The fission yeast methyl phosphate capping enzyme Bmc1 guides 2′-O-methylation of the U6 snRNA

Abstract Splicing requires the tight coordination of dynamic spliceosomal RNAs and proteins. U6 is the only spliceosomal RNA transcribed by RNA Polymerase III and undergoes an extensive maturation process. In humans and fission yeast, this includes addition of a 5′ γ-monomethyl phosphate cap by members of the Bin3/MePCE family as well as snoRNA guided 2′-O-methylation. Previously, we have shown that the Bin3/MePCE homolog Bmc1 is recruited to the S. pombe telomerase holoenzyme by the LARP7 family protein Pof8, where it acts in a catalytic-independent manner to protect the telomerase RNA and facilitate holoenzyme assembly. Here, we show that Bmc1 and Pof8 are required for the formation of a distinct U6 snRNP that promotes 2′-O-methylation of U6, and identify a non-canonical snoRNA that guides this methylation. We also show that the 5′ γ-monomethyl phosphate capping activity of Bmc1 is not required for its role in promoting snoRNA guided 2′-O-methylation, and that this role relies on different regions of Pof8 from those required for Pof8 function in telomerase. Our results are consistent with a novel role for Bmc1/MePCE family members in stimulating 2′-O-methylation and a more general role for Bmc1 and Pof8 in guiding noncoding RNP assembly beyond the telomerase RNP.


INTRODUCTION
Pre-mRNA splicing, comprised of intron excision and subsequent exon ligation, relies on dynamic RN A-RN A and RNA-protein interactions in the spliceosome (re vie wed in ( 1 )).The spliceosome contains upwards of 100 proteins ( 2 ) and 5 urid yla te-rich small nuclear RN As (snRN As): U1, U2, U4, U5 and U6.The U6 snRN A, w hich forms part of the catalytic core of the spliceosome ( 3 ), undergoes se v eral conformational changes during pre-spliceosome assembly and splicing catal ysis, w hich enables its interaction with other spliceosomal RNAs and the switch between a catalytically acti v e and inacti v e state ( 4 ).As such, U6 biogenesis and ma tura tion is complex and tightly regula ted to ensure correct functioning in the spliceosome (re vie wed in ( 5 )).
In addition to being the most highly conserved of the snRNAs, U6 is also the only snRNA transcribed by RNA Pol ymerase III (RN AP III) ( 6 ).In humans and Schizosacchar om y ces pombe , transcription of U6 by RNAP III is associated with the addition of a 5 ␥ -monomethyl phosphate ca p catal yzed by enzymes of the Bin3 / MePCE (methyl phosphate capping enzyme) family (7)(8)(9).U6 contains a 5 stem loop critical for 5 capping ( 10 ), as well as an internal stem loop (ISL) that forms during splicing catalysis.
The ISL is m utuall y e xclusi v e with U4 / U6 base pairing that occurs in pre-spliceosome snRNPs ( 11 ).U6 also contains 2 -O-methyla ted, pseudourid yla ted, and m6A-modified nucleotides, with pseudouridines largely present towards the 5 end and 2 -O-methylations tending to cluster in the ISL ( 12 , 13 ).Moreover, U6 ma tura tion in fission yeast involves the splicing of an mRNA-type intron, thought to arise from re v erse splicing, as the intron is located near the catalytic nucleotides of U6 (14)(15)(16)(17)(18).Most information about the timing of U6 processing e v ents has come from elegant studies in budding yeast (re vie wed in ( 5 )).Howe v er, since budding yeast U6 lacks 2 -O-methylations and a Bmc1 homolog ( 19 , 20 ), se v eral questions remain as to the timing and importance of post-transcriptional modifications with respect to other U6 processing steps in organisms like fission yeast and humans.
In addition to 5 ␥ -monomethyl phosphate capping enzymes, se v eral other proteins have been linked to U6 processing.These include the La protein, which associates with nascent U6 transcripts through the 3 urid yla te tail ( 21 ), and the Lsm2-8 complex, which binds end-matured U6 and remains stably associated through spliceosome assembly (22)(23)(24).Recent work revealed that mammalian LARP7, a La-Related Protein (LARP) previously linked to MePCE in the context of the 7SK snRNP ( 25 ), is also involved in post-transcriptional processing of U6.LARP7 promotes 2 -O-methylation of U6 by the methyltr ansfer ase fibrillarin, which in turn contributes to splicing fidelity at elevated temperatures in humans and in male germ cells in mice ( 26 , 27 ).Conversely, ciliate and fission yeast LARP7 homologs have been well studied for their roles in telomerase biogenesis (28)(29)(30)(31)(32)(33).We and others have reported that the S. pombe LARP7 protein Pof8 associates with the Bin3 / MePCE homolog Bmc1 and a fission yeast-specific protein, the telomerase holoenzyme component 1 (Thc1) ( 20 , 34 ), which shares homology with the nuclear cap binding complex and poly-adenosine ribonuclease, both of which are involved in human telomerase RN A bio genesis ( 35 , 36 ).We have also shown that the Bmc1 / Pof8 interaction is important for optimal telomerase activity, and that the link between these proteins is evolutionarily conserved across diverse fungal species ( 19 , 20 , 34 ).Thus, while much has been learned about MePCE / Bmc1 function in the 7SK snRNP and telomerase, its precise role in U6 biogenesis and function remains unknown.
In this work, we set out to examine the role of Bmc1 in U6 biogenesis and spliceosome function.We describe a new RNP containing the U6 snRNA and the telomerase components Bmc1, Pof8 and Thc1, and show that this complex is r equir ed for wild type levels of 2 -O-methylation in the U6 ISL and U6 snRNP assembly.We also show that while Bmc1's 5 capping catalytic activity is not r equir ed for its function in promoting 2 -O-methylation of U6, an intact Pof8-Lsm2-8 interaction is.Finally, we show that Bmc1 deletion influences the splicing of some introns under wildtype and heat-shock conditions, consistent with previous work linking human LARP7 to splicing robustness.Using our transcriptome-wide pre-mRNA splicing data set, we also show that increased intron retention upon heat shock in S. pombe is linked to particular intron features including strength of 5 splice site sequences.Together, these data point towards an intricate network of post-transcriptional processing e v ents that are critical for normal U6 biogenesis, and provide the first direct evidence for a function of the Bin3 / MePCE family in promoting U6 snRNP ma tura tion.

