Characterisation and quality control: what we measure, what we miss

Reading time: 7 minutes

The previous article in this series introduced characterisation as one of the core methods in stem cell science. This article looks at it more closely, because the gap between what we currently measure and what we need to know is one of the most consequential bottlenecks in the field. If you are building tools for stem cell workflows, this gap is where many of the most valuable problems sit.

What characterisation means in practice

Characterisation is the process of establishing what a cell is, what state it is in, and whether it is fit for its intended purpose. In stem cell science, this spans every stage of the workflow: confirming the identity and quality of starting materials, monitoring cells during culture and expansion, and assessing differentiated products before they are used in research, screening, or therapy [1].

The challenge is that stem cells are not static objects. They exist in shifting states influenced by culture conditions, passage number, the mechanical and biochemical environment, and the history of signals they have received. Two cell lines that look identical under a microscope and express the same surface markers can behave differently in a differentiation protocol. Characterisation must therefore capture not just identity but function, stability, and predictive potential [2].

The standard toolkit

Current characterisation of pluripotent stem cells typically involves several layers of assessment.

Identity confirmation relies on short tandem repeat (STR) profiling to verify that the cells are the correct line and have not been contaminated with or replaced by another. This is a basic but necessary check: cell line misidentification has been a persistent problem across biomedical research for decades [3].

Pluripotency assessment uses a combination of surface markers (SSEA-3, SSEA-4, TRA-1-60, TRA-1-81), transcription factor expression (Oct4, Nanog, Sox2), and functional assays. The most stringent functional test has historically been the teratoma assay, in which cells are injected into immunodeficient mice and assessed for their ability to form tissues from all three germ layers. This assay is slow, variable, expensive, and ethically problematic from the perspective of animal usage and welfare. Alternatives based on gene expression profiling, such as the PluriTest and ScoreCard assays, offer faster and more standardised readouts but do not fully replace functional assessment [1,4].

Genetic integrity is traditionally assessed by G-banded karyotyping, which detects large chromosomal abnormalities such as trisomies and translocations. Pluripotent stem cells are known to acquire recurrent genetic changes during prolonged culture, with gains of chromosomes 1, 12, 17, and 20 being among the most common. Some of these confer a growth advantage and can affect differentiation capacity and safety [5]. Higher-resolution methods, including single nucleotide polymorphism (SNP) arrays and next-generation sequencing, can detect sub-karyotypic changes such as copy number variants and loss of heterozygosity, but are not yet routinely applied in all contexts [5].

Sterility and mycoplasma testing confirm the absence of microbial contamination, a non-negotiable requirement for any cell product destined for clinical use and a basic quality standard for research materials [1].

Where the gaps are

The standard toolkit described above is necessary but not sufficient. It answers the question "is this cell what it claims to be?" reasonably well. It answers the question "will this cell do what I need it to do?" much less reliably.

Several specific gaps deserve attention.

Epigenetic stability is poorly captured by existing assays. Pluripotent stem cells carry a complex landscape of DNA methylation and histone modifications that influence which genes are accessible and how the cell responds to differentiation signals. These marks can drift during culture, and iPSCs can retain epigenetic memory of their tissue of origin, which biases their differentiation potential [6]. Standard characterisation does not routinely assess the epigenome, even though epigenetic state may be a better predictor of functional performance than gene expression or surface markers alone.

Heterogeneity within populations is another blind spot. A culture of pluripotent stem cells is not a homogeneous pool. It contains cells in subtly different states, including naive-like, primed, and transitional states, as well as cells that have begun to differentiate spontaneously. Standard bulk assays report an average across the population, which can mask the presence of subpopulations that may compromise the quality or consistency of downstream products [7]. Single-cell methods (transcriptomics, proteomics) can reveal this heterogeneity but are not yet practical as routine quality control tools.

Functional potency, the ability of cells to generate specific functional derivatives, is the most important property for most applications and the hardest to measure prospectively. A cell line may express all the right pluripotency markers and have a normal karyotype, yet consistently fail to produce mature cardiomyocytes or functional neurons. Current assays do not provide a reliable way to predict this in advance, short of running the differentiation protocol itself, which can take weeks and is not practical as a quality gate for incoming materials [2,4].

Metabolic profiling is an emerging area. Stem cells in different states have distinct metabolic signatures, with pluripotent cells relying more on glycolysis and differentiated cells shifting toward oxidative phosphorylation. Metabolic assays could provide rapid, non-destructive readouts of cell state, but the field has not yet standardised what metabolic profiles predict about downstream performance [8].

What this means for technology developers

Each of these gaps represents an opportunity for ancillary technology companies. The field needs assays and instruments that can characterise stem cells more deeply, more quickly, and at the single-cell level, ideally without destroying the cells being assessed.

