article published online: 20 July 2014 | doi: 10.1038/nchembio.1587
Crystal structure and mechanistic investigation of the twister ribozyme Yijin Liu, Timothy J Wilson, Scott A McPhee & David M J Lilley*
npg
© 2014 Nature America, Inc. All rights reserved.
We present a crystal structure at 2.3-Å resolution of the recently described nucleolytic ribozyme twister. The RNA adopts a previously uncharacterized compact fold based on a double-pseudoknot structure, with the active site at its center. Eight highly conserved nucleobases stabilize the core of the ribozyme through the formation of one Watson-Crick and three noncanonical base pairs, and the highly conserved adenine 3′ of the scissile phosphate is bound in the major groove of an adjacent pseudoknot. A strongly conserved guanine nucleobase directs its Watson-Crick edge toward the scissile phosphate in the crystal structure, and mechanistic evidence supports a role for this guanine as either a general base or acid in a concerted, general acid-base–catalyzed cleavage reaction.
R
NA-mediated catalysis is important in protein synthesis1, RNA splicing2 and a variety of other RNA processing events3 and provides insight into the chemical origins of life on the planet4. The nucleolytic ribozymes are a group of RNA species that undergo self-cleavage or ligation at a particular site. Originally discovered in plant pathogens5–7, these were mostly involved in processing replication intermediates and later were found to be a means of regulating genetic expression in bacteria8. However, it has recently become clear that ribozyme sequences are very widespread in many genomes, located within noncoding RNA sequences and strongly conserved and expressed inside the cell. Ribozymes of the hepatitis delta virus (HDV)/cytoplasmic polyadenylation element binding protein 3 (CPEB3) (ref. 9) and hammerhead10,11 families have been found in many eukaryotic species, including humans, and their location and expression indicate that they are functional in vivo. A new member of the nucleolytic ribozyme class called twister was very recently described12. Twister sequences are extremely widespread, with 2,700 examples distributed over bacteria, fungi, plants and animals including insects, fish and worms. The RNA exhibits nucleolytic ribozyme activity both in vitro and in vivo, with cleavage rates comparable to those of other nucleolytic ribozymes. The distribution and activity of these noncoding sequences suggests a role in genetic control. Twister ribozymes have a highly conserved secondary structure (Fig. 1a), with a core comprising a stem-loop interrupted by two internal loops. The cleavage site is located within loop L1, one nucleotide 3′ to the P1 stem. Phylogenetic analysis strongly indicates the presence of two long-range tertiary interactions, T1 and T2, so that the structure should fold into a double pseudoknot. Strongly conserved nucleotides are contained within both L4 and L1 (ref. 12), including ten nucleotides with >97% conservation. In the majority of the ribozymes, there is an additional stem-loop (P3) between T2 and P4, and in a few cases another stem-loop (P5) connects to the 3′ side of L2. Circular permutations exist, with a P1 stem-loop and either the P3 helix or the 3′ side of L2 being open12. The mechanism of action of catalytic RNA species is incompletely understood in general and has been sometimes controversial. In principle, there are two main ways of catalyzing phosphoryl transfer reactions, using general acid-base catalysis or metal ion catalysis,
and protein enzymes use one or the other mechanism. RNA enzymes similarly divide into these two groups, with the self-splicing introns and RNase P acting as metalloenzymes, whereas the nucleolytic ribozymes seem to use general acid-base catalysis13. Yet within the nucleolytic ribozymes, the detailed mechanism of catalysis and the functional groups involved are quite variable. All of them use one or more nucleobases, and guanines are involved in the mechanisms of all the ribozymes except the HDV ribozyme14–19. The VS and hairpin ribozymes use a combination of guanine and adenine nucleobases19–23. By contrast, the HDV ribozyme uses a cytosine nucleobase as a general acid in cleavage24–26 together with a hydrated metal ion to activate the nucleophile, acting either as a general base24 or Lewis acid27. Lastly, for the hammerhead ribozyme, crystallographic and mechanistic evidence suggests that a 2′-hydroxyl group may act as the general acid28,29. We have now solved a crystal structure of a basic form of the twister ribozyme. The RNA adopts a double-pseudoknot structure in which all of the conserved nucleotides form key structural elements. Mechanistic studies, together with our structure, support nucleobase-mediated general acid-base catalysis and point to the involvement of a conserved guanine.
