Kinetic Studies on the Binding Affinity of Human Hemoglobin for the 4th Carbon Monoxide Molecule, L,*

L,, the affinity of hemoglobin for the 4th CO molecule, has been determined for human adult hemoglobin (HbA) as a function of pH and the presence of organic phosphates by measuring the kinetic parameters for the reaction. of L, human fetal hemoglobin phosphate buffers determined pH-dependent. These results cannot be reconciled within the framework of the two-state allosteric model. Additional structures in the conformational equilibrium due to either intermediates in the T to R transition or two or more R states must exist.

The differences in the structural and functional properties of liganded and unliganded hemoglobin are well established and have become the basis for essentially all models explaining heme-heme interaction. Of these, the Monod, Wyman, Changeux model (2) has excited the most interest. In its simplest form, it postulates the existence of only two conformational states of the hemoglobin tetramer. These are a tense (T) state in which the heme groups have a low affinity for ligand, and a relaxed (R) state in which the heme groups have a high affinity for ligand. To a first approximation the main functional properties of human adult hemoglobin, its sigmoidal ligand saturation curve, the Bohr effect, and the effects of DPG' and other polyphosphates on ligand affinity, can be explained by this two-state model. It is the concerted structural transition from the low affinity state to the high affinity state that is * This work was supported by Grant HE 12524 from the National Heart and Lung Institute, National Institutes of Health, and funds from the Veterans Administration Research Project 6098.01. A preliminary report of this work was presented at the Meeting of the Biophysical Society (1). *Present address, Department of Biological Sciences, State University of New York at Binghamton, Binghamton, N.Y. 13901. 8 Established investigator of the American Heart Association. 'The abbreviations used are: DPG, 2,3-diphosphoglycerate; IHP, inositol hexaphosphate; HbA, human adult hemoglobin; HbF, human fetal hemoglobin; HbA-CPA, HbA digested with carboxypeptidase A; HbA-IHP, stripped HbA in the presence of IHP; HbA + DPG, stripped HbA in the presence of DPG; bis-Tris, [2,2-bis(hydroxymethyl)-2,2',2"-nitriloethanol].
Throughout this article the symbol I, denotes the equilibrium binding constant for the reaction of hemoglobin with CO. At no point does it denote the value of the equilibrium constant for the interconversion between the R and T conformational states.
responsible for the observed sigmoidal, rather than hyperbolic, ligand saturation curve. In addition, protons and organic polyphosphates exert their effects by preferentially binding to the tense structure, shifting the allosteric R++ T equilibrium, and resulting in a decrease in ligand affinity. One of the major premises of this two-state model is that the properties of the T and R states are not changed by allosteric effecters but that only the free energy of the R++ T transition is altered. The correctness of this premise relative to the R state has already been questioned by McDonald and Noble (3) who demonstrated a pH dependence of the rates of 0, dissociation (k,) and CO dissociation (1,) from both fully liganded human adult and fetal hemoglobins, and by Lindstrom and Ho (4) who detected changes in the NMR spectra of liganded hemoglobin with variations of pH. However, it is conceivable that both of these parameters might vary with pH while the ligand affinity, the parameter of primary importance, remains unchanged. Ligand affinity can be computed from kinetic parameters by taking the ratio of the combination to the dissociation rate constants. Therefore, determination of l',, the rate of combination of carbon monoxide with the triliganded hemoglobin molecule, when coupled with knowledge of 1, permits the calculation of the affinity constant (L, = 1',/1,) for the binding of the 4th CO molecule to the hemoglobin tetramer.
The use of flash photolysis techniques to measure the kinetics of the reaction of carbon monoxide with hemoglobin is well documented, beginning with the work of Gibson in 1956 (5). When the photodissociation of the carbon monoxyhemoglobin is limited to 10% or less, the predominant contribution to the observed recombination reaction comes from the binding of the 4th CO molecule to the triliganded tetramer (1',). The Studies on the 4th Step in Ligand Binding to Hemoglobin 6693 observed reaction is generally homogeneous and is at least 20.fold faster than the overall reaction of CO with fully deoxygenated hemoglobin. This latter reaction is rate-limited by the reaction of CO with the low affinity (T) state of the hemoglobin molecule. We have measured 1', for HbA as a function of both pH and the presence of organic phosphates, the most common heterotropic allosteric effecters of hemoglobin. In addition, we have extended the measurements of McDonald and Noble (3) to examine the effects of these same ions on 1,, the CO dissociation rate constant. Not surprisingly, I', is greatly affected by the presence of inositol hexaphosphate.
