The Arctic sponge Haliclona viscosa as a source of a wide array of 3-alkyl pyridine alkaloids

Abstract

The Arctic sponge Haliclona viscosa has proven to be a promising source for 3-alkyl pyridinium and 3-alkyl tetrahydropyridine alkaloids (3-APAs). H. viscosa was the origin of the first cyclic monomeric 3-APA (haliclocyclin F), the first cyclic trimeric 3-APA (viscosamine C), the first dimeric 3-APA amino acid adduct (viscosaline C), and the first 3‐alkyl tetrahydropyridine alkaloids with saturated alkyl chains (haliclamines C and D). This review focuses on the challenging structure elucidations of 3-APAs isolated from H. viscosa using a combination of NMR spectroscopy and mass spectrometry.

Introduction

Sponges from warm and tropical waters have drawn attention for more than 30 years because of their high potential in producing bioactive secondary metabolites. In fact, the most densely populated marine habitats are located in the tropics, which shaped the hypothesis that predation pressure is higher in the tropics than in temperate areas, and consequently, that tropical sponges are chemically better defended than temperate ones (Ruzicka and Gleason 2008). As a result, the chemistry of sponges from cold and Arctic waters is little-known, despite their capacity of producing a wide array of bioactive pyridine/pyridinium alkaloids. Of particular interest is the Arctic sponge Haliclona viscosa. The sponge has been revealed as the most prolific organism out of a total of 17 organisms in an investigation of sessile or slow moving Arctic invertebrate species in the Kongsfjord, Spitsbergen (Lippert et al. 2004). The study shows that the crude extract of H. viscosa was significantly feeding deterrent to the Arctic amphipod Anonyx nugax and to the North Sea starfish Asterias rubens which are both general predators in their respective habitats (Lippert et al. 2004). The investigation also demonstrated the inhibition of the growth of five bacterial strains that had been isolated from the vicinity of the sponge (Lippert et al. 2003; Timm et al. 2010) and of a number of bacteria used in general bioactivity screening assays (Volk and Köck 2004).

In this review we focus on the species H. viscosa, a sponge known to produce a wide array of 3-alkyl pyridinium and 3-alkyl tetrahydropyridine alkaloids (both termed 3-APAs). The structural diversity of these alkaloids will be described. A special emphasis is placed on structure elucidation issues which were solved by a combined approach using NMR spectroscopy and mass spectrometry.

The Arctic sponge Haliclona viscosa

Haliclona viscosa belongs to the order Haplosclerida, class Demospongiae. Sponges of this order are widely distributed around the world and can represent high biomass portions in shallow and deep water communities. Among taxonomists the order Haplosclerida is known to be one of the most difficult and ambiguous groups of the Demospongiae (de Weerdt 1985). Reasons for this are the few characters which are available for taxonomic identification, their variability (and thus assigned taxonomic weight), and the large number of species (de Weerdt 1986). This is especially true for most species of the genus Haliclona, where growth form is highly variable while spicule form and arrangement is highly homogeneous (de Weerdt 1985).

The species H. viscosa has been described by different authors, under different names and from various locations. A detailed taxonomic overview and revision of the species was prepared by de Weerdt (1985, 1986, 1987) and is accessible in the World Porifera Database (van Soest et al. 2011). Most reports on H. viscosa assign the habitat of the sponge to the coasts of France, the British Isles, Ireland, and the Mediterranean, where it ranges from the sublittoral downwards but can also be found in the intertidal area (Picton and Morrow 2010).

3-APAs in Haliclona species and Haliclona viscosa

3-Alkyl pyridine alkaloids are specific for the marine environment and have been isolated to date only from marine sponges. Interestingly, 3-APA-containing sponges are widely distributed throughout the world. The largest number of 3-APAs was reported from sponges of the order Haplosclerida with the genera Haliclona, Amphimedon, and Xestospongia. In this section a focus will be given in 3-APAs isolated from Haliclona species and, more specifically, from H. viscosa.

