Article | Open | Published:

High frequency of +1 programmed ribosomal frameshifting in Euplotes octocarinatus

Scientific Reports volume 6, Article number: 21139 (2016) | Download Citation


Programmed −1 ribosomal frameshifting (−1 PRF) has been identified as a mechanism to regulate the expression of many viral genes and some cellular genes. The slippery site of −1 PRF has been well characterized, whereas the +1 PRF signal and the mechanism involved in +1 PRF remain poorly understood. Previous study confirmed that +1 PRF is required for the synthesis of protein products in several genes of ciliates from the genus Euplotes. To accurately assess the frequency of genes requiring frameshift in Euplotes, the macronuclear genome and transcriptome of Euplotes octocarinatus were analyzed in this study. A total of 3,700 +1 PRF candidate genes were identified from 32,353 transcripts, and the gene products of these putative +1 PRFs were mainly identified as protein kinases. Furthermore, we reported a putative suppressor tRNA of UAA which may provide new insights into the mechanism of +1 PRF in euplotids. For the first time, our transcriptome-wide survey of +1 PRF in E. octocarinatus provided a dataset which serves as a valuable resource for the future understanding of the mechanism underlying +1 PRF.


As a typical ciliate, Euplotes octocarinatus exhibits nuclear dimorphism [micronucleus (MIC) and macronucleus (MAC)]. The MIC is diploid and transcriptionally inert during most of its life cycle, enabling the transmission of genetic information between generations by sexual reproduction. The MAC is considered the somatic nucleus, which is transcriptionally active during the vegetative growth1. During conjugation, the MAC is degraded and a new MAC is developed from the zygote nucleus accompanied by DNA rearrangements. Similar to that of other hypotrichous ciliates, the MAC of E. octocarinatus contains abundant gene-sized DNA molecules (‘nanochromosome’, with mean length ~2 kb), each of which is differentially amplified2. All nanochromosomes have telomeric repeats 5′-(C4A4)n-3′ at their ends3.

The following unique features distinguish Euplotes from other ciliates: 1) the conventional stop codon UGA is reassigned as cysteine4 or selenocysteine5, which means that only the UAA and UAG are used as stop codons in Euplotes; and 2) the high frequency of +1 programmed ribosomal frameshifting (PRF) in Euplotes6. PRF is a recoding event by which the translating ribosome switches from the initial (0) reading frame to the −1 or +1 reading frame at a specific position, and then continues its translation7. Although the first reported frameshifting sequence has been found in viruses, it is becoming increasingly apparent that PRF is also widespread and likely exists in all branches of life from bacteria to higher eukaryotes8,9,10.

On the basis of the reading frame shift, two main PRFs (−1 and +1) were reported in viruses and other cellular organisms. The −1 PRF is prevalent and abundant and the most well-defined −1 PRF phenomena are directed by an mRNA sequence motif composed of the following three crucial elements11: 1) the so-called slippery sequence composed of seven nucleotides; 2) a short spacer sequence (usually less than 12 nt); and 3) a downstream stimulatory structure (usually a pseudoknot or a stem-loop). Compared with −1 PRF, +1PRF has fewer examples found in bacteria, fungi, mammals and ciliated protozoa of Euplotes. In the majority of bacteria, +1 PRF reportedly regulates the expression of release factor 2 (RF2)12,13; in fungi and mammals, +1 PRF purportedly regulates the expression of ornithine decarboxylase antizyme (OAZ), the negative regulator of cellular polyamine levels14,15.

Unlike −1 PRF, which has only one well-understood type of frameshift signal, +1 PRF involves highly diverse mechanisms. In Escherichia coli, RF2 autoregulates its production by the in-frame UGA premature termination codon found within the slippery site U CUU UGA. Peptide chain termination is efficient when adequate RF2 is present, thereby suppressing +1 PRF and limiting its translational production. However, low RF2 levels result in the inefficient recognition of the UGA codon and thus increased efficiency of +1 PRF, thereby allowing the expression of the RF2 protein16. In addition, a Shine–Dalgarno–like (SD–like) element located upstream of the slippery site can stimulate a +1 frameshift by interacting with the anti–SD sequence on the 16S rRNA17. In eukaryotes, +1 PRF is driven by other mechanisms. In the case of human OAZ mRNA, the crucial stimulatory element is the mRNA secondary structure located downstream of the slippery sequence. Similar to RF2 from E. coli, the OAZ +1 frameshift is stimulated by a 0-frame UGA codon, and is also autoregulated. Ornithine decarboxylase (ODC) catalyses the first step in polyamine biosynthesis, whereas OAZ downregulates polyamine synthesis by stimulating the ubiquitin-independent degradation of ODC by the proteasome. Thus, the increased levels of polyamines cause a negative feedback on polyamine synthesis by stimulating +1 PRF, and hence OAZ synthesis18.

Euplotes contains several +1 PRF genes, such as the Tec2 transposon ORF2 protein19,20, membrane occupation and recognition nexus (MORN) repeat protein, C2H2-type zinc finger protein, Ser/Thr protein kinase6, cAMP-dependent protein kinase21, nuclear protein kinase22, La motif protein23, mitogen-activated protein kinase (MAPK1)24,25, and the reverse transcriptase subunits of telomerase26,27,28. All of these genes have some common features. Their slippery sequences usually have the motif 5′-AAA-TAR-3′ (where R=A or G, the underlined sequence denotes the 0-frame codons), and all genes require a +1 PRF to produce complete protein products. In addition, a previous survey6 of Euplotes crassus macronuclear genes by random sequencing has found three new putative +1 PRF genes from 23 macronuclear genes, suggesting that the frequency of genes requiring frameshifts may exceed 10%.

