A System for the Targeted Amplification of Bacterial Gene Clusters Multiplies

A system for the targeted ampliﬁcation of bacterial gene clusters multiplies antibiotic yield in Streptomyces coelicolor

Takeshi Murakamia,1, Jan Buriana, Koji Yanaib, Mervyn J. Bibbc, and Charles J. Thompsona

aDepartment of Microbiology and Immunology, University of British Columbia, Vancouver, BC, Canada, V6T 1Z3; bBioscience Laboratories, Meiji Seika Pharma, Odawara-shi, Kanagawa 250-0852, Japan; and cDepartment of Molecular Microbiology, John Innes Centre, Colney, Norwich NR4 7UH, United Kingdom

Edited* by Arnold L. Demain, Drew University, Madison, NJ, and approved August 12, 2011 (received for review May 25, 2011) Gene clusters found in bacterial species classified as Streptomyces ceutical and agricultural activities. However, when such gene encode the majority of known antibiotics as well as many phar- clusters are transferred to heterologous hosts, levels of pro- maceutically active compounds. A site-specific recombination sys- duction are often low (13–16), presumably reflecting important tem similar to those that mediate plasmid conjugation was metabolic and/or regulatory interactions with their native hosts. engineered to catalyze tandem amplification of one of these gene An important factor in the commercial production of anti- clusters in a heterologous Streptomyces species. Three genetic biotics and other pharmaceutically active secondary metabolites elements were known to be required for DNA amplification in is the cost of strain improvement for increased yields. Yield S. kanamyceticus: the oriT-like recombination sites RsA and RsB, improvement typically involves years of repeated cycles of mu- and ZouA, a site-specific relaxase similar to TraA proteins that tagenesis and screening, potentially generating mutants in both catalyze plasmid transfer. We inserted RsA and RsB sequences primary and secondary metabolism (17). For example, these into the S. coelicolor genome flanking a cluster of 22 genes (act) mutations may change primary metabolic flux to increase pre- responsible for biosynthesis of the polyketide antibiotic actino- cursor availability (17). Rate-limiting steps for antibiotic pro- rhodin. Recombination between RsA and RsB generated zouA- duction are also often associated with the secondary metabolic dependent DNA amplification resulting in 4–12 tandem copies of pathway itself, and some overproducing strains of Penicillium MICROBIOLOGY the act gene cluster averaging nine repeats per genome. This chrysogenum, S. lincolnensis, and S. kanamyceticus obtained in resulted in a 20-fold increase in actinorhodin production compared traditional screening programs contain amplifications of their with the parental strain. To determine whether the recombination antibiotic biosynthetic gene clusters (18–21). Detailed analysis of event required taxon-specific genetic effectors or generalized bac- penicillin production strains derived by repeated rounds of mu- terial recombination (recA), it was also analyzed in the heterolo- tagenesis and screening at different pharmaceutical companies gous host Escherichia coli. zouA was expressed under the control revealed that independent lineages contained a progressively of an inducible promoter in wild-type and recA mutant strains. A increasing number of tandem amplifications of the 57-kb bio- plasmid was constructed with recombination sites RsA and RsB synthetic gene cluster (18, 19). Our studies of a kanamycin-over- zouA bordering a drug resistance marker. Induction of expression producing strain of S. kanamyceticus revealed that its genome fi generated hybrid RsB/RsA sites, evidence of site-speci c recombi- contained 36 tandem copies of a 145-kb DNA sequence that recA nation that occurred independently of . ZouA-mediated DNA included the kanamycin biosynthetic gene cluster (21). A major fi ampli cation promises to be a valuable tool for increasing the group of secondary metabolites, the polyketides, have commercial activities of commercially important biosynthetic, degradative, and importance or potential applications as antibacterials, antifungals, photosynthetic pathways in a wide variety of organisms. antiparasitics, animal growth promotants, or immunosuppressants (22). Polyketide biosynthetic pathways are encoded by clusters gene duplication | mutagenesis of genes ranging in size from around 20 kb to more than 100 kb (23). These observations suggest that controlled, stable amplifi- andem amplifications of genomic DNA occur in all domains cation of entire antibiotic biosynthetic gene clusters would be Tof life including humans, plants, insects, yeast, and bacteria a valuable and generally applicable tool for engineering high- (1–4) and are proposed to be “the principle source of new genes” yielding production strains. (for references, see ref. 4). In bacteria, DNA amplification plays In Escherichia coli and Salmonella, gene duplication–amplifi- a role in antibiotic resistance, chromosome instability, gene cation (GDA) is typically initiated via recA-independent re- evolution, and increasing the level of gene expression (4–8). combination between microhomologous sequences (imperfect Regulated amplification of genes or gene clusters could also be matches of less than 20 bp) to generate a tandem duplication (3, a means of activating gene expression that is inherited over 6, 7, 24–27). Subsequent amplification requires recA-dependent subsequent generations (6). As a biotechnological tool, the in- recombination. Such amplifications are generally restricted in ducible amplification of specific regions of microbial genomes size and unstable without continuous selection or recA inacti- (amplifiable or amplified units of DNA; AUDs) could have im- vation (7, 25, 28). Although gene amplifications in Streptomyces portant applications in strain improvement for a wide variety of species have been described (29–31), the specific enzymes and complex multigene processes, such as the biosynthesis of pharmaceutically active metabolites and vitamins, bioconversions, photosynthesis, and the degradation of toxic compounds. Author contributions: T.M. and C.J.T. designed research; T.M. performed research; T.M., J.B., About half of all agriculturally and pharmaceutically impor- K.Y., M.J.B., and C.J.T. analyzed data; and T.M., J.B., K.Y., M.J.B., and C.J.T. wrote the paper. tant compounds, including the majority of antibiotics, are pro- The authors declare no conflict of interest. duced by Actinomycetes (most belonging to the Streptomyces *This Direct Submission article had a prearranged editor. . “ ” genus) or fungi (9) Almost all of these secondary metabolites Data deposition: The sequence reported in this