Repair of infarcted tissue via transplanting human cerebral organoids

To further test the long-term differentiation of grafted human cerebral organoids in the cortex of stroke mice, we detected the cellular composition of cerebral organoids in the cortex of stroke mice on day 180 after transplantation. As shown in Supplementary Fig. 7, the proportion of HN⁺ cells expressing TBR1, CTIP2 and SATB2 was 19.0%, 42.9% and 44.1% (Supplementary Fig. 7a–c, j), respectively, and the proportion of HN⁺ cells expressing NeuN, DCX (an immature neuron marker), Ki67, GFAP, oligodendrocyte transcription factor 2 (OLIG2) and GABA was 44.5%, 43.9%, 0.6%, 6.2%, 4.6% and 2.1% respectively (Supplementary Fig. 7d–I, j). We analyzed the source of OLIG2 in the grafts and found that OLIG2⁺/HN⁺ and OLIG2⁺/HN⁻ cells were 92.7% and 7.3% respectively (Supplementary Fig. 7k–l), revealing that oligodendrocytes are mainly derived from grafted human cerebral organoids. These data suggest that grafted human cerebral organoids are mainly composed of mature and immature glutaminergic neurons in the infarcted cortex, with a small number of astrocytes, oligodendrocytes and GABAergic neurons 180 d after transplantation.

Cell therapy with single neural stem cell is difficult to repair the tissue structure in the infarcted core^31,32. To test whether the grafted human cerebral organoids can repair infarcted tissue, we transplanted the human cerebral organoids transduced with LV-EF1a-EGFP into the junction of the infarct core and the peri-infarct zone of stroke mice on day 7 after stroke and assessed the long-term survival and differentiation of grafts in the infarct core on day 180 after transplantation (Fig. ​(Fig.2a).2a). We observed that the infarcted core was fully filled by grafted organoids, with astrocytes labeled by STEM123⁺ (specifically with GFAP of human cells marker) (Fig. ​(Fig.2b)2b) and mature neurons labeled by NeuN (Fig. ​(Fig.2c),2c), suggesting long-term survival of organoids in the infarcted core. To determine whether the transplantation of dissociated single cells from organoids can repair infarcted tissue-like organoids transplantation, we prepared the dissociated single cells from equal numbers of organoids and transplanted them into the junction area as above. Although the total number of cells was comparable and the location of transplants was the same between the single cells and organoids transplantation, only a few cells survived at the infarct margin on day 180 after the dissociated organoids cells transplantation, and in the peri-infarct area, although there were many GFP⁺ fibers, few GFP⁺ neurons were observed (Supplementary Fig. 8). Thus, organoids but not dissociated organoids cells transplantation can repair the infarcted tissue after stroke.

Open in a separate window

Fig. 2

Survival and differentiation and functions of grafted human cerebral organoids in the infarct core of stroke mice.

a Design of the experiments for b–m. Representative images showing the coronal section stained with STEM123 and GFAP in animal A (b) or with NeuN and GFAP in animal B (c) on d 180 after organoids transplantation, in which, b1 and c1 showing infarct area (scale bar, 200 μm), b2 and c2 showing the survival of organoids (scale bar, 200 μm), the boxed area in b2 and c2 was shown in high magnification in b3 and c3 respectively (scale bar, 50 μm), the boxed area in b3 and c3 was shown in high magnification in b4 and c4 respectively (scale bar, 5 μm). Representative images showing HN⁺ graft cells co-labeled with NeuN (d) or GFAP (e) (left, scale bar, 200 μm) on d 180 after transplantation. The boxed area in leftward image was shown in high magnification on the middle image (scale bar, 50 μm), and the boxed area in middle image was shown in high magnification on the right three panels (scale bar, 5 μm). f–h Bar graphs show cerebral organoids’ cellular composition at 180 d after transplantation. n=6 animals. More than 4,500 cells from random fields were manually counted in each condition. i Representative image showing organoids neurons (GFP⁺) and astrocytes (GFAP⁺) (upper, scale bar, 50 μm) and high magnification images from a selected area in the upper image (lower, scale bar, 20 μm) at 180 d after transplantation. j Images showing whole-cell patch-clamp recordings from organoids neurons (GFP⁺). Black dotted lines indicate electrode. Scale bar, 40 μm. k Voltage traces showing APs from GFP⁺ neuron in response to current-step, in which, red traces showing the first AP induced by 0 pA injected current. l Number of APs evoked by various current steps from GFP⁺ neurons. n=13 neurons from 5 mice. m Representative traces of sPSCs from the organoids neurons. n=34 neurons from 7 mice. In b–e, the area outlined by the yellow dotted line indicates the infarct core. In f–h and l, data were presented as mean±standard deviation. HO Hoechst 33258 dye, HN Human Nuclei.