RESULTS Crystallization and structure determination
We crystallized a basic P1 form of the twister ribozyme lacking both P3 and P5 helices based on the Osa-1-4 sequence from Oryza sativa. RNA of this sequence cleaved readily during transcription in vitro (Supplementary Results, Supplementary Fig. 1). The RNA was chemically synthesized with a 2′ H substitution at U6, thus removing the nucleophile to prevent cleavage occurring. A derivative with 5-bromocytosine substitutions in P1 and loop 2 was also synthesized. Both crystallized in the hexagonal space group P6522, and data were collected from the native and derivative crystals to 2.3 Å and 2.8 Å, respectively (Supplementary Table 1). The asymmetric unit contained a single ribozyme molecule. The bromine atoms were located using single-wavelength anomalous diffraction (Supplementary Fig. 2), from which models of the complete structure were built and refined. All except two nucleotides could be placed with confidence, together with a number of Mg2+ ions. Some conformational heterogeneity of the phosphodiester backbone was observed in the region of A7 (Supplementary Fig. 3).
Cancer Research UK Nucleic Acid Structure Research Group, Medical Sciences Institute/Wellcome Trust Building (MSI/WTB) Complex, University of Dundee, Dundee, UK. *e-mail: [email protected] nature CHEMICAL BIOLOGY | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
739
article
Nature chemical biology doi: 10.1038/nchembio.1587
a
b T1
29 30
25 24
P3
L4 23
c
T2
P3
P4
T2 40
32 15
P4
19 30
14
P5
19
T2
T2
13
P2 36
P2
G45
20
P4 A29
T1
G45
T1
31
12
L2
A29
P4
P1
P1
15
P2 T2 13
24
41
P2 7
45
8
45
46
28
29
T1
7
T1
L1
8
49
50
5
npg
© 2014 Nature America, Inc. All rights reserved.
1
5′
3′
54
G22
G30 C32 G31
G30 C32
G23
G31
25
A29
U24
49
P1
G22
G23
6
6
d
A29
U24
P1 54
1 5′
C25
3′
A28 C26
C27
C25
A28 C26
C27
Figure 1 | The sequence and structure of the twister ribozyme. (a) The sequence and secondary structure of the 54-nucleotide O. sativa Osa-1-4 twister as crystallized. Red nucleotides are >97% conserved. Grey nucleotides are not included in our final model. The arrow indicates the position of ribozyme cleavage. (b) Schematic to illustrate the connectivity of the twister ribozyme in the three-dimensional structure. (c) Parallel-eye stereoscopic image of the complete ribozyme, shown in cartoon mode. The color coding of the nucleotides matches that in a. The scissile phosphate is indicated by a red ball. (d) Parallel-eye stereoscopic image of the L4 loop, showing the 2Fo-Fc map of the twister ribozyme contoured at 2σ.
Fitting this to a double conformation improved the electron density map. The conformational difference is purely local; the overall structure of the ribozyme is indistinguishable in the two conformations.
The overall structure of the ribozyme
The overall structure of the twister ribozyme is fully consistent with the proposed12 secondary structure (Fig. 1b,c and Supplementary Fig. 4) and is distinct from all of the other known nucleolytic ribozyme structures15,25,30–32. The terminal loop L4 (Fig. 1d) participates in two tertiary interactions of opposite polarity with L1 and L2, generating the helices T1 and T2, respectively, so that the overall structure is a double pseudoknot. Helices P1, T1, P2 and T2 are coaxially aligned, although there is a large helical twist angle between helices P1 and T1. In addition, base pairs formed between A28 and A46 (trans Watson-Crick) and A8 and G45 (cis-Hoogsteen–sugar edge) are located coaxially in between T1 and P2. Indeed, the A28-A46 pair should be formally included within T1, and the A8-G45 pair should be included in P2. T2 is capped by a final Watson-Crick pair formed between C15 and G19 (Supplementary Fig. 5). This would be the first base pair of the P3 helix, if present, as there are almost invariably no intervening nucleotides between T2 and P3 or between P3 and P4 (ref. 12). Thus, the C15-G19 pair indicates how P3 must be connected in the majority of twister ribozymes (details below; Supplementary Fig. 6). In our basic form of the ribozyme, C15 and G19 are connected by A16, A17 and U18, although electron density for A17 and U18 was not visible in the 2Fo-Fc map, so these nucleotides were not included in our final model. The axis of helix P4 is approximately parallel to that of P1 through T2. It comprises four Watson-Crick pairs, including the essentially invariant G23-C32 base pair, and includes an additional trans Watson-Crick–Hoogsteen pair formed between U24 and A29. Indeed, every nucleotide of L4 is involved in base-base interactions, and all of the substitutions tested result in lower cleavage activity (Supplementary Table 2). Nucleotides A36 through A40 form the 74 0
connection between P4 and P2 and will be the location of P5, when present. A39 and A40 interact with the minor groove of T2 and the C15-G19 base pair, and more extensive A-minor interactions would be possible with an extended P3. Four well-defined magnesium ions are observed bound within the structure of the ribozyme fold (Supplementary Fig. 7). Two have complete inner coordination shells of water molecules, whereas two have exchanged some inner-sphere water ligands for phosphate nonbridging oxygen atoms.
