Crystal Structures and Solution Studies of Oxime Adducts of Mitochondrial Aspartate Aminotransferase

Eur. J. Biochem. 236, 1025-1032 (1996) 0 FEBS 1996

Crystal structures and solution studies of oxime adducts of mitochondrial aspartate aminotransferase Zora MARKOVIC-HOUSLEY I, Tilman SCHIRMER’, Erhard HOHENESTER’, Alex R. KHOMUTOV2, Radii M. KHOMUTOV ’, Marat Y. KARPEISKY’, Erika SANDMEIER“, Philipp CHRISTEN4 and Johan N. JANSONIUS’

I Abteilung Strukturbiologie, Biozentrum der Universitat Basel, Basel, Switzerland Institute of Organic Chemistry, Russian Academy of Sciences, Moscow, Russia Institute of Molecular Biology, Russian Academy of Sciences, Moscow, Russia Biochemisches Institut der Universitat Zurich, Zurich, Switzerland

(Received 6 OctoberAS December 1995) - EJB 95 1632/4

The interaction of mitochondrial aspartate aminotransferase with hydroxylamine and five derivatives (in which the hydroxyl hydrogen is replaced by the side chain of naturally occurring amino acids) was investigated by X-ray diffraction as well as by kinetic and spectral measurements with the enzyme in solution. The inhibitors react with pyridoxal 5‘-phosphate in the enzyme active site, both in solution and in the crystalline state, in a reversible single-step reaction forming spectrally distinct oxime adducts. Dissociation constants determined in solution range from lo-’ M to M depending on the nature of the side-chain group.

The crystal structures of the adducts of mitochondrial aspartate aminotransferase with the monocarboxylic analogue of L-aspartate in the open and closed enzyme conformation were determined at 0.23- nm and 0.25-nm resolution, respectively. This inhibitor binds to both the open and closed crystal forms of the enzyme without disturbing the crystalline order. Small differences in the conformation of the cofactor pyridoxal phosphate were detected between the structures of both oxime complexes and the 2- methylaspartate adduct. The crystal structures indicate that the interaction between the w-carboxylate of the inhibitor and Arg292* of the neighbouring subunit is mainly responsible for the attainment of near- coplanarity of the aldimine bond with the pyridine ring in the oxime adducts. Studies with a fluorescent probe aimed to detect shifts in the open/closed conformational equilibrium of the enzyme in oxime complexes showed that the hydroxylamine-derived inhibitors, even those containing a carboxylate group, do not induce the ‘domain closure’ in solution. This is probably due to the absence of the a-carboxylate group in the monocarboxylic hydroxylamine-derived inhibitors, emphasizing that both carboxylates of the substrates L - A s ~ and L - G ~ are essential for stabilizing the closed form of aspartate aminotransferase.

Ke-ywords: B, enzyme ; oxime inhibitor; ligand binding ; conformational change ; crystal structure.

Hydroxylamine and five analogues of type H,NO-R, where R represents the side chain of naturally occurring L-amino acids [H,N-0-CH, (H,N-0-Ala), HZN-O-CH,COOH (H,N-0-Asp),

Phe)] are potent inhibitors of aspartate aminotransferase (Asp-

Correspondence to J. N. Jansonius, Abteilung Strukturbiologie, Bio- zentrum der Universitat Basel, Klingelbergstrasse 70, CH-4056 Basel, Switzerland

H,N-OCH,CH,COOH (H,N-O-Glu), HZN-OCH,-C,H, (H,N-0-

_____

Fux: +41 61 2672109. Ahbra idom. (m)AspAT, (mitochondrial) aspartate aminotransfer-

ase : pyridoxal-P, pyridoxal 5’-phosphate; pyridoxamine-P, pyridoxamine 5’-phosphate: R-group, substituent replacing hydrogen in hydroxylamine; H,N-0-Ala, H,N-0-CH,; H,N-0-Asp, H,N-0-CH,COOH; H,N-0-Glu, H,N-O-CHZ-CH2COOH; H,N-0-Phe, H,N-0-CH,-C,H,; HOL, crystalline holoenzyme at pH 7.5 (unprotonated internal aldimine; PDB entry 7AAT); LPH, crystalline holoenzyme at pH 5.1 (protonated internal aldimine; PDB entry 8AAT); 2-MeAsp, 2-methylaspartate; AMA, crystalline enzyme complex with 2-MeAsp (external aldimine an- alog; PDB entry IAMA); OxOP, oxime formed in the triclinic crystal form (open enzyme conformation); OxCL, oxime formed in the orthorhombic crystal form (closed enzyme conformation); Arg292“ : the :$ indicates the residue from the second subunit.

6zzynes. Aspartate aminotransferase (EC 2.6.1 . l ) ; malate dehydrogenase (EC 1.1.1.37).

AT) and other pyridoxal 5’-phosphate (pyridoxal-P)-dependent enzymes (Karpeisky et al., 1963; John et al., 1978, and references therein). They react readily with the cofactor of the enzyme to form oximes (Scheme 1). The monocarboxylic hydroxylamine derivatives mimic amino acids that lack the a-carboxylate group (Khomutov et al., 1961, 1963, 1987; Karpeisky et al., 1963). The oxime with aminooxyacetate, the aspartate analogue, has been studied particularly well. It strongly binds to AspAT (Kd--10~7-10-8 M ; John et al., 1978; Raunio et al., 1984). Analysis of the interactions of H,N-0-Asp with aspartate aminotransferase may be expected to provide information about the role of the a-carboxylate group in the binding of the natural substrate L-aspartate.