To assess the long-term differentiation of human cerebral organoids in the infarcted core, immunofluorescence staining was performed. As shown in Fig. ​Fig.2d,2d, ​d,e,e, a large number of HN⁺ organoids cells were co-labeled by GFAP or NeuN. Approximately 70% of the cells in the infarcted core were HN-positive, and the percentage of HN⁺ cells expressing GAFP and NeuN was 65.7% and 29.6% respectively (Fig. ​(Fig.2f),2f), suggesting that the cerebral organoids in the infarcted core are more likely to differentiate into glial cells, different from the neuronal differentiation of the grafts in the peri-infarct area. The source analysis of GFAP- and NeuN-positive cells in the infarcted core showed that GFAP⁺/HN⁺ and GFAP⁺/HN⁻ cells were 53.1% and 46.9% respectively, and NeuN⁺/HN⁺ and NeuN⁺/HN⁻ cells was 90.4% and 9.6% respectively (Fig. ​(Fig.2g,2g, ​g,h),h), indicating that more than half of astrocytes are derived from the host, while neurons are predominantly derived from grafted human cerebral organoids. However, in the neural induction medium (NIM)-treated stroke mice, the infarcted core had become a cavity with few cells surviving on day 180 after NIM microinjection (Supplementary Fig. 9).

To test whether the organoids-derived neurons in the infarcted core have a normal physiological function, we detected their electrophysiological characteristics on day 180 after transplantation (Fig. ​(Fig.2a,2a, 2i–m). The electrophysiological recordings showed that the EGFP⁺ organoids neurons in the infarcted core generated action potentials (APs) in response to membrane depolarization induced by current-step (Fig. 2i–l) and spontaneous postsynaptic currents (sPSCs) (Fig. ​(Fig.2m),2m), suggesting that grafted cerebral organoids neurons exhibit electrophysiological properties of mature neurons.

Integration of grafted cerebral organoids into the host neural circuits

The mammalian neocortex is responsible for processing multiple modalities of sensory information and controlling motor output. Glutamatergic projection neurons in the cortex send axons to distant brain targets³³, in which, callosal projection neurons primarily located in layers II/III send axons to the contralateral cortex, intrahemispheric projection neurons located in layer IV send axons within a single cortical, corticostriatal projection neuron located in layer V send axons to contralateral and ipsilateral striatum, corticospinal motor neurons located in layer V send axons to spinal cord and corticothalamic projection neurons located in layer VI send axons to thalamus³⁴. On day 180 after transplantation, massive of GFP⁺ nerve fibers appeared in the contralateral and ipsilateral corpus callosum (CC), striatum and hippocampus, ipsilateral internal capsule (IC), sensory cortex (S2) and ventral posterior thalamic nucleus (VP), suggesting that the grafted human cerebral organoids can establish commissural, associative and corticofugal projections in host brain, similar to the neocortex of mammals (Fig. ​(Fig.3a).3a). More interestingly, the GFP⁺ axon bundles from the grafted human cerebral organoid passed through the CC to the striatum, and then passed through the IC and the cerebral peduncle (CP) to the brainstem (Fig. ​(Fig.3b),3b), suggesting the establishment of the corticospinal tract (CST), a crucial circuit for the recovery of motor function from stroke. In addition, GFP⁺ fibers were also observed in the spinal trigeminal nucleus (Sp5), a somatosensory nucleus.

