Trypanosoma brucei 20 S Editosomes Have One RNA Substrate-binding Site and Execute RNA Unwinding Activity*

Background: Mitochondrial transcripts in African trypanosomes undergo U insertion/deletion-type RNA editing that is catalyzed by a protein complex known as the editosome. Results: Editosomes have a single RNA substrate-binding site and catalyze RNA unwinding. Conclusion: Both U insertion and U deletion are executed within a single, multifunctional reaction center. Significance: RNA binding is followed by RNA unwinding, which represents a novel activity of the editing machinery. Editing of mitochondrial pre-mRNAs in African trypanosomes generates full-length transcripts by the site-specific insertion and deletion of uridylate nucleotides. The reaction is catalyzed by a 0.8 MDa multienzyme complex, the editosome. Although the binding of substrate pre-edited mRNAs and cognate guide RNAs (gRNAs) represents the first step in the reaction cycle, the biochemical and biophysical details of the editosome/RNA interaction are not understood. Here we show that editosomes bind full-length substrate mRNAs with nanomolar affinity in a nonselective fashion. The complexes do not discriminate–neither kinetically nor thermodynamically–between different mitochondrial pre-mRNAs or between edited and unedited versions of the same transcript. They also bind gRNAs and gRNA/pre-mRNA hybrid RNAs with similar affinities and association rate constants. Gold labeling of editosome-bound RNA in combination with transmission electron microscopy identified a single RNA-binding site per editosome. However, atomic force microscopy of individual pre-mRNA-editosome complexes revealed that multiple editosomes can interact with one pre-mRNA. Lastly, we demonstrate a so far unknown activity of the editing machinery: editosome-bound RNA becomes unfolded by a chaperone-type RNA unwinding activity.

Mitochondrial gene expression in kinetoplastid protozoa such as African trypanosomes requires RNA editing. The reaction converts cryptic pre-mRNAs into mature transcripts and is characterized by the insertion and deletion of exclusively U 3 nucleotides (for recent reviews see Refs. 1 and 2). The process involves a specific class of small, noncoding RNAs, known as guide RNAs (gRNAs), which represent transacting genetic elements that function as templates in the process (3). The basic reaction steps of the editing reaction cycle involve ribonuclease (4 -9), uridylyl transferase (10,11), RNA ligase (12)(13)(14), and perhaps nucleotidyl phosphatase activities (15). In addition, accessory factors such as matchmaking-type RNA/RNA annealing factors (16 -22) and RNA helicases (23,24) have been shown to contribute. Protein candidates for every step of the basic reaction cycle have been identified, thereby confirming the general features of the enzyme-driven multistep pathway (3,25). Importantly, all key editing enzymes are assembled in a single, high molecular mass mitochondrial complex, which has been termed the editosome (reviewed in Refs. 2 and 26). Editosomes provide a reaction platform for the individual steps of the processing reaction and have been isolated from steady state mitochondrial detergent extracts of African trypanosomes and Leishmania tarentolae (27,28). Trypanosome editosomes have an apparent hydrodynamic size in the range of 20 S and have been visualized by cryoelectronmicroscopy at a resolution of approximately 2 nm. They are composed of two globular subdomains of roughly equal size that are connected by a small interface region. Editosomes have a calculated molecular mass of 0.8 MDa (27) and depending on the enrichment protocol consist of 13-20 polypeptides (for a recent review see Ref. 1). All of the proteins are nuclear encoded and the complex is free of RNA. Because the number of complex-assembled polypeptides exceeds the number of required enzyme activities, it has been suggested that some components have only a structural function (29). However, it cannot be excluded that editosomes harbor additional, so far unidentified enzymatic activities. 20 S editosomes have the ability to bind short, synthetic substrate RNAs and catalyze both U insertion and U deletion editing in a gRNA-dependent fashion (7,30). Although the situation seems to differ between Leishmania and trypanosomes (27,28), steady state isolates of Trypanosoma brucei editosomes contain editing complexes that are loaded with endogenous pre-mRNA/ gRNA substrate RNAs. Because of the different sizes of the various pre-mRNAs, these complexes form a heterogeneous ensemble of particles with hydrodynamic values varying between 35 and 40 S (27). 35-40 S editosomes have a calculated molecular mass of up to 1.45 MDa (27) and consist of a slightly convex platform element that extends on both ends into globular head-like and foot-like protuberances. A three-dimensional alignment of the 20 S and 35-40 S complexes in conjunction with biochemical data identified that the 20 S complex represents a major part of the 35-40 S complex (27) and as a consequence supports a scenario in which substrate RNA binding and release are central determinants in the interconversion of the two complexes. Unfortunately, the molecular details of the editosome/RNA interaction are only marginally understood. Neither the number of RNA-binding sites per editosome, nor kinetic, thermodynamic, and/or selectivity and specificity issues have been addressed systematically. The situation is further complicated by the fact that compositionally different editosomes seem to exist in transgenic trypanosomes (31,32) and that the complexes can associate with multiple RNA ligands: pre-edited, partially edited, and fully edited mRNAs as well as gRNAs and cognate gRNA/pre-mRNA pairs.
