Type-I Hemins and Free Porphyrins from a Western Australian Sponge Isabela sp.

Two novel free porphyrins, isabellins A and B, as well as the known compounds corallistin D and deuteroporphyrin IX were isolated from a marine sponge Isabela sp. LC-MS analysis of the crude extract revealed that the natural products were present both as free porphyrins and iron(III) coordinated hemins, designated isabellihemin A, isabellihemin B, corallistihemin D and deuterohemin IX, respectively. Structures were determined via high-resolution mass spectrometry, UV-Vis spectroscopy and extensive NOESY NMR spectroscopic experiments. The type-I alkyl substitution pattern of isabellin A and isabellihemin A was assigned unambiguously by single crystal X-ray diffraction. Biological evaluation of the metabolites revealed potent cytotoxicity for isabellin A against the NS-1 murine myeloma cell line.


Introduction
Tetractinellid sponges of the Corallistes genus (Order: Tetractinellida; Family: Corallistes) have been reported to yield the microtubule stabilising macrolactone dictyostatin [1], and the poly-nitrogen compound corallistine [2], as well as the free porphyrins corallistins A-E and deuteroporphirin IX [3,4]. The assigned structures of corallistins A, B, C and E have been confirmed via total synthesis [5,6]. Recently, taxonomic re-identification of the sponges reported to produce corallistins A-E has suggested that the sponges are in fact members of the genus Isabela [7].
Metallated porphyrins and related macrocycles are expressed in most living organisms while synthetic porphyrins have been applied broadly in the area of photodynamic therapy [8]. They are however rarely encountered as functionalised secondary metabolites constituting a significant proportion of an organism's metabolic extract [4]. Given the biological importance of porphyrins as by-products of heme biosynthesis, their biosynthetic pathway has been well studied [9,10]. Beginning with glycine and succinyl-CoA, the enzymes ALA-synthase and ALA-dehydratase yield porphobilinogen in animals, fungi and α-proteobacteria. In plants, Archea and most other Bacteria, porphobilinogen is biosynthesised from two tRNA bound glutamyl starter units [10]. The enzyme PBG-deaminase then leads to the linear hydroxymethylbilane. From here, cyclisation can either occur chemically to afford the D 4h symmetrical uroporphyrinogen-I, or enzymatically with D-ring inversion via the UPG-III synthase, PBG-deaminase complex to afford uroporphirinogen-III [9,11]. In the case of uroporphirinogen-III, stepwise decarboxylation via the UPG-III decarboxylase chemically to afford the D4h symmetrical uroporphyrinogen-I, or enzymatically with Dring inversion via the UPG-III synthase, PBG-deaminase complex to afford uroporphirinogen-III [9,11]. In the case of uroporphirinogen-III, stepwise decarboxylation via the UPG-III decarboxylase enzyme affords coproporphyrinogen-III [11], which then loses a further two CO2 groups via the CPG-oxidase enzyme leading to the formation of protoporphyrinogen IX [11]. From here, PPG-oxidase removes six hydrogens to give protoporphyrin IX, whereby two protons are substituted for one unit of Fe 2+ via a ferrochelatase enzyme to afford heme B [10,11].
Our ongoing investigations [12,13] into the marine sponges of the Western Australian Marine Bio-resources Library (WAMBL) [14] yielded two novel free porphyrins from an Isabela marine sponge, herein named isabellins A and B (1a, 2a, Figure 1). LC-MS analysis of the crude extract revealed that the compounds were present as both free porphyrins and as ferric hemin compounds (1b, 2b). Further analysis of the mixture revealed the presence of the known compounds corralistin D (3a) and deuteroporphyrin IX (4a) [4] as well as their corresponding hemin counterparts (3b, 4b). Nuclear magnetic resonance experiments of the compounds were complicated by intense signal suppression of the nuclei present on the aromatic scaffolds, as well as the difficulty in replication of experiments, presumably due to the effects of aromatic ring current on the rapidly aggregating/de-aggregating porphyrins in solution, as has been noted in prior literature [3][4][5]. To this end, the structures of the compounds were assigned unambiguously via detailed interpretation of 2D-NOESY experiments and single crystal X-ray crystallography of 1a/b. The paramagnetic iron(III) hemin complexes 2b, 3b and 4b were provisionally assigned via HR-ESI-MS. Biological evaluation of the isolated metabolites revealed compound 1a as a potent inhibitor of the NS-1 murine myeloma cell line with strong selectivity for mammalian cell lines over bacterial pathogens and other Eukaryotes.

