An essential function of Saccharomyces cerevisiae RNA triphosphatase Cet1 is to stabilize RNA guanylyltransferase Ceg1 against thermal inactivation.

Saccharomyces cerevisiae RNA triphosphatase (Cet1) and RNA guanylyltransferase (Ceg1) interact in vivo and in vitro to form a bifunctional mRNA capping enzyme complex. Here we show that the guanylyltransferase activity of Ceg1 is highly thermolabile in vitro (98% loss of activity after treatment for 10 min at 35 degrees C) and that binding to recombinant Cet1 protein, or a synthetic peptide Cet1(232-265), protects Ceg1 from heat inactivation at physiological temperatures. Candida albicans guanylyltransferase Cgt1 is also thermolabile and is stabilized by binding to Cet1(232-265). In contrast, Schizosaccharomyces pombe and mammalian guanylyltransferases are intrinsically thermostable in vitro and they are unaffected by Cet1(232-265). We show that the requirement for the Ceg1-binding domain of Cet1 for yeast cell growth can be circumvented by overexpression in high gene dosage of a catalytically active mutant lacking the Ceg1-binding site (Cet1(269-549)) provided that Ceg1 is also overexpressed. However, such cells are unable to grow at 37 degrees C. In contrast, cells overexpressing Cet1(269-549) in single copy grow at all temperatures if they express either the S. pombe or mammalian guanylyltransferase in lieu of Ceg1. Thus, the cell growth phenotype correlates with the inherent thermal stability of the guanylyltransferase. We propose that an essential function of the Cet1-Ceg1 interaction is to stabilize Ceg1 guanylyltransferase activity rather than to allosterically regulate its activity. We used protein-affinity chromatography to identify the COOH-terminal segment of Ceg1 (from amino acids 245-459) as an autonomous Cet1-binding domain. Genetic experiments implicate two peptide segments, (287)KPVSLYVW(295) and (337)WQNLKNLEQPLN(348), as likely constituents of the Cet1-binding site on Ceg1.

RNA triphosphatase catalyzes the first step in mRNA cap formation entailing the cleavage of the ␤-␥ phosphoanhydride bond of 5Ј triphosphate RNA to yield a 5Ј diphosphate end that is then capped with GMP by RNA guanylyltransferase. The genetic and physical organization of these two capping enzymes differs in higher versus lower eukaryotes (1). Mammals encode a bifunctional capping enzyme (Mce1; 597 aa) 1  The budding yeast Saccharomyces cerevisiae encodes separate triphosphatase (Cet1; 549 aa) and guanylyltransferase (Ceg1; 459 aa) proteins that interact in trans to form a heteromeric capping enzyme complex.
The binding of yeast Cet1 to Ceg1 elicits two apparently beneficial outcomes. First, Cet1-Ceg1 interaction stimulates the guanylyltransferase activity of Ceg1 by increasing the extent of formation of the covalent Ceg1-GMP reaction intermediate (2,3). Second, the physical tethering of Cet1 to Ceg1 may facilitate recruitment of the capping apparatus to the RNA polymerase II elongation complex. Ceg1 binds in vitro and in vivo to the phosphorylated carboxyl-terminal domain (CTD) of the largest subunit of RNA polymerase II (4 -7), whereas Cet1 by itself does not interact in vitro with the phosphorylated CTD (8). It has also been suggested that Cet1 binding to Ceg1 antagonizes negative effects of CTD-PO 4 binding on the guanylyltransferase activity of Ceg1 (8).
Cet1 consists of three domains: (i) a 230-aa NH 2 -terminal segment that is dispensable for catalysis in vitro and for Cet1 function in vivo, (ii) a protease-sensitive segment from residues 230 to 275 that is dispensable for catalysis, but essential for Cet1 function in vivo, and (iii) a catalytic domain from residues 276 to 539 (9). The catalytic domain by itself is a monomeric protein and does not support yeast cell growth, whereas the biologically active triphosphatase has a homodimeric quaternary structure (9,10). Mutational disruption of the Cet1 homodimer interface is uniquely deleterious in vivo when the yeast RNA triphosphatase functions in concert with the endogenous yeast guanylyltransferase Ceg1. Lethal or severe temperature-sensitive (ts) growth phenotypes elicited by mutations of the Cet1 homodimer interface are suppressed by fusion of the mutated triphosphatase to the guanylyltransferase domain of mammalian capping enzyme (11).