Here we present an analysis of the RNA binding characteristics of native T. brucei 20 S editosomes. We show high affinity RNA binding without discrimination between mRNA, gRNA, and mRNA/gRNA hybrid molecules and without discrimination between different mitochondrial transcripts or edited and unedited versions of the same transcript. We further demonstrate that 20 S editosomes have a single RNA-binding site, whereas multiple editosomes can interact with one RNA molecule. Finally, we demonstrate that RNA binding to 20 S editosomes is followed by an RNA unfolding reaction, which is catalyzed by a chaperone-type RNA unwinding activity of the editosome.
RNA Transcription-Unedited (UE) and fully edited (FE) A6 (344 nucleotides/762 nucleotides), UE and FE apocytochrome b (Cyb) (1080 nucleotides/1113 nucleotides), and never edited cytochromoxidase subunit I (COI) (1647 nucleotides) transcripts were PCR-amplified from T. brucei Lister 427 cDNA preparations. PCR products were cloned into pBS SK Ϫ (Stratagene) and sequenced. RNAs were synthesized by run-off transcription from linear DNA plasmid templates using T7 RNA polymerase following standard procedures. Guide RNA gA6-14 was in vitro transcribed as in Ref. 36. Radioactive RNA preparations were generated by in vitro transcription in the presence of ␣-[ 32 P]ATP (specific activity, 3000 Ci/mmol). All of the transcripts were gel electrophoretically purified, eluted from the gel slices, and EtOH-precipitated. RNA folding was performed in EB by heating to 75°C (5min) and cooling to room temperature at 1°C/min.
Substrate RNA Competition of in Vitro RNA Editing-Precleaved RNA editing in vitro insertion and deletion assays were conducted as described (7,30) using preannealed, synthetic pre-mRNA/gRNA hybrid RNAs. The reactions were initiated by a 5 min preincubation at 27°C with 20 S editosomes in EB in the presence of 0.5 mM ATP and 40 M UTP. Competitor pre-mRNA/gRNA hybrid RNA was added (up to a 380-fold molar excess) and incubated for an additional 2 h at 27°C. The reactions were terminated by phenol extraction, EtOH-precipitated, and analyzed in denaturing (8 M urea), 15% (w/v) polyacrylamide gels followed by phosphorimaging.
Surface Plasmon Resonance-gRNAs, pre-edited, and edited mRNAs were 3Ј-oxidized at 4°C overnight in the dark in 10 mM NaIO 4 , 50 mM NaOAc, pH 4.8, 10 mM MgCl 2 , 100 mM NaCl (37). The samples were desalted by size exclusion chromatography and EtOH-precipitated. RNAs were covalently attached to the surface of an amino silane-derivatized microcuvette (Neosensors, UK) in coupling buffer (100 mM Na x H y PO 4 , pH 7.0, 150 mM NaCl, 50 mM NaBH 3 CN) for 3 h at 27°C. Binding was monitored in real time as a shift in the resonant angle. The data were fitted by nonlinear regression, and k diss and k ass values were determined by plotting observed on rates (k on(obs) ) as a function of the complex concentration (k on(obs) ϭ k ass [RNA/20 S complex] ϩ k diss ). Equilibrium dissociation constants (K d ) were calculated as K d ϭ k diss /k ass . The number of binding sites was calculated based on the equation K a R max Ϫ R eq K a ϭ R eq / [20 S]. A plot of R eq versus R eq /[20 S] yields a slope of ϪK a with a x intercept of R max and a y intercept of K a R max . RNA/editosome half-lives (t1 ⁄ 2 ) were determined as t1 ⁄ 2 ϭ ln2/k diss .