Mitochondria1 AspAT (mAspAT) is an a, dimer which catal- yses the reversible transfer of the a-amino group of aspartate or glutamate to 2-oxoglutarate and oxalacetate, respectively, in a ping-pong mechanism. Each subunit of AspAT consists of a large pyridoxal 5’-phosphate(pyridoxal-P)-binding domain, a small domain and an N-terminal segment which non-covalently attaches to the adjacent subunit (for reviews, see Jansonius et al., 1985; Jansonius and Vincent, 1987). In the unliganded pyridoxal-P enzyme, the cofactor is covalently linked through an aldimine double bond to Lys258 (‘internal’ aldimine, see

1026

Scheme 1.

Lys 258 < Internal aldimine ( 1 m a 360nm)

R ‘ 0

Maleate d 1

0

Markovid-Housley et a1

H

Michaelis complex ( max 440nm)

Oxime ( 1 max 380nm)

External aldimine AMA ( 1 max 430nm)

Scheme 1). The binding of the substrate and its transformation into product is accompanied by changes in the conformation of the enzyme (Birchmeier et al., 1973; Gehring and Christen, 1975, 1978; Sandmeier and Christen, 1980; Pfister et al., 1985; Picot et al., 1991 ; McPhalen et al., 1992a; Hohenester and Jan- sonius, 1994). Two distinct conformations of mAspAT are found in the crystalline state. The ‘open’ conformation is characteristic of the unliganded enzyme form (Ford et al., 1980; McPhalen et al., 1992b). The ‘closed’ conformation is induced and maintained by the binding of dicarboxylic C, and C, substrate analogues and inhibitors (Picot et al., 1991 ; McPhalen et al., 1992a; Malashkevich et al., 1993 ; Hohenester and Jansonius, 1994) such as maleate (a non-covalently binding C, substrate analogue) and 2-methylaspartate (2-MeAsp) which binds covalently to pyridoxal-P through an aldimine linkage (Fasella et al., 1966; Hammes and Haslam, 1968). Transition from the open to the closed conformation involves rigid-body rotation of the small domain towards the large domain closing off the active-site pocket from the solvent.

In an earlier study, absorption, linear and circular dichroism spectroscopy were used to investigate the spectral properties of mAspAT oximes in solution and in the crystalline state (Del- baere et al., 1989). The results indicated that the orientations of the cofactor in these oximes are similar to that found in the external aldimine adduct with 2-MeAsp (Vincent et al., 1984; Picot, 1987) suggesting a similar binding mode for the inhibitor. However, rnonocarboxylic H,N-0-Asp, unlike the dicarboxylic inhibitor 2-MeAsp, could not induce domain closure in cocrystallization experiments. This led to the conclusion that the pres-

( E m J . Biochem. 236)

ence of an a-carboxyl,ate group is important for productive substrate binding and closure of the active-site cleft.

The present study is an extension of the previous work. We report the crystal structures of the mAspAT oxime complexes with H,N-0-Asp in the open and closed conformations. The ef- fects of size and charge of various R groups on the kinetics of oxime formation were investigated. The openklosed confcrrma- tional equilibrium of the mAspAT oxime complex in solution was probed by a covalently bound fluorescent label and bq limited proteolysis experiments.

EXPERIMENTAL PROCEDURES

mAspAT was prepared from chicken heart according to Gehring et al. (1977). The protein was judged to be pure on the basis of SDS/PAGE and analytical ultracentrifugation. The enzyme concentration was determined photometrically at 280 nm using a molar absorption coefficient of the subunit t: = 7.0X lo4 M-’ cm-’. Unless stated otherwise, the experiments were carried out in 20 mM sodium phosphate, pH 7.5. H! droxylamine was purchased from Merck and its derivatives were synthesized as described earlier (Delbaere et al., 1989). 2- Methyl-DL-aspartate was from Sigma.

The activity of mAspAT was determined in a coupled assay with malate dehydrogenase and NADH. The standard assay mixture contained 20 mM L-aspartate, 2 mM 2-oxoglutarate, 0.2 mM NADH, 2 IJ/ml of malate dehydrogenase in l ( 0 mM Hepes, pH 7.5. Enzyme activity in the presence of inhibitor was determined as follows. Oxime adducts were formed by incubating the enzyme (13 rtM) with a large excess (500 pM) of inhibitor in 20 mM sodium phosphate, pH 7.5, for 2 h at 22°C. To avoid possible reactivation by dilution, the residual activity was assayed after 2 h by adding 0.2 ml of a concentrated assay mixture to 1 ml of enzyme solution.

Kinetic measurements. The association rate constants k,, of the inhibitors were determined by measuring the incrcase in absorbance at wavelengths slightly higher than A,,,,, (in the 380- 390-nm range) as a function of time according to John et al. (1978). These authors have shown that the same rate constant ( k + ] ==* 400 M I s-I) was obtained if the inhibitor concentration was low (comparable to that of the enzyme; reaction time in minute range) and if a 300-fold molar excess of inhibitor over enzyme was used (reaction time in millisecond range). Here, we chose inhibitor and enzyme concentrations such that the reaction times were of the order of minutes. This allowed the use of conventional mixing techniques in the spectrophotometer. The kinetic measurements were performed at several inhibitor concentrations using an excess of inhibitor over enzyme concentrations (1 -14 pM) with constant [I]/[E] ratios (7 for H,NOH, H,N-0-Asp and H,N-0-Glu, 120 for H,N-0-Ala and 109 for H,N-0-Phe). For each inhibitor concentration, the time--course was recorded. The apparent pseudo-first-order rate constants, k,,,,, were calculated by non-linear least squares regression fitting of a single exponential function [A;,,, = A;:;;; (1 -e-‘~~l,~‘) I to the absorbance data. Single relaxation times were obtained tor each inhibitor and all concentrations used. The second order rate constants ( k + ] ) were calculated from the slopes of the straight lines obtained by plotting kc>,,, against the inhibitor concentration

To follow the dissociation reaction of the oxinie Ldducts. stoichiometric amounts of enzyme and inhibitors were incubated for 2 h at 22°C. Recording of the dissociation reaction was started by addition of excess cysteine sulphinate (4 mM: John et al., 1978). This reagent reacts with unliganded pyridoxal-P enzyme in less than 10 s. forming the pyridoxamine-P form of

(kc,,,, = ki-1 . LIl+k- 1).

