Vol. 55 No. 1/2008, 151–160 Regular paper on-line at: www.actabp.pl

Asymmetrical diadenosine 5',5''-P(1)P(4) tetraphosphate (Ap(4)A) hydrolases are key enzymes controlling the in vivo concentration of Ap(4)A--an important signaling molecule involved in regulation of DNA replication and repair, signaling in stress response and apoptosis. Sequence homologies indicate that the genome of the model plant Arabidopsis thaliana contains at least three open reading frames encoding presumptive Ap(4)A hydrolases: At1g30110, At3g10620, and At5g06340. In this work we present efficient overexpression and detailed biochemical characteristics of the AtNUDX25 protein encoded by the At1g30110 gene. Aided by the determination of the binding constants of Mn(Ap(4)A) and Mg(Ap(4)A) complexes using isothermal titration calorimetry (ITC) we show that AtNUDX25 preferentially hydrolyzes Ap(4)A in the form of a Mn(2+) complex.


INTRODUCTION
Bis (5´-nucleosidyl) tetraphosphates are found in both eukaryotic and prokaryotic organisms at submicromolar concentrations (Garrison & Barnes, 1992;Guranowski, 2003).They were initially characterized as toxic side-products of protein synthesis, but more recently have been proposed to play a range of roles in processes such as control of DNA replication and repair, signaling in stress response, and apoptosis (for reviews see McLennan, 2000;2006).Their in vivo concentration seems to be precisely controlled by a set of different enzymes including Ap 4 A hydrolases (for reviews see Guranowski, 2000;2004).Asymmetrical Ap 4 A hydrolases [asymmetrical bis (5´-nucleosidyl) tetraphosphatases, EC 3.6.1.17]are enzymes which hydrolyse bis (5´-nucleosidyl) polyphosphates (Np 4-6 N') in such a way that a nucleosidyl triphosphate (NTP) is always one of the products.The favoured substrates of these enzymes are Ap 4 A, Gp 4 G and other bis (5´-nucleosidyl) tetraphosphates.Such enzymes have been identified in eukaryotes, among others in human and plants, and recently also in prokaryotes, including invasive bacteria (Warner & Finamore, 1965;Ogilvie & Antl, 1983;Jakubowski & Guranowski, 1983;Conyers & Bessman, 1999).The hydrolytic activity of asymmetrical Ap 4 A hydrolases is based on a highly conserved sequence motif called the Nudix box (Gx 5 Ex 7 REUxEExGU, where x is any amino acid and U is a bulky hydrophobic residue) (Bessman et al., 1996).A conserved

2008
B. Szurmak and others Tyr residue located 16-18 amino acids downstream of the Nudix box is considered responsible for the preference of Ap 4 A hydrolases for Ap 4 A as their main substrate (Dunn et al., 1999).Ap 4 A hydrolases may be divided into two groups according to sequence homology.The plant and bacterial Ap 4 A hydrolases have sequences similar to each other, but are substantially different from the human and animal ones.A comparison of structural data available for Ap 4 A hydrolases from Caenorhabditis elegans and Lupinus angustifolius has shown that, despite of differences in protein sequences, these two proteins contain similar catalytic centers of three conserved glutamates in a spatial proximity (Swarbrick et al., 2000;Maksel et al., 2001;Bailey et al., 2002).Two of these glutamates are within the Nudix motif and one is located elsewhere in the sequence.
A strict requirement for divalent metals to maintain their hydrolytic activity is a characteristic feature of all Nudix hydrolases (Mildvan et al., 2005).In many cases more then one cation is needed to activate the enzyme.The precise roles of the metal ions in Nudix enzymes are not fully understood.It is proposed that one of the cations is bound to a conserved amino acid from the Nudix box, while the others may be bound to the oligophosphate substrate, orient the incoming nucleophile or neutralise the anionic leaving group (Mildvan et al., 2005).
Since the announcement of its complete genome sequence, Arabidopsis thaliana has become an important model organism for plant molecular biology and genetics.Sequence homologies indicate that the Arabidopsis thaliana genome contains at least three open reading frames encoding presumptive Ap 4 A hydrolases: At1g30110, At3g10620, and At5g06340.The protein products of two of these genes, At1g30110, At3g10620, have been obtained very recently as hexahistidine-tagged fusion proteins AtNUDX25 and AtNUDX26, respectively (Yoshimura et al., 2007).Their activity toward different Nudix hydrolase substrates has been checked in the presence of 5 mM Mg 2+ ions.Under such conditions the highest hydrolytic activity was observed for Ap 4 A hydrolysis by AtNUDX25 but it was very low with the V max of only 26.4 nmol min -1 mg protein -1 .
In this work we overexpressed of At-NUDX25 using pQE80 expression vector and performed a detailed biochemical characteristics of this protein with the main focus on its activation by different metal ions.Aided by the determination of the binding constants for the Mn(Ap 4 A) and Mg(Ap 4 A) complexes using isothermal titration calorimetry (ITC) we show that AtNUDX25 efficiently hydrolyzes Ap 4 A in the form of a Mn 2+ complex.

