V. Bouriotis, Department of Biology, Division of Applied Biology and Biotechnology, University of Crete, PO Box 1470, Heraklion 711 10, Crete, Greece. Fax: + 30 81 394408; Tel.: + 30 81 394375; E-mail: Bouriotis@nefeli.imbb.forth.gr
The gene encoding alkaline phosphatase (AP) from the psychrophilic strain TAB5 was cloned, and its nucleotide sequence was determined. A single open reading frame consisting of 1125 base pairs which encodes a polypeptide consisting of signal peptide of 22 amino acids and a mature protein of 353 amino acids was identified. The deduced protein sequence of AP exhibits a 38% identity to the AP III and AP IV sequences of Bacillus subtilis and conserves the typical sequence motifs of the core structure and active sites of APs from various sources. Based on the crystal structure of the mutated Escerichia coli AP D153H, a homology-based 3D model of the TAB5 AP was constructed on the basis of which various features of the enzyme amino-acid sequence can be interpreted in terms of potential psychrophilic adaptations.
The AP gene was expressed in E. coli BL21(DE3) cells, the recombinant protein was isolated to homogeneity from the membrane fraction of the cells and its properties were examined. The purified TAB5 AP shows typical features of a cold enzyme: high catalytic activity at low temperature and a remarkable thermosensitivity.
The use of this heat-labile enzyme, for dephosphorylation of nucleic acids, simplifies dephosphorylation protocols.
Alkaline phosphatases (APs) are dimeric, zinc-containing nonspecific phosphomonoesterases that exist in various organisms from bacteria to mammals . Sequence comparisons of APs from a variety of species combined with structural information from the Escherichia coli AP [2–4], suggest that the functionally important domains are conserved [5–7]. However, although the amino-acid sequences of AP from bacteria and mammals show sequence conservation of 25–30%, mammalian APs are 10–20 times more active than the E. coli enzyme . Comparisons between the E. coli and the mammalian enzymes reveal two striking differences close to the active site: in the E. coli AP, two of the catalytic residues (153 and 328) are Asp and Lys, while in the mammalian enzymes the equivalent residues are both His. Interestingly, the Asp→His substitution at position 153 in the E. coli AP results in a mutant enzyme (D153H), which exhibits typical properties of a mammalian AP, including a shift in the pH of optimal activity, low activity in the absence of Mg2+ ions, a time dependent activation by Mg2+ ions and a reduction in activity in the presence of Zn2+ ions [9,10].
Although, the origin of the observed properties of the D153H mutant is not yet well understood, the molecular basis of increased enzymatic activity represents a problem which is both challenging and biotechnologically important. Generally, higher catalytic efficiencies at low and moderate temperatures, along with remarkable thermolability are properties of psychrophilic enzymes [11,12]. Therefore, an AP from psychrophiles exhibiting such properties could be of significant biotechnological interest, while on the other hand it could provide an attractive model system to study protein adaptations to cold.
Recently, crystal structures have been elucidated for bacterial cold-active enzymes such as malate dehydrogenase , citrate synthase , triose phosphate isomerase , and amylase . These studies provide new insights into the nature of psychrophilic adaptations in proteins. For example the catalytic and physicochemical properties [17,18], of psychrophilic enzymes had been earlier interpreted in terms of an increased structural flexibility which enables a more efficient use of the lower thermal energy of the environment. The requirement of an inherently increased flexibility however, has not been always confirmed by the recent crystallographic studies. It appears that frequently it is a relative increase in local flexibility (rather than in the overall protein flexibility) which constitutes frequently the adaptation required for psychrophilic enzyme function. Based on extensive sequence analyses, several types of sequence adaptations have been proposed in the past to be associated with increased structural flexibility, e.g. the clustering of Gly residues close to catalytic and metal binding sites , and/or the decrease of the sequence content in rigid Pro residues, and/or the decrease of electrostatic and hydrogen bonding interactions. Other potential psychrophilic adaptations deduced from the crystallographic structures include the presence of favorable surface charge distributions for substrate and cofactor binding and an increase of active site accessibility. Shorter surface loops in psychrophilic enzymes had been earlier proposed to increase the substrate accessibility of the catalytic site thus enhancing catalytic activity .
