|Home | About | Journals | Submit | Contact Us | Français|
The synthesis of selenocysteine-containing proteins (selenoproteins) involves the interaction of selenocysteine synthase (SelA), tRNA (tRNASec), selenophosphate synthetase (SelD, SPS), a specific elongation factor (SelB), and a specific mRNA sequence known as selenocysteine insertion sequence (SECIS). Because selenium compounds are highly toxic in the cellular environment, the association of selenium with proteins throughout its metabolism is essential for cell survival. In this study, we demonstrate the interaction of SPS with the SelA-tRNASec complex, resulting in a 1.3-MDa ternary complex of 27.0 ± 0.5 nm in diameter and 4.02 ± 0.05 nm in height. To assemble the ternary complex, SPS undergoes a conformational change. We demonstrated that the glycine-rich N-terminal region of SPS is crucial for the SelA-tRNASec-SPS interaction and selenoprotein biosynthesis, as revealed by functional complementation experiments. Taken together, our results provide new insights into selenoprotein biosynthesis, demonstrating for the first time the formation of the functional ternary SelA-tRNASec-SPS complex. We propose that this complex is necessary for proper selenocysteine synthesis and may be involved in avoiding the cellular toxicity of selenium compounds.
Selenium has been recognized as an essential trace element for many life forms, although it is toxic at high levels due to the high chemical reactivity of its metabolites (1, 2). Organisms in all three domains of life (bacteria, archaea, and eukarya) synthesize selenocysteine (Sec)3 as the main form of organic selenium in the cells, which is incorporated into specialized proteins, known as selenoproteins, that are involved in several functions including oxidoreductions, redox signaling, and antioxidant defense (1, 3).
Sec is synthesized on the specific l-serine-aminoacylated tRNA (Ser-tRNASec) and incorporated into selenoproteins at UGA codons via a complex pathway that works through transient protein-RNA and protein-protein interactions. In bacteria, this pathway requires the specific tRNASec (SelC) and an mRNA-specific structure called selenocysteine insertion sequence (SECIS) (1, 3). Escherichia coli tRNASec has 8- and 5-bp stems in the acceptor and T arms, respectively, whereas the canonical tRNAs have a 7+5 secondary structure. The D arm of E. coli tRNASec has a 6-bp stem and a 4-nucleotide loop, whereas the canonical tRNAs have a 3–4-bp D stem and 7–12-nucleotide D loop. In addition, the extra arms of the bacterial tRNASer have 5–7-bp stems, in contrast to the 6–9-bp stem observed in E. coli tRNASec (4).
Sec biosynthesis is initiated by the conversion of l-seryl-tRNASec, aminoacylated with serine by seryl-tRNA synthetase (SerRS), to l-selenocysteyl-tRNASec in a reaction catalyzed by selenocysteine synthase (E.C. 184.108.40.206., SelA), which is a pyridoxal 5′-phosphate (PLP)-dependent homodecameric enzyme of ~500 kDa (5). The co-factor PLP is covalently linked to the Lys295 amino acid residue in each monomer of E. coli SelA prior to Ser-Sec conversion (5). Therefore, seryl-tRNASec is linked to SelA in the cofactor site, resulting in a binary complex consisting of one SelAdecamer:10 tRNASec (6). Recently, the structures of Aquifex aeolicus SelA and its binary complex SelA-tRNASec were resolved by x-ray crystallography, highlighting that the decameric conformation is mandatory to provide the catalytic site for binding the tRNA molecule (4).
To achieve Ser-Sec conversion, selenium is transferred to the binary complex on its biologically active form, selenophosphate, a product of the reaction catalyzed by the 72.4-kDa dimeric enzyme selenophosphate synthetase (E.C. 220.127.116.11, SelD or SPS), from selenide and ATP (8). Selenophosphate is produced in a two-step reaction, in which selenide is phosphorylated by the ATP γ-phosphate moiety and then ADP is hydrolyzed, releasing selenophosphate, AMP, and orthophosphate (8,–11). Selenide originates from selenite reduction, from converted methylated selenium compounds, or through selenium removal from selenoprotein degradation (12).
Because the Km value of 20 μm for selenide in vitro results in toxic levels of this compound in the cellular environment, it was hypothesized that SPS in vivo obtains selenide from the PLP-dependent NifS-like enzymes CsdB, CSD, and IscS (12). In E. coli, these PLP-donor enzymes act as β-lyases, catalyzing the cleavage of the C–S bond from Cys or the C-Se bond from Sec to Ala and S0 or Se0, respectively (3, 11, 13). However, an interaction between SPS and NifS-like enzymes has not been described, although a structural basis for the interaction of E. coli CsdB and A. aeolicus SPS was proposed because the molecular surfaces surrounding the active sites of CsdB and SPS exhibit complementarity by molecular docking (10). It is possible that thioredoxin reductase, which is involved in selenite reduction, is also involved in delivering selenide for SPS (3, 13). After selenophosphate is synthesized, it remains bound to the active-site cavity of SPS until ADP hydrolysis occurs and the product release is completed (7, 10).
