Copyright � 2008 by the Genetics Society of AmericaDOI: 10.1534/genetics.107.085688

On the Evolution and Expression of Chlamydomonas reinhardtiiNucleus-Encoded Transfer RNA Genes

Valerie Cognat,* Jean-Marc Deragon,† Elizaveta Vinogradova,*,1 Thalia Salinas,*Claire Remacle‡ and Laurence Marechal-Drouard*,2

*Institut de Biologie Moleculaire des Plantes, Unite Propre de Recherche 2357, Conventionnee avec l’Universite Louis Pasteur(Strasbourg 1), Centre National de la Recherche Scientifique, 67084 Strasbourg Cedex, France, †Universite de Perpignan

Via Domitia, CNRS UMR5096 Laboratoire Genome et Developpement des Plantes, 66860 Perpignan Cedex,France and ‡Genetique des Microorganismes, Departement des Sciences de la Vie, Institut de

Botanique, Universite de Liege, B-4000 Liege, Belgium

Manuscript received December 11, 2007Accepted for publication March 18, 2008

ABSTRACT

In Chlamydomonas reinhardtii, 259 tRNA genes were identified and classified into 49 tRNA isoacceptingfamilies. By constructing phylogenetic trees, we determined the evolutionary history for each tRNA genefamily. The majority of the tRNA sequences are more closely related to their plant counterparts than toanimals ones. Northern experiments also permitted us to show that at least one member of each tRNAisoacceptor family is transcribed and correctly processed in vivo. A short stretch of T residues known to be asignal for termination of polymerase III transcription was found downstream of most tRNA genes. It allowedus to propose that the vast majority of the tRNA genes are expressed and to confirm that numerous tRNAgenes separated by short spacers are indeed cotranscribed. Interestingly, in silico analyses and hybridizationexperiments show that the cellular tRNA abundance is correlated with the number of tRNA genes and isadjusted to the codon usage to optimize translation efficiency. Finally, we studied the origin of SINEs, shortinterspersed elements related to tRNAs, whose presence in Chlamydomonas is exceptional. Phylogeneticanalysis strongly suggests that tRNAAsp-related SINEs originate from a prokaryotic-type tRNA either hori-zontally transferred from a bacterium or originally present in mitochondria or chloroplasts.

A survey of the Chlamydomonas reinhardtii sequenceproject has resulted in the identification of 259

tRNA genes, allowing the decoding of every sensecodon using the classical codon–anticodon wobble rules(Merchant et al. 2007). The C. reinhardtii genome is notfully sequenced but as stated by the Joint GenomeInstitute, the depth of sequence coverage for the maingenome scaffolds is estimated at�12.83. Thus, the 259tRNA genes are expected to represent nearly all the totaltRNA gene complement. The presence on the Chlamy-domonas nuclear genome of both key plant and animalprotein genes is one of the most exciting conclusionsdrawn from the sequencing and annotation project.However, the origin of noncoding RNAs such as tRNAshas not been studied yet. Several Northern experimentsperformed on a whole-cell tRNA fraction from Chlamy-domonas and using as probes oligonucleotides specificto different higher plant tRNAs failed to give hybridiza-tion signals (C. Remacle and L. Marechal-Drouard,unpublished data), indicating that at least some Chlamy-

domonas tRNA sequences are different from theirhigher-plant counterparts. Furthermore and in contrastto higher plants, a selenocysteine (Sec) protein insertionmachinery was found in C. reinhardtii. A selenocysteinetRNASec was found expressed in this green alga (Rao et al.2003) and the corresponding single gene was identifiedin frame of the annotation project (Merchant et al.2007). Thus, this tRNASec represents the first nonanimaleukaryotic tRNASec. To determine whether the Chlamy-domonas tRNA genes are evolutionarily closer to plantsor to animals, we traced the evolutionary origin of mostof Chlamydomonas tRNA genes by constructing phylo-genetic trees. We then provided several in silico andexperimental lines of evidence that most of these tRNAgenes are expressed from mono- or polycistronic operonsand correctly processed in vivo. As a consequence ofthe numerous gene duplications, the isoaccepting tRNAgene copy number varies from 1 to 17 copies. Further-more, there is a codon bias in C. reinhardtii, with somecodons highly used in highly expressed genes. In severalorganisms such as Escherichia coli, yeast, and Caenorhabditiselegans or in chloroplasts, an adjustment of tRNA popu-lationtothecodonusageexists (Bulmer1987;Pfitzinger

etal.1987;Duret2000).Wethereforeasked thequestionwhether a coadaptation between tRNA gene number,tRNA abundance, and codon usage occurs in this green

1Present address: Faculty of Biology, Lomonosov Moscow State University,Leninskie Gory, Moscow 119992, Russia.

2Corresponding author: Institut de Biologie Moleculaire des Plantes duCNRS, UPR2357 du CNRS, 12 rue du General Zimmer, 67084 StrasbourgCedex, France. E-mail: laurence.drouard@ibmp-ulp.u-strasbg.fr

Genetics 179: 113–123 (May 2008)

alga. In silico analysis as well as hybridization experimentsshowed that also in Chlamydomonas, there is likely agene dosage effect so as to adapt the tRNA populationto an optimal cytosolic translation. Finally, tRNAs are atthe origin of tRNA-related short interspersed elements(SINEs). SINE retroelements are widespread amongeukaryotic organisms and have an important impacton the structure, evolution, and expression of the hostgenome. Remarkably for a unicellular microorganism,six SINE families were identified in Chlamydomonas(Jurka et al. 2005; Merchant et al. 2007). Becauseretroelements are one of the major sources of genomebiodiversity, understanding their evolution is important.To get more insight on the origin of ChlamydomonastRNA-related SINEs, phylogenies were constructed. Thedata obtained permitted us to propose that a prokaryotictRNA is at the origin of tRNAAsp-related SINE families.

