Selectivity in the Modification of the a-Amino Groups of Hemoglobin on Reductive Alkylation with Aliphatic Carbonyl Compounds INFLUENCE OF DERIVATIZATION ON THE POLYMERIZATION OF HEMOGLOBIN S*

The reactivity of the a-amino groups of the a- and 8-chains of hemoglobn toward reductive alkylation using limiting concentrations of the aliphatic carbonyl compounds, acetaldehyde (ethylation), glyoxylic acid (car-boxymethylation), glycolaldehyde (hydroxyethyla- tion), glyceraldehyde (dihydroxypropylation), and dihydroxyacetone (dihydroxyisopropylation) has been investigated. Hemoglobin A reductively ethylated at the a-amino groups eluted on CM-52 ahead of unmod- ified hemoglobin A, and hemoglobin A reductively ethylated at the e-amino groups. This observation is similar to that seen on hydroxyethylation and dihydroxypropylation of the a-amino group of hemoglobin A. The presence of the a-hydroxyl or the car- boxyl group in the carbonyl component used in the reductive alkylation influences considerably the selec- tivity pattern during the derivatization. The a-amino groups of the a- and 8-chains are modified to nearly the same degree during reductive hydroxyethylation as well as during reductive dihydroxypropylation. Re- ductive ethylation (aldehyde lacking the a-hydroxyl group) exhibited a slight preferential reaction at Val- 1(B). The presence of a negatively charged carboxyl group in for helping in the preparation of Our special amino acid


INFLUENCE OF DERIVATIZATION ON T H E POLYMERIZATION OF HEMOGLOBIN S*
A. Seetharama AcharyaS, Leslie G . Sussman, and James M. Manning From The Rockefeller University, New York, New York I0021 The reactivity of the a-amino groups of the a-and 8chains of hemoglobn toward reductive alkylation using limiting concentrations of the aliphatic carbonyl compounds, acetaldehyde (ethylation), glyoxylic acid (carboxymethylation), glycolaldehyde (hydroxyethylation), glyceraldehyde (dihydroxypropylation), and dihydroxyacetone (dihydroxyisopropylation) has been investigated. Hemoglobin A reductively ethylated at the a-amino groups eluted on CM-52 ahead of unmodified hemoglobin A, and hemoglobin A reductively ethylated at the e-amino groups. This observation is similar to that seen on hydroxyethylation and dihydroxypropylation of the a-amino group of hemoglobin A. The presence of the a-hydroxyl or the carboxyl group in the carbonyl component used in the reductive alkylation influences considerably the selectivity pattern during the derivatization. The a-amino groups of the a-and 8-chains are modified to nearly the same degree during reductive hydroxyethylation as well as during reductive dihydroxypropylation. Reductive ethylation (aldehyde lacking the a-hydroxyl group) exhibited a slight preferential reaction at Val-1(B). The presence of a negatively charged carboxyl group in the carbonyl component, i.e. glyoxylic acid, made this preferential reaction at Val-l(B) even more pronounced. When the reductive alkylation is carried out with dihydroxyacetone (a ketone instead of an aldehyde), the dihydroxyisopropylation occurred at a slower rate and exclusively at Val-l(B).
The ethylation, hydroxyethylation, carboxymethylation, and dihydroxypropylation of the a-amino groups of hemoglobin S increased its solubility from the value of 16 g/dl for the unmodified protein to about 25 g/dl for the modified protein. Thus, the alkyl chains on the a-amino groups on the polymerization have a strong inhibitory influence. In order to determine the influence of the alkyl chains at the a-amino groups of CYand &chains on polymerization, hybrid hemoglobin S tetramers with hydroxyethylation either at Val-l(a) * This work is supported in part by Grant BRSG 507 RR07065 from the Biomedical Research Support Grant, Division of Research Resources, National Institutes of Health to Rockefeller University, National Institutes of Health Grant HL-27183 to A. S. A., and National Institutes of Health Grant HL-18819 to J. M. M. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This manuscript is dedicated to the memory of Dr. Stanford Moore, who encouraged us to pursue the study of the correlation between the nature of the alkyl chain introduced on the a-amino group and the apparent lowering of the pK. of the a-amino group on reductive alkylation.
$ Established Fellow of the New York Heart Association.
or at Val-l(@ have been prepared. The solubility of each hybrid is about 26 g/dl. Thus, the hydroxyethyl group either on the a-or the &chain appears to interfere with the polymerization of deoxygenated HbS to the same degree. The inhibitory influence of the hydroxyethyl chain at Val-l(a) on the polymerization, compared with the lack of such an influence when this a-amino group is modified by cyanate, suggests that a carbamoyl group on Val-1(a) can be accommodated in the intermolecular contact region involving this segment of the molecule without seriously perturbing the molecular fit of this contact region, whereas the hydroxyethyl group cannot be accommodated easily at this site, and hence the inhibition of polymerization.
