ABSTRACT
The Cbp2 protein is encoded in the nucleus and is required for the splicing of the terminal intron of the mitochondrial COB gene in Saccharomyces cerevisiae. Using a yeast strain that lacks this intron but contains a related group I intron in the precursor of the large ribosomal RNA, we have determined that Cbp2 protein is also required for the normal accumulation of 21S ribosomal RNA in vivo. Such strains bearing a deletion of the CBP2 gene adapt slowly to growth in glycerol/ethanol media implying a defect in derepression. At physiologic concentrations of magnesium, Cbp2 stimulates the splicing of the ribosomal RNA intron in vitro. Nevertheless, Cbp2 is not essential for splicing of this intron in mitochondria nor is it required in vitro at magnesium concentrations >5 mM. A similar intron exists in the large ribosomal RNA (LSU) gene of Saccharomyces douglasii. This intron does need Cbp2 for catalytic activity in physiologic magnesium. Similarities between the LSU introns and COB intron 5 suggest that Cbp2 may recognize conserved elements of the these two introns, and protein-induced UV crosslinks occur in similar sites in the substrate and catalytic domains of the RNA precursors.
Group I introns serve as models for understanding large catalytic RNA molecules. The group I intron in the large ribosomal RNA of Tetrahymena thermophila was the first RNA shown to have catalytic activity, and many other introns of this class are self-splicing in vitro. Ribozymes based on the Tetrahymena intron can catalyze a variety of reactions with other RNA molecules, including cleavage, ligation, polymerization and even the cleavage of an amide bond between ribonucleotides (1 -5 ). Nevertheless, it is widely believed that the splicing of group I introns is stimulated by proteins in the cell (6 -8 ). In fungal mitochondria, where genetic insight has been applied, genes for splicing proteins were discovered within both group I and II introns (9 -13 ). Such proteins are termed maturases. In addition, proteins encoded in the nucleus and imported into mitochondria are required for the splicing of specific introns (14 -17 ). The most closely scrutinized of these proteins is called Cyt18. Cyt18 is the mitochondrial tyrosyl tRNA synthetase in Neurospora crassa, but is also essential for the splicing of several mitochondrial introns (18 ,19 ). In addition to specific splicing factors, Coetzee et al. (8 ) have shown that a variety of RNA-binding proteins, including the ribosomal protein S12 of Escherichia coli, stimulate the splicing of several group I introns, presumably by stimulating the appropriate folding of the RNA.
The mitochondrial gene COB, which encodes apocytochrome b, has as many as five introns in Saccharomyces cerevisiae. The terminal intron, designated bI5, is a group I intron that splices autocatalytically in the presence of high concentrations of magnesium and monovalent salt (20 -22 ). In yeast cells, removal of this intron requires the product of the nuclear gene CBP2. The Cbp2 protein (hereinafter called Cbp2) comprises 630 amino acids and is rich in basic and aromatic amino acids but contains none of the common RNA-binding motifs (23 ). Gampel et al. (24 ) showed that Cbp2 stimulates the splicing of transcripts containing bI5 in vitro under conditions in which no autocatalytic splicing was observed. Gampel and Cech (25 ) demonstrated that Cbp2 binds the catalytic core of intron 5. Detailed analysis by Weeks and Cech (26 -28 ) and by us (29 ) have recently shown that binding of Cbp2 stabilizes tertiary interactions necessary for the catalytic activity of this ribozyme. Cbp2 can suppress mutations in conserved helices, even though these mutations interfere with autocatalytic splicing (30 ). We have also demonstrated that Cbp2 can stimulate splicing during transcription, presumably by directing the folding of the nascent RNA transcript (31 ).
McGraw and Tzagoloff (23 ) originally cloned CBP2 by complementation of a nuclear mutation affecting the splicing of the COB transcript. Cbp2 mutants are deficient in respiration and accumulate transcripts retaining the terminal intron. (This was intron 2 in their strain but is the fifth intron in many laboratory strains of S.cerevisiae.) Later work by Hill et al. (32 ) showed that manganese induced revertants of the original cbp2 mutant were caused by a precise deletion of the intron encoding sequences from the mitochondrial DNA. These revertants could grow on glycerol plates (were respiring) and showed no defect in processing of COB RNA. This result implied that the Cbp2 protein was involved in the splicing of only bI5.
