Cylindrofridins AâC, Linear Cylindrocyclophane-Related

Article pubs.acs.org/jnp

Cylindrofridins A−C, Linear Cylindrocyclophane-Related Alkylresorcinols from the Cyanobacterium Cylindrospermum stagnale Michael Preisitsch,† Timo H. J. Niedermeyer,‡,§,⊥ Stefan E. Heiden,∥ Inga Neidhardt,†,# Jana Kumpfmüller,∥,◊ Martina Wurster,† Kirsten Harmrolfs,∇ Christoph Wiesner,□ Heike Enke,⊥ Rolf Müller,∇ and Sabine Mundt*,† †

Institute of Pharmacy, Department of Pharmaceutical Biology, Ernst-Moritz-Arndt-University, Friedrich-Ludwig-Jahn-Straße 17, 17489 Greifswald, Germany ‡ Interfaculty Institute of Microbiology and Infection Medicine, Eberhard Karls University, Auf der Morgenstelle 28, 72076 Tübingen, Germany § German Centre for Infection Research (DZIF), Partner Site Tübingen, Germany ⊥ Cyano Biotech GmbH, Magnusstraße 11, 12489 Berlin, Germany ∥ Institute of Pharmacy, Department of Pharmaceutical Biotechnology, Ernst-Moritz-Arndt-University, Felix-Hausdorff-Straße 3, 17489 Greifswald, Germany ∇ Helmholtz Institute for Pharmaceutical Research Saarland, Helmholtz Centre for Infection Research, and Department of Pharmaceutical Biotechnology, Saarland University, Campus E8.1, 66123 Saarbrücken, Germany □ Sealife PHARMA GmbH, Technopark 1/Obj.C/EG, 3430 Tulln, Austria S Supporting Information *

ABSTRACT: A rapid and exhaustive one-step biomass extraction as well as an enrichment and cleanup procedure has been developed for HPLC-UV detection and quantification of closely related [7.7]paracyclophanes and structural derivatives based on a two-phase solvent system. The procedure has been validated using the biomass of the carbamidocyclophane- and cylindrocyclophane-producing cyanobacterium Nostoc sp. CAVN2 and was utilized to perform a screening comprising 102 cyanobacterial strains. As a result, three new cylindrocyclophane-related alkylresorcinols, cylindrofridins A−C (1−3), and known cylindrocyclophanes (4− 6) were detected and isolated from Cylindrospermum stagnale PCC 7417. Structures of 1−3 were elucidated by a combination of 1D and 2D NMR experiments, HRMS, and ECD spectroscopy. Cylindrofridin A (1) is the first naturally occurring [7.7]paracyclophane-related monomeric derivative. In contrast, cylindrofridins B (2) and C (3) represent dimers related to 1. Due to chlorination at the alkyl carbon atom in 1−3, the site of [7.7]paracyclophane macrocycle formation, the cylindrofridins represent linearized congeners of the cylindrocyclophanes. Compounds 1−3 were not toxic against nontumorigenic HaCaT cells (IC50 values >25 μM) compared to the respective cylindrocyclophanes, but 1 was the only cylindrofridin showing moderate activity against methicillin-resistant Staphylococcus aureus (MRSA) and Streptococcus pneumoniae with MIC values of 9 and 17 μM, respectively.

C

approximately one-quarter of all marine cyanobacterial metabolites are derived from the order Nostocales, predominately from the genus Nostoc.4 The major structural classes found in Nostoc spp. are peptides, glycolipopeptides, and depsipeptides, often with unusual amino acids. Furthermore, alkaloids, fatty acids, terpenoids, derivatives of amino acids, and also polyketides such as [7.7]paracyclophanes have been structurally elucidated, the latter only from terrestrial and

yanobacteria are considered a prolific source of structurally diverse and biologically active natural products. Various compounds have been identified as new potent lead structures for the development of novel pharmaceuticals against cancer or other diseases.1−7 Of all secondary metabolites isolated from marine cyanobacteria, over half are biosynthesized from Oscillatoriales such as Moorea species and very often represent complex and bioactive lipopeptides. These compounds are derived from a combination of genes encoding for nonribosomal peptide synthetases (NRPSs) and polyketide synthases (PKSs).4,8−11 Nevertheless, © XXXX American Chemical Society and American Society of Pharmacognosy

