Euryjanicins E–G, poly-phenylalanine and poly-proline cyclic heptapeptides from the Caribbean sponge Prosuberites laughlini

Abstract

A new investigation of the active sponge extracts of Prosuberites laughlini collected off the West coast of Puerto Rico has yielded three new cyclic heptapeptides, namely euryjanicins E (1)–G (3), containing multiple phenylalanine and proline residues. In CDCl₃ solution, each euryjanicin F (2) and G (3) exists as an inseparable complex mixture of conformational isomers. The molecular structures of 1–3 were elucidated by a combination of chemical degradation, extensive ESI-MS/MSⁿanalyses, and 2D NMR methods. The elucidation of the absolute configuration was achieved by HPLC following analysis of the acid hydrolysates after derivatization with Marfey’s reagent. When assayed against the National Cancer Institute 60 tumor cell line panel, the new cyclic peptides did not display significant in vitro cytotoxicity.

1. Introduction

Marine invertebrates, especially sponges and tunicates, represent an abundant resource of novel bioactive metabolites including terpenoids, alkaloids, macrolides, polyethers, and nucleoside derivatives.¹ Marine sponges in particular, are a rich source of non-ribosomal cyclic peptides with seven to ten amino acids that often exhibit a wide range of desirable pharmacological properties, such as antibiotic, anti-inflammatory, and cytotoxic activity.² Non-ribosomal cyclic peptides are often of considerable interest because they can also be used as models for studies of the structural features of proteins.^3–6 A particularly important class of marine cyclic peptides is the so called proline-rich compounds. Among the 20 naturally occurring amino acids, proline is the only one in which the side chain atoms form a pyrrolidine ring with the backbone atoms. This ring structure leads to some conspicuous properties of proline, namely, it induces conformational constraints among the atoms in the pyrrolidine ring, and it is the reason for the slow isomerization between cis/trans conformations and for the secondary structure preferences of poly-proline sequences.⁷ Remarkably, the cis/trans preference of proline–AA peptide bonds (where AA is any amino acid) varies in different solvent environments.⁸ Cyclic peptides having between seven to ten amino acids of which two to four are proline residues are typically designated as “proline-rich”.⁹ Similarly, “phenylalanine-rich” cyclic peptides have been associated with strong anticancer and antimicrobial activity.¹⁰ Some examples of this class of natural products, expressly isolated from marine sponges, are the hymenamides B¹¹ and E,¹² phakellistatin 11,¹³ axinellin A,¹⁴ and stylissamides C¹⁵ and F.¹⁶

In 2006, the marine sponge Prosuberites laughlini (Diaz, Alvarez & van Soest, 1987)¹⁷ (phylum: Porifera; class: Demospongiae; order: Hadromerida; family: Suberitidae) was collected during an underwater expedition from a fringing reef off the northwestern coast of Puerto Rico, near the town of Aguadilla (18°27’46.90’N, 67°10’06.11”W). The sponge specimens were triturated and repeatedly extracted with MeOH, and the combined extracts concentrated and partitioned between H₂O and EtOAc followed by n-butanol. In-house biological screening of both extracts quickly revealed that the EtOAc fraction displayed in vitro cytotoxicity. Routine ¹H and ¹³C NMR analysis revealed that the EtOAc extract comprised a complex mixture of cyclic peptides, which were subsequently characterized as euryjanicins A–D^18,19 and dominicin.²⁰ Oddly, while the EtOAc extract was active, none of the individual peptides could account for the primary level of cytotoxicity.¹⁹ This led us to hypothesize that this sponge might contain trace amounts of highly toxic peptides. To test this hypothesis more than twice as much sponge material was re-collected from the original location and extracted. Upon re-investigating the chemical composition of the new extracts, we now report on the isolation, structure elucidation, and biological evaluation of euryjanicins E (1)–G (3), three sparse cyclic peptides rich in phenylalanine and proline residues.

2. Results and Discussion

2.1. Extraction and isolation of cyclic peptides

Freeze-dried samples of P. laghlini (1.6 kg) were repeatedly extracted with MeOH and the extracts were combined and concentrated in vacuo. The crude MeOH-soluble material was subsequently purified by liquid/liquid partitioning between EtOAc, n-BuOH and H₂O. The EtOAc phase was concentrated to dryness to give an oily residue consisting of substances of a peptidic nature. TLC-guided fractionation of the EtOAcsoluble components via flash chromatography over Si gel using a gradient of increasing polarity with MeOH/NH₃ in CHCl₃ (98:2–1:1) yielded 12 fractions. Fractions eluted with 5–10% of MeOH/NH₃ in CHCl₃ were further purified by reversed-phase HPLC on a C-18 Spheri-5 column eluting with MeOH/H₂O to yield pure euryjanicin E (1) (7.2 mg; 0.0005%), euryjanicin F (2) (10.8 mg; 0.0007%) and euryjanicin G (3) (5.2 mg; 0.0003%), along with the same previously isolated cyclic peptides.^18–20

