Search tips
Search criteria 


Logo of narLink to Publisher's site
Nucleic Acids Res. 2017 April 20; 45(7): 3634–3642.
Published online 2017 March 15. doi:  10.1093/nar/gkx171
PMCID: PMC5397171

Iron mediates catalysis of nucleic acid processing enzymes: support for Fe(II) as a cofactor before the great oxidation event


Life originated in an anoxic, Fe2+-rich environment. We hypothesize that on early Earth, Fe2+ was a ubiquitous cofactor for nucleic acids, with roles in RNA folding and catalysis as well as in processing of nucleic acids by protein enzymes. In this model, Mg2+ replaced Fe2+ as the primary cofactor for nucleic acids in parallel with known metal substitutions of metalloproteins, driven by the Great Oxidation Event. To test predictions of this model, we assay the ability of nucleic acid processing enzymes, including a DNA polymerase, an RNA polymerase and a DNA ligase, to use Fe2+ in place of Mg2+ as a cofactor during catalysis. Results show that Fe2+ can indeed substitute for Mg2+ in catalytic function of these enzymes. Additionally, we use calculations to unravel differences in energetics, structures and reactivities of relevant Mg2+ and Fe2+ complexes. Computation explains why Fe2+ can be a more potent cofactor than Mg2+ in a variety of folding and catalytic functions. We propose that the rise of O2 on Earth drove a Fe2+ to Mg2+ substitution in proteins and nucleic acids, a hypothesis consistent with a general model in which some modern biochemical systems retain latent abilities to revert to primordial Fe2+-based states when exposed to pre-GOE conditions.


Iron was abundant, benign and soluble when life originated on the ancient earth (15). The geological record indicates that for ~2 billion years the oceans contained vast quantities of soluble Fe2+, with concentrations on the order of 10−4 M. The reducing conditions of the ancient Earth favored Fe2+ over Fe3+ and mitigated destructive iron-mediated processes such as Fenton chemistry (6,7). During the first half of Earth's history iron became broadly involved in protein-mediated biochemistry (8,9). The wide distribution of iron in extant biological systems (10), despite its harmful effects and low solubility on the surface of the extant Earth, highlights this element's catalytic utility and deep evolutionary history.

Approximately 2 billion years ago, O2 began accumulating in the atmosphere, triggering global shifts in biochemistry and microbiology. This Great Oxidation Event (GOE) brought the modern condition of iron scarcity (10−9 M in the oceans) and iron-mediated oxidative damage to biological systems (11). The GOE drove substitution of copper, zinc, manganese and other metals for iron in metabolic enzymes (8,1218) as well as tight regulation of cellular distributions and concentrations (10).

We hypothesize that on the ancient earth, Fe2+ was a primary divalent cofactor in protein-based nucleic acid processing enzymes as well as in ribozymes (23,24). In this model, the GOE drove biosphere-wide Fe2+→Mg2+ substitutions in both enzymes and ribozymes. The abilities of these two metals to substitute for each other are consistent with similarities in their coordination chemistries (Table (Table1).1). An earth-wide Fe2+→Mg2+ substitution in nucleic acid processing enzymes and in ribozymes has analogy to well-established substitutions of Cu2+, Zn2+, Mn2+ for Fe2+ in metabolic enzymes (8,1218). In our Fe2+→Mg2+ model, the early emergence of polymerases, ligases, nucleases, repair enzymes and ribozymes was facilitated by interactions with Fe2+.

Table 1.
Mg2+ versus Fe2+

Magnesium seems to be essential for function of extant DNA polymerases, RNA polymerases and DNA ligases. A consensus of data supports a two divalent metal cation mechanism of phosphoryl transfer by these enzymes (2530). Magnesium is thought to interact with phosphate groups and to accept and donate protons during catalysis. In a generally accepted transition state in polymerases, a hexacoordinated Mg2+ stabilizes the triphosphate of the newly base paired (d)NTP, while a partially hydrated Mg2+ activates the 3΄-OH of the primer for nucleophilic attack on the 5΄-phosphate of the (d)NTP, thereby completing the phosphodiester bond. A similar two metal transition state has been proposed for the last step of phosphodiester bond formation by DNA ligase, whereby one metal stabilizes the phosphates of the temporarily adenylated DNA substrate and another activates the hydroxyl group. Probable metal ligands of polymerases and DNA ligase active sites include protein carboxylates, water molecules, and the non-bridging phosphate oxygens and hydroxyls of nucleic acids. Recent work proposes that a third divalent metal ion may also be essential for catalysis (31).

We use experiments to characterize Fe2+-mediated biochemistry in simulated ancient Earth conditions. We have recreated anoxic, Fe2+-rich conditions (pre-GOE conditions) in the laboratory. We test predictions of the Fe2+→Mg2+ model by removing Mg2+ from three nucleic acid processing proteins, and replacing it with Fe2+ in the absence of O2. We test abilities of a DNA polymerase, an RNA polymerase and a DNA ligase to function using Fe2+ as a cofactor in place of Mg2+. Specifically, we substituted Fe2+ for Mg2+ in a thermostable DNA polymerase (Deep Vent exo-) (32), in T7 RNA polymerase (33) and in T4 DNA ligase (34). The results show that Fe2+ can substitute for Mg2+ in initiation and elongation by the DNA polymerase, in RNA synthesis from a DNA template by the RNA polymerase, and in the joining of DNA oligonucleotides by the ligase.

