Separating Target-Mediated Uptake From Linker Instability

Why the Three Time Courses Matter

A fall in measured intact antibody-drug conjugate (ADC) does not, by itself, say why the ADC disappeared. It may reflect target-mediated cellular uptake, or it may reflect linker cleavage or another form of ADC loss. Likewise, an increase in released payload does not uniquely indicate greater uptake: the linker may simply be cleaving faster. Separating these mechanisms matters because each points to a different source of altered intracellular drug exposure.

The useful evidence is the joint time course: intact ADC remaining outside the cells, drug still conjugated to antibody in the cells, and free payload in the cells. The model below treats these as linked amounts. It is an effective model for comparing conditions, not a claim that every intracellular trafficking step is represented explicitly.

Definition: Let \(E(t)\) be extracellular intact ADC concentration, in nM ADC; \(C(t)\) be intracellular conjugated-drug concentration, in nM payload equivalents; and \(P(t)\) be intracellular released-payload concentration, in nM payload equivalents. The effective uptake rate constant is \(k_u\), the baseline extracellular intact-ADC loss rate constant is \(k_0\), the linker-cleavage rate constant is \(k_{\mathrm{cl}}\), the non-cleavage loss rate constant for conjugated drug is \(k_C\), and the released-payload loss rate constant is \(k_P\). All rate constants have units of h\(^{-1}\). Let \(q\) be the mean number of payload equivalents delivered per ADC molecule.

A Minimal Joint Model

For a batch experiment with no ADC input after dosing, assume the effective first-order model

\frac{dE}{dt}=-(k_0+k_u)E,\qquad \frac{dC}{dt}=qk_uE-(k_{\mathrm{cl}}+k_C)C,\qquad \frac{dP}{dt}=k_{\mathrm{cl}}C-k_PP.

The first equation says that extracellular intact ADC disappears through baseline loss and uptake. The second says that uptake supplies conjugated payload to the cells, while cleavage and other conjugated-drug loss remove it. The third says that cleavage supplies free payload, which is itself lost or removed at rate \(k_P\). The factor \(q\) converts an ADC uptake flux into payload-equivalent units.

The model assumes that the effective rates are constant over the observation interval; that the assay distinguishes intact ADC, conjugated drug, and free payload; and that the reported concentrations represent comparable, properly normalized samples. It also assumes that the relevant payload entering \(C\) arrives with the ADC. Extracellularly released payload must be measured or controlled for, rather than silently attributed to intracellular cleavage.

Here, \(k_u\) is an effective uptake rate under the tested antigen density and dose. It need not remain constant at other doses or antigen densities. A matched target-negative condition or a target-blocking condition helps estimate baseline loss \(k_0\), provided that the comparison does not otherwise change ADC stability or cell handling. These controls are essential: an intact-ADC decline alone identifies \(k_0+k_u\), not the two rates separately.

How the Rates Are Separated

The intact-ADC equation has the solution \(E(t)=E(0)e^{-(k_0+k_u)t}\). Thus, if \(E\) is measured over time, the negative slope of \(\log E(t)\) estimates total disappearance. Subtracting \(k_0\), estimated in an appropriate control, gives \(k_u\). Conjugated drug then shows whether the inferred uptake flux is consistent with cellular delivery, while released payload identifies cleavage when payload loss is accounted for.

k_u=-\frac{d}{dt}\log E(t)-k_0,\qquad k_{\mathrm{cl}}=\frac{\frac{dP}{dt}+k_PP}{C}\quad(C>0),\qquad k_u=\frac{\frac{dC}{dt}+(k_{\mathrm{cl}}+k_C)C}{qE}\quad(E>0).

The last two expressions are rearrangements of the model equations. They show why the measurements complement one another: \(P\) informs cleavage only after correcting for payload loss, and \(C\) informs uptake only after correcting for cleavage and other conjugated-drug loss. In practice, noisy measurements make direct differentiation unstable. A joint fit to all time points is usually preferable; the equations clarify what the fit must learn and what the data need to contain.

Identifiability condition: Separating uptake from cleavage requires information about baseline extracellular loss, the payload conversion factor \(q\), and relevant loss rates such as \(k_P\) and \(k_C\), or enough data and experimental conditions to estimate them. If these quantities are unknown and the time courses are sparse, several rate combinations may fit similarly well. A fitted distinction is credible only if the data and controls constrain the rates separately.

