|Home | About | Journals | Submit | Contact Us | Français|
Rpp20 and Rpp25 are two key subunits of the human endoribonucleases RNase P and MRP. Formation of an Rpp20–Rpp25 complex is critical for enzyme function and sub-cellular localization. We present the first detailed in vitro analysis of their conformational properties, and a biochemical and biophysical characterization of their mutual interaction and RNA recognition. This study specifically examines the role of the Rpp20/Rpp25 association in the formation of the ribonucleoprotein complex. The interaction of the individual subunits with the P3 arm of the RNase MRP RNA is revealed to be negligible whereas the 1:1 Rpp20:Rpp25 complex binds to the same target with an affinity of the order of nM. These results unambiguously demonstrate that Rpp20 and Rpp25 interact with the P3 RNA as a heterodimer, which is formed prior to RNA binding. This creates a platform for the design of future experiments aimed at a better understanding of the function and organization of RNase P and MRP. Finally, analyses of interactions with deletion mutant proteins constructed with successively shorter N- and C-terminal sequences indicate that the Alba-type core domain of both Rpp20 and Rpp25 contains most of the determinants for mutual association and P3 RNA recognition.
Ribonucleoproteins (RNPs) are functional units formed by the association of protein-coding and non-coding RNAs with proteins. RNPs are involved in a large spectrum of molecular activities and govern key cellular functions such as gene expression and its regulation; the significance of their roles is emphasized by the many diseases caused by mutations that disrupt either the RNA or the protein component of the RNP, or the factors required for their correct assembly (1–3). Several RNPs are implicated in the biogenesis of RNA, including the related endoribonucleases RNase mitochondrial RNA processing (MRP) and RNase P, both composed of an RNA molecule and several protein subunits. RNase MRP, identified only in Eukarya, is involved in pre-rRNA processing, in particular in the formation of the mature 5′-end of the 5.8S rRNA (4,5). Very recently, however, a more prevalent role of RNase MRP in ribosome biogenesis has emerged, specifically for entry of 35S pre-rRNA processing into the canonical maturation pathway (6). Despite its predominant localization in the nucleolus (7–9), RNase MRP also functions in mitochondrial DNA replication, by cleaving an RNA primer required for this process (10), and it has been shown to partake in the degradation of the mRNA encoding the mitosis specific cyclin B2 in yeasts (11). The significance of RNase MRP’s role in human growth and differentiation is substantiated by the link between mutations in the human RNA subunit and the pleiotropic condition termed cartilage–hair hypoplasia (CHH), a severe form of dwarfism (12).
Unlike RNase MRP, RNase P is ubiquitous in all domains of life; in all cases, it processes precursor tRNA transcripts, removing the 5′ leader sequences to generate their 5′ mature termini (13). Moreover, the Saccharomyces cerevisiae enzyme has been shown to be involved in C/D small nucleolar RNA (snoRNA) processing (14).
The RNA components of RNase MRP and P are essential for enzymatic activity and a high degree of similarity has been found or predicted in their structural features across species (15–22). Such structural correspondence has prompted endorsement for an evolutionary relation between RNase P and MRP, and in support of this hypothesis both complexes have been found to share many protein subunits that co-purify with respective endoribonuclease activities (13,22,23). In particular, to date up to 10 subunits have been identified in both human enzymes (hPop1, Rpp38, Rpp21, Rpp29/hPop4, hPop5, Rpp25, Rpp20, Rpp14, Rpp30, Rpp40), of which nine have homologues in yeast; in addition, RNase MRP from S. cerevisiae contains two specific proteins: Smn1 and Rmp1, and reports are discordant as to whether yeast Rpp21 (Rpr2) is unique to RNase P (22,24–27). Although in eukaryotes the RNA moieties of RNase P and MRP are thought to embody the catalytic core of the respective enzymes, they are reliant on the protein subunits for function in vitro and in vivo (13,24). The specific reasons for this requirement remain however unknown; alleged roles of the protein components, that await to be demonstrated, include maintaining the correct and active 3D fold of the RNA molecule, determining sub-cellular localization and contributing to substrate RNA discrimination (13).
The elucidation of the overall composition and architecture of the eukaryotic RNase P and MRP holoenzymes has been stalled by the challenges encountered in the isolation of the RNP complexes from native sources on the one hand, and in obtaining pure, stable individual recombinant components for reconstitution studies on the other (13). As a first step towards a depiction of the spatial organization of the holoenzymes, binary interactions between protein subunits, and with the RNA component, have been investigated in both human and yeast systems (28–32). However, such studies posed the question as to whether the inconsistencies and lack of completeness encountered in the investigations were a reflection of experimental artefacts, leading to false positive or false negative results. First, numerous attempts foiled by technical difficulties in obtaining recombinant proteins or by their high susceptibility to misfold, aggregate, degrade or be altered by their fusion tag have been reported (13,28,31,32); second, these experiments suffer from the drawback that synergic actions are not taken into account (13). The detailed molecular interaction of some of the subunits has been well established by structural data, in particular concerning the archaeal proteins of RNase P (that have homologues in eukaryotic systems) namely Rpp29–Rpp21 (33,34) and Pop5–Rpp30 (35); however, to our knowledge no systematic biophysical analysis of eukaryotic subunits has been reported thus far. We have therefore undertaken a comprehensive biochemical and biophysical characterization of two essential subunits of human RNase P and MRP, namely Rpp20 and Rpp25, with the aim of understanding their molecular properties and identifying potential problems with the recombinant products that may encumber assembly/interaction analyses. Rpp20 and Rpp25 have been reported to associate stably with one another and to interact with the P3 arm of the RNase MRP RNA (25,31,36,37). We find that Rpp20/Rpp25 association is very strong and that the heterodimerization activity is located largely in the central Alba-type core domain of both proteins. Importantly our results clearly show that Rpp20 and Rpp25 operate in tandem and that the formation of a 1:1 heterodimer is an obligate prerequisite for RNA binding.
