Understanding YTHDF2-mediated mRNA degradation by m6A-BERT-Deg.

N6-methyladenosine (m6A) is the most abundant mRNA modification within mammalian cells, holding pivotal significance in the regulation of mRNA stability, translation and splicing. Furthermore, it plays a critical role in the regulation of RNA degradation by primarily recruiting the YTHDF2 reader protein. However, the selective regulation of mRNA decay of the m6A-methylated mRNA through YTHDF2 binding is poorly understood. To improve our understanding, we developed m6A-BERT-Deg, a BERT model adapted for predicting YTHDF2-mediated degradation of m6A-methylated mRNAs. We meticulously assembled a high-quality training dataset by integrating multiple data sources for the HeLa cell line. To overcome the limitation of small training samples, we employed a pre-training-fine-tuning strategy by first performing a self-supervised pre-training of the model on 427 760 unlabeled m6A site sequences. The test results demonstrated the importance of this pre-training strategy in enabling m6A-BERT-Deg to outperform other benchmark models. We further conducted a comprehensive model interpretation and revealed a surprising finding that the presence of co-factors in proximity to m6A sites may disrupt YTHDF2-mediated mRNA degradation, subsequently enhancing mRNA stability. We also extended our analyses to the HEK293 cell line, shedding light on the context-dependent YTHDF2-mediated mRNA degradation.

Early-life infection depletes preleukemic cells in a mouse model of hyperdiploid B-cell acute lymphoblastic leukemia.

Epidemiological studies report opposing influences of infection on childhood B cell acute lymphoblastic leukemia (B-ALL). Although infections in the first year of life appear to exert the largest impact on leukemia risk, the effect of early pathogen exposure on the fetal preleukemia cells (PLC) that lead to B-ALL has yet to be reported. Using cytomegalovirus as a model early-life infection, we show that virus exposure within one week of birth induces profound depletion of transplanted B-ALL cells in two mouse models and of in situ-generated PLC in Eu-ret mice. The age-dependent depletion of PLC results from an elevated STAT4-mediated cytokine response in neonates, with high levels of IL-12p40-driven IFN-g production inducing PLC death. Similar PLC depletion can be achieved in adult mice by impairing viral clearance. These findings provide mechanistic support for an inhibitory effect of early-life infection on B-ALL progression and could inform development of therapeutic or preventative approaches.

Deep learning tackles single-cell analysis-a survey of deep learning for scRNA-seq analysis.

Since its selection as the method of the year in 2013, single-cell technologies have become mature enough to provide answers to complex research questions. With the growth of single-cell profiling technologies, there has also been a significant increase in data collected from single-cell profilings, resulting in computational challenges to process these massive and complicated datasets. To address these challenges, deep learning (DL) is positioned as a competitive alternative for single-cell analyses besides the traditional machine learning approaches. Here, we survey a total of 25 DL algorithms and their applicability for a specific step in the single cell RNA-seq processing pipeline. Specifically, we establish a unified mathematical representation of variational autoencoder, autoencoder, generative adversarial network and supervised DL models, compare the training strategies and loss functions for these models, and relate the loss functions of these models to specific objectives of the data processing step. Such a presentation will allow readers to choose suitable algorithms for their particular objective at each step in the pipeline. We envision that this survey will serve as an important information portal for learning the application of DL for scRNA-seq analysis and inspire innovative uses of DL to address a broader range of new challenges in emerging multi-omics and spatial single-cell sequencing.

Analysis of the Memristor-Based Crossbar Synapse for Neuromorphic Systems.

In this study, we analyzed the memristor device typically used as a synapse in neuromorphic architecture and confirmed that the synaptic memristor device can be adopted to perform the machine learning algorithm. The nonlinear characteristics of the memristor complicates its use as the neuromorphic hardware in an artificial neural network (ANN) with a back-propagation algorithm. Using a memristor device with a nonlinear characteristic, we demonstrated that pattern classification can be implemented in ANNs using the Guide training algorithm without back-propagation. Furthermore, the memristor characteristics required to achieve accurate learning results are analyzed.