The present study conducted a genome-wide investigation of +1 PRF in E. octocarinatus through genome and transcriptome sequencing. A total of 3,700 (approximately 11%) putative +1 PRF genes were identified in E. octocarinatus. To the best of our knowledge, this frequency of +1 PRF is the highest found in all living organisms. Based on the functional annotation of Pfam, GO and KEGG, we systematically investigated the putative functions of +1 PRF gene products, which were mainly enriched in protein kinases. We also found a novel suppressor tRNA of UAA which is a potential key factor of +1 PRF in euplotids. This work provides the first comprehensive genome-wide investigation of +1 PRF in E. octocarinatus, and thus lays a foundation for further exploring the mechanism of PRF.

Results and Discussion

Constructing the transcripts of E. octocarinatus by genome and transcriptome sequencing

The PRF occurs at the post-transcriptional level; thus, transcripts should first be assembled to analyse the PRF. Reference-based transcriptome assembly is reported as the best method to construct transcripts, especially full-length transcripts, from short high-throughput sequencing reads29. The MAC genome and transcriptome of E. octocarinatus were sequenced to construct a high quality transcripts set.

In consideration of the unique properties of the highly fragmented macronuclear genome of Euplotes, two short paired-end sequencing libraries with insert sizes of 180 bp (100 bp × 2 by Illumina Hiseq2000 platform) and 500 bp (300 bp × 2 by Illumina Miseq platform) were constructed and sequenced. In total, about 11 gigabases (Gb) were obtained (Table S1). Several popular short-read assemblers were tested and compared to obtain full-length nanochromosomes and the minimum number of contigs simultaneously (Table 1). Finally, we adopted the strategy in Figure S1. In general, HiSeq and Miseq data were independently assembled using the assembler with the best performance. Specifically, the Miseq data (300 bp × 2) were assembled using Mira. This assembly produced a large proportion of 2-telomere contigs (61.0%) and long contigs (N50 length 2,683 bp). However, the Miseq assembly missed many nanochromosomes which were shorter than 500 bp because of the library insert size limitation (Table 1). Therefore, the Hiseq data (100 bp × 2) was independently assembled using SPAdes. This assembly produced 24.8% of 2-telomere capped contigs with a N50 length of 1,129 bp. Then, the two assemblies were merged using CAP3, and all redundant contigs were removed on the basis of the result of LASTZ (see Methods and Fig. S1). However, some telomereless contigs (5,494) shorter than 500 bp were not removed from the final assembly. We speculate that those ‘chaff’ contigs were fragments of the MIC or the inter-genic regions of some long chromosomes with a low GC content (Fig. S2).

Table 1: Comparison of Euplotes octocarinatus macronuclear genome assemblies.

Based on the GC content results of all contigs (Fig. S3), we suspected that the initial assembly contained a mixture of target DNA, bacterial DNA (endosymbionts of E. octocarinatus) and mitochondrial DNA (some sub-peaks were located behind the major peak). Therefore, a series of filters was applied to exclude the contamination (see Methods), and 1,628 bacterial contigs and 72 mitochondrial contigs were removed from the initial assembly. Finally, a total of 41,980 contigs with an average length of 2,117 bp were used as the E. octocarinatus MAC genome assembly, and most (70.1%) of these contigs were capped with telomeres on both ends. The completeness of the genome was supported by the assessment results (see Methods). Subsequently, we compared two reported highly fragmented macronuclear genomes30,31 with the E. octocarinatus assembly. Similar to Oxytricha and Stylonychia, few nanochromosomes were assembled at either extremities of the length distribution in Euplotes (Fig. 1). Only 283 were shorter than 500 bp and 15 were longer than 15 kb.

Figure 1: Length distribution of nanochromosomes of three highly fragmented macronuclear genomes.
Figure 1

X axis is the contig length (nucleotides), Y axis is the frequency of contigs with the indicated lengths. The histograms show normalized frequencies for 29,413 nanochromosomes of Euplotes octocarinatus, 15,085 nanochromosomes of Oxytricha trifallax and 16,029 nanochromosomes of Stylonychia lemnae.

To construct the transcript set, high-throughput RNA-seq (125 bp × 2) of E. octocarinatus was performed (see Methods). We obtained 39,478,354 short reads, with a total length of more than 4.9 Gb through sequencing. Low-quality reads were filtered by fastq_quality_filter with the parameters -q 20 -p 80. Then high-quality reads of RNA-seq data were mapped to the Euplotes macronuclear genome by Tophat32, and all mapped reads were assembled using Cufflinks33. Finally, 32,353 transcripts were generated with a mean transcript length of 1,300 bp and a N50 of 1,578 bp.

High frequency of +1 PRF in E. octocarinatus

A similarity search-based method was used to identify the +1 PRF transcripts in E. octocarinatus. The strategy was to find out-of-frame ORFs first and then identify the frameshift motif in the in-frame ORF which could potentially redirect ribosomes from the upstream ORF into the downstream one, resulting in the translation of a complete protein. As depicted in Fig. 2, all transcripts were aligned to the NCBI non-redundant (nr) protein database by using BLASTX with a cut-off of 10−5. A total of 6,064 transcripts having two or more high score fragments with different reading frames in the same hit protein sequence were extracted on the basis of BLASTX results. Considering that the intron retention transcripts may also direct the production of out-of-frame ORFs and lead to BLASTX results similar to PRF genes, we identified and excluded intron retention transcripts by using transcriptome information. Based on the typical arrangement of +1 frameshift genes in Euplotes, the initial open reading frame is expected to terminate with the sequence 5′-AAA TAR-3′. So the stop codon (TAA or TAG) was searched in the initial open reading frame, and the ‘T’ of the stop codon was artificially removed. Subsequently, this new fs-gene was aligned to the NCBI nr protein database by BLASTX again. Once a C-terminally extended protein was produced, this gene would be marked as a +1 PRF gene. Using this strategy, we identified 3,700 putative +1 PRF genes from the 32,353 E. octocarinatus transcripts. In addition to the 3,489 +1 PRF genes with the classical ‘Euplotes frameshift motif’ (5′-AAA-TAR-3′), we also identified 211 novel +1 PRF genes with different types of slippery sequences. Among these novel +1 PRF genes, the most abundant slippery sequence motif was the 5′-TTT-TAR-3′ motif with 54 genes (Table S2), followed by the 5′-AAG-TAR-3′ motif with 41 genes, the 5′-AAT-TAR-3′ motif with 29 genes, and the 5′-ATT-TAR-3′ motif with 28 genes.