In this paper, we report cloning, sequencing and expression of the gene encoding an AP from the Antarctic psychrophile TAB5, as well as purification and partial characterization of the recombinant enzyme. Furthermore, a 3D homology-based model of the enzyme was used to identify potential psychrophilic adaptations of the enzyme sequence. The suitability of recombinant TAB5 AP for molecular cloning is also demonstrated.
Materials and methods
Biochemicals and reagents
All biochemicals were of analytical grade. DNA restriction and modifying enzymes, were from MINOTECH except for NdeI restriction enzyme, exonuclease III, T4 polynucleotide kinase (PNK) and calf intestinal alkaline phosphatase (CIP) which were supplied from New England Biolabs. Reactions with these enzymes were carried out as recommended by the suppliers. All chromatography media, molecular mass markers, the sequencing kit and S1 nuclease were obtained from Pharmacia Biotech. Mimetic Blue AP A6XL-Agarose was purchased from ACL (Isle of Man, UK). Ultrafiltration membranes were purchased from Amicon, whereas cellulose nitrate filters (filter type GV, pore size 0.22 µm) and poly(vinylidene difluoride) membranes were from Millipore.
Organisms and culture conditions
The psychrophilic TAB5 strain was isolated at the Dumont d’Urville antarctic station (location: 60°40′ S, 40°01′ E) and kindly provided to us by C. Gerday (Laboratory of Biochemistry, Institute of Chemistry, University of Liege, Belgium). This strain is able to grow from 0–20 °C and is therefore refered to as a psychrophilic bacterium. TAB5 was cultured aerobically at 8 °C for four days in Luria–Bertani broth (LB) supplemented with MgCl2 (1.3 g·L−1) and KCl (0.4 g·L−1). E. coli XL1-MRF, DH5α, and BL21(DE3) were used as cloning hosts and these strains were grown in LB or on LB plates at 25 °C, 30 °C or 37 °C. When required, antibiotics were added to media at the following concentrations: 100 µg·mL−1 ampicillin, 50 µg·mL−1 tetracycline.
Generally, all manipulations were performed according to standard methods . Plasmid DNA was isolated by the alkaline lysis method . TAB5 genomic DNA was isolated as described previously . To isolate the AP encoding gene of TAB5 a size fractionated genomic library containing 3 to 8-kb DNA fragments (from a partial Sau3A I digestion), cloned into the BamHI site of the vector pBluescript KSII (Stratagene) was transformed into E. coli DH5α. Colonies were transferred onto cellulose nitrate filters and were incubated in 1 m diethanolamine/HCl, pH 8.5, containing 5 mm 5-bromo-4-chloro-3-indolyl phosphate (BCIP) and 5 mm MgCl2. After 5–10 min incubation at room temperature, clones containing the AP-encoding gene from TAB5 were identified by their blue colour.
DNA sequence determination, sequence comparisons and 3D modelling
The sequence was determined in both orientations, using the dideoxy chain termination method  after generating exonuclease III/S1 nuclease deletions . DNA sequencing was performed with an automatic DNA sequencer (Pharmacia Biotech). Sequence data were compiled and analyzed using the gcg (Genetics Computer Group, Madison, WI, USA) software package  and software provided by the National Center for Biotechnology Information network service. clustal x and the programs pileup, blast, and bestfit as implemented in the gcg package were used for amino-acid sequence comparisons and alignments.
The 3D-model of the TAB5 psychrophilic AP was based on the crystal structure of the mutated E. coli AP D153H (PDB code 2anh ), because sequence alignments suggest that the TAB5 AP and E. coli AP mutant have in common an important His residue of the metal binding site (in the wild-type E. coli AP this residue is an Asp). APs from TAB5 and E. coli share an overall sequence similarity of 33%, so that homology based modelling of at least the core region of the TAB5 protein using the E. coli AP as a template structure can be reliably performed . Molecular modelling was performed with the modeler V3.0  program on a Silicon Graphics INDIGO2 workstation running irix 6.2. Energy minimizations with the xplor program  were used to improve the overall packing of side chains in the final model. The quality of the atomic model was evaluated using procheck. Protein structures were visualized with the rasmol program . Analysis of atomic contacts was performed using the programs contacts of the ccp4 suite . No attempts were made to model a dimeric form of the enzyme due to the lack of biochemical evidence supporting the presence of this form.
