tRNAHis guanylyltransferase catalyzes a3؅-5؅ polymerization reaction that is distinct from G؊1 addition Jane E. Jackman and Eric M. Phizicky†

Department of Biochemistry and Biophysics, University of Rochester School of Medicine, Rochester, NY 14642

Communicated by Fred Sherman, University of Rochester Medical Center, Rochester, NY, April 15, 2006 (received for review April 1, 2006)

Yeast tRNAHis guanylyltransferase, Thg1, is an essential protein His that adds a single guanine to the 5؅ end (G؊1) of tRNA . This G؊1 residue is required for aminoacylation of tRNAHis by histidyl-tRNA synthetase, both in vitro and in vivo. The guanine nucleotide addition reaction catalyzed by Thg1 extends the polynucleotide ,chain in the reverse (3؅-5؅) direction of other known albeit by one nucleotide. Here, we show that alteration of the 3؅ end of the Thg1 substrate tRNAHis unleashes an unexpected re- verse activity of wild-type Thg1, resulting in the 3؅-5؅ addition of multiple nucleotides to the tRNA, with efficiency comparable to the G؊1 addition reaction. The addition of G؊1 forms a mismatched G⅐A base pair at the 5؅ end of tRNAHis, and, with monophosphorylated tRNA substrates, it is absolutely specific for His ⅐ tRNA . By contrast, reverse polymerization forms multiple G Cor Fig. 1. Nucleotide addition reactions catalyzed by 5Ј-3Ј polymerases and C⅐G base pairs, and, with preactivated tRNA species, it can initiate tRNAHis guanylyltransferase (Thg1). ,at positions other than ؊1 and is not specific for tRNAHis. Thus wild-type Thg1 catalyzes a templated polymerization reaction acting in the reverse direction of that of canonical DNA and RNA consistent with the demonstration that GϪ1 is necessary and polymerases. Surprisingly, Thg1 can also readily use dNTPs for sufficient for histidyl-tRNA synthetase activity in vitro (11–13). -nucleotide addition. These results suggest that 3؅-5؅ polymeriza- Because GϪ1 modification is a strong determinant of histidyl tion represents either an uncharacterized role for Thg1 in RNA or tRNA synthetase activity, Thg1 must be extraordinarily specific DNA repair or metabolism, or it may be a remnant of an earlier for tRNAHis to prevent misacylation of other tRNAs. Indeed, catalytic strategy used in nature. Thg1 prefers its physiological substrate, monophosphorylated tRNAHis (p-tRNAHis), which arises after cleavage by RNase P, polymerase ͉ Saccharomyces cerevisiae ͉ THG1 ͉ tRNA modification by Ͼ10,000-fold over other p-tRNA species (14). As for amino- acyl-tRNA synthetases, where the anticodon is often a major ll known polymerases catalyze the addition of nucleotides in determinant of tRNA specificity, the tRNAHis GUG anticodon Athe 5Ј-3Ј direction, including classical DNA and RNA is the source of Thg1 specificity because it is both necessary and polymerases, as well as , , and a sufficient for Thg1 activity in vitro (14). The three-step enzymatic number of template-independent RNA polymerases (1–3). reaction catalyzed by Thg1 involves the activation of the 5Ј end These polymerases universally catalyze the same reaction, in- of the tRNA by adenylylation, attack of the 3Ј-OH of GTP on volving attack by the polynucleotide 3Ј-OH on the 5Ј- the activated intermediate, and removal of the 5Ј-pyrophos- His triphosphate of an incoming NTP, to yield an extended polynu- phate, yielding mature GϪ1-containing tRNA and AMP (Fig. cleotide and a pyrophosphate side product from the NTP (Fig. 1B) (7, 8). These chemical steps are also strikingly similar to 1A). However, polymerization could just as easily occur chem- those catalyzed by aminoacyl-tRNA synthetases (14). However, ically in the reverse (3Ј-5Ј) direction with the same functional despite these similarities between Thg1 and synthetases, as well groups, by attack of the 3Ј-OH of the incoming nucleotide on a as the proximity of their sites of action at the top of the tRNA 5Ј-triphosphorylated polynucleotide (Fig. 1B). We show here aminoacyl-acceptor stem, there is no identifiable homology that tRNAHis guanylyltransferase can act in this capacity as a between Thg1 and the synthetases or even between Thg1 and any reverse polymerase, catalyzing the 3Ј-5Ј addition of multiple other family. nucleotides to the polynucleotide chain in a reaction that is Formally, the Thg1 GϪ1 addition reaction is the 3Ј-5Ј extension distinct from its known physiological role. of a polynucleotide chain, albeit by one nucleotide. Moreover, tRNAHis species are unique among tRNAs because they there is an alternative mode of Thg1 activity that more closely contain an additional universally conserved G residue at the Ϫ1 resembles a polymerase, in which triphosphorylated tRNAHis His position (GϪ1), whereas only one other characterized tRNA has (ppp-tRNA ) is used directly for the addition of GϪ1 (Fig. 1B), any nucleotide at that position (4, 5). In prokaryotes, GϪ1 is bypassing the ATP requirement for activation (7). genome-encoded, forms a standard base pair with C73, and is We report here that wild-type Thg1 can catalyze multiple retained during tRNAHis maturation because RNase P cleaves rounds of reverse