Genetic evidence indicates that the Cet1-Ceg1 interaction is important. ceg1-ts mutations are suppressed in an allele-specific manner by overexpression of CET1 (2,8). In turn, cet1-ts mutations can be suppressed by overexpression of CEG1 (9). The guanylyltransferase-binding and guanylyltransferase-stimulation functions of Cet1 localize to a 21-amino acid segment from residues 239 to 259 (3). The guanylyltransferase-binding domain is located on the protein surface (10) and is conserved in the Candida albicans RNA triphosphatase CaCet1 (3), but not in the RNA triphosphatase Pct1 from the fission yeast Schizosaccharomyces pombe (12). Alanine-cluster mutations of a WAQKW motif within the Ceg1-binding domain of Cet1 abolish guanylyltransferase binding in vitro and Cet1 function in vivo, but do not affect the triphosphatase enzymatic activity (3,9,13).
Here we use biochemical and genetic methods to address three key questions concerning the Cet1-Ceg1 interaction. (i) What is the basis for stimulation of Ceg1 guanylyltransferase activity by Cet1? Specifically, is the stimulation attributable to allosteric interactions or to other factors, such as protein stabilization? (ii) Is the stimulation of Ceg1 by Cet1 important in vivo? (iii) Is there a discrete triphosphatase-binding site on yeast guanylyltransferase?
Guanylyltransferase Assay-Guanylyltransferase activity was assayed by the formation of the covalent enzyme-GMP intermediate. Reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 8.0), 5 mM MgCl 2 , 0.17 M [␣-32 P]GTP, and enzyme as specified were incubated for 10 min at 22°C. The reaction was halted by adding SDS to 1% final concentration. The samples were analyzed by SDS-PAGE. The guanylyltransferase-32 P[GMP] complex was resolved and visualized by autoradiography of the dried gel and quantitated by scanning the gel with a FUJIX BAS2500 PhosphorImager.
ATPase Assay-Reaction mixtures (20 l) containing 50 mM Tris-HCl (pH 7.5), 5 mM dithiothreitol, 2 mM MnCl 2 , 1 mM [␥-32 P]ATP, and enzyme were incubated for 15 min at 30°C. The reactions were quenched by adding 2 l of 5 M formic acid. Aliquots (2 l) of the mixtures were applied to a polyethyleneimine-cellulose TLC plate, which was developed with 1 M formic acid, 0.5 M LiCl. The extent of 32 P i release was quantitated by scanning the chromatogram with a PhosphorImager.
Thermal Inactivation of Cet1-Purified Cet1(201-549) (2.2 M) was incubated for 15 min on ice either alone or with 12 M purified Ceg1. Aliquots (5 l) of the mixtures were preincubated for 10 min at 30, 35, 40, 45, 50, or 55°C and then quenched on ice. Control aliquots were kept on ice throughout the pretreatment. An aliquot (2 l) of each sample was then assayed at 30°C for manganese-dependent ATP hydrolysis. The extent of ATP hydrolysis by preheated enzyme was normalized to that of the unheated control enzyme (14 nmol of 32 P release; defined as 100%). The normalized activities are plotted as a function of preincubation temperature in Fig. 1F.
Yeast 2 plasmids encoding chimeric capping enzymes composed of mammalian RNA triphosphatase Mce1(1-210) fused to either Ceg1 or Pce1 were constructed as follows. The Mce1(1-210) reading frame was subjected to two-stage PCR amplification using primers designed to eliminate an internal NcoI site (without altering the protein sequence), to introduce an NcoI site at the start codon, and to change the Glu 211 codon to alanine and introduce an NcoI restriction site at the codons for the Ser 210 -Ala 211 dipeptide. The PCR product was digested with NcoI and inserted into the NcoI site of pYX232-CEG1 (2 TRP1) or pYX232-PCE1 (2 TRP1) to yield plasmids encoding the in-frame fusion proteins Mce1(1-210)-Ceg1 and Mce1(1-210)-Pce1. Expression of the chimeric mammalian-fungal capping enzymes in these plasmids is under the control of the yeast TPI1 promoter. The MCE1(1-210) inserts were sequenced completely to confirm the in-frame fusion to CEG1 or PCE1 and to exclude the introduction of unwanted coding changes during amplification and cloning.
Yeast CEN plasmids containing the chimeric genes were constructed by excising from the 2 plasmids an AatI-NheI fragment containing the fusion gene and transferring the fragment into pYX132 (CEN TRP1). Expression of the fusion genes in the CEN plasmids is also under control of the TPI1 promoter.
Expression and Purification of GST-Ceg1 Fusion Proteins-The ORF encoding full-length Ceg1(1-459) fused to an NH 2 -terminal His 10 tag leader peptide was inserted into the pGEX-KG vector between the NcoI and BamHI sites. The resulting plasmid encodes a glutathione S-transferase (GST)-His 10 Ceg1(1-549) fusion protein.
