Synthetic Data Enhancement and Network Compression Technology of Monocular Depth Estimation for Real-Time Autonomous Driving System.

Accurate 3D image recognition, critical for autonomous driving safety, is shifting from the LIDAR-based point cloud to camera-based depth estimation technologies driven by cost considerations and the point cloud's limitations in detecting distant small objects. This research aims to enhance MDE (Monocular Depth Estimation) using a single camera, offering extreme cost-effectiveness in acquiring 3D environmental data. In particular, this paper focuses on novel data augmentation methods designed to enhance the accuracy of MDE. Our research addresses the challenge of limited MDE data quantities by proposing the use of synthetic-based augmentation techniques: Mask, Mask-Scale, and CutFlip. The implementation of these synthetic-based data augmentation strategies has demonstrably enhanced the accuracy of MDE models by 4.0% compared to the original dataset. Furthermore, this study introduces the RMS (Real-time Monocular Depth Estimation configuration considering Resolution, Efficiency, and Latency) algorithm, designed for the optimization of neural networks to augment the performance of contemporary monocular depth estimation technologies through a three-step process. Initially, it selects a model based on minimum latency and REL criteria, followed by refining the model's accuracy using various data augmentation techniques and loss functions. Finally, the refined model is compressed using quantization and pruning techniques to minimize its size for efficient on-device real-time applications. Experimental results from implementing the RMS algorithm indicated that, within the required latency and size constraints, the IEBins model exhibited the most accurate REL (absolute RELative error) performance, achieving a 0.0480 REL. Furthermore, the data augmentation combination of the original dataset with Flip, Mask, and CutFlip, alongside the SigLoss loss function, displayed the best REL performance, with a score of 0.0461. The network compression technique using FP16 was analyzed as the most effective, reducing the model size by 83.4% compared to the original while maintaining the least impact on REL performance and latency. Finally, the performance of the RMS algorithm was validated on the on-device autonomous driving platform, NVIDIA Jetson AGX Orin, through which optimal deployment strategies were derived for various applications and scenarios requiring autonomous driving technologies.

Targeted protein degradation systems to enhance Wnt signaling.

Molecules that facilitate targeted protein degradation (TPD) offer great promise as novel therapeutics. The human hepatic lectin asialoglycoprotein receptor (ASGR) is selectively expressed on hepatocytes. We have previously engineered an anti-ASGR1 antibody-mutant RSPO2 (RSPO2RA) fusion protein (called SWEETS) to drive tissue-specific degradation of ZNRF3/RNF43 E3 ubiquitin ligases, which achieved hepatocyte-specific enhanced Wnt signaling, proliferation, and restored liver function in mouse models, and an antibody-RSPO2RA fusion molecule is currently in human clinical trials. In the current study, we identified two new ASGR1- and ASGR1/2-specific antibodies, 8M24 and 8G8. High-resolution crystal structures of ASGR1:8M24 and ASGR2:8G8 complexes revealed that these antibodies bind to distinct epitopes on opposing sides of ASGR, away from the substrate-binding site. Both antibodies enhanced Wnt activity when assembled as SWEETS molecules with RSPO2RA through specific effects sequestering E3 ligases. In addition, 8M24-RSPO2RA and 8G8-RSPO2RA efficiently downregulate ASGR1 through TPD mechanisms. These results demonstrate the possibility of combining different therapeutic effects and degradation mechanisms in a single molecule.

Design principles and therapeutic applications of novel synthetic WNT signaling agonists.

Wingless-related integration site or Wingless and Int-1 or Wingless-Int (WNT) signaling is crucial for embryonic development, and adult tissue homeostasis and regeneration, through its essential roles in cell fate, patterning, and stem cell regulation. The biophysical characteristics of WNT ligands have hindered efforts to interrogate ligand activity in vivo and prevented their development as therapeutics. Recent breakthroughs have enabled the generation of synthetic WNT signaling molecules that possess characteristics of natural ligands and potently activate the pathway, while also providing distinct advantages for therapeutic development and manufacturing. This review provides a detailed discussion of the protein engineering of these molecular platforms for WNT signaling agonism. We discuss the importance of WNT signaling in several organs and share insights from the initial application of these new classes of molecules in vitro and in vivo. These molecules offer a unique opportunity to enhance our understanding of how WNT signaling agonism promotes tissue repair, enabling targeted development of tailored therapeutics.