The present work grew out of our attempts to identify the functional groups of hemoglobin S that are present at the intermolecular contact regions and accessible to chemical manipulation. The ultimate objective of these studies is to develop an antisickling agent of therapeutic value targeted to one or more critical amino acid residues at the intermolecular contact regions of deoxy-HbS.' One of the early approaches to inhibit the sickling of the erythrocytes from patients with sickle cell disease was the use of cyanate (Cerami and Manning, 1971). Detailed chemical studies of HbS reacted with cyanate revealed that the major contribution for the inhibition of sickling of erythrocytes is a result of carbamoylation of the a-amino group of a-chain. This derivatization increases the oxygen affinity of HbS, thus perturbing the oxy-HbS * deoxy-HbS equilibrium, to decrease the concentration of the deoxy-HbS at a given O2 tension. This in turn influences the polymerization of HbS (Nigen et al., 1974). However, this derivatization of HbS at its a-amino group of achain has very little direct influence on the polymerization of deoxy-HbS (i.e. it does not increase the minimum gelling concentration). On the other hand, the selective carbamoylation of the a-amino group of the /3-chain of HbS slightly increased the minimum gelling concentration of the protein, i e . decreased the propensity of the deoxy-HbS to polymerize (Nigen et al., 1974). In contrast to the lack of inhibitory influence of carbamoylation of the a-amino group of a-chain on the polymerization of deoxy-HbS, Benesch et al. (1974) have shown that selective blocking of the a-amino group of ~ The abbreviations used are: deoxy-HbS, deoxygenated hemoglobin S; HbS, hemoglobin S; HbA, hemoglobin A; oxy-HbS, oxygenated hemoglobin S; E, ethyl; HE, hydroxyethyl; Cm, carboxymethyl; DHP, 2,3-dihydroxypropyl; DHIP, 1,3-dihydroxyisopropyl; HPLC, high performance liquid chromatography; HEPES, 4-(2-hydroxyethyl)-lpiperazineethanesulfonic acid; Ac, acetyl; PBS, phosphate-buffered saline; HMB, hydroxymercuribenzoate.

6039
the a-chain of the protein by pyridoxal sulfate caused a marked increase in the minimum gelling concentration. These results suggest that the influence of substitution of a-amino group on gelation may be strongly dependent on the nature of the blocking groups. In view of the significant difference in the size of the blocking groups (carbamoyl group as opposed to pyridoxal sulfate), it was of interest to introduce small alkyl chains on the a-amino group of HbS to determine whether the inhibitory influence is dependent on the size or the nature of the substituting group introduced. In the present study we have investigated the influence of ethyl, hydroxyethyl, carboxymethyl, dihydroxypropyl, and dihydroxyisopropyl groups at the a-amino groups of HbS on the polymerization behavior of the protein.
The possibility of using reductive alkylation of hemoglobin with aliphatic aldehydes to introduce the alkyl chains on the a-amino groups has been now explored. We have previously shown that reductive alkylation of carbonmonoxyhemoglobin A at pH 7.4, 37 "C with a limiting concentration of glyceraldehyde (Acharya et al., 1983a) resulted in the selective dihydroxypropylation of the a-amino groups of carbonmonoxy-HbA. Reductive alkylation has been subsequently investigated with glycolaldehyde (hydroxyethylation) and the selectivity appears to be nearly the same as that with reductive dihydroxypropylation . Thus, this approach of condensation of a carbonyl compound with the a-amino groups of HbA in the presence of sodium cyanoborohydride appears to provide a general procedure to derivatize the a-amino groups of the protein with an alkyl group of choice by appropriate selection of the carbonyl compounds. By employing this procedure in the present study, we have introduced ethyl (acetaldehyde), hydroxyethyl (glycolaldehyde), carboxymethyl (glyoxylic acid), 2,3-dihydroxypropyl (glyceraldehyde), and 1,3-dihydroxyisopropyl (dihydroxyacetone) groups at the amino terminus of HbS to study their influence on gelation.* By changing the carbonyl component from acetaldehyde to glycolaldehyde to glyoxylic acid in the reductive alkylation procedure (Fig. l), the property of the substituted alkyl group changes from being relatively hydrophobic (ethyl groups) to neutral hydrophilic (hydroxyethyl) to negatively charged hydrophilic (carboxymethyl groups). With glyceraldehyde and dihydroxyacetone, bulkier, neutral hydrophilic 3-carbon alkyl chains are introduced.

MATERIALS AND METHODS
Erythrocytes from normal adults were isolated by centrifugation and washed with carbon monoxide-saturated phosphate-buffered saline (PBS), pH 7.4. Washed cells were lysed, dialyzed first against PBS, and then dialyzed overnight against 50 mM Tris-Ac, pH 8.5. HbA was isolated by chromatographing the dialyzed lysate on DE52 as described earlier (Acharya and Manning, 1980a).