3-APAs consist of a pyridinium or tetrahydropyridine (THPy) moiety that is connected, usually in the 1- and/or 3-position, to an alkyl chain of variable length. These compounds exist as open, cyclic, or polycyclic structures. The open and cyclic compounds form monomers, dimers, trimers, and polymers (Figs. 1 and 2). The first 3-alkyl pyridinium alkaloid, halitoxin (1) (Fig. 1), a linear polymeric 3-APA with branched alkyl chains, was isolated in 1978 from the sponges H. rubens, H. viridis, and H. erina (Schmitz et al. 1978), species that were later reclassified to the genus Amphimedon. Additionally, the sarains (2) were identified in Reniera (=Haliclona) sarai (Cimino et al. 1986; van Soest et al. 2011). The first member of the structurally complex manzamines, manzamine A (3) was also isolated from Haliclona sp. in 1986 (Sakai et al. 1986). Haliclona sponges have also yielded the 1-oxaquinolizidine 3α-methylaraguspongine C (4) isolated in 1994 from the sponge H. exigua (Venkateswarlu et al. 1994) as well as the halicyclamine A (5) isolated from Haliclona sp. (Jaspars et al. 1994). All of the above mentioned compounds are cytotoxic and ichtyotoxic which has stimulated a search for similar compounds represented today by the large number of complex structures identified from five different sponge orders¹ from around the world.

Fig. 1

Selected 3-APAs, halitoxin (1), sarain-1 (2a), manzamine A (3), 3α-methylaraguspongine C (4), and halicyclamine A (5) isolated from Haliclona species

Fig. 2

Selected 3-APAs, haliclocyclins (6), cyclostellettamines (7), viscosamines (8), haliclamines (9), and viscosalines (10) isolated from Haliclona viscosa

To date, five different structural classes are known among the 3-APAs contained in the Arctic sponge H. viscosa (Fig. 2); the macrocyclic monomeric haliclocyclins (6) (Timm et al. 2008; Schmidt et al. 2011), the macrocyclic dimeric cyclostellettamines (7) (Timm 2007), the macrocyclic trimeric viscosamines (8) (Volk and Köck 2003), the macrocyclic dimeric haliclamines (9) (Volk et al. 2004; Schmidt et al. 2009; Cychon et al. 2012) which are the only representatives of 3-alkyl THPy’s in H. viscosa, and the linear dimeric viscosalines (10) (Volk and Köck 2004; Schmidt et al. 2012).

The n-butanol fraction of the crude extract of an Arctic H. viscosa (collected in 2000) contained a number of 3-alkyl pyridinium and 3-alkyl THPy alkaloids (3-APAs) that were identified as the responsible components for their antibacterial (Kelman et al. 2001) and antifungal activity (Fusetani et al. 1989), cytotoxicity (Fusetani et al. 1989), and feeding deterrency (Lippert et al. 2004). The difference among the members of each structural class is in the length of the alkyl chains and the number of double bonds therein. There is still a large number of 3‐APAs with unknown structures in this specimen of H. viscosa.

Structure elucidation of 3-alkyl pyridine alkaloids

The structure elucidation of the secondary metabolites isolated from H. viscosa represented a real challenge. Despite their apparent simplicity and their large number of protons, the structures of 3-APAs needed the combination of both NMR spectroscopy and mass spectrometry to finally be elucidated.

NMR spectroscopy

Dimeric 3-APAs revealed a major problem for the structure elucidation by NMR spectroscopy as more than 50 % of their protons have the same resonance frequency (located in one signal). As an example, the mass of haliclamine D (9d) m/z = 457.4522 [M + H]⁺, calculated as C₃₁H₅₇N₂, indicates 56 protons (if a neutral molecule is considered) whereas the 1D proton spectrum of 9d only shows 29 protons (Fig. 3) (Volk et al. 2004). This represents only about half of the protons (including protonated amines, 10.3 ppm) indicating that this compound is symmetric or pseudo-symmetric. The main problem in the structure elucidation of the dimeric compounds by NMR spectroscopy was that the two heterocycles were connected by saturated alkyl chains where only the first two positions in proximity to the pyridinium (or THPy) ring were distinguishable (Fig. 4). This hampered the assignment of the length of the alkyl chains. Thus, NMR is not ideally suited in the structure elucidation of these molecules (Figs. 3 and 4). As a result it is possible to determine by NMR spectroscopy whether 3-APA compounds contain a pyridine or a THPy moiety, however the number of such units and their chain lengths cannot be deduced. Therefore NMR spectroscopy should be used in combination with mass spectrometry including MS/MS fragmentation for an unambiguous structure elucidation of 3-APAs (Fig. 4). Moreover, as 3-APAs in the H. viscosa extract are structurally very similar, their separation and isolation with preparative HPLC is difficult or even impossible.

