Journal of Protein Chemistry, VoL 1l, No. 2, 1992
The Functional, Oxygen-Linked Chloride Binding Sites of Hemoglobin Are Contiguous Within a Channel in the Central Cavity Hiroshi Ueno 1 and James M. Manning 1'2
Received November 11, 1991
Chloride ion is a major aUosteric regulator for many hemoglobins and particularly for bovine hemoglobin. A site-directed reagent for amino groups, methyl acetyl phosphate, when used for global rather than selective modification of R (oxy) and T (deoxy) state bovine hemoglobin, can acetylate those functional amino groups involved in binding of chloride; the extensively acetylated hemoglobin tetramer retains nearly full cooperativity. The chloride-induced decrease in the oxygen affÉnityparallels the acetylation of bovine hemoglobin (i.e., their effects are mutually exclusive), suggesting that methyl acetyl phosphate is a good probe for the functional chloride binding sites in hemoglobins. Studies on the overall alkaline Bohr effect indicates that the part of the contribution dependent on chloride and reduced by 60% after acetylation is due to amino groups, Val-1 (a) and Lys-81(fl) ; the remaining 40% is contributed by the imidazole side chain of His-146(fl), which is not acetylated by methyl acetyt phosphate, and is not dependent on chloride. The five amino groups--Val-l(a), Lys-99(a), Met-l(fl), Lys-81(fl), and Lys-103(fl)--of bovine hemoglobin that are acetylated in an oxygen-linked fashion are considered functional chloride binding sites. Molecular modeling indicates that these functional chloride binding sites are contiguous from one end of the central cavity of hemoglobin to the other; some of them are aligned within a chloride channel connecting each end of the dyad axis. A generalization that can be made about hemoglobin function from these studies is that the blocking of positive charges within this channel either by binding of chloride or other anions, by covalent chemical modification such as acetylation, or by sitespecific mutagenesis to create additional chloride binding sites each accomplish the same function of lowering the oxygen affinity of hemoglobin. KEY WORDS: Hemoglobin; chloride binding; allosteric regulator; chemical modification.
1. I N T R O D U C T I O N
and Brunori, 1970; Chiancone et al., 1970; Perutz, 1980, 1990). In order to understand more completely the aliosteric mechanism of chloride binding to Hb, it is necessary to identify all the major oxygen-linked binding sites for chloride. However, studies with selectively modified or mutant hemoglobins provide information on only one binding site at a time. It appeared desirable to attempt to modify all the major chloride binding sites in Hb in order to assess the overall contribution of each site to the allosteric effects of chloride, especially in view of the discussion on the amount of functional importance o f chloride in the alkaline Bohr effect (Matthew et al., 1977; Perutz et al., 1980;
Chloride binding to proteins is always reversible, frequently weak, and usually difficult to distinguish with respect to functional from nonfunctional binding. For hemoglobin (Hb), the R (oxy) structure favors the binding of oxygen and the T (deoxy) structure favors oxygen release assisted by the allosteric binding of chloride (and CO2) (Monod et al., 1965; Antonini
The Rockefeller University, 1230 York Avenue, New York, NY 10021. z To whom all correspondence should be addressed.
177 027%8033/92/0400-0177506,50/0 © 1992PlenumPublishingCorporation
178 Ho and Russu, 1987). In addition, it seemed appropriate to determine whether any interrelationship exists among the functional oxygen-linked sites in the structure of the molecule. Several observations suggested to us that methyl acetyl phosphate (MAP), a site-directed reagent previously shown to react specifically with the amino groups in the DPG cleft of Hb, when used in a selective fashion (Ueno et al., 1986, 1989), might be a good choice to probe all of the chloride binding sites if it could be successfully employed for global modification without a large loss in hemoglobin function. We had previously shown that MAP, when used to achieve selective and minimal modification, mimicks the effect of chloride by reacting with certain sites in human Hb to produce a lowered oxygen affinity. We now describe a more extensive treatment of R and T state bovine hemoglobin with MAP in order to identify all of those amino groups that bind chloride in a manner that depends on the conformation of Hb (i.e., oxygen-linked). Bovine Hb was chosen for this study because chloride is a major aliosteric regulator for this Hb (Fronticelli et al., 1984, 1988). Indeed, bovine erythrocytes do not contain 2,3-DPG, a major allosteric regulator for human Hb (Bunn, 1971; Arnone, 1972). Bucci and Fronticelli showed that the intrinsic oxygen affinities of human and bovine Hbs were nearly the same (i.e., bovine Hb is not a low affinity hemoglobin but it responds to chloride to a greater extent than human Hb to attain a lower oxygen affinity) (Fronticelli et at., 1984). The more rapid and extensive acetylation of bovine Hb by 14C-MAP compared to human Hb correlates very well with the effect of chloride on bovine Hb (Ueno et al., 1989) and reinforces the idea that MAP would be a useful probe for amino groups that bind chloride.
