Disturbance of the cholinergic system plays a crucial role in the pathological progression of neurological diseases that cause dyskinesia-like behaviors. However, the molecular mechanisms underlying this disturbance remain elusive. Here, we showed that cyclin-dependent kinase 5 (Cdk5) was reduced in cholinergic neurons of midbrain according to the single-nucleus RNA sequencing analysis. Serum levels of CDK5 also decreased in patients with Parkinson’s disease accompanied by motor symptoms. Moreover, Cdk5 deficiency in cholinergic neurons triggered paw tremors, abnormal motor coordination, and motor balance deficits in mice. These symptoms occurred along with cholinergic neuron hyperexcitability and increases in the current density of large-conductance Ca²⁺-activated K⁺ channels (BK channels). Pharmacological inhibition of BK channels restrained the excessive intrinsic excitability of striatal cholinergic neurons in Cdk5-deficient mice. Furthermore, CDK5 interacted with BK channels and negatively regulated BK channel activity via phosphorylation of threonine-908. Restoration of CDK5 expression in striatal cholinergic neurons reduced dyskinesia-like behaviors in ChAT-Cre;Cdk5^f/f mice. Together, these findings indicate that CDK5-induced phosphorylation of BK channels involves in cholinergic-neuron-mediated motor function, providing a potential new therapeutic target for treating dyskinesia-like behaviors arising from neurological diseases.

Soft pneumatic actuators (SPAs) have attracted enormous attention in the growing field of robotics. Among different SPAs, composite reinforced actuators (CRAs) are widely used because of their simple structure and high controllability. However, multistep molding, a time-consuming method, is still the predominant fabrication method. Here, we propose a multimaterial embedded printing method (ME3P) to fabricate CRAs. In comparison with other 3-dimensional printing methods, our method improves fabrication flexibility greatly. Via the design and fabrication of the reinforced composites’ patterns and different geometries of the soft body, we demonstrate actuators with programmable responses (elongation, contraction, twisting, bending, and helical and omnidirectional bending). Finite element analysis is employed for the prediction of pneumatic responses and the inverse design of actuators based on specific actuation needs. Lastly, we use tube-crawling robots as a model system to demonstrate our ability to fabricate complex soft robots for practical applications. This work demonstrates the versatility of ME3P for the future manufacturing of CRA-based soft robots.

Seeking new strategies to tune the intrinsic defect and optimize the thermoelectric performance via no or less use of external doped elements (i.e., plain optimization) is an important method to realize the sustainable development of thermoelectric materials. Meanwhile, creating dislocation defects in oxide systems is quite challenging because the rigid and stiff ionic/covalent bonds can hardly tolerate the large strain energy associated with dislocations. Herein, taking BiCuSeO oxide as an example, the present work reports a successful construction of dense lattice dislocations in BiCuSeO by self-doping of Se at the O site (i.e., Se_O self-substitution), and achieves plain optimization of the thermoelectric properties with only external Pb doping. Owing to the self-substitution-induced large lattice distortion and the potential reinforcement effect by Pb doping, high-density (about 3.0 × 10¹⁴ m⁻²) dislocations form in the grains, which enhances the scattering strength of mid-frequency phonon and results in a substantial low lattice thermal conductivity of 0.38 W m⁻¹ K⁻¹ at 823 K in Pb-doped BiCuSeO. Meanwhile, Pb_Bi doping and Cu vacancy markedly improve the electrical conductivity while maintaining a competitively high Seebeck coefficient, thereby contributing to a highest power factor of 942 μW m⁻¹ K⁻². Finally, a remarkably enhanced zT value of 1.32 is obtained at 823 K in Bi_0.94Pb_0.06Cu_0.97Se_1.05O_0.95 with almost compositional plainification. The high-density dislocation structure reported in this work will also provide a good inspiration for the design and construction of dislocations in other oxide systems.

