Tailoring modification of deoxysugars during biosynthesis of the antitumour drug chromomycin A3 by Streptomyces griseus ssp. griseus



Chromomycin A3 is a member of the aureolic acid group family of antitumour drugs. Three tailoring modification steps occur during its biosynthesis affecting the sugar moieties: two O-acetylations and one O-methylation. The 4-O-methylation in the 4-O-methyl-D-oliose moiety of the disaccharide chain is catalysed by the cmmMIII gene product. Inactivation of this gene generated a chromomycin-non-producing mutant that accumulated three unmethylated derivatives containing all sugars but differing in the acylation pattern. Two of these compounds were shown to be substrates of the methyltransferase as determined by their bioconversion into chromomycin A2 and A3 after feeding these compounds to a Streptomyces albus strain expressing the cmmMIII gene. The same single membrane-bound enzyme, encoded by the cmmA gene, is responsible for both acetyl transfer reactions, which convert a relatively inactive compound into the bioactive chromomycin A3. Insertional inactivation of this gene resulted in a mutant accumulating a dideacetylated chromomycin A3 derivative. This compound, lacking both acetyl groups, was converted in a two-step reaction via the 4E-monoacetylated intermediate into chromomycin A3 when fed to cultures of S. albus expressing the cmmA gene. This acetylation step would occur as the last step in chromomycin biosynthesis, being a very important event for self-protection of the producing organism. It would convert a molecule with low biological activity into an active one, in a reaction catalysed by an enzyme that is predicted to be located in the cell membrane.


The aureolic acid group of antitumour drugs comprises of mithramycin, the chromomycins, the olivomycins, chromocyclomycin, UCH9 (Rohr et al., 1999) and the recently discovered durhamycin A (Jayasuriya et al., 2002). They show antibacterial activity against Gram-positive bacteria and also antitumour activity (Remers, 1979; Skarbek and Speedie, 1981). One of them, durhamycin A, has also been shown to be an inhibitor of HIV Tat transactivation (Jayasuriya et al., 2002). More recently, chromomycin and mithramycin have also been found to be powerful inducers of erythroid differentiation of K562 cells, which makes them candidates for therapeutics of certain haematological diseases (Bianchi et al., 1999), and to be potent inhibitors of neuronal apoptosis, making them possible candidates for the treatment of neurological diseases (Chatterjee et al., 2001). According to the structure of the aglycon and its biosynthetic origin, they belong to the large and important family of aromatic polyketides. Structurally, all members of the family consist of a tricyclic chromophore (chromocyclomycin being an exception with a tetracyclic chromophore) that is glycosylated at two different positions with oligosaccharides of various chain length. All the single sugar moieties belong to the 6-deoxyhexoses family, and different L- and D-sugars are constituents of the saccharide chains (Rohr et al., 1999).

Chromomycin A3(Fig. 1A) is produced by different streptomycete species, usually as the main component of a mixture. At least five components have been described, differing in minor modifications of the sugar moieties (Korzybski et al., 1978). Chromomycin A2 differs from chromomycin A3 in the acyl residue bound to the 4-hydroxyl group of the 4-O-acetyl-L-chromose B (sugar E) of the trisaccharide chain (Fig. 1A): acetyl and isobutyryl in chromomycin A3 and A2 respectively. Chromomycins (and all aureolic acid members) interact with GC-rich DNA regions in a non-intercalative way in the presence of Mg2+ ions, which are essential for activity (Gao and Patel, 1989; Aich et al., 1992; Majee et al., 1997). In this respect, chromomycin A3 has different binding properties to DNA from mithramycin (Majee et al., 1997; Mir et al., 2003). It has been suggested that sugar residues play an important role in this binding process, and that the differences between both drugs regarding the stereochemistry and functional group substitutions of the sugars could be responsible for this (Majee et al., 1997; Mir et al., 2003). Thus, the acetoxy groups in the sugars D-oliose (sugar A) and L-chromose B (sugar E) of chromomycin contribute distinctively in the DNA complex formation by providing an additional H-bond with the 2-amino groups of G-bases and thus adding more specificity to the DNA binding (Silva and Kahne, 1993; Silva et al., 1993; Majee et al., 1997; Chakrabarti et al., 2001).

Figure 1.

A. Chemical structures of chromomycins A2 and A3, and of mithramycin.
B. Organization of the chromomycin biosynthetic gene cluster showing in detail the DNA regions containing the cmmMIII and cmmA genes. B, BamHI; EI, EcoRI; EV, EcoRV; Bg, BglII; S, SalI; P, PstI. Restriction sites indicated by an asterisk (SalI and PstI sites) are not unique sites in the region shown. For details on the organization of the gene cluster, see Menéndez et al. (2004).

