Choosing a Hydrogel for Stem Cell Culture: What Actually Matters

Team Dynamic Matrices | 9 April 2026

We spend a lot of time talking to stem cell labs about their matrix, and the same pattern repeats. The hydrogel was chosen years ago because a protocol paper used it, nobody has revisited the decision, and it is now the quiet cause of half the variability in the lab. Stem cells read their microenvironment continuously. The gel is not a container; it is an input.

Here are the six criteria we use, in the order we would apply them, with the evidence behind each.

Criterion 1: Is the composition defined?

This is the first fork. Basement membrane extracts such as Matrigel deliver a rich, native-like biochemistry, but Hughes and colleagues (Proteomics, 2010) identified over 1,800 proteins in it, and the growth factor content differs from lot to lot. Decellularised ECMs have the same issue in a different wrapper. Defined matrices (synthetic PEG, polyacrylamide-based systems, engineered proteins) give you a known composition and a clean translational path, at the cost of having to supply adhesion ligands yourself.

Our recommendation: for any quantitative, longitudinal or translational stem cell work, start from a defined matrix and add biology deliberately. New Approach Methodology initiatives from FDA and EMA are pushing the same direction, and reviewers increasingly ask why an undefined animal extract was used.

Criterion 2: Can you set stiffness, and over what range?

Engler et al. (Cell, 2006) showed MSCs differentiate along neural, myogenic or osteogenic lines depending on substrate elasticity alone. That result has been replicated many times, including in 3D. Native tissues span roughly 0.1 kPa (brain) to tens of kPa (pre-calcified bone), so a hydrogel that only reaches one part of that range limits what you can model.

Ask two things of any candidate: what range of storage modulus can it reach, and can stiffness be changed without also changing ligand density or mesh size? In many natural matrices you change stiffness by changing protein concentration, which changes everything else too.

Criterion 3: Does it relax stress like tissue?

This is the criterion most labs skip, and the one we care most about. Living tissues are viscoelastic: they dissipate stress over seconds to minutes. Chaudhuri et al. (Nature Materials, 2016) showed that in alginate gels of identical stiffness, faster stress relaxation increased MSC spreading, proliferation and osteogenic differentiation. Later work extended this to neural progenitors, intestinal organoids and iPSC-derived systems (reviewed in Chaudhuri et al., Nature, 2020).

A purely elastic gel, which is what most synthetic PEG gels are, presents a mechanical environment that does not exist in the body. If the hydrogel cannot reproduce tissue-like stress relaxation, you are running a mechanics experiment you did not intend to run.

Criterion 4: Can you get the cells back, alive?

Passaging, sorting, single-cell sequencing and re-embedding all require recovery. Harsh enzymatic or mechanical dissociation damages sensitive populations and skews the cells you keep, and dissociation itself triggers an immediate-early gene stress signature in single-cell data (van den Brink et al., Nature Methods, 2017). A matrix that releases cells gently and quickly is worth more than most people realise until a harvest fails.

Criterion 5: Imaging and handling

Stem cell work is imaging-heavy. Check optical clarity at your working thickness, autofluorescence in your channels, and whether the gel tolerates repeated confocal or light-sheet sessions. On handling: room-temperature stability, a simple gelation step and low viscosity all reduce technique-driven variability. Anything that requires ice, fast pipetting or precise temperature windows will introduce operator effects.

Criterion 6: Reproducibility, batch to batch

Underlying all of the above: does the gel behave the same way every time? For biological extracts the answer is structurally no. For defined materials it should be yes, but ask the supplier for a certificate of analysis with mechanical data, not just a lot number.

A one-week validation plan we recommend

1. Embed your line in your current matrix and one candidate, three technical replicates each.
2. Day 3: viability (live/dead) and morphology.
3. Day 7: a lineage marker panel relevant to your question (OCT4/NANOG for maintenance, or your differentiation markers).
4. Recover cells from both and count viable yield.
5. Repeat once with a second lot of each matrix.

If the candidate matches or beats your current gel on these five points, switch. If not, you have learned something about your line for the price of two plates.

Where DyNAtrix® fits

We built DyNAtrix against exactly this list. It is fully synthetic and animal-free, with a defined composition (a linear polyacrylamide backbone crosslinked by short synthetic DNA strands, plus adhesion peptides such as RGD). Stiffness and stress relaxation are set independently through the DNA crosslinker design, and stress relaxation is tunable over roughly five orders of magnitude. Cells are recovered enzymatically with a nuclease step that leaves them intact, and in our hands recovery viability is high. It is optically clear, stable at room temperature and thermoresponsive, so embedding does not require an ice bucket. We have validated it with human iPSCs, mesenchymal stromal cells, trophoblast organoids and MDCK cysts, with cultures maintained beyond 60 days (Peng et al., Nature Nanotechnology, 2023). Explore DyNAtrix further.