A family of RRM-type RNA-binding proteins specific to plant mitochondria Matthieu Vermel*, Benoit Guermann*, Ludovic Delage, Jean-Michel Grienenberger, Laurence Mare´chal-Drouard, and Jose´ M. Gualberto† Institut de Biologie Mole´culaire des Plantes du Centre National de la Recherche Scientifique, 12 rue du ge´ne´ral Zimmer, 67084 Strasbourg, France Edited by Sharon R. Long, Stanford University, Stanford, CA, and approved March 6, 2002 (received for review January 11, 2002)
Expression of higher plant mitochondrial (mt) genes is regulated at the transcriptional, posttranscriptional, and translational levels, but the vast majority of the mtDNA and RNA-binding proteins involved remain to be identified. Plant mt single-stranded nucleic acid-binding proteins were purified by affinity chromatography, and corresponding genes have been identified. A majority of these proteins belong to a family of RNA-binding proteins characterized by the presence of an N-terminal RNA-recognition motif (RRM) sequence. They diverge in their C-terminal sequences, suggesting that they can be involved in different plant mt regulation processes. Mitochondrial localization of the proteins was confirmed both in vitro and in vivo and by immunolocalization. Binding experiments showed that several proteins have a preference for poly(U)-rich sequences. This mt protein family contains the ubiquitous RRM motif and has no known mt counterpart in non-plant species. Phylogenetic and functional analysis suggest a common ancestor with RNA-binding glycine-rich proteins (GRP), a family of developmentally regulated proteins of unknown function. As with several plant, cyanobacteria, and animal proteins that have similar structures, the expression of one of the Arabidopsis thaliana mt RNA-binding protein genes is induced by low temperatures.
H
igher plants require mitochondrial (mt) function for their survival, which depends on proper mtDNA maintenance and expression (1). At the structural level, the mtDNA of plants is relatively large, and in most species it is constantly reorganized by recombination between repeated sequences (2). Although large portions of mtDNA have been moved around during evolution, the plant mt genome evolves very slowly through nucleotide substitution; plant mt gene sequences have remained remarkably constant, suggesting the existence of very efficient DNA repair systems. On the other hand, the expression of plant mt genes is also complexly regulated, both at the transcriptional, posttranscriptional, and translational levels. For proper maturation of its transcripts, mitochondria puts to play complex processes of intron splicing, 5⬘ and 3⬘ RNA trimming, extensive RNA editing by C to U conversions, and regulation of transcript stability by secondary structures and polyadenylation (3, 4). Despite the importance of these mt processes in plant development, little is known about the factors involved, but at the core of the protein complexes must be DNA-binding proteins involved in mtDNA replication, recombination, repair and transcription, and RNA-binding proteins involved in posttranscriptional RNA maturation and translation. As a first step in the dissection of these complexes, we undertook to purify and identify nucleic acid-binding proteins from plant mitochondria. Most of the proteins identified are plant-specific, and many belong to a previously undescribed family of mt RNA-binding proteins (mRBP) characterized by the presence of an N-terminal RNA recognition motif (RRM). This family of plant mRBPs is phylogenetically and structurally related to plant RNA-binding glycine-rich proteins (GRP), suggesting that their genes evolved from recruitment of a preexisting plant gene that acquired a mt targeting sequence.
