要理解大脑类器官的体积神经连接，监测类器官内的时空电生理信号（称为类器官内信号）至关重要。然而，以前的方法有可能通过切片或插入刚性针状电极来破坏类器官的三维 （3D） 细胞结构。此外，具有不可调电极形状的固定位置的有限数量的电极不足以检查整个类器官的复杂神经活动。在此，我们提出了一种磁性可翻转的 3D 多电极阵列 （MEA），使用液态金属的直接打印对脑类器官进行电生理分析。这些印刷电极的适应性分布和柔软性有助于类器官内信号的时空记录。此外，使用磁场在类器官内重塑这些软电极的独特能力允许 MEA 中的单个电极从多个点进行记录，从而有效地增加记录位点密度，而无需额外的电极。 要理解脑类器官的立体神经连接性，关键在于监测类器官内部的时空电生理信号，即类器官内信号。然而，既往方法无论是通过切片还是插入刚性针状电极，都可能破坏类器官的三维（3D）细胞结构。此外，固定位置电极数量有限且形状不可调，难以全面检测类器官中复杂的神经活动。本文提出一种采用液态金属直接打印的磁性可重构 3D 多电极阵列（MEA），用于脑类器官电生理分析。这些印刷电极的适应性分布和柔软特性，有助于实现类器官内信号的时空记录。更重要的是，利用磁场在类器官内重塑这些软电极的独特能力，可使 MEA 中的单个电极从多个位点进行记录，无需增加额外电极即可有效提升记录位点密度。
在此，我们提出了一种液态金属 （LMs） 的磁性可翻转 3D MEA，用于脑类器官的电生理分析。生物相容性液态金属，即共晶镓铟合金 （EGaIn;75.5 wt % 镓，24.5wt%24.5 \mathrm{wt} \%indium） 的 3D 柱状 （tid） 通过喷嘴直接打印成各种高度的 3D 柱状，以检测大脑类器官内不同 3D 坐标上的类器官内部信号。这些 3D LM 电极的杨氏模量与类器官的模量相当。它们的柔软和精细的突出结构允许对类器官的最小侵袭，这适用于神经活动的慢性监测。该 MEA 平台包括以下主要优势。首先，微型 3D LM 电极的位置和高度可以通过打印条件轻松控制，从而可以精确定制它们，以解决单个类器官的特定 3D 形态。特别是，这种 3D LM 打印的高分辨率形成了柱尖的细胞级尺寸，这在结构和机械上类似于大脑类器官中的神经元，同时最大限度地减少了侵入性。因此，3D LM 电极的精确定制布置有助于监测单个类器官的整个 3D 空间点的类器官内信号。此外，铂 （Pt） 纳米团簇仅覆盖这些 3D LM 柱的尖端，其中电池-电极界面已建立，从而降低了电极阻抗，增加了其表面积，并提高了信号读出质量。其次，使用我们的 3D LM MEA 对类器官内信号进行深入分析，揭示了通过在 3D 体积类器官中开发的神经网络进行电生理通讯。此外，3D LM 电极的最小侵入性允许在电生理网络水平上跟踪脑类器官的成熟。最后，EGaIn 的固有变形性和柔软性有助于通过铁磁层涂层重塑这些 3D LM 电极的磁性控制。例如，3D LM 柱的尖端根据磁场的方向进行调整，允许单个电极检测大脑类器官内的多个区域，有效地增加记录位点密度，而无需额外的电极。这种多点可检测的 MEA 有利于类器官内部神经网络电路的 3D 映射，它提供了一种有前途的解决方案来克服以前方法的局限性，并为大脑类器官的电生理体积网络提供了详尽的见解。 在此，我们提出了一种可磁性重塑的液态金属三维微电极阵列（MEA），用于脑类器官的电生理分析。该生物相容性液态金属，即共晶镓铟合金（EGaIn；75.5 wt%镓， 24.5wt%24.5 \mathrm{wt} \% 铟），通过喷嘴直接打印成不同高度的三维柱状结构，以检测脑类器官内跨三维坐标的类器官内信号。这些三维液态金属电极的杨氏模量与类器官相当，其柔软性和精细突起结构可实现最小侵入性，适用于神经活动的长期监测。该 MEA 平台具有以下关键优势：首先，通过打印条件可轻松控制微米级三维液态金属电极的排布与高度，从而精确匹配单个类器官的三维形态特征。尤其值得注意的是，这种三维液态金属打印技术的高分辨率使柱体尖端达到细胞级尺寸，其结构和力学特性与脑类器官中的神经元相似，同时最大程度降低了侵入性。 因此，3D LM 电极的精确定制排列有助于监测单个类器官所有 3D 空间点上的内部信号。此外，铂（Pt）纳米团簇仅覆盖这些 3D LM 柱的尖端，此处细胞-电极界面已良好建立，从而降低了电极阻抗，增加了表面积，并提高了信号读取质量。其次，利用我们的 3D LM MEA 对类器官内部信号进行深入分析，揭示了通过 3D 体积类器官内发育的神经网络进行的电生理通信。此外，3D LM 电极的微创性允许在电生理网络水平上追踪脑类器官的成熟过程。最后，EGaIn 固有的可变形性和柔软性通过铁磁层涂层促进了磁性控制以重塑这些 3D LM 电极。 例如，3D LM 柱的尖端会根据磁场方向进行调整，使得单个电极能够检测脑类器官内的多个区域，有效提高记录位点密度而无需增加额外电极。这种可多点检测的微电极阵列（MEA）对于脑类器官内部神经网络的 3D 映射具有优势，并为克服以往方法的局限性提供了有前景的解决方案，从而能更精细地解析脑类器官的电生理体积网络。
Results 结果
3D LM MEA for intra-organoid recording 用于类器官内记录的 3D LM 微电极阵列
Figure 1a presents a schematic image on the intra-organoid recording using 3D LM MEA. 3D LM pillars with desired placements and heights were directly printed to form electrodes (Fig. 1b). The printing system 图 1a 展示了使用 3D LM 微电极阵列进行类器官内记录的示意图。通过直接打印形成具有预定位置和高度的 3D LM 柱状电极（图 1b）。该打印系统
Fig. 1 | 3D multi-electrode array of liquid metals as an electrophysiological monitoring platform. a Schematic illustration of intra-organoid signal recording by 3D LM MEA. b Schematic illustration of printing system and printing process. c Various heights of 3D LM pillars by control printing velocity. Each data point indicates the average of 12 measurements, and the error bars represent the s.e.m. Inset presents SEM image, which was taken by 30^(@)30^{\circ} tilting of the substrate, presenting 3D LM pillars with various heights. Scale bar, 100 mum100 \mu \mathrm{~m}. This experiment was independently repeated more than ten times with similar results. d Optical images of LM interconnects and 3D LM pillars with height-variance. Scale bars, 100 mum100 \mu \mathrm{~m}. 图 1 | 作为电生理监测平台的液态金属三维多电极阵列。a 通过 3D LM MEA 进行类器官内信号记录的示意图。b 打印系统及打印过程的示意图。c 通过控制打印速度实现不同高度的 3D LM 柱状结构。每个数据点代表 12 次测量的平均值，误差条表示标准误差均值。插图展示了 SEM 图像，拍摄时基底倾斜 30^(@)30^{\circ} 度，呈现不同高度的 3D LM 柱状结构。比例尺为 100 mum100 \mu \mathrm{~m} 。该实验独立重复超过十次，结果相似。d LM 互连结构和高度变化的 3D LM 柱状结构的光学图像。比例尺为 100 mum100 \mu \mathrm{~m} 。
e Schematic illustrations of 3D LM MEA, and single 3D LM electrode. f SEM image of platinum nanoclusters coated on tip of the electrode (The substrate is 30^(@)30^{\circ} tilted when the image was taken). Scale bar, 1mum1 \mu \mathrm{~m}. This experiment was independently repeated more than ten times with similar results. gg Comparison of impedances between pristine 3D LM electrode and Pt nanocluster coated LM electrode ( n=9n=9 electrodes for each group; p=2.27 xx10^(-13)p=2.27 \times 10^{-13} ). All data are presented as mean +-\pm s.e.m. Statistical differences were determined with unpaired, one-sided t-test; ^(******)p < 0.001{ }^{* * *} p<0.001. LM liquid metal; MEA multi-electrode array; SEM scanning electron microscopy. e 三维液态金属多电极阵列（3D LM MEA）及单个 3D LM 电极的示意图。f 电极尖端铂纳米团簇的扫描电子显微镜（SEM）图像（拍摄时基底倾斜 30^(@)30^{\circ} ）。比例尺， 1mum1 \mu \mathrm{~m} 。本实验独立重复超过十次，结果相似。 gg 原始 3D LM 电极与铂纳米团簇包覆 LM 电极的阻抗对比（每组 n=9n=9 个电极； p=2.27 xx10^(-13)p=2.27 \times 10^{-13} ）。所有数据均以平均值 +-\pm 标准误差表示。统计差异采用非配对单侧 t 检验确定； ^(******)p < 0.001{ }^{* * *} p<0.001 。LM：液态金属；MEA：多电极阵列；SEM：扫描电子显微镜。
was composed of an ink reservoir connected to a glass capillary nozzle in a fixed position, a pneumatic pressure controller, and a precisely controllable 6 -axis stage ( x,y,zx, y, z-axis, 2 tilting axes, and a rotating axis in xy-plane). EGaIn, which exists in the liquid phase at room temperature (melting point of 15.5^(@)C15.5^{\circ} \mathrm{C} ), was used as the ink. The Young’s modulus of EGaIn (2.1 xx10^(5)(Pa))\left(2.1 \times 10^{5} \mathrm{~Pa}\right) is 4 to 6 orders of magnitude lower than that of the rigid, solid-state metals (e.g., Au or Pt ) that are conventionally used for bioelectronics ^(30,31){ }^{30,31}. The mechanical softness and good biocompatibility of EGaIn ensures the avoidance of inflammatory responses that can be triggered by the mechanical modulus mismatch between biological tissue and materials ^(32){ }^{32}. Supplementary Note 1 describes the printing process, and Supplementary Movie 1 shows a movie to print interconnects and 3D pillars. When a pressure of 36 psi was applied to a glass capillary (inner diameter: 18 mum18 \mu \mathrm{~m} ), pillars of varying heights above 该装置由一个固定位置的墨水储存器连接至玻璃毛细管喷嘴、一个气压控制器以及一个精确可控的六轴平台（ x,y,zx, y, z 轴、2 个倾斜轴和 xy 平面内的旋转轴）组成。室温下呈液态的共晶镓铟合金（EGaIn，熔点 15.5^(@)C15.5^{\circ} \mathrm{C} ）被用作墨水。EGaIn 的杨氏模量 (2.1 xx10^(5)(Pa))\left(2.1 \times 10^{5} \mathrm{~Pa}\right) 比传统生物电子器件中使用的刚性固态金属（如金或铂）低 4 至 6 个数量级。EGaIn 的机械柔软性和良好生物相容性确保了可避免因生物组织与材料间机械模量不匹配引发的炎症反应 ^(32){ }^{32} 。补充说明 1 描述了打印过程，补充视频 1 展示了打印互连结构和 3D 柱体的动态过程。当对玻璃毛细管（内径 18 mum18 \mu \mathrm{~m} ）施加 36 psi 压力时，可形成不同高度的柱体结构。 310 mum310 \mu \mathrm{~m} (diameter: 14 mum14 \mu \mathrm{~m} in minimum) were formed by controlling the printing velocity (Fig. 1c). This controllability over the pillar height allowed precise positioning of the recording sites, which were formed only at the pillar tips (after selectively encapsulating the side walls of pillars using a thin insulating layer), throughout the internal volume of the organoid. For example, Fig. 1d shows photographs of a 3D LM MEA consisting of LM interconnects and pillars of varying heights; these pillars are capable of collecting electrophysiological information across different spaces in the x,yx, y, and zz axes within an organoid (Supplementary Fig. 1). 通过控制打印速度形成了直径最小为 14 mum14 \mu \mathrm{~m} 的 310 mum310 \mu \mathrm{~m} （图 1c）。这种对柱体高度的可控性使得记录位点能够精确定位，这些位点仅在柱体顶端形成（在选择性包裹柱体侧壁的薄绝缘层之后），遍布类器官的内部空间。例如，图 1d 展示了一个由液态金属互连和不同高度柱体构成的三维液态金属微电极阵列照片；这些柱体能够在类器官内部沿 x,yx, y 和 zz 轴方向的不同空间收集电生理信息（补充图 1）。
Figure 1e presents schematic illustrations of our 3D LM MEA and the pillar tip part. This MEA consisted of interconnects, a parylenepassivation layer, 3D LM electrodes (i.e., pillars), and elastomer wells. The elastomer wells included (i) an inner microwell that guided the 图 1e 展示了我们的三维液态金属微电极阵列及柱体顶端部分的示意图。该微电极阵列由互连结构、聚对二甲苯钝化层、三维液态金属电极（即柱体）和弹性体微孔组成。弹性体微孔包括（i）一个用于引导的内侧微孔。
positioning of an organoid to the desired location of this MEA device and (ii) an outer macrowell that contained media for the cultivation of organoids. The side walls of 3D LM pillars were encapsulated with parylene-C, a biocompatible elastomeric layer (thickness: 2mum2 \mu \mathrm{~m} ). Then, using the isotropy of reactive ion etching system, only the tips of pillars were opened selectively, resulting in a small open area of approximately 78.5 mum^(2)78.5 \mu \mathrm{~m}^{2} (Supplementary Fig. 2). Additionally, the Pt nanoclusters were electrodeposited only on the open tip areas of EGaIn, and the rough surface of these Pt nanoclusters increased the surface area of the pillar tip to establish a good cell-electrode interface (Fig. 1f). For example, Fig. 1g and Supplementary Fig. 3 show that the 3D LM electrode coated with Pt nanoclusters exhibited an impedance that was four times lower than the impedance of the pristine 3D EGaIn electrode (without Pt nanoclusters). Furthermore, when we investigate the stability of the electrical performance of 3D LM electrodes by accelerated aging test, the impedance shows negligible changes for 12 days at 87^(@)C(18.005+-0.083kOmega87{ }^{\circ} \mathrm{C}(18.005 \pm 0.083 \mathrm{k} \Omega to 18.774+-0.145kOmega)18.774 \pm 0.145 \mathrm{k} \Omega), which corresponds to 12 months in the incubator (37^(@)C)\left(37^{\circ} \mathrm{C}\right) (Supplementary Fig. 4). 将类器官定位至该 MEA 设备的期望位置，以及(ii)一个容纳培养基用于类器官培养的外部宏孔。3D LM 柱的侧壁被聚对二甲苯-C（一种生物相容性弹性层，厚度： 2mum2 \mu \mathrm{~m} ）包裹。随后，利用反应离子蚀刻系统的各向同性，仅选择性打开柱的尖端，形成一个约 78.5 mum^(2)78.5 \mu \mathrm{~m}^{2} 的小开放区域（附图 2）。此外，Pt 纳米团簇仅电沉积在 EGaIn 的开放尖端区域，这些 Pt 纳米团簇的粗糙表面增加了柱尖端的表面积，从而建立了良好的细胞-电极界面（图 1f）。例如，图 1g 和附图 3 显示，涂覆有 Pt 纳米团簇的 3D LM 电极的阻抗比原始 3D EGaIn 电极（无 Pt 纳米团簇）的阻抗低四倍。此外，当我们通过加速老化测试研究 3D LM 电极电性能的稳定性时，在 87^(@)C(18.005+-0.083kOmega87{ }^{\circ} \mathrm{C}(18.005 \pm 0.083 \mathrm{k} \Omega 至 18.774+-0.145kOmega)18.774 \pm 0.145 \mathrm{k} \Omega) 的 12 天内阻抗变化可忽略不计，相当于培养箱中的 12 个月 (37^(@)C)\left(37^{\circ} \mathrm{C}\right) （附图 4）。
This direct printing of LM offers flexibility in choice of structural designs, such as 1 ) variation in the position and height of the 3D electrodes and 2) addressing the diverse needs for achieving electrophysiological information from brain organoids. In particular, an electrophysiological monitoring system for drug screening requires a high-throughput platform to analyze the efficacy of an identical drug to various groups of organoids or to monitor the responses of brain organoids to numerous types of drugs. As an example, Supplementary Fig. 5 demonstrates a high-throughput platform where a total of 9 MEAs (with 3D LM electrodes) were integrated into a single chip to simultaneously monitor the electrophysiological signals from 9 different organoids. This ease of modifying the configurations of the 3D LM electrodes provides new opportunities to explore neural dynamics in brain organoids using customizable designs of devices. 这种直接打印液态金属（LM）的技术为结构设计提供了灵活性选择，例如：1）三维电极位置和高度的变化；2）满足从脑类器官获取电生理信息的多样化需求。特别地，用于药物筛选的电生理监测系统需要一个高通量平台，以分析同一药物对不同组别类器官的效果，或监测脑类器官对多种药物的反应。举例来说，补充图 5 展示了一个高通量平台，其中集成了 9 个带有三维液态金属电极的微电极阵列（MEA），可同时监测来自 9 个不同类器官的电生理信号。这种易于调整三维液态金属电极配置的特性，为利用可定制化设备探索脑类器官中的神经动力学提供了新机遇。
Integration of 3D LM MEAs with human cortical organoids 三维液态金属微电极阵列与人皮层类器官的整合
Cortical organoids were generated from hiPSCs based on previously established protocols with minor modifications (Fig. 2a) ^(33){ }^{33}. Briefly, single-cell-dissociated hiPSCs were aggregated with the addition of dual-SMAD inhibition for neural induction. Then, they were further differentiated by the sequential addition of growth factors to the differentiation medium. Early developmental cortical organoids showed an expanded epithelium with bright, smooth, optically translucent edges, and further differentiated organoids increased in size, reaching over 3 mm on day 60 after which little additional growth was observed for up to 120 days of culture (Fig. 2b). The cytoarchitecture of the generated cortical organoids was confirmed by immunofluorescence staining for SOX2, a radial glial cell marker, and TUJ1, a neurofilament protein marker (Fig. 2c). The expression of SOX2 was observed in the cells that lined the ventricular zone, similar to in-vivo development of brain tissue. At 30 days, the majority of ventricular units had strong SOX2+ signals. However, by 120 days, the SOX2+ ventricular zone became ambiguous. After 150 days of culture, we confirmed the presence of glial cells, which occur during the late stages of cortical development in humans, via the expression of glial fibrillary acidic protein (GFAP), an astrocyte marker (Fig. 2d) ^(34){ }^{34}. The GFAP+ cells, with the characteristic morphological feature of astrocytes, were distributed throughout the organoids. In addition, Vesicular glutamate transporter 1 (VGLUT1)+ mature excitatory neurons were also observed (Fig. 2e). 皮质类器官是基于先前建立的方案，并稍作修改后从 hiPSCs 生成的（图 2a）。简而言之，将单细胞解离的 hiPSCs 聚集，并添加双 SMAD 抑制以进行神经诱导。随后，通过向分化培养基中依次添加生长因子进一步分化。早期发育的皮质类器官显示出扩张的上皮层，边缘明亮、光滑、半透明，进一步分化的类器官体积增大，在第 60 天时超过 3 毫米，之后在培养至 120 天时几乎没有观察到额外的生长（图 2b）。生成的皮质类器官的细胞结构通过免疫荧光染色 SOX2（一种放射状胶质细胞标记物）和 TUJ1（一种神经丝蛋白标记物）得到确认（图 2c）。SOX2 的表达在脑室区排列的细胞中观察到，类似于脑组织的体内发育。在第 30 天时，大多数脑室单元具有强烈的 SOX2+信号。然而，到 120 天时，SOX2+脑室区变得模糊不清。 经过 150 天的培养后，我们通过星形胶质细胞标志物胶质纤维酸性蛋白（GFAP）的表达，确认了人类皮质发育晚期出现的胶质细胞存在（图 2d） ^(34){ }^{34} 。具有星形胶质细胞典型形态特征的 GFAP 阳性细胞遍布类器官各处。此外，还观察到囊泡谷氨酸转运体 1（VGLUT1）阳性的成熟兴奋性神经元（图 2e）。
