Evolution of putrescine N-methyltransferase from spermidine synthase demanded alterations in substrate binding

FEBS Letters 583 (2009) 3367–3374

journal homepage: www.FEBSLetters.org

Evolution of putrescine N-methyltransferase from spermidine synthase demanded alterations in substrate binding Stefan Biastoff a, Nicole Reinhardt a, Veaceslav Reva b, Wolfgang Brandt c, Birgit Dräger a,* a

Institute of Pharmacy, Faculty of Science I, Martin Luther University Halle-Wittenberg, Hoher Weg 8, D-06120 Halle, Germany Faculty of Biology and Soil Science, State University of Moldova, Str. Mateevici 60, MD-2009 Chisinau, Republic of Moldova c Department of Bioorganic Chemistry, Leibniz Institute of Plant Biochemistry, Weinberg 3, D-06120 Halle, Germany b

a r t i c l e

i n f o

Article history: Received 10 July 2009 Revised 20 September 2009 Accepted 25 September 2009 Available online 29 September 2009 Edited by Takashi Gojobori Keywords: Putrescine N-methyltransferase S-Adenosylmethionine Spermidine synthase Alkaloid biosynthesis Protein evolution

a b s t r a c t Putrescine N-methyltransferase (PMT) catalyses S-adenosylmethionine (SAM)-dependent methylation of putrescine in tropane alkaloid biosynthesis. PMT presumably evolved from the ubiquitous spermidine synthase (SPDS). SPDS protein structure suggested that only few amino acid exchanges in the active site were necessary to achieve PMT activity. Protein modelling, mutagenesis, and chimeric protein construction were applied to trace back evolution of PMT activity from SPDS. Ten amino acid exchanges in Datura stramonium SPDS dismissed the hypothesis of facile generation of PMT activity in existing SPDS proteins. Chimeric PMT and SPDS enzymes were active and indicated the necessity for a different putrescine binding site when PMT developed. Ó 2009 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved.

1. Introduction Putrescine N-methyltransferase (PMT, EC 2.1.1.53) is a S-adenosylmethionine (SAM)-dependent methyltransferase and catalyses the first metabolic step of tropane, nortropane, and nicotine alkaloid formation in Solanaceae and Convolvulaceae [1]. When PMT cDNA was initially cloned from tobacco, the translated amino acid sequences showed little similarity to other plant methyltransferases. Sequence identity of ca. 60% to plant SPDS, however, suggested that PMT evolved from SPDS by gene duplication and change of function [2,3]. Spermidine synthase (SPDS, EC 2.5.1.16) produces spermidine using the same substrate as PMT, putrescine, and a similar cosubstrate, decarboxylated S-adenosylmethionine (dcSAM) (Fig. 1). Crystal structures of SPDS proteins, e.g. from Thermotoga maritima [4], Plasmodium falciparum [5] and Homo sapiens [6] revealed an overall protein fold similar to many methyltransferases [7] and enabled to locate substrate

Abbreviations: dcSAM, decarboxylated S-adenosylmethionine; his-tag, hexahistidine tag; pmt, putrescine N-methyltransferase (cDNA); DsPMT, Datura stramonium putrescine N-methyltransferase; DsSPDS, Datura stramonium spermidine synthase; SAM, S-adenosylmethionine; spds, spermidine synthase (cDNA) * Corresponding author. Fax: +49 345 55 27 021. E-mail address: [email protected] (B. Dräger).

and cosubstrate arrangements in the active site of SPDS. Subsequent comparison between PMT and SPDS focussed on 20 amino acids in contact with substrate and cosubstrate in SPDS [4]. Among them, only a few active site residues systematically differed between PMT and SPDS. They were assumed to alter the selectivity for SAM versus dcSAM in the cosubstrate binding region [3,8]. We hypothesised that PMT evolution from an ancient SPDS may have required only a few amino acid exchanges in the active site. Recently, several PMT cDNAs were cloned from Solanaceae and a Convolvulaceae [9–11]. Enzyme proteins are available by heterologous synthesis, but PMT up to now proved recalcitrant to crystal structure elucidation. In order to test the hypothesis of facile conversion of SPDS into a methyl transfer catalysing protein we mutated up to 10 amino acids in the active site of Datura stramonium SPDS. As no active PMT resulted from those experiments we merged the substrate binding regions of D. stramonium PMT and SPDS by constructing chimeric proteins and found that the amino acids decisive for catalytic activity reside in the N-terminal parts of both proteins. Taking crystal structures of Arabidopsis thaliana and human SPDS as templates, protein structure modelling of D. stramonium PMT and SPDS and substrate docking suggested a binding position for putrescine in PMT that varied from that of putrescine in SPDS.

