Marine natural products

Abstract

This review covers the literature published in 2024 for marine natural products (MNPs), with 617 citations (578 for the period January to December 2024) referring to compounds isolated from marine microorganisms and phytoplankton, green, brown and red algae, sponges, cnidarians, bryozoans, molluscs, tunicates, echinoderms, the submerged parts of mangroves and other intertidal plants. The emphasis is on new compounds (1256 in 336 papers for 2024), together with the relevant biological activities, source organisms and country of origin. Pertinent reviews, biosynthetic studies, first syntheses, and syntheses that led to the revision of structures or stereochemistries, have been included. An analysis of the role of artificial intellegence in marine natural products research is discussed.

Graphical Abstract

A comprehensive review of 1256 new MNPs including a novel cytotoxic alkaloid PM742 from the sponge Discoderma du Bocage

1. Introduction

This review is of the literature for 2024 and describes 1256 new compounds from 336 papers, compared to 1220 new compounds in 340 papers reported for 2023.¹ In addition, 30 known NPs were reported from a marine source for the first time, 3 artefacts were identified and 48 known MNPs had their structures revised. Only new MNP structures or previously reported compounds where there has been a structural revision, or a newly established stereochemistry are shown in this review. The review also covers previously reported MNPs with significant new bioactivities or ones that have been synthesised for the first time, but their structures are generally not shown. A † symbol on the identifying diagram number is used to distinguish structures where the absolute configuration has been determined for all stereogenic centres, axes and/or planes in a compound. Reports of new MNPs that were identified based solely on a combination of biosynthetic gene cluster (BGC) information, MS/MS data and/or Global Natural Products Social (GNPS)-based molecular networking, with compounds not isolated and no NMR data recorded, are excluded from the review. Only a selection of highlighted structures (54) is shown. Compound numbers for structures not highlighted in the review are italicised, and all structures are available for viewing, along with their names, taxonomic origins, collection locations, and biological activities, in an associated ESI^†† document. Over the past 20 years we have discussed a variety of topics relevant to MNP research in the conclusion section. Given that these treatises introduce new data derived from analysis of historical trends, we feel that it is more appropriate to include this discussion in a stand-alone section rather than the conclusion section. Therefore, this year we introduce a new “Perspective” section to the review, leaving the conclusion section to sum-up the overall insights gained from documenting the years MNP research outputs. Access to the curated MNP data held in the Marinlit database² provides all the structural and literature data used to prepare this review.

A pioneer of research on MNPs from Red Sea organisms, Professor Yoel Kashman from Tel Aviv University, passed away in mid-2025. Over his long career, Prof Kashman made significant discoveries, mostly associated with sponges, ascidians and soft corals. He will be sadly missed.

Trends in the number of new MNPs reported annually over the decade show that numbers of new sponge MNPs dropped again to a new low, while numbers of mollusc, (2) tunicate (6) and red algae MNPs (3) also dropped to decadal lows and numbers of reported cnidarian MNPs dropped to a near decadal low. No new bryozoan NPs were reported for the second year in a row and the number of new brown algae MNPs was lower than the decadal average. In contrast, numbers of MNPs isolated from fungi continue to rise and based on our changed reporting criteria, that now only includes fungi isolated from submerged environments, it is likely that 2024 had the highest number of new fungi MNPs reported over the 42-year history of this review. New MNPs reported from bacteria resurged to a decadal average level (Fig 1).

Fig. 1.

Trends in new MNPs. The bars represent the total number of new MNPs reported each year over the last 10 years. Fungi MNPs only includes those isolated from submerged sources.