IFN-γ directly inhibits murine B-cell precursor leukemia-initiating cell proliferation early in life.

The early-life immune environment has been implicated as a modulator of acute lymphoblastic leukemia (ALL) development in children, with infection being associated with significant changes in ALL risk. Furthermore, polymorphisms in several cytokine genes, including IL-10 and IFN-γ, are associated with leukemia development. However, the mechanisms and timing of these influences remain unknown. Here, we use the Eµ-ret transgenic mouse model of B-cell precursor ALL to assess the influence of IFN-γ on the early-life burden of leukemia-initiating cells. The absence of IFN-γ activity resulted in greater numbers of leukemia-initiating cells early in life and was associated with accelerated leukemia onset. The leukemia-initiating cells from IFN-γ-knockout mice had reduced suppressor of cytokine signaling (SOCS-1) expression, were significantly more sensitive to IFN-γ, and exhibited more rapid expansion in vivo than their wild-type counterparts. However, sensitivity to this inhibitory pathway was lost in fully transformed IFN-γ-knockout leukemia cells. These results demonstrate that the influence of IFN-γ on ALL progression may not be mediated by selection of nascent transformed cells but rather through a general SOCS-mediated reduction in B-cell precursor proliferation. Thus, while cytokine levels may influence leukemia at multiple points during disease progression, our study indicates a significant early influence of basal, infection-independent cytokine production on leukemogenesis.

Understanding YTHDF2-mediated mRNA Degradation By m6A-BERT-Deg.

N6-methyladenosine (m6A) is the most abundant mRNA modification within mammalian cells, holding pivotal significance in the regulation of mRNA stability, translation, and splicing. Furthermore, it plays a critical role in the regulation of RNA degradation by primarily recruiting the YTHDF2 reader protein. However, the selective regulation of mRNA decay of the m6A-methylated mRNA through YTHDF2 binding is poorly understood. To improve our understanding, we developed m6A-BERT-Deg, a BERT model adapted for predicting YTHDF2-mediated degradation of m6A-methylated mRNAs. We meticulously assembled a high-quality training dataset by integrating multiple data sources for the HeLa cell line. To overcome the limitation of small training samples, we employed a pre-training-fine-tuning strategy by first performing a self-supervised pre-training of the model on 427,760 unlabeled m6A site sequences. The test results demonstrated the importance of this pre-training strategy in enabling m6A-BERT-Deg to outperform other benchmark models. We further conducted a comprehensive model interpretation and revealed a surprising finding that the presence of co-factors in proximity to m6A sites may disrupt YTHDF2-mediated mRNA degradation, subsequently enhancing mRNA stability. We also extended our analyses to the HEK293 cell line, shedding light on the context-dependent YTHDF2-mediated mRNA degradation.

reguloGPT: Harnessing GPT for Knowledge Graph Construction of Molecular Regulatory Pathways.

Motivation: Molecular Regulatory Pathways (MRPs) are crucial for understanding biological functions. Knowledge Graphs (KGs) have become vital in organizing and analyzing MRPs, providing structured representations of complex biological interactions. Current tools for mining KGs from biomedical literature are inadequate in capturing complex, hierarchical relationships and contextual information about MRPs. Large Language Models (LLMs) like GPT-4 offer a promising solution, with advanced capabilities to decipher the intricate nuances of language. However, their potential for end-to-end KG construction, particularly for MRPs, remains largely unexplored. Results: We present reguloGPT, a novel GPT-4 based in-context learning prompt, designed for the end-to-end joint name entity recognition, N-ary relationship extraction, and context predictions from a sentence that describes regulatory interactions with MRPs. Our reguloGPT approach introduces a context-aware relational graph that effectively embodies the hierarchical structure of MRPs and resolves semantic inconsistencies by embedding context directly within relational edges. We created a benchmark dataset including 400 annotated PubMed titles on N6-methyladenosine (m6A) regulations. Rigorous evaluation of reguloGPT on the benchmark dataset demonstrated marked improvement over existing algorithms. We further developed a novel G-Eval scheme, leveraging GPT-4 for annotation-free performance evaluation and demonstrated its agreement with traditional annotation-based evaluations. Utilizing reguloGPT predictions on m6A-related titles, we constructed the m6A-KG and demonstrated its utility in elucidating m6A's regulatory mechanisms in cancer phenotypes across various cancers. These results underscore reguloGPT's transformative potential for extracting biological knowledge from the literature. Availability and implementation: The source code of reguloGPT, the m6A title and benchmark datasets, and m6A-KG are available at: https://github.com/Huang-AI4Medicine-Lab/reguloGPT.

