processing of human pre-U1 snRNA requires a combination of RNA and pro
http://www.100md.com
《核酸研究医学期刊》
Sir William Dunn School of Pathology, University of Oxford, South Parks Road, Oxford OX1 3RE, UK
*To whom correspondence should be addressed. Tel: +44 1865 275616; Fax: +44 1865 275556; Email: shona.murphy@pathology.ox.ac.uk
Correspondence may also be addressed to Patricia Uguen at present address: Signalisation, Développement et Cancer, Batiment 442 bis, Université Paris XI, 91405 Orsay cédex, France. Tel: +33 1 69 15 65 93; Fax: +33 1 69 15 68 02; Email: patricia.uguen@ibaic.u-psud.fr
ABSTRACT
Using an in vitro system we have recently shown that the 3' ends of human pre-snRNAs synthesized by RNA polymerase II are produced by RNA processing directed by the snRNA gene-specific 3' box. Towards a complete characterization of this processing reaction we have further investigated the in vitro requirements for proper 3' end formation of pre-U1 snRNA. Here we show that the 5' cap plays a stimulatory role and processing requires creatine phosphate. Our results also indicate that the pre-U1 processing activity is heat sensitive and that an RNA component is required. In addition, the exact sequence adjacent to the 3' box influences the position of the pre-U1 3' end produced in vitro. Interestingly, the processing extract active for 3'-box-dependent processing also contains an activity that converts the 3' end of RNA containing the U1 Sm protein binding site and the 3' terminal stem–loop into the mature form.
INTRODUCTION
The majority of vertebrate U snRNA genes (e.g. U1–U5) are transcribed by RNA polymerase II (pol II) to yield short non-polyadenylated 3'-elongated pre-snRNAs. The 3' ends of these RNAs are produced by a processing reaction dependent on the 13–16 nucleotide (nt) cis-acting 3' box element, located 9–19 nt downstream of the 3' end of the RNA-encoding region (1). Most of the 3' extension is removed by further processing in the cytoplasm before nuclear re-import (2,3). A compatible snRNA promoter is also required for proper 3' end formation of pre-snRNAs (4,5), highlighting the tight link between transcription and the function of the 3' box. Relevant to this we have recently demonstrated that the C-terminal domain (CTD) of pol II is required for 3'-box-dependent RNA processing in vivo and that processing in vitro is activated by phospho-CTD (1,6).
3' End formation of pre-snRNAs is likely to involve divalent-cation-dependent endonucleolytic activity, in common with 3' end formation of vertebrate pre-mRNAs or yeast pre-snRNAs (1,7). However, neither subunits of the cleavage–polyadenylation complex factors cleavage/polyadenylation-specificity factor (CPSF), cleavage-stimulation factor (CstF) and cleavage factor I (CFI), nor the human homologues of yeast pre-snRNA processing factors hRNase III and PM-Scl100 appear to co-purify with the human pre-snRNA 3' processing activity (1). This suggests that there is little overlap between known RNA 3' end processing factors and those used for 3' end formation of human pre-snRNAs. Two cis-acting elements, the U2 3' box and the upstream sequence element (USE) act together to efficiently direct accurate pre-U2 snRNA 3' end formation both in vitro and in vivo (1), whereas 3' end formation of pre-U1 snRNA appears to depend largely on the 3' box. Thus, there may also be differences in the cis-acting sequences and trans-acting factors involved in 3' end formation of different pre-snRNAs.
Further investigation of the biochemical requirements of pre-U1 snRNA 3' end formation indicates that, in addition to the phospho-CTD of pol II, creatine phosphate and a 5' cap activate processing. We also demonstrate that 3' processing requires an RNA component and that the activity is heat labile. Thus, 3'-box-dependent processing has some characteristics in common with both AAUAAA-dependent processing (8) and 3' end formation of the replication-activated histone mRNAs (9) as well as significant differences. Furthermore, we show that the efficiency of in vitro pre-U1 snRNA 3' end formation is not affected by additional cis elements although the exact sequence around the cleavage site influences the site of the final 3' end formed in vitro. The S100 fraction active for 3' end formation of pre-U1 snRNA is also active for 3' end maturation and we show that these processes can occur independently in vitro.
MATERIALS AND METHODS
DNA constructs
In 10end U1 3' box, 17end U1 3' box and 59end U1 3' box templates, the 10, 17 and 59 nt, respectively, located upstream of the 3' box of U1 3' box template described by Uguen and Murphy (1) were replaced by the 10, 17 and 59 nt of the endogenous U1 snRNA gene (GenBank accession number J00318 ). The U1 3' box sequence, GTTTCAAAAGTAGA, is deleted from the 59end 3' box template. DNA constructs used to produce synthetic RNA were obtained by deleting the U1 promoter region by EcoRI/BglII digestion as described previously (1).
S100 fraction preparation
HeLa cell S100 extract was prepared using a slight modification of the procedure of Shapiro et al. (10). After cell lysis in buffer A (10 mM HEPES pH 7.9, 0.75 μM spermidine, 0.15 μM spermine, 10 mM KCl, 0.5 μM DTT) sucrose was added to a final concentration of 10% before spinning.
In vitro RNA synthesis and in vitro processing
Synthetic, capped RNA was produced as described previously (1). To produce uncapped RNA, m7G(5')ppp(5')G was omitted from the reaction. Processing was carried out as described in (1) for 2.5 h at 30°C in the presence of 3 mM MnCl2 unless stated otherwise in the figure legends. To reduce the level of endogenous ATP, the S100 fraction was bound to heparin–sepharose at 100 mM KCl and eluted by 1 M KCl. This fraction was then incubated for 15 min at 30°C with 3 mM MgCl2, 2 mM glucose and 0.6 U of hexokinase (Boehringer Mannheim) to deplete any residual ATP.
S1 mapping analysis
After incubation with S100, RNA was extracted, annealed to the S1 probe, digested and analysed on a 6% polyacrylamide–8 M urea gel as described previously (1). The S1 probes used for each construct (U1 3' box, 10end U1 3' box, 17end U1 3' box, 59end U1 3' box and 59end 3' box) were prepared as described previously (1).
Western blotting
Western blotting was carried out as described by Medlin et al. (6). Anti-stem–loop-binding protein (SLBP) antibody was kindly provided by Berndt Mueller (University of Aberdeen) and used diluted at 1/100.
RESULTS
Improving U1 3'-box-dependent in vitro processing
We recently described an in vitro system where 3'-box-dependent processing of substrate RNAs was detected for the first time (1). However, long incubation times (>5 h) are required to detect any accurate 3' end processing directed by the U1 3' box in unpurified S100. We therefore undertook to improve the efficiency of the processing activity by modifying the extract preparation procedure (see Materials and Methods) as a prelude to characterization of the requirements for 3' processing. Modification of the extract preparation method described by Shapiro et al. (10) led to the production of S100 that yields up to 8% properly processed RNA after 2.5 h incubation rather than the 13 h needed with S100 produced by our previously described protocol (1). The increased sucrose concentration used in production of this extract (see Materials and Methods) may reduce the release of nucleases from the nuclei upon spinning and consequently reduce turnover of substrate RNA, which proved to be a major problem for in vitro processing (1). Relevant to this, the amount of pol II and factors involved in mRNA 3' end formation is reduced in this S100 fraction (data not shown and see Fig. 4B).
Figure 4. U1 3'-box-dependent processing activity is heat labile. (A) U1 3' box RNA was incubated with S100 preheated for 10 min at the indicated temperature (temp.). The graph represents the percentage processing compared with the level obtained when S100 is preheated at 30°C. (B) Western blot with SLBP antibody on the S100 used for processing in Figures 1–5 (lane 1) and on the S100 described by Uguen and Murphy (1), which is less active for 3' processing (lane 2). SLBP is 45 kDa. (C) Western blot with anti-SLBP antibody on the different fractions resulting from purification of the processing activity on different sepharose resins as described by Uguen and Murphy (1). B-heparin and B-Q represent proteins bound to heparin– and Q-sepharose, respectively; UB-SP represents proteins that did not bind to SP-sepharose.
Efficient U1 3'-box-dependent processing requires creatine phosphate as a co-factor and is ATP independent
Since creatine phosphate (CP) (usually used to transfer Pi to ADP, creating ATP) improves the efficiency of some mRNA processing steps (11), we tested its effect on U1 3'-box-dependent processing (Fig. 1). Little processing (1%) is detected in the absence of added CP after incubation for 2.5 h at 30°C (Fig. 1A, lanes 2 and 4), whereas 7% of the input RNA is processed in the presence of 20 mM CP (Fig. 1A, lanes 3 and 5). Processing efficiency is only slightly improved by increasing the concentration of CP further (8% of input RNA is processed in the presence of 40 or 60 mM CP; Fig. 1A, lanes 6 and 7). Furthermore, nucleoside triphosphates, including ATP, do not support U1 3'-box-dependent processing in the absence of CP (Fig. 1A, lane 4, and B, lanes 3–6), indicating that CP does not act simply to regenerate the energy source. Serine phosphate (SP) can also activate processing although not as effectively as CP (Fig. 1C, lanes 3–5) suggesting that the effect of CP is not entirely due to the presence of phosphate groups.