paper has been deposited in the GenBank are encoded by gene clusters (9–12). When these clusters of database (accession no. JN005928). genes are expressed, intermediates of primary metabolism are 1To whom correspondence should be addressed. E-mail: [email protected]. redirected to alternative pathways, generating antibiotics and This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. other compounds with unusual structures and useful pharma- 1073/pnas.1108124108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1108124108 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 Cassette I Cassette II A 145kb zouA RsA RsB Fig. 1. Engineering zouA and its recombination kan cluster S. kanamyceticus sites RsA and RsB for amplification of an antibiotic 35kb biosynthetic gene cluster in a heterologous host. zouA RsA RsB (A) Regions of the S. kanamyceticus genome containing the three genetic elements essential for amplification (zouA, RsA, and RsB) were act cluster S. coelicolor cloned into two cassettes and targeted to flank (MT617) the actinorhodin biosynthetic gene cluster in 110kb the S. coelicolor genome by homologous re- zouA RsA RsB combination (Materials and Methods). act, actinorhodin biosynthetic genes; kan, kanamycin biosynthetic genes. (B) A schematic representa- act cluster S. coelicolor tion of the vectors delivering cassettes I and II and (MT6h17) the genomic area that is targeted. Double-cross- over clones were identified for pAB606 and B kanamycinR pAB606 pAB1004 pAB607 by screening for viomycin resistance and loss of kanamycin resistance. Double-cross-over viomycinR zouA RsA apramycinR RsB clones for pAB1004 were confirmed by Southern blot analyses. These plasmids were used to con- Cassette I Cassette II struct strains MT17 (pAB1004), MT617 (pAB1004 and pAB606), and MT717 (pAB1004 and pAB607). Actinorhodin Biosynthesis Cluster S. coelicolor Cassette I was inserted either between positions Cassette I SCO5070-SCO5092 138,040 and 138,051 (MT617) or at position RsA 23kb 63,080 (MT6h17) (S. coelicolor genome accession number AL6458821). In both MT617 and MT6h17, viomycinR kanamycinR pAB607 cassette II was inserted at nucleotide position 167,240.

sequences required to generate and maintain them are not Results known. In S. lividans, such amplifications are relatively short, Use of zouA and Flanking RsA and RsB Sites to Amplify an Antibiotic relying on two 4.7-kb repeats that amplify a 1-kb section of in- Biosynthetic Gene Cluster in S. coelicolor. To introduce zouA and tervening sequence (29, 31). RsA and RsB at sites flanking the actinorhodin biosynthetic gene A previous report described DNA amplification in S. kana- cluster (act)ofS. coelicolor, the three genetic elements were first myceticus that required zouA, two recombination sites (32), and subcloned as two cassettes (I and II) into an E. coli vector able perhaps other host-specific factors, whereas possible mechanisms to transfer to S. coelicolor by conjugation. These two plasmids of amplification were discussed in ref. 21. zouA encodes a TraA- contained, in addition, sequences homologous to those flanking like protein with two domains that are homologous to those that the act gene cluster (Fig. 1; strain and plasmid constructions are mediate plasmid conjugation in a wide variety of bacteria. One described in SI Materials and Methods). Because the plasmids are domain encodes a putative relaxase (similar to TrwC), and the unable to replicate in Streptomycetes, selection for the antibiotic other encodes a helicase/single-strand exonuclease (similar to resistance genes carried by each vector resulted in their insertion Tra_Ti). They are usually part of a protein complex that interacts into the S. coelicolor genome by homologous recombination with with specific sequences in plasmid origins of transfer (oriT)- the sequences flanking the act gene cluster. MT617 contained all mediating replication and recombination events that initiate and three elements, with RsA and RsB flanking the act cluster and terminate conjugal transfer of single-stranded DNA. In addition with an apramycin resistance gene located in the potential AUD. to TraA homologs, many plasmids rely on additional plasmid- or Two control strains were constructed that did not have a func- host-encoded proteins to catalyze a series of reactions needed tional zouA gene; one contained both recombination sites for plasmid transfer, including DNA bending, nicking, helical (MT717) and the other contained only RsB (MT17). unwinding, and recombination (33, 34). Transfer of Streptomyces plasmids is unlike all other bacterial conjugation systems in that Generating Actinorhodin-Overproducing Cultures by Selective Enrich- it requires only one plasmid-encoded protein (translocase; TraB) ment. To enrich for genetic variants that had amplified the act that recognizes plasmid sequence motifs and catalyzes transfer of gene cluster, M17, MT617, and MT717 were passaged five times unnicked double-stranded DNA (35). Although TraA homologs (generations one–five; G1–G5) by 1/25 dilution in medium are encoded by some Streptomyces plasmids, their role in plasmid containing progressively increasing concentrations of apramycin transfer has been questioned (36). In S. kanamyceticus, site- (50–800 μg/mL) (Fig. 2A). Samples of these cultures were in- − specific recombination takes place between two sequences (RsA oculated onto R5 agar medium (apramycin 50–500 μg/mL) to and RsB) flanking the 145-kb AUD. RsA and RsB, which are visualize production of the blue-pigmented and diffusible acti- likely to have functions analogous to oriT, are nearly identical in norhodin and to produce spore stocks (Fig. 2A). Whereas their first 16 bp (two mismatches). Tra genes (similar to zouA) MT617 G5 made large amounts of actinorhodin, little was pro- catalyze RecA-independent site-specific recombination (34) that duced by MT17 G5 and MT717 G5 (Fig. 2B). Spores from G1 may generate gene amplifications in plasmids (37). Here we and G5 were then used to inoculate liquid cultures lacking describe the use of genetic cassettes containing S. kanamyceticus apramycin. MT617 G5 produced progressively more intense blue zouA and recombination sites RsA and RsB to amplify a poly- pigment, reaching 450 μg/mL actinorhodin after 3 d in culture ketide (actinorhodin) biosynthetic gene cluster in its native (Fig. 2C). Much smaller amounts of actinorhodin were produced S. coelicolor A3(2) host to generate an overproducing strain. We by cultures of the parental strain and derivatives (MT17 and also demonstrate that the zouA, RsA/RsB-dependent recombi- MT717) that lacked a functional zouA gene (all produced less nation system functions in E. coli. than 20 μg/mL) (Fig. 2C).

