Báo cáo Y học: The a1b1 contact of human hemoglobin plays a key role in stabilizing the bound dioxygen Further evidence from the iron valency hybrids potx

The a1b1 contact of human hemoglobin plays a key role in stabilizing the bound dioxygen Further evidence from the iron valency hybrids Jun pei Yasuda1, Takayuki Ichikawa1, Mie Tsuruga1,

The a1b1 contact of human hemoglobin plays a key role in stabilizing the bound dioxygen

Further evidence from the iron valency hybrids

Jun pei Yasuda1, Takayuki Ichikawa1, Mie Tsuruga1, Ariki Matsuoka2, Yoshiaki Sugawara3and Keiji Shikama1,4

1 Biological Institute, Graduate School of Science, Tohoku University, Sendai, Japan; 2 Fukushima Medical University, Fukushima, Japan; 3 Hiroshima Prefectural Women's University, Hiroshima, Japan; 4 PHP Laboratory for Molecular Biology, Sendai, Japan

When the a and b chains were separated from human

oxyhemoglobin (HbO2), each individual chain was oxidized

easily to the ferric form, their rates being almost the same

with a very strong acid-catalysis In the HbO2tetramer, on

the other hand, both chains become considerably resistant to

autoxidation over a wide range of pH values (pH 5±11)

Moreover, HbA showed a biphasic autoxidation curve

containing the two rate constants, i.e kffor the fast oxidation

due to the a chains, and ksfor the slow oxidation to the

b chains The kf/ksratio increased from 3.2 at pH 7.5±7.3 at

pH 5.8, but became 1 : 1 at pH values higher than 8.5 In the

present work, we used the valency hybrid tetramers such as

(a3+)2(bO2)2and (aO2)2(b3+)2, and demonstrated that the

autoxidation rate of either the a or b chains (when O2 -ligated) is independent of the valency state of the corre-sponding counterpart chains From these results, we have concluded that the formation of the a1b1 or a2b2 contact suppresses remarkably the autoxidation rate of the b chain and thus plays a key role in stabilizing the HbO2tetramer Its mechanistic details were also given in terms of a nucleophilic displacement of O2 from the FeO2center, and the emphasis was placed on the proton-catalyzed process performed by the distal histidine residue

Keywords: Hb oxidation; chain nonequivalence; valency hybrids; a1b1 contact; acid-catalysis

The reversible and stable binding of molecular oxygen to the

heme iron(II) is the basis of hemoglobin function However,

the oxygenated form of hemoglobin, as well as of

myoglo-bin, is known to be oxidized easily to the ferric met-form,

which cannot bind dioxygen and is therefore physiologically

inactive, with generation of the superoxide anion [1±7] In

this reaction, human oxyhemoglobin (HbO2) shows a

biphasic autoxidation curve containing the two rate

con-stants, a fast one due to the a chains and a slow one for the b

chains, respectively [3] Such chain heterogeneity could be

maintained even in the low concentrations of hemoglobin

corresponding to appreciable dissociation into a1b1 and

a2b2 dimers [5] When the a and b chains are separated

from the HbO2tetramer, both chains were oxidized much

more rapidly than those in the tetrameric parent, and

become freed from their rate differences

of pH 5±11 [8] These recent new ®ndings have led us to

conclude that the formation of the a1b1 or a2b2 contact

produces a conformational constraint in the b chain

where-by the distal (E7) histidine at position 63 is tilted slightly

away from the bound dioxygen, so as to prevent the

proton-catalyzed displacement of O2 from the FeO2center by an

entering water molecule The b chains have thus acquired a

remarkably delayed oxidation rate in the HbO2tetramer, and this is the origin of such chain heterogeneity found in the hemoglobin autoxidation at acidic pH [8]

To further characterize the nature of the a1b1 or a2b2 interface in stabilizing the heme-bound dioxygen, we have constructed iron valency hybrid hemoglobins, and studied their autoxidation behavior at several different pH values as compared with the native or reconstructed HbO2 Such examinations seem to be of primary importance, not only for a full understanding of the molecular mechanism of hemoglobin autoxidation, but also for planning new molecular designs for synthetic oxygen carriers that are highly resistant against the heme oxidation under physio-logical conditions Finally, we will revisit the hemoglobin function as seen from the two different types of the ab contact, and try to reconcile the cooperative oxygen binding with the stabilization of the bound dioxygen With respect

to this, we will also give possible implications for the unstable hemoglobin mutants leading to the formation of Heinz bodies in red blood cells, resulting in hemolytic anemia

M A T E R I A L S A N D M E T H O D S Chemicals

Sodium p-hydroxymercuribenzoate (p-MB) was from

Sig-ma Mes, Mops, Hepes, Tris and Caps for buffer systems, 2-mercaptoethanol, and all other chemicals were of reagent grade from Wako Pure Chemicals, Osaka Solutions were made with deionized and glass-distilled water

Correspondence to K Shikama, PHP Laboratory for Molecular

Biology, Nakayama-Yoshinari 1-16-8, Sendai 989±3203, Japan.

Fax: + 81 22 278 6180, E-mail: shikama@mail.cc.tohoku.ac.jp

Abbreviations: p-MB, sodium p-hydroxymercuribenzoate.

(Received 22 August 2001, revised 23 October 2001, accepted 29

October 2001)

Preparation of human oxyhemoglobin

Human hemoglobin A was prepared from freshly drawn

blood (30 mL each time) by the method of Williams &

Tsay [9], with our previous speci®cations [5,8] The major

band of HbA, which was developed on a DEAE-cellulose

column (3.5 ´ 12 cm), was eluted out completely with

20 mM Hepes buffer at pH 7.9 The HbO2 solution was

then concentrated by centrifugation in a Centriprep-10

tube (Amicon), and kept at low temperature (4 °C) until

use The concentration of hemoglobin was determined as

heme, after conversion into cyanomet form, using the

absorption coef®cient of 10.4 mM)1ácm)1at 540 nm This

value was obtained on the basis of the pyridine

hemo-chromogen method [10]

