Theopapuamide: a Cyclic Depsipeptide from a Papua New Guinea Lithistid Sponge Theonella swinhoei

Marine sponges have proven to be a significant source of biologically active cyclic peptides and depsipeptides.1 Among these sponge depsipeptides, callipeltin A (from New Caledonian sponge Callipelta sp.),2 neamphamide A (from a Papua New Guinea sponge Neamphius huxleyi)3 and Papuamides A-B (from Papua New Guinea sponges Theonella mirabilis and Theonella swinhoei)4 are well known for their potent HIV-inhibitory activity and their structurally unique features incorporating several modified amino acid residues. For instance, the atypical amino acid residues 3,4-dimethyl-L-glutamine and β-methoxytyrosine are common to all of the above mentioned marine depsipeptides but to date have not been described elsewhere. The rarity of these atypical amino acid residues has inspired their chemical synthesis.5 The structural diversity found among lithistid sponge metabolites (genus Theonella and Callipelta) has been attributed to symbiotic microorganisms.1 As part of our continuing studies on Theonella swinhoei from Papua New Guinea,6 the aqueous CH3CN extract of the sponge was analyzed and proved active in an in vitro anti-HIV assay. Fractionation of the active extract resulted in the isolation of a new cyclic depsipeptide, theopapuamide (1). This paper describes the isolation, structure elucidation, and stereochemical analysis of theopapuamide (1). The crude aqueous CH3CN extract of T. swinhoei (family Theonellidae) was concentrated under vacuum and fractionated by C18 flash column chromatography. Further purification on Diaion HP-20 resin followed by CN-HPLC afforded the new cyclic undecapeptide, theopapuamide (1, 15.8 mg, 3.95 × 10−3 % yield wet wt) as an off-white amorphous solid ([α]25D −3.0, c 0.86, MeOH). For compound 1, a molecular formula of C69H123N17O23 was established based on the divalent molecular ion [M+2H]2+ observed at m/z 779.9563 by HRESI-FTMS. The protonated molecular ion at m/z 1558.9053 [M+H]+ in the HRESI-FTMS was in agreement with a molecular weight of 1557.8980 Da (Δ = −0.2 mmu) for the neutral compound (1). The molecular formula suggested seventeen units of unsaturation. Subsequent examination of 1D-spectra of 1 reflected characteristic features of a peptide, such as an abundance of exchangeable N-H protons (δH 6.55–9.05) and carbonyl signals (δC 169.9–179.6) in the 1H and 13C NMR spectra, respectively. However, the interpretation of NMR spectra of 1 was hampered by the existence of multiple conformations in several of the common NMR solvents (MeOH-d4, CD3CN:DMSO-d6 and DMSO-d6). Attempts to improve resolution by employing acetone-d6 and DMF-d7 as solvents failed due to poor solubility of 1. Moreover, efforts to achieve a favorable conformational ratio by addition of LiCl7 (0.5–4 equi., CD3CN:DMSO-d6, 25 °C) was also ineffective. Gratifyingly, one major conformation, with improved resolution, was observed using a mixture of CD3CN:H2O (4:1) at 25 °C. The aliphatic nature of the peptide was supported by the absence of any sp2-hybridized carbon resonances in the region between 110–150 ppm, as well as localization of resonances in the upfield region of the 13C NMR spectrum of 1. Due to the extensive overlap of aliphatic resonances in 1, the assembly of individual amino acid residues required combined analysis of 1H-1H-COSY, gHSQC, gHSQC-TOCSY, 1D-TOCSY and z-DIPSI-tocsy8 spectra. In conjunction with NMR analysis, the gross structure elucidation of 1 was guided by standard amino acid analysis,9 which revealed molar concentrations of ~1:2:1 for Asx, Thr and Leu respectively. The presence of two N-methylated amino acid residues were suggested based on the characteristic 1H and 13C chemical shifts of the N-methyl groups at δH 2.81 (δC 30.8) and δH 2.88 (δC 31.8). HMBC correlations were used to identify these N-methylated amino acids as NMeLeu and NMeGln, respectively. Additionally, the presence of a methoxy-bearing amino acid residue was suggested by the characteristic 1H and 13C chemical shifts of the O-methyl group at δH 3.34 (δC 60.3). Based on HMBC correlations, the methoxy-bearing amino acid residue was identified as β-OMeAsn. An ester linked threonine residue was suggested from a typical ~1.0 ppm downfield shift of the β-hydroxymethine proton (δH 5.55). A subsequent 15N-HSQC experiment showed four pairs of signals for the primary amide protons of 3,4-diMeGln, Asn, β-OMeAsn and NMeGln residues. However, the 1H-15N-correlations for the primary amino protons of Orn and Dpr could not be observed under the given experimental conditions [CD3CN-H2O (4:1), 25 °C]. The