Received: August 27, 2015

A

DOI: 10.1021/acs.jnatprod.5b00768 J. Nat. Prod. XXXX, XXX, XXX−XXX

Journal of Natural Products

Article

Figure 1. One-step biomass extraction as well as enrichment and cleanup procedure for the analysis of [7.7]paracyclophane-related compounds. (A) Overview of key steps from biomass to compound detection. (B) Validation results for the extraction and HPLC-UV analysis of the lower phase sample from Nostoc sp. CAVN2.

freshwater Nostoc and Cylindrospermum strains. [7.7]Paracyclophanes share a unique scaffold consisting of two resorcinol moieties, which are connected via two aliphatic chains and are divided into different subclasses including cylindrocyclophanes,12−14 nostocyclophanes,15 carbamidocyclophanes,16−18 and merocyclophanes.19 These cyanobacterial metabolites show a wide range of bioactivities, e.g., antimicrobial activity against Gram-positive bacteria such as methicillinresistant Staphylococcus aureus (MRSA), Streptococcus pneumoniae,18 Mycobacterium tuberculosis, and Enterococcus faecalis.17 Furthermore, proteasome-inhibiting effects and potent antiproliferative activities against various cancer cell lines have been reported. These activities are mostly independent of the individual substitution pattern of the compounds. Cytotoxicity against nontumorigenic cells is usually equally pronounced.13−21 However, isolation of pharacine, a p-cyclophane from the marine bacterium Cytophaga sp. AM13, indicated that the occurrence of cyclophanes is not restricted to the phylum Cyanobacteria.22 Recently, Nakamura et al. identified the cylindrocyclophane biosynthetic gene cluster from Cylindrospermum licheniforme UTEX ‘B 2014’ (ATCC 29412). The authors characterized most of the key enzymatic steps and discussed the biosynthesis of a putative, monomeric precursor that possibly forms the dimeric cylindrocyclophanes.23 Moreover, a homologous gene cluster can be found in the genome from Cylindrospermum stagnale PCC 7417 (also designated as Cylindrospermum sp. ATCC 29204, formerly known as C. licheniforme Kützing ATCC 29204),24 which was the very first reported producer of cylindrocyclophane A.12 The data of Nakamura et al.23 corroborate earlier results of feeding studies by Bobzin and Moore,25 which showed that [7.7]paracyclophanes are derived from a polyketide biosynthetic pathway. Interestingly, and in contrast to earlier proposals, cyclization of the two monomeric precursors possibly involves an unactivated carbon center

located within the recruited fatty acid that initiates the unusual cylindrocyclophane biosynthesis.23 The subsequent elongation and aromatization steps have also been attributed to different, partly uncommon-acting PKSs.26 However, the exact dimerization mechanism as well as the process and timing of halogenation of various [7.7]paracyclophane derivatives is still unclear. Due to their interesting bioactivities as well as challenging structures and biosynthesis, there is an ongoing interest in isolating new cyanobacterial cyclophanes. Here a rapid and convenient extraction method for the separation and enrichment of [7.7]paracyclophanes, as well as its application on 102 cyanobacterial strains, is described. The screening resulted in the isolation and characterization of three novel cylindrocyclophane-related compounds, named cylindrofridins A−C (1−3), along with three known cylindrocyclophanes (4−6) from the cyanobacterium C. stagnale PCC 7417. Their structures were determined using a combination of spectroscopic and spectrometric methods including HRMS, 1D and 2D NMR analyses, and ECD experiments. Furthermore, the antibacterial activities of 1−6 against selected Gram-positive and Gramnegative bacteria as well as the cytotoxicity against nontumorigenic HaCaT cells were evaluated.