The cDNAs of full-length human Rpp20 and Rpp25 in pCR4_TOPO were used as templates for further cloning. Rpp20 and Rpp25, and deletion mutants thereof were subcloned in many different expression vectors, but in many cases the recombinant products turned out to be insoluble, aggregated or affected by their fusion tag. We report here only the plasmids that have been utilized in this work.
Full-length Rpp25 (encompassing residues 1–199) was subcloned with an N-terminal hexahistidine tag into a pPROEX-HTb expression vector (Invitrogen) using Nco1/Not1 restriction sites. Full-length Rpp20 (residues 1–140) was subcloned by PCR into a pET-30 expression vector using the LIC methodology (Novagen); the pET-30 vector was modified to bear a TEV cleavage site to remove the N-terminal hexahistidine tag. Rpp20(35–140) was amplified from pET-30 Rpp20 by PCR to introduce 5′ Asc1 and 3′ Not1 sites and then ligated into a modified version of pET-15b vector (Novagen) with an Asc1 site inserted directly after the his tag.
Fragments subcloned in pETDuet-1 vector (Novagen) with an N-terminal hexahistidine tag were: Rpp20, Rpp20(16–140), Rpp25(25–170) and co-expressed Rpp20/Rpp25 (in which Rpp20 bears the his tag). Rpp25(25–170) was also subcloned in a pCDFDuet-1 vector (Novagen) also with an N-terminal hexahistidine tag (Supplementary Data). For cloning into pETDuet-1 and pCDFDuet-1, PCR primers were designed to complement appropriate regions of DNA in the pET-30 and pPROEX-htb clones encoding Rpp20 and Rpp25 proteins, respectively. Forward PCR primers used for producing his-tagged proteins encoded a TEV-cleavage site (ENLYFQG). Restriction sites incorporated in the forward/reverse PCR primers were as follows: EcoRI/NotI for Rpp20 and Rpp20(16–140); EcoRI/PstI for Rpp25(25–170); DNA encoding non-his-tagged proteins was amplified by PCR forward/reverse primers with restriction sites NdeI/XhoI. PCR products were cut with the appropriate restriction enzymes and ligated into similarly digested pETDuet-1 or pCDFDuet-1. Escherichia coli KRX cells (Promega) were transformed with the ligation mixtures and transformants, obtained on LB amp plates (for pETDuet-1 clones) or LB streptomycin plates (for pCDFDuet-1 clones), were screened by colony PCR and recombinant clones were sequenced.
Rpp25, Rpp20 (from pET-30 vector) and Rpp20(35–140) were expressed in E. coli strain BL21(DE3)pLysS at 37 and 18°C, respectively, induced with 1 mM IPTG (isopropyl β-d-thiogalactoside) at an OD600 of ~0.6 and incubated for a further 3 h. All the other proteins subcloned into pETDuet-1 or pCDFDuet-1 vectors were expressed in E. coli KRX strain at 25°C, induced using 1 mM IPTG and 0.1% (w/v) rhamnose at an OD600 of ~0.6 and incubated overnight.
Cell pellets were lysed by sonication in 20 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, pH 7.4 and centrifuged to separate the soluble and insoluble fractions of the cell. All recombinant proteins listed above were soluble except Rpp20 expressed in pET-30. In this case, the insoluble fraction of the pellet was completely denatured in a buffer containing 8 M urea (see below).
As all the soluble recombinant proteins contained a hexahistidine-tag, or were co-expressed with hexahistidine-tag proteins, they were purified by affinity chromatography on 5 ml HisTrap columns (GE Healthcare) with a gradient of 20–500 mM imidazole. The His-affinity tag was removed by addition of TEV protease and incubation at 30°C overnight in 25 mM sodium phosphate, 100 mM NaCl, 1 mM DTT (Dithiothreitol), pH 7; the reaction mixture was subsequently applied to a Ni–NTA column (Qiagen) to remove the cleaved tags, the His-tagged TEV and any undigested product. The proteins were further purified on a 5 ml Hi-Trap heparin column, mainly to remove nucleic acids contaminants. The proteins eluted with a linear 0–2 M NaCl gradient were dialysed overnight in buffer A [phosphate buffered saline (PBS, Sigma) containing 10 mM phosphate, 2.7 mM KCl and 137 mM NaCl, adjusted to pH 7, plus 1 mM DTE (Dithioerythritol)] and loaded on a Superdex™ 75 column in the same buffer (see below). The tag was not removable for Rpp20(35–140).