MATERIALS AND METHODS

Computational methods: All available nucleic acid se-quence data for C. reinhardtii were used. The majority of thenuclear sequence data were produced by the U.S. Departmentof Energy Joint Genome Institute (http://www.jgi.doe.gov/).All tRNA genes discussed here may be obtained at http://genome.jgi-psf.org/Chlre3/Chlre3.home.html. In this study,the most representative sequence for each isoaccepting tRNAgene family was chosen. This information was used to designoligonucleotide probes specific for the corresponding tRNAs(see supplemental Table 2). The cytosolic codon usage of C.reinhardtii has been obtained through the website http://www.kazusa.or.jp/codon/. For phylogenetic trees, the tRNAsequences from a set of organisms (Figure 1) were downloadedfrom the Genomic tRNA Database (http://lowelab.ucsc.edu/GtRNAdb/). Two tRNA sets were added: Cyanidioschyzon merolaeand Dictyostelium discoideum; they were downloaded and ex-tracted from their own websites (http://merolae.biol.s.u-tokyo.ac.jp/ and http://dictybase.org/, respectively). For each aminoacid type, duplicated sequences were removed and alignmentsof tRNA sequences were done using the TFAM package(Ardell and Andersson 2006). Only the COVEMF programwas used: this program aligns a set of sequences to an existingalignment and takes into account both primary and secondarytRNA structures. All sites in the alignments that contained toomany variations or large numbers of gaps were excluded,especially in the variable loop because they could not be reliablyaligned. Phylogenetic trees were produced using the PHASE2.0package (Hudelot et al. 2003) and the Markov chain MonteCarlo (MCMC) method. This package is designed for rRNA andtRNA sequences and their conserved secondary structure. Asrecommended by the authors we used a gamma distributionwith four rate categories and we executed three different runswith variable random seeds to check reproducibility and obtainsimilar consensus trees. The branch lengths were calculatedon the consensus tree topology with the maximum-likelihoodmethod. The trees were drawn using the Treedyn package(http://www.treedyn.org/).

Strain and growth conditions: The strain of C. reinhardtiiused in this work is the cell-wall-less mutant cw15 mt1. Cells wereroutinely grown under mixotrophic conditions, i.e., under light(4000 lux) on Tris-acetate phosphate (TAP) medium (Harris

1989).Total tRNA isolation: Total tRNA extract was prepared from

whole cells as described in Marechal-Drouard et al. (1995).

Northern analysis: Total tRNAs from C. reinhardtii wereseparated on a 15% denaturing polyacrylamide gel and electro-transferred onto Hybond-N membrane (Amersham, ArlingtonHeights, IL). Oligonucleotides (see supplemental Table 2) were59-end labeled using T4 polynucleotide kinase and ½g-32P�ATPunder standard conditions. Hybridization and washing wereperformed as described in Delage et al. (2003).

RNA labeling and hybridization to oligonucleotides: Iso-lated tRNAs were 39-end labeled using 59½32P�pCp and RNA ligase(Marechal-Drouard et al. 1990). Oligonucleotides separatedon a 15% denaturing polyacrylamide gel and electrotransferredonto Hybond-N membrane (Amersham) were hybridized against39-end-labeled tRNAs, using conditions described above forNorthern analysis.

Cloning and sequencing of RT–PCR products: TransferRNAs purified on a 15% denaturing polyacrylamide gel(Marechal-Drouard et al. 1995) were used as a substratefor RT–PCR amplification. Gel-isolated RT–PCR productswere cloned into PCR-2-TOPO (Invitrogen, San Diego).

RESULTS AND DISCUSSION

Phylogenetic analysis of the 259 nuclear tRNA genes:Taking into account 2 families for tRNAAla(CGC) and2 families for tRNAAla(UGC) (see below), 49 isoacceptorfamilies were identified in Chlamydomonas, allowingthe decoding of every sense codon using the classicalcodon–anticodon wobble hypothesis (Table 1). To deter-mine the evolutionary history for each of these families,phylogenetic trees were constructed. Different situationswere observed. Figure 1A shows that tRNACys(GCA) isclosely related to Arabidopsis thaliana tRNACys(GCA). Thesituation is the same for tRNAAsn, tRNAGln(CTG), tRNAGlu,tRNAHis, tRNALys, elongator tRNAMete, tRNAPhe, tRNAPro,and tRNAGly ½except for tRNAGly(CCC)� (supplementalFigure 6). All these families appear to be more closelyrelated to the plant tRNA genes than to the animal ones.In contrast, other tRNA genes are more evolutionarilyrelated to animal ones. For example, ChlamydomonastRNATyr isoacceptors clearly branch among animal andfungus tRNA genes (Figure 1B). For two amino acids, Ileand Leu depending on the isoacceptor tRNA considered,tRNA gene sequences are more related to either the plantsequences ½tRNAIle(TAT), tRNALeu(TAA, CAA)� or theanimal ones ½tRNAIle(AAT), tRNALeu(CAG, AAG, TAG)�(Figure 1C and supplemental Figure 6). The single-copytRNAIle(GAT) gene (not shown in Table 1) was identifiedby blast as a plastidial tRNAIle gene. It is part of a 900-ntchloroplastic fragment that was found to be inserted intothe nuclear genome (Merchant et al. 2007). For tRNAArg

and tRNASer, a few tRNA gene sequences are linked tothe plant sequences ½tRNAArg(ACG), tRNASer(GCT)�, andsome others are more related to animal tRNA species ½e.g.,tRNAArg(CCG, TCG)�, although in this latter case theBayesianposteriorprobabilitiesareweakandcannotunam-biguously ascertain this conclusion (supplemental Figure6). The tRNATrp sequences are close to the C. elegans andArabidopsis sequences. Initiator tRNAMeti andtRNAGln(TTG)has slightly diverged from the other eukaryotic sequen-ces (supplemental Figure 6). No reliable phylogenetic

