In Vivo Manipulation of the Bleomycin Biosynthetic Gene Cluster in Streptomyces verticillus ATCC15003 Revealing New Insights into Its Biosynthetic Pathway*

Bleomycin (BLM), an important clinically used antitumor compound, and its analogs are challenging to prepare by chemical synthesis. Genetic engineering of the biosynthetic pathway in the producer strain would provide an efficient and convenient method of generating new derivatives of this complex molecule in vivo. However, the BLM producing Streptomyces verticillus ATCC15003 has been refractory to all means of introducing plasmid DNA into its cells for nearly two decades. Several years after cloning and identification of the bleomycin biosynthetic gene cluster, this study demonstrates, for the first time, genetic accessibility of this pharmaceutically relevant producer strain by intergeneric Escherichia coli-Streptomyces conjugation. Gene replacement and in-frame deletion mutants were created by λRED-mediated PCR targeting mutagenesis, and the secondary metabolite profile of the resultant mutants confirmed the identity of the BLM biosynthetic gene cluster and established its boundaries. Ultimately, the in-frame blmD deletion mutant strain S. verticillus SB5 resulted in the production of a bleomycin intermediate. The structure of this compound, decarbamoyl-BLM, was elucidated, and its DNA cleavage activity was compared with the parent compounds.

The bleomycins (BLMs) 3 (1,2) are important clinically used hybrid peptide-polyketide antitumor compounds. Combined with other agents, the BLMs are used clinically for the treatment of several types of tumors and marketed under the trade name Blenoxane, with BLM A2 and B2 as the principle constituents (3). Early development of drug resistance and cumu-lative pulmonary toxicity are the major limitations of BLMs in chemotherapy (3,4). Therefore, it is an important research goal to develop strategies to produce novel BLM analogs by microbial fermentation, particularly those unavailable or extremely difficult to prepare by chemical synthesis.
The BLMs are thought to exert their biological effects through a sequence-selective, metal-dependent oxidative cleavage of DNA and RNA in the presence of oxygen (5,6). They can be dissected into four functional domains: (i) the metal-binding domain, which consists of the pyrimidoblamic acid subunit along with the adjacent ␤-hydroxyl histidine; (ii) the bithiazole and C-terminal amine, which confers the majority of BLM-DNA affinity; (iii) the (2S,3S,4R)-4-amino-3-hydroxy-2-methylpentanoic acid linker subunit, which plays an important role in the efficiency of DNA cleavage by BLMs; and (iv) the carbamoylated disaccharide moiety (4). The exact functional role of the sugars and the attached carbamoyl group remains controversial. They possibly contribute to cell recognition, cellular uptake of BLMs, metal ion coordination, and/or DNA affinity (4,6).
The investigation of biosynthetic pathways and generation of new natural product analogs by genetic engineering critically depends on the availability of tools for recombinant DNA work in the producing organism. In general, this can be achieved by the development of an expedient genetic system for the native producer of the respective compound. Alternative solutions to this most commonly used approach are represented by the heterologous expression of the entire biosynthetic gene cluster in a suitable host strain or by the identification of genetically accessible producers of structurally closely related compounds. Although mobilization of entire biosynthetic pathways is often challenging because of their size or host-specific limitations (7,8), producer strains of similar compounds can be found only in very few cases.
We have been studying the biosynthesis of BLM in Streptomyces verticillus ATCC15003 (9 -16) and the closely related compounds tallysomycin in Streptoalloteichus hindustanus E465-94 ATCC31158 (17) and zorbamycin in Streptomyces flavoviridis ATCC21892 (18). Feeding experiments with isotopelabeled precursors and isolation of biosynthetic intermediates suggested that the BLM aglycone is derived from nine amino acids and one acetate, with methionine serving as two methyl donors (16). Cloning and analysis of the BLM, and more recently tallysomycin, and zorbamycin biosynthetic gene clus-ters suggested that this family of compounds is formed via a hybrid nonribosomal peptide synthetase (NRPS)-polyketide synthase (PKS)-catalyzed enzymology from amino acid and carboxylic acid precursors (10,12,14,17,18).