Results
HPLC-photo diode array detector analysis of the crude solvent extract of the Isabela sponge revealed a series of etio-type porphyrins with a characteristic Soret band at ap- proximately 390 nm and Q-bands with intensities in the order of IV > III > II > I [15]. Large scale isolation of the compounds was achieved using a combination of normal phase chromatography and reversed phase HPLC under acidic conditions. Compound 1a was isolated as a dark red crystalline solid. HR-ESIMS gave a protonated molecular ion [M + H] + at m/z 367.1924 consistent with the molecular formula C 24 H 23 N 4 (requires 367.1923). To our surprise, 1 H NMR of the compound in CDCl 3 (600 MHz) revealed the presence of a compound with four-fold rotational symmetry as revealed by three unique electronic environments (δ H 10.12; CH-5, 9.14; CH-3, 3.76; C-2-CH 3 ). We note that appropriate 1 H NMR integration was achieved by applying a 10 second relaxation time to the pulse sequence. All attempts to obtain reliable 13 C NMR data for compound 1a were hampered by significant signal suppression, presumably caused by the significant aromatic ring current present in the system. The difficulty in obtaining reliable 13 C NMR spectra for free base porphyrins has been documented elsewhere [5]. The structure of compound 1a was subsequently assigned unambiguously via single crystal X-ray diffraction ( Figure 2) and assigned the trivial name isabellin A. The extremely reduced state of compound 1a is a biosynthetically anomaly, and to the best of our knowledge is the first report of a highly reduced geo-porphyrinoid [16] isolated from a living organism.

Results
HPLC-photo diode array detector analysis of the crude solvent extract of the Isabela sponge revealed a series of etio-type porphyrins with a characteristic Soret band at approximately 390 nm and Q-bands with intensities in the order of IV > III > II > I [15]. Large scale isolation of the compounds was achieved using a combination of normal phase chromatography and reversed phase HPLC under acidic conditions. Compound 1a was isolated as a dark red crystalline solid. HR-ESIMS gave a protonated molecular ion [M + H] + at m/z 367.1924 consistent with the molecular formula C24H23N4 (requires 367.1923). To our surprise, 1 H NMR of the compound in CDCl3 (600 MHz) revealed the presence of a compound with four-fold rotational symmetry as revealed by three unique electronic environments (δH 10.12; CH-5, 9.14; CH-3, 3.76; C-2-CH3). We note that appropriate 1 H NMR integration was achieved by applying a 10 second relaxation time to the pulse sequence. All attempts to obtain reliable 13 C NMR data for compound 1a were hampered by significant signal suppression, presumably caused by the significant aromatic ring current present in the system. The difficulty in obtaining reliable 13 C NMR spectra for free base porphyrins has been documented elsewhere [5]. The structure of compound 1a was subsequently assigned unambiguously via single crystal X-ray diffraction ( Figure 2) and assigned the trivial name isabellin A. The extremely reduced state of compound 1a is a biosynthetically anomaly, and to the best of our knowledge is the first report of a highly reduced geo-porphyrinoid [16] isolated from a living organism. Compound 1b was isolated as a brown solid which recrystallised in chloroform to give dark brown needles. HR-ESIMS of the compound gave a molecular ion cluster [M − 2H + Fe(III)] + with an isotopic distribution characteristic of an iron atom at m/z 420.1032, consistent with the molecular formula C24H20N4 56 Fe + (requires 420.1037), as well as a prominent acetonitrile adduct at m/z 461.1302 consistent with the molecular formula C26H23N5 56 Fe + (requires 461.1303), allowing us to conclude that we had isolated the ferric hemin counterpart to compound 1a. The 1 H and 13 C NMR analysis of the paramagnetic Fe(III) complex proved unproductive. Ultimately, the structure of compound 1b was assigned via single crystal X-ray diffraction as the trifluoroacetate (TFA) salt ( Figure 3). Compound 1b was isolated as a brown solid which recrystallised in chloroform to give dark brown needles. HR-ESIMS of the compound gave a molecular ion cluster [M − 2H + Fe(III)] + with an isotopic distribution characteristic