The kinetics of assembly of normal and variant human oxyhemoglobins.

The kinetics of assembly have been monitored spectrophotometrically for normal and variant human oxyhemoglobins in 0.1 M Tris, 0.1 M NaCl, 1 mM Na2EDTA, pH 7.4, at 21.5 degrees C. Oxyhemoglobin versus oxy chain static difference spectra were performed and revealed subtle but significant absorption changes in both the visible and Soret regions. Kinetic experiments were performed by rapidly mixing equivalent (in heme) concentrations of alpha and beta A chains and following the change in absorbance at 583 nm with time. Over a protein concentration range of 10-100 microM in heme prior to mixing, these time courses were homogeneous and followed first-order kinetics, yielding a value of 0.069 s-1 for the apparent rate constant of dissociation of oxygenated beta A chain tetramers. Under these conditions, the overall assembly of oxyhemoglobins S (beta 6Glu----Val) and N-Baltimore (beta 95Lys----Glu) were also governed by the rates of dissociation of their respective oxygenated beta S and beta N-Baltimore chain tetramers with the apparent first-order rate constants of 0.044 and 0.15 s-1, respectively. In the Soret region, the alpha, beta monomer combination reaction could be observed if the protein concentration (heme basis) was lowered and if protein nonequivalency (beta chain exceeded alpha chain concentration) mixing experiments were performed. A kinetic oxyhemoglobin A, oxy-alpha, oxy-beta A monomer difference spectrum could be generated, and simple second-order kinetics were observed (415 nm) yielding rate constants of 2.3, 3.3, and 4.8 X 10(5) M-1 s-1 for the assembly of oxyhemoglobins S, A, and N-Baltimore, respectively. To our knowledge, this is the first kinetic study to reveal significant differences between the rate of association of alpha and beta monomers of hemoglobin A and those of two distinctly charged hemoglobin variants.

The kinetics of assembly have been monitored spectrophotometrically for normal and variant human oxyhemoglobins in 0.1 M Tris, 0.1 M NaC1, 1 mM Na2EDTA, pH 7.4, at 21.5 "C. Oxyhemoglobin versus oxy chain static difference spectra were performed and revealed subtle but significant absorption changes in both the visible and Soret regions. Kinetic experiments were performed by rapidly mixing equivalent (in heme) concentrations of a and @A chains and following the change in absorbance at 583 nm with time. Over a protein concentration range of 10-100 PM in heme prior to mixing, these time courses were homogeneous and followed first-order kinetics, yielding a value of 0.069 s-l for the apparent rate constant of dissociation of oxygenated @A chain tetramers. Under these conditions, the overall assembly of oxyhemoglobins S (@6Glu+Val) and N-Baltimore (@95Lys+Glu) were also governed by the rates of dissociation of their respective oxygenated @' and @N-Ba'timore chain tetramers with the apparent first-order rate constants of 0.044 and 0.15 s-', respectively. In the Soret region, the a,@ monomer combination reaction could be observed if the protein concentration (heme basis) was lowered and if protein nonequivalency @ chain exceeded a chain concentration) mixing experiments were performed. A kinetic oxyhemoglobin A, oxy-a,oxy-@* monomer difference spectrum could be generated, and simple second-order kinetics were observed (415 nm) yielding rate constants of 2.3, 3.3, and 4.8 X 1 0 ' M" s-' for the assembly of oxyhemoglobins S, A, and N-Baltimore, respectively. To our knowledge, this is the first kinetic study to reveal significant differences between the rate of association of a and @ monomers of hemoglobin A and those of two distinctly charged hemoglobin variants.
The kinetics of subunit assembly of normal adult hemoglobin (Hb A) were first investigated 2 decades ago by Antonini et al. (1). They monitored deoxyhemoglobin tetramer assem-* This work was supported by Grants HL 36089 and HL 16927 from the National Institutes of Health and the Howard Hughes Medical Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ bly spectrophotometrically in a stopped flow apparatus by exploiting the known spectral differences between deoxygenated subunits and intact deoxyhemoglobin. Subsequent spectrophotometric (2) and circular dichroism (3) studies have revealed three sequential processes in the reaction mechanism of assembly. A first-order dissociation of oligomeric hemecontaining subunits into a and p monomeric precursors must take place before two consecutive second-order processes can occur. The a and P monomers combine to form a/3 dimers which then self-associate into intact a& hemoglobin tetramers. Experimental variables such as pH, phosphate, and protein concentration were shown to alter the kinetic profile of deoxyhemoglobin assembly and to dictate whether the ratedetermining step is a first-or second-order process.