We then examined whether the cortical organoids exhibited mature cortical organization during long-term culture. To determine if developing cortical organoids could recapitulate the cortical spatial organization, the organoids were stained with markers for cortical layer, i.e., CTIP2 and SATB2, which identify layers V (deep) and II-IV (upper), respectively, on day 60 and 130 of the culture (Fig. 2f, g). The cortical plate of 60 -day organoids contained both the deep-layer 随后我们检测了长期培养的皮质类器官是否呈现成熟皮质组织结构。为确定发育中的皮质类器官能否重现皮质空间组织特征，分别在培养第 60 天和 130 天用皮质层标志物 CTIP2（标记第 V 层即深层）和 SATB2（标记第 II-IV 层即上层）对类器官进行染色（图 2f, g）。60 天类器官的皮质板同时包含深层
marker CTIP2 and the upper-layer marker, SATB2, without layer preference. In contrast, when the cortical organoids were cultured up to day 130, the laminar distribution of the markers became layer-specific. The upper layers were mainly populated by SATB2+ neurons, separated by a distinguishable boundary. We analyzed quantitatively the laminar expression patterns of SATB2 and CTIP2 (Fig. 2g). On day 60, the distributions of the two markers overlapped, but on day 130, they appeared as two separate populations, representing the upper and deep layers. Altogether, these results demonstrate that the generated cortical organoids recapitulate the neurogenesis and formation mature cortical plates that have distinct upper and deep cortical layers with glial cells. 标记物 CTIP2 和上层标记物 SATB2 无分层偏好。相比之下，当皮质类器官培养至第 130 天时，标记物的层状分布变得具有层次特异性。上层主要由 SATB2+神经元占据，并通过可区分的边界分隔。我们定量分析了 SATB2 和 CTIP2 的层状表达模式（图 2g）。在第 60 天，两种标记物的分布重叠，但在第 130 天，它们表现为两个独立的群体，分别代表上层和深层。总之，这些结果表明，生成的皮质类器官重现了神经发生和成熟皮质板的形成过程，该皮质板具有由胶质细胞构成的不同上层和深层皮质结构。
Figure 3a and Supplementary Movie 2 illustrates the integration of this cortical organoid to our 3D LM MEA. After positioning the cortical organoid floating in the medium near the electrode area and using a pipette to aspirate the medium, this organoid settled down. Then, 3D LM electrodes became naturally inserted inside this organoid. After the aspiration for electrode insertion, this organoid recovered its original form by resupplying the medium. Immunostaining of the organoid section confirmed that the 3D LM electrodes had been inserted successfully into the organoid (Fig. 3b). To investigate the biocompatibility of these electrodes on the cortical organoid, a cytotoxicity test using a live/dead staining kit was performed for this sample, and this result was compared with the control case (i.e., organoid with no electrode) for 2 weeks (Fig. 3c). There was no significant difference in viability between these two cases (Fig. 3d). Furthermore, we performed optical clearing of the organoid followed by whole-mount 3D imaging (Supplementary Fig. 6), showing that the 3D LM electrodes were inserted into the organoid straight without causing any structural damage on brain organoids, allowing for reliable monitoring of neural network within the organoid. To further investigate the effect of 3D LM electrodes on neuronal gene expression in the organoid, quantitative real-time polymerase chain reaction (qPCR) was performed on 7 and 14 days after the electrode insertion (Fig. 3e). Expression of neuronal differentiation markers (PAX6, Nestin, TUJ1) showed no significant difference between organoids with or without the 3D LM electrodes. These data suggest that insertion of 3D LM electrodes does not affect the viability or neurogenesis of cortical organoids during the recording period. 图 3a 和补充影片 2 展示了该皮质类器官与我们 3D LM MEA 的整合过程。将漂浮在培养基中的皮质类器官移至电极区域附近后，用移液管吸除培养基，类器官随即沉降。随后，3D LM 电极自然插入类器官内部。完成电极插入的吸液操作后，通过重新补充培养基使类器官恢复原始形态。类器官切片的免疫染色结果证实 3D LM 电极已成功插入（图 3b）。为评估这些电极对皮质类器官的生物相容性，使用活/死细胞染色试剂盒对该样本进行细胞毒性测试，并与对照组（即无电极的类器官）进行为期 2 周的对比（图 3c）。两种情况下细胞存活率无显著差异（图 3d）。此外，我们对类器官进行光学透明化处理并实施整体三维成像（补充图。 6）显示，3D 液态金属电极被直接插入类器官而未对其造成结构损伤，从而能够可靠地监测类器官内部的神经网络活动。为进一步探究 3D 液态金属电极对类器官神经元基因表达的影响，在电极插入后第 7 天和第 14 天进行了实时定量聚合酶链反应（qPCR）分析（图 3e）。神经元分化标志物（PAX6、巢蛋白、TUJ1）的表达在植入与未植入 3D 液态金属电极的类器官间无显著差异。这些数据表明，在记录期间，3D 液态金属电极的插入不会影响皮质类器官的存活或神经发生。
Intra-organoid recordings using 3D LM MEAs 利用 3D 液态金属微电极阵列进行类器官内记录
For intra-organoid recordings, 3D volumetric electrophysiological activities of a brain organoid, a 4 -month-old cortical organoid that resembles the cerebral cortex of the human brain was monitored using the 3D LM MEA. The intra-organoid signals with frequencies ranging from 0.1 to 3000 Hz were recorded and transferred to a computer passing through a multi-channel signal processor (Tucker-Davis Technologies Inc, USA). The signal processor applied band-pass filters on the data and displayed the results through the software (Synapse). We examined the electrophysiological activities of the organoid by acquisition of filtered data directly from the software. We then utilized custom MATLAB code to gain deeper insight into the 3D neural networking circuitry that was developed within this organoid. 对于类器官内记录，使用 3D LM MEA 监测了一个 4 月龄皮质类器官（类似人脑大脑皮层）的 3D 容积电生理活动。记录频率范围为 0.1 至 3000 Hz 的类器官内信号，并通过多通道信号处理器（Tucker-Davis Technologies Inc, USA）传输至计算机。信号处理器对数据应用带通滤波器，并通过软件（Synapse）显示结果。我们通过直接从软件获取滤波数据来检查类器官的电生理活动。随后利用定制 MATLAB 代码深入分析该类器官内发育的 3D 神经网络回路。
Figure 4 a and Supplementary Fig. 7 present the single unit (SU) potentials and local field potentials (LFPs) of this 4-month-old cortical organoid, respectively. The sufficiently low impedance and the cellular-scale size of our 3D LM electrodes allowed the detection of SU potentials, the rapid fluctuations in voltage traces (noise level: ∼13 muV\sim 13 \mu \mathrm{~V} ). To examine the neuronal activity of intra-organoid signals, we conducted the spike detection using a threshold of -5xx-5 \times standard deviation of the SU potential. The fluctuations in the voltage traces that exceeded the threshold within 1-2 milliseconds were determined as spikes. Principal component analysis (PCA), which is commonly used to classify clusters of neural spikes, was adopted to examine the spikeforms. The mean spikeforms of each cluster were detected from 60 channels 图 4a 及补充图 7 分别展示了该 4 月龄皮层类器官的单单位（SU）电位与局部场电位（LFP）。我们研发的三维光微电极凭借足够低的阻抗和细胞级尺寸，成功捕获了单单位电位——即电压轨迹中快速波动的信号（噪声水平： ∼13 muV\sim 13 \mu \mathrm{~V} ）。为分析类器官内部信号的神经元活动，我们采用 -5xx-5 \times 倍单单位电位标准差作为阈值进行尖峰检测，将 1-2 毫秒内超过该阈值的电压波动判定为动作电位。采用主成分分析（PCA）这一常用于神经尖峰聚类的方法对尖峰波形进行分类研究，并从 60 个通道中检测出各簇的平均尖峰波形。
a
b
C
TUJ1 SOX2 DAPI
d
MAP2 GFAP DAPI e\mathbf{e}
TUJ1 VGLUT1 DAPI
f
Day 60 第 60 天
Day 130 第 130 天
Day 60 第 60 天
Day 130 第 130 天
Fig. 2 Formation and characterization of cortical organoids. a Schematic of the protocol for generating human cortical organoids from hiPSCs. DM, Dorsomorphin; SB, SB-431542; FGF2, Fibroblast growth factor 2; EGF, Epidermal growth factor; BDNF, brain-derived neurotrophic factor; NT3, Neurotrophin-3. b Representative bright-field images of cortical organoids at 15,30,45,6015,30,45,60, and 120 days of culture ( n=4-8n=4-8, biological replicates). Scale bars, 500 mum500 \mu \mathrm{~m}. c Immunofluorescence staining for neuronal progenitor marker SOX2 and neuronal marker TUJ1 at 30, 60, and 120 days of culture ( n=5n=5, biological replicates). Scale bars, 100 mum100 \mu \mathrm{~m}. d Immunofluorescence staining for neuronal marker MAP2 and astrocyte marker GFAP at day 150 in cortical organoids ( n=5n=5, biological replicates). 图 2 皮层类器官的形成与表征。a 从 hiPSCs 生成人类皮层类器官的方案示意图。DM，Dorsomorphin；SB，SB-431542；FGF2，成纤维细胞生长因子 2；EGF，表皮生长因子；BDNF，脑源性神经营养因子；NT3，神经营养素-3。b 皮层类器官在培养 0 天和 120 天时的代表性明场图像（n=3，生物学重复）。比例尺，500 μm。c 培养 30、60 和 120 天时神经元前体标志物 SOX2 与神经元标志物 TUJ1 的免疫荧光染色（n=3，生物学重复）。比例尺，100 μm。d 皮层类器官培养 150 天时神经元标志物 MAP2 与星形胶质细胞标志物 GFAP 的免疫荧光染色（n=3，生物学重复）。
of 3D LM MEA, and representative results are presented in Fig. 4b. Each single electrode detected only two or three clusters, indicating that these 3D electrodes could collect intra-organoid signals without degradation to distinguish differences in spikeforms. We further examined the response of these cortical organoids to potassium chloride ( KCl ), in order to investigate changes in neural activity. The 10 mM of KCl was added to the culture medium for intra-organoid recordings, and electrophysiological responses were analyzed as shown in Supplementary Fig. 8a. The spiking rate significantly increased by this KCl treatment, indicating that the recorded intraorganoid signals originated from the activity of neurons (Supplementary Fig. 8b). Along with the treatment of KCl , we examined the neural response of brain organoid to tetrodotoxin (TTX) to confirm that the recorded signals originated from neuronal activities (Supplementary Fig. 9). 3D LM MEA 的部分，代表性结果如图 4b 所示。每个单电极仅检测到两到三个簇群，表明这些 3D 电极能采集类器官内部信号且无衰减，可区分不同峰电位形态。为进一步探究神经活动变化，我们检测了这些皮层类器官对氯化钾（KCl）的响应：在类器官内记录时向培养基添加 10 mM KCl，电生理响应分析如补充图 8a 所示。KCl 处理使峰电位发放率显著增加（补充图 8b），证实记录的类器官内信号源自神经元活动。同步进行河豚毒素（TTX）处理以验证脑类器官的神经响应（补充图 9），进一步确认记录信号源于神经元活动。
Scale bar, 50 mum50 \mu \mathrm{~m}. e Immunofluorescence staining for TUJ1 and glutamatergic neuron marker VGLUT1 at day 150 in cortical organoids ( n=4n=4, biological replicates). Scale bar, 50 mum50 \mu \mathrm{~m}. f Comparative immunohistochemical images and g\mathbf{g} quantifications of the distribution of SATB2+ and CTIP2+ neurons in the cortical organoids at days 60 and 130 . The yy-axis position of the images was evenly divided into 10 bins following the basal-to-apical direction within 200 mum200 \mu \mathrm{~m} from the pial surface. The curve showing the normalized cell number within each bin was calculated by measuring the number of positively stained cells in a bin and dividing them by the total cells. Values were presented as mean +-\pm s.e.m. (n=4(n=4, biological replicates). Scale bars, 25 mum25 \mu \mathrm{~m}. hiPSCs, human induced pluripotent stem cells. 比例尺， 50 mum50 \mu \mathrm{~m} 。e 皮质类器官在第 150 天时 TUJ1 和谷氨酸能神经元标记物 VGLUT1 的免疫荧光染色（ n=4n=4 ，生物学重复）。比例尺， 50 mum50 \mu \mathrm{~m} 。f 皮质类器官在第 60 天和第 130 天时 SATB2+和 CTIP2+神经元分布的对比免疫组化图像及 g\mathbf{g} 定量分析。图像沿 yy 轴位置从软脑膜表面开始向基底到顶端方向在 200 mum200 \mu \mathrm{~m} 范围内均匀划分为 10 个区间。通过测量每个区间内阳性染色细胞数并除以总细胞数，计算显示各区间内标准化细胞数量的曲线。数值以均值 +-\pm s.e.m.表示（ (n=4(n=4 ，生物学重复）。比例尺， 25 mum25 \mu \mathrm{~m} 。hiPSCs，人诱导多能干细胞。
Brain organoids have attracted researchers’ interest because they recapitulate the complex functionality of the human brain. Thus, brain organoids have been exploited in a diverse range of studies, ranging from discovering the dysfunction in neuropathological circuits of specific diseases to excavating patient-specific drugs with patientdriven organoids. The functionality of the brain organoid is represented by communications among neurons through the neural networking circuitry developed in the 3D intra-organoid volume. In other words, simply observing the voltage traces arising from neurons is insufficient to fully understand the connoted electrophysiological information of organoids. Thus, comprehensive examination of 3D intra-organoid neural networking circuitry can exhaustively utilize the merit of brain organoids. In this purpose, we investigated neural networking circuitry based on the evaluation of synchronized activity by calculating spike train synchrony (referred to as synchronization 脑器官体因其重现人脑复杂功能而吸引了研究者的兴趣。因此，脑器官体已被广泛应用于多项研究中，从发现特定疾病神经病理回路的功能障碍，到利用患者驱动的器官体挖掘针对患者的药物。脑器官体的功能体现在神经元之间通过三维器官体内形成的神经网络回路进行的交流。换句话说，仅观察神经元产生的电压轨迹不足以全面理解器官体所蕴含的电生理信息。因此，对三维器官体内神经网络回路的全面检查可以充分利用脑器官体的优势。为此，我们基于同步活动的评估，通过计算尖峰序列同步性（称为同步化）来研究神经网络回路。
Fig. 4 | Intra-organoid signals from 4-month-old cortical organoids. 图 4 | 4 月龄皮层类器官的类器官内信号。
a, b Representative intra-organoid signals of 4 -month-old cortical organoid integrated on 3D LM MEA; (a) SU potential. Scale bars, 50 muV50 \mu \mathrm{~V} (vertical), 10 s (horizontal). (b) Spikeforms. Scale bars, 50 muV50 \mu \mathrm{~V} (vertical), 2 ms (horizontal). c\mathbf{c} Color mapping of synchronization score between the electrodes located inside the 4 -month-old cortical organoid. The color bar indicates the synchronization score between two electrodes. d Neural networking map based on synchronization score and neural community of 4 -month-old cortical organoid integrated on 3D LM MEA. Each node indicates individual channels of 3D LM MEA. The lines connecting two electrodes are presented when the synchronization score exceeds 0.5 . The size of the node represents the number of lines connecting with other nodes, and the color of the a、b 4 月龄皮层类器官在 3D LM MEA 上整合的代表性类器官内信号；(a) SU 电位。比例尺， 50 muV50 \mu \mathrm{~V} （垂直方向），10 秒（水平方向）。(b) 尖峰波形。比例尺， 50 muV50 \mu \mathrm{~V} （垂直方向），2 毫秒（水平方向）。 c\mathbf{c} 4 月龄皮层类器官内部电极间同步得分的颜色映射。色条表示两个电极间的同步得分。d 基于同步得分和 4 月龄皮层类器官在 3D LM MEA 上整合的神经群落的神经网络图。每个节点代表 3D LM MEA 的独立通道。当同步得分超过 0.5 时，显示连接两个电极的线条。节点大小表示与其他节点连接的线条数量，节点颜色表示...