Plasmids for expression of Ceg1 domains and domain fragments fused to GST were engineered as follows. An ORF encoding the NH 2terminal domain Ceg1(1-244) was amplified by PCR using an antisense primer that introduced a translation stop codon in lieu of the codon for Leu 245 and a BamHI site 3Ј of the stop codon. An ORF encoding the COOH-terminal domain Ceg1(245-459) was amplified by PCR using a sense primer that changed the Ser 243 -Leu 244 codons to Met 243 -Gly 244 and introduced an NcoI site at the new start codon. An ORF encoding the fragment Ceg1(245-360) was amplified by PCR using an antisense primer that introduced a translation stop codon in place of the codon for Gly 361 and a BamHI site 3Ј of the stop codon. An ORF encoding Ceg1(361-459) was amplified by PCR using a sensestrand primer that introduced an NcoI site and a methionine codon in lieu of the codon for Thr 360 . The PCR products were digested with NcoI and BamHI and then inserted into pGEX-KG to generate plasmids encoding fusion proteins GST-Ceg1(1-244), GST-Ceg1(245-459), GST-Ceg1(245-360), and GST-Ceg1(361-459).
Protein-Protein Interaction Assays-Purified GST and the GST-Ceg1 fusion proteins (either 10 -15 g of the proteins specified in in  Fig. 6B) in 50 l of binding buffer. After incubation for 1 h on ice, the beads were concentrated by microcentrifugation and the supernatant was withdrawn. The beads were resuspended in 1 ml of binding buffer and subjected to three rounds of concentration and washing. After the third wash, the protein bound to the beads was eluted by beads were resuspended in 50 l of binding buffer containing 10 mM glutathione. Aliquots (20 l) of the input Cet1(201-549) sample, the first supernatant fraction (containing the "free" protein), and the bead-bound fraction were mixed with 5 l of SDS sample buffer, heated at 90°C for 3 min, and then analyzed by SDS-PAGE. Polypeptides were visualized by staining the gel with Coomassie Blue dye.
Mutagenesis of Yeast Guanylyltransferase-Alanine substitution mutations were introduced into the CEG1 gene by using the PCR-based two-stage overlap extension method as described (16). The mutated CEG1 genes were inserted into the yeast plasmid pGYCE-358 (CEN TRP1) (17), where expression of the CEG1 gene is under the control of the natural CEG1 promoter. The inserts were sequenced completely to confirm the presence of the programmed changes and to exclude the introduction of unwanted coding changes during amplification and cloning. The in vivo activity of the protein encoded by the mutated CEG1 alleles in supporting the growth of a ceg1⌬ strain was tested by plasmid shuffle as described previously (16,17).

Cet1
Binding to Ceg1 Stabilizes Ceg1 against Thermal Inactivation-Prior studies showed that the binding of full-length Cet1 or the truncated enzyme Cet1(201-549) to Ceg1 stimulated the guanylyltransferase activity of Ceg1 by an order of magnitude when guanylyltransferase activity was measured at 37°C (2). An equivalent stimulation was elicited by a synthetic peptide Cet1(232-265), which binds quantitatively and with high affinity to Ceg1 in vitro (3). We show here that the purified recombinant Ceg1 protein is extremely thermolabile in vitro (Fig. 1A). Guanylyltransferase activity was abolished by preincubation of the protein for 10 min at 35°C or higher and reduced by a factor of 5 after 10 min at 30°C. Thus, Ceg1 is rapidly inactivated at physiological temperatures.
The instructive finding was that mixture of Ceg1 with recombinant Cet1(201-549) protein effected a dramatic shift to the right in the guanylyltransferase thermal inactivation profile, such that the Cet1(201-549)-Ceg1 complex was impervious to preincubation for 10 min at 30°C and retained 50% of the original activity after incubation at 35°C (Fig. 1A). The synthetic peptide Cet1(232-265) also had a profound protective effect against thermal inactivation of Ceg1, whereby the Cet1(232-265)-Ceg1 complex retained 70% of the original activity after incubation for 10 min at 35°C (Fig. 1B). A synthetic Cet1(232-265) peptide containing a quadruple alanine-cluster mutation of the WAQKW motif had no salutary effect on the stability of Ceg1 (Fig. 1B). The 4xAla mutant peptide does not bind Ceg1 and does not stimulate its activity at 37°C (3). We conclude that the stimulation of Ceg1 by Cet1 is attributable to protein stabilization.
In contrast to the instability of Ceg1, we found that the triphosphatase activity of Cet1(201-549) was stable to preincubation at 30 -35°C and its thermal inactivation profile was unaffected by prior binding to the yeast guanylyltransferase Ceg1 (Fig. 1F). Thus, the protective effects of Cet1-Ceg1 complex formation on enzyme stability in vitro are not reciprocal.