0014-5793/$36.00 Ó 2009 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2009.09.043

3368

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

NH2

S-adenosylL-homocysteine

5´-methylthioadenosine spermidine synthase (SPDS) EC 2.5.1.16

NH2

putrescine N-methyltransferase (PMT) EC 2.1.1.53

putrescine NH2

NH2

NH

NH2

NH2 CH3

N

+

H2N

NH2

N

S

O

N

CH3

N

spermidine

H2N

H N

H N

decarboxylated S-adenosylL-methionine (dcSAM)

NH2

primary metabolism

NH

N

CH3

+

H2N

S

O

N

N

COOH

N-methylputrescine

OH OH

OH OH spermine synthase EC 2.5.1.22

N

S-adenosylL-methionine (SAM)

secondary metabolism

nicotine, calystegines, tropane alkaloids

spermine Fig. 1. Putrescine metabolism. Spermidine occurs in all living organisms, while N-methylputrescine is known from alkaloid producing plants only.

2. Materials and methods 2.1. cDNA mutagenesis and chimerisation After cloning wild-type cDNA of D. stramonium spds1 (Accession No. Y08252) into a pET-21d vector (Novagen), DsSPDS was synthesised in Escherichia coli BL21(DE3) cells (Novagen) with a hexahistidine tag (his-tag) at the C-terminus. D. stramonium pmt (Accession No. AJ583514) was cloned into vector pQE-30 (QIAGEN), and the enzyme was synthesised in M15[pREP4] cells (QIAGEN) with a N-terminal his-tag. Mutant DsSPDS proteins were synthesised in BL21(DE3) cells or in BL21-CodonPlus (DE3)-RP cells (Stratagene) after mutagenesis with the QuikChange II Site-Directed Mutagenesis Kit (Stratagene). cDNA for deletion of the N-terminal 54 triplets of pmt was amplified with Pfu DNA Polymerase (Fermentas), cloned into pET-47b vector (Novagen) providing a N-terminal his-tag and synthesised in BL21-CodonPlus (DE3)-RP cells (primers Table S1). D. stramonium spds and D. stramonium pmt were restricted with BamHI (New England Biolabs), spds at position 432, and for pmt a BamHI-site was constructed at position 546 (primers N182E, Table S1). Crossover ligations with T4-ligase (Fermentas) produced two chimeric cDNAs used for protein synthesis from pET-21d vectors in BL21-CodonPlus (DE3)-RP cells. Calculated molecular weights per subunit are SPDS–PMT: 34.9 kDa; PMT–SPDS: 39.0 kDa. 2.2. Protein synthesis and purification E. coli cells were grown in terrific broth medium [12] with appropriate antibiotic at 37 °C. Enzyme synthesis was induced with 1 mM IPTG for 4–6 h. Bacteria were lysed in buffer (50 mM sodium phosphate, 300 mM NaCl, 10 mM imidazole, Sigma protease inhibitor cocktail for proteins with his-tags, 800 lg/ml lysozyme, pH 8.0) by sonication (Branson Sonifier 250). After addition of 10 lg/ml DNase I, centrifugation (40 min, 15 000g, 4 °C), and filtration (0.45 lm, cellulose acetate), the supernatant was used for purification. A HisTrap HP 1 ml column (GE Healthcare) equilibrated with buffer A (20 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, pH 7.8) was washed with 15 vol-

umes of buffer A (1 ml/min) after sample application. Protein was eluted in a gradient from 20 to 500 mM imidazole in 30 column volumes (wild-type PMT and SPDS) or 20 column volumes (mutant and chimeric proteins). The imidazole buffer was immediately exchanged by PMT storage buffer B (100 mM HEPES, pH 8.0) or the SPDS storage buffer C (20 mM HEPES, pH 8.0, 2 mM DTT, 1 mM ascorbic acid) on a Sephadex G-25-Fine column (13.5  1.6 cm) at a flow rate of 2 ml/min. Purified proteins were stored at 4 °C or at 80 °C after addition of 10% (v/v) glycerol. Purified 54-NPMT was enzymatically separated from the his-tag by incubation with recombinant HRV 3C protease (Novagen) at 4 °C for 16 h. Protein concentrations were determined by Coomassie Brilliant Blue with bovine serum albumin as standard [13]. For analytical gel filtration, a Sephacryl S-200 HR column (59  1.6 cm) was calibrated with standard proteins and Blue Dextran (Sigma). 2.3. Enzyme activity DsPMT and 54-N-PMT activities were measured by a coupled enzymatic and colorimetric methyltransferase assay [14] with 0.4– 1.0 lg PMT protein per sample. Presence of reducing agents necessary to maintain SPDS enzyme activity interfered with the colorimetric PMT assay. HPLC [11] was used to measure formation of spermidine or N-methylputrescine in wild-type SPDS, in mutated and in chimeric proteins. Reaction mixtures contained 10–25 lg purified protein, 20 mM HEPES, pH 8.0, 2 mM DTT, 1 mM ascorbic acid, 500 lM putrescine and 400 lM SAM or dcSAM in a total volume of 250 ll. Enzyme assays were incubated for 30 min at 37 °C and stopped with 250 ll saturated sodium carbonate solution. Two hundred microliters of the alkaline mixture were dansylated as described [15]. Dansylated spermidine and N-methylputrescine were separated by HPLC and quantified by DAD detection at 220 nm. For kinetic characterisation, enzyme protein was reduced to 1 lg/assay. 2.4. Protein modelling and substrate docking Homologous proteins with a resolved X-ray structure were identified by BLASTp search in the Protein Data Bank (PDB). The