Non-destructive, label-free methods are of particular interest. Techniques based on Raman spectroscopy, biophysical measurements such as cell elasticity and dielectric properties, and machine learning-enhanced imaging offer the potential to assess cell quality in real time during culture, rather than at fixed checkpoints after the fact [9,10]. Our own collaborative research explored biophysical methods including dielectrophoresis for label-free discrimination and separation of pluripotent stem cells, demonstrating that physical properties of cells can carry information about their identity and state [10].

Computational approaches are also becoming important. Machine learning models trained on imaging data, gene expression profiles, or process parameters can potentially predict cell quality and differentiation outcomes earlier and more reliably than traditional assays. The challenge is that these models require large, well-annotated datasets, and the stem cell field has not yet generated these at the scale or standardisation needed for robust training [9].

For regulatory purposes, there is a pressing need for potency assays that are validated, quantitative, reproducible, and predictive. Regulators require evidence that a cell product will perform its intended function, and the assays used to demonstrate this must be fit for purpose. Developing and validating such assays is a substantial undertaking, but it is one of the clearest bottlenecks to clinical translation of stem cell therapies [1,2].

The bigger picture

Characterisation is not a standalone step. It is the thread that connects raw material quality to process control to product performance. Every other method described in this series, from directed differentiation to cryopreservation to manufacturing, depends on characterisation to confirm that the process is working and the product is acceptable.

The current state of stem cell characterisation is roughly where clinical chemistry was before the development of standardised reference ranges and automated analysers. The methods exist, but they are not yet mature enough to support the scale, speed, and reliability that clinical and commercial applications demand. Closing this gap will require new tools, new standards, and new ways of integrating data across the workflow. The Pillar 2 series on why ancillary technologies fail examines the commercial consequences of these characterisation gaps, including how reproducibility and scale-up challenges amplify the measurement problem.

References

[1] Stacey GN, Crook JM, Hei D, Ludwig T. Banking human induced pluripotent stem cells: lessons learned from embryonic stem cells? Cell Stem Cell. 2013;13(4):385-388. DOI: 10.1016/j.stem.2013.09.007

[2] Steeg R, Mueller SC, Mah N, et al. EBiSC best practice: how to ensure optimal generation, qualification and distribution of iPSC lines. Stem Cell Reports. 2021;16(8):1853-1867. DOI: 10.1016/j.stemcr.2021.07.009

[3] Horbach SPJM, Halffman W. The ghosts of HeLa: how cell line misidentification contaminates the scientific literature. PLoS One. 2017;12(10):e0186281. DOI: 10.1371/journal.pone.0186281

[4] Bock C, Kiskinis E, Verstappen G, et al. Reference maps of human ES and iPS cell variation enable high-throughput characterization of pluripotent cell lines. Cell. 2011;144(3):439-452. DOI: 10.1016/j.cell.2010.12.032

[5] Baker DEC, Harrison NJ, Maltby E, et al. Adaptation to culture of human embryonic stem cells and oncogenesis in vivo. Nat Biotechnol. 2007;25(2):207-215. DOI: 10.1038/nbt1285

[6] Kim K, Doi A, Wen B, et al. Epigenetic memory in induced pluripotent stem cells. Nature. 2010;467(7313):285-290. DOI: 10.1038/nature09342

[7] Hough SR, Thornton M, Mason E, Mar JC, Wells CA, Pera MF. Single-cell gene expression profiles define self-renewing, pluripotent, and lineage primed states of human pluripotent stem cells. Stem Cell Reports. 2014;2(6):881-895. DOI: 10.1016/j.stemcr.2014.04.014

[8] Zhang J, Nuebel E, Daley GQ, Koehler CM, Bhatt AP. Metabolic regulation in pluripotent stem cells during reprogramming and self-renewal. Cell Stem Cell. 2012;11(5):589-595. DOI: 10.1016/j.stem.2012.10.005

[9] Kusumoto D, Lachmann M, Kunber T, et al. Automated deep learning-based system to identify endothelial cells derived from induced pluripotent stem cells. Stem Cell Reports. 2018;10(6):1687-1695. DOI: 10.1016/j.stemcr.2018.04.007

[10] Pethig R, Menachery A, Pells S, De Sousa P. Dielectrophoresis: a review of applications for stem cell research. J Biomed Biotechnol. 2010;2010:182581. DOI: 10.1155/2010/182581

About StemCells.Help

StemCells.Help is an advisory consultancy that aids innovation and real-world impact of life science applications built on developmental and stem cell biology. Founded by Dr Paul De Sousa, it draws on over four decades of experience spanning early embryo development, animal cloning, pluripotent stem cell manufacturing, and technology commercialisation. If you build tools for these domains or work in an emerging application where the biology is the enabling technology, StemCells.Help can provide experienced scientific counsel to ground your decisions. To discuss your needs, talk to Paul.

ORCID: 0000-0003-0745-2504

Web: stemcells.help