Since isolated a-and P-heme-containing subunits possess a high affinity for ligand, it is the oxygenated form of the hemoglobin subunit which is almost certainly assembled into tetramers in uiuo. The assembly of liganded Hb A has also been studied and has been found to follow the identical reaction sequence (and apparent dependence on experimental variables) seen for deoxyhemoglobin. Oxyhemoglobin assembly has been monitored by pH changes (4), time-resolved small-angle x-ray scattering techniques (5), and circular dichroism measurements (6, 7), as well as visible absorbance measurements (8)(9)(10).
These visible spectrometry studies, which have been conducted in our laboratory, include the investigation of variant as well as normal oxyhemoglobin assembly. The kinetics of reconstitution of several human hemoglobins were examined to test the hypothesis that macromolecular assembly may be a post-translational determinant of hemoglobin phenotype. Our previous investigations were aimed at defining the firstorder process of oxyhemoglobin assembly and were, therefore, conducted under experimental conditions which promote the stability of the oligomeric structure of liganded ,8 chain subunits. In this report, we extend these studies to include evaluation of the a and p monomer combination process.

EXPERIMENTAL PROCEDURES
Preparation of Human Hemoglobins and Subunits-Erythrocytes from normal donors or from individuals with hemoglobinopathies were washed three times with 0.15 M NaCl and either used immediately or frozen in liquid nitrogen until needed. The washed red blood cells were lysed in distilled water, and the subunits of human Hb A, Hb S, and Hb N-Baltimore were prepared from these hemolysates by appropriate modifications of the method of Bucci and Fronticelli (11) as described in detail elsewhere (12). Both hemoglobins and subunits were freed of phosphate (8) prior to dialysis against the standard experimental buffer (0.1 M Tris, 0.1 M NaC1, 1 mM Na2EDTA, pH 7.4, at 21.5 "C).
Static Difference Spectrophotometry-Static visible and Soret difference spectra of intact oxyhemoglobin uersus its constituent oxy-

5952
Subunit Assembly of genated a and p subunits were generated using a Cary 2200 recording spectrophotometer in the auto-gain mode with a spectral bandwidth of 1.0 nm. Protein solutions of oxyhemoglobin and subunits were made equivalent in the 576 (or 415) nm region of the spectrum just prior to use, and an t value of 15.37 (or 131) mM" cm" was employed to calculate protein concentration on a heme basis. Rectangular tandem-mixing cuvettes containing the appropriate oxygenated a chain, oxygenated PA chain, and oxyhemoglobin A solutions were placed in a temperature-regulated compartment a t 21.5 'C. The resultant spectra were recorded at the absorbance range and scan speed noted under "Results." Spectrophotometric Kinetic Studies-Spectral changes accompanying formation of oxyhemoglobin from its oxygenated a and p subunits were monitored as a function of time in a Gibson-Durrum stopped flow device temperature-regulated a t 21.5 "C. Multiple analyses were carried out on several preparations. Data collection was performed by the microcomputer-based OLIS 3820 system (On-Line Instruments Systems, Jefferson, GA). Absolute static spectra were taken immediately prior to and following each kinetic experiment to confirm protein concentration (on a heme basis) and nativity of the isolated ferrous subunits. Complete recombination of the oxygenated subunits was independently verified by electrophoresis and functional criteria as previously described (8, 12).
Analysis of Spectrophotometric Data-The overall assembly of oxyhemoglobin from its constituent oxygenated a and subunits in vitro can be described by the following reaction sequence:' Human Hemoglobin diate species has allowed us to define precisely two kinetic constants:

k,(P) and k;.
At concentrations where significant oligomeric p chain exists, the rate of dissociation of these oligomers into monomers (kl(P)) was limiting and the kinetics were first-order. An integrated rate equation of the form, log(AA,/hA0) = -(1/2.303) kl(P)t, was used to analyze the spectral data (see "Results").
At low protein concentrations, the a,@ monomer combination reaction ( k ; ) was rate-limiting, and a second-order kinetic profile was observed. These kinetic time courses were analyzed by an integrated rate equation of the form, l/AAt = (CT/AA,,) k ; t + l/AAo, where CT = fCo; f = percent p monomer (derived from K,(P)), and Co = heme concentration after mixing (see "Results"). All data were analyzed by the method of least squares, and standard deviations have been presented.

