Effect of Amidination of Lysyl Residues on the Oxygen Affinity of Human Hemoglobin SPECIFICITY

Treatment of human oxyhemoglobin with methylacetimidate results in selective amidination of the epfilon-amino group of lysin C5(40)alpha. The modified hemoglobin exhibits increased oxygen affinity, high cooperatively, and normal Bohr effect. Hybrid molecules containing amidinated beta chains and normal alpha chains have normal ligand-binding properties, whereas hybrid molecules containing amidinated alpha chains have ligand-binding properties identical with fully amidinated hemoglobin. Amidination of deoxyhemoglobin produces only minimal changes in ligand-binding properties. We propose that amidination of lysine C5(40)alpha prevents its participation in the salt bond with histidine HC3(146)beta in deoxyhemoglobin, thus shifting the allosteric equilibrium in favor of the high affinity oxy conformation.

guanidinium group of arginme HC3(141)cu and the /3-carboxyl group of aspartate H9(126)ar of the opposite chain (2). A modification of the hemoglobin structure which specifically interferes with the formation of these salt bonds is expected to result in a lower stability of the deoxy (T) conformation, and a shift in the position of equilibrium toward the oxy (R) conformation. This shift could be observed as an increase in the oxygen affinity of the modified hemoglobin.
In this investigation we have utilized chemical modification to elucidate the role of the amino groups of human hemoglobin in the binding of oxygen. We have selected methyl acetimidate as the reagent of choice for two reasons. The conditions for extensive modification are rather gentle, pH 9 at room temperature for 15 to 60 min; and the modification does not alter the net charge of the protein at neutral pH, the acetamidino group having a pK above 12 (3). In addition, a number of proteins have been subjected to extensive amidination with few if any accompanying changes in physical or biological properties (4)(5)(6)(7)(8)(9).
A preliminary account of this work has been presented (10). for 30 min at O', and the red cell stromata were removed by centrifugation.

Synthesis
The hemolysate was concentrated 4-fold by ultrafiltration and dialyzed against 0.01 M sodium phosphate, pH 6.8. Hemoglobin A was purified by chromatography on CM-Sephadex C-50 (ll), and was concentrated to 60 mg/ml by ultrafiltration. Hemoglobin concentration and ferrihemoglobin content were determined spectrophotometrically (12). Preparation of globin was achieved by the acid-acetone precipitation method of Winterhalter and Huehns (11). Native and amidinated hemoglobin were dissociated into a and /3 chains by treatment with excess p-hydroxymercuribenzoate (13) followed bv chromatoeranhv on DEAE-Senhadex A-50 (14). Purity of tde separated 'Eha& was ascertained-by disc gel ele&& phoresis at pH 8.9 and by amino acid analysis. Sulfhydryl groups of the a chains were regenerated on a column of CM-Sephadex C-50 by treatment with 0.3 M 2-mercaptoethanol in 0.01 M sodium phosphate, pH 6.7 (14). Sulfhydryl groups of the fl chains were regenerated on a column of DEAE-Sephadex A-50 by treatment with 0.03 M 2-mercaptoethanol in 0.1 M Tris-HCl, pH 8. Hybrid hemoglobins were formed by mixing stoichiometric amounts of the appropriate Q and 0 chains at 0" for 1 hour. The samples were then dialyzed against 0.01 M sodium phosphate, pH 6.8, and purified by chromatography on CM-Sephadex (11). Amidination of Hemoglobin-A stock solution of methyl  in 0.1 M Tris-HCI, pH 9, was prepared and neutralized with 1 eq of NaOH immediately before use. This solution was diluted to the desired concentration with 0.04 M NaCl in 0.1 M Tris-HCl, pH 9, and was added to purified hemoglobin which had been dialyzed against 0.01 Y Tris-HCl, pH 9. The hemoglobin concentration in the reaction mixture was 0.27 mM. After 15 min at room temperature the reaction was terminated by the addition of 0.2 mmol of NaHPPOd per ml of sample. The amidinated samples were dialyzed against 0.01 M Tris-HCl, pH 7.4, or against 0.05 M sodium phosphate, pH 6.0, containing 0.1 M NaCl and 1 mu EDTA.