AP activity was measured spectrophotometrically utilizing p-nitrophenyl phosphate (pNPP) as the substrate . Reaction mixtures (1 mL) contained 1 m diethanolamine/HCl, pH 8.5, and 10 mm MgCl2 except where otherwise indicated, 10 mm pNPP, and enzyme as indicated. The release of p-nitrophenolate was monitored at 405 nm. All assays were carried out at 25 °C except where otherwise indicated. The reaction was stopped by the addition of NaOH at 1.2 m. One unit of enzyme activity is defined as the amount of enzyme which hydrolyzes one μmole of pNPP to p-nitrophenol in 1 min at 25 °C in a volume of 1 mL.
N-terminal sequencing was performed on samples of the purified recombinant enzyme, which had been electrophoresed on SDS/polyacrylamide gels. The protein was electroblotted onto a polyvinylidene difluoride membrane following standard procedures. The membrane was stained with Coomassie Brilliant Blue, and the region containing the AP protein was excised and sequenced at the University of Liege (Laboratory of Biochemistry, Institute of Chemistry B6).
Protein concentration was determined according to Bradford , using bovine serum albumin as standard. Denaturing gel electrophoresis was performed in 12.5% SDS/PAGE minigels according to Laemmli . Proteins were visualized by Coomassie Brilliant Blue R-250 staining.
Overexpression of the TAB5 AP gene
The T7 expression vector pRsetA (Invitrogen) was used to harbor and direct the synthesis of the following construct in E. coli strain BL21(DE3). The putative gene was amplified by polymerase chain reaction from pPhoA1.9 using the upstream primer 5′-d(GCTAGCATATGAAGCTTAAAAAAATTG)-3′ and the downstream primer 5′-d(CAAGTTTAGTTTTCATATGAG)-3′. An NdeI site (underlined) was engineered into the upstream primer to facilitate cloning. The downstream primer contained an NdeI site (underlined) that exists downstream of the native stop codon (nucleotide 1579, Fig. 1). PCR primers were synthesized using an Applied Biosystems DNA synthesizer. The PCR reaction mixtures were incubated on a Pelkin-Elmer thermal cycler for 30 cycles of 94 °C for 1 min, 49 °C for 1 min, and 72 °C for 1 min. The amplified product was isolated by agarose gel electrophoresis, gel purified and digested with NdeI restriction enzyme. The resulting NdeI/NdeI fragment was ligated into the NdeI site of pRsetA by standard methods using T4 DNA ligase. The ligation mixture was used to transform competent cells of E. coli strain XL1-MRF. The correct recombinant plasmid, called pN1, was used to transform E. coli BL21(DE3) cells.
For overexpression of the AP gene freshly transformed BL21(DE3) cells were grown in 50 mL LB containing 150 μg·mL ampicillin overnight at 28 °C. This culture was used to inoculate 30 L of LB medium containing 4 mm MgSO4, 0.4 mm ZnSO4, 10 mm KCl, 150 μg·mL−1 ampicillin, in a 30-L stainless steel fermentor (Bioengineering). The culture was grown to an D600 of 0.8, isopropyl thio-β-d-galactoside (IPTG) at 1 mm was added to the culture, and incubation continued at 28 °C to an D600 of 1.5–1.8. Cells were harvested by centrifugation using a continuous flow centrifuge (Sharpless, Rennwalt Ltd, UK), and frozen at −70 °C.
Purification of the recombinant TAB5 AP
All purification steps were carried out at 0–4 °C. Frozen cell paste (30 g) was thawed with 30 mL of 20 mm Tris/HCl, pH 7.6, 10 mm MgCl2 (Buffer A), disrupted by French press and centrifuged at 4 °C for 15 min at 5000 g to remove intact cells and cell debris. The cell extract was centrifuged at 105 000 g and the precipitate corresponding to the total membrane fraction resuspended in Buffer A containing 0.2 m NaCl and 2% Triton X-100. The mixture was extracted with continuous and gentle stirring at 4 °C for 4 h, and insoluble material was removed by centrifugation at 105 000 g for 1 h. The supernatant containing AP activity was retained for further fractionation. The precipitate was extracted once more under the same conditions and the suspension was centrifuged at 105 000 g for 1 h. The supernatant from this second extraction containing most of AP activity was joined with the first for futher fractionation.