Figure 1. Efficient U1 3'-box-dependent processing requires CP as a co-factor and is ATP independent. (A) U1 3' box RNA was incubated with S100 in the presence of ATP (1 mM) and different concentrations of CP as indicated. The percentage of input RNA processed (% proc.) is indicated below this and subsequent gels. The scheme represents the position and size of the S1 probe, input (IP) and processed (proc.) RNA. The diagram on the left side of this and subsequent figures indicates the position of the U1 3' box in the input RNA. (B) U1 3' box RNA was incubated with S100 in the presence of 20 mM CP or 1 mM of the nucleosides triphosphates indicated. (C) U1 3' box RNA was incubated with S100 in the presence of 20 mM CP or different concentrations of SP as indicated. (D) U1 3' box RNA was incubated with S100 in the presence or absence of 20 mM CP and in the presence of phospho-CTD as indicated. (E) U1 3' box RNA was incubated with S100 fractionated on heparin–sepharose in the presence of 20 mM CP, and with or without ATP, hexokinase and glucose. The processing reaction was carried out in the presence of 2 mM MgCl2.
It has been proposed that CP activates pre-mRNA 3' cleavage by mimicking the effect of the pol II CTD rather than affecting the reservoir of energy (11,12) and may facilitate a conformational change of some factors. In support of this notion, the phosphorylated form of the CTD of pol II activates 3'-box-dependent processing in vitro better than the unphosphorylated form (1). However, no processing is detected in the absence of CP when phospho-CTD is added (Fig. 1D). This indicates that CP does not simply substitute for phospho-CTD in our system.
To eliminate the ATP present in the S100 fraction, we fractionated the processing activity by binding to heparin–sepharose at 100 mM KCl and eluting by 1 M KCl. Although this fractionation procedure should markedly reduce the level of ATP in the active fraction it does not result in a requirement for added ATP (Fig. 1E, lanes 2 and 6). Furthermore, addition of both glucose and hexokinase (Fig. 1E, lane 5), which together deplete the available pool of ATP (11), has no greater effect than the addition of glucose or hexokinase alone (Fig. 1E, lanes 3 and 4). These results indicate that 3'-box-dependent processing is ATP independent. Interestingly, AAUAAA-dependent 3' cleavage is also CP dependent and ATP independent in vitro (11).
U1 3'-box-dependent processing is activated by a 7-methyl-G cap
To examine whether 5' cap binding proteins activate 3'-box-dependent RNA processing in addition to splicing, 3' processing, transport and translation of mRNA (13), we compared the efficiency of 3' processing of capped and uncapped U1 3' box RNA. In the absence of a 5' cap structure, U1 3'-box-dependent processing is slower (Fig. 2A), suggesting that either the 5' cap itself, or more likely, cap binding proteins present in the S100, activate processing. The decrease of proper 3' end formation of uncapped RNA is not due simply to RNA instability since input RNA is still detected after 3 h incubation (Fig. 2A). Addition of 7-methyl-G cap analogue to the in vitro reaction inhibits processing and the highest concentration of cap analogue tested effectively abolished processing (Fig. 2B, lanes 8–12). This suggests that the cap analogue can sequester cap binding proteins and interacting 3' processing factors. Consistent with this, the effect of the cap structure of the RNA on processing efficiency is lost after partial purification of the 3' processing activity as described by Uguen and Murphy (1) (data not shown), indicating that the 5' cap per se is not sufficient. Moreover, addition of unmethylated cap analogue G(5')ppp(5')G to the reaction has little effect on processing (Fig. 2B, lanes 3–7). Taken together, these results suggest that the nuclear cap binding complex (14) may be involved in the recruitment of pre-snRNA 3' processing factor(s) in vivo.
Figure 2. U1 3'-box-dependent processing is activated by a 7-methyl-G cap. (A) Capped (empty square) or uncapped (filled diamond) U1 3' box RNA was incubated with S100 for 0–3 h. The graph below represents the percentage relative processing activity calculated by comparison with the amount of processing obtained after incubating capped RNA for 3 h. (B) Capped U1 3' box RNA was incubated with S100 in the presence of different concentrations of unmethylated cap analogue (lanes 3–7) or methylated cap analogue (lanes 8–12) as indicated. The graph below represents the average percentage relative processing activity obtained in the presence of increasing concentrations of unmethylated (filled triangle) or methylated (empty square) cap analogue in two independent experiments.
U1 3'-box-dependent processing requires an RNA co-factor
Small non-coding RNAs function as co-factors in both splicing and 3' processing of pre-mRNAs (8,9). To determine whether 3' processing of U1 3' box RNA substrate is dependent on a trans-acting nucleic acid component, the S100 was treated with micrococcal nuclease (MN) before incubation with U1 3' box RNA (Fig. 3). U1 3'-box-dependent processing is completely abolished by this treatment (lane 3). However, when EGTA, which chelates the Ca2+ ions essential for this nuclease to function, is added to the S100 before MN, no reduction in processing is observed (lane 4). This suggests the involvement of an RNA or DNA co-factor in U1 3'-box-dependent processing. To distinguish between these two possibilities, we treated the S100 with pancreatic DNase I before incubation with U1 3' box RNA (lane 8). This has no effect on 3' processing, indicating that the nucleic acid co-factor is composed of RNA. In order to test whether this RNA component is the sole catalytic component of the 3' processing activity, we extracted ribonucleic acid from the S100 by acidic phenol/chloroform treatment and incubated it with U1 3' box RNA. This fraction does not support 3' processing on its own (lane 5). In addition, 3' processing is not recovered when the MN-treated S100 is complemented with this RNA fraction (lane 6). These data suggest that U1 3'-box-dependent processing involves both RNA and protein factor(s) and leads us to propose that processing is carried out by an RNA–protein complex.
Figure 3. U1 3'-box-dependent processing requires an RNA co-factor. U1 3' box RNA was incubated with S100 in the presence of 2 mM of MgCl2 before or after pretreatment with MN or DNase I and/or complemented with the extracted RNA. MN (lane 3), S100 pretreated for 15 min at 25°C with MN (5 U/μl of S100; Roche) in the presence of 1 mM CaCl2. The reaction was stopped by adding 2 mM EGTA. MN in. (lane 4), S100 pretreated for 15 min at 25°C with MN inactivated by the addition of 2 mM EGTA; RNA extracted (lane 5), S100 extracted by acidic phenol/chloroform treatment and ethanol precipitated before resuspension in buffer D (15); RNA extracted + MN (lane 6), S100 pretreated for 15 min at 25°C with MN in the presence of 1 mM CaCl2 and complemented by the extracted RNA. RNA extracted + MN in. (lane 7), S100 pretreated for 15 min at 25°C with MN inactivated by the presence of 2 mM EGTA and complemented by the extracted RNA; DNase I (lane 8), S100 pretreated for 15 min at 30°C with pancreatic DNase I (10 U/μl of S100).
U7 snRNA is required for 3' end formation of the mRNAs for replication-activated histones and has a region of complementarity to the substrate RNA (9). Since mammalian U7 snRNAs contain no conserved sequences complementary to 3' box sequences it is unlikely that U7 snRNA participates in 3' end formation of U1 snRNA. Unfortunately, experiments to test the effect of depletion of U7 snRNA and other U snRNAs from the processing extract using antibodies to the trimethylguanosine cap present at the 5' end were inconclusive.
U1 3'-box-dependent processing activity is heat labile
To analyse the sensitivity of the 3'-box-dependent processing activity to heat, the S100 fraction was heated for 10 min at temperatures from 30 to 55°C before incubation with U1 3' box RNA (Fig. 4). The efficiency of 3' processing is slightly reduced at 45°C (inhibition of 33%; Fig. 4A, lane 4). However, 3' processing is inhibited by 83% after preheating the S100 fraction at 50°C (lane 5) and is lost completely after treatment at 55°C (lane 6), suggesting that U1 3'-box-dependent processing involves heat-sensitive factor(s).
3' Processing of replication-activated histone mRNAs is similarly heat labile (16), raising the possibility of shared factors with U1 3'-box-dependent processing. A well-characterized SLBP is required for histone 3' end formation (9) and we have used an anti-SLBP antibody to analyse our processing extract by western blot (Fig. 4B). SLBP is not readily detected in the S100 used for the in vitro processing reactions shown in Figure 4A (Fig. 4B, lane 1) although it is relatively abundant in our previously described S100 preparation (1) (Fig. 4B, lane 2). In addition, no SLBP is detected in the partially purified 0.4 M KCl SP fraction active for pre-snRNA 3' end formation (1) (Fig. 4C, lane 6). These results suggest that SLBP is not involved in 3'-box-dependent processing.
