Applying Molecular Networking for the Detection of Natural Sources

Jun 26, 2018 - The cyclic depsipeptide FR900359 (FR), isolated from the traditional Chinese medicine plant Ardisia crenata, is a potent Gq protein inh...
0 downloads 0 Views 2MB Size
Article Cite This: J. Nat. Prod. 2018, 81, 1628−1635

pubs.acs.org/jnp

Applying Molecular Networking for the Detection of Natural Sources and Analogues of the Selective Gq Protein Inhibitor FR900359 Raphael Reher,† Markus Kuschak,§ Nina Heycke,‡ Suvi Annala,‡ Stefan Kehraus,† Hao-Fu Dai,⊥ Christa E. Müller,§ Evi Kostenis,‡ Gabriele M. König,† and Max Crüsemann*,† †

Institute of Pharmaceutical Biology, ‡Molecular, Cellular, and Pharmacobiology Section, Institute of Pharmaceutical Biology, and Pharmaceutical Institute, Institute of Pharmaceutical Chemistry I, University of Bonn, Bonn 53113, Germany ⊥ Institute of Tropical Bioscience and Biotechnology, Chinese Academy of Tropical Agricultural Sciences, Haikou 571101, Hainan, China

Downloaded via KAOHSIUNG MEDICAL UNIV on July 27, 2018 at 06:35:09 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

§

S Supporting Information *

ABSTRACT: The cyclic depsipeptide FR900359 (FR), isolated from the traditional Chinese medicine plant Ardisia crenata, is a potent Gq protein inhibitor and thus a valuable tool to study Gq-mediated signaling of G protein-coupled receptors. Two new FR analogues (3 and 4) were isolated from A. crenata together with the known analogues 1 and 2. The structures of compounds 3 and 4 were established by NMR spectroscopic data and MS-based molecular networking followed by in-depth LCMS2 analysis. The latter approach led to the annotation of further FR analogues 5−9. Comparative bioactivity tests of compounds 1−4 along with the parent molecule FR showed high-affinity binding to Gq proteins in the low nanomolar range (IC50 = 2.3− 16.8 nM) for all analogues as well as equipotent inhibition of Gq signaling, which gives important SAR insights into this valuable natural product. Additionally, FR was detected from leaves of five other Ardisia species, among them the non-nodulated leaves of Ardisia lucida, implying a much broader distribution of FR than originally anticipated.

P

pharmacological research, but it also offers new strategies for the treatment of complex diseases with the pathogenesis involving many GqPCRs, such as asthma, or constitutively activated Gq proteins, as in uvea melanoma.5,11−16 To evaluate the molecular interaction of FR with its target, further derivatives of this molecule are needed. Also, as FR is described as binding in a pseudoirreversible manner to Gq proteins, there is an urgent need for FR analogues with an altered kinetic profile.5 Our previous studies revealed that a bacterial endophyte, i.e., Candidatus Burkholderia crenata, located in the leaf nodules of A. crenata, harbors the genetic information for FR production.11 This cyclic depsipeptide is synthesized by two nonribosomal peptide synthetases (NRPSs), one responsible for the heptacyclic part of the molecule and the second for the side chain comprising N-propionylhydroxyleucine. Recently, it was suggested that FR serves to protect the host plant against

lants are one of the major sources for therapeutically applied compounds, e.g., paclitaxel, vinca alkaloids, and camptothecin. In most cases, the higher plants are themselves the producers of these metabolites. However, all of the estimated 300 000 higher plants are suspected to contain endosymbiotic organisms, which have been neglected by researchers until recently.1−3 Thus, endophytes open up new research fields in terms of natural products for medical use, for investigations into the chemical ecology, and for evolutionary aspects of symbioses. The natural product FR900359 (FR) (Chart 1), isolated from the traditional Chinese medicine plant Ardisia crenata (Primulaceae, subgenus Crispardisia),4 is the most potent and selective Gq protein inhibitor so far discovered.5,6 Gq proteins play a major role in the signal transduction of G proteincoupled receptors (GPCRs) and thus are involved in many physiological processes. Traditionally, the roots of A. crenata have been applied as a medicine for the treatment of respiratory tract infections.7−10 In this regard, FR is not solely of outstanding importance as a chemical tool for fundamental © 2018 American Chemical Society and American Society of Pharmacognosy

