| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Cubist Pharmaceuticals, Inc., 65 Hayden Avenue, Lexington, MA 02421, USA
2 Cubist Pharmaceuticals, Slough, 545 Ipswich Road, Slough SL1 4EQ, UK
Correspondence
Vivian Miao
vmiao{at}cubist.com
Richard H. Baltz
Rbaltz{at}cubist.com
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
The GenBank/EMBL/DDBJ accession number for the sequence of pCV1 is AY787762.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
; Debono et al., 1988). It has potent antibacterial activity in vitro against Gram-positive pathogens, including vancomycin-resistant Staphylococcus aureus, methicillin-resistant S. aureus (MRSA), penicillin-resistant Streptococcus pneumoniae, vancomycin-resistant enterococci and other antibiotic-resistant strains (Akins & Rybak, 2001; Baltz, 1997; Barry et al., 2001; Cha et al., 2003; King & Phillips, 2001; Tally & DeBruin, 2000; Tally et al., 1999; Wise et al., 2001). Daptomycin has a novel mechanism of action that is not yet fully understood (Alborn et al., 1991; Allen et al., 1991; Silverman et al., 2001), but that apparently accounts for its activity against microbes resistant to antibiotics in current clinical use. Cubicin® (Daptomycin-for-injection) has been approved recently for the treatment of Gram-positive infections of skin and skin structures.
Daptomycin is a member of the A21978C complex of acidic lipopeptide antibiotics produced by S. roseosporus (Baltz, 1997; Debono et al., 1988). The depsipeptide portion of A21978C contains a 13 amino acid chain linked by an ester bond between the carboxyl group of L-kynurenine13 (kyn) and the hydroxyl group of L-Thr4 to form a 10 amino acid ring with a three amino acid tail (Fig. 1a). The three major components, A21978C1–3, have 11-, 12- or 13-carbon branched-chain fatty acids, respectively, attached to the terminal amino group of L-Trp (Debono et al., 1987). The fatty acid side chains of A21978C are readily removed by incubation with Actinoplanes utahensis (Boeck et al., 1988) or with the A. utahensis deacylase enzyme produced by a recombinant strain of Streptomyces lividans (Kreuzman et al., 2000), and the cyclic peptide can be reacylated with n-decanoyl fatty acid to produce daptomycin. Alternatively, for the production of daptomycin, decanoic acid can be added to the fermentation of S. roseosporus (Huber et al., 1988) to direct incorporation of the straight-chain supplement.
|
; Wessels et al., 1996). NRPSs are typically very large enzyme complexes composed of several subunits, each with specialized catalytic domains arranged in an orderly fashion to recognize specific amino acids and join them together in the proper order (Marahiel et al., 1997). Transposon mutagenesis, partial DNA sequence analysis of cosmid clones and gene-disruption experiments were used to identify the genes for the daptomycin (dpt) synthetase, while PFGE data mapped the locus to one end of the linear chromosome of S. roseosporus (McHenney et al., 1998). S. roseosporus can be manipulated by a variety of molecular genetic methods (Hosted & Baltz, 1996, 1997; McHenney & Baltz, 1996) and presents a suitable system to explore the genetic engineering of the NRPS genes to produce derivatives of daptomycin altered in peptide structure. In this study, a 128 kb DNA region from S. roseosporus including the dpt locus was cloned, validated by heterologous expression in S. lividans and sequenced completely. The core NRPS and modifier genes were characterized and a prediction, based on sequence analysis, that the Asn2 residue has D- rather than L- stereochemistry as reported earlier (Debono et al., 1987) was confirmed chemically. |
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
) with pCV1. Staphylococcus aureus SA399 and SA42, Enterococcus faecium EF14 and Enterococcus faecalis EF201 were from the Cubist culture collection and are all susceptible to daptomycin.
Media.
Composition is given as % (w/v) for solid components and % (v/v) for the trace salts solution, Tween 80 and antifoam A. Medium F10A was composed of 0·3 % CaCO3, 0·5 % distillers solubles (Sigma), 2·5 % soluble starch, 0·5 % yeast extract, 0·2 % glucose and 0·5 % bactopeptone. Medium A9 was composed of 1·8 % agar, 2 % oats, 0·7 % tryptone, 0·2 % soya peptone, 0·5 % NaCl and 0·1 % trace salts solution (per litre, 1 ml 1 M H2SO4, 1·722 g ZnSO4.7H2O, 1·112 g FeSO4.7H2O, 0·223 g MnSO4.4H2O, 0·062 g H3BO3, 0·125 g CuSO4.5H2O, 0·048 g Na2MoO4.2H2O, 0·048 g CoCl2.6H2O, 0·083 g KI). Seed medium A355 was composed of 1 % glucose, 1·5 % glycerol, 1·5 % soya peptone, 0·3 % NaCl, 0·5 % malt extract, 0·5 % yeast extract, 0·1 % Tween 80 and 2 % MOPS, pH 7. Medium A345 was similar to A355, but with 4·6 % MOPS. Production medium A346 was composed of 1 % glucose, 2 % soluble starch, 0·5 % yeast extract, 0·5 % casein and 4·6 % MOPS, pH 7.
Culture methods.
Spores and hyphal fragments of S. lividans transformants grown on A9 (with 100 mg apramycin l–1) were inoculated into 250 ml baffled flasks containing 40 ml A345 and shaken at 240 r.p.m. for 44 h at 30 °C. Five per cent of this seed was transferred to 250 ml baffled flasks containing 50 ml A346 and shaken at 240 r.p.m. for up to 7 days at 30 °C. For production in fermenters, biomass from 9-day cultures of S. lividans CBUK 136742 on A9 was inoculated into a 2 l baffled Erlenmeyer flask containing 250 ml A355 (with 0·1 % antifoam A, 25 mg apramycin l–1) and shaken at 240 r.p.m. for 2 days at 30 °C. This seed was used to inoculate 14 l A346 (with 0·1 % antifoam A) in a 20 l fermenter stirred at 350 r.p.m., aerated at 0·7 v.v.m. and maintained at 30 °C. After 20 h, a 50 % (w/v) glucose feed at 5 g h–1 was introduced. After 40 h, a 50 : 50 (w/w) blend of decanoic acid : methyl oleate feed at 0·5 g h–1 was initiated.
