Molecular Characterization of Three PRORP Proteins in the Moss Physcomitrella patens: Nuclear PRORP Protein Is Not Essential for Moss Viability

RNase P is a ubiquitous endonuclease that removes the 5′ leader sequence from pre-tRNAs in all organisms. In Arabidopsis thaliana, RNA-free proteinaceous RNase Ps (PRORPs) seem to be enzyme(s) for pre-tRNA 5′-end processing in organelles and the nucleus and are thought to have replaced the ribonucleoprotein RNase P variant. However, the evolution and function of plant PRORPs are not fully understood. Here, we identified and characterized three PRORP-like proteins, PpPPR_63, 67, and 104, in the basal land plant, the moss Physcomitrella patens. PpPPR_63 localizes to the nucleus, while PpPPR_67 and PpPPR_104 are found in both the mitochondria and chloroplasts. The three proteins displayed pre-tRNA 5′-end processing activity in vitro. Mutants with knockout (KO) of the PpPPR_63 gene displayed growth retardation of protonemal colonies, indicating that, unlike Arabidopsis nuclear RPORPs, the moss nuclear PpPPR_63 is not essential for viability. In the KO mutant, nuclear-encoded tRNAAsp (GUC) levels were slightly decreased, whereas most nuclear-encoded tRNA levels were not altered. This indicated that most of the cytosolic mature tRNAs were produced normally without proteinaceous RNase P-like PpPPR_63. Single PpPPR_67 or 104 gene KO mutants displayed different phenotypes of protonemal growth and chloroplast tRNAArg (ACG) accumulation. However, the levels of all other tRNAs were not altered in the KO mutants. In addition, in vitro RNase P assays showed that PpPPR_67 and PpPPR_104 efficiently cleaved chloroplast pre-tRNAArg (CCG) and pre-tRNAArg (UCU) but they cleaved pre-tRNAArg (ACG) with different efficiency. This suggests that the two proteins have overlapping function but their substrate specificity is not identical.


Introduction
RNase P is an endonuclease that removes the 59 leader sequence from precursor tRNAs (pre-tRNAs). This endonucleolytic cleavage is an essential step in the production of mature tRNAs in all organisms, as well as in mitochondria and chloroplasts [1,2]. Bacterial, archaeal, and eukaryotic nuclear RNase P enzymes are ribonucleoprotein complexes composed of a catalytic RNA component and one or several proteins [3]. Therefore, it was believed that 59-maturation of tRNAs in all organisms is catalyzed by ubiquitous ribonucleoprotein (RNP) RNase P enzymes. However, human mitochondrial RNase P was identified as an RNA-free enzyme composed of three proteins called mitochondrial RNase P protein 1 (MRPP1), MRPP2, and MRPP3 [4]. MRPP3 contains two RNA-binding pentatricopeptide repeat (PPR) motifs and a conserved NYN metallonuclease domain [5], which are involved in the catalytic activity of mitochondrial RNase P. In a model plant, Arabidopsis thaliana, three MRPP3 homologs have been identified as RNase P enzymes; they are termed proteinaceous RNase P 1 (PRORP1), PRORP2, and PRORP3 [6]. In vitro cleavage assays using recombinant PRORP proteins have demonstrated that three PRORP proteins display RNase P activity [6,7]. PRORPs are single-protein RNase P enzymes, whereas human mitochondrial RNase P is a multisubunit enzyme. PRORP1 localizes in both chloroplasts and mitochondria, whereas PRORP2 and 3 localize in the nucleus. PRORP1 is an essential gene, because prorp1 mutants are embryonic lethal [6]. Single PRORP2 or PRORP3 knockout mutant lines show wild-type phenotypes, whereas the homozygous double mutation in prorp2 prorp3 results in embryonic lethality [7]. This indicates that PRORP2 and 3 have redundant functions and are essential for embryogenesis and plant growth.
The Arabidopsis and rice genomes do not encode RNase P RNA, whereas they do code for several homologs of RNP RNase P protein subunits found in mammals and yeasts [8]. The presence of ribonucleoprotein RNase P enzyme in plants remains obscure. Recently, Gutmann et al. (2012) suggested that Arabidopsis had entirely replaced the ribonucleoprotein RNase P enzyme by PRORP enzymes for tRNA maturation during plant evolution [7]. However, functional characterization of PRORP proteins is limited in Arabidopsis.
Unlike in land plant organelles, an RNA subunit of bacterialtype RNase P is encoded in the organellar genomes of the Glaucophyta (Cyanophora paradoxa) [9], the red alga Poryphyra purpurea [10], and green algae (Nephroselmis olivacea, Ostreococcus tauri [11,12]. These algae also possess a single nuclear gene for PRORP [6]. In C. paradoxa, RNase P activity in an extract of the photosynthesis organelle (cyanelle) is sensitive to micrococcal nuclease [13,14], indicating the existence of bacterial-type RNase P. The nuclear-encoded PRORP in the green alga O. tauri has been shown to cleave the 59 leader of pre-tRNA in vitro [12,15]. Thus, algae seem to possess both a ribonuceoprotein RNase P and a PRORP. However, the latter is not known to localize in either the organelles or the nucleus. From these differences in the status of PRORP enzymes in algae and plants, the following questions have arisen: When were plant PRORP genes duplicated during plant evolution, which PRORP protein was targeted to the organelles and which to the nucleus, and was ribonucleoprotein RNase P enzyme replaced by PRORP in plants? To answer these questions, we focused on PRORPs of the basal land plant bryophyte Physcomitrella patens. Here, we report the molecular characterization of three P. patens PRORPlike proteins (PpPPR_63, 67, and 104) that are members of the PPR protein family [16,17]. We also report the disruption mutants of each gene and show that nuclear-localized PpPPR_63 is dispensable for moss viability and that organelle-localized PpPPR_67 and PpPPR_104 have functions that are redundant, but not completely so.

