Correspondence: Bin Hong, Key Laboratory of Biotechnology of Antibiotics of Ministry of Health, Institute of Medicinal Biotechnology, Chinese Academy of Medical Sciences and Peking Union Medical College, No. 1 Tiantan Xili, Beijing 100050, China. Tel.: +86 10 6302 8003; fax: +86 10 6301 7302; e-mail: email@example.com
C-1027 is a chromoprotein of the nine-membered enediyne antitumour antibiotic family, comprising apoprotein to stabilize and transport the enediyne chromophore. The disruption of apoprotein gene cagA within the C-1027 biosynthetic gene cluster abolished C-1027 holo-antibiotic production detected by an antibacterial assay, as well as the expression of the apoprotein and C-1027 chromophore extracted following protein precipitation of the culture supernatant. Complementation of the cagA-disrupted mutant AKO with the intact cagA gene restored C-1027 production, suggesting that cagA is indispensable for holo-antibiotic production. Overexpression of cagA in the wild-type strain resulted in a significant increase in C-1027 production as expected. Surprisingly, electrospray ionization (ESI)-MS and ESI-MS/MS analyses suggested that the AKO mutant still produced the C-1027 enediyne chromophore [m/z=844 (M+H)+] and its aromatized product [m/z=846 (M+H)+]. Consistent with this, the results from gene expression analysis using real-time reverse transcriptase-PCR showed that transcripts of the positive regulator sgcR3 and the structural genes sgcA1, sgcC4, sgcD6 and sgcE were readily detected in the AKO mutant as well as in the wild-type and the complementation strain. These results provided, for the first time, evidence suggesting that the apoprotein of C-1027 is not essential in the self-resistance mechanism for the enediyne chromophore.
The enediyne antibiotics are a class of natural products that fall into two categories: nine-membered ring enediynes and 10-membered ring enediynes (Thorson et al., 1999). Most of the former comprise a 1 : 1 complex of an apoprotein and a noncovalently bound chromophore, while the latter lack any associated apoproteins. With the exception of maduropeptin, the enediyne apoproteins such as CagA for C-1027, NcsA for neocarzinostatin, Ked for kedarcidin, AxnA for actinoxanthin and McmA for macromomycin (also known as auromomycin) are highly homologous acidic proteins of 108–114 amino acid residues (Thorson et al., 1999). The structures of all five enediyne apoproteins have been elucidated (PDB accession nos 1J48, 1NOA, 1AKP, 1ACX and 2MCM), and the structures of the C-1027 and neocarzinostatin holo-antibiotic are also known (PDB accession nos 1HZL and 1NCO). The chromophores of these homologous chromoproteins are highly labile and unstable to nucleophiles, heat, UV light and extremes of pH. The chromophores of macromomycin and actinoxanthin are too labile to yet allow structure determinations (Baker et al., 2007). The apoproteins have a highly conserved tertiary structure to bind, protect and transport the highly reactive enediyne chromophores, which are potent cytotoxic compounds exerting their effect by abstracting hydrogen atoms from DNA to cause strand cleavage, resulting in the formation of potent highly effective antitumour agents (Baker et al., 2007). The best-studied member of this class, neocarzinostatin conjugated with poly(styrene-co-maleic acid/anhydride), has been approved (Maeda et al., 1997) and is currently in use as a chemotherapeutic in Japan.
