Abstract
Nascent pre-tRNAs are transcribed by RNA polymerase III and immediately bound by La proteins on the UUU-3’OH sequence, using a tandem arrangement of the La motif and an adjacent RNA recognition motif-1 (RRM1), resulting in protection from 3’-exonucleases and promotion of pre-tRNA folding. The Tetrahymena thermophila protein Mlp1 has been classified as a genuine La protein, despite the predicted absence of the RRM1. We found that Mlp1 functions as a La protein through binding of pre-tRNAs and affecting processing in Tetrahymena thermophila and when expressed in fission yeast. However, unlike in other examined eukaryotes, depletion of Mlp1 results in 3’-trailer stabilization. We also observed that 3’-trailers in Tetrahymena thermophila are uniquely short relative to other examined eukaryotes, and that 5’-leaders have evolved to disfavour pre-tRNA leader/trailer pairing. Our data indicate that this variant Mlp1 architecture is linked to an altered, novel mechanism of tRNA processing in Tetrahymena thermophila.
Introduction
RNA polymerase III transcription of nascent pre-tRNAs terminates after the synthesis of a stretch of uridylates on a 3’-trailer extension (UUU-3’OH). La is the first protein to bind these nascent pre-tRNAs on the uridylate stretch and assists with pre-tRNA folding1. Once the nascent pre-tRNA has acquired a tRNA-like structure, the endonuclease RNase P removes the 5’-leader of La-bound pre-tRNA, followed by endonucleolytic cleavage of the 3’-trailer by RNase Z2,3. As a result of 3’-end cleavage, the La protein no longer associates with the tRNA and can be recycled for processing of new nascent pre-tRNAs4. In addition to the La-dependent pathway, an alternative La-independent pre-tRNA processing pathway exists but the order of pre-tRNA processing is reversed: without 3’-trailer protection by La, the exonuclease Rex1 digests the 3’-trailer of the pre-tRNA before RNase P endonucleolytic cleavage of the 5’-leader5, and misfolded pre-tRNAs are also more prone to degradation through nuclear surveillance6. Thus, La binding to pre-tRNAs is hypothesized to establish and determine the order of 5’-leader versus 3’-trailer processing3. Once the 5’-leader and 3’-trailer sequences are processed, the nucleotidyltransferase adds a CCA sequence to the discriminator base at the 3’-end of the mature tRNA which will serve as the site of amino acid charging7.
Genuine La proteins are members of the La-related proteins (LARPs)8. Most LARP family members, and all genuine La proteins, contain an N-terminal La module consisting of two adjacent RNA-binding domains: a La motif (LaM) and an RNA recognition motif-1 (RRM1) (Figure 1A)9–11. In Tetrahymena thermophila, the LARP7 ortholog p65 was until recently the only characterized LARP in this species8,12, but a recent study has grouped the Tetrahymena Macronucleus localized protein of unknown function (Mlp1)13 with the genuine La proteins, based on its primary sequence conservation in the LaM12. Interestingly, Mlp1 only contains a highly conserved LaM, unlike all previously studied La proteins which contain the tandem LaM-RRM arrangement. This atypical La protein has been identified in several members of the alveolates12.
The absence of the RRM1 in the La module of a genuine La protein in alveolates is highly unexpected, due to the mechanism by which La proteins bind the uridylate containing 3’-trailer. Structural studies of the hLa LaM-RRM bound to UUU-3’OH revealed that both domains sandwich this RNA, and functional studies have demonstrated that the La module is indispensable and sufficient for uridylate binding9,14, as deletion of either the LaM or RRM1 results in complete loss of binding9,10,15,16. Another unusual feature of this new type of La protein is the presence of a domain of unknown function-3223 (DUF3223) in the C-terminal region12. Thus, questions remain as to whether Mlp1 functions as a genuine La protein, and if so, whether it uses a mode of RNA binding dissimilar to previously studied La proteins, with possible associated changes in how Mlp1 directs pre-tRNA processing.
Here, we present evidence that despite the apparent lack of the RRM1, Mlp1 functions as a genuine La protein. Using ribonucleoprotein immunoprecipitation (RIP)-Seq of Mlp1, we show association with UUU-3’OH containing pre-tRNAs in vivo, and preferential binding of pre-tRNA substrates over mature tRNA substrates in vitro. Furthermore, we demonstrate that Mlp1 expression in Schizosaccharomyces pombe promotes pre-tRNA processing and tRNA mediated suppression but without typical La-associated 3’-end protection. Genetic depletion of Mlp1 in Tetrahymena reveals that in contrast to previously studied La proteins, Mlp1 destabilizes pre-tRNA 3’-ends, thus acting as a factor that accelerates 3’-end processing. In addition, when comparing pre-tRNAs found in Tetrahymena with other eukaryotic species, we found that the 3’-trailer sequences are considerably shorter, and that 5’-leader sequences have evolved accordingly to disfavor base pairing at these shortened 3’-trailers. Together, our data are consistent with a model in which Mlp1 fulfills many expected functions of a genuine La protein but uses alternate RNA binding modes to promote a pre-tRNA processing pathway in Tetrahymena that differs from those found in other studied eukaryotic systems.
Results
Mlp1 is predicted to lack an RRM adjacent to the LaM
To investigate primary amino acid sequence conservation, multiple sequence alignments were conducted for La proteins from different eukaryotic lineages. We confirm that residues important for uridylate binding in the LaM are mostly conserved in Tetrahymena (Figure 1B, Figure S1A,C-D). In contrast, we found little to no conservation of residues in the region of the adjacent domain where the RRM1 would be located (Figure S1B).
When comparing primary sequences of the La module between different LARPs, the presence of an RRM1 domain could often only be inferred from secondary structure predictions8. Therefore, we compared the secondary structure predictions for Tetrahymena La with secondary structures for different eukaryotes (Figure 1C). These consistently predicted an RRM fold immediately C-terminal to the LaM in all species examined, except for candidate alveolate La proteins.
Mlp1 preferentially binds pre-tRNAs in Tetrahymena thermophila
We hypothesized that should Mlp1 function as a genuine La protein, it should bind and promote the processing of RNA polymerase III transcripts, as has been demonstrated in budding and fission yeast2,3. We immunoprecipitated Mlp1-associated RNAs followed by detection by Northern blot (Figure 2A, Figure S2A). Using probes specific for pre-tRNA 3’-extensions, we confirmed that Mlp1 immunoprecipitation enriched pre-tRNA species relative to mature tRNAs, which were relatively more abundant in input RNA fractions. We also observed lesser enrichment of the U5 small nuclear RNA (snRNA), which is consistent with data indicating association of La proteins with processing intermediates of U5 in Saccharomyces cerevisiae17. We conclude that based on its expected RNA target cohort, Mlp1 behaves as expected for a genuine La protein.
To investigate Mlp1-associated RNAs more extensively, we sequenced Mlp1-immunoprecipitated RNAs after removing larger RNAs (> 300 nt) by gel electrophoresis. As expected for a genuine La protein, we observed an enrichment of reads mapping to pre-tRNA genes carrying 3’-uridylate extensions relative to their normalized abundance in size-matched input samples (Figure 2B – Mlp1). In contrast, mature tRNAs are not enriched in Mlp1-bound samples (Figure 2B – Mlp1; averaged enrichment values for tRNA isotypes presented in Figure S2B). We compared our dataset with a previous study in which hLa-bound tRNA reads were obtained by photoactivatable crosslinking and immunoprecipitation (PAR-CLIP) and where tRNA sequencing of input and immunoprecipitated RNA was done using hydro-tRNAseq18 and found a similar enrichment for 3’-uridylate containing pre-tRNAs (Figure 2B – hLa; averaged enrichment values for tRNA isotypes presented in Figure S2C). The relative enrichment of pre-tRNAs over mature tRNAs, are strongly indicative of Mlp1 functioning as a genuine La protein in Tetrahymena.