Fig. 3

400 MHz 1D ¹H-NMR spectrum of haliclamine D (9d) in DMSO-d ₆

Fig. 4

Schematic representation of the structure elucidation of haliclamine D (9d) using NMR spectroscopy in combination with mass spectrometry

Liquid chromatography mass spectrometry (LC–MS)

For this purpose, the advantage of mass spectrometry in contrast to NMR spectroscopy lies in its higher sensitivity and its applicability to non-pure sample fractions. The entire crude extract or single fractions can be subjected to HPLC separation coupled to high-resolution MS or MSⁿ fragmentation. Improved chromatographic peak separation leads to better compound differentiation and more specific mass spectra in HRMS. This facilitates the structure elucidation of unknown compounds (Fig. 5). Despite the application of improved chromatographic separation conditions, the large number of similar 3-APAs in the crude extract of H. viscosa hindered the complete baseline-separation of all chromatographic peaks; several of the 3-APAs still co-eluted or overlapped. 3-APAs of the same structural class but with different chain lengths could be separated well (Fig. 6). However 3-APAs associated with different structural classes often co-eluted in one peak. Although this resulted in superimposed mass spectra, compounds of different compound classes could still confidently be recognized by their exact mass, characteristic molecular ions, and fragmentation pattern.

Fig. 5

Comparison of the HPLC–MS chromatogram under standard (left) and improved (right) conditions of Haliclona viscosa collected in the year of 2000. Standard conditions: H₂O/MeCN gradient with 0.1 % TFA in both solvents, at 35 °C [0 min: 100/0; 40 min: 20/80; 45 min: 0/100 with a flow rate of 1 mL min⁻¹] with a Knauer Kromasil-100 C18 column (4.6 x 250 mm, 5 μm). Improved conditions: H₂O/MeCN gradient with 0.1 % HCOOH in both solvents, at 35 °C [0 min: 80/20; 25 min: 45/55; 27 min: 0/100 with a flow rate of 0.4 mL min⁻¹] with a Waters XTerra C18 column (3.0 x 150 mm, 3.5 μm)

Fig. 6

HPLC–MS chromatogram of the n-butanol fraction of H. viscosa (2000). Experimental conditions: H₂O/MeCN gradient with 0.1 % in both solvents, at 35 °C [0 min: 80/20; 25 min: 45/55; 27 min: 0/100 with a flow rate of 0.4 mL min⁻¹] with a Waters XTerra C18 column (3.0 x 150 mm, 3.5 μm)

The compound identification was then possible directly from a LC–MS or LC–MS/MS run. The identification relied on three different mass spectrometric methods: accurate mass measurement followed by in-source CID (collision-induced dissociation) fragmentation on a high-resolution mass spectrometer, MSⁿ fragmentation on a low-resolution ion trap mass spectrometer, and in-source CID combined with MSⁿ fragmentation on the ion trap MS. The accurate mass measurement allowed the assignment of the sum formula and the in-source CID fragmentation generated a unique fragmentation pattern. This fragmentation pattern was specific for each of the five structural classes present in H. viscosa. Comparison of this “mass fingerprint” with reference mass fingerprints obtained from synthetic compounds allocated the compound class and assigned the alkyl chains lengths. In-source CID combined with MSⁿ fragmentation on the ion trap MS aided in the structure elucidation of more complex cases like, as an example, differentiation of the viscosalines B₁ (10a), B₂ (10b), E₁ (10d), and E₂ (10e) (Schmidt et al. 2012) (see “Linear viscosalines”). Their structures were finally solved by a combined approach of NMR spectroscopy, MS spectrometry (including MS/MS), and chemical synthesis. The different schematic MS patterns for cyclic monomeric, dimeric, trimeric 3-APAs and viscosaline are shown in Fig. 7.

