A system hierarchy for brain-inspired computing.

Neuromorphic computing draws inspiration from the brain to provide computing technology and architecture with the potential to drive the next wave of computer engineering1-13. Such brain-inspired computing also provides a promising platform for the development of artificial general intelligence14,15. However, unlike conventional computing systems, which have a well established computer hierarchy built around the concept of Turing completeness and the von Neumann architecture16-18, there is currently no generalized system hierarchy or understanding of completeness for brain-inspired computing. This affects the compatibility between software and hardware, impairing the programming flexibility and development productivity of brain-inspired computing. Here we propose 'neuromorphic completeness', which relaxes the requirement for hardware completeness, and a corresponding system hierarchy, which consists of a Turing-complete software-abstraction model and a versatile abstract neuromorphic architecture. Using this hierarchy, various programs can be described as uniform representations and transformed into the equivalent executable on any neuromorphic complete hardware-that is, it ensures programming-language portability, hardware completeness and compilation feasibility. We implement toolchain software to support the execution of different types of program on various typical hardware platforms, demonstrating the advantage of our system hierarchy, including a new system-design dimension introduced by the neuromorphic completeness. We expect that our study will enable efficient and compatible progress in all aspects of brain-inspired computing systems, facilitating the development of various applications, including artificial general intelligence.

Towards artificial general intelligence with hybrid Tianjic chip architecture.

There are two general approaches to developing artificial general intelligence (AGI)1: computer-science-oriented and neuroscience-oriented. Because of the fundamental differences in their formulations and coding schemes, these two approaches rely on distinct and incompatible platforms2-8, retarding the development of AGI. A general platform that could support the prevailing computer-science-based artificial neural networks as well as neuroscience-inspired models and algorithms is highly desirable. Here we present the Tianjic chip, which integrates the two approaches to provide a hybrid, synergistic platform. The Tianjic chip adopts a many-core architecture, reconfigurable building blocks and a streamlined dataflow with hybrid coding schemes, and can not only accommodate computer-science-based machine-learning algorithms, but also easily implement brain-inspired circuits and several coding schemes. Using just one chip, we demonstrate the simultaneous processing of versatile algorithms and models in an unmanned bicycle system, realizing real-time object detection, tracking, voice control, obstacle avoidance and balance control. Our study is expected to stimulate AGI development by paving the way to more generalized hardware platforms.

Non B Cell-Derived Immunoglobulin, A Brighter Horizon for the Future.

The canonical theory of immunology stating that "Immunoglobulin (Ig) is produced by B lymphocytes and exerts antibody activity" has been established since the 1970s. However, the discovery of non B cell-derived Igs (non B-Igs), which can exert multiple biological activities in addition to their antibody activities, necessitates a reevaluation of the classic concept of Ig. This has been documented with a number of characteristics related to their structure, modification, genetic regulation as well as the functions associated with clinical conditions, particularly multiple cancers. The discovery of non B-Ig provides us with a new perspective to better understand not only basic immunology, but also various Ig-related clinical manifestations including autoimmune diseases, chronic inflammation, and anaphylaxis. Notably, non B-Ig can directly promote the occurrence of malignant tumours.

Regulating the Sequence Structure of Conjugated Block Copolymers Enables Large-Area Single-Component Organic Solar Cells with High Efficiency and Stability.

Single-component organic solar cells (SCOSCs) based on conjugated block copolymers (CBCs) by covalently bonding a polymer donor and polymer acceptor become more and more appealing due to the formation of a favorable and stable morphology. Unfortunately, a deep understanding of the effect of the assembly behavior caused by the sequence structure of CBCs on the device performance is still missing. Herein, from the aspect of manipulating the sequence length and distribution regularity of CBCs, we synthesized a series of new CBCs, namely D18(20)-b-PYIT, D18(40)-b-PYIT and D18(60)-b-PYIT by two-pot polymerization, and D18(40)-b-PYIT(r) by traditional one-pot method. It is observed that precise manipulation of sequence length and distribution regularity of the polymer blocks fine-tunes the self-assembly of the CBCs, optimizes film morphology, improves optoelectronic properties, and reduces energy loss, leading to simultaneously improved efficiency and stability. Among these CBCs, the D18(40)-b-PYIT-based device achieves a high efficiency of 13.4 % with enhanced stability, which is an outstanding performance among SCOSCs. Importantly, the regular sequence distribution and suitable sequence length of the CBCs enable a facile film-forming process of the printed device. For the first time, the blade-coated large-area rigid/flexible SCOSCs are fabricated, delivering an impressive efficiency of 11.62 %/10.73 %, much higher than their corresponding binary devices.

