MicroRNAs (miRNAs) are 20–24 nt non-coding RNAs that play important regulatory roles in a broad range of eukaryotes by pairing with mRNAs to direct post-transcriptional repression. The mechanistic details of miRNA-mediated post-transcriptional regulation have been well documented in multicellular model organisms. However, this process remains poorly studied in algae such as Chlamydomonas reinhardtii, and specific features of miRNA biogenesis, target mRNA recognition and subsequent silencing are not well understood. In this study, we report on the characterization of a Chlamydomonas miRNA, cre-miR1174.2, , which is processed from a near-perfect hairpin RNA. Using Gaussia luciferase (gluc) reporter genes, we have demonstrated that cre-miR1174.2 is functional in Chlamydomonas and capable of triggering site-specific cleavage at the center of a perfectly complementary target sequence. A mismatch tolerance test assay, based on pools of transgenic strains, revealed that target hybridization to nucleotides of the seed region, at the 5′ end of an miRNA, was sufficient to induce moderate repression of expression. In contrast, pairing to the 3′ region of the miRNA was not critical for silencing. Our results suggest that the base-pairing requirements for small RNA-mediated repression in C. reinhardtii are more similar to those of metazoans compared with the extensive complementarity that is typical of land plants. Individual Chlamydomonas miRNAs may potentially modulate the expression of numerous endogenous targets as a result of these relaxed base-pairing requirements.
MicroRNAs (miRNAs) are endogenous 20–24 nt non-coding RNAs that regulate gene expression by base pairing to mRNAs and direct their degradation or inhibit productive translation. To date, thousands of miRNAs have been found in a diverse range of eukaryotes and shown to have critical roles in diverse biological processes, including development, proliferation and stress responses (Huntzinger and Izaurralde, 2011; Pasquinelli, 2012). The canonical miRNA biogenesis pathway begins with the transcription of long RNAs that fold back to form distinctive hairpin structures. These transcripts are subsequently processed by protein complexes, which contain RNase III endonucleases and double-stranded RNA binding proteins, into miRNA/miRNA* duplexes with 2 nt 3′ overhangs. One of the strands of each miRNA/miRNA* duplex (proper miRNA) is loaded into the RNA-induced silencing complex (RISC) and serves as a guide to identify complementary mRNAs targeted for silencing (Voinnet, 2009; Axtell et al., 2011; Yang and Lai, 2011). Plants and animals share several features of the miRNA biogenesis pathway, but there are also appreciable mechanistic differences between these organisms (Flynt and Lai, 2008; Voinnet, 2009; Axtell et al., 2011; Yang and Lai, 2011).
The biogenesis of plant miRNAs has been studied primarily in Arabidopsis thaliana. Plant primary miRNAs (pri-miRNAs) are generally transcribed by RNA polymerase II and subjected to post-transcriptional processing, including 5′ capping, intron splicing and 3′ polyadenylation (Bartel, 2004; Xie et al., 2005). Local secondary structures of long pri-miRNAs are recognized by the DCL1/HYL1/Serrate protein complex, and are usually cleaved at the double-stranded bases to release hairpin structures as precursor miRNAs (pre-miRNAs), which in turn are cleaved again by DCL1 to generate miRNA/miRNA* duplexes (Han et al., 2004; Kurihara and Watanabe, 2004; Kurihara et al., 2006; Lobbes et al., 2006). This sequential processing may occur in two directions: stem-to-loop or loop-to-base (Naqvi et al., 2012). However, the processing of most plant pri-miRNAs appears to follow the stem-to-loop orientation, and only pri-miR159 and pri-miR319 are known to be processed in the reverse orientation (Bartel, 2004; Addo-Quaye et al., 2009; Bologna et al., 2009).
Most land plant miRNAs display perfect or near-perfect complementarity to their targets. Perfect pairing usually triggers endonucleolytic target cleavage (slicing) at the center of the miRNA/mRNA hybrid, and subsequent degradation of the target transcript (Llave et al., 2002; Voinnet, 2009). In contrast, in animals, mature miRNAs generally form imperfect hybrids with target transcripts, leading to reduced protein expression through translation inhibition and/or accelerated exonucleolytic RNA decay. Most animal miRNAs display perfect pairing in the seed region, between nucleotides 2 and 8, and imperfect hybridization in the central region, between nucleotides 9 and 12 (Lewis et al., 2005; Bartel, 2009). Because of this limited sequence complementarity, individual animal miRNAs appear to have multiple target transcripts (Bartel, 2009), whereas individual plant miRNAs are predicted to have a reduced number of target mRNAs given their requirement for more extensive sequence complementarity (Schwab et al., 2005; Voinnet, 2009). Nonetheless, systematic mutagenesis studies of miRNA/target site hybrids have revealed that the miRNA seed region is more critical than the 3′ region for target recognition in A. thaliana (Mallory et al., 2004). Moreover, plant and algal small RNAs also induce translational repression of perfectly complementary target mRNAs without, or with only minimal, transcript destabilization, thereby adding to the mechanistic complexity of miRNA action (Brodersen et al., 2008; Ma et al., 2013).