■

RESULTS AND DISCUSSION Method Development and Validation. An exhaustive one-step biomass extraction and enrichment as well as cleanup procedure for effective detection and quantification of [7.7]paracyclophanes and related compounds via HPLC-UV-MS analysis (Figure 1) was developed. To evaluate the extraction efficiency, the well-characterized cylindrocyclophane- and carbamidocyclophane-producing cyanobacterium Nostoc sp. CAVN2 was used.18 On the basis of initial observations during the development of a liquid−liquid chromatography separation method for [7.7]paracyclophane isolation using the ARIZONA B



Article

Figure 2. Overview of screened cyanobacterial strains with available geographical origin information. Dots represent exact isolation locations. Circles represent approximate isolation locations based on available data. Screening numbers refer to Table S2.

solvent series,27 a biphasic solvent system consisting of nheptane, EtOAc, EtOH, and H2O has been optimized to the ratios 5:2:5:2 (v/v/v/v). This solvent system was usable for both the direct extraction and prepurification of all cyclophanerelated derivatives present in CAVN2. The extraction step was reduced to only 30 min, and the accumulation of compounds in the lower phase showed no significant difference up to 48 h (Figure S1). The total extract yielded 27.6 ± 2.0% of dry biomass, of which 23.2 ± 2.1% has been obtained from the lower phase and 4.4 ± 1.6% from the upper phase. This total extract amount was slightly higher than that obtained by using a previously reported extraction regime including five MeOH extraction steps for the analysis of [7.7]paracyclophanes in strain CAVN2 (23.5 ± 0.4%, n = 4) (Figure S2).18 In addition, the new protocol is statistically more effective, revealing a 9.9% higher average total [7.7]paracyclophane amount than the recently described extraction procedure using MeOH (Figure S3).21 Furthermore, the application of the novel biphasic solvent system to the one-step procedure revealed higher [7.7]paracyclophane yields compared to MeOH and other solvents or solvent mixtures, such as CH2Cl2, CH2Cl2/MeOH (1:1, v/v), CH2Cl2/IPA (1:1, v/v), or EtOH/H2O (7:3, v/v), that were also reported for [7.7]paracyclophane extraction (Figure S4).12−15 Although extraction with 70% EtOH or MeOH did not result in statistically lower yields, the major advantage of the novel solvent system is a quantitative accumulation of [7.7]paracyclophanes and related derivatives in the lower, more hydrophilic, phase (Figure 1). Pigments such as chlorophyll (measured as phytol species in GC-MS analysis) as well as fatty acids were highly enriched in the upper, more lipophilic, phase (Table S1), and thus were separated from the compounds of interest.

The method has been validated according to the guidelines of the Association of Official Analytical Chemists (AOAC).28 Validation includes HPLC-UV separation and quantification at 226 nm for compounds eluting in the retention time range of 5−25.2 min using carbamidocyclophane A as external reference substance. Due to a detected total [7.7]paracyclophane content of 5.3% in the biomass of strain CAVN2, obtained validation parameter values correspond well to acceptable limits or values, making this one-step procedure not only an effective tool for cyclophane-directed high-throughput biomass processing but also suitable for quantitative analysis of closely related congeners (Figure 1). Screening of Cyanobacteria. The developed method was utilized to perform a [7.7]paracyclophane-targeted screening including 102 cyanobacterial strains that belong to the orders Nostocales and Oscillatoriales. Strain selection was based on the knowledge about cyclophane-producing genera as previously reported in the literature.12,14,16−19,29 Thus, special focus was placed on terrestrial or freshwater Nostoc and Cylindrospermum species, but genera such as Anabaena, Nodularia, Pseudanabaena, and several so far unidentified cyanobacteria of the orders mentioned above were also included. The screened strains originated from a widespread range of habitats and geographic regions (Figure 2, Table S2). Notably, a phylogenetic analysis based on available partial 16S rRNA gene sequences revealed A. azotica strain Ley HB686 (accession number: AJ488603) as a promising candidate for cyclophane production, as it is a member of a clade including [7.7]paracyclophane-producing Nostoc sp. strains (Figure S5). Interestingly, a second partial 16S rRNA gene sequence (accession number: AY422691) of A. azotica strain FACHB118 showed only 94% sequence identity (92% query coverage, E-value: 0.0; Figure S6), but has also been reported to be strain C