To test whether the inactivation of Ceg1 at physiological temperatures can be reversed ex post facto by Cet1, we varied the order of addition of the protective Cet1(232-265) peptide with respect to the heat treatment. Mixture of Ceg1 with the Cet1 peptide on ice before heating the protein for 10 min at 35°C resulted in 11-fold higher guanylyltransferase activity compared with heat-treated Ceg1 that had not been exposed to Cet1(232-265). Yet, mixing the already heat-treated Ceg1 with the Cet1 peptide on ice had no restorative effect on the guanylyltransferase activity (Fig. 2). Thus, Cet1 could not reverse the inactivation of Ceg1, implying that Cet1 is not serving as a chaperone that promotes refolding of Ceg1 or resumption of an active conformation. Rather, Cet1 binding to Ceg1 stabilizes the guanylyltransferase only prospectively against thermal inactivation.
The Stabilizing Effect of Cet1 on Fungal Guanylyltransferase Is Species-specific-The guanylyltransferase-binding peptide domain is conserved in C. albicans RNA triphosphatase CaCet1 and we have shown that the C. albicans guanylyltransferase Cgt1 binds avidly to the Cet1(232-265) peptide in vitro (3). To gauge if stabilization of guanylyltransferase by triphosphatase is a general phenomenon in fungal systems, we examined the thermal stability of Cgt1 in the absence and presence of Cet1(232-265). Two notable findings emerged. First, Candida guanylyltransferase was more stable than the Saccharomyces enzyme. Cgt1 retained 40% of the original activity after 10 min at 35°C (Fig. 1C) compared with 2% activity for Ceg1 (Fig. 1A). Second, the Candida guanylyltransferase was protected from thermal inactivation by the Cet1(232-265) peptide. This was especially clear during treatment at 40°C, whereby the Cet1(232-265)-Cgt1 complex retained 55% activity compared with 5% for Cgt1 alone (Fig. 1C). In contrast, the S. pombe RNA guanylyltransferase Pce1 (18) was clearly more thermostable than either Ceg1 or Cgt1. Pce1 was unaffected by preincubation at up to 35°C (which abolished Ceg1 activity), retained 70% of basal activity after heating at 40°C (compared with 5% activity for Cgt1) and 25% activity after treatment a 45°C. Moreover, its thermal inactivation profile was unaffected by the Cet1(232-265) peptide (Fig. 1D). We found that the Cet1(232-265) peptide interacts very weakly with Pce1 in vitro as gauged by peptide-affinity chromatography (data not shown) (3). These experiments show that the stabilizing effect of RNA triphosphatase on guanylyltransferase is conserved in two species of budding yeast, but not in fission yeast.
The Interaction of Cet1 with Ceg1 Stabilizes Ceg1 in Vivo-The catalytic domain Cet1(276 -549) lacks the high-affinity guanylyltransferase-binding site and does not interact with Ceg1 in vitro (9). Cet1(269 -549) also lacks the Ceg1-binding site. Neither CET1(269 -549) nor CET1(276 -549) was able to complement growth of S. cerevisiae cet1⌬ cells, even when the truncated enzymes were expressed in high gene dosage under the control of the strong constitutive yeast TPI1 promoter (Ref. 9 and data not shown). Remarkably, the in vivo function of Cet1(276 -549) was restored when it was fused to the guanylyltransferase domain of the mammalian capping enzyme (9). We proposed that the mammalian domain, Mce1(211-597), which binds avidly to the phosphorylated CTD (15,20), can act as a vehicle to deliver the fused RNA triphosphatase to the RNA polymerase II elongation complex (9,21). Also, because Mce1(211-597) is thermostable (unlike Ceg1; see Fig. 1), the chimeric capping enzyme likely bypasses the need for the Ceg1stabilization function of the 232-259 domain of Cet1.
We reasoned that if the only defect of the truncated Cet1 proteins missing the Ceg1-binding site was a lack of targeting to the CTD, then fusing them in cis to Ceg1 would restore their function in vivo, just as the fusion to Mce1(211-597) does. However, if the Cet1(232-259) peptide is as important for stabilizing Ceg1 in vivo as it is in vitro, then the chimeric Cet1-Ceg1 proteins lacking that domain would either not function in vivo or else would evince a temperature-sensitive growth defect. To test if Ceg1 would tolerate a large fusion peptide at its N terminus, we expressed a chimeric yeast-yeast capping enzyme in which the biologically active truncated triphosphatase Cet1(201-547) was linked to full-length Ceg1. The CET1(201-547)-CEG1 fusion complemented the growth of cet1⌬ ceg1⌬ cells in a plasmid shuffle assay when the chimeric gene was expressed on either a 2 plasmid or a CEN plasmid (Fig. 3). Moreover, 2 CET1(201-547)-CEG1 and CEN CET1(201-547)-CEG1 cells grew as well as wild type CET1 CEG1 cells on YPD agar at 22, 30, and 37°C (scored as ϩϩϩ growth in Fig. 3). Thus, an NH 2 -terminal fusion per se did not perturb the in vivo activity of Ceg1.