of an iron atom at m/z 420.1032, consistent with the molecular formula C 24 H 20 N 4 56 Fe + (requires 420.1037), as well as a prominent acetonitrile adduct at m/z 461.1302 consistent with the molecular formula C 26 H 23 N 5 56 Fe + (requires 461.1303), allowing us to conclude that we had isolated the ferric hemin counterpart to compound 1a. The 1 H and 13 C NMR analysis of the paramagnetic Fe(III) complex proved unproductive. Ultimately, the structure of compound 1b was assigned via single crystal X-ray diffraction as the trifluoroacetate (TFA) salt ( Figure 3  Compound 2a was isolated as dark brown amorphous solid. HR-ESIMS gave a protonated molecular ion [M + H] + at m/z 439.2133 consistent with the molecular formula C27H27N4O2 (requires 439.2134). The 1 H NMR analysis of the compound in acidified DMSO-d6 revealed the presence of four meso-proton environments (δH 10.37-10.34), three pyrrolic protons (δH 9.39-9.34), four methyl groups (δH 3.77, 3.75, 3.73 and 3.66) and a single propionate group as evidenced by two broad spin coupled triplets at δH 4.40 and δH 3.21. As with compound 1a, extensive 13 C NMR experiments failed to resolve the majority of the 13 C nuclei attributable to a compound of this size. A range of NMR solvents and experimental parameters were trialled.
Given the capacity of the organism in question to produce porphyrins with both type-I (derived from a symmetrical APAPAPAP bearing uroporphirinogen precursor, with A = Acetyl and P = Propionyl) and type-III (derived from an asymmetrical APAPAPPA bearing uroporphirinogen precursor, featuring D-ring inversion) topology, configurational assignment of the alkyl substituents present on compound 2a became a non-trivial exercise, with one potential type-I isomer and four potential type-III isomers possible. After some consideration it became clear that a simple experiment would be able to distinguish between naturally occurring type-I and type-III configurational isomers with four methyl groups present. The method can be summarised with the conditional statement: If every meso-proton of the porphyrin macrocycle shows a 2D-NOE correlation to a corresponding methyl group, then the porphyrin must be a type-I derived porphyrin. Extending the rationale, if one meso-proton does not show a correlation to a methyl group, then the porphyrin is either a natural type-III derived porphyrin, or a type-IV porphyrin, of which there are no naturally occurring derivatives.
Porphyrin 2a showed clear correlations from each meso-proton to each methyl group, and was therefore assigned the type-I structure depicted (Figure 4). The compound was given the trivial name isabellin B (The authors here suggest that the new trivial name Isabellin be used to designate type-I porphyrins within the lithistid family of compounds, whereas the trivial name Corallistin be retained for porphirins with a type-III alkyl substitution pattern.). In similar fashion to 2a, the type-III alkyl substitution patterns of the known metabolites corallistin D (3a) and deuteroporphyrin IX (4a) were confirmed using the methodology described above. Given the capacity of the organism in question to produce porphyrins with both type-I (derived from a symmetrical APAPAPAP bearing uroporphirinogen precursor, with A = Acetyl and P = Propionyl) and type-III (derived from an asymmetrical APAPAPPA bearing uroporphirinogen precursor, featuring D-ring inversion) topology, configurational assignment of the alkyl substituents present on compound 2a became a non-trivial exercise, with one potential type-I isomer and four potential type-III isomers possible. After some consideration it became clear that a simple experiment would be able to distinguish between naturally occurring type-I and type-III configurational isomers with four methyl groups present. The method can be summarised with the conditional statement: If every mesoproton of the porphyrin macrocycle shows a 2D-NOE correlation to a corresponding methyl group, then the porphyrin must be a type-I derived porphyrin. Extending the rationale, if one meso-proton does not show a correlation to a methyl group, then the porphyrin is either a natural type-III derived porphyrin, or a type-IV porphyrin, of which there are no naturally occurring derivatives.