Bioassay against Gram-positive pathogens.
MICs against test organisms were determined by broth microdilution according to NCCLS guidelines, except that Mueller–Hinton broth was supplemented to 50 mg Ca2+ l–1 and all assays were performed at 37 °C (Silverman et al., 2001; NCCLS, 2003). MIC values were reported as the lowest concentration of a compound that prevented growth of the bacteria.
Construction of S. roseosporus BAC library.
S. roseosporus cells from an overnight culture in 25 ml F10A were washed in sterile water and embedded in 1 % Seakem GTG agarose (FMC Bioproducts) to make plugs. The plugs were incubated in 2 mg lysozyme ml–1 for 2 h and then transferred to a solution of 0·2 mg proteinase K ml–1 and 1 % sarcosine (w/v) at 50 °C overnight to lyse cells and release genomic DNA. Plugs were washed extensively in 0·1 mM EDTA and treated with BamHI. Partially digested DNA was fractionated by PFGE and 200 ng of 75–145 kb DNA fragments recovered by electroelution was ligated to 75 ng BamHI-digested, phosphorylated pStreptoBAC V vector DNA (M.-F. Coëffet-LeGal and V. Miao, manuscript in preparation) and electroporated into Escherichia coli DH10B. Cells were plated on LB plates with 100 µg apramycin ml–1 and 5 % sucrose. Approximately 2000 clones were archived in microtitre plates as the B12 library.
Identification of the dpt gene cluster on a BAC plasmid.
Screening for clones carrying the dpt genes was conducted by PCR on cell lysates of the library with primers P61 (5'-GCTCGTCCCCCTCCCCGCACT-3') and P62 (5'-CGAACAGGTGGGCTTTGAGTGG-3') using Taq polymerase (Gibco-BRL) under the following conditions: 94 °C for 45 s, 54 °C for 30 s and 72 °C for 1 min, 32 cycles. These primers, based on GenBank accession no. AF021262 (McHenney et al., 1998), target the NRPS genes. All reactions included 5 % DMSO. Positive clones (pCV1 to pCV6) were mapped and end-sequenced to determine their position on the final contig.
DNA sequencing and analysis.
Cosmids pRHB159 and pRHB160 (McHenney et al., 1998) were shotgun sequenced (Genome Therapeutics) while portions of BAC clone pCV1 were sequenced by primer walking or by sequencing (SeqWright) of a subclone. Assembly and analysis of the final contig (GenBank accession no. AY787762) was performed using MacVector (Accelrys), BLAST (Altschul et al., 1990) queries of NCBI databases and COG (Marchler-Bauer et al., 2003), as well as the Streptomyces coelicolor (http://streptomyces.org.uk/) and Streptomyces avermitilis (http://avermitilis.ls.kitasato-u.ac.jp/) genome project databases. Deduced protein alignments and dendrograms (constructed by the neighbour-joining method, default parameters) were prepared using MacVector. The binding-pocket residues for amino-acid-activating (A) domains were recognized by alignment to conserved residues at positions 235, 236, 239, 278, 299, 301, 322 and 330 of PheA (Stachelhaus et al., 1999).
Heterologous expression of BAC clones in S. lividans.
Transformation of S. lividans TK64 was performed under standard conditions (Kieser et al., 2000). Fresh protoplasts from 40 h cultures were mixed with 250 ng plasmid DNA and incubated for 18 h at 30 °C before application of 100 µg apramycin ml–1 (final) in a 20 % glycerol overlay for selection; transformants were isolated after 3 days. For the initial test of expression, purified transformants were inoculated into 25 ml F10A in 125 ml flasks shaken at 250 r.p.m. at 30 °C. Broths collected after 6 days were centrifuged and 10–20 µl supernatant was bioassayed by disc diffusion against Staphylococcus aureus. Clarified samples were also assessed for A21978C lipopeptides by HPLC and electrospray ionization liquid chromatography-mass spectrometry (ESI LC-MS) (below). For Southern analysis, chromosomal DNA was isolated from streptomycetes (Kieser et al., 2000), digested with BamHI, fractionated on a 1 % gel and transferred to a nylon membrane. The blot was probed with 32P-labelled pCV1 (DECAprime II; Ambion) at 55 °C overnight (Ultrahyb; Ambion), washed twice in 2x SSC/0·1 % SDS (w/v) for 15 min at room temperature and twice in 0·5x SSC/0·1 % SDS for 30 min at 65 °C and exposed to film.
Isolation of daptomycin from 20 l fermentation broth.
The 20 l fermentation was harvested after 115 h by centrifugation and the supernatant (approx. 10 l) was loaded onto a 60x300 mm column of Diaion HP20 resin (Mitsubishi Chemical Co.) pre-equilibrated with water at 100 ml min–1. The column was washed with 2 l water and then with 1·5 l 80 % aqueous methanol. Bound material was eluted with 2 l methanol and evaporated under vacuum to yield an aqueous concentrate that was diluted to 1 l with deionized water, washed three times with ethyl acetate (each 700 ml) and lyophilized. Further purification was achieved by HPLC using a radially compressed cartridge column consisting of two 40x100 mm Waters Nova-Pak C18 60Å 6 µm units and a 40x10 mm Guard-Pak with identical packing. The lyophilized material (175 mg) was dissolved in water and chromatographed using a water/acetonitrile gradient increasing linearly from 10 to 80 % acetonitrile over 10 min and then ramped up to 100 % acetonitrile over 1 min. The absorbance of the eluate was monitored at 223 nm. The peak eluting at 9 min was collected from repeated chromatographic runs and dried in vacuo to yield 30 mg purified compound. This was analysed by HPLC, ESI LC-MS and NMR (below).
Characterization of lipopeptides A21978C1–3 and daptomycin.
