Characterization of the metallocenter of rabbit skeletal muscle AMP deaminase. Evidence for a dinuclear zinc site

Biochimica et Biophysica Acta 1774 (2007) 312 – 322 www.elsevier.com/locate/bbapap

Characterization of the metallocenter of rabbit skeletal muscle AMP deaminase. Evidence for a dinuclear zinc site Stefano Mangani a,d , Manuela Benvenuti a , Arthur J.G. Moir b , Maria Ranieri-Raggi c , Daniela Martini c , Antonietta R.M. Sabbatini c , Antonio Raggi c,⁎ a Dipartimento di Chimica, Università di Siena, Via Aldo Moro, 53100-Siena, Italy Krebs Institute, Department of Molecular Biology and Biotechnology, University of Sheffield, Sheffield S10 2UH, UK Dipartimento di Scienze dell'Uomo e dell'Ambiente, Chimica e Biochimica Medica, Università di Pisa, Via Roma 55, 56126 Pisa, Italy d CERM, Università di Firenze, Via Luigi Sacconi 6, 50019 Firenze, Italy b

c

Received 11 September 2006; received in revised form 6 December 2006; accepted 18 December 2006 Available online 23 December 2006

Abstract XAS of Zn–peptide binary and ternary complexes prepared using peptides mimicking the potential metal binding sites of rabbit skeletal muscle AMP deaminase (AMPD) strongly suggest that the region 48–61 of the enzyme contains a zinc binding site, whilst the region 360–372 of the enzyme is not able to form 1:1 complexes with zinc, in contrast with what has been suggested for the corresponding region of yeast AMPD. XAS performed on fresh preparations of rabbit skeletal muscle AMPD provides evidence for a dinuclear zinc site in the enzyme compatible with a (μ-aqua)(μ-carboxylato)dizinc(II) core with an average of two histidine residues at each metal site and a Zn–Zn distance of about 3.3 Å. The data indicate that zinc is not required for HPRG/AMPD interaction, both zinc ions being bound to the catalytic subunit of the enzyme, one to the three conserved amino acid residues among those four assumed to be in contact with zinc in yeast AMPD, and the other at the N-terminal region, probably to His-52, Glu-53 and His-57. Tryptic digests of different enzyme preparations demonstrate the existence of two different protein conformations and of a zinc ion connecting the N-terminal and C-terminal regions of AMPD. © 2007 Elsevier B.V. All rights reserved. Keywords: AMP deaminase; Zinc binding site; Histidine-proline-rich-glycoprotein; X-ray absorption spectroscopy

1. Introduction AMP deaminase (AMPD, EC 3.5.4.6) catalyzes the hydrolytic deamination of AMP to IMP and ammonia. Its level of activity in skeletal muscle is particularly high when compared with that found in all other tissues, including heart and smooth muscle [1]. The function of AMPD in muscle operation is not clearly defined, but the considerably higher specific activity of the isoform specific to skeletal muscle indicates that the biochemical role of AMPD is likely to be associated with the specialized chemistry of this tissue. This view is supported by Abbreviations: AMPD, AMP deaminase; HPRG, histidine-proline-richglycoprotein; XAS, X-ray absorption spectroscopy; EXAFS, Extended X-ray Absorption Fine Structure; FT, Fourier Transforms; BVS, bond valence sum; CD, Circular Dichroism; SOD1, superoxide dismutase ⁎ Corresponding author. Tel.: +39 0502218679; fax: +39 0502218660. E-mail address: [email protected] (A. Raggi). 1570-9639/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bbapap.2006.12.005

our observation of the association of rabbit skeletal muscle AMPD with a histidine-proline-rich-glycoprotein (HPRG)-like molecule [2]. We have also demonstrated that in healthy human skeletal muscle an anti-HPRG antibody selectively binds to type IIB fibers that contain the highest level of AMPD activity [3] and in a recent immunohistochemical study performed on human skeletal muscle biopsies from patients with AMPD deficiency we have demonstrated a correlation between the muscle content of the HPRG-like protein and the level of AMPD activity [4]. A tetrameric structure has been suggested for skeletal-muscle AMPD from various species. However, a range of molecular masses (from 238 to 326 kDa) has been reported for the enzyme [5]. In the light of our observation that the rabbit enzyme undergoes fragmentation on storage, the N-terminal domain (10 kDa) being removed [6], an effect that can be reproduced in vitro by limited trypsinization which cleaves the enzyme subunit (85 kDa) after lysine-95 [7,8], we interpreted the different

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

reports on the molecular mass of skeletal-muscle AMPD as being due to the inherent instability of the enzyme during the prolonged extraction step that is performed at room temperature. Our determination by sedimentation-equilibrium analysis of the molecular mass of freshly prepared rabbit skeletal muscle AMPD indicated the presence of two species of 173 kDa and 309 kDa, which is consistent with the existence of a dimer– tetramer equilibrium [5]. In the light of the subsequent observation of the association of rabbit skeletal muscle AMPD with a HPRG-like molecule [2], the 309 kDa molecular mass determined for the native enzyme is in total agreement with a new model for AMP deaminase, a 1:1 molecular adduct in which two 85-kDa catalytic subunits assemble with two HPRG subunits (approximately 70 kDa each). The characterization of skeletal muscle AMPD as a zinc metalloenzyme was reported for the rat enzyme [9] and for the rabbit enzyme [10] on the basis of its interaction with chelating agents and metal ions. Atomic absorption analysis established the presence of 2.0 and 2.6 g atoms of zinc, respectively, per mole of rat enzyme (MW = 290 000) and rabbit enzyme (MW = 278 000) [9,10]. More recently, 1 zinc atom content was reported for the 80 kDa subunit of a form of AMPD from baker's yeast missing the 192 N-terminal amino acids due to proteolysis during purification [11]. Alignment of the amino acid sequence for yeast AMPD with that for mouse adenosine deaminase demonstrated conservation of the 4 amino acid residues (3 His and 1 Asp) known from the X-ray crystal structure of adenosine deaminase to bind zinc in contact with the attacking water nucleophile [11,12], suggesting for the 810 amino acid monomer of yeast AMPD the same model of a pentacoordinated zinc bound at the catalytic site that was described for adenosine deaminase, a 352 amino acids protein [11]. The alignment of the amino acid residues that have been proposed to be in contact with zinc in yeast AMPD [11] with the available amino acid sequences of skeletal muscle AMPD (rat and human) [13] demonstrates His-422, His-630 and Asp-707 are conserved but that His-424 in yeast AMPD is replaced by Gly-365 in both skeletal muscle enzymes. We have previously observed that the HPRG component can be isolated from rabbit skeletal muscle AMPD by zinc-affinity chromatography, indicating the presence in the HPRG-like protein of a specific metal binding site [14]. An investigation by X-ray absorption spectroscopy (XAS) of the zinc binding behaviour of the isolated HPRG showed that the protein can bind zinc, most probably in a dinuclear cluster where each Zn2+ ion is coordinated, on average, by three histidine and one heavier ligand, most likely to be a sulfur from a cysteine [15]. Taken together, these observations confirm that zinc is a firmly bound component of skeletal muscle AMPD and is essential for enzyme activity but also suggest that the model of the zinc binding site proposed for yeast AMPD cannot be extended to skeletal muscle AMPD. The present paper reports an investigation by X-ray absorption spectroscopy (XAS) of the zinc binding sites of rabbit skeletal muscle AMPD. We have collected X-ray fluorescence data on Zn–peptide binary and ternary complexes prepared using a number of synthetic peptides that mimic the potential metal binding sites of the enzyme. X-ray absorption spectro-