Received: March 17, 2018 Published: June 26, 2018 1628

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products



Chart 1. Structures of FR900359 (FR) and FR Analogues 1−4a

Article

RESULTS AND DISCUSSION

Detection and Structure Elucidation of FR Analogues from A. crenata. A. crenata leaves were extracted with MeOH, and after liquid−liquid separation, a butanolic extract was obtained. Subsequently, the n-BuOH phase was further fractionated using vacuum liquid chromatography and size exclusion chromatography on Sephadex LH-20. The depsipeptide-containing fraction was analyzed with uHPLC-MS/MS and subsequent GNPS molecular networking25 to visualize the FR molecular family26 (Figure 1). This MS2-based approach allows comparisons of fragmentation patterns of molecular ions leading to clustering of structurally similar compounds and their visualization in a molecular network. Here, molecular ions represent nodes, which are connected by edges, highlighting the structural relationships and similarities of the molecules. Molecular networking has been successfully applied to dereplicate natural products from complex extracts,27 to compare the metabolomes of various natural product producers,28 and to improve and accelerate natural product isolation and structure elucidation workflows.29−32 The molecular network of the FR family (Figure 1) shows 22 nodes. Protonated FR (m/z 1002.54) clusters with 1 (m/z 1032.55),12 2 (m/z 988.53),23 and additional analogous peptides, e.g., m/z 974.51, 1016.55, and 817.44. We intended to assign some of these structures of putative FR analogues via manual analysis of the MS2 spectra. For that reason, one characteristic MS fragmentation pathway, observed for proton adducts of FR and presumably all putative FR analogues, was annotated as shown in Scheme 1. In a first step, the core structure is cleaved between alanine and N-methylalanine and linearized. In cyclic peptides, breakage of the amide bond (or ester bond for depsipeptides, respectively) is reported to lead to the formation of a linear peptide with a C-terminal oxazolone ring,33 which sequentially loses amino acids. This is evident when LCMS2 spectra of FR are compared with those of the known and structurally proven analogues 1 and 2.12,23 The different acylation patterns in 1 and 2 lead to a characteristic change in the mass of the respective fragments (Scheme 1). In a former study we assigned the molecule with m/z 817.44 as a putative biosynthetic FR precursor FR-SC (5), denoting FR without side chain.12 Analysis of fragmentation spectra of m/z 831.46 (6, FR-SC-1), m/z 803.42 (7, FR-SC-2), m/z 847.45 (8, FR-SC-3), and m/z 888.47 (9, FR-SC-4), clustering with m/z 817.44, gives additional evidence for this assumption and suggests that there is a multitude of further FR derivatives differing from the major metabolite by lacking the side chain (Figure S2, Scheme S2, Supporting Information). In that way, 6 was assigned to be the precursor of 3 with Npropionylhydroxyleucine1 instead of N-acetylhydroxyleucine1, 7 with a modified N-methylalanine residue lacking one methylene group, 8 to be the precursor of 1, thus harboring N-3-hydroxypropionylhydroxyleucine1 instead of N-acetylhydroxyleucine1, and most interestingly 9 to be an analogue of 5 with an additional alanine moiety in its cyclic backbone. These derivatives underline the hypothesis that the transesterification step of the side chain to the cyclic core peptide is the last step in the biosynthesis of FR, after all modifications, such as acylations or epimerizations, have occurred.12 Compound 9 also demonstrates unexpected flexibility of the FR-NRPS, as

a

Building blocks are colored and annotated for FR.

