Eur. J. Biochem. 208, 607-615 (1992)

0FEBS 1992

The interaction of acetate and formate with cobalt carbonic anhydrase An NMR study Ivano BERTINI', Claudio LUCHINAT', Roberta PIERATTELLI' and Alejandro J. VILA' Department of Chemistry, University of Florence, Italy Institute of Agricultural Chemistry, University of Bologna, Italy (Received April 6/June 18,1992) - EJB 92 0478

The interaction of formate and acetate ions with cobalt-substituted carbonic anhydrase (CA) has been investigated through 3C-NMR and one-dimensional and two-dimensional 'H-NMR spectroscopy. 13C data on formate are consistent with a regularly coordinated ligand, as previously proposed for the acetate anion [Bertini, I., Luchinat, C. & Scozzafava, A. (1977) J . Chem. SOC. Dalton Trans., 1962- 19651. 'H-NOE experiments on both anions give evidence of through-space interactions between ligand protons and protein protons. The latter are assigned to specific residues in the active cavity through nuclear Overhauser effect spectroscopy (NOESY) experiments. The 3Cderived and 'H-derived constrains allow reliable docking of these ligands in the active-site cavity. The resulting geometries are similar to one another and consistent with five-coordinated structures around the metal ion, as previously proposed from electronic spectroscopy [Bertini, I., Canti, G., Luchinat, C. & Scozzafava, A. (1978) J. Am. Chem. SOC.100,4873-48771. The results are discussed in light of the current debate on anion binding to metal ions in carbonic anhydrase [Lindahl, M., Svensson, A. & Liljas, A . (1992) Proteins, in the press]; Bertini, I., Luchinat, C., Pierattelli, R. & Vila, A. J. (1992) Znorg. Chem., in the press; Banci, L. & Merz, K. (1992) unpublished results] and, in particular, of the proposed long Zn-0 distance found in the recent X-ray results on the formate adduct [Hakanson, K., Carlsson, M., Svensson, A. & Liljas, A. (1992) J . Mol. Biol., in the press].

'

'

Carbonic anhydrase (CA) is a zinc enzyme which catalyzes the reversible hydration of COz [l]. As this reaction is fundamental in biology, detailed structural [2, 31 and mechanistic [4- 61 studies have been undertaken since its discovery. Amongst them, the understanding of the interaction of the metal ion with inhibitors constitutes a key step in the unraveling of its function [7 - 101. The zinc ion is coordinated to three histidines (His94, His96 and His1 19) and to a water molecule, which is located at the bottom of a cavity (Fig. 1) [ll]. On the basis of spectroscopic investigations, mostly of the cobalt-substituted enzyme, binding of the inhibitors was proposed to proceed in two ways; by substitution of the water molecule, yielding a pseudo tetrahedral coordination to the metal, or by addition as a fifth ligand, leaving the water molecule coordinated to zinc [8, 121. Consistent with spectroscopy, X-ray analyses have shown that hydrogen sulphide and sulphonamides are examples of the Correspondence to I. Bertini, Department of Chemistry, University of Florence, Via Gino Capponi 7,I-50121, Florence, Italy Abbreviations. CA, carbonic anhydrase; H, human; B, bovine; NOESY, nuclear Overhauser effect spectroscopy; FID, free induction decay; TPPI, time-proportional phase incrementation. Enzyme. Carbonic anhydrase (EC 4.2.1.1).

former class of inhibitors [13, 141, whereas thiocyanate is an example of the latter [15]. As far as cyanide and cyanate are concerned, spectroscopy was indicative of a tetracoordinated arrangement of the ligands [8, 121. The X-ray analysis in the solid state has shown that they do not bind at the metal, leaving the chromophore unaltered, but they do bind in the vicinity of the metal [16]. They are hydrogen bonded to the N H group of Thr199, replacing the so-called 'deep-water' molecule [16]. Further heteronuclear NMR investigations suggest direct binding to both cobalt and zinc [17] and water-proton relaxation points to substitution of the coordinated water by the anions [18]. This could be a case of a substantial difference between the solid state and solution. Crystallographic results are also available on the formate adduct of the native enzyme [19]. The anion is located in a similar way as the pentacoordinated ligand NCS-, but at a unusually long Zn-0 distance (0.25 nm). This result is quite unusual in the coordination chemistry of zinc and could be justified only by strong hydrogen bonds and/or other interactions within the enzymatic cavity. The spectroscopic study of CA has always been limited by the scarcity of information which could be obtained from the zinc ion. The cobalt(I1)-substituted CA (COCA) has proved