Itoh et al. (10) hypothesized that SPS could interact with SelA in a manner similar to that of NifS-like proteins, facilitating the efficient transfer of selenophosphate from SPS to SelA; however, this interaction has never been formally proven. Interestingly, the human SepSecS was reported to interact in vivo with the SPS1 isoform (14), but little is known about the mechanism of this interaction. The elucidation of SPS-catalyzed selenium metabolism is important because SPS, rather than the less specific SelA, is responsible for the discrimination between selenium and sulfur in the process of Sec-tRNASec biosynthesis. The structural basis for this specificity is not yet understood.
In this study, we show that SPS functionally interacts with the SelA-tRNASec binary complex, forming the SelA-tRNASec-SPS complex. The macromolecular assembly of the ternary complex follows a stoichiometric ratio of 1SelAdecamer:10tRNASec:5SPSdimer, resulting in a macromolecular structure of ~1.3 MDa, and we provide structural insights into the organization of the ternary complex.
SelA was expressed and purified according to Manzine et al. (15) in binding buffer consisting of 20 mm potassium phosphate (pH 7.5), 100 mm sodium chloride, 5% glycerol, 2 mm β-mercaptoethanol, and 10 μm PLP. The Δ28-SelA truncated N-terminal domain was amplified from selA-pET29a vector using 5′-CATATGGCTATTGATCGCTTATTG-3′ forward and 5′-GCGGCCGCTCATTTCAACAACATCTCC-3′ reverse primers and then ligated into the same vector used by Manzine et al. (15) and transformed into the selA(−) E. coli strain JS1. The DNA sequence of E. coli SPS was amplified from E. coli genomic DNA using 5′-ACTGTATCATATGAGCGAGAACTCGATTCGTTTGACCCAATAC-3′ forward and 5′-TGCACTCGAGTCATTAACGAATCTCAACCATGGCACGACCGAC-3′ reverse primers and ligated into pET28a(+) vector (GE Healthcare). Recombinant SPS was overexpressed at 37 °C overnight in the E. coli BL21 (λDE3) (Stratagene) in LB medium and then harvested at 12,000 × g for 15 min at 4 °C. The pellet was resuspended in buffer A (50 mm Tris/HCl, pH 8.0, 10 mm imidazole, 300 mm NaCl) and lysed by six cycles of 30 s of sonication and 1 min of rest using the 550 Sonic Dismembrator (Fisher Scientific). The soluble fraction was applied to a metal-chelate affinity matrix (nickel-nitrilotriacetic acid, Qiagen) and eluted with 250 mm imidazole, followed by cleavage of the affinity tag using 1 unit of thrombin protease (GE Healthcare) for 100 μg of E. coli SPS. The product was purified to homogeneity using size exclusion chromatography (Superdex 200, GE Healthcare) in 50 mm Tris/HCl buffer, pH 8.0, 300 mm NaCl, and 5 mm DTT. Limited proteolysis of E. coli SPS was performed using chymotrypsin protease (Sigma). SPS (5 mg/ml) was incubated at a protease:protein ratio of 1:50 w/w for 20 min at 18 °C and analyzed by SDS-PAGE. A stable proteolytic fraction was subjected to N-terminal sequencing by Edman degradation (Department of Biochemistry, University of Cambridge). The result from the proteolytic digestion was used to confirm the truncation of the N-terminal sequence of E. coli SPS after the 11th amino acid residue. The Δ11-SPS construct lacking the first 11 amino acid residues was obtained by DNA sequence amplification from E. coli genomic DNA using 5′-AGCATATGAGCCACGGAGCTGGTTGCGGCTG-3′ forward and 5′-AGCTCGAGTTAACGGATCTCAACCATGGCACG-3′ reverse primers and ligated into pET28a(+) vector (GE Healthcare).