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

crystal structure of A. thaliana SPDS1 (AtSPDS; PBD 1XJ5) with 68% amino acid identity to DsPMT was used for homology modelling. Human spermidine synthase (PDB 2O06) [6] has 46% overall sequence identity to PMT. AtSPDS forms a homo-tetrameric structure, human SPDS appears as homo-dimer in the crystal structure and served as template for homology modelling of the dimeric structure of DsPMT. Based on these two structures, the homology model of DsPMT was created by MOE (Molecular Operating Environment, Chemical Computing Group Inc., Montreal, Canada) and minimised using CHARMM22 force field (gradient below 0.05)

Table 1 Kinetic properties of DsSPDS, DsPMT, and chimeric enzymes; data are average of 2–3 determinations; n.d. = not determined. Protein

Km putrescine (lM)

Km dcSAM (lM)

Km SAM (lM)

kcat (s 1)

DsSPDS wild-type DsPMT wild-type DsSPDS–PMT chimera DsPMT–SPDS chimera 54-N-PMT

33.1 205.4 75.8

64.2 56.1

0.44 5.14 0.40

529.9

78.5

0.69

170.2

n.d.

3.09

38.4

3369

[16] and Born solvation [17]. DsSPDS with sequence identity of 81% to A. thaliana SPDS1 (PBD 1XJ5) was modelled accordingly. Models of chimeric proteins were constructed by MOE starting from the equivalent FASTA sequences. The sequences were aligned with those of DsSPDS and DsPMT, and the protein with the longest identical chain was taken for homology modelling of the chimera. The quality of the models was verified with PROCHECK [18], PROSA [19], and ProQ [20]. All methods indicated extremely good or very good models and showed slight uncertainties only for the N- and C-terminal tails (using ModFold [21–23] and MetaMQAP [24]). Docking of SAM and dcSAM was achieved by superposition of the backbone structures of the protein models with the X-ray structure of human spermidine synthase (PDB 2O06), followed by merging the cosubstrate to the model and subsequent energy optimisation. The optimal SAM position in PMT was evaluated by calculation of the conformation energy of 48.4 kcal/mol. Alternative SAM conformations, e.g. mimicking exactly that of dcSAM in SPDS required 58.6 kcal/mol or more. The substrate putrescine was docked to the active site by the program GOLD (Genetic Optimized Ligand Docking, Cambridge Crystallographic Data Centre, 1998, Cambridge, UK) [25,26] using standard values of the program that allow early termination if the top 3 solutions are within 1.5 Å RMSD. Thus, only three almost identical docking arrangements remained. The one with a non-protonated amino group oriented

Fig. 2. Sequence alignment by Clustal W. Arabidopsis thaliana (At), Homo sapiens (Hs), Atropa belladonna (Ab), Hyoscyamus niger (Hn). Secondary structures in the DsSPDS model are indicated by arrows (b-strands) and spirals (a-helices). Identical amino acids in the alignment are shaded in grey, residues for cumulative mutagenesis are printed white on black. Below the sequences, restriction site for chimera construction (N), deletion site for the N-terminus of DsPMT (j), and residues responsible for AdoDATO binding in T. maritima SPDS (o) are indicated.