Mlp1 preferentially binds pre-tRNA substrates in vitro
The hLa protein preferentially binds pre-tRNA substrates through engagement of multiple sites, including the pre-tRNA body, 3’-trailer and 5’-leader4,19. To determine whether enrichment of Mlp1 associated pre-tRNAs relative to mature tRNAs correlated with changes in affinity for such ligands, we compared binding affinity of Mlp1 for in vitro transcribed mature tRNA versus pre-tRNA substrates using electromobility shift assays (EMSAs), as well as versions of these that included either or both of 5’-leader or 3’-trailer extensions (Figure 2C). We found that Mlp1 preferentially binds pre-tRNAs containing both 5’- and 3’-extensions, followed by pre-tRNA containing the 3’-trailer (Figure S2D,E, Table S1). Lowest binding affinities were consistently found for mature tRNAs, confirming the in vivo preferential binding of pre-tRNAs (see Figure S2D,E, Table S1).
To further investigate Mlp1 tRNA binding, we compared the affinity of full-length Mlp1 or Mlp1 mutants to hLa for radioactively labeled tRNA targets by EMSA (Figure 2D-I, Figure S3A, Table 1). To test the importance of 3’-uridylates binding, we compared Mlp1 to an Mlp1 mutant in which conserved amino acids predicted to recognize the UUU-3’OH motif were substituted (Q11A/Y14A)10,16 (see Figure 1B, Figure S1D, Q20/Y23 numbering in hLa), as well as a mutant in which the entire LaM was deleted (Mlp1 95-340). We found that the increased affinity for pre-tRNAs was lost in the Mlp1 Q11A/Y14A mutant (compare Figure 2E and F) as well as for the LaM deletion mutant (compare Figure 2E and G). In contrast, the C-terminal DUF3223 is not important to discriminate between pre-tRNA and mature tRNAs since removal of this domain (Mlp1 1-250) maintains the binding affinity difference (Figure 2H), while the Q11A/Y14A mutations in the context of the deleted DUF3223 (Mlp1 1-250 Q11A/Y14A) again resulted in decreased affinity for pre-tRNAs (Figure 2I). To investigate UUU-3’OH binding more directly, we compared these Mlp1 mutants in protein-RNA binding experiments using a 3’-trailer sequence CUGUGUUUU-3’OH and we found that the predicted uridylate binding residues located in the LaM (Q11A/Y14A) were required for binding (Figure 2J). Additionally, binding of 3’-uridylate RNA occurs with the same affinity for Mlp1 1-250 compared to full length Mlp1, whereas affinity for uridylate RNA is lost when both domains are used individually (Mlp1 1-95 and Mlp1 95-250) (Figure S3B). These data suggest that despite the apparent lack of the RRM1, Mlp1 functions in a similar manner as the hLa protein in preferentially binding pre-tRNA substrates, and that predicted, conserved uridylate binding residues in the LaM promote higher affinity binding associated with the UUU-3’OH motif.
The hLa protein is known to also interact with the 5’-triphosphate containing end of the nascent pre-tRNA through a short basic motif located in the C-terminal part of the protein19. We tested the affinity of the 5’-triphosphate for Mlp1 by competition EMSA, using a 5’-leader containing, 3’-trailer processed in vitro transcribed radioactively labeled tRNA (+5’PPP) and unlabelled competitor tRNAs with the 5’-triphosphate removed after phosphatase treatment (−5’PPP; +/- 5’PPP demonstrated in Figure S3C). We found that the unlabelled –5’PPP substrate competed poorly relative to an unlabelled +5’PPP substrate for the radioactive +5’PPP substrate on hLa, (Figure 2K – compare -5’PPP and +5’PPP hLa, Figure S3D) but the difference in competition between these RNAs on Mlp1 was smaller (Figure 2K – compare -5’PPP and +5’PPP Mlp1, Figure S3E). These data are consistent with a lesser degree of 5’-leader discrimination for 5’PPP in Mlp1 relative to hLa.
Mlp1 binding to 3’-trailer RNAs is different from hLa
To further study the distinct binding modes, we performed competition experiments between 32P-labeled uridylate RNA (U10) and unlabelled competitor pre-tRNA and mature tRNA substrates. We used hLa as a control and found that, as expected from previous work4, pre-tRNAs were a stronger competitor for the uridylate binding pocket than mature tRNAs (Figure 3A, Figure S4A). In contrast, uridylate RNA was competed off of Mlp1 using only low amounts of either pre-tRNA or mature tRNA (Figure 3B, Figure S4A), suggesting that binding of uridylates on Mlp1 is weaker compared to hLa.
Previous high-resolution structural characterization of hLa bound to UUU-3’OH established that the penultimate uridylate (UUU-3’OH) has the greatest importance for sequence specific, high affinity binding14,15. Previous high-resolution structural characterization of human La bound to UUU-3’OH established that the penultimate uridylate (UUU-3’OH) has the greatest importance for sequence specific, high affinity binding15,16. Notably, this uridylate is the only residue in the UUU-3’OH motif that contacts the RRM1 in hLa (Figure 3C). To compare the importance of the position of the penultimate uridylate (two nucleotides from the 3’-end: U-2), we performed competition EMSAs for hLa and Mlp1 using the radioactively labeled UUU-3’OH containing RNA CUGCUGUUUU-3’OH RNA (hence referred to as wild type 4U) and unlabelled RNAs carrying specific variations to this sequence. As expected, mutating the penultimate uridylate (U-2) into a cytidylate (U-2C – CUGCUGUUCU-3’OH) resulted in this RNA functioning as a poor competitor with hLa, whereas changes at the most 3’-terminal position (U-1C) and the third last position (U-3C) has a lesser effect on competition capability (Figure 3D, Figure S4B). Mutating the four last positions (4C – CUGCUGCCCC-3’OH) resulted in a total inability to compete with the wild type 4U RNA (Figure 3D, Figure S4B). In contrast, competition for Mlp1 between radioactively labeled 4U and any unlabelled variant RNA competitor (U-1C, U-2C and U-3C) showed similar competition levels, indicating that the position of the penultimate uridylate U-2 is not as important for interactions between Mlp1 and 3’-trailer sequences (Figure 3E, Figure S4C). Interestingly, unlabelled competitor 4C was more capable of competition with wild type 4U RNA for binding on Mlp1. These data are also consistent with Mlp1 discrimination for uridylate RNA not occurring as strongly as for hLa, and that the specificity of binding to the penultimate uridylate is weaker or absent for Mlp1.
The previous high-resolution work on hLa in complex with UUU-3’OH containing RNA also revealed the importance of the 2’- and 3’-hydroxyls for high affinity binding in the hydrophobic binding pocket between LaM and RRM115,16, with two hydrogen bonds formed with an aspartate (Figure 3F) that is conserved in Mlp1 (see Figure 1B, Figure S1D). To investigate the importance of the free 3’-terminal hydroxyl groups for Mlp1 binding, we used competition EMSAs comparing a regular U10 and a 3’-phosphorylated unlabeled competitor RNA. We found that for both hLa and Mlp1 the phosphorylated substrate makes a poor competitor, confirming that the presence of a 3’-hydroxyl end is important for interactions between Mlp1 and uridylates (Figure 3G,H, Figure S4D). Together, these results demonstrate that Mlp1 shows the same preference and LaM amino acid dependence for pre-tRNAs and UUU-3’OH binding as hLa, but that the altered architecture correlates with diminished discrimination for these substrates relative to hLa.
RNA chaperone function of Mlp1 in an Schizosaccharomyces pombe based heterologous system
In addition to 3’-end protection of nascent pre-tRNAs from exonucleases, La proteins also possess RNA chaperone activity to enhance correct folding of nascent pre-tRNAs4,20,21. We used a tRNA-mediated suppression assay to investigate the ability of Mlp1 to rescue a misfolded suppressor tRNA22. We transformed Sla1p, full-length Mlp1 and multiple Mlp1 mutants into a sla1-Schizosaccharomyces pombe strain (ySH9) which encodes a defective stop codon UGA-decoding suppressor tRNA (tRNA-SerUCA) as well as the ade6-704 allele, which is decoded by a tRNA-SerUCA to suppress red pigment accumulation during growth on low adenine. Successful suppression in this system relies on the presence of a La protein or other suitable RNA chaperone, resulting in white colonies versus red colonies in unsuppressed cells.