The detergent-solubilized enzyme was applied onto a Q-Sepharose fast flow column (12.5 × 3.2 cm, 100 mL) previously equilibrated in buffer B (20 mm Tris/HCl, pH 7.6, 10 mm MgCl2, 0.2% Triton X-100, 10% glycerol). The column was washed with buffer B until no absorption at A280 was evident in the effluents and then developed with a linear gradient of NaCl (600 mL, 0–0.6 m) in buffer B. Fractions (6 mL) with AP activity corresponding to ≈ 0.25–0.35 m NaCl in the gradient were pooled (135 mL), concentrated with Amicon ultrafiltration cell and after dialysis against buffer B were loaded onto a Mimetic Blue AP A6XL-Agarose column (5.3 × 2.2 cm, 20 mL) previously equilibrated in buffer B. The column was washed with buffer B until no absorption at A280 was evident in the effluents and then developed with a linear gradient of KPO4, pH 6 (100 mL, 0–0.05 m) in buffer B. Fractions (2 mL) with AP activity corresponding to ≈ 10 mm KPO4 in the gradient were pooled (10 mL) and dialyzed against storage buffer (10 mm Tris/HCl, pH 7.6, 50 mm NaCl, 10 mm MgCl2, 0.5 mm ZnCl2, 50% glycerol). Purified enzyme preparation was stored at −20 °C. Fractions collected from each step of purification were assayed for AP activity and protein concentration (Table 1).
Table 1. Purification of recombinant TAB5 alkaline phosphatase. See Materials and methods for more details.
The plasmid vector Bluescript KSII was digested with various restriction enzymes using standard protocols. In general 10 μg of DNA in a volume of 100 μL was digested for 1 h with 50 U of restriction enzyme under the appropriate conditions. Following phenol/chloroform extraction and EtOH precipitation, 1.5 U of TAB5 AP was added to the digested DNA and the mixture was incubated in AP reaction buffer for 30 min at 25 °C, in a total reaction volume of 50 μl. After 10 min inactivation of the TAB5 AP at 55 °C the ligation efficiency of the phosphatase treated DNA ends was tested.
Cloning of the AP encoding gene
An AP encoding gene was isolated from a TAB5 gene library by screening for expression of AP activity in E. coli DH5α, which expressed only a weak background of intrinsic AP activity under the conditions used. About 10 000 clones were grown on ampicillin containing LB plates overnight at 25 °C to circumvent thermal denaturation of the cloned gene product. The colonies were subsequently transferred onto cellulose nitrate filters and were incubated in AP assay buffer containing 5 mm BCIP. By this chromogenic assay one colony which expressed AP activity was detected. The recombinant plasmid carrying the phoA gene was isolated and characterized by restriction analysis. The plasmid (pPhoA4.3) contained an insert of 4.3 kb. A probe for the phosphatase gene was prepared and hybridized to restriction digests of TAB5 genomic DNA (data not shown). The results indicate that the cloned gene corresponded to a single, uninterrupted DNA sequence. To select the smallest DNA fragment still encoding for AP activity, plasmid pPhoA4.3 was subjected to exonuclease III/S1 treatment to generate nested deletions. The deletion derivatives were tested on cellulose nitrate filters for expression of AP activity. The minimum deletion derivative (pPhoA1.9) which expressed AP activity contained a 1.9-kb TaqI/XbaI fragment.
Nucleotide sequence of phoA gene
The full 1852-bp DNA insert in pPhoA1.9 was sequenced in both directions (Fig. 1). An open reading frame was identified between nucleotides 301 and 1426. This ORF contains a start codon at nucleotide position 301 and is terminated by two stop codons at nucleotide positions 1426 and 1432 and one inverted repeat sequence that may function as transcriptional termination signal (underlined in Fig. 1). No typical procaryotic ribosome binding site for the initiation of translation has been identified. The translated amino-acid sequence from phoA is also shown in Fig. 1. The deduced primary stucture of the protein is 375 amino acids with a predicted molecular mass of 40 433 Da and an isoelectric point of 5.48. Hydropathy plot analysis revealed a hydrophobic segment near the N-terminus consisting of 22 amino acids with typical features of procaryotic signal peptide; the positively charged N-terminus is followed by regions containing residues with a high propensity for α-helix formation. The predicted cleavage site for the putative signal peptide is between residues 22 Ser and 23 Val. As for other extracellular proteins this signal peptide is predicted to be necessary for the translocation across the cytoplasmic membrane.
