Catalytic Activity of Vaccinia mRNA Capping Enzyme Subunits Coexpressed in Escherichia coZi*

assays performed as described under “Experimental Procedures.” Enzyme was assayed at three dilutions to establish linearity. The values reported for SP5PW pool and glycerol pool fractions represent the sum of the activity and protein present in the individual fractions (assayed separately) that comprise the

These results indicate that the Dl and D12 gene products are together sufficient to catalyze all three enzymatic steps in cap synthesis.
A model for the domain structure of this enzyme is proposed.
Capping of the 5' terminus of vaccinia virus mRNA occurs by a series of three reactions catalyzed, respectively, by RNA triphosphatase, RNA guanylyltransferase, and RNA (guanine-7)-methyltransferase (1). These three enzymes are components of a heterodimeric capping enzyme complex containing subunit polypeptides of M, 95,000 and M, 31,000 (2-4). In addition to its role in 5' RNA modification, the vaccinia * This work was supported by Grant GM 42498-01 from the National Institutes of Health and by Junior Faculty Research Award JFRA-2'74 from the American Cancer Society.
The costs of publication of this article were defrayed in part by the payment of page charges.
This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
capping enzyme is involved in correct 3' end formation of vaccinia early mRNAs by acting as a transcription termination factor for the vaccinia DNA-dependent RNA polymerase (5). Attempts to analyze the domain structure of vaccinia capping enzyme have been frustrated previously by the inability to dissociate the subunits with preservation of catalytic activity. Recently, the identification of the viral genes encoding the M, 95,000 and M, 31,000 polypeptides has made feasible a molecular genetic approach to the problem. The gene encoding the large enzyme subunit has been mapped to the Dl ORF (open reading frame) of the viral genome (6, 7). The large subunit participates in the transguanylation step in the capping pathway by forming an enzyme-guanylate intermediate (8). The intermediate consists of a GMP residue attached covalently to the M, 95,000 enzyme subunit via a phosphoamide bond to the t-amino group of a single lysine (8-10). It is not clear, though, whether the large subunit is sufficient to catalyze GMP transfer, or if participation of the small subunit is required as well. The gene encoding the small subunit has been mapped to the D12 ORF of the vaccinia genome (ll), yet no biochemical function has been assigned to this polypeptide.
The present report demonstrates the synthesis of active vaccinia virus mRNA capping enzyme in Escherichia coli as a consequence of coexpression of the Dl and D12 ORFs. The enzyme has been purified 1000-fold and shown to catalyze all three steps in cap formation. A separate paper (17) describes the heterologous expression of the large subunit alone and the assignment of specific enzymatic properties to that polypeptide per se. Construction of PET-Dl and PET-Dl/ Dl2 is described under "Experimental Procedures." The figure depicts the salient functional elements of the plasmids (not drawn to scale). The direction of transcription from the T7 promoter is indicated by the arrows.

Construction of Plasmids
relative amounts of the two induced polypeptides was influenced by the bacterial host strain, eg. HMS174 (a strain that contains the lon and ompTproteases that are lacking in BL21) accumulated proportionately more M, 39,000 species than M, 95,000 protein upon phage infection.
In the case of cells bearing PET-D12, XCE6 infection resulted in the time-dependent appearance of a single M, 31,000 polypeptide in either BL21 or HMS174, although the extent of accumulation appeared higher in HMS174.' When T7 RNA polymerase was provided instead by IPTG induction of E. coli BL21(DE3) carryingpET-Dl or PET-D12, the pattern of inducible protein accumulation was the same as that found with phage infection. Cells bearing plasmid pETDl/DlB were induced to accumulate both the M, 95,000 and the M, 31,000 polypeptides when T7 RNA polymerase was provided by either method. All subsequent studies described in this and the succeeding report (17) involve coexpression of Dl and D12 ORFs from PET-Dl/DlS or expression of Dl ORF alone from PET-Dl.