Yeast strains and growth
Strains were grown at 32˚C in yeast extract with supplements (YES) or Edinburgh Minimal Media (EMM), as indica ted.Tag integra tion and knockouts were generated as described in ( 20 ) (primer sequences provided in Supplementary Table S1).A list of yeast strains is provided in Supplementary Table S2.

RNA pr epar ation, northern blotting, 2 -O-methylation detection, and solution hybridization
Total RNA was extracted with hot phenol, separated on 10% TBE-urea polyacrylamide gels, and transferred to positi v ely charged nylon membranes (Per kin Elmer, NEF988001) as per ( 38 ).For nati v e RNA e xtraction to detect U4 / U6 duplexes, RNA was extracted with cold phenol, as per ( 39 ).Solution hybridization was performed as per ( 40 ) and resolved on 9% TBE gels.Probe sequences f or 32 P ␥ -ATP-labeled DNA probes f or northern blotting are provided in Supplementary Table S3.Primer extensions to detect 2 -O-methylation were performed based on protocols from ( 41 ).Briefly, 5 g RNA was incubated for 5 min at 85˚C in a 10 l reaction containing 32 P ␥ -ATPlabeled probe, 50 mM Tris-HCl pH 7.4, 60 mM NaCl, then transferred to 55˚C for 20 min to allow the probe to anneal.Re v erse transcription was carried out with 1.5 mM (high concentration) or 0.1 mM (limiting concentration) dNTP mix and 2.5 U AMV-RT (NEB, M0277S) and 1 h incuba tion a t 42˚C .cDNA products were separa ted on 8% TBE-urea sequencing gels, dried, and exposed to Phosphor screens ov ernight.Relati v e 2 -O-methyla tion was calcula ted by determining the ratio of each RT stop relati v e to the total signal in each lane (all RT stops and full length U6). 2 -O-methylations were also detected by RNase H (NEB, M0297S) digestion of 2 g with 25 pmol chimeric RNA-DNA probes, as per ( 42 ).Probe sequences targeting C57 and A64 2 -O-methylations are provided in Supplementary Table S3.

qRT-PCR and semi-quantitative RT-PCR
1 g TURBO DNase-treated RNA was re v erse transcribed with the iScript cDNA re v erse transcription kit (Biorad, 1708890) or 5 U AMV-RT (NEB, M0277S) and genespecific re v erse primers.qRT-PCR was performed with the SensiFAST SYBR No-Rox kit (Bioline, BIO-98005) and 1 M of each primer, with settings outlined in ( 20 ).For semiquantitati v e RT-PCR, cDNA was amplified with Taq polymerase (NEB, MO273L) using the following cycling conditions: 5 min initial dena tura tion a t 94˚C , 26 (pud1, alp41) or 27 (rpl1603, bor1) cycles of 30 s a t 94˚C , 30 s a t 50˚C , and 1 min at 72˚C, and a final 5 min extension at 72˚C.cDNA was resolved on 10% TBE gels.

Nativ e y east extr act pr epar ation, native snRNP gels and glycerol gradient sedimentation
Pellets from 1 L yeast cultur es wer e r esuspended to 1 g / ml in AGK400 buffer (10 mM HEPES-KOH pH 7.9, 400 mM KCl, 1.5 mM MgCl 2 , 0.5 mM DTT, 1 mM PMSF, and protease inhibitor cocktail (Sigma, P8215)), frozen in liquid nitr ogen, and gr ound to fine po w der with a mortar and pestle.Po w der was thawed on ice and spun in a JA 25.50 rotor (Beckman) for 16 min at 15000 rpm and the supernatant was subsequently spun in a 70.1 Ti rotor (Beckman) for 45 min at 50 000 rpm to pellet ribosomes and heavy molecular weight complexes.Supernatants were flash frozen and stored a t −80˚C .For na ti v e snRNP gels, gly cerol with xylene cyanol and bromophenol blue was added to 30 g cell extract (final glycerol concentration = 10%) and fractionated on 4% 19:1 acrylamide: bis-acrylamide nati v e gels (15 cm × 18 cm) for 220 min at 240 V and 4 • C, then transferred to nylon membranes for northern b lotting.For gly cerol gradients, cell extracts from 1.0 g frozen cell po w der were layered on an 11 ml 10-30% glycerol gradient (50 mM Tris-HCl pH 7.4, 25 mM NaCl, 5 mM MgCl 2 ) and spun in an SW41Ti rotor (Beckman) for 20 hours at 30 900 rpm.Fractions were collected starting from the top of the gradient and RNA and proteins were extracted with phenol: chloroform: isoamyl alcohol (25:24:1) and T CA pr ecipitation, respecti v ely.

UV melt curves
UV melt curves were recorded on a Cary BIO 100 spectrometer with a 6 × 6 temperature-controlled cell holder.2 l 10 mM U4 and modified or unmodified U6 RNA oligos in 96 l buffer (10 mM KH 2 PO 4 pH 7.0 and 200 mM KCl) was heated and cooled from 50˚C to 65˚C at a rate of 2˚C per minute without collecting data, then re-heated and cooled while monitoring absorbance at 260 nm at 1˚C intervals.Absorbance at 260 nm at each temperature point was normalized to absorbance at 50˚C and absorbance curves were fitted with an equation for one site specific binding with a Hill slope to determine T m values.RNA sequences are provided in Supplementary Table S3.