The deduced amino-acid sequence of TAB5 AP was compared with published sequences of other APs using the BLAST program. TAB5 AP shares the greatest similarity with the two Bacillus subtilis AP sequences AP III and AP IV with identity scores of 38%. Lower degrees of similarity with identity scores of 28–33% were found with the APs from three enteric bacteria (E. coli, Escherichia fergusonii and Serratia marcescens) and APs from Streptomyces cerevisiae, Schizosaccharomyces pombe and human intestinal. The residues forming the three metal binding sites, the phosphorylation site and catalytic site are conserved between the TAB5 and the E. coli APs (Fig. 2). However, residues Asp153 and Lys328, which form the Mg2+ coordination sphere in the E. coli AP, are replaced by His and Trp in the TAB5 AP, the B. subtilis AP III and AP IV  and the B. licheniformis AP . There is a reduced number of Gly and Pro residues in the psychrophilic enzyme (32 Gly and 8 Pro residues in the TAB5 AP vs. 45 Gly and 20 Pro residues in the E. coli enzyme) although the percent content of Gly in both sequences is roughly constant. The most drastic decrease of Pro residues occurs in the loops of the modeled enzyme. The Arg/(Arg + Lys) ratio which has been discussed in connection with enzyme stability , is lower in the psychrophilic enzyme (6/6 + 29 in the TAB5 protein vs 13/13 + 28 in the E. coli AP).
The TAB5 AP model
The homology based 3D model of the psychrophilic enzyme shows that only minor structural changes (such as deletions) with respect to the E. coli AP should be expected, and these should be mainly clustered in peripheral residues and loops. The number of salt bridges in the model is almost half of those found in the structure of the mesophilic E. coli AP. The total number of hydrogen bonds in the model is slightly decreased. On the other hand, there are no changes in aromatic contacts.
Although psychrophilic enzymes are frequently associated with longer polypeptide chains compared to their mesophilic counterparts , this is not the case for the psychrophilic phosphatase modeled in the present work. Compared to the majority of the APs in the SwissProt Database (including the E. coli enzyme) the TAB5 AP is quite small (40 kDa).
Overexpression of the AP gene in E. coli and purification of the recombinant AP protein
The putative gene was cloned into pRsetA by standard methods. Induced BL21(DE3) cells carrying the pN1 recombinant vector produced significant amount of AP activity whereas AP activity was not detectable in the pRsetA control (data not shown). When cell pellets from cultures grown under protein expression conditions were disrupted and centrifuged, only a small quantity of AP activity (≈ 10%) was detectable in the cytoplasmic fraction; however, the majority of the AP was found in the insoluble membrane fraction. Although distribution between those two fractions varied from preparation to preparation, purification of the AP protein was achieved using the membrane fraction.
The purification of AP is summarized in Materials and methods. The enzyme was purified to homogeneity employing two chromatographic steps with an overall yield of 18.9% and a purification factor of 241. The final enzyme preparation was shown by SDS/PAGE to be homogeneous with an apparent molecular mass of approximately 38 kDa (Fig. 3). This value is approximately 2.5 kDa lower than that calculated from the predicted amino-acid sequence (40.43 kDa). When purified AP was subjected to gel filtration on Superdex 200 HR column it was eluted as a single peak with an apparent molecular mass of about 480 kDa (data not shown), indicating that the native enzyme may exist as a multimer. When Triton X-100 was incorporated in the buffer (at concentrations ranging from 0.1 to 1%) the molecular mass was estimated to be 240 kDa.
The overexpressed TAB5 AP was isolated by SDS/PAGE and subjected to N-terminal amino-acid sequence analysis. Determination of the seven N-terminal residues revealed the following sequence: NH2-V-L-V-K-N-E-P which is identical to the predicted amino-acid sequence after the 22nd residue (bold in Fig. 1). The sequencing yield was more than 80%.
