Definitive Phenotypes in Hematopoeisis

Standardization may not be the most exciting or cutting-edge topic in biomedical research but in an era where we lament the lack of reproducibility and distribute blame to reagents, sample prep and general technical know-how, standardization is something every researcher should spend time considering. We are in a time where generating big data is attainable but sifting through that data to make relevant, reliable biological relationships can still be a challenge on many levels. The challenge I’ll address here is how do we “name” a cell and assign its identity. How do we all know that we are talking about the same cells? How do we agree on what it takes to confidently correlate previous with current research when the depths of our technical capabilities have advanced so much farther? I like to think of the cell as a person. Think of your own name. If you search for your name, how many others on the planet share your name? So, expand your search to include a middle name. Still too many people? What about a maiden name? What about restricting a search just to a country of residence? Or just your institution? This is how we create trees to delineate the relationship between cells and accuracy will depend on which criteria we employ that is definitively correlative.

I’ve recently been working on a project here at FluoroFinder that we call the Cell Ontology Tool. The intent is to help guide researchers to standardize definitive phenotypes in flow cytometry. Ontology is the best term to describe that specific purpose as it describes not just delineating all the subsets upon subsets of cells that compose our organism but the origins and function of those subsets and their relationship to their neighbors and progeny. We often use the term definitive phenotype in flow cytometry which describes the least number of markers to define specific cell populations. Not too long ago, flow cytometry assays were limited to 8-10 fluorescent markers to define the basic landscape of cell subsets. With a limited number of fluorescent features, the focus was on cell surface markers, the proteomic level of phenotype, thus allowing us to sort live cells into categories of cell subsets. As flow cytometric technology advanced to 18 colors, staining for intracellular cytokines and transcription factors confirmed a surface marker phenotype with functional markers. A commonly used example are helper and cytotoxic T cell subsets defined by an expression pattern of 5 chemokine receptors but confirmed by cytokine and transcription factor expression. Check out our interactive T Cell Maturation page to learn more.

Another example of how functional markers define true members of a subset are NK T cells which express both NK and T cell surface markers. In Immunoprofiling flow cytometry assays (General Immunophenotyping page), where the purpose is to give broad strokes to cell subsets by using the least number of markers to define the most subsets in a single assay, CD3+CD56+ is a simple marker phenotype that groups together all NK T cells. In a 2021 paper from Lenart et al¹, the prevalence of iNKT cells when defined only by CD3+/CD56+ is between 8-13% of CD3+ T lymphocytes in healthy individuals. However, by including a functional probe like a CD1d tetramer loaded with a-GalCer, that percent representation of iNKT cells falls to 0.3-0.5%. How big of a difference is that when trying to draw conclusions across the literature about the function of NK T cells?

Now we approach 50 fluorescent parameters on a single cell sample in flow cytometry. At the same time, cellular ontology at the transcriptomics level is being driven full speed ahead by single cell sequencing technologies. For researchers interested in subjects like hematopoiesis, progenitor maturation and fate choice, the marriage of these two technologies is truly paradigm breaking. I want us to consider Human Hematopoiesis for a minute, even if this is not your area of expertise, it is a glaring example of how standardization is imperative. Human bone marrow is a hot commodity, not easily attainable from healthy subjects. Therefore, most research has been done in rodents, cell lines or humanized models. Stem cells generally don’t produce proteins at a high abundance which limits the sensitivity of our assays. Also, most transitional subsets are rare which requires more sample than may be accessible to get statistical significance or to distribute into appropriate controls for the confidence required in a high-parameter assay. And then, we must agree on which protein expression patterns confirm the transition between developmental stages. For example, there are 4 oligopotent progenitor populations regularly defined by differential expression of CD123 and CD45RA.

Common Myeloid Progenitor (CMP): CD45RA-/CD123^low

Lymphoid-Primed Multipotent Progenitor (LMPP):CD45RA+/CD123-

Megakaryocyte-Erythroid Progenitor (MEP): CD45RA-/CD123-

Granulocyte-Myeloid Progenitor (GMP): CD45RA+/CD123^low

If CD123 isn’t detected with a bright enough fluorophore or an appropriate gating control (like an FMO) isn’t employed to define the edges of this population gate or the assay involves more than 30 fluorescent parameters and this marker has been assigned to a fluorophore with heavy spectral overlap and thus is a victim of spillover spreading error in the data analysis, the confidence that these fundamental gates contain only the progenitor of interest could be confounded.

Another example of errors in experimental design I’ve encountered in the literature are the cells referred to as double negative (DN) T cells, the name describing cells that lack CD4 and CD8 expression. They are detailed in the T cell maturation page. There are two important markers, CD34 and CD3, that are required to distinguish different types of DN T cells in any assay. DN T cells expressing CD34 are members of an immature developmental stage that occurs predominantly in bone marrow. Yet many papers that are specifically investigating DN T cells often don’t include CD34 in a flow cytometry panel, likely reasoning that it is not relevant when using peripheral tissue and blood. And yet, CD34+ cells are not found exclusively in bone marrow. Second is the inclusion of CD3. Some papers will call out the “misnaming” of DN T cells when some are triple negative, lacking CD3, CD4 and CD8 in development. This distinction is a big difference when defining ILC, gamma/delta T cells and MAIT cell subsets that are CD3+/CD4-/CD8-. Without including these markers, you may be inadvertently mixing mature and immature populations.

Finally, as a last example of the importance of defining ontogeny to construct a consistent framework across all cell assay platforms, let’s consider the many examples where markers “shouldn’t” be positive on a cell subset, and yet they are. For this topic, we are in luck that dimensionality reduction analysis techniques like ViSNE or UMAP, help to ensure that our biases in gating data along a linear progression of including and excluding events from analysis doesn’t lead to blindly ignoring relevant relationships simply because we didn’t know to look for them. Innate Lymphoid Cells (ILC) have only recently gained a heightened recognition. In previous assays, some ILC would have been lost in CD56+ gates in cases like the subset ILC3 where 50% express CD56. I always like to say, you don’t know what you don’t know. Marker expression, especially on these cells whose function is to bridge innate and adaptive immunity, might be more interchangeable than previously accepted.

I suspect that the ontology trees we’ve constructed in these pages will some day be rectified when 30+ color comprehensive high-parameter proteomic-level flow cytometry profiles are appropriately matched to single cell sequencing and Cite-Seq assay design and both are correlated with downstream functional confirmation. I suspect these big data results will create a tabula rosa, a clean slate, and new beginning from which to build our understanding of the nature of immature cells, their maturation, fate choice and functional commitment. We would like to encourage you to submit your own publications to keep this tool relevant and accurate to the best of our abilities, that it can reflect the rapidly expanding understanding of cellular ontology.