Effects of Two Ionizing Groups on the Active Site of Human Carbonic Anhydrase B*

Investigation of some pH-dependent properties of human erythrocyte carbonic anhydrase B indicate that the active site is influenced by at least two charged groups. The properties studied include the pH dependence of inhibition of native, monocarboxamidomethyl, and monocarboxymethyl enzymes by iodide ion and the pH dependence of the visible spectra of the cobalt derivatives of these enzymes. One ionizing group has a pK, of about 7.3 in the native enzyme, 8.2 in the carboxyamidomethyl enzyme, and 9.0 in the carboxymethyl enzyme. It has a major influence on activity and anion inhibition, and on the visible spectra of the cobalt eznymes. A second group has a pK, of about 6.1 in native and modified enzymes. When zinc is at the active site, the secondary group in its acidic form decreases the K, for I-. With the carboxamidomethyl and carboxymethyl enzymes, the K, decreases by about an order of magnitude. However, if cobalt is substituted for zinc in the modified enzymes, this group does not influence the K, for I- and the binding of I- does not influence the pK, of the spectral transitions caused by ionization of this secondary group. In the case of nonalkylated Co”+ -enzyme, another ionizing group with a pK of about 6.2 prevents the binding

From the Department of Biochemistry, University of Miami School of Medicine, Miami, Florida 33152 Investigation of some pH-dependent properties of human erythrocyte carbonic anhydrase B indicate that the active site is influenced by at least two charged groups. The properties studied include the pH dependence of inhibition of native, monocarboxamidomethyl, and monocarboxymethyl enzymes by iodide ion and the pH dependence of the visible spectra of the cobalt derivatives of these enzymes. One ionizing group has a pK, of about 7.3 in the native enzyme, 8.2 in the carboxyamidomethyl enzyme, and 9.0 in the carboxymethyl enzyme. It has a major influence on activity and anion inhibition, and on the visible spectra of the cobalt eznymes.
A second group has a pK, of about 6.1 in native and modified enzymes. When zinc is at the active site, the secondary group in its acidic form decreases the K, for I-. With the carboxamidomethyl and carboxymethyl enzymes, the K, decreases by about an order of magnitude. However, if cobalt is substituted for zinc in the modified enzymes, this group does not influence the K, for I-and the binding of I-does not influence the pK, of the spectral transitions caused by ionization of this secondary group. In the case of nonalkylated Co"+ -enzyme, another ionizing group with a pK of about 6.2 prevents the binding of I-at low pH. These results show that the active site is altered when cobalt is substituted for zinc in carbonic anhydrase B.
An important test in probing the active site of an enzyme is the study of the pH dependence of specific properties of the active site. Many approaches have been applied to carbonic anhydrase B from human erythrocytes to get this type of information.
These include the pH dependence of activity and its inhibition (l-8), binding of inhibitors (g-12), chemical modification (6)(7)(8), the visible spectrum of the Co*+-enzyme (4,6,13,14), and NMR spectra of histidine protons (15)(16)(17)(18)(19)(20). Most of the active site properties have been analyzed in terms of the simplest case where only one ionizing group influences them. However, the pH dependence of CO,-hydrating activity (1,2) and the visible spectrum of the Co+enzyme (6) did not follow a simple titration curve. In addition, the pH dependence of inhibition by bromoacetate suggested that more than one ionizing group influenced the binding of this anion (6) (6,7,21,22). These alkylating agents are affinity reagents which bind at the active site as reversible inhibitors and then proceed to react with the 3'-nitrogen of histidine 200 (7,23,24). The modified enzymes have been purified and shown to have residual activity (6,7,22,25). The sedimentation coefficient and optical rotatory dispersion of the carboxymethyl enzyme were the same as for the native enzyme (7) indicating that no large conformational changes occurred.
X-ray diffraction studies (26) determined that histidine 200 is a surface residue in the active site region, and is located within 7 A of the zinc ion which is at the base of the active site cavity.

MATERIALS AND METHODS
Carbonic anhydrase B was prepared from human erythrocytes and employed to prepare Co?+-, Cam-' and Cm-enzyme derivatives (22). CamCo'+-enzyme was prepared from Co*+-enzyme by reaction at room temperature with 2.5 mM iodoacetamide at pH 7 for 24 hours. CmCoZ+-enzyme was prepared by reacting Coz+-enzyme with 1 mM bromoacetate at pH 7 for 5 hours. Each was then titrated to about pH 8 and immediately chromatographed on columns (1.8 x 80 cm) of Sulfamylan-Sepharose (25) at 4" equilibrated with 5 rnM Tris sulfate/22 mM Na,SO,, pH 7.6, and eluted with 10 mM sodium phosphate/20 rnM Na,SO,, pH 6.5. It was observed that although these modified enzymes were retarded, they slowly moved down the column in starting buffer. Thus, after applying the sample, the column was 'The abbreviations used are: Cam, carboxamidomethyl; Cm, carboxymethyl; bis-Tris, bis(2-hydroxyethyl)imino-tris(hydroxymethyl)aminomethane.