We used the protocol described by Manzine et al. (6) to obtain the E. coli tRNASec (5). For fluorescence spectroscopy assays, E. coli tRNASec was labeled with fluorescein maleimide using the 5′ EndTagTM nucleic acid labeling system (Vector Laboratories, Burlingame, CA) according to Manzine et al. (6). E. coli tRNASec oligonucleotides were designed with modified regions (in bold) replaced by corresponding regions of one isoform of the amino acid serine tRNA carrier (tRNASer) of E. coli, as follows: E. coli serine tRNA gene: ′-GGTGAGGTGTCCGAGTGGCTGAAGGAGCACGCCTGGAAAGTGTGTATACGGCAACGTATCGGGGGTTCGAATCCCCCCCTCACCGCCA-3′; E. coli selC gene: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTCGTCTCCGGTGAGGCGGCTGGACTTCAAATCCAGTTGGGGCCGCCAGCGGTCCCGGGCAGGTTCGACTCCTGTGATCTTCCGCCA-3′; acceptor arm mutant: 5′-ACGAATTCTAATACGACTCACTATAGGGTGAGGGTCGTCTCCGGTGAGGCGGCTGGACTTCAAATCCAGTTGGGGCCGCCAGCGGTCCCGGGCAGGTTCGACTCCTGTCCTCACCGCCA-3′; D-loop arm mutant: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTTCCGAGTGGCTGAAGGAGCTGGACTTCAAATCCAGTTGGGGCCGCCAGCGGTCCCGGGCAGGTTCGACTCCTGTGATCTTCCGCCA-3′; anticodon arm mutant: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTCGTCTCCGGTGAGGCGGCACGCCTGGAAAGTGTGTTGGGGCCGCCAGCGGTCCCGGGCAGGTTCGACTCCTGTGATCTTCCGCCA-3′; deleted variable arm construct: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTCGTCTCCGGTGAGGCGGCTGGACTTCAAATCCAGTGGCAGGTTCGACTCCTGTGATCTTCCGCCA-3′; variable arm mutant: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTCGTCTCCGGTGAGGCGGCTGGACTTCAAATCCAGTATACGGCAACGTATGGCAGGTTCGACTCCTGTGACTTTCCGCCA-3′; and TΨC arm mutant: 5′-ACGAATTCTAATACGACTCACTATAGGGAAGATCGTCGTCTCCGGTGAGGCGGCTGGACTTCAAATCCAGTTGGGGCCGCCAGCGGTCCCGGGGGGGTTCGAATCCCCCGATCTTCCGCCA-3′. The amplification, in vitro transcription, and folding were performed as described previously (6)
The functional complementation experiments were conducted according to Sculaccio et al. (16) for N-terminally truncated SPS. Briefly, the E. coli strain WL400 (DE3), which lacks the functional selD gene (7), was transformed with the full-length E. coli SPS sequence and the SPS construct lacking the N-terminal 11 residues (Δ11-SPS). These cells were tested for the presence of an active selenoprotein formate dehydrogenase H (FDH H) using the benzyl viologen assay under anaerobic conditions (16). Similarly, the SelA and N-terminally truncated SelA complementation experiments were performed using this methodology using the E. coli strain JS1 (DE3), which lacks the functional selA gene, under the same anaerobic conditions (16) for 48 h in 30 °C.
Fluorescence anisotropy measurements were performed in an ISS-PC spectrofluorometer (ISS, Champaign, IL). The uncharged tRNASec was fluorescein-labeled, and its interaction with SelA was conducted using 500 nm SelA with 490 nm unlabeled tRNASec and 10 nm fluorescein-labeled tRNASec incubated in binding buffer for 30 min at 25 °C to form the covalently bound binary complex SelA-tRNASec in a final equimolar stoichiometry, according to previous publications (5, 6). The isothermal fluorescence anisotropy assay was performed with fluorescence anisotropy measurements in “L” geometry at 25 °C. A concentrated SPS sample was titrated to a SelA-tRNASec sample, homogenized, and equilibrated for 5 min at 25 °C prior to steady-state anisotropy measurements. The same experimental conditions were applied to fluorescence anisotropy assays using mutant tRNASec constructs. Excitation was set to 480 nm, and emission was recorded through an orange cut-off filter at 515 nm (6). Anisotropy fluorescence values, r, and total intensity of fluorescence were calculated with the ISS program. In all cases, maximal dilution was less than 20%. The resulting fluorescence anisotropy values were fitted, using the program Origin 8.0, to the Hill equation
with r0 and rf representing the initial and final fluorescence anisotropy measures. [SPSmonomer] is the titrated SPS concentration in units of monomers. Thus, the apparent dissociation constant (Kd) and the Hill constant (n) were determined.
Experiments for determination of the stoichiometry of SelA-tRNASec-SPS binding were performed using 5000 nm SelA bound to 4990 nm unlabeled tRNASec and 10 nm fluorescein-labeled tRNASec. The same procedures as described above were used during the SPS titration. Mutant tRNASec molecules were also tested for interaction with SelA by fluorescence anisotropy assays, as described previously (6).