Despite exhaustive efforts, S. verticillus remained refractory to all means of introducing plasmid DNA into its cells, a fact that has so far prevented us from carrying out genetic analysis of the BLM biosynthetic locus in vivo. Protoplast-mediated transformation was inhibited by high lysozyme resistance of the S. verticillus mycelia as well as low frequencies of protoplast formation and regeneration, Escherichia coli-Streptomyces conjugation was complicated by poor sporulation characteristics of S. verticillus. All attempts to carry out transformation by electroporation proved to be unsuccessful. Only in one case was successful plasmid DNA transfer into S. verticillus reported by phage transduction. The plasmid, however, could not be recovered from these mutants (19). Here we present (i) the development of a genetic system for the BLM producer S. verticillus, (ii) in vivo confirmation and boundary determination for the BLM biosynthetic gene cluster by gene inactivation in S. verticillus, and (iii) isolation of a BLM pathway intermediate, decarbamoyl-BLM, followed by its chemical and biochemical characterization.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-S. verticillus ATCC15003 (American Type Culture Collection, Manassas, VA) and recombinant strains generated in this study were routinely grown at 28°C in tryptic soy broth (TSB) liquid medium (20). ISP4 medium supplemented with 0.05% yeast extract, 0.1% tryptone, and MgCl 2 to a final concentration of 20 mM was used to prepare S. verticillus spores and to plate out conjugation mixtures. For BLM production, S. verticillus and recombinant strains were cultured in 250-ml baffled flasks containing 50 ml of the previously described seed medium (10) for 2 days at 28°C and 250 rpm. Then, 5 ml of the seed cultures were inoculated into 250-ml baffled flasks containing 50 ml of the production medium (10) supplemented with 5 l of 3-morpholino-propylamine (3-MOP) and fermented at 28°C and 250 rpm for 6 -8 days. E. coli XL1 Blue MR (Stratagene, La Jolla, CA) and E. coli DH5␣ (21) were used for routine subcloning, plasmid preparations, and cosmid library preparation. E. coli S17-1 (20) or E. coli ET12567 (pUZ8002) (22) were the donor strains in E. coli-S. verticillus conjugation, and E. coli BW25113/pIJ790 was the host for RED-mediated PCR targeting mutagenesis (23).
Genetic Manipulation of S. verticillus ATCC15003-E. coli-S. verticillus conjugation was carried out using a protocol based on standard literature conditions (20). In brief, ϳ10 9 -10 10 S. verticillus spores were heat-shocked in 500 l of TSB medium supplemented with 0.4% glycine and 10.0% sucrose at 50°C for 10 min, followed by incubation at 30°C for 3-4 h as S. verticillus recipients. E. coli S17-1 containing the respective plasmid was grown in LB medium to an A 600 of 0.4 -0.6. Cells from 10 ml of culture were pelleted, washed twice with LB medium, and resuspended in 500 l of LB medium as the E. coli donors. For conjugation, the donors (500 l) and recipients (500 l, 10 9 -10 10 spores) were mixed and distributed onto ISP4 plates supplemented with 0.05% yeast extract, 0.1% tryptone, and MgCl 2 to a final concentration of 20 mM. The plates were incubated at 30°C for 16 -20 h and overlaid with 1 ml of water containing final concentrations of 50 g/ml nalidixic acid to select against E. coli and 100 g/ml apramycin to select for S. verticillus exconjugants. Incubation was continued at 30°C until exconjugants appeared.
DNA Isolation and Manipulation-DNA isolations and manipulations in E. coli (21) and Streptomyces (20) were carried out according to standard procedures. For Southern analysis, digoxigenin labeling of DNA probes, hybridization, and detection were performed according to the protocols provided by the manufacturer (Roche Applied Science). A cosmid library of S. verticillus chromosomal DNA was constructed in SuperCos1 according to the manufacturer protocol (Stratagene, La Jolla, CA). This library was screened with digoxigenin-labeled fragments derived from the BLM biosynthetic gene cluster (10) as probes. A set of four cosmids, pBS37, pBS38, pBS39, and pBS40, covering the entire BLM biosynthetic gene cluster, was isolated and used for gene inactivation experiments.