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1108124108 Murakami et al. Downloaded by guest on October 2, 2021 A increase in copy number of the AUD correlated with increasing levels of selection for apramycin resistance. However, MT717 + + − − + − Spores (RsA , RsB , zouA ) and MT17 (RsA , RsB , zouA ), which did not possess the amplification, also grew in the presence of increasing concentrations of apramycin (G1–G5). Because the parental strain (S. coelicolor MT1110) did not grow, growth of the control strains was presumably caused by spontaneous mutations that increased the level of expression of the apramycin resistance gene or the specific activity of the enzyme it encodes. To analyze the correlation between the degree of amplification and the increase in actinorhodin production, total genomic B DNAs were digested with NheI, an enzyme that did not cleave the AUD (Fig. 3A). After separating fragments by pulsed-field gel electrophoresis (PFGE), they were hybridized to a probe adjacent to RsB (probe 1). The probe hybridized to the expected fragment (∼50 kb) containing a single copy of the AUD in genomic DNA from MT17 (G1, G5), MT617 (G1), and MT717 (G1, G5). In contrast, only amplified bands were observed in MT617 (G5) cultures that had been subjected to selection (Fig. 3C). After five generations, there were at least seven bands C ranging in size from 155 kb to >350 kb, demonstrating the presence of genomes that contained from 4 to more than 10 copies of the 35-kb AUD (Fig. 3C). AUD copy numbers were also estimated by comparing the strength of hybridization signals corresponding to a fragment within the tandem repeats with a fragment present as a single copy at the genome–AUD border. This was done using a restriction enzyme (BsrG1) that had one

cleavage site within the AUD sequence and a hybridization MICROBIOLOGY probe (probe 1) adjacent to RsB (Fig. 3A). The relative in- tensities of the repeated and border fragments were quantified, indicating that the AUD in this mixed culture was amplified on Fig. 2. Selection for increased copy number of an AUD containing the average about 10 times (Fig. 3D). Conceptually similar experi- actinorhodin gene cluster generates pigmented, overproducing cultures. (A) ments using a different restriction enzyme to probe smaller Overall experimental design used to select for amplification of the actino- fragments separated on standard agarose gels confirmed these rhodin gene cluster. Genetic components were transferred by conjugation results (Fig. S1). from E. coli to S. coelicolor MT1110 (53). Transconjugants were selected on To determine whether larger segments of the S. coelicolor Mannitol Soya Flour Agar (55) containing viomycin or apramycin as appro- genome could be similarly amplified, a strain was constructed priate, and recombinants were passaged in modified SOB medium (2% tryptone, 0.5% yeast extract, 0.05% NaCl, 2.5 mM KCl, MgCl2 omitted) containing cassette II in the same position adjacent to the act containing increasing concentrations of apramycin. Generations 1–5 were cluster but with cassette I now inserted at a site about 60 kb away − plated on R5 agar medium (apramycin 50–500 μg/mL) to visualize actino- from the act cluster, potentially generating an AUD of 110 kb rhodin production. (B) Visual comparison of actinorhodin production (blue (Fig. 1A). After growth in the presence of increasing concen- pigmentation) of G1 (Left) and G5 (Right) cultures of MT17 (Upper), MT617 trations of apramycin (MT6h17), the 1.2-kb PCR fragment in- − (Right), and MT717 (Left) grown on R5 agar medium (lacking apramycin) dicative of an RsB/RsA junction in the mixed culture was for 4 d at 30 °C. (C) Actinorhodin titers of liquid cultures of MT617(G1), detected (Fig. S2A) and confirmed by sequencing. This proved MT717(G1), MT617(G5), MT717(G5), and the parent strain MT1110 grown in that the recombination event occurred between RsA and RsB 25 mL of R5MS (58) at 30 °C, 150 rpm. Titer was calculated by measuring the sites separated by 110 kb. However, in contrast to the strain absorbance at 640 nm of culture supernatant after incubation in 1 N KOH. containing the 35-kb AUD (MT617), overproduction of actinorhodin was not observed (Fig. S2B). Quantitative PCR showed PCR was performed to determine whether increased actino- that the RsB/RsA junctions were infrequent (about 1 per 228 rhodin production in MT617 G5 correlated with tandem repeats genomes) for the 110-kb AUD, whereas the 35-kb AUD had ∼ of the act gene cluster within the presumed AUD. Genomic about eight junctions per genome, indicating an 1,800-fold fi DNAs of MT617 and MT717 from generations 1, 3, and 5 were difference in ampli cation. These genetic experiments demonstrated that zouA-dependent analyzed using primers designed to detect a potential re- fi combination event generated at the junctions of head-to-tail site-speci c recombination between RsA and RsB was needed to generate tandem amplifications of the act gene cluster encoded tandem repeats (Fig. 3A). If such tandem copies were generated within a 35-kb AUD and that this resulted in a 20-fold increase by site-specific recombination between RsA and RsB, they would in actinorhodin production. When the act gene cluster was in- generate a 1.2-kb PCR product that included a hybrid RsB/RsA cluded in a larger AUD of 110 kb, recombination occurred at sequence (Fig. 3A). A single PCR fragment of this size was in- a much reduced frequency and the cultures did not overproduce deed detected, and its intensity progressively increased as MT617 actinorhodin. was passaged in medium containing