Also, the computed values of L, for both HbA and HbF are highly pH-dependent. Less expected is a clear effect of 2,3-diphosphoglycerate on the L, values of HbA. This results from significant alteration of both l', and 1, by this ion.
A number of hemoglobin components or derivatives have been suggested as good models for the high affinity (R) state of hemoglobin in that they bind ligand non-cooperatively and with high affinity. Two of these are the isolated subunits of hemoglobin (6) and HbA-CPA (7), hemoglobin from which the COOH-terminal histidine and penultimate tyrosine residues of the /3 chains have been removed by digestion with carboxypeptidase A. In an attempt to further elucidate the functional properties of the R state, flash photolysis studies of LYSH chains and HbA-CPA have been carried out. The results are reported and compared to those obtained with normal HbA. Apparently, carboxypeptidase A digestion largely eliminates the pH dependence of the high affinity state observed in unmodified hemoglobin.

MATERIALS AND METHODS
Hemoglobins-The hemolysate of human adult hemoglobin was prepared from freshly drawn blood according to the method of Geraci et al. (8). After lysis and a preliminary centrifugation to remove the cell membranes, the hemolysate solution was dialyzed overnight against distilled water and again centrifuged at 48,000 x g for 20 min to remove the remaining cell membranes. Fetal hemoglobin (HbF) was prepared by CM-cellulose chromatography as described by McDonald and Noble (3). The HbF was found to be free of contamination by alkaline denaturation (9). cy. sH chains were prepared according to the Geraci et al. (8) modification of the method of Bucci and Fronticelli (10).
The organic phosphate 2,3-diphosphoglycerate which is present in HbA hemolysate was removed by the method of Benesch et al. (11). Phosphate analysis, as described by Ames and Dubin (In), established that the amount of DPG remaining did not exceed 2% of the concentration of hemoglobin tetramers.
In most experiments where IHP or DPG was added to the stripped HbA, the final concentration of these organic phosphates was 1.0 mM. HbA-CPA was prepared by using a modification of the method described by Moffat (13). HbA was dialyzed for 15 h against 0.025 M barbital buffer, pH 8.3, and its concentration was adjusted to 30 mg/ml with the same buffer. Then 5.3 ml of 1 mg/ml of carboxypeptidase A (Sigma, C 6510) were added to 9 ml of the HbA solution and immersed in a 25" water bath. The rate of digestion was followed once per h by measuring the CO combination rates at 420 nm using the methods and apparatus of Gibson and Milnes (14). To measure the rate of CO combination, the digestion was stopped by adding 0. and HbA-CPA were phosphate/citrate from pH 5 to 6.5, phosphate from pH 6.5 to 7.5, phosphate/borat'e from pH 7.5 to 8, and borate from pH 8 to 9. All were 0.1 M in anion concentration.
These same buffers were used in the studies of stripped hemoglobin to which organic phosphates were added. For the experiments on stripped HbA, it was necessary to use buffers that did not affect the ligand affinity like the phosphate buffers. These buffers were 0.1 M cacodylate from pH 5 to 6.0, 0.1 M bis-Tris [2,2- This longer delay did not appear to affect the 1'. determinations greatly. The reaction was followed at 420 nm where there is a large optical density difference between the deoxy and carboxy Hb spectra and auramine absorbs strongly. In order to limit dissociation of the hemoglobin to dimers, the hemoglobin concentrations varied from 40 to 65 PM in heme equivalents.
This required the use of a cuvette with a l-mm optical path length. The cell was surrounded by a water jacket through which water at 20" was circulated.
