Mutational Analysis of the Terminal Protein Tpg of Streptomyces Chromosomes: Identification of the Deoxynucleotidylation Site

The linear chromosomes and linear plasmids of Streptomyces are capped by terminal proteins (TPs) covalently bound to the 5′ ends of the DNA. The TPs serve as primers for DNA synthesis that patches in the single-stranded gaps at the telomeres resulting from the bi-directional replication (‘end patching’). Typical Streptomyces TPs, designated Tpgs, are conserved in sequence and size (about 185 amino acids), and contain a predicted helix-turn-helix domain and a functional nuclear localization signal. The Tpg-encoding gene (tpg) is often accompanied by an upstream gene tap that encodes an essential telomere-associating protein. Five lone tpg variants (not accompanied by tap) from various Streptomyces species were tested, and three were found to be pseudogenes. The lone tpg variant on the SLP2 plasmid, although functional, still requires the presence of tap on the chromosome for end patching. Using a combination of in vitro deoxynucleotidylation, physical localization, and genetic analysis, we identified the threonine at position 114 (T114) in Tpg of Streptomyces lividans chromosome as the deoxynucleotidylated site. Interestingly, T114 could be substituted by a serine without destroying the priming activity of Tpg in vitro and in vivo. Such T114S substitution is seen in and a number of pseudogenes as well as functional Tpgs. T114 lies in a predicted coil flanked by two short helixes in a highly hydrophilic region. The location and structural arrangement of the deoxynucleotidylated site in Tpg is similar to those in the TPs of phage ø 29 and adenoviruses. However, these TPs are distinct in their sequences and sizes, indicating that they have evolved independently during evolution. Using naturally occurring and artificially created tpg variants, we further identified several amino acid residues in the N-terminus and the helix-turn-helix domain that were important for functionality.


Introduction
Soil bacteria of the genus Streptomyces possess linear chromosomes and linear plasmids that are capped by terminal proteins (TPs) covalently bound to the 59 ends of the DNA [1,2]. Replication of these linear replicons is accomplished in two steps: (i) bidirectional replication initiated from an internal origin, which results in single-stranded gaps at the 39 end; and (ii) patching of the single-stranded gaps by TP-primed DNA synthesis [3]. The TPs remain covalently attached to the telomeres of the linear Streptomyces replicons after replication.
The patching TP-primed DNA synthesis in Streptomyces differs from the replicative TP-primed synthesis involved in the replication of adenoviruses [4] and phage ø29 [5], which have been extensively studied. In end patching, only about 300 nt of single-stranded gaps need to be filled [3], whereas replicative TPprimed synthesis duplicates the whole replicons end-to-end.
In Streptomyces, most TPs are highly conserved in sequences and size (about 185 amino acids). These conserved TPs are designated archetypal TPs, and are encoded by tpg gene, which typically lie downstream of a tap gene in the same operon [6,7]. Tap is also essential for replication of linear Streptomyces replicons. It binds specifically to a secondary structure formed by the single-stranded 39 overhang during replication, and presumably recruits Tpg to the telomere location for the end patching reaction [8].
The tap-tpg operon generally lies in the terminal region of the Streptomyces chromosomes. A few linear plasmids also contain the tap-tpg operon or a lone tpg homolog. In addition, apparent tpg pseudogenes are found in sequenced Streptomyces genomes. The first lone tpg homolog was discovered in the 15.4-kb terminal sequences shared by the S. lividans chromosome and the right end of SLP2 [9]. This homolog, designated tpg SLP2. 38 , exhibited atypical codon usage and encoded a product (183-amino acid) with a number of deletions and insertions. It was presumably a pseudogene. Subsequently, more putative tpg pseudogenes have been found in the linear plasmids and chromosomes in many Streptomyces species, for example, S. avermitilis [10], S. ambofaciens [11], S. griseus [12,13], and S. violaceoruber [14].