The last intron of COB belongs to a subgroup designated Ia. These are characterized by lack of the extensive P5a,b,c domain present in the Tetrahymena rRNA intron and by the presence of an extra stem-loop structure called P7.1a,b or P7.1/P7.2 (33 ). Yeast mitochondria have another group Ia intron in the precursor to the large ribosomal RNA. This intron, known as [omega] or as the ScLSU intron, was originally described as a mobile genetic element in yeast (34 ,35 ). The intron encodes a double-stranded endonuclease essential for its transposition at the DNA level but not involved in splicing. In this study, we show that Cbp2 stimulates the splicing of [omega], but is not essential for its processing either in mitochondria or in vitro.
The yeast strains used in this study are described in Table 1 . D273-10B/G1 and DM445-R1 are isonuclear but differ in mitochondrial genotype. D273-10B contains introns 4 and 5 of COB, introns 1-4 and 5[gamma] of COX1 and the LSU intron, [omega]. DM445-R1, contains the LSU intron, intron 4 of COB and introns 1-4 and 5[gamma] of COX1. DM445-R1 was constructed by Dr G. Faye by methods described in Séraphin et al. (36 ). GF157-4B contains neither [omega] nor bI5, but retains aI1, aI3 and aI5[gamma]. These strains were converted to uracil auxotrophs by transformation with the ura3-52 allele (37 ). YAL7 contains bI5, but not [omega]. This strain was derived by cytoduction (38 ) in a cross between GF110-8C and MC109. Disruptions of the CBP2 gene were made by inserting a HindIII fragment containing URA3 at the HindIII site internal to CBP2. Linear restriction fragments containing this construct were used to transform strains to uracil prototropy. Disruption of the chromosomal allele of CBP2 was confirmed by Southern hybridization (data not shown). Yeast were grown on SD media (0.67% yeast nitrogen base, 1% glucose) or YP media (1% w/v yeast extract, 1% w/v Bacto-peptone) containing 2% glucose (YPD), 2% galactose (YPGal) or 2% ethanol and 2% glycerol (YPGE). Culture densities were measured by absorbance at 600 nm.
RNA was extracted from whole yeast cells using hot phenol/chloroform/isoamyl alcohol according to the method of Schmitt et al. (39 ). Cells were grown to late logarithmic growth phase on YPGal medium. RNA was fractionated on 1.2% agarose formaldehyde gels and transferred to nylon membranes (Hybond Plus, Amersham) by capillary blotting. Hybridization was carried out overnight at 60oC in 0.9 M NaCl, 90 mM sodium citrate, 1 mM sodium pyrophosphate, 0.1% SDS, 1 [mu]g/ml tRNA and Denhardt's solution (0.02% Ficoll, 0.02% polyvinylpyrrolidone and 0.02% bovine serum albumin). The COB probe was a 32P-labeled EcoRI-BamHI fragment of 1.2 kb that annealed to exons 4 and 5, introns 4 and 5 and downstream sequences. The LSU probe was a 1.7 kb EcoRI-SalI fragment that included the [omega] intron and flanking sequences. The [omega]-specific probe was an end-labeled oligonucleotide complementary to positions 209-228 of the intron. Compared lanes were loaded equivalently based on A260 measurements and staining of the cytoplasmic ribosomal RNAs. Equal loading and transfer was confirmed by hybridization with an 18S rRNA probe (Fig. 3 C). Blots were exposed to X-ray film and quantitated using a Molecular Dynamics Phosphorimager. Analysis of a series of radioactive dots exposed simultaneously indicated that our exposures were within the linear range of the instrument.
Total DNA was extracted from S.douglasii (ATCC 76859) cells and PCR was used to amplify the intron and portions of the flanking exons similar to those present in our clone of [omega]. Homology to the 5' primer began 365 nt upstream from the 5' splice junction. Homology to the downstream primer extended to 55 nt past the 3' splice junction. EcoRI and BamHI restriction sites were added to the 5' and 3' ends during amplification, and the resulting EcoRI-BamHI fragment was cloned in pT7/T3-18 to produce pLS2545.
