Hemoglobin Structure and Quaternary Change

Hemoglobin's T/R switch is not a museum piece — it is actively exploited and studied across medicine and biotechnology. Sickle cell disease arises from a single β6 Glu→Val substitution that creates a hydrophobic patch exposed specifically on deoxy (T-state) hemoglobin, allowing deoxy-HbS tetramers to polymerize into rigid fibers; understanding that the mutation's effect is conditional on the T-state surface (not the R-state one) explains why sickling occurs preferentially in venous, deoxygenated blood, and underlies drugs like voxelotor, which stabilizes the R state pharmacologically by covalently modifying Val1α, raising oxygen affinity and reducing the fraction of polymerization-prone deoxy-HbS. Allosteric effector design for blood substitutes and high-altitude physiology research directly manipulates the central-cavity 2,3-BPG site described here. Carbon monoxide and cyanide poisoning are understood through the same iron-coordination chemistry: CO binds with far higher intrinsic affinity than O2 and still triggers the T→R doming mechanism, which is why CO poisoning both blocks oxygen transport and left-shifts the oxygen affinity of the remaining unblocked hemes (a right-shift in normal Bohr terms would help unloading, but CO does the opposite, worsening tissue hypoxia). Structural comparisons of the T and R hemoglobin conformations remain the standard textbook and pedagogical model for allosteric drug design in general — from aspartate transcarbamoylase inhibitors to allosteric kinase inhibitors — precisely because hemoglobin was the first system where both end-states of an allosteric switch were solved atom-by-atom. Finally, engineered hemoglobins with altered interface residues are used as scaffolds for oxygen-carrying blood substitutes, where suppressing the T-state dimer dissociation (a side reaction not covered in the two-state model above) is itself an active area of protein engineering.

Ask the tutor: How does the sequential (KNF) model of hemoglobin binding differ from the two-state MWC picture developed here, and which one better fits real intermediate-saturation crystal structures?

Hemoglobin's basic fold is the globin fold you already know from myoglobin: eight α-helices (A–H) wrapping a heme group whose iron is anchored on its proximal face by a histidine (F8) and left open on the distal face for O2, CO, or NO, with a second histidine (E7) hydrogen-bonding the bound ligand. Each hemoglobin subunit — α (141 residues) or β (146 residues) — is essentially this fold with a shortened D-helix in α chains. What is new in hemoglobin is what happens when four of these globins are packed into one particle. The tetramer is best understood not as four equivalent subunits in a square but as a dimer of dimers. Two tight, extensive interfaces (α1–β1 and α2–β2, related by the molecule's two-fold axis) bury roughly 1700 Ų of surface each and involve both helical packing and interlocking loops; these α1β1/α2β2 contacts barely change between oxy and deoxy forms and are not thought to break during normal function. The two copies of this rigid αβ unit are held together by a second, much smaller and more mobile interface — α1β2 and its symmetric partner α2β1 — burying only about 1000–1600 Ų depending on the quaternary state. This is the interface that does essentially all the moving. At the center of the tetramer, a solvent-filled central cavity runs down the molecular two-fold axis, wide enough in deoxyhemoglobin to admit one molecule of 2,3-bisphosphoglycerate (2,3-BPG), which bridges the two β-chain N-termini and additional basic residues, stabilizing the T state electrostatically. This cavity narrows sharply upon the T→R transition, squeezing 2,3-BPG out — a preview of the allosteric-effector chemistry developed as The Bohr Effect and Allosteric Effectors.