Properties of the recombinant TAB5 AP
The purified recombinant TAB5 AP was further characterized using pNPP as the substrate. The temperature optimum of the enzyme was 25 °C and the respective activities at 0, 5, 10, 15 and 20 °C were 38, 56, 68, 78, and 93% of that at the optimal temperature (25 °C). Figure 4 shows the temperature effect on the stability of the AP. The enzyme was preincubated for 15 min in the absence of substrate at temperatures ranging from 20 °C to 70 °C and then assayed in the presence of substrate at 25 °C. Preincubation of TAB5 AP for 15 min at 45 °C and at 50 °C resulted in 50% and 100% inactivation, respectively.
The pH dependence of the enzyme activity was examined using either 10 mm or 1 m Tris/HCl buffer in a pH range from 7.0 to 8.5 and 1 m diethanolamine/HCl buffer in the range between 7.5 and 10.5 at 25 °C. Maximum activity was observed in 1 m diethanolamine/HCl buffer, pH 8.5 (data not shown). The effect of divalent metal ions were examined for their influence on the enzyme activity. Similarly to all other members of the AP family, a divalent cation is required for activity , and Mg2+ ion is the most effective. The optimal concentration was determined to be 10 mm. However, Zn2+ ions exhibited an inhibitory effect to the TAB5 AP, even at 0.05 mm, most likely by replacing the essential Mg2+ from its specific binding site . CaCl2 and MnCl2 had no effect on the AP activity at concentrations up to 20 mm. TAB5 AP was inhibited 50% by as little as 0.1 mm EDTA and almost 100% by 2 mm EDTA. In the presence of Triton X-100 (at concentrations up to 0.5%) AP activity increased by a factor of 1.3. NaCl or KCl had no effect on the AP activity.
Dephosphorylation of DNA by recombinant TAB5 AP
The recombinant TAB5 AP was tested for its ability to remove 5′-terminal phosphates from DNA fragments. The plasmid vector Bluescript KSII was digested with various restriction enzymes, and the resulting fragment(s) treated with recombinant TAB5 AP as described in Materials and methods. Following treatment with T4 DNA ligase the degree of dephosphorylation of the DNA fragments was tested.
The suitability of TAB5 AP for molecular cloning was demonstrated (Fig. 6). When plasmid DNA was cleaved by EcoRI and dephosphorylated with TAB5 AP, it could no longer be religated (Fig. 6, lane 3). To show that TAB5 AP-treated EcoRI ends are completely amenable to ligation, T4 PNK was used to restore the 5′ phosphate groups and then T4 DNA Ligase was added. Following PNK and Ligase treatment, a recircularized plasmid was obtained which was cleavable with EcoRI as before (Fig. 6, lanes 4, 5). Under the conditions used, dephosphorylation was also very efficient on 3′-overhangs or blund ends (data not shown).
Although the isolation and characterization of two other APs from psychrophiles has been earlier reported by other authors , , this is the first report describing cloning and expression of a psychrophilic AP.
The recombinant TAB5 AP was purified to homogeneity employing both conventional and affinity chromatographic techniques. The specific activity of the purified enzyme (1650 U·mg−1) is one of the highest among bacterial APs reported so far (Table 1). The apparent molecular mass of the purified enzyme preparation as determined by SDS/PAGE was 38 kDa (Fig. 3) approximately 2.5 kDa smaller than that estimated from the deduced amino-acid sequence (Fig. 1). The native molecular mass as determined by gel filtration was estimated to be 480 kDa indicating that the native enzyme may exist as a multimer. The optimum temperature for enzyme activity was determined to be 25 °C while the optimum pH 8.5. The psychrophilic enzyme exhibited a remarkable thermal lability. Incubation of the enzyme at 50 °C for 15 min resulted in complete denaturation of the enzyme, while CIP retained 90% of its original activity (Fig. 4).
The 3D-model of the enzyme indicates that the sequence conserves all characteristic features found in the core and active site of APs from various sources; furthermore the model provides insights into the sequence-function correlations of a protein which displays typical characteristics of psychrophilic enzymes, i.e, a high catalytic efficiency at low and moderate temperatures and an increased thermosensitivity [11,12]. Numerous residues can be identified which might play a key role in the establishment of these characterictics; the discussion will focus on the metal binding site, the catalytic site and residues proposed by other authors to be associated with the specific properties of psychrophilic enzymes.
