中文导读
基于 PDO 的高通量筛选（HTS）正在走向临床。本文介绍筛选流程、关键质控点与自动化整合方案。
以下为英文全文（English full text）
1. Introduction
High-throughput screening (HTS) using patient-derived organoids (PDOs) has emerged as one of the most promising applications of organoid technology in precision oncology. Unlike conventional cell-based assays that rely on immortalized lines with accumulated genetic drift, PDOs retain the genomic landscape, histological architecture, and drug response profiles of the original patient tumor. This article presents detailed protocols, validation frameworks, and analytical methodologies for implementing PDO-based HTS in drug discovery and clinical translation pipelines.
2. Establishment and Quality Control of Patient-Derived Organoids
2.1 Tissue Collection and Processing
Successful PDO establishment begins with proper tissue collection and processing. Tumor specimens should be obtained from fresh surgical resections or core needle biopsies and transported in organoid culture medium on ice within 2–4 hours. The tissue is minced into 1–2 mm fragments using sterile scalpels, then digested with collagenase-dispase (2–5 mg/mL) and DNase I (10 U/mL) in Advanced DMEM/F12 supplemented with 1× GlutaMAX, 10 mM HEPES, and 100 U/mL penicillin-streptomycin for 30–60 minutes at 37°C with gentle shaking.
2.2 Culture Conditions and Medium Composition
Tumor organoids typically require niche-specific growth factor cocktails. For colorectal cancer organoids, the culture medium includes EGF (50 ng/mL), Noggin (100 ng/mL), R-spondin-1 (500 ng/mL), A83-01 (500 nM), and SB202190 (10 µM). For pancreatic cancer organoids, additional Wnt3a-conditioned medium and FGF10 (100 ng/mL) are often required. Matrigel or basement membrane extract (BME) at a concentration of 10–12 mg/mL serves as the extracellular matrix scaffold, with organoids embedded as 50 µL domes in pre-warmed 24-well plates.
2.3 Quality Control and Authentication
Each PDO line must undergo rigorous quality control before entering screening workflows. Short tandem repeat (STR) profiling confirms genetic identity and rules out cross-contamination. Whole-exome sequencing (WES) or targeted next-generation sequencing (NGS) panels verify that the PDO retains the driver mutations present in the original tumor. Immunohistochemistry (IHC) staining for lineage-specific markers (e.g., CK20 for colorectal, CA19-9 for pancreatic) confirms histological fidelity. Passage number should be limited to P5–P8 for drug screening to minimize culture adaptation artifacts.
3. High-Throughput Screening Protocols
3.1 Organoid Dispensing and Plating
For HTS, PDOs are dissociated into single cells or small clusters (3–10 cells) using TrypLE Express or Accutase for 5–10 minutes at 37°C. Cell viability is assessed by trypan blue exclusion, and only preparations with >90% viability are used. Organoids are plated in 384-well or 1536-well plates at densities of 1,000–5,000 cells per well in 20–50 µL of Matrigel-medium mixture. Automated liquid handlers (e.g., Hamilton, Tecan) are recommended for plate preparation to ensure uniformity.
3.2 Compound Library Preparation and Dosing
Compound libraries are typically prepared as 10 mM DMSO stock solutions and stored at −80°C. For screening, compounds are diluted in culture medium to 3× or 5× final concentrations and dispensed using acoustic dispensers (e.g., Echo 550) or pin tools. Standard dose-response screening uses 8-point serial dilutions (1:3 or 1:5) spanning 0.1 nM to 100 µM, with a final DMSO concentration not exceeding 0.5% to avoid solvent toxicity. Each compound is tested in technical triplicates per PDO line.
3.3 Positive and Negative Controls
Every screening plate includes appropriate controls. Negative controls (vehicle-only wells, typically 0.1% DMSO) establish baseline viability. Positive controls include benchmark chemotherapeutic agents known to be effective for the tumor type (e.g., 5-fluorouracil for colorectal, gemcitabine for pancreatic). Staurosporine (1 µM) serves as a universal cytotoxic positive control. Each plate should contain at least 8 control wells per condition.
4. Assay Readouts and Endpoint Measurements
4.1 Cell Viability Assays
CellTiter-Glo 3D (Promega) is the gold standard for PDO viability assessment. This luminescent assay quantifies intracellular ATP, which correlates with metabolically active cells. After 72–96 hours of drug exposure, 25–50 µL of CellTiter-Glo reagent is added per well, and plates are incubated for 30 minutes at room temperature with gentle shaking. Luminescence is read on a multi-mode plate reader (e.g., PerkinElmer EnVision, Molecular Devices SpectraMax). The assay demonstrates a linear dynamic range of 1,000–50,000 cells per well and a Z-prime factor >0.5 in validated PDO screens.
4.2 High-Content Imaging
High-content imaging systems (e.g., Opera Phenix, ImageXpress Micro) enable multiparametric phenotypic readouts. Organoids are stained with Hoechst 33342 (nuclei), calcein-AM (live cells), and ethidium homodimer-1 (dead cells) for 1 hour at 37°C. Automated image acquisition captures 4–9 fields per well at 10× or 20× magnification. Image analysis pipelines quantify organoid number, size, circularity, and live/dead ratios using algorithms developed in CellProfiler or Harmony software.
4.3 Apoptosis and Mechanistic Readouts
Caspase-3/7 activity (Caspase-Glo 3/7 Assay) provides early indicators of apoptosis, typically measured at 24 and 48 hours post-treatment. For mechanistic profiling, ATP-binding cassette (ABC) transporter activity can be assessed using fluorescent substrates (e.g., calcein-AM for MRP, Hoechst 33342 for BCRP). Reactive oxygen species (ROS) generation is measured with CellROX Deep Red Reagent, and mitochondrial membrane potential with JC-1 or TMRE dyes.