The cell pellet containing Rpp20 expressed from pET-30 vector was lysed by sonication in 20 mM sodium phosphate, 500 mM NaCl, 20 mM imidazole, pH 7.4. After centrifugation, the pellet containing inclusion bodies was resuspended in 100 mM sodium phosphate, 20 mM imidazole, 8 M urea, pH 7 and then filtered through a 0.2 µm filter. The protein was purified on a 5 ml HisTrap column (GE Healthcare) under denaturing conditions with a 20–500 mM imidazole gradient. Rpp20 was then refolded by stepwise dialysis as follows: the eluted protein was placed in a dialysis membrane with molecular weight cut-off of 6000 Da and equilibrated for 24 h in a base buffer (0.4 M l-Arginine, 25 mM sodium phosphate, 100 mM NaCl, 1 mM DTE, pH 7) containing 4 M urea at 4°C. Denaturant was slowly removed by a series of overnight equilibrations with buffers of decreasing urea concentration (2, 1, 0 M). Finally, the sample was dialysed into TEV protease buffer for removal of the tag. Subsequent purification steps were as described for the other proteins. The correct refolding of Rpp20 was validated by CD analysis. When recombinant Rpp20 (in pET-30 vector) was co-expressed with Rpp25(25–170) (in pCDFDuet-1) for analytical ultracentrifugation (AUC) studies (see below), the complex was found in the soluble fraction and did not require refolding.
Protein concentrations were calculated based upon the near-UV absorption (ε280) using theoretical extinction coefficients derived from ExPASy (38).
Size exclusion chromatography (SEC) was performed using an AKTA Basic System (GE Healthcare) in combination with a Superdex™ 75 column (10/300). All samples were exchanged into identical buffer conditions (buffer A). The column was calibrated using separate injections of aprotinin (6.5 kDa), ribonuclease A (13.7 kDa), carbonic anhydrase (29.0 kDa), ovalbumin (44.0 kDa) and conalbumin (77.0 kDa) from the Low Molecular Weight Gel Filtration Calibration Kit (GE Healthcare) under the same buffer conditions.
The P3 arm of RNase MRP RNA was purchased from IBA GmbH (Göttingen, Germany) and dissolved in buffer A. The RNA were annealed by heating at 95°C for 5 min followed by slow cooling to room temperature as previously described (39). The concentration of the dissolved oligonucleotide was evaluated by UV measurement at 95°C, using a molar extinction coefficient at 260 nm calculated by the nearest-neighbour model (463 700 M−1 cm−1) (40).
The protein and RNA solutions were prepared in buffer A. Experiments titrating Rpp25 with Rpp20 (and deletion mutants) were performed at three different temperatures (293, 298 and 303 K) using a high-sensitivity ITC-200 microcalorimeter from Microcal (GE Healthcare). RNA–protein titrations were carried out at 298 K on an ITC-200 microcalorimeter. Before each isothermal titration calorimetry (ITC) experiment, the pH of each solution was checked, the reference cell was filled with deionized water, and the protein solutions were degassed for 2–5 min to eliminate air bubbles. The first addition was executed only after achieving baseline stability. Measurement from the first injection was discarded from the analysis of the integrated data, in order to avoid artefacts due to the diffusion through the injection port occurring during the long equilibration period, locally affecting the protein concentration near the syringe needle tip. To investigate protein–protein interactions, in each titration volumes of 2 µl of a solution of Rpp20 or mutants thereof at a concentration of 160–180 µM were injected into a solution of Rpp25 or mutants thereof (20 µM) in the same buffer, using a computer-controlled 40-µl microsyringe. For the RNA–protein studies, the P3 RNA at a concentration of 170 µM was injected into a solution containing individual proteins or their pre-assembled complexes at a concentration of 20 µM in buffer A. Notably, the titration of Rpp25 with Rpp20 was also repeated in perchlorate buffer (25 mM sodium phosphate, 10 mM sodium perchlorate, 1 mM DTE, pH 7) to check any effect of this buffer (used in the CD experiments, see below) on protein behaviour. A spacing of 200 s between each injection was applied to enable the system to reach the equilibrium. Heat produced by titrant dilution was verified to be negligible by performing a control experiment, titrating it into the buffer alone, under the same conditions. Integrated heat data obtained for the titrations were fitted using a non-linear least-squares minimization algorithm to a theoretical titration curve, using the MicroCal-Origin 7.0 software package. ΔH° (reaction enthalpy change in kJ mol−1), Kb (binding constant in M−1), and n (molar ratio between the two proteins in the complex) were the fitting parameters. The reaction entropy was calculated using the relationships ΔG° = −RT ln Kb (R = 8.314 J mol−1 K−1, T = 298 K) and ΔG° = ΔH° − TΔS°. In addition, from the experiments at different temperatures (293, 298 and 303 K), the change in heat capacity ΔCp° upon binding has been calculated as the resulting slope in a plot of ΔH° versus the experimental temperature.
The Rpp20 self-association was characterized following a dilution protocol (41), injecting 3 µl of a Rpp20 solution at a concentration of 6 µM into the calorimetric cell containing 200 µl of buffer A at 298 K. The resulting dissociation isotherm was interpolated using a non-linear regression fitting procedure based on a simple dissociation model (Supplementary Data).
