Wei Chen

Tumorigenesis typically follows a multi-hit trajectory, driven by accumulating oncogenic mutations. Colorectal cancer (CRC) has long served as a paradigmatic model of multi-hit tumorigenesis, characterized by adenoma-carcinoma transition accompanied by acquisition of specific oncogene and tumor suppressor mutations. However, how the temporal order of early mutations influences CRC initiation remains poorly understood. To address this, we established a CRC tumorigenesis model using murine intestinal organoids. By introducing defined combinations of key CRC driver mutations ( Kras , Apc , and Trp53 ) in distinct orders, we systematically investigated how the order of mutation accumulation affects tumor initiation. Our results reveal that the mutation accumulation confers growth advantages in both in vitro and in vivo models. Strikingly, mutation order also influenced the tumorigenic properties of the organoids. Whereas organoids with Trp53 loss before or after Apc loss similarly affected organoid phenotypes in vitro or tumorigenicity in immunodeficient mice, organoids with Trp53 loss preceding Apc inactivation exhibited reduced tumor-forming potential in immunocompetent mice, likely due to their distinct immunological features. Collectively, our study reveals a critical role of ordered mutation accumulation in CRC initiation, an insight that may hold clinical relevance.

Tumorigenesis typically follows a multi-hit trajectory, driven by accumulating oncogenic mutations. Colorectal cancer (CRC) has long served as a paradigmatic model of multi-hit tumorigenesis, characterized by adenoma-carcinoma transition accompanied by acquisition of specific oncogene and tumor suppressor mutations. However, how the temporal order of early mutations influences CRC initiation remains poorly understood. To address this, we established a CRC tumorigenesis model using murine intestinal organoids. By introducing defined combinations of key CRC driver mutations (Kras, Apc, and Trp53) in distinct orders, we systematically investigated how the order of mutation accumulation affects tumor initiation. Our results reveal that the mutation accumulation confers growth advantages in both in vitro and in vivo models. Strikingly, mutation order also influenced the tumorigenic properties of the organoids. Whereas organoids with Trp53 loss before or after Apc loss similarly affected organoid phenotypes in vitro or tumorigenicity in immunodeficient mice, organoids with Trp53 loss preceding Apc inactivation exhibited reduced tumor-forming potential in immunocompetent mice, likely due to their distinct immunological features. Collectively, our study reveals a critical role of ordered mutation accumulation in CRC initiation, an insight that may hold clinical relevance.

Measurement of mRNA translation level has heavily relied on quantifying the amount of polysome-associated RNAs, which would not always reflect efficient translation. To address this, we developed ReleAsing Factor Tethered Editing of RNA (RAFTER), which combined the RNA editing enzyme with translation releasing factor to selectively label successfully translated RNAs. The translation level estimated based on RAFTER correlated well with that by Ribo-seq and Polysome-seq, but with a greatly simplified workflow, which enables it for single-cell analysis.

Measurement of mRNA translational efficiency has heavily relied on quantifying the amount of polysome-associated RNAs which would not always reflect the efficient translation. To address this, we developed RAFTER, which combined RNA editing enzyme with translation releasing factor to selectively label successfully translated RNAs. The translational efficiency estimated based on RAFTER correlated well with that by Ribo-seq and Polysome-seq, but with greatly simplified workflow, which enables it for single-cell analysis.

Zygotic genome activation is tightly associated with the modulation of chromatin accessibility via maternal transcription factors. Understanding how chromatin accessibility is established and identifying key maternal regulators are crucial to comprehending this process. Here, by developing CANTAC-seq, we generate a genome-wide map of accessible chromatin of early Xenopus tropicalis embryos and find that the open chromatin landscape is progressively established at cis-regulatory elements during zygotic genome activation. Based on the motif analysis and perturbation experiments, we demonstrate that E2f1 maintains a repressive chromatin environment and inhibits zygotic gene transcription before the mid-blastula transition. Moreover, we identify that Otx1, another maternal transcriptional activator, coordinates with E2f1 in regulating chromatin accessibility and zygotic genome activation. Together, E2f1 and Otx1 determine the timely expression of a subset of genes required for zygotic gene transcription and germ layer differentiation.

RNA, an essential component of the central dogma of molecular biology, plays versatile roles in all cellular processes. RNA large language models (LLMs) are emerging as powerful methods in RNA research to decipher its intricate network of function and regulation. However, previous RNA LLMs were based on the Transformer model and pre-trained on short segment of non-coding RNAs, which limits their general usability. Here we present the first full-length RNA foundation model, HydraRNA, which is based on a hybrid architecture of bidirectional state space model and multi-head attention mechanism. HydraRNA is pre-trained on a large amount of both protein-coding mRNAs and non-coding RNAs. Despite being pre-trained with the fewest parameters and the least GPU resources, HydraRNA learns better RNA representations and outperforms the existing foundation models on a variety of mRNA-related tasks, including coding/non-coding RNA classification, prediction of RNA secondary structure, RBP binding sites, splicing and polyadenylation sites, mRNA stability and translation efficiency. Furthermore, HydraRNA can accurately predict the effect of mutations and estimate the relative contributions of different mRNA regions to the RNA stability and translation. Our results demonstrate that the hybrid architecture outperforms the pure Transformer architectures in RNA language modeling. We anticipate that HydraRNA will enable dissecting the diverse properties of mRNA, accelerating the research of mRNA regulation and facilitating the optimal design of mRNA therapeutics.

Processing bodies (PBs) are dynamic, membraneless organelles consisting of RNAs and proteins. While PB proteins have been extensively characterized, the methods for systematically profiling PB-associated RNAs are limited. To address this, we developed PB-TRIBE-STAMP, a tool based on two orthogonal RNA editing enzymes. Simultaneously applying APOBEC1-DDX6 and LSM14A-ADAR2dd, PB-TRIBE-STAMP identified 1,639 and 2,577 PB-associated mRNAs in HCT116 and HEK293T cells, respectively. Further biochemical isolation of PBs followed by RNA-seq validated that edited transcripts of these mRNAs were indeed enriched in PBs. Integration of PB-TRIBE-STAMP with long-read sequencing revealed that the PB-associated transcripts possessed shorter poly(A)-tails. Many mRNA 3’ UTR isoforms exhibited isoform-specific PB association patterns. Moreover, we established a TRIBE-ID-based tool to characterize the mRNA-LSM14A/PB association at high temporal resolution and unveiled a higher splicing efficiency of LSM14A-associated XBP1 transcripts during unfolded protein response (UPR). Finally, based on single-cell LSM14A-TRIBE-ID (sc-LSM14A-TRIBE-ID), we demonstrated the dynamic pattern of mRNA-LSM14A/PB association during cell cycle progression.