increasing concentrations of apramycin (G1, G3, and G5) (Fig. 3B). Nucleotide sequencing Inducible, ZouA-Mediated Recombination of RsA and RsB in E. coli. confirmed that it contained the recombinant RsB/RsA junction The discovery that only three components, RsA, RsB, and zouA, (Fig. 3A), as observed previously in S. kanamyceticus (21). Sim- catalyzed recombination in a heterologous Streptomyces species ilar PCR analyses of genomic DNAs isolated from the MT717 or led us to investigate whether this system could function in a dis- MT17 cultures (G1–G5) failed to amplify a 1.2-kb fragment. tantly related bacterium such as E. coli, and whether it was de- These results established that a specific amplification had oc- pendent on the universal bacterial enzyme (RecA) catalyzing − curred in MT617, that it was dependent on zouA, and that an homologous recombination. Strains of E. coli that were recA

Murakami et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 AB

Fig. 3. Evidence for amplification of the actinorhodin AUD. (A) A schematic representation of the single-copy AUD and a representative amplified tandem repeat (four copies). Positions of NheI, BsrGI, probe 1, and primers KM-16′ and KM-17′ are indicated with expected fragment sizes. (Inset) The sequences of native and recombinant Rs sites, along with their conserved 6-bp core. (B) Agarose gel electrophoresis of PCR products generated using primers KM-16′ and KM-17′ with genomic DNAs of MT617(G1), MT617(G3), MT617(G5), MT717(G1), MT717(G3), MT717(G5), and the parent strain MT1110 (lanes 1–7, respectively). Lanes 1–3 show the 1.2-kb fragment containing the RsB/RsA junction. (C and D)SYBRGold(Invitrogen)-stainedPFGE(Left) and Southern blot hybridization using probe 1 (Right). Genomic DNAs digested by (C)NheIor(D) BsrGI from MT617(G1), MT17(G1), MT717(G1), MT617(G5), MT17(G5), and MT717(G5) are shown in lanes 1–6, respectively. In C, the copy numbers of the AUD, determined from the lengths of the restriction fragments containing the total amplification, are indicated. PFGE was carried out under the following conditions: (C) 6 V/cm, angle 120, switch time 2–27 s, and run time 27 h; or (D) 6 V/cm, angle 120, switch time 0.5–8.5s,andruntime20h.

[JM109(DE3)] or recA+ [BL21(DE3)] were transformed with ieved by isolating large fragments of genomic DNA in heterol- two plasmids: pRSFZ, with zouA expression under the control of ogous hosts on plasmid vectors; however, the zouA-based system an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible pro- allowed relatively facile and stable amplification of a large region moter (Fig. S3), and pAB707 (GenBank accession no. JN005928), of a genome in its native host. The recombinant DNA methods which carried an engineered AUD containing an apramycin re- used here should be broadly applicable to Streptomyces and other sistance gene (Fig. 4A). The strains were grown in the presence Actinomycetes species, and could be used to increase the level of of apramycin (50 μg/mL) with and without IPTG. The formation production of commercially useful secondary metabolites or to of an RsB/RsA junction was examined by PCR using primers discover new pharmaceutically active compounds by the specific designed to amplify a 584-bp DNA fragment (Fig. 4B) containing amplification of cryptic gene clusters (38). the hybrid junction. Formation of the RsB/RsA junction was − induced by IPTG in both the recA and recA+ strains (Fig. 4B), and the predicted RsB/RsA junction (Fig. 3A) was confirmed by A sequencing. Thus, RsB/RsA recombination had occurred in a zouA-dependent, recA-independent manner. Due to the ex- tremely high levels of resistance conferred by the apramycin resistance gene in pAB707 (>4 mg/mL), selective enrichment of plasmids containing tandem amplifications was not possible. Discussion Our studies demonstrated that three genetic elements identified in S. kanamyceticus (zouA, encoding a putative relaxase, and the specific recombination sites, RsA and RsB) were able to amplify specific segments of DNA in S. coelicolor. Four- to 12-fold (with B an average of ninefold) amplification of a 35-kb region containing the act gene cluster resulted in a greater than 20-fold increase in actinorhodin production (MT617 G5). In contrast, when the same experiment was done using a strain (MT6h17) in which one of the recombination sites was located 60 kb away Fig. 4. Induced zouA expression catalyzes recombination in E. coli. E. coli from the act cluster (Fig. 1A), amplification occurred in less than cultures were assayed for site-specific recombination by PCR (Materials and 1% of the genomes. Although this difference could reflect the Methods). (A) Plasmid map of pAB707 indicating the RsA and RsB sites along increased distance that separated the Rs sites in MT6h17 com- with PCR primers specific for the right or left sides of both recombination sites. (B) Agarose gel electrophoresis of PCR products generated from ge- pared with MT617, it may also reflect the deleterious effects − fi fl nomic DNA isolated from E. coli JM109(DE3) (recA ; lanes 1–4) or BL21(DE3) caused by ampli cation of sequences anking the act gene (recA+; lanes 5–8) using primers KM-16′ and SbfLN to detect the 584-bp RsB/ cluster. In either case, it highlighted the importance of using site- RsA fusion fragment. E. coli hosts contained no plasmid (lanes 1 and 5), specific recombination to limit the amplification so as to include pAB707 (lanes 2 and 6), pAB707 and pRSFZ uninduced (lanes 3 and 7), or only the secondary metabolic gene cluster. This could be ach- pAB707 and pRSFZ induced with IPTG (lanes 4 and 8).