Materials.
The A. thaliana cDNA library was from Clontech.Oligonucleotide primers for polymerase chain reactions (PCR) were provided by the DNA Sequencing and Oligonucleotide Synthesis Service at the Institute of Biochemistry and Biophysics, Polish Academy of Sciences (Warszawa).Enzymes used in standard cloning reactions and Taq polymerase were from Fermentas.The pGEM-T Easy vector system I was supplied by Promega.The Escherichia coli expression vector pQE 80 and QIA-express kit type IV were from Qiagen.The E. coli strain Rosetta (DE3) was from Novagen.The complete EDTA-free protease inhibitor cocktail was from Roche Applied Science (Warszawa, Poland).All biochemicals and calf intestinal alkaline phosphatase were obtained from Sigma.
Cloning.The cDNA corresponding to the genomic sequence of At1g30110 was amplified from a commercial A. thaliana cDNA library with forward and reverse primers incorporating KpnI and PstI restriction sites, respectively.The PCR product was gel-purified and ligated into the pGEM-T Easy vector.The resulting plasmid was digested with KpnI and PstI, the insert was gel-purified, then ligated into the pQE80 vector and verified by sequencing.The resulting plasmid, pQE80At1g30110, with T7 lac promotor and a His-tag fusion upstream of the coding sequence was used to transform E. coli strain Rosetta (DE3) for expression.
Expression and purification of recombinant AtNUDX25 protein.A single colony was used to inoculate 10 ml of LB medium containing 100 µg/ml ampicillin and 24 µg/ml chloramphenicol.The culture was grown overnight at 37 o C and 0.25 ml of the overnight culture was transferred to 50 ml of Luria-Bretani (LB) medium including the antibiotics, and grown to OD 600 of 0.6.The expression was induced by adding IPTG to the final concentration of 0.5 mM, then the culture was incubated for additional 3 h.The cells were harvested by centrifugation, washed with an isotonic saline solution, resuspended in 1 ml of lysis bufer (0.5 M NaCl, 0.1 M Tris/HCl, pH 8.0, 5 mM β-mercaptoethanol plus 50 µl of the complete EDTA-free protease inhibitor cocktail), subjected to 4 or 5 freeze and thaw cycles (liquid N 2 /room temp.), and centrifuged at 14 000 r.p.m. for 10 min at 4 o C. The supernatant was then mixed with 0.25 ml of 50% nickel-nitrilotriacetic acid agarose (Qiagen).The recombinant protein was eluted with 250 mM imidazole, and dialysed overnight as described previously (Dobrzańska et al., 2002).Unless indicated otherwise, either 5 mM β-mercaptoethanol or 1 mM DTT were used during the purification procedure.

