Role of β112 Cys (G14) in Homo- (β4) and Hetero- (α2β2) Tetramer Hemoglobin Formation*

In order to assess the role of β112 Cys in homo- and hetero-tetrameric hemoglobin formation, we expressed four β112 variants (β112Cys→Asp, β112Cys→Ser, β112Cys→Thr, and β112Cys→Val) and studied assembly with α chainsin vitro. β112 Cys is normally present at β1β2 and α1β1interaction sites in homo- (β4) and hetero-tetramers (α2β2). β4 formation in vitro was influenced by the amino acid at β112. β112 Asp completely inhibited formation of homo-tetramers, whereas β112 Ser showed only slight inhibition. In contrast, β112 Thr or Val enhanced homo-tetramer formation compared with βA chains. Association constants for homo-tetramer formation increased in the order of β112Cys→Ser, βA, β112Cys→Thr, and β112Cys→Val, whereas the value for β112Cys→Asp was zero under the same conditions. These β112 changes also affected in vitroα2β2 hetero-tetramer formation. Order of α2β2 formation under limiting α-globin chain conditions showed Hb βC112S > Hb A > Hb S = Hb βC112T = Hb βC112V >>> Hb βC112D. Hb β112D can form tetrameric hemoglobin, but this β112 change promotes dissociation into α and β chains instead of αβ dimer formation upon dilution. These results indicate that amino acids at α1β1 interaction sites such as β112 on the G helix play a key role in stable αβ dimer formation. Our findings suggest, in addition to electrostatic interaction between α and β chains, that dissociation of β4 homo-tetramers to monomers and hydrophobic interactions of the β112 amino acid with α chains governs stable α1β1 interactions, which then results in formation of functional hemoglobin tetramers. Information gained from these studies should increase our understanding of the mechanism of assembly of multi-subunit proteins.

Equimolar amounts of ␣and ␤-globin chains of human hemoglobin self assemble to form ␣ 2 ␤ 2 tetramer. In addition, isolated ␤ chains also assemble to form ␤ 4 homo-tetramers (1). Extensive previous studies using naturally occurring variants and our recent studies using recombinant ␤ chain variants showed that affinity between ␣ and ␤ chains is promoted by negatively charged ␤ chains and is independent of charge location on the surface except at the ␣ 1 ␤ 1 interaction site (2)(3)(4)(5). Our previous studies showed that affinity is promoted by negatively charged ␤ chains up to a maximum of two additional net negative charges (5). In addition, we showed that ␤112 Cys located at an ␣ 1 ␤ 1 interaction site on the G helix is critical for facilitating formation of stable ␣␤ dimers which then form functional hemoglobin tetramers. We also demonstrated that ␤ 112Cys3 Asp inhibits formation of stable ␣ 1 ␤ 1 and ␤ 1 ␤ 2 interactions in ␣ 2 ␤ 2 and ␤ 4 tetramers, respectively (5).
The ␣ chains are in monomer-dimer equilibrium and dissociation into monomers is favored, whereas the ␤ chains are in monomer-tetramer equilibrium and association into tetramers is favored. It is generally assumed that dissociation of these oligomeric subunits (Reactions I and II) into monomers must occur before they can combine to form the ␣␤ dimers (Reaction III). Two ␣␤ dimers then associate to form tetrameric hemoglobin (Reaction IV). The dissociation of oligomeric ␤ subunits is a first-order reaction, while assembly of ␣␤ dimers from ␣ and ␤ monomers (Reaction III) is a second-order reaction (2,4,7). Formation of functional hemoglobin tetramers is dependent on these reactions (7). In vitro assembly of the liganded forms of ␣and ␤-globin chains show that the rate of dissociation of ␤ 4 homo-tetramers is a rate-limiting step in formation of ␣␤ hetero-tetramers (2). Therefore, clarification of the role of ␤112 Cys at the ␣ 1 ␤ 1 and ␤ 1 ␤ 2 interaction site on ␣␤ and ␤ 4 formation, respectively, is critical in order to understand the mechanism of hemoglobin assembly. In this report, we expressed four ␤112 variants and characterized effects of these changes on ␣ 2 ␤ 2 and ␤ 4 tetramer formation in vitro in order to further assess the role of ␤112 Cys on assembly of hemoglobin.