Oligomer-Assisted Photoactive Layers Enable >18 % Efficiency of Organic Solar Cells.

Although ternary organic solar cells (OSCs) have unique advantages in improving device performance, the morphology assembly in the ternary-phase would be more uncertain or complex than that in the binary-phase. Here, we propose a new concept of oligomer-assisted photoactive layers for high-performance OSCs. The formed alloy-like phase of the oligomer : host polymer blend enabled the oligomer-assisted OSCs to fuse the advantages of both binary and ternary devices, exhibiting substantially enhanced performance and stability compared to the control devices. With the addition of oligomers, outstanding efficiencies of 17.33 % for a PM6 : Y6 device, 18.32 % for a PM6 : BTP-eC9 device, and 17.13 % for a PM6/Y6 pseudo-bilayer device were achieved, all of which are one of the highest values in their corresponding fields. The improved performance originated from the downshift energy levels, enhanced light absorption, optimized blend morphology, favorable charge dynamics, and reduced non-radiative energy loss.

Binding of the monoclonal antibody RP215 to immunoglobulin G in metastatic lung adenocarcinomas is correlated with poor prognosis.

AIMS: Cancer cell-derived immunoglobulin (Ig)G (cancer-IgG) has been found to be involved in the pathogenesis and progression of many cancers, including lung cancer. The aim of the present study was to investigate the relationship between cancer-IgG expression in lung adenocarcinoma (ADC) and clinicopathological characteristics and clinical outcome. METHODS AND RESULTS: Immunohistochemical analysis was performed using an RP215 monoclonal antibody to determine cancer-IgG expression in 140 lung ADC patients. Cell migration and invasion were analysed in A549 cell line after short interfering RNA (siRNA) knockdown of IgG and cell sorting by flow cytometry. Our results show that RP215 immunostaining score is correlated significantly with local invasion (P < 0.05) and tumour differentiation (P < 0.05) in ADC. Moreover, RP215 staining was significantly higher in metastatic tumours than in primary tumours (P < 0.0001). The knockdown of IgG resulted in a reduction of cell migration and invasion. In contrast, RP215-positive cells displayed greater migration and invasion ability than RP215-negative cells. Additionally, a higher RP215 immunostaining score was associated significantly with poor prognosis. CONCLUSIONS: RP215 staining is correlated strongly with differentiation, local invasion, metastasis and clinical outcome of patients with lung ADC. Our results suggest that RP215 can serve as a biomarker for prognosis of lung ADC.

The immunoglobulin heavy chain VH6-1 promoter regulates Ig transcription in non-B cells.

BACKGROUND: Non-B cell immunoglobulins (Igs) are widely expressed in epithelial cancer cells. The past 20 years of research have demonstrated that non-B cell Igs are associated with cancer cell proliferation, the cellular cytoskeleton and cancer stem cells. In this study we explored the transcriptional mechanism of IgM production in non-B cells. METHODS: The promoter region of a V-segment of the heavy mu chain gene (VH6-1) was cloned from a colon cancer cell line HT-29. Next, the promoter activities in non-B cells and B-cells were detected using the dual-luciferase reporter assay. Then the transcription factor binding to the promoter regions was evaluated by electrophoretic mobility shift assays (EMSAs) and gel supershift experiments. RESULTS: Our data showed that the sequence 1200 bp upstream of VH6-1 exhibited promoter activity in both B and non-B cells. No new regulatory elements were identified within the region 1200 bp to 300 bp upstream of VH6-1. In addition, Oct-1 was found to bind to the octamer element of the Ig gene promoter in cancer cells, in contrast to B cells, which utilize the transcriptional factor Oct-2. CONCLUSION: The regulatory mechanisms among different cell types controlling the production of IgM heavy chains are worth discussing.

Influence of hydrocolloids and natural emulsifier in the physical stability of UHT oat beverage.

This study aimed to improve the physical stability of ultra-high temperature (UHT) oat beverage by adding hydrophilic colloids (guar gum [GG] and xanthan gum [XG]) and a natural emulsifier (soluble soybean polysaccharide [SSPS]). The stability of the oat beverage was characterized by particle size, zeta potential, rheological properties, Fourier-transform infrared (FTIR) spectroscopy, backscattered light intensity (ΔBS), and microstructure. The results indicated that XG reduced the average particle size and size distribution of the beverage, indicating that XG could prevent particle aggregation. GG increases the apparent viscosity of the oat beverage without affecting the zeta potential. When SSPS was added to the oat beverage, it increased the absolute value of the zeta potential and the infrared absorption peak intensity, while the average particle size and backscattered light intensity (ΔBS) decreased, resulting in a more uniform microstructure. The zeta potential reached a maximum value of 32.12 when GG, XG, and SSPS were combined, indicating that the physical stability of the oat beverage was effectively improved when all three were present simultaneously. This study may provide some suggestions for the industrial production of low-viscosity cereal beverages with good stability.