608 were multiplied by an exponential function to improve the signallnoise ratio, introducing a 20 - 30-Hz line broadening. One-dimensional spectra at 200 MHz were recorded using the superweft [23] pulse sequence with a recycle delay of 100 200 ms, while those at 600 MHz were performed presaturating the residual water signal during the relaxation delay (1 s). T I measurements were performed with the inversion recovery method [24]. T, values were obtained from the bandwidth at half-height according to T , = l/xAv. 'H-NOE experiments at 600 MHz were performed using a recycle time of 2 s and a saturation pulse of 4.5 W applied for 300 ms, and collected using the previously reported methodology [25, 261. The CH3 signal of acetate or the proton of formate, in large excess with respect to the protein, were saturated at 40-70%. Due to the fast exchange with bound Fig. 1. Stereoview of the active site of carbonic anhydrase as taken from ligand, NOE can be observed for protein signals. The difference spectra were compared with spectra obtained by the X-ray structure of HCA I1 1111. irradiating, at the same chemical shift, the corresponding adducts with deuterated ligands. Final difference spectra, typito have a catalytic behavior similar to that of the native cally consist of 4- 5 blocks of 3200 scans each. The selective T,s of signals exhibiting NOES were meaenzyme [6], and it is suitable for spectroscopic investigations; sured using the saturation-recovery technique [27] applying a therefore, it has been extensively used as a probe of the native enzyme, although minor differences may exist between the selective 180O Gaussian-shaped pulse [28]. These pulses were calibrated as 180" Gaussian-shaped pulses with a 10% of two derivatives [8, 121. truncation (4 K points; duration, 8.192 ms) upon each signal We report here on a 13C-NMR investigation of HCOOin the presence of CoBCA I1 (B, bovine). The aim is to check of interest. NOESY spectra [29] were recorded with presaturation of whether the NMR data are consistent with the long distance the HDO and the methyl acetate signal during the recycle found in the solid state. Further, 'H-NMR spectra were recorded in order to gain information on the binding mode of delay (1 s) and the mixing time (100 ms or 20 ms), except when indicated. They were collected in the phase-sensitive mode formate and acetate. In particular, well-designed 'H-NOE could provide information regarding the distances between using the time proportional phase incrementation method the ligand protons and the protein residues. The latter, in (TPPI) [30]. 512 FID were collected using 1024 data points their turn, were assigned through two-dimensional nuclear each. Zero filling in the F1 dimension was applied in order to Overhauser effect spectroscopy (NOESY) experiments in obtain a 1 K x 1 K data-point matrix. To the data a sine-bellanalogy to the assignment available for the C10, adduct [20]. squared weighting function of 90 phase shift was applied. 13C spectra were recorded on a Bruker MSL 200 specIn this way, significant connectivities have been found for the formate and for the acetate which have allowed a useful trometer at 50.33 MHz, in a 10-mm widebore broadband docking of the anions in the cavity. In the case of formate, probe. the picture is quite consistent with the X-ray data. All the approaches are based on the NMR investigation of nuclei which sense the paramagnetic center and are representative RESULTS AND DISCUSSION examples of the potentiality of NMR studies of paramagnetic Relaxation measurements molecules in the investigation of relatively large molecules. O

Experimental section

BCA I1 was purchased from Sigma Chemical Co. and used without further purification. Human (H) carbonic anhydrase I was purified from recently outdated human red cells according to the procedure of Khalifah [21]. Apo-CA was prepared by removing zinc(I1) ions by dialysis against solutions of 50 mM pyridine-2,6-dicarboxylic acid in 0.2 M sodium phosphate, pH 6.9 [22]. Cobalt(I1) addition was performed by titration of the apoenzyme with a CoS04 solution followed spectrophotometrically. 'H-NMR spectra were performed on a Bruker MSL 200 and on a Bruker AMX 600 at 200.13 MHz and 600.14 MHz, respectively. All 'H-NMR samples were 2 - 3 mM at the pH indicated in the figures. All D 2 0 solutions were prepared by exchanging the aqueous solution with DzO. The pH meter readings were uncorrected for isotopic effects. In all cases, the anions were added until total formation of the final adducts. Chemical shifts were calculated by assigning a value of 4.8 ppm to the solvent signal. The free induction decays (FID)

One way to probe direct ligand binding to the paramagnetic metal ion is that of measuring the relaxation induced by the paramagnetic center on the ligated nuclei. Under fastexchange conditions, a nucleus in the vicinity of the paramagnetic ion senses the presence of the unpaired electrons which provide efficient nuclear-relaxation pathways and therefore shorten T1and T2 [31-331. If the ligand is in excess with respect to the enzyme and is under fast-exchange conditions, the measured longitudinal relaxation rate of a nucleus belonging to the ligand will be the sum of the following two contributions: T-' Id

+ T;;,

TC: = fM

. TCd,

T;'

(1) where T;: is the relaxation of the resonating nucleus in the presence of an equimolar solution of ZnBCA I1 and TF: is the effect of the paramagnetic ion. The latter is given by =

(2)

wherefM is the molar fraction of bound ligand, and T;A is the full paramagnetic effect. The values of fM are estimated by assuming that all the enzyme is bound in accordance with the

609 Table 1. Effect of the addition of sub-stoichiometricamounts of CoBCA I1 on the nuclear relaxation times of a 15 mM HI3COO- at pH 6.0. T1 was measured with an error of 10%. HCOO- + ZnBCA HCOO- + CoBCA parameters were measured using a formate:enzyme ratio of 100.

