closing the circle: replicating rna with rna

Closing the Circle: Replicating RNA with RNA

Leslie K.L. Cheng and Peter J. Unrau

Simon Fraser University, 8888 University Drive, Burnaby, BC. V5A 1S6, Canada

Correspondence: [email protected]

How life emerged on this planet is one of the most important and fundamental questions ofscience. Although nearly all details concerning our origins have been lost in the depths oftime, there is compelling evidence to suggest that the earliest life might have exploited thecatalytic and self-recognition properties of RNA to survive. If an RNA based replicatingsystem could be constructed in the laboratory, it would be much easier to understand thechallenges associated with the very earliest steps in evolution and provide importantinsight into the establishment of the complex metabolic systems that now dominate thisplanet. Recent progress into the selection and characterization of ribozymes that promotenucleotide synthesis and RNA polymerization are discussed and outstanding problems inthe field of RNA-mediated RNA replication are summarized.

Cell division is a fundamental biologicalprocess in which genetic information is

duplicated and shared between daughter cells.In extant cellular life, DNA serves as the reposi-tory of genetic information, but its replication iscomplicated by the daunting size and complexstructural organization of modern genomes.For this reason, multiple enzymes are requiredto ensure faithful genomic replication in allhigher life forms. Notably, simpler replicatingsystems such as viruses, have smaller genomesand tend to use correspondingly more error-prone replicative machinery (Kunkel and Bebe-nek 2000; Gago, Elena et al. 2009). Presumably,if the initial organisms on this planet also hadsmall genomes, then the earliest genomic repli-cation could have been a relatively simple and

error-prone process compared with the com-plex replicative strategies of modern life.

ABC of Life: Abiotic to Biotic Chemistry, theEmergence of the RNA World

Modern biology is built up from a dense web ofchemical reactions that are maintained by thecatalytic reactions of hundreds of enzymes.These catalysts are all one dimensional, aperi-odic polymers, built from protein or nucleicacids. Such polymers are able to adopt a diverserange of complex three-dimensional folds andby virtue of their linear sequence are simpleto encode. A particular advantage of these bio-logical polymers is that each polymer type isdefined by a small set of monomers that can be

Editors: David Deamer and Jack W. Szostak

Additional Perspectives on The Origins of Life available at www.cshperspectives.org

Copyright # 2010 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/cshperspect.a002204

Cite this article as Cold Spring Harb Perspect Biol 2010;2:a002204

1

polymerized by a single self-consistent chemis-try. This type of polymer construction empow-ers evolution with a mechanism to rapidly adaptexisting polymer folds to new functions by themutation/recombination of polymer sequence.Based on the intrinsic simplicity of modern bio-logical polymers, the earliest biological systemsappear overwhelmingly likely to have beenpolymer based as well (Joyce 2002).

The emergence of the earliest biological rep-licating systems must have required consider-able abiotic chemical organization. Such abioticchemistry would not have immediately disap-peared upon the emergence of life and wouldhave provided chemical “sustenance” to nurturethe increasing metabolic complexity of emer-gent life as it evolved. As a result if the first bio-logical polymer can ever be reliably identifiedit would place an important constraint on theabiotic environments possible on the early Earth.This polymer would be the bridge between theworld of abiotic chemistry and the biotic worldof enzymatic reactions.

Based on this logic, the earliest replicativepolymers should satisfy three primary condi-tions: (1) The initial polymers should have anintrinsic mechanism to facilitate their replica-tion either abiotically or biotically. Ideally thepolymer’s monomer subunits should be ableto be polymerized by a single uniform chemis-try. (2) Abiotically, monomers should be easilysynthesized and this synthesis should be com-patible with abiotic polymerization. Abioticpolymerization from abiotically synthesizedmonomers would have presumably resulted inthe synthesis of more monomer than polymerand would provide the earliest replicating sys-tems with a source of monomers before theevolution of a biological metabolism. (3) Theresulting polymers should intrinsically be en-dowed with the ability to promote a broad rangeof chemistry. In particular the metabolic syn-thesis of monomer units from abiotic sourcesshould be tenable using the polymers them-selves as catalysts. This monomer synthesiswould serve as the basis for one of the earliestmetabolic reactions associated with replication.

RNA is the simplest aperiodic polymer thatbiologically and chemically has been shown to

satisfy the first and last of these three conditions.RNA is comprised of four distinct monomers,which can form a double-stranded RNA helixby simple and predictable pairing rules. Theability to form a regular homoduplex providesa fundamental mechanism for the templatedreplication of RNA strands and is used by simpleRNA systems such as viruses to replicate.Equally important, RNA can fold into a varietyof complex three-dimensional shapes that pro-mote a broad range of chemical reactions(Wilson and Szostak 1999; Ellington, Chenet al. 2009). Biologically, RNA plays a funda-mental role in modern life being responsiblefor ribosomal translation of mRNA into protein(Nissen, Hansen et al. 2000), RNase P and tRNAmaturation (Pannucci, Haas et al. 1999; Mar-quez, Chen et al. 2006), riboswitches and generegulation (Tucker and Breaker 2005), and as atemplate for the extension of telomeres by telo-merase (Qiao and Cech 2008). The ribosometogether with other critical biological RNA cat-alysts, such as RNase P, has been interpreted as arelic of an “RNA World,” where RNA and notprotein was the dominant catalyst (Crick1958; Gilbert 1986; Orgel 2004). This evidencetogether with the metabolic importance of RNAand the nucleotide cofactors across biology(White 1976; Benner et al. 1989), strongly suggeststhat RNA played an important role early in evolu-tion and that it might have been involved in thevery first autocatalytic reactions.

Abiotic Nucleotide Synthesis

Until recently a primary difficulty in declaringRNA the earliest biological polymer has beenthe absence of convincing evidence that acti-vated RNA monomers can be produced byabiotic processes. This has caused some to spec-ulate that another even simpler polymer pre-ceded RNA in the early evolution of life(Orgel 2004; Joyce and Orgel 2006). Althoughit has been known for some time that the purineand pyrimidine nitrogenous bases (Robertsonand Miller 1995; Zubay and Mui 2001; Hilland Orgel 2002) and ribose sugar (Ricardoet al. 2004; Gesteland et al. 2006) required toconstruct ribonucleosides can be synthesized

L.K.L. Cheng and P.J. Unrau

2 Cite this article as Cold Spring Harb Perspect Biol 2010;2:a002204

separately from plausible prebiotic compounds,the creation of a glycosidic linkage has beenproblematic. Heating purine bases togetherwith ribose in dehydrating conditions leads tothe production of nucleosides in low yield(Fuller et al. 1972) but the synthesis ofpyrimidine nucleotides has proven difficult(Orgel 2004).