Whole blood from patients homozygous for sickle cell anemia was collected into heparinized tubes by venipuncture. The erythrocytes were isolated by centrifugation and washed with phosphate-buffered saline (pH 7.4), lysed, and then dialyzed extensively against phosphate-buffered saline.
Reductiue Alkylation of HbA-Purified HbA (1 mM) in the carbonmonoxy form was dialyzed against phosphate-buffered saline (pH 7.4) and treated with 10 mM aliphatic aldehyde (acetaldehyde, glycolaldehyde, glyoxylic acid, glyceraldehyde, or dihydroxyacetone) in the presence of 20 mM sodium cyanoborohydride (NaCNB3H3) at 37 "C for 30 min. After the incubation period, the excess reagents were removed by gel filtration on a Sephadex G-25 column, equilibrated, and eluted with 10 mM phosphate buffer, pH 6.0, or 50 mM Tris-Ac, pH 8.5.
The preparation of p-hydroxymercuribenzoate a-and 8-chains of * A preliminary account of this work has been presented . derivatized HbA (Bucci and Fronticelli, 1965;Acharya and Manning, 1980a), the analysis of the tryptic peptides by reverse phase HPLC (Acharya et al., 1983b), and the amino acid analysis were carried out as described earlier (Acharya et al., 1983a). Reductiue Alkylation of HbS-HbS (dialyzed lysate) was reacted at a concentration of 1 mM (tetramer) with aldehydes (10 mM) as described above for HbA, except for the fact that the reaction was carried out in the oxy form of HbS. The excess reagents were removed by gel filtration on a Sephadex G-25 column equilibrated and eluted with phosphate buffer, pH 6.9.
Oxygen Equilibrium Measurement-The oxygenation curves of HbS and reductively alkylated HbS were recorded at 37 "C using an Aminco Hem-0-Scan, as described by Benesch et al. (1978). Polymerization of HbS-The concentration at which the onset of polymerization both of HbS and of the reductively alkylated HbS occurs was determined by the method of Benesch et al. (1978) as described earlier (Acharya et al., 1984, andSeetharam et d , 1983).

Reductive Ethylation of HbA-Incubation of hemoglobin A
(1 mM in tetramer) with 10 mM acetaldehyde and 20 mM NaCNB3H3 for 30 min at 37 "C resulted in the incorporation of 3H label into the protein, suggesting the reductive alkylation of the protein. From the amount of the label incorporated into the protein, it is calculated that nearly four to five alkyl (ethyl) groups are introduced into the protein. Amino acid analysis of an acid hydrolysate of the labeled ethylated HbA showed that nearly 30% of the label eluted at the position corresponding to t-N-ethyl lysine (Means and Feeney, 1968), slightly after the position of unmodified lysine on the short column of a Moore-Stein amino acid analyzer. The remainder of the label eluted near the void volume of this column and apparently is a-ethyl valine. The modification of a-amino groups is also confirmed by the tryptic peptide analysis (see below).
Reductive alkylation of the amino groups of protein in general does not significantly influence the net charge of the protein, since the derivatized amino grou still retains its original positive charge (Jentoft and Dear orn, 1979). Consistent with this observation, when HbA reductively ethylated, was chromatographed on DE52 around pH 8.0, the modified protein eluted at the position corresponding to that of unmodified protein. The reductively ethylated HbA has been chromatographed on CM-52 and the chromatographic behavior ( Fig. 2) has been compared with that of reductively hydroxyethylated HbA. It is clear that the chromatographic behavior of the ethylated HbA ( Fig. 2 A ) is very similar to that of hydroxyethylated HbA (Fig. 2B). In the case of the hydroxyethylated protein (as well as dihydroxypropylated HbA), we have shown previously that the two components eluting earlier than unmodified HbA (peaks A and B) are the derivatives modified at a-amino groups. The component eluting around 240-280 ml (peak A) is a derivative of HbA in which all four a-amino groups of the tetramer are modified, whereas the component eluting around 280-320 ml (peak B) is a derivative in which, on an average, two of the four aamino groups of the tetramer are modified (Fig. 2B). The similarity in the chromatographic behavior of ethylated HbA and hydroxyethylated HbA suggests that the component elut- ing around 240-280 ml (peak A) in the ethylated protein is atetraethylated HbA, while the one eluting around 280-320 ml (peak B) is a-diethylated HbA. Consistent with this suggestion is the finding that both of these fractions contained aethyl valine. Furthermore, the a-ethyl valine content in fraction A (-4 mol/tetramer) is nearly twice of that in fraction B (-2 mol/tetramer). All the counts in fraction C were associated with t-N-ethyl lysine. From these results it is clear that ethylation of a-amino groups of HbA results in an earlier elution of the protein on CM-52 compared to that of the unmodified protein, as seen earlier with the hydroxyethylation studies. Peak A from ethylated HbA (Fig. 2B) appears to be the tetramer in which all four amino groups are ethylated. Hence, analysis of the HMB chains of peak B should provide the information about the relative reactivity of the a-amino groups of a-and P-chains of HbA. Chromatography of the HMB chains of peak B of ethylated HbA on CM-52 is shown in Fig. 3. The chromatogram shows three peaks; peak B1 is the P-chain whereas peaks B2 and B3 are a-chains. Amino acid analysis of these fractions showed that B1 and B2 contain a-ethyl valine, while B3 did not contain the a-N-ethyl valine.