We used hydrogen/deuterium exchange coupled with mass spectrometry to map the surfaces of SelA and SPS following the formation of the SelA-tRNASec binary complex and the SelA-tRNASec-SPS ternary complex. The various samples (SelA, SelA-tRNASec, SPS, tRNASec, and SelA-tRNASec-SPS) were prepared using a published protocol (17). Briefly, the samples were labeled by diluting the sample to a final concentration of D2O of ~90%. At each time point analyzed, aliquots (20 μl) were taken out of the exchange tube and quenched by mixing the solution with a 1:1 ratio of the quenching buffer (D2O, 100 mm sodium phosphate, pH 2.5) and cooled to 0 °C to slow down the H/D exchange. These sample aliquots were digested for 5 min at 0 °C after the addition of 1 μl of a precooled pepsin solution (1 mg/ml in 5% (v/v) formic acid) and were injected directly to the mass spectrometer using a flow of 80 μl/min. The MS experiments were performed with an electrospray ionization triple quadrupole instrument, model Quatro II (Micromass UK), using the same procedures described by Figueira et al. (17). The spectral data were acquired and monitored using the MassLynx software (Micromass); the spectra deconvolution of the intact protein samples was performed with the program Transform (Waters). The theoretical digest was performed using the MS-Digest web server, and the error at each data point was determined to be 0.3 Da (based on multiple measurements).
The structural model of E. coli SelA decamer was obtained using the I-TASSER server (18) that joins multiple threading alignments to rounds of iterative structural assembly simulations for protein structure modeling.
Infrared spectra of protein solutions were collected in a Nicolet Nexus 670 FTIR spectrometer equipped with a DTGS KBr detector, corresponding to 512 scans at a resolution of 2 cm−1 over the wavenumber range 4000–400 cm−1 at 25 °C. During data acquisition, the spectrometer was continuously purged with nitrogen. The buffer spectrum was subtracted digitally from the sample spectrum. The second derivative was used to identify the peak positions of the major components of the amide I band on the original (non-smoothed) protein vibrational spectra. To estimate the secondary structure content, Gaussian curve fitting was performed in the region of 1500–1700 cm−1 using GRAMS/386 software package (Galactic Industries). For FTIR analyses, SelA and SPS were prepared isolated in solution but also in the combinations 1SPS:1SelA, 1SelA:1tRNASec, and 1SelA:1tRNASec:1SPS (molar ratios in monomer units). Difference infrared spectra were used to monitor the initial and the final state of SelA after SelA-tRNASec complex formation obtained by spectrum subtraction of the complex with the isolated samples. The final state of SPS after SelA-tRNASec-SPS complex formation was assessed by subtracting the experimental FTIR signal for the ternary complex, previously subtracted by the FTIR signal of the binary complex, to the SPS spectrum. The combination 1SelA:1SPS was also analyzed.
To analyze the external dimensions, 1 μl of each sample, at 0.5 mg/ml, was incubated in binding buffer without PLP for 40 min at 25 °C, deposited on a mica square (10 × 10 mm), and dried at room temperature for 3 h. This mica square was fixed in a metal base and analyzed in a Bruker Digital Instruments Nanoscope IIIA atomic force microscope (LNNano, CNPEM) using the non-contact mode and silicon tip of 1-nm diameter with 256 lines of scanning (19). The n-Surf 1.0 beta software was used to analyze the images and determine the dimensions of the ternary complex.
To test the hypothesis that SelA-tRNASec interacts with SPS, we isothermally titrated 500 nm SelA-tRNASec binary complex fluorescein labeled with increasing amounts of dimeric SPS in the absence of their substrates. Fluorescence anisotropy of labeled tRNASec, covalently bound to SelA, progressively increased as a function of free SPS concentration (Fig. 1A), resulting in a specific sigmoidal binding pattern. The Hill equation (Equation 1) fitted to the experimental data with an effective dissociation constant of 610 ± 79 nm and n = 2.1 ± 0.4, indicating positive binding cooperativity. Such a dissociation constant value is consistent with a transient biomolecular interaction.
The binding stoichiometry of the ternary complex was determined by isothermal titration of SPS in 5000 nm SelA-tRNASec fluorescein-labeled complex, which is above its dissociation constant value for the interaction. A progressive increase in fluorescence anisotropy was observed as a function of SPS concentration (Fig. 1B) up to 5000 nm. Thereafter, no further change in fluorescence anisotropy was observed, indicating the saturation of the interaction sites at the inflection point of the curve. This pattern is consistent with a SelA-tRNASec binary complex composed of 10 SelA monomers of 50.6 kDa each (resulting in a 506-kDa decamer) bound to 10 tRNASec molecules of 31 kDa each (contributing with 310 kDa to the complex) interacting with 5 dimers of SPS of 72.4 kDa (contributing with 362 kDa to the complex), forming the predicted ternary SelA-tRNASec-SPS complex of ~1.3 MDa, maintained by surface contact between each component.