5866 –5871 兩 PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9
Materials and Methods Purification of Potato mt Single-Stranded Binding Proteins. Mito-
chondria were prepared from 160 kg of potato (Solanum tuberosum, var. Bintje) tubers and purified on Percoll gradients as described (5). Mitochondria stored at ⫺80°C were thawed and lysed in buffer A (50 mM Tris䡠HCl, pH 8.0兾0.5 mM EDTA兾1 mM DTT兾7.5% glycerol) plus 1.0 M KCl, 0.5% Triton X-100, 1 mM PMSF, and a mixture of protease inhibitors (Complete, Roche Molecular Biochemicals). After centrifugation at 50,000 ⫻ g, the supernatant (2.1 g) was diluted to 0.35 M KCl and loaded on a 30-ml Q-fast flow column (Pharmacia). Unbound proteins were fractionated by ammonium sulfate precipitation. The predominantly hydrophilic nucleic acid-binding proteins were present in the fraction precipitating at high salt concentration (30–80% saturation). The 30–80% ammonium sulfate fraction (840 mg) was equilibrated in buffer A plus 100 mM NaCl by gel filtration and loaded onto a 5-ml single-stranded DNA (ssDNA) cellulose column (Sigma). Proteins bound to the column were stepwise eluted with 0.3, 0.6, and 2.0 M NaCl. Proteins from each fraction were analyzed by SDS兾PAGE, blotted on poly(vinylidene difluoride) membranes, and the N-terminal sequences of the predominant proteins were determined (on an Applied Biosystems 473A sequencer). MT Localization. Precursor proteins were synthesized from the
full-length cDNA clones by coupled transcription-translation in the presence of [35S]methionine (TNT-coupled reticulocyte lysate system, Promega), and in vitro import assays were conducted as described (6). For in vivo intracellular localization, the gene coding sequences were cloned in the NcoI–BamHI sites of plasmid pCK-GFP3A (7), which allows the expression of protein-eGFP (enhanced green fluorescent protein) fusions under the control of a 35S promoter. The resulting plasmids were transfected in tobacco leaves by bombardment (8), and images were obtained with a Zeiss LSM510 confocal microscope. As an internal mt marker, the pCK-mRFP plasmid was cotransfected with the tested constructions. The pCK-mRFP plasmid harbors the DsRED1 gene (CLONTECH) that codes for the red fluorescent protein (RFP) fused to the yeast COXIV mt presequence. For immunolocalization, the complete At-mRBP1a sequence and the last 41 codons of At-mRBP2b were amplified and cloned in the Escherichia coli expression vectors pQE60 (Quiagen) and pRSET (Invitrogen), respectively. The recombi-
This paper was submitted directly (Track II) to the PNAS office. Abbreviations: mt, mitochondrial; mRBP, mitochondrial RNA-binding protein; EST, expressed sequence tag; GR, glycine-rich; GRP, glycine-rich protein; RRM, RNA-recognition motif; ABA, abscissic acid. Data deposition: The sequences reported in this paper have been deposited in the GenBank database (accession nos. AY048972 and AY048973). *M.V. and B.G. contributed equally to this work. †To
whom reprint requests should be addressed. E-mail:
[email protected].
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.092019599
nant proteins were purified and injected into rabbits to raise antibodies, which were immunopurified against the antigens as described (9). The purified antibodies were used to probe Western blots of subcellular Arabidopsis fractions purified from cell cultures as described (10). cDNA Screening and Sequence Analysis. Potato and tobacco
mRBP2 cDNA clones were obtained from the screening of potato flower and tobacco BY2 cells cDNA libraries. The accession numbers of the Nt-mRBP2b and St-mRBP2 cDNA sequences are AY048972 and AY048973, respectively. Complete At-mRBP1a, At-mRBP1b, At-mRBP2a, and At-mRBP2b expressed sequence tag (EST) clones (accession nos. T22144, T13737, AV551113, and AI998233, respectively) were obtained from the Arabidopsis Biological Resource Center (Ohio State Univ., Columbus) and from the Kazusa DNA Research Institute (Chiba, Japan). The MACVECTOR software (Oxford Molecular) was used for sequences and phylogentic analysis, and organellar targeting predictions determined with PREDOTAR and TARGETP (www.inra.fr兾Internet兾Produits兾Predotar兾; www.cbs.dtu.dk兾 services兾TargetP兾).
Nucleic Acid-Binding Assay. The sequences coding for the mature
proteins (without targeting peptides) were amplified, cloned in pBluscript SK⫺ (Stratagene), and labeled proteins were synthesized by coupled transcription-translation in the presence of [ 35 S]methionine. Homoribopolymers [poly(A), poly(C), poly(G), and poly(U), Sigma] were coupled to Sepharose 4B (Pharmacia) at 2.0, 1.4, 1.6, and 1.9 mg/ml swollen beads, respectively, and ssDNA-cellulose and double stranded DNA (dsDNA)-cellulose were obtained from Sigma. The nucleic acid binding assay was carried out essentially as described (12). The bound proteins were analyzed by SDS兾PAGE, and the relative protein amounts were calculated by PhosphorImager quantification. Results Isolation of mt ssDNA-Binding Proteins. Sufficient amounts of highly
purified mitochondria were necessary for mRBP purification. After preliminary experiments in which different sources of plant mitochondria were tested, mitochondria purified from potato tubers on Percoll gradients were found to be virtually free of cytosolic, nuclear, or plastidic contamination (13), and were chosen as starting material in the mRBP purification. The purification protocol chosen was based on the affinity chromatography to ssDNA. A similar approach had previously been successfully used to purify chloroplast RNA-binding proteins (14). Potato mt proteins displaying high affinity for ssDNA were purified (Fig. 1), and the N-terminal sequences of the predominant proteins were determined (Table 1). Among the sequences obtained, few correspond to known proteins (HSP60 and histone H3), several could not be matched to any known protein (P100, P50, P24, P19), and two (P18 and P26) match proteins of unknown functions encoded by several tomato EST clones that were obtained and sequenced. They code for precursor proteins with N-terminal extensions that probably are mt targeting sequences. That the tomato P18 protein is a mt protein has been confirmed by in vitro impact experiments (not shown). Vermel et al.