+

HCOOHCOO300 K 305 K HCOO-

+ CoBCA

+ ZnBCA

S

Hz

12.70

18

0.44 0.51 3.52

56 63 20

s-l

1.89 1.60

189 160

reported value of the affinity constant of formate (log& = 4.1 at pH 6.0) [8]. The longitudinal-relaxation-rate enhancement caused by the paramagnetic ion is induced by the unpaired electronnucleus coupling [31]. This interaction may be dipolar and contact in origin. Assuming that the coupling is exclusively dipolar, this enhancement is given by the Solomon equation [341

. I-

37,

\1

.

7%

\

(3)

+ wl"z," l+wsza)' +

where p o is the permeability of vacuum, yN is the proton magnetogyric ratio, g, is the electron g factor, pB is the Bohr magneton, r is the electron-nucleus distance, and w, and us are the nuclear and electron Larmor precession frequencies, respectively. The correlation time z, is given by

+

7;' + zM1, (4) where z, is the electronic relaxation time, z, is the rotational correlation time and zu is the chemical exchange time. In this case, the correlation time is given by the electronic relaxation time of the metal ion z,. Hence, this equation allows us to correlate TFA with structural information. if other relaxation mechanisms are operative, the obtained distances represent upper limits. An addition of 1% CoBCA I1 to a 50 mM solution of H13COONa at pH 6.0 induced a considerable broadening (from 18 Hz to 56 Hz) of the 13Cresonance, with a concomitant reduction of the longitudinal relaxation time T1 (see Table 1). In this case, we have found a TlA of 189 s-', which is consistent with that previously reported for the acetate adduct (130 s-') [35]. Therefore, it can be inferred that both anions bind essentially in the same way, as is also shown by electronic [8] and 'H-NMR spectroscopy. Experiments were performed at two different temperatures showing that T;: decreased at higher temperatures. This indicates that TlP is in fast-exchanging conditions. The inverse effect observed for TZpis not contradictory, but indicates that Tg; experiences a considerable contribution from the chemical-shift difference between bound and free species [311. A Co-C distance of 0.34 nm, as found in the solid-state structure [19], would require a z, of 1.25 x lo-'' s to account for the observed T;& value. Such a value is too large for a five-coordinated cobalt derivative [18]. However, taking a z, 7,'

-1

= 7,

of 4.2 ps as found for the acetate adduct from water-proton relaxation measurement of CoBCA I1 [18], a cobalt-carbon distance of 0.29 nm for formate is obtained from Solomon's equation. This value is consistent with that calculated in a similar way for the acetate adduct, i.e. 0.31 nm [35]. The use of the same z, value for both adducts is justified by the similarity of the T1 values of the hyperfine shifted signals and of all the other spectroscopic data which point to a very similar coordination geometry. These calculated distances indicate a direct Co-0 binding, although no conclusion can be made on the Co-0 distances since they depend on the Co-0-C angle. However, Co-C distances, as found here, are in the same range of those proposed for regularly coordinated carboxylate complexes [36- 381. In most ligands coordinated to paramagnetic ions, both contact and ligand-centered effects are operative, which simulate shorter metal-carbon distances 131, 39,401. Apparently this is not the case for 3C in carboxylate complexes [41, 421.