In addition, we use calculations to reveal differences in energetics, structures and reactivities of Mg2+ and Fe2+ complexes. We observe that conformations and geometries of hexa aquo or first shell phosphodiester-complexes are conserved when Fe2+ is replaced by Mg2+. Compared to Mg2+, Fe2+ more effectively withdraws electrons from first shell ligands, causing increases in the electrophilicity of phosphorus atoms of first shell phosphodiester-complexes. By the same mechanism, compared to Mg2+, Fe2+ increases the acidity of first shell water molecules. Therefore, Fe2+ is expected to be a more effective cofactor than Mg2+. The combined experimental and computational results are consistent with the Fe2+→Mg2+ model and suggest that some modern nucleic acid processing enzymes retain latent abilities to revert to primordial Fe2+-based states when exposed to pre-GOE conditions.


Polymerase chain reactions

Polymerase chain reactions (PCR) was performed in a Coy anaerobic chamber in 20 mM Tris–HCl (pH 8.8), 10 mM KCl, 10 mM (NH4)2SO4, 0.1% Triton® X-100, 100 μg/ml nuclease-free bovine serum albumin, 1 mM dNTPs, 500 nM Cy3-labeled reverse primer, 500 nM forward primer, 0.5 nM template, 1 U DeepVent® (exo-) DNA polymerase (New England Biolabs). Reaction solutions contained either 2 mM MgSO4, 2 mM FeCl2, 2 mM MnCl2 or in negative controls, no divalent cations. Master mix solutions were prepared with all components except the polymerase and divalent cation, and lyophilized to dryness. We have previously demonstrated that our protocols remove oxygen from reaction mixtures (23,24). Using these methods there is no observable Fenton degradation of RNA in the presence of Fe2+ and Fe2+ substitutes for Mg2+ in RNA folding and ribozyme catalysis.

The DNA polymerase was added to the dry master mix, which was then transferred into the anaerobic chamber. Divalent cation solutions and nuclease-free H2O, pre-equilibrated in the anoxic atmosphere, were added to produce four PCR mixtures, one with 2 mM Mg2+, one with 2 mM Fe2+, one with 2 mM Mn2+ and one lacking divalent cations. Ten microliter of each PCR solution were aliquoted to eight PCR tubes for a total of 32 reactions. One of these was kept at room temperature and served as a negative PCR control (Reaction 0). Tubes were heated to 95°C for 2 min in a thermal cycler and cycled through 2, 4, 8, 12, 16, 20 or 24 PCR cycles. A cycle consisted of (i) denaturation at 95°C for 30 s, (ii) annealing at 52°C for 30 s and (iii) extension at 72°C for 30 s. Divalent cations were removed from the reaction mixtures by incubation with Bio-Rad Chelex 100 Resin. The resin was removed with 0.22 μm centrifugal filters, rendering the samples stable in the presence O2. The metal-free solutions were taken out of the anaerobic chamber and diluted 1:10 with nuclease-free water. One microliter of each diluted PCR solution was mixed with 9 μl of 10% glycerol and loaded onto a 12% 19:1 polyacrylamide gel buffered in Tris–Borate-Ethylenediaminetetraacetic acid (EDTA), pH 8.4. The reverse primer was also diluted in 10% glycerol and loaded onto the same gel. Each of the polyacrylamide gels were run at 100 V for 70 min at ambient temperature. Gels were imaged using a General Electric Typhoon Trio+ Imager. PCR template and primer sequences are provided in the Supplementary Information.

In vitro transcription

For in vitro transcription reactions, DNA template was generated by digestion of intact plasmid pUC19 containing the Thermus thermophilus Domain III rRNA gene (35) with HindIII, overnight, in NEB CutSmart buffer. The enzyme was heat inactivated at 80°C for 20 min and linearized DNA purified with an IBI Scientific Gel/PCR DNA Fragment Extraction kit.

The in vitro transcription reactions were performed in a Coy anaerobic chamber. Master mix solutions were prepared with all components except the T7 RNA polymerase and divalent cation, lyophilized to dryness, then introduced to the anaerobic chamber. In the chamber, the master mix was re-suspended in water that had been pre-equilibrated to the chamber atmosphere. Final reaction conditions were 1× RNA polymerase reaction buffer (40 mM Tris–HCl, pH 7.9, 2 mM spermidine, 1 mM dithiothreitol), 0.375 mM each nucleoside triphosphate, 5.4 μM linearized DNA template and 5–6 U NEB T7 RNA Polymerase (cat no. M0251S) per microliter of reaction volume, in addition to variable concentrations of MgCl2, FeCl2 or water (control reactions lacked divalent cations). Reactions were incubated at 37°C for 1 h then quenched with excess EDTA. Quenched reactions were removed from the anoxic chamber, mixed with at least an equal volume of ambion gel loading buffer II (95% formamide, 18 mM EDTA and 0.025% each of sodium dodecylsulphate (SDS), xylene cyanol and bromophenol blue), heated to 95°C for 5 min and loaded onto a 5% denaturing polyacrylamide gel alongside a single stranded RNA marker. Gels were run at 120 V for 1 h 20 min, stained with EtBr and imaged on a GE/Amersham Imager 600.

The supplementary material (Supplementary Figure S1) demonstrates that the intensity of the DNA template band decreases when the transcription reaction is successful and that the intensity of RNA transcription product band increases when the completed reaction mixture is treated with DNase [TURBO DNase (Ambion)]. These changes in band intensities arise because complexes of the DNA template and RNA transcript can be so stable that they survive the denaturing conditions and shift the DNA template out of the primary band on the gel. However, these complexes do not survive DNase treatment and the intensity of the transcript band is increased by DNase treatment. These phenomena are illustrated by a series of controls shown in Supplementary Figure S1 and in Figure Figure22 of the manuscript (note the three lanes with the least amount transcript appear to have the most template). The topmost band of the gels (Figure (Figure2,2, Supplementary Figures S1 and S2) contains aggregates and long nucleic acids that do not enter the gel.