In 2009, Powner et al. elegantly demon-strated that pyrimidine ribonucleotides couldbe formed from the same precursor moleculesused to make a pyrimidine and ribose, with theexception that these molecules first be reactedto generate an intermediate, 2-aminooxazole.This clever approach bypasses the need to syn-thesize a glycosidic linkage and produces a rea-sonable yield of a pyrimidine 20,30-cyclicphosphate nucleotide monomer (Powner et al.2009). It is currently unknown how purinemonomers could be efficiently synthesized byabiotic mechanisms. Nevertheless, if prebioticroutes for both pyrimidine and purine nucleo-tides can be found, then the abiotic synthesis ofRNA polymers via their condensation (Ferriset al. 1996; Monnard et al. 2003) or via the effi-cient polymerization of 30-50 cyclic nucleotides(Costanzo et al. 2009) might be sufficient totrigger the emergence of an RNA based replicat-ing system. Such additional experimental evi-dence would remove the primary objections toan RNA early hypothesis and is an exciting areaof current research (Ricardo and Szostak 2009).

TOWARD A REPLICATING SYSTEM:NUCLEOTIDE SYNTHESIS BY RNA

Prebiotic sourcesofnucleotides wouldhave beenquickly used by a rapidly growing population ofribo-organisms and would have presented abottleneck to early evolution. To flourish, earlyRNA based life would have needed to developthe ability to synthesize nucleotide monomersfrom some more abundant supply of prebioticmaterial. Based on the chemistry of nucleotidesynthesis in modern metabolism it would besatisfying if these abundant prebiotic com-pounds included ribose and the nucleotide bases(Joyce 1989; Robertson and Miller 1995; Orgel1998). If so a solid abiotic chemical framework

would exist for the evolution of RNA-mediatednucleotide synthesis reactions that resemblethose found in modern metabolism.

Pyrimidine nucleotides and purine nucleo-tides (via salvage pathways) are synthesized bythe formation of a glycosidic linkage using phos-phoribosyl 1-pyrophosphate (PRPP) and theappropriate nucleobase. Given the fundamentalimportance of this chemistry to modern metab-olism and its potential compatibility with abio-tic supplies of ribose and nucleobases (Lau andUnrau 2009), the ability of RNA to mediate suchchemistry serves as a potential bridge betweenthe abiotic and biotic RNAWorld of nucleotidesynthesis. However, the chemistry of glycosidicbond formation presents a number of hurdlesfor RNA catalysts. Relative to protein, RNA hasrelatively few functional groups and the nucleo-tide monomers from which an RNA catalystmust be built are considerably bulkier. Further,the chemistry of nucleotide synthesis is stronglyinfluenced by nucleobase composition. Pyri-midine nucleotides are thermodynamically andkinetically much more difficult to form thanpurine nucleotides (Lau et al. 2004) makingnucleotide synthesis by RNA an interesting en-zymatic challenge.

One way to evaluate the ability of RNA tomediate such chemistries is to artificially selectand evolve ribozymes in the laboratory usingin vitro selection (Ellington and Szostak 1990;Robertson and Joyce 1990; Tuerk and Gold1990). We have used this approach to isolateboth pyrimidine nucleotide and purine nucleo-tide synthase ribozymes that are able to pro-mote the formation of a glycosidic linkagebetween a base (4-thiouracil and 6-thioguanine,respectively) and tethered PRPP (Unrau andBartel 1998; Lau et al. 2004). As expected basedon the difficulty of pyrimidine glycosidic bondformation, purine nucleotide synthases weremuch more abundant in sequence space andwere 50–100 times more efficient than theirpyrimidine synthase counterparts. Interestinglypyrimidine nucleotide synthase ribozymes usecharge stabilization to promote glycosidic bondformation (Unrau and Bartel 2003) presumablyby the precise positioning of the ribozyme’sphosphodiester backbone (Dinner et al. 2001).

Replicating RNA with RNA

Cite this article as Cold Spring Harb Perspect Biol 2010;2:a002204 3

Recently we made the interesting discovery thata ribozyme selected for its ability to mediatechemistry between ribose and 6-thioguanosinecould also promote purine nucleotide synthesiswhen the ribose substrate was substituted withPRPP (Lau and Unrau 2009). This promiscuousribozyme suggests that metabolically relevantribozymes making use of a small metabolite suchas ribose could have easily evolved to promotenucleotide synthesis with PRPP. Together withour purine and pyrimidine nucleotide syn-thases, this in vitro evidence strongly suggeststhat RNA folds able to promote nucleotide syn-thesis would not have been difficult to discoverearly in the evolution of a RNAWorld.

Satisfyingly, both purine and pyrimidinenucleotide synthase ribozymes were able torobustly discriminate between even quiteclosely related nucleobase substrates. The familyA pyrimidine nucleotide synthase showed amarked preference for 4-thiouracil over anyother pyrimidine tested (including uracil),whereas the purine nucleotide synthases reactedtwo to three orders of magnitude slower when6-thiopurine was substituted for 6-thioguanine(Unrau and Bartel 1998; Lau et al. 2004). Thisability to accurately distinguish small substratesis a hallmark of the metabolic enzymes and pro-vides encouraging evidence that in an earlyRNA World metabolically relevant ribozymeswould have been able to efficiently distinguishand hence regulate the synthesis of importantsmall molecule metabolites essential to life.

RNA REPLICATION: RNA POLYMERASESAND LIGASES

A RNA-mediated metabolism could only havebeen sustained if the RNA catalysts required tosustain metabolism were produced or repairedfaster than they degraded. Although RNA repairis relatively uncommon in modern metabolism(Chan, Zhou et al. 2009), there exist numerousexamples of RNA modification and editing thatmake use of short guide RNAs to promotechemistry at specific sites within a target RNA(Kiss 2002; Madison-Antenucci, Grams et al.2002; Matera, Terns et al. 2007). Guide RNAs,which generally speaking are RNAs having the

reverse complement of some target sequence,could have been very useful early in evolutionand might have directed RNA ligase ribozymesto specifically ligate together short RNA frag-ments in a sequence specific fashion. This wouldallow the construction of larger and morecomplex metabolically useful ribozymes fromsimpler RNA elements. If in turn these RNAfragments and guide sequences were tran-scribed from short RNA “genomic elements”this would not only allow the replication of arelatively complex, but fragmented RNA ge-nome, it would only require two replicative ac-tivities: RNA ligation and a RNA polymerasecapable of transcribing and replicating theshort RNA genomic elements. As the phospho-diester chemistry of RNA ligation and poly-merization are related (Fig. 1) their emergencefrom an abiotic system could be evolutionarilylinked and provide a simple route to biologicalreplication.