Results
The moss P. patens has three PRORP homologs Among 105 Physcomitrella PPR proteins [17], PpPPR_63 (protein ID, Phypa_174001), PpPPR_67 (Phypa_177191) and PpPPR_104 (Phypa_152956) appeared to be PRORP homologs (Fig. 1A), because these three PPR proteins contained at least two PPR motifs and an NYN metallonuclease domain and had 42%-53% amino acid identity with Arabidopsis PRORP proteins. Although algae and Chara have a single PRORP homolog, mosses and vascular plants have two or three PRORP homologs (Fig. 1B  and Table S1). This suggests that PRORP genes were duplicated after the emergence of land plants.
PpPPR_63 is localized in the nucleus, whereas PpPPR_67 and 104 are found in the organelles PpPPR_67 and 104 possess an N-terminal transit peptide-like sequence targeting the organelles, whereas PpPPR_63 does not. This suggests that PpPPR_67 and 104 are likely organellelocalized PRORP1 homologs, whereas PpPPR_63 seems to be a nuclear-localized PRORP2/PRORP3 homolog. To confirm this prediction we generated PpPPR_63-green fluorescent protein (GFP) knockin (KI) mosses. The four independent KI lines obtained showed normal growth (Fig. S1), and GFP fluorescence was observed in the nuclei ( Fig. 2A, b and c). GFP fluorescence was observed in the protonemal cells, buds, and young leaves ( Fig. 2A, d-i) but not in mature leaves ( Fig. 2A, j and k).
To verify the organellar localization of PpPPR_67 and 104, their N-terminal sequences fused to GFP (N67-GFP and N104-GFP) were transiently expressed in the protonemata (Fig. 2B). GFP fluorescence of N67-GFP and N104-GFP was observed in both mitochondria and chloroplasts, coinciding with mitochon-dria-localized red fluorescent protein (Mt-RFP) signals and Nomarski images of chloroplasts (DIC).

