Disentangled self-attention neural network based on information sharing for click-through rate prediction.

With the exponential growth of network resources, recommendation systems have become successful at combating information overload. In intelligent recommendation systems, the prediction of click-through rates (CTR) plays a crucial role. Most CTR models employ a parallel network architecture to successfully capture explicit and implicit feature interactions. However, the existing models ignore two aspects. One limitation observed in most models is that they focus only on the interaction of paired term features, with no emphasis on modeling unary terms. The second issue is that most models input characteristics indiscriminately into parallel networks, resulting in network input oversharing. We propose a disentangled self-attention neural network based on information sharing (DSAN) for CTR prediction to simulate complex feature interactions. Firstly, an embedding layer transforms high-dimensional sparse features into low-dimensional dense matrices. Then, the disentangled multi-head self-attention learns the relationship between different features and is fed into a parallel network architecture. Finally, we set up a shared interaction layer to solve the problem of insufficient information sharing in parallel networks. Results from experiments conducted on two real-world datasets demonstrate that our proposed method surpasses existing methods in predictive accuracy.

Synthesis of spiro[indolenine]-methanofullerenes via Deoxofluor promoted deoxygenative cyclopropanation of 1,2-(3-indole)-fullerenols.

Deoxofluor-promoted intramolecular cyclopropanation of 1,2-(3-indole)fullerenols has been developed as a straightforward and efficient protocol for the synthesis of various spiro[indolenine]-methanofullerenes. This approach exhibits low cost, operational simplicity, and convenient conditions, and thus has potential application value.

Structural insight into H4K20 methylation on H2A.Z-nucleosome by SUV420H1.

DNA replication ensures the accurate transmission of genetic information during the cell cycle. Histone variant H2A.Z is crucial for early replication origins licensing and activation in which SUV420H1 preferentially recognizes H2A.Z-nucleosome and deposits H4 lysine 20 dimethylation (H4K20me2) on replication origins. Here, we report the cryo-EM structures of SUV420H1 bound to H2A.Z-nucleosome or H2A-nucleosome and demonstrate that SUV420H1 directly interacts with H4 N-terminal tail, the DNA, and the acidic patch in the nucleosome. The H4 (1-24) forms a lasso-shaped structure that stabilizes the SUV420H1-nucleosome complex and precisely projects the H4K20 residue into the SUV420H1 catalytic center. In vitro and in vivo analyses reveal a crucial role of the SUV420H1 KR loop (residues 214-223), which lies close to the H2A.Z-specific residues D97/S98, in H2A.Z-nucleosome preferential recognition. Together, our findings elucidate how SUV420H1 recognizes nucleosomes to ensure site-specific H4K20me2 modification and provide insights into how SUV420H1 preferentially recognizes H2A.Z nucleosome.

Structural insight into BRCA1-BARD1 complex recruitment to damaged chromatin.

The BRCA1-BARD1 complex directs the DNA double-strand break (DSB) repair pathway choice to error-free homologous recombination (HR) during the S-G2 stages. Targeting BRCA1-BARD1 to DSB-proximal sites requires BARD1-mediated nucleosome interaction and histone mark recognition. Here, we report the cryo-EM structure of BARD1 bound to a ubiquitinated nucleosome core particle (NCPUb) at 3.1 Å resolution and illustrate how BARD1 simultaneously recognizes the DNA damage-induced mark H2AK15ub and DNA replication-associated mark H4K20me0 on the nucleosome. In vitro and in vivo analyses reveal that the BARD1-NCPUb complex is stabilized by BARD1-nucleosome interaction, BARD1-ubiquitin interaction, and BARD1 ARD domain-BARD1 BRCT domain interaction, and abrogating these interactions is detrimental to HR activity. We further identify multiple disease-causing BARD1 mutations that disrupt BARD1-NCPUb interactions and hence impair HR. Together, this study elucidates the mechanism of BRCA1-BARD1 complex recruitment and retention by DSB-flanking nucleosomes and sheds important light on cancer therapeutic avenues.

Structural basis for shieldin complex subunit 3-mediated recruitment of the checkpoint protein REV7 during DNA double-strand break repair.

Shieldin complex subunit 3 (SHLD3) is the apical subunit of a recently-identified shieldin complex and plays a critical role in DNA double-strand break repair. To fulfill its function in DNA repair, SHLD3 interacts with the mitotic spindle assembly checkpoint protein REV7 homolog (REV7), but the details of this interaction remain obscure. Here, we present the crystal structures of REV7 in complex with SHLD3's REV7-binding domain (RBD) at 2.2-2.3 Å resolutions. The structures revealed that the ladle-shaped RBD in SHLD3 uses its N-terminal loop and C-terminal α-helix (αC-helix) in its interaction with REV7. The N-terminal loop exhibited a structure similar to those previously identified in other REV7-binding proteins, and the less-conserved αC-helix region adopted a distinct mode for binding REV7. In vitro and in vivo binding analyses revealed that the N-terminal loop and the αC-helix are both indispensable for high-affinity REV7 binding (with low-nanomolar affinity), underscoring the crucial role of SHLD3 αC-helix in protein binding. Moreover, binding kinetics analyses revealed that the REV7 "safety belt" region, which plays a role in binding other proteins, is essential for SHLD3-REV7 binding, as this region retards the dissociation of the RBD from the bound REV7. Together, the findings of our study reveal the molecular basis of the SHLD3-REV7 interaction and provide critical insights into how SHLD3 recognizes REV7.