Drug repurposing screening and mechanism analysis based on human colorectal cancer organoids

Yunuo Mao, Wei Wang, Jingwei Yang, Xin Zhou, Yongqu Lu, Junpeng Gao, Xiao Wang, Lu Wen, Wei Fu, Fuchou TangProtein 2023 · DOI

TL;DR

This paper presents a novel organoid-based drug screening platform for colorectal cancer (CRC), combining computational prediction with experimental validation. They identified 34 potential repurposed drugs and characterized their mechanisms of action through transcriptome analysis, revealing five distinct drug response patterns. These findings were further validated in patient-derived organoid xenograft (PDOX) models, offering a promising avenue for accelerated CRC drug discovery.

What Problem Does This Paper Address?

Colorectal cancer (CRC) is a complex and highly heterogeneous disease, making effective treatment challenging. While targeted therapies have improved outcomes, many patients still lack effective options or develop drug resistance. Traditional drug discovery is time-consuming and expensive. This study addresses the need for more efficient and predictive methods to identify novel therapeutic agents for CRC, particularly by leveraging drug repurposing and advanced preclinical models. The authors aim to overcome the limitations of traditional cell line screening, which often fail to recapitulate the heterogeneity and molecular characteristics of actual tumors.

Methods

The researchers developed a multi-pronged approach:

  • Organoid Establishment: They established long-term, stable human CRC tumor-derived organoid lines that accurately reflect the morphological and molecular heterogeneity of the original tumors.
  • Organoid-Based Drug Screening: A robust screening platform was created using these organoids. They tested 335 approved small-molecule drugs and computationally predicted candidates. Drug sensitivity was measured using the Growth Rate Inhibition (GR50) metric, which accounts for cell proliferation.
  • Computational Drug Prediction: A ridge regression-based approach was employed, integrating existing drug response databases (CTRP, PRISM, GDSC1, GDSC2) and CRC disease signatures from TCGA. This helped predict drugs with potential inhibitory effects on CRC.
  • Transcriptome Analysis (RNA-seq): This was a key component for deciphering the mechanisms of action (MOA) of the identified drugs. By analyzing gene expression changes in organoids treated with drugs, they classified drug responses into five distinct patterns.
  • Patient-Derived Organoid Xenograft (PDOX) Models: To validate the findings in a more complex in vivo environment, the efficacy of selected drug candidates was tested in PDOX models.
  • Comparative Analysis: The study also investigated the concordance and discordance of drug response signatures between in vitro organoid models and in vivo PDOX models.

Datasets

  • Drug Response Databases: CTRP, PRISM, GDSC1, and GDSC2 were used for computational drug prediction.
  • Disease Signature Databases: TCGA (The Cancer Genome Atlas) was used for CRC disease signatures.
  • Single-Cell RNA-seq Data: A published single-cell RNA-seq dataset was used to predict potential drug responses for CRC.
  • Newly Established Datasets: The study generated new datasets through the establishment of eight human CRC organoid lines and their subsequent drug treatment and RNA-seq analysis. PDOX models were also generated for in vivo validation.

Key Insights and Ideas

  • Organoids as Superior Models: The study strongly reinforces the utility of patient-derived organoids as a more faithful preclinical model for CRC compared to traditional cell lines, capturing both histological and molecular heterogeneity.
  • Five Distinct Drug Response Signatures: The identification and characterization of five distinct transcriptome-based drug response patterns (differentiation induction, growth inhibition, metabolism inhibition, immune response promotion, and cell cycle inhibition) provide a valuable framework for understanding how different drugs affect CRC cells and for predicting treatment outcomes.
  • Bridging In Vitro and In Vivo: The comparison between organoid and PDOX models highlights the challenges and nuances of translating in vitro findings to in vivo efficacy, suggesting that while organoids are powerful, in vivo validation remains crucial.
  • Drug Repurposing Power: The success in identifying 34 drugs with anti-CRC effects, many of which were not originally developed for cancer, underscores the efficiency and potential of drug repurposing strategies.

Key Results

  • 34 Anti-CRC Drugs Identified: Out of 335 tested drugs, 34 showed significant anti-CRC effects in organoid screening.
  • Five Response Patterns Defined: Transcriptome analysis revealed five distinct patterns of drug response, offering insights into MOA.
  • In Vivo Validation: Fedratinib, trametinib, and bortezomib demonstrated effective anticancer effects in PDOX models, comparable to the first-line drug 5-fluorouracil.
  • Limited In Vitro-In Vivo Concordance: Only about 10% of differentially expressed genes overlapped between organoids and PDOX models after drug treatment, indicating significant differences in response signatures.
  • Potential Drug Combinations: The study explored the feasibility of predicting drug combinations, identifying potential synergistic effects between fedratinib and bortezomib.

Limitations and Future Work

  • Limited Sample Size for Mutation Correlation: While organoids captured diverse mutation subtypes, the limited number of samples prevented drawing definitive conclusions on the correlation between specific gene mutations and drug sensitivities. The authors suggest a larger panel of organoids with various mutations and from different tumor sites is needed.
  • In Vitro vs. In Vivo Discordance: The observed differences in transcriptome profiles between in vitro organoids and in vivo PDOX models warrant further investigation. Factors like drug dosage, tumor microenvironment, and mouse-specific evolution may contribute to these discrepancies.
  • Mouse-Specific Evolution: The authors acknowledge that xenografted tumors can undergo mouse-specific evolution, which might alter tumor characteristics. Further exploration is needed to understand how this impacts drug mechanism studies.
  • Future Directions: The study suggests that integrating organoid screening with transcriptome analysis and clinical data could accelerate the discovery of drugs targeting specific CRC subtypes or overcoming drug resistance. The potential for organoid systems to explore drug combination therapies is also highlighted.

Summary updated

Abstract

Colorectal cancer (CRC) is a highly heterogeneous cancer and exploring novel therapeutic options is a pressing issue that needs to be addressed. Here, we established human CRC tumor-derived organoids that well represent both morphological and molecular heterogeneities of original tumors. To efficiently identify repurposed drugs for CRC, we developed a robust organoid-based drug screening system. By combining the repurposed drug library and computation-based drug prediction, 335 drugs were tested and 34 drugs with anti-CRC effects were identified. More importantly, we conducted a detailed transcriptome analysis of drug responses and divided the drug response signatures into five representative patterns: differentiation induction, growth inhibition, metabolism inhibition, immune response promotion, and cell cycle inhibition. The anticancer activities of drug candidates were further validated in the established patient-derived organoids-based xenograft (PDOX) system in vivo. We found that fedratinib, trametinib, and bortezomib exhibited effective anticancer effects. Furthermore, the concordance and discordance of drug response signatures between organoids in vitro and pairwise PDOX in vivo were evaluated. Our study offers an innovative approach for drug discovery, and the representative transcriptome features of drug responses provide valuable resources for developing novel clinical treatments for CRC.
BibTeX
@inproceedings{mao2023drug,
  author = {Yunuo Mao and Wei Wang and Jingwei Yang and Xin Zhou and Yongqu Lu and Junpeng Gao and Xiao Wang and Lu Wen and Wei Fu and Fuchou Tang},
  title = {{Drug repurposing screening and mechanism analysis based on human colorectal cancer organoids}},
  booktitle = {Protein},
  year = {2023},
  doi = {10.1093/procel/pwad038},
}
Figure