Production of lipopeptides A21978C1–3 and daptomycin was monitored by HPLC performed at ambient temperature using a Waters Alliance 2690 HPLC system and a 996 photodiode array detector with a Waters Symmetry column (4·6x50 mm, 3·5 µm particle size) and a Phenomenex Security Guard C8 cartridge. The mobile phase, buffered with 0·01 % trifluoroacetic acid (TFA), was initially held at 90 : 10 water/acetonitrile for 2·5 min, followed by a linear gradient over 6 min to 100 % acetonitrile; the flow rate was 1·5 ml min–1. All lipopeptide concentrations were determined by comparison with reference daptomycin purified from S. roseosporus fermentations and provided by Cubist Pharmaceuticals, Inc., Manufacturing Department. ESI LC-MS was performed on a Finnigan SSQ710C instrument interfaced to a Waters 600-MS HPLC system. Chromatographic separation was achieved on a Symmetry C8 column (2·1x50 mm, 3·5 µm) eluted at ambient temperature with a linear water/acetonitrile gradient containing 0·01 % formic acid, increasing from 10 to 100 % acetonitrile over 6 min after an initial delay of 0·5 min and then remaining at 100 % acetonitrile for 3·5 min before reequilibration. The flow rate was 0·35 ml min–1. ESI MS data were collected in positive ion mode with a scan range of 200–2000 Da and 2 s scans. The electrospray capillary voltage was 21·2 V with a collisional induced dissociation offset of 0 or –10 V. The capillary temperature was maintained at 250 °C. 1H and 13C NMR spectra were recorded at 308 K on a Bruker ACF400 spectrometer at 400 and 100 MHz, respectively.
Hydrolysis and enzymic cleavage of daptomycin to yield lipohexapeptide fragment (compound 2; Fig. 6a).
Daptomycin (1 mg) was dissolved in 1 ml of 1 mg LiOH (aq.) ml–1 and incubated at room temperature for 50 min. The pH was reduced to <4·5 with concentrated formic acid (0·5 ml) to stop hydrolysis. The sample was loaded onto an SPE Bondesil 40 µM C8 cartridge, washed with 3 ml water and then eluted with 3 ml methanol. After evaporation to remove methanol, the linear peptide hydrolysis product of daptomycin (500 µg) was dissolved in 1 ml water and 2 µg Asp-N (Roche) was added. After incubation at 37 °C for 16 h, the solution was frozen in dry ice to stop the reaction and lyophilized to recover the products.
|
2 ml and added to cold ether (10 ml). The resultant precipitate was pelleted by centrifugation (10 000 r.p.m., 5 min) and dried under vacuum. Reverse-phase (C18) HPLC purification [30 % acetonitrile, 70 % aqueous ammonium hydrogen phosphate buffer (0·5 % w/v) over 30 min] followed by desalting and freeze-drying gave the desired peptides as white solids (D-Asn-containing peptide, 5·7 mg; L-Asn-containing peptide, 6·1 mg); D-Asn mass spectrometry (MS) m/z 860·10 (MH+), L-Asn MS m/z 860·24 (MH+).
HPLC analysis of diastereoisomers.
Analytical HPLC was conducted on a 250x4·6 mm IBSIL 5 C8 column eluted at a flow rate of 1·5 ml min–1 with an acetonitrile/aqueous ammonium hydrogen phosphate (0·5 % w/v) gradient starting at 30 % acetonitrile for 3 min and then increasing linearly to 40 % acetonitrile over 15 min, holding at this composition for 5 min before returning to the initial conditions over 3 min and reequilibrating for 2 min. The absorbance of the eluate was monitored at 254 nm. The retention times of the D-asn and L-Asn hexapeptides were 9·7 min and 10·1 min, respectively.
General Fmoc deprotection.
The resin-bound, Fmoc-protected amino acid or peptide (0·2 mmol) was twice suspended and agitated in a solution of 20 % piperidine in 5 ml dimethylformamide (DMF). The first wash was for 5 min, the second for 25 min. The resin was then washed three times with each of these solvents: DMF, methanol and DMF. Kaiser tests on two to three beads gave a positive blue colour at this stage, indicating the presence of a free amine.
General coupling method.
Solutions (0·5 M) of the Fmoc-protected amino acid to be coupled in DMF (1 mmol), 1-hydroxy-7-azabenzotriazole in DMF (1 mmol) and 1,3-diisopropylcarbodiimide in DMF (1 mmol) were added to the resin-bound amino acid or peptide (0·2 mmol). The resultant slurry was agitated for 1 h and then filtered and the coupling procedure was repeated for 1 h. The resin was then washed as above.
Preparation of deacyl-Orn-(NHAlloc)-daptomycin (compound 3; Fig. 6c
).
Allyl-1-benzotriazolylcarbonate (1·35 g) was added to a solution containing 10 g daptomycin in 40 ml dry DMF at 0 °C and stirred for 18 h, after which the solution was warmed to room temperature. DMF was removed under vacuum, leaving crude Orn-(NHAlloc)-daptomycin (Alloc; allyloxycarbonyl) as a brown oil (10·5 g) [MS m/z 1705·61 (MH+)]. This was dissolved in water (1·9 l) at pH 8 and deacylase, produced from recombinant S. lividans (Boeck et al., 1988; Debono et al., 1987; Kreuzman et al., 2000) and dissolved in 10 ml aqueous ethylene glycol, was added. The mixture was stirred at room temperature for 18 h, adjusted to pH 8 with 1 M NaOH and then poured on to Bondesil 40 µM C8 resin (400 g, prewashed with 1 l methanol and 1 l water). The product was eluted with 1 l water and freeze-dried to give compound 3 as a yellow solid (8·68 g, 91 % over two steps) [MS m/z 1551·32 (MH+), 776·22 ([M+2H]2+)].
