DNA Polymerase I Is Not Required for Replication of Linear Chromosomes in Streptomyces

Abstract

Both polA (encoding DNA polymerase I; Pol I) and a paralog were deleted from Streptomyces strains. Despite the UV sensitivity and slow growth caused by the ΔpolA mutation, the double mutant was viable. Thus, in contrast to a previous postulate, Pol I and its paralog are not essential for replication of Streptomyces chromosomes.

Replication of the linear chromosomes and linear plasmids of Streptomyces is initiated at an internal origin, and bidirectional replication leaves single-stranded gaps at the 3′ ends of the replicons. The 3′ overhangs are about 280 nt long (5) and contain several conserved and tightly packed palindrome sequences, which presumably form extensive secondary structures. The 3′ overhangs are believed to be “patched” by DNA synthesis primed by a terminal protein (TP) that remains covalently bound at the 5′ ends of the DNA (6). The TPs that cap the typical telomeres (Tpgs) are highly conserved in sequence and size (184 to 185 amino acids [aa]) (3, 20).

End patching requires another protein encoded by a gene (tap) upstream of tpg (1, 3). Tap binds to the secondary structure of the telomeres and to Tpg and presumably is involved in recruiting Tpg to the telomere for end patching (1). The chromosomes of mutants defective in tpg or tap lose the telomeres and become circularized.

In the extensively studied replication of bacteriophage φ29 and adenoviruses by TP-primed DNA synthesis, the DNA polymerases (Pols) involved belong to family B. No family B DNA polymerase gene (polB homolog) is found in Streptomyces coelicolor (4) or Streptomyces avermitilis (10). Genes encoding Pol I (polA), Pol III, and Pol IV (two homologs) are present in these Streptomyces genomes.

Of these DNA polymerases, Pol I was found to bind strongly to the Streptomyces telomere DNA (Y.-R. Lin et al., submitted for publication). Using Tap as a scaffold, Bao and Cohen (2) identified Pol I as a component of the Streptomyces telomere complex, which makes Pol I an attractive candidate for the polymerase that catalyzes end patching synthesis. However, this possibility could not be resolved, because these authors were unable to knock out the polA gene (SCO2003) in S. coelicolor M145 (with a linear chromosome) or Streptomyces lividans BKK019 (with a circularized chromosome). Instead, these authors proposed that Pol I was essential for replication of Streptomyces chromosomes. While that paper was being published, we had deleted polA in two related strains of S. coelicolor. Here we report the construction and partial characterization of these mutants and discuss their implications.

To delete polA in S. coelicolor, we employed the Redirect gene replacement procedure (7). A polA disrupting cassette consisting of the aac3(IV) gene (apramycin resistance) and an oriT was generated by PCR and used to transform Escherichia coli BW25113/pIJ790 harboring cosmid St7H2 of S. coelicolor (18), which contains polA, to create a ΔpolA::aac3(IV) allele on the cosmid by recombination (Fig. 1A). The ΔpolA::aac3(IV) allele was subsequently transferred conjugally from E. coli ET12567/pUZ8002 (17) to six S. coelicolor strains, M130 (8), M145 (8), M146 (8), J1508 (9), 3454 (9), and 3456 (19).

FIG. 1.

Creation and characterization of ΔpolA mutants. (A) Design. The polA and relevant surrounding genes in the wild type (top) are shown as open arrows. In the mutant (bottom), polA was replaced by a cassette (filled box) containing aac3(IV) (white arrow). Two hybridization probes, A and B, are indicated by open boxes. Ps, PstI site. The sizes (in kilobases) of the hybridization PstI fragments are indicated. (B) Hybridization analysis of the ΔpolA mutants. PstI-digested total DNA of each culture was electrophoresed in an agarose gel and hybridized to digoxigenin-labeled probes A (left panel) and B (right panel), as described by Kieser et al. (12). (C) UV sensitivity of the mutants (9). Filled circles, strain 3456; open circles, strain HT1-3; filled triangles, strain 3454; open triangles, strain HT4-4. The error bars indicate 1 standard deviation calculated by Poisson distribution based on the number of colonies scored. (D) Growth retardation of the mutants. The cultures were plated on LB agar (12) and incubated at 30° for 4 days.

Most of the apramycin-resistant (Apr^r) exconjugants selected were resistant to kanamycin (conferred by neo on the cosmid) and contained the cosmid integrated through a single crossover. Kanamycin-sensitive (Kan^s) segregants, in which the polA⁺ allele was removed through a second crossover after two to three sporulation cycles, were readily isolated on nutrient agar (Difco) from the Apr^r exconjugants of strains 3454 (7 Kan^s segregants/480 exconjugants) and 3456 (7/510) but not of strains M130 (0/1,000), M145 (0/200), M146 (0/1,300), and J1508 (0/1,200). In E. coli, either the 5′-to-3′ or 3′-to-5′ exonuclease activity of Pol I is required for growth on rich medium (11). An attempt to isolate double crossover mutants from M145 on a minimal medium (12) was made without success (0/400). Deletion of polA in the Apr^r Kan^s isolates from strains 3454 and 3456 was confirmed by Southern hybridization (Fig. 1A and B). Two ΔpolA mutants from each lineage were used for subsequent studies; only the results from one representative each (HT4-4 from 3454 and HT1-3 from 3456) are reported here.

