768
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
REVIEW Addressing the Problems of Base Pairing and Strand Cyclization in TemplateDirected Synthesis A Case for the Utility and Necessity of #Molecular Midwives( and Reversible Backbone Linkages for the Origin of proto-RNA by Nicholas V. Hud* a ) b ), Swapan S. Jain a ) b ), Xiaohui Li a ) c ), and David G. Lynn* a ) c ) a
) Center for Fundamental and Applied Molecular Evolution, Georgia Institute of Technology and Emory University, Atlanta, Georgia, U.S.A. b ) School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, GA 30332, U.S.A. (phone: þ 1-404-3851162; fax: þ 1-404-8942295; e-mail:
[email protected]) c ) Departments of Chemistry and Biology, Emory University, Atlanta, GA 30322, U.S.A. (phone: þ 1-404-7279348; fax: 1-404-7276586; e-mail:
[email protected]) Dedicated to Leslie Orgel on the occasion of his 80th birthday
Nucleic acid synthesis is precisely controlled in living organisms by highly evolved protein enzymes. The remarkable fidelity of information transfer realized between template and product strands is the result of both the spatial selectivity of the polymerase active site for Watson – Crick base pairs at the point of nucleotide coupling and subsequent proof-reading mechanisms. In the absence of naturally derived polymerases, in vitro template-directed synthesis by means of chemically activated mononucleotides has proven remarkably inefficient and error-prone. Nevertheless, the spontaneous emergence of RNA polymers and their protein-free replication is frequently taken as a prerequisite for the hypothetical :RNA world;. We present two specific difficulties that face the de novo synthesis of RNA-like polymers in a prebiotic (enzyme-free) environment: nucleoside base selection and intramolecular strand cyclization. These two problems are inherent to the assumption that RNA formed de novo from pre-existing, chemically-activated mononucleotides in solution. As a possible resolution to these problems, we present arguments and experimental support for our hypothesis that small molecules (referred to as :molecular midwives;) and alternative backbone linkages (under equilibrium control) facilitated the emergence of the first RNA-like polymers of life.
1. Introduction. – The discovery of catalytic RNA molecules and the central role of RNA in contemporary life have been used as support for the hypothesis that an early form of life, before the advent of DNA or proteins, used RNA for both information storage and catalysis [1]. Some investigators are apparently under the impression that this :RNA world; existed as far back as the earliest stages of life, and that RNA was the first polymer of life. However, many prebiotic chemists consider it implausible that RNA polymers formed de novo on the prebiotic Earth [2] [3]. The chemical steps required to select the nucleoside bases, ribose, and phosphate from a solution of related compounds, and to connect these components appropriately to create an RNA polymer are generally considered too improbable to occur in a prebiotic environment. As an alternative scenario, a number of investigators have proposed that a more accessible E 2007 Verlag Helvetica Chimica Acta AG, ZGrich
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
769
RNA-like polymer, termed proto-RNA, must have come before RNA [2] [3]. Presumably, this polymer would have gradually evolved into RNA by the incremental replacement of individual elements (e.g., specific bases, backbone sugar), but it would have nevertheless had an initial structure still recognizable as an RNA-like polymer. If we accept that the chemical structure of RNA has changed over the course of evolution, we are still faced with the problem of how the first proto-RNA molecules came into existence. We are at the same time burdened and liberated by the prospect that the chemistry and structure of proto-RNA differed from that of present-day RNA. We are burdened by the prospect of sifting through innumerable possibilities for a proto-RNA that can spontaneously assemble from simple precursor molecules, but we are liberated by the possibility that alternative processes might greatly facilitate the formation of RNA-like polymers in a plausible prebiotic environment. Our laboratories are using a functional approach to investigate possible routes by which the first RNA-like molecules could have formed de novo. We have previously proposed that small-molecule self-assembly and reversible backbone linkages could have brought about the spontaneous formation and subsequent replication of such polymers [2] [4] [5]. In the present communication, we provide additional arguments for the appropriateness and importance of these concepts to the origin of an :RNA world;. We also discuss new and previously published results from our laboratories that support our collective proposal for how a :small-molecule world; could have selfassembled into a proto-RNA world. 2. Basic Questions. – We begin with the statement of two problems that currently guide our search for the origin of the first proto-RNA polymers. These are also problems that we feel have been overlooked or underappreciated by most researchers that subscribe to the :RNA world; hypothesis. The first problem relates to the original selection of the nucleoside bases (or their predecessors) from the diverse pool of molecules likely present on the prebiotic Earth. The second problem relates to backbone formation during oligonucleotide growth and the potential for deleterious intramolecular strand cyclization during the same process. 2.1. The Paradox of Base Pairing. The Watson – Crick base pairs are the essence of information storage and information transfer in all living organisms. It, therefore, seems most certain that the nucleotide bases were originally selected for use by an early form of life because they were able to form Watson – Crick base pairs. However, the free nucleoside bases do not form H-bonded base pairs in aqueous solution [6] [7]. Even at relatively high concentration (ca. 1m), the bases, either by themselves or as nucleosides or nucleotides, form coplanar stacks with their H-bonding edges exposed to the solvent. This situation is as expected from a thermodynamics perspective. The formation of vertical stacks reduces solvent-exposed hydrophobic surface area, while maintaining Hbonds with solvent H2O molecules. These H-bonds with H2O are essentially equivalent in energy to the H-bonds that are formed between Watson – Crick base pairs [8]. The lack of Watson – Crick base-pair formation by the free bases in aqueous solution presents a paradox regarding the origin of the first RNA-like polymers: How would life have anticipated Watson – Crick base pairing, and, therefore, coupled the bases into a polymer, if the nucleoside bases were not involved in such pairings prior to their attachment along a common backbone? One can always argue that the formation of