Isolation of mercuribenzoated a and b chains

All separations were carried out with fresh HbO2solutions

at low temperature (0±4 °C) by a two-column method The

procedure was essentially the same as described by Geraci

et al [11] and by Turci & McDonald [12], with our previous

speci®cations [8] Each time, p-MB (100 mg) was dissolved

in 2 mL of 0.1M NaOH and neutralized with 1M

CH3COOH This was reacted with 10 mL of HbO2solution

(4±7 mMas heme) in 50 mMphosphate buffer, pH 6.0, and

in the presence of 0.1 M NaCl After passing through a

Sephadex G-25 column (2.5 ´ 40 cm), the mercurated

HbO2 solution was applied on a DEAE-cellulose column

(3.5 ´ 12 cm) to elute out the ap-Mb chains, or on a

CM-cellulose column (3.5 ´ 12 cm) for the bp-Mbchains In

each case the counterpart chain had remained on the top of

the column

Regeneration of SH groups from mercuribenzoated

a and b chains

To recover sulfhydryl groups from the mercuribenzoated

protein, 75 mL of the ap-Mbor bp-Mbsolution ( 200 lMas

heme) were treated with 20 mM 2-mercaptoethanol for

10 min in 10 mM phosphate buffer at 0 °C, as described

previously [8] The mixture ( 150 mL) was placed on a

CM-cellulose column (2.5 ´ 6 cm) for the a chain, or on a

DEAE-cellulose column (5 ´ 6 cm) for the b chain, to

remove excess amounts of the reagent After washing each

column with a large volume of the buffer alone, the

regenerated a or b chains were eluted out completely as the

oxy-form by changing the buffer, and kept stably in liquid

nitrogen until use The concentration of each separated

chain was determined, after conversion into

cyanomet-form, using the following absorption coef®cients at 540 nm:

10.5 mM)1ácm)1for the a chain and 11.2 mM)1ácm)1for the

b chain These values were obtained on the basis of the

pyridine hemochromogen method [10]

Titration of SH groups

According to the method of Boyer [13], free sulfhydryl

groups of the regenerated a or b chains were titrated

spectrophotometrically at 250 nm with

p-hydroxy-mercuribenzoate in 0.1MMops buffer, pH 7.0 The

result-ing contents were 1.0 (1.05 ‹ 0.08) for the a chain and 2.0

(2.01 ‹ 0.08) for the b chain, respectively, as might be

expected from the number of cysteines located at positions a104(G11), b93 (F9) and b112(G14) for HbA

Preparation of valency hybrid hemoglobins from separated a and b chains

Reconstructed HbA (aO2)2(bO2)2 A 2-mL solution of the oxygenated a chain ( 300 lM) was mixed with an equal volume of the b chain ( 300 lM) in 10 mM phos-phate buffer at pH 6.8 The mixed solution was then applied

to a CM-cellulose column (2.5 ´ 3 cm) equilibrated with the same buffer After a small peak of unassociated bO2chains passed through the column, the buffer was changed to

20 mM Hepes (pH 7.9) to completely elute out the major peak of the reconstructed HbO2 Under this condition, a small quantity of unassociated aO2chains remained on the top of the column

Valency hybrids (a3+)2(bO2)2 and (aO2)2(b3+)2 For conversion of the separated a or b chains from the oxy form

to the ferric met-form, 2.5 mL of each solution ( 0.5 mM

as heme) were oxidized with 1.5 mMpotassium ferricyanide

in 0.1Mphosphate buffer, pH 6.8, and in the presence of 10% (v/v) glycerol To remove the residual oxidizing agent, the resultant met-species was immediately passed through a Sephadex G-25 column (2.5 ´ 10 cm) equilibrated with

10 mMphosphate buffer, pH 6.8 All these procedures were carried out at low temperature (0±4 °C) to avoid hemi-chrome formation as well as protein denaturation In preparing the valency hybrid (a3+)2(bO2)2, a 1.8-mL solution of ferric a chains ( 150 lM) was mixed with

360 lL of bO2solution ( 750 lM) The resultant mixture was then applied to a CM-cellulose column (2.5 ´ 3 cm) equilibrated with 10 mMphosphate buffer, pH 6.8 After a small quantity of unassociated bO2chains passed through the column, the buffer was changed to 50 mM Hepes (pH 7.9) to elute out the major peak of the hybrid tetramer (a3+)2(bO2)2 Under this condition, unassociated a3+

chains had remained on the top of the column Essentially the same procedure can be used for the preparation of another hybrid (aO2)2(b3+)2 In this case, a small quantity

of unassociated b3+chains passed through a CM-cellulose column (2.5 ´ 3 cm) with 10 mMphosphate buffer, pH 6.8 The hybrid tetramer was then eluted out completely by changing the buffer to 20 mMHepes, pH 7.9

Autoxidation rate measurements According to our previous procedures [5,8], the rate of autoxidation of HbA and its derivatives was measured at

35 °C in 0.1M buffer containing 1 mM EDTA To meet various hemoglobin concentrations required, a 1-cm cell was used for a 3-mL sample containing 10±50 lM heme, while a 1-mm cell was employed for a 0.5-mL sample containing 300 lM heme For spectrophotometry, the reaction mixture was quickly transferred to a quartz cell held at 35 ‹ 0.1 °C, and changes in the absorption spectrum from 450 to 700 nm were recorded on the same chart at measured intervals of time For separated a and

b chains, the rate measurement was usually carried out with

10 lMprotein (as heme) and in the presence of 20% (v/v) glycerol As the ®nal state of each run, the hemoglobin was completely converted to the ferric met-form by the addition

of potassium ferricyanide The buffers used were Mes,

maleate, Mops, and Caps The pH of the reaction mixture

was carefully checked, before and after the run, with a

Hitachi±Horiba pH meter (Model F-22)

Spectrophotometric measurements

Absorption spectra were recorded in a Hitachi

two-wave-length double-beam spectrophotometer (model 557, U-3210

or U-3300) or in a Beckman spectrophotometer (model

DU-650), each being equipped with a thermostatically

controlled (within ‹ 0.1 °C) cell holder

Curve ®ttings

Biphasic autoxidation curves were analyzed by an iterative

least-squares method on a computer (NEC PC-9821 V12)

with graphic display, according to our previous

speci®ca-tions [5,8]

