Small RNA-guided gene silencing is an evolutionarily conserved process that operates by a variety of molecular mechanisms. In multicellular eukaryotes, the core components of RNA-mediated silencing have significantly expanded and diversified, resulting in partly distinct pathways for the epigenetic control of gene expression and genomic parasites. In contrast, many unicellular organisms with small nuclear genomes seem to have lost entirely the RNA-silencing machinery or have retained only a basic set of components. We report here that Chlamydomonas reinhardtii, a unicellular eukaryote with a relatively large nuclear genome, has undergone extensive duplication of Dicer and Argonaute polypeptides after the divergence of the green algae and land plant lineages. Chlamydomonas encodes three Dicers and three Argonautes with DICER-LIKE1 (DCL1) and ARGONAUTE1 being more divergent than the other paralogs. Interestingly, DCL1 is uniquely involved in the post-transcriptional silencing of retrotransposons such as TOC1. Moreover, on the basis of the subcellular distribution of TOC1 small RNAs and target transcripts, this pathway most likely operates in the nucleus. However, Chlamydomonas also relies on a DCL1-independent, transcriptional silencing mechanism(s) for the maintenance of transposon repression. Our results suggest that multiple, partly redundant epigenetic processes are involved in preventing transposon mobilization in this green alga.

---

RNA-MEDIATED silencing is an evolutionarily conserved process by which double-stranded RNA (dsRNA) induces the inactivation of cognate sequences through a variety of mechanisms, including translation inhibition, RNA degradation, transcriptional repression, or DNA elimination (BAULCOMBE 2004; MEISTER and TUSCHL 2004; MATZKE and BIRCHLER 2005; MATRANGA and ZAMORE 2007). dsRNA-triggered silencing was initially characterized in Caenorhabditis elegans and termed RNA interference (RNAi) (FIRE et al. 1998), but this phenomenon is now known to occur in a wide spectrum of eukaryotes (CERUTTI and CASAS-MOLLANO 2006). Moreover, despite the mechanistic diversity of RNA-mediated repression, all characterized pathways appear to involve small RNAs (20–30 nucleotides in length) generated by the processing of dsRNAs, with the possible exception of Piwi-associated RNAs and certain small RNAs produced directly by polymerization (BARTEL 2004; BRODERSEN and VOINNET 2006; VAUCHERET 2006; AOKI et al. 2007; ARAVIN et al. 2007; MATRANGA and ZAMORE 2007; PAK and FIRE 2007). Intriguingly, recent results indicate that small RNAs not only function in gene silencing but also may participate in activation of expression (LI et al. 2006; JANOWSKI et al. 2007; VASUDEVAN et al. 2007).

Genetic and biochemical studies from multiple organisms have led to the identification of three core components of the RNAi machinery, namely Dicer, Argonaute-Piwi (AGO-Piwi), and RNA-dependent RNA polymerase (RDR) (COGONI and MACINO 2000; BAULCOMBE 2004; ZAMORE and HALEY 2005). Long or hairpin dsRNAs are processed into small RNAs by the RNaseIII-like endonuclease Dicer (BERNSTEIN et al. 2001; MEISTER and TUSCHL 2004; ZAMORE and HALEY 2005). These small RNAs are then incorporated into effector complexes, which include members of the AGO-Piwi family of proteins. This family consists of two main classes of polypeptides, one named after Arabidopsis thaliana ARGONAUTE1 and the other after Drosophila melanogaster Piwi (CARMELL et al. 2002; CERUTTI and CASAS-MOLLANO 2006; ARAVIN et al. 2007). AGO-Piwi proteins contain two conserved motifs: the PAZ (Piwi/Argonaute/Zwille) domain, which binds to the 3'-ends of small RNAs, and the Piwi domain, which is structurally related to RNase H (CERUTTI et al. 2000; MA et al. 2004; SONG et al. 2004; YUAN et al. 2005). Some AGO-Piwi polypeptides function as small RNA-guided endonucleases that cleave complementary RNAs (LIU et al. 2004; MEISTER et al. 2004; BAUMBERGER and BAULCOMBE 2005). Others, in contrast, are not endonucleolytically active (LIU et al. 2004; MEISTER et al. 2004; RIVAS et al. 2005) and may be part of effector complexes involved in nondegradative RNAi processes (WIENHOLDS and PLASTERK 2005; ZAMORE and HALEY 2005). In certain organisms such as nematodes, land plants, and fungi, RDRs also play an important role in RNAi (COGONI and MACINO 2000; SIJEN et al. 2001; BAULCOMBE 2004). In these species, RDR activity may initiate RNAi by producing dsRNA from single-stranded transcripts or dramatically enhance the RNAi response by amplifying the amounts of small RNAs (CERUTTI 2003; BAULCOMBE 2004; AOKI et al. 2007; PAK and FIRE 2007).

In multicellular eukaryotes, duplication and diversification of the protein components for small RNA biogenesis and/or of those involved in effector functions have resulted in complex, partly overlapping pathways for the epigenetic control of gene expression (OKAMURA et al. 2004; MATZKE and BIRCHLER 2005; BRODERSEN and VOINNET 2006; VAUCHERET 2006; CHAPMAN and CARRINGTON 2007; MATRANGA and ZAMORE 2007). Three major classes of small RNAs have been identified in animals: microRNAs (miRNAs), Piwi-interacting RNAs (piRNAs), and small interfering RNAs (siRNAs) (ARAVIN et al. 2007; MATRANGA and ZAMORE 2007). Plant species lack Piwi proteins (CERUTTI and CASAS-MOLLANO 2006; ARAVIN et al. 2007) and contain only miRNAs and siRNAs. MicroRNAs originate from endogenous RNA transcripts that fold into imperfect stem-loop structures. They often modulate the expression of genes with roles in development, physiological processes, or stress responses (BARTEL 2004; WIENHOLDS and PLASTERK 2005; CHAPMAN and CARRINGTON 2007; MATRANGA and ZAMORE 2007). piRNAs are longer than miRNAs or siRNAs and interact specifically with Piwi proteins, and some of them seem to function in the control of mobile genetic elements (ARAVIN et al. 2007; MATRANGA and ZAMORE 2007). Their biogenesis is not clearly understood but piRNAs appear to derive from single-stranded transcripts by a Dicer-independent mechanism (ARAVIN et al. 2007; MATRANGA and ZAMORE 2007). siRNAs are produced from long dsRNAs of diverse origins, including the transcripts of long inverted repeats, the products of convergent transcription or RDR activity, viral RNAs, or dsRNA experimentally introduced into cells (BAULCOMBE 2004; CHAPMAN and CARRINGTON 2007; MATRANGA and ZAMORE 2007). On the basis of their origins and functions, siRNAs can be further classified into several subclasses such as primary siRNAs, secondary siRNAs, heterochromatic siRNAs, trans-acting siRNAs, and natural antisense transcript-derived siRNAs. These siRNAs play various roles in post-transcriptional regulation of gene expression, suppression of viruses and transposable elements, and heterochromatin formation (BRODERSEN and VOINNET 2006; VAUCHERET 2006; CHAPMAN and CARRINGTON 2007; DING and VOINNET 2007; MATRANGA and ZAMORE 2007).

Unicellular eukaryotes appear to have fewer RNAi components and small RNA-mediated pathways than multicellular ones (CERUTTI and CASAS-MOLLANO 2006). Yet many unicellular organisms also possess relatively small nuclear genomes and therefore it is not clear whether diversification of RNA-mediated silencing is a unique innovation coupled to the evolution of multicellularity or whether it is associated mainly with complex genomes, regardless of cellularity, where the additional level of regulation may confer a selective advantage. In this respect, the unicellular green alga Chlamydomonas reinhardtii may provide valuable insights since its 120-Mb nuclear genome (MERCHANT et al. 2007) is similar in size to that of the higher plant A. thaliana. Chlamydomonas possesses a functional RNAi machinery, as reflected by the generation of siRNAs from inverted repeat transgenic transcripts and the corresponding suppression of expression of target genes (ROHR et al. 2004; IBRAHIM et al. 2006; SCHRODA 2006). Moreover, this alga has recently been shown to contain a complex set of small RNAs including miRNAs, phased siRNAs, as well as siRNAs originating from transposons and repeats (MOLNAR et al. 2007; ZHAO et al. 2007). Yet the RNAi machinery and its biological role(s), in particular with regard to pathway specialization, have not been explored in detail in Chlamydomonas. Here we examine the core RNAi components encoded in the C. reinhardtii genome and demonstrate that DCL1 is involved in the post-transcriptional silencing of the TOC1 retrotransposon. This function is not compensated by two other Dicer paralogs, indicating that RNAi pathway diversification does occur in Chlamydomonas. However, when grown under standard laboratory conditions, this alga relies primarily on a transcriptional silencing mechanism(s) to control mobile genetic elements. In fact, C. reinhardtii appears to have several, at least partly independent, transposon silencing pathways that operate at either the transcriptional or the post-transcriptional levels.

