Prebiotic Iron Originates the Peptidyl Transfer Origin

Abstract The ribosome is responsible for protein synthesis in all living organisms. It is best known to exist around 3.5–3.7 Ga whereat life on Earth inhabited anoxic environment with abundant soluble irons. The RNAs and proteins are the two biopolymers that constitute the ribosome. However, both proteins and RNAs require metal cations to fold and to function. There are four Mg-microcluster (Mg2+-μc) structures conserved in core of large subunit, and the 23S ribosomal RNA (rRNA) was shown to catalyze electron transfer in an anoxic environment in the presence of Fe2+. The Mg2+-μc features two idiosyncratic Mg2+ ions that are chelated and bridged by a common phosphate group and along with that, the adjacent residues of RNA backbone together forming ten-membered chelation ring(s). Here, we utilized four rRNA fragments of the large subunit 23S rRNA of Haloarcula marismortui, that includes the residues that form the four Mg2+-μc’s. These four rRNA fragments are shown competent to assemble with Mg2+. Our results show that when these rRNA fragments fold or assembly in the presence of Fe2+ under anoxic conditions, each Fe2+-microcluster can catalyze electron transfer. We propose that Fe2+-microclusters of the ribosome, which use Fe2+ as a cofactor to regulate electron transfer, are pivotal and primordial and may be an origin in evolution of the ribosome.


Introduction
The ribosome is the operating system (Bowman et al. 2015) for protein synthesis in all living organisms. It is best known to exist around 3.5-3.7 Ga whereat life on Earth inhabited anoxic environment with abundant soluble irons (Schopf et al. 2002;Anbar 2008;Ohtomo et al. 2014). The RNAs and proteins are the two biopolymers that constitute the ribosome. However, protein requires metal cations to fold and to function, so as RNA. There are four Mg-microcluster (Mg 2þ -lc) structures conserved in core of large subunit (LSU) ribosomes, where the three Mg 2þ -lc's were proposed to provide a framework for the peptidyl transfer center (PTC) ). The Mg 2þ -lc's ) are seen to 1) exhibit two idiosyncratic Mg 2þ ions, 2) possess a common phosphate group that chelates and bridges the two Mg 2þ ions, and 3) form ten-membered chelation ring(s) (supplementary fig. S1, Supplementary Material online). In addition, metals are essential constituents that confer varieties of catalytic functions on biological macromolecules (Riordan 1977). For example, the T4 DNA ligase requires Mg 2þ ions to catalyze formation of the phosphodiester bond of two dsDNA fragments (Lehman 1974). The Class Ia ribonucleotide reductase (RNR) contains dinuclear iron clusters that are required to catalyze the reduction of ribonucleotide to deoxyribonucleotide (Cotruvo and Stubbe 2011). The L1 ribozyme ligase performs phosphoryl transfer in the presence of Mg 2þ or Fe 2þ ions (Athavale, Petrov, et al. 2012).
Previous work has shown that the substitution of iron for magnesium confers the 23S ribosomal RNA (rRNA) with a new catalytic function: single-electron transfer (Hsiao, Chou, et al. 2013). Here, we report that each of the four Mg 2þ -lc's of the 23S rRNA when substituted with Fe 2þ demonstrates electron transfer activity ( fig. 1A). The structure-based comparisons of these Mg 2þ -lc's within the 23S rRNA and the dinuclear metal centers in the Class I RNRs revealed that the geometries of the two are closely related. For example, we superimpose one of the Mg 2þ -lc's that is closest to the PTC of the Haloarcula marismortui LSU (PDB entry 1JJ2) with the dinuclear Fe 2þ cluster, the catalytic center of the Escherichia coli RNR R2 (PDB entry 1PIY) ( fig. 2A). The superimposition shows that the three carboxylic groups of the amino acid residues (Glu238, Asp84, and Glu115) that are directly contacted by the Fe 2þ within the RNR are mimicked by the phosphate groups of the nucleotides (G877, G2623, and A2624) that are in direct contact with the Mg 2þ in the 23S rRNA ( fig. 2B and C). Likewise, the bridging phosphate of the residue G877 corresponds to the bridging carboxylate of the amino acid residue Glu238. In analogy to the dinuclear metal centers conferring the RNRs catalytic reduction activities, we hypothesize that one or more of the microcluster structures in the primordial 23S rRNA conferred it with the catalytic capability of electron transfer when iron (II) was abundantly available and could take the place of Mg 2þ in the anoxic environment of early earth.
To test our hypothesis, we used the high-resolution crystallographic data of the ribosomes (Klein et al. 2001;Selmer et al. 2006) to guide and dissect the 23S rRNA into fragments that form the direct contacts with the four Mg 2þ -lc's ( fig. 1B). Four 23S rRNA fragments were in vitro transcribed from DNA templates (supplementary table S1, Article ß The Author(s) 2019. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.