313

scopy was also performed on two fresh preparations of rabbit skeletal muscle AMPD with different contents of HPRG. The extended X-ray absorption fine structure (EXAFS) results strongly suggest that the region 48–61 of rabbit muscle AMPD contain a zinc binding site whilst region 360–372 is not able to form 1:1 complexes with zinc, in contrast with what has been suggested for the corresponding region of the yeast enzyme. Evidence is also provided for a dinuclear zinc site in the enzyme compatible with a (μ-aqua)(μ-carboxylato)dizinc (II) core with an average of two histidine residues at each metal site and a Zn–Zn distance of about 3.3 Å. The data indicate that zinc is not required for HPRG/AMPD interaction, both zinc ions being bound to the catalytic subunit of the enzyme, one to the three conserved amino acid residues among those four assumed to be in contact with zinc in yeast AMPD, and the other at the Nterminal region, probably to His-52, Glu-53 and His-57. 2. Materials and methods 2.1. Reagents Trypsin and AMP were purchased from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Phosphocellulose resin (P-11) was supplied by Whatman International Ltd. (Maidstone, UK). All of the chemicals and other reagents were of analytical grade.

2.2. AMP deaminase AMPD was prepared as previously described from fresh muscle dissected from the back and hind leg of rabbits [15]. Homogenization of the muscle and phosphocellulose purification of the enzyme was carried out using at each step a buffer system containing 5 mM NaN3. Under these conditions, the enzyme preparation is very stable, the 85- to 70-kDa SDS gel band transition of the purified enzyme occurring with a half-time of one month [15]. X-ray absorption spectroscopy was performed on two fresh different rabbit skeletal muscle AMPD preparations that were obtained by eluting the enzyme bound to the phosphocellulose column either by direct elution using 1.0 M KCl or with a two-step elution procedure. The enzyme samples used for recording XAS spectra were as follows: Sample 1 Holo AMPD–HPRG complex. The enzyme prepared following the one-step elution procedure was concentrated by ammonium sulphate precipitation and redissolved in 5 mM phosphate buffer and 0.2 M KCl at pH 7.0. The protein concentration of this sample was about 10 mg/ml, which was the maximum concentration attainable without the appearance of insoluble aggregates. The Zn(II) concentration in this sample was 170 μM as indicated by atomic absorption data, consistent with the presence of about 4 Zn ions in the putative AMPD–HPRG tetramer, composed of two catalytic and two HPRG subunits [15]. Sample 2 AMPD partially depleted of HPRG. The enzyme was prepared following the two-step elution procedure. As has been previously established, the elution of the cellulose phosphate-bound enzyme with 1.0 M KCl, after the column had been washed with 0.6 M KCl, selectively elutes about 30% of the HPRG component from the adsorbed AMPD complex, thereby causing an about 20% increase of the specific activity of AMPD [15]. The enzyme preparation was concentrated as described for sample 1. The protein concentration of sample 2 was about 16 mg/ml and a Zn(II) concentration of 135 μM was determined by atomic absorption. The protein concentration of the two AMPD samples was determined spectrocm photometrically by using the A1%,1 values of 12.6 and 9.1 for the enzymes 280 obtained with the one-step and the two-step elution procedures, respectively [15].

314

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

2.3. Fractionation of a tryptic limited digest of AMPD Limited proteolysis of AMPD with trypsin (trypsin/AMPD 1:80, w/w) was carried out in the presence of 1 M KCl adjusted to pH 7.0 with 1 M K2HPO4 (20 °C). After 60 min of digestion, 1 μl of 1 M HCl was added to the reaction mixture (0.5 ml) to cause precipitation of AMPD. The constituent peptides of the supernatant were separated by reversed-phase HPLC using an Alphasil C18 column equilibrated with 0.1% (v/v) trifluoroacetic acid. Peptides were eluted using a gradient, at 1 ml/min, of acetonitrile containing 0.1% trifluoroacetic acid/0–21% (v/v) acetonitrile over 2 min, followed by 21–28% acetonitrile over 18 min and 28–70% acetonitrile over 17 min. Peptides were detected by their UV absorption at 214 nm. Peptide peaks were collected and freeze-dried. The HPLC apparatus consisted of a Model 600 E from Waters Associates. The peptides that were recovered after HPLC fractionation of the AMPD digest were characterized by N-terminal analysis. N-terminal sequencing was performed on an ABI 476A protein sequencer.

2.4. Peptide synthesis In order to define the zinc binding mode to AMPD, we prepared a number of synthetic peptides mimicking the potential metal binding sites of the enzyme. The following four peptides were synthesised: (1) Ac-PISHHEMQAHILHM-NH2 corresponding to residues 48–61 of rabbit muscle AMPD [8]; (2) Ac-LDVHAGRQTFQRF-NH2 corresponding to residues 360–372 of rat and human muscle AMPD [13]; (3) Ac-LQKGLMISLSTDDPMQF-NH2, corresponding to residues 638– 654 of rat and human muscle AMPD [13] and totally conserved also in the rabbit enzyme, as shown by the partial amino-acid sequence data reported in the present paper. Both peptides 2 and 3 show sequence homology with regions of the yeast enzyme that have been predicted to contain putative metal binding sites [11]. (4) Ac-LDVHAHRQTFQRF-NH2 corresponding to the 360–372 region of rat and human AMPD but with Gly365 replaced with a His residue in order to introduce the HAH motif that is predicted to bind zinc in the homologous region of the yeast enzyme. Peptide syntheses were performed using a Milligen 9050 peptide synthesiser. Rink Amide NovaGel resin (Novabiochem) was used to prepare peptide amides. N-α-Fmoc protected amino acids, N-α-acetyl-L-leucine and Nα-acetyl-L-proline (Novabiochem) were added in a 4-fold molar excess and were activated in situ using equimolar amounts of HCTU and HOBt (Novabiochem) in 0.6 M N-methylmorpholine in DMF. Peptides were cleaved from the resin using 95% TFA in the presence of appropriate scavengers and were purified by HPLC using a Vydac C18 (25 cm × 2.4 cm) column.