polymerization with certain substrates. While pre-tRNA at the Ϫ1 position (6). By contrast, in eukaryotes, GϪ1 is added posttranscriptionally by tRNAHis guanylyltransferase, which is encoded in yeast by the essential THG1 gene (7), and Conflict of interest statement: No conflicts declared. it catalyzes the addition of a guanine nucleotide to the 5Ј end of Abbreviations: CIP, calf intestinal alkaline phosphatase; PEI, polyethyleneimine; p-tRNA, His monophosphorylated tRNA; ppp-tRNA, triphosphorylated tRNA. tRNA across from residue A73 of the tRNA, the nucleotide † ࿝ immediately preceding the CCA 3Ј end (7–9). GϪ1 addition To whom correspondence should be addressed. E-mail: eric [email protected]. activity is crucial for aminoacylation of tRNAHis in vivo (10), © 2006 by The National Academy of Sciences of the USA

8640–8645 ͉ PNAS ͉ June 6, 2006 ͉ vol. 103 ͉ no. 23 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0603068103 Downloaded by guest on September 26, 2021 Fig. 2. Thg1 exhibits 3Ј-5Ј reverse polymerase activity with tRNAHis variants containing altered 3Ј ends. (A) Schematic of Thg1 assay with 5Ј-32P-labeled tRNAHis. After RNase A and phosphatase treatment, the Thg1 reaction prod- 32 32 ucts [G[ P]pGpC (Gp*GpC) and longer species] are resolved from Pi derived His from substrate by TLC. (B) Thg1 adds guanine nucleotides to A73C tRNA 32 His His variants. Activity with 5Ј- P-labeled wild-type tRNA (a), tRNA C73CCA (b), His His tRNA C73CAA (c), or tRNA C73ACA (d) variants is shown using 5-fold serial dilutions of purified Thg1 (50 ␮g͞ml to 0.4 ␮g͞ml); Ϫ, no Thg1 added. (C) Confirmation of identity of additional 5Ј reaction products formed by Thg1. BIOCHEMISTRY His Comigration of tRNA C73CCA GϪ1,GϪ2, and GϪ3 nucleotide addition prod- ucts with [3Ј-32P]Cp-labeled, calf intestinal alkaline phosphatase (CIP)-treated oligonucleotide standards (GpGp*C, GpGpGp*C, and GpGpGpGp*C; lanes c, a, and b, respectively) is shown after resolution by silica TLC. Fig. 3. Thg1 adds guanine nucleotides to the 5Ј end of tRNAHis variants in a template-dependent reaction. (A) Thg1 reverse polymerization does not re- quire ATP with ppp-tRNA. Primer extension analysis of Thg1 reaction products His His investigating the tRNA determinants for Thg1 activity, we is shown with wild-type (a), C73CCA (b), C73CAA (c), or C73ACA (d) ppp-tRNA His found that alteration of the conserved nucleotide A73 of tRNA species in the presence of ATP (a–d) or in the absence of ATP (e). The horizontal unleashes a previously unexpected reverse polymerase function line denotes the expected primer extension stop with unreacted transcript of wild-type Thg1, resulting in the 3Ј-5Ј extension of the polynu- (position ϩ1), and additional nucleotide incorporations are indicated by (Ϫ1), (Ϫ2), and (Ϫ3). The primer position is indicated by an arrow on the tRNA cleotide chain by multiple nucleotides. Characterization of Thg1 His reverse polymerization shows that this activity, unlike the addi- diagram; GϪ1,GϪ2, and GϪ3 nucleotides incorporated by Thg1 with the tRNA ⅐ C73CCA variant are indicated in parentheses. (B) Thg1 reverse polymerization tion of GϪ1, is template-dependent, recognizing G C Watson– does not depend on the source of purified Thg1 protein. Shown is the primer Crick base pairs. Moreover, reverse polymerization is not spe- extension of reaction products from the addition of GϪ1 (with wild-type His Ϫ His His cific for tRNA or for starting at the 1 position of tRNA, tRNA ) and reverse polymerization (with tRNA C73CCA) using yeast Thg1 provided that the activated (triphosphorylated) form of the purified from either Saccharomyces cerevisiae (Sc)orEscherichia coli (Ec). (C) tRNA substrate is used in the assays. Thg1 adds guanosine residues according to the number of C residues at the 3Ј His end of the tRNA species. Thg1 reaction products with tRNA C73CCCCA were Results analyzed by primer extension, demonstrating the addition of up to five nucleotides to the 5Ј end of the tRNA. Thg1 Adds Multiple Guanine Residues to the 5؅ End of A73C Variant tRNAHis Species. Although Thg1 normally adds a single guanine His nucleotide to the 5Ј end of tRNA , Thg1 can add multiple His a tRNA C CAA variant results in a GϪ product at higher guanine nucleotides to a variant tRNAHis that bears C instead 73 2 73 Ϫ His concentrations of Thg1 but no G 3 product (Fig. 2Bc), whereas of the A73 that is universally conserved in eukaryotic tRNA reaction with a tRNAHis C ACA variant yields only a GϪ species (4, 15). We demonstrated this phenomenon with a 73 1 product (Fig. 2Bd). Thus, it appears that formation of additional sensitive assay that exploits the phosphatase resistance that products depends on the identity of the A73CCA 3Ј end of tRNA. occurs when GϪ is added to 5Ј end-labeled tRNAHis (14). After 1 Examination of Thg1 activity with the corresponding ppp- incubation of tRNAHis with Thg1, treatment with RNase A and phosphatase yields the internally labeled trimer G[32P]pGpC if tRNA substrates shows that Thg1 has the same spectrum of products with this alternatively activated tRNAHis substrate as