R E S U L T S

Biphasic nature of the autoxidation reaction

for human HbO2

In air-saturated buffers, the oxygenated form of HbA is

oxidized easily to the ferric met-form (metHb) with

generation of the superoxide anion [1,2],

HbO2ƒƒƒƒ!kobs

metHb ‡ Oÿ

where kobsrepresents the ®rst-order rate constant observed

at a given pH in terms of the constituent chains This

autoxidation reaction can be monitored by the spectral

changes with time, after fresh HbO2 was placed in 0.1M

buffer containing 1 mM EDTA at 35 °C The spectra

evolved to the ®nal state, which was identi®ed as the usual

ferric met-form, with a set of isosbestic points

Consequent-ly, the process was followed by a plot of experimental data

as ±ln([HbO2]t/[HbO2]0) vs time t, where the ratio of HbO2

concentration after time t to that at time t ˆ 0 can be

obtained by the absorbance changes at 576 nm for the

a-peak of human HbO2

Figure 1 shows such examples of the ®rst-order plot for

the autoxidation reaction of human HbO2at two different

pH values At pH 6.2, HbA showed a biphasic curve that

can be described completely by the ®rst-order kinetics

containing two rate constants as follows:

‰HbO2Št

‰HbO2Š0 ˆ P
exp…ÿkf
t† ‡ …1 ÿ P†
exp…ÿks
t† …2†

In this equation, a fast ®rst-order rate constant kf is

attributed to the a chains and a slow rate constant ksis for

the b chains in the HbO2tetramer P is the molar fraction of

the rapidly reacting hemes This conclusion is based on the

rapid chain separation experiment of partially (30%)

oxidized HbO2on a 7.5% polyacrylamide gel [8], this being

in good agreement with that of Mansouri & Winterhalter

[3]

By iterative least-squares procedures inserting various

values for kf and ks into Eqn (2), the best ®t to the

experimental data was obtained as a function of time t In

these computations, the initial value for each of the rate constants was taken from the corresponding slope of a biphasic curve (as delineated in Fig 1 by two dotted lines), and was re®ned by the step sizes of 0.01±0.001 h)1to ®nd out the best values of kfand ks, according to our previous speci®cations [5] The value of P was also allowed to vary a large range (from 0.40 to 0.60) in all cases In this way, the following parameters were established at pH 6.2;

kfˆ 0.82 ‹ 0.03 ´ 10)1h)1, ksˆ 0.13 ‹ 0.01 ´ 10)1h)1, and P ˆ 0.52 ‹ 0.04 in 0.1M Mes buffer at 35 °C At

pH 9.2, on the other hand, the reaction could be described completely by a single ®rst-order rate constant of 0.99 ‹ 0.02 ´ 10)2h)1 (i.e kfˆ ks with P ˆ 0.50) in 0.1MCaps buffer at 35 °C

Table 1 represents such examples for a pair of the ®rst-order rate constants deduced from each autoxidation curve

at different values of pH From the kf/ksratios, one can easily realize the biphasic nature emerged in the autoxida-tion of HbA Moreover, we have found that such chain heterogeneity can be kept even in very diluted concentra-tions of hemoglobin from 1.0 ´ 10)3 Mto 3.2 ´ 10)6 Mas heme [5] When the HbO2sample is diluted

species is known to dissociate into ab dimers along the a1b2

or a2b1 interface, so that the dimers formed are of the a1b1

or a2b2 type [14,15] From these results, we can unequiv-ocally conclude that the remarkable stability of the b chain

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

]t

]0

Time (h)

pH 9.2

pH 6.2

HbA(α2β2 )

Fig 1 First-order plots for the autoxidation reaction of human HbO 2 in 0.1 M bu€er at 35 °C The rate measurements were carried out with

75 l M HbO 2 (300 l M as heme) in the presence of 1 m M EDTA Each curve (±±) was obtained by a least-squares ®tting to the experimental points (s), based on Eqn (2) At pH 6.2, HbA showed a biphasic autoxidation curve containing two rate constants, k f and k s , respec-tively At pH 9.2, however, the reaction was monophasic The bu€er used was Mes for pH 6.2 and Caps for pH 9.2.

against the acidic autoxidation must have been produced by

the formation of the a1b1 or a2b2 contact To see more

quantitatively the effect of the a1b1 or a2b2 contact on the

autoxidation reaction, our next step was to construct the

iron valency hybrid tetramers containing either the a or

b chains in the ferric state, and to examine for their stability

properties as compared with the native HbO2 and its

separated chains

Preparation of the valency hybrid hemoglobins

and their autoxidation behavior

By mixing equivalent amounts of the separated a and b

chains whose sulfhydryl groups were completely recovered,

we have prepared the reconstructed HbO2and its valency

hybrid tetramers Figure 2 shows such an example for the

chromatographic separation of the hybrid tetramer

(a3+)2(bO2)2 from unassociated chains In this case, the

mixed chain solution (2.2 mL) was applied to a

CM-cellulose column (2.5 ´ 3 cm) that had been equilibrated

with 10 mMphosphate buffer, pH 6.8 After a small band of the unassociated bO2passed through the column, the buffer was changed to 50 mMHepes, pH 7.9, to obtain the major peak of the hybrid tetramer

When the iron valency hybrids are placed in air-saturated buffers, the oxygenated chains of each tetramer are oxidized easily to the ferric met-form Figure 3 shows such an example of the spectral changes with time for the autoxi-dation reaction of hybrid Hb (a3+)2(bO2)2 in 0.1M Mes buffer pH 6.2, and in the presence of 1 mMEDTA at 35 °C

In this tetramer, even if freshly prepared, the a-peak (at

577 nm) was always lower than the b-peak (at 541 nm) with

an absorbance ratio of a/b ˆ 0.90, this being in contrast to

a value of 1.06 for the native or reconstructed HbO2 The

Table 1 Comparison of the two rate constants involved in the autoxidation reaction of human HbO 2 at various pH values and 35 °C.

pH

k obs (h )1 )

k f /k s

Concentration (l M as heme)

6.2 0.82 ´ 10 )1 0.13 ´ 10 )1 6.3 300

6.5 0.56 ´ 10 )1 0.90 ´ 10 )2 6.2 300

7.5 0.16 ´ 10 )1 0.50 ´ 10 )2 3.2 300

9.0 0.48 ´ 10 )2 0.48 ´ 10 )2 1.0 300

9.2 0.99 ´ 10 )2 0.99 ´ 10 )2 1.0 300

0

2

4

6

8

10

12

Fraction Number (4 ml / tube)

0 2 4 6 8 10 12

A280

CM-cellulose Valency hybrid

A415

3+

O 2

)

Fig 2 CM-cellulose chromatography of the valency hybrid

(a 3+ ) 2 (bO 2 ) 2 tetramer The equimolar mixture (2.2 mL) of the a 3+ and

bO 2 chains was applied to a CM-cellulose column (2.5 ´ 3 cm)

equilibrated with 10 m M phosphate bu€er, pH 6.8 A small band of

unassociated bO 2 chains passed through the column with the same

bu€er To elute out the major peak of the hybrid tetramer, the bu€er

was changed to 50 m M Hepes (pH 7.9) at the point indicated by the

®rst arrow The unassociated a 3+ chains could be removed by the

addition of 1 M NaCl as indicated by the second arrow The protein

and the heme protein levels were monitored by the absorbances at

280 nm (s) and 415 nm (d), respectively.