The observed difference in the fluorescence anisotropy initial values shown in Fig. 1, A and B, for the binary complex was larger than would typically be expected from instrument variation. It may be related to the variation of local viscosity due to the initial binary complex sample concentration being 10 times higher for the stoichiometry measurement experiment when compared with the binding measurement experiment.
H/DEx-MS followed by peptide mapping allowed the specific identification of solvent-accessible exchange sites in the dimeric SPS, the homodecameric SelA, the SelA-tRNASec binary complex, and the SelA-tRNASec-SPS ternary complex. Because SPS binding to SelA-tRNASec disturbs secondary structure elements of both proteins of the binary complex, altering the solvent accessibility of the contact regions, binding interfaces could be mapped by comparing the rates of H/D exchange on proteins in the bound and unbound states (17, 20).
Overall, 41 peptides (including those with overlapping sequences), covering 57% of the SelA primary structure, were identified by tandem MS/MS, as shown by the coverage map (Fig. 2A). The region from Ala14 to Arg17, the SelA N-terminal domain, and regions Ala104–Thr117, Asp146–Cys148, and Ile304–Lys321, show small percentages of deuterium incorporation, even after 30 min of deuterium exposure. Thus, these amino acid residues were hidden within the protein structure, as surface contacts in E. coli SelA decamer in solution, as observed in the crystallographic structure of the homologous A. aeolicus SelA (4).
Following SelA-tRNASec covalent binding, we detected 41 peptides (including those with overlapping sequences), covering 63% of the SelA amino acid sequence (Fig. 2A). Characterization of the solvent accessibility of the N-terminal domain shows that regions Leu27–Gly31 and Leu40–Ile51 are hidden after SelA-tRNASec binding (Fig. 2B). These regions were recently observed to interact with tRNASec D-loop in the crystallographic structure of A. aeolicus SelA-tRNASec (4). Other regions, including fragment Leu137–Ala154 and the amino acid residues near the active site (Lys295), also have low incorporation of deuterium even after 30 min of exposure (Fig. 2, B and C and supplemental Tables 1A and 1B). These regions must be non-covalent SelA-tRNASec contacts on the surface of SelA.
In addition, evaluation of the effect of stereo chemical block in tRNASec interaction with SelA by qualitative fluorescence anisotropy spectroscopy assays showed a decrease in SelA-tRNASec observed binding when the acceptor arm, D-loop, and variable arm were mutated for the corresponding E. coli tRNASer region, highlighting the importance of these regions in SelA-tRNASec specific interaction (Fig. 3, A–G). As a negative control, we titrated fluorescein-labeled single-stranded DNA (Fig. 3G). The interaction pattern of SelA-(mutant) tRNASec binding is similar to that previously observed by Manzine et al. (6) and does not present a saturation plateau because decameric SelA can stack side-by-side and one on top of each other.
The anticodon and TΨC arms variations did not affect the SelA-tRNASec interaction (Fig. 3, A and B, respectively); however, the substitution of the D-loop by a fragment from E. coli tRNASer D-loop caused a decrease in the binary complex interaction (Fig. 3C). These results highlight the D-loop as responsible for the specificity of SelA-tRNASec recognition, which corroborates with the SelA-tRNASec binary complex crystallographic structure from A. aeolicus (Protein Data Bank (PDB) ID 3W1K (4)). Based on amino acid sequence alignment between E. coli and A. aeolicus SelA (data not shown) and A. aeolicus SelA-tRNASec structure analysis (4), we identified by H/D-Ex MS the E. coli SelA Leu27–Gly31 and Leu40–Ile51 regions as interaction points to E. coli tRNASec D-loop and TΨC arms (Fig. 2, A and B).
The deletion of the variable arm or its substitution by the E. coli tRNASer variable arm (Fig. 3, D and E, respectively) and the acceptor arm reduction from 8+5 to 7+5 (Fig. 3F) caused a marked decrease in the anisotropy values. The 8+5 folding is a key difference to other tRNAs and must be an important SelA recognition point that was not identified based on structural analysis (4).
Mapping the surface interactions of SelA to form the ternary complex shows that the N-terminal region (Glu46–Arg52) of SelA and two small loops (Glu67–Asp69 and Ala111–Thr117) have low deuterium incorporation when compared with SelA in the binary complex (Fig. 2B). We believe that these are the most important SelA-SPS interaction regions.
For SPS, 41 peptides were identified, covering 68.7% of the primary structure. Amino acid residues Leu136–Asp143, Ser239–Gly245, and Pro271–His283 presented low rates of deuterium incorporation even after 30 min of exposure. These regions are hidden within the protein and either are near or participate in the SPS dimerization interface (Fig. 2, B and D). The SPS N-terminal loop showed a high deuterium incorporation rate after 5 min of exposure, indicating that it is a flexible region.