The chemical trigger for all of this sits at the heme. In deoxyhemoglobin, Fe(II) is five-coordinate: four pyrrole nitrogens of the porphyrin plus the Nε of the proximal His F8. A five-coordinate high-spin Fe(II) has an ionic radius too large to fit inside the porphyrin's central hole, so the iron sits domed out of the heme plane, toward the proximal histidine, by roughly 0.4–0.6 Å. When O2 binds on the distal face, it forms a bent Fe–O–O geometry (as in myoglobin) and forces a change to low-spin Fe(II) — a smaller ion that fits in-plane. The iron moves into the porphyrin plane by that same ~0.4–0.6 Å, and because His F8 is still coordinated, the F-helix is pulled along with it. That local, few-tenths-of-an-Ångström motion is the whole story mechanically. The F-helix is packed against the FG corner (the loop between helices F and G), which in turn makes the principal contacts across the α1β2 interface. A shift of the F-helix therefore changes the packing at α1β2 in a way that is incompatible with the T-state salt-bridge network, and the interface responds not by breaking randomly but by rotating as a rigid body — one αβ dimer swings against the other by about 15°, sliding along a well-defined set of contacts often described as a ball-and-socket joint, where Cβ2's His97 shifts from packing against Thrα38 in the T state to packing against Thrα41 in the R state (the "switch" region), while an adjacent "hinge" region (around α94–Val β98) stays in near-continuous contact and pivots without a discrete jump.

The essential quantitative relationships in hemoglobin structure are geometric and thermodynamic, not a single "governing equation" the way a rate law is. Four relationships organize the chapter.

1. The two-state (MWQ) partition of quaternary structure. Hemoglobin exists overwhelmingly in only two quaternary conformations, T and R, whose populations are governed by an allosteric equilibrium constant

where: $[T_0]$ and $[R_0]$ are the concentrations of unliganded tetramer in the T and R quaternary states; $L$ is large (on the order of $10^4$–$10^6$) for hemoglobin, meaning unliganded protein sits almost entirely in T.

Reading: the number that matters for physiology is not the intrinsic affinity of a single heme but the ratio $L$, because it sets how much "pull" ligand binding must exert before the tetramer flips. This equation is developed quantitatively as the backbone of Cooperative Binding and Allostery; here we only need that $L\gg1$ and ask why structurally.

2. Free energy stored in the T-state salt bridge network. Each T-state-specific ion pair contributes a favorable free energy of formation, and the sum over the network sets the T–R energy gap:

where: $\Delta G_{\text{intrinsic}}$ is the conformational free energy difference of the polypeptide backbone/interface alone (rotation cost), and the sum runs over T-state salt bridges (e.g., Val1α–Argα141, His146β–Lysα40, Aspβ94–Hisβ146) that must be broken to reach R.

Reading: this is a bookkeping equation, not a fitted formula — it says explicitly that the observed preference for T in deoxyhemoglobin is a sum of many modest (2–5 kcal/mol) electrostatic and hydrogen-bond terms, none of which alone would lock the structure, but which cooperate combinatorially exactly the way the hemes do.

3. Iron-out-of-plane displacement and the doming trigger. The correlation between spin state, coordination number, and iron displacement can be written as a simple geometric statement:

where: $\Delta z_{\text{Fe}}$ is the iron displacement from the mean porphyrin plane, and $r_{\text{HS}}$, $r_{\text{LS}}$ are the effective ionic radii of high-spin and low-spin Fe(II) (roughly 0.78 Å vs. 0.61 Å), so the geometric mismatch with the ~2.0 Å porphyrin hole radius accounts for the observed ~0.4–0.6 Å motion.

Reading: this equation converts a piece of ligand-field theory (spin-state radius) directly into a mechanical lever arm — the same logic used for the myoglobin heme but now serving as the input signal to a quaternary switch rather than an end in itself.

4. Hill coefficient as a phenomenological readout of the switch. The oxygen-binding curve is fit empirically by

where: $\theta$ is fractional saturation, $p\mathrm{O_2}$ the partial pressure of oxygen, $P_{50}$ the half-saturation pressure (≈26 mmHg for human blood at pH 7.4, 37 °C), and $n_H$ the Hill coefficient (≈2.8–3.0 for hemoglobin, exactly 1 for myoglobin).

