A Conserved Domain of Yeast RNA Triphosphatase Flanking the Catalytic Core Regulates Self-association and Interaction with the Guanylyltransferase Component of the mRNA Capping Apparatus*

The 549-amino acid yeast RNA triphosphatase Cet1p catalyzes the first step in mRNA cap formation. Cet1p consists of three domains as follows: (i) a 230-amino acid N-terminal segment that is dispensable for catalysis in vitro and for Cet1p function in vivo; (ii) a protease-sensitive segment from residues 230 to 275 that is dispensable for catalysis but essential for Cet1p function in vivo; and (iii) a catalytic domain from residues 275 to 539. Sedimentation analysis indicates that purified Cet1(231–549)p is a homodimer. Cet1(231–549)p binds in vitro to the yeast RNA guanylyltransferase Ceg1p to form a 7.1 S complex that we surmise to be a trimer consisting of two molecules of Cet1(231–549)p and one molecule of Ceg1p. The more extensively truncated protein Cet1(276–549)p, which cannot support cell growth, sediments as a monomer and does not interact with Ceg1p. An intermediate deletion protein Cet1(246–549)p, which supports cell growth only when overexpressed, sediments principally as a discrete salt-stable 11.5 S homo-oligomeric complex. These data implicate the segment of Ceg1p from residues 230 to 275 in regulating self-association and in binding to Ceg1p. Genetic data support the existence of a Ceg1p-binding domain flanking the catalytic domain of Cet1p, to wit: (i) the ts growth phenotype of 2μCET1(246–549) is suppressed by overexpression of Ceg1p; (ii) a ts alanine cluster mutationCET1(201–549)/K250A-W251A is suppressed by overexpression of Ceg1p; and (iii) 15 othercet-ts alleles with missense changes mapping elsewhere in the protein are not suppressed by Ceg1p overexpression. Finally, we show that the in vivo function of Cet1(275–549)p is completely restored by fusion to the guanylyltransferase domain of the mouse capping enzyme. We hypothesize that the need for Ceg1p binding by yeast RNA triphosphatase can by bypassed when the triphosphatase catalytic domain is delivered to the RNA polymerase II elongation complex by linkage in cis to the mammalian guanylyltransferase.

The Saccharomyces cerevisiae RNA 5Ј-triphosphatase Cet1p is an essential enzyme that catalyzes the first step of mRNA cap formation: the hydrolysis of the ␥-phosphate of triphosphate-terminated pre-mRNA to form a diphosphate end that serves as the substrate for capping by RNA guanylyltransferase (1,2). Cet1p is the prototype of a newly identified family of divalent cation-dependent nucleoside triphosphatases that includes the RNA 5Ј-triphosphatases encoded by two families of DNA viruses, poxviruses and baculoviruses (3). The enzyme family is defined by the presence of three conserved collinear motifs (A, B, and C) that include clusters of acidic and basic amino acids essential for triphosphatase activity (3)(4)(5)(6)(7). Motifs A-C of yeast Cet1p are located within the C-terminal half of the 549-amino acid protein (Fig. 1A). The N-terminal 200amino acid segment of Cet1p is dispensable for catalytic activity in vitro (2). Moreover, the N⌬200 allele CET1(201-549) is fully capable of supporting yeast cell growth (2).
Fungi and mammals have evolved different strategies to assemble bifunctional capping enzymes with triphosphatase and guanylyltransferase activities. In S. cerevisiae, separately encoded triphosphatase (Cet1p) and guanylyltransferase (Ceg1p) proteins interact to form a heteromeric complex (1,2,8), whereas in mammals, autonomous triphosphatase and guanylyltransferase domains are linked in cis within a single polypeptide (Mce1p) (9 -13). Genetic evidence suggests that the physical linkage of the mammalian triphosphatase and guanylyltransferase domains is essential in vivo. For example, a lethal yeast cet1 deletion can be complemented by expression of MCE1 but only if the catalytically active mouse triphosphatase domain is linked in cis to a catalytically active guanylyltransferase domain (2,12). Our isolation of CET1 in a genetic screen for high copy suppressors of temperature-sensitive mutations of Ceg1p suggested that interaction of the endogenous yeast triphosphatase and guanylyltransferase in trans may also be essential in vivo (2). Although analyses of the spectrum of ceg1-ts mutations that could be suppressed by Cet1p overexpression showed clustering of such lesions within a C-terminal domain of Ceg1p (2,14), there is, as yet, no physical identification of a Cet1p-binding site on Ceg1p.
A variety of qualitative methods have been employed to document Cet1p-Ceg1p interaction in vivo (by yeast two-hybrid reporter assay) or in vitro (far Western blotting or co-immunoprecipitation/Western blotting) (1,14,15). A drawback of these studies is that they provide limited insight into the nature of the complexes formed, and the results have not been integrated with a genetic assessment of the function of a presumptive Ceg1p-binding site on Cet1p. Our approach has been to study the interaction of purified recombinant Cet1p with purified recombinant Ceg1p by zonal velocity sedimentation (2). In this method, the purity and concentration of the input proteins, as well as the extent of complex formation (manifest as a shift in the sedimentation peaks), are apparent from the gradient profiles. Moreover, the native size of the Cet1p-Ceg1p complexes can be gauged from the sedimentation behavior. We have shown (2) that the N-terminal deletion mutant Cet1(201-549)p forms a discrete complex with Ceg1p, in near quantitative yield, that is stable to centrifugation in a glycerol gradient. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Thus, the catalytic domain and the Ceg1p-binding site reside within Cet1(201-549)p, which is fully functional in vivo.
Here we present further physical and genetic analysis of the yeast RNA triphosphatase that includes the following: (i) delineation of a domain structure of Cet1(201-549)p by limited proteolysis; (ii) definition by serial N-and C-terminal deletions of Cet1(241-539)p as a minimum functional domain sufficient for cell growth at 25-37°C; (iii) physical characterization of the Cet1p N-terminal deletion mutants and their complexes with Ceg1p; (iv) demonstration of suppression of the temperaturesensitive growth phenotype of the deletion mutant CET1(246 -549) by overexpression of Ceg1p; (v) characterization of a lethal deletion mutant, CET1(276 -549), which encodes a monomeric enzyme with full catalytic activity that is unable to interact with Ceg1p; and (vi) restoration of the in vivo function of Cet1(276 -549)p by fusion in cis with the guanylyltransferase domain of the mouse capping enzyme. Our results suggest that a protease-sensitive segment of Cet1p located proximal to the core catalytic domain regulates both Cet1p-Cet1p self-association and heteromerization with Ceg1p.