'H-NMR studies on the formate and acetate adducts After confirming direct binding of the ligand with the metal ion, we turned to analyze the 'H-NMR spectrum of the adduct in an attempt to characterize the active cavity. Fig. 2 shows the NMR spectra at 200 MHz of the final adduct of CoBCA I1 with formate and acetate in H 2 0 at pH 6.0. The latter is equal to that previously published 1431. In the case of formate, four paramagnetically shifted signals (A - D) are seen in the downfield region, three of them exchangeable in D 2 0 (A, C, D). The strong paramagnetic effect observed on these signals indicates that they belong to protons of residues strongly interacting with the metal ion, i. e. the coordinated histidines. On this basis, following assignments of related adducts, the three exchangeable signals are attributed to the three imidazolic NH protons, while the non-exchangeable proton corresponds to the H62 of His119 (see Fig. l), which is 0.5-nm far from the metal ion as this histidine is coordinated by N61. The other non-exchangeable protons of these residues are barely observed at this magnetic field, due to the large linewidth induced by the proximity to the paramagnetic center. The region of the spectrum near the diamagnetic envelope (20/ - 10 ppm) is quite rich in paramagnetically shifted signals compared to the anion-free CoBCA I1 [43]. Amongst them, we can point out the presence of signals E and z at 15.4 ppm and -8.5 ppm, respectively, which, according to their intensity, correspond to methyl groups. All these signals do belong to protons which experience a larger pseudocontact shift upon anion binding. The signals of the two adducts were correlated by titration of the acetate adduct with formate. in such a way, an assignment made on any of the two adducts can be transferred to the other. In Table 2, the measured chemical shifts are summarized, and the same letters are used to identify the signals which correspond to the same protons in both adducts. Signals E and z were previously assigned in the acetate adduct to the methyl groups of Thr2OO and Thr199, respectively [43]. Paramagnetic-induced shifts influencing non-coordinated residues are pseudocontact in origin, i. e. they depend on the magnitude and orientation of the magnetic anisotropy tensor, Pentacoordinated cobalt(i1) complexes experience a larger orbital contribution to the magnetic susceptibility than four coordinated ones 1441and, therefore, a larger anisotropy. This will induce larger dipolar shifts and therefore a higher spreading of the dipolarly shifted signals. This feature has been

610

S I

I Z U

I

1

A

I

$0

B

I

60

I

I

80

I

20 chemical shift (ppm) 40

I

0

40

20

-20

d

I

60

J

0

-20

chemical shift (pprn)

Fig. 2.300-K 'H-NMR spectra in H 2 0 of the (A) formate and (B) acetate adducts of cobalt (11)-substitutedbovine carbonic anhydrase at pH 6.0 at 200 MHz. The shaded signals correspond to exchangeable protons in DzO.

pointed out as characteristic of pentacoordinated adducts [43]. Furthermore, pentacoordinated adducts have short z, and longer nuclear T I values with respect to pseudotetrahedral ones. The T I values of the formate and acetate adducts at 200 MHz (Table 2) are typical of pentacoordinated CA adducts. The spectra of both adducts at 600 MHz are reported in Figs 3A and 4A. Since at such a high field there is a large contribution of the Curie-spin-relaxation mechanism [45,46], some of the broadest signals observed at 200 MHz are broadened beyond detection. In fact, by taking the metalproton distances from the crystallographic data of the NCSadduct of HCA I1 and a rotational correlation time of 15 ns, by means of the following equation

it can be estimated that protons in a sphere of 0.6-nm radius centered on the metal will experience an exceedingly large Curie broadening.

'H-NOE measurements NOE qij is defined as the fractional change of the intensity of a signal i when another signalj, dipolarly coupled to it, is selectively saturated [47]. If the saturation time is much larger than the spin-lattice relaxation time of signal i (i. e. a steady state is reached), the NOE intensity for a multi-spin system is given by qi{j} = l/ei[cij - Cxqx(J}oixI, (6) where ei is the selective spin-lattice relaxation rate of signal i (the signal undergoing NOE), oijis the cross-relaxation rate between protons i and j , qx(j} is the NOE on other nearby protons, and oixis the cross-relaxation rate between the observed nucleus i and the protons x experiencing also NOE

61 1 Table 2. 'H-NMR chemical shifts and non-selective TIvalues for the formate and acetate adducts of CoBCA I1 at 298 K. T I values were measured with an error of 10%: Signals A, C and D were exchangeable signals in D20. Signals A, B, C, D, E and z were measured at 200 MHz. Signals r, s, u, v and z were measured at 600 MHz.

Signal Formate adduct

A B C D E r S

U V Z

Acetate adduct

Assignment

6

TI

6

TI

PPm 76.4 66.2 62.6 53.2 15.4 - 3.06 -3.25 -7.20 -4.40 -8.5

K 10 15 14 17 31

PPm 76.3 67.5 66.9 55.7 16.3

K 19 20 20 22 24

NHg2 Hisl 19 H62 Hisl 19 NH61 His94 (96) NH61 His96 (94) yCH, Thr2OO

232 137 223 38

-3.12 -4.00 -4.15 -12.1

399 230 320 40

yCH, yCH, yCH, yCH,

Val207 Val143 Val207 Thr199

upon irradiation of signalj . The second term of Eqn 6 accounts for spin-diffusion effects which turn out to be negligible in the present case (see later), yielding the so-called two-spin approximation y.I = o../ I] el. (7) The cross relaxation rate between i and j is given by: 0.. = I]