Simultaneous UV and circular dichroism (CD) spectra of 0.2 mg ml−1 protein solutions were recorded on the Applied Photophysics Ltd. Chirascan Plus Spectrometer (Leatherhead, UK), continuously flushed with nitrogen. The molar ellipticity per mean residue, [θ] (deg cm2 dmol−1), was calculated from the equation: [θ] = [θ]obs × (mrw)/10 × l×C, where [θ]obs is the ellipticity (deg), mrw is the mean residue molecular weight, C is the protein concentration (g ml−1), and l is the optical path length of the cell (cm). Rectangular Suprasil cells with 1 and 0.05 cm path lengths were employed to record spectra in the regions between 400–230 and 260–180 nm, respectively. The parameters used to acquire the spectra were: spectral bandwidth of 1 nm, data step-size of 1 nm with a time-per-data-point of 1.5 and 3.0 s in the near-UV and the far-UV, respectively. All the spectra were baseline corrected by subtracting the buffer spectrum. Experiments were conducted in buffer A at 298 K. The sodium chloride content of many buffers absorbs light very strongly below 200 nm making CD measurements below 200 nm very difficult and secondary-structure analysis less reliable (42). A full far-UV CD spectrum down to 180 nm was only possible replacing NaCl with sodium perchlorate in the buffer medium (see above). To estimate the secondary structure content, curve fitting was performed using DICHROWEB (43).
Recombinant 15N-labelled Rpp25 and Rpp25(25–170) were prepared on minimal media as described previously (44) and dissolved in buffer A at concentrations of 0.1–0.3 mM. Nuclear magnetic resonance (NMR) spectra were recorded at 298 K on a Bruker Avance spectrometer operating at 16.4 T equipped with a triple resonance cryoprobe. Spectra were processed and analysed as previously reported (44).
Sedimentation equilibrium experiments were performed using a Beckman Optima XL-A analytical ultracentrifuge as described previously (45). Samples were prepared in buffer A and data were acquired with an average of 25 absorbance measurements at a wavelength of 280 nm and a radial spacing of 0.001 cm. Sedimentation equilibrium experiments were performed at 4°C and rotor speeds of 17 000, 14 500 and 12 000 rpm for both Rpp20 and co-expressed Rpp20/Rpp25(25–170). Protein concentrations were in the range of 15–30 µM for Rpp20/Rpp25(25–170) and 90–180 µM for Rpp20. For data analysis, the partial specific volumes and monomeric molecular weights of the different proteins were calculated from amino acid composition using SEDNTERP (http://www.rasmb.bbri.org/) and gave the following values: Rpp20 from pETDuet-1 vector (0.7269 cm3 g−1; 15 706 Da); Rpp20/Rpp25(25–170) (0.7253 cm3 g−1; 16026 Da and 0.7303 cm3 g−1; 15 435 Da, respectively). The solvent density (ρ) was calculated to be 1.00722 g ml−1 at 4°C. The monomeric buoyant molecular mass M(1 − ) was calculated to be: 4206 for Rpp20 from pETDuet-1, 4319 for Rpp20 from pET-30 and 4081 for Rpp25(25–170). Data for all concentrations and speeds were analysed simultaneously using a range of models in the Sigmaplot software package as described previously (45). Residuals were calculated by subtracting the best fit of the model from the experimental data.
Rpp20 and Rpp25 are postulated to function in tandem and to be components of both human RNase P and MRP complex (25,31,36,37). Comparative genomics and primary sequence profile analysis have unveiled an evolutionary connection for Rpp20, Rpp25 and the archaeal protein Alba; in particular, the phylogenetic tree derived from the alignment of the Alba domain has subdivided this superfamily into two eukaryotic families (containing Rpp20 and Rpp25 orthologues, respectively) and one archaeal Alba family (46). The primary function ascribed to Alba proteins relates to DNA packaging and chromosomal organization in archaea, in an equivalent role to that of eukaryotic histones (47). Structural studies have unveiled a compact α/β fold for the Alba domain with a topology akin to the C-terminal domain of IF3 (48). Since the first structure of the archaeal Alba1 was published by Wardleworth et al. (48), several structures have been determined for Alba1 and Alba2 from different organisms, and the first structure available within the eukaryotic subgroups belongs to the At2g34160 protein from Arabidopsis thaliana, of unknown function, which has not been characterized beyond the deposition of the coordinates (PDB ID 2Q3V). The primary sequence alignment of Rpp20 and Rpp25 with proteins of known structure (Figure 1) complemented with their secondary-structure prediction analysis [PredictProtein (49), data not shown] would suggest the presence of an Alba-type core domain flanked by additional secondary-structure elements and/or unstructured regions. Guided by such analyses, we embarked on mutagenesis studies of Rpp20 and Rpp25 to identify regions required for their domain stability, their mutual interaction and RNA recognition.
Human Rpp20 is a 16 kDa protein predicted to consist of an Alba-like domain and an N-terminal tail. To characterize the molecular properties of Rpp20, far-UV CD analysis was applied to full-length protein (encompassing residues 1–140) and to two truncated C-terminal fragments, spanning amino acids 16–140 and 35–140, henceforth referred to as Rpp20(16–140) and Rpp20(35–140), respectively. The deletion mutants were designed to retain the regions conserved within the superfamily (46) (Figure 1), and presumably an intact Alba-type domain. It is noteworthy that extensive molecular biology and biochemistry work had to be devoted to obtain reproducibly soluble and stable proteins (see ‘Materials and Methods’ section and below).