A Self-Propagating c-Met-SOX2 Axis Drives Cancer-Derived IgG Signaling That Promotes Lung Cancer Cell Stemness.

SIGNIFICANCE: Sialylated cancer IgG activates c-Met-SOX2 signaling to promote stemness properties in cancer cells and can be targeted to suppress tumor growth.

Immunotherapy using slow-cycling tumor cells prolonged overall survival of tumor-bearing mice.

BACKGROUND: Despite considerable progress in the development of anticancer therapies, there is still a high mortality rate caused by cancer relapse and metastasis. Dormant or slow-cycling residual tumor cells are thought to be a source of tumor relapse and metastasis, and are therefore an obstacle to therapy. In this study, we assessed the drug resistance of tumor cells in mice, and investigated whether vaccination could promote survival. METHODS: The mouse colon carcinoma cell line CT-26 was treated with 5-fluorouracil to assess its sensitivity to drug treatment. Mice with colon tumors were immunized with inactivated slow-cycling CT-26 cells to estimate the efficacy of this vaccine. RESULTS: We identified a small population of slow-cycling tumor cells in the mouse colon carcinoma CT-26 cell line, which was resistant to conventional chemotherapy. To inhibit tumor recurrence and metastasis more effectively, treatments that selectively target the slow-cycling tumor cells should be developed to complement conventional therapies. We found that drug-treated, slow-cycling tumor cells induced a more intense immune response in vitro. Moreover, vaccination with inactivated slow-cycling tumor cells caused a reduction in tumor volume and prolonged the overall survival of tumor-bearing mice. CONCLUSIONS: These findings suggest that targeting of slow-cycling tumor cells application using immunotherapy is a possible treatment to complement traditional antitumor therapy.

Doxorubicin-enriched, ALDH(br) mouse breast cancer stem cells are treatable to oncolytic herpes simplex virus type 1.

BACKGROUND: The primary objective of this study was to test whether oncolytic herpes simplex virus type 1 (HSV1) could eradicate chemoresistant cancer stem cells (CSCs). METHODS: The fluorescent aldefluor reagent-based technique was used to identify and isolate ALDH(br) cells as CSCs from the 4T1 murine breast cancer cell line. The presence of ALDH(br) 4T1 cells was also examined in 4T1 breast cancer transplanted in immune-competent syngeneic mice. RESULTS: Compared with ALDH(lo) cells, ALDH(br) cells had a markedly higher ability to form tumor spheres in vitro and a higher tumorigenic potential in vivo. ALDH(br) cells also exhibited increased doxorubicin resistance in vitro, which correlated with a selective increase in the percentage of ALDH(br) cells after doxorubicin treatment and an increased expression of P-glycoprotein (P-gp), a known chemoresistance factor. In contrast, oncolytic HSV1 was able to kill ALDH(br) cells in vitro and even more markedly in vivo. Furthermore, in in vivo studies, systemic administration of doxorubicin followed by intratumoral injection of oncolytic HSV1 resulted in much more significant suppression of tumor growth with increased median survival period compared with each treatment given alone (p<0.05). Though more CD8(+) T lymphocytes were induced by oncolytic HSV1, no significant specific T cell response against CSCs was detected in vivo. CONCLUSIONS: These results suggested that the use of oncolytic HSV1 following doxorubicin treatment may help eradicate residual chemoresistant CSCs in vivo.

[Generation of a herpes simplex virus-permissive mouse melanoma cell line B16RHSV].

OBJECTIVE: To generate an oncolytic herpes simplex virus (oHSV) permissive mouse melanoma cell line B16RHSV, preserving the tumorigenic ability in syngeneic mice. METHODS: The herpes simplex virus entry mediator (HVEM) gene was amplified by PCR from human melanoma cell line A375, and cloned into pGEM-T Easy vector for sequencing. The HVEM gene was then cloned into pcDNA3 vector to generate pcDNA3-HVEM for transfection of mouse melanoma cell line B16-F10 cells. After that, the putative transfected cells were selected in full growth medium containing G418. The HVEM-expressing cells were isolated by immunomagnetic bead separation. The mouse melanoma cell line expressing oHSV receptor-HVEM, designated as B16RHSV, was generated. The permissibility of B16RHSV cells to oHSV infection was examined with green fluorescence protein (GFP)-expressing oHSV (oHSVGFP). To investigate the tumorigenic ability of both cells in vivo, 2×10(5) cells in 100 µl were subcutaneously inoculated into the right flanks of C57/BL mice. RESULTS: In vitro, the B16RHSV mouse melanoma cells were shown by fluorescence microscopy capable of being infected by oHSVGFP. In vivo, the B16RHSV cells, like their wild type counterpart, grew to form melanoma in syngeneic mice. CONCLUSION: A herpes simplex virus-permissive mouse melanoma cell line was established. Its tumorigenicity remained unchanged.