MATERIALS AND METHODS
Expression of Soluble Recombinant Human ␤-Globin Chain Variants in Escherichia coli-Four ␤112 globin chain variants (e.g. ␤112 Ser, ␤112 Val, ␤112 Thr, and ␤112 Asp) were constructed and expressed using the pHE2␤ plasmid vector that contains cDNAs coding for each ␤ chain variant and methionine aminopeptidase which was originally developed to express ␣ and ␤ chains at the same time (8,9). The basic strategy for generation of these variants by site-specific mutagenesis of the normal ␤ chain involves recombination/polymerase chain reaction as described previously (10). Clones were subjected to DNA sequence analysis of the entire ␤-globin cDNA region using site-specific primers and fluorescently tagged terminators in a cycle sequencing reaction in which extension products were analyzed on an automated DNA sequencer. Plasmids were transfected into E. coli (JM 109) (Promega Co.), bacteria were grown at 30°C, and soluble ␤-globin chain variants were isolated as described (5,8). Expression and purification of ␤-globin variants were basically as described previously (5,8). However, cationexchange chromatography on a Source 15 S column (Amersham Pharmacia Biotech) instead of Superose 12 gel filtration was used for purification of the ␤ chain variants. Authentic human hemoglobin, ␣-, ␤ A (␤6 Glu)-and ␤ S (␤6 Val)-globin chains were purified from erythrocyte lysates from normal controls and patients with sickle cell disease, respectively, according to previously described methods (11). Removal of p-mercuribenzoate from ␤ chains was accomplished using 20 mM dithiothreitol (DTT), 1 and globin chains were isolated using gel filtration on a Superose 12 column for the final purification step.
Biochemical Characterization of Purified ␤-Globin Chains-Molecular mass and sample purity were assessed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) as described (12). Mutations in each purified ␤-globin chain variant were confirmed using mass spectral analysis. Electrospray ionization mass spectrometry was performed on a VG BioQ triple quadrapole mass spectrometer (Micromass, Altrincham, UK) using the multiple charged ion peaks from the ␣-globin chain (M r ϭ 15,126.4) as reference for mass scale calibrations (13). Data analysis employed the MassLynx ® software package (Micromass).
Purified ␤-globin chains were also analyzed by cellulose acetate electrophoresis, and mobilities were compared with those of authentic human globin chains. Isoelectric focusing of purified ␤-globin variants, Hb A and Hb S, was performed on Ampholine PAG plates (Amersham Pharmacia Biotech), pH 5.5-8.5, using a Multiphor II system (Amersham Pharmacia Biotech). After focusing for 2 h at a constant 25 watts at 4°C, the gel plate was cut in half, one portion was stained using the JB-2 staining system (Isolab inc., Akron, Ohio) to detect heme proteins, and the other half was stained with Coomassie Brilliant Blue R-250 (Sigma) to detect protein. Isoelectric point of each ␤-globin variant was estimated using a calibration curve prepared with isoelectric focusing standards (Bio-Rad).
Absorption spectra of purified ␤-globins in the CO form were recorded using a Hitachi U-2000 spectrophotometer (Hitachi Instruments, Inc. Danbury, CT). Globin concentration was determined spectrophotometrically using a millimolar extinction coefficient of 13.4 at 540 nm for carbonmonoxy hemoglobin (14). Circular dichroism (CD) spectra of ␤-globin variants were recorded using an Aviv model 62 DS instrument (Varian Analytical Instruments, San Fernando, CA) employing a 0.1-cm light path cuvette at 10 M globin concentration. CD ellipticity of ␤-globin variants compared with normal ␤ A was monitored between 190 and 260 nm. Oxygen dissociation curves of hemoglobin tetramers were determined in 50 mM Bis-Tris buffer containing 0.1 M NaCl and 5 mM EDTA, pH 7.2, at 20°C using a Hemox Analyzer (TCS Medical Products, Huntingdon Valley, PA) (15).
Dissociation of ␤-globin homo-tetramers was studied by fast protein liquid chromatography (FPLC) on a Superose 12 gel filtration column. ␤-Globin was mixed with blue dextran and vitamin B 12 , internal markers for determination of void volume (V 0 ) and total column volume (V t ), respectively. The final ␤-globin solution (1.5 to 75 M) was injected into a Superose 12 column, and gel filtration was done using 0.1 M potassium phosphate buffer, pH 7.0 at 4°C. Elution coefficients were calculated using the following equation, where V e represents elution volume of each sample.