Fig. 1. Purification of potato ssDNA-binding proteins. The fractions analyzed are the ones described in Materials and Methods. Equivalent protein amounts were loaded. Only the proteins from which peptide sequence data could be obtained are indicated. Proteins with a same name in different elution fractions were verified to have identical N-terminal sequences. Those that are underlined contain a RNP-2 consensus sequence (LFVGGL) at their N terminus.
All other proteins identified (Fig. 1) seem to belong to a family of mt RNA-binding proteins characterized by the presence of an N-terminal RRM motif. mRBPs, A Family of MT RRM-Containing GRP. Eight of the proteins
sequenced (P12, P13, P30, P32, P35, P39, P41, and P75) have N-terminal sequences that match the RNP-2 consensus of the RRM motif (15). Thus, they apparently are RNA-binding proteins with N-terminal RRM sequences (Table 1). We named this family of mitochondrial RNA-Binding Proteins as mRBP, and subdivided some of them in subfamilies (mRBP1and mRBP2) according to their sizes and sequences (Table 1). From the peptide sequences, corresponding genes could be identified in different plant species. The potato P13 and P12 proteins (mRBP1 subfamily) have in Arabidopsis two orthologues that we named At-mRBP1a and At-mRBP1b (see Table 1). The Arabidopsis precursor proteins display a targeting peptide sequence, a single RRM motif and a short C-terminal glycine-rich (GR) sequence mainly constituted by the repetition of a GGGGGY motif (Fig. 2). The major difference between At-mRBP1a and b is the size of the GR sequences. Homologous sequences could also be found in other plant species, including the previously identified Nicotiana sylvestris RGP-3 gene (16). The potato P35, P32, and P30 proteins (mRBP2 subfamily) have identical N-terminal sequences that identified them (24 of 25 amino acids identity) as homologues of the N. sylvestris RGP-2 protein (17). Although RGP-2 was considered to be a nuclear or cytosolic protein, the PREDOTAR and TARGETP software programs predict that it has a potential mt targeting peptide. By screening Nicotiana tabacum and potato cDNA libraries, complete cDNA clones were obtained, corresponding to two different tobacco genes (63% amino acid identity) and to a single potato gene. We called the corresponding proteins Nt-mRBP2a (previously RGP-2), Nt-mRBP2b, and St-mRBP2 (Fig. 2). By its size and sequence, St-mRBP2 seems to correspond to the P35 protein that we had purified (Fig. 1). When compared, the tobacco and potato mRBP2 proteins have equivalent structures, PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5867
BIOCHEMISTRY
RNA Isolation and Northern Blot Analysis. Arabidopsis thaliana (ecotype Columbia) RNA were extracted from 4- to 6-week-old plants as described (11), and Northern blots were hybridized with random-primed labeled probes. Stress conditions tested were the following: cold treatment (24 h at 4°C), wounding (with forceps), drought (1 week without watering), and abscissic acid (ABA) treatment (by spraying with a 50 M ABA solution in 0.01% Tween-20, control plants were sprayed with the same solution, but without ABA).