C-1027 (also called lidamycin) is a potent enediyne antitumour antibiotic, which has entered a phase II clinical trial in China (Shao & Zhen, 2008). In the family of chromoproteins, C-1027 is the most potent, highly active antitumour antibiotic. C-1027 was isolated from the fermentation broth of Streptomyces globisporus C-1027 (Hu et al., 1988) and was found to exhibit antimicrobial activity against Gram-positive bacteria and to be highly effective against various human cancer cells in vitro and in vivo (Zhen et al., 1989). The enediyne chromophore of C-1027 consists of four distinct moieties: an enediyne core, a deoxy aminosugar, a benzoxazolinate and a β-tyrosine moiety (Fig. 1) (Minami et al., 1993; Yoshida et al., 1993). The enediyne core, characterized by two acetylenic groups conjugated to a double bond within a nine-membered ring, undergoes a Bergman cyclization to generate a highly reactive cycloaromatized diradical that is capable of abstracting hydrogen atoms and then converts to its aromatized product (Fig. 1) (Inoue et al., 2006). The apoprotein of C-1027 (CagA, previously called C-1027-AG) was purified and its primary structure was determined (Otani et al., 1991) shortly after the discovery of C-1027, and then, the cagA gene and its flanking regions were cloned and sequenced (Sakata et al., 1992). CagA is a 110-amino acid acidic protein with two internal disulphide bridges, derived from the preapoprotein with a 33-amino acid signal peptide. The three-dimensional structure of the apoprotein and the holocomplex of C-1027 was determined by both crystallographic (Chen et al., 2002) and nuclear magnetic resonance spectroscopic techniques (Tanaka et al., 2001). The overall structure of CagA, similar to those of other chromoprotein apoproteins such as neocarzinostatin and kedarcidin, has a β-sheet-rich fold that provides a hydrophobic pocket for chromophore binding (Fig. 1).
The cagA gene was determined to be clustered within the biosynthetic gene cluster for C-1027 (GenBank accession no. AY048670), which was the first cloned enediyne gene cluster (Liu & Shen, 2000). The biosynthetic gene cluster for C-1027 contains a total of 56 ORFs in a region of 75 kb, providing a convergent biosynthetic strategy for C-1027 from four biosynthetic building blocks (Liu et al., 2002). Further cloning and characterization of biosynthetic gene clusters for other enediynes [calicheamicin (Ahlert et al., 2002), neocarzinostatin (Liu et al., 2005), maduropeptin (Van Lanen et al., 2007) and dynemicin (Gao & Thorson, 2008)] confirmed the unifying paradigm for both nine- and 10-membered enediyne chromophore biogenesis. Early cloning studies showed that cagA and mcmA were expressed early, preceding that of chromophore production (Sakata et al., 1989, 1992), and cagA was expressed even in the chromophore nonproductive medium. Therefore, it was proposed that the chromophore produced was first sequestered by binding to the preapoprotein to form a complex before being transported out of the cell to confer self-resistance to the producing organism (Thorson et al., 1999). However, no further direct data have been reported to support this hypothesis.
Here, we investigated the role of the cagA gene in C-1027 holo-antibiotic biosynthesis in S. globisporus C-1027 by analysing the chromoprotein and chromophore production in the constructed cagA disruption, complementation and overexpression mutants. Intriguingly, it was shown that the cagA disruption mutant still generates chromophore, which suggested that apoprotein is not essential in the self-resistance mechanism for the enediyne antibiotics.
Materials and methods
Bacterial strains and culture conditions
Streptomyces globisporus C-1027 was used as the wild-type strain to generate different derivatives. Streptomyces globisporus strains were grown at 28 °C on S5 agar (Wang et al., 2009) for sporulation production, on mannitol soya flour (MS) agar (Kieser et al., 2000) for conjugation, in trypticase soy broth (BD, Sparks) for isolation of total DNA or plasmid DNA and in fermentation medium FMC-1027-1 (Wang et al., 2009) for antibiotic production. Bacillus subtilis CMCC(B) 63501 was used as the test organism for the antibacterial activity assay of C-1027 (Zhao et al., 2005), grown on solid F403 agar (Wang et al., 2009) at 37 °C. Escherichia coli DH5α (Sambrook & Russell, 2001) was used for general gene manipulation. The methylation-deficient E. coli ET12567 carrying a driving vector pUZ8002 was used as a donor for conjugal transfer of genes into Streptomyces (Kieser et al., 2000). They were grown either on solid or in liquid Luria–Bertani medium at 37 °C (Sambrook & Russell, 2001). When antibiotic selection of bacteria was needed, strains were incubated with apramycin (50 μg mL−1), ampicillin (100 μg mL−1), thiostrepton (30 μg mL−1), kanamycin (50 μg mL−1) and chloramphenicol (30 μg mL−1). The bacterial strains used and constructed in this study are listed in Supporting Information, Table S1.