It is worth noting that within 30 min, 68.7% of the amide hydrogen atoms in SPS were replaced with deuterium, whereas only 62.5% were replaced in the presence of the SelA-tRNASec binary complex, indicating that some amide protons were protected from deuterium exchange upon ternary complex formation. The SPS N-terminal flexible loop (Met1–Thr9) is hidden from H/D exchange after the interaction of SPS with the binary complex, resulting in lower deuterium incorporation. Two other loop regions (Leu43–Val54 and Met71–Pro72) and an α-helix region (Glu120–Cys129) that are near the catalytic site of the dimeric enzyme become inaccessible to the solvent after the interaction (Fig. 2, B and 2D). Our data identify the regions of molecular contact between the various components of the ternary complex and indicate that the regions near the active sites are crucial to the interaction between SPS and SelA-tRNASec to form a ternary complex.
Structural changes due to SPS binding to the SelA-tRNASec binary complex were investigated via FTIR spectroscopy because the amide I region (1600–1700 cm−1) of the FTIR spectra is sensitive to changes in the protein secondary structure (21,–24). SPS and SelA amide I bands were resolved into seven bands each. The bands appearing at 1628 and 1676 cm−1 are attributed to the low- and high-frequency components of the β-sheet, whereas the band centered at 1665 and 1640 cm−1 (Fig. 4, A and B, respectively) can be assigned to turns and unordered structures, respectively (22,–24). These results indicate that SPS consists of 43.1% α-helix, 15.4% turn, 10.2% β-sheet, and 31.3% random coil, whereas SelA secondary structures consist of 37.0% α-helix, 18.0% turn, 20.0% β-sheet, and 25.0% random coil, consistent with the crystal structure of E. coli SPS PDB ID 3U0O (10) and the circular dichroism spectrum deconvolution of SelA (15), respectively which corroborates the crystallographic structure (PDB.ID 3WCN) (27).
We observed that the amide I absorption band of SelA did not change upon SelA-tRNASec interaction when we analyzed the difference spectrum between SelA-tRNASec binary complex and SelA, which implies that SelA does not have a significant secondary structure variation upon tRNASec binding. Additionally, concerning the ternary complex formation, we propose that the most significant secondary structure change is more likely to be in SPS.
Indeed, there is an evident shift in the amide I absorption band of SPS (Fig. 4C) upon its binding to the SelA-tRNASec binary complex when compared with the SPS sample, indicating that SPS undergoes a conformational change to form the ternary complex. Such a shift was not observed in the absence of tRNASec, implying that the SelA-SPS interaction is dependent on previous tRNASec interaction with SelA.
To further analyze the change in the secondary structure of SPS after its interaction with the SelA-tRNASec binary complex, we obtained a difference spectrum by subtracting the spectrum of free SPS from that of the bound protein, which was previously subtracted by the contribution of SelA-tRNASec (Fig. 4D). The result shows a large negative band of ~1653 cm−1 and a positive band in the 1640–1620-cm−1 range. This pattern can be due to the loss of an α-helical component, as first described by Trewhella et al. (24), indicating that a structural element in the α-helix configuration in SPS loses conformation to enable the formation of the ternary complex SelA-tRNASec-SPS.
Because the H/D change experiment strongly suggested the participation of the SPS N-terminal loop in the SelA-tRNASec-SPS complex assembly, we investigated the potential role of this region in selenoprotein biosynthesis. Previous in situ limited proteolysis experiments with chymotrypsin protease removed the first 11 residues of E. coli SPS (Δ11-SPS) (data not shown), but the catalytic residues Cys17 and Lys20 were preserved. Fluorescence anisotropy of SelA-tRNASec is not altered with Δ11-SPS titration, indicating a lack of specific interaction between Δ11-SPS and the binary complex (Fig. 5A). Functional complementation assays in E. coli strain WL400, which lacks the SPS gene, transformed with Δ11-SPS, were unable to restore the selenoprotein biosynthesis (Fig. 5B), despite the presence of the known catalytic residues. The positive control WL400 transformed with the E. coli SPS gene (Fig. 5C) developed the purple color characteristic of selenoprotein biosynthesis. We also investigated whether this region is required for assembly of the SelA-tRNASec-SPS complex. SPS multiple sequence alignment analysis revealed three highly conserved residues (Leu8, Thr9, and Tyr11) in the SPS N-terminal sequence; however, the biological significance of these residues has not yet been investigated. Together, these results suggest that the SPS N-terminal region is essential to SelA-tRNASec-SPS complex assembly and that its deletion impairs selenoprotein biosynthesis.
