This study uncovers dynamic immune adaptations during pregnancy through single-cell RNA sequencing of peripheral blood mononuclear cells at different pregnancy stages. A gradual reduction in cytotoxicity of T and NK cells, along with decreased MHC-II and CD40 signaling in T and B cells, suggests weakened adaptive immunity. Meanwhile, upregulated pro-inflammatory genes in monocytes may compensate for this reduction. Late pregnancy shows a transition toward immune activation in dendritic and CD4+ T cells. Notably, we highlight a novel pro-aging effect of pregnancy, which may reverse postpartum. These findings enhance our understanding of pregnancy immunity and its impact on disease risk.

Traditional terahertz (THz) polarization conversion devices show the unchanged polarization state on each output plane along the propagation path. This paper proposes a THz polarization controlled metasurface device that is dependent on propagation distance, which can continuously modify the polarization state of each output plane along the propagation path. The designed metasurface device can control the phase difference between the left-handed and right-handed circularly polarized components in the incident THz linearly polarized wave. This phase difference changes with the propagation distance, and ultimately outputs a linearly polarized wave that rotates along the propagation path. The polarization rotation angle can cover 0 to π. This device may have potential applications in fields such as variable matter excitation, THz communication, THz radar, THz sensing.

For the first time, researchers have revealed the high-affinity interaction between mitochondrial proteins HAX1 (intrinsically disordered) and CLPB (α-helical), uncovering their structural and functional synergy. Using NMR, the study shows HAX1's disordered region binds CLPB's helical domain with low micromolar affinity, providing mechanistic insights into diseases like cancer and mitochondrial encephalomyopathy. These findings open new avenues for targeted therapies against mitochondrial dysfunction-related disorders.

This work developed a new deep learning framework, MULGONET, to predict cancer recurrence and identify key biomarkers by integrating multi omics data (such as mRNA, DNA methylation, copy number variation). By utilizing the gene ontology (go) hierarchy, the model overcomes the challenges of data dimensionality and interpretability, and achieves higher accuracy in bladder cancer, pancreatic cancer, and gastric cancer datasets. This innovation enables clinicians to pinpoint key genes and biological pathways associated with cancer recurrence, paving the way for personalized treatment strategies.

A comprehensive review titled “From the Test Tube to the Cell: A Homecoming for DNA Computing Circuits?" published Mar. 4 in Intelligent Computing, a Science Partner Journal, highlights progress in operating DNA computing circuits for operation within living cells. According to the authors, dynamic nanodevices powered by DNA strand displacement reactions could soon enable real-time computing, sensing and actuation inside biological systems—and usher in a new era of “molecular robots” that interact with cellular environments.