Molecular mass determination. Electrospray ionization mass spectrometry (ESI-MS) was used to
An Ap 4 A hydrolase of Arabidopsis thaliana determine the molecular mass of the recombinant protein.A sample of the protein was desalted using Vydac C18 HPLC column, then injected into a Q-Tof1 spectrometer (Micromass) at 4 µl/min, using a Hamilton syringe pump.Bovine pancreatic trypsin inhibitor was used as an internal mass standard.Positive ion spectra were deconvoluted using the MaxEnt 1 program (Waters Corp.).
Enzyme assay and kinetic studies.The progress of enzymatic reactions was followed either colorimetrically, at 800 nm, measuring inorganic phosphate liberated from the products of Ap 4 A hydrolysis by calf intestine phosphatase (Ames & Dubin, 1960) or by ESI-MS in the negative ion mode, as described previously (Dobrzańska et al., 2002).Briefly, the reaction mixture (20-50 µl) contained 50 mM Tris/HCl, pH 8.0, 3.5 mM MnCl 2 , 1 mM DTT, variable Ap 4 A concentrations ranging from 0.025 to 5.5 mM, 2-4 units of alkaline phosphatase, and 20 ng of the hydrolase.Reactions were stopped either by adding 280 µl of 10 mM EDTA or by heating at 95 o C for 4 min.Calf intestine phosphatase was either omitted (negative ion spectra) or added after the reaction had been terminated by heating.In several cases Ap 4 A (0.02 mM) the hydrolysis was followed spectrophotometrically.This general assay is based on an increase of A 260 as a result of hydrolysis of dinucleoside polyphosphates (Jakubowski & Guranowski, 1983).
Isotermal titration calorimetry.The stability constants of Mn 2+ and Mg 2+ complexes of Ap 4 A in 0.1 M Hepes buffer (pH 7.4), were measured at 25 o C on a Microcal OMEGA ultrasensitive titration calorimeter (MicroCal Inc.).The solutions in the cell were stirred using a syringe at 400 r.p.m.Portions of the titrant (5 µl) were injected over 20 s with an interval of 240 s between injections from a 250 µl injection syringe into the sample cells (containing Ap 4 A) in a series of several controlled pulses.The sample cell volume was 1.3611 cm 3 .The integrated heat effects of each injection were corrected by subtraction of the corresponding integrated heat effects of MnCl 2 or MgCl 2 injection to pure buffer and the heat effects of buffer injection to the Ap 4 A solution.Solutions of reactants were degassed prior to measurements.The Ap 4 A concentration was determined by A 260 absorbance, using the absorption coefficient ε 260 = 27 100 M -1 ⋅ cm -1 (Holler et al., 1983).The experimental data obtained from calorimetric titrations (five experiments done with two different sets of stock solutions) were analyzed assuming the model of a single set of identical sites (ITC Tutorial Guide).

Sequence analysis
The amino-acid sequences of all putative Ap 4 A hydrolases encoded by the A. thaliana genomic sequences At1g30110, At3g10620, and At5g06340, were compared using the Blast 2.0 program (National Center for Biotechnology Information, Bethesda, MD, USA) with the sequence of the best characterized plant Ap 4 A hydrolase from Lupinus angustifolius L. (accession number U89841).The search revealed that all three sequences resembled the sequence of the lupin enzyme, with at least 53% of identical amino acids.All amino-acid residues shown to be crucial for the lupin Ap 4 A hydrolase catalysis (Glu55, Glu59, Glu125) and specificity (Tyr77) were also conserved in the A. thaliana sequences (Maksel et al., 2001).A multialignment of the above proteins is presented in Fig. 1.The least similar of these protein sequences is that encoded by the At1g30110 gene and contains two unique blocks of amino-acid residues (aa 107-124 and 161-175) with no homology to the L. angustifolius Ap 4 A hydrolase.

Expression and purification of recombinant Ap 4 A hydrolase
The His 6 -tagged At1g30110 gene product was expressed in E. coli transformed with the pQE80At1g30110 plasmid.Affinity chromatography on an immobilized-nickel column yielded a highly pure recombinant protein as revealed by SDS/PAGE (Fig. 2).The slightly higher than expected molecular mass of the major SDS/PAGE band (21.6 kDa) is typical for Nudix hydrolases, in particular for Ap 4 A hydrolases, which migrate slowly during electrophoresis (Abdelghany et al., 2001).

