Structure of a bacterial ribonuclease P holoenzyme in complex with tRNA

Abstract

Ribonuclease (RNase) P is the universal ribozyme responsible for 5′-end tRNA processing. We report the crystal structure of the Thermotoga maritima RNase P holoenzyme in complex with tRNA^Phe. The 154 kDa complex consists of a large catalytic RNA (P RNA), a small protein cofactor, and mature tRNA. The structure shows that RNA-RNA recognition occurs through shape complementarity, specific intermolecular contacts, and base pairing interactions. Soaks with a pre-tRNA 5′ leader sequence with and without metal help identify the 5′ substrate path and potential catalytic metal ions. The protein binds on top of a universally conserved structural module in P RNA and interacts with the leader, but not with mature tRNA. The active site is composed of phosphate backbone moieties, a universally conserved uridine nucleobase, and at least two catalytically important metal ions. The active site structure and conserved RNase P/tRNA contacts suggest a universal mechanism of catalysis by RNase P.

Ribonuclease P (RNase P) is a ribonucleoprotein complex responsible for processing many different RNA molecules in the cell (for recent reviews, see 1–3). It is found in almost all organisms and is composed of one essential RNA subunit and one or more protein subunits. The RNA component is responsible for catalysis and can process RNA in vitro in the absence of protein, albeit with reduced efficiency⁴. The discovery that the RNA component is the catalytic moiety⁴ helped cement the notion that RNA can be directly involved in catalysis. RNase P is considered a remnant of an ancient RNA-based world and an example of an RNA-based catalyst with many features in common with protein-based catalysts.

RNase P recognizes its substrate in trans and is a multiple turnover enzyme. The preferred substrate is pre-tRNA and recognition involves features distant from the cleavage site, such as the TΨC loop of the tRNA acceptor stem⁵. RNA cleavage requires divalent metals^4,6,7, yet the chemical mechanism and the location of the active site remain largely undefined as well as the exact role of the protein components. In the case of bacterial RNase P, the single essential protein improves the reaction rate by 2 to 3 orders of magnitude^8,9, helps to stabilize the active P RNA fold^8,10, binds the 5′ leader region of the pre-tRNA substrate^11,12, and assists in product release¹³.

Structural studies of the RNA component reveal a two domain (S- and C-domains) molecule formed by single and coaxial stems linked together by a variety of tertiary interactions^14–17, including five conserved regions I to V (CR-I to CR-V) of P RNA that are common to all organisms¹⁸. These conserved regions cluster into two areas, one involved in substrate recognition and the other forming the active site scaffold¹⁹.

Here we present the crystal structure of Thermotoga maritima RNase P holoenzyme in complex with mature tRNA^Phe, and also the structure of the complex in the presence of a post-cleavage tRNA leader. The two structures help answer key questions about the mechanism of this crucial ribozyme with implications for a broader understanding of the general mechanisms of RNA-RNA based recognition and catalysis.

Structure determination

The components of the complex were purified separately and assembled by mixing and heating prior to crystallization (see full methods). The pre-tRNA was processed into mature tRNA and hence the structure represents a ribozyme/product complex. To promote crystal formation, two interaction modules²⁰ were introduced, which did not affect catalytic activity (Fig. S1 and Table STI). The crystals diffract anisotropically to 3.8 Å and ~4.0 Å. An initial 6 Å map was obtained from phases from a Ta₆Br₁₂ derivative; these phases helped locate heavy atoms in other derivative data sets. MIRAS phases produced an excellent map to 4.1 Å where all three components were visible (Figs. 1, S2–3, and Tables STII-IV). Density was particularly clear for the RNA molecules, while density was only clear for the protein backbone and hence the high resolution model of the T. maritima protein²¹ was positioned without significant rebuilding. The P RNA was built into the map using the structures of T. maritima¹⁷ and B. stearothermophilus¹⁴ P RNA as guides, while T. maritima tRNA^Phe utilized yeast tRNA^Phe as a guide²². The structure was refined using anisotropic data to 3.8 Å resolution. Crystals with a tRNA leader present were obtained by soaking a short, oligonucleotide with and without samarium chloride and this structure was refined to 4.2 Å resolution.

Figure 1. Crystal structure of the T. maritima RNase P holoenzyme in complex with tRNA.

a, Structure of bacterial RNase P, composed of a large RNA subunit (338 nucleotides, ~110 kDa) and a small protein component (117 amino acids, ~14.3 kDa), in complex with tRNA (76 nucleotides, ~26 kDa). The RNA component serves as the primary biocatalyst in the reaction and contains two domains, termed the catalytic (C, blue) and specificity (S, light blue) domains. The RNase P protein (green) binds the 5′ leader region of the pre-tRNA substrate and assists in product release. Transfer RNA (tRNA^Phe) (red) makes multiple interactions with the P RNA (see Fig. 2 and S1 for details). Regions in grey denote additional RNA nucleotides required for crystallization. b, Alternate view of the RNase P/tRNA complex, identifying the tRNA recognition regions: the 5′ end where catalysis occurs, the 3′ CCA end, and the highly conserved TΨC and D loop regions. c, View of the 4.1 Å experimental electron density map centered on the 5′ end of tRNA. The map is represented as a dark grey mesh, contoured at 1.4 rmsd.