Fig. 7

Schematic MS pattern of cyclic monomeric, dimeric, trimeric 3-APAs and viscosaline

Structural classes of 3-alkyl pyridinium and 3-alkyl tetrahydropyridine alkaloids isolated from Haliclona viscosa

Cyclic monomers

Two cyclic monomers, haliclocyclins C (6a) and F (6b) have been isolated from the Arctic sponge H. viscosa (Fig. 8) (Timm et al. 2008; Schmidt et al. 2011). The haliclocyclins (6) consist of one pyridinium moiety connected in 1- and 3-position by a single alkyl chain of variable length.

Fig. 8

Structures of haliclocyclins C (6a) and F (6b) isolated from H. viscosa

The haliclocyclins (6) are characterized in mass spectrometry by a singly-charged molecular ion that does not form fragments under in-source CID fragmentation conditions. The HRMS spectrum of haliclocyclin C (6a) and haliclocyclin F (6b) are shown in Fig. 9. The analysis of their masses is shown in Table 1. Both compounds were identified in H. viscosa by comparing their chromatographic and spectrometric characteristics to those of a synthetic analogue (Timm et al. 2008; Schmidt et al. 2011). Haliclocyclin C (6a) has been already reported as a side product in the synthesis of the dimeric compound cyclostellettamine C (7a) (Timm et al. 2008).

Fig. 9

HRMS spectra of haliclocyclin C (6a, natural product and synthetic) and haliclocyclin F (6b)

Table 1 Compound masses of haliclocyclin C (6a, natural product and synthetic) and haliclocyclin F (6b)

Macrocyclic dimeric cyclostellettamines

Cyclostellettamines (7) consist of two pyridinium moieties connected in 1- and 3-position by two alkyl chains of variable length. They are structurally similar to the macrocyclic dimeric 3-alkyl THPy’s, the haliclamines (9). Cyclostellettamine C (7a) (Fig. 10) with two alkyl chains of 13 methylene groups is so far the only compound of this structural class identified in the Arctic sponge H. viscosa (Timm 2007). In fact, many cyclostellettamines (7) were already isolated from non-Haliclona sponges (Fusetani et al. 1994; de Oliveira et al. 2004; Oku et al. 2004; Grube et al. 2006; Xu et al. 2007).

Fig. 10

Structures of cyclostellettamine C (7a) isolated from H. viscosa and cyclostellettamine L (7b) isolated from Pachychalina sp.

The in-source CID MS spectrum of cyclostellettamines (7) is characterized by a singly- and doubly- charged molecular ion. For each alkyl chain, one fragment, corresponding to one pyridinium moiety connected to one alkyl chain, is observed. (F₁ and F₂, Fig. 11). The accurate mass of these fragments allows the determination of the length of each alkyl chain (Table 2) (Grube et al. 2006).

Fig. 11

HRMS spectrum of cyclostellettamines C (7a) and L (7b). The latter is shown for comparison since 7a has two alkyl chains of equal length

Table 2 Compound and fragment masses of cyclostellettamines C (7a) and L (7b)

Macrocyclic trimeric viscosamines

Cyclic trimers in H. viscosa are represented by viscosamine C (8a) that contains three saturated alkyl chains of 13 methylene groups each (Volk and Köck 2003). Two similar compounds with monounsaturated alkyl chains (8b/c) have been reported from Haliclona sp. from the Pacific coast of Guatemala (Casapullo et al. 2009) and from Bise Island, Japan (Teruya et al. 2006). The two derivatives differ in the position of the double bond and all three compounds are different in the length of the alkyl chain (Fig. 12).

Fig. 12

Structures of viscosamine C (8a) isolated from H. viscosa, the cyclic trimeric 3-APA (8b), and cyclohaliclonamine B (8c) both isolated from Haliclona sp.