Porphyrin 2a showed clear correlations from each meso-proton to each methyl group, and was therefore assigned the type-I structure depicted ( Figure 4). The compound was given the trivial name isabellin B (The authors here suggest that the new trivial name Isabellin be used to designate type-I porphyrins within the lithistid family of compounds, whereas the trivial name Corallistin be retained for porphirins with a type-III alkyl substitution pattern.). In similar fashion to 2a, the type-III alkyl substitution patterns of the known metabolites corallistin D (3a) and deuteroporphyrin IX (4a) were confirmed using the methodology described above.
Reanalysis of the sponge crude extract by LC-HRMS revealed a series of ferric metabolites consistent in accurate mass measurements to be the hemin counterparts of 2a, 3a and 4a, in similar relationship to that between 1a and 1b (Table 1). Given our previous difficulty in obtaining 1 H and 13 C NMR data for the paramagnetic 1b, isolation of the compounds was not pursued. In support of their identity as the ferric counterparts to 2a, 3a and 4a, treatment of the crude extract with concentrated H 2 SO 4 led to the disappearance of the ferric metabolites when analysed by LC-MS. No new peaks (aside from those ascribed to 2a, 3a and 4a) were seen in the chromatogram allowing for the provisional assignment of 2b, 3b and 4b as illustrated.
Mar. Drugs 2022, 20, x  Reanalysis of the sponge crude extract by LC-HRMS revealed a series o tabolites consistent in accurate mass measurements to be the hemin counterpa and 4a, in similar relationship to that between 1a and 1b (Table 1). Given ou difficulty in obtaining 1 H and 13 C NMR data for the paramagnetic 1b, isolation pounds was not pursued. In support of their identity as the ferric counterpar and 4a, treatment of the crude extract with concentrated H2SO4 led to the disa of the ferric metabolites when analysed by LC-MS. No new peaks (aside from cribed to 2a, 3a and 4a) were seen in the chromatogram allowing for the prov signment of 2b, 3b and 4b as illustrated. Biosynthetically, compounds 1a/b and 2a/b are presumably derived fro phyrinogen-I, whereas compounds 3a/b and 4a/b are divergent uroporphyrino rivatives. Given the degree of elaboration observed on the respective type-I a derived compounds, it seems that 1a/b, 2a/b act as more efficient substrates fo decarboxylase, CPG-oxidase and PPG oxidase enzymes, as well as the subsequ ing and devinylating enzymes active on the scaffolds. Whether the iron was its ferric state as isolated or in the ferrous state as has been reported for fer enzymes [10,11]    Biosynthetically, compounds 1a/b and 2a/b are presumably derived from uroporphyrinogen-I, whereas compounds 3a/b and 4a/b are divergent uroporphyrinogen-III derivatives. Given the degree of elaboration observed on the respective type-I and type-III derived compounds, it seems that 1a/b, 2a/b act as more efficient substrates for the UPG-decarboxylase, CPGoxidase and PPG oxidase enzymes, as well as the subsequent reducing and devinylating enzymes active on the scaffolds. Whether the iron was chelated in its ferric state as isolated or in the ferrous state as has been reported for ferrochelatase enzymes [10,11] remains unknown, however LC-MS analysis of the sponge crude extract failed to detect any iron(II) metabolites.
