Structure-based functional mechanisms and biotechnology applications of anti-CRISPR proteins.

CRISPR loci and Cas proteins provide adaptive immunity in prokaryotes against invading bacteriophages and plasmids. In response, bacteriophages have evolved a broad spectrum of anti-CRISPR proteins (anti-CRISPRs) to counteract and overcome this immunity pathway. Numerous anti-CRISPRs have been identified to date, which suppress single-subunit Cas effectors (in CRISPR class 2, type II, V and VI systems) and multisubunit Cascade effectors (in CRISPR class 1, type I and III systems). Crystallography and cryo-electron microscopy structural studies of anti-CRISPRs bound to effector complexes, complemented by functional experiments in vitro and in vivo, have identified four major CRISPR-Cas suppression mechanisms: inhibition of CRISPR-Cas complex assembly, blocking of target binding, prevention of target cleavage, and degradation of cyclic oligonucleotide signalling molecules. In this Review, we discuss novel mechanistic insights into anti-CRISPR function that have emerged from X-ray crystallography and cryo-electron microscopy studies, and how these structures in combination with function studies provide valuable tools for the ever-growing CRISPR-Cas biotechnology toolbox, to be used for precise and robust genome editing and other applications.

Type IV-A CRISPR-Csf complex: Assembly, dsDNA targeting, and CasDinG recruitment.

Type IV CRISPR-Cas systems, which are primarily found on plasmids and exhibit a strong plasmid-targeting preference, are the only one of the six known CRISPR-Cas types for which the mechanistic details of their function remain unknown. Here, we provide high-resolution functional snapshots of type IV-A Csf complexes before and after target dsDNA binding, either in the absence or presence of CasDinG, revealing the mechanisms underlying CsfcrRNA complex assembly, "DWN" PAM-dependent dsDNA targeting, R-loop formation, and CasDinG recruitment. Furthermore, we establish that CasDinG, a signature DinG family helicase, harbors ssDNA-stimulated ATPase activity and ATP-dependent 5'-3' DNA helicase activity. In addition, we show that CasDinG unwinds the non-target strand (NTS) and target strand (TS) of target dsDNA from the CsfcrRNA complex. These molecular details advance our mechanistic understanding of type IV-A CRISPR-Csf function and should enable Csf complexes to be harnessed as genome-engineering tools for biotechnological applications.

Bacterial origins of cyclic nucleotide-activated antiviral immune signaling.

Second-messenger-mediated signaling by cyclic oligonucleotides (cOs) composed of distinct base, ring size, and 3'-5'/2'-5' linkage combinations constitutes the initial trigger resulting in activation of signaling pathways that have an impact on immune-mediated antiviral defense against invading viruses and phages. Bacteria and archaea have evolved CRISPR, CBASS, Pycsar, and Thoeris surveillance complexes that involve cO-mediated activation of effectors resulting in antiviral defense through either targeted nuclease activity, effector oligomerization-mediated depletion of essential cellular metabolites or disruption of host cell membrane functions. Notably, antiviral defense capitalizes on an abortive infection mechanism, whereby infected cells die prior to completion of the phage replication cycle. In turn, phages have evolved small proteins that target and degrade/sequester cOs, thereby suppressing host immunity. This review presents a structure-based mechanistic perspective of recent advances in the field of cO-mediated antiviral defense, in particular highlighting the ancient evolutionary adaptation by metazoans of bacterial cell-autonomous innate immune mechanisms.

Second Messenger cA4 Formation within the Composite Csm1 Palm Pocket of Type III-A CRISPR-Cas Csm Complex and Its Release Path.

Target RNA binding to crRNA-bound type III-A CRISPR-Cas multi-subunit Csm surveillance complexes activates cyclic-oligoadenylate (cAn) formation from ATP subunits positioned within the composite pair of Palm domain pockets of the Csm1 subunit. The generated cAn second messenger in turn targets the CARF domain of trans-acting RNase Csm6, triggering its HEPN domain-based RNase activity. We have undertaken cryo-EM studies on multi-subunit Thermococcus onnurineus Csm effector ternary complexes, as well as X-ray studies on Csm1-Csm4 cassette, both bound to substrate (AMPPNP), intermediates (pppAn), and products (cAn), to decipher mechanistic aspects of cAn formation and release. A network of intermolecular hydrogen bond alignments accounts for the observed adenosine specificity, with ligand positioning dictating formation of linear pppAn intermediates and subsequent cAn formation by cyclization. We combine our structural results with published functional studies to highlight mechanistic insights into the role of the Csm effector complex in mediating the cAn signaling pathway.

