Structure, Folding and Activity of Branched Nucleic Acids

 

The Nucleic Acid Structure Research Group

 

Helical branchpoints are important elements in both DNA and RNA. In DNA they constitute central intermediates in recombination processes, while in RNA they are important architectural elements.
 
DNA junctions – their structure and dynamics
Genetic recombination is a process by which similar DNA molecules become transiently broken and then re-joined in a new combination. It provides genetic diversity, the raw material of evolution, and an important process by which cells can repair their genetic material against double-stranded breaks. The central intermediate of this process is a four-way junction (or Holliday junction), a structure in which four double-helical DNA molecules are connected by mutually exchanging component strands. We determined the structure of this species in 1988, using new electrophoretic and fluorescence approaches [Duckett et al Cell 55, 79-89, 1988].
 
The stacked X-structure
In the presence of divalent metal ions, the DNA junction adopts the stacked X-structure by pairwise, coaxial stacking of helices in a right-handed, antiparallel structure. The structure has been confirmed in virtually every detail by X-ray crystallography in the last few years.

 

The four-way DNA junction solved by crystallography. This was solved by Shing Ho and colleagues [Eichman et al Proc. Natl. Acad. Sci. USA  97, 3971-3976, 2000].

 

Electrostatics, metal ions and the stacked X-structure
Nucleic acids are highly charged polyelectrolyte molecules, and like most such species their structure depends strongly on the presence or absence of metal ions such as magnesium. In the absence of such ions, the structure is open and extended.

The four-way DNA junction. It exists as an extended structure in the absence of divalent metal ions. On addition of ions it folds by coaxial pairwise stacking of arms into the stacked X-structure

 

There is a cluster of charged phosphate groups at the centre of the junction, providing a regions of very high electrostatic potential that would be expected to attract metal ions. Substitution of central phosphate groups by electrically-neutral methyl phosphonates can lead to ion-independent folding into the stacked X-structure. However, the effect is strongly dependent on stereochemistry, and can depend on whether the proR or proS oxygen atom is substituted. It is likely that the stereochemical environment of the methyl group affects the interaction with metal ions in the centre of the junction [Liu et al Chem. Biol. 12, 217-228, 2005].

 

The dynamic character of the DNA junction
We have come to appreciate that the DNA junction is a very dynamic object, and single-molecule spectroscopy in collaboration with Taekjip Ha and Sean McKinney in Urbana has been especially revealing. This readily shows that junctions are constantly flipping between the two alternative ways of pairwise helical stacking [McKinney et al Nature Struct. Biol. 10, 93-97, 2003]. The bias between conformers varies between junctions of different central sequence.

Stacking conformer exchange in a four-way DNA junction. A FRET time trace for a single molecule of a junction (with the fluorophores Cy3 attached to H arm and Cy5 to the B arm) showing repeated flipping between the two possible stacking conformers, giving high and low FRET states [McKinney et al Nature Struct. Biol. 10, 93-97, 2003].

 

Branch migration is a second process that is very important in natural junctions created from homologous sequences, in which there is a step-wise exchange of basepairs. The mechanism of branch migration is very difficult to study by conventional means because there is no way to synchronise the process in a population of molecules. However, single-molecule FRET has allowed us to observe branch migration occurring in DNA junctions [McKinney et al Proc. Natl. Acad. Sci. USA 102, 5715-5720, 2005]. We have found that the free energy landscape of branch migration can be highly non-uniform and is governed by two types of sequence-dependent barriers.
 
The interaction of DNA junctions with proteins
Four-way junctions must ultimately be cut and rejoined (resolved) to re-create regular duplex DNA. There are a group of enzymes – the DNA junction-resolving enzymes -  that are dimeric nucleases that are exquisitly selective for the structure of the junction. These have been isolated from a wide variety of organisms from bacterial viruses to mammals. Despite being almost paradigms for structure-recognition, all these proteins distort the structure of the junction, ie they appear to alter the very property that they recognise. Yet it turns out that this is quite central to their function. The release of structural strain appears to be coupled to the cleavage process, thereby ensuring apparent simultaneous cleavage of the two strands and thus a productive resolution event.
 
In collaboration with Simon Phillips and John Hadden in Leeds we have obtained the crystal structure of one of the enzymes, from the bacteriophage T7. This has an unusual architecture in which the two polypeptides of the dimer are intertwined, and both contribute aminoacid side chains to the active sites.