2.2. Structure elucidation of euryjanicin E

Compound 1, the most polar compound, was isolated as an optically active white semi-solid with a molecular formula of C₄₄H₆₀N₈O₈ as deduced from HRESI-MS [M+H]⁺ (calcd 829.4607), requiring 19 sites of unsaturation. The presence of intense absorption peaks at 3317 and 1645 cm⁻¹ in the IR, corresponding to amino and carbonyl groups, respectively, along with some characteristic peaks in the ¹H- and ¹³C-NMR spectra (Table 1), strongly suggested a peptide like compound. In as much as eight amide carbonyls [δ 172.2, 172.1, 2 × 171.7, 171.6, 170.9, 170.8, and 170.4] but only seven peaks for amino acid α-carbons [δ 61.4, 61.2, 61.2, 56.2, 54.1, 52.8, 49.2] were detected in the ¹³C NMR spectrum, we surmised that 1 was a heptapeptide bearing either a glutamine or an asparagine residue. Additionally, eight resonances about the alkene region hinted at the presence of two monosubstituted phenyl rings [δ 135.8, 135.3, 129.6, 129.2, 128.9, 129.8, 127.5, 127.3], which suggested the occurrence in 1 of two phenylalanine residues. Analysis of the COSY, TOCSY and HSQC–TOCSY NMR data revealed only five peptide-bond secondary amide proton signals thus implying the presence of two proline units in the heptapeptide. Further analysis of these data allowed us to quickly assemble seven independent spin systems. TOCSY correlations between NHIle¹, αHIle¹, βHIle¹, γH₂Ile¹, δH₃Ile¹, and εH₃Ile¹ clearly established the presence of an isoleucine residue. Likewise, another spin system of the type –CH-CH(CH₃)-CH₂-CH₃ was observed between NHIle⁵, αHIle⁵, βHIle⁵, γH₂Ile⁵, δH₃Ile⁵, and εH₃Ile⁵ that was attributed to another isoleucine unit. Moreover, the COSY spectrum allowed us to established two independent spin systems of the type X-CH-CH₂-CH₂-CH₂-X’ that were ascribable to a pair of proline residues. The first one, Pro², was profiled from a combination of COSY (αHPro², βH₂Pro², γH₂Pro², and δH₂Pro²) and TOCSY correlations (αHPro² and γH₂Pro²), while Pro⁶ was similarly outlined from COSY correlations between αHPro⁶, βH₂Pro⁶, γH₂Pro⁶, and δH₂Pro⁶. Of the three spin systems remaining, two were assigned as phenylalanine residues from the strong COSY correlations between NHPhe³, αHPhe³, and βH₂Phe³, and those of NHPhe⁷, αHPhe⁷, and βH₂Phe⁷. As for the final spin system in 1, we attributed it to an asparagine unit from TOCSY correlations between αHAsn⁴ and βHAsn⁴ and from subsequent analysis of the combined HMBC (Table 1) and electrospray tandem mass spectrometry (ESI-MS/MSⁿ) data (see Supplementary data).

Table 1.

¹H NMR (500 MHz), ¹³C NMR (125 MHz), TOCSY, HMBC, and ROESY correlations for euryjanicin E (1)^a

Amino acid	Position	δC (mult)b	δH, mult (J in Hz)	TOCSY	HMBCc	ROESYd
L-isoleucine1	CO	171.7, C			α
	α	56.2, CH	4.13, t (9.2)	β, γ	β, γ	NHPhe7, β
	β	31.5, CH	2.02, m	α, γ, δ	α, γ, δ	α, γ, ε
	γ	24.3, CH2	1.41, m 1.05, m	a, β, δ, ε	α, β, δ, ε	β
	δ	9.8, CH3	0.78, t (7.5)	β, γ, ε	β, γ, ε	γ
	ε	16.1, CH3	0.87, d (6.6)	γ, δ	δ	β
	NH		7.94, d (8.5)	α		α, αPhe3
L-proline2 cis	CO	172.2, C			α, β, NHIle1
	α	61.4, CH	3.71, d (7.9)	β, γ	β, γ	αPhe3, β
	β	31.3, CH2	2.02, m 1.04, m	α, γ, δ	α, γ, δ	α,γ
	γ	21.7, CH2	1.68, m 1.52, m	α, β, γ, δ	α, β, δ	β, δ
	δ	45.9, CH2	3.50, m 3.32, m	β, γ	β, γ	γ
L-phenylalanine3	CO	171.6, C			α,β
	α	52.8, CH	4.67, dd (7.8, 15.9)	β	β,γ	αPro2, NHIle1, β
	β	37.9, CH2	3.00, m	α	α, γ	α, γ
	γ	135.8, C			β, ortho
	ortho	129.6, CH	7.21, m		γ, meta	meta
	meta	128.9, CH	7.32, m		ortho, para	ortho, para
	para	127.5, CH	7.28, m		meta, ortho	meta
	NH		7.46, d (6.0)	α		α, αAsn4
L-asparagine4	CO	172.1, C			β, NH, NHPhe3
	α	49.8, CH	4.38, m	β	β	NHPhe3
	β	37.0, CH2	2.99, dd (5.8, 15.8) 2.84, dd (5.8, 15.8)	α	α	α
	γ	171.7, C
	γ NH		6.48, s			γ’ NH
	γ’ NH		5.34, s			γ NH
	NH		7.69, d (5.2)	α		α, NHIle5
L-isoleucine5	CO	170.4, C			NH, NHAsn4
	α	61.2, CH	3.93, t (8.5)	β, γ	β, γ	β
	β	35.8, CH	1.96, m	α, γ, δ	a, γ, δ	α, γ, ε
	γ	25.6, CH2	1.48, m 1.12, m	α, β, δ, ε	a, β, δ, ε	β
	δ	10.8, CH3	0.81, t (7.4)	β, γ, ε	β, γ, ε	γ
	ε	15.8, CH	0.90, d (6.8)	γ, δ	δ	β
	NH		8.18, d (8.2)	α		α, αβhe7, NHAsn4
L-proline6 cis	CO	170.9, C			a, β, NHIle
	α	61.2, CH	3.76, t (7.6)	β, γ	β, γ	αPhe7,β
	β	30.4, CH2	2.13, m 0.85, m	α, γ, δ	a, γ, δ	α, γ
	γ	21.6, CH2	1.65, m 1.43, m	α, β, γ, δ	a, β, δ	β, δ
	δ	46.2, CH2	3.51, m 3.36, m	β, γ	β, γ	γ
L-phenylalanine7	CO	170.8, C			β
	α	54.1, CH	4.53, m		β, γ	αPro6 , β
	β	38.4, CH2	3.03, m		a, γ	α, γ
	γ	135.3, C			β, ortho
	ortho	129.2, CH	7.23, m		γ, meta	meta
	meta	128.8, CH	7.32, m		ortho, para	ortho, para
	para	127.3, CH	7.28, m		meta, ortho	meta
	NH		7.50, d (3.1)	α		α, αIle1