Markovit-Housley et al. ( E m J. Biochem. 236) 1027

mAspAT (absorption maximum at 330 nm) which does not react with hydroxylamine derivatives. Thus, the (slow) appearance of an absorption maximum at 330 nm, indicative of the pyridoxamine-P form of the enzyme, is a measure for the dissociation of the inhibitor from its adduct with AspAT. The dissociation rate constants ( k - , ) were calculated by fitting a theoretical curve [A:,,, , , ,~ = AT;;,,,,, (1 - e-'-l')] to the experimental data. The increase of A,,,, with time followed pseudo-first-order kinetics.

Fluorescence experiments with mAspAT and its oxime adducts. Cysl66 was selectively labelled with the fluorescent probe monobromotrimethyl-ammoniobimane as described earlier (Kosower et al., 1979; Picot et al., 1991). 1 mol tri- methylammoniobirnane incorporated/mol monomer. The emission maximum of bimane-labelled mAspAT (7 pM subunit concentration; excitation at 380 nm) in 20 mM sodium phosphate, pH 7.5, is at 473 nm. With the same excitation wavelength, the emission of unlabelled mAspAT is negligible. Oxime adducts were prepared by incubating bimane-labelled mAspAT (7 pM subunit concentration in 20 mM sodium phosphate, pH 7.5) with the various inhibitors (1 mM) at room temperature in the dark overnight. Excess inhibitor was removed by gel filtration. Fluo- rescence of bimane-labelled mAspAT was measured on a Hi- tachi-Perkin Elmer MPF-2A spectrofluorimeter equipped with a 150 W Xenon lamp. The excitation band width was set at 7 nm (0.90-mm slit width) and the emission band width at 10 nm (1.29-mm, slit width).

Proteolytic cleavage of mAspAT oxime adducts by trypsin. Limited trypsin digestion of mAspAT and its adducts with the various inhibitors was carried out as described by Sandmeier and Christen (1980).

Crystallographic methods. Triclinic crystals of mAspAT (holoenzyme, space group P t ; McPhalen et al., 1992b) and orthorhombic crystals (Michaelis complex analogue with maleate, space group C222,; Jansonius et al., 1985; Picot et al., 1991) were grown as described elsewhere (Thaller et al., 1981 ; Picot, 1987). Crystals of the oxime adduct with H,N-0-Asp were prepared as follows. Triclinic crystals were soaked overnight in the presence of a stabilizing solution [20 % poly(ethy1ene glycol) 4000 in 20 mM sodium phosphate, pH 7.51 containing 5 mM H,N-0-Asp. The maleate bound in the orthorhombic crystals was replaced by H,N-0-Asp in a step-wise fashion by soaking the crystals in increasing concentrations of H,N-0-Asp over 24 h. Completion of the reaction was determined in both cases by monitoring the absorption spectrum of small crystals (Eichele et a]., 1978) treated similarly. The 360-nm and 430-nm absorption peaks, characteristic of unprotonated and protonated internal aldimine of the holoenzyme and the mAspAT-maleate complex, respectively, were shifted to 380 nm, the absorption maximum characteristic for the H,N-0-Asp oxime.

X-ray diffraction data were collected on a FAST area detec- tor and processed with the MADNES software (Messerschmidt and Pflugrath, 1987). The CCP4 program package (Collabora- tive Computing Project No. 4, 1994) was used for scaling and merging. The triclinic data set (OxOP) to 0.23-nm resolution comprises 25 752 independent reflections. Because of low re- dundancy, scaling of the individual batches with the program ROTAVATA was performed against a reference consisting of calculated structure factors obtained from the model of the holoenzyme structure (HOL) at pH7.5 (McPhalen et al., 1992b). After scaling, the agreement between symmetry-related reflections, R,,,,,, was 4.1 %. The 11 838 independent reflections of the 0.25-nm orthorhombic data set (OxCL) scaled with an RIyn, of 5.6%.

The oxime adducts proved to be isomorphous with the parent mAspAT structures. Therefore, (2F0-F,), a, electron-density n ~ p s in which the cofactor pyridoxal-P and the maleate ion (in

Table 1. Spectroscopic and catalytic properties of mAspAT in the presence and absence of inhibitors. The enzyme (13 nM, in 20 mM sodium phosphate, pH 7.5) and the indicated inhibitor (0.5 mM) were incubated in 1 ml for 1 h at 20°C. The residual activity after 1 h was assayed by adding 0.2 ml of the concentrated assay mixture so that standard assay conditions were maintained. Activity was remeasured after dialysis of the oxime adducts against 20 mM sodium phosphate, pH 7.5, for 24 h at 4°C.

Inhibitor Absorption/activity Activity after after incubation with inhibitor subsequent

L, activity .~ - dialysis

nm %

- None 356 100 HZN-O-Asp 383 < 0.5 7 H,N-O-Glu 380 2 23 H,N-0-Phe 378 6 54 H,N-0-Ala 379 8 76 Hydroxylamine 372 < 0.5 74

OxCL) had been excluded from the phasing model, immediately revealed density for the pyridoxal-P oxime adduct with H,N-0- Asp into which the oxime model could be built. The R-factors for these models were 17.5% (OxOP) and 24.1% (OxCL). Least-squares refinement, using the program PROLSQ quickly converged to R-factors of 10.9% (OxOP) and 14.9% (OxCL). The coordinates of the OxOP and OxCL structures have been deposited with the Brookhaven Protein Data Bank, with entry codes 1 0 x 0 and 1 OXP, respectively.