Viscosamines (8) can be recognized in in-source-CID MS spectra by a characteristic pattern that consists of a very prominent triply-charged molecular ion and two lower-intensity fragments with an adduct ion each. The first, and more intense, doubly-charged fragment F₁ (8d, Fig. 13; Table 3) holds one proton more and one charge less than the triply-charged molecular ion. This is reminiscent of the cyclostellettamine (7) fragmentation under in-source-CID conditions. The F₁ fragment actually appears only in orthogonal time of flight (oTOF) mass spectrometry measurements under concomitance of formic acid and is assigned as a reduced form of a Hofmann fragment (Grube et al. 2006). Since the HPLC–HRMS measurements of H. viscosa crude extracts use similar MS conditions, the same mechanism can be proposed for viscosamines (8). The mechanism involves opening of the macrocycle and loss of one charge. Fragment F₁ (8d) (m/z = 390.8637) forms a doubly-charged formic acid salt (8f) at m/z = 412.8512 (Schmidt 2010).

Fig. 13

HR in-source-CID MS spectrum of viscosamine C (8a)

Table 3 Compound and fragment masses of viscosamine C (8a)

The second doubly-charged fragment, F₂ (8e) (m/z = 389.8637) is present only at low intensity. It is two mass units lower than the F₁ ion (8d) and constitutes the “regular” Hofmann fragment (8e) (Fig. 14) that is also observed in the related cyclostellettamines (7). It forms a doubly-charged salt with trifluoracetic acid (8g) at m/z = 446.8519, which was probably present in the instrument in trace amounts.

Fig. 14

MS adducts of viscosamine C (8a): 8d (reduced Hofmann adduct), 8e (Hofmann adduct), 8f (formic acid salt of 8d), and 8g (TFA salt of 8e)

Macrocyclic dimeric haliclamines

The haliclamines (9) are cyclic dimers bearing two THPy rings that are connected in 1- and 3-position by two alkyl chains of variable length. The haliclamines (9) are structurally similar to the cyclostellettamines (7) but possess THPy instead of pyridinium moieties. To date, four naturally occurring haliclamines were isolated from the Arctic H. viscosa; haliclamines C–F (9c–f), contain saturated alkyl chains (Fig. 15) (Volk et al. 2004; Schmidt et al. 2009). Haliclamines G (9g) and H (9h) have also been recently isolated from H. viscosa (Cychon et al. 2012). Another two compounds, haliclamines A (9a) and B (9b), containing mono- and/or bis-unsaturated alkene chains were originally described from Haliclona sp. from Hiburi-jima, Japan (Fusetani et al. 1989).

Fig. 15

Structures of haliclamines A (9a) and B (9b) isolated from Haliclona sp. and haliclamines C–H (9c–h) isolated from H. viscosa

The haliclamines (9) MS fragmentation pattern during in-source CID experiments resembles very much that of the cyclostellettamines (7) but haliclamines (9) are discriminable from cyclostellettamines (7) by their molecular weight. The similarity in the fragmentation pattern facilitated their original structure elucidation; the spectra show singly- and doubly-charged pseudomolecular ions as well as two fragments that represent the mass equivalent of a THPy moiety connected to either alkyl chain (Fig. 16; Table 4). These fragments allow the assignment of the alkyl chain lengths as well as potential double bonds. However, the exact structure of the fragments remains speculative an iminium ion resulting from a retro‐Diels‐Alder reaction is considered most likely (Schmidt et al. manuscript in preparation).

Fig. 16

HRMS spectrum (A) and in-source CID fragmentation spectrum (B) of haliclamine C (9c)

Table 4 Compound and fragment masses of haliclamine C (9c)

Fragmentation of the singly-charged molecular ion also produced ions with masses lower than the expected masses for the THPy ring and the alkyl chain daughter ion by two and four atomic units. Thus, leading to the same daughter ions as the fragmentation of the corresponding cyclostellettamines (7). This is due to a stepwise oxidation of the THPy ring to a pyridine ring which was also observed in NMR experiments in the presence of oxygen (Timm 2007).