770
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
RNA-like polymers with appropriate bases was a complete accident. However, we argue that the probability for the spontaneous coupling of small molecules from a complex mixture to give rise to an RNA-like polymer with a majority of bases capable of Watson – Crick pairing is impossibly low, unless there is a mechanism that preorganizes only those molecules that can form base pairs prior to their coupling into a polymer. 2.2. The Obstacle of Strand Cyclization to Oligonucleotide Growth. Strand cyclization is a second problem that would have plagued de novo oligonucleotide synthesis if this synthesis had taken place with unpaired and chemically activated mononucleotides. It has long been appreciated that a polymer growing in solution tends to undergo intramolecular cyclization when grown to sufficient length, so that its two ends can come within close proximity and with a relative orientation that allows for covalent-bond formation [9]. Association of a nucleic acid product strand with a template strand can, in principle, inhibit intramolecular cyclization by constraining the ends of the product strand to be separated in space. However, strand cyclization can still be a problem if the growing product strand is in exchange with solution during monomer addition. In this case, product-strand cyclization can occur once the product strand has grown to sufficient length to allow cyclization, but is still short enough to be in exchange with solution. If we assume that the structural properties of proto-RNA were similar to those of contemporary RNA, then strand cyclization would have been a major impediment to oligonucleotide growth (assuming an irreversible coupling chemistry), as single-stranded RNA is very flexible, with a persistence length of approximately five nucleotides [10]. Studies of single-stranded DNA cyclization illustrate the likely magnitude of the strand-cyclization problem. Chemically activated oligonucleotides as short as dinucleotides exhibit rapid cyclization [11], and dinucleotides formed on a template strand would certainly be in exchange with solution if Watson – Crick base pairing were the only factor maintaining association with the template. 2.3. Proposed Resolutions to the Problems of Base Selection and Oligonucleotide Growth. We put forth that the most plausible resolution to the paradox of base pairing and oligonucleotide self-cyclization is that the bases were paired within linear molecular assemblies before becoming incorporated into polymers, and that the polymer backbone was formed with reversible chemical linkages under equilibrium control. These two principles, regardless of whether achieved precisely by the mechanisms proposed below, would have both selected the bases from a mixture of similar molecules and promoted linear polymer growth rather than self-cyclization. 3. #Molecular Midwives( as a Solution to Base Selection and Preorganization. – Over the past several decades, a number of chemical reactions have been presented as plausible models for the prebiotic synthesis of the nucleic acid bases [12]. The fact that these reactions produce many more products than just the DNA and RNA nucleoside bases has important implications regarding the problem of nucleoside base selection from the prebiotic chemical inventory. As we have suggested, heterocyclic compounds with chemical properties similar to the bases would have disrupted proto-RNA formation if the appropriate bases were not selectively preorganized prior to polymer formation. On the other hand, molecules present in the prebiotic chemical inventory
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
771
could have facilitated the selection of the appropriate bases through non-covalent interactions. We have, thus, coined the term :molecular midwives; 1) to designate prebiotic molecules facilitating the formation of the first RNA-like molecules, but without themselves being covalently attached to these polymers [2]. These molecules would have aided the :birth; of proto-RNA, but would have no longer been necessary once RNA-like polymers evolved alternative (e.g., protein-based) means for replication. We envision molecular midwives facilitating the selection and preorganization of the free bases by acting as nanometer-scale templates upon which the nucleoside bases stack in aqueous solution (Fig. 1, a). In the prebiotic environment, these midwife molecules would have been similar in size and shape to the numerous small, planar molecules that are known to intercalate the bases of RNA and DNA [13]. In principle, such molecules could act as templates for base assemblies composed of base pairs, base triplets, and base tetrads. Specific examples of planar molecules with shapes similar to H-bonded base assemblies are shown in Fig. 2. It is not difficult to imagine that abiotic
Fig. 1. Schematic representation of the -molecular midwife. hypothesis [2]. a) A midwife molecule, proflavine (in blue) acts as a template for the formation of a Watson – Crick base pair. H-atoms of the nucleoside bases that will be replaced by C-atoms upon backbone formation are shown in magenta. b) In the case of molecular midwives with a greater association constant for base assembly than for selfassociation, columnar stacks containing alternating midwife molecules and base assemblies spontaneously form under certain conditions. These columnar stacks preorganize the bases such that the introduction of a linkage chemistry that couples the bases when spaced by 6.8 J leads to the formation of RNA-like polymers. Because the midwife molecules are only associated with the resulting polymers through non-covalent interactions, changes in solution conditions can lead to the removal of the intercalating midwives. 1)
For convenience, this term will not be hyphenated in the following discussion.