Overall structure

In the complex, the tRNA sits with the acceptor stem against RNase P, making several tRNA-P RNA intermolecular contacts (Figs. 1, S1). The TΨC and D loops of the tRNA contact the S-domain, while the acceptor stem extends from the S-domain into the C-domain crossing the main P1/P4/P5 coaxial stem (Figs. 2, S4). The 3′ CCA end of the tRNA enters a tunnel formed by P6/P15/P16/P17 and base pairs with nucleotides in the L15 region (Figs. 2, S4, S5), an interaction recognized previously²³. The 5′ end of the tRNA indicates the location of the active site, which is close to the region where P4, P5, and CR-IV intersect. The protein component is also adjacent to the 5′ end of the tRNA, but does not contact it. The protein contacts include the CR-IV and CR-V regions, the P15 stem, and the P2/P3 helix interface (Figs. 3, S1, S6). The pre-tRNA leader makes extensive contacts with the protein, but few with the PRNA.

Figure 2. tRNA recognition by RNase P is mediated by RNA-RNA interactions.

a, Schematic of the P RNA secondary structure mapping the tRNA-P RNA contacts observed in the crystal structure. The tRNA nucleotides (1•72, 2, 3, 64, and 65) and regions (5′, 3′, TΨC loop, D loop, and acceptor) involved in direct interactions are shown in red. Intermolecular base pairs form between the 3′ end of tRNA (DCCA) and loop 15 (L15), where D is the discriminator nucleotide that serves as an identity element in tRNA biogenesis. P RNA nucleotides that are universally conserved (black, uppercase), conserved among all bacteria (grey, uppercase), or highly conserved in bacteria (black, lowercase) are identified. Metal ions are shown as filled pink circles, and denote the location of the active site (M1, M2), and other structurally important regions (M3, M4). Single and double dashes in red represent minor groove and base stacking interactions, respectively. All identified tRNA/P RNA contacts are within 4 Å. The crystallized T. maritima P RNA consists of eighteen paired helices (P), five universally conserved regions (CRI to CR-V) (black), two junctions containing conserved nucleotides in bacteria (dark grey), several loop (L) regions, and an engineered tetraloop region (T, light grey). The coaxial P1/P4/P5 stem is shown in blue, P2/P3 stems in cyan, P6/P15/P16 and L15/L17 in yellow, P7 and P10/P11/P12 in orange, P8/P9 in light green, and P13/P14 in pink (see Fig. S1 for additional details). b, Recognition of tRNA by the P RNA of RNase P. The acceptor stem of tRNA (red) docks onto the P RNA (colored as in a) making a series of interactions, including base stacking in the TΨC/D loops of tRNA and the S-domain, an A-minor interaction, and base pairing, ribose zippers, and stacking interactions between the 5′ and 3′ ends of tRNA and the C-domain. The protein (green) makes no direct contacts with mature tRNA. Critical metals ions (M1–M4) identified are shown as magenta spheres. c, tRNA recognition by the S-domain. Two universally conserved P RNA regions (CR-II and III, dark grey) facilitate base stacking interactions with unstacked bases in the structurally conserved TΨC and D loops of tRNA. Dashed circles highlight this stacking interaction between P RNA residues A112, G147 and tRNA residues G19, C56. A conserved P RNA adenosine (A198) stacks into the minor groove of the acceptor tRNA stem. d, Recognition of the tRNA 3′ CCA by the C-domain. Intermolecular base pairs form between the 3′ tRNA (ACC) and the L15 (GGU) loop of P RNA. This interaction is stabilized by a structural metal (M3, magenta sphere) and a L15 ribose zipper conformation.

Figure 3. Protein-RNA contacts within the RNase P holoenzyme.

a, The protein sits on the P RNA surface formed by conserved regions I, IV, and V. The protein (green, shown as ribbons) additionally contacts the L15/P15 junction and the P2/3 helices (P RNA as colored in Fig. 2). Labeled P RNA nucleotides make protein contacts (within 4 Å) and include: A45 in CR-I, U257 and G258 in the L15/P15 junction, U293, U294, G295, A296, U297 in CR-IV, and A311, G312, and A313 in CR-V. Bold nucleotides are universally conserved. b, Surface representation of the protein colored by sequence conservation (Variable: tan, Neutral: light green, Conserved: green). A highly conserved patch in the protein extends from the vicinity of the 5′ end of the tRNA, and interacts with P RNA conserved regions IV (U293–U297) and V (A311–A313). Other P RNA nucleotides that make protein contacts include: the P2 helix (C18-G22, G298-A299), the P3 helix (G37), and the L15/P15 junction (U257-G258). Four hundred and ninety bacterial RNase P proteins were included in the analysis of the sequence conservation using the ConSurf server⁴⁸. Panels (c) and (d) show different orientations to emphasize that high sequence conservation is concentrated in the region of the protein that faces the conserved regions of the P RNA. Neutral or slightly conserved regions shown in these two orientations correspond to a patch that interacts with the leader.