Cross Catalytic Ligation Strategies

An ingenious continuous molecular evolutionapproach, developed by Kim and Joyce (Kimand Joyce 2004) and further optimized byLincoln and Joyce (Lincoln and Joyce 2009)has made possible the study of cross-catalyticsystems built entirely of RNA. In this ribozyme-based system an RNA ligase ribozyme hy-bridizes to two oligonucleotide substrates andspecifically catalyzes their ligation (Fig. 1A).The ligated RNA results in a second ligase ribo-zyme that in turn can hybridize to two otheroligonucleotide substrates catalyzing in turnthe production of the first ribozyme. This cross-catalytic system was found to promote thecontinuous exponential doubling of RNA. Asimilar system was developed and optimizedby the Lehman laboratory, which divided theAzoarcus Group I ribozyme into four partsand made use of its intrinsic ligation capabilityto autocatalytically reassemble the ribozyme(Hayden and Lehman 2006; Draper et al. 2008).These fascinating systems all make use of guidesequences built into the RNA catalyzing liga-tion. If in the future these systems can be engi-neered to use generic guide RNAs supplied in



trans these systems would have tremendous invitro evolutionary potential and make possiblethe synthesis of arbitrary RNA componentsfrom a set of short RNAs and their correspond-ing guide sequences.

Althoughthese systemshaveelegantlyshownthe potential for RNA to replicate without theaddition of other components and provideinsight into how a complex multicomponentRNA world might function, these experimentsdo not address how the RNA fragments beingligated together are themselves produced.Returning to our initial argument, if the emer-gence of life rapidly consumed abioticallygenerated RNA fragments it might be expectedthat early in evolution ligation strategies couldhave evolved hand in hand with ribozymepolymerases able to transcribe short nucleotidesequences that ultimately become substrates forligation. In this model, as the RNAWorld became

more established and complex, improvementsin RNA polymerization would make possiblethe gradual concatenation of RNA genomic ele-ments into longer and longer elements makingthe importance of RNA ligation less and lessimportant as overall genome size grew.

Polymerization from Short RNA GenomicElements

Although nature abounds with replicative strat-egies, nucleic acid based replication alwaysinvolves the synthesis of sense and antisensestrands following the canonical base pairingrules of nucleic acid. The fundamental symme-try of this copying process—where sense standis copied to antisense to sense ad infinitum—is always broken in biological systems with thesynthesis of one stand being elevated over theother so as to allow gene expression. This is

Ligase ribozyme

Ligase ribozyme′

Ligase ribozyme′

Ligase ribozyme

A BRz′-1

RNA primer

RNA primer

RNA template

RNA template

RNA duplex

RNA polymeraseribozyme

RNA polymeraseribozyme

Rz-1 + Rz-2

Rz-2

Rz′-2ÖH5′

5′

5′

5′

5′

5′

5′

5′ 3′

3′

5′

5′

5′

5′

3′5′

– –pppÖH

OHNTPs

NTPs

–

ÖH–

– –

5′ 3′OH– –

ppp

ppp

–– ppp

ppi

ppi

ppi

ppi

etc…

ppi

Rz-1

Rz′-1 +Rz′-2

ÖH

Figure 1. RNA-catalyzed polymerization and cross-replication by ligation. (A) Cartoon schematic ofcross-replication of RNA ligase ribozymes: A ligase ribozyme (colored in two shades of green) catalyzes theligation of two orange oligonucleotides (Rz0-1 and Rz0-2) to generate a ligase ribozyme that catalyzes theligation of two green oligonucleotides (Rz-1 and Rz-2) to regenerate the first ligase ribozyme. (Adapted fromLincoln and Joyce 2009.) (B) Cartoon schematic scheme of ribozyme-catalyzed RNA polymerization. Thecomplete extension of an RNA primer (pink) according to the sequence of a template (blue) by an RNApolymerase produces an RNA duplex.



most clear in the highly evolved organismswhere dsDNA is the genetic material andmRNA is transcribed, but it is also true for thesimplest homopolymeric systems, such as theRNA based plant viroids, where more plusstrand is expressed during rolling circle replica-tion than the minus. Notably in the case of theplant viroids this asymmetry does not dependon translation but only on the functional as-pects of the expressed RNAs (Soll et al. 2001).In an early RNA world such asymmetry wouldhave been equally important for the stableexpression of RNA based metabolic enzymes.If such “transcriptional” asymmetry could becombined with the replication of short RNAgenomic elements, then together with the liga-tion strategies just discussed a self-consistentsystem of RNA replication could be constructedthat simultaneously allows genomic replicationand expression of metabolic ribozymes.

We favor a replicative model whereby theearliest genetic components were short dsRNAs.Although not fundamental for replication,

dsRNA has a number of interesting virtues.First and in contrast to single-strand or foldedRNAs the RNA duplex has a uniform andpredictable double helix that makes it easilyrecognizable as a genetic component by replica-tive enzymes. Second, it is intrinsically difficultto copy single-stranded RNA without creatinga RNA duplex. Thus generation of a reversecomplement strand from a single-strandedRNA genome requires some structural mecha-nism to avoid the genome becoming double-stranded by default. Although such strategiesare possible and are used by degenerate RNAsystems such as RNA-X (Konarska and Sharp1989) it is unclear how such strategies couldbe generalized to arbitrary sequence (Bartel1999). A RNA polymerase ribozyme able todifferentially transcribe from both strands of ashort RNA duplex could however producedouble-stranded genomic copies and an excessof single-stranded RNA that can be ligated bysubsequent steps into functional RNA compo-nents (Fig. 2).

Guide element 1/2 Genome element 2Genome element 1

5′ 5′ 5′

5′ 5′ 5′

5′

5′

5′

5′

5′

5′

5′5′

5′

5′+ +

+

5′

A

B

C

Figure 2. Genomic elements, their asymmetrical transcription resulting in genomic replication and synthesis offunctional RNAs by ligation. (A) Short double-stranded elements define the total RNA genome together withguide sequences important for the later synthesis of full length functional RNAs. (B) Asymmetrical transcriptionfrom either strand of the duplex genomic element results in the synthesis of multiple genomic element copiestogether with an excess of one strand. This asymmetry could be generated by having one end of the genomicelement fray with a higher propensity than the other. (C) Excess single-stranded transcripts can in turnhybridize to guide element sequences making them substrates for ligase enzymes and allowing the synthesisof long highly functional RNAs.