Table 1. N-terminal sequences of the potato proteins indicated in Figure 1 Protein P100 P75 P60 P50 P41 P41-a P41-b P41-c P41-d P39 P35 P32 P30 P26 P24 P19 P18 P17 P13 P12
Sequence
Identification兾Given name
AQAAPVPlllVLLlLxDLvf prxLFVxGl aAKDIKFGvEArGLxLQGVEQL xSNGLDvGNANCPNCGEsIF ASKLFVGGLSFGTDDQSLKEAF LFVGGLSYGTDETSL PVVAGVKAHEFDAYA IEIYDEAEDVSSWPKPSEI NFLLDENDNQHSDY xxKLFVxaGLxy SSsKLFVGGLSyGTDEtSLR SSSKLFVGGLSYGTDETSLRxAF sSSKLFVGGLSYGTDETSLREAFSQxG sTVRQNVVADGrKEGRVFAPYSVF WTGTcFGxGDFPxxRNlEFAP PWKfGVGPgexVGPGIGf sAKASSsPKDKDRERLMFDxERET xARTKQTARKSTGGKAPRKQLATKA aSKLFVGGLSxGTDDQxxRDAF sSKLFVGGLsxgTDDQaLKEAFs
– mRBP? HSP60 – mRBP
– – HSP60: At3g23990 – –
mRBP
– At-mRBP2a: At1g74230 At-mRBP2b: At5g61031
mRBP2 subfamily* Tomato AW624491 – – Tomato AI486749 Histone H3 mRBP1 subfamily†
Arabidopsis homologues
At1g71260; At2g02740 – – At1g71310 Histone H3: 8 genes At-mRBP1a: At4g13850 At-mRBP1b: At3g23830
Resides that match the RNP-2 consensus motif are indicated in bold and underlined. Characters in lower case indicate ambiguities in the sequence. Sequences P41-a to -d correspond to tryptic peptides. Accession number of genes discussed in text are indicated. *Potato St-mRBP2: AY048973; Tobacco Nt-mRBP2a (previous RGP-2): BAA05170; and Nt-mRBP2b: AY048973. †Similarity to wood tobacco RGP-3: BAA22083.
consisting of a targeting peptide, an N-terminal RRM, a GR sequence, and a C-terminal acidic domain. In Arabidopsis, two genes code for proteins with equivalent sizes and structures (At-mRBP2a and At-mRBP2b, see Table 1). Their predicted sequences were confirmed by sequencing full-length EST clones. They diverge from the tobacco and potato mRBP2 proteins in their GR and acidic domains, but the RRM sequences and, to a
lesser extent, the targeting peptides are relatively well conserved (Fig. 2). The potato mRBP2 proteins that we have purified range in size between 30 and 35 kDa and their N-terminal sequences are identical. However, the screening of the flower cDNA library yielded clones from just a single gene. It could be possible that alternative splicing is responsible for the synthesis of smaller
Fig. 2. Alignment of mRBP1 and mRBP2 sequences identified in this work, together with other putative Arabidopsis mRBPs and a representative GRP (At-GRP8). The RNP-2 and RNP-1 consensus sequences of the RRM are boxed. The arrow indicates the site of targeting peptide processing in the mRBP1 and mRBP2 proteins. At, A. thaliana; St, S. tuberosum; Nt, N. tabacum. Similarities between the C-terminal parts of potato and tobacco mRBP2 are in white characters on gray background, whereas the similarities between the C-terminal parts of the Arabidopsis mRBP2 proteins are in black characters on gray background. 5868 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.092019599
Vermel et al.
proteins, but reverse transcription (RT)-PCR experiments failed to detect such alternatively spliced products in total RNA of potato tubers and sprouts. Therefore, the P30 and P32 proteins might be posttranslation processing products of St-mRBP2. Among the proteins purified and sequenced, the P41 protein is one of the predominant (Fig. 1). To amplify the corresponding cDNA, additional sequence data were obtained from tryptic peptides (Table 1), and degenerate oligonucleotides were used in RT-PCR experiments using RNA from potato tubers or sprouts, but up to now without success. The Arabidopsis databases were also screened for genes coding for a corresponding protein, that is, coding for a protein of equivalent size, predicted to be targeted to plant organelles and containing an N-terminal RRM motif, but that gene could not be identified. Therefore, it is possible that in Arabidopsis the corresponding function is either dispensable or is fulfilled by a different protein. Other Predicted mRBP Proteins. The mRBP proteins that we have
identified have a distinctive structure (they have a well conserved N-terminal RRM motif, a mt targeting sequence predicted as organellar by at least one of the target prediction programs, and they diverge from chloroplast RNA-binding proteins that typically contain two RRM motifs that are not N-terminal). We used these criteria to screen the complete predicted proteome of Arabidopsis for additional putative mRBP proteins. Apart AtmRBP1a and -b and At-mRBP2a and -b, seven other proteins were identified (At5g54580, At1g18630, At2g37510, At3g08000, At5g06210, At4g13860, and At5g47320), ranging in size between 126 and 212 aa (Fig. 2). These include the nuclear-encoded mt RPS19, a protein that consists in the fusion of a RRM domain to the mt ribosomal protein S19 (18). The predicted mature proteins contain a single RRM sequence and diverge mainly in their C-terminal sequences. So far, mt targeting of At5g54580 has already been confirmed by expression of protein–eGFP fusions (data not shown).