The components of the RNase P holoenzyme are largely unchanged when bound to tRNA (Fig. S7). A comparison between T. maritima P RNA alone¹⁷ and in the complex reveals an overall similar fold (backbone normalized rmsd ~1.1Å) with a small change in the relative orientation of the two domains (Fig. S7). The only major change in the P RNA structure occurs in the vicinity of the P15–P17 stems (Figs. S8, S9). A few additional residues at the N-terminus were clear and follow a similar path to the B. subtilis protein²⁴ (Fig. S10); no changes in the structure of the protein component were detected. The structures of yeast and T. maritima tRNA^Phe show remarkable resemblance (backbone normalized rmsd for acceptor stem ~0.8 Å) (Fig. S11). Further, a comparison with previous models reveals an excellent agreement with the predicted secondary structure²⁵ and a good agreement with the models of the complex^19,26,27 (Fig. S12).

tRNA recognition

The observed RNA-RNA interactions involved in substrate recognition agree with previous biochemical studies^5,23,28 and include: 1) stacking between bases in the tRNA TΨC and D loops and the P RNA S-domain, 2) an A-minor interaction at the acceptor stem, and 3) the formation of canonical base pairs at the 3′ end of tRNA (Figs. 2, S1). The first interaction identifies the TΨC loop as a key element in recognition. Both the tRNA D and TΨC loops have unstacked bases (G19 and C56) that interact with unstacked bases in the P RNA (A112 and G147), forming G19-A112 and C56-G147 stacks in the complex. The second major interaction involves a highly conserved unstacked adenosine (A198) in the P11 stem entering the minor groove of the tRNA acceptor stem. These interactions facilitate shape complementarity and help explain the central role of the S-domain in recognition. The third major interaction involves intermolecular base pairing between the tRNA 3′ CCAD motif and the L15 loop. This interaction is likely conserved in all bacterial and most archaeal RNase Ps, but not in organisms where CCA is added post-transcriptionally¹. The fourth to last nucleotide, A73, forms a Watson-Crick (WC) base pair with nucleotide U256. C74 and C75 form WC base pairs with G255 and G254, while the terminal A76 forms a weak interaction with G253. To accommodate these intermolecular base pairs, the two strands of L15 fold into a ribose zipper. In addition, a structural metal ion (M3) (Figs. 2, S13) binds adjacent to this P RNA-tRNA region and likely corresponds to a metal ion identified biochemically⁷. In the complex, the 3′ end of the tRNA separates from the 5′ end and enters a wide opening formed by P6/P15/P16/P17 (Figs. 1, 2, S4, S5). This opening is ~20 Å in diameter, can easily accommodate a single stranded RNA molecule, and is created when the P6/L17 pseudoknot forms (Figs. S1, S5).

Protein/RNA interactions

The bacterial RNase P protein structure is highly conserved, but has little or no sequence or structural similarity with the protein components of archaea or eukarya²⁹. In the complex, the protein is near the 5′ end of tRNA, but is too far (over 6 Å) to make direct contacts. The protein sits between the P15 and P3 stems (Figs. 3, S6), and also contacts the CR-IV and CR-V loop regions of P RNA. Comparison of bacterial sequences reveals that the protein has a large, contiguous area with high sequence conservation (Figs. 3, S6) including important residues identified previously^11,30,31. The conserved area extends in an arch along the surface of the protein, starting from a point close to the 5′ end of the tRNA and faces the universally conserved modules.

To investigate the interactions with the leader, crystals were soaked with a short oligoribonucleotide in the presence and absence of Sm³⁺. Fourier difference maps to 4.2 Å show 5 phosphates of the leader along the conserved surface of the protein (Figs. 4, S6), but the position of the nucleobases was ambiguous. The structure shows that the leader contacts residues Phe17, Phe21, Lys51, Arg52, and Lys90 and likely interacts with Ser26, Gln28, Lys56, Arg89, in agreement with biochemical results^6,9,12,30,31. The 3′ end of the leader is located adjacent to the 5′ end of tRNA and near two conserved residues (Arg52 and Lys56). A metal ion is present in between the leader 3′ and the 5′ end of mature tRNA (Figs. 4, 5, S14), but is too distant (> 4 Å) to directly ligate protein residues. Leader nucleotides −1 to −3 are poised to interact with nucleotides A213, U294, G295, and A314 of P RNA (Figs. 4, S1, S6). These results indicate that the major role of the protein component is to interact with the leader to align the pre-tRNA in the complex, as previously observed^9,11,31.

Figure 4. Pre-tRNA leader - protein interactions in the RNase P holoenzyme.

a, Surface representation of the protein colored by sequence conservation as in Fig. 3. The pre-tRNA 5′ leader (purple, with purple and orange spheres for the phosphorous and non-bridging oxygens, respectively) was modeled as a polyphosphate chain with five phosphates (P_-1 to P_-5). The leader follows a highly conserved patch in the protein extending from the 5′ end of the mature tRNA (red) and away from the P RNA. The addition of a 5′ leader with metal (Sm³⁺) reveals a second metal ion (M2). b, Alternative view of the pre-tRNA leader/protein interaction. Each phosphate position (P_-1 though P_-5) was visible in a 4.2 Å difference Fourier map (mF_o-DF_c) calculated from crystals where only the leader was soaked into the crystals (blue mesh, 3 rmsd contour levels). A second 4.2 Å difference Fourier map (mF_o-DF_c) calculated from crystals where the leader and Sm³⁺ metal were soaked into the crystals shows clearly the position of the second metal ion (magenta mesh, 3.5 rmsd contour level). P RNA residues poised to make contacts are labeled. Nucleotide U52 serves as a reference point in a and b and does not interact with the 5′ leader oligonucleotide.