RNA seq and intron retention analysis
DNase-treated RNA was rRNA-depleted (Qiagen, 334215) and stranded libraries wer e pr epar ed by Genome Qu ébec.cDNA libraries were sequenced on a NovaSeq6000 with 150 bp paired-end reads.Reads were aligned to the fission yeast genome (ASM294v2) with Bowtie2 ( 43 ).Intron retention was quantified using IRFinder (version 2.0.1), as per ( 44 , 45 ).Any introns flagged as having a low sequencing depth or fewer than 4 reads to support splicing were not considered for sta tistical analysis.Dif fer ential intron r etention was calculated using DESeq2 ( 46 ).Sequence extraction for S. pombe introns was carried out using BED-Tools v2.3.0 ( 47 ) and sequences are provided in dataset 1. 5 splice sites (3 bases in the exon and 6 bases in the intron) and 3 splice sites (20 bases in the intron and 3 bases in the exon) wer e scor ed with MaxEntScan using a maximum entropy model ( 48 ).Intron free energy of the thermodynamic ensemble (kcal / mol) was calculated using RNAfold v2.5.1 ( 49 ).

Bmc1 forms a U6-containing complex with the telomerase proteins Pof8 and Thc1
Our previous work characterizing Bmc1 as a component of the telomerase holoenzyme also re v ealed interactions between Bmc1 and various other noncoding RNAs, including the U6 snRNA (Supplementary Figure S1A) ( 20 ).We ther efor e tested whether Bmc1 has a role in the biogenesis, stability, or function of these transcripts, and if this function is linked to the Bmc1-interacting telomerase components Pof8 and Thc1.Having already demonstrated that Pof8 is r equir ed to r ecruit Bmc1 to the telomerase RNA TER1 ( 20 ), we determined the protein binding r equir ements for U6.In contrast to what has been reported for TER1, for which (reduced) binding to Pof8 persists in the absence of Bmc1 ( 20 , 34 ), we found that all three proteins are necessary for an interaction with U6 (Figure 1 A, Supplementary Figure S1B).Our results indicating an interaction between Pof8 and U6 are also consistent with previous work identifying mammalian LARP7 as a U6-interacting protein ( 26 , 27 ), suggesting that LARP7 family members have conserved functions related to U6, in addition to LARP7 function in telomerase in S. pombe and ciliates (29)(30)(31).
As an additional means to confirm U6 snRNP formation, we fractionated nati v e cell e xtracts on gly cerol gradients and compared protein and RNA sedimentation in wild type and knockout yeast strains (Figure 1 B, Supplementary Figure S1C).A substantial fraction of Bmc1 ).We propose that Bmc1, Pof8, and Thc1 associate with U6 sim ultaneousl y, with all three proteins required to be present to initiate formation of the Bmc1containing U6 snRNP.We also fractionated cell extracts on higher percentage glycerol gradients to compare sedimenta-tion of the U6-and Bmc1-containing snRNP to the telomerase holoenzyme (Supplementary Figure S1D).The telomerase RNA TER1 migrated in heavier sedimenting fractions (fractions 6-9) compared to the U6 peak that was sensiti v e to Pof8 deletion (fractions 2-4), suggesting that the U6 and telomerase peaks are not overlapping, and thus distinct complexes.
Together, these data point towards the existence of a new U6-containing complex that also shares components with the telomerase holoenzyme, providing a surprising link between two seemingly disparate fission yeast noncoding RNA pathways.

Bmc1, Pof8, and Thc1 promote 2 -O-methylation of U6
To gain further insight into the role of the Bmc1-containing U6 snRNP, we examined our Bmc1 RIP-Seq dataset ( 20 ), which re v ealed an interaction between Bmc1 and snoZ30, which guides 2 -O-methylation of U6 at position 41 ( 41 ) (Supplementary Figures S1A, S2A, B).Further supporting the idea that U6 complex formation is contingent on the presence of all three proteins, we observed a loss of snoZ30 binding to Bmc1 upon knockout of any member of the complex (Supplementary Figure S2A).The observed interaction between Bmc1 and snoZ30, coupled with the recently described function of mammalian LARP7 in facilitating snoRNA-guided 2 -O-methylation of U6 by the methyltr ansfer ase fibrillarin ( 26 , 27 ) provided initial clues as to the function of this new Bmc1-and U6-containing snRNP.To determine if Bmc1, Pof8, and Thc1 influence 2 -O-methylation, we mapped U6 2 -O-methylation sites by performing primer extensions at low dNTP concentrations ( 41 ).Although snoZ30 is the sole annotated U6-modifying snoRNA in fission yeast ( 41 ), se v eral other 2 -O-methylated sites have been identified in U6, including A64 ( 13 ).Deletion of Bmc1, Pof8, and Thc1 resulted in no observable changes in 2 -O-methylation at the snoZ30-modified A41, but we did detect a reproducible decrease in modification at se v eral other sites, most notably A64 (Figure 1 C, D, Supplementary Figure S3).Initial attempts at identification of the U6 A64methylating snoRNA using box C / D snoRNA consensus sequences and base pairing rules ( 50 ) yielded no other obvious snoRNA candidates, so we instead turned to our Bmc1 RIP-Seq dataset in the hope we might identify novel snoRNAs (Supplementary Figure S1A).The uncharacterized fission yeast noncoding RNA, SPNCRNA.530(henceforth r eferr ed to as sno530), contains a D box, a putati v e C box one nucleotide different from the C box consensus motif, and a region with 12 nucleotides of complementarity with U6, including a single non-Watson Crick base pair (Supplementary Figure S2C).It is also noteworthy that the predicted secondary structure of sno530 does not position the C and D boxes flanking a hairpin, as is common for canonical box C / D snoRNAs (Supplementary Figure S2C).We validated the interaction between Bmc1 and sno530 by RNP immunoprecipitation / qPCR and showed that much like snoZ30 and U6, this interaction is dependent on the presence of the assembled Bmc1-Pof8-Thc1 complex (Figure 1 A).Deletion of snoZ30 and sno530 resulted in a loss of 2 -O-methylation at A41 and A64, respecti v ely, suggesting that sno530 is indeed the A64 U6-modifying snoRNA (Figure 1 C, D, Supplementary Figure S3).We obtained similar results using a complementary method that exploits the tendency for 2 -O-methylations to block RNase H cleavage following the annealing of a chimeric DN A-2 -O-methylated RN A oligo targeting the suspected 2 -O-methylated site ( 42 , 51 ) (Supplementary Figure S4A).This also served to provide evidence for 2 -O-methylation a t C57, suggesting tha t it, too, is another site in U6 whose modification is similarly guided by Bmc1 and Pof8 (Sup-plementary Figure S4B).While our knockout studies unambiguously identify sno530 as the A64 U6-modifying snoRNA, the unusual sequence and ar chitectur e of sno530 relati v e to snoZ30 is more reminiscent of the di v ergent box C' / D' motifs tha t stimula te rRNA 2 -O-methyla tion by providing additional regions of complementarity surrounding the methylated site ( 52 , 53 ).We were unable to identify any candidate snoRNA(s) responsible for methylating C55, C56, and C57, by searching the fission yeast genome or through our Bmc1 RIP-Seq dataset, although this may be because the snoRNA does not conform to canonical snoRNA motifs or RNA-snoRNA base pairing rules.