2.5. XAS data collection and analysis Zn–peptide binary complexes in a stoichiometric ratio 1:1 were prepared for all 4 synthetic peptides described in the previous section, by dissolving ZnCl2 and peptides in 50 mM HEPES buffered water solutions at pH 7.2. The final Zn–peptide concentration was 1.5 mM. Zn–peptide 1 + 2, 1 + 3 and 2 + 3 ternary complexes, in a stoichiometric ratio 1:1:1, were also prepared with the same salt and buffer solution, with a final Zn–peptide concentration of 1.5 mM. For the XAS measurements 45 μL of the above Zn–peptide solutions were filled into plastic cells covered with Kapton windows. Both the cells and the Kapton foils used for the windows were thoroughly washed with ultra pure water and absolute ethanol and then dried before use. X-ray fluorescence data on the Zn– peptide complexes were collected at DESY (Hamburg, Germany) on the EMBL bending magnet beam line D2 using Si(111) double crystal monochromator for the measurement at the zinc edge. The data collection protocol used for all peptide samples is similar to that already reported [15]. In the X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) regions steps of 0.3 and 0.5–1.2 eV were used respectively. An absolute energy calibration of the spectra was obtained by recording known

Bragg reflections of a Si(220) crystal in back reflection geometry following a reported procedure [16] (E0,Zn(II) = 9663.7 eV). The signal from each sample was excellent and 14–20 scans were averaged to obtain good signal/noise statistics. The study of Zn-AMPD solutions poses several experimental problems since at high protein concentration a loss of catalytic activity is observed, due to the formation of insoluble aggregates. By ammonium sulphate precipitation in the presence of phosphate we have obtained stable, active enzyme preparations which contained approximately 100–200 μM zinc ion. This concentration is one order of magnitude lower than that needed to obtain useful data from a conventional XAS spectrometer. For this reason, the experimental setup especially built for ultradiluted samples on the ID26 beamline at the European Synchrotron Radiation Facility (ESRF, Grenoble, Fr) was used [17]. Measurements on samples where metal concentration is in the 100 μM range entail careful control of experimental conditions due to possible contamination of the sample and/or the cryostat with exogenous metal. This is particularly true with zinc because of the ubiquitous nature of the ion. For this reason, the whole experimental apparatus was carefully cleaned, the sample holders and the Kapton foil used to cover the liquid sample holders where treated with NaEDTA solutions to remove all possible contaminating metals and then washed four times with excess of the buffer solution in ultra-pure water. Before data collection, blank scans of the cryostat were taken using empty sample holders to ascertain the absence of trace zinc metal. Furthermore, before every measurement the sample fluorescence was monitored. In each case the only metal present in the samples was confirmed to be zinc. Samples were loaded in 1 mm thick plastic holders covered with Kapton windows and flash-frozen into liquid nitrogen. All the spectra were collected at 20 K by using a liquid helium flow cryostat. The experimental details have already been reported elsewhere [18]. The data collection and reduction was performed with the XOP/XAID software [19] and the EXPROG set of programs [20]. Series of 40–120 scans for each sample were collected and averaged. The intrinsic low concentration of zinc present in all samples prevented us to obtain useful XAS spectra beyond k = 11.6–11.7 Å− 1. The whole experimental spectra were compared with theoretical simulations obtained by the set of programs EXCURVE9.20 [21] using the same protocols used in similar studies [15]. The quality of the analysis was assessed by the standard statistical criteria [22] and expressed by the goodness of fit function ε and by the residual R-value [21].

3. Results 3.1. X-ray absorption spectroscopy of Zn–peptide complexes Alignment of the partial amino acid sequence data of residues 1–95 of rabbit skeletal muscle AMPD [8] with the predicted amino-acid sequence of the rat and human enzymes reveals that the rabbit and human enzymes share a region (HHEMQAH, residues 51–57) that is similar to the HExxH zinc binding motif of various zinc peptidases [23]. The hypothesis that the HExxxH motif present in the N-terminal sequence of rabbit skeletal muscle AMPD could contain the ligands for a zinc ion of the enzyme metallocenter is strengthened by the observation that in the region HELLGH (residues 450–455) of rat liver dipeptidyl peptidase III residues His-450, Glu-451 and His-455 are involved in zinc coordination and catalytic activity [24]. The same authors showed that changing the motif of the zinc binding site (HExxxH) to HExxH by deletion of Leu-453 barely influenced the enzyme activity and the zinc content. An explanation to this finding was given by the observation that when the zinc binding site (HExxH) in thermolysin is superimposed onto the iron binding site of tyrosine hydroxylase (HExxxH), the metal binding residues (two His and one Glu) of the two enzymes are located in the same positions [24].

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

In order to ascertain whether the HEXXXH motif present in the N-terminal sequence of human and rabbit muscle AMPD could represent a zinc binding site we carried out the EXAFS analysis of a 1:1 solution of ZnCl2 and a synthetic peptide AcPISHHEMQAHILHM-NH2 (peptide 1) corresponding to residues 48–61 of the rabbit enzyme [8]. XAS is a technique uniquely suited to identify the occurrence of coordination bonds between a metal ion (Zn(II) in the present case) and a peptide containing histidine residues because of the unequivocal features, due to multiple scattering effects, arising in the XAS spectrum when histidine is bound to the metal ion [15,25]. The study was extended to investigate the interaction of zinc ion with two further synthetic peptides corresponding to regions of rat and human muscle AMPD [13] which show sequence homology with regions of the yeast enzyme that have been predicted to contain putative metal binding sites [11], namely residues 360–372 (AcLDVHAGRQTFQRF-NH2, peptide 2) and 638–654 (AcLQKGLMISLSTDDPMQF-NH2, peptide 3), the last one being totally conserved also in the rabbit enzyme, as shown by our partial amino-acid sequence data (see Table 3). The same 360–372 amino acid residues sequence is deduced from both available genes of skeletal muscle AMPD (rat and human) [13], but only one out of the two amino acid residues of this region supposed to be in contact with zinc in yeast AMPD [11] is conserved in the skeletal muscle enzyme (His363, corresponding to His-422 in yeast AMPD), whilst His424 of the yeast enzyme sequence is replaced by Gly-365 in the skeletal muscle enzyme. Therefore, a fourth peptide, corresponding to the 360–372 region of rat and human AMPD but with Gly-365 replaced with a His residue (AcLDVHAHRQTFQRF-NH2, peptide 4), so as to introduce the HAH motif that is predicted to bind zinc in the homologous region of the yeast enzyme, was also synthesised. The results of the EXAFS analysis of the solutions of the various Zn–peptide binary complexes in a stoichiometric ratio 1:1 are summarized in Table 1. The XAS spectrum of the Zn– peptide 1 solution showed the unequivocal camelback features due to the multiple scattering effect arising from the histidine imidazole coordination to zinc [15]: analysis of the XAS data confirms the formation of a complex, which occurs with the binding of the metal ion to two of the His residues present in the Table 1 EXAFS analysis of Zn–peptide complexes Zn–peptide complex

Zn ligand number and type

Distance (Å)

Debye–Waller factor 2σ2 (Å2)

