Closely Related CC- and A-adding Enzymes Collaborate to Construct and Repair the 3′-Terminal CCA of tRNA in Synechocystis sp. and Deinococcus radiodurans *

The 3′-terminal CCA sequence of tRNA is faithfully constructed and repaired by the CCA-adding enzyme (ATP(CTP):tRNA nucleotidyltransferase) using CTP and ATP as substrates but no nucleic acid template. Until recently, all CCA-adding enzymes from all three kingdoms appeared to be composed of a single kind of polypeptide with dual specificity for adding both CTP and ATP; however, we recently found that in Aquifex aeolicus, which lies near the deepest root of the eubacterial 16 S rRNA-based phylogenetic tree, CCA addition represents a collaboration between closely related CC-adding and A-adding enzymes (Tomita, K. and Weiner, A. M. (2001) Science 294, 1334–1336). Here we show that inSynechocystis sp. and Deinococcus radiodurans, as in A. aeolicus, CCA is added by homologous CC- and A-adding enzymes. We also find that the eubacterial CCA-, CC-, and A-adding enzymes, as well as the related eubacterial poly(A) polymerases, each fall into phylogenetically distinct groups derived from a common ancestor. Intriguingly, the Thermatoga maritima CCA-adding enzyme groups with the A-adding enzymes, suggesting that these distinct tRNA nucleotidyltransferase activities can intraconvert over evolutionary time.

The 3Ј-terminal CCA sequence (positions 74 -76 in the standard cloverleaf representation) is universally present in the tRNAs of all organisms (1) and is important for many aspects of gene expression. The CCA terminus is required for the aminoacylation of tRNA (2,3) and for translation on the ribosome where the CCA sequences of the aminoacyl-and peptidyl-tRNA pair with the large ribosomal RNA near the peptidyltransferase center (4 -6). In eubacteria, the CCA sequence is required for the efficient maturation of the 5Ј-end of tRNA by RNase P (7,8). In eukaryotes, the CCA sequence serves as an antideterminant to block 3Ј-exonuclease activity (9), and it is essential for the export of mature tRNA from the nucleus to the cytoplasm (10,11).
The CCA-adding enzyme (ATP(CTP):tRNA nucleotidyltransferase) builds and repairs the 3Ј-terminal CCA sequence of all tRNAs (12). CCA-adding activity has been identified in all three kingdoms, suggesting conservation of function and per-haps structure throughout evolution (13). CCA-adding activity is essential in some eubacteria as well as in all archaea and eukaryotes where some or all tRNA genes do not encode CCA (14). Yet, even in organisms such as Escherichia coli where all tRNA genes do encode CCA, CCA-adding activity confers a substantial selective advantage, probably by repairing tRNAs that have been subject to errant nucleolytic attack (15).
The CCA-adding enzyme belongs to the nucleotidyltransferase (NTR) 1 family, a large protein superfamily that encompasses template-dependent DNA polymerases (DNA polymerase ␤) and template-independent RNA and DNA polymerases (poly(A) polymerase, terminal deoxynucleotidyltransferase, and CCA-adding enzymes) as well as metabolic regulators (GlnB uridylyltransferase, glutamine synthase adenylyltransferase) and antibiotic resistance factors (kanamycin nucleotidyltransferase and streptomycin adenylyltransferase) (13,16,17). The CCA-adding enzyme is unique among these NTRs because, unlike other the other DNA and RNA polymerases in the superfamily, it does not use a nucleic acid template, yet it faithfully constructs a defined nucleotide sequence by the addition of mononucleotides. Several models have been proposed to explain how the CCA addition could be templated by protein alone or by a tRNA/protein complex (18 -23), but the detailed mechanism of the CCA addition remains unknown.
Until recently, all CCA-adding enzymes characterized from all three kingdoms were composed of a single kind of polypeptide with dual specificity for the 3Ј-terminal addition of CTP and ATP to tRNA primers. However, we found that CCAadding activity in Aquifex aeolicus reflects a collaboration between two closely related nucleotidyltransferases (24). The smaller polypeptide (Aa.S) is a CC-adding enzyme that adds CTP at positions 74 and 75 of a tRNA primer. The larger polypeptide (Aa.L) is an A-adding enzyme that adds a single ATP at position 76 of a tRNA primer. Remarkably, the A. aeolicus Aa.S and Aa.L polypeptides do not appear to bind to each other or to work in concert. We also found that Thermatoga maritima encodes a polypeptide that is homologous to Aa.L over its entire length, yet it has CCA-instead of A-adding activity; moreover, the homologous N-terminal halves of the T. maritima CCA-adding enzyme and the Aa.L A-adding enzyme are dispensible for NTR activity.