Determination of Number of Residues Amidinated-The radioactivity of samples of amidinated hemoglobin or hemoglobin peptides was measured in a Beckman LSSOOB liquid scintillation spectrometer using a scintillation fluid of 0.45% 2,5-diphenyloxazole in toluene-Triton X-100 (2:l). The aqueous sample plus water made up 10% of the total volume. Counting efficiency was determined using an internal standard of [14C]toluene. The number of residues amidinated was calculated by dividing the radioactivity incorporated per mmol of protein or peptide by the specific radioactivity of the methyl [IaClacetimidate.
The number of lysyl e-amino residues amidinated was calculated by subtracting the number of amidinated NH*-terminal residues from the total number of amidinated residues.
The number of NH*-terminal valyl a-amino groups amidinated was determined indirectly by quantitating the nonamidinated amino groups with KN"CO (15). Amidinated a-or B-alobin chains were dissolied in 8 M urea'cbntaining 2% N-ethylmorpholine, DH 8. and 0.6 M KN"CO (7.3 mCi/mmol). The nrotein concentralion kas 10 mg/ml. After.24 ho&s at 4d0 unreacted cyanate was removed by exhaustive dialysis against 10-' N HCl. The sample was lyophilized and dissolved in 6 N HCI and a portion was hydrolyzed at 145" for 4 hours (16) and the protein concentration was determined by amino acid analysis. The remainder of the sample was heated ai 100" for 1 hour tocyclize the carbamylated residues. The resulting hvdantoins were eluted from Dowex 50 H+ with -I water, and the radioactivity was quantitated. The radioactive product was identified as valine hydantoin by chromatography on Dowex 1 (17), and by amino acid analysis of a basic hydrolysate (15). Treatment of native hemoglobin gave 1.02 and 0.98 mol of valine hydantoin per mol of a and /3 chain, respectively.
Cyanogen Bromide Cleavage of cr-Globin Chains-The a chain was cleaved into three fragmenta by treatment with cyanogen bromide (18), after reduction of methionine sulfoxide residues with 25% 2-mercaptoethanol (19). After 20 hours of incubation with CNBr the protein solution was evaporated to dryness, redissolved in 7yo formic acid, and chromatographed on Sephadex G-50. Peptides were detected by their absorbance at 280 nm and by their radioactivity. Fragments corresponding to residues 1 to 32 and 33 to 76 were further purified by rechromatography on Sephadex G-50, and were identified and quantitated by amino acid analysis. The fragment corresponding to residues 77 to 141 was purified by chromatography on phosphocellulose with a gradient of KC1 in HaPOd (20), followed by gel filtration on Sephadex G-75. The peptide was identified and quantitated by amino analysis. Chymotryptic Digeslion of CNBr Fragments-The CNBr fragments were-digested (21) at-pH 8 with chymotrypsin (Sigma type II) which weshowed to be free of trvnsin activitv (22). The chvmotriptic digests were dissolved in 01625 N phosp'hdric acid an; the pH was adjusted to 2.5. The samples were chromatographed on phosphocellulose (20) using the elution systems described later. The radioactive fractions were pooled and further purified by gel filtration as described later. Peptides were identified and quantitated by amino acid analysis.
The partial pressure of oxygen was messured continuously using a YSI 5331 polarographic oxygen sensor with a high sensitivity Teflon membrane. The sensor output was amplified by a YSI model 53 oxygen monitor. The ferrihemoglobin content of the samples did not exceed 5$&.