Open in a separate window

Fig. 3

Grafted cerebral organoids extend long-distance projections to specific brain areas in stroke mice 180 days after transplantation.

a Schematic showing the GFP⁺ grafts in the infarct core and peri-infarct area on the coronal sections of brain (middle), and images showing graft-derived projections in the contralateral and ipsilateral corpus callosum (CC), hippocampus (Hip) (upper) and striatum (Str), ipsilateral S2 and internal capsule (IP). Scale bar, 50 μm. b Schematic showing the GFP⁺ grafts in the infarct core and peri-infarct area on the sagittal sections of brain (middle), and images showing graft-derived projections in the CC, and striatum, IC, cerebral peduncle (CP), ventral posterior thalamic nucleus (VP), spinal trigeminal nucleus (Sp5) and brainstem. Scale bar, 20 μm.

To determine whether the neurons of grafted organoids receive the synaptic afferent from the host neuron and integrate into the host neural circuit, we microinjected an AAV-syn-hChR2(E123A)-mCherry into the adjacent area of the grafts on day 150 after transplantation, and 30 days later, we performed electrophysiological recordings (Fig. ​(Fig.4a).4a). Immunostaining showed that less than 3% of mCherry⁺ cells were HN-positive (mCherry⁺/HN⁺) (Fig. ​(Fig.4b,4b, Supplementary Fig. 10), indicating that almost all infected cells were host cells. Next, we illuminated the cortex in the acute slices with blue light (470nm) and found that trains of short pulses of blue light-induced AP firing in ChR2-expressing neurons in current-clamp mode, and the AP firing probability followed the frequency of photostimulation, validating the functionality of expressed ChR2 (Fig. ​(Fig.4c).4c). As shown in Fig. ​Fig.4d,4d, EFGP⁺ organoids neurons were surrounded by abundant mCherry⁺ fibers in the cortex, suggesting possible interactions between organoids neurons and host neurons. Next, we recorded postsynaptic currents (PSCs) in the GFP⁺ neurons by photostimulation of the mCherry⁺ fibers from the host neurons (Fig. ​(Fig.4e)4e) and found that the illumination resulted in inward currents in GFP⁺ neurons under voltage-clamp recording conditions at a holding potential of −65 mV (Fig. ​(Fig.4f)4f) and the PSCs were blocked by the application of ionotropic glutamate receptor antagonist kynurenic acid³⁵ (Fig. ​(Fig.4g),4g), suggesting a glutaminergic synaptic afferent from host neurons to the neurons of grafted organoids.

Open in a separate window

Fig. 4

The grafted organoid neurons receive synaptic input from host cells and synaptic efferent from the organoid neurons to distant host neurons.

a Design of the experiments for b–g. b A representative image showing that mCherry⁺ cells were not co-labeled with HN (left, scale bar, 20 μm) and high magnification images from a selected area in the leftward image (right, scale bar, 5 μm). c Diagram (left) showing the strategy to evoke APs in the host neuron and APs traces evoked by blue photo stimuli pulses (470nm, 10mW/mm², 5ms duration) of various frequencies on the ChR2-mCherry-expressing host neuron (right). d A representative image showing mCherry⁺ fibers extending to GFP⁺/HN⁺ human organoids neurons (left, scale bar, 20 μm) and high magnification images from a selected area in the leftward image (right, scale bar, 5 μm). e Diagram (left) showing the strategy to test whether organoids neurons receive synaptic input from host neurons and images (right) showing whole-cell patch-clamp recordings from grafted organoids neurons (GFP⁺). Black dotted lines indicate the placement of patch-clamp electrode. Scale bar, 20 μm. f Representative traces of PSCs evoked by 1 or 3 blue photo stimuli pulses at a holding potential −65 mV. g Representative traces of PSCs evoked by 1 blue photo stimuli pulse with or without kynurenic acid (α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) and N-methyl-D-aspartate (NMDA) receptors blocker). h Design of the experiments for i and j. i, j Cerebral organoids expressing ChR2 were grafted into the junction area between infarct core and peri-infarct zone of stroke mice and blue light-induced LFPs were recorded in the contralateral motor cortex and ipsilateral striatum. i Representative LFP traces recorded in the contralateral motor cortex (left, average of 20 responses) and input-output curves of peak amplitude for LFPs (right) (n=7 slices from 3 animals). j Representative LFP traces recorded in the ipsilateral striatum (left, average of 20 responses) and input-output curves of peak amplitude for LFPs (right) (n=6 slices from 3 animals). In f and g, n=13 neurons from 5 mice, 6 recorded neurons were responsive to photostimulation. In i and j, data were presented as mean±standard deviation. HO Hoechst 33258 dye, HN Human Nuclei.

