The oncogenic fusion protein BCR-ABL is the driving force of leukemogenesis in chronic myeloid leukemia (CML). Despite great progress for CML treatment through application of tyrosine kinase inhibitors (TKIs) against BCR-ABL, long-term drug administration and clinical resistance continue to be an issue. Herein, we described the design, synthesis, and evaluation of novel proteolysis-targeting chimeric (PROTAC) small molecules targeting BCR-ABL which connect dasatinib and VHL E3 ubiquitin ligase ligand by extensive optimization of linkers. Our efforts have yielded SIAIS178 (19), which induces proper interaction between BCR-ABL and VHL ligase leading to effective degradation of BCR-ABL protein, achieves significant growth inhibition of BCR-ABL + leukemic cells in vitro, and induces substantial tumor regression against K562 xenograft tumors in vivo. In addition, SIAIS178 also degrades several clinically relevant resistance-conferring mutations.Our data indicate that SIAIS178 as efficacious BCR-ABL degrader warrants extensive further investigation for the treatment of BCR-ABL + leukemia.

A Ni/BPh catalyzed [2+2+2] cycloaddition of alkyne-nitriles with alkynes has been developed, which provides an efficient route to fused pyridines under mild reaction conditions. Mechanistic studies indicate that an azanickelacycle via heterocoupling of an alkyne with a nitrile moiety is possibly formed as a key reaction intermediate. The Lewis acid catalyst is crucial to the successful transformation, which is suggested to promote the oxidative cyclization process.

The rational design of PROTACs is difficult due to their obscure structure-activity relationship. This study introduces a deep neural network model - DeepPROTACs to help design potent PROTACs molecules. It can predict the degradation capacity of a proposed PROTAC molecule based on structures of given target protein and E3 ligase. The experimental dataset is mainly collected from PROTAC-DB and appropriately labeled according to the DC50 and Dmax values. In the model of DeepPROTACs, the ligands as well as the ligand binding pockets are generated and represented with graphs and fed into Graph Convolutional Networks for feature extraction. While SMILES representations of linkers are fed into a Bidirectional Long Short-Term Memory layer to generate the features. Experiments show that DeepPROTACs model achieves 77.95% average prediction accuracy and 0.8470 area under receiver operating characteristic curve on the test set. DeepPROTACs is available online at a web server (https://bailab.siais.shanghaitech.edu.cn/services/deepprotacs/) and at github (https://github.com/fenglei104/DeepPROTACs).