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1108124108 Murakami et al. Downloaded by guest on October 2, 2021 The complete genomic sequences of six actinobacteria (39–44) Controlled amplification of gene clusters is likely to have have revealed many previously unknown secondary metabolic important applications for increasing the productivity of com- pathways. Each of these genomes encodes from 20 to 36 putative mercially important processes in a wide range of bacteria. The secondary metabolic gene clusters whose unknown products zouA-mediated gene amplification in S. coelicolor reported here potentially include novel structures with pharmaceutical appli- suggests that it could be used readily and immediately to increase cations. zouA-based amplification of these cryptic gene clusters levels of antibiotic biosynthesis in many Actinomycetes and could be used to activate or enhance the production of their possibly other microorganisms. corresponding secondary metabolites, and contribute to much- needed drug (particularly antibiotic) discovery. Materials and Methods The concept that GDA allows accelerated evolution of Bacterial Strains. E. coli DH5α, JM109(DE3) (Promega; P9801), BW25113 (52) functions encoded by gene clusters is also important in the harboring pIJ790 (53), and ET12567 (54) harboring pUZ8002 (55) were used context of industrial strain development. As noted in the In- in this work. E. coli strains were cultivated in LB medium containing 1% troduction, overproducing strains generated by many cycles of tryptone, 0.5% yeast extract, and 0.5% NaCl supplemented with antibiotics empirical mutagenesis and screening often contain amplifica- as needed. Construction of plasmids for targeted recombination (PCR tar- tions of antibiotic gene clusters. If the genome contains multiple geting) in S. coelicolor MT1110 (56) was performed as described by Gust et al. (53). All other Streptomyces genetic manipulations were done according copies of a metabolic gene cluster, it can sustain higher rates of to Kieser et al. (55). mutagenesis, thus increasing the probability of beneficial mutations in genes encoding rate-limiting steps without loss of overall Oligonucleotides. Oligonucleotide sequences are presented in Table S1. cluster function. Thus, not only are strains with ZouA-catalyzed amplifications of gene clusters likely to produce more product Plasmid and Strain Constructions. Construction of pAB601, pAB701, and initially, they may also be ideal strains to use in traditional yield pAB801 was described previously (57). All other plasmids and strain con- improvement programs of mutagenesis and screening. structions are described in SI Materials and Methods. Many species of Streptomyces generate stable genomic amplifications at high frequencies (around 1% of the colonies) in the Culture Conditions for Actinorhodin Production. Seed cultures were grown at absence of obvious selective pressure (29, 45–47), in contrast to 30 °C in 25 mL modified SOB medium (2% tryptone, 0.5% yeast extract, the instability of recA-dependent gene amplifications in E. coli 0.05% NaCl, 2.5 mM KCl, MgCl2 omitted) with shaking at 140 rpm for 2 d. (7, 25, 28). In Streptomyces, as in other bacteria, the first step, After homogenization, 5 mL was inoculated into 25 mL of R5MS (58) liquid medium. The culture was grown with shaking at 150 rpm at 30 °C for 4 d. namely duplication, is rate-limiting (47). Our data strongly sug- − MICROBIOLOGY R5 medium (59) (pH 7.2) was used for agar-grown cultures (30 °C, 4 d). gest that zouA, or perhaps other homologs traditionally associ- Actinorhodin production was quantified by measuring the absorbance at ated with plasmid transfer, catalyze the initial duplication and/or 640 nm in 1 N KOH (55). the subsequent amplification. A TBLASTN analysis of the 30 sequenced Streptomyces genomes currently available on the Na- Pulsed-Field Gel Electrophoresis. S. coelicolor strains were grown in YEME (55) tional Center for Biotechnology Information/BLAST/microbe medium containing 0.5% glycine and 50 μg/mL apramycin for 48 h. Agarose site (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi) revealed plugs containing mycelium were prepared and restriction endonuclease significant matches only in Streptomyces sp. S4 [CADY01000194; digestions were performed (55). A CHEF-DR III Pulsed-Field Electrophoresis a block of 1,092/1,439 (76%) identical amino acids] and Strep- System (Bio-Rad) was used to resolve DNA digests (60) in 1% agarose gels tomyces sp. Mg1 [ABJF01000113; a block of 374/852 (44%) using 0.5% TBE (45 mM Tris-borate/1 mM EDTA) supplemented with 100 μM identical amino acids]. It is not known whether these proteins thiourea as the running buffer. Electrophoresis conditions varied depending fi catalyze gene amplification. Whereas apramycin was needed on the size of the DNA fragments to be resolved ( gure legends). to select for amplification of the act gene cluster in S. coelicolor, Southern Hybridization. DNA fragments from agarose gels were transferred to it was not needed to maintain the overproduction phenotype + ∼ fi Amersham Hybond-N sheets (GE Healthcare). Probe 1 (Fig. 3A) was an 3.4- over a period of ve passages in liquid culture; actinorhodin kb SbfI-NdeI fragment of pAB1002 containing the apramycin resistance gene overproduction, and presumably the corresponding GDA, was located adjacent to RsB. Probe 2 was an ∼3.5-kb NotI-XbaI fragment of stably maintained in the absence of apramycin selection (Fig. S4). pAB602 (Fig. 3A). These probes were isolated using QIAquick Gel Extraction Studies carried out in E. coli established that zouA could act as Kits (QIAGEN) and nonradioactively labeled using the ECL Amersham Kit (GE a site-specific recombinase in an evolutionary divergent taxon of Healthcare). Hybridization, washing, and detection were performed accord- bacteria. Remarkably, RsB/RsA junction fragments (Fig. 4B) ing to the manufacturer’s instructions. − were IPTG-inducible in both recA+ and recA hosts. By analogy to its ability to amplify genes in Streptomyces, ZouA may catalyze IPTG-Induced Duplication in E. coli JM109(DE3)- and E. coli BL21(DE3)-Containing amplification of the sequence flanked by RsA and RsB in Plasmids pAB707/pRSFZ. E. coli JM109(DE3)/pAB707/pRSFZ and E. coli BL21 pAB707 (Fig. 4A). Many site-specific recombination systems (DE3)/pAB707/pRSFZ were cultured overnight at 37 °C, 200 rpm in 5 mL LB medium supplemented with kanamycin, ampicillin, and apramycin (50 μg/mL have evolved to eliminate deleterious duplications in plasmids or each). Two hundred fifty microliters of the culture was used to inoculate chromosomes (48, 49). For example, Cre recombinase catalyzes fi 25 mL of LB medium supplemented with glucose (0.5% wt/vol), kanamycin, site-speci c recombination at the loxA site (34 bp) to resolve ampicillin, and apramycin (50 μg/mL each), and grown at 37 °C, 200 rpm to

multimers of the P1 plasmid (49). Whereas Cre and related an OD600 of 0.6. IPTG was added to the cultures to a final concentration of recombinases favor resolution of duplications, they also catalyze 1 mM, and the cultures were incubated overnight at 30 °C, 150 rpm. DNA the less favored reverse reaction that can generate gene dupli- was isolated from the cultures using the illustra Kit (GE Healthcare) per the cation (50, 51). Interestingly, although ZouA catalyzes a similar manufacturer’s instructions. Control cultures of E. coli JM109(DE3) and E. coli reaction between RsA and RsB, in contrast to Cre, we observed BL21(DE3) were grown in parallel without antibiotics and IPTG. PCR analysis gene amplification in Streptomyces. This could simply reflect of the DNA was carried out at an annealing temperature of 55 °C. selective pressure for high copy number of the resistance gene; however, the fact that cluster amplifications were stably main- PCR Conditions. All PCR primers are listed in Table S1. PCR was performed tained in S. coelicolor in the absence of selection suggested using TaKaRa LA Taq with GC Buffer (Takara Bio) according to the manufacturer’s instructions, using an annealing temperature of 52 °C unless that ZouA favors duplication. Although the ZouA-catalyzed otherwise specified. Reaction mixtures were supplemented with 0.5% di- recombination observed in E. coli may be indicative of gene am- methyl sulfoxide. All PCR was carried out with DNA isolated from cultures plification, further experiments will be required to isolate the using the illustra Kit (GE Healthcare) per the manufacturer’s instructions. For resulting recombinant plasmids using selectable drug resis- quantitative PCR, DNA samples were used undiluted or 1/4 diluted. In total, tance markers. 2.5 μL DNA was used per 25 μL reaction. A mixture of PerfeCTa SYBR Green

Murakami et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 Supermix (Quanta BioSciences), cDNA, and primers (0.5 μM each) was run on ACKNOWLEDGMENTS. We are grateful to Carol Ng, Gaye Sweet, Joshua a Bio-Rad Opticon2 (95 °C 3 min, 95 °C 30 s, 55 °C 30 s, 72 °C 30 s, read, and Chang Mell, and Dongchang Sun for their generous support in the lab, repeated from the second step 39 times). Plasmid DNA was used as a stan- Toyomi Sato for helpful discussions and encouragement in executing this ′ research, and Ryszard Brzezinski for plasmid pMG302M. This work was dard to calculate concentrations. Primers KM-16 and RsAL were used to supported by Canadian Natural Sciences and Engineering Research Council amplify the RsB/RsA junction, and primers OutU and OutL were used to Grant 293171-06 (to C.J.T.), Meiji Seika Kaisha Ltd, and funding from the UK amplify a region outside of the AUD as an internal control. Biotechnology and Biological Sciences Research Council (M.J.B.).