Solubility
and Activity of Coexpressed Subunits-Electrophoretic analysis of soluble and insoluble fractions derived from lysates of XCEG-infected BL2lpETDl/D12 cells or IPTG-induced BL21(DE3)pET-Dl/D12 cells revealed that the Dl and D12 polypeptides were detectable only in the insoluble pellet (not shown). The M, 95,000 and M, 31,000 proteins could be solubilized readily in either 8 M urea or 5 M guanidine HCl; however, both proteins precipitated upon removal of either denaturant by dialysis. Other manipulations, such as phage infection or IPTG induction at lowered temperatures (either 30 "C! or 25 "C), failed to enhance the solubility of the expressed proteins. Guanylyltransferase activity was assayed in crude extracts by the formation of covalent protein-GMP complex, this being a highly sensitive and specific method for detection of capping enzyme. Formation of a 32P-labeled M, 95,000 polypeptide was mediated by insoluble protein from cells induced to express the Dl ORF, but not by the soluble protein fraction. Thus, the ability to bind GMP covalently was retained in some part by the large subunit even in insoluble form (see accompanying article (17)). Crude extracts of bacteria that did not carry the plasmid-borne vaccinia Dl ORF did not catalyze formation of the 32P-labeled M, 95,000 polypeptide (not shown). Additional studies showed that BL21(DE3)pET-Dl/D12 cells accumulated appreciable amounts of soluble guanylyltransferase activity without IPTG induction. This can be attributed to the basal level of T7 RNA polymerase expression * M. Spector and S. Shuman, unpublished work. from the lacUV5 promoter in the DE3 prophage. This permitted the purification of guanylyltransferase from the soluble fraction, as described under "Experimental Procedures" and summarized below.

Enzyme Purification and Characterization: Coexpressed
Subunits-Incubation of the polymin P supernatant fraction (containing soluble protein depleted of nucleic acid) with [a-32P]GTP and MgCl* resulted in the formation of a M, 95,000 guanylylated polypeptide, as well as two minor labeled species of lower molecular weight (Fig. 2). The electrophoretic mobility of the M, 95,000 labeled protein was identical with that of the enzyme-GMP intermediate formed by capping enzyme purified from vaccinia virions (not shown). The 20-40% ammonium sulfate fraction derived from the polymin P supernatant contained most of the guanylyltransferase activity. This fraction catalyzed formation of a major M, 95,000 protein-GMP complex and a minor labeled polypeptide of M, 60,000 (Fig. 2). Residual M, 95,000 EpG-forming activity was recovered in the 40-60% and 60-80% ammonium sulfate fractions. The apparent resolution of the more rapidly migrating labeled polypeptides during ammonium sulfate fractionation (Fig. 2) may reflect the separation of endogenous proteases responsible for the formation of these smaller species.
The 20-40% ammonium sulfate fraction was purified further by sequential chromatography on columns of phosphocellulose and SP5PW. M, 95,000 EpG formation activity was retained on the SP5PW column and eluted at 0.16 M NaCl ( Fig. 3; fraction 22, top panel). The minor M, 60,000 EpGforming activity eluted at lower salt and was partially resolved from the M, 95,000 species ( Fig. 3; fractions 18 and 20, top panel). Since the formation of EpG (an obligate step in mRNA capping) is a less stringent assay of guanylyltransferase than is the ability to transfer the GMP moiety to the 5' end of RNA, capping activity with a triphosphate-terminated poly(A) acceptor was assayed across the column. A peak of GMP incorporation eluted in parallel with the EpG-forming activity (Fig. 3, middle panel). The broad appearance of the guanylyltransferase peak was attributable to assay conditions Aliquots (1 ~1) of the indicated fractions were assayed for protein-guanylate complex formation as described under "Experimental Procedures." An autoradiograph of the protein gel is shown. The position of the labeled M, 95,000 capping enzyme subunit is indicated by the arrow. The protein concentrations of the fractions were: polymin P supernatant, 4.11 mg/ml; O-20% ammonium sulfate, (A.S.), 0.19 mg/ml; 20-40% ammonium sulfate, 1.69 mg/ml; 40-60% ammonium sulfate, 7.58 mg/ ml; 60-80% ammonium sulfate, 2.55 mg/ml. Protein concentration was measured by dye binding using bovine serum albumin as a standard (23).