Preparation of deacyl-deTrp-deAsn-Orn-(NHAlloc)-daptomycin (compound 4; Fig. 6c).
n-Decylisothiocyanate (1·46 g, 7·35 mmol) was added to a suspension of compound 3 (10·35 g, 6·7 mmol) in 100 ml dry DMF. The mixture was stirred at room temperature for 18 h. DMF was removed by evaporation at reduced pressure and the residue was dissolved in water; this was poured on to Bondesil 40 µM C8 resin prepared as before. Unreacted starting material 3 (1 g) was eluted with 20 % acetonitrile in water and the product was eluted with methanol. Evaporation of methanol gave the Orn-(NHAlloc)-Trp-decylthiourea daptomycin derivative as a yellow solid (3 g, 27 %) [MS m/z 1687·52 (MH+), 844·24 ([M+2H]2+)]. This solid (3 g, 1·77 mmol) was stirred at room temperature in 25 % TFA in dry CH2Cl2 (30 ml) for 2 h and then evaporated to dryness. The residue was dissolved in water (20 ml) and poured on to Bondesil 40 µM C8 resin prepared as above. The product was eluted with a 20–40 % gradient of acetonitrile in water and freeze-dried to give deacyl-deTrp-Orn-(NHAlloc)-daptomycin as a yellow solid (1·2 g, 49 %) [MS m/z 1365·31 (MH+), 683·66 ([M+2H]2+)]. n-Decylisothiocyanate (80 µl, 0·418 mmol) was added to a suspension of deacyl-deTrp-Orn-(NHAlloc)-daptomycin (0·57 g, 0·418 mmol) in 5 ml dry DMF. The reaction was stirred at room temperature for 18 h and evaporated to dryness. The residue was suspended in methanol and evaporated to a crude solid. Trituration with ether gave the Orn-(NHAlloc)-deTrp-deAsn-decylthiourea daptomycin derivative as a yellow powder (0·54 g), which was dried under high vacuum and used without further purification [MS m/z 1565 (MH+), 783 ([M+2H]2+)]. The decylthiourea intermediate (0·54 g) was stirred in 50 % TFA in dry CH2Cl2 (4 ml) for 2 h and evaporated to dryness. The residue was dissolved in water (25 ml) and poured on to Bondesil 40 µM C8 resin (50 g), prewashed with methanol (100 ml) and water (100 ml). The product was eluted with 20 % acetonitrile in water after first being washed with water (100 ml). The eluate was evaporated to dryness to give the title compound 4 as a yellow solid (0·40 g, 76 % over two steps) [MS m/z 1252 (MH+), 626 ([M+2H]2+)].
Preparation of N-decanoyl-L-trp-L-Asn(Trt)-O-pentafluorophenyl (OPfp).
The acylated N-decanoyl-L-trp-L-Asn(Trt)-ClTrityl resin-bound peptide (Trt, trityl) was made using the standard deprotection and coupling methods above and commercially available Fmoc-N-L-Asn(Trt)-ClTrityl resin (2·67 g, 1·60 mmol). The resin-bound peptide was suspended in 3 ml CH2Cl2, 1 ml trifluoroethanol and 1 ml acetic acid for 3 h, filtered and washed well with CH2Cl2. The combined washings were evaporated and then repeatedly redissolved and evaporated from CH2Cl2/hexane to give the crude, fully protected peptide, N-decanoyl-L-trp-L-Asn(Trt)-OH, as a cream solid (1·1 g, 96 %). A portion of this peptide (0·50 g, 0·69 mmol) was dissolved in 6 ml dry tetrahydrofuran and 1,3-dicyclohexylcarbodiimide (DCC) (144 mg, 0·69 mmol) with pentafluorophenol (128 mg, 0·69 mmol). Hexane (6 ml) was added after 2 h. The reaction was filtered and the filtrate was evaporated to give the crude product (0·60 g, 97 %) [MS m/z 881 (MH+), 903 (MNa+)].
Preparation of N-decanoyl-L-trp-D-Asn(Trt)-OPfp.
N-Fmoc-D-Asn(Trt)-OH (4·04 g, 6·77 mmol) was suspended in 37 ml CH2Cl2 and dissolved upon addition of N,N-diisopropylethylamine (4·7 ml, 27·04 mmol). 2-Chlorotrityl resin (3·7 g, 1·4 mmol g–1) was added to this solution and the mixture was agitated for 2 h. The resin was collected and washed with excess CH2Cl2, DMF and again with CH2Cl2. The resin loading was determined (Bernhardt et al., 1997) to be 0·6 mmol g–1. Using this resin (670 mg, 0·40 mmol), the acylated N-decanoyl-L-trp-D-Asn(Trt)-ClTrityl resin-bound peptide was made using standard deprotection and coupling methods, cleaved and isolated as described for the L peptide, to give the crude, fully protected peptide N-decanoyl-L-trp-D-Asn(Trt)-OH as a cream solid (227 mg, 79 %). This peptide (71 mg, 0·11 mmol) was dissolved in 3 ml dry tetrahydrofuran and DCC (21 mg, 0·10 mmol) with pentafluorophenol (20 mg, 0·11 mmol). After 2 h, 3 ml hexane was added and the crude product, collected as above, was used without further purification.
Preparation of L-Asn daptomycin (compound 5a; Fig. 6c).
n-Decanoyl-L-trp-L-Asn(Trt)-OPfp (0·085 g, 0·096 mmol) in 1 ml dry DMF was added to compound 4 (0·1 g, 0·08 mmol). The mixture was stirred at room temperature for 18 h before being evaporated to dryness. The residue was triturated with 5 ml diethylether to give Orn-Alloc-Asn(Trt)-daptomycin as a yellow powder (0·144 g). This was dissolved in 0·5 M HCl (1 ml) and 3 ml 1,4-dioxane and 0·1 ml N-methylmorpholine was added, followed by 0·1 g tetrakis-(triphenylphosphine)palladium. The mixture was stirred at room temperature for 24 h under argon and then filtered and concentrated to give a semi-dry solid of crude Asn(Trt)-protected daptomycin (0·4 g). This compound (0·2 g) and 0·1 ml triisopropylsilane were stirred in 25 % TFA in 4 ml dry CH2Cl2 at room temperature for 2 h before being evaporated to dryness. The residue was purified by preparative HPLC with a 250x21·2 mm IBSIL 5 µ C8 column eluted with 20–60 % acetonitrile in 0·5 % aqueous ammonium hydrogen phosphate buffer. After evaporation of acetonitrile from the collected fractions, the solution was loaded onto Bondesil 40 µM C8 resin (1 g) (prewashed with 10 ml each methanol and water) and washed with 10 ml water. The product was eluted with 20 ml methanol and evaporated to dryness to give daptomycin stereoisomeric compound 5a as a yellow solid (1·0 mg) [MS m/z 1621 (MH+), 811 ([M+2H]2+)].