Figure 5. Structure of the RNase P active site environment.

a, The active site is inferred from the location of the mature 5′ end of tRNA. The diagram shows the position of the mature tRNA (red), the leader (purple), the protein component (green), and the P RNA (blue and grey). A group of conserved P RNA nucleotides (A49 - U52, A213, A313, and A314) form part of the active site. Two metal ions (magenta spheres) are found in the active site. b, The two active site metal ions (M1 and M2) are within 4 Å of the 5′ phosphate of tRNA and the M1–M2 metal-metal distance is ~4.8 Å. The M1 metal makes contacts (≤2.1 Å, solid grey bonds, labeled) with tRNA (G1 O1P) and P RNA (A50 O1P and U52 O4) oxygens. Other possible ligands within 3.5 Å of M1 or M2 are represented by dashed grey lines (Table STV). The figure shows two isomorphous difference Fourier (mF_o-DF_c) maps. The green mesh corresponds to a Eu³⁺ soak in the absence of leader and is contoured at the 9.5 rmsd level. The magenta mesh corresponds to a Sm³⁺ and 5′ leader soak and is contoured at the 5.5 rmsd level. The second metal is clearly visible only when the leader is present. c, Schematic diagram of the interactions around the active site. The diagram shows all residues within 8 Å of the 5′ phosphorus atom of tRNA. Short dashed lines represent metal ligand distances within 2.2 Å and longer dashed lines represent nucleotides which form canonical base pairs. Nucleotides in bold are universally conserved in P RNA. The P RNA, tRNA, 5′ leader, and protein side chains are shown in blue, red, purple and green, respectively. d, Proposed reaction mechanism for the endonucleoytic cleavage of pre-tRNA by RNase P based on the structure of the enzyme-product (E-P) complex and previous mechanistic studies⁴⁹,⁵⁰. The M1 metal distance to the 5′ phosphate ligands (Table STV) in the E-P complex is consistent with the proposed enzyme-substrate (E-S) transition state. In this proposed reaction scheme, M1 is ~180° from the apical O3′ position and activates a hydroxyl nucleophile for an in-line nucleophillic displacement, creating a new bond and displacing the 3′ scissile phosphate oxygen. As RNase P proceeds through an S_N2 reaction pathway, the stereochemistry around the phosphorus atom undergoes a net inversion of configuration. If the pro-R_P (O2P) oxygen coordinates metal in the E-S complex during catalysis, as previously observed⁴⁹,⁵⁰, this would subsequently allow for the pro-S_P (O1P) oxygen to coordinate metal in the E-P complex, as observed in the crystal structure. Product release could be facilitated by a metal (M2) coordinated water, which would enable proton transfer to the 3′ scissile oxygen. The exact active site geometry and identity of other metal ligands in an E-S complex has yet to be established.

Active site

The location of the active site is inferred from the 5′ end of mature tRNA (Figs. 5, S15). The phosphate backbone of tRNA nucleotides (+1 to +3) sits on the major groove of the P4 stem (near A50, G51, and U52), and places the tRNA 5′ end next to the P4 phosphate backbone and nucleotides A313 and A314 (Figs. 5 and S15). The universally conserved U52 nucleotide is unstacked from the P4 stem and faces the tRNA 5′ end. In addition, the tRNA 1•72 base pair is stabilized by an adenosine stack with A213, a nucleotide conserved in all bacteria.

A metal ion (M1), putatively magnesium, is found trapped between the tRNA 5′ end, the A50 and G51 phosphates, and the O4 oxygen of the universal U52 nucleotide and was confirmed using crystals soaked with Sm³⁺ and Eu³⁺ (Figs. S14 and S15). Putative M1 metal contacts include the A50 non-bridging phosphoryl oxygen, the O4 oxygen of the U52 nucleobase, and the O1P oxygen at the 5′ end of tRNA. Other metal-ligand interactions may include: the backbone of A50, the phosphoryl oxygen of G51, and the 5′ end of tRNA (Table STV). Many of these oxygen ligands have been implicated in metal coordination and catalysis^32–34. The M1 site may also coincide with a site (M6) observed in the structure of B. stearothermophilus P RNA³⁵. The structure of the complex suggests that M1 participates in catalysis by directly binding P RNA and the 5′ phosphate of tRNA.

A second metal (M2) was located in experiments where the leader was soaked in the presence of Sm³⁺. The M2 metal is in close proximity to the phosphoryl oxygens of G51, the O3′ of the leader, and the 5′ end of tRNA (Table STV). The two metals observed in crystals soaked with the leader and Sm³⁺ are ~4.8 Å apart (Figs. 5, S15). The structures indicate that the active site includes at least two metal ions upon complex formation with pre-tRNA. Due to its location, the M2 metal ion could make additional contacts with both the tRNA and the P RNA during catalysis.