Figure 2: Pipeline of prediction of +1 programmed ribosomal frameshifted transcripts.
Figure 2

Thus, the present study has increased the number of previously known +1 PRF genes in E. octocarinatus by three orders of magnitude. As expected, two previously reported +1 PRF genes in E. octocarinatus –cAMP-dependent protein kinase21 (CUFF.28794.1) and putative nuclear protein kinases22 (CUFF.8279.1) – were identified by our pipeline, suggesting that the method we used was robust. Detailed information on the putative +1 PRF transcripts, including length, GC content, coordinates of predicted slippery sit and E-value of BLASTX is presented in Table S3.

The genome-wide analysis of the Saccharomyces cerevisiae genome11 and 1,106 complete prokaryotic genomes34 suggests a high frequency of −1 PRF in these organisms. However, only a few of the +1 PRF genes have been described (Table 2). By contrast, no −1 PRF gene has been reported in Euplotes so far, but several +1 PRF cases have been reported. Our results showed that approximately 11.4% genes required +1 PRF to produce a functional protein in E. octocarinatus. Our results provide evidence supporting the notion that euplotids contain an extremely high number of genes requiring +1 frameshifts for expression at the post-transcriptional level. The observed number of +1 PRF genes was higher in E. octocarinatus than in other organisms (Table 2), but the true percentage of frameshifted genes in E. octocarinatus should be more abundant, because only 52.4% of the transcripts (16,950 of 32,353) have a homologous gene in other organisms. In specific, about half of the transcripts whose functions are unknown may also require a frameshift for expression.

Table 2: Summary of +1 programmed ribosomal frameshifted genes in diverse organisms.

Hypothetical function of +1 PRF gene products is significantly enriched in protein kinases

We systematically investigated the hypothetical function of 3,700 identified +1 PRF genes. A total of 2,336 putative +1 PRF genes were found to contain at least one protein domain by searching the Pfam database. The most abundant protein domain found in those genes was ‘Pkinase’ domain (PF00069.20), with a total of 362 genes (Table S4), followed by the ‘MORN’ domain (PF02493.15) with 265 genes, the ‘WD40’ domain (PF00400.27) with 179 genes, the ‘SHIPPO-rpt’ domain (PF07004.7) with 146 genes, and the ‘cNMP_binding’ domain (PF00027.24) with 130 genes. All putative +1 PRF genes were mapped to the Kyoto Encyclopedia of Genes and Genomes (KEGG)35 pathway to investigate the biological pathways where the putative +1 PRF genes may be involved. In general, 813 genes were assigned to 282 KEGG pathways (Table S5). The pathways represented by the putative +1 PRF genes included the PI3K-Akt signalling pathway (18 members), the sphingolipid signalling pathway (14 members) and the MAPK signalling pathway (12 members). Furthermore, a total of 1,629 putative +1 PRF transcripts were annotated with at least one GO term and categorised into 26 functional groups on the basis of sequence homology (Fig. S4). In each of the three main categories, namely, GO classification cellular component, molecular function and biological process, the terms ‘cell’ and ‘cell part’, ‘binding’ and ‘catalytic’, and ‘cellular process’ were dominant, respectively.

Functional annotations indicated that the putative +1 PRF genes in E. octocarinatus possessed various functions involved in multiple cellular processes and pathways. As reported previously, most putative +1 PRF genes encode proteins with enzymatic functions, especially protein kinases6. However, none of the highly expressed genes in cells have been reported to require a frameshift which is proven by the fact that the expression abundance of putative +1 PRF genes [mean fragments per kilobase of transcript per million mapped fragments (FPKM) value: 10.59] was significantly lower than that of normal genes (mean FPKM value: 44.38) (p < 0.01, t test). The most abundant representative proteins in the cell were ribosomal proteins. All 79 of the standard eukaryotic ribosomal proteins of E. octocarinatus (32 small subunit and 47 large subunit proteins) were identified and analysed, and none of them required a frameshift for expression.

A GO enrichment analysis of putative +1 PRF genes was performed to investigate the functional enrichment of putative +1 PRF genes. Results showed that the identified putative +1 PRF genes were significantly overrepresented in the regulation of various biological processes such as dephosphorylation, protein amino acid phosphorylation, and ubiquitin-dependent protein catabolic process (Fig. 3).

Figure 3: Enriched GO terms of +1 PRF transcripts were analyzed with Bingo.
Figure 3

Each circle represents a GO term, arrows indicate pairs of GO terms with a parent-child relationship. Colored circles are statistically significant overrepresented GO terms (functions); the deeper of the color, the smaller the corrected p-value (more significant). Overrepresented molecular functions (A) and biological processes (B).

These results suggest that the products of these putative +1 PRF genes in E. octocarinatus are significantly enriched in protein kinases. Protein kinases are important regulatory components of every eukaryotic intracellular signal transduction pathway. Some protein kinases, such as MAPK1, are associated with cell proliferation and cell cycle events; MAPK1 is a homologous kinase with intestinal–cell kinases in mammals. The expression of the MAPK1 gene requires +1 translational frameshifting in both Euplotes raikovi24 and Euplotes nobilii25. This Euplotes kinase is related to the autocrine signalling loop that promotes vegetative growth. Furthermore, the MAPK1 of E. raikovi resides in the nuclear apparatus, where it appears either phosphorylated in growing cells which interact in autocrine fashion with their own specific (self) signalling pheromones or dephosphorylated in cells which are induced to mate and temporarily arrest their growth by paracrine interactions with foreign (non-self) signalling pheromones24. These results suggest that +1 PRF genes may have important functions in cell growth.

