Quaternary Structure and Subunit Assembly

Quaternary assembly chemistry underlies entire classes of modern therapeutics and diagnostics. Monoclonal antibody drugs (IgG) rely on precisely engineered disulfide-linked heavy-heavy and heavy-light chain interfaces; bispecific antibody engineering deliberately re-designs the heavy-chain $C_H3$-$C_H3$ dimer interface ("knobs-into-holes" mutations, introducing a bulky Trp "knob" on one chain and a complementary cavity on the other) to force heterodimerization instead of the thermodynamically preferred homodimer, a direct engineering application of the buried-surface-area energetics developed above. Insulin's storage form is a zinc-coordinated hexamer (three isologous dimers arranged with 32-point-group symmetry) — a quaternary trick that keeps the hormone inertly packaged in the pancreatic $\beta$-cell secretory granule until dilution upon secretion drives dissociation to the active monomer, a design principle now exploited in rapid-acting insulin analogs engineered to favor faster hexamer-to-monomer dissociation. Viral capsid assembly is a direct application of the closed-symmetry mathematics in Derivation 2, generalized by Caspar and Klug to icosahedral "quasi-equivalence," and is now a major target for capsid-assembly-disrupting antivirals (e.g., HIV maturation inhibitors that block the correct quaternary rearrangement of Gag protein). The bacterial ribosome's 30S and 50S subunit interfaces are targeted by numerous clinical antibiotics (aminoglycosides, macrolides) that jam subunit-subunit or subunit-tRNA quaternary contacts rather than any single active site. Proteasome assembly chaperones, GroEL/GroES heptameric ring closure, and the AAA+ ATPase hexameric unfoldases all rely on the same closed-ring energetics; single-particle cryo-EM has, in the last decade, made it routine to directly visualize these symmetric and pseudo-symmetric quaternary architectures at near-atomic resolution, turning what Perutz spent twenty years extracting from one crystal into a now-routine structural determination.

Ask the tutor: How does the T-to-R quaternary transition in hemoglobin actually move atoms at the α1β2 interface, step by step?

A folded polypeptide chain presents a closed, largely hydrophobic core to itself and a mixed polar/hydrophobic surface to solvent. If two such folded chains happen to have complementary patches of exposed hydrophobic residues, bringing those patches together to exclude water is thermodynamically favorable — exactly the hydrophobic-effect logic that drives folding in the first place, now operating between molecules instead of within one. This is why quaternary interfaces are frequently described as "the same forces as the hydrophobic core, turned inside out and shared between two chains." Three points from tertiary-structure chemistry carry over unchanged. First, shape complementarity: interfaces bury surface with few voids, meaning side chains interdigitate almost as tightly as in a folded core (packing density near 0.72–0.75, close to the Kihara packing limit for spheres). Second, specific polar contacts at the rim of the interface — hydrogen bonds and salt bridges — that are not the main source of binding energy (water could have made similar bonds) but provide specificity: they are why subunit A recognizes subunit B and not some other similarly hydrophobic protein floating in the cytoplasm. Third, conformational strain: subunits are often not in their lowest-energy tertiary conformation when isolated in solution; binding to a partner can pull a subunit into a slightly higher-energy shape, "pre-loading" it with strain energy that is released, or re-distributed, when a ligand binds elsewhere in the complex. This strain-storage is the physical currency of cooperativity. Symmetry is not an aesthetic afterthought — it is a direct consequence of the fact that most oligomers are built from identical subunits encoded by one gene. If subunit A binds subunit B through one interface, and A and B are chemically identical, then by the isologous/heterologous distinction (Monod, Wyman, Changeux, 1965) the same interface chemistry appropriately dictates that A must also present a complementary surface to a second copy of itself, and the simplest way to close this ring of identical contacts without leaving a dangling, un-matched surface is a closed rotational symmetry group. This is why a dimer almost always has a two-fold ($C_2$) axis, a tetramer often has $222$ ($D_2$) symmetry, and a viral capsid can display 5-3-2 icosahedral symmetry.

