Ceg1 depletion reveals mechanisms governing degradation of non-capped RNAs

Most functional eukaryotic mRNAs contain a 7-methylguanosine (m7G) cap which serves as a platform that recruits proteins to support essential biological functions such as mRNA processing, nuclear export and translation. Although capping is accomplished during the first steps of transcription the fate and turnover of uncapped transcripts have not been studied extensively. Here, we employed fast nuclear depletion of the capping enzymes in Saccharomyces cerevisiae to uncover the turnover of the transcripts that failed to be capped. We show that the levels of non-capped mRNAs are determined principally by the abundance of their synthesis. Nuclear depletion of the capping enzymes increases the levels of lowly expressed mRNAs and decreases mRNAs that are highly transcribed altogether mimicking the effects observed in cells lacking the predominantly cytoplasmic 5’-3’ exonuclease Xrn1. The nuclear 5’-3’ exonuclease Rat1 is not involved in the degradation of cap-defective transcripts and the lack of the capping does not affect the distribution of RNA Polymerase II on the chromatin. Our data indicate that the mRNAs that failed to be capped are not directed to a specific quality-control pathway and that the 5’ cap role is associated with the Xrn1-dependent buffering of the cellular mRNA levels, along with protecting from 5’-3’ degradation.


INTRODUCTION
7 m7G cap antibody (H-20). The amount of capped ADH1 mRNA (normalised to human 151 GAPDH spike) was reduced by 80% in ceg1-AA (Rap) when compared to WT (control 152 cells grown on DMSO). The nuclear depletion of either Ceg1 or Cet1 (ceg1-AA or cet1-153 AA) resulted in a similar decrease in the ADH1 mRNA ( Fig. 1D and Fig. S1B) and 154 therefore, in the subsequent analyses, we focused mainly on the effects of Ceg1 155 removal. We also confirmed that the nuclear depletion of CBC did not affect ADH1 156 mRNA ( Fig. 1D and Fig. S1B). 157 Next, we performed a global analysis to investigate genome-wide the mRNA levels 158 after Ceg1 nuclear depletion. We depleted Ceg1 for 45 minutes (min) as this time was 159 sufficient to decrease the ADH1 level (Fig. S1B, 1D). Our RNA-seq analysis identified 160 1205 mRNAs affected by the nuclear depletion of Ceg1 (displaying log2 fold change 161 < -1 and > 1) (Fig. 1E). 736 (61%) of these mRNAs decreased when compared to the 162 control (WT), reflecting the fraction of non-capped transcripts that were subjected to 163 degradation. Unexpectedly, the levels of 469 genes (39%) increased (Fig. 1E). Both 164 effects were on average similar to each other (log2 fold change -1.4 and +1.5, 165 respectively) (Fig. S1C). We also tested how the levels of the mRNAs changed over 166 time during the nuclear depletion of Ceg1 (Fig. 1F). We selected ADH1, PYK2 and 167 PMA1 genes whose levels decreased upon Ceg1 depletion in the RNA-seq data.  qPCR analysis revealed that their mRNAs decreased even further after 90 min and 169 stabilised after 120 min. We also tested PML39, STP2 and CTA1 whose levels