Bmc1, Pof8 and Thc1 are involved in U6 snRNP assembly
We then tested how disruption of the Bmc1-containing U6 snRNP might impact spliceosome assembly by fractionating cell extracts in a glycerol gradient and assaying protein and snRNA distribution.In wild-type cells, we noted a lowly abundant, lighter sedimenting complement of U6 that does not appear to co-sediment with U4 (compare U6 and U4 in lanes 3-6 in Figure 1 B and Supplementary Figure S1C), suggesting a U6-containing complex outside of the more abundant U4 / U6 di-snRNP.Consistent with this, we note that the migration of Pof8 and Bmc1 does not fully overlap with U4 / U6 in the gradient, but is rather shifted towards lighter fractions, arguing against the inclusion of the Bmc1-Pof8-Thc1 complex in the U4 / U6 di-snRNP (Figure 1 B, Supplementary Figure S1C and below).As the lighter sedimenting U6 species is not evident in Bmc1 and Pof8 KO strains (compare U6 and U4 in lanes 3-6 in Figure 1 B and Supplementary Figur e S1C r elati v e to Bmc1 and Pof8 KO strains), we hypothesized that Bmc1, Pof8 and Thc1 interact with U6 before the U4 / U6 di-snRNP.
To obtain clearer resolution of distinct U6-containing complexes, we ran cell extracts on native gels and anal yzed spliceosomal RN As by northern blotting.We observed a single, prominent band for all spliceosomal RNAs except U6, which migrated as 2 distinct complexes (Figure 2 A).We could assign the higher molecular weight complex, which comigrates with U4 but not U2 or U5, as the U4 / U6 di-snRNP.Upon deletion of any of Bmc1, Pof8 or Thc1, we observed a significant and reproducible decrease in the intensity of the lower molecular weight U6containing snRNP (Figure 2 A, B), consistent with this band r epr esenting the Bmc1-containing U6 snRNP.The persistence of this complex upon loss of sno530 suggests that complex formation is not reliant on the ability to modify U6 at A64.Although U6 and sno530 are associated with Bmc1 (Figure 1 A), sno530 is ther efor e not r equir ed for the stability of the U6 snRNP observed in native gels.
To understand when Bmc1 interacts with U6 with respect to spliceosome formation, we immunoprecipitated Bmc1 associated RNPs under nati v e conditions and ran total and Bmc1-associated RNA on nati v e gels.Bmc1 immunoprecipitated only the lower molecular weight U6 snRNP and not the species tha t co-migra tes with the U4 / U6 di-snRNP (Figure 2  U6 complex, which promotes 5 capping and 2 -Omethylation, is distinct from the U4 / U6 di-snRNP. Nati v e fission yeast cell extracts do not form detectable amounts of the U4 / U6.U5 tri-snRNP ( 54 , 55 ), so we focused our further efforts on examining U4 / U6 base pairing by performing a solution hybridization assay on cold phenol-extracted total RNA to maintain U4 / U6 base pairing (Figure 2 D).This differs from nati v e spliceosomal snRNP gels (Figure 2 A) in that it only assesses RNA-RNA interactions, without changes in mobility due to protein binding.We detected minor defects in U4 / U6 assembly upon Bmc1, Pof8, or Thc1 deletion, as measured by the increase in 'free' U4 relati v e to U4 complexed in the di-snRNP, although the increase in the fraction of free U4 only reached statistical significance upon Pof8 deletion (Figure 2 D, E).Consistent with the increase in free U4 in the knockout strains, glycerol gradients re v ealed an increase in lighter sedimenting U4 in the knockout strains (Figure 1 B, Supplementary Figure S1B, compare lanes 1-3 in wild type versus knockouts).Although U6 is in e xcess ov er U4, the incr ease of fr ee U4 in the knockouts suggests that the absence of Bmc1, Pof8 and Thc1 may result in a non-functional, alterna te pa thwa y f or U4 that does not in volve U4 / U6 di-snRNP formation.The lack of U4 / U6 pairing defects upon the loss of sno530 further suggests that it is largely the Bmc1-Pof8-Thc1 protein complex dictating U4 / U6 pairing, not the single A64 2 -O-methylation.Still, UV melt analysis of the U6-interacting region of U4 and the U6 internal stem loop (ISL), with or without 2 -O-methylation of A64, re v ealed a slight increase in U4-U6 duplex stability with 2 -O-methylation, consistent with previous findings reporting on the stabilizing properties of 2 -O-methylation on RNA duplex formation ( 56-58 ) (Figure 2 F).
Subsequent steps in spliceosome f ormation in volve unwinding of the U6 ISL and base pairing between U4 and U6, both of which are promoted by the U4 / U6 di-snRNP assembly factor Prp24 ( 59 , 60 ).We generated an endogenously tagged Prp24 strain and assessed the interaction between Prp24 and U4 and U6 in wild-type and Bmc1 or P of8 deleted cells.P of8 and Bmc1 deletion resulted in a decreased interaction between Prp24 and U4 and U6, suggesting that Bmc1 and Pof8 promote the association of U4 and U6 with Prp24, which in turn may promote the formation of the U4 / U6 di-snRNP (Figure 2 G, H).In sum, our results are consistent with the existence of a Bmc1 / Pof8 / Thc1containing U6 snRNP, with Bmc1 / Pof8 / Thc1 dissociating from U6 during establishment of the U4 / U6 di-snRNP.