Unless otherwise indicated, transcription reactions were performed according to Grodberg and Dunn (40 ) using T7 RNA polymerase prepared as described by those authors. The template for synthesis of bI5 and flanking exons was pSPI5 cut with SmaI (21 ). The ScLSU intron was transcribed from pT7[omega], which contained 1.7 kb of mitochondrial DNA including the entire 1143 nt intron, 501 nt of upstream exon plus 31 nt of vector sequence and 55 nt of downstream exon. This construct was made linear with SalI prior to transcription. The SdLSU intron was transcribed from pLS2525 that had been linearized with BamHI. RNA was either unlabeled or was labeled with [[alpha]-32P]UTP (ICN) during synthesis. RNA was purified from unincorporated nucleotides by chromatography on G-50 Sephadex (Pharmacia). Splicing reactions were typically conducted for 45 min at 37oC using buffer and salt conditions indicated in the text. Reactions typically contained Cbp2 at 0.9 [mu]M and RNA at 7 [mu]M. Reactions with labeled transcripts included 200 [mu]M GTP, while guanosine labeling experiments employed 10 [mu]Ci [[alpha]-32P]GTP. Products of splicing were displayed on 4% acrylamide-8 M urea gels run as described by Partono and Lewin (21 ), and products were detected by autoradiography.
UV crosslinking of the [omega] transcript in the presence and absence of Cbp2 and reverse transcription reactions were performed as described in Shaw and Lewin (29 ). All crosslink sites were documented in triplicate.
Cbp2 was prepared according to our published procedure (29 ). Reagents for electrophoresis including urea were purchased from Schwartz/Mann. Radioisotopes and yeast lytic enzyme were bought from ICN. Nucleotides and Sequenase were obtained from US Biochemicals. Restriction enzymes were purchased from New England Biolabs. Taq DNA polymerase was obtained from Boehringer-Mannheim. Dehydrated culture media were purchased from Difco. Other reagents were purchased from Fisher Chemical Company or Sigma Chemical Company.
We employed four strains to monitor the requirement of the Cbp2 protein for splicing of [omega] in vivo. Strain DM445-R1 contains the [omega] (ScLSU) intron but lacks intron 5 of COB. It has the same nuclear background as D273-10B, the strain used to identify the original cbp2 mutations (23 ). D273-10B contains both [omega] and bI5. We also tested a strain that contained bI5 and not [omega] (YAL7) and a strain containing neither of these introns (GF157-4B). The CBP2 gene was disrupted in all strains using the wild-type URA3 allele. Disruption of CBP2 in D273-10B (bI5, [omega]) prevented growth on glycerol/ethanol plates (Fig. 1 A). This is the result expected from the work of McGraw and Tzagoloff (23 ). In contrast, disruption of CBP2 in the strain that retained [omega] but not bI5 (DM445-R1) resulted in cells that could grow on this medium, though at a reduced rate relative to same strain with CBP2 intact (Fig. 1 B). Disruption of CBP2 blocked glycerol growth of YAL7 (bI5, [omega]) (Fig. 1 C), but had no impact on the growth of GF157-4B (bI5, [omega]) on non-fermentable carbon sources (Fig. 1 D). Séraphin et al. (36 ) showed that a cbp2 mutation had no effect on glycerol growth in a strain lacking both introns.
