Ginseng Sprouts Attenuate Mortality and Systemic Inflammation by Modulating TLR4/NF-κB Signaling in an LPS-Induced Mouse Model of Sepsis.

Sepsis leads to multi-organ failure due to aggressive systemic inflammation, which is one of the main causes of death clinically. This study aimed to evaluate whether ginseng sprout extracts (GSE) can rescue sepsis and explore its underlying mechanisms. C57BL/6J male mice (n = 15/group) were pre-administered with GSE (25, 50, and 100 mg/kg, p.o) for 5 days, and a single injection of lipopolysaccharide (LPS, 30 mg/kg, i.p) was administered to construct a sepsis model. Additionally, RAW264.7 cells were treated with LPS with/without GSE/its main components (Rd and Re) to explain the mechanisms corresponding to the animal-derived effects. LPS injection led to the death of all mice within 38 h, while GSE pretreatment delayed the time to death. GSE pretreatment also notably ameliorated LPS-induced systemic inflammation such as histological destruction in both the lung and liver, along with reductions in inflammatory cytokines, such as TNF-α, IL-6, and IL-1ß, in both tissues and serum. Additionally, GSE markedly diminished the drastic secretion of nitric oxide (NO) by suppressing the expression levels of inducible nitric oxide synthase (iNOS) and cyclooxygenase 2 (COX2) in both tissues. Similar changes in TNF-α, IL-1ß, NO, iNOS, and COX2 were observed in LPS-stimulated RAW264.7 cells, and protein expression data and nuclear translocation assays suggested GSE could modulate LPS-binding protein (LBP), Toll-like receptor 4 (TLR4), and NF-κB. Ginsenoside Rd could be a major active component in GSE that produces the anti-sepsis effects. Our data support that ginseng sprouts could be used as an herbal resource to reduce the risk of sepsis. The corresponding mechanisms may involve TLR4/NF-κB signaling and a potentially active component.

Low Complexity Gradient Computation Techniques to Accelerate Deep Neural Network Training.

Deep neural network (DNN) training is an iterative process of updating network weights, called gradient computation, where (mini-batch) stochastic gradient descent (SGD) algorithm is generally used. Since SGD inherently allows gradient computations with noise, the proper approximation of computing weight gradients within SGD noise can be a promising technique to save energy/time consumptions during DNN training. This article proposes two novel techniques to reduce the computational complexity of the gradient computations for the acceleration of SGD-based DNN training. First, considering that the output predictions of a network (confidence) change with training inputs, the relation between the confidence and the magnitude of the weight gradient can be exploited to skip the gradient computations without seriously sacrificing the accuracy, especially for high confidence inputs. Second, the angle diversity-based approximations of intermediate activations for weight gradient calculation are also presented. Based on the fact that the angle diversity of gradients is small (highly uncorrelated) in the early training epoch, the bit precision of activations can be reduced to 2-/4-/8-bit depending on the resulting angle error between the original gradient and quantized gradient. The simulations show that the proposed approach can skip up to 75.83% of gradient computations with negligible accuracy degradation for CIFAR-10 dataset using ResNet-20. Hardware implementation results using 65-nm CMOS technology also show that the proposed training accelerator achieves up to 1.69× energy efficiency compared with other training accelerators.

Korean Red Ginseng Ameliorates Fatigue via Modulation of 5-HT and Corticosterone in a Sleep-Deprived Mouse Model.

Central fatigue, which is neuromuscular dysfunction associated with neurochemical alterations, is an important clinical issue related to pathologic fatigue. This study aimed to investigate the anti-central fatigue effect of Korean red ginseng (KRG) and its underlying mechanism. Male BALB/c mice (8 weeks old) were subjected to periodic sleep deprivation (SD) for 6 cycles (forced wakefulness for 2 days + 1 normal day per cycle). Simultaneously, the mice were administered KRG (0, 100, 200, or 400 mg/kg) or ascorbic acid (100 mg/kg). After all cycles, the rotarod and grip strength tests were performed, and then the changes regarding stress- and neurotransmitter-related parameters in serum and brain tissue were evaluated. Six cycles of SD notably deteriorated exercise performance in both the rotarod and grip strength tests, while KRG administration significantly ameliorated these alterations. KRG also significantly attenuated the SD-induced depletion of serum corticosterone. The levels of main neurotransmitters related to the sleep/wake cycle were markedly altered (serotonin was overproduced while dopamine levels were decreased) by SD, and KRG significantly attenuated these alterations through relevant molecules including brain-derived neurotropic factor and serotonin transporter. This study demonstrated the anti-fatigue effects of KRG in an SD mouse model, indicating the clinical relevance of KRG.