Sequences upstream from the U1 3' box can influence 3' processing
We have shown that accurate pre-U2 snRNA 3' end formation requires two cis-acting elements, USE and U2, acting synergically. Although the U1 3' box is sufficient to direct accurate formation of RNA 3' ends in vitro and in vivo (1,4,17), the last 5–6 nt of the mature 3' end of U1 snRNA can affect the in vivo accuracy of pre-U1 snRNA 3' end formation (17). We therefore analysed the influence of the natural sequences preceding the U1 3' box on the efficiency and accuracy of 3' processing in vitro. When RNA containing the 10 nt just upstream of U1 3' box (10end U1 3' box RNA) is incubated with S100, no properly processed RNA is detected (Fig. 5A, lanes 6–10, and C), even after 3 h incubation. This may be due to instability of processed RNA rather than to a complete loss of processing. When the 17 nt upstream of the U1 3' box are present (17end U1 3' box), processing is detected (Fig. 5A, lanes 11–15, and C) although the efficiency is reduced by 45% with respect to U1 3' box RNA (Fig. 5A, lanes 1–5, and C). A small amount of the 3' end processed RNA maps to the same nucleotides seen with the U1 3' box RNA but most map a few nucleotides upstream, corresponding to the most highly represented 3' ends detected in vivo, which end 6–8 nt upstream of the 3' box (17,18) (Fig. 5A, compare lanes 1–5 and 11–15, and C). These results may indicate that the addition of 17 nt of the natural U1 sequence influences the position of 3' endonucleolytic cleavage. Alternatively, 3' exonucleolytic trimming of the RNA occurs after cleavage 3–5 nt upstream from the 3' box.
Figure 5. Sequences upstream from the U1 3' box can influence 3' processing. (A) U1 3' box, 10end U1 3' box and 17end U1 3' box RNA substrates were incubated with S100 for 0–3 h. The scheme below indicates the position and size of the S1 probes, input (IP) and processed (proc.) RNA. (B) U1 3' box, 59end U1 3' box and 59end 3' box RNA substrates were incubated with S100. The scheme below indicates the position and size of the S1 probes, input (IP), processed (proc.) and mature RNA. (C) The sequences of the 3' ends of the substrate RNAs in (A) and (B) are shown with the position of in vivo and in vitro processing sites and mature ends. The endogenous U1 snRNA sequence is highlighted in grey. The U1 3' box is boxed. The vertical line indicates the boundary between the end of the mature snRNA and the first nucleotide of the non-coding RNA. The bold nucleotides represent the processing sites observed in vivo (17,18). The arrows above the sequence indicate the pre-snRNA processing sites observed in vitro. The mature ends observed in vitro are indicated by unfilled arrows above the sequence.
We have also tested the effect of placing a longer endogenous sequence (59 nt) upstream of the U1 3' box. This substrate RNA contains both the Sm protein binding site and the 3' terminal stem–loop, which are not required in vivo for pre-U1 snRNA 3' end formation but are required for efficient nuclear export (19). When 59end U1 3' box RNA (Fig. 5B, lanes 3 and 4) is incubated with S100, RNA with the mature 3' end mapping 10–12 nt upstream of the 3' box (17) is produced. This indicates that the activities required for production of mature U1 3' ends (20) are present in the S100, in addition to factors required for pre-U1 3'-box-dependent processing. Some accurately processed pre-U1 RNA 3' ends are still detected and these map to the same sites as are observed when only 17 nt of natural U1 sequence are present upstream of the 3' box (17end U1 3' box RNA; Fig. 5A, lanes 11–15, and C). This may suggest that processing proceeds in a stepwise manner as it does in vivo and the precise pre-snRNA 3' ends are required for the next step, which matures the 3' end of the RNA, as has been shown for 3' end formation of U2 snRNA (21). 3' Ends mapping 6–8 nt upstream from the 3' box are no longer detected when the 3' box is removed (Fig. 5B, lanes 5 and 6, and C). However, production of the mature U1 3' ends is unaffected (Fig. 5B, lane 6), indicating that these two processing events occur independently in our in vitro system. This suggests that any RNAs containing the Sm protein binding site and the 3' terminal stem–loop can be 3' end matured, presumably by exonucleolytic trimming. Since the 59end 3'box RNA has 27–29 nt downstream from the sites of the mature 3' ends, these results also indicate that RNAs longer than pre-U1 at the 3' end are readily 3' matured in vitro. In contrast, only pre-U1 snRNAs with a short 3' extension (no more than 10 nt) are further processed after injection into Xenopus oocytes (20). This may highlight differences between species in the 3' end maturation of snRNAs. Alternatively, the exact sequence of the 3' extension may influence trimming and/or some in vivo constraints may be lost in our in vitro system.
DISCUSSION
The first step of expression of the vertebrate U1 snRNA genes involves the production of a 3' extended pre-U1 snRNA, which occurs by a co-transcriptional 3'-box-dependent mechanism (4,5). Proper 3' end formation of pre-U1 snRNA is important for the stability of the snRNA in the nucleus, in addition to nucleocytoplasmic export (19,20). The exact 3' ends of U1 snRNA precursors can also influence the cytoplasmic 3' trimming of the RNA required for the nuclear translocation of U1 snRNPs and subsequent maturation steps (20). Towards the dissection of the molecular mechanism of 3' end formation of human pre-U1 snRNA we have determined several of the co-factor requirements of U1 3'-box-dependent in vitro processing. We have also investigated the influence of sequences upstream of the U1 3' box on 3' end formation. The results of these studies highlight similarities and differences between 3' end formation of pre-snRNAs and other vertebrate pol II transcripts. In addition, we show that 3' extended RNAs containing the U1 Sm protein binding site and the 3' terminal stem–loop can be processed to the mature U1 3' end in vitro.
Comparing U1 3'-box-dependent processing and formation of mRNA 3' ends
Some co-factor requirements for 3'-box-dependent processing are shared with AAUAAA-dependent processing. Both processing events require divalent cations and CP, are ATP independent and are activated by 5' cap binding proteins (8,11,13) (Fig. 6). In addition, both require the CTD of pol II in vivo (6,13) and are activated by phospho-CTD in vitro (1,12) (Fig. 6). So far there is no indication that any subunits of CPSF, CstF and CFI factors required for AAUAAA-dependent processing participate in pre-snRNA 3' end formation (1). However, it has been shown that CFI participates in 3' processing of some yeast snoRNAs and snRNAs (22).
Figure 6. Requirements for U1 3'-box-dependent RNA processing. 3'-Box-dependent 3' end formation of pre-U1 snRNA requires an RNA–protein complex, CP and divalent cations (div. ion). The reaction is also activated by 5' cap binding proteins (diamond) and the CTD of pol II. In addition, the processing activity is abolished by relatively mild heat treatment, which suggests the involvement of a heat-labile factor (HLF).
There are also some similarities between 3' end formation of U1 snRNA and replication-activated histone mRNAs. In both cases a stem–loop structure is present just upstream of the mature 3' end, the processing activities are heat labile and a trans-acting RNA component is required (9). Our results therefore point to mechanistic similarities in 3' end formation of non-polyadenylated histone mRNA and snRNA. However, our data indicate that the SLBP is not involved in U1 snRNA 3' end formation. In addition, the lack of complementarity between U7 snRNA and 3' box sequences makes it likely that a distinct RNA is required.
Divalent cations are also necessary for efficient activity of the yeast RNase III homologues, Pac1p and Rnt1p (23,24) which are involved in yeast snRNA 3' end formation (25). However, human RNase III does not co-purify with the 3'-box-dependent processing activity and the cis-acting elements do not correspond to RNase III recognition sequences (1). Divalent cations may be required for the function of a 3'-box-specific processing factor or play a more direct catalytic role.
The currently known requirements for 3' end processing of pre-U1 snRNA are summarized in Figure 6. Taken together our results indicate that there is little overlap in the factors required for 3'-box-dependent processing and mRNA 3' end formation apart from the shared co-factors noted above. This is consistent with the differences in cis-acting sequences and the differential promoter dependence of these signals in vivo (4,5).
Sequences immediately upstream of the U1 3' box influence 3' end formation
The efficient production of accurately 3' processed pre-U2 snRNA requires two distinct sequences: the U2 3' box and the USE, which contains a ‘minihelix’ stem–loop structure (1). The U2 3' box alone directs formation of RNA a few nucleotides longer than cellular pre-U2 snRNA while RNA containing the USE is processed to the authentic pre-U2 snRNA 3' end. In addition, the amount of processed product detected in vitro is increased when both the U2 3' box and the USE are present. The minihelix may therefore influence processing by interacting with a processing factor and/or prevent RNA cleaved by a 3'-box-dependent endonuclease from being rapidly turned over by exonucleases (1). Addition of 17 nt of the natural U1 gene-encoded sequence upstream of the U1 3' box also causes an apparent shift in the position of the final 3' ends, some of which now map to the most abundant pre-U1 3' ends detected in cellular RNA (17,18). The sequences directly upstream from the U1 3' box may therefore play a similar role to the U2 USE to determine the final position of the pre-U1 3' end. In both cases the site of endonucleolytic cleavage may be influenced by the exact sequence of the target RNA or changes in processing–factor RNA interaction. Alternatively, the natural sequences may be preferential substrates for limited trimming by exonucleases following 3' box-directed cleavage. However, unlike the U2 USE, the 17 nt upstream from the U1 3' box do not cause an increase in the amount of processed RNA detected but rather a decrease. In addition, unlike the U2 USE (1), the 59 nt of U1 sequence upstream of the 3' box do not direct formation of pre-U1 3' ends, suggesting functional differences. In agreement with this, little authentic pre-U1 RNA is produced from transfected U1 templates lacking the 3' box (17).