5. Data Analysis and Statistical Considerations
5.1 Normalization and Quality Metrics
Raw viability data are normalized to plate-specific negative controls (set to 100% viability) and positive controls (set to 0% viability). Percent viability is calculated as: (Sample signal - Mean positive control) / (Mean negative control - Mean positive control) × 100%. Assay quality is evaluated using the Z-prime factor: Z' = 1 - (3×SD_positive + 3×SD_negative) / |Mean_positive - Mean_negative|. Plates with Z' < 0.5 are flagged for repeat.
5.2 Dose-Response Curve Fitting
Dose-response data are fitted to a four-parameter logistic (4PL) model using software such as GraphPad Prism, R (drc package), or Pipeline Pilot. The model is defined as: Y = Bottom + (Top - Bottom) / (1 + 10^((LogEC50 - X) × HillSlope)), where X is the log concentration, Y is the percent response, Bottom is the minimum response plateau, Top is the maximum response plateau, LogEC50 is the log concentration producing 50% of maximal response, and HillSlope is the slope factor.
Goodness-of-fit is assessed by R-squared values >0.85 and visual inspection of residual plots. Curve fitting failures (due to flat responses, high variability, or insufficient dynamic range) are flagged and reviewed manually.
5.3 Key Pharmacological Parameters
The primary output parameters from dose-response curves are:
- IC50: The concentration inhibiting 50% of cell viability relative to controls
- AUC (Area Under the Curve): Integrated measure of drug sensitivity across all tested concentrations, calculated using the trapezoidal rule
- EC50: The concentration producing 50% of the maximal effect
- Emax: The maximal achievable effect at saturating concentrations
- DSS (Drug Sensitivity Score): A normalized metric accounting for both potency and efficacy, calculated as the integral of the fitted curve normalized to the testing range
5.4 Statistical Analysis
For comparing drug responses across multiple PDO lines, one-way ANOVA with Tukey's post-hoc test or Kruskal-Wallis with Dunn's test (for non-parametric data) is used. For paired comparisons (e.g., pre/post treatment or mutant/wild-type), paired t-tests or Wilcoxon signed-rank tests are applied. Multiple testing correction using the Benjamini-Hochberg false discovery rate (FDR) is essential when testing many compounds or lines simultaneously. A threshold of FDR < 0.05 is standard for significance.
6. Validation and Clinical Correlation
6.1 Concordance with Clinical Response
The clinical validity of PDO-based HTS is established by comparing in vitro drug responses with patient outcomes. Vlachogiannis et al. (2018) demonstrated a positive predictive value (PPV) of 88% and negative predictive value (NPV) of 100% for PDO-based predictions of colorectal cancer patient responses to chemotherapy. Pasch et al. (2019) reported 100% accuracy in predicting clinical response to chemotherapy in pancreatic cancer patients using PDOs, with a median time of 4 weeks from tissue collection to actionable results.
6.2 Reproducibility Metrics
Inter-batch reproducibility is assessed by calculating the coefficient of variation (CV) for IC50 values across independent experiments. A CV < 25% is considered acceptable for clinical decision-making. Intra-batch reproducibility (technical replicates within the same experiment) should achieve CV < 15%. Concordance correlation coefficients (CCC) between biological replicates should exceed 0.8.
6.3 Reference Standards and Benchmarking
Validation panels should include FDA-approved drugs with established clinical efficacy for the tumor type, enabling benchmarking against known standards. The National Cancer Institute (NCI) 60-cell line panel or commercially available reference organoid lines (e.g., ATCC, HUB Organoids) can serve as external controls. Cross-laboratory validation through ring trials is recommended before clinical implementation.
7. Automation and Scalability
7.1 Robotic Integration
Fully automated workflows integrate robotic liquid handlers, plate washers, and incubators with barcode tracking for chain-of-custody documentation. Systems such as the Tecan Freedom EVO or Hamilton VANTAGE can process 50–100 PDO lines per week in 384-well format, screening up to 5,000 compounds per line. Automated organoid dispensers (e.g., CellAssembler) maintain viability during plating by controlling temperature and shear stress.
7.2 Data Management and LIMS Integration
High-throughput screens generate substantial datasets requiring structured management. Laboratory Information Management Systems (LIMS) such as Benchling, STARLIMS, or custom platforms built on PostgreSQL/MongoDB are used to track sample provenance, assay metadata, and results. Integration with computational pipelines for curve fitting, hit calling, and report generation is essential for turnaround times compatible with clinical decision-making.
8. Challenges and Future Directions
Despite remarkable progress, several challenges remain. The establishment success rate varies by tumor type (40–90%), and the time required for expansion (4–8 weeks) may exceed the window for clinical decision-making in aggressive cancers. Standardization of culture conditions, assay endpoints, and data analysis pipelines across laboratories is an active area of development. The integration of immune components, stromal cells, and vascular networks into PDO screening platforms will further enhance physiological relevance.
9. Conclusion
Patient-derived organoid-based high-throughput screening represents a transformative approach to precision drug discovery. By combining rigorous protocols for organoid establishment, validated assay platforms, and robust statistical frameworks, researchers can generate clinically actionable drug sensitivity profiles. As automation and standardization continue to advance, PDO-HTS is poised to become a cornerstone of both pharmaceutical drug development and personalized oncology.