connectivity between amino acid residues was established based on careful analysis of HMBC, WATERGATE-NOESY and WATERGATE-ROESY10 spectra. Further inspection of HMBC data supported formation of an 8-residue ring moiety in 1 via an ester linkage between the C-3-hydroxyl group of Thr (δH-3 5.55) and the C-1-carboxylic acid of NMeLeu (δC-1 171.5). Additionally, a NOESY cross peak between Thr (δH-3 5.55) and NMeLeu (δH-2 5.07) was in agreement with such a macrocyclic ring formation. The constituents of the macrocycle, identified based on sequential α-NH/NH and α-NH/NCH3 NOESY correlations, consisted of NMeLeu (δCH3 2.81), Asn (δH 8.18), β-OMeAsn (δH 6.69), NMeGln (δCH3 2.88), Leu (δH 7.20), Orn (δH 8.00), and two residues of Thr (Thr1 δH 8.25, Thr2 δH 8.89). Furthermore, based on HMBC and NOESY data, the α-NH (δH 8.89) of the ester linked Thr was attached to a 4-residue sequence of 3,4-diMeGln, 4-amino-5-methyl-2,3,5-trihydroxy-hexanoic acid (Amtha), 2,3-diaminopropionic acid (Dpr) and 3-hydroxy-2,4,6-trimethyl-octanoic acid (Htoa), forming the linear portion of the peptide. Of the component amino acid residues identified in undecapeptide 1, 5 of the residues were also described in both callipeltin A and neamphamide A (3,4-diMeGln, Thr (×2), Leu, and NMeGln).2,3 A Htoa dimer, bourgeanic acid has been isolated from several Ramalina sp. lichens.11 Concurrent attempts to sequence the peptide by Edman degradation12 were unsuccessful signifying a concealed or chemically modified N-terminus in 1. Additionally, theopapuamide (1) was resistant to digestion by a variety of proteases (trypsin, thermolysin and pepsin)13 likely due to collective effects of N-methylated amino acid residues, peptide bonds involving D-amino acids and the cyclic nature of 1. Initial attempts to clarify the amino acid sequence by MS/MS (ESI and MALDI) analysis of the intact cyclic depsipeptide (1) was not informative, and yielded random cleavage products. Subsequently, the ester linkage in theopapuamide (1) was subjected to base hydrolysis (1N KOH, r.t., 2 h) to generate the acyclic peptide 2 [m/z 1576.9 (M + H)+]. As anticipated, the acyclic peptide (2) was amenable to tandem MS approach.14 Accordingly, successful sequence analysis of 2 was carried out on the basis of fragment ion spectra generated by SORI-CID ESI-FTMS/MS (Figure 1). Figure 1 SORI-CID MS/MS fragmentations of the acyclic peptide (2) The absolute configurations of the amino acid constituents of theopapuamide (1) were determined by acid hydrolysis of 1 (6N HCl, 110 °C, 12 h), followed by chiral HPLC analysis of the hydrolysate, and RP-HPLC analysis of FDAA derivatives.15 By chiral HPLC [Chirex phase 3126(D), iPrOH-2mM CuSO4 (5:95)], diagnostic peaks were observed for L-NMeLeu, D-Asp, L-Leu, L-NMeGlu, D-Orn and D-Dpr residues. The chrial HPLC analysis failed to give sufficient separation of the D/L-allo-Thr standards [D/L-allo-Thr (6.3)], but ruled out the presence of D/L-Thr in the hydrolysate of 1. Subsequently, acid hydrolysate of 1 was derivatized with FDAA and analyzed by RP-HPLC, which allowed for the assignment of D-configuration for the two allo-Thr residues. To facilitate configurational analysis of the remaining stereocenters in 1, X-ray crystallographic study was attempted. Unfortunately, under variety of conditions, theopapuamide (1) only yielded fine crystals that were unsuitable for X-ray diffraction studies. In summary, theopapuamide (1) was cytotoxic against CEM-TART and HCT-116 cell lines with EC50 values of 0.5 μM and 0.9 μM respectively. Theopapuamide (1) contains a high degree of D-amino acids and N-methylated amino acids along with several other modified amino acid residues of non-ribosomal origin. Theopapuamide (1), neamphamide A and callipeltins share the same basic structural skeleton of a 7- or 8-residue ring moiety formed by cyclization through a β-hydroxyl group of a Thr residue and a poly-substituted side chain linked to the amino terminus of a 3,4-diMeGln residue. Since the 3,4-diMeGln residue is conserved in all of the above mentioned HIV-inibitory marine depsipeptides (callipeltin A, neamphamide A, and Papuamides AB), it has been postulated by Acevedo et al.5c that the 3,4-diMeGln residue may play a role in their biological activity. However, theopapuamide (1), which also carries the atypical 3,4-diMeGln residue, failed to show any appreciable HIV-activity. The major difference between theopapuamide (1) and related HIV-inhibitory peptides is the absence of a β-methoxytyrosine residue in 1. Therefore, this work indicates the potential significance of the β-methoxytyrosine residue for biological activity.