Reading: $n_H$ is not a mechanistic parameter — it is a curve-fitting index whose value in the range 1 < $n_H$ ≤ 4 (the subunit number) reports the degree of cooperativity, and its structural origin is precisely the T→R rotation this article dissects. The quantitative decomposition of $n_H$ into microscopic binding constants belongs to Cooperative Binding and Allostery.

Example 1 (foundational) — Counting and classifying the α1β2 interface contacts in T vs. R. Deoxyhemoglobin (T) and oxyhemoglobin (R) crystal structures both show extensive α1β2 contact, but the specific residue pairs in contact differ at the "switch" region while the "hinge" region stays essentially constant. Take the switch region residue pair Hisβ97–Thrα38 (T state) vs. Hisβ97–Thrα41 (R state), and the hinge region pair Proα44–Thrβ41 (present in both states). Task: given that the α1β2 interface rotates by $\phi \approx 15°$ about an axis passing near the hinge contact, and that Hisβ97's Cα lies at a radial distance $r \approx 15$ Å from that axis, estimate the arc-length displacement of Hisβ97 during the T→R transition, and compare it to typical van der Waals contact tolerances (~0.3–0.5 Å) to justify why the switch region must break contact while the hinge does not. Solution. Arc length for a rigid-body rotation is

with $\phi$ in radians:

A displacement of 3.9 Å is roughly an order of magnitude larger than the ~0.3–0.5 Å tolerance within which van der Waals and hydrogen-bond contacts remain intact without steric clash or bond-length distortion. Hisβ97 must therefore break its T-state contact with Thrα38 and slide into a new resting position near Thrα41 — precisely the "switch" behavior seen crystallographically. By contrast, a residue lying very close to the rotation axis (small $r$) would have small $s = r\phi$ even for the same 15° rotation; the hinge-region residues are, structurally, those situated near the effective pivot, so their local contacts persist (small arc length, within contact tolerance) throughout the rotation. This example shows quantitatively why a single global rotation angle produces two qualitatively different local behaviors (switch-like jump vs. hinge-like slide) depending purely on distance from the axis — no separate mechanism is needed. Limiting case check. If $r\to0$ (a contact sitting exactly on the rotation axis), $s\to0$ for any rotation angle: such a contact would never need to break, consistent with the observed hinge behavior in the limit of proximity to the pivot.

Example 2 (applied) — Estimating the physiological consequence: oxygen delivery with vs. without cooperativity. Human arterial blood equilibrates at $p\mathrm{O_2} \approx 100$ mmHg (lungs) and venous blood at $p\mathrm{O_2}\approx 30$ mmHg (resting tissue). Using the Hill equation with hemoglobin's real parameters ($P_{50}=26$ mmHg, $n_H=2.8$), compare the oxygen unloaded per unit hemoglobin to a hypothetical non-cooperative ($n_H=1$) carrier with the same $P_{50}$. Solution. Fractional saturation:

Cooperative case ($n_H=2.8$): At $p\mathrm{O_2}=100$ mmHg:

Compute $100^{2.8} = 10^{2\times2.8} = 10^{5.6} = 3.98\times10^{5}$, and $26^{2.8}$: $\ln 26 = 3.258$, so $2.8\times3.258 = 9.122$, $26^{2.8}=e^{9.122}=9.16\times10^3$.

At $p\mathrm{O_2}=30$ mmHg: $30^{2.8}$: $\ln30=3.401$, $2.8\times3.401=9.523$, $30^{2.8}=e^{9.523}=1.37\times10^4$.