The structural roles of the highly conserved nucleotides
Of the ten highly conserved nucleotides in the twister ribozyme, eight are found in three noncanonical base pairs (Fig. 2) and in one Watson-Crick base pair (Supplementary Fig. 8) in the core of the folded structure (Fig. 2). The formation of the A28-A46 trans Watson-Crick base pair at the end of T1 is facilitated by a turn in the backbone between C27 and A28 and enables A28 to accept a hydrogen bond from O2′ of A7, thus helping position A7 and presumably the scissile phosphate as well. Notably, the A28G variant ribozyme had much lower activity than the A46G variant (Supplementary Table 2), possibly because the A28G variant would project its N2 exocyclic amine into the major groove close to the scissile phosphate. Stacked on A28-A46 is the A8-G45 cis-Hoogsteen–sugar edge base pair of P2. This pair directly interacts with the scissile phosphate with the N2 exocyclic amine of G45, donating a hydrogen bond to the pro-R oxygen (Supplementary Fig. 8). The other two highly conserved base pairs are found at the end of P4 and seem to stabilize the formation of the two pseudoknots. The first is the G23-C32 Watson-Crick base pair, where C32 N4 donates a hydrogen bond to the pro-S nonbridging oxygen of the phosphate of C31, stabilizing the sharp turn in the backbone at the transition between T2 and P4. The potential importance of this bond is illustrated by the 104-fold lower activity of a C32U ribozyme (Supplementary Table 2). U24 and A29, the two nucleotides flanking T1, form a trans Watson-Crick–Hoogsteen base pair at the end of P4. The positioning of A29, which links T1 and T2, is further
nature chemical biology | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
article
Nature chemical biology doi: 10.1038/nchembio.1587 a
other nucleotide at position 40, but only a guanine at position 30 is able to donate the hydrogen bond to the sugar edge of the base at position 40. The other highly conserved nucleotide is A7. Its structural context is described in detail below.
A28
A46 A7
The conformation of the active site of the ribozyme
b G45
A8
c
A29
© 2014 Nature America, Inc. All rights reserved.
U24
Figure 2 | The structural role of conserved nucleotides in the twister ribozyme. (a) The A28-A46 trans Watson-Crick base pair at the end of T1. N3 of A28 accepts a hydrogen bond from the O2′ of A7. (b) The A8-G45 cis-Hoogsteen–sugar edge base pair of P2. The N2 exocyclic amine of G45 donates a hydrogen bond to the pro-R oxygen of the scissile phosphate. (c) The U24-A29 trans Watson-Crick–Hoogsteen base pair at the end of P4.