Testing of the metabolites against a panel of micro-organisms and cell lines (Tables 2 and 3) revealed potent cytotoxicity of compound 1a against the NS-1 myeloma cell line, with an MIC of 0.4 µg/mL at the 72-h time-point and 0.8 µg/mL at the 96-h time-point, comparable to the sparsomycin positive control. Testing against the neonatal foreskin fibroblast (NFF) cell line revealed marginal selectivity for the tumorigenic cell line with an MIC of 1.6 µg/mL at both time-points tested. In addition to this the compound was found to be moderately bacteriostatic towards the Gram-positive pathogens Bacillus subtilis and Staphylococcus aureus with an MIC of 25 µg/mL at the 24-h time-point, and inhibited the growth of Giardia duodenalis (MIC = 6.3 µg/mL). In stark contrast, metabolites 1b, 2a, 3a and 4a failed to display significant activity against any of the organisms and celllines tested, with the exception of mild activity towards NS-1 displayed by compound 3a (MIC = 50 µg/mL at the 72-h time-point), and mild anti-bacterial activity displayed by compound 4a towards Staphylococcus aureus (MIC = 50 µg/mL at the 72-h time-point). The cytotoxicity displayed by compound 1a is in line with other porphyrin macrocycles [8]. We postulate that the discrepancy in activity between metabolite 1a and other metabolites tested may be in part due to an increase in membrane permeability afforded to 1a by its lipophilic structure, over that of the Fe(III) salt 1b and the carboxylate bearing 2a, 3a and 4a.

Discussion
Two new free porphyrins, isabellins A (1a) and B (2a), an iron (III) coordinated porphyrin, isabellihemin A (1b), and the known compounds corallistin D (3a) and deuteroporphyrin IX (4a) were isolated from a marine sponge Isabela sp. The type-I alkyl substitution pattern of 1a, 1b and 2a was assigned unambiguously by NOESY NMR spectroscopic experiments and single crystal X-ray diffraction (Supplementary Materials). Testing of the isolated metabolites against a panel of micro-organisms and cell lines revealed potent cytotoxicity for 1a against the NS-1 and NFF cell lines that was comparable to that of the sparsomycin positive control.

General Experimental
UV/Vis spectra were acquired on an Agilent Cary 60 UV/Vis spectrometer. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance IIIHD 500 MHz spectrometer (500.1 MHz for 1 H and 125.8 MHz for 13 C) and a Bruker Avance IIIHD 600 MHz spectrometer (600.1 MHz for 1 H and 150 MHz for 13 C). Chemical shifts were calibrated against the residual solvent present: in CDCl 3 ( 1 H, δ 7.26 and 13 C, δ 77.16 ppm), in CD 3 OD ( 1 H, δ 3.31 and 13 C, δ 49.00 ppm), in (CD 3 ) 2 SO ( 1 H, δ 3.50 and 13 C, δ 39.52 ppm), in (CD 3 ) 2 CO ( 1 H, δ 2.05 and 13 C, δ 29.84 ppm) and expressed relative to TMS [17]. NMR spectra measured in neat TFA were measured at 273 K. Deuterium lock was maintained via insertion of a (CD 3 ) 2 CO standard capillary tube insert. H 2 O was suppressed via selective presaturation at 11.5 ppm. HPLC-mass spectrometry and HRMS were conducted using a Waters Alliance e2695 HPLC connected to a Waters 2998 diode array detector and Waters LCT Premier XE time-of-flight mass spectrometer using either an atmospheric pressure chemical ionization (APCI) source or an electrospray ionisation (ESI) source in either positive or negative mode. HRMS was conducted with either APCI or ESI in Wmode, using leucine enkephalin (200 pg/µL) as internal lock mass. For LC-MS separation an Altima C 18 column (150 mm × 2.1 mm, 5 µm, Grace Discovery Sciences, Columbia, MD, USA) was used with a flow rate of 0.3 mL/min. Rapid silica filtration (RSF) under reduced pressure was conducted on a sintered glass column using chromatographic silica (Davisil LC60A 40-63 micron, Grace Discovery Sciences, Columbia, MD, USA). Flash silica chromatography was conducted using the Reveleris X2 flash chromatography system equipped with a cartridge containing silica gel as the stationary phase (120 g, 40 µm, p/n 145). Semi-preparative and analytical HPLC were performed using either an Agilent 1200 HPLC system with a diode array detector (DAD) and fraction collector or using a Hewlett Packard 1050 equipped with a DAD and Pharmacia Biotech RediFrac fraction collector. Analytical work was conducted using an Apollo C 18 reversed phase column (250 mm × 4.6 mm, 5 µm, Grace Discovery Sciences) utilising 20 µL injections, and with a flow rate of 1.0 mL/min, and semi-preparative HPLC was undertaken with an Apollo C 18 reversed-phase column (250 mm × 10 mm, 5 µm, Grace Discovery Sciences) with 300 µL injections at a flow rate of 4.0 mL/min.

