Skip to content
← Back to browse
BRAIN Publication-derived

Deep-learning analysis of micropattern-based organoids enables high-throughput drug screening of Huntington's disease models

Source Metzger et al., 2022 · The Rockefeller University · 10.1016/j.crmeth.2022.100297

👤 Jakob J. Metzger, Carlota Pereda, Arjun Adhikari, Tomomi Haremaki, Szilvia Galgoczi, Eric D. Siggia, Ali H. Brivanlou, Fred Etoc ⏱ 47 days 📋 10 phases 🧫 Human ESC (RUES2)

Abstract

This protocol describes a high-throughput screening platform combining micropatterned neural organoids ('neuruloids') with deep-learning image analysis to identify compounds that rescue Huntington's disease phenotypes. The approach uses immunofluorescence imaging of organoids, convolutional neural networks for classification, and autoencoders to quantify drug efficacy and adverse effects, enabling discovery of bromodomain inhibitors as rescue agents for HD developmental defects.

Cell source
Human ESC (RUES2)
Application
Disease modeling and drug screening

Protocol overview

38 steps across 10 phases

Cell preparation and micropattern coating Day -1 to Day 0
  1. 1 Coat micropatterned plates with laminin
  2. 2 Wash coated plates
  3. 3 Prepare single-cell suspension
Neuruloid formation and differentiation Day 0 to Day 7
  1. 1 Seed cells onto micropatterned plates
  2. 2 First medium change with differentiation factors (Day 3)
  3. 3 Second medium change (Day 5)
  4. 4 Harvest organoids at Day 7
Compound treatment and media management Day 3 to Day 7
  1. 1 Prepare and pin compound library
  2. 2 Maintain control wells during screen
  3. 3 Continue differentiation through Day 7
Immunofluorescence staining Day 7
  1. 1 Fix organoids in micropattern plates
  2. 2 Block and permeabilize
  3. 3 Incubate with primary antibodies
  4. 4 Incubate with secondary antibodies and DAPI
  5. 5 Mount coverslips
High-content imaging and image acquisition Day 7 (post-staining)
  1. 1 Acquire confocal images on fixed organoids
  2. 2 Image full wells using high-content imager
  3. 3 Stitch and extract individual organoids
Deep learning analysis: classifier training and phenotype classification Post-imaging
  1. 1 Data cleaning with DAPI neural network
  2. 2 Train CNN classifier on WT and HD control images
  3. 3 Apply classifier to screen compounds
Deep learning analysis: adverse effect quantification using autoencoders Post-imaging (parallel to Phase 6)
  1. 1 Train convolutional autoencoder on control organoid images
  2. 2 Encode all organoid images into latent space
  3. 3 Quantify rescue and adverse effect in phenotypic space
  4. 4 Define hit compounds and validate with MTT assay
Validation assays: 500 μm micropatterns and dose-response Post-screening (validation phase)
  1. 1 Re-validate hits on 500 μm diameter micropatterns
  2. 2 Generate dose-response curves
  3. 3 Analyze individual channel contributions
Secondary assay: neuronal differentiation and BDNF withdrawal apoptosis assay Day 0 to Day 47
  1. 1 Differentiate hESC lines to cortical neurons
  2. 2 Passage neurons and continue maturation
  3. 3 Plate neurons in 96-well glass-bottom plates
  4. 4 BDNF withdrawal and compound treatment (Days 40–47)
  5. 5 TUNEL apoptosis assay and quantification
  6. 6 Quantify TUNEL intensity
  7. 7 Measure neuronal projections (optional)
Hit characterization: mechanism of action studies Post-screening (validation phase)
  1. 1 Screen bromodomain inhibitor panel
  2. 2 Quantify HTT protein levels
  3. 3 Explore mechanism of action (mechanistic studies)

Full SOP

🔬

Create a free account to access this protocol

Join OrganMatch to unlock step-by-step procedures, reagent concentrations, QC checklists, and downloadable batch record templates.

Create free account

Already registered? Log in

Attribution

This SOP was authored by Organthis based on the published method in Metzger et al., 2022. The originating laboratory holds no rights in this SOP and has not endorsed it unless marked Verified.

This wording is awaiting legal review.

Something wrong with this entry? Report an issue with this protocol

Need a commercial licence?
Use this protocol in your therapeutic or diagnostic pipeline.
Request Commercial Licence