Far-UV CD spectral analyses indicated that Rpp20 adopts an α/β fold, compatible with the Alba structure, alongside flexible regions, and consists of ~18% α-helix, 25% β-strand, 24% turns and 33% of disordered/other structures (Figure 2). As the shape of the curves is very similar for Rpp20 and its deletion mutants (Figure 2A), the comparative CD study supports the hypothesis that the folded core of the protein is retained in all the constructs examined, and that the most N-terminal 34 residues are likely to be predominantly unstructured. The secondary-structure content for the deletion mutants of Rpp20, however, could not be estimated because of the instability and aggregation of Rpp20(16–140) and Rpp20(35–140) in perchlorate and other buffers tested which would be transparent below 200 nm (see ‘Materials and Methods’ section).
During purification of recombinant proteins, we noticed that full-length Rpp20 and its deletion mutants Rpp20(16–140) and Rpp20(35–140) eluted in SEC at apparent molecular weights of about 47, 43 and 40 kDa, respectively (Figure 3 and Table 1), indicating the formation of oligomers in solution. Subsequent AUC sedimentation equilibrium experiments on full-length Rpp20 could not confirm the presence of a single ideal species, as systematic residuals were observed (Figure 3); nevertheless, a good fit was obtained with a model of self-association and non-interacting mixtures, signifying that under the conditions used Rpp20 exists primarily in a homodimeric state, alongside small amounts of contaminating aggregates not in equilibrium with the dimer itself. This may be due to part of the protein irreversibly unfolding and associating during the AUC experiments. Interestingly, most of the Alba proteins studied also form dimers in solution.
The process of association of Rpp20 was characterized further by conducting dilution experiments using ITC, where the amount of heat measured upon each injection is governed by the enthalpy change of dissociation and the monomer–dimer equilibrium constant (Supplementary Data) (41). Analysis of the data shown in Supplementary Figure S1 revealed an enthalpic change of 20 kJ mol−1 and a derived association constant of 8 × 105 M−1 (Supplementary Data).
Human Rpp25 is a 20.6 kDa protein that belongs to the eukaryotic group most divergent from the archaeal Alba subfamily. A distinctive trait of Rpp25 and its orthologues is the occurrence of a long C-terminal extension bearing the conserved GYQXP signature (at position 150–154 in human Rpp25), whose function remains unidentified. Bioinformatic analyses of Rpp25 (Figure 1 and see above) favour the presence of a putative Alba-type core domain flanked by N- and C-terminal extensions and to characterize this further a number of deletion mutants were designed accordingly. Nonetheless, for several, the lack of expression or solubility of the recombinant products, or else instability of the purified proteins could not be overcome (data not shown), hence efforts converged towards the attainment of the full-length protein and one deletion mutant truncated at both N- and C-terminal tails, specifically Rpp25(25–170).
To evaluate the effects of such N- and C-terminal truncations on the secondary structure of Rpp25, far-UV CD spectra were recorded for the wild-type and Rpp25(25–170) mutant, revealing comparable curves for the two proteins and bearing the hallmarks of α/β domains. Nonetheless, the molar ellipticities are significantly less negative for the full-length Rpp25, in concomitance with the negative peak shifted towards 200 nm. It is therefore conceivable that longer Rpp25 has a lower amount of structured content, supporting the notion of partially disordered N- and C-terminal tails. Although a full secondary structure content prediction for Rpp25(25–170) was impeded by the poor behaviour of the protein in perchlorate buffer, the qualitative CD results are in excellent agreement with the NMR analysis: here the comparison of 1H–15N HSQC spectra of full-length Rpp25 and its mutant Rpp25(25–170) demonstrated that the well-dispersed resonances belonging to the folded portion of the protein remain unaltered in the shorter construct, plus the sharp signals exclusive to the intact Rpp25 spectrum manifest a very narrow range of chemical shifts, symptomatic of flexible conformations (50) (Figure 2).
On a size exclusion column, Rpp25 migrated with an apparent molecular weight of ~39 kDa; however, the partially unstructured nature of the N- and C-terminal appendices (see above) is expected to influence the effective hydrodynamic radius of the macromolecule, and may therefore result in an overestimation of the protein mass by this technique. Unfortunately, accurate determination of the molecular weight using shape-independent AUC equilibrium sedimentation was precluded by the inherent instability of the full-length product at concentrations required for this experiment over a period of 2–3 days. Nonetheless, the presumably more globular mutant Rpp25(25–170) (15.4 kDa), behaved plausibly as a monomer on the SEC, with an apparent molecular mass of ~21 kDa (Figure 3 and Table 1). On the basis that full-length Rpp25 elutes later than full-length Rpp20 from the size exclusion column in the same experimental conditions (Figure 3), despite a greater formula molecular weight and conceivably a less globular shape, we concluded that Rpp25 in solution exists largely in a monomeric state.
Despite earlier reports that Rpp20 and Rpp25 interact strongly in vivo and in vitro, no quantitative measure of this interaction has ever been reported. This would be important to determine, with the aim of understanding whether the two proteins are likely to operate as a single working pair or act individually; also the analysis of the energetics of the interaction might provide insights into their mechanism of action. To this aim, we employed ITC, which is largely used to investigate molecular binding reactions by measuring the heat generated or absorbed in the binding event and thereby providing the binding constant, the molar ratio of the two proteins in the complex and the enthalpy change (ΔH°) of the interaction. For the Rpp25–Rpp20 system the integrated heat data, at each investigated temperature, showed that the binding process is composed of one clear event centred on a molar ratio of one (Figure 4). The binding isotherm curves corresponding to this reaction have been interpolated using an independent-sites model, revealing that at 298 K full-length proteins interact with each other with an association constant (Kb) of 5.3 × 107 M−1, with an enthalpy change (ΔH°) of −94 kJ mol−1 and an entropic contribution (defined as −TΔS°) of 50 kJ mol−1. The thermodynamics of the interaction indicate that the binding event is enthalpically driven.