Figure 2.
Fe2+ can replace Mg2+ as a cofactor in transcription by T7 RNA polymerase. Full length RNA transcript is observed with either Mg2+ or Fe2+. No product is observed in the no-divalent negative control. The far left lane contains ssRNA size markers. The ...

Ligase reactions

Ligase reactions were performed in the Coy anaerobic chamber in 50 mM Tris–HCl (pH 7.4), 1 mM adenosine triphosphate and 10 mM dithiothreitol. Two semi-complementary DNA 12-mers, one with 5΄-phosphate, were used in the ligation reaction. The sliding half-complementary oligonucleotide sequences were 5΄-pGAATGGGTAGAC-3΄ and 5΄-CCATTCGTCTAC-3΄. Seven μM of each oligonucleotide was incubated with 200 U of T4 DNA ligase (New England Biolabs) and 10 mM MgCl2 or 10 mM FeCl2. Negative control experiments lacked divalent cations.

DNA oligonucleotides and a ligase master mix containing all reaction components except divalent cations and ligase were lyophilized to dryness, transferred to the anaerobic chamber, equilibrated with the anoxic atmosphere and dissolved in degassed nuclease-free water. T4 DNA ligase was brought into the chamber and equilibrated with the anoxic atmosphere. Once equilibrated, the ligase was then added to the master mix. The solution was divided into three aliquots. Mg2+, Fe2+ or nuclease-free water was added to each of the tubes. The reactions were incubated at room temperature for 1 h. Reactions were terminated by incubation with Bio-Rad Chelex 100 Resin. The resin was removed with 0.22 μm centrifugal filters transferred out of the chamber for analysis. Ten microliter of the filtrate was added to 10 μl of loading dye (47.5% formamide, 0.01% SDS, 0.01% bromophenol blue, 0.005% Xylene Cyanol, 0.5 mM EDTA), followed by heating to 90°C for 5 min and quick cooling on ice for 10 min. Samples were loaded onto a 6% denaturing polyacrylamide gel buffered in tris–borate EDTA, pH 8.4. Gels were run at 100 V for 30 min. at ambient temperature, stained with SYBR Green I and imaged on a General Electric Typhoon Trio+ Imager.


Hexa aquo complexes [M2+(H2O)6] and M2+-RNA clamps (where M = Fe2+ or Mg2+) were optimized at the unrestricted B3LYP/6–31G(d,p) level of theory. For Fe2+ the spin was two and multiplicity was five. For Mg2+, the spin was zero and the multiplicity was one. Single point energies for these complexes were further obtained at the (U)B3LYP/6–311++G(d,p) level of theory using SCF options DIIS, NOVARACC, VTL and MaxCyc =  1000.

The coordinates of the Mg2+-RNA clamp were extracted from the x-ray structure of the Haloarcula marismortui large ribosomal subunit (PDB entry: 1JJ2) (36) as previously described (37). The free 5΄ and 3΄ termini of the phosphate groups were capped with methyl groups in lieu of the remainder of the RNA polymer, and hydrogen atoms were added where appropriate. An Fe2+-RNA clamp was constructed from the Mg2+ clamp by converting Mg2+ to Fe2+ as described (23). The binding of an Mg2+ or Fe2+ ion to an RNA fragment is described by the following reaction:

equation M1

where, M2+  =  Mg2+ or Fe2+.

The reactants and products of this reaction were fully optimized using the density functional theory with the hybrid B3LYP functional, which combines the generalized gradient approximation (GGA) exchange three-parameter hybrid functional of Becke (38) and the correlation functional of Lee-Yang-Parr (39), with the 6–311++G(d,p) basis set as implemented in Gaussian 09 (40).

Natural Bond Order (NBO) (41) and natural energy decomposition analysis (NEDA) (41,42) were performed on the optimized complexes at the (U)B3LYP/6–31G(d,p) level of theory using the GAMESS package (43) and the NBO 5.0 routine. For NEDA calculations, metal-phosphate clamps were treated as products and the free sugar-phosphate backbones and hexa aquo metals were reactants.


We have investigated the ability of a DNA polymerase, an RNA polymerase and a DNA ligase to function using Fe2+ instead of Mg2+ as a cofactor under simulated ancient earth conditions. All reactions were performed in a Coy anaerobic chamber in an atmosphere of 95% argon and 5% hydrogen. Each reaction was run multiple times to ensure reproducibility.

DNA polymerase

Fe2+ substitutes for Mg2+ as a cofactor for DNA polymerase. We have investigated polymerization using Deep Vent (exo-) DNA polymerase. We used PCR in the presence of Mg2+ or Fe2+ or Mn2+ or in the absence of divalent cations to amplify a 72 nucleotide DNA fragment. Mg2+, Fe2+ and Mn2+ each facilitate formation of polymerization product. We determined the amount of reactants (primers) and product in the reaction mixture at every other PCR cycle (Figure (Figure1).1). For divalent cations Mg2+ or Fe2+ or Mn2+, reaction product is first visible on the gel at cycle eight. The primer is fully consumed in all reactions by cycle twenty. It can be seen that these three divalent cations produce identical yields under the conditions of the reaction within the error of the experiment. If the yields were not similar for each divalent cation at each cycle, the cycle numbers at which primer disappears and at which product appears would vary between the different divalent cations. The absence of product in the reaction lacking divalent cations demonstrates that Mg2+ extraction methods are efficient and that the product observed in the Fe2+ reaction is not attributable to contaminating Mg2+.