When comparing Sla1p-transformants to full-length Mlp1-transformants, we found that Mlp1 can stabilize the defective suppressor pre-tRNA similar to Sla1p (Figure 4A). Maturation of the defective suppressor tRNA can occur via 3’-terminal protection of uridylates from exonucleases, or general RNA chaperone activity, or a combination of these to assist with folding of pre-tRNAs6. To test the importance of the uridylate binding residues for suppression, we compared the Mlp1 Q11A/Y14A mutant and the LaM deletion mutant (Mlp1 95-340) to wild type Mlp1 and observed a pink phenotype indicating intermediate suppression levels despite equal levels of protein expression (Figure 4A, Figure S5A), suggesting that these uridylate binding residues function in maturation of the pre-tRNA in vivo. Next, we studied the function of the C-terminal DUF3223 and found that removal of this domain (Mlp1 1-250) led to near complete loss of suppression (Figure 4A), and combination of DUF3223 removal and uridylate binding inactivation (Mlp1 1-250 Q11A/Y14A) resulted in a complete loss of suppression (Figure 4A). These results indicate that to obtain complete tRNA mediated suppression, both the conserved uridylate binding residues (Q11 and Y14) in the LaM and the C-terminal DUF3223 are required.
To investigate 3’-end protection more directly, we extracted total RNA from Mlp1- and Sla1p-expressing strains and analysed suppressor pre-tRNA SerUCA processing by Northern blot (Figure 4B). We found that higher levels of suppressor pre-tRNA stabilization in Sla1p- and Mlp1-transformed strains (lanes 2, 3) corresponded to more mature suppressor tRNA which correlates with the white phenotype observed in the tRNA-mediated suppression results (Figure 4A), as opposed to the lack of pre-tRNA stabilization and subsequent mature suppressor tRNA in pRep4 control (lane 1) resulting in a red phenotype (Figure 4A,B). Interestingly, the most abundant pre-tRNA species was slightly smaller in Mlp1 transformed cells relative to Sla1p (see asterisk). We also detected endogenous pre-tRNAs LysCUU processing intermediates using an intron, 5’-leader and a 3’-trailer probe (Figure 4C). The presence of the three La-dependent pathway pre-tRNA intermediates are detected after transformation of the positive control Sla1p23 (Figure 4C – intron probe, compare lanes 1 and 2). Notably, we observed the appearance of a distinct pre-tRNA intermediate in Sla1 expressing cells when using a probe specific for the 5’-leader of pre-tRNA LysCUU that did not co-migrate with any of the major species observed using the intron or 3’-trailer probe (Figure 4C – 5’-leader, indicated by a double asterisk, beneath the full length pre-tRNA band). We hypothesize that this species represents a subset of pre-tRNAs that are not stably bound by Sla1p and whose 3’-ends have been nibbled by a 3’-exonuclease, yet have retained their 5’-leaders as has been described during La-independent pre-tRNA processing. As expected from the tRNA-mediated suppression assay, we found that Mlp1 also stabilized endogenous precursor pre-tRNA LysCUU, however, this precursor tRNA co-migrated with the 5’-leader containing, 3’-trailer exonuclease processed intermediate observed in Sla1p -transformants (Figure 4C – 5’-leader, compare lane 2 and 3), consistent with Mlp1 stabilizing a 5’-leader containing, 3’-processed or nibbled pre-tRNA intermediate. As expected, all Mlp1 mutants defective in tRNA-mediation suppression were defective in stabilizing pre-tRNA intermediates. Together, these data are consistent with Mlp1 engaging pre-tRNAs and promoting tRNA-mediated suppression in Schizosaccharomyces pombe, but with impaired protections of pre-tRNA 3’-ends relative to Sla1p.
To further compare Mlp1 and Sla1p function in the processing of pre-tRNA intermediates by Mlp1 in Schizosaccharomyces pombe, we immunoprecipitated Sla1p and Mlp1 ribonucleoprotein-complexes (Figure S5B) and sequenced the 3’-ends of their associated pre-tRNA LysCUU and pre-tRNA TyrGUA by 3’-rapid amplification of cDNA ends (3’RACE). While Sla1p-associated pre-tRNAs were enriched for species containing primarily four and five nucleotide long uridylate 3’-trailers, as has been described previously24, Mlp1-associated pre-tRNAs were largely depleted for 3’-trailer containing species (Figure 4D), consistent with the Northern blotting results (Figure 4B,C). These data are consistent with our Mlp1 RNA binding data in vitro, indicating that Mlp1 promotes pre-tRNA processing but does not offer as strong protection of the 3’-ends of associated pre-tRNAs as other examined genuine La proteins.
Reduced Mlp1 expression in Tetrahymena thermophila leads to impaired 3’-trailer processing
To investigate the effect of Mlp1 depletion on pre-tRNA processing, we generated a partial MLP1 knockout strain and confirmed genomic integration of the selection marker by Southern blot and PCR (Figure S6A,B and data not shown) and reduced protein expression of Mlp1 by Western blot (Figure 5A) and indirect immunofluorescent staining (Figure S6C). Notably, complete deletion of the MLP1 locus in Tetrahymena macronuclei through increasing drug selection (phenotypic assortment) was not achieved, indicating that Mlp1 is likely essential in this system, similar to Mus musculus, Drosophila melanogaster, Trypanosoma brucei and Arabidopsis thaliana25–28. We extracted total RNA followed by pre-tRNA intermediate detection by Northern blot for pre-tRNA IleUAU, LeuUAA and ValAAC. We found that in absence of Mlp1, a 3’-trailer containing pre-tRNA intermediate was stabilized (Figure 5B – intron probe, bottom band). These data provide evidence for Tetrahymena being the first eukaryote in which La protein levels correlate inversely with 3’-trailer stabilization. These results are also consistent with our heterologous expression data in Schizosaccharomyces pombe indicating impaired protection of pre-tRNA 3’-trailers by Mlp1. These data suggest that the variant domain architecture of Mlp1 relative to other genuine La proteins is associated with an altered pre-tRNA processing pathway in this system.
To study the effect of Mlp1 depletion on a transcriptome-wide scale, we performed total RNA-sequencing of small RNA-enriched samples, similar to our Mlp1 RIP-Seq (see Figure 2B), in which counts for each pre-tRNA isoacceptor were determined for each unique uridylate tail length at the 3’-end of the tRNA. We found that the average length of the 3’-uridylate tail increased for the majority of tRNA species as a result of Mlp1 depletion (Figure 5C,D). These findings indicate that Mlp1 promotes removal of the 3’-trailer and that when Mlp1 is limiting, the 3’-trailer sequences are not processed as efficiently. Together, these data are consistent with Mlp1 accelerating 3’-end processing, relative to other examined species in which genuine La proteins stabilize 3’-trailers.
Previous work has demonstrated that in eukaryotes, the number of genomic tRNA gene copies correlates strongly with the expression of mature tRNA transcripts and thus can function as a predictor for tRNA expression levels29,30. We observed that in Tetrahymena the number of genomic tRNA genes correlates with the expression levels of tRNAs (R2 = 0.55, r = 0.75) (Figure 5E). To investigate if depletion of Mlp1 results in changes of total tRNA levels, we performed the same analysis for our Mlp1 partial knockout strain (R2 = 0.53, r = 0.73) (Figure 5F) and found that when comparing tRNA expression between WT and partial KO strains the correlation was almost perfect (R2 = 0.94, r = 0.97) (Figure 5G), indicating that Mlp1 depletion did not influence mature tRNA expression. Northern blot analysis for representative tRNAs confirmed equal amounts of mature tRNA between WT and partial KO strains (Figure 5H).