The N-terminal halves of Tpgs contain a number of predicted functional and structural domains [7]: (i) a helix-turn-helix (HTH) domain overlapping a segment similar to part of the thumb domain of HIV reverse transcriptase; (ii) a mono-partite nuclear localization signal (NLS) that is functional in targeting nuclei of human [15] and plant [16] cells, but is not essential for end patching [15]; (iii) an amphiphilic ß-strand that may be involved in protein-protein interactions or protein-membrane interactions. Interactions between the Streptomyces telomeres in vivo have been recently demonstrated [17]. SCP1, a 350-kb linear plasmid in Streptomyces coelicolor A3(2), encodes its own TP, Tpc [18]. Tpc shares no homology with Tpgs, and is significantly larger (259 amino acids). The coding gene, tpc, is also downstream from a gene tac that is essential for replication of SCP1. Tpc, like Tpgs, contains an HTH domain in the Nterminal region and a functional bi-partite NLS in the central region [15].
Recently, Yang et al. [19] demonstrated in vitro deoxynucleotidylation of Tpg Sli of S. lividans, in which dCMP, the first nucleotide at the 59 ends of the S. lividans chromosome, is specifically covalently attached to Tpg Sli in the in vitro reactions. The dCMP was attached to a Thr residue of Tpg Sli . There are 11 Thr residues in Tpg Sli . In comparison, the adenovirus [20] and ø29 [21] DNAs are attached to a Ser residue on their TPs. Other than these, little is known about the mechanism of end patching in replication of linear Streptomyces replicons.
In this study, combining biochemical and genetic approaches, we identified T114 at the C-terminus of Tpg Sli as the attachment site of the telomere DNA. Interestingly, this residue may be substituted by a Ser without destroying the deoxynucleotidylation function of Tpg in vitro and the priming function in vivo. Examination of five lone tpg homologs identified three of them to be pseudogenes. Interestingly all these pseudogenes have the T114S substitution, and therefore the defect must lie somewhere else. Moreover, we identified several amino acid residues in the neighborhood of T114, the N-terminus, and the helix-turn-helix (HTH) domain of Tpg Sli that were important for the functionality of Tpg Sli .

Bacterial Cultures and Molecular Manipulations
Genetic manipulations of E. coli and Streptomyces were performed according to the methods of Kieser et al. [22]. E. coli BL21(DE52) (Stratagene) and S. lividans MR04 [23] were used for expression of protein. pLUS980 plasmid containing a 4.5-kb BglII fragment spanning the tap-tpg operon from S. coelicolor cosmid 8D11 [24] was obtained from Chia-Hui Ke.

Protein Expression in E. coli BL21(DE52) and Purification
All the proteins used here were constructed in vector pet-15b (Novagen) where His6 is tagged to the N-terminal, or in pet-22b to the C-terminal with the purpose of purification, or vector pRSET A (Invitrogen) without His6-tagging. Protein expression was induced by IPTG (isopropyl-b-D-1-thiogalactopyranoside). The cells were centrifuged and sonicated in A8 buffer (50 mM TrisHCl pH8, 20 mM NaCl, 10% glycerol) and the His6-tagged protein purified using nickel-bound resin (Bioman) according to the protocols provided by the manufacturer. Insoluble His6-tagged proteins were purified in denatured state using nickel-bound resin and eluted in Elution buffer containing 6 M urea, refolded in Elution buffer by step-wise dialysis, and stored at 220uC in Elution buffer containing 5% DMSO (dimethyl sulfoxide). The concentrations of proteins were determined by the Bradford method [25]. Tpg without the His6-tag was purified by electrophoresis in 12% SDS-PAGE as described previously [19].

In vitro Deoxynucleotidylation and Purification of the Labeled Tpg
In vitro deoxynucleotidylation of untagged Tpg in cell extracts with alpha-[ 32 P]-dCMP was based on the procedure described previously [19]. The labeled proteins were purified by SDS-PAGE or by immuno-precipitation using protein A-Sepharose and anti-Tpg Sli antibody. With this procedure, Tpg with various lengths of deoxynucleotides were recovered.