Article

HB68630 and is maintained in the Culture Collection Yerseke (CCY) as laboratory culture A. azotica CCY 0405. Due to these contradictory results and a relatively short sequence length of FACHB-118, we redetermined the partial 16S rRNA gene sequence of CCY 0405. Our results showed a 2 bp difference from previous data of FACHB-118 but a complete sequence identity to Trichormus azollae Kom BAI/1983 (accession number: AJ630454),31 which is phylogenetically rather distinct from known [7.7]paracyclophane-producing cyanobacteria (see Figures S5 and S6). This result was congruent with our cyclophane-negative screening of that strain. Among the cyanobacteria screened for the production of new [7.7]paracyclophanes or related derivatives, only C. stagnale PCC 7417 revealed three compounds with carbamidocyclophane-like UV spectra18 in the range 200−400 nm and with corresponding monoisotopic [M − H]− ions at m/z 369.2, 661.4, and 703.4 in addition to known cylindrocyclophanes 4− 6. Due to our phylogenetic results of A. azotica and its tangled strain history, we repeated phylogenetic marker gene analysis of strain PCC 7417, such as for hetR, rbcLX, and cpcB-cpcA-IGS, to confirm its identity. Data analysis of obtained 381, 785, and 704 bp products revealed no differences from reported sequence data (Table S3).24 Structure Elucidation of Cylindrofridins. Cultured C. stagnale PCC 7417 biomass was extracted with the two-phase solvent system under scaled-up conditions. The lower-phase extract showed significant inhibitory activity against MRSA (MIC = 0.04−0.08 μg/mL) and S. pneumoniae (MIC = 0.2 μg/ mL) as well as cytotoxicity against HaCaT cells (IC50 = 2.8 ± 1.1 μg/mL). A portion of this extract (181 mg) was directly subjected to semipreparative reversed-phase HPLC. Multiple rounds of isolation yielded cylindrofridins A (1, 1.5 mg), B (2, 5.1 mg), and C (3, 3.9 mg) and three previously known [7.7]paracyclophanes, cylindrocyclophanes A (4, 5.3 mg), B (5, 5.4 mg), and C (6, 3.4 mg).

