Why Occupancy Curves May Not Be Hyperbolic
For a single ligand binding to a single class of independent sites at equilibrium, fractional occupancy follows a hyperbolic concentration-response curve. A plot of occupancy against free ligand concentration rises gradually and approaches saturation. Some binding data instead rise more sharply around half occupancy. A Hill-type model represents that non-hyperbolic shape and provides a compact way to describe its steepness.
The model is useful for asking two separate questions: at what free concentration is occupancy one half, and how rapidly does occupancy change around that point? Those features are controlled by different parameters. A horizontal shift in the curve is not the same as a change in its steepness, and neither parameter by itself proves a particular molecular binding mechanism.
The model assumes that the measured quantity can be interpreted as fractional occupancy, that equilibrium has been reached, and that the concentration used is the free concentration available to bind. It also assumes that a single Hill curve adequately summarizes occupancy over the concentration range of interest. If the assay signal is not already a fraction, it must first be related to occupancy using appropriate baseline and saturation values; otherwise a signal change may be mistaken for an occupancy change.
The Hill model is a shape model, not a full molecular description. Its use does not require that every binding event follow a literal concerted mechanism. Conversely, observing a steep curve is not by itself proof that binding is cooperative. Experimental design, concentration measurement, and the adequacy of the assumed occupancy scale matter to that interpretation.
What the Parameters Mean
Set \(C=K_{0.5}\) in the model. The numerator and the first term in the denominator are then equal, so \(\theta=1/2\). Thus \(K_{0.5}\) locates the midpoint of the curve. A lower \(K_{0.5}\) means that half occupancy is reached at a lower free concentration; it is often described as greater apparent affinity. For a Hill curve with \(n_H\ne1\), \(K_{0.5}\) should not automatically be identified with a microscopic dissociation constant.
The Hill coefficient controls steepness. If \(n_H=1\), the equation reduces to the hyperbolic occupancy model. Values \(n_H>1\) give a steeper transition around \(K_{0.5}\), a pattern often described as positive cooperativity. Values \(0<n_H<1\) give a shallower transition, sometimes described as negative cooperativity. These are descriptions of the curve’s shape. The coefficient is not generally the number of ligand-binding sites; inferring a molecular stoichiometry requires stronger mechanistic assumptions than fitting this empirical equation.
Deriving the Log-Odds Form
The model’s interpretation becomes clearer by comparing occupied and unoccupied fractions. For \(0<\theta<1\), the unoccupied fraction is \(1-\theta\). Dividing the Hill equation by its complement gives the occupancy odds:
Taking natural logarithms yields a straight-line relationship between log odds and log concentration:
Consequently, on a log-concentration scale, \(n_H\) is the slope of log occupancy odds against log free concentration. The intercept is \(-n_H\log(K_{0.5})\). This derivation also shows why midpoint and steepness are distinct: the midpoint fixes where the log-odds line crosses zero, while the slope determines how rapidly it moves through zero.
A local slope on the original occupancy scale provides another interpretation. Differentiating with respect to \(\log C\), and using \(dC/d\log C=C\), gives
At half occupancy, this becomes \(n_H/4\). Thus a larger Hill coefficient means a larger change in occupancy for a given small change in log concentration near the midpoint. The formula describes a local slope, not a constant change in occupancy per concentration unit across the entire curve.
Worked Example: Comparing Hyperbolic and Steeper Occupancy
Consider hypothetical equilibrium binding curves with \(K_{0.5}=1.00\) nM. Compare a hyperbolic model with \(n_H=1\) to a steeper Hill model with \(n_H=2\), at free concentrations of 0.500, 1.00, and 2.00 nM.
At \(C=0.500\) nM, the hyperbolic model gives \( \theta=0.500/(1.00+0.500)=0.333 \), or 33.3% occupancy. The \(n_H=2\) model gives \( \theta=0.500^2/(1.00^2+0.500^2)=0.250/1.250=0.200 \), or 20.0%.
At \(C=1.00\) nM, both models give \( \theta=1.00/(1.00+1.00)=0.500 \), or 50.0%, because this concentration equals \(K_{0.5}\).
At \(C=2.00\) nM, the hyperbolic model gives \( \theta=2.00/(1.00+2.00)=0.667 \), or 66.7%. The \(n_H=2\) model gives \( \theta=2.00^2/(1.00^2+2.00^2)=4.00/5.00=0.800 \), or 80.0%.
Both curves have the same midpoint, but the \(n_H=2\) curve moves more sharply from low to high occupancy near that midpoint. These calculations illustrate a shape difference; they do not establish a molecular mechanism.
Estimating the Shape From Occupancy Measurements
The log-odds derivation allows a direct calculation when two idealized occupancy measurements are given. For two concentrations \(C_1,C_2\) and corresponding fractional occupancies \(\theta_1,\theta_2\), the model implies
This calculation requires occupancies strictly between 0 and 1, since their log odds are otherwise undefined. With real data, estimates from only two observations are sensitive to measurement error. Multiple concentrations and an assessment of fit are needed before treating an estimated slope as a reliable description.
Worked Example: Calculating a Hill Coefficient and Midpoint
Suppose hypothetical equilibrium measurements give \(\theta_1=0.200\) at \(C_1=0.500\) nM and \(\theta_2=0.800\) at \(C_2=2.00\) nM. Calculate the Hill coefficient and \(K_{0.5}\), assuming the measurements follow a Hill curve.