h2y47, 40~~1-5

-__

where rij is the i-j interproton distance and z, the reorientational correlation time of the molecule. In this way, the irradiation of the proton resonances of the inhibitors can yield significant information pointing to the identification of the protein residues in the vicinity of the inhibitors, thus providing useful clues for the docking of the inhibitor molecules in the cavity. In the case of the formate adduct, the metal-bound formate is in fast exchange with the free formate, thus providing an averaged signal at 8.65 ppm. By irradiation of this signal, NOE are detected on signal r at -3.06 ppm, on signal s at -3.25 ppm, and on signal u at -7.20 ppm and a smaller NOE is also observed on signal v at -4.40 ppm (Fig. 3B). Signals s, u and v, according to their intensities, correspond to methyl groups. A similar NOE experiment was performed on the acetate adduct in an attempt to correlate the spatial orientation of both anions in the active cavity. When irradiating the fastexchanging acetate signal, three NOE were detected in the upfield region on signals s, u, and v located at -3.12, -4.00 and -4.15 ppm (see Fig. 4B). In order to confirm that the observed NOE originated from the CH3 moiety of acetate and not of signals belonging to the diamagnetic region of the protein, the same experiment was performed on the adduct with CD3COO- by saturating at the resonating frequency of the methyl signal. The NOE from signals s, u and v were absent (see Fig. 4C), thus confirming that the observed enhancements shown in Fig. 4B are originated by a true interaction of protein residues with the acetate moiety. To obtain quantitative information from these NOE data, we performed the measurement of the selective T I values of the signals giving NOE. These are reported in Table 3,

Fig. 3. NMR spectrum and NOE difference spectrum of the formate adduct. (A) 600-MHz 300-K 'H-NMR spectrum of the formate adduct of CoBCA I1 at pH 6.0; (B) NOE difference spectrum of the formate adduct upon irradiation of the formate proton signal (8.65 ppm).

together with the measured percent intensity of the NOE. The latter values were calculated taking into account the amount of saturation of the irradiated signal in each adduct. The small number of detected NOE in this case allows us to use the two-spin approximation. A check has been performed on the validity of the two-spin approximation in the present case. By using reasonable estimates of the nixvalues from the X-ray data and of qx(j>from the interproton distances obtained here (Table 3), we concluded that the second term in Eqn 6 is indeed negligible in all cases. Then, using Eqns 7 and 8, the interproton distances were calculated, and their values are also reported in Table 3. It is noteworthy that the distances of both anions to the methyl protons corresponding to signals s and u are identical (0.42 nm and 0.35 nm, respectively). These values point to similar orientations of both anions in the cavity.

Two-dimensional NMR experiments At this point, it was necessary to assign signals s, u, and v to identify the protein residues located near the inhibitors in the cavity. Two-dimensional NMR is known to yield a considerable amount of structural information [48,49], and is regarded as the most powerful spectroscopic tool for proteins. In paramagnetic systems, the fast relaxation of the nuclei gives rise to a situation in which the magnetization transfer could be lost if not using appropriate delays in the pulse sequence. For this reason, reports have appeared using two dimensional experiments in paramagnetic molecules only in recent years, in particular for iron(II1) systems [SO - 521. Two-dimensional NMR of high spin Co(I1) systems is still in its early stages and

612

0

\I

ChaHlical shift (ppm)

Fig.4. NMR spectrum and NOE difference spectrum of the acetate adduct. (A) 600-MHz 300-K 'H-NMR spectrum of the acetate adduct of CoBCA 11; (B) NOE difference spectrum of the acetate adduct upon irradiation of the methyl acetate signal (1.95 pprn); (C) NOE difference spectrum of the CDJCOO- adduct of CoBCA 11 upon irradiation at 1.95 ppm. (D) Slice of a NOESY spectrum of the acetate adduct at 600 MHz with a mixing time of 20 ms, performed without presaturation of the acetate methyl signal. All the samples were in D 2 0 at pH 6.0. Table 3. Observed nuclear Overhauser enhancements,selective TIvalues and calculated interproton distances in the formate and acetate adducts of CoBCA 11. The NOE data are given relative to 100% saturation of the irradiated signal. values were measured with an error of f 10%. Values for rHHwere calculated from Eqn 6 , assuming a t, of 15 ns [54].

v'

F'

rH€I

-0.9 0.1% -1.4 0.2% -0.8+0.1%

+

ms 58.2 30.2" 54.5

nm 0.42 k 0.03 0.35 f 0.04 0.43k0.04

-0.8 & 0.1 Yo -3.0 f 0.3%

51.7 71.0

0.42 f 0.03 0.35 & 0.03

Adduct

Signal

NOE

Acetate adduct

s U V

Formate adduct

s U

a

Measured with an error of

k 20%.

a report on the perchlorate derivative of CoBCA I1 is available

POI.