Phosphoamino Acids Analysis
The procedures of Pargellis et al. [26] and Garcia et al. [27] were followed with minor modifications. Purified alpha-[ 32 P]-dCMPlabeled Tpg Sli protein was treated in 50 ml of 5.7 N HCl at 110uC for 1-2 h. The hydrolytes were dried and resuspended in 10 ml of water. A sample containing about 50 cpm was mixed with nonradioactive standards (1 mg each phosposerine, phosphothreonine, and phosphotyrosine; Sigma), and spotted on a cellulose thin layer plate (Merck) and subjected to two-dimensional electrophoresis in the Hunter Thin Layer Electrophoresis System (C.B.S. Scientific Company). The first dimension was carried out at pH 1.9 at 1.5 kV for 20 m and the second dimension in pH 3.5 at 1.3 kV for 16 m. The internal standards (marked with dotted circles) were visualized by spraying with 0.25% ninhydrin in acetone, and the radioactivity was imaged by autoradiography.

Fragmentation of the TP-dCMP Adduct
Alpha-[ 32 P]-dCMP-labeled Tpg Sli was isolated from a polyacrylamide gel by elution and precipitation with TCA, and cleaved with CNBr (Sigma) according to the published procedure [28] with minor modifications. The reaction products were separated on a 16% polyacrylamide gel formulated for the analysis of small peptides [29], followed by autoradiography without drying.
For proteolytic digestion with endoproteinase LysC (Roche), alpha-[ 32 P]-dCMP-labeled Tpg Sli was isolated using protein A-Sepharose beads, precipitated with TCA, and cleaved with LysC at 25uC for 20 h in a 50-ml solution containing [ 32 P]-labeled TP-dCMP, 1 mg enzyme, 25 mM Tris-HCl (pH 8.8), 1 mM EDTA and 0.1 M or 0.3 M of urea, and stopped by vacuum drying. The products were analysis by SDS-PAGE and autoradiography.

Mass Spectrometric Analysis
Native dCMP-TP complexes for the purpose of mass spectrometric analysis were gained from linear plasmid pLUS980L harbored in Streptomyces through four steps of purification as described [19], whereas TpgC was just separated by 12% SDSPAGE because it locates alone against a clear background. Mass spectrometric analysis was operated at either Genomics Center, National Yang Ming University (ESI and MALDI) or Molecular Medicine Research Center, Chang-Gung University (MALDI).

Construction of Mutated tpg Genes
Two-step PCR was employed to create mutations in tpg. The altered sequences were confirmed by sequencing before being used.
Construction of Linear Plasmids and Test for Linearity of the DNA Linear plasmids were constructed following the general procedure of Qin et al. [30]. The linearity of the plasmid DNA in the transformants was confirmed by restriction digestion and Southern hybridization. To facilitate manipulation of the tpg sequence, an NdeI site was added immediately upstream of and including the initiation codon of tpg on pLUS980. The resulting plasmid was designated pLUS980(Nd). Linear plasmids were generated from these plasmids and their derivative using the previously described procedure [30,31], i.e., linearization of the plasmid DNA by AseI digestion in the E. coli vector sequence followed by transformation of Streptomyces.
Various tpg homologs, including the putative pseudogenes, were generated by PCR and used to substitute tpg Sli on pLUS980 or pLUS980(Nd) through a series of restriction, subcloning, and ligation manipulations such that tpg in the end products was precisely replaced by these sequences without any alteration of the upstream tap sequence.
To test the linearity of plasmid DNA, genomic DNA was isolated from Streptomyces host (mostly MR04) harboring pLUS980, pLUS980(Nd), or their derivatives, digested with SacI or SpeI (which cuts uniquely in these plasmids), electrophoresed in agarose gel, with or without further Southern hybridization. The DNA isolation step involved the use of pronase E, which removed the TP caps on any linear plasmid DNA, allowing it to enter the gel. The linearity is demonstrated by the presence of two restriction fragments of the expected sizes. Circular plasmids, resulting from either incomplete linearization (by AseI digestion) before transformation or spontaneous circularization after transformation, would produce only a restrictions fragment.