0 0.1 0.2 0.3 0.4 0.5

Wavelength (nm)

pH 6.2

Valency hybrid

Start

Finish

(α3+) (βO 2)

Fig 3 Spectral changes with time for the autoxidation of valency hybrid

Hb (a 3+ ) 2 (bO 2 ) 2 in 0.1 M Mes bu€er at pH 6.2 and 35 °C Scans were made at 270-min intervals in the presence of 1 m M EDTA The ®nal spectrum was for the acidic metHb with a set of isosbestic points at 526 and 592 nm Hb concentration: 50 l M as heme.

hybrid tetramer also exhibited very intensive charge-transfer

bands at 501 nm as well as 631 nm All these features

seemed to be produced by a spectral overlapping of ferric

a3+chains Furthermore, the reaction spectra evolved to the

®nal state of a run, which was identi®ed as the usual acidic

(or aquo) metHb

If the contribution of ferric a3+ chains could be

subtracted from the oxidation spectra of the (a3+)2(bO2)2

tetramer on a computer, we may have the spectral changes

that can be ascribed to the autoxidation of the b chains

alone Such computations have disclosed that the reaction

started from the fully oxygenated b chains with an

absor-bance ratio of a/b ˆ 1.05, and that the oxidation proceeded

to the usual acidic met-form with a set of isosbestic points at

526 and 592 nm, as depicted in Fig 4 This process was

therefore followed by absorbance changes at 578 nm for the

a-peak of the b chain, and could be described completely by

a single ®rst-order rate constant of kobsˆ 0.19 ´ 10)1h)1

This oxidation rate is essentially the same with that of the

b chains in the HbO2tetramer (see Fig 1)

In separated chain solutions, the protein is known to

exist in an equilibrium of a ÿÿ*)ÿÿ a2or b ÿÿ*)ÿÿ b4,

respec-tively Under our experimental conditions (10±25 lM as

heme), the monomeric form (87%) was predominant in the

a chain, while the tetrameric form (99%) was for the

b chain This estimation was made on the basis of the

results by McDonald et al [16] In a previous paper [8], we

have reported that the separated a and b chains are both

oxidized much more rapidly than those in parent HbO2

tetramer over the wide range of pH 5±10 Figure 5 shows such spectral changes with time for the autoxidation of separated bO2chains in 0.1Mmaleate buffer, pH 6.2, and

in the presence of 1 mMEDTA plus 20% (v/v) glycerol at

35 °C The oxidation began with an absorbance ratio of a/

b ˆ 1.04, and proceeded very rapidly with a ®rst-order rate constant of kobsˆ 0.10 h)1 This rate is several times higher than the corresponding ks value for the b chains either in the hybrid tetramer (a3+)2(bO2)2or reconstructed HbO2 Moreover, the ®nal state of the run was not for the usual acidic met-form but for an admixture with hemi-chrome

For the oxidation product of separated b chains, we have already carried out 8K EPR analysis in 10 mM maleate buffer at pH 6.2 [8] In addition to a high spin EPR spectrum attributed to the usual aqua-met species with g values of 5.86 and 1.99, the b chains exhibited a low spin spectrum with g1ˆ 2.77, g2ˆ 2.27, and g3ˆ 1.68, which differentiates this species from that of the hydroxide-type complex According to Rifkind et al [17], such low spin complexes characterized by the highest g value in the range

of 2.83±2.75 and the lowest g value in the range of 1.69±1.63 have been designated as complex B, indicating crystal ®eld parameters of the reversible hemichrome They also suggest that the bis-histidine complex B may still have a water molecule retained in the heme pocket, and therefore in solution it is in rapid equilibrium with the high spin

aquo-0

0.1

0.2

0.3

0.4

Wavelength (nm)

2 in

pH 6.2

Start

Finish

(βO2) ( α 3+ ) ( β O2)2

Fig 4 Spectral changes with time for the autoxidation of the b chains of

valency hybrid Hb (a 3+ ) 2 (bO 2 ) 2 in 0.1 M Mes bu€er at pH 6.2 and

35 °C Spectral subtraction of the (a 3+ ) 2 part from the hybrid Hb was

made at 270-min intervals on a computer The b chains were found to

oxidize from the fully oxygenated form to the usual acidic met-form.

Heme concentration: 25 l M

0 0.1 0.2 0.3 0.4

Wavelength (nm)

4

pH 6.2

Finish

Start

(βO2)

Fig 5 Spectral changes with time for the autoxidation of separated b chains in 0.1 M maleate bu€er at pH 6.2 and 35 °C Scans were made at 70-min intervals in the presence of 1 m M EDTA and 20% (v/v) glyc-erol The ®nal spectrum was not for the acidic met-form, but an ad-mixture with hemichrome having a peak at 530 nm and a shoulder near 560 nm Heme concentration: 25 l M