node represents the neural community. The color of lines represents the synchronization score. e Configuration of 3D LM electrodes with 3-level of heights. f, g\mathbf{g} Representative intra-organoid signals depending on the heights of the electrodes. Each color represents the height of the electrodes; Orange, 50 mum50 \mu \mathrm{~m}, green, 100 mum100 \mu \mathrm{~m}, blue, 200 mum200 \mu \mathrm{~m}. (f) SU potentials. Scale bars, 100 muV100 \mu \mathrm{~V} (vertical), 0.5 s (horizontal). (g) Spikeforms. Scale bars, 100 muV100 \mu \mathrm{~V} (vertical), 0.5 ms (horizontal). h 3D neural networking map corresponding to the 3-level height variance of 3D LM MEA. i,j\mathbf{i}, \mathbf{j} Changes in neural activity depending on the heights of the electrodes ( n=5n=5 electrodes per each height); (i) Spiking rate, (j) Number of bursts. All data are presented as mean +-\pm s.e.m. LM, liquid metal; MEA, multi-electrode array; SU, single-unit. 节点代表神经群落。线条颜色表示同步得分。e 具有 3 层高度的 3D 液态金属电极配置。f， g\mathbf{g} 根据电极高度呈现的代表性类器官内信号。每种颜色代表电极的高度；橙色， 50 mum50 \mu \mathrm{~m} ，绿色， 100 mum100 \mu \mathrm{~m} ，蓝色， 200 mum200 \mu \mathrm{~m} 。(f) 单单元电位。比例尺， 100 muV100 \mu \mathrm{~V} （垂直），0.5 秒（水平）。(g) 尖峰波形。比例尺， 100 muV100 \mu \mathrm{~V} （垂直），0.5 毫秒（水平）。h 对应 3D 液态金属多电极阵列 3 层高度变化的 3D 神经网络映射图。 i,j\mathbf{i}, \mathbf{j} 神经活动随电极高度的变化（ n=5n=5 每个高度的电极数）；(i) 尖峰率，(j) 爆发次数。所有数据均以平均值 +-\pm 标准误差表示。LM，液态金属；MEA，多电极阵列；SU，单单元。
score ^(35){ }^{35} ). The representative synchronized activities of the brain organoid are presented in Supplementary Fig. 10. The mechanism to determine the synchronization score with this platform is described in Supplementary Note 2. Figure 4c presents the synchronization score mapping of this 4 -month-old cortical organoid, providing an intuitive, color-by-color representation of active neural networking. Furthermore, we also performed an evaluation of neural networks by identifying neural communities. We defined a neural community among neurons where a modularity, a complexity of neural networking 得分 ^(35){ }^{35} ）。大脑类器官的代表性同步活动展示在补充图 10 中。该平台用于确定同步得分的机制在补充说明 2 中有详细描述。图 4c 展示了这个 4 个月大皮质类器官的同步得分映射，通过直观的逐色表示法呈现活跃的神经网络。此外，我们还通过识别神经群落对神经网络进行了评估。我们将神经元之间的模块化定义为神经群落，即神经网络的复杂性。
circuitry, is maximized ^(36,37){ }^{36,37}. (The computational method for neural community is described in Supplementary Note 3) Fig. 4d presents the neural community and connectivity of this 4 -month-old organoid as a neural networking map. Nodes were placed according to the distribution of 3D LM electrodes, and the node size indicated its connectivity (i.e., the number of connecting lines with other nodes). Nodes in the same neural community were shown in the same color. As shown by the color bar in Fig. 4d, the line color indicated the synchronization score. These results from neural network mapping indicated that 电路结构在 ^(36,37){ }^{36,37} 处达到最大化。（神经群落的计算方法详见补充说明 3）图 4d 以神经网络图谱形式展示了这一 4 个月大类器官的神经群落及其连接性。节点根据 3D LM 电极的分布进行排布，节点大小反映其连接度（即与其他节点的连线数量）。同一神经群落中的节点采用相同颜色标示。如图 4d 色条所示，连线颜色代表同步化评分。神经网络图谱分析结果表明，
neurons in contact with electrodes, within a neural community, developed robust and complex circuitry. And the networking circuitry of neurons near electrodes with larger nodes reached diverse regions of this organoid, indicating that the neurons actively communicated with other neuron groups. For example, neurons near the electrodes connected by red lines were connected strongly by neural networking circuitry. 与电极接触的神经元在神经群落内形成了稳健而复杂的电路结构。靠近较大节点电极的神经元网络回路可延伸至类器官的广泛区域，表明这些神经元与其他神经元群存在活跃通信。例如，由红色连线连接的电极附近神经元，通过神经网络回路形成了强效连接。
Since the height of 3D LM electrodes can be controlled readily during the direct printing process, the electrodes can be positioned in a variety of configurations throughout the internal volume of organoids to investigate changes in neural activity depending on the recording site. For example, we printed 3D LM pillars with 3 different heights to record SU potentials and spikes at various coordinates within a single organoid (Fig. 4e and Supplementary Fig. 11). Figure 4f,g4 \mathrm{f}, \mathrm{g} show the results plotted in different colors depending on the electrode height. The neural network map was also displayed in 3D space corresponding to the configuration of electrodes, representing the neural networks developed in this 3D volumetric organoid (Fig. 4h). In addition, the neural activity of this organoid along the electrode height was analyzed by the parameters of spiking rate and burst number (Fig. 4i, j, and Supplementary Fig. 12). The spiking rate detected at 50-50- and 200-mum200-\mu \mathrm{m}-high electrodes were higher than those detected at 100-mum100-\mu \mathrm{m}-high electrodes. On the other hand, the number of bursts, representing the mature activity of the neurons, showed negligible changes with electrode height. These observations indicated that neural activity of this cortical organoid varied longitudinally, while showing an equivalent maturation degree of neurons irrespective of these variances in neural activity. The neural activities from other organoids monitored by identical electrode configuration are shown in Supplementary Figs. 13, 14. Given the variability in cytoarchitecture between brain organoids, the results differ to other organoids. For instance, in Fig. 4i, the spiking rate of the 100 mum100 \mu \mathrm{~m}-height electrodes is lower than that of the 50 mum50 \mu \mathrm{~m} and 200 mum200 \mu \mathrm{~m} height-electrodes, whereas in Supplementary Fig. 13, the spiking rates increase with electrode height. A similar tendency was also observed in the analysis of the intraorganoid signals obtained using another 3D LM MEA with 2 different heights of electrodes (Supplementary Figs. 15-17). The variations in intra-organoid signals corresponding to the electrode heights are consistent with layer-specific expression of discriminative neural markers (Fig. 2f, g) and findings from a previous report ^(38){ }^{38}, which demonstrated distinct electrical properties for each sub-neuronal type constituting the cortical layer. These results indicate that the ability of our 3D LM MEA to detect various 3D coordinates within the organoid facilitates capturing variations in neural activities throughout the organoid, which is inconsistent across different organoids due to the randomly developed cytoarchitectures. Also, these analyses along the electrode height as well as 3D neural networking map provide insight into the neural circuitry between neurons across cortical layers, a capability that has not been achievable with previous approaches using surface-type MEAs. 由于 3D LM 电极的高度在直接打印过程中易于控制，因此可以在类器官内部体积中以多种配置方式定位电极，以研究神经活动随记录位点的变化。例如，我们打印了 3 种不同高度的 3D LM 柱状电极，以在单个类器官内的不同坐标记录单单位电位和尖峰信号（图 4e 和补充图 11）。图 4f,g4 \mathrm{f}, \mathrm{g} 根据不同电极高度以不同颜色显示结果。神经网络图也在与电极配置相对应的 3D 空间中展示，呈现了该 3D 体积类器官中发育的神经网络（图 4h）。此外，通过尖峰率和爆发次数参数分析了该类器官沿电极高度的神经活动（图 4i、j 和补充图 12）。在 50-50- 和 200-mum200-\mu \mathrm{m} 高度电极检测到的尖峰率高于 100-mum100-\mu \mathrm{m} 高度电极。另一方面，代表神经元成熟活动的爆发次数随电极高度变化可忽略不计。 这些观察结果表明，该皮质类器官的神经活动在纵向方向上存在变化，同时无论神经活动如何变化，神经元的成熟程度均保持一致。由相同电极配置监测的其他类器官的神经活动如补充图 13、14 所示。鉴于不同脑类器官的细胞结构存在差异，结果与其他类器官有所不同。例如，在图 4i 中， 100 mum100 \mu \mathrm{~m} 高度电极的放电频率低于 50 mum50 \mu \mathrm{~m} 和 200 mum200 \mu \mathrm{~m} 高度电极，而在补充图 13 中，放电频率随电极高度增加而上升。在使用另一种具有两种不同高度电极的 3D LM MEA 获取的类器官内信号分析中也观察到类似趋势（补充图 15-17）。类器官内信号随电极高度变化的现象与区分性神经标记物的层特异性表达（图 2f、g）及先前研究 ^(38){ }^{38} 的发现一致，该研究证实构成皮质层的各亚神经元类型具有独特的电生理特性。 这些结果表明，我们的 3D LM MEA 能够检测类器官内各种三维坐标的能力，有助于捕捉整个类器官神经活动的变化，由于随机发育的细胞结构，这种变化在不同类器官间并不一致。此外，沿电极高度方向的分析以及 3D 神经网络图谱，为跨皮质层神经元间的神经回路提供了见解，这是以往使用表面型 MEA 的方法所无法实现的。
Analysis of electrophysiological maturation of cortical organoids 皮质类器官电生理成熟度的分析
The maturation of intra-neural networks in cortical organoids was investigated using our 3D LM MEA. The minimal invasiveness and organoid-like softness of these electrodes can aid in long-term monitoring of the electrophysiological activities in cortical organoids. Additionally, the high flexibility of electrode configuration (position and height) through direct printing facilitates examination of the intraneural networks during their maturation, as the electrodes can be tailored to individual organoids that can be mature into a variety of morphologies. 利用我们的 3D LM MEA 研究了皮质类器官内神经网络的成熟过程。这些电极的微创性和类器官般的柔软特性，有助于长期监测皮质类器官的电生理活动。此外，通过直接打印实现电极配置（位置和高度）的高度灵活性，便于在神经网络成熟过程中对其进行检测，因为电极可以根据成熟为不同形态的单个类器官进行定制。
A brain organoid is composed of neurons that generate electrical action potentials, and glial cells that support the neurons and are involved in immune responses. These cells mature as the organoid is 脑类器官由产生电动作电位的神经元和支持神经元并参与免疫反应的胶质细胞组成。这些细胞随着类器官的成熟而
cultured and developed. Thus, the neural networking circuitry between cells becomes more complex and stronger. These features of cellular maturation lead to electrophysiological maturation, as presented in Fig. 5 and Supplementary Fig. 18. SU potentials were recorded during the maturation period of this cortical organoid (Fig. 5a), and its electrophysiological properties were analyzed using custom MATLAB codes (Fig. 5b-e). During its maturation from 2 to 6 months, the spiking rate increased (from 0.330+-0.241Hz0.330 \pm 0.241 \mathrm{~Hz} to 8.90+-5.87Hz8.90 \pm 5.87 \mathrm{~Hz} ) and inter-spike intervals (ISIs) decreased (from 3.832+-1.60s3.832 \pm 1.60 \mathrm{~s} to 0.755+-0.790s0.755 \pm 0.790 \mathrm{~s} ) as shown in Fig. 5b, c. Furthermore, bursts, repeatedly arising spikes over a minimum duration of 50 ms , were examined during the culture period from 2 to 6 months and both burst number (from 1.5+-1.51.5 \pm 1.5 to 9+-5.2099 \pm 5.209 ) and burst duration (from 0.991+-0.165s0.991 \pm 0.165 \mathrm{~s} to 1.490+-0.562s1.490 \pm 0.562 \mathrm{~s} ) increased at the same time (Fig. 5d, e). These results showed the similar trends reported in previous studies ^(26,39){ }^{26,39}, suggesting that bursts serve as an indirect indicator of the maturation degree in neural networking circuitry. Figure 5 f presents neural network maps of 2-, 4-, and 6-month-old organoids, showing the increase in complexity and connectivity among neurons, evidenced by a rising number of lines connecting nodes. Also, as presented in Fig. 5 g -j, the changes in properties related to neural networking circuitry were evaluated within the maturation period. The average number of connecting lines per node indicates the degree of connectivity of neurons (Fig. 5g). The total number of lines in the map, as well as the number of nodes with connecting lines, indicate the overall strength of neural networking circuitry throughout this organoid (Fig. 5h), and the number of nodes with connecting lines indicates the quantity of actively networking neurons (Fig. 5i). The increase in these properties over 2 to 6 months of maturation means that neural networking circuitry becomes stronger and more connected, with more neurons communicating interactively. As shown in Fig. 5j, the number of communities increased, while the number of nodes that did not belong to any community decreased. In addition, the average number and the maximum number of nodes included in a single community also increased as this organoid matured (Supplementary Fig. 19). As nodes within one community interact through more robust circuitry than the nodes outside this community, these results indicate that the strength and size of the neural community were enhanced by electrophysiological maturation of this organoid. Long-term monitoring of intra-organoid analysis using our 3D LM MEA can provide the comprehensive information required to fully understand the evolution of the functional activity of brain organoids. 培养和发育。因此，细胞间的神经网络回路变得更加复杂和强大。这些细胞成熟的特征导致了电生理成熟，如图 5 和补充图 18 所示。在该皮质类器官的成熟期记录了 SU 电位（图 5a），并使用自定义 MATLAB 代码分析了其电生理特性（图 5b-e）。在 2 至 6 个月的成熟过程中，如图 5b、c 所示，放电率增加（从 0.330+-0.241Hz0.330 \pm 0.241 \mathrm{~Hz} 到 8.90+-5.87Hz8.90 \pm 5.87 \mathrm{~Hz} ），放电间隔（ISIs）减少（从 3.832+-1.60s3.832 \pm 1.60 \mathrm{~s} 到 0.755+-0.790s0.755 \pm 0.790 \mathrm{~s} ）。此外，在 2 至 6 个月的培养期间，检测了爆发（在至少 50 毫秒内重复出现的放电），同时爆发次数（从 1.5+-1.51.5 \pm 1.5 到 9+-5.2099 \pm 5.209 ）和爆发持续时间（从 0.991+-0.165s0.991 \pm 0.165 \mathrm{~s} 到 1.490+-0.562s1.490 \pm 0.562 \mathrm{~s} ）均增加（图 5d、e）。这些结果显示出与先前研究 ^(26,39){ }^{26,39} 报告的相似趋势，表明爆发可作为神经网络回路成熟程度的间接指标。 图 5f 展示了 2、4、6 月龄类器官的神经网络图谱，通过节点间连接线数量的增加，显示出神经元间复杂性和连接性的提升。此外，如图 5g-j 所示，在成熟期内评估了与神经网络回路相关的特性变化。每个节点的平均连接线数量反映了神经元的连接程度（图 5g）。图谱中连接线总数及带连接线的节点数量，体现了整个类器官神经网络回路的总体强度（图 5h），而带连接线的节点数量则代表活跃参与网络连接的神经元数量（图 5i）。这些特性在 2 至 6 个月成熟期内的增强，意味着神经网络回路变得更强大且连接更紧密，更多神经元实现了交互通信。如图 5j 所示，群落数量增加，而不属于任何群落的节点数量减少。 此外，随着类器官的成熟，单个社群中包含的平均节点数和最大节点数也有所增加（补充图 19）。由于同一社群内的节点通过比社群外节点更强大的神经回路相互作用，这些结果表明神经社群的强度和规模均因类器官的电生理成熟而增强。利用我们的三维光微电极阵列进行长期类器官内部分析监测，可提供全面理解脑类器官功能活动演变所需的综合信息。
Magnetic reshaping of 3D LM MEAs for multi-spot detection of intra-organoid signals 三维光微电极阵列的磁性重塑用于类器官内信号的多点检测