The odds at the first concentration are \(0.200/(1-0.200)=0.250\). At the second they are \(0.800/(1-0.800)=4.00\). Therefore, \( n_H=\log(4.00/0.250)/\log(2.00/0.500)=\log(16)/\log(4)=2.00 \).
To find \(K_{0.5}\), use the first observation and the odds equation: \(0.250=(0.500/K_{0.5})^{2.00}\). Taking the positive square root gives \(0.500=0.500/K_{0.5}\), so \(K_{0.5}=1.00\) nM. Substitution at the second point checks the result: \((2.00/1.00)^2=4.00\), matching the observed odds. Under the assumed model, the fitted midpoint is 1.00 nM and the fitted Hill coefficient is 2.00.
Separating a Midpoint Shift From a Slope Change
A change in \(K_{0.5}\) shifts the curve along the concentration axis: it changes the concentration at which occupancy is one half. A change in \(n_H\) changes the steepness around that midpoint. To distinguish these possibilities, measurements should include concentrations on both sides of the midpoint as well as observations near it. A measurement only at half occupancy identifies the midpoint but contains no information about the Hill coefficient.
This distinction is especially important when describing affinity. If two curves have the same \(n_H\) but different \(K_{0.5}\), their midpoints differ. If they share \(K_{0.5}\) but have different \(n_H\), their midpoints agree while their shapes differ. With sparse data, however, a midpoint shift and a slope change can produce similar predictions over a limited concentration range.
Worked Example: Comparing Affinity and Steepness Changes
Compare three hypothetical models: A has \(K_{0.5}=2.00\) nM and \(n_H=1\); B has \(K_{0.5}=2.00\) nM and \(n_H=2\); C has \(K_{0.5}=4.00\) nM and \(n_H=1\). Calculate occupancy at 1.00, 2.00, and 4.00 nM.
At \(C=1.00\) nM, A gives \(1/(2+1)=0.333\). B gives \(1^2/(2^2+1^2)=1/5=0.200\). C gives \(1/(4+1)=0.200\). A single observation at this concentration would not distinguish B from C.
At \(C=2.00\) nM, A gives \(2/(2+2)=0.500\), B gives \(2^2/(2^2+2^2)=0.500\), and C gives \(2/(4+2)=0.333\). A and B share a midpoint, while C has a midpoint at 4.00 nM.
At \(C=4.00\) nM, A gives \(4/(2+4)=0.667\), B gives \(4^2/(2^2+4^2)=16/20=0.800\), and C gives \(4/(4+4)=0.500\). B differs from A through a steeper curve with the same midpoint; C differs through a rightward midpoint shift with the same slope parameter. Concentrations on both sides of the midpoint make the distinction clearer than a single measurement does.
Diagnostics, Alternatives, and Limits
A fitted Hill curve should be checked against the observations, not accepted simply because it produces a plausible coefficient. Useful checks include plotting occupancy against log free concentration, comparing observed and predicted occupancies across the full range, and inspecting residuals for systematic patterns. The measurements should include enough of the rising portion of the curve to inform both midpoint and steepness; data only far below or far above the midpoint may provide little information about one or both parameters.
- Check the occupancy scale. Predictions must lie between 0 and 1 and should approach the expected lower and upper limits. A signal that has not been properly converted to fractional occupancy may violate this interpretation.
- Check concentration and equilibrium assumptions. The model uses free concentration at equilibrium. Using total concentration when binding materially depletes free ligand, or using measurements taken before equilibrium, can distort the apparent curve.
- Look for systematic residuals. If deviations are consistently positive in one concentration region and negative in another, the Hill curve may not represent the observed shape adequately.
- Check parameter precision. A narrow concentration range, few observations near the midpoint, or substantial measurement error can leave \(K_{0.5}\) and \(n_H\) poorly determined. A numerical estimate alone does not guarantee that the two parameters have been separated by the data.
The hyperbolic model is the special case \(n_H=1\). It is simpler and is appropriate when that shape adequately describes the data. The Hill model adds a steepness parameter and is useful when a non-hyperbolic curve is supported by measurements. A more detailed mechanistic binding model may be preferable when molecular assumptions and suitable data justify describing binding steps explicitly. The Hill equation is often more compact, but its parameters should not be assigned mechanistic meanings that the model does not establish.
The Hill model is limited to the occupancy pattern it represents and the conditions under which the concentration and occupancy measurements are meaningful. It does not by itself explain why a curve is steep, establish the number of sites, or ensure that a single curve adequately summarizes all observations. Its main value is disciplined description: it separates midpoint from slope and makes the consequences of each parameter explicit.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- For a Hill curve with \(K_{0.5}=3.00\) nM and \(n_H=2\), what is the occupancy at \(C=3.00\) nM?
- For \(K_{0.5}=2.00\) nM and \(n_H=1\), calculate occupancy at \(C=1.00\) nM.
- In the Hill model, what changes when \(K_{0.5}\) increases while \(n_H\) is held constant?
- Why does an estimated \(n_H=2.00\) not, by itself, demonstrate two binding sites?
Solutions. 1. Substituting \(C=K_{0.5}=3.00\) nM gives \(\theta=3.00^2/(3.00^2+3.00^2)=9/18=0.500\), or 50.0%. 2. \(\theta=1.00/(2.00+1.00)=0.333\), or 33.3%. 3. The concentration for half occupancy increases, shifting the curve toward higher concentrations; the slope parameter and curve steepness characterization remain unchanged. 4. The Hill coefficient is the slope of log occupancy odds against log concentration under the model. It describes curve steepness and is not generally a direct count of binding sites.