Streptavidin Gold Labeling and Transmission EM-20 S editosome preparations were dialyzed to remove glycerol and incubated with synthetic 5Ј-biotinylated (bio) substrate RNAs (bio gRNA, bio pre-edited mRNA, or gRNA/bio pre-edited mRNA hybrid molecules) in EB in a 1:1 molar ratio (27°C, 60 min). Editosome-bio RNA complexes were allowed to adhere to carbon films and were further incubated with streptavidin-derivatized 6-nm gold particles (Aurion, Wageningen, The Netherlands) (room temperature, 30 min). Following an additional 20 min incubation with 2% (w/v) uranyl acetate in H 2 O, the streptavidin-decorated editosome-bio RNA complexes were transferred onto 400 mesh square copper grids (Plano, Wetzlar, Germany) and sandwiched with another carbon film. Air-dried specimens were analyzed by transmission electron microscopy at 50 keV, and pictures were taken at 85,000ϫ magnification. The images were processed using ImageJ (38). The images were smoothed to reduce noise and false colored by applying the ImageJ interactive three-dimensional surface plot plug-in.
Atomic Force Microscopy (AFM)-Editosomes were dialyzed in EB to remove glycerol and incubated with substrate RNAs (pre-mRNA, gRNA, and pre-mRNA/gRNA hybrids) in a 1:1 molar stoichiometry (60 min at 27°C). Editosome-RNA complexes were deposited onto freshly cleaved mica and allowed to adhere for 5 min. The mica was washed (three times with 1 ml of EB) and dried in a mild stream of N 2 . The images were taken in tapping mode in air using a MFP-3D AFM (Asylum Research, Santa Barbara, CA) and OMCL-AC240TS silicon cantilevers (Olympus, Hamburg, Germany) with a nominal spring constant of 2 N/m and a resonance frequency of approximately 70 kHz. All of the images were scanned at a frequency of 1 Hz and were analyzed using the MFP-3D software (Asylum Research). Apparent particle volumes (V app ) were calculated as V app ϭ h(w) 2 /6 (where h is height, and w is width), assuming oblate spheroid shapes of the complexes (39). Contour length measurements were performed by manually tracing using MFP-3D and fitting of the contour length histogram to a Gaussian distribution.

S Editosomes Bind Mitochondrial RNAs Indiscriminately-
T. brucei 20 S editosomes have been shown to bind short (Յ30 nucleotides), synthetic oligoribonucleotides that mimic substrate gRNA and pre-mRNA molecules as well as short, antiparallel gRNA/pre-mRNA hybrid RNAs with nanomolar affinity (27). To analyze the binding of 20 S editosomes to their natural substrates, we examined their interaction with full-length mRNA and gRNA molecules. For that we generated a set of mitochondrial mRNAs that are edited to a different extent. We cloned the UE and FE versions of the ATPase subunit 6 (A6) transcript, which undergoes editing throughout its entire primary sequence (447 U insertions and 28 U deletions) (40). Furthermore, we cloned the UE and FE versions of the Cyb transcript, which contains a single editing domain at its 5Ј-end (34 U insertions) (41), and lastly we used the COI transcript, which is never edited. To derive quantitative data we performed real time binding experiments using a surface plasmon resonancebased read-out system. Fig. 1A shows representative binding isotherms of 20 S editosomes to surface immobilized A6-UE mRNA. The two binding partners interact in a concentrationdependent fashion. Binding is complete within 3-4 min and is characterized by an association rate constant (k ass ) of 4.1 ϫ 10 5 M Ϫ1 s Ϫ1 and a macroscopic equilibrium dissociation constant (K d ) of 4.9 nM (Fig. 1B). Identical experiments with full-length gRNA gA6-14 and gA6-14 hybridized to A6-UE resulted in K d values of 6.5 and 5.5 nM. This indicates high affinity binding and confirms that 20 S editosomes do not discriminate between full-length mRNA, full-length gRNA, and full-length mRNA/ gRNA hybrid molecules. Furthermore, we showed that Cyb-UE transcripts and the never edited COI transcript interact with 20 S editosomes with K d values of 2.9 and 8.6 nM, indicating that pan-edited pre-mRNAs, pre-mRNAs that are edited to a lower extent, and never edited transcripts are bound with similar affinities. Experiments with A6-FE and Cyb-FE resulted in K d values of 3.6 and 12 nM with calculated half-lives (t1 ⁄ 2 ) for the different complexes varying between 2 and 10 min. Taken together ( Fig. 1C and supplemental Fig. S1), the determined association rate constants (k ass ) vary by a factor of 2, the k diss values vary by a factor of 6, and the corresponding macroscopic K d values vary by a factor of 4. This indicates that 20 S editosomes by and large interact with different RNA species in a kinetically and thermodynamically nondiscriminative fashion.