Ϫ 32 A G 1 has been added, or Pi from unreacted substrate (Fig. 2 ), His which are well separated by TLC (Fig. 2Ba). However, reaction with the p-tRNA species, which require activation by ATP His His (Fig. 3A). As measured by primer extension product length and of Thg1 with a tRNA variant bearing A73C (tRNA C73CCA) also yields a prominent second product and a small amount of intensity, Thg1 efficiently adds a single guanine nucleotide His a third product at higher concentrations of Thg1 (Fig. 2Bb). residue to ppp-tRNA (Fig. 3Aa), whereas up to three nucle- His These additional products comigrate on TLC with chemically otides are added to a ppp-tRNA C73CCA variant (Fig. 3Ab), His synthesized oligonucleotide standards corresponding to GϪ2 and two nucleotides are added to a ppp-tRNA C73CAA variant His GϪ3 reaction products (Fig. 2C), and they are presumed to be (Fig. 3Ac), and a single nucleotide is added to a ppp-tRNA these species because only GTP and ATP are present in the C73ACA variant (Fig. 3Ad). We conclude that Thg1 catalyzes reactions, and all reaction products are sensitive to RNase T1, authentic 3Ј-5Ј reverse polymerization, defined as extension of producing Gp (data not shown). Similarly, reaction of Thg1 with the 5Ј end of a polynucleotide chain by more than a single

Jackman and Phizicky PNAS ͉ June 6, 2006 ͉ vol. 103 ͉ no. 23 ͉ 8641 Downloaded by guest on September 26, 2021 nucleotide. Because addition of multiple guanine nucleotides to the ppp-tRNAHis variants is at least as efficient in the absence of ATP (Fig. 3Ae), we conclude that Thg1 does not require ATP to activate the 5Ј end of the RNA for multiple-nucleotide additions; rather, we infer that the second and third guanine nucleotide additions, like the first guanine nucleotide addition, are powered by hydrolysis of pyrophosphate from the growing polynucleotide chain. Moreover, the slightly enhanced addition activity ob- His served with the ppp-tRNA C73CCA variant in the absence of ATP suggests that ATP in fact inhibits the multiple-addition reaction. Because yeast Thg1 purified from E. coli exhibits the same multiple-nucleotide addition activity, we conclude that this activity is intrinsic to Thg1 protein (Fig. 3B). The addition of multiple guanine nucleotides by Thg1 is an efficient RNA reverse polymerization activity. Thg1 reverse His polymerization with tRNA C73CCA and other variants occurs with similar efficiency (Figs. 2B and 3A) and on the same time scale (data not shown) as for the addition of GϪ1 to wild-type tRNAHis, whether the compared substrates are ppp-tRNAHis species or p-tRNAHis species. Furthermore, there is an element of template dependence in the reverse polymerization because the number of guanine nucleotides added to the 5Ј end of both p-tRNAHis and ppp-tRNAHis variants appears to be directed by the number of consecutive cytidine residues at the 3Ј end, starting at nucleotide 73 (Figs. 2B and 3). Indeed, up to five guanine residues are added across from the 3Ј end of tRNAHis Fig. 4. Nucleotide dependence of Thg1 reverse polymerization compared C73CCCCA (Fig. 3C). with the addition of GϪ1.(A) Nucleotide dependence of Thg1 activity with ppp-tRNAHis species. Thg1 reaction products formed in the presence of indi- Thg1 Reverse Polymerase Activity Is Template-Dependent Beyond His Ј ؊ vidual NTPs and ppp-tRNA substrates (with 3 end sequence indicated below Position 1. Closer examination of the nucleotide dependence of each panel) were analyzed by primer extension. Lanes G, A, U, and C show each the reverse polymerase activity of Thg1 demonstrates that it is NTP added as indicated; Ϫ, no nucleotide; NP, no Thg1. (B) Thg1 is specific for strictly template-dependent and recognizes G⅐C and C⅐G base the addition of G at position Ϫ1 in the presence of all four NTPs. Thg1 reaction pairs. Consistent with a templated nucleotide addition reaction, products with 5Ј-32P-labeled wild-type tRNAHis are shown in the presence of Thg1 can catalyze multiple rounds of reverse polymerization at GTP and ATP (a) or a mixture of all four NTPs (GTP, ATP, UTP, and CTP) (b–d), His each at a final concentration of 1 mM (b), 0.4 mM (c), or 0.1 mM (d); Ϫ,no the 5Ј end of tRNA C73CCA only in the presence of GTP (Fig. 4Ab), whereas with each of the other NTPs, Thg1 can catalyze added nucleotide; NP, no Thg1. (Left) RNase A and CIP treatment of reaction products, resolved by standard silica TLC. (Right) RNase T2 digestion of reac- only a single addition reaction. Furthermore, reverse polymer- tion products, to release 3Ј-labeled Np resulting from the addition of nucle- ization is not restricted to the addition of guanine nucleotide otide at position Ϫ1, resolved by polyethyleneimine (PEI)-cellulose TLC in 0.5 because up to three cytidine nucleotides are added to the 5Ј end M sodium formate (pH 3.5). Gp, Ap, Up, and Cp, migration of 3Ј-phosphory- His of tRNA G73GGA with CTP (Fig. 4Ac), but only one nucle- lated standards as visualized by fluorescence quenching. otide is added with each of the other NTPs. However, Thg1 does not appear to recognize U⅐A base pairs under these conditions