Activation-Dependent Expansion and Contraction of Engineered Cells

Why Expansion Can Turn Into Contraction

Engineered cells may expand while stimulation is strong and contract as that stimulation wanes. A useful model must distinguish the activation signal from the cell count: the signal changes the per-cell rates of proliferation and loss, while the cell count accumulates the net effect of those rates over time. This distinction makes it possible to predict not only whether cells are growing now, but also when growth is expected to peak.

The model below treats stimulation as a continuously decaying input rather than as an abrupt on/off switch. Activation follows a saturable relationship with the signal, proliferation is proportional to activation, and the per-cell loss rate is constant. These assumptions describe a deliberately compact mechanism; they do not represent every process that can shape an engineered-cell time course.

The model assumes that the signal is externally specified and is not depleted by the cells; activation responds instantaneously to the current signal; all modeled cells share the same rates; and no density limit or additional compartment changes the count. Consequently, \(p_{\max}A(t)-d\) is the net per-cell growth rate. A positive value means the count increases at that time, and a negative value means it decreases.

Deriving the Expansion-to-Contraction Threshold

At a given signal level \(S\), the net per-cell rate is \(g(S)=p_{\max}S/(K_A+S)-d\). If \(d>0\) and \(p_{\max}>d\), set this rate to zero to find the signal level at which its sign changes:

Because the activation fraction increases with \(S\), the net rate is positive above this threshold and negative below it. With exponentially decaying stimulation, an initial signal \(S_0>S_{\mathrm{crit}}\) therefore produces an initial expansion phase followed by contraction. The switch occurs at the finite time

This condition matters: if \(S_0\leq S_{\mathrm{crit}}\), the cells do not have an initial period of positive net growth. If \(p_{\max}\leq d\) with \(d>0\), no finite signal can make the net rate positive. If \(d=0\) and \(S_0>0\), the net rate is positive at every finite time because the decaying exponential remains positive; there is no finite-time expansion-to-contraction switch. These cases should not be conflated with a model that has an abrupt signal cutoff.

Cell Counts Accumulate the Changing Net Rate

The sign of the net rate identifies whether the count is rising or falling at an instant, but the count at a later time depends on the rate over the whole interval. Starting from \(N(0)=N_0\), divide the cell-count equation by \(N(t)\) and integrate:

For the activation term, let \(v=K_A+S_0e^{-k_s u}\). Then \(dv/du=-k_sS_0e^{-k_s u}\), so its integral from 0 to \(t\) is \(k_s^{-1}\ln[(K_A+S_0)/(K_A+S(t))]\). The constant loss term integrates to \(dt\). Thus the explicit solution is

This solution is continuous through the peak: there is no instantaneous jump in cell count when the net rate crosses zero. At the peak the count is maximal, because its derivative changes from positive to negative. The signal may continue to activate the cells after the peak; contraction begins when activation is no longer sufficient to offset the modeled loss rate.

Different Starting Conditions, Different Trajectories

The same mechanism can produce contraction from the outset. If the initial signal is below threshold, the population begins with negative net growth; because the signal continues to fall, it cannot later cross upward through the threshold. The following calculation checks this case with the same rate parameters as above.

A post-peak count can also remain above its initial value. That does not contradict contraction: contraction describes the current direction of change, whereas the count reflects the history of earlier expansion as well as later loss.

Assumptions, Diagnostics, and Alternatives

The model is most interpretable when the measured count represents the modeled viable population, the activation signal is a reasonable driver of proliferation, and loss is adequately approximated by a constant per-cell rate over the interval. A useful diagnostic is to compare observed time-course features with the model’s implications: a single peak if \(S_0>S_{\mathrm{crit}}\), a smooth change in slope at that peak, and subsequent contraction under continuing signal decay. A peak followed by renewed expansion, multiple peaks, or an extended plateau may indicate that the assumed input or rate structure is inadequate.

The model also predicts that changing \(S_0\) affects both the initial activation and, when the threshold is crossed, the peak time. Changing \(k_s\) changes how quickly the signal traverses the threshold. The peak count depends on the accumulated balance of proliferation and loss, not on peak timing alone. In practice, cell counts by themselves may not uniquely identify signal decay, activation sensitivity, maximum proliferation, and loss: different parameter combinations can produce similar trajectories. Independent information on stimulation or cell biology can therefore be important when interpreting fitted parameters.

This formulation differs from a constant net-growth model, which cannot represent a signal-driven switch unless its rate is changed by hand. It also differs from the saturable expansion models for infused cells, where expansion may be limited by the cell population or a capacity term. Here, saturation describes the response of activation to stimulation; it does not impose a maximum cell count. Adding a density limit, multiple cell states, signal feedback, or changing loss may be warranted for a particular application, but each adds assumptions that require evidence.

The introductory material on cell and gene therapy state representations distinguishes cell states from measured quantities; in this model, \(N(t)\) is one viable-cell state and \(S(t)\) is a time-varying driver. The model does not explain why cells ultimately persist, nor does it distinguish loss from movement out of a measured compartment. Those interpretations require information beyond this single-population expansion–contraction equation.