Localization of the Deoxynucleotidylation Site in a Cterminal Fragment of Tpg Sli
In vitro deoxynucleotidylation of TPs typically selects specifically the nucleotide corresponding to the first nucleotide at the 59 end of the replicons, such as dCMP for adenovirus-2 [35], dAMP for phage ø29 [21], dCMP for linear plasmid pAL1of Arthrobacter nitroguajacolicus [36], dGMP for linear plasmid SCP1 of Streptomyces (Tsai, H.-H, unpublished data), and dCMP for the S. lividans chromosome [19]. In the case of ø29, the in vitro deoxynucleotidylation site was shown to be the attaching site for the telomere DNA in vivo [21]. In the case of S. lividans chromosome, dCMP was shown to attach to a Thr residue of its TP (designated Tpg Sli ) in in vitro deoxynucleotidylation [19]. Therefore, it was likely that the Thr residue of Tpg Sli deoxynucleotidylated in vitro was also deoxynucleotidylated in vivo. There are 11 Thr residues on Tpg Sli . We set out to determine the deoxynucleotidylated Thr.
The [ 32 P]-labeled Tpg Sli protein generated in in vitro deoxynucleotidylation was fragmented by CNBr and LysC endoproteinase digestion. Of the three polypeptides produced by CNBr digestion, S 2 ,M 28 (2.7 kD), R 29 ,M 44 (1.8 kD), and L 45 ,L 185 (16.4 kD), the isotope was present in L 45 ,L 185 (Fig. 1A). In the LysC digest, the isotope was present in the largest polypeptide fragment, which exhibited an apparent molecular weight of about 9 kD (Fig. 1B). The largest LysC polypeptide fragment, A 91 ,K 156 , of Tpg Sli , was expected to have a molecular weight of 7.0 kD. We interpreted the increased apparent molecular weight to result from the attachment of the dCMP residue(s). A 91 ,K 156 was a subset of L 45 ,L 185 that contained seven Thr residues. These results suggested that the deoxynucleotidylated Thr residue lay in the C-terminal region of Tpg Sli .
To narrow down the location of the deoxynucleotidylated Thr in the C-terminal region, we employed mass spectrometry on trypsin-digested Tpg Sli to identify the trypsin fragment that contained the deoxynucleotidylated Thr by its altered molecular weight. Tpg Sli -capped DNA was isolated by guanidine HCl-CsCl density gradient centrifugation followed by glass bead binding, and the DNA was trimmed by benzonase, an endonuclease that hydrolyzes DNA into 2-to 3-bp fragments (according to the manufacturer's specification). The resulting Tpg Sli was isolated by SDS-PAGE, digested with trypsin, and subjected to mass spectrometric analysis by ESI or MALDI ionization. It was anticipated that, if a fragment contained the deoxynucleotidylated Thr, the fragment with the native molecular weight would be absent from the sample (Fig. 1C).
In the C-terminal region, the only expected trypsin fragment that was not detected by either ESI or MALDI was the L 105 ,R 118 fragment. A fragment with a mass corresponding to that of (dCMP) 2 -L 105 ,R 118 was detected in the MALDI spectrum. In contrast, trypsin digestion of Tpg Sli protein produced in E. coli (presumably without deoxynucleotidylation) gave rise to the expected L 105 ,R 118 trypsin fragment in both ESI and MALDI spectra. These results indicated that the L 105 ,R 118 fragment contained the deoxynucleotidylation. The L 105 ,R 118 fragment lies in the A 91 ,K 156 LysC fragment and contains two Thr residues, T108 and T114.

Identification of the Deoxynucleotidylation Site at T114
Site-directed mutagenesis was subsequently employed to replace T108 and T114 with different residues in tpg Sli to test the functionality of the resulting mutant Tpg Sli . To do this, the mutant tpg gene was placed on pLUS980 or its derivative pLUS980(Nd) (Fig. 2A). These two plasmids contained a linear plasmid sequence consisting of a pair of telomeres of the S. lividans chromosome flanking a thiostrepton resistance gene (tsr), an autonomously replicating sequence (ARS) of linear plasmid pSLA2, and the taptpg operon of the S. lividans chromosome. These two plasmids, upon linearization (by AseI digestion in the E. coli sequence) and transformation [30,31], generated linear plasmids with Tpg-capped telomeres, designated pLUS980L and pLUS980(Nd)L, respectively (Fig. 2B).