R-factor

1

2 His 2 O/N 3 O/N 1 Cl/S 3 O/N 1 Cl/S 2 His 2 O/N 3 His 2 N/O 2 His 3 O/N

2.00 1.90 1.99 2.25 2.01 2.26 2.00 1.88 2.03 1.91 2.02 1.95

0.002 0.013 0.011 0.004 0.010 0.007 0.004 0.013 0.002 0.002 0.004 0.008

30.8

2 3 4 1+2 1+3

29.6 33.1 32.5 33.5 30.9

315

peptide, the zinc coordination being completed by the binding of two further O/N atoms from peptide or water molecules (Table 1). Secondary structures for metalloproteases containing HExxH (zincins) normally show the motif as an α helix. Furthermore, secondary structure prediction indicates that the region around the HELLGH motif of rat DPP III also is situated in an α helix [24]. By applying a number of secondary structure prediction algorithms to peptide 1 [26,27], we have demonstrated that the motif is most likely to exist in the form of an αhelix. Experimental justification for this conclusion was made by determining the CD spectrum of the peptide 1 solution and of its Zn(II) 1:1 complex (data not shown); this revealed that the peptide possesses about 20% α-helical secondary structure which increased slightly (to about 25%) in the zinc complex. In contrast to the results obtained with peptide 1, the solutions containing peptide 2 or 3 and Zn(II) in 1:1 ratio gave almost identical EXAFS spectra with no sign of Zn-His binding, indicating that zinc binds to 3–4 O-like atoms and 1 Cl ion. This suggests that although zinc might be bound to the Asp or Gln residues of either peptide, it does not bind to the DVH motif present in peptide 2 thereby casting some doubt as to whether the zinc ion is complexed by either peptide rather than being free in solution. The above results indicate that peptide 1 has a different coordination behaviour towards the Zn(II) ion with respect to peptides 2 and 3 , thereby strongly suggesting that the region 48–61 of rabbit muscle AMPD contain a zinc binding site, whilst there is no evidence that regions 360–372 (peptide 2) and 638–654 (peptide 3) are able to form 1:1 complexes with zinc, in contrast with what has been suggested for the corresponding regions of the yeast enzyme. The lack of Zn-binding observed for peptide 2 can be explained by taking into account that His424 of the yeast enzyme sequence is replaced by Gly-365 in mammalian muscle AMPD. This suggestion is strengthened by the EXAFS analysis of the Zn(II)–peptide 4 solution, which demonstrates that the 360–372 fragment acquires zinc binding properties upon the G365H substitution, since the data give unambiguous evidence of the involvement of the HAH motif in zinc coordination (Table 1). Concerning the lack of interaction with zinc shown by peptide 3 (residues 638–654), it should be noted that this peptide contains the zinc binding motif SLSTDDP that is conserved in all sequenced AMPD [11,13]. It has been hypothesized that in yeast AMPD a zinc ion would bridge between the carboxylate of an Asp residue of this region (possibly Asp-707, which corresponds to Asp-649 in mammalian muscle enzyme) and 3 His residues of other regions (namely His-422, His-424 and His-630 [11]. This hypothesis is strengthened by the results of the EXAFS analysis carried out on samples containing stoichiometric amounts of ZnCl2 and a mixture of two of the above described synthetic peptides in a 1:1:1 ratio (Table 1), revealing the formation of the ternary complexes Zn–peptide 1 + 2 and Zn–peptide 1 + 3. In the presence of both peptides 1 and 3, the zinc coordination increases above 4 while the two His coordination is maintained, thus suggesting the binding of peptide 3 in a ternary complex. Even more striking are the results of the

316

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

EXAFS analysis of the Zn(II)–peptide 1 + 2 solution which show an increase of the histidine coordination to zinc compared to that found in the binary complex Zn–peptide 1, although the EXAFS data of the Zn plus peptide 2 solution did not show the involvement of histidine in zinc coordination. This finding strongly supports the presence of a Zn–peptide 1 + 2 ternary complex and suggests that the single His residue present in the AMPD region corresponding to peptide 2 might be involved in the overall zinc coordination of the enzyme, although the lack of the HAH motif due to the H365G substitution compared to the yeast enzyme sequence clearly indicates that a novel model is required for the metal binding site of mammalian muscle AMPD. 3.2. X-ray absorption spectroscopy of rabbit skeletal muscle AMPD The Zn-K-edge regions of the two AMPD samples described in “Materials and methods" (sample 1, holo AMPD–HPRG complex; sample 2, AMPD partially depleted of HPRG) are reported in Fig. 1, while the EXAFS spectra and the Fourier Transforms (FT), together with the best simulations, are reported in Figs. 2 and 3, respectively. The zinc K-edge is located at 9664.5 eV in both samples and shows two sharp absorptions of identical intensity, at 9668,0 eV and at 9674,0 eV (Fig. 1). The presence of the characteristic camelback features in the EXAFS spectra denotes zinc binding to a histidine. Fig. 2 shows that the signal/noise ratio for each sample is satisfactory up to about k = 11.8 Å− 1, allowing the extraction of useful structural information from the spectra. Table 2 reports the structural parameters relative to the zinc coordination in the two samples obtained from the best fits to the spectra. The EXAFS spectrum of AMPD sample 1 (Fig. 2A) was reproduced by an average four-coordinated zinc bound to nitrogen/oxygen ligands, two of which belong to histidine residues as shown by the parameters reported in Table 2 (Fit

1). The possible presence of further zinc ligands (up to an overall number of six) and up to four histidines was tested by using the protocol described in Materials and methods. Introduction of multiple scattering effects from two histidine imidazole rings was able to fully reproduce the spectrum. The possibility of higher coordination numbers by other N/O ligands was also tested. Only a fit with a five-coordinated zinc resulted in relatively low R and ε values (Table 2, Sample 1, Fit 2), although worse than Fit 1. This result poses the problem of estimating the average zinc coordination number in AMPD. Our results point to either a four- or fivecoordinated zinc ion in the enzyme active site. The question appears more complex in the light of the fact that the EXAFS analysis, although performed on data sets limited in k range, provides some evidence for the presence of a dinuclear zinc site in the enzyme active site. The spectrum FT of AMPD sample 1 (Fig. 3A) shows a peak at about 3.5 Å in between the 2nd (at ∼ 3.0 Å) and 3rd (at ∼ 4.0 Å) shell histidine peaks. Such a feature is easily reproduced by the single scattering contribution of a zinc ion at about 3.3 Å with significant improvement of the fit (e.g. Sample 1, Fit 1 vs. Fit 3, Table 2). In contrast, a shell of 4–5 light atoms (C/N/O) at 3.0 to 3.5 Å from zinc was unable to reproduce this feature and all fits tried resulted in cancelling the contributions from these atoms by ending with extremely large, unphysical, Debye–Waller factors (∼ 0.200 Å2). The position of this FT peak does not change if the FT is performed over a different k range and by using a different FT window, confirming that this feature is not an artefact, but is a real component of the AMPD spectrum. The same feature is observed in the FT of the EXAFS spectra of AMPD sample 2 (Fig. 3B). providing further evidence for the presence of a dinuclear zinc cluster in the enzyme. The best EXAFS and FT simulations of sample 1 result in an average zinc coordination polyhedron composed of two histidines at 2.12 Å, and 2 N/O ligands at 1.92 Å (Table 2, Sample 1, Fit 1). The presence of a short Zn–(N/O) interaction

Fig. 1. Normalized Zn K-edge spectra of AMPD sample 1 (open triangles) and AMPD sample 2 (open circles).

