资源环境科技发展态势分析平台

Skin organoids generated in vitro from human pluripotent stem cells form complex, multilayered skin tissue with hair follicles, sebaceous glands and neural circuitry, and integrate with endogenous skin when grafted onto immunocompromised mice.

The skin is a multilayered organ, equipped with appendages (that is, follicles and glands), that is critical for regulating body temperature and the retention of bodily fluids, guarding against external stresses and mediating the sensation of touch and pain(1,2). Reconstructing appendage-bearing skin in cultures and in bioengineered grafts is a biomedical challenge that has yet to be met(3-9). Here we report an organoid culture system that generates complex skin from human pluripotent stem cells. We use stepwise modulation of the transforming growth factor beta (TGF beta) and fibroblast growth factor (FGF) signalling pathways to co-induce cranial epithelial cells and neural crest cells within a spherical cell aggregate. During an incubation period of 4-5 months, we observe the emergence of a cyst-like skin organoid composed of stratified epidermis, fat-rich dermis and pigmented hair follicles that are equipped with sebaceous glands. A network of sensory neurons and Schwann cells form nerve-like bundles that target Merkel cells in organoid hair follicles, mimicking the neural circuitry associated with human touch. Single-cell RNA sequencing and direct comparison to fetal specimens suggest that the skin organoids are equivalent to the facial skin of human fetuses in the second trimester of development. Moreover, we show that skin organoids form planar hair-bearing skin when grafted onto nude mice. Together, our results demonstrate that nearly complete skin can self-assemble in vitro and be used to reconstitute skin in vivo. We anticipate that our skin organoids will provide a foundation for future studies of human skin development, disease modelling and reconstructive surgery.

Most cortical synapses are local and excitatory. Local recurrent circuits could implement amplification, allowing pattern completion and other computations(1-4). Cortical circuits contain subnetworks that consist of neurons with similar receptive fields and increased connectivity relative to the network average(5,6). Cortical neurons that encode different types of information are spatially intermingled and distributed over large brain volumes(5-7), and this complexity has hindered attempts to probe the function of these subnetworks by perturbing them individually(8). Here we use computational modelling, optical recordings and manipulations to probe the function of recurrent coupling in layer 2/3 of the mouse vibrissal somatosensory cortex during active tactile discrimination. A neural circuit model of layer 2/3 revealed that recurrent excitation enhances sensory signals by amplification, but only for subnetworks with increased connectivity. Model networks with high amplification were sensitive to damage: loss of a few members of the subnetwork degraded stimulus encoding. We tested this prediction by mapping neuronal selectivity(7) and photoablating(9,10) neurons with specific selectivity. Ablation of a small proportion of layer 2/3 neurons (10-20, less than 5% of the total) representing touch markedly reduced responses in the spared touch representation, but not in other representations. Ablations most strongly affected neurons with stimulus responses that were similar to those of the ablated population, which is also consistent with network models. Recurrence among cortical neurons with similar selectivity therefore drives input-specific amplification during behaviour.

Computational modelling, imaging and single-cell ablation in layer 2/3 of the mouse vibrissal somatosensory cortex reveals that recurrent activity in cortical neurons can drive input-specific amplification during behaviour.

Memristor-enabled neuromorphic computing systems provide a fast and energy-efficient approach to training neural networks(1-4). However, convolutional neural networks (CNNs)-one of the most important models for image recognition(5)-have not yet been fully hardware-implemented using memristor crossbars, which are cross-point arrays with a memristor device at each intersection. Moreover, achieving software-comparable results is highly challenging owing to the poor yield, large variation and other non-ideal characteristics of devices(6-9). Here we report the fabrication of high-yield, high-performance and uniform memristor crossbar arrays for the implementation of CNNs, which integrate eight 2,048-cell memristor arrays to improve parallel-computing efficiency. In addition, we propose an effective hybrid-training method to adapt to device imperfections and improve the overall system performance. We built a five-layer memristor-based CNN to perform MNIST10 image recognition, and achieved a high accuracy of more than 96 per cent. In addition to parallel convolutions using different kernels with shared inputs, replication of multiple identical kernels in memristor arrays was demonstrated for processing different inputs in parallel. The memristor-based CNN neuromorphic system has an energy efficiency more than two orders of magnitude greater than that of state-of-the-art graphics-processing units, and is shown to be scalable to larger networks, such as residual neural networks. Our results are expected to enable a viable memristor-based non-von Neumann hardware solution for deep neural networks and edge computing.