Why Relate Antagonist Concentration to Dose Ratio?
A concentration-response curve measured in the presence of an antagonist may shift to the right: more agonist is needed to produce the same response. The size of that shift alone does not identify antagonist affinity. The Schild relationship makes a stronger prediction: for reversible competitive antagonism at equilibrium, the dose ratio depends linearly on antagonist concentration after a particular transformation. When the data support that prediction, the relationship can estimate antagonist affinity in a way that is less dependent on the agonist’s response-generating machinery.
The key evidence is not simply a rightward shift. It is a consistent relationship between antagonist concentration and the agonist dose ratio at matched responses, with a Schild slope compatible with one. A non-unit slope is a reason to investigate assumptions and data quality, not an invitation to report an unqualified affinity estimate.
Deriving the Schild Relationship
Let \(A\) denote free agonist concentration and \(B\) free antagonist concentration. Let \(K_A\) and \(K_B\) be their equilibrium dissociation constants at a single receptor site. Under reversible competition, equilibrium binding gives the fraction of receptors occupied by agonist as
Without antagonist, the agonist occupancy at concentration \(A_0\) is \(f_{A,0}=A_0/(A_0+K_A)\). With antagonist, the occupancy at agonist concentration \(A_B\) is \(f_{A,B}=A_B/[A_B+K_A(1+B/K_B)]\). For these occupancies to be equal, equate the odds of occupancy, \(f_A/(1-f_A)\):
Thus, the agonist concentration required for any fixed occupancy is multiplied by \(1+B/K_B\). If response is a stable, monotonic function of agonist occupancy or of the resulting stimulus, the same multiplicative shift applies at a matched response. Define the dose ratio \(DR\) as the agonist concentration producing a specified response in the presence of antagonist divided by the agonist concentration producing that same response without antagonist. Then:
Taking base-10 logarithms gives the Schild equation:
If concentrations are expressed in mol/L, define \(pA_2=-\log_{10}(K_B)\). When the Schild slope is one, the equation becomes \(\log_{10}(DR-1)=\log_{10}(B)+pA_2\). At \(DR=2\), \(\log_{10}(DR-1)=0\), so the antagonist concentration producing a twofold agonist dose ratio is \(K_B\), and its negative base-10 logarithm in mol/L is \(pA_2\).
Assumptions and Experimental Requirements
The algebra is exact for the stated equilibrium competition model. Interpreting a fitted Schild plot as antagonist affinity additionally requires the experiment to represent that model adequately:
- Reversible, equilibrium competition: agonist and antagonist compete at the same receptor site, and concentrations at the receptor have reached equilibrium during measurement.
- Known free antagonist concentrations: the \(B\) values used in the plot should represent relevant free concentrations, rather than nominal added concentrations when binding, depletion, or other losses are substantial.
- Comparable response conditions: response should be stable and measured consistently. The antagonist should not independently change the response being measured, and matched effects must be attainable in both conditions.
- Reliable dose ratios: control and antagonist curves should cover the response range used for matching. A dose ratio should not be inferred by extrapolating far beyond observed data.
- Appropriate time course: changing agonist or antagonist concentrations should not create appreciable kinetic disequilibrium, desensitization, or other time-dependent shifts during the measurements.
A monotonic response function and receptor reserve do not, by themselves, invalidate the derivation: a stable transduction relationship can preserve the matched-response shift. The important condition is that antagonist presence changes agonist access through the specified competition, rather than changing the transduction relationship or the measured response independently.
Worked Example: Calculating a Dose Ratio and Predicting a Shift
In a hypothetical equilibrium experiment, an agonist produces a response of 40 response units at \(3.0\ \mathrm{nM}\) without antagonist and at \(12.0\ \mathrm{nM}\) with \(9.0\ \mathrm{nM}\) antagonist. Suppose matched-response measurements are reliable and the curves are parallel with the same maximum. Calculate the dose ratio and the \(K_B\) implied by the competitive model.