CRISPR-Cas III-A Csm6 CARF Domain Is a Ring Nuclease Triggering Stepwise cA4 Cleavage with ApA>p Formation Terminating RNase Activity.

Type III-A CRISPR-Cas surveillance complexes containing multi-subunit Csm effector, guide, and target RNAs exhibit multiple activities, including formation of cyclic-oligoadenylates (cAn) from ATP and subsequent cAn-mediated cleavage of single-strand RNA (ssRNA) by the trans-acting Csm6 RNase. Our structure-function studies have focused on Thermococcus onnurineus Csm6 to deduce mechanistic insights into how cA4 binding to the Csm6 CARF domain triggers the RNase activity of the Csm6 HEPN domain and what factors contribute to regulation of RNA cleavage activity. We demonstrate that the Csm6 CARF domain is a ring nuclease, whereby bound cA4 is stepwise cleaved initially to ApApApA>p and subsequently to ApA>p in its CARF domain-binding pocket, with such cleavage bursts using a timer mechanism to regulate the RNase activity of the Csm6 HEPN domain. In addition, we establish T. onnurineus Csm6 as an adenosine-specific RNase and identify a histidine in the cA4 CARF-binding pocket involved in autoinhibitory regulation of RNase activity.

Type III-A CRISPR-Cas Csm Complexes: Assembly, Periodic RNA Cleavage, DNase Activity Regulation, and Autoimmunity.

Type ΙΙΙ CRISPR-Cas systems provide robust immunity against foreign RNA and DNA by sequence-specific RNase and target RNA-activated sequence-nonspecific DNase and RNase activities. We report on cryo-EM structures of Thermococcus onnurineus CsmcrRNA binary, CsmcrRNA-target RNA and CsmcrRNA-target RNAanti-tag ternary complexes in the 3.1 Å range. The topological features of the crRNA 5'-repeat tag explains the 5'-ruler mechanism for defining target cleavage sites, with accessibility of positions -2 to -5 within the 5'-repeat serving as sensors for avoidance of autoimmunity. The Csm3 thumb elements introduce periodic kinks in the crRNA-target RNA duplex, facilitating cleavage of the target RNA with 6-nt periodicity. Key Glu residues within a Csm1 loop segment of CsmcrRNA adopt a proposed autoinhibitory conformation suggestive of DNase activity regulation. These structural findings, complemented by mutational studies of key intermolecular contacts, provide insights into CsmcrRNA complex assembly, mechanisms underlying RNA targeting and site-specific periodic cleavage, regulation of DNase cleavage activity, and autoimmunity suppression.

Target ssDNA activates the NADase activity of prokaryotic SPARTA immune system.

Argonaute proteins (Agos), which use small RNAs or DNAs as guides to recognize complementary nucleic acid targets, mediate RNA silencing in eukaryotes. In prokaryotes, Agos are involved in immunity: the short prokaryotic Ago/TIR-APAZ (SPARTA) immune system triggers cell death by degrading NAD+ in response to invading plasmids, but its molecular mechanisms remain unknown. Here we used cryo-electron microscopy to determine the structures of inactive monomeric and active tetrameric Crenotalea thermophila SPARTA complexes, revealing mechanisms underlying SPARTA assembly, RNA-guided recognition of target single-stranded DNA (ssDNA) and subsequent SPARTA tetramerization, as well as tetramerization-dependent NADase activation. The small RNA guides Ago to recognize its ssDNA target, inducing SPARTA tetramerization via both Ago- and TIR-mediated interactions and resulting in a two-stranded, parallel, head-to-tail TIR rearrangement primed for NAD+ hydrolysis. Our findings thus identify the molecular basis for target ssDNA-mediated SPARTA activation, which will facilitate the development of SPARTA-based biotechnological tools.

The molecular chaperone mtHSC70-1 interacts with DjA30 to regulate female gametophyte development and fertility in Arabidopsis.

