Search

Halogenation plays a unique role in the design of agrochemicals. Enzymatic halogenation reactions have attracted great attention due to their excellent specificity and mild reaction conditions. S-adenosyl-l-methionine (SAM)-dependent halogenases mediate the nucleophilic attack of halide ions (X–) to SAM to produce 5′-XDA. However, only 11 SAM-dependent fluorinases and 3 chlorinases have been reported, highlighting the desire for additional halogenases. SAM-dependent hydroxide adenosyltransferase (HATase) has a similar reaction mechanism as halogenases but uses water as a substrate instead of halide ions. Here, we explored a HATase from the thermophile Thermotoga maritimaMSB8 and transformed it into a halogenase. We identified a key dyad W8L/V71T for the halogenation reaction. We also obtained the best performing mutants for each halogenation reaction: M1, M2 and M4 for Cl–, Br–and I–, respectively. The M4 mutant retained the thermostability of HATase in the iodination reaction at 80 °C, which surpasses the natural halogenase SalL. QM/MM revealed that these mutants bind halide ions with more suitable angles for nucleophilic attack of C5′ of SAM, thus conferring halogenation capabilities. Our work achieved the halide ion specificity of halogenases and generated thermostable halogenases for the first time, which provides new opportunities to expand the halogenase repertoire from hydroxylase.

The development of catalysts serves as the cornerstone of innovation in synthesis, as exemplified by the recent discovery of photoenzymes. However, the repertoire of naturally occurring enzymes repurposed by direct light excitation to catalyze new-to-nature photobiotransformations is currently limited to flavoproteins and keto-reductases. Herein, we shed light on imine reductases (IREDs) that catalyze the remote C(sp3)–C(sp3) bond formation, providing a previously elusive radical hydroalkylation of enamides for accessing chiral amines (45 examples with up to 99% enantiomeric excess). Beyond their natural function in catalyzing two-electron reductive amination reactions, upon direct visible-light excitation or in synergy with a synthetic photoredox catalyst, IREDs are repurposed to tune the non-natural photoinduced single-electron radical processes. By conducting wet mechanistic experiments and computational simulations, we unravel how engineered IREDs direct radical intermediates toward the productive and enantioselective pathway. This work represents a promising paradigm for harnessing nature’s catalysts for new-to-nature asymmetric transformations that remain challenging through traditional chemocatalytic methods.

P450 NascB catalyzes the coupling of cyclo-(l-tryptophan-l-proline) (1)to generate (−)-naseseazine C (2)through intramolecular C–N bond formation and intermolecular C–C coupling. A thorough understanding of its catalytic mechanism is crucial for the engineering or design of P450-catalyzed C–N dimerization reactions. By employing MD simulations, QM/MM calculations, and enhanced sampling, we assessed various mechanisms from recent works. Our study demonstrates that the most favorable pathway entails the transfer of a hydrogen atom from N7–H to Cpd I. Subsequently, there is a conformational change in the substrate radical, shifting it from the Re-face to the Si-face of N7 in Substrate 1. The Si-face conformation of Substrate 1 is stabilized by the protein environment and the π–π stacking interaction between the indole ring and heme porphyrin. The subsequent intermolecular C3–C6′ bond formation between Substrate 1 radical and Substrate 2 occurs via a radical attack mechanism. The conformational switch of the Substrate 1 radical not only lowers the barrier of the intermolecular C3–C6′ bond formation but also yields the correct stereoselectivity observed in experiments. In addition, we evaluated the reactivity of the ferric-superoxide species, showing it is not reactive enough to initiate the hydrogen atom abstraction from the indole NH group of the substrate. Our simulation provides a comprehensive mechanistic insight into how the P450 enzyme precisely controls both the intramolecular C–N cyclization and intermolecular C–C coupling. The current findings align with the available experimental data, emphasizing the pivotal role of substrate dynamics in governing P450 catalysis.

In order to realize the 1000-s long pulse operation of the electron cyclotron resonance heating (ECRH) system on experimental advanced superconducting tokamak (EAST), we developed an ECRH steady-state operation control system. Four working modes were designed, namely, manual restart mode, automatic restart mode, timed alternate operation mode, and protection trigger alternate operation mode. The field-programmable gate array (FPGA)-based NI cRIO was used as the control device, and the LabVIEW was used to write the program. The system has a quick response and accurate timing sequence and can realize the long-pulse steady-state operation of the gyrotrons in EAST ECRH system. The steady-state operation control system has been verified in nearly two rounds of EAST experiments, helping EAST achieve the 1056-s long pulse discharge.