A precise pre-U1 3' end is not required for RNA maturation in vitro
Although pre-U2 snRNA can be further 3' processed in vitro (2,21), maturation of the 3' ends of pre-U1 snRNA in vitro has not previously been reported. However, in our in vitro system the inclusion of U1 snRNA sequences containing the Sm protein binding site and the 3' terminal stem–loop (19) in the RNA substrate results in the production of properly processed mature U1 3' ends. Interaction of Sm protein with its binding site is required for the cytoplasmic 5' cap modification of U1 snRNA, which is necessary for nuclear re-import (26) and it also plays a role in maturation of the 3' ends of yeast U1 snRNA (27, reviewed in 3,28). Surprisingly, in vitro 3' end maturation occurs independently of 3'-box-dependent processing, suggesting that no intermediate processing step is required even though the substrate U1 RNAs have relatively long 3' extensions (27–29 nt). Instead, 3' end extensions longer than 10 nt are not processed to the mature form when human pre-U1 RNAs are injected into Xenopus oocytes (20). In addition, specialized sequences distinct from the Sm protein binding site are required for cytoplasmic trimming of the 3' end of human U2 snRNA (29), and pre-U2 RNA with the natural 3' ends are preferential substrates for the reaction in vitro (2,21). There may therefore be differences in 3' end maturation of Xenopus and human U1 snRNAs, and between human U1 and U2 snRNAs. The availability of in vitro systems for both 3'-box-dependent processing and maturation of 3' extended pre-U1 snRNAs will facilitate the molecular analysis of all steps leading to accurate formation of U1 3' ends.
ACKNOWLEDGEMENTS
We thank Alice Taylor for excellent technical assistance. We are grateful to Berndt Mueller for the kind gift of the anti-SLBP antibody. We also thank Zbigniew Dominski, Nick Proudfoot and Andre Furger for helpful suggestions and comments on the manuscript. P.U. was supported by MRC Co-operative Component grant No. G9900343 and the Edward Penley Abraham Trust. S.M. was supported by MRC Senior Fellowship No. G117/309.
REFERENCES
Uguen,P. and Murphy,S. ( (2003) ) The 3' ends of human pre-snRNAs are produced by RNA polymerase II CTD-dependent RNA processing. EMBO J., , 22, , 4544–4554.
Huang,Q., Jacobson,M.R. and Pederson,T. ( (1997) ) 3' Processing of human pre-U2 small nuclear RNA: a base-pairing interaction between the 3' extension of the precursor and an internal region. Mol. Cell. Biol., , 17, , 7178–7185.
Will,C.L. and Lührmann,R. ( (2001) ) Spliceosomal UsnRNP biogenesis, structure and function. Curr. Opin. Cell Biol., , 13, , 290–301.
Hernandez,N. and Weiner,A.M. ( (1986) ) Formation of the 3' end of U1 snRNA requires compatible snRNA promoter elements. Cell, , 47, , 249–258.
Neuman de Vegvar,H.E., Lund,E. and Dahlberg,J.E. ( (1986) ) 3' End formation of U1 snRNA precursors is coupled to transcription from snRNA promoters. Cell, , 47, , 259–266.
Medlin,J.E., Uguen,P., Taylor,A., Bentley,D. and Murphy,S. ( (2003) ) The C-terminal domain of pol II and a DRB-sensitive kinase are required for 3' processing of U2 snRNA. EMBO J., , 22, , 925–934.
Ryan,K., Calvo,O. and Manley,J.L. ( (2004) ) Evidence that polyadenylation factor CPSF-73 is the mRNA 3' processing endonuclease. RNA, , 10, , 565–573.
Wahle,E. and Rüegsegger,U. ( (1999) ) 3'-End processing of pre-mRNA in eukaryotes. FEMS Microbiol. Rev., , 23, , 277–295.
Dominski,Z. and Marzluff,W.F. ( (1999) ) Formation of the 3' end of histone mRNA. Gene, , 239, , 1–14.
Shapiro,D.J., Sharp,P.A., Wahli,W.W and Keller,M.J. ( (1988) ) A high-efficiency HeLa cell nuclear transcription extract. DNA, , 7, , 47–55.
Hirose,Y. and Manley,J.L. ( (1997) ) Creatine phosphate, not ATP, is required for 3' end cleavage of mammalian pre-mRNA in vitro. J. Biol. Chem., , 272, , 29636–29642.
Hirose,Y. and Manley,J.L. ( (1998) ) RNA polymerase II is an essential mRNA polyadenylation factor. Nature, , 395, , 93–96.
McCracken,S., Fong,N., Rosonina,E., Yankulov,K., Brothers,G., Siderovski,D., Hessel,A., Foster,S., Shuman,S. and Bentley,D.L. ( (1997) ) 5'-Capping enzymes are targeted to pre-mRNA by binding to the phosphorylated carboxy-terminal domain of RNA polymerase II. Genes Dev., , 11, , 3306–3318.
Flaherty,S.M., Fortes,P., Izaurralde,E., Mattaj,I.W. and Gilmartin,G.M. ( (1997) ) Participation of the nuclear cap binding complex in pre-mRNA 3' processing. Proc. Natl Acad. Sci. USA, , 94, , 11893–11898.
Dignam,J.D., Lebovitz,R.M. and Roeder,R.G. ( (1983) ) Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res., , 11, , 1475–1489.
Gick,O., Kramer,A., Vasserot,A. and Birnstiel,M.L. ( (1987) ) Generation of histone mRNA 3' ends by endonucleolytic cleavage of the pre-mRNA in a snRNP-dependent in vitro reaction. Proc. Natl Acad. Sci. USA, , 84, , 8937–8940.
Hernandez,N. ( (1985) ) Formation of the 3' end of U1 snRNA is directed by a conserved sequence located downstream of the coding region. EMBO J., , 4, , 1827–1837.
Patton,J. and Wieben,E. ( (1987) ) U1 precursors: variant 3' flanking sequences are transcribed in human cells. J. Cell Biol., , 104, , 175–182.
Terns,M.P., Dahlberg,J.E. and Lund,E. ( (1993) ) Multiple cis-acting signals for export of pre-U1 snRNA from the nucleus. Genes Dev., , 7, , 1898–1908.
Neuman de Vegvar,H.E. and Dahlberg,J.E. ( (1990) ) Nucleocytoplasmic transport and processing of small nuclear RNA precursors. Mol. Cell. Biol., , 10, , 3365–3375.
Huang,Q. and Pederson,T. ( (1999) ) A human U2 RNA mutant stalled in 3' end processing is impaired in nuclear import. Nucleic Acids Res., , 15, , 1025–1031.
Morlando,M., Greco,P., Dichtl, B., Fatica,A., Keller,W. and Bozzoni,I. ( (2002) ) Functional analysis of yeast snoRNA and snRNA 3'-end formation mediated by uncoupling of cleavage and polyadenylation. Mol. Cell. Biol., , 22, , 1379–1389.
Rotondo,G. and Frendewey,D. ( (1996) ) Purification and characterization of the Pac1 ribonuclease of Schizosaccharomyces pombe. Nucleic Acids Res., , 24, , 2377–2386.
Lamontagne,B. and Elela,S.A. ( (2001) ) Purification and characterization of Schizosaccharomyces pombe Rnt1p nuclease. Methods Enzymol., , 342, , 159–167.
Allmang,C., Kufel,J., Chanfreau,G., Mitchell,P., Petfalski,E. and Tollervey,D. ( (1999) ) Functions of the exosome in rRNA, snoRNA and snRNA synthesis. EMBO J., , 18, , 5399–5410.
Mattaj,I.W. ( (1986) ) Cap trimethylation of U snRNA is cytoplasmic and dependent on U snRNP protein binding. Cell, , 46, , 905–911.
Seipelt,R.L., Zheng,B., Asuru,A. and Rymond,B.C. ( (1999) ) U1 snRNA is cleaved by RNase III and processed through an Sm site-dependent pathway. Nucleic Acids Res., , 27, , 587–595.
Yong,J., Pellizzoni,L. and Dreyfuss,G. ( (2002) ) Sequence-specific interaction of U1 snRNA with the SMN complex. EMBO J., , 21, , 1188–1196.