The 3’-trailer lengths are shorter in Tetrahymena thermophila compared to other examined eukaryotes
To explore tRNAs and pre-tRNA processing in Tetrahymena more extensively, we compared their predicted pre-tRNA architecture to those from other eukaryotic species. First, we compared 3’-trailer lengths of each pre-tRNA, as determined by the number of nucleotides between the discriminator base and the genomic stretch of at least four consecutive thymines31 that give rise to the 3’UUU-OH motif in the nascent transcript. We found that Tetrahymena has shorter 3’-trailer lengths than any other species analyzed, with the most common 3’-trailer length being zero nucleotides (Figure 6A, Table 2). A similar analysis for the total length of the mature tRNA revealed that, as expected, the average length of mature tRNAs is the same as other examined species (Figure 6B, Table 2). We previously noted differences in enrichment of pre-tRNAs isotypes (Figure S2B – pre-tRNAs) in our Mlp1 RIP-seq results and hypothesized that the 3’-trailer length could be a determinant for binding affinity. We plotted fold enrichment for pre-tRNAs between Mlp1-immunoprecipitated and input tRNAs against 3’-trailer length and found that these are moderately negatively correlated (R2 = 0.14, r = -0.38) (Figure S7A), suggesting that nascent pre-tRNAs containing a longer 3’-trailer sequence may have a slightly lower binding affinity for Mlp1. When comparing this to the hLa data, we found a lesser correlation (R2 = 0.05, r = -0.22) (Figure S7B).
Processing of the 5’-leader by RNase P is based on tertiary structure recognition of the tRNA followed by endonucleolytic cleavage at the –1/+1 position32,33. Previous studies demonstrated structural conservation of the catalytically active RNA in RNase P in Tetrahymena, indicating that processing occurs in an similar manner33,34. Optimal efficiency of RNase P cleavage is dependent on the presence of a bulge that includes the last nucleotide of the 5’-leader (the N–1 position; Figure 6C), as extensive base pairing between 5’-leader and 3’-trailer sequences inhibits 5’-leader cleavage35. We hypothesized that Tetrahymena tRNAs should have adapted 5’-leader sequences to ensure the generation of a mismatched bulge against the shorter 3’-trailer sequences. We determined the most common 5’-leader sequences by logo generation for Tetrahymena thermophila, Schizosaccharomyces pombe and Homo sapiens and found that the most frequent nucleotide at the most 3’-residue of the 5’-leader sequence (N-1) in Tetrahymena is an adenine (± 75%) (Figure 6D, Figure S7C). In contrast, the distribution of nucleotides at this position is more diverse in other species (Figure 6D, Figure S7C). At the 3’-end of the mature transcript, the discriminator base (N73) is mostly adenine (± 50%), followed by guanine as second most frequent nucleotide (± 25%) (Figure S7D) which is common for all eukaryotes in this study.
We then split our data based on the identity of the discriminator base (N73) (Figure 6E, Figure S7E), which can pair with the first nucleotide in the 5’-leader (N-1). We observed a strong lack of Watson–Crick base pairing for Tetrahymena between the opposing nucleotides in the discriminator base pairing with the typical adenine (A73 – A-1; C73 – A-1 and G73 – A-1). Conversely, tRNAs containing a thymine as discriminator base (T73) strongly avoided an adenine as the N-1 5’-leader base to avoid base pairing. This partitioned discrimination is not evident for Schizosaccharomyces pombe and Homo sapiens (Figure 6E). Since shortened 3’-trailers in Tetrahymena should have a greater dependence on leader-trailer bulges occurring through the discriminator base, these data are thus consistent with the 3’ most nucleotide of the 5’-leader having a greater evolutionary pressure to avoid base pairing with the discriminator base which is seen most strongly for Tetrahymena (Figure S7F) relative to other examined species.
Discussion
Previously studied genuine La proteins contain a conserved tandem arrangement of a LaM and RRM1 collectively referred to as a La module. Previous phylogenetic predictions12 and our continued computational analysis have indicated that the previously characterized Tetrahymena protein Mlp1 may group with the genuine La proteins, despite the predicted lack of the La module’s RRM1 domain. In addition to the absence of the RRM1, Mlp1 is predicted to have a previously uncharacterized domain of unknown function (DUF3223), which is absent in all other examined members of the LARP superfamily. Since this arrangement of a genuine La protein is unprecedented, it was not clear how this protein might perform La associated functions and what effects this might have, if any, on the processing of La RNA targets.
In this work we present data consistent with Mlp1 functioning as a genuine La protein. Using in vitro binding assays, we demonstrate that the identity of the Mlp1 residues Q11 and Y14, analogous to UUU-3’OH binding residues in hLa, are also required for high affinity binding of 3’-uridylate containing trailer sequences. We demonstrate that binding of uridylate RNA is retained when both the predicted LaM (Mlp1 1-95) and middle domain (Mlp1 95-250) are included, however, uridylate binding does not occur when either the LaM (Mlp1 1-95) or the middle domain (Mlp1 95-250) are tested in isolation. Given the relative paucity of contacts from the RRM relative to the LaM during hLa binding to UUU-3’OH, it seems likely that the 95-250 region might serve an analogous function, making contacts important for UUU-3’OH binding despite the absence of the RRM fold.
While the Mlp1 LaM and middle domain combine to support UUU-3’OH binding similar to the classic La module, other key differences exist. Using competition experiments, we found that short UUU-3’OH containing trailers were more easily displaced from Mlp1 by pre-tRNA and mature tRNA substrates, relative to hLa, suggesting that Mlp1 binding to UUU-3’OH may be less stable compared to La proteins with the classic LaM-RRM1 arrangement. The previous hLa-UUU-3’OH co-crystals revealed that RRM1 only makes a single contact with the penultimate uridylate RNA (U-2) during UUU-3’OH binding, and it is this uridylate that is recognized with the greatest specificity. Unlike hLa, the U-2C RNA was no less effective a competitor against the 4U RNA than the U-3C and U-1C RNAs when Mlp1 was tested. The impaired ability of Mlp1 to discriminate the U-2 residue, as well as uridylates more generally, suggest that the lack of the RRM1 results in a relatively lower ability of Mlp1 to differentiate UUU-3’OH containing RNAs.
Previous work has demonstrated that the La module (LaM+RRM1) of different LARPs, and more specifically their RRM regions, possess RNA chaperone activity in the absence of 3’-uridylate protection from exonucleases20,21. Using tRNA-mediated suppression in a La null strain (sla1-) of Schizosaccharomyces pombe, we demonstrated that while the uridylate binding residues in the LaM of Mlp1 are important for suppression, the DUF3223 region also promotes the correct folding of defective pre-tRNAs. These data raise the possibility that the DUF3223 region of Mlp1 may serve an analogous function in RNA chaperone activity previously associated with the RRM1 and other C-terminal regions of genuine La proteins.
However, genuine La proteins are well known to also stabilize 3’-trailer containing intermediates through high-affinity binding of the 3’-uridylate tail and thereby providing protection against 3’-exonucleases. Using Northern blotting and RIP-3’-RACE, we demonstrated that while Schizosaccharomyces pombe La stabilizes 5’-leader, 3’-trailer, and intron-containing pre-tRNA intermediates, Mlp1 stabilized a 5’-leader and intron containing, but 3’-trailer lacking, pre-tRNA species, suggesting that Mlp1 may accelerate 3’-processing of nascent pre-tRNAs, relative to other La proteins. This could imply that binding of Mlp1 to the 3’-ends of the nascent pre-tRNAs is not as strong as compared to Sla1p, leaving the RNA exposed to 3’-exonucleases. Alternatively, Mlp1 could increase access of the 3’-trailer to the endonuclease RNase Z.