Suppressor tRNA may play an important role in +1 frameshifting in E. octocarinatus

With sufficient samples of +1 PRF genes, the conserved sequence elements which might facilitate frameshifting were checked. To search the potential conserved sequence elements, 30 bp upstream and downstream of the conserved slippery sequence motif from 4,545 predicted slippery sites were extracted and analysed using WebLogo36. Consistent with a previous report, no conserved sequence element was found except the slippery site sequence 5′-WWW TAR-3′ (W=A or T, Fig. 4), suggesting that frameshifting does not depend on other sequence motifs.

Figure 4: WebLogo displaying conserved sequences associated with frameshift sites.
Figure 4

Sizes of letters denote information content, or sequence conservation, at each position. The analysis is based on the alignment of 30 bp preceding and following the frameshift motif from 4,545 predicted slippery sites.

Klobutcher and Farabaugh37 suggested that altering eRF1 to ignore UGA might impair its recognition of other termination codons in Eupotes. In addition, Vallabhaneni et al.38 proved that the reassignment of UGA to Cys in E. octocarinatus increases +1 slippery stop frameshifting at both UAA and UAG. Based on this finding, we analysed the stop codon usage in E. octocarinatus at the transcriptome level, and compared the usage between the ‘normal’ stop codon and the slippery stop codon. Results showed that UAA was preferentially used in both the ‘normal’ termination signal (79.6%) and the slippery signal (89.4%) (Fig. 5A). Moreover, the frequency of UAA codon usage in slippery signal is significantly higher than that in ‘normal’ termination signal (P < 0.01, Fisher exact test) which suggested that UAA may be favourable for frameshifting in E. octocarinatus. The release factor recognises a tetranucleotide sequence consisting of the termination codon and its nearest 3′ neighbour nucleotide in both prokaryotes and eukaryotes37. Thus, we also analysed the frequency of the tetranucleotide sequence in both normal and +1 PRF slippery sites (Fig. 5B). A similar trend was found in both cases, where UAA-A was the most frequently used tetranucleotide (49.7% in the slippery signal and 32.0% in the ‘normal’ termination signal) and UAG-G was the least frequently used (0.5% in the slippery signal and 1.8% in the ‘normal’ termination signal). However, the usage frequency of UAA-A (49.7%) was considerable higher than that of UAA-U (16.9%) in the slippery signals, even though they have similar frequencies in the “normal” termination signals (32.0% vs. 31.0%). These results suggest that the tetranucleotide sequence UAA-A may be favourable for frameshifting in E. octocarinatus.

Figure 5: Frequency of stop codon usage in Euplotes octocarinatus.
Figure 5

(A) Frequency of stop codon-trinucucleotide. (B) Frequency of stop codon-tetranucleotide. The red and yellow bars indicate normal termination signal and slippery signal, respectively.

While no suppressor tRNA of UAA had been previously reported in Euplotes, a putative suppressor tRNA of UAA (Contig36094) was predicted from the genomic sequences of E. octocarinatus. Contig36094 (343 bp) was predicted to encode a 72 nucleotide tRNA that can fold into the characteristic cloverleaf secondary structure (Fig. 6B). An intervening sequence of 11 base pairs located at the canonical 37/38 position39 was also predicted in this gene (Fig. 6A). The gene contains the characteristic internal split promoter and a typical termination signal40 (Fig. 6A). Furthermore, a perfect consensus sequence, ‘TATAAAA’, for the TATA-binding protein (TBP) was located at position −35 to −29 relative to the +1 nt of the tRNAUAA.

Figure 6: Nucleotide sequences of novel tRNA from different eukaryotic organisms shown in the cloverleaf arrangement.
Figure 6

(A) The nucleotide sequence of Contig36094 is shown. The coding region is indicated by bold face and the terminal C4A4 and G4T4 telomeric sequences are shown in italic. The intragenic promoter boxes, the termination signal and a putative binding for the TATA-box binding protein are underlined. The putative intron is shown in red. The predicted cloverleaf structures of three novel tRNAs from (B) Euplotes octocarinatus (C) Schizosaccharomyces pombe and (D) Astasia longa are shown.

Further analysis indicated a base mismatch in the anticodon stem of the molecule, which increased the loop of unpaired bases from the typical seven to nine (Fig. 6B). Such an unusual structure was not unprecedented, and two examples of apparently nine-base anticodon loops in presumably wild-type, functional tRNAs were observed, namely, a tRNALeu in Schizosaccharomyces pombe41 (Fig. 6C) and a tRNAMet in Astasia longa42 (Fig. 6D). In addition, the similar suppressor tRNAs that have been isolated from both bacteria43,44,45,46 and yeast47,48, contained additional nucleotides in their anticodon loops. Expanded or modification-deficient anticodon stem loops have been proven to cause the ribosome to decode four rather than three nucleotides, resulting in a +1 translational frameshifting49,50,51. Therefore, we proposed that the particular suppressor tRNAUAA enters the ribosomal A site and decode 4 nucleotides when the translating ribosome meets the slippery stop codon. Then translation would continue in the +1 frame. Further experimental verification is needed to investigate how suppressor tRNAUAA regulates +1 PRF in E. octocarinatus.


We reported a genome-wide investigation of +1 PRF in E. octocarinatus on the basis of its genome and transcriptome sequencing. We identified 3,700 (about 11%) putative +1 PRF genes, which to the best of our knowledge, is the highest frequency of +1 PRF found in all living organisms up to date. We also found a novel suppressor tRNA of UAA, which is potentially the key factor of +1 PRF in euplotids. This work provided the first comprehensive genome-wide investigation of +1 PRF and contributed to the mechanism of underlying programmed translational frameshift.


Cell culture, DNA isolation, and genome sequencing

E. octocarinatus line 69 was cultured in 2 liter flasks containing in synthetic medium52 at room temperature with the photosynthetic flagellate Chlorogonium elongatum as a food source. This strain was kindly provided by Klaus Heckmann (Universität Münster, Germany). Prior to harvesting, Euplotes cells were starved for 7–10 days to allow them to exhaust most of the food. Then, 8–10 liters of starved cells were harvested by filtering through several layers of gauze to remove large particles, and then a filter paper was used to concentrate cells and remove bacteria and small contaminants. Cells were collected by centrifugation (4 °C,4,000 rpm, 5 min) and then lysed in Urea buffer (0.01 M Tris-HCl, 0.01 M EDTA, 0.35 M NaCl, 1% SDS, 42% Urea, pH 7.4) for 5 min at 4 °C. After phenol/chloroform extraction, total DNA was dialyzed against isopropanol followed by ethanol precipitation. RNase was then added and incubated for 1hr at 37 °C.