Figure 1.
Fe2+ or Mn2+ can replace Mg2+ as a cofactor for Deep Vent (exo-) DNA polymerase. Reaction mixtures with Fe2+ or Mg2+ or Mn2+ or no divalent cation were analyzed for amount of reactants and products at every other PCR cycle for 24 cycles. Reactant consumption ...

RNA polymerase

Fe2+ substitutes for Mg2+ as a cofactor for T7 RNA polymerase. Digested plasmid encoding a template for a 376 nucleotide fragment of Domain III rRNA (44) was used with varying concentrations of Mg2+ or Fe2+. Full length RNA product is observed at all concentrations of Mg2+ investigated here, and at concentrations of Fe2+ up to 6 mM (Figure (Figure2).2). No RNA product is observed in control reactions without added divalent metals, confirming that the RNA products in the Fe2+ reactions do not result from Mg2+ contamination. The divalent cation concentration that gives a maximum RNA yield is less for Fe2+ than for Mg2+. At 0.75 mM of either divalent cation, the product yield is greater for Fe2+ than for Mg2+. Therefore, at low divalent metal concentrations T7 RNA polymerase appears more active in the presence of Fe2+ than in Mg2+. These experiments were independently replicated, giving consistent results in each replica. The identities of other bands on the gels are discussed in ‘Materials and Methods’ section and Supplementary Figure S1.

The Mg2+ optimum for T7 DNA polymerase is around 10 mM, depending on the presence of other cations (45). The concentration optimum is an order of magnitude lower for Fe2+ than for Mg2+. Under the conditions of these reactions, the T7 RNA polymerase yield decreases with increasing Fe2+. At 60 mM Fe2+ the yield drops to zero (Supplementary Figure S2).

DNA ligase

We have investigated a ligation reaction using sliding-half complementary DNA oligonucleotides ('Materials and Methods' section). The oligonucleotides were ligated using T4 DNA ligase in the presence of either Mg2+or Fe2+ and in the absence of divalent cations. The results demonstrate that T4 DNA ligase is functional with either Mg2+ or Fe2+. The Fe2+ reaction forms less product and shorter fragments than the Mg2+ reaction. We observe ligation products up to 14 substrates in length with T4 ligase in the presence of Mg2+, and ligation products seven substrates in length in the presence of Fe2+ (Figure (Figure3).3). On the gel, each successive band of increasing DNA length corresponds to an increase in product length by the ligation of one additional oligonucleotide. The absence of significant product in the control reaction confirms that the product observed in the Fe2+ reaction is not be attributed to contaminating Mg2+.

Figure 3.
Fe2+ can replace Mg2+ as a cofactor for T4 DNA ligase. Ligation products of sliding-half complementary oligonucleotides are observed with both Mg2+ and Fe2+. No product is observed in the divalent-minus negative control. See ‘Materials and Methods’ ...


The geometries of RNA2−–Fe2+(H2O)4 and RNA2−–Mg2+(H2O)4 clamps (37) and of hexa aquo Fe2+ and hexa aquo Mg2+ were optimized. Each clamp consists of a (deoxy)ribose with 5΄ and 3΄ phosphates and four water molecules. M2+ is six-coordinate in all complexes.

The conformations of RNA clamps are nearly identical with Fe2+ and Mg2+, with very similar metal-ligand distances (Supplementary Table S1). The average metal to oxygen (water) distance is 2.16 Å for Mg2+ and 2.15 for Fe2+. The interaction energies are also similar between Fe2+ and Mg2+, although the Fe2+ clamp is slightly more stable than the Mg2+ clamp (Table (Table2)2) due to charge transfer to the vacant d-orbitals of Fe2+ (Figure (Figure4,4, Supplementary Tables S2 and 3). RNA and DNA clamp more tightly to either Fe2+ or Mg2+ than to Na+ or Ca2+ (37). RNA clamps are more stable than DNA clamps.

Figure 4.
(A) A three-dimensional representation of a common M2+ complex in RNA in which the M2+ ion is coordinated directly by two phosphate oxygens in an RNA-M2+ clamp. The M2+ is six coordinate; the four first shell water molecules are omitted for clarity. ( ...
Table 2.
Interaction energies in RNA/DNA clamps of Fe2+ and Mg2+

Hexa aquo complexes

The calculations show why hexa aquo Fe2+ is a stronger acid than hexa aquo Mg2+. Fe2+ confers greater positive charge on its first shell waters than Mg2+. In comparison with Mg2+, Fe2+ withdraws 0.142 more electrons from its first shell water molecules (Supplementary Tables S4 and S5). The net charge on Fe2+ in a hexa aquo complex is +1.631 while the net charge on Mg2+ in a hexa aquo complex is +1.773. The calculations show that 0.369 electrons are transferred from the six first shell water molecules to Fe2+, while only 0.227 electrons are transferred to Mg2+. The net number of electrons transferred from the average first shell water of hexa aquo Fe2+ is 0.060 electrons per water molecule. The net number of electrons transferred from the average first shell water of hexa aquo Mg2+ is 0.037 electrons per water molecule.


Here we recapitulate the reductive potential of the Archean atmosphere (4648). Kasting's high CO2 model of the Archean atmosphere (46) has been re-evaluated (48); geological data is considered to be incompatible with this model (47,49). Although the specific chemical composition of the Archean atmosphere remains unresolved, it is accepted that it was reductive and that the biosphere was iron-rich and was lacking O2. Life originated and first proliferated in a reductive iron-rich environment, which persisted until around 2 billion years ago, when the GOE began depleting iron from the biosphere (14,23) and fostering Fe2+/O2 mediated cellular damage (7).