␣ 2 ␤ 2 tetramer formation was assessed by mixing purified ␤ chain variants (75 M) with varying amounts of ␣-globin chain in the CO form in 0.1 M potassium phosphate buffer, pH 7.0 at 25°C (2,5), and tetramer formation was monitored by FPLC using a Mono-S column. Assembled tetramers were characterized after separation from excess, free globin chains by FPLC using a Source 15 S column. Dissociation of in vitro assembled ␣ 2 ␤ 2 tetramer into subunits was assessed at a flow rate of 0.5 ml/min at 4°C by large zone chromatography using a Superose 12 gel filtration column (1 ϫ 31 cm) as described previously (16,17). Sample volumes of ␣ 2 ␤ 2 variant tetramers (25 ml in 0.1 M potassium phosphate buffer, pH 7.0) were equal to the column bed volumes. Elution was monitored at 405 nm using a 5-mm flow cell. Elution volume (V e ) was measured at the centroid of the ascending leading boundary (16,17). Dissociation constants of ␣ 2 ␤ 2 tetramers to ␣␤ dimers were estimated by evaluation of the change in V e as a function of hemoglobin concentration (17,18).

Expression and Purification of Soluble ␤-Globin
Chain Variants-We previously reported expression of ␤ 112Cys3 Asp chains in bacteria (5). In order to further define the role of ␤112 Cys in ␣ 2 ␤ 2 and ␤ 4 tetramer formation, three additional ␤-globin chain variants (e.g. ␤ 112Cys3 Ser , ␤ 112Cys3 Thr , and ␤ 112Cys3 Val ) were expressed and characterized. Ser is similar in size to Cys, whereas Thr and Val are slightly larger. In addition, Ser and Thr are hydrophilic like Cys, whereas Val is highly hydrophobic. After DNA sequence confirmation, the three ␤-globin chain variants were expressed in bacteria and purified by a combination of two anion-exchange chromatography steps using DEAE-cellulose and Mono-Q columns (9). Further purification was achieved using cation-exchange chromatography on a Source 15 S column. Purified ␤ A chains migrate predominantly as 32-kDa dimers (9), whereas the four ␤112 variants migrate mainly as 16-kDa monomers with small traces of dimers on SDS-PAGE (Fig. 1). Addition of 20 mM DTT to the chains in solution prior to SDS-PAGE converts the dimers to monomers, indicating that small amounts of ␤93 Cys oxidize and form disulfide-linked dimers.
Mass spectral analysis of the four ␤ chain variants using electrospray ionization mass spectrometry resulted in values of 15,878.9, 15,850.3, 15,865.8, and 15,863.5 for ␤ 112Asp , ␤ 112Ser , ␤ 112Thr , and ␤ 112Val , respectively, which were in agreement with expected masses for the variant ␤-globin chains. Carbonmonoxy forms of all four variants showed typical absorption spectral characteristics of human hemoglobin chains with peaks at 568, 540, 419, 344, and 276 nm (14), indicating correct heme insertion into the variant ␤-globin chains. CD spectra were also measured to examine structures of each ␤-globin variant compared with ␤ A -globin chains (Fig. 2). Ellipticities between 190 and 260 nm for the ␤-globin variants were almost identical to that of normal human and recombinant ␤-globin, but there are small differences in the region of the peak around 213 nm that are likely to be characteristic of these different ␤ chains. Interestingly, the region of the peak of the spectrum for ␤ 112Val chains around 213 nm was slightly left shifted compared with those of ␤ A , ␤ 112Asp , and ␤ 112Ser chains (Fig. 2). These CD results indicate that ␤-globin chain variants made in the E. coli were properly folded and had secondary structures similar to authentic ␤-globin. In addition, the differences in the spectrum for the ␤ 112Val chains may be a reflection of increased conformational changes.
Characterization of Purified ␤-Globin Chain Variants-We previously reported that electrophoretic mobility on cellulose acetate of ␤ 112Cys3 Asp chains was identical to ␤ s chains (5), indicating that these chains migrate as monomers (charge of -2) rather than tetramers (charge of -4) like ␤ 4 A chains (Fig. 3). In contrast, mobility of ␤ 112Cys3 Val and ␤ 112Cys3 Thr chains was identical to authentic human ␤ A . These results indicate that conversion from Cys to Val or Thr at ␤112 does not cause significant changes in surface charge while conversion to Asp appears to increase net positive charge of the ␤-globin molecule because formation of monomers is favored rather than tetramers. Effect of ␤112 amino acid on ␤-globin charge was studied further by isoelectric focusing on polyacrylamide gel plates. ␤ 112Cys3 Ser and ␤ 112Cys3 Asp chains focused as sharp bands with pI values of 6.68 and 6.35, respectively, whereas ␤ 112Cys3 Val and ␤ 112Cys3 Thr chains had pI values of 6.30 and 6.68, respectively, but focused as diffuse bands possibly because they exist as mixtures of ␤ chain monomer, dimer, and tetramer during electrophoresis (19).