Accumulation of non-capped mRNAs depends on their expression level 197
To understand the differential expression observed upon depletion of the capping 198 machinery we analysed the features of the affected mRNAs. First, we tested the basal 199 expression in WT cells (ceg1-AA on DMSO) for the differentially expressed mRNAs in 200 Ceg1-depleted cells. We found that increased mRNAs were generally lower 201 expressed (median transcript per million, log2 TPM = 2.7) than the decreased mRNAs 202 (median log2 TPM = 6.2) in WT ( Fig. 2A). Next, we split all 1205 protein-coding genes 203 differentially expressed in ceg1-AA into three bins according to their expression levels 204 in WT cells: the top 25% were classified as highly expressed (High), the bottom 25% 205 as lowly expressed (Low) and the remaining 50% as medially expressed (Mid) (Fig.  206 2B). Most of the lowly expressed genes increased their mRNA levels (291 out of 301), 207 while most of the highly expressed mRNAs were decreased (300 out of 301) after the however, they displayed the tendency to decrease the levels in Ceg1-depleted cells 210 (426 decreased versus 176 increased), correlating with increasing expression (Fig.  211 2B-C). Overall, we observed a progressive decrease of the log2 fold change for 212 mRNAs in ceg1-AA corelated with their increase in the basal expression levels in WT. 213 This was also reflected at the single gene level as shown for ZRT1, STB4 and ADH1 214 (Fig. 2D). Next, we employed available datasets estimating mRNAs half-life (Geisberg 215 et al., 2015), transcription rates (TR) and Pol II density in S. cerevisiae (Pelechano,216 Wei and Steinmetz, 2013) and applied them to the sets of differentially expressed 217 mRNAs identified in the ceg1-AA. These analyses revealed that the mRNAs which 218 increased upon Ceg1 nuclear depletion are more stable as they display longer half-219 lives (average half-life 41.9 vs 30.7 min for decreased mRNAs) ( Fig. 2E and S2A), 220 they also have lower transcription rates ( Fig. 2F and S2B) and Pol II density (Fig. S2C) 221 than the genes whose mRNAs decreased in Ceg1-depleted cells. The mRNAs 222 accumulating in ceg1-AA were also longer than those from the decreased set (average 223 gene length 1430 vs 1113 nt) (Fig. S2D) consistently with the fact that shorter genes 224 are generally associated with higher expression (Urrutia and Hurst, 2003). 225 Interestingly, the MEME suite (https://meme-suite.org/meme/) analysis identified a 226 very pronounced sequence motif (A/G)GAAAA strongly enriched in the upregulated 227 mRNAs (Fig. 2G) however, its function remains unknown. 228 Gene ontology (GO) analysis showed that some mRNAs upregulated in ceg1-AA (45 229 out of the 469 tested) were associated with genes coding for DNA binding proteins 230 directly associated with transcription ( Fig. S3A) including the zinc finger protein PPR1, 231 the regulator of stress response STB5 and PCF11, a component of the 3' end 232 processing factor (Fig. S3B). GO terms analysis did not reveal consistent terms   63/rat1-1 did not rescue their abundance (Fig. 3A). The profile of the FMP27 levels 286 expressed from the GAL1 promoter were mirrored by the levels of endogenous GAL1 287 mRNA (Fig. 3A). In contrast to the temperature sensitive mutants ceg1-63 and ceg1-288 63/rat1-1, native FMP27 mRNA expressed at low level did not decrease in ceg1-AA 289 but rather it remains stable (Fig. 3B). This may be caused by the difference in using ts 290 mutants versus rapid nuclear depletion to induce the capping defects as well as by a 291 significant difference between the expression from the native FMP27 and GAL1 292 promoters. Moreover, expression from the latter was already affected in the WT cells 293 shifted to 37 °C. However, we cannot exclude that FMP27 mRNA in ceg1-63 is 294 rescued by Rat1 inactivation at the sub-permissive temperature (34 o C) as previously 295