2. Marine microorganisms and phytoplankton

2.1. Marine-sourced bacteria

Actinobacteria continue to be the most prolific source of MNPs reported from bacteria (with 165 new structures), even though they represent approximately 10% of total marine bacterial biodiversity. A sediment-derived Actinomadura sp. yielded five phenyl polyenes, maduraflavacins A−E 1−5.³ Two rare genera of bacteria yielded MNPs for the first time in 2024, linear peptides cellulamides A 6 and B 7 were reported from Cellulosimicrobium funkei,⁴ and glycoglycerolipids testacosides A−D 8−11 were the first MNPs reported from Microbacterium testaceum.⁵ A sponge-derived Microbispora sp. strain yielded okichromanone, a chromanone NP isolated as a mixture of two interconverting epimers 12 and 13 due to hemiketal ring opening and recyclisation.⁶

The genus Micromonospora yielded two anthracyclinones 14 and 15 and one anthraquinone 16,⁷ three nona-2,7-dienoic acid derivatives apocimycins A−C 17−19,⁸ as well as three spiroketal-containing macrolides 20−22, and a diketopiperazine 23. ⁹ X-Ray diffraction (XRD) analysis was used to define the absolute configuration of the co-occurring known MNP IB96212 24 for the first time.⁹ A large scale (96 L) fermentation of Nocardiopsis maritima yielded two cyclic hexapeptides, maritiamides A 25 and B 26, which showed moderate activity against S. epidermidis and weak activity against E. coli and P. fluorescens.¹⁰ Two α-pyrone polyketides, nocardiopyrones D 27 and E 28, as well as an alkaloid, nocarterphenyl I 29, were reported from a sediment-derived Nocardiopsis sp.¹¹ Nocarterphenyl I 29 possesses a rare 2,2’-bithiazole-p-terphenyl scaffold, the structure of which was confirmed via XRD, and showed weak to moderate activity against four bacterial strains with MIC values ranging from 0.8 to 1.6 μM with no cytotoxicity against methicillin-resistant S. aureus (MRSA) or four human tumour cell lines (HTCLs).

The genus Saccharomonospora yielded four pyridine-, thiazole-, and chloroquinoline-containing alkaloids lodopyridones D−G 30−33,¹² and two prenylated indole alkaloids, penipaline D 34 and N-acetyl-6-dimethylallyl-L-tryptophan 35.¹³ Three sulfur-bridged angucycline dimers, spongisulfins A−C 36, 37 and 38, and one monomer, rubiginone A3 39, were reported from a sponge-derived Spongiactinospora rosea, with the structures of 37 and 39 confirmed via XRD.¹⁴

As for previous years, the genus Streptomyces was the most studied, with 125 new MNPs reported. A discovery campaign targeting the identification of terminal oxazole-bearing NPs using PCR screening for oxazole cyclase encoding genes coupled with NMR-based screening measuring ¹J_CH one-bond coupling constants specific to oxazole rings, identified two polyketide MNPs, methyl-oxazolomycins A 40 and B 41.¹⁵ The two compounds showed moderate cytotoxicity against two estrogen receptor (ER) positive breast cancer cell lines MCF7 and T-47D, but no activity against ER-negative cell line MDA-MB-231.

Two linear polyketides, alpiniamides H 42 and I 43 were isolated from a sediment-derived collection of Streptomyces sp.,¹⁶ and a coral-derived Streptomyces sp. yielded polyketide chromone 44, polyketide 45 and fatty acids 46 and 47, and 48 and 49, each as racemates but purified to separate each enantiomeric pair.¹⁷

A series of methylene-bridged, dimeric, 4-hydroxy-2H-pyran-2-one (HPO)-containing pyrones, the phaeochromycins M−U 50−59 were produced by a large-scale culture of sediment-derived Streptomyces sp.¹⁸ The authors showed that simple HPO-containing molecules can react with formaldehyde to generate methylene-bridged dimers, proposing a chemical defence mechanism where HPO-producing bacteria can sequester, and effectively neutralise formaldehyde from the environment for competitive gain. Cloning and heterologous expression of a type II polyketide BGC spi1 sourced from a marine sediment-derived Streptomyces sp. into a S. coelicolor host, led to the isolation and identification of six angucyclines 60, 61−65, including spirocyclinone A 60 with a rare oxaspiro[5,5]undecane motif.¹⁹ Madeirone 66, a new mixed terpene- and polyketide-derived MNP, was reported from a sediment-derived Streptomyces aculeoletus.²⁰

Glycosylated angucyclines reported from Streptomyces sp. included seven C-glycosides, chrysomycins F−J 67−71,²¹ and grincamycins V 72 and U 73.²² A mangrove rhizosphere-derived Streptomyces sp. yielded three glycosylated polycyclic xanthones, kebanmycins A−C 74, 75 and 76, and the aglycone kebanmycin D 77.²³ Kebanmycin A showed potent activity against S. aureus, including four MRSA isolates, with MIC values ranging from 0.125 to 0.5 µg/mL, but did not inhibit B. subtilis growth. Based on the annotations from the genome mining tool antiSMASH, the candidate BGC keb was proposed to be responsible for assembly of the compounds.