3370

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

close to the active methyl group of SAM was used for further energy refinement to capture slight induced fit modifications of the proteins active site. 2.5. Phylogenetic analysis and other software Twenty-seven full-length PMT protein sequences from Solanaceae and Convolvulaceae were identified by extensive database search. Putative PMTs from Capsicum annuum (Solanaceae, EST Accession Nos. CA518000; CA518369; CA518560; CA520943) and Ipomoea batatas (Convolvulaceae, EST Accession Nos. DV038077; EE876509) were extracted from the NCBI EST database and assembled (Table S2). Twenty SPDS sequences from various organisms were included in the phylogenetic analysis (Table S3). Multiple sequence alignments were performed with Clustal W2. Phylogenetic relationship was inferred by the Neighbour-Joining (NJ) method and the maximum parsimony (MP) calculation using amino acid sequences and cDNA sequences. Poisson correction or JTT matrix for amino acid exchange proportions were applied in NJ tress. Computing was done by MEGA4 [27]. Enzyme kinetic parameters were calculated using the Sigmaplot 9.0 software (Systat Inc.). Protein structure images were designed with PyMol 0.98 (DeLano Scientifc LLC). Sequence alignment was edited with ESPript 2.2 [28].

3. Results and discussion 3.1. Active site mutagenesis of Datura stramonium SPDS D. stramonium PMT and SPDS wild-type proteins were synthesised in E. coli. After purification, kinetic parameters were measured (Table 1). DsSPDS, like other plant SPDS [29,30], possessed a high affinity for putrescine. The comparatively lower affinity for putrescine in DsPMT (Km 205 lM) may prevent the putrescine pool from decreasing at the expense of polyamine metabolism. Activity measurements after analytical gel filtration demonstrated native DsPMT and DsSPDS to be active as homo-dimers (data not shown). Many methyltransferases, in particular those of plant secondary metabolism, were described as monomeric proteins [7,31], and earlier gel filtration results indicated a monomeric enzyme for PMT isolated from D. stramonium roots [32]. PMT sequences, however, are related more closely to those of plant SPDS (Fig. 2) than to any methyltransferases. PMT amino acid sequence identity to plant methyltransferases is below 10% [33]. The PMT protein dimer obtained by homology modelling (Fig. 3A) does not locate the active site near the dimerisation plane. Monomeric proteins after dimer separation possibly retain some activity, which would explain earlier findings of a monomeric PMT. When docking of SAM to PMT was optimised, it adopted a position slightly different from dcSAM in SPDS. The adenine and ribose moieties stayed unaltered, but the aminocarboxyl end was rotated around the S–Cc bond of the methionine moiety and pointed into the enzyme cavity that corresponded to the putrescine binding site of SPDS (Fig. 4A and B). Putrescine in PMT was docked best at a position that accommodated the aminopropyl moiety of dcSAM in DsSPDS. In the SPDS active site, L70 and Q79 interacted with the aminopropyl side chain of dcSAM (Fig. 4A). The corresponding residues in PMT, H108 and T117, due to different putrescine positioning, participated in putrescine binding (Fig. 4B). The SPDS mutants L70H and Q79T retained SPDS activity with 100% and 75%, respectively, of the wild-type enzyme. PMT activity was not detected. Thus, functional positioning of dcSAM in SPDS was possible after exchange of L70H and Q79T. A strictly conserved aspartate in SPDS (D103 in DsSPDS) was considered as decisive for the selectivity of the cosubstrate dcSAM versus SAM [4,8]. Aspartate was assumed to prevent SAM from binding by repelling the carboxyl group in

Fig. 3. (A) Dimeric D. stramonium PMT. (B) Subunit of the chimeric protein PMTSPDS. (C) Subunit of chimeric protein SPDS–PMT. Fifty-one N-terminal amino acids were omitted in (B) and 13 in (C) for clarity. Parts of PMT (green), parts of SPDS (grey), putrescine (cyan), SAM (magenta), dcSAM (golden); N-termini (N) and Ctermini (C) are indicated.

the methionine end. With dcSAM, a salt bridge would stabilise binding. The corresponding position in all PMTs is a non-charged isoleucine, I141 in DsPMT (Fig. 4A and B). Mutation D103I abolished DsSPDS activity but did not produce PMT activity. A similar exchange in T. maritima SPDS, D101I, also resulted in an almost inactive SPDS enzyme [6]. Probably, positioning of dcSAM is impaired by the non-charged isoleucine. In the PMT amino acid sequence, a region that differs consistently between PMT and SPDS contains residues H108–T117 that line the putrescine binding site predicted by substrate docking to PMT. The PMT-typical residues were introduced stepwise into

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

3371

Fig. 4. (A) Active site model of DsSPDS with dcSAM (golden) and putrescine (cyan). Residues for single mutations in green. (B) Active site model of DsPMT with SAM (magenta) and putrescine (cyan). Residues corresponding to (A) in green. (C) Putrescine binding site of DsSPDS mutated 10-fold with SAM and putrescine (put). (D) Putrescine binding site of DsPMT docked with SAM and putrescine (put).