Biosynthetic studies have shown that the chromomycin polyketide skeleton derives from one acetyl-CoA and nine malonyl-CoA units (Montanari and Rosazza, 1988; 1990). We have reported recently the cloning and characterization of the chromomycin A3 biosynthetic gene cluster (Menéndez et al., 2004). Experimental evidence of the involvement of this cluster in chromomycin A3 biosynthesis was obtained by insertional inactivation of the cmmWI that encodes a ketoreductase and the subsequent accumulation of three novel derivatives (Menéndez et al., 2004). Here, we report the functional characterization of two genes from this cluster, cmmMIII and cmmA, encoding an O-methyltransferase and an O-acyltransferase, respectively, that modify the architecture of three of the sugars in chromomycin A3 after these have been transferred to the aglycon. We also show the isolation of some biosynthetic intermediates with different degrees of methylation and acetylation at the sugar moieties. Evidence is also provided indicating that the acylation steps are essential to convert biologically inactive intermediates into the fully active compound.


Chromomycins and mithramycin possess the same aglycon and only differ in a few minor structural features regarding the sugar moieties. Both contain a trisaccharide chain attached to C-2 of the aglycon that differs in its sugar profile: D-olivose, D-oliose and D-mycarose in mithramycin and D-olivose (sugar C), D-olivose (sugar D) and 4-O-acetyl-L-chromose B (sugar E) in chromomycin A3. They also differ in the disaccharide chain attached at C-6, which consists of two D-olivose units in mithramycin and 4-O-acetyl-D-oliose (sugar A) and 4-O-methyl-D-oliose (sugar B) in chromomycin A3 (Fig. 1A). The biosynthesis of three of the sugars in chromomycin A3 requires methyl and acetyl transfer steps respectively: O-methylation at C-4 of sugar B and O-acetylations at C-4 of sugars A and E (Fig. 1). Altogether, 36 genes were identified in the chromomycin A3 biosynthetic cluster encoding the enzymatic functions required for the biosynthesis of this polyketide. Although several gene candidates were found encoding methyltransferases, only one seemed to code for an acyltransferase (Fig. 1B).

Insertional inactivation of cmmMIII

Analysis of the chromomycin gene cluster showed four genes (cmmMI, cmmMII, cmmMIII and cmmC) as potential candidates to encode the methyltransferase responsible for catalysing the O-methyl transfer step at C-4 of sugar B in the disaccharide chain of chromomycin A3. The deduced products of two of them (CmmMI and CmmMII) resemble two polyketide methyltransferases (MtmMI and MtmMII) of the mithramycin cluster: 48.9% identity between CmmMI and MtmMI and 52.4% identity between CmmMII and MtmMII. These two mithramycin methyltransferases have been shown to catalyse the 4-O- and 9-C-methylations of premithramycin intermediate steps during mithramycin biosynthesis (Fernández-Lozano et al., 2000). As these two methyl groups exist at the same positions of the chromomycin A3 aglycon, we assumed that CmmMI and CmmMII catalyse the same methyl transfer steps in chromomycin A3 biosynthesis. Database comparisons showed that the products of the other two genes (CmmC and CmmMIII) are more likely to be involved in the biosynthesis of deoxysugars. The cmmC gene product resembles C-methyltransferases involved in deoxysugar biosyntheses, e.g. a C-3 methyl transferase involved in the methylation of dehydrovancosamine in balhimycin biosynthesis, 63.2% identity (Pelzer et al., 1999), and MtmC, a 3-C-methyltransferase that introduces the 3-methyl group of the D-mycarose building block during the biosynthesis of mithramycin, 39.5% identity (González et al., 2001). Therefore, we concluded that CmmC is responsible for an analogous step in chromomycin biosynthesis, namely the C-methylation during the formation of sugar E, which is the last sugar of the chromomycin trisaccharide chain. Consequently, the only unassigned MT gene, cmmMIII, became the most probable candidate for catalysing the O-methylation step at C-4 of sugar B. This is also supported by the Blast analysis data, which revealed that the cmmMIII gene product shows significant similarities to various sugar O-methyltransferases: ElmMIII from the elloramycin cluster, 61.4% identity (Patallo et al., 2001); SpnH from the spinosyn cluster, 57% identity (Waldron et al., 2001); and CloP from the chlorobiocin cluster, 57.6% identity (Pojer et al., 2002). To test the possible involvement of CmmMIII in the O-methyl transfer step at C-4 of sugar B, we inactivated this gene. An apramycin resistance (AmR) cassette, containing the aac3(IV) gene, was subcloned into the unique BglII site within cmmMIII(Fig. 2A) and, after intergeneric conjugation (Escherichia coli to Streptomyces griseus), AmR-ThioS transconjugants were isolated. The replacement was verified by Southern hybridization: using a 2.7 kb PstI fragment as probe (sites 2–5 in Fig. 1B), two hybridizing PstI bands (1.5 and 2.6 kb) were observed in the C22MIII mutant in comparison with the 2.7 kb PstI band of the wild-type strain (Fig. 2A and B). High-performance liquid chromatography (HPLC) analysis of cultures of the C22MIII mutant showed the absence of chromomycin A3 and the presence of three major new compounds (Fig. 2C) with the characteristic absorption spectrum of chromomycin A3. The ability to produce chromomycin A3 by the mutant was recovered by expressing cmmMIII in trans (data not shown).

Figure 2.