herbivores.11,12 It is the first discovered defense chemical that acts by inhibition of Gq proteins, thus representing an effective and unique mode of action. FR is suspected to act on many organisms based on the fact that G proteins are highly conserved in the evolution of metazoans.17 Many organisms suffer adverse or toxic effects after exposure to FR, which we have shown for insects and mammals.12 A characteristic feature of plants of the subgenus Crispardisia, which includes among many other species A. crenata, are so-called “bacterial leaf nodules”, i.e., morphological structures housing endosymbiotic bacteria of the genus Burkholderia.18−22 Recently, we demonstrated colocalization of FR and endosymbionts via MALDI imaging. We were also able to isolate the first natural FR analogues, i.e., FR-1 (1, formerly called AC-1)12 and FR-2 (2, formerly called AC-0),23 both containing altered N-acylation patterns at the hydroxyleucine residues (Scheme 1). Hence, the question arose whether there are more FR derivatives in A. crenata or in other species belonging to the subgenus Crispardisia. The work presented here provides the structure elucidation of new FR analogues, i.e., FR-3 (3) and FR-4 (4), from A. crenata, predominantly achieved by in-depth LCMS2 data analysis and confirmed by NMR spectroscopic data analysis (Chart 1). Interestingly, FR-3 has the same 2D structure as sameuramide, a compound previously isolated from a marine ascidian.24 Biological evaluation of 3 and 4 revealed potent Gqinhibiting properties comparable to those of FR, 1, and 2 (Figure 2). By applying Global Natural Products Molecular Networking (GNPS), it was also possible to discover other FR producers besides A. crenata. Indeed, we demonstrate that FR is much more broadly distributed in nature than was previously known. It is present not only in A. crenata, but also in A. hanceana, A. mamillata, A. villosa, A. crispa, and even the nonnodulated A. lucida. 1629

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products

Article

Scheme 1. (A) Fragmentation Pathway of FR900359 (FR) Proton Adduct;a (B) Fragmentation Pathway of 1 (FR-1) (L* = N3-hydroxypropionylhydroxyleucine1); (C) Fragmentation Pathway of 2 (FR-2) (L^ = N-acetylhydroxyleucine2)

a FR ring structure is first cleaved, followed by sequential loss of amino acids (m/z 1002.54 → (-L‴) m/z 817.43 → (-L″) m/z 688.35 →(-A′) m/z 603.30 → (-T′) m/z 474.22 → (-L′) m/z 303.14 → (-A) m/z 232.10). Peptide fragment ions are labeled according to a nomenclature system developed by Ngoka et al.34 based on Biemann’s modifications35 of Roepstorff’s nomenclature36 in one-letter amino acid code. L‴ = Npropionylhydroxyleucine2, L″ = hydroxyleucine, A′ = N-methylalanine, T′ = N,O-dimethylthreonine, L′ = N-acetylhydroxyleucine1, A = alanine, F′ = phenyllactic acid, A″ = N-methyldehydroalanine.

apparently module 5 has been used twice to incorporate an additional alanine. These results encouraged us in attempting to isolate FR analogues clustering with FR in the network. From the molecular network (Figure 1), an ion with m/z 1016.55 attracted attention, because it directly clustered with the FR

ion (m/z 1002.54) and appeared in sufficient intensities for isolation. Investigation of the extracted ion chromatogram of m/z 1016.55 of the isolated sample indicated that it consisted of two isomeric compounds, as evidenced by slightly different retention times (ΔtR 0.2 min, Figure S3, Supporting Information). These compounds were named FR-3 (3) and 1630