pH-dependent
Properties of Human Carbonic Anhydrase B 3863 washed with not more than 100 ml of starting buffer before changing to the second buffer to bring the modified enzyme off in a sharp peak. The reactive reagents and inactive protein came off at the front of the elution pattern. Unmodified enzyme remained bound to the affinity gel and was eluted later. The fractions containing modified enzyme were pooled and concentrated by vacuum dialysis. The modified enzymes were dialyzed against three changes of deionized water during a total of about 6 hours. Alkylated Co*+-enzymes were prepared by modifying the Co+ enzymes because it is very difficult to remove zinc from the alkylated enzymes. The modified Co*+ -enzymes prepared by alkylation of the CoZ+-enzyme or by alkylating and then changing the metal have only one modified histidine, behave the same on affinity chromatography, and have the same kinetic properties. Most of the spectral properties are also the same.
The concentrations of native and modified enzymes were determined from their absorbance at 280 nm (c = 47,000 M-' cm-l). The extinction coefficient was calculated from the absorbance of a 1 mg/ml solution at 280 nm of 1.63 (27), and a molecular weight of 28,850 (23).
Esterase activity was assayed with p-nitrophenyl acetate as described (6), except that 20 ~1 of 50 mM substrate in dimethylsulfoxide was used.
Iodoacetate, bromoacetate, and p-nitrophenyl acetate were purified as previously described (6) and carboxymethyl enzymes were determined in the pH range of 5.3 to 9.8 ( Fig. 1). A simple titration curve for a group with a pK, of 6.8 was calculated to fit the data with the native enzyme. The deviation from the theoretical curve between pH 6 and 7.5, while not large, was reproducible and suggestive that more than one ionizing group influences I-binding. A better fit could be obtained with a theoretical curve with two pK, values, but the deviation from the simple curve is too small to permit a reliable, unique solution.
Results for the modified enzymes definitely show more than one inflection. The theoretical curves are shown for a simple two-group case of two independent ionizing groups in the enzyme which influence the binding of I-. The ionization of the two groups on the enzyme are illustrated in Equation 1, with the most acidic form arbitrarily assigned to be neutral, and the dissociated protons omitted.
Hy-E-xHK\ ,,.,,$A K,, K,, K,, and K, are acid dissociation constants, and it is assumed that K, = K, and K, = K, and that pK, > pK,. The binding of iodide to HY-E-XH is assumed to be strongest; it is weaker to -Y-E-XH, and not significant to the other two forms.
Although pK, and pK, cannot be assigned with great accuracy, the data closely fit the theoretical lines generated when pK, = 6.1 and pK, = 8.2 for the Cam-enzyme and 6.1 and 9.0 for the Cm-enzyme.
The pH dependence of Cll inhibition of native enzyme was investigated at ionic strengths of 0.075 and 0.2. The minimum K, at low pH was 2 mM at an ionic strength of 0.075 and 3 mM at an ionic strength of 0.2. The pH dependence was similar to that for iodide with an apparent pK, near 6.8 for the case where a single pK, is assumed.
Evidence that the primary binding site is the same for Cl-, Br, and I-is shown in Table I. The degree of inhibition of esterase activity at pH 6.5 by Cl-+ Br-, Cl-+ I-, or Br-+ Iagrees with that calculated for the case where the binding of one inhibitor is strictly competitive with the other. Binding of Halide Ions to Co'+-enzymes-The pH dependence of the K, of iodide for the Co'+-enzymes are shown in Fig.  2. A surprising result is that the group with a pK of 6.1 has absolutely no effect on the binding of iodide to the Co*+ form of the Cam-or Cm-enzyme.