Bmc1 5 capping catalytic activity is not r equir ed f or promoting 2 -O-methylation of U6
With previous studies indica ting tha t Bmc1 5 ␥ -phosphate methyltr ansfer ase catalytic activity is dispensable for telomerase activity ( 34 ), we assayed a combination of previously described and newly constructed putati v e Bmc1 catalytic mutants for the ability to promote U6 2 -O-methylation.We mutated residues that are both highly conserved between Bmc1 and human MePCE, and well-positioned in structure predictions to interact with the methyltransferase byproduct SAH (S-Adenosyl-L-Homocysteine) (Figure 3 A).HAtagged Bmc1 mutants were transformed into a Bmc1 knockout yeast strain and profiled for U6 2 -O-methylation as above (Figure 3 B, C).While the Bmc1 mutants were more lowly expressed than wild type Bmc1, some mutants still promoted 2 -O-methylation to a greater extent than Bmc1 knockout, empty vector transformed cells (Figure 3 C).Further, normaliza tion of rela ti v e 2 -O-methylation le v els to Bmc1 expression confirmed a statistically significant increase in 2 -O-methylation for all Bmc1 mutants compared to the empty vector (Figure 3 D).This suggests that, as in telomerase, Bmc1 5 ␥ -phosphate methyltransferase catalytic activity may not be critical for its function in promoting snoRNA-directed U6 2 -O-methylation.
For further characterization of Bmc1 catalytic mutants, we chose the Bmc1 L153A V155A m utant, w hich showed the highest expression across biological replicates.As measur ed by co-immunopr ecipitation, L153A V155A still interacted with Pof8, suggesting tha t ca talytic activity is also not r equir ed f or complex f ormation (Figure 3 E).Further, L153A V155A interacted with U6, indicating that 5 ␥phosphate methyltr ansfer ase catal ytic activity is likel y not r equir ed for U6 binding (Figure 3 E).

The xRRM and Pof8-Lsm2-8 interaction are important determinants for U6 2 -O-methylation
As an established member of the LARP7 family of proteins, the protein-interacting and RNA binding domains of Pof8 have been well-characterized in the context of the telomerase RNP (29)(30)(31)(32)(33). Pof8 contains a di v ergent La motif that lacks the conserved urid yla te-binding residues typically seen in LARP7 proteins ( 31 ), so its interaction with the telomerase RNA TER1 is mediated by the RRM1, xRRM and the N-terminal region that makes direct pr otein-pr otein contacts to Lsm2-8, which in turn binds the urid yla te-rich 3 end of TER1 (29)(30)(31)(32).As muta tions to these regions have been shown to impair Pof8 binding to TER1 and telomere length homeostasis, we looked at the impact of these same mutations on U6 2 -O-methylation (Figure 4 A).In contrast to what has been observed for TER1, where both RRMs are important for binding, only mutations to the xRRM and the Lsm2-8 binding region caused a significant reduction in 2 -O-methylation at A64 (Figure 4 B, C).
To further understand the molecular basis for the drop in 2 -O-methyla tion, we immunoprecipita ted Bmc1 in a Pof8 knockout strain r e-expr essing the Pof8 mutants (Figur e 4 D, E).Bmc1 co-immunoprecipitated all Pof8 mutants, suggesting that the 2 -O-methylation defect is not due to complete disruption of the Bmc1-Pof8 interaction (Figure 4 D).The Bmc1-U6 interaction, which is dependent on the presence of Pof8 (Figure 1 A), was almost completely lost in the Lsm2-8-binding mutant ( 2-10), suggesting that its 2 -O-methylation defect may be due to a loss in U6 association with the Bmc1-Pof8-Thc1 complex (Figure 4 D, E). Surprisingly, we detected no loss in U6 binding with the xRRM mutant, indica ting tha t while U6 still interacts with the Bmc1-Pof8-Thc1 snRNP in the context of the xRRM mutant, the xRRM may have another function in facilita ting U6 2 -O-methyla tion (Figure 4 D, E).To distinguish between the possibility that the xRRM mutant is defecti v e in snoRNA binding or another activity, we repeated Bmc1 immunoprecipitations in Pof8 knockout strains reexpressing the Pof8 mutants and analyzed sno530 by qRT-PCR (Figure 4 F).Unlike human LARP7, which uses its xRRM to bind U6-modifying snoRNA ( 27 ), we only detected defects in sno530 binding with the Lsm2-8-binding m utant, not the xRRM m utant, suggesting that Pof8 may interact with snoRNAs through Lsm2-8.As the xRRM in the ciliate LARP7 protein p65 has been suggested to possess RN A cha perone activity to remodel the ciliate telomerase RNA ( 61 , 62 ), it is tempting to speculate that the decrease in U6 2 -O-methylation in the context of the xRRM mutant may result from defects in xRRM-mediated RN A cha perone activity, which could play a role in correctly positioning U6 and the snoRNA for 2 -O-methylation.