Insertional inactivation of the cmmMIII gene.
A. Scheme representing the replacement in the chromosome of the wild-type cmmMIII gene by the in vitro mutated one. aac3(IV), apramycin resistance gene; tsr, thiostrepton resistance gene; bla, ampicillin resistance gene.
B. Southern hybridization using the 2.7 kb PstI fragment as probe. Lane 1, PstI-digested chromosomal DNA from the wild-type strain. Lane 2, PstI-digested chromosomal DNA from mutant C22MIII.
C. HPLC analysis of a culture of mutant C22MIII. The arrows indicate the mobility of chromomycin A2 (CA2) and chromomycin A3 (CA3).

Isolation and structure elucidation of compounds accumulated by the C22MIII mutant

The compounds accumulated by the C22MIII mutant were purified by preparative HPLC. The yields of the different compounds were: 3.5 mg l−1 (peak 1), 28.8 mg l−1 (peak 2) and 2.4 mg l−1 (peak 3). Their structures were elucidated by nuclear magnetic resonance (NMR) and mass spectrometry (MS).

Peak 1. 4B-O-demethyl-4B-O-acetyl-4 A-O-deacetylchromomycin A3 (1).  The molecular ions identified in the negative fast atom bombardment (FAB) mass spectrum at m/z 1167 (MH) and in the positive FAB MS at m/z 1190 (M+Na)+ allowed the deduction of the molecular formula of C56H80O26, which was also supported by the NMR data and was confirmed by HR-ESI-MS (1167.4843 observed, calculated for C56H79O26: 1167.4865). Compound 1 was obtained as a yellow amorphous solid, which showed identical UV and similar IR data to chromomycin A3 (Menéndez et al., 2004). The 1H- and 13C-NMR spectra (see Supplementary material) of compound 1 closely resembled that of chromomycin A3, except that a characteristic signal of the methoxy group of chromomycin A3 was missing, which indicated the lack of a 4B-methoxy group. The analysis of the one- and two-dimensional NMR spectra (COSY, TOCSY, HSQC and HMBC) indicated that the acetoxy group is attached at C-4B-O rather than at C-4A-O, most obvious from the downfield shift in 4B-H of ≈ 1.5 p.p.m. Thus, structure 1 was deduced for this compound, called 4B-O-demethyl-4B-O-acetyl-4A-O-deacetylchromomycin A3.

Peak 2. 4B-O-demethyl-chromomycin A3 (2).  The mole-cular ions identified in the negative FAB mass spectrum at m/z 1167 (MH) and in the positive FAB MS at m/z 1190 (M+Na)+ are in agreement with a molecular formula of C56H80O26, which also was supported by the NMR spectra and the high-resolution ESI mass spectrum (observed: 1167.4852, calculated for C56H79O26: 1167.4865). Compound 2 was obtained as a yellow amorphous solid, which also showed almost identical UV and similar IR data to chromomycin A3 (Menéndez et al., 2004). As expected from the design of the inactivation experiment, the 1H-NMR, 13C-NMR and two-dimensional NMR spectra of the major compound 2 showed that the compound lacks the 4B-methoxy group, whereas the acetoxy group remains at its normal 4A position. Thus, the MS and NMR data (see Supplementary material) analysis in comparison with the data of chromomycin A3 reveal structure 2, which is 4B-O-demethyl-chromomycin A3.

Peak 3. 4B-O-demethyl-chromomycin A2 (3).  The mole-cular ion identified in the negative FAB mass at m/z 1195 (MH) and the positive FAB MS at m/z 1218 (M+Na)+ allowed the deduction of the molecular formula to be C58H84O26, which is supported by 58 observed signals in the 13C-NMR spectrum and the highly resolved ESI MS (observed: 1195.5200, calculated for C58H83O26: 1195.5178). Compound 3 was isolated as a yellow amorphous solid, which showed similar UV and IR spectra to chromomycin A3 as well as the derivatives described above. Again, the 1H-NMR was similar to that of chromomycin A3, except for the missing signals of a methoxy and an acetyl group and the presence of an isobutoxy group, indicated by two additional methyl group doublets at δ 1.17 (d, J = 7Hz) and 1.15 (d, J = 7 Hz) and the multiplet at δ 2.59 integrating for one proton. These conclusions were also supported by the 13C-NMR spectrum (absence of the methoxy and acetoxy methyl signals, while a downfield shift at δ 176.9 along with the presence of two methyl signals at δ 19.9 and 19.8 and a signal at δ 34.1 for a methine group indicate the presence of an isobutoxy group). The two-dimensional NMR spectra confirmed that the remaining acetoxy group is located at the 4A position and the isobutoxy group at 4E position. In summary, the structure of 3 was deduced for the compound as 4B-O-demethyl-chromomycin A2.

Thus, the structure elucidation of these compounds (the complete spectroscopic assignments are given in Supplementary material) showed that all lack the methyl group at C-4 of the corresponding sugar B of the disaccharide chain. The compounds from peaks 2 and 3 also differed in the acyl residue at C-4 of sugar E: acetyl and isobutyryl respectively. Therefore, these two compounds (Fig. 3) were designated 4B-O-demethyl-chromomycin A3 (peak 2; DMC-A3) and 4B-O-demethyl-chromomycin A2 (peak 3; DMC-A2). The third compound (Fig. 3) possesses two acetyl groups: one in its normal position (C-4E at sugar E) and the other at C-4B, the position at which the O-methyl group is normally located at sugar B. This novel compound was named 4B-O-demethyl-4B-O-acetyl-4A-O-deacetylchromomycin A3 (peak 1; DMAC-A3).