RNA POLYMERIZATION: EVOLVINGRNA LIGASES

The advent of in vitro selection provided amechanism to directly select template directedRNA ligase ribozymes. RNA ligation, whichinvolves the nucleophilic attack of a 30-hydroxylfrom one substrate onto the 50-triphosphate of asecond, is identical to the chemistry required forRNA polymerization (Fig. 1). Initial efforts inthe 1990s led to the isolation and structuralcharacterization of the highly efficient Class Iligase ribozyme (Bartel and Szostak 1993;Ekland and Bartel 1995; Ekland et al. 1995).Different classes of ligase ribozymes were subse-quently isolated by in vitro selection, includingthe hc ribozyme and its variants isolated fromthe Tetrahymena group 1 intron (Jaeger, Wrightet al. 1999; Yoshioka, Ikawa et al. 2004) andsmaller ligases such as the L1 ligase (Robertsonand Ellington 1999; Robertson and Scott 2007).Unfortunately, these ribozymes were all muchslower than the class I ligase implying thathighly efficient ligases might be rare in sequencespace. Recently however, the DSL ligase (Ikawa,Tsuda et al. 2004; Voytek and Joyce 2007) wasevolved to rival the catalytic efficiency of theclass I ligase. This second example of an efficientRNA ligase ribozyme strongly suggests that adiverse range of ligase ribozymes could havebeen available early in evolution. Further char-acterization of these new ligases should general-ize our understanding of RNA mediatedphosphodiester bond formation and by virtueof their identical chemistry provide insightinto the function of RNA polymerases.

Template-Directed RNA Polymerization:Evolving the Class I Ligase

Surprisingly, Ekland and Bartel showed that avariant of the class I ligase ribozyme could besimply engineered to extend a primer by sixnucleotide incorporations after a four day incu-bation (Ekland and Bartel 1996). This template-mediated extension required that the primeritself be hybridized to the ligase ribozyme. Invitro selection was used to overcome this prob-lem and led to the isolation of the Round 18

ribozyme (Johnston, Unrau et al. 2001). ThisRNA polymerase ribozyme contains a 76-nucleotide accessory domain selected from arandom sequence pool that was appendedonto the 30 end of the class I ligase ribozyme(now called the ligase core). The Round 18 ribo-zyme was capable of using nucleotide tri-phosphates in a template-dependent mannerto extend 14 nucleotides of a trans RNA primer-template in 24 hours (i.e., see Fig. 1B). The engi-neering of the Round 18 ribozyme from the classI ligase directly demonstrates that polymeriza-tion can be evolved from RNA ligation and hasyet to be shown for any other RNA ligase family.

The selection of the Round 18 ribozymerequired that the RNA primer being extendedby the polymerase be covalently attached tothe ribozyme so as to maintain a correlationbetween RNA function (polymerization) andphenotype (RNA sequence). To overcome thislimitation and allow a true trans selection forRNA polymerization, we used a compartmen-talized approach (Fig. 3) to select for an im-proved variant of the Round 18 ribozymecalled B6.61. This polymerase is able to extenda trans primer-template duplex by 20 nucleo-tides and has improved fidelity relative to theRound 18 ribozyme (Zaher and Unrau 2007).In the selection for B6.61, a diverse DNA poolcontaining approximately 9 �1014 variants ofthe Round 18 ribozyme was ligated to an RNAprimer-template complex. These tagged DNAgenomes were encapsulated into water-in-oilvesicles (Tawfik and Griffiths 1998; Milleret al. 2006) where they were transcribed by T7RNA polymerase. By this strategy RNA pheno-types were confined together with their DNAgenotype. If an expressed RNA could extendthe primer-template, enrichment of the corre-sponding DNA genome was made possible. Bymimicking natural selection, the in vitro encap-sulated selection of B6.61 allowed for a truetrans correlation between the genotype andphenotype. Similar approaches might in futureallow the isolation of RNA systems that workcooperatively in order to survive within a com-mon compartment and allow the developmentof artificially evolving RNA systems (Barteland Unrau 1999).



The Evolutionary Power of ConstructingModular RNA PolymerasesProtein polymerases are modular enzymes thathave clearly delimited functional folds (Werner2007). This is seen most clearly in the well stud-ied DNA dependent RNA polymerases whereinitiation requires the static recognition of adsDNA promoter element (by the s-factors inbacterial RNAPs) and where elongation requiresa structural rearrangement of the polymerase soas to produce a DNA–RNA heteroduplex thatmoves within the DNA transcription bubbleto ensure processive elongation (Yin and Steitz2002; Mooney et al. 2005). Although the B6.61polymerase cannot rival the complexity ofsuch beautiful machines, it does compare withthe much less processive, primer dependentDNA repair enzymes such as Taq polymerase.Interestingly, experiments with this commonlyused polymerase have found that appending

a nonspecific DNA binding protein to theenzyme can dramatically increase its processiv-ity (Wang et al. 2004). It might therefore bepossible to enhance the ability of the B6.61polymerase to extend long templates by addingadditional RNA domains that enhance the abil-ity of the polymerase to nonspecifically bindRNA duplex. Such modularity is extensivelyfound in large biological RNAs such as the ribo-some where the functions of peptide bond for-mation and mRNA decoding are apportionedbetween the large and small subunits. Similarly,the RNase P RNA can be dissected into twomajor submotifs, one responsible for substraterecognition and the other catalysis (Krasilnikovet al. 2004). Presumably the evolution of suchmodularity would have driven the early emer-gence of new function and is an importantelement in understanding complex molecularmachines.

Elute, PCR reamplify;ligate primer anneal template

Streptavidin

Recover nucleic acidand dehybridize template

Encapsulateinto vesicles

Gel purify,hybridize biotinylated oligo

4SU

4SUTP + NTPs

4SUTP + NTPs

4SU

APM gel

T7PT7P

T7P

T7RNAPT7RNAP

T7P

T7P

B

T7P4SU

Figure 3. In vitro encapsulated selection scheme for trans acting RNA polymerase ribozymes. A DNA pool wasgenerated that contained a T7 promoter allowing transcription of mutagenized Round 18 ribozyme sequencelibrary. After ligating a RNA primer (orange)–template (green) complex to the DNA pool, genomes wereencapsulated and RNA transcribed by T7 RNAP. Active RNA polymerase ribozymes extend the RNA primertethered to their DNA genome and in the process incorporate 4-thiouridine residues in the growing strand.This allows selection of functional genomes using thiol-sensitive mercury gels and hybridization-basedcapture using biotinylated oligonucleotides. The captured DNA was then PCR amplified and used in afurther round of selection (Zaher and Unrau 2007 and reprinted here with permission by the author).



The laboratory evolution of the B6.61 ribo-zyme models this incremental expansion in en-zyme functionality and was recently exploredby characterizing the two domains of the poly-merase. The ligase core and accessory domainof B6.61 are modular domains that fold inde-pendently, yet act cooperatively, to extend atrans primer-template substrate (Cheng et al.,

unpubl). Both domains fold well in trans anddo not, based on chemical probing, appear tochange their folds when transcribed separately.Notably nucleotide incorporation was com-pletely abolished by removal of the accessorydomain, but as shown in Figure 4A, addingthe ligase and accessory domain in trans re-sulted in polymerization (last 5 lanes), albeit

B6.61

B6.61

765

432

1

*P16

765

432

1

*P16

10 m

in.