Small RNA-mediated silencing remains poorly understood in the unicellular green alga Chlamydomonas reinhardtii (Ibrahim et al., 2006, 2010; Molnar et al., 2007; Zhao et al., 2007; Casas-Mollano et al., 2008; Cerutti et al., 2011; Ma et al., 2013). Considering the physiological and gene complement similarities between C. reinhardtii and A. thaliana (Merchant et al., 2007), many features of the RNA interference (RNAi) pathway(s) are expected to be conserved between these organisms. However, the chlorophytan green algae lineage diverged from land plants over one billion years ago, which is similar to the time of divergence between yeast and mouse (Yoon et al., 2004; Merchant et al., 2007). Therefore, it is possible that differences in the basic mechanisms of miRNA biogenesis and function do exist between green algae and higher plants. In this respect, the present study shows that release of an miRNA/miRNA* duplex in C. reinhardtii appears to occur from the loop of a near-perfect hairpin precursor (i.e. at least partial loop-to-base processing). In addition, pairing of the miRNA seed region alone was sufficient to cause moderate repression of a target transcript. These results suggest that certain aspects of miRNA biogenesis and mode of action in green algae differ somewhat from those in land plants and may resemble RNAi mechanisms operating in metazoans.
An miRNA, named cre-miR1174.2. is processed from the transcript of the Cre10.g444300 gene
By deep sequencing of small RNAs in C. reinhardtii(Ibrahim et al., 2010), we identified a 21 nt putative miRNA, which we named cre-miR1174.2, A homology search against the C. reinhardtii genome (http://www.phytozome.net/chlamy.php), using the cre-miR1174.2 sequence as the query, revealed a single precise match in a predicted gene, Cre10.g444300. The RNAfold secondary structure analysis tool (http://rna.tbi.univie.ac.at/cgi-bin/RNAfold.cgi) predicted that nucleotides 789–922 from the 5′ end of the Cre10.g444300 transcript form a hairpin with a 64 bp stem (including the cre-miR1174.2 sequence) and a 6 nt loop (Figures 1a and 2). Northern hybridization analyses showed that the Cre10.g444300 transcript was not detected in the wild-type CC-124 strain, probably as a result of fast processing of the pri-miRNA to mature miRNA (Figure 1b). However, we observed a shorter RNA of approximately 0.7 kb in length (Figure 1b) that presumably corresponded to the 5′ end of the pri-miRNA after Dicer processing. In contrast, a band of approximately 2.2 kb in size that corresponded to the predicted Cre10.g444300 transcript was clearly detected in an RNAi-defective mutant, Mut-20, which was isolated from a library of insertional mutants in the Maa7-IR44s background (Ma et al., 2013) (Figure 1b). The specific molecular alteration in Mut-20 remains uncharacterized; however, as the mutant showed greatly reduced levels of miRNA (Figure 1c), it is possible that a defect in miRNA biogenesis allowed detection of the full-length Cre10.g444300 RNA.
The data obtained by deep sequencing of small RNAs also supported processing of cre-miR1174.2 from the Cre10.g444300 hairpin because both miRNA and miRNA* matching sequences were identified; however, as expected, miRNA* reads were detected at a much lower abundance (Figure 2). Moreover, the Cre10.g444300 hairpin appeared to code for two distinct miRNAs, one on each arm, which show 5′ end homogeneity in over 98% of the reads and marked asymmetry in the accumulated reads (Figure 2, miRNA versus miRNA* reads). These are all expected features of canonical miRNAs (Tarver et al., 2012). We also performed 5′ RACE analyses on polyadenylated RNA from the wild-type CC-124 strain to identify cleavage sites within the Cre10.g444300 hairpin structure. A Cre10.g444300-specific reverse primer was designed to anneal approximately 100 nt downstream of the 3′ end of the hairpin, and PCR amplicons differing in size by approximately 40 bp were obtained. Cloning and subsequent sequencing of 27 amplicons revealed that the predominant sites of 5′ termini truncation (putative cleavage sites) corresponded to the ends of the cre-miR1174.2 sequence in the 3′ arm of the Cre10.g444300 hairpin (Figure 2).
Chlamydomonas Cre10.g444300 is predicted to encode a 189 amino acid polypeptide of unknown function and lacks homology to proteins in any other organism, even closely related algal species (http://www.phytozome.net/chlamy.php). Thus, it is tempting to speculate that Cre10.g444300 actually produces a non-coding RNA, probably transcribed by RNA polymerase II as for pri-miRNAs in other eukaryotes, that serves as a precursor for the biogenesis of miRNAs.
Reporter transcripts containing cre-miR1174.2 target sites are repressed in Chlamydomonas, probably by cre-miR1174.2 induced site-specific RNA cleavage
Currently, 19 526 protein-coding transcripts, including splice variants, are predicted to be encoded in the Chlamydomonas genome (http://www.phytozome.net/chlamy.php). None of the transcript models, except for Cre10.g444300, contained a sequence that was perfectly complementary to cre-miR1174.2 however, several predicted mRNAs formed imperfect hybrids resembling the typical metazoan miRNA–target interaction. Thus, a series of four Gaussia luciferase (gluc) reporter genes (Ruecker et al., 2008; Shao and Bock, 2008) were introduced into C. reinhardtii to demonstrate the ability of cre-miR1174.2 to induce target RNA repression. The 3′ UTRs of these reporters showed no homology to cre-miR1174.2 or one, four or eight copies (1 ×, 4 × or 8 ×, respectively) of a sequence that was perfectly complementary to cre-miR1174.2 (Figure S1). Genomic PCR was performed to screen for transgenic strains that integrated the entire transforming DNA (Table S1), and individual transgenic lines were examined for gluc activity. Strains transformed with gluc-aph7 (no cre-miR1174.2 target site) showed variable luciferase activities, presumably as a result of integration site-dependent position effects (Figure S2). Nonetheless, approximately 83% of the strains tested exhibited adequate gluc activity, i.e. higher than 1000 relative light units (RLU) (Figure S2). In contrast, the majority of Chlamydomonas transformants containing reporter constructs with cre-miR1174.2 target sites showed low gluc activity (Figure S2). For instance, approximately 50% of the gluc(1 × )-aph7-transformed strains exhibited gluc activity of <1000 RLU, and none of the gluc(4 × )-aph7 or gluc(8 × )-aph7-transformed strains exhibited luciferase activity above 1000 RLU (Figure S2). These results suggested that Chlamydomonas cre-miR1174.2 may trigger silencing of gluc transgenes containing cre-miR1174.2 target sites in their transcripts.