772
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
chemical reactions that produce nucleoside bases and porphyrin-like molecules [14] could also give rise to heterocycles with shapes similar to the putative midwife molecules shown in Fig. 2. If the association constant of a midwife molecule and a planar nucleoside-base assembly is greater than that of midwife self-association (e.g., in the case of a positively charged midwife), then columnar stacks of alternating midwives and bases would be expected to spontaneously assemble in aqueous solution (Fig. 1, b). On the prebiotic Earth, such columnar stacks could have both selected the bases from a mixture of molecules, and organized the bases for incorporation into RNA-like polymers (Fig. 1, b). We note that the distance between the planes defined by bases stacked on opposite faces of an intercalating molecule is 6.8 1, which is also the length of the fully extended backbone of present-day RNA. Thus, the action of midwife molecules in assembling the bases could have also dictated the length of the proto-RNA backbone linkage, which may have remained essentially constant throughout evolution due to other advantageous properties (e.g., the ability to adopt a variety of folded structures). 3.1. Molecular Midwives vs. Macroscopic Surfaces. The association of organic molecules with inorganic surfaces has long been cited as a possible means by which the building blocks of life were locally concentrated on the prebiotic Earth [15]. Direct imaging of the nucleoside bases deposited on macroscopic surfaces has revealed, as might be expected, that the bases lay flat on the mineral surface [16]. It is difficult to imagine how such an arrangement would facilitate the incorporation of bases into RNA-like polymers, as the planes of paired bases are perpendicular to the helical axis of a nucleic acid polymer. Furthermore, if either the bases or the backbone of an RNAlike polymer possess a particular affinity for a given macroscopic surface, then the association constant of the polymer for the surface must increase with polymer length. This trend inevitably leads to the irreversible attachment of a polymer to the surface above some threshold length [17], which would likely impair the functionality of any proto-RNA. The planar faces of organic, polycyclic molecular midwives can be viewed as surfaces as well, more specifically, as nanometer-scale surfaces with the potential for selective and reversible associations with the nucleoside bases. Furthermore, as mentioned above, the formation of columnar stacks with alternating midwives and nucleoside bases would locally concentrate and organize the bases in an arrangement that facilitates oligonucleotide synthesis. Additionally, because the non-covalent interactions between bases and midwife molecules remain local, regardless of polymer length, RNA-like polymers synthesized as part of midwife – nucleoside-base columnar stacks would have the freedom to grow in length without the associations between midwife molecules and bases becoming irreversible. Thus, changes in solution conditions, such as dilution, temperature, or ionic strength, could conceivably release polymers and midwives from their columnar stacks, regardless of polymer length. Such a feature would become indispensable as life increased in complexity (and genome length). 3.2. Demonstrating the Promotion of Base Pairing by Molecular Midwives. Small molecules that intercalate nucleic acid duplexes exhibit association constants with DNA and RNA that are typically in the range of 105 – 106 m 1 [18]. Thus, it would be expected that intercalating molecules at concentrations above ca. 10 mm would,
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
773
Fig. 2. Nucleoside base assemblies and heterocyclic molecules with similar shapes. Proflavine has already been shown to act as a midwife in ligation assays that include duplex formation between the template and substrate strands [19]. In the present communication, data is presented that supports the ability for coralyne to act as a midwife when the template – substrate ligation complex includes base triplets. The phthalocyanine analogue with a similar shape to the base tetrads is shown for illustrative purposes, as a molecular midwife has not yet been demonstrated for a tetrad system.