␤ 4 tetramer and ␤-monomer levels for each of the ␤ chain variants were assessed by size-exclusion chromatography (17,20,21). ␤ A -Globin chains in solution exist as homo-tetramers (␤ 4 ) rather than monomers in the absence of ␣-globin chains (1). In contrast, we found previously that ␤ 112Cys3 Asp chains exist as monomers rather than ␤ 4 tetramers (5). In order to further characterize effects of the different ␤112 amino acids on tetramer formation, each ␤-globin variant was chromatographed on a Superose 12 gel-filtration column equilibrated with 0.1 M phosphate buffer, pH 7.0, at 4 O C in order to separate tetrameric and monomeric species. The gel-filtration pattern of ␤ A globin, which contains ␤112 Cys, depended on con-centration and temperature. At 75 M, ␤ A chains eluted mainly as tetramers (86%) with a minor shoulder of monomers (14%) (Fig. 4a). The patterns of the variants depended on the ␤112 amino acid. When the same concentration as ␤ A globin for each ␤ chain variant was applied to the column, results showed that the ␤ 112Cys3 Ser variant contained 51% tetramer and 49% monomer (Fig. 4c), whereas the ␤ 112Cys3 Thr variant contained a slightly higher percentage of tetramer than that of ␤ A (Fig. 4d). The ␤ 112Cys3 Val chains eluted only as tetramers when the concentration was 75 M. Dilution of samples of the other ␤ chain variants and ␤ A chains decreased the amounts of tetramer while monomer amounts increased. In contrast, ␤ 112Cys3 Asp chains eluted only as monomers at all hemoglobin concentrations used. The ratio of tetramer/monomer decreased as a function of ␤112 amino acid in the following order: Values for ␤ 112Cys3 Thr and ␤ 112Cys3 Val were 2and 600-fold, respectively, higher than that for ␤ A(112Cys) chains. Association constants for ␤ 4 homo-tetramer formation for the ␤112 variants was calculated using results from gelfiltration chromatography and are summarized in Table I. The ␤ A -globin chain value (6 ϫ 10 16 M Ϫ3 ) was similar to results reported previously (22). The association constant for the ␤ 112Cys3 Asp variant was 0 since this variant existed only as a monomer. It is interesting to note that dissociation of ␤ S chains to monomers with decreasing concentrations was significantly less compared with that of ␤ A . The tetramer association constant for ␤ S chain monomers was about 30-fold higher than that of ␤ A , indicating that the ␤6 Val mutation in ␤ S also affects tetramermonomer equilibrium and promotes homo-tetramer assembly.
Hetero-tetramer Formation (␣ 2 ␤ 2 ) in Vitro Using ␤-Globin Variants and Native ␣-Globin-␤-Globin readily forms heterotetramers (␣ 2 ␤ 2 ) in vitro in the presence of ␣-globin. Tetramers were formed in vitro, purified and separated from unreacted globins by chromatography, and then were analyzed by cellulose acetate electrophoresis (Fig. 3B). We previously reported that ␤112 Asp chains did not form ␤ 4 tetramers but did form ␣ 2 ␤ 2 112Asp tetramers. All three new recombinant ␤-globin variants formed ␣ 2 ␤ 2 tetramers in vitro. Cellulose acetate electrophoresis showed cathodic shifts in migration of tetramers compared with ␤ chain variants upon binding of ␣-globin chains. It is interesting to note that, although individual ␤-globin chain variants migrated differently, all ␣ 2 ␤ 2 tetramers containing the different ␤112 variants co-migrated to the same position as that of Hb A (Fig. 3B). We also performed functional studies of the variant tetramers in 50 mM BisTris buffer, pH 7.2, containing 0.1 M NaCl at 20°C in the presence and absence of 2,3-biphosphoglycerate and compared results with those of Hb A. Results of oxygenbinding properties for the ␤112 variant ␣ 2 ␤ 2 tetramers are summarized in Table II. Hb ␤C112S, Hb ␤C112T and Hb ␤C112V exhibited similar oxygen affinities and n values to those of Hb A tetramers. Oxygen affinity of recombinant Hb A was previously reported to be the same as that of normal human Hb A (5,8), while oxygen affinity of Hb ␤C112D (P 50 ϭ 2.2) was higher than that of Hb A with a similar cooperativity.
