Stem, Progenitor, and Differentiated Compartments

Why Separate a Lineage Into Compartments?

A total cell count can conceal important changes in population structure. The same total might consist mostly of stem cells at one time and mostly of differentiated cells at another. A useful lineage model therefore distinguishes cell states that have different roles: stem cells maintain the lineage, progenitors expand and supply downstream cells, and differentiated cells represent the terminal output being tracked.

The goal is not to describe every individual division. Instead, a compartment model records the rates at which cells enter, leave, or are added within each state. Keeping these processes distinct makes it possible to ask whether a changing differentiated-cell count reflects altered supply from progenitors, altered loss from the differentiated pool, or both.

Define the Compartments and Fluxes

Let \(N_S(t)\), \(N_P(t)\), and \(N_D(t)\) denote the numbers of stem, progenitor, and differentiated cells at time \(t\). Consider the unidirectional hierarchy \(S\to P\to D\). In this model, stem cells contribute cells to the progenitor compartment, progenitors contribute differentiated cells, and no cells return upstream.

We distinguish two types of input from the transitions. Self-renewal supplies the stem compartment, helping maintain its population. Progenitor amplification adds progenitor cells within the progenitor compartment. For a simple linear model, represent those effective additions by \(r_S N_S\) and \(r_P N_P\), respectively. Here \(r_S\) and \(r_P\) have units of inverse time. They describe effective net cell additions per source cell per unit time, not counts of individual division events.

The transition flux from stem to progenitor is \(k_{SP}N_S\), and the transition flux from progenitor to differentiated cells is \(k_{PD}N_P\). The coefficients \(k_{SP}\) and \(k_{PD}\) also have units of inverse time. Let \(L_S\), \(L_P\), and \(L_D\) denote the respective cell-loss fluxes, measured in cells per unit time. They are written as fluxes rather than assigned a particular loss mechanism or rate law.

With these definitions, the model is a set of cell-count balances:

Each term has units of cells per unit time. Self-renewal adds cells to \(S\); commitment from \(S\) removes cells from \(S\) and supplies \(P\). Amplification adds cells within \(P\); differentiation removes cells from \(P\) and supplies \(D\). Loss removes cells from the compartment in which it occurs. This accounting makes clear that differentiation is not itself a loss from the whole modeled lineage: it is a transfer between compartments.

What Happens to the Total Cell Count?

Adding the three balance equations gives a useful check on the model. Each transition appears once as an outflow and once as an inflow: \(k_{SP}N_S\) leaves \(S\) and enters \(P\), while \(k_{PD}N_P\) leaves \(P\) and enters \(D\). The two contributions therefore cancel in the total.

This identity follows directly from cell accounting: internal transfers redistribute cells but do not change the total number across the three compartments. In the preceding example, total change is \(2+20-(1+8+20)=-7\) cells per day. The same result follows by adding the compartment rates, \(0+3-10=-7\) cells per day.

The cancellation is also a diagnostic. If a proposed set of equations makes a one-way differentiation flux decrease total cell number without an explicit loss, the transition has likely been counted as an outflow without its matching inflow. Conversely, if a transition is added to both compartments as an input, it has not been removed from its source.

Interpret Fluxes Before Interpreting Counts

The rates \(r_S\) and \(r_P\) represent distinct biological roles in this coarse-grained model. A change in \(r_S\) changes the supply to the stem compartment; a change in \(r_P\) changes the addition of progenitor cells. By contrast, \(k_{SP}\) redirects cells from the stem compartment to the progenitor compartment, and \(k_{PD}\) redirects cells from progenitors to differentiated cells. Changes in those transition coefficients affect both the source and destination compartments, with opposite signs.

A differentiated-cell count is consequently an output of upstream supply and local loss, not a direct readout of any one process. In particular, a low \(N_D\) alone does not establish that differentiation is slow. It could also arise from a small progenitor pool, weak progenitor amplification, or substantial differentiated-cell loss. The model helps organize these explanations, but deciding among them requires information about counts or fluxes beyond a single terminal measurement.

Assumptions, Diagnostics, and Alternative Representations

The three-compartment structure is useful when the main question concerns the balance of stem maintenance, progenitor expansion, and terminal output. Its simplicity depends on assumptions that should be checked against the intended application:

• Compartment meaning is consistent. Each state must group cells whose modeled roles are sufficiently similar. If a “progenitor” label combines states with very different expansion or differentiation behavior, one effective rate may not describe them well.
• The lineage is treated as one-way. The equations omit dedifferentiation, reversible switching, and branching into separate fates. If those processes matter, the specified hierarchy is incomplete.
• Rates and fluxes have appropriate units and signs. Transition and addition coefficients are nonnegative and have units of inverse time; loss fluxes are nonnegative and have units of cells per time. A negative loss flux would represent an unmodeled source, not ordinary loss.
• The coarse-grained inputs have a defensible interpretation. The \(r_S N_S\) and \(r_P N_P\) terms are effective net additions in their respective compartments. They should not be interpreted automatically as raw division frequencies.

A practical diagnostic is to write a flux table before fitting or simulating the equations: list each source and destination, its units, and whether it adds cells, transfers them, or removes them. Then verify the total-count identity. If observed trajectories cannot be represented without implausible negative fluxes or inconsistent compartment meanings, the model structure or its inputs may be inadequate.

Alternative representations are appropriate for different questions. A maturation chain separates several successive intermediate stages when their residence times or counts matter. A branching-flux model is more suitable when the important feature is a split into distinct downstream fates. A single total-population model is more compact, but it cannot describe how cells are distributed among stem, progenitor, and differentiated states. The three-compartment model is a middle ground: it preserves these functional distinctions without representing every division or every intermediate state.

A common mistake is to treat differentiation as both a loss from the whole lineage and a source of differentiated cells. The incorrect reasoning is that \(k_{PD}N_P\) should be subtracted from the total because it leaves the progenitor pool, while also being added to the differentiated pool. The correction is to distinguish a compartment outflow from a lineage loss: differentiation leaves \(P\) but enters \(D\), so it cancels in the total balance. Only fluxes that leave the modeled lineage reduce its total cell count.