RESULTS

Reversibility of enzyme inhibition with hydroxylamine and hydroxylamine-derived inhibitors. The pK of protonation of the mAspAT internal aldimine in mAspAT is 6.2. The unprotonated form has an absorption maximum at 356 nm, the protonated form at 430 nm. The hydroxylamine-derived inhibitors H2N-O- Ala, H2N-O-Asp, H,N-0-Glu, H,N-0-Phe and hydroxylamine react with both forms of mAspAT. The resulting oximes are characterised by an absorption maximum near 380 nin (Table 1). To quantify the extent of binding of these inhibitors. the enzymic activity of preformed oxime adducts was measured. The activity was close to zero in the presence of H,N-0-Asp and hydroxylamine. With H,N-0-Glu, H,N-0-Ala and H,N-0-Phe the re- maining activities were less than 10% of the initial (Table 1). The substrates L-aspartate and 2-oxoglutarate (at the concentrations used in the activity assays) prevented the inhibition of the enzyme by the hydroxylamine derivatives, confirming that the inhibitors and substrates compete for the same binding site.

Reactivation of the enzyme was monitored by measuring enzymic activity and by following the absorbance at 356 nm. Ac- tivity was partially recovered upon dialysis of the preformed oxime adducts against 20 mM sodium phosphate, pH 7.5. The highest reactivation of the enzyme was obtained from adducts with carboxylate-free inhibitors and the lowest with H,N-0-Asp (Table 1). This result indicates strong binding of H,N-0-Asp and is consistent with the preference of mAspAT for four-carbon substrates. The absorption band at 356 nm did not reappear in the case of the carboxylate-containing inhibitors during several days of dialysis. The H,N-0-Asp and H,N-0-Glu oxime adducts thus were judged to be amenable to X-ray studies. With H,N-0- Ala and H,N-0-Phe, complete reversal of the spectral properties

1028 Markovid-Housley et al. (Eur: J. Biochern. 236)

Table 2. Kinetic parameters of mAspAT-inhibitor interactions.

M-Is-I s ’ M

H,N-O- Asp 412 l . lxlO-s 2.7X10 ’ H,N-O-Glu 835 4.9 x 1 5 . 9 ~ 1 0 - ~

H,N-O- Ala 13.4 Hydroxylamine 730 9.8X 10- 1.3X lo-‘

- - H,N-0-Phe 9.7 - -

Table 3. Fluorescence properties of bimane-labelled aspartate aminotransferase in liganded and unliganded form. For conditions see Experimental Procedures.

Ligand intensity Relative fluorescence

None 1 .OO Maleate 0.28 2-Methylaspartate 0.48 HLN-0- ASP 0.88 H2N-O-Glu 0.86 H,N-0-Phe 1.09 H,N-0-Ala 1.01 Hydroxy lamine 0.90

of the pyridoxal-P enzyme and considerable recovery of enzymic activity were observed, suggesting much weaker binding of carboxylate-free inhibitors. If H,N-0-Ala and H,N-0-Phe were incubated with the enzyme for longer than 5 h, a slow con- tinuous decrease of absorbance at 380nm was observed concomitant with appearance of an absorption band at 330-340 nm. This is due to diffusion of the pyridoxal-P oxime out of the active site, as demonstrated earlier (Delbaere et al., 1989). Crys- tallographic studies of H,N-0-Ala and H,N-0-Phe complexes were therefore not possible. The oxime complex with hydroxylamine was relatively stable, as absorption at 372 nm persisted for several days and only a very small shoulder at 340 nm was observed.

Kinetic studies. Table 2 lists the association and dissociation rate constants ( k , I and k - , , calculated as described in Experi- mental Procedures) as well as the dissociation equilibrium constants (&, calculated as the ratio of k- , /k , , ) for the reactions of the hydroxylamine-derived inhibitors with mAspAT. For H,N- 0-Ala and H,N-0-Phe, no accurate estimates of k - , could be obtained because of the occurrence of two concomitant pro- cesses, formation of pyridoxamine-P mAspAT in the reaction of regenerated pyridoxal-P mAspAT with cysteine sulphinate and dissociation of the pyridoxal-P oxime moiety, which both cause an increase of ahorbance in the 330- 340-nm range.

Conformation of the enzyme in solution in the presence and absence of inhibitors. The changes in the relative fluorescence yield of bimane-labelled Cys166 of mAspAT have been shown to be a sensitive indicator for detecting shifts in the openklosed conformational equilibrium of the enzyme (Sandmeier and Christen, 1984). The relative fluorescence yield varies between 0.3 and 0.6 for the closed enzyme forms compared with the value of 1 in the open form (Picot et al., 1991). The relative fluorescence yield was measured with the unliganded enzyme and upon prior incubation with the inhibitors (Table 3). Only the values obtained for the maleate-liganded and 2-MeAsp-liganded enzyme (0.28 and 0.48, respectively) are indicative of the closed

Table 4. Cofactor conformation and orientation. For nomenc tature see Scheme 2. In the structures OxOP, HOL and LPH the enzyme ih in the open conformation with two crysta~lographically independent sub- units. In OxCL and AMA, the enzyme is in the closed conformation with one subunitlasymmetric unit. Cofactor orientation is defined JS the orientation of a line connecting N1 and C4 with respect to its orientation in HOL.

Dihedral angle Value for angle in

OxOP OxCL AMA HOL LPH

Deg.