Correction: Mohd Azmi et al. Application of Plant Proteases in Meat Tenderization: Recent Trends and Future Prospects. Foods 2023, 12, 1336.

In the original publication [...].

Effects of Fc glycosylation on the activity of WNT mimetic agonistic antibodies.

Monoclonal antibodies have been explored in a broad range of applications including receptor agonism. Given the importance of receptor conformation in signaling, the agonistic activity of antibodies that engage these receptors are influenced by many parameters. Tetravalent bispecific antibodies that target the frizzled and lipoprotein receptor-related protein receptors and subsequently activate WNT ("Wingless-related integration site" or "Wingless and Int-1" or "Wingless-Int") signaling have been constructed. Because WNT activation stimulates stem cell proliferation and tissue regeneration, immune effector functions should be eliminated from therapeutic antibodies targeting this pathway. Here, we report an unexpected effect of Fc glycosylation on the agonistic activity of WNT mimetic antibodies. Our findings underscore the importance of antibody format, geometry and epitope in agonistic antibody design, and highlight the need to establish appropriate early discovery screening strategies to identify hits for further optimization.

A new exceptionally well-preserved basal actinopterygian fish in the juvenile stage from the Upper Triassic Amisan Formation of South Korea.

The study of the large paraphyletic group of extinct 'palaeoniscoid' fishes has shed light on the diversity and evolutionary history of basal actinopterygians. However, only a little ontogenetic information about 'palaeoniscoids' is known because their records in the early stages of development are scarce. Here, we report on a growth series of 'palaeoniscoids' in the juvenile stage from the Upper Triassic Amisan Formation of South Korea. Fourteen specimens, including five counterpart specimens, represent a new taxon, Megalomatia minima gen. et sp. nov., exhibiting ontogeny and exceptional preservation with the eyes possibly containing the crystalline lens, the otoliths, and the lateral line canals without covering scales. This discovery allows us to discuss the adaptations and evolution of basal actinopterygians in more detail than before. The otoliths in situ of Megalomatia support the previous interpretation that basal actinopterygians have a sagitta as the largest otolith. The trunk lateral line canal, which runs under the scales instead of passing through them, represents a plesiomorphic gnathostome trait. Notably, the large protruded eyes suggest that Megalomatia probably has binocular vision, which would have played a significant role in targeting and catching prey with the primitive jaw structure. In addition, the firstly formed skeletal elements such as the jaws, pectoral girdle, and opercular series, and the posteroanterior pattern of squamation development are likely linked to the adaptation of young individuals to increase their viability for feeding, respiration, and swimming.

Photoprotective effects of Lithospermum erythrorhizon and Pueraria lobata extracts on UVB-induced photoaging: A study on skin barrier protection.

AIM: Lithospermum erythrorhizon and Pueraria lobata exhibit promising potential as cosmetic additives for mitigating skin barrier impairment induced by photoaging. Despite their potential, the precise mechanisms underlying their protective and ameliorative effects remain elusive. This study sought to assess the reparative properties of Lithospermum erythrorhizon and Pueraria lobata extracts (LP) on UVB-irradiated human skin keratinocytes (HaCaT cells) and explore the therapeutic potential of LP as a skin barrier protection agent. MATERIALS AND METHODS: Antioxidant activities were gauged through 2,2-diphenyl-1-picrylhydrazyl (DPPH), 2,2'-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS), and reactive oxygen species (ROS) assays. The expression levels of skin barrier-related markers, encompassing metalloproteinases (MMPs) and hyaluronidase (HYAL) were scrutinized using enzyme-linked immunosorbent assay (ELISA), reverse transcriptase (RT)-PCR, and Western blotting, with a particular focus on the involvement of the transforming growth factor (TGF)-ß/Smad and nuclear factor-κB (NF-κB) signaling pathways. RESULTS: The study revealed that LP effectively scavenges free radicals, diminishes ROS production in a dose-dependent manner, and significantly attenuates UVB-induced expression of MMP-1 and MMP-3 through modulation of the hyaluronan synthase (HAS)2/HYAL1 signaling axis in UVB-irradiated HaCaT cells. Additionally, LP demonstrated enhanced TGF-ß signaling activation, fostering procollagen type I synthesis, and concurrently exhibited mitogen-activated protein kinases (MAPK)/NF-κB signaling inactivation, thereby mitigating pro-inflammatory cytokine release and alleviating UVB-induced cellular damage. CONCLUSION: In conclusion, the observed protective effects of LP on skin cellular constituents highlight its substantial biological potential for shielding against UVB-induced skin photoaging, positioning it as a promising candidate for both pharmaceutical and cosmetic applications.