Additional results show that the difference between Cm-enzyme and CmCoa+-enzyme is due to FIG. 1. pH dependence of inhibition by iodide ion. The inhibition of e&erase activity was determined at 25" and at an ionic strength of 0.2. Below pH 7, 50 mM bis-Tris sulfate buffer was used and 50 mM Tris sulfate was used above that pH. The curves are theoretical, calculated on the following bases. Native enzyme (dashed line), pK, = 6.8 and the limiting K, at low pH is 0.096 mM. Cam-enzyme (solid line), PK.' = 6.1, pKaZ = 8.2, limiting K, = 0.35 mM, and the K, with only the more basic group in its acidic form is 2.2 mM. Cm-enzyme (dotted line); pK,' = 6.1, PK.' = 9.0, limiting K, = 0.6 msr, and the K, with only the more basic group in its acidic form is 10 mM.  Fig. 1. The theoretical curves were calculated on the following bases. Coz+-enzyme (dashed line), pK,' = 6.2 and pKaz = 7.2, the K, with group 1 in its basic form and group 2 in its acidic form is 0.66 mM; CamCo*+ -enzyme (solid line); pK = 8.2 and limiting K, = 0.36 mM; CmCol+-enzyme (dotted line), pK = 9.0 and limiting K, = 2.1 mM. the substitution of Co"+ for Zn2+ and not to the procedure of removing the metal. Cm-enzyme was dialyzed against ophenanthroline at pH 5.5 for 20 days, and the apoenzyme separated from the metalloenzyme by affinity chromatography using the same conditions as described under "Materials and Methods" for purification of Cm-enzyme. Since the metal dissociates slowly from Cm-or Cam-enzyme, most (80%) of the enzyme was recovered as CmZn*+-enzyme, and this enzyme showed the same complex pH dependence of K, for iodide ion as the untreated Cm-enzyme.
Cobalt was added to half of the apoenzyme and zinc was added to the other half. The Zn'+enzyme showed the complex pH dependence and the Co'+enzyme did not.
The pH dependence of iodide inhibition of Coz+-enzyme shows a maximum at pH 6.7 (Fig. 2). The data were fitted with a curve with one group (pK = 7.2) which must be in its acidic form and a second group (pK = 6.2) which must be in its basic form for binding to occur. The esterase activity falls off rapidly below pH 6. Since the metal dissociates from the nonalkylated enzyme much more rapidly than from the Cam-or Cmenzymes, the possibility that the metal may be dissociating from the Co'+-enzyme was investigated. It was found that neither the activity nor the inhibition was affected by addition of 20 PM excess Co2+ at pH 5.5.
The effects of halide ions on the spectra of Co'+-enzyme and its carboxamidomethyl and carboxymethyl derivatives are shown in Fig. 3. The spectra of the native Coz+-enzyme with Cl-and Br-are very similar, but the spectrum with I-shows more resolved peaks in the 600 to 650 nm region and an additional large band below 350 nm. The spectra of the modified Co*+-enzymes with Cl-and I-are similar to those for anions with the native Coz+ -enzyme, but the modified enzymes have much smaller peaks in the 500 to 650 nm region and a pronounced shoulder in the spectra with I-is evident near 340 nm. The basic similarity of the spectra of the three Co'+enzymes with I-is more evident at lower pH values (Fig. 4). The changes in the spectra as the pH value increases are also quite similar.
It seems reasonable to conclude that the site and general manner of binding of I-is the same for the three Co*+-enzymes.
Some of the dissociation constants of these ions with the different Co*+-enzymes were determined by spectral changes. In Table II

pH-dependent Spectral
Changes-The pH-dependent changes in the spectra of the three Co+enzymes in the presence (Fig. 4) and absence (Fig. 5) of I-were determined in the region of pH 5.5 to 8 (Fig. 6)  The solution was quickly filtered, scanned, and titrated to higher pH. Since this laboratory previously reported that the CamCoZ+-enzyme did not show a secondary transition in its spectrum at low pH (22), the present results were checked with three different preparations of enzyme. The reason it was not observed before may be due to slightly different methods of preparation of the CamCo*+-enzyme. The procedure used for these studies was more gentle, since the dialysis time against o-phenanthroline at pH 5.5 was much shorter. The alkaline spectrum of Co2+-enzyme is given in Fig. 4A to show that the spectral changes at low pH are small compared to the changes which occur in the major pH-dependent spectral transition. The alkaline spectra for Cam-and CmCo2+-enzymes are similar to that of the CoZ+-enzyme (6,22). Since these spectral changes were small compared to the major transitions which occur at higher pH, the pH dependence of the minor transitions was followed at wavelengths near an isosbestic point for the major transition (Fig. 6). The data were fitted with theoretical titration curves with pK, values of 6.2 for native Co*+-enzymes and 6.0 for the modified enzymes. The magnitude of the changes were similar in all six cases but the direction of the changes in the presence of I-was opposite from those in its absence. Since the same pK, is obtained in the presence or absence of I-, it is likely that the same pH-dependent ionization is responsible for the observed spectral changes in the Co*+-enzyme and the Co?+. enzyme. I-complex. native enzyme and two chemically modified derivatives. In addition, the effects of halide ions on the visible spectra on the Co*+-enzymes were studied. From the data of the pH dependence of inhibition of native carbonic anhydrase B by anions it is difficult to make any definite conclusions about the pK of the ionizing group or groups which control the binding. Assuming that a single ionizing group controls the binding of iodide and chloride, its apparent pK, is about 6.8. This agrees with previous results for inhibition by iodoacetate (7) and bromoacetate (6). However, Ward studied the pH dependence of chloride ion binding using "Cl NMR techniques and calculated a pK, of 8.2 (11). This value was calculated using a literature value of 51 mM for the K, of chloride.