Sequencing and Computer-assisted Sequence Analysis-Sequencing of an additional 5385 bp beyond the downstream end of the published sequences (accession numbers AF210249 (10) and L26955 (28)) was performed by primer walking using the dideoxynucleotide chain termination method. Detailed procedures are provided in the supplemental data. The DNA sequence reported in this work was determined from pBS40 and has been deposited at GenBank TM under accession number EU805703.
Cloned Inactivation Constructs for Gene Replacement of blmIV (NRPS)-A 6.4-kb DNA fragment containing part of blmIV was cloned into pSET151 and separated into a 4.1-kb left and 2.3-right arm by the inserted aac(3)IV gene to afford the first inactivation construct for blmIV, pBS41.
To generate the second blmIV inactivation construct, a 22.5-kb DNA fragment containing blmIV was cloned into pSET151, and a 5.9-kb internal fragment was then replaced with the aac(3)IV gene to afford pBS42. Regions of 9.8-and 7.8-kb in size flanking the aac(3)IV gene at the upstream and downstream end, respectively, were available for homologous recombination.
To generate an in-frame deletion of blmD, the aac(3)IV-oriT cassette was then removed by FLP recombination from pBS47, resulting in pBS48. The bla gene on the SuperCos1 backbone of pBS48 was finally replaced by the aac(3)IV-oriT cassette from pSET152 via RED-mediated PCR targeting (23) using the previously described oligonucleotides f-pMT3 and r-pMT3 (17) to yield pBS49.
Isolation and Analysis of BLM and Decarbamoyl-BLM-BLM and decarbamoyl-BLM were first isolated as copper complexes. Copper-free BLM and decarbamoyl-BLM were obtained by treating the copper complexes with 0.5 M EDTA-Na, pH 7.3, solution. The identity of BLM was confirmed by liquid chromatography-electrospray ionization-mass spectrometry on an Agilent 1100 HPLC-MSD SL quadrupole mass spectrometer. The structure of decarbamoyl-BLM was established by spectroscopic analyses. Decarbamoyl-BLM-Cu complex: light blue powder; electro- DNA Cleavage Assay-The DNA cleavage assays were performed in 10 l (total volume) of 25 mM Tris-HCl buffer, pH 7.5, containing ϳ20 ng of pBluescript SK II(ϩ) plasmid DNA, 10 M Fe(NH 4 ) 2 (SO 4 ) 2 ⅐6 H 2 O (freshly prepared solution in 1 mM H 2 SO 4 ) and the appropriate concentrations of BLM A2, BLM B2, or 3-MOP-decarbamoyl-BLM. The reactions were incubated at 35°C for 30 min and stopped by the addition of 5 mM EDTA and 5 l of loading dye (30% glycerol containing 0.25% (w/v) bromphenol blue). The samples were applied to a 0.8% agarose gel containing 1 g/ml ethidium bromide, and gel electrophoresis was carried out in 40 mM Tris acetate buffer, pH 8.0, containing 1 mM disodium EDTA at 90 V for 1 h. The DNA bands were evaluated under UV light.
Nucleotide Sequence Accession Number-The nucleotide sequences reported in this study are available in the GenBank TM data base under accession numbers L26955, AF210249, and EU805703.

Development of a Genetic System for S. verticillus
ATCC15003-Aiming at the establishment of a genetic system for the BLM producer, the growth and sporulation conditions were evaluated and optimized. S. verticillus grows optimally between 28 and 30°C. Common media such as TSB, ISP2, and R2YE (20) are suitable for vegetative growth of S. verticillus. However, for genomic DNA isolation, it is necessary to supplement TSB with 10% (w/v) sucrose and 0.4% (w/v) glycine to facilitate lysozyme digestion. SGGP medium (29) supplemented with 10% or 20% (w/v) sucrose and YEME medium (20) supplemented with 34% (w/v) sucrose proved most suitable for protoplast preparation. S. verticillus sporulates very poorly on common growth media such as TSB, ISP2, R2YE, and CM medium. Slightly green spore formation to an acceptable extent (ϳ10 7 spores/plate) was observed after 10 days of growth at 30°C on ISP4 medium or ISP4 medium supplemented with 0.05% yeast extract, 0.1% tryptone, and MgCl 2 to a final concentration of 20 mM. The BLM producer is fairly sensitive to apramycin, thiostrepton, and kanamycin but fully resistant to even high concentrations (up to 200 g/ml) of hygromycin; therefore, 50 -150 g/ml of apramycin and 25-60 g/ml of thiostrepton were used throughout this study for selection.