Routine DNA manipulations with E. coli were carried out as described previously (Sambrook & Russell, 2001). The plasmids used and constructed in this study are listed in Table S1. Recombinant DNA techniques in Streptomyces species were performed as described previously (Kieser et al., 2000). The primers for PCR amplification are listed in Table 1 and oligonucleotides were synthesized by Sangon (Shanghai, China). Southern blot analysis was performed on a Hybond-N+ nylon membrane (Amersham Biosciences, Buckinghamshire, UK) with a fluorescein-labelled probe using the Gene Images Random Prime Labelling Module and CDP-Star Detection Kit (Amersham Biosciences).
Table 1. Primers used during this study for gene cloning and RT-PCR analysis
Restriction endonuclease recognition sequences introduced by these oligonucleotides are underlined.
Amplifying 1.1-kb upstream arm used in knockout of cagA
Amplifying 1.2-kb downstream arm used in knockout of cagA
Amplifying 0.4-kb cagA gene coding region
Amplifying 0.9-kb cagA gene and its flanking regions
Amplifying 1.0-kb thiostrepton resistance gene tsr
Confirming the integration of pSET152 and its derivatives
Detection of hrdB transcripts
Detection of sgcA1 transcripts
Detection of sgcC4 transcripts
Detection of sgcD6 transcripts
Detection of sgcE transcripts
Detection of sgcR3 transcripts
Construction of cagA-disrupted mutant
To inactivate the cagA gene, two pairs of primers (AupF/R and AdownF/R) were used to amplify the upstream fragment (1.1 kb) and the downstream fragment (1.2 kb) of the cagA gene, respectively. The PCR products obtained were cloned in pGEM-T vector (Promega, WI) and verified by DNA sequencing. The EcoRI–BglII-digested upstream fragment and the NdeI–HindIII-digested downstream fragment, together with the 1.5-kb BglII–NdeI apramycin resistance gene [aac(3)IV] cassette from plasmid pUO9090 (Smith et al., 2000), were then cloned into EcoRI–HindIII sites of pBS03 (a suicide vector for Streptomyces containing RK2 oriT and the thiostrepton resistance gene tsr) (Li et al., 2007). The resultant plasmid pBSA bearing the disrupted cagA gene was first transformed into E. coli ET12567/pUZ8002 and then transferred into S. globisporus C-1027 by intergeneric conjugation, to replace the original cagA gene via double crossover. Apramycin-resistant, thiostrepton-sensitive exconjugants were selected on MS agar and deletions within cagA were confirmed by PCR and Southern blot analysis.
Construction of cagA expression plasmids
A 0.4-kb fragment of the cagA ORF and a 0.9-kb fragment of the cagA gene with its 5′- and 3′-flanking regions were amplified by PCR as an NdeI–BamHI and an EcoRI–BamHI fragment, respectively, using total DNA of S. globisporus C-1027 as a template (primers A400F/R and A900F/R) and then cloned into pGEM-T vector. After verification by DNA sequencing, EcoRI–BamHI digested 0.4- and 0.9-kb fragments were placed under the control of the strong constitutive promoter ermE*p by cloning into the corresponding sites of pL646 (a derivative of pSET152 containing ermE*p) (Hong et al., 2007) to yield pLA400 and pLA900. Similarly, the 0.9-kb fragment was also ligated into the EcoRI–BamHI sites of integrative vector pSET152 and multicopy plasmid pKC1139 (Bierman et al., 1992) to yield pSETA900 and pKCA900, respectively. For cagA overexpression, these plasmids were transferred into S. globisporus C-1027, and apramycin-resistant exconjugants were selected and confirmed by restriction enzyme digestion or PCR analysis. As the cagA-disrupted mutant is apramycin resistant, a thiostrepton resistance gene (tsr) cassette amplified from pIJ680 (Kieser et al., 2000) using primers tsrF/R was inserted into these expression plasmids to yield the corresponding complementary plasmids pLTA400, pLTA900, pSETTA900 and pKCTA900. The plasmids obtained were then introduced into the cagA-disrupted mutant, and the thiostrepton-resistant exconjugants were selected and confirmed. The homologous empty vectors were used as negative controls.