The reduced growth of DM445-R1 on YPGE in the absence of CBP2 was the result of a prolonged lag phase after shifting from medium with a fermentable carbon source (SD) to glycerol/ethanol medium (Fig. 2 A). Though the exponential growth rate was nearly identical in the presence and absence of CBP2, the onset of exponential growth was delayed by 25 h in the CBP2 disruptant. Once cells had adapted to growth on a non-fermentable carbon source, i.e. in a subculture from cells grown in YPGE, this delay was not observed (data not shown). Likewise, the two strains grew identically on glucose media regardless of the CBP2 disruption (Fig. 2 B).ab
The LSU intron splices autocatalytically in considerably lower salt conditions than bI5 (21 ,41 ). We assayed the splicing of the LSU precursor in 25 mM Tris-HCl pH 7.5, 25 mM NH4Cl, 1 mM MgCl2, by labeling with [[alpha]-32P]GTP to detect addition of guanosine to the 5' ends of intron products resulting from splicing (Fig. 4 ). Under these conditions, Cbp2 was required for the 5' cleavage reaction (lanes 1 and 2). Incubation in buffer containing 5 mM MgCl2 and 50 mM NH4Cl resulted in much greater labeling of the intron (lane 3). This result was compared with the effect of Cbp2 on bI5: incubation in the 5 mM MgCl2, 50 mM NH4Cl buffer did not result in splicing (lane 4), addition of Cbp2 led to labeling of the COB intron in that buffer (lane 5), while incubation in high salt buffer (50 mM MgCl2, 0.5 M KCl) resulted in labeling of intron products and cleavage of the intron at an internal cyclization site (lane 6) (21 ,22 ). Since the amount of unlabeled RNA present in the [omega] and bI5 reactions was the same, we conclude that under magnesium limiting conditions, Cbp2 is required for the splicing of [omega].
In the same buffer containing 2 mM MgCl2, the [omega] intron was reactive for a short time without Cbp2, but the reaction was soon attenuated. This behavior could be followed either by monitoring the guanosine labeling of the intron (Fig. 5 A) or by measuring the processing of the 32P-labeled precursor RNA (Fig. 5 B). In either case, the initial rate of splicing (determined from the level of intron product accumulated in the first 2.5 min) was identical with or without the protein, but splicing nearly ceased after 5 min in its absence. After the initial burst of splicing in the absence of the protein, the accumulation of guanosine labeled products was >6-fold higher in the presence of the protein, as determined from the accumulation of guanosine labeled intron over the next 10 min. This result implies that a fraction of the precursor RNA had attained the splicing competent tertiary structure in the absence of Cbp2 but that most of the precursor was misfolded. The fact that the reaction continued at a constant rate in the presence of protein suggests that Cbp2 could reorganize the RNA structure during the course of the reaction (29 ) perhaps by capturing transient intermediates (28 ).
If the transcript containing [omega] intron and flanking exons was prepared under the usual conditions for T7 RNA polymerase, which contained 8 mM MgCl2 and 4 mM spermidine, then splicing occurred during transcription (Fig. 6 , lane 4). Splicing of [omega] also occurred during a transcription reaction catalyzed by SP6 RNA polymerase (41 ). Lowering the magnesium and spermidine to 3 and 0.4 mM respectively, prevented co-transcriptional splicing (Fig. 6 , lane 1), but adding Cbp2 during transcription under these conditions stimulated splicing (lane 2). Co-transcriptional splicing required native Cbp2, since heat denatured protein did not stimulate splicing (lane 3). We have demonstrated elsewhere that adding Cbp2 facilitated co-transcriptional splicing of precursors containing bI5 (31 ). This experiment suggests that Cbp2 also assists the folding of [omega] during the transcription reaction.
Li et al. (42 ) have demonstrated that the CBP2 gene is required for the growth of S.douglasii despite the fact that this yeast does not contain bI5. Since S.douglasii does contain an LSU intron nearly identical to that in S.cerevisiae, we tested whether the protein was required for splicing of this intron. At physiologic ionic strength (50 mM NH4Cl, 50 mM Tris-HCl pH 7.5) Cbp2 stimulated splicing of SdLSU beginning at 3 mM MgCl2 (Fig. 7 ). In these salt and buffer conditions splicing was not observed in the absence of Cbp2 even at elevated MgCl2 (data not shown). In the presence of 0.5 M KCl, autocatalytic splicing of SdLSU was observed beginning at 20 mM MgCl2 (Fig. 7 ).
