Titration behavior of histidines in human, horse, and bovine hemoglobins.

Abstract The titration curves of the human hemoglobins A2 and F, the slow and fast components of horse hemoglobin, and bovine hemoglobin B have been analyzed, mainly to obtain information about the titration behavior of the histidines in these hemoglobins. The results were also compared with earlier data concerning human hemoglobin A and bovine hemoglobin A. In both hemoglobin A2 and F, 18 titratable histidines were found, two less than in hemoglobin A. Assuming that those histidyl residues which in the three-dimensional model occupy the same sites will show similar titration behavior, we reached the conclusion that in human hemoglobin histidine G19 and histidine H21 are titratable and that histidine G18 is not titratable. Both the slow and fast components of horse hemoglobin were found to contain 22 titratable histidines and 70 titratable carboxyl groups. Combining these results with the known number of amides in the α chain we calculated that the β chain should contain 11 amides. It appeared that in bovine hemoglobin B the histidines at B1(18)β, which are absent in bovine hemoglobin A, are titratable. The pK of these histidines, as estimated from the difference titration curve, is about 7.8, both in oxy- and deoxyhemoglobin. This high pK is probably caused by the formation of a saltbridge with the carboxyl group of aspartic acid B3(20)β in the same chain. Both hemoglobins contain 72 titratable carboxyl groups. Comparing the human, horse, and bovine hemoglobins, assuming again similar titration behavior for structurally identical histidines, we were able to correlate the titration results of human and horse hemoglobin. The human and bovine hemoglobins failed to show an equally good correlation, although the discrepancy was small.


It has been established b y Steinhardt and Zaiser (1), Steinhardt and H irem ath
, and Geddes and Steinhardt (3) th a t at acid pH groups are becom ing titratable in hemoglobin which are masked when th e protein is in its native state. In the case of horse carboxyhemoglobin it was found (3) th at out of a to ta l of 24 groups liberated, a t least 12 were im idazole groups. An estim ation of the number of groups unm asked a t acid pH is com phcated b y the fact th at the electrostatic interaction Factor w changes sim ultaneously with unmasking. Therefore to find the number of titratable histidines in hemoglobin we have followed a different approach which consists of a direct mcasurcm ent or counting of the total number of histidines whieh are titrated in the neutral pH region where the protein is in its na tive state. A direct counting procedure, although different from ours, was also followed b y Tanford and N ozaki (4) for human and horse hemoglobin and by Bucci et al. (5) for hum an hemo globin and its subunits. T he accuracy of their analyzing pro cedure w as about one-half histidine per hem e or tw o groups per tetram er. In earlier reports on bovine and hum an hemoglobin (6, 7) we have introduced a measuring and analyzing procedure for titration curves which permits a greater accuracy in the de term ination of the number of titratable histidines. In this paper we have extended our study to human hem oglobins A 2 and F, horse hemoglobin, including the slow and fast com ponents, and bovine hem oglobin B.
In our analyses of the several hemoglobins we also present an efiort to correlate the data of the several hem oglobins in order to get information about the titratability and pK values of particular histidines. e x p e r i m e n t a l p r o c e d u r e Hem oglobin solutions were prepared following the toluene m ethod of D rabkin (8).
T he tw o components of horse hemoglobin were isolated as de scribed b y Perutz et al. (9). B y this m ethod the so-called slow and fast com ponents are obtained as m ethem oglobin. T he two fractions were concentrated b y means of a D iaflo ultrafiltration cell and subsequently converted to th e cyanom et derivatives. T he separation was checked with polyacrylam ide gel electrophoresis a t pH 8.
T h e m ethod of B em in i (10) for the isolation of hum an hemo globin A 2 w as som ewhat modified in order to obtain larger quantities. As buffers were used 0.2 m Tris to w hich 0.2 m N a H 2PC>4 was added until the pH was 8.5, and a solution of 0.035 m Tris adjusted to pH 8.5 w ith 0.035 M N a H 2PC>4. Chromatography was performed on D EA E -Sephadex ty p e A-50. T en grams of this resin were allo wed to swell in the first buffer for 12 hours and subsequently washed several times w ith the second buffer. Before operating the column (4 X 20 cm) was equilibrated with the second buffer. T he colum n w as charged w ith 40 m l of a hemoglobin solution, concentration about 100 g per liter, previously dialyzed against the second buffer. Subsequently the colum n was eluted w ith th is buffer. T he hemoglobin A2 m oved rapidly through the colum n and could satisfactorily be separated from the main com ponent. T he solution containing hemoglobin A 2 was concentrated b y means of ultrafiltration and dialyzed against distilled water. T he separation was checked b y polyacrylam ide gel electrophoresis a t pH 8. T he whole procedure was carried out at 4°. To one sam ple of hem oglobin A2, N a 2S20.i was added and the solution im m ediately deionized (11) in order to reduce ariy methem oglobin w hich m ight have been formed during the whole procedure. T he titration curve obtained from this sample was not significantly different from that of an untreated sample.
B ovine hemoglobin B w as obtained from a B B genotype of the Limousine breed.
Hem oglobin isolated from um bilical cord blood was used for the hemoglobin F studies. T he am ount of hemoglobin F was determ ined b y th e alkali denaturation te st (12).
Before titration all hem oglobin solutions were dialyzed against distilled water and subsequently deionized b y repeated passing through a mixed bed ion exchange column.
T he concentrations of the hem oglobin solutions were deter mined b y drying to constant w eight a t 105°. T he titrations were performed a t 25° at an ionic strength of 0.1 (KC1) as de scribed earlier (7, 13).
A ll reagents were an alytical grade.  ( 2 ) where Zmax is the m axim um positive charge of the native pro tein, so 2max = ïïHiü + ^a-NHï + ^Lys + ^Arg (3) in which wH;8, n a_Ni , 2, n Lys, and nArg represent the number of titratable histidines, a-am ino groups, lysines, and arginines. E quations 1 and 2 describe the norm al titration curve of a pro tein. From these equations th e following expression can be derived.