The biosynthesis of blasticidin S has drawn attention due to the participation of the radical S-adenosyl methionine (SAM) enzyme BlsE. The original assignment of BlsE as a radical-mediated, redox-neutral decarboxylase is unusual because this reaction appears to serve no biosynthetic purpose and would need to be reversed by a subsequent carboxylation step. Furthermore, with the exception of BlsE, all other radical SAM decarboxylases reported to date are oxidative in nature. Careful analysis of the BlsE reaction, however, demonstrates that BlsE is not a decarboxylase but instead a lyase that catalyzes the dehydration of cytosylglucuronic acid (CGA) to form cytosyl-4′-keto-3′-deoxy-d-glucuronic acid, which can rapidly decarboxylate nonenzymatically in vitro. Analysis of substrate isotopologs, fluorinated analogues, as well as computational models based on X-ray crystal structures of the BlsE·SAM (2.09 Å) and BlsE·SAM·CGA (2.62 Å) complexes suggests that BlsE catalysis likely proceeds via direct elimination of water from the CGA C4′ α-hydroxyalkyl radical as opposed to 1,2-migration of the C3′-hydroxyl prior to dehydration. Biosynthetic and mechanistic implications of the revised assignment of BlsE are discussed.

Enzymes are increasingly explored for use in asymmetric synthesis1–3, but their applications are generally limited by the reactions available to naturally occurring enzymes. Recently, interest in photocatalysis4has spurred the discovery of novel reactivity from known enzymes5. However, so far photoinduced enzymatic catalysis6has not been used for the cross-coupling of two molecules. For example, the intermolecular coupling of alkenes with α-halo carbonyl compounds through a visible-light-induced radical hydroalkylation, which could provide access to important γ-chiral carbonyl compounds, has not yet been achieved by enzymes. The major challenges are the inherent poor photoreactivity of enzymes and the difficulty in achieving stereochemical control of the remote prochiral radical intermediate7. Here we report a visible-light-induced intermolecular radical hydroalkylation of terminal alkenes that does not occur naturally, catalysed by an ‘ene’ reductase using readily available α-halo carbonyl compounds as reactants. This method provides an efficient approach to the synthesis of various carbonyl compounds bearing a γ-stereocentre with excellent yields and enantioselectivities (up to 99 per cent yield with 99 per cent enantiomeric excess), which otherwise are difficult to access using chemocatalysis. Mechanistic studies suggest that the formation of the complex of the substrates (α-halo carbonyl compounds) and the ‘ene’ reductase triggers the enantioselective photoinduced radical reaction. Our work further expands the reactivity repertoire of biocatalytic, synthetically useful asymmetric transformations by the merger of photocatalysis and enzyme catalysis.

The proliferation of large‐scale distributed Internet of Things (IoT) applications has resulted in a surge in demand for network models such as deep neural networks (DNNs) to be trained and inferred at the edge. Due to the central data transmission mechanism, heterogeneity of edge devices, and resource constraints, the existing single data‐parallel, and model‐parallel distributed training mechanisms frequently fail to fully utilize the computing power of edge devices, network topology and bandwidth resources. In light of the shortcomings mentioned earlier, this article proposes EdgeMesh, a hybrid parallel training mode based on the Mesh‐Tensorflow framework, consisting of an adaptive meshing strategy and a dynamic model convolutional partitioning strategy. The computing power of IoT edge devices significantly speeds up the DNN training process. In a resource‐constrained environment, each node only supports a subset of the model's parallel computing tasks, reducing communication and memory overhead while retaining high scalability. Experiments show that when compared to single‐machine training and data parallel mode, EdgeMesh distributed training mechanism can reduce the average delay by 3.2 times and average memory overhead by 43% while maintaining model accuracy. The computing power of IoT edge devices effectively accelerates the DNN training process. This article analyzes the advantages and disadvantages of existing distributed deep learning training methods in edge scenes, and proposes EdgeMesh, a hybrid distributed training method, which includes: A hybrid distributed training framework. The framework is a model‐parallel parameter server architecture among multiple devices, which integrates data parallelism and model parallelism. The parameter server architecture composed of multiple model‐parallel nodes ensures the accuracy of model training. It solves the resource limitation of deep neural network training on edge devices and accelerates the model training process.EdgeMesh device meshing algorithm. The algorithm divides a layout that adapts to the hybrid distributed training framework according to the difference in floating‐point computing capabilities of edge devices, aiming to reduce the training delays caused by performance differences between clusters.Partitioning strategy for dynamic model convolution. It is a dynamic load balancing strategy based on a hybrid distributed training framework that takes into account the bandwidth of the runtime environment, the floating point computing capacity of the device, and the network latency. The algorithm is deployed in a hybrid training scenario of data parallelism and model parallelism based on Tensorflow and Mesh‐Tensorflow, which effectively reduces the problem of long training latency on heterogeneous edge devices.