EXPERIMENTAL PROCEDURES
CET1 Deletion Mutants-A series of N-terminal deletion mutants of CET1 was constructed by PCR 1 amplification with mutagenic sensestrand primers that introduced an NdeI restriction site and a methionine codon in lieu of the codons for Glu 210 , Asn 220 , Asn 230 , Lys 240 , and Ile 275 . The PCR products were digested with NdeI and BamHI and then inserted into yeast plasmid pCET1-5Ј3Ј (CEN TRP1) (2). The mutated genes were named according to the amino acid coordinates of their polypeptide products, i.e. CET1(211-549), CET1(221-549),CET1(231-549), CET1(241-549), and CET1(276 -549). C-terminal truncation mutants CET1(201-539) and CET1(201-529) were constructed by PCR amplification using antisense primers that introduced translation stop codons in lieu of the codons for Glu 540 and Ile 530 and BamHI sites immediately 3Ј of the new stop codons. The PCR products were digested with NdeI and BamHI and then inserted into pCET1-5Ј3Ј. Expression of the deleted alleles in these plasmids is under the control of the natural CET1 promoter. 2 CET1 Expression Vectors-Restriction fragments containing CET1, CET1(201-549), CET1(246 -549), and CET1(276 -549) were excised from the respective pCET1-5Ј3Ј-based plasmids and inserted into the yeast expression vector pYX232 (2 TRP1). In the resulting plasmids, expression of the yeast RNA triphosphatase is under the control of the yeast TPI1 promoter.
Expression and Purification of Recombinant Ceg1p-The CEG1 gene was inserted into a customized T7-based expression plasmid (a derivative of pET16b) in such a way as to fuse the 459-amino acid CEG1coding sequence in frame to an N-terminal 29-amino acid leader peptide containing 10 consecutive histidine codons sequence (MGSHHHHHH-HHHHSSGHIEGRHSRRASVH). The plasmid was transformed into Escherichia coli BL21(DE3). A 500-ml culture was grown at 37°C in LB medium containing 0.1 mg/ml ampicillin until the A 600 reached 0.6. The culture was adjusted to 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside and 2% ethanol and then incubation was continued for 21 h at 17°C. Cells were harvested by centrifugation and stored at Ϫ80°C. All subsequent procedures were performed at 4°C. Thawed bacteria were resuspended in 25 ml of buffer A (50 mM Tris-HCl, pH 7.5, 0.2 M NaCl, 10% sucrose). Lysozyme was added to a final concentration of 50 g/ml; the suspension was incubated on ice for 10 min and then sonicated for 30 s. Triton X-100 was added to a final concentration of 0.1%, and sonication was repeated to reduce the viscosity of the lysate. Insoluble material was removed by centrifugation for 45 min at 18,000 rpm in a Sorvall SS34 rotor. The soluble extract was applied to a 2-ml column of Ni 2ϩ -NTA-agarose (Qiagen) that had been equilibrated with buffer A containing 0.1% Triton X-100. The column was washed with the same buffer and then eluted stepwise with 5 ml of buffer B (50 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 10% glucose, 0.05% Triton X-100) containing 50, 100, 200, 500, and 1000 mM imidazole. The polypeptide compositions of the column fractions were monitored by SDS-PAGE. Ceg1p was retained on the Ni 2ϩ -agarose column and recovered in the 0.2 M imidazole eluate. The 0.2 M imidazole fraction was dialyzed against buffer C. The Ceg1p enzyme preparation was stored at Ϫ80°C.
Glycerol Gradient Sedimentation-Aliquots of the Ni 2ϩ -agarose preparations of truncated Cet1p proteins (typically ϳ30 g) were mixed with catalase (25 g), BSA (25 g), and cytochrome c (25 g) in 0.2 ml of buffer G (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 1 mM EDTA, 2 mM dithiothreitol, 0.05% Triton X-100). Where indicated, the mixtures also included recombinant Ceg1p (ϳ30 g). The mixtures were incubated on ice for 45 min and then layered onto a 4.8-ml 15-30% glycerol gradient containing buffer G. The gradients were centrifuged in a Beckman SW50 rotor at 50,000 rpm for 16 h at 4°C. Fractions (0.2 ml) were collected from the bottoms of the tubes. Aliquots (20 l) of odd-numbered fractions were analyzed by SDS-PAGE along with samples of the input protein mixtures for each gradient. Polypeptides were visualized by staining with Coomassie Blue dye.
Isolation of cet1-ts Mutants-The CET1 gene was amplified in vitro by Taq DNA polymerase. The standard PCR reaction mixture was modified to contain either (i) a reduced concentration of dATP (0.08 mM) relative to the other three dNTPs (each at 2 mM); (ii) 0.08 mM dATP plus 0.2 mM manganese chloride; or (iii) 2 mM dATP plus 0.2 mM manganese chloride. The PCR products were digested with NdeI and BamHI. The mutagenized CET1 DNA fragments from 10 separate PCR reactions (4 using PCR reaction condition i, and 3 each using conditions ii and iii) were ligated separately into pCET1-5Ј3Ј, and the ligation mixtures were transformed into E. coli. After amplification in vivo, 10 pooled plasmid libraries were prepared, each from approximately 3,000 ampicillin-resistant colonies harvested directly from the agar plates. These DNA libraries were transformed into yeast strain YBS20. Trp ϩ transformants (150 for each transformed pool) were patched on agar medium lacking tryptophan at 30°C and then replica-plated on 5-FOA medium at 30°C to eliminate the wild type CET1 allele on a CEN URA3 plasmid. Six to fifteen percent of the transformants were unable to grow on 5-FOA. The surviving isolates, which had lost the CET1 URA3 plasmid, were replica-plated on YPD agar and incubated at 30°C (permissive temperature) and 37°C (nonpermissive temperature). Those able to grow at 30 but not 37°C were selected. Plasmid DNA was recovered from candidate ts mutants, amplified in vivo in E. coli, and retested for the conditional growth phenotype by plasmid shuffle. In this way, we obtained a collection of cet1-ts mutants. Fifteen of these mutant cet1 clones were mapped at the nucleotide level by DNA sequencing.