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products

Article

Thus, together with further evidence from other fragmentation pathways, compound 3 contains a modified N-acetylhydroxyleucine building block. Taking biosynthetic considerations into account reduces the possible structures of the modified residue to either N-propionylhydroxyleucine1 or N-acetylhydroxyhomoleucine1. Compound 4 differs from FR as follows: Instead of alanine, 4 contains a residue with a 14.01 Da higher mass, i.e., homoalanine or alternatively an N-methylalanine building block, as revealed from the LCMS2 data (Figure S4, Supporting Information). This was again evident from different Δm/z values resulting from sequential loss of amino acids. Compound 4, but not FR, produced a fragment with m/z 470.22. Instead of a Δm/z 71.03 Da loss as for FR, 4 loses a Δm/z 85.05 Da fragment, leading to the same m/z 385.17 fragment. Collectively, MS2 analysis permitted reduction of the number of possible structures for 3 and 4 to two for each compound. Applying a further HPLC purification step permitted isolation of 3 as a white powder, only containing minor amounts of 4 (ratio 3/4 = 3:1). After recording and analyzing 1 H NMR and 1H and 13C HSQC spectra, the structures of 3 and 4 were unambiguously assigned (Table 1, Figures S5−S11, Supporting Information) in comparison to the NMR spectra of FR and 1.12 Evidence for the structure of 3 was provided from (i) absence of the NMR signals for an acetate moiety at δH 2.23/δC 22.6 for FR; (ii) the additional NMR signals for a methylene group CH2-24 and methyl group CH3-25, resulting in resonances at δH 2.55/δC 28.7 and δH 1.24/δC 10.1, respectively (Table 1); and (iii) the COSY correlations of H224 and H2-25 (Figure S5, Supporting Information) and HMBC correlations of H2-24 and H2-25 to the neighboring amide carbonyl group (C-23, δC 175.6). Thus, compound 3 contains N-propionylhydroxyleucine1 instead of N-acetylhydroxyleucine1 in the cyclic backbone. After the assignment of 3, 4 was assigned from the NMR spectra of the 1:1 mixture. Evidence for the homoalanine residue of 4 was provided from (i) the shifted methine αCH-2 δH 4.69/δC 51.4, as compared to δH 4.90/δC 45.9 for FR; (ii) the additional NMR signals for the methylene group CH2-3a/3b (δHa 1.78/δHb 2.00/δC 26.4) and methyl group CH3-4 (δH 0.97/δC 10.3), respectively (Table 1); and (iii) the COSY correlations of H-2 to H2-3 and H2-3 and H3-4 and HMBC correlations of H2-4 to the neighboring methine αCH-2 (δH 4.69/δC 51.4). Thus, 4 contains a homoalanine instead of an alanine moiety (Scheme

Figure 1. Molecular network of the FR molecular family from Ardisia leaves (Sephadex LH-20 fraction). Nodes display distinct m/z features, displaying their parent mass. Width of edges is proportional to cosine. Sizes of the nodes correspond to number of species it is found in (small nodes all stem from A. crenata extract). Highlighted in red are FR and the structurally related derivatives FR-1 to FR-4 (1− 4). Orange-labeled nodes are FR analogues 5−9, lacking the side chain N-propionylhydroxyleucine2. For (putative) structures of 1−9, see Schemes S1 and S2 (Supporting Information).

FR-4 (4). 1H NMR spectroscopy revealed that the compounds were present in a 1:1 mixture, making an assignment of resonances impossible at this stage. The molecular formulas of 3 and 4 were determined to be C50H77N7O15 (calcd 1016.5550; obsd 1016.5544 (3) and 1016.5553 (4), respectively) for [M + H]+. MS data revealed a mass difference of 14.01 Da between FR and compound 3 and 4 and thus suggested an additional methyl or methylene group present in 3 and 4. In-depth MS2 analysis (Figure 2) permitted the separate analysis of 3 and 4. FR, 3, and 4 can be distinguished by different Δm/z values resulting from differently modified amino acids that were sequentially lost (Figure S4, Supporting Information). In the fifth fragmentation step, m/z 474.22 of FR loses N-acetylhydroxyleucine (Δm/z 171.08 Da) to yield an m/z 303.14 ion. The corresponding m/z 488.24 ion of 3 loses a Δm/z 185.10 Da fragment to give the same m/z 303.14 ion.