DsSPDS. The residues corresponding to I141, F143 and F146 that are located in a loop or short helix also facing the substrate were changed accordingly. None of the resulting mutants Nos. 2–10 (Fig. 2; Table S1) had PMT activity. Decrease of enzyme solubility, in particular after exchange R73N (mutation No. 6) was observed during synthesis of mutant proteins, suggesting that the conserved arginine 73 located at the protein surface of SPDS ensures enzyme solubility. Each mutated DsSPDS enzyme, however, was synthesised with sufficient soluble protein to allow detection of PMT activity if there had been any. Modelling of the 10-fold mutated DsSPDS showed amino acids in the active site arranged very similar to PMT (Fig. 4C), but slight differences in the individual amino acid positions and in the protein structural framework were evident. Putrescine docking showed that the putrescine-binding pocket in the DsSPDS mutant was narrower than in wild-type PMT (Fig. 4D). Exchange of six amino acids in the active site was assumed to be sufficient for conversion of SPDS into PMT activity [8], but did not produce PMT activity, and additional four exchanges as performed here did not convert catalytic activity. Rather, larger alterations in protein structure and in substrate binding are required and took place during evolution from SPDS to PMT. 3.2. Catalytic specificities of DsPMT and DsSPDS depend on N-terminal protein domains Because PMT-specific amino acids in the active site did not suffice to generate PMT activity in the SPDS protein, single position mutagenesis was not longer pursued. Two chimeric proteins were constructed consisting of (1) 144 N-terminal amino acids of DsSPDS and 162 C-terminal amino acids of DsPMT (SPDS–PMT) and of (2)

181 N-terminal amino acids of DsPMT and 165 C-terminal amino acids of DsSPDS (PMT–SPDS) (Fig. 3B and C). The SPDS–PMT chimera showed SPDS activity with kinetic parameters similar to wild-type DsSPDS (Table 1). The chimeric PMT–SPDS catalysed methyl transfer, but considerably slower (kcat 0.69 s 1) than wildtype DsPMT (kcat 5.14 s 1). The Km value for SAM was slightly higher in the chimeric enzyme while the Km value for putrescine of 0.5 mM was strikingly high compared to wild-type PMT, suggesting that the altered protein framework in the C-terminal part did not support optimal putrescine binding. Each chimeric enzyme catalysed only one reaction, methyl transfer or aminopropyl transfer, which was determined by the respective N-terminal protein moiety. DsPMT, like other PMT proteins, contains a N-terminal extension that appeared largely as unordered loops with short b-strand sections, partially in proximity to the active site (Fig. 3A). To examine whether this N-terminal extension participates in PMT substrate specificity or catalysis, 54 amino acids from the Nterminus were omitted. The truncated PMT protein had essentially the same properties as wild-type PMT with a Km value for putrescine slightly lower than in the wild-type enzyme (Table 1). Obviously, the N-terminal part is not essential for PMT catalytic activity. Removal of the N-terminal amino acids from PMT decreased protein solubility. The conclusion that the N-terminus procured PMT protein solubility was tested by transfer of 54 Nterminal amino acids from PMT to the 10-fold mutated SPDS. Protein solubility of mutated and N-terminally extended SPDS slightly increased, but no PMT activity was observed. Both chimeric proteins were modelled, and the substrate putrescine and the cosubstrates were docked for their optimal positions. Again, as in the native enzymes, the adenosine moiety of SAM in PMT–SPDS was placed with the carboxyl-amino termi-

3372

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

95 Nic. tabacum PMT2 53 Nic. sylvestris PMT1 43

Nic. attenuata PMT2

78

Nic. tabacum PMT4

100 Nic. sylvestris PMT3 99 Nic. tabacum PMT1 Nic. attenuata PMT1 Nic. tabacum PMT3

83

Nic. benthamiana PMT

76 86

Sol. dulcamara PMT Sol. tuberosum PMT Dat. stramonium PMT

88 97

Dat. metel PMT Dat. inoxia PMT2

97 97

Dat. inoxia PMT1

96

Phy. divaricata PMT Cap. annuum PMT (put.)

80

Sol. lycopersicum PMT Hyo. niger PMT

93 100

Atr. belladonna PMT1

93

Ani. tanguticus PMT

80 63 Ani. acutangulus PMT1 100

Atr. belladonna PMT2 Ani. acutangulus PMT2

100 62

Sco. parviflora PMT (put.) Cal. sepium PMT

100

Ipo. batatas PMT (put.)

