Heterogeneous Receptor Populations and Biphasic Curves

Why a Concentration-Response Curve Can Have Two Transitions

A concentration-response curve is often summarized by a single transition from baseline toward a plateau. That description can be inadequate when more than one receptor population contributes to the measured response. If the populations differ in ligand affinity, their contributions may rise over different concentration ranges. If they also differ in coupling efficiency, one population may contribute much more response than its receptor abundance alone would suggest. The combined curve can then show two transitions, often perceived as a shoulder or as two regions of steep change.

Here, biphasic describes this two-transition shape as concentration increases. It does not mean that the response first rises and then falls. The model below assumes two populations whose responses add; it is a useful description of curve shape, not by itself proof that two distinct receptor types exist.

Definition: Let $C$ be the free agonist concentration and $E(C)$ the measured response at equilibrium. An additive model for two receptor populations is $$ E(C)=E_0+A_1\frac{C}{K_1+C}+A_2\frac{C}{K_2+C}. $$ Here $E_0$ is the response at zero agonist concentration, $K_j$ is the concentration at which the occupancy of population $j$ is one half under the one-site equilibrium binding assumption, and $A_j$ is that population’s maximum response contribution. The total response approaches $E_0+A_1+A_2$ as $C$ becomes large.

The parameters $A_1$ and $A_2$ summarize response contributions, not receptor fractions. They may reflect receptor abundance together with coupling and downstream signal gain. Thus, a population that is small in number can still make a substantial response contribution if it is efficiently coupled. This distinction is consistent with the earlier material on receptor reserve and response amplification and on operational model parameters.

Assumptions and a Short Derivation

The additive form follows from a specific set of assumptions. Each population binds agonist reversibly at equilibrium according to a one-site occupancy relation; both populations experience the same free concentration $C$; each population’s response contribution is proportional to its fractional occupancy; and the two contributions add to a common baseline. The model also treats the parameters as constant over the concentration range and assumes the measured response is not distorted by concentration-dependent assay artifacts. If downstream signaling is nonlinear or the populations interact, the equation need not describe the data.

For a single population, one-site equilibrium occupancy is $C/(K+C)$. If full occupancy of that population would add $A$ response units, its contribution at concentration $C$ is $A C/(K+C)$. Adding the baseline and the contributions from two populations gives the stated model. Its limiting values follow directly: at $C=0$, both occupancy terms are zero, so $E(0)=E_0$; as $C$ grows without bound, each fraction approaches one, so $E(C)$ approaches $E_0+A_1+A_2$.

The location and prominence of each transition can be understood by considering the slope with respect to log concentration. For one population,

\frac{d}{d\ln C}\left(A\frac{C}{K+C}\right) =A\frac{KC}{(K+C)^2}.

This slope is largest at $C=K$, where it equals $A/4$: differentiating $KC/(K+C)^2$ with respect to $C$ gives $K(K-C)/(K+C)^3$, which is positive for $C<K$, zero at $C=K$, and negative for $C>K$. For two populations, the total slope is the sum of their slopes. If $K_1$ and $K_2$ are far apart and each $A_j$ is large enough to matter, the two slope increases can be separated enough to produce a visible shoulder or two transitions. If the affinities are close, or one contribution is small, the summed curve may look like a single smooth transition instead.

Calculating a Two-Population Curve

Worked Example: Locating Two Transitions

Suppose a hypothetical assay has $E_0=5$ response units, $A_1=20$ response units, $K_1=0.3$ nM, $A_2=80$ response units, and $K_2=30$ nM. Calculate the response at concentrations around the two affinity scales and at a high concentration.

At $C=0.3$ nM, population 1 is half occupied and contributes $20(0.3)/(0.3+0.3)=10$ response units. Population 2 contributes $80(0.3)/(30+0.3)=0.7921$ response units. Therefore,

E(0.3)=5+10+0.7921=15.79\text{ response units}.