Open Access
Supplementary Material online). They are referred to here as the 23S rRNA subelements (rRNAsubs) which are rRNAsub_32, rRNAsub_48, rRNAsub_61, and rRNAsub_73. The number of each rRNAsub indicates the helix number of the 23S rRNA that comprises the terminal helical stem of the 23S rRNA fragment ( fig. 1B). These 23S rRNA fragments are different in lengths and proven an induction of folding with Mg 2þ (supplementary fig. S2, Supplementary Material online). They provided a test of the hypothesis that one or more of the four Mg 2þ -lc's responsible for the electron transfer activity of the 23S rRNA in association with Fe 2þ .

Results
Two of the four rRNA fragments (rRNAsub_48 and rRNAsub_73) dissected from the 23S rRNA were found to catalyze electron transfer in the presence of Fe 2þ (fig. 1A). The other two (rRNAsub_32 and rRNAsub_61) showed no enzymatic activity alone, however, the catalysis was observed upon their assembly ( fig. 1A). This outcome is consistent with the prediction based on the 23S rRNA structure that each Mg 2þ -lc , with Fe 2þ substituting for Mg 2þ , is capable of electron transfer ( fig. 1B).
Continuous variation analysis (Job 1928;Cantor and Schimmel 1984) was carried out on the complexation of In order to gain better understanding of the stoichiometries found for Fe 2þ and the individual rRNAsub fragments, continuous variation experiments were employed to examine the interaction of rRNAsub_32 with rRNAsub_61 and rRNAsub_32 with rRNAsub_73, and the stoichiometry of these complexes with Fe 2þ . Two of the Mg 2þ -lc's of the 23S rRNA are formed from these pairs of rRNAsub fragments The rRNAsub_32 and rRNAsub_61 alone by itself showed no activity. (B) The four rRNAsubs taken from the 23S rRNA of Haloarcula marismortui were highlighted on the 3D structure-based 23S rRNA secondary structure map (Petrov, Bernier, Gulen, et al. 2014). The helix number of the 23S rRNA used for nomenclature of the rRNAsubs is shown as number colored in cyan. The rRNAsub_48 is color green, the rRNAsub_73 red, and black for both the rRNAsub32 and rRNAsub_61. The four Mg 2þ -lc's within the rRNAsubs of the ribosome  were colored by the thick lines that represent the molecular interactions of the phosphate oxygen and the direct contact of the two Mg 2þ . The rRNAsub_48 can form one Mg 2þ -lc, shown by the yellow thick lines. The rRNAsub_61 and rRNAsub_32 together assemble one Mg 2þ -lc, shown by the blue thick lines. The rRNAsub_73 by itself has one Mg 2þ -lc (the yellow thick lines) and forms the second    ) that is closest to the PTC (red circle) of the Haloarcula marismortui 23S rRNA (wheat ribbon, PDB entry 1JJ2) was superimposed to the catalytic center of the Escherichia coli RNR R2 (cyan ribbon, PDB entry 1PIY). (B) A close-up view of the catalytic site of this superimposition shows the di-Fe 2þ geometry in RNR R2 is well conserved in Mg 2þ -lc of the 23S rRNA. The three carboxylic groups of the amino acid residues that are directly contacted by the Fe 2þ are mimicked by the phosphate groups of the nucleotides that are direct contact with the Mg 2þ . (C) Another view of this superimposition. Green spheres denote Fe 2þ , yellow: Mg 2þ , orange: phosphorous, red: oxygen, and gray: carbon. RNA is colored wheat, and protein colored cyan.