RESULTS
Visible and Soret Static Difference Spectra-There exist small but significant spectral differences between intact oxyhemoglobin and the oxygenated derivatives of a and PA chains in both the Soret and visible regions of the spectrum (19, 20). Fig. 1 displays the results of static difference spectroscopy of oxyhemoglobin A versus its constituent oxygenated subunits under experimental conditions (0.1 M Tris, 0.1 M NaC1, 1 mM Na,EDTA, pH 7.4, at 21.5 "C) identical to those of kinetic experiments described below. At protein concentrations of 100 and 10 FM (in heme), the visible spectra for oxyhemoglobin A versus oxy-a subunits, versus 0 x y -8~ subunits, and versus oxy-a,oxy-PA subunits are displayed in Fig. 1 ( A and  B ) . The shape and magnitude of the oxyhemoglobin A versus oxy-a subunit are unaltered between 10 and 100 W M (in heme) consistent with previous findings which demonstrated an invariance in the spectrum of oxygenated a chains, as well as that of oxyhemoglobin A with protein concentration (8,18). In contrast, Philo et al. (18) documented concentration-dependent changes in the spectrum of oxygenated PA subunits, and this is readily seen in the oxyhemoglobin A versus oxy-PA chain difference spectra and hence, the oxyhemoglobin A versus oxy-a,oxy-PA subunit difference spectra shown in Fig.  1 ( A and B ) at 100 and 10 /IM (in heme). A At value of 0.18 mM" cm" was seen at 582 and 579 nm for oxyhemoglobin A versus oxy-a,oxy-PA subunit difference spectra at 100 and 10 FM (in heme), respectively.
The Soret spectra for oxyhemoglobin A versus oxy-a subunits, versus oxy-pA subunits, and versus oxy-a,oxy-PA subunits are displayed in Fig. 1 (C and D ) for 10 and 1.0 PM (in heme). In this region of the spectrum, a 10-fold dilution again appears to alter the spectrum of the oxygenated PA subunits, and the oxyhemoglobin A versus oxy-a,oxy-PA subunit difference spectrum is different at 10 and 1.0 FM (in heme), respectively. A At value of 2.5 mM" cm" was observed at 421 and 418 nm for oxyhemoglobin A versus oxy-a,oxy-PA subunit difference spectra at 10 and 1.0 PM (in heme), respectively.
The optical difference spectra presented in Fig. 1 are by no means unique. Knowles et al. (21) showed that a difference spectrum of this type could arise from any number of alterations of structure or solvent conditions which result in a shift in the position of absorption bands. This being the case, it is imperative that difference spectra be generated for the exact experimental conditions utilized in the kinetic experiments The treatment of p chain tetramer kinetics may be simplified by assuming an irreversible dissociation into monomers in the presence of a chains (p4 -2a2 -4P). Then by using steady state approximation with respect to the @ dimer intermediate (i.e. if klb >> k,, and d(P2)/dt = 0), the total reaction may be treated as p4 -4p. This value of k,. is the first-order rate constant apparently measured in this report and will be referred to in the text as k,(B).  . . and furthermore, that there be excellent correlation between measured absorption change and protein concentration. As demonstrated below, this is indeed the case for the studies presented here.
p Tetramer Dissociation Reaction-We have previously shown (8-10) that under appropriate experimental conditions, the dissociation of the oxygenated non-a chain tetramer is the rate-limiting step in the overall assembly of oxyhemoglobin. Here we investigate the assembly of oxyhemoglobins S, A, and N-Baltimore under experimental conditions of 0.1 M Tris, 0.1 M NaC1, 1 mM Na2EDTA, pH 7.4, at 21.5 "C and find that over a protein concentration range of 25-110 PM in heme prior to mixing, the assembly profile follows first-order kinetics (see Fig. 2). Semilogarithmic plots of time courses following the mixing of equivalent concentrations (on a heme basis) of oxygenated a and PA chains are shown in Fig. 2B.
These time courses are homogeneous and followed first-order kinetics, yielding a value of 0.069 (k0.0073) s" for the appar- displayed is for a 2-cm path length; a A6 value (consistent with that in Fig. 1, A and B ) of 0.18 mM" cm" at 583 (+3) nm for the DA and Os tetramer dissociation reaction was obtained. (The AA value accommore dissociated into monomer at any given protein concentration (lo).) The first-order rate constants presented in the text are the result of a minimum of five independent determinations of five kinetic runs each.
chain tetramers of Hb A (kl(pA)). This value agrees remarkably well with that of Kawamura and Nakamura (7) who reported a value of 0.064 (k0.0072) s" for the oxygenated PA chain tetramer dissociation rate under identical experimental conditions using a circular dichroism stopped flow technique (see "Discussion"). In addition, the absorbance changes ( Fig.  2B) accompanying the assembly of oxyhemoglobin A from its constituent oxygenated subunits correlate well with the static difference spectrometry experiments presented above (see Fig.  1A). Both the static and kinetic spectral measurements yield a millimolar difference extinction coefficient value (at) of 0.18 m"' cm" at 583 nm.