Extent of Am&u&n
of Hemoglatin-The extent of amidination of the amino groups in hemoglobin is proportional to the concentration of methyl acetimidate employed. Hemoglobin (0.27 mM> was treated with different concentrations of methyl ["Clacetimidate from 2.7 to 108 mu. The amidinated samples were separated into (Y and @ chains and the extent of amidination of lysyl and valyl residues was determined ( Table I). The (Y and p chains are amidinated to essentially the same extent. The e-amino groups of the lysyl residues are considerably more reactive than are the a-amino groups of the valyl residues. E$ect of Amidination on Oxygen Afinity-With increasing extent of amidination, the reoxygenation equilibrium curves are shifted to the left, corresponding to an increase in oxygen af%nity (Fig. 1). The ~60 values for the native, 29% amidinated, and 99% amidinated samples are 1.75, 1.04, and 0.47 mm Hg, respectively. The oxygen afKnities of a number of hemoglobin samples amidinated to varying extents were measured at pH 6.0, 6. 8, 7.4, and 9.0, and at pH 7.4 in the presence of 5 mM 2,3-diphosphoglycerat (Table II). An increase in oxygen aEnity with increasing extent of amidination is seen at all pH values and in the presence of DPG.lThe relationship between p&o and the extent of amidination of lysyl and valyl amino groups is shown in Fig. 2. An initial sharp decrease in ~50 is observed corresponding to amidination of up to 60% of the lysyl residues. At this point, only 8% of the valyl residues have been amidinated. Further amidination of up to 100% of the lysyl and 28% of the valyl residues does not result in any further increase in oxygen afEnity. Thus it is obvious that the effect on oxygen affinity is due to amidination of lysyl e-amino groups, and not to amidination of valyl a-amino groups. Furthermore, since the maximal effect on oxygen aflinity is observed after amidination of only 60% of the lysyl residues, it appears as l The abbreviations used are: DPG, 2,3-diphosphoglycerate; Hb, hemoglobin.
if the reagent is somewhat selective toward certain lysyl residues which are functionally involved in oxygen binding.
E&d of Amidination on Cooperativity-The cooperativity of oxygen binding, ss measured from the slope of the Hill plot at 50% ligand saturation, is only slightly decreased by amidiition. The n values for native, 290/ and 99% amidinated hemoglobins ( Fig. 1) are 2.7, 2.5, and 2.3, respectively. The n values for the samples presented in Table II are all above 2.0. Since the cooperativity of ligand binding is high, it is unlikely that extensive structural changes occur upon amidiuation.
E&ct of Amidinatkn on Alkaline Bohr Effect and DPG Effect-The pso values of native and amidinated hemoglobin samples decrease as the pH is increased from 6.0 to 9.0 ( Table II). The percentage of the Bohr effect which is maintained on amidinated samples is calculated from the following formula (24) To Bohr effect = where p,, and p,,. represent the oxygen afhnities of native and amidinated hemoglobin at the indicated pH. The Bohr effect present in native hemoglobin is defiued as lOO'%. Amidination of hemoglobin to the extent of 95% decreases the Bohr effect by only 25 to 30%. We recognize that there are inherent limitations in these calculations of the relative Bohr effect, but the data (Table III) taken as a whole show that amidination of hemoglobin causes minimal change in the Bohr effect between pH 6.0 and 7.4. The ~60 values of native and amidinated hemoglobin are increased in the presence of DPG (Table II, Fig. 2). For purposes of presentation only, we have expressed the DPG effect in amidinated hemoglobin as a percentage, according to the following formula:   (Table III).
Oxygen Ajinity Studies of Hybrid Hemoglobins-Tetrameric hemoglobin was amidinated to the extent of 29 or 99%. A portion of each sample was separated into amidinated a! and amidinated /I chains, which were recombined with native cr and /I chains to form the hybrid hemoglobins Hba@* and Hba*@, containing native (Y and amidinated @ chains, or native /3 and amidinated a! chains, respectively. The hybridization showed no apparent selectivity against the amidinated chains, as yields for all recombinants were comparable, and the specific radioactivities of the hybrid products were those predicted from random reassociation (Table IV). Hemoglobins reconstituted from native cx and /3 chains, or from amidinated (Y and @ chains (Hbafi, Hbcr*p*) do not differ in oxygen aflinity from the hemoglobins from which the chains were derived. The oxygen equilibrium curves of hybrids containing amidinated /3 chains and native (Y chains (Hba/3*) are superimposable with those of native hemoglobin in the absence of DPG. The oxygen equilibrium curves of hybrids containing amidinated (Y chains and native p chains (Hba*& are superimposable with those of the corresponding amidinated hemoglobin in the absence of DPG. These data show conclusively that the increased oxygen Sample Per cent ho (-DPG) ho (+DPG) Per cent amidination DPG effect Native Hb Oxy-Hb* Deoxy-Hb* afhnity of amidinated hemoglobin is due to selective modification of 1 or more lysyl residues in the (Y chain. In the presence of DPG the oxygen-binding curves of the hybrids containing 99% amidinated chains are no longer superimposable with those of the control or 99% amidinated hemoglobin (Table IV) (see "Discussion").