The dose ratio compares agonist concentrations producing the same response:
Under the competitive relationship \(DR=1+B/K_B\), rearrange to \(K_B=B/(DR-1)\). Substitution gives
This is a single-concentration estimate conditional on the model assumptions; it is not yet a test of the Schild slope. In molar units, \(K_B=3.00\times10^{-9}\ \mathrm{M}\), so the implied \(pA_2=-\log_{10}(3.00\times10^{-9})=8.523\). Multiple antagonist concentrations are needed to assess whether the predicted relationship holds.
Estimating pA2 with a Schild Plot
For each antagonist concentration, estimate \(DR\) from agonist concentrations at the same response level, then calculate \(y=\log_{10}(DR-1)\) and \(x=\log_{10}(B\ [\mathrm{M}])\). Plot \(y\) against \(x\). A linear model \(y=a+bx\) is a useful summary: the competitive equilibrium prediction is \(b=1\). If the slope is compatible with one and the rest of the assumptions are credible, estimate pA2 from the intercept, or equivalently from the concentration where the fitted line crosses \(y=0\).
Worked Example: Estimating pA2 from a Unit-Slope Schild Plot
Consider hypothetical antagonist concentrations \(1.00\), \(10.0\), and \(100\ \mathrm{nM}\). Suppose the corresponding matched-response dose ratios are \(1.10\), \(2.00\), and \(11.0\). Assume the underlying curves support reliable matching and the Schild slope is compatible with one.
For \(B=1.00\ \mathrm{nM}=1.00\times10^{-9}\ \mathrm{M}\), \(DR-1=0.100\), so \(y=\log_{10}(0.100)=-1.000\) and \(x=-9.000\). For \(B=10.0\ \mathrm{nM}=1.00\times10^{-8}\ \mathrm{M}\), \(DR-1=1.00\), so \(y=0.000\) and \(x=-8.000\). For \(B=100\ \mathrm{nM}=1.00\times10^{-7}\ \mathrm{M}\), \(DR-1=10.0\), so \(y=1.000\) and \(x=-7.000\). The transformed points lie on \(y=x+8.000\), which has slope one and intercept 8.000.
The fitted line crosses \(y=0\) at \(x=-8.000\), corresponding to \(B=10^{-8}\ \mathrm{M}\). Therefore:
The conclusion is a model-based affinity estimate of \(K_B=10.0\ \mathrm{nM}\), or \(pA_2=8.000\), conditional on a unit-slope Schild relationship and the experiment’s equilibrium and competition assumptions. It is not a claim that the antagonist concentration was measured without error.
What a Non-Unit Slope Means
A Schild slope that differs materially from one weakens the interpretation of the intercept as a competitive antagonist affinity. It can reflect departures from the simple model, but also uncertainty in dose ratios, concentration errors, limited concentration range, or an unsuitable linear fit after transformation. The slope alone does not identify which explanation is correct.
A fitted line \(y=a+bx\) with \(b\ne1\) still describes the transformed data over the observed range. Its \(y=0\) crossing is \(x=-a/b\), so the corresponding concentration is \(10^{-a/b}\ \mathrm{M}\). However, labelling the negative log of that concentration as a definitive \(pA_2\) is not justified by the unit-slope derivation. It is at most an empirical twofold-shift concentration under that fitted line; the simple affinity interpretation requires further support.
Worked Example: Diagnosing a Non-Unit Slope
Suppose a hypothetical dataset gives antagonist concentrations \(1.00\), \(10.0\), and \(100\ \mathrm{nM}\), with dose ratios \(1.10\), \(1.562\), and \(4.162\). The transformed values are approximately \(x=-9,-8,-7\) and \(y=-1.000,-0.250,0.500\), respectively. Calculate the fitted slope and explain the implication.
The mean values are \(\bar{x}=-8\) and \(\bar{y}=-0.250\). The least-squares slope is
The intercept is \(a=\bar{y}-b\bar{x}=-0.250-(0.750)(-8)=5.750\). Thus the descriptive fit is \(y=5.750+0.750x\), with slope 0.750 rather than one. Its zero crossing is \(x=-5.750/0.750=-7.667\), or approximately \(2.15\times10^{-8}\ \mathrm{M}\). That crossing should not be reported as an established competitive \(pA_2\): the non-unit slope does not support the simple Schild affinity interpretation. The next step is to inspect the response curves, concentration and timing assumptions, and uncertainty in the dose ratios; these three points alone do not establish a cause.