S. lividans MR04, in which the tap Sli -tpg Sli operon was deleted along with large stretches of terminal DNA from the chromosome [23], was chosen as the host. If the tpg variant on pLUS980 and pLUS980(Nd) were defective, transformation of MR04 using the AseI-linearized plasmid DNA would produce either no transformants, or a few transformants harboring only circular plasmids. The latter would result from the plasmids that had escaped AseI digestion or linear fragments that had circularized in the transformants [37]. The topology of the plasmids in the transformants was determined by restriction digestion. For example, SpeI, which cuts singly in pLUS980 and pLUS980(Nd) DNA, would produce two fragments of 5.1 and 8.0 kb from the linear plasmid DNA, but would produce a single larger fragment from the circular plasmid DNA (Fig. 2C).
Firstly, T108 and T114 were individually mutated to Ser in Tpg Sli . Interestingly, both the resulting plasmids, pLUS980-T108S ( Fig. S1 in Supporting Information) and pLUS980-T114S (Fig. 2C, lane 6), could replicate in linear form in MR04, indicating that neither of the mutations inactivated Tpg Sli . Similarly, the T108S and T114S mutations were individually created in tpg SLP2. 19 , and again these mutations did not inactivate Tpg SLP2. 19 (data not shown). These results suggested that either none of these Thr residues was the deoxynucleotidylation site, or the deoxynucleotidylated Thr could be substituted by a Ser without losing its function. Substitutions of four other Thr residues at 101, 123, 143 and 176 positions in the C-terminal region also did not destroy the ability of Tpg to support replication of the linear plasmid ( Fig. S2 in Supporting Information).
Next, T108 and T114 of Tpg Sli were substituted by three other residues -Ala, Cys, and Tyr. None of these substitutions at T108 inactivated Tpg Sli . In contrast, the substitutions at T114 gave rise to transformants at very low frequencies (about three orders of magnitude lower), which harbored only circular plasmids (Fig. 2C,  lanes 2, 4, 5). Substitution of T114 by a His, which might provide an amino group to form a covalent bond with the nucleotide [38], also failed to support replication of the linear plasmid (Fig. 2C,  lane 3). These results eliminated the role of T108 as the site of deoxynucleotidylation, leaving T114 as the final candidate.
If T114 was the deoxynucleotidylation site in vivo, the Ser residue that substituted it in the T114S mutants would be expected to be deoxynucleotidylated, and this was tested in an in vitro deoxynucleotidylation assay using His-tagged Tpg Sli and Tpg Sli -T114S proteins produced in E. coli. In the in vitro deoxynucleotidylation reaction, approximately equal weights of Tpg Sli and Tpg Sli -T114S were labeled by radioactive dCMP. The alpha-[ 32 P]-dCMP labeled Tpg Sli and Tpg Sli -T114S were subjected to acid hydrolysis, and chromatography on cellulose thin layer plate [19]. The results showed that the labeled dCMP was attached to a Ser residue on Tpg Sli -T114S (Fig. 3 middle panel), while it was attached to a Thr residue of Tpg Sli (Fig. 3. left panel). The efficiency of deoxynucleotidylation for Tpg Sli -T114S was approximately 30% of that for Tpg Sli (Fig. 3, right panel). These results supported the idea that T114 was the deoxynucleotidylation site in vivo and that it could be functionally substituted by Ser albeit possibly with a reduced efficiency of deoxynucleotidylation.
In the Tpg Sli sequence, T114 is followed closely by two negatively charged Asp residues (D115 and D117 in Tpg Sli ). In the ø29 TP, the deoxynucleotidylated S232 is also flanked by two negatively charged Asp (D231) and Glu (E233). Introduction of a D115A mutation destroyed the Tpg function, whereas a D117A mutation had no effect on the Tpg function (Fig. S3B, Supporting Information). Interestingly, D115 could be substituted by a Glu (D115E mutation) but not by a Asn (D115N mutation) without destroying the Tpg function (Fig. S3B, Supporting Information), indicating that a negative charge there is important.