The mRBP Proteins Are Strictly Localized in Mitochondria. Several
plant proteins have been studied that display protein structures
Vermel et al.
similar to the mature mRBP1 proteins (a single N-terminal RRM and a C-terminal GR sequence) and have been localized in the nucleus (16, 19). To confirm that the mRBP proteins are indeed targeted to mitochondria, import experiments were performed with the in vitro synthesized proteins. All proteins tested [At-mRBP1a, At-mRBP1b, At-mRBP2b, and St-mRBP2 (Fig. 3A), and Nt-mRBP2b (data not shown)] could be imported efficiently into purified potato mitochondria in a membrane potential dependent manner and with cleavage of the targeting peptide. In addition, protein–eGFP fusions of At-mRBP1a, At-mRBP2a, and At-mRBP2b (Fig. 3C), and St-mRBP2 (data not shown), when expressed transiently in tobacco leaf cells were only found in mitochondria. The strict mt localization was further demonstrated in the case of At-mRBP1a and AtmRBP2b, for which highly specific antibodies have been obtained (Fig. 3B). The ratio between the signals obtained in the mt and total cellular extract are comparable to the one obtained for a mt encoded complex I protein (NAD9). Thus, a hypothetical alternative translation initiation at the second ATG codon do not account for a double targeting of the mRBPs to the nucleus or cytosol. With the At-mRBP2b antibody, two predominant proteins are detected. Apparently both are RRM proteins, because they are also detected (faintly) by the At-mRBP1a antibody, which is directed against the full size mature protein. The smaller protein probably is At-mRBP2b, because it has the same apparent size has the in vitro synthesized protein. The upper band might be At-mRBP2a, or a posttranslationally modified form of At-mRBP2b. mRBPs Are Poly(U)-Rich RNA-Binding Proteins. For a preliminary
analysis of the nucleic acids binding specificity of mRBPs, the At-mRBP1a, At-mRBP2b, and St-mRBP2 mature proteins were synthesized in vitro, and their capability to bind to homoribopolymers, ssDNA, and dsDNA was tested. The three proteins showed much higher affinity to poly(U) than to the other three homoribopolymers or to DNA (Fig. 4), and a significant proportion of the proteins bound to poly(U) even at high salt concentrations (up to 1.0 M, data not shown). This result is consistent with the
PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5869
BIOCHEMISTRY
Fig. 3. Mitochondrial localization of mRBP proteins. (A) In vitro import of the proteins into purified potato mitochondria. Mitochondrial internalization is demonstrated by the resistance to proteinase K treatment (10 g/ml). If mitochondria are lysed with 0.5% Triton X-100, the mature proteins are rapidly degraded by the protease. Protein import depends on mt membrane potential, as demonstrated by inhibition of import by the uncoupling agent valinomycin (250 nM final). (B) Immunolocalization of At-mRBP1a, At-mRBP2b, NAD9 (subunit 9 of NADH dehydrogenase), and RBCS (small subunit of Rubisco) in Arabidopsis extracts. T, total extract; C, cytosol; Cl, chloroplast. (C) Mitochondrial localization of the mRBP1 proteins fused to eGFP in N. tabacum stomata (see Materials and Methods). In each case, only the lower guard cell was transfected. (i) Autofluorescence of chlorophyll (white) and fluorescence of the mt marker (red). (ii) Fluorescence of eGFP fusion protein. (iii) Merged images, including image obtained under visible light.
cold stress, and the protein might be involved in the modulation of mt gene expression concomitant with plant acclimation to low temperatures.
Fig. 4. Nucleic acid-binding specificity of mRBP proteins. The mature (without targeting sequences) At-mRBP1a, At-mRBP2b, and St-mRBP2 proteins were synthesized in vitro in the presence of [35S]methionine, and their binding to homoribopolymers, ssDNA, and dsDNA were tested. The relative amounts of bound proteins are shown, taking the binding to poly(U) as standard.