C. reinhardtii strains, culture conditions, and generation of transgenic strains:

The CC-124 and Mut-11 strains have been previously described (HARRIS 1989; ZHANG et al. 2002). CC-3491 was obtained from the Chlamy Center (http://www.chlamy.org). C. reinhardtii cells were routinely grown in Tris–acetate–phosphate (TAP) medium (HARRIS 1989) under moderate light conditions at 21° (JEONG et al. 2002). To suppress DCL1 expression by RNAi, cells were transformed by the glass-beads procedure (KINDLE 1990) with an inverted repeat transgene designed to produce dsRNA homologous to the DCL1 mRNA. An 360-bp fragment corresponding to part of the DCL1 coding sequence was amplified by reverse transcriptase–PCR (RT–PCR) with primers Dcr-3 (5'-GCTGGAGACCCTGGGTGA-3') and Dcr-2 (5'-CTGCGCGTCATTGCTGTT-3'). This segment was then inserted in sense and antisense orientations, flanking a spacer sequence, in the Maa7/X IR vector (ROHR et al. 2004). RNAi-positive transgenic strains were identified as previously described (ROHR et al. 2004). Chlamydomonas cells transformed with the empty Maa7/X IR vector, as a control, showed no effect on TOC1 or GULLIVER reactivation (VAN DIJK et al. 2005 and data not shown).

DNA sequence and phylogenetic analyses:

The coding sequence of the predicted DCL1 gene was confirmed by sequencing a truncated cDNA and products generated by reverse transcriptase–PCR amplification from polyadenylated RNA. Polypeptides homologous to AGO-Piwi, Dicer, or RDR were identified by BLAST or PSI-BLAST searches of protein and/or translated genomic DNA databases. Sequences corresponding to conserved Argonaute or Dicer domains were extracted by comparison to the SMART database and aligned with ClustalX (THOMPSON et al. 1997). The neighbor-joining method (SAITOU and NEI 1987) and the MEGA program v3.1 (KUMAR et al. 2004) were used to obtain phylogenetic trees with Poisson-corrected amino acid distances. The coding sequence of DCL1 has been deposited with the EMBL/GenBank Data Libraries (accession no. EU368690).

DNA and RNA analyses:

Standard protocols were used for nucleic acid isolation, fractionation by gel electrophoresis, and hybridization with ³²P-labeled probes (CERUTTI et al. 1997a; SAMBROOK and RUSSELL 2001). To detect small RNAs by Northern blotting, TRI reagent (Molecular Research Center) isolated total RNA was fractionated through Microcon YM-100 centrifugal devices (Millipore, Bedford, MA) to remove high-molecular-weight transcripts. Small RNAs were concentrated from the filtrate by ethanol precipitation, resolved in 15% polyacrylamide/7 M urea gels and electroblotted to Hybond-XL membranes (GE Healthcare). Blots were then hybridized as previously described (ROHR et al. 2004; IBRAHIM et al. 2006).

RT–PCR analyses:

Total RNA was isolated with TRI reagent, according to the manufacturer's instructions (Molecular Research Center), and contaminant DNA was removed by DNase-I treatment (Ambion). First-strand cDNA synthesis and PCR reactions were performed as previously described (ROHR et al. 2004; VAN DIJK et al. 2005). PCR products were resolved on 2% agarose gels and visualized by ethidium bromide staining. The number of cycles showing a linear relationship between input RNA and the final product was determined in preliminary experiments. Controls included the use as template of reactions without RT and verification of PCR products by hybridization with specific probes (data not shown). The primer sequences were as follows: for DCL1, Dcr-7 (5'-GCAGCATCGATGGTACTGATAGC-3') and Dcr-10 (5'-CTGCACGTGCTTGCTTGGAT-3'); for DCL2, Dicer2-F4 (5'-CAGTGTGTGCGAGCTGGAG-3') and Dicer2-R3 (5'-GGACATGGCCTCGGCACT-3'); for DCL3, Dicer3-F3 (5'-GTGGTGTCCTTCTGGCTGTTC-3') and Dicer3-R3 (5'-GTTGCTGACCAGCGCCTTG-3'); and for ACT1, ACT-cod-F (5'-GACATCCGCAAGGACCTCTAC-3') and ACT-cod-R (5'-GATCCACATTTGCTGGAAGGT-3').

Nuclear run-on transcription assays:

Chlamydomonas cells grown to mid-log phase were concentrated by centrifugation, resuspended to a density of 1 x 10⁸ cells/ml in wash buffer (10 mM Tris–HCl, pH 8.0, 0.5 mM EDTA, 50 mM KCl, 250 mM sucrose, and 1 mM DTT), and frozen in liquid nitrogen. These permeabilized cells were then used to determine transcriptional activity as previously described for isolated nuclei (KELLER et al. 1984; CERUTTI et al. 1997a). Nuclear transcripts, labeled with [³²P]rUTP, were employed as probes on filters containing an excess of target DNA. The radioactivity retained on the membranes was quantified with a PhosphorImager (GE Healthcare) (CERUTTI et al. 1997a,b).

Immunoblot analyses:

Histone methylation status was examined in vivo by Western blotting, as previously described (VAN DIJK et al. 2005; CASAS-MOLLANO et al. 2007). The different methylated states of histone H3 Lys 4 were detected with antibodies against H3K4me1 (Abcam, ab8895), H3K4me2 (Upstate, 07-030), or H3K4me3 (Abcam, ab8580). A modification-insensitive anti-H3 antibody (Abcam, ab1791) was used to adjust sample loading.

Chromatin immunoprecipitation assays:

The H3K4 methylation status of the chromatin associated with the TOC1 long terminal repeat (LTR) was examined by chromatin immunoprecipitation (ChIP), following a previously described protocol (VAN DIJK et al. 2005; CASAS-MOLLANO et al. 2007). Since the antibody against H3K4me3 (Abcam) lacks specificity and cross-reacts with several nonhistone proteins (VAN DIJK et al. 2005), this modification was not examined by ChIP. Rabbit IgG (Sigma, St. Louis) was employed as a negative control. Immunoprecipitated DNA was quantified by real-time PCR on a Bio-Rad (Hercules, CA) iCycler using SYBR Green (CASAS-MOLLANO et al. 2007). After each run, a melting curve was performed to ensure that no primer dimers interfered with the quantification. The primers used for amplification of the target locus have been previously reported (VAN DIJK et al. 2005).

Cytoplasmic/nuclear fractionation of RNA:

Cells from CC-3491, which lacks a cell wall, were grown to mid-log phase in TAP medium, collected by centrifugation, and resuspended in solution I from KELLER et al. (1984). Triton X-100, instead of Nonidet P-40, was used to lyse the cells, and nuclei were pelleted by centrifugation as previously described (KELLER et al. 1984). The supernatant was used for the isolation of cytoplasmic RNA by a SDS-based protocol (SAMBROOK and RUSSELL 2001). The pellet, consisting mostly of nuclei and cell debris, was washed twice with solution I without detergent (KELLER et al. 1984) and then used for nuclear RNA isolation by the TRI reagent procedure. Total cell RNA was purified from CC-3491 as previously reported (ROHR et al. 2004). For analysis by Northern blotting, we used cell-volume equivalent amounts (RUDENKO et al. 2003) of total (11.5 µg), cytoplasmic (10 µg), and nuclear (1.5 µg) RNAs. Small RNAs were purified from the cytoplasmic, nuclear, and total cell RNA fractions as described under DNA and RNA analyses.