Next-Generation Sequencing Revealed a Distinct Immunoglobulin Repertoire with Specific Mutation Hotspots in Acute Myeloid Leukemia.

Immunoglobulin (Ig) is known as a hallmark of B-lymphocytes exerting antibody functions. However, our previous studies demonstrated that myeloblasts from acute myeloid leukemia (AML) patients could also express Ig with distinct roles. Here, we quantified Ig (IGHG and IGK) transcripts by real-time PCR and performed a comprehensive analysis of Ig repertoire (both heavy chains and light chains) in AML blasts. We found that Ig was frequently expressed by AML blasts. A higher level of AML-derived IGHG expression correlated with a significantly shorter disease-free survival. Next-generation sequencing revealed dysregulated transcripts of all five Ig classes (IGHA, IGHD, IGHE, IGHG, and IGHM) and two Ig types (IGK and IGL) in AML. VH-D-JH rearrangements in myeloblasts were biased with individual specificity rather than generally diverse as in B-cells. Compared to AML-derived IgH, AML-derived IGK was more conserved among different AML samples. The frequently shared Vκ-Jκ patterns were IGKV3-20*01/IGKJ1*01, IGKV2D-28*01/IGKJ1*01, and IGKV4-1*01/IGKJ1*01. Moreover, AML-derived IGK was different from classical IGK in B-cells for the high mutation rates and special mutation hotspots at serine codons. Findings of the distinct Ig repertoire in myeloblasts may facilitate the discovery of a new molecular marker for disease monitoring and target therapy.

Brain-inspired global-local learning incorporated with neuromorphic computing.

There are two principle approaches for learning in artificial intelligence: error-driven global learning and neuroscience-oriented local learning. Integrating them into one network may provide complementary learning capabilities for versatile learning scenarios. At the same time, neuromorphic computing holds great promise, but still needs plenty of useful algorithms and algorithm-hardware co-designs to fully exploit its advantages. Here, we present a neuromorphic global-local synergic learning model by introducing a brain-inspired meta-learning paradigm and a differentiable spiking model incorporating neuronal dynamics and synaptic plasticity. It can meta-learn local plasticity and receive top-down supervision information for multiscale learning. We demonstrate the advantages of this model in multiple different tasks, including few-shot learning, continual learning, and fault-tolerance learning in neuromorphic vision sensors. It achieves significantly higher performance than single-learning methods. We further implement the model in the Tianjic neuromorphic platform by exploiting algorithm-hardware co-designs and prove that the model can fully utilize neuromorphic many-core architecture to develop hybrid computation paradigm.

A widely expressed free immunoglobulin κ chain with a unique Vκ4-1/Jκ3 pattern promotes colon cancer invasion and metastasis by activating the integrin ß1/FAK pathway.

Historically, immunoglobulin (Ig) has been known as an antibody and is expressed only in B lineage cells; importantly, Ig light chains are conjugated to heavy chains to form intact Igs. However, in this study, we found a free Igκ light chain with a unique Vκ4-1/Jκ3 rearrangement (Vκ4-1/Jκ3-FLC) that was widely expressed in different non-B lineages and was overexpressed in cancer cells. Further study indicated that Vκ4-1/Jκ3-FLC was hydrophobic, formed obvious insoluble deposits in the extracellular matrix (ECM) and existed in free form. Functional analyses demonstrated that Vκ4-1/Jκ3-FLC promoted the proliferation, migration and metastasis of colon cancer cells in vitro and in vivo. Mechanistically, Vκ4-1/Jκ3-FLC bound to integrin ß1 and activated the FAK and Src pathways. More importantly, specific antibodies against the variable region of Vκ4-1/Jκ3-FLC significantly inhibited the growth of colon cancer tumors. Our findings suggested that Vκ4-1/Jκ3-FLC is a novel ECM protein and integrin ß1 ligand and that it is involved in cancer progression and is a potential therapeutic target in cancer, particularly colon cancer.

A framework for the general design and computation of hybrid neural networks.