Tryptic peptide mapping of B1 and B2 confirmed the derivatization of the a-amino group of 8and a-chains, respectively (data not shown). Nearly 30% of the modification of B1 was present at sites other than the a-amino groups (apparently tamino groups, whereas with BP almost all the modification was associated with the a-amino group. All the 3H counts of B3 and is associated with t-ethyl lysine. The chromatographic position of a-N-ethyl a-chain is close to that of a-hydroxyethyl a-chain and ahead of that of the unmodified a-chain and a-chain modified at the a-amino group. This behavior is apparently a reflection of the lowering of the pK, of the aamino group as a result of ethylation. This influence is similar  Fig.   2 A ) was treated with p-hydroxymercuribenzoate as described earlier (Acharya and Manning, 1980a), dialyzed extensively against 10 mM potassium phosphate buffer, pH 5.85, containing 1 mM EDTA. The dialyzed sample was loaded on to a CM-cellulose column (0.9 X 30 mm) equilibrated to pH 5.85 with potassium phosphate buffer, pH 5.85. The HMB chains were eluted with a linear gradient from 150 ml each of 10 mM potassium phosphate buffer, pH 5.85 (1 mM in EDTA) and 15 mM potassium phosphate buffer, pH 7.6 (1 mM EDTA). The column is operated at 4 "C and all the buffers were saturated with CO.

Reductive Alkylation
of Hemoglobin S to that observed earlier ) on reductive hydroxyethylation of a-amino groups of HbA.
The amount of a-ethyl a-chain in this chromatogram (Fig.  3) accounts for about 35% of the a-chain present in fraction B. Since the rest of the a-chains are either unmodified or modified only at t-amino groups, these did not contribute to chromatographic behavior of the tetramer (fractions from which these chains are prepared), i.e. to an early elution on CM-52 (compared with unmodified HbA). Therefore, the elution of these chains in fraction B (HbA ethylated at two of the four a-amino groups) is apparently due to their hybridization with @-chains ethylated at the a-amino groups. Thus, the distribution of ethylation at the a-amino groups of fraction B is that 65% of the @-chains are modified, while only 35% of the a-chains are modified. Thus, studies demonstrate that the a-amino group of @-chain shows a somewhat higher selectivity for reductive ethylation compared to the a-amino groups of a-chain.
Reductive Carboxymethylation of HbS-The reductive carboxymethylation would introduce charge difference due to the negative charge (-COO-) incorporated into the protein. In view of the negative charge of the glyoxylic acid (the carbonyl component) compared with its absence in acetaldehyde and glycolaldehyde, it is of interest to determine whether the carboxylate of glyoxylic acid influences the selectivity pattern of a-amino groups of Hb for reductive alkylation. Previously we have shown that reductive carboxymethylation of the HbA is reasonably selective towards the a-amino groups of HbA (Acharya et al., 1982. The earlier studies on carboxymethylation were, however, carried out at pH 7.2 using HEPES buffer and without chloride for 40 min. Besides, a larger excess (lo-fold over the aldehyde) of sodium cyanoborohydride had been used, as compared with a 2-fold molar excess (over aldehyde) used in the studies with ethylation, hydroxyethylation, and dihydroxypropylation. Therefore, carboxymethylation of HbS has been now carried out in PBS at pH 7.4 and 37 "C for 30 min using 10 mM glyoxylic acid and 20 mM NaCNB3H3 so that the reactivity of the aamino groups could be compared in a systematic way.
On DE52 chromatography of the carboxymethylated HbS (lysate), three chromatographically distinct components were obtained (Fig. 4). The minor component eluting around 450 ml is apparently the unmodified HbS.3 The two major radioactive components are the carboxymethylated derivatives. These are designated HbS-Cml and HbS-Cm2 and were isolated. Thus, the presence of the carboxylate group in the carbonyl component has made the alkylated protein more acidic and hence more retarded on the DE52 columns. Reductive alkylation with the neutral carbonyl compounds (acetaldehyde and glycolaldehyde) does not influence the chromatographic behavior of HbS on DE52.
The HbS-Cml and HbS-Cm2 contain two and four carboxymethyl groups, respectively, hence appearing to correspond to Hbl and Hb, previously isolated from HbA reductively carboxymethylated using a greater excess of sodium cyanoborohydride over the aldehyde .
