1. Protein-Protein Interactions
Biochemistry 412
Protein-Protein Interactions
February 22, 2005

2. Macromolecular Recognition by Proteins
• Protein folding is a process governed by intramolecular recognition.
• Protein-protein association is an intermolecular process.
Note: the biophysical principles are the same!

3. Special Features of Protein-Protein Interfaces
• Critical for macromolecular recognition
• Typically, ca Å2 of surface buried upon complex
formation by two globular proteins
• Epitopes on protein surface thus may have a “hybrid” character,
compatible with both a solvent-exposed (‘free”) state
and a buried, solvent-inaccessible (“bound”) state
• Energetics of binding primarily determined by a few critical residues
• Flexibility of surface loops may be quite important for promoting
“adaptive” binding and for allowing high specificity interactions
without overly-tight binding (via free state entropy effects)
• Most contacts between two proteins at the interface involve
amino acid side chains, although there are some
backbone interactions

4. Formalisms for Characterizing Binding Affinities
For a protein (P), ligand (A), and complex (P • A)
P + A P • A
where [P]total = [P] + [P • A] ka kd
The association constant: Ka = [P • A]/[P][A] = ka/kd
The dissociation constant: Kd = 1/Ka = [P][A]/[P • A]
…please note that Kd has units of concentration, and so when Kd = [A]
then [P] = [P • A], and thus Kd is equal to the concentration of the
ligand A at the point of half-maximal binding.

5. At a given ligand concentration [A] the free energy of binding,
in terms of the difference in free energies between the free
and the bound states, can be described as
DG°binding = -RT ln ([A]/Kd)
It is also often useful to describe the difference in binding affinity
between a wild type protein and a mutant of the same protein,
which is an intrinsic property independent of the ligand
concentration. In that case we can express this as
DDG°wt-mut = -RT ln (Kdmut/Kdwt)

6. Mapping Antigen-Antibody Interaction Surfaces (Binding Epitopes)
Using Hydrogen-Deuterium Exchange and
NMR Spectroscopy

11. Mapping Protein-Protein Interactions
Using Alanine-Scanning Mutagenesis

12. “If amino acids had personalities, alanine
would not be the life of the party!”
- George Rose
Johns Hopkins Univ.

13. Auguste Rodin
The Kiss
1886 (100 Kb); Bronze, 87 x 51 x 55 cm; Musee Rodin, Paris

20. Clackson et al (1998) J. Mol. Biol. 277, 1111.

21. Most mutations that markedly affect the binding affinity
(Ka) do so by affecting the off-rate (kd or koff).
In general, mutational effects on the on-rate (ka or kon)
are limited to the following circumstances:
• Long-range electrostatic effects (steering)
• Folding mutations masquerading as affinity mutations
(i.e., mutations that shift the folding equilibrium to
the non-native [and non-binding] state)
• Inadvertent creation of alternative binding modes
that compete with the “correct” binding mode

22. Cunningham & Wells (1993) J. Mol. Biol. 234, 554.

23. Cunningham & Wells (1993) J. Mol. Biol. 234, 554.

24. Clackson et al (1998) J. Mol. Biol. 277, 1111.

27. Reference Molecule: Turkey Ovomucoid Third Domain
(a Serine Protease Inhibitor)
• All nineteen possible
amino acid substitutions
were made for each of
the residues shown in
blue (total = 190).
• For each inhibitor,
binding constants
were measured precisely
for each of six different
serine proteases.
• X-ray structures were
performed on a subset
of the mutant complexes.

28. Structure of the complex to TKY-OM3D P1 Pro with
Streptomyces griseus Protease B
Bateman et al (2001) J. Mol. Biol. 305, 839.

29. The Principle of Additivity
Consider the double mutant, AB, composed of mutation
A and mutation B. In general (but not always -- see below),
the binding free energy perturbations caused by single mutations
are additive, in other words
DDG°wt-mutAB = DDG°wt-mutA + DDG°wt-mutB + DDG°i
where DDG°i ≈ 0.
DDG°i has been termed the “interaction energy” (see (Wells
[1990] Biochemistry 29, 8509). If DDG°i ≠ 0, then mutations A and
B are said to be nonadditive and it can therefore be inferred that
the two residues at which these mutations occur must physically
interact, directly or indirectly, in the native structure.
Note: this has important implications regarding
how evolution shapes proteins.

30. Qasim et al (2003)
Biochemistry 42, 6460.

31. …and the theorists are now beginning to mine this data
to refine their docking programs.
“Bad” prediction
“Good” prediction
Lorber et al (2002) Protein Sci. 11, 1393.

32. If you want to be “hard core” and really
understand protein-protein interactions,
you need to know more than just the
free energies of association. You (ultimately)
will need to know something about enthalpies,
entropies, and heat capacities, too.

33. Makarov et al (1998) Biopolymers 45, 469.

34. Makarov et al (2000) Biophys. J. 76, 2966.

35. Makarov et al (2002) Acc. Chem. Res. 35, 376.

36. When two proteins form a complex, solvent
must be displaced from the interfacial regions
and the conformational freedom (configurational
entropy) of the main chain and side chain atoms
will change also.

37. Jelesarov and Bosshard (1999) J. Molec. Recognition 12, 3.

38. Jelesarov and Bosshard (1999) J. Molec. Recognition 12, 3.

39. Isothermal Titration Calorimetry Yields DH of Binding

40. …and when you have DH and DG
(= -RTlnKa), you can calculate DS.

41. Some examples of experimentally-measured thermodynamic quantities for
interacting proteins, measured using isothermal titration calorimetry:
Note: isothermal titration calorimetry also directly yields n, the stoichiometry of binding.
Weber and Salemme (2003) Curr. Opin. Struct. Biol. 13, 115.