Crystal structure of thermospermine synthase from Medicago truncatula and substrate discriminatory features of plant aminopropyltransferases

Abstract

Polyamines are linear polycationic compounds that play a crucial role in the growth and development of higher plants. One triamine (spermidine, SPD) and two tetraamine isomers (spermine, SPM, and thermospermine, TSPM) are obtained by the transfer of the aminopropyl group from decarboxylated S-adenosylmethionine to putrescine and SPD. These reactions are catalyzed by the specialized aminopropyltransferases. In that respect, plants are unique eukaryotes that have independently evolved two enzymes, thermospermine synthase (TSPS), encoded by the gene ACAULIS5, and spermine synthase, which produce TSPM and SPM, respectively. In this work, we structurally characterize the ACAULIS5 gene product, TSPS, from the model legume plant Medicago truncatula (Mt). Six crystal structures of MtTSPS — one without ligands and five in complexes with either reaction substrate (SPD), reaction product (TSPM), or one of three cofactor analogs (5′-methylthioadenosine, S-adenosylthiopropylamine, and adenosine) — give detailed insights into the biosynthesis of TSPM. Combined with small-angle X-ray scattering data, the crystal structures show that MtTSPS is a symmetric homotetramer with an interdomain eight-stranded β-barrel. Such an assembly and the presence of a hinge-like feature between N-terminal and C-terminal domains give the protein additional flexibility which potentially improves loading substrates and discarding products after the catalytic event. We also discuss the sequence and structural features around the active site of the plant aminopropyltransferases that distinguish them from each other and determine their characteristic substrate discrimination.

Introduction

The ubiquitous polyamine biosynthesis route in eukaryota, which yields the simple diamine putrescine [PUT, NH₂(CH₂)₄NH₂], relies on the decarboxylation of ornithine by ornithine decarboxylase (EC 4.1.1.17). Ornithine is produced from arginine by the action of arginase (EC 3.5.3.1). However, plants have also developed a parallel branch of the polyamine biosynthesis pathway, which has its origins in the horizontal gene transfer from the cyanobacterial ancestor of the chloroplast [1]. This three-step pathway starts with arginine [2]. In the first reaction, arginine decarboxylase (EC 4.1.1.19) provides agmatine, which, in the second step, is hydrolyzed to N-carbamoylputrescine by agmatine iminohydrolase (EC 3.5.3.12). In the third step, N-carbamoylputrescine is hydrolyzed to PUT by an octameric enzyme, N-carbamoylputrescine aminohydrolase (EC 3.5.1.53), whose quaternary structure resembles an incomplete left-handed helix [3]. Some species in the plant kingdom, including Arabidopsis thaliana (At) and Physcomitrella patens, lack an ornithine decarboxylase gene [4] and rely only on the agmatine pathway. Higher polyamines are produced by aminopropyltransferases which use decarboxylated S-adenosylmethionine (dc-SAM) as a donor of the aminopropyl group. Therefore, in the first reaction catalyzed by spermidine synthase (SPDS, EC 2.5.1.16), the transfer of the aminopropyl group to PUT yields triamine spermidine [SPD, NH₂(CH₂)₃NH(CH₂)₄NH₂]. Another transfer to SPD, catalyzed by either spermine synthase (SPMS, EC 2.5.1.22) or thermospermine synthase (TSPS, EC 2.5.1.79), provides the tetraamine isomers: symmetrical spermine [SPM, NH₂(CH₂)₃NH(CH₂)₄NH(CH₂)₃NH₂] or thermospermine [TSPM, NH₂(CH₂)₃NH(CH₂)₃NH(CH₂)₄NH₂].

For a long time, it was unclear whether both SPMS and TSPS are present in plant organisms. However, it has been recently shown that the enzyme engaged in TSPM synthesis, encoded by a gene called ACAULIS5, has the activity of TSPS but not SPMS [5]. Moreover, TSPM is evolutionally more ancient than SPM in plant and algae organisms [6]. Owing to the similarity between ACAULIS5 and prokaryotic triamine/agmatine aminopropyltransferase (TAAPT) genes, it is also plausible that plant TSPS has its origin in prokaryotic TAAPTs [7,8].

Polyamines, including TSPM, play a crucial role in the growth and development of higher plants [9–13]. Owing to their cationic character and ability to interact with nucleic acids and proteins, polyamines are able to regulate transcription and translation [9,14,15]. Moreover, they can affect the rate of membrane transport [16,17]. Polyamines also have a stimulating effect on the antioxidant enzymes, which leads to the decrease in reactive oxygen species concentration [18–20]. The accumulation of polyamines in plant tissues under stress conditions leads to an increase in stress tolerance [21–23]. In addition, the application of exogenous polyamines increases tolerance to high temperatures [24] and salinity [25–27]. Malfunctions of the polyamine pathway lead to growth retardation, sterility, and other developmental defects [28]. In this aspect, the dysfunction of ACAULIS5 expression has more severe consequences on the development of the plant than similar defects in SPMS gene expression [6]. In A. thaliana, it causes severe dwarfism and overproliferation of xylem vessels [29]. The application of TSPM can reverse these effects to some extent [30].