Jacobson,M.R., Rhoadhouse,M. and Pederson,T. ( (1993) ) U2 small nuclear RNA 3' end formation is directed by a critical internal structure distinct from the processing site. Mol. Cell. Biol., , 13, , 1119–1129.(Patricia Uguen and Shona Murphy*)
*To whom correspondence should be addressed. Tel: +44 1865 275616; Fax: +44 1865 275556; Email: shona.murphy@pathology.ox.ac.uk
Correspondence may also be addressed to Patricia Uguen at present address: Signalisation, Développement et Cancer, Batiment 442 bis, Université Paris XI, 91405 Orsay cédex, France. Tel: +33 1 69 15 65 93; Fax: +33 1 69 15 68 02; Email: patricia.uguen@ibaic.u-psud.fr
ABSTRACT
Using an in vitro system we have recently shown that the 3' ends of human pre-snRNAs synthesized by RNA polymerase II are produced by RNA processing directed by the snRNA gene-specific 3' box. Towards a complete characterization of this processing reaction we have further investigated the in vitro requirements for proper 3' end formation of pre-U1 snRNA. Here we show that the 5' cap plays a stimulatory role and processing requires creatine phosphate. Our results also indicate that the pre-U1 processing activity is heat sensitive and that an RNA component is required. In addition, the exact sequence adjacent to the 3' box influences the position of the pre-U1 3' end produced in vitro. Interestingly, the processing extract active for 3'-box-dependent processing also contains an activity that converts the 3' end of RNA containing the U1 Sm protein binding site and the 3' terminal stem–loop into the mature form.
INTRODUCTION
The majority of vertebrate U snRNA genes (e.g. U1–U5) are transcribed by RNA polymerase II (pol II) to yield short non-polyadenylated 3'-elongated pre-snRNAs. The 3' ends of these RNAs are produced by a processing reaction dependent on the 13–16 nucleotide (nt) cis-acting 3' box element, located 9–19 nt downstream of the 3' end of the RNA-encoding region (1). Most of the 3' extension is removed by further processing in the cytoplasm before nuclear re-import (2,3). A compatible snRNA promoter is also required for proper 3' end formation of pre-snRNAs (4,5), highlighting the tight link between transcription and the function of the 3' box. Relevant to this we have recently demonstrated that the C-terminal domain (CTD) of pol II is required for 3'-box-dependent RNA processing in vivo and that processing in vitro is activated by phospho-CTD (1,6).
3' End formation of pre-snRNAs is likely to involve divalent-cation-dependent endonucleolytic activity, in common with 3' end formation of vertebrate pre-mRNAs or yeast pre-snRNAs (1,7). However, neither subunits of the cleavage–polyadenylation complex factors cleavage/polyadenylation-specificity factor (CPSF), cleavage-stimulation factor (CstF) and cleavage factor I (CFI), nor the human homologues of yeast pre-snRNA processing factors hRNase III and PM-Scl100 appear to co-purify with the human pre-snRNA 3' processing activity (1). This suggests that there is little overlap between known RNA 3' end processing factors and those used for 3' end formation of human pre-snRNAs. Two cis-acting elements, the U2 3' box and the upstream sequence element (USE) act together to efficiently direct accurate pre-U2 snRNA 3' end formation both in vitro and in vivo (1), whereas 3' end formation of pre-U1 snRNA appears to depend largely on the 3' box. Thus, there may also be differences in the cis-acting sequences and trans-acting factors involved in 3' end formation of different pre-snRNAs.
Further investigation of the biochemical requirements of pre-U1 snRNA 3' end formation indicates that, in addition to the phospho-CTD of pol II, creatine phosphate and a 5' cap activate processing. We also demonstrate that 3' processing requires an RNA component and that the activity is heat labile. Thus, 3'-box-dependent processing has some characteristics in common with both AAUAAA-dependent processing (8) and 3' end formation of the replication-activated histone mRNAs (9) as well as significant differences. Furthermore, we show that the efficiency of in vitro pre-U1 snRNA 3' end formation is not affected by additional cis elements although the exact sequence around the cleavage site influences the site of the final 3' end formed in vitro. The S100 fraction active for 3' end formation of pre-U1 snRNA is also active for 3' end maturation and we show that these processes can occur independently in vitro.
MATERIALS AND METHODS
DNA constructs
In 10end U1 3' box, 17end U1 3' box and 59end U1 3' box templates, the 10, 17 and 59 nt, respectively, located upstream of the 3' box of U1 3' box template described by Uguen and Murphy (1) were replaced by the 10, 17 and 59 nt of the endogenous U1 snRNA gene (GenBank accession number J00318 ). The U1 3' box sequence, GTTTCAAAAGTAGA, is deleted from the 59end 3' box template. DNA constructs used to produce synthetic RNA were obtained by deleting the U1 promoter region by EcoRI/BglII digestion as described previously (1).
S100 fraction preparation
HeLa cell S100 extract was prepared using a slight modification of the procedure of Shapiro et al. (10). After cell lysis in buffer A (10 mM HEPES pH 7.9, 0.75 μM spermidine, 0.15 μM spermine, 10 mM KCl, 0.5 μM DTT) sucrose was added to a final concentration of 10% before spinning.
In vitro RNA synthesis and in vitro processing
Synthetic, capped RNA was produced as described previously (1). To produce uncapped RNA, m7G(5')ppp(5')G was omitted from the reaction. Processing was carried out as described in (1) for 2.5 h at 30°C in the presence of 3 mM MnCl2 unless stated otherwise in the figure legends. To reduce the level of endogenous ATP, the S100 fraction was bound to heparin–sepharose at 100 mM KCl and eluted by 1 M KCl. This fraction was then incubated for 15 min at 30°C with 3 mM MgCl2, 2 mM glucose and 0.6 U of hexokinase (Boehringer Mannheim) to deplete any residual ATP.
S1 mapping analysis
After incubation with S100, RNA was extracted, annealed to the S1 probe, digested and analysed on a 6% polyacrylamide–8 M urea gel as described previously (1). The S1 probes used for each construct (U1 3' box, 10end U1 3' box, 17end U1 3' box, 59end U1 3' box and 59end 3' box) were prepared as described previously (1).
Western blotting
Western blotting was carried out as described by Medlin et al. (6). Anti-stem–loop-binding protein (SLBP) antibody was kindly provided by Berndt Mueller (University of Aberdeen) and used diluted at 1/100.
RESULTS
Improving U1 3'-box-dependent in vitro processing
We recently described an in vitro system where 3'-box-dependent processing of substrate RNAs was detected for the first time (1). However, long incubation times (>5 h) are required to detect any accurate 3' end processing directed by the U1 3' box in unpurified S100. We therefore undertook to improve the efficiency of the processing activity by modifying the extract preparation procedure (see Materials and Methods) as a prelude to characterization of the requirements for 3' processing. Modification of the extract preparation method described by Shapiro et al. (10) led to the production of S100 that yields up to 8% properly processed RNA after 2.5 h incubation rather than the 13 h needed with S100 produced by our previously described protocol (1). The increased sucrose concentration used in production of this extract (see Materials and Methods) may reduce the release of nucleases from the nuclei upon spinning and consequently reduce turnover of substrate RNA, which proved to be a major problem for in vitro processing (1). Relevant to this, the amount of pol II and factors involved in mRNA 3' end formation is reduced in this S100 fraction (data not shown and see Fig. 4B).
Figure 4. U1 3'-box-dependent processing activity is heat labile. (A) U1 3' box RNA was incubated with S100 preheated for 10 min at the indicated temperature (temp.). The graph represents the percentage processing compared with the level obtained when S100 is preheated at 30°C. (B) Western blot with SLBP antibody on the S100 used for processing in Figures 1–5 (lane 1) and on the S100 described by Uguen and Murphy (1), which is less active for 3' processing (lane 2). SLBP is 45 kDa. (C) Western blot with anti-SLBP antibody on the different fractions resulting from purification of the processing activity on different sepharose resins as described by Uguen and Murphy (1). B-heparin and B-Q represent proteins bound to heparin– and Q-sepharose, respectively; UB-SP represents proteins that did not bind to SP-sepharose.
Efficient U1 3'-box-dependent processing requires creatine phosphate as a co-factor and is ATP independent
Since creatine phosphate (CP) (usually used to transfer Pi to ADP, creating ATP) improves the efficiency of some mRNA processing steps (11), we tested its effect on U1 3'-box-dependent processing (Fig. 1). Little processing (1%) is detected in the absence of added CP after incubation for 2.5 h at 30°C (Fig. 1A, lanes 2 and 4), whereas 7% of the input RNA is processed in the presence of 20 mM CP (Fig. 1A, lanes 3 and 5). Processing efficiency is only slightly improved by increasing the concentration of CP further (8% of input RNA is processed in the presence of 40 or 60 mM CP; Fig. 1A, lanes 6 and 7). Furthermore, nucleoside triphosphates, including ATP, do not support U1 3'-box-dependent processing in the absence of CP (Fig. 1A, lane 4, and B, lanes 3–6), indicating that CP does not act simply to regenerate the energy source. Serine phosphate (SP) can also activate processing although not as effectively as CP (Fig. 1C, lanes 3–5) suggesting that the effect of CP is not entirely due to the presence of phosphate groups.