Cbp2 showed no effect on the splicing of two other group Ia introns, the td intron from bacteriophage T4 and the LSU intron from the mitochondria of Aspergillus nidulans, whether added during or after transcription. The Aspergillus LSU intron resembles the [omega] intron and bI5 in that important joining regions (J3/4 and J6/7) and other elements in the catalytic core are identical or nearly so. Like the yeast mitochondrial introns, the Aspergillus intron has the capacity to form a P11 pairing between loops designated L6a and L7.1a (43 ). Even though the LSU intron from Aspergillus, like [omega], has the capacity to form a seven membered P1 stem containing three G-C pairs, it did not splice even in the presence of high concentrations of magnesium and monovalent salt and showed no response to Cbp2 (data not shown). In contrast, the td intron, which differs more in sequence, was active even in low magnesium (1 mM), and splicing was not stimulated by Cbp2 (data not shown). Similarly, the group Ib intron Nox1C spliced autocatalytically in 50 mM MgCl2, 0.4 M KCl, but Cbp2 did not stimulate splicing either in this buffer or in lower salt incubations (data not shown).
In our earlier work on bI5 (29 ), we employed UV induced crosslinking of the RNA to map potential sites of interaction between the protein and the intron. By comparing sites of crosslinks in low salt with and without Cbp2, we were able to identify crosslink sites that occurred only in the presence of the protein. These were mapped as strong blocks to reverse transcription, using intron-specific DNA oligonucleotides as primers. As we noted previously, blocks to reverse transcription may result from protein-RNA crosslinks, but may also result from RNA-RNA crosslinks that depend on the presence of Cbp2. To help identify the latter class, we have compared Cbp2-dependent crosslinks with those that occur in autocatalytic conditions (in which case the RNA has attained its functional conformation). Note that UV-induced blocks to reverse transcription also occur at reactive nucleotides without Cbp2 (Fig. 8 , lanes 2 and 6). We are scoring the enhancement of these stops at specific sites as the Cbp2-dependent crosslinks.
The central core (P3, P4, P6, P7 and joining segments) of [omega] and bI5 are identical in sequence with the exception of an extra uridine in the P4 stem of bI5 (Fig. 9 ). There is also substantial identity in non-conserved structures such as L5, P6a, P9, P9.0 and P9.1. The major differences lie in P1, L1, P2 and in L8. The latter structure contains the coding region for SceI, a double-stranded exonuclease required for the transposition of [omega]. In bI5, this structure is 152 nt and encodes no protein. As mentioned above, P1 of [omega] is relatively stable, containing 7 bp, including three contiguous G-C pairs. The P1 stem of bI5 is less stable by itself but may be stabilized by the extensive (221 nt) L1 structure that is substantially helical (29 ).
Experiments by Weeks and Cech (26 ) and by us (29 ) determined that Cbp2 binds one face of the catalytic core composed of the helical stacks P5-P4-P6 and P8-P3-P7 in bI5. This binding appears to nest the P1-P2 substrate domain between the other two domains. The similarity of the two introns suggests that Cbp2 might recognize similar sites in [omega]. Our UV crosslinking data (Figs 8 and 9 ) do, in fact, show that Cbp2 may contact both introns in P4 and in the P2 region of the substrate domain. Nevertheless, the [omega] intron is active at lower salt concentrations in the absence of Cbp2, suggesting that the substrate domain is able to dock with the catalytic core more readily in this intron. It is likely that the peripheral structures of substrate domain of bI5 affect its binding by the catalytic core. Deletion of extensive L1 of bI5 does not render the intron autocatalytic in low salt (26 ,29 ), suggesting that differences in P2 or J2/3 may be significant.
It is likely that the effect of Cbp2 on [omega] detected in vitro accounts for the differences in growth on non-fermentable carbon sources between strains with and without CBP2. Once adapted to ethanol/glycerol medium, cells containing [omega] but not bI5 grew at nearly the same rate whether Cbp2 was present or not. However, adaptation to YPGE medium in cells lacking Cbp2 was considerably slower, which may be a consequence of a reduced ability to assemble mitochondrial ribosomes. This interpretation is supported by the reduced accumulation of 15S and 21S rRNA in cbp2 mutant cells. In contrast, Cbp2 is absolutely required for the splicing of COB mRNA and for growth on non-fermentable carbon sources. It is formally possible that another protein in mitochondria can assist the splicing of [omega] and that Cbp2 plays a subsidiary role. Such a protein has not been detected in genetic screens, and is probably not the intron encoded endonuclease, since mutations in this protein do not block splicing (48 ,49 ).