Although crystal structures of different aminopropyltransferases, such as human SPDS [31] and SPMS [32], several other eukaryotic [33,34] and prokaryotic [35] representatives of this group of enzymes, including examples from thermophilic organisms [36–38], are available, not much is still known about features that are responsible for the substrate discrimination in plant SPDS, SPMS, and TSPS. In this work, we describe six crystal structures of TSPS from the model legume plant Medicago truncatula (Mt). These are as follows: the structure of MtTSPS without ligands and its complexes with the reaction substrate (SPD), the reaction product (TSPM), 5′-methylthioadenosine (MTA), S-adenosylthiopropylamine (ATAM), and adenosine (ADN). Crystallographic results combined with an ab initio envelope obtained with the use of small-angle X-ray scattering (SAXS) reveal a tetrameric biological assembly of MtTSPS and give insights into the catalytic action of the enzyme responsible for the production of an essential tetraamine. We also combine available structural and sequence data of plant aminopropyltransferases to emphasize the differences around the active site which are responsible for the specificity of SPDS, SPMS, and TSPS.

Materials and methods

Cloning, overexpression, and purification of MtTSPS

Complementary DNA (cDNA) of M. truncatula was obtained with the use of SuperScript II reverse transcriptase (Life Technologies), as well as oligo dT (15 and 18) primers and total RNA isolated from leaves with an RNeasy Plant Mini Kit (Qiagen). The MtTSPS open reading frame, annotated in the GenBank entry XM_003610659.2 [39] as locus MTR_5g006140, was isolated by polymerase chain reaction with primers (forward: TACTTCCAATCCAATGCCATGGGTGAAGTAGCTTACACAAATGGAAAT and reverse: TTATCCACTTCCAATGTTATTATGCATTCTTTCCATGACCATATATGAACCTA) and cDNA as a template. A ligase-independent cloning [40] protocol was applied prior to incorporating the MtTSPS gene into a pMCSG68 vector (Midwest Center for Structural Genomics). The vector with the MtTSPS gene was used for the transformation of BL21 Gold Escherichia coli competent cells (Agilent Technologies). The overexpression yielded the protein with the N-terminal His₆-tag followed by the tobacco etch virus (TEV) protease cleavage site. The culture was carried out at 37°C in lysogeny broth medium with the addition of ampicillin (150 μg/ml) until OD₆₀₀ reached value 1.0. Before induction with the final concentration of 0.5 mM of isopropyl-β-D-thiogalactopyranoside, the culture was cooled to 18°C. After 16 h of overexpression, the culture was cooled to 4°C. Next, the cells were pelleted by centrifugation at 3500 g for 20 min. Cell pellets were resuspended in 35 ml of the binding buffer [50 mM HEPES (pH 7.4); 500 mM NaCl; 20 mM imidazole; 1 mM tris(2-carboxyethyl)phosphine, TCEP] and frozen at −80°C. Sonic disruption of the cells was carried out on the thawed culture placed in an ice/water bath for 4 min of total sonication time with bursts of 4-s and 26-s intervals. Cell debris was then pelleted by centrifugation at 25 000 g for 30 min at 4°C. The supernatant was applied on the column packed with 5 ml of HisTrap HP resin (GE Healthcare) connected to Vac-Man (Promega). After the application of the supernatant, the column was washed five times with 40 ml of the binding buffer. His₆-tagged MtTSPS protein was eluted with 20 ml of elution buffer [50 mM HEPES (pH 7.4), 500 mM NaCl, 400 mM imidazole, and 1 mM TCEP]. His₆-tagged TEV protease (final concentration of 0.1 mg/ml) was mixed with the protein solution in order to cleave the His₆-tag from MtTSPS. Afterward, the sample was subjected to overnight dialysis at 4°C against the buffer: 50 mM HEPES (pH 8.0), 500 mM NaCl, and 1 mM TCEP. Cleaved His₆-tag and His₆-tagged TEV protease were removed on HisTrap HP resin. The final step of MtTSPS purification was performed on the AKTA FPLC system (Amersham Biosciences) through size-exclusion chromatography on a HiLoad Superdex 200 16/60 column (GE Healthcare) equilibrated with the buffer: 50 mM HEPES (pH 7.4), 100 mM KCl, 50 mM NaCl, and 1 mM TCEP.

The MtTSPS-R37E mutant was designed by introducing the R37E mutation to the MtTSPS gene-containing vector used for the production of wild-type MtTSPS, according to the Polymerase Incomplete Primer Extension method [41]. The protocol of overexpression and purification was the same as for the wild-type protein. Sequences of MtTSPS and MtTSPS-R37E were confirmed by DNA sequencing of the isolated plasmids.

Crystallization and data collection

The MtTSPS sample was concentrated with Amicon concentrators (Millipore) to the final concentration of ~10 mg/ml, determined by the absorbance measurement at 280 nm, with an extinction coefficient of 47 330. Initial crystallization conditions for MtTSPS were established as the 92nd reagent of Index Screen HT (Hampton Research). The protein was then crystallized by the hanging drop method in conditions containing 10% polyethylene glycol (PEG) 3350, 0.1 M MgCl₂, and 0.1 M buffer of either 2-(N-morpholino)ethanesulfonic acid (MES) or Bis–Tris propane, as well as a pH of 6.5.