The residue upstream of T114 is also a Ser. A S113A mutation had no effect on the Tpg functionality ( Fig. S1 in Supporting Information).

Testing Tpg Variants for Support of Replication of Linear Plasmids
Examination of the aa sequences of 17 Tpg homologs (including the conceptual translation products of the putative pseudogenes; Fig. 4) revealed that 12 contained T114 and five contained S114. Of the five S114-containing tpg homologs, tpg Sav_39 , which encodes a product of only about one half length (95 aa), is most likely a pseudogene. The other four S114-containing homologs, tpg SLP2.38 , tpg pFRL1.6 , tpg SAP1_11 , and tpg pSV2.102 , along with the T114containing tpg SLP2. 19 were tested for their ability to support the replication of linear plasmids. All these tpg homologs are present on a linear plasmid unaccompanied by a tap homolog. Of these, tpg SLP2. 19 and tpg SLP2.38 are on the same (SLP2) plasmid [9], with tpg SLP2.19 lying in the left arm and tpg SLP2.38 lying in the 15.4-kb right arm sequence, which is shared by the ends of the S. lividans chromosome [39].
These tpg homologs were placed on pLUS980 and pLUS980(Nd), and tested for their ability to support linear replication in MR04. Of the five putative pseudogenes tested, tpg SLP2.38 , tpg SAP1_11 , and tpg pSV2.102 could not support replication of linear plasmids in MR04 (Fig. S4 in Supporting Information),   indicating that they were indeed pseudogenes. tpg SLP2. 19 and tpg pFRL1.6 could support replication of linear plasmids in MR04 (Fig. S4 in Supporting Information). It was probably not surprising that these two genes are functional, because the encoded Tpg homologs are relatively conserved and the codon usage is typical for Streptomyces (high G/C at the third positions of the codons). However, Tpg pFRL1.6 contains a 13-amino acid extension at the Nterminus. Interestingly, the pseudogene product Tpg SAP1_1 is highly similar to Tpg pFRL1.6 in sequence including the 13-amino acid extension at the N-terminus. The defect of this product was likely to be due to one or more amino acid substitutions elsewhere. tpg SLP2. 19 , when placed downstream of tap on pLUS980L, was presumably transcribed from the same promoter as tap. To test whether the lone tpg SLP2.19 on SLP2 was expressed, and, if so, whether it alone was sufficient for supporting the replication of SLP2, attempts were made to use SLP2tsr, an SLP2 derivative with tsr inserted in Tn4811 [39], to transform MR04 (Dtap-tpg) and ZX7 (parent of MR04, tap-tpg + ). Thio r transformants were readily obtained in ZX7, but not in MR04. This indicated that either tpg SLP2. 19 was not expressed on SLP2, or the participation of tap was required as shown in the replication of the Streptomyces chromosomes [8]. To check this, tap Sli was inserted into the øC31 att site on the MR04 chromosome using the integrative plasmid pSET152 [40]. The resulting strain, CK03, could then be transformed successfully with SLP2tsr. This result indicated that tpg SLP2.19 on SLP2 was functional, and the presence of tap elsewhere was required for the replication of SLP2.

Analysis of the N-terminal Region of Tpg
Compared to most other Tpgs, the pseudogene product Tpg pSV2.102 lacks eight amino acid residues at the N-terminus (Fig. 4). To examine the importance of these eight amino acids, three mutations in Tpg Sli were created with a deletion of eight (residue 2 to 9), four (2 to 5), and two (2 to 3) amino acids from the N-terminus, and the effects of these deletions were tested. The results showed that, the deletions of eight and four amino acids inactivated Tpg Sli , but that of two amino acids did not (Fig. S5, Supporting Information). In addition to the N-terminal truncation, the 47 th residue (based on Tpg Sli ) of Tpg pSV2.102 , located in the 'turn' of the predicted HTH motif of Tpg, is a Thr, while Ala or Ile are present at this position in the other Tpgs (Fig. 4). To test the possible effect of the Thr substitution, a V47T mutation was introduced into Tpg Sli . The resulting Tpg Sli was functional, indicating that the V47T substitution had not effect on the functionality of Tpg Sli .
