Towards understanding non-equivalence of α and β subunits within human hemoglobin in conformational relaxation and molecular oxygen rebinding

Picosecond to millisecond laser time-resolved transient absorption spectroscopy was used to study molecular oxygen (O2) rebinding and conformational relaxation following O2 photodissociation in the α and β subunits within human hemoglobin in the quaternary R-like structure. Oxy-cyanomet valency hybrids, α2(Fe2+–O2)β2(Fe3+–CN) and α2(Fe3+–CN)β2(Fe2+–O2), were used as models for oxygenated R-state hemoglobin. An extended kinetic model for geminate O2 rebinding in the ferrous hemoglobin subunits, ligand migration between the primary and secondary docking site(s), and nonexponential tertiary relaxation within the R quaternary structure, was introduced and discussed. Significant functional non-equivalence of the α and β subunits in both the geminate O2 rebinding and concomitant structural relaxation was revealed. For the β subunits, the rate constant for the geminate O2 rebinding to the unrelaxed tertiary structure and the tertiary transition rate were found to be greater than the corresponding values for the α subunits. The conformational relaxation following the O2 photodissociation in the α and β subunits was found to decrease the rate constant for the geminate O2 rebinding, this effect being more than one order of magnitude greater for the β subunits than for the α subunits. Evidence was provided for the modulation of the O2 rebinding to the individual α and β subunits within human hemoglobin in the R-state structure by the intrinsic heme reactivity through a change in proximal constraints upon the relaxation of the tertiary structure on a picosecond to microsecond time scale. Our results demonstrate that, for native R-state oxyhemoglobin, O2 rebinding properties and spectral changes following the O2 photodissociation can be adequately described as the sum of those for the α and β subunits within the valency hybrids. The isolated β chains (hemoglobin H) show similar behavior to the β subunits within the valency hybrids and can be used as a model for the β subunits within the R-state oxyhemoglobin. At the same time, the isolated α chains behave differently to the α subunits within the valency hybrids.


Preparation of isolated hemoglobin chains
Human Hb was purified from freshly drawn blood. The isolated α and β chains were subsequently obtained by the p-mercuribenzoate (PMB) method 1 with minor modifications. The isolated chains with bound PMB (α PMB and β PMB chains) were obtained in either the oxy or the carbonmonoxy form. Separation of the PMB-reacted chains was carried out on a column with DEAE-Sepharose CL-6B (GE Healthcare) equilibrated with 20 mМ Tris HCl buffer, pH 8.1.
Elution was carried out by a linear gradient produced by mixing equal volumes of 20 and 380 mM Tris HCl buffers, both at pH 8.1, in a gradient mixer GM-1 (GE Healthcare). The first eluted fraction contained the α PMB chains, while the other main fraction, eluted later, contained the β PMB chains. The isolated α and β chains were regenerated to their -SH forms (α SH and β SH chains) using dithiothreitol (DTT). 2  chains were eluted with 220 mМ Tris HCl buffer at the same pH. All the procedures were carried out at 4°C. All the buffers used for the sulfhydryl groups regeneration were purged carefully with nitrogen. The molecular weights of the α and β chains were verified by electrospray ionization mass spectroscopy.

Preparation of hemoglobin valency hybrids
The Hb valency hybrids of the type α 2 (Fe 2+ -O 2 )β 2 (Fe 3+ -CN) and α 2 (Fe 3+ -CN)β 2 (Fe 2+ -O 2 ) were prepared by mixing the isolated chains with free -SH groups in the oxy form with their partner chains in the cyanomet form. 3 The cyanomet forms of the isolated α SH and β SH chains were obtained from the carbonmonoxy-derivatives by oxidation with potassium ferricyanide in the presence of sodium cyanide. The complete oxidation of the isolated chains with the concomitant formation of the cyanomet-derivatives was controlled spectrophotometrically. Ferri-and ferrocyanide were removed by gel filtration on a Sephadex G-25 column equilibrated with 50 mM Tris HCl buffer, pH 8.2. All the procedures were carried out at 4°C. The valency hybrids were prepared within one day before experimental measurements. To ensure as complete recombination as possible of the oxygenated partner, an excess of the cyanomet chains (between S4 20 and 50%) was added. The valency hybrids were not separated from the unrecombined cyanomet chains, which were spectroscopically inactive in the present O 2 rebinding studies in a specially chosen time frame (see Section 3.1 in the main text).

Maximum entropy method (MEM) analysis
The obtained time-dependent amplitudes of the basis spectra were subjected to the maximum entropy method (MEM) analysis which extracts model independent lifetime distributions from the kinetics. 7 To analyze the time-dependent amplitudes in terms of distributed lifetimes, the program MemExp (version 3.0) 8,9 was used. One or two distributions of effective log-lifetimes, g(log τ) and h(log τ), were extracted from the data. The fit F i to datum D i at time t i can be written where D 0 is a normalization constant, g(log τ) and h(log τ) are the lifetime distributions that correspond to decaying and rising kinetics, respectively. The quality of the fit was evaluated by the χ 2 value, 10  TEST. 12 To improve the fidelity of the recovered distributions as well as to determine which features in the lifetime distributions are required by the data, the fit was performed with the MEM using different prior distributions. 8,9 Namely, uniform distributions as well as distributions derived by uniform or differential blurring of intermediate MEM results were used as the prior. 8,9 Moreover, fits to the data were performed several times using different expected lifetime limits at a maximum number of anticipated lifetimes.

Simulation of experimental data
In the present work we performed modeling the data matrix D so that to produce a set of timedependent species populations (which comprise a matrix P) and a set of corresponding species spectra (which comprise a matrix S), such that the product SP provides the best approximation to the matrix D , i.e. D ≈ SP. P is a 2 × n matrix whose rows contain the total populations of unliganded subunits in each of the two tertiary structures r * and r. S is a m × 2 matrix whose columns contain difference spectra between unliganded and O 2 -liganded ferrous subunits in each of the two tertiary structures r * and r. Each difference spectrum is assumed to be a linear combination of the first two basis spectra, U 1 and U 2 . Moreover, to quantitatively relate the experimental data to the kinetic models (see first two components stand out above the others, which, in their turn, form almost a straight line.
The rank of matrix D is estimated as two by identification of two singular value components above the systematic decline.
The peak position was obtained by fitting the data in the region of the peak to a third-order  The values were taken from Yamada et al. 20 The uncertainties shown are 90% confidence levels in the curve fitting. f In hybrid Hb, the native protoheme is replaced with mesoheme in either the α or β subunits. 17 g The values were taken from Jones et al. 17

Table S3
List of model parameters