2. EXPERIMENTAL PROCEDURES 2.1. Preparation of Bovine Hemoglobin Bovine blood (Pel'freeze) was washed with 0.9% saline and then resuspended in distilled water to its original volume. This lysate was treated with CHC13 and the stroma-free bovine Hb obtained after centrifugation was dialyzed against phosphate-buffered saline (Fronticelli et al., 1984; Ueno et al., 1989). Residual organic phosphate was removed by passing the dialyzed lysate through a mixed ion exchange resin (BioRad) and eluted with distilled water. Two major components of bovine Hb were separated on DE-52
Ueno and Manning cellulose (Whatman) since the blood originated from a hybrid strain of domestic and European cows. The amino acid composition of the second component used for the studies described in this figure matched that of the bovine Hb used in the previous studies (Ueno et al., 1989).
2.2. Treatment of Bovine Hemoglobin with J4C-MAP Four micromoles of bovine Hb in 50 mM hepes buffer a t p H 7.5 was treated with 20 pM of J4C-MAP under either aerobic or anaerobic conditions for 1 hr as described previously (Ueno et al., 1989). Two separate preparations of ~4C-MAP were synthesized according to the original (Kluger and Tsui, 1980) and improved (Kluger et al., 1990) procedures; they had specific activity of 10 cpm/nmol and 500 cpm/nmol, respectively. The results of experiments with each preparation of ~4C-MAP gave consistent and reproducible results. For studies with R state Hb, the buffers were saturated with CO gas. For studies with T state Hb, the solutions were continuously bubbled with N2. The formation ofmetHb amounted to 3-6%, as determined spectrophotometrically. After I hr, the reaction mixture was bubbled with CO gas and dialyzed overnight against 50 mM Hepes buffer saturated with CO gas.
2.3. Isolation and Proteolytic Digestion of Giobin Chain Bovine Hb was separated into its globin chains (Shelton et al., 1982) under the conditions described previously (Ueno et al., 1989). The modified a and fl chains of bovine Hb are separated by this HPLC procedure. Fractions containing radioactivity were pooled and dialyzed against 5% acetic acid overnight at 4°C. Lyophilized globin chains were suspended in 10 ml of 0.1 M ammonium bicarbonate and incubated overnight at 37°C with TPCK-treated trypsin ( i / 5 0 w / w ) ; the solution was further treated with chymotrypsin (1/25w/w) overnight at 37°C, and then lyophilized. The digests were subjected to peptide mapping on reversed-phase HPLC method (Ueno et al., 1989). In some experiments, prior to HPLC, the digestion was separated on TrisAcryl GF-05 column chromatography (TSK) eluted with 0.1 M ammonium bicarbonate. Fractions containing radioactivity were pooled and reapplied to reversed-phase HPLC. The identity of the radioactive peptides after acid hydrolysis in vacuo was established by amino acid analysis on a Beckman 6300 amino acid analyzer equipped
Chloride Binding Sites of Hemoglobin with System Gold software. The assignment of acetylation sites was made as described previously.
2.4. Determination of Oxygen Binding Properties of Acetylated Hemoglobin The oxygen equilibrium curves of unacetylated or MAP-treated bovine hemoglobin either in the absence or presence of 0.1 M or 1.0 M NaCI at 37°C were measured on a modified Hemox analyzer.
2.5. Molecular Modeling of Acetylated Bovine Hb Molecular modeling was performed on Stellar supermini computer using Quanta at the Rockefeller University Computing Facility. The hemoglobin coordinates from Brookhaven protein database (PDB format) were used.
3. RESULTS
3.1. Global Acetylation of Hemoglobin Amino Groups Under Aerobic or Anaerobic Conditions The amino groups of Hb that are acetylated after extensive treatment with ~4C-MAP under aerobic or anaerobic conditions are shown in Table I and Fig. la and b. The identification and the yields of all the lysine-containing peptides that were globally labeled by ~4C-MAP are shown in Table I. There was 75% and 85% modification (the amounts that chromatographically separated from unmodified hemoglobin) of each oxy and deoxy Hb tetramer, respectively, with 2-6 acetylated sites distributed throughout the molecule when Hb was treated with five-fold molar excess of MAP. After separation of the Hb chains, the amounts of modification for the a and /3 chains of oxy Hb were 46% and 83%, respectively, and the amounts of modification for' the a and /3 chains of deoxy Hb were 72% and 80%, respectively. The major oxygen-linked sites (i.e., those that reacted more in deoxy Hb than in oxy Hb) were Met-l, Lys-81, and Lys-103 of the /3-chain; in the a-chain, the major oxygen-linked chloride binding sites were Val-1 and Lys-99 (Fig. la and b). The oxygen-linked binding sites found in this study comprise 70% and 76% of the total reactive sites of the/3-chain under aerobic and anaerobic conditions, respectively, and 87% and 84% of the total reactive sites of the a-chain under aerobic and anaerobic conditions, respectively (Table I).
179 3.2. Effect of Acetylation and Chloride on Oxygen Equilibrium Properties of Hemoglobin Bovine Hb extensively acetylated with MAP maintained a high degree of cooperativity (n= 1.92.2) (Table II), permitting a systematic study of marly of its functional properties. Thus, the oxygen affinity of extensively acetylated bovine Hb was considerably reduced from a Pso of 16 mm Hg or unmodified Hb to a P5o of 44 mm Hg without added NaCl (AP50= 28mm Hg (Table lI)). Furthermore, whereas unmodified bovine Hb shows a large response to chloride (APs0=24 mm Hg), acetylated Hb showed only a slight response (APso = 5 mm Hg) (Table tl)o These results indicate that acetylation with MAP was equally effective in lowering the oxygen affinity by acetylating the amino groups of bovine Hb listed above as was chloride in neutralizing presumably these same positive charges (i.e., their effects are mutually exclusive). Since MAP reacts only with amino groups, other types of side chains [i.e., His146(/3)] would not be labeled by this reagent.
3.3. Effects of Acetylation or Chloride Binding on the Alkaline Bohr Coefficient Studies on human HbA had previously shown that chloride-dependent and chloride-independent sites contribute to alkaline Bohr coefficient (Perutz et al., 1980; Fronticelli et al., 1984; Ho and Russu, 1987; Fermi and Perutz, 1981 ; Bucci and Fronticelli, 1985). Thus, Val-l(a) is primarily responsible for the chloride-dependent alkaline Bohr coefficient and the chloride-independent alkaline Bohr coefficient is contributed predominantly by His-146(fl) (Perutz et al., 1980). Our results (Fig. 2) on the influence of chloride on the alkaline Bohr coefficient of extensively acetylated bovine Hb reinforce these conclusions because we can distinguish between the relative contributions of amino or imidazole groups in the same molecule since MAP reacts with Val-l(a) but not wflh His146(/3). Hence, in the presence of 0.1 M chloride, unmodified bovine hemoglobin has an alkaline Bohr coefficient of 0.48 (Table II), a value consistent with that previously reported for bovine Hb (Fronticelli et aL, 1984) and close to that found for human HbA (Perutz et al,, 1980). In low concentrations of chloride, the Bohr coefficient is reduced to a value of 0.20, which is 40% of the total Bohr coefficient (Table II); this part of the alkaline Bohr coefficient is likely contributed by His-146 (/3) (Perutz et al., 1980; Ohe and Kajita, 1980; Garner et al., 1963; Hill and Davis,
180
Ueno and Manning
Table I. Sites of Acetylation of Oxy and Deoxy Bovine Hemoglobin
The Functional, Oxygen-Linked Chloride Binding Sites of Hemoglobin Are Contiguous Within a Channel in the Central Cavity Hiroshi Ueno 1 and James M. Manning 1'2
Received November 11, 1991