At $C=30$ nM, population 2 is half occupied. The contributions are $20(30)/(0.3+30)=19.80$ and $80(30)/(30+30)=40.00$ response units, giving

E(30)=5+19.80+40.00=64.80\text{ response units}.

At $C=300$ nM, the contributions are $20(300)/(300.3)=19.98$ and $80(300)/(330)=72.73$ response units. Thus,

E(300)=5+19.98+72.73=97.71\text{ response units}.

The asymptotic plateau is $5+20+80=105$ response units. The response at 300 nM is below, and approaching, that plateau. The low-affinity-scale transition is associated mainly with population 1, while population 2 contributes substantially over the higher concentration range.

The example illustrates why “two populations” does not necessarily mean two equally sized steps. The high-concentration transition is larger here because $A_2$ is larger. Nor does the observed response at any one concentration reveal how many receptors are present: it reflects occupancy and the response contribution parameters together.

When a Single-Site Model Misses Structure

A single-site proportional-response model has the form $E(C)=E_0+A C/(K+C)$. It has one affinity scale and one transition. It can be an adequate approximation when the two population affinities overlap, when one response contribution is negligible, or when data are too sparse or noisy to resolve separate transitions. A visually smooth curve is not, by itself, evidence for a single receptor population.

Conversely, a two-population model is worth considering when a single-site fit leaves systematic deviations across concentration: for example, it follows the lower-concentration rise but consistently misses a second rise at higher concentrations. Random scatter around a fitted curve is different from a repeated, structured residual pattern. The response range, concentration spacing, assay precision, and evidence that equilibrium was reached all affect whether the apparent structure is interpretable.

Worked Example: Testing a Single-Site Approximation

Use the hypothetical parameters in the previous example. The two-population model has baseline 5 response units, total possible increment $A_1+A_2=100$ response units, and gives an increment of $15.792-5=10.792$ response units at $C=0.3$ nM. Suppose a single-site curve is forced to have the same baseline and plateau and is chosen to match that low-concentration point. Find its $K$, then compare its prediction at 30 nM with the two-population response.

For the single-site curve, the increment at $0.3$ nM must satisfy $10.792=100(0.3)/(K+0.3)$. Solving for $K$ gives

K=0.3\left(\frac{100}{10.792}-1\right)=2.480\text{ nM}.

At $C=30$ nM, this single-site curve predicts an increment of $100(30)/(2.480+30)=92.36$ response units, or a total response of $97.36$ response units. The two-population model gives $64.80$ response units at the same concentration. The difference is large because a single affinity scale that matches the low-concentration response cannot also represent the later, distinct transition in this example.

This calculation demonstrates a limitation of that particular single-site approximation; it is not a statistical test and does not establish that real observations arise from two receptor types. Actual data have measurement error, and fitted parameters need not equal the hypothetical values. Curve shape alone supports considering heterogeneity, not a unique biological explanation.

Affinity and Coupling Both Shape the Curve

Worked Example: A Smaller Population Can Contribute More

Consider a second hypothetical assay with $E_0=5$ response units, $K_1=1$ nM, and $K_2=20$ nM. Suppose population 1 is half of the receptors but has a response contribution $A_1=10$ response units, while population 2 is also half of the receptors but has $A_2=90$ response units because its coupling is more efficient. Calculate the response at 1 nM and 20 nM.

At 1 nM, population 1 contributes $10(1)/(1+1)=5$ response units. Population 2 contributes $90(1)/(20+1)=4.286$ response units. Hence $E(1)=5+5+4.286=14.29$ response units.

At 20 nM, population 1 contributes $10(20)/(1+20)=9.524$ response units and population 2 contributes $90(20)/(20+20)=45.00$ response units. Therefore $E(20)=5+9.524+45.00=59.52$ response units. At saturation, the response approaches $5+10+90=105$ response units.

Although the receptor fractions are equal in this hypothetical setup, their maximum response contributions are not. The curve’s upper transition is dominated by population 2. Inferring receptor proportions directly from the size of curve segments would therefore be unjustified without additional information about coupling.

