The Effect of 2 , 3-Diphosphoglyceric Acid on the Changes in p-p Interactions in Hemoglobin during Oxygenation*

SUMMARY During the binding of oxygen to the artificial intermediate of saturation of hemoglobin, CYZ+~~PZ, a significant change in the interactions between the two p subunits is observed. The same change in interaction is found both in the presence and absence of the hemoglobin effector 2,3-diphosphoglyceric acid (DPG). The observed change in 6-p interactions even in the absence of DPG suggests that contacts between the two p chains contribute to cooper$ivity as well as a bridge formed by the DPG molecule. The study of the mechanism of oxygen binding by hemoglobin has in recent years centered on the interactions between the four hemoglobin subunits. During this period a variety of experimental approaches have all indicated that during the course of oxygenation the conformation of each subunit changes when oxygen is bound to that subunit (l-4), so that the coopera-tive oxygen-binding best described by a sequential relates cooperativity to changes in the strengths of various interactions between subunits as each subunit changes

The study of the mechanism of oxygen binding by hemoglobin has in recent years centered on the interactions between the four hemoglobin subunits. During this period a variety of experimental approaches have all indicated that during the course of oxygenation the conformation of each subunit changes when oxygen is bound to that subunit (l-4), so that the cooperative oxygen-binding mechanism is best described by a sequential change model (5). This model relates cooperativity to changes in the strengths of various interactions between subunits as each subunit changes shape. We have previously reported some experiments which measured some of the changes in subunit interactions (6). These results, similar to those obtained by two other groups (7, S), showed that there was a significant change in the interaction between the two /3 chains but little or no change in the interactions of the two OL chains during oxygenation.
The finding of a change in the p-p interaction during oxygenation emerged at the same time that the Benesches and their co-workers demonstrated that organic phosphates such as 2,3diphosphoglyceric acid bound to hemoglobin and affected the midpoint of the saturation curve (9 bound between the two fi chains (ll), and Perutz has reported that DPG could be bound in the cavity between the two fl subunits in the deoxy form (12). Bunn and Briehl (13) have studied DPG binding to normal and modified hemoglobins and concluded that the a-amino groups of the two p chains are involved in the binding. Even though the position of the oxygen binding curve is shifted to higher oxygen pressures in the presence of DPG or even phosphate buffer, there is little or no change in the shape of the binding curve (9) and hence in the degree of cooperativity among the subunits.
In view of the fact that a small but sig-n&cant change in /3-p interaction can be measured during oxygenation (6) it was of interest to examine whether the observed interactions between the two fi subunits were the result of direct contacts of amino acid side chains or were mediated through a, bridge of DPG or two phosphate ions. The nature of the p-0 interaction could be ascertained by measuring the change in /3-p binding energy in the presence and absence of DPG and other phosphate compounds.
The basic approach to these questions was to use artificial intermediates of saturation formed by freezing some of the subunits (in this case, the a chains) into an oxygenated conformational state. This approach, first developed by Rntonini et al. (14), was the basis of our previous measurement of CY-O( and p-/3 interactions (6). In Fig. 1 the method for measuring the change in /3-p interaction in the intermediate oc;tCNflz is illustrated.
Oxygen is bound only by the unmodified p chains. Since the same changes in crl-/3i and cry-& interactions occur following the binding of each oxygen molecule, these changes cancel out in measuring the ratio of the two binding constants, Kl and Kz. Thus, a value for KnnS/(KABB)2 is obtained.
If oxygen saturation is measured in the presence and absence of DPG under conditions when all other phosphate has been removed, the effect of DPG on the ,&3 linkage will be seen by the change in the ratio of the two binding constants.