Extraction and Isolation
A portion of the Isabela sp. sponge (14.9 g, frozen weight) was sheared with scissors and extracted overnight, three times in MeOH:DCM (1:1, v/v) solution (3 × 500 mL) to which was added TFA (3 × 0.5 mL). The crude dark brown extract was filtered (Whatman No. 1, 18.5 cm) and reduced in vacuo to give a dark brown gum. The sample was reconstituted in acidified CH 2 Cl 2 (1.0% TFA) and absorbed on celite (1 g) before separating with a Reveleris automated flash chromatography module. The column was eluted with an isocratic solvent system consisting of 100% EtOAc for 20 min. The mobile phase was then increased from EtOAc to 100% MeOH over a further 5 min and held for 10 min. The flow rate was set at 25 mL/min and fractions were collected in 25 mL aliquots throughout the run. The fraction eluting at six minutes was separated using semi-preparative reversed phase HPLC, eluting with an isocratic solvent system of 95% ACN/H 2 O with 0.1% TFA to yield 1a (3.5 mg). The fraction eluting at seven minutes was separated using semi-preparative reversed phase HPLC at a flow rate of 4 mL/min, eluting with an isocratic solvent system of 75% ACN/H 2 O with 0.1% TFA over 40 min to yield 1b (2.2 mg) eluting at 5 min and 2a (3.2 mg) eluting between 15-20 min. The fraction eluting at twelve minutes was separated using semi-preparative reversed phase HPLC at a flow rate of 4 mL/min, eluting with an isocratic solvent system of 55% ACN/H 2 O with 0.1% TFA over 40 min to yield 3a (5.0 mg) and 4a (4.2 mg).
ProTOX is a generic bioassay platform for antibiotic discovery. Bacillus subtilis (ATCC 6633) and Staphylococcus aureus (ATCC 25923) were used as indicative species for antibacterial activity. A bacterial suspension (50 mL in a 250 mL flask) was prepared in nutrient broth by cultivation for 24 h at 100-250 rpm, 28 • C. The suspension was diluted to an absorbance of 0.01 absorbance unit per mL, and 10 µL aliquots were added to the wells of a 96-well microtiter plate, which contained the test compounds dispersed in nutrient agar (Amyl) with resazurin (12.5 µg/mL). The plates were incubated at 28 • C for 48 h, during which time the negative control wells change from a blue to light pink color. MIC end points were determined visually.
EuTOX is a generic bioassay platform for antifungal discovery. The yeast Candida albicans (ATCC 10231) and Saccharomyces cerevisiae (ATCC 9763) wereused as indicative species for antifungal activity. A yeast suspension (50 mL in a 250 mL flask) was prepared in 1% malt extract broth by cultivation for 24 h at 250 rpm, 24 • C. The suspension was diluted to an absorbance of 0.005 and 0.03 absorbance units per mL for C. albicans and S. cerevisiae, respectively. Aliquots (20 µL and 30 µL) of C. albicans and S. cerevisiae, respectively, were applied to the wells of a 96-well microtiter plate, which contained the test compounds dispersed in malt extract agar containing bromocresol green (50 µg/mL). The plates were incubated at 24 • C for 48 h, during which time the negative control wells change from a blue to yellow color. MIC end points were determined visually.