His ؅ ؅ because tRNA U73UUA or A73AAA species are not sub- 3 -5 Reverse Polymerization of ppp-tRNA Substrates by Thg1 Is Not strates for reverse polymerization in the presence of ATP or Specific for tRNAHis. Although our previous results demonstrate UTP, respectively (Fig. 4 Ad and Ae). Thus, polymerization that the addition of Thg1 GϪ1 activity is at least 10,000-fold more beyond the Ϫ1 position appears to require formation of a specific for tRNAHis than for other tRNAs when assayed in vitro His canonical GϪ1⅐C73 or CϪ1⅐G73 base pair and appears to extend with the physiological p-tRNA substrate (14), Thg1 is not further only if the next templated nucleotides at position 74 and nearly as specific when assayed with ppp-tRNA substrates for beyond are G or C residues. Moreover, Thg1 is not restricted to either GϪ1 addition activity (14) or reverse polymerization multiple-nucleotide additions of a single NTP because addition activity. Thus, Thg1 adds a single G residue to wild-type His Phe Ј to a tRNA G73CCA variant is somewhat more extensive in the ppp-tRNA , which normally contains an A73CCA 3 end (Fig. presence of both GTP and CTP than it is with GTP alone (Fig. 5A), and Thg1 efficiently adds up to three guanine residues to a Phe 4Af). The relatively weaker extension observed with this sub- ppp-tRNA C73CCA variant (Fig. 5A). These results demon- strate suggests that some nontemplated GϪ1 addition occurs strate that reverse polymerase activity does not inherently His opposite G73, which would create a nonpaired and therefore depend on recognition of tRNA and therefore could poten- nonextendable substrate. Indeed, as shown below, the addition tially occur with other substrates in vivo. of G is preferred at the Ϫ1 position. Reverse polymerization by Thg1 also does not need to initiate His Templated reverse polymerization by Thg1 beyond position at the Ϫ1 position. Indeed, a GϪ1-containing tRNA C73CCA Ϫ1 is in sharp contrast to the addition reaction that occurs at the transcript, which is the same molecule as the GϪ1 addition His Ϫ1 position because Thg1 can add any nucleotide at the Ϫ1 product that results from Thg1 activity on tRNA C73CCA, is position with each of the ppp-tRNA species examined for reverse a substrate for continued addition of up to two additional polymerization (Fig. 4 Ab–Af), as well as with wild-type tRNAHis nucleotides as demonstrated by primer extension (Fig. 5B). His (Fig. 4Aa). However, consistent with the known role of Thg1, However, the corresponding GϪ1-containing wild-type tRNA when all four NTPs are present, Thg1 adds exclusively G at A73CCA 76-mer transcript cannot be extended further, presum- His His position Ϫ1 of wild-type p-tRNA opposite A73, as indicated by ably because it is the same molecule as mature tRNA , with an RNase T2 analysis of reaction products, which yields only Gp unpaired GϪ1 opposite A73 (Fig. 5B). Likewise, Thg1 adds two (Fig. 4B). nucleotides to a tRNAHis transcript that initiates at residue ϩ2;

8642 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0603068103 Jackman and Phizicky Downloaded by guest on September 26, 2021 Fig. 5. Thg1 reverse polymerase activity is not specific for tRNAHis and does Phe not need to begin at the Ϫ1 position. (A) tRNA C73CCA is a substrate for Thg1 reverse polymerization. Thg1 catalyzes the addition of GϪ1 to wild-type BIOCHEMISTRY Phe Phe ppp-tRNA and addition of multiple G residues to ppp-tRNA C73CCA, as analyzed by primer extension. (B) Reverse polymerization can initiate at His position Ϫ1. Thg1 catalyzes addition of GϪ2 and GϪ3 to a tRNA 76-mer transcript that already contains GϪ1 when this GϪ1 residue is base-paired with C73, but not when it is opposite A73, as analyzed by primer extension. (C) Reverse polymerization can initiate at position ϩ2. Thg1 catalyzes the addi- His tion of Gϩ1 and GϪ1 to a tRNA variant that begins at the ϩ2 position, yielding the same GϪ1-containing reaction product as a control RNA species that begins Ј with Gϩ1, as analyzed by primer extension. (D) Thg1 does not add additional Fig. 6. Thg1 adds deoxynucleotides and ribonucleotides to the 5 end of His 32 His residues to p-tRNA species that do not begin with Gϩ1.5Ј- P-labeled tRNA .(A) Comparison of Thg1 activity in the presence of GTP or dGTP. Thg1 Ј 32 His wild-type or variant tRNAHis and tRNAPhe species, as indicated below each was assayed with 5 - P-labeled wild-type or C73CCA tRNA and the standard panel, were assayed in the presence of GTP and ATP. GpGpC is the reaction RNase A͞CIP assay in the presence of either GTP or dGTP as indicated. (B) product observed with wild-type tRNAHis; GpGpGpC is the tetramer RNase A Verification of deoxyguanosine nucleotide incorporation into tRNA. (a) Sam- reaction product expected for Gϩ2 control RNA, which has an altered G2⅐C71 ples from A were resolved by TLC on PEI-cellulose in 0.5 M sodium formate (pH base pair. 3.5), where dG-containing reaction products migrate differently from the corresponding ribonucleotide-containing products. (b) Samples from A were treated with RNase T1 and CIP and resolved in the same system.