To test the possibility that Pol I is required for end patching, the chromosome of the mutants was examined for the possible loss of the telomere by using Southern hybridization with the terminal 1.3-kb BamHI DNA of the S. coelicolor chromosome (9). All of the ΔpolA mutants retained intact telomeres (see Fig. S1 in the supplemental material), indicating that polA is not essential for end patching. This result is consistent with the ability of crude extracts from HT1-3 to support in vitro deoxynucleotidylation of TP (21).

The ΔpolA mutants exhibited increased UV sensitivity, as expected (Fig. 1C), and grew more slowly than the polA⁺ strains on various solid media, such as LB agar (Fig. 1D), nutrient agar (Difco), PYM (14), SFM, and minimal medium (12) (data not shown). The sizes of the colonies formed by the mutants were smaller than those of the wild type and variable, showing arrest of growth at different times (Fig. 1D). Such a growth defect was similar to that of ΔrecA mutants of Streptomyces (9) and probably reflects retardation in joining the Okazaki fragments (15, 16) and/or a deficiency in repair of chromosomal breakage during replication.

Our failure to introduce a ΔpolA mutation into M145 is consistent with the result of Bao and Cohen (2). The reason for the ease of generating ΔpolA mutations in strains 3454 and 3456 is not clear. Strain 3454 is a pgl mutant derived from M145 after UV irradiation (H. Kieser, personal communication), and 3456 (pgl SCP1^NF SCP2⁻) was derived from mating between 3454 and J1508.

Among the chromosomes of six strains studied, those of J1508 and M130 have long (1-Mb) terminal inverted repeats (9, 19), and those of M145, 3454, and 3456 are short (22 kb) (9, 19). Thus, the terminal-inverted-repeat length does not appear to affect the viability of the ΔpolA mutants. The pgl system, which is defective in strains 3454 and 3456, confers protection against phage φC31 by a specific attenuation system (13). J1508 and M146 are also pgl mutants, and their polA genes could not be deleted here. S. lividans is naturally pgl negative, and its polA gene could also not be deleted (2). Therefore, it is unlikely that the lack of pgl in strains 3454 and 3456 was essential for the viability of the ΔpolA mutants.

A polA paralog is present in the chromosome of S. coelicolor (SCO3434), S. avermitilis (SAV4636), and many other actinomycetes. The products of SCO3434 and SAV4636 are shorter than their polA paralogs (563 to 564 aa versus 907 aa), lacking the N-terminal 5′-3′ exonuclease domain but retaining the 3′-5′ exonuclease and polymerase domains. To test the possibility that this gene can suppress a lethal effect of a polA mutation in strains 3454 and 3456, the same gene replacement procedure was used to replace SCO3434 in HT1-3 with an aadA (spectinomycin resistance; Spc^r) cassette (Fig. 2A) (7). Spc^r Kan^s exoconjugants were readily isolated, and the deletion of SCO3434 in HT1-3 was confirmed by Southern hybridization. Figure 2B shows the results for one of the representative ΔSCO3434 ΔpolA double mutants, HD21-16. HD21-16 exhibited growth retardation similar to that of HT1-3 but a slight but reproducible increase in UV sensitivity (Fig. 2C). Its chromosome also remained linear (see Fig. S1 in the supplemental material). Knockout of SCO3434 in strains M145 and 3456 was similarly achieved using an aac3(IV) cassette (Fig. 2B), and the mutants (HL-5m and HL-13n) exhibited no detectable phenotypic changes (Fig. 2C). These results indicate that SCO3434 does not suppress a putative lethal effect of the ΔpolA mutation in strains 3454 and 3456, although it might play a minor role in DNA repair, which was detectable only in the ΔpolA background.

FIG. 2.

Creation and characterization of ΔSCO3434 ΔpolA double mutants. (A) Design. SCO3434 and relevant surrounding genes in the wild type (top) are shown as open arrows. In the mutant (bottom), SCO3434 was replaced by a cassette (filled box) containing aadA or aac3(IV) (white arrow). The digoxigenin-labeled hybridization probe C is indicated by an open box. Ba, BamHI site. The sizes of the hybridization BamHI fragments are indicated by numbers. (B) Hybridization analysis of the ΔSCO3434 mutants. BamHI-digested total DNA of each culture was electrophoresed in an agarose gel and hybridized to probe C. HL-5m, ΔSCO3434 mutant of M145; HL-13n, ΔSCO3434 mutant of 3456; HD21-16, ΔSCO3434 mutant of HT1-3. (C) UV sensitivity of the mutants (9). Filled circles, strain 3456; open circles, strain HL-13n; filled triangles, strain HT1-3; open triangles, strain HD21-16. The error bars indicate 1 standard deviation calculated by Poisson distribution based on the number of colonies scored. The results for M145 and HL-5m were identical to those for 3456 and HL-13n, respectively.

SCO4495 of S. coelicolor is incorrectly annotated as encoding a “putative DNA polymerase related protein” due to a UDP glycosylase domain shared with a phage SP01 DNA polymerase. No homolog of this gene is found in S. avermitilis (based on the genome sequence) or 14 other Streptomyces species (based on microarray analysis; R. Kirby, personal communication). Deletion of this gene did not cause detectable changes in growth, UV sensitivity, or chromosome topology. Besides containing polA and SCO3434/SAV4636, both S. coelicolor and S. avermitilis chromosomes contain two dinB genes encoding error-prone translesion repair DNA polymerases (Pol IV) and 10 genes encoding subunits of DNA polymerase III (Pol III), including duplicate dnaE (α subunit) and dnaN (β subunit) genes and triplicate dnaQ (ɛ subunit) genes. It is likely that end patching involves one of the polymerases encoded by these genes.