The main difference appears to be in the amount and the distribution of carboxymethylated derivatives. The component corresponding to Hb3 (containing both a-N-carboxymethylated valine and t-N-carboxymethyl lysine) was absent in the present study. Besides HbS-Cml (corresponding to HbJ, the dicarboxymethylated HbS was the major carboxymethylated product in the present study, whereas Hb2 The exact chemistry for the incorporation of tritium label into the HbS peak is not clear. It is conceivable that glycosylated HbS is reduced by NaCNB3H3 (Acharya and Sussman, 1984). Carbonmonoxy HhS (1 mM) was treated with 10 mM glyoxylic acid in the presence of 20 mM sodium cyanoborohydride (NaCNB3H3). The reaction mixture was then passed through a column (2.2 X 45 cm) of Sephadex G-25 equilibrated and eluted with 50 mM Tris-Ac, pH 8.5, to separate the carboxymethylated protein from the excess reagents. The fractions containing the protein were pooled, concentrated, and applied to a column (2.2 X 40 cm) of DE52 equilibrated with 50 mM Tris-Ac buffer, pH 8.5. The protein is eluted with a linear pH gradient of 500 ml each of 50 mM Tris-Ac buffer, pH 8.3, and 50 mM Tris-Ac buffer, pH 7.3. The position of HbS is indicated. The derivatized HbS were pooled as indicated and designated HbS-Cml and HbS-Cm2, respectively.
(tetracarboxymethylated HbA) was the major reaction product with HbA (DiDonato et al., 1983). This apparently reflects the influence of using phosphate buffer with NaCl and lower amounts of NaCNB3H3 (over the aldehyde) that was used in the present study.
The separation of the a-and @-chains of HbS-Cml and HbS-Cm2 are shown in Fig. 5. HbS-Cml had almost all of the modifications on the @-chains and had very little carboxymethylated a-chain (Fig. 5A). HbS-Cm2, on the other hand, has carboxymethylation on the a-amino groups of a as well as that of the @-chains (Fig. 5B). The a-Cm a-chain eluted at a position much ahead of a-HE a-chain, demonstrating the contribution of the carboxyl group toward the separation at p H values above 7.0. Some amounts of unmodified a-chain was present in HbS-Cm2. The presence of unmodified achains in this tetracarboxymethylated tetramer is apparently a reflection of some carboxymethylation on the t-amino groups (of the @-chains) of the protein.
The peptide mapping of the derivatized a-and @-chains of HbS-Cml and HbS-Cm2 showed that the modification is at Val-l(a) and Val-l(@), respectively. HbS-Cml is the derivative of HbS with modifications only at Val(@), while HbS-Cm2 contains modifications at Val-l(a) and Val-l(@). A small amount of carboxymethylation on the t-amino group of the @-chain is indicated by peptide mapping. The presence of small amounts of unmodified a-chain (Fig. 5B) is consistent with this observation.
The detailed chromatographic analysis of HbS, reductively carboxymethylated, and the peptide mapping of the isolated chains demonstrate that, even in the presence of chloride, the major sites of reductive carboxymethylation are the a-amino groups of the protein when a limiting concentration of glyoxylate is used. HbS-Cml is the major carboxymethylated derivative formed under the conditions studied. Thus, a-CM-52 columns. Two chromatographically distinct components were detected. One eluted at the position of the unmodified HbS and had no radioactivity in it. This component accounted for about 75 to 80% of the protein. The second minor component was radioactive and accounted for about 20% of the protein. This material eluted slightly ahead of the position of HbS and corresponded to the position of a2@$(a-HE). The results demonstrate that, when HbS is subjected to reductive alklyation with dihydroxyacetone, about 20% of its @-chain gets modified at its a-amino group in 30 min. This dihydroxyisopropylation of the a-amino group of the @-chain results in an early elution of the tetramer as has been observed with ethylation, hydroxyethylation, and dihydroxypropylation.
Thus, the propensity of dihydroxyacetone, a ketone, to form Schiff base adducts with the a-amino groups of Hb at pH 7.4 appears to be significantly lower than that of other aliphatic aldehydes studied. This possibly results in the higher selectivity of dihydroxyacetone towards the a-amino group of the @chain (see "Discussion").

Influence of Reductive Alkylation on the Polymerization of
HbS-The influence of the selective introduction of charged carboxymethyl groups, uncharged hydroxyethyl groups, or weakly hydrophobic ethyl groups at the a-amino groups of HbS on the polymerization of deoxy-HbS has been investigated (Fig. 6). Ethylation and hydroxyethylation increased the O2 affinity of HbS; on the other hand, carboxymethylation decreased the O2 affinity. However, these modifications had very little influence on the cooperativity. The Hill coefficient of the modified proteins is nearly the same as that of unmodified protein (-2.5).