114 V. Cognat et al.

TABLE 1

Nuclear tRNA genes and cytosolic codon usage in Chlamydomonas reinhardtii

Amino acids Anticodonsa Codonsb Gene nos.a % cytoc Probesd

Ala AGC GCU–GCC 13 7.24 AlCGC GCG 6 1 4 4.29 A1–A2UGC GCA 3 1 1 1.06 A1–A2

Arg CCG CGG 3 1.1 RiACG CGU–CGC 11 4.04 R2CCU AGG 2 0.27 R3UCG CGA 1 0.19 R4UCU AGA 3 0.07 R5

Asn GGU AAU–AAC 7 3.09 NAsp GUC GAU–GAC 11 4.83 DCys GCA UGU–UGC 7 1.39 CGln UUG CAA 1 0.43 Qi

CUG CAG 6 3.6 Q2Glu UUC GAA 1 0.28 El

CUC GAG 13 5.41 E2Gly GCC GGU–GGC 17 7.27 Gi

UCC GGA 1 0.5 G2CCC GGG 1 0.94 G3

His GUG CAC 5 1.9 HIle AAU AUU–AUC 7 3.49 Ii

UAU AUA 1 0.1 12Leu CAG CUG 10 6.48 Li

CAA UUG 2 0.39 L2UAA UUA 1 0.06 L3UAG CUA 1 0.26 L4AAG CUU–CUC 3 1.73 L5

Lys CUU AAG ii 4.42 KiUUU AAA 1 0.24 K2

Mete CAU AUG i 8 0.21 MiCAU AUG e 6 2.38 Me

Phe GAA UUU–UUC 9 3.24 FSeC UCA UGA 1 ND SeCPro AGG CCU–CCC 13 3.65 P

CGG CCG 6 2.02 PUGG CCA 1 0.52 P

Ser AGA UCU–UCC 5 2.09 SiUGA UCA 1 0.32 S2CGA UCG 5 1.63 S3GCU AGC–AGU 8 2.52 S4

Thr AGU ACU–ACC 6 3.35 TiCGU ACG 3 1.55 T2UGU ACA 2 0.41 T3

Trp CCA UGG 5 1.31 WTyr GAC UAU–UAC 8 2.56 YVal AAC GUU–GUC 7 2.1 Vi

CAC GUG 10 4.61 V2UAC GUA 1 0.21 V3

a Each tRNA isoacceptor family is identified by its anticodon and the number of corresponding genes is givenaccording to Merchant et al. (2007). For tRNAAlaCGC and tRNAAlaUGC, the two gene numbers refer to the twotypes of sequences (groups I and II, see Figure 2) identified for each isoacceptor.

b Minimal potential codon recognition pattern of tRNA isoacceptors listed according to the anticodons in theprevious column.

c Cytosolic (% cyto) codon usage as obtained from http://www.kazusa.or.jp/codon/.d Names of probes are according to supplemental Table 2. The plastidial tRNAIleGAT gene inserted into the

nuclear genome is not indicated.e For Met codons AUG i and AUG e, i and e stand for initiator and elongator tRNAMet, respectively.

Nucleus-Encoded tRNAs in Chlamydomonas 115

Figure 1.—Phylogenetic analyses of (A) cysteine, (B) tyrosine, and (C) isoleucine tRNA genes show their evolutionary linkeither to plant or to animal counterparts. The trees are representative of three different situations: Chamydomonas tRNACys

branches with Arabidopsis tRNACys sequences, tRNATyr branches with fungi and human tRNATyr sequences, and tRNAIle with eitherArabidopsis or human tRNAIle, depending on the tRNA isoacceptor considered. The phylogenetic trees constructed for the otheramino acids are presented in supplemental Figure 6. Construction of the trees is described in detail in materials and methods.Phyla are differentiated by colors: bacteria (red), archaebacteria (pink), eukaryotes (blue), algae (green), and protozoans (black).Species used for the phylogenetic analysis are as follows: Aeropyrum pernix (Ape), Archaeoglobus fulgidus (Afu), Methanococcus janna-schii (Mja), Methanopyrus kandleri (Mka), Methanosarcina mazei (Mma), Methanobacterium thermoautotrophicum (Mth), Pyrococcus abyssi(Pab), Pyrobaculum aerophilum (Paer), Sulfolobus solfataricus (Sso), and Thermoplasma acidophilum (Tac) for Archae; Agrobacterium tu-mefaciens C58 Cereon (Atu), Bacillus subtilis (Bsu), Bordetella parapertussis (Bpa), Brucella melitensis (Bme), Campylobacter jejuni (Cje),Caulobacter crescentus (Ccr), Chlamydophyla pneumoniae AR39 (Cpn), Clostridium perfringens (Cpe), Escherichia coliK12 (Eco), Haemophi-lus influenzae (Hin), Helicobacter pylori 26695 (Hpy), Listeria monocytogenes (Lmo), Mycoplasma pneumoniae (Mpn), Mycobacterium

116 V. Cognat et al.

tree was constructed from the alignment of tRNAThr

sequences. Finally, the case of tRNAAla constitutes aninteresting situation. Two groups of tRNAAla sequenceswere retrieved from the genome sequence. The firstgroup (group I) contains 13, 6, and 1 tRNA isoacceptorswith AGC, CGC, and UGC anticodons, respectively;the second group (group II) contains 4 and 3 tRNAisoacceptors with CGC and UGC anticodons, respec-tively. Whereas within each group, mature tRNA sequen-ces are either identical or differ only by point mutations(in particular at the level of the anticodons), the se-quences between the two groups of tRNAAla genes differsignificantly (Figure 2A). Phylogenetic analysis showsthat they form two distinct branches and thus haveindependent evolutionary history (Figure 2B). In thefirst group, tRNAs are related to higher plant and animaltRNAAla genes whereas tRNAs from the second groupform a separate branch on the tree and are more closelyrelated to Saccharomyces cerevisiae and Schizosaccharomycespombe tRNAAla genes. It is also worth noting that all groupI tRNAAla genes have an intron located at position 37/38;by contrast, only 1 of the 7 group II tRNAAla genes hasacquired an intron at the same position. Alanine tRNAswith an AGC anticodon are present only in group I buttRNAs with the two other anticodons, CGC and UGC,are found in both groups and thus there is a priori aredundancy of function, leading to the following ques-tion: Are both types of tRNAAla genes expressed inChlamydomonas? To address this question, the expres-sion of all tRNA isoacceptor families was analyzed.