Intergeneric conjugation allowed the introduction of two of six plasmid vectors, pSET152 and pRT801, into S. verticillus. Although S. verticillus mycelium of the early exponential growth phase can be protoplasted in a respectable yield (20 -30%), the protoplasts cannot be regenerated efficiently under all conditions tested (Յ10%). All attempts to introduce plasmid DNA into S. verticillus by protoplast-mediated transformation failed to produce any true transformants, and this technique was subsequently abandoned. In contrast, we were successful in developing a protocol for conjugation between E. coli S17-1 and S. verticillus. Following a literature procedure (20) with modifications, we found that conjugation between E. coli S17-1 and S. verticillus occurred with the highest frequency on modified ISP-4 agar freshly supplemented with 20 mM MgCl 2 when 10 9 -10 10 S. verticillus recipient cells and E. coli S17-1 donor cells harvested from 10 ml of culture grown to an A 600 of 0.4 -0.6 were used. Utilization of the methylation deficient E. coli ET12567/ pUZ8002 as donor strain failed to result in higher exconjugant frequencies than the methylation proficient E. coli S17-1. The true nature of exconjugants was verified by PCR and Southern blot hybridization.
Homologous Recombination in S. verticillus ATCC15003-Gene replacement for blmIV and blmVIII via double cross-over homologous recombination was carried out to confirm, for the first time, the identity of the BLM biosynthetic gene cluster by gene inactivation. Construct pBS41, which contains a 6.4-kb fragment of blmIV separated into a 4.1-kb upstream and a 2.3-kb downstream region by the insertion of the aac(3)IV apramycin resistance gene, was first chosen for this experiment. Construct pBS42, which harbors a 9.8-kb upstream fragment and a 7.8-kb downstream fragment, was also generated for homologous recombination. No true exconjugants were obtained after conjugal transfer of these plasmids into S. verticillus.
A newly prepared cosmid library of S. verticillus genomic DNA was then employed for gene inactivation by REDmediated PCR targeting mutagenesis (23). Replacement of the NRPS encoding blmIV gene by the aac(3)IV gene on pBS38 generated 18.9-kb upstream and a 12.9-kb downstream flanking regions accessible for homologous recombination in pBS43 (Fig. 1A). Similarly, inactivation of the PKS encoding blmVIII gene on pBS37 yielded 23.5-kb upstream and 12.2-kb downstream regions in pBS44 (Fig. 1C). After introduction of pBS43 and pBS44 into S. verticillus by conjugation, exconjugants resistant to apramycin were isolated, and the double cross-over mutants SB1 and SB2, respectively, were selected (Fig. 1, B and D). Both mutant strains completely lost their ability to produce BLM (Fig. 2, traces IV and III), confirming that the NRPS encoding blmIV and the PKS encoding blmVIII are indeed required for BLM biosynthesis. The conjugation frequencies observed for both inactivation constructs were in the same range as for the integrative model plasmid pSET152 (10 Ϫ7 -10 Ϫ8 exconjugants/cfu).
Determination of the Boundaries for the BLM Biosynthetic Locus by Gene Inactivation-Through inactivation of selected genes (blm-orf31/32 and blm-orf(Ϫ1)) residing at distal ends of the BLM biosynthetic gene cluster, the upstream and downstream blm cluster boundaries were assigned to reside between blm-orf31/32 and blm-orf30 and between blm-orf7 and blmorf(Ϫ1), respectively (supplemental Fig. S2). At the upstream end of the cluster, blm-orf30 and blm-orf29 had previously been proposed to play a role in the regulatory and self-resistance mechanisms in BLM production, respectively (10). The predicted gene products of blm-orf31 (a protein with similarity to phosphonoacetate hydrolases) and blm-orf32 (an NRPS protein) were thought to be dispensible for BLM biosynthesis. Gene replacement of blm-orf31 together with part of blm-orf32 yielded the mutant strain SB3 that exhibited BLM production levels similar to the wild-type S. verticillus strain (ϳ95% of the wild-type strain; Fig. 2, traces II and I).