Preparation of D-Asn daptomycin (compound 5b; Fig. 6c).
Treatment of compound 4 (50 mg, 0·04 mmol) as described above for L-Asn daptomycin, except that N-decanoyl-L-trp-D-asn(trt)-opfp was used for n-decanoyl-L-trp-L-Asn(Trt)-OPfp, gave the desired compound 5b as a yellow solid (1·0 mg) [MS m/z 1621 (MH+)].
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
) that allows maintenance of the library in E. coli as well as site-specific integration in streptomycetes at the 
C31 attachment site in the bacterial chromosome (Kuhstoss & Rao, 1991) (Fig. 1b). Although a cosmid library was constructed previously (McHenney et al., 1998), the theoretical minimum size of the biosynthetic pathway for A21978C of about 45 kb indicated that a BAC system enhanced the probability of cloning an intact pathway that could be validated by heterologous expression. The calculation was based on the estimate for the NRPS genes: 3·2 kb for each L-amino acid-incorporating module, 4·6 kb for the two known D-amino acid-incorporating modules and 0·8 kb for a chain-terminating domain (see below). Approximately 2000 BAC clones were collected into a library where 98 % of clones had inserts, as demonstrated by restriction digestion of 44 randomly sampled clones, where the mean size of the inserts was 71·4±14·7 kb. The library was screened by PCR with primers corresponding to known portions of the daptomycin NRPS genes (McHenney et al., 1998), and six positive clones with inserts ranging from 46 to 128 kb were identified (Fig. 1c).
Heterologous expression of the dpt cluster
Five of the six BAC clones were introduced into S. lividans TK64 by protoplast transformation and replicate apramycin-resistant transformants were analysed for production of daptomycin. A disc diffusion test for biological activity in crude broths showed that only replicate transformants with pCV1, a clone carrying a 128 kb insert (Fig. 1d), inhibited growth of Staphylococcus aureus. For chemical analysis, transformants were fermented in 250 ml shake flasks and, after 2 days, three compounds eluting as a cluster of peaks from 5·5 to 5·9 min were identifiable as the native S. roseosporus lipopeptides A21978C1–3 (Debono et al., 1987) by their HPLC retention times, UV/visible spectra (all exhibited the same spectrum: 
max 224, 262 and 365 nm) (Fig. 2a, b) and ESI MS molecular ions (MH+) at m/z of 1634·7, 1648·7 and 1662·7 for only the strains carrying pCV1. Southern analyses indicated that the entire pCV1 plasmid had been integrated into the chromosome of S. lividans (Fig. 2c). The combined titre under these conditions was typically 15–20 mg A21978C1–3 l–1.
|
). Chromatographic analysis demonstrated the presence of daptomycin based on its HPLC retention time (5·5 min) and UV absorption spectrum (max as for A21978C1–3), and ESI LC-MS analysis confirmed the expected molecular ion (MH+) as 1620·6. The fermentation was harvested after 115 h, when the titre was 18 mg daptomycin l–1 (about 2 % of the yield compared to decanoic acid-fed S. roseosporus fermented under similar conditions). Daptomycin purified from the S. lividans CBUK 136742 fermentation broth was identical, by both 1H and 13C NMR spectroscopic analyses, to authentic daptomycin purified from S. roseosporus fermentations. These results indicated that pCV1 contains all the genetic information required for the production of A21978C factors and daptomycin in S. lividans.
NRPS genes
A 128 kb contig comprising the entire region cloned in pCV1 was assembled by sequencing pRHB159 and pRHB160, two cosmids previously determined to carry daptomycin biosynthetic genes (Fig. 1d) as well as portions of pCV1. Sequence analyses suggested the presence of up to 64 ORFs (Fig. 3), most of which had orthologues in S. coelicolor (Bentley et al., 2002) or S. avermitilis (Ikeda et al., 2003) or both (Table 1). The three largest, ORFs 42–44, contained typical NRPS motifs, and detailed analyses (below) supported the conclusion that these ORFs and their corresponding genes, dptA, dptBC and dptD, encoded the subunits of daptomycin synthetase. No other BAC clone tested included all three of these ORFs, consistent with the observation that only S. lividans transformants carrying pCV1 synthesized A21978C factors.
|
|
) (Kleinkauf & Von Dohren, 1996; Marahiel et al., 1997), delimiting 13 modules were found: five modules in DptA, six in DptBC and two in DptD (5 : 6 : 2 organization) (Fig. 3a). Previous biochemical studies (Wessels et al., 1996) also suggested that daptomycin NRPS was composed of three proteins, but the total size of 1·5 MDa determined by immunodetection (670, 630 and 250 kDa) was less than predicted for the required activities, and the proposed modular organization was 6 : 5 : 2. Gel resolution, protein stability, as well as accurate sizing of proteins >500 kDa could have been limitations, however, as the sizes of very large proteins have been underestimated in other instances; for example, initial biochemical evidence indicated a size of 1·4 MDa for cyclosporin synthetase, but this was revised to 1·7 MDa when the full gene sequence was obtained (Weber et al., 1994). The new information provided by sequencing the dpt genes clarified the sizes of the proteins and showed that the first five residues in daptomycin are associated with DptA and the next six residues with DptBC. All data were consistent for the smallest subunit and the 250 kDa protein, shown to activate L-Kyn by an ATP/[32P]pyrophosphate exchange assay (Wessels et al., 1996), clearly corresponded to the DptD subunit.
|
). Between these, domains for activation of specific amino acids (A-domains), attachment to the NRPS by a panthethienyl tether (T-domains) and C-domains were found in the expected number and order to comprise a total of 13 modules, one for incorporation of each amino acid. Domains responsible for epimerization (E-domains) in modules 8 and 11 for D-ala and D-Ser, respectively, were expected given the reported activation of L- but not D-forms of Ala and Ser in biochemical studies (Wessels et al., 1996). An additional E-domain in module 2 (Fig. 3a, Table 2) was inconsistent with the published stereochemistry of A21978C lipopeptides (Debono et al., 1987); this observation implied the presence of D-asn instead of L-Asn in position 2.