Expression of the tRNA isoacceptor families: North-ern blot experiments performed using as probes oligo-nucleotides specific to each tRNA species (supplementalTable 2) permitted us to obtain an overview of the ex-pression of all nuclear tRNA gene families from C.reinhardtii (Figure 3A), except tRNAAla and tRNAPro. Forthese two tRNA isoacceptor families, RT–PCR experi-ments were additionally performed since sequences mostlydiffer only at the level of the anticodon (supplementalTable 2). RT–PCR experiments on total C. reinhardtiitRNA fraction followed by cloning and sequence analysisshow that tRNAAla of group I (with AGC, UGC, or CGCanticodons) and group II (with UGC or CGC antico-dons) as well as tRNAPro (with AGG or CGG anticodons)are indeed expressed and correctly processed (Figure3B). In particular, for all tRNAs, splicing of introns waseffective. It is worth noting that .20 years ago, Tyc et al.(1983) reported that a Chlamydomonas tRNA prepara-

tion contains small amounts of tRNA 59 halves andcorresponding 39 halves. They postulated that Chlamy-domonas tRNAAla and tRNAPro genes could have intronsand that the tRNA halves they detected may be inter-mediates of the tRNA splicing process (Tyc et al. 1983).As mentioned above, part of the tRNAAla and tRNAPro

gene population indeed contains introns. However, theposition of the splicing site (between A37:U38 for tRNAAla

and between G37:U38 for tRNAPro) does not correlatewith the position of the sites of tRNA half-molecule li-gation (between U34:G35 for tRNAAla and between G35:G36 for tRNAPro) found by these authors. The partialRNA sequences they obtained are close to the tRNA genesequences now available. This discrepancy suggests, con-trary to their hypothesis, that the tRNA halves detected inChlamydomonas do not correspond to true intermedi-ates of tRNA splicing and that the RNA kinase and ligaseactivities they observed may rather be involved in the tRNArepair. Concerning tRNAAla(AGC) and tRNAPro(AGG), allsequences obtained presented a GGC or a GGG antico-don. This means that the A at the wobble position of theanticodon is very likely post-transcriptionaly converted toinosine (I) since the reverse transcriptase recognizes I asa G. The presence of I at position 34 has already beenreported for several plant tRNAs (Glover et al. 2001). NotRNAPro with a UGG anticodon was found out of 50 se-quences. There is only a single-copy gene coding for atRNAPro(UGG) as compared to 13 and 6 copies for tRNAPro

with AGG and CGG anticodons, respectively. Eitherwe did not sequence enough clones or this gene is nottranscribed. As a whole, Northern analysis indicatesthat at least one copy of each tRNA family but one½tRNAPro(UGG)� is expressed. However, it is more difficultto know whether all currently annotated Chlamydomo-nas tRNA genes are transcribed. In an attempt to answerthis question, we used another mark of the expression ofeukaryotic tRNA genes. A specific feature of virtually allnewly synthesized eukaryotic RNA polymerase III tran-scripts is the presence of a short poly(U) motif at their 39

ends. We therefore looked for the presence of a shortpoly(T) stretch a few nucleotides downstream of Chla-mydomonas tRNA genes. For 66% of them, a shortstretch of four to five T’s was found 1–10 nucleotidesdownstream of the predicted 39 end of each tRNA gene(see, for example, supplemental Figure 7, tRNAsVal nos. 1,4, 6, 7, 10, and 12). In particular, a short stretch of four T’sis present 2 nucleotides downstream of the 39 end of thesingle tRNAPro(TGG) gene, thus indicating that this gene

tuberculosis CDC1551 (Mtu), Pseudomonas aeruginosa (Paea), Rickettsia prowazekii (Rpr), Salmonella enterica typhi (Sen), and SynechocystisPCC6 (Syn) for bacteria; and Arabidopsis thaliana (Ath), Caenorhabditis elegans (Cel ), Chlamydomonas reinhardtii (Cre), Cyanidioschyzonmerolae (Cme), Dictyostelium discoideum (Ddi), Drosophila melanogaster (Dme), Homo sapiens (Hsa), Saccharomyces cerevisiae (Sce), andSchizosaccharomyces pombe (Spo). The nomenclature used is illustrated with one example as follows: CreYcGTA is chloroplastic tRNATyr

with a GTA anticodon from Chlamydomonas reinhardtii. When a mitochondrial tRNA is on the tree, the letter m follows the one-lettercode amino acid. When a number is after the anticodon, it corresponds to one of the tRNA isoacceptors of the same family differingby point mutations. Branch lengths are the maximum-likelihood values for the consensus tree topology. The values along eachbranch are the node Bayesian posterior probability (only values . 50 are indicated).

Nucleus-Encoded tRNAs in Chlamydomonas 117

is likely transcribed by polymerase III. Surprisingly, forone-third of the tRNA gene sequences, the poly(T)stretch is absent. In Chlamydomonas, �160 tRNA genesare clustered on the genome. They are associated on thesame or opposite DNA strands and can be separatedby spacers as short as 3–7 nt (Merchant et al. 2007).This unusual genomic organization in Chlamydomonasleads to the hypothesis that a few tRNA genes could becotranscribed. As shown in Figure 3C for a few scaffolds, apoly(T) motif is mostly present only at the most down-stream tRNA gene of each cluster (see also supplementalFigure 7). When tRNA genes are on opposite strands, thena stretch of T’s is found in both directions. Thus, thesedata strongly support that polycistronic precursor tRNAs

do exist in Chlamydomonas and explain why a poly(T)motif is not always present downstream of each tRNAgene. Polycistronic tRNA gene units are not the rule ineukaryotic systems and in yeast, animals, and plants, eachtRNA gene is usually independently transcribed. Suchunusual genomic organization has been reported onlyin a few cases, in particular in the protozoan Trypanosomabrucei where dicistronic tRNA precursors are found(LeBlanc et al. 1999; Mottram et al. 1991a,b). The pres-ence of a short poly(T) motif at the 39 end of monocis-tronic and polycistronic tRNA gene units strongly suggeststhat most of the predicted tRNA genes are expressed inChlamydomonas. However, for a few tRNAMet and tRNALys

genes, no poly(T) motif was observed, although they are

Figure 2.—Two phylogeneti-cally distant tRNAAla gene groupscoexist in the nuclear genome ofChlamydomonas. (A) Predictedcloverleaf structures of the tRNAAla

isoacceptors of the two groups (Iand II) as deduced from the nu-cleus-encoded tRNAAla sequences.Only one representative structurefor each family is represented. Nu-cleotide differences among isoac-ceptors are indicated by arrows.Uppercase letters indicate thethree (AGC, UGC, and CGC) orthe two anticodons (UGC andCGC) found in groups I and II, re-spectively. (B) Phylogenetic treeanalysis of alanine tRNAs. The treeis as described in the Figure 1legend.