Sediment-derived collections of Streptomyces sp. yielded four ansamycin-type macrolides 78−81,²⁴ five tetramate-containing macrolactams aburatubolactams D-E 82−84,²⁵ as well as hydroxycapsimycin 85 and brokamycin 86.²⁶ Two 6,6-spiroketal-containing MNPs, streptospirodienoic acids D 87 and E 88 were reported from a coral-derived S. cavourensis together with two known compounds bafilomycins P 89, and Q 90 whose absolute configurations were revised based on nOe correlation analysis, DP4+ calculations, and XRD data.²⁷ Compounds 89 and 90 showed moderate activity against the A-549 and HCT-116 HTCLs. Five oligomycins 91−95 were reported from a sediment-derived Streptomyces sp.,²⁸ four polyene macrolides 96−99, were isolated from a mangrove-derived S. hiroshimensis,²⁹ and a sponge-derived Streptomyces sp. yielded a tricyclic polyene macrolactam, weddellamycin 100.³⁰

Albusamides A-G, 101−107, a series of new long-chain hydroxylated acetamide MNPs, were reported from S. albus,³¹ and a sediment-derived collection of Streptomyces sp. yielded additional acetamide-containing MNPs, streptothiomycin F 108 and N-(5-nitropentyl)acetamide 109, the latter reported as a first time MNP.³² Albusamide C 103 displayed moderate cytotoxicity against a panel of fourteen HTCLs and leukaemias, but with no selectivity margin against the normal B lymphocyte cell line RPMI-1788. A genome-based mining strategy targeting the presence of the carpatamide BCG ctd, identified five acylated arylamines, carpatamides I−M 110−114 from S. parvus.³³

A series of shorter chain aryl amines 115−119, a phenylpropanoid 120, unsaturated fatty acids 121, 122, and two thiazinone-containing MNPs 123, 124 were reported from a sediment-derived Streptomyces sp., with 119 and 121-124 representing new MNPs and others known compounds reported from the marine environment for the first time.³⁴ A mudflat-derived Streptomyces sp. yielded a phenylthiazole anithiactin D 125,³⁵ and heterologous expression of an NRPS BGC grsc, sourced from a marine sediment-derived Streptomyces sp. into three different Streptomyces surrogate hosts, yielded eight new thiazole-containing MNPs, grisechelins E 126, F 127, and I−N 128−133, as well as two quinoline derivatives, grisechelins G 134 and H 135.³⁶ Notably, grisechelin E 127 showed moderate antimycobacterial activity, with an MIC of 8 μg/mL against M. tuberculosis.

A pyrrolo-pyrido-benzoquinone, streptoquinoneazaindole A 136, a urea-containing pyrrole alkaloid, streptoureipyrrole A 137 and 6-hydroxybenzothiazole 138, reported as a MNP for the first time, were isolated from a deep-sea sediment-derived Streptomyces sp.³⁷ A thioether-linked quinoline-quinazoline, quinosumycin 139 and a chromone, chromycone 140, were reported from a sponge derived S. diastaticus.³⁸ A dithiolopyrrolone, 141 was isolated from a tunicate-derived Streptomyces sp.³⁹ and a staurosporine analogue, streptomholyrine A 142 was reported from a sediment-derived Streptomyces sp.⁴⁰

Pattern-based genome mining, combining mass spectrometry (MS) isotopic signature and biosynthetic sequence data, led to the identification of two chlorinated pyrroloketoindanes, indanopyrroles A 143 and B 144, with their candidate BGC idp identified and biosynthetic assembly proposed.⁴¹ The trichlorinated indanopyrrole A showed moderate to potent activity against several Gram-positive and -negative bacteria, while the dichlorinated analogue indanopyrrole A was inactive.