Additionally, because H/D exchange experiments showed that the N-terminal domain of SelA is part of its decamerization interface, we also tested its requirement in Sec synthesis in a functional complementation assay in the E. coli strain JS1, which lacks the selA gene. N-terminally truncated SelA was unable to restore Sec synthesis (Fig. 5D) as seen in the positive control E. coli SelA (Fig. 5E). It is worth noting that Methanocaldococcus jannaschii SelA, which lacks an equivalent N-terminal domain but shares 30% amino acid sequence identity with E. coli SelA, is organized as a non-functional dimer and does not interact with tRNASec (25).
Engelhardt et al. (26) were the first to visualize, in 1992, the decamers of SelA and SelA-tRNASec by transmission electron microscopy of negative stained samples. Manzine et al. (6) determined the stoichiometry of the binary complex (SelA-tRNASec) as 1SelAdecamer:10tRNASec. This stoichiometric ratio, different from the accepted 1SelAdecamer:5tRNASec, was fundamental for investigating the conformational changes occurring in the transition from a binary to a ternary complex. We used AFM to measure the low-resolution dimensions of SelAdecamer as 20.8 ± 0.5 nm in diameter and 3.96 ± 0.05 nm in height as the average for 58 single particles. After the binding of 10 tRNASec, the average dimensions of the binary complex were 22.0 ± 0.5 nm in diameter and 3.56 ± 0.05 nm in height from 86 single particles. The decrease in height is consistent with the size of the predicted SPS interaction surface (Fig. 6A) allowing the SelA-tRNASec-SPS interaction. The low-resolution dimensions of the SelA-tRNASec-SPS ternary complex were 27.0 ± 0.5 nm in diameter and 4.02 ± 0.05 nm in height, as determined from the average of 58 single particles.
Sec biosynthesis in E. coli requires 10 molecules of Ser-tRNASec covalently bound to homodecameric SelA to catalyze the conversion of Ser to Sec (5). The SelA-tRNASec binary complex can thus be interpreted as a reservoir of cellular tRNASec.
It was observed by H/DEx-MS presented here that the N-terminal region of SelA is required for SelA oligomerization, as it becomes hidden from the surface of homodecamers exposed to solvent. Therefore, homodecamerization, and consequently, the Ser-Sec conversion and selenoprotein biosynthesis, is dependent on the N-terminal region (or N terminus), as we observed by functional complementation with the N-terminally truncated E. coli SelA. Similar results from A. aeolicus SelA N-terminal mutants (27) and the non-functional dimeric M. jannaschii SelA (25), which do not interact with tRNASec, strengthen our findings.
A Schiff base is formed between the α-amino group of the Ser residue with the formyl group of PLP following SelA-tRNASec interaction, resulting in the synthesis of the intermediate aminoacrylyl-tRNASec upon dehydration of the amino acid residue (5). FTIR experiments show that SelA does not undergo a secondary structure change upon its interaction with tRNASec as was also observed in the crystallographic structure of A. aeolicus SelA-tRNASec complex (4), and fluorescence anisotropy spectroscopy with tRNASec mutants has shown that this interaction is dependent on the tRNASec acceptor arm, D-loop, and variable arm.
In addition to the D-loop arm (4) as the recognition point of tRNASec to SelA, we observed that the difference in the acceptor arm pairing number (8 to 7) is essential for tRNASec affinity to E. coli SelA. Selenium is transferred to the aminoacrylyl-tRNASec intermediate complex in the form of selenophosphate, a product of dimeric SPS selenide water dikinase catalytic activity (7, 10), to form Sec-tRNASec. The SPS dimerization interface is composed of the β-sheet domain of each monomer, a common structural characteristic of the PurM protein superfamily (7). This dimerization domain was confirmed by our H/DEx-MS experiments. In addition, consistent with the SPS crystallographic structures (PDB ID 3U0O) that were previously described, the glycine-rich N-terminal region of SPS was observed to be flexible in solution, showing high levels of deuterium exchange even after low deuterium exposure time. This flexibility allows the formation of the SPS active site on its “closed” form, upon ATP binding, releasing the catalysis product in its “open” form (7, 10).
Because one SelAdecamer molecule and 10 tRNASec molecules form a covalently bound binary complex (6), we analyzed the interaction of SPS with the SelA-tRNASec complex. Using fluorescein-labeled tRNASec, we observed an increase in fluorescence anisotropy following SPS isothermal titration, revealing a specific binding leading to the formation of the ternary complex, with a stoichiometric ratio of one SelA decamer covalently bound to 10 tRNASec molecules interacting with five SPS dimers. The SelA-tRNASec-SPS interaction dissociation constant of 610 ± 79 nm is consistent with the expected values for biomolecular transient interactions. Hill's plot (Fig. 1A) indicates a positive cooperativity, with n = 2.1 ± 0.4, for the formation of the ternary complex. Based on this observation, we propose that trapping the selenium compounds in the SelA-tRNASec-SPS complex would be an efficient mechanism to avoid the high cellular toxicity posed by free selenium. Additional experiments are necessary to verify this hypothesis.