Oxygen Ajinities of Hemoglobin Amidinated in Oxy and Deoxy States-Oxyhemoglobin was amidinated using 6.7 mM methyl [i4C]acetimidate as described above. Another sample of hemoglobin was deoxygenated under nitrogen, and the amidination was carried out in a glove bag filled with nitrogen. The extent of amidination was approximately the same for the two samples (Table V). However, the oxygen aflinity of the sample amidinated in the deoxy state was not altered nearly as much as was that of the sample amidinated in the oxy state. It therefore appears that the key lysyl residue (or residues) in the a! chains are less accessible to the reagent when the hemoglobin molecule is in the deoxy conformation. One such lysyl residue which is exposed in the oxy and not in the deoxy conformation is lysine C5(4O)(r. This residue is involved in a salt bridge which has been proposed to stabilize the deoxy conformation (2).
Localization of Amidinaicd Residues in cr Chain. CNBr Frag-ment&on-Hemoglobin was treated with methyl ["C]acetimidate (3.2 x lo6 dpm/pmol) and was separated into (Y and /3 chains as described above. The heme was removed from the a! chain and the specific radioactivity of the cr-globin was determined to be 3.8 x lo6 dpm/pmol. Since the cy chain contains 11 lysyl residues, the average specific activity of the lysyl residues is 3.4 x 10' dpm/pmol lysine. The cu-globin chain was treated with CNBr which should result in cleavage at methionines 32 and 76. The resulting mixture was separated on Sephdex G-50 (Fig. 3). Five peaks were observed from measurements of radioactivity and absorbance at 280 run. The order of elution is intact a-globin (residues 1 to 141), peptide BC (residues 33 to 141), peptide AB (residues 1 to 76), and the three desired CNBr fragments, C (residues 77 to 141), B (residues 33 to 76), and A (residues 1 to 32). Fragments A, B, and C were separated from each other by chromatography on phosphocellulose (20). Peptides A, B, and C were identified and quantitated by amino acid analysis (Table VI). The specific radioactivities of peptides A, B, and C are 4.2 x 104, 4.6 x 104, and 1.8 X 10' dpm/pmol of lysine. These specific activities are more conveniently expressed relative to the specific activity of the average lysine in the (Y chain, taken as 1.0. Thus the relative specific activities of peptides A, B, and C are 1.  Eight peaks were detected by absorbance at 215 nm, and six of these peaks were radioactive (Fig. 4). Samples from Peaks AI, AZ, Aa, and Ad were further purified by gel filtration. A1 was separated by chromatography on Sephadex G-15 into one major and two very minor fractions. The major fraction contained all the radioactivity and was identified as residues 10 to 14 (Table VII). AS was separated on Sephadex G-25 into three fractions of approximately equal quantity based on absorbance at 215 nm. The first two fractions, As and A'z, contained nearly equal amounts of radioactivity. The third fraction was not radioactive. Peptides AZ and A'2 are identified as residues 1 to 11 and 1 to 14 (Table VII). It is assumed that the tryptophanyl residue (No. 14), which was detectable by absorbance at 280 run, is responsible for the retardation of peptide A'2 on Sephadex G-25. Aa was separated into three fractions on Sephadex G-25. The first fraction was the major one and contained all the radioactivity applied to the Sephadex column. It is identified as residues 3 to 11 (Table VII). Ad was separated into one major and two Three-milliliter fractions were collected. 1 UEE NUMBER very minor fractions on Sephadex G-25. The major fraction contained all the radioactivity and is identified as residues 15 to 24 (Table VII).