EXPERIMENTAL RESULTS
Experimental methods and procedures for preparation of hemoglobin hybrids were as described in previous publication (6). %Ieasurcmcnts of oxygen saturation were made in the presence of 0.2 M HEPES buffer, 0.06 in N&l, and 5 x 10-4 I\I EDTA, pH 7.2. In this solution the isolated subunits nud the recombined cr& and OLZ+'~/~ 2 e ramers were found to be more t t stable than in a totally unbuffered solution.
HEPES buffer was not totally analogous to unbuffered solution since the satura-   tion curve of hemoglobin (oc&) in the HEPES solution has a midpoint of 6.5 mm of O2 as opposed to a midpoint value of 2.7 mm of O2 reported by Benesch et al. (9) for an unbuffered solution between pH 7.3 and 7.0. In an unbuffered solution a shift in pH results from the release of protons during oxygenation (9). It was felt that this small shift in pH, with a concomitant slight change in oxygen affinity, might obscure the small changes in slope which are being measured in these experiments.
Therefore the more stable HEPES buffer system was employed.
Oxygen binding to a& and to o12+c.cN/& was measured in the presence and absence of 8 X 10M4 M 2,3-diphosphoglyceric acid. In each case, the hemoglobin concentration was approximately 8 X 10e5 M. The saturation curves are presented in the form of Hill plots in Fig. 2, A and B. In Fig. 2R the saturation curves for reconstituted hemoglobin have a Hill slope of 2.7 which is in agreement with values for untreated hemoglobin. Although there is no change in the slope, the presence of DPG clearly shifts the midpoint of the saturation curve.
In the case of 0l2+'~& a similar shift of the midpoint is observed when DPG is added. In both the presence and absence of DPG, however, the slope of the Hill plot is 1.3. As we observed before (B), a slope of 1.3 represents a significant and consistently observed deviation from 1.0, the expected slope if there were no changes in p-/3 interactions during oxygenation. It would thus appear that the 6-p interaction changes remain constant whether DPG or other phosphates are present or not.
In Table I, the midpoints of the saturation curves are listed. The ratios of the midpoint in the presence of DPG to that in the absence of a phosphate compound is 1.7 in the case of CY&J and 1.85 for CY~+~~/~Z.
The fact that a similar shift in the midpoint of saturation occurs for both ~~$32 and a~+'~& when DPG is added to the solution is a direct demonstration that DPG exerts its effect on the two /3 chains regardless of the state of oxygenation of the ar subunits.
This observation is consistent with the fact that DPG is bound between the two 6 chains (10-12). DISCUSSION The observation that /3-p interactions are maintained in the presence or absence of phosphate ions or 2,3 diphosphoglyceric acid suggests that the changes in interaction between the two /3 subunits during oxygenation are not mediated through a phosphate or organic phosphate bridge.
There must be some direct changes in the interactions between the amino acid side chains of the two p subunits which accompany the sequential changes during oxygen binding.
It should be emphasized, of course, that, since the cooperativity depends on the change in subunit contacts, the contacts may be present only in the deoxy form and may be broken in the 0x7 form. Furthermore, these results lead to the conclusion that the binding site for DPG does not involve all of the amino acid side chains of the two p subunits which are implicated in changes in stabilization of the two subunits.
Recently Per&z (12) has suggested that the binding of DPG involves three /3 chain amino acid groups: the a-amino group of the NHz-terminal valine (l), the e-amino group of lysine (82), and histidine (143). Perutz also implicates a salt linkage between histidine (146) and the a-chain lysine (40). Since the p-p interactions are observed to change whether organic phosphate or phosphate ion is present or not, there appear to be additional amino acid contacts which are either broken or formed during the oxygenation process.
One additional salt linkage which has been previously suggested from crystallographic observations (15) is between the carboxyl group of the COOH-terminal histidine (146) and one p subunit and the e-amino group of lysine (132) in the other /3 chain. Such a bond may be important in t,he change of interaction energy between the two fl subunits.
The demonstration of a change in 6-p interactions during the oxygenation of ~r2+~~fl 2 was taken as evidence that ligand binding to hemoglobin follows a mechanism of sequential conformational changes rather than a concerted pathway (6). The effect of DPG further substantiates this conclusion.
The Hill coefficient for the binding of oxygen to crs+CN/3s is related to the binding constants K1 and K, by Equation I I If oxygen binding follows the (1) sequential model (cf. Fig. 1A) and DPG binds to the various mixed state hemoglobins as in Equation 2, then the Hill coefficient equation becomes Equat'ion 3. Kl(1 + KID1)z (3) ml + K"[Dl) (I + Kr1lDl) + 2 To satisfy the observation that YLE did not change in the presence or absence of DPG, the condition that  (6) Since the midpoint of 02 binding is shifted to higher concentrations of O2 in the presence of DPG, fulfillment of the conditions described in Equations 5 and 6 lead to the conch&on that li;, > Kr > Kzz, i.e. binding of DPG is strongest to the least liganded form. The observations that there was no change in Q but a significant shift in the binding curve midpoint when DPG is introduced thus are consistent with a sequential model.
A consideration of the concerted mechanism (16) (cf. Fig. 1B) yields a less satisfactory conclusion.
The Hill coefficient equa-Lion for the concerted model treated in its general (nonexclusive binding) form is: and KTD and 1c~~ represent the association constants for DPG of the T and R forms, respectively.
It is argued that in a concerted mechanism the binding of one or more ligand to hemoglobin should stabilize the molecule in the R state. In this case, then, the presence or absence of DPG should exert little or no effect, since the binding of DPG to the fully liganded molecule (that is, the R state) has been found to be very weak (11). One could postulate that the binding of DPG to a~z+'~ fiz would indeed be the same as to 012 +cNpo2/3 or to 012+CNP202. In this case, however, there should be no signilicant effect on the midpoint of saturation when DPG is added. Clearly, this is not the case.
The results reported here are in conflict with a recent report (17) based on the nuclear magnetic resonance heme spectrum of a2+CN& which postulates that the p chain heme are in an oxy form in the absence of DPG (in the presence of 2,2-bis(hydroxymethyl)-2,2',2"-nitriloethanol buffer) but in the deoxy form when DPG is introduced. From this no further conformational changes would be expected when DPG is absent, thus leading to a loss of p-0 interactions.
Using the HEPES buffer in the absence of DPG we find that p-0 interactions are maintained.
It should be pointed out that the conclusions reached here are based on t.he assumption that cyanomet derivatives are indeed close analogues of the oxy form of hemoglobin.
This assumption has formed the basis of considerable work on the mechanism of ligand binding hemoglobin (2, 6-8, 14, 17) and is in turn based on the observations by Perutz and his co-workers that the complete cyanomet hemoglobin is structurally analogous to oxy hemoglobin (15). Whether the CQ@~& form is therefore also a close analogue of an intermediate form of saturation azo2& can only be determined indirectly, although, for example, the data of Shulman et al. (2) show only two distinct nuclear magnetic resonance signals in the mixed state hemoglobins (as might be expected if a sequential model was followed).
It is also possible that a third concerted state exists for intermediates of the type o(itCNpz. A, vain, this alternative seems unlikely in terms of nuclear magnetic resonance data (2). Such a third state would also seem to negate many of the hypotheses on which a concerted mechanism for hemoglobin is founded.
From a direct measurement of the change in /3-p interactions in the presence and absence of 2,3-diphosphoglyceric acid we conclude that, although the binding of DPG acts to stabilize the hemoglobin molecule in the deoxy conformation with respect to Bohr proton release (and to the midpoint of oxygen saturation), DPG does not alter the changes in interactions of the two 6 subunits.