Discovery and total synthesis of doscadenamide A: a quorum sensing signaling molecule from a marine cyanobacterium

Abstract

Quorum sensing (QS) plays a critical role in the regulation of bacterial pathogenesis. Doscadenamide A (1a) was isolated from a marine cyanobacterium, its structure elucidated by NMR and activity linked to QS induction. The total synthesis of 1a was developed and the absolute configuration confirmed through comparison of the isolated natural product with synthetic diastereomers. Our preliminary investigation indicated that 1a could activate QS signaling in a LasR dependent manner.

Graphical Abstract

Quorum sensing (QS) is an intercellular communication process adopted by a number of bacteria to regulate diverse physiological activities. This process involves the production and release of diffusible extracellular signaling molecules named autoinducers (AIs), which would accumulate with increasing bacterial population density.^1–3 QS plays a pivotal role in regulating bacterial pathogenesis. For example, QS modulates the production of virulence factors such as pyocyanin and elastase in Pseudomonas aeruginosa during bacterial growth and infection.⁴ Thus, QS signaling pathways are an attractive target for the development of antimicrobial therapeutic agents. P. aeruginosa is a Gram-negative opportunistic pathogen that can cause serious lung infections in cystic fibrosis patients⁵ and microbial keratitis during contact lens wear.⁶ The AIs that control QS signaling in P. aeruginosa include two acylhomoserine lactones with varying alkyl chain lengths and oxidation states at C-3 (C4-HSL and 3-oxo-C12-HSL, Figure 1) and a group of quinolone compounds (Pseudomonas quinolone signal, PQS). They can diffuse freely across cell membranes and bind intracellularly with corresponding receptor proteins (R proteins).⁷ These signaling systems form a complex hierarchical quorum sensing network, where the Las system is considered to be the apex of the hierarchy.⁸ Therefore, the LasR receptor is usually considered to be the target for antagonist and agonist development in P. aeruginosa.^{8, 9} For example, a synthetic non-native AHL, QSI-1, was demonstrated to be a potent LasR antagonist.¹⁰ The structurally unrelated mimic of AHL, TP-1, is a highly selective superagonist of the LasR quorum sensing system; while its derivative TP-5 turned into a moderate QS antagonist.^{11, 12} Moreover, there is evidence showing that N-octanoyl homoserine lactone (C8-HSL, Figure 1) can be produced in the cyanobacterium culture of Gloeothece PCC6909 and its accumulation corresponds to a characteristic pattern of autoinduction.¹³

Figure 1.

Structures of the endogenous quorum sensing signaling molecules in P. aeruginosa (C4-HSL, 3-oxo-C12-HSL), C8-AHL produced in cyanobacterium culture and natural quorum sensing inhibitors from marine cyanobacteria, lyngbyoic acid and pitinoic acid A.

Marine cyanobacteria have been a valuable source for the discovery of biologically active and structurally unique natural products including peptides, polyketides and hybrid of peptide-polyketides.¹⁴ It is noteworthy that marine cyanobacteria also produce various AHL-dependent QS inhibitors.¹⁵ For instance, lyngbyoic acid (Figure 1), a small cyclopropane-containing fatty acid, was isolated from Lyngbya cf. majuscula and proved to strongly inhibit the activity of LasR.⁷ Pitinoic acid A (Figure 1) was also reported to be a P. aeruginosa quorum sensing inhibitor.¹⁶ In addition to these QS inhibitors, here we report the isolation, total synthesis and preliminary biological investigation of a structurally unprecedented QS modulator, doscadenamide A (1a, Figure 2), as a non-HSL QS agonist in a LasR dependent manner.

Figure 2.

Structure of doscadenamide A (1a).