according to the Law of Mass Action, shift the equilibrium of nucleoside bases in solution towards assemblies that contain intercalated base pairs. We have initiated experimental investigations of this method for driving nucleic acid assembly. In a previous report, we demonstrated that proflavine (Fig. 2), a well-characterized intercalator, substantially increases the spontaneous coupling of two short oligonu-
774
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
cleotide substrates along a complimentary oligonucleotide template [19]. Specifically, we monitored the formation of a phosphorothioate linkage between two oligonucleotides, dT4 with a 5’-iodo substituent, and dT3 with a 3’-phosphorothiate, to yield a heptamer product in the presence of a (dA)7 template and proflavine (Fig. 3, a). Coupling yields in this ligation test system, for otherwise identical reaction conditions, were found to increase as much as three orders of magnitude over that observed in the absence of proflavine [19]. Although the coupling chemistry used in these experiments
Fig. 3. Experimental system used to explore the ability of proflavine to act as a midwife in a templatedirected ligation reaction including Watson – Crick base pairs. a) 3’-Phosphorothioate-(dT)3 and 5’-iodo(dT)4 substrate strands are too short to form a stable complex with the (dA)7 template strand. The addition of proflavine promotes the assembly of a ligation-active complex, which leads to the formation of the (dT)7 product within an internal phosphorothioate linkage. b) Image of a denaturing polyacrylamide gel after electrophoretic analysis, illustrating the ability of proflavine to act as a midwife for the ligation test system described in a. Lanes M: molecular-weight markers of (dT)8 , (dT)7, and (dT)6 ; lane C1: 32P-labeled 3’-phosphorothioate-(dT)3 only; lane C2 : substrate 32P-labeled 3’-phosphorothioate(dT)3 and 5’-iodo-(dT)4 ; remaining lanes: 32P-labeled 3’-phosphorothioate-(dT)3 , 5’-iodo-(dT)4 , template strand (dA)16 , and proflavine at 0 – 250 mm concentration. All mixtures were incubated for 24 h at 48. For each experiment, substrate and template strands (where present) were added to the reaction buffer to a final concentration of 1.0 mm for each strand. Experimental conditions correspond to those reported previously [19].
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
775
is not prebiotically plausible, our results demonstrate that the desired duplex assembly is promoted by proflavine, and that backbone coupling is possible in an intercalationstabilized assembly. The substantial increase observed in oligonucleotide coupling reactions when proflavine was present is certainly supportive of the molecular-midwife hypothesis. Nevertheless, mononucleotide coupling and continuous oligonucleotide growth also needs to be demonstrated, as these capabilities most directly address the problem of de novo polymer synthesis from monomer building blocks. However, the realization of such a system will require a coupling chemistry that is reversible and under equilibrium control. As indicated above, a growing product strand created from bifunctional 5’iodo – 3’-phosphorothioate-mononucleotides will simply lead to bifunctional oligonucleotides that principally form cyclization products as small as dinucleotides [11]. One might argue that the molecular-midwife hypothesis introduces an additional complication to origin-of-life scenarios by requiring one more ingredient. We would counter that the added benefits of a molecular midwife far outweigh this additional requirement and, overall, the participation of a midwife makes the spontaneous formation of proto-RNA much more probable. For example, the ability for midwife molecules to promote template-directed synthesis with templates and substrates at much lower concentrations implies that the bulk concentration of midwife molecules and nucleic acid components required for synthesis of RNA-like polymers will be far lower than the concentrations required for the same reactions to occur in the absence of molecular midwives. As an illustration, we found that 5’-iodo-dT4 and 3’-phosphorothioate-dT3 coupling along a dA7 template can be achieved for substrate oligonucleotide concentrations of only 1 mm [19] with 125 mm proflavine, whereas much less product is observed in the absence of proflavine for oligonucleotide concentrations higher by up to a factor of 103 (unpublished result by S. S. J. and N. V. H.). 3.3. Molecular-Midwife Dictation of Base-Assembly Geometry. A question related to how the bases were originally selected from the prebiotic chemical inventory is why purine – pyrimidine base pairs were selected as the fundamental pairing unit, as opposed to other possible combinations (i.e., purine – purine, pyrimidine – pyrimidine). Double-helix formation by H-bonded pairs of planar heterocycles is clearly not limited to purine – pyrimidine base pairs [20]. Furthermore, duplexes formed by purine – pyrimidine pairs do not have a particular advantage with respect to duplex stability over duplexes formed with larger base pairs (e.g., with two bicyclic bases) [21]. The midwife hypothesis also provides a possible answer to the question of how and why the purine – pyrimidine Watson – Crick base pairs were selected. The shape of the original midwife molecules would have exerted a substantial influence, perhaps the greatest single influence, on the selection of purine – pyrimidine pairs if it had provided a surface for base stacking that was the size and shape of a Watson – Crick pair (or a base assembly that included these pairs as a substructure). Assemblies that were smaller (e.g., pyrimidine – pyrimidine pairs) or larger (e.g., purine – purine pairs) would have formed less-stable associations with the midwife surface, and would have thereby been excluded from columnar-stack formation and, ultimately, from polymer formation. As a means to demonstrate the potential for molecular midwives to dictate the structure of base assemblies utilized in a template-directed ligation reaction, we have qualitatively compared oligonucleotide-coupling efficiency as a function of the match