118 V. Cognat et al.

apparently not clustered with other tRNA genes. The moststriking exception is the case of tRNAPhe genes. Ninecopies of the tRNAPhe(GAA) gene were identified on threeindependent scaffolds (4, 61, and 85) in the Chlamydo-monas genome. None of them is close to other tRNAgenes and none of them ends with a stretch of Tresidues.As a hybridization signal was obtained by a Northernexperiment (Figure 3A) and as no other tRNAPhe geneshave been predicted, at least some of these 9 copies mustbe expressed so as to decode the TTT and TTC codons.Kruszka et al. (2003) showed that plant tRNA genes canbe cotranscribed with snoRNA genes, but neither we norOlivier Vallon (Institut de Biologie Physico-Chimique, Paris;O. Vallon, personal communication) found snoRNAs inthe vicinity of any tRNAPhe genes. Strikingly, however, for8 of the 9 tRNAPhe genes, a poly(T) stretch can be found

�75 nt downstream of the 39 end of each gene (see sup-plemental Figure 8). It is thus likely that for the tRNAPhe

gene family, these poly(T) stretches are important fortranscription termination and for the stabilization oftranscripts. Why the poly(T) stretches are not linked tothe 39 end of the tRNAPhe genes will need to be clarified inthe future. It must also be noted that 6 of 9 of the tRNAPhe

genes are found in introns of protein-coding genes (datanot shown). Thus we cannot exclude the possibility thatsome tRNAs are generated during intron splicing, aprocess known to be essential for the biosynthesis of thesnoRNA-coding region (e.g., Vincenti et al. 2007). Thesedata hold true for all the tRNA genes where no poly(T)stretches were found in the close vicinity of their 39 ends.Thus, to conclude, most nuclear tRNA genes are likelyfunctional in Chlamydomonas.

Figure 3.—Most Chlamydomonas nuclear tRNA genes are likely expressed. (A) Northern experiments on whole-cell RNA frac-tions with probes specific to cytosolic tRNA genes (see supplemental Table 2). The name of the probe is given below each North-ern experiment. (B) RT–PCR experiments on whole-cell RNA fractions with pairs of primers specific to tRNAPro (P), tRNAAla1 (A1),and tRNAAla2 (A2) (see supplemental Table 2). (C) The clustered tRNA genes present on scaffolds 20, 35, 68, and 76 are depictedas arrows that indicate the orientation on the chromosome. For scaffold 20, the tRNAVal genes are numbered in the arrows from 1to 12. The anticodon is shown under each gene. The spacing in base pairs between the tRNA genes is indicated. Poly(T) stretchesare depicted as solid rectangles. (D) Schematic of the three putative La proteins (Cr-La1, Cr-La2, and Cr-La3; also called LAL1,LAL2, and LAL3 in http://genome.jgi-psf.org//Chlre3/Chlre3.info.html) identified in the nuclear genome of Chlamydomonas.They have the following accession numbers: Chlre2_kg.scaffold_32000083, OVA_estExt_fgenesh2_pg.C_50237, and OVA_fgenesh1_pg.C_scaffold_10000098. La domains are shaded boxes and RRM1 and RRM2 are solid boxes. Numbers indicatethe lengths of the proteins in amino acids.

Nucleus-Encoded tRNAs in Chlamydomonas 119

The Chlamydomonas La protein: In a wide range ofeukaryotes the poly(U) motif is specifically recognizedby an RNA-binding protein called the La protein(Wolin and Cedervall 2002). One of the majorfunctions of the La protein is to stabilize newly synthe-sized small RNAs. For example, the binding of the Laprotein to primary tRNA transcripts prevents exonu-cleolytic nibbling of their 39 trailer and promotes theirendonucleolytic removal. Blast and advanced searchesin the Chlamydomonas genome using the recently char-acterized A. thaliana La protein (Fleurdepine et al. 2007)revealed the presence of three La or La-like proteins.They were subsequently called Cr-La1, Cr-La2, andCr-La3 (Figure 3D). The N-terminal domain (NTD) oftrue La proteins is extremely well conserved. It alwayscontains the eukaryotic La motif of�60–80 amino acids(PF05383; http://pfam.sanger.ac.uk) at the very begin-ning of the NTD and closely followed by a typical RNArecognition motif (RRM) (PF00076, http://pfam.sanger.ac.uk/). On the basis of sequence alignments of variousLa proteins and the Pfam database, an La motif wasfound on the three Cr-La proteins. However in Cr-La3,the La domain is not located in the NTD but rather inthe middle of the protein. The La domain is followedby the typical RRM in Cr-La1. No RRM is retrieved onCr-La2 using Pfam but such a motif was identified bysequence alignments, suggesting that it has slightlydiverged from the consensus RRM sequence. No RRMwas found on Cr-La3. The C-terminal domain (CTD) ofLa proteins is more variable. A second atypical RRM(PF08777) is found in La proteins from all vertebratesand higher plants but is absent in yeast. Here, on thebasis of sequence alignments and secondary structureprediction, this second RRM was identified in Cr-La1but not in Cr-La2 and Cr-La3. It is worth mentioningthat Cr-La3 contains DM15 domains in the CTD, a motifof unknown function already found in La-like proteinsof various organisms, and Cr-La1 contains a nuclear lo-

calization signal as the genuine La proteins (Fleurdepine

et al. 2007). Altogether, our in silico data provide evi-dence that Cr-La1 is the bona fide ChlamydomonasLa protein whereas Cr-La2 and Cr-La3 are more relatedto La-like proteins (also called LARP) (Wolin andCedervall 2002). In Arabidopsis, two ‘‘true’’ La pro-teins are expressed, although so far only one was shownto fulfill the genuine La function (Fleurdepine et al.2007). It is worth noting that in contrast to higher plantsand as in vertebrates and fungi, the nuclear genome ofChlamydomonas encodes only one true La homolog.