Electrospray ionization mass spectrometry (ESI-MS)
Mass spectrometry was used to precisely determine the molecular mass of the recombinant protein.A single peak with molecular mass of 21 671.25 ± 1.8 Da was observed.This mass is identical to the calculated mass of hexahistidine-tagged At1g30110 gene product (21 671.14 Da).In rare cases, we observed an additional peak at 21 823.50 ± 1.8 Da, corresponding to a mass increase of 152.25 Da.The peak disappeared after treating the sample with DTT and thus was assigned to the formation of a mixed disulfide with two 2-mercaptoethanol molecules per protein.Such samples were discarded.

Initial characterization of the recombinant protein in the presence of Mn 2+ ions
Electrospray mass spectrometry of negative ions was applied for an initial characterization of the recombinant protein specificity.This method allows direct observation of the negatively charged substrates and products of the enzyme without a need to isolate them from the reaction mixture (Dobrzańska et al., 2002).Optimal conditions (buffer, pH and metal ion concentration) were searched for using a set of substrates including Ap 3 A, Ap 4 A, Ap 5 A, Ap 6 A, NADH, NAD, ADP-ribose, FAD, and 5-phosphoribosyl 1-pyrophosphate (PRPP) (Fisher et al., 2002).While almost no enzymatic activity was observed for any of these substrates in the presence of magnesium ions, products of Ap 4 A hydrolysis, namely ATP and AMP, were clearly seen when manganese ions were used in the reaction mixture.A mass spectrum obtained for the hydrolysis of 2.0 mM Ap 4 A in the presence of Mn 2+ ions (3.5 mM) and 20 ng of AtNUDX25 at pH 8.0 is presented in Fig. 3.Among the other substrates tested only a slight cleavage of Ap 5 A was observed, with ATP among the reaction products.In the presence of Mn 2+ (3.5 mM) and Ap 4 A (2.0 mM), the enzyme was active over a wide pH range, from 6.8 to 9.5, showing an optimum at around pH 8.0 (not shown).Relative activities of the enzyme towards Ap 4 A, Ap 5 A and Ap 6 A were also measured using the colorimetric assay.This indirect test confirmed Ap 4 A cleavage, and indicated enzymatic hydrolysis of Ap 5 A and Ap 6 A, at 10.4% and 6.5% of the rate of Ap 4 A hydrolysis, respectively.

Activation of AtNUDX25 by divalent cations
The results presented in Table 1 summarize the data on activation of AtNUDX25 by various divalent cations as obtained by colorimetric experiments.These data confirm the preliminary MS observations An Ap 4 A hydrolase of Arabidopsis thaliana that Mn 2+ provides optimal activation for AtNUDX25.In the presence of Ap 4 A (1.5 mM) no activity was seen with Mg 2+ , Cd 2+ , or Ca 2+ at 1 mM concentrations, however, the enzyme exhibited a residual activity in the presence of Zn 2+ ions.An increase of divalent cation concentrations to 5 mM resulted in weak enzyme activation for Cd 2+ , Mg 2+ , and Zn 2+ ions (ordered by increasing activity); no activity was seen in the presence of Ca 2+ ions.The enzyme was also inactive in the presence of 3.5 mM Mg 2+ and two low Ap 4 A concentrations, 0.1 and 0.02 mM (Table 1).

Stability constants of Ap 4 A complexes
Varied stabilities of Ap 4 A complexes with metal ions have been proposed in the literature.The constants provided previously by calorimetry (Tanner et al., 2002) appear to be significantly overesti-mated, as discussed by Wszelaka-Rylik et al. (2007).A spectroscopic determination of the Mn(Ap 4 A) stability constant allows for the same conclusion (Conyers et al., 2000).The conditional binding constant c K for the Mn(Ap 4 A) complex was determined here by ITC as 1.7 ± 0.2 × 10 4 Mˉ1, corresponding to a dissociation constant K d = 59 ± 8 µM.The analogous constants for the Mg(Ap 4 A) complex, also determined here, were 5.5 ± 0.2 × 10 3 Mˉ1 and 183 ± 8 µM, respectively.The stabilities of these complexes are lower than those previously reported by Tanner et al. (2002) and Conyers et al. (2000).