The SelA-tRNASec-SPS interaction is dependent on stereochemical recognition, involving the structural accommodation of one molecule to the other. Remarkably, the height of the SelA-tRNASec complex is consistent with the SPS dimer size in the interaction region above 3 nm. However, FTIR analysis has indicated a modification in the α-helix segment of E. coli SPS only upon binding to the SelA-tRNASec complex. We suggest that this conformational change corresponds to the region from Glu159 to Val161 (α-helix 4) of E. coli SPS. Although no information about H/D exchange was obtained for this α-helix, analysis of the E. coli SPS crystal structure (7) indicates that it is in the middle of the linker between the aminoimidazole ribonucleotide synthetase (AIRS)-related and C-terminal domains, which is consistent with the SelA-tRNASec-SPS interaction. Regions close to the active sites of SPS were observed to be hidden from H/D exchange, and we conclude that these regions may be the interaction points that enable the formation of the ternary complex. Our analysis suggests that the formation of the ternary complex occurs via SPS opening its active sites to deliver selenophosphate to the active site of SelA. Furthermore, this represents a probable sequence of events in the synthesis of selenoproteins, with SelA binding to tRNASec prior to SPS (Fig. 7).
A flexible conformation of SPS is certainly required to facilitate its interaction with the SelA-tRNASec complex, as observed for the NifS-like-SPS interaction (7, 12). In fact, the glycine-rich N-terminal region of SPS is hidden from the solvent after SelA-tRNASec-SPS formation, as observed by H/DEx-MS, and SPS with an N-terminal truncation does not interact with SelA-tRNASec, as shown by fluorescence anisotropy spectroscopy experiments. In vivo studies of Δ11-SPS show that it does not complement SPS function in the SPS-deficient E. coli strain WL400. These data show that SelA-tRNASec-SPS complex formation is essential for selenoprotein biosynthesis in E. coli and that it follows a sequence of events, i.e. SelA interacts with tRNASec and undergoes tertiary structure rearrangements allowing the interaction with SPS, without changing its secondary structures (Fig. 7).
As previously noted by Yoshizawa and Böck (3), a second level of fidelity control in selenoprotein pathway, in addition to UGA stop-codon recognition, is the discrimination of Sec from its isosteric form Cys (3). Although widely studied, selenophosphate formation from selenide and ATP in a reaction catalyzed by SPS is not completely understood, and the structural basis for the substrate specificity has not yet been solved (3, 7). Our results provide new insights into the biosynthesis of selenoproteins, for the first time demonstrating the functional macromolecular assembly of the SelA-tRNASec-SPS. The significance of this finding centers on the ability of this complex to enable selenium delivery to Sec biosynthesis in the presence of tRNASec. We propose that once the ternary complex is formed, selenophosphate can be transferred from SPS to SelA active sites and to the tRNASec, concealing the toxic selenium compounds from the cytoplasm. Further investigation awaits to address this hypothesis.
I. R. S. and V. H. B. S. took part in the planning, data acquisition, treatment, and interpretation of all experiments and drafted this paper; L. R. M. contributed to the design and analysis of the fluorescence anisotropy experiments and produced all tRNA mutants; M. T. A.d. S. and L. M. F. contributed to the SPS N-terminally truncated constructions and their functional analysis; R. M. contributed to the SelA N-terminally truncated constructions and their functional analysis; D. S. and M. S. P. contributed to the design and analysis of H/DEx-MS experiments; M. L. C. contributed to FTIR experiments and data analysis; and O. H. T. served as group and project leader.
We thank Professor Luis Maurício Trambaioli Lima (UFRJ-Brazil) for important suggestions and discussions during the development of this project. We also kindly acknowledge the support received from the Department of Biochemistry of the University of Cambridge and Professor Dr. Tom Blundell, as well as the technical assistance provided in the N-terminal sequencing facilities. We also acknowledge the support from LNNano for AFM measurements and Vinicius Lago Pimentel for technical support in AFM measurements.
*This work was supported by Research Grant 2008/57910-0 from the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), CAPES and by the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq Grant 550514/2011-2). This work was also supported by FAPESP Fellowship 2010/04429-3 (to I. R. S.). The authors declare that they have no conflicts of interest with the contents of this article.
This article contains supplemental Tables 1A and 1B.
3The abbreviations used are: