From Cell States to a Compartment Model
A progenitor-to-mature-cell chain represents a biological sequence with a small set of cell populations and explicit flows between them. Its central modeling task is bookkeeping: a maturation flux removes cells from one compartment and adds the same flux to the next, while proliferation and loss change the population in the state where they occur. Writing those contributions separately makes the model interpretable and helps prevent counting a flow twice.
Consider three cell states: progenitor cells \(P\), an intermediate population \(I\), and mature cells \(M\). The model below describes their counts over time. It assumes that maturation proceeds in one direction, \(P\to I\to M\); progenitor cells have an effective expansion contribution; and each state has its own loss process.
For the basic model, assume the populations are well mixed within each state; maturation and loss rates are constant and first order; there are no reverse transitions or direct jumps from progenitor to mature cells; and mature cells do not proliferate. Progenitor proliferation is represented by an effective expansion coefficient: it describes the contribution of proliferation to the progenitor count, not the frequency of raw division events. An externally supplied progenitor input can be added when relevant, but is omitted from the baseline chain.
Deriving the Three Balance Equations
Let \(k_P\) be the per-capita maturation rate from \(P\) to \(I\), and \(k_I\) the per-capita maturation rate from \(I\) to \(M\). Let \(d_P\), \(d_I\), and \(d_M\) be the per-capita loss rates from the corresponding states. Let \(\rho_P\) be the effective per-capita progenitor expansion coefficient. All five coefficients have units of inverse time and are nonnegative. With population counts measured in cells, each flux has units of cells per unit time.
Under first-order assumptions, the flux out of a state is its count multiplied by the applicable per-capita rate. Thus the maturation fluxes are \(k_P P\) and \(k_I I\), and the state-specific loss fluxes are \(d_P P\), \(d_I I\), and \(d_M M\). Applying “change equals inflows minus outflows, plus any within-state expansion” gives the model:
The sign of each term follows from the compartment being modeled. Maturation is an outflow in the equation for the state a cell leaves and an inflow in the equation for the state it enters. Loss appears only as an outflow from the state where it occurs. Progenitor expansion appears only in the progenitor equation. These signs are not chosen to make a particular population increase or decrease; they follow from the flow definitions.
Worked Examples: Applying the Flow Bookkeeping
Worked Example: Building the Equations and Calculating Instantaneous Changes
Suppose hypothetical cell counts are \(P=100\) cells, \(I=40\) cells, and \(M=20\) cells. Let \(\rho_P=0.30\ \text{day}^{-1}\), \(k_P=0.20\ \text{day}^{-1}\), \(d_P=0.05\ \text{day}^{-1}\), \(k_I=0.40\ \text{day}^{-1}\), \(d_I=0.10\ \text{day}^{-1}\), and \(d_M=0.08\ \text{day}^{-1}\). Calculate the three instantaneous rates of change.
The equations use the same rates and signs regardless of the numerical values. For progenitors, the expansion contribution is \(0.30(100)=30\) cells/day, maturation removes \(0.20(100)=20\) cells/day, and loss removes \(0.05(100)=5\) cells/day. Therefore,
For the intermediate population, the progenitor maturation flux supplies \(0.20(100)=20\) cells/day. The intermediate maturation and loss fluxes are \(0.40(40)=16\) and \(0.10(40)=4\) cells/day, respectively. Hence,
For mature cells, the input flux is \(0.40(40)=16\) cells/day and the loss flux is \(0.08(20)=1.6\) cells/day:
These are instantaneous slopes at the specified state, not changes that must persist for an entire day. The calculation shows that the intermediate count is momentarily unchanged even though cells enter and leave that state at positive rates.
Worked Example: Checking That Maturation Is Counted Once
Use the same hypothetical counts and rates. Calculate the total-population rate of change by adding the three compartment equations, and identify which flows remain.
The total count is \(N=P+I+M\), so its derivative is the sum of the three derivatives:
The internal maturation terms cancel: \(-k_P P\) is paired with \(+k_P P\), and \(-k_I I\) with \(+k_I I\). Substituting the hypothetical values gives
This matches the sum of the compartment slopes, \(5+0+14.4=19.4\) cells/day. Maturation redistributes cells among the modeled states; it does not by itself change the total count. Expansion and losses do change the total.
Worked Example: Separating a High Maturation Flux From a Growing Compartment
Suppose a second hypothetical system has \(P=60\) cells and \(I=30\) cells, with \(k_P=0.50\ \text{day}^{-1}\), \(\rho_P=0.20\ \text{day}^{-1}\), \(d_P=0.05\ \text{day}^{-1}\), \(k_I=0.30\ \text{day}^{-1}\), and \(d_I=0.10\ \text{day}^{-1}\). Calculate the maturation flux from progenitors and the slopes of \(P\) and \(I\).