To delineate the regions of Rpp20 responsible for Rpp25 recognition, ITC measurements were carried out on the deletion mutants Rpp20(16–140) and Rpp20(35–140) truncated in the N-terminal tail. Interestingly, in both cases the global thermodynamics of the interaction with Rpp25 is seemingly unaffected (Table 2), suggesting that the N-terminal region of Rpp20 does not participate in this recognition.
A second set of experiments was performed monitoring complex formation between the Rpp25(25–170) mutant and all the available versions of Rpp20, to evaluate the contribution to binding of the N- and C-terminal regions of Rpp25 (Figure 4 and Table 2). The thermodynamic signature of the association is fully preserved in all cases (Table 2), thereby demonstrating that the elements required for the mutual interaction remain intact in the deletion mutants and that both the N- and C-terminal tails of Rpp25 do not partake in the recognition process. Furthermore, the Kb values obtained in the reactions with the truncated Rpp25 mutant are slightly higher than the ones found with full-length Rpp25, indicative of an improved efficiency of the interaction and probably linked to the absence of the mainly unstructured tails.
Taken together, the ITC experiments reveal that the regions within the mutants Rpp20(35–140) and Rpp25(25–170) are sufficient for mutual interaction, indicating that this recognition could be mediated largely, if not exclusively, by the Alba-type core domains. In support, CD spectroscopy showed that the association between Rpp20 and Rpp25 has no detectable influence on their secondary/tertiary structure (Figure 2); in other words, this interaction is not accompanied by structural rearrangements, thus ruling out refolding events in the N- and/or C-terminal extensions upon binding.
To investigate the Rpp20–Rpp25 interaction in more depth, we measured the values for the change in heat capacity ΔCp° upon binding; these are all negative and centred around −3 kJ mol−1 K−1 (Supplementary Table S1). From the ΔCp° values, an estimation of the solvent-accessible surface area buried upon association could in principle be derived; this however requires an accurate determination of the ionization and protonation contributions to the ΔH°, achieved by performing titration experiments in different buffer/salt conditions (51). Unfortunately, we could not find a sufficient number of experimental conditions for such measurements because of protein misbehaviour. Nonetheless, the large negative ΔCp° values point towards a large loss of solvent-accessible surface area upon complex formation, suggesting that each protein participates in the association through a large interacting surface. This finding is consistent with the high values measured for the Kb.
ITC experiments indicate that Rpp20 and Rpp25 interact with one another in a 1:1 molar ratio. Nonetheless, the homodimeric nature of Rpp20 invites the question whether the complex consists of a 1:1 (Rpp20/Rpp25) heterodimer or a 2:2 (Rpp20/Rpp25) heterotetramer. Understanding the exact stoichiometry of the Rpp20/Rpp25 complex is a matter of key importance: since many of the RNase P/MRP subunits do indeed self-associate, it has been suggested that the holoenzyme(s) might contain multiple copies of its constituents (13,26).
To address this point, AUC sedimentation equilibrium experiments were attempted on a number of complexes with various combinations of full-length and deletion mutant proteins, including co-expressed recombinant fragments. Some of the results, however, were ambiguous and deemed inconclusive, once more thwarted by the high propensity of the complexes to aggregate and in some cases by their susceptibility to degrade over a short period of time. The best results were obtained for the co-expressed Rpp20/Rpp25(25–170), though yet again the data did not correspond to a single species but gave a reasonable fit using a model of self-association with non-interacting aggregates. The best fit was found for a 1:1 heterodimer in the presence of, but not in equilibrium with, other species of larger molecular weight (Figure 4). Consistent with this, the pattern of the SEC profiles clearly supports the Rpp20/Rpp25 heterodimer model, with apparent molecular masses of their complexes (including a variety of combinations with truncation mutants) well below the expected formula weights for putative tetramers (Figure 3 and Table 1). This not only implies a significant overlap in the surfaces of Rpp20 implicated in homodimerization and Rpp25 binding, but also that the two subunits are optimized to maximize heterotypic complementarity. A quantitative treatment of the equilibrium between the Rpp20 homodimer and the Rpp20/Rpp25 heterodimer revealed that the preferential heterodimer formation arise from a higher association constant of Rpp20/Rpp25 with respect to the self-association of Rpp20 (Supplementary Data). Interestingly, this closely resembles the molecular behaviour of Alba1/Alba2 proteins, which form an obligate heterodimer though existing individually as tight homodimers (52). A 1:1 stoichiometry is also in agreement with what was found for the yeast homologue Pop6/Pop7 complex (53), although the oligomeric state of Pop7 (homologous to Rpp20) was not characterized beyond observations of self-association in GST-pull down assays (32).