Analysis of C-1027 production
For the production of secondary metabolites, S. globisporus strains were grown in liquid medium FMC-1027-1 as described previously (Wang et al., 2009). The dry weight of mycelia was measured in cultures taken at different time points during the course of fermentation and the patterns of growth curves were monitored consistently among the relevant strains. To obtain statistically significant results, at least two independent exconjugants of each recombinant strain were selected and the fermentations were repeated at least three times. To determine the antibacterial activity of the strains studied, the fermentation supernatant (270 μL) was added to stainless-steel cylinders, which were placed on F403 agar plates containing B. subtilis spores (0.4% v/v). For the bioassay of C-1027 produced on S5 agar, 5 μL spores was spotted onto the S5 agar plates and incubated at 28 °C for the indicated time, and then the plates were overlaid with F403 soft agar containing B. subtilis spores (0.4% v/v). Antibacterial activity was observed based on the inhibition zone diameter measured after incubation for 18 h at 37 °C.
For the detection of C-1027 chromophore complexed with the chromoprotein, isolation and HPLC were carried out as described previously (Wang et al., 2009). In this procedure, the C-1027 chromoprotein complex was precipitated by (NH4)2SO4 from the fermentation supernatant, followed by extraction with ethyl acetate, and is therefore described here as a protein precipitation and extraction method. Reverse-phased HPLC analysis was performed using a Phenomenex C18 column (250 × 4.6 mm, 5 μm, CA) maintained at 30 °C with 50% acetonitrile/50% 20 mM potassium phosphate (pH 6.86) as the mobile phase at a flow rate of 1 mL min−1. Absorbance was monitored at 350 nm. Peaks consistent with the C-1027 enediyne chromphore standard, which was confirmed by electrospray ionization-MS (ESI-MS), were determined to have a UV/Vis spectrum with a definite absorption shoulder between 340 and 360 nm as reported previously (Otani et al., 1988).
For investigation of C-1027 chromophore production in the cagA-disrupted mutant, a direct extraction method was used. The fermentation supernatant was collected and extracted with an equal volume of ethyl acetate three times and concentrated in vacuum. The crude extracts were then dissolved in methanol and subjected to ESI-MS analysis using a Q-Trap LC/MS/MS System with the TurboIonSpray ion source (Applied Biosystems/MDS Sciex). The presence of enediyne [(M+H)+ ion at m/z=844] and aromatized [(M+H)+ ion at m/z=846] C-1027 chromophore was further validated with ESI-MS/MS captured at a collision energy setting of 35–50 V.
Western blot analysis
To assess the CagA production of various S. globisporus strains, a Western blot analysis was performed as described previously (Hong et al., 2005). Briefly, 1 mL of ice-chilled culture supernatants was precipitated by 0.5 mL ice-cold 60% trichloroacetic acid. The pellets were washed with 95% ethanol, dried in vacuum and redissolved in 2 × loading buffer. Proteins were separated by 15% sodium dodecyl sulphate-polyacrylamide gel electrophoresis and then blotted onto polyvinylidene fluoride membranes using a semi-dry electroblotter (Bio-Rad). The CagA protein was detected using a mouse anti-CagA monoclonal antibody (Zhou et al., 1997), followed by horseradish peroxidase-conjugated goat anti-mouse IgG antibody (Sigma). An enhanced chemiluminescence Western blotting detection system (Amersham Biosciences) was used to visualize the protein.