Non-cooperative case ($n_H=1$, same $P_{50}=26$):

Interpretation. Cooperativity raises the fraction of oxygen delivered per hemoglobin molecule between lung and resting tissue from about 26% (non-cooperative, same $P_{50}$) to about 38% — roughly a 46% increase in useful oxygen delivery over the same physiological pressure swing, without needing a higher blood oxygen affinity. This is the quantitative payoff of the T→R switching mechanism developed above: the sigmoidal curve is steep in exactly the pressure window between lung and tissue. Limiting case check. As $n_H\to1$ from above, $\Delta\theta$ shrinks toward the non-cooperative value computed here (0.258); as $n_H\to4$ (the theoretical maximum for a tetramer, full concertedness), the curve would become even steeper and $\Delta\theta$ over this same pressure window would increase further, though real hemoglobin's $n_H\approx2.8$–$3.0$ reflects that the MWC two-state switch is not perfectly all-or-none.

Example 3 (exam/competition level) — Predicting the phenotype of an interface mutation from structure. A hypothetical mutation replaces Aspβ99 (a hinge-region residue that in wild type hydrogen-bonds to Tyrα42 and helps stabilize the R-state α1β2 interface) with Asn. Structural data show Aspβ99 sits close to the T→R rotation axis at $r\approx4$ Å (a hinge contact), and its carboxylate accepts a hydrogen bond from Tyrα42-OH specifically in the R state; the corresponding T-state contact does not involve this hydrogen bond. Given that this hydrogen bond contributes roughly $\Delta G_{\text{Hbond}}\approx 1.5$ kcal/mol of R-state-specific stabilization, and assuming the mutation abolishes the bond entirely while leaving all other interactions unchanged, predict qualitatively and quantitatively how the mutation shifts $L$ (using the MWC relation $\ln L = \Delta G_{T\to R}^{\circ}/RT$) and hence the expected clinical phenotype. Solution.

Step 1 — Identify what changes. Removing an R-state-stabilizing interaction is equivalent to raising the free energy of R relative to T by $+1.5$ kcal/mol (R becomes less stable, so R is disfavored, so T becomes relatively more favored, i.e., $L$ increases).

Step 2 — Quantify the shift. Using $\Delta(\ln L) = \Delta\Delta G / RT$ at $T=310$ K:

If wild-type $L_{\text{wt}} \approx 9\times10^4$ (from the earlier bookkeeping estimate), the mutant:

Step 3 — Translate to $P_{50}$ and cooperativity. In the MWC framework, increasing $L$ (T more favored at zero ligand) requires more ligand-binding events to tip the equilibrium toward R, which manifests as a higher $P_{50}$ (lower oxygen affinity) and, up to a point, can preserve or even slightly steepen cooperativity if the intrinsic affinities of T and R sites are otherwise unchanged. Because this particular residue sits at the hinge and only affects the R-state-specific hydrogen bond without altering the T-state salt-bridge network computed earlier, the intrinsic affinities of individual T- and R-state hemes are essentially unperturbed — only the quaternary equilibrium shifts.

Step 4 — Clinical reasoning. A mutation that raises $L$ by roughly an order of magnitude while leaving intrinsic site affinities alone predicts a low-oxygen-affinity hemoglobin variant — right-shifted oxygen dissociation curve, higher $P_{50}$, but with cooperativity retained (since the T and R states themselves, and the trigger mechanism, are structurally intact — only their relative populations shifted). Clinically, such variants typically present not with cyanosis-causing hypoxia at the tissue level, but rather can trigger a compensatory erythrocytosis (the body increases red cell mass to compensate for the lower affinity), or in some real cases are essentially silent unless combined with other stresses. This reasoning mirrors real low-affinity variants such as Hb Kansas (β102 Asn→Thr, an α1β2 interface substitution) which show right-shifted curves and reduced but not abolished cooperativity — validating the qualitative logic even though the specific residue and numbers here are illustrative. Limiting case check. If instead the hydrogen bond had been T-state specific (stabilizing T rather than R), the sign of $\Delta\Delta G$ would flip, $L$ would decrease, and the predicted phenotype would invert to a high-oxygen-affinity variant with a left-shifted curve — exactly the behavior seen in real T-state-destabilizing mutants like Hb Kempsey.