To better understand how the altered architecture of Mlp1 might influence pre-tRNA processing, we generated a partial MLP1 knockout strain in Tetrahymena and performed RNA-Seq and Northern blots of endogenous tRNA species. We found that, consistent with our Mlp1 data from Schizosaccharomyces pombe, Mlp1 appears to promote the removal of 3’-trailers, as depletion of Mlp1 in vivo leads to a greater relative abundance of UUU-3’OH trailer extensions. In yeast, depletion of La leads to exonucleolytic nibbling and a lower abundance of UUU-3’OH ends, while in the presence of La the 3’-end is stabilized until endonucleolytic cleavage by RNase Z directly 3’ to the discriminator base (N73) (Figure 7).
Previous work has demonstrated that hLa impedes access by RNase Z, delaying the 3’-endonucleolytic and resulting in 5’-leader processing proceeding 3’-trailer processing36. We hypothesize that the presence of an RRM1 in hLa could be contributing to this function, seeing that in Mlp1, a La protein lacking the RRM1, the 3’-processing appears to not be blocked. Thus, the apparent destabilization of 3’-trailers by Mlp1 in Tetrahymena suggests that the function of La in this system, differing from other examined La proteins, may be to increase access of the pre-tRNA 3’-trailer to the tRNA processing machinery (Figure 7). The difference in processing of pre-tRNAs is likely caused by Mlp1 since the 3’-exonuclease Rex1p and 3’-endonuclease RNase Z are conserved in Tetrahymena (Figure S8A,B). An alternate but not mutually exclusive hypothesis may be that reduction of Mlp1 levels may alter access of pre-tRNA 3’-ends to the Lsm2-8 complex, and that this may be linked to 3’-trailer accumulation relative to wild-type cells, as previous work in yeast has also linked the Lsm2-8 complex to the processing of pre-tRNA 3’-ends37,38. To confirm that the Lsm2-8 complex functions similar as in other eukaryotic systems, we compared primary amino acid sequence alignments and found that the LSm2-8 complex in Tetrahymena appears to be conserved in Tetrahymena, indicating similar functionality (Figure S8C,D). We also observed that Mlp1 depletion did not lead to changes in mature tRNA expression levels, suggesting that an Mlp1-independent tRNA maturation pathway also likely exists in Tetrahymena, similar to budding and fission yeast.
Along with the alternate Mlp1-associated tRNA processing in Tetrahymena, we observed several differences in features of pre-tRNA 5’-leaders and 3’-trailers compared to other eukaryotic species. Our genome-wide 3’-trailer length analysis in eukaryotes revealed that Tetrahymena pre-tRNAs have very short 3’-trailer sequences. Since 3’-trailers in Tetrahymena are dominated by the uridylate stretch, we studied the prevalence of the 3’-most terminal residue of the 5’-leader (N-1) and found that Tetrahymena pre-tRNAs appear to have evolved to avoid perfect matches with the discriminator base (N73) (preceding the uridylate tail), ensuring a lack of complete base pairing to enhance RNase P cleavage in the context of a minimized 3’-trailer sequence. Further upstream in the 5’-leader (N-2) is often an adenosine, but then typically uridylates, which would not be expected to pair with the 3’UUU-OH motif. Thus, 5’-leaders and 3’-trailers in Tetrahymena may have evolved to have minimal base pairing between pre-tRNA 5’-leaders and 3’-trailers, compared to other eukaryotes in which a bulge proximal to the mature tRNA ends is often followed by a paired region closer to the 5’-leader and 3’-trailer extremities. For example, human pre-tRNAs have an approximately equal distribution of N-1 nucleotide for discriminator bases A, T and G. The most prevalent N-1 nucleotide for discriminator bases A and T, are T and A respectively, indicating that human pre-tRNAs, which generally contain a longer 3’-trailer sequence, have more flexibility in N-1 sequence to introduce a mismatch bulge at the N-1 site for optimal 5’-leader processing by RNase P. It will be interesting to further investigate whether this alternate arrangement is linked to the altered functional roles described here for Mlp1 in this system.
Materials and Methods
Conservation analysis
Accession numbers used to obtain primary amino acid sequences from the National Center for Biotechnology Information (NCBI) for La, Rex1p, RNase Z/ELAC 2 and LSm2-8 complex are shown in Table S6 and primary amino acid alignments were obtained using Clustal Omega (EMBL-EBI)39, followed by analysis using a custom python script to annotate identical and conserved amino acids using human protein sequences as a reference. Amino acids were grouped as conserved based on side chain diversity: (1) Asp (D), Glu (E), Asn (N), Gln (Q); (2) Lys (K), Arg (R), His (H); (3) Phe (F), Trp (W), Tyr (Y); (4) Val (V), Ile (I), Leu (L), Met (M); and (5) Ser (S), Thr (T).
Secondary structure predictions were obtained using Phyre240 and color coded based on predicted β-sheet or α-helices. High-resolution structures of the LaM of Homo sapiens, Trypanosoma brucei, and Dictyostelium discoideum La proteins were obtained from PDB: 2VOD, 1S29, and 2M5W, respectively. Structure of the La module (LaM+RRM1) in complex with uridylate RNA of Homo sapiens La was obtained from PDB: 2VOD. Structure prediction of the LaM of Tetrahymena thermophila was obtained using the lomets2 tool41. Structure of the LSm2-8 complex in complex with uridylate RNA of Schizosaccharomyces pombe and Saccharomyces cerevisiae were obtained from PDB: 6PNN and 4M7D, respectively. Modeling was done using PyMOL.
DNA constructs
The MLP1 coding sequence encoding Tetrahymena thermophila La protein (NCBI Reference Sequence: XP_001019287.2) was obtained from the Tetrahymena Genome Database http://www.ciliate.org/42, and codon optimized for expression in E. coli. gBlocks of the optimized codon sequence were ordered from integrated DNA technologies (IDT) (sequence can be found in Table S5).
Mlp1 and Mlp1 mutants used for in vitro electromobility shift assays (EMSAs) were cloned in the NheI and BamHI restriction enzyme sites of the pET28a vector using the plasmid encoded N-terminal 6XHis-tag for protein purification. pET28a hLa was previously cloned in the NcoI and BamHI sites including a reverse primer-encoded 6X-His tag. Mlp1 and Mlp1 mutants used for Schizosaccharomyces pombe transformations were cloned in the SalI and BamHI sites in pRep4 plasmid (ura+) incorporating a 6X-His tag N-terminally in the forward primer during PCR amplification. All primers are listed in Table S5.
T7 DNA templates for in vitro transcription were generated by PCR amplification using pre- or mature tRNA specific primers (listed in Table S5) to obtain DNA templates containing an upstream T7 promotor. The DNA template was gel purified using 7 M urea denaturing 10% polyacrylamide gel and DNA was eluted from the gel by overnight rotation in 150 mM sodium acetate, in 50% phenol:chloroform:isoamylalcohol at 4°C. The aqueous layer obtained by centrifugation at 20,000 g for 10 minutes at 4°C was ethanol precipitated overnight at -80°C.
Tetrahymena thermophila cultivation and knockout strain generation
Liquid cultures were grown to mid-log phase (0.1 - 1 x 106 cells/mL) at 30°C shaking at 90 RPM in SPP media (1% proteose peptone, 0.1% yeast extract, 0.2% glucose, 0.003% Fe-EDTA)43. Cell pellets were harvested by centrifugation for 3 minutes at 1000g. Following two washes in 10 mM Tris-HCl, pH 7.4, pellets were stored at -80°C.
PCR-amplification from Tetrahymena thermophila genomic DNA to obtain flanking 5’- and 3’-regions was completed using primers containing KpnI/XhoI and NotI/SacI, respectively. The flanking regions of the MLP1 gene (http://www.ciliate.org/ - TTHERM_00384860) were cloned into pNeo4 plasmid44 flanking the paromomycin (Neo4) drug resistance cassette using restriction enzyme sites KpnI XhoI and SacI NotI, respectively. The Neo4 cassette is located downstream of the CdCl2 inducible metallothionein (MTT1) promoter. The resulting pNeo4 MLP1 knockout plasmid DNA construct was linearized using the ScaI restriction enzyme prior to transformation. Biolistic transformation of Tetrahymena thermophila was performed as described previously45. Integration of the DNA construct is based on homologous recombination and transformants were grown under increasing concentration of paromomycin starting at 60 μg/mL to a final concentration of 1000 μg/mL (phenotypic assortment)46. Correct integration was determined using Southern blot and PCR.