results obtained with other proteins displaying similar RRM sequences that also have high affinity for poly(U) (16, 17, 20). Thus, the mRBP proteins analyzed probably have a preference for poly(U)-rich sequences, but it is also possible that they have other preferred RNA sequence targets and are unable to bind other homoribopolymers because of steric constraints. It is surprising that they bind poorly to ssDNA, the affinity matrix used for their purification. A possible explanation is that, in the total mt extract, there is cooperative binding of the mRBPs to the nucleic-acids matrix. The Arabidopsis mRBP1 Gene Is Induced by Cold Treatment. The mRBP1 proteins are strikingly similar to several plant and non-plant proteins that have a similar structure, except for the presence of the mt targeting peptide. They are the plant family of RNA-binding GRP (21), the animal cold inducible RNAbinding proteins (22), and the cyanobacteria Rbp proteins (23). Transcription of the corresponding genes is regulated by different stresses, mainly by cold treatment, but their precise functions remain unknown. We tested whether the Arabidopsis AtmRBP1a gene is induced by several abiotic stress treatments and to the stress-associated phytohormone abscissic acid. Total RNAs of treated Arabidopsis plants were probed with the At-mRBP1a cDNA. A faint RNA band of the expected size could be detected, and was significantly (7-fold) induced in cold-treated plants (Fig. 5). No effect was observed in wounded, drought-treated, and ABA-treated plants. The same cold treatment had no effect in the accumulation of the transcript of another nuclear-encoded mt protein (apocytochrome c). Thus, at least the Arabidopsis At-mRBP1a gene seems to be induced by
Fig. 5. Expression of the At-mRBP1a gene is induced by cold treatment. A Northern blot of total RNA (40 g) extracted from treated (⫹) and control (⫺) Arabidopsis plants was probed with a specific At-mRBP1a gene probe and subsequently with an apocytochrome c probe. PhosphorImager quantification estimated a 7-fold increase of At-mRBP1a mRNA in cold-treated plants. 5870 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.092019599
Discussion We have purified plant mt proteins that have affinity to singlestranded nucleic acids by a conventional biochemical approach. The presence of histone H3 among the purified proteins does not seem to reflect a nuclear contamination of the mitochondria preparations, because no other known nuclear protein has been detected, including histones H2 and H4, which are associated with H3 in the nucleosome. In vitro import experiments suggested that the Arabidopsis H3 can indeed bind to the mt outer membrane, but is not internalyzed (data not shown). HSP60 was also found in the purified fractions, and it cannot be excluded that in plant mitochondria it assumes a double function as a nucleic acids-binding protein, as it seems to be the case of the yeast mt HSP60 (24). Apart from these two known proteins, several of the purified proteins still require further analysis for their identification, but many could been assigned as belonging to a previously uncharacterized important family of mt RNAbinding proteins. As chloroplast RNA-binding proteins (14), mRBPs contain the characteristic RRM motif (also known as the consensus-type RNA-binding domain, CS-RBD). It is present as one or more copies in animals, fungi, plants, and cyanobacteria proteins involved in multiple posttranscriptional processes (15). However, up to now it had not been found in mt proteins, or in proteins from ␣ purple bacteria (such as Rickettsia prowazekii), which are close relatives to the protosymbiont ancestor of mitochondria (25). Thus, the plant mt mRBP genes probably evolved from a preexisting plant gene that has been recruited to assume specific mt functions. Origin of the mRBP Gene Family: GR RNA-Binding Proteins Regulated by Stress. The mRBP1 proteins are strikingly similar to several
plant, animal and cyanobacteria proteins that have a similar structure, except for the presence of the mt targeting peptide (21–23). These proteins are generally induced by different abiotic stresses, but their functions remain largely unknown. That is the case of the Arabidopsis At-GRP7 and At-GRP8 proteins, whose genes are regulated during the circadian cycle, are induced by low temperatures, and are repressed by ABA (26). Animal cold inducible RNA-binding proteins and cyanobacteria Rbp proteins are also cold-induced, but phylogenetic analysis suggested that the presence of GR domains and cold regulation in the proteins of such different species is the result of convergent evolution (27). Compared with plant and animal GRPs, the mRBP1 proteins differ by the presence of the targeting peptide and the length of the GR sequence, that in GRPs is longer (70–80 aa compared with 14–29 in mRBP1) and contain repeats of the RGG-box, which has also been defined as an RNA-binding motif (15). In that respect mRBP1 proteins seem closer to cyanobacterial Rbp proteins, that have a similar N-terminal RRM and a short C-terminal GR domain. However, sequence comparison between the Arabidopsis GRP and mRBP genes strongly suggest a direct relationship between the two gene families; Arabidopsis GRP genes (such as At-GRP7 and At-GRP8) contain a single intron, whereas the mRBP genes conserved the same intron, at the same position, and acquired additional introns (Fig. 6A). A phylogenetic analysis of the RRM sequences found in plant and bacteria proteins also support the hypothesis that mRBPs evolved from GRPs: their RRM sequences branch together in the tree, with a strong bootstrap value of 86 (Fig. 6B). The comparison of the intron positions has in addition confirmed the common origin of the mRBP1 and mRBP2 genes of Arabidopsis: all four genes share the same three introns in the RRM coding Vermel et al.