cet1⌬ ceg1⌬ cells expressing the Cet1(269 -547)-Ceg1 chimera at high gene dosage grew as well as wild-type cells on YPD at 30°C (ϩϩϩ), but failed to grow at all at 37°C (ts) (Fig.  3). Cells expressing Cet1(269 -547)-Ceg1 from a CEN plasmid formed pinpoint colonies on YPD at 30°C (scored as ϩ growth) and failed to grow at 37°C (ts). Thus, the fusion to Ceg1 bypassed the requirement for a high-affinity Ceg1-binding site on Cet1, but conferred a profound ts growth defect that was exacerbated when the chimera was expressed at low gene dosage. Given that cells expressing a Cet1(269 -547)-Pce1 fusion grew well at 37°C (see below), we attribute the ts growth to a defect in the Ceg1 component of the chimeric capping enzyme.
cet1⌬ ceg1⌬ cells expressing the more extensively truncated Cet1(276 -547)-Ceg1 fusion protein from a 2 plasmid formed smaller colonies than wild-type cells on YPD at 30°C (ϩϩ  cet1⌬ ceg1⌬ cells were transformed with the 2 TRP1 or CEN TRP1 plasmids encoding the indicated chimeric capping enzymes under the control of the TPI1 promoter. Trp ϩ isolates were selected and then streaked on agar plates containing 0.75 mg/ml 5-FOA. Chimeras that failed to yield colonies on 5-FOA after 7 days of incubation at 18, 22, and 30°C were scored as lethal (Ϫ). Individual FOA-resistant colonies were picked and patched on YPD agar. Two isolates of each mutant were then streaked on YPD agar at 30 and 37°C. Growth was assessed after 4 days of incubation as follows: ϩϩϩ indicates colony size indistinguishable from wild-type; ϩϩ indicates colony size smaller than the wild-type; ϩ indicates pinpoint colonies. Symbols in parentheses are included in cases where the growth phenotype at 37°C differed from that at 30°C. Temperature-sensitive (ts) mutants were those that did not form colonies at 37°C. growth) and did not form colonies at 37°C (ts). Expression of Cet1(276 -547)-Ceg1 from a CEN plasmid resulted in failure to recover viable colonies during the 5-FOA selection step of the plasmid shuffle conducted at either 18, 22, or 30°C. Thus, the CEN CET1(276 -547)-CEG1 allele was lethal in single copy (scored as Ϫ in Fig. 3). Comparing the profound ts and lethal phenotypes of the Cet1(276 -547) fusion to Ceg1 to the normal function in vivo of the Cet1(276 -547) fusion to mammalian guanylyltransferase (9) again prompts the conclusion that it is the Ceg1 component that is thermolabile or defective in vivo. The more severe phenotypes observed for Cet1(276 -547)-Ceg1 fusions versus Ceg1(269 -547)-Ceg1 fusions may reflect the fact that the shorter derivative of Cet1 has lost the ability to homodimerize (9) because resides Phe 272 and Leu 273 are key components of the dimer interface (10,11). This theme is underscored by the effects of fusing the monomeric mammalian RNA triphosphatase domain Mce1(1-210) to Ceg1 (Fig. 3), whereby expression of the Mce1(1-210)-Ceg1 chimera from a 2 plasmid resulted in slow growth at 30°C (ϩϩ) and no growth at 37°C (ts), while expression of Mce1(1-210)-Ceg1 from a CEN plasmid was lethal. Thus, the identical growth defects were elicited by fusing Ceg1 to catalytically active monomeric triphosphatases from yeast or mammals. This suggests that indirect dimerization of Ceg1 (via the homodimeric yeast triphosphatase) enhances Ceg1 function at 37°C in vivo.
We extended this analysis to chimeric capping enzymes consisting of an NH 2 -terminal truncated Cet1 polypeptide linked in cis to the full-length S. pombe guanylyltransferase Pce1. The chimeric enzyme Cet1(201-547)-Pce1 complemented growth of cet1⌬ ceg1⌬ cells when provided on either 2 or CEN plasmids and CET1(201-547)-PCE1 cells grew as well as wild-type yeast at 22, 30, and 37°C (ϩϩϩ in Fig. 3). Thus, Pce1 function was not compromised by the NH 2 -terminal fusion. The salient finding was that expression of Cet1(269 -547)-Pce1 from a 2 or a CEN plasmid fully complemented growth of cet1⌬ ceg1⌬ and that the fusion to the S. pombe guanylyltransferase was functional in vivo at 37°C (ϩϩϩ growth), unlike the fusion of Cet1(269 -547) to Ceg1, which was growth impaired when expressed in single copy and was defective in vivo at 37°C whether expressed at low or high gene dosage (Fig. 3). Thus, when the essential Ceg1-binding domain of the triphosphatase is eliminated, the activity of the chimeric capping enzymes in vivo reflects faithfully the thermal stability of the guanylyltransferase component in vitro (i.e. unstable for Ceg1 versus stable for Pce1).
