Discovery, isolation, heterologous expression and mode-of-action studies of the antibiotic polyketide tatiomicin from Amycolatopsis sp. DEM30355 |…

The actinomycete DEM30355 was isolated from a soil sample, collected from the El Tatio geyser field within an arid part of the Atacama Desert in Chile17. Strain DEM30355 was recovered in the genus Amycolatopsis, based on 16S rRNA analysis, forming a subgroup with Amycolatopsis vancoresmycina DSM 44592T and Amycolatopsis bullii SF27T (see ESI). The genus Amycolatopsis contains 94 species and four subspecies encompassing both extremophiles and producers of bioactive secondary metabolites, including the clinically used vancomycin and rifamycin antibiotics19,20. Preliminary bioactivity screening showed that extracts of Amycolatopsis sp. DEM30355 displayed promising antibiotic activity against B. subtilis, thus we decided to examine the genome of Amycolatopsis sp. DEM30355 for novel biosynthetic potential. Purified genomic DNA from Amycolatopsis sp. DEM30355 was analysed using both PacBio and Illumina sequencing technologies and genome assembly was performed using the combined datasets to give a 9.6Mb draft genome, in 13 contigs. The draft genome of Amycolatopsis sp. DEM30355 was examined using the secondary metabolite analysis software AntiSMASH 6.0.121. Of the 31 biosynthetic gene clusters (BGCs) detected, a PKS cluster was identified showing moderate overall similarity (81%) to that which encodes for rishirilides A and B22,23,24,25,26. These compounds are anthracenone polyketides, originally isolated from Streptomyces rishiriensis OFR-1056, with no reported antibiotic activity. Rishirilide B has been shown to be a moderately potent inhibitor of 2-macroglobulin, glutathione S-transferase and asparaginyl-tRNA synthetase, whilst little is known about the biological role of rishirilide A (Fig.1)18,27,28.

Top ()-Rishirilide A (1) and (+)-rishirilide B (2). Relative stereochemistry of ()-1 and absolute stereochemistry of (+)-2 shown. Bottom. Structure of ()-tatiomicin (3) as derived from NMR and SCXRD experiments. Key COSY (red) and HMBC (blue) correlations shown. Absolute stereochemistry as shown by both vibrational circular dichroism (VCD) and single-crystal X-ray diffraction (SCXRD) resonant scattering experiments. Structural differences of rishirilide A and tatiomicin are highlighted (magenta).

Further inspection of the BGC from Amycolatopsis sp. DEM30355 revealed a highly altered gene synteny (see ESI), compared to the rishirilide BGC, along with the presence of several new genes: one postulated to be involved in PKS biosynthesis (tatS1), two encoding methyltransferases (tatM1 and tatM2), two encoding cyclases (tatC4 and tatC5) and one cytochrome p450 oxidoreductase (tatO11) (Fig.2). Due to the significant variation in the genetic make-up of the BGC, we postulated that it may code for the production of an as yet undiscovered polyketide and as such we set about attempting to identify this molecule from the metabolome of Amycolatopsis sp. DEM30355.

Organization of the tatiomicin BGC. Genes coding for polyketide biosynthesis (red; tatS=starter unit biosynthesis, tatK=chain biosynthesis), polyketide modification (blue; tatO=oxidoreductases, tatC=cyclases, tatM=methyltransferases), regulation (yellow; tatR), transport (green, tatT) and others (grey; tatP=phosphorylase, black; genes not assigned to the tatiomicin BGC based on homology to the rishirilide BGC and proposed biosynthetic pathway)).

Preliminary analysis of the fermentation supernatant of Amycolatopsis sp. DEM30355 by HPLC-HRMS showed the presence of a large number of secondary metabolites, in keeping with the predicted number of BGCs, including a compound with activity against Gram-positive bacteria (MW of 402Da, m/z=403 [M+H]+, m/z=425 [M+Na]+, ()-tatiomicin (3)). Fermentation of Amycolatopsis sp. DEM30355, removal of the biomass, extraction of the supernatant and bioactivity guided fractionation by multiple chromatography steps resulted in a fraction which retained antimicrobial activity and contained two closely related compounds. HRMS analysis suggested that these compounds were stereoisomers of each other, the major compound showing m/z=425.1221 [M+Na]+ corresponding to a molecular formula of C21H22O8 for both molecules (see ESI).

Structural determination of the major component was initially performed by NMR, which provided the majority of molecular connectivity with the exception of the ordering of the three contiguous quaternary centres at the C-3, C-4 and C-4a positions. Structural assignment was completed via single-crystal X-ray diffraction (XRD) analysis, revealing a highly oxygenated anthracenone polyketide, structurally consistent with the BGC of interest, which we named ()-tatiomicin (3) (Fig.1)29.