similar targeting sequence (40% identity to the At-mRBP1a targeting peptide). This protein is mitochondrial, as demonstrated by in vitro import experiments (not shown). It is thus conceivable that the mRBP gene ancestor acquired the targeting sequence from the At5g49210 gene. As an opposite example, the sequence (Fig. 2) and introns conservation (Fig. 6A) show that the targeting and RRM sequences of the Arabidopsis rps19 gene have originated from a mRBP1 gene. The N-terminal RRM acquired by RPS19 was postulated to complement the absence of mt ribosomal protein S13 in Arabidopsis (18).
sequence. The mRBP2 subfamily probably evolved from mRBP1 genes by the acquisition of a different C-terminal sequence (Fig. 6A). Because mRBP genes probably evolved from preexisting GRP genes, we considered the possibility that they retained the developmental regulation of the ancestor gene. Indeed, Northern blot analysis suggest that at least the Arabidopsis mRBP1a gene is cold-induced. If mRBPs evolved from an ancestor GRP gene by the acquisition of a targeting peptide sequence, it is possible that it acquired the targeting peptide sequence from another gene, as described for other genes coding for mt proteins (28, 29). An Arabidopsis protein of unknown function (At5g49210) has a 1. Mackenzie, S. & McIntosh, L. (1999) Plant Cell 11, 571–586. 2. Wolstenholme, D. R. & Fauron, C. M. R. (1995) in The Molecular Biology of Plant Mitochondria, eds. Levings, C. S. I. & Vasil, I. K. (Kluwer Academic, Dordrecht, The Netherlands), pp. 1–60. 3. Binder, S., Marchfelder, A. & Brennicke, A. (1996) Plant Mol. Biol. 32, 303–314. 4. Gagliardi, D. & Leaver, C. J. (1999) EMBO J. 18, 3757–3766. 5. Douce, R., Bourguignon, J., Brouquisse, R. & Neuburger, M. (1987) Methods Enzymol. 148, 403–414. 6. Wischmann, C. & Schuster, W. (1995) FEBS Lett. 374, 152–156. 7. Menand, B., Marechal-Drouard, L., Sakamoto, W., Dietrich, A. & Wintz, H. (1998) Proc. Natl. Acad. Sci. USA 95, 11014–11019. 8. Sanford, J. C., Smith, F. D. & Russell, J. A. (1993) Methods Enzymol. 217, 483–509. 9. Harlow, E. & Lane, D. (1988) Antibodies: A Laboratory Manual (Cold Spring Harbor Lab. Press, Plainview, NY). 10. Laloi, C., Rayapuram, N., Chartier, Y., Grienenberger, J. M., Bonnard, G. & Meyer, Y. (2001) Proc. Natl. Acad. Sci. USA 98, 14144–14149. 11. Goodall, G. J., Wiebauer, K. & Filipowicz, W. (1990) Methods Enzymol. 181, 148–161. 12. Ye, L. & Sugiura, M. (1992) Nucleic Acids Res. 20, 6275–6279. 13. Neuburger, M., Journet, E. P., Bligny, R., Carde, J. P. & Douce, R. (1982) Arch. Biochem. Biophys. 217, 312–323. 14. Li, Y. & Sugiura, M. (1990) EMBO J. 9, 3059–3066. 15. Burd, C. G. & Dreyfuss, G. (1994) Science 265, 615–620. 16. Moriguchi, K., Sugita, M. & Sugiura, M. (1997) Plant J. 12, 215–221.
Vermel et al.
