Energetic components of the allosteric machinery in hemoglobin measured by hydrogen exchange.

A hydrogen exchange (HX) functional labeling method was used to study allosterically active segments in human hemoglobin (Hb) at the alpha-chain N terminus and the beta-chain C terminus. Allosterically important interactions that contact these segments were removed one or more at a time by mutation (Hbs Cowtown, Bunbury, Barcelona, Kariya), proteolysis (desArg141alpha, desHis146beta), chemical modification (N-ethylsuccinimidyl-Cys93beta), and the withdrawal of extrinsic effectors (phosphate groups, chloride). The effects of each modification on HX rate at the local and the remote position were measured in the deoxy Hb T-state and translated into change in structural free energy at each position.The removal of individual salt links destabilizes local structure by 0.4 to 0.75 kcal/mol (pH 7.4, 0 degreesC, 0.35 M ionic strength) and often produces cross-subunit effects while hemoglobin remains in the T-state. In doubly modified hemoglobins, different changes that break the same links produce identical destabilization, changes that are structurally independent show energetic additivity, and changes that intersect show energetic overlap. For the overall T-state to R-state transition and for some but not all modifications within the T-state, the summed loss in stabilization free energy measured at the two chain termini matches the total loss in allosteric free energy measured by global methods. These observations illustrate the importance of evaluating the detailed energetics and the modes of energy transfer that define the allosteric machinery.

Signal transmission between subunits in the hemoglobin T-state.

To study allosteric mechanism in hemoglobin, a hydrogen-exchange method was used to measure ligand-dependent changes in structural free energy at defined allosterically sensitive positions. When the two alpha-subunits are CN-met liganded, effects can be measured locally, within the alpha-subunit, and also remotely, within the beta-subunit, even though the quaternary structure remains in the T conformation. When the two beta-subunits are liganded, effects occur at the same positions. The effects seen are the same, independently of whether ligands occupy the alpha-chain hemes or the beta-chain hemes. Control experiments rule out modes of energy transfer other than programmed cross-subunit interaction within the T-state. Cross-subunit transfer may depend on pulling the heme trigger (moving the heme iron into the heme plane) rather than on liganding alone.

Measurement of protein structure change in active muscle by hydrogen-tritium exchange.

A hydrogen-tritium exchange method was developed to study protein structure changes at the molecular level in active muscle. Skinned rabbit psoas fibers mounted on a specially designed holder were selectively tritium labeled at peptide group NH sites that change from a highly protected form in rigor to an easily exchangeable, essentially random coil condition when muscle is activated. The number of sites found to show this behavior varies linearly with thick filament-thin filament overlap, and would correspond to 83 amino acids per myosin molecule in the muscle, although the experiments do not yet place these sites in any given protein. Half of the sensitive sites respond to relaxing conditions as well to activation.

Mechanisms and uses of hydrogen exchange.

Recent work has largely completed our understanding of the hydrogen-exchange chemistry of unstructured proteins and nucleic acids. Some of the high-energy structural fluctuations that determine the hydrogen-exchange behavior of native macromolecules have been explained; others remain elusive. A growing number of applications are exploiting hydrogen-exchange behavior to study difficult molecular systems and elicit otherwise inaccessible information on protein structure, dynamics and energetics.

Thermodynamic parameters from hydrogen exchange measurements.

Just as exchangeable hydrogens that are controlled by global unfolding can be used to measure thermodynamic parameters at a global level, hydrogens that are exposed to exchange by local unfolding reactions may be used to obtain locally resolved energy parameters. Results with the hemoglobin system demonstrate the ability of HX methods to locate functionally important changes in a protein and to measure the energetic contribution of each. These results offer the promise that HX measurements may be used to delineate, in terms of definable bonds and their energies and interactions, the network of interactions that Hb and other proteins use to produce their various functions.

Hydrogen exchange measurement of the free energy of structural and allosteric change in hemoglobin.