Bmc1 deletion has a minor effect on pre-mRNA splicing
Having observed Bmc1-dependent defects in U6 2 -Omethylation and U6 snRNP assembly, we tested the effects of Bmc1 deletion on pre-mRNA splicing.To that end, we performed short-r ead, pair ed-end sequencing on RNA extracted from wild type and Bmc1 knockout yeast strains and quantified intron retention as a proxy for splicing ( 44 , 45 ).We also measured intron retention in wild type and knockout cells heat shocked for 15 min at 42˚C, which has been shown to impact splicing in fission yeast ( 63 ).We observed significant increases in intron retention following heat shock, similar to what has been reported in mammalian cells ( 64 ) (Supplementary Figure S5A, B).Although we observed slight increases in intron retention in Bmc1 knockout cells compared to wild type cells, very few of these splicing e v ents at 32˚C passed our significance cutoff, and no splicing e v ents at 42˚C were statistically significant (Supplementary Figure S5C, D), although this may be due to greater sample to sample variability across our tripli-ca te replica tes for this data set (Supplementary Figure S6).Still, as mean intron retention values indeed showed an increase upon Bmc1 deletion (Figure 5 A), we chose se v eral r epr esentati v e intron retention e v ents to validate with semiquantitati v e RT-PCR (one of which, intron 1 of pud1 , displayed a statistically significant increase upon Bmc1 deletion at 32˚C in our RNA Seq dataset).We observed an increase in intron retention following heat shock, and confirmed their further impaired splicing in the context of the Bmc1 deletion (Figure 5 B, Supplementary Figure S7).Conversely, a ribosomal protein gene, which have been reported to be efficiently spliced relati v e to non-ribosomal protein genes in budding yeast ( 65 , 66 ), and did not show heat shock or Bmc1 associated changes in our RNA-Seq data set, was confirmed to have no changes in intron retention in response to heat shock or Bmc1 deletion (Figure 5 B, Supplementary Figure S7).We note that these validated Bmc-1 affected introns have higher than average intron retention rates in normal cells and as such, may not be r epr esentati v e of the average splicing event.Together, these data indicate that Bmc1 does not have a major effect on pre-mRNA splicing, but that Bmc1 likely contributes to splicing robustness, similar to what has been described for mammalian LARP7 deletion in human cells ( 26 , 27 ).We also examined other factors that could contribute to heat shock-sensiti v e splicing defects.We compared intronic features between heat shock-sensiti v e introns, classified as introns exhibiting a > 2-fold increase in intron retention upon heat shock and a false discovery rate less than 0.05, and remaining introns (heat shock insensiti v e) (Figure 5 C-E, Supplementary Figure S5A, B).We noted that heat shock-sensiti v e introns are enriched for weaker 5 splice site scores in the context of both the wild type and bmc1 Δ strains, with no significant differences in 3 splice site scores (Figure 5 C,D), in line with previous data indicating that intron retention in fission yeast is linked to weak 5 splice sites ( 67 ),.Additionally, heat shock-sensiti v e introns displayed lower minimum free energy, indicati v e of a link between intron structure and splicing changes in response to heat shock (Figure 5 E).