In the E. coli AP (PDB code 1alk) residue D153 (corresponding to residue His135 in the TAB5 enzyme) is part of the Mg2+ binding site. Substitution of this residue by His (e.g. in the E. coli AP mutant D153H, PDB code 2anh) results in an enhancement of enzymatic activity when the Zn bound in this site is replaced by Mg2+. Generally, the most active APs (e.g. the mammalian APs, 650–13000 U·mg−1[37,38,39]), also exhibit His at this position while all other residues in the environment of the metal are strictly conserved. Similarly the TAB5 AP and the E. coli D153H mutant (which have identical metal binding residues) are both characterized by high enzymatic activities (1650 U·mg−1 and 2150 U·mg−1, respectively [8,40]. This could be related to the geometry of the site formed: in both the TAB5 model (Fig. 5) and in D153H E. coli mutant, the coordination sphere of the metal is defined exclusively by side chains of the protein. In contrast, in the less active E. coli AP (670 U·mg−1) the Mg2+-binding site possibly represents a less favorable configuration as some of the surrounding side chains interact with the metal only indirectly, via water molecules, thus affecting negatively the binding of the Mg2+ ion which is essential for activity.
Although the catalytic sites are to a large extent conserved between the TAB5 AP and its mesophilic counterparts, sequence alignments and the 3D-model of the protein reveal one active site residue corresponding to residue Lys328 in the E. coli AP which is rather variable (Figs 2 and 7B). This catalytic residue is His in mammalian APs and Trp in Bacillus and TAB5 APs (Figs 2 and 7A). On the basis of the 3D-model, the variability of these catalytic residues may be linked to the variability of other residues against which they pack in the enzyme structure, i.e. Asn263 in the E. coli AP and the equivalent residue Ala219 in the TAB5 AP (Figs 7A,B). Ala219 in the TAB5 AP which packs against the massive catalytic residue Trp is a small residue; this is also the case for the Bacillus AP which also packs against a catalytic Trp. In contrast, a smaller catalytic residue, e.g. Lys328 in the E. coli AP packs against larger residues, i.e. Asn263. It is thus likely that the volume changes of this active site residue are compensated by suitable changes of the residue against which this packs. A comparison of the structure of the D153H mutant  with the model of the TAB5 AP suggests that in the psychrophilic enzyme the catalytic Trp might form weak hydrogen bonds with the substrate. In the E. coli protein the catalytic Lys makes indirect (i.e. water mediated) hydrogen bonding interactions with the substrate. This altered pattern of interactions could affect enzymatic activity.
A striking feature of the modelled structure which could be associated with psychrophilicity  is the propensity of Gly residues to cluster at the termini of secondary structural elements of the Mg2+-binding site and the active site: two consecutive Gly residues are positioned immediately after the catalytic Trp260 residue (positions 261, 262). The Zn2+-binding residues Asp301 and His302 are also followed by two consecutive Gly residues. In the mesophilic mammalian and Bacillus APs the catalytic residue is followed by only one Gly, while no Gly is present at a structurally equivalent position in the less active E. coli AP. Mutagenesis work reported by other authors  provides experimental support to the hypothesis related the functional significance of Gly at the termini of secondary structural elements in psychrophilic enzymes. Overall, these observations and other characteristics of the TAB5 AP sequence (e.g. the decreased Arg/Arg + Lys ratio and the reduced number of salt bridges and hydrogen bonds) are in agreement with a pattern of sequence adaptations proposed by other authors for psychrophilic enzymes .
TAB5 AP was used to remove 5′-terminal phosphates from DNA fragments. Our results showed that under the conditions used TAB5 AP is effective in removing 5′ phosphate groups in 5′ overhangs as well as in 3′-overhangs or blunt ends, without damaging these ends (Fig. 6). Furthermore a simple heat treatment inactivates TAB5 AP, in contrast to APs from E. coli and calf intestine, and subsequently allows phosphorylation of DNA fragments successively by polynucleotide kinase.
The determination of the 3D structure of TAB5 AP and the construction of site-directed mutants are now required for a detailed analysis of the adaptation parameters suggested by molecular modeling.
We thank Prof. C. Gerday for providing the TAB5 strain and assisting in amino terminal sequencing. We also thank Dr D. Alexandraki, for her advice in DNA sequencing. This work was supported by the TMR Network FMRX-CT97-0131.
Enzyme: alkaline phosphatase (E.C. 188.8.131.52)Note: the first two authors have equally contributed to this work.