facilitated by the donation of a hydrogen bond from its O2′ to the pro-R nonbridging oxygen of the phosphate of C31. The highly conserved G30 is part of the second base pair of T2, which hydrogen bonds to A40 and presumably helps stabilize T2 (Supplementary Fig. 8). A similar interaction is possible with any a
The scissile phosphate linking dU6 and A7, and thus the active site of the ribozyme, is located in the center of the RNA structure, lying in the major groove of the T1-P2 helix (Fig. 3a). An O2′ atom modeled onto dU6 deviates ~90° from the in-line orientation of the O2′ nucleophile, P and O5′ leaving group that would be optimal for activity (Fig. 3b). This clearly cannot be the active geometry, but our structure provides an explanation for this. dU6 makes no interactions with this ribozyme, but instead its O4 accepts a hydrogen bond from G23 N2 of a symmetry-related ribozyme molecule in the crystal lattice (Supplementary Fig. 9). The extrusion of dU6 from the ribozyme core is essentially an artifact of crystallization and possibly results from the use of a deoxyribonucleotide to prevent cleavage. There can be no doubt that the active ribozyme has the global structure observed in the crystal, which is completely consistent with all of the phylogenetic data, and thus any deviation from the active structure is local. Moreover, A7 is very specifically positioned at the base of the P4 helix. A7 is stacked on the U24-A29 pair at the lower end of P4, and its O2′ is hydrogen bonded to A28 N3 (Fig. 3c). In addition, the two N6 protons are donated to the backbone nonbridging oxygen atoms P26 pro-R and P25 pro-S. The latter phosphate lies at the sharp turn between the P4 and T1 helices, which is stabilized by a Mg2+ ion (Supplementary Fig. 7). Collectively, these interactions with L4 hold A7 in a pocket in an unusual syn conformation with a C2′-endo sugar pucker. It was therefore likely that the lattice-induced extrusion of dU6 was responsible for the departure from an in-line trajectory. Examination of the immediate environment of the scissile phosphate suggested that in free solution it was likely that U6 could b
A29
A29 A7
A7
U24
U24
A8
npg
A8
G45
G45 A7
A46 dU6
G45
U6
c
A7
A46
dU6
G45
U6
d
A29
A29
U24
U24
G45 A7
A7
U6
C25 A28
A7
G45
A7
U6
C25 A28
C26 C27
C26 C27
Figure 3 | The active center of the twister ribozyme. (a) The immediate environment of the scissile phosphate between dU6 and A7 (cyan), showing the nearby G45 and A29 nucleotides. The P1 (blue) and T1 (green) helices are shown in space-filling representation. (b) A view into the active center, with an oxygen atom added in the 2′ position of U6 (red sphere), i.e., showing the position of the nucleophile. This is clearly far from the position required for inline nucleophilic attack on the scissile phosphate. (c) A7 is held in a pocket below the P4 helix, making interactions with the L4 nucleotides of the T1 helix. Electron density is the 2Fo-Fc map contoured at 2σ. (d) The structure was modified by a local rotation of U6, as explained in the text. The nucleobase is stacked with that of G45, and the O2′ added to U6 is close to an in-line trajectory. It is also close to N1 of G45. nature CHEMICAL BIOLOGY | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
741
Nature chemical biology doi: 10.1038/nchembio.1587
npg
© 2014 Nature America, Inc. All rights reserved.
The chemical mechanism of the ribozyme
Like other nucleolytic ribozymes, the reaction follows an SN2related mechanism, yielding 2′,3′-cyclic phosphate and 5′-hydroxyl products (Supplementary Fig. 11)12. Efficient catalysis requires activation of the nucleophile and stabilization of the leaving group through proton transfer, and the precedent set by other nucleolytic ribozymes led us to anticipate that general acid-base catalysis might have a major role17,19,23,24,26,33. We used a two-stranded P3 version of the twister ribozyme, ES2 (derived from the ES ribozyme12) for further mechanistic study (Fig. 4a and Supplementary Fig. 12). The reaction rate is dependent on Mg2+ ion concentration, with an apparent affinity for Mg2+ ions of Kd = 2.8 mM. However, Mg2+ ions are not essential for activity as the cleavage reaction also proceeds in cobalt hexammine chloride and in high concentrations of monovalent ions12 (Supplementary Table 2). The pH dependence of cleavage rate has the shape of a flattened bell (Fig. 4b) and can be analyzed in terms of two ionizations, corresponding to apparent pKa values of 6.9 and 9.5 but with a further rate-limiting process that flattens the peak of the bell. The data are in good agreement with earlier studies12 and are consistent with a concerted general acid-base–catalyzed reaction (Supplementary Fig. 13). To identify nucleobases that contribute directly to catalysis, we examined the rates of twister ribozymes with substitutions of conserved nucleotides at three pH values (Fig. 4c). If a given nucleotide has a direct role in proton transfer in catalysis, we would expect that the rate of cleavage would be greater close to the pKa of the introduced nucleotide. It is notable that the only variant that showed a marked increase in activity at low pH was that with mutation G45A, i.e., a substitution of the guanine close to the putative active center in the structure. The base pair with A8 holds the Watson-Crick edge of G45 close to the scissile phosphate, with N1 free to participate in proton transfer. The N2 exocyclic amine of G45 donates hydrogen bonds to both N7 of A8 and to the pro-R oxygen of the scissile phosphate. Cleavage by a G45A ribozyme was 300-fold slower at neutral pH, but, in contrast to the native ribozyme, the G45A variant was maximally active at pH 5; the data fit apparent pKa values of
Crystal structure and mechanistic investigation of the twister ribozyme Yijin Liu, Timothy J Wilson, Scott A McPhee & David M J Lilley*
npg
© 2014 Nature America, Inc. All rights reserved.