Candida albicans CET1 Deletion Mutants-The complete C. albicans open reading frame encoding the 520-amino acid RNA triphosphatase CaCet1p (15) was amplified by PCR from a C. albicans DNA library (16) using a sense primer that introduced an NdeI site at the translation start codon and an antisense primer that introduced a BamHI site immediately 3Ј of the translation stop codon. The PCR product was digested with NdeI and BamHI and then inserted into a customized yeast expression vector pYN132 (CEN TRP1), a derivative of pYX132 in which a unique NdeI site replaced the NcoI site of pYX132. The result-ing plasmid was named pCaCet1(1-520). N-terminal deletion mutants of CaCET1 were constructed by PCR amplification with mutagenic sense-strand primers that introduced an NdeI restriction site and a methionine codon in lieu of the codon for Asp 178 , Gln 195 , Tyr 202 , or Asn 216 . The PCR products were digested with NdeI and BamHI and then inserted into pYN132 to yield plasmids pCaCet1(179 -520), pCaCet1(196 -520), pCaCet1(203-520), and pCaCet1(217-520). In these vectors, expression of the C. albicans polypeptide is under the control of the TPI1 promoter.
Chimeric Yeast-Mammalian Capping Enzymes-Genes encoding fusion proteins composed of an N-terminal segment derived from yeast Cet1p and the catalytically active C-terminal guanylyltransferase domain of the mouse capping enzyme [Mce1(211-597)p] were constructed as follows. The CET1 coding sequences from residues 201-549, 275-549, and 301-549 were PCR-amplified using the respective pET-based CET1 plasmids as templates and an antisense primer that changed Val 548 to His and introduced an NdeI restriction site at the C-terminal dipeptide His 548 -Met 549 . The PCR products were digested with NdeI and then inserted into the Expression of the chimeric genes in these plasmids is under the control of the TPI1 promoter.

RESULTS
Probing the Structure of Yeast RNA Triphosphatase by Limited Proteolysis-Recombinant Cet1(201-549)p containing a short N-terminal His tag was subjected to proteolysis with increasing concentrations of chymotrypsin, trypsin, and V8 proteases (Fig. 2). SDS-PAGE analysis of the undigested protein preparation revealed a 44-kDa polypeptide ( Fig. 2A) corresponding to His-tagged Cet1(201-549)p (predicted size of 43 kDa). Sequencing of this species by automated Edman chemistry after transfer to a polyvinylidene difluoride membrane confirmed that the N-terminal sequence (GHHHHH) corresponded to that of the recombinant protein beginning from the second residue of the His tag. Apparently, the triphosphatase suffered removal of the initiating methionine during expression in E. coli. Initial scission of Cet1(201-549)p at low concentrations of chymotrypsin yielded a predominant ϳ39-kDa species with N-terminal sequence RNVPIW resulting from chymotryptic cleavage at Tyr 241 /Arg 242 ( Fig. 2A). At 2-fold higher protease concentration, this species was converted to an ϳ36-kDa polypeptide with N-terminal sequence QSINVK, arising via cleavage at Leu 258 /Gln 259 ( Fig. 2A). Even higher chymotrypsin concentrations yielded two major products, a ϳ26-kDa polypeptide with its N terminus at Gln 259 and a ϳ12-kDa fragment with N-terminal sequence KSQSPI resulting from scission at Tyr 445 /Lys 446 ( Fig. 2A). These two species persisted at the highest levels of chymotrypsin tested (8-fold greater than the amount sufficient to cleave all of the input protein at least once), although some of the protein was digested to a mixed cluster of low molecular fragments. One of these fragments retained an N terminus at Gln 259 , whereas two novel fragments were generated via cleavage at Tyr 380 /Arg 381 (RVGLST) or Tyr 461 /Ile 462 (IHNDSXT). The evolution of the digestion pattern with increasing protease concentration suggested the sequence of chymotryptic cleavages diagrammed in Fig. 1B.
The key point here is that the principal chymotrypsin-accessible sites (Tyr 241 and Leu 258 ) are located in the N-terminal portion of Cet1(201-549)p.
Initial scission by V8 protease occurred between Glu 231 and Ile 232 to yield a ϳ39-kDa polypeptide with N-terminal sequence ISASSK (Fig. 2C). At higher V8 concentrations, this fragment was converted transiently to a ϳ32-kDa species with the same N terminus (Fig. 2C). Even higher levels of V8 led to the transient appearance of a ϳ26-kDa fragment with N-terminal sequence LDAHLT via scission at Glu 334 /Leu 335 (Fig. 2C). Also arising was a ϳ12-kDa species with N-terminal residue Ile 232 ; this fragment was presumably generated when the larger precursor was cleaved at Glu 334 . This 12-kDa species persisted at the highest V8 levels tested, whereas the 26-kDa fragment originating at Leu 335 was converted into an ϳ19-kDa species with the same N terminus (Fig. 2C).
Initial cleavage by trypsin yielded a large fragment retaining the N-terminal His tag (Fig. 2B). At the highest level of trypsin tested, the protein was converted to a stable ϳ27-kDa doublet arising via closely spaced cleavages at Lys 256 /Ala 257 (leaving N-terminal sequence ALQSIN) and Lys 264 /Asp 265 (sequence DLKIDP) (Fig. 2B).
The experimentally determined protease cleavage sites within Cet1(201-549)p are annotated on the primary structure in Fig. 1A. The instructive point of this analysis is that the principal or prominent sites of accessibility to all three proteases are clustered within a short polypeptide segment from Glu 231 to Lys 264 . We infer that this segment is either disordered or surface-exposed. In contrast, the distal portion of the protein is relatively protease-insensitive and likely to comprise a folded domain.