Tetramer Formation in Vitro with Limiting Amounts of ␣ Chains-Competition for ␣ 2 ␤ 2 hetero-tetramer formation between two different ␤-globin chains can be assessed by monitoring tetramer assembly in the presence of limiting amounts of ␣ chains using 1:1 mixtures of the two different ␤-globin chains (3,5). Previous studies in vitro showed that variant hemoglobin percentages were higher when using more negatively charged ␤ chains like J-Baltimore (␤ 16Gly3 Asp ) and N-Baltimore (␤ 95Lys3 Glu ) (4,5). These studies suggested that more negatively charged ␤ chains bind positively charged ␣ chains more readily than ␤ A chains (2)(3)(4)(5). Furthermore, Hb A tetramers formed twice as readily as Hb S, suggesting that ␤ A chains interact with ␣ chain more readily than ␤ S under limiting ␣-globin conditions and that this was due to the positive net charge of Val instead of Glu at the ␤6 position (3)(4)(5). In addition, our previous subunit competition studies with equimolar mixtures of ␤ S and ␤ 112Cys3 Asp chains in the presence of limiting ␣ chains suggested that ␤ 112Cys (G14) is a key amino acid in formation of stable ␣␤ dimers (5). The ␤112 position is known to be localized at the ␣␤ subunit interface after hetero-tetramers assemble (6).
In order to further clarify the effect of ␤112 amino acid side chain on assembly, tetramer formation in vitro was evaluated for each of the purified ␤-globin variants after addition of ␣-globin chain isolated from human red blood cells. Competition experiments were done in which varying amounts of ␣ chains were added to equimolar mixtures of ␤ S and either ␤ 112Cys3 Val , ␤ 112Cys3 Thr , ␤ 112Cys3 Ser , or ␤ A chains, and assembled tetramers were separated from free chains by ion-exchange chromatography (Fig. 5). Our previous competition ex- FIG. 4. Superose 12 gel-filtration chromatogram of purified ␤-globin chains. Purified ␤-globin variants were mixed with internal markers (blue dextran for identification of void volume (V 0 ) and vitamin B 12 for total bed volume (V t )) in 150 l (final globin concentration ϭ 75 M) and applied to a Superose 12 column (1 ϫ 30 cm). Gel filtration was accomplished at a flow rate of 0.5 ml/min using 100 mM potassium phosphate buffer, pH 7.0. a, ␤ A chains; b, ␤ S chains; c, ␤ 112Cys3 Ser chains; d, ␤ 112Cys3 Thr chains; e, ␤ 112Cys3 Asp chains; and f, ␤ 112Cys3 Val chains.

TABLE II Oxygen-binding properties of ␤112-variant hemoglobin tetramers
Oxygen equilibrium curves of hemoglobins were determined using 35 M Hb concentration in 50 mM BisTris/HCl buffer, pH 7.2, containing 100 mM NaCl and 5 mM EDTA at 20°C. P 50 is partial oxygen pressure required to give 50% oxygen saturation of hemoglobin. n max values were calculated from the Hill plot of oxygen-equilibrium curves. Concentration of 2,3-biphosphoglycerate (BPG) when present (ϩ) is 2 mM. Mean P 50 values (n ϭ 5) for Hb A with and without BPG are shown with a mean Ϯ S.E. of 0.1 and 0.3, respectively. K 4,2 values for Hb A, Hb ␤C112T, and Hb ␤C112V were calculated as shown in Fig. 6. However, the K 4,2 values for Hb ␤C112T and Hb ␤C112V were not calculated since these two tetramers dissociated directly to monomers instead of dimers upon dilution as shown in Fig. 6 periments in vitro using mixtures of purified ␣ and ␤ chains showed that ␣␤ A dimers form about twice as readily as ␣␤ s dimers when the concentration of ␣ chains becomes limiting (5). This results in assembly of less Hb S relative to Hb A when equimolar amounts of ␤ A and ␤ S chains compete for limiting amounts of ␣-globin (Fig. 5). Similar amounts of variant tetramers (Hb ␤C112T and Hb ␤C112V) and Hb S were formed in mixtures at all ␣-globin concentrations tested. These results indicate that levels of Hb ␤C112T and Hb ␤C112V would be less than that of Hb A in competition studies with the ␤ A chain. Total amounts of Hb ␤C112S were always more than that of Hb S, with the Hb ␤C112S/Hb S ratio approaching 3.5 when ␣ chain/total ␤ chain approaches 0.1. On the other hand, as shown previously, much less Hb ␤C112D formed in mixtures of ␤ 112Cys3 Asp and ␤ S , and almost all ␣-globin chains assembled with ␤ S chains to form Hb S, when the ratio of ␣/total ␤ chains was Ͻ 0.5 (5).