Arabidopsis mitochondria-targeted heat shock protein 70 (mtHSC70-1) plays important roles in the establishment of cytochrome c oxidase-dependent respiration and redox homeostasis during the vegetative growth of plants. Here, we report that knocking out the mtHSC70-1 gene led to a decrease in plant fertility; the fertility defect of the mutant was completely rescued by introducing the mtHSC70-1 gene. mtHSC70-1 mutants also showed defects in female gametophyte (FG) development, including delayed mitosis, abnormal nuclear position, and ectopic gene expression in the embryo sacs. In addition, we found that an Arabidopsis mitochondrial J-protein gene (DjA30) mutant, j30+/- , had defects in FG development and fertility similar to those of mtHSC70-1 mutant. mtHSC70-1 and DjA30 had similar expression patterns in FGs and interacted in vivo, suggesting that these two proteins might cooperate during female gametogenesis. Further, respiratory chain complex IV activity in mtHSC70-1 and DjA30 mutant embryo sacs was markedly downregulated; this led to the accumulation of mitochondrial reactive oxygen species (ROS). Scavenging excess ROS by introducing Mn-superoxide dismutase 1 or catalase 1 gene into the mtHSC70-1 mutant rescued FG development and fertility. Altogether, our results suggest that mtHSC70-1 and DjA30 are essential for the maintenance of ROS homeostasis in the embryo sacs and provide direct evidence for the roles of ROS homeostasis in embryo sac maturation and nuclear patterning, which might determine the fate of gametic and accessory cells.

Training machine learning potentials for reactive systems: A Colab tutorial on basic models.

In the last several years, there has been a surge in the development of machine learning potential (MLP) models for describing molecular systems. We are interested in a particular area of this field - the training of system-specific MLPs for reactive systems - with the goal of using these MLPs to accelerate free energy simulations of chemical and enzyme reactions. To help new members in our labs become familiar with the basic techniques, we have put together a self-guided Colab tutorial (https://cc-ats.github.io/mlp_tutorial/), which we expect to be also useful to other young researchers in the community. Our tutorial begins with the introduction of simple feedforward neural network (FNN) and kernel-based (using Gaussian process regression, GPR) models by fitting the two-dimensional Müller-Brown potential. Subsequently, two simple descriptors are presented for extracting features of molecular systems: symmetry functions (including the ANI variant) and embedding neural networks (such as DeepPot-SE). Lastly, these features will be fed into FNN and GPR models to reproduce the energies and forces for the molecular configurations in a Claisen rearrangement reaction.

Visual Sensor with Host-Guest Specific Recognition and Light-Electrical Co-Controlled Switch.

Perception of UV radiation has important applications in medical health, industrial production, electronic communication, etc. In numerous application scenarios, there is an increasing demand for the intuitive and low-cost detection of UV radiation through colorimetric visual behavior, as well as the efficient and multi-functional utilization of UV radiation. However, photodetectors based on photoconductive modes or photosensitive colorimetric materials are not conducive to portable or multi-scene applications owing to their complex and expensive photosensitive components, potential photobleaching, and single-stimulus response behavior. Here, a multifunctional visual sensor based on the "host-guest photo-controlled permutation" strategy and the "lock and key" model is developed. The host-guest specific molecular recognition and electrochromic sensing platform is integrated at the micro-molecular scale, enabling multi-functional and multi-scene applications in the convenient and fast perception of UV radiation, military camouflage, and information erasure at the macro level of human-computer interaction through light-electrical co-controlled visual switching characteristics. This light-electrical co-controlled visual sensor based on an optoelectronic multi-mode sensing system is expected to provide new ideas and paradigms for healthcare, microelectronics manufacturing, and wearable electronic devices owing to its advantages of signal visualization, low energy consumption, low cost, and versatility.

CProMG: controllable protein-oriented molecule generation with desired binding affinity and drug-like properties.

MOTIVATION: Deep learning-based molecule generation becomes a new paradigm of de novo molecule design since it enables fast and directional exploration in the vast chemical space. However, it is still an open issue to generate molecules, which bind to specific proteins with high-binding affinities while owning desired drug-like physicochemical properties. RESULTS: To address these issues, we elaborate a novel framework for controllable protein-oriented molecule generation, named CProMG, which contains a 3D protein embedding module, a dual-view protein encoder, a molecule embedding module, and a novel drug-like molecule decoder. Based on fusing the hierarchical views of proteins, it enhances the representation of protein binding pockets significantly by associating amino acid residues with their comprising atoms. Through jointly embedding molecule sequences, their drug-like properties, and binding affinities w.r.t. proteins, it autoregressively generates novel molecules having specific properties in a controllable manner by measuring the proximity of molecule tokens to protein residues and atoms. The comparison with state-of-the-art deep generative methods demonstrates the superiority of our CProMG. Furthermore, the progressive control of properties demonstrates the effectiveness of CProMG when controlling binding affinity and drug-like properties. After that, the ablation studies reveal how its crucial components contribute to the model respectively, including hierarchical protein views, Laplacian position encoding as well as property control. Last, a case study w.r.t. protein illustrates the novelty of CProMG and the ability to capture crucial interactions between protein pockets and molecules. It's anticipated that this work can boost de novo molecule design. AVAILABILITY AND IMPLEMENTATION: The code and data underlying this article are freely available at https://github.com/lijianing0902/CProMG.

