RNA secondary structure and ribozymes

Intuition Beginner

DNA is the cell's tidy archive. RNA is the cell's working molecule. Unlike DNA, which almost always lives as a clean double helix, a single RNA strand folds back on itself. Its bases pair up — A with U, G with C — forming stems, loops, and bulges. This folding is what lets RNA do chemistry, not just carry a message.

Some folded RNA molecules are enzymes. We call them ribozymes. They speed up chemical reactions inside living cells: cutting RNA strands, joining them back together, and even building proteins. The fact that RNA can both store information and catalyse reactions hints that life may have started in an RNA world, before proteins and DNA existed.

The shape of an RNA molecule decides what it can do. A ribozyme's active site is built from its folded geometry, exactly as a protein enzyme's active site is built from its folded chain. Predicting these shapes — and designing new RNA molecules that bind chosen targets — is a central goal of modern biochemistry and drug design.

Visual Beginner

A folded RNA molecule looks like a knot of stems and loops. The stems are short double-helical regions where the strand pairs with itself. At the end of a stem sits a loop of unpaired bases. When a stem curves sharply back on itself, the loop at the turn is called a hairpin. Two stems that cross each other form a pseudoknot — the feature that lets some RNAs do clever chemistry.

The picture shows why G-C-rich stems (three hydrogen bonds each) anchor the most stable parts of a structure, while loops and bulges mark the flexible joints. Ribozyme active sites sit at junctions where several stems meet, because those junctions bring distant bases close enough to catalyse a reaction.

Worked example Beginner

Count the hydrogen bonds that hold together a small RNA hairpin, and compare it to an A-U-rich version.

A hairpin has a stem of six base pairs and a loop of four unpaired bases. The stem contains four G-C pairs and two A-U pairs.

Step 1. Hydrogen bonds from the G-C pairs: each G-C pair has three hydrogen bonds, so 4 × 3 = 12.

Step 2. Hydrogen bonds from the A-U pairs: each A-U pair has two hydrogen bonds, so 2 × 2 = 4.

Step 3. Hydrogen bonds holding the stem closed: 12 + 4 = 16 in total.

Now compare. If all six stem pairs were A-U, the stem would have only 6 × 2 = 12 hydrogen bonds. The G-C-rich stem is more stable by four hydrogen bonds. This is why the cores of ribozymes and other structural RNAs are G-C-rich: more hydrogen bonds mean a stem that resists being pulled apart by the thermal motion of water.

What this tells us: the base composition of a stem controls its stability, and cells exploit this to build RNA structures that either last or that open and close on cue during catalysis.

Check your understanding Beginner

Formal definition Intermediate+

Primary structure of an RNA molecule is its linear sequence of nucleotides, read 5′ → 3′. Secondary structure is the set of intramolecular base pairs the strand forms when it folds back on itself, ignoring the three-dimensional arrangement of those pairs. Tertiary structure is the full three-dimensional fold — how the secondary-structure elements pack against one another through coaxial stacking, tertiary hydrogen bonds (often involving the 2′-OH and the ribose), and metal-ion coordination. The progression primary → secondary → tertiary mirrors the sequence → contact-map → 3D-coordinate pipeline used by every RNA-structure prediction program ^{[Saenger 1984]}.

Base-pairing rules. Watson-Crick pairs (A-U, G-C, and the DNA analogue A-T) constitute the dominant class. G-U wobble pairs, with their shifted hydrogen-bond geometry, are the most common non-canonical pair in structured RNA and are essential to the codon-anticodon decoding at the ribosome. Beyond wobble, structured RNA exploits Hoogsteen and sheared pairs (which use the major-groove and backbone faces of the bases), base triples, and sugar-edge contacts. The isostericity relation — two base pairs are isosteric if they occupy the same helical position with the same C1′-to-C1′ geometry — explains why evolution can swap a G-C pair for an A-U pair at conserved helical positions without distorting the fold.

Secondary-structure motifs. A stem-loop (or hairpin) is a helical stem closed by a loop of unpaired nucleotides at its end; the smallest stable loop, the tetraloop, commonly has the consensus sequences GNRA or UNCG and acts as a nucleation site for tertiary packing. An internal loop is a region of unpaired bases on both strands of an otherwise helical segment; a bulge is unpaired on one strand only. A multibranch loop (or helical junction) is a point from which three or more helices radiate — the hammerhead and hairpin ribozymes are built on such junctions. A pseudoknot forms when bases inside a loop pair with a region outside the loop that encloses it, crossing the conventional nesting of base pairs; pseudoknots are central to ribosomal frameshifting and to telomerase RNA function.