Separation and Identification of Chymotryptic Peptides from Fragment B-A chymotryptic digest of CNBr Fragment B was chromatographed on phosphocellulose. The peptides were eluted with two successive salt gradients. The first gradient was prepared in a three-chamber gradient mixer with 200 ml of 0.025 N HsPOd in each chamber, containing 0, 0.10, and 0.20 M KCl, respectively. This was followed by a gradient formed from 200 ml of 0.025 N HaPOd + 0.20 M KC1 and 200 ml of 0.025 N HaPOd + 0.50 M KCl. Eight distinct peaks were detected by absorbance at 215 nm and five were radioactive (Fig. 5). Samples from Peaks B,, Bz, Bs, Bd, and B6 were further purified by gel filtration on Sephadex G-25. B1 was separated into two fractions, the second of which contained all the radioactivity and was identified as residues 33 to 42 (Table VIII). & was separated into two equal fractions, the first of which contained all the radioactivity and was identified as residues 34 to 48 (Table VIII). B3 was homogeneous on Sephadex G-25 and was identified as residues 47 to 56 (Table VIII). Bc was separated into three fractions. The first fraction contained all the radioactivity and was identified as    IX in Table IX. Lysyl residues 11, 16, 40, and 56 are each uniquely Extent of amiclination of lysine residues of (Y chain present in one or more peptides, and their specific activities are Specific activities (S.A.), relative specific activities (R.S.A.), directly obtainable. The specific activity of lysine 7 is calculated and per cent amidination of each radioactive chymotryptic pep-from the specific activity of peptide Al, containing lysine 11, and tide isolated from Fragments A and B.
_-_peptides &, A's, and Aa, containing lysines 7 and 11. The average specific activity for lysine 7 in these peptides is 3.6 X lo6 dpm/ rmol of lysine, corresponding to a relative specific activity of 1.1. No unique peptides were isolated for lysines 60 and 61, and an average value for the 2 residues is presented. Of all of the lysyl residues shown in the table, lysine 40 has by far the highest specific activity, showing a selectivity of the reagent for this residue. This finding is consistent with our other findings that increased oxygen affinity is a result of selective amidination of an a! chain lysyl residue which is less accessible to the reagent when hemoglobin is in the deoxy conformation. Amid&&ion of hemoglobin results in a rather specific increase in oxygen aflinity with minimal effects on cooperativity and alkaline Bohr effect. We propose that this increase in oxygen aflinity is due to amidination of 1 specific lysyl residue, lysine C5(4O)(r, resulting in displacement of the allosteric equilibrium between oxy (R) and deoxy (T) conformations toward the high affinity oxy (R) form.
Extensively aminidinated hemoglobin has a very high oxygen aflinity, comparable to that of myoglobin or of isolated hemoglobin chains (Table II). The ligand-binding curves are, however, definitely sigmoid in character, indicating cooperative binding of oxygen. The Hill plots for extensively amidinated samples are somewhat asymmetrical (Fig. 1)) but all have slopes greater than 2.0 at 50% ligand saturation. The unexpectedly high oxygen a S.A. units are disintegrations per min per ~01 lysine x 10-a. * Based on the specific radioactivity of methyl [Wlacetimidate (3.2 X 106 dpm/mol).
residues 59 to 66 (Table VII). B5 was separated into three fractions of which the second contained all the radioactivity and was identified as residues 57 to 66 (Table VIII).
The distribution of radioactivity within CNBr Fragment C was not investigated, since the relative specific activity was low relative to that of the other two fragments.
Specijic Radioactivity of Individual Lysyl R&dues-The specific radioactivity of each of the chymotryptic peptides is presented afhnity may possibly reflect an effect of extensive amidiition on a! chains. The increase in oxygen affinity seen for this hemoglobin the oxygen arsnity of the R form of hemoglobin, in addition to sample is 40% of the over-all difference in oxygen afhnity bean effect on the allosteric equilibrium between T and R forms.
tween control and totally amidinated hemoglobin (Fig. 2). We The change in oxygen afhnity is proportional to the extent of therefore conclude that there exists a direct cause and effect relaamidination of lysyl residues, and is not related to the extent of tionship between amidination of lysine C5(4O)a! and an increase amidination of NHz-terminal valyl residues (Fig. 2). The p60 for in oxygen afRnity of the tetrameric hemoglobin. Since lysine oxygen decreases progressively with amidination of up to 60 to C5(4O)cr is known to participate in a salt bond with histidine 70% of the lysyl residues. Further amidination causes only a HC3(146)/3 which stabilizes the deoxy (T) conformation, we minimal additional increase in oxygen affinity (Fig. 2, Table II).