Tpgs possess a putative helix-turn-helix (HTH) motif despite the fact that the in vitro DNA-binding specificity appears to be low [6](Yi-Hong Chen, unpublished results). The atypical TP, Tpc, also contains a HTH domain with a different sequence [18]. To examine the importance of the predicted HTH domain in Tpg sequences (Fig. 4), several amino acids in the helix domains of Tpg Sli were substituted by different residues to lower their potential to form HTH (Table 1). In the first helix of the HTH motif, the G37A and A40P mutations (which reduced the probability of HTH formation to 25%) did not cause a defective Tpg. Quadruple mutation, R18S-A26V-G37A-M44V, generated by error-prone PCR, inactivated Tpg Sli (data not shown). Of these four mutations, G37A and M44V were in the HTH motif. Tpg Sli containing the G37A-M44V double mutation had a 71% probability of HTH formation, and exhibited no effect. Therefore, the defect caused by the quadruple mutations must be attributed to the other two upstream mutations. Moreover, an A42D mutation (insignificant probability of HTH formation) inactivated Tpg Sli (Fig. S6, Supporting Information). Perhaps the collapse of the first helix might not be critical, but the defect was caused by repulsion of DNA by the negative charge of the Asp residue. These results are summarized in Table 1 and Fig. 4.
The second helix of the HTH motif is more hydrophilic. The R54P and Y55F mutations (71% and 90% probability of HTH formation, respectively) resulted in a defective Tpg Sli (Fig. S6, Supporting Information). Possibly the positively charge R54 was involved in interaction with a phosphate group on the DNA. A V52D mutation in this region that introduced a negatively charged amino acid and reduced the predicted probability of HTH formation to 71% also resulted in a defective Tpg Sli (Fig. S6, Supporting Information).
These results indicated the importance of the HTH motif of Tpg Sli in end patching. This is in contrast to the NLS motif that immediately follows the HTH motif, which may be mutated without affecting replication of the linear plasmids [15],

Discussion
In this study, we have identified T114 to be the site of deoxynucleotidylation of Tpg Sli . A substitution of T114 with Tyr, Cys, or His inactivated Tpg Sli . A substitution with Ser did not, which is not surprising, because the functional lone Tpg pFRL1.6 also contains a Ser at this position (Fig. 4). For ø29 TP, a substitution of the deoxynucleotidylated residue S232 with Thr inactivates the TP [41], but a substitution with Cys produces a TP with a reduced (7%) efficiency of deoxynucleotidylation [42].
It is noteworthy that the amino acid residue immediate upstream of T114 is also a Ser. When T114 became defective in deoxynucleotidylation through a His/Ala/Cys/Tyr substitution, S113 could not serve as the deoxynucleotidylation site as S114 could. Thus, it appeared that, while Ser might serve as a deoxynucleotidylation, the deoxynucleotidylation is specific at position 114.
The different effects of T114Y and T114S substitutions are interesting. While both Tyr and Ser possess a priming hydroxyl group for DNA synthesis, the inability of Tyr to serve the priming function may be due to its bulkier and relatively more hydrophobic side chain. Secondary structure predication by the methods of Garnier, et al. [33] places T114 in a coiled domain flanked by two short helix segments in Tpg Sli (Fig. 5 top). This is similar to the situations of the TPs of ø29 and adenoviruses. That the deoxynucleotidylated Ser residues in these two viral TPs lie in a predicted coil segment flanked by two short helixes has been previously noticed [21]. In the case of the TPs of ø29, this prediction was confirmed in the TP-DNA polymerase c-crystal structure [44].