Nearest-neighbour thermodynamics. The dominant stabilising contribution to an RNA helix is not the hydrogen bonding between paired bases but the vertical stacking of each base on its neighbour along the helix axis. This observation grounds the Turner nearest-neighbour model (treated formally in the next section): the folding free energy of a duplex is, to good approximation, a local function of successive dinucleotide stacks.

Counterexamples to common slips

• RNA secondary structure is not the same as protein secondary structure. In proteins, secondary structure means α-helices and β-sheets formed by backbone hydrogen bonds. In RNA, secondary structure means the base-pairing pattern (the contact map), which is a discrete combinatorial object computable from sequence. The two terms share a name by analogy, not by mechanism.

• Hairpins are not the same as duplexes. A hairpin forms from a single strand folding on itself; its stem has a continuous backbone at the loop end and a free 5′/3′ end at the other. A duplex forms from two separate strands. The Turner nearest-neighbour model parameterises both, but hairpins carry an additional loop-initiation penalty (the entropic cost of closing a loop) that scales with loop size.

• G-U pairs are not mistakes. A G-U wobble pair is a genuine, thermodynamically stable structural element with geometry close to a Watson-Crick pair. Excluding G-U pairs from secondary-structure prediction discards real biology; the Turner parameter set includes G-U-containing nearest neighbours explicitly.

• Catalysis is not caused by magnesium alone. Divalent metal ions stabilise RNA tertiary structure and shield the polyanionic backbone, but in several small ribozymes (hammerhead, hairpin, HDV) the catalytic acid-base chemistry is performed by nucleobases (for instance guanine G12 in the hammerhead and cytosine C75 in HDV), not by the metal. Conflating structural metal binding with catalytic metal chemistry is a recurrent slip.

Key theorem with proof Intermediate+

Theorem (Turner nearest-neighbour free-energy model). Let $D$ be a Watson-Crick RNA duplex of $n$ base pairs with a specified sequence. Under standard conditions (1 M NaCl, pH 7, 37 °C), the standard folding free energy of $D$ is approximated, to within experimental error of optical melting data, by

$$\Delta G_{37}^{\circ }(D)=\Delta G_{\mathrm{init}}^{\circ }+\sum _{i=1}^{n-1}\Delta G_{\mathrm{NN}}^{\circ }(5^{\prime }\text{-}{X}_i{X}_{i+1}\text{-}{3}^{\prime }\Vert 3^{\prime }\text{-}{Y}_i{Y}_{i+1}\text{-}{5}^{\prime })+\Delta G_{\mathrm{AU}\mathrm{end}}^{\circ }\cdot \tau ,$$

where $X_i$ denotes the $i$-th base of the top strand (5′→3′), $Y_i$ its complementary base on the bottom strand, $\tau$ counts terminal A-U pairs, and $\Delta G_{\mathrm{init}}^{\circ }$ is a single strand-association initiation term. The ten unique nearest-neighbour parameters, together with the symmetry, initiation, and terminal-AU penalties, reproduce measured duplex free energies with root-mean-square deviation below $0.5\mathrm{kcal}\mathrm{mo}{\mathrm{l}}^{-1}$ over hundreds of independently measured duplexes ^{[Xia 1998]}.

Proof. Write the duplex as two antiparallel strands with the phosphodiester backbone covalently fixing the register between them: base $X_i$ on the top strand faces $Y_i$ on the bottom strand, for $i=1,\dots ,n$. The three-dimensional geometry of an A-form RNA helix places each base pair in van der Waals contact only with its two immediate neighbours along the helix axis. Long-range stacking contacts across two or more base pairs are geometrically forbidden by the A-form rise and twist, so the stacking component of the folding enthalpy decomposes into local contributions, one per adjacent pair of base pairs. There are exactly $n-1$ such adjacent pairs in a duplex of length $n$.