BRAIDing receptors for cell-specific targeting.

Systemic toxicity is a major challenge in the development of therapeutics. Consequently, cell-type-specific targeting is needed to improve on-target efficacy while reducing off-target toxicity. Here, we describe a cell-targeting system we have termed BRAID (BRidged Activation by Intra/intermolecular Division) whereby an active molecule is divided into two inactive or less active parts that are subsequently brought together via a so-called 'bridging receptor' on the target cell. This concept was validated using the WNT/ß-catenin signaling system, demonstrating that a multivalent WNT agonist molecule divided into two inactive components assembled from different epitopes via the hepatocyte receptor ßKlotho induces signaling specifically on hepatocytes. These data provide proof of concept for this cell-specific targeting strategy, and in principle, this may also allow activation of multiple signaling pathways where desirable. This approach has broad application potential for other receptor systems.

Development of a neutralization monoclonal antibody with a broad neutralizing effect against SARS-CoV-2 variants.

BACKGROUND: The emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants has challenged the effectiveness of current therapeutic regimens. Here, we aimed to develop a potent SARS-CoV-2 antibody with broad neutralizing effect by screening a scFv library with the spike protein receptor-binding domain (RBD) via phage display. METHODS: SKAI-DS84 was identified through phage display, and we performed pseudovirus neutralization assays, authentic virus neutralization assays, and in vivo neutralization efficacy evaluations. Furthermore, surface plasmon resonance (SPR) analysis was conducted to assess the physical characteristics of the antibody, including binding kinetics and measure its affinity for variant RBDs. RESULTS: The selected clones were converted to human IgG, and among them, SKAI-DS84 was selected for further analyses based on its binding affinity with the variant RBDs. Using pseudoviruses, we confirmed that SKAI-DS84 was strongly neutralizing against wild-type, B.1.617.2, B.1.1.529, and subvariants of SARS-CoV-2. We also tested the neutralizing effect of SKAI-DS84 on authentic viruses, in vivo and observed a reduction in viral replication and improved lung pathology. We performed binding and epitope mapping experiments to understand the mechanisms underlying neutralization and identified quaternary epitopes formed by the interaction between RBDs as the target of SKAI-DS84. CONCLUSIONS: We identified, produced, and tested the neutralizing effect of SKAI-DS84 antibody. Our results highlight that SKAI-DS84 could be a potential neutralizing antibody against SARS-CoV-2 and its variants.

Optimal Configuration of Multi-Task Learning for Autonomous Driving.

For autonomous driving, it is imperative to perform various high-computation image recognition tasks with high accuracy, utilizing diverse sensors to perceive the surrounding environment. Specifically, cameras are used to perform lane detection, object detection, and segmentation, and, in the absence of lidar, tasks extend to inferring 3D information through depth estimation, 3D object detection, 3D reconstruction, and SLAM. However, accurately processing all these image recognition operations in real-time for autonomous driving under constrained hardware conditions is practically unfeasible. In this study, considering the characteristics of image recognition tasks performed by these sensors and the given hardware conditions, we investigated MTL (multi-task learning), which enables parallel execution of various image recognition tasks to maximize their processing speed, accuracy, and memory efficiency. Particularly, this study analyzes the combinations of image recognition tasks for autonomous driving and proposes the MDO (multi-task decision and optimization) algorithm, consisting of three steps, as a means for optimization. In the initial step, a MTS (multi-task set) is selected to minimize overall latency while meeting minimum accuracy requirements. Subsequently, additional training of the shared backbone and individual subnets is conducted to enhance accuracy with the predefined MTS. Finally, both the shared backbone and each subnet undergo compression while maintaining the already secured accuracy and latency performance. The experimental results indicate that integrated accuracy performance is critically important in the configuration and optimization of MTL, and this integrated accuracy is determined by the ITC (inter-task correlation). The MDO algorithm was designed to consider these characteristics and construct multi-task sets with tasks that exhibit high ITC. Furthermore, the implementation of the proposed MDO algorithm, coupled with additional SSL (semi-supervised learning) based training, resulted in a significant performance enhancement. This advancement manifested as approximately a 12% increase in object detection mAP performance, a 15% improvement in lane detection accuracy, and a 27% reduction in latency, surpassing the results of previous three-task learning techniques like YOLOP and HybridNet.