the first nucleotide addition restores the Gϩ1 5Ј end, and the second adds GϪ1 opposite A73, yielding the same GϪ1-containing product as is observed with a control tRNAHis transcript that incorporation are resistant to RNase T1 treatment (Fig. 6Bb). Thus, dGTP is incorporated and not contaminating GTP. starts at Gϩ1 (Fig. 5C). However, the addition of a nucleotide Phe initiating at either GϪ1 or at Gϩ2 or to the tRNA C73CCA Discussion variant was not detected when the corresponding p-tRNA species were assayed for activity with 5Ј-end-labeled p-tRNA; in We have demonstrated that wild-type Thg1 has an unexpected this assay, the only p-tRNA species that was a substrate for Thg1 second catalytic activity in addition to its known role of adding Ј His was the wild-type control p-tRNAHis bearing the altered 2⅐71 GϪ1 to the 5 end of tRNA (7, 10): Thg1 can catalyze a reverse Ј Ј Ј base pair (Fig. 5D). The lack of reaction with all other p-tRNA (3 -5 ) polymerase activity, adding multiple nucleotides to the 5 variants is consistent with the high selectivity of Thg1 for mature end of variant tRNA species in a template-dependent reaction. p-tRNAHis. Like the normal GϪ1 addition reaction, reverse polymeriza- tion requires ATP with p-tRNAHis variants that begin at ϩ1.

Thg1 Efficiently Incorporates dGTP in both the G؊1 Addition and Reverse polymerization with all other substrate tRNAs requires Reverse Polymerization Reactions. To explore further the general- ppp-tRNA and can occur in the absence of ATP. Presumably, as ity of the Thg1 reaction, we investigated Thg1 activity with with the normal GϪ1 addition, these triphosphorylated sub- deoxynucleotides. Surprisingly, Thg1 appears to incorporate strates bypass the ATP requirement because they are already dGTP with efficiency comparable with that of GTP, as indicated activated for the first nucleotide addition, and this and subse- by the similar amounts of addition products observed with both quent nucleotide addition steps are powered by pyrophosphate nucleotides, using 5Ј-end-labeled wild-type p-tRNAHis and p- release of the ␤- and ␥-phosphates from the triphosphorylated His Ј tRNA C73CCA substrates (Fig. 6A). Although the deoxy- and nucleotide at the 5 end of the growing polynucleotide chain ribonucleotide reaction products are not resolved from one (Fig. 1B). In support of this mechanism for reverse polymeriza- another in this TLC system (Fig. 6A), they are easily resolved tion, a substrate tRNA that begins at Ϫ1 and therefore emulates with another TLC assay system to analyze the same reaction the product of the first addition reaction is active only if it is products (Fig. 6Ba), and as expected, the products of dGTP triphosphorylated (Fig. 5B). This mechanism of reverse poly-

Jackman and Phizicky PNAS ͉ June 6, 2006 ͉ vol. 103 ͉ no. 23 ͉ 8643 Downloaded by guest on September 26, 2021 His Phe merization is very similar to the pyrophosphate release that tRNA and tRNA C73CCA substrates at least as efficiently powers conventional 5Ј-3Ј polymerization, although in 5Ј-3Ј as it catalyzes its known physiological GϪ1 addition reaction on polymerization the pyrophosphate originates from the incoming wild-type tRNAHis (Figs. 2B,3A, and 5A) and that both of these NTP instead of the polynucleotide (Fig. 1B). reactions occur on the same time scale (data not shown). Thus, Under normal circumstances in the cell, tRNAs other than it is tempting to speculate that the reverse polymerase activity tRNAHis are presumably not substrates for reverse polymeriza- has a function. It is possible that the reverse polymerization tion or GϪ1 addition by Thg1 because there is no ready source activity of Thg1 might simply be a remnant of an earlier 3Ј-5Ј of activated 5Ј ends of the tRNAs. There is no naturally occurring polymerase activity of this family of proteins, which could have mature ppp-tRNA that would be a substrate for Thg1, and competed with or acted together with the activities of conven- activation by adenylylation occurs only with 75-mer p-tRNAHis tional 5Ј-3Ј polymerases for replication or repair functions. through Thg1 recognition of the anticodon (14), consistent with However, it is particularly difficult to reconcile the evolutionary His His the crucial role of GϪ1-containing tRNA in charging tRNA retention of a distinct Thg1 reverse polymerization activity that in vitro (11–13) and in vivo (10). Furthermore, tRNAHis nucle- forms G⅐C and C⅐G base pairs because such Watson–Crick base otide addition is restricted to a single guanosine residue because pair recognition and use would not be expected of an enzyme His Thg1 adds only the single required GϪ1 opposite the discrimi- whose only role is to add GϪ1 opposite A73 of tRNA in vivo. nator A73 (Figs. 2B and 3A). This feature emphasizes the Rather, the reverse polymerization described here may reflect an His importance of the conserved A73 in eukaryotic tRNA because additional Thg1 activity that is used in the cell. Indeed, there is Thg1 could add additional residues opposite the conserved an apparent 3Ј-5Ј polymerization activity that functions to edit prokaryotic discriminator, C73. the 5Ј end of mitochondrial tRNA in Acanthamoeba castellanii His The incorporation of GϪ1 into tRNA in archaeal species is and Spizellomyces punctatus although, unlike Thg1, this activity more complicated. Despite the nearly universal conservation of does not proceed beyond the normal 5Ј end of the tRNA C73 in Archaea, there are at least some