Stabilization of Ceg1 by Cet1 Versus Targeting of Cet1 by Ceg1-The preceding analysis of chimeric capping enzymes (Fig. 3) provides evidence that an essential function of the Cet1-Ceg1 interaction in vivo is the stabilization of the inherently labile guanylyltransferase activity of Ceg1. Yet the experiments do not probe the putative role of the Cet1-Ceg1 interaction in helping to target the triphosphatase to the RNA polymerase II transcription complex. To approach this issue, we expressed RNA triphosphatase and RNA guanylyltransferase in trans from separate CEN plasmids, marked either with TRP1 (for the triphosphatase) or ADE2 (for the guanylyltransferase). Expression of both components was under the control of the strong constitutive yeast TPI1 promoter, the same promoter used to drive expression of the fused capping enzymes. The coexpression of the biologically active domain Cet1(201-549) with Ceg1 supported normal growth of cet1⌬ ceg1⌬ cells at 30 and 37°C, as expected (Fig. 4). In contrast, cet1⌬ ceg1⌬ cells co-transformed with the TPI1-CET1(269 -549) and TPI1-CEG1 alleles yielded few viable FOA-resistant colonies, which then grew very slowly on YPD agar at 30°C (ϩ growth) and failed to grow on YPD at 37°C (Fig. 4). Thus, the slow growth and ts phenotypes of cells expressing Cet1(269 -549) and Ceg1 as separate proteins were the same as those observed for the Cet1(269 -549)-Ceg1 fusion protein (Fig. 3). Note that the TPI-CET1(269 -549) allele, which encodes a protein that lacks the Ceg1-binding domain, is lethal in single copy or in high copy in yeast cells that express CEG1 in single copy under the control of its natural promoter. Thus, we surmise that the gain of function of TPI1-CET1(269 -549) in the TPI1-CEG1 background is attributable to overexpression of the guanylyltransferase, which, although it is thermolabile as a consequence of the loss of the stabilizing influence of Cet1 binding, is able via overexpression to attain a threshold level of active guanylyltransferase to support very slow cell growth at 30°C.
The instructive finding was that the TPI1-CET1(269 -549) allele was fully functional at both 30 and 37°C when present in trans with TPI1-PCE1 (Fig. 4). TPI1-CET1(269 -549) was also fully functional at 30 and 37°C in the presence of a separate CEN plasmid expressing the mammalian guanylyltransferase gene TPI1-MCE1(211-597) (not shown). These results undescore two key points: (i) that the ts growth phenotype elicited by removal of the Ceg1-binding domain of Cet1 correlates with the inherent thermolability of the coexisting guanylyltrans- FIG. 4. A thermostable guanylyltransferase bypasses the in vivo requirement for the Ceg1-binding domain of Cet1. cet1⌬ ceg1⌬ cells were co-transformed with CEN TRP1 plasmids encoding the indicated RNA triphosphatases under control of the TPI1 promoter and CEN ADE2 plasmids encoding either CEG1 or PCE1 under the control of the TPI1 promoter. Trp ϩ Ade ϩ isolates were selected and then streaked on agar plates containing 0.75 mg/ml 5-FOA. Strains that failed to yield colonies on 5-FOA after 7 days of incubation at 18, 22, and 30°C were scored as lethal (Ϫ). Individual FOA-resistant colonies were picked and patched on YPD agar. Two isolates of each mutant were then streaked on YPD agar at 30 and 37°C. Growth was assessed after 4 days of incubation as follows: ϩϩϩ indicates colony size indistinguishable from wild-type; ϩϩ indicates colony size smaller than the wild-type; ϩ indicates pinpoint colonies. Symbols in parentheses are included in cases where the growth phenotype at 37°C differed from that at 30°C. Temperature-sensitive (ts) mutants were those that did not formed colonies at 37°C. ferase component even when the two enzymes are not linked physically and (ii) cells overexpressing Cet1(269 -549) grow normally only when the need for Ceg1 stabilization is bypassed. The implication of the latter point is that the triphosphatase Cet1(269 -549) can, when overexpressed, gain access to pre-mRNAs without the benefit of its guanylyltransferasebinding domain. Similar gain of function results have been obtained by Takase et al. (13) for the closely related deletion mutant Cet1(265-549) and by Schwer et al. (22) for C. albicans RNA triphosphatase mutants deleted in the conserved guanylyltransferase-binding domain.