The different 20 S/RNA binding isotherms can be further analyzed to determine the number of 20 S binding sites/RNA substrate molecule (supplemental Fig. S2). For gRNA gA6-14 (70 nucleotides), A6-UE (344 nucleotides), and the A6-UE/ gRNA hybrid molecule (344/70 nucleotides) we derived Յ1 bound editosome per RNA molecule. For Cyb-FE (1113 nucleotides) and Cyb-UE (1080 nucleotides), we identified two interaction sites and the never edited COI transcript (1647 nucleotides) was bound by four editosomes. This demonstrates that multiple editosomes can bind to a single mRNA and further suggests that the editosome/mRNA interaction is characterized by a defined spatial arrangement of Յ470 nucleotides of RNA per 20 S particle. 20 S Editosomes Have One RNA-binding Site-As a follow up, these data raised the question of the number of RNA-binding site(s) per 20 S editosome. To address this issue, we directly visualized editosome-bound RNA by gold labeling in combination with transmission electron microscopy. Affinity-purified 20 S complexes were incubated with 5Ј-biotin-derivatized, synthetic oligoribonucleotides mimicking gRNA, pre-mRNA, and gRNA/pre-mRNA hybrid RNAs. After binding we localized the biotin-modified RNAs with the help of streptavidin-derivatized gold cluster with a mean diameter of 6 nm. Fig. 2 (A and B) shows representative electronmicrographs of gold-labeled mRNA/gRNA hybrid molecules bound to 20 S editosomes. Fig.  2C shows individual gold-labeled editosome-RNA complexes derived from experiments with all three RNA species (gRNA, pre-mRNA, and gRNA/mRNA hybrid RNAs). The complexes are characterized by dimensions of 21 ϫ 26 nm (27), and as expected, the bound gold cluster covers approximately onefourth of the surface area of the individual particles (Fig. 2D). For each RNA binding experiment, we analyzed 10 3 editosome-RNA complexes. Approximately 97% of the complexes showed one bound gold cluster indicating that editosomes have a single substrate RNA-binding site.
This result has the following consequences: First, it implicates that editing substrate RNAs should compete for the editosome-binding site, and second, it suggests that the two types of editing reactions (U deletion and U insertion) might be catalyzed within a single, bifunctional reaction center. To test this hypothesis we conducted substrate RNA competition experiments in combination with in vitro RNA editing activity assays (7,30). The experiments are based on the rationale that a prebound insertion type RNA editing substrate should be competed out of its binding site by an excess of deletion type RNA substrate (and vice versa). Supplemental Fig. S3 shows the result. As anticipated, both the U deletion editing activity as well as the U insertion activity can be inhibited by increasing amounts of the "reciprocal" editing substrate. In both cases half-maximal inhibition was achieved at a Յ10-fold molar excess of competitor RNA, which verifies that the different substrate RNAs act as archetypical competitive inhibitors. Furthermore it shows that the identified substrate-binding site is at or in close proximity to the catalytically active reaction center of the editosome.