the spectrum, indicating that the chlorine could not be located at the end of the aliphatic chain. One of the methyl signals was a singlet shifted to δH 2.08, suggesting the presence of an acetate residue. The two other methyl signals were found as a triplet at δH 0.93 (indicating its position at the end of the aliphatic chain) and a doublet at δH 0.91 (corresponding to the branching methyl group found in all cylindrocyclophanes13,14). A H-multiplet at δH 3.87 with an integral corresponding to one proton and the corresponding carbon signal at δC 64.8 (CH from HSQC-DEPT) confirmed that the chlorine is located within the aliphatic chain. On the basis of COSY and HMBC spectra, the aliphatic system could be assembled. Despite poorly resolved signals in the HSQC spectrum, the proton resonances could unambiguously be assigned to the corresponding carbon atoms in the aliphatic chain based on the HMBC spectrum. The coupling pattern in the aromatic region complied with a 1,3,5-trisubstituted benzene ring. The three substructures could be connected by evaluating the signals in the HMBC spectrum; especially the proton doublet at δH 5.41 (H-11) proved to be of high indicative value, as it showed coupling to carbons in all three substructures. All chemical shifts are summarized in Table 1, and the key correlations in the two-dimensional NMR spectra are shown in Figure 3. HRMS data of 2 support the molecular formula C38H59ClO7. As its molecular formula is roughly twice that of 1, we first suspected it to be a dimer derivative of 1, with a structure similar to other cylindrocyclophanes. However, nine degrees of unsaturation did not fit this assumption, as this would require 10 DBE. Its 1H NMR and HSQC-DEPT spectra did resemble those of 1, featuring aromatic as well as numerous aliphatic protons. Five methyl group signals, one of them again an acetate singlet, were observed in the spectrum, confirming the assumption that a dimeric structure of 1 could be present. This was further strengthened by the observation of two oxymethine doublets at δH 4.14 and δH 5.40, suggesting the presence of free and esterified hydroxy groups, respectively. Interestingly, a triplet at δH 6.15 with an integral corresponding to one proton in conjunction with the absence of another signal corresponding to one aromatic proton indicated that one of the two aromatic systems still featured a 1,3,5-trisubstitution. Again, the presence of the chlorine in the side chain was clearly visible by comparison of the CH signal at δC/H 65.0/3.85 in the HSQCDEPT spectrum with C/H-5 of 1. As this chlorine is located in the position that is usually found to be involved in cylindrocyclophane macrocyclization, we suspected that ring closure could not be possible in this compound due to the presence of the chlorine. Indeed, detailed evaluation of the HMBC spectrum revealed the structure of 2 to be as depicted (Figure 3). Key correlations in the HMBC spectrum were found to be H-24/26 to C-15 and H-25 to C-14/16, confirming the connection of the two monomers via C-25 of the aliphatic side chain and the aromatic carbon, C-15, located ortho to the two phenolic hydroxy groups, as is commonly found in the [7.7]paracyclophanes. HMBC correlations also showed that the acetate moiety is not connected to the monomer also featuring the chlorine atom, but via the oxygen at C-31. HRMS of 3 showed an [M − H]− ion at m/z 703.4014, corresponding to the molecular formula C40H61ClO8. The 1H NMR and HSQC-DEPT spectra showed great similarities to the spectra of 2. An additional C2H2O as well as an additional DBE (10 in 3) suggested the presence of an additional acetate group compared to 2. Indeed, the 1H doublet at δH 4.14, corresponding to the proton located at the carbon with the free

HRMS data for 1 were consistent with the molecular formula C20H31ClO4. Five degrees of unsaturation suggested the presence of one aromatic ring system and an additional double bond. On the basis of the molecular formula and the doublebond equivalents (DBE), we suspected 1 to be a hemicylindrocyclophane D analogue13 and the chlorine to be located at the end of the aliphatic chain, as is common for chlorinated cylindrocyclophanes.14,16−18 Inspection of the 1H NMR spectrum of 1 in MeOH-d4 did indeed reveal close structural similarities of 1 to other cylindrocyclophanes. However, the presence of three methyl groups was clearly recognizable from D



Article

Table 1. 1H (500 MHz) and 13C (125 MHz) NMR Data for Cylindrofridins A−C (1−3) in MeOH-d4 1

2

3

δC, type

δH (J in Hz)

δC, type

δH (J in Hz)

δC, type

δH (J in Hz)

1 2 3a 3b 4

14.6, CH3 23.2, CH2 29.6, CH2

14.3, CH3 23.3, CH2 29.7, CH2

64.8, CH 39.4, CH2

7

27.3, CH2

8 9

27.1, CH2 33.6, CH2

0.92, 1.34, 1.37, 1.47, 1.61, 1.67, 3.85, 1.58, 1.64, 1.35, 1.41, 1.31, 1.00, 1.27, 1.71, 4.14,

14.3, CH3 23.3, CH2 29.7, CH2

5 6

0.93, 1.34, 1.38, 1.49, 1.64, 1.72, 3.86, 1.61, 1.68, 1.40, 1.45, 1.35, 1.09, 1.32, 1.89, 5.41,

0.91, 1.34, 1.38, 1.47, 1.61, 1.67, 3.85, 1.58, 1.64, 1.35, 1.42, 1.31, 0.99, 1.25, 1.86, 5.36,

position

10 11 12 13/17 14/16 15 18 19 20 21 22 23

39.4, CH2

39.4, 80.7, 143.3, 106.0, 159.5, 102.7, 15.3, 172.3, 20.9,

CH CH C CH C CH CH3 C CH3

t (7.6) m m m m m m m m m m m m m m d (6.4)