The NOESY NMR experiments were performed on the acetate derivative in the 20 ppm to - 10 ppm spectral window. The NOESY map recorded with a recycle delay of 1 s and a mixing time t , of 100 ms (Fig. 5 A and B) exhibits a large number of connectivities between the paramagnetically shifted signals and peaks within the diamagnetic envelope. A cross peak between signals s and v and among these signals and a signal at - 1.85 ppm (signal p) are observed. There are also

cross peaks due to the dipolar interaction of the CH3 moieties with the methyl group of CH3COO-. The latter also gives a cross peak with signal u, consistently with the one-dimensional NOE spectrum. In addition, signals p, s and v give a large number of NOE connectivities, some of which may be attributable to spin diffusion that occurs with such a long mixing time. In fact, to avoid the buildup of the secondary NOE, a NOESY spectrum with shorter mixing time was recorded ( t , = 20 ms, Fig. 5C). In this spectrum, the same typical crosspeak pattern between signals p, s and v can still be seen. The cross peaks between signals s and v and the acetate methyl are not seen, because this experiment was performed with presaturation of the acetate methyl signal during the relaxation delay and the mixing time in order to avoid dynamic range problems. By performing the same experiment without presaturation of the acetate methyl signal, cross peaks with s, u and v are again apparent, of intensity comparable to those of s and v with p. This is a further proof that the ligandprotein NOE connectivities are primary. A slice of the abovementioned two-dimensional NOESY experiment, showing the acetate-protein connectivities, is reported in Fig. 4D. As we are now able to monitor only primary dipolar connectivities, the p-s-v pattern may correspond to the throughspace coupling of the geminal methyl groups of a valine residue between them and with their BCH proton. In order to identify inequivocally the scalar connectivities, a total correlation spectroscopy (TOCSY) experiment was attempted without success, possibly due to the large linewidths (150 Hz) of signals s and v. However, in another study on the perchlorate derivative [20], a COSY cross pattern of a valine has been detected, whose signals have been correlated by titration with those of the present acetate adduct, giving further support to this assignment. At this point, we are able to state that the acetate molecule bound to the metal ion orients its methyl moiety towards the hydrophobic pocket of the cavity. The possible residues that can be located in this region are Va1121, Va1143, Val207 and Leu1 98. Without ruling out the possibility that some connectivities may be lost in the diamagnetic portion of the spectrum, the cross-peak pattern points to the identification of a valine rather than a leucine residue. The two methyl groups (s and v) belonging to the same residue, should correspond to Va1207, since both Val143 and Val121 have one of their methyl groups too close to the metal ion to be detectable at 600 MHz. As to signal u, which is broader than s and v, it may be due to a methyl group of one of these closer valines. Taking into account that, in the isoenzyme HCA I, Val121 is replaced by an alanine [53],we performed NOE and NOESY spectra on the acetate adduct of CoHCA I. As shown in Fig. 6, all three characteristic NOE upfield are present, so that the NOE on signal u should correspond to Va1143. Considering that the signals of the acetate and the formate adducts are correlated by means of titration, we are now able to extend these assignments to the formate case. The formate proton gives NOE spectra with the same proton as acetate protons do. However, the distances are different. The formate proton is close to a methyl of Val143 and oriented towards one methyl of Va1207. The second methyl of Val207 (signal v) is now further away at a distance which can be estimated to be around 0.5 nm. This is consistent with the X-ray data [19]. If acetate were placed exactly as the formate, its methyl group would suffer a sizable steric hindrance with Trp209. Therefore, the acetate lies in a slightly modified position. This fact explains why the calculated distances from the protein residues are not shorter

61 3 V

T-

-?-

"I

P fI 0

I

t

0

I

ppm

I

I

8

I

6

I 4

I

2

I

I

0

chemical shift (ppm)

Fig.5. NOESY spectra of the acetate adduct of CoBCA 11. (A) NOESY spectrum of the acetate adduct of CoBCA I1 at 600 MHz. The experiment was performed with a mixing time of 100 ms, at 300 K and pH 6.0. (B) Expanded region (- 1.5/-5 pprn) of the precedent spectrum. (C) NOESY spectrum of the acetate adduct using a mixing time of 20 ms.

than for the formate. As the acetate is tilted, its methyl group locates equidistant to both methyls of Va1207. Concluding remarks