Diagnostics, Alternatives, and Limitations

A useful assessment begins with the concentration-response plot on a log concentration axis, including the individual observations and replicate variability. Concentrations should cover both the lower transition and the approach to the upper plateau; a sparse series can miss a shoulder or make one appear by chance. A candidate two-population curve should be checked for systematic lack of fit, plausible parameter values, and sensitivity to which concentrations are included. If the second response contribution is close to zero or the affinity scales are poorly separated, the parameters can be weakly determined even when a two-term equation can be fitted.

Interpretation also depends on experimental conditions. The model assumes free concentration is adequately represented by the value used in the analysis, the assay is at equilibrium, and the response is measured consistently across concentrations. Ligand depletion, time-dependent responses, solubility limits, assay saturation, or concentration-dependent nonspecific effects can create curve features that a two-population equation may appear to accommodate. These possibilities should be considered before attributing a shoulder to receptor heterogeneity.

The main alternatives answer different questions. A one-site occupancy model is the more parsimonious choice when one transition is adequate. A Hill equation can summarize a sigmoidal curve empirically, but a Hill coefficient different from one does not by itself identify multiple receptor populations. A single-receptor operational model allows the relationship between occupancy and response to be nonlinear; it may explain curve shapes that proportional occupancy-response models cannot. The two-population additive model is useful when distinct affinity scales and additive response contributions are plausible, but its fit does not uniquely establish that mechanism.

Ultimately, response data alone can be compatible with several biological and measurement mechanisms. A biphasic-looking curve is a reason to examine model adequacy and experimental conditions, not a definitive receptor count. The subsequent material on diagnosing receptor-theory models with binding and response data addresses how those sources of evidence can be considered together.

Common Pitfalls

Adding a response contribution incorrectly. The total is baseline plus each population’s occupied contribution. At $C=300$ nM in the first example, $5+19.98+72.73=97.71$, not 117.7 response units.
Calling every shoulder proof of two receptor types. The additive model can describe a two-transition shape, but downstream nonlinearity and experimental artifacts can also produce structured curves.
Equating response amplitude with receptor abundance. Each $A_j$ combines receptor capacity with coupling and response scaling; it is not a receptor fraction.
Assuming two populations must be visible as two steps. Similar affinities or a small contribution can make the sum look effectively monophasic.
Confusing a high-concentration response with the plateau. The plateau is a limiting value; a finite concentration generally gives a response below it under this model.

Key takeaway: Distinct receptor affinities can create separated transitions, while distinct coupling efficiencies determine how strongly each population contributes. A two-population curve is a mechanistic hypothesis with explicit assumptions—not a conclusion that can be read uniquely from curve shape.

Check Your Understanding

Answer each question, then compare with the worked solutions below.

For $E_0=2$, $A_1=12$, $K_1=1$ nM, $A_2=18$, and $K_2=10$ nM, calculate $E(1)$ and the asymptotic plateau.
Why might two receptor populations with different affinities fail to produce a visibly biphasic curve?
In the two-population model, what does $A_2$ represent, and why is it not necessarily the fraction of receptors in population 2?
A one-site curve fits the lower-concentration observations but consistently underpredicts the response at higher concentrations. What does this pattern suggest, and what does it not prove?

Solutions. (1) At 1 nM, population 1 contributes $12(1)/(1+1)=6$ response units and population 2 contributes $18(1)/(10+1)=1.636$ response units. Thus $E(1)=2+6+1.636=9.636$ response units. The plateau is $2+12+18=32$ response units. (2) The transitions may overlap if the affinities are not sufficiently separated, or one population may contribute too little response to distinguish from measurement variability. (3) $A_2$ is the maximum response contribution of population 2 in the model. It reflects receptor capacity and coupling, so response contribution need not be proportional to receptor fraction. (4) The systematic pattern suggests the single-site model may be inadequate and that another transition or mechanism should be considered. It does not prove the existence of a second receptor population, because other biological or assay processes can produce similar deviations.