Treatment of Titration D ata--The titration results are presented as normal titration curves (pH versus Z H), differential titration curves ( -d p E /d Z H versus Z h) or difference titration curves (AZH versus p H ), in which
T he inflection points in a normal titration curve of a protein, term inating the neutral region at the acid and alkaline side, are found in a differential titration curve as peaks. T he positions of these peaks are indicated by Z T and Z n . In the case of hem o globin the second peak at Z n is found near pH 9 and is alw ays a sharp one. Calculations have shown th a t for Z lr the following relation holds.

Z l l = Z ma,x -TICOOH -nH's -na-NH2 (5)
in which wCOOH is the number of titratable carboxyl groups. W ith the help of E quation 3 this relation can be w ritten as T he first peak at Z , (near pH 6) represents the titration end point of the carboxyl groups, at least when the peak is sharp. So for Z i we have in first approxim ation the equation Combining the E quations 6 and 7 we have

Calculations have shown th at E quation 7 m ay be applied to a good approxim ation when the first peak is well resolved. In th a t case th e number of carboxyl groups is m ostly found only one group too large. A s a consequence the number of histidines and a-am ino groups, obtained w ith the help of E quation 8, is found one group too small. T his im plies th at to the number of titratable histidines estim ated in this w ay about one group has
to be added. B y means of calculated curves the result can be checked. Therefore, in view of the fact th a t the first peak in the differential titration curve of deoxyhem oglobin is much better resolved than in oxyhemoglobin, we have studied the deoxy form of the several hemoglobins to find the number of titratable histidines. In deoxyhem oglobin the acid Bohr groups are m ainly titrated a t the acid side of the first peak. In our calculations we have assumed th a t there are four acid Bohr groups and, moreover, th a t th ey are carboxyl groups (15).
From th e experim ental differential titration curves Z t and Z Ir can be read off w ith an accuracy of about 0.1.
A ll data and discussions are based on hem oglobin as a tetramer. In Fig. 1 the differential titration curve of hum an hemoglobin A 2 is given. Z n for hemoglobin A 2 is -6.0 while it is -10.0 for hem oglobin A (Table I I Table I we find n Lys + nAls = 58 and n COoH = 68; therefore we expect Z n = -10, as is found experim entally (see F ig. 1). Z j is 11.6, so Zr -Z n is 21.6 as compared w ith 23.0 for hemoglobin A. Correcting for th e fact th a t the sample studied consisted of 75% hem oglobin F and 25% hem oglobin A w e find Z i -Z n = 21.1 and so we conclude th a t in hemoglobin F tw o histidines less are titratable th an in hem oglobin A . T he equal number of titratable histidines in hem oglobins A2 and F  Human A 2............. 15.0 -6 . 0 17. Human F A............ 11.6 -1 0 . 0 17 A i .............. 11.3 -1 0 .3 17.6 18 h Obtained b y rounding off upward to the next even integer the numbers in the preceding column.

D a ta from Dayhoff (17). d Calculated w ith Equation 6 and rounded ofï to the next even integer.
' From Reference 7.
1 Checked b y comparing w ith calculated curves. 0 In accord w ith amino acid composition. h The hemoglobin studied contained 75% hemoglobin F and 26% hemoglobin A.
' Taking into account one extra carboxyl group from the pm ercuribenzoate.
is also shown b y th e nearly identical shape of the two curves. T he main difference is the expected shift of four groups.