The amino groups of @-chains appear to have reacted significantly to a higher degree than the a-amino groups of the a-chain. Thus, the a-amino group of the a-chain is more reactive towards hydroxyethylation than towards the reductive carboxymethylation reaction. Apparently, the presence of a carboxyl group in the alkyl chain decreases the propensity of the glyoxylate to form a Schiff base adduct at the a-amino group of the a-chain (compared to that with glycolaldehyde) in liganded Hb.
Reductive Dihydroxyisopropylation of HbS-The generality of the high reactivity of a-amino groups of HbS towards carbonyl compounds in the presence of a reducing agent has also been investigated with dihydroxyacetone to determine whether aliphatic ketones also show selectivity similar to that seen with the aliphatic aldehydes studied. Incubation of HbS with 10 mM dihydroxyacetone in the presence of 20 mM NaCNB3H3 resulted in the incorporation of the tritium label into the protein. Separation of HMB chains of modified HbS showed that almost all of the radioactivity is associated with the @-chain. Analysis of the tryptic peptides of the modified @-chain by reverse phase HPLC demonstrated that all the radioactivity is associated with the @-TI, clearly showing that the dihydroxyisopropylation has taken place exclusively at the a-amino group of the p-chain. Modified B-chain was hybridized with unmodified a-chain and chromatographed on concentration of the protein needed for the onset of polymerization from the control value of 16 to about 24 g/dl for the hydroxyethylated protein, and to about 26 g/dl for the eth-  . 6. Relation between oxygen affinity and hemoglobin concentraton of W S and HbS reductively alkylated with 2carbon aldehydes. P w was obtained from the complete oxygenation curve recorded using an Aminco Hem-0-Scan at 37 "C and pH 6.9 at each concentration shown. A, HbS; 0, ethylated HbS; 0, hydroxyethylated HbS; and A, carboxymethylated HhS. Alkylation of Hemoglobin S droxyethyl, or ethyl groups on the a-amino groups appears to be nearly the same.
The O2 affinity (at pH 6.9) for the ethylated and/or the hydroxyethylated HbS is nearly the same and is higher than that of unmodified HbS. The influence of ethylation and hydroxyethylation on the oxygen affinity of HbS is nearly the same. This reflects a similarity in the structural perturbations that these two modifications bring about in the molecule. On the other hand, with carboxymethylation of the a-amino groups, the Pm of the molecule is increased . Thus, the propensity of the carboxyl group of the alkyl chains is to reduce the O2 affinity of the protein. Nonetheless, this modification still increases the concentration of HbS that is needed for the onset of polymerization to nearly the same point. The nature of alkyl groups does not appear to be crucial for the inhibition of the polymerization. The substitution of the a-amino group by alkylation is sufficient to cause this inhibition of polymerization. The influence of reductive 2,3-dihydroxypropylation of the a-amino groups of HbS, as well as that of reductive 1,3dihydroxyisopropylation of the a-amino group of the @-chain of HbS, is shown in Fig. 7. Nearly the same influence of reductive hydroxyethylation and reductive dihydroxypropylation on the polymerization of deoxy-HbS demonstrates  that the increase in the chain length of the alkyl group from 2 to 3 carbons has limited additional inhibitory influence on the propensity of the molecule to polymerize. The lower influence of dihydroxyisopropylation on the solubility of HbS, seen in the present study, is due to lower levels of derivatization (0.8 mol/tetramer as opposed to nearly 3.0 mol/tetramer for hydroxyethylation, ethylation, and dihydroxypropylation).
Polymerization of HbS Selectively Hydroxyethyluted at the a-Amino Group of Either a-or f-Chains-During the reductive alkylation of HbS, a-amino groups of both the a-and @chains are derivatized. Therefore, it was of interest to determine whether the alkyl group present either on the a-amino group of the a-chain or the one on the a-amino group of the @-chain has contributed to the inhibition observed on reductive alkylation. For a detailed study, we have examined the reductive hydroxyethylation of HbS. HMB a-and @-chains of reductively hydroxyethylated HbS were prepared and mixed with HMB @-and a-chains, respectively, in the presence of @-mercaptoethanol (the procedures are similar to those used earlier for the preparation of hybrids from hydroxyethylated HbA . The hybrids a2@2a(n-HE)and a2(a-HE)@2~ were purified by CM-cellulose chromatography and used for the polymerization studies. The O2 affinity of the hybrid HbS prepared was slightly higher than that of the native HbS. The Hill coefficient of the hybrids is about 2.6, nearly the same as that of HbS. The polymerization studies (Fig. 8) have shown that hydroxyethylation of the aamino group of either the a-or @-chain increases the concentrations of HbS at which the onset of polymerization occurs to about 26 g/dl, from a control value of 14.7 g/dl for the purified HbS. Thus, it is clear that the hydroxyethylation of the a-amino group of a-chain inhibits polymerization, as has been observed earlier on the modification of a-amino groups of a-chain by pyridoxal sulfate (Benesch et al., 1974). This influence is distinct from that seen on carbamoylation of the a-amino group of a-chain of HbS.