Figure 2. Bioactivity of FR, 1, 2, and 3/4 (3:1 ratio). (A) Inhibition of agonist-induced second messenger production in CHO cells expressing the Gq-sensitive muscarinic M1 receptor. Indicated analogues were evaluated against the muscarinic agonist carbachol at its EC80. (B) Competition binding study of FR, 1, 2, and 3/4 (3:1 ratio) versus [3H]PSB-15900 (5 nM), the tritiated derivative of FR, by cell membrane preparations of platelets. Values represent means ± SEM of three independent experiments. pIC50 values ranging between 7.79 and 8.64 were determined. 1631

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products Table 1. 1H and

13

Article

C NMR Spectroscopic Data of 3 (FR-3) and 4 (FR-4) in CDCl3 (1H: 600 MHz; 13C: 150 MHz) FR-3 (3)

c

residue Ala

N-Me-Dha

Pla

N-Pr-β-HyLeu1

N,O-Me2-Thr

β-HyLeu

N-Me-Ala

N-Pr-β-HyLeu2

no. C/H 1 2 2-NH 3 4 5 6a 6b 7 8 9 10a 10b 11 12/16 13/15 14 17 18 18-NH 19 20 21 22 23 24a 24b 25 26 27 28 29 30 31 32 33 33-NH 34 35 36 37 38 39 40 41 42 43 43-NH 44 44-OH 45 46 47 48 49a 49b 50

b

FR-4 (4)

δC, mult

δH (J [Hz])

a

172.6, C 45.6, CH 18.0, CH3 164.0, C 145.5, C 106.9, CH2 36.3, CH3 167.9, C 72.8, CH 36.8, CH2 136.1, C 129.4, CH 128.8, CH 127.2, CH 169.4, C 50.3, CH 77.8, CH 29.1, CH 18.9, CH3 18.8, CH3 175.6, C 28.7, CH2 10.3, CH3 166.6, C 64.4, CH 72.3, CH 16.2, CH3 28.8, CH3 57.0, CH3 171.4, C 46.9, CH 77.1, CH 30.4, CH 19.3, CH3 18.3, CH3 170.1, C 56.5, CH 14.2, CH3 31.3, CH3 170.4, C 56.8, CH 78.3, CH 30.0, CH 20.3, CH3 18.7, CH3 174.9, C 28.8, CH2 10.0, CH3

a

residue

c

Homo-Ala 4.91 (dq, 9.1, 6.7) 8.52 (d, 9.1) 1.41 (d, 6.7)

N-Me-Dha

a 5.32 (d, 2.2) b 5.07 (d, 2.2) 3.16 (s) 5.19 (dd, 4.1, 8.3) a 3.11 (dd, 4.1, 14.8) b 2.98 (dd, 8.3, 14.8)

Pla

7.26d 7.32d 7.25d 5.30 (dd, 1.3, 10.0) 7.45, (d, 10,0) 5.12 (dd, 1.3, 10.0) 1.91 (m) 1.01 (d, 6.8) 0.85 (d, 6.8)

N-Ac-β-HyLeu1

2.59 (m) 2.53 (m 1.24 (m) N,O-Me2-Thr 4.08 3.76 1.18 2.70 3.41

(d, 9.8) (m) (d, 4.8) (s) (s) β-HyLeu

5.36 (d, 9.9) 6.75, (d, 9.9) 5.31 (m) 1.75 (m) 1.10 (d, 6.7) 0.83 (d, 6.7) N-Me-Ala 4.71 (q, 6.8) 1.41 (d, 6.8) 2.90 (s) N-Pr-β-HyLeu2 4.57 7.25 3.73 6.87 1.99 1.17 0.89

(dd, 1.8, 7.8) (d, 7.8) (m) (d, 4.2) (m) (6.7) (6.7)

a 2.57 (dq, 14.9, 7.5) b 2.50 (dq, 14.9, 7.5) 1.19 (t, 7.5)

no. C/H 1 2 2-NH 3a 3b 4 5 6 7a 7b 8 9 10 11a 11b 12 13/17 14/16 15 18 19 19-NH 20 21 22 23 24 25 26 27 28 29 30 31 32 33 33-NH 34 35 36 37 38 39 40 41 42 43 43-NH 44 44-OH 45 46 47 48 49a 49b 50