Discussion
Metal cations play critical roles in all biological macromolecules, including proteins, DNAs, and RNAs. The coordination chemistry and electrostatic forces of cations have conformational effects on folding and assembly for these biological macromolecules (Brown 1988;Williamson et al. 1989 ribosomes, which allows one to trace back life to the last universal common ancestor by several billions of years ago.

A Ribosomal Switch On and Off Electron Transfer
It is now known that substitution for Mg 2þ with Fe 2þ confers the 23S rRNA an electron transfer function (Hsiao, Chou, et al. 2013). As of the RNR, the dinuclear Fe or Mn clusters are essential entities for the catalytic reduction of ribonucleotides (Cotruvo and Stubbe 2011). When we compare the coordination geometries of the catalytic centers in the Class Ia RNR and one of the Mg 2þ -lc's in the 23S rRNA, surprisingly, they share an almost identical metal-ligand conformation, despite the two biological macromolecules are totally distinct ( fig. 2). Our rRNAsub fragments folding and assembly experiments with Fe 2þ reveal a Fe 2þ -lc that remarks the electron transfer activity. It is when Fe 2þ substitutes the two idiosyncratic Mg 2þ cations within the Mg 2þ -lc conferring the 23S rRNA a capability of electron transfer function. When rRNAsub_32 is alone in the presence of Fe 2þ , there was no catalytic activity in despite a 1:2 stoichiometric rRNA-Fe 2þ complexes were observed (figs. 1A, 3A, and 4C), suggesting that the Fe 2þ -lc was not formed. Notably, this catalytic activity can be summoned by the addition of rRNAsub_61, upon with their assembly (figs. 1A, 3E, and 4E). Thus, the Fe 2þ -lc of the 23S rRNA is a ribosomal switch that can turn on and off the electron transfer.