The complexes of MtTSPS with TSPM, ADN, MTA, and ATAM were obtained by the co-crystallization of MtTSPS with a 10-molar excess of ligands. The complex of MtTSPS–SPD was obtained by soaking the MtTSPS crystal in a crystallization drop containing 5 mM SPD for 30 min. Crystals were cryoprotected by the transfer to the solution containing 10% PEG 3350, 0.1 M MgCl₂, 0.1 M Bis–Tris propane (pH 6.5), and 33% PEG 400 or 25% glycerol.

The diffraction data were collected at the SER-CAT 22-ID and SBC 19-ID beamlines at the Advanced Photon Source (APS), Argonne National Laboratory, U.S.A. The diffraction data were processed with XDS [42] and HKL-3000 [43]; for details, see Table 1.

Table 1.

Data collection and refinement statistics.

Structure	MtTSPS	MtTSPS–ATAM	MtTSPS–ADN	MtTSPS–MTA	MtTSPS–TSP	MtTSPS–SPD
Data collection
Beamline	22-ID	22-ID	19-ID	19-ID	19-ID	19-ID
Wavelength (Å)	1.00	1.00	0.979	0.979	0.979	0.979
Temperature (K)	100	100	100	100	100	100
Space group	P43212	P43212	P43212	P43212	P43212	P43212
Unit cell parameters a, c (Å)	80.5, 164.4	81.1, 163.1	81.0, 163.8	82.8, 164.2	80.3, 163.2	80.0, 163.5
Oscillation range (°)	0.3	0.5	0.3	0.3	0.3	0.3
Resolution (Å)	1.68 (1.78–1.68)	1.91 (2.00–1.91)	1.89 (2.01–1.89)	1.80 (1.91–1.80)	1.65 (1.75–1.65)	1.95 (2.07–1.95)
Reflections collected/unique	585 291/62 270	239 861/43 034	566 760/44 135	464 637/53 857	835 184/65 012	335 996/39 529
Completeness (%)	99.7 (98.4)	99.9 (100)	99.8 (98.9)	99.8 (99.1)	99.8 (99.1)	99.9 (99.6)
Multiplicity	9.4 (9.1)	5.6 (5.6)	12.8 (13.0)	8.6 (8.6)	12.8 (12.8)	8.5 (8.8)
Rmerge (%)	5.9 (105.8)	7.5 (74.1)	6.7 (126.8)	6.2 (104.9)	6.8 (128.7)	4.9 (77.9)
〈I/σ(I)〉	20.9 (2.0)	17.4 (2.1)	23.9 (2.4)	22.2 (2.5)	23.5 (2.1)	20.7 (2.2)
Refinement
Rfree reflections	1059	1060	1016	1078	1041	1028
No. of atoms (non-H)
Protein	4678	4718	4686	4708	4745	4685
Ligands	6	46	81	96	28	20
Solvent	302	163	180	317	341	113
Rwork/Rfree (%)	17.6/21.7	18.9/23.8	18.7/23.5	17.6/21.6	17.3/23.0	18.7/24.4
Mean ADP (Å2)	31.3	41.7	43.1	37.0	29.9	66.0
RMSD from ideal geometry
Bond lengths (Å)	0.01	0.01	0.02	0.02	0.01	0.01
Bond angles (°)	1.6	1.6	1.8	1.7	1.7	0.92
Ramachandran statistics (%)
Favored	98	98	97	98	97	96
Allowed	2	2	3	2	3	4
Outliers	0	0	0	0	0	0
PDB code	6BQ2	6BQ3	6BQ4	6BQ5	6BQ6	6BQ7

Values in parentheses refer to the highest resolution shell. Abbreviations: ADP: atomic displacement parameter.

Structure determination and refinement

The structure of MtTSPS was solved with the use of Phaser [44]. SPDS from Thermotoga maritima [38] (chain A of Protein Data Bank [PDB] entry 1INL) was used as the search model. The initial solution with two monomers in the asymmetric unit was rebuilt in PHENIX AutoBuild [45]. The model was then taken for the subsequent steps of manual and automatic refinement with Coot [46] and Refmac [47]. This was used as an initial model for the determination of other MtTSPS complexes through rigid-body refinement. TLS parameters [48,49] were applied at the later stages of the structure refinement. All ligands were refined using standard CCP4 libraries [50]. The quality of refined structures was controlled by R_work, R_free factors [51], and geometric parameters. PROCHECK [52] and MolProbity [53] were used for the evaluation of the final models. The final refinement statistics are given in Table 1.