By replotting Ward's data using the K, determined here (2 mM), an apparent pK, of 7.0 was obtained which is in reasonable agreement with the pK obtained in the inhibition studies. (A similar replot of Ward's data for carbonic anhydrase C, using a K, of 200 mM which was estimated from results at pH 6.5 (25), yielded an apparent pK, = 6.7.) The pK, of 6.8 obtained here for enzyme B does not agree with the pK. near 7.3 for the pH dependence of esterase activity and the visible spectrum of the Co*+-enzyme (6,13,14). In addition, it was found previously that if sufficient data on bromoacetate inhibition are obtained, they do not exactly fit the simple case for a single PK., but can be fitted to a case where two ionizing groups control the binding (6). The pK, values could not be assigned with great accuracy, but they were estimated to be 7.3 and 6.4. Thus, one of the pK, values is the same as the major one affecting activity and the visible spectrum.
The ambiquities and uncertainties in the determination of the number and pK values of groups influencing anion binding to this enzyme serves to re-emphasize the fact that it is difficult to assign reliable pK values from one set of data on a pH-dependent property, since an undetected secondary group may affect the curve in subtle ways if its pK is near that of the primary group which controls binding. To reinforce the data showing that more than one ionizing group influences the active site, two chemically modified forms of the enzyme were used. From the pH dependence of esterase activity and the visible spectra of the Co*+-enzymes, it was observed that the pK, of the primary group affecting these properties was shifted from 7.3 for the native enzyme to about 8.3 for the Cam-enzyme and 9.2 for the Cm-enzyme (6,22). Hence, in these derivatives the ionization of the primary group is shifted enough to make a secondary group with a lower pK more evident. This is clearly demonstrated by the results in Figs. 1 and 2. The major pH-dependent transition can now be seen to have the same pK as observed for esterase activity and the visible spectra.
The pK, for the secondary group is about 6.1 in both modified enzymes.
The pK, values of 6.4 estimated previously for native enzyme (22), and the pK of 6.1 determined here for the modified enzymes are within experimental uncertainty, so it is likely that the pK of this group is affected very little by alkylation of histidine 200 with either a neutral or negatively charged group. The most surprising finding is that the substitution of cobalt for zinc causes such marked changes in the pH dependence of anion binding at low pH. The finding that the secondary group (pK, = 6.1) does not influence the K, for iodide with CmCo'+or CamCo*+-enzyme is in accord with the results showing that the binding of iodide does not shift the pK, of the spectral transition.
However, the case with the native enzyme is not so simple, since there are pH-dependent transitions with pK, values near 6.1 which promote the binding of I-to Zn2+enzyme, prevent I-binding to Co2+-enzyme, and produce spectral changes in the Co'+-enzyme and the Co'+.enzyme.Icomplex. If the same ionization which prevents binding of I-to Coz+-enzyme at low pH was responsible for the spectral transition, then the spectral transition would have a higher pK, in the presence of I-. Such is not the case, so another ionization is implicated. It is unlikely that the loss in ability to bind I-at low pH is due to the binding of a proton to a nearby basic group, so it may be due to a small pH-dependent conformational change which shows up in the native Co'+enzyme but not in the alkylated Co*+-enzymes. The identities of the ionizing groups involved in enzymic functions are not known. The ionizing group of primary importance has a pK, of 7.3 in the native enzyme and is thought to be a zinc-bound water molecule (2,28,29), although alternatives involving histidyl side chains have been proposed (18,30,31). The group with a pK near 6.1 could be due to histidines 64, 67, or 200 (26), all of which are in the active site region. Histidine 200 is the one modified in the Cm-and Cam-enzyme (7,23,24). In the native enzyme, this histidine has a pK, of 5.6 in the presence of reversibly bound bromoacetate (6) and 5.8 with iodoacetate (7). Its pK, in the absence of inhibitor could be higher and it would not be expected to change much upon alkylation (21). Therefore, any of these histidyl residues are possible candidates for affecting anion binding.
An important question is whether the secondary group has any other functional significance.
The pH dependence of CO*-hydrating activity of enzyme B does not follow a simple titration curve (l), and this may be due to a positive or negative influence by this secondary group. Perhaps the effects of the secondary group will show up more distinctly in studies using dehydrating activity, since this has a maximum at low pH. Some dehydration studies have been done (3), but, due to experimental limitations, these were not extended to low enough pH to see these effects. The involvement of a second ionizing group has also been postulated to account for hydrogen isotope effects on the kinetics of CO, hydration and HCO, dehydration by human carbonic anhydrase C (32). Possible roles of the secondary groups may be to enhance binding of anions, to influence the water structure in the active site, and to function in concert with the primary group which is responsible for enzymic activity.