N-C4’-C4-C3 -381-28 -16 -8 76/87 0/12 4 OS’-C5’-C5-C4 51/49 57 43 -4.5-47 -1b-12 0 P-O5‘-C5’-CS 1631156 148 161 -1751-171 1691175 y 02-P-O5’-C5’ 24/27 19 27 82/83 61/52

Cofactor orientation 28/27 27 25 010 1 0/0

Table 5. Lengths of potential H-bonds between cofactor-ligand moiety and the protein in OxOP and in OxCL (this work) and in AMA (POB entry IAMA). In OxOP, the values in parentheses refer to the second, crystallographically independent subunit.

Cofactor-ligand Protein group H-bond length moiety in enzyme form

OxOP OxCL AMA

nm

a-Carboxylate 0 1 Arg386 NH1 - - 0.30 oxygen 0 2 Arg386 NH2 - - 0.31

w-Carboxylate 0 1 Arg292* NH1 0.36 (-) 0.32 0.30 oxygen 0 2 Arg292* NH2 - (0.47) 0.25 0.30

Pyridoxal-P 03’ Tyr225 OH 0.35 (0.26) 0.27 0.28 03’ As11194 ND2 0.25 (0.27) 0.23 0.24 N4’” pyridoxal-P 03’b 0.33 (0.30) 0.32 0.24 N4’d Tyr225 OH 0.40 (0.37) 0.39 0.34

Oxime 0 Lys258 NZ 0.30 (0.31) 0.34 --

0 Wat767 - (-) 0.26 -

“ Oxime nitrogen (OxOP, OxCL) or aldimine nitrogen (AMA) Intramolecular distance.

form, whereas the oximes formed with hydroxylamine and hydroxylamine-derived inhibitors all have relative fluorescence yields near 1, indicative of the open enzyme conformation. It is noteworthy that H,N-0-Asp, H,N-0-Glu and hydroxylamine, which bind more strongly to the enzyme than H,N-0-Ala and H,N-0-Phe (Table 2), have slightly lower fluorescence yields ( ~ 0 . 9 ) compared with those of H,N-0-Ala and H,N--0-Phe (=1).

The second method used to probe the enzyme conformation was based on the observation that mAspAT in the open conformation is cleaved selectively by trypsin at the peptide bonds after Arg26 and Lys31, yielding a new band of 42000 instead of 45 000 Da on SDS/PAGE. In contrast, the closed form of the enzyme is cleaved rnarkedly more slowly by trypsin under similar experimental conditons (Sandmeier and Christen, 1980). The electrophoretic patterns of digested oxime complexes were iden- tical to that of the unliganded enzyme, thus demonstrating that the oxime adducts of the enzyme remain in an essentially open conformation (data not shown).

MarkoviC-Housley et al. ( E m J . Biochem. 236)

Scheme 2.

1029

H+

8"1 A

$'

B f' f"

C

T 015 T D15

Fig. 1. The active-site regions of AMA (a), OxCL (b) and OxOP (c). The large domains (residues 48-325) are oriented in the same way for all three structures. The coenzyme-inhibitor complexes are shown to- gether with some selected amino acid residues. Hydrogen bonds are de- noted with dashed lines. In the OxOP structure, the side chain of Arg292" is disordered and only one of several possible conformations is shown here. A bound water molecule in the OxCL structure is drawn as a double circle.

Crystal structures. Formation of the mAspAT-H,N-0-Asp oxime adduct in the closed crystal form (OxCL structure; see Ex- periniental Procedures) produced only insignificant changes in cofactor orientation and conformation in the active site. How- ever, in OxOP, binding of H,N-0-Asp to HOL decreases the torsion angle x from approximately 80" to approximately -30" ;

Fig.2. Stereo views of (F,-F,) difference maps contoured at 1.5 (F

around the oxime structure of pyridoxal-P with H,N-0-Asp. The F, structure factors have been calculated from the final models with the cofactor-oxime omitted. Carbon atoms are coloured in yellow, oxygens in red, nitrogens in blue and phosphorus in magenta. (a, top) The cofactor -H,N-0-Asp oxime as found in the OxCL structure, (b, bottom) idem for one subunit of the OxOP structure (the non-crystallographically related cofactor shows electron density of similar quality).

the oxime double bond (C4'=N4') is closer to coplanarity with the pyridine ring while the cofactor is tilted by about 30" (Table 4). The pyridoxal-P ring tilt is measured with the holoenzyme structure at pH 7.5 as reference. Definitions of the variable dihedral angles in the pyridoxal-P cofactor are given in Scheme 2.

Table 5 provides contact distances for the interactions of the oximes in OxOP and OxCL and of 2-MeAsp complexed with the enzyme (AMA) with selected active-site residues. In AMA, the a-carboxylate and w-carboxylate groups of 2-MeAsp form double-hydrogen-bonded ion pairs with the positively charged guanidinium groups of Arg386 and Arg292*, respectively (Fig. 1 A). The o-carboxylate group of H,N-0-Asp in OxCL interacts with Arg292" in a manner very similar to that of the w- carboxylate group of 2-MeAsp (Fig. 1 B). The oxime carboxylate group of OxOP is in essentially the same position as in OxCL (Fig. 1C). However, the interaction with Arg292* is much weaker. Arg292* forms a salt bridge with Asp15 in the unliganded open structure (HOL) of the holoenzyme, but is partly disordered in the OxOP structure, where it adopts two alternative conformations, of which only the most occupied con- former is shown i n Fig. 1C. Stereo views of cofactor-oxime models in OxOP and OxCL within their respective (F,-F,) difference electron densities (Fig. 2) give an indication of the qualities of these structures. In Fig. 3, selected features within the active sites of the structures OxOP, OxCL, HOL and AMA are super- imposed.

