CD8 T-Cell Expansion and Virus-Specific Killing

Linking Antigen, CD8 Effectors, and Viral Burden

Virus-specific CD8 T cells can expand when they encounter antigen, kill infected cells, and contract when antigenic stimulation wanes. These processes create a time-dependent immune effect: the same viral replication process may initially dominate, then be overcome as effectors accumulate. Consequently, the timing of CD8 activation can matter as much as the eventual effector level. A delay may allow a larger viral peak even if the later immune response is unchanged in shape.

A mechanistic within-host model distinguishes uninfected target cells \(T\), infected cells \(I\), free virus \(V\), and virus-specific effectors \(E\). One useful structure is:

Here, \(A\) is an antigen measure, such as an infected-cell burden or a related presentation signal. The saturating function \(f(A)\) is near zero when \(A\) is small and approaches one when \(A\) is large. The source \(s_E f(A)\) represents antigen-dependent recruitment or priming, while \(\rho f(A)E\) represents antigen-stimulated expansion of the existing effector pool. The term \(d_EE\) describes loss of effectors, and \(kEI\) is CD8-associated infected-cell killing. Parameters are nonnegative; \(K_A\) has the units of \(A\), and \(\beta,p,c,\delta_0,k,s_E,\rho,d_E\) have units that make each equation dimensionally consistent.

The equations are a model structure, not a claim that measured antigen instantaneously equals the signal seen by T cells. They assume that \(A\) is an adequate driver of stimulation and that one effective effector population captures the relevant CD8 activity. In applications, activation delays, tissue distribution, and variation in cytotoxic function may require additional states.

Expansion and Contraction Conditions

If antigen is approximately constant over a short interval and recruitment is negligible relative to the existing effector pool, the effector equation becomes \(dE/dt=r_EE\), where \(r_E=\rho f(A)-d_E\). This approximation gives exponential expansion when \(r_E>0\), constant abundance when \(r_E=0\), and contraction when \(r_E<0\). With recruitment retained, the full equation is affine in \(E\), so its trajectory is not generally a pure exponential.

In a time-varying infection, antigen and effector abundance influence one another: infection supplies antigen, and effectors remove infected cells, potentially reducing later stimulation. This feedback can produce an expansion phase followed by contraction without requiring a fixed, externally imposed activation schedule. The model must represent that feedback explicitly if its timing is central to the question.

A Reduced Model for Viral Peak Timing

The mechanistic infected-cell equation makes the route of CD8 action explicit: effectors increase infected-cell loss. For a transparent timing calculation, it is useful to work with a reduced approximation for viral burden during a period when target-cell depletion and other changing processes are limited. Write the effective viral growth equation as:

Here \(g_0\) is the effective per-day viral growth rate without CD8 suppression, and \(qE\) is the effective per-day suppression associated with the effector level. The product \(qE\) has units of inverse time. This is a reduced description: it approximates the combined consequences of infected-cell killing and virus production, rather than replacing the explicit \(I\) equation in every setting.

For an analytic timing example, assume no effective CD8 suppression before activation at time \(\tau\), and let effectors grow exponentially afterward: \(E(t)=E_0e^{r(t-\tau)}\) for \(t\geq\tau\), where \(E_0\) is the effector level at activation and \(r>0\) is the net expansion rate. Before activation, \(V(t)=V_0e^{g_0t}\). After activation, the instantaneous viral growth rate is \(g_0-qE(t)\). The viral peak occurs when this rate first reaches zero, provided it changes from positive to negative and no other process changes the growth rate first.

The logarithm follows by substituting \(E(t)=E_0e^{r(t-\tau)}\) into the zero-growth condition and solving for \(t\). To obtain the peak burden, integrate \(d\log V/dt=g_0-qE(t)\) from time zero to the crossing:

The first term is the viral growth that would accrue at rate \(g_0\) throughout the interval; the second term subtracts the accumulated CD8 suppression after activation. Omitting this suppression would correspond to a different model in which CD8 activity switches on abruptly only at the peak, not one in which effectors act while they expand.

Contraction and the Duration of Viral Suppression

When antigen falls, a simple contraction approximation is \(E(t)=E_s e^{-d_E(t-t_s)}\) after time \(t_s\), where \(E_s\) is the effector level at the start of contraction and \(d_E>0\) is the contraction rate. In the reduced viral model, the instantaneous growth rate then becomes \(g_0-qE_s e^{-d_E(t-t_s)}\). Viral burden declines only while \(qE(t)>g_0\). If \(qE_s>g_0\), that condition ends at \(t-t_s=\ln(qE_s/g_0)/d_E\). This describes the duration of negative instantaneous growth under the approximation; it does not by itself establish long-term control.

Assumptions, Diagnostics, and Alternatives

These calculations depend on stated simplifying assumptions. The analytic peak calculation assumes constant \(g_0\), an exponential effector trajectory after a specified activation time, and an effective suppression term proportional to \(E\). It also assumes that the viral burden remains well represented by a single effective state during the interval. If target cells become depleted, infected-cell stages produce a delay, or antigen and effectors substantially alter one another’s trajectories, \(g_0\) and the mapping \(qE\) may not remain constant or adequate. In those settings, solve the coupled \(T,I,V,E\) equations rather than applying the reduced peak formula.

Useful diagnostics include checking whether the observed viral slope is consistent with \(g_0-qE(t)\), whether measured effector abundance actually expands at an approximately constant log-rate, and whether the predicted peak occurs when the modeled net growth rate changes sign. Sparse viral sampling near the peak can obscure its timing, while measuring blood effectors alone may not represent their abundance in infected tissue. A fitted model that requires implausible expansion rates, or predicts viral decline before appreciable effector activity, signals possible model misspecification or an unmodeled mechanism.

A full mechanistic ODE model is preferable when target-cell depletion, infected-cell kinetics, or feedback from infection to antigen and effectors matters. The reduced model is useful for transparent timing calculations and for interpreting local growth slopes, but its parameters are effective quantities and may not separately identify CD8 killing from other causes of infected-cell loss. A hard-threshold alternative that assumes no suppression until the crossing would yield \(V_{\mathrm{peak}}=V_0e^{g_0t_{\mathrm{cross}}}\); that result is valid only for that abrupt-switch assumption. It is not valid when the stated model includes positive suppression during effector expansion.

The central interpretation is conditional: earlier activation lowers the peak in the worked reduced model because it starts suppression sooner, with all other model features held fixed. It does not prove that an intervention that changes activation timing will have that exact effect in a biological system. The result depends on the modeled effector kinetics, killing effectiveness, and viral growth processes.