^aSpectra were recorded in CDCl₃ at 25 °C. Chemical shift values are in ppm relative to the residual CHCl₃ (7.26 ppm) or CDCl₃ (77.0 ppm) signals.

^b13C NMR multiplicities were obtained from DEPTQ experiments.

^cHMBC mixing time = 50 ms.

^dROESY mixing time = 200 ms.

The amino acid composition of euryjanicin E (1) was corroborated by HPLC analysis of the acid hydrolysate obtained from the reaction of the peptide with 6 N HCl followed by Marfey’s derivatization with FDAA.²¹ One mol of aspartate (Asp) and two moles each of phenylalanine (Phe), proline (Pro), and isoleucine (Ile) residues were generated and determined to have the l-configuration at the α-carbon by comparison to amino acid standards. These seven amino acids, minus two hydrogens, accounted for all the atoms in the molecular formula of euryjanicin E (1) and all but one site of unsaturation. Hence, the degree of unsaturation leftover could be accounted for if 1 were cyclic, a contention supported by its high solubility in organic solvents and the fact that it tested negative to a ninhydrin test.

The amino acid sequence in 1 was elucidated by interpretation of HMBC and ROESY data as illustrated in Fig. 1, and was later confirmed by ESI-MS/MSⁿ. The identity of all of the carbonyls in the cyclic heptapeptide was established from HMBC correlations to the corresponding protons in the α, β or NH positions within each residue. For instance, COPro² correlations to αHPro², βH₂Pro², and NHIle¹, placed Pro² adjacent to Ile¹. The COAsn⁴ cross peaks to NHAsn⁴ and NHPhe³, together with correlations between COIle⁵ and NHAsn⁴ indicated that Asn⁴ was flanked by Phe³ and Ile⁵. This generated two amino acid fragments, namely, Pro²-Ile¹ and Phe³-Asn⁴-Ile⁵. In turn, COPro⁶ correlated to αHPro⁶, βH₂Pro⁶ and NHIle⁵, conclusively elongating the latter amino acid chain to Phe³-Asn⁴-Ile⁵-Pro⁶. Key cross peaks in the ROESY spectrum between αHIle¹ and NHPhe⁷ and those of αHPhe⁷ and αHPro⁶ allowed the merging of fragments Pro²-Ile¹-Phe⁷ and Phe³-Asn⁴-Ile⁵-Pro⁶-Phe⁷. Additional cross peaks between αHPro² and αHPhe³ allowed us to bring about the entire amino acid sequence of euryjanicin E (1) as cyclo-(-Phe-Pro-Ile-Asn-Phe-Pro-Ile-).

Fig. 1.

Key HMBC (C→H) and ROESY correlations (dashed arrows) for euryjanicin E (1) and euryjanicin F (2).

The amide bond conformations of Pro² and Pro⁶ within 1 were determined to be cis on the basis of the differential value of ¹³C chemical shifts of C_β and C_γ for each proline residue (Δδ_βγ), ^22,23 namely, 9.6 and 8.8 ppm, respectively, and the strong ROESY cross peaks between αH of Pro² and Pro⁶ with the αH of the adjacent amino acids (i.e. αHPro²/αHPhe³ and αHPro⁶/αHPhe⁷). Hence, the absolute stereostructure for the solution conformation of euryjanicin E (1) in CDCl₃ at 25 °C was established as cyclo-(-Phe-cis-Pro-Ile-Asn-Phe-cis-Pro-Ile-).

2.3. Structure elucidation of euryjanicin F

Compound 2 was obtained as a pure white amorphous powder. The molecular weight of euryjanicin F (2) (m/z 884.4756 [M+Na]⁺) was established from HES-MS and indicated the molecular formula C₄₉H₆₃N₇O₇Na, implying 22 degrees of unsaturation. The ¹H NMR spectrum in CDCl₃, like that of euryjanicin E (1), exhibited tell-telling resonances indicative of a compound of a peptidic nature. However, the H- and ¹³C NMR spectra of 2 were visibly more complex than those of 1 as many, if not all, of the signals were doubled or broadened, suggesting that two interconverting solution conformations of comparable stability (1:1 ratio) existed at room temperature. Attempts to separate the two conformers by reversed-phased HPLC under a variety of experimental conditions (columns, solvent compositions, and flow rates) invariably led to a single sharp peak and a “clean” molecular ion in the HPLC-ESI-MS/MSⁿ. Neither changing the solvent (MeOH-d₄, DMSO-d₆, MeCN-d₃, acetone-d₆, and pyridine-d₅), heating the sample to ∼60 °C or cooling it to −15 °C, nor adjusting the pH by addition of TFA or Et₃N, shifted the equilibrium in favor of a single conformer. Notwithstanding, when recorded in pyridine-d₅, the NMR signals were doubled in a 3:1 ratio making spectral interpretation more amenable to a structure elucidation study.²⁴