There is a growing trend to design hybrid neural networks (HNNs) by combining spiking neural networks and artificial neural networks to leverage the strengths of both. Here, we propose a framework for general design and computation of HNNs by introducing hybrid units (HUs) as a linkage interface. The framework not only integrates key features of these computing paradigms but also decouples them to improve flexibility and efficiency. HUs are designable and learnable to promote transmission and modulation of hybrid information flows in HNNs. Through three cases, we demonstrate that the framework can facilitate hybrid model design. The hybrid sensing network implements multi-pathway sensing, achieving high tracking accuracy and energy efficiency. The hybrid modulation network implements hierarchical information abstraction, enabling meta-continual learning of multiple tasks. The hybrid reasoning network performs multimodal reasoning in an interpretable, robust and parallel manner. This study advances cross-paradigm modeling for a broad range of intelligent tasks.

AIE based GSH activatable photosensitizer for imaging-guided photodynamic therapy.

A novel ferrocene decorated vinyl pyridinium-substituted tetraphenylethylene (TPEPY-S-Fc) linked by a disulfide bond was designed as a GSH activatable photosensitizer by aggregation-induced emission for imaging-guided photodynamic therapy of cancer cells.

A hybrid and scalable brain-inspired robotic platform.

Recent years have witnessed tremendous progress of intelligent robots brought about by mimicking human intelligence. However, current robots are still far from being able to handle multiple tasks in a dynamic environment as efficiently as humans. To cope with complexity and variability, further progress toward scalability and adaptability are essential for intelligent robots. Here, we report a brain-inspired robotic platform implemented by an unmanned bicycle that exhibits scalability of network scale, quantity and diversity to handle the changing needs of different scenarios. The platform adopts rich coding schemes and a trainable and scalable neural state machine, enabling flexible cooperation of hybrid networks. In addition, an embedded system is developed using a cross-paradigm neuromorphic chip to facilitate the implementation of diverse neural networks in spike or non-spike form. The platform achieved various real-time tasks concurrently in different real-world scenarios, providing a new pathway to enhance robots' intelligence.

Cancer-derived sialylated IgG promotes tumor immune escape by binding to Siglecs on effector T cells.

To date, IgG in the tumor microenvironment (TME) has been considered a product of B cells and serves as an antitumor antibody. However, in this study, using a monoclonal antibody against cancer-derived IgG (Cancer-IgG), we found that cancer cells could secrete IgG into the TME. Furthermore, Cancer-IgG, which carries an abnormal sialic acid modification in the CH1 domain, directly inhibited effector T-cell proliferation and significantly promoted tumor growth by reducing CD4+ and CD8+ T-cell infiltration into tumor tissues. Mechanistic studies showed that the immunosuppressive effect of sialylated Cancer-IgG is dependent on its sialylation and binding to sialic acid-binding immunoglobulin-type lectins (Siglecs) on effector CD4+ and CD8+ T cells. Importantly, we show that several Siglecs are overexpressed on effector T cells from cancer patients, but not those from healthy donors. These findings suggest that sialylated Cancer-IgG may be a ligand for Siglecs, which may serve as potential checkpoint proteins and mediate tumor immune evasion.

Novel chemotactic-antigen DNA vaccine against cancer.

Dendritic cells play a pivotal role in immune induction. Dendritic cells perform antigen uptake, processing and presentation to T cells only when they are matured and in the functional state. In the development of a vaccine, it is of utmost importance to consider how to make dendritic cells' functions immunologically adequate. In this paper, we report the development of a series of antitumor DNA vaccines with similar structural framework, in which a gene encoding tumor-associated antigenic peptide is ligated upstream to the gene coding secondary lymphoid-tissue chemokine and downstream to the gene encoding the Fc portion of IgG (named chemotactic-antigen DNA vaccine [CADV]). CCR7(+) T, B, natural killer and dendritic cells can be attracted by secondary lymphoid-tissue chemokine, and Fc facilitates antigen uptake via Fc receptors expressed on dendritic cells. In a series of experiments in mice vaccinated by CADV with such tumor-associated antigenic specificities as HPV-16 E7, PSA-PSM-PAP, HER-2/neu, p53 and hTERT, CADV can attract immune cells to the vaccine inoculation site, remarkably inhibit tumor growth and extend survival time in tumor-bearing mice. The antitumor effect is more efficacious than that in mice treated with SLC-Ag or Ag-Fc hybrid gene. Tumor-associated antigenic-specific cytotoxic T lymphocytes can be induced by in vitro experiment in a human system. When combined with measures blocking the negative immune feedback circuits, the therapeutic effect of the vaccine can be further enhanced. Large-scale production of CADV is possible for clinical application.