10 m

in.

10 m

in.

10 m

in.

50 m

in.

50 m

in.

50 m

in.

250

min

.

250

min

.

10 m

in.

50 m

in.

250

min

.

250

min

.

10 m

in.

60 m

in.

60 m

in.

3 hr

.9

hr.

9 hr

.3

hr.

23 h

r.

23 h

r.

25 h

r.

25 h

r.

25 h

r.

25 h

r.

L.1+A.1

L.1+A.1

0tL.1:tA.1

0tL.1:tA.1

79tL.1:tA.1 106tL.1:tA.1

44tL.1:tA.1

44tL.1:tA.1 79tL.1:tA.1 106tL.1:tA.1

RCtagTag

RCtag

Tag

Tag

RC

tag

RCtag

Tag

A

C

B

Figure 4. RNA polymerase ribozyme modularity: The B6.61 Ligase core and accessory domain in a range ofequimolar (0.5 mm) contexts. (A) Comparison of polymerization activity of trans bimolecular constructs(L.1 þ A.1) with that of B6.61. (B) Polymerization activity assay of different assemblies of the two hybridizedtrans bimolecular constructs as shown in panel C. (C) Cartoon schematic of the four bimolecular assemblies.



�100-fold slower than unimolecular construct(first 5 lanes). Remarkably the two domainscan be joined together by a range of hybridiza-tion sites some of which retain nearly wild-typelevels of activity (Fig. 4 B,C). Orientationappears crucial based on the existence of a spe-cific cross-linking pattern between the twodomains that suggest that weak tertiary interac-tions are responsible for the correct positioningof the two domains with respect to each other.Consistent with the three-dimensional struc-ture of the class I ligase ribozyme (Bergmanet al. 2004; Shechner et al. 2009), the 50 and 30

tethering sites are close together and on thesame side of the ligase core as the active site,whereas tethering the accessory domain to thetwo internal loops of the ligase core that aredistal to the active site lowers polymerizationactivity. This is consistent with the accessorydomain being positioned over the class I li-gase active site so at to constrain the primer–template and presumably nucleotide triphos-phates in the vicinity of the enzyme active site.Most interestingly, a ten-fold increase in poly-merization rate was observed by tethering theprimer-template to the polymerase. As this in-crease was dependent on the accessory domainbeing present, improvements in the accessorydomain or the selection of new complementaryfunctional modules might be expected to fur-ther improve RNA polymerase.

FUTURE RESEARCH DIRECTIONS: STRANDDISPLACEMENT AND INITIATION

The B6.61 polymerase fills in a primer-templateto produce a RNA duplex. Currently, thisdsRNA product is not a substrate for the poly-merase, making it impossible to construct a rep-licating system of the sort sketched in Figure 2.If, however, this polymerase could be evolved torecognize and transcribe a RNA duplex, then aself-consistent RNA system could be con-structed that requires only nucleotide triphos-phates to replicate. Two critical challengesremain to be achieved: First and most problem-atic, a mechanism must be developed to allowthe initiation of polymerization from a duplexRNA. Second a mechanism to expose the

template strand of the duplex must be engi-neered. These two functionalities would allowthe polymerase ribozyme to transcribe onestrand of the duplex, resulting in a total of threeRNA strands, one of which is by necessity mustbe single-stranded (Fig. 2). Because this ssRNAcan hybridize to an antisense strand producedby transcription in the opposite direction asimple and effective mechanism for copyinggenomic elements would be built into thisform of transcription. Asymmetric transcrip-tion from any particular genomic elementwould therefore not only provide a mechanismto copy the RNA genome but would naturallyprovide a source of ssRNA that could be ligatedtogether to form larger more complex func-tional RNAs (Fig. 2).

RdRP Initiation: A Protein Point of View

Biologically, there are two main types (Nget al. 2008) of initiation: primer-independent(de novo synthesis) and primer-dependent. Asshown in Figure 5A, de novo synthesis beginswith the base pairing of an initiation nucleotidetriphosphate (often this is GTP, Ng et al. 2008)to the 30 end of the RNA template. Because thisinteraction is not sufficient to stabilize the sin-gle base pair thus formed, other interactionssuch as base stacking of aromatic protein resi-dues with the initiation nucleotide are typicallyemployed (Butcher et al. 2001). The extensionof this first nucleotide by additional templatednucleotides can then trigger a rearrangementof protein structure that allows the formationof a stable elongation complex via an abor-tive cycling process (Ng et al. 2008). Primer-dependent mechanisms of initiation use ahydroxyl from a nearby source for nucleophilicattack on the triphosphate of the incomingnucleotide. Figure 5B–D depict the mechanismemployed by three viral RdRPs in which a freehydroxyl is derived from: (1) a nearby proteinresidue (Paul et al. 1998), (2) a short oligo-nucleotide originating either from abortivecycling (McClure 1985) or from a cleavedmRNA (Hagen et al. 1995), or (3) by foldingback a RNA template so as to produce a hairpinand a free 30-hydroxyl that can be extended by a



suitable polymerase (Laurila et al. 2002; Laurilaet al. 2005).

In all of these examples the template strandmust first be exposed so that a short primer caneither be synthesized (primer-independent) orprovided by an external source (primer depend-ent). While transcribing from a RNA duplex itwould be essential that the duplex be firstmelted by some form of helicase activity. Thereare several ways to design a solution to this prob-lem using RNA, although none appear partic-ularly straightforward. The simplest approachmight be to make use of the intrinsic propertiesof RNA to pry open the end of the duplex. Thiscould be achieved by first making a temporaryand reversible anchor to the 30 terminus of thetemplate strand that would ultimately orientthe template strand correctly with respect tothe polymerase’s active site. This ligation mightbe expected to mildly destabilize the terminalduplex and could be dramatically enhanced byappending a binding motif that specifically rec-ognizes either the 50 triphosphate on the RNAduplex or equally a 50 cap (Huang and Yarus1997; Zaher et al. 2006). Correctly positionedthis combination of a covalent anchor on the

30 terminus of the template strand together withtight binding to the 50 terminus would serve tounwind the RNA duplex in a fashion sufficientto allow either primer independent or depend-ent initiation. After several nucleotide incorpo-rations the polymerase structure would change,induced by forces acting through the site ofRNA ligation. This change in geometry wouldthen mediate changes in the ligation site causingRNA cleavage to be favored instead of ligationand triggering the release of the template strandfrom the polymerase (potentially the polymer-ase could use a cyclic phosphate to mediate thisinteraction in analogy with the reversible liga-tion and cleavage mediated by the hammerheadribozyme). This same structural change wouldhopefully distort the geometry of the 50 bindingmotif, releasing the emerging single strand andallowing a functional elongation complex toform.