Diagnostics, Alternatives, and Limitations
Evaluate a Schild analysis alongside the original concentration-response curves, not as a substitute for them. Check that matched responses are inside the observed range, that the dose ratios are stable across the chosen response levels, and that no single point dominates the fitted line. Examine uncertainty in dose ratios and in the slope; with few antagonist concentrations, a point estimate near one may be too imprecise to support a strong conclusion. Because \(DR-1\) is transformed logarithmically, values near \(DR=1\) are especially sensitive: small errors in \(DR\) can produce large changes in \(\log_{10}(DR-1)\). A zero or negative \(DR-1\) cannot be used in this transformation.
If the slope is compatible with one, analysts commonly estimate the intercept under a unit-slope model and report its uncertainty, along with the concentration range and experimental conditions. An unconstrained regression is useful for checking whether the slope departs from one, but a good-looking unconstrained line is not evidence that the mechanistic assumptions hold. Conversely, uncertainty that includes one does not prove those assumptions; it means the data have not clearly distinguished the slope from one.
| Approach | What it provides | Main caution |
|---|---|---|
| Single-concentration dose-ratio calculation | A conditional \(K_B=B/(DR-1)\) estimate at one antagonist concentration | Cannot assess the Schild slope or consistency across concentrations |
| Schild regression with slope estimated | A slope diagnostic and a descriptive relation across concentrations | A non-unit slope undermines the simple affinity interpretation |
| Schild regression with slope fixed at one | An affinity estimate using the competitive model’s predicted slope | Can conceal model mismatch if the unit-slope assumption is imposed without checking it |
The Schild transformation also changes the error structure: dose ratios are estimated quantities, and the log transformation can make their uncertainty uneven across concentrations. Ordinary least squares may therefore be an imperfect estimation method when dose-ratio uncertainty varies substantially. More detailed analysis can model the underlying concentration-response curves jointly, but that does not remove the need to check the biological assumptions. A Schild plot is a compact affinity tool, not a universal test of all mechanisms of antagonism.
Common Mistakes
- Using \(DR\) instead of \(DR-1\). The Schild ordinate is \(\log_{10}(DR-1)\), not \(\log_{10}(DR)\). The subtraction follows directly from \(DR=1+B/K_B\).
- Mixing concentration units. For \(pA_2=-\log_{10}(K_B)\), express the concentration in mol/L. A value in nM must be converted before taking the logarithm.
- Calling every intercept \(pA_2\). The intercept has that affinity interpretation when the slope is one and the competitive model is suitable. A non-unit slope calls for caution, not automatic relabelling.
- Treating a non-unit slope as a diagnosis. It is evidence that the simple relationship may not describe the data adequately; it does not identify the biological or experimental cause on its own.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- At \(5.0\ \mathrm{nM}\) antagonist, a matched-response dose ratio is 3.00. Under the competitive model, what \(K_B\) and \(pA_2\) are implied?
- Why is \(\log_{10}(DR-1)\), rather than \(\log_{10}(DR)\), used in a Schild plot?
- A fitted Schild slope is 1.04 with substantial uncertainty. What does this establish, and what does it not establish?
- A dose ratio is estimated as 0.95. Can its Schild ordinate be calculated? Explain.
Solutions. (1) \(K_B=B/(DR-1)=5.0\ \mathrm{nM}/2.00=2.50\ \mathrm{nM}=2.50\times10^{-9}\ \mathrm{M}\). Hence \(pA_2=-\log_{10}(2.50\times10^{-9})=8.602\), conditional on the model. (2) The competitive relationship is \(DR=1+B/K_B\); subtracting one yields the quantity proportional to antagonist concentration, \(DR-1=B/K_B\), whose logarithm is linear in \(\log_{10}(B)\). (3) The estimate is near one, but substantial uncertainty means the data may not distinguish it from other slopes. It neither proves the competitive assumptions nor identifies the mechanism. (4) No. Since \(DR-1=-0.05\), its logarithm is not defined over the real numbers; inspect the curve and dose-ratio estimation rather than forcing a Schild value.