Antileukemic effects of topoisomerase I inhibitors mediated by de-SUMOylase SENP1.

SUMO-specific proteases (SENPs) play pivotal roles in maintaining the balance of SUMOylation/de-SUMOylation and in SUMO recycling. Deregulation of SENPs leads to cellular dysfunction and corresponding diseases. As a key member of the SENP family, SENP1 is highly correlated with various cancers. However, the potential role of SENP1 in leukemia, especially in acute lymphoblastic leukemia (ALL), is not clear. This study shows that ALL cells knocking down SENP1 display compromised growth rather than significant alterations in chemosensitivity, although ALL relapse samples have a relatively higher expression of SENP1 than the paired diagnosis samples. Camptothecin derivatives 7-ethylcamptothecin (7E-CPT, a monomer compound) and topotecan (TPT, an approved clinical drug) induce specific SENP1 reduction and severe apoptosis of ALL cells, showing strong anticancer effects against ALL. Conversely, SENP1 could attenuate this inhibitory effect by targeting DNA topoisomerase I (TOP1) for de-SUMOylation, indicating that specific reduction in SENP1 induced by 7E-CPT and/or topotecan inhibits the proliferation of ALL cells.

A γ-glutamyl hydrolase lacking the signal peptide confers susceptibility to folates/antifolates in acute lymphoblastic leukemia cells.

A key cofactor of several enzymes implicated in DNA synthesis, repair, and methylation, folate has been shown to be required for normal cell growth and replication and is the basis for cancer chemotherapy using antifolates. γ-Glutamyl hydrolase (GGH) catalyzes the removal of γ-polyglutamate tails of folylpoly-/antifolylpoly-γ-glutamates to facilitate their export out of the cell, thereby maintaining metabolic homeostasis of folates or pharmacological efficacy of antifolates. However, the factors that control or modulate GGH function are not well understood. In this study, we show that intact GGH is not indispensable for the chemosensitivity and growth of acute lymphoblastic leukemia (ALL) cells, whereas GGH lacking N-terminal signal peptide (GGH-ΔN ) confers the significant drug resistance of ALL cells to the antifolates MTX and RTX. In addition, ALL cells harboring GGH-ΔN show high susceptibility to the change in folates, and glycosylation is not responsible for these phenotypes elicited by GGH-ΔN . Mechanistically, the loss of signal peptide enhances intracellular retention of GGH and its lysosomal disposition. Our findings clearly define the in vivo role of GGH in ALL cells and indicate a novel modulation of the GGH function, suggesting new avenues for ALL treatment in future.

Knowledge-driven feature component interpretable network for motor imagery classification.

Objective. The end-to-end convolutional neural network (CNN) has achieved great success in motor imagery (MI) classification without a manual feature design. However, all the existing deep network solutions are purely datadriven and lack interpretability, which makes it impossible to discover insightful knowledge from the learned features, not to mention to design specific network structures. The heavy computational cost of CNN also makes it challenging for real-time application along with high classification performance.Approach. To address these problems, a novel knowledge-driven feature component interpretable network (KFCNet) is proposed, which combines spatial and temporal convolution in analogy to independent component analysis and a power spectrum pipeline. Prior frequency band knowledge of sensory-motor rhythms has been formulated as band-pass linear-phase digital finite impulse response filters to initialize the temporal convolution kernels to enable the knowledge-driven mechanism. To avoid signal distortion and achieve a linear phase and unimodality of filters, a symmetry loss is proposed, which is used in combination with the cross-entropy classification loss for training. Besides the general prior knowledge, subject-specific time-frequency property of event-related desynchronization and synchronization has been employed to construct and initialize the network with significantly fewer parameters.Main results.Comparison of experiments on two public datasets has been performed. Interpretable feature components could be observed in the trained model. The physically meaningful observation could efficiently assist the design of the network structure. Excellent classification performance on MI has been obtained.Significance. The performance of KFCNet is comparable to the state-of-the-art methods but with much fewer parameters and makes real-time applications possible.