According to the whole genome shotgun strategy, genomic DNA was broken into random fragments. Two libraries with different paired-end (PE) length distributions were created from Euplotes DNA sequence data (Table S1). The 500 bp library was sequenced using the Illumina MiSeq platform, and the 180 bp library was sequenced using the Illumina HiSeq 2000 platform.

The genome assembly and assembly cleanup

The genome was assembled by a meta-assembly method (Fig. S1). All sequence data were used to build a reference genome.

MiSeq reads were assembled with Mira (4.0)53. High quality reads were selected for assembly. Read1 and Read2 were trimmed by fastx_trimmer (from the FASTX-Toolkit) with the parameters -l 290 and -l 250, respectively, and then filtered by fastq_quality_filter (from the FASTX-Toolkit) with the parameters -q 20 -p 80. FLASH (1.2.10)54 was used to merge these processed paired-end reads with the parameters –M 100. Finally, approximately 3.5 Gb reads were assembled with Mira (default parameters). SPAdes (2.5.0)55 was run with the “careful” option on HiSeq reads. Then we merged two assembly results with the CAP3 assembler with strict overlap parameters (-o 50 -p 99).

To remove redundant contigs from the assembly, LASTZ56 was used to align every contig to each other. Contigs were discarded (13,623 in total) if they had longer non-self matches that are identical or almost identical (≥90% coverage and ≥90% sequence identical) (Fig. S1).

The final assembly contained a mixture of bacterial DNA and mitochondrial DNA. To identify bacterial genomic sequences, all telomereless contigs were searched against NCBI non-redundant protein sequences database using BLASTX (E-value ≤ 1e−5). Any contig that belongs to bacteria or archaea was removed. To exclude mitochondrial contamination in our final assembly, these telomereless contigs which had substantial TBLASTX matches (E-value ≤ 1e−4) to the Euplotes minuta and Euplotes crassus mitochondrial genome57 were removed. A total of 1628 bacterial contigs and 72 mitochondrial contigs were identified and removed. In addition, we also removed 35 contigs that were shorter than 100 bp.

Assessment of genome completeness

To assess the completeness of the draft Euplotes macronuclear genome, we used a strategy similar to that used to assess completeness of Oxytricha30.

Firstly, we evaluated the percentage of reads mapping to the final assembly. All reads and reads containing telomere sequences of HiSeq reads were separately mapped to the final assembly with BWA58 (default parameters; version 0.7.5). Nearly all high-quality reads mapped to our final assembly (96% of all PE reads and 92% of telomeric reads). Furthermore, the majority of contigs (70.1%) had both 5′ and 3′ telomeres (Table 1). This simple assessment indicated that our assembly was largely complete.

Then we analyzed the completeness of two gene sets: ribosomal proteins and tRNAs. Based on the reciprocal blast results, all 80 of the standard eukaryotic ribosomal proteins except L41 were identified (32 small subunit and 47 large subunit proteins). Considering that the coding sequence of human L41 is too short (only 75 bp), we speculated that the L41of Euplotes probably was missed in the process of library construction. In addition, tRNAscan (version 1.3.1)59 with default parameters was used to search for tRNAs. A total of 95 contigs of Euplotes’s macronuclear genome encode a comprehensive set of tRNAs for all 20 standard amino acids (Table S6) including a novel suppressor tRNAs of traditional stop codon UAA.

We also assessed the completeness of the macronuclear genome by searching protein sequences from Euplotes against the core eukaryotic genes (CEGs)60. Matches from BLASTP with E-values lower than 1e-10 and a sequence coverage ≥70% of the CEG sequence were counted as a match. Of the predicted proteins, 218 proteins were predicted for Euplotes and had substantial sequence similarity to the CEG protein sequences. 21 of the 30 remaining CEGs were found by TBLASTN matches or using HMMER361 domain searches because of the deep evolutionary divergences of ciliates from these eukaryotes. After these more sensitive searches in Euplotes, only 6 CEGs are missing from the 245 ciliate-specific CEGs. Of the six undetectable ciliate-specific CEGs, one, KOG3285, is also missing from Oxytricha30 and Stylonychia31. Thus, the macronuclear genomes of Euplotes encode 97.6% of the ciliate-specific CEGs.

RNA isolation and transcriptome sequencing

Cell culture and collection were the same as described for DNA isolation. Total RNA was extracted using the RNeasy Plus Mini Kit Cell Mini Kit (Qiagen) per manufacturer’s instructions. Total RNA concentrations was determined using Qubit RNA Assay Kit in Qubit 2.0 Flurometer (Life Technologies, CA, USA) and RNA integrity was assessed using the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies,CA, USA).

Poly-A mRNAs was purified using Dynal magnetic beads (Invitrogen). Double-stranded cDNAs were synthesized using reverse transcriptase and random hexamer primers. cDNAs were fragmented by nebulization, and the standard Illumina protocol was followed thereafter to construct mRNA-seq libraries. The normalized cDNA population was sequenced using the Illumina 2000 platform, with paired-end 125 bp mode. About 4.9 Gb of raw RNA-seq data were obtained.

Gene prediction

Given that the existence of PRF genes will influence the accuracy of gene prediction, all +1 PRF candidate genes were excluded. The de novo prediction software AUGUSTUS (version 3.0.2)62 was used to predict complete genes on the non-PRF contigs (38,615 in total).