The total CO concentration varied from 60 to 95 WM. Data collection and processing was achieved by means of an on-line digital computer essentially as described by DeSa (15) could not be introduced into the system. A 1.5-ml volume of concentrated carbon monoxyhemoglobin was added directly to a stoppered, anaerobic cuvette with a l-cm path length. An equal volume of deoxygenated buffer that had been equilibrated with 1 atm of NO and had been kept at a temperature of 20" was added to the cuvette. The concentration of NO was in excess of 1.0 x lo-' M, while that of the hemoglobin for most measurements was approximately 1.0 x 10m5 M in heme. Since large optical density changes were desired, the HbA concentration was greater than that previously used by McDonald and Noble (3), requiring the reference cuvette to also contain carbon monoxyhemoglobin, its concentration being approximately 20% that of the experimental solution. The replacement reaction was followed at 420 nm by a Cary 14 spectrophotometer with a water-jacketed cell holder. The measured rate was limited by and equal to the rate of dissociation of the first CO molecule. In order to determine the effects of the dimer population on these rate constants, I, for stripped HbA was also measured at concentrations of 1.0 x lo-' M and 1.0 x 1Om6 M in heme at pH 6 and pH 9. Stopped flow techniques and the previously mentioned on-line computer system were used for these measurements, There is an assumption that is made for both the measurement of 1, and the assignment of the rate measured by partial flash photolysis to the constant, I',. This assumption is that the species, Hb(CO),, undergoes primarily the reaction which the technique presumes; i.e. the reaction with CO to form Hb(CO), in the measurement of 1',, or the reaction with NO to form Hb(CO),NO in the measurement of 1,. There exists the possibility that if L,, the affinity of the hemoglobin for the third CO, were low enough, and if I,, the rate of dissociation of 6694 Studies on the 4th Step in Ligand Binding to Hemoglobin Hb(CO), to Hb(CO)% + CO, were fast enough, our assumption for one or both techniques might be incorrect. However, these circumstances would require the existence of a functional state of the hemoglobin molecule for which no evidence exists. In addition, the value of 1, necessary to invalidate our assumption would be orders of magnitude greater than any reported rate of CO dissociation from a hemoglobin derivative.
Measurements of 0, Dissociation Kinetics-Both the overall rate of O2 dissociation (k) and the rate of O2 dissociation with replacement by carbon monoxide (k,) were measured with a stopped flow apparatus as described by Gibson and Milnes (14) using the same automatic data collection system as with the flash photolysis experiments. A 2-mm optical path length was used. k was followed at 435 and 415 nm and k, was followed at 420 and 406 nm. The dithionite solution was prepared by dissolving the reagent in a carefully deoxygenated solution of 1 mu NaOH.
The Hb concentration for the determination of both k and k, was 2.5 x 10m5 M after mixing in the stopped flow. The CO concentration in the k, determination was 5 x lo-' M. Oxygen Equilibrium Measurement-The oxygen affinity of HbA-CPA was measured by a modification (16)  In fact, at pH 9, the rate constants are essentially the same.
In the upper graph of Fig. 2     Oxygen Equilibrium of HbA-CPA-- Fig.   6 shows the Hill plots of the oxygen equilibrium curves at pH 5.7, 6.5, 7.2, and 7.5  Oxygen Dissociation from HbA-CPA--In order to explore the effect of this cooperativity on the kinetics of ligand dissociation, the overall rate of 0, dissociation, k, was compared to k,, the dissociation rate of oxygen from the fully liganded Hb tetramer. Measurements of these two first order reactions were made at 20" and the rate constants are plotted as a function of pH in Fig. 7. The values of k and k, for HbA of the cooperativity observed in the oxygen equilibrium experiments.
The 0, dissociation rates for HbA-CPA are 10 s' from pH 5.2 to pH 9.0 and are nearly the same as the values of k and k, obtained  (11,30,31). Below pH 8, IHP causes a very large reduction in 1',. At pH 7 this reduction is lo-fold, while at pH 5 and 6 it is somewhat smaller. Similar effects of IHP on oxygenation were observed by Gray and Gibson (31). They calculated the binding affinity for the 4th 0, molecule, K,, and found that in the presence of IHP at pH 7 the value was 13.fold less than for stripped hemoglobin. Although these results can be interpreted as a modification of the R structure, the occurrence of appreciable amounts of the T state at high levels of saturation appears likely. The biphasic kinetic data observed after partial photolysis under these conditions have more than one potential explanation. The two phases could be attributed to the hemoglobin populations in the R and in the T conformations only if the R-T relaxation time were slow relative to the CO combination reaction. Alternatively, they could represent subunit heterogeneity in the IHP-modified R state. The data presented here are inadequate to choose between these and other possibilities. The inadequacy of a model invoking only two conformations to explain the properties of the hemoglobin molecule has been apparent for some time. In 1970 Perutz (32) distinguished between tertiary and quaternary conformational states in order to permit a sequential release of protons during ligand binding.
Lau et al. (33) have clearly shown a pH dependence in the properties of the T state of a trout hemoglobin, and such a dependence of the properties of the T state of human hemoglo-bin is clear from the results of Imai (27). It now appears that the R state cannot be assumed to be invariant. Thus, a significant contribution to the Bohr effect may come from pH-dependent equilibria between R states and between T states, as well as from the pH-dependent transition from the T to the R conformation.