The Eµ-Ret mouse is a novel model of hyperdiploid B-cell acute lymphoblastic leukemia.

The presence of supernumerary chromosomes is the only abnormality shared by all patients diagnosed with high-hyperdiploid B cell acute lymphoblastic leukemia (HD-ALL). Despite being the most frequently diagnosed pediatric leukemia, the lack of clonal molecular lesions and complete absence of appropriate experimental models have impeded the elucidation of HD-ALL leukemogenesis. Here, we report that for 23 leukemia samples isolated from moribund Eµ-Ret mice, all were characterized by non-random chromosomal gains, involving combinations of trisomy 9, 12, 14, 15, and 17. With a median gain of three chromosomes, leukemia emerged after a prolonged latency from a preleukemic B cell precursor cell population displaying more diverse aneuploidy. Transition from preleukemia to overt disease in Eµ-Ret mice is associated with acquisition of heterogeneous genomic abnormalities affecting the expression of genes implicated in pediatric B-ALL. The development of abnormal centrosomes in parallel with aneuploidy renders both preleukemic and leukemic cells sensitive to inhibitors of centrosome clustering, enabling targeted in vivo depletion of leukemia-propagating cells. This study reveals the Eµ-Ret mouse to be a novel tool for investigating HD-ALL leukemogenesis, including supervision and selection of preleukemic aneuploid clones by the immune system and identification of vulnerabilities that could be targeted to prevent relapse.

Age and ligand specificity influence the outcome of pathogen engagement on preleukemic and leukemic B-cell precursor populations.

Common infections have long been proposed to play a role in the development of pediatric B-cell acute lymphoblastic leukemia (B-ALL). However, epidemiologic studies report contradictory effects of infection exposure on subsequent B-ALL risk, and no specific pathogen has been definitively linked to the disease. A unifying mechanism to explain the divergent outcomes could inform disease prevention strategies. We previously reported that the pattern recognition receptor (PRR) ligand Poly(I:C) exerted effects on B-ALL cells that were distinct from those observed with other nucleic acid-based PRR ligands. Here, using multiple double-stranded RNA (dsRNA) moieties, we show that the overall outcome of exposure to Poly(I:C) reflects the balance of opposing responses induced by its ligation to endosomal and cytoplasmic receptors. This PRR response biology is shared between mouse and human B-ALL and can increase leukemia-initiating cell burden in vivo during the preleukemia phase of B-ALL, primarily through tumor necrosis factor α signaling. The age of the responding immune system further influences the impact of dsRNA exposure on B-ALL cells in both mouse and human settings. Overall, our study demonstrates that potentially proleukemic and antileukemic effects can each be generated by the stimulation of pathogen recognition pathways and indicates a mechanistic explanation for the contrasting epidemiologic associations reported for infection exposure and B-ALL.

Inflammatory Immune Responses Trigger Rejection of Allogeneic Fibroblasts Transplanted into Mouse Skin.

Fibroblasts, or their homolog stromal cells, are present in most tissues and play an essential role in tissue homeostasis and regeneration. As a result, fibroblast-based strategies have been widely employed in tissue engineering. However, while considered to have immunosuppressive properties, the survival and functionality of allogeneic fibroblasts after transplantation remain controversial. Here, we evaluated innate and adaptive immune responses against allogeneic fibroblasts following intradermal injection into different immune-deficient mouse strains. While allogeneic fibroblasts were rejected 1 week after transplantation in immunocompetent mice, rejection did not occur in immunodeficient γ chain-deficient NOD-SCID (NSG) mice. T-cell- and B-cell-deficient RAG1 knockout mice showed greater loss of fibroblasts by day 5 after transplantation compared with NSG mice (P ≤ 0.05) but prolonged persistence compared with wild-type recipient (P ≤ 0.005). Loss of fibroblasts correlated with the expression of proinflammatory chemokine genes and infiltration of myeloid cells in the transplantation site. Depletion of macrophages and neutrophils delayed rejection, revealing the role of innate immune cells in an early elimination of fibroblasts that is followed by T-cell-mediated rejection in the second week. These findings indicate that the application of allogeneic fibroblasts in tissue engineering products requires further improvements to overcome cell rejection by innate and adaptive immune cells.