We present a crystal structure at 2.3-Å resolution of the recently described nucleolytic ribozyme twister. The RNA adopts a previously uncharacterized compact fold based on a double-pseudoknot structure, with the active site at its center. Eight highly conserved nucleobases stabilize the core of the ribozyme through the formation of one Watson-Crick and three noncanonical base pairs, and the highly conserved adenine 3′ of the scissile phosphate is bound in the major groove of an adjacent pseudoknot. A strongly conserved guanine nucleobase directs its Watson-Crick edge toward the scissile phosphate in the crystal structure, and mechanistic evidence supports a role for this guanine as either a general base or acid in a concerted, general acid-base–catalyzed cleavage reaction.
R
NA-mediated catalysis is important in protein synthesis1, RNA splicing2 and a variety of other RNA processing events3 and provides insight into the chemical origins of life on the planet4. The nucleolytic ribozymes are a group of RNA species that undergo self-cleavage or ligation at a particular site. Originally discovered in plant pathogens5–7, these were mostly involved in processing replication intermediates and later were found to be a means of regulating genetic expression in bacteria8. However, it has recently become clear that ribozyme sequences are very widespread in many genomes, located within noncoding RNA sequences and strongly conserved and expressed inside the cell. Ribozymes of the hepatitis delta virus (HDV)/cytoplasmic polyadenylation element binding protein 3 (CPEB3) (ref. 9) and hammerhead10,11 families have been found in many eukaryotic species, including humans, and their location and expression indicate that they are functional in vivo. A new member of the nucleolytic ribozyme class called twister was very recently described12. Twister sequences are extremely widespread, with 2,700 examples distributed over bacteria, fungi, plants and animals including insects, fish and worms. The RNA exhibits nucleolytic ribozyme activity both in vitro and in vivo, with cleavage rates comparable to those of other nucleolytic ribozymes. The distribution and activity of these noncoding sequences suggests a role in genetic control. Twister ribozymes have a highly conserved secondary structure (Fig. 1a), with a core comprising a stem-loop interrupted by two internal loops. The cleavage site is located within loop L1, one nucleotide 3′ to the P1 stem. Phylogenetic analysis strongly indicates the presence of two long-range tertiary interactions, T1 and T2, so that the structure should fold into a double pseudoknot. Strongly conserved nucleotides are contained within both L4 and L1 (ref. 12), including ten nucleotides with >97% conservation. In the majority of the ribozymes, there is an additional stem-loop (P3) between T2 and P4, and in a few cases another stem-loop (P5) connects to the 3′ side of L2. Circular permutations exist, with a P1 stem-loop and either the P3 helix or the 3′ side of L2 being open12. The mechanism of action of catalytic RNA species is incompletely understood in general and has been sometimes controversial. In principle, there are two main ways of catalyzing phosphoryl transfer reactions, using general acid-base catalysis or metal ion catalysis,
and protein enzymes use one or the other mechanism. RNA enzymes similarly divide into these two groups, with the self-splicing introns and RNase P acting as metalloenzymes, whereas the nucleolytic ribozymes seem to use general acid-base catalysis13. Yet within the nucleolytic ribozymes, the detailed mechanism of catalysis and the functional groups involved are quite variable. All of them use one or more nucleobases, and guanines are involved in the mechanisms of all the ribozymes except the HDV ribozyme14–19. The VS and hairpin ribozymes use a combination of guanine and adenine nucleobases19–23. By contrast, the HDV ribozyme uses a cytosine nucleobase as a general acid in cleavage24–26 together with a hydrated metal ion to activate the nucleophile, acting either as a general base24 or Lewis acid27. Lastly, for the hammerhead ribozyme, crystallographic and mechanistic evidence suggests that a 2′-hydroxyl group may act as the general acid28,29. We have now solved a crystal structure of a basic form of the twister ribozyme. The RNA adopts a double-pseudoknot structure in which all of the conserved nucleotides form key structural elements. Mechanistic studies, together with our structure, support nucleobase-mediated general acid-base catalysis and point to the involvement of a conserved guanine.