To confirm that repression of the gluc reporters was indeed miRNA-mediated, we selected a single-copy gluc-aph7-transformed strain, termed gluc-10, and a single copy gluc(1 × )-aph7-transformed strain, termed gluc(1 × )-10, for further analyses. We then crossed these strains with the Mut-20 RNAi-defective mutant. Interestingly, gluc expression from the gluc-aph7 transgene in the Mut-20 background was comparable to that observed in the wild-type gluc-10 strain (Figure 3a–c). In contrast, gluc(1 × ) expression from the gluc(1 × )-aph7 transgene was substantially increased in the Mut-20 background compared with that observed in the gluc(1 × )-10 strain (Figure 3a–c). As Mut-20 had virtually undetectable levels of cre-miR1174.2 (Figure 3c), these observations are consistent with direct involvement of this miRNA in repression of the gluc(1 × )-aph7 transgene in the wild-type background, with a marked reduction of transcript (Figure 3c) and protein (Figure 3b) levels.
In both animals and land plants, RISC-mediated, site-specific target cleavage (slicing) occurs predominantly between the nucleotides pairing to positions 10 and 11 from the 5′ end of a guide small RNA (Elbashir et al., 2001; Llave et al., 2002). Similar miRNA-mediated site-specific cleavage has been reported for some endogenous target transcripts in Chlamydomonas (Molnar et al., 2007; Zhao et al., 2007). Thus, we performed 5′ RACE analyses to verify whether cre-miR1174.2 induced slicing of the gluc(1 × ) transcript. As expected for RISC-triggered RNA cleavage, the assays revealed that the gluc(1 × ) mRNA was frequently truncated at the center of the cre-miR1174.2 target site (Figure 3d). In addition, a truncated transcript, probably corresponding to the RISC-cleaved 5′ end of the gluc(1 × ) mRNA, was detected in Northern hybridization analyses of the gluc(1 × )-10 strain (Figure 3c). Taken together, these results strongly suggest that cre-miR1174.2 is a functional miRNA in Chlamydomonas, and that it mediates site-specific cleavage of a perfectly complementary target transcript. Moreover, the gluc reporter system appeared to serve as a good indicator to estimate the potency of miRNA-mediated silencing.
The efficiency of cre-miR1174.2-mediated gene silencing may be examined using pools of transgenic strains in Chlamydomonas
Base pairing of the miRNA seed region (positions 2–8 from the 5′ end of an miRNA) in animals is critical for target recognition and repression (Bartel, 2009; Pasquinelli, 2012). In contrast, most evidence indicates that miRNAs in land plants require more extensive pairing to their targets (Schwab et al., 2005; Voinnet, 2009). Mismatch tolerance tests have revealed different requirements with regard to the degree of complementarity and length of base pairing among miRNA/mRNA hybrids in plants and animals (Kloosterman, 2004; Mallory et al., 2004; Brennecke et al., 2005). However, similar analyses have not been performed in algae. To address this, we first developed a reliable method to estimate the efficiency of miRNA-mediated repression using the gluc reporter genes. As already discussed, stable transgenic strains integrated luciferase genes in various genomic locations and displayed variable expression levels (Figure S2). However, as previously demonstrated (Lodha et al., 2008), the position effects may be averaged by analyzing pools of Chlamydomonas transformants instead of individual transgenic lines.
For a comparative analysis of miRNA-mediated repression, pools of transgenic strains were generated by transformation of C. reinhardtii with gluc reporter constructs linked on the same plasmid to a gp64 epitope-tagged aph7 selectable marker (Figure S3). Transgenic pools were prepared by electroporating approximately 2.5 × 107 CC-3491 cells, followed by selection in liquid medium containing hygromycin B (see Experimental procedures). A single electroporation generated over 500 hygromycin B-resistant transformants and, according to previous estimations (Lodha et al., 2008), the size of these transgenic pools is acceptable for comparative analyses of gluc expression. The gluc transgene gluc-tagaph7, which lacks an miRNA target site, was used to determine baseline levels of luciferase activity, protein abundance and mRNA quantity (Figure S4). We also used gluc transgenes containing reversed or randomized cre-miR1174.2 target sequences as controls; these exhibited gluc expression levels equivalent to those of gluc-tagaph7 (Figure S4). In contrast, expression of the cre-miR1174.2-targeted gluc transgenes gluc(1 × )-tagaph7, gluc(4 × )-tagaph7 and gluc(8 × )-tagaph7 was substantially lower (Figure S4). Moreover, the degree of repression appeared to correlate positively with the number of repeated cre-miR1174.2 target sites in the gluc transcripts (Figure S4). Experimental controls revealed that gp64-tagged aph7 was expressed equally in all transgenic pools (Figure S4), and the mean level of nuclear-integrated gluc transgenes was also comparable among the examined pools (Figure S4). Silencing of gluc reporter genes containing target sites for other Chlamydomonas miRNAs (miR910, miR912 and candidate 82; Molnar et al., 2007; Zhao et al., 2007) was also observed in pools of transgenic strains (Figure S5). Therefore, our results demonstrate that the degree of miRNA-mediated gene repression may be examined using transgenic pools in C. reinhardtii, thereby providing a useful tool to perform mismatch tolerance tests.