318
To avoid using the temperature-sensitive mutants we tested the Ceg1 nuclear 319 depletion combined with the deletion of the genes encoding the Rat1 binding partner 320 Rai1 (ceg1-AA/rai1∆) or the exonuclease Xrn1 (ceg1-AA/xrn1∆). The 5' to 3' transcripts. In contrast, the deletion of XRN1 rescued the levels of the non-capped than WT and was non-viable on Rap-containing medium (Fig. 3D), underlining the 333 importance of m7G cap for cellular processes (e.g., translation). 334 The rescue effect observed for ADH1 in ceg1-AA/xrn1Δ (Rap) prompted us to test 335 whether the differential expression in ceg1-AA could be rescued globally by the 336 deletion of the Xrn1 exonuclease. The XRN1 null mutants were shown to stabilise non-337 capped mRNAs in CEG1 temperature-sensitive mutants (Schwer, Mao and Shuman, 338 1998). We performed global differential expression analysis on the ceg1-AA/xrn1∆ 339 strain growing on either DMSO (xrn1D) or Rap for 45 min. The Ceg1 nuclear depletion 340 in the xrn1D background resulted in only 18 decreased mRNAs, 13 of which were also 341 reduced in ceg1-AA (Rap) (Fig. 3E-F). Unexpectedly, the number of increased mRNAs 342 was also significantly lower compared to ceg1-AA as we detected only 23 upregulated 343 mRNAs (Fig 3E-F). Consistently, upregulated genes displayed longer half-time compared to the 359 downregulated with a median of 30 min and 17 min, respectively (Fig. 4E). Next, we 360 tested if the effect of nuclear depletion of Ceg1 in xrn1∆ strain was masked by the pre-361 existing aberrant accumulation of mRNAs caused by the only XRN1 deletion. We 362 assessed how many differentially expressed mRNAs (log2 fold change < -1 and > 1) 363 in ceg1-AA were also differentially expressed in xrn1∆ prior to nuclear depletion of 364 Ceg1. We found that 190 out of 469 mRNAs increased and 264 out of 736 mRNAs 365 decreased in ceg1-AA were also increased or downregulated respectively in xrn1∆ 366 (Fig. 4F). This overlap identified only genes with log2 fold change over the ±1 367 threshold in both data sets therefore, we also tested the levels of differentially 368 expressed mRNAs in ceg1-AA in the total mRNA fraction in xrn1∆. Generally, most of 369 the increased and decreased mRNAs identified in ceg1-AA were similarly affected in 370 xrn1∆ (Fig. 4G) displaying a mean log2 fold change of 1.28 for the increased and -371 1.01 for the decreased mRNAs (Fig. 4H). We also performed reverse analysis and 372 tested the levels of increased and decreased mRNA in xrn1∆ (log2 fold change < -1 373 and > 1) on all mRNAs in ceg1-AA (Fig. 4I). Both groups are accordingly increased or 374 decreased in ceg1-AA (Rap) by a mean log2 fold change of 1.2 and -1.1 respectively 375 (Fig. 4J). This analysis indicates that both XRN1 deletion and lack of capping 376 deregulate mRNA levels through a common mechanism and therefore the Xrn1 and 377 Ceg1 co-inactivation did not result in differentially expressed genes (Fig. 3E). Finally, 378 we compared the RNA-seq dataset from ceg1-AA (Rap) and ceg1AA/xrn1∆ (Rap) to 379 exclude the genes similarly affected by either XRN1 deletion or Ceg1 depletion. We 380 identified 482 up-and 536 downregulated (Fig. S4A), both groups are similarly