Atomic Force Microscopy of Native Editosomes-To derive a more detailed picture of the RNA/editosome interaction, we aimed at visualizing individual RNA/editosome complexes by AFM. As a first experiment we analyzed affinity-purified (TAPtagged) editosome preparations in the absence of exogenously added RNA. Fig. 3A shows a representative result. As anticipated from the published EM data (27) TAP-tagged editosome preparations contain both 20 and 35-40 S complexes. Both particles are characterized by a well defined, roundish appearance and can be distinguished by their diameter and height (Fig.  3, B and C). 20 S complexes have a mean diameter of 24.4 nm and an average height of 4 nm (Fig. 3D). 35-40 S particles have a mean diameter of 49.3 nm and an average height of 7 nm (Fig.  3E). These numbers deviate from the published EM-based measurements (27) and are likely influenced by two phenomena: (i) flattening of the complexes during the AFM analysis (39, 42, 43) and (ii) shrinking of the particles caused by dehydration on the mica surface (44,45). Nevertheless, the dimensions can be used to calculate apparent volumes (V app ) for the complexes, which revealed Gaussian distributions with an average volume of 3-4 ϫ 10 3 nm 3 for the 20 S editosome and 6 -7 ϫ 10 3 nm 3 for the 35-40 S editosome (Fig. 3, D and E). In agreement with recently published data (46), AFM of pure RNA preparations showed that the molecules are highly folded. Fig. 3F illustrates as an example a preparation of the A6-UE transcript. The molecules appear as a monodisperse population of small roundish structures with diameters varying between 7 and 20 nm (mean diameter, 13 nm) and an average height of Յ2 nm (Fig. 3G).
Atomic Force Microscopy of Editosome-RNA Complexes-For the visualization of individual editosome-RNA complexes, we incubated the two reactants at equimolar concentrations for 0 -60 min. Fig. 4 shows representative experiments using the transcripts A6-UE, A6-FE, Cyb-UE, and COI. At zero incubation time, the two reactants are well separated and can be distinguished by their characteristic dimensions (Fig. 4A). After 5 min (Fig. 4B) binding can be detected, which surprisingly is accompanied by an unfolding of the RNA structure (see inset in Fig. 4B). The highly folded RNAs (Fig. 3F) are opened up through the interaction with 20 S editosomes, and the reaction increases over time (Fig. 4C). Furthermore, and in line with the binding data, more than one editosome interacts with a fully unfolded mRNA molecule. Fig. 4 (D and F) shows a representative example of an A6-UE mRNA-editosome complex after 5 min of incubation. Multiple 20 S particles (on average 2-3) are bound to one A6-UE transcript and are distributed over the entire length of the RNA. No preferential binding sites were identified, confirming the non-sequence-specific binding of RNA by 20 S editosomes. After 60 min, RNA-associated structures with dimensions even greater than that of a 20 S editosome were observed (Fig. 4, E and G). These structures likely represent editosome multimers (likely dimers and trimers). All tested mRNAs (Fig. 4, H, J, and K) were substrates of the unwinding reaction although with different kinetic (after 60 min): Cyb-UE Ͼ A6-UE Ͼ COI ϭ A6-FE. This likely reflects the different higher order foldings, i.e., thermodynamic stabilities of the various RNAs. Measuring the contour length of three of the transcripts (COI, A6-UE, and Cyb-UE) after 60 min identified that the A6-UE and Cyb-UE transcripts were fully unfolded, whereas the COI mRNA was resolved to only approximately 30% of its theoretical length (supplemental Fig. S4). The average thickness of an unfolded RNA strand was determined as Յ2 nm in line with previously published data (47). 20 S Editosomes Execute RNA Unwinding Activity-To confirm the editosome-driven RNA unfolding reaction, we measured RNA unfolding by biochemical means. The assay is based on the rationale that opening up the higher order structure of an RNA should make the molecule RNase susceptible. For that, we generated radioactively labeled mRNA preparations, which were incubated with 20 S editosomes in the presence of structure-specific RNases. The reaction was performed at limiting enzyme conditions to ensure that in the absence of editosomes, all input RNA remained undigested during the incubation. Fig.  5 shows a representative experiment using the A6-UE transcript as an example. The RNA was hydrolyzed with RNase T1 (Fig. 5A), which degrades RNA 3Ј of single-stranded G-residues, as well as with cobra venom RNase V1 (Fig. 5D), which cleaves base-paired nucleotides. The generated cleavage products were separated in denaturing polyacrylamide gels, and the percentage of degradation was determined. Incubation of 2 nM A6-UE mRNA with increasing concentrations (0 -40 nM) of 20 S editosomes for 60 min at 27°C led to an increased degradation of the RNA by the two ribonucleases (Fig. 5, B and E). The degradation follows a general decay function (Fig. 5, C and F), and the RNase T1 digestion is Ն90% complete at an editosome concentration of Ն10 nM. The RNase V1 digest requires 40 nM editing complex to achieve the same result. Half-maximal degradation is accomplished at approximately 5 nM editosomes in both experiments (Fig. 5, C and F). The insets in Fig. 5 (C and F) show the results of two time course experiments. The degradation follows a decay kinetic with half-maximal degradation after approximately 15 min and approximately 80 -90% degradation after 60 min in both cases. As expected for a RNA chaperonetype activity (48), the reaction does not require exogenous ATP.