Figure 3.

Chemical structures of the compounds accumulated by mutants C22MIII and C10A. The arrows indicate modified positions with respect to chromomycin A3.

The absence of the methyl group in the 4B position of all these compounds indicates that the cmmMIII product acts as a methyltransferase responsible for the methyl transfer step affecting this position.

Bioconversion of compound accumulated by the C22MIII mutant

In order to determine whether these mutant compounds could serve as substrates for the CmmMIII O-methyltransferase, two of them (DMC-A2 and DMC-A3) were tested by in vivo biotransformation. The third compound (DMAC-A3) was not assayed because the hydroxyl group at C-4 of sugar B was already modified by acetylation. The cmmMIII gene was expressed in Streptomyces albus using the expression vector pEM4, in which the inserted genes are under the control of the strong and constitutive promoter of the erythromycin resistance gene, ermEp*. The resultant recombinant strain was grown and then cultivated independently in the presence of DMC-A3 or DMC-A2. After incubation, the biotransformation products were analysed by HPLC. Both compounds were converted by the recombinant strain into new compounds that showed identical HPLC retention times and absorption spectra to chromomycin A3 (in the case of DMC-A3) and chromomycin A2 (in the case of DMC-A2) respectively. Efficiencies of bioconversion were ≈ 38% (for DMC-A3) and 30% (for DMC-A2). Mass spectra analysis of these methylation products showed molecular masses at m/z values of 1205.8 and 1234.1 (sodium adducts), respectively, which confirm their identity with chromomycin A3 and A2 respectively. As a control, S. albus (containing the vector without insert) did not convert DMC-A2 and DMC-A3 into any product.

These experiments not only confirm the above-mentioned conclusion regarding the role of the cmmMIII gene product for chromomycin biosynthesis, but also demonstrate that both DMC-A2 and DMC-A3 are substrates for the methyltransferase. These results also indicate that this methylation step occurs after the sugars have been transferred to the aglycon.

Insertional inactivation of cmmA

Comparison of the different gene products of the chromomycin biosynthetic gene cluster with proteins in databases revealed that there is only one gene, cmmA, that could code for an acyltransferase (Menéndez et al., 2004). The cmmA gene product resembles different acyltransferases that have been shown to acylate macrolide antibiotics. The highest similarities were with two enzymes acting on the aglycons, such as MdmB from the midecamycin pathway (37.7% identity; Hara and Hutchinson, 1992) and AcyA from the carbomycin pathway (35.7% identity; Arisawa et al., 1995), and two acyltransferases acting on the sugars attached to the aglycons: MegY from the megalomicin pathway (34% identity; Volchegursky et al., 2000) and CarE from the carbomycin pathway (31.2% identity; Epp et al., 1989). Interestingly, hydropathy profile analysis of the CmmA protein showed 10 transmembrane domains (Fig. 4), consistent with a location of the CmmA protein in the cell membrane. To test a possible role for CmmA in introducing acetyl groups in two of the sugars of chromomycin A3, cmmA was inactivated by gene replacement. The AmR resistance cassette was used to interrupt the cmmA gene (replacement of an internal DNA fragment from cmmA by the resistance cassette; Fig. 5A). After intergeneric conjugation, several AmR-ThioS transconjugants were isolated, and the gene replacement was verified by Southern hybridization: using as probe a 11.8 kb PstI fragment (sites 6–17 in Fig. 1B), two hybridizing PstI bands (5.7 and 6.8 kb) were observed in the C10A mutant in comparison with the 11.8 kb PstI band of the wild-type strain (Fig. 5A and B). Analysis by HPLC of cultures of the C10A mutant showed a single peak with the characteristic absorption spectrum of chromomycin A3 but differing in the HPLC retention (Fig. 5C). The ability of the mutant to produce chromomycin A3 was recovered by expressing cmmA in trans (data not shown).

Figure 4.

Hydropathy profile of the CmmA acyltransferase.

Figure 5.

Insertional inactivation of the cmmA gene.
A. Scheme representing the replacement in the chromosome of the wild-type cmmA gene by the in vitro mutated one. aac3(IV), apramycin resistance gene; tsr, thiostrepton resistance gene; bla, ampicillin resistance gene.
B. Southern hybridization using the 11.8 kb PstI fragment as probe. Lane 1, PstI-digested chromosomal DNA from the wild-type strain. Lane 2, PstI-digested chromosomal DNA from mutant C10A.
C. HPLC analysis of a culture of mutant C10A. The arrows indicate the mobility of chromomycin A2 (CA2) and chromomycin A3 (CA3).