complex [18] As shown in Fig 5, the molar fraction of the

hemichrome (complex B) was estimated to be 75% at

pH 6.2 Furthermore, Borgstahl et al [19] reported the

1.8 AÊ structure of carbonmonoxy-b4 (COb4) tetramer of

human hemoglobin, and compared subunit±subunit

con-tacts between three types of interfaces (a1b1, a1b2 and

a1a2) of HbO2and the corresponding COb4interfaces As a

result, they found that, in contrast to the stable b1b4

interface, the b1b2 interface of the COb4 tetramer is less

stable and more loosely packed than its a1b1 counterpart in

HbO2

At all rates, the present spectral examinations clearly

indicate that the formation of the a1b1 or a2b2 contact

suppresses remarkably the acidic autoxidation of the

b chain, and prevents its hemichrome formation as well

This is true no matter which valency state the partner

a chains may take, the oxy-form or the ferric met-form

Unlike separated b chains, the spontaneous formation of

hemichrome was at variance with separated a chains in the

pH range 5±10 Therefore, in another valency hybrid

(aO2)2(b3+)2 the oxidation of the a chains was found to

start with an absorbance ratio of a/b ˆ 1.07 and to proceed

as usual as in the HbO2tetramer Under our experimental

conditions, the valency hybrid hemoglobins were suf®ciently

stable for the rate measurements over a long period of time

In separated a and b chains, on the other hand, the addition

of 20% (v/v) glycerol was most effective in preventing

occasional precipitations

Kinetic analysis of the autoxidation reaction

of valency hybrid hemoglobins

Figure 6 represents ®rst-order plots to show wide differences

in the autoxidation rate of the b chain, when it exists as the

separated (bO2)4, valency hybrid (a3+)2(bO2)2, and

recon-structed HbO2tetramers in 0.1MMes buffer at pH 6.2 and

35 °C In this way, all the spectrophotometric data were

subjected to ®rst-order kinetics using Eqn (2) The resulting

rate constants for the native, separated, reconstructed, and

valency hybrid hemoglobins are summarized in Tables 2±4

at three different values of pH At pH 6.2, for example, the

HbO2 tetramer exhibited a biphasic autoxidation curve

with the rate constants of kfˆ 0.82 ´ 10)1h)1 and

ksˆ 0.13 ´ 10)1h)1 Almost the same oxidation rates were

obtained for the reconstructed HbO2 giving a value of

kf/ksˆ 6.1 Among those, the most remarkable effect was

found on the b chain Separated b chains undergo quite

rapid oxidation with a rate constant of kobsˆ 0.10 h)1, but

this inherent rate was dramatically suppressed in the

reconstructed as well as the native HbO2 More importantly,

such a retarded ksvalue could be kept almost completely in

the valency hybrid (a3+)2(bO2)2 tetramer, too All these

features were essentially the same at other pH values as seen

in Tables 3 and 4 Certainly, the biphasic nature of the

autoxidation rate became much less steep at pH 7.5, and

even disappeared at pH 9.0 Nevertheless, the rate of

oxidation of the separated b chain was markedly reduced

by up to 15-fold at pH 7.5 and up to 23-fold at pH 9.0 in the

tetrameric hemoglobin, either it is native or reconstructed or

valency hybrid species

The similar situation was also found in the a chain, but its

effect on the HbO2tetramer was much less crucial than the

b chain At pH 9.0, the rate of oxidation of the separated

a chain was reduced by up to 16-fold in the HbO2tetramer, but such a rate suppression was decreased with increasing hydrogen ion concentration This is due to the fact that the

a chain exhibits a very strong proton-catalysis not only in the separated chain but also in the HbO2tetramer Among those, an unexpected result was found at pH 6.2 At present,

we do not know exactly why the oxidation rate of the

a chains was more suppressed in the hybrid tetramer (aO2)2(b3+)2than in the native HbO2 The most probable case was in the hemichrome formation that might have occurred in part to the b3+chains when preparing the corresponding valency hybrid However, it should be noted that such a valency state can never occur in the autoxidation reaction of the HbO2 tetramer, because kf³ ks at any physiological pH

D I S C U S S I O N

In hemoglobin research, the central problem is understand-ing the cooperative bindunderstand-ing of molecular oxygen to the a2b2

tetramer For human HbA, the a and b chains contain 141 and 146 amino-acid residues, respectively, and a represen-tative set of the successive oxygen-binding constants is given

in terms of mmáHg)1as follows: K1ˆ 0.0188, K2ˆ 0.0566,

0

0.2

0.4

0.6

0.8

1.0

1.2

]t

]0

Time (h)

in Valency hybrid

pH 6.2

2

2

4

2

Reconstructed

kf

ks

Fig 6 First-order plots to show di€erent autoxidation rates of the

b chain between three di€erent hemoglobin derivatives in 0.1 M maleate bu€er at pH 6.2 and 35 °C Each curve (±±) was obtained by a least-squares ®tting to the experimental points, based on Eqn (2) The oxidation of separated b chains could be described by a single rate constant of k obs ˆ 0.10 h )1 in the presence of 20% (v/v) glycerol This inherent rate was dramatically suppressed not only in reconstructed HbO 2 but also in valency hybrid (a 3+ ) 2 (bO 2 ) 2 as well Heme concen-tration: 25 l M for separated b chains, 300 l M for reconstructed HbO 2 , and 50 l M for valency hybrid Hb.

K3ˆ 0.407 and K4ˆ 4.28 in 0.1M Bis/Tris buffer

con-taining 0.1MKCl at pH 7.4 and 25 °C [20] In this reaction,

major differences have been de®ned between

deoxyhemo-globin and oxyhemodeoxyhemo-globin by comparing their X-ray crystal

structures These include a movement of the iron atom into

the heme plane with a simultaneous change in the

orienta-tion of the proximal (F8) histidine, a rotaorienta-tion of the a1b1

dimer relative to the other a2b2 dimer about an axis P by

12±15 degrees, and a translation of one dimer relative to the

other along the P axis by  1 AÊ The latter two changes are

accompanied with sequential breaking of the so-called salt

bridges by C-terminal residues [21±25] Therefore, the two

types of the ab contact are de®ned in the molecule One is

the a1b1 (or a2b2) contact involving B, G, and H helices

and the GH corner, and other is the a1b2 (or a2b1) contact

involving mainly helices C and G and the FG corner [19,24]

When HbA goes from the deoxy to the oxy con®guration,

the a1b2 and a2b1 contacts undergo the principal changes

associated with the cooperative oxygen binding, so that

these are named the sliding contacts At the a1b1 and a2b2 interfaces, on the other hand, negligible changes are found insofar as the crystal structure was examined Consequently, these are called simply the packing contacts, and their role in hemoglobin function was not clear for a very long period of time To these packing contacts, we have recently assigned a key role in stabilizing the HbO2tetramer, as the formation

of the a1b1 or a2b2 contact greatly suppresses the autoxidation rate, particularly of the b chains [8]

In the autoxidation reaction of HbO2, as well as MbO2,

pH can affect the rate in many different ways Recent kinetic and thermodynamic studies of the stability of mammalian oxymyoglobins have shown that the autoxidation reaction

is not a simple, dissociative loss of O2 from MbO2but is due to a nucleophilic displacement of O2 from MbO2by a water molecule or a hydroxyl ion that can enter the heme pocket from the surrounding solvent The iron is thus converted to the ferric met-form, and the water molecule or the hydroxyl ion remains bound to the Fe(III) at the sixth

Table 2 Comparison of the autoxidation rate constants between the whole, separated, reconstructed, and hybrid hemoglobins in 0.1 M bu€er at pH 6.2 and 35 °C.