Hydropathy prediction showed that T114 lay in a highly hydrophilic region, presumably close to the surface of the protein (Fig. 5 bottom). The deoxynucleotidylated S232 of ø29 [45] and S557 of adenovirus [20] also lie in highly hydrophilic regions (Fig. 5 bottom). In contrast to these, the deoxynucleotidylated Y190 of PRD1 TP lies in a ß-structure enclosed in a hydrophobic area [45].
Despite these similarities in geometric properties, there is no similarity in sequence and size among the TPs in the different systems. The patching TP-primed DNA synthesis in Streptomyces also differs from the replicative TP-primed DNA synthesis (ø29, adenovirus, and PRD1) in that the former requires a proper DNA template for in vitro deoxynucleotidylation [19], whereas the latter does not. Moreover, the telomere sequences in the former consist of extensive palindromic sequences with potentials of forming complex secondary structures, which is not found in the telomeres of the latter. It follows that the telomere associating protein, Tap, that recognizes the secondary structure formed by the 39 overhang at the telomeres is also absent from the replicative TP-primed synthesis systems. All these considerations, plus the use of different DNA polymerases by these systems [43], lead to a conclusion that these different TP-primed DNA synthesis systems have evolved independently, recruiting different proteins with a suitable priming function in the process. Even among Streptomyces, there are atypical, heterologous TPs capping linear plasmids and linear chromosomes, one of which (Tap of SCP1) has been characterized [18]. In addition to the variations in TP families in Streptomyces, the frequent occurrence of tpg pseudogenes is also remarkable for an essential housekeeping gene. In this study, we have demonstrated three defective tpg pseudogenes out of five candidates. It is unlikely that the pseudogenes have arisen by gene duplication, and the absence of an accompanying tap gene argues against it. More likely, the lone tpg homologs were acquired through horizontal transfer from a plasmid. The findings of the tpg pseudogenes in the terminal regions of the linear replicons of Streptomyces, where exchanges with other linear replicons are frequent, also support this notion. The abundance of tpg pseudogenes suggests that such horizontal events are relatively frequent.
SLP2 is interesting in that it contains two lone tpg homologs, the functional tpg SLP2.19 and the pseudogene tpg SLP2. 38 . Although tpg SLP2.19 is functional, a tap gene present elsewhere is nonetheless required for the replication of SLP2. This finding answers the previous observations that SLP2 could not replicate in certain S. lividans mutants, whose chromosomes had suffered circularization and deletions (in the terminal regions where the tap-tpg operon resides) [44]. Apparently, tap was the chromosomal gene in the deletion that was required for replication of SLP2. The requirement of tap for replication of SLP2 is the same for that of the linear chromosomes in Streptomyces [8]. The evolutionary significance of the emergence of lone functional tpg is obscure. Figure S1 Topology of pLUS980(Nd) derivatives containing substitutions at T108 or T114 of tpg. (A) Physical maps of plasmids pLUS980 and pLUS980(Nd). (B) Genomic DNA containing a pLUS980 derivative (except lanes 2 and 3) was isolated from MRO4, digested with SpeI (Sp), and subjected to agarose gel electrophoresis. The four substitutions at T108 and at T114 are indicated by the substituting amino acids. Lane M, I kb DNA ladders as size markers. Lanes 1-5, plasmid DNA (purified or in total genomic DNA) of known topology and size serving as controls and markers: 1, SpeI-digested genomic DNA from MR04 containing pLUS980L (fragment size: 8.0 and 5.1 kb); 2, AseIdigested pLUS980 (largest fragment 14.3 kb); 3, SpeI-digested pLUS980 (15.6 kb); lane 4: SpeI-digested genomic DNA from MR04 containing SpeI-digested pLUS980(Nd)L (8.0 and 5.1 kb); 5, SpeI-digested genomic DNA from MR04 containing pLUS980(Nd)L with a S113A substitution in tpg (8.0 and 5.1 kb). The sizes (in kb) of the SpeI fragments are indicated. The sizes of the circularized plasmid DNA derived from the four linearized pLUS980(Nd) derivatives with a substitution at T114 varied from transformants to transformants, depending on the end joining sites. Shown here are representative cases.