Linear viscosalines

Five linear 3-alkyl pyridinium compounds have been isolated so far from H. viscosa. Viscosaline C (10c) combines two 1,3-linked alkyl pyridines with the amino acid β-alanine (Volk and Köck 2004). The viscosalines B (10a/b) and E (10d/e) possess alkyl chains of unequal length and therefore exist in two constitutional isomers each, B₁ (10a), B₂ (10b), E₁ (10d), and E₂ (10e), with the β-alanine moiety connected to either the longer or the shorter chain (Fig. 17). The identification of β-alanin-containing 3-APAs is unique to H. viscosa (Volk and Köck 2004; Schmidt et al. 2012).

Fig. 17

Structures of viscosalines B₁ (10a), B₂ (10b), C (10c), E₁ (10d), and E₂ (10e) isolated from H. viscosa

Viscosalines (10) show a unique peak pattern in mass spectrometry (Fig. 18; Table 5). It consists of the singly-, doubly-, and triply-charged molecular ions as well as a fingerprint-type fragment pattern that forms during in-source CID fragmentation and allows the assignment of different chain length (Fig. 17). In more detailed MS experiments, constitutional information about the viscosalines (10) can also be retrieved, i.e. the assignment of chain connectivity if alkyl chains of different length are present.

Fig. 18

HR in-source CID MS spectrum of viscosaline C (10c)

Table 5 Compound and fragment masses of viscosaline C (10c)

The structure elucidation of the viscosalines (10) was challenging. The central pyridinium unit could be established straightforward by NMR experiments, but not the connection of the β-alanine moiety to each alkyl chain.

Since viscosaline C (10c) contains two alkyl chains of equal length, the connectivity of the alkyl chains on the central pyridinium ring was not an issue. Based on the NMR data it was not possible to distinguish between the constitutional isomers I and II (Fig. 19) because the ends of the alkyl chains and the pyridinium ring are separated by 13 methylene groups. Extensive NMR studies that used a ten-step TOCSY transfer finally assigned the β-alanine to the alkyl chain that is connected in 3-position to the central pyridinium, i.e. isomer I in Fig. 19 (Volk and Köck 2004).

Fig. 19

Schematic representation of the issue for the constitutional assignment of viscosaline C (10c) as an example of viscosalines with alkyl chains of equal length

Another problem arose for viscosalines B (10a/b) and E (10d/e) as they possess alkyl chains of unequal length. The connection of each alkyl chain of different length to the 1 or 3 positions of the central pyridinium and to the pyridinium and β-alanine moieties could form 4 constitutional isomers III–VI (Fig. 20).

Fig. 20

Schematic representation of the issue for the constitutional assignment of viscosaline B (10a/b) as an example for viscosalines with alkyl chains of different length

The different isomers did not show conclusively different fragmentation in standard MSⁿ as the loss of the β-alanine moiety leads to the loss of structural information which hampered the elucidation of the connectivity of both alkyl chains. Additionally, only slight differences in intensity were observed during in-source CID experiments (Schmidt 2010). A lack of natural material initiated the synthesis of the constitutional isomers of viscosalines B (10a/b) and E (10d/e), which allowed extensive MS investigations. The isomers were finally distinguishable by a combination of in-source CID fragmentation performed prior to detailed MSⁿ experiments on an ion trap mass spectrometer. Viscosaline B₁ (10a) was then assigned to isomer I and viscosaline B₂ (10b) to isomer III (Fig. 20). It was also demonstrated that methylation of the natural products followed by MSⁿ allowed a distinction between the two constitutional isomers (Timm 2007; Schmidt et al. 2012).

Variation in secondary metabolites of Haliclona viscosa

The classification of our bioactive sponge sample from Spitsbergen by W. H. de Weerdt as H. viscosa represented the first report of this species in Arctic waters. However, sample amounts of the Spitsbergen sponges are limited by the small size of the individuals the small size of the individuals, their slow growth rate, and the attention to sustainable sampling in the fragile Arctic ecosystem. The reported occurrence of H. viscosa around the Orkney Islands, a much more accessible region than the Arctic, stimulated an examination of the Orkneys specimens’ secondary metabolites in hope of finding more organisms with similar bioactive metabolites.