propose that amidmation of the e-amino group prevents forma-These data indicate that certain residues which are more sus-tion of the salt bond, thus shifting the position of allosteric equiceptible to amidination than the bulk of the lysines are responsi-librium toward the high afhnity oxy (R) conformation. ble for the increased oxygen aflinity upon amidination. It is ap-Kihnartin and Hewitt (26) conclude that, since at pH 8.5 des parent that the NHrterminal valyl residues are not the residues (Arg-14lol) hemoglobin lacks cooperativity (n = l), the salt in question, since nearly maximal effect on oxygen aflinity is seen bridge involving lysme 4Oa is too weak to stabilize the deoxy upon amidination of only 8 to 10% of the valyl o-amino groups.
conformation in the absence of the salt bridges involving arginme Further amidination to an extent of 28% does not appreciably 141a! and the imidazole group of histidine 1468. We have shown affect the oxygen afRnity (Fig. 2).
that amidination of lysine 4Oar, which we propose to prevent The reactive lysyl residues which are responsible for the effect formation of the salt bridge between lysine 4Oat and histidine 146, of amidiition on oxygen aflinity reside in the (Y chain. Hybrid does not appreciably affect the cooperativity of oxygen binding. hemoglobin molecules, containing approximately 3 amidmolysyl Although the degree of sigmoidicity of the oxygen equilibrium residues per chain, were prepared from native hemoglobin and curves is apparently decreased in the more extensively amidinated from hemoglobin which had been amidinated in the tetrameric samples, the Hill plots in all cases have slopes greater than 2.0 state. The oxygen-binding curves of native hemoglobin (Hbafl) at 50% ligand saturation. Since the accuracy of polarographic and of hybrid hemoglobin containing amidmated /3 chains measurement of very low oxygen concentrations (po, < 1 mm (Hba/3*) are superimposable in stripped hemoglobin and in the Hg) is poor, it is difficult to determine whether the altered shape presence of 2 mM DPG. The oxygen-binding curves of amidinated of the oxygen equilibrium curves and Hill plots is due to a change hemoglobin (Hbar*@*) and of hybrid hemoglobin containing in heme-heme interaction, or is simply due to instrumental error. amidinated a! chains (Hbru*@ are likewise superimposable in the In any case, even the exhaustively amidinated hemoglobin is still presence and absence of DPG. Hybrid hemoglobin molecules highly cooperative in its oxygen binding. Amidination of lysine containing 11 amidinolysyl and 0.3 amidinovalyl residues per 4Ocu is thus proposed to result in a destabilization of the deoxy (T) chain were similarly prepared. The oxygen-binding curves of conformation to an extent which permits the T to R transition to Hba/3 and of HbaB* are superimposable in the absence of DPG. occur at a lower ligand saturation than seen in native hemoglobin. The oxygen-binding curves of Hbcu*/3* and of Hbcr*P are likewise This destabilization is not so great as to prevent the formation superimposable in the absence of DPG. In the presence of DPG, of the deoxy quaternary conformation, however, so cooperativity however, Hbafl* has a higher oxygen s,fhnity than Hbaj3, and of ligand binding is retained (n > 2). Our results are therefore Hba*/3 has a lower oxygen afhnity than Hbcr*P*. The significance consistent with the proposal of Perutz and TenEyck (27) that of this result will be discussed below.
These reactive lysyl residues of the (Y chain appear to be amidinated to a greater extent in the oxy conformation than in the deoxy conformation, although the over-all extent of amidination of hemoglobin is the same in both conformations. Amidination of oxyhemoglobin to an extent of 260/, lowers the ~SO from 1.75 to 1.08 mm Hg. Amidination of deoxyhemoglobin to an extent of 28% only lowers the ps to 1.45 mm Hg (Table V). It selective removal or blockage of salt bridges leads to a specific increase of oxygen aflinity and reduction of the Hill coe5cient.