The copy number of nuclear tRNA genes is cor-related to the codon usage of the nuclear genome inC. reinhardtii: The 259 nuclear-encoded tRNA genes aregrouped into 49 isoacceptor families (Table 1). Asmentioned above, tRNA gene clusters and recent tRNAgene duplications are present on the Chlamydomonasgenome. Consequently the isoaccepting tRNA genecopy number varies from 1 copy to 17 copies (Table 1).Codon usage is strongly biased in the nuclear genome ofC. reinhardtii; some codons are mostly not used whereasothers are very frequent (Table 1). Interestingly, thecomparison between codon usage and gene copynumber reveals that all the favored codons are decodedby the isoaccepting tRNAs that have the highest genecopy numbers. There is only one exception to this rule,the case of initiator AUG methionine codons where thenumber of genes is far higher than expected fromthe abundance of the initiation codon. It is likely thatthe initiator tRNAMeti has to be abundant to not be alimiting factor in the initiation of translation, as alreadysuggested in chloroplasts (Pfitzinger et al. 1987). Thismechanism of coadaptation between tRNA gene num-ber and codon usage seems to be frequent and hasalready been reported in different organisms such asE. coli and C. elegans (Bulmer 1987; Duret 2000).Furthermore, hybridization of pCp-labeled total tRNAsfrom C. reinhardtii with oligonucleotides specific to dif-ferent tRNA isoacceptors permitted us to show that indeedfor the studied tRNAs (Figure 4), the most abundanttRNA isoacceptors (e.g., tRNALeuCAG, tRNAValCAC,tRNAGlyGCC, tRNALysCUU, and tRNAGlnCUG) recognizethe most frequent codons whereas scarce tRNA iso-acceptors (tRNALeuUAG, tRNAValAAC, tRNAGlyUCC,tRNALysUUU, and tRNAGlnUUG) recognize rare codons.In various organisms and chloroplasts, the abundanceof specific tRNAs was shown to be directly correlated tothe frequency of the cognate codons (Pfitzinger et al.1987; Dong et al. 1996). Here, in Chlamydomonas, inrelation to the number of the corresponding genes, itseems that there is also a good adjustment of the tRNApopulation to the cytosolic codon usage.

tRNA-related SINEs: SINE retroelements are found ina wide variety of eukaryotes (Kramerov and Vassetzky

2005). These short (,500 bp) nonautonomous ele-ments are transcribed by RNA polymerase III and relyon long interspersed elements (LINEs) for their prop-

Figure 4.—The tRNA population is adjusted to the cyto-solic codon usage. (A) Histogram showing the frequency ofa few codons as indicated in Table 1. (B) Comparison ofthe steady-state level of tRNAs hybridized with 1 mg of specificoligonucleotides separated on a polyacrylamide gel and trans-ferred onto Hybond-N membrane. The tRNAs correspondingto the analyzed oligonucleotides are in Table 1, and they areable to decode the codons presented in the histogram.

120 V. Cognat et al.

agation. Their copy number ranges from a few hundredto more than a million. Most eukaryotic SINEs areancestrally derived from tRNA genes (and in rare casesfrom 7SL or 5S RNAs), although the typical tRNAcloverleaf structure is not apparent anymore for mostSINE RNAs (Sun et al. 2007). SINEs derived from tRNAsusually have a composite structure made of a 59 tRNA-related portion followed by a tRNA-unrelated portion(Okada and Ohshima 1995; Deragon and Zhang

2006). In all cases, an internal promoter (composed ofA and B boxes and recognized by the RNA polymeraseIII machinery (Arnaud et al. 2001) is present in theSINE tRNA-related portion.

With .200 copies of tRNA-related SINEs ½belonging tosix families named SINEX-1 to -6_CR, see supplementalAppendix 1 for consensus sequences and Repbase (http://www.girinst.org/repbase/index.html)�, C. reinhardtii isthe first unicellular organism where such elements havebeen identified (Jurka et al. 2005; Merchant et al.2007). SINEX-3_CR and SINEX-4_CR consensus se-quences have no significant homologies with otherC. reinhardtii SINE consensus sequences and thereforetruly represent founder elements of two distinct fami-lies. In contrast, SINEX-1_CR and SINEX-2_CR consen-sus sequences have highly similar tRNA-related portionsbut divergent tRNA-unrelated portions (see supplemen-tal Figures 9 and 10), suggesting that founder copiesof these two families evolved from a common ancestorfollowing the acquisition of a new tRNA-unrelatedregion (Deragon and Zhang 2006). SINEX-5_CR andSINEX-6_CR consensus sequences are also homolo-gous, suggesting a common origin but neverthelesspresent patches of divergence (at the beginning of thetRNA-related and -unrelated portions; see supplementalFigures 9 and 10), likely resulting from gene conversionevents (Kass et al. 1995; Lenoir et al. 1997). Still SINEX-5and SINEX-6 were defined as two different families sincetheir consensuses share 75% sequence identity and aretherefore under the 80% threshold usually used to in-clude sequences in a given family (Wicker et al. 2007).

Since the SINE/LINE partnership is, in some cases,based on common 39-regions, we looked for LINEs in C.reinhardtii that would share primary sequence homolo-gies in their 39 ends with SINEs. A putative SINE/LINEpartnership between L1-1_CR and SINEX-3_CR hasalready been described in Repbase (http://www.girinst.org/2004/vol4/issue2/SINEX-3_CR.html; see supple-mental Figure 11A for its description). We observed thatall other SINE families also possess primary sequencehomologies in their 39 ends with at least one LINEconsensus sequence from the RandI family (see supple-mental Figure 11B). This suggests that most SINE/LINE partnerships in C. reinhardtii are based on primarysequence homologies and are not of the ‘‘relaxed’’ type(Okada et al. 1997).

Remarkably, the SINE tRNA-related portions fromSINEX-3_CR, SINEX-5_CR, and SINEX-6_CR families

have apparently kept an authentic tRNA structure aswell as an intron between positions 37 and 38 of thetRNA sequences (Figure 5A). The 59 tRNA-relatedsequences of SINEX-3_CR, SINEX-5_CR, and SINEX-6_CR copies resemble a tRNAArg(CCG) (a few of themshow point mutations in the anticodon leading to ACG,CCA, CCC, and CTG anticodons), a tRNAAsp(AUC), anda tRNAAsp(GUC), respectively. These families contain40, 11, and 18 members, respectively. The tRNA-relatedregion of 36 SINEX-3_CR copies ends with a genome-encoded CCA, a motif that is post-transcriptionalyadded in true eukaryotic tRNAs. Two interesting ques-tions can be addressed: Can we trace the origin of thetRNA domains of these SINEs and are these tRNAdomains folded in the predicted canonical tRNA clover-leaf structure? By constructing phylogenetic trees, wefound that the tRNA-related regions of SINE-3_CRsequences are closely related to eukaryotic tRNAArg (seesupplemental Figure 6). Interestingly, whereas the trueChlamydomonas tRNAsAsp are, as expected, evolution-arily related to the eukaryotic aspartyl tRNAs (and branchwith metazoan tRNA sequences), the tRNAAsp(AUC andGUC)-related SINE domains of SINEX-5_CR and SINEX-6_CR are evolutionarily linked to bacterial and organ-ellar aspartyl tRNAs (Figure 5B). On the basis of theseobservations, we propose that tRNAAsp-related SINEs ofChlamydomonas derive initially from a prokaryotic tRNAthat has been horizontally transferred during evolution.We cannot also exclude the possibility that the tRNAat the origin of these two SINE families comes from theplastidial or mitochondrial tRNAAsp integrated in thenuclear genome during evolution.