Effect of ␤112 Variants on Dimer (␣␤) and Tetramer (␣ 2 ␤ 2 ) Equilibrium-In order to evaluate effects of ␤112 amino acid on dimer-tetramer equilibrium, large-zone gel-filtration chromatography of ␤112-variant ␣ 2 ␤ 2 tetramers was performed using a Superose 12 HR column at various hemoglobin concentrations (17). Decreasing concentrations of ␣ 2 ␤ 2 variant tetramers in the CO form (25 ml in 0.1 M potassium phosphate buffer, pH 7.0), whose volumes were equal to the column bed volume, were loaded on the column in order to prevent dilution of samples during chromatography. Elution profiles of hemoglobin variants and Hb A exhibited sharp leading boundaries which then plateaued at different levels depending on the concentration applied. The elution volume (V e ), assessed by the centride of the leading boundary, depends on hemoglobin concentration below 1.3 M; the lower the concentration then the higher the V e . V e values were constant above 1.3 M hemoglobin, indicating most molecules exist as tetramers (Fig. 6). With decreasing concentrations of Hb A, the V e values approached 15.5 ml, which corresponds to the value for ␣␤ dimers. A plot of V e as a function of hemoglobin concentration showed similar curves for Hb ␤C112T and Hb A, whereas the curve for Hb ␤C112V was slightly shifted to the left. These findings suggest that ␤112 Cys3 Val inhibits tetramer dissociation. The tetramer-dimer dissociation constants for Hb A, Hb ␤C112T and Hb ␤C112V were calculated as described previously (23) and were 0.11, 0.12, and 0.07 M, respectively (Table I). In contrast, the plots of elution volume as a function of concentration for Hb ␤C112S and Hb ␤C112D were right shifted from that of Hb A (Fig. 6). Furthermore, V e values for Hb ␤C112S and Hb ␤C112D below 0.1 M were greater than that for ␣␤ dimers and approached 16.25 ml at lower concentrations, which corresponds to the V e value for monomeric forms determined using ␤ 112Cys3 Asp chains. The plot for Hb ␤C112D was more right shifted than that for Hb ␤C112S (Fig. 6). These results indicate that tetrameric forms of Hb ␤C112S and Hb ␤C112D continue to dissociate to ␣ and ␤ monomers upon dilution rather than remaining as ␣␤ dimers and that Hb ␤C112D dissociates more than Hb ␤C112S.

DISCUSSION
Effects of ␤112 Amino Acids on Homo-tetramer Assembly-X-ray analysis of ␤ 4 homo-tetramers at 1.8 Å resolution showed that ␤112 Cys (G14) is located at the ␤1 and ␤2 chain interface and that the side chains of ␤ 1 112 Cys and ␤ 2 112 Cys in the ␤ 4 tetramer are very close to the molecular dyad at this interface (6). These two residues exist on the surface of the ␤ chains and may be involved in weak interactions with other residues. Our present results showing absence or trace amounts of disulfide dimer formation for all four of the ␤112 chain variants in contrast to ␤ A and other ␤ chain variants, suggest that ␤ 1 112 Cys and ␤ 2 112 Cys are physically close together and that disulfide ␤ chain dimer formation is governed in part by these two cysteine residues. Our results also show that ␤112 Val chains exist mainly as ␤ 4 tetramers in solution, which is in contrast to the previous results with ␤ 1 112 Asp chains that exist as monomers (5). In addition, ␤112 Thr and ␤112 Ser chains can form ␤ 4 homo-tetramers, whose levels are intermediate between those of ␤112 Val and ␤112 Asp. Relative order of tetramer formation among the different ␤112 variants is a direct function of hydrophobicity of the ␤112 amino acid, the higher the hydrophobicity then the more tetramers are formed. These results suggest that ␤1 and ␤2 interactions in homotetramers are governed to a large extent by hydrophobic interactions involving the ␤112 amino acid and that hydrophobic interactions between ␤ 1 112 Val and ␤ 2 112 Val are strongest followed by ␤ 1 112 Thr-␤ 2 112 Thr, ␤ 1 112 Cys-␤ 2 112 Cys, and then ␤ 1 112 Ser-␤ 2 112 Ser. In contrast, ␤112 Asp chains exist as monomers since ␤ 1 112 Asp and ␤ 1 112 Asp cannot participate in hydrophobic interactions, which results in lack of formation of homo-tetramers. The ellipticity values for the CD spectra in the far ultraviolet region (190 -280 nm) are expected not to be dependent on the heme moiety but on overall protein conformation (24). The spectral difference in ␤112 Val compared with those of ␤ A -and other ␤ chain variants may be related to the high homo-tetramer formation of the ␤112 Val variant. Detailed studies of the conformational change differences between homo-tetramer and monomer forms are now in progress.