Grasping through dynamic weaving with entangled closed loops.

Pick-and-place is essential in diverse robotic applications for industries including manufacturing, and assembly. Soft grippers offer a cost-effective, and low-maintenance alternative for secure object grasping without complex sensing and control systems. However, their inherent softness normally limits payload capabilities and robustness to external disturbances, constraining their applications and hindering reliable performance. In this study, we propose a weaving-inspired grasping mechanism that substantially increases payload capacity while maintaining the use of soft and flexible materials. Drawing from weaving principles, we designed a flexible continuum structure featuring multiple closed-loop strips and employing a kirigami-inspired approach to enable the instantaneous and reversible creation of a woven configuration. The mechanical stability of the woven configuration offers exceptional loading capacity, while the softness of the gripper material ensures safe and adaptive interactions with objects. Experimental results show that the 130 g·f gripper can support up to 100 kg·f. Outperforming competitors in similar weight and softness domains, this breakthrough, enabled by the weaving principle, will broaden the scope of gripper applications to previously inaccessible or barely accessible fields, such as agriculture and logistics.

Novel Frizzled-specific antibody-based Wnt mimetics and Wnt superagonists selectively activate WNT/ß-catenin signaling in target tissues.

WNTs are essential factors for stem cell biology, embryonic development, and for maintaining homeostasis and tissue repair in adults. Difficulties in purifying WNTs and their lack of receptor selectivity have hampered research and regenerative medicine development. While breakthroughs in WNT mimetic development have overcome some of these difficulties, the tools developed so far are incomplete and mimetics alone are often not sufficient. Here, we developed a complete set of WNT mimetic molecules that cover all WNT/ß-catenin-activating Frizzleds (FZDs). We show that FZD1,2,7 stimulate salivary gland expansion in vivo and salivary gland organoid expansion. We further describe the discovery of a novel WNT-modulating platform that combines WNT and RSPO mimetics' effects into one molecule. This set of molecules supports better organoid expansion in various tissues. These WNT-activating platforms can be broadly applied to organoids, pluripotent stem cells, and in vivo research, and serve as bases for future therapeutic development.

Therapeutic blood-brain barrier modulation and stroke treatment by a bioengineered FZD4-selective WNT surrogate in mice.

Derangements of the blood-brain barrier (BBB) or blood-retinal barrier (BRB) occur in disorders ranging from stroke, cancer, diabetic retinopathy, and Alzheimer's disease. The Norrin/FZD4/TSPAN12 pathway activates WNT/ß-catenin signaling, which is essential for BBB and BRB function. However, systemic pharmacologic FZD4 stimulation is hindered by obligate palmitoylation and insolubility of native WNTs and suboptimal properties of the FZD4-selective ligand Norrin. Here, we develop L6-F4-2, a non-lipidated, FZD4-specific surrogate which significantly improves subpicomolar affinity versus native Norrin. In Norrin knockout (NdpKO) mice, L6-F4-2 not only potently reverses neonatal retinal angiogenesis deficits, but also restores BRB and BBB function. In adult C57Bl/6J mice, post-stroke systemic delivery of L6-F4-2 strongly reduces BBB permeability, infarction, and edema, while improving neurologic score and capillary pericyte coverage. Our findings reveal systemic efficacy of a bioengineered FZD4-selective WNT surrogate during ischemic BBB dysfunction, with potential applicability to adult CNS disorders characterized by an aberrant blood-brain barrier.

DNA lattice growth with single, double, and triple double-crossover boundaries by stepwise self-assembly.