k obs (h )1 )

Concentration (l M as heme)

Whole HbO 2 0.82 (‹ 0.03) ´ 10 )1 0.13 (‹ 0.01) ´ 10 )1 300

Separated chains (aO 2 ) 1 0.89 (‹ 0.03) ´ 10 )1 ± 10

Reconstructed (aO 2 ) 2 (bO 2 ) 2 0.85 (‹ 0.06) ´ 10 )1 0.14 (‹ 0.04) ´ 10 )1 300

Hybrid (a 3+ ) 2 (bO 2 ) 2 ± 0.19 (‹ 0.02) ´ 10 )1 50

(aO 2 ) 2 (b 3+ ) 2 0.77 (‹ 0.03) ´ 10 )1 ± 50

Table 4 Comparison of the autoxidation rate constants between the whole, separated, reconstructed, and hybrid hemoglobins in 0.1 M bu€er at pH 9.0 and 35 °C.

k obs (h )1 )

Concentration (l M as heme)

Whole HbO 2 0.48 ´ 10 )2 0.48 ´ 10 )2 300

Separated chains (aO 2 ) 1 0.78 ´ 10 )1 ± 10

Reconstructed (aO 2 ) 2 (bO 2 ) 2 0.67 ´ 10 )2 0.67 ´ 10 )2 50

Hybrid (a 3+ ) 2 (bO 2 ) 2 ± 0.62 ´ 10 )2 50

(aO 2 ) 2 (b 3+ ) 2 0.61 ´ 10 )2 ± 50

Table 3 Comparison of the autoxidation rate constants between the whole, separated, reconstructed, and hybrid hemoglobins in 0.1 M bu€er at pH 7.5 and 35 °C.

k obs (h )1 )

Concentration (l M as heme)

Whole HbO 2 0.16 ´ 10 )1 0.50 ´ 10 )2 300

Separated chains (aO 2 ) 1 0.35 ´ 10 )1 ± 10

Reconstructed (aO 2 ) 2 (bO 2 ) 2 0.23 ´ 10 )1 0.50 ´ 10 )2 50

Hybrid (a 3+ ) 2 (bO 2 ) 2 ± 0.63 ´ 10 )2 50

(aO 2 ) 2 (b 3+ ) 2 0.63 ´ 10 )2 ± 50

coordinate position so as to form aqua- or

hydroxide-metMb Even the complicated pH-dependence for the

autoxidation rate can thereby be explained primarily in

terms of the following three types of displacement process

[6,7,26±28]:

Mb…II†…O2† ‡ H2O ƒƒƒƒ!k0

Mb…III†…OH2† ‡ Oÿ

2 …3†

Mb…II†…O2† ‡ H2O‡H‡ƒƒƒƒ!kH Mb…III†…OH2† ‡ HO2

…4†

Mb…II†…O2† ‡ OHÿ ƒƒƒƒ!kOH

Mb…III†…OHÿ† ‡ Oÿ

2 …5†

In these equations, k0is the rate constant for the basal

displacement by H2O, kHis the rate constant for the

proton-catalyzed displacement by H2O, and kOHis the rate constant

for the displacement by OH± The extent of contribution of

these elementary processes to the observed or overall

autoxidation rate, kobs in Eqn (1), can vary with the

concentrations of H+ or OH± ion Consequently, the

autoxidation rate exhibits a very strong parabolic

depen-dence on pH The reductive displacement of the bound

dioxygen as O2 by H2O can proceed without any

proto-nation, but it has been clearly shown that the rate is greatly

accelerated with the proton assistance by a factor of more

than 106mol)1, as formulated by Eqn (4) In this proton

catalysis, the distal histidine, which forms a hydrogen bond

to the bound dioxygen [29], appears to facilitate the effective

movement of a catalytic proton from the solvent to the

bound, polarized dioxygen via its imidazole ring and by a

proton-relay mechanism [6,7]

In our previous paper [8], such a nucleophilic

displace-ment mechanism was successfully applied to detailed

pH-dependence studies of the kf and ks values, both for

the HbO2tetramer and its separated chains, at more than

70 different values of pH from 5 to 11 in 0.1Mbuffer at

35 °C When the a and b chains were separated from the

HbO2 tetramer, each individual chain was oxidized much

more rapidly than in the parent HbO2, exhibiting a

proton-catalyzed displacement process performed by its own distal

histidine residue with pKaˆ 6.1 At the same time, the

oxidation rates of both chains were essentially the same

over the wide range of pH 5±11, so that their

pH-dependences could be formulated in terms of an

Ôacid-catalyzed two-state modelÕ However, this is not the case

with the HbO2 tetramer The value of kfincreased very

rapidly with increasing hydrogen ion concentration,

in-volving a proton-catalysis by the distal (a58) histidine with

pKaˆ 6.2, as with the separated chains The value of ks

also increased with increasing hydrogen ion concentration

but much less so than for kf Rather, the ksvalue showed a

rate saturation behavior with pKaˆ 5.1 on the acidic side

This pH-pro®le was therefore explained as a single

dissociation process for the distal histidine at position

b63, and described in terms of a Ôtwo-state modelÕ without

any proton catalysis Such a unique stability of the HbO2

tetramer was found to remain even in the low

concentra-tions of hemoglobin corresponding to appreciable

dissoci-ation into a1b1 or a2b2 dimers [5]

We have recently proposed that the distal histidine

residue can play a dual role in the nucleophilic

displace-ment of O2 from MbO2 or HbO2 [30] One is in a proton-relay mechanism via its imidazole ring, as random and undirected access of a proton to the bound dioxygen cannot yield such an enzyme-like, catalytic effect on the autoxidation rate of MbO2 or HbO2 Insofar as we have examined for more than a dozen of myoglobins, such a proton-catalyzed process could never be observed in the autoxidation of myoglobins lacking the usual distal histidine residue, no matter what the protein is, the naturally occurring or the distal His mutant as well [30] The other role is in the maximum protection of the FeO2

center against a water molecule or a hydroxyl ion that can enter the heme pocket from the surrounding solvent The latter case may be in the b chains of the HbO2 tetramer