b

δC,a mult 172.2, C 51.4, CH 26.4, CH2 10.3, CH3 164.4, C 145.5, C 106.9, CH2 36.3, CH3 167.9, C 72.7, CH 36.8, CH2 136.1, C 129.7, CH 128.7, CH 127.1, CH 169.4, C 50.5, CH 77.8, CH 29.0, CH 19.1, CH3 19.0, CH3 171.6, C 22.6, CH3 166.6, C 64.6, CH 72.5, CH 16.2, CH3 28.9, CH3 57.3, CH3 171.4, C 46.8, CH 77.1, CH 30.7, CH 19.5, CH3 18.3, CH3 170.1, C 56.4, CH 14.4, CH3 31.5, CH3 170.4, C 57.0, CH 78.4, CH 30.1, CH 20.6, CH3 18.7, CH3 174.9, C 28.8, CH2 10.1, CH3

δHa (J [Hz]) 4.69 (m) 8.49 (d, 9.1) a 1.78, m b 2.00, m 0.97 (t, 7.2)

a 5.37 (d, 2.2) b 5.09 (d, 2.2) 3.20 (s) 5.21 (dd, 4.1, 8.3) a 3.11 (dd, 4.1, 14.8) b 2.98 (dd, 8.3, 14.8) 7.26d 7.29d 7.23d 5.26 (dd, 1.3, 10.0) 7.55, (d, 10,0) 5.12 (dd, 1.3, 10.0) 1.89 (m) 1.01 (d, 6.8) 0.85 (d, 6.8) 2.26 (s) 4.07 (d, 9.8) 3.75 (m) 1.23, (d, 4.8) 2.70 (s) 3.41 (s) 5.38 (d, 9.9) 6.70, (d, 9.9) 5.31 (m) 1.72 (m) 1.10 (d, 6.7) 0.83 (d, 6.7) 4.69 (q, 6.8) 1.41 (d, 6.8) 2.88 (s) 4.55 7.17 3.73 6.81 1.98 1.17 0.88

(m) (d, 7.8) (m) (d, 4.3) (m) (6.7) (6.7)

2.57 (dq, 14.9, 7.5) 2.50 (dq, 14.9, 7.5) 1.19 (t, 7.5)

a

Assignments are based on 1D and 2D NMR measurements (HMBC, HSQC, COSY). 13C NMR spectra were recorded at 150 MHz. bNumbers according to Scheme S1. cResidues: Ala = alanine, Homo-Ala = homoalanine, N-Me-Dha = N-methyldehydroalanine, Pla = 3-phenyllactic acid, N1632

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products

Article

Table 1. continued Ac-β-HyLeu1 = N-acetylhydroxyleucine1, N,O-Me2-Thr = N,O-dimethylthreonine, β-HyLeu = hydroxyleucine, N-Me-Ala = N-methylalanine, NPr-β-HyLeu2 = N-propionylhydroxyleucine2. dOverlaying resonances. Numbering of atoms is depicted in Scheme S3.

and purified membrane preparations of platelets (Figure 2B).5 All of the tested compounds show Ki and IC50 values in the low nanomolar range, with the new 3/4 (3:1) having the highest Gq affinity for any compound so far discovered (Ki ± SEM = 1.24 ± 0.06 nM). In terms of potential to inhibit Gq signaling, no compound surpassed the activity of the parent substance FR. This might be explained by slightly reduced cell permeability of the analogues compared to FR, for which a specific transport system can be assumed. With respect to structure−activity relationship (SAR), it may be concluded that alterations in the length of the acyl chains at the hydroxyleucine moieties as well as a substitution of an alanine for a homoalanine moiety hardly affect Gq binding and inhibition of downstream signaling. Thus, in contrast to most of the synthetically derived Gq inhibitors, all of the analogues provided by nature (1, 2, and 3/4 (3:1)) are potent and efficacious. In summary, the utility of GNPS molecular networking for the detection and dereplication of a family of FR derivatives in the traditional Chinese medicine plant A. crenata is demonstrated. Applying in-depth MS2 analysis, a multitude of FR analogues lacking the side chain were dereplicated, which confirms the biosynthetic hypothesis. Based on the molecular network two new isomeric FR analogues, FR-3 (3) and FR-4 (4), were identified. These new FR analogues demonstrate, together with the known FR-1 (1) and FR-2 (2), high affinities to the Gq protein, with IC50 values in the low nanomolar range. It is obvious that the known natural FR derivatives have similar affinity and downstream effects mediated by Gq proteins. In contrast, synthetic FR analogues were mostly devoid of activity.6,2338 Additionally, new FR producers, namely, the nodulated A. hanceana, A. villosa, A. mamillata, and A. crispa and unexpectedly also the non-nodulated A. lucida, were discovered. This finding, together with reports on the FR-related compounds YM-254890 and sameuramide (with the same 2D structure as FR-3), that were found in a soil bacterium and a marine tunicate, respectively, suggests a broad distribution of the FR molecular family in nature. This may implicate an important ecological role for this natural product and its congeners, not only for Ardisia plants but also in several other contexts.