6.20, d (2.0) 6.16, t (2.2) 0.91, d (7.1)

39.3, CH2 65.0, CH 39.3, CH2 27.5, CH2 27.5, CH2 34.1, CH2 40.8, 79.4, 143.7, 106.9, 157.9, 117.6, 15.8,

CH CH C CH C C CH3

6.23, br s

0.95, d (6.6)

2.08, s 14.6, CH3 23.9, CH2 31.6, CH2

24

34.4, CH2

25 26

36.2, CH 34.4, CH2

27 28

29.3, CH2 28.1, CH2

29 30 31 32 33/37 34/36 35 38 39 40

t (7.4) m m m m m m m m m m m m m m d (6.8)

30.6, 39.6, 80.6, 143.6, 105.9, 159.3, 102.6, 15.2, 172.2, 21.0,

CH2 CH CH C CH C CH CH3 C CH3

hydroxy group in 2, was shifted to δH 5.36. Assignment of the signals in the HMBC spectrum was straightforward based on the structure of 2, and the additional acetate residue could as expected be connected at C-11. Thus, the structure of 3 was identified to be as shown. Configurational Analysis. The absolute configuration of all stereogenic centers of 4 was determined by Moore et al. using data acquired via Mosher’s method.13 On the basis of these results, the absolute configurations of 5 and 6 were determined by a comparison of their ECD data with those obtained for 4. According to the benzene sector rule,32 substituents at the benzene-adjacent α and β stereogenic carbons may significantly affect Cotton effects that arise from

0.82, 1.28, 1.06, 1.19, 1.49, 1.93, 3.18, 1.51, 1.97, 1.10, 1.15, 1.39, 1.29, 1.78, 5.40,

t (7.1) m m m m m m m m m m m m m d (6.0)

6.17, d (2.0) 6.15, t (2.0) 0.85, d (6.7) 2.06, s

39.3, CH2 65.0, CH 39.3, CH2 27.3, CH2 27.3, CH2 34.0, CH2 39.3, 81.0, 139.1, 106.2, 159.3, 180.2, 15.6, 172.1, 20.8, 14.5, 23.8, 31.6,

CH CH C CH C C CH3 C CH3 CH3 CH2 CH2

34.2, CH2 36.2, CH 34.2, CH2 29.3, CH2 28.1, CH2 30.5, 39.6, 80.7, 143.5, 106.0, 159.3, 102.5, 15.2, 172.5, 20.9,

CH2 CH CH C CH C CH CH3 C CH3

m m m m m m m m m m m m m m m d (6.9)

6.19, br s

0.92, m 2.06, 0.81, 1.28, 1.07, 1.17, 1.47, 1.90, 3.18, 1.51, 1.94, 1.10, 1.15, 1.38, 1.30, 1.78, 5.40,

s t (7.1) m m m m m m m m m m m m m d (6.0)

6.17, d (1.9) 6.15, t (1.9) 0.84, d (6.6) 2.07, s

electronic transitions of the benzene chromophore. These effects were used to deduce configurational information not only on cylindrocyclophane derivatives but also of other [7.7]paracyclophanes, comparing ECD spectra in the ranges 215−240 and 265−285 nm.15,17−19 In contrast to nonacetylated 4, compounds 5 and 6 displayed ECD spectra with a positive Cotton effect at 231 nm. Because deacetylation of 5 and 6 resulted in identical ECD spectra as observed for 4, the absolute stereoconfigurations of 4−6 were concluded to be identical.13 Moreover, stereocenters that are more distant from the benzene moiety have no impact on the above-mentioned Cotton effects.19,32,33 The ECD data of 4 do not significantly differ from those obtained for the nostocyclophanes, which E



