In the exploding field of stem cell biology, confusion pervades among some newcomers, and even veterans. The question is simple: When do we call a cell - stem cell?

Some would argue that a fertilized egg is the ultimate stem cell. It is totipotent, giving rise to the embryo and extra-embryonic structures. The more fate-restricted cells of the inner cell mass of a blastocyst give rise to all body tissues and, with appropriate culture, to embryonic stem cells, which are considered to be pluripotent. Yet another term is reserved for the even more fate-restricted often multipotent stem cells, found widely from 8-day-old embryos to 90-year-old bone marrow. But these landmark definitions are slippery. Until a few years ago, for instance, immunologists routinely referred to stem cells in the blood as pluripotent, even though their fates are normally restricted to the blood. This is one potential source of confusion.

The criteria for a stem cell remain ambiguous. Must stem cells self-renew? If so, the fertilized egg and the cells of the inner cell mass of a blastocyst do not meet their current stem cell appellation. Both will be exhausted as they give rise to derivatives that are unlike themselves. Notably, true self-renewal is a misnomer, because mutations accumulate in the genome with every round of DNA replication; thus, the daughter of a stem cell is marginally distinct, at least by this criterion, from its parent. For practical purposes, however, this imprecision is tolerated.

Credit: Jessica Kolman

No naming scheme will be perfect, but certainly alternatives preferable to the current ad hoc situation can be readily imagined.

Historically, the word stammzelle (German for stem cell1,2; see also3,4) had a dual meaning: the first being the evolutionary unicellular ancestor of multicellular organisms, and the second being the ancestral (ontological) stem cell of a tissue in an organism, initially in the germ line. Subsequently, the term became more widely applied to other tissues. But as the notion of stem cells evolved, the definition also became somewhat muddled, particularly due to loose standards for naming such cells and distinguishing them from their progeny.

Probably the most puzzling terms are progenitor and precursor — here, the definitions are murky. Sometimes the progenitor (from Latin pro-, meaning forth, and gignere, meaning to beget) or precursor (from Latin prae-, meaning pre-, and currere, meaning to run) designations were already in place when a more ancestral cell type was discovered, thereby limiting the terms available to identify a new cell state. In other instances, stem cells are referred to as progenitors when the use of the stem cell appellation appears to leave a doubt about their true nature. Hence, the label of 'stem/progenitor' cell. In addition, developmental biologists have traditionally referred to ancestral embryonic cells in their favourite lineage as precursors. Indeed, for any given cell in the embryo, there exists a precursor cell that gives rise to it. Confusion arises also when nomenclature is used in a generic sense to denote any ancestral cell type (for example, precursor = stem cell), as well as in a specific context to denote a specific cell type (for example, precursor = 'blast' cell (neuroblast, myoblast and so forth)).

The notion of transit amplifying cells also needs to be parked somewhere in this constantly shifting scheme, and their relationship with progenitors and precursors should be clarified. Generally, the term applies to a class of cells, with a finite life span, that arise from stem cells, proliferate and then differentiate. However, one study in skin proposed that stem cells can yield differentiated cells in the absence of a transit amplifying population5.

Another intriguing example is the pancreas, in which stem/progenitor cells appeared to play no significant role in adults6 — yet some cells mobilized in an adult pancreas injury model do give rise to all islet cell types, including beta cells7. Will these pancreatic 'progenitors' self-renew for the lifetime of the organism, and, if so, should they be considered as stem cells – or will they be exhausted during that round of regeneration?

The model of blood or haematopoietic stem cells provides an elegant paradigm for stem cell biology and tissue generation: an assortment of highly divergent cell types are derived from a well-characterized stem cell. The more committed haematopoietic progenitors may be viewed as self-renewing, to a limited extent, or simply proliferating - and they give rise to multiple cell types before being exhausted. In this context, 'progenitor' reflects an unspecified cell that can normally generate multiple cell types — as is also the case in the central nervous system. Notably, during stem cell expansion, self-renewal needs to be operationally distinguished from proliferation; sometimes this distinction, which requires a sort of balance between symmetric and asymmetric cell divisions, is not clear-cut.

The haematopoietic system has provided numerous insights, yet it has also inflicted a tragic nomenclature upon us — 'short-term' and 'long-term' stem cells — referring to the capacity of the cells to generate the lineage over a given time interval following their transplantation. Although useful in that context and assay, a terminology that can be translated to other tissues and species may be more convenient. A mayfly lives for about a day, whereas some turtles persist for a couple of centuries. Self-renewal is a luxury in the former, but a necessity for tissue regeneration in the latter. Thus, for comparative purposes across species and tissues, transposable criteria, other than time, are desirable for relating different cell states.