The cyanobacterium Moorea bouillonii was collected at Fingers Reef, Guam, and previous investigation of this cyanobacterium led to the isolation of apratoxin A,¹⁷ lyngbyaloside, 2-epi-lyngbyaloside, 18E-lyngbyaloside C, 18Z-lyngbyaloside C,¹⁸ and apratyramide.¹⁹ The cyanobacterial sample was fractionated as described previously¹⁸ and the isolation was achieved by silica gel column chromatography and several rounds of reversed-phase HPLC to yield doscadenamide A (1a) {white solid, [α]²⁰_D +40 (c 0.07, MeOH)}. The HRESIMS of 1a in the positive mode exhibited a [M + H]⁺ peak at m/z 457.3066, which suggested a molecular formula C₂₇H₄₀N₂O₄ with 9 degrees of unsaturation. The structure of 1a was elucidated using a combination of 1D and 2D NMR techniques. The ¹H and ¹³C NMR spectra in CDCl₃ (Figures S1 and S2) indicated the presence of several characteristic signals corresponding to one O-methyl group (δ_H 3.85 ppm, δ_C 58.9 ppm), two alkyne groups (δ_H 1.92–1.94 ppm, δ_C 68.5, 68.5, 84.6, 84.7 ppm), one α- proton (δ_H 4.64–4.66 ppm, δ_C 59.2 ppm), two α- methyl groups (δ_H 1.12–1.14 ppm, δ_C 16.3–18.1 ppm) and several methylene groups (δ_H 1.30–1.90, 2.06–2.18 ppm, δ_C 18.4–39.4 ppm). Examination of the 2D NMR spectra (COSY, TOCSY, HSQC, HMBC and NOESY, Figures S3–7, Table S1) in CDCl₃ revealed the structural skeleton of 1a, which features two linear alkyne amide side chains (Moya1 and Moya2, Figure 2) and one pyrrolinone core (pyLys-OMe, Figure 2). To establish the absolute configuration, a portion of 1a (2 mg) was treated with ozone at 25 ℃ for 30 min, followed by oxidative workup and acid hydrolysis (Scheme S1). The hydrolyzate was concentrated and partitioned between water and EtOAc. The resulting aqueous phase was analyzed by chiral HPLC-MS revealing the presence of L-Lys, establishing the S configuration of C4 at the pyLys-OMe moiety. The organic residue was coupled with (S)- and (R)-phenylglycine methyl ester (PGME)²⁰ (Scheme S2) to analyze the configuration of the α-methine in the side chain of 1a. The Δδ values (Δδ = δ_S – δ_R) for the PGME derivatives (Figure 3) indicated that the configurations of the α-methine in both side chains of 1a are highly likely R and that the overall configuration is 4S,11R,20R. However, during the investigation, we found a minor diastereomer signal in addition to the major NMR signal corresponding to the α-methyl group, which could be observed in the COSY spectra of the two PGME derivatives (Figures S8 and S9). To further validate the configuration of 1a and provide sufficient material for thorough biological investigation, the total synthesis of 1a was accomplished.

Figure 3.

PGME analysis. Δδ (δ_S – δ_R) values for PGME derivatives of 1a.

As depicted in Scheme 1, the retrosynthetic analysis of 1a relied on the disconnection at the two amide linkages between the pyrrolinone ring and two side chain carboxylic acids, which in the case of 1a are the same as (R)-2-methyloct-7-ynoic acid (Moya, 2a). The pyrrolinone ring can be obtained via the reaction between the double protected amino acid Fmoc-L-Lys (Boc)-OH (5) and Meldrum’s acid (6). As for the synthesis of 2a, the target compound can be achieved in 14 steps using a method reported in 2005.²¹ To improve the efficiency and introduce more flexibility into the production of diverse carboxylic acids with α-substituted alkyl groups, we developed an optimized synthetic method (Scheme 2), where 2a can be obtained in 3 steps in 38% overall yield. The commercially available oct-7-ynoic acid (2c) was activated using pivaloyl chloride followed by addition of the lithium salt of the oxazolidinone chiral auxiliary at −78 ℃. The resulting 7a was methylated under conventional conditions to yield 8a as single diastereomer.^22,23 The target compound 2a was obtained following alkaline hydrolysis by lithium hydroperoxide.²⁴

Scheme 1.

Retrosynthetic analysis of doscadenamide A (1a).

Scheme 2.

Optimized synthesis of Moya (2a).

The total synthesis of 1a was accomplished using the synthetic carboxylic acid 2a (Scheme 3), including generation of the pyrrolinone core and two coupling processes to assemble the structure. The pyrrolidine-2,4-dione 9a was prepared through condensation of 5 with 6 in the presence of EDCI and DMAP, followed by thermolysis.²⁵ This intermediate was used in the next step without purification. Conversion of 9a into its O-methylated derivative 4 was achieved by treatment with trimethylsilyldiazomethane. Subsequently, the N-Fmoc protecting group in 4 was removed using piperidine to yield the secondary amide of tetramic acid 3. The first coupling was accomplished by condensation of the anion derived from deprotonation of 3 by nBuLi and the active ester 10a derived from activation of 2a by pentafluorophenol to yield 11a.²⁶ Based on the ¹H NMR spectrum of 11a, there was less than 5% impurity (Figure S12). In addition, from comparison of the ¹H NMR spectra of 11a and its epimer at C20 11b, we found that this impurity may be introduced due to minor epimerization at C20 (Figure S13). After removal of the Boc-protecting group with TFA, the intermediate 12a was condensed with 2a using typical coupling conditions to afford the target compound 1a in 6 steps with 30% overall yield from 3. The ¹H NMR spectrum of 1a indicated less than 5% impurity and that synthetic 1a had higher diastereomeric purity than the natural product doscadenamide A (see methyl region, Figure 4).

Scheme 3.

Total synthesis of doscadenamide A (1a).

Figure 4.

Selected regional ¹H NMR spectra comparison of the isolated natural product doscadenamide A and synthetic diastereomers 1a, 1b, 1c and 1d (bottom to top, maroon, olive, green, navy and purple) in CDCl₃ (600 MHz) at 27 ℃, with structures of the synthetic diastereomers 1a, 1b, 1c and 1d shown on the left.