Worked Example: Estimating Target-Dependent Uptake

In a hypothetical batch experiment, extracellular intact ADC falls from 100 nM to 45.0 nM over 4.00 h in target-positive cells. A matched target-negative control estimates baseline loss as \(k_0=0.0200\) h\(^{-1}\). Assume first-order disappearance and comparable baseline stability in the two conditions.

The total disappearance rate in the target-positive condition is

k_0+k_u=-\frac{\log(E(4)/E(0))}{4.00\ \mathrm{h}} =-\frac{\log(45.0/100)}{4.00\ \mathrm{h}} =\frac{0.7985}{4.00\ \mathrm{h}} =0.1996\ \mathrm{h}^{-1}.

Subtracting baseline loss gives

k_u=0.1996-0.0200=0.1796\ \mathrm{h}^{-1}.

For comparison, the target-negative ADC concentration is 92.3 nM at 4.00 h. Its estimated disappearance rate is \(-\log(0.923)/4.00=0.0200\) h\(^{-1}\), consistent with the baseline estimate and therefore with little detectable target-dependent uptake in that control. These calculations support greater effective uptake in the target-positive condition, but they do not establish the mechanism unless the matched control assumptions are reasonable.

Worked Example: Estimating Linker Cleavage From Payload Accumulation

In a hypothetical intracellular time course, \(C\) rises from 2.00 to 2.20 nM payload equivalents between 1.00 and 1.50 h. Over the same interval, \(P\) rises from 0.50 to 0.74 nM. Let \(k_P=0.100\) h\(^{-1}\), estimated independently from a payload washout condition. Approximate local derivatives and state concentrations by their midpoint values, and assume the change is sufficiently gradual for this approximation.

The midpoint values are \(C=2.10\) nM and \(P=0.62\) nM. The estimated payload accumulation rate is

\frac{dP}{dt}\approx\frac{0.74-0.50}{1.50-1.00} =0.480\ \mathrm{nM\,h}^{-1}.

From \(dP/dt=k_{\mathrm{cl}}C-k_PP\), the estimated cleavage rate constant is

k_{\mathrm{cl}}\approx \frac{0.480+(0.100)(0.62)}{2.10} =\frac{0.542}{2.10} =0.258\ \mathrm{h}^{-1}.

The positive correction \(k_PP\) is needed because some released payload is lost while it is being produced. Ignoring that loss would give \(0.480/2.10=0.229\) h\(^{-1}\), an underestimate under this model. The estimate is an interval-based approximation; a full time-course fit is more appropriate when measurements are noisy or rates change substantially within the interval.

Contrasting the Mechanistic Signatures

The simplest qualitative comparison is between how much conjugated drug is delivered and how rapidly it is converted to free payload. Increased uptake tends to increase the input \(qk_uE\) to \(C\). Increased cleavage raises the conversion rate \(k_{\mathrm{cl}}C\), tending to reduce \(C\) relative to its input while increasing the source of \(P\). The released-payload curve may still be affected by \(k_P\), so its height alone is not a measure of cleavage.

Worked Example: Comparing Uptake and Cleavage Changes at Controlled ADC Exposure

Consider hypothetical conditions with \(q=3.00\), \(k_C=0.100\) h\(^{-1}\), and \(k_P=0.200\) h\(^{-1}\). For this comparison, extracellular \(E\) is maintained at 1.00 nM by an input that balances its loss. This is a controlled-exposure condition: the input rate is \((k_0+k_u)E\), so \(E\) is constant even though ADC is continually being removed. The intracellular steady-state calculations below are conditional on that maintained \(E\).

In the reference condition, let \(k_u=0.0500\) h\(^{-1}\) and \(k_{\mathrm{cl}}=0.100\) h\(^{-1}\). The steady-state conjugated payload satisfies \(0=qk_uE-(k_{\mathrm{cl}}+k_C)C\), giving

C=\frac{(3.00)(0.0500)(1.00)}{0.100+0.100} =0.750\ \mathrm{nM}.

At steady state for released payload, \(0=k_{\mathrm{cl}}C-k_PP\), so \(P=(0.100)(0.750)/0.200=0.375\) nM.

For an uptake increase to \(k_u=0.150\) h\(^{-1}\), with cleavage unchanged, the required input maintains the same extracellular \(E\), while

C=\frac{(3.00)(0.150)(1.00)}{0.100+0.100} =2.25\ \mathrm{nM},\qquad P=\frac{(0.100)(2.25)}{0.200}=1.125\ \mathrm{nM}.