The specificity-conferring residues (‘amino-acid-binding pocket’) for the A-domains of the 13 modules were deduced by aligning the A4–A5 regions of the predicted dpt NRPS proteins with the corresponding portion of PheA, an A-domain of gramicidin synthetase for which a crystal structure complexed with its substrate, phenylalanine, has been determined (Stachelhaus et al., 1999). In most cases, the pocket residues were consistent with those suggested for other streptomycete NRPS (Challis et al., 2000; Stachelhaus et al., 1999). For example, the set of Thr4-activating pocket residues in DptA (DFWSVGMV) was nearly identical to those deduced from other pathways incorporating Thr, such as the actinomycin, pristinamycin, calcium-dependent antibiotic (CDA) (DFWNVGMV) and bleomycin (DFWSVGMI) biosynthetic pathways (Du et al., 2000). The residues for the Orn6-activating module (DTWDMGYV) in DptBC differed from those of other known Orn-activating modules; however, the latter were from Bacillus species, so differences may be attributable to the source organisms. Module 13, responsible for incorporating Kyn13, an amino acid so far known only in daptomycin, had a new pocket code (DAWTTTGV). Comparison of the full A4–A5 regions from dpt A-domains with those from other NRPS provide further information. The A-domains for modules 1 and 13 are related, even though the pocket code, per se, for the Trp1 module is quite different (DVSSIGAV) from that for Kyn13 (Fig. 4a). Isolated DptD protein activates Trp as well as Kyn in vitro (Wessels et al., 1996), but reduced efficiency for incorporation of the former (26·6 % of maximum) suggests that module 13 does not primarily incorporate Trp (followed by post-synthetic modification for production of A21978C), but rather incorporates Kyn or another intermediate. The grouping of the 3mGlu-activating A-domains from dpt and the CDA NRPS with those for Asn and Asp is consistent with an evolutionary commonality for modules that activate acidic amino acids and their derivatives.
|
). CI, the most common class, follows L-amino acid-activating modules; cii follows donor modules associated with D-amino acids. While C-domains are known to have specificity to their acceptor A-domains (Doekel & Marahiel, 2000), these comparisons suggest that CII domains have additional features, distributed over the domain, that influence association with certain upstream donor A-T- or A-T-E-domains. Correlation of C-domains for upstream E-domains was recently demonstrated in tyrocidine synthetase (Clugston et al., 2003). The C-domain in module DptBC1, responsible for condensation between the N-terminal pentapeptide ending in Gly5 and Orn6 (module DptBC1 in Fig. 3), was notable as it groups with the CII domains, even though it is not downstream of an E-domain. An alternative consideration is that CII domains may either follow E-domains or be subunit starters. The CI and CII categories are supported when other C-domains, such as those from the S. coelicolor CDA NRPS gene cluster (Hojati et al., 2002), are included in the comparison (Fig. 4b; unpublished results), but more examples of the DptBC1-type subunit starter CII domain are needed to explore any structural implications in this connection. A third category of C-domains (CIII), represented at the start of DptA, may exemplify a type that interacts with an acyl rather than a peptidyl substrate. Distinct starter C-domains are also encoded at the start of the first CDA NRPS gene and in lipopeptide-encoding genes in Bacillus species, where they are responsible for coupling a fatty acid to the peptide (Mootz & Marahiel, 1997). CIII domains may be heterogeneous among themselves; the peptide starter C-domain in the CDA pathway was different from that in DptA, perhaps reflecting physical accommodations for different fatty acids or interactions with specific carrier proteins that must dock and transfer activated acyl-thioesters to begin building the lipopeptide.
Each module in the daptomycin NRPS also has a requisite T-domain, recognized by the conserved Ser needed for post-translational attachment of a 4'-phosphopanthetheine group in the holoprotein (Schlumbohm et al., 1991). An LGGDS, rather than the more common LGGHS, motif was found in T-domains in the modules associated with the incorporation of D-amino acids. This signature has functional significance in the interaction between E- and T- domains (Linne et al., 2001). In this regard, it is noteworthy that the T-domain in the module (DptA5) that incorporates Gly5 had an LGGDS motif, in contrast to that for incorporating Gly10 (DptBC6), which had an LGGHS sequence. As noted above, the C-domain at the DptA/DptBC junction (from module DptBC1) is of the CII type. Here, instead of following an E-domain directly, the CII-domain follows a T-domain carrying a motif that has an association with E-domains.
The DptD subunit, comprising modules responsible for incorporating 3mGlu12 and Kyn13, is terminated by a Te-domain typically required for release of the peptide from the synthetase. Te-domains use a catalytic triad that includes a Ser residue (GXSXG) in a conserved three-dimensional configuration, relative to a His and an acidic residue. Macrocyclization to form the lariat structure of daptomycin might occur by transfer of the linear peptide from the last T-domain (in the Kyn module) to the conserved Ser in the Te, forming an acyl-O-Te intermediate. Crystal structure studies (Bruner et al., 2002) of the Te from surfactin synthetase, also responsible for cyclizing an acylated peptide, revealed a lidded bowl-shaped hydrophobic cavity that could accommodate folding of the acylpeptide intermediate into a conformation productive for intramolecular nucleophilic attack by the hydroxyl group from Thr4 (if in the case of daptomycin) and allow formation of the ring-closing bond and release of the mature peptide from the synthetase (Kohli & Walsh, 2003). The structure might also facilitate exclusion of water, a competing reaction that could lead to release of a linear peptide.