The core RNAi machinery in C. reinhardtii and other green algae:

We identified polypeptides related to AGO-Piwi, Dicer, or RDR by BLAST or PSI-BLAST searches of protein and/or translated genomic DNA databases. Since several of the examined genomes are in draft stage, an important caveat in our analyses is that some proteins may be missing from the databases whereas others may have errors in the predicted gene structure. However, we considered as potential homologs only proteins that exhibited enough sequence similarity to be aligned together and to be used for phylogenetic tree construction. The core RNAi machinery components seem to be entirely absent from algal species with small nuclear genomes such as the red alga Cyanidioschyzon merolae (an 16.5-Mb nuclear genome) and the green algae Ostreococcus lucimarinus and Ostreococcus tauri (an 12.6-Mb nuclear genome) (data not shown). This is consistent with the proposal that the RNAi mechanism appears to have been lost independently several times during eukaryotic evolution (CERUTTI and CASAS-MOLLANO 2006; NAKAYASHIKI et al. 2006). In contrast, obvious AGO and Dicer-like homologs are present in Chlamydomonas, the related volvocine alga Volvox carteri f. nagariensis as well as in another green alga, Chlorella sp. NC64A, an endosymbiont of the ciliated protozoan Paramecium bursaria (Figures 1 and 2). As in the case of land plants (CERUTTI and CASAS-MOLLANO 2006; ARAVIN et al. 2007), algae appear to lack Piwi proteins.

View larger version (24K): In this window In a new window Download PPT slide   FIGURE 1.— Neighbor-joining tree showing the phylogenetic relationship among Argonaute proteins. Sequences corresponding to the PAZ and Piwi domains of each polypeptide were aligned using the ClustalX program and the tree was drawn using the MEGA v3.1 program. Numbers indicate bootstrap values, >60%, based on 1000 pseudoreplicates. Polypeptides belonging to the green alga lineage are shaded. Species are designated by a two-letter abbreviation preceding the name of each protein: At, A. thaliana; Cr, C. reinhardtii; Cs, Chlorella sp. NC64A; Dm, D. melanogaster; Hs, H. sapiens; Os, O. sativa; and Vc, V. carteri f. nagariensis. Accession numbers of proteins used to draw the tree are the following: At AGO1, AAC18440; At AGO2, NP_174413; At AGO3, NP_174414; At AGO4, NP_565633; At AGO5, NP_850110; At AGO6, NP_180853; At AGO7(ZIP), NP_177103; At AGO8, NP_197602; At AGO9, NP_197613; At AGO10(PNH), CAA11429; Cr AGO1, XP_001694840; Cr AGO2, XP_001698670; Cr AGO3, XP_001698906; Dm Ago1, BAA88078; Dm Ago2, Q9VUQ5; Cs NC64A scaffold-12-45 gene25, v2_NC64A_scaffold_12_45_gene25; Hs AGO1(eIF2C-1), AAH63275; Hs AGO2(eIF2C-2), AAL76093; Hs AGO3(eIF2C-3), BAB14262; Hs AGO4(eIF2C-4), BAB13393; Os AGO701, XP_468547; Os AGO702, BAB96813; Os AGO703, NP_912975; Os AGO704, XP_478040; Os AGO705, AL606693; Os AGO706, NP_909924; Os AGO707, AP005750; Os AGO708, XP_473529; Os AGO709, XP_473887; Os AGO710, XP_469312; Os AGO711, AP004188; Os AGO712, XP_476934; Os AGO713, XP_473888; Os AGO714, XP_468898; Os AGO715, XP_477327; Os AGO716, XP_464271; Os AGO717, XP_469311; Os AGO719, AP000836; Vc_105714, v1_105714; Vc_92637, v1_92637. Accession numbers in italics are according to the annotated draft genomes of V. carteri (http://genome.jgi-psf.org/Volca1/Volca1.home.html) and Chlorella sp. NC64A (http://greengene.uml.edu/chlorella/chlorella.html).

 

View larger version (25K): In this window In a new window Download PPT slide   FIGURE 2.— Phylogenetic tree of Dicer-like proteins. Sequences corresponding to the RNaseIII domains were aligned using the ClustalX program, and a neighbor-joining tree was constructed using MEGA v3.1. Numbers show bootstrap values, >60%, based on 1000 pseudoreplicates. Polypeptides belonging to the green alga lineage are shaded. Species are designated by a two-letter abbreviation preceding the name of each protein, as described in the legend to Figure 1. Accession numbers of proteins used to draw the tree are the following: At DCL1, NP_171612; At DCL2, NP_566199; At DCL3, NP_189978; At DCL4, NP_197532; Cr DCL1, EU368690; Cr DCL2, XP_001698921; Cr DCL3, XP_001692436; Cs NC64A scaffold-5-57 gene7, v2_NC64A_scaffold_5_57_gene7; Dm Dcr1, Q9VCU9; Dm Dcr2, BAB69959; Hs DCR1, NP_803187; Os DCL1, NP_912466; Os DCL2a, XP_463068; Os DCL2b, BAD34005; Os DCL3a, XP_463595; Os DCL3b, NP_922059; Os DCL4, XP_473129; Vc_106340, v1_106340. Accession numbers in italic are according to the annotated draft genomes of V. carteri (http://genome.jgi-psf.org/Volca1/Volca1.home.html) and Chlorella sp. NC64A (http://greengene.uml.edu/chlorella/chlorella.html).

 
The dual domain structure of the AGO-Piwi polypeptides, namely a PAZ domain followed by a Piwi domain (Figure 3), has been well conserved in the three species of algae, although some of the predicted AGO gene models may be partly incorrect or incomplete due to gaps in the available genomic sequences (data not shown). The Dicer enzymes initially characterized in D. melanogaster and humans are multidomain proteins consisting of RNA helicase domains, a motif of unknown function (DUF283) that appears to be a divergent dsRNA-binding domain (dsRB) (DLAKIC 2006), a PAZ domain, two RNaseIII catalytic domains, and a dsRNA-binding domain (BERNSTEIN et al. 2001; MEISTER and TUSCHL 2004; CERUTTI and CASAS-MOLLANO 2006; MARGIS et al. 2006). This overall organization is maintained in Dicer-like proteins from the volvocine algae except for the absence of obvious PAZ and dsRB domains (Figure 3; ZHAO et al. 2007). In contrast, an RDR was not identified in the genomes of Chlamydomonas, Volvox, or Chlorella sp. NC64A, but a clear homolog was predicted from the sequence of the free-living Chlorella vulgaris C-169 (data not shown). Interestingly, we could not detect AGO or Dicer-related polypeptides in the genome of the latter species, which, considering their existence in Chlorella sp. NC64A, suggests that the draft genome of C. vulgaris C-169 is most likely incomplete. This issue may also apply to the other algal species. In Arabidopsis, the biogenesis of phased trans-acting siRNAs requires RDR activity (ALLEN et al. 2005; YOSHIKAWA et al. 2005; CHAPMAN and CARRINGTON 2007). Chlamydomonas also contains phased siRNAs (MOLNAR et al. 2007; ZHAO et al. 2007), despite an apparent lack of an RDR. However, a long dsRNA generated by convergent transcription and annealing of the complementary RNAs or a foldback transcript of long inverted repeats could also be cleaved by Dicer in a processive fashion to generate phased, RDR-independent siRNAs. Given these uncertainties, the question of whether C. reinhardtii encodes an RDR homolog will likely remain open until all gaps in the current version of the genome are filled.

View larger version (16K): In this window In a new window Download PPT slide   FIGURE 3.— Genomic organization of C. reinhardtii DCL1 and AGO1 and schematic of the corresponding proteins. (Top) The chromosomal arrangement of the DCL1 and AGO1 genes, which are transcribed in divergent orientation (arrows) on linkage group II (http://genome.jgi-psf.org/Chlre3/Chlre3.home.html). Polyadenylation sites are indicated by the stop signs. (Middle) The precursor messenger RNAs (excluding 5' and 3' untranslated regions) with exons indicated by solid boxes. The annealing sites of primers used for RT–PCR amplification are shown below the exons. (Bottom) The domain architecture of the DCL1 and AGO1 proteins. Black oval, nuclear localization signal; PAZ, Piwi/Argonaute/Zwille domain; Piwi, Piwi domain; DEXHc, DEAD/DEAH-like helicase superfamily domain; HelC, helicase superfamily C-terminal domain; DUF283, putative divergent dsRNA-binding fold; RIII, ribonucleaseIII C-terminal catalytic domain.