For a cleavage increase to \(k_{\mathrm{cl}}=0.300\) h\(^{-1}\), with uptake at its reference value, the corresponding steady states are

C=\frac{(3.00)(0.0500)(1.00)}{0.300+0.100} =0.375\ \mathrm{nM},\qquad P=\frac{(0.300)(0.375)}{0.200}=0.5625\ \mathrm{nM}.

In this example, the uptake change increases both intracellular conjugated drug and released payload, whereas the cleavage change reduces conjugated drug and raises released payload more modestly. The distinctions follow from the stated model and controlled exposure; they are not universal patterns when, for example, conjugated-drug loss or payload clearance also changes.

Diagnostics, Alternatives, and Limitations

A useful analysis checks whether the inferred rates explain all three curves at once. Examine whether the fitted model systematically misses early ADC depletion, the timing of the conjugated-drug peak, or the rise and decline of released payload. Such patterns may indicate time-varying uptake, more than one intracellular release step, recycling, or a changing cell-associated fraction. Residuals should be considered against time and assay range, not only summarized by a single fit statistic.

Check that the target-negative or blocking condition provides a defensible estimate of \(k_0\); different ADC stability between conditions can mimic uptake.
Check that \(C\) and \(P\) use the same payload-equivalent units and that the assumed \(q\) matches the ADC preparation. A changing drug-to-antibody ratio distribution can invalidate a single fixed conversion factor.
Include early samples to characterize ADC disappearance and conjugated-drug delivery, and later samples to observe payload accumulation or washout. If only one time point is measured, the rates will generally be difficult to distinguish.
Check assay recovery, cell normalization, and whether extracellular released payload contaminates the intracellular measurement. An assay that counts free payload as conjugated drug, or vice versa, defeats the intended separation.

Endpoint comparisons and direct mechanistic controls can complement, but not replace, a time-course model. An endpoint measurement of free payload conflates delivery, cleavage, and payload loss. Target blocking or antigen-negative cells can test whether ADC disappearance depends on target engagement. A cleavage-resistant linker control or a direct measure of linker products can test cleavage more directly. These approaches are valuable because they constrain different parts of the model; no single assay automatically resolves every ambiguity.

The effective first-order model is limited when uptake saturates over the tested concentration range, antigen abundance changes over time, ADC recycles, intracellular processing has important delays, or linker cleavage differs across compartments. If those effects matter, the model needs additional states or concentration-dependent rates and sufficient measurements to support them. The purpose of the minimal model is to make the key distinction explicit and show which observations are needed—not to force complex biology into constant rates.

Common Mistake

Pitfall: “Released payload is higher, so uptake must be higher.” That conclusion is not justified: a larger \(k_{\mathrm{cl}}\) can increase \(P\) even when uptake is unchanged, and a smaller \(k_P\) can also raise \(P\). Correct the reasoning by checking intact-ADC depletion and conjugated-drug delivery, then interpreting released payload with its production and loss terms. Similarly, a faster fall in extracellular ADC is not automatically target-mediated uptake; baseline loss must be estimated with a suitable control.

Key takeaway: Intact-ADC depletion, conjugated-drug accumulation, and released-payload formation constrain different fluxes. Attribute a change to uptake or linker cleavage only when the matched controls and joint time-course model support that distinction.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions below.

In a batch experiment, intact ADC has an estimated disappearance rate of 0.16 h\(^{-1}\), while a matched control estimates \(k_0=0.04\) h\(^{-1}\). What is the effective uptake rate?
If \(dP/dt=0.30\) nM h\(^{-1}\), \(P=0.40\) nM, \(k_P=0.10\) h\(^{-1}\), and \(C=1.00\) nM, estimate \(k_{\mathrm{cl}}\).
Why can a high released-payload concentration at a single time point not distinguish increased uptake from faster cleavage?
In the model, what additional information is needed to infer uptake from the conjugated-drug curve rather than relying on extracellular ADC depletion?

Solutions. (1) \(k_u=0.16-0.04=0.12\) h\(^{-1}\), assuming the control estimates the applicable baseline loss. (2) \(k_{\mathrm{cl}}=(dP/dt+k_PP)/C=[0.30+(0.10)(0.40)]/1.00=0.34\) h\(^{-1}\). (3) Released payload reflects both its production, \(k_{\mathrm{cl}}C\), and its loss, \(k_PP\); increased uptake can increase the available conjugated drug, while faster cleavage can increase conversion at unchanged uptake. (4) The conjugated-drug equation requires \(q\), \(E\), \(k_{\mathrm{cl}}\), and \(k_C\), in addition to the measured change in \(C\). Without those terms, the observed conjugated-drug trajectory does not uniquely identify \(k_u\).