RESULTS Crystallization and structure determination
We crystallized a basic P1 form of the twister ribozyme lacking both P3 and P5 helices based on the Osa-1-4 sequence from Oryza sativa. RNA of this sequence cleaved readily during transcription in vitro (Supplementary Results, Supplementary Fig. 1). The RNA was chemically synthesized with a 2′ H substitution at U6, thus removing the nucleophile to prevent cleavage occurring. A derivative with 5-bromocytosine substitutions in P1 and loop 2 was also synthesized. Both crystallized in the hexagonal space group P6522, and data were collected from the native and derivative crystals to 2.3 Å and 2.8 Å, respectively (Supplementary Table 1). The asymmetric unit contained a single ribozyme molecule. The bromine atoms were located using single-wavelength anomalous diffraction (Supplementary Fig. 2), from which models of the complete structure were built and refined. All except two nucleotides could be placed with confidence, together with a number of Mg2+ ions. Some conformational heterogeneity of the phosphodiester backbone was observed in the region of A7 (Supplementary Fig. 3).
Cancer Research UK Nucleic Acid Structure Research Group, Medical Sciences Institute/Wellcome Trust Building (MSI/WTB) Complex, University of Dundee, Dundee, UK. *e-mail: [email protected] nature CHEMICAL BIOLOGY | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
739
article
Nature chemical biology doi: 10.1038/nchembio.1587
a
b T1
29 30
25 24
P3
L4 23
c
T2
P3
P4
T2 40
32 15
P4
19 30
14
P5
19
T2
T2
13
P2 36
P2
G45
20
P4 A29
T1
G45
T1
31
12
L2
A29
P4
P1
P1
15
P2 T2 13
24
41
P2 7
45
8
45
46
28
29
T1
7
T1
L1
8
49
50
5
npg
© 2014 Nature America, Inc. All rights reserved.
1
5′
3′
54
G22
G30 C32 G31
G30 C32
G23
G31
25
A29
U24
49
P1
G22
G23
6
6
d
A29
U24
P1 54
1 5′
C25
3′
A28 C26
C27
C25
A28 C26
C27
Figure 1 | The sequence and structure of the twister ribozyme. (a) The sequence and secondary structure of the 54-nucleotide O. sativa Osa-1-4 twister as crystallized. Red nucleotides are >97% conserved. Grey nucleotides are not included in our final model. The arrow indicates the position of ribozyme cleavage. (b) Schematic to illustrate the connectivity of the twister ribozyme in the three-dimensional structure. (c) Parallel-eye stereoscopic image of the complete ribozyme, shown in cartoon mode. The color coding of the nucleotides matches that in a. The scissile phosphate is indicated by a red ball. (d) Parallel-eye stereoscopic image of the L4 loop, showing the 2Fo-Fc map of the twister ribozyme contoured at 2σ.
Fitting this to a double conformation improved the electron density map. The conformational difference is purely local; the overall structure of the ribozyme is indistinguishable in the two conformations.
The overall structure of the ribozyme
The overall structure of the twister ribozyme is fully consistent with the proposed12 secondary structure (Fig. 1b,c and Supplementary Fig. 4) and is distinct from all of the other known nucleolytic ribozyme structures15,25,30–32. The terminal loop L4 (Fig. 1d) participates in two tertiary interactions of opposite polarity with L1 and L2, generating the helices T1 and T2, respectively, so that the overall structure is a double pseudoknot. Helices P1, T1, P2 and T2 are coaxially aligned, although there is a large helical twist angle between helices P1 and T1. In addition, base pairs formed between A28 and A46 (trans Watson-Crick) and A8 and G45 (cis-Hoogsteen–sugar edge) are located coaxially in between T1 and P2. Indeed, the A28-A46 pair should be formally included within T1, and the A8-G45 pair should be included in P2. T2 is capped by a final Watson-Crick pair formed between C15 and G19 (Supplementary Fig. 5). This would be the first base pair of the P3 helix, if present, as there are almost invariably no intervening nucleotides between T2 and P3 or between P3 and P4 (ref. 12). Thus, the C15-G19 pair indicates how P3 must be connected in the majority of twister ribozymes (details below; Supplementary Fig. 6). In our basic form of the ribozyme, C15 and G19 are connected by A16, A17 and U18, although electron density for A17 and U18 was not visible in the 2Fo-Fc map, so these nucleotides were not included in our final model. The axis of helix P4 is approximately parallel to that of P1 through T2. It comprises four Watson-Crick pairs, including the essentially invariant G23-C32 base pair, and includes an additional trans Watson-Crick–Hoogsteen pair formed between U24 and A29. Indeed, every nucleotide of L4 is involved in base-base interactions, and all of the substitutions tested result in lower cleavage activity (Supplementary Table 2). Nucleotides A36 through A40 form the 74 0
connection between P4 and P2 and will be the location of P5, when present. A39 and A40 interact with the minor groove of T2 and the C15-G19 base pair, and more extensive A-minor interactions would be possible with an extended P3. Four well-defined magnesium ions are observed bound within the structure of the ribozyme fold (Supplementary Fig. 7). Two have complete inner coordination shells of water molecules, whereas two have exchanged some inner-sphere water ligands for phosphate nonbridging oxygen atoms.