Gene expression analysis by reverse transcriptase-PCR (RT-PCR)
Fresh spores of S. globisporus from the S5 agar plates were harvested and then spread on cellophane membranes placed on S5 agar plates and incubated at 28 °C. Mycelia were collected at 48 h, and RNA was isolated as described previously (Uguru et al., 2005). Contaminated DNA was removed by digestion with DNAse I (Promega). The quantity and integrity of the RNA preparations were determined by spectrophotometry and agarose gel electrophoresis. The first-strand synthesis of cDNA was performed using a SuperScript III First-strand Synthesis System (Invitrogen, CA) using random hexamers as primers following the manufacturer's instructions. Oligonucleotides were designed to amplify fragments of about 100–150 bp from the target genes (Table 1). The relative levels of transcripts in samples from mycelia patches were analysed by semi-quantitative end-point PCR and the final products were subjected to agarose gel electrophoresis. Quantitative real-time PCR of selected genes were performed as described previously (Wang et al., 2009). PCR was also performed without reverse transcriptase to confirm that the RNA preparations were not contaminated with genomic DNA. The relative mRNA level of target genes was normalized to the level of hrdB according to Pfaffl's method (Pfaffl, 2001).
Data analysis and statistical analysis
Quantitative data are expressed as mean±SE. The statistical significance of the data was evaluated using Student's t-test. Differences were considered significant at P<0.05.
Inactivation of cagA
In order to determine the significance of cagA for the production of C-1027, a 0.3-kb fragment within the cagA gene-coding region was eliminated from the C-1027 biosynthetic cluster of S. globisporus C-1027 (deletion from 14 281 to 14 589 nt of AY048670) by replacing it with an apramycin resistance gene [aac(3)IV] cassette (Fig. 2a). The correct cagA-disrupted mutant (two independent exconjugants), designated as S. globisporus AKO, was confirmed by PCR using primers AupF and AdownR (Table 1, Fig. 2b). Southern blot hybridization further authenticated the site-specific disruption of cagA using the right and the left arm for crossover as probes (Fig. 2c and d). Compared with the wild-type strain, no apparent difference in the growth characteristics and morphologies was observed in the AKO mutant both on solid medium and in liquid medium (data not shown). The antibacterial bioassay against B. subtilis showed that disruption of cagA completely abolished C-1027 production in the AKO mutant, while C-1027 production could be detected from days 3 to 5 in the wild-type strain (Fig. 3a). Western blot analysis of the CagA protein showed that the AKO mutant no longer produced apoprotein CagA (Fig. 3d). HPLC analysis of the C-1027 chromophore isolated by the protein precipitation and extraction method revealed that no C-1027 chromophore could be detected in the AKO mutant culture supernatant (Fig. 3e).
Complementation of the S. globisporus AKO
Complementation experiments were performed in order to test whether the loss of C-1027 production in S. globisporus AKO was caused solely by the cagA gene disruption rather than any polar effect or an unanticipated mutation in another locus. A multicopy plasmid pKC1139 derivative pKCTA900, containing a 0.9-kb fragment of cagA ORF with its 5′-flanking region including the native promoter and the 3′-flanking region including a transcriptional terminator, was introduced into the AKO mutant by conjugation. The resultant strain with pKCTA900 restored the C-1027 production to the wild-type level throughout the cultivation, and even to a slightly higher level than the wild-type strain on day 5 (Fig. 3a). Western blot assay of the CagA protein (Fig. 3d) and HPLC analysis of the C-1027 chromophore (Fig. 3e) also showed the restored production. These results strongly confirmed that the loss of C-1027 production in the AKO mutant was due to the lack of the cagA gene.