Each adjacent pair of base pairs is fully specified by the two consecutive bases on the top strand, $X_i{X}_{i+1}$, together with the two complementary bases $Y_i{Y}_{i+1}$ on the bottom strand. Read 5′→3′ on the top strand, this is the Watson-Crick nearest neighbour $5^{\prime }\text{-}{X}_i{X}_{i+1}\text{-}{3}^{\prime }\Vert 3^{\prime }\text{-}{Y}_i{Y}_{i+1}\text{-}{5}^{\prime }$. The folding free energy is therefore additive over these $n-1$ local terms, plus a single initiation term that pays the entropic cost of bringing the two strands together at one helix end, plus a small terminal correction $\Delta G_{\mathrm{AU}\mathrm{end}}^{\circ }$ applied once for each helix end terminated by an A-U pair (A-U ends are less constrained than G-C ends).

For the parameter count: there are four choices for each base in $X_i{X}_{i+1}$, giving $16$ ordered dinucleotides. A duplex and its reverse complement present the same set of nearest-neighbour stacks read from the opposite strand, so the $16$ reduce by symmetry to $10$ unique nearest-neighbour free-energy parameters. Fitting these $10$ parameters (plus symmetry, initiation, and terminal-AU penalties) to a large body of optical melting data on short duplexes yields a self-consistent set. When applied to duplexes not used in the fit, the model predicts measured $\Delta G_{37}^{\circ }$ values within experimental error, which validates the decomposition. $\square$

Bridge. The nearest-neighbour model builds toward 17.05.01 DNA replication, where primer-template duplex stability sets the stringency of primer annealing and the conditions for polymerase extension, and appears again in 15.14.01 enzyme mechanism, where the same base-stacking thermodynamics governs how RNA-binding enzymes and ribozymes recognise their substrates. The foundational reason nearest neighbours suffice is that the phosphodiester backbone fixes the register between strands, so every stacking contact is local. This is exactly the locality that makes RNA folding computable from sequence alone, and the model generalises from duplexes to hairpins, internal loops, and junctions once loop-initiation penalties are added. The bridge is between the discrete combinatorics of base sequences and the continuous thermodynamics of molecular folding, and the same additivity underwrites every algorithm that predicts melting temperatures from primary sequence.

Exercises Intermediate+

Catalytic RNA: the ribozyme families Master

Eight biochemically characterised ribozyme families establish the breadth of RNA catalysis. They divide naturally by mechanism and by size into small self-cleaving ribozymes, large self-splicing ribozymes, the ribonucleoprotein ribozyme RNase P, and the ribosome itself. Each catalyses a phosphoryl-transfer reaction at a scissile phosphate, and each positions its substrate by Watson-Crick or Watson-Crick-like pairing and activates the attacking nucleophile by acid-base chemistry — performed either by a divalent metal ion, by a nucleobase, or by both.

Small self-cleaving ribozymes: hammerhead, hairpin, HDV, VS, and the twister / pistol / hatchet clans. The hammerhead ribozyme, identified by Symons and co-workers in 1987 in plant viroid and satellite RNAs, is a three-helix junction of ~40 nucleotides. Crystal structures (Pley, Flaherty and McKay 1994 of a minimal hammerhead; Martick and Scott 2006 of the full-length active conformation) revealed the active-site geometry that places G12 and the scissile phosphate for general-base catalysis ^{[Cech 1981]}. The hairpin ribozyme, also from a plant satellite RNA, folds as a four-way junction that brings two internal loops into coaxial helical stacking; its catalysis uses nucleobase acid-base chemistry (G8 and A38) and can proceed without divalent metal. The hepatitis delta virus (HDV) ribozyme, the only ribozyme encoded in a human pathogen genome, uses a cytosine (C75) as a general acid in concert with a metal-bound water as general base. The Neurospora VS ribozyme is the largest of the small cleavers and shares the hairpin's substrate-ribose zipper. The more recently discovered twister, twister-sister, pistol, hatchet, and lariat capping ribozymes (Breaker and colleagues, 2014 onward), identified by comparative genomics, expand the small-ribozyme repertoire but follow the same mechanistic template: in-line 2′-OH attack producing a 2′,3′-cyclic phosphate and a 5′-OH. The repeated, independent emergence of this mechanism is itself the structural evidence that the 2′-OH of RNA is the chemical enabler of phosphoryl-transfer catalysis.

Large self-splicing ribozymes: group I and group II introns. Group I introns, of which the Tetrahymena rRNA intron characterised by Cech and colleagues in 1981–1982 is the paradigm, fold into a conserved catalytic core of ~250 nucleotides that uses an exogenous guanosine cofactor as the first-step nucleophile. Group II introns fold into a six-domain catalytic core and use the 2′-OH of a branch-point adenosine as nucleophile, producing a lariat intermediate; they are retrotransposable and encode a reverse transcriptase in domain IV. The mechanistic and structural homology between group II introns and the eukaryotic spliceosome — same two transesterifications, same branch-point adenosine, same lariat intermediate — establishes that the spliceosome is a trans-acting group II intron remodelled around snRNAs, with U2 and U6 snRNA forming the catalytic centre.