Currently, the generic nomenclature used for identifying cell states may hold us hostage to biased thinking. Bias and confusion can arise when a cell state is named before it is fully characterised. If some cell types follow a non-canonical lineage progression during tissue regeneration, their differentiating cell states may not fit a generic nomenclature scheme. In addition, just as skin may sometimes not require a transit–amplifying population to maintain the tissue,5 not all embryonic tissues will produce all of the anticipated generic cell states during normal organogenesis. This would result in apparent 'gaps' in the naming scheme.

Is there a solution to the confusion?

Some measure of clarity could be obtained if the fertilized egg and other cells that establish tissues are referred to as founder cells rather than as stem cells. This can distinguish them from the self-renewing cells that emerge as organogenesis concludes. Subsequently, one can identify tissue-specific stem cells, typically postnatal, that can self-renew.

Furthermore, with generic terms, there is an implicit assumption that all lineages have the same number of cell states (for example, stem cell to transit amplifying cell (progenitor? + precursor?) to differentiated cell). As indicated above, this may not be the case5. Instead, cells could be named within their specific context — their lineage — of parents and grandparents, daughters and granddaughters. The notion of stem cells arose with investigations in embryology, hence envisaging these cells in a developmental biology context is constructive. Here, the position that a cell occupies within a lineage progression would be a landmark (though reprogramming, cell state reversal and transdifferentiation would have to be factored into this equation). With this view, one can temporarily dispense with problematic terms (stem cell, progenitor cell, precursor cell) and identify criteria that distinguish cell states in the form of A → B → C → D and so forth in which the last state depicts the differentiated and mature phenotype. In this way, all states can be designated if they display a characteristic chromatin/epigenetic signature and gene-expression pattern. This would apply also to mesenchymal stem cells, which give rise to multiple cell types, and are found in diverse tissues and organs. They would slip into different lineages in tissues and organs, presumably as specific states, to achieve the differentiated phenotype. Cultured cells, cancer cells, cells undergoing regeneration, or reprogrammed cells may not follow a predictable lineage progression. Nevertheless, defining distinct cell states in those cases is also a necessity for their proper characterisation.

Credit: Jessica Kolman

Another bias with the nomenclature currently in use is the assumption that only one stem cell entity exists in any given tissue. This may also be incorrect; one can imagine having A1, A2...An as different stem cells in any given tissue (B1, B2...Bn and C1, C2...Cn for their daughters and granddaughters). Indeed, some haematopoietic stem cells were suggested to already exhibit lineage bias8. Skeletal muscle provides another example — the embryonic counterparts of adult satellite 'stem' cells are born with distinct genetic signatures that depend on their location. Interestingly, they can maintain this ontological signature to some extent in adults9. Hence, this tissue is composed of different stem cells with distinct genetic signatures.

Therein lies another challenge. Due to current technical limitations, arguably every stem cell population being examined prospectively is heterogeneous, consisting of 'stem-like' cells and more committed cells. In some cases, this represents a single stem cell state 'breathing' within the limits of transcriptional noise. In others, bona fide distinct cell states are being observed. Stem cells may also differ from each other at the single cell level due to fluctuations in gene expression. In some cases, stem/progenitor cells may refrain from participating in normal tissue turnover, and thus escape detection until they are unveiled after an acute injury7. Defining the true stem cell state, be it fixed or not, is one of the major challenges in this field.

No naming scheme will be perfect, but certainly alternatives preferable to the current ad hoc situation can be readily imagined.

But, the unbiased scheme also has some pitfalls. In the subventricular zone (SVZ) of the central nervous system, type 'B' and 'A' cells were designated historically prior to knowledge regarding their lineage relationships. Another cell type in this lineage was eventually called 'C'. Further characterisation revealed the stem cells to be 'B', the neuroblast 'A', and the intermediary progenitor to be 'C' — so oddly, we are left with B → C → A (ref. 10).

No naming scheme will be perfect, but certainly alternatives preferable to the current ad hoc situation can be readily imagined (See Box 1). In the field of reprogramming and induced pluripotent cells, reference cell states, preferably not associated with inherently biased definitions, are crucial for clarity. Of course, what a cell is called may not matter if it can reconstitute a failing organ. Although precise language may not be a formal necessity, it could nevertheless lower barriers of conception and communication and thereby alleviate some of the confusion to achieve at least one key goal — using the right cell types to get cell therapies to work.

Ultimately, what's in a name? A lot, but often, very little.

Box 1. Terminology summary

  • The designations 'stem', 'progenitor' and 'precursor' are currently used as somewhat generic terms to denote an ancestral cell — as well as occasionally being used in a specific context to denote a particular cell state.

  • Fertilized eggs and other embryonic stem cells that establish tissues and organs can be viewed as 'founder' cells. Self-renewing cells that emerge as organogenesis is concluding may be considered 'stem cells'.

  • Non-generic terms may be employed that designate a cell state, its parent and its daughter in a lineage progression.