Nearly all laboratory strains of S.cerevisiae retain the terminal intron of COB (bI5) and, therefore, require the function of Cbp2. Yet bI5 is not essential for normal growth and respiration. Recently, Li et al. (42 ) showed that S.douglasii contains a Cbp2 homologue (87% protein sequence identity), even though bI5 is absent from the mitochondrial genome of this yeast. Disruption of CBP2 resulted in respiratory deficiency, except in a strain lacking mitochondrial introns. They concluded that CBP2 is required for the processing of a mitochondrial intron in S.douglasii and showed that the CBP2 from S.cerevisiae could functionally replace the endogenous gene. The large ribosomal RNA gene of S.douglasii mitochondria contains a group IA intron that is nearly identical to [omega] of S.cerevisiae, except in L8, which encodes the transposase function of [omega]. Our data suggest (Fig. 7 ) that Sd LSU requires Cbp2 for splicing. We found that the Cbp2 protein from S.cerevisiae stimulates the splicing of SdLSU at 3 mM MgCl2 and that the threshold of autocatalytic splicing is 20 mM MgCl2 (at 500 mM monovalent salt). Therefore, the SdLSU intron has splicing requirements similar to bI5 and not to [omega] (ScLSU).
The LSU introns from S.cerevisiae and S.douglasii are nearly identical in sequence and predicted secondary structure. Nevertheless, ScLSU (4 ) splices autocatalytically at low salt and SdLSU does not. SdLSU lacks a conserved G-C base pair that forms part of P9.0. (It has the potential for an A-U pair, instead.) We have recently shown that Cbp2 can suppress the effects of P9.0 mutations in bI5 in vitro (30 ). The less stable P9.0a may explain the Cbp2 dependence of the SdLSU intron and the requirement for high ionic strength and magnesium for autocatalysis. Mutational analysis will allow us to test this hypothesis.
One might suppose that proteins like Cbp2, which are required to express certain mitochondrial genes, serve to coordinate the activities of nuclear and mitochondrial genome. Nevertheless, synthesis of CBP2 mRNA is not subject to glucose repression (Lawson and Lewin, unpublished data) while synthesis of COB mRNA is repressed by glucose. Therefore a regulatory function of this protein remains elusive. It may be possible that Cbp2, like Cyt18, has a second function. If so, this activity is clearly not essential for mitochondrial function or for cell viability.
This work was supported by a grant from the National Institute of General Medical Sciences (RO1 GM12228). Support facilities were provided by the Interdisciplinary Center for Biotechnology Research and the Center for Mammalian Genetics at the University of Florida. The authors thank Drs Gerard Faye and Fred Winston for the gift of yeast strains and plasmid pMRFW2 respectively. Dr Thomas Fox provided strain MC109. A plasmid encoding the td intron was the gift of Dr Marlene Belfort, and Dr Richard Waring sent us a plasmids encoding the AnLSU and Nox1C introns.
*To whom correspondence should be addressed. Tel: +1 352 392 0676; Fax: +1 352 392 3133; Email: lewin@college.med.ufl.edu
Table 1
Strain
Nuclear genotype
Mitochondrial genotype
D273-10B/G1
Mat[alpha] met6 lys2
[omega], aI1, aI2, aI3, aI4, aI5[gamma], bI4, bI5
DM445-R1
Mat[alpha] met6 lys2
[omega], aI1, aI2, aI3, aI4, aI5[gamma], bI4
GF157-4B
Mat[alpha] lys2 canR
aI2, aI3, aI5[gamma]
GF110-8C
Mata his1 ade2 canR
aI2, aI3, aI5[gamma], bI4, bI5
MC109
Mat[alpha] ade2 ura3-52 kar1-1
[rho]o
YAL7
Mat[alpha] ade2 ura3-52
aI2, aI3, aI5[gamma], bI4, bI5
<>P
REFERENCES
Return