Construction of various nanostructures with nanometre-scale precision through various DNA building blocks depends upon self-assembly, base-pair complementarity and sequence programmability. During annealing, unit tiles are formed by the complementarity of base pairs in each strand. Enhancement of growth of target lattices is expected if seed lattices (i.e. boundaries for growth of target lattices) are initially present in a test tube during annealing. Although most processes for annealing DNA nanostructures adopt a one-step high temperature method, multi-step annealing provides certain advantages such as reusability of unit tiles and tuneability of lattice formation. We can construct target lattices effectively (through multi-step annealing) and efficiently (via boundaries) by multi-step annealing and combining boundaries. Here, we construct efficient boundaries made of single, double, and triple double-crossover DNA tiles for growth of DNA lattices. Two unit double-crossover DNA tile-based lattices and copy-logic implemented algorithmic lattices were introduced to test the growth of target lattices on boundaries. We used multi-step annealing to tune the formation of DNA crystals during fabrication of DNA crystals comprised of boundaries and target lattices. The formation of target DNA lattices was visualized using atomic force microscopy (AFM). The borders between boundaries and lattices in a single crystal were clearly differentiable from AFM images. Our method provides way to construct various types of lattices in a single crystal, which might generate various patterns and enhance the information capacity in a given crystal.

Application of Plant Proteases in Meat Tenderization: Recent Trends and Future Prospects.

Papain, bromelain, and ficin are commonly used plant proteases used for meat tenderization. Other plant proteases explored for meat tenderization are actinidin, zingibain, and cucumin. The application of plant crude extracts or powders containing higher levels of compounds exerting tenderizing effects is also gaining popularity due to lower cost, improved sensory attributes of meat, and the presence of bioactive compounds exerting additional benefits in addition to tenderization, such as antioxidants and antimicrobial effects. The uncontrolled plant protease action could cause excessive tenderization (mushy texture) and poor quality due to an indiscriminate breakdown of proteins. The higher cost of separation and the purification of enzymes, unstable structure, and poor stability of these enzymes due to autolysis are some major challenges faced by the food industry. The meat industry is targeting the recycling of enzymes and improving their stability and shelf-life by immobilization, encapsulation, protein engineering, medium engineering, and stabilization during tenderization. The present review critically analyzed recent trends and the prospects of the application of plant proteases in meat tenderization.

Neurexin-2 restricts synapse numbers and restrains the presynaptic release probability by an alternative splicing-dependent mechanism.

α- and ß-neurexins are extensively alternatively spliced, presynaptic cell-adhesion molecules that are thought to organize synapse assembly. However, recent data revealed that, in the hippocampus in vivo, the deletion of one neurexin isoform, Nrxn2, surprisingly increased excitatory synapse numbers and enhanced their presynaptic release probability, suggesting that Nrxn2 restricts, instead of enabling, synapse assembly. To delineate the synaptic function and mechanism of action of Nrxn2, we examined cultured hippocampal neurons as a reduced system. In heterologous synapse formation assays, different alternatively spliced Nrxn2ß isoforms robustly promoted synapse assembly similar to Nrxn1ß and Nrxn3ß, consistent with a general synaptogenic function of neurexins. Deletion of Nrxn2 from cultured hippocampal neurons, however, caused a significant increase in synapse density and release probability, replicating the in vivo data that suggested a synapse-restricting function. Rescue experiments revealed that two of the four Nrxn2ß splice variants (Nrxn2ß-SS4+/SS5- and Nrxn2ß-SS4+/SS5+) reversed the increase in synapse density in Nrxn2-deficient neurons, whereas only one of the four Nrxn2ß splice variants (Nrxn2ß-SS4+/SS5+) normalized the increase in release probability in Nrxn2-deficient neurons. Thus, a subset of Nrxn2 splice variants restricts synapse numbers and restrains their release probability in cultured neurons.

Rethinking Breast Cancer Diagnosis through Deep Learning Based Image Recognition.