The environment of the ancient Earth is consistent with a Fe2+→Mg2+ model in which Fe2+ was an important cofactor for both nucleic acids and proteins during early evolution. Extant protein enzymes process nucleic acids using Mg2+ as a cofactor but might have used Fe2+ on the early earth. It is possible that extant biopolymers retain intrinsic adaptation to Fe2+. Indeed, previous experimental substitution of Fe2+ for Mg2+ in association with RNA demonstrated that Fe2+ can facilitate RNA folding and expand RNA catalytic breadth (23,24).

To determine if the Fe2+→Mg2+ model is plausible, and if ancient nucleic acid processing enzymes might have used Fe2+ instead of Mg2+ as a primary cofactor, here we substituted Fe2+ for Mg2+ in a DNA polymerase, an RNA polymerase and a DNA ligase. Polynucleotide polymerases are ubiquitous enzymes that perform some of the most critical and universal enzymatic activities in the biological world. Polymerases synthesize polynucleotides from (d)NTPs by covalently joining nucleotides as directed by a template. Polymerases use di-metal centers with metals coordinated by the protein and during catalysis, by phosphate groups of the substrate (25). The di-metal centers in DNA and RNA polymerases are thought to (i) stabilize a pentacovalent transition state, (ii) facilitate the leaving of pyrophosphate and (iii) lower the pKa of the 3΄-hydroxyl of the terminus. Mg2+ is the thought to be the preferred divalent ion in vivo for polymerases. The results here demonstrate that Fe2+, in a reductive environment, can substitute for Mg2+ in function of both DNA and RNA polymerases. It was shown previously that Mn2+ (50,51), or with reduced functionality, Ca2+ (52), can substitute for Mg2+ in some polymerases in vitro.

DNA ligase, another ancient protein enzyme, is required for DNA replication and repair (34,53). DNA ligase catalyzes the joining of terminal 5΄-phosphoryl and 3΄-hydroxyl groups of DNA fragments. DNA ligase joins Okazaki fragments produced by lagging strand DNA synthesis and seals nicks after DNA excision repair. Ligation uses three sequential nucleotidyl transfers, each of which requires a divalent metal cofactor. Mg2+ is thought to be the preferred ion in vivo. A ferric iron-containing ligase with an exceptionally low pH optimum has been isolated from an acidophilic ferrous iron-oxidizing archaeon (54). Structural and possibly catalytic roles for the ferric iron are possible. Unlike the systems we investigate here, the ferric ligase is not active in the presence of Mg2+.

Like DNA and RNA polymerases, DNA ligases are believed to contain di-metal centers within their active sites. Metals are coordinated by the protein and, during catalysis, by substrate phosphate groups (34,53). The di-metal center activates hydroxyl groups for nucleophilic attack and stabilizes leaving groups. The results here indicate that Fe2+ can substitute for Mg2+ in a ligase di-metal center. In the presence of Fe2+, the ligase is functional but may be less active than in the presence of Mg2+.

Mechanisms of phosphoryl transfer and geometries of di-metal centers are highly conserved and thought to be evolutionarily ancient (25). Formation of di-metal centers appears to be required for activity of the enzymes investigated here. Therefore, it appears functional di-iron centers form in DNA and RNA polymerases and in DNA ligases under our experimental conditions. The facility of substitution of Fe2+ for Mg2+ demonstrated here suggests that di-metal centers, including di-iron centers in enzymes such as ribonucleotide reductase (15) and di-magnesium centers in polymerases (25,55) and in large RNAs (56), may be related by common ancestry or by origins in a common chemical environment. Observed activity of polynucleotide polymerases and a DNA ligase in the presence of Fe2+ is consistent with similar coordination geometries and chemistries between Mg2+ and Fe2+ (Table (Table1).1). However, decreased activity of these enzymes in the presence of Fe+2 instead of Mg2+ might be expected after nearly 2 billion years of evolution that would have optimized the use of Mg2+ as a cofactor, rather than Fe2+.

Although Mg2+ and Fe2+ are characterized by similar coordination geometries and chemistries (Table (Table1),1), and can substitute for each other in di-metal centers, they are distinguished by important differences. We have observed that Fe2+ can be a more potent cofactor than Mg2+ in RNA folding. RNA folds at lower concentrations of Fe2+ than Mg2+, and that at least a subset of ribozymes are more active in Fe2+ than in Mg2+ (23). In addition, we observe here that at low concentrations of divalent cations, T7 RNA polymerase is more active in the presence of Fe2+ than Mg2+. To attempt to understand the origins of these differences, we have used computation to determine the effects of substitution of Fe2+ for Mg2+ in hexa aquo complexes and in complexes with first shell phosphodiester ligands.

We have modeled phosphate complexes of Fe2+ and Mg2+. The results allow us to understand the roles of divalent cations in RNA folding and during important catalytic steps of polymerization and ligation reactions. The forces, energetics and electronic perturbations within phosphate complexes of Fe2+ and Mg2+complexes were characterized by density functional methods (5763). Components of the interaction energy, such as charge transfer, polarization and exchange were investigated with NBO and NEDA (41,42). The computations reveal subtle yet critical differences between Mg2+ and Fe2+.