Kinetic properties of the recombinant enzyme in the presence of Mn 2+ ions
A series of experiments was performed to analyze in detail the influence of substrate and Mn 2+ The concentrations of the Ap 4 A hydrolysis products, ATP and AMP, were estimated after terminating the enzymatic reaction by heating, using indirect colorimetric assays which measure the concentration of phosphate released from these mononucleotides by alkaline phosphatase.In some experiments the phosphatase was added during the enzymatic reaction to avoid the hypothetical product inhibition of AtNUDX25.The presence of alkaline phosphatase did not affect the rates of the catalyzed reaction.The dependence of the rate of hydrolysis on Mn 2+ at the constant Ap 4 A concentration of 1 mM (expressed as inverse logarithm of the total Mn 2+ concentration, pMn) is presented in Fig. 4. The initial concentrations of the Mn(Ap 4 A) complex and of free Ap 4 A in these enzymatic reactions are also plotted.These concentrations were calculated from the binding constant determined above here.The curve generated by plotting the rate of Ap 4 A hydrolysis as a function of pMn has two sigmoidal sectors of opposite inclinations.The optimal enzyme activity range is very narrow.The maximal activity of the enzyme in this series of experiments, 27.0 ± 1.0 µmol/min per mg, coincided with total Mn 2+ and Ap 4 A concentrations of 1 mM.At such concentrations the Mn(Ap 4 A) complex dominates and the concentration of its components (denoted by square brackets) [Mn 2+ ] = [Ap 4 A] = 83 µM.An excess of either [Mn 2+ ] or [Ap 4 A] over [Mn(Ap 4 A)] results in a decrease of the reaction rate.The fits of the sigmoidal sectors of the pMn curve to the Hill equation (Hill, 1910;Ac-erenza & Mizraji, 1997) yielded apparent K d values of 195 ± 13 µM with n 1 = 2.0 ± 0.2 for the low Mn 2+ branch and of 1.82 ± 0.12 mM with n 2 = 2.9 ± 0.8 for the high Mn 2+ branch (n 1 and n 2 are respective Hill coefficients).The n 1 value indicates that at least two Mn 2+ binding sites must be occupied to activate the enzyme.The value of n 2 was determined with a much lower accuracy, not allowing firm conclusion about the stoichiometry of the manganese-related inhibition.
Curves generated by plotting the velocities of Ap 4 A hydrolysis by AtNUDX25 as a function of to-

Table 1. Effects of divalent metal ions (added as chlorides) on the activity of AtNUDX25 measured in the presence of Ap 4 A.
All experiments were done using the colorimetric method except for measurements in the presence of 0.1 or 0.02 mM Ap 4 A, which were followed spectrophotometrically.The dependence of V max of Ap 4 A hydrolysis by AtNUDX25 on log concentrations of Mn(Ap 4 A) is presented in Fig. 7.The fitting of this sigmoidal relationship to the Hill equation yielded K d of 2.5 ± 1 mM, and the corresponding Hill coefficient n 3 = 0.94 ± 0.17.Extrapolations of V max values to high and low complex concentrations yielded the limiting maximal and minimal velocities of 104 ± 18 and 4.4 ± 2 µmol/min per mg, respectively.