species that appear to (19–21). Because the gene product(s) that catalyze this reaction require posttranscriptional GϪ1 addition opposite C73 because are unknown, the relationship between these activities is unclear. His their tRNA genes lack a genomically encoded GϪ1, and the Thg1 might act in a similar or related capacity in 3Ј-5Ј polymer- corresponding organisms have a predicted Thg1 ortholog. Thus, ization of tRNA substrates, although this role would depend on archaebacterial Thg1 homologs may possess additional mecha- some additional mechanism to activate the 5Ј ends of the tRNA nisms to ensure addition of only a single guanine residue across species. from C73. Consistent with this possibility, the archaeal THG1 The ability of Thg1 to use deoxynucleotides may also have orthologs are much less conserved than eukaryotic orthologs, implications for Thg1 function. Thg1 incorporates deox- and, although both human and Candida albicans THG1 or- yguanosine with an efficiency similar to its ribonucleotide coun- His thologs can complement a thg1 conditional yeast strain, the terpart for both reverse polymerization with tRNA C73CCA His archaeal ortholog from Methanobacterium thermoautotrophicum and the addition of GϪ1 with wild-type tRNA (Fig. 6). This does not complement a thg1 mutant under the same expression property of Thg1 is in contrast to the highly selective preferences conditions (data not shown). exhibited by most 5Ј-3Ј RNA and DNA polymerases for their The results presented here, coupled with previous results, illus- nucleotide substrates (22). However, several members of the trate a bewildering array of nucleotide recognition and use by Thg1 polymerase X family of DNA damage repair polymerases also protein. During the addition of GϪ1, Thg1 recognizes the 5Ј end of exhibit a lack of discrimination toward nucleotide sugar species ATP for adenylylation, the 3Ј end of incoming NTPs for nucleotidyl in vitro, which may be important for their proposed functions in transfer, and the 5Ј-pyrophosphate of the product tRNA for vivo, including polymerization at points in the cell cycle when pyrophosphatase activity (Fig. 1). In addition, reverse polymeriza- cellular dNTP levels are low and signaling the site of repair by tion of ppp-tRNAHis substrates by Thg1 at positions other than Ϫ1 marking the chromosome with incorporated ribonucleotides (23, requires formation of a G⅐CorC⅐G base pair and the potential to 24). Thus, the ability of Thg1 to use deoxynucleotides suggests form similar base pairs at subsequent addition sites, whereas any that any additional functions of Thg1 might employ either nucleotide can be added at position Ϫ1 whether or not a base pair dNTPs or rNTPs. In this regard, Thg1 has recently been shown can be formed (Fig. 4). Thus, Thg1 has the potential both to to associate with the origin recognition complex and to be recognize and select for nucleotides that participate in Watson– implicated in the G2͞M transition in yeast (25), suggesting a Crick base pairing for polymerization and to recognize and select potential role at this site. guanosine in a non-Watson–Crick interaction opposite A73.Al- though Thg1 base pair recognition is limited to G⅐C but apparently Materials and Methods not U⅐A base pairs (Fig. 4), it is not limited by the normal tRNA 3Ј Sources of Thg1. Thg1 protein was purified from both S. cerevisiae end in the number of nucleotides that can be added because a tRNA and E. coli as described in ref. 14. Unless otherwise indicated, the His with five C residues at the 3Ј end (tRNA C73CCCCA) can be Thg1 protein used for all assays was the yeast purified protein. extended by up to five nucleotides (Fig. 3C). The multiple modes of nucleotide selection by the relatively small (28 kDa) Thg1 suggest Variant tRNA Species. All tRNA variants with the exception of His conformational changes during the course of the reaction, and they C73CCCCA tRNA were cloned into a previously described are reminiscent of the multiple use of different nucleotides by the pUC13-based plasmid under control of the T7 RNA polymerase CCA-adding enzyme, which restructures its single for promoter for purposes of in vitro transcription (7). tRNAHis and accommodation of each new nucleotide substrate during nontem- tRNAPhe variants were all created by mutagenic PCR, with the plated addition to the 3Ј end of the tRNA (16–18). T7 RNA polymerase promoter encoded by the 5Ј PCR primer We emphasize that the reverse polymerase activity described and subsequent ligation of the amplified PCR product into the His here is distinct from the addition of GϪ1 to tRNA that occurs PstI͞BamHI sites of the previously described EMP835 (7). All His in vivo during tRNA maturation because the addition of GϪ1 tRNA-containing plasmids were verified by sequencing. The His is not dependent on base pair formation, does not require prior tRNA C73CCCCA RNA substrate was made by ligation of a activation of the tRNAHis 5Ј end, and is absolutely specific for phosphorylated synthetic RNA oligonucleotide (5Ј-ggagauggc- p-tRNAHis that arises after cleavage of pre-tRNA by RNase P. CCCCCA) to the 3Ј end of a truncated tRNAHis transcript, By contrast, reverse polymerization depends on base pair for- ending at A63, using 10 units͞␮l T4 DNA (USB Corp.) in mation and is not specific for the correct 5Ј end of tRNAHis.We the presence of a complementary DNA oligonucleotide span- also emphasize that Thg1 catalyzes reverse polymerization on ning the ligation site.

