Half a Century of Hemoglobin’s Allostery

Main Text

In 1965, Monod et al. (1) published a plausible model on the nature and role of allosteric transitions in proteins. That article triggered the intense interest of a vast number of talented scientists starting with Perutz (2), the father of structural biology. The competition with the induced-fit model of Koshland et al. (3), started an exciting scientific debate. In science, the general validity of an idea can best be appreciated in the long run. After half a century, allostery is still alive and flourishing; and interesting, novel applications of the theory have led to original pharmacological approaches for the therapy of central-nervous system diseases and cancer.

The term “Allostery” was conceived to account for the experiments carried out by Changeux (4) on the enzyme threonine deaminase. Nevertheless, at the time it seemed quite natural to adopt hemoglobin as the paradigmatic case of an allosteric protein, for the following reasons: 1) The oxygen binding curve of Hb is cooperative, implying heme-heme interactions; 2) the position and shape of the binding curve depend on allosteric effectors (H⁺, bisphospoglycerate (BPG), Cl⁻, etc.); 3) the tertiary and quaternary structures of oxygenated and deoxygenated Hb are different; 4) a quickly reacting state of Hb has been discovered, by flash photolysis of the CO derivative; and 5) many natural Hb variants, normal and abnormal, have been proven to be very useful tools for exploring the mechanism. Therefore, hemoglobin was promoted to the rank of an “honorary enzyme” (5).

The energy level diagram depicted in Fig. 1 illustrates the basic parameters of the theory, as outlined in the legend. Perutz (2) proposed a structural mechanism to explain the difference in ligand affinity and overall stability between the T- and R-quaternary states of Hb, which was basically consistent with the Monod-Wyman-Changeux (MWC) model. The iron position out of the porphyrin plane in deoxyHb shifts into the plane upon ligand binding, acting as a trigger; a number of salt bridges across the α₁-β₂ interface, present in deoxy Hb but broken upon oxygenation, would account for the low affinity of the T state, the difference in stability among T₀ and R₀ and the Bohr effect. Perutz’s stereochemical mechanism was the first attempt to infer the energetics of a complex process by examination of a three-dimensional structure, a challenging task. His mechanism was tackled using a battery of advanced biophysical methods. Some review articles summarizing the level of understanding and consistency with the MWC model may be useful to read (6–8).

Figure 1.

Picasso’s dove is an attractive image conveying a universal message; no comment required. Absit inuria verbo: conformational selection, the founding concept behind allostery, is condensed in a sketchy energy level diagram, a blend of the principles conceived by Le Châtelier and Darwin. The model postulates that the hemoglobin tetramer populates two conformational states, different in quaternary and tertiary structure, called R for relaxed and T for tense. The 10 species (T₀ to T₄, and R₀ to R₄) are in equilibrium at every level of ligation; but in the absence of oxygen, T₀ prevails over R₀ (by as much as 1,000,000 folds), while the opposite occurs for the fully saturated Hb; at intermediate saturations (in the figure, T₂-R₂) the switchover occurs. The equilibrium parameters are: L₀ = T₀/R₀; K_T and K_R are the dissociation constants for the reaction of oxygen with the two states.

The article by Henry et al. (9), offered in this issue of the Biophysical Journal, discusses the experimental basis for proposing the “simplest possible extension of the MWC model”, a variant termed by them the tertiary two-state (TTS) model, which includes tertiary as well as quaternary pre-equilibria. Over and above presenting their data, Henry et al. (9) discuss the TTS model by comparison with previous articles, and particularly the statistical mechanical description of Perutz’s stereochemical mechanism published by Szabo and Karplus (10).