At the downstream boundary, blm-orf7 encoding a transport protein with a possible function in BLM resistance mechanisms of the producer strain represented the last gene of the sequenced region (accession number L26955) (28). Sequencing and analysis of ϳ5.4-kb beyond the published nucleotide sequence revealed a ϳ3.2-kb gap without any apparent open reading frames downstream of blm-orf7 followed by blmorf(Ϫ1). The blm-orf(Ϫ1) gene product (802 amino acids) exhibits high similarity to hydrolases such as a putative glycosyl hydrolase from Streptomyces ambofaciens (accession number CAJ88378; 67% identity and 76% similarity) and a hypothetical protein from Streptomyces coelicolor (accession number NP_624928; 68% identity and 76% similarity). Inactivation of blm-orf(Ϫ1) yielded the mutant strain SB4 that had little effect on BLM production (ϳ95% of the wild-type strain; Fig. 2, traces VI and I), confirming that its product is not required for BLM biosynthesis.
In-frame Deletion of blmD Resulting in the Accumulation of a Pathway Intermediate-Inactivation of the blmD gene resulted in the accumulation of a pathway intermediate, decarbamoyl-BLM, which was characterized by MS and NMR spectroscopic analyses. In-frame deletion was chosen for the inactivation of the carbamoyl transferase encoding blmD gene. This strategy ensures the generation of nonpolar mutants, in contrast to gene replacement mutants bearing an antibiotic resistance cassette. The selected double cross-over mutant, SB5, completely lost its ability to produce BLM and instead accumulated a new compound, decarbamoyl-BLM, with a slightly different retention time (Fig. 2, traces V and I). This new compound (6 -8 mg/liter) was isolated, and the pure intermediate (10.0 mg) was obtained as a copper complex. The decarbamoyl-BLM-Cu complex was further treated with EDTA to remove copper and subsequently subjected to HPLC on a C18 column to afford the copper-free molecule (8.0 mg) as a pale white powder.
DNA Cleavage Activity of Decarbamoyl-BLM-Loss of the carbamoyl group attached to the BLM-disaccharide resulted in a decrease in DNA cleavage activity by a factor of 10. Decarbamoyl-BLM was compared with BLM A2 and B2 for its ability to cleave pBluescript II SK(ϩ) supercoiled plasmid DNA in the presence of Fe 2ϩ . BLM-mediated single-strand cleavage first results in the conversion of supercoiled plasmid DNA (form A) to open circular plasmid DNA (form B), and double-strand cleavage subsequently generates linearized plasmid DNA (form C). In this assay, BLM A2 and B2 showed nearly 100% plasmid relaxation and ϳ50% plasmid linearization at concentrations of 1 M each, whereas decarbamoyl-BLM exhibited ϳ10-fold reduced DNA cleavage activity, resulting in a mixture of supercoiled, relaxed, and linearized plasmid DNA at a concentration of 5 M (Fig. 5). The observed DNA cleavage activity was concentration-dependent for all of the compounds tested.

DISCUSSION
More than four decades after the discovery of the clinically important anticancer compound BLM (1), its native producer S. verticillus unambiguously proved genetically accessible for the first time. Although some improvement of protoplast formation and regeneration was achieved during these studies, DNA transfer by protoplast-mediated transformation failed to yield any true transformants. Intergeneric E. coli-Streptomyces conjugation under optimized conditions, however, facilitated the development of a genetic system for this pharmaceutically important microorganism.
From an entire set of six different replicative and integrative vectors employed as test plasmids, two of three integrative vectors, pSET152 and pRT801, yielded true exconjugants at frequencies of 10 Ϫ7 -10 Ϫ8 exconjugants/cfu, which is 10 5 -10 6 -fold lower than reported for Streptomyces lividans (ϳ10 Ϫ2 exconjugants/cfu) under similar conditions (31). Failure to introduce the self-replicating vectors originating from three different Streptomyces replicons indicated that this host strain is unable to propagate any of the well known E. coli-Streptomyces shuttle vectors, thereby significantly complicating expression and complementation experiments in this strain.