8644 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0603068103 Jackman and Phizicky Downloaded by guest on September 26, 2021 Thg1 Activity Assay with [5؅-32P]tRNA. [5Ј-32P]tRNA species were Determination of 5؅ Nucleotide Identity by RNase T2 Digestion. 5Ј prepared by phosphatase treatment of in vitro transcripts fol- 32P-labeled wild-type tRNAHis was used in Thg1 reactions in the lowed by labeling with [␥-32P]ATP using T4 polynucleotide buffer described above with varied concentrations of all four (Roche Applied Biosciences) to a final specific activity of NTPs (1 mM, 0.4 mM, or 0.1 mM each) and 1 ␮M purified Thg1 Ϸ7,000 Ci͞mmol (1 Ci ϭ 37 GBq). The [5Ј-32P]tRNA was used in a 10-␮l reaction mixture for6hatroom temperature. After as the substrate in a Thg1 assay containing 10 mM MgCl2,3mM 6 h, a portion (4 ␮l) of the reaction was stopped, digested, and DTT, 125 mM NaCl, 0.2 mg͞ml BSA, and, unless otherwise analyzed as described above with EDTA and RNase A. Another indicated, 1 mM each GTP and ATP. Enzyme titration assays 4-␮l aliquot was removed to a new tube containing 0.5 unit of consisted of 5-fold serial dilutions (Ϸ50 ␮g͞ml to 0.4 ␮g͞ml) of CIP in 1ϫ alkaline phosphatase buffer in a final volume of 10 ␮l purified Thg1. Reactions were started by the addition of enzyme to remove the background from the remaining unreacted sub- and carried out at room temperature. After 5–6 h, the reaction strate tRNA. After 30 min at 37°C, the reaction products were was stopped by removal of an aliquot (4 ␮l) to a new tube purified by extraction with phenol͞chloroform͞isoamyl alcohol containing 0.5 ␮l of 0.5 M EDTA and 0.5 ␮lof10mg͞ml RNase and precipitation with ethanol. The resulting tRNA was resus- A followed by digestion for 10–30 min at 50°C and treatment pended in a 10-␮l reaction mixture containing 1 unit of RNase with 0.5 unit of CIP in 1ϫ phosphatase reaction buffer in a final T2 and 0.5 ␮g of RNase A in 20 mM sodium acetate (pH 5.2)͞1 volume of 6 ␮l at 37°C for 30 min. Reactions were spotted to mM EDTA and digested for 30 min at 50°C. Two microliters of silica TLC plates (EM Science) and resolved in an n-propyl each digestion reaction mixture was spotted to PEI-cellulose alcohol͞NH4OH͞H2O (55:35:10) solvent system. The plates plates and developed in a 0.5 M sodium formate (pH 3.5) solvent were visualized and quantified with a PhosphorImager and system. The positions of unlabeled nucleotide standards were IMAGEQUANT software. determined by spotting 2 ␮l of 10 mM solutions of GMP, UMP, AMP, or CMP and visualizing the resultant spots by fluores- .5؅ End Analysis by Primer Extension. Thg1 reactions were carried cence quenching outfor6hatroom temperature in the same buffer described above with 1 ␮M unlabeled triphosphorylated tRNA in a 10-␮l Determination of Deoxyguanosine Incorporation by Resistance to reaction volume containing 1 ␮M purified Thg1. Unless other- RNase T1 Digestion. Thg1 reactions were carried out as described 32 His wise noted, each reaction containeda1mMconcentration of the above with 5Ј- P-labeled tRNA (wild-type and C73CCA indicated NTP. Thg1-treated tRNAs were purified by extraction species) in the presence of 0.1 mM ATP and either 1 mM GTP with phenol͞chloroform͞isoamyl alcohol, concentrated by pre- or1mMdGTPfor5hatroom temperature. After digestion with BIOCHEMISTRY cipitation with ethanol, and resuspended in 10 ␮l of buffer RNase A and CIP, as described above, a 2-␮l portion of the containing 10 mM Tris⅐HCl (pH 7.5) and 1 mM EDTA. Two reaction mixture was removed to a new tube containing 2 ␮gof microliters of each purified tRNA was used as the template in torula yeast tRNA and 0.5 unit of