The Case for Retiring Basement Membrane Matrices

Team Dynamic Matrices | 11 January 2026

We are going to make an argument that is obviously in our commercial interest, so we will make it with citations and we will say where it does not apply. The argument: basement membrane extracts (Matrigel, Cultrex, Geltrex and their relatives) should be treated as legacy defaults, kept for the narrow cases where nothing else yet works, and no longer the automatic first choice for new protocols.

1. How it is made contradicts why we use it

Basement membrane extract is harvested from Engelbreth-Holm-Swarm sarcomas grown in mice. The tumours are allowed to reach a substantial fraction of the animal's body weight before harvest; Kleinman and Martin (Seminars in Cancer Biology, 2005) describe the original production, and current supply requires large numbers of animals per year for a global market. Organoid research is justified, in grants and in regulatory submissions, as a way to reduce and replace animal experiments. Growing those organoids in a mouse tumour extract is a contradiction that reviewers and ethics committees are starting to notice.

2. It is not adequately controlled, and everyone knows it

Hughes, Postovit and Lajoie (Proteomics, 2010) found more than 1,800 proteins in Matrigel, most uncharacterised, with lot-variable growth factor content. The universal response is to qualify favoured lots and stockpile them. That practice is sensible. It is also an admission that the material is not under control. A reagent whose users routinely hoard specific batches is not a reagent that meets the standard we apply to everything else in a rigorous protocol. Kozlowski, Crook and Ku (Communications Biology, 2021) lay out the case in detail.

3. Regulators are moving

The FDA Modernization Act 2.0 (December 2022) removed the statutory requirement for animal testing before human trials. The FDA's April 2025 roadmap for reducing animal testing in preclinical safety studies explicitly names organoids and microphysiological systems as acceptable New Approach Methodologies. The EMA and OECD are on the same path. When a regulator evaluates organoid data as animal-replacement evidence, an animal-derived, undefined matrix inside that model is a weakness in the submission. Defined, animal-free materials are not yet required, but they are where the direction of travel points.

4. The alternatives are now good enough

A decade ago the honest objection was that nothing else worked. That is no longer true across a widening range of applications:

• Intestinal organoids in fully synthetic PEG (Gjorevski et al., Nature, 2016)
• Enteroids and endometrial organoids in defined synthetic gels (Hernandez-Gordillo et al., Biomaterials, 2020)
• Liver organoids in mechanically defined synthetic niches (Sorrentino et al., Nature Communications, 2020)
• Epithelial organoids in defined fibrin-laminin gels (Broguiere et al., Advanced Materials, 2018)
• iPSCs, MSCs, trophoblast organoids and kidney cysts in DNA-crosslinked viscoelastic gels for over 60 days (Peng et al., Nature Nanotechnology, 2023)

The gap has narrowed to specific lineages with specific uncharacterised requirements. It is no longer a general gap.

A practical transition path

We are not proposing anyone throw out their Matrigel tomorrow.

1. New protocols start on a defined matrix by default. If it works, you never inherit the lot problem.
2. Existing protocols migrate at a natural break, a new passage or a new study, after a side-by-side validation on a subset of conditions.
3. Keep BME for the narrow cases: matching a published baseline exactly, or a lineage whose synthetic requirements are not yet enumerated. Report the lot number when you do.
4. Report matrix identity, lot and mechanics in methods, whichever matrix you use.

The narrow exceptions, stated plainly

Basement membrane extract still earns its place when exact comparison to an existing literature dataset is the point of the experiment, and when a lineage depends on a component nobody has yet identified and replaced. Those are real cases. They are exceptions, and they should be treated as such rather than as the default the rest of the field inherits.