Isolation and structure elucidation of compound accumulated by the C10A mutant

The compound in this peak was purified from cultures of C10A  mutant  by  preparative  HPLC  obtaining  20.04 mg l−1 pure compound, and its structure was elucidated by NMR and MS. The molecular formula C53H78O24 was deduced for this compound with the help of negative and positive FAB MS spectra showing ions at m/z 1097 (MH) and m/z 1121 (M+Na)+, respectively, which was also supported by the 13C-NMR spectrum, in which 54 carbon signals were observed. The 1H-NMR (acetone-d6, 400 MHz) spectrum closely resembled that of chromomycin A3, the major differences being the absence of two acetoxy group signals along with the upfield shifted 4A- and 4E-proton signals at δ 3.21 (br d, 3 Hz) and 3.10 (d, 10 Hz) respectively. This indicates the loss of both acetoxy groups that are normally located in these positions, a conclusion that was further supported by the 13C-NMR spectrum, in which both acetoxy methyl groups and two ester carbonyls were missing in comparison with the 13C-NMR spectrum of chromomycin A3. Thus, the MS and H-NMR data (see Supplementary material) in comparison with the data of chromomycin A3 reveal structure 4, in which both acetyl groups present in chromomycin A3 were missing in this compound (Fig. 3). On the basis of its structure, this compound was designated as 4A,4E-O-dideacetyl-chromomycin A3 (DDAC-A3).

Bioconversion of the compound accumulated by the C10A mutant

This dideacetylated derivative was also assayed in biotransformation experiments as a possible substrate for the CmmA acyltransferase and also to verify whether this transferase was responsible for introducing both acetyl groups in chromomycin A3. The cmmA gene was overexpressed into S. albus under the control of ermEp*, and the resultant recombinant strain was grown in the presence of DDAC-A3. After incubation, two biotransformation products were detected by HPLC analysis, representing 58% (major peak) and 18% (minor peak) bioconversion of the substrate. These products were not detected in control experiments in which S. albus (containing the vector without insert) was used for the bioconversion experiment. The major peak was identified as chromomycin A3 on account of its identical spectral characteristics, HPLC mobility and mass spectroscopic analysis (molecular peaks at m/z 1205.5 and 1221.5, respectively, for the sodium and potassium adducts). The second (minor) peak showed a lower retention time, and its mass spectroscopic analysis was consistent with a monoacetylated derivative (molecular peaks at m/z 1163.8 and 1179.8, respectively, for the sodium and potassium adducts). Its structure elucidation revealed 4A-O-deacetyl-chromomycin A3 (DAC-A3; 5) for the following reasons. The molecular formula C55H60O25 was deduced for this compound with the help of the positive FAB MS, showing ions at m/z 1163.8 (M+Na)+ and m/z 1179.8 (M+K)+. The 1H-NMR (acetone-d6, 400 MHz), which needed to be accumulated for 17 h as only 45 µg of the compound was available, was similar to that of chromomycin A3, except for the absence of an acetyl group signal and the upfield shifted 4A-proton at δ 3.25 (br d, 3 Hz). This indicates that the acetoxy group normally positioned at C-4A-O was missing, whereas the 4E acetoxy group was present, indicated by the 4E-proton in the downfield region at δ 5.14 (d, 10 Hz), as in chromomycin A3. Thus, the MS and 1H-NMR data in comparison with the data of chromomycin A3 reveal structure 5 (Fig. 3).

These results not only prove that CmmA is responsible for the introduction of both acetyl groups attached at C-4A and C-4E, but also reveals the sequence in which this enzyme probably acts. Furthermore, the experiments demonstrate that the acetylations occur after the attachment of the sugars to the aglycon.

Biological activity

The biological activity of the isolated compounds was tested. All three demethylated derivatives (DMAC-A3, DMC-A3 or DMC-A2) showed antibiotic activity against Micrococcus luteus comparable to that of chromomycin A3[minimum inhibitory concentration (MIC) 2.5 µg ml−1]; however, the dideacetylated derivative (DDAC-A3) was less active (MIC 20 µg ml−1). The major compounds accumulated by both mutants (DMC-A3 and DDAC-A3) were also tested for antitumour activity against a panel of human tumour cell lines (Table 1). Both derivatives showed antitumour activity against all cell lines tested; however, this was lower than that of chromomycin A3, especially for compound DDAC-A3.

Table 1. . Antitumour activity tests of chromomycin derivatives generated in this work.
Tumour cell lines Log(GI50a)
Chromomycin A34B-O-demethyl-chromomycin A34A, 4E-O-dideacetyl-chromomycin A3
  • a

    . GI50, 50% growth inhibition.