Typically, the frequency of site-specific integration into the host chromosome is much higher than that of homologous recombination for a specific Streptomyces strain. Based on the observation of site-specific integration frequencies as low as 10 Ϫ7 -10 Ϫ8 exconjugants/cfu for S. verticillus, failure to obtain gene inactivation mutants by homologous recombination was not surprising. Because the frequency for homologous recombination is also proportional to the size of the DNA regions available for the recombination event (20), a recently developed technology ( RED-mediated PCR targeting mutagenesis) (23) provided a promising new strategy to genetically modify this strain allowing the facile increase of inactivation constructs. As opposed to 2.3-and 4.1-kb homologous arms and 9.8-and 7.8-kb homologous arms in the earlier constructs pBS41 and pBS42, respectively, flanking regions of Ͼ12-kb on at least one side of the targeted gene in constructs pBS43, pBS44, pBS45, pBS46, and pBS49 enabled the generation of the desired mutant strains at surprisingly high frequencies, very comparable with site-specific integration of pSET152 and pRT801. From these data we conclude that for S. verticillus, the minimal size range of DNA fragments allowing for homologous recombination events to occur lies between 10 and 12 kb, which is uncommonly large compared with other Streptomyces strains (20).
To date, many other microbial producers of important natural products have been reported to be genetically inaccessible or extremely hard to manipulate (32)(33)(34). However, the availability of an efficient genetic system is one of the key prerequisites for successful engineering of natural product biosynthetic pathways to generate new biologically active derivatives using combinatorial biosynthesis. The results from our studies indicate a potential solution for this widely experienced obstacle in microbiology, especially in cases where the alternative approach of heterologously expressing the desired biosynthetic pathway has failed or is not feasible.
Multiple gene knock-out experiments by gene replacement and in-frame deletion confirmed the identity of the BLM biosynthetic locus in vivo, experimentally defined the BLM cluster boundaries, and revealed new insights into the timing of BLMcarbamoylation by BlmD. Five gene inactivation experiments proved the involvement of blmIV, blmVIII, and blmD in BLM biosynthesis and confirmed the predicted cluster boundaries to be located downstream of blm-orf31/32 and upstream of blmorf(Ϫ1). All gene replacement mutants showed either the expected BLM producing phenotype for targeted genes located

No.
␦ outside the cluster boundaries (blm-orf31/32 and blm-orf(Ϫ1)) or complete abolishment of BLM production for genes proposed to be involved in BLM biosynthesis (blmIV and blmVIII).
Only the in-frame deletion of blmD, however, yielded the anticipated pathway intermediate, decarbamoyl-BLM. It may be speculated that early pathway intermediates stay tethered to their respective NRPS or PKS enzymes and therefore could not be isolated from the respective mutant fermentations. Moreover, intermediates lacking the sugar moiety may not be recognized by the respective transport proteins for their export out of the cells and hence escape detection in the culture broth. Isolation of 3-MOP-decarbamoyl-BLM from the cell free supernatant of the respective fermentations, in contrast, indicates efficient transport of late biosynthetic intermediates into the production medium. A one-step purification procedure using Amberlite IRC-50 resin (H ϩ type) allowed us to estimate the titer of 3-MOP-decarbamoyl BLM to be 6 -8 mg/liter representing ϳ70% of the 3-MOP-BLM production observed in wild-type S. verticillus (8 -10 mg/liter) under equivalent fermentation conditions. The 3-MOP-decarbamoyl BLM was then purified from preparative scale fermentation. Extensive analysis of MS, 1 H, 13 C, 1 H-1 H COSY, TOCSY, HSQC, and HMBC data for this compound resulted in the full 1 H and 13 C NMR spectroscopic assignments ( Fig. 4 and Table  1), and their comparison with the BLM A2 spectroscopic data reported in the literature (30) confirmed that the isolated intermediate indeed represents 3-MOP-decarbamoyl-BLM. Decarbamoyl-BLM has previously been reported as a partial hydrol-  OCTOBER 17, 2008 • VOLUME 283 • NUMBER 42 JOURNAL OF BIOLOGICAL CHEMISTRY 28243 ysis product of BLM under basic aqueous conditions after prolonged storage but has only been accessible in very small quantities (35,36). Additionally, decarbamoyl-BLM demethyl-A2, lacking one methyl group in the A2 terminal amine, was isolated as a byproduct during the tedious multi-step total chemical synthesis of BLM demethyl-A2 (37). From a structural point of view, the carbamoyl group has been speculated to represent one of the ligands involved in formation of the active metal complexed BLM. Detailed mechanistic investigations, however, have so far been hampered by the limited availability of decarbamoyl-BLM. The accessibility of this particularly important compound in large quantities by simple fermentation and compound isolation procedures from the mutant of a highly appreciated, pharmaceutically relevant Streptomyces strain now facilitates such studies.