Small-angle X-ray scattering measurement

SAXS data were collected from 5 mg/ml protein solution at the BioCAT 18-ID beamline [54] at APS. Prior to the SAXS analysis, the sample was subjected to size-exclusion chromatography using a Superdex 200 column (GE Healthcare), which was directly coupled to the SAXS cell. The position of the protein peak was determined by examining the scattered intensity. Using a PILATUS 3 1 M detector (Dectris), data were recorded at a 1.03 Å wavelength at room temperature, with 0.5-s exposures every 3 s. The sample-to-detector distance was 3.5 m, and the data range was 0.0052–0.37 Å⁻¹ (q = 4π sin θ/λ, where 2θ is the scattering angle and λ is the X-ray wavelength). Frames corresponding to the elution peak of the chromatogram were averaged to maximize the signal-to-noise ratio. Several frames immediately proximal to the sample peak (buffer frames) were averaged and subtracted from the sample scattering to obtain the final SAXS curve (Figure 1A). Data reduction was performed using the BioCAT beamline software pipeline, which utilizes ATSAS [55]. PRIMUS [56] was used for the Guinier analysis (Figure 1B) and radius of gyration (R_g) calculation. GNOM [57] was used to calculate the R_g and the pair distribution function, P(r) (Figure 1C), with the use of the entire scattering pattern. The R_g value calculated from the Guinier and distance distribution analysis was 37 Å. The qR_g limits for further calculations were 0.34–1.30. The calculated maximum dimension of the particle (D_max) was 105 Å. DAMMIF [58] was used for the calculation of low-resolution ab initio models. The models were averaged with DAMAVER [59], refined with DAMMIN [60], and filtered with DAMFILT. The symmetry 222 was applied as the restraint for the model building. SUPCOMB [61] was used for the superimposition of the SAXS model, with co-ordinates of the tetrameric assembly obtained from the crystal structure of MtTSPS.

Figure 1. SAXS data.

(A) The experimental curve for MtTSPS. (B) Guinier plot (blue dots) of the scattering curve with the best fit shown as a dashed black line. (C) Pair–distance distribution function for MtTSPS SAXS data.

Other software used

Molecular illustrations were created with UCSF Chimera [62] and PyMOL (Schrödinger, LLC). The Ramachandran plot was calculated in Rampage [63]. The secondary structure was recognized with ProMotif [64] within the PDBsum server [65]. Sequence alignments were performed in CLUSTAL W [66] and MUSCLE [67] and edited in BioEdit [68]. MEGA7 [69] was used for phylogenetic analysis. HingeProt server [70] was used for the analysis of flexible regions. For the calculation of the sequence conservation, WebLogo was used [71].

Database deposition

The coordinates and structure factors of the related structures were deposited in the PDB: 6BQ2 (MtTSPS), 6BQ3 (MtTSPS–ATAM), 6BQ4 (MtTSPS–ADN), 6BQ5 (MtTSPS–MTA), 6BQ6 (MtTSPS–TSPM), and 6BQ7 (MtTSPS–SPD).

Results and discussion

MtTSPS structure

The polypeptide chain of MtTSPS contains 328 amino acids, and its calculated molecular mass is 37 kDa. It crystallizes in the P4₃2₁2 space group, with a dimer (chains A and B) as the asymmetric unit and 35% solvent content. The protein chain is well defined in electron density, except for 23 N-terminal residues. The monomer of MtTSPS consists of two domains and has a fold typical for polyamine biosynthesis proteins. The N-terminal domain is built by six β-strands (N-terminal β-hairpin followed by a four-stranded antiparallel β-sheet), whereas the C-terminal domain has Rossman fold-like topology. That is, β-strands (five parallel and two antiparallel) placed in the core are covered from both sides by five and seven helical fragments (Figure 2).

Figure 2. Monomer of MtTSPS.

(A) Topological diagram of MtTSPS with secondary structure elements. Helices (cylinders) and sheets (arrows) are depicted in red and blue, respectively. (B) Structure of MtTSPS monomer with highlighted active site as solid surface representation. SPD- and dc-SAM-binding sites are shown with light green and purple surfaces, respectively.

The analysis of the assembly through the Proteins, Interfaces, Structures, and Assemblies server [72] showed that both the dimeric and the tetrameric forms of MtTSPS may be stable in the solution. This, together with the inconclusive estimation of the molecular mass of MtTSPS based on elution volume during size exclusion, was the motivation for the further analysis of MtTSPS biological assembly. The high-quality data obtained from SAXS experiments have led to the ab initio model of MtTSPS (Figure 3A) and the determination that the D_max of the protein is 105 Å. Both the model and calculated D_max from SAXS data are consistent with the crystallographic structure, where the asymmetric unit actually presents only half of the biological assembly of a tetrameric MtTSPS (Figure 3B). The tetramer in the crystal lattice is obtained by a twofold crystallographic symmetry operation on the axis diagonal between crystallographic directions a and b. The longest dimension of MtTSPS calculated from the crystal structure is ~100 Å. The four monomers of MtTSPS create an intersubunit eight-stranded antiparallel β-barrel in the center of the tetramer, which is formed by the N-terminal β-hairpin of each subunit (Figure 3B). The MtTSPS molecule has 222 symmetry, where the three perpendicular twofold axes are crossed in the center of the β-barrel. The tight dimers, which are formed by chains A and B in the asymmetric unit, have the longest diameter of ~87 Å. In the tetrameric formation, two A–B dimers are mutually twisted ~72° around the longer axis of the tetramer (Figure 3B). The SAXS envelope exhibits a significantly less profound twist (Figure 3A), presenting the shape of the MtTSPS tetramer more similar to that of red blood cells. This feature stems from the presence of a hinge-like, flexible connection between the N-terminal and C-terminal domains of MtTSPS subunits. The additional analysis of the structure with the HingeProt server [70] indicates that in MtTSPS, there are three distinctive structural regions (Figure 4): rigid core fragment, which complies with almost entire C-terminal domain, two helices α3-η4/α4 (residues Ile130-Phe149) with moderate flexibility, and third, most labile block, formed by N-terminal domain up to Tyr84 of α1 helix. Therefore, the hinge residues are Tyr84, Ile130, and Phe149. Figure 4 shows superposition of the predicted chain conformations obtained by adding the fluctuation vectors at the extreme position. The highest conformational adaptability is presented by N-terminal β-hairpin and β-sheet β3–β6 (marked with cyan and red in Figure 4).