Figure 1. Efficient U1 3'-box-dependent processing requires CP as a co-factor and is ATP independent. (A) U1 3' box RNA was incubated with S100 in the presence of ATP (1 mM) and different concentrations of CP as indicated. The percentage of input RNA processed (% proc.) is indicated below this and subsequent gels. The scheme represents the position and size of the S1 probe, input (IP) and processed (proc.) RNA. The diagram on the left side of this and subsequent figures indicates the position of the U1 3' box in the input RNA. (B) U1 3' box RNA was incubated with S100 in the presence of 20 mM CP or 1 mM of the nucleosides triphosphates indicated. (C) U1 3' box RNA was incubated with S100 in the presence of 20 mM CP or different concentrations of SP as indicated. (D) U1 3' box RNA was incubated with S100 in the presence or absence of 20 mM CP and in the presence of phospho-CTD as indicated. (E) U1 3' box RNA was incubated with S100 fractionated on heparin–sepharose in the presence of 20 mM CP, and with or without ATP, hexokinase and glucose. The processing reaction was carried out in the presence of 2 mM MgCl2.
It has been proposed that CP activates pre-mRNA 3' cleavage by mimicking the effect of the pol II CTD rather than affecting the reservoir of energy (11,12) and may facilitate a conformational change of some factors. In support of this notion, the phosphorylated form of the CTD of pol II activates 3'-box-dependent processing in vitro better than the unphosphorylated form (1). However, no processing is detected in the absence of CP when phospho-CTD is added (Fig. 1D). This indicates that CP does not simply substitute for phospho-CTD in our system.
To eliminate the ATP present in the S100 fraction, we fractionated the processing activity by binding to heparin–sepharose at 100 mM KCl and eluting by 1 M KCl. Although this fractionation procedure should markedly reduce the level of ATP in the active fraction it does not result in a requirement for added ATP (Fig. 1E, lanes 2 and 6). Furthermore, addition of both glucose and hexokinase (Fig. 1E, lane 5), which together deplete the available pool of ATP (11), has no greater effect than the addition of glucose or hexokinase alone (Fig. 1E, lanes 3 and 4). These results indicate that 3'-box-dependent processing is ATP independent. Interestingly, AAUAAA-dependent 3' cleavage is also CP dependent and ATP independent in vitro (11).
U1 3'-box-dependent processing is activated by a 7-methyl-G cap
To examine whether 5' cap binding proteins activate 3'-box-dependent RNA processing in addition to splicing, 3' processing, transport and translation of mRNA (13), we compared the efficiency of 3' processing of capped and uncapped U1 3' box RNA. In the absence of a 5' cap structure, U1 3'-box-dependent processing is slower (Fig. 2A), suggesting that either the 5' cap itself, or more likely, cap binding proteins present in the S100, activate processing. The decrease of proper 3' end formation of uncapped RNA is not due simply to RNA instability since input RNA is still detected after 3 h incubation (Fig. 2A). Addition of 7-methyl-G cap analogue to the in vitro reaction inhibits processing and the highest concentration of cap analogue tested effectively abolished processing (Fig. 2B, lanes 8–12). This suggests that the cap analogue can sequester cap binding proteins and interacting 3' processing factors. Consistent with this, the effect of the cap structure of the RNA on processing efficiency is lost after partial purification of the 3' processing activity as described by Uguen and Murphy (1) (data not shown), indicating that the 5' cap per se is not sufficient. Moreover, addition of unmethylated cap analogue G(5')ppp(5')G to the reaction has little effect on processing (Fig. 2B, lanes 3–7). Taken together, these results suggest that the nuclear cap binding complex (14) may be involved in the recruitment of pre-snRNA 3' processing factor(s) in vivo.
Figure 2. U1 3'-box-dependent processing is activated by a 7-methyl-G cap. (A) Capped (empty square) or uncapped (filled diamond) U1 3' box RNA was incubated with S100 for 0–3 h. The graph below represents the percentage relative processing activity calculated by comparison with the amount of processing obtained after incubating capped RNA for 3 h. (B) Capped U1 3' box RNA was incubated with S100 in the presence of different concentrations of unmethylated cap analogue (lanes 3–7) or methylated cap analogue (lanes 8–12) as indicated. The graph below represents the average percentage relative processing activity obtained in the presence of increasing concentrations of unmethylated (filled triangle) or methylated (empty square) cap analogue in two independent experiments.
U1 3'-box-dependent processing requires an RNA co-factor
Small non-coding RNAs function as co-factors in both splicing and 3' processing of pre-mRNAs (8,9). To determine whether 3' processing of U1 3' box RNA substrate is dependent on a trans-acting nucleic acid component, the S100 was treated with micrococcal nuclease (MN) before incubation with U1 3' box RNA (Fig. 3). U1 3'-box-dependent processing is completely abolished by this treatment (lane 3). However, when EGTA, which chelates the Ca2+ ions essential for this nuclease to function, is added to the S100 before MN, no reduction in processing is observed (lane 4). This suggests the involvement of an RNA or DNA co-factor in U1 3'-box-dependent processing. To distinguish between these two possibilities, we treated the S100 with pancreatic DNase I before incubation with U1 3' box RNA (lane 8). This has no effect on 3' processing, indicating that the nucleic acid co-factor is composed of RNA. In order to test whether this RNA component is the sole catalytic component of the 3' processing activity, we extracted ribonucleic acid from the S100 by acidic phenol/chloroform treatment and incubated it with U1 3' box RNA. This fraction does not support 3' processing on its own (lane 5). In addition, 3' processing is not recovered when the MN-treated S100 is complemented with this RNA fraction (lane 6). These data suggest that U1 3'-box-dependent processing involves both RNA and protein factor(s) and leads us to propose that processing is carried out by an RNA–protein complex.
Figure 3. U1 3'-box-dependent processing requires an RNA co-factor. U1 3' box RNA was incubated with S100 in the presence of 2 mM of MgCl2 before or after pretreatment with MN or DNase I and/or complemented with the extracted RNA. MN (lane 3), S100 pretreated for 15 min at 25°C with MN (5 U/μl of S100; Roche) in the presence of 1 mM CaCl2. The reaction was stopped by adding 2 mM EGTA. MN in. (lane 4), S100 pretreated for 15 min at 25°C with MN inactivated by the addition of 2 mM EGTA; RNA extracted (lane 5), S100 extracted by acidic phenol/chloroform treatment and ethanol precipitated before resuspension in buffer D (15); RNA extracted + MN (lane 6), S100 pretreated for 15 min at 25°C with MN in the presence of 1 mM CaCl2 and complemented by the extracted RNA. RNA extracted + MN in. (lane 7), S100 pretreated for 15 min at 25°C with MN inactivated by the presence of 2 mM EGTA and complemented by the extracted RNA; DNase I (lane 8), S100 pretreated for 15 min at 30°C with pancreatic DNase I (10 U/μl of S100).
U7 snRNA is required for 3' end formation of the mRNAs for replication-activated histones and has a region of complementarity to the substrate RNA (9). Since mammalian U7 snRNAs contain no conserved sequences complementary to 3' box sequences it is unlikely that U7 snRNA participates in 3' end formation of U1 snRNA. Unfortunately, experiments to test the effect of depletion of U7 snRNA and other U snRNAs from the processing extract using antibodies to the trimethylguanosine cap present at the 5' end were inconclusive.
U1 3'-box-dependent processing activity is heat labile
To analyse the sensitivity of the 3'-box-dependent processing activity to heat, the S100 fraction was heated for 10 min at temperatures from 30 to 55°C before incubation with U1 3' box RNA (Fig. 4). The efficiency of 3' processing is slightly reduced at 45°C (inhibition of 33%; Fig. 4A, lane 4). However, 3' processing is inhibited by 83% after preheating the S100 fraction at 50°C (lane 5) and is lost completely after treatment at 55°C (lane 6), suggesting that U1 3'-box-dependent processing involves heat-sensitive factor(s).
3' Processing of replication-activated histone mRNAs is similarly heat labile (16), raising the possibility of shared factors with U1 3'-box-dependent processing. A well-characterized SLBP is required for histone 3' end formation (9) and we have used an anti-SLBP antibody to analyse our processing extract by western blot (Fig. 4B). SLBP is not readily detected in the S100 used for the in vitro processing reactions shown in Figure 4A (Fig. 4B, lane 1) although it is relatively abundant in our previously described S100 preparation (1) (Fig. 4B, lane 2). In addition, no SLBP is detected in the partially purified 0.4 M KCl SP fraction active for pre-snRNA 3' end formation (1) (Fig. 4C, lane 6). These results suggest that SLBP is not involved in 3'-box-dependent processing.