Computation helps explain why Fe2+ can be a potent cofactor for nucleic acid processing enzymes. Conformations and geometries are nearly identical between complexes of Fe2+ or Mg2+. These similarities are observed in both hexa aquo and phosphate complexes. Important differences between Mg2+ and Fe2+ follow. (i) Interactions with water. Hexa aquo Fe2+ is a stronger acid than hexa aquo Mg2+ (Table (Table1).1). Our calculations show greater depletion of electrons from water molecules that coordinate Fe2+ than those that coordinate Mg2+. Greater frequency of M2+(H2O)5(OH) as predicted for Fe2+ over Mg2+ would facilitate important reactions in which a ribose hydroxyl gives up a proton while acting as a nucleophile. (ii) Nucleophilic attack at phosphorus. Because of low lying d orbitals, Fe2+ has greater electron withdrawing power than Mg2+ from first shell phosphate ligands (Figure (Figure4,4, Supplementary Tables S2 and 3). In coordination complexes with phosphate groups, the phosphorus atom is a better electrophile when M2+ = Fe2+ than when M2+ = Mg2+. Modulation of rates of nucleophilic attack on phosphorus is important in many biological reactions, including formation of phosphodiester bonds in polymerization and ligation. This difference between Mg2+ and Fe2+ is apparent in ribozyme reactions. We have observed that phosphoryl transfer ribozymes, including a ribozyme selected in the presence of Mg2+, are more active with Fe2+ as a cofactor than with Mg2+ (23). (iii) Phosphate affinity. Also because of low-lying d orbitals, Fe2+ interacts with slightly greater affinity than Mg2+ with the oxygen atoms of first shell phosphate ligands. Tighter binding of M2+ to RNA appears to improve folding and would be important during the catalytic reactions assayed here.


Popović and Ditzler performed what was, to our knowledge, the first-ever in vitro RNA selection under plausible pre-GOE conditions (64) and demonstrated that ribozymes obtained by selection with Fe2+ are active upon substitution of Mg2+ for Fe2+. Here we begin to test the hypothesis that Fe2+ was a divalent cation that actively facilitated processing of nucleic acids by proteins on the ancient Earth. We have investigated ancient Earth biochemistry of proteins that play important roles in nucleic acid processing. Polymerases and ligases are both important in DNA replication, RNA transcription and DNA repair. The ability of Fe2+ to substitute for Mg2+ in the polymerases and the ligase suggests that Fe2+, in the absence of O2, could substitute for Mg2+ in a broad variety of enzymes involved in nucleic acid processing. We have used calculations to show that Fe2+ is a viable substitute for Mg2+ in nucleic acid-cation interactions. Therefore, Fe2+ might have played a role in the early evolution of proteins and nucleic acids.

Supplementary Material

Supplementary Data


Supplementary Data are available at NAR Online.


National Aeronautics and Space Administration [NNX16AJ28G, in part]. Funding for open access charge: NASA [NNX16J28G].

Conflict of interest statement. None declared.