776
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
between the structure of a model molecular midwife and the base-pairing assembly required for ligation. For these studies, we compared the Watson – Crick ligation system described above and an analogous system based upon a pyrimidine triplex (Fig. 4, a). The small molecule coralyne (see Fig. 2) was used as a midwife for the triplex ligation system, as this molecule has a shape similar to a base triplet, and is known to selectively bind triplex over duplex nucleic acids [22]. The resulting electrophoresis gel image (Fig. 4, b) demonstrates that coralyne promotes the ligation of substrate strands along an existing duplex through the formation of a coralyne-stabilized triplex. In contrast, proflavine, which promotes duplex-based ligation (Fig. 3), is not nearly as active in the triplex-based ligation system (Fig. 4). Likewise, coralyne is ineffective in the promotion of duplex-based ligation (data not shown). These results illustrate that the match between the shape of a small planar molecule and a particular base assembly is essential for midwife activity. By the same principle, we propose that, from a complex mixture of heterocyclic molecules, midwife molecules could have selected the original bases and their Watson – Crick pairing structure. 3.4. Potential Limits on the Minimum Size of a Molecular Midwife. Size, shape, solubility, electrostatic charge, and polarizability are among the molecular properties that are likely to determine whether or not a particular molecule can act as a midwife for the assembly of nucleoside bases in aqueous solution. We have hypothesized that the size and shape of the original molecular midwife was close to either that of a Watson – Crick base pair, or to that of tetrads formed by two such base pairs (Fig. 2). Theoretical and experimental investigations of self-assembly in aqueous solution may provide some insights into whether this size for a molecular midwife is governed by fundamental physical principles. It is generally accepted that the free energy per unit of surface area associated with the hydrophobic effect becomes more significant as solute surface area increases above approximately 1 nm2 due to the perturbation of H-bonded H2O networks near a hydrophobic surface of around these dimensions [23]. Chandler has proposed that this property of the hydrophobic effect places a lower limit on the possibility of creating stable self-assembled structures in aqueous solution from molecules with dimensions smaller than 1 nm [23]. Smaller molecules (e.g., purines and pyrimidines) can transiently form assemblies in aqueous solution, but these assemblies are short-lived compared to the thermodynamically stable assemblies that can be formed by larger solutes (e.g., micelles formed by surfactants). While the purine and pyrimidine bases are less than 1 nm in any dimension, the width of a Watson – Crick base pair reaches the 1-nm limit at its widest point. Thus, it is feasible that a molecular midwife of this dimension could have promoted assembly of the bases. We note that molecular midwives that mimic the size and shape of a base tetrad (Fig. 2) would have a dimension of ca. 1 nm 1 nm. This similarity between the predicted lower-size limit for self-assembly in aqueous solution and the size of purine – pyrimidine H-bonded assemblies suggests to us that the size of the original molecular midwife, and the size of the Watson – Crick base pairs, may be a direct result of the physical limits of selfassembly in an aqueous environment. 4. Reversible Backbone Linkages and Oligonucleotide Growth. – Our laboratories are also exploring the possibility that alternative backbone linkages can likewise promote the self-assembly of RNA-like polymers, and that such backbones preceded
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
777
Fig. 4. Experimental system used to explore the ability of coralyne to act as a midwife in a templatedirected ligation reaction including A · T · T base triplets. a) 3’-Phosphorothioate-(dT)3 and 5’-iodo-(dT)4 substrate strands are too short to form a stable triplex with the indicated DNA hairpin. The addition of coralyne promotes the assembly of a ligation-active complex, which leads to the formation of (dT)7 within an internal phosphorothioate linkage. b) Image of a denaturing polyacrylamide gel after electrophoresis, illustrating the ability of coralyne to act as a midwife for the ligation test system described in a. Each sample was run in triplicate on gel in adjacent lanes. Lanes 1: 32P-labeled 3’phosphorothioate-(dT)3 only; lanes 2: both substrate strands, 32P-labeled 3’-phosphorothioate-(dT)3 and 5’-iodo-(dT)4 ; lanes 3: both substrate strands with DNA hairpin; lanes 4: both substrate strands with coralyne; lanes 5: both substrate strands with proflavine; lanes 6: both substrate strands with DNA hairpin and coralyne; lanes 7: both substrate strands with DNA hairpin and proflavine. All reaction mixtures were incubated for 24 h at 48. For each experiment, substrate and template strands (where present) were added to the reaction buffer to a final concentration each of 1.0 mm in strand. For experiments containing proflavine or coralyne, these molecules were present at 125 mm. Additional experimental details are reported in [19].