Manipulation of the Bleomycin Biosynthetic Gene Cluster
Our data presented here support the hypothesis that carbamoylation is the final step of BLM biosynthesis, revising the previous assumptions for disaccharide formation and attachment ( Fig. 6) (10). This is in agreement with the novobiocin biosynthetic pathway for which carbamoylation of the noviose sugar was also reported to be the final catalytic step (38). The previously discussed alternative of the mannose moiety first being carbamoylated by BlmD and then transferred onto the BLM-monosaccharide by the respective glycosyltransferase (BlmE or BlmF) seems less likely, because this pathway would require broad substrate flexibility of the glycosyltransferase to catalyze decarbamoyl-mannose transfer to form 3-MOP-decarbamoyl-BLM in the blmD defective mutant strain.
Investigation of 3-MOP-decarbamoyl-BLM for DNA relaxation and linearization in comparison with BLM A2 and B2 indicates an important role of the carbamoyl group for efficient cleavage activity. Although several in vitro studies have shown that deglyco-BLMs exhibit reduced DNA cleavage activities compared with BLM A2 or B2 (36,39,40), the role of the carbamoyl group remained controversial. Decarbamoyl-BLM was reported to show significantly reduced malondialdehyde forming activity, but similar oligonucleotide cleavage activity compared with BLM A2 (36), and similar DNA cleavage activity as BLM in the presence of Fe 2ϩ , but reduced activity in the presence of Cu 2ϩ and dithiothreitol (39), and oligonucleotide cleavage activities much closer to deglyco-BLM A2 rather than BLM A2 (41). Our results obtained with 3-MOP-decarbamoyl-BLM, however, indicate a significant impact of the carbamoyl group on DNA cleavage activity resulting in ϳ10-fold reduced plasmid relaxation efficiency compared with BLM A2 and B2. This observation may originate from the carbamoyl group directly contributing to the coordination of the metal ion for DNA cleavage or from being involved in DNA binding. Although there has been speculation about the former hypothesis (41), the latter is supported by the recently published crystal structure of DNA-bound Co(III)-BLM B2; the carbamoyl NH 2 forms a hydrogen bond with the minor grove, whereas the function of the disaccharide seems to be to correctly position and stabilize the entire complex (42). In addition to the likely reduced DNA affinity of 3-MOP-decarbamoyl-BLM caused by the lack of the carbamoyl group, the different terminal amines of 3-MOP-decarbamoyl-BLM, BLM A2, and B2 may also contribute to the variation in DNA cleavage activity. In depth spectroscopic and mechanistic studies of 3-MOP-decarbamoyl-BLM in comparison with BLM will now allow elucidation of the exact contribution of the carbamoyl group to DNA binding, cleavage activity, cytotoxicity, and bioavailability of the BLM family of compounds.
In summary, since the identification and sequence analysis of the BLM biosynthetic gene cluster several years ago (10,12,14), the inaccessibility of S. verticillus to genetic manipulation has prevented us from in depth in vivo investigations of its biosynthetic pathway. Moreover, all attempts to express the entire ϳ65-kb BLM biosynthetic gene cluster in a heterologous host and to engineer the respective DNA sequence therein have been unsuccessful as well. Although progress in genetic manipulation of the related tallysomycin and zorbamycin producers has been made recently (17,18), this study represents the first report of a new BLM derivative generated in vivo by engineering of the respective producer strain and demonstrates the feasibility of BLM analog production in suitable mutants on a large scale in the future. Moreover, the availability of an S. verticillus mutant strain producing 3-MOP-decarbamoyl-BLM in respectable yields will now greatly facilitate spectroscopic and mechanistic studies, providing insight into the exact mode of action of the BLM family of antibiotics.