The structure of T7 endonuclease I. The two polypeptides comprising the dimer are shown in blue and green. Note the exchange such that each domain contains aminoacids from both polypeptide [Hadden et al Nature Struct. Biol.  8, 62-67, 2000]

 

The structure and nature of these active sites are remarkably similar to a group of restriction enzymes, and we have been able to propose a chemical mechanism for the hydrolysis of the phosphodiester bond.

The active site of T7 endonuclease I. It comprises three acidic residues and a lysine. One of the acidic residues (Glu 20) is contributed from the other polypeptide of the dimer. The acidic residues coordinate two metal ions that probably provide the hydrolytic water molecule [Dèclais et al J. Molec. Biol. 307, 1145-1158, 2001].

 

We have proposed a detailed model for the structure of the endonuclease I dimer with a DNA junction by combining our data distorted structure of the DNA in the complex and the local structure of BglI with DNA which has a very similar active site structure.

A model of the structure of a complex between T7 endonuclease I and a DNA junction [Dèclais et al EMBO J. 22, 1398-1409, 2003].

 

The sequence of Kt-7 that occurs in the 23S rRNA.

 

These elements have been identified in U4 snRNA, box C/D and H/ACA snoRNAs, mRNA and riboswitches, and are thus involved in RNA splicing and modification, and in gene regulation. In the ribosome they were found to be tightly kinked structures. However, somewhat to our surprise we found that in free solution the RNA exists the kinked conformation and a more extended structure, and that the equilibrium is driven towards the kinked form by divalent metal ion concentration [Goody et al RNA  10, 254–264, 2004]. Many natural K-turns are protein binding site, and the ribosomal L7 protein (and related proteins such as the human 15.5 kDa protein) bind to K-turns with high affinity. We have recently shown that the kinked conformation is strongly stabilized by the binding of L7Ae protein, a good example of induced fit in RNA structure [Turner et al RNA  11, 1192-1200, 2005].

L7Ae bound to a K-turn, with fluorophores attached as used in our FRET analysis [Turner et al RNA  11, 1192-1200, 2005].

 

It seems likely that K-turns provide a flexible joint that allows an RNA to explore many conformations during its folding, that may then be locked into place by the binding of a protein that fixes it in the kinked geometry.
 
RNA catalysis
Ribozymes are RNA species that accelerate chemical reactions. These provide a fascinating challenge for the biological chemist, to understand the origins of the catalysis, and perhaps help us to clarify ideas about biocatalysis in general. Moreover, many people now feel that in the development of life on this planet, life may have gone through an early stage Òthe RNA worldÓ during which RNA was simultaneously both the informational and catalytic molecular species. Thus, we have the feeling that in studying these ribozymes that we are peering back 2 billion years, perhaps examining very ancient molecular fossils.
 
We have made a special study of three nucleolytic ribozymes. The structures of all three ribozymes are based around helical junctions – these are clearly very effective folding strategies in small, autonomously-folding RNA species. These ribozymes break or re-ligate the phosphodiester backbone of RNA by a transesterification reaction.

The mechanism of the nucleolytic ribozymes. The cleavage reaction involves attack of the 2'-hydroxyl on the 3'-phosphate group in an S_N2 reaction, generating a bipyramidal phosphorane transition state. Departure of the 5'-oxygen leads to the formation of a cyclic 2'-3' phosphate. The ligation reaction is the reverse of this, with attack of the 5' oxygen on the cyclic phosphate. The possible involvement of general acid-base catalysis by X and Y is shown for the two reactions.

 

We can identify a number of potential catalytic strategies including acid-base catalysis to generate better nucleophiles and leaving groups, transition state stabilization, and conformational effects. These effects combine together to generate a rate enhancement of around a million fold
 
The hammerhead ribozyme
The hammerhead ribozyme is the smallest member of the nucleolytic ribozymes, and is based on an elaborated three-way junction.  We have previously shown that this RNA folds in two distinct ion-induced steps, corresponding to the formation of a domain mediating coaxial helical stacking, followed by folding of the catalytic core [Bassi et al EMBO J.  16, 7481-7489, 1997]. However, the second step required millimolar concentrations of magnesium ions to occur. It has been shown recently that the activity of the ribozyme at micromolar magnesium ion concentrations is markedly increased by the inclusion of loops in two helices. We have found that folding into the active conformation occurs in a single step with these loops present, in the micromolar range of magnesium ion concentration [Penedo et al RNA  10, 880-888, 2004]. The loops are not explicity required for catalytic activity, but enhance the folding of the ribozyme so that it may occur in physiological conditions – this now appears to be a fairly general phenomenon.
 