H,N-0-Asp is lacking the a-carboxylate group which in the natural substrate or in 2-MeAsp interacts with Arg386 in the

1030 MarkoviC-Housley et al. (Eul: J. Biochern. 236)

Fig.3. Stereo view of a superposition of the active-site regions of the unliganded enzyme (HOL; blue), the external aldimine with 2- methylaspartate (AMA; yellow), OxOP (green) and OxCL (red). A water molecule in the OxCl structure (near the m-carboxylate in Ah4A) is indicated by an orange sphere. The coenzyme ring plane has the same orientation in AMA, OxOP and OxCL.

closed form of the enzyme. In OxCL its position is occupied by a water molecule mediating an indirect contact between the oxime oxygen and Arg386 (Fig. 1 B). No such interaction is seen i n OxOP because Arg386, from the small domain, is too distant from the cofactor in the open crystal form.

DISCUSSION

Comparison of enzyme inhibition by 2-methylaspartate and hydroxylamine-derived inhibitors. The dissociation constants ( K J of the oxime complexes in solution vary between M and lo-' M, depending on the R-group (Table 2), the nature of which influences the association and dissociation rate constants. Oximes formed with the carboxylate-containing inhibitors H2N- 0-Asp and H,N-0-Glu are more stable than oximes formed with carboxylate-free inhibitors, H,N-0-Ala and H,N-0-Phe. The difference must be due to the stabilizing interaction of the LU-

carboxylate group with Arg292*, which anchors the inhibitor to the protein. In the absence of a carboxylate group, the pyridoxal- P oxime diffuses out of the enzyme active site.

Aspartate aminotransferase reacts with hydroxylamine-derived inhibitors H2N-OR in a slow, reversible single-step reaction to produce tight oxime complexes. In contrast, the formation of the enzyme-substrate aldiniine with asp or L-Clu via the Michaelis complex is a fast, two-step reaction. The weak binding of natural substrates, reflected in values of k - I = 10'- 10' s~', is required for efficient catalysis. An initial binding complex (Michaelis complex) observed with the substrates L- Asp and L-GIu (Fasella and Hammes, 1967) and with the inhibitor 2-MeAsp (Hammes and Haslam, 1968) was not detected in the reaction with H,N-0-Asp (John et al., 1978; this study).

The rate of enzyme-ligand complex formation depends on two main factors: the chemical reactivity of the compound to- ward the enzyme and the molecular complementarity (fit) between the ligand and thc enzyme. The driving force for the binding of asp and ~ G l u to AspAT is the high molecular complementarity of the natural substrate and the enzyme, reflected, for cytosolic AspAT, in the value of the second-order association rate constant, k,-107- 10' M-' S K I (Fasella and Hammes,

1967). However, hydroxylamine belongs to a class of coni- pounds that exhibit unusually high nucleophilic reactivity due to an electronegative atom with a free electron pair adjacent 11) the nucleophilic atom ( ' ( x effect', Edwards and Pearson, 1062). Since hydroxylamine has an association rate constant k , for mAspAT similar to H,N-0-Asp and H,N-0-Glu (Table 2), uoni- plementarity to the active site clearly does not play a signit'icant role in the reaction. The low complementarity is expre k+,-values that are 5-7 orders of magnitude smaller than those for the natural substrates, whereas the high nucleophilicity causes the k - , rates to' be 8-11 orders of magnitude smaller.

Several other observations are explained by the large d~ffer- ences in reactivity and complementarity between the two cl;tsses of inhibitors. The k , , values for the structurally widely different hydroxylamine deriva1.ives differ by less than two orders of inag- nitude (Table 2), while the replacement of a single H atom in L- Asp by a methyl group reduces the k+,-value of L - A s ~ from 10"-I s-l to lo4 M-' s- ' (Hammes and Haslam, 1968). Due to its high reactivity H2N-O-Asp binds to both the open and closed active site of mAspAT. It is an effective inhibitor of inany different pyridoxal-P enzymes, with necessarily distinct substrate-binding sites (John et al., 1978). For 2-MeAsp, X-ray studies on different A.spAT isozymes have shown that it binds readily to the enzyme in its closed form while it has hard! any affinity for the open conformation, in spite of its close resem- blance to the natural substrate L - A s ~ (Jansonius and Vincent, 1987).

Conformation of the oxime complexes in solution. Recently published studies on the conformational equilibrium of mAspAT in solution and in the crystalline state (Picot et al.. 1991 ; Hohen- cster and Jansonius, 1994) confirmed earlier indications that the unliganded enzyme adopts the open conformation and that the closed conformation is stabilized by C, and C, dicarboxylic acid inhibitors. The question arose as to whether binding of monocarboxylic inhibitors lacking the a-carboxylate group would a150 be able to shift the conformational equilibrium of the enLyme towards the closed form. Our results from experiments with bimane-labelled Cys166 and with tryptic digcstion showed that this is not the case. The inability of monocarboxylic and car-box-

MarkoviC-Housley et al. (EUK J. Biochem. 236) 103 1

ylate-free inhibitors to induce the closed conformation in solution is in agreement with the results from cocrystallization experiments (Delbaere et al., 1989).

Crystal structures. Binding of H,N-0-Asp to preformed crystals of the enzyme in the closed and open conformation did not alter either of the two crystal structures except for local structural changes within the active site. Significant conformational changes of the cofactor are observed when comparing the structure of OxOP with that of the unliganded holoenzyme. In OxCL, the orientation and conformation of the cofactor is very similar to that in the external aldimine with 2-MeAsp (Fig. 3), in agreement with the results from spectroscopic studies of the enzyme-oxime complexes in solution and in the crystalline state (Delbaere et al., 1989).

Comparison of the structures OxOP and OxCL revealed that the oxime double bond (C4'=N4') is only slightly closer to coplanarity with the pyridine ring in the closed than in the open structure. This is probably due to a weaker interaction of Arg292" with the carboxylate group of H,N-0-Asp in OxOP than in OxCL that is caused by different side-chain conformations of Arg292* in the open and the closed enzyme structure. In OxCL, the space of the missing carboxylate group is occupied by a water molecule that mediates an indirect interaction between the oxime oxygen and Arg386. There is no such interaction in the open form since Arg386 of the small domain is too far away.