Article

cluster found in C. stagnale PCC 7417,24 which shares 97% identity at the nucleotide level (accession number: NC_019757; range: 2 221 058 to 2 193 183; 89% query coverage, E value: 0.0) with that reported for the cylindrocyclophanes D−F producing strain C. licheniforme UTEX ‘B 2014’ (gene cluster accession number JX477167).13,23 This small difference between both gene clusters might be a key factor for the natural coproduction of cylindrofridins and cylindrocyclophanes. According to the experimental data and the putative biosynthetic route, we postulate that the absolute configurations of 2 and 3 at C-11 and C-31 are “R”, while C-10, C-25, and C-30 are “S”. As the ECD spectrum of 1 closely resembled those of 2 and 3 (positive Cotton effect at 226 nm, Δε = 1.25; negative Cotton effect at 273.5 nm, Δε = −0.63), 1 should have the same absolute configurations at C-10 and C-11 as 2 and 3, respectively (Figures S12, S18, and S24). Due to the spatial distance from the benzene ring to the chlorine-bearing stereogenic center, the absolute configuration of C-5 could not be determined based on the available data. Attempts to crystallize the amorphous compounds for X-ray crystallography were not successful. Biological Evaluation. Compounds 1−6 were evaluated for antibacterial activities and cytotoxicity against a set of various clinically relevant pathogens as well as HaCaT cells (Table 2), as recently carried out for other [7.7]paracyclophanes.18 The cylindrocyclophanes 4−6 showed strong antibacterial effects against MRSA and against S. pneumoniae (MICs in the range 0.1−2.4 μM) as well as cytotoxic activity (IC50’s in the range 2.9−11 μM) similar to reported results of carbamidocyclophanes A−F and H−L and cylindrocyclophanes A1−A4.18 Notably, only 1, but not 2 and 3, showed activity against MRSA and S. pneumoniae (MICs in the range 8.6−17 μM), although its activity is remarkably reduced compared to the [7.7]paracyclophanes. All three novel compounds exhibited decreased cytotoxicity. Especially cylindrofridin A (1) showed up to 35-fold reduction in the toxicity against HaCaT cells compared with the cylindrocyclophanes. This low cytotoxicity of 1 might be attributed to the single resorcinol moiety. Activity is enhanced as soon as two such motifs are present, as seen in 2 and 3, and even more if these groups are embedded in the robust [7.7]paracyclophane framework as identified for 4 by Yamakoshi et al.20 The “open-chained” dimeric structure of 2 and 3, however, seems to prevent any antibacterial activity against the pathogens tested in this study.

Figure 3. Structures of 1 and 2. Bold bonds indicate COSY correlations. Arrows indicate HMBC correlations. Chemical shifts are summarized in Table 1.

contain chlorine atoms at positions equivalent to C-3 and C-16, and lack the C-2/C-15 methyls. In addition, stereogenic centers bearing the cylindro- and carbamidocyclophane-typical branched methyl groups do no not contribute significantly to the ECD spectra of these molecules. The absolute configurations of these stereocenters are commonly achieved indirectly by determination of their relative configurations using the 3JH‑α,H‑β coupling constants.12,14,17,18 Comparative ECD spectroscopy in the range 200−300 nm revealed positive Cotton effects at 225 nm (Δε = 2.23) for 2 and 228.5 nm (Δε = 4.72) for 3 and negative Cotton effects at 277.5 nm (Δε = −1.11) for 2 and 278 nm (Δε = −1.02) for 3. The spectra were very similar to those of the corresponding cylindrocyclophanes 5 and 6. In contrast to the rigid paracyclophane structure, the monoalkyl resorcinol scaffold displays a greater conformational flexibility of the adjacent alkyl stereocenters. Thus, the 3JH‑11,H‑10 coupling constants cannot be used to deduce the relative configurations. Additional experiments on the conformational analysis of these atoms are therefore needed. However, the presence of both the linear cylindrofridins and the macrocyclic dimerized cylindrocyclophanes in strain PCC 741713 as well as their structural similarity suggests that these metabolites are derived from the same biosynthetic pathway. Therefore, the absolute configurations at the benzene-adjacent α and β carbons are most likely identical. This is also corroborated by the cylindrocyclophane gene