The inability to localize and measure the free energy of protein structure and structure change severely limits protein structure-function investigations. The local unfolding model for protein hydrogen exchange quantitatively related the free energy of local structural stability with the hydrogen exchange rate of concerted sets of structurally related protons. In tests with a number of modified hemoglobin forms, the loss in structural free energy obtained locally from hydrogen exchange results matches the loss in allosteric free energy measured globally by oxygen-binding and subunit dissociation experiments.

Allosteric energy at the hemoglobin beta chain C terminus studied by hydrogen exchange.

When hemoglobin switches from the deoxy (T) to the liganded (R) form, several of its peptide group NH experience a great increase in their rate of exchange with water. Selective labeling and fragment isolation experiments identify some of the sensitive protons as three to four near-neighbor H-bonded peptide NH placed between Ala140 beta and the C-terminal His146 beta residue. These NH have differing solvent accessibilities, yet all exchange at about the same rate, and they maintain a common rate in the face of modifications that change their exchange rate over a 1000-fold range. This suggests that their exchange is mediated by a concerted transient unfolding reaction. The removal of allosterically important salt links at the distant alpha subunit N termini (des-Arg141 alpha hemoglobin) has little if any effect on the indicator NH at the beta C terminus. This demonstrates the restricted reach of the separate allosteric interactions in the T form as well as the localized nature of the H-exchange probe. Breakage of a salt link at the beta chain C terminus (His146 beta to Asp94 beta) by chemical modification (NES-Cys93 beta hemoglobin) speeds exchange of the indicator peptide NH in T-state hemoglobin by six-fold, which corresponds to an allosteric destabilization at the C-terminal segment of 1 kcal (pH 7.4, 0 degrees C), according to local unfolding theory. This is in quantitative agreement with energy values obtainable from other measurements. These NH exchange with an average halftime of five hours in deoxy hemoglobin and 15 seconds in oxy hemoglobin. According to the unfolding model for protein H-exchange, the 1200-fold increase in rate indicates a loss of 3.8 kcal in structural stabilization free energy at or near the C terminus of each beta chain in the T to R transition (pH 7.4, 0 degrees C, with 2,3-diphosphoglycerate). This result together with other available data places about 70% of hemoglobin's total allosterically significant structural energy change at the beta chain C termini.

Salt, phosphate and the Bohr effect at the hemoglobin beta chain C terminus studied by hydrogen exchange.

Hydrogen exchange experiments using functional labeling and fragment separation methods were performed to study interactions at the C terminus of the hemoglobin beta subunit that contribute to the phosphate effect and the Bohr effect. The results show that the H-exchange behavior of several peptide NH at the beta chain C terminus is determined by a transient, concerted unfolding reaction involving five or more residues, from the C-terminal His146 beta through at least Ala142 beta, and that H-exchange rate can be used to measure the stabilization free energy of interactions, both individually and collectively, at this locus. In deoxy hemoglobin at pH 7.4 and 0 degrees C, the removal of 2,3-diphosphoglycerate (DPG) or pyrophosphate (loss of a salt to His143 beta) speeds the exchange of the beta chain C-terminal peptide NH protons by 2.5-fold (at high salt), indicating a destabilization of the C-terminal segment by 0.5 kcal of free energy. Loss of the His146 beta 1 to Asp94 beta 1 salt link speeds all these protons by 6.3-fold, indicating a bond stabilization free energy of 1.0 kcal. When both these salt links are removed together, the effect is found to be strictly additive; all the protons exchange faster by 16-fold indicating a loss of 1.5 kcal in stabilization free energy. Added salt is slightly destabilizing when DPG is present but provides some increased stability, in the 0.2 kcal range, when DPG is absent. The total allosteric stabilization energy at each beta chain C terminus in deoxy hemoglobin under these conditions is measured to be 3.8 kcal (pH 7.4, 0 degrees C, with DPG). In oxy hemoglobin at pH 7.4 and 0 degrees C, stability at the beta chain C terminus is essentially independent of salt concentration, and the NES modification, which in deoxy hemoglobin blocks the His146 beta to Asp94 beta salt link, has no destabilizing effect, either at high or low salt. These results appear to show that the His146 beta salt link, which participates importantly in the alkaline Bohr effect, does not reform to Asp94 beta or to any other salt link acceptor in a stable way in oxy hemoglobin at low or high salt conditions.