RNase P. Ribonuclease P is the endoribonuclease that matures the 5′ end of transfer RNA. In bacteria it is a ribonucleoprotein consisting of a ~350–400 nucleotide catalytic RNA (the M1 RNA) and a small protein cofactor. Guerrier-Takada, Gardiner, Marsh, Pace and Altman showed in 1983 that the RNA subunit alone catalyses accurate 5′ maturation of precursor tRNA at high salt in vitro; the protein cofactor stabilises the complex under physiological ionic conditions and improves substrate affinity but is not the catalyst. RNase P was, with the group I intron, one of the two founding examples of a catalytic RNA, and it remains the only naturally occurring ribozyme that acts on a separate substrate in trans rather than on itself.

The ribosome. The ribosome's peptidyl transferase centre is the catalytic heart of protein synthesis and is built entirely of 23S/28S rRNA. Noller's 1992 protein-extraction assay and the 2000 atomic-resolution crystal structures of the 50S subunit (Ban, Nissen, Hansen, Moore and Steitz) and the 30S subunit (Wimberly, Brodersen, Clemons, Morgan, Vonrhein, Ramakrishnan and colleagues) showed that there is no protein atom within bonding reach of the reacting substrates. The ribosome is therefore a ribozyme — by far the largest and the most quantitatively important in the modern cell, catalysing peptide-bond formation at rates of tens of bonds per second per ribosome. The mechanism is now understood as substrate-assisted catalysis in which the 2′-OH of the P-site A76 ribose acts as a general base, assisted by the entropic effect of precise substrate positioning in the A- and P-sites.

Aptamers and SELEX. Catalytic RNA is not limited to naturally occurring sequences. Tuerk and Gold (1990) and, independently, Ellington and Szostak (1990) introduced systematic evolution of ligands by exponential enrichment (SELEX) — repeated selection of binding or catalytic function from a random-sequence pool of $10^{14}$–$10^{15}$ RNA or DNA molecules, followed by amplification. SELEX has produced aptamers that bind small molecules, proteins, and whole cells with nanomolar to picomolar affinity, and has selected de novo ribozymes that catalyse ligations, phosphoryl transfers, alkylations, and (with appropriate cofactors) Diels-Alder and amide-bond-forming reactions. The demonstrated evolvability of random RNA pools is the operational evidence behind the RNA-world hypothesis: a single polymer class can store information (by sequence) and catalyse chemistry (by folding), satisfying the two requirements for a self-replicating system.

Synthesis. The four small self-cleaving ribozymes, the two large splicing ribozymes, RNase P, and the ribosomal peptidyl transferase together establish that RNA is a genuine catalyst, not merely a passive carrier of genetic information. The foundational reason is the same 2′-OH that makes RNA chemically fragile in solution: positioned in-line by a folded tertiary structure, it becomes the nucleophile of phosphoryl-transfer catalysis. This is exactly the chemical versatility that identifies the ribosome's active site with rRNA rather than protein, and the central insight — verified by Noller's domain-mutation assays and the Steitz-Moore-Yonath 50S structures — is that all known ribozymes position substrates by Watson-Crick pairing and activate nucleophiles by nucleobase or metal-ion general acid-base chemistry. Putting these together with SELEX, which shows that random-sequence pools repeatedly evolve both catalysts and ligand binders, the bridge is to an RNA-world origin of life in which RNA stored information and catalysed its own replication before proteins existed. The pattern generalises from a handful of natural ribozymes to a selection-accessible universe of artificial RNA catalysts, and the same in-line S${}_{\mathrm{N}}$2-like mechanism recurs across every family.

Full proof set Master

Proposition (in-line phosphodiester cleavage mechanism for small self-cleaving ribozymes). The four classical small self-cleaving ribozymes — hammerhead, hairpin, HDV, and VS — all catalyse internal phosphodiester cleavage by an in-line nucleophilic attack of the 2′-OH of an upstream ribose on the adjacent scissile phosphate, producing a 2′,3′-cyclic phosphate on the upstream fragment and a 5′-OH on the downstream fragment, with rate enhancement arising from (i) in-line positioning of the nucleophile, (ii) general acid-base catalysis by a nucleobase or a metal-bound water, and (iii) electrostatic stabilisation of the pentacoordinate transition state.