1). Configurations of the compounds 3 and 4 were deduced from the NRPS assembly line and in analogy to FR. Detection of FR by Molecular Networking in Further Species of the Subgenus Crispardisia. Leaves of several species belonging to Crispardisia, i.e., A. hanceana, A. villosa, and A. mamillata, were collected in 2015 from tropical forest habitats in Hainan, China. Additionally, plants of A. crispa, A. polycephala, and the non-nodulated species A. lucida (Figure S12, Supporting Information) were cultivated in the Botanical Garden Bonn, Germany. The dried leaves were processed in a similar way to that described above for A. crenata. In the MeOH extract of A. crispa and the n-BuOH extracts of A. hanceana, A. villosa, and A. mamillata, an ion with the same mass and retention time as FR was detected, although its concentration was significantly lower than in A. crenata extracts (Figure S13, Supporting Information). In crude extracts of A. lucida and A. polycephala, such an ion could not be detected. Further fractionation of the extracts permitted detection of the FR signal in a subfraction of the A. lucida extract, but not for A. polycephala (Figure S13, Supporting Information). The FRcontaining samples were networked with the A. crenata extract to confirm the identity of FR by MS2 comparison and matching with an authentic FR standard uploaded to the GNPS database (Figure 1). Minor FR derivatives could not be detected in the other plant species. However, it may be that they are also produced as in A. crenata, but appear in quantities below detection level. The detection of FR in the non-nodulated A. lucida suggests the presence of an endosymbiont that is not located in the usual nodules and may thus represent the very early stage of a plant−bacterial symbiosis. Overall, the results emphasize a much broader distribution of FR in the Ardisia genus than previously known, since six out of seven analyzed Ardisia species do contain FR. Additionally, FR derivatives like YM254890 and sameuramide were found in a soil bacterium37 and a marine tunicate,24 respectively. This points to a much broader distribution of this family of molecules than currently thought. Bioactivity of FR Analogues. FR is characterized by its ability to inhibit Gq proteins and, in this way, the signal transduction of many GPCRs. In the context of the current study the question arose whether FR analogues differ in their affinity toward and their effect on the downstream signaling of Gq proteins. To investigate this, competitive radioligand binding studies and IP-1 accumulation assays were performed. Owing to the minute amounts of the 3:1 mixture (approximately 500 μg from 2 kg of leaves) of compounds 3 and 4, the mixture was evaluated biologically without further purification. Compounds 3/4 (3:1) were evaluated for their Gq-inhibiting properties via an IP1 accumulation assay, using a CHO cell line, expressing the Gq-coupled muscarinic M1 receptor. Intriguingly, while most of the synthetic FR analogues6,38 have shown a dramatic decrease in potency, the new FR analogues 3/4 (3:1) are almost equipotent to FR in terms of inhibiting Gq signaling (Figure 2A). The same picture can be drawn by comparing the affinities of FR, 1, 2, and 3/4 (3:1) to the Gq protein revealed from competitive binding assays, applying the tritiated FR-radioligand ([3H]PSB-15900)