Repeated DNA within dptBC
The high level of functional conservation among the modules of an NRPS predicts a commensurate degree of amino acid and DNA sequence conservation. However, there are several unexpectedly large stretches of sequence identity in dptBC, interspersed by small regions with less conservation. The A- and T-domains of modules for Asp7 and Asp9 are very similar, as are the segments comprising the end of the A-domains, the T-domains and start of the E-domains of modules for Ala8 and Ser11 (Fig. 5a). Within these stretches is an approximately 0·4 kb region, between A-domain motifs A6 and A10 (Marahiel et al., 1997), that is distinctive in modules for Asp7, Ala8, Asp9, Gly10 and Ser11, but not for Orn6. Centred here is a segment of DNA including a 137 bp invariant segment, followed by a 58–64 bp variable region and then another 71 bp invariant segment (an ‘IVI’ box). Despite the large tracts of identity between the Asp7 and Asp9 modules, not only are their IVI box regions distinct from each other, but they are instead identical to those of nearby modules. The IVI box sequence of Asp7 (‘type 7’) is found not in Asp9, but in modules Ala8 and Gly10, while the only other occurrence of the IVI box sequence of module Asp9 (‘type 9’) is in module Ser11. The sequence similarity extends somewhat beyond the IVI boxes of these modules, and the distribution is suggestive of some relation among them. Further study may address whether these repeats reflect some aspect of the origin of the dptBC gene, perhaps by duplications of a smaller ancestral gene in an evolutionary sequence towards a multimodular NRPS system.
|
) by insertional inactivation of dptE (described previously as a region with similarity to masC of Mycobacterium leprae) is consistent with a role for this gene in daptomycin biosynthesis. The dptE protein may be one of the 60 kDa proteins found by immunodetection using antibodies to the conserved phosphate-binding loop sequence (SGTTGRPKG) common to many adenylate-forming enzymes (Saraste et al., 1990) in biochemical studies of daptomycin synthetase (Wessels et al., 1996).
The role of pathway-associated acyl CoA synthetases as well as accompanying carrier proteins (Marshall et al., 2002) in determination of specificity and the biochemistry of their interaction with the starter C-domain for the NRPS are unclear, but the relative abundance and variety of acyl groups can be altered during fermentation of A21978C by fatty acid supplementations (Huber et al., 1988). When the dpt pathway was expressed in S. lividans, the fatty acids incorporated into A21978C factors remain those typical of S. roseosporus, and no A21978C factor carrying the 2,3-epoxyhexanoyl side chain of CDA, which was produced concurrently by S. lividans, was observed. Acylation of CDA with the shorter fatty acid may occur by a different route, however, as no dptE or dptF equivalent is associated with CDA NRPS. Further experiments to clarify the mechanism of acylation by dptE and dptF proteins, their specificity for a specific starter C-domain and the nature of their response to perturbation of the intracellular fatty acid pool during fermentation can be performed with the availability of the cloned dpt genes.
Genes for accessory biosynthetic capabilities
Other genes likely to participate in biosynthesis of A21978C reside downstream of those for the NRPS. The dptG gene belongs to a group of conserved ORFs associated with NRPS-derived siderophore and other secondary metabolite biosynthetic pathways in E. coli and actinomycetes. In S. coelicolor, SCO0489 is associated with coelichelin biosynthetic genes (Challis & Ravel, 2000) and SCO3218 is found in the CDA biosynthetic cluster (Hojati et al., 2002). In Streptomyces tendae Tü901, the small, approximately 70 amino acid ORF is incorporated as the first domain in an NRPS module (Lauer et al., 2001). Comparative domain analyses suggest involvement of dptG in regulation of expression or export of siderophores or antibiotics (Yeats et al., 2003).
The hydrolase-like protein encoded by the dptH gene bears the active site GXSXG motif of known thioesterases. Free thioesterases (Te II) that are not components of multidomain proteins (e.g. Te in DptD), but whose genes are found in biosynthetic gene clusters, participate in the synthesis of peptide and polyketide secondary metabolites (Heathcote et al., 2001). Disruption of Te II in the tylosin biosynthetic gene cluster in Streptomyces fradiae reduced antibiotic production by >85 % (Butler et al., 1999); similar observations were made for a Te II associated with pikromycin biosynthesis in Streptomyces venuezuelae (Kim et al., 2002) and for an external Te associated with surfactin biosynthesis in Bacillus subtilis (Schneider & Marahiel, 1998). Te II enzymes may be responsible for removal of misprimed substrates from T- or ACP domains; dptH thus may enhance the efficiency of daptomycin production by clearing misincorporated substrates that block the pathway.
The ORF designated dptI has similarity to the glmT gene in the CDA pathway (Hojati et al., 2002). The latter is proposed to encode a methyltransferase required for C3-methylation of Glu, prior to its activation by the cognate A-domain in the CDA NRPS. A pre-synthetic role of this enzyme was argued on the observation that both Glu- and 3mGlu-containing factors in the CDA family have been isolated and that the A-domain associated with 3mGlu was distinct from others associated with incorporation only of Glu. It is interesting to note that, for another Glu-/3mGlu-containing 13 amino acid cyclic lipodepsipeptide, A54145 produced by S. fradiae, there is a temporal shift toward 3mGlu-containing products over the course of fermentation; a rate-limiting post-peptide-modification enzyme was one hypothesis proposed to explain the observation (Baltz et al., 1997). No Glu-containing A21978C factor has been reported to date, but, given the resemblance between dptI and glmT and the similarity between daptomycin and CDA and between their respective 3mGlu-activating A-domains, one can suppose that the mechanism for elaborating the 3mGlu moiety might be similar among the three pathways.
Comparison with known sequences indicates that the next gene, dptJ, encodes a tryptophan 2,3-dioxygenase (TDO). The apparent translation start of this gene overlaps the stop of dptI and could allow for co-regulation during daptomycin biosynthesis. TDO, the first enzyme in the degradation of tryptophan, has been purified from Streptomyces parvulus, where its expression is correlated with actinomycin production (Hitchcock & Katz, 1988). Putative genes for TDO occur in both S. coelicolor (SCO3646) and S. avermitilis (SAV4526), but while the deduced SCO and SAV proteins are similar (82 % identity), and both genes are flanked by a gene probably encoding kynureninase, an enzyme downstream in the tryptophan catabolic pathway, the dptJ sequence is divergent (29 % identity with SCO3646) and there is no similarity in the neighbouring ORFs. The divergence is unlikely to be attributable to evolutionary distance between organisms: instead, dptJ may encode a paralogous enzyme, a primary metabolism function duplicated in S. roseosporus and diverged for A21978C biosynthesis. Extra copies of genes for primary metabolic functions that are located in and co-regulated with secondary metabolism are known; trpC2, trpD2 and trpE2 in the CDA gene cluster (Hojati et al., 2002; Huang et al., 2001) are paralogues (Bentley et al., 2002). While the genes involved in primary metabolism are conserved between species, the paralogous copy in the genome appears diverged. For example, the S. coelicolor TrpC1 protein (SCO2039) is more similar in size and sequence to its S. avermitilis orthologue (SAV6175) than it is to TrpC2 (SCO3211), associated with the CDA pathway. The 70 % G+C content and codon usage in dptJ are consistent with other streptomycete genes, but the proteins most similar to dptJ are found not in other actinomycetes, but rather in other Gram-positive or proteobacteria, thus providing another potential origin of dptJ.