The progenitor-to-intermediate flux is \(k_P P\):
For progenitors, expansion contributes \(0.20(60)=12\) cells/day and the two outflows are maturation at \(30\) cells/day and loss at \(0.05(60)=3\) cells/day. Thus
For intermediates, the same 30 cells/day maturation flux is an inflow. Intermediate maturation removes \(0.30(30)=9\) cells/day, and intermediate loss removes \(0.10(30)=3\) cells/day. Therefore,
The large maturation flux does not imply that progenitors increase: here it contributes to a net decline in \(P\). At the same time, the flux supplies the intermediate state, which grows at this instant because its inflow exceeds its two outflows.
Why the Structure Matters
The chain separates a change in the number of cells in a state from a change in the total number of cells. A progenitor can leave its state by maturing while contributing a cell to the intermediate population. If the model instead treated maturation as a loss from the whole system, it would understate the intermediate input and make maturation appear to reduce the total population. Conversely, including the same maturation flux as an inflow without its source-state outflow would create cells.
A useful diagnostic is therefore to inspect every transition before interpreting a trajectory. For each arrow, write its flux once, then place it with opposite signs in the two relevant state equations. Separately list expansion and state-specific loss. Finally, check dimensions: a rate in day\(^{-1}\) multiplied by a count in cells gives cells/day, as required for a derivative of a cell count.
| Model feature | Contribution to the relevant equation | Interpretation |
|---|---|---|
| Progenitor expansion | \(+\rho_P P\) in \(dP/dt\) | Effective gain in progenitor count |
| Progenitor maturation | \(-k_P P\) in \(dP/dt\); \(+k_P P\) in \(dI/dt\) | Transfer from progenitor to intermediate |
| Intermediate maturation | \(-k_I I\) in \(dI/dt\); \(+k_I I\) in \(dM/dt\) | Transfer from intermediate to mature |
| State-specific loss | \(-d_j X_j\) in the equation for state \(j\) | Removal from that state, not transfer to the next one |
This explicit chain is more informative than a single mature-cell equation with an unspecified production term when the timing of maturation or the sizes of intermediate populations matter. A simpler model may be adequate if only mature-cell counts are observed and the intermediate stages cannot be distinguished; in that case, its production term should not be mistaken for a directly measured maturation flux. The chain also has limitations: constant first-order rates may be inadequate when rates vary with time or population state, and unmodeled pathways can make fitted rate values hard to interpret. Adding such mechanisms should be guided by data that can distinguish them, rather than by adding compartments without evidence.
Common Mistakes and Model Checks
- Counting maturation as loss from the entire chain. A cell leaving \(P\) by maturation is an input to \(I\), not a disappearance from the modeled population.
- Adding an inflow without its matching outflow. The same internal flux must have equal magnitude in the source and destination equations.
- Putting progenitor expansion in every compartment. In this model, expansion is defined only for \(P\). Other proliferating states require their own justified terms.
- Confusing a positive flux with a positive slope. A compartment’s slope is the sum of all its inflows minus all its outflows; a large inflow can still be outweighed by larger outflows.
- Using incompatible units. If counts are cells and time is days, the rate coefficients must be in day\(^{-1}\), and each flux must be in cells/day.
If observed trajectories cannot be reconciled with the model’s implied direction of flow, or if one fitted rate must compensate implausibly for another, the assumptions may be inadequate. Possible explanations include omitted pathways, a changing rate, or measurements that do not correspond cleanly to the defined states. Such patterns are reasons to reassess the model structure, not proof of a specific alternative mechanism.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- Write the intermediate-state equation for a chain with progenitor input flux \(k_P P\), intermediate maturation rate \(k_I\), and intermediate loss rate \(d_I\).
- Suppose \(I=50\) cells, \(k_I=0.20\ \text{day}^{-1}\), and \(d_I=0.04\ \text{day}^{-1}\). Calculate total outflow from \(I\) and name its components.
- Why do the maturation terms cancel when the three state equations are added?
- A model includes \(-k_P P\) in \(dP/dt\) and \(+k_P P\) in \(dI/dt\). Should \(k_P P\) also be subtracted from the total-population derivative? Explain.
Solutions. (1) \(dI/dt=k_P P-(k_I+d_I)I\): progenitor maturation is the inflow, while intermediate maturation and loss are outflows. (2) The maturation outflow is \(0.20(50)=10\) cells/day and the loss outflow is \(0.04(50)=2\) cells/day, so total outflow is \(12\) cells/day. (3) Each internal transfer is an outflow from one state and an equal inflow to another, so its negative and positive terms sum to zero. (4) No. The two terms cancel in the total derivative; the flux redistributes cells within the chain rather than removing them from it.