By means of 13C relaxation-rate measurement on 13C enriched formate, the binding of formate and acetate to the metal ion in CA in solution is confirmed. The estimated CoC distances are the same as found in carboxylate complexes. It is possible, therefore, that the metal-oxygen distances in the protein derivatives fall in a range of regular values. 'H-NMR spectra show that both acetate and formate bind in a similar pentacoordinated fashion. Through NOE experiments, it is clearly shown that the methyl group of acetate and the proton of formate are oriented towards the hydrophobic portion of the cavity, in particular, towards Val143 and Va1207. The carboxylate moiety of acetate has a

somewhat different tilt to allow more room for the methyl group in the hydrophobic pocket. In both cases, one oxygen is maintained at hydrogen-bonding distance from the N H of Thr199, as proposed from the X-ray structure of the formate derivative [19]. The distances from the ligand protons to three protein residues have been calculated. The resulting pictures are refined enough to proceed with a molecular-dynamics simulation which is in progress independently of these results. Calculations should be helpful in clarifying how many hydrogen bonds with terminal atoms of the ligands are able to weaken a coordinative bond. The use of two-dimensional NMR together with NOE experiments is shown to provide helpful hints in the assignment of a medium-size protein ( M , 30000) possessing a strong paramagnetic ion as high-spin cobalt(I1) ( S = 3/2). The paramagnetic center allows the discrimination of protons close to it and hence to the catalytic center from the bulk protons.

614

I

-4

-2

0

2

4

6

'

8

. 10

rm I

ppm

I

I

-4

chernical shift (ppm) Fig. 6. Expanded region (- 1.5/ - 6 ppm) of the 300 K NOESY spectrum of the acetate adduct of CoHCA I at 600 MHz.

More important, this approach has been shown to be valid to locate inhibitors within the cavity of enzymes. We would like to thank Prof. A. Liljas for kindly sharing with us .his crystallographic results. The Blood Center of the Polyclinic of Careggi, Florence, is acknowledged for providing human red cells for the purification of the HCA I isoenzyme. R. P. is grateful to Bruker spectrospin Italiana s. r. 1. for a research fellowship. A. J. V. thanks the International Centre for Genetic Engineering and Biotechnology (UNIDO) for a fellowship, as well as Fundacion Antorchas for further support.

REFERENCES 1. Lindskog, S. (1983) in Zinc enzymes (Spiro, T. G., ed.) pp. 77102, J. Wiley and Sons, New York. 2. Kannan, K. K., Notstrand, K., Fridborg, S., Lovgren, A,, Ohlsson, A. & Petef, M. (1975) Proc. Natl Acad. Sci. USA 72, 51 - 54.

3. Liljas, A., Kannan, K. K., Bergsten, P. C., Waara, I., Fridborg, K., Strandberg, B., Carlbom, U., Jarup, L., Lovgren, S. & Petef, M. (1972) Nature New Biol. 235, 131-137. 4. Pocker, Y. & Sarkanen, S. (1978) Adv. Enzymol. 47, 149-188. 5. Silverman, D. N. & Lindskog, S. (1 988) Arc. Chem. Res. 21,30 36. 6. Lindskog, S. & Coleman, J. E. (1973) Proc. Natl Acad. Sci. USA 70,2505 - 2508. 7. Lindskog, S. & Malmstrom, B. G. (1962) J . B i d . Chem. 237, 1129 - 1137. 8. Bertini, I., Canti, G., Luchinat, C. & Scozzafava, A. (1978) J . Am. Chem. Soc. 100,4873-4877. 9. Banci, L., Bertini, I., Luchinat, C. & Moratal Mascarell, J. M. (1990) in Enzymatic and model carboxylation and reduction reactions f o r carbon dioxide utilization (Aresta, M. & Schloss, J. V., eds) pp. 181 - 197, Reidel Publ. 10. Khalifah, R. G. (1971) J. Biol. Chem. 246, 2561 -2573. 11. Eriksson, A. E., Jones, T. A. & Liljas, A. (1989) Proteins 4,274282. 12. Bertini, I. & Luchinat, C. (1983) Arc. Chem. Res. 16, 272-279.