The stability of an oligomer is a standard free energy of association. For a dimer formed from monomers M,

where:

• $K_d$ = dissociation constant (M), the concentration of free monomer at which half the protein is dimeric
• $[\text{M}]$, $[\text{M}_2]$ = equilibrium concentrations of monomer and dimer

The associated standard free energy is

where:

• $R = 8.314\ \text{J mol}^{-1}\text{K}^{-1}$, $T$ in kelvin
• $K_a = 1/K_d$, the association constant (M$^{-1}$)

This free energy decomposes into physically distinct terms, the working "energy budget" of an interface:

where:

• $\Delta G_{\text{hydrophobic}} \approx -\gamma \,\Delta A_{\text{nonpolar}}$, with empirical $\gamma \approx 20$–$25\ \text{cal mol}^{-1}\text{Å}^{-2}$ (roughly the same surface-tension coefficient used for hydrophobic folding)
• $\Delta A_{\text{nonpolar}}$ = nonpolar solvent-accessible surface area buried at the interface (Ų)
• $\Delta S_{\text{conform}}$ = conformational entropy lost by side chains immobilized at the interface (unfavorable, i.e. $\Delta S<0$ so $-T\Delta S>0$)
• $\Delta S_{\text{translation/rotation}}$ = loss of independent translational and rotational freedom when two molecules become one — the single largest unfavorable term, roughly 15–25 kcal/mol per binding event at 298 K for typical protein-sized partners, largely independent of interface size

Point-group symmetry gives the combinatorial/geometric relationship that fixes how many subunits close into a ring:

where:

• $n$ = number of subunits related by a single rotation axis
• $\theta$ = rotation angle per subunit-subunit step imposed by the interface geometry

And the operational readout of interface quality, buried surface area from crystal structures:

where each $A$ is solvent-accessible surface area (SASA, Ų) computed by rolling a 1.4 Å probe sphere over the structure; typical protein-protein interfaces bury 800–2000 Ų of total surface (each partner contributing roughly half).

Reading these equations: the first two say that oligomerization is an ordinary binding equilibrium, no different in form from ligand binding to a receptor — subunit association is a bimolecular binding reaction, just between two folded proteins instead of a protein and a small molecule. The energy budget equation says stability is a tug-of-war: burying nonpolar area and making van der Waals contacts pays for the steep entropic cost of losing translational and rotational freedom, exactly the same tug-of-war that determines whether a small-molecule ligand binds tightly or not at all. The symmetry equation is pure geometry: it explains, without any reference to sequence, why certain oligomer sizes ($n=2,3,4,6,12,60\ldots$) are common and others (5, 7 outside very specific contexts) are geometrically awkward for closed rotational assembly of identical subunits.

Example 1 (foundational): Converting a measured $K_d$ into an assembly free energy. A calorimetry experiment on a small viral capsid protein finds that dimerization has $K_d = 3.2\ \mu\text{M} = 3.2\times10^{-6}$ M at 298 K. Find $\Delta G^\circ_{\text{assoc}}$ and estimate the buried nonpolar area consistent with this value if the translational/rotational entropy penalty is 11.5 kcal/mol. Using $\Delta G^\circ = RT\ln K_d$ with $R = 1.987\times10^{-3}\ \text{kcal mol}^{-1}\text{K}^{-1}$:

So the dimer is favorable by about $-7.5$ kcal/mol. Setting up the energy budget (ignoring smaller polar/vdW/conformational terms for this estimate) as

With $\gamma = 0.0225$ kcal mol$^{-1}$Å$^{-2}$,

This is a modest interface (consistent with a moderately weak micromolar dimer), well below the 1200–2000 Ų typical of tight, obligate interfaces — a useful diagnostic: weak $K_d$ correlates with small buried area, exactly as the surface-tension model predicts.

Example 2 (applied): Cooperativity signature from a Hill-type analysis of assembly-linked ligand binding. A tetrameric enzyme (four identical, $D_2$-symmetric subunits) binds substrate S with an apparent Hill coefficient $n_H = 2.3$ measured from a plot of $\log[\theta/(1-\theta)]$ versus $\log[S]$, where $\theta$ is fractional saturation. At $[\text{S}] = 4.0\times10^{-4}$ M the enzyme is 62% saturated. Estimate $K_{0.5}$ (the substrate concentration at half-saturation) and comment on what $n_H$ says about interface-mediated communication between subunits. The Hill equation is

Rearranging with $\theta = 0.62$:

Take both sides to the power $1/2.3 = 0.4348$:

So $K_{0.5} \approx 0.32$ mM. Because $n_H = 2.3$ is well above 1 (and below the theoretical ceiling of 4 for a tetramer with maximal cooperativity), the interfaces in this tetramer must be conformation-transmitting: substrate binding at one subunit measurably distorts the neighboring subunits' active sites through the shared quaternary contacts, exactly the mechanism you will see worked out atom-by-atom for hemoglobin oxygen binding when studying Cooperative Binding and Allostery. An $n_H$ near 1 would instead indicate independent, non-communicating sites — geometrically oligomeric but functionally behaving like four separate monomers.