DISCUSSION
Asymmetrical Ap 4 A hydrolases are suggested to be key enzymes participating in the regulation of intracellular concentration of Ap 4 A, and to a smaller extent, of other adenosine polyphosphates (Guranowski, 2000).Despite their importance in many organisms, no efficient Ap 4 A hydrolases have been characterized so far in the model plant -A.thaliana.At1g30110 is one of the three genomic sequences from A. thaliana proposed to encode an Ap 4 A hydrolase.The full-length At1g30110 gene, including six introns, comprises 1940 bp, and the open reading frame encodes a 175-amino-acid protein with a deduced molecular mass of 21 671.14Da and pI 4.8.We have developed an efficient method for bacterial overexpression of the AtNUDX25 protein using pQE 80 expression system.The recombinant protein is capable of efficiently degrading Ap 4 A in the cleavage pattern typical for Ap 4 A hydrolases (Lobatón et al., 1975;Jakubowski & Guranowski, 1983).The alkaline pH optimum of the enzyme is typical for asymmetrically acting Ap 4 A hydrolases.Under the conditions used in this study none of other standard Nudix hydrolase substrates, such as Ap 3 A, NADH, NAD, ADP-ribose, PRPP and FAD, was hydrolysed by At-NUDX25.A notable difference between this enzyme and the plant hydrolase studied earlier is in the hydrolysis of substrates with more than four phosphate groups.AtNUDX25 digested Ap 5 A and Ap 6 A much more slowly than did the other plant Ap 4 A hydrolase (Jakubowski & Guranowski, 1983).The rates of Ap 5 A and Ap 6 A hydrolysis were only 10.4% and 6.5% of the rate of Ap 4 A hydrolysis, in comparison to 42% and 34%, respectively, reported for the lupin enzyme (Jakubowski & Guranowski, 1983).In this regard AtNUDX25 resembles the Caenorhabditis elegans Ap 4 A hydrolase (Abdelghany et al., 2001).
The presence of metal ions, often more than one per enzyme molecule, is a prerequisite for activation of all Nudix enzymes, but the number and type of the cations needed for efficient catalysis varies depending on the subtype of the hydrolases (for a review see Mildvan et al., 2005).In most cases, Mg 2+ ions have been shown to be preferred by eukaryotic Ap 4 A hydrolases (Guranowski et al., 2000).Several of these hydrolases, although preferentially activated by Mg 2+ ions, were shown to maintain  their activity in the sole presence of Mn 2+ ions (Jakubowski & Guranowski, 1983).Recently, very inefficient Ap 4 A hydrolysis was detected by others for AtNUDX25 in the presence of 5 mM magnesium ion (Yoshimura et al., 2007).Table 1 presents the effect of different divalent cations on the Ap 4 A hydrolysis by AtNUDX25.The strong preference of AtNUDX25 for Mn 2+ ions is remarkable.We were unable to observe any hydrolysis of 1.5 mM Ap 4 A in the presence of Ca 2+ .The enzyme required high (5 mM) Mg 2+ , Cd 2+ , and Zn 2+ concentrations to cleave some substrate (<10%), but never reached the activity provided by Mn 2+ ions.Since the human Mg 2+ -dependent Ap 4 A hydrolase was strongly inhibited when substrate concentrations exceeded 10 µM (Hankin et al., 1995), we additionally tested the activity of At-NUDX25 in the presence of 3.5 mM Mg 2+ and two low Ap 4 A concentrations, 0.1 and 0.02 mM, using a more sensitive spectrophotometric assay (Jakubowski & Guranowski, 1983).In both cases we failed to detect even a trace of Ap 4 A hydrolysis.Therefore we can state that Mn 2+ , and not Mg 2+ , is a true activator of AtNUDX25.Recently, an enzyme from Deinococcus radiodurans was characterized which showed a strong Mn 2+ dependence of Ap n A hydrolysis, with Mg 2+ ions at 15 mM giving only 5% of the activity accomplished by 2 mM Mn 2+ (Fisher et al., 2006).
The involvement and number of Mn 2+ ions participating in Ap 4 A hydrolysis was studied earlier for the hydrolase from Bartonella bacilliformis, which required three Mn 2+ ions.One of them was proposed to be substrate-bound (Conyers et al., 2000).Similarly, two Mn 2+ ions were found to activate another Nudix hydrolase, 8-oxo-dGTPase (MutT) (Frick et al., 1994).The kinetic data presented above indicated that at least two Mn 2+ ions participated cooperatively in the AtNUDX25-catalyzed reaction, when [Ap 4 A] was in excess over [Mn 2+ ] (Fig. 4).This view is supported both by the Hill coefficient value n 1 = 2.0, and by an absence of systematic deviations of the fit.Interestingly, the maximum of enzymatic activity, regardless of the concentration of Ap 4 A and Mn 2+ , always coincided with their equimolarity (Figs.4-6), indicating that one Mn 2+ ion interacted with Ap 4 A directly.It is generally known that enzymes involved in the hydrolysis of oligophosphates prefer as true substrates their complexes with M 2+ rather than free ligands (Toscano et al., 2003).For AtNUDX25 this view is confirmed by the analysis presented in Fig. 7, which shows that the maximum activity of the enzyme, measured at equimolar Ap 4 A and Mn 2+ , depends directly on the concentration of the Mn(Ap 4 A) complex.The value