Next, we investigated whether there is synaptic efferent from the organoids neurons to host neurons. We generated human cerebral organoids from a previously reported hESC line expressing ChR2 fused with EYFP²⁹, named H9-CAG-ChR2-EYFP cell line, and transplanted the organoids into the junction of the infarct core and the peri-infarct zone of stroke mice, and 180 days later, performed electrophysiological recordings (Supplementary Fig. 11a). Immunostaining showed that EYFP⁺ cells were HN-positive, indicating that they are grafted organoids-derived cells (Supplementary Fig. 11b). Trains of short pulses of blue light-induced AP firing in ChR2-expressing neurons in current-clamp mode, and the AP firing probability followed the frequency of photostimulation, validating the functionality of expressed ChR2 (Supplementary Fig. 11c). HN⁻ host neurons were surrounded by abundant GFP⁺ fibers in the cortex, suggesting possible interactions between organoids neurons and host neurons (Supplementary Fig. 11d). Next, we recorded PSCs in the GFP⁻ host neurons by photostimulating the GFP⁺ fibers from the organoids neurons and found that the illumination resulted in inward currents in GFP⁻ host neurons under voltage-clamp recording conditions at a holding potential of −65 mV and the PSCs were blocked by the application of ionotropic glutamate receptor antagonist kynurenic acid (Supplementary Fig. 11e–g), suggesting a glutaminergic synaptic efferent from the organoids neurons to host neurons. To confirm that the cells recorded electrophysiological signals are host cells but not organoids cells, we infused biocytin into the recorded neuron through an electrode after patch-clamp recording. Immunofluorescence staining showed that biocytin-filled cells did not express EYFP and HN, indicating that they are host neurons but not organoids neurons (Supplementary Fig. 11h).

To further demonstrate the synaptic efferent from the organoids neurons to host neurons in distant targets, we transplanted the organoids derived from H9-CAG-ChR2-EYFP cell line into the junction of the infarct core and the peri-infarct zone of stroke mice, and on day 180 after transplantation, we induced local field potentials (LFPs) in the contralateral cortex and the ipsilateral striatum by photostimulating the fibers projected from the grafted organoids (Fig. ​(Fig.4h).4h). Because of no GFP⁺ neurons in the contralateral cortex and the ipsilateral striatum (Fig. ​(Fig.3),3), the recordings of LFPs in these areas absolutely reflected the activity of host neurons receiving projections from grafted organoids neurons. As shown in Fig. ​Fig.4i,4i, ​i,j,j, the illumination of GFP⁺ fibers in both the contralateral cortex and the ipsilateral striatum induced LPFs dose-dependently. Therefore, there is synaptic efferent from the organoids neurons to host neurons, including local and distant targets.

Human pluripotent stem cells culture

H9-CAG-ChR2-EYFP cell line (a gift from Dr. Su-Chun Zhang, University of Wisconsin-Madison, Madison, USA), hESCs (line H9, WiCell Agreement No. 16-W0060) and hiPSCs (IMR90–4, WiCell Agreement No.17-W0063) were cultured on xeno-free conditions in the Essential 8 medium and passaged every 5–7 days.

Generation of human cerebral organoids

Generation of human cerebral organoids was performed based on the previously reported protocols^29,30. Briefly, hiPSCs or hESCs were detached by using dispase to form embryonic bodies (EBs) on day 0 and cultured in the neural induction medium (NIM; DMEM/F12, 1%NEAA, 1% N2). Half of NIM was changed every day from day 2 to day 6. Then, EBs were attached on 6-well plates for the rosette-like neuroepithelial cells (NE) formation from day 7 to day 15. On day 16, the rosettes were gently blown off by a 1ml pipette tip and resuspended in NIM containing B27. On day 20, cerebral organoids with multiple neural tube structures were obtained. The medium was refreshed every 5–6 days.

Lentivirus infection of cerebral organoids

Cerebral organoids were collected and placed in a 1.5ml centrifuge tube, each tube containing about 10 cerebral organoids. In our experiment, there were two kinds of lentivirus (LV) infection solution: solution 1 containing 15μl LV-EF1a-EGFP, 100μl NIM, 0.2μl polybrene and 2.3μl B27; solution 2 containing 30μl rLV-EF1a-hM4D(Gi)-mCherry-WPRE, 100μl NIM, 0.2μl polybrene and 2.6μl B27. Both A and B are virus infection enhancers. The prepared lentivirus infection solutions were added into two 1.5ml centrifuge tubes containing cerebral organoids and incubated at 37°C for 2h, and then, transferred to the T12.5 cell culture flask, added to 4ml NIM. The organoids were transplanted at 10 d after infection with Lentivirus.

Photothrombotic stroke model and organoids or dispersed organoids cells transplantation

Focal cortical stroke was induced in adult male mice through photothrombosis using a well-established procedure^6,45. In brief, under 2% isoflurane anesthesia (RWD Life Science), mice were fixed in a stereotaxic apparatus (Model 902, David Kopf Instruments) and their skulls were exposed through a midline incision, cleared of connective tissue, and dried. A cold light source (Z-LITE-Z, World Precision Instruments) attached to a custom-made opaque template giving a 2mm diameter spot was precisely positioned at two sites (PT-1 and PT-2) independently, following stereotactic coordinates (relative to bregma): (PT-1) anterior-posterior = 0mm, medial-lateral=+ 1.5mm, dorsal-ventral = 0mm; (PT-2) anterior-posterior=−1.9mm, medial-lateral=+ 1.5mm, dorsal-ventral = 0mm. Rose bengal solution (100mg/kg, i.p.; Sigma-Aldrich) was intraperitoneally injected, 5min later, the brain was illuminated with the intensity of 18500 Lux through the exposed intact skull for 15min. During illumination, rose bengal generated singlet oxygen leading to vascular endothelium damage and occlusion resulting in focal cortical stroke. Following illumination, the skin was surgically glued and the mice were allowed to recover. Control mice received the same surgery and the same dose of rose bengal but did not receive illumination.

One week later, one organoid (~600 μm in diameter and ~70,000 cells in total) or three organoids (~600 μm in diameter and ~200,000 cells in total), single cells (~200,000 cells) dissociated from 3 organoids of similar size with TrypLE, or 3μl NIM were microinjected into the junction of the infarct core and the peri-infarct zone, using a borosilicate glass capillary. For the microinjections of three organoids, single cells or NIM, the microinjection sites were left, right and inferior to the infarcted core, respectively, and following stereotactic coordinates (relative to bregma): (1) anterior-posterior = 0mm, medial-lateral=+ 0.7mm, dorsal-ventral=−1.5mm; (2) anterior-posterior = 0mm, medial-lateral=+ 1.5mm, dorsal-ventral=−1.7mm; (3) anterior-posterior = 0mm, medial-lateral=+ 2.5mm; dorsal-ventral=−2.0mm. For the microinjections of one organoid, the microinjection site was inferior to the infarcted core following stereotactic coordinates (relative to bregma): anterior-posterior = 0mm, medial-lateral=+ 1.5mm; dorsal-ventral=−1.7mm. Core body temperature was measured by a rectal probe and maintained at 37°C throughout the surgery.

Assessment of infarct volume

Mice were sacrificed under an overdose of chloral hydrate anesthesia (400mg/kg, i.p.) and brains were removed. 1mm coronal slices were made with a vibrating blade microtome (VT1200s, Leica) in ice-cold PBS. Then, Brain slices were incubated with 2% 2,3,5-triphenyl tetrazolium chloride (TTC) at 37°C for 20min. Brain sections were captured with a microscope and infarct volume was calculated with Image-J software.

Recombinant virus and stereotaxic microinjection

Recombinant adeno-associated virus (AAV) AAV-syn-hChR2(E123A)-mCherry (5×10¹² virus particles/ml) was purchased from GeneChem Co., Ltd (Shanghai, China). Recombinant lentivirus rLV-EF1a-hM4D(Gi)-mCherry-WRPE (5×10¹² virus particles/ml) were purchased from BrainVTA Co., Ltd (WuHan, China). Lentivirus LV-EF1a-EGFP (5×10⁹ virus particles/ml) were purchased from Nantong Gadgetzan Co., Ltd (NanTong, China) and stored at −80°C freezer until the day of infusion.

Mice were anesthetized with isoflurane gas (2% induction; 1% maintenance) mixed in oxygen and fixed on a stereotaxic apparatus. A small craniotomy was made targeting the motor cortex with stereotaxic coordinates (anterior-posterior = 0mm, medial-lateral=+ 2.0mm, dorsal-ventral=−2.0mm, relative to bregma). Glass capillaries with a tip size of 30 μm in outer diameter were loaded with AAV-syn-hChR2(E123A)-mCherry. Then, 0.35μl of the virus-containing solution was injected into the motor cortex at an injection rate of 5nl/s. After injection, injection needles were left in place for an additional 10min to assure even distribution of the virus and were then slowly withdrawn.

Immunostaining for organoids and brain slices

Organoids were aspirated with a 1ml pipette and fixed with 4% PFA for 2h. Mice were sacrificed under an overdose of chloral hydrate anesthesia (400mg/kg, i.p.; HuShi) and then were perfused with PBS followed by 4% PFA, and the brains were removed. Organoids or brains were dehydrated with 20% and 30% sucrose solution until completely sunk. Serial sections were obtained by frozen section (CM1950, Leica) at a thickness of 10 μm for organoids and 30 μm for mice brains. Sections were washed by PBS 3 times and treated with 5% donkey serum and 1% triton for 1h at room temperature. The sections were incubated in primary antibodies, which were prepared with 5% donkey serum and 0.2% Triton at 4°C overnight. The next day, sections were washed by PBS 3 times and incubated in secondary antibodies diluted in 5% donkey serum at room temperature for 1h. Finally, the sections were washed with PBS 3 times and mounted by Fluoromount-G. The primary and secondary antibodies are listed in Supplementary Table 1. Images were captured on Nikon Eclipse 80i fluorescence microscope and Zeiss LSM 700B laser confocal microscope. Image J software was used for cell counting, at least 8 random fields were chosen.

Behavioral test

In this study, the behavioral tests include cylinder test, grid-walking test, and adhesive removal test. Investigators were blinded to group allocation when assessing animal behaviors.

Cylinder test

Cylinder tasks were performed as previously described in detail with minor modifications. Briefly, each mouse was placed individually in a clear plexiglass cylinder (height, 15cm; diameter, 10cm). The mouse would spontaneously rear to a standing position while supporting its weight with either one or both of its forelimbs. A video camera (C525, Logitech) was positioned 20cm in front of a cylinder to record the free exploration of the mouse for 5min. The video footage was analyzed offline in slow motion (1/5^th real-time speed) by calculating the time (s) during each rear that the animal spent on either its right forelimb, left forelimb, or both forelimbs. Only rears in which both forelimbs could be clearly seen were included. The percentage of time spent on each limb was calculated and these data were used to derive an asymmetry index as follows: (% ipsilateral use) − (% contralateral use)⁴⁶.

Grid-walking test

The grid-walking test can evaluate the spontaneous movement defects of rodents by using the instinct of mice to explore new environments. To perform this test, each mouse was placed individually on top of an elevated wire grid (length × width × height: 32 × 20 × 50cm, each square wire meshes size is 1.2 × 1.2cm) and allowed to walk freely for 5min. A video camera (C525, Logitech) was positioned 30cm beneath the wire grid to capture stepping errors (foot faults). The video footage was analyzed offline in slow motion (1/5th real time speed) by a rater who was blind to the treatment groups. If a step did not provide support and the foot went through the grid it was considered a fault. A step was also considered a foot fault if the mouse was resting with the grid at the level of the wrist. The total number of steps for each limb was counted and the ratio between foot fault steps (F) and the total steps (T) of forelimbs was calculated: the number of forelimb foot faults/the total steps of forelimbs × 100%⁴⁷.

Adhesive removal test

The adhesive removal test was used to evaluate the sensorimotor function, time for a mouse to touch and remove adhesive stickers from the paws was measured as previously described^36,48. In brief, each mouse was placed individually in a plastic cage to explore the environment for 5min. Then a sticker (length × width: 0.3 × 0.2cm) was adhered to on the contralateral forepaw of the stroke mouse to record the whole process of contacting and removing the sticker. Mice were trained three times before surgery. The latency to touch and remove the sticker was counted by video slow-speed playback.

Electrophysiology

Brains were sectioned with a vibrating blade microtome (VT1200s, Leica) in ice-cold cutting solution containing 2.5mM KCl, 0.5mM CaCl₂, 7mM MgCl₂, 1.3mM NaH₂PO₄, 25mM NaHCO₃, 1.3mM Sodium-ascorbate, 0.6mM Sodium-Pyruvate, 110mM Choline chloride and 20mM D-Glucose. The slices were transferred to a recording chamber and were continuously perfused with normal artificial cerebrospinal fluid (ACSF) containing 125mM NaCl, 2.5mM KCl, 2mM CaCl₂, 1.3mM MgCl₂, 1.3mM NaH2_PO₄, 5mM NaHCO₃, 10mM D-Glucose, 1.3mM Sodium-ascorbate, 0.6mM Sodium-Pyruvate with 95% O₂ and 5% CO₂ at 34°C for 1h before recording (flow rate, 4–6ml/min). The slices were visualized using an upright microscope (Olympus X51W, Nomasky) with a 5 × 0.1 NA or 60 × 1.0 NA objective and infrared and differential interference contrast optics. The fluorescent neurons were visualized under the Olympus microscope equipped with a 60 × water-immersion lens and illuminated with a mercury lamp. For APs recordings, borosilicate glass pipettes were loaded with an intracellular solution containing 4mM Na-ATP, 5mM QX-314, 10mM HEPES, 0.5mM EGTA, 132.5mM Cs-Gluconate, 7.5mM CsCl and 2mM MgCl₂. For sPSCs recordings, we used internal solution containing 70 mM K-gluconate, 70mM KCl, 2mM NaCl, 2mM MgCl₂, 10mM HEPES, 1mM EGTA, 2mM Mg-ATP and 0.3mM Na2-GTP. EGFP⁺ neurons were clamped at −65 mV, and a series of current steps, going from 0 to 100 pA in 10 pA steps were injected to evoke APs, and sPSCs were recorded at −65 mV.

For ChR2-mediated APs and sPSCs recordings, optogenetic stimulation during recordings was performed using a 470nm light-emitting diode (LED) source coupled to a fiber guide and a collimator lens to focus the light spot around the recorded neuron or axons. For ChR2-EYFP-expressing grafted cells, blue light pulse (5ms duration, 10mW/mm²) was illustrated to EYFP⁺ neuron to evoke APs of this neuron or to the EYFP⁺ axons to evoke postsynaptic currents (PSCs) in EYFP⁻ neurons. After recording, the brain slices were washed with PBS, fixed with 4% PFA for immunostaining. To identify the source of GFP⁻ neuron recorded, we filled internal solution containing biocytin into the recorded neuron through microelectrodes after recording and then fixed the slice. For hChR2-mCherry-expressing host cells, blue light pulse (5ms duration, 10mW/mm²) was illustrated to the mCherry⁺ neuron to evoke APs of this neuron or to mCherry⁺ axons to evoke PSCs in EGFP⁺ graft neurons. In this experiment, kynurenic acid (3mM) was used to block AMPARs- and NMDARs-mediated EPSCs.

For rLV-EF1a-hM4D(Gi)-mCherry-expressing graft cells, mCherry⁺ neurons were clamped at −65 mV, and spontaneous APs of the neuron were recorded at 0 pA. Then bath application of CNO (5μM) inhibited AP firing for 10min and then washed out.

For local field potential (LFP) recordings, microelectrodes (1–2 MΩ) were filled with ACSF. A collimated 470nm LED was used to photostimulate ChR2-expressing GFP⁺ fibers in contralateral motor cortex and ipsilateral striatum. These fibers were projected from cerebral organoids neurons in the ipsilateral cortex. The slices were stimulated by a single 20ms light pulse repeated every 4s for 25 repetitions.

All recordings were acquired with a Digidata 1440A at 2kHz (Axon Instruments). Data were collected with pClamp 10.3 software and analyzed using Clamfit 10.3 (Molecular Devices).

We thank Dr. Su-Chun Zhang (University of Wisconsin-Madison, Madison, USA) for the gift of the H9-CAG-ChR2-EYFP cell line. This work was supported by grants from the National Key Research and Development Program of China (2021YFA1101803), the National Natural Science Foundation of China (82090042, 81870912, 82171293) and the Collaborative Innovation Center for Cardiovascular Disease Translational Medicine.