate in vivo after transplantation to replace lost cell types in a stroke-injured brain. At 12 weeks after transplantation, we observed long-term survival of hNSCs in the ischemic brain of rats, which were differentiated mainly into mature neurons, less than 10% oligodendrocytes (NG2), and virtually no astrocytes (GFAP) (Kelly et al., 2004; Darsalia et al., 2007; Trotter et al., 2010). NSCs can differentiate into GFAP-positive astrocytes in vitro, thus, cell differentiation may be influenced by the microenvironment. Most grafts preferentially differentiated into neurons, not astrocytes in vivo, which would hinder glial scar formation and promote nerve regeneration. Moreover, we also observed that the majority of grafted cells differentiated into neurons with functional cholinergic transmission and only a few with GABAergic synaptic transmission, consistent with previous reports (Thonhoff et al., 2009; Palma-Tortosa et al., 2020; Park et al., 2020). Consistent with previous findings, transplanted hNSCs gave rise to neurons that migrated toward the ischemic lesion in rodents, further increasing the possibility of cell replacement (Kelly et al., 2004; Mine et al., 2013). Furthermore, hNSCs survived for at least 6 months without tumor formation (Lu et al., 2017; Palma-Tortosa et al., 2020; Xiong et al., 2021), and the percentage of Ki67-positive cells was less than 2% (Additional Figure 2 ^{(2.2MB, tif)} A and B).

We then investigated whether the differentiated neurons generated synaptic connections with host neurons and exhibited fundamental neuronal functions. Using retrograde neuronal tracing (Grealish et al., 2015; Tornero et al., 2017), we observed direct synaptic inputs from host neurons to grafted hNSCs, confirming that host-to-graft connectivity was stable between 8 and 24 weeks after transplantation. The vast majority of GFP⁺/mCherry⁺ starter neurons (representing transplanted cells) in rats were observed not only near the transplant core but also in distant areas of the injured brain, which was likely caused by cell migration. Further, retrograde labeling of mCherry⁺ neurons that did not co-express GFP (thus representing host cells that formed the monosynaptic host-to-graft connectivity) were observed near the graft core over time, and appeared in distant areas from the graft, such as the contralateral cortex, thalamus, and hippocampus. Furthermore, none of these findings were observed in animals with the control vector. Thus, the findings suggest that transplanted hNSCs sent widespread axonal projections to both ischemic lesions and the contralateral areas, established monosynaptic afferent inputs from host neurons, and repaired monosynaptic neuronal circuits, thereby replacing the damaged host neurons and integrating into the host’s existing neuronal network. These data were consistent with previous reports and indicated that grafted hNSCs with long-term survival could not only efficiently differentiate into mature neurons in vivo (Palma-Tortosa et al., 2020), but also receive stable synaptic input from host neurons. Additionally, at about 16 weeks after transplantation into the cortex and striatum, grafted cells exhibited electrophysiological properties and developed excitatory postsynaptic currents, consistent with physiologic neuronal properties and indicating integration into host neural circuitry (Oki et al., 2012; Lu et al., 2017).

This study has some limitations that should be noted. First, long-term functional improvement after transplantation may overlap with the natural recovery ability of animals, making it difficult to distinguish whether improvement is the result of cell therapy or the natural recovery process. Second, there are differences between the pathophysiological processes after stroke in rodent models and humans, and they may not be able to fully simulate the complex conditions after human stroke. In addition, although we observed that transplanted hNSCs migrated to areas away from the transplant core, the specific mechanisms of migration (including chemotaxis, cell-to-cell interactions, and microenvironmental factors) are not yet fully understood. This limits the ability to further regulate and optimize cell migration after transplantation.

In conclusion, we aimed to comprehensively investigate the behavioral recovery of rats in a severe tMCAO model through long-term observation after hNSCs grafting, as well as the proliferation, differentiation, angiogenesis, paracrine, and replacement effects in vivo. We showed that transplanted hNSCs survived long-term, differentiated into a high proportion of functional mature neurons with electrophysiological properties, and exhibited functional synaptic integration into the existing neuronal network. These findings suggest that the transplanted cells replaced the damaged host neurons, and provide evidence that neuronal replacement in the stroke-injured brain achieves therapeutic effects of long-term behavioral functional recovery. These findings provide insight for the effectiveness of hNSC transplantation as a treatment approach for ischemic stroke with long-term safety, and support that hNSC treatment may be a beneficial method for clinical application in cerebral infarction.

Additional files:

Additional Figure 1 ^{(2MB, tif)} : hNSC transplantation enhances angiogenesis after focal ischemic stroke.

Additional Figure 1

hNSC transplantation enhances angiogenesis after focal ischemic stroke.

Angiogenesis, as indicated by vWF staining (red) in infarct regions, was obviously augmented in NSCs-treated rats versus vehicle-treated rats at 12 weeks after stroke. Nuclei were stained blue. Scale bar: 100 μm. (B) Bar graph illustrating the number of vWF-positive microvessels in hNSCs-treated group was sixfold higher compared with that of the vehicle group at 12 weeks after stroke, indicating that transplanted hNSCs promoted angiogenesis in the ischemic region after stroke. Data are expressed as the mean ± SEM (n = 5/group), ^***P < 0.0001. DAPI: 4’,6-Diamidino-2-phenylindole; hNSCs: human neural stem cells; vWF: von Willebrand factor.

Additional Figure 2 ^{(2.2MB, tif)} : Ki67 immunolabeling and the survival status of graft cells in the ischemic brain.

Additional Figure 2

Ki67 immunolabeling and the survival status of graft cells in the ischemic brain.

(A, B) Representative images of coimmunostaining of cell proliferating marker Ki67 (green) and Stem121⁺ graft cells (red) decreased at later time points after tMCAO induction. (C) The estimated total number of surviving hNSCs in the striatum or cortex regions (above 50% of the initial cells) of stroke-injured rat brain at 12 weeks after grafting. Data are presented as the mean ± SEM (n = 5). DAPI: 4’,6-Diamidino-2-phenylindole; hNSCs: human neural stem cells; tMCAO: transient middle cerebral artery occlusion.

Data availability statement:

All data in this study are available within the manuscript and its Additional files.