The Rpp20/Rpp25 complex and the isolated Rpp25 have been reported to interact with the P3 arm of the RNase MRP RNA (hereafter referred to as P3 RNA) (36,37). To gain a better insight into the nature of these interactions, we performed ITC titrations between the P3 RNA and an array of Rpp20 and Rpp25 proteins and complexes. To begin with, the RNA binding ability of the individual full-length proteins was appraised: as shown in Figure 5, the ITC results indicated that isolated Rpp20 was unable to interact with P3 RNA, whereas Rpp25 displayed a weak association, with a dissociation constant lower than the threshold of binding that could be fully characterized by ITC, i.e. around 0.1 mM. Next, under the same experimental conditions, we measured the binding of a previously assembled Rpp20/Rpp25 complex with the P3 RNA (Figure 5 and Table 2): in this case a much stronger interaction was detected, with a binding affinity of 1.3 × 107 M−1 and favourable changes in enthalpy and entropy (as − TΔS°) of −28 kJ mol−1 and −13 kJ mol−1, respectively. The negative enthalpy change denotes the formation of new intermolecular contacts between the protein complex and the RNA molecule, whereas the negative entropic contribution suggests an increased degree of freedom of the system upon binding, probably because of displacement of bound water molecules from interacting surfaces. It is noteworthy that, as the protein concentrations used in the ITC measurements are a 1000-fold higher than the Kd value describing protein–protein binding, we anticipate that the Rpp20/Rpp25 mixture would exist solely in the heterodimeric form prior to RNA interaction, as confirmed by gel filtration analysis (Figure 5). Our results therefore reveal a truly synergic mechanism by which heterodimer formation augments the affinity of the individual subunits for the P3 RNA by at least a 1000-fold, thereby conferring P3 RNA binding proficiency to the Rpp20 and Rpp25 subunits. The strong specific binding to the P3 stem–loop is in complete agreement with previously reported GST pull-down studies (36).
Thus it appears that the generation of the required RNA-binding surface for this RNA target strictly hinges on the heterodimerization of the protein subunits, and to determine whether the partially disordered extensions of Rpp20 and Rpp25 would participate in RNA interaction, a second set of ITC experiments using the truncated heterodimer Rpp20(35–140)/Rpp25(25–170) was undertaken. The results in Figure 5 and Table 2 show that this complex binds to P3 RNA with slightly lower affinity than the full-length counterparts, suggesting that, whereas the extensions beyond the domain core of Rpp20 and Rpp25 had essentially no effect on protein–protein interaction, they may well play a role in RNA recognition. Nevertheless, since the decrease in RNA binding affinity exhibited by the deletion mutant complex is relatively small, the core of the Rpp20/Rpp25 proteins plausibly dominates the RNA interaction, with the tails only yielding a minor contribution to stabilization of the RNA complex.
Contrary to archaeal Alba proteins, which interact with double stranded DNA at stoichiometries of 12 bp or 6 bp per dimer (48,52), our isothermal binding curves clearly indicate that each Rpp20/Rpp25 heterodimer binds to one molecule of 50 nt P3 RNA. A stoichiometry of 1:1:1 (Rpp20:Rpp25:P3) agrees well with the size exclusion profiles (Figure 3 and Table 1).
The purpose of this study was to shed light on the organization and function of the RNase P and MRP, in particular focussing on the essential Rpp20 and Rpp25 subunits. The recombinant proteins were subjected to an array of biochemical and biophysical methodologies, with the aim of understanding their molecular behaviour and the details of molecular recognition within the RNP complex. The value of these investigations is 2-fold: first, such molecular characterizations are per se key to elucidate RNP architecture and mode of action, and they are generally not afforded by studies on purified native particles; secondly, a detailed scrutiny of the molecular behaviour of recombinant products could provide a solid platform for the design of subsequent experiments. It is noteworthy that a further dimension of complexity and interest of these systems is the nearly identical composition of the protein subunit element for the RNase P and MRP of a given organism, despite their distinctive functions; furthermore, it was very recently reported that the human RNase MRP RNA associates with the telomerase reverse transcriptase to generate an RNP with RNA-dependent RNA polymerase activity (54). This adds conviction to the view that the determinants of the function for a given RNP particle reside in its entirety, underscoring therefore the importance of elucidating the molecular basis of recognition between the different components within the RNP systems.
Rpp20 and Rpp25 have previously been shown in vivo and in vitro to associate with one another, with functional relevance for holoenzyme association and localization (36). Nonetheless, it remained unclear whether the Rpp20/Rpp25 complex would constitute a single working unit, in other words whether this complex would represent the minimal protein and RNA-binding unit within the RNP particle. Such a conclusion can in fact only be derived from systematic biophysical and biochemical analyses with multiple components, beyond the largely qualitative screening of binary Rpp25–RNA and Rpp20–RNA interactions hitherto reported (31,36,37,55,56). Although Rpp20 has been said to have the ability to enhance the interaction between Rpp25 and the RNA (36,37), the results presented here firmly indicate that Rpp20 and Rpp25 operate as a single working unit, whereby their assembly is an obligate prerequisite for P3 RNA interaction. In fact, Rpp20–Rpp25 nanomolar dissociation constant would strongly argue for a preponderance of the heterodimeric form also in vivo, which agrees well with particle composition analysis studies (31); in addition, the ITC results unambiguously prove that the binding to the P3 arm of the RNase MRP RNA is a fully synergic event, with an association constant for the complex at least a 1000-fold higher than for Rpp25 alone. Given the similarity between the P3 domain of RNase P and RNase MRP RNA, it is likely that the same molecular recognition occurs in both holoenzymes. A similar conclusion was also proposed for the yeast homologues Pop6/Pop7, although the suggested synergic mechanism here stopped short of full characterization because of difficulties in obtaining both individual proteins (53).