Here we show that the addition of CCA to tRNA is the joint responsibility of homologous CC-and A-adding enzymes in two other eubacteria, Synechocystis sp. and Deinococcus radio-durans, just as it is in A. aeolicus. A phylogenetic analysis of eubacterial CCA-, CC-, and A-adding enzymes, as well as of poly(A) polymerases, places the CC-and A-adding enzymes in distinct groups from CCA-adding enzymes and poly(A) polymerases. We discuss several different evolutionary scenarios that could explain these data.

MATERIALS AND METHODS
Cloning of Nucleotidyltransferase Homologs from Synechocystis sp. and Deinococcus radiodurans-As shown in Fig. 1A, Synechocystis sp. encodes two nucleotidyltransferase homologs, designated Sy.S and Sy.L for short and long polypeptides, with accession numbers NP_442458 and NP_441479 respectively (25). The coding regions were amplified from Synechocystis sp. PCC 6803 genomic DNA (generous gift from T. Kaneko of the Kazusa DNA Research Institute) using PCR primer pairs Sy.SF (5Ј-GGGCATATGCTTTGTCCTGTGAGCCACCTG-3Ј) and Sy.SR (5Ј-CCCCTCGAGAGATTTTCCCAGTTCTTGAGCATA-3Ј) for Sy.S and Sy.LCF (5Ј-GGGCATATGCGGACCGATGTGTTACGGCAATTATTG-3Ј) and Sy.LCR (5Ј-CCCCTCGAGAACTTTGTTCGGTAAAGGATAAT-G-3Ј) for the C-terminal half of Sy.L (residues 399 -942). To allow cloning of the coding region between the NdeI and XhoI sites of the pET22b(ϩ) expression vector (Novagen), the Sy.SF and Sy.LCF primers contained an NdeI site (underlined) overlapping the start codon, and the Sy.SR and Sy.LCR primers contained a XhoI site (underlined and italicized) overlapping the natural stop codon, which was changed to a serine TCG codon. The resulting plasmids expressing Sy.S and Sy.LC are pETSy.S and pETSy.LC, respectively. Note that Sy.LC contains about 50 additional N-terminal residues beyond the region of homology with Aa.LC and Tm.C ( Fig. 1 and Ref. 24); without these N-terminal residues, Sy.LC was insoluble whether it had an N-terminal or a Cterminal hexahistidine tag.
As shown in Fig. 1A, Deinococcus radiodurans encodes two NTR homologs, designated DR.1 and DR.2, with accession numbers NP_294707 and NP_294915 (26). The coding regions were amplified from genomic DNA (generous gift from S. Wolin of Yale University) using PCR primer pairs DR.1F (5Ј-TTTCATATGTTTCGTCGCCGTCC-GCCCCTGCCGCCGTTTCCT-3Ј) and DR.1R (5Ј-TTCGGATCCGAACT-TCCCGGCGTTTCCTCCGCCCA-3Ј) for DR.1 and DR.2F (5Ј-TTTCAT-ATGGCGACCCCAGACGGCGAGCAGGTCTGG-3Ј) and DR.2R (5Ј-TT-CGGATCCGAGGTCGCCTTGGGGTTTGCGCCGAGGTA3Ј) for DR.2. As in the Synechocystis protocol, the DR.1F and DR.1R primers contain an NdeI site (underlined) overlapping the start codon, and the DR.1R and DR.2R primers contain a BamHI site (underlined and italicized) overlapping the natural stop codon, which was changed to a serine TCG codon. PCR failed using Taq and Pfu polymerase, presumably due to the high GϩC content of D. radiodurans DNA, but was successful with Herculase Hotstart DNA polymerase (Stratagene) and Me 2 SO. The 50-l PCR reaction contained 1 ϫ Herculase DNA polymerase buffer, 200 M dNTP, 25 pmol of each primer, 2.5 units of polymerase, and 6% Me 2 SO. The PCR reaction conditions were one cycle at 98°C for 3 min followed by 35 cycles consisting of 98°C, 65°C, and 72°C for 45 s, 45 s, and 1 min 45 s, respectively. The PCR fragments were purified by gel electrophoresis and cloned into pCR 2.1-TOPO vector (Invitrogen), and the sequences were verified by sequencing. The NdeI-BamHI DNA fragment of DR.1 and a partial NdeI and BamHI digest of DR.2 (which contains an NdeI site within the open reading frame) were cloned between the NdeI and BamHI sites of the pET22b(ϩ) expression vector. The resulting plasmids expressing DR.1 and DR.2 are pETDR.1 and pETDR.2, respectively.