. 60+ . GPPPA  (Fig. 3). A parallel set of reactions included 50 pM AdoMet. After incubation for 30 min at 37 "C. reactions were halted bv the addition of 10% trichloroacetic acid. The RNA product was recovered by two cycles of precipitation with trichloroacetic acid, followed by sequential extractions with phenohchloroform and chloroform, and a final ethanol precipitation step. The labeled RNA pellet was resuspended in 30 ~1 of 10 mM Tris-HCl, pH 8.0, 1 mM EDTA. Aliquots The dialyzed phosphocellulose fraction was chromatographed on SPBPW as described in the text. Fractions (1 ml) were collected during salt gradient elution. Top panel, aliquots (1.5 ~1) of the indicated fractions were assayed for protein-guanylate complex formation. An autoradiograph of the protein gel is shown. The position of the labeled M, 95,000 capping enzyme subunit is indicated by the arrow. Middle panel, guanylyltransferase reaction mixtures (20 ~1) contained 50 mM Tris-HCl, pH 7.5, 1.25 mM MgCl?, 5 mM dithiothreitol, 25 pM [~u-~*P]GTP, 11 pmol (of ends) of triphosphate-terminated poly(A), and 1.5 ~1 of the indicated column fractions. After incubation for 30 min at 37 "C, reactions were halted by the addition of 5% trichloroacetic acid. Acid-insoluble material was collected by filtration and counted in liquid scintillation fluid. Activity is plotted as GMP incorporation into acid-insoluble material (picomoles). Methyltransferase reaction mixtures (10 ~1) contained 50 mM Tris-HCl, pH 7.5, 5 mM dithiothreitol, 50 pM AdoMet, 60 fmol of cap-labeled poly(A), and 1 ~1 of a 20-fold dilution (in buffer A) of the indicated column fractions. After incubation for 10 min at 37 "C, the reaction mixtures were made 50 mM in sodium acetate, pH 5.5, and incubated for an additional 60 min at 37 "C with 5 pg of nuclease PI. The digests were spotted on polyethyleneimine cellulose TLC plates that were then developed with 0.45 M ammonium sulfate. Labeled dinucleotides corresponding to "'GpppA and GpppA were detected by autoradiography, then cut out and counted in liquid scintillation fluid. Activity is plotted as the amount (cpm X 10-3.3) of labeled miGpppA synthesized. The activity of fraction 22 represents quantitative methylation of the input RNA substrate. Bottom panel, NaCl concentration was determined using a Radiometer conductivity meter and is plotted as M (X 10-l). Protein concentration is plotted as (pg/ml) X lo-*. EpG is plotted as the concentration (nM X lo-*) of active guanylyltransferase in the indicated fractions, as determined by titration of enzyme-guanylate formation. that reflected yield rather than rate of reaction.

AdOhm
Analysis of the products of the capping reaction by the E. coli enzyme is shown in Fig. 4. Triphosphate-terminated poly(A) was capped with [(Y-"'P]GTP and the nature of the cap dinucleotide was determined by thin layer chromatography after digestion with nuclease Pl. Under standard condi-tions, the SP5PW fraction synthesized the unmethylated cap GpppA. Inclusion of AdoMet in the reaction resulted in the synthesis of the methylated cap m7GpppA; the efficiency of the methylation reaction was underscored by the finding that 99% of the capped product was m7GpppA. The cap products made by the SP5PW fraction were identical chromatographically to the dinucleotides synthesized under the same conditions by capping enzyme purified from vaccinia virions (not shown).