By combining ¹³C and DEP135 NMR spectra with data obtained from HSQC experiments, it was established that the major solution conformer of euryjanicin F had 49 well-resolved carbon resonances. Of these, seven resonances possessed chemical shifts appropriate for amide carbonyls [δ 173.4, 171.6, 171.5, 171.4, 171.1, 170.4, 169.9], twelve were for three monosubstituted phenyls [δ 137.9, 136.3, 136.1, 130.9, 129.9, 129.7, 129.1, 129.0, 128.7, 127.7, 127.6, 127.0], and seven had chemical shifts for amino acid α-carbons [δ 62.0, 61.7, 61.4, 56.8, 56.7, 55.5, 53.7] consistent with a heptapeptide with three phenylalanine residues (Table 2). Further inspection of the ¹H NMR spectrum revealed five amide NH signals indicative of a peptide containing two proline units. Careful interpretation of the HSQC, COSY, TOCSY, and HSQC-TOCSY NMR spectra led to the identification of the following seven amino acids: 2 x Pro, 2 x Ile, and 3 x Phe. As these amino acid residues accounted for only 21 of the 22 sites of unsaturation, and no further evidence was found for a terminal amino or carboxylic acid functionality, it stands to reason that heptapeptide 2 was also cyclic. The presence of only proline, isoleucine, and phenylalanine in this cyclic peptide was confirmed by acid hydrolysis of 2 followed by Marfey’s HPLC analysis of the hydrolysate,²¹ which also established the absolute configuration of all the amino acids to be l.

Table 2.

¹H NMR (500 MHz), ¹³C NMR (125 MHz), TOCSY, HMBC, and ROESY correlations for euryjanicin F (2)^a

Amino acid	Position	δC (mult)b	δH, mult (J in Hz)	TOCSY	HMBCc	ROESYd
L-phenylalanine1	CO	169.9, C			β, NH
	α	55.5, CH	4.58, m		β, γ	ΑPro7 , β
	β	37.9, CH2	3.06, t (12.2) 2.79, dd (4.2, 12.1)		α, γ	α, γ
	γ	136.1,C			β, Ortho
	ortho	129.9, CH	7.19, m	meta	γ, meta	meta
	meta	129.1, CH	7.20, m	ortho	ortho, para	ortho, para
	para	127.6, CH	7.22, m		meta, ortho	meta
	NH		10.16, s			αPhe2, α, β
L-phenylalanine2	CO	171.6, C			NHPhe1
	α	56.7, CH	4.83, m		β, γ	β
	β	38.8, CH2	3.56, m 3.44, m		α, γ	α, γ
	γ	137.9, C			β, ortho
	ortho	130.9, CH	7.71, d (7.5)	meta	γ, meta	meta
	meta	129.0, CH	7.27, m	ortho	ortho, para	ortho, para
	Para	127.0, CH	7.22, m		meta, ortho	meta
	NH		8.42, s			αIle3
L-isoleucine3	CO	171.5, C			α
	α	61.4, CH	4.76, t (9.0)	β, γ	β,γ	β
	β	37.7, CH	3.20, dd (7.8, 14.8)	α, γ, δ	α, γ, δ	α, γ, ε
	γ	25.6, CH2	1.67, m 1.27, m	α, β, δ, ε	α, β, δ, ε	β
	δ	10.9, CH3	0.90, t (7.3)	β, γ,ε	β, γ, ε	γ
	ε	16.0, CH3	1.11, d (6.7)	γ, δ	δ	β
	NH		8.43, d (2.6)			αPhe5
L-proline4 cis	CO	170.4, C
	α	61.7, CH	3.98, d (7.0)	β, γ	β, γ	αPhe5 , β
	β	29.0, CH2	2.30, dd (11.7, 6.5) 0.49, m	α, γ, δ	α, γ, δ	α, γ
	γ	21. 9, CH2	1.56, m	α, β, γ, δ	α, β, δ	β, δ
	δ	45.9, CH2	3.77, m 3.41, m	β, γ	β, γ	γ
L-phenylalanine5	CO	171.4, C
	α	53.7, CH	4.94, m		β, γ, NH	αPro4 , β
	γ	38.4, CH2 136.3, C	3.16, t (12.2) 2.94, dd (5.4, 12.7)		α, γ β, ortho	α, γ
	ortho	129.7, CH	6.86, d (6.4)	meta	γ, meta	meta
	meta	128.7, CH	7.27, m	ortho	ortho, para	ortho, para
	Para	127.7, CH	7.22, m		meta, ortho	meta
	NH		10.82, d (6.3)			αIle6, α, β
L-isoleucine6	CO	173.4, C			α, NHPhe5
	α	56.8, CH	4.88, t (10.4)	β, γ	β, γ	β
	β	34.3, CH	2.54, m	α, γ, δ	α, γ, δ	α, γ, ε
	γ	25.0, CH2	1.50, m	α, β, δ, ε	α, β, δ, ε	β
	δ	9.3, CH3	0.68,t (7.4)	β, γ, ε	β, γ, ε	γ
	ε	15.8, CH3	1.03, d (6.7)	γ, δ	δ	αPhe1 , α, β
	NH		9.78, d (9.3)
L-proline7 cis	CO	171.1, C			αIle6, NHIle6
	α	62.0, CH	3.79, d (7.5)	β, γ	β, γ	αPhe1,β
	β	30.9, CH2	2.44, dd (5.9, 12.1) 0.97, m	α, γ, δ	α, γ, δ	α, γ
	γ	22.5, CH2	1.51, m	α, β, γ, δ	α, β, δ	β, δ
	δ	46.7, CH2	3.67, m	β, γ	β, γ	γ