To obtain a reliable training data set, all non-PRF transcripts of Cufflinks33 assembly were aligned to the Euplotes macronuclear genome and reassembled by PASA (version r20140417, run with default parameters)63. These transcripts which were marked as “complete” (had both ATG and TAA/TAG) were searched against NCBI non-redundant protein sequences database using BLASTX. Matches from BLASTX with E-values lower than 1e-10 and a sequence coverage ≥95% of the top hit sequence were extracted. In addition, we only chose proteins that were < 70% identical to each other according to the recommendations in the AUGUSTUS documentation. Ultimately, 551 Euplotes genes were used to train AUGUSTUS. The final data set of 551 genes was split into training and test data sets of 401 and 150 genes respectively. For gene prediction, AUGUSTUS was run with the following parameters: “–species=euplotes –UTR=on –extrinsicCfgFile=install/augustus.3.0.2/config/extrinsic/extrinsic.M.RM.E.W.cfg –genemodel=complete –codingseq=on”. To produce additional constraints (hints) for AUGUSTUS, the RNA-seq data were processed according to the instructions on the website ( We also recompiled AUGUSTUS after decreasing the default minimum length of intron hints in the source code from 39 to 25 bp to allow AUGUSTUS to evaluate hints for shorter introns30.

Overall, 29,076 putative protein-coding genes were obtained, and 90% of them were supported by RNA-Seq Reads. About 89% nanochromosomes were predicted containing one or more complete gene. Key properties of Euplotes’s gene predictions are given in Table S7. Like other ciliates, the noncoding regions of Euplotes are more AT-richer than coding regions (e.g., 23.3% GC content in introns, versus 31.3% GC content in exons).

Functional annotation and enrichment analysis

The ‘T’ of the stop codon within the slippery sequence was artificially removed. Then these new fs-genes and non-PRF transcripts were translated into amino acid sequences using the GetOrf program in the EMBOSS package64 with the Euplotid Nuclear Code, and the longest CDSs were obtained by using a custom Perl script. Subsequently, the amino acid sequences were loaded into Pfamscan65 to perform protein domain annotation, and only Pfam-A database was searched.

Gene ontology (GO) annotations were performed by mapping of GO terms to Pfam entries. This mapping was generated from data supplied by InterPro66 for the InterPro2GO mapping. Functional enrichment analysis of +1 PRF genes was performed to determine the significantly enriched GO terms and relevant proteins by using BINGO67 plugin in the Cytoscape platform (version 3.1.1)68. Enrichment analysis of GO term assignment was performed in reference to the entire E. octocarinatus transcripts (containing 7,060 proteins). The corrected (corr) p-values were derived from a hypergeometric test followed by Benjamini and Hochberg false discovery rate (FDR) correction. The FDR ≤ 0.05 was regarded as significant.

In addition, the +1 PRF transcripts were annotated according to the Kyoto Encyclopedia of Genes and Genomes (KEGG)35 orthology (KO) using the KEGG Automatic Annotation Server (KAAS)69. The amino acid sequences of translated +1 PRF transcripts were used as the query sequence, and the bi-directional best hit (BBH) method was employed to obtain the KO terms for the query sequences.

Availability of supporting data

The BioProject accession number for the genome is PRJNA294366. The raw genome sequences reads have been deposited in Sequence Read Archive (SRA) under accession SRX1267944 and SRX1270715. Transcriptome data has also been deposited in SRA under accession SRX1270740.

Additional Information

How to cite this article: Wang, R. et al. High frequency of +1 programmed ribosomal frameshifting in Euplotes octocarinatus. Sci. Rep. 6, 21139; doi: 10.1038/srep21139 (2016).


  1. 1.

    , & Isolation of micronuclei from Euplotes octocarinatus and identification of an internal eliminated sequence in the micronuclear gene encoding γ-tubulin 2. European Journal of Protistology 35, 208–216 (1999).

  2. 2.

    & The molecular biology of ciliated protozoa. Academic Press, New York, 111–154 (1986).

  3. 3.

    , , & Characterization of transcription initiation, translation initiation, and poly(A) addition sites in the gene-sized macronuclear DNA molecules of Euplotes. Nucleic Acids Research 22, 214–221 (1994).

  4. 4.

    et al. UGA is translated as Cysteine in pheromone-3 of Euplotes octocarinatus. Proc Natl Acad Sci. USA 88, 3758–3761 (1991).

  5. 5.

    et al. Genetic code supports targeted insertion of two amino acids by one codon. Science 323, 259–261 (2009).

  6. 6.

    Sequencing of random Euplotes crassus macronuclear genes supports a high frequency of +1 translational frameshifting. Eukaryotic Cell 4, 2098–2105 (2005).

  7. 7.

    , & Changed in translation: mRNA recoding by −1 programmed ribosomal frameshifting. Trends in Biochemical Sciences 40, 265–274 (2015).

  8. 8.

    , , & Reprogrammed genetic decoding in cellular gene expression. Mol Cell 13, 157–168 (2004).

  9. 9.

    , & Translational recoding in archaea. Extremophiles 16, 793–803 (2012).

  10. 10.

    , & Recoding: translational bifurcations in gene expression. Gene 286, 187–201 (2002).

  11. 11.

    , , & Identification of functional, endogenous programmed −1 ribosomal frameshift signals in the genome of Saccharomyces cerevisiae. Nucleic Acids Research 35, 165–174 (2007).

  12. 12.

    & Expression of peptide chain release factor 2 requires high-efficiency frameshift. Nature 322, 273–275 (1986).

  13. 13.

    , & ARFA: a program for annotating bacterial release factor genes, including prediction of programmed ribosomal frameshifting. Bioinformatics 22, 2463–2465 (2006).

  14. 14.

    & Ribosomal frameshifting in decoding antizyme mRNAs from yeast and protists to humans: close to 300 cases reveal remarkable diversity despite underlying conservation. Nucleic Acids Research 35, 1842–1858 (2007).

  15. 15.

    , , & Polyamine sensing by nascent ornithine decarboxylase antizyme stimulates decoding of its mRNA. Nature 477, 490–U147 (2011).

  16. 16.