Conserved functions for LARP7 family proteins in splicing and U6 2 -O-methylation
This work r epr esents the first r eport of an MePCE homolog with a role in splicing and U6 snRNP assembly, beyond 5 methyl phosphate cap addition of U6.In our efforts to in vestigate functions f or Bmc1 bey ond telomerase, we revealed an unanticipated overlap between components of the y east telomer ase holoenzyme and a U6-containing snRNP.While it is surprising that Bmc1, Pof8, Thc1, and Lsm2-8 interact with 2 very distinct non-coding RNAs produced by different polymerases, both RNAs possess urid yla terich sequences recognized by Lsm2-8 and highly structured regions, including stem loops in U6 and pseudoknots in telomerase, that act as scaffolds to recruit other RNP components.These common features may provide an explanation as to why these di v ergent RNAs share a common set of protein binding partners.Such RNP plasticity is not unique to fission y east telomer ase and U6, but may r epr esent a shar ed featur e of LARP7 and MePCE family proteins.Mammalian LARP7 and MePCE are particularly well-studied for their roles in capping and stabilizing the 7SK snRNP ( 8 , 25 , 68 , 69 ), transcriptional control through DDX21 ( 51 ), directing U6 modification ( 26 , 27 ), and snRNP assembly through the SMN complex ( 70 ).Thus, continuing to study the RNA interactome of MePCE and LARP7 homologs across species will likely yield additional insight into how these proteins associate with and influence various classes of non-coding RNAs.It is also possible that Bmc1, Pof8, and Thc1 interactions with U6 ar e mediated entir ely by dir ect interactions with Lsm2-8, which in turn directly contacts U6, much like Prp24 interacts with U6 by directly binding Lsm2-8 ( 71 ).Future structural and biochemical studies will lend insight into the pr otein-pr otein and pr otein-RNA interactions that cooperate to form the U6 snRNP.
Se v eral fungal species, including S. cerevisiae , lack both a LARP7 and MePCE homolog ( 19 , 20 ).As deletion or depletion of LARP7 / Pof8 or MePCE / Bmc1 does not influence U6 stability in species where this has been investigated ( 8 , 20 , 26 , 27 , 34 ), the function of LARP7 and MePCE family members in U6 biogenesis and function has remained unclear.This work expands our understanding of the evolutionary conservation of LARP7 family members, with shared or unique functions relating to the telomerase, U6, and 7SK RNAs, depending on the species under investigation (Figure 6 A).Further links can be drawn between the functional consequences of LARP7-and Pof8-mediated promotion of U6 2 -O-methylation.LARP7 or Pof8 deletion and the subsequent decrease in 2 -O-methylation of U6 results in no functional consequences under standard physiological conditions, but becomes important for maintaining splicing fidelity under heat stress, in the case of human LARP7, and male germ cells in mice ( 26 , 27 ).As the loss of A64 modification alone results in no changes to U6 snRNP assembly, compared to the changes observed upon Bmc1, Pof8, and Thc1 deletion, we anticipa te tha t it is either a combination of the loss of se v eral 2 -O-methylations or the loss of the Bmc1-Pof8-Thc1 U6 snRNP that leads to the slight increase in intron retention at elevated temperatures upon Bmc1 deletion (Figure 5 ).Future studies aimed at teasing apart this mechanism in mammalian and yeast cells will provide additional insight into the intertwining role of RNA modifications and RNP biogenesis complexes in spliceosome assembly.
We note that the observed heat shock-dependent splicing defect in Bmc1 knockout cells is relati v ely minor.Pre vious studies on mammalian LARP7 proposed that LARP7guided 2 -O-methylation of U6 is not an important factor for splicing as a whole, but rather contributes to splicing robustness ( 26 , 27 ).Although r ecent r eports indica te tha t alternati v e splicing in fission yeast may be more widespread than previously thought ( 63 , 72 , 73 ), splicing complexity in fission yeast is still less than that observed in mammalian cells, which may explain why we do not observe any drastic splicing changes upon Bmc1 deletion.It remains to be determined whether Bmc1 affects other aspects of splicing that have not been tested here, such as splicing efficiency and fidelity, or whether Bmc1-associated defects in splicing might be greater under differ ent str esses.In addition, Bmc1 promotes 2 -O-methylation in the internal stem loop of U6, which does not base pair with the 5 or 3 splice site.Thus, modula tion of ISL modifica tions might not be expected to manifest as a robust splicing defect.This is in contrast to what has been reported for the loss of m 6 A in S. pombe U6, where affected introns are enriched for an adenosine at the fourth position of the intron, which directly base pairs with the m 6 A ( 74 ).While it is surprising that Bmc1, Pof8, and Thc1 deletion have no effect on A41 methylation levels, despite the recovery of snoZ30 in our Bmc1 RIP-Seq, A41 is the only 2 -O-methylation occurring outside the U6 ISL.This suggests that the Bmc1-containing U6 snRNP may only have an important role in guiding ISL modifications, much like LARP7 facilitates the analogous ISL modifications in mammals ( 26 , 27 ).

Emerging importance of the xRRM in RNA folding and function
Fission yeast, possessing a LARP7 homolog that functions in telomerase like its ciliate counterpart ( 20 , 29-34 ), and U6 2 -O-methylation in an analogous manner to its mammalian homologs, may r epr esent an evolutionary intermediate bridging RNA binding proteins between ciliates and mammals.The 7SK snRN A, w hich has onl y been found in animals ( 75 ) likely arose independently from the more widely distributed LARP7 and MePCE, suggesting the need for continued studies into 7SK-independent functions for LARP7 and MePCE.Of note, the conservation of the xRRM between fungal, mammalian, and ciliate LARP7 proteins, rather than the La motif ( 19 , 20 , 29-31 ) may provide a reason explaining the diverse RNA substrates bound by LARP7 homologs, compared to the more well-conserved classes of RNA binding partners of other LARPs across species ( 19 ).xRRM-mediated binding to structured stem loops like the telomerase RNA pseudoknot ( 32 ), SL4 of 7SK ( 76 ), and U6-modifying snoRNAs ( 27 ) may be a better determinant than 3 terminal urid yla te stretches for predicting LARP7 binding.The importance of the xRRM in the biogenesis and stability of telomerase RNA, 7SK, and U6 may be linked to its RN A cha perone activity, w hich has been proposed to have a role in promoting RNA folding ( 61 , 62 ).Our finding that mutation of the xRRM of Pof8 impairs 2 -O-methylation of U6 without disrupting U6 binding (Figure 5 D) may provide further evidence that the xRRM has functions beyond U6 binding and raises additional questions as to the mechanism by which RNA chap-erones can coordinate snoRNA and target RNA binding to carry out ef ficient 2 -O-methyla tion.Importantly, xRRM chaperone activity is not limited to LARP7 family proteins, as the RRM2 / xRRM of the human La protein has also been shown to promote RNA folding (77)(78)(79).