Many bioactive natural products are glycosylated compounds in which the sugar moieties are important, sometimes essential, for the biological activity. These sugars usually belong to the wide family of the 6-deoxyhexoses (Piepersberg, 1994; Kirschning et al., 1997; Weymouth-Wilson, 1997), and they are transferred to the corresponding aglycon by specific glycosyltransferases after their biosynthesis into NDP-activated species through successive enzymatic steps (Liu and Thorson, 1994; Trefzer et al., 1999). Only some deoxysugars undergo further tailoring modifications once they have been transferred to the aglycon. Many of these modifications are sugar O-methylations, for example the conversion of an L-olivosyl into an L-oleandrosyl residue by the 3-O-methyltransferase OleY during oleandomycin biosynthesis in Streptomyces antibioticus (Rodriguez et al., 2001) or the conversion of erythromycin C into erythromycin A in Saccharopolyspora erythraea. This step consists of a 3-O-methylation of an L-mycarosyl into the final L-cladinosyl moiety catalysed by the methyltransferase EryG (Paulus et al., 1990). In Streptomyces fradiae, two O-methylation steps catalysed by the methyltransferases TylE and TylF convert the deoxyallose moiety into mycinose as last steps in tylosin biosynthesis (Bate and Cundliffe, 1999). Less frequently found is sugar tailoring modification by acylation. The carE gene from Streptomyces thermotolerans encodes an acyltransferase modifying the mycarose moiety during biosynthesis of carbomycin through the introduction of an isovaleryl group (Epp et al., 1989). The chromomycin biosynthesis requires two acetylation steps and one methylation step. Evidence shown in this paper points to the cmmA and cmmMIII gene products as the enzymes responsible for catalysing these tailoring modifications. These two enzymes also act after all deoxysugars in chromomycin have been transferred as: (i) the insertional inactivation of both genes generated compounds with a complete set of deoxysugars attached but specifically lacking the methyl or acetyl groups depending on the gene inactivated; and (ii) when the fully glycosylated (but not methylated or acetylated) appropriate substrates were fed to S. albus expressing either cmmMIII or cmmA, the corresponding modified (methylated or acetylated) compounds were recovered. Based on the following observations, we believe that the CmmMIII methyltransferase acts before the CmmA acyltransferase. First, the C10A mutant in which the cmmA gene has been inactivated only accumulated one compound, 4A,4E-O-dideacetylchromomycin A3; this compound lacks both acetyl groups, but already possesses the methyl group at C-4 of sugar B. Secondly, bioactivity testing of the compounds accumulated by both mutants showed that the presence of the acetyl groups is absolutely required for high biological activity of the compound, whereas the methyl group is not so important. Therefore, it is likely that the acetylations, which ultimately lead to bioactive drugs, occur at the very end of the biosynthesis (Fig. 6). This is also supported by observations that the acyltransferase is probably located in the cell membrane (see below), thus activating the compounds during the export out of the cell. Nevertheless, the acyltransferase can also act on unmethylated derivatives as it is deduced from the isolation of compounds lacking the methyl group but containing acetyl groups, which reflects some kind of flexibility. This enzyme also has some substrate flexibility with respect to both the acylation position (C-4 A or C4B, see 4B-O-demethyl-chromomycin A3) and the acyl group transferred (acetyl or isobutyryl).

Figure 6.

Proposed biosynthetic pathway for late steps in chromomycin A3 biosynthesis involving the participation of the CmmMIII and CmmA enzymes. Methyl and acetyl groups incorporated by the CmmMIII and CmmA enzymes are indicated in bold.

Interestingly, in spite of the existence of two acetyl groups in chromomycin A3, only one gene, cmmA, coding for an acyltransferase was found in the chromomycin cluster. Several lines of experimental evidence support the view that its product is responsible for the introduction of both acetyl groups. Insertional inactivation of cmmA only produced a unique accumulated intermediate, 4A,4E-O-dideacetylchromomycin A3, lacking both acetyl groups. Furthermore, this compound was readily converted into chromomycin A3 when fed to a recombinant S. albus strain expressing cmmA, thus showing that: (i) this compound is a real biosynthetic intermediate and not a shunt product; and (ii) expression of the CmmA acyltransferase is sufficient to convert the dideacetylated derivative into chromomycin A3. It is worth mentioning that the CmmA acyltransferase is capable of acting on both L- and D-sugars (sugars E and A). The fact that a monoacetylated derivative containing an acetyl group at C-4 of sugar E was isolated as a minor compound after adding the dideacetylated derivative to cultures of an S. albus strain expressing cmmA strongly suggests that the acyltransferase acts first on sugar E and then on sugar A. This view is also supported by the work of Kawano et al. (1990), who isolated a monoacetylated chromomycin derivative lacking the acetyl group at sugar A from fermentation broths of Streptomyces avellaneus, another chromomycin producer.

The acetylation steps can be considered as crucial enzymatic events during chromomycin biosynthesis. The dideacetylated chromomycin shows low biological activity compared with the acetylated compound (i.e. chromomycin A3). Therefore, these acetylation steps are essential to convert low active biosynthetic intermediates into a fully active compound. Considering survival of the producer organism during the production phase, it would be beneficial for the organism to activate the final product in the vicinity of the cell membrane. Hydropathy profile of the CmmA protein shows 10 transmembrane domains, characteristic of membrane-embedded proteins, thus supporting the view that the acetylation events could occur associated with the cell membrane. It has to be mentioned that the ABC transporter coded by the cmrA and cmrB genes, which is present in the chromomycin cluster, confers a high level of resistance to dideacetyl chromomycin A3 but a low level to chromomycin A3 and demethyl-chromomycin A3 (N. Ménendez, unpublished results). This suggests that, during chromomycin A3 biosynthesis, the ABC transporter recognizes and transports dideacetyl chromomycin A3, and this intermediate would be now acetylated by the CmmA acyltransferase. This would require a close co-operation between the acyltransferase and the chromomycin export system. This system would provide the producer organism with an efficient self-resistance mechanism to avoid its suicide. There have been reports on resistance mechanisms in producer organisms in which the organism produces and secretes an inactive compound that is activated outside the cell by removing a glucose moiety from its structure (Vilches et al., 1992). However, in chromomycin biosynthesis, the situation would be quite different, as the formation of a fully active antibiotic would require the incorporation of functional groups (i.e. acetyl groups) during the exportation process. Experiments now in progress are focused on determining the membrane location of the acyltransferase, understanding the biochemistry of the acetylation process and its interaction with the transport system.