DISCUSSION
The studies described here were undertaken with two major objectives. The first one was to determine whether the condensation of the a-amino groups of Hb around neutral pH using limiting concentration of the aliphatic aldehydes or ketones in the presence of sodium cyanoborohydride would provide a general route for derivatizing the a-amino groups of Hb. The second objective was to determine whether reductive alkylation of the a-amino group influences the polymerization properties of HbS and, if it did, to determine whether the nature of the alkyl group introduced has any strong influence on the inhibition of polymerization.
The results of the present study clearly demonstrate that around pH 7.4 the a-amino groups of HbA show a high selectivity to form Schiff bases when a limiting concentration of the aliphatic aldehyde was used. All four aldehydes used resulted in the modification of the a-amino groups of Hb. Glycolaldehyde and glyceraldehyde modified the a-amino group of both chains to nearly the same degree Sussman, 1983, Acharya et al., 1983a). Although acetaldehyde reacted with the a-amino groups of both the chains, it showed some preferential reaction at the Val-l(@). With glyoxylic acid, the preferential reactivity of Val-l(@) is even more pronounced. The significantly higher reactivity of Val-l(@) with glyoxylic acid is apparently a reflection of the refractory influence of the environment of Val-l(a) to accommodate the negatively charged alkyl chain of glyoxylic acid when these two aldehydes form the Schiff bases at these two sites. Prelim-inary studies with glyceraldehyde 3-phosphate also have indicated that the anionic charge of this aldehyde contributes a higher selectivity for this reagent to Val-1(/3) as opposed to glyceraldehyde? This would imply that if lower levels of aldehydes (over the a-amino groups) are used in the reductive alkylation, a selective modification of Val-l(P) would occur. Reductive alkylation using ketones which have a very low propensity to form Schiff bases at neutral pH could be considered as equivalent to using lower levels of any of these aldehydes. Dihydroxyisopropylation, which proceeds at a much slower rate, indeed showed nearly exclusive reaction at Val-l(B).
The anionic nature of the glyoxylic acid significantly influences the relative selectivity of the amino groups of the two chains of Hb. Besides, significant differences are also seen in the composition of the Hb species containing alkyl chains on two of its four a-amino groups that are isolated after reductive hydroxyethylation (or ethylation) and carboxymethylation. On modification with the uncharged aldehydes (acetaldehyde, glycolaldehyde, and glyceraldehyde), the Hb species, containing two of its four a-amino groups modified (material eluting at the position corresponding to that of peak B in Fig. 2) had the derivatizations on the a-amino groups of both a-and bchains. Thus, the species chromatographing in this position could be a mixture of two different symmetrical forms of tetramer, one form containing reductive alkylation on the aamino groups of Val-l(@) and the other on that of Val-l(a) of a-chains. Alternatively, the reductive alkylation could be assymmetrical. Apparently these symmetrical and/or assymmetrical forms do not readily segregate to form native Hb (unmodified) and Hb species with derivatization on all four a-amino groups. Thus, the reductive alkylation with uncharged aldehydes (i.e. ethylation, hydroxyethylation, and dihydroxypropylation) does not significantly perturb (destabilize) the intersubunit interactions.
In contrast, the tetramers of Hb containing two carboxymethylated a-amino groups (HbS-Cml) are species with modifications of Val-l(B). This result suggests that the symmetrical and/or assymmetrical carboxymethylated Hb formed during carboxymethylation appears to segregate readily to the native and tetracarboxymethylated derivatives. Apparently the carboxymethyl groups on the a-amino group destabilize the subunit interactions, permitting their ready segregation to native and tetracarboxymethylated derivative.
Substitution of the a-amino groups of HbS by ethylation, hydroxyethylation, carboxymethylation, and dihydroxypropylation had nearly the same influence of inhibiting the polymerization of deoxy-HbS. The ethyl group is weakly hydrophobic, hydroxyethyl and dihydroxypropyl groups are hydrophilic, and the carboxymethyl group is negatively charged and also hydrophilic. The nature of these groups is also readily reflected in the elution behavior of t-N-ethyl lysine, c-N-hydroxyethyl lysine, t-N-dihydroxypropyl lysine, Manning, 1983, Geoghegan et al., 1979;Acharya et al., 1984a), and the t-Ncarboxymethyl lysine on the amino acid analyzer. It is conceivable that these alkyl groups, when present at the aminoterminal, induce similar changes around the substituted amino group. The dihydroxypropyl group is bulkier than the hydroxyethyl groups introduced. However, in spite of these differences in their hydrophilicity or size, the alkylation of the a-amino group by these derivatizations appears to influence the polymerization to nearly the same degree. This would imply that the groups introduced perturbed the stereochemi-A. S. Acharya, unpublished results. cal orientation of the HbS tetramer during the polymerization around the contact regions involving Val-l(a) and/or Val-1(b), respectively. Since the reductive alkylation by acetaldehyde, glycolaldehyde, glyoxylic acid, or glyceraldehyde has nearly the same inhibitory influence, the inhibition of polymerization is apparently related to alkylation of a-amino groups rather than the charge or hydrophobicity of the alkyl chain introduced. The studies have also shown that the influence of reductive hydroxyethylation and reductive dihydroxypropylation on polymerization of HbS is nearly the same. Thus, the small changes in size of the alkyl group also appear to have little influence on the observed inhibition.