^aSpectra were recorded in pyridine-d₅ at 25 °C. Chemical shift values are in ppm relative to the residual pyridine (8.71 ppm) or pyridine-d₅ (149.9 ppm) signals.

^b13C NMR multiplicities were obtained from DEPTQ experiments.

^cHMBC mixing time = 50 ms.

^dNOESY mixing time = 200 ms.

Following identification of all of the amide protons in 2 from COSY correlations to their corresponding amino acid α-protons, the amino acid sequence for euryjanicin F (2) was easily established by combination of the ROESY and HMBC data. For instance, COPhe² gave a strong HMBC cross peak with NHPhe¹, which in turn correlated to COPhe¹ and βH₂Phe¹, establishing the first fragment as Phe¹ -Phe². HMBC correlations COPro⁷/αHIle⁶, COIle⁶/NHPhe⁵, and COPro⁷/NHIle⁶ provided the segment Phe⁵ -Ile⁶ -Pro⁷. Further analysis of the ROESY spectrum revealed a strong correlation between αHPhe¹/αHPro⁷ that allowed us to interconnect the latter two segments thus establishing the sequence -Phe⁵ -Ile⁶ - Pro⁷ -Phe¹ -Phe² -. Additional HMBC and ROESY cross peaks between αHPhe⁵/αHPro⁴ and NHPhe²/αHIle³ brought about the complete amino acid sequence of euryjanicin F (2) as cyclo-(- Phe-Phe-Ile-Pro-Phe-le-Pro-) (see Table 2 and Fig. 1).²⁴

Close inspection of Δδ_βγ in the ¹³C NMR spectrum for the two conformational variants of euryjanicin F in pyridine-d₅ revealed the geometry for each proline peptide bond.²⁵ In the major solution conformer 2, both proline amide bonds adopt the cis geometry (Δδ_29.0–21.9 = 7.1 ppm for Pro⁴; Δδ_30.9–22.5 = 8.4 ppm for Pro⁷). This geometry was further supported by the ROESY cross peaks αHPhe¹/αHPro⁷ and αHPhe⁵/αHPro⁴ (Fig. 1). In the minor solution conformer (not shown), one of the proline amide bonds (Pro⁴) adopts the trans geometry (Δδ_29.9–25.8 = 4.1 ppm) whereas the same bond of Pro⁷ assumes the cis geometry (Δδ_314–22.2 = 9.2 ppm).²⁶ Additional support for the latter geometries stems from ROESY cross peaks NHIle³/HαPro⁴ and HαPhe¹/HαPro⁷ plus the conspicuous absence of cross peaks between αHPhe⁵/αHPro⁴. Interestingly, the sequences of euryjanicin E (1) and euryjanicin F (2) differ by only one amino acid, namely, the fragment Phe-Asn in 1 is replaced by Phe-Ile in 2.

2.4. Structure elucidation of euryjanicin G

Peptide 3 was the least abundant peptide component isolated from the sponge, with an accurate mass measurement of m/z 868.4421 ([M+Na]⁺ Δ 4.7 mDa from that calculated for C₄₈H₅₉N₇O₇). Like euryjanicin F (2), euryjanicin G (3) generated a single molecular ion in the HRESI-MS and always gave a sharp well-resolved peak when analyzed by reversed-phase HPLC under a variety of solvent systems. Despite these indications of homogeneity, the ¹H and ¹³C NMR spectra of 3 in CDC_l3 were difficult to analyze due to the complexity caused by multiple conformations resulting in poor resolution. Subsequently, a single set of well-resolved resonances could be obtained at 25 °C when pyridine-d₅ was used as the solvent, revealing that 3 can exist as a single solution conormer.²⁷

The structure of euryjanicin G (3) was elucidated by interpretation of 1D and 2D NMR data in pyridine-d₅ (Table 3). The ¹H NMR spectrum of 3 showed four amide protons at δ_H 10.42 (1H, br s, NHPhe⁵), 9.79 (1H, d, J 9.1 Hz, NHIle¹), 9.73 (1H, d, J = 9.7 Hz, NHPhe⁷), and 7.35 (1H, d, J = 4.5 Hz, NHPhe³), while the ¹³C NMR spectrum contained seven carbonyl signals at δ_C 173.0, 172.8, 172.0, 171.9, 171.4, 170.4, and 169.3 along with seven amino acid αC’s (δ_C 62.3, 61.3, 59.0, 56.5, 55.0, 54.6, and 53.2). The presence of 13 methylene carbons along with 23 sites of unsaturation suggested a substantial number of proline and phenylalanine moieties constituting the structural backbone. Further analysis of the ¹H–¹H COSY, HSQC, HMBC, and HSQC–TOCSY spectra revealed spin systems corresponding to three phenylalanines, one isoleucine, and three proline residues (Table 3). The amino acid composition was confirmed by HPLC analyses of the acid hydrolysate, and since only 22 degrees of unsaturation could be accounted for by the functionality present in the seven individual amino acids, it was apparent that 3 too was a cyclic peptide. Marfey’s analysis was used to establish the absolute congiguration as L for all the constituent amino acids of euryjanicin G (3).²¹

Table 3.