Fe 2þ -lc May Be an Origin in Evolution of the Ribosome
There are four Mg 2þ -lc's in LSU of the ribosome, conserved in sequences and structures over three domains of life. One of the Mg 2þ -lc's formed a portion of the peptidyl exit tunnel was suggested to be added late in ribosomal evolution . The other three are close to the PTC that provides a framework for ribosomal peptidyl transfer (PT) ). The PTC is the catalytic site of the ribosome that catalyzes the peptide formation. It consists of two terminal loops, the A and P loops that position the aminoacyl-tRNA and peptidyl-tRNA in an appropriate conformation to catalytically facilitate the formation of peptide bond in an entropic manner (Sievers et al. 2004). Structural analysis of the PTCs in 2D and 3D structures reveal a symmetric pocketlike assembly. This pocketlike assembly is a pseudo 2-fold symmetric entity, named protoribosome (Agmon 2009;Belousoff et al. 2010). Further structural investigations on models of the proto-ribosomes, we observed that there is one conserved Mg 2þ -lc that is intimate bridging the P-and A-regions of the 23S rRNA. The sequence contents of the proto-ribosomes are highly conserved in all organisms across the phylogenic tree. In addition, the conformations and geometries of the local structure formed by the Mg 2þ -lc and the 23S rRNA are very well conserved. As of the translation is believed to appear originally not to synthesize proteins, but allow small peptides bound to small RNAs such that to extend RNA's structural capabilities (Fahnestock et al. 1970;Noller 2004). When we examined the 2D structures of the proto-ribosome and our four rRNA fragments, the proto-ribosome is strikingly similar to the rRNAsub_73. Comparisons of ribosomal evolution on rRNAsub_73 indicate that it might be the most ancient part of the ribosome goes beyond last universal common ancestor (Bokov and   In models of ribosomal evolution, rRNA fragments were "elements" to be added as layers on top of the PT origin throughout its history (Bokov and Steinberg 2009;. Likewise, Fe 2þ and Mn 2þ are suggested to play a role in early ribosomal evolution. Bray et al. (2018) demonstrated that substitution for Mg 2þ with Fe 2þ and Mn 2þ as cofactors in ribosomes can mediate in vitro protein production. Their work supports the translation machinery matured when ancient Earth conditions were anoxic and abundant with soluble Fe 2þ and Mn 2þ (Schopf et al. 2002;Anbar 2008;Johnson et al. 2016). Our study here shows that rRNAsub_73 formed a Fe 2þ -lc in the presence of Fe 2þ is capable of catalyzing electron transfer. Thus, it is possible that Fe 2þ -lc complexes with a very ancient rRNA fragment such as rRNAsub_73 to form a PT origin and act as a nucleation for evolution of the translation just like a proto-ribosome proposed by Yonath and coworkers (Belousoff et al. 2010;Krupkin et al. 2011). However, the PT origin might have originated in electron transfer rather than PT. Interestingly, rRNAsub_73, rRNAsub_48, [rRNAsub_61 and rRNAsub_32], and [rRNAsub_73 and rRNAsub_32] failed of catalyzing electron transfer in the presence of Mn 2þ , and in the absence of Fe 2þ and O 2 (data not shown) suggests that electron transfer might be, in part, a selection pressure in Fe 2þ -lc that originated the PT origin. Additionally, we observed the initial rate of reaction for each Fe 2þ -lc increased (ca. 1-to 9-folds, see supplementary table S2, Supplementary Material online) when Fe 2þ and Mn 2þ were both present in the reaction, which implies that prebiotic Mn 2þ may play a role as an ancillary element that stabilizes the Fe 2þ -lc structures in early ribosomal evolution.
In this study, rRNAsub_32 is recruited and assembled with rRNAsub_73 and formed a Fe 2þ -lc ( fig. 1B), where we think that rRNAsub_61 fragment might be an evolutionary selection through the formation of Fe 2þ -lc with rRNasub_32 and was added after the assembly of [rRNAsub_73 and rRNAsub_32] complex. Thus, our results suggest that rRNAsub_73, rRNAsub_32, and rRNAsub_61 together formed a complex with three Fe 2þ -lc's embedded may be a core of LSU in early ribosomal evolution. These three rRNA fragments are proposed as the essential elements of core of LSU (Fox 2010;Hsiao, Lenz, et al. 2013;Petrov et al. 2015). We show that Fe 2þ -lc of the ribosome is pivotal and primordial and may be an origin in evolution of the ribosome.

Design of the 23S rRNAsub Genes
The rRNAsubs were designed to include residues that have direct first-shell contact of the di-Mg 2þ , guided by dissecting the high-resolution crystallographic data of the LSU of the H. marismortui ribosome (PDB entry 1JJ2). The dissected 23S rRNA fragments were categorized by its domain. This dissecting process left two fragments in each domain III, IV, and V ( fig. 1B). The two fragments were then joined by a tetraloop with a sequence (5 0 -GGGGTAACCC-3 0 ). Therefore, the four rRNAsub's genes were obtainable and commercially purchased from Omics Bio. In addition, each rRNAsub gene was added a T7 promoter at the 5 0 -teminus and the two restriction sites, HindIII at 5 0 -and BamHI at 3 0 -end. The gene sequence of the rRNAsubs was listed in the supplementary table S1, Supplementary Material online.