1. Hazen R.M., Ferry J.M. Mineral evolution: mineralogy in the fourth dimension. Elements. 2010; 6:9–12.
2. Anbar A.D. Oceans. elements and evolution. Science. 2008; 322:1481–1483. [PubMed]
3. Holland H.D. The oxygenation of the atmosphere and oceans. Philos. Trans. R. Soc. Lond. B Biol. Sci. 2006; 361:903–915. [PMC free article] [PubMed]
4. Klein C. Some precambrian banded iron-formations (BIFs) from around the world: their age, geologic setting, mineralogy, metamorphism, geochemistry, and origin. Am. Mineral. 2005; 90:1473–1499.
5. Holland H.D. The oceans: a possible source of iron in iron-formations. Econ. Geol. 1973; 68:1169–1172.
6. Kozlowski H., Kolkowska P., Watly J., Krzywoszynska K., Potocki S. General aspects of metal toxicity. Curr. Med. Chem. 2014; 21:3721–3740. [PubMed]
7. Prousek J. Fenton chemistry in biology and medicine. Pure Appl. Chem. 2007; 79:2325–2338.
8. Harel A., Bromberg Y., Falkowski P.G., Bhattacharya D. Evolutionary history of redox metal-binding domains across the tree of life. Proc. Natl. Acad. Sci. U.S.A. 2014; 111:7042–7047. [PubMed]
9. Dupont C.L., Yang S., Palenik B., Bourne P.E. Modern proteomes contain putative imprints of ancient shifts in trace metal geochemistry. Proc. Natl. Acad. Sci. U.S.A. 2006; 103:17822–17827. [PubMed]
10. Theil E.C., Goss D.J. Living with iron (and oxygen): questions and answers about iron homeostasis. Chem. Rev. 2009; 109:4568–4579. [PMC free article] [PubMed]
11. Valko M., Morris H., Cronin M.T.D. Metals, toxicity and oxidative stress. Curr. Med. Chem. 2005; 12:1161–1208. [PubMed]
12. Aguirre J.D., Culotta V.C. Battles with iron: manganese in oxidative stress protection. J. Biol. Chem. 2012; 287:13541–13548. [PMC free article] [PubMed]
13. Ushizaka S., Kuma K., Suzuki K. Effects of Mn and Fe on growth of a coastal marine diatom talassiosira weissflogii in the presence of precipitated Fe(III) hydroxide and EDTA-Fe(III) complex. Fish. Sci. 2011; 77:411–424.
14. Martin J.E., Imlay J.A. The Alternative aerobic ribonucleotide reductase of Escherichia Coli, NrdEF, is a manganese-dependent enzyme that enables cell replication during periods of iron starvation. Mol. Microbiol. 2011; 80:319–334. [PMC free article] [PubMed]
15. Cotruvo J.A., Stubbe J. Class I ribonucleotide reductases: metallocofactor assembly and repair in vitro and in vivo. Annu. Rev. Biochem. 2011; 80:733–767. [PMC free article] [PubMed]
16. Anjem A., Varghese S., Imlay J.A. Manganese import is a key element of the oxyr response to hydrogen peroxide in Escherichia Coli. Mol. Microbiol. 2009; 72:844–858. [PMC free article] [PubMed]
17. Wolfe-Simon F., Starovoytov V., Reinfelder J.R., Schofield O., Falkowski P.G. Localization and role of manganese superoxide dismutase in a marine diatom. Plant Physiol. 2006; 142:1701–1709. [PubMed]
18. Torrents E., Aloy P., Gibert I., Rodriguez-Trelles F. Ribonucleotide reductases: divergent evolution of an ancient enzyme. J. Mol. Evol. 2002; 55:138–152. [PubMed]
19. Brown I.D. What factors determine cation coordination numbers. Acta Crystallogr. Sect. B. 1988; 44:545–553.
20. Wulfsberg G. Principles of descriptive inorganic chemistry. 1991; Sausalito: University Science Books.
21. Rashin A.A., Honig B. Reevaluation of the born model of ion hydration. J. Phys. Chem. 1985; 89:5588–5593.
22. Uudsemaa M., Tamm T. Calculation of hydration enthalpies of aqueous transition metal cations using two coordination shells and central ion substitution. Chem. Phys. Lett. 2004; 400:54–58.
23. Athavale S.S., Petrov A.S., Hsiao C., Watkins D., Prickett C.D., Gossett J.J., Lie L., Bowman J.C., O’Neill E., Bernier C.R. et al. RNA folding and catalysis mediated by iron (II). PLoS One. 2012; 7:e38024. [PMC free article] [PubMed]
24. Hsiao C., Chou I.-C., Okafor C.D., Bowman J.C., O’Neill E.B., Athavale S.S., Petrov A.S., Hud N.V., Wartell R.M., Harvey S.C. et al. Iron(II) plus RNA can catalyze electron transfer. Nat. Chem. 2013; 5:525–528. [PubMed]
25. Steitz T.A. DNA polymerases: structural diversity and common mechanisms. J. Biol. Chem. 1999; 274:17395–17398. [PubMed]
26. Doherty A.J., Dafforn T.R. Nick recognition by DNA ligases. J. Mol. Biol. 2000; 296:43–56. [PubMed]
27. Lee J.Y., Chang C., Song H.K., Moon J., Yang J.K., Kim H.K., Kwon S.T., Suh S.W. Crystal structure of nad+‐dependent DNA ligase: modular architecture and functional implications. EMBO J. 2000; 19:1119–1129. [PubMed]
28. Ellenberger T., Tomkinson A.E. Eukaryotic DNA ligases: structural and functional insights. Annu. Rev. Biochem. 2008; 77:313–338. [PMC free article] [PubMed]
29. Lykke-Andersen J., Christiansen J. The C-Terminal carboxy group of T7 RNA polymerase ensures efficient magnesium ion-dependent catalysis. Nucleic Acids Res. 1998; 26:5630–5635. [PMC free article] [PubMed]
30. Yin Y.W., Steitz T.A. The Structural mechanism of translocation and helicase activity in T7 RNA polymerase. Cell. 2004; 116:393–404. [PubMed]
31. Gao Y., Yang W. Capture of a third Mg2+ is essential for catalyzing dna synthesis. Science. 2016; 352:1334–1337. [PubMed]
32. Jannasch H.W., Wirsen C.O., Molyneaux S.J., Langworthy T.A. Comparative physiological studies on hyperthermophilic archaea isolated from deep-sea hot vents with emphasis on pyrococcus strain Gb-D. Appl. Environ. Microbiol. 1992; 58:3472–3481. [PMC free article] [PubMed]
33. Sousa R., Mukherjee S. T7 RNA polymerase. Prog. Nucleic Acid Res. Mol. Biol. 2003; 73:1–41. [PubMed]
34. Shuman S. DNA ligases: progress and prospects. J. Biol. Chem. 2009; 284:17365–17369. [PMC free article] [PubMed]
35. Athavale S.S., Gossett J.J., Hsiao C., Bowman J.C., O’Neill E., Hershkovitz E., Preeprem T., Hud N.V., Wartell R.M., Harvey S.C. et al. Domain III of the T. thermophilus 23S rRNA folds independently to a near-native state. RNA. 2012; 18:752–758. [PubMed]