Impact of Hepatic Pedicle Clamping on Long-Term Survival Following Hepatectomy for Hepatocellular Carcinoma: Stratified Analysis Based on Intraoperative Blood Transfusion Status.

BACKGROUND: Hepatic pedicle clamping (HPC) is frequently utilized during hepatectomy to reduce intraoperative bleeding and diminish the need for intraoperative blood transfusion (IBT). The long-term prognostic implications of HPC following hepatectomy for hepatocellular carcinoma (HCC) remain under debate. This study aims to elucidate the association between HPC and oncologic outcomes after HCC resection, stratified by whether IBT was administered. PATIENTS AND METHODS: Prospectively collected data on patients with HCC who underwent curative resection from a multicenter database was studied. Patients were stratified into two cohorts on the basis of whether IBT was administered. The impact of HPC on long-term overall survival (OS) and recurrence-free survival (RFS) between the two cohorts was assessed by univariable and multivariable Cox regression analyses. RESULTS: Of 3362 patients, 535 received IBT. In the IBT cohort, using or not using HPC showed no significant difference in OS and RFS outcomes (5-year OS and RFS rates 27.9% vs. 24.6% and 13.8% vs. 12.0%, P = 0.810 and 0.530). However, in the non-IBT cohort of 2827 patients, the HPC subgroup demonstrated significantly decreased OS (5-year 45.9% vs. 56.5%, P < 0.001) and RFS (5-year 24.7% vs. 33.3%, P < 0.001) when compared with the subgroup without HPC. Multivariable Cox regression analysis identified HPC as an independent risk factor of OS and RFS [hazard ratios (HR) 1.16 and 1.12, P = 0.024 and 0.044, respectively] among patients who did not receive IBT. CONCLUSIONS: The impact of HPC on the oncological outcomes following hepatectomy for patients with HCC differed significantly whether IBT was administered, and HPC adversely impacted on long-term survival for patients without receiving IBT during hepatectomy.

Binding patterns of inhibitors to different pockets of kinesin Eg5.

The kinesin-5 family member, Eg5, plays very important role in the mitosis. As a mitotic protein, Eg5 is the target of various mitotic inhibitors. There are two targeting pockets in the motor domain of Eg5, which locates in the α2/L5/α3 region and the α4/α6 region respectively. We investigated the interactions between the different inhibitors and the two binding pockets of Eg5 by using all-atom molecular dynamics method. Combined the conformational analysis with the free-energy calculation, the binding patterns of inhibitors to the two binding pockets are shown. The α2/L5/α3 pocket can be divided into 4 regions. The structures and binding conformations of inhibitors in region 1 and 2 are highly conserved. The shape of α4/α6 pocket is alterable. The space of this pocket in ADP-binding state of Eg5 is larger than that in ADP·Pi-binding state due to the limitation of a hydrogen bond formed in the ADP·Pi-binding state. The results of this investigation provide the structural basis of the inhibitor-Eg5 interaction and offer a reference for the Eg5-targeted drug design.

The clinical trajectory of peripheral blood immune cell subsets, T-cell activation, and cytokines in septic patients.

OBJECTIVE AND DESIGN: Changes in the immune status of patients with sepsis may have a major impact on their prognosis. Our research focused on changes in various immune cell subsets and T-cell activation during the progression of sepsis. METHODS AND SUBJECTS: We collected data from 188 sepsis patients at the First Affiliated Hospital of Zhejiang University School of Medicine. The main focus was on the patient's immunocyte subset typing, T-cell activation/Treg cell analysis, and cytokine assay, which can indicate the immune status of the patient. RESULTS: The study found that the number of CD4+ T cells, CD8+ T cells, NK cells, and B cells decreased early in the disease, and the decrease in CD4+ and CD8+ T cells was more pronounced in the death group. T lymphocyte activation was inhibited, and the number of Treg cells increased as the disease progressed. T lymphocyte inhibition was more significant in the death group, and the increase in IL-10 was more significant in the death group. Finally, we used patients' baseline conditions and immunological detection indicators for modeling and found that IL-10, CD4+ Treg cells, CD3+HLA-DR+ T cells, and CD3+CD69+ T cells could predict patients' prognosis well. CONCLUSION: Our study found that immunosuppression occurs in patients early in sepsis. Early monitoring of the patient's immune status may provide a timely warning of the disease.