This paper explored techniques for diagnosing breast cancer using deep learning based medical image recognition. X-ray (Mammography) images, ultrasound images, and histopathology images are used to improve the accuracy of the process by diagnosing breast cancer classification and by inferring their affected location. For this goal, the image recognition application strategies for the maximal diagnosis accuracy in each medical image data are investigated in terms of various image classification (VGGNet19, ResNet50, DenseNet121, EfficietNet v2), image segmentation (UNet, ResUNet++, DeepLab v3), and related loss functions (binary cross entropy, dice Loss, Tversky loss), and data augmentation. As a result of evaluations through the presented methods, when using filter-based data augmentation, ResNet50 showed the best performance in image classification, and UNet showed the best performance in both X-ray image and ultrasound image as image segmentation. When applying the proposed image recognition strategies for the maximal diagnosis accuracy in each medical image data, the accuracy can be improved by 33.3% in image segmentation in X-ray images, 29.9% in image segmentation in ultrasound images, and 22.8% in image classification in histopathology images.

A non-avian dinosaur with a streamlined body exhibits potential adaptations for swimming.

Streamlining a body is a major adaptation for aquatic animals to move efficiently in the water. Whereas diving birds are well known to have streamlined bodies, such body shapes have not been documented in non-avian dinosaurs. It is primarily because most known non-avian theropods are terrestrial, barring a few exceptions. However, clear evidence of streamlined bodies is absent even in the purported semiaquatic groups. Here we report a new theropod, Natovenator polydontus gen. et sp. nov., from the Upper Cretaceous of Mongolia. The new specimen includes a well-preserved skeleton with several articulated dorsal ribs that are posterolaterally oriented to streamline the body as in diving birds. Additionally, the widely arched proximal rib shafts reflect a dorsoventrally compressed ribcage like aquatic reptiles. Its body shape suggests that Natovenator was a potentially capable swimming predator, and the streamlined body evolved independently in separate lineages of theropod dinosaurs.

Nipah and Hendra Viruses: Deadly Zoonotic Paramyxoviruses with the Potential to Cause the Next Pandemic.

Nipah and Hendra viruses are deadly zoonotic paramyxoviruses with a case fatality rate of upto 75%. The viruses belong to the genus henipavirus in the family Paramyxoviridae, a family of negative-sense single-stranded RNA viruses. The natural reservoirs of NiV and HeV are bats (flying foxes) in which the virus infection is asymptomatic. The intermediate hosts for NiV and HeV are swine and equine, respectively. In humans, NiV infections result in severe and often fatal respiratory and neurological manifestations. The Nipah virus was first identified in Malaysia and Singapore following an outbreak of encephalitis in pig farmers and subsequent outbreaks have been reported in Bangladesh and India almost every year. Due to its extreme pathogenicity, pandemic potential, and lack of established antiviral therapeutics and vaccines, research on henipaviruses is highly warranted so as to develop antivirals or vaccines that could aid in the prevention and control of future outbreaks.

Selective Activation of the Wnt-Signaling Pathway as a Novel Therapy for the Treatment of Diabetic Retinopathy and Other Retinal Vascular Diseases.

Retinal ischemia, often associated with various disorders such as diabetic retinopathy (DR), retinal vein occlusion, glaucoma, optic neuropathies, stroke, and other retinopathies, is a major cause of visual impairment and blindness worldwide. As proper blood supply to the retina is critical to maintain its high metabolic demand, any impediment to blood flow can lead to a decrease in oxygen supply, resulting in retinal ischemia. In the pathogenesis of DR, including diabetic macular edema (DME), elevated blood glucose leads to blood-retina barrier (BRB) disruptions, vascular leakage, and capillary occlusion and dropouts, causing insufficient delivery of oxygen to the retina, and ultimately resulting in visual impairment. Other potential causes of DR include neuronal dysfunction in the absence of vascular defect, genetic, and environmental factors. The exact disease progression remains unclear and varies from patient to patient. Vascular leakage leading to edema clearly links to visual impairment and remains an important target for therapy. Despite recent advances in the treatment of DME and DR with anti-VEGFs, effective therapies with new mechanisms of action to address current treatment limitations regarding vessel regeneration and reperfusion of ischemic retinal areas are still needed. The Wnt signaling pathway plays a critical role in proper vascular development and maintenance in the retina, and thus provides a novel therapeutic approach for the treatment of diabetic and other retinopathies. In this review, we summarize the potential of this pathway to address treatment gaps with current therapies, its promise as a novel and potentially disease modifying therapy for patients with DR and opportunities in other retinal vascular diseases.