Experimental procedures

Microorganisms, culture conditions and vectors

Streptomyces griseus ssp. griseus ATCC13273, chromomycin A3 producer, was used as the donor of chromosomal DNA. For sporulation on solid medium, it was grown at 30°C on plates containing A medium (Fernández et al., 1998). For growth in liquid medium, the organisms were grown on either TSB medium (trypticase soy broth; Oxoid) or R5A medium (Fernández et al., 1998). Streptomyces albus J1074 (ilv-1, sal-2) (Chater and Wilde, 1980) was used as host for gene expression. Escherichia coli DH10B (Invitrogen) was used as host for subcloning, and E. coli ET12567 (pUB307) (Flett et al., 1997) was used as donor for intergeneric conjugation. pUC18 (Yanisch-Perron et al., 1985), pBSK (Stratagene) and pUK21 (Vieira and Messing, 1991) were used for subcloning in E. coli. pEM4 (Quirós et al., 1998) and pHZ1358 (Sun et al., 2002) were used for gene expression and gene replacement respectively. When plasmid-containing clones were grown, the medium was supplemented with the appro-priate antibiotics: 5 or 25 µg ml−1 thiostrepton for liquid or solid cultures respectively; 100 µg ml−1 for ampicillin; 25 µg ml−1 apramycin or 20 µg ml−1 tobramycin.

DNA manipulation and sequencing

Plasmid DNA preparations, restriction endonuclease digestions, alkaline phosphatase treatments, DNA ligations and other DNA manipulations were performed according to standard techniques for E. coli (Sambrook et al., 1989) and Streptomyces (Kieser et al., 2000). Preparation of S. albus protoplasts, transformation and selection of transformants was carried out as described previously (Kieser et al., 2000). Intergeneric conjugation from E. coli ET12567 (pUB307) to S. griseus ssp. griseus was performed as described previously (Mazodier et al., 1989). Computer-aided database searching and sequence analyses were made using the GCG sequence analysis software package of the University of Wisconsin Genetics Computer group and the Blast program (Altschul et al., 1990). Hydropathy profile was determined using the Tmhmm transmembrane prediction program (Krogh et al., 2001).

Insertional inactivation

cmmMIII.  A 3.4 kb BamHI fragment (sites 1–4 in Fig. 1B) was subcloned into the BamHI site of pUC18. Then, a 1.5 kb BamHI–BglII fragment containing an apramycin resistance (AmR) cassette, aac3(IV) gene, from pUO9090 (M. C. Martin, unpublished results) was inserted into a unique BglII site located within the cmmMIII gene (site 3 in Fig. 1B) and in the same direction of transcription of cmmMIII. The insert was then rescued by digestion with BamHI and subcloned into the same site of pHZ1358, generating pC22MIII (Fig. 2).

cmmA.  A 4.8 kb SalI fragment (sites 8–13 in Fig. 1B) was subcloned into the same site of pUC18, and the insert was rescued as an XbaI–HindIII fragment and subcloned into the same sites of pBSK. Then, the AmR cassette was inserted as a 1.5 kb BamHI–HindIII (both ends blunt-ended) fragment into the EcoRV–EcoRI (the latter blunt-ended) sites (sites 9–10 in Fig. 1B) located within cmmA. The in vitro mutagenized fragment, now 5.4 kb, was then rescued as a XbaI–HindIII fragment and subcloned into the same sites of pUK21. The fragment was then rescued as a SpeI fragment and finally subcloned into the XbaI site of pHZ1358, thus generating pC10A (Fig. 5).

PCR amplification and gene expression

cmmMII.  The following oligoprimers were used: 5′-GGACTA GTCGCGCAGGAGGAAGCATG-3′ (SpeI site underlined) and 5′-GCTCTAGACCTAGGTCACCCGGTCCTGCGCC-3′ (XbaI and AvrII sites underlined).

cmmA.  The following oligoprimers were used: 5′-GGACTA GTCCTAGGGGACGAAAGAGGCAGGATG-3′ (SpeI and AvrII sites underlined) and 5′-GCTCTAGAGCTAGCTCACCT CTCCACCATGTGTG-3′ (XbaI and NheI sites underlined).