The structures of the active site of the complex and the apo-ribozyme structures are similar (Figs. S7, S16, and S17), including the presence of a metal ion next to the P4 helix³⁵. With the exception of the U52 nucleobase (Fig. S16), no large changes are observed in the active site region. A fully occupied M2 site is observed only in the presence of leader, suggesting that a local metal-dependent conformation change may occur, as previously reported⁶. The structure also reveals that the tRNA 5′ and 3′ ends splay and separate to interact with the P RNA (Fig. S11), confirming the need for movement of the tRNA ends^36,37. While accommodating the upstream RNA leader likely requires local protein and P RNA structural changes, the location of the active site is not significantly altered and is largely pre-assembled.

Mechanistic implications

RNase P can cleave a variety of substrates^1,10,38, but pre-tRNA is the only one that is common among all organisms. To decipher its function, it is important to understand two different aspects of pre-tRNA processing by RNase P: substrate specificity and the chemical mechanism of cleavage.

tRNA recognition by RNase P involves the highly conserved tRNA TΨC and D loops and the CR-II and CR-III in the S-domain of P RNA. Thus, regions with high sequence and structure conservation are involved in specific tertiary interactions, suggesting a universal mode of recognition among all RNase P’s. The presence of unpaired nucleotides next to the cleavage site is also an important feature for pre-tRNA recognition, although it is unclear whether this is a universal feature of all natural substrates¹. Finally, pre-tRNA is usually processed to form a 7 base pair long acceptor stem. An additional role of the interactions between CR-II and CR-III and tRNA may be to serve as a ‘ruler’ that ensures that the correct lengths are processed, although there is some flexibility as tRNA’s with acceptor stems 8 base pairs long can be processed³⁹. The interaction with the 3′ CCA end is also a key recognition feature, but may not be necessarily an RNA-RNA interaction in higher organisms. The L15 loop of P RNA is not found in eukarya or some archea⁴⁰ and its function may be replaced by additional protein(s), suggesting that 3′ CCA intermolecular base pairing is not a universal interaction.

The second important aspect of RNase P function is the chemical mechanism of cleavage. Hydrolysis of a phosphodiester bond generates the mature 5′ RNA product. While it is not possible to propose a complete mechanism from a structure at this resolution, the RNase P-tRNA structures, together with extensive biochemical information, help identify the major active site components. The structure suggests that at least two distinct metals play a direct role. It is possible to propose a transition state model (Fig. 5d) where the M1 metal directly positions the scissile phosphate oxygens of the substrate and enables a hydroxyl ion to perform an S_N2-type nucleophilic substitution. In this scenario, the M2 metal ion stabilizes the transition state and mediates proton transfer to the 3′ scissile oxygen during product release, as previously proposed⁷. Other universally conserved nucleotides in the vicinity appear to play a structural role in forming the correct structure and are not directly involved in catalysis, consistent with proposals that sequence conservation is largely the result of strong structural constraints¹⁹. Hence, the RNaseP/tRNA complex reveals how the P RNA structure can serve as a scaffold to bind and orient metals and substrate properly. It appears that RNase P employs a two-metal ion catalytic mechanism, similar to other mechanisms proposed based on other large ribozyme structures^41,42 and originally put forth as a general mechanism for many ribozymes⁴³.

The structural studies of the holoenzyme/tRNA complex help to show that all RNase P ribozymes share a common, RNA-based mechanism of RNA cleavage and recognition that involves two universally conserved structural modules. Adaptation through the addition of protein increases RNase P functionality by accurately positioning the 5′ leader pre-tRNA substrate and by contacting conserved regions of the P RNA structure. The unique tertiary fold of the P RNA utilizes shape complementarity, specific RNA-RNA contacts, and intermolecular base pairing to recognize its substrate efficiently. Within this tertiary fold, the universally conserved regions are crucial to form the active site scaffold and to create regions involved in tRNA recognition. In addition, both P RNA and the pre-tRNA help to coordinate two catalytically important metal ions essential for the putative mechanism of pre-tRNA cleavage. The RNase P/tRNA complex offers a glimpse into the transition from an ancient, RNA-based world to the present, protein-catalyst dominated world and affirms that RNA molecules can display comparable versatility and complexity.

Methods summary

Crystallization

Preparation, purification, and folding of T. maritima RNase P and tRNA^Phe have been described^8,17,44. For crystallization, the components were mixed in a 1:1.1:1 (P RNA: pre-tRNA: protein) molar ratio to a concentration of 45 μM. The mixture was heated to 94 °C (2 min), cooled to 4 °C (2 min), and after the addition of MgCl₂ to a final 10 mM concentration, further incubated at 50 °C (10 min) and 37 °C (40 min). Crystals were obtained by mixing 1 μl of complex with 1 μl of reservoir solution (1.8 M Li₂SO₄, 50 mM sodium cacodylate (pH 6.0)) and equilibrated by vapor diffusion at 30 °C. Crystals were cryo-protected using 15% xylitol plus reservoir solution.