615 13. Mangani, S., & Hakansson, K. (1992) Eur. J . Biochem., in the press. 14. Vidgren, J., Liljas, A. & Walker, N. P. C. (1990) Znt. J . B i d . Macromol. 12, 342 - 344. 15. Eriksson, A. E., Kylsten, P. M., Jones, T. A. & Liljas, A. (1989) Proteins 4 , 283 - 293. 16. Lindahl, M., Svensson, A. & Liljas, A. (1992) Proteins, in the press. 17. Bertini, I., Luchinat, C., Pierattelli, R. & Vila, A. (1992) Znorg. Chem., in the press. 18. Bertini, I., Canti, G. & Luchinat, C. (1981) Znorg. Chim. Acta 56, 99-107. 19. Hakansson, K., Carlsson, M., Svensson, A. & Liljas, A. (1992) J. Mol. Biol., in the press. 20. Banci, L., Dugad, L. B., Keating, K. A,, La Mar, G. N., Luchinat, C. & Pierattelli, R. (1992) Biophys. J., in the press. 21. Khalifah, R. G., Strader, D. J., Bryant, S. H. & Gibson, S. M. (1977) Biochemistry 16, 2241 -2241. 22. Hunt, J. B., Rhee, M. J. & Storm, C. B. (1977) Anal. Biochem. 55,614-617. 23. Inubushi, T. & Becker, E. D. (1983) J. Magn. Reson. 51, 128133. 24. Vold, R. L., Waugh, J. S., Klein, M. P. & Phelps, D. E. (1968) J . Chem. Phys. 48, 3831 - 3835. 25. Thanabal, V., De Ropp, J. S. & La Mar, G. N. (1987) J . Am. Chem. Soc. 109,265 -272. 26. Banci, L., Bertini, I., Luchinat, C., Piccioli, M., Scozzafava, A. & Turano, P. (1989) Inorg. Chem. 28,4650-4656. 27, Freeman, R. & Hill, H. D. W. (1971) J . Chem. Phys. 48, 33673370. 28. Bauer, C . , Freeman, R., Frenkiel, T., Keeler, J. & Shaka, A. J. (1984) J . Magn. Reson. 58, 442-451. 29. Macura, S., Wuthrich, K. & Ernst, R. R. (1982) J . Mugn. Reson. 47,351 -357. 30. Marion, D. & Wuthrich, K. (1982) Biochem. Biophys. Res. Commun. 113, 967-971. 31. Bertini, I. & Luchinat, C. (1986) N M R ofparamagnetic molecules in biological systems, Benjamin Cummings Publishing Company, Inc., Menlo Park, CA. 32. Cohn, M. & Mildvan, A. S. (1970) A h . Enzymol. 33, 1 - 12.

33. Dwek, R. A. (1973) Nuclear magnetic resonance in biochemistry, Oxford University Press, Oxford. 34. Solomon, I. & Bloembergen, N. (1956) J . Chem. Phys. 25, 261 266. 35. Bertini, I., Luchinat, C. & Scozzafava, A. (1977) J. Chem. Soc. Dalton Trans., 1962- 1965. 36. Carrell, C. J., Carrell, H. L., Erlebacher, J. & Gluster, J. P. (1988) J . Am. Chem. Soc. 110, 8651 -8656. 37. Horrocks, W. D. Jr, Ishley, J. N. & Whittle, R. R. (1982) Znorg. Chem. 21, 3265 - 3269. 37a. Horrocks, W. D. Jr, Ishley, J. N. &Whittle, R. R. (1982) Znorg. Chem. 21,3270 - 3274. 38. Kirchner, C. & Krebs, B. (1987) Znorg. Chem. 26, 3569-3576. 39. Bertini, I., Lanini, G., Luchinat, C., Haas, C., Maret, W. & Zeppezauer, M. (1987) Eur. Biophys. J . 14,431 -439. 40. Bertini, I., Monnanni, R., Pellacani, G. C., Sola, M., Vallee, B. L. & Auld, D. S. (1988) J . Inorg. Biochem. 32, 13 -20. 41. Led, J. J. & Grant, D. M. (1977) J. Am. Chem. Soc. 99, 58455858. 42. Henry, B., Boubel, J. C. & Delpuech, J. J. (1986) Znorg. Chem. 25,623 - 631. 43. Banci, L., Bertini, I., Luchinat, C., Donaire, A., Martinez, M.J. & Moratal Mascarell, J. M. (1990) Comments Znorg. Chem. 9,245 - 261. 44. Bertini, I. & Luchinat, C. (1984) Adv. Znorg. Biochem. 6,71- 111. 45. Gueron, M. (1975) J . Mugn. Reson. 19, 58-66. 46. Vega, A. J. &Fiat, D. (1976) Mol. Phys. 31, 347-355. 47. Neuhaus, D. &Williamson, M. P. (1989) The nuclear Overhauser effect in structural and conformational analysis, VCH Publishers, New York. 48. Ernst, R. R., Bodenhausen, G. & Wokaun, A. (1987) Principles of nuclear magnetic resonance in one and two dimensions, Clarendon Press, Oxford. 49. Bax, A. (1989) Annu. Rev. Biochem. 58,223-256. 50. Yu, L. P., La Mar, G. N. & Rajarathnam, K. (1990) J . Am. Chem. SOC.112,9527-9534. 51. Emerson, S . D. & La Mar, G. N. (1990) Biochemistry 29, 15451556. 52. Bertini, I., Briganti, F., Luchinat, C., Messori, L., Monnanni, R., Scozzafava, A. & Vallini, G. (1991) FEBSLett. 289, 253-256. 53. Coleman, J. E. (1973) Znorg. Biochem. I , 488-548. 54. L a n k A. & Navon, G. (1974) Biochim. Biophys. Acta 341, 6574.

The interaction of acetate and formate with cobalt carbonic anhydrase. An NMR study.

The interaction of formate and acetate ions with cobalt-substituted carbonic anhydrase (CA) has been investigated through 13C-NMR and one-dimensional ...
818KB Sizes 0 Downloads 0 Views