DISCUSSION
20 S editosomes are the catalytic machinery of the RNA editing reaction. They represent high molecular mass, "protein only" assemblies that contain all key activities to convert preedited transcripts into edited mRNAs. A crucial step in the reaction cascade is the binding of pre-edited mRNAs and gRNAs to form "substrate RNA-loaded" editosomes. Here we analyzed the thermodynamic and kinetic characteristics, as well as the molecularity of the editosome/RNA interaction in detail. 20 S editosomes have previously been shown to bind short, synthetic gRNA, pre-mRNA, and gRNA/pre-mRNA hybrid oligonucleotides with nanomolar affinity (K d ) (27). In the present study we used a panel of "in vivo sized" substrate RNAs (70 -1647 nucleotides). We determined that full-length mRNAs, gRNAs, and mRNA/gRNA hybrid RNAs interact with editosomes with similar (nanomolar) K d values and almost identical association and dissociation rate constants (k ass /k diss ). Furthermore, we showed that editosomes do not discriminate between transcripts that are extensively (A6) or moderately edited (Cyb) or between pre-edited and edited versions of the same transcript. Even never edited mRNAs (COI) bind with nanomolar affinity. Despite the high affinity of the different 20 S-RNA complexes, the lack of substrate specificity demonstrates that editosomes can interact with RNA molecules in a nonselective fashion, and it further shows that RNA binding is not restricted to exclusively RNA editing substrate RNAs. This suggests an additional, perhaps in parts even RNA editing-independent function of the editosome (see below). Whether that reflects the situation in vivo remains to be tested. Given the fact that the RNA editing machinery has been shown to interact with various gRNA-and mRNA-containing ribonucleoprotein complexes, it seems plausible that some level of discrimination can be achieved by binding protein-associated RNAs rather than "naked" RNA molecules. Potential candidates are the identified mitochondrial RNA binding and mRNA-stabilizing proteins and complexes as well as chaperone-type RNA annealing factors and RNA helicases (reviewed in Refs. 2 and 49 -51).
We also found no evidence for a discrimination between transcripts that undergo insertion type editing only and transcripts that undergo both insertion and deletion type RNA editing. This raised the questions of whether editosomes have one or multiple reaction centers and whether the two different editing reactions are catalyzed by the same or by separate (sub)complexes with separate RNA-binding sites. Using transmission electron microscopy, we were able to demonstrate that editosomes have only one substrate RNA-binding site. All three RNA classes: mRNAs, gRNAs, and mRNA/gRNA hybrids, bind to a single binding domain on the surface of the complex. In only 3% of the examined particles, we identified Ն2 RNA-binding sites. However, these cases could be characterized as aggregates of two or more 20 S complexes. The transmission electron microscopyderived binding site data were supported by visualizing editosome-RNA complexes by atomic force microscopy. AFM confirmed the presence of a single RNA-binding site on the surface of 20 S editosomes; however, the data also showed that multiple 20 S particles can interact with one RNA transcript. On average two or three 20 S particles bound to one mRNA molecule. No preferential binding domains were identified, and the result was confirmed by surface plasmon resonance-based real time binding experiments. For long substrate RNAs (A6-FE, Cyb-FE, Cyb-UE, and COI), we identified multiple bound editosomes and calculated an approximate mean spacing in the range of 300 -500 nucleotides-20 S complex. Importantly, the identified RNA-binding domain is the interaction site for both types of RNA editing substrates. U deletion as well as U insertion pre-mRNA/gRNA model substrates interact with the very same 20 S binding site and as a consequence can act as competitive inhibitors in a reciprocal fashion. Because that also leads to an inhibition of the in vitro editing activity this further demonstrates that the identified substrate-binding domain is structurally linked or located near the catalytically active center of the editosome. It also shows that the two types of editing reactions are catalyzed within a single bifunctional reaction center. These data find additional support in the fact that individual gRNAs can guide both deletion and insertion editing within one editing domain (52). After 60 min of incubation we also observed the formation of editosome dimers, trimers, and even assemblies/ aggregates of higher complexity. The significance of these "multimeric" complexes awaits further testing.