Data collection and structure determination

Diffraction data were collected at 100 °K at the LS-CAT sector at the APS. Complete native and Ta₆Br₁₂, SmCl₃, EuCl₃, and iridium hexammine (Ir(NH₃)₆)³⁺ derivatives were collected. A weak Molecular Replacement⁴⁵ solution using a trimmed model of the tRNA/P RNA complex¹⁹ located the Ta₆Br₁₂ cluster. MAD phases⁴⁶ from the cluster extended to ~6 Å, with the map showing a clear envelope. These phases were used to locate the other heavy atoms that were used to calculate a 4.1 Å MIRAS map. To locate the pre-tRNA leader, crystals were soaked with a T. maritima 5′ tRNA 7 nucleotide leader sequence (final 0.2 mM concentration), with and without 14 mM SmCl₃. Difference maps allowed the placement of five pre-tRNA nucleotides and the unambiguous identification of a second active site metal. The experimental electron density map was of excellent quality and allowed model building of nearly all RNA phosphate and nucleobase positions and accurate placing of the protein. Model building was guided by the known structures^14,17,21. Final R_work and R_free are 24.9% and 27.0%, respectively, with rmsd of 0.007Å and 1.24° for bonds and angles. Figures were made with Pymol⁴⁷.

Full Methods

Preparation of the T. maritima RNase P holoenzyme - tRNA^Phe ternary complex

RNA transcriptions were performed in vitro using purified His6-tagged T7 RNA polymerase using standard protocols⁵¹. Sequences from the T. maritima RNase P RNA and tRNA^Phe genes were inserted into a pUC19 vector at Fok I and Bsm AI restriction sites, respectively, allowing for run-off transcription of the DNA plasmid after digestion with the appropriate restriction enzyme (NEB). Constructions of modified RNA molecules with either mutations or additions were performed using the QuickChange mutagenesis kit (Strategene). RNA samples were purified by 6% denaturing polyacrylamide gel electrophoresis (PAGE), identified by UV absorbance, recovered by diffusion into 50 mM potassium acetate (pH 7) and 0.2 M potassium chloride, and precipitated with ethanol. tRNA was further purified by anion exchange [MonoQ (5/50 GL)] and gel filtration [HiPrep 26/60, Sephacryl S-200] chromatography (GE Health Sciences). Overexpression and purification of the RNase P protein from T. maritima was performed as described previously⁴⁴.

To form the RNase P holoenzyme-tRNA complex, unfolded P RNA, unfolded tRNA, and P protein molecules were mixed at a 1:1.1:1 molar ratio in 66 mM HEPES, 33 mM Tris (pH 7.4), 0.1 mM EDTA (1X THE), and 100 mM NH₄OAc (Ref. ⁸). The ternary mix, at a final concentration of 45 μM, was incubated at 94 °C for 2 minutes and then cooled to 4 °C over 2 minutes. After addition of MgCl₂ to a final 10 mM concentration, the reaction mixture was incubated at 50 °C for 10 minutes, followed by incubation at 37 °C for 40 minutes, and finally cooled to 4 °C over 30 seconds.

Rational design of an RNA tertiary module to build a crystal lattice

In order to promote formation of a crystal lattice, intermolecular interactions were facilitated by introducing a tertiary structure interaction module. Based on the T. maritima RNA sequence and a proposed model of the P RNA-tRNA complex¹⁹, constructs were designed where a tetraloop was inserted into the P12 loop (L12) of the P RNA and a tetraloop-receptor into the anticodon stem of tRNA (Fig. S1). These two RNA regions were chosen as they were deemed to be far from the active site or other regions involved in specific interactions. In addition, the P12 stem of P RNA has a highly variable helix length across all organisms, lacks sequence conservation, and is non-essential or absent in several organisms⁴⁰. The P12 and the anticodon loop of tRNA are not known to form any functional contacts. The length of the anticodon and the P12 stems were systematically varied by single base pair insertions adjacent to the tetraloop and tetraloop receptor module, thus altering the position (~2.7 Å per base pair added) and orientation (~36o per base pair added) of the tetraloop receptor and the tetraloop. Forty two combinations of molecules were screened for crystallization conditions using a sparse matrix approach employing a set of crystallization conditions developed locally. A few combinations of RNA molecules produced crystals, with most of them diffracting poorly. The best crystals were obtained from a construct where the P12 and anti-codon stems were elongated by five and three base pairs respectively. Insertion of two G-U wobble pairs adjacent to the tetraloop-tetraloop receptor module further improved diffraction, and also created a binding site for an iridium hexammine cation.

Crystallization and Data Collection

Crystals were obtained by mixing 1 μL of complex with 1 μL of reservoir solution (1.8 M LiSO₄, 50 mM sodium cacodylate (pH 6.0)) and equilibrated by vapor diffusion hanging or sitting drops at 30 °C. Gel analysis of washed crystals show that all three components were present (data not shown). Attempts to crystallize the complex in the absence of protein yielded no crystals. Crystals suitable for data collection grew in approximately three weeks and were cryo-cooled in liquid nitrogen immediately after transfer to a 15% xylitol plus reservoir solution. Crystals of the RNase P holoenzyme/tRNA ternary complex suitable for data collection grew to approximately ~80–300 μm per side/edge, and diffract anisotropically to 3.8 Å in the best direction and ~4.0 Å in other directions. Crystals belong to space group P3₁21 (a=b=169.3 Å, c=185 Å) and contain 1 molecule per asymmetric unit.