The overall assembly of oxyhemoglobins S and N-Baltimore (Fig. 2, A and C) is also governed by the rates of dissociation of their respective oxygenated ps and /3N-Ba'timore chain tetramers. The apparent first-order rate constants k ( Ps) and k ( PN-Ba'timore ) were 0.044 (k0.0069) and 0.15 (k0.032) s-', respectively. Thus, under conditions where the /3 tetramer dissociation reaction is rate-limiting, the assembly of oxyhemoglobin S precedes 1.6-fold slower than normal oxyhemoglobin A assembly and the reconstitution of oxyhemoglobin N-Baltimore is 2.2-fold more rapid than oxyhemoglobin A assembly. These ratios for P tetramer dissociation rates are similar to those reported previously (10) under experimental conditions of 0.1 M potassium phosphate buffer, pH 7.0, at 20 "C.
In an effort to monitor a subunit assembly reaction which was subsequent to the p tetramer dissociation reaction, namely the second-order monomer combination step, we focused our attention on mixing experiments which could be conducted in the Soret region at protein concentrations 510 p~ in heme prior to mixing. Normalized semilogarithmic plots of time courses following mixing of equivalent concentrations This combination reaction was monitored over the spectral region from 380 to 440 nm. Time courses at several wavelengths yielded k; values identical to that seen in Fig. 3B, and a kinetic difference spectrum for this monomer combination reaction was developed (see Fig. 4, A and B). A value of kL(Hb A) of 3.3 (f0.42) X lo5 M-' s-l was observed. This kinetic difference spectrum is similar to the static Soret spectra depicted in Fig. 1C and can be attributed to a concentration-dependent spectral change for oxygenated PA chains, The A6 value calculated at the peak region (418 k 3 nm) was 2.5 mM" cm".
The rate of association of cx and PA monomers at 415 nm was followed at five distinct concentrations (Fig. 5 A ) . By limiting the concentration of (Y chains, a monomer equivalency experiment could be attempted. Under the present experimental conditions, the concentration of a chain monomer exceeds that of /3 chain monomer, and a deviation from simple second-order kinetics would be expected. This is indeed the case, as seen in Fig. 5A. As the time courses develop, the fitted and experimental curves diverge. Exclusive of curue I , the error introduced by the presence of a possible slow phase (first-order / 3 tetramer reaction) is approximately 10%. This error is very tolerable and within experimental range for the absorbance changes seen in these experiments. In the case of curue I, a greater error, nearly 20%, is possible. (Nevertheless, a consistent difference (see Fig. 5B) is seen between assembly rates of Hb A, Hb S, and Hb N-Baltimore.) The time courses and second-order rate plots reveal a second-order rate constant (within experimental error) identical to that calculated in Fig. 4.
The overall rates of assembly of oxyhemoglobins S and N-Baltimore (Fig. 5B) are also governed by the rates of combination of oxygenated a chain Monomers with oxygenated Ps chain monomers and pN-Ba'timore chain monomers, respectively.
The apparent second-order rate constants k;(Hb S) and &(Hb N-Baltimore) were 2.3 (k0.13) and 4.8 (f0.30) X lo5 M" s-l, respectively. The reconstitution of oxyhemoglobin S proceeds 1.4-fold slower than that of normal oxyhemoglobin A, and the assembly of oxyhemoglobin N-Baltimore is 1.5fold more rapid than oxyhemoglobin A assembly. These findings constitute the first kinetic measurements of variant human hemoglobin assembly and are discussed in detail below.

DISCUSSION
Of prime importance is the origin of the spectral change accompanying hemoglobin assembly. The difference spectrum for oxyhemoglobin A and its constituent chains is distinct from that reported for deoxyhemoglobin A and its constituent subunits. The reconstitution of unliganded hemoglobin yields color changes that are 2-fold greater in the visible and at least 5-fold larger in the Soret region of the spectrum than the corresponding changes accompanying liganded hemoglobin formation (20). This larger color change is what prompted earlier studies of assembly of deoxyhemoglobin subunits by Antonini et al. (1). McGovern et al. (2), upon reinvestigation of the assembly process, found that the majority of absorbance change following mixing of unliganded subunits was associated with the (YP dimer to a& tetramer aggregation step. In direct contrast to that seen for the assembly of deoxyhemoglobin, no absorbance change is associated with the cy0 dimer aggregation step of liganded hemoglobin formation. Both this laboratory and that of Philo et al. (18) reported an invariance in the spectrum of liganded Hb A with concentration. This means that the formation of liganded hemoglobin involves two and only two possible color-producing reactions, one of which is first-order (PA tetramer dissociation) and the other of which is second-order ( a and PA monomer combination). Our results indicate that the overall color change in both kinetic instances is attributed to a change in the oxygenated PA chain tetramer with concentration.