778
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
the contemporary phosphate – ribose backbone of RNA. The idea that a different backbone preceded RNA is not new and has been proposed for several reasons, including the unfavorable enthalpy of phosphodiester-bond formation discussed above, as well as the problem that ribose was likely only a minor component of the prebiotic chemical inventory [2] [3]. Alternative thermodynamically favored backbones do not require monomer activation to form and provide a means for error correction, when information transfer is under equilibrium control. This latter feature has not proven to be possible for nonenzymatic backbone synthetic schemes that rely upon chemical activation of mononucleotides. We aim to emulate the fundamental features that are critical to the high-fidelity template-directed synthesis of the nucleic acids in living organisms, but without the use of highly evolved enzymes. Within the active site of a polymerase, phosphodiester-bond formation is achieved by the reversible condensation of nucleoside triphosphates that are complementary to a template strand. The thermodynamic product (with correct Watson – Crick pairing) is trapped by subsequent hydrolysis of the resulting pyrophosphate leaving group (Scheme 1). In our attempts to mimic this basic two-step process, we see as critical an initial coupling reaction that exploits the stability of template association, as well as a subsequent reaction that traps this thermodynamically favored product. Scheme 1. Minimal Coupling Steps in Biological Template-Directed Polymerization
4.1. Demonstrating the Merits of Reversible-Backbone Coupling by Experiment. A process we have developed that maintains the functional features of reversible coupling and irreversible trapping in a template-directed synthesis is the reductive amination of amines and aldehydes. In order to assess the ability of imines to function as isosteric replacements of the phosphodiester in duplex association, amine/aldehyde function-
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
779
ality was synthetically incorporated into DNA nucleosides and nucleotide oligomers [4] [24] [25]. As shown in the upper part of Scheme 2, the estimated equilibrium constant K3 for the hexameric duplex containing a single backbone imine greatly favors duplex formation. Surprisingly, only imine, no hemi-aminal intermediate, was observed in the associated complex contributing to the equilibrium. Therefore, even in H2O, where hydration is expected to contribute significantly to the equilibrium, the conformation of the imine more significantly stabilizes the duplex and functions as an adequate, if not superior, replacement for the phosphate. Scheme 2. Catalytic Activity of a Simple DNA Template, d(GCAACG), on the Reductive Amination of 5’-NH2-dTGC and d(CGT)-3’-CH2CHO. The equilibrium constants were determined by NMR.
Upon chemical reduction to the amine (lower panel in Scheme 2), duplex stability is reduced by a factor of nearly 106 [4] [24] [26]. This destabilization was attributed to the increased flexibility arising from the removal of two conformationally restricted sp2hybridized centers (N and C) in the linking backbone. This dramatic destabilization derived from such a simple change in the product backbone is both instructive and quite useful. In a practical sense, the simple hexameric DNA template can serve as a robust catalytic ligase in a reaction that is virtually immune to the product inhibition that so limits linear template reactions (e.g., ribozyme functions). Here, the product concentration must reach a 106-fold excess over the substrate concentration, before it
780
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
competes for catalytic sites along the template. More generally, these studies clearly establish that alternative backbone linkages can dramatically regulate the thermodynamics of template association, and this thermodynamic control can be exploited for catalytic cycles involving template-directed synthesis (Scheme 2). Amine-nucleoside polymers (ANPs) with uniformly modified backbone linkages were also successfully synthesized on DNA templates. Monomeric building blocks were prepared, with an NH2 group replacing the 5’-OH function, and a CH2CHO moiety replacing the 3’-OH group, to afford 5’-NH2-dT-3’-CH2CHO (T1). While T1 can form intramolecular imines, no reaction was detected under mild reducing conditions in an aqueous environment [5]. However, in the presence of the (dAp )8 template, the reaction gave high yields of a single T8 ANP product. This example represents the first clean, nonenzymatic template-directed polymerization reaction that gives chain-lengthspecific products.
4.2. Exploiting the Fidelity of Equilibrium-Controlled Template-Directed Synthesis. The product distribution of the template-directed reductive polymerization of T1 on a (dAp )8 template in the presence of NaBH3CN is displayed in Table 1. The process is characteristic of step-growth kinetics. For this exponential growth in product distribution, the reaction rates of the amine-mononucleotide T1 as starting material and the Tn amine-oligonucleotide intermediate products (dTN )n must follow the overall reaction-rate order T1 > T2 > T4 in the formation of the full-length T8 product on a (dAp )8 template. These relative rates are clearly indicated by the absence of T3 , T5 , T6 , and T7 products. Two features are uniquely and critically important to this templatedirected polymerization. First, reductive amination of the amine linkage reduces duplex stability (Scheme 2). Therefore, the association of T2 with the DNA template would incur a greater entropic penalty than the association of two T1 monomers coupled through the imine linkage (Scheme 2). Second, as reductive amination occurs only on the template, and since the reaction rates are directly dependent on the imine concentration, the number of binding sites is expected to contribute critically to the overall reaction rate. On a template of length n, which is only read in an antiparallel direction [5], there are (n 1) monomer (T1) binding sites, (n 3) T2 binding sites, etc. This decreasing number of sites as the product chain length increases further favors the rate order T1 > T2 > T4 . Most importantly, equilibrium control overcomes completely the obstacle of strand cyclization.