Base pairing of a target site to positions 2–12 from the 5′ end of cre-miR1174.2 is important for gene repression
Ten dinucleotide substitutions were introduced into the cre-miR1174.2 target site in the gluc reporter gene to determine the possible negative effects of mismatches on cre-miR1174.2-mediated repression (Figure S6). Pairs of nucleotide mismatches within the region hybridizing to positions 1–12 of cre-miR1174.2 resulted in increased reporter expression levels. This was determined by both luciferase activity and immunoblot assays in comparison with a perfectly complementary control (Figure 4a,b). In particular, dinucleotide mismatches in the target site corresponding to positions 1–6 in cre-miR1174.2 led to noticeably higher expression levels relative to other mismatches (Figure 4a,b). In contrast, repression of the gluc reporter gene was not significantly affected by substitutions in the region pairing with the 3′ end of the miRNA from nucleotide positions 13–20 (Figure 4a,b).
To examine reporter mRNA levels by real-time PCR, we designed a gluc primer set that amplifies a fragment located immediately upstream of the cre-miR1174.2 target site. However, this fragment may be amplified from both the full-length gluc transcript and from a truncated RNA that was presumably generated by slicing (Figure 3c), leading to erroneous quantification of the intact transcript. To avoid this complication, we used an oligo(dT)18 primer for reverse transcription and generation of cDNA samples for PCR amplification, which prevented copying of the non-polyadenylated truncated RNA. Interestingly, the measured amounts of gluc mRNA (Figure 4c) broadly coincided with the relative levels of gluc activity for the various reporters, with the exception of gluc(9·10). In this transgenic pool, polyadenylated gluc mRNA accumulated to higher levels than expected relative to the observed gluc protein activity and amount (Figure 4a–c). Mismatches flanking positions 10 and 11 are known to have an inhibitory effect on RISC slicing activity (Brodersen and Voinnet, 2009); therefore, the increased gluc mRNA accumulation in gluc(9·10) was considered to result from a defect in cre-miR1174.2-mediated target cleavage, which may have been compensated for by some degree of translational repression. The gp64-tagged aph7 control showed virtually identical expression in all transgenic pools (Figure 4b), and the mean level of the integrated gluc transgenes was also comparable (Figure 4d).
These observations suggest that hybridization of a target transcript to positions 2–12 of an miRNA is an important requirement for RNAi-mediated silencing in C. reinhardtii. The position 1 of the target sequence was also mutated, but complementarity at this position is probably inconsequential for repression because the 5′ terminal nucleotide of miRNAs is bound in a pocket of the Argonaute protein and is unavailable for RNA pairing (Boland et al., 2011). Our analysis also indicated that, as expected, mismatches flanking the cleavage site may be detrimental for slicing and transcript degradation, but target mRNAs may still be silenced by translational repression.
Seed matching alone is sufficient to induce moderate miRNA-mediated repression of target gene expression
In land plants, miRNA target selection only tolerates one mismatch in the region complementary to positions 2–12 of a given miRNA, provided that this does not occur at positions 10 and 11 (Schwab et al., 2005). In addition, a total of up to four mismatches throughout the length of the miRNA, but no more than two mismatches in a row, are tolerated (Schwab et al., 2005). Many miRNA targets have been successfully predicted in land plants by following these rules. However, there appeared to be some exceptions to these rules in terms of the detected activity of cre-miR1174.2 towards target reporters in the green alga Chlamydomonas. For instance, cre-miR1174.2 induced silencing when the miRNA/mRNA hybrid had dinucleotide mismatches in the region flanking the cleavage site (Figure 4). These observations imply that limited complementarity to a target site may be sufficient to induce cre-miR1174.2-mediated repression in C. reinhardtii. To address this question, the expression of a gluc–cre-miR1174.2 seed reporter containing a target site complementary to positions 1–8 of cre-miR1174.2 was tested in transgenic pools (Figure 5a). Moderately reduced expression levels were observed for the gluc–cre-miR1174.2 seed transgenic pool compared with the gluc-tagaph7 transgenic pool control that lacks a cre-miR1174.2 target site (Figure 5b–d), despite virtually identical amounts of integrated gluc transgenes in both transgenic pools. A similar result was obtained when analyzing the interaction of Chlamydomonas miR910 with its target site (Figure S7). These results suggested that pairing to the seed region alone may be sufficient to induce mild cre-miR1174.2 and miR910-mediated repression in C. reinhardtii.