Horse Hemoglobin--Horse hemoglobin consists of tw o components roughly in equal am ounts. T hey are called th e slow and fast com ponents according to their electrophoretic behavior (19, 20). In T able I the amino acid composition of the slow and fast com ponents is given. In the fast component lysine E 9 (6 0 )a has been replaced b y a glutamine residue (16). Crystallographically the tw o components are indistinguishable (9).
T he differential titration curves of the tw o com ponents are shown in Fig. 2  term ination of the number of titratable im idazole groups in horse hemoglobin it is perm itted to use the naturally occurring mixture of the tw o com ponents. T he normal and differential titration curve of the deoxygenated and oxygenated form of this m ixture are shown in Fig. 3. (6, 7). T able III gives the parameters which were used to obtain the calculated curves of Fig. 3. T he table shows, besides the to tal number and p K value of each class, the nature and presumed identity of the groups titrated. T he lysines, tyrosines, and cysteines were considered as one class. From th e 12 tyrosines present 8 were assum ed to be titratable (21); the four cysteines and all lysines were considered as being titratable. T he number of lysines is the m ean value of the number of lysines occurring in the two components. A and B are known (19, 22). In a previous report we studied the titration behavior of bovine hem oglobin A (6). T his m ol ecule difïers only a t three sites from hem oglobin B : the residues glycine-15, lysine-18, and lysine-119 in the 8 chain of hem oglo bin A are replaced b y serine, histidine, and asparagine in hem o globin B (17). Fig. 4 shows th e differential titration curves of the deoxy form of b oth hemoglobins. T h e second peak is found at -14.0 for th e B com ponent, while in hemoglobin A Z u is -10.   Fig. 3 of Reference 23). Studying the atom ic m odel of horse hemoglobin and assuming structural correspondence w ith bovine hem oglobin it is even possible to bring both carboxyl groups w ithin saltbridge distance to histidine B l . 1 T he observation th a t th e pK has the same value in b oth deoxy-and oxyhemoglobin can be understood in view of th e fact th a t the three residues are found nearly in sequence in th e peptide chain and are not located a t interfaces so th at they probably are little sensitive to the structural changes which take place upon oxygenation of deoxyhem oglobin. Assuming that th e p K of a normal histidine is near 7 it can easily be calculated th a t a p K shift to 7.8 is associated w ith a Standard free energy change of about 1 kcal per mole which is a reasonable value for th e form ation of a saltbridge (24). Table I

Masked Imidazole Oroups in Horse Hemoglobin-T he number of 16 m asked imidazole groups we found in horse hemoglobin is four larger than reported b y Geddes and Steinhardt (3) and
Tanford and N ozaki (4) who found 12 m asked groups. The experim ental approach of Geddes and Steinhardt was quite different from ours and this could b e an explanation for the dif ference in outcom e. Tanford and N ozaki, however, also followed a procedure in which th ey measure th e number of titrata ble histidines, so a better correspondence w ith our reported value could be expected. In view of the fa ct th at the number of ti tratable histidines calculated from th e titration data in their w ay of analysis is greatly influenced b y th e number of titratable carboxyl groups, the observed discrepancy can be explained as probably caused b y the fact that th ey assumed a number of 78 titratable carboxyl groups while our results indicate a number of 70. M oreover the value of 9.1 th ey assumed for pK BH seems to b e too low (26) and has influenced their results too. In their analysis of the titration data of hum an hemoglobin A B ucci et al. (5) and Tanford and N ozaki (4) reported a number of 16 m asked im idazole groups, only tw o less than we found (7). T h is better agreement is probably due to the fact thay th ey use for hem oglobin A the proper number of ionizable carboxyl groups. I t is further noteworthy to m ention here th a t Geddes and Steinhardt (3) suppose th at at acid pH , besides imidazole groups, also other groups such as neutral «-amino or ionized phenoxy groups, are getting unmasked. H owever, an objection to this supposition could be made. In the case of hum an hemoglobin A w e found th a t Equation 6 could be applied. If neutral eamino groups exist at the pH of the second peak (pH about 9) come masked upon form ation of the tetram er as w ith our values.
Comparison of Hum an, Horse, and Bovine Hemoglobins--For th e comparison of th e human hem oglobins w ith horse and bovine hemoglobin we will start from th e sam e basic assum ption that residues at structurally identical positions w ill show similar ti tration behavior; in addition we will use the knowledge about the titratability of special groups obtained b y comparing the several hum an hemoglobins w ith each other. Because the ce chains of th e hem oglobins studied are all alike as far as the posi tions of histidines are concem ed, the difference in titration be havior of the hem oglobins should be found in th e j3 chains. For th is reason the difference in titratable histidine content of tw o groups betw een horse and hum an hem oglobins A can only be understood if histidines E 13 and E 20 are titratable; this interpretation is supported b y the fact th a t both residues are external (35). A similar comparison betw een hum an hem oglobin A and bovine hem oglobin A failed to give th e right answer; based on th e data of T able IV hum an hem oglobin is expected to have four titratable histidines in excess, while actually only tw o are found. A s y e t we have no explanation for this. One m ay wonder, however, whether it is justified to compare hem oglobins of different m am m als in th e w ay we did, because th ey show besides dif ferences in histidine content m any other am ino acid replacem ents which m ight introducé som e structural differences. T his could make our basic assum ption less valid. A lthough some doubt is justified in this respect we do n o t think th a t this qualification is needed when hem oglobins of th e sam e mam mal are compared because the amino acid replacem ents in these cases are n ot very numerous, which m akes structural correspondence more likely.