To investigate more exactly the effect of the a1b1 or a2b2 contact on the stability of human HbO2, we have used this time the valency hybrid tetramers As a result, the b chain was found to acquire a noticeable resistance against the acidic autoxidation in a manner of contacting with the a chain, no matter which valency state the latter partner is in, the ferrous or the ferric These new ®ndings have led us to conclude that the packing contact produces a conforma-tional constraint in the b chain whereby the distal (E7) histidine at position 63 is tilted slightly away from the bound dioxygen, so as to prevent the acid-catalyzed displacement

of O2 from the FeO2center by an entering water molecule Thus, the remarkable stability of the HbO2tetramer can be ascribed mainly to the delayed autoxidation of the b chains

in acidic pH range More speci®cally, the b chain has acquired this stability by blocking out the proton catalysis performed by the distal histidine residue (Eqn 4)

Similarly, Shaanan [31] reported the stereochemistry of the iron-dioxygen bond in human HbO2by single-crystal X-ray analysis In the a chain, the distance between Neof His (E7) and the terminal oxygen atom (O-2) is found to be 2.7 AÊ, and the geometry favors a similar hydrogen bond as

in oxymyoglobin [29] In the b chain, however, Neof His (E7) is located further away from both O-2 and O-1 (3.4 and 3.2 AÊ, respectively), indicating that the hydrogen bond, even

if formed, must be very weak Recently, Lukin et al [32] claimed that a hydrogen bond is formed between O2and the distal histidine in both a and b chains of human HbO2, as revealed by heteronuclear NMR spectra of the chain-selectively labeled samples In 0.1M phosphate buffer at

pH 8.0 and 29 °C, the (He2, Ne2) cross-peaks of the distal histidyl residues were clearly observed as doublets in the (1H, 15N) spectrum of HbO2, at 1H chemical shifts of 4.79 p.p.m for b63His and 5.42 p.p.m for a58His These were taken as an indication that the He2proton is stabilized against solvent water exchange by a hydrogen bond between the distal His and the O2ligand in both a and b chains At the same time, they reported that much wider separation of 1.17 p.p.m appears on the He1resonances of the two distal histidine residues, showing that b63His is different from a58His in either the orientation or distance or both, with respect to the heme-bound dioxygen Such marked differ-ences between the two distal heme pockets may also be responsible for our kinetic results of the a and b chains in the HbO2 tetramer In this context, NMR spectra of the separated bO2chain must be most informative if available, because the autoxidation reaction of the b chain contains a very strong proton-catalysis in the isolated form but not in the HbO2tetramer

As for the dimer, as well as the tetramer, effect on the

oxidation rate, our explanations are as follows At basic pH,

both isolated a and b chains are quite susceptible to

autoxidation Each heme pocket seems to be suf®ciently

open to allow easier attack of the solvent hydroxyl ion on

the FeO2 center As a result, there occurs a very rapid

formation of hydroxide-metHb, the rate being dependent

directly on the concentrations of OH±ion In a1b1 dimers,

conformational constraints would greatly suppress

accessi-bility of the displacing nucleophile to each heme pocket

However, OH± ion is one of the strongest nucleophiles

in vivo, so that practically no rate difference was observed

between the a and b chains, resulting in the monophasic

autoxidation rate over the basic pH range To the acidic

autoxidation, essentially the same explanation is valid At

acidic pH, the displacing nucleophile is an entering water

molecule, but its concentration is always taken as 55.5Min

aqueous solution Therefore, participation of the catalytic

proton should be of primary importance to give a strong pH

dependence on the autoxidation rate As the HbO2sample is

diluted, the heme pocket of the a chain becomes freed from

conformational constraints that would decrease accessibility

of a water molecule and a catalytic proton as well As a

consequence, the rate of displacing O2 from the FeO2

approaches to that of the isolated a chain In contrast to

this, the heme pocket of the b chain still obstructs easy

access of a water molecule as well as a proton, so that the

b chains can keep a constant resistance against the acidic

autoxidation, even if the HbO2tetramer is diluted into ab

dimers Indeed, this is the most characteristic feature of

hemoglobin autoxidation

In relevance to a clinical aspect, it should be noted that a

quite large number of unstable hemoglobins have been

reported so far [24,33] Many of the mutants which occur at

the a1b2 interface have altered oxygen af®nity, but bulk of

evidence suggests that the a1b1 interface is much more

important in maintaining normal hemoglobin stability than

is the a1b2 interface As a matter of fact, hemolytic anemia

is known to result from substitutions affecting the a1b1

interface or the heme pocket If such mutations occur, the

heme iron will be more easily oxidized, and a sequence of

events leads to the globin precipitation or Heinz body

formation in red blood cells that causes hemolytic anemia

Typical examples of such variants are: E [b26(B8)Glu ®

Lys], Volga [b27(B9)Ala ® Asp], Genova

[b28(B10)-Leu ® Pro], St Louis [b28(B10)[b28(B10)-Leu ® Gln], Tacoma

[b30(B12)Arg ® Ser], Abraham Lincoln [b32(B14)Leu ®

Pro], Castilla [b32(B14)Leu ® Arg], Philly [b35(C1)Tyr ®

Phe], Rush [b101(G3)Glu ® Gln], Peterborough

[b111(G13)Val ® Phe], Madrid [b115(G17)Ala ® Pro],

Khartoum [b124(H2)Pro ® Arg], J Guantanamo

[b128-(H6)Ala ® Asp], Wien [b130(H8)Tyr ® Asp], Leslie

[b131(H9)Gln ® deleted], Torino [a43(CD1)Phe ® Val],

L.Ferrara [a47(CD5)Asp ® Gly], Setif [a94(G1)Asp ®

Tyr], St Lukes [a95(G2)Pro ® Arg] Surprisingly, almost

all of these pathological mutations are found on the b chain,

especially in the a1b1 contact regions It follows from

our present study that in these variant hemoglobins the

a1b1 contact becomes loose or disruptive, and that the

autoxidation of the b chain must have been accelerated, just

like the separated one, with irreversible hemichrome

formation

In conclusion, human hemoglobin seems to differentiate the two types of the ab contact quite properly for its functional properties The a1b2 or a2b1 contact is associ-ated with the cooperative oxygen binding, whereas the a1b1

or a2b2 contact is used for controlling the stability of the bound O2 We can thus form, for the ®rst time, a uni®ed picture of hemoglobin function by closely integrating the reversible and the stable binding of molecular oxygen by iron(II) in protic, aqueous solvent

A C K N O W L E D G E M E N T

This work was partly supported by grants-in-aid for Scienti®c Research (07640896 & 10440248) from the Ministry of Education, Culture and Science of Japan.