of n 3 = 0.94 ± 0.17 for the fitted Hill curve indicates a simple, non-cooperative interaction.The dissociation constant for the Mn(Ap 4 A) complex interaction with the enzyme, provided by this calculation, is 2.5 ± 1 mM.This value is somewhat underestimated, due to an inhibition by free Ap 4 A and free Mn 2+ , always present as a result of the relatively low stability of the Mn(Ap 4 A) complex (see Figs. 4 and 6).A semiquantitative analysis of the inhibitory branches of the kinetic curves at high Ap 4 A, presented in Figs. 5 and 6 suggests, however, that the error in peak velocity determinations, introduced by this effect, is small, approx.3-5%.The V max value for Mn(Ap 4 A) as substrate, determined from the fit, was 104 ± 18 µmol/min per mg.
These facts suggest a situation akin to that found for the Bartonella bacilliformis Ap 4 A hydrolase and MutT, where the Mn(Ap 4 A) and Mn(8-oxo-dGTP) complexes, respectively, were proposed to be true substrates for these enzymes, while the remaining Mn 2+ ion(s) interacted directly with the protein residues (Frick et al., 1994;Conyers et al., 2000).The latter view is also supported by the selectivity of AtNUDX25 activation, which is certainly due to the protein-based metal binding site(s).The formation of the Mn(Ap 4 A) complex may not be responsible for the selectivity of AtNUDX25 activation.Ap 4 A alone is not so strongly selective for Mn 2+ , as indicated by the similar stabilities of its Mn 2+ , Mg 2+ , and Zn 2+ complexes: K d values of 59, 183, and 138 µM, respectively; the latter value was recalculated from published data (Wszelaka-Rylik et al., 2007).One has to note that our protein contains a hexahistidine affinity tag, which might provide a binding site for Mn 2+ ions.The chemistry of such interactions has not been reported, but a study of Ni 2+ interactions with the His-tag suggests that a corresponding interaction with Mn 2+ ions would not be effective (Valenti et al., 2006).
A series of Ap 4 A titrations at fixed Mn 2+ levels demonstrated that an excess of Ap 4 A concentration over Mn 2+ concentration resulted in a partial inhibition of the AtNUDX25 activity (Figs. 5 and  6).The overall profiles of such Ap 4 A titrations were analogous to the profile of Mn 2+ titration presented in Fig. 4, but the branches of these curves could not be fitted to the Hill equation (Acerenza & Mizraji, 1997).This complex behavior seems to be due to overlapping effects of the cooperativity between Ap 4 A and Mn 2+ , to form free and enzyme-bound Mn(Ap 4 A) complexes, and of the competition between Ap 4 A and AtNUDX25 for Mn 2+ ions.Moreover, at low Ap 4 A concentrations AtNUDX25 may be activated by Ap 4 A, because it will first remove the inhibitory excess of Mn 2+ ions.
We have previously reported that another Nudix protein, AtNUDX1, is active toward NADH in the presence of 5 mM Mn 2+ (Dobrzańska et al., 2002).Later, Klaus et al. (2004) reported that the activity fell drastically as Mn 2+ concentration approached the usual 5 µM physiological range, and An Ap 4 A hydrolase of Arabidopsis thaliana at that concentration the enzyme activity was only 0.001% of that found at 5 mM.An extrapolation of AtNUDX25 activity from millimolar to micromolar concentrations may be attempted by comparing the saturation of Ap 4 A by Mn 2+ at these two concentration ranges.This leads to an extrapolated activity of ATNUDX25 at 5 µM substrate as low as 0.025% of that interpolated for 5 mM, thus about 0.017 µmol/ min/mg.The same reason may be responsible for the behavior of AtNUDX1.Of course, this extrapolation is very tentative, due to the complicated and only partially elucidated character of AtNUDX25 inhibition by free Ap 4 A.
Our experiments characterised the AtNUDX25 protein as an atypical eukaryotic Ap 4 A hydrolase, which is most selectively activated by manganese ions.In general, the complicated character of the catalytic process of Nudix, and in particular of Ap 4 A hydrolases (Mildvan et al., 2005), makes predictions of their in vivo role very speculative, especially as neither the presence nor the level of Ap 4 A in plant cells have ever been reported.Low levels (0.05-5 µM) of various oligophosphates, including Ap 4 A, have been shown to exist in other eukaryotic cells (Garrison & Barnes, 1992).The intracellular concentration of Mn 2+ is in the micromolar range (Schinkmann & Blenis, 1997).When the intracellular Mn 2+ and Ap 4 A concentrations in eukaryotic organisms are taken into account, the catalytic power of AtNUDX25 must be low, as estimated above.However, it might be important that AtNUDX25 hydrolyzes Ap 4 A when its concentration approaches the concentration of Mn 2+ ions.We can therefore speculate that the protein acts in signal transduction to record intracellular manganese concentration with respect to Ap 4 A or contrariwise in the case of spikes of their concentrations.Proteins behaving in such a fashion are known to respond to Ca 2+ spikes (Berridge et al., 2003).