Protein isolation from Tetrahymena thermophila
Tetrahymena thermophila cell pellets were resuspended in 10% Trichloroacetic acid (TCA) in PBS, followed by incubation at –20°C overnight to enhance protein precipitation. The white fluffy protein pellet was collected by centrifugation for 15 minutes at 10,000g at 4°C, then washed twice with 100% ice-cold acetone. The pellet was air-dried and resuspended in 50μL/mL culture 2.5X SDS loading dye [5X SDS loading dye: 5% β-mercaptoethanol (v/v), 0.02% bromophenol blue (w/v), 30% glycerol (v/v), 10% sodium dodecyl sulfate (SDS) (w/v), 250 mM Tris-HCl, pH 6.8].
RNA-immunoprecipitation from Tetrahymena thermophila
Tetrahymena thermophila cell pellets from 100 mL cultures were collected at log phase (0.1 - 1 x 106 cells/mL) and washed twice with 1X PBS. The cells were cross-linked with 1% formaldehyde for 10 minutes followed by quenching with 0.25 M glycine for 5 minutes at room temperature. Cells were washed twice in 1X PBS before lysis in 2 mL buffer A [30 mM Tris-HCl pH 7.4, 150 mM NaCl, 20 mM KCl, 2 mM MgCl2, 0.1% Triton-X100, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1X Halt Protease Inhibitor Cocktail (PIC) (ThemoFisher Scientific), 100U SUPERase In RNase Inhibitor (ThermoFisher Scientific)] followed by sonication (25% amplification with 15s intervals for 4 minutes). The lysate was clarified by centrifugation at 20,000 g for 45 minutes at 4°C and treated with 10 U TurboDNase. The lysates were pre-cleared using rabbit isotype immunoglobin G (IgG) control-bound Protein G Dynabeads rotating for 1 hour at 4°C. Immunoprecipitations were performed using an affinity purified rabbit anti-Mlp1 antibody (ThermoFisher Scientific – Custom Antibodies) and a rabbit IgG control coupled to Protein G Dynabeads rotating 1 hour at 4°C. The Protein G Dynabeads were washed 5 times using buffer A (without PIC, PMSF and RNase inhibitor). Input and eluted RNA was isolated following reverse cross-linking for 45 minutes at 70°C in buffer B [50 mM Tris-HCl pH 7.4, 5 mM EDTA pH 8.0, 10 mM DTT, 1% SDS], followed by Trizol extraction.
tRNA library preparation and TGIRT sequencing and analysis
Total and Mlp1-associated RNAs were size-selected (20-300 nucleotides) on a 7 M urea denaturing 10% polyacrylamide gel. RNA was eluted from the gel by overnight rotation in 150 mM sodium acetate, pH 5.1 in 50% phenol:chloroform:isoamylalcohol at 4°C. The aqueous layer obtained by centrifugation at 20,000g for 10 minutes at 4°C was ethanol precipitated overnight at -80°C. The cell pellet was washed once with 70% ethanol, air dried and resuspended in RNase-free H2O. To promote sequencing of highly structured tRNA species carrying modifications that are inhibitory to cDNA synthesis, we used TGIRT-III reverse transcriptase (InGex) template-switching to prepare cDNA libraries as described previously47,48. Libraries were sequenced on a NextSeq500 platform (Illumina) (2x 75 bp).
tRNA-Seq processing pipeline
Details about tools and parameters of bioinformatics analyses are regrouped in a reproducible Snakemake workflow that can be found at https://github.com/etiennefc/t_thermophila_RNA_Seq.git and are also described below. The raw data (.fastq files) are available for download on the Gene Expression Omnibus (GEO) under the accession number XXX. Briefly, paired-end reads were trimmed using Trimmomatic v0.3649, and FastQC v0.11.5 was used to evaluate read quality before and after trimming, as described previously47. Resulting trimmed reads were then aligned stringently (i.e. allowing no mismatch) to the Tetrahymena thermophila genome assembly version T_Thermophila_MAC_202150 (accessible on the Tetrahymena Genome Database at http://www.ciliate.org/system/downloads/1-upd-Genome-assembly.fasta) using STAR v2.6.1a51 (with the parameters described previously47 and also the following parameter and value: --outFilterMismatchNmax 0. The index needed to align reads to the genome was produced using STAR v2.6.1a as described previously47 using the genome assembly described previously and a custom Tetrahymena thermophila annotation (.gtf) file available at https://zenodo.org/record/6391187#.YkH9v-fMK3A. The annotation file was built by converting .gff files (one for protein-coding genes, one for tRNA genes and one for 5S ribosomal RNA genes; these .gff files are also available at https://zenodo.org/record/6391187#.YkH9v-fMK3A) into .gtf files and by concatenating these files into one final (.gtf) annotation file using custom bash scripts. This annotation was corrected for embedded genes using CoCo v0.2.5.p152 with the correct_annotation mode and default parameters. Counts were attributed to genes and normalized as transcripts per million (TPM) as previously described47 using CoCo v0.2.5.p1 with the correct_count mode. Bedgraph files were generated using again CoCo v0.2.5.p1 (with the correct_bedgraph mode with default parameters). Differential expression analysis was performed using DESeq253 with default parameters and the count output of CoCo correct_count to compare the WT, Mlp1 IP and Mlp1 partial Mlp1 KO samples. Normalized counts measured in TPM and differential expression data for tRNAs are shown in Additional Table 1.
tRNA read fishing and binning into pre-mature and mature tRNAs
Raw counts for pre-tRNA (3’-UUU) and mature tRNAs (3’-CCA) were generated using custom python scripts. A list of unique sequences was generated for each tRNA isoacceptor (e.g. Gln TTG: AATCCTCTGACCTGGGTTCGAATCCCAGTACGACCT) (Additional Table 2) and used to obtain (“fish”) all reads from the unmapped raw sequence files (.fastq file format). Each sequence was grouped in corresponding bins based on the 3’-end of the reads: -CCA (mature tRNA), 1U, 2U, 3U, 4U, 5U, 6U, 7U, 8U, 9U or 10U (premature tRNA). Raw counts for each tRNA isotype were normalized as counts per million (CPM) by dividing raw counts by the total number of fished read per bin for each replicate divided by 106 (Table S2 and Table S3). Fold-enrichment for Mlp1-bound tRNAs was calculated after summation of pre-tRNA counts, followed by taking the ratio of CMP data for Mlp1-immunoprecipitated tRNAs and WT input tRNAs and displayed after log2 transformation in heatmaps (Table S2). Fold-enrichment for hLa-bound tRNAs from 18 was obtained from the GEO database under accession number GSE95683 where counts for pre-tRNAs and mature tRNAs were normalized as CPM and transformed identically as Mlp1-bound tRNAs. The cumulative abundance in CPM for each pre-tRNA isoacceptor was calculated for WT and partial KO tRNAs and displayed after log10 transformation in heatmaps (Table S3).
Electromobility shift assays (EMSA)
U10 and CUGCUGUUUU (20 pmol) were chemically synthesized (Integrated DNA Technologies (IDT)) and 20 pmol RNA was radioactively labeled using [γ-32P]-ATP (PerkinElmer, 10mCi/ml) and 10 units T4 Polynucleotide Kinase (PNK) enzyme (New England Biolabs, cat#M0201S) for 2 hours at 37°C in 1X T4 PNK buffer (New England Biolabs, cat#B0201S). Radioactively labeled tRNAs were produced by T7 in vitro transcription in the presence of [α-32P]-UTP (PerkinElmer, 10mCi/ml) using PCR products containing an upstream T7 promotor (see DNA constructs). Dephosphorylation of the 5’-triphosphate pre-tRNA was done using 5 units QuickCIP (New England Biolabs, cat#M0525S) for 30 minutes at 37°C in 1X rCutSmart Buffer (New England Biolabs, cat# B6004S), followed by Trizol extraction. All radioactively labeled RNAs were PAGE purified and eluted in 0.5M NaCl overnight at room temperature.