To date, 50 Chlamydomonas miRNA sequences have been deposited in miRBase (http://www.mirbase.org; Molnar et al., 2007; Zhao et al., 2007). However, a recent re-evaluation of these miRNAs suggested that many may actually correspond to endogenous small interfering RNAs (e.g. multiple products from the same long hairpin without corresponding star sequences and/or lacking stable 5′ processing; Tarver et al., 2012). Two reports indicated that only five (Nozawa et al., 2012) or six (Tarver et al., 2012) currently recognized miRNAs in C. reinhardtii may be genuine. In addition, no miRNA precursor has been experimentally characterized in this alga (Molnar et al., 2007; Zhao et al., 2007). In this study, we provide evidence that Cre10.g444300 may actually encode a pri-miRNA transcript, which is only detected in an RNAi-defective mutant (Figure 1). Based on established criteria for miRNA annotation, including evidence of expression of both miRNA and miRNA*, 2 nt 3′ overhangs in the proposed miRNA/miRNA* duplex, and consistent 5′ processing of candidate miRNAs (Bartel, 2004; Axtell et al., 2011; Tarver et al., 2012), the Cre10.g444300 hairpin appears to serve as the precursor for two Chlamydomonas miRNAs: cre-miR1174.2 produced from the 3′ arm and cre-miR1174.1 produced from the 5′ arm (Figure 2). Moreover, cre-miR1174.2 was shown to function as a bona fide miRNA because reporter transgenes containing perfectly complementary cre-miR1174.2 target sequences in their 3′ UTRs were efficiently silenced in C. reinhardtii (Figure 3 and Figure S4).
5′ RACE analyses revealed predominant sites of 5′ termini transcript truncation (putative cleavage sites) corresponding to the ends of candidate miRNA or miRNA* sequences in the Cre10.g444300 hairpin 3′ arm (Figure 2). Because 5′ RACE products contain sequences downstream of the hairpin (approximately 100 bp to the annealing site for the reverse primer), a major PCR fragment truncated at the 3′ end of the cre-miR1174.2 sequence (at the base of the double-stranded RNA stem) is expected if Dicer processing began from the base of the hairpin structure. In contrast, detection of several putative cleavage intermediates on the 3′ arm, including one of particular significance at the 5′ end of the cre-miR1174.2 sequence (Figure 2), suggested that processing may start from the loop of the hairpin, as previously illustrated for biogenesis of plant miR319 and miR159 (Bologna et al., 2009). However, we cannot rule out the possibility that processing of the hairpin structure may start from either end or in the middle of the double-stranded stem.
Dual luciferase reporter assays have frequently been used in cultured animal cells to study transcription factors, mRNA processing and miRNA-mediated repression (Dougherty and Sanders, 2005; Lynch et al., 2007; Lee et al., 2008). In this assay, the expression of two independent luciferase reporters that require different substrates is examined. Firefly luciferase and Renilla luciferase are typically used, with one reflecting the effect of a specific experimental condition and the other serving as a control for transfection efficiency (Dougherty and Sanders, 2005). To the best of our knowledge, a similar system is not currently available in green algae. In C. reinhardtii, two coelenterazine luciferases, from Renilla reniformis and Gaussia princeps, have been used for transgene expression analyses (Fuhrmann et al., 2004; Ruecker et al., 2008; Shao and Bock, 2008), but no luciferin-luciferase has been adapted for expression from the nuclear genome. Therefore, we developed a semi-quantitative gene expression assay combining Gaussia luciferase as the experimental reporter and gp64-tagged aph7 as the control, to test the pairing requirements for miRNA-mediated gene silencing in this alga. Gene expression was examined in pools of transgenic strains that were generated by electroporation and selected in liquid medium containing hygromycin B. In addition, the plasmid gluc-tagaph7 (Figure S3) was designed for easy replacement of the PsaD promoter and terminator sequences by digestion at unique restriction enzyme sites to facilitate utilization of this strategy for other studies in Chlamydomonas such as promoter strength testing.
Numerous experimental and computational analyses in metazoans have demonstrated that miRNA target recognition is achieved predominantly through seed pairing (Bartel, 2009). In contrast, most land plant miRNAs pair with perfect or near-perfect complementarity to their targets (Schwab et al., 2005; Voinnet, 2009). Because of these distinctive features regarding the extent of miRNA–target complementarity, approximately 30% of transcripts are predicted to contain miRNA-binding sites in mammals, whereas fewer than 1% of the transcripts in a land plant such as Arabidopsis are envisaged to be miRNA targets (Bartel, 2009; Axtell et al., 2011). Given the close phylogenetic relationship between green algae and land plants, it was anticipated that C. reinhardtii miRNAs recognize their targets according to plant-like complementarity rules (Molnar et al., 2007; Zhao et al., 2007). However, our mismatch tolerance test using pools of transgenic strains revealed that cre-miR1174.2 induced moderate silencing of Gaussia luciferase reporter genes when miRNA/target mRNA hybrids contained certain dinucleotide mismatches in the miRNA seed region or in regions flanking the cleavage site, and dinucleotide mismatches in the 3′ miRNA region had no obvious detrimental effect on repression (Figure 4). Moreover, base pairing in the seed region alone (for both cre-miR1174.2 and miR910) triggered mild silencing of target reporter genes (Figure 5 and Figure S7). Furthermore, G:U wobble pairing appeared to be well tolerated, as a reporter containing a miR910 target site with one mismatch and five G:U matches, including one in the seed region and one flanking the cleavage site, was efficiently repressed in Chlamydomonas (Figure S5).
These results suggested that some aspects of miRNA biogenesis and function in green algae, at least for certain miRNAs, bear more resemblance to mechanisms operating in metazoans than those in land plants. In particular, relatively limited complementarity between miRNA and target site may be enough to trigger gene silencing in Chlamydomonas. Although the magnitude of repression in these cases may be modest, individual miRNAs may potentially target multiple endogenous transcripts, thereby fine-tuning gene expression levels, as proposed in metazoans (Bartel, 2009). Our observations also indicated that, in Chlamydomonas, miRNAs may trigger RISC-mediated slicing of perfectly complementary target mRNAs (Figure 3). However, as in animal systems (Lewis et al., 2005; Bartel, 2009), the silencing mechanism may involve translation repression when the target contains dinucleotide mismatches around the central miRNA region, flanking the predicted cleavage site (Figure 4). Therefore, as previously described for small interfering RNAs (Ma et al., 2013), miRNAs may mediate silencing of target transcripts in C. reinhardtii by inducing RNA degradation and/or translation inhibition.