Cap methylation by native vaccinia capping enzyme is not obligately coupled to cap synthesis (4) and can be assayed with high sensitivity by the AdoMet-dependent conversion of 5'guanylylated poly(A) to methylated capped poly(A), as described (16). Assay of RNA (guanine-7)-methyltransferase activity across the SP5PW column showed that this activity eluted broadly, but in parallel with RNA guanylyltransferase activity (Fig. 3, middle panel), suggesting that capping and methylating activities may be associated physically. In order to test this possibility, an aliquot of the peak SP5PW fraction 22 was centrifuged through a 15-30% glycerol gradient in buffer A containing 0.5 M NaCl. M, 95,000 EpG-forming activity sedimented as a single component of 6.5 S relative to marker proteins that were sedimented in a parallel gradient (Fig. 5, toppanel). The ability to cap triphosphate-terminated poly(A) cosedimented at 6.5 S. This sedimentation constant of the expressed guanylyltransferase was identical with the S value of the heterodimeric capping enzyme purified from vaccinia virions (3) and was higher than the 5.5 S value obtained for monomeric large subunit per se (17), suggesting that the coexpressed large and small subunits might form a heterodimeric complex in E. coli. Methyltransferase activity remained associated with guanylyltransferase during sedimentation (in gradient fractions 13-15), but the activity profile was diffuse and peaked in less rapidly sedimenting fractions that contained little guanylyltransferase (Fig. 5). Sedimentation of SP5PW fraction 22 in a 15-30% glycerol gradient is described in the text. Fractions (0.185 ml) were collected from the bottom of the tube. Top panel, aliquots (1 ~1) of the indicated fractions were assayed for protein-guanylate complex formation. An autoradiograph of the protein gel is shown. The position of the labeled M, 95,000 capping enzyme subunit is indicated by the arrow. Bottom panel, aliquots (1 @cl) of the gradient fractions were assayed for guanylyltransferase activity as described in the legend to Fig. 3, except that the incubation was for 8 min at 37 "C. Activity is plotted as GMP incorporation into acid-insoluble material (picomoles, open triangles).
Methyltransferase reaction mixtures were constituted as described in the legend to Fig.  3 and contained 1~1 of a 20-fold dilution (in buffer A) of the indicated gradient fractions. Incubation was for 5 min at 37 "C. After nuclease digestion and chromatographic separation, labeled dinucleotides corresponding to "'GpppA and GpppA were detected by autoradiography, then cut out and counted in liquid scintillation fluid. Activity is plotted as the percent of the input RNA substrate converted to methylated form (open circles). The positions of marker proteins (that had been sedimented in a parallel gradient) are indicated by the arrows.
Capping of triphosphate-terminated RNA requires both RNA y-phosphate cleavage and GMP transfer to the resulting diphosphate RNA end. RNA triphosphatase activity is intrinsic to the native capping enzyme from vaccinia virions, as is the ability to cleave the y-phosphate of purine nucleoside triphosphates (3). Assay of the glycerol gradient fractions for RNA triphosphatase and ATPase (Fig. 6) revealed single peaks of enzyme activity at 6.5 S that were coincident with each other and with the activity profile of the guanylyltransferase. Characterization of the RNA triphosphatase and ATPase activities associated with the expressed capping enzyme is presented in Table I. RNA triphosphate cleavage required magnesium. Manganese was able to activate the triphosphatase partially, while calcium could not satisfy the divalent cation requirement.
ATPase required a divalent cation; magnesium, manganese, and cobalt activated the ATPase, while calcium did not support activity. These features correspond exactly to the reaction requirements reported for the capping enzyme complex purified from vaccinia virions (3), suggesting that the triphosphate phosphohydrolase activity of the glycerol gradient fraction can be attributed to the expressed guanylyltransferase.