The detailed molecular explanation for such synergic behaviour remains to be uncovered; nonetheless, our current results point to the likelihood that the interface required for P3 RNA binding is generated upon the formation of the heterodimer. The bioinformatic analysis in conjunction with the CD and ITC investigations presented here show that Rpp20 and Rpp25 contain a core Alba-type domain flanked by extensions (N-terminal for Rpp20 and N- and C-terminal for Rpp25) which do not contribute to core domain stabilization or to protein–protein interaction, but may play a minor role in RNA recognition. Specifically, our study explored whether truncation of these extensions affected the structural and functional properties relative to the wild-type. As none of the mutants analysed elicit structural alterations or exhibited differences in mutual recognition compared to the full-length proteins, we reason that the extensions are probably largely unstructured and that Rpp20 and Rpp25 interact with one another mainly through the Alba-type core domain without undergoing substantial conformational rearrangements, forming a tight 1:1 heterodimer. Within the structure catalogue of archaeal Alba proteins, the heterodimeric organization is exemplified by the Alba1/Alba2 complex from Sulfolobus solfataricus (52) (Figure 6).
This quaternary arrangement exhibits a high degree of conservation amongst the homodimeric archaeal Alba structures and, interestingly, is extremely similar to the very recent structure of the yeast Pop6/Pop7 heterodimer bound to the P3 arm of the yeast RNase MRP (57). The protein–protein interface in these complexes is formed by helix α2 and the last two β-strands (β3 and β4) from both monomers, engaging in an extensive network of hydrophobic interactions and specific hydrogen bond contacts (Figure 6). Our results would support an overall structural resemblance between Rpp20/Rpp25 and the yeast homologue system: akin to Pop6/Pop7, Rpp20 and Rpp25 in fact interact with one another via the Alba-like domain with a very extensive solvent-accessible surface area buried by this molecular association, as suggested by our ITC experiments (specifically the large negative ΔCp° values). Interestingly, Pop6 and Pop7 displayed an Alba-type domain fold with additional short secondary-structure elements in the N-terminal region (in particular a β-strand and an α-helix) (57); whether this is also the case for Rpp20 and Rpp25 remains to be established. Bioinformatics analysis, however, shows that for Pop6 these elements are comprised in a region not present at all in Rpp25 [Figure 1 and ref. (57)], whereas a very low degree of conservation is displayed in this amino acid stretch for Pop7 and Rpp20 [Figure 1 and ref. (57)].
The Rpp20/Rpp25 heterodimer has been shown to interact with the P3 RNA with a 1:1 stoichiometry, in contrast with the situation in archaeal proteins where the binding to DNA entails a much higher protein:nucleic acid ratio. This divergence though is not entirely unexpected, and conforms to Alba’s main role in DNA packaging and organization. One may speculate that the Rpp20/Rpp25 heterodimer would exhibit a nucleic acid binding mechanism distinct from that of archaeal homologues, following modifications in the course of evolution in response to different functional requirements. Consistent with this idea, the recent evidence for the yeast system showed that Pop6/Pop7 recognize specifically a stem-bulge element in yeast P3 RNase MRP RNA (57) compared to the non-specific double-helix DNA recognition which is characteristic of Alba.
Although the binding to the P3 RNA appears to be the key factor in mediating the association of Rpp20 and Rpp25 to the holoenzyme, previous work has suggested that other protein subunits might further stabilize their assembly into the RNP complex (31). Nonetheless, these investigations through binary screening might not provide a faithful reflection of the molecular recognition process in the context of the holoenzymes, where synergic interactions are likely to take place. In the case of Rpp20 and Rpp25, our results clearly indicate that the outcome of some of the previous investigations might have been perturbed because the interacting entities used were the individual proteins instead of the working unit defined as a pair, and the isolated proteins may be unlikely to retain authentic binding behaviour. Crucially, it may also appear that this is not a prerogative of Rpp20–Rpp25, as other pairs which are likely to perform as a working unit have already been identified, namely Rpp29–Rpp21 (33,34) and Pop5–Rpp30 (35), although this remains to be confirmed by quantitative measurements of their association. The identification of pair-wise interactions between subunits represents an important step forward in the study of the assembly, structure and dynamics of the RNase P and MRP holoenzymes.
In conclusion, the biochemical and biophysical characterization of Rpp20 and Rpp25, and their interaction with RNA will help elucidate the mechanistic basis of their actions and will be an invaluable contribution to understanding the architecture and function of the eukaryotic RNase P and MRP.
Supplementary Data are available at NAR Online.
The Wellcome Trust for the Biomolecular Centre for Molecular Spectroscopy (to M.R.C., A.F.D.); EPSRC-case PhD studentship (to K.L.D.H-T.). Funding for open access charge: The Wellcome Trust.
Conflict of interest statement. None declared.
We are grateful to the MRC Biomedical NMR Centre, Mill Hill and its staff, for a generous allocation of NMR time and for expert technical assistance. We thank Drs Tim Welting and Sandy Mattijssen for helpful discussions.