Culture conditions, transformation and genetic crosses
C. reinhardtii cells were grown photoheterotrophically in Tris/acetate/phosphate (TAP) medium as previously described (Yamasaki and Ohama, 2011). Nuclear transformation was performed by electroporation (Shimogawara et al., 1998) by applying an exponential electric pulse of 0.7 kV at a capacitance of 50 μF (BT600 Rev.G; BTX, http://www.btxonline.com/). The plasmids gluc-aph7, gluc(1 × )-aph7, gluc(4 × )-aph7 and gluc(8 × )-aph7 (Figure S1) were digested using KpnI, and the transgene fragments were purified by agarose gel electrophoresis. Excised DNA (200 ng) was used for transformation of CC-124. Transgenic strains were selected on TAP/agar plates containing 15 μg ml−1 hygromycin B (Wako, http://www.wako-chem.co.jp/english/; Berthold et al., 2002). Approximately 2.5 × 107 cells of the CC-3491 strain (cell wall-deficient) in 250 μl TAP medium supplemented with 50 mm sucrose (TAP/sucrose) were used for preparation of transgenic pools. Electroporation was performed using 300 ng of KpnI-linearized plasmids: tagaph7, gluc-tagaph7, gluc(1 × )-tagaph7, gluc(no pair)-tagaph7, gluc(random1)-tagaph7, gluc(random2)-tagaph7, gluc(4 × )-tagaph7, or gluc(8 × )-tagaph7 (Figure S3). Electroporated cells were incubated in 5 ml of fresh TAP/sucrose medium for 24 h. The cells were collected by centrifugation (1000 × g, 5 min) at room temperature, washed once with TAP medium, and then inoculated into 100 ml TAP medium containing 6 μg ml−1 hygromycin B. Hygromycin B-resistant transgenic strains were selectively grown in this medium for 5 days; subsequently, approximately 2 × 107 cells were collected by centrifugation (1000 × g, 5 min) at room temperature. Pelleted cells were washed twice with TAP medium, and then grown in 100 ml of antibiotic-free TAP medium for 2 days. The cell density usually reached approximately 4 × 106 cells/ml after this period of cultivation, and this population of mixed strains was used as a transgenic pool. Control cells were electroporated in the absence of transforming DNA, and these cells did not survive the 5-day hygromycin B selection regime. Genetic crosses were performed as described by Harris (1989).
Construction of the gluc reporter genes
DNA fragments containing cre-miR1174.2 target site(s) were constructed as previously described (Iwasaki et al., 2009). In brief, a DNA fragment containing one cre-miR1174.2 target site was amplified by PCR from the oligo-1 × cre-miR1174.2 oligonucleotide (5′-TGCGGCCGCCTCGACGCCGATAAGAAGGAGCCGTAACTCGACCCTACTAGTGGCC-3′; the target site is underlined), using primers cre-miR1174.2-left-F (5′-TGCGGCCGCCTCGAC-3′) and cre-miR1174.2-right-R (5′-GGCCACTAGTAGGGTCGAG-3′), and cloned into the T-vector pMD20 (TaKaRa Bio, http://www.takara-bio.com/). A DNA fragment containing four cre-miR1174.2 target sites was constructed from two PCR products (A and B). Both PCR products were amplified from the oligo-2 × cre-miR1174.2 oligonucleotide (5′-TGCGGCCGCGTCGACGCCGATAAGAAGGAGCCGTAACTCGACGCCGATAAGAAGGAGCCGTAACTCGAGCCTACTAGTGGCC-3′) using the following primer sets: cre-miR1174.2-left-R and cre-miR1174.2-left-R (5′-GGCCACTAGTAGGCTCGAG-3′) for fragment A, and cre-miR1174.2-right-F (5′-TGCGGCCGCGTCGAC-3′) and cre-miR1174.2-right-R for fragment B. Fragment A was digested using XhoI and fragment B was digested using SalI. The two fragments were ligated using the Mighty Mix DNA ligation kit (TaKaRa Bio). The ligation product was digested using XhoI/SalI and used as a template for PCR amplification with cre-miR1174.2-left-F and cre-miR1174.2-right-R to obtain a DNA fragment containing four cre-miR1174.2 target sites. This fragment was cloned into pMD20 (TaKaRa Bio). The DNA product containing eight cre-miR1174.2 target sites was amplified by PCR using a similar strategy but using the plasmid containing the fragment with four cre-miR1174.2 target sites as the template. DNA fragments containing mutated cre-miR1174.2 target sequences or target sites (underlined) for other Chlamydomonas miRNAs were generated as described for the PCR product with one cre-miR1174.2 target site, using the following oligonucleotides as templates: cre-miR1174.2 no pair target (5′-CGTGCGGCCGCCTCGACCGGCTATTCTTCCTCGGCATTCTCGACCCTACTAGTGGC-3′), cre-miR1174.2 randomized