1. Sebat J, et al. (2004) Large-scale copy number polymorphism in the human genome. 30. Hornemann U, Zhang XY, Otto CJ (1993) Transferable Streptomyces DNA amplifica- Science 305:525e528. tion and coamplification of foreign DNA sequences. J Bacteriol 175:1126e1133. 2. Haber JE, Debatisse M (2006) Gene amplification: Yeast takes a turn. Cell 125: 31. Volff JN, Eichenseer C, Viell P, Piendl W, Altenbuchner J (1996) Nucleotide sequence 1237e1240. and role in DNA amplification of the direct repeats composing the amplifiable ele- 3. Andersson DI, Hughes D (2009) Gene amplification and adaptive evolution in bacte- ment AUD1 of Streptomyces lividans 66. Mol Microbiol 21:1037e1047. ria. Annu Rev Genet 43:167e195. 32. Murakami T, Sumida N, Bibb MJ, Yanai K (2011) ZouA, a putative relaxase, is essential 4. Bergthorsson U, Andersson DI, Roth JR (2007) Ohno’s dilemma: Evolution of new for DNA amplification in Streptomyces kanamyceticus. J Bacteriol 193:1815e1822. genes under continuous selection. Proc Natl Acad Sci USA 104:17004e17009. 33. Byrd DR, Matson SW (1997) Nicking by transesterification: The reaction catalysed by 5. Brochet M, Couvé E, Zouine M, Poyart C, Glaser P (2008) A naturally occurring gene a relaxase. Mol Microbiol 25:1011e1022. fi ampli cation leading to sulfonamide and trimethoprim resistance in Streptococcus 34. Garcillán-Barcia MP, Francia MV, de la Cruz F (2009) The diversity of conjugative re- e agalactiae. J Bacteriol 190:672 680. laxases and its application in plasmid classification. FEMS Microbiol Rev 33:657e687. fi 6. Sandegren L, Andersson DI (2009) Bacterial gene ampli cation: Implications for the 35. Vogelmann J, et al. (2011) Conjugal plasmid transfer in Streptomyces resembles e evolution of antibiotic resistance. Nat Rev Microbiol 7:578 588. bacterial chromosome segregation by FtsK/SpoIIIE. EMBO J 30:2246e2254. 7. Roth JR, et al. (1996) Rearrangements of the bacterial chromosome: Formation and 36. Zhang R, Zeng A, Fang P, Qin Z (2008) Characterization of replication and conjugation applications. Escherichia coli and Salmonella: Cellular and Molecular Biology, eds of Streptomyces circular plasmids pFP1 and pFP11 and their ability to propagate in Neidhardt FC, et al. (Am Soc Microbiol, Washington, DC), Vol 2, pp 2256e2276. linear mode with artificially attached telomeres. Appl Environ Microbiol 74: 8. Normark S, Edlund T, Grundström T, Bergström S, Wolf-Watz H (1977) Escherichia coli 3368e3376. K-12 mutants hyperproducing chromosomal β-lactamase by gene repetitions. J Bac- 37. Francia MV, Clewell DB (2002) Amplification of the tetracycline resistance de- teriol 132:912e922. terminant of pAM&alpha1 in Enterococcus faecalis requires a site-specificre- 9. Demain AL, Sanchez S (2009) Microbial drug discovery: 80 years of progress. J Antibiot combination event involving relaxase. J Bacteriol 184:5187e5193. (Tokyo) 62:5e16. 38. Zerikly M, Challis GL (2009) Strategies for the discovery of new natural products by 10. Zhang Y-Q, Wilkinson H, Keller NP, Tsitsigiannis D (2004) Secondary metabolite gene e clusters. Handbook of Industrial Microbiology, ed An Z (Marcel Dekker, New York), genome mining. ChemBioChem 10:625 633. pp 355e386. 39. Bentley SD, et al. (2002) Complete genome sequence of the model actinomycete e 11. Keller NP, Hohn TM (1997) Metabolic pathway gene clusters in filamentous fungi. Streptomyces coelicolor A3(2). Nature 417:141 147. Fungal Genet Biol 21:17e29. 40. Ikeda H, et al. (2003) Complete genome sequence and comparative analysis of the e 12. Strohl WR (1997) Introduction and approaches for improvement in antibiotic pro- industrial microorganism Streptomyces avermitilis. Nat Biotechnol 21:526 531. duction. Biotechnology of Antibiotics, ed Strohl WR (Marcel Dekker, New York), Vol 41. Ohnishi Y, et al. (2008) Genome sequence of the streptomycin-producing microor- e 82, pp 1e48. ganism Streptomyces griseus IFO 13350. J Bacteriol 190:4050 4060. 13. Penn J, et al. (2006) Heterologous production of daptomycin in Streptomyces lividans. 42. Oliynyk M, et al. (2007) Complete genome sequence of the erythromycin-producing J Ind Microbiol Biotechnol 33:121e128. bacterium Saccharopolyspora erythraea NRRL23338. Nat Biotechnol 25:447e453. 14. Thapa LP, et al. (2007) Heterologous expression of the kanamycin biosynthetic gene 43. Ishikawa J, et al. (2004) The complete genomic sequence of Nocardia farcinica IFM cluster (pSKC2) in Streptomyces venezuelae YJ003. Appl Microbiol Biotechnol 76: 10152. Proc Natl Acad Sci USA 101:14925e14930. 1357e1364. 44. Zhao W, et al. (2010) Complete genome sequence of the rifamycin SV-producing 15. Blodgett JA, Zhang JK, Metcalf WW (2005) Molecular cloning, sequence analysis, and Amycolatopsis mediterranei U32 revealed its genetic characteristics in phylogeny and heterologous expression of the phosphinothricin tripeptide biosynthetic gene cluster metabolism. Cell Res 20:1096e1108. from Streptomyces viridochromogenes DSM 40736. Antimicrob Agents Chemother 45. Fishman SE, Hershberger CL (1983) Amplified DNA in Streptomyces fradiae. J Bacteriol 49:230e240. 