Synthesis and Purification of the rRNAsubs
The commercially purchased rRNAsub genes from Omics Bio were already cloned in pUC57 vector. We transformed the plasmid containing the rRNAsub gene into DH5a competent cells by heat shock technique and incubated at 37 C for 16 h prior to plating to Luria-Bertani (LB) agar containing 100 mg/ ml ampicillin. Colonies positive for insert were used to inoculate 3 ml of LB containing 100 mg/ml ampicillin. After 16 h of incubation at 37 C, plasmids were purified by QIAPrep Spin Miniprep Kit (Qiagen).
The in vitro transcripts. All rRNAsubs were prepared and synthesized by using MEGAscript transcription kit (Ambion) to avoid any in vivo contaminations for the experiments. To synthesize the in vitro rRNAsubs, 1 mg of linearized template DNA with a T7 promoter was mixed with NTPs, 10Â reaction buffer, enzyme mix and nuclease-free water to make the total reaction volume 20 ml, and incubated at 37 C. The optimum in vitro transcription time for each rRNAsub production is as follows: 7 h for rRNAsub_32, 6 h for rRNAsub_48, 6 h for rRNAsub_61, and 4 h for rRNAsub_73. After in vitro transcription reaction, the transcripts were precipitated by adding equal volume of a fresh prepared 5-M ammonium acetate solution, and placed it on ice for 15 min, followed by centrifuged at speed 10,000 Â g, at 4 C for 15 min. The RNA pallet was then washed twice by using 80% ethanol, lyophilized, and resuspended in 50-ml nuclease-free water.

The Single-Electron Transfer Assay
The in vitro transcripts used in the experiments were first mixed with the cation exchange resin (Chelex 100 Resin, Bio-RAD) and incubated at 60 C for 1 min. The RNAs were then recovered by using a 0.22-lm Ultrafree-MC Centricon (Millipore), centrifuged at 10,000 Â g to remove resin. The resulting RNA solution was then lyophilized for later use. The single-electron transfer assay was described detail in Hsiao, Chou, et al. (2013). The resin-treated rRNAsubs were resuspend in 20-mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)-Tris buffer, pH 7, and mixed with 500-lM 3,3 0 ,5,5 0 -tetramethylbenzidine in Dimethyl sulfoxide (DMSO) and 50-mM H 2 O 2 solution. The mixtures were sealed in a 1.5-ml centrifuged tube and deoxygenated by bubbling argon for 5 min. After deoxygenated, a 20-mM FeSO 4 solution was added to the RNA containing solution to the final Fe 2þ concentration of 100 lM and incubated at room temperature for 5 min followed by the addition of 200-mM deoxygenated H 2 SO 4 solution to pH 6.1. The reaction was then monitored the UV-Vis absorption at 652 nm by using Ocean optics USB2000þ spectrophotometer. The data readings were recorded by using the commercial software SpectraSuite ver. 1.0 (Ocean Optics). The rRNAsub final concentrations used in each test were 0.1 lM for every rRNAsub and their assemblies.

The Iron Dependency Analysis
The electron transfer experiments were described above and performed by varying the Fe 2þ concentrations at 8-200 lM. The initial rate of each electron transfer at different Fe 2þ concentration was determined by fitting linear regression to data. The iron dependency is then plotted natural logarithm of the initial rate of electron transfer versus natural logarithm of Fe 2þ concentration (supplementary fig. S5, Supplementary Material online).