In order to assess the effect of different cagA gene constructs on the complemented C-1027 production, three pSET152-based cagA expression plasmids (pSETTA900, pLTA900 and pLTA400) were introduced into the AKO mutant separately, integrating into the ΦC31 attB site on the chromosome as a single copy. pSETTA900 and pLTA900 contained the same cagA gene fragment as in pKCTA900, while pLTA900 contained the strong constitutive promoter ermE*p in addition to the cagA native promoter. pLTA400 contained the coding region of the cagA gene directly downstream of the strong constitutive promoter ermE*p. As shown in Fig. 3b, pSETTA900 appears to restore C-1027 production, but with a reduced production level compared with pKCTA900. The AKO mutant containing pLTA900 produced slightly more C-1027 than the mutant containing pSETTA900, as expected. Surprisingly, the expression of cagA ORF directly under the control of ermE*p (pLTA400) could not restore C-1027 production in the AKO mutant (Fig. 3b). These results showed that the production of C-1027 in complementation strains was related to the cagA gene copy number and the promoter used.
Overexpression of the cagA gene increased the production of C-1027
To further investigate whether the increased expression of cagA might enhance C-1027 production, the expression plasmids including pLA900 and pKCA900 were introduced into the wild-type strain. As shown in Fig. 3c, the wild-type strain with pLA900 appears to increase the production of C-1027 as determined based on antibacterial activity. Consistent with this, cagA expression (Fig. 3d) and chromophore production (Fig. 3e) were also increased compared with those of the wild-type strain containing pSET152. The wild-type strain with pKCA900 showed a similar increased level of C-1027 production (data not shown). These results suggested that overexpression of the cagA gene in the wild-type strain was effective in enhancing C-1027 chromoprotein production, whether by an extra copy under the strong constitutive promoter ermE*p or by a multicopy of the cagA gene with its native promoter.
Streptomyces globisporus AKO still produces C-1027 chromophore
The free C-1027 chromophore was reported to be extremely labile without the protection of apoprotein, i.e. the half-life of free C-1027 chromophore (t1/2=0.8 h in ethanol) was significantly shorter than with the apoprotein (t1/2=37 h in neutral buffer) (Inoue et al., 2006). Thus, the first step of the conventional C-1027 isolation method is the precipitation of proteins by (NH4)2SO4. However, this method only allows the isolation of the C-1027 chromophore that forms the chromoprotein complex with the apoprotein CagA, and thus it is expected that the C-1027 chromophore would not be isolated from the AKO mutant using the standard procedure (Fig. 3e). Therefore, a different isolation method was used that is independent of apoprotein production. As the production level of C-1027 was the highest on day 5 (as determined by antibacterial activity; Fig. 3a), the culture supernatants of the AKO mutant on day 5 were directly extracted by ethyl acetate and subjected to ESI-MS analysis. Intriguingly, the molecular weights corresponding exactly to that of the authentic C-1027 standard [i.e. C-1027 chromophore with m/z=844 (M+H)+ and its aromatized product with m/z=846 (M+H)+] were detected from the extracts of the AKO mutant (Fig. 4). Furthermore, ESI-MS/MS spectra of these two peaks were taken for validation of the chromophore. The ‘fingerprints’ of the two peaks matched essentially those of the C-1027 standard, with a major peak of 188 Da corresponding to the deoxy aminosugar part of the C-1027 chromophore (Fig. 4). These results provided conclusive confirmation of the C-1027 chromophore in the extracts of the AKO mutant culture supernatant.