155:459e466. 16. Olano C, Lombó F, Méndez C, Salas JA (2008) Improving production of bioactive 46. Hasegawa M, et al. (1985) Certain chromosomal regions in Streptomyces glaucescens secondary metabolites in Actinomycetes by metabolic engineering. Metab Eng 10: tend to carry amplifications and deletions. Mol Gen Genet 200:375e384. 281e292. 47. Dyson P, Schrempf H (1987) Genetic instability and DNA amplification in Streptomyces 17. Baltz RH (1997) Molecular genetic approaches to yield improvement in Actino- lividans 66. J Bacteriol 169:4796e4803. mycetes. Biotechnology of Antibiotics, ed Strohl WR (Marcel Dekker, New York), Vol 48. Nash H (1996) Rearrangements of the bacterial chromosome: Formation and appli- 82, pp 49e79. cations. Escherichia coli and Salmonella: Cellular and Molecular Biology, eds fi 18. Fierro F, et al. (1995) The penicillin gene cluster is ampli ed in tandem repeats linked Neidhardt FC, et al. (Am Soc Microbiol, Washington, DC), Vol 2, pp 2363e2376. e by conserved hexanucleotide sequences. Proc Natl Acad Sci USA 92:6200 6204. 49. Austin S, Ziese M, Sternberg N (1981) A novel role for site-specific recombination in 19. Newbert RW, Barton B, Greaves P, Harper J, Turner G (1997) Analysis of a commer- maintenance of bacterial replicons. Cell 25:729e736. cially improved Penicillium chrysogenum strain series: Involvement of recombino- 50. Wang Y, Yau YY, Perkins-Balding D, Thomson JG (2011) Recombinase technology: fi genic regions in ampli cation and deletion of the penicillin biosynthesis gene cluster. Applications and possibilities. Plant Cell Rep 30:267e285. e J Ind Microbiol Biotechnol 19:18 27. 51. Sauer B (2002) Cre/lox: One more step in the taming of the genome. Endocrine 19: 20. Peschke U, Schmidt H, Zhang HZ, Piepersberg W (1995) Molecular characterization of 221e228. the lincomycin-production gene cluster of Streptomyces lincolnensis 78-11. Mol Mi- 52. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Es- crobiol 16:1137e1156. cherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97:6640e6645. 21. Yanai K, Murakami T, Bibb M (2006) Amplification of the entire kanamycin bio- 53. Gust B, Challis GL, Fowler K, Kieser T, Chater KF (2003) PCR-targeted Streptomyces synthetic gene cluster during empirical strain improvement of Streptomyces kana- gene replacement identifies a protein domain needed for biosynthesis of the ses- myceticus. Proc Natl Acad Sci USA 103:9661e9666. quiterpene soil odor geosmin. Proc Natl Acad Sci USA 100:1541e1546. 22. Staunton J, Weissman KJ (2001) Polyketide biosynthesis: A millennium review. Nat 54. MacNeil DJ, et al. (1992) Analysis of Streptomyces avermitilis genes required for Prod Rep 18:380e416. avermectin biosynthesis utilizing a novel integration vector. Gene 111:61e68. 23. Kim CG, et al. (2008) Biosynthesis of rubradirin as an ansamycin antibiotic from 55. Kieser T, Bibb MJ, Buttner MJ, Chater KF, Hopwood DA (2000) Practical Streptomyces Streptomyces achromogenes var. rubradiris NRRL3061. Arch Microbiol 189:463e473. Genetics (John Innes Foundation, Norwich, UK). 24. Edlund T, Normark S (1981) Recombination between short DNA homologies causes 56. Hindle Z, Smith CP (1994) Substrate induction and catabolite repression of the tandem duplication. Nature 292:269e271. 25. Tlsty TD, Albertini AM, Miller JH (1984) Gene amplification in the lac region of E. coli. Streptomyces coelicolor glycerol operon are mediated through the GylR protein. Mol e Cell 37:217e224. Microbiol 12:737 745. fi 26. Whoriskey SK, Nghiem VH, Leong PM, Masson JM, Miller JH (1987) Genetic re- 57. Murakami T, Sumida N, Yanai K (2009) Method for ampli cation of DNA in cell. arrangements and gene amplification in Escherichia coli: DNA sequences at the Patent Cooperation Treaty Appl WO 2009148149 (October 12, 2009). junctures of amplified gene fusions. Genes Dev 1:227e237. 58. Okamoto S, Taguchi T, Ochi K, Ichinose K (2009) Biosynthesis of actinorhodin and 27. Anderson RP, Roth JR (1977) Tandem genetic duplications in phage and bacteria. related antibiotics: Discovery of alternative routes for quinone formation encoded in Annu Rev Microbiol 31:473e505. the act gene cluster. Chem Biol 16:226e236. 28. Tyo KE, Ajikumar PK, Stephanopoulos G (2009) Stabilized gene duplication enables 59. Huang J, Lih CJ, Pan KH, Cohen SN (2001) Global analysis of growth phase responsive long-term selection-free heterologous pathway expression. Nat Biotechnol 27: gene expression and regulation of antibiotic biosynthetic pathways in Streptomyces 760e765. coelicolor using DNA microarrays. Genes Dev 15:3183e3192. 29. Young M, Cullum J (1987) A plausible mechanism for large-scale chromosomal DNA 60. Taupp M, Lee S, Hawley A, Yang J, Hallam SJ (2009) Large insert environmental ge- amplification in Streptomycetes. FEBS Lett 212:10e14. nomic library production. J Vis Exp 31: pii:1387.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1108124108 Murakami et al. Downloaded by guest on October 2, 2021