NMR and HPLC experiments demonstrated that the minor compound was the C-2 epimer, capable of equilibrating with ()-(3) under acidic conditions (see ESI).

Determination of the absolute stereochemistry of ()-3 was undertaken in parallel via vibrational and electronic circular dichroism spectroscopies and additional single-crystal X-ray diffraction (SCXRD) experiments.

Absolute configuration determination by vibrational circular dichroism (VCD) was based on a comparison of experimental and computationally predicted spectra, taking into account the presence of two epimers of ()-3. Conformational analysis (see ESI), removal of redundant geometries and final optimization at the B3LYP/6311++G(d,p) level allowed Boltzmann-weighted VCD spectra for both epimers of ()-3 to be constructed. The final predicted spectrum was obtained by applying a 3:1 ratio to account for the experimentally analysed mixture of epimers. Numerical analysis was used to establish agreement between experiment and theory, the neighbourhood similarity values (IR=92.0, VCD=71.2, ESI=57.8) suggesting an absolute stereochemical assignment of (2S,3S,4R,4aR,10R) (Fig.3 and ESI)30. The assignment was supported through similar electronic circular dichroism (ECD) experiments; however, in this case correlation between experiment and prediction was weaker (see ESI).

Experimental IR (top) and VCD spectra (bottom) of ()-tatiomicin 3 (CDCl3) with predicted spectra obtained at the B3LYP/PCM/6311++G(d,p) level of theory. VCD: Solid line=(2R,3R,4S,4aS,10S), dashed line=(2S,3S,4R,4aR,10R). Spectra have been frequency scaled Black line (=0.987) to yield maximal similarity grey line between the computed and experimental VCD spectra.

A suitable, albeit small, single-crystal of tatiomicin (3) was grown via slow evaporation from a benzene solution. Due to the crystals dimensions, diffraction data were collected at beam line I19 at the Diamond Light Source using synchrotron radiation at standard operating wavelength (=0.6889), providing a data set of sufficient quality to allow for structural confirmation. ()-Tatiomicin (3) crystallized as an H-bonded dimer in the unit cell (Z=2) along with a single molecule of solvent (benzene). To validate the absolute stereochemical assignment a further single-crystal X-ray diffraction experiment was undertaken at I19, employing non-typical, longer wavelength synchrotron radiation (=1.4879) to enhance resonant scattering contributions (also known inappropriately as anomalous dispersion). The absolute-structure (Flack) parameter (0.05(6)) was insignificantly different from zero and with a small standard uncertainty, indicating the correct absolute configuration in the refined (2S,3S,4R,4aR,10R) structure (see ESI)29. Interestingly, following extensive stereochemical debate and several reported total syntheses, the absolute stereochemistry of the congeneric (+)-rishirilide B (2) was recently revised (2S,3S,4S), matching that of ()-(3) over the three common stereocentres, suggesting a similar biosynthetic pathway for both sets of natural products (Fig.4)31,32,33,34,35.

Displacement ellipsoid plot of the molecular structure of ()-tatiomicin (3), absolute stereochemistry as shown determined by resonant scatteringthe dimer molecular structure (Flack parameter=0.05(6)). Displacement ellipsoids shown at 50% probability level.

To verify that the BGC previously identified does indeed encode the biosynthetic pathway for tatiomicin (3), a high molecular-weight P1 artificial chromosome (PAC) library was obtained, consisting of 2,688 clones with an average insert size of 138kb which contained resistant markers for kanamycin (for E. coli) and thiostreptone (for S. coelicolor). The PAC library was screened by PCR, using four primer pairs for the putative BGC. A single PAC clone was identified with the required PCR profile, which was then transferred into E. coli strain ET12567/pR9604 (dam- dcm-), the plasmid was subsequently transferred into S. coelicolor M1152 via conjugation. Exconjugants containing the plasmid integrated on the chromosome were selected for resistance to thiostrepton. Ninety-six putatively identified exconjugants were arrayed into 24 well plates and screened for the production of tatiomicin (3) by TLC, with detection based on the characteristic fluorescence upon UV irradiation at 365nm. Based on these screening parameters, S. coelicolor M1152::tat was identified as a producer of tatiomicin (3) (see ESI).

Growth of S. coelicolor M1152::tat was examined on solid medium, the agar was extracted (EtOAc) and analysed by LCMS alongside similar fermentation extracts from both the parent strain M1152, Amycolatopsis sp. DEM30355 and a tatiomicin (3) standard. An LCMS peak corresponding to tatiomicin (3) was observed in the extract from S. coelicolor M1152::tat but was absent in that of the parent strain M1152 (Fig.6).

Tatiomicin (3) was subsequently isolated from the fermentation of S. coelicolor M1152::tat in liquid medium (GYMG), as demonstrated by HRMS ([M+H]+=403.1403), with a production level in the heterologous host estimated at 0.57mg/L, confirming the identity of the tatiomicin BGC (Fig.5).