Purinergic P2X7 receptor involves in anti-retinal photodamage effects of berberine.

Berberine (BBR) is a Chinese herb with antioxidant and anti-inflammatory properties. In a previous study, we found that BBR had a protective effect against light-induced retinal degeneration in BALB/c mice. The purinergic P2X7 receptor (P2X7R) plays a key role in retinal degeneration via inducing oxidative stress, inflammatory changes, and cell death. The aim of this study was to investigate whether BBR can induce protective effects in light damage experiments and whether P2X7R can get involved in these effects. C57BL/6 J mice and P2X7 knockout (KO) mice on the C57BL/6 J background were used. We found that BBR preserved the outer nuclear layer (ONL) thickness and retinal ganglion cells following light stimulation. Furthermore, BBR significantly suppressed photoreceptor apoptosis, pro-apoptotic c-fos expression, pro-inflammatory responses of Mϋller cells, and inflammatory factors (TNF-α, IL-1ß). In addition, protein levels of P2X7R were downregulated in BBR-treated mice. Double immunofluorescence showed that BBR reduced overexpression of P2X7R in retinal ganglion cells and Mϋller cells. Furthermore, BBR combined with the P2X7R agonist BzATP blocked the effects of BBR on retinal morphology and photoreceptor apoptosis. However, in P2X7 KO mice, BBR had an additive effect resulting in thicker ONL and more photoreceptors. The data suggest that the P2X7 receptor is involved in retinal light damage, and BBR inhibits this process by reducing histological impairment, cell death, and inflammatory responses.

Potent inhibitors targeting cyclin-dependent kinase 9 discovered via virtual high-throughput screening and absolute binding free energy calculations.

Due to the crucial regulatory mechanism of cyclin-dependent kinase 9 (CDK9) in mRNA transcription, the development of kinase inhibitors targeting CDK9 holds promise as a potential treatment strategy for cancer. A structure-based virtual screening approach has been employed for the discovery of potential novel CDK9 inhibitors. First, compounds with kinase inhibitor characteristics were identified from the ZINC15 database via virtual high-throughput screening. Next, the predicted binding modes were optimized by molecular dynamics simulations, followed by precise estimation of binding affinities using absolute binding free energy calculations based on the free energy perturbation scheme. The binding mode of molecule 006 underwent an inward-to-outward flipping, and the new binding mode exhibited binding affinity comparable to the small molecule T6Q in the crystal structure (PDB ID: 4BCF), highlighting the essential role of molecular dynamics simulation in capturing a plausible binding pose bridging docking and absolute binding free energy calculations. Finally, structural modifications based on these findings further enhanced the binding affinity with CDK9. The results revealed that enhancing the molecule's rigidity through ring formation, while maintaining the major interactions, reduced the entropy loss during the binding process and, thus, enhanced binding affinities.

Sodium tanshinone IIA sulfonate protects vascular relaxation in ApoE-knockout mice by inhibiting the SYK-NLRP3 inflammasome-MMP2/9 pathway.

BACKGROUND: Hyperlipidemia damages vascular wall and serves as a foundation for diseases such as atherosclerosis, hypertension and stiffness. The NOD-like receptor family pyrin domain-containing 3 (NLRP3) inflammasome is implicated in vascular dysfunction associated with hyperlipidemia-induced vascular injury. Sodium tanshinone IIA sulfonate (STS), a well-established cardiovascular protective drug with recognized anti-inflammatory, antioxidant, and vasodilatory properties, is yet to be thoroughly investigated for its impact on vascular relaxant imbalance induced by hyperlipidemia. METHODS: In this study, we treated ApoE-knockout (ApoE-/-) mouse with STS and assessed the activation of the NLRP3 inflammasome, expression of MMP2/9, integrity of elastic fibers, and vascular constriction and relaxation. RESULTS: Our findings reveal that STS intervention effectively preserves elastic fibers, significantly restores aortic relaxation function in ApoE-/- mice, and reduces their excessive constriction. Furthermore, STS inhibits the phosphorylation of spleen tyrosine kinase (SYK), suppresses NLRP3 inflammasome activation, and reduces MMP2/9 expression. CONCLUSIONS: These results demonstrate that STS protects vascular relaxation against hyperlipidemia-induced damage through modulation of the SYK-NLRP3 inflammasome-MMP2/9 pathway. This research provides novel insights into the mechanisms underlying vascular relaxation impairment in a hyperlipidemic environment and uncovers a unique mechanism by which STS preserves vascular relaxation, offering valuable foundational research evidence for its clinical application in promoting vascular health.