Solving the Strand Displacement Problem

The initiation just described would not lead toa transcriptional complex that is directly an-alogous to a biological transcriptional complex.

Initiationnucleotide

A

pppHO: HO

ppp

O

G G

C C U G A

O

Op p p p p…

O O O O 5′3′

3′

Incomingnucleotide

RNA template

B

HO: :O

ppp

H

O

U

A A A A A

VPg 5′-terminal protein

Incoming nucleotide

3′ Op…p

Op

Op

O Op

5′

Oligonucleotide fromabortive cycling

C

5′-capped oligonucleotidefrom cleaved mRNA

OR

ppp

…pHO: HO

Incomingnucleotide

O O

G A

C U U3′

5′

RNA template

O O O…p p… 5′p p

3′poly(A) RNA template

DHO

Incomingnucleotide

pppHO:

…p

O3′

O

G A

C U U

…p p… 5′p pO O O

RNA template

…

YT

T

GLPP

IRT

PN NVKK

…

Figure 5. Comparison of initiation mechanisms. (A) Primer-independent (de novo) initiation B, C, and D—Primer-dependent initiation strategies: (B) “Borrowing” a hydroxyl from a nearby protein residue (C) Use ofa short oligonucleotide from abortive cycling in de novo initiation or from a cleaved mRNA; (D) Templatefolds back to form a stable hairpin that is then extended. Adapted from (Paul et al. 1998; van Dijk et al. 2004and reprinted with permission from The Journal of General Virology #2004; Ng et al. 2008 and reprintedwith express permission from the authors).



DNA dependent RNAPs form a stable tran-scription bubble complex (Yin and Steitz2004; Mooney et al. 2005; Borukhov and Nudler2008) that makes use of a RNA exit tunnel toprecisely position a RNA–DNA heteroduplexwith respect to the polymerase active site. Atranscription bubble where the two downstreamstrands of DNA are reunited upstream of theRNA–DNA heteroduplex is only required fortranscription of mixed polymer types andmakes DNA dependent RNA transcriptionmore complex. If RNA is being transcribedfrom a RNA duplex simpler topological nucleicacid structures can be used and the freshly syn-thesized strand need not be displaced, butinstead can remain hybridized to the templatestrand. This change results in the original non-template strand being displaced and makeselongation simpler as a consequence.

Thermodynamically the free energy of thestates sampled by polymerization should benearly equivalent at any particular extension:

DG8(n,m)complex ¼ DG8(n,m)RNA-RNA

þ DG8(n,m)RNA-polymerase ¼ 0

where n and m are the nth and mth nucleotideextension, RNA–RNA represents the interac-tions formed in the substrate strands, and RNA-polymerase the interactions made specificallyby the polymerase with the substrate. As theformation of a base pair in the downstream prod-uct helix is on average compensated by the dis-ruption of an upstream base pair, DG8RNA-RNA

should remain small except for pathologicalsequences. Ideally DG8RNA-polymerase can bemade small as well by ensuring that the poly-merase is only topologically associated withsubstrate RNA strands and does not make anyspecific contacts with the replicating nucleicacid. This can be achieved by building RNAdomains onto the polymerase that are able tosurround but not bind to either the upstreamor downstream helixes. This clamp would pro-vide a mechanism to ensure that the polymerasealways stays associated with its template makingit much more processive (Fig. 6, Accessorydomain). Equally important the emerging

single-strand must be prevented from hybridiz-ing to the template strand of the duplex immedi-ately upstream of the site of nucleotideincorporation. A helix placed transverselybetween the template strand and the strandbeing displaced by polymerization mightachieve this aim provided it was rigidly anchored

A

B

C

5′

5′

5′

5′

5′

5′

5′

NTPs

NTPs

5′

5′

L

L

B

B

A

A

Figure 6. Evolving a processive strand-displacingRNA polymerase ribozyme from the B6.61 RNApolymerase. (A) After formation of an initiationcomplex, which exposes the template strand, a shortRNA primer sequence is either synthesized de novoor allowed to hybridize. (B) Elongation requires thatnucleotides immediately upstream of the site ofnucleotide incorporation are single-stranded.If interactions between the ligase core (L) andthe accessory domain (Circled A) can be improvedso that the freshly synthesized strand together withtemplate are held robustly and nonspecificallythe polymerase should become much moreprocessive. If in addition these two domains can berigidly connected to a third domain (Circled B) thatensures dehybridization of the incoming duplex,transcription from dsRNA as well as ssRNAtemplates would be possible. This elongationcomplex should remain invariant in shape as itslides along the template. (C) Transcription endswith the release of one strand from the originalduplex RNA and the duplex RNA being regeneratedwith a freshly synthesized RNA strand.



to the duplex sliding clamp (Fig. 6, thirddomain). Failure to achieve strand displacementwould prevent polymerization and is essential tosuccess. Similarly, improving the stability of theB6.61 accessory domain and ligase core by fur-ther engineering and in vitro selection couldwell result in a suitable duplex sliding clamp.This clamp could then be selected to functionwith the new strand displacement motif byselecting for polymerization from artificially pre-pared mimics of a replication complex (Fig. 7).

Selecting polymerases for strand displace-ment should be the first priority for improvingRNA dependent polymerase ribozymes. Such

an enzyme even if it lacks the ability to initiatecould still allow a number of very interestingexperiments. For example, with the appropriateprimers such a polymerase could mediate roll-ing circle replication of an RNA genome. Ifthis circular genome contained a hammerheadribozyme motif together with the polymeraseribozyme sequence, then primed rolling circletranscription of both plus and minus strandswould allow the creation of a simple autocata-lytic system requiring only primer and nucleoti-des to replicate. This would make possible theartificial continuous evolution of the RNA pol-ymerase itself and with suitable selection

Pool

Active variants

Active variants

5′

5′

5′

5′

5′

5′

5′

5′

5′

Figure 7. Selecting for strand displacement in an RNA polymerase ribozyme. A pool of potential stranddisplacing RNA polymerase ribozymes is tethered by a long flexible linker to a RNA (red strand) that ispartially complementary to the template strand (bottom black strand). The unhybridized nucleotides in thered strand are not able to hybridize to the template strand resulting in the formation of an elongationcomplex mimic. Transcription by the tethered polymerase that is able to displace the red strand results in thepolymerase disassociating from the extending primer template duplex (black strands). Freed polymerases canthen be differentially recovered and amplified.



pressure might rapidly lead to the evolution ofinitiation strategies that allow the controlledexpression of other potentially metabolicallyrelevant RNAs.

CONCLUSIONS

Based on a number of independent lines oflaboratory research RNA appears well suitedto have served as the first replicative polymeron this planet. More importantly the catalyticand genetic properties intrinsic to RNA mightin the near future make possible the first artifi-cial forms of life. Such systems would allow theexploration of many important and outstand-ing questions concerning the robustness andearly evolution of life and provide criticalinsight into the difficulty of evolving life else-where in the universe.