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

317

Fig. 2. EXAFS spectra (continuous line) superimposed to the best fit (open circles) obtained with the parameters reported in Table 2. (A) AMPD sample 1; (B) AMPD sample 2.

is needed to reproduce the characteristic camelback feature of the first EXAFS oscillation. This suggests the presence of a negatively charged, carboxylate-type, oxygen in the first coordination shell. The 3.31 Å distance between the two zinc ions in the enzyme active site suggests the possibility that one or two ligands, such as a carboxylate and a hydroxide anion, may bridge between the two metal ions. The analysis of the EXAFS spectrum of sample 2 provides an average zinc coordination that is compatible with the structural parameters obtained from sample 1 as two histidines and 2 N/O ligands at short distance from zinc (1.91–1.92 Å) are reproducing the main features of the spectrum (Table 2, Sample 2, Fits 1–3). The presence of a Zn–Zn interaction at about 3.3 Å is also confirmed by these data. A slight, although not significant, improvement of the fit index and of the R-factor was obtained by adding a shell consisting of 0.5 N/O atoms at 2.03 Å (Table 2, Sample 2, Fit 2).

For this reason, the presence of five- or of a mixture of four- and five-coordinated zinc sites cannot be distinguished on the basis of our data. It is known indeed that coordination numbers obtained from EXAFS analysis alone are affected by large errors and the above difference in coordination numbers is well within the error threshold [28,29]. The bond valence sum (BVS) analysis [30] of the observed distances is consistent with either a four- or a mixture of four- and five-coordinated Zn(II) sites in AMPD (Table 2). It is interesting to note the difference between the present spectra (Zn–AMPD–HPRG complexes with varying HPRG content) and the previously reported EXAFS obtained for the Zn–HPRG complex [15]. In contrast to the data obtained for the Zn–HPRG complex, the present data do not show any evidence of sulfur coordination to zinc, thus demonstrating that no zinc is bound to HPRG when the ternary complex of HPRG with Zn and the AMPD catalytic subunit is formed.

318

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

Fig. 3. Fourier transforms (continuous line) superimposed to the best fits (open circles) obtained with the parameters reported in Table 2. (A) AMPD sample 1; (B) AMPD sample 2.

3.3. A model of dinuclear Zn site in rabbit skeletal muscle AMPD Taken together, our EXAFS data suggest that rabbit skeletal muscle AMPD is likely to possess an active site consisting of a dinuclear zinc site with two 4-coordinated (or a mixture of 4and 5-coordinated) zinc ions. The unambiguous detection of metal–metal scattering at distances larger than 3.0 Å in protein EXAFS is always difficult since factors such as multiple scattering from other ligands and metal–carbon scattering usually interfere with the metal–metal scattering [31]. Nevertheless, our data point conclusively to the presence of a dinuclear zinc site. The use of fitting protocols identical to that used to establish metal–metal interactions in other systems [22] give us confidence in drawing such conclusions. The Zn–Zn distance of 3.3 Å and the short Zn(II) first coordination distances suggest the presence of bridging ligands between the two metal ions. The analogy with other dinuclear

zinc hydrolytic enzymes of known structure such as dihydroorotase [32] and phosphotriesterase [33] strongly suggests that the bridging ligands might be a carboxylate and a water/hydroxyde anion. By coupling the above observations with the X-ray fluorescence data on the Zn–peptide complexes reported in the present paper, it can reasonably be inferred that in rabbit skeletal muscle AMPD zinc is bound in a dinuclear cluster, one of the Zn ions (Zn1 in Scheme 1) being bound at the enzyme C-terminal to the three conserved amino acid residues among those four supposed to be in contact with zinc in yeast AMPD, and the other (Zn2 in Scheme 1) being bound at the N-terminal region, probably to His-52 (or His-51), Glu-53 and His-57. As suggested by the 3.3 Å distance between the two zinc ions which results from the best EXAFS and FT simulations, a carboxylate (probably Asp-649) as well as a water/hydroxide molecule would likely bridge between the two zinc ions so that Zn1 and Zn2 might be 4- and 5-coordinated, respectively (Scheme 1).

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322 Table 2 EXAFS curve fitting results for AMPD samples 1 and 2. ESDs in parentheses are from the least-squares covariance matrix Ligand Sample 1 Fit 1 2 N/O 2 N-His 1 Zn BVS 1.89 Fit 2 3 N/O 2 N-His 1 Zn BVS 2.34 Fit 3 2 N/O 2 N-His BVS 1.89 Sample 2 Fit 1 2 N/O 2 N-His 1 Zn BVS 2.07 Fit 2 2 N/O 2 N-His 1 Zn 0.5 N/O BVS 2.15 Fit 3 2 N/O 2 N-His BVS 2.07

Distance (Å)

DW-factor 2σ2 (Å2)

ε2 × 101

R-factor

1.92(1) 2.12(1) 3.31(2)

0.003(2) 0.011(5) 0.017(6)

13.5

0.382

1.93(1) 2.15(1) 3.31(2)

0.008(2) 0.010(4) 0.017(6)

15.4

0.396

1.92(1) 2.12(1)

0.003(2) 0.011(5)

15.2

0.395

1.91(1) 2.05(1) 3.29(2)

0.008(3) 0.008(5) 0.024(8)

5.58

1.92(1) 2.05(1) 3.30(2) 2.03(3)

0.009(3) 0.004(5) 0.022(8) 0.009(5)

5.32

0.236

1.91(1) 2.05(1)

0.008(3) 0.008(5)

5.95

0.263

0.245

3.4. Justification for a structurally bridged dinuclear metallocenter in rabbit skeletal muscle AMPD In a previous paper [8] we reported that all of the peptides removed from rabbit skeletal muscle AMPD by digestion for 60 min at pH 7.0 with trypsin (trypsin/AMPD 1:100, w/w) were derived exclusively from residues 1–95 of AMPD (Table 3A). We have subsequently shown [2] that by slightly increasing the amount of the proteolytic enzyme (trypsin/AMPD 1:80, w/w), fragments that are derived from digestion of the HPRG-like component – ALDLINK (res16–22), DGYLFQLLR (26–34)

Scheme 1. A model of dinuclear Zn site in rabbit skeletal muscle AMPD. The amino-acid sequence of the 51–60-region is from our partial amino acid sequence data of the N-terminal region of the rabbit skeletal muscle AMPD [8].