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
781
Table 1. Product Distribution upon Reductive Polymerization of T1 on a (dAp )8 Template. Conditions: 20 mol-equiv. of NaBH3CN in H2O at 248. Time [h]
0 6 24 48 b ) 60 a
Yield [%] a ) Monomer ( T1)
Dimer ( T2 )
Tetramer ( T4 )
Octamer ( T8 )
100 0 0 0 0
0 99 15 0 0
0 0 80 22 0
0 0 0 70 91
) Determined by HPLC analysis. b ) Additional reducing agent added to complete the reaction.
Predicting that both the total number of binding sites and the entropic activation of assembly contribute to the rate of product formation, it follows that dimer formation will be defined by adjacent base pairs. It should, therefore, be possible to control monomer polymerization and translational fidelity further with longer templates. Accordingly, a 32-mer DNA template strand, i.e., 5’-dAAAAAT(AAAAAAAT)3AA3’, approximating the length of the smallest catalytic ribozyme and the molecular weight of simple functional protein domains, was used to template the polymerization of T1 and the amide dimers TNT and ANT. The 32-mer amide/amine nucleoside polymer was formed as the sole product (unpublished result by X. L and D. G. L.). No truncations resulting from strand cyclization were detected within the resolution of HPLC and MALDI-TOF-MS analyses. No misincorporation of TNT or ANT monomers (which would constitute a change in product molecular weight of 9 mass units) were detected at levels > 1%. Polymerization of T1 and ANT substrates, which require more bonds to form along the same template to give a product of equal length, also gave a single 32-mer ANP product (Table 2). These reactions clearly demonstrate the ability to use present-day nucleic acid polymers as templates to accurately translate sequence and chain-length information into alternative polymer backbones. It has also been possible to demonstrate that these new polymer backbones are able to direct phosphodiesterbond formation [26], translating that information back into present-day polymers. The fact that these simple reactions simultaneously solve the problems of high-fidelity sequence transfer, uniform product chain length, and elimination of premature
Table 2. Time-Dependent Yield of 32-mer Amine-Nucleoside Polymers ( ANPs) upon Reductive Polymerization of Different Substrates. Conditions: 20 mol-equiv. of NaBH3CN in H2O at 248. Substrates
TNT/ANT T1/ANT
Yield [%] a ) 12 h
24 h
48 h b )
72 h
5 n.d.
18 6
81 27
n.d. c ) 76
a ) Determined by HPLC analysis. b ) Additional reducing agent added to complete the reaction. c ) Not determined.
782
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
termination due to chain cyclization underscore the power of reversible backbonecoupling reactions. 4.3. What was the Structure of the Original proto-RNA Backbone? Our investigations of reversible backbone linkages, discovered through a functional rather than a structural search, illustrate the utility of this general approach in obtaining high-fidelity transfer of nucleotide-sequence information. The imine/amine system is only one of the possible chemical linkages that we are now certain will achieve this end. We do not propose that this particular linkage was necessarily the original linkage of proto-RNA, only that the concept of monomer assembly under equilibrium control is, at present, the most robust approach to achieving nonenzymatic template-directed synthesis. As such, we reaffirm our earlier proposals that reversible backbone linkages are much more likely to have produced the molecular order necessary for the original RNA-like polymers and the chemical strategies necessary for their replication, as opposed to processes relying upon monomer activation. With regard to the actual chemical structure of the original proto-RNA backbone, we are now positioned to look back in structural space for entities that maintain the functional advantages of reversible backbone-linkage chemistry. In other words, we seek a backbone with a structure as similar as possible to the phosphodiester linkage of today, but with chemical properties that allow equilibrium control of the templatedirected polymer products. A class of linkages that potentially satisfy these criteria are acetals [2] [27]. The tetrahedral C-atom of an acetal linkage between two nucleotides would have the same number of atoms and the same basic geometry as the phosphate linkage of contemporary RNA. However, acetals have low-energy bonds and low activation barriers of formation. We have recently demonstrated that simply drying nucleosides with glyoxylate (the ionized form of glyoxylic acid) in the presence of divalent metal ions spontaneously produces (glyoxylate – acetal)-linked dinucleotides [27]. The acetal formed by glyoxylate is a close structural and electrostatic analogue of a phosphodiester linkage [27]. Thus, such backbones satisfy our vision of a prebiotic backbone that is structurally similar to RNA, but functionally much more easily assembled and quite possibly replicated. 5. Theory Falsification, Validation, and Reinvention of the #RNA World(. – The studies and hypotheses described here provide perspective on the most critical issue regarding the de novo assembly and replication of biopolymers: the simultaneous growth of molecular order and transferable sequence information. We have framed this issue in the context of two functional challenges, the paradox of base pairing and abiotic template-directed polymerizations. We have demonstrated that molecular midwives, simple :organic catalysts;, can order materials selectively via molecular recognition, and function to direct the growth of structural complexity. We have further demonstrated that kinetic trapping of products selected under equilibrium control enables one to achieve high-fidelity template-directed polymerization. We cannot yet define which molecular structures allowed life to take hold, but we can define the physical principles and chemical functions that were likely essential from the beginning, and remain critical to contemporary living systems. Our search for the principles that guided the formation of the first RNA-like polymers is carried out with continued feedback from laboratory experimentation.