REFERENCES

Bartel DP. 1999. Re-creating an RNA replicase. In RNAWorld, 2nd ed. Cold Spring Harber Laboratory Press,Cold Spring Harber, NY.

Bartel DP, Szostak JW. 1993. Isolation of new ribozymesfrom a large pool of random sequences. Science 261:1411–1418.

Bartel DP, Unrau PJ. 1999. Constructing an RNA world.Trends Cell Biol 9: M9–M13.

Benner SA, Ellington AD, Tauer A. 1989. Modern meta-bolism as a palimpsest of the RNA world. Proc NatlAcad Sci 86: 7054–7058.

Bergman NH, Lau NC et al. 2004. The three-dimensionalarchitecture of the class I ligase ribozyme. RNA 10:176–184.

Borukhov S, Nudler EC. 2008. RNA polymerase: The vehicleof transcription. Trends Microbiol 16: 126–134.

Butcher SJ, Grimes JM, Makeyev EV, Bamford DH, StuartDI. 2001. A mechanism for initiating RNA-dependentRNA polymerization. Nature 410: 235–240.

Chan CM, Zhou C, Huang RH. 2009. Reconstituting bacte-rial RNA repair and modification in vitro. Science 326:247.

Costanzo G, Pino S, Cicirielli F, Di Mauro E. 2009. Gene-ration of long RNA chains in water. J Biol Chem 284:33206–33216.

Crick FH. 1958. On protein synthesis. Symp Soc Exp Biol 12:138–163.

Dinner AR, Blackburn GM, Karplus M. 2001. Uracil-DNAglycosylase acts by substrate autocatalysis. Nature 413:752–755.

Draper WE, Hayden EJ, Lehman N. 2008. Mechanisms ofcovalent self-assembly of the Azoarcus ribozyme from

four fragment oligonucleotides. Nucleic Acids Res 36:520–531.

Ekland EH, Bartel DP. 1995. The secondary structure andsequence optimization of an RNA ligase ribozyme.Nucleic Acids Res 23: 3231–3238.

Ekland EH, Bartel DP. 1996. RNA-catalysed RNA polymer-ization using nucleoside triphosphates. Nature 382:373–376.

Ekland EH, Szostak JW, Bartel DP. 1995. Structurally com-plex and highly active RNA ligases derived from randomRNA sequences. Science 269: 364–370.

Ellington AD, Szostak JW. 1990. In vitro selection of RNAmolecules that bind specific ligands. Nature 346:818–822.

Ellington AD, Chen X, Robertson M, Syrett A. 2009. Evolu-tionary origins and directed evolution of RNA. Int JBiochem Cell Biol 41: 254–265.

Ferris JP, Hill AR Jr, Lui R, Orgel LE. 1996. Synthesis of longprebiotic oligomers on mineral surfaces. Nature 381:59–61.

Fuller WD, Sanchez RA, Orgel LE. 1972. Studies in prebioticsynthesis. VI. Synthesis of purine nucleosides. J Mol Biol67: 25–33.

Gago S, Elena SF, Flores R, Sanjuan R. 2009. Extremelyhigh mutation rate of a hammerhead viroid. Science323: 1308.

Gesteland RF, Cech T, Atkins JF. 2006. The RNAWorld: Thenature of modern RNA suggests a prebiotic RNA world.Cold Spring Harbor Laboratory Press, NY.

Gilbert SD. 1986. The RNA world. Nature 319: 618.

Hagen M, Tiley L, Chung TD, Krystal M. 1995. The role oftemplate-primer interactions in cleavage and initiationby the influenza virus polymerase. J Gen Virol 76:603–611.

Hayden EJ, Lehman N. 2006. Self-assembly of a group Iintron from inactive oligonucleotide fragments. ChemBiol 13: 909–918.

Hayden EJ, von Kiedrowski G, Lehman N. 2008. Systemschemistry on ribozyme self-construction: Evidence foranabolic autocatalysis in a recombination network.Angew Chem Int Ed Engl 47: 8424–8428.

Hill A, Orgel LE. 2002. Synthesis of adenine from HCNtetramer and ammonium formate. Orig Life Evol Biosph32: 99–102.

Huang F, Yarus M. 1997. Versatile 50 phosphoryl coupling ofsmall and large molecules to an RNA. Proc Natl Acad Sci94: 8965–8969.

Ikawa Y, Tsuda K, Matsumura S, Inoue T. 2004. De novo syn-thesis and development of an RNA enzyme. Proc NatlAcad Sci 101: 13750–13755.

Jaeger L, Wright MC, Joyce GF. 1999. A complex ligase ribo-zyme evolved in vitro from a group I ribozyme domain.Proc Natl Acad Sci 96: 14712–14717.

Johnston WK, Unrau PJ, Lawrence MS, Glasner ME, BartelDP. 2001. RNA-catalyzed RNA polymerization: Accurateand general RNA-templated primer extension. Science292: 1319–1325.

Joyce GF. 1989. RNA evolution and the origins of life. Nature338: 217–224.



Joyce GF. 2002. The antiquity of RNA-based evolution.Nature 418: 214–221.

Joyce GF, Orgel LE. 2006. Progress towards understandingthe origin of the RNA world. In RNA World, 3rd ed.Cold Spring Harber Laboratory Press, Cold SpringHarber, NY.

Kim DE, Joyce GF. 2004. Cross-catalytic replication of anRNA ligase ribozyme. Chem Biol 11: 1505–1512.

Kiss T. 2002. Small nucleolar RNAs: An abundant group ofnoncoding RNAs with diverse cellular functions. Cell109: 145–148.

Konarska MM, Sharp PA. 1989. Replication of RNA by theDNA-dependent RNA polymerase of phage T7. Cell 57:423–431.

Krasilnikov AS, Xiao Y, Pan T, Mondragon A. 2004. Basis forstructural diversity in homologous RNAs. Science 306:104–107.

Kunkel TA, Bebenek K. 2000. DNA replication fidelity. AnnuRev Biochem 69: 497–529.

Lau MW, Unrau PJ. 2009. A promiscuous ribozyme pro-motes nucleotide synthesis in addition to ribose chemis-try. Chem Biol 16: 815–825.

Lau MW, Cadieux KE, Unrau PJ. 2004. Isolation of fastpurine nucleotide synthase ribozymes. J Am Chem Soc126: 15686–15693.

Laurila MR, Makeyev EV, Bamford DH. 2002. Bacterio-phage w6 RNA-dependent RNA polymerase: Moleculardetails of initiating nucleic acid synthesis without primer.J Biol Chem 277: 17117–17124.