Attainment of aldimine-pyridine ring coplanarity in covalent mAspAT-inhibitor complexes. Structural and biochemical studies on mAspAT complexes with dicarboxylate substrates and inhibitors have shown that ligand binding induces the protonation of N4', a significant conformational change of the cofactor, and coplanarity of the C4'=N4' double bond with the pyridine ring (McPhalen et al., 1992a). These events take place in the early stages of the transamination reaction and appear to be a prerequisite for further reaction steps to occur. The conjugation of the aldimine double bond with the pyridine ring is a necessary condition for the subsequent tautomeric conversion of aldimine to ketimine intermediate. In the unliganded enzyme at physio- logical pH, the internal aldimine between pyridoxal-P and Lys258 is unprotonated and the C4'=N4' double bond is almost orthogonal to the plane of the pyridine ring. The pK of the aldimine is i n the range 6.2-6.5. Dicarboxylate substrates and inhibitors such as maleate neutralize the positive charges of Arg292* and Arg386 and thereby increase the aldimine pK by about pH 2 (Kirsch et al., 1984 and references therein). The pK of the u-amino group of the substrate, L-aspartate or L-glutamate, is lowered due to the neutralisation of its carboxylate groups by the two arginine residues. In combination with the effect on the aldimine pK this allows the first covalent step in the catalytic reaction, the transaldimination, to occur (Karpeisky and Ivanov, 1966; Ivanov and Karpeisky, 1969). This mechanism discrimi- nates against other amino acids and amino-group-containing rea- gents.

All of the high-resolution structures of mAspAT have shown coincidence of aldimine conjugation with protonation of N4'. Clearly, the hydrogen-bonded ion pair N4'H' and 03'- is instru- mental in attaining the aldimine-pyridine ring coplanarity. For example in the native pyridoxal-P enzyme structure at low pH (LPH), the mere protonation of N4' of the internal aldimine at low pH (as demonstrated by crystal spectra) is sufficient to induce the coplanarity (McPhalen et al., 1992b). In the external aldimine with 2-methylaspartate, N4' is partially protonated. Al- dimine-pyridine ring coplanarity in this adduct is assisted by strong interactions of the a-carboxylate and w-carboxylate of

AMA with Arg386 and Arg292*, respectively, upon closure of the active site (Kirsch et al., 1984; McPhalen et al., 1992a).

In the OxCL and OxOP structure?, the oxime double bond is nearly coplanar with the cofactor pyridine ring (Fig. 1 b and c), despite the absence of the attracting force between NH' and 03- . The double-hydrogen-bonded ion pair between the p-carboxylate group of L-aspartate and the guanidinium group of Arg292* alone apparently suffices already to bring about the coplanarity of the external aldimine double bond with the pyridine ring plane required for effective catalysis. The same orientation of the coenzyme ring plane observed when superimposing the active-site structures of OxCL, OxOP and AMA (Fig. 3) indicates that the relative positions of the small and large domains in mAspAT do not significantly affect the geometry of the cofactor-oxime moiety. This is understandable, as predominantly residues from the large domain interact directly with the cofactor, in contrast with complexes with natural substrates which make strong hydrogen-bonded ion pairs between their a-carboxylate group and Arg386 from the small domain.

Conclusions. Hydroxylamine-derived inhibitors (general for- mula H,NOR), interact readily and reversibly with pyridoxal phosphate in the active site of mAspAT to form tight enzyme- inhibitor complexes, in solution and in the crystalline state. The binding of these inhibitors to the enzyme is apparently driven by the high chemical reactivity of the H,N-0 moiety, rather than by molecular complementarity as in the case of natural substrates.

Unless the hydroxylamine derivatives contain a carboxylate group, as in H,N-0-Asp and H,N-0-Glu, the oxime adducts formed are unstable since the pyridoxal-P oxime tends to disso- ciate from the enzyme.

The cocrystallization experiments showed that the H,N-0- Asp and H,N-0-Glu oximes do not stabilize the closed form of the enzyme. This was confirmed by the fluorescence experiments of the bimane-labelled complexes in solution, and by proteolytic digestion experiments.

The successful exchange of maleate for H,N-0-Asp in crystals of the closed form of mAspAT and the stability of the resulting crystalline oxime adduct (allowing its X-ray structure OxCL to be determined) demonstrated that the crystal lattice forces are strong enough to stabilize the H,N-0-Asp oxime adduct of mAspAT in its closed conformation.

The OxOP and OxCL crystal structures suggest that the interaction between Arg292* and the w-carboxylate of L-aspartate by itself suffice to orient this substrate in a productive way in the active site. However, the lack of an cx-carboxylate group prevents domain closure. The missing a-carboxylate group would, if present, presumably interact with Arg386 of the small domain and induce the domain movement necessary for closure.

Thank? are due to Mrs U. Griitter for her invaluable help in the preparation of the manuscript, and to Mrs M . Jaggi for preparing the schemes. Support by the Swiss National Science Foundation Grants 31- 36432.92 to J . N. J. and 31-36542.92 to P. C. is gratefully acknowledged.

REFERENCES Birchmeier, W., Wilson, K. J . & Christen, P. (1973) 1. B i d . Chern. 248,

1751 -1759. Collaborative Computing Project No.4 (1 993) Acfu Cryrfuhgr Sect. D

Rid . Crystullogl: 50, 760-763. Delbaere, L. T. J., Kallen, J., Markovit-Housley, Z., Khoinutov, A. R.,

Khomutov, R. M., Karpeisky, M. Ya. & Jansonius, J . N. (1989) Bi- oclzirnie (Puris) 71, 449-459.