target1 (random1) (5′-CGTGCGGCCGCCTCGACAGGGCCAAAATAAGCCGGATGCTCGACCCTACTAGTGGC-3′), cre-miR1174.2 randomized target2 (random2) (5′-CGTGCGGCCGCCTCGACTCACCAAGGGGAACAAAGTGGCTCGACCCTACTAGTGGC-3′), cre-miR1174.2(1·2) (5′-CGTGCGGCCGCCTCGACGCCGATAAGAAGGAGCCGTTTCTCGACCCTACTAGTGGC-3′), cre-miR1174.2(3·4) (5′-CGTGCGGCCGCCTCGACGCCGATAAGAAGGAGCCCAAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(5·6) (5′-CGTGCGGCCGCCTCGACGCCGATAAGAAGGAGGGGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(7·8) (5′-CGTGCGGCCGCCTCGACGCCGATAAGAAGGTCCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(9·10) (5′-CGTGCGGCCGCCTCGACGCCGATAAGAACCAGCCGTAACTCGACCCTACTAGTGGC-3′, cre-miR1174.2(11·12) (5′-CGTGCGGCCGCCTCGACGCCGATAAGTTGGAGCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(13·14) (5′-CGTGCGGCCGCCTCGACGCCGATATCAAGGAGCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(15·16) (5′-CGTGCGGCCGCCTCGACGCCGAATAGAAGGAGCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(17·18) (5′-CGTGCGGCCGCCTCGACGCCCTTAAGAAGGAGCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2(19·20) (5′-CGTGCGGCCGCCTCGACGGGGATAAGAAGGAGCCGTAACTCGACCCTACTAGTGGC-3′), cre-miR1174.2 seed target (5′-CGTGCGGCCGCCTCGACCGGCTATTCTTCCAGCCGTAACTCGACCCTACTAGTGGC-3′), miR-912 target (5′-CGTGCGGCCGCCTCGACGCCTGTGGGGATCAATCCACTCGACCCTACTAGTGGC-3′), miR-912P target (5′-CGTGCGGCCGCCTCGACGCCTGGCTGGGATCAATCCACTCGACCCTACTAGTGGC-3′), miR-910 target (5′-CGTGCGGCCGCCTCGACGTGGATGGGCTCGGCGCTGCGCTCGACCCTACTAGTGGCC-3′), miR910 perfect target (5′-CGTGCGGCCGCCTCGACGCGGTCGAGCCCGACGCTGCTCTCGACCCTACTAGTGGC-3′), miR910 seed target (5′-CGTGCGGCCGCCTCGACCGCCAGCTCGGGCACGCTGCTCTCGACCCTACTAGTGGC-3′), miR910 no pair target (5′-CGTGCGGCCGCCTCGACCGCCAGCTCGGGCTGCGACGACTCGACCCTACTAGTGGC-3′), and candidate 82 target (5′-CGTGCGGCCGCCTCGACGGCGCTGTGCAGCTTTTCGAACTCGACCCTACTAGTGGC-3′).
A codon-optimized gluc open reading frame (ORF) was amplified by PCR from the pHsp70A/RbcS2-cgLuc plasmid (Ruecker et al., 2008) using primers NdeI-cgluc-F1 (5′-AGGATGCTCCATATGGGCGTGAAGGTG-3′) and EcoRI-NotI-cgluc-R1 (5′-ATGAATTCATGCGGCCGCTTAATCGCCGCCAGC-3′). The amplified gluc fragment was then inserted into the pGenD plasmid (Fischer and Rochaix, 2001) at the NdeI and EcoRI sites. The EcoRI site was replaced with a SpeI site by substituting the PsaD terminator region in the pGenD plasmid, by digestion at the NotI and HindIII sites, with a modified PsaD terminator fragment [amplified from pGenD using primers NotI-SpeI-PsaDter-F (5′-TAAGCGGCCGCATACTAGTTGGCAG-3′) and PsaDter-R2 (5′-GTGCCAAGCTTGATTGCACAGTCAC-3′). The PsaDpro::gluc::NotI::SpeI::PsaDter fragment obtained was then excised by BamHI/HindIII digestion and inserted into the corresponding sites of the pKF18-2-Δ-EcoRI plasmid [generated from the pKF18-2 plasmid (TaKaRa Bio) by EcoRI digestion, treatment with T4 DNA polymerase (TaKaRa Bio), and subsequent self-ligation]. For construction of the gluc-aph7 plasmid (Figure S1), the aph7 expression cassette (Berthold et al., 2002) was inserted into the HindIII site of the above-described vector.
For expression of gp64-tagged aph7 (tagaph7 plasmid, Figure S3), the aph7 expression cassette was removed from NE-689 (a kind gift from J. Rohr, University of Nebraska-Lincoln) by SpeI–NotI digestion, treatment with T4 DNA polymerase, and subsequent self-ligation. The aph7 fragment was amplified by PCR from the pHyg3 plasmid (Berthold et al., 2002) using primers NcoI-aph7-F (5′-CGCAACCATGGCACAAGAATCCCTG-3′) and PstI-aph7-R2 (5′-CACGCTGCAGGGATCCTTATCAGGC-3′). This fragment was re-integrated into the modified NE-689 vector at the NcoI and PstI sites. The gluc-tagaph7 plasmid (Figure S3) was constructed by excising the gluc expression cassette from the gluc-aph7 plasmid by KpnI–HindIII digestion and generating blunt ends by treatment with T4 DNA polymerase. The tagaph7 plasmid was digested with XbaI, treated with T4 DNA polymerase and Alkaline Phosphatase (Calf intestine) (TaKaRa Bio), and then the gluc expression cassette was ligated to the blunt ends. DNA fragments containing one, four or eight copies of the mutated cre-miR1174.2 target site and target sites for other Chlamydomonas miRNAs were inserted into the gluc-aph7 or gluc-tagaph7 plasmids at the NotI and SpeI sites to generate the various reporters (Figures S1, S3 and S5–S7).