Effects of ␤112 Amino Acid on Hetero-tetramer Assembly- ␤112 Cys is located at the interface of ␣ 1 ␤ 1 in ␣ 2 ␤ 2 hemoglobin tetramers and interacts with Val ␣107 (G14) and Ala ␣110 (G11), which are critical for stabilization of the ␣␤ interface (6). Amino acid changes at ␤112 also affected in vitro assembly of ␣ 2 ␤ 2 variant tetramers. Order of ␣ 2 ␤ 2 formation under limiting ␣-globin chain conditions showed Hb ␤C112S Ͼ Hb A Ͼ Hb S ϭ Hb ␤C112T ϭ Hb ␤C112V Ͼ Ͼ Ͼ Hb ␤C112D. In addition, apohemoglobin dimers were reported previously to dissociate into monomers 170-fold faster than normal ␣␤ dimers because no interactions occur between ␤ 112Cys and ␣ 104 Cys as a result of loss of the heme moiety (25). These findings and our results indicate that relative affinity of ␣ for ␤ chains is highly dependent on direct ␣ 1 and ␤ 1 interaction sites even though surface charge of the chains also affects interactions during assembly (2)(3)(4)(5). Amino acids at ␣ 1 ␤ 1 interaction sites such as ␤112 Cys may be critical for ␣␤ assembly.
Although the precise steps in ␣ 2 ␤ 2 tetramer assembly in vivo are not completely understood, functionally intact ␣ and ␤ chains can be isolated and readily reconstituted into ␣ 2 ␤ 2 tetramers by mixing equimolar amounts of ␣ and ␤ chains. As noted previously, it is generally assumed that isolated ␣ and ␤ chains exist in a monomer-dimer and monomer-tetramer equilibrium at least in vitro, respectively (see Reactions I and II in the Introduction). Dissociation of ␣ 2 dimers and ␤ 4 tetramers follows kinetics of a first-order reaction, while the monomer and dimer combination steps (Reaction III in the Introduction) are second-order processes that depend on protein concentration (7). Interaction of ␣ and ␤ chain monomers in Reaction III is very rapid compared with dissociation of ␤ 4 tetramers to monomers (7). The second-order rate constant of Reaction III for ␣ and ␤ A ( ␣␤ k 5 ) is about 10 5 M Ϫ1 s Ϫ1 (20,26), which means that ␣ monomer binds monomeric ␤ A chain very rapidly, roughly following second-order reaction kinetics. In contrast, dissociation of ␤ 4 34␤ is very slow exhibiting a dissociation constant ( ␤ k 4 ) of about 10 Ϫ3 s Ϫ1 (20). These results indicate that isolated single ␣ and ␤ chains assemble quickly but that formation of tetrameric hemoglobin depends on ␤ 4 stability and total amounts of monomer in solution. Substitution of ␤112 Cys with Val as well as ␤6 Glu with Val in ␤ S chains inhibits tetramer dissociation, whereas substitution of ␤112 Cys with Ser or Asp promotes dissociation. When these ␤ chain variants competed with ␤ S chain for assembly with limiting ␣ chains, the order of assembly of the ␤112 variant hetero-tetramers (e.g. Hb ␤C112S Ͼ Hb A Ͼ Hb ␤C112V) shown in Fig. 5 implies that ␤112 Val forms a weaker ␣ 1 ␤ 1 interface even though the hydrophobic substitution to Val is expected to strengthen hydrophobic interactions with ␣107 Val (G14) and ␣110 Ala (G11) (27). These results, however, may be reconciled by the fact that ␤112 Val facilitates increased tetramer stability compared with ␤ A chains and that dissociation of ␤112 Val homo-tetramers to monomers is much slower than that of ␤112 Cys homo-tetramers. This slow dissociation of ␤112 Val homo-tetramers leads to less monomer formation and therefore less Hb C␤112V compared with Hb A. Furthermore, homo-tetramers containing ␤112 Ser dissociate to monomers more readily than those containing ␤112 Thr. This results in ␤112 Ser promoting ␣ 2 ␤ 2 tetrameric formation compared with ␤112 Thr under limiting ␣ chain conditions. In contrast, even though ␤112 Asp promotes formation of monomers, the results of the order of tetramer assembly (e.g. Hb ␤C112T ϭ Hb ␤C112V Ͼ Ͼ Ͼ Hb ␤C112D) indicates that stability of the interaction between ␣ and ␤ chains plays a key role in assembly in addition to ability to form monomers. However, when ␤112 Val associates with ␣ chains, this substitution of the strong hydrophobic amino acid Val for Cys at ␤112 could stabilize ␣ 1 ␤ 1 interactions due to its ability to exclude water. This would facilitate interactions with the ␣ chain at ␣107 Val (G14) and ␣110 Ala (G11) (27). In contrast, a ␤112 Cys substitution to a hydrophilic amino acid like Asp would inhibit this hydrophobic interaction and result in no homo-tetramer formation and weaken ␣␤ interactions. This also would destabilize formation of hetero-tetramers. In vivo, hemoglobin synthesis occurs in erythroblasts and reticulocytes (28). Even though the assembly scheme in vitro is clear, it is not clear whether these results in vitro apply to assembly in vivo. The newly synthesized ␤ chains in the cytoplasm probably exist at low concentrations (Ͻ10 M), which would favor formation of monomers rather than homo-tetramers (3). However, gel-filtration experiments using ␤ chains at 1.5 M indicate that more than 80% exist as tetramers when analyzing ␤ S and ␤112 Val chains. The combination of heme and apoglobin is so rapid that newly translated subunits probably bind heme prior to assembly (3,29). Our present results on production of soluble ␤-globin chains in bacteria in the presence of heme but in the absence of ␣-globin also suggest that these conditions in vivo promote correct ␤-globin chain folding in the absence of ␣-globin chains. The relevance of ␤ 4 tetramer formation and dissociation under in vivo conditions in which ␣ and ␤ chain synthesis is balanced is not clear. Hb H (␤ chain homo-tetramers) is formed in a severe form of ␣ thalassemia in which patients produce only about one-fourth of the normal amount of ␣ chains. As a result, ␤ chain homo-tetramers form in cells. The ease of dissociation of ␤ chain homo-tetramers to monomers in addition to direct ␣ and ␤ interactions may be rate-limiting for assembly in vivo.
Effect of ␤112 Variants on Dimer-tetramer Equilibrium-Interactions between amino acid residues at the ␣ 1 ␤ 1 and ␣ 1 ␤ 2 interfaces are different (6,28,30). The ␣ 1 ␤ 1 interface remains relatively fixed during deoxygenation, and contacts between the ␣ 1 and ␤ 1 subunits are identical in oxyhemoglobin and deoxyhemoglobin. In deoxyhemoglobin, there are about 40 contacts, including 19 hydrogen bonds. In contrast, there is considerable movement at the ␣ 1 ␤ 2 interface during oxygenation and deoxygenation. When hemoglobin is oxygenated, the total number of contacts drops to about 23, including 12 hydrogen bonds. In view of the important role that conformational isomerization plays in hemoglobin function, it is not surprising that residues at the ␣ 1 ␤ 2 interface are invariant and highly conserved throughout vertebrate evolution (28,30).
Interestingly, recent studies showed that ␤ 112Cys3 Gly chains in tetrameric hemoglobin unexpectedly stabilized ␣ 1 ␤ 2 interactions (34). Our studies also showed that ␤ 112Cys3 Val stabilizes the ␣ 1 ␤ 2 interface without significantly changing oxygen affinity. Substitution of Cys at the ␤112 position with Val appears to stabilize ␣ 1 ␤ 2 interactions but may not affect local quaternary conformational changes induced during the oxygenation-deoxygenation process. Furthermore, even though the ␤112 Ser variant decreased ␣␤ stability by promoting dissociation into monomers, the oxygen affinity of Hb ␤C112S was not influenced by this mutation. These findings suggest that local quaternary conformational changes induced during oxygenation/ deoxygenation are not influenced by this substitution at an ␣ 1 ␤ 1 interaction site. In contrast, the ␤112 Asp variant destabilizes the ␣ 1 ␤ 1 interaction sites and slightly reduced oxygen affinity. Furthermore, this substitution also increased dimer dissociation from hetero-tetramers compared with normal Hb A. These results indicate that the ␤112 amino acid plays a critical role in tetramer-dimer stability and suggest that structural changes at the ␣ 1 ␤ 1 contact can be propagated through the protein and lead to stability at the ␣ 1 ␤ 2 interface. We are now investigating the relationship between mutation at the ␣ 1 ␤ 1 interface and dimer-tetramer stability of hemoglobin. These studies should aid in our understanding of hemoglobin assembly as well as in how changes at one site influence tertiary and quaternary structure at a distant location.