Deletion Analysis of Yeast RNA Triphosphatase Defines a Minimal Functional Domain in Vivo-We reported previously that expression of the truncated derivative Cet1p(201-549)p on a single copy plasmid under the control of the natural CET1 promoter fully complemented the growth of a ⌬cet1 deletion strain of S. cerevisiae, whereas expression of the more extensively truncated derivatives Cet1(246 -549)p and Cet1(301-549)p did not complement (2). In light of the proteolysis results The margins of the latter two alleles were placed at Glu 231 and Tyr 241 , the residues that were cleaved by V8 and chymotrypsin, respectively. The deleted genes were cloned into a CEN TRP1 vector so as to place them under the control of the CET1 promoter. These plasmids were transformed into a ⌬cet1 strain YBS20, in which the chromosomal CET1 locus has been deleted and replaced by LEU2. Growth of YBS20 is contingent on the maintenance of a wild type CET1 allele on a CEN URA3 plasmid. Hence, YBS20 is unable to grow on agar medium containing 5-fluoroorotic acid (5-FOA) unless it is transformed with a biologically active CET1 allele such as CET1(201-549) ( Fig. 3A; ⌬200) or a functional homologue from another organism (2). We found that growth on 5-FOA was complemented by CET1(211-549), CET1(221-549), CET1(231-549), and CET1(241-549) (Fig. 3, ⌬210, ⌬220, ⌬230, and ⌬240), but not by the more extensively truncated alleles CET1(246 -549) or CET1(276 -549) (Fig. 3B, ⌬245 and ⌬275). This experiment delineates a short peptide segment between residues 241 and 245 that is required for Cet1p function in vivo when the triphosphatase is expressed from its own promoter.
Previously, we showed that deletion of 30 or 60 amino acids from the C terminus of Cet1p resulted in loss of function in vivo (2). Here, we tested by plasmid shuffle a finer series of C-terminal truncation mutants and found that CET1(201-539) complemented growth of YBS20 on 5-FOA, whereas CET1(201-529) did not (data not shown). A mutant allele, CET1(241-539), in which both termini were truncated to their respective functional boundaries, also complemented ⌬cet1 (data not shown).
Effects of N-terminal Deletions on Triphosphatase Activity-The N-terminal deletion mutants of Cet1p were expressed in bacteria as His-tagged fusions and purified from soluble lysates by Ni 2ϩ -agarose column chromatography (Fig. 4A). Triphosphatase activity was determined by assaying manganesedependent ATP hydrolysis (3) as a function of input enzyme. The specific activities of Cet1(211-549)p, Cet1(221-549)p, Cet1(231-549)p, and Cet1(241-549)p were comparable to that of Cet1(201-549)p (Fig. 4B). This was in keeping with the expectation that any Cet1p derivatives that were active in vivo should retain catalytic activity in vitro. Note that Cet1(246 -549)p, which did not complement ⌬cet1 growth, was also active in vitro (Fig. 4B). The instructive finding was that Cet1(276 -549)p also retained full triphosphatase function in vitro ( A parallel set of reaction mixtures was resolved by SDS-PAGE, and the indicated polypeptide bands were subjected to automated N-terminal sequencing as described (22). The N-terminal sequences are denoted in single-letter code.

FIG. 3. Effect of N-terminal deletions on CET1 function in vivo.
Yeast strain YBS20 was transformed with CEN TRP1 plasmids containing the indicated N-terminal deletion alleles of CET1. A control transformation was performed using the TRP1 vector. Trp ϩ isolates were streaked on agar plates containing 0.75 mg/ml 5-FOA. The plates were photographed after incubation for 3 days at 30°C. trols), into 2 vectors wherein their expression is driven by the yeast TPI1 promoter. These genes were tested for function by plasmid shuffle; the counterselection on 5-FOA was performed at 25, 30, and 37°C. Whereas the 2 TPI1-CET1 and 2 TPI1-CET1(201-549) transformants gave rise to FOA-resistant colonies at all temperatures, the 2 TPI1-CET1(276 -549) transformants were unable to grow on 5-FOA under any conditions (data not shown). Thus, Cet1(276 -549)p, which is catalytically active in vitro, cannot sustain cell growth even when overexpressed in vivo. The salient finding was that ⌬cet1 cells transformed with 2 TPI1-CET1(246 -549) did give rise to FOA-resistant colonies at 25 and 30°C but not at 37°C (data not shown).
To characterize better the growth of the 2 TPI1-CET1(246 -549) strain, FOA-resistant cells selected at 25°C were streaked on YPD plates at 25, 30, 34, and 37°C in parallel with CET1 and CET1(201-549) controls. The 2 TPI1-CET1(246 -549) strain displayed a clear-cut ts phenotype at the restrictive temperatures of 34 or 37°C (Fig. 5A, ⌬245). Additional experiments revealed that growth of ⌬cet1 on 5-FOA at 25°C could also be complemented by a CEN plasmid containing CET1(246 -549) driven by the TPI1 promoter. The CEN TPI1-CET1(246 -549) strain was unable to grow on YPD medium at 34 or 37°C and grew more slowly than the 2 TPI1-CET1(246 -549) strain at 30°C (not shown). Thus, whereas increased promoter strength and gene dosage both serve to enhance the in vivo activity of Cet1(246 -549)p at 25-30°C, these maneuvers are insufficient to sustain cell growth at higher temperatures.