With a few exceptions (Matsumoto et al. 1986; Lin

et al. 2001), the majority of SINEs derived from tRNAsdo not fold as their ancestral tRNA molecule and itwas recently shown that most tRNA-related SINEs havevery similar RNA structures (Rozhdestvensky et al.2001; Sun et al. 2007). Remarkably, when the intron isspliced, SINEX-3_CR, SINEX-5_CR, and SINEX-6_CRtRNA-related sequences can be folded in the theoreticalcloverleaf structure. They present only one or twomismatches in the stems and all invariant and semi-invariant nucleotides are present (Figure 5A). On onehand, RT–PCR experiments performed with a pair ofprimers specific to SINEX-3_CR tRNAArg-related se-quence allowed us to amplify the tRNA-related domainof this SINE family but in all 20 clones tested the intronwas present (data not shown). This result shows thatthe folding of the SINE RNA is not compatible with thecorrect splicing of a pre-tRNA molecule. On the otherhand, Northern experiments on a whole-cell tRNAfraction using as probes oligonucleotides specific toSINEX-5_CR and SINEX-6_CR tRNAAsp(AUC)- andtRNAAsp(GUC)-related SINEs gave no signal (data notshown). This demonstrates that no mature tRNA can beexpressed from these SINE RNAs, very likely because thecorrect cloverleaf folding recognized by tRNA-processing

Nucleus-Encoded tRNAs in Chlamydomonas 121

enzymes (e.g., RNAse P and RNAse Z) is not present. As awhole, it is likely that, despite their almost perfecttheoretical cloverleaf secondary structure, Chlamydo-monas tRNA-related SINE RNAs do not fold as theirancestral molecule.

Conclusion: On the basis of the data provided here onthe evolution and expression of Chlamydomonas tRNAgenes, it is clear that this green alga has retained somefeatures of the plant–animal common ancestor, asattested, for instance, by an La protein containing twoRRM domains. While the tRNASec gene is specific to the

animal lineage, only few tRNA genes are linked to theiranimal counterparts and most of them appear to bephylogenetically related to the plant ones. Whether theset of amino-acyl-tRNA synthetases able to recognizethese tRNAs has followed the same evolution will needto be answered in the future. In most cases, animalmitochondrial genomes have retained a complete set oftRNA genes whereas in higher plants, one-third to one-half of the mitochondrial tRNAs are nucleus encodedand used both in the cytosolic and in the mitochondrialtranslation machineries (Schneider and Marechal-

Figure 5.—tRNA-related SINEs in Chlamydomonas. (A) tRNA-like structure of putative tRNAArg-related and tRNAAsp-relatedSINE transcripts. Localization of the intron-like sequence is indicated by a black triangle. Point mutations leading to changesin the anticodon of the tRNAArg-related SINE family are indicated. All tRNA-related SINE sequences are available in the supple-mental data of Merchant et al. (2007). (B) Evolutionary origin of the ‘‘true’’ tRNAAsp and of tRNAAsp-like sequence at the origin ofthe tRNAAsp-related SINEs in Chlamydomonas. The phylogenetic tree was constructed as described in the Figure 1 legend.

122 V. Cognat et al.

Drouard 2000). The mitochondrial genome of C.reinhardtii encodes only three tRNA genes and thusthe import of nucleus-encoded tRNAs is expected. Thefact that Chlamydomonas cytosolic tRNAs resemblemore their plant than their animal counterparts canbe linked to the occurrence of this import process.Finally, we report the presence of several polycistronictRNA units in C. reinhardtii, which is quite unusual in aeukaryotic organism (Haeusler and Engelke 2006).Why and how Chlamydomonas has developed such aspecific mode of tRNA gene expression will need to befurther addressed.

This work was supported by the French Centre National de laRecherche Scientifique, by the Belgian Fonds de la RechercheFondamentale Collective (grant nos. 2.4582.05 and 2.4638.05 toC.R.), by a joint France–Wallonie Tournesol grant (to L.M.D. andC.R.), and by a short-term European Molecular Biology Organizationfellowship (to E.V.).

LITERATURE CITED

Ardell, D. H., and S. G. Andersson, 2006 TFAM detects co-evolutionof tRNA identity rules with lateral transfer of histidyl-tRNA synthe-tase. Nucleic Acids Res. 34: 893–904.

Arnaud, P., Y. Yukawa, L. Lavie, T. Pelissier, M. Sugiura et al.,2001 Analysis of the SINE S1 Pol III promoter from Brassica;impact of methylation and influence of external sequences. PlantJ. 26: 295–305.

Bulmer, M., 1987 Coevolution of codon usage and transfer RNAabundance. Nature 325: 728–730.

Delage, L., A. M. Duchene, M. Zaepfel and L. Marechal-Drouard,2003 The anticodon and the D-domain sequences are essentialdeterminants for plant cytosolic tRNA(Val) import into mito-chondria. Plant J. 34: 623–633.

Deragon, J. M., and X. Zhang, 2006 Short interspersed elements(SINEs) in plants: origin, classification, and use as phylogeneticmarkers. Syst. Biol. 55: 949–956.

Dong, H., L. Nilsson and C. G. Kurland, 1996 Co-variation oftRNA abundance and codon usage in Escherichia coli at differ-ent growth rates. J. Mol. Biol. 260: 649–663.

Duret, L., 2000 tRNA gene number and codon usage in the C. elegansgenome are co-adapted for optimal translation of highly expressedgenes. Trends Genet. 16: 287–289.

Fleurdepine, S., J. M. Deragon, M. Devic, J. Guilleminot and C.Bousquet-Antonelli, 2007 A bona fide La protein is requiredfor embryogenesis in Arabidopsis thaliana. Nucleic Acids Res. 35:3306–3321.