PpPPR_63, 67, and 104 have RNase P activity
To investigate whether the three Physcomitrella PRORP-like PPR proteins are RNase P enzymes, we performed in vitro RNase P assays with recombinant proteins composed of the respective PRORP-like protein fused with thioredoxin at the N-terminus and six histidines and a VP5 tag at the C-terminus (Fig. 3A, B). The three recombinant proteins, r67, r104, and r63, removed the 59 leader sequences from chloroplast pre-tRNA Phe , mitochondrial pre-tRNA Cys , and nuclear pre-tRNA Asp (Fig. 3C). The 59-end of the tRNA produced was verified by primer extension analysis (Fig. 3D, E). We also confirmed the precise cleavage of pre-tRNAs by circular reverse transcription (cRT)-polymerase chain reaction (PCR) analysis (data not shown). To verify the RNase P activity of these proteins, we also tested two recombinant proteins, rM63 and rC63. rM63 possessed a mutated NYN domain, the two adjacent aspartates of which (D 460 and D 461 ) were altered to alanines. These two aspartates have been shown to be essential for the activity of Arabidopsis PRORP1 [6]. The rC63 was composed of the tag sequences and the NYN domain only. The rM63 and rC63 proteins did not cleave nuclear-encoded pre-tRNA Asp (Fig. 3C).  Table S1. doi:10.1371/journal.pone.0108962.g001 These results indicated that the three moss PPR proteins had RNase P activity similar to that of Arabidopsis PRORPs [6,7,18].

Nuclear PpPPR_63 is dispensable for viability
To investigate the in vivo function of PpPPR_63, we generated two independent knockout (KO) mutants of PpPPR_63, D63-1 andD63-15. Both KO lines were verified to be null mutants. Gene structure of theD63-1 line is given as Figs. 4A and S2, and its absence of PpPPR_63 transcripts was verified by reverse transcription (RT)-PCR (Fig. 4B). Therefore, we used theD63-1 line for further analyses (hereafter named D63). In addition, to generate the complementation lines of theD63-1, we transformed the D63 moss with a transgene encoding the full length of PpPPR_63 or the mutated version and generated F63 and M63, respectively (Fig. S3). In M63, two aspartates (D 460 and D 461 ) in the NYN domain were mutated to alanines (A 460 and A 461 ). In the complementation lines F63 and M63, the introduced transgene was overexpressed (Fig. 4B). The KO mutantD63 and the complementation line M63 mosses displayed smaller protonemal colonies than the wild type (WT) (Fig. 4C), and they showed abnormal regeneration from leaves (especially defective growth of the sub-apical cells) (Fig. 4D). The complementation line F63 exhibited a phenotype the same as that of the WT. These results Knockout of PpPPR_63 slightly affects specific nuclearencoded tRNAs If nuclear PpPPR_63 could function as a bona fide RNase P, knockout of the PpPPR_63 gene could result in severely reduced accumulation of nuclear-encoded tRNAs. To investigate this possibility, we analyzed the accumulation levels of several nuclear-encoded tRNAs in theD63 mutant. For this analysis nuclearencoded tRNA Asp (GUC) and tRNA Gln (CUG) were chosen, because expression of these tRNAs has been demonstrated to significantly decrease in the knockdown mutants of nuclear PRORP2/3 plants [7]. tRNA Asp (GUC) in theD63 mutant and M63 complementation line was slightly reduced to 70% of that in the WT and F63. Whereas, tRNA Gln (CUG) in the D63 mutant, M63 and F63 complementation lines accumulated to the same level as in the WT (Fig. 4E). Similarly, nuclear-encoded tRNA Ser (AGA) and tRNA Lys (CUU) levels were not affected in the mutants. This suggests that PpPPR_63 PNase P activity may be limited to specific tRNA(s) including tRNA Asp (GUC).
Knockout of PpPPR_104 results in a significant reduction in chloroplast tRNA Arg (ACG) To investigate the in vivo function of the PpPPR_67 and 104 genes, we generated and characterized their KO mutants (Figs. 5A, S4, S5, S6). The single gene KO mutants obtained, D67 andD104, were null mutants (Fig. 5B), and the protonemal colony ofD104 moss was smaller than that of the WT (Fig. 5C). A peunumbra and central part of the protonemal colony mostly consists of caulonemal and chloronemal filaments, respectively. In theD104 moss, caulonemal filaments were poorly induced, resulting in a smaller colony. In contrast, PpPPR_67 KO moss (D67) showed little difference from the WT. To generate double KO mutants, we carried out four times transformation experiments, such as introducing the PpPPR_104 KO plasmid into the PpPPR_67 KO line (D67-1-17 line) and vice versa. However, no double KO mutants were obtained from the genotyping of 262 moss plants. This indicates that PpPPR_67 and 104 genes cannot be knocked out simultaneously. Since they may have redundant functions, single KO mutants are viable but double KO is probably lethal.
To investigate whether organellar tRNAs were affected in the PpPPR_67 and 104 KO mutants, we performed northern blot analyses by using all organelle-encoded tRNA probes. The chloroplast tRNA Arg (ACG) level significantly decreased inD104 but not inD67 orD63 (Fig. 5D). All the other tRNAs, including chloroplast tRNA Phe (GAA) and mitochondrial tRNA Cys (GCA) that were used for the in vitro RNase P cleavage assay, were not altered in the WT and the three KO mutants (Figs. 5D, S7, S8).
This result suggests that PpPPR_67 and PpPPR_104 have redundant functions as RNase P in moss organelles.
PpPPR_67 and 104 cleave chloroplast pre-tRNA Arg (ACG) with different efficiency Because chloroplast tRNA Arg (ACG) levels decreased in the D104 but not in theD67, we assumed that PpPPR_104 and PpPPR_67 would have distinct RNase P activity against chloroplast pre-tRNA Arg (ACG). To test this possibility, three chloroplast pre-tRNA Arg isoacceptors were used as substrates for in vitro RNase P cleavage assay. This assay revealed that PpPPR_67 and 104 efficiently cleaved pre-tRNA Arg (CCG) and pre-tRNA Arg (UCU) but they cleaved pre-tRNA Arg (ACG) with different efficiency (Fig. 6). These results suggest that the function of their RNase P is not completely redundant.