Overexpression of Guanylyltransferase Suppresses the ts Phe-notype of 2 TPI1-CET1(246 -549)-We showed previously that overexpression of Cet1p from a 2 plasmid suppressed certain ts mutations in the yeast guanylyltransferase Ceg1p. We hypothesized that the ts guanylyltransferase has diminished affinity for Cet1p at restrictive temperature (37°C) and that increasing the level of wild type Cet1p drives Ceg1p-Cet1p heteromerization by simple mass action (2). We have since found that the ceg1-25 mutant can be suppressed by 2 TPI1-CET1(201-549) but not by 2 TPI1-CET1(246 -549) (data not shown). These results suggest that deletion of the Cet1p segment from residues 201 to 245 may compromise its interaction with Ceg1p at 37°C, a scenario that may explain the ts phenotype of the 2 TPI1-CET1(246 -549) strain. Is the genetic interaction between these cap-forming enzymes reciprocal, i.e. can overexpression of the guanylyltransferase suppress the growth defect of a conditional mutant of the triphosphatase? To address this question, we transformed the 2 TPI1-CET1(246 -549) strain with a 2 URA3 CEG1 plasmid and tested the growth of Ura ϩ isolates at permissive and restrictive temperatures. Control cells transformed with a 2 URA3 CET1 plasmid grew at 25 and 37°C, whereas cells transformed with the 2 URA3 vector grew at 25°C only (Fig.  5B). Remarkably, the introduction of CEG1 on a 2 plasmid rescued growth at 37°C and did so as effectively as wild type CET1, as judged by colony size (Fig. 5B). 2 CEG1 also suppressed the ts growth phenotype of the CEN TPI1-CET1(246 -549) strain (not shown). These experiments suggest that Cet1(246 -549)p is defective for interaction with Ceg1p at 37°C in vivo.
Suppression of cet1-ts Mutations by 2 CEG1 Is Allele-specific-In order to examine the allele specificity of the genetic interaction between the yeast triphosphatase and guanylyltransferase, we isolated a collection of temperature-sensitive cet1 alleles from separate pools of mutagenized CET1 clones. To do this, we performed PCR amplification of the CET1 gene under reaction conditions designed to promote nucleotide misincorporation by Taq polymerase. The PCR products from separate amplification reactions were restricted and cloned into a CEN TRP1 expression plasmid; in this context, expression of the inserted gene is under the control of the natural CET1 promoter. After amplification in vivo in E. coli, pooled libraries of CEN TRP1 DNAs were transformed into yeast containing wild type CET1 on a URA3 plasmid. After selection for Trp ϩ growth at 30°C, followed by selection on 5-FOA to eliminate the wild type allele, we obtained populations of cells that were screened for growth at 37°C. We identified cet1-ts isolates that readily formed colonies at 30°C but were unable to form colonies at 37°C. Plasmid DNA was recovered from conditional isolates and retested by plasmid shuffle for the ts phenotype.
Fifteen cet1 mutant DNA clones were selected and sequenced (Fig. 6). Six of the ts alleles contained a single missense mutation. Nine ts alleles had multiple missense changes. Substitutions at residues Asn 46 , Met 108 , Glu 408 , Ser 518 , and Phe 523 were encountered twice. The ts alleles containing repeat "hits" represented independent isolates, insofar as each clone contained unique mutations not found in the other clones. There were 33 coding changes affecting 28 different residues (Fig. 6). None of the mutations were located within the essential triphosphatase motifs A, B, and C. Many of the changes were tightly clustered within several discrete segments of the carboxyl domain of the enzyme, e.g. the S518P, L519P, and F523L cluster (Fig. 6). All of the cet1-ts alleles were recessive to wild type CET1, i.e. growth at 37°C could be restored to the cet1-ts strains by transforming them with a CEN URA3 plasmid bearing the wild type CET1 gene (not shown). Remarkably, none of the cet1-ts mutants could be rescued to growth at the restrictive temperature by transformation with a 2 URA3 plasmid bearing the CEG1 gene (not shown). Therefore, suppression of the conditional growth phenotype of yeast triphosphatase mutations by overexpression of guanylyltransferase was highly specific for the 2 CET1(246 -549) allele.
Physical Characterization of N-terminal Deletion Mutants of Cet1p and Their Interaction with Ceg1p-The native sizes of several of the N-terminal deletion mutants of Cet1p were investigated by sedimentation of the recombinant enzymes through 15-30% glycerol gradients containing 100 mM NaCl. At the same time, we tested the interaction of the Cet1p deletion mutants with recombinant Ceg1p by mixing the proteins in vitro and sedimenting the mixture through a glycerol gradient. Marker proteins (catalase, BSA, and cytochrome c) were included as internal standards in every glycerol gradient. After centrifugation, the polypeptide compositions of the gradient fractions were analyzed by SDS-PAGE.
The sedimentation profile for Cet1(201-549)p (⌬200) is shown in Fig. 7A. The ⌬200 triphosphatase (a 43-kDa polypeptide) sedimented as a discrete peak (fraction 15) just slightly heavier than the BSA peak (68-kDa). A plot of the S values of the three standards versus fraction number yielded a straight line (not shown). An S value of 5.0 was determined for Cet1(201-549)p from the internal standard curve. These results suggest that Cet1(201-549)p is a homodimer. (Note that our earlier estimate of an S value of 4.1 for recombinant Cet1(201-549)p (2) was based on a comparison to external markers sedimented in a different glycerol gradient.) The sedimentation profile for the mixture of Cet1(201-549)p and Ceg1p is shown in Fig. 7B. Here, the yeast proteins form a discrete complex (peak fraction 11) that sediments at a position between catalase and BSA. The sedimentation coefficient of the Cet1(201-549)p-Ceg1p complex was 7.3. This value suggested a heterotrimeric subunit structure which, given that Cet1(201-549)p per se sediments as a dimer, whereas Ceg1p alone is monomeric (see below), likely consists of two molecules of Cet1(201-549)p and one molecule of Ceg1p.
The sedimentation profile for Cet1(276 -549)p (⌬275) is shown in Fig. 8A. The ⌬275 protein (a 34-kDa polypeptide) sedimented as a single discrete component at a position be- FIG. 6. cet1-ts mutants. The coding changes detected in each of 15 ceg1-ts alleles are listed in the left panel. The mutations are annotated above the sequence of the wild type Ceg1p protein in the right panel. Amino acids that were altered more than once in the ts collection are underlined.
tween BSA and cytochrome c. The observed sedimentation coefficient of 3.6 suggests that Cet1(276 -549)p is a monomer. Additional sedimentation experiments confirmed that the triphosphatase activity cosedimented with the Cet1(276 -549)p polypeptide (data not shown). The sedimentation behavior of Cet1(276 -549)p was unaltered in the presence of Ceg1p (Fig.  8B). Ceg1p sedimented a discrete monomeric peak intermediate between BSA and ⌬275 (Fig. 8B). Ceg1p displayed the same profile when centrifuged in a gradient containing just Ceg1p and markers (data not shown). These results show that Cet1(276 -549)p, although catalytically active, lacks the capac-ity to bind to the guanylyltransferase. We hypothesize that the lack of Ceg1p binding accounts for the failure of Cet1(276 -549)p to support yeast cell growth.