Chloride ion is a major aUosteric regulator for many hemoglobins and particularly for bovine hemoglobin. A site-directed reagent for amino groups, methyl acetyl phosphate, when used for global rather than selective modification of R (oxy) and T (deoxy) state bovine hemoglobin, can acetylate those functional amino groups involved in binding of chloride; the extensively acetylated hemoglobin tetramer retains nearly full cooperativity. The chloride-induced decrease in the oxygen affÉnityparallels the acetylation of bovine hemoglobin (i.e., their effects are mutually exclusive), suggesting that methyl acetyl phosphate is a good probe for the functional chloride binding sites in hemoglobins. Studies on the overall alkaline Bohr effect indicates that the part of the contribution dependent on chloride and reduced by 60% after acetylation is due to amino groups, Val-1 (a) and Lys-81(fl) ; the remaining 40% is contributed by the imidazole side chain of His-146(fl), which is not acetylated by methyl acetyt phosphate, and is not dependent on chloride. The five amino groups--Val-l(a), Lys-99(a), Met-l(fl), Lys-81(fl), and Lys-103(fl)--of bovine hemoglobin that are acetylated in an oxygen-linked fashion are considered functional chloride binding sites. Molecular modeling indicates that these functional chloride binding sites are contiguous from one end of the central cavity of hemoglobin to the other; some of them are aligned within a chloride channel connecting each end of the dyad axis. A generalization that can be made about hemoglobin function from these studies is that the blocking of positive charges within this channel either by binding of chloride or other anions, by covalent chemical modification such as acetylation, or by sitespecific mutagenesis to create additional chloride binding sites each accomplish the same function of lowering the oxygen affinity of hemoglobin. KEY WORDS: Hemoglobin; chloride binding; allosteric regulator; chemical modification.
1. I N T R O D U C T I O N
and Brunori, 1970; Chiancone et al., 1970; Perutz, 1980, 1990). In order to understand more completely the aliosteric mechanism of chloride binding to Hb, it is necessary to identify all the major oxygen-linked binding sites for chloride. However, studies with selectively modified or mutant hemoglobins provide information on only one binding site at a time. It appeared desirable to attempt to modify all the major chloride binding sites in Hb in order to assess the overall contribution of each site to the allosteric effects of chloride, especially in view of the discussion on the amount of functional importance o f chloride in the alkaline Bohr effect (Matthew et al., 1977; Perutz et al., 1980;
Chloride binding to proteins is always reversible, frequently weak, and usually difficult to distinguish with respect to functional from nonfunctional binding. For hemoglobin (Hb), the R (oxy) structure favors the binding of oxygen and the T (deoxy) structure favors oxygen release assisted by the allosteric binding of chloride (and CO2) (Monod et al., 1965; Antonini
The Rockefeller University, 1230 York Avenue, New York, NY 10021. z To whom all correspondence should be addressed.
177 027%8033/92/0400-0177506,50/0 © 1992PlenumPublishingCorporation
178 Ho and Russu, 1987). In addition, it seemed appropriate to determine whether any interrelationship exists among the functional oxygen-linked sites in the structure of the molecule. Several observations suggested to us that methyl acetyl phosphate (MAP), a site-directed reagent previously shown to react specifically with the amino groups in the DPG cleft of Hb, when used in a selective fashion (Ueno et al., 1986, 1989), might be a good choice to probe all of the chloride binding sites if it could be successfully employed for global modification without a large loss in hemoglobin function. We had previously shown that MAP, when used to achieve selective and minimal modification, mimicks the effect of chloride by reacting with certain sites in human Hb to produce a lowered oxygen affinity. We now describe a more extensive treatment of R and T state bovine hemoglobin with MAP in order to identify all of those amino groups that bind chloride in a manner that depends on the conformation of Hb (i.e., oxygen-linked). Bovine Hb was chosen for this study because chloride is a major aliosteric regulator for this Hb (Fronticelli et al., 1984, 1988). Indeed, bovine erythrocytes do not contain 2,3-DPG, a major allosteric regulator for human Hb (Bunn, 1971; Arnone, 1972). Bucci and Fronticelli showed that the intrinsic oxygen affinities of human and bovine Hbs were nearly the same (i.e., bovine Hb is not a low affinity hemoglobin but it responds to chloride to a greater extent than human Hb to attain a lower oxygen affinity) (Fronticelli et at., 1984). The more rapid and extensive acetylation of bovine Hb by 14C-MAP compared to human Hb correlates very well with the effect of chloride on bovine Hb (Ueno et al., 1989) and reinforces the idea that MAP would be a useful probe for amino groups that bind chloride.