Two prokaryotic enzymes that specifically cleave the y-phosphate from RNA (and that also have RNA triphosphatase reaction mixtures (10 ~1) containing 50 mM Tris-HCl, pH 7.5,5 mM MgC12, 5 mM dithiothreitol, 25 pmol (of ends) y-'12P-poly(A) and enzyme (1 ~1 of a 50-fold diluted sample of the indicated glycerol gradient fractions) were incubated for 5 min at 37 "C. The samples were then spotted on polyethyleneimine cellulose TLC plates that were developed with 0.75 M potassium phosphate, pH 3.5. Reaction products corresponding to "Pi were located by autoradiography, then cut out and counted in liquid scintillation fluid. The activity profile is indicated by the open circles. The activity of peak fraction 13 in this experiment corresponds to cleavage of 15% of the input RNA ends. ATPase reaction mixtures (50 ~1) containing 40 mM Tris-HCl, pH 8.0, 4 mM MgCl,, 2 mM dithiothreitol, 10 mM [y-'12P]ATP, and 5 ~1 of the indicated glycerol gradient fractions were incubated for 30 min at 37 "C. Release of "Pi was quantitated as described (3); the activity profile is indicated by the solid triangles. Experiment 1. Complete RNA triphosphatase reactions were constituted as described in Fig. 6 and included 1 ~1 of a lo-fold diluted sample of glycerol gradient fraction 14. Divalent cations (chloride salt) were substituted for magnesium as indicated at 5 mM concentration. The activity of the complete reaction represents cleavage of 95% of the input RNA triphosphate ends. Experiment 2. Complete ATPase reactions contained 60 mM Tris-HCl, pH 8.0, 4 mM MgC12, 1 mM [-y"'P]ATP, and 5 ~1 of glycerol gradient fraction 14. Incubation was for 30 min at 37 "C. Divalent cations (chloride salt) were substituted for magnesium as indicated at 4 mM concentration. NTPase activity) have been identified in E. coli (24). One of the bacterial enzymes, referred to as "alkaline RNA triphosphatase," is active without specificity for the 5'-RNA NTP base. The alkaline triphosphatase does not require magnesium for activity, however, thus ruling out the possibility that this enzyme is contributing to the RNA triphosphatase present in the capping enzyme preparation.
The second bacterial enzyme, "ATP-terminated RNA triphosphatase," is substratespecific as indicated. The sedimentation properties and cation specificity of this enzyme have not been reported.
The polypeptide composition of glycerol gradient fractions was analyzed by SDS-polyacrylamide gel electrophoresis (Fig.  7). RNA triphosphatase, ATPase, EpG formation, and RNA guanylyltransferase activity profiles all correlated with the presence of a major polypeptide of M, 95,000 (indicated by *) in fractions 13-15. This M, 95,000 polypeptide comigrated with the Dl gene product expressed inducibly in E. coli (not shown). A polypeptide of M, 31,000 (that comigrated with the D12 gene product expressed inducibly in E. coli) was also detected in the peak fractions 13-15 (Fig. 7, lower *). This polypeptide was present as well in less rapidly sedimenting fractions and appeared to correlate with methyltransferase activity. These data are consistent with the expressed capping enzyme purified from E. coli being a heterodimer of the Dl and and D12 gene products. The existence of free methyltransferase activity and the implications for the domain structure of the capping enzyme are discussed below.
The purification of RNA guanylyltransferase from E. coli BL21(DE3)pET-Dl/D12 is summarized in Table II. The EpG assay permits direct determination of the molar concentration of active enzyme molecules, assuming a stoichiometry of one GMP residue bound per M, 95,000 polypeptide.
The enzyme was purified lOOO-fold at the SP5PW step (peak fraction) with a yield at this step of 224%. The apparent increase in activity relative to the polymin P supernatant is probably attributable to the removal of interfering activities, particularly bacterial GTPases. The amount of active purified en- Aliquots (35 ~1) of the indicated glycerol gradient were made 1% in sodium dodecyl sulfate and electrophoresed through a 10% polyacrylamide gel containing 0.1% sodium dodecyl sulfate. Protein was visualized by staining with Coomassie Blue dye. A photograph of the stained gel is shown. The positions and sizes (in kilodaltons) of coelectrophoresed marker proteins are indicated on the right. The polypeptides comigrating with the Dl and D12 gene products are indicated by the asterisks (*) on the left. zyme obtained at this step from 1 liter of bacteria is comparable to the amount of guanylyltransferase purified from 1100 Ate0 of purified vaccinia virions (roughly, the yield of virus obtained from infecting lo-12 liters of HeLa suspension cells) (5). At the glycerol gradient step, the expressed guanylyltransferase had been purified 1300-fold. The specific activity of the peak glycerol fraction was 2580 pmol/mg of protein, compared with a theoretical maximum of 7874 for fully active homogeneous heterodimeric enzyme of M, 127,000. The implication that the glycerol fraction is therefore 32% pure is not out of line with the electrophoretic analysis of polypeptide composition.