Exploiting Retraining-Based Mixed-Precision Quantization for Low-Cost DNN Accelerator Design.

For successful deployment of deep neural networks (DNNs) on resource-constrained devices, retraining-based quantization has been widely adopted to reduce the number of DRAM accesses. By properly setting training parameters, such as batch size and learning rate, bit widths of both weights and activations can be uniformly quantized down to 4 bit while maintaining full precision accuracy. In this article, we present a retraining-based mixed-precision quantization approach and its customized DNN accelerator to achieve high energy efficiency. In the proposed quantization, in the middle of retraining, an additional bit (extra quantization level) is assigned to the weights that have shown frequent switching between two contiguous quantization levels since it means that both quantization levels cannot help to reduce quantization loss. We also mitigate the gradient noise that occurs in the retraining process by taking a lower learning rate near the quantization threshold. For the proposed novel mixed-precision quantized network (MPQ-network), we have implemented a customized accelerator using a 65-nm CMOS process. In the accelerator, the proposed processing elements (PEs) can be dynamically reconfigured to process variable bit widths from 2 to 4 bit for both weights and activations. The numerical results show that the proposed quantization can achieve 1.37 × better compression ratio for VGG-9 using CIFAR-10 data set compared with a uniform 4-bit (both weights and activations) model without loss of classification accuracy. The proposed accelerator also shows 1.29× of energy savings for VGG-9 using the CIFAR-10 data set over the state-of-the-art accelerator.

Physiological responses of insects to microbial fermentation products: Insights from the interactions between Drosophila and acetic acid.

Acetic acid is a fermentation product of many microorganisms, including some that inhabit the food and guts of Drosophila. Here, we investigated the effect of dietary acetic acid on oviposition and larval performance of Drosophila. At all concentrations tested (0.34-3.4%), acetic acid promoted egg deposition by mated females in no-choice assays; and females preferred to oviposit on diet with acetic acid relative to acetic acid-free diet. However, acetic acid depressed larval performance, particularly extending the development time of both larvae colonized with the bacterium Acetobacter pomorum and axenic (microbe-free) larvae. The larvae may incur an energetic cost associated with dissipating the high acid load on acetic acid-supplemented diets. This effect was compounded by suppressed population growth of A. pomorum on the 3.4% acetic acid diet, such that the gnotobiotic Drosophila on this diet displayed traits characteristic of axenic Drosophila, specifically reduced developmental rate and elevated lipid content. It is concluded that acetic acid is deleterious to larval Drosophila, and hypothesized that acetic acid may function as a reliable cue for females to oviposit in substrates bearing microbial communities that promote larval nutrition.

Type I interferons and the cytokine TNF cooperatively reprogram the macrophage epigenome to promote inflammatory activation.

Cross-regulation of Toll-like receptor (TLR) responses by cytokines is essential for effective host defense, avoidance of toxicity and homeostasis, but the underlying mechanisms are not well understood. Our comprehensive epigenomics approach to the analysis of human macrophages showed that the proinflammatory cytokines TNF and type I interferons induced transcriptional cascades that altered chromatin states to broadly reprogram responses induced by TLR4. TNF tolerized genes encoding inflammatory molecules to prevent toxicity while preserving the induction of genes encoding antiviral and metabolic molecules. Type I interferons potentiated the inflammatory function of TNF by priming chromatin to prevent the silencing of target genes of the transcription factor NF-κB that encode inflammatory molecules. The priming of chromatin enabled robust transcriptional responses to weak upstream signals. Similar chromatin regulation occurred in human diseases. Our findings reveal that signaling crosstalk between interferons and TNF is integrated at the level of chromatin to reprogram inflammatory responses, and identify previously unknown functions and mechanisms of action of these cytokines.