A notable finding with respect to the issue of capping enzyme targeting was that separate expression of the mammalian RNA triphosphatase Mce1(1-210) and the S. pombe guanylyltransferase Pce1 failed to sustain cell growth (Fig. 4), whereas expression of the Mce1(1-210)-Pce1 fusion protein at the same gene dosage and driven by the same promoter resulted in normal growth of cet1⌬ ceg1⌬ cells (Fig. 3). We surmise that the Mce1(1-210) domain is not properly targeted in vivo in yeast unless it is chaperoned to the pre-mRNA by fusion to a themostable guanylyltransferase, which can be either Pce1 or Mce1(211-597). The fission yeast and mammalian guanylyltransferases bind directly and specifically to the phosphorylated CTD in vitro (4,15,20,23).
Delineation of a Cet1-binding Site on Ceg1-RNA guanylyltransferases consist of two structural domains: a larger NH 2terminal domain (domain 1) that contains the GMP binding pocket and a smaller COOH-terminal domain (domain 2) that interacts with the ␤ and ␥ phosphates of GTP (24). The active site of guanylyltransferase is composed of six conserved motifs (I, III, IIIa, IV, V, and VI) that define the covalent nucleotidyl transferase superfamily (16,24) (Fig. 5A). Domain 1 includes motifs I, III, IIIa, IV, and the proximal half of motif V; domain 2 includes the distal part of motif V as well as motif VI.
Initial studies of the Ceg1 side of the yeast triphosphataseguanylyltransferase interface entailed proteolytic footprinting of Ceg1 in the presence and absence of Cet1 (3). The principal tryptic cleavage sites in Ceg1, located within domain 2 at Arg 304 and Lys 306 (denoted by arrowheads in Fig. 5A), were protected from trypsin digestion by Cet1(201-549), whereas secondary tryptic sites located close to the NH 2 terminus in domain 1 were not shielded. The Cet1(232-265) peptide also protected Ceg1 from proteolysis by trypsin, whereas the Cet1(232-265)4xAla mutant peptide did not (3). These results suggested that at least part of a Cet1-binding site on Ceg1 is located within domain 2, but they did not address whether Ceg1 contains a discrete triphosphatase-binding epitope or whether the domain 1 is required for interaction with Cet1.
Here we used affinity chromatography to address these issues. Fusion proteins containing GST linked to full-length Ceg1 (aa 1-459), Ceg1 domain 1 (aa 1-244), or Ceg1 domain 2 (aa 245-459) were produced in bacteria and purified. The GST fusions proteins were immobilized on glutathione-Sepharose beads, which were then mixed with purified Cet1(201-549) protein (shown in Fig. 6A, lane L). The material that did not bind to the beads (free fraction, F) was analyzed by SDS-PAGE along with the material that had bound to the resin and was subsequently stripped off with glutathione and SDS (bound fraction, B). The Cet1(201-549) protein was retained quantitatively on the beads containing GST-Ceg1(1-459), but not at all on beads containing GST alone or GST-Ceg1(1-244) (Fig. 6A). A majority of the input Cet1(201-549) did adsorb to the resin when the affinity ligand was GST-Ceg1(245-459). We conclude that domain 2 comprises an autonomous Cet1-binding module. The fine structure of the binding site within domain 2 is likely to be complex, insofar as the splitting of domain 2 fusion into  (16, 18; denoted by ! above the sequence) and the crystal structure of the Chlorella virus enzyme (24). The secondary structure elements of the Chlorella virus guanylyltransferase are shown below the amino acid sequence; ␤ strands are depicted as arrows; and ␣ helices as bars. Ceg1 residues between motifs V and VI that were mutated in the present study are highlighted in shaded boxes. The boxes also embrace side chain identity or similarity in the other aligned guanylyltransferase sequences at the positions selected for mutagenesis. B, ceg1⌬ cells were transformed with CEN TRP1 plasmids containing the wild-type and mutant alleles of CEG1 under the control of the CEG1 promoter. A control transformation was performed using the TRP1 vector. Trp ϩ isolates were selected and then streaked on agar plates containing 0.75 mg/ml 5-FOA. Only one mutant, L347A N348A, failed to yield colonies on 5-FOA after 7 days of incubation at 18, 22, and 30°C; this mutant was scored as lethal (Ϫ). Individual FOA-resistant colonies were picked and patched on YPD agar. Two isolates of each mutant were then streaked on YPD agar at 30 and 37°C. Growth was assessed as follows: ϩϩϩ indicates colony size indistinguishable from wild-type; ϩϩ indicates colony size smaller than the wild-type; ϩ indicates pinpoint colonies. Temperature-sensitive (ts) mutants were those that did not form colonies at 37°C after 4 days.