401
Our data indicate that non-capped mRNAs are not subjected to any specific quality-402 control degradation upon Ceg1 depletion, but they are rather degraded according to 403 their expression. This and the similarities with the xrn1∆ mutation suggest that the cap 404 structure is required for the processes of sensing and buffering the global mRNA levels 405 by Xrn1. Non-capped mRNAs are not rapidly degraded upon Ceg1 depletion as they 406 are not optimal substrates for the 5'-3' exonucleases and require modifications of the 407 5' ends by other enzymes. We hypothesise that such modifications are efficiently 408 performed if the capping enzymes are in place which may represent a condition for 409 the cap quality control to occur. In this scenario, even lowly expressed non-capped 410 mRNAs with a suitable 5' end would be efficiently removed in WT cells. To test this 411 hypothesis, we introduced the self-cleaving hepatitis delta virus (HDV) ribozyme 412 sequence into the 5' UTR of the FMP27 gene (fmp27-RZ) (Fig. 5A). The internal self-413 cleavage of fmp27-RZ generated a non-capped transcript with a 5'-OH end which is & Doudna, 2014). As a control we used the catalytically inactive ribozyme mutant 416 harbouring the substitution C76:U (fmp27-C76U) (Perrotta, Shih and Been, 1999; Bird 417 et al., 2005). FMP27 levels did not change upon Ceg1 nuclear depletion (Fig. 3B). 418 However, in WT cells the introduction of the HDV ribozyme decreased the levels of 419 fmp27-RZ by 64% compared to the catalytically inactive fmp27-C76U (Fig. 5B). This 420 indicates that the generation of the correct substrate for exonucleases was required Mins on Rapamicyn Relative RNA levels ADH1 ceg1-AA cet1-AA cbp20-AA cbp80-AA WT Pol II CTD revealed only a slight reduction of Pol II occupancy 0.5 kb from the 436 transcription start site but not further downstream of the gene (Fig. 6A). Since the 437 average gene length in yeast is 1.4 kb (Hurowitz and Brown, 2003), we investigated 438 how transcription was affected by Ceg1 nuclear depletion on shorter genes (Fig 6B). 439 We depleted Ceg1 for 45 min and analysed the distribution of Pol II phosphoisoforms 440 on highly and lowly expressed genes: ADH1 and PML39, respectively (Fig. 6C-D). 441 ChIP analysis revealed that total Pol II was partially depleted mainly on the 5' ends for 442 both highly and lowly expressed genes. We used 8WG16 antibody recognising non-443 modified heptapeptides of the CTD to analyse total Pol II distribution. Thus, we 444 speculate that this decrease of the Pol II signal detected by this antibody may reflect 445 prematurely terminated unphosphorylated or hypophosphorylated Pol II. Indeed, both 446 ser5-P and ser2-P Pol II phosphoisoforms, that represent actively transcribing Pol II, 447 were only marginally affected by Ceg1 depletion (Fig. 6C-D).  S4B). We also investigated if longer nuclear depletion of Ceg1 resulted in 452 increased premature transcription termination. We shifted cells to the medium 453 containing Rap for up to 120 min and tested the levels of Pol II over the 3' end of high 454 and mid-range expressed ADH1, PYK1 and PMA1 genes (Fig. 6E). Consequently, we 455 observed only ~20% decrease in Pol II signal at 45 min which was not exacerbated 456 by longer nuclear depletion of Ceg1. 457   The nuclear depletion of Ceg1 resulted in the depletion of highly expressed mRNAs 489 and accumulation of lowly expressed and more stable mRNAs. We observed the same 490 pattern in cells lacking the 5'-3' exonuclease Xrn1 which is the major enzyme 491 degrading mRNA in the cytoplasm. Consistently, we did not detect any significant 492 number of differentially expressed genes in the xrn1D strain following nuclear 493 depletion of Ceg1. These observations allowed us to draw a few conclusions. First, 494 our data indicate that non-capped mRNA is not quality controlled but randomly 495 degraded. Therefore, the most abundant and so most accessible RNA species 496 decrease upon nuclear depletion of the capping machinery. This in turn, may increase 497 the relative concentration of lowly transcribed mRNAs in the overall fraction which we 498 detected as upregulated mRNAs. We speculate that the m7G cap may not be essential 499 for the protection from exonucleolytic 5'-3' degradation upon Ceg1 depletion, as the 5' 500 triphosphate non-capped mRNAs synthesised are not immediate substrates of the 5'- in the strain lacking Xrn1. Almost all genes upregulated and downregulated upon Ceg1 528 nuclear depletion were also respectively affected in the xrn1∆ strain. We speculate 529 that the lack of the cap may disturb the correct buffering of mRNA levels (Fig. 7) and Finally, we did not observe a significant increase in premature transcription termination 544 upon depletion of the capping enzymes. This is consistent with our data showing that 545 Rat1 and Rai inactivation did not restore the levels of non-capped mRNAs, but in 546 contrasts with previous models suggesting that such RNAs are degraded co-547 transcriptionally by Rat1 eliciting the removal of Pol II from DNA by the "torpedo 548 termination" mechanism, where 5'-3' exonuclease Rat1 attacks the 5' unprotected 549 end, chases the Pol II by degrading the nascent RNA, and ultimately displaces the Cells were cultured to OD600 = 1 and pelleted by centrifugation for 5 min at 2 400 x g 611 at room temperature (RT). The pellet was resuspended in sterile 10 mM Tris-HCl pH 612 7.5 or water and centrifuged again as before. The pellet was resuspended in filtered 613 LiT (10 mM Tris-HCl pH 7.5;100 mM Lithium acetate), 1 M DTT and incubated at RT 614 for 40 min with gentle shaking. After incubation the cells were pelleted again (as 615 before) and resuspended in LiT and 1 M DTT. The suspension was added with LiT; 616 and dsDNA (10 mg/mL); transforming DNA (0.1 to 1 ug or more) and incubated at RT 617 for 10 min. Next, PEG solution (1:1 PEG4000 : LiT) was added to the suspension and 618 incubated at RT for 10 min. DMSO was then added, and the suspension incubated at 619 42° C for 15 min. After the incubation the suspension was pelleted and resuspended 620 in 1 mL of YPD (1% Bacto yeast extract, 2% Bacto peptone, 2% glucose/dextrose). 621 Then the suspension was incubated at 30° C for 1 hour, pelleted with 10 seconds (sec) 622 spin, max speed and inoculated on selective plates.

RNA extraction 625
RNA was purified by phenol/chloroform method. The yeasts were cultured to the final 626 concentration OD600 = 1, pelleted in 50 mL falcon tubes by centrifugation for 2 min at 627 1000 rpm at room temperature (RT). The pellet was frozen at -80° C before further 628 manipulation. Next, the pellet was resuspended in 1 mL of ice-cold RNase free water, 629 moved to a 1.5 mL tube, centrifuged for 10 sec at 4° C max speed.

6-azauracil (6-AU) treatment 678
Overnight culture was prepared inoculating a single colony on minimal media as 679 described above. The next day the culture was diluted to OD600 = 0.2. The diluted 680 culture was grown with agitation at 30° C to OD600 = 1 and split in 4. Two cultures 681 were added of 6-azauracil (6-AU) and the other two with ammonium hydroxide 682