New insights into U6 biogenesis in fission yeast
This work also sheds light on the timing of U6 biogenesis steps in fission yeast (Figure 6 B).We have previously shown that Lsm2-8 interacts with both mature and introncontaining U6, suggesting that intron removal occurs after 3 end processing and the switch from La to Lsm2-8 ( 20 , 23 ).Conversely, Bmc1 and Pof8 interact solely with the spliced form of U6 ( 20 ).This, coupled with our finding that the Lsm2-8-interacting region of Pof8 is r equir ed for the Bmc1-U6 interaction (Figure 5 D, E), indicates that Lsm2-8 binding occurs prior to splicing and recruitment of the Bmc1-Pof8-Thc1 complex.Our da ta indica ting tha t Bmc1 co-purifies with U6-modifying snoRNAs (Figure 1 and Supplementary Figure S2) suggests that U6 then undergoes 5 capping by Bmc1 and 2 -O-methylation, prior to Bmc1-Pof8-Thc1 dissociation from U6 and U4 / U6 di-snRNP assembly mediated by Prp24.Since deletion of Bmc1 or Pof8 results in decreased association of Prp24 with U4 and U6 (Figure 2 ), the Bmc1-Pof8-Thc1 complex may play a role in the handoff to Prp24.This role may be mediated by xRRM-linked chaperone activity that remodels U6 to better position it to interact with Prp24 and U4.Our finding of a new U6 biogenesis complex thus adds another layer of regulation to spliceosome assembly.Still, it remains unknown whether Bmc1, Pof8, and Thc1 only interact with U6 during its biogenesis, or re-associate with U6 when it is reassembled into the U4 / U6 di-snRNP for subsequent rounds of splicing catalysis, although the significant decrease in abundance of the U4-lacking U6 snRNP (Figure 2 A) as well as the impaired binding of U4 and U6 to Prp24 upon deletion of any of Bmc1, Pof8 or Thc1 (Figure 2 G) may be more consistent with re-engagement of the Bmc1 / Pof8 / Thc1 complex after dissociation of U6 from the newly spliced pre-mRNA.Indeed, U6 is the only spliceosomal RNA that is released from the post-catalytic spliceosome as a free RNA, rather than a snRNP, and therefore must re-associate with Lsm2-8 and Prp24 to form the U4 / U6 di-snRNP with each round of splicing ( 80 ).Additionally, our finding of a mono-U6 snRNP containing Bmc1 and Pof8 that promotes internal modifications of U6 is consistent with earlier reports of the human m 6 A methyltr ansfer ase METTL16 present in a mono-U6 snRNP with MePCE and LARP7 ( 81 ).Since mammalian U6 also undergoes 5 methyl phosphate capping by MePCE and LARP7media ted 2 -O-methyla tion, it will be interesting to examine the interplay between MePCE, LARP7, and METTL16, and how these factors may function in promoting the formation of the U4 / U6 di-snRNP in higher systems.
Tak en together, this w ork adds to the growing body of literature on the catalytic-independent functions of RNA modification enzymes (re vie wed in ( 82 )).While this raises questions as to the precise function of Bmc1 catalytic activity on the 5 end of U6, in vitro binding assays showed that catalytic activity of the human MePCE promotes 7SK retention following catalysis ( 83 ).It remains to be found if this extends to other MePCE / Bmc1 targets like U6, and how U6 snRNP assembly may be regulated in species lacking MePCE / Bmc1 and LARP7 homologs.

DA T A A V AILABILITY
The data supporting the findings of this study are available from the corresponding author upon reasonable request.RNA Seq data have been deposited in NCBI's Sequence Read Archi v e (SRA) database under BioProject number PRJNA918556.

SUPPLEMENT ARY DA T A
Supplementary Data are available at NAR Online.

Figure 2 .
Figure 2. Bmc1, Pof8, and Thc1 promote U6 snRNP assembly.( A ) Native northern blot analysis of spliceosomal and non-spliceosomal (U3) snRNPs from nati v e yeast cell extracts.( B ) Quantification of U6-containing snRNPs from wild type and knockout yeast cell extracts (mean ± standard error, two-tailed paired t test) ( n = 4 biological replica tes).( C ) Na ti v e northern b lot analysis of total and Bmc1-immunoprecipitated U6. ( D ) Solution hybridization of U4 / U6 pairing in wild type and knockout yeast strains using radiolabeled probes targeting the 5 end of U4 and 3 end of U6. ( E ) Quantification of U4 / U6 pairing from solution hybridization assay, expressed as the fraction of non-duplexed U4 and U6 ('free RNA') (mean ± standard error, two-tailed paired t test) ( n = 3 biological replicates).( F ) T m values from UV melt curve analysis of U4 / U6 pairing with unmodified and A64-2 -O-methylated U6 oligos (mean ± standard error, two-tailed paired t test) ( n = 6 technical replicates).( G ) Northern and western blot analysis of U4, U6 and myc-tagged Prp24 from total cell extracts and myc-immunoprecipita tes.( H ) Quantifica tion of Prp24-immunoprecipita ted U4 and U6, relati v e to Prp24 myc (mean ± standard error, two-tailed paired t test) ( n = 4 biological replicates).

Figure 3 .
Figure 3. Bmc1 catalytic activity is not a r equir ement for 2 -O-methylation of U6. ( A ) AlphaFold ( 84 ) structure prediction of Bmc1 aligned to the SAHbound (yellow) catalytic domain of MePCE (PDB 6DCB) ( 83 ) with mutations indicated in red.Inset: side chain interactions with SAH. ( B ) U6 2 -Omethylation primer extension in bmc1 Δ cells transformed with the indicated plasmid. 2 -O-methylated sites are indicated.Western blots for Bmc1-HA expression and b-actin are indicated below.( C ) Quantification of relati v e 2 -O-methylation-induced re v erse transcriptase stops at A64, compared to wild type Bmc1-HA (mean ± standard error, two-tailed paired t test) ( n = 4 biological replica tes).( D ) Quantifica tion of relati v e 2 -O-methylation-induced re v erse transcriptase stops at A64, compared to wild type Bmc1-HA, normalized to average Bmc1-HA expression relati v e to b-actin (mean ± standard error, two-tailed paired t test) ( n = 4 biological replicates).( E ) Western blot and northern blot analysis of co-immunoprecipitation of HA-tagged Bmc1, myc-tagged Pof8 and U6.