¹H NMR (500 MHz), ¹³C NMR (125 MHz), TOCSY, HMBC, and ROESY correlations for euryjanicin G (3)^a

Amino acid	Position CO	δC (mult)b	δH, mult (J in Hz)	TOCSY	HMBCc	ROESYd
L-isoleucine1		173.0, C
	α	55.0, CH	4.96, t (9.3)		NH	αPro2
	β	38.3, CH	1.89, m	β, γ	β, γ	β
	γ	24.0, CH2	1.55, m 1.16, m	α, γ, δ	α, γ, δ	α, γ, ε
	δ	11.1, CH3	0.69, t (7.3)	α, β, δ, ε	α, β, δ, ε	β
	ε	15.3, CH3	0.83, d (6.8)	β, γ, ε	β, γ, ε	γ
	NH		9.79, d (9.1)	γ, δ	δ	αPhe7, β
L-proline2 cis	CO	171.4, C			α, NHPhe3
	α	62.3, CH	4.83, d (7.9)	β, γ	β, γ	αIle1
	β	32.3, CH2	2.23, m	α, γ, δ	α, γ, δ	α, γ
	γ	22.5, CH	2.00, m	α, β, γ, δ	α,β, δ	β, δ
			1.69, m
	δ	47.0, CH2	3.79, m	β, γ	β,γ	γ
			3.53, m
L-phenylalanine3	CO	169. 3, C			α,β, NH
	α	53.2, CH	5.30, m			αPro4
	β	37.9, CH2	3.47, m		α, γ	α, γ
			3.22, m
	γ	136.5, C			β, Ortho
	ortho	129.7, CH	7.15, m	meta	γ, meta	meta
	meta	128.5, CH	7.17, m	ortho	ortho, para	ortho, para
	para	127.2, CH	7.19, m		meta, ortho	meta
	NH		7.35, d (4.5)
L-proline4 cis	CO	172.8, C			β, NHPhe5
	α	59.0, CH	4.69, d (8.0)	β,γ	β, γ	αPhe3
	β	31.2, CH2	2.11, m 1.88, m	α, γ, δ	α, γ, δ	α, γ
	γ	22.2, CH2	2.28, m 1.69, m	α, β, γ, δ	α, β, δ	β, δ
	δ	47.6, CH2	4.03, m	β, γ	β, γ	γ
			3.62, m
L-phenylalanine5	CO	170.4, C			α, β, NH
	α	54.6, CH	5.07, m			αPro6
	β	37.2, CH2	3.15, m		α, γ	α, γ
			3.03, m
	γ	136.2, C			β, ortho
	ortho	129.6, CH	7.22, m	meta	γ, meta	meta
	meta	128.4, CH	7.17, m	ortho	ortho, para	ortho, para
	para NH	126.4, CH	7.19, m 10.42, s		meta, ortho	meta αPro4, α, β
L-proline6 cis	CO	172.0, C			α, NHPhe7
	α	61.3, CH	3.58, d (8.5)	β, γ	β, γ	αPhe5
	β	31.1, CH2	2.38, m 1.13, m	α, γ, δ	α, γ, δ	α, γ
	γ	22.0, CH2	1.69, m 1.29, m	α, β, γ, δ	α, β, δ	β, δ
	δ	46.8, CH2	3.53, m 3.49, m	β, γ	β, γ	γ
L-phenylalanine7	CO	171.9, C			α, β, NH
	α	56.5, CH	4.99, m		NH
	β	37.2, CH2	3.45, m		α, γ	α, γ
			3.17, m
	γ	139.9, C			β, ortho
	ortho	130.4, CH	7.07, m	meta	γ, meta	meta
	meta	129.1, CH	7.29, m	ortho	ortho, para	ortho, para
	par a	127.3, CH	7.19, m		meta, ortho	meta
	NH		9.73, d (8.3)			αPhe5, β

^aSpectra were recorded in pyridine-d₅ at 25 °C. Chemical shift values are in ppm relative to the residual pyridine (8.71 ppm) or pyridine-d₅ (149.9 ppm) signals.

^b13C NMR multiplicities were obtained from DEPTQ experiments.

^cHMBC mixing time = 50 ms.

^dNOESY mixing time = 200 ms.

The amino acid sequence in 3 was established by combination of HMBC, ROESY, and ESI-MS/MSⁿ data. The fragment -Pro² -Phe³ - was assigned from the HMBC correlations of COPro²/NHPhe³ and COPhe³/NHPhe³ Furthermore, cross peaks for COPro⁴/NHPhe⁵ and COPhe⁵/δH₂Pro⁶ determined the partial sequence -Pro⁴-Phe⁵-Pro⁶-while the correlations between COPro⁶/NHPhe⁷ and COPhe⁷/NHIle¹ lengthened the chain to Pro⁴-Phe⁵-Pro⁶-Phe⁷-Ile¹. Strong ROESY cross peaks between αHPro²/αHIle¹ made it possible to link fragments –Pro²-Phe³- and Pro⁴-Phe⁵-Pro⁶-Phe⁷-Ile¹. Overall, this sequence was supported by the ROESY correlations between NH_(i) and αH_(i–1) and, in case of proline residues, αH_(i)Pro and αH_(i–1). Pivotal NOE cross peaks between αHPhe³ and αHPro⁴ allowed us to close the ring.