Further insights into the self-association and Ceg1p-binding properties of yeast RNA triphosphatase emerged from sedimentation analyses of intermediate deletion mutants. For example, Cet1(231-549)p by itself sedimented as a discrete 4.9 S component (a presumptive homodimer) just slightly ahead of BSA (Fig. 9A). Mixture with Ceg1p resulted in the near quantitative formation of a Cet1(231-549)p-Ceg1p complex sedimenting at 7.1 S (Fig. 9B). By the same reasoning applied to the ⌬200 mutant, we presume this complex to be a heterotrimer. This experiment shows that the segment from residues 201 to 230 of Ceg1p is not required for Ceg1p binding in vitro.
A dramatic change in the sedimentation behavior of the triphosphatase occurred when the protein was truncated from residue 231 to 245. The ⌬245 enzyme by itself was resolved into two discrete peaks as follows: a heavy component (11.5 S) that sedimented slightly faster than catalase and a lighter component of 4.8 S (Fig. 10A). Although trace amounts of a heavy component can be discerned during gradient centrifugation of the ⌬200 and ⌬230 proteins (Figs. 7A and 9A), in the case of ⌬245, the heavy component comprises the majority of species. The position of the heavy component of ⌬245 relative to catalase (a 248-kDa tetramer of a 62-kDa subunit) suggests that the species is at least a hexamer, if not an octamer, of the 37-kDa Cet1(246 -549)p polypeptide. The 4.8 S light component is a presumptive homodimer. Note that additional sedimentation studies showed that the triphosphatase activity profile mirrored the biphasic distribution of the ⌬245 polypeptide, with the majority of the activity associated with the heavy component (data not shown). The stability of the oligomeric form of Cet1(246 -549)p was investigated by increasing the ionic strength of the protein sample (to 0.25, 0.5, or 1.0 M NaCl) and then centrifuging the ⌬245 and internal standards through glycerol gradients containing 0.25, 0.5, or 1.0 M NaCl. These experiments showed that the sedimentation coefficient of the heavy and light components and the distribution of the ⌬245 polypeptide between the two components was unaffected by salt up to 1 M NaCl (data not shown).
Mixing ⌬245 with Ceg1p resulted in a shift of the light component to form a Cet1(246 -549)p-Ceg1p complex sedimenting at 7.1 S (Fig. 10B, fractions 11-13). A fraction of the Ceg1p also became associated with the heavy component (Fig. 10B,  fraction 3). The ratio of the Ceg1p to ⌬245 proteins was much lower in the heavy component complex than in the 7.1 S complex (a putative heterotrimer).
Finally, the sedimentation behavior of Cet1(241-549)p (⌬240) constituted a forme fruste of the aberrant behavior of ⌬245, meaning that ⌬240 by itself was resolved into two discrete peaks of 12 S (Fig. 11A, fraction 3) and 4.8 S (Fig. 11A,  fraction 15), but the distribution of the protein was biased toward the light component. Mixture with Ceg1p resulted in the formation of a 7.2 S Cet1(241-549)p-Ceg1p complex (Fig.  11B, lane 11) and the association of some Ceg1p with the heavy component.
The conclusions of the sedimentation analysis can be summarized as follows: (i) the C-terminal domain of Cet1p from residues 231 to 549 has an intrinsic capacity to self-associate to form a homodimer, which binds stably to Ceg1p; (ii) deletion of the segment from residues 231 to 245 unmasks a latent capacity to form a salt-stable 11-12 S oligomer consisting of at least 6 Ceg1p protomers; and (iii) deletion of the segment from residues 246 to 275 abolishes self-association and the ability of Cet1p to bind to Ceg1p.
A Putative Guanylyltransferase-binding Element Is Conserved and Essential in C. albicans RNA Triphosphatase-The C. albicans CaCET1 gene encodes a 520-amino acid RNA triphosphatase (15). The C-terminal segment of CaCet1p from residues 186 to 520 displays extensive sequence similarity to the carboxyl portion of S. cerevisiae Cet1p from positions 225 to 538 (95 identical residues and 64 positions with side chain similarity). In contrast, the N-terminal segments of CaCet1p and Cet1p are not conserved. We cloned wild type CaCET1 and a series of N-terminal deletion alleles of CaCET1 into CEN vectors under the control of the TPI1 promoter and then tested them by plasmid shuffle for complementation of the S. cerevisiae ⌬cet1 mutant. The CaCET1(179 -520), CaCET1(196 -520), and CaCET1(203-520) mutants were viable (Fig. 12A, ⌬178, ⌬195, ⌬202), whereas CaCET1(217-520) was lethal (Fig. 12A,  ⌬216). According to the sequence alignment shown in Fig. 12B, the viable ⌬185 and ⌬202 deletions of CaCet1p correspond to deletions of 235 and 242 amino acids from the N terminus of Cet1p, whereas the lethal ⌬216 mutation corresponds to a deletion of 256 residues from Cet1p. These results show that the upstream functional borders of the two fungal RNA triphosphatases are quite similar.
The viable S. cerevisiae strains bearing deletion mutants of CaCET1 were tested for growth at 37°C in rich medium. The CaCET1(179 -520) and CaCET1(196 -520) mutants grew at 37°C, whereas CaCET1(203-520) cells did not (not shown). The ts growth phenotype of CaCET1(203-520) was suppressed by the introduction of a 2 plasmid containing the S. cerevisiae guanylyltransferase gene CEG1 (not shown). Thus, the CaCET1 ⌬202 mutant displayed the same phenotypes with respect to conditional growth and high copy suppression as did the ⌬245 mutant of S. cerevisiae CET1.