The structural roles of the highly conserved nucleotides
Of the ten highly conserved nucleotides in the twister ribozyme, eight are found in three noncanonical base pairs (Fig. 2) and in one Watson-Crick base pair (Supplementary Fig. 8) in the core of the folded structure (Fig. 2). The formation of the A28-A46 trans Watson-Crick base pair at the end of T1 is facilitated by a turn in the backbone between C27 and A28 and enables A28 to accept a hydrogen bond from O2′ of A7, thus helping position A7 and presumably the scissile phosphate as well. Notably, the A28G variant ribozyme had much lower activity than the A46G variant (Supplementary Table 2), possibly because the A28G variant would project its N2 exocyclic amine into the major groove close to the scissile phosphate. Stacked on A28-A46 is the A8-G45 cis-Hoogsteen–sugar edge base pair of P2. This pair directly interacts with the scissile phosphate with the N2 exocyclic amine of G45, donating a hydrogen bond to the pro-R oxygen (Supplementary Fig. 8). The other two highly conserved base pairs are found at the end of P4 and seem to stabilize the formation of the two pseudoknots. The first is the G23-C32 Watson-Crick base pair, where C32 N4 donates a hydrogen bond to the pro-S nonbridging oxygen of the phosphate of C31, stabilizing the sharp turn in the backbone at the transition between T2 and P4. The potential importance of this bond is illustrated by the 104-fold lower activity of a C32U ribozyme (Supplementary Table 2). U24 and A29, the two nucleotides flanking T1, form a trans Watson-Crick–Hoogsteen base pair at the end of P4. The positioning of A29, which links T1 and T2, is further
nature chemical biology | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
article
Nature chemical biology doi: 10.1038/nchembio.1587 a
other nucleotide at position 40, but only a guanine at position 30 is able to donate the hydrogen bond to the sugar edge of the base at position 40. The other highly conserved nucleotide is A7. Its structural context is described in detail below.
A28
A46 A7
The conformation of the active site of the ribozyme
b G45
A8
c
A29
© 2014 Nature America, Inc. All rights reserved.
U24
Figure 2 | The structural role of conserved nucleotides in the twister ribozyme. (a) The A28-A46 trans Watson-Crick base pair at the end of T1. N3 of A28 accepts a hydrogen bond from the O2′ of A7. (b) The A8-G45 cis-Hoogsteen–sugar edge base pair of P2. The N2 exocyclic amine of G45 donates a hydrogen bond to the pro-R oxygen of the scissile phosphate. (c) The U24-A29 trans Watson-Crick–Hoogsteen base pair at the end of P4.
