Why High-Throughput Screening Still Breaks on 3D Models
Team Dynamic Matrices | 20 December 2025
Organoids and spheroids predict clinical response better than 2D monolayers, and most screening groups still cannot run them at scale. The bottleneck is not biology. It is materials handling. We have spent a good deal of time on screening decks with partners and the same four failures keep appearing. Here is what breaks, why, and what a matrix has to do to fix it.
Failure 1: cold gelation and the liquid handler
Basement membrane extracts are liquid at 4 °C and gel between about 10 and 37 °C. Every automated liquid handler was calibrated for aqueous reagents at room temperature. Put a matrix that starts gelling the moment it warms into that system and you get partial gelation in the tip, clogging, volume inconsistency, and domes that vary in shape. The workarounds (chilled decks, chilled tips, cooled reservoirs, frequent tip changes, small batches) eat the throughput advantage automation was meant to deliver. Several groups have published custom cold-dispensing hardware for exactly this reason (Brandenberg et al., Nature Biomedical Engineering, 2020, for an engineered microwell approach that avoids domes altogether).
Failure 2: well geometry and miniaturisation
Screening economics push toward 384- and 1536-well plates. An embedded organoid needs a centred, uniform gel droplet, and as well diameter shrinks the placement tolerance shrinks with it. A dome that is slightly off-centre in a 96-well plate images fine; in a 1536-well plate it touches the wall, wicks up the side, or leaves the organoid outside the imaging plane. Successful 3D screening platforms have generally built custom dispensing rather than reusing existing infrastructure, which is expensive and non-transferable.
Failure 3: the suspension compromise
To escape dispensing problems, many labs culture organoids in suspension or in low-attachment plates with a small percentage of matrix in the medium. That solves automation. It creates a biological problem: organoids without a surrounding matrix lack the mechanical and biochemical cues that shape transcriptome and drug sensitivity, which are the readouts the screen exists to measure (our physiological relevance post covers why). The compromise can quietly undo the reason for going 3D in the first place.
Failure 4: production variability
Organoid generation is slower and less predictable than 2D expansion, and matrix lot variation makes it worse: formation efficiency changes between lots, so the number of usable organoids per production run is not a reliable target. A campaign that needs 50,000 organoids on a fixed date cannot plan around a yield that moves with the matrix lot.
What a screening-compatible matrix has to do
From the four failures, the specification writes itself:
| Requirement | Why | What to ask a supplier |
|---|---|---|
| Gelation independent of ambient temperature | So it can sit on a room-temperature deck and gel on demand in the incubator | Gelation trigger and the time window at 20 to 25 °C |
| Low, stable viscosity before gelation | So standard tips dispense accurate volumes at 384 and 1536 | Viscosity at room temperature |
| Viscosity tunable separately from final mechanics | So you can lower viscosity for dispensing without softening the gel | Whether polymer content and stiffness are coupled |
| Batch-to-batch consistency | So production yield is predictable | Lot-specific rheology and formation-efficiency data |
| Gentle on-demand dissolution | So organoids can be recovered for endpoint assays on the same deck | Dissolution method and time |
| Optical clarity | So high-content imaging works through the gel | Transmission and autofluorescence data |
Animal-derived extracts fail most of these by nature. Chemically defined synthetic matrices can be engineered to meet them, which is a design-first rather than retrofit position.
How DyNAtrix was built for this
DyNAtrix is mixed with cells at room temperature and remains a low-viscosity liquid for a window long enough to fill a 384-well plate on a standard liquid handler; gelation is thermoresponsive and happens in the incubator, so nothing gels in the tip. We reduced viscosity and removed swelling specifically for dispensing and imaging. Stiffness is set by polymer content and stress relaxation by the DNA crosslinker, both consistent lot to lot. A nuclease dissolves the gel in minutes for cell recovery, which automates as an addition and an incubation. We are working with a high-content screening group at Imperial College London on 384-well protocols and would be glad to share parameters.
References
- Brandenberg N et al. High-throughput automated organoid culture via stem-cell aggregation in microcavity arrays. Nat Biomed Eng 4, 863–874 (2020).
- Du Y et al. Development of a miniaturized 3D organoid culture platform for ultra-high-throughput screening. J Mol Cell Biol 12, 630–643 (2020).
- Driehuis E, Kretzschmar K, Clevers H. Establishment of patient-derived cancer organoids for drug-screening applications. Nat Protoc 15, 3380–3409 (2020).
- Aisenbrey EA, Murphy WL. Synthetic alternatives to Matrigel. Nat Rev Mater 5, 539–551 (2020).
- Peng Y-H et al. Dynamic matrices with DNA-encoded viscoelasticity support the development of organoids and stem cells. Nat Nanotechnol (2023).
FAQs
They were calibrated for water-like liquids at room temperature. A reagent that begins gelling as it warms behaves unpredictably in the tip and the well.
Yes: a wide temperature window before gelation, consistent low viscosity during that window, and a gelation trigger that happens after dispensing. These are engineering choices in a synthetic material.
With a room-temperature-stable, low-viscosity matrix and an imager that handles the well depth, yes. Most groups start at 384 and move down once dispensing and imaging are validated.