In a first approach, to test specific predictions of the MWC model, Eaton et al. (7) pursued the strategy of characterizing ligand binding by human hemoglobin locked in either one of the two allosteric states (R and T). Perutz et al. (6) had discovered that deoxy human Hb, crystallized from polyethylene glycol at low pH, maintains the salt bridges intact even upon ligand saturation. Accurate oxygen binding curves of these crystals obtained by microspectrophotometry (Henry et al. (9) and references therein) showed that 1) the saturation curve is noncooperative (Hill’s coefficient n = 1); and 2) the oxygen affinity is very low and independent of pH, BPG, and other effectors. These findings are consistent with predictions of the MWC model whereby stabilization of the T state is associated to low affinity, and inconsistent with the Koshland, Nemethy, and Filmer-induced fit mechanism. However, the result contradicts Perutz, who stated that “If it is only the salt bridges which constrain the deoxy tetramer, then part of the energy released by the reaction of the haem groups with oxygen must be expended to break them”. As a matter of fact, this inconsistency had been already suggested by data on Hb Kansas, a single mutant of the α₁-β₂ interface (Ans¹⁰³(G4)β to Thr) destabilizing the R state; this variant is noncooperative and saturates without rupture of the salt bridges (8).

The second approach produced a surprising result that inspired the proposal of the TTS model. Shibayama and Saigo (11) had demonstrated that it was possible to lock either the T-state or the R state of human Hb by encapsulation into silica gel cavities. Although the physical state of the protein in the cavity is by-and-large a conundrum, both entrapped allosteric states are stable for hours to days and display noncooperative oxygen binding with either low affinity (the T state) or high affinity (the R state). The silica-gel locking protocol has been exploited successfully by the Bethesda-Parma group (Henry et al. (9) and references therein). They confirmed that R- and T-state Hb locked-in silica gels bind oxygen noncooperatively; however, at variance with data with the crystals (see above,) oxygen affinity of entrapped T-state Hb is sensitive to pH and BPG as the protein in solution (12). The surprise came from laser photolysis experiments carried out on silica gel entrapped Hb. This is a powerful methodology (13) to investigate Hb dynamics by breaking the ligand-iron bond with a short and intense flash of visible light. The CO rebinding time course after photolysis of T-state HbCO locked in the gel was multiphasic: a substantial fraction of the photoproduct reacted with CO very rapidly with a rate constant, in a manner essentially identical to that characteristic of the R state; and vice versa, upon photolysing the locked R state, a fraction of the rebinding occurred at the slow rate typical of the T state. Thus, the system populates a more complex ensemble of states encompassing T-quaternary with t and r, and R-quaternary with t and r. The TTS model is discussed by Henry et al. (9) in comparison with the models of Monod et al. (1) and the Szabo and Karplus (10). Henry et al. (9) state that quantitative analysis of the time-resolved CO rebinding experiments on gel-encapsulated Hb is consistent only with the TTS model. This extension of the MWC allosteric model will stimulate comments and new contributions given that some aspects need attention.

Henry et al. (9) point out that “inequivalence of the tertiary equilibrium constants for the α-and β-subunits in both the T and R quaternary structure” needs to be taken into consideration to account for the data. This was expected in view of the results of T. Spiro and co-workers by time-resolved resonance Raman spectroscopy and computational analysis (quoted by Henry et al. (9)), showing that the lower affinity conformation pertains mainly to the α-subunits. The statement that tertiary effects may control ligand binding is reasonable and in fact consistent with data on some natural variants; the paradigmatic case is Hb Zürich (distal HisE7 of the β-chains mutated to Arg), which displays a quickly reacting CO binding component in the allosteric T-state due to the opening of the distal pocket (as shown by kinetics and crystallography) (14).

The structural basis for the difference in reactivity between tertiary t and tertiary r is, as of this writing, only speculative. Indeed, Henry et al. (9) are working to obtain high-resolution structural information on unliganded t in the quaternary R-state, and vice versa of liganded r in the quaternary T state. Because the slow and fast CO rebinding bimolecular phases (r and t) seen in the locked quaternary states are essentially the same as the canonical R and T rates, it is assumed that the fast and slow tertiary conformations monitor the high- and low-affinity states. However, extrapolation from CO rebinding rates to oxygen affinities may demand some care given that 1) in the case of oxygen, cooperativity is largely expressed in the oxygen dissociation rate constant (which increases by 50–100-fold in going from R to T state) and the role of the H-bond with distal HisE7 is of paramount importance; while 2) in the case of CO, cooperativity is expressed largely with an increase in the combination rate constant (by ∼100-fold) and the role of the water molecule trapped in the distal cavity of the deoxy α-subunits is still debated. Possibly, we may expect more surprises in the future.

Editor: H. Jane Dyson.