100

Pis. sativum SPDS2

64

Pis. sativum SPDS1 Ara. thaliana SPDS1

99

Ara. thaliana SPDS2 100

Mal. domestica SPDS1 Cof. arabica SPDS

41

Dat. stramonium SPDS1

88

Hyo. niger SPDS2 Nic. sylvestris SPDS

83

Dat. stramonium SPDS2 100 76

Hyo. niger SPDS1

67

Sol. lycopersicum SPDS

62 100

Sol. tuberosum SPDS

Sac. cerevisiae SPDS 100

Neu. crassa SPDS Hom. sapiens SPDS Cae. elegans SPDS Pyr. furiosus SPDS The. maritima SPDS

96 92

Bac. subtilis SPDS

0.1 Fig. 5. Phylogenetic tree of PMT and SPDS proteins. The tree was calculated by the Neighbour-Joining method. The optimal tree with the sum of branch length = 4.83879196 is shown. Bootstrap values above 40% (1000 replicates) are printed next to the branch points. Evolutionary distances were calculated with Poisson correction. Positions with alignment gaps were eliminated only in pairwise sequence comparisons (total of 463 positions in the final dataset). PMT proteins indicated as putative (put) were retrieved as non-annotated cDNA sequences or EST fragments, assembled and translated into full-length proteins.

nus of methionine rotated and located in the equivalent putrescine binding region of SPDS (Fig. 4B). Putrescine in PMT–SPDS was lo-

cated adjacent to Ile141. An acidic amino acid residue at this position, like D103 in SPDS, will not prevent SAM from binding as

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

hypothesised before [4,8], but will detract putrescine and thus prevent a reaction with SAM. Other protein modelling results supported this view. Two crystal structures of SPDS, human SPDS (PDB 1ZDZ) and Trypanosoma cruzi SPDS (PDB 3BWC) contain bound SAM instead of dcSAM in the same orientation as docking proposed for SAM in PMT. Conversely, when SAM was placed into a N. tabacum PMT protein model in an orientation exactly like that of dcSAM in SPDS, a position for putrescine was not found [3]. We tried to complete the docking of putrescine into PMT in the position that corresponds to that in SPDS, but SAM was then oriented for aminoproyl transfer, not for methyl transfer, because the sulfonium cation in SAM is chiral and naturally in (S)-configuration. The methionine moiety of SAM was rotated in order to find a conformation, from which methyl transfer was possible onto putrescine that remained located in the position of putrescine in SPDS. The only possible orientation of SAM demanded 58.6 kcal/mol conformation energy, which is 10.2 kcal/mol higher than the conformation optimised before (48.4 kcal/mol; Fig. 4B). These comparisons together indicate that the difficulty in the transition from SPDS into PMT activity was not binding of SAM, but rather binding of putrescine in a position appropriate for methyl group transfer. 3.3. Phylogenetic analysis reveals long distance between PMT and SPDS All available PMT amino acid sequences were aligned to SPDSs of various organisms. Non-annotated sequences from C. annuum, I. batatas, and Scopolia parviflora found by BLAST search were included after translation into amino acids; they clustered together with PMT sequences. Identification of calystegines that require PMT activity for biosynthesis in C. annuum [34], I. batatas [35], and Scopolia species [36] confirmed the prediction for those sequences to code for PMT. The pertinent question is, how a transition from SPDS to PMT that demanded the formation of a different putrescine binding site may have occurred. The phylogenetic tree of amino acid sequences is bifurcated between PMT and all plant SPDS sequences (Fig. 5). When cDNA sequences and different tree calculation methods, e.g. maximum parsimony were used, essentially the same branching pattern resulted with each approach. It was important to note that all PMTs were divided from the whole group of plant SPDS suggesting that separation of PMT from SPDS occurred before Convolvulaceae separated from Solanaceae, which happened ca. 62 million years ago [37]. If PMTs had recently evolved from SPDS, PMT sequences would have been placed intermingled with plant SPDS in a phylogenetic tree. The long evolutionary distance between the most recent common ancestor of SPDS and PMT and extant active PMT proteins supports a route with many intermediate mutations. After gene duplication one copy of the ancient SPDS accumulated mutations most of which probably led to inactive proteins, but the respective genes were not removed from the genomes and evolved further until putrescine binding for methyl transfer was possible. In contrast, due to the chirality of the sulphur atom in natural SAM, a marginal adaptation of the cosubstrate binding site in SPDS for acceptance of SAM is discouraged as decisive event in PMT evolution. At present, PMT sequence data are only known for Solanaceae and Convolvulaceae. Other taxonomically unrelated plant families also contain tropane alkaloids, such as Erythroxylaceae [38] and Brassicaceae [39]. For a deeper understanding of PMT development, it is essential to identify and examine PMT proteins and coding sequences in those unrelated plant families. Clearly, isolation of proteins that perform intermediate PMT and SPDS activity and generation of further mutants and chimera that form stepping stones for a potential transition between SPDS and PMT will help to illustrate a feasible path for the evolution of PMT.