A study on the application of radiomics based on cardiac MR non-enhanced cine sequence in the early diagnosis of hypertensive heart disease.

BACKGROUND: The prevalence of hypertensive heart disease (HHD) is high and there is currently no easy way to detect early HHD. Explore the application of radiomics using cardiac magnetic resonance (CMR) non-enhanced cine sequences in diagnosing HHD and latent cardiac changes caused by hypertension. METHODS: 132 patients who underwent CMR scanning were divided into groups: HHD (42), hypertension with normal cardiac structure and function (HWN) group (46), and normal control (NOR) group (44). Myocardial regions of the end-diastolic (ED) and end-systolic (ES) phases of the CMR short-axis cine sequence images were segmented into regions of interest (ROI). Three feature subsets (ED, ES, and ED combined with ES) were established after radiomic least absolute shrinkage and selection operator feature selection. Nine radiomic models were built using random forest (RF), support vector machine (SVM), and naive Bayes. Model performance was analyzed using receiver operating characteristic curves, and metrics like accuracy, area under the curve (AUC), precision, recall, and specificity. RESULTS: The feature subsets included first-order, shape, and texture features. SVM of ED combined with ES achieved the highest accuracy (0.833), with a macro-average AUC of 0.941. AUCs for HHD, HWN, and NOR identification were 0.967, 0.876, and 0.963, respectively. Precisions were 0.972, 0.740, and 0.826; recalls were 0.833, 0.804, and 0.863, respectively; and specificities were 0.989, 0.863, and 0.909, respectively. CONCLUSIONS: Radiomics technology using CMR non-enhanced cine sequences can detect early cardiac changes due to hypertension. It holds promise for future use in screening for latent cardiac damage in early HHD.

Development of Vogt-Koyanagi-Harada disease-like uveitis during treatment by anti-programmed death-1 antibody: a case report.

BACKGROUND: Several immune checkpoint inhibitors (ICIs) have been linked to the occurrence of Vogt-Koyanagi-Harada disease (VKHD)-like uveitis. Among the ICIs, there has been no report of immune-related adverse events (irAEs) caused by a new programmed death protein-1(PD-1) monoclonal antibody (Toripalimab). CASE PRESENTATION: This paper presents a case of VKHD-like uveitis that arose following Toripalimab therapy for urothelial cancer of the bladder, and the patient experienced symptoms 10 days after the final dosage of 20 months of medication treatment. This patient with bladder uroepithelial carcinoma had severe binocular acute panuveitis with exudative retinal detachment after receiving Toripalimab therapy. Binocular VKHD-like uveitis was suggested as a diagnosis. Both eyes recovered after discontinuing immune checkpoint inhibitors and local and systemic corticosteroid treatment. CONCLUSIONS: This report suggests that VKHD-like uveitis can also occur in patients receiving novel PD-1 antibodies and the importance of paying attention to eye complications in patients receiving treatment over a long period.

Federated Learning-Oriented Edge Computing Framework for the IIoT.

With the maturity of artificial intelligence (AI) technology, applications of AI in edge computing will greatly promote the development of industrial technology. However, the existing studies on the edge computing framework for the Industrial Internet of Things (IIoT) still face several challenges, such as deep hardware and software coupling, diverse protocols, difficult deployment of AI models, insufficient computing capabilities of edge devices, and sensitivity to delay and energy consumption. To solve the above problems, this paper proposes a software-defined AI-oriented three-layer IIoT edge computing framework and presents the design and implementation of an AI-oriented edge computing system, aiming to support device access, enable the acceptance and deployment of AI models from the cloud, and allow the whole process from data acquisition to model training to be completed at the edge. In addition, this paper proposes a time series-based method for device selection and computation offloading in the federated learning process, which selectively offloads the tasks of inefficient nodes to the edge computing center to reduce the training delay and energy consumption. Finally, experiments carried out to verify the feasibility and effectiveness of the proposed method are reported. The model training time with the proposed method is generally 30% to 50% less than that with the random device selection method, and the training energy consumption under the proposed method is generally 35% to 55% less.