We thank Monique Le Ret and Philippe Hammann for their assistance in protein and DNA sequencing, and Patrice Imbault and Dominique Gagliardi for helpful discussions and critical reading of the manuscript. Potato flower and tobacco BY2 cells cDNA libraries were kindly provided by Drs. M. Klein (Potsdam) and W-H. Shen (Strasbourg), respectively. This work was supported by the Centre National de la Recherche Scientifique (UPR2357) and by Ministe`re de la Recherche fellowships (to M.V., B.G., and L.D.). 17. Hirose, T., Sugita, M. & Sugiura, M. (1994) Mol. Gen. Genet. 244, 360–366. 18. Sa´nchez, H., Fester, T., Kloska, S., Schro ¨der, W. & Schuster, W. (1996) EMBO J. 15, 2138–2149. 19. Alba`, M. M., Culia´n ˜ez-Macia`, F. A., Goday, A., Freire, M. A., Belen, N. & Page`s, M. (1994) Plant J. 6, 825–834. 20. Ludevid, M. D., Freire, M. A., Gomez, J., Burd, C. G., Albericio, F., Giralt, E., Dreyfuss, G. & Pages, M. (1992) Plant J. 2, 999–1003. 21. Alba`, M. M. & Page`s, M. (1998) Trends Plant Sci. 3, 15–21. 22. Nishiyama, H., Itoh, K., Kaneko, Y., Kishishita, M., Yoshida, O. & Fujita, J. (1997) J. Cell Biol. 137, 899–908. 23. Sugita, M. & Sugiura, M. (1994) Nucleic Acids Res. 22, 25–31. 24. Kaufman, B. A., Newman, S. M., Hallberg, R. L., Slaughter, C. A., Perlman, P. S. & Butow, R. A. (2000) Proc. Natl. Acad. Sci. USA 97, 7772–7777. 25. Lang, B. F., Gray, M. W. & Burger, G. (1999) Annu. Rev. Genet. 33, 351–397. 26. van Nocker, S. & Vierstra, R. D. (1993) Plant Mol. Biol. 21, 695–699. 27. Maruyama, K., Sato, N. & Ohta, N. (1999) Nucleic Acids Res. 27, 2029–2036. 28. Kadowaki, K., Kubo, N., Ozawa, K. & Hirai, A. (1996) EMBO J. 15, 6652–6661. 29. Kubo, N., Harada, K., Hirai, A. & Kadowaki, K. (1999) Proc. Natl. Acad. Sci. USA 96, 9207–9211. 30. Schuster, G. & Gruissem, W. (1991) EMBO J. 10, 1493–1502. 31. Hirose, T. & Sugiura, M. (2001) EMBO J. 20, 1144–1152. 32. Thomashow, M. F. (2001) Plant Physiol. 125, 89–93. 33. Jiang, W., Hou, Y. & Inouye, M. (1997) J. Biol. Chem. 272, 196–202. 34. Bae, W., Xia, B., Inouye, M. & Severinov, K. (2000) Proc. Natl. Acad. Sci. USA 97, 7784–7789.
PNAS 兩 April 30, 2002 兩 vol. 99 兩 no. 9 兩 5871
BIOCHEMISTRY
Fig. 6. The plant mRBP and RNA-binding GRP proteins have a common ancestor. (A) Structure of Arabidopsis mRBP1 and mRBP2 proteins as compared with the mt ribosomal protein S19 (At-RPS19) and to a characteristic RNA-binding GRP (At-GRP8). Position of introns are indicated by arrowheads. (B) Phylogenetic analysis of the RRM sequences found in representative plant and bacteria proteins. The tree was generated by the neighbor-joining method using the software MACVECTOR. Significant Bootstrap values are indicated. a, At-mRBP1a; b, Nt-RGP3; c, At-mRBP1b; d, At-RPS19; e, At-mRBP2a; f, At-mRBP2b; g, St-mRBP2; h, Nt-mRBP2b; i, Nt-mRBP2a; j, Ns-RGP1a:S41771; k, Ns-RGP1c:S59529; l, Ns-RGP1b:S41772; m, At-GRP7:At2g21660; n, AtGRP8:At4g39260; o, Dc-GRP:Q03878,; p, Zm-GRPA:P10979; q, Sb-GRP2: S12312.
Possible Functions of mRBP Proteins. Proteins containing RRM domains can be associated to multiple posttranscriptional processes (21) and mRBP proteins can a priori be actors in the multiple posttranscriptional mechanisms involved in mt gene expression. They mainly diverge in their C-terminal sequences, which probably are responsible for their function specificity. As an example, chloroplast RNA-binding proteins have up to now been found associated with the regulation of transcript stability (30) and RNA editing (31). In the latter example, an acidic domain of the chloroplast protein seems important for its function. It is possible that mRBP proteins are associated with similar functions, and in that respect the presence of acidic domains in the proteins of the mRBP2 subfamily is probably relevant. In the case of the mRBP1 proteins, the apparent gene induction by low temperatures and the similarity to plant, cyanobacteria, and animal cold-induced RNA-binding GRPs suggest similar functions. That function could be regulation of mt gene expression during plant acclimation to cold, a process that only recently starts to be understood (32). It is for instance possible that, like CspA, the major cold-shock protein of E. coli, mRBP1 proteins function as mRNA chaperones to destabilize secondary structures in mt mRNAs (33, 34). Such a function could be crucial for efficient mt transcription or translation at low temperatures. These hypotheses need now to be tested.
