Hua Wu

Recent advances in text-to-video (T2V) generation with diffusion models have garnered significant attention. How ever, they typically perform well in scenes with a single object and motion, struggling in compositional scenarios with multiple objects and distinct motions to accurately reflect the semantic content of text prompts. To address these challenges, we propose StarVid, a plug-and-play, training-free method that improves semantic alignment between multiple subjects, their motions, and text prompts in T2V models. StarVid first employs large language models (LLMs) to perform two-stage motion trajectory planning based on the input prompt, providing object-level spatial priors that guide the generation process. These priors are then used to impose a spatial-aware loss that refines cross-attention (CA) maps, encouraging attention to be focused on distinct regions corresponding to different subjects. Furthermore, a syntax guided contrastive constraint is introduced to explicitly enhance the association between verbs and their corresponding nouns in the CA maps, thereby improving motion-subject binding. Both qualitative and quantitative evaluations demonstrate that the proposed framework significantly outperforms baseline methods, delivering videos of higher quality with improved semantic consistency.

Ubiquitin specific protease 4 (USP4) is a deubiquitinating enzyme with key roles in the regulation of p53 and TGFβ signaling, suggesting its importance in tumorigenesis. However, the mechanisms and regulatory roles of USP4 in cancer, including colorectal cancer, remain largely elusive. Here, we present the first evidence that USP4 regulates the growth, invasion, and metastasis of colorectal cancer. USP4 expression was significantly elevated in colorectal cancer tissues and was significantly associated with tumor size, differentiation, distant metastasis, and poor survival. Knockdown of USP4 diminished colorectal cancer cell growth, colony formation, migration, and invasion in vitro and metastasis in vivo. Importantly, we found that phosphatase of regenerating liver-3 (PRL-3) is indispensable for USP4-mediated oncogenic activity in colorectal cancer. Mechanistically, we observed that USP4 interacted with and stabilized PRL-3 via deubiquitination. This resulted in activation of Akt and reduction of E-cadherin, critical regulators of cancer cell growth and metastasis. Examination of clinical samples confirmed that USP4 expression positively correlates with PRL-3 protein expression, but not mRNA transcript levels. Taken together, our results demonstrate that aberrant expression of USP4 contributes to the development and progression of colorectal cancer and reveal a critical mechanism underlying USP4-mediated oncogenic activity. These observations suggest that the potential of harnessing proteolytic degradation processes for therapeutic manipulation may offer a much-needed new approach for improving colorectal cancer treatment strategies. Cancer Res; 76(1); 83–95. ©2015 AACR.

Correlation of RARγ expression with patients' clinicopathological variables in 91 cases of CRC (S1); Human RARγ shRNA target sequences (S2); Primers for qPCR (S3).

Legends for Supplementary Figures S1-S4

Correlation of RARγ expression with patients' clinicopathological variables in 91 cases of CRC (S1); Human RARγ shRNA target sequences (S2); Primers for qPCR (S3).