Influence of cagA on the expression of C-1027 biosynthetic genes
To investigate further the production of the C-1027 chromophore in the AKO mutant, C-1027 biosynthetic gene expression analysis was conducted by quantitative real-time RT-PCR. Total RNA from the wild-type strain, the AKO mutant, the AKO mutant with pKCTA900 and the wild-type strain with pLA900 was extracted under conditions at which the wild-type strain commenced C-1027 production at about 48-h growth on S5 agar (Fig. 5a). The cDNA was synthesized and then used as a template in semi-quantitative and quantitative PCR. Figure 5b shows the end-point RT-PCR results for cagA and hrdB in the wild-type strain, the AKO mutant and the AKO mutant with pKCTA900. As expected, the cagA transcripts were undetectable in the AKO mutant, but were present at a similar level in the wild-type strain and the complementation strain. The relative level of the transcripts of the positive regulatory gene sgcR3 (Wang et al., 2009) and four structural genes (sgcA1, sgcC4, sgcD6 and sgcE) were then analysed together with cagA. SgcA1 and SgcC4 were reported to catalyse the first step in the biosynthesis of the deoxy aminosugar and the β-amino acid moiety of C-1027, respectively (Christenson et al., 2003; Murrell et al., 2004). sgcD6 is putatively involved in the biosynthesis of the benzoxazolinate moiety based on bioinformatic analysis (Liu et al., 2002). sgcE encodes the enediyne polyketide synthase responsible for enediyne core biosynthesis (Liu et al., 2002). Consistent with the presence of chromophore in the AKO mutant culture supernatant, the positive regulatory gene sgcR3 and four structural genes described above were readily detected in the AKO mutant at nearly the same level as the wild-type strain, the complementation strain and the overexpression strain (Fig. 5c).
The enediyne antibiotics represent an unusual example of anticancer agents with unique molecular scaffolds and remarkable biological activity for DNA damaging, offering a distinct opportunity to study their biosynthetic mechanisms and modes of self-resistance. Notable developments have been achieved to decipher the genetic and biochemical basis for the biosynthesis of enediyne chromophores (Christianson et al., 2007; Van Lanen et al., 2008; Lin et al., 2009) since the biosynthetic gene cluster for C-1027 was cloned and characterized about 10 years ago. The biosynthesis of both nine- and 10-membered ring enediynes was found to be catalysed by a common distinctive type I enediyne polyketide synthase (sgcE for C-1027 and calE8 for calicheamicin) (Liu et al., 2002) and convergently assembled from variant building blocks (Ahlert et al., 2002; Liu et al., 2002; Van Lanen et al., 2007). In this study, we focused on the role of apoprotein gene cagA in C-1027 holo-antibiotic and chromophore production. Residing between sgcA1 and sgcA4, which are involved in the biosynthesis of deoxy aminosugar (Liu et al., 2002), cagA was reported to be transcribed as a monocistronic mRNA under its native promoter (Sakata et al., 1992). The cagA disruption mutant was constructed by double crossover, and its ability to produce C-1027 was completely eliminated as detected by an antibacterial assay (Fig. 3a). The apoprotein (Fig. 3d) and its associated chromophore (Fig. 3e) were also abolished. Together with the results obtained in the complementation experiments, it was demonstrated that the expression of cagA is indispensable for the production of the holo-antibiotic (Fig. 3). The results of complementation of the AKO mutant with different strategies (Fig. 3b) showed the existence of a gene-dosage effect. Furthermore, the presence of at least one copy of the authentic native promoter of cagA seemed to be required for its intricate regulation for C-1027 production.