Edwards, J. 0. & Pearson, R. G. (1962) J. Am. Clirm. Soc. 84, 16-24.

1032 MarkoviC-Housley et al. (Eur: J. Biochem. 236)

Eichele, G., Karahelnik, D., Halonhrenner, R., Jansonius, J. N. & Chris-

Fasella, P., Giartosio, A. & Hammes, G. G. (1966) Biochemisfry 5, 197-

Fasella, P. & Hammes, G. G. (1967) Biochemistry 6 , 1798-1804. Ford, G. C., Eichele, G. & Jansonius, J. N. (1980) Proc. Nut1 Acad. Sci.

USA 77, 2559-2563. Gehring, H. & Christen, P. (1975) Biochern. Biophys. Res. Commun. 63,

441 -447. Gehring, H., Christen, P., Eichele. G., Glor, M., Jansonius, J. N., Reimer,

A,-S., Smit, J . D. G. & Thaller, C. (1977) J. Mol. B i d . 115, 97- 101.

ten. P. (1978) J . B i d . Chem. 253, 5239-5242.

202.

Gehring, H. & Christen, P. (1978) J. Bid . Chem. 253, 3158-3163. Hammes, G. G. & Haslam, J. L. (1968) Biochemistry 7, 1519-1525. Hohenester, E. & Jansonius, J. N. (1994) J. Mol. B i d . 236, 963-968. Ivanov, V. I. & Karpeisky, M. Ya. (1969) Adv. Enzymol. 32, 21 -53. Jansonius, J. N., Eichele, G., Ford, G. C., Picot, D., Thaller. C. & Vin-

cent, M. G. (1985) in Transaminases (Christen, P. & Metzler, D., eds) pp. 109-138, John Wiley & Sons, New York.

Jansonius, J. N. & Vincent, M. G. (1987) in Biological niacromo1ecule.s and assemblies (Jurnak, F. A. & McPherson, A,, eds) vol. 3, pp. I87-285, John Wiley & Sons, New York.

John, R. A,, Charteris, A. & Fowler, L. Y. (1978) Biochem. J. 171, 771 - 779.

Karpeisky, M. Ya. & Ivanov, V. I. (1966) Nature 210, 493-496. Karpeisky, M. Ya., Khomutov, R. H., Severin, E. S. & Breusov, Yu. N.

(1 963) in Chemical arid biological aspects of pyridoxal catalysis (Snell, E. E., Fasella, P. M., Braunstein, A. E. & Rossi-Fanelli, A,, eds) pp. 323-331, Pergamon Press, New York.

Khomutov, A. R., Karpeisky, M. Ya. & Severin, E. S . (1961) Biokhimiu

Khomutov, R. M., Karpeisky, M. Ya. & Severin, E. S. (1963) in Chemi- cal and biological aspects of pyridoxal catalysis (Snell, E. E., Fa- sella, P. M., Braunstein, A. E. & Rossi-Fanelli, A,, eds) pp. 313- 322, Pergamon Press, New York.

26, 772-781.

Khomutov, R. M., Gabibov, A. G., Khurs, E. N., Tolosa, E. A,, Shiister, A. M., Goryachenkova, E. V. & Khomutov, R. M. (1987) in Bio- chemist? of vitamin B, (Korpela, T. & Christen, P., eds) pp. 317- 320, Birkhauser Verlag, Basel.

Kirsch, J. F., Eichele, G., Ford, G. C., Vincent, M. G., Jansonius. 1. N., Gehring, H. & Christen, P. (1984) J . Mol. B i d . 174, 497-525.

Kosower, N. S., Kosower. E. M., Newton, G. L. & Ranney, H. M. ( 1979) Proc. Natl Acad. Sci. USA 76, 3382-3386.

Malashkevich, V. N., Toriey, M. D. & Jansonius, J. N. (1993) Biochrmis- try 32, 13451-13462.

McPhalen, C. A,, Vincent, M. G., Picot, D., Jansonius, J. N.. Lek . A. M. & Chothia, C. (1092a) J. Mol. B i d . 227, 197-213.

McPhalen, C. A,, Vincent, M. G. & Jansonius, J . N. (1992b) J . Mol.

Messerschmidt, A. & Pflugrath, J. W. (1987) J . Appl. Crystal/o,ec 20,

Pfister, K., Sandmeier, E., Berchtold, W. & Christen, P. (1985) ./. Biol.

Picot, D. (1987) PhD thesis, University of Basel. Picot, D., Sandmeier, E., Thaller, C., Vincent, M. G., Christen. P. &

Raunio, R. P., Lindberg. R. K. & Jenkins, W. T. (1984) Arch. Biochem.

Sandmeier, E. & Christen, P. (1980) J. Biol. Chem. 255, 10284- 10289. Sandmeier, E. & Christen, P. (1984) in Chemical and biological mpects

of vitamin B, catalysis (Evangelopoulos, A. E., ed.) part B. pp. 1 17- 124, Liss, New York.

Thaller, C., Weaver, L. H., Eichele, G., Wilson, E., Karlsson, R. k Jan- sonius, J. N. (1981) J. Mol. Biol. 147, 465-469.

Vincent, M. G., Picot, D., Eichele, G., Jansonius, J. N., Kirsten. H. & christen, P. (1984) in Chemical and biological aspects of vitumin B, catalysis (Evangelopoulos, A. E., ed.) part B, pp. 233-243, Liss, New York.

Biol. 225, 495-517.

306-315.

Chem. 260, 11414-11421.

Jansonius, J. N. (1991) Eur: J. Biochem. 196, 329-431.

Biophys. 233, 43-49.

Date post:	01-Dec-2023
Category:	Documents
Upload:	independent
View:	0 times
Download:	0 times

Crystal Structures and Solution Studies of Oxime Adducts of Mitochondrial Aspartate Aminotransferase

Documents