A series of derivatized crystals were also prepared by soaking in heavy metal compounds. Derivatives were prepared by soaking the crystals in mother liquor plus the derivative and incubating for 2–24 hours before transferring them to cryo-protectant with the derivative present and freezing them in liquid nitrogen. Successful derivatizations were obtained by soaking the crystals in the following compounds: 2 mM Ta₆Br₁₂, 15 mM samarium chloride (Sm³⁺), 15 mM europium chloride (Eu³⁺), and 15 mM iridium hexammine (Ir(NH₃)₆)³⁺). However, several of the compounds partially precipitated upon addition to the mother liquor solution and hence the final concentration is not known precisely. In addition, crystals with a leader present were obtained by soaking in a 0.2 mM 7-mer oligonucleotide (5′-A_-7 A_-6 G_-5 G_-4 C_-3 G_-2 U_-1 -3′) (ThermoFisher) for 4 hours with and without 14 mM samarium chloride present. The sequence was chosen by selecting the most common nucleotide in the T. maritima tRNA leaders at each position.

All diffraction data were collected at 100 °K at the Life Science-Collaborative Access Team (LS-CAT) sector located at the Advance Photon Source (APS) (Argonne, IL) using Rayonix CCD detectors. As the crystals are very radiation sensitive, the data collection range was optimized using the program MOSFLM⁵² to collect the most complete native or anomalous data set using the minimal rotation range. Multi-wavelength anomalous dispersion (MAD) data were collected from a tantalum bromide cluster (Ta₆Br₁₂) derivative at three different wavelengths. Single or multiple wavelength anomalous dispersion data were also collected fromthe samarium chloride (Sm³⁺), europium chloride (Eu³⁺₃₊), and iridium hexammine (Ir(NH₃ )₆) derivatives. Data were processed with XDS⁵³ and scaled with SCALA⁵⁴. All other processing was done with programs from the CCP4 suite⁵⁴, except when noted. Data collection statistics for native and derivative data sets are shown in Table STII.

In all cases, the diffraction limits of the data were anisotropic. The extent of the anisotropy was determined using the Anisotropy Server⁵⁵ and the data were treated in three different ways: 1) without any anisotropy correction, 2) carving the data to the limits suggested by the anisotropy server (3σ cut-off level on amplitudes), and 3) applying an anisotropic correction to the data using the server. For the second case, the integrated data from XDS was carved to the limits suggested by the server and then merged and scaled with SCALA before final processing. In many instances, the phasing and refinement calculations were done separately with the complete and carved data sets and the results compared. Overall, the different ways of treating the data had little effect on the final results, even though the data collection statistics were better for the carved data set (see Table STII and STIII).

Structure determination and model refinement

Molecular Replacement (MR) studies with the program PHASER⁴⁵ using a proposed partial model of the P RNA/tRNA complex¹⁹ gave a weak low resolution (25–8 Å) MR solution (Z-scores: 5.4 and 9.0 for the rotation and translation functions respectively). Phases calculated from the MR solution were used to locate the position of the 3 sites in the Ta₆Br₁₂ cluster data set. The program SHARP⁴⁶ was used to calculate MAD phases using data from three different wavelengths and spherically averaged form factors for the cluster. The solvent-flattened MAD map was of excellent quality but the phases were only good to ~6 Å resolution. The positions of the Eu³⁺, Sm³⁺, and (Ir(NH₃ )₆)³⁺ heavy atoms were determined using the cluster phases. The parameters from the cluster and other derivatives could not be refined simultaneously and instead Multiple Isomorphous Replacement with Anomalous Scattering (MIRAS) phases to 4.1 Å resolution were calculated using data from the single-atom derivatives together with phase information to 6 Å from the cluster data. The SOLOMON⁵⁶ solvent-flattened map was very clear (Fig. S2) and all three molecules were apparent in the map. The model for the P RNA/tRNA complex¹⁹ fit well in many areas, but the map showed regions where the model needed to be changed, regions that were missing in the model, like the P12 extension and the pseudoknot region, and the position of the protein. The models for the tRNA and P RNA were rebuilt completely using the high resolution model of yeast tRNA^Phe (ref. ²²), T. maritima P RNA¹⁷, and B. stearothermophilus P RNA¹⁴ as guides. All regions of the RNA molecules were visible in the map and regions that were missing in the original T. maritima P RNA model were built. Some minor corrections to the original model were needed, but overall the models for P RNA agree well. The protein density was clear for the backbone, but not for the side chains and hence the high resolution model of the T. maritima protein²¹ was placed on the experimental electron density map as a rigid body with minimal rebuilding.