    , & Competition between frameshifting, termination and suppression at the frameshift site in the Escherichia coli release factor-2 mRNA. Nucleic Acids Research 21, 5074–5078 (1993).

  17. 17.

    Mechanisms and implications of programmed translational frameshifting. Wiley interdisciplinary reviews. RNA 3, 661–673 (2012).

  18. 18.

    & Polyamines regulate the expression of ornithine decarboxylase antizyme in vitro by inducing ribosomal frameshifting. Proc Natl Acad Sci. USA 91, 3959–3963 (1994).

  19. 19.

    , , & Selection on the genes of Euplotes crassus Tec1 and Tec2 transposons: Evolutionary appearance of a programmed frameshift in a Tec2 gene encoding a tyrosine family site-specific recombinase. Eukaryotic Cell 2, 95–102 (2003).

  20. 20.

    , , , & Structures of the Euplotes crassus Tec1 and Tec2 elements - Identification of putative transposase coding regions. Gene 133, 71–78 (1993).

  21. 21.

    , & Analysis of micronuclear, macronuclear and cDNA sequences encoding the regulatory subunit of cAMP-dependent protein kinase of Euplotes octocarinatus: Evidence for a ribosomal frameshift. Journal of Eukaryotic Microbiology 48, 80–87 (2001).

  22. 22.

    , , & Programmed translational frameshifting is likely required for expressions of genes encoding putative nuclear protein kinases of the ciliate Euplotes octocarinatus. Journal of Eukaryotic Microbiology 48, 575–582 (2001).

  23. 23.

    et al. Euplotes telomerase contains an La motif protein produced by apparent translational frameshifting. Embo J 19, 6230–6239 (2000).

  24. 24.

    , , , & A novel protein kinase from the ciliate Euplotes raikovi with close structural identity to the mammalian intestinal and male-germ cell kinases: Characterization and functional implications in the autocrine pheromone signaling loop. Protist 161, 250–263 (2010).

  25. 25.

    , , & Characterization and expression of the gene encoding En-MAPK1, an intestinal cell kinase (ICK)-like kinase activated by the autocrine pheromone-signaling loop in the polar ciliate, Euplotes nobilii. International Journal of Molecular Sciences 14, 7457–7467 (2013).

  26. 26.

    et al. Developmentally programmed gene elimination in Euplotes crassus facilitates a switch in the telomerase catalytic subunit. Cell 113, 565–576 (2003).

  27. 27.

    , & Evolution of programmed ribosomal frameshifting in the TERT genes of Euplotes. Journal of Molecular Evolution 58, 701–711 (2004).

  28. 28.

    , & Oligomerization of the telomerase reverse transcriptase from Euplotes crassus. Nucleic Acids Research 30, 4032–4039 (2002).

  29. 29.

    & Next-generation transcriptome assembly. Nature Reviews Genetics 12, 671–682 (2011).

  30. 30.

    et al. The Oxytricha trifallax macronuclear genome: a complex eukaryotic genome with 16,000 tiny chromosomes. PLoS Biology 11, e1001473 (2013).

  31. 31.

    et al. The draft assembly of the radically organized Stylonychia lemnae macronuclear genome. Genome Biology and Evolution 6, 1707–1723 (2014).

  32. 32.

    , & TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25, 1105–1111 (2009).

  33. 33.

    et al. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nature Biotechnology 28, 511–515 (2010).

  34. 34.

    , & GeneTack database: genes with frameshifts in prokaryotic genomes and eukaryotic mRNA sequences. Nucleic Acids Research 41, D152–156 (2013).

  35. 35.

    et al. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Research 27, 29–34 (1999).

  36. 36.

    , , & WebLogo: A sequence logo generator. Genome Research 14, 1188–1190 (2004).

  37. 37.

    & Shifty ciliates: frequent programmed translatinal frameshifting in Euplotids. Cell 111, 763–766 (2002).

  38. 38.

    , , & Connection between stop codon reassignment and frequent use of shifty stop frameshifting. RNA 15, 889–897 (2009).

  39. 39.

    & tRNomics: analysis of tRNA genes from 50 genomes of Eukarya, Archaea, and Bacteria reveals anticodon-sparing strategies and domain-specific features. RNA 8, 1189–1232 (2002).

  40. 40.

    , , , & Identification of new eukaryotic tRNA genes in genomic DNA databases by a multistep weight matrix analysis of transcriptional control regions. Nucleic Acids Research 22, 1247–1256 (1994).

  41. 41.

    et al. The sup8 tRNALeu gene of Schizosaccharomyces pombe has an unusual intervening sequence and reduced pairing in the anticodon stem. Mol Gen Genet 197, 447–452 (1984).

  42. 42.

    , & Genes for the plastid elongation factor Tu and ribosomal protein S7 and six tRNA genes on the 73 kb DNA from Astasia longa that resembles the chloroplast DNA of Euglena. Mol Gen Genet 220, 425–432 (1990).

  43. 43.

    & Frameshift suppression: a nucleotide addition in the anticodon of a glycine transfer RNA. Nature: New Biology 242, 230–234 (1973).

  44. 44.

    , & Expanding the genetic code: selection of efficient suppressors of four-base codons and identification of “shifty” four-base codons with a library approach in Escherichia coli. J Mol Biol 307, 755–769 (2001).

  45. 45.

    , , & Insertion (sufB) in the anticodon loop or base substitution (sufC) in the anticodon stem of tRNA(Pro)2 from Salmonella typhimurium induces suppression of frameshift mutations. Nucleic Acids Research 20, 3463–3469 (1992).

  46. 46.

    Externally suppressible +1 “glycine” frameshift: possible quadruplet isomers for glycine and proline. Nature: New Biology 239, 219–221 (1972).

  47. 47.

    , & Nucleotide sequence of the SUF2 frameshift suppressor gene of Saccharomyces cerevisiae. Proc Natl Acad Sci. USA 79, 3565–3569 (1982).

  48. 48.

    & Codon recognition during frameshift suppression in Saccharomyces cerevisiae. Molecular and Cellular Biology 4, 2052–2061 (1984).