Sequences upstream from the U1 3' box can influence 3' processing
We have shown that accurate pre-U2 snRNA 3' end formation requires two cis-acting elements, USE and U2, acting synergically. Although the U1 3' box is sufficient to direct accurate formation of RNA 3' ends in vitro and in vivo (1,4,17), the last 5–6 nt of the mature 3' end of U1 snRNA can affect the in vivo accuracy of pre-U1 snRNA 3' end formation (17). We therefore analysed the influence of the natural sequences preceding the U1 3' box on the efficiency and accuracy of 3' processing in vitro. When RNA containing the 10 nt just upstream of U1 3' box (10end U1 3' box RNA) is incubated with S100, no properly processed RNA is detected (Fig. 5A, lanes 6–10, and C), even after 3 h incubation. This may be due to instability of processed RNA rather than to a complete loss of processing. When the 17 nt upstream of the U1 3' box are present (17end U1 3' box), processing is detected (Fig. 5A, lanes 11–15, and C) although the efficiency is reduced by 45% with respect to U1 3' box RNA (Fig. 5A, lanes 1–5, and C). A small amount of the 3' end processed RNA maps to the same nucleotides seen with the U1 3' box RNA but most map a few nucleotides upstream, corresponding to the most highly represented 3' ends detected in vivo, which end 6–8 nt upstream of the 3' box (17,18) (Fig. 5A, compare lanes 1–5 and 11–15, and C). These results may indicate that the addition of 17 nt of the natural U1 sequence influences the position of 3' endonucleolytic cleavage. Alternatively, 3' exonucleolytic trimming of the RNA occurs after cleavage 3–5 nt upstream from the 3' box.
Figure 5. Sequences upstream from the U1 3' box can influence 3' processing. (A) U1 3' box, 10end U1 3' box and 17end U1 3' box RNA substrates were incubated with S100 for 0–3 h. The scheme below indicates the position and size of the S1 probes, input (IP) and processed (proc.) RNA. (B) U1 3' box, 59end U1 3' box and 59end 3' box RNA substrates were incubated with S100. The scheme below indicates the position and size of the S1 probes, input (IP), processed (proc.) and mature RNA. (C) The sequences of the 3' ends of the substrate RNAs in (A) and (B) are shown with the position of in vivo and in vitro processing sites and mature ends. The endogenous U1 snRNA sequence is highlighted in grey. The U1 3' box is boxed. The vertical line indicates the boundary between the end of the mature snRNA and the first nucleotide of the non-coding RNA. The bold nucleotides represent the processing sites observed in vivo (17,18). The arrows above the sequence indicate the pre-snRNA processing sites observed in vitro. The mature ends observed in vitro are indicated by unfilled arrows above the sequence.
We have also tested the effect of placing a longer endogenous sequence (59 nt) upstream of the U1 3' box. This substrate RNA contains both the Sm protein binding site and the 3' terminal stem–loop, which are not required in vivo for pre-U1 snRNA 3' end formation but are required for efficient nuclear export (19). When 59end U1 3' box RNA (Fig. 5B, lanes 3 and 4) is incubated with S100, RNA with the mature 3' end mapping 10–12 nt upstream of the 3' box (17) is produced. This indicates that the activities required for production of mature U1 3' ends (20) are present in the S100, in addition to factors required for pre-U1 3'-box-dependent processing. Some accurately processed pre-U1 RNA 3' ends are still detected and these map to the same sites as are observed when only 17 nt of natural U1 sequence are present upstream of the 3' box (17end U1 3' box RNA; Fig. 5A, lanes 11–15, and C). This may suggest that processing proceeds in a stepwise manner as it does in vivo and the precise pre-snRNA 3' ends are required for the next step, which matures the 3' end of the RNA, as has been shown for 3' end formation of U2 snRNA (21). 3' Ends mapping 6–8 nt upstream from the 3' box are no longer detected when the 3' box is removed (Fig. 5B, lanes 5 and 6, and C). However, production of the mature U1 3' ends is unaffected (Fig. 5B, lane 6), indicating that these two processing events occur independently in our in vitro system. This suggests that any RNAs containing the Sm protein binding site and the 3' terminal stem–loop can be 3' end matured, presumably by exonucleolytic trimming. Since the 59end 3'box RNA has 27–29 nt downstream from the sites of the mature 3' ends, these results also indicate that RNAs longer than pre-U1 at the 3' end are readily 3' matured in vitro. In contrast, only pre-U1 snRNAs with a short 3' extension (no more than 10 nt) are further processed after injection into Xenopus oocytes (20). This may highlight differences between species in the 3' end maturation of snRNAs. Alternatively, the exact sequence of the 3' extension may influence trimming and/or some in vivo constraints may be lost in our in vitro system.
DISCUSSION
The first step of expression of the vertebrate U1 snRNA genes involves the production of a 3' extended pre-U1 snRNA, which occurs by a co-transcriptional 3'-box-dependent mechanism (4,5). Proper 3' end formation of pre-U1 snRNA is important for the stability of the snRNA in the nucleus, in addition to nucleocytoplasmic export (19,20). The exact 3' ends of U1 snRNA precursors can also influence the cytoplasmic 3' trimming of the RNA required for the nuclear translocation of U1 snRNPs and subsequent maturation steps (20). Towards the dissection of the molecular mechanism of 3' end formation of human pre-U1 snRNA we have determined several of the co-factor requirements of U1 3'-box-dependent in vitro processing. We have also investigated the influence of sequences upstream of the U1 3' box on 3' end formation. The results of these studies highlight similarities and differences between 3' end formation of pre-snRNAs and other vertebrate pol II transcripts. In addition, we show that 3' extended RNAs containing the U1 Sm protein binding site and the 3' terminal stem–loop can be processed to the mature U1 3' end in vitro.
Comparing U1 3'-box-dependent processing and formation of mRNA 3' ends
Some co-factor requirements for 3'-box-dependent processing are shared with AAUAAA-dependent processing. Both processing events require divalent cations and CP, are ATP independent and are activated by 5' cap binding proteins (8,11,13) (Fig. 6). In addition, both require the CTD of pol II in vivo (6,13) and are activated by phospho-CTD in vitro (1,12) (Fig. 6). So far there is no indication that any subunits of CPSF, CstF and CFI factors required for AAUAAA-dependent processing participate in pre-snRNA 3' end formation (1). However, it has been shown that CFI participates in 3' processing of some yeast snoRNAs and snRNAs (22).
Figure 6. Requirements for U1 3'-box-dependent RNA processing. 3'-Box-dependent 3' end formation of pre-U1 snRNA requires an RNA–protein complex, CP and divalent cations (div. ion). The reaction is also activated by 5' cap binding proteins (diamond) and the CTD of pol II. In addition, the processing activity is abolished by relatively mild heat treatment, which suggests the involvement of a heat-labile factor (HLF).
There are also some similarities between 3' end formation of U1 snRNA and replication-activated histone mRNAs. In both cases a stem–loop structure is present just upstream of the mature 3' end, the processing activities are heat labile and a trans-acting RNA component is required (9). Our results therefore point to mechanistic similarities in 3' end formation of non-polyadenylated histone mRNA and snRNA. However, our data indicate that the SLBP is not involved in U1 snRNA 3' end formation. In addition, the lack of complementarity between U7 snRNA and 3' box sequences makes it likely that a distinct RNA is required.
Divalent cations are also necessary for efficient activity of the yeast RNase III homologues, Pac1p and Rnt1p (23,24) which are involved in yeast snRNA 3' end formation (25). However, human RNase III does not co-purify with the 3'-box-dependent processing activity and the cis-acting elements do not correspond to RNase III recognition sequences (1). Divalent cations may be required for the function of a 3'-box-specific processing factor or play a more direct catalytic role.
The currently known requirements for 3' end processing of pre-U1 snRNA are summarized in Figure 6. Taken together our results indicate that there is little overlap in the factors required for 3'-box-dependent processing and mRNA 3' end formation apart from the shared co-factors noted above. This is consistent with the differences in cis-acting sequences and the differential promoter dependence of these signals in vivo (4,5).
Sequences immediately upstream of the U1 3' box influence 3' end formation
The efficient production of accurately 3' processed pre-U2 snRNA requires two distinct sequences: the U2 3' box and the USE, which contains a ‘minihelix’ stem–loop structure (1). The U2 3' box alone directs formation of RNA a few nucleotides longer than cellular pre-U2 snRNA while RNA containing the USE is processed to the authentic pre-U2 snRNA 3' end. In addition, the amount of processed product detected in vitro is increased when both the U2 3' box and the USE are present. The minihelix may therefore influence processing by interacting with a processing factor and/or prevent RNA cleaved by a 3'-box-dependent endonuclease from being rapidly turned over by exonucleases (1). Addition of 17 nt of the natural U1 gene-encoded sequence upstream of the U1 3' box also causes an apparent shift in the position of the final 3' ends, some of which now map to the most abundant pre-U1 3' ends detected in cellular RNA (17,18). The sequences directly upstream from the U1 3' box may therefore play a similar role to the U2 USE to determine the final position of the pre-U1 3' end. In both cases the site of endonucleolytic cleavage may be influenced by the exact sequence of the target RNA or changes in processing–factor RNA interaction. Alternatively, the natural sequences may be preferential substrates for limited trimming by exonucleases following 3' box-directed cleavage. However, unlike the U2 USE, the 17 nt upstream from the U1 3' box do not cause an increase in the amount of processed RNA detected but rather a decrease. In addition, unlike the U2 USE (1), the 59 nt of U1 sequence upstream of the 3' box do not direct formation of pre-U1 3' ends, suggesting functional differences. In agreement with this, little authentic pre-U1 RNA is produced from transfected U1 templates lacking the 3' box (17).