Figure 3. Biological assembly of MtTSPS.

(A) Ab initio averaged envelope (blue mesh) obtained from SAXS data modeling with a superposed crystallographic tetramer. (B) Crystal structure of MtTSPS with transparent surface representation. Chains A and B of the asymmetric unit are shown as cyan and green ribbons, respectively.

Figure 4. Molecular flexibility of the MtTSPS monomer.

The predicted extent of molecular movements is presented as a superposition of two states (I and II) with a maximal movement amplitude. Monomer of MtTSPS was divided into three regions: core region with lowest flexibility colored in pale cyan (state I) and light pink (state II), α₃-η4/α4 region with moderate flexibility depicted with blue (I) and purple (II) and N-terminal domain with the highest conformational adaptability, cyan (I) and red (II). Hinge residues were marked with black dots. Analysis was made with Hingeprot [70].

In solution, the respective dimers are able to rotate to some extent, maintaining the intersubunit β-barrel. This conformational flexibility may improve the efficiency of the catalysis, since the active site is formed at the interface of the N-terminal and C-terminal domains (see the section ‘Catalytic site,’ below). The ability of the enzyme to alter the relative orientation of its domains and change the shape of its catalytic site (especially in the region of polyamine binding) most probably facilitates the removal of the TSPM product after reaction. The seven residues from β2 strands and five residues from β1 form an antiparallel intersubunit β-barrel with the topology β1_A–β2_A–β2_B–β1_B–β1_A★–β2_A★–β2_B★–β1_B★ (asterisks indicate symmetry-related monomers), where each β-strand is antiparallel to the neighboring one. Additionally, within the barrel, a pair of salt bridges are created between Arg37 and Glu29 of the symmetry-related monomers. Interestingly, in the designed MtTSPS-R37E, the introduction of a repulsive interaction within the barrel shifted the retention volume of protein toward smaller sizes by ~7 ml, which indicated half of the size of wild-type MtTSPS. The stability of the dimeric form of MtTSPS might be important within the cell, since A. thaliana studies show that two other aminopropyltransferases, AtSPDS and AtSPMS, can form aminopropyltransferase heteromultimers [73]. This would be an interesting aspect for further studies on TSPS enzymes, especially since TSPS was initially misidentified in the literature as a protein with SPMS activity [28].

Catalytic site

The active site of MtTSPS is a narrow, elongated, and negatively charged groove placed in the interface between the N-terminal and C-terminal domains. This groove can be divided into two compartments, dc-SAM and polyamine-binding sites (Figure 2B).

Three crystal structures — (i) with the cofactor after reaction (MtTSPS–MTA, resolution 1.80 Å), (ii) with the demethylated cofactor analog (MtTSPS–ATAM, 1.91 Å), and (iii) with adenosine (MtTSPS–ADN, 1.89 Å) — show that the cofactor binding location is inside the deeper cleft above the loop connecting β4–β5. In the three complexes, the ligands are bound with the planes of the purine bases placed between Ile130 and Leu179 (Figure 5). The N⁶-amine group of the adenine base creates hydrogen bonds with the carboxylate moiety of Asp160 and with the carbonyl oxygen of Pro187. The N1 atom of adenine also creates a hydrogen bond with the backbone amide of Ala161. The hydroxyl groups of ribose form two hydrogen bonds with the carboxylate of Asp129 and a water-mediated hydrogen bond with the peptide amide of Glu109. O2′ also interacts with the side-chain amide of Gln54. The cofactor analogs bind within the dc-SAM-binding site in a way that one distant chamber, close to His85, remains empty. Based on the complexes of human SPDS (HsSPDS, PDB ID: 2O0L) [32] and SPDS from Plasmodium falciparum (PfSPDS, PDB ID: 2PT6) [33] with dc-SAM, the aminopropyl group of dc-SAM fills this chamber during the catalytic event. Such location of the aminopropyl group secures the proper position of the carbon adjacent to sulfur, essential for the aminopropyl transfer to the polyamine substrate bound in the neighboring groove. In MtTSPS, the placement of the aminopropyl moiety in this compartment would lead to the creation of hydrogen bonds between the terminal amine of dc-SAM and the side chains of Asp178 and His185. This is not the case in the MtTSPS–ATAM complex, in which the aminopropyl moiety is shifted toward Tyr84, Tyr251, and Gln75, where terminal amine creates three hydrogen bonds with these residues (Figure 5C). In general, polyamine synthases seem to be highly specific in the context of the aminopropyl donor. In the bacterial aminopropyltransferase study, the substitution of the six-amine group of the adenine base with a hydroxyl group led to a loss of activity of the compound as the aminopropyl donor [74]. The MtTSPS–ATAM structure where MtTSPS is complexed with the dc-SAM analog, which lacks the methyl group on the sulfur atom, shows that the enzyme also requires a specific configuration and a charge on the sulfur atom. Otherwise, the aminopropyl moiety binds in a conformation, which precludes SPD from binding in the polyamine cleft.