facilitated by the donation of a hydrogen bond from its O2′ to the pro-R nonbridging oxygen of the phosphate of C31. The highly conserved G30 is part of the second base pair of T2, which hydrogen bonds to A40 and presumably helps stabilize T2 (Supplementary Fig. 8). A similar interaction is possible with any a
The scissile phosphate linking dU6 and A7, and thus the active site of the ribozyme, is located in the center of the RNA structure, lying in the major groove of the T1-P2 helix (Fig. 3a). An O2′ atom modeled onto dU6 deviates ~90° from the in-line orientation of the O2′ nucleophile, P and O5′ leaving group that would be optimal for activity (Fig. 3b). This clearly cannot be the active geometry, but our structure provides an explanation for this. dU6 makes no interactions with this ribozyme, but instead its O4 accepts a hydrogen bond from G23 N2 of a symmetry-related ribozyme molecule in the crystal lattice (Supplementary Fig. 9). The extrusion of dU6 from the ribozyme core is essentially an artifact of crystallization and possibly results from the use of a deoxyribonucleotide to prevent cleavage. There can be no doubt that the active ribozyme has the global structure observed in the crystal, which is completely consistent with all of the phylogenetic data, and thus any deviation from the active structure is local. Moreover, A7 is very specifically positioned at the base of the P4 helix. A7 is stacked on the U24-A29 pair at the lower end of P4, and its O2′ is hydrogen bonded to A28 N3 (Fig. 3c). In addition, the two N6 protons are donated to the backbone nonbridging oxygen atoms P26 pro-R and P25 pro-S. The latter phosphate lies at the sharp turn between the P4 and T1 helices, which is stabilized by a Mg2+ ion (Supplementary Fig. 7). Collectively, these interactions with L4 hold A7 in a pocket in an unusual syn conformation with a C2′-endo sugar pucker. It was therefore likely that the lattice-induced extrusion of dU6 was responsible for the departure from an in-line trajectory. Examination of the immediate environment of the scissile phosphate suggested that in free solution it was likely that U6 could b
A29
A29 A7
A7
U24
U24
A8
npg
A8
G45
G45 A7
A46 dU6
G45
U6
c
A7
A46
dU6
G45
U6
d
A29
A29
U24
U24
G45 A7
A7
U6
C25 A28
A7
G45
A7
U6
C25 A28
C26 C27
C26 C27
Figure 3 | The active center of the twister ribozyme. (a) The immediate environment of the scissile phosphate between dU6 and A7 (cyan), showing the nearby G45 and A29 nucleotides. The P1 (blue) and T1 (green) helices are shown in space-filling representation. (b) A view into the active center, with an oxygen atom added in the 2′ position of U6 (red sphere), i.e., showing the position of the nucleophile. This is clearly far from the position required for inline nucleophilic attack on the scissile phosphate. (c) A7 is held in a pocket below the P4 helix, making interactions with the L4 nucleotides of the T1 helix. Electron density is the 2Fo-Fc map contoured at 2σ. (d) The structure was modified by a local rotation of U6, as explained in the text. The nucleobase is stacked with that of G45, and the O2′ added to U6 is close to an in-line trajectory. It is also close to N1 of G45. nature CHEMICAL BIOLOGY | vol 10 | september 2014 | www.nature.com/naturechemicalbiology
741
Nature chemical biology doi: 10.1038/nchembio.1587
npg
© 2014 Nature America, Inc. All rights reserved.
The chemical mechanism of the ribozyme
Like other nucleolytic ribozymes, the reaction follows an SN2related mechanism, yielding 2′,3′-cyclic phosphate and 5′-hydroxyl products (Supplementary Fig. 11)12. Efficient catalysis requires activation of the nucleophile and stabilization of the leaving group through proton transfer, and the precedent set by other nucleolytic ribozymes led us to anticipate that general acid-base catalysis might have a major role17,19,23,24,26,33. We used a two-stranded P3 version of the twister ribozyme, ES2 (derived from the ES ribozyme12) for further mechanistic study (Fig. 4a and Supplementary Fig. 12). The reaction rate is dependent on Mg2+ ion concentration, with an apparent affinity for Mg2+ ions of Kd = 2.8 mM. However, Mg2+ ions are not essential for activity as the cleavage reaction also proceeds in cobalt hexammine chloride and in high concentrations of monovalent ions12 (Supplementary Table 2). The pH dependence of cleavage rate has the shape of a flattened bell (Fig. 4b) and can be analyzed in terms of two ionizations, corresponding to apparent pKa values of 6.9 and 9.5 but with a further rate-limiting process that flattens the peak of the bell. The data are in good agreement with earlier studies12 and are consistent with a concerted general acid-base–catalyzed reaction (Supplementary Fig. 13). To identify nucleobases that contribute directly to catalysis, we examined the rates of twister ribozymes with substitutions of conserved nucleotides at three pH values (Fig. 4c). If a given nucleotide has a direct role in proton transfer in catalysis, we would expect that the rate of cleavage would be greater close to the pKa of the introduced nucleotide. It is notable that the only variant that showed a marked increase in activity at low pH was that with mutation G45A, i.e., a substitution of the guanine close to the putative active center in the structure. The base pair with A8 holds the Watson-Crick edge of G45 close to the scissile phosphate, with N1 free to participate in proton transfer. The N2 exocyclic amine of G45 donates hydrogen bonds to both N7 of A8 and to the pro-R oxygen of the scissile phosphate. Cleavage by a G45A ribozyme was 300-fold slower at neutral pH, but, in contrast to the native ribozyme, the G45A variant was maximally active at pH 5; the data fit apparent pKa values of