A concentration-dependent spectral change for oxygenated PA chains in the standard buffer system employed here has been previously demonstrated by Philo et al. (18). These workers reported a red-shifted spectrum for the PA tetramer to PA monomer reaction. We have found a 3-nm red shift for the oxyhemoglobin A versus oxygenated PA spectrum and a corresponding 3-nm blue shift for the oxyhemoglobin A versus oxygenated n,PA chain spectrum upon protein dilution. Furthermore, the spectra depicted in Fig. l (A and B)  and oxyhemoglobin N-Baltimore (W) assembly. Protein concentrations correspond to those in A. A solid line in the reciprocal plots represents the oxyhemoglobin A control. Values of C T were calculated (as described under "Experimental Procedures") from a previously reported equilibrium value (10) for os and estimated for pN-Baltimore from absorbance changes. This was appropriate since only the pN-Be'timom monomer species is being reacted, and the accompanying absorbance change should be similar to pS and PA monomer combination into hemoglobin.
tatively similar to those we reported (8) in 0.1 M potassium phosphate buffer, pH 7.0, at 20 "C except that the Ac value reported here is 2-fold less. We attribute this difference in the value of At to a phosphate-induced spectral change in the PA chain. Phosphate-induced spectral changes for oxyhemoglobin A have been reported (21).
The effect of ligation on the rates of subunit assembly may also be related to the self-association properties of the P  globin from isolated human carbon monoxy-a and -PA chains by time-recorded small-angle x-ray scattering techniques. They found that at pH 7.4 in 0.1 M potassium phosphate buffers and a protein concentration of 2 mM (in heme), the overall kinetic profile of assembly followed first-order kinetics. They attributed this process to the dissociation of the PA tetramer to monomer. This agrees with previous findings (6,8,9) which demonstrate that at high protein concentrations the PA tetramer dissociation reaction can be rate-limiting in the assembly of hemoglobin. Kawamura   Subunit Assembly of Human Hemoglobin concentrations (heme basis) of a and PA chains, that the time courses of the circular dichroism change in the Soret region exhibited both a rapid and slow phase. In 0.1 M potassium phosphate buffer, pH 7.5, at 25 "C (6), this slow phase was a first-order reaction with a rate constant of 0.0028 s" and was attributed to the dissociation of the PA chain tetramers. The rapid phase assigned to the combination of oxy-a and -P monomers to dimers yielded a second-order rate constant of 7.5 x LO5 M" s -I . More recent studies by these workers (7) in the buffer system employed here have shown an even greater rate of assembly (1.9 X lo6 M" S K I ) .
Our spectrophotometric studies here on nonequivalent mixtures of oxygenated cy and PA chains indicate a ki value for oxyhemoglobin A nearly 10-fold less than that reported in this identical buffer system by Kawamura and Nakamura (7). This is especially noteworthy since our value of k,(pA) was identical to that found by these workers (see "Results" and Fig. 2 8 ) . The two spectroscopic techniques may be probing different stages of assembly. Circular dichroism captures any alteration in secondary structure. Such structural perturbations could occur in a region of the protein remote from the heme pocket and coud take place prior to the heme pocket, itself, undergoing a conformational change. It would be this later event at the heme environment which our spectrophotometric technique would detect. This could account for the differences in k; values obtained by these two distinct techniques. Nevertheless, it would be of special interest if the assembly of several variant hemoglobins could be monitored by circular dichroism.
A recent report from our laboratory (23) has shown that electrostatic attraction may govern oxyhemoglobin assembly. Although direct measurements of rate constants are not possible with chain competition experiments, nonetheless, relative rates of monomer association can be estimated. The ratio of experimental values of ki(Hb X)/ki (Hb A) was 2.6 and 0.41 for Hb N-Baltimore and Hb S, respectively. These ratios are in qualitative agreement with the direct kinetic measurements reported here, which yielded ratios of 1.5 and 0.66, respectively. Quantitative differences are most likely attributed to different buffer, pH, and temperature conditions. Although the technique we employed here has an inherent error due to the small color changes seen upon oxyhemoglobin assembly (and large deviations are presented for any given value of ki), paired kinetic experiments between normal and variant P subunits nevertheless showed a consistent, reproducible difference in monomer assembly rates,