Endowing universal CAR T-cell with immune-evasive properties using TALEN-gene editing.

Universal CAR T-cell therapies are poised to revolutionize cancer treatment and to improve patient outcomes. However, realizing these advantages in an allogeneic setting requires universal CAR T-cells that can kill target tumor cells, avoid depletion by the host immune system, and proliferate without attacking host tissues. Here, we describe the development of a novel immune-evasive universal CAR T-cells scaffold using precise TALEN-mediated gene editing and DNA matrices vectorized by recombinant adeno-associated virus 6. We simultaneously disrupt and repurpose the endogenous TRAC and B2M loci to generate TCRαß- and HLA-ABC-deficient T-cells expressing the CAR construct and the NK-inhibitor named HLA-E. This highly efficient gene editing process enables the engineered T-cells to evade NK cell and alloresponsive T-cell attacks and extend their persistence and antitumor activity in the presence of cytotoxic levels of NK cell in vivo and in vitro, respectively. This scaffold could enable the broad use of universal CAR T-cells in allogeneic settings and holds great promise for clinical applications.

Mapping of a Major QTL, qBK1Z, for Bakanae Disease Resistance in Rice.

Bakanae disease is a fungal disease of rice (Oryza sativa L.) caused by the pathogen Gibberella fujikuroi (also known as Fusarium fujikuroi). This study was carried out to identify novel quantitative trait loci (QTLs) from an indica variety Zenith. We performed a QTL mapping using 180 F2:9 recombinant inbred lines (RILs) derived from a cross between the resistant variety, Zenith, and the susceptible variety, Ilpum. A primary QTL study using the genotypes and phenotypes of the RILs indicated that the locus qBK1z conferring bakanae disease resistance from the Zenith was located in a 2.8 Mb region bordered by the two RM (Rice Microsatellite) markers, RM1331 and RM3530 on chromosome 1. The log of odds (LOD) score of qBK1z was 13.43, accounting for 30.9% of the total phenotypic variation. A finer localization of qBK1z was delimited at an approximate 730 kb interval in the physical map between Chr01_1435908 (1.43 Mbp) and RM10116 (2.16 Mbp). Introducing qBK1z or pyramiding with other previously identified QTLs could provide effective genetic control of bakanae disease in rice.

Differential Depletion of Bone Marrow Resident B-ALL after Systemic Administration of Endosomal TLR Agonists.

Acute lymphoblastic leukemia (ALL) is the most common pediatric malignancy. While frontline chemotherapy regimens are generally very effective, the prognosis for patients whose leukemia returns remains poor. The presence of measurable residual disease (MRD) in bone marrow at the completion of induction therapy is the strongest predictor of relapse, suggesting that strategies to eliminate the residual leukemic blasts from this niche could reduce the incidence of recurrence. We have previously reported that toll-like receptor (TLR) agonists achieve durable T cell-mediated protection in transplantable cell line-based models of B cell precursor leukemia (B-ALL). However, the successful application of TLR agonist therapy in an MRD setting would require the induction of anti-leukemic immune activity specifically in the bone marrow, a site of the chemotherapy-resistant leukemic blasts. In this study, we compare the organ-specific depletion of human and mouse primary B-ALL cells after systemic administration of endosomal TLR agonists. Despite comparable splenic responses, only the TLR9 agonist induced strong innate immune responses in the bone marrow and achieved a near-complete elimination of B-ALL cells. This pattern of response was associated with the most significantly prolonged disease-free survival. Overall, our findings identify innate immune activity in the bone marrow that is associated with durable TLR-induced protection against B-ALL outgrowth.