2. EXPERIMENTAL PROCEDURES 2.1. Preparation of Bovine Hemoglobin Bovine blood (Pel'freeze) was washed with 0.9% saline and then resuspended in distilled water to its original volume. This lysate was treated with CHC13 and the stroma-free bovine Hb obtained after centrifugation was dialyzed against phosphate-buffered saline (Fronticelli et al., 1984; Ueno et al., 1989). Residual organic phosphate was removed by passing the dialyzed lysate through a mixed ion exchange resin (BioRad) and eluted with distilled water. Two major components of bovine Hb were separated on DE-52
Ueno and Manning cellulose (Whatman) since the blood originated from a hybrid strain of domestic and European cows. The amino acid composition of the second component used for the studies described in this figure matched that of the bovine Hb used in the previous studies (Ueno et al., 1989).
2.2. Treatment of Bovine Hemoglobin with J4C-MAP Four micromoles of bovine Hb in 50 mM hepes buffer a t p H 7.5 was treated with 20 pM of J4C-MAP under either aerobic or anaerobic conditions for 1 hr as described previously (Ueno et al., 1989). Two separate preparations of ~4C-MAP were synthesized according to the original (Kluger and Tsui, 1980) and improved (Kluger et al., 1990) procedures; they had specific activity of 10 cpm/nmol and 500 cpm/nmol, respectively. The results of experiments with each preparation of ~4C-MAP gave consistent and reproducible results. For studies with R state Hb, the buffers were saturated with CO gas. For studies with T state Hb, the solutions were continuously bubbled with N2. The formation ofmetHb amounted to 3-6%, as determined spectrophotometrically. After I hr, the reaction mixture was bubbled with CO gas and dialyzed overnight against 50 mM Hepes buffer saturated with CO gas.
2.3. Isolation and Proteolytic Digestion of Giobin Chain Bovine Hb was separated into its globin chains (Shelton et al., 1982) under the conditions described previously (Ueno et al., 1989). The modified a and fl chains of bovine Hb are separated by this HPLC procedure. Fractions containing radioactivity were pooled and dialyzed against 5% acetic acid overnight at 4°C. Lyophilized globin chains were suspended in 10 ml of 0.1 M ammonium bicarbonate and incubated overnight at 37°C with TPCK-treated trypsin ( i / 5 0 w / w ) ; the solution was further treated with chymotrypsin (1/25w/w) overnight at 37°C, and then lyophilized. The digests were subjected to peptide mapping on reversed-phase HPLC method (Ueno et al., 1989). In some experiments, prior to HPLC, the digestion was separated on TrisAcryl GF-05 column chromatography (TSK) eluted with 0.1 M ammonium bicarbonate. Fractions containing radioactivity were pooled and reapplied to reversed-phase HPLC. The identity of the radioactive peptides after acid hydrolysis in vacuo was established by amino acid analysis on a Beckman 6300 amino acid analyzer equipped
Chloride Binding Sites of Hemoglobin with System Gold software. The assignment of acetylation sites was made as described previously.
2.4. Determination of Oxygen Binding Properties of Acetylated Hemoglobin The oxygen equilibrium curves of unacetylated or MAP-treated bovine hemoglobin either in the absence or presence of 0.1 M or 1.0 M NaCI at 37°C were measured on a modified Hemox analyzer.
2.5. Molecular Modeling of Acetylated Bovine Hb Molecular modeling was performed on Stellar supermini computer using Quanta at the Rockefeller University Computing Facility. The hemoglobin coordinates from Brookhaven protein database (PDB format) were used.