Table 2. Antibacterial and Cytotoxic Activities of Compounds 1−6 antimicrobial susceptibility testing Gram-positivea

Gram-negative

cytotoxicity testing

MRSA

S. pneumoniae

E. coli

K. pneumoniae

P. aeruginosa

HaCaTb

compound

MIC (μM)

MIC (μM)

MIC (μM)

MIC (μM)

MIC (μM)

IC50 (μM)

1 2 3 4c 5 6

8.6 >75 >71 0.1−1.1 0.1 0.9

17 >75 >71 0.3−2.2 0.3 2.4

>135 >75 >71 >85 >80 >75

>135 >75 >71 >85 >80 >75

>135 >75 >71 >85 >80 >75

100 25 25 5.0 2.9 11

± ± ± ± ± ±

20 2 3 1.9 1.0 1

Positive control: vancomycin (MIC = 1.4 μM) and fusidic acid (MIC = 3.8 μM). bPositive control: mitoxantrone (IC50 = 3.9 μM). cSimilar antimicrobial and cytotoxicity data have previously been reported by Preisitsch et al.18

a

F



Article

in addition to known cylindrocyclophanes in C. stagnale PCC 7417. Consequently, this method may be a supporting tool not only for the analysis of [7.7]paracyclophanes but also for structurally related derivatives, such as alkylresorcinols. In contrast to 2 and 3, 1 shows antimicrobial activity against Gram-positive bacteria including MRSA with a pronounced reduction of cytotoxicity, which is commonly associated with [7.7]paracyclophanes. Research regarding the identification of further derivatives to increase the understanding of the natural role of cylindrofridins is in progress. As indicated by the most recent publications on biosynthesis research of halogenated natural products, including cylindrocyclophanes,23,34−36 the remaining enzymatic key steps, such as C−C bond formation and halogenation, are under active investigation. Efforts to discover their biosynthetic pathways will expand the knowledge of the complex cylindrofridin as well as [7.7]paracyclophane assembly.

While the cylindrofridins do not appear to exhibit strong bioactivity, these compounds provide potential insights into the key assembly steps of [7.7]paracyclophane biosynthesis in cyanobacteria, such as macrocyclization or halogenation. Nakamura et al.23 showed that the cylindrocyclophane biosynthetic enzymes could construct the nonacetylated and nonhalogenated resorcinol analogue of 1 as a potential monomeric precursor. Furthermore, the alkyl carbon, which is the site of monomer dimerization, may enter the pathway as an unfunctionalized carbon center. However, the putative compound could not be detected in culture extracts. This has been explained by a tight coupling of monomeric precursor generation and dimerization. On the basis of bioinformatic analyses,34 the macrocyclization event in cylindrocyclophane biosynthesis is hypothesized to involve a C−H functionalization step featured by cryptic halogenation. Whether chlorination at C-5 of 1−3 or a lower enzyme activity could have a negative impact on the dimerization step in C. stagnale PCC 7417 cannot be answered based on existing data. On the other hand, a direct oxidative, intermolecular formation of the C−C bond catalyzed by Rieske oxygenase homologues is discussed for the macrocyclization event as well.34 Specifically, the biosynthesis of 1−3 along with cylindrocyclophanes in strain PCC 7417 as well as the occurrence of a Rieske [2Fe-2S] iron−sulfur domain-containing open reading frame in the downstream region of the cylindrocyclophane gene cluster may strengthen this assumption. Interestingly, Nakamura et al. reported that the above-mentioned resorcinol analogue of 1 has been toxic to the cylindrocyclophane-producing cyanobacterium C. licheniforme UTEX ‘B 2014’ at low concentrations (

Cylindrofridins AâC, Linear Cylindrocyclophane-Related

Recommend Documents

Cylindrofridins AâC, Linear Cylindrocyclophane-Related