His-tagged proteins hLa, Mlp1 and Mlp1 mutants were purified from Escherichia coli BL21 cells using Co2+ beads, followed by heparin column purification. The proteins were buffer exchanged in 1X PBS and quantified using Bovine Serum Albumin (BSA) quantifications on SDS-PAGE gels.
EMSAs were performed as described54. Briefly, 3000 CPM of radioactive RNA substrates (∼ 0.1 nM) were incubated in 1X EMSA buffer [10% glycerol, 20 mM Tris-HCl, pH 7.4, 100 mM KCl, 1 mM EDTA, 5 mM β-mercaptoethanol and bromophenol blue] at 95°C for 5 minutes, followed by slow cooling to room temperature. For competition EMSAs, unlabelled RNA was added to the radioactively labeled RNA prior to incubation at 95°C. The concentration of unlabelled RNA depends on the protein concentration used to bind >85% of the radioactively labeled RNA. Protein was added and incubated for 30 minutes at 30°C. Then, the protein-RNA complexes were immediately snap cooled on ice for 5 minutes and separated on a 8% native polyacrylamide gel at 4°C at 100V. The gels were dried for 45 minutes at 80°C on a Gel Dryer (Bio-Rad) and exposed to a storage phosphor screen overnight. The screens were developed on a Typhoon. Quantification of bound and free RNA was done using ImageJ and binding curves were fit using a nonlinear specific binding curve fitting program and Kd values were calculated (GraphPad Prism).
tRNA mediated suppression assay in Schizosaccharomyces pombe
tRNA mediated suppression assays were performed as described previously22. In brief, the sla1-Schizosaccharomyces pombe ySH9 strain encoding a defective UGA-decoding suppressor tRNA (tRNA-SerUCA) and the ade6-704 allele was transformed using a pRep4 plasmid encoding Sla1p, Mlp1 and multiple Mlp1 mutants. Following transformation of Schizosaccharomyces pombe suppressor strains using pRep4 plasmid (ura4+), strains were grown on selective media (Edinburgh Minimal Medium (EMM) –ura –leu) and grown in liquid EMM –ura – leu to mid-log phase (OD 0.6-0.9). Spotting was done by transferring 4 μL of liquid culture on low adenine-containing plates (EMM –ura –leu ade10), followed by a 4-day incubation at 32°C. Yeast pellets for protein purification or RNA extraction were obtained by centrifugation of mid-log phase cells at 1800 g for 10 minutes following by two washes with ddH2O.
Protein extraction was done by resuspension of cell pellets in NET-2 buffer [50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.05% NP40, 1 mM PMSF and 1X PIC], followed by lysis through bead-beating for 2 minutes total (20 sec ON – 20 sec OFF intervals) at 4°C. Cell lysates for protein analysis were obtained following centrifugation for 15 minutes at 20,000 g.
RNA extraction was completed by resuspending the cell pellets in complete RNA extraction buffer A [50 mM NaOAc, pH 5.1, 10 mM EDTA, 1% SDS], followed by adding 37°C buffer A [50 mM NaOAc, pH 5.1]-saturated acid phenol and incubation at 65°C for 4 minutes with frequent vortexing. The aqueous top layer was extracted following centrifugation for 3 minutes at 20,000 g and extracted again using phenol:chloroform:isoamyl alcohol (25:24:1). RNA was precipitated from the aqueous layer by ethanol precipitation and incubation at -80°C for at least 1 hour.
RNA-immunoprecipitation and 3’RACE in Schizosaccharomyces pombe
Sla1p- and Mlp1-transformed Schizosaccharomyces pombe sla1-strains (ySH9) were grown to mid-log phase (OD 0.6-0.9) in EMM –ura –leu. The culture was cross-linked at 200 RPM in 0.5% formaldehyde at room temperature for 20 minutes, followed by adding 200 mM glycine for 10 minutes. The yeast pellet was collected by centrifugation for 10 minutes at 4000 RPM (Beckman Coulter JLA9.1000 rotor), washed with ddH2O and collected by centrifugation for 10 minutes at 1800g, followed by one wash in resuspension buffer [1.2% polyvinylpyrrolidone (PVP), 20 mM HEPES, pH 7.4, 1 mM PMSF, 1X PIC, 1 mM DTT]. The yeast pellet was flash frozen in liquid nitrogen as a continuous bead, followed by cryogrinding in liquid nitrogen using a mortar and pestle. Yeast powder was lysed in RNP buffer [20 mM HEPES, pH 7.4, 110 mM KOAc, 100 mM NaCl, 0.5% Triton-X100, 0.1% Tween-20, 1 mM PMSF, 1X PIC and 0.05 U/µL RNase inhibitor]. The lysate was clarified by centrifugation at 20,000g for 10 minutes at 4°C. Next, 0.005 U/µL TurboDNase (ThermoFisher Scientific AM2238) was added, and the cell lysate was pre-cleared using Protein G Dynabeads coated with rabbit isotype immunoglobin G (IgG) control rotating 1 hour at 4°C. Immunoprecipitations were performed using an affinity purified rabbit anti-Mlp1 and anti-Sla1p antibody (ThermoFisher Scientific – Custom Antibodies) coated to Protein G Dynabeads rotating 1 hour at 4°C. As a control, the same antibodies were used for immunoprecipitations from ySH9 transformed with empty pRep4 plasmid. The Protein G Dynabeads were washed 5 times using RNP buffer (without PIC, PMSF and RNase inhibitor). RNA was isolated following reverse cross-linking for 45 minutes at 70°C in buffer B [50 mM Tris-HCl pH 7.4, 5 mM EDTA pH 8.0, 10 mM DTT, 1% SDS], followed by Trizol extraction.
RNA samples were polyadenylated and reverse transcribed into cDNA using qScript® microRNA cDNA Synthesis Kit (QuantaBio). Using a pre-tRNA intron-specific forward primer (see Table S5), the substrate of interest was PCR-amplified using Taq DNA polymerase in combination with the reverse PerfeCTa Universal PCR Primer (QuantaBio) which anneals with the oligo-dT adapter sequence incorporated during cDNA synthesis. The PCR products were purified on a 1% agarose gel and ligated into a pGEM-T Easy Vector Systems (Promega) plasmid followed by transformation in E. coli cells. Plasmid DNA was extracted, and sequences determined by Sanger sequencing at the Hospital for Sick Children – The Centre for Applied Genomics (TCAG).
tRNA 5’-leader and 3’-trailer computational analysis
A custom python script was used to scrape tRNA information for Homo sapiens, Mus musculus, Drosophila melanogaster, Arabidopsis thaliana, Saccharomyces cerevisiae and Schizosaccharomyces pombe from the Genomic tRNA Database (GtRNAdb)55 and Tetrahymena thermophila sequences were obtained from the UCSC Genome Browser (https://genome.ucsc.edu/) (Additional Table 3). The number of tRNA genes encoded in the genome was determined for different eukaryotes for each isotype and isodecoder (Table S4). Trailer lengths were calculated as the number of nucleotides found between the discriminator base, the last annotated mature tRNA nucleotide, and a stretch of minimum four Ts in the genomic DNA. Trailer lengths longer than 20 nucleotides were excluded from the analysis. Mature tRNA sequence lengths were determined starting at nucleotide +1 and ending at discriminator base N73, excluding introns (official tRNA numbering as described previously56). Statistical significance (P < 0.05) was calculated using a one-way ANOVA and Tukey’s multiple comparison test. The 5’-leader logo were generated for Homo sapiens, S pombe and Tetrahymena thermophila using WebLogo57 and divided based on discriminator base identity.