Quantitative RNA Capping by Expressed Capping Enzyme-A time course of GMP incorpration into a 5'-triphosphateterminated poly(A) cap acceptor is shown in Fig. 8. Under conditions of enzyme excess, the expressed capping enzyme (SP5PW fraction) capped one-half of the input RNA ends. Inclusion of 50 PM AdoMet in the reaction increased the rate and the extent of the reaction and resulted in the quantitative modification of all 5' ends. The effect of AdoMet on RNA modification by the expressed enzyme mimics that observed with the enzyme purified from virions (18). This is attributable to the fact that cap methylation renders the RNA terminus resistant to pyrophosphorolysis of the capped product (19) and to transfer of the guanylate moiety back to the enzyme large subunit (20), and therefore serves to drive the reaction to completion at equilibrium. The ability to cap quantitatively suggests that the SP5PW preparation does not contain high levels of (1) nonspecific phosphatases that would convert the 5' ends to 5'-monophosphate or 5'-hydroxyl forms that are not cap acceptors for the vaccinia enzyme, or (2) nucleotidyl pyrophosphatases that would degrade the triphosphate cap bridge after cap formation. The exact levels of these activities in the enzyme preparation have not been determined directly, however.

DISCUSSION
The capping enzyme purified from infectious vaccinia virus particles contains polypeptides of M, 95,000 and M, 31,000 that are encoded, respectively, by the Dl and D12 viral genes. The present study proves that these two gene products, when coexpressed in a heterologous system, are sufficient to mediate all three enzymatic reactions (RNA triphosphatase, RNA guanylyltransferase, and RNA (guanine-7)-methyltransferase) leading to the formation of "cap zero" 5' ends. The enzymatic and physical properties of the recombinant guanylyltransferase purified from E. coli are consistent with those of the enzyme obtained from virions. In particular, the large and small enzyme subunits appear to form a heterodimeric complex in the prokaryotic millieu, much as they do during the early phase of the vaccinia growth cycle when they are newly synthesized (21) and during their encapsidation into the virus core. The recombinant heterodimeric guanylyltransferase catalyzes cap methylation, yet clearly not all methyltransferase expressed in E. coli is associated physically with guanylyltransferase (Fig. 5). A plausible explanation for this finding is implicit in the following model of the organization of functional domains within the capping enzyme complex. I draw upon a previous model (16) and propose that (1) the RNA triphosphatase and RNA guanylyltransferase domains reside on the M, 95,000 subunit and that this protein per se is sufficient to catalyze y-phosphate cleavage and nucleotidyl transfer; (2) the methyltransferase domain resides on the M, 31,000 subunit, which is itself sufficient to catalyze cap methylation; (3) association of the two subunits with 1:l stoichiometry is an inherent property of the subunits and requires no other viral factors. The existence of free methyltransferase is thereby accounted for by a molar excess of the small subunit relative to the large in E. coli BLZl(DE3)pET-Dl/D12. This excess is precisely what is expected based on the nature of the coexpression plasmid PET-Dl/DlB. D12 expression is driven from two T7 promoters while Dl expression is directed by only one promoter (Fig. 1). Therefore, all other factors being equal (e.g. mRNA stability, translation efficiency, protein solubility, protein stability, etc.), the small subunit should accumulate to a higher steady-state level than the large subunit.
Testing of various aspects of this proposal should be feasible through a molecular genetic approach involving expression of individual subunits in E. coli. The accompanying paper (17) describes the catalytic properties of the large capping enzyme subunit and substantiates the domain structure of this protein discussed above.
Finally, the vaccinia mRNA capping enzyme has been a valuable reagent for the manipulation of the 5'-terminal structure of RNA (22). Purification of the enzyme from large quantities of virions necessitates a facility for cell culture and exposure of personnel to infectious pathogen. The ability to purify the capping enzyme from bacteria (in good amounts and with high yield) using the relatively simple procedure presented above may prove useful for studies of the role of the 5' RNA cap in mRNA biogenesis and function.