Effects of Domain 2 Mutations on Ceg1 Function in Vivo-
The trypsin-sensitive sites of Ceg1, Arg 304 and Lys 306 , that are shielded from proteolysis when Cet1 is bound to Ceg1, are located between nucleotidyl transferase motifs V and VI in domain 2 (Fig. 5A). An alignment of the sequences of the S. cerevisiae, C. albicans, S. pombe, and mammalian guanylyltransferases based on the crystal structure of the Chlorella virus guanylyltransferase (24) reveals that this intervening region is variable in length and relatively poorly conserved. We predict from the Chlorella virus enzyme structure that this segment comprises a surface loop. To gauge whether this loop and sequences flanking it might include functionally important components of the triphosphatase-binding site of Ceg1, we performed alanine-cluster mutagenesis of pairs of vicinal amino acids and also introduced single alanine substitutions of four positions, including the two basic residues at the tryptic sites. A total of 38 residues (7% of the complete Ceg1 polypeptide) were changed to alanine in this analysis (Fig. 5B). The CEG1-Ala genes were cloned into a CEN vector under the control of the natural CEG1 promoter and then tested by plasmid shuffle for their ability to complement a ceg1⌬ mutant. The mutational effects are tabulated in Fig. 5B.
Only one mutant allele, L347A N348A, failed to support the growth of ceg1⌬ cells on 5-FOA at all temperatures tested (18,22, and 30°C); thus, this mutation was lethal in vivo. The 347-348 dipeptide is predicted to comprise a short turn between two ␤ strands in Ceg1 and other cellular capping enzymes; the ␤ strands are connected by a loop in the Chlorella virus enzyme (Fig. 5A). The strand immediately distal to the essential dipeptide corresponds to motif Vc described by Wang et al. (16). This motif is located just upstream of the essential nucleotidyl transferase motif VI in all cellular capping enzymes and in the Chlorella virus guanylyltransferase (Fig. 5A). Mutation of the conserved motif V glutamate (Glu 353 ) had no effect on the in vivo activity of Ceg1 (Fig. 5B). The double mutant E353A C354A grew normally at 30°C, but colony size at 37°C was smaller than wild-type (ϩϩ). Several groups had previously isolated C354Y mutants in a screen for ceg1-ts alleles (25)(26)(27). We surmise in light of the present results that the tight growth defect of C354Y at 37°C is caused not by the loss of the cysteine functional group (alanine being better tolerated), but rather by steric effects of introducing a bulky tyrosine at this position. The observation that the ts growth defect of C354Y was suppressed by overexpression of Cet1 (8) is consistent with a perturbation of the (nearby) triphosphatase-binding site by a Tyr 354 side chain.
Alanine cluster mutations L343A E344A and Q345A P346A in the ␤ strand immediately proximal to the essential 347-348 dipeptide had no effect on cell growth. Thus, we infer that the ts growth phenotype noted previously for the P346L mutant, which can be suppressed by overexpression of Cet1 (8,27), reflects the perturbation of a triphosphatase-binding site by the bulky leucine rather than a specific contribution of proline to Cet1 binding (alanine being able to function in place of Pro 346 ). Here, we noted ts phenotypes for the neighboring mutants, L340A K341A and W337A Q338A, located just upstream at the takeoff point of the proposed surface loop of the cellular guanylyltransferases (Fig. 5). L340A K341A cells failed to grow at 37°C; however, growth at the restrictive temperature was restored by provision of CET1(201-549) on a multicopy (2) plasmid (Fig. 7). These effects of simple side chain removal imply either that the Leu 340 /Lys 341 dipeptide is a component of the Cet1-binding site or that the loss of these side chains affects the binding site indirectly via a local conformational change.
The Ceg1 segment immediately upstream is a strongly hy- vector. Ura ϩ isolates were selected and streaked on SC agar medium lacking uracil. The plate was photographed after incubation for 4 days at 37°C. normally contingent on CTD phosphorylation. We have observed that yeast cells expressing Mce1(211-597) as their only source of guanylyltransferase display a slow growth phenotype when Cet1 is expressed from its natural promoter and that normal growth is restored by overexpressing Cet1 (2,3). Thus, we suspect that Cet1-Ceg1 complex formation does facilitate the function of the triphosphatase, presumably via assisting in targeting it to the CTD, even if the requirement for CTD targeting via Ceg1 is not absolute. The triphosphatase targeting function is more clearly established for the mammalian RNA triphosphatase domain, which does not sustain yeast cell growth unless fused to a guanylyltransferase (Figs. 3 and 4), either the natural mammalian guanylyltransferase domain or the heterologous guanylyltransferase of S. pombe.
Finally, biochemical and genetic experiments presented here implicate domain 2 of Ceg1 in Cet1 binding and show that domain 1 is not required for Cet1 binding in vitro. Two segments within domain 2, 287 KPVSLYVW 295 and 337 WQNLKN-LEQPLN 348 , emerge as likely constituents of a Cet1-binding site on Ceg1. Finer analysis of the Cet1-Ceg1 interface now hinges on crystallization of the yeast guanylyltransferase complexed either to the native yeast triphosphatase or to the Ceg1binding peptide of Cet1.