Immunofluorescent staining in Tetrahymena thermophila
Tetrahymena thermophila WT and partial Mlp1 KO strains were grown to mid-log phase and prepared for indirect immunofluorescent staining as described in58. Centrifugation steps including washes were performed for 3 minutes at 1000g. Briefly, fixative (two parts saturated HgCl2 and one part 95% ethanol) was added to the cell suspension and incubated for 5 minutes at room temperature, followed by collecting and resuspending the cell pellet in 100% ice-cold methanol twice. The cell pellet was washed with PBS and incubated with primary rabbit anti-Mlp1 antibody (1:500 dilution) rotating overnight at 4°C. The cell pellet was washed three times with PBS, followed by incubation with secondary goat anti-rabbit IgG (1:10,000, ThermoFisher Scientific A11008) rotating for 1 hour at room temperature. The cell pellet was washed three times with PBS. The cell suspension was dropped on a coverslip, airdried and mounted with Vectashield Antifade Mounting Medium with DAPI (Vector Laboratories) onto a microscope slide. Microscopy images were taken at 63x magnification on a LSM700 confocal laser scanning microscope (Zeiss) and processed in ZEN3.3 (blue edition).
Northern blotting
RNA was obtained from storage solution at -80°C [100% ethanol slurry containing 150 mM NaOAc, pH 5.1 and 30 µg GlycoBlue Coprecipitant (ThermoFisher Scientific AM9515)] by centrifugation at 20,000 g for 10 minutes. The RNA pellet was washed with 70% ethanol and airdried, followed by resuspension in RNase-free ddH2O. Samples were prepared by adding an equal volume of 2X formamide loading dye [80% deionized formamide, 0.06% (w/v) bromophenol blue, 0.06% (w/v) xylene cyanol, 10 mM EDTA, pH 8.0], followed by incubation at 95°C for 5 minutes and snap cooling on ice. RNA samples were separated on a 7 M urea denaturing polyacrylamide gel at 4°C at 100V and transferred onto a charged nylon transfer membrane (PerkinElmer), followed by cross-linking using a UV Stratalinker and drying at 80°C on a Gel Dryer (Bio-Rad).
The membranes were hybridized for 2 h at the corresponding probe melting temperature (Tm) -10°C in hybridization buffer [6X SSC (1X SSC: 150 mM NaCl and 15 mM sodium citrate), 1% sodium dodecyl sulfate (SDS) and 4X Denhardt’s solution (Bio Basic D0062)], followed by overnight incubation with T4 Polynucleotide Kinase (PNK) 32P-labeled probes (see Table S5). Following three 20-minute washes in wash buffer [2X SSC and 0.1% SDS] the membrane was exposed to a storage phosphor screen overnight and developed on a Typhoon. To strip hybridized probe, the membrane was incubated three times with stripping buffer [0.1X SSC and 0.1% SDS] for 20 minutes each at 70°C.
Western blotting
Protein concentration was obtained using Bradford assays (ThermoFisher Scientific, cat#23300). Protein samples were incubated at 95°C for 10 minutes, separated by electrophoresis on a 12% SDS polyacrylamide gel and transferred onto a nitrocellulose membrane. The membrane was blocked in 0.5% (w/v) skim milk powder in Tris-Buffered Saline [20 mM Tris-HCl, 150 mM NaCl] + 0.1% Tween20 (TBST) for 1 hour at room temperature (or overnight at 4°C), followed by incubation with primary antibodies in TBST for 1 hour at room temperature (or overnight at 4°C). The membrane was washed 5 times with TBST, followed by incubation with HRP-conjugated secondary antibodies at 1:10,000 dilutions and incubated for 1 hour at room temperature. Primary antibodies used in this study: mouse anti-beta actin (abcam ab8224), rabbit anti-histone H3 (abcam ab1791) and affinity purified antibodies (ThermoFisher Scientific – Custom Antibodies) rabbit anti-Mlp1 and rabbit anti-Sla1p. Secondary HRP-coupled antibodies used in this study: goat anti-rabbit IgG (Cell Signaling Technology 7074) and horse anti-mouse IgG (Cell Signaling Technology 7076).
Genomic DNA extraction and southern blotting in Tetrahymena thermophila
Genomic DNA extraction from Tetrahymena thermophila cell pellets was performed as described previously43. Briefly, cell pellets from 25 mL cultures were harvested and resuspended in 0.5 mL 10 mM Tris-HCl, pH 7.4, followed by addition of 3.5 mL urea buffer [42% (w/v) urea, 350 mM NaCl, 10 mM Tris, pH 7.4, 10 mM EDTA, 1% SDS, 0.1 mg/mL Proteinase K] and incubated for 5 minutes at 50°C. DNA was extracted twice with an equal volume of phenol:chloroform:isoamylalcohol (25:24:1), followed by chloroform:isoamylalcohol (24:1) extraction. Next, one-third volume of 5 M NaCl was added to the aqueous phase and the DNA was precipitated with an equal volume of isopropanol. The DNA was pelleted by centrifugation at 20,000g for 10 minutes at 4°C and the DNA pellet resuspended in 50 μL Tris-EDTA (TE), pH 8.0 [10 mM Tris-HCl, pH 8.0, 1 mM EDTA, pH 8.0]. The DNA suspension was treated with RNase A (10 mg/mL) overnight at 55°C and stored at -20°C.
Genomic DNA from WT and partial Mlp1 KO strains was digested with EcoRI restriction enzyme overnight at 37°C and separated by electrophoresis on a 1% agarose gel. The DNA was transferred onto a nitrocellulose membrane by capillary forces, followed by cross-linking using a UV Stratalinker and drying at 80°C on a Gel Dryer (Bio-Rad). The membrane was probed using a T4 Polynucleotide Kinase (PNK) 32P-labeled PCR product as described in “Northern blotting”.
Data reproducibility and statistics
Mlp1 RIP Northern blot and associated Western blot analysis were performed in biological triplicates. Mlp1 partial KO strain Northern blots and associated Western blots were performed in biological triplicates. TGIRT-sequencing of size-excluded Mlp1-RIP, WT and Mlp1 partial KO strains were performed in biological triplicates. EMSAs and competition EMSAs were performed in triplicates unless stated otherwise in figure legends. tRNA-mediates suppression assays in Schizosaccharomyces pombe ySH9 strains and associated Northern blots and Western blots were performed in biological triplicates unless stated otherwise in the figure legend. RNA-immunoprecipitation and associated Western blots from ySH9 and Sanger sequencing of clonal isolates derived from 3’-RACE were performed in biological duplicates. Comparison of tRNA features in different eukaryotes was done using the one-way ANOVA for 3’-trailer length (F = 289.3, DF=6, P value < 0.0001) and mature tRNA length (F = 0.9931, DF=6, P value = 0.428).
Code availability
The RNAseq analysis code is available in a Snakemake workflow that can be found at https://github.com/etiennefc/t_thermophila_RNA_Seq.git.
Data availability
TGIRT-sequencing has been deposited to the Gene Expression Omnibus (GEO) under the accession number XXXX).
Author Contributions
K.K. performed most experiments. J.G. and R.P. provided Tetrahymena WT strains, generated the Mlp1 partial KO strain and associated confirmation (southern blot and PCR). S.A.E. performed cDNA TGIRT library preparation and sequencing and E.F-C., M.S. the associated bioinformatic analysis. K.K and M.A.B. designed the study, analysed the data, and wrote the paper.
Additional Information
Competing Interests Statement
The authors declare that they have no competing interests.
Correspondence and Materials Requests should be addressed to M.A.B.
Acknowledgements
We thank R.J. Maraia and J-M. Deragon for comments on the manuscript. This work was funded by the Canadian Institutes of Health Research’s Institute of Genetics (to M.A.B).