3373

Acknowledgements N-methylputrescine was a generous gift of Prof. T. Hashimoto, NAIST, Japan. Prof. K. Samejima, Musashino University, Japan, kindly provided dcSAM. We thank Dr. Y. Sichhart, Institute of Pharmacy, MLU, for helpful discussions. U. Mönch and K. Klimpel, Institute of Pharmacy, MLU, helped with cloning and protein synthesis. Experiments were financially supported by the German Research Foundation, Priority programme 1152 ‘‘Evolution of metabolic diversity”. V. Reva was financially supported by the Martin Luther University Halle-Wittenberg. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.febslet.2009.09.043. References [1] Biastoff, S. and Dräger, B. (2007) Calystegines (Cordell, G.A., Ed.), The Alkaloids, vol. 64, pp. 49–102, Academic Press, New York. [2] Hashimoto, T., Tamaki, K., Suzuki, K. and Yamada, Y. (1998) Molecular cloning of plant spermidine synthases. Plant Cell Physiol. 39, 73–79. [3] Minguet, E.G., Vera-Sirera, F., Marina, A., Carbonell, J. and Blazquez, M.A. (2008) Evolutionary diversification in polyamine biosynthesis. Mol. Biol. Evol. 25, 2119–2128. [4] Korolev, S., Ikeguchi, Y., Skarina, T., Beasley, S., Arrowsmith, C., Edwards, A., Joachimiak, A., Pegg, A.E. and Savchenko, A. (2002) The crystal structure of spermidine synthase with a multisubstrate adduct inhibitor. Nat. Struct. Biol. 9, 27–31. [5] Dufe, V.T., Qiu, W., Muller, I.B., Hui, R., Walter, R.D. and Al Karadaghi, S. (2007) Crystal structure of Plasmodium falciparum spermidine synthase in complex with the substrate decarboxylated S-adenosylmethionine and the potent inhibitors 4MCHA and AdoDATO. J. Mol. Biol. 373, 167–177. [6] Wu, H., Min, J., Ikeguchi, Y., Zeng, H., Dong, A., Loppnau, P., Pegg, A.E. and Plotnikov, A.N. (2007) Structure and mechanism of spermidine synthases. Biochemistry 46, 8331–8339. [7] Martin, J.L. and McMillan, F.M. (2002) SAM (dependent) I AM: the Sadenosylmethionine-dependent methyltransferase fold. Curr. Opin. Struct. Biol. 12, 783–793. [8] Ikeguchi, Y., Bewley, M.C. and Pegg, A.E. (2006) Aminopropyltransferases: function, structure and genetics. J. Biochem. 139, 1–9. [9] Hibi, N., Higashiguchi, S., Hashimoto, T. and Yamada, Y. (1994) Gene expression in tobacco low-nicotine mutants. Plant Cell 6, 723–735. [10] Hashimoto, T., Shoji, T., Mihara, T., Oguri, H., Tamaki, K., Suzuki, K.i. and Yamada, Y. (1998) Intraspecific variability of the tandem repeats in Nicotiana putrescine N-methyltransferases. Plant Mol. Biol. 37, 25–37. [11] Teuber, M., Azemi, M.E., Namjoyan, F., Meier, A.-C., Wodak, A., Brandt, W. and Dräger, B. (2007) Putrescine N-methyltransferases – a structure–function analysis. Plant Mol. Biol. 63, 787–801. [12] J. Sambrook, E.F. Fritsch, T. Maniatis, Molecular cloning. In: A Laboratory Manual, second ed. Cold Spring Harbour Laboratory Press, Plainview, New York, 1989. [13] Bradford, M.M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. [14] Biastoff, S., Teuber, M., Zhou, Z.S. and Dräger, B. (2006) Colorimetric activity measurement of a recombinant putrescine N-methyltransferase from Datura stramonium. Planta Med. 72, 1136–1141. [15] Smith, M.A. and Davies, P.J. (1985) Separation and quantitation of polyamines in plant tissue by high performance liquid chromatography of their dansyl derivatives. Plant Physiol. 78, 89–91. [16] MacKerell, A.D., Bashford, D., Bellott, M., Dunbrack, R.L., Evanseck, J.D., Field, M.J., Fischer, S., Gao, J., Guo, H., Ha, S., Joseph-McCarthy, D., Kuchnir, L., Kuczera, K., Lau, F.T.K., Mattos, C., Michnick, S., Ngo, T., Nguyen, D.T., Prodhom, B., Reiher, W.E., Roux, B., Schlenkrich, M., Smith, J.C., Stote, R., Straub, J., Watanabe, M., Wiorkiewicz-Kuczera, J., Yin, D. and Karplus, M. (1998) Allatom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616. [17] Pellegrini, E. and Field, M.J. (2002) A generalized-born solvation model for macromolecular hybrid-potential calculations. J. Phys. Chem. A 106, 1316– 1326. [18] Laskowski, R.A., MacArthur, M.W., Moss, D.S. and Thornton, J.M. (1993) PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallogr. 26, 283–291. [19] Sippl, M.J. (1990) Calculation of conformational ensembles from potentials of mean force – an approach to the knowledge-based prediction of local structures in globular proteins. J. Mol. Biol. 213, 859–883. [20] Cristobal, S., Zemla, A., Fischer, D., Rychlewski, L. and Elofsson, A. (2001) A study of quality measures for protein threading models. BMC Bioinform. 2, 5.