An important issue for service providers to consider before building a mobile edge services network is the limited budget for edge server deployment. In addition, the geographic position of unmanned aerial vehicle (UAV) edge server will affect its energy consumption. Therefore, we have established a UAV‐assisted mobile edge computing (MEC) system. UAV acts as a mobile edge server to provide computing services for user equipment (UE). This system aim to minimize the total energy consumption and deployment cost required for the UAVs to complete offload tasks and hover. To optimize the total energy consumption and deployment cost of UAVs, we proposed an improved mean shift (IMS) algorithm, which jointly optimizing the location and number of UAV edge servers. Various simulation results show the effectiveness of our proposed scheme in reducing energy consumption compared to other deployment methods. Furthermore, our research is of great significance for service providers in control the trade‐off between investment cost and energy consumption. This article proposes an improved mean shift algorithm to deploy UAV edge servers. First, we improve the bandwidth value in the kernel function, and get the optimal bandwidth by calculating the number of clusters under different bandwidth. Then we use the optimal bandwidth as the input value of the mean shift radius to get the final hovering number and position of UAVs. The simulation results show that the scheme can significantly reduce the energy consumption and deployment cost of UAV, and has better delay performance.

To study the possible mechanism of the technetium-99m complex of dimercaptosuccinic acid 99mTc(V)-DMSA accumulation in a model of Staphylococcus aureus-induced bacterial arthritis regulated by glucose.

We have examined the interactions between NMDA receptors and opioid effects in isolated neonatal rat spinal cord. Electrical stimulation of a lumbar dorsal root evoked a nociceptive-related slow ventral root potential (sVRP) recorded at the corresponding ipsilateral ventral root. The κ opiate receptor agonist U69,593 (2.5 nM-1 μM) depressed sVRP area by a maximum of 80%; EC50was approximately 33 nM. Both the non-specific antagonist naloxone and the κ-specific antagonist nor-binaltorphimine (nor-BNI) antagonized the effects of U69,593. Morphine, a μ agonist, (1 nM-1 μM) depressed sVRP area with an approximate EC50of 90 nM. The effects of both μ and κ opioid agonists were selective for the very slow metabotropically mediated components of the sVRP, compared to the relatively fast NMDA receptor-mediated components. The non-competitive N-methyl- d-aspartate (NMDA) antagonist MK-801 (20 nM) had no effect on sVRP area when applied alone but co-applied with morphine significantly potentiated the depressant effects of morphine. In contrast, MK-801 either had no effect on or slightly antagonized the depressant effects of U69,593. Naloxone following morphine produced a significant increase in sVRP area above pre-morphine control values; the increase lasted 30 min or more. Neither naloxone nor nor-BNI was associated with an increase in sVRP area when given alone or following U69,593. MK-801 co-applied with morphine blocked the rebound increase in sVRP area following naloxone. These results suggest that (1) both μ and κ receptor agonists exert similar selective depressant effects on spinal nociceptive neurotransmission; (2) μ but not κ agonists exert prolonged excitatory effects that oppose the depression; and (3) NMDA receptors play a role in determining opioid analgesic potency and naloxone-precipitated hyperresponsiveness. The results may be related to initial steps in the development of acute tolerance to μ opioids, and suggest that tolerance to κ opioids may have a different mechanism.