PCR conditions were as follows: 100 ng of template DNA were mixed with 30 pmol of each primer and 2 units of Pfx DNA polymerase (New England Biolabs) in a total reaction volume of 50 µl containing 2 mM each dNTP, 10 mM KCl, 10 mM (NH4)2SO4, 20 mM Tris-HCl (pH 8.8), 2 mM MgSO4 and 0.1% Triton X-100. The polymerization reaction was performed in a thermocycler (MiniCycler; MJ Research) under the following conditions: an initial denaturation of 2 min at 94°C; 30 cycles of 15 s at 94°C, 30 s at 55°C and 1 min at 68°C; after the 30 cycles, an extra extension step of 5 min at 68°C was added.

After PCR amplification, the amplicons containing the cmmMIII and cmmA genes were independently subcloned as SpeI–XbaI fragments into the XbaI site of pEM4 under the control of the ermE*p, generating the final constructs pNMS-MIII and pNMS-A respectively.

Biotransformation experiments

The appropriate S. albus recombinant clones expressing the cmmMIII and cmmA genes were grown on TSB liquid medium for 24 h. This preinoculum was used to inoculate Erlenmeyer flasks containing 50 ml of R5A liquid medium and 5 µg ml−1 thiostrepton. After 24 h incubation, the cultures were filtrated, and the mycelial paste was resuspended in 3 ml of R5A medium containing thiostrepton. Then, 100 µg ml−1 4B-O-demethyl-chromomycin A3, 4-B-O-demethyl-chromomycin A2 or 4A,4E-O-dideacetyl-chromomycin A3 (depending on the experiment) was added, and further incubation was carried out for 24 h. After incubation, samples (1 ml) were centrifuged, and the supernatant was extracted with 0.5 volumes of ethyl acetate. The organic solvent was dried under vacuum, and the residue was finally suspended in a small volume of methanol before HPLC analysis.

HPLC analysis and isolation of intermediates

HPLC analysis of chromomycin-related compounds was performed as described previously (Fernández-Lozano et al., 2000). For purification of the compound accumulated by mutant C10A, spores of this strain were initially grown in TSB medium for 48 h at 30°C and 250 r.p.m. This seed culture was used to inoculate (at 2% v/v) 50 250 ml Erlenmeyer flasks containing 50 ml of R5A medium. After incubation for 6 days in the above conditions, the cultures were centrifuged, filtered and extracted (Fernández et al., 1998). The accumulated product was purified by preparative HPLC in a µBondapak C18 radial compression cartridge (PrepPak cartridge, 25 × 100 mm; Waters). A mixture of acetonitrile and 0.1% trifluoroacetic acid (TFA) in water (37:63) at 10 ml min−1 was used for elution in isocratic conditions. The purified material was diluted fourfold with water, applied to a solid-phase extraction column (SepPak Vac C18; Waters), washed with water, eluted with methanol and dried in vacuo.

Mutant C22MIII was grown on R5A solid medium. One hundred and forty plates were inoculated with spores and incubated for 4 days at 28°C. Agar cultures were removed from the plates, placed in five 2 l Erlenmeyer flasks, covered with ethyl acetate and extracted for 3 h at 30°C and 150 r.p.m. in an orbital shaker. The organic extracts were evaporated in vacuo, and the extraction procedure was repeated twice. The pooled extracts were redissolved in 7 ml of a mixture of DMSO and methanol (50:50) and chromatographed using the preparative column described above. Three peaks were collected in a first purification with acetonitrile and 0.1% TFA in water (60:40) as solvents. Peak 3 was repurified in the same conditions, but with acetonitrile reduced to 55%. Peak 1 required two repurifications: the first one with acetonitrile and 0.1% TFA in water (40:60) and, finally ,with methanol and 0.1% TFA in water (70:30). The material collected at every step was desalted and concentrated using C18 solid-phase extraction cartridges, as explained above, and finally dried in vacuo.

Antibacterial and antitumour activities

Antibacterial activity was tested against Micrococcus luteus as described previously (Vilches et al., 1990). The antitumour activity of the compounds was tested against a variety of tumour cell lines (for details, see Table 1). Quantitative measurement of cell growth and viability was carried out using a colorimetric type of assay, using sulphorhodamine reaction (Skehan et al., 1990).

Mass spectra analysis

Analysis of the bioconversion products was carried out using matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectra analysis on a Voyager-DE STR. The samples were dissolved in an acetonitrile−0.1% TFA in water (1:1) solution and mixed with the 2,5-dihydroxybenzoic acid matrix using different serial dilutions. Spectra were carried out on a reflector mode.


This work was supported by grants from the Programa Nacional de Promoción General del Conocimiento (PB98-1572 to C.M.), Proyecto de Investigación Científica y Desarrollo Tecnológico (BMC2002-03599 to C.M.), the Plan Regional de Investigación del Principado de Asturias (GE-MEDO1-05 to J.A.S.) and the US National Institutes of Health (CA 91901, to J.R.). N.M. was the recipient of a predoctoral fellowship of the FICYT.

Supplementary material

The following material is available from http://www.blackwellpublishing.com/products/journals/suppmat/mmi/mmi4166/mmi4166sm.htm

Appendix S1.  Other physicochemical properties of some of the new compounds.

TableS1.  1H-NMR data* of the chromomycin derivatives 1–5 at 400 MHz.

TableS2.  13C-NMR data of the chromomycin derivatives 1–4 at 100.6 MHz, d in p.p.m.