319

Table 3 Peptides that were produced by limited tryptic digestion of different fresh preparations of rabbit skeletal muscle AMPD A

B

C

AMPD

AMPD

HPRG

AMPD

HPRG

2–18 4–18 19–23 24–73 31–72 36–72 80–95 81–95

4–11 6–11 12–18 19–23 24–30 36–57 36–71 58–71 76–79 80–95 81–95

16–22 26–34 460–477

4–11 6–11 12–18 19–23 24–30 36–41 42–52 175–189 190–196 456–461 462–467 632–640 641–654

16–22 26–34 35–42 50–56 460–477

Numbers indicate their collocation in the sequence of skeletal muscle AMPD [13] and rabbit plasma HPRG [34]. (A) The peptides were isolated by fractionation of the digest by gel filtration method (data from ref. [8]). (B) and (C) The peptides were isolated by HPLC of the soluble fraction obtained after acid precipitation of the digests of two different preparations of AMPD. (B) Data from ref. [2]. (C) Present data.

and GEVLPLPEANFPSFXLRH (460–477) – were also produced (Table 3B). In order to ascertain whether a further characterization of the primary structure of the HPRG-like component of AMPD could be obtained by N-terminal analysis of the peptides liberated by limited proteolysis of different AMPD preparations, we carried out the trypsinization of a new preparation of the enzyme under the conditions that we have more recently followed [2]. Reversed-phase HPLC of the supernatant revealed the presence of a large number of peaks that were isolated and subjected to N-terminal analysis. Table 3 compares the AMPD and HPRG fragments that have been identified (Table 3C) with those previously described. Two additional peptides derived from the HPRG N-terminal region, i.e. peptide 50–56, YLVLDVK, and peptide 35–42, VADAHLDK, in which G-42 of the rabbit plasma protein is replaced by K, were detected. Moreover, the previously detected 36–57 fragment of AMPD was itself split at F-41; several additional fragments were found that were derived from the middle and C-terminal regions of AMPD. These results show that the extent of digestion has increased in the middle and C-terminal regions of AMPD in comparison to what was observed with the previous preparation; in contrast no evidence of digestion within the second half of the 1–95 region of the new preparation was obtained. The qualitative differences observed between the products of the two limited digestions reveal the existence of two different protein conformations due to specific interactions between the N-terminal and C-terminal regions of rabbit skeletal muscle AMPD. Some of the proteolytic fragments of AMPD that have been isolated during this study are apparently non-tryptic since they result from cleavage C-terminal to Phe-41, His-52, His-57, Trp-455 and Phe-654. The observed cleavage at bulky aromatic residues may be due to the well-documented production of pseudotrypsin resulting from autocatalytic digestion of trypsin.

320

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

Cleavage by trypsin C-terminal to histidine is very unusual but since both His residues, His-52 (or His-51) and His-57, are bound to zinc, according to our model of a zinc dinuclear site in AMPD (Scheme 1), we would suggest that this has introduced a metal-dependent His specificity that is identical to one previously described [35], where a histidine substrate specificity was engineered into a trypsin subsite (S2′) by creating metal sites for Ni2+ and Zn2+ ions; it was suggested that a metal ion bridging the enzyme and substrate in the primary S1 binding pocket might result in cleavage C-terminal to a histidine residue [35]. On this basis, we can advance the hypothesis that the cleavage of AMPD by trypsin C-terminal to H-52 and H-57 is due to Zn2+ bridging between the side chain of these residues and Asp-189 that is located at the bottom of the trypsin S1 binding pocket, resulting in a histidine specificity of the proteolytic enzyme. 4. Discussion It was previously shown that native rabbit skeletal muscle AMPD, as isolated, contained 2.6 and 3.2 g atoms of zinc per mol (based on a molecular mass of 278 kDa) when analyzed by atomic absorption spectroscopy and by titration with 8-hydroxyquinoline-5-sulfonate, respectively [10]. The problem of assigning a precise stoichiometry for the zinc binding to the enzyme in vivo is complicated by the observation that the apoenzyme binds 4 g atoms of zinc per mol, but the increase of Vmax due to the addition of the fourth zinc atom is only 28% of that expected. This suggests that the fourth zinc atom is not directly associated with activity [10]. The discrepancy in zinc content between native and zinc-reconstituted enzymes may represent isolation of AMPD without its full complement of zinc, but the function of additional zinc bound, that is, whether or not the metal also participates in binding of the effectors (as was suggested by the competitive nature of the citrate inhibition of ADP activation), or participates in maintaining either tertiary or quaternary structure of the enzyme remained to be investigated [10]. We have recently reported [2] the association of the enzyme with an HPRG-like molecule, which is able to bind zinc, probably in a dinuclear metal binding site [15]. The complete separation of the HPRG component by zinc-affinity chromatography induced a marked reduction in the solubility of AMPD, indicating a role of this protein in assuring the molecular integrity of the enzyme [14]. Our determination of a 309-kDa molecular mass for the native enzyme by sedimentation– equilibrium centrifugation [5] is in agreement with an AMPD quaternary structure composed of two 85-kDa catalytic subunits assembled with two approximately 70-kDa HPRG subunits. As reported in the present paper, the Zn(II) concentration determined in the freshly prepared enzyme by atomic absorption spectroscopy is consistent with the presence of about 4 Zn ions in the putative AMPD–HPRG tetramer. As suggested by the data of the present paper, in rabbit skeletal muscle AMPD zinc appears to be bound in a dinuclear cluster with bridging ligand(s) between two zinc ions. One Zn (II) ion (Zn1) corresponds to the penta-coordinated zinc bound at the catalytic site that was described for adenosine deaminase

and yeast AMPD, whereas the other Zn(II) ion (Zn2) is probably localized in the region HHEMQAH (residues 51–57) at the enzyme N-terminus (Scheme 1). Since this putative Zn2 binding site is contained in the Nterminal region of the enzyme which is removed by the proteolytic process that occurs during storage of the enzyme, earlier data on the metal analysis of rabbit AMPD is likely to be very imprecise because the enzyme preparation used for these studies did not contain this N-terminal region. Consequences similar to those brought about by trypsin on the subunit structure of rabbit skeletal muscle AMPD, that is an apparent 10 kDa reduction in molecular mass of the native 85 kDa enzyme subunit with its conversion to a 75 kDa core (ΔL96) [8], were observed on storage of AMPD, as well as of the muscle used for the enzyme preparation [6]. A similar result has been observed for human AMPD recombinant protein, which is cleaved on storage to give a mixture of the truncated polypeptides ΔI86 and ΔH98 [36]. The model presented in this paper, of AMP binding at a dinuclear catalytic site in rabbit skeletal muscle AMPD, is in apparent contrast with that proposed for the yeast enzyme [11], which located a mononuclear catalytic site near the C-terminal end of the protein of 810 amino acids, with the region of amino acids 472–708 suggested as being in contact with the enzymebound zinc and the C6 hydrate of purine ribonucleoside, corresponding to amino acids 15–296 (out of a total of 352) of mouse adenosine deaminase [12]. The same authors suggested that the N-terminal 1–421 amino acid region of yeast AMPD must provide for the allosteric interactions by binding 2 mol of ATP per subunit, this function being absent in adenosine deaminase. It should be noted, however, that the form of yeast AMPD that was used for determination of zinc stoichiometry (1 zinc atom/subunit) was a fragment of the intact protein that did not contain residues 1–192 [11]. Therefore, it cannot be excluded that the presumed homology between yeast AMPD and mouse adenosine deaminase metallocenters could stand only for proteolysed AMPD. We therefore advance the hypothesis that the catalytic mechanism described for adenosine deaminase could also be shared by trypsinized skeletal muscle AMPD which loses the additional zinc atom and the sensitivity to ADP activation (and also shows an about 30% reduction of Vmax) as a consequence of the removal of residues 1–95 by proteolysis [5,7,8]. The results of the present paper demonstrate clearly that two different AMPD preparations subjected to the same proteolytic treatment may produce significantly different peptide profiles which must result from an altered susceptibility of each separate domain of AMPD to cleavage by trypsin (Table 3B and C). Since this observation is consistent with the presence of specific interactions between the N-terminal and C-terminal regions of the enzyme, it can be taken as structural evidence for the above described model of the dinuclear Zn site in which the metal could play a role in stabilizing one among the possible enzyme conformations. When the limited proteolysis affects only the Nterminal 1–95 residue region of the enzyme (preparation B), the consequence on the constitution of the enzyme metallocenter is probably the loss of the putative Zn2 ion without any major