As before, the amide bond configuration of Pro², Pro⁴, and Pro⁶ was determined to be cis on the basis of Δδ_βγ in all three proline residues, which was ≥ 9.0 ppm. Furthermore, pivotal ROESY cross peaks between aH’s of Pro², Pro⁴, and Pro⁶ and the αH’s of the adjacent amino acid, which are present in the cis conformation but absent in the trans-proline,²⁸ and the strong correlations between αHPro²/αHIle¹, αHPro⁴/αHPhe³, and αHPro⁶/αHPhe⁵, confirmed that all three proline amide bonds adopted a cis geometry. Thus, the entire amino acid sequence for 3 was established as cyclo-(Ile¹-cis- Pro²-Phe³-cis-Pro⁴-Phe⁵-cis-Pro⁶-Phe⁷). Remarkably, the sequences of euryjanicin G (3) and stylissamide C, a cyclic heptapeptide reported in 2007 from the Bahamian sponge Stylissa caribica, differ by only one amino acid.¹⁵ While all three proline amide bonds in these two compounds adopt the cis geometry, Phe⁵ in euryjanicin G is replaced by tyrosine in stylissamide C.

2.5. Biological activity

Euryjanicins 1–3 were tested at a single high dose (10 µM) in the full NCI 60 cell panel, but none satisfied the NCI’s predetermined threshold inhibition criteria for evaluation against the 60 cell panel at five concentration levels. Most of the cells displayed growths above 90%, with the exception of leukemia cell lines MOLT-4 and RPMI-8226, wherein euryjanicin F (2) elicited growth percentages of 49% and 43%, thus implying growth suppressions of 51% and 57%, respectively.

3. Conclusion

Like their predecessors, the new peptides could not account for the incipient cytotoxicity of the primary extracts.^{18, 19} This may be due to changes in the conformation of the cyclic peptides during the isolation process.²⁹ Compounds 1–3 are characterized by the co-occurrence of two to three proline and phenylalanine residues, one or two strongly hydrophobic isoleucine side chains and, in the case of 1, a polar uncharged asparagine residue. Remarkably, these metabolites are the first examples of cyclic peptides from a sponge to possess the same cis-Pro-Phe-Ile-cis-Pro-Phe segment. Each 2 and 3, exists as an inseparable complex mixture of conformational isomers about the proline residues, supporting the contention that proline-rich cyclic peptides often adopt well-defined conformations in solution, most likely from distinctive folds, hydrogen bonding patterns, and solvent accessible surfaces that contribute to the independent stability of the dominant conformer.³⁰ To no small extend, the fact that compounds 1–3 display a remarkable analogy to several other sponge-derived proline-rich antitumor cyclopeptides ¹¹–¹⁴ as to their amino acid content and sequential homology, might suggest that the construction of a suitable pharmacophore model for rationalizing their structure-activity relationship is possible. Thus, cyclopeptides such as 1–3 isolated from marine sponges could ultimately become indispensable tools for understanding the role of prolyl cis⇆trans isomerization as a common means of switching between alternative functional states of folded proteins.²⁶

4. Experimental section

4.1. General procedures

Optical rotations were recorded with a Rudolph Autopol IV polarimeter. UV and IR spectra were recorded with Shimadzu UV-2401P and Nicolet Magna IR 750 spectrometers, respectively. ID and 2D NMR spectra were recorded on a Bruker DRX-500 or a Bruker AV-500 FT-NMR spectrometer. HPLC was performed on an Agilent 1260 infinity equipped with an Agilent 1260 photodiode array detector. Column chromatography was performed on Analtech Si gel (35–75 mesh) and monitored by TLC analysis carried out on Analtech glass precoated Si gel plates and visualized using UV light (λ = 254 nm) or I₂ vapors. The percentage yield of each compound is based on the weight of the dry sponge.

4.2. Animal material

The sponge Prosuberites laughlini (Diaz, Alvarez & van Soest, 1987)¹⁷ (phylum: Porifera; class: Demospongiae; order: Hadromerida; family: Suberitidae) was collected in April 2009 and identified by J. Vicente. A voucher specimen (No. PLAG09–02) is stored at the Chemistry Department of the University of Puerto Rico.¹⁹