The COVID-19 pandemic caused by SARS-CoV-2 virus is an ongoing global health burden. Severe cases of COVID-19 and the rare cases of COVID-19 vaccine-induced-thrombotic-thrombocytopenia (VITT) are both associated with thrombosis and thrombocytopenia; however, the underlying mechanisms remain inadequately understood. Both infection and vaccination utilize the spike protein receptor-binding domain (RBD) of SARS-CoV-2. We found that intravenous injection of recombinant RBD caused significant platelet clearance in mice. Further investigation revealed the RBD could bind platelets, cause platelet activation, and potentiate platelet aggregation, which was exacerbated in the Delta and Kappa variants. The RBD–platelet interaction was partially dependent on the β3 integrin as binding was significantly reduced in β3^−/− mice. Furthermore, RBD binding to human and mouse platelets was significantly reduced with related αIIbβ3 antagonists and mutation of the RGD (arginine-glycine-aspartate) integrin binding motif to RGE (arginine-glycine-glutamate). We developed anti-RBD polyclonal and several monoclonal antibodies (mAbs) and identified 4F2 and 4H12 for their potent dual inhibition of RBD-induced platelet activation, aggregation, and clearance in vivo, and SARS-CoV-2 infection and replication in Vero E6 cells. Our data show that the RBD can bind platelets partially though αIIbβ3 and induce platelet activation and clearance, which may contribute to thrombosis and thrombocytopenia observed in COVID-19 and VITT. Our newly developed mAbs 4F2 and 4H12 have potential not only for diagnosis of SARS-CoV-2 virus antigen but also importantly for therapy against COVID-19.

Natural killer (NK) cells, as key immune cells, play essential roles in tumor cell immune escape and immunotherapy. Accumulating evidence has demonstrated that the gut microbiota community affects the efficacy of anti-PD1 immunotherapy and that remodeling the gut microbiota is a promising strategy to enhance anti-PD1 immunotherapy responsiveness in advanced melanoma patients; however, the details of the mechanism remain elusive. In this study, we found that Eubacterium rectale was significantly enriched in melanoma patients who responded to anti-PD1 immunotherapy and that a high E. rectale abundance was related to longer survival in melanoma patients. Furthermore, administration of E. rectale remarkably improved the efficacy of anti-PD1 therapy and increased the overall survival of tumor-bearing mice; moreover, application of E. rectale led to a significant accumulation of NK cells in the tumor microenvironment. Interestingly, conditioned medium isolated from an E. rectale culture system dramatically enhanced NK cell function. Gas chromatography–mass spectrometry/ultrahigh performance liquid chromatography–tandem mass spectrometry-based metabolomic analysis showed that l-serine production was significantly decreased in the E. rectale group; moreover, administration of an l-serine synthesis inhibitor dramatically increased NK cell activation, which enhanced anti-PD1 immunotherapy effects. Mechanistically, supplementation with l-serine or application of an l-serine synthesis inhibitor affected NK cell activation through Fos/Fosl. In summary, our findings reveal the role of bacteria-modulated serine metabolic signaling in NK cell activation and provide a novel therapeutic strategy to improve the efficacy of anti-PD1 immunotherapy in melanoma.

Microneedles (MNs) have drawn rising attention owing to their merits of convenience, noninvasiveness, flexible applicability, painless microchannels with boosted metabolism, and precisely tailored multifunction control. MNs can be modified to serve as novel transdermal drug delivery, which conventionally confront with the penetration barrier caused by skin stratum corneum. The micrometer-sized needles create channels through stratum corneum, enabling efficient drug delivery to the dermis for gratifying efficacy. Then, incorporating photosensitizer or photothermal agents into MNs can conduct photodynamic or photothermal therapy, respectively. Besides, health monitoring and medical detection by MN sensors can extract information from skin interstitial fluid and other biochemical/electronic signals. Here, this review discloses a novel monitoring, diagnostic, and therapeutic pattern by MNs, with elaborate discussion about the classified formation of MNs together with various applications and inherent mechanism. Hereby, multifunction development and outlook from biomedical/nanotechnology/photoelectric/devices/informatics to multidisciplinary applications are provided. Programmable intelligent MNs enable logic encoding of diverse monitoring and treatment pathways to extract signals, optimize the therapy efficacy, real-time monitoring, remote control, and drug screening, and take instant treatment.