 
To gain insight into the evolution and diversity of the RNA-mediated silencing machinery among algae, phylogenetic analysis of the Argonaute and Dicer-like proteins was performed with the available algal sequences as well as with those from two land plants (A. thaliana and Oryza sativa) and two metazoans (D. melanogaster and Homo sapiens). In C. reinhardtii, three paralogs of AGO (AGO1, AGO2, and AGO3) and of Dicer-like (DCL1, DCL2, and DCL3) proteins have been identified (Figures 1 and 2). Moreover, 23-nt transgenic siRNAs (ROHR et al. 2004; IBRAHIM et al. 2006) and 21-nt transposon small RNAs (VAN DIJK et al. 2006) have been detected by Northern blotting, suggesting that, in Chlamydomonas, different Dicers may be functional and generate small RNAs of various lengths. Volvox carteri contains two AGO paralogs and one Dicer-like polypeptide whereas Chlorella sp. NC64A has only one copy of each protein in the respective draft versions of their genomes (Figures 1 and 2). The examined Argonaute and Dicer-like polypeptides fell into three relatively well-supported groups that included exclusively proteins from land plants, algae, or animals (Figures 1 and 2). The only exception was D. melanogaster Ago2 that clustered with the algal Argonaute polypeptides (Figure 1), but this is likely a methodological artifact resulting from the unusually fast evolution of the fly gene (see below). AGO proteins have undergone a marked degree of expansion in most of the eukaryotic lineages analyzed (Figure 1), conceivably associated with extensive functional diversification. In land plants, several duplications of AGO polypeptides appear to have occurred both before and after the divergence of dicots and monocots, represented by A. thaliana and O. sativa, respectively (Figure 1). Expansion of Argonaute proteins has also occurred in the algal lineages, particularly in C. reinhardtii (Figure 1). However, our phylogenetic analysis indicates that gene duplication and, likely, pathway diversification occurred after the divergence of green algae from the lineage leading to land plants. A phylogenetic tree of Dicer-like proteins supports a similar conclusion (Figure 2).

Most animals appear to encode a single Dicer gene, with the exception of insects that contain two. Whereas insect Dcr1 clusters with all other animal Dicers, Dcr2 is much more divergent and forms a paralogous clade (CERUTTI and CASAS-MOLLANO 2006). Insect Ago2 is also much more divergent than Ago1 and does not cluster with most other animal Argonaute-like proteins (Figure 1; CERUTTI and CASAS-MOLLANO 2006). In D. melanogaster, Dcr1 and Ago1 are involved in the miRNA pathway whereas the divergent Dcr2 and Ago2 play a role in siRNA-mediated processes (OKAMURA et al. 2004; ZAMORE and HALEY 2005). Moreover, Dcr2 and Ago2 have been implicated in antiviral immune responses (ZAMBON et al. 2006; DING and VOINNET 2007) and a recent report suggests that both Dcr2 and Ago2 are among the fastest-evolving genes in flies, perhaps as a result of a co-evolutionary "arms race" with viral pathogens (OBBARD et al. 2006). In Arabidopsis, DCL2, DCL3, and DCL4, which function in a variety of siRNA-mediated silencing pathways (including virus defense and transposon repression) (BRODERSEN and VOINNET 2006; VAUCHERET 2006; CHAPMAN and CARRINGTON 2007; DING and VOINNET 2007), are also more divergent than DCL1, which generates miRNAs (BARTEL 2004; MARGIS et al. 2006). Interestingly, C. reinhardtii DCL1 and AGO1 do not cluster tightly with their counterparts in the closely related alga V. carteri and appear to be more divergent than the other paralogs (Figures 1 and 2).

C. reinhardtii DCL1 and AGO1 are encoded by adjacent, divergently transcribed genes:

By analogy to insects and land plants, we hypothesized that Chlamydomonas DCL1 and AGO1, which seem to be more divergent than the other paralogs, may be involved in defense responses against genomic parasites such as viruses and transposable elements. Therefore, we decided to examine their genes in more detail. The amino acid sequence of DCL1 has been previously inferred from genomic DNA sequences by computational prediction of exon/intron regions. To obtain more accurate information on this protein, a truncated cDNA and RT–PCR products were used to determine the actual coding sequence and exon/intron boundaries (Figure 3). The coding region of AGO1 was not verified experimentally but both the 5'- and the 3'-ends of the predicted model are supported by available ESTs. Interestingly, the genes encoding DCL1 and AGO1 are arranged head to head near the distal end of linkage group II. Their respective translation start sites are separated by only 300 bp (Figure 3), suggesting that these genes may be transcriptionally coregulated by expression from a common bidirectional promoter and, conceivably, part of the same RNAi pathway.

DCL1 appears to play a minor role in transposon silencing in wild-type C. reinhardtii grown under standard laboratory conditions:

Several partly active transposable elements have been described in Chlamydomonas, including two class I retrotransposons, TOC1 and REM1, and a number of class II DNA elements, GULLIVER, PIONEER, TCR1, TCR2, and TCR3 (DAY et al. 1988; FERRIS 1989; SCHNELL and LEFEBVRE 1993; WANG et al. 1998; PÉREZ-ALEGRE et al. 2005). To test directly the role of DCL1 in defense responses against some of these mobile genetic elements, we decided to suppress its expression by RNAi. The wild-type Chlamydomonas strain CC-124 was transformed with an inverted repeat construct designed to produce dsRNA homologous to DCL1 (Dcl1-IR). However, the DNA sequence chosen to build the inverted repeat is quite divergent from the corresponding regions in DCL2 or DCL3, to avoid the unintended downregulation of expression of the latter genes. By using this approach, we recovered several transformants displaying DCL1 suppression (to 25% of the wild-type transcript amounts) without altering DCL2 or DCL3 mRNA levels [Figure 4A, CC-124(Dcl1-IR); data not shown].

View larger version (57K): In this window In a new window Download PPT slide   FIGURE 4.— RNAi-mediated suppression of DCL1 affects the post-transcriptional silencing of the TOC1 retrotransposon. (A) RT–PCR analysis of DCL1, DCL2, and DCL3 expression in the indicated strains. Reactions using RNA not treated with reverse transcriptase (–RT) as the template were employed as a negative control. Amplification of ACT1 (encoding actin) transcripts is shown as an input control. Numbers below the panels indicate relative levels of specific transcripts normalized to the ACT1 mRNA amount. CC-124, wild-type C. reinhardtii; CC-124(Dcl1-IR), CC-124 transformed with an inverted repeat (IR) transgene designed to produce dsRNA homologous to DCL1; Mut-11, mutant defective in a core subunit of H3K4 methyltransferase complexes; Mut-11(Dcl1-IR), Mut-11 transformed with an IR transgene designed to induce RNAi of DCL1. (B) Northern blot detection of transposon siRNAs. Column-fractionated small RNAs were separated in a 15% denaturing polyacrylamide gel, electroblotted onto a nylon membrane, and hybridized with a probe corresponding to the right terminus of the GULLIVER transposon (middle). The same blot was then sequentially reprobed for the TOC1 LTR (top) and for the U6 small nuclear RNA (bottom) as a loading control. (C) Northern blot of total cell RNA probed sequentially for TOC1 (top) to examine transcript levels and for ACT1 (bottom) to test for equivalent loading of the lanes. The asterisk indicates the full-length, 5.5-kb TOC1 transcript. (D) Southern blot analysis of TOC1 transposition. Genomic DNA from parallel subcultures (clones) of the indicated strains was digested with HincII and probed for TOC1. The asterisks indicate newly integrated transposon copies in the subclones of Mut-11(Dcl1-IR).

 
Long antisense RNAs corresponding to the TOC1 retrotransposon have been previously observed in certain Chlamydomonas strains and postulated to arise by readthrough transcription from a promoter(s) flanking a specific transposon copy (DAY and ROCHAIX 1991). Moreover, by genetic tetrad analyses, the production of sense and antisense transcripts in the same cell, presumably annealing into dsRNA that triggers RNAi, has been shown to inhibit TOC1 RNA accumulation (DAY and ROCHAIX 1991). In agreement with these findings, we have detected siRNAs derived from the TOC1 LTRs in CC-124 (Figure 4B), although they appear to be present at very low levels. This observation is consistent with the small number of transposon siRNAs identified by high-throughput sequencing in Chlamydomonas (MOLNAR et al. 2007; ZHAO et al. 2007). In addition, we were unable to detect GULLIVER siRNAs by Northern blotting in the wild-type background (Figure 4B).

RNAi-mediated suppression of DCL1 did reduce the steady-state level of TOC1 siRNAs (Figure 4B). However, this resulted in only a very slight increase in TOC1 transcripts in comparison with the wild type (Figure 4C). Furthermore, the TOC1 transposition frequency did not appear to be affected. Southern blot analyses of 10 parallel subcultures of CC-124 and CC-124(Dcl1-IR) revealed no additional TOC1 copies in the strain where DCL1 expression was downregulated (Figure 4D and data not shown). GULLIVER transposition was also unchanged by RNAi-mediated DCL1 suppression (data not shown). Interestingly, the full-length, transposition competent TOC1 RNA, which is 5.5 kb (DAY et al. 1988; DAY and ROCHAIX 1991), was virtually undetectable in Northern blots of the CC-124 and CC-124(Dcl1-IR) strains (Figure 4C). The majority of the observed TOC1 transcripts were <5.5 kb and very heterogeneous in size, giving rise to a smeary signal on the RNA blots (Figure 4C). Indeed, on the basis of the available Chlamydomonas ESTs, most TOC1 transcripts in wild-type cells appear to be prematurely terminated RNAs, solo LTRs, or transcripts initiated from flanking promoters that extend and terminate into a TOC1 element (data not shown).