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

change in the conformation of the remaining part of the enzyme. In contrast, when the digestion affects the binding region of the Zn1 ion at the C-terminus (preparation C), profound changes within the AMPD structure ensue as suggested by the observed fragmentation in the middle- and C-terminal regions of the enzyme, whilst the 58–95 region appears to be unaffected. Our early report [9] indicated that rat skeletal muscle AMPD contains 2 g atoms of zinc per mol (MW = 290 000). The significantly lower zinc content observed in rat AMPD in comparison with that determined for the rabbit enzyme can now be explained by the observed higher susceptibility of the rat skeletal muscle enzyme to proteolysis; whereas AMPD from fresh rabbit muscle yields an 85 kDa subunit even in the absence of protease inhibitors [15], rat skeletal muscle AMPD subunit has always been isolated as an approximately 70 kDa fragment, indicating that proteolysis occurring during the isolation of the enzyme had cleaved the N-terminal domain from the protein [8]. Our recent paper [37] has demonstrated that a calpain-like proteolytic activity produces the limited cleavage at the N-terminal regulatory domain of rabbit skeletal muscle AMPD and has given evidence that residues 10–17 of the rabbit enzyme, but not the corresponding highly divergent region of the rat enzyme, is intimately involved in a mechanism that protects the protein from N-terminal proteolysis. Further studies are necessary to establish which domains of HPRG are required for contact with AMPD and whether a mechanism exists of transferring zinc between the two proteins. However, the data of the present paper indicate that zinc is not required for HPRG/AMPD interaction since in contrast to the previously reported EXAFS obtained for the Zn–HPRG complex [15], the spectrum of the Zn–AMPD–HPRG complexes with various HPRG content does not show any evidence of sulfur coordination to zinc, thus demonstrating that no zinc is bound to HPRG when the ternary complex of HPRG with Zn and the AMPD catalytic subunit is formed. This conclusion is supported by our previous observation that the removal of zinc by EDTA from rat skeletal muscle AMPD does not abolish the tetrameric complex, since the apoenzyme behaves as a species with an apparent molecular mass of 254 kDa, which is somewhat lower than that of 285 kDa calculated for the native enzyme [38]. It has also been shown that alkalinization has consequences similar to those of the removal of zinc in provoking both a slight reduction of the sedimentation velocity of the enzyme and an enhancement of the reactivity of a class of SH groups essential for the AMPD activity [38]. These observations have been interpreted as due to an isomerization process which might be provoked by the deprotonation of specific residues influencing the properties of the metal binding sites of the catalytic subunit of AMPD. The above observations strengthen the hypothesis that the HPRG component behaves as a zinc metallochaperone [14,15]. Our findings about the AMPD–HPRG complex recall the welldocumented association of superoxide dismutase (SOD1) with its copper chaperone (CCS) for which also a crystal structure exists [39]. CCS brings CXC and CXXC metal binding motifs located in different protein domains and transfers copper to a histidine only site in SOD1 by a mechanism which most

321

probably involves the formation of an intermolecular disulfide bridge [39]. The presence in the HPRG molecule dissociated from rabbit skeletal muscle AMPD of a dinuclear cysteinebridged zinc-binding site [15] allows one to envisage the possibility that the novel component of AMPD could behave as CCS in enhancing the stability of the metalloenzyme through insertion of zinc. Acknowledgements The authors are thankful to the personnel of ESRF beamline ID26 and to Dr. Wolfram Meyer-Klaucke for assistance in X-ray absorption data collection at ESRF and EMBL Hamburg, respectively. This work was supported by the Italian Ministero dell'Istruzione, dell'Università e della Ricerca PRIN03–04 and by the European Community contract number RII3/CT/2004/5060008 for the work performed at EMBL Hamburg. References [1] E.J. Conway, R. Cooke, The deaminases of adenosine and adenylic acid in blood and tissues, Biochem. J. 33 (1939) 479–492. [2] M. Ranieri-Raggi, U. Montali, F. Ronca, A. Sabbatini, P.E. Brown, A.J.G. Moir, A. Raggi, Association of purified skeletal-muscle AMP deaminase with a histidine-proline-rich-glycoprotein-like molecule, Biochem. J. 326 (1997) 641–648. [3] A.R.M. Sabbatini, M. Ranieri-Raggi, L. Pollina, P. Viacava, J.R. Ashby, A. J.G. Moir, A. Raggi, Presence in human skeletal muscle of an AMP deaminase-associated protein that reacts with an antibody to human plasma histidine-proline-rich glycoprotein, J. Histochem. Cytochem. 47 (1999) 255–260. [4] A.R. Sabbatini, A. Toscano, M. Aguennouz, D. Martini, E. Polizzi, M. Ranieri-Raggi, A.J. Moir, A. Migliorato, O. Musumeci, G. Vita, A. Raggi, Immunohistochemical analysis of human skeletal muscle AMP deaminase deficiency. Evidence of a correlation between the muscle HPRG content and the level of the residual AMP deaminase activity, J. Muscle Res. Cell Motil. 27 (2006) 83–92. [5] M. Ranieri-Raggi, A. Raggi, Regulation of skeletal muscle AMP deaminase. Evidence for a highly pH-dependent inhibition by ATP of the homogeneous derivative of the rabbit enzyme yielded by limited proteolysis, Biochem. J. 272 (1990) 755–759. [6] M. Ranieri-Raggi, A. Raggi, Effects of storage on activity and subunit structure of rabbit skeletal-muscle AMP deaminase, Biochem. J. 189 (1980) 367–368. [7] M. Ranieri-Raggi, A. Raggi, Regulation of skeletal muscle AMP deaminase. Effects of limited proteolysis on the activity of the rabbit enzyme, FEBS Lett. 102 (1979) 59–63. [8] F. Ronca, M. Ranieri-Raggi, P.E. Brown, A.J.G. Moir, A. Raggi, Evidence of a species-differentiated regulatory domain within the N-terminal region of skeletal muscle AMP deaminase, Biochim. Biophys. Acta 1209 (1994) 123–129. [9] A. Raggi, M. Ranieri, G. Taponeco, S. Ronca-Testoni, G. Ronca, C.A. Rossi, Interaction of the rat muscle AMP aminohydrolase with chelating agents and metal ions, FEBS Lett. 10 (1970) 101–104. [10] C.L. Zielke, C.H. Suelter, Rabbit muscle adenosine 5′-monophosphate aminohydrolase. Characterization as a zinc metalloenzyme, J. Biol. Chem. 246 (1971) 2179–2186. [11] D.J. Merkler, V.L. Schramm, Catalytic mechanism of yeast adenosine 5′-monophosphate deaminase. Zinc content, substrate specificity, pH studies, and solvent isotope effects, Biochemistry 32 (1993) 5792–5799. [12] D.K. Wilson, F.B. Rudolph, F.A. Quiocho, Atomic structure of adenosine deaminase complexed with a transition-state analog: understanding catalysis and immunodeficiency mutations, Science 252 (1991) 1278–1284.