Detection of tatiomicin from the fermentation of heterologous host S. coelicolor M1152::tat. Top) Extracted ion chromatogram (EIC) based on m/z=827.25. S. coelicolor M1152 (purple), S. coelicolor M1152::tat (blue), Amycolatopsis sp. DEM30355 (black) and tatiomicin standard (red). Bottom) MS spectrum of tatiomicin (3) purified from the heterologous host S. coelicolor M1152::tat.

Based on a comparison between the tatiomicin and rishirilide BGCs22,23,24,25,26 we propose the following biosynthetic pathway operates for the assembly of tatiomicin (3) (See ESI). The modular type I polyketide synthase TatS1 is likely responsible for the biosynthesis of the polyketide starter unit, cis-crotonyl-ACP, which is then elongated via the attachment of eight malonyl-CoA by minimal PKS enzymes TatK1, TatK2, and TatK3. TatC1, TatC2, TatC3 and TatO10 show close homology to rishirilide cyclases RslC1, RslC2, and RslC3 and C9-ketoreductase RslO10, respectively. Thus, TatC1 and TatO10 likely act together to form the A ring of tatiomicin (3), whilst TatC2 and TatC3 catalyse the formation of the B and C rings. Tailoring of the polyketide core likely involves oxidation of the C ring by TatO4, and installation of the C ring epoxide by flavin mononucleotide (FMN)-dependent monooxygenase TatO1 together with a putative flavin reductase, TatO2. Opening of the epoxide is proposed to be mediated by NADPH:acceptor oxidoreductase TatO5, followed by the key BaeyerVilliger oxidation/rearrangement controlled by TatO9 and finally reduction of the B ring ketone by ketoreductase TatO8.

Three additional tailoring enzymes are present in the BGC of tatiomicin (3) for which no homologues are present in that of rishirilide, TatO11, TatM1 and TatM2. TatO11 is a cytochrome p450 oxidoreductase, likely responsible for oxidation of the A ring to the hydroquinone form, followed by double methylation by the two methyl transferases TatM1 and TatM2 to yield the completed molecule (Fig.6).

Top) Proposed pathway for the biosynthesis of ()-tatiomicin (3) based on homology with the biosynthetic gene cluster for the rishirilides. Enzymes shown in red have no direct congener in the rishirilide BGC and their biosynthetic role is hypothesised, based on BLAST analysis. Bottom) comparison of the rishirilide and (-) tatiomicin gene cluter based on BLAST analysis. (red; tatS=starter unit biosynthesis, tatK or rslK=chain biosynthesis), polyketide modification (blue; tatO or rslO=oxidoreductases, tatC or rslO=cyclases), regulation (yellow; tatR or rslR), transport (green, tatT or rslT) and others (grey; tatP or rslP=phosphorylase; tatM=methyltransferases, black; genes not assigned to the tatiomicin BGC based on homology to the rishirilide BGC and proposed biosynthetic pathway)).

The enzymes TatC4 and TatC5, which are not present in the rishirilide cluster, encode for a dehydrogenase and a monooxygenase and are located in the centre of the biosynthetic gene cluster. The tatiomicin BGC contains all orthologous genes responsible for the synthesis of rishirilide. The function of these additional genes is therefore not immediate obvious and might be a result of evolutionary divergence.

()-Tatiomicin (3) showed no detectable antimicrobial activity (MIC>64g/mL) against ten Gram-negative bacteria and two eukaryotic microorganisms (Candida spp.) (see ESI). However, antibacterial activity was observed against a sub-set of Gram-positive bacteria (MIC=48g/mL), namely Staphylococcus and Streptococcus species. Due to the interest in developing new antibiotics against drug-resistant Staphylococcus infections, we further evaluated ()-3 against a panel of MRSA clinical isolates, including twenty-four EMRSA-15 and EMRSA-16 strains (the main causative agents of nosocomial epidemic MRSA bacteraemia in the UK, with resistance to penicillin, ciprofloxacin and erythromycin)36, and twelve MRSA strains isolated from Belgian, Finnish, French and German hospitals (see SI). In all cases antibiotic activity was maintained (MIC=48g/mL), suggesting that ()-3 does not operate via a mode-of-action previously encountered by these strains, prompting us towards further investigation.

Elucidation of the mode-of-action (MOA) for a new antibacterial agent is a significant experimental challenge. The characterization of resistance mutations can be informative, however all attempts to isolate Bacillus subtilis mutants resistant to ()-tatiomicin (3) proved unsuccessful (see ESI). Also, no positive responses were seen with a panel of B. subtilis strains containing lacZ reporter genes used to indicate common antibacterial mechanisms of action, including: fatty acid synthesis (fabHA), DNA damage (105 prophage induction), RNA polymerase (RNAP) inhibition (helD), cell wall damage (ypuA), gyrase inhibition (gyrA), and cell envelope stress (liaI)) (see ESI)37,38,39.