Lastly, the AFM experiments revealed a so far unknown biochemical activity of the editing machinery: RNA binding is followed by RNA unwinding, which resolves the higher order structure of editosome bound RNAs. This implies that 20 S editosomes execute a chaperone-type RNA unwinding activity. Biochemical unfolding experiments confirmed the presence of the activity and demonstrated that the reaction features defined stoichiometric and kinetic characteristics. RNA contour length measurement further demonstrated that some of the bound transcripts (A6-UE, 344 nucleotides; Cyb-UE, 1080 nucleotides) were fully unfolded (after 60 min), whereas others (COI, 1647 nucleotides) were only partially resolved (approximately 30%). This likely reflects a dependence on the length and/or thermodynamic stability of the bound RNAs. RNA chaperones are proteins that recognize substrate RNA with a broad specificity. They play a role in a multitude of biological processes (for a review see Refs. 48 and 53) and modulate RNA folding to generate biochemically active RNA conformations. Generally, the binding of a chaperone protein destabilizes "incorrectly" folded RNA elements, and therefore, within the context of the editing reaction, the most likely function of the activity is to resolve the highly structured conformations of editosomebound pre-edited transcripts. Misfolded and/or conformationally trapped RNAs have been shown to be rate-limiting in other RNA-guided biochemical processes (54 -56), and as a consequence, it seems plausible that the editing machinery exhibits a catalytic activity to side step this potential roadblock. Whether the activity contributes to other editing-specific reaction steps such as the presentation of anchor sequences or the unfolding of gRNA secondary structure elements (57, 58) cannot be deduced from the data here. However, based on the observation that never edited mRNAs are substrates of the unwinding reaction, it seems conceivable that the activity represents a rather promiscuous feature of the editing machinery. The reaction possibly contributes to other mitochondrial RNA-driven processes and thus provides a rationale for the above described lack of substrate specificity.
Which of the protein component(s) of the editosome mediates the unfolding activity is an open question. Importantly, the activity is intrinsic to the 20 S editosome and thus is different from the various accessory RNA structure modulating activities that have been identified within the context of the editing cycle (reviewed in Ref. 59). Therefore, all so far functionally unassigned editosomal proteins represent potential candidates. However, of special importance are the six oligonucleotide/oligosaccharide binding (OB)-fold proteins of the 20 S complex (29). OB-fold proteins in other RNA-based biochemical systems have been shown to provide RNA chaperone-type activities (reviewed in Refs. 48 and 60), and TbMP24/KREPA4, one of the editosomal OB-fold proteins, has recently been identified to execute RNA annealing activity (22). Within that context the yeast exosome core protein Rrp44 is of special interest (61,62). Rrp44 has been shown to catalyze RNA unwinding, and the activity was structurally attributed to a multimeric OB-fold domain. Evidence for a heteromultimeric association of five of the editosomal OB-fold proteins has recently been published (63), which makes this hypothetical "OB-fold core" a prime candidate for the unwinding activity of the 20 S editosome.
Taken together, the presented data support the following scenario ( Fig. 6): mitochondrial RNA molecules bind with high affinity to a single RNA-binding site of the 20 S editosome. Both types of editing substrate RNAs (insertion/deletion) interact with the same binding site, suggesting a bifunctional reaction center of the editing machinery. RNA binding is followed by RNA unwinding. The activity is intrinsic to 20 S editosomes and unfolds higher order RNA structure elements, thereby resolving structural roadblocks. Ultimately the activity facilitates the interaction of multiple editosomes with one pre-mRNA. The unwinding activity is promiscuous, which suggests that editosomes may contribute their unwinding function to other RNAdriven mitochondrial pathways in trypanosomes.