Hydrogen-tritium exchange survey of allosteric effects in hemoglobin.

The oxy and deoxy forms of hemoglobin display major differences in H-exchange behavior. Hydrogen-tritium exchange experiments on hemoglobin were performed in the low-resolution mode to observe the dependence of these differences on pH (Bohr effect), organic phosphates, and salt. Unlike a prior report, increasing pH was found to decrease the oxy-deoxy difference monotonically, in general accordance with the alkaline Bohr effect. A prior report that the H-exchange difference between oxy- and deoxyhemoglobin vanishes at pH 9, and thus appears to reflect the Bohr effect alone, was found to be due to the borate buffer used, which at high pH tends to abolish the oxy-deoxy difference in a limited region of the H-exchange curve. Effects on hemoglobin H exchange due to organic phosphates parallel the differential binding of these agents (inositol hexaphosphate more than diphosphoglycerate, deoxy more than oxy, at low pH more than at high pH). Added salt slows H exchange of deoxyhemoglobin and has no effect on the oxy form. These results display the sensitivity of simple H-exchange measurements for finding and characterizing effects on structure and dynamics that may occur anywhere in the protein and help to define conditions for higher resolution approaches that can localize the changes observed.

Biochemistry without oxygen.

Published procedures for experimentation under anoxic conditions generally involve specialized apparatus that hinders the easy manipulation of experimental samples. We describe here some procedures that rapidly remove oxygen from experimental solutions, maintain anoxia with simple equipment for long periods of time, and do not interfere with normal sample addition and removal, spectrometric measurements, chromatographic manipulations, and the like. Anoxia can be achieved and maintained by the use of an enzyme system (glucose oxidase, glucose, catalase), or an inorganic oxygen-reducing system (ferrous pyrophosphate), or dithionite. Physical isolation of experimental samples from atmospheric oxygen can be maintained by continuous flushing with treated argon gas and/or by an overlay of heavy mineral oil.

Protein hydrogen exchange studied by the fragment separation method.

The potential of hydrogen-exchange studies for providing detailed information on protein structure and structural dynamics has not yet been realized, largely because of the continuing inability to correlate measured exchange behavior with the parts of a protein that generate that behavior. J. Rosa and F. M. Richards (1979, J. Mol. Biol. 133, 399-416) pioneered a promising approach to this problem in which tritium label at exchangeable proton sites can be located by fragmenting the protein, separating the fragments, and measuring the label carried by each fragment. However, severe losses of tritium label during the fragment separation steps have so far rendered the results ambiguous. This paper describes methods that minimize losses of tritium label during the fragment separation steps and correct for losses that do occur so that the label can be unambiguously located and even quantified. Steps that promote adequate fragment isolation are also described.

Identification of an allosterically sensitive unfolding unit in hemoglobin.

Hydrogen-exchange studies locate a set of seven allosterically sensitive amide NH protons side by side around two turns of the F-FG helical segment in the hemoglobin beta chain. Some of these protons are on the aqueous protein surface and some deeply inside, yet they all exchange with solvent protons at similar rates. Further, they move in unison to a new common rate when hemoglobin changes its allosteric form. These observations and analogous results for other proteins appear to be inconsistent with penetration-dependent models which relate H-exchange rate to solvent accessibility in the native state. Rather, these results point to sizeable fluctuational distortions that make small sets of protons more or less equally accessible in some transient H-exchange transition state, as visualized in the local unfolding model. The set of allosterically sensitive protons studied here exchanges 30-fold faster in liganded hemoglobin than in the deoxy form. In terms of the unfolding model, this means that the F-FG structure is relatively destabilized in oxyhemoglobin, so that the allosterically linked change in structural free energy at F-FG favors the deoxy state. The 30-fold change in H-exchange rate suggests a contribution to the allosteric free energy by this segment of 2 kcal (1 cal = 4.184 J). These experiments utilized a labeling technique, described earlier, that selectively places tritium on sites whose H-exchange rates are sensitive to the protein functional state, and used a method introduced by Rosa & Richards (1979,1981) to locate this label in the protein. The latter method, which rapidly separates protein fragments under conditions that can preserve exchangeable label, was here brought to a more quantitative level. Taken together, these techniques provide a "functional labeling" method capable of selectively labeling and identifying protein segments that participate in functional interactions.