Figure 5. Cofactor binding site.

MTA (A), ADN (B), and ATAM (C) bound in the catalytic site of MtTSPS. 2F_O – F_C electron density maps (blue mesh) are contoured at 1 σ.

The polyamine-binding compartment of the active site is perfectly shaped to facilitate the SPD substrate during catalysis. It is a groove that runs between the β6 strand from one side and the β11, β13 strands from the other side. In the MtTSPS–SPD complex, the N10 amine of SPD (at the aminobutyl end of the ligand) interacts with the carboxylate of Glu30, the residue that restricts the length of the polyamine groove at this side of the pocket (Figure 6A). The central part of the binding site is constituted mostly by aromatic residues (Tyr251, Trp255, and Tyr84), which create hydrophobic interactions with a polyamine. The sole exception is Gln75, which makes a hydrogen bond with N5 of SPD. Inside the catalytic venue, the N1 amine of SPD interacts with the carboxylate of Asp178. During the catalysis, Asp178 most probably deprotonates the amine group of SPD, thus enabling it to perform a nucleophilic attack on the carbon of the aminopropyl group of dc-SAM, which yields TSPM. TSPM in the MtTSPS-TSPM complex is bound with aminopropyl group pointing outside the pocket (Figure 6B), showing conformation of the product without the cofactor discarded after catalysis. Interestingly, in the complex of MtTSPS–MTA, the polyamine groove is occupied by Bis–Tris propane, bound from the crystallization buffer. The molecule creates several hydrogen bonds with the protein at both ends of the polyamine site (Figure 6C). It would be very interesting to investigate the inhibitory properties of novel compounds with a Bis–Tris propane backbone or in fusion with MTA, especially since MTA has already been reported as an efficient inhibitor of some aminopropyltransferases [75].

Figure 6. Substrate-binding site.

(A) SPD-binding site. (B) Catalytic site of MtTSPS with superimposed bound ligands: ATAM (purple), SPD (light green), and TSPM (dark green). (C) Bis–Tris propane (BTP) bound in the catalytic site in the MtTSPS–MTA complex. 2F_O – F_C electron density maps (blue mesh) are contoured at 1 σ.

Comparison of MtTSPS tetrameric assembly with other aminopropyltransferases

The closest analog of MtTSPS with a known crystal structure is the polyamine aminopropyltransferase from Thermus thermophilus (TtPAAPT) with 34.8% sequence identity (PDB ID: 1UIR) [37]. This protein, similar to TSPS, presents a G-G-G-E-G motif within the binding site of dc-SAM. TtPAAPT also has a very similar tetrameric assembly to that observed in MtTSPS. The β-barrel created by the N-terminal β-hairpins of four units of TtPAAPT is analogous to that of MtTSPS. However, inside the β-barrel of TtPAAPT, there is a highly hydrophobic environment, while in MtTSPS there are two Arg37-Glu29 salt bridges. Similar features are observed in SPDS from T. maritima (PDB ID: 1INL) [38]. Other structurally similar proteins in the PDB, despite the presence of the N-terminal β-hairpin, do not form tetramers as MtTSPS does. This is also the case of the only plant aminopropyltransferase with a known structure, SPDS1 from A. thaliana (AtSPSD1, PDB ID: 1XJ5). AtSPSD1 presents 24.9% sequence identity with MtTSPS, and the lowest sequence homology is within the N-terminal region (Supplementary Figure S1). Although monomers of both proteins are structurally very similar [root-mean-square deviation (RMSD) 1.64 Å], as are their pairs of the tight dimers (A–B, A^★–B^★ of MtTSPS and A–D, C–B of AtSPDS1), the tetrameric assembly of MtTSPS is significantly different from that of AtSPDS1 (Figure 7A). In AtSPDS1, there is no β-barrel of the topology observed in MtTSPS, β1_A–β2_A–β2_B–β1_B–β1_A^★–β2_A^★–β2_B^★–β1_B^★ (Figure 7B). Instead, the respective pairs of monomers in AtSPDS1 (A–D and B–C) are shifted in the tetrameric assembly, together creating two separate β-sheets. These β-sheets have the following topology: β1_D–β2_D–β2_A–β1_A–β3_C–β4_C–β5_C–β6_C for interacting monomers D, A, C (Figure 7C), and, analogously, β1_B–β2_B–β2_C–β1_C–β3_A–β4_A–β5_A–β6_A for monomers B, C, A.

Figure 7. The comparison of MtTSPS with AtSPDS1.