Expression and Purification of Recombinant Proteins-The expression vectors were transformed into E. coli strain BL21(DE3) carrying a plasmid encoding the minor tRNA Arg (argU). Transformants were grown in LB broth containing 50 g/ml ampicillin and 25 g/ml kanamycin at 37°C to an A 600 of 0.8, and expression was induced by the addition of 0.1 mM isopropyl ␤-D-thiogalactopyranoside (IPTG) for 8 h at 25°C. The cells were lysed by sonication in buffer A (50 mM Tris-HCl (pH 7.8), 10 mM MgCl 2 , 0.5 M KCl, 6 mM ␤-mercaptoethanol, 5% glycerol, and 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and the lysates were cleared by centrifugation for 30 min at 4°C at 10,000 ϫ g. The cleared lysates were passed over nickel nitrilotriacetic agarose columns (Qiagen), washed extensively with buffer A plus 20 mM imidazole, and the histidine-tagged proteins were eluted with buffer A containing 150 mM imidazole. The eluted proteins were dialyzed against buffer A containing 200 mM KCl and 50% glycerol, and stored at Ϫ20°C. Protein concentrations were assayed with the Bio-Rad protein assay kit. The purity of the histidine-tagged proteins was about 90% as judged by SDS-PAGE. A. aeolicus A-adding and CC-adding enzymes were prepared as described (24).
tRNA Substrates Lacking CCA, CA, and A-tRNA transcripts lacking CA and A (tRNA-DC and tRNA-DCC, where D is a discriminator nucleotide at position 73) were prepared by in vitro transcription of the pmBSCCA plasmid (a kind gift of N. R. Pace) (8) linearized with FokI and BbsI, respectively, as described (27). For tRNA lacking CCA (tRNA-D), the plasmid was mutated using the QuikChange mutagenesis kit (Stratagene) as described (24) and digested with FokI prior to in vitro transcription. Transcripts were purified by denaturing polyacrylamide gel electrophoresis with one nucleotide resolution, excised after UV shadowing, and then eluted, ethanol precipitated, washed with cold 70% ethanol, and dried. Uniformly labeled tRNAs were prepared identically using 800 nM [␣-32 P]UTP (Amersham Biosciences, 3000 Ci/ mmol), 25 M UTP, and 500 M CTP, ATP, and GTP. The desired products were located by autoradiography and then excised, eluted, and concentrated as for unlabeled products.

RESULTS AND DISCUSSION
Two NTR Homologs in Synechocystis sp. and D. radiodurans-We recently found that the addition of CCA to tRNA in A. aeolicus reflects a collaborative effort between two different but closely related polypeptides, one that adds CC at positions 74 and 75 and another that adds A at position 76 (24). A BLAST search (37) using the A. aeolicus CC-adding and A-adding enzyme sequences revealed that Synechocystis sp. (25) encodes two NTRs designated Sy.S and Sy.L (accession numbers NP_442458 and NP_441479 respectively). Sy.S is homologous to the A. aeolicus CC-adding enzyme (Aa.S), and Sy.L is homologous to the A. aeolicus A-adding enzyme (Aa.S). Sy.L also possesses the same N-terminal 44-kDa extension of unknown function seen in the A. aeolicus A-adding enzyme (Aa.L) and T. maritima CCA-adding enzyme (Tm) (Fig. 1).
A BLAST search using the A. aeolicus CC-adding polypeptide (Aa.S) revealed that the genome of D. radiodurans (26) encodes an A.aS homolog designated DR.1 (accession number NP_294915). Amino acid identity between DR.1 and Aa.S is high over the entire length of the proteins. A BLAST search for homologs of the A. aeolicus A-adding enzyme (Aa.L) did not reveal any full-length homologs in D. radiodurans, but rather a smaller homolog, designated DR.2, corresponding to the Cterminal half of Aa.L (Aa.LC) (accession number NP_294707). Thus, three eubacteria (Synechocystis sp., A. aeolicus, and T. maritima) have NTR homologs with homologous N-terminal extensions of unknown function (Sy.L, Aa.L, and Tm), but D. radiodurans does not.
These data indicate that the Synechocystis Sy.S and Deinococcus DR.1 polypeptides are most closely related to A. aeolicus CC-adding enzyme, and the Synechocystis Sy.L and Deinococcus DR.2 polypeptides most closely related to the A. aeolicus A-adding enzyme. The implication is that CCA-adding activity in Synechocystis sp. and D. radiodurans, as in A. aeolicus, may be the joint responsibility of two distinct but related polypeptides, although the N-terminal extension found in the Aa.L and Tm homologs is absent from DR.2.