Discussion
The first two questions to be addressed here were: When were plant PRORP genes duplicated during plant evolution and where were the respective PRORP proteins localized or targeted to? Our present study provides evidence that PRORP genes were duplicated after the emergence of the early land plants, i.e. mosses. The moss P. patens has three PRORP-like proteins displaying RNase P activity in vitro. Arabidopsis PRORP1 is localized in both the mitochondria and chloroplasts, and PRORP2 and 3 are localized in the nucleus [6]. On the other hand, the moss P. patens has two PRORP1-like proteins, PpPPR_67 and 104, both of which are dual-targeted to the mitochondria and chloroplasts, and one nuclear-localized PRORP2/3-like PpPPR_63. Thus, duplicated PRORP gene products are likely targeted to either the organelles or the nucleus. In Arabidopsis, the double KO mutation of PRORP2 and 3 resulted in embryonic lethality [6]. In contrast, a null mutant of PpPPR_63 displayed smaller protonemal colonies than the WT and also showed abnormal branch formation of filamentous protonemata (Fig. 4). This indicates that unlike Arabidopsis nuclear RPORPs, the moss nuclear PpPPR_63 is dispensable for plant viability. The nuclear-encoded tRNA Asp (GUC) level in the PpPPR_63 KO mutant was slightly decreased (to 70% of that in WT mosses), but most nuclear-encoded tRNA levels were not altered. This result indicates that most of the cytosolic mature tRNAs were produced normally, without proteinaceous RNase Plike PpPPR_63. This does not concur with the results observed from functional analysis of Arabidopsis PRORP2 and 3 [6,7]. In Arabidopsis mutant plants where PRORP3 was absent and the PRORP2 level was decreased, the steady-state levels of five nuclear-encoded tRNAs investigated were significantly reduced, to 30% of the tRNA levels in control plants [7]. These observations indicate that Arabidopsis nuclear PRORP2 and 3 function as master RNase P enzymes, while the moss PpPPR_63 is not the sole RNase P enzyme in the nucleus.
It is unlikely that the slight reduction of the tRNA Asp (GUC) level resulted in growth retardation and abnormal branch formation in the PpPPR_63 KO mutant. PpPPR_63 may have a versatile function rather than pre-tRNA 59-end processing. Arabidopsis PRORP proteins are involved in the maturation of not only tRNAs but also mRNA or snoRNA [7]. Similarly, PpPPR_63 is probably involved in the maturation of unknown mRNA or small RNA-containing tRNA-like structures.

