A. We obtained binding thermodynamic parameters by ITC from 14 proteins, comprising 72 protein-ligand pairs. Using multiconformer models, we calculated estimates of protein and solvent conformational entropies to identify correlations between ITC binding entropy and structural estimates of entropy. B. Distribution of the thermodynamic binding contributions across all protein-ligand pairs ordered by entropy (n = 72). C. Correlation between protein conformational entropy as measured by crystallographic multiconformer structures and ITC-measured binding entropy. Each dot represents an individual protein-ligand complex, with each color representing a different protein (slope = 0.85, R² = 0.59). Higher binding entropy/protein conformational entropy indicates a more unfavorable binding entropy.

A. Estimates of differences in heat capacity with binding entropy (Slope = 0.020, R² = 0.066); B. Difference in the number of water molecules bound corrected for resolution and correlation with binding entropy (Slope = -0.021, R² = 0.11)

A. Correlation between predicted binding entropy values using estimates of heat capacity (R² = 0.59; TΔS = -1.044 + 0.891*[Protein Entropy] + 1.097*[change in heat capacity]) B. Correlation between predicted binding entropy values using change in number of bound water molecules corrected for resolution (R² = 0.69; (TΔS = -1.45 + 0.64×[Protein Entropy] - 3.8×[Number of Water]) C. Protein conformational entropy and solvent differences between apo TGT (PDB: 4PUN) and two ligand-bound structures with dramatically different changes in binding entropy. Both ligands pay an entropic penalty, with 4Q8T paying a lower entropic penalty, derived from a smaller reduction in side chain alternative conformers (observed based on color differences of stick representations in the first row) and a smaller difference in the number of water molecules in the bound structure. 4PUK has more water molecules bound per residue with corrected resolution, further explaining the larger entropic penalty.

A. Resolution distribution across all PDBs (Median: 1.30 [IQR: 1.08-1.60]). B. Resolution differences between paired PDBs (Median: 0.06 [IQR: -0.06 - 0.15]). C. Rfree distribution across all PDBs (Median: 0.16 [IQR: 0.15-0.19]). D. RFree differences between paired PDBs (Median: 0.010 [IQR: -0.010-0.027].

A. Distribution of protein size across the dataset by number of residue numbers. B. Alpha carbon RMSD between bound and unbound PDB pairs (Median: 0.27 [IQR: 0.19-0.46].

Ligand information

A. Molecular weight distribution across the dataset (Median: 284.68(102.46-461.50]) B. Hydrogen Bond Acceptor(Median: 4 [1–7]) C. Hydrogen Bond Donor (Median: 2 [0–5])

Correlation between ITC-measured binding entropy and protein conformational entropy metrics.

Each point represents the mean binding entropy from ITC experiments, with error bars indicating the range of all measurements. Linear regression parameters: slope = 0.88, intercept = 1.63, R2 = 0.61.

Distribution of change in SASA values across pairs

A. Total SASA changes (Median: 0.06, Mean: 0.05, Std: 0.62); B. Apolar SASA changes (Median: 0.16, Mean:-0.03, Std: 0.84); C. Polar SASA changes (Median: 0.07, Mean: 0.03, Std: 0.74)

Correlation of SASA changes with binding entropy

A. Total ΔSASA: R2=0.0058; B. apolar ΔSASA: R2=0.0174; C. polar ΔSASA: Rz=0.0027.

A. Distribution of average number of waters gained or lost between bound and unbound. B. Correlation between resolution differences and mean water count change per residue (R2=0.53, Slope=-1.232).

A. Distribution of average number of waters gained or lost between bound and unbound corrected for resolution differences between bound and apo. B. Correlation between resolution differences and mean water count change per residue (R2=0, Slope=O).

Contribution of solvent-accessible surface area components along with protein conformational entropy to experimental binding entropy.

A. Model including total SASA (R2 = 0.588). B. Model including apolar SASA (R2 = 0.589). C. Model including polar SASA (R2 =0.587).

Linear model relating non-corrected protein conformational entropy and non-corrected water count to the entropic binding (R2 = 0.681)

Hydrogen bond protein only

A. Correlation between the number of nodes and experimental entropy measurement (R2=0.0734) B. Correlation between the number of edges and experimental entropy measurement (R2=0.0489)

Packing and entropy measurement

A. Correlation between the difference in packing and protein conformational entropy estimates (R2=0.1976). B. Correlation between average number of waters gained or lost between bound and unbound with protein structural entropy estimates (R2=0.1318).

Hydrogen bond protein and solvent

A. Correlation between the number of nodes and experimental entropy measurement (R2=0.3765) B. Correlation between the number of edges and experimental entropy measurement (R2=0.2423)