Luciferase activity assay
A 96-well microplate luminometer (MTP-800, Corona Electric Co. Ltd, http://www.corona-el.co.jp/global/) was used for luminescence analyses. Cells were grown in 3 ml TAP medium for 3 days before measuring the luciferase activity of individual transgenic strains. A 50 μl aliquot of this culture was inoculated into 3 ml of fresh TAP medium, and cells were grown for an additional 2 days. Approximately 1 × 106 cells were collected by centrifugation (5000 × g, 3 min) at 4°C, resuspended in 400 μl TAP medium containing 1 mm of EDTA, vortexed for 10 sec, and quickly frozen at −20°C. Cells were then thawed at 25°C for 5 min, and 10 μl aliquots of the cell suspension were mixed with 10 μl of 10 μm coelenterazine (Gold Biotechnology Inc., https://www.goldbio.com/) in TAP medium in wells of a black 96-well microplate. Luciferase activity of the transgenic pools was evaluated by mixing 30 μl of each pool culture with 20 μl of 12.5 μm coelenterazine in TAP medium in a microplate well. Relative light units were normalized to the same number of cells for all samples examined, to permit comparative analyses of luciferase activity.
DNA and RNA analyses
Genomic PCR was performed as described previously (Cao et al., 2009). Whole transforming DNA fragments were amplified using primers PsaD-pro-F2 (5′-GACAGGAAGTGAACGCATGTCGAG-3′) and 3′rbcS2-R1 (5′-TCAACGGAGGATCGTTACAACCAAC-3′). Total RNA was purified using TRI reagent (Molecular Research Center, http://www.mrcgene.com/), according to the manufacturer's instructions. Northern hybridization analyses using digoxigenin-labeled probes or 32P-labeled probes were performed as previously described (Rohr et al., 2004; Yamasaki et al., 2008). RNA blots were stained with methylene blue prior to hybridization to confirm equal lane loading (Herrin and Schmidt, 1988). DNA probes for hybridization were generated by PCR amplification with the following primers: au5.g528-t1-F1 (5′-GTCGCCTACCACGTCACCAATC-3′) and au5.g528-t1-R1 (5′-CTTCACGTCCACGTCCTTCTTCTTC-3′) for the Cre10.g444300 ORF probe. au5.g528-t1-F4 (5′-GACGTGAACAATTGGCCATGACG-3′) and au5.g528-t1-R4 (5′-CCACTGCACCTGCAGACTACAC-3′) for the Cre10.g444300 3′ UTR probe, gcluc-F2 (5′-ATCTGCCTGTCCCACATCAAGTG-3′) and gluc-R1 (5′-CGCCCTTAATCTTGTCCACCTGG-3′) for the gluc probe, CBLP-F (5′-GAGTCCAACTACGGCTACGCC-3′) and CBLP-R (5′-CTCGCCAATGGTGTACTTGCAC-3′) for the CBLP probe, and U6snRNA-F (5′-GTGCTTCGGCACAACTGTTAAAAT-3′) and U6snRNA-R (5′-AAAATTTGGAACCATTTCTCGATTTATGC-3′) for the U6 snRNA probe.
Quantitative PCR analyses
Purified total RNA was treated with DNase I (TaKaRa Bio) to remove contaminant DNA. First-strand cDNA was synthesized using an oligo(dT)18 primer or random heptamers and PrimeScript reverse transcriptase (TaKaRa Bio) according to the manufacturer's instructions. Primer pairs for the quantitative PCR amplifications were as follows: gcluc-F2 and gcluc-R1 for the gluc probe, and CBLP-F and CBLP-R for CBLP. DNA fragments were amplified and quantified using SYBR Premix Ex Taq II (Tli RNaseH Plus; TaKaRa Bio) using the StepOne real-time PCR system (Life Technologies, http://www.appliedbiosystems.com/).
5′ RACE assay
Putative cleavage sites of the Cre10.g444300 transcript in strain CC-124 and putative slicing sites of the gluc(1 × ) transcript in the gluc(1 × )-10 strain were examined using a GeneRacer kit (Life Technologies) as previously described (Molnar et al., 2007). First and nested PCR amplifications were performed using the following primers: 444300-5′ RACE-R (5′-GGCAGCCGTCAGATGACAGCAACTA-3′) and 444300-5′ RACE-nested-R (5′-CGGCCCATGCTTCTAGCAGCATACT-3′) for Cre10.g444300, and gluc(cre-miR1174.2) -5′ RACE-R (5′-CACTGGCTCACGCACACGCTAACAT-3′) and gluc(cre-miR1174.2) -5′ RACE-nested-R (5′-TCCCCCTGATCCTCCTGTGGCTAAT-3′) for gluc(1 × ).
Immunodetection was performed as described previously (Rohr et al., 2004). Gaussia luciferase protein was detected using an anti-gluc antibody (E8023S, New England BioLabs, https://www.neb.com/). gp64-tagged aph7 protein was detected using an anti-baculovirus envelope gp64 polypeptide antibody (14-6995, eBioscience, http://www.ebioscience.com/). Protein blots were stained using 0.5% w/v Ponceau S in 1% acetic acid.
This work was supported in part by Japan Society for the Promotion of Science (JSPS) KAKENHI (grant number 70512060 to T.Y.). We also acknowledge the support of the US National Science Foundation (to H.C.).