4.3. Extraction and isolation of cyclic peptides

The freshly collected sponge was freeze-dried for 3 days, and the dried organism (1.6 kg) was repeatedly extracted with MeOH (20 L). The combined MeOH extracts were evaporated to dryness, and the resulting brown oil (196 g) was partitioned between EtOAc (5 × 1 L), nBuOH (5 × 1 L), and H₂O (1×2 L). The combined EtOAc extracts were concentrated in vacuo to give 18 g of a dark brown oil that was flash chromatographed over Si gel (600 g) using a gradient of increasing polarity with MeOH/NH₃ in CHC1₃ (98:2–1:1) as mobile phase and separated into 12 fractions on the basis of TLC and ¹H NMR analyses. Fraction 4 (2.8 g) was subsequently chromatographed over Si gel (100 g) using a gradient of increasing polarity with MeOH/NH₃ in CHC1₃ (98:2–4:1) as eluent resulting in 13 fractions. Subfractions 2–7 contained all of the cyclic peptides. Fraction 2 was purified by reversed-phase chromatography with a 5 mm × 250 mm Spheri-5 C18 column, 5 µm, with an isocratic solvent composition of (A) 42% H₂O and (B) 58% MeOH over 25 min (flow rate = 1 mL/min) with UV detection set at 220 nm to yield pure euryjanicin C (71 mg; 0.005%), euryjanicin D (80 mg; 0.005%), euryjanicin G (3) (5.2 mg; 0.0003%), and euryjanicin F (2) (10.8 mg; 0.0007%).¹⁹ Fraction 3 contained a mixture of euryjanicins C and D (143.0 mg, 0.009%). Fractions 4, 5 and 6 consisted of pure dominicin (40.0 mg, 0.003%), euryjanicin A (65.0 mg, 0.004%) and euryjanicin B (90.0 mg, 0.006%), respectively.^18–20 Purification of fraction 7 via CI8 reversed-phase HPLC using a 10 mm × 25 cm Ultrasphere ODS column, 5 µm, with 20% H₂O in MeOH yielded pure euryjanicin E (1) (7.2 mg; 0.0005%).

4.3.1. Euryjanicin E (1)

White semi- solid; [α]²⁰ _D = −78.0 (c 1.0, CHCl₁₃); IR (film) v_max 3317 (br), 3063, 3030, 2963, 2932, 2878, 1645 (br), 1526, 1437, 750, 702 cm⁻¹; UV (MeOH) λ_max (log ε) 192 (ε 4.8) nm; ¹H NMR (500 MHz, CDC1₃) and ¹³C NMR (125 MHz, CDC1₃) (see Table 1); HR(+)ESI-MS m/z [M+H]⁺ 829.4606 (calcd for C₄₄H₆₁N₈O₈, 829.4607).

4.3.2. Euryjanicin F (2)

White powder; [α]²⁰ _D =−102.0 (c 1.0, CHCl₃); IR (film) v_max 3312 (br), 3063, 3030, 2963, 2932, 2878, 1643 (br), 1520, 1447, 750, 702 cm⁻¹; UV (MeOH) λ_max (log ε) 203 (ε 4.7) nm; ¹H NMR (500 MHz, C5D5N) and ¹³C NMR (125 MHz, C₅D₅N) (see Table 2); HR(+)ESI-MS m/z [M+Na]⁺ 884.4756 (calcd for C₄₉H₆₃N₇O₇Na, 884.4681).

4.3.3. Euryjanicin G (3)

White semi-solid; [α]²⁰ _D =− 62.0 (c 1.0, CHCl₃); IR (film) v_max 3282 (br), 3062, 3030, 2956, 2852, 1714, 1658, 1633 (br), 1552, 1452, 1346, 1247, 1188, 748, 702 cm⁻¹; UV (MeOH) λ_max (log ε) 203.5 (ε 4.5) nm; ¹H NMR (500 MHz, C₅D₅N) and ¹³C NMR (125 MHz, C₅D₅N) (see Table 3); HR(+)ESI-MS m/z [M+Na]⁺ 868.4421 (calcd for C₄₈H₅₉N₇O₇Na, 868.4374).

4.4. Acid hydrolysis of euryjanicins E (1)–G (3)

Pure euryjanicins E–G (0.5 mg) were hydrolyzed in 0.5 mL of 6 N HC1 at 110 °C for 12 h in a 1.0 mL reaction vial. The cooled reaction mixture was evaporated to dryness and traces of HC1 were removed from the residual hydrolysate by repeated evaporation from H₂O (3 × 0.5 mL) using N₂ gas.

4.5. Absolute configuration of amino acids²¹

To a 4 mL vial containing 1 µmol of pure amino acid standards in 200 µL of H₂O was added 1 µmol of N-R-(2,4-dinitro-5-fluorophenyl)-l-alanine amide (l-FDAA) in 400 µL of acetone followed by 100 µL of 1 N NaHCO₃. The mixture was heated for 1 h at 40 °C. After cooling to rt, 100 µL of 2 N HC1 was added and the resulting solution was filtered through a small 4.5 mm filter and stored in the freezer until HPLC analysis. Half of each peptide hydrolysate mixture was dissolved in 200 µL of H₂O, and to this was added 1.5 µmol of L-FDAA in 400 µL of acetone followed by 100 µL of 1 N NaHCO₃. The derivatization reaction was carried out and worked up as described above. An 8 µL aliquot of the resulting mixture of l-FDAA derivatives was analyzed by reversed-phase HPLC. A 5 mm 250 mm Spheri-5 CI8 column, 5 µm, with a linear gradient of (A) 9:1 triethylammonium phosphate (50 mM, pH 3.0)/CH₃CN and (B) CH₃CN with 0% B at start to 40% B over 55 min (flow rate = 1 mL/min) was used to separate the l-FDAA derivatives with UV detection set at 340 nm. Each chromatographic peak was identified by comparing its retention time with the l-FDAA derivative of the pure l-amino acid standard and by co-injection. In all cases a peak at 33.9 min was observed, which was attributed to excess l-FDAA. Retention times (min) are given in parentheses: l-Pro (30.37), d-Pro (34.14), l-Ile (43.52), d-Ile (51.25), l-Phe (43.54), d-Phe (49.35), l-Asp (20.98), d-Asp (22.13).