Worked Example: Endosomal Sequestration With Finite Membrane Flux

Why Follow Endosomal Concentration Separately?

A tissue model that combines all intracellular drug into one compartment cannot describe a delay between cellular entry and accumulation in endosomes. When both membranes have finite permeability, the extracellular, intracellular, and endosomal concentrations can change at different rates. Following each concentration separately makes that timing visible and lets us check that the modeled transfers do not create or destroy drug.

The example below is deliberately closed and simple: drug moves only between three well-mixed spaces. It has no blood input, elimination, metabolism, or irreversible binding. Thus, any change in one compartment must be balanced by changes in the others. The model illustrates finite-rate endosomal sequestration; it does not represent every biological process that can influence endosomal concentration.

Set Up the Three-Compartment Model

The amount balances follow directly from the two membrane fluxes. The extracellular compartment loses the amount crossing its membrane; the intracellular compartment gains that flux and loses the flux crossing into the endosome; and the endosomal compartment gains the latter flux.

These equations assume constant volumes, spatially well-mixed concentrations within each compartment, fixed conductances, and flux proportional to the concentration difference across each membrane. The same freely dissolved drug species is being tracked on both sides of each membrane, so equal concentrations imply zero net flux in this example. There is no external input or loss. If transport depended on a partition coefficient, saturable binding, pH gradients, or active transport, the flux law would need to represent that mechanism rather than simply using these concentration differences.

The equations are in amounts, which makes their bookkeeping explicit. Dividing each amount balance by its compartment volume gives concentration equations; the volumes must be included unless they are all equal. In the numerical example, all three volumes are \(1\ \mathrm{L}\), which lets amounts in milligrams and concentrations in milligrams per litre have the same numerical values.

Derive the Concentration-Time Courses

With all three volumes equal to \(1\ \mathrm{L}\) and both conductances equal to \(1\ \mathrm{L/h}\), the amount balances become the following concentration system. Time is measured in hours, so the numerical rate constants in the system are per hour.

To solve the system, the matrix has three independent modes. The vector \([1,1,1]^T\) has eigenvalue \(0\): equal concentrations produce no flux and describe the equilibrium mode. The vectors \([1,0,-1]^T\) and \([1,-2,1]^T\) have eigenvalues \(-1\) and \(-3\), respectively, as can be checked by multiplying each vector by the matrix. Therefore, a solution is a constant mode plus terms proportional to \(e^{-t}\) and \(e^{-3t}\).

Write the initial vector as a sum of these modes: \([6,0,0]^T=2[1,1,1]^T+3[1,0,-1]^T+[1,-2,1]^T\). The coefficients reproduce the initial values in all three compartments: the first component is \(2+3+1=6\), the second is \(2-2=0\), and the third is \(2-3+1=0\). Applying the corresponding exponential to each nonconstant eigenmode gives the concentration courses.

As \(t\) increases, both exponential terms decay to zero, so all three concentrations approach \(2\ \mathrm{mg/L}\). This is consistent with the closed model: the total amount is \(6\ \mathrm{mg}\), the combined volume is \(3\ \mathrm{L}\), and equal concentrations at equilibrium must therefore be \(6/3=2\ \mathrm{mg/L}\).

Check Conservation and Interpret the Delay

The conservation result follows from the amount balances themselves, not from a special feature of the numerical solution. Adding the three derivatives cancels both internal fluxes:

Thus \(A_e+A_i+A_z\) is constant in time. In this example the volumes are all \(1\ \mathrm{L}\), so the amount sum at one hour can also be checked by summing the numerical concentrations and multiplying each by its volume: \(3.153425392+1.900425863+0.9461487448=6.000000000\ \mathrm{mg}\), up to rounding. This matches the initial \(6\ \mathrm{mg}\).

The delay in the endosomal profile is a consequence of the two sequential finite-rate transfers. At time zero, the first membrane has a concentration difference of \(6\ \mathrm{mg/L}\), but the second has none. As intracellular concentration rises above endosomal concentration, the second membrane develops a positive inward flux. The endosomal concentration therefore responds to the evolving intracellular concentration rather than being forced to match it immediately.

Assumptions, Diagnostics, and Alternatives

The profiles are useful only to the extent that the assumptions are reasonable. A model with constant volumes and conductances predicts smooth, linear exchanges: doubling the concentration difference doubles the flux. It does not include finite binding capacity, pH-dependent ion trapping, nonlinear transport, or changes in membrane area. Nor does the closed system represent the continued delivery of drug from blood or its removal by metabolism and clearance.

Several checks can reveal mistakes or a poor fit to observations. Concentrations should start at the specified initial conditions; in this example they should remain nonnegative and approach a common equilibrium value. The total amount should remain \(6\ \mathrm{mg}\) in a closed simulation. A computed increase in total amount, a negative concentration, or endosomal accumulation before there is a driving concentration difference suggests an implementation or assumption problem. With measured profiles, systematic delays or plateaus inconsistent with the predicted curves may indicate that this simple model does not capture the governing processes; the profiles alone do not prove that a particular biological mechanism is present.

Finite-rate exchange differs from the main simplifying alternative, instantaneous equilibrium partitioning. An equilibrium approximation constrains compartments to their partition relationship at every time, eliminating the transient difference across the membrane. It can be appropriate when exchange is much faster than the timescale of interest, but it cannot represent the sequential lag shown here. At the other extreme, a perfusion-limited tissue model treats tissue equilibration with blood as rapid; it does not resolve these two internal barriers. The three-compartment finite-flux model is more explicit, but requires additional conductance and volume information and can be difficult to distinguish from alternative mechanisms using sparse concentration data.