As the enediynes are among the most potent natural products, there is great interest in understanding how the antibiotic-producing organisms escape their lethal effects. An unprecedented mode of self-resistance to calicheamicin, a 10-membered enediyne without apoprotein, was reported to present a self-sacrificing paradigm in which detoxification of enediyne is accomplished at the expense of the specific proteolysis of CalC through Gly113 backbone hydrogen abstraction (Biggins et al., 2003). However, a search using blastp (protein–protein blast) at NCBI yielded no homologues of CalC in other (including nine- and 10-membered) enediyne biosynthetic gene clusters (data not shown), suggesting that it is a specific mechanism used by calicheamicin. For the nine-membered enediyne chromoproteins, the apoproteins serving to tightly bind and stabilize the enediyne have been intuitively proposed to play a role in self-resistance to the producing organism by sequestering the synthesized chromophore (Thorson et al., 1999). A recent report seems to be in accordance with this hypothesis that the C-1027 chromophore mediated the decomposition of apoprotein by Gly96 backbone hydrogen abstraction (Inoue et al., 2006). However, the results presented here suggested that the apoprotein might not be indispensable for the self-resistance of C-1027 chromophore. The production of the C-1027 chromophore was determined (Fig. 4b) in the cagA disruption mutant AKO, which no longer expressed apoprotein (Fig. 3d). Although without the protection of apoprotein only a tiny amount of the chromophore could be detected in the extract of the broth of the mutant strain, the presence of a similar amount of chromophore in the broth of the wild-type strain (similar signal-to-noise ratio of the ESI-MS spectrum, data not shown) suggested a similar production level in the two strains. Further analyses by quantitative real-time RT-PCR showed that the positive regulatory gene sgcR3 and several structural genes (sgcA1, sgcC4, sgcD6 and sgcE) for chromophore biosynthesis were expressed to nearly the same level in the AKO mutant as in the wild-type strain (Fig. 5c), which also implies that the chromophore was produced normally in the AKO mutant. Kudo et al. (1982) reported the production of free neocarzinostatin chromophore by the producing strain Streptomyces carzinostaticus var. F-41 grown in a semi-synthetic medium containing MgSO4 in spite of the limited production of apoprotein. As the neocarzinostatin holo-antibiotic is a 1 : 1 complex of chromophore and apoprotein, their finding might offer an example that the nine-membered enediyne chromophore was produced independent of apoprotein under some natural conditions. Overexpression of SgcB, a transmembrane efflux protein of C-1027, was demonstrated to increase C-1027 production significantly in the early stage of fermentation (Liu & Shen, 2000). Its homologous genes also exist in the neocarzinostatin and maduropeptin biosynthetic gene cluster. Thus, an ingenious, tightly sequestered enediyne biosynthesis exportation system might exist to protect the organism that produces the nine-membered enediyne chromophore regardless of the production of the respective apoprotein. As there are still some uncharacterized ORFs in the C-1027, neocarzinostatin and maduropeptin biosynthetic gene clusters, the presence of a self-sacrifice protein functioning like CalC cannot be excluded at present.
In view of their unique structure, mode of action and potent cytotoxicity, enediyne chromoproteins have generated intense interest for their economic production at an industrial scale. Our results (Fig. 3c) suggested that overexpression of their corresponding apoprotein gene might provide an additional and rational choice to construct overproducing strains of enediynes and their analogues. On the other hand, the natural enediynes have seen limited uses as clinical drugs mainly because of their short half-life and severe toxicity. Protein engineering studies on the apoproteins might aim at addressing these limitations (Baker et al., 2007), as the apoproteins are utilized as a packing carrier protein to stabilize and deliver the biologically active chromophores. For example, a kinetically stabilized supra C-1027 (fourfold more stable than C-1027) was rationally designed to have deuterium (D) instead of protium (H) at the α-hydrogen position of Gly96 by expression of the d-Gly apoprotein in E. coli (Usuki et al., 2004). The cagA disruption mutant complemented with the pKC1139-based expression plasmid reported here may provide a useful and convenient system for apoprotein engineering studies, where the recombinant holo-antibiotic may be biosynthesized, exported and evaluated directly. This may avoid the complicated procedure of extraction of unstable enediyne chromophore and then its incorporation into recombinant apoprotein (Usuki et al., 2004), and thus facilitate future efforts of apoprotein engineering for the search for novel drugs as potential anticancer agents.
We gratefully acknowledge Dr K. McDowall for providing plasmid pL646 and Prof. Rongguang Shao for providing mouse anti-CagA monoclonal antibodies. We also thank Prof. Lianfang Jin and Prof. Shuyi Si for technical assistance with HPLC analysis of C-1027. This work was funded by grants from the Key New Drug Creation and Manufacturing Program (2009ZX09501-008), the National High-tech R&D Program (2006AA02Z223), China Ministry of Education (NCET-06-0157) and the National Natural Science Foundation of China (30572274).