The above conclusion appears to be reminiscent of the conclusions that we had drawn earlier with regard to another intermolecular contact region of HbS, i.e. one involving Lys-16(a) (Acharya et al., 1984b). The mutation of Lys-lG(a) to Glu there by replacing the positive charge at the intermolecular contact region with a negative charge inhibits the polymerization (Benesch et al., 1977). Dihydroxypropylation of the e-amino group of Lys-lG(a), which would retain the original positive charge of this amino group under the physiological condition, also results in the inhibition of polymerization. These results, as well as the results of the present study, could be interpreted as suggesting that perturbation of the intermolecular contact region by the modification, which would make the proper "fit" of the region during polymerization difficult, rather than the changes in the charge distribution at a particular intermolecular contact region is possibly the general molecular basis for most of the polymerization inhibitors. Such a conclusion is simply a reflection of the fact that intermolecular contact involving Val-6(8) provides the dominant "force" for the polymerization process. This is also consistent with the fact that none of the chemical modification done so far on HbS has been able to completely neutralize the influence of Val-G(P), i.e. to restore a solubility to HbS similar to that of HbA.
The modification of the a-amino group of Val-1(a) as well as that of Val-l(p) by reductive hydroxyethylation results in reducing the propensity of HbS to polymerize. The influence of reductive hydroxyethylation of Val-l(a) is consistent with the influence of modification of this amino group by pyridoxal sulfate. This influence is distinctly different than that observed with the carbamoylation of this amino group. On the other hand, the influence of modification of the a-amino groups of Val-l(a) on the oxygen affinity is consistent with that seen on carbamoylation of this amino group (Nigen et al., 1974). As pointed out earlier, there is a close similarity in the influence of reductive hydroxyethylation and carbamoylation of a-amino groups on the oxygen affinity of HbA . This is probably related to the apparent loss of nearly one positive charge under the physiological condition that occurs as a result of both hydroxyethylation and carbamoylation of the a-amino group. With hydroxyethylation, this is apparently due to the decrease in the apparent pK, of the a-amino group when it is modified to a secondary amino group. Carbamoylation of the a-amino group results in the loss of the positive charge of this amino group. The fact that the carbamoylation of Val-l(a) has no influence on the propensity of the molecule to polymerize, whereas the reductive hydroxyethylation increases the solubility of deoxy-HbS, possibly suggests the difference in orientation of the carbamoyl group, and the hydroxyethyl groups from the nitrogen atom of the amino group of Val-l(a). The carbamoyl group could probably be accommodated in the intermolecular contact regions involving Val-l(a) without seriously perturbing the molecular fit of this region during polymerization.
The carbamoyl group on the Val-(/?) apparently perturbs the intermolecular contact region(s) involving this residue as in the case with reductive alkylation. Therefore, it would be of interest to study the influence of the substitution at Val-1(a) by a methyl group on the polymerization of deoxy-HbS to determine whether this group could be accommodated at the intermolecular contact region without significantly perturbing the molecular fit of this region.
We have previously shown that, on incubation of HbS with glyceraldehyde, the a-amino group of Val-l(@) and the tamino groups of Lys-lG(a), Lys-82(@), Lys-59(/3), and Lys-120(@) are modified (Acharya and Manning, 1980a and b). The derivatization of the a-amino group of Lys-lG(a) has been shown to inhibit the polymerization (Acharya et al., 1984b). The results of the present study suggest that the derivatization of Val-l(/?) by glyceraldehyde should also contribute, at least to some extent, toward the inhibition of polymerization of deoxy-HbS when the erythrocytes from sickle cell patients are treated with glyceraldehyde. Thus, the inhibition of polymerization of HbS by glyceraldehyde is a reflection of the derivatization of at least two intermolecular contact regions. Therefore, it will be of interest to prepare a doubly modified HbS, i.e. a derivative modified at Val-l(P) and also at the t-amino group of Lys-lG(a), to determine the additivity of the perturbation of these two intermolecular contact regions to inhibit the polymerization of deoxy-HbS.