36. Ban N., Beckmann R., Cate J.H., Dinman J.D., Dragon F., Ellis S.R., Lafontaine D.L., Lindahl L., Liljas A., Lipton J.M. et al. A New system for naming ribosomal proteins. Curr. Opin. Struct. Biol. 2014; 24:165–169. [PMC free article] [PubMed]
37. Petrov A.S., Bowman J.C., Harvey S.C., Williams L.D. Bidentate RNA-magnesium clamps: on the origin of the special role of magnesium in RNA folding. RNA. 2011; 17:291–297. [PubMed]
38. Becke A.D. Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A. 1988; 38:3098–3100. [PubMed]
39. Lee C.T., Yang W.T., Parr R.G. Development of the colle-salvetti correlation energy formula into a functional of the electron density. Phys. Rev. B Condens. Matter. 1988; 37:785–789. [PubMed]
40. Frisch M.J., Trucks G.W., Schlegel H.B., Scuseria G.E., Robb M.A., Cheeseman J.R., Scalmani G., Barone V., Mennucci B., Petersson G.A. et al. Gaussian 09, Revision A.0.1. 2009; Wallingford: Gaussian, Inc.
41. Glendening E.D., Feller D. Dication-water interactions: M2+(H2o)N clusters for alkaline earth metals M=Mg,Ca,Sr,Ba, and Ra. J. Phys. Chem. 1996; 100:4790–4797.
42. Schenter G.K., Glendening E.D. Natural energy decomposition analysis: the linear response electrical self energy. J. Phys. Chem. 1996; 100:17152–17156.
43. Gordon M.S., Schmidt M.W. Dykstra C.E., Frenking G, Kim KS, Scuseria GE Advances in electronic structure theory: GAMESS a decade later. Theory and Applications of Computational Chemistry: the First Forty Years. 2005; Amsterdam: Elsevier science; 1167–1189.
44. Lanier K.A., Athavale S.S., Petrov A.S., Wartell R., Williams L.D. Imprint of ancient evolution on rRNA folding. Biochemistry. 2016; 55:4603–4613. [PubMed]
45. Fuchs E. The interdependence of magnesium with spermidine and phosphoenolpyruvate in an enzyme‐synthesizing system in vitro. FEBS J. 1976; 63:15–22. [PubMed]
46. Kasting J.F. Theoretical constraints on oxygen and carbon dioxide concentrations in the precambrian atmosphere. Precambrian Res. 1987; 34:205–229. [PubMed]
47. Ueno Y., Johnson M.S., Danielache S.O., Eskebjerg C., Pandey A., Yoshida N. Geological sulfur isotopes indicate elevated ocs in the archean atmosphere, solving faint young sun paradox. Proc. Natl. Acad. Sci. U.S.A. 2009; 106:14784–14789. [PubMed]
48. Kasting J.F., Siefert J.L. Life and the evolution of earth's atmosphere. Science. 2002; 296:1066–1068. [PubMed]
49. Rosing M.T., Bird D.K., Sleep N.H., Bjerrum C.J. No climate paradox under the faint early sun. Nature. 2010; 464:744–747. [PubMed]
50. Tabor S., Richardson C.C. Effect of manganese ions on the incorporation of dideoxynucleotides by bacteriophage T7 DNA polymerase and Escherichia Coli DNA polymerase I. Proc. Natl. Acad. Sci. U.S.A. 1989; 86:4076–4080. [PubMed]
51. Litman R.M. The differential effect of magnesium and manganese ions on the synthesis of poly (Dg-Dc) and micrococcus luteus DNA by Micrococcus Luteus DNA polymerase. J. Mol. Biol. 1971; 61:1–23. [PubMed]
52. Irimia A., Zang H., Loukachevitch L.V., Eoff R.L., Guengerich F.P., Egli M. Calcium is a cofactor of polymerization but inhibits pyrophosphorolysis by the Sulfolobus Solfataricus DNA polymerase Dpo4. Biochemistry. 2006; 45:5949–5956. [PubMed]
53. Subramanya H.S., Doherty A.J., Ashford S.R., Wigley D.B. Crystal structure of an ATP-dependent DNA ligase from bacteriophage T7. Cell. 1996; 85:607–615. [PubMed]
54. Ferrer M., Golyshina O.V., Beloqui A., Böttger L.H., Andreu J.M., Polaina J., De Lacey A.L., Trautwein A.X., Timmis K.N., Golyshin P.N. A purple acidophilic di-ferric DNA ligase from ferroplasma. Proc. Natl. Acad. Sci. U.S.A. 2008; 105:8878–8883. [PubMed]
55. Steitz T.A., Steitz J.A. A general two-metal-ion mechanism for catalytic RNA. Proc. Natl. Acad. Sci. U.S.A. 1993; 90:6498–6502. [PubMed]
56. Hsiao C., Williams L.D. A recurrent magnesium-binding motif provides a framework for the ribosomal peptidyl transferase center. Nucleic Acids Res. 2009; 37:3134–3142. [PMC free article] [PubMed]
57. Rulisek L., Sponer J. Outer-shell and inner-shell coordination of phosphate group to hydrated metal ions (Mg2+, Cu2+, Zn2+, Cd2+) in the presence and absence of nucleobase. The role of nonelectrostatic effects. J. Phys. Chem. B. 2003; 107:1913–1923.
58. Gresh N., Sponer J.E., Spackova N., Leszczynski J., Sponer J. Theoretical study of binding of hydrated Zn(II) and Mg(II) cations to 5΄-guanosine monophosphate. Toward polarizable molecular mechanics for DNA and RNA. J. Phys. Chem. B. 2003; 107:8669–8681.
59. Munoz J., Sponer J., Hobza P., Orozco M., Luque F.J. Interactions of hydrated Mg2+ cation with bases, base pairs, and nucleotides. Electron topology, natural bond orbital, electrostatic, and vibrational study. J. Phys. Chem. B. 2001; 105:6051–6060.
60. Trachtman M., Markham G.D., Glusker J.P., George P., Bock C.W. Interactions of metal ions with water: Ab initio molecular orbital studies of structure, bonding enthalpies, vibrational frequencies and charge distributions. 1. Monohydrates. Inorg. Chem. 1998; 37:4421–4431. [PubMed]
61. Murashov V.V., Leszczynski J. Theoretical study of complexation of phosphodiester linkage with alkali and alkaline-earth cations. J. Phys. Chem. B. 1999; 103:8391–8397.
62. Petrov A.S., Lamm G., Pack G.R. Calculation of the binding free energy for magnesium-RNA interactions. Biopolymers. 2005; 77:137–154. [PubMed]
63. Petrov A.S., Pack G.R., Lamm G. Calculations of magnesium-nucleic acid site binding in solution. J. Phys. Chem. B. 2004; 108:6072–6081.
64. Popović M., Fliss P.S., Ditzler M.A. In vitro evolution of distinct self-cleaving ribozymes in diverse environments. Nucleic Acids Res. 2015; 43:7070–7082. [PMC free article] [PubMed]

Articles from Nucleic Acids Research are provided here courtesy of Oxford University Press