DCL1 is required for the post-transcriptional repression of the TOC1 retrotransposon in a mutant strain defective in transcriptional silencing:

We reasoned that the very modest effect of DCL1 suppression on TOC1 reactivation in wild-type Chlamydomonas might be due to the fact that this retrotransposon (in particular its full-length, transposition-competent form) is silenced mainly at the transcriptional level by a DCL1-independent mechanism(s). Indeed, we have previously demonstrated that both TOC1 and GULLIVER are repressed by effectors that operate at the transcriptional level (JEONG et al. 2002; ZHANG et al. 2002). MUT11 encodes a subunit of histone methyltransferase complexes and its deficiency results in the loss of histone H3 lysine 4 monomethylation (H3K4me1), enrichment in H3K4 dimethylation (H3K4me2), a decrease in H3K4 trimethylation (H3K4me3), and the reactivation of silenced transgenes and transposons (JEONG et al. 2002; VAN DIJK et al. 2005). In nuclear run-on assays, TOC1 transcription was significantly enhanced in the Mut-11 mutant whereas DCL1 suppression in an otherwise wild-type background had no effect [Figure 5A, cf. CC-124(Dcl1-IR) and Mut-11]. We also observed a substantial increase in transposon-derived siRNAs (both TOC1 and GULLIVER small RNAs) in Mut-11 (Figure 4B). We speculate that an enhancement in overall genome transcription in Mut-11, due to defects in chromatin repression, results in greater production of both sense and antisense transposon transcripts that likely anneal, producing increased amounts of dsRNAs processed by Dicer into siRNAs.

View larger version (60K): In this window In a new window Download PPT slide   FIGURE 5.— DCL1 is dispensable for transcriptional silencing of the TOC1 retrotransposon. (A) Transcriptional activity of TOC1 in nuclear run-on assays of the indicated strains, described in the legend to Figure 4. Transcription of TUBA (encoding -tubulin) and ACT1 (encoding actin) was evaluated as a control. (B) Southern blot analysis of TOC1 and GULLIVER cytosine DNA methylation with isoschizomeric restriction enzymes (REs). Total cell DNA was digested with the indicated REs, resolved by agarose gel electrophoresis, and probed with the TOC1 LTR (left) or with the right terminus of GULLIVER (right). In the absence of cytosine methylation, HinfI/MspI or HinfI/HpaII cleavage would result in a TOC1 fragment of 140 bp, whereas XbaI/MspI or XbaI/HpaII cleavage would result in a GULLIVER fragment of 210 bp. The multiple segments obtained with HinfI or XbaI digestions reflect sequence heterogeneity among different copies of TOC1 or GULLIVER. (C) Immunoblot analysis of in vivo H3K4 methylation states. Whole-cell protein extracts from the indicated strains were separated by SDS–PAGE, transferred to nitrocellulose, and probed with antibodies raised against mono-, di-, or trimethyl H3K4. Sample loading was calibrated on the basis of immunoblots with a modification-insensitive anti-H3 antibody. (D) ChIP assay of H3K4 modifications associated with the TOC1 LTR. ChIP was performed with anti-H3K4me1, anti-H3K4me2, or anti-H3 antibodies. A rabbit IgG antibody was used as a negative control. Immunoprecipitated DNA was examined by real-time PCR and enrichment was calculated relative to the anti-H3 immunoprecipitate. For illustration purposes, the level of H3K4me1 in the CC-124 strain and the level of H3K4me2 in the Mut-11 mutant were set to 1.0 and the remaining samples adjusted accordingly in the bar graph. Results represent the mean ±SD of three independent experiments.

 
To further explore the connection between DCL1 and transposon silencing, we also suppressed DCL1 expression by transgenic RNAi in the Mut-11 mutant background [Figure 4A, Mut-11(Dcl1-IR); data not shown]. This resulted in a decrease in the levels of TOC1 and GULLIVER siRNAs (Figure 4B), which correlated with a pronounced increase in TOC1 transcripts (Figure 4C). Moreover, an 5.5-kb, presumably transposition-competent, TOC1 RNA was clearly detectable in Northern blots of the Mut-11(Dcl1-IR) strain (Figure 4C). The occurrence of TOC1 transposition was accordingly enhanced in Mut-11(Dcl1-IR). Parallel subcultures of the transgenic strain showed additional TOC1 copies integrated into the genome at a much higher frequency than those from Mut-11 (Figure 4D and data not shown), although the latter strain also accumulated, over a longer time scale, extra TOC1 insertions relative to the wild-type CC-124. However, by nuclear run-on assays, DCL1 suppression in the Mut-11 mutant background did not enhance global TOC1 transcription [Figure 5A: cf. Mut-11 and Mut-11(Dcl1-IR)].

We also examined whether downregulation of DCL1 expression had any effect on DNA or chromatin modifications associated with the silenced transposable elements. Cytosine DNA methylation was analyzed by digestion with methylation-sensitive isoschizomers. HpaII and MspI recognize the same DNA sequence (5'-CCGG-3'), but HpaII is inhibited by methylation of either cytosine, whereas MspI is sensitive only to methylation of the outer cytosine residue. Thus, if HpaII/MspI sites become methylated, the inability of the enzymes to cleave will result in the appearance of DNA fragments of higher molecular weight. By using this approach, we observed that both TOC1 and GULLIVER showed very little, if any, cytosine DNA methylation and that this modification was not affected by RNAi-mediated suppression of DCL1 (Figure 5B and data not shown). As previously reported (VAN DIJK et al. 2005), loss of MUT11 causes substantial changes in global levels of H3K4 methylation, detected by immunoblotting of whole-cell proteins with modification-specific antibodies (Figure 5C). However, DCL1 suppression had virtually no effect on the overall levels of these histone modifications (Figure 5C). We also examined directly the chromatin environment of the TOC1 retrotransposon by chromatin immunoprecipitation assays. In these experiments, we did detect somewhat reduced amounts of H3K4me1 associated with the TOC1 transcription units in CC-124(Dcl1-IR) and slightly lower levels of H3K4me2 in Mut-11(Dcl1-IR) in comparison with the respective control strains (Figure 5D). However, these are conflicting outcomes, since a decrease in H3K4me1 would promote gene activation whereas a reduction in H3K4me2 would counteract transcriptional competence (VAN DIJK et al. 2005), and they are likely the result of indirect or stochastic effects. Moreover, neither of these changes seemed to have any consequence for the global transcriptional activity of TOC1 (Figure 5A). Thus, our results suggest that DCL1 functions mainly in the post-transcriptional silencing of retrotransposons such as TOC1. In contrast, it appears to have no role in the maintenance of their transcriptional repression or, alternatively, such an activity may be redundantly compensated by one of the other Dicer paralogs. Although all transcripts with homology to the TOC1 LTRs (the sequence corresponding to the TOC1 siRNAs) appear to be targeted by the DCL1-dependent mechanism [Figure 4C: cf. Mut-11 and Mut-11(Dcl1-IR)], the key to preventing transposition is likely the suppression of the full-length TOC1 RNA.

TOC1 transcripts and siRNAs are predominantly located in the nucleus:

To gain insight into the subcellular localization of the DCL1-dependent mechanism of TOC1 suppression, we examined the distribution of TOC1 long RNAs as well as siRNAs between the nucleus and the cytoplasm. For these experiments, we used a C. reinhardtii strain lacking the cell wall (CC-3491), which facilitated the isolation of intact nuclei (KELLER et al. 1984). DAY and ROCHAIX (1991) have previously reported that TOC1 transcripts are mainly nonpolyadenylated, suggesting that they may be retained in the nucleus. Our cell fractionation analyses indicated that the majority of the long TOC1 transcripts are present in the nucleus (Figure 6A), although an abundant RNA of 2.7 kb appears to be mostly cytoplasmic. This particular RNA is shorter than the described full-length TOC1 element (DAY et al. 1988) and was not prevalent in the CC-124 strain (Figure 4C), indicating that it may have originated from a newly inserted TOC1 element in CC-3491. Interestingly, the TOC1 small RNAs were also found predominantly in the nuclear fraction (Figure 6B). The localization of the full-length 5.5-kb TOC1 RNA could not be assessed because, as is the case for CC-124, it was undetectable in Northern blots of the CC-3491 strain (Figure 6A).