To further validate the configuration of the stereocenters in the side chain, the other three diastereomers (1b, 1c and 1d, Figure 4) of 1a were also synthesized using the described synthetic method. Similarly, there is less than 5% impurity shown in the ¹H NMR spectra of 1b, 1c and 1d. Specifically, the absolute configurations of the three stereocenters are 4S,11R,20R for 1a, 4S,11S,20R for 1b, 4S,11S,20S for 1c and 4S,11R,20S for 1d. From comparison of the NMR spectra of all the four diastereomers with those of the isolated natural product doscadenamide A (Figures 4, S10 and S11), the four diastereomers displayed different ¹H NMR signals in δ 5.42–5.55 ppm, δ 3.11–3.28 ppm, δ 2.04–2.21 ppm, δ 1.90–1.95 ppm, δ 1.32–1.56 ppm and δ 1.09–1.22 ppm. Only the ¹H and ¹³C NMR spectra of the synthetic 1a matched those of the natural product. Meanwhile, 1a and 1b exhibited virtually the same ¹³C NMR spectrum as well as the natural product, while 1c and 1d displayed different ¹³C NMR spectra from the isolated doscadenamide A (Figure S11). In addition, the optical rotation value of 1a {[α]²⁰_D +54.3 (c 0.07, MeOH)} further confirmed that the absolute configuration of synthetic diastereomer 1a is consistent with the isolated doscadenamide A, verifying the proposed configuration of doscadenamide A.

Doscadenamide A (1a) and QS signaling molecule 3-oxo-C12-HSL (C12, Figure 1) share structural similarities since they both feature a five-membered ring core and long alkyl side chains. In addition, 1a superficially resembles tetramic acid derivative compounds, and C12 has been previously shown to undergo rearrangement to form a tetramic acid derivative.²⁷ Thus, we proposed 1a may exhibit QS modulatory activities. Our preliminary screening results indicated that 1a can activate the 3-oxo-C12-HSL-responsive reporter plasmid pSB1075,²⁸ a plasmid encoding LasR and containing a light-producing luxCDABE cassette expressed in E. coli (Figure 5A). However, 1a was not able to activate the related reporter pTIM5319,⁷ which is identical to pSB1075 but lacks the AHL-binding site LasR (Figure 5B), which suggests that 1a activates QS via the AHL-binding site. To validate the activity of 1a, we tested its effect on wild-type P. aeruginosa using activating doses (10 μM, 100 μM and 1000 μM) based on reporter assay results. Since 10 μM already caused almost maximal induction of the QS pigment pyocyanin production (Figure S15), we selected 10 μM for subsequent comparative biological investigations. After further optimization of the assay condition using a more concentrated cell density, we included the other three diastereomers. As shown in Figure 5C, after treatment with 1a, 1b, 1c and 1d, pyocyanin levels were elevated in response to all diastereomers after 6 h. Our preliminary results indicated that the C11 and C20 configurations are not crucial for the QS activation by the doscadenamide scaffold but may play a slight modulatory role. However, the assay is sensitive to cell density, timing and other factors. In-depth SAR studies are ongoing to understand the molecular basis. The observed agonistic activities may provide a starting point for the development of potential superagonists with new structural skeletons. Those superagonists are expected to artificially regulate virulence factor production such as pyocyanin and activate QS at lower bacterial cell populations, thus to stimulate the host immune system to clear the infection when fewer bacterial cells are present.³

Figure 5.

Preliminary determination of the activating activity of doscadenamide A (1a) in (A) pSB1075, the lasR-luxCDABE reporter constructs expressed in E. coli, and (B) related reporter pTIM5319, which lacks a functional AHL-binding domain. The bacterial cultures were treated with 1a in a dose-response manner and solvent control at 37 ℃ for 6 h before luminescence was measured. Results are expressed as fold activation compared to solvent control. Effect of doscadenamide A (1a) and its diastereomers 1b, 1c, 1d as well as positive control 3-oxo-C12-HSL (C12) at 10 μM on the (C) production of pyocyanin in wild-type P. aeruginosa after 6 h shaking at 37 ℃. Data are presented as mean ± SD, **P < 0.01, ****P < 0.0001, compared to solvent control using one-way ANOVA (n = 3).

In summary, we have isolated a new quorum sensing modulating agent, doscadenamide A (1a) from a marine cyanobacterium. Its total synthesis has been accomplished through development of an efficient and flexible scheme. Specifically, we developed an efficient method to synthesize α-alkylated carboxylic acids. The quorum sensing activating activity of 1a was verified in the Gram-negative bacterium P. aeruginosa. Doscadenamide A (1a) could serve as a new template for the development of QS superagonists with new skeletons to explore the potential of these activators as therapeutics or chemical tools.