3. RESULTS
3.1. Global Acetylation of Hemoglobin Amino Groups Under Aerobic or Anaerobic Conditions The amino groups of Hb that are acetylated after extensive treatment with ~4C-MAP under aerobic or anaerobic conditions are shown in Table I and Fig. la and b. The identification and the yields of all the lysine-containing peptides that were globally labeled by ~4C-MAP are shown in Table I. There was 75% and 85% modification (the amounts that chromatographically separated from unmodified hemoglobin) of each oxy and deoxy Hb tetramer, respectively, with 2-6 acetylated sites distributed throughout the molecule when Hb was treated with five-fold molar excess of MAP. After separation of the Hb chains, the amounts of modification for the a and /3 chains of oxy Hb were 46% and 83%, respectively, and the amounts of modification for' the a and /3 chains of deoxy Hb were 72% and 80%, respectively. The major oxygen-linked sites (i.e., those that reacted more in deoxy Hb than in oxy Hb) were Met-l, Lys-81, and Lys-103 of the /3-chain; in the a-chain, the major oxygen-linked chloride binding sites were Val-1 and Lys-99 (Fig. la and b). The oxygen-linked binding sites found in this study comprise 70% and 76% of the total reactive sites of the/3-chain under aerobic and anaerobic conditions, respectively, and 87% and 84% of the total reactive sites of the a-chain under aerobic and anaerobic conditions, respectively (Table I).
179 3.2. Effect of Acetylation and Chloride on Oxygen Equilibrium Properties of Hemoglobin Bovine Hb extensively acetylated with MAP maintained a high degree of cooperativity (n= 1.92.2) (Table II), permitting a systematic study of marly of its functional properties. Thus, the oxygen affinity of extensively acetylated bovine Hb was considerably reduced from a Pso of 16 mm Hg or unmodified Hb to a P5o of 44 mm Hg without added NaCl (AP50= 28mm Hg (Table lI)). Furthermore, whereas unmodified bovine Hb shows a large response to chloride (APs0=24 mm Hg), acetylated Hb showed only a slight response (APso = 5 mm Hg) (Table tl)o These results indicate that acetylation with MAP was equally effective in lowering the oxygen affinity by acetylating the amino groups of bovine Hb listed above as was chloride in neutralizing presumably these same positive charges (i.e., their effects are mutually exclusive). Since MAP reacts only with amino groups, other types of side chains [i.e., His146(/3)] would not be labeled by this reagent.
3.3. Effects of Acetylation or Chloride Binding on the Alkaline Bohr Coefficient Studies on human HbA had previously shown that chloride-dependent and chloride-independent sites contribute to alkaline Bohr coefficient (Perutz et al., 1980; Fronticelli et al., 1984; Ho and Russu, 1987; Fermi and Perutz, 1981 ; Bucci and Fronticelli, 1985). Thus, Val-l(a) is primarily responsible for the chloride-dependent alkaline Bohr coefficient and the chloride-independent alkaline Bohr coefficient is contributed predominantly by His-146(fl) (Perutz et al., 1980). Our results (Fig. 2) on the influence of chloride on the alkaline Bohr coefficient of extensively acetylated bovine Hb reinforce these conclusions because we can distinguish between the relative contributions of amino or imidazole groups in the same molecule since MAP reacts with Val-l(a) but not wflh His146(/3). Hence, in the presence of 0.1 M chloride, unmodified bovine hemoglobin has an alkaline Bohr coefficient of 0.48 (Table II), a value consistent with that previously reported for bovine Hb (Fronticelli et aL, 1984) and close to that found for human HbA (Perutz et al,, 1980). In low concentrations of chloride, the Bohr coefficient is reduced to a value of 0.20, which is 40% of the total Bohr coefficient (Table II); this part of the alkaline Bohr coefficient is likely contributed by His-146 (/3) (Perutz et al., 1980; Ohe and Kajita, 1980; Garner et al., 1963; Hill and Davis,
180
Ueno and Manning
Table I. Sites of Acetylation of Oxy and Deoxy Bovine Hemoglobin