Organelle-localized PpPPR_67 and 104 have functions not completely redundant for tRNA accumulation
In this study, we generated single KO mutants of PpPPR_67 and PpPPR_104 but failed to obtain double KO mutants. Similarly, KO mutation of Arabidopsis PRORP1 was shown to  result in embryonic lethality [6]. This suggests that PpPPR_67 and 104 have redundant functions. However, it is interesting to observe that a single mutant of PpPPR_67 and 104 displayed different phenotypes of protonemal growth and chloroplast tRNA Arg (ACG) accumulation. Chloroplast tRNA Arg (ACG) significantly decreased in the PpPPR_104 KO mutant but not in PpPPR_67 KO mutants. The P. patens chloroplast genome codes for three tRNA Arg isoacceptors [19]. An in vitro RNase P cleavage assay revealed that pre-tRNA Arg (ACG) was more efficiently cleaved by PpPPR_104 than by PpPPR_67, whereas pre-tRNA Arg (UCU) and pre-tRNA Arg (CCG) were cleaved with similar efficiency by both proteins. This is the first evidence that PRORP enzymes have substrate specificity of tRNA processing. The anticodon of P. patens tRNA Arg (ACG) is modified that including inosine in chloroplasts, and tRNA Arg (ICG) can recognize all four codons. Chloroplast tRNA Arg (CCG) recognizing only the GGC codon is dispensable [20]. Processing of chloroplast pre-tRNA Arg (ACG) by PpPPR_104 is essential for the accumulation of mature tRNA Arg (ACG). Therefore, the PpPPR_104 KO mutant may have displayed small protonemal colonies because of a great reduction in mature tRNA Arg (ACG) levels.
Unlike Arabidopsis [6,7], pre-tRNAs were hardly detected in the moss organelles (Figs. S7 and S8). It is conceivable that precursors might be turned over faster than mature tRNAs in mosses.

Was RNP-type RNase P enzyme entirely replaced by PRORP protein in mosses?
The last question is whether ribonucleoprotein RNase P enzyme was replaced by PRORP in both the organelles and the nucleus. In this study, we showed that disruption of PpPPR_63 did not result in a significant reduction in nuclear-encoded tRNAs. This raises the possibility that unidentified RNase P activity may contribute to pre-tRNA maturation in the moss nucleus (Fig. 7). However, it remains unknown whether 59-end processing of nuclear pre-tRNAs is supported by RNP-type RNase P. In P. patens, an RNA component of RNP-type RNase P was not identified, yet but protein components (Pop4, Rpp1, Pop1, and Pop5) are encoded in the genome [21]. Identification of RNase P activity rather than PRORP1-type PpPPR_63 will be a challenge for the future.

Accession numbers
The sequences reported in this paper have been deposited in the DDBJ/GenBank/EMBL databases with accession numbers: PpPPR_63 mRNA, AB983707; PpPPR_67 mRNA, AB983708; PpPPR_104 mRNA, AB808584.

Plant material and preparation of DNA and RNA
The moss P. patens subsp. patens was grown at 25uC under continuous light (30 mmol photon m 22 s 21 ), and genomic DNA and total cellular RNA were prepared from the moss protonemata as described previously [22].

Preparation of recombinant proteins
DNAs corresponding to the proteins (1-595 residues for PpPPR_63, 119-791 for PpPPR_67, and 125-993 for PpPPR104) were amplified by using the primers listed in Table S2 and cloned into pBAD/Thio-TOPO vector (Invitrogen). Proteins were expressed for 6 h at 18uC in E. coli BL21 and purified as described previously [23].

Preparation of pre-tRNA
DNAs representing precursors of chloroplast tRNA Phe , mitochondrial tRNA Cys , and nuclear tRNA Asp were amplified by PCR from P. patens genomic DNA with oligonucleotide pairs T7-cpF5 and cpF3, T7-mtC5 and mtC3, and T7-nuDGTC5 and nuDGTC, respectively (Table S2). The 59 oligonucleotides contained the T7 promoter sequence. The pre-tRNA Phe , pre-tRNA Cys , and nuclear pre-tRNA Asp were respectively 142, 153, and 140 nt long and were designed with respective 59 leaders of 42, 51, and 40 nt and 39 trailers of 27, 32, and 25 nt. Amplified DNAs (200 ng) were transcribed with T7 RNA polymerase (TaKaRa) for 1 h at 37uC, then digested with 5 U of RNasefree DNase I (TaKaRa) and 20 U of RNase Inhibitor (TaKaRa) for 15 min at 37uC. Synthesized pre-tRNAs were purified by phenol chloroform extraction.
For preparation of chloroplast pre-tRNA Arg isoacceptors, the corresponding genes were cloned and in vitro transcribed. The obtained pre-tRNAs were dephosphorylated with Thermo Sensitive Alkaline Phosphatase (Promega) and were 59-end radiolabeled with [c 32 P] ATP and polynucleotidekinase (TaKaRa). Labeled RNAs were purified from the gel after 8% polyacrylamide gel electrophoresis (PAGE) containing 7M urea.