Laurila MR, Salgado PS, Stuart DI, Grimes JM, BamfordDH. 2005. Back-priming mode of w6 RNA-dependentRNA polymerase. J Gen Virol 86: 521–526.

Lincoln TA, Joyce GF. 2009. Self-sustained replication of anRNA enzyme. Science 323: 1229–1232.

Madison-Antenucci S, Grams J, Hajduk SL. 2002. Editingmachines: the complexities of trypanosome RNA editing.Cell 108: 435–438.

Marquez SM, Chen JL, Evans D, Pace NR. 2006. Structureand function of eukaryotic Ribonuclease P RNA. MolCell 24: 445–456.

Matera AG, Terns RM, Terns MP. 2007. Non-coding RNAs:Lessons from the small nuclear and small nucleolarRNAs. Nat Rev Mol Cell Biol 8: 209–220.

McClure WR. 1985. Mechanism and control of transcrip-tion initiation in prokaryotes. Annu Rev Biochem 54:171–204.

Miller OJ, Bernath K, Agresti JJ, Amitai G, Kelly BT, Mastro-battista E, Taly V, Magdassi S, Tawfik DS, Griffiths AD.2006. Directed evolution by in vitro compartmentali-zation. Nat Methods 3: 561–570.

Monnard PA, Kanavarioti A, Deamer DW. 2003. Eutecticphase polymerization of activated ribonucleotide mix-tures yields quasi-equimolar incorporation of purineand pyrimidine nucleobases. J Am Chem Soc 125:13734–13740.

Mooney RA, Darst SA, Landick R. 2005. Sigma and RNApolymerase: an on-again, off-again relationship? MolCell 20: 335–345.

Ng KK, Arnold JJ, Cameron CE. 2008. Structure-functionrelationships among RNA-dependent RNA polymerases.Curr Top Microbiol Immunol 320: 137–156.

Nissen P, Hansen J, Ban N, Moore PB, Steitz TA. 2000. Thestructural basis of ribosome activity in peptide bond syn-thesis. Science 289: 920–930.

Orgel LE. 1998. The origin of life—a review of facts andspeculations. Trends Biochem Sci 23: 491–495.

Orgel LE. 2004. Prebiotic chemistry and the origin of theRNA world. Crit Rev Biochem Mol Biol 39: 99–123.

Pannucci JA, Haas ES, Hall TA, Harris JK, Brown JW. 1999.RNase P RNAs from some Archaea are catalytically active.Proc Natl Acad Sci 96: 7803–7808.

Paul AV, van Boom JH et al. 1998. Protein-primed RNA syn-thesis by purified poliovirus RNA polymerase. Nature393: 280–284.

Powner MW, Gerland B et al. 2009. Synthesis of activatedpyrimidine ribonucleotides in prebiotically plausibleconditions. Nature 459: 239–242.

Qiao F, Cech TR. 2008. Triple-helix structure in telomeraseRNA contributes to catalysis. Nat Struct Mol Biol 15:634–640.

Ricardo A, Szostak JW. 2009. Origin of life on earth. Sci Am301: 54–61.

Ricardo A, Carrigan MA, Olcott AN, Benner SA. 2004.Borate minerals stabilize ribose. Science 303: 196.

Robertson MP, Ellington AD. 1999. In vitro selection of anallosteric ribozyme that transduces analytes to ampli-cons. Nat Biotechnol 17: 62–66.

Robertson DL, Joyce GF. 1990. Selection in vitro of an RNAenzyme that specifically cleaves single-stranded DNA.Nature 344: 467–468.

Robertson MP, Miller SL. 1995. An efficient prebiotic syn-thesis of cytosine and uracil. Nature 375: 772–774.

Robertson MP, Scott WG. 2007. The structural basis of ribo-zyme-catalyzed RNA assembly. Science 315: 1549–1553.

Shechner DM, Grant RA et al. 2009. Crystal structure of thecatalytic core of an RNA-polymerase ribozyme. Science326: 1271–1275.

Soll D, Nishimura S, Moore PB. 2001. RNA. PergamonPress.

Tawfik DS, Griffiths AD. 1998. Man-made cell-like com-partments for molecular evolution. Nat Biotechnol 16:652–656.

Tucker BJ, Breaker RR. 2005. Riboswitches as versatile genecontrol elements. Curr Opin Struct Biol 15: 342–348.

Tuerk C, Gold LC. 1990. Systematic evolution of ligands byexponential enrichment: RNA ligands to bacteriophageT4 DNA polymerase. Science 249: 505–510.

Unrau PJ, Bartel DP. 1998. RNA-catalysed nucleotidesynthesis. Nature 395: 260–263.

Unrau PJ, Bartel DP. 2003. An oxocarbenium-ionintermediate of a ribozyme reaction indicated by kineticisotope effects. Proc Natl Acad Sci 100: 15393–15397.

van Dijk AA, Makeyev EV, Bamford DH. 2004. Initiation ofviral RNA-dependent RNA polymerization. J Gen Virol85: 1077–1093.

Voytek SB, Joyce GF. 2007. Emergence of a fast-reactingribozyme that is capable of undergoing continuous evo-lution. Proc Natl Acad Sci 104: 15288–15293.



Wang Y, Prosen DE, Mei L, Sullivan JC, Finney M,Vander Horn PB. 2004. A novel strategy to engineerDNA polymerases for enhanced processivity andimproved performance in vitro. Nucleic Acids Res 32:1197–1207.

Werner F. 2007. Structure and function of archaeal RNApolymerases. Mol Microbiol 65: 1395–1404.

White HBIII. 1976. Coenzymes as fossils of an earlier meta-bolic state. J Mol Evol 7: 101–104.

Wilson DS, Szostak JW. 1999. In vitro selection of functionalnucleic acids. Annu Rev Biochem 68: 611–647.

Yin YW, Steitz TA. 2002. Structural basis for the transitionfrom initiation to elongation transcription in T7 RNApolymerase. Science 298: 1387–1395.

Yin YW, Steitz TA. 2004. The structural mechanism of trans-location and helicase activity in T7 RNA polymerase. Cell116: 393–404.

Yoshioka W, Ikawa Y, Jaeger L, Inoue T. 2004. A ligase ribo-zyme obtained from a structured pool. Nucleic AcidsSymp Ser (Oxf ) 48: 209–210.

Zaher HS, Unrau PJ. 2007. Selection of an improved RNApolymerase ribozyme with superior extension and fidel-ity. RNA 13: 1017–1026.

Zaher HS, Watkins RA, Unrau PJ. 2006. Two independentlyselected capping ribozymes share similar substraterequirements. RNA 12: 1949–1958.

Zubay G, Mui T. 2001. Prebiotic synthesis of nucleotides.Orig Life Evol Biosph 31: 87–102.



closing the circle: replicating rna with rna

Documents