The opening and closure of the salt bridges involving the imidazole of histidine HC3(146)/3 and the amino group of valine NA(l)a! accounts for most, or possibly all, of the alkaline Bohr effect (27). Amidination of lysine HC3(40)(~ causes a very minimal reduction of the Bohr effect (Table III) while presumably preventing the formation of the salt bond between lysine 4Oar therefore appears as if the key lysyl residues are less accessible to and the carboxyl of histidine 146/3. It is therefore probable that the reagent in the deoxy conformation, implicating lysine C5-the salt bridges between the imidazole of histidine 1466 and as-(4O)cu, which forms a salt bond with histidine HC3(146)/3 in the partate FGl(94)P and between vahne la and the carboxyl of deoxy conformation, but is free in the oxy conformation (2).
The extent of amidination of a number of lysyl residues of the u chain was determined in a hemoglobin sample which contained 1.2 amidinolysyl residues per a! chain. This represents an average of 0.11 amidino group at each of the 11 lysyl positions. We found the following extent of amidination (Table IX): lysine 7, 0.11; lysine 11, 0.13; lysine 16, 0.16; lysine 40, 0.40; lysine 56, 0.03; lysine 60-61 (average), 0.10; lysine 90-99-127-139 (average), 0.05. Lysine C5(4O)(w is amidinated 2.5 times as extensively as is lysine A14(16)(r, and 3 times as extensively as any other lysyl residue in the (Y chain. We consider it unlikely that lysine A14 (16)cu is involved functionally in the effect of amidination on oxygen affinity, since the variant hemoglobin I (lysine A14(16)a to glutamate) has no apparent abnormal properties (25).
Amidination of hemoglobin to the extent of 1.2 lysyl residues per chain results in amidination of lysine C5(4O)a! in 40% of the arginine HC3(141)a remain intact in amidinated hemoglobin.
Deoxyhemoglobin binds DPG on the dyad axis in the central cavity so that its acidic groups are coordinated to the imidazole groups of histidine H21(143)8 and the amino groups of valine NA l(l)/3 of both /3 chains and to the e-amino group of lysine EF6(82)@ of one of the /3 chains (27). The ability of DPG to lower the oxygen aflinity decreases progressively as more amino groups are amidinated (Table III). Amidination of 29% of the amino groups of hemoglobin reduces to about 75% the ability of DPG to lower oxygen afhnity (Table IV). This effect is due to amidination of residues in the a! chain, since Hba@* exhibits the same DPG effect as native Hb, whereas Hbcu*P exhibits the same DPG effect as amidinated hemoglobin. Amidination of 100% of the lysine and 28% of the valine residues reduces to about 35% the ability of DPG to lower oxygen aflinity (Table IV). This effect is due in part to amidination of residues in the /3 chain, presumably valine l/3 and lysine 82& since the DPG effect of Hbaj3* is reduced to about 60% that of native hemoglobin. The DPG effect of Hba*@ is reduced to about 75% of control, regardless of whether 29 or 99% of the amino groups of the a! chain are amidinated. The reduction of the ability of DPG to stabilize the deoxy conformation of hemoglobin upon amidination of the (Y chains may be directly due to diminished binding of DPG, or may be an indirect result of destabilization of the deoxy conformation by prevention of the salt bridge involving lysine 40~ The reagent methyl acetimidate exhibits great selectivity toward reaction with the e-amino group of lysine 4Ooa! in the oxy conformation of hemoglobin. Since the majority of the lysyl residues of hemoglobin, including lysine 40a, are exposed on the surface of the molecule, it is difficult to explain this selectivity on steric grounds. It therefore appears likely that the positively charged methyl acetimidate interacts with a carboxylate group on the surface of the molecule which is in the immediate vicinity of lysine 40a, thus facilitating nucleophilic attack on the imidoester by the e-amino group of lysine 4Oa. It is of interest to note that guanidination of about 60% of the amino groups of hemoglobin, presumably using the uncharged reagent 0 methylisourea, resulted in only a slight increase in the oxygen affinity (28). The extent of modification of lysine 40a under these conditions is not known. Since the structure of e-acetamidinolysine is very similar to that of homoarginine, one might predict that neither derivative would be able to form the salt bridge with histidine HC3(146)/3 in deoxyhemoglobin. Thus, comparable effects on oxygen afllnity should result from comparable extents of modification of lysine 40~ The fact that the oxygen affinity of guanidinated hemoglobin is only slightly increased strongly suggests that lysine 4Oa! has not been extensively modified, and emphasizes the selectivity of amidination for this residue.