The hairpin ribozyme
The hairpin ribozyme comprises a four-way helical junction that presents two adjacent arms containing formally-unpaired internal loops (A and B). A cleavage/ligation reaction occurs within the smaller A loop. Using FRET we showed that ribozyme junction folds by coaxial stacking of arms such that the two loops are on adjacent arms, and upon addition of magnesium ions there is an antiparallel rotation that leads to an intimate association between the loops [Murchie et al Molec. Cell  1, 873-881, 1998]. This structure was later confirmed by Adrian Ferrè d'Amarè by crystallography. Although the ribozyme retains activity in the absence of the four-way junction (with the two arms simply hinged by a single phosphodiester linkage), the junction reduces the required magnesium concentration for folding by orders of magnitude.
 
We have studied folding and catalysis using single-molecule FRET in an extensive collaboration with Taekjip Ha and coworkers in Urbana. We have found that the junction introduces a discrete and obligate intermediate into the folding process, that accelerates the docking of the loops by three orders of magnitude [Tan et al .  Proc. Natl. Acad. Sci. USA  100, 9308-9313, 2003]. We have also been able to observe repeated cycles of cleavage and ligation reactions in single ribozyme molecules, and thus measure the rates of both processes in a uniquely direct way – an example of single-molecule enzymology in RNA [Nahas et al  Nature Struct. Molec. Biol.  100, 9308-9313, 2003]. The position of the internal equilibrium is biased toward ligation, but the cleaved ribozyme undergoes several undocking events prior to ligation, during which products may dissociate. This makes good biological sense in terms of the requirement of the viral RNA.

Single-molecule enzymology on the hairpin ribozyme. A time trace for a ribozyme molecule showing interconversion of cleaved and ligated forms. The cleaved form of the ribozyme exhibits rapid exchange between docked and undocked forms, whereas the ligated form remains stably docked. The points at which cleavage and ligation occur can therefore be clearly identified [Nahas et al  Nature Struct. Molec. Biol.  100, 9308-9313, 2003].

 

The active site of the ribozyme contains the nucleobases of G8 and A38 that are H-bonded to the scissile phosphate, ideally placed to act in general acid-base catalysis. We find that both cleavage and ligation reactions are pH dependent, corresponding to the titration of a group with pK_A = 6.2, consistent with A38 with an elevated pK_A. We are currently extending these studies with a chemo-genetic approach.
 
The VS ribozyme
The VS ribozyme is the largest of the nucleolytic ribozymes, and the only one for which there is no crystal structure. The ribozyme is organised by two three-way helical junctions. We have deduced the structure of both junctions by a combination of electrophoretic and FRET approaches, and shown that they fold in response to the non-cooperative binding of magnesium ions. We have determined the dihedral angle between the two junctions, and thereby established the global structure of the five-helix ribozyme [Lafontaine et al . EMBO J. 21, 2461-2471, 2002].  We have deduced the general manner of binding of the substrate stem-loop, docked into the cleft formed between helices II and VI. In that position the cleaved phosphate can interact closely with the A730 loop.

A model for the structure of the VS ribozyme with bound substrate. The substrate is shown in green, with the scissile phosphate highlighted in red [Lafontaine et al . EMBO J. 21, 2461-2471, 2002].

 

We have shown that the cleavage activity of the ribozyme is strongly impaired by most single substitutions of the four nucleotides of this loop [Lafontaine et al  J. Molec. Biol.  312, 663-674, 2001]. Cleavage and ligation activity is especially sensitive to changes in A756 within the probable active site. Replacement by any other nucleotide leads to reduction in cleavage rate by three orders of magnitude, as does ablation of the base. The Watson-Crick edge of the base is particularly critical, notably the presence and position of the exocyclic amine group [Lafontaine et al  J. Molec. Biol. 323, 23-34, 2002]. This base is a good candidate for direct participation in the cleavage reaction. Replacement of A756 by a novel nucleoside in which the nucleobase is replaced by imidazole leads to significant rates of cleavage and ligation, consistent with a role in general acid-base catalysis [Zhao et al . J. Amer. Chem. Soc. 127, 5026-5027, 2005].

Cleavage activity in a VS ribozyme in which the nucleobase of A756 has been replaced by an imidazole group.