The Continuous Variation Assay
A deionized, nuclease-free water was first deoxygenated by bubbling argon. The deoxygenated water was then transferred into the anaerobic chamber (Coy). All chemical powders used in the experiments were deoxygenated prior to transferring into the chamber. The reaction buffer 100-mM Tris-Cl, pH 8 and the 10 mM FeSO4 solution were freshly prepared inside the chamber. The 10-mM MgCl 2 solution was prepared outside the chamber. All the RNAs used in the experiments were mixed with the cation exchange resin (Chelex 100 Resin, Bio-RAD) and incubated at 60 C for 1 min. The RNAs were then recovered by using a 0.22lm Ultrafree-MC Centricon (Millipore), lyophilized, and then transferred into anaerobic chamber for later use. The continuous variation experiments with Mg 2þ were performed in an aerobic environment.
The continuous variation analysis on the rRNAsubs with Fe 2+ /Mg 2+ The rRNAsub and Fe 2þ /Mg 2þ were mixed and held constant at 0.4 lM, whereas the mole fractions of the two components were varied from 0.0 to 1.0. For the rRNAsub assemblies (e.g., rRNAsub_73 þ rRNAsub_32) with Fe 2þ /Mg 2þ , the two components were mixed and held constant at 0.8 lM. The mixtures were prepared in 50 mM Tris-Cl buffer, pH 8, incubated at 25 C for 15 min and then stored at 4 C. The UV-Vis absorption of the RNA-Fe 2þ mixtures were measured at 260 nm at 25 C using a Nanodrop (Thermo) inside the anaerobic chamber, and the RNA-Mg 2þ mixtures were measured outside the chamber.
Prebiotic Iron Originates the Peptidyl Transfer Origin . doi:10.1093/molbev/msz034 The continuous variation analysis on the rRNAsubs in the presence of Fe 2+ /Mg 2+ The study of the two rRNAsubs were mixed and held constant at 0.2 lM, whereas the mole fractions of the two components were varied from 0.0 to 1.0. The RNA mixtures were prepared in 50-mM Tris-Cl buffer, pH 8 and 32-lM Fe 2þ solution, incubated at 25 C for 20 min and then stored at 4 C. The RNA-Mg 2þ mixtures were prepared in 10 mM Mg 2þ and 100 mM Na þ , incubated at 25 C for 20 min and then stored at 4 C. The UV-Vis absorption of the rRNAsubs interactions in the presence of Fe 2þ were measured at 260 nm at 25 C using a Nanodrop inside the anaerobic chamber, and the mixture samples were measured in an aerobic environment for the samples in the presence of Mg 2þ .

The continuous variation experiment data analysis
The UV-Vis absorption of each mixture sample was measured repetitively five times. Every experiment was performed twice. The measured outliers of each data set were excluded using the cutoff of 1.5 standard deviation (1.5r). After excluded the outliers, the experimental data were then averaged and expressed in OD 260 per nucleotide. The discontinuity at equivalent mole fractions was then determined by solving the intersection of the two linear regression lines.

Michaelis-Menten Kinetics
The resin-treated rRNAsubs were resuspended in 20-mM HEPES-Tris buffer, pH 7, and mixed with 500-lM 3,3 0 ,5,5 0tetramethylbenzidine (in DMSO) and different concentrations of H 2 O 2 solution at 15, 16.5, 18, 21.5, 25, 50, and 100 lM. The mixture samples were deoxygenated by bubbling Argon for 5 min followed by the addition of 20-mM Fe 2 SO 4 solution to the final Fe 2þ concentration that is optimized for the kinetics experiment of each RNA: 100 lM Fe 2þ for rRNAsub_48, 64 lM Fe 2þ for rRNAsub_73, 32 lM Fe 2þ for [rRNAsub_61 þ rRNAsub_32], and 32 lM Fe 2þ for [rRNAsub_73 þ rRNAsub_32] and incubated at room temperature for 5 min. The UV-Vis absorption of each electron transfer reaction was measured at 652 nm after the addition of 200-mM deoxygenated H 2 SO 4 solution to pH 6.1. The enzymatic kinetic parameters performed by the rRNAsubs and its assemblies can then be obtained by nonlinear regression curve fitting to the experimental data using the Michaelis-Menten model.

Michaelis-Menten data analysis
The V max and K m obtained by nonlinear regression fit to the kinetics experimental data were performed by using JMP Pro 13.1 (SAS Institute Inc.) with the Michaelis-Menten formalism.
The Gel Mobility of rRNAsub Folding with Mg 2þ The RNA samples were treated with cation exchange resin (Chelex 100 Resin, Bio-RAD) described as above. The study on the rRNAsub folding with varied Mg 2þ concentration: 0, 25, 50, 100, 250, 500, 1,000, 5,000 lM, 200 ng of the resin-treated RNAs were mixed with Mg 2þ solution, heated at 90 C for 30 s and annealed by cooling to 25 C at a linear rate of