Although conventional MEA platforms, including flat surface-type electrodes or 3D electrodes, have been extensively utilized for the electrophysiological analysis of neural tissues, limitations on the number and position of electrodes as well as the inability to adjust their structures remain disadvantages. In our study, by depositing a thin ferromagnetic cobalt (Co) layer (thickness: 120 nm ) only on half of the sidewall of each LM pillar before the parylene encapsulation (Supplementary Fig. 20), the 3D LM electrodes could be reshaped by applying external magnetic fields (Fig. 6a). This reshaping capability of the 3D LM electrodes exploited their inherent deformability and softness of EGaIn to prevent these pillars from cracking. To precisely control the tilting of electrodes, we constructed a magnetic tilting system with a 6 -axis stage (H-820 6-Axis Hexapod, Physik Instrumente, minimum displacement: 0.5 mum)0.5 \mu \mathrm{~m}). This system offers minute and uniform controllability over the direction and intensity of the magnetic field through software-based operation (see Supplementary Fig. 21 and Supplementary Movie 3). The magnet (NdFeB magnet, N52 grade, size: 5.5cmxx5.5cmxx2.5cm5.5 \mathrm{~cm} \times 5.5 \mathrm{~cm} \times 2.5 \mathrm{~cm} ) was mounted on the other xyz-axis linear stage (M-460P-XYZ, Newport Corporation), which was fixed on the 6 -axis stage. Additionally, by integrating the recording interface 尽管传统的 MEA 平台（包括平面型电极或 3D 电极）已广泛应用于神经组织的电生理分析，但电极数量和位置的限制以及无法调整其结构仍是其缺点。在我们的研究中，通过在聚对二甲苯封装前仅在每个 LM 柱体的一半侧壁沉积一层薄铁磁钴（Co）层（厚度：120 纳米），3D LM 电极可通过施加外部磁场进行重塑（图 6a）。这种 3D LM 电极的重塑能力利用了 EGaIn 固有的可变形性和柔软性，防止这些柱体破裂。为了精确控制电极的倾斜，我们构建了一个带有 6 轴平台（H-820 6 轴六足平台，Physik Instrumente，最小位移： 0.5 mum)0.5 \mu \mathrm{~m}) ）的磁倾斜系统。该系统通过基于软件的操作，提供了对磁场方向和强度的微小且均匀的控制（见补充图 21 和补充视频 3）。 磁体（钕铁硼磁体，N52 等级，尺寸： 5.5cmxx5.5cmxx2.5cm5.5 \mathrm{~cm} \times 5.5 \mathrm{~cm} \times 2.5 \mathrm{~cm} ）被安装在另一个 xyz 轴线性平台（M-460P-XYZ，Newport Corporation）上，该平台固定在 6 轴平台上。此外，通过集成记录接口
Fig. 6 | Magnetically reshapable 3D LM MEA. a Schematic illustration of magnetically reshapable 3D LM MEA. b SEM image (top) and EDS analysis (bottom) of partial magnetic layer on 3D LM electrode. Scale bars, 10 mum10 \mu \mathrm{~m}. This experiment was independently repeated more than ten times with similar results. c Magnetization of magnetically reshapable 3D LM electrode. d Trajectories of electrode tip during 20 repetitive tilting cycles. The color bar represents the cycle of repetitive tilting. e Magnetic tilting degrees (blue) and displacements (yellow) of the electrode tip for the repetitive cycles. f\mathbf{f} Expanded detectable area of the single tiltable electrode in 图 6 | 磁性可重构 3D 液态金属微电极阵列。a 磁性可重构 3D 液态金属 MEA 示意图。b 3D 液态金属电极表面局部磁性层的 SEM 图像（上）及 EDS 能谱分析（下）。比例尺， 10 mum10 \mu \mathrm{~m} 。本实验独立重复十余次，结果具有一致性。c 磁性可重构 3D 液态金属电极的磁化特性。d 电极尖端在 20 次重复倾斜运动中的轨迹。色条表示倾斜循环周期。e 电极尖端在重复循环中的磁倾角度（蓝色）与位移量（黄色）。 f\mathbf{f} 单侧可倾斜电极在
various directions. 多方向上的可探测区域扩展。g\mathbf{g}, h\mathbf{h} Multi-spot detection of intra-organoid signals by magnetically reshapable 3D LM MEA. SU potential, colored raster plot, and spikeforms when the degree of reshaping (theta)(\theta) is (g)8.2^(@)(\mathbf{g}) 8.2^{\circ} and (h)15.6^(@)(\mathbf{h}) 15.6^{\circ} are presented. Scale bars, 50 muV50 \mu \mathrm{~V} (vertical), 1s (horizontal) for single unit potential; 20 muV20 \mu \mathrm{~V} (vertical), 1 ms (horizontal) for spikeforms. Each color of spikeforms represent individual cluster of spikes. LM, liquid metal; MEA, multi-electrode array; SEM, scanning electron microscopy. EDS, energy-dispersive X-ray spectroscopy. SU, single-unit. g\mathbf{g} , h\mathbf{h} 通过磁性可重构 3D LM MEA 对类器官内信号进行多点检测。展示了当重构程度 (theta)(\theta) 为 (g)8.2^(@)(\mathbf{g}) 8.2^{\circ} 和 (h)15.6^(@)(\mathbf{h}) 15.6^{\circ} 时的 SU 电位、彩色光栅图和尖峰波形。比例尺：单单位电位垂直 50 muV50 \mu \mathrm{~V} ，水平 1 秒；尖峰波形垂直 20 muV20 \mu \mathrm{~V} ，水平 1 毫秒。尖峰波形的每种颜色代表不同的尖峰簇。LM，液态金属；MEA，多电极阵列；SEM，扫描电子显微镜。EDS，能量色散 X 射线光谱。SU，单单位。
(MZ60, Tucker-Davis Technologies Inc, USA) into the system, we enabled simultaneous recording of intra-organoid signals while tilting the electrodes magnetically. The scanning electron microscopy (SEM) image (top inset in Fig. 6b) and energy dispersive spectroscopy (EDS) analysis (bottom inset) verified this selective deposition of Co only on half of the pillar’s sidewall. Figure 6 c shows the magnetization of this 3D LM electrode evaluated by a vibrating sample magnetometer (VSM). As a demonstration, with the base of each pillar fixed to the substrate, the position of the electrode’s recording site within an organoid shifted according to the direction of the magnetic field, 将(MZ60, Tucker-Davis Technologies Inc, USA)接入系统后，我们实现了在磁力倾斜电极的同时对类器官内信号进行同步记录。扫描电子显微镜(SEM)图像（图 6b 顶部插图）和能量色散谱(EDS)分析（底部插图）证实了钴仅选择性沉积在柱状结构侧壁的一半区域。图 6c 通过振动样品磁强计(VSM)评估了该三维液态金属电极的磁化特性。作为演示，当每个柱状结构的基底固定在基板上时，电极在类器官内的记录位点位置会随磁场方向发生偏移。
achieved by tilting the electrode. To investigate the precise controllability of our magnetic tilting system, we theoretically studied the minimum degree of electrode deflection under our system, as shown in Supplementary Note 4 . Our system allows the electrode to tilt with a resolution of 0.002^(@)0.002^{\circ}, resulting in a minimum displacement of electrode tip of 0.009 mum0.009 \mu \mathrm{~m}. When examining the tilting trajectory of an electrode (height: 260 mum260 \mu \mathrm{~m}, magnet speed: 5mm//s5 \mathrm{~mm} / \mathrm{s} ) as shown in Supplementary Fig. 22, the tilting degrees of the electrode are linearly increased along the magnet displacement with a slope of 3.34 , which represents 0.00334^(@)0.00334^{\circ} of tilting per 1mum1 \mu \mathrm{~m} of magnet displacement. The 通过倾斜电极实现。为了研究我们磁倾斜系统的精确可控性，我们从理论上分析了该系统下电极的最小偏转角度，如补充说明 4 所示。我们的系统允许电极以 0.002^(@)0.002^{\circ} 的分辨率进行倾斜，从而使电极尖端的最小位移达到 0.009 mum0.009 \mu \mathrm{~m} 。在考察电极（高度： 260 mum260 \mu \mathrm{~m} ，磁铁速度： 5mm//s5 \mathrm{~mm} / \mathrm{s} ）的倾斜轨迹时（如补充图 22 所示），电极的倾斜角度随磁铁位移呈线性增加，斜率为 3.34，代表每 1mum1 \mu \mathrm{~m} 磁铁位移产生 0.00334^(@)0.00334^{\circ} 的倾斜。
displacements of electrode tip are also changed along the magnet displacement with a slope of 13.99,0.0140 mum13.99,0.0140 \mu \mathrm{~m} of electrode tip displacement per 1mum1 \mu \mathrm{~m} of magnet displacement, which is corresponded to our theoretical study. Owing to the magnetic field induced by the magnetic tilting system, the array of electrodes exhibited reliable tilting with constant tilting degrees (see Supplementary Fig. 23 and Supplementary Movie 4). Considering the size of neuron soma (10-12 mum\mu \mathrm{m} ) as shown in Fig. 3b, c, the electrode tip (electrically open for neural recording) must be shifted at least 24 mum24 \mu \mathrm{~m} to contact another group of neurons. Since the displacement of the electrode tip by magnetic tilting increases with the electrode height, the electrode’s capability to contact other groups of neurons depends on the electrode height under an identical magnetic tilting degree. For example, we calculated the minimum height of the electrode required to achieve a 24 mum24 \mu \mathrm{~m} displacement with 10^(@)10^{\circ} tilting, a feasible tilting degree, using the following Eq. (1); 电极尖端的位移也随磁体位移变化，斜率为每 1mum1 \mu \mathrm{~m} 磁体位移对应 13.99,0.0140 mum13.99,0.0140 \mu \mathrm{~m} 电极尖端位移，这与我们的理论研究相符。由于磁倾斜系统产生的磁场作用，电极阵列表现出可靠的倾斜，且倾斜角度恒定（参见补充图 23 和补充影片 4）。考虑到如图 3b、c 所示的神经元胞体大小（10-12 mum\mu \mathrm{m} ），电极尖端（用于神经记录的电开放端）需至少移动 24 mum24 \mu \mathrm{~m} 才能接触另一组神经元。由于磁倾斜引起的电极尖端位移随电极高度增加而增大，在相同倾斜角度下，电极接触其他神经元组的能力取决于电极高度。例如，我们通过以下公式(1)计算了在可行倾斜角度 10^(@)10^{\circ} 下实现 24 mum24 \mu \mathrm{~m} 位移所需的最小电极高度；
h=(d xx360^(@))/(2pi xx theta)h=\frac{d \times 360^{\circ}}{2 \pi \times \theta}
where h\boldsymbol{h} is the electrode height, d\boldsymbol{d} is the displacement of the electrode tip by tilting, and theta\boldsymbol{\theta} is the tilting degree. Although an electrode with a height of 138 mum138 \mu \mathrm{~m} can detect another group of neurons under 10^(@)10^{\circ} tilting, we printed 3D LM electrodes with a height of 260 mum260 \mu \mathrm{~m} for larger displacement to contact numerous neuron groups. To verify the reproducible magnetic tilting of 3D LM electrodes, we conducted 20 cycles of repetitive tilting, sufficient to monitor weekly electrophysiological signals from different neuron groups within a brain organoid over 5 months of maturation. The repetitive tilting was observed through a microscope camera (QImaging Micropublisher 5.0 RTV, Teledyne Photometrics) from the perspective of the xz-plane. The magnet was positioned directly above the 3D LM electrodes (distance between electrodes and the magnet: 2 cm ) and tilted the electrodes by shifting the magnet at speed of 5mm//s5 \mathrm{~mm} / \mathrm{s} until reaching the maximum tilting degree. Then, video analysis software (Tracker, Open Source Physics) was used to track the position of the electrode tip and obtain x,z\mathrm{x}, \mathrm{z} coordinates for calculating the tilting degree, the angle between the electrodes before and after tilting (Fig. 6d and Supplementary Movie 5). The 260 mum260 \mu \mathrm{~m} height electrode showed a maximum tilting degree of 35.5+-0.05^(@)35.5 \pm 0.05^{\circ} and a displacement of the electrode tip of 161 mum161 \mu \mathrm{~m} in a single direction along consistent trajectories over 20 cycles of repetitive tilting (Fig. 6d, e), implying reliable tilting to detect the same group of neurons over multiple cycles of tilting. Furthermore, magnetic tilting in both directions along the identical trajectory in the xz-plane was investigated using the same method. The 3D LM electrode was placed under the magnet with a spacing of 2 cm , and the magnet moved horizontally in both directions at 5mm//s5 \mathrm{~mm} / \mathrm{s} until reaching the maximum tilting in both directions. The total maximum tilting degree of 3D LM electrodes showed 51.6+-0.21^(@)51.6 \pm 0.21^{\circ}, and the displacement of the electrode tip was 234 mum234 \mu \mathrm{~m} over 20 cycles of repetitive tilting in both directions (Supplementary Fig. 24 and Supplementary Movie 6). These results indicate that accurate control of the magnetic field by the magnetic tilting system facilitated reproducible tilting of 3D LM electrodes and ensured reliable contact with the same neuron group over multiple tilting cycles. Furthermore, with the system’s facile magnet control, the 3D LM electrodes can be tilted in various directions (Supplementary Movie 7). By integrating data from 8 different trajectories into a single graph (Fig. 6f), we inferred the detectable area of a single electrode with magnetic tilting. As a result, the single electrode can cover an expanded area of 6.994 xx10^(4)mum^(2)6.994 \times 10^{4} \mu \mathrm{~m}^{2} within the organoid, indicating that the electrode exhibited about 891 times larger detectable area than that of an electrode without magnetic tilting. Considering the diameter of neurons ( 10-12 mum10-12 \mu \mathrm{~m} ) as shown in Fig. 3b and c, this expanded detectable area allows the electrode to reach other groups of neurons. 其中 h\boldsymbol{h} 为电极高度， d\boldsymbol{d} 为电极尖端因倾斜产生的位移， theta\boldsymbol{\theta} 为倾斜角度。虽然高度为 138 mum138 \mu \mathrm{~m} 的电极在 10^(@)10^{\circ} 倾斜下可检测另一组神经元，但我们打印了高度为 260 mum260 \mu \mathrm{~m} 的 3D LM 电极以实现更大位移，从而接触更多神经元群。为验证 3D LM 电极的可重复磁倾斜特性，我们进行了 20 次重复倾斜循环，足以在 5 个月的成熟期内监测脑器官中不同神经元群的每周电生理信号。通过显微镜摄像头（QImaging Micropublisher 5.0 RTV，Teledyne Photometrics）从 xz 平面视角观察重复倾斜过程。磁体直接置于 3D LM 电极上方（电极与磁体间距：2 厘米），并以 5mm//s5 \mathrm{~mm} / \mathrm{s} 速度移动磁体使电极倾斜，直至达到最大倾斜角度。随后使用视频分析软件（Tracker，Open Source Physics）追踪电极尖端位置，获取 x,z\mathrm{x}, \mathrm{z} 坐标以计算倾斜角度，即倾斜前后电极之间的夹角（图。 6d 和补充影片 5）。 260 mum260 \mu \mathrm{~m} 高度电极在 20 次重复倾斜循环中沿一致轨迹单方向的最大倾斜度为 35.5+-0.05^(@)35.5 \pm 0.05^{\circ} ，电极尖端位移为 161 mum161 \mu \mathrm{~m} （图 6d,e），表明其能可靠倾斜以在多次循环中检测同一神经元群。此外，采用相同方法研究了 xz 平面内沿相同轨迹的双向磁倾斜。将 3D LM 电极置于磁铁下方 2 厘米间距处，磁铁以 5mm//s5 \mathrm{~mm} / \mathrm{s} 速度双向水平移动直至达到双向最大倾斜。3D LM 电极在双向 20 次重复倾斜循环中的总最大倾斜度为 51.6+-0.21^(@)51.6 \pm 0.21^{\circ} ，电极尖端位移为 234 mum234 \mu \mathrm{~m} （补充图 24 和补充影片 6）。这些结果表明，磁倾斜系统对磁场的精确控制实现了 3D LM 电极的可重复倾斜，并确保在多次倾斜循环中与同一神经元群保持可靠接触。 此外，借助该系统便捷的磁控功能，3D 液态金属电极可朝不同方向倾斜（补充影片 7）。通过将 8 种不同轨迹的数据整合至同一图表（图 6f），我们推算出单个电极在磁倾条件下的可探测区域。结果表明，该单电极能覆盖类器官内约 14#的扩展区域，意味着其可探测面积较无磁倾状态的电极增大约 891 倍。结合图 3b 和 c 所示的神经元直径（15#），这一扩大的探测范围使电极能够触及其他神经元群组。
The detectable areas of the MEAs in previous studies were compared in Supplementary Table 1, validating the superior spatial resolution of our system’s recording sites. In addition, SEM images of 3D LM electrodes before and after magnetic tilting showed no vestige of leakage, which might occur by the physical deformation from magnetic tilting, showing the structural stability owing to intrinsic softness and self-healability of the LM (Supplementary Fig. 25). 补充表 1 对比了既往研究中微电极阵列的可探测区域，验证了本系统记录位点更优的空间分辨率。此外，磁倾前后 3D 液态金属电极的 SEM 图像显示无泄漏痕迹——这种物理形变可能由磁倾导致，从而证明了液态金属因其固有柔软性和自修复特性带来的结构稳定性（补充图 25）。
As the viscosity affects the tilting of electrodes, we studied the relationship between electrode tilting and medium viscosity, as shown in Supplementary Note 5^(40)5^{40}. The viscosity of the medium induces a drag force against the magnetic force exerted on the 3D LM electrode, resulting in a reduction in the tilting speed. To experimentally examine tilting behavior, we used a 0.6wt%0.6 \mathrm{wt} \% agarose gel instead of brain organoids, as imaging the inside of brain organoids is challenging due to their opacity. Since brain organoids exhibit similar mechanical properties and viscosity compared to the brain, we adopted this transparent 0.6wt%0.6 \mathrm{wt} \% agarose gel, commonly used as a brain phantom due to its similar mechanical properties, to substitute for brain organoids ^(41){ }^{41}. When tracking the positions of the electrode tip (height: 250 mum250 \mu \mathrm{~m} ) along the magnet movement at 3mm//s3 \mathrm{~mm} / \mathrm{s} in air and in the agarose gel (Supplementary Fig. 26 and Supplementary Movie 8), magnetic tilting in this gel (red dots) was slower than in air (blue dots), corresponding to the theoretical study. Although it took more time to overcome the drag force of the viscous medium, the electrode tips reached identical positions (when sufficient time is provided for the electrode to tilt) regardless of the medium’s viscosity due to the facile deformability of the 3D LM electrode and sufficient magnetic field to exert the driving force for tilting. This result assures reliable tilting of the electrodes within brain organoids, further demonstrated by the change in the neural recording results. 由于粘度会影响电极的倾斜，我们研究了电极倾斜与介质粘度之间的关系，如补充说明 5^(40)5^{40} 所示。介质的粘度会对 3D 液态金属电极施加的磁力产生阻力，导致倾斜速度降低。为了实验研究倾斜行为，我们使用 0.6wt%0.6 \mathrm{wt} \% 琼脂糖凝胶替代脑类器官，因为脑类器官的不透明性使得内部成像具有挑战性。由于脑类器官与大脑具有相似的机械特性和粘度，我们采用这种透明的 0.6wt%0.6 \mathrm{wt} \% 琼脂糖凝胶（因其相似的机械特性常被用作脑模型）来替代脑类器官 ^(41){ }^{41} 。当追踪电极尖端（高度： 250 mum250 \mu \mathrm{~m} ）在空气和琼脂糖凝胶中沿磁体移动（ 3mm//s3 \mathrm{~mm} / \mathrm{s} ）时的位置（补充图 26 和补充影片 8），该凝胶中的磁倾斜（红点）比在空气中（蓝点）更慢，这与理论研究相符。 尽管需要更多时间来克服粘性介质的阻力，但由于 3D LM 电极易于变形且磁场足以提供倾斜驱动力，电极尖端最终到达相同位置（在给予足够时间让电极倾斜的情况下），与介质粘度无关。这一结果确保了电极在脑类器官中的可靠倾斜，神经记录结果的变化进一步证实了这一点。
As the brain organoids are securely immobilized onto the substrate of 3D LM MEAs with the aid of cell-adhesive coating, the brain organoids do not shift under the magnetic tilting of electrodes within them (Supplementary Movie 9). Moreover, we assessed the potential damage caused by the magnetic tilting of 3D LM electrodes within brain organoids. In the magnetic tiling group, the organoids were subjected to an applied magnetic field, causing the electrodes to tilt at an angle of 20^(@)20^{\circ} for 3 hours. To evaluate the influence of this tilting on the viability of the brain organoids, we conducted quantitative real-time polymerase chain reaction (qPCR) analysis for the neuronal marker (MAP2), astrocyte marker (S100B), and apoptosis marker (Caspase-3). The results were compared with a control group that was not exposed to the magnetic field. As shown in Supplementary Fig. 27, the expression levels of MAP2, S100B, and CASP3 did not significantly differ between the magnetic tilting group and the control group. The low modulus and high mechanical deformability of LM electrodes contributed to minimizing the risk of damage to the brain organoids. In addition, the 3D visualization of the organoids both without and with magnetic tilting (electrodes at the tilted position) was performed to examine the damage that might be induced due to the 20 cycles of magnetic tilting (magnet speed of 5mm//s5 \mathrm{~mm} / \mathrm{s} for 4 seconds per each tilting). As shown in Supplementary Fig. 28, the magnetically tiltable electrodes remained intactly and the cytoarchitecture of the organoid was well preserved even after the 20 cycles of tilting. Additionally, a comparison of phosphorylated Tau ( p -Tau) expression, a marker of brain damage, confirms that the magnetic tilting of 3D LM electrodes within the organoid does not cause any damage to the inserted organoid while altering the interaction of the electrodes with neural cells, potentially allowing for a broader recording of neural network activity within the organoid. 在细胞粘附涂层的辅助下，脑类器官被牢固固定在 3D LM MEA 基底上，即使内部电极发生磁性倾斜，脑类器官也不会移位（补充影片 9）。此外，我们评估了 3D LM 电极在脑类器官内磁性倾斜可能造成的损伤。在磁性倾斜组中，类器官受到外加磁场作用，使电极以 20^(@)20^{\circ} 角度倾斜 3 小时。为评估这种倾斜对脑类器官活性的影响，我们对神经元标记物（MAP2）、星形胶质细胞标记物（S100B）和凋亡标记物（Caspase-3）进行了实时定量聚合酶链反应（qPCR）分析。结果显示，如补充图 27 所示，磁性倾斜组与对照组的 MAP2、S100B 和 CASP3 表达水平无显著差异。LM 电极的低模量和高机械变形能力有助于最大限度降低对脑类器官的损伤风险。 此外，还对未进行磁倾斜和进行磁倾斜（电极处于倾斜位置）的类器官进行了 3D 可视化，以检查 20 次磁倾斜循环（每次倾斜磁铁速度为 5mm//s5 \mathrm{~mm} / \mathrm{s} ，持续 4 秒）可能造成的损伤。如补充图 28 所示，磁可倾斜电极保持完好，类器官的细胞结构在 20 次倾斜循环后仍保存良好。此外，通过比较磷酸化 Tau（p-Tau）表达（一种脑损伤标志物）证实，类器官内 3D LM 电极的磁倾斜不会对插入的类器官造成任何损伤，同时改变了电极与神经细胞的相互作用，可能有助于更广泛地记录类器官内的神经网络活动。
As shown in Supplementary Fig. 29, the electrode impedance did not change significantly by Co deposition because the electrode tip was preserved without a Co coating. This impedance showed negligible changes under 20 cycles of repetitive magnetic tilting ( 20.697+-0.502kOmega20.697 \pm 0.502 \mathrm{k} \Omega to 20.781+-0.480kOmega20.781 \pm 0.480 \mathrm{k} \Omega ), assuring the stability of electrical performance of magnetically tiltable electrodes 如补充图 29 所示，由于电极尖端未形成钴涂层而得以保留，钴沉积并未显著改变电极阻抗。在 20 次重复磁倾斜循环（ 20.697+-0.502kOmega20.697 \pm 0.502 \mathrm{k} \Omega 至 20.781+-0.480kOmega20.781 \pm 0.480 \mathrm{k} \Omega ）下，该阻抗变化可忽略不计，确保了磁倾斜电极电性能的稳定性。
(Supplementary Fig. 30). Also, when we examined the influence of the magnetic field on intra-organoid signal recording, there was a slight increase in the noise level only during magnet movement (maximum noise level: ∼52 muV\sim 52 \mu \mathrm{~V} ), due to changes in the magnetic field. However, the noise level reduced to its initial state again after the movement ceased. Since our electrophysiological recordings were conducted only when the magnet remained stationary, the noise induced during the magnet movement did not affect these recordings significantly (Supplementary Fig. 31). For accurate analysis on the effect of magnetic field on firing rate by excluding any possible deformation of electrodes, we loaded brain organoid on 3D LM MEA with non-magnetic stationary electrodes. Significant changes in spiking rate along the existence of the magnetic field generated by the magnet were not observable, indicating that the magnetic field did not affect on the spontaneous neural activity of the organoids since the strength of the magnet of our system ( 285 mT on the surface) is not large enough to induce any changes in neural activities while deflecting magnetic tiltable electrodes (Supplementary Fig. 32). （补充图 30）。此外，当我们研究磁场对类器官内信号记录的影响时，仅在磁体移动期间由于磁场变化导致噪声水平略有增加（最大噪声水平： ∼52 muV\sim 52 \mu \mathrm{~V} ）。然而，移动停止后噪声水平又恢复到初始状态。由于我们的电生理记录仅在磁体保持静止时进行，磁体移动期间产生的噪声对这些记录没有显著影响（补充图 31）。为了准确分析磁场对放电频率的影响并排除电极可能发生的形变，我们将脑类器官装载在带有非磁性固定电极的 3D LM MEA 上。未观察到沿磁体产生的磁场存在时放电频率的显著变化，这表明磁场未影响类器官的自发神经活动，因为我们系统磁体的强度（表面 285 mT）不足以在偏转磁性可倾斜电极时引起神经活动的任何变化（补充图 32）。
Furthermore, Fig. 6g, h and Supplementary Fig. 33 demonstrate that a single electrode in MEA can collect electrophysiological signals from multiple points within a cortical organoid, effectively densifying the recording sites without needing additional electrodes. Although LFPs and SU potentials did not differ significantly depending on the degree of magnetic tilting, changes in spikeforms were evident due to this tilting. These changes indicated that when the electrode was tilted, it came into contact with a different group of neurons, each exhibiting a unique spikeform. The number of spike clusters was detected as one before tilting, and it became two clusters after tilting (Fig. 6g). By further tilting this electrode, different spikeforms were detected while maintaining the number of clusters (Fig. 6h). To further verify the multi-spot detection by magnetically tilted 3D LM electrodes, we fabricated the non-magnetic stationary electrodes and magnetically tiltable electrodes in a single substrate and recorded neural signals by them from a single organoid (Supplementary Fig. 34). The spiking rates of both non-magnetic electrodes and magnetic electrodes did not show specific tendency in response to the magnetic titling, indicating that the spiking rates reflect neural activities of brain organoids regardless of the magnetic tilting (Supplementary Fig. 35a). On the other hand, the spike waveform similarity of magnetically tiltable electrodes showed much lower value ( 0.740 ) compared to that of nonmagnetic stationary electrodes ( 0.911 ), showing the detection of different group of neurons by magnetic tilting (Supplementary Fig. 35b). These results indicate that the non-magnetic electrodes are not influenced by the magnetic field gradient generated by magnet movement, while magnetic electrodes shift their recording spots through the magnetic tilting. The change in SU spikes indicated alternation in contact groups of neurons by moving the recording position of the electrode. Therefore, this magnetic reshaping of our 3D LM MEA allows a single electrode to record from multiple spots within the brain organoid, which can be effective in forming high-density recording sites. We also examined intra-organoid signals to verify that the reproducible tiling of the electrode can detect the same group of neurons over 20 tilting cycles. The intra-organoid signals of a 4 -monthold organoid loaded onto the 3D LM MEA (electrode height: 260 mum260 \mu \mathrm{~m} ) were recorded during 20 repetitive tilting cycles. The magnet (spacing from MEA: 2 cm ) was moved horizontally at a speed of 5mm//s5 \mathrm{~mm} / \mathrm{s} for 4 seconds, corresponding to an electrode tip displacement of 120 mum120 \mu \mathrm{~m}, enabling contact with other groups of neurons. To validate the detection of the identical neuron group, similarities in spikeforms were calculated using Pearson’s correlation coefficient, as in previous studies ^(42){ }^{42}. When comparing the spikeforms between the first and the 20th tilting, the signal similarities were 0.923 and 0.973 at the initial position and tilted position, respectively. This indicates that reliable magnetic tilting facilitated reproducible detection of neuron groups over repetitive tilting cycles (Supplementary Fig. 36). 此外，图 6g、h 和补充图 33 表明，MEA 中的单个电极可以从皮质类器官内的多个点收集电生理信号，有效增加记录位点密度而无需额外电极。尽管 LFP 和 SU 电位未因磁倾斜程度而显著差异，但波形变化明显，表明电极倾斜时接触了不同神经元群，各自呈现独特波形。倾斜前检测到一个尖峰簇，倾斜后变为两个簇（图 6g）。进一步倾斜该电极时，在保持簇数量的同时检测到不同波形（图 6h）。为验证磁倾斜 3D LM 电极的多点检测能力，我们在同一基底上制备了非磁性固定电极和磁可倾斜电极，并记录单个类器官的神经信号（补充图 34）。 非磁性电极和磁性电极的尖峰频率均未显示出对磁倾斜的特定响应趋势，表明尖峰频率反映了脑类器官的神经活动，与磁倾斜无关（补充图 35a）。另一方面，可磁倾斜电极的尖峰波形相似度（0.740）远低于非磁性固定电极（0.911），表明磁倾斜检测到了不同的神经元群（补充图 35b）。这些结果表明，非磁性电极不受磁体运动产生的磁场梯度影响，而磁性电极通过磁倾斜改变了其记录位点。SU 尖峰的变化表明通过移动电极记录位置改变了神经元接触群。因此，我们 3D LM MEA 的这种磁性重塑使单个电极能够记录脑类器官内的多个位点，这对于形成高密度记录位点非常有效。 我们还检测了类器官内部信号，以验证电极的可重复倾斜能在 20 次倾斜循环中检测到同一群神经元。将一个 4 月龄类器官装载到 3D LM MEA（电极高度： 260 mum260 \mu \mathrm{~m} ）上，在 20 次重复倾斜循环期间记录其内部信号。磁铁（与 MEA 间距：2 厘米）以 5mm//s5 \mathrm{~mm} / \mathrm{s} 的速度水平移动 4 秒，对应电极尖端位移 120 mum120 \mu \mathrm{~m} ，使其能与其他神经元群接触。为验证同一神经元群的检测，如先前研究 ^(42){ }^{42} 所述，采用皮尔逊相关系数计算了波形相似性。比较首次与第 20 次倾斜的波形时，初始位置与倾斜位置的信号相似度分别为 0.923 和 0.973。这表明可靠的磁倾斜实现了神经元群在重复倾斜循环中的可重复检测（补充图 36）。
Discussion 讨论
We developed magnetically reshapable 3D LM MEAs for intra-organoid analysis. High-resolution direct printing of the biocompatible LMs enabled the formation of 3D soft electrodes with adjustable configurations of recording sites, leading to the collection of intra-organoid signals across a wide range of 3D coordinates within brain organoids. The softness and fine structure of our electrodes made them suitable for chronic monitoring of neural activity with minimal invasion into organoids. And the flexibility in placement and height of the 3D LM electrodes allowed for customization to address the specific 3D morphology of individual organoids. Our MEAs recorded intra-organoid signals with sufficient spatiotemporal resolutions to extract electrophysiological information such as LFPs, SU potentials, and spikeforms as well as neural circuitry developed in the 3D interior volume of organoids. Furthermore, we demonstrated the magnetic reshaping of 3D LM electrodes for recording from multiple sites within the organoid. This provides a solution that increases the recording density without having to fabricate additional electrodes. Our approach can offer an opportunity to precisely explore the electrophysiological dynamics of the brain and to clarify the origins of neurodegenerative disorders. For instance, integrating our 3D LM MEAs into diseasemodeled brain organoids can facilitate the discovery of dysfunctions in the neuropathological circuits of specific diseases. In addition, these MEAs can be utilized to monitor the responses of patient-driven organoids to drugs, further excavating patient-specific drugs. Nonetheless, several potential shortcomings should be addressed to expand the applicability of this technique. Firstly, the printing system uses a single nozzle, making it time-consuming for mass production. Additionally, since the liquid metal is printed under the ambient conditions, this method has structural constraints compared to other printing methods, such as direct laser writing or inkjet printing, which utilize photocurable, thermoset, or rigid materials. While the liquidous properties of EGaIn present these challenges, it also offers irreplaceable advantages including minimized tissue damage and robust magnetic tilting for high-resolution electrophysiological analysis of brain organoids. 我们开发了磁性可重构的 3D 液态金属微电极阵列（LM MEA），用于类器官内部分析。通过高分辨率直接打印生物相容性液态金属，形成了具有可调记录位点配置的 3D 柔性电极，从而能够在大脑类器官内的广泛 3D 坐标范围内采集内部信号。电极的柔软性和精细结构使其适合对类器官进行长期神经活动监测，且侵入性极小。3D 液态金属电极在位置和高度上的灵活性允许根据单个类器官的特定 3D 形态进行定制。我们的微电极阵列以足够的时空分辨率记录了类器官内部信号，可提取低频电位（LFPs）、单单元电位（SU potentials）和尖峰波形等电生理信息，以及类器官 3D 内部体积中发育的神经回路。此外，我们还展示了通过磁性重塑 3D 液态金属电极从类器官内多个位点进行记录的能力。这提供了一种无需制造额外电极即可提高记录密度的解决方案。 我们的方法为精确探索大脑的电生理动态和阐明神经退行性疾病的起源提供了机会。例如，将我们的 3D LM MEA 集成到疾病模型脑器官中，可以促进特定疾病神经病理回路功能障碍的发现。此外，这些 MEA 可用于监测患者驱动器官对药物的反应，进一步挖掘患者特异性药物。然而，为了扩大该技术的适用性，仍需解决几个潜在的缺点。首先，打印系统使用单喷嘴，使得大规模生产耗时。此外，由于液态金属在环境条件下打印，与其他打印方法（如直接激光写入或喷墨打印）相比，该方法存在结构限制，这些方法使用光固化、热固性或刚性材料。尽管 EGaIn 的液态特性带来了这些挑战，但它也提供了不可替代的优势，包括最小化组织损伤和强大的磁倾斜能力，用于脑器官的高分辨率电生理分析。
Furthermore, we note that challenges still remain, given the need for the development of reliable brain organoids that resemble the full functionality of the human brain. For example, the disparities that exist in neural activity and neuronal density between brain organoids and invivo brain tissue impede the acquisition of electrophysiological information from the organoids. Likewise, the results of our intraorganoid signal recording showed abated neural activities in amplitudes and spiking rates in comparison to in-vivo neural signals. In addition, insufficient delivery of oxygen and nutrients to the core of organoids becomes more problematic as their sizes increase, which hinders chronic analysis due to the reduced viability of the organoids. Therefore, in conjunction with the advancement in cultivating robust brain organoids, the use of 3D LM MEAs as an intra-organoid analysis platform provides substantial potential to expand the applicability of brain organoids by offering in-depth insights into their functional activities. 此外，我们注意到仍存在挑战，因为需要开发出能够模拟人脑全部功能的可靠脑类器官。例如，脑类器官与活体脑组织在神经活动和神经元密度上的差异阻碍了从类器官获取电生理信息。同样，我们的类器官内信号记录结果显示，与活体神经信号相比，其神经活动的振幅和峰值频率有所减弱。此外，随着类器官体积增大，其核心部位氧气和营养供应不足的问题愈发突出，这因类器官存活率降低而阻碍了长期分析。因此，结合培育强健脑类器官的技术进步，使用 3D 液态金属微电极阵列作为类器官内分析平台，通过深入探究其功能活动，为拓展脑类器官的应用潜力提供了重要可能。
Methods 方法
Liquid metal printing 液态金属打印
A glass capillary (Sutter Instrument) with an outer diameter of 1.0 mm and an inner diameter of 0.5 mm was pulled with a pipette puller ( P-1000\mathrm{P}-1000, Sutter Instrument) to prepare the nozzle which had an inner diameter of 18 mum18 \mu \mathrm{~m}. EGaIn ( 75.5%75.5 \% gallium, 24.5%24.5 \% indium alloy by weight; Changsha Santech Materials Co. Ltd.) was used as an ink for liquid metal printing. All printing steps of EGaln were monitored using a microscope camera (QImaging Micropublisher 5.0 RTV, Teledyne Photometrics) to control the nozzle position from a substrate using a 6-axis stage (H-820 6-Axis Hexapod, Physik Instrument) during the printing process. In the case of 18 mum18 \mu \mathrm{~m}-diameter nozzle, using a 使用外径 1.0 毫米、内径 0.5 毫米的玻璃毛细管（Sutter Instrument），通过移液管拉制仪（ P-1000\mathrm{P}-1000 ，Sutter Instrument）拉制出内径为 18 mum18 \mu \mathrm{~m} 的喷嘴。采用 EGaIn（ 75.5%75.5 \% 镓与 24.5%24.5 \% 铟按重量比的合金；长沙三特材料有限公司）作为液态金属打印的墨水。在打印过程中，使用显微镜摄像头（QImaging Micropublisher 5.0 RTV，Teledyne Photometrics）监控 EGaIn 的所有打印步骤，并通过六轴平台（H-820 6-Axis Hexapod，Physik Instrument）控制喷嘴与基板的位置。对于 18 mum18 \mu \mathrm{~m} 直径的喷嘴，利用
compressed dry air, the air pressure of 80 psi was applied to EGaIn to be extruded from the syringe (i.e., an ink reservoir) to the tip of a glass capillary nozzle before starting printing. 压缩干燥空气，在开始打印前对 EGaIn 施加 80 psi 的气压，使其从注射器（即墨水储存器）挤出至玻璃毛细管喷嘴尖端。
Fabrication of 3D LM MEA 三维液态金属微电极阵列的制备
The fabrication steps of the 3D LM MEA are as follows; (1) A slide glass was prepared in size of 4.5 xx4.5cm4.5 \times 4.5 \mathrm{~cm}. First, interconnect pads were formed by deposition of Cr//Au\mathrm{Cr} / \mathrm{Au} and Cr//Pt\mathrm{Cr} / \mathrm{Pt} through e-beam evaporator (Korea Vacuum Tech Co. Ltd) and photolithography. Then, LM interconnect lines and 3D LM pillars were printed. For passivation, parylene-C were patterned with a shadow mask to open the contact pad for recording system and electrode tip. 三维液态金属微电极阵列的制备步骤如下：(1)准备尺寸为 4.5 xx4.5cm4.5 \times 4.5 \mathrm{~cm} 的载玻片。首先，通过电子束蒸发器（韩国真空技术有限公司）和光刻技术沉积 Cr//Au\mathrm{Cr} / \mathrm{Au} 和 Cr//Pt\mathrm{Cr} / \mathrm{Pt} 形成互连焊盘。随后，印刷液态金属互连线路及三维液态金属柱。为进行钝化处理，采用阴影掩模对聚对二甲苯-C 进行图案化，以开放记录系统与电极尖端的接触焊盘。
Sidewall passivation of 3D LM pillar 三维液态金属柱的侧壁钝化
After the printing of 3D LM pillars, 2mum2 \mu \mathrm{~m} of parylene-C was deposited on whole surface of the device. Parylene-C on the tip of the pillars was selectively dry-etched by reactive ion etching system with the condition of 200W,240200 \mathrm{~W}, 240 seconds. 完成三维液态金属柱印刷后，在整个器件表面沉积 2mum2 \mu \mathrm{~m} 厚度的聚对二甲苯-C。通过反应离子刻蚀系统在 200W,240200 \mathrm{~W}, 240 秒条件下选择性干法刻蚀柱尖端的聚对二甲苯-C。
Electrodeposition of platinum nanoparticles 铂纳米粒子的电沉积
For preparing 25 mL of an electrodepostion solution, 25 mL of isopropyl alcohol, 5 mg of lead acetate trihydrate (Sigma-Aldrich), and 0.25 g of platinum tetrachloride (Sigma-Aldrich) were mixed at room temperature. This electrodeposition solution was stirred under ultrasonic vibration for 20 minutes and filtered to remove impurities. The electroplating was performed by ion transfer between the cathode and anode in the Pt electrodeposition solution. A cathode (the 3D LM electrode) and an anode (Pt electrode) were immersed in this electrodeposition solution, and each was connected to a source meter (Keithley 2400, Tektronix). The electroplating reaction occurred under an electrical voltage of 2 V and compliance current of 0.1 mA for 120 seconds. 为制备 25 毫升电沉积溶液，将 25 毫升异丙醇、5 毫克三水合乙酸铅（Sigma-Aldrich）和 0.25 克四氯化铂（Sigma-Aldrich）在室温下混合。该电沉积溶液在超声振动下搅拌 20 分钟并过滤以去除杂质。电镀通过在铂电沉积溶液中的阴极和阳极之间的离子转移进行。将阴极（3D LM 电极）和阳极（铂电极）浸入此电沉积溶液中，并分别连接到源表（Keithley 2400, Tektronix）。电镀反应在 2 伏电压和 0.1 毫安合规电流下进行 120 秒。
Attachment of elastomer wells 弹性体井的附着
Elastomer wells were fabricated with polydimethylsiloxane (PDMS). For the macrowell, uncured PDMS was poured in square petri dish with 2 cm -thickness. Once PDMS was cured, 19 mm -diameter hole was created with a punch. For the microwell, micromold was designed with 3D printer (Form3, Formlabs), and uncured PDMS was poured onto the mold. Cured microwell was attached on the 3D LM MEA, and macrowell was attached with the uncured PDMS as an adhesive material. 弹性体孔槽采用聚二甲基硅氧烷（PDMS）制备。对于大孔槽，将未固化的 PDMS 倒入边长为 2 厘米的正方形培养皿中。待 PDMS 固化后，用打孔器制作直径为 19 毫米的孔洞。对于微孔槽，使用 3D 打印机（Form3，Formlabs）设计微模具，并将未固化的 PDMS 浇注到模具上。固化后的微孔槽附着在 3D 液态金属微电极阵列上，大孔槽则用未固化的 PDMS 作为粘合材料固定。
Impedance spectroscopy 阻抗谱分析
The impedance measurements of the 3D LM electrodes were conducted in a PBS solution (Sigma-Aldrich). All impedance measurements were performed over a frequency range of 0.01 to 100 kHz using a multichannel potentiostat (PMC-1000, AMETEK). 3D 液态金属电极的阻抗测量在磷酸盐缓冲盐水溶液（PBS，Sigma-Aldrich）中进行。所有阻抗测量均在 0.01 至 100 千赫兹的频率范围内使用多通道恒电位仪（PMC-1000，AMETEK）完成。
hiPSC maintenance 人诱导多能干细胞培养维持
hiPSC line KYOU-DXR0109B (#ACS-1023; American Type Culture Collection, Manassas, VA, USA) was used for cortical organoid generation ^(43,44){ }^{43,44}. The use of hiPSCs was approved by the Institutional Review Board (IRB) of Yonsei University (Permit Number: 7001988-202309-BR-1066-02E). hiPSCs were cultured on Matrigel-coated dishes (#354277, Corning) with mTeSR-Plus (#5825, STEMCELL Technologies) and maintained at passages 20 to 45 using ReLeSR (#5872, STEMCELL Technologies). The cells were regularly checked by MycoAlert PLUS Mycoplasma Detection Kit (#LT07-705, Lonza) for mycoplasma contamination and the infection was prevented by MycoGuard ^("TM "){ }^{\text {TM }} Mycoplasma Elimination Reagent (#SMD022, Biomax). 采用人诱导多能干细胞系 KYOU-DXR0109B（编号 ACS-1023，美国典型培养物保藏中心，弗吉尼亚州马纳萨斯）进行皮质类器官生成 ^(43,44){ }^{43,44} 。该 hiPSCs 的使用已获得延世大学机构审查委员会（IRB）批准（许可号：7001988-202309-BR-1066-02E）。hiPSCs 培养于 Matrigel 包被培养皿（货号 354277，康宁公司）中，使用 mTeSR-Plus 培养基（货号 5825，STEMCELL Technologies 公司），并通过 ReLeSR（货号 5872，STEMCELL Technologies 公司）维持在 20 至 45 代。定期使用 MycoAlert PLUS 支原体检测试剂盒（货号 LT07-705，龙沙公司）检测支原体污染，并采用 MycoGuard ^("TM "){ }^{\text {TM }} 支原体清除试剂（货号 SMD022，Biomax 公司）预防感染。
Generation of cortical organoids 皮质类器官的生成
Cortical organoids were generated and cultured with a slightly modified protocol from a previous study ^(33,45){ }^{33,45}. hiPSCs were dissociated into 皮质类器官的生成与培养采用基于先前研究 ^(33,45){ }^{33,45} 的改良方案。将 hiPSCs 解离为
single cells with TrypLE Express Enzyme (#12604013, Thermo Fisher Scientific) and transferred to ultra-low attachment 96 -well plate (#CLS7007, Corning), with each well containing 10,000 cells in mTeSR-Plus medium supplemented with the ROCK inhibitor Y-27632 ( 20 muM20 \mu \mathrm{M}; #1293823, BioGems). For the first 5 days, the medium was changed every day and supplemented with dorsomorphin (2.5 muM(2.5 \mu \mathrm{M}; #P5499, Sigma), SB-431542 ( 10 muM10 \mu \mathrm{M}; #1614, Tocris), and XAV939 ( 2.5 muM2.5 \mu \mathrm{M}; # X3004, Sigma). On the sixth day in suspension, neural spheroids were transferred to ultra-low attachment 24 -well plate (#cls3473, Corning) in neural medium containing neurobasal-A (#0888-022, Thermo Fisher Scientific), 1X B-27 supplement without vitamin A (#12587010, Thermo Fisher Scientific), 1X GlutaMax (#35050061, Thermo Fisher Scientific), 1X Penicillin/Streptomycin (#15140122, Thermo Fisher Scientific) and supplemented with the epidermal growth factor (EGF; 20ngml^(-1)20 \mathrm{ng} \mathrm{ml}^{-1}; #AF100-15, Peprotech) and fibroblast growth factor 2 (FGF2; 20ngml^(-1);#100-18 B20 \mathrm{ng} \mathrm{ml}^{-1} ; \# 100-18 B, Peprotech) until day 24. From day 25 to 42 , the medium was supplemented with brain-derived neurotrophic factor (BDNF; 20ngml^(-1),#450-0220 \mathrm{ng} \mathrm{ml}^{-1}, \# 450-02, Peprotech) and neurotrophin 3 (NT3; 20ngml^(-1)20 \mathrm{ng} \mathrm{ml}^{-1}; #450-03, Peprotech). Medium was changed every other day. From day 43 onward, cortical organoids were maintained in neurobasal-A medium supplemented with 1X B-27 supplement without vitamin A with medium changes every 4-6 days. 使用 TrypLE Express 酶（#12604013，Thermo Fisher Scientific）分离单细胞，并将其转移至超低吸附 96 孔板（#CLS7007，Corning），每孔在添加 ROCK 抑制剂 Y-27632（ 20 muM20 \mu \mathrm{M} ；#1293823，BioGems）的 mTeSR-Plus 培养基中包含 10,000 个细胞。前 5 天每天更换培养基，并添加 dorsomorphin（ (2.5 muM(2.5 \mu \mathrm{M} ；#P5499，Sigma）、SB-431542（ 10 muM10 \mu \mathrm{M} ；#1614，Tocris）和 XAV939（ 2.5 muM2.5 \mu \mathrm{M} ；#X3004，Sigma）。悬浮培养第 6 天，将神经球转移至超低吸附 24 孔板（#cls3473，Corning），使用含 Neurobasal-A（#0888-022，Thermo Fisher Scientific）、1X 不含维生素 A 的 B-27 补充剂（#12587010，Thermo Fisher Scientific）、1X GlutaMax（#35050061，Thermo Fisher Scientific）、1X 青霉素/链霉素（#15140122，Thermo Fisher Scientific）的神经培养基，并添加表皮生长因子（EGF； 20ngml^(-1)20 \mathrm{ng} \mathrm{ml}^{-1} ；#AF100-15，Peprotech）和成纤维细胞生长因子 2（FGF2； 20ngml^(-1);#100-18 B20 \mathrm{ng} \mathrm{ml}^{-1} ; \# 100-18 B ，Peprotech）培养至第 24 天。从第 25 天至第 42 天，培养基中添加脑源性神经营养因子（BDNF； 20ngml^(-1),#450-0220 \mathrm{ng} \mathrm{ml}^{-1}, \# 450-02 ，Peprotech）和神经营养素 3（NT3； 20ngml^(-1)20 \mathrm{ng} \mathrm{ml}^{-1} ；#450-03，Peprotech）。 培养基每隔一天更换一次。从第 43 天起，皮质类器官在添加了不含维生素 A 的 1X B-27 补充剂的神经基础-A 培养基中维持，每 4-6 天更换一次培养基。
Immunohistochemistry 免疫组织化学
The cortical organoids were fixed in 10% formalin solution (#HT501640, Sigma) for two hours at room temperature and washed in phosphate-buffered saline (PBS, Biosesang, Seongnam, Korea). Subsequently, they were immersed in 1:1(v//v)1: 1(\mathrm{v} / \mathrm{v}) OCT compound (#HCP-0100-00A, CellPath):30% sucrose (#84097, Sigma) dissolved in PBS overnight at 4^(@)C4{ }^{\circ} \mathrm{C} for cryoprotection. The organoids were embedded in OCT compound and frozen in liquid nitrogen, and sectioned at 10-15mum\mu \mathrm{m}-thickness using a cryostat (Leica). Cryosections were washed with PBS to remove excess OCT compound and permeabilized with 0.25%0.25 \% ( v//v\mathrm{v} / \mathrm{v} ) Triton X-100 (#X100, Sigma) in PBS for 20 min . Then, the sections were treated with 4%(w//v)4 \%(\mathrm{w} / \mathrm{v}) bovine serum albumin (#216006980, MP Biomedicals), 2%(v//v)2 \%(\mathrm{v} / \mathrm{v}) horse serum (#16050130, Thermo Fisher Scientific), and 0.02%(v//v)0.02 \%(\mathrm{v} / \mathrm{v}) Triton X-100\mathrm{X}-100 in PBS for 1 h at room temperature, and incubated with primary antibodies diluted in PBS overnight at 4^(@)C4^{\circ} \mathrm{C}. The stained samples were then washed with PBS and incubated with Alexa Fluor 488, 555, or 594-conjugated secondary antibodies in PBS (1:200; Thermo Fisher Scientific) for 1 h at room temperature. The nuclei were counterstained with 4^('),64^{\prime}, 6-diamidino-2-phenylindole (DAPI; #A2412, TCI America) for 20 min and washed with PBS. The samples were mounted using a fluorescent mounting medium (#H1400, Vector laboratories). Images were acquired with a confocal microscope (LSM 880, Carl Zeiss) and processed with ImageJ (National Institutes of Health). The following primary antibodies were used for immunohistochemistry: anti-TUJ1 (mouse, 1:500, #801213, Biolegend), anti-SOX2 (rabbit, 1:200, #AB5603, Millipore), anti-MAP2 (rabbit, 1:100, #4542S, Cell Signaling Technology), anti-GFAP (mouse, 1:200, #MAB3402, Millipore), anti-VGLUT1 (rabbit, 1:500, #135 302, Synaptic Systems), anti-CTIP2 (rat, 1:250, #ab18465, Abcam), and anti-SATB2 (mouse, 1:25, #ab51502, Abcam). 皮质类器官在室温下用 10%福尔马林溶液（#HT501640，Sigma）固定两小时，并用磷酸盐缓冲盐水（PBS，Biosesang，韩国城南）洗涤。随后，它们浸入 1:1(v//v)1: 1(\mathrm{v} / \mathrm{v}) OCT 化合物（#HCP-0100-00A，CellPath）与 30%蔗糖（#84097，Sigma）溶解于 PBS 的溶液中，在 4^(@)C4{ }^{\circ} \mathrm{C} 过夜进行冷冻保护。类器官被包埋在 OCT 化合物中，并在液氮中冷冻，使用冷冻切片机（Leica）切成 10-15 mum\mu \mathrm{m} 厚度的切片。冷冻切片用 PBS 洗涤以去除多余的 OCT 化合物，并用 0.25%0.25 \% （ v//v\mathrm{v} / \mathrm{v} ）Triton X-100（#X100，Sigma）在 PBS 中渗透 20 分钟。然后，切片在室温下用 4%(w//v)4 \%(\mathrm{w} / \mathrm{v}) 牛血清白蛋白（#216006980，MP Biomedicals）、 2%(v//v)2 \%(\mathrm{v} / \mathrm{v}) 马血清（#16050130，Thermo Fisher Scientific）和 0.02%(v//v)0.02 \%(\mathrm{v} / \mathrm{v}) Triton X-100\mathrm{X}-100 在 PBS 中处理 1 小时，并在 4^(@)C4^{\circ} \mathrm{C} 与稀释于 PBS 中的一抗孵育过夜。染色后的样品用 PBS 洗涤，并在室温下与 Alexa Fluor 488、555 或 594 标记的二抗（1:200；Thermo Fisher Scientific）在 PBS 中孵育 1 小时。 细胞核用 4^('),64^{\prime}, 6 -二脒基-2-苯基吲哚（DAPI；#A2412，TCI America）复染 20 分钟，并用 PBS 洗涤。样品使用荧光封片剂（#H1400，Vector laboratories）封片。图像通过共聚焦显微镜（LSM 880，Carl Zeiss）采集，并用 ImageJ（美国国立卫生研究院）处理。免疫组化使用了以下一抗：抗-TUJ1（小鼠，1:500，#801213，Biolegend）、抗-SOX2（兔，1:200，#AB5603，Millipore）、抗-MAP2（兔，1:100，#4542S，Cell Signaling Technology）、抗-GFAP（小鼠，1:200，#MAB3402，Millipore）、抗-VGLUT1（兔，1:500，#135 302，Synaptic Systems）、抗-CTIP2（大鼠，1:250，#ab18465，Abcam）和抗-SATB2（小鼠，1:25，#ab51502，Abcam）。
Assessment of biocompatibility 生物相容性评估
The viability of cortical organoids with 3D LM electrodes was measured using Live/Dead assay kit (Thermo Fisher Scientific). The staining procedure was conducted following the instructions provided by the manufacturer. The cortical organoids were washed with PBS and stained with Live/Dead solution consisting of 2muM2 \mu \mathrm{M} calcein-AM (to label live cells) and 4muM4 \mu \mathrm{M} ethidium homodimer- 1 (to label dead cells) with incubation at 37^(@)C37^{\circ} \mathrm{C} for 60 minutes, followed by washing with PBS. Fluorescent images of cortical organoids were obtained using a confocal microscope (LSM 880), and organoid viability was assessed 使用 Live/Dead 检测试剂盒（Thermo Fisher Scientific）测量了配备 3D LM 电极的皮质类器官的存活率。染色步骤按照制造商提供的说明进行。皮质类器官经 PBS 洗涤后，用含 2muM2 \mu \mathrm{M} 钙黄绿素-AM（标记活细胞）和 4muM4 \mu \mathrm{M} 乙锭同型二聚体-1（标记死细胞）的 Live/Dead 溶液染色，在 37^(@)C37^{\circ} \mathrm{C} 条件下孵育 60 分钟，随后用 PBS 冲洗。通过共聚焦显微镜（LSM 880）获取皮质类器官的荧光图像，并评估其存活率
based on Live/Dead stained images, and analyzed with ImageJ software. 基于 Live/Dead 染色图像，并利用 ImageJ 软件进行分析。
Real-time quantitative PCR analysis 实时定量 PCR 分析
mRNA was isolated using TaKaRa MiniBEST Universal RNA Extraction Kit (#9767 A; TaKaRa Bio Inc.) and template cDNA was synthesized from the extracted mRNA using TaKaRa PrimeScript II First-strand cDNA synthesis kit (#6110 A; TaKaRa). Subsequently, qPCR analysis was conducted with the cDNA samples, TaqMan Fast Universal PCR MasterMix (#4366073; Thermo), and the following TaqMan gene expression assay kits: PAX6 (Hs00240871_m1), Nestin (Hs04187831_g1), TUBB3 (Hs00801390_s1), MAP2 (Hs00258900_m1), S100B (Hs00902901_m1), and CASP3 (Hs_00234387_m1) using a StepOnePlus Real-Time PCR System (Applied Biosystems). The relative gene expression of each target was calculated by the comparative Ct method and normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH; HsO2786624_g1). 使用 TaKaRa MiniBEST Universal RNA 提取试剂盒（#9767 A；TaKaRa Bio Inc.）分离 mRNA，并采用 TaKaRa PrimeScript II 第一链 cDNA 合成试剂盒（#6110 A；TaKaRa）从提取的 mRNA 合成模板 cDNA。随后，利用 cDNA 样本、TaqMan Fast Universal PCR MasterMix（#4366073；Thermo）及以下 TaqMan 基因表达检测试剂盒：PAX6（Hs00240871_m1）、Nestin（Hs04187831_g1）、TUBB3（Hs00801390_s1）、MAP2（Hs00258900_m1）、S100B（Hs00902901_m1）和 CASP3（Hs_00234387_m1），在 StepOnePlus 实时 PCR 系统（Applied Biosystems）上进行 qPCR 分析。各靶基因的相对表达量通过比较 Ct 法计算，并以甘油醛-3-磷酸脱氢酶（GAPDH；HsO2786624_g1）作为内参进行标准化。
3D imaging of the cleared cortical organoids 透明化皮质类器官的 3D 成像
3D imaging of the cleared cortical organoids was performed using a slightly modified protocol from previous studies ^(46,47){ }^{46,47}. The organoids were fixed in 10%10 \% formalin solution (#HT501640, Sigma) for two hours at room temperature and washed in phosphate-buffered saline (PBS, Biosesang, Seongnam, Korea). The fixed organoids were washed with a washing buffer and treated with a blocking buffer for 8 h at 37^(@)C37^{\circ} \mathrm{C}. Next, the organoids were incubated with primary antibodies for 67 h at 37^(@)C37^{\circ} \mathrm{C}. After washing with the washing buffer, the organoids were incubated with secondary antibodies and DAPI solution for 68 h at 37^(@)C37^{\circ} \mathrm{C}. Following this, the organoids were washed again with the washing buffer and cleared by replacing the Ce3D solution overnight. Images were acquired using a confocal microscope (LSM 700 and 880, Carl Zeiss) and processed with ImageJ (National Institutes of Health). 采用略微改进自先前研究 ^(46,47){ }^{46,47} 的协议对透明化皮质类器官进行 3D 成像。类器官在室温下用 10%10 \% 福尔马林溶液（#HT501640，Sigma）固定两小时，并用磷酸盐缓冲液（PBS，Biosesang，韩国城南市）清洗。固定后的类器官用洗涤缓冲液冲洗，并在 37^(@)C37^{\circ} \mathrm{C} 条件下用封闭缓冲液处理 8 小时。随后，类器官在 37^(@)C37^{\circ} \mathrm{C} 条件下与一抗孵育 67 小时。洗涤缓冲液冲洗后，类器官与二抗及 DAPI 溶液在 37^(@)C37^{\circ} \mathrm{C} 条件下孵育 68 小时。此后，类器官再次用洗涤缓冲液洗涤，并通过 Ce3D 溶液置换过夜实现透明化。使用共聚焦显微镜（LSM 700 和 880，Carl Zeiss）采集图像，并用 ImageJ（美国国立卫生研究院）进行处理。
Electrophysiological recording 电生理记录
For the data analysis of LFPs and SU potentials, an electrophysiological recording system was used, which consisted of a RZ2 amplifier processor, PZ5 Neurodigitizer, MZ60 MEA interface (Tucker-Davis Technologies Inc, USA), and a computer with Synapse program. A sampling rate of 24,414Hz24,414 \mathrm{~Hz} and 60 Hz notch were used during recording. Mostly 0.1-300Hz0.1-300 \mathrm{~Hz} bandpass filter was used for recording LFPs, and a 300-3,000Hz300-3,000 \mathrm{~Hz} bandpass filter was used for recording single-unit spikes. For long-term monitoring of cortical organoids, the cortical organoids of 2, 4, and 6 months were placed on the 3D MEAs respectively, and the neural signals from them were recorded. 针对局部场电位（LFPs）和单单元电位的数据分析，采用了由 RZ2 放大器处理器、PZ5 神经数字转换器、MZ60 多电极阵列接口（美国 Tucker-Davis Technologies Inc 公司）及搭载 Synapse 软件的计算机组成的电生理记录系统。记录过程中使用 24,414Hz24,414 \mathrm{~Hz} 采样率并施加 60Hz 陷波滤波。多数情况下，采用 0.1-300Hz0.1-300 \mathrm{~Hz} 带通滤波器记录 LFPs，而单单元尖峰记录则使用 300-3,000Hz300-3,000 \mathrm{~Hz} 带通滤波器。为长期监测皮质类器官，将 2、4、6 月龄的皮质类器官分别置于 3D 多电极阵列上，并记录其神经信号。
Fabrication of magnetically reshapable 3D LM MEA 可磁性重塑 3D 液态金属多电极阵列的制备
Magnetically reshapable 3D LM electrodes were fabricated by deposition of a ferromagnetic layer directly on the electrodes. Before printing 3D LM pillars on the substrate, shadow mask was positioned upon the substrate in order to form magnetic layer only on the 3D LM pillars. The ferromagnetic layer (cobalt) was vacuum deposited only on the half side of LM pillars by tilting the substrate during the deposition. The later processes were proceeded same as the fabrication process of 3D LM MEA. 可磁性重塑 3D 液态金属电极通过在电极表面直接沉积铁磁层制成。在基底上打印 3D 液态金属柱前，先放置阴影掩模以确保磁性层仅形成于柱体部分。通过倾斜基底进行真空沉积，使铁磁层（钴）仅覆盖液态金属柱的半侧。后续工艺流程与 3D 液态金属多电极阵列的制备过程相同。
VSM analysis 振动样品磁强计分析
For quantitative analysis, the magnetic hysteresis loops were measured with a vibrating sample magnetometer (VSM, 7404-S, Lake Shore Cryotronics) for all samples. Since Co coated EGaIn was in a liquid state, an appropriate amount ( ∼100mg\sim 100 \mathrm{mg} ) was trapped in the cartridge container and fixed to a sample holder to prevent unintended vibration. 为进行定量分析，所有样品均使用振动样品磁强计（VSM，型号 7404-S，Lake Shore Cryotronics）测量磁滞回线。由于钴包覆的 EGaIn 呈液态，需将适量（ ∼100mg\sim 100 \mathrm{mg} ）样品捕获于卡槽容器中并固定于样品支架上，以避免非预期振动。
Magnetic tilting system 磁倾角系统
For precise control of the magnetic field for accurate tilting of electrodes, a magnet (NdFeB magnet, N 52 grade, size: 5.5cmxx5.5cmxx5.5 \mathrm{~cm} \times 5.5 \mathrm{~cm} \times 为实现电极精确倾斜的磁场精准控制，采用钕铁硼磁体（N52 等级，尺寸： 5.5cmxx5.5cmxx5.5 \mathrm{~cm} \times 5.5 \mathrm{~cm} \times
2.5 cm ) is mounted on an xyz-axis linear stage (M-460P-XYZ, Newport Corporation), which is fixed on the 6 -axis stage ( H-820\mathrm{H}-820 6-Axis Hexapod, Physik Instrumente). The movements of the 6 -axis stage (minimum displacement: 500 nm ) are controlled by software (LabVIEW, National Instruments). Furthermore, the MEA interface (MZ60, Tucker-Davis Technologies Inc.) is integrated into this system for neural recording of brain organoids. 2.5 厘米）安装于 xyz 轴线性平台（M-460P-XYZ，Newport Corporation），该平台固定于六轴位移台（ H-820\mathrm{H}-820 6-Axis Hexapod，Physik Instrumente）。六轴台最小位移精度为 500 纳米，其运动由 LabVIEW（National Instruments）软件控制。此外，该系统还集成了 MEA 接口（MZ60，Tucker-Davis Technologies Inc.）用于脑类器官的神经信号记录。
Analysis of magnetic tilting of 3D LM electrode 三维液态金属电极磁倾斜分析
The magnetic tilting of 3D LM electrodes was recorded using a microscope camera (QImaging Micropublisher 5.0 RTV, Teledyne Photometrics) from the perspective of the xz-plane. Subsequently, the tilting was analyzed using a video analysis tool (Tracker, Open Source Physics). The electrode tips were targeted for tracking, providing x,zx, z coordinates. The tilting degree was then calculated using the following Eq. (2); 通过显微镜摄像头（QImaging Micropublisher 5.0 RTV，Teledyne Photometrics）从 xz 平面视角记录三维液态金属电极的磁倾斜现象，随后使用视频分析工具（Tracker，开源物理项目）对倾斜进行分析。电极尖端被选为追踪目标，提供 x,zx, z 坐标位置。倾斜角度随后通过以下公式(2)计算得出；
where theta\boldsymbol{\theta} is the tilting degree, x_(1)x_{1} is the xx-coordinate of the electrode tip before tilting, and x_(2)x_{2} and z_(2)z_{2} are the x -, z -coordinates of the electrode tip after tilting, respectively. 其中 theta\boldsymbol{\theta} 表示倾斜角度， x_(1)x_{1} 为倾斜前电极尖端的 xx 坐标， x_(2)x_{2} 和 z_(2)z_{2} 分别为倾斜后电极尖端的 x、z 坐标。
Data analysis 数据分析
Spike detection and PCA clustering. All analyses of neural recording data were done using MATLAB R2022b with four open-source toolboxes (Statistics and Machine Learning Toolbox, Signal Processing Toolbox, Bioinformatics Toolbox, and Circular Statistics Toolbox by Philipp Berens). Any timepoint was classified as a spike peak if its signal amplitude was lower than -5 times of the signal standard deviation. A spike waveform was defined as a 4 -milliseconds-long slice of the signal centered at a spike peak. 尖峰检测与主成分分析聚类。所有神经记录数据的分析均使用 MATLAB R2022b 及四个开源工具箱完成（统计与机器学习工具箱、信号处理工具箱、生物信息学工具箱以及 Philipp Berens 开发的环形统计工具箱）。当信号幅度低于信号标准偏差的-5 倍时，该时间点被判定为尖峰峰值。尖峰波形定义为以尖峰峰值中心、时长 4 毫秒的信号片段。
In order to sort the detected spike waveforms that originated from different neurons, we applied dimension reduction and clustering algorithms. We first used principal component analysis (MATLAB function pca) to reduce each high-dimensional spike waveform into vectors residing in a two-dimensional feature space. These vectors were then clustered into pre-specified number of groups, using the k -means algorithm (MATLAB function kmeans). We set the number of groups as 2 or 3 . 为区分来自不同神经元的检测到的尖峰波形，我们采用了降维与聚类算法。首先使用主成分分析（MATLAB 函数 pca）将高维尖峰波形降维至二维特征空间的向量，随后通过 k 均值算法（MATLAB 函数 kmeans）将这些向量聚类为预设数量的组别。我们设定分组数量为 2 或 3 组。
Burst detection. We defined bursts using five pre-defined threshold parameters-maximum interspike interval at start of burst, maximum interspike interval in burst, minimum burst duration, minimum interburst interval, and minimum number of spikes in burst. Any series of spikes that satisfies all of these thresholds was considered as a burst. We set the threshold as 300,301,10,50,200300,301,10,50,200, respectively. 爆发检测。我们采用五个预定义的阈值参数来定义爆发——爆发起始的最大峰电位间隔、爆发期间的最大峰电位间隔、最小爆发持续时间、最小爆发间期以及爆发中的最小峰电位数量。任何满足所有这些阈值条件的峰电位序列均被视为一次爆发。我们将阈值分别设为 300,301,10,50,200300,301,10,50,200 。
2D and 3D neural network map. We quantified the synchronization level between two electrodes using the SPIKE-distance between the spike trains ^(35){ }^{35}. For the calculation of SPIKE-distance, we used Python package Pyspike (version 0.7.0) on a Python environment (version 3.9.2 )^(48))^{48}. The SPIKE-distance was normalized to average of 1,000 randomly generated baseline values; each baseline value is the SPIKEdistance between two uniformly randomly generated spike trains. Then, synchronization score was defined as SPIKE-distance values subtracted from 1 . This score ranged from 0 to 1 , where a higher score indicated a high level of synchrony. 二维与三维神经网络图谱。我们通过计算两个电极间峰电位序列的 SPIKE 距离来量化其同步水平 ^(35){ }^{35} 。SPIKE 距离的计算采用 Python 环境（版本 3.9.2）下的 Pyspike 软件包（版本 0.7.0） )^(48))^{48} 。该距离通过 1,000 次随机生成的基线值进行归一化处理，每个基线值为两条均匀随机峰电位序列间的 SPIKE 距离。同步得分定义为 1 减去 SPIKE 距离值，得分范围 0 至 1，数值越高表明同步程度越高。
Based on this synchronization score, community structure on the electrodes was examined by graph-theoretic Louvain algorithm which is based on modularity optimization ^(36,49){ }^{36,49}. In the analysis, we modelled the electrodes as nodes (e.g., circles in the network map), whose 3D positions represented the position of the actual electrode. We set the 基于此同步评分，通过基于模块度优化的图论 Louvain 算法检测了电极上的群落结构 ^(36,49){ }^{36,49} 。分析中，我们将电极建模为节点（如网络图中的圆圈），其三维位置代表实际电极的位置。我们设定
synchronized scores between every pair of active electrodes as edge (e.g., lines in the network map) weights between the corresponding nodes. Edges with synchronized scores less than 0.5 were filtered out. The result of the Louvain algorithm was visualized through color and size, using the Python package plotly (version 5.9.0). Nodes with the same color (i.e., electrodes) represented electrodes belonging to the same neural community. Synchronization scores were color-mapped, ranging from 0 (yellow) to 1 (red). In addition, the node size indicated the number of connected electrodes. 每对活跃电极间的同步评分作为对应节点间边的权重（如网络图中的连线）。同步评分低于 0.5 的边被过滤。Louvain 算法的结果通过颜色和大小可视化，使用 Python 包 plotly（版本 5.9.0）。相同颜色的节点（即电极）表示属于同一神经群落的电极。同步评分采用颜色映射，范围从 0（黄色）到 1（红色）。此外，节点大小表示连接电极的数量。
Statistics and reproducibility 统计与可重复性
For Fig. 2b-e and Fig. 3a, b, statistical analyzes were conducted using GraphPad Prism 10 (GraphPad, La Jolla, CA, USA). Statistical significance was determined using unpaired, two-sided Student’s tt-tests and one-way analysis of variance (ANOVA) with Tukey’s multiple comparisons tests based on the test requirements. For the other figures, statistical analyzes were conducted using Excel (Microsoft, Washington, USA), and statistical significance was determined using unpaired, one-tailed tt tests. All details regarding experimental replications, biological replicates ( n ), and statistical tests are included in the corresponding figure legends. 对于图 2b-e 及图 3a、b，统计分析采用 GraphPad Prism 10 软件（GraphPad 公司，美国拉霍亚）进行。根据检验需求，统计显著性通过双样本双尾学生 tt t 检验及单因素方差分析（ANOVA）结合 Tukey 多重比较检验确定。其余图表的数据分析使用 Excel（微软公司，美国华盛顿）完成，统计显著性采用单尾 tt 检验判定。所有关于实验重复次数、生物学重复样本数（n）及统计检验的详细信息均标注于对应图注中。
Ethics 伦理声明
This study does not involve experiment involving animals, human participants, or clinical samples. 本研究未涉及动物实验、人类受试者或临床样本。
Reporting summary 报告摘要
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. 有关研究设计的更多信息可在本文链接的《自然》作品集报告摘要中获取。
Data availability 数据可用性
All data supporting the findings of this study are available within the article and its supplementary files. Any additional requests for information can be directed to, and will be fulfilled by, the corresponding authors. The data used in this study are available in the Figshare (https://doi.org/10.6084/m9.figshare.24764508) ^(50){ }^{50}. Source data are provided with this paper for reproducing all Figures in the manuscript and Supplementary Information. 本研究所有支持发现的数据均可在文章及其补充文件中找到。如需额外信息，请联系通讯作者，我们将予以提供。本研究使用的数据已存放于 Figshare 平台（https://doi.org/10.6084/m9.figshare.24764508） ^(50){ }^{50} 。随文附带的源数据可用于复现稿件及补充信息中的所有图表。
Code availability 代码可用性
The custom code used for this study is available at Code Ocean (https://doi.org/10.24433/CO.7867348.v1) ^(51){ }^{51}. 本研究所用自定义代码可在 Code Ocean 平台获取（https://doi.org/10.24433/CO.7867348.v1） ^(51){ }^{51} 。
References 参考文献
Lancaster, M. A. et al. Cerebral organoids model human brain development and microcephaly. Nature 501, 373-379 (2013). Lancaster, M. A. 等。脑类器官模拟人类大脑发育与小头畸形。《自然》501 卷，373-379 页（2013 年）。
Di Lullo, E. & Kriegstein, A. R. The use of brain organoids to investigate neural development and disease. Nat. Rev. Neurosci. 18, 573-584 (2017). Di Lullo, E. 与 Kriegstein, A. R. 利用脑类器官研究神经发育与疾病。《自然·神经科学评论》18 卷，573-584 页（2017 年）。
Cho, A.-N. et al. Microfluidic device with brain extracellular matrix promotes structural and functional maturation of human brain organoids. Nat. Commun. 12, 4730 (2021). Cho, A.-N. 等。含有脑细胞外基质的微流控装置促进人脑类器官的结构与功能成熟。自然-通讯 12, 4730 (2021)。
Jeong, E., Choi, S. & Cho, S.-W. Recent advances in brain organoid technology for human brain research. ACS Appl. Mater. Interfaces 15, 200-219 (2023). Jeong, E., Choi, S. 与 Cho, S.-W. 脑类器官技术在人脑研究中的最新进展。ACS 应用材料与界面 15, 200-219 (2023)。
Luo, C. et al. Cerebral organoids recapitulate epigenomic signatures of the human fetal brain. Cell Rep. 17, 3369-3384 (2016). Luo, C. 等。大脑类器官重现人类胎儿大脑的表观基因组特征。细胞报告 17, 3369-3384 (2016)。
Gordon, A. et al. Long-term maturation of human cortical organoids matches key early postnatal transitions. Nat. Neurosci. 24, 331-342 (2021). Gordon, A. 等。人皮层类器官的长期成熟过程匹配出生后早期关键转变。自然-神经科学 24, 331-342 (2021)。
Qian, X. et al. Brain-region-specific organoids using minibioreactors for modeling zikv exposure. Cell 165, 1238-1254 (2016). 钱，X. 等。使用微型生物反应器构建脑区特异性类器官以模拟寨卡病毒暴露。《细胞》165 卷，1238-1254 页（2016 年）。
Bagley, J. A., Reumann, D., Bian, S., Lévi-Strauss, J. & Knoblich, J. A. Fused cerebral organoids model interactions between brain regions. Nat. Methods 14, 743-751 (2017). 巴格利，J. A. 等。融合脑类器官模拟脑区间相互作用。《自然·方法》14 卷，743-751 页（2017 年）。
Xiang, Y. et al. hESC-derived thalamic organoids form reciprocal projections when fused with cortical organoids. Cell Stem Cell 24, 487-497 (2019). 向，Y. 等。人胚胎干细胞来源的丘脑类器官与皮层类器官融合后形成相互投射。《细胞·干细胞》24 卷，487-497 页（2019 年）。
Faustino Martins, J.-M. et al. Self-organizing 3D human trunk neuromuscular organoids. Cell Stem Cell 26, 172-186 (2020). 福斯蒂诺·马丁斯，J.-M. 等。自组织三维人体躯干神经肌肉类器官。《细胞·干细胞》26 卷，172-186 页（2020 年）。
Park, J. et al. A 3D human triculture system modeling neurodegeneration and neuroinflammation in Alzheimer’s disease. Nat. Neurosci. 21, 941-951 (2018). Park, J. 等。一种模拟阿尔茨海默病神经退行性变与神经炎症的 3D 人类三培养系统。《自然·神经科学》21 卷, 941-951 页 (2018 年)。
Bershteyn, M. et al. Human iPSC-derived cerebral organoids model cellular features of lissencephaly and reveal prolonged mitosis of outer radial glia. Cell Stem Cell 20, 435-449 (2017). Bershteyn, M. 等。人 iPSC 来源的大脑类器官模拟无脑回症的细胞特征并揭示外放射状胶质细胞的有丝分裂延长。《细胞·干细胞》20 卷, 435-449 页 (2017 年)。
Kim, H. et al. Modeling G2O19S-LRRK2 sporadic parkinson’s disease in 3d midbrain organoids. Stem Cell Rep. 12, 518-531 (2019). Kim, H. 等。在 3D 中脑类器官中模拟 G2O19S-LRRK2 基因型散发性帕金森病。《干细胞报告》12 卷, 518-531 页 (2019 年)。
Ormel, P. R. et al. Microglia innately develop within cerebral organoids. Nat. Commun. 9, 4167 (2018). Ormel, P. R. 等。小胶质细胞在脑类器官中天然发育。《自然·通讯》9 卷, 4167 页 (2018 年)。
Opitz, T., De Lima, A. D. & Voigt, T. Spontaneous development of synchronous oscillatory activity during maturation of cortical networks in vitro. J. Neurophysiol. 88, 2196-2206 (2002). Opitz, T., De Lima, A. D. 与 Voigt, T. 体外培养皮质网络成熟过程中自发性同步振荡活动的形成。《神经生理学杂志》88 卷，2196-2206 页（2002 年）。
Bullmore, E. & Sporns, O. Complex brain networks: graph theoretical analysis of structural and functional systems. Nat. Rev. Neurosci. 10, 186-198 (2009). Bullmore, E. 与 Sporns, O. 复杂脑网络：结构与功能系统的图论分析。《自然·神经科学评论》10 卷，186-198 页（2009 年）。
Sharf, T. et al. Functional neuronal circuitry and oscillatory dynamics in human brain organoids. Nat. Commun. 13, 4403 (2022). Sharf, T. 等。人脑类器官中的功能性神经元回路与振荡动力学。《自然·通讯》13 卷，4403 页（2022 年）。
Trujillo, C. A. et al. Complex oscillatory waves emerging from cortical organoids model early human brain network development. Cell Stem Cell 25, 558-569 (2019). Trujillo, C. A. 等。皮质类器官中涌现的复杂振荡波模拟早期人脑网络发育。《细胞·干细胞》25 卷，558-569 页（2019 年）。
Park, Y. et al. Three-dimensional, multifunctional neural interfaces for cortical spheroids and engineered assembloids. Sci. Adv. 7, eabf9153 (2021). Park, Y. 等。用于皮质球体和工程组装体的三维多功能神经接口。《科学进展》7 卷，eabf9153（2021 年）。
Kim, J. et al. Intraocular pressure monitoring following islet transplantation to the anterior chamber of the eye. Nano Lett. 20, 1517-1525 (2019). Kim, J. 等。胰岛移植至眼前房后的眼压监测。《纳米快报》20 卷，1517-1525 页（2019 年）。
Jang, J. et al. Human-interactive, active-matrix displays for visualization of tactile pressures. Adv. Mater. Technol. 4, 1900082 (2019). Jang, J. 等。用于触觉压力可视化的人机交互有源矩阵显示器。《先进材料技术》4 卷，1900082（2019 年）。
Park, J. et al. Research on flexible display at Ulsan National Institute of Science and Technology. npj Flex. Electron. 1, 9 (2017). 蔚山国立科学技术研究院 Park, J.等人关于柔性显示器的研究。《npj 柔性电子》第 1 卷，第 9 页（2017 年）。
Quadrato, G. et al. Cell diversity and network dynamics in photosensitive human brain organoids. Nature 545, 48-53 (2017). Quadrato, G.等人关于光敏人脑类器官中细胞多样性与网络动态的研究。《自然》第 545 卷，48-53 页（2017 年）。
Shin, H. et al. 3D high-density microelectrode array with optical stimulation and drug delivery for investigating neural circuit dynamics. Nat. Commun. 12, 492 (2021). Shin, H.等人研发的具备光刺激与药物递送功能的 3D 高密度微电极阵列用于探究神经回路动态。《自然通讯》第 12 卷，第 492 页（2021 年）。
Polikov, V. S., Tresco, P. A. & Reichert, W. M. Response of brain tissue to chronically implanted neural electrodes. J. Neurosci. Methods 148, 1-18 (2005). Polikov, V. S., Tresco, P. A. 和 Reichert, W. M. 大脑组织对长期植入神经电极的反应。《神经科学方法杂志》148, 1-18 (2005).
Lai, J. D. et al. KCNJ2 inhibition mitigates mechanical injury in a human brain organoid model of traumatic brain injury. Cell Stem Cell 31, 519-536 (2024). Lai, J. D. 等。KCNJ2 抑制减轻了创伤性脑损伤人类脑器官模型中的机械损伤。《细胞干细胞》31, 519-536 (2024).
Le Floch, P. et al. Stretchable mesh nanoelectronics for 3d singlecell chronic electrophysiology from developing brain organoids. Adv. Mater. 34, 2106829 (2022). Le Floch, P. 等。可拉伸网状纳米电子器件用于发育中脑器官的三维单细胞长期电生理记录。《先进材料》34, 2106829 (2022).
Park, Y.-G. et al. High-resolution 3D printing for electronics. Adv. Sci. 9, 2104623 (2022). Park, Y.-G. 等。高分辨率 3D 打印技术用于电子器件制造。《先进科学》9, 2104623 (2022).
Park, Y.-G. et al. Liquid metal-based soft electronics for wearable healthcare. Adv. Healthc. Mater. 10, 2002280 (2021). Park, Y.-G. 等。基于液态金属的可穿戴医疗软电子器件。《先进医疗材料》10 卷，2002280 (2021)。
Kim, M. et al. Multimodal characterization of cardiac organoids using integrations of pressure-sensitive transistor arrays with threedimensional liquid metal electrodes. Nano Lett. 22, 7892-7901 (2022). Kim, M. 等。利用三维液态金属电极集成压敏晶体管阵列对心脏类器官进行多模态表征。《纳米快报》22 卷，7892-7901 (2022)。
Yoon, S.-J. et al. Reliability of human cortical organoid generation. Nat. Methods 16, 75-78 (2019). Yoon, S.-J. 等。人类皮质类器官生成的可靠性。《自然方法》16 卷，75-78 (2019)。
Zourray, C., Kurian, M. A., Barral, S. & Lignani, G. Electrophysiological properties of human cortical organoids: current state of the art and future directions. Front. Mol. Neurosci. 15, 839366 (2022). Zourray, C., Kurian, M. A., Barral, S. & Lignani, G. 人类皮质类器官的电生理特性：当前研究现状与未来方向。《分子神经科学前沿》15 卷，839366 (2022)。
Kreuz, T., Chicharro, D., Houghton, C., Andrzejak, R. G. & Mormann, F. Monitoring spike train synchrony. J. Neurophysiol. 109, 1457-1472 (2013). Kreuz, T., Chicharro, D., Houghton, C., Andrzejak, R. G. & Mormann, F. 监测尖峰序列同步性。《神经生理学杂志》109 卷, 1457-1472 页 (2013 年)。
Blondel, V. D., Guillaume, J.-L., Lambiotte, R. & Lefebvre, E. Fast unfolding of communities in large networks. J. Stat. Mech. Theory Exp. 2008, P10008 (2008). Blondel, V. D., Guillaume, J.-L., Lambiotte, R. & Lefebvre, E. 大型网络中社区的快速展开。《统计力学理论与实验》2008 年, P10008 (2008 年)。
Lam, D. et al. Tissue-specific extracellular matrix accelerates the formation of neural networks and communities in a neuron-glia coculture on a multi-electrode array. Sci. Rep. 9, 4159 (2019). Lam, D. 等。组织特异性细胞外基质加速多电极阵列上神经元-胶质细胞共培养中神经网络及群落形成。《科学报告》9 卷, 4159 页 (2019 年)。
Harb, K. et al. Area-specific development of distinct projection neuron subclasses is regulated by postnatal epigenetic modifications. Elife 5, e09531 (2016). Harb, K. 等。区域特异性投射神经元亚类的分化受出生后表观遗传修饰调控。《生命》5 卷, e09531 (2016 年)。
Livesey, M. R. et al. Maturation and electrophysiological properties of human pluripotent stem cell-derived oligodendrocytes. Stem Cells 34, 1040-1053 (2016). Livesey, M. R. 等。人类多能干细胞来源的少突胶质细胞的成熟及电生理特性。《干细胞》34 卷，1040-1053 页（2016 年）。
Liu, Y. et al. Ferromagnetic flexible electronics for brain-wide selective neural recording. Adv. Mater. 35, 2208251 (2023). Liu, Y. 等。用于全脑选择性神经记录的铁磁柔性电子器件。《先进材料》35 卷，2208251（2023 年）。
Jiang, S. et al. Spatially expandable fiber-based probes as a multifunctional deep brain interface. Nat. Commun. 11, 6115 (2020). Jiang, S. 等。作为多功能深脑接口的空间可扩展纤维基探针。《自然通讯》11 卷，6115（2020 年）。
Schoonover, C. E., Ohashi, S. N., Axel, R. & Fink, A. J. P. Representational drift in primary olfactory cortex. Nature 594, 541-546 (2021). Schoonover, C. E., Ohashi, S. N., Axel, R. & Fink, A. J. P. 初级嗅觉皮层中的表征漂移。《自然》594 卷，541-546 页（2021 年）。
Jang, J. et al. Disease-specific induced pluripotent stem cells: a platform for human disease modeling and drug discovery. Exp. Mol. Med. 44, 202-213 (2012). 张杰等。疾病特异性诱导多能干细胞：人类疾病建模与药物发现的平台。实验与分子医学 44, 202-213 (2012)。
Takahashi, K. et al. Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131, 861-872 (2007). 高桥和弥等。通过限定因子从成人成纤维细胞诱导多能干细胞。细胞 131, 861-872 (2007)。
Paşca, A. M. et al. Functional cortical neurons and astrocytes from human pluripotent stem cells in 3D culture. Nat. Methods 12, 671-678 (2015). 帕斯卡·A·M 等。从人类多能干细胞在 3D 培养中生成功能性皮质神经元和星形胶质细胞。自然方法 12, 671-678 (2015)。
Li, W., Germain, R. N. & Gerner, M. Y. Multiplex, quantitative cellular analysis in large tissue volumes with clearing-enhanced 3D microscopy (Ce3D). PNAS 114, E7321-E7330 (2017). 李伟等。通过透明增强 3D 显微镜（Ce3D）对大体积组织进行多重定量细胞分析。美国国家科学院院刊 114, E7321-E7330 (2017)。
Lee, J. et al. Generation and characterization of hair-bearing skin organoids from human pluripotent stem cells. Nat. Protoc. 17, 1266-1305 (2022). 李, J. 等。利用人类多能干细胞生成并鉴定带毛发的皮肤类器官。《自然-实验手册》17 卷, 1266-1305 页 (2022 年)。
Mulansky, M. & Kreuz, T. PySpike-A Python library for analyzing spike train synchrony. SoftwareX 5, 183-189 (2016). 穆兰斯基, M. 与 克鲁兹, T. PySpike——用于分析脉冲序列同步性的 Python 库。《SoftwareX》5 卷, 183-189 页 (2016 年)。
Newman, M. E. J. & Girvan, M. Finding and evaluating community structure in networks. Phys. Rev. E 69, 026113 (2004). 纽曼, M. E. J. 与 吉尔万, M. 网络中社区结构的发现与评估。《物理评论 E》69 卷, 026113 (2004 年)。
Kim, E. et al. Source data of “Magnetically reshapable 3D multielectrode arrays of liquid metlas for electrophysiological analysis of brain organoids”. Figshare. Online resource. https://doi.org/10. 6084/m9.figshare. 24764508 (2024). 金, E. 等。“液态金属可磁控重塑 3D 多电极阵列用于脑类器官电生理分析”的源数据。Figshare。在线资源。https://doi.org/10.6084/m9.figshare.24764508 (2024 年)。
Kim, E. et al. Code for “Magnetically reshapable 3D multi-electrode arrays of liquid metlas for electrophysiological analysis of brain organoids”. Code ocean. Online resource. https://doi.org/10. 24433/CO.7867348.v1 (2024). Kim, E. 等。《用于脑类器官电生理分析的液态金属磁重塑 3D 多电极阵列》代码。Code ocean。在线资源。https://doi.org/10.24433/CO.7867348.v1（2024 年）。
Acknowledgements 致谢
This work was supported by the Ministry of Science & ICT (MSIT) through the National Research Foundation, STEAM Research Business - Korea Global Cooperative Convergence Research Program (RS-202400460364) (J.-U. P) and ERC Program (RS-2024-00406240) (J.-U. P). 本研究由韩国科学技术信息通信部（MSIT）通过国家研究基金会支持，包括 STEAM 研究事业-韩国全球合作融合研究计划（RS-202400460364）（J.-U. P）和 ERC 计划（RS-2024-00406240）（J.-U. P）。
Author contributions 作者贡献
E.K. and E.J. carried out the experiment, analyzed the data, and wrote the manuscript. E.K., Y.-M.H., I.J., Y.W.K., Y.-G.P., and J.-Y.K. were involved in device fabrications and electrophysiological analysis. E.J., J.K., J.L., and S.C. were involved in organoid cultivation and assessment of biocompatibility. J.-U.P., S.-W.C., and J.-H.L., an oversaw all of the research phases and revised the manuscript. All authors discussed and commented on the manuscript. E.K.和 E.J.负责实验操作、数据分析及论文撰写。E.K.、Y.-M.H.、I.J.、Y.W.K.、Y.-G.P.和 J.-Y.K.参与了器件制备与电生理分析。E.J.、J.K.、J.L.和 S.C.负责类器官培养及生物相容性评估。J.-U.P.、S.-W.C.和 J.-H.L.监督研究全过程并修改论文。所有作者均参与讨论并对论文提出意见。
Competing interests 利益冲突
S.-W.C. is a chief technology officer (CTO) of Cellartgen, Inc., Republic of Korea. The remaining authors declare no competing interests. S.-W.C.担任韩国 Cellartgen 公司首席技术官(CTO)，其余作者声明无利益冲突。
Additional information 补充信息
Supplementary information The online version contains supplementary material available at 补充信息 在线版本包含补充材料，可查阅于 https://doi.org/10.1038/s41467-024-55752-3.
Correspondence and requests for materials should be addressed to Jae-Hyun Lee, Seung-Woo Cho or Jang-Ung Park. 通信及材料请求请寄送至 Jae-Hyun Lee、Seung-Woo Cho 或 Jang-Ung Park。
Peer review information Nature Communications thanks David Gracias, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. 同行评审信息 《自然-通讯》感谢 David Gracias 及其他匿名审稿人对本工作的同行评审贡献。同行评审文件可供查阅。
Reprints and permissions information is available at http://www.nature.com/reprints 重印及许可信息详见 http://www.nature.com/reprints
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 出版商注：施普林格·自然对出版地图和机构附属关系中的管辖权主张保持中立。
^(1){ }^{1} Department of Materials Science and Engineering, Yonsei University, Seoul 03722, Republic of Korea. ^(2){ }^{2} Center for Nanomedicine, Institute for Basic Science (IBS), Yonsei University, Seoul 03722, Republic of Korea. ^(3){ }^{3} Department of Biotechnology, Yonsei University, Seoul 03722, Republic of Korea. ^(4){ }^{4} Graduate Program of Nano Biomedical Engineering (NanoBME), Advanced Science Institute, Yonsei University, Yonsei, Republic of Korea. ^(5){ }^{5} Department of Neurosurgery, Yonsei University College of Medicine, Yonsei, Republic of Korea. ^(6){ }^{6} Yonsei-KIST Convergence Research Institute, Seoul 03722, Republic of Korea. ^(7){ }^{7} These authors contributed equally: Enji Kim, Eunseon Jeong, Yeon-Mi Hong. ee-mail: jhyun_lee@yonsei.ac.kr; seungwoocho@yonsei.ac.kr; jang-ung@yonsei.ac.kr ^(1){ }^{1} 延世大学材料科学与工程系，韩国首尔 03722。 ^(2){ }^{2} 延世大学基础科学研究院纳米医学中心，韩国首尔 03722。 ^(3){ }^{3} 延世大学生物技术系，韩国首尔 03722。 ^(4){ }^{4} 延世大学先进科学研究所纳米生物医学工程研究生项目，韩国延世。 ^(5){ }^{5} 延世大学医学院神经外科，韩国延世。 ^(6){ }^{6} 延世-KIST 融合研究院，韩国首尔 03722。 ^(7){ }^{7} 这些作者贡献均等：Enji Kim、Eunseon Jeong、Yeon-Mi Hong。电子邮箱：jhyun_lee@yonsei.ac.kr；seungwoocho@yonsei.ac.kr；jang-ung@yonsei.ac.kr