Due to the presence of an, albeit electron-rich, ,-unsaturated carbonyl moiety, we postulated that the observed biological activity of ()-tatiomicin (3) may involve the covalent modification of thiol-containing enzymes through a conjugate or Michael addition of the thiol to the ,-unsaturated carbonyl. Thus, ()-tatiomicin (3) was reacted with L-cysteine hydrochloride, L-cysteine methyl ester hydrochloride and a short thiol-containing peptide (LcrV (271291)) as an enzyme proxy, under biologically relevant conditions. In all cases thiol adducts could be detected by LCMS, suggesting that ()-tatiomicin (3) may have biologically relevant Michael acceptor activity (see ESI).

To gain further insight into a potential mode-of-action, we undertook a bacterial cytological profiling experiment in which antibacterial induced changes in the morphology of test bacteria are compared to those induced by known mode-of-action antibacterials40,41. B. subtilis 168CA-CRW419 expresses two fusion proteins, HbsU-GFP and WALP23-mCherry, allowing simultaneous visualization of both the chromosomal DNA and the bacterial cell membrane by fluorescence microscopy. The cytoplasmic membrane was unaffected unlike in the control compound nisin, which forms large pores in the membrane42. Interestingly, treatment with ()-tatiomicin (3) induced chromosome decondensation in B. subtilis 168CA-CRW419, similar to the effects elicited by the RNAP inhibitor rifampicin (Fig.7).

Single-cell analysis of chromosome and membrane integrity. Phase contrast (top panels) and fluorescence microscopy of B. subtilis cells treated with various antibiotics (indicated above). DNA was visualized with an HsbU-GFP fusion (middle panels) and the cytoplasmic membrane with a WALP23-mCherry fusion (bottom panels).

The combination of the negative result observed with the helD reporter strain, cell lysis after prolonged incubation with the compound and the inability to create resistant mutants suggest that direct RNAP inhibition is unlikely. We therefore attempted to examine the integrity of the cytoplasmic membrane using the voltage sensitive dye DiSC3(5). This dye accumulates in well-energised cells in the cytoplasmic membrane15,43 but is released upon depolarisation of the membrane, and this release can be measured by fluorescence microscopy. DiSC3(5) is used in parallel with Sytox Green, a membrane-impermeable DNA stain used as a reporter for pore formation44. Upon addition of nisin, which forms large pores in the B. subtilis membrane42, both a loss of DiSC3(5) and uptake of Sytox Green was observed. In contrast gramicidin, which forms small cation-specific channels45, showed loss of DiSC3(5) without Sytox Green staining. Treatment with ()-tatiomicin (3) showed a similar effect to that of gramicidin, i.e. loss of DiSC3(5) without Sytox Green staining. Hence tatiomicin probably acts to dissipate the membrane potential without the formation of large pores (Fig.8).

Single-cell measurement of membrane potential and permeability. Phase contrast (top panels) and fluorescence microscopy of B. subtilis cells stained with the voltage-sensitive dye DiSC3(5) (middle panels) and the membrane permeability indicator Sytox Green (bottom panels) in the presence and absence of 32 g/mL of tatiomicin. As positive controls, the cells were treated with 5 g/mL of gramicidin (membrane depolarisation without pore formation) and 10 M nisin (membrane depolarisation through pore formation). Cellular DiSC3(5) and Sytox Green fluorescence values were quantified for cells treated with tatiomicin (32 g/mL), gramicidin (5 g/mL), and nisin (10 M) (see SI).

In an attempt to ascertain whether the observed loss of membrane potential is a downstream effect or occurs at the same time as chromosome depolarisation we performed a time-course experiment using DiSC3(5) in combination with a HsbU-GFP fusion to assess chromosome decondensation with images taken every two minutes. This showed that the loss of membrane potential occurred simultaneously with the chromosome decondensation, between 2 to 4min, suggesting that they are closely linked events (Fig.9).

Single-cell measurement of chromosome decondensation and membrane potential in a time course experiment in the presence of tatiomicin (32 g/mL). Phase contrast (top panels), fluorescence microscopy of B. subtilis HsbUGFP (chromosome marker) (middle panel) and stained with the voltage sensitive dye DiSC3(5) bottom panel. Cellular DiSC3(5) fluorescence values where quantified over time. The bar chart depicts the fluorescent intensity values of individual cells (> 30) (see SI).

Read more:
Discovery, isolation, heterologous expression and mode-of-action studies of the antibiotic polyketide tatiomicin from Amycolatopsis sp. DEM30355 |...