The incremental deletions that result in lethality or conditional lethality of the TPI1-driven alleles highlight a conserved motif, PIWAQKWXP, that may comprise part of the guanylyltransferase-binding site of yeast RNA triphosphatases (Fig.  12B). To begin to test this idea, we constructed two mutated versions of Cet1(201-549)p, P245A/I246A and K250A/W251A, in which pairs of neighboring amino acids within this conserved motif were replaced by alanine. Additional alanine cluster mutations, D225A/L226A, K237A/P238A, and K240A/ Y241A, were introduced at conserved dipeptides just upstream of the PIWAQKWXP motif. The mutated residues are denoted by asterisks in Fig. 12B. The mutant genes were cloned into a CEN TRP1 vector under the control of the CET1 promoter and then tested by plasmid shuffle for complementation of a ⌬cet1 deletion strain. The D225A/L226A, K237A/P238A, K240A/ Y241A, P245A/I246A, and K250A/W251A mutants were viable after selection on 5-FOA at 25 or 30°C. D225A/L226A, K237A/ P238A, K240A/Y241A, and P245A/I246A cells grew as well as wild type cells on rich medium at 25, 30, and 37°C. In contrast, the K250A/W251A strain grew at 25 and 30°C but not at 37°C (Fig. 12B). The ts phenotype of the K250A/W251A mutant was suppressed by the introduction of CEG1 on a 2 plasmid (Fig.  12B). These findings implicate the Lys 250 and/or Trp 251 side chains as constituents of a guanylyltransferase-binding site on yeast RNA triphosphatase.

Enzyme Bypasses the in Vivo Requirement for a Ceg1p
Interaction Domain-The experiment in Fig. 13 shows that expression of mammalian triphosphatase-guanylyltransferase in yeast can complement the growth of a ⌬cet1 ⌬ceg1 strain (YBS50) in a plasmid shuffle assay. Growth of YBS50 depends on maintenance of a CEN URA3 CET1 CEG1 plasmid. Transformation of YBS50 with a CEN TRP1 plasmid containing MCE1 (the gene encoding the bifunctional mouse capping enzyme) under control of the yeast TPI1 promoter allowed growth of YBS50 on medium containing 5-FOA. A control transformation showed that MCE1(211-597), which encodes only the guanylyltransferase domain of mouse capping enzyme, was incapable of complementing ⌬cet1 ⌬ceg1 (Fig. 13). We then tested the activity of two novel genes, CET1(201-547)-MCE1(211-597) and CET1 (276 -547)-MCE1(211-597), encoding chimeric capping enzymes in which segments of the yeast RNA triphosphatase [Cet1(201-547)p and Cet1(276 -547)p] are fused to the mouse guanylyltransferase domain. Both chimeric genes were perfectly capable of supporting growth of the ⌬cet1 ⌬ceg1 strain on 5-FOA (Fig. 13). Both chimeric strains grew normally on YPD medium at 37°C (data not shown). The salient point of this experiment is that the in vivo requirement for Cet1p residues 241-275 (containing the putative binding site for yeast guanylyltransferase) can be obviated by linking the triphosphatase catalytic domain in cis to a heterologous guanylyltransferase.
Additional experiments showed that the chimeric gene CET1(301-547)-MCE1(211-597) was incapable of complementing ⌬cet1 ⌬ceg1 (not shown). We infer from this result that the segment from residues 276 to 300 is important for Cet1p activity in a capacity other than guanylyltransferase binding. Efforts to test whether Cet1(301-549)p has catalytic activity have been hampered by the insolubility of the recombinant protein expressed in bacteria (2). DISCUSSION The work presented here contributes to an emerging model for the structural and functional organization of the yeast capping apparatus. Physical and genetic evidence shows that the triphosphatase component Cet1p consists of three domains as follows: (i) a 230-amino acid N-terminal segment that makes no discernible contribution to catalysis and is dispensable for Cet1p function in vivo; (ii) a protease-sensitive segment from residues 230 to 275 that is essential for Cet1p function in vivo and that mediates both Cet1p self-association and Cet1p binding to the yeast guanylyltransferase Ceg1p; (iii) an essential C-terminal catalytic domain that includes the conserved metaldependent triphosphatase motifs A, B, and C.
Two lines of evidence argue strongly that the interaction of Cet1p with Ceg1p is essential. First, the complete loss of in vivo function with incremental N-terminal deletion of Cet1p to position 275 correlates with the loss of Ceg1p binding in vitro. Second, the in vivo function of Cet1(276 -549)p can be restored by fusing this protein in cis to mouse guanylyltransferase, effectively bypassing the need for the guanylyltransferase interaction domain.
A requirement for the assembly of a bifunctional triphosphatase-guanylyltransferase enzyme (whether in trans as in yeast or in cis as in mammals and the yeast-mouse chimeras) highlights an evolutionarily conserved strategy for targeting the cellular capping apparatus to nascent pre-mRNAs via the binding of the guanylyltransferase component to the phospho- A, deletion analysis of CaCet1p. YBS20 was transformed with CEN TRP1 plasmids containing CET1 and the indicated alleles of C. albicans CET1 (CaCET1). A control transformation was performed using the TRP1 vector. Trp ϩ isolates were streaked on agar plates containing 0.75 mg/ml 5-FOA. The plates were photographed after incubation for 7 days at 25°C. B, double-alanine mutagenesis of Cet1p. The amino acid sequence of S. cerevisiae (Sc) Cet1p from residues 225 to 265 is aligned with the homologous segment of C. albicans (Ca) Cet1p. Identical amino acids are denoted by a colon; positions of side chain similarity are denoted by a dot. The margins of the viable N-terminal deletion alleles of CET1 and CaCET1 are denoted by arrowheads above and below the aligned sequences. The N terminus of the lethal CaCET1(217-520) mutant is marked by a small arrow with a cross. Amino acid pairs in Cet1p that were targeted for double-alanine replacement are indicated by asterisks above the sequence. CET1(K250A-W251A) cells were transformed with a 2 URA3 vector or 2 URA3 plasmids containing CEG1 or CET1. Individual Ura ϩ transformants were streaked on SC agar medium lacking uracil and tryptophan. The plates were incubated at either 25 (4 days) or 37°C (3 days) as indicated. rylated C-terminal domain (CTD) of elongating RNA polymerase II (9 -11, 17). Both full-length mouse capping enzyme Mce1p and the C-terminal guanylyltransferase domain Mce1(211-597)p bind directly to the phosphorylated CTD, whereas the isolated N-terminal mouse triphosphatase domain Mce1(1-211)p has no capacity to bind the CTD (11,18). In yeast, the guanylyltransferase per se also binds to CTD-PO 4 (10), whereas the yeast triphosphatase alone does not (14). 2 In effect, the guanylyltransferase chaperones the triphosphatase to the transcription complex. It is noteworthy that the mammalian guanylyltransferase can act as chaperone for the yeast triphosphatase, even though Cet1p is completely divergent in structure and catalytic mechanism from the mammalian RNA triphosphatase domain. This result implies that there is no vital functional interaction between the mammalian triphosphatase and guanylyltransferase domains that is unique or specific to the mammalian triphosphatase. It also suggests that the mammalian guanylyltransferase can act as a convenient vehicle to direct heterologous proteins to the transcription complex in vivo.