Example 3 (exam/competition level): Predicting the effect of an interface-disrupting point mutation on both stability and cooperativity. An $\alpha_2\beta_2$-type heterotetramer (two types of chain, related the way hemoglobin's chains are) has a wild-type dimer-of-dimers dissociation constant for splitting into two $\alpha\beta$ dimers of $K_{d,\text{tetramer}} = 1.0\times10^{-6}$ M, and each $\alpha\beta$ dimer further dissociates into free monomers with $K_{d,\text{dimer}} = 1.0\times10^{-9}$ M — reflecting the well-known fact that the $\alpha_1\beta_2$ "sliding" interface (the one that moves during the allosteric transition) is far weaker than the tight, largely fixed $\alpha_1\beta_1$ interface. A point mutation at the sliding interface reduces the buried nonpolar area there from 1100 Ų to 650 Ų, with all other terms unchanged. Estimate the new tetramer $K_d$, and predict qualitatively what happens to the cooperativity of ligand binding. The change in buried area alters only the hydrophobic term of $\Delta G^\circ_{\text{assoc}}$ for the tetramer-forming interface:

The positive sign means the mutant interface is destabilized by 10.13 kcal/mol relative to wild type (less favorable burial). Converting the wild-type tetramer $K_d$ to $\Delta G^\circ$:

The mutant's tetramer association free energy is

A positive $\Delta G^\circ$ means the tetramer is now unstable relative to separated dimers under standard conditions — solving for $K_d$:

This astronomically weaker $K_d$ (26.9 M — i.e., essentially never assembled at any physiological concentration) shows that even a modest-looking loss of ~450 Ų of buried nonpolar surface can be catastrophic, because the interface is fighting the same fixed ~11 kcal/mol translational/rotational entropy toll from Derivation 1 with a much smaller hydrophobic "payment." Functionally, if this sliding interface is the one that transmits the allosteric conformational signal (as it is in hemoglobin's actual $\alpha_1\beta_2$ contact), the mutant protein would exist predominantly as separate, non-communicating $\alpha\beta$ dimers — each dimer binding ligand with simple, non-cooperative (hyperbolic, $n_H \to 1$) kinetics, having lost the tetrameric scaffold needed to couple one subunit's ligation state to another's affinity. This is the structural logic behind real human hemoglobin variants (e.g., Hb Kansas) that weaken the $\alpha_1\beta_2$ interface and produce anemia through loss of cooperative oxygen delivery, not through loss of oxygen-binding chemistry itself.

Max Ferdinand Perutz spent over two decades — from the late 1930s to 1959 — solving the crystal structure of horse hemoglobin, a project so protracted that his own supervisor doubted it could ever succeed with the X-ray methods of the time. Working at the Cavendish Laboratory in Cambridge, Perutz pioneered the method of isomorphous replacement (soaking heavy atoms like mercury into protein crystals to solve the crystallographic phase problem) specifically to crack hemoglobin's quaternary architecture, and in doing so gave structural biology its founding example of subunit assembly: four chains, two kinds, related by near-$D_2$ symmetry, gripping each other at interfaces that move when oxygen binds. His 1959–1970 work with collaborators established that hemoglobin exists in two discrete quaternary states — the low-affinity "tense" (T) and high-affinity "relaxed" (R) forms — related by a rigid-body rotation of one $\alpha\beta$ dimer relative to the other by roughly 15° about a single axis, with the $\alpha_1\beta_2$ interface as the sliding hinge. This single structure became the Rosetta Stone for understanding how oligomeric assembly converts local, single-site ligand binding chemistry into a global, cooperative, physiologically tunable response — precisely the mechanistic template later generalized into the Monod-Wyman-Changeux and Koshland-Nemethy-Filmer models of allostery.