CCA-adding Activity Can Be Reconstituted by the Synechocystis Sy.S CC-adding (Sy.S) and A-adding (Sy.LC) Polypeptides-Synechocyctis
Sy.S and the C-terminal half of Sy.L (Sy.LC; amino acid residues 399 -942, where Thr-399 is changed to Met) were expressed in E. coli as hexahistidine-tagged proteins ( Fig. 2A). The recombinant polypeptides were assayed in the presence of both ATP and CTP using tRNA substrates lacking CCA, CA, or A (tRNA-D, tRNA-DC, and tRNA-DCC, respectively, where D is discriminator base at position 73). Sy.S adds one or more CMPs to tRNA-D and tRNA-DC, but not to tRNA-DCC, and does not add AMP to any tRNA substrate (Fig. 2B, lanes 1-3). Sy.LC adds AMP to tRNA-DCC, but not to tRNA-D or tRNA-DC, and does not add CMP to any tRNA substrate (Fig. 2B, lanes 4 -6). These data suggest that Sy.S and Sy.LC are likely to be CC-adding and A-adding enzymes, respectively. Indeed, reconstitution of the CCA-adding activity was observed when Sy.S and Sy.LC were combined (Fig. 2B, lanes 7-9), as demonstrated previously for the A. aeolicus polypeptides A.aLC and Aa.S (ref. 24; see also Fig.  2C). When the assays were performed in the presence of all four ribonucleotide triphosphates, only one of which was labeled, Sy.S added only CMP to tRNA-D and tRNA-DC, Sy.LC added only AMP to tRNA-DCC, and neither Sy.S nor Sy.LC added GMP or UMP to any tRNA substrate (Fig. 2D). We conclude that CCAadding activity in Synechocystis is divided between two closely related polypeptides with different activity and specificity, just as in A. aeolicus.

CCA-adding Activity Can Be Reconstituted by the D. radiodurans CC-adding (DR.1) and A-adding (DR.2) Polypeptides-
The D. radiodurans DR.1 and DR.2 polypeptides were expressed as C-terminally hexahistidine-tagged polypeptides in E. coli (Fig. 3A) and assayed using the same three tRNA substrates as for the Synechocystis polypeptides. DR.1 adds one or more CMPs to tRNA-D and tRNA-DC, but not to tRNA-DCC, and does not add AMP to any tRNA substrate (Fig. 3B, lanes 1-3). DR.2 adds AMP to tRNA-DCC, but not to tRNA-D or tRNA-DC, and does not add CMP to any tRNA substrate (Fig. 3B, lanes 4 -6). Thus DR.1 and DR.2 are likely to be CC-adding and A-adding enzymes, respectively. Moreover, as expected, CCA adding activity could be reconstituted when DR.1 and DR.2 were combined (Fig. 3B, lanes 7-9). The D. radiodurans DR.2 polypeptide lacks the N-terminal extension of unknown function found in the homologous Aa.L, Sy.L, and Tm polypeptides (Fig. 1). When the assays were performed using uniformly labeled tRNA substrates in the presence of all of four nucleotides, DR. 1 added one nucleotide to tRNA-D and two nucleotides to tRNA-DCC, DR.2 added only one nucleotide to tRNA-DCC, and together DR. 1 and DR.2 added three nucleotides to tRNA-D (Fig. 3C). DR.1 incorporates only CMP into tRNA lacking CCA and CA, whereas DR.2 incorporates only AMP into tRNA lacking 3Јterminal A (Fig. 3D). These are the same results obtained for the A. aeolicus Aa.S and Aa.LC polypeptides (24) and the Synechocystis Sy.S and Sy.LC polypeptides (Fig. 2, B-D). We conclude that the D. radiodurans CCA-adding activity is also divided between two closely related polypeptides of different activity and specificity as in A. aeolicus and Synechocystis sp.