Proof. Fix a ribozyme and let $S$ denote the scissile phosphate linking the upstream ribose (carrying the 2′-OH nucleophile) to the downstream 5′-oxygen. Three independent lines of evidence constrain the mechanism.

First, the product structure is invariant. In every published cleavage assay of the four ribozymes, gel electrophoresis after cleavage reveals an upstream fragment terminating in a 2′,3′-cyclic phosphate (reactive to periodate only after opening of the cyclic phosphate) and a downstream fragment with a free 5′-OH (kinase-extendable). The cyclic phosphate is the signature product of nucleophilic attack by a vicinal 2′-oxyanion on a phosphorus, closing a five-membered ring; no other nucleophile accessible in the active site yields this product.

Second, the transition-state geometry is in-line. Crystallographic and molecular-dynamics analyses of precursor and vanadate-transition-state analogues show that the 2′-O, the scissile P, and the 5′-O leaving group are collinear (apical angle near 180°), the geometry required for an S${}_{\mathrm{N}}$2-like associative substitution at phosphorus. Departures from collinearity correlate quantitatively with reduced cleavage rate, so in-line positioning is a catalytic contribution, not merely a structural observation.

Third, the rate enhancement decomposes into general acid-base and electrostatic components. pH-rate profiles for each ribozyme reveal a kinetically influential ionisable group whose pK${}_a$ matches a specific active-site nucleobase: G12 in the hammerhead (general base), A38 in the hairpin, C75 in HDV (general acid). Substitution of these nucleobases by analogues with shifted pK${}_a$ values shifts the pH-rate profile commensurately. Divalent metal ions (Mg${}^{2+}$ or Mn${}^{2+}$) contribute through electrostatic stabilisation of the developing negative charge on the non-bridging oxygens of the pentacoordinate transition state and, in some families, through a metal-bound water acting as the general acid.

Together these three constraints fix the mechanism: the products require a 2′-O nucleophile; the transition-state geometry requires in-line attack; the rate enhancement requires acid-base and electrostatic catalysis. No step is left unaccounted for, and the mechanism is therefore established for all four ribozymes up to the acid-base identity of the specific nucleobase in each active site. $\square$

Proposition (SELEX enrichment of high-affinity binders). Let $P_0$ be a pool of $N$ nucleic acid sequences, and let $p_s$ denote the binding probability of sequence $s$ to the target under fixed selection conditions. Under one SELEX round that retains bound sequences and amplifies them, the frequency $f_k(s)$ of $s$ after $k$ rounds satisfies $f_k(s)=f_0(s)(p_s/{\bar{p}}_k)(p_s/{\bar{p}}_{k-1})\cdots (p_s/{\bar{p}}_1)$, where ${\bar{p}}_j$ is the mean binding probability of the round-$j$ pool; in particular, any sequence with $p_s>{\bar{p}}_j$ for all $j$ is monotonically enriched.

Proof. Let $f_j(s)$ be the frequency of $s$ in the pool at round $j$ and $p_s$ its probability of binding (and therefore of being retained). After selection, the retained copy number of $s$ is proportional to $f_j(s)p_s$, and after amplification the frequencies renormalise to one. The mean binding probability of the round-$j$ pool is ${\bar{p}}_j={\sum }_t{f}_j(t)p_t$, so the retained fraction of the pool is ${\bar{p}}_j$. Renormalisation gives

$$f_{j+1}(s)=\frac{f_j(s)p_s}{{\bar{p}}_j}.$$

Iterating from $f_0$ to $f_k$ yields $f_k(s)=f_0(s){\prod }_{j=0}^{k-1}(p_s/{\bar{p}}_j)$, with ${\bar{p}}_j$ recomputed at each round from the current frequencies. Because the denominator ${\bar{p}}_j$ is a pool average, every sequence with $p_s$ strictly above the round-$j$ pool mean contributes a factor $p_s/{\bar{p}}_j>1$ to its own frequency and is enriched; every sequence with $p_s<{\bar{p}}_j$ is depleted. As high-affinity sequences are enriched, the pool mean ${\bar{p}}_j$ rises, progressively raising the bar for further enrichment and driving the pool toward the highest-$p_s$ sequences present in $P_0$. The pool converges, after a finite number of rounds bounded by the dynamic range of $p_s$ across $P_0$, to a population dominated by the highest-affinity sequences in the initial library. $\square$