View larger version (51K): In this window In a new window Download PPT slide   FIGURE 6.— Subcellular compartmentalization of TOC1 transcripts and siRNAs. (A) Northern blot analysis of TOC1 long RNAs in whole cells (WHOLE), the cytoplasmic fraction (CYT), or the nuclear fraction (NUC) from a cell-wall-less strain (CC-3491). Purified RNAs were resolved on agarose–formaldehyde gels and hybridized to a TOC1 probe. The same blot was reprobed for ACT1, a predominantly cytoplasmic transcript, and for the U6 small nuclear RNA, retained in the nucleus, to assess the effectiveness of the cell fractionation. (B) Northern blot detection of TOC1 siRNAs in the different subcellular fractions. Column-purified small RNAs were separated in a 15% denaturing polyacrylamide gel, electroblotted onto a nylon membrane, and sequentially hybridized with probes corresponding to the TOC1 LTR and the U6 small nuclear RNA.

 
As controls in our compartmentalization analyses, we examined the distribution of the actin mRNA, which is mainly a cytoplasmic transcript (RUDENKO et al. 2003), and the U6 small nuclear RNA, which is thought to remain in the nucleus (STANEK and NEUGEBAUER 2006) (Figure 6, A and B). Our findings indicate that both the long TOC1 transcripts, which are the targets of post-transcriptional degradation by a DCL1-dependent mechanism (Figure 4C), and the TOC1 small RNAs, which appear to be generated by DCL1 (Figure 4B), are mainly in the nucleus. Moreover, the DCL1 (as well as the AGO1) protein contains a predicted nuclear localization signal (Figure 3). Thus, it seems likely that the DCL1-dependent pathway of post-transcriptional retrotransposon silencing operates partly, if not entirely, in the nucleus of C. reinhardtii.

In the past few years, RNA-mediated silencing has emerged as a widespread mechanism to control gene expression in eukaryotes. Moreover, in land plants, animals, and filamentous fungi, the core RNAi machinery has significantly expanded and diversified (CARMELL et al. 2002; CERUTTI and CASAS-MOLLANO 2006; MARGIS et al. 2006; NAKAYASHIKI et al. 2006). In these organisms, various small RNA-silencing pathways often involve different sets of Dicer, AGO-Piwi, and/or RDR (when present) proteins, although there is also a considerable degree of operative overlap (OKAMURA et al. 2004; BRODERSEN and VOINNET 2006; NAKAYASHIKI et al. 2006; VAUCHERET 2006; CHAPMAN and CARRINGTON 2007). This functional diversification is just beginning to be elucidated, but the presence of duplicated RNAi components appears to allow the evolution of new gene control mechanisms that use small RNAs as sequence-specific determinants, as well as more effective strategies to counteract the action of invading genomic parasites such as viruses and transposable elements (BRODERSEN and VOINNET 2006; CERUTTI and CASAS-MOLLANO 2006; VAUCHERET 2006; CHAPMAN and CARRINGTON 2007; MATRANGA and ZAMORE 2007). In contrast, many unicellular eukaryotes, particularly those with small nuclear genomes, seem to have lost entirely the RNAi machinery or have retained only a basic set of RNAi components (CERUTTI and CASAS-MOLLANO 2006; NAKAYASHIKI et al. 2006). Indeed, it has been proposed that the deployment of small RNA pathways for spatiotemporal regulation of the transcriptome has shaped the evolution of eukaryotic genomes and contributed to the complexity of multicellular organisms (CHAPMAN and CARRINGTON 2007).

Our results indicate that the unicellular eukaryote C. reinhardtii has also undergone extensive duplication of Argonaute and Dicer polypeptides after the divergence of the green algae and land plant lineages. Moreover, Chlamydomonas contains an assortment of siRNAs as well as miRNAs (MOLNAR et al. 2007; ZHAO et al. 2007). Other unicellular organisms with relatively large nuclear genomes, such as the ciliate Tetrahymena thermophila and the social amoeba Dictyostelium discoideum, also show expansion of the core RNAi machinery and different classes of small RNAs (LEE and COLLINS 2006; HINAS et al. 2007). These observations suggest that diversification of RNA-silencing components is not limited to multicellular lineages and appears to have been a common occurrence during the evolution of eukaryotes. However, it might have provided a selective advantage, resulting in its genetic fixation only in organisms with relatively complex genomes and life cycles (regardless of cellularity).

Despite considerable advances in our mechanistic understanding of RNA-mediated silencing, the biological functions of different RNAi pathways remain largely uncharacterized in most eukaryotes. To begin addressing this issue in C. reinhardtii, we focused on the role(s) of DCL1. This protein and AGO1 are more divergent than the other paralogs encoded in the Chlamydomonas genome, suggesting, by analogy to D. melanogaster Dcr2 and Ago2 (OBBARD et al. 2006), that they may be involved in siRNA-mediated defense responses against genomic intruders. The head-to-head chromosomal arrangement of their genes also indicated that they might be coregulated and, possibly, part of the same pathway. However, RNAi-mediated suppression of DCL1 had virtually no effect on the reactivation of the retrotransposon TOC1 or the DNA element GULLIVER in a wild-type Chlamydomonas strain. In contrast, when TOC1 transcription was elevated, due to a defect in chromatin-mediated silencing in Mut-11 (ZHANG et al. 2002; VAN DIJK et al. 2005), DCL1 played a substantial role in the post-transcriptional suppression of this retrotransposon. Interestingly, the deficiency in post-transcriptional silencing caused by DCL1 suppression was not compensated by DCL2 or DCL3. Thus, our findings suggest that DCL1 (and presumably AGO1) may be part of an RNAi pathway that has specialized for the control of transposable elements in Chlamydomonas. Moreover, based on the subcellular distribution of TOC1 transcripts and siRNAs, this mechanism most likely operates (at least partly) in the nucleus. To our knowledge, viruses naturally infecting C. reinhardtii have not been isolated and, therefore, it remains to be explored whether DCL1 could also be involved in antiviral immunity. In addition, DCL1 appears to have no effect on miRNA processing since the levels of several endogenous miRNAs were not affected in the DCL1-suppressed strains (data not shown), although we cannot rule out a redundant, overlapping function of DCL2 and/or DCL3.

The TOC1 and GULLIVER transposons are present as dispersed repeats in the C. reinhardtii genome at 10–40 copies/haploid genome and appear to integrate preferentially in euchromatic regions (DAY et al. 1988; FERRIS 1989; HALL and LUCK 1995; KIM et al. 2006; MERCHANT et al. 2007). In fact, some of these transposable elements were first isolated due to the mutant phenotypes caused by their insertion into active genes (DAY et al. 1988; SCHNELL and LEFEBVRE 1993). Both TOC1 and GULLIVER seem to be controlled primarily at the transcriptional level, when cells are grown under standard laboratory conditions (this work and JEONG et al. 2002). However, their DNA does not appear to be heavily methylated, suggesting that, in Chlamydomonas, transposon siRNAs do not direct methylation of (presumably trans-) complementary DNA sequences. In addition, transcriptional silencing mediated by the MUT11 machinery and involving H3K4 monomethylation seems to be independent of RNAi. We have never detected small RNAs corresponding to single-copy transgenes silenced by this mechanism either in wild-type or in mutant strains, and DCL1 suppression had no effect on the transcriptional repression of, or on the chromatin modifications associated with, TOC1 or an RbcS2:aadA:RbcS2 transgene (this work and data not shown). However, we cannot exclude a redundant involvement of DCL2 and/or DCL3 in the maintenance of transcriptional silencing in Chlamydomonas.