Memristor Neural Network Training with Clock Synchronous Neuromorphic System.

Memristor devices are considered to have the potential to implement unsupervised learning, especially spike timing-dependent plasticity (STDP), in the field of neuromorphic hardware research. In this study, a neuromorphic hardware system for multilayer unsupervised learning was designed, and unsupervised learning was performed with a memristor neural network. We showed that the nonlinear characteristic memristor neural network can be trained by unsupervised learning only with the correlation between inputs and outputs. Moreover, a method to train nonlinear memristor devices in a supervised manner, named guide training, was devised. Memristor devices have a nonlinear characteristic, which makes implementing machine learning algorithms, such as backpropagation, difficult. The guide-training algorithm devised in this paper updates the synaptic weights by only using the correlations between inputs and outputs, and therefore, neither complex mathematical formulas nor computations are required during the training. Thus, it is considered appropriate to train a nonlinear memristor neural network. All training and inference simulations were performed using the designed neuromorphic hardware system. With the system and memristor neural network, the image classification was successfully done using both the Hebbian unsupervised training and guide supervised training methods.

Study on the mechanisms of seizure-like events suppression effect by electrical stimulation using a microelectrode array.

In this paper, we studied the mechanisms underlying the suppression of seizure-like events (SLEs) by electrical stimulation. We conducted an in-vitro experiment using entorhinal cortex combined hippocampal slices and two convulsant drugs, bicuculline and 4-aminopyridine, to induce spontaneous SLEs. We used a microelectrode array to observe network dynamics over the entire hippocampal area simultaneously, including regions far from the stimulation site. We stimulated the entorhinal cortex region, which has been determined to be a focus of SLEs by Granger causality analysis of multichannel time series data, by an external electrode. In bicuculline application, electrical stimulation showed a marked suppression effect, even though the sizes of the effective region differed. In 4-aminopyridine application, however, stimulation under the same conditions did not suppress the activities in ∼80% of SLEs. The suppression effect was more remarkable in the areas surrounding the stimulation site in both cases. Our experimental results could support the neuronal depolarization blockade mechanism by accumulation of extracellular potassium ions, which is one of the most convincing mechanisms to understand seizure suppression phenomena because of electrical stimulation. Computer simulation using a neuronal network model also confirmed the mechanism.

Prediction of the Seizure Suppression Effect by Electrical Stimulation via a Computational Modeling Approach.

In this paper, we identified factors that can affect seizure suppression via electrical stimulation by an integrative study based on experimental and computational approach. Preferentially, we analyzed the characteristics of seizure-like events (SLEs) using our previous in vitro experimental data. The results were analyzed in two groups classified according to the size of the effective region, in which the SLE was able to be completely suppressed by local stimulation. However, no significant differences were found between these two groups in terms of signal features or propagation characteristics (i.e., propagation delays, frequency spectrum, and phase synchrony). Thus, we further investigated important factors using a computational model that was capable of evaluating specific influences on effective region size. In the proposed model, signal transmission between neurons was based on two different mechanisms: synaptic transmission and the electrical field effect. We were able to induce SLEs having similar characteristics with differentially weighted adjustments for the two transmission methods in various noise environments. Although the SLEs had similar characteristics, their suppression effects differed. First of all, the suppression effect occurred only locally where directly received the stimulation effect in the high noise environment, but it occurred in the entire network in the low noise environment. Interestingly, in the same noise environment, the suppression effect was different depending on SLE propagation mechanism; only a local suppression effect was observed when the influence of the electrical field transmission was very weak, whereas a global effect was observed with a stronger electrical field effect. These results indicate that neuronal activities synchronized by a strong electrical field effect respond more sensitively to partial changes in the entire network. In addition, the proposed model was able to predict that stimulation of a seizure focus region is more effective for suppression. In conclusion, we confirmed the possibility of a computational model as a simulation tool to analyze the efficacy of deep brain stimulation (DBS) and investigated the key factors that determine the size of an effective region in seizure suppression via electrical stimulation.