Connections Master

• Nucleic acid chemistry 15.13.01. RNA secondary structure and ribozyme catalysis are built from the Watson-Crick base pairing, phosphodiester bonding, and 2′-OH chemistry established in the nucleic-acid-chemistry unit. The 2′-OH that enables the ribozyme mechanism here is the same 2′-OH whose intramolecular nucleophilic attack makes RNA chemically less stable than DNA there, and the A-form helix geometry that underpins every RNA fold is the A-form characterised in the sibling unit.

• Amino acids and protein chemistry 15.12.01. The ribosome is the machine that polymerises amino acids into proteins, and its peptidyl transferase centre — an all-RNA catalyst — forms the same peptide bond whose chemistry and Ramachandran geometry are defined in the protein-chemistry unit. The structural isostericity of the ribosome's A- and P-site substrates (aminoacyl- and peptidyl-tRNA) mirrors the backbone-angle analysis applied to polypeptide chains.

• Enzyme mechanism 15.14.01. Ribozymes use the same catalytic strategies as protein enzymes — general acid-base catalysis, transition-state stabilisation, proximity and orientation, metal-ion cofactors — but deploy them from an RNA scaffold. The hammerhead's G12 general base and the HDV ribozyme's C75 general acid are direct RNA analogues of the histidine and lysine acid-base residues treated in the enzyme-mechanism unit, and the comparison isolates which features of catalysis are generic to any folded biopolymer and which are specific to protein chemistry.

• DNA replication 17.05.01. The nearest-neighbour free-energy model that predicts RNA duplex folding also predicts DNA primer-template duplex stability, which sets the stringency of primer annealing, the melting behaviour of replication intermediates, and the conditions under which polymerase proofreading operates. The additivity that makes RNA folding computable from sequence is the same additivity that underlies DNA hybridisation design.

• Translation 17.05.03. The ribosome-as-ribozyme result developed here is the structural foundation of the translation unit's account of peptide-bond formation. The wobble-pairing chemistry that permits third-position codon degeneracy, introduced here as a non-canonical RNA base pair, becomes the decoding rule at the A site of the ribosome in the translation unit.

Historical & philosophical context Master

The discovery that RNA can act as an enzyme overturned a hard-won dogma of twentieth-century biochemistry — that all biological catalysts are proteins — and reopened the question of how life began. Thomas Cech and co-workers reported in 1981 that the intervening sequence of the Tetrahymena ribosomal RNA precursor is excised and cyclised without any protein in vitro ^{[Cech 1981]}; the 1982 paper by Kruger, Grabowski, Zaug, Sands, Gottschling and Cech named the phenomenon self-splicing and characterised the reaction as an RNA-catalysed transesterification. Sidney Altman and co-workers reached the complementary conclusion by a different route: the RNA component of ribonuclease P, studied by Guerrier-Takada, Gardiner, Marsh, Pace and Altman in 1983, is itself catalytically competent ^{[Guerrier-Takada 1983]}. The 1989 Nobel Prize in Chemistry was awarded jointly to Cech and Altman for the discovery of catalytic RNA.

The extension of the ribozyme principle to the ribosome took a further decade. Harry Noller and colleagues showed in 1992 that peptidyl transferase activity survives the removal of most ribosomal protein ^{[Noller 1992]}, localising the activity to rRNA. The decisive structural confirmation came in 2000 with the atomic-resolution crystal structures of the 50S ribosomal subunit by Ban, Nissen, Hansen, Moore and Steitz ^{[Ban 2000]} and of the 30S subunit by Wimberly, Brodersen, Clemons, Morgan, Vonrhein and Ramakrishnan, which showed that the peptidyl transferase centre contains no protein atoms within bonding reach of the substrates. The 2009 Nobel Prize in Chemistry was awarded to Ada Yonath, Venki Ramakrishnan and Tom Steitz for the ribosome structures.

The selection of functional RNA from random pools by SELEX, introduced by Tuerk and Gold and by Ellington and Szostak in 1990 ^{[Tuerk 1990]}, provided the operational counterpart to the natural ribozymes: arbitrary RNA sequences can be evolved in the test tube to bind chosen targets or catalyse chosen reactions, demonstrating that catalytic function is not rare in sequence space. Walter Gilbert's 1986 paper framed these results as the RNA world ^{[Gilbert 1986]} — a proposed stage of early life in which RNA both stored genetic information and catalysed its own replication, with DNA and protein appearing only later as specialisations of information storage and catalysis.