A Phylogeny of Eubacterial CCA-, CC-, and A-adding Enzymes and Poly(A) Polymerases-Having found that CCA-adding activity in Synechocystis sp. and D. radiodurans represents a collaboration between two distinct but related polypeptides (Figs. 2 and 3) just as in A. aeolicus (24), we wanted to understand the phylogenetic relationship between eubacterial CCA-, CC-, and A-adding enzymes, as well as the related eubacterial poly(A) polymerases. All known CCA-adding enzymes belong to the nucleotidyltransferase superfamily, which can be divided into Class I and Class II enzymes (13). The archaeal CCAadding enzymes (Class I) share a highly homologous 45-kDa core, and the eubacterial and eukaryotic CCA-adding enzymes (Class II) share a highly homologous 25-kDa core; however, Class I and Class II enzymes exhibit little obvious homology with each other outside of the immediate vicinity of the nucleotidyltransferase active site signature (17). The prokaryotic poly(A) polymerases belong to Class II and share the 25-kDa core of the CCA-adding enzymes (13); the eukaryotic poly(A) polymerases belong to Class I (13) and exhibit very modest homology with the archaeal CCA-adding enzymes over roughly 25 kDa of the 45-kDa core. 2 To avoid needless complication, we compared only Class II enzymes. Because it is not yet possible to distinguish Class II CCA-adding enzymes from poly(A) polymerases based solely on sequence analysis (38), we included only those enzymes whose enzymatic activities had been experimentally demonstrated or could be confidently assumed. These are the H. influenzae CCAadding enzyme and poly(A) polymerase, both of which are almost identical to their experimentally characterized E. coli counterparts (21,23,39), the B. subtilis and M. leprae CCA-adding enzymes (38), and the CCA-adding enzyme from M. tuberculosis, which is almost identical to M. leprae enzyme (31). A multiple alignment of the 25-kDa core regions of these Class II enzymes, including the active site signature, is shown in Fig. 4.
As shown in Fig. 5, eubacterial nucleotidyltransferases are divided into four main groups (Groups 1-4) and two subgroups (Group 4a and 4b). dyltransferase, originally predicted to be poly(A) polymerase, is in fact a CCA-adding enzyme (38). The CC-adding enzymes (AA-CC, SY-CC, and DR-CC; Group 1) and A-adding enzymes (AA-A, SY-A, and DR-A; Group 3) group separately except for one CCA-adding enzyme (TM-CCA) that groups anomalously with the A-adding enzymes (see below for discussion).
Interestingly, although the three CC-adding enzymes group with each other, as do the three A-adding enzymes, the CC-and A-adding enzymes from each organism do not group together. This suggests that CC-and A-adding enzymes do not reflect gene duplication and diversification within each lineage. Instead, the division of CCA-adding activity between CC-and A-adding enzymes may have been the primitive state, consistent with placement of A. aeolicus, D. radiodurans, and T. maritima near the deepest root of the 16 S rRNA-based phylogenetic tree (40,41). This could also explain why the T. maritima CCA-adding enzyme groups with A-adding enzymes (Group 3) rather than with other CCA-or CC-adding enzymes; the progenitor T. maritima A-adding enzyme could have acquired CC-adding activity to become a CCA-adding enzyme, thus rendering the CC-adding enzyme redundant and subject to loss.
Synechocystis does not branch as early as A. aeolicus, T. maritima, and D. radiodurans in the phylogenetic tree based on rRNA (40,41). Yet, the Synechocystis CC-adding enzyme (SY-CC in Group 1) and A-adding enzyme (SY-A in Group 3) both group with the corresponding CC-and A-adding enzymes from A. aeolicus, D. radiodurans, and T. maritima. One possible explanation is that phylogenetic analysis based on proteins including DnaJ/K, EF-Tu, and DNA polymerase I revealed a close relationship between the Deinococcus-Thermus group and cyanobacteria (42,43). Thus, cyanobacterial CC-and A-adding activities may also represent the primitive state. Alternatively, A. aeolicus, D. radiodurans, and Synechocystis sp. may have independently acquired the CC-and A-adding enzymes by horizontal transfer; however, this would not explain why horizontal transfer was mainly or exclusively restricted to deeply rooted eubacteria.
The CCA-adding enzymes build a defined nucleic acid se-quence without using a nucleic acid template. A variety of models have been proposed to explain this remarkable behavior (18 -23), but the mechanism of CCA addition remains mysterious. Moreover, although eubacterial CCA-adding enzymes and poly(A) polymerases are closely related in sequence (13,38,44), it is still difficult to predict the activity from sequence alone (38,45). We have now found that CCA-adding activity is divided between two closely related nucleotidyltransferases in Synechocystis sp. and D. radiodurans (this paper) as well as in A. aeolicus (24). At least in A. aeolicus these two nucleotidyltransferases recognize the same face of tRNA 3 and work independently. Comparative biochemical and structural analysis of the related eubacterial CCA-, CC-, and A-adding enzymes, as well as eubacterial poly(A) polymerases, may help to explain the specificity of these unique enzymes and the evolutionary relationships among them.