Pre-tRNA cleavage assay
Cleavage assays were performed as described previously [7]. Recombinant protein (100 ng) and in vitro-synthesized pre-tRNA (2 mg) were incubated for 1 h at 25uC in a 20 ml mixture of 20 mM Tris-HCl (pH 8.0), 30 mM KCl, 4.5 mM MgCl 2 , 20 mg/ ml bovine serum albumin, and 2 mM DTT. A half of the reaction mixture was separated on 8% polyacrylamide-7 M urea gels and stained with ethidium bromide.
For cleavage assays of three chloroplast pre-tRNA Arg isoacceptors, the 59 end-radiolabeled pre-tRNAs (500 nM each) and the recombinant proteins (500 nM each) were incubated and aliquots were withdrawn at the indicated times. Dried PAGE gels were analyzed by Storm 820 (GE Healthcare).

Determination of the 59-end of tRNA
The 32 P 59-end-labeled oligonucleotides cp-F3 or mt-C3 (Table  S2) and 200 ng of in vitro-cleaved RNAs were mixed and reverse transcribed with ReverTraAce (ToYoBo) for 1 h at 42uC, then digested with RNase A and precipitated with ethanol. The sequence ladders were obtained by using template DNA, the radiolabled primer cp-F3 or mt-C3, as described previously [22].

Subcellular localization
The DNA region from the 6th intron to the last codon of the PpPPR_63 gene was inserted in-frame into the GFP coding region of pGFPmutNTPII (http://moss.nibb.ac.jp). The 39flanking region of PpPPR_63 was inserted downstream of the nptII cassette of pGFPmutNTPII. The resultant plasmid p63-GFPKI (Fig. S1) was linearized with NotI and introduced into P. patens by particle bombardment as described previously [23]. For N67-GFP and N104-GFP, cDNAs corresponding to an Nterminal part of the proteins (residues 1-112 and 1-105, respectively) were amplified by using the appropriate primers (Table S2) and cloned into pKSPGFP9 [23]. The resultant plasmid was cotransfected with pMt-RFP into P. patens protonemata and GFP and RFP fluorescence were observed as described previously [24].

Generation of KO lines
The 59 region (2960 bp) of PpPPR_63, a 2958-bp region of PpPPR_67, and a 3032-bp region of PpPPR_104 were amplified from moss genomic DNA by PCR using the appropriate primers (Table S2) and cloned into pGEM-T Easy (Promega). The KO plasmid p63KO4-13A (Fig. S2) was generated by insertion of the nptII cassette into the 4th intron of the PpPPR_63 gene by using the GPS-M mutagenesis System (NEB). The KO plasmids p67KO-1 and p67KO-2 (Figs. S4 and S5) were generated by insertion of the nptII cassette into HindIII in the 4th exon, or into StuI in the 5th exon, respectively, of the PpPPR_67 gene. The KO plasmid p104KO (Fig. S6) was generated by insertion of the nptII cassette into the BglII site in the 4th exon of the PpPPR_104 gene. These KO plasmids were digested with NotI and introduced into P. patens by particle bombardment.

Generation of complementation lines
For F63, the PpPPR_63 cDNA coding region (1785 bp) amplified by PCR using specific primers (Table S2) was cloned into the SwaI site of the overexpression vector pOX9WZ1 [25]. The mutated PpPPR_63 (M63: D 460 A/D 461 A) was amplified from the F63 construct by using the primers 63DA-F and 63DA-R. Both constructs were linearized with NotI and introduced into the KO line D63.