Re-examination of rhodopsin structure by hydrogen exchange.

The hydrogen exchange behavior of rhodopsin was re-examined by studies of the protein in the disc membrane and after solubilization in octyl glucoside. The methods used measure either the peptide hydrogens alone (hydrogen-deuterium exchange by infrared spectroscopy) or all slowly exchanging hydrogens (hydrogen-tritium exchange by hel filtration). Under mild exchange conditions, disc membranes and solubilized lipid-free proteins show very similar exchange behavior, indicating the absence of slowly exchanging lipid protons. At high temperature, exchange of an additional large group of very slow peptide NH can be detected. The total number of slow hydrogens significantly exceeds the amide content, and apparently includes slowly exchanging protons from perhaps 40% of the protein's non-amide side chains. This is thought to require the involvement of many polar side chains in internal H-bonding. The exchange rates of the non-amide side chains sites have not been determined. However, to the extent that these contribute to the fast time region of the measured kinetic H-exchange curve, previously identified with exposed, non-H-bonded peptides, the estimate of freely exposed rhodopsin peptides must be reduced. The fraction of free peptides could range from a remarkably high value of 70% down to about 45%.

A high energy structure change in hemoglobin studied by difference hydrogen exchange.

The hydrogen exchange behavior of a small allosterically responsive set of exchanging hydrogens was studied in hemoglobin A and in some chemically modified hemoglobins. The set experiences an exceptionally large change in exchange rate through hemoglobin's allosteric transition. This indicates, according to the local unfolding model of H-exchange, that a large change in allosteric free energy impinges on the opening segment that exposes these protons to exchange. In oxyhemoglobin the set consists of 5 to 6 protons which exchange with a half-time of 20 s at pH 7.4 and 0 degrees C. In deoxyhemoglobin the set splits into a slower and a faster half. The slower 3 protons exchange more slowly than in oxyhemoglobin by a factor of 5000 (26 h half-time) and are 5-fold slower still in the presence of pyrophosphate or inositol hexaphosphate (136 h half-time). The other 2 to 3 protons exchange about 20-fold faster in both cases (about 2 h and 10 h half-times). The effect of some chemical modifications was tested, including reaction with iodoacetamide and N-ethylmaleimide and cleavage with carboxypeptidases A and B. In all cases the 3 slower protons continue to behave as a cohesive set and in the various modified deoxyhemoglobins their exchange is accelerated by factors ranging between 1 and 3 decades. These factors correlate with the effect of the different modifications on hemoglobin cooperativity.

Individual breathing reactions measured in hemoglobin by hydrogen exchange methods.

Protein hydrogen exchange is generally believed to register some aspects of internal protein dynamics, but the kind of motion at work is not clear. Experiments are being done to identify the determinants of protein hydrogen exchange and to distinguish between local unfolding and accessibility-penetration mechanisms. Results with small molecules, polynucleotides, and proteins demonstrate that solvent accessibility is by no means sufficient for fast exchange. H-exchange slowing is quite generally connected with intramolecular H-bonding, and the exchange process depends pivotally on transient H-bond cleavage. At least in alpha-helical structures, the cooperative aspect of H-bond cleavage must be expressed in local unfolding reactions. Results obtained by use of a difference hydrogen exchange method appear to provide a direct measurement of transient, cooperative, local unfolding reactions in hemoglobin. The reality of these supposed coherent breathing units is being tested by using the difference H-exchange approach to tritium label the units one at a time and then attempting to locate the tritium by fragmenting the protein, separating the fragments, and testing them for label. Early results demonstrate the feasibility of this approach.