Refinement was performed using Refmac5⁵⁷ and BUSTER⁵⁸. Due to the resolution of the data, the models were restrained to enforce good hydrogen bonding distance between Watson-Crick (WC) base pairs, planarity between base pairs (both for WC and non-WC base pairs), and C3′-endo sugar puckering for recognizable secondary structure elements. In addition, during BUSTER refinement the protein was restrained by the high resolution structure of the protein²¹. Model building with Coot⁵⁹ was interspersed with either Refmac5 or BUSTER refinement. During rebuilding, missing nucleotides were added as well as some missing residues at the N-terminus of the protein. Mg²⁺ ions were included at positions that had high density peaks in residual maps and also coincided with heavy atom sites. Other large peaks in the native data set that coincided with phosphate positions in the leader-soaked crystals were modelled as PO₄ ions. No individual atomic or group temperature factors were refined, only an overall anisotropic temperature factor. The final stages of the refinement were done with the program BUSTER. The refinement was done both with a carved data set where data outside the anisotropic diffraction limits (3σ cut-off) were excluded and also with a complete data set (isotropic) to the highest resolution limit (Table STIII). No significant difference was noted in the two refinements and the refinement statistics and electron density maps calculated from either data set were also virtually identical. It appears that the anisotropic temperature factor correction in the refinement programs adequately modeled the modest anisotropy of the data.

The final model for the P RNA includes nucleotides 1 to 338. Only the phosphate backbone was modelled for nucleotides 39, 241, and 314–317. In addition, 9 nucleotides were inserted between nucleotides 130 and 136 to account for the extension added for crystallization (Fig. S1). The final model for the tRNA includes nucleotides 1–76, but only the phosphate backbone was modelled for nucleotides 16, 17, and 20. The crystallization module added 8 extra nucleotides incorporated at the end of the anticodon stem (Fig. S1). Nearly the entire anticodon stem and anticodon loop were altered to accommodate the tetraloop receptor and altered anitcodon loop. The protein model includes residues 6 to 117. The positions of all side chains were ambiguous in the map and were not rebuilt, but kept as much as possible as in the original 1.2 Å model (PDB ID: 1NZ0) during refinement. Side chains that collided with the RNA were rebuilt when needed. There are 4 Mg²⁺ and 2 PO₄ ions in the model of the complex. The final model to 3.8 Å resolution has an overall R_work of 24.9% and R_free of 27.0% with a root mean square deviation (rmsd) from target values of 0.007 Å and 1.24° for bonds and angles respectively. The model in the presence of the leader includes an additional polyphosphate molecule with five phosphates and two Mg²⁺ ions coinciding with metals ions M1 and M2. A total of 5 Mg²⁺ ions were modeled into the complex that contains the 5′ polyphosphate leader backbone. The final model to 4.21 Å resolution has an overall R_work of 25.8% and R_free of 26.7% with an rmsd of 0.007 Å and 1.23° for bonds and angles respectively (see Table STIII and STIV).

Model superpositions were done with programs from the CCP4 suite⁵⁴, lsqman⁶⁰, and Coot⁵⁹. Diagrams were made with PYMOL⁴⁷. Coordinates and structure factors have been deposited in the PDB with accession numbers 3OK7 and 3OKB.

Activity assays of RNase P holoenzyme

Cleavage assays measuring k_cat/K_M under single turnover conditions were performed on the RNase P and pre-tRNA constructs that gave the best diffracting crystals. The pre-tRNA (with a single nucleotide leader (−1)) which yielded crystals and a control pre-tRNA (containing a T. maritima nine nucleotide leader (−9)) were radioactively labelled at their 5′ ends. Labelled substrates were purified over a 10% denaturing polyacrylamide gel and identified by ³²P-phosphorimaging. The holoenzyme was folded and cleavage reactions were performed in identical conditions as the folding reaction (1X THE, 10 mM MgCl₂, 0.1 M NH₄OAc, 37 °C). The enzyme activity of both the modified RNase P which gave crystals and the T. maritima wild-type RNase P were tested. The reaction was initiated by mixing pre-folded RNase P holoenzyme (25, 50, and 100 nM) with pre-folded pre-tRNA substrate (< 4 nM), incubated for various times (t = 0, 0.25, 1, 4, and 16 minutes), and subsequently quenched by adding 9 M urea, 50 mM EDTA. All reaction mixtures were directly loaded onto a 15% denaturing polyacrylamide gel which separated the substrate from the product(s). To observe unambiguously the products of the leader (−1) pre-tRNA, thin layer chromatography (TLC) was also performed with polyethyleneimine (PEI)-cellulose coated plates, where the quenched reaction mixture was spotted and run in a 5% acetic acid/100 mM NH₄Cl solution. The dried gels and the TLC plates were exposed to a phosphorimaging screen and the reaction profile was quantified by a phosphorimager (Fuji Medical) using ImageGauge software. A plot of the percentage of product over time gave the cleavage reaction rate for each concentration. Single turnover conditions assuming a first order reaction follow the equation: [P] = [P]_∞ (1 − e^−kobs^t), where [P] is the fraction of pre-tRNA cleaved, [P]_∞ is the fraction of uncleaved pre-tRNA at the end of the reaction, and k_obs is the observed reaction rate constant. By measuring k_obs at different concentrations it is possible to obtain K_cat/K_M assuming Michaelis-Menten kinetics.