A precise pre-U1 3' end is not required for RNA maturation in vitro
Although pre-U2 snRNA can be further 3' processed in vitro (2,21), maturation of the 3' ends of pre-U1 snRNA in vitro has not previously been reported. However, in our in vitro system the inclusion of U1 snRNA sequences containing the Sm protein binding site and the 3' terminal stem–loop (19) in the RNA substrate results in the production of properly processed mature U1 3' ends. Interaction of Sm protein with its binding site is required for the cytoplasmic 5' cap modification of U1 snRNA, which is necessary for nuclear re-import (26) and it also plays a role in maturation of the 3' ends of yeast U1 snRNA (27, reviewed in 3,28). Surprisingly, in vitro 3' end maturation occurs independently of 3'-box-dependent processing, suggesting that no intermediate processing step is required even though the substrate U1 RNAs have relatively long 3' extensions (27–29 nt). Instead, 3' end extensions longer than 10 nt are not processed to the mature form when human pre-U1 RNAs are injected into Xenopus oocytes (20). In addition, specialized sequences distinct from the Sm protein binding site are required for cytoplasmic trimming of the 3' end of human U2 snRNA (29), and pre-U2 RNA with the natural 3' ends are preferential substrates for the reaction in vitro (2,21). There may therefore be differences in 3' end maturation of Xenopus and human U1 snRNAs, and between human U1 and U2 snRNAs. The availability of in vitro systems for both 3'-box-dependent processing and maturation of 3' extended pre-U1 snRNAs will facilitate the molecular analysis of all steps leading to accurate formation of U1 3' ends.
ACKNOWLEDGEMENTS
We thank Alice Taylor for excellent technical assistance. We are grateful to Berndt Mueller for the kind gift of the anti-SLBP antibody. We also thank Zbigniew Dominski, Nick Proudfoot and Andre Furger for helpful suggestions and comments on the manuscript. P.U. was supported by MRC Co-operative Component grant No. G9900343 and the Edward Penley Abraham Trust. S.M. was supported by MRC Senior Fellowship No. G117/309.
REFERENCES
Uguen,P. and Murphy,S. ( (2003) ) The 3' ends of human pre-snRNAs are produced by RNA polymerase II CTD-dependent RNA processing. EMBO J., , 22, , 4544–4554.
Huang,Q., Jacobson,M.R. and Pederson,T. ( (1997) ) 3' Processing of human pre-U2 small nuclear RNA: a base-pairing interaction between the 3' extension of the precursor and an internal region. Mol. Cell. Biol., , 17, , 7178–7185.
Will,C.L. and Lührmann,R. ( (2001) ) Spliceosomal UsnRNP biogenesis, structure and function. Curr. Opin. Cell Biol., , 13, , 290–301.
Hernandez,N. and Weiner,A.M. ( (1986) ) Formation of the 3' end of U1 snRNA requires compatible snRNA promoter elements. Cell, , 47, , 249–258.
Neuman de Vegvar,H.E., Lund,E. and Dahlberg,J.E. ( (1986) ) 3' End formation of U1 snRNA precursors is coupled to transcription from snRNA promoters. Cell, , 47, , 259–266.
Medlin,J.E., Uguen,P., Taylor,A., Bentley,D. and Murphy,S. ( (2003) ) The C-terminal domain of pol II and a DRB-sensitive kinase are required for 3' processing of U2 snRNA. EMBO J., , 22, , 925–934.
Ryan,K., Calvo,O. and Manley,J.L. ( (2004) ) Evidence that polyadenylation factor CPSF-73 is the mRNA 3' processing endonuclease. RNA, , 10, , 565–573.
Wahle,E. and Rüegsegger,U. ( (1999) ) 3'-End processing of pre-mRNA in eukaryotes. FEMS Microbiol. Rev., , 23, , 277–295.
Dominski,Z. and Marzluff,W.F. ( (1999) ) Formation of the 3' end of histone mRNA. Gene, , 239, , 1–14.
Shapiro,D.J., Sharp,P.A., Wahli,W.W and Keller,M.J. ( (1988) ) A high-efficiency HeLa cell nuclear transcription extract. DNA, , 7, , 47–55.
Hirose,Y. and Manley,J.L. ( (1997) ) Creatine phosphate, not ATP, is required for 3' end cleavage of mammalian pre-mRNA in vitro. J. Biol. Chem., , 272, , 29636–29642.
Hirose,Y. and Manley,J.L. ( (1998) ) RNA polymerase II is an essential mRNA polyadenylation factor. Nature, , 395, , 93–96.
McCracken,S., Fong,N., Rosonina,E., Yankulov,K., Brothers,G., Siderovski,D., Hessel,A., Foster,S., Shuman,S. and Bentley,D.L. ( (1997) ) 5'-Capping enzymes are targeted to pre-mRNA by binding to the phosphorylated carboxy-terminal domain of RNA polymerase II. Genes Dev., , 11, , 3306–3318.
Flaherty,S.M., Fortes,P., Izaurralde,E., Mattaj,I.W. and Gilmartin,G.M. ( (1997) ) Participation of the nuclear cap binding complex in pre-mRNA 3' processing. Proc. Natl Acad. Sci. USA, , 94, , 11893–11898.
Dignam,J.D., Lebovitz,R.M. and Roeder,R.G. ( (1983) ) Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res., , 11, , 1475–1489.
Gick,O., Kramer,A., Vasserot,A. and Birnstiel,M.L. ( (1987) ) Generation of histone mRNA 3' ends by endonucleolytic cleavage of the pre-mRNA in a snRNP-dependent in vitro reaction. Proc. Natl Acad. Sci. USA, , 84, , 8937–8940.
Hernandez,N. ( (1985) ) Formation of the 3' end of U1 snRNA is directed by a conserved sequence located downstream of the coding region. EMBO J., , 4, , 1827–1837.
Patton,J. and Wieben,E. ( (1987) ) U1 precursors: variant 3' flanking sequences are transcribed in human cells. J. Cell Biol., , 104, , 175–182.
Terns,M.P., Dahlberg,J.E. and Lund,E. ( (1993) ) Multiple cis-acting signals for export of pre-U1 snRNA from the nucleus. Genes Dev., , 7, , 1898–1908.
Neuman de Vegvar,H.E. and Dahlberg,J.E. ( (1990) ) Nucleocytoplasmic transport and processing of small nuclear RNA precursors. Mol. Cell. Biol., , 10, , 3365–3375.
Huang,Q. and Pederson,T. ( (1999) ) A human U2 RNA mutant stalled in 3' end processing is impaired in nuclear import. Nucleic Acids Res., , 15, , 1025–1031.
Morlando,M., Greco,P., Dichtl, B., Fatica,A., Keller,W. and Bozzoni,I. ( (2002) ) Functional analysis of yeast snoRNA and snRNA 3'-end formation mediated by uncoupling of cleavage and polyadenylation. Mol. Cell. Biol., , 22, , 1379–1389.
Rotondo,G. and Frendewey,D. ( (1996) ) Purification and characterization of the Pac1 ribonuclease of Schizosaccharomyces pombe. Nucleic Acids Res., , 24, , 2377–2386.
Lamontagne,B. and Elela,S.A. ( (2001) ) Purification and characterization of Schizosaccharomyces pombe Rnt1p nuclease. Methods Enzymol., , 342, , 159–167.
Allmang,C., Kufel,J., Chanfreau,G., Mitchell,P., Petfalski,E. and Tollervey,D. ( (1999) ) Functions of the exosome in rRNA, snoRNA and snRNA synthesis. EMBO J., , 18, , 5399–5410.
Mattaj,I.W. ( (1986) ) Cap trimethylation of U snRNA is cytoplasmic and dependent on U snRNP protein binding. Cell, , 46, , 905–911.
Seipelt,R.L., Zheng,B., Asuru,A. and Rymond,B.C. ( (1999) ) U1 snRNA is cleaved by RNase III and processed through an Sm site-dependent pathway. Nucleic Acids Res., , 27, , 587–595.
Yong,J., Pellizzoni,L. and Dreyfuss,G. ( (2002) ) Sequence-specific interaction of U1 snRNA with the SMN complex. EMBO J., , 21, , 1188–1196.
Jacobson,M.R., Rhoadhouse,M. and Pederson,T. ( (1993) ) U2 small nuclear RNA 3' end formation is directed by a critical internal structure distinct from the processing site. Mol. Cell. Biol., , 13, , 1119–1129.(Patricia Uguen and Shona Murphy*)