3374

S. Biastoff et al. / FEBS Letters 583 (2009) 3367–3374

[21] McGuffin, L.J. (2007) Benchmarking consensus model quality assessment for protein fold recognition. BMC Bioinform. 8, 345. [22] McGuffin, L.J. (2008) The ModFOLD server for the quality assessment of protein structural models. Bioinformatics 24, 586–587. [23] Pettitt, C.S., McGuffin, L.J. and Jones, D.T. (2005) Improving sequence-based fold recognition by using 3D model quality assessment. Bioinformatics 21, 3509–3515. [24] Pawlowski, M., Gajda, M.J., Matlak, R. and Bujnicki, J.M. (2008) MetaMQAP: a meta-server for the quality assessment of protein models. BMC Bioinform. 9, 403. [25] Jones, G., Willett, P., Glen, R.C., Leach, A.R. and Taylor, R. (1997) Development and validation of a genetic algorithm for flexible docking. J. Mol. Biol. 267, 727–748. [26] Verdonk, M.L., Cole, J.C., Hartshorn, M.J., Murray, C.W. and Taylor, R.D. (2003) Improved protein-ligand docking using GOLD. Proteins 52, 609–623. [27] Tamura, K., Dudley, J., Nei, M. and Kumar, S. (2007) MEGA4: molecular evolutionary genetics analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596–1599. [28] Gouet, P., Courcelle, E., Stuart, D.I. and Metoz, F. (1999) ESPript: multiple sequence alignments in PostScript. Bioinformatics 15, 305–308. [29] Graser, G. and Hartmann, T. (2000) Biosynthesis of spermidine, a direct precursor of pyrrolizidine alkaloids in root cultures of Senecio vulgaris L. Planta 211, 239–245. [30] Sindhu, R.K. and Cohen, S.S. (1984) Propylamine transferases in Chinese cabbage leaves. Plant Physiol. 74, 645–649.

[31] Lam, K.C., Ibrahim, R.K., Behdad, B. and Dayanandan, S. (2007) Structure, function, and evolution of plant O-methyltransferases. Genome 50, 1001–1013. [32] Walton, N.J., Peerless, A.C.J., Robins, R.J., Rhodes, M.J.C., Boswell, H.D. and Robins, D.J. (1994) Purification and properties of putrescine Nmethyltransferase from transformed roots of Datura stramonium L. Planta 193, 9–15. [33] Stenzel, O., Teuber, M. and Dräger, B. (2006) Putrescine N-methyltransferase in Solanum tuberosum L., a calystegine-forming plant. Planta 223, 200–212. [34] Asano, N., Kato, A., Matsui, K., Watson, A.A., Nash, R.J., Molyneux, R.J., Hackett, L., Topping, J. and Winchester, B. (1997) The effects of calystegines isolated from edible fruits and vegetables on mammalian liver glycosidases. Glycobiology 7, 1085–1088. [35] Schimming, T., Tofern, B., Mann, P., Richter, A., Jenett-Siems, K., Dräger, B., Asano, N., Gupta, M.P., Correa, M.D. and Eich, E. (1998) Phytochemistry and chemotaxonomy of the Convolvulaceae. 6. Distribution and taxonomic significance of calystegines in the Convolvulaceae. Phytochemistry 49, 1989–1995. [36] Dräger, B., van Almsick, A. and Mrachatz, G. (1995) Distribution of calystegines in several Solanaceae. Planta Med. 61, 577–579. [37] Paape, T., Igic, B., Smith, S.D., Olmstead, R., Bohs, L. and Kohn, J.R. (2008) A 15Myr-old genetic bottleneck. Mol. Biol. Evol. 25, 655–663. [38] Brock, A., Bieri, S., Christen, P. and Dräger, B. (2005) Calystegines in wild and cultivated Erythroxylum species. Phytochemistry 66, 1231–1240. [39] Brock, A., Herzfeld, T., Paschke, R., Koch, M. and Dräger, B. (2006) Brassicaceae contain nortropane alkaloids. Phytochemistry 67, 2050–2057.