The genetic interaction whereby Cet1p overexpression suppresses a conditional mutation in Ceg1p (2, 14) is now reciprocated by the observation that Ceg1p overexpression suppresses the conditional phenotype of the 2 CET1(246 -549) strain. The simplest interpretation of this result is that Cet1(246 -549)p is defective in its interaction with Ceg1p. However, our analysis of the sedimentation behavior of recombinant Cet1(246 -549)p hints that the situation is more complex. The predominant form of the ⌬245 protein sediments as discrete salt-stable multimer that does not interact in a stoichiometric fashion with Ceg1p. Yet, the minor "light" component of ⌬245 does interact with Ceg1p in vitro. Hence, the in vivo phenotype of the ⌬245 mutation might arise as follows: (i) at low gene dosage and under the control of its own promoter, there is insufficient light ⌬245 to function in concert with Ceg1p and therefore yeast cells cannot grow; (ii) increased expression of ⌬245 at 25 or 30°C increases the level of triphosphatase available to interact with Ceg1p above the threshold required for viability; (iii) at 37°C, the light-heavy equilibrium might be skewed toward the multimeric state such that the level of functional ⌬245 dips back below the threshold for viability; and (iv) increased expression of Ceg1p in vivo drives the equilibrium toward ⌬245-Ceg1p heteromerization (and away from ⌬245 multimerization) by mass action, thus restoring a suprathreshold level of active capping enzyme complex at 37°C. The in vivo phenotype is apparently sensitive to changes in the multimerization equilibrium of the triphosphatase, given that the viable ⌬240 mutant also forms a stable multimer in vitro, albeit to a lesser extent that does ⌬245.
Overexpression of Ceg1p also suppressed the ts phenotype elicited when the K250A/W251A mutant of Cet1(201-549)p was expressed in single copy from the native CET1 promoter. Preliminary characterization of recombinant Cet1(201-549)-K250A/W251A protein revealed the following: (i) as expected, the mutant enzyme displayed full activity in manganese-dependent ATP hydrolysis in vitro, and (ii) the mutant protein sedimented as a discrete peak coincident with BSA (this peak comprised 94% of the ATPase activity), and only trace amounts of a heavy component (comprising 6% of the ATPase activity) were detected in the 11-12 S size range. 3 Thus, the in vivo effects of the K250A/W251A mutation are unlikely to be mediated by the overzealous multimerization observed for the ⌬245 deletion mutant. Instead, we hypothesize that elimination of the Lys 250 and Trp 251 side chains simply lowers the affinity of Cet1p for Ceg1p at restrictive temperature and that Ceg1p overexpression can compensate to restore a threshold level of the heteromeric complex. In preliminary sedimentation analysis, we found that Cet1(201-549)-K250A/W251A interacted normally with Ceg1p 3 ; this is not surprising given that the sedimentation is performed at low temperature. Our working model is that Lys 250 and/or Trp 251 comprise part of the guanylyltransferase-binding site on S. cerevisiae RNA triphosphatase. Sequence similarity between S. cerevisiae and C. albicans Cet1p suggests that the putative guanylyltransferasebinding site is conserved in other fungal RNA triphosphatases.
There has been uncertainty in the literature concerning the native size and subunit stoichiometry of the yeast triphosphatase-guanylyltransferase complex. Itoh et al. (19) initially reported the purification of a bifunctional complex from yeast cell extracts that consisted of a 45-kDa guanylyltransferase ␣-subunit and a 39-kDa RNA triphosphatase ␤-subunit. This enzyme sedimented at 7.3 S in a glycerol gradient calibrated with internal standards. From this datum, the authors estimated a native size of 140-kDa and proposed that the enzyme was a ␣2␤2 heterotetramer (19). An improved purification from yeast extracts yielded an enzyme, composed of a 52-kDa guanylyltransferase ␣-subunit and a 80-kDa RNA triphosphatase ␤-subunit, that sedimented at 9.0 S in a glycerol gradient (20). It is now known that the 80-kDa triphosphatase subunit corresponds to full-length Cet1p, which has an actual size of 62-kDa, but migrates anomalously slowly during SDS-PAGE (1,2). We surmise that the 39-kDa triphosphatase subunit present in the enzyme purified initially is a proteolytic fragment of Cet1p comprising the carboxyl catalytic domain. The 45-kDa form of guanylyltransferase is probably a proteolytic fragment lacking a nonessential segment at the C terminus (21). Yamada-Okabe et al. (15) recently suggested that the 9.0 S enzyme capping enzyme isolated from yeast cells (with an estimated size of 180 kDa) is either an ␣2␤1 or ␣1␤2 trimer. Based on the sedimentation analysis in the present study, we suggest that the enzyme complex reconstituted in vitro from separately expressed Cet1(201-549)p and Ceg1p proteins is a Ceg1p-[Cet1(201-549)p]2 trimer and that Cet1(201-549)p by itself is a homodimer. An ␣1␤2 trimer structure for the capping enzyme complex raises the prospects that either (i) the binding of one Ceg1p to one of the Cet1(201-549)p protomers of the dimer occludes the Ceg1p-binding surface of the other Cet1(201-549)p protomer or (ii) moieties on both triphosphatase protomers contribute to form a single guanylyltransferase-binding site.