Glover, K. E., D. F. Spencer and M. W. Gray, 2001 Identificationand structural characterization of nucleus-encoded transferRNAs imported into wheat mitochondria. J. Biol. Chem. 276:639–648.

Haeusler, R. A., and D. R. Engelke, 2006 Spatial organization oftranscription by RNA polymerase III. Nucleic Acids Res. 34:4826–4836.

Harris, E. H., 1989 The Chamydomonas Sourcebook. Academic Press,San Diego.

Hudelot, C., V. Gowri-Shankar, H. Jow, M. Rattray and P. G.Higgs, 2003 RNA-based phylogenetic methods: applicationto mammalian mitochondrial RNA sequences. Mol. Phylogenet.Evol. 28: 241–252.

Jurka, J., O. Kohany, A. Pavlicek, V. V. Kapitonov and M. V. Jurka,2005 Clustering, duplication and chromosomal distribution ofmouse SINE retrotransposons. Cytogenet. Genome. Res. 110:117–123.

Kass, D. H., M. A. Batzer and P. L. Deininger, 1995 Gene conver-sion as a secondary mechanism of short interspersed element(SINE) evolution. Mol. Cell. Biol. 15: 19–25.

Kramerov, D. A., and N. S. Vassetzky, 2005 Short retroposons ineukaryotic genomes. Int. Rev. Cytol. 247: 165–221.

Kruszka, K., F. Barneche, R. Guyot, J. Ailhas, I. Meneau et al.,2003 Plant dicistronic tRNA-snoRNA genes: a new mode ofexpression of the small nucleolar RNAs processed by RNase Z.EMBO J. 22: 621–632.

LeBlanc, A. J., A. E. YermovskyKammerer and S. L. Hajduk,1999 A nuclear encoded and mitochondrial imported dicis-tronic tRNA precursor in Trypanosoma brucei. J. Biol. Chem.274: 21071–21077.

Lenoir, A., B. Cournoyer, S. Warwick, G. Picard and J. M. Deragon,1997 Evolution of SINE S1 retroposons in Cruciferae plantspecies. Mol. Biol. Evol. 14: 934–941.

Lin, Z., O. Nomura, T. Hasyahsi, Y. Wada and H. Yasue,2001 Characterization of a SINE species from vicuma and its dis-tribution in animal species including the family Camelidae.Mamm. Genome 12: 305–308.

Marechal-Drouard, L., P. Guillemaut, A. Cosset, M. Arbogast,F. Weber et al., 1990 Transfer RNAs of potato (Solanum tuberosum)mitochondria have different genetic origins. Nucleic Acids Res. 18:3689–3696.

Marechal-Drouard, L., I. Small, J.-H. Weil and A. Dietrich,1995 Transfer RNA import into plant mitochondria, pp. 310–327 in Mitochondrial Biogenesis and Genetics, Vol. 260, edited byG. M. Attardi and A. Chomym. Spectrum Publisher Services,New York.

Matsumoto, K., K. Murakami and N. Okada, 1986 Gene for lysinetRNA1 may be a progenitor of the highly repetitive and transcrib-able sequences present in the salmon genome. Proc. Natl. Acad.Sci. USA 83: 3156–3160.

Merchant, S. S., S. E. Prochnik, O. Vallon, E. H. Harris, S. J.Karpowicz et al., 2007 The Chlamydomonas genome revealsthe evolution of key animal and plant functions. Science 318:245–250.

Mottram, J. C., S. D. Bell, R. G. Nelson and J. D. Barry,1991a tRNAs of Trypanosoma brucei. Unusual gene organizationand mitochondrial importation. J. Biol. Chem. 266: 18313–18317.

Mottram, J. C., Y. Shafi and J. D. Barry, 1991b Sequence of a tRNAgene cluster in Trypanosoma brucei. Nucleic Acids Res. 19: 3995.

Okada, N., and K. Ohshima, 1995 Evolution of tRNA-derivedSINEs, pp. 61–80 in The Impact of Short Interspersed Elements (SINEs)on the Host Genome, edited by R. Maraia. Landes Company/Springer, Austin, TX.

Okada, N., I. M. Hamada, I. Ogiwara and K. Phshima, 1997 SINEsandLINEssharecommon39 sequences: areview.Gene205:229–243.

Pfitzinger, H., P. Guillemaut, J. H. Weil and D. T. N. Pillay,1987 Adjustment of the tRNA population to the codon usagein chloroplasts. Nucleic Acids Res. 15: 1377–1386.

Rao, M., B. A. Carlson, S. V. Novoselov, D. P. Weeks, V. N. Glady-

shev et al., 2003 Chlamydomonas reinhardtii selenocysteinetRNA½Ser�Sec. RNA 9: 923–930.

Rozhdestvensky, T. S., A. M. Kopylov, J. Brosius and A.Huttenhofer, 2001 Neuronal BC1 RNA structure: evolution-ary conversion of a tRNA(Ala) domain into an extended stem-loop structure. RNA 7: 722–730.

Schneider, A., and L. Marechal-Drouard, 2000 MitochondrialtRNA import: are there distinct mechanisms? Trends Cell Biol.10: 509–513.

Sun, F. J., S. Fleurdepine, C. Bousquet-Antonelli, G. Caetano-Anolles and J. M. Deragon, 2007 Common evolutionarytrends for SINE RNA structures. Trends Genet. 23: 26–33.

Tyc, K., Y. Kikuchi, M. Konarska, W. Filipowicz and H. J. Gross,1983 Ligation of endogenous tRNA 39 half molecules to their cor-responding 59 halves via 29-phosphomonester,3959-phosphodiesterbonds in extracts of Chlamydomonas. EMBO J. 2: 605–610.

Vincenti, S., V. De Chiara, I. Bozoni and C. Presutti, 2007 Theposition of yeast snoRNA-coding regions within host introns isessential for their biogenesis and for efficient splicing of the hostpre-mRNA. RNA 13: 138–150.

Wicker, T., F. Sabot, A. Hua-Van, J. L. Bennetzen, P. Capy et al.,2007 A unified classification system for eukaryotic transposableelements. Nat. Rev. Genet. 8: 973–982.

Wolin, S. L., and T. Cedervall, 2002 The La protein. Annu. Rev.Biochem. 71: 375–403.

Communicating editor: O. Vallon

Nucleus-Encoded tRNAs in Chlamydomonas 123