Bibliography Master

@article{Cech1981,
  author = {Cech, T. R. and Zaug, A. J. and Grabowski, P. J.},
  title  = {In vitro splicing of the ribosomal {RNA} precursor of {Tetrahymena}},
  journal = {Cell},
  volume = {27},
  year   = {1981},
  pages  = {487--496},
}

@article{Kruger1982,
  author = {Kruger, K. and Grabowski, P. J. and Zaug, A. J. and Sands, J. and Gottschling, D. E. and Cech, T. R.},
  title  = {Self-splicing {RNA}: autoexcision and autocyclization of the ribosomal {RNA} intervening sequence of {Tetrahymena}},
  journal = {Cell},
  volume = {31},
  year   = {1982},
  pages  = {147--157},
}

@article{GuerrierTakada1983,
  author = {Guerrier-Takada, C. and Gardiner, K. and Marsh, T. and Pace, N. and Altman, S.},
  title  = {The {RNA} moiety of ribonuclease {P} is the catalytic subunit of the enzyme},
  journal = {Cell},
  volume = {35},
  year   = {1983},
  pages  = {849--857},
}

@article{Noller1992,
  author = {Noller, H. F. and Hoffarth, V. and Zimniak, L.},
  title  = {Unusual resistance of peptidyl transferase to protein extraction procedures},
  journal = {Science},
  volume = {256},
  year   = {1992},
  pages  = {1416--1419},
}

@article{Ban2000,
  author = {Ban, N. and Nissen, P. and Hansen, J. and Moore, P. B. and Steitz, T. A.},
  title  = {The complete atomic structure of the large ribosomal subunit at {2.4} {\AA} resolution},
  journal = {Science},
  volume = {289},
  year   = {2000},
  pages  = {905--920},
}

@article{Wimberly2000,
  author = {Wimberly, B. T. and Brodersen, D. E. and Clemons, W. M. and Morgan-Warren, R. J. and Carter, A. P. and Vonrhein, C. and Hartsch, T. and Ramakrishnan, V.},
  title  = {Structure of the {30S} ribosomal subunit},
  journal = {Nature},
  volume = {407},
  year   = {2000},
  pages  = {327--339},
}

@article{TuerkGold1990,
  author = {Tuerk, C. and Gold, L.},
  title  = {Systematic evolution of ligands by exponential enrichment: {RNA} ligands to bacteriophage {T4 DNA} polymerase},
  journal = {Science},
  volume = {249},
  year   = {1990},
  pages  = {505--510},
}

@article{EllingtonSzostak1990,
  author = {Ellington, A. D. and Szostak, J. W.},
  title  = {In vitro selection of {RNA} molecules that bind specific ligands},
  journal = {Nature},
  volume = {346},
  year   = {1990},
  pages  = {818--822},
}

@article{Xia1998,
  author = {Xia, T. and SantaLucia, J. and Burkard, M. E. and Kierzek, R. and Schroeder, S. J. and Jiao, X. and Cox, C. and Turner, D. H.},
  title  = {Thermodynamic parameters for an expanded nearest-neighbor model for formation of {RNA} duplexes with {Watson-Crick} base pairs},
  journal = {Biochemistry},
  volume = {37},
  year   = {1998},
  pages  = {14719--14735},
}

@article{Gilbert1986,
  author = {Gilbert, W.},
  title  = {The {RNA} world},
  journal = {Nature},
  volume = {319},
  year   = {1986},
  pages  = {618},
}

@article{MartickScott2006,
  author = {Martick, M. and Scott, W. G.},
  title  = {Tertiary contacts distant from the active site prime a ribozyme for catalysis},
  journal = {Cell},
  volume = {126},
  year   = {2006},
  pages  = {309--320},
}

@book{Saenger1984,
  author = {Saenger, W.},
  title  = {Principles of Nucleic Acid Structure},
  publisher = {Springer-Verlag},
  year   = {1984},
}

@book{AtkinsGestelandRNAWorld,
  author = {Atkins, J. F. and Gesteland, R. F. and Cech, T. R.},
  title  = {The {RNA} World},
  publisher = {Cold Spring Harbor Laboratory Press},
  year   = {2011},
}