CHEMISTRY & BIODIVERSITY – Vol. 4 (2007)
783
Albert Eschenmoser has stated that the origin of life cannot be discovered, it must be :reinvented; (personal communication by R. Krishnamurthy). We fully agree with this statement. Furthermore, in regards to the many theories put forth for the origin of the first RNA-like polymers, we feel that the proof of the pudding is in the eating. A tremendous limitation of many models for the origin of RNA (and the origin of life) is that they are almost impossible to verify or falsify, often due to a complete lack of chemical details. Experimental demonstration of progress towards the spontaneous assembly of polymers akin to RNA from plausible prebiotic building blocks should be the universal standard by which such theories are judged for merit. We thank the NSF (CHE-0404677), the NASA Exobiology Program (NNG04GJ32G), and DOE (ER15377) for financial support. We also thank Frank A. L. Anet and Aaron Engelhart for helpful discussions.
REFERENCES [1] :The RNA World: The Nature of Modern RNA Suggests a Prebiotic RNA World;, 3rd edn., Eds. R. Gesteland, J. F. Atkins, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2006. [2] N. V. Hud, F. A. L. Anet, J. Theor. Biol. 2000, 205, 543. [3] G. F. Joyce, L. E. Orgel, in :The RNA World: The Nature of Modern RNA Suggests a Prebiotic RNA World;, 3rd edn., Eds. R. F. Gesteland, J. F. Atkins, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, 2006, p. 23. [4] Y. Gat, D. G. Lynn, Biopolymers 1998, 48, 19. [5] X. Li, Z.-Y. J. Zhan, R. Knipe, D. G. Lynn, J. Am. Chem. Soc. 2002, 124, 746. [6] P. Ts;o, I. Melvin, A. Olson, J. Am. Chem. Soc. 1963, 85, 1289. [7] P. O. P. Ts;o, in :Basic Principles in Nucleic Acid Chemistry;, Ed. P. O. P. Ts;o, Academic Press, New York, 1974, p. 453. [8] P. Yakovchuk, E. Protozanova, M. D. Frank-Kamenetskii, Nucleic Acids Res. 2006, 34, 564. [9] M. A. Winnik, Acc. Chem. Res. 1985, 18, 73. [10] J. B. Mills, E. Vacano, P. J. Hagerman, J. Mol. Biol. 1999, 285, 245. [11] M. Smietana, E. T. Kool, Angew. Chem., Int. Ed. 2002, 41, 3704. [12] R. Saladino, C. Crestini, G. Costanzo, E. DiMauro, Curr. Org. Chem. 2004, 8, 1425. [13] H. Ihmels, D. Otto, Top. Curr. Chem. 2005, 258, 161. [14] E. Anders, R. Hayatsu, M. H. Studier, Science 1973, 182, 781. [15] J. D. Bernal, Proc. Phys. Soc. Lond. B 1949, 62, 597. [16] R. E. A. Kelly, L. N. Kantorovich, J. Mater. Chem. 2006, 16, 1894. [17] L. E. Orgel, Origins Life Evol. Biosphere 1998, 28, 227. [18] J. B. Chaires, Biopolymers 1997, 44, 201. [19] S. S. Jain, F. A. L. Anet, C. J. Stahle, N. V. Hud, Angew. Chem., Int. Ed. 2004, 43, 2004. [20] A. H. F. Lee, E. T. Kool, J. Am. Chem. Soc. 2006, 128, 9219. [21] A. H. F. Lee, E. T. Kool, J. Am. Chem. Soc. 2005, 127, 3332. [22] J. S. Ren, J. B. Chaires, Biochemistry 1999, 38, 16067. [23] D. Chandler, Nature 2005, 437, 640. [24] Z. Y. J. Zhan, D. G. Lynn, J. Am. Chem. Soc. 1997, 119, 12420. [25] P. Z. Luo, J. C. Leitzel, Z. Y. J. Zhan, D. G. Lynn, J. Am. Chem. Soc. 1998, 120, 3019. [26] J. D. Ye, Y. Gat, D. G. Lynn, Angew. Chem., Int. Ed. 2000, 39, 3641. [27] H. D. Bean, F. A. L. Anet, I. R. Gould, N. V. Hud, Origins Life Evol. Biosphere 2006, 36, 39. Received December 28, 2006