Two Haliclona specimen from the Orkney Islands were examined, H. viscosa and Haliclona rosea, and a comparison of their HPLC–MS profiles was carried out. Large differences in the quality and the quantity of the secondary metabolites were observed between the Arctic and the Orkneys H. viscosa (Fig. 21).

Fig. 21

Comparison of the HPLC–MS chromatograms of H. viscosa from Spitsbergen and the Orkney Islands. Conditions: H₂O (0.1 % HCOOH)/MeCN gradient, at 35 °C [0 min: 80/20; 30 min: 20/80 with a flow rate of 0.4 mL min⁻¹] with a Waters XTerra C18 column (3.0 x 150 mm, 3.5 μm)

However, a comparison of the Arctic and Orkneys H. rosea, did not yield such differences; most importantly the main 3-APA metabolites present in the Arctic sponge H. viscosa were absent from the crude extracts of H. rosea (Fig. 22). In addition, the Arctic H. viscosa showed variation of its metabolites between different sampling years (Fig. 23) while individuals from the Orkneys and H. rosea did not show such variations. The reason for the differences in secondary metabolites in H. viscosa between the geographic locations has not been addressed yet. However, the different environments where the organisms were sampled as well as seasonal changes between the sampling years for the Spitsbergen samples suggest an influence of these conditions.

Fig. 22

Comparison of the HPLC–MS chromatograms of Haliclona rosea from Spitsbergen and the Orkney Islands. Conditions: H₂O (0.1 % HCOOH)/MeCN gradient, at 35 °C [0 min: 80/20; 30 min: 20/80 with a flow rate of 0.4 mL min⁻¹] with a Waters XTerra C18 column (3.0 x 150 mm, 3.5 μm)

Fig. 23

Variation in secondary metabolites of H. viscosa samples from Spitsbergen (different collecting years). Conditions: H₂O/MeCN gradient with 0.1 % HCOOH in both solvents, at 35 °C [0 min: 80/20; 25 min 45/55; 27 min 0/100] with a Waters XTerra C18 column (3.0 x 150 mm, 3.5 μm)

Haliclamines C (9c) and D (9d) constituted the major components in the crude extracts of all H. viscosa samples obtained so far, with exception of the year 2001 (Fig. 23). While the years 1999–2001 yielded an array of 3-alkyl pyridinium and 3-alkyl THPy alkaloids, no 3-alkyl pyridinium alkaloids were detected in any of our sample obtained after 2001. Whether this variation in the extract composition is due to differences between sampled individuals or due to environmental factors could not yet be fully assessed, but unpublished results point to a strong influence of the developmental stage of the sponges.

Summary

The Arctic sponge H. viscosa was identified as a rich source of 3-APAs. Two 3-APA structural classes were described for the first time from this sponge: the viscosamines (8) and the viscosalines (10). The cyclic monomeric 3-APAs, the haliclocyclins (6) were also first isolated from this sponge but these compounds were known before as side products from the syntheses of cyclostellettamines (7). Finally, the presence of a large number of haliclamines (9) could be assessed although many of them have not yet been identified.

Five 3-APA structural classes were identified from H. viscosa, the main structural differences came from the length of the alkyl chains which varies with the existence of a pyridinium or THPy moiety. The pyridinium-based compounds were isolated with chain lengths of 12–14: viscosamines (8) and viscosalines (10) with chain lengths of 12, 13, and 14 whereas for the dimeric pyridinium compounds only cyclostellettamine C (7a, 13/13) was isolated. Only chain lengths of 13 and 14 were identified for the cyclic monomeric 3-APAs haliclocyclins (6). Surprisingly the THPy alkaloids haliclamines (9) were identified bearing a large variety of chain lengths less than or equal to 12.

The structure elucidation of the five structural classes of 3-APAs showed the importance of a combined approach using NMR spectroscopy and mass spectrometry. Since these molecules contain long alkyl chains the information from MSⁿ spectra is essential to provide unambiguous structural information. This allowed us to investigate the MS fragmentation pathways of 3-APAs in detail. As a result, the identification of new 3-APA alkaloids without isolating them could now be achieved. This would be of great help for further investigation of the Arctic marine sponge H. viscosa as small sample amounts do not often enable the isolation of the secondary metabolites.