RNase T1 in 30 mM sodium primer extension reactions employing a 5Ј-end-labeled primer acetate (pH 5.2) buffer in a final volume of 5 ␮l. The RNase T1 specific for tRNAHis (5Ј-ACTAACCACTATACTAAGA-3Ј)as digestion was allowed to proceed for 30 min at 50°C, and then the shown in Fig. 2A or a similar tRNAPhe-specific primer (5Ј- reaction mixtures were treated with 0.5 unit of CIP in 1ϫ alkaline TGGCGCTCTCCCAACTG-3Ј). After the Thg1-treated tRNA phosphatase reaction buffer in a final volume of 6 ␮l. After 30 template was annealed to 1 pmol of primer in a final volume of min at 37°C, 2 ␮l of each reaction mixture (both the original 5 ␮l, extension reactions were carried out with 8 units of avian RNase A-treated and RNase A- ϩ RNase T1-treated reactions) myeloblastosis virus reverse transcriptase (AMV-RT) (20 units͞ were spotted to PEI-cellulose TLC plates and resolved in the 0.5 ␮l; Promega) in the presence of a 0.4 mM concentration of each M sodium formate (pH 3.5) solvent system. dNTPin1ϫ AMV-RT reaction buffer (Promega) in a final volume of 10 ␮l at 37°C for 1–2 h. The resulting products were We thank S. Crary, S. Fields, M. Gorovsky, M. Gray, E. Grayhack, R. resolved on 10% polyacrylamide͞4 M urea gels, which were Green, and A. Hopper for advice and comments. This work was dried and visualized with the PhosphorImager. supported by National Institutes of Health Grant GM52347.

1. Steitz, T. A. (1999) J. Biol. Chem. 274, 17395–17398. 12. Rudinger, J., Florentz, C. & Giege, R. (1994) Nucleic Acids Res. 22, 5031–5037. 2. Aphasizhev, R., Sbicego, S., Peris, M., Jang, S. H., Aphasizheva, I., Simpson, 13. Rudinger, J., Felden, B., Florentz, C. & Giege, R. (1997) Bioorg. Med. Chem. A. M., Rivlin, A. & Simpson, L. (2002) Cell 108, 637–648. 5, 1001–1009. 3. Schurer, H., Schiffer, S., Marchfelder, A. & Morl, M. (2001) Biol. Chem. 382, 14. Jackman, J. E. & Phizicky, E. (April 2006) RNA, 10.1261/.54706. 1147–1156. 15. Marck, C. & Grosjean, H. (2002) RNA 8, 1189–1232. 4. Sprinzl, M., Horn, C., Brown, M., Ioudovitch, A. & Steinberg, S. (1998) Nucleic 16. Xiong, Y., Li, F., Wang, J., Weiner, A. M. & Steitz, T. A. (2003) Mol. Cell 12, Acids Res. 26, 148–153. 1165–1172. 5. Schnare, M. N., Heinonen, T. Y., Young, P. G. & Gray, M. W. (1985) Curr. 17. Shi, P. Y., Maizels, N. & Weiner, A. M. (1998) EMBO J. 17, 3197–3206. Genet. 9, 389–393. 18. Yue, D., Weiner, A. M. & Maizels, N. (1998) J. Biol. Chem. 273, 29693–29700. 6. Orellana, O., Cooley, L. & Soll, D. (1986) Mol. Cell. Biol. 6, 525–529. 19. Lonergan, K. M. & Gray, M. W. (1993) Science 259, 812–816. 7. Gu, W., Jackman, J. E., Lohan, A. J., Gray, M. W. & Phizicky, E. M. (2003) 20. Lonergan, K. M. & Gray, M. W. (1993) Nucleic Acids Res. 21, 4402. Genes Dev. 17, 2889–2901. 21. Bullerwell, C. E. & Gray, M. W. (2005) J. Biol. Chem. 280, 2463–2470. 8. Jahn, D. & Pande, S. (1991) J. Biol. Chem. 266, 22832–22836. 22. Rose, A. M., Joyce, P. B., Hopper, A. K. & Martin, N. C. (1992) Mol. Cell. Biol. 9. Cooley, L., Appel, B. & So¨ll, D. (1982) Proc. Natl. Acad. Sci. USA 79, 12, 5652–5658. 6475–6479. 23. Nick McElhinny, S. A. & Ramsden, D. A. (2003) Mol. Cell. Biol. 23, 2309–2315. 10. Gu, W., Hurto, R. L., Hopper, A. K., Grayhack, E. J. & Phizicky, E. M. (2005) 24. Bebenek, K., Garcia-Diaz, M., Patishall, S. R. & Kunkel, T. A. (2005) J. Biol. Mol. Cell. Biol. 25, 8191–8201. Chem. 280, 20051–20058. 11. Nameki, N., Asahara, H., Shimizu, M., Okada, N. & Himeno, H. (1995) Nucleic 25. Rice, T. S., Ding, M., Pederson, D. S. & Heintz, N. H. (2005) Eukaryot. Cell Acids Res. 23, 389–394. 4, 832–835.

Jackman and Phizicky PNAS ͉ June 6, 2006 ͉ vol. 103 ͉ no. 23 ͉ 8645 Downloaded by guest on September 26, 2021