In higher plants, transposons appear to be regulated by a variety of epigenetic mechanisms (FESCHOTTE et al. 2002; LIPPMAN et al. 2003; RUDENKO et al. 2003; LIPPMAN and MARTIENSSEN 2004; WOODHOUSE et al. 2006; RANGWALA and RICHARDS 2007; SLOTKIN and MARTIENSSEN 2007). In Arabidopsis, the analysis of a number of mutants revealed that DNA methylation, post-translational histone modifications as well as chromatin packaging and condensation contribute to transposon silencing, although to different extents, depending on the specific transposable element (LIPPMAN et al. 2003; RANGWALA and RICHARDS 2007; SLOTKIN and MARTIENSSEN 2007). RNAi seems to play a role in the sequence-specific targeting of transposons and other repeated sequences for heterochromatin formation (WOODHOUSE et al. 2006). Many endogenous siRNAs in Arabidopsis correspond to silenced transposable elements, and an RDR (RDR2), a Dicer homolog (DCL3), and an Argonaute (AGO4) are required for the production of transposon-derived siRNAs (LIPPMAN and MARTIENSSEN 2004; XIE et al. 2004; MATZKE et al. 2007; SLOTKIN and MARTIENSSEN 2007). However, most repressed transposons are not immediately reactivated in mutants of the RNAi machinery (LIPPMAN et al. 2003; ZILBERMAN et al. 2003; WOODHOUSE et al. 2006), even when siRNAs are completely lost (XIE et al. 2004). This suggested that RNAi may be involved in the establishment of transcriptional transposon silencing, as demonstrated for FWA transgenes in Arabidopsis (CHAN et al. 2004), but perpetuation of their repression can occur by alternative means, such as maintenance DNA methylation (LIPPMAN and MARTIENSSEN 2004; TRAN et al. 2005; MATZKE et al. 2007; SLOTKIN and MARTIENSSEN 2007). Yet a redundant role of RNAi components in the maintenance of transcriptional silencing and/or the existence of transcriptional repression mechanisms entirely independent of RNAi is also possible (WOODHOUSE et al. 2006; RANGWALA and RICHARDS 2007; SLOTKIN and MARTIENSSEN 2007). In addition, a function of the RNAi machinery in the post-transcriptional silencing of transposons remains to be examined in higher plants.

In Neurospora crassa, the RNAi machinery is not required for heterochromatin formation, H3K9 methylation, or DNA methylation (CHICAS et al. 2004; FREITAG et al. 2004), suggesting that it has no role in transcriptional silencing. However, it controls the expression of transposon and other repetitive sequences post-transcriptionally by directing the degradation of cognate transcripts (CHICAS et al. 2004; NOLAN et al. 2005). In C. elegans, silencing of the Tc1 DNA transposon also appears to be (at least in part) post-transcriptional (SIJEN and PLASTERK 2003). Likewise, our findings in the green alga Chlamydomonas indicate that the TOC1 retrotransposon can be suppressed post-transcriptionally by RNA interference. Yet this alga also relies on a DCL1-independent, transcriptional silencing mechanism(s) to control mobile genetic elements. Interestingly, chromatin-mediated repression involving the MUT11 machinery is sensitive to temperature, being much more effective at 17° than at 25° (CERUTTI et al. 1997b). Conversely, post-transcriptional gene silencing by RNAi, in both invertebrates and higher plants, appears to be more efficient at 25°–29° than at lower temperatures (FORTIER and BELOTE 2000; SZITTYA et al. 2003). We speculate that, in C. reinhardtii, multiple, partly independent silencing mechanisms may allow a more effective control of transposon mobilization over a wide range of environmental conditions.

Similarly to our observations in Chlamydomonas, the LINE1 retrotransposon in humans and the P element in Drosophila appear to be repressed by multiple pathways (SOIFER and ROSSI 2006; ARAVIN et al. 2007; JOSSE et al. 2007). Since transposable elements are infectious agents in sexual populations, hosts are under selective pressure to develop defensive functions that alleviate the fitness penalty imposed by active transposons (BESTOR 2003). Therefore, it is perhaps not surprising that eukaryotes seem to have evolved multiple epigenetic silencing mechanisms, including specialized RNAi pathways, as a defense strategy against the massive expansion of transposable elements. Moreover, redundant, partly independent pathways likely increase the genetic robustness of transposon repression against mutational and environmental perturbations.

We thank Joint Genome Institute scientists for allowing access to the Volvox and Chlorella genome sequences prior to publication and members of the Cerutti lab for critical reading of the manuscript. This work was supported by grants from the National Institutes of Health and the National Science Foundation to H.C. We also acknowledge the support of the Nebraska Experimental Program to Stimulate Competitive Research.

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession no. EU368690.

¹ These authors contributed equally to this work.

² Present address: Department of Biology, Creighton University, Omaha, NE 68178.

ALLEN, E., Z. XIE, A. M. GUSTAFSON and J. C. CARRINGTON, 2005 microRNA-directed phasing during trans-acting siRNA biogenesis in plants. Cell 121: 207–221.[CrossRef][Medline]

AOKI, K., H. MORIGUCHI, T. YOSHIOKA, K. OKAWA and H. TABARA, 2007 In vitro analyses of the production and activity of secondary small interfering RNAs in C. elegans. EMBO J. 26: 5007–5019.[CrossRef][Medline]

ARAVIN, A. A., G. J. HANNON and J. BRENNECKE, 2007 The Piwi-piRNA pathway provides an adaptive defense in the transposon arms race. Science 318: 761–764.[Abstract/Free Full Text]

BARTEL, D. P., 2004 MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116: 282–297.

BAULCOMBE, D., 2004 RNA silencing in plants. Nature 431: 356–363.[CrossRef][Medline]

BAUMBERGER, N., and D. C. BAULCOMBE, 2005 Arabidopsis ARGONAUTE1 is an RNA slicer that selectively recruits microRNAs and short interfering RNAs. Proc. Natl. Acad. Sci. USA 102: 11928–11933.[Abstract/Free Full Text]

BERNSTEIN, E., A. A. CAUDY, S. M. HAMMOND and G. J. HANNON, 2001 Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409: 363–366.[CrossRef][Medline]

BESTOR, T. H., 2003 Cytosine methylation mediates sexual conflict. Trends Genet. 19: 185–190.[CrossRef][Medline]

BRODERSEN, P., and O. VOINNET, 2006 The diversity of RNA silencing pathways in plants. Trends Genet. 22: 268–280.[CrossRef][Medline]

CARMELL, M. A., Z. XUAN, M. Q. ZHANG and G. J. HANNON, 2002 The Argonaute family: tentacles that reach into RNAi, developmental control, stem cell maintenance, and tumorigenesis. Genes Dev. 16: 2733–2742.[Free Full Text]

CASAS-MOLLANO, J. A., K. VAN DIJK, J. EISENHART and H. CERUTTI, 2007 SET3p monomethylates histone H3 on lysine 9 and is required for the silencing of tandemly repeated transgenes in Chlamydomonas. Nucleic Acids Res. 35: 939–950.[Abstract/Free Full Text]

CERUTTI, H., 2003 RNA interference: Traveling in the cell and gaining functions? Trends Genet. 19: 39–46.[CrossRef][Medline]

CERUTTI, H., and J. A. CASAS-MOLLANO, 2006 On the origin and functions of RNA-mediated silencing: from protists to man. Curr. Genet. 50: 81–99.[CrossRef][Medline]

CERUTTI, H., A. M. JOHNSON, N. W. GILLHAM and J. E. BOYNTON, 1997a A eubacterial gene conferring spectinomycin resistance on Chlamydomonas reinhardtii: integration into the nuclear genome and gene expression. Genetics 145: 97–110.[Abstract]

CERUTTI, H., A. M. JOHNSON, N. W. GILLHAM and J. E. BOYNTON, 1997b Epigenetic silencing of a foreign gene in nuclear transformants of Chlamydomonas. Plant Cell 9: 925–945.[Abstract/Free Full Text]

CERUTTI, L., N. MIAN and A. BATEMAN, 2000 Domains in gene silencing and cell differentiation proteins: the novel PAZ domain and redefinition of the Piwi domain. Trends Biochem. Sci. 25: 481–482.[CrossRef][Medline]

CHAN, S. W., D. ZILBERMAN, Z. XIE, L. K. JOHANSEN, J. C. CARRINGTON et al., 2004 RNA silencing genes control de novo DNA methylation. Science 303: 1336.[Free Full Text]

CHAPMAN, E. J., and J. C. CARRINGTON, 2007 Specialization and evolution of endogenous small RNA pathways. Nat. Rev. Genet. 8: 884–896.[CrossRef][Medline]

CHICAS, A., C. COGONI and G. MACINO, 2004 RNAi-dependent and RNAi-independent mechanisms contribute to the silencing of RIPed sequences in Neurospora crassa. Nucleic Acids Res. 32: 4237–4243.[Abstract/Free Full Text]

COGONI, C., and G. MACINO, 2000 Post-transcriptional gene silencing across kingdoms. Curr. Opin. Genet. Dev. 10: 638–643.[CrossRef][Medline]

DAY, A., and J. D. ROCHAIX, 1991 Structure and inheritance of sense and anti-sense transcripts from a transposon in the green alga Chlamydomonas reinhardtii. J. Mol. Biol. 218: 273–291.[CrossRef][Medline]