(A) Superposition of the MtTSPS (cyan) and AtSPDS1 (purple) tetramers. (B) Intersubunit β-barrel created by the N-terminal β-hairpins of the four monomers of MtTSPS (symmetry-related monomers are marked with an asterisk). (C) N-terminal intersubunit β-sheet of AtSPSD1.

Most of the other structurally characterized aminopropyltransferases are homodimeric proteins. This also includes human enzymes HsSPMS (PDB ID: 3CM6) [32] and HsSPDS (PDB ID: 2O05) [31]. HsSPMS has a much larger N-terminal domain (~120 additional amino acids), consisting of six antiparallel β-strands and two helices. In the dimer, these N-terminal domains of HsSPMS are swapped. But, HsSPMS has a shorter C-terminus, which lacks the three helices present in MtTSPS, α10, α11, and α12. These features of HsSPMS promote C-terminal domains to be even more labile than in MtTSPS. Conversely, HsSPDS, which has only 25% identity with MtTSPS, shows a much higher structural homology with MtTSPS (RMSD 1.75 Å), but in this case, due to a dimeric assembly, the N-terminal β-sheets form only half of the β-barrel.

Substrate discrimination by plant aminopropyltransferases

The phylogenetic tree (Figure 8) of the flowering plants aminopropyltransferases shows a clear divergence of TSPS from the other three clades of related enzymes (SPDS, SPMS, and putrescine N-methyltransferase, PMT). But, SPDS and SPMS are more closely related. Also, the sequence comparison of M. truncatula aminopropyltransferases shows that MtTSPS shares only 25.7% sequence identity with MtSPDS and 23.4% with MtSPMS. However, among plant TSPS homologs, there are enzymes, like TSPS from chickpea, that show over 90% sequence identity with MtTSPS. A. thaliana TSPS (AtTSPS), a more distant homolog, has 60.5% identity. The closer relation between TSPS enzymes from different species than the relation between TSPS with other polyamine synthases of the same species clearly suggests the separate evolutionary routes of TSPS from SPDS and SPMS enzymes.

Figure 8. Phylogenetic tree of flowering plant aminopropyltransferases.

The analyzed protein sequences were assigned to the thermospermine synthases (TSPS, 171 sequences), spermine synthases (SPMS, 188), spermidine synthases (SPMS, two subgroups of proteins with 197 and 47 sequences marked in violet and light violet, respectively), and putrescine N-methyltransferases (PMT, 42). The tree was made in MEGA7 [69] using the Neighbor-Joining method [76]. The analysis involved 645 sequences of flowering plant proteins classified to the Polyamine Biosynthesis Domain (IPR030374) by InterPro [77] with the length between 210 and 460 amino acids; significant outliers were manually excluded.

Several distinctive features of the catalytic cleft of plant TSPS, SPMS, and SPDS, which determine their role in the production of SPD, SPM, and TSPM, draw attention during the structural and sequence analysis. The first aspect that differs TSPS from SPDS and SPMS is the dc-SAM-binding site. Three very conserved positions are present only in TSPS enzymes: His85, Glu109, and Asp129 (Figure 9; sequence positions of each residue refer to the MtTSPS sequence). Glu109 and His85 are placed in the compartment responsible for interaction with terminal amine of dc-SAM aminopropyl group, whereas Asp 129 recognizes ribose residue of dc-SAM. In SPMS and SPDS, these residues are Gln85, Asp109, and Glu129. This difference does not change the binding mode of dc-SAM; however, due to the exchange between Asp and Glu, the cofactor bound in TSPS is slightly shifted. This allows for somewhat different positioning of the polyamine substrate correlated with the location of its N5 amine.

Figure 9. Sequence conservation around the catalytic site of plant aminopropyltransferases.

Sequences of TSPS, SPMS, and SPDS are aligned and numbers refer to the sequence position in MtTSPS. Representation prepared in WebLogo [71] using multiple sequence alignment of the protein sequences assigned to each group of enzymes by phylogenetic analysis. Each position is presented as stacks of symbols, where the overall height of the stack indicates the sequence conservation at that position, while the height of symbols within the stack indicates the relative frequency of each amino acid at that position. Violet color denotes residues mentioned in the text, whereas red highlights residues which distinguish TSPS from SPMS and SPDS.

Far more extensive differences exist in the polyamine-binding groove of aminopropyltransferases that influence the substrate recognition. Especially striking are changes in the distant part of the pocket, in particular the region of N-terminal β-hairpin (close to Glu30 in TSPS), also crucial for the multimerization of aminopropyl transferases. TSPS enzymes have the shortest turn between β₁ and β₂ with only two residues in the loop (positions 34, 35) (Figure 9). SPDS and TSPS have a conserved Glu residue in position 30; however, the loop region is two residues longer in SPDS and presents a much more hydrophobic character with bulky Trp33. As seen in Figure 10A,B, the β₁ strand in AtSPDS is shorter than in MtTSPS and the whole β₁–β₂ region is moved toward the active site. Additionally, Trp33 side chain is placed inside the pocket, close to Pro252 in the loop between η₉ and β₁₃. These features of SPDS shorten the polyamine groove in keeping with the shorter length of the substrate, PUT. The absence of Pro252 and Trp33 in TSPS provides the possibility of forming a hydrogen bond between Glu30 and the longer SPD substrate. In SPMS, the loop region of β₁–β₂ is significantly extended with up to 20 residues and conserved two Asn residues close to the β₂ region. One of these two Asn residues is a good potential binder of the terminal amine of bound SPD, although there is no structural support available for this concept. Most probably, the terminal amine of SPD bound in plant SPMS is placed similar to that of SPD bound by HsSPMS (Figure 10C).