322

S. Mangani et al. / Biochimica et Biophysica Acta 1774 (2007) 312–322

[13] R.L. Sabina, T. Morisaki, P.R.H. Clarke, R. Eddy, T.B. Shows, C.C. Morton, E.W. Holmes, Characterization of the human and rat myoadenylate deaminase genes, J. Biol. Chem. 265 (1990) 9423–9433. [14] M. Ranieri-Raggi, D. Martini, A.R.M. Sabbatini, A.J.G. Moir, A. Raggi, Isolation by zinc-affinity chromatography of the histidine-proline-richglycoprotein molecule associated with rabbit skeletal muscle AMP deaminase. Evidence that the formation of a protein–protein complex between the catalytic subunit and the novel component is critical for the stability of the enzyme, Biochim. Biophys. Acta 1645 (2003) 81–88. [15] S. Mangani, W. Meyer-Klaucke, A.J.G. Moir, M. Ranieri-Raggi, D. Martini, A. Raggi, Characterization of the Zn binding site of the HPRG protein associated with rabbit skeletal muscle AMP deaminase, J. Biol. Chem. 278 (2003) 3176–3184. [16] R.F. Pettifer, C. Hermes, Absolute energy calibration of X-ray radiation from synchrotron sources, J. Appl. Crystallogr. 18 (1985) 404–412. [17] C. Gauthier, V.A. Sole, R. Signorato, J. Goulon, E. Moguiline, The ESRF beamline ID26: X-ray absorption on ultra dilute sample, J. Synchrotron Radiat. 6 (1999) 164–166. [18] M. Ranieri-Raggi, A. Raggi, D. Martini, M. Benvenuti, S. Mangani, XAS of dilute biological samples, J. Synchrotron Radiat. 10 (2003) 69–70. [19] M. Sánchez del Río, J. Dejus, XOP: a multiplatform graphical user interface for synchrotron radiation spectral and optics calculations, SPIE Proc. 3152 (1997) 148–157. [20] H.F. Nolting, C. Hermes, EXPROG: EMBL EXAFS data analysis and evaluation program package for PC/AT, European Molecular Biology Laboratory, DESY, Hamburg, Germany, 1993. [21] N. Binsted, S.S. Hasnain, State-of-the-art analysis of whole X-ray absorption spectra, J. Synchrotron Radiat. 3 (1996) 185–196. [22] F.W. Lytle, D.E. Sayers, E.A. Stern, Report of the international workshop on standards and criteria in X-ray absorption spectroscopy, Phys., B 158 (1989) 701–722. [23] W.N. Lipscomb, N. Strater, Recent advances in zinc enzymology, Chem. Rev. 96 (1996) 2375–2434. [24] K. Fukasawa, K.M. Fukasawa, Y. Iwamoto, J. Hirose, M. Harada, The HELLGH motif of rat liver dipeptidyl peptidase III is involved in zinc coordination and the catalytic activity of the enzyme, Biochemistry 38 (1999) 8299–8303. [25] G.N. George, B. Hedman, K.O. Hodgson, An edge with XAS, Nat. Struct. Biol., Suppl. 1 (1998) 645–647. [26] J.A. Cuff, M.E. Clamp, A.S. Siddiqui, M. Finlay, G.J. Barton, Jpred: a consensus secondary structure prediction server, Bioinformatics 14 (1998) 892–893.

[27] J. Cheng, A. Randall, M. Sweredoski, P. Baldi, SCRATCH: a protein structure and structural feature prediction server, Nucleic Acids Res. 33 (2005) w72–w76 (Web Server Issue). [28] N.J. Blackburn, R.W. Strange, L.M. McFadden, S.S. Hasnain, Constrained and restrained refinement in EXAFS data analysis with curved wave theory, Biochemistry 31 (1992) 12117–12125. [29] S. Cramer, in: D.C. Koningsberger, R. Prins (Eds.), X-ray absorption: Principles, applications, techniques of EXAFS, SEXAFS and XANES, John Wiley and Sons Inc, NY, 1988. [30] W. Liu, H.H. Thorp, Bond valence sum analysis of metal–ligand bond lengths in metalloenzymes and model complexes. 2. Refined distances and other enzymes, Inorg. Chem. 32 (1993) 4102–4105. [31] P.J. Riggs-Gelasco, T.L. Stemmler, J.E. Penner-Hahn, XAFS of dinuclear metal sites in proteins and model compounds, Coord. Chem. Rev. 144 (1995) 245–286. [32] J.B. Thoden, G.N. Phillips Jr, T.M. Neal, F.M. Raushel, H.M. Holden, Molecular structure of dihydroorotase: a paradigm for catalysis through the use of a binuclear metal center, Biochemistry 40 (2001) 6989–6997. [33] M.M. Benning, J.M. Kuo, F.M. Raushel, H.M. Holden, Three-dimensional structure of phosphotriesterase: an enzyme capable of detoxifying organophosphate nerve agents, Biochemistry 33 (1994) 15001–15007. [34] D.B. Borza, F.M. Tatum, W.T. Morgan, Domain structure and conformation of histidine-proline-rich glycoprotein, Biochemistry 35 (1996) 1925–1934. [35] W.S. Willett, S.A. Gillmor, J.J. Perona, R.J. Fletterick, C.S. Craik, Engineered metal regulation of trypsin specificity, Biochemistry 34 (1995) 2172–2180. [36] D.K. Mahnke-Zizelman, P.C. Tullson, R.L. Sabina, Novel aspects of tetramer assembly and N-terminal domain structure and function are revealed by recombinant expression of human AMP deaminase isoforms, J. Biol. Chem. 273 (1998) 35118–35125. [37] D. Martini, U. Montali, M. Ranieri-Raggi, A.R.M. Sabbatini, S.J. Thorpe, A.J.G. Moir, A. Raggi, A calpain-like proteolytic activity produces the limited cleavage at the N-terminal regulatory domain of rabbit skeletal muscle AMP deaminase: evidence of a protective molecular mechanism, Biochim. Biophys. Acta 1702 (2004) 191–198. [38] M. Ranieri-Raggi, A. Raggi, Effect of pH and KCl on aggregation state and sulphydryl groups reactivity of rat skeletal muscle AMP deaminase, Ital. J. Biochem. 33 (1984) 155–176. [39] A.L. Lamb, A.S. Torres, T.V. O'Halloran, A.C. Rosenzweig, Heterodimeric structure of superoxide dismutase in complex with its metallochaperone, Nat. Struct. Biol. 8 (2001) 751–755.