We overview several properties—old and new—of training overparameterized deep networks under the square loss. We first consider a model of the dynamics of gradient flow under the square loss in deep homogeneous rectified linear unit networks. We study the convergence to a solution with the absolute minimum ρ, which is the product of the Frobenius norms of each layer weight matrix, when normalization by Lagrange multipliers is used together with weight decay under different forms of gradient descent. A main property of the minimizers that bound their expected error for a specific network architecture is ρ. In particular, we derive novel norm-based bounds for convolutional layers that are orders of magnitude better than classical bounds for dense networks. Next, we prove that quasi-interpolating solutions obtained by stochastic gradient descent in the presence of weight decay have a bias toward low-rank weight matrices, which should improve generalization. The same analysis predicts the existence of an inherent stochastic gradient descent noise for deep networks. In both cases, we verify our predictions experimentally. We then predict neural collapse and its properties without any specific assumption—unlike other published proofs. Our analysis supports the idea that the advantage of deep networks relative to other classifiers is greater for problems that are appropriate for sparse deep architectures such as convolutional neural networks. The reason is that compositionally sparse target functions can be approximated well by “sparse” deep networks without incurring in the curse of dimensionality.

Cancer treatments always face challenging problems, particularly drug resistance due to tumor cell heterogeneity. The existing datasets include the relationship between gene expression and drug sensitivities; however, the majority are based on tissue-level studies. Study drugs at the single-cell level are perspective to overcome minimal residual disease caused by subclonal resistant cancer cells retained after initial curative therapy. Fortunately, machine learning techniques can help us understand how different types of cells respond to different cancer drugs from the perspective of single-cell gene expression. Good modeling using single-cell data and drug response information will not only improve machine learning for cell–drug outcome prediction but also facilitate the discovery of drugs for specific cancer subgroups and specific cancer treatments. In this paper, we review machine learning and deep learning approaches in drug research. By analyzing the application of these methods on cancer cell lines and single-cell data and comparing the technical gap between single-cell sequencing data analysis and single-cell drug sensitivity analysis, we hope to explore the trends and potential of drug research at the single-cell data level and provide more inspiration for drug research at the single-cell level. We anticipate that this review will stimulate the innovative use of machine learning methods to address new challenges in precision medicine more broadly.

Host defense peptides (HDPs) are one of the potentially promising agents for infection diseases due to their broad spectrum and low resistance rate, but their clinical applications are limited by proteolytic instability, high-cost, and complicated synthesis process. Here, we report a host-defense-peptide-mimicking β-peptide polymer that resists proteolysis to have enhanced the activity under physiological conditions, excellent antimicrobial efficiency even at high density of bacteria, and low cost for preparation. The β-peptide polymer demonstrated quorum sensing (QS) interference and bactericidal effect against both bacterial communities and individual bacterium to simultaneously block bacterial communication and disrupt bacterial membranes. The hierarchical QS network was suppressed, and main QS signaling systems showed considerably down-regulated gene expression, resulting in excellent biofilm eradication and virulence reduction effects. The dual-modal antibacterial ability possessed excellent therapeutic effects in Pseudomonas aeruginosa pneumonia, which could inhibit biofilm formation and exhibit better antibacterial and anti-inflammatory efficiency than clinically used antibiotics, levofloxacin. Furthermore, the β-peptide polymer also showed excellent therapeutic effect Escherichia coli pyogenic liver abscess. Together, we believed that the β-peptide polymer had a feasible clinical potential to treat bacterial infection diseases.

As a naturally occurring cytolytic peptide, melittin (Mel) has strong cytolytic activity and is a potent therapeutic peptide for cancer therapy. However, the serious hemolytic activity of Mel largely impedes its clinical applications. In this work, based on the strong interactions between proteins/peptides and polyphenols, we develop a tannic acid–Fe³⁺ metal–phenolic network (MPN)-based strategy that can convert Mel from foe to friend via shielding its positive charges and reducing its hemolytic activity. Besides, an immune adjuvant resiquimod (R848) is also introduced for immunostimulation, affording the final Mel- and R848-coloaded nanodrug. The Mel-caused membrane disruption can induce immunogenic cell death for immunostimulation, R848 can act as an immune adjuvant to further facilitate the immunostimulatory effect, and the tannic acid–Fe³⁺ MPN-mediated Fenton reaction can produce reactive oxygen species for cancer treatment. Further experiments reveal that the nanodrug can effectively cause immunogenic cell death of tumor cells and arouse robust intratumoral and systemic antitumor immunostimulation. In the bilateral tumor-bearing mouse models, the nanodrug considerably destroys the primary tumor and also boosts the abscopal effect to ablate the distant tumor. Collectively, the MPN-facilitated “foe-to-friend” strategy may promote the practical applications of Mel and foster the development of cancer immunotherapeutics.