The immediate product of TDO, N-formyl-L-kynurenine, must be converted by kynurenine formamidase (KF) to produce Kyn. Two isozymes for KF have been purified in S. parvulus (Brown et al., 1986). The higher molecular mass form, KFI (42 000 kDa), was constitutive, but expression of the lower molecular mass form, KFII (25 000 kDa), appeared correlated with actinomycin biosynthesis. Pathway-coordinated expression of KF would be advantageous for daptomycin production, but further experiments are required to determine whether one of the unassigned ORFs in the sequenced dpt region encodes this function. It should be noted, however, that only KFI was isolated from another actinomycin-producing organism, Streptomyces antibioticus. Gene disruption and expression studies of the sequences represented on pCV1 would be instructive as to whether dptJ and others are coordinately expressed with the NRPS genes, as reported for genes flanking the CDA NRPS genes in S. coelicolor (Huang et al., 2001).
Genes flanking the dpt region
In addition to genes immediately involved in the biosynthesis of daptomycin, there are many ORFs associated with multicomponent transporter systems. These use proton-dependent transmembrane electrochemical potential to import materials such as iron or other metals from the environment or to export antibiotics or other cellular products to confer self-resistance to toxic metabolites (Mendez & Salas, 1998). Some ORFs near the dpt genes may have such a role, as genes encoding transporters are often near or in secondary metabolite biosynthetic gene clusters in producer organisms. For example, ORFs 38 and 39, immediately upstream of dptE and transcribed in the same direction, bear similarity to genes clustered with the biosynthetic pathways for candicidin or friulimicin (Campelo & Gil, 2002; Heinzelmann et al., 2003). Alternatively or in addition, resistance may be served by one or more of a number of ORFs encoding hypothetical proteins. Regulatory genes are also expected in the gene cluster: ORF 15, a possible marR-like regulator (Martin & Rosner, 1995), and ORFs 51 and 52, which bear similarity to transcriptional activators such as brpA (Raibaud et al., 1991) or catabolite repressors such as deoR (Zeng & Saxild, 1999), respectively, are potential candidates.
Most S. roseosporus ORFs flanking the dpt genes had similarity to genes in S. coelicolor and S. avermitilis. An instance of conservation of relative gene order was observed for the group of ORFs from 4 to 20, inclusive, where consistently low E-values indicated a high level of confidence in the significance of the detected matches (Table 1). Orthologues to these ORFs are located in the subtelomeric region and lie in a conserved but inverted (SAV relative to SCO) region, designated ‘A’ (Ikeda et al., 2003), that is defined by SAV1418–1593 (positions 1750041–1957682) and SCO1010–1168 (positions 7602649–7438541). Subtelomeric regions do not have essential genes and are enriched for mobile elements, strain-specific genes and genes such as those related to some specific secondary metabolic pathways. The data for S. roseosporus are consistent with this pattern in that the dpt locus is located 400–500 kb from the end of the chromosome (McHenney et al., 1998) and that, overall, about 10 % of the ORFs in the sequenced region appear strain specific, with a series of ten (ORFs 27–36) bearing no significant similarity to the S. avermitilis genome. The notion that gene duplication may occur preferentially in subtelomeric regions of Streptomyces chromosomes, and that new genes and structural variability may emerge from the subtelomeric regions of linear bacterial chromosomes (Ikeda et al., 2003), provides an interesting background for future interpretations of the organization and features of the dpt cluster relating to the tracts of repeated DNA, for example, as well as to the assembly of associated biosynthetic functions encoded by paralogous genes.
Stereochemistry of Asn2
Although the structure of daptomycin was thought to contain only two D-amino acids (Debono et al., 1987), the sequence for the DptA subunit included an E-domain in module 2, suggesting that Asn2 would have a D-configuration. Two methods were used to support the assignment. In the first, the N-terminal lipohexapeptide tail comprising decanoyl-Trp1 to Orn6 was excised from daptomycin by alkaline hydrolysis of the depsipeptide bond and enzymic cleavage of the Orn6 to Asp7 amide bond (Fig. 6a), and compared with equivalent fragments containing either L-asn or D-Asn synthesized by standard Fmoc solid-phase peptide synthesis methods (Atherton & Sheppard, 1989; Chan & White, 2000; Fields & Noble, 1990) from commercially available Wang-Orn-Fmoc resin (Fig. 6b). Under HPLC conditions that separated the synthetic stereoisomers, the tail lipohexapeptide isolated from daptomycin was observed as a single peak. When this tail lipohexapeptide was co-injected with a synthetic lipohexapeptide containing D-Asn2, only a single peak was observed. When the tail lipohexapeptide was co-injected with a synthetic lipohexapeptide containing L-Asn2, two distinct peaks were observed. These data indicated that the tail peptide, and thus natural daptomycin, contain D-Asn at position 2.
The same conclusion was reached when L-asn2 or D-Asn2 versions of daptomycin were synthesized and assayed for bioactivity. An Orn-protected daptomycin derivative, from which the decanoyl moiety was removed enzymically (Boeck et al., 1988; Debono et al., 1987), was treated with successive Edman-type cleavages to remove the first two amino acids, Trp1 and Asn2 (Fig. 6c). The remaining undecapeptide core (compound 4 in Fig. 6c) was coupled with activated esters of N-decanoylated L-trp1-D-asn2 or L-trp1-L-asn2 dipeptides to generate protected daptomycin with D-asn2 or L-Asn2. After deprotection and purification, the semi-synthetic stereoisomers (compounds 5a and 5b in Fig. 6c) were evaluated for antibacterial activity against a panel of Gram-positive strains using microdilution assays. The MICs of the D-Asn isomer, at 0·78 µg ml–1 against Staphylococcus aureus (strain 42 and MRSA strain 499) and 6·25 µg ml–1