EXPERIMENTAL SECTION

General Experimental Procedures. NMR spectra were recorded on a Bruker Ascend 600 NMR spectrometer operating at 600 MHz (1H) and 150 MHz (13C) using CDCl3 as solvent from Deutero GmbH (99.8% D). NMR spectra were processed using Bruker Topspin version 1.3 software and MestReNova 8.0.1. Spectra were referenced to residual solvent signals with resonances at δH/C 7.26/77.1 for CDCl3. LCMS data were analyzed by uHPLC-MS/MS on a micrOTOF-Q mass spectrometer (Bruker) with an ESI source coupled with an HPLC Dionex Ultimate 3000 (Thermo Scientific) using a Zorbax Eclipse Plus C18 1.8 μm column, 2.1 × 50 mm (Agilent). The column temperature was 45 °C. MS data were acquired over a range from 100 to 3000 m/z in positive mode. Auto MS/MS fragmentation was achieved with rising collision energy (35− 50 keV over a gradient from 500 to 2000 m/z) with a frequency of 4 Hz for all ions over a threshold of 100. uHPLC begins with 90% H2O containing 0.1% acetic acid. The gradient starts after 0.5 min to 100% 1633

DOI: 10.1021/acs.jnatprod.8b00222 J. Nat. Prod. 2018, 81, 1628−1635

Journal of Natural Products

Article

acetonitrile (0.1% acetic acid) in 4 min. A 2 μL amount of sample solution was injected at a flow of 0.8 mL/min. All solvents were LCMS grade. Organism Collection and Identification. A. crenata was purchased commercially and afterward cultivated at the Botanical Garden Bonn of the University of Bonn together with A. polycephala and A. lucida. Herbarium specimens are located at the Institute for Pharmaceutical Biology of the University of Bonn. The A. villosa, A. mamillata, and A. hanceana samples were collected in January 2016 at a height of 800 m by excursion near Bawang Mountain in Changjiang County, Hainan, China. The specimens were dried at 25 °C until extraction. Botanical identification was performed by H.D., and a voucher specimen was deposited at the Herbarium of the Institute of Tropical Bioscience and Biotechnology, Chinese Academy of Tropical Agricultural Sciences. Extraction and Isolation. Bioassay-, LC-MS-, and 1H NMRguided fractionation of the crude extracts obtained from the plant A. crenata yielded the cyclic depsipeptide FR. The plant material was cultivated in the greenhouse of the Botanical Garden Bonn. Dried plant leaves (200 g) were coarsely ground and extracted with MeOH. Further purification steps included liquid−liquid extraction, RP-18 vacuum liquid chromatography, and size exclusion chromatography. Final purification was done by HPLC with a semipreparative YMC Hydrosphere RP-18 column, 250 × 4.6 mm, 3 μm, using 80% MeOH isocratic and 0.7 mL/min flow for 35 min. The resulting crude peptide was purified again with a Nucleoshell RP18+ column, 250 × 4.6 mm, 5 μm using 76.5% MeOH and 0.7 mL/min. Pure FR900359 was isolated as a white powder (tR: 27 min, 10 mg). A shoulder of the FR peak eluting after FR was collected (tR: 30.2 min). This fraction included a peptide mixture of FR-3 and FR-4 as judged from LC-MS data. HPLC conditions had to be adjusted to separate the new peptides from each other. The composition of the mobile phase was changed to isocratic 70% MeOH for 8 min, followed by a linear gradient to 89% MeOH in 9 min, then to 100% MeOH in 1 min, applying a Nucleoshell RP18+ column, 5 μm, 250 × 4.6 mm, with a flow rate of 0.9 mL/min. The new cyclic depsipeptides FR-3/FR-4 = 3:1 were obtained as a white powder (tR: 13.8 min, 0.2 mg from 200 g of dried leaves). The isolation process was repeated until 0.8 mg of FR-3/FR-4 was collected for structure elucidation. Compound 3/4 (FR-3/FR-4):