Figure 1 .
Figure 1.Sequence alignment of putative Ap 4 A hydrolases from Arabidopsis thaliana encoded by genomic sequences At3g10620, At5g06340 and At1g30110 with the well-characterized Ap 4 A hydrolase from Lupinus angustifolius L. (accession number U89841).Residues strictly conserved among the four protein sequences are shaded black and those shared among three of the four sequences are shaded grey.The catalytic Nudix motif is underlined.The glutamate residues crucial for catalysis by lupin Ap 4 A hydrolase are marked with black dots.The alignment was performed using MultiAlign program (INRA).

Figure 3 .
Figure 3. Substrate specificity of AtNUDX25 analyzed using negative-ion mass spectrometry.Substrate solutions at 2 mM concentration were tested either before or after addition of the enzyme.The positions of standards are indicated.

a
Related to the rate of Ap 4 A hydrolysis obtained in the reaction mixture containing 1.5 mM Ap 4 A and 1.5 mM MnCl 2 .

Figure 4 .
Figure 4. AtNUDX25 activity at total Ap 4 A concentration of 1 mM, as a function of Mn 2+ ion concentration, along with the fit to the Hill equation (-).Initial concentrations of Mn(Ap 4 A), free Mn 2+ , and free Ap 4 A are shown for comparison.The activity was determined colorimetrically at 800 nm(Ames & Dubin, 1960).

Figure 5 .
Figure 5. AtNUDX25 activity as a function of total Ap 4 A concentration.The data are means of triplicate assays.The data points outside the maxima, as well as the standard errors of the mean (typically within 4% of the respective values) are omitted for clarity.The total concentrations of MnCl 2 in individual curves are indicated.The activity was determined colorimetrically at 800 nm(Ames & Dubin, 1960).

Figure 6 .
Figure 6.AtNUDX25 activity as a function of Ap 4 A concentration, at the MnCl 2 concentration of 1.5 mM, compared with initial concentrations of Mn(Ap 4 A), free Mn 2+ and free Ap 4 A.The activity was determined colorimetrically at 800 nm(Ames & Dubin, 1960).

Figure 7 .
Figure 7. Peak AtNUDX25 activities as a function of Mn(Ap 4 A) concentrations, along with the fit to the Hill equation (-).