R E F E R E N C E S

1 Wever, R., Oudega, B & Van Gelder, B.F (1973) Generation of superoxide radicals during the autoxidation of mammalian oxyhemoglobin Biochim Biophys Acta 302, 475±478.

2 Gotoh, T & Shikama, K (1976) Generation of the superoxide radical during the autoxidation of oxymyoglobin J Biochem (Tokyo) 80, 397±399.

3 Mansouri, A & Winterhalter, K.H (1973) Nonequivalence of chains in hemoglobin oxidation Biochemistry 12, 4946±4949.

4 Zhang, L., Levy, A & Rifkind, J.M (1991) Autoxidation of hemoglobin enhanced by dissociation into dimers J Biol Chem.

266, 24698±24701.

5 Tsuruga, M & Shikama, K (1997) Biphasic nature in the autoxidation reaction of human oxyhemoglobin Biochim Biophys Acta 1337, 96±104.

6 Shikama, K (1988) Stability properties of dioxygen-iron(II) porphyrins: an overview from simple complexes to myoglobin Coordination Chem Rev 83, 73±91.

7 Shikama, K (1998) The molecular mechanism of autoxidation for myoglobin and hemoglobin: a venerable puzzle Chem Rev 98, 1357±1373.

8 Tsuruga, M., Matsuoka, A., Hachimori, A., Sugawara, Y & Shikama, K (1998) The molecular mechanism of autoxidation for human oxyhemoglobin: tilting of the distal histidine causes nonequivalent oxidation in the b chain J Biol Chem 273, 8607±8615.

9 Williams, R.C Jr & Tsay, K (1973) A convenient chromato-graphic method for the preparation of human hemoglobin Anal Biochem 54, 137±145.

10 De Duve, C (1948) A spectrophotometric method for the simul-taneous determination of myoglobin and hemoglobin in extracts

of human muscle Acta Chem Scand 2, 264±289.

11 Geraci, G., Parkhurst, L.J & Gibson, Q.H (1969) Preparation and properties of a- and b-chains from human hemoglobin J Biol Chem 244, 4664±4667.

12 Turci, S.M & McDonald, M.J (1985) Isolation of normal and variant human hemoglobin subunits J Chromatogr 343, 168± 174.

13 Boyer, P.D (1954) Spectrophotometric study of the reaction of protein sulfhydryl groups with organic mercurials J Am Chem Soc 76, 4331±4337.

14 Rosemeyer, M.A & Huehns, E.R (1967) On the mechanism of the dissociation of haemoglobin J Mol Biol 25, 253±273.

15 Edelstein, S.J., Rehmar, M.J., Olson, J.S & Gibson, Q.H (1970) Functional aspects of the subunit association-dissociation equi-libria of hemoglobin J Biol Chem 245, 4372±4381.

16 McDonald, M.J., Turci, S.M., Mrabet, N.T., Himelstein, B.P & Bunn, H.F (1987) The kinetics of assembly of normal and variant human oxyhemoglobins J Biol Chem 262, 5951±5956.

17 Rifkind, J.M., Abugo, O., Levy, A & Heim, J (1994) Detection,

formation and relevance of hemichromes and hemochromes.

Methods Enzymol 231, 449±480.

18 Levy, A., Kuppusamy, P & Rifkind, J.M (1990) Multiple heme

pocket subconformations of methemoglobin associated with distal

histidine interactions Biochemistry 29, 9311±9316.

19 Borgstahl, G.E.O., Rogers, P.H & Arnone, A (1994) The 1.8 AÊ

structure of carbonmonoxy-b 4 hemoglobin J Mol Biol 236,

817±830.

20 Imai, K (1994) Adair ®tting to oxygen equilibrium curves of

hemoglobin Methods Enzymol 232, 559±576.

21 Perutz, M (1990) Mechanisms of Cooperativity and Allosteric

Regulation in Proteins Cambridge University Press, Cambridge,

UK.

22 Fermi, G & Perutz, M.F (1981) Haemoglobin and myoglobin.

In Atlas of Molecular Structure in Biology, Vol 2 (Phillips, D.C.

& Richards, F.M., eds), Clarendon Press, Oxford, UK.

23 Perutz, M.F., Wilkinson, A.J., Paoli, M & Dodson, G.G (1998)

The stereochemical mechanism of the cooperative e€ects in

hemoglobin revisited Annu Rev Biophys Biomol Struct 27,

1±34.

24 Dickerson, R.E & Geis, I (1983) Hemoglobin: Structure,

Func-tion, Evolution and Pathology The Benjamin/Cummings

Publish-ing Co, Inc Menlo Park, CA, USA.

25 Baldwin, J & Chothia, C (1979) Haemoglobin: the structural

changes related to ligand binding and its allosteric mechanism.

J Mol Biol 129, 175±220.

26 Satoh, Y & Shikama, K (1981) Autoxidation of oxymyoglobin:

a nucleophilic displacement mechanism J Biol Chem 256, 10272±10275.

27 Shikama, K (1984) A controversy on the mechanism of autoxi-dation of oxymyoglobin and oxyhaemoglobin: oxiautoxi-dation, disso-ciation, or displacement? Biochem J 223, 279±280.

28 Shikama, K (1990) Autoxidation of oxymyoglobin: a meeting point of the stabilization and the activation of molecular oxygen Biol Rev (Cambridge) 65, 517±527.

29 Phillips, S.E.V & Schoenborn, B.P (1981) Neutron di€raction reveals oxygen-histidine hydrogen bond in oxymyoglobin Nature (London) 292, 81±82.

30 Suzuki, T., Watanabe, Y.-H., Nagasawa, M., Matsuoka, A & Shikama, K (2000) Dual nature of the distal histidine residue in the autoxidation reaction of myoglobin and hemoglobin: com-parison of the H64 mutants Eur J Biochem 267, 6166±6174.

31 Shaanan, B (1982) The iron-oxygen bond in human oxyhaemo-globin Nature (London) 296, 683±684.

32 Lukin, J.A., Simplaceanu, V., Zou, M., Ho, N.T & Ho, C (2000) NMR reveals hydrogen bonds between oxygen and distal histi-dines in oxyhemoglobin Proc Natl Acad Sci USA 97, 10354±10358.

33 Winslow, R.M & Anderson, W.F (1978) The hemoglobinopa-thies In The Metabolic Basis of Inherited Disease (Stanbury, J.B., Wyngaarden, J.B & Fredrickson, D.S., eds), 4th edn Part 10, Chapter 62, pp 1465±1507 McGraw-Hill, Inc., New York, USA.