Figure 10. Architecture of the catalytic site of aminopropyltransferases.

Close-up to the polyamine groove in the structure of MtTSPS in complex with SPD (A), AtSPDS (B), and HsSPMS in complex with SPD (C). For clarity of the comparison, all sequence positions refer to the sequence of MtTSPS.

The next aspect of SPD binding by TSPS and SPMS, and thus the preference toward the production of either TSPM or SPM, is the position of the middle amine group of SPD inside the active site. In the middle part of the pocket, TSPS has Trp255 instead of Ile in the other enzymes, which pushes the substrate toward the same direction as the shift of the cofactor (discussed above). The middle amine group of SPD is directed away from the hydrophobic Trp255 and points toward Gln75 forming hydrogen bond with its side chain (Figure 10A). This favors the position of the middle amine group closer to the catalytic Asp178, thus orienting the aminopropyl and aminobutyl parts of SPD to produce TSPM instead of SPM. The highly conserved in all plant enzymes Asp181, placed in the loop between β₁₀ and η₆, in TSPS is H-bonded with Gln214 and does not interact with the substrate. This is made possible by the presence of Gly216 instead of Glu in other enzymes. In plant SPMS and SPDS, there is no Trp255 or Gly216 equivalent and the loop after Asp181 is shorter. As a consequence, Asp181 is closer to the polyamine groove and one of its Oε atoms is hydrogen bonded with Glu216 (one of the carboxyl groups has to be protonated). The second Oε atom of Asp181 is well poised to H-bond with the N5 amine of SPD oriented further from the catalytic Asp178, opposite to TSPS. Gln75 in SPDS is pushed away from the groove (Figure 10B).

The differences in specificity of plant aminopropyltransferases can be plausibly explained as follows. In SPDS and SPMS, the substrate is recognized through the length of the diaminobutyl moiety by ensuring the appropriate distance between two Asp178 and Asp181. The difference between these two enzymes is that SPMS has an extended binding pocket to accommodate the longer substrate, which is connected with the presence of a long loop between the β₁–β₂ sheet. In TSPS, the different orientation of SPD is favored by the interaction of the middle amine group with Gln75, which is made possible by the presence of Trp255 and a different conformation of Asp181 due to the absence of Glu216.

Conclusions

The presented work describes the first structural investigation of a plant translation product of the ACAULIS5 gene, thermospermine synthase from M. truncatula. The high-resolution crystal structure of MtTSPS combined with the low-resolution solution SAXS envelope revealed the tetrameric assembly of this protein. The MtTSPS protein forms a homotetramer through the interaction of the N-terminal β-hairpin from each subunit, which collectively build an eight-stranded β-barrel. Such a tetrameric assembly gives the enzyme an ability to adjust the volume of the active site through a relative movement of C-terminal domains with respect to the N-terminal intersubunit β-barrel. This feature can be useful for discarding the products after reaction or acquir-ing the reaction substrates before the catalytic event. Moreover, five complexes of MtTSPS — with the triamine substrate (SPD); with the product (TSPM); with the cofactor after aminopropyl group transfer (MTA); and with two cofactor analogs, demethylated dc-SAM (ATAM) and ADN — give insights into the architecture of the active site of MtTSPS and the catalytic mechanism of the reaction of aminopropyl group transfer to SPD. A detailed sequence and structural comparison with other aminopropyltransferases identified the unique features of plant aminopropyltransferases that are probably responsible for the characteristic substrate discrimination and differentiate TSPS, SPDS, and SPMS enzymes. Since TSPM is essential for plant development, our crystallographic data can be utilized for the design of herbicides based on the TSPS inhibitors.

Abbreviations

ADN

adenosine

APS

Advanced Photon Source

At

Arabidopsis thaliana

ATAM

S-adenosylthiopropylamine

AtSPSD1

SPDS1 from A. thaliana

cDNA

complementary DNA

dc-SAM

decarboxylated S-adenosylmethionine

HsSPDS

human SPDS

Mt

Medicago truncatula

MTA

5′-methylthioadenosine

PAAPT

polyamine aminopropyltransferase

PDB

Protein Data Bank

PEG

polyethylene glycol

PMT

putrescine N-methyltransferase

PUT

putrescine

RMSD

root-mean-square deviation

SAXS

small-angle X-ray scattering

SPD

spermidine

SPDS

spermidine synthase

SPM

spermine

SPMS

spermine synthase

TCEP

tris(2-carboxyethyl)phosphine

TEV

tobacco etch virus

TAAPT

triamine/agmatine aminopropyltransferase

TSPM

thermospermine

TSPS

thermospermine synthase

TtPAAPT

polyamine aminopropyltransferase from Thermus thermophilus.