DAY, A., M. SCHIRMER-RAHIRE, M. R. KUCHKA, S. P. MAYFIELD and J. D. ROCHAIX, 1988 A transposon with an unusual arrangement of long terminal repeats in the green alga Chlamydomonas reinhardtii. EMBO J. 7: 1917–1927.[Medline]

DING, S.-W., and O. VOINNET, 2007 Antiviral immunity directed by small RNAs. Cell 130: 413–426.[CrossRef][Medline]

DLAKIC, M., 2006 DUF283 domain of Dicer proteins has a double-stranded RNA-binding fold. Bioinformatics 22: 2711–2714.[Abstract/Free Full Text]

FERRIS, P. J., 1989 Characterization of a Chlamydomonas transposon, Gulliver, resembling those in higher plants. Genetics 122: 363–377.[Abstract/Free Full Text]

FESCHOTTE, C., N. JIANG and S. R. WESSLER, 2002 Plant transposable elements: where genetics meets genomics. Nat. Rev. Genet. 3: 329–341.[CrossRef][Medline]

FIRE, A., S. XU, M. K. MONTGOMERY, S. A. KOSTAS, S. E. DRIVER et al., 1998 Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391: 806–811.[CrossRef][Medline]

FORTIER, E., and J. M. BELOTE, 2000 Temperature-dependent gene silencing by an expressed inverted repeat in Drosophila. Genesis 26: 240–244.[CrossRef][Medline]

FREITAG, M., D. W. LEE, G. O. KOTHE, R. J. PRATT, R. ARAMAYO et al., 2004 DNA methylation is independent of RNA interference in Neurospora. Science 304: 1939.[Free Full Text]

HALL, J. L., and D. J. L. LUCK, 1995 Basal body-associated DNA: in situ studies in Chlamydomonas reinhardtii. Proc. Natl. Acad. Sci. USA 92: 5129–5133.[Abstract/Free Full Text]

HARRIS, E. H., 1989 The Chlamydomonas Sourcebook. Academic Press, San Diego.

HINAS, A., J. REIMEGÅRD, E. G. WAGNER, W. NELLEN, V. R. AMBROS et al., 2007 The small RNA repertoire of Dictyostelium discoideum and its regulation by components of the RNAi pathway. Nucleic Acids Res. 35: 6714–6726.[Abstract/Free Full Text]

IBRAHIM, F., J. ROHR, W-J. JEONG, J. HESSON and H. CERUTTI, 2006 Untemplated oligoadenylation promotes degradation of RISC-cleaved transcripts. Science 314: 1893.[Abstract/Free Full Text]

JANOWSKI, B. A., S. T. YOUNGER, D. B. HARDY, R. RAM, K. E. HUFFMAN et al., 2007 Activating gene expression in mammalian cells with promoter-targeted duplex RNAs. Nat. Chem. Biol. 3: 166–173.[CrossRef][Medline]

JEONG, B. R., D. WU-SCHARF, C. ZHANG and H. CERUTTI, 2002 Suppressors of transcriptional transgenic silencing in Chlamydomonas are sensitive to DNA-damaging agents and reactivate transposable elements. Proc. Natl. Acad. Sci. USA 99: 1076–1081.[Abstract/Free Full Text]

JOSSE, T., L. TEYSSET, A. L. TODESCHINI, C. M. SIDOR, D. ANXOLABEHERE et al., 2007 Telomeric trans-silencing: an epigenetic repression combining RNA silencing and heterochromatin formation. PLoS Genet. 3: 1633–1643.[Medline]

KELLER, L. R., J. A. SCHLOSS, C. D. SILFLOW and J. L. ROSENBAUM, 1984 Transcription of - and β-tubulin genes in vitro in isolated Chlamydomonas reinhardtii nuclei. J. Cell Biol. 98: 1138–1143.[Abstract/Free Full Text]

KIM, K. S., S. KUSTU and W. INWOOD, 2006 Natural history of transposition in the green alga Chlamydomonas reinhardtii: use of the AMT4 locus as an experimental system. Genetics 173: 2005–2019.[Abstract/Free Full Text]

KINDLE, K. L., 1990 High-frequency nuclear transformation of Chlamydomonas reinhardtii. Proc. Natl. Acad. Sci. USA 87: 1228–1232.[Abstract/Free Full Text]

KUMAR, S., K. TAMURA and M. NEI, 2004 MEGA3: Integrated software for Molecular Evolutionary Genetics Analysis and sequence alignment. Brief. Bioinform. 2: 150–163.

LEE, S. R., and K. COLLINS, 2006 Two classes of endogenous small RNAs in Tetrahymena thermophila. Genes Dev. 20: 28–33.[Abstract/Free Full Text]

LI, L. C., S. T. OKINO, H. ZHAO, D. POOKOT, R. F. PLACE et al., 2006 Small dsRNAs induce transcriptional activation in human cells. Proc. Natl. Acad. Sci. USA 103: 17337–17342.[Abstract/Free Full Text]

LIPPMAN, Z., and R. MARTIENSSEN, 2004 The role of RNA interference in heterochromatic silencing. Nature 431: 364–370.[CrossRef][Medline]

LIPPMAN, Z., B. MAY, C. YORDAN, T. SINGER and R. MARTIENSSEN, 2003 Distinct mechanisms determine transposon inheritance and methylation via small interfering RNA and histone modification. PLoS Biol. 1: E67.[Medline]

LIU, J., M. A. CARMELL, F. V. RIVAS, C. G. MARSDEN, J. M. THOMSON et al., 2004 Argonaute2 is the catalytic engine of mammalian RNAi. Science 305: 1437–1441.[Abstract/Free Full Text]

MA, J. B., K. YE and D. J. PATEL, 2004 Structural basis for overhang-specific small interfering RNA recognition by the PAZ domain. Nature 429: 318–322.[CrossRef][Medline]

MARGIS, R., A. F. FUSARO, N. A. SMITH, S. J. CURTIN, J. M. WATSON et al., 2006 The evolution and diversification of Dicers in plants. FEBS Lett. 580: 2442–2450.[CrossRef][Medline]

MATRANGA, C., and P. ZAMORE, 2007 Small silencing RNAs. Curr. Biol. 17: R789–R793.[CrossRef][Medline]

MATZKE, M. A., and J. A. BIRCHLER, 2005 RNAi-mediated pathways in the nucleus. Nat. Rev. Genet. 6: 24–35.[CrossRef][Medline]

MATZKE, M., T. KANNO, B. HUETTEL, L. DAXINGER and A. J. MATZKE, 2007 Targets of RNA-directed DNA methylation. Curr. Opin. Plant Biol. 10: 512–519.[CrossRef][Medline]

MEISTER, G., and T. TUSCHL, 2004 Mechanisms of gene silencing by double-stranded RNA. Nature 431: 343–349.[CrossRef][Medline]

MEISTER, G., M. LANDTHALER, A. PATKANIOWSKA, Y. DORSETT, G. TENG et al., 2004 Human Argonaute2 mediates RNA cleavage targeted by miRNAs and siRNAs. Mol. Cell 15: 185–197.[CrossRef][Medline]

MERCHANT, S. S., S. E. PROCHNIK, O. VALLON, E. H. HARRIS, S. J. KARPOWICZ et al., 2007 The Chlamydomonas genome reveals the evolution of key animal and plant functions. Science 318: 245–251.[Abstract/Free Full Text]

MOLNAR, A., F. SCHWACH, D. J. STUDHOLME, E. C. THUENEMANN and D. C. BAULCOMBE, 2007 miRNAs control gene expression in the single-cell alga Chlamydomonas reinhardtii. Nature 447: 1126–1129.[CrossRef][Medline]

NAKAYASHIKI, H., N. KADOTANI and S. MAYAMA, 2006 Evolution and diversification of RNA silencing proteins in fungi. J. Mol. Evol. 63: 127–135.[CrossRef][Medline]

NOLAN, T., L. BRACCINI, G. AZZALIN, A. DE TONI, G. MACINO et al., 2005 The post-transcriptional gene silencing machinery functions independently of DNA methylation to repress a LINE1-like retrotransposon in Neurospora crassa. Nucleic Acids Res. 33: 1564–1573.[Abstract/Free Full Text]

OBBARD, D. J., F. M. JIGGINS, D. L. HALLIGAN and T. J. LITTLE, 2006 Natural selection drives extremely rapid evolution in antiviral RNAi genes. Curr. Biol. 16: 580–585.