  49. 49.

    , , & Structural insights into +1 frameshifting promoted by expanded or modification-deficient anticodon stem loops. Proc Natl Acad Sci. USA 111, 12740–12745 (2014).

  50. 50.

    , , & Seven, eight and nine-membered anticodon loop mutants of tRNA (2Arg) which cause +1 frameshifting. Tolerance of DHU arm and other secondary mutations. J Mol Biol 228, 1042–1054 (1992).

  51. 51.

    , , , & Structural insights into translational recoding by frameshift suppressor tRNASufJ. RNA 20, 1944–1954 (2014).

  52. 52.

    Identification of cell surface polypeptides of the hypotrich ciliate Euplotes octocarinatus. Arch.Protistenkd 143, 311–318 (1992).

  53. 53.

    , & Genome sequence assembly using trace signals and additional sequence information. German Conference on Bioinformatics, 45–56 (1999).

  54. 54.

    & FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963 (2011).

  55. 55.

    et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. Journal of Computational Biology 19, 455–477 (2012).

  56. 56.

    Improved pairwise alignment of genomic DNA Doctor of Philosophy thesis, Pennsylvania State University (2007).

  57. 57.

    et al. The mitochondrial genomes of the ciliates Euplotes minuta and Euplotes crassus. BMC Genomics 10, 514 (2009).

  58. 58.

    & Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009).

  59. 59.

    & tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Research 25, 955–964 (1997).

  60. 60.

    et al. The COG database: an updated version includes eukaryotes. BMC Bioinformatics 4, 41 (2003).

  61. 61.

    A new generation of homology search tools based on probabilistic inference. Genome informatics. International Conference on Genome Informatics 23, 205–211 (2009).

  62. 62.

    et al. AUGUSTUS: ab initio prediction of alternative transcripts. Nucleic Acids Research 34, W435–439 (2006).

  63. 63.

    et al. Improving the Arabidopsis genome annotation using maximal transcript alignment assemblies. Nucleic Acids Research 31, 5654–5666 (2003).

  64. 64.

    , & EMBOSS: the European molecular biology open software suite. Trends in Genetics 16, 276–277 (2000).

  65. 65.

    et al. The Pfam protein families database. Nucleic Acids Research 36, D281–D288 (2008).

  66. 66.

    et al. The InterPro protein families database: the classification resource after 15 years. Nucleic Acids Research 43, D213–221 (2015).

  67. 67.

    , & BiNGO: a Cytoscape plugin to assess overrepresentation of gene ontology categories in biological networks. Bioinformatics 21, 3448–3449 (2005).

  68. 68.

    et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Research 13, 2498–2504 (2003).

  69. 69.

    , , , & KOBAS server: a web-based platform for automated annotation and pathway identification. Nucleic Acids Research 34, W720–724 (2006).

  70. 70.

    et al. +1 Frameshifting as a novel mechanism to generate a cryptic cytotoxic T lymphocyte epitope derived from human interleukin 10. The Journal of Experimental Medicine 195, 353–358 (2002).

  71. 71.

    , & A second mammalian antizyme: Conservation of programmed ribosomal frameshifting. Genomics 52, 119–129 (1998).

  72. 72.

    , , , & Discovery of a spermatogenesis stage-specific ornithine decarboxylase antizyme: antizyme 3. Proc Natl Acad Sci. USA 97, 4808–4813 (2000).

  73. 73.

    et al. Autoregulatory frameshifting in decoding mammalian ornithine decarboxylase antizyme. Cell 80, 51–60 (1995).

  74. 74.

    , , , & Polyamines regulate both transcription and translation of the gene encoding ornithine decarboxylase antizyme in mouse. European Journal of Biochemistry 250, 223–231 (1997).

  75. 75.

    , , , & Conservation of polyamine regulation by translational frameshifting from yeast to mammals. Embo J 19, 1907–1917 (2000).

  76. 76.

    et al. Two zebrafish (Danio rerio) antizymes with different expression and activities. Biochem J 345, 99–106 (2000).

  77. 77.

    , , , & The Drosophila gene for antizyme requires ribosomal frameshifting for expression and contains an intronic gene for snRNP Sm D3 on the opposite strand. Molecular and Cellular Biology 18, 1553–1561 (1998).

  78. 78.

    & Programmed translational frameshifting in a gene required for yeast telomere replication. Curr Biol 7, 969–976 (1997).

  79. 79.

    et al. Isolation and characterization of a novel actin filament-binding protein from Saccharomyces cerevisiae. Oncogene 16, 121–130 (1998).

  80. 80.

    , , , & Genome and proteome of Listeria monocytogenes phage PSA: an unusual case for programmed+1 translational frameshifting in structural protein synthesis. Molecular Microbiology 50, 303–317 (2003).

  81. 81.

    et al. Ribosomal frameshifting used in influenza A virus expression occurs within the sequence UCC_UUU_CGU and is in the +1 direction. Open Biology 2, 120109 (2012).

Download references


This project is supported by grants from National Natural Science Foundation of China (No. 31372199) to A. Liang.

Author information


  1. Key Laboratory of Chemical Biology and Molecular Engineering of Ministry of Education, Institute of Biotechnology, Shanxi University, Taiyuan 030006, China

    • Ruanlin Wang
    • , Wei Wang
    •  & Aihua Liang
  2. Key Laboratory of Aquatic Biodiversity and Conservation, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan 430072, China

    • Jie Xiong
    •  & Wei Miao


  1. Search for Ruanlin Wang in:

  2. Search for Jie Xiong in:

  3. Search for Wei Wang in:

  4. Search for Wei Miao in:

  5. Search for Aihua Liang in:


A.L. and R.W. conceived and conducted the project. J.X. and W.M. guided data analysis. J.X. and W.W. participated in correcting the manuscript critically. R.W. wrote the manuscript. All authors have read and approved the final manuscript.

Competing interests

The authors declare no competing financial interests.

Corresponding author

Correspondence to Aihua Liang.

Supplementary information

About this article

Publication history





Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.