In a high-throughput screening (HTS) discovery campaign against Mycobacterium avium complex (MAC) pathogens, seven disaccharide pyrimidine nucleosides, mavintramycins A−G 145, 146−151 were identified from a sediment-derived Streptomyces sp.⁴² Mavintramycin A 145 showed potent activity against two different MAC strains, with only moderate activity against two TB-associated mycobacterial strains and no cytotoxicity against other bacterial and fungal pathogens and one mammalian cell line. Moreover, 145 was found to be active against 40 clinically isolated M. avium strains, including those that were multidrug-resistant. Mechanism of action studies found that mavintramycin A binds to 23S ribosomal DNA and inhibits protein synthesis. The structure of 145 was later unequivocally confirmed via total synthesis.⁴³

A Korean sediment-derived Streptomyces sp. yielded six linear depsipeptides, homiamides A−C 152−154,⁴⁴ and cavomycins A−C 155−157,⁴⁵ all of which contained a core amino acid sequence of α-hydroxyisovaleric acid (Hiv)-valine (Val)-lactic acid (Lac)-valine (Val) residues like the valinomycin family of cyclic peptides. Three linear lipopeptides, albubactins A−C 158−160 were isolated from a tunicate-derived S. albidoflavus,⁴⁶ with the structure of 158 confirmed via total synthesis. A sediment-derived Streptomyces sp. yielded N-acylated cyclic octapeptides, acyl-surugamides A1−A4 161−164,⁴⁷ with the structure of acyl-surugamide A2 162 also reported, but not fully characterised, from a tunicate-derived S. albidoflavus.⁴⁸ A sponge-derived Streptomyces sp. yielded four glycosylated cyclopeptides, pyridapeptides F−I 165−168,⁴⁹ and three cyclic lipopeptides olenamidonins A−C 169−171 were reported from a deep-sea derived S. olivaceus.⁵⁰ Six terpenoid structures reported from Streptomyces spp. included the iridoid lucknolide A 172, reported as a MNP for the first time,⁵¹ and five labdanes, chlorolabdans A−C 173−175, and epoxylabdans A 176 and B 177.⁵²

Only one MNP from the phylum Bacteroidota was reported in 2024; an alga-derived Algoriphagus sp. strain yielded a new carotenoid, 2-hydroxyflexixanthin 178.⁵³ Five MNPs were identified from the phylum Fimicutes, which included three glycosylated 24-membered polyene macrolides, amylomacrolactines A−C 179−181 sourced from a sediment-derived Bacillus amyloliquefaciens,⁵⁴ and a new 25-membered polyene macrolide, macrolactin XY 182 and a new fatty alcohol 183 from a sponge-derived B. subtilis.⁵⁵

Fourteen MNPs were reported from the phylum Proteobacteria in 2024. A deep-sea sediment-derived Alcanivorax dieselolei strain yielded two p-aminoacetophenonic acid analogues, mohangic acids E 184 and F 185,⁵⁶ and four lipopeptides, bokeelamides A−D 186−189 were isolated from snail egg mass-derived Ectopseudomonas khazarica.⁵⁷ Two ureido-containing linear peptides, alteropeptilides A 190 and B 191 were reported from an arthropod-derived Pseudoalteromonas flavipulchra.⁵⁸ Structures of sulfur-containing amino lipids cysteinolides A 192 and B 193 were reported from bacterial species Ruegeria pomeroyi and R. meonggei and represent the first members of the sulfonolipid structural class of MNPs that have been fully characterised by NMR spectroscopy and their structures confirmed by total synthesis.⁵⁹ A seawater-derived Vibrio ruber yielded a new pyrrolidine, vibripyrrolidine A 194, a piperazine vibripiperazine A 195, and two diazinanes, vibridiazinanes A 196 and B 197, with 196 showing potent activity against S. aureus with a MIC value of 0.98 μg/mL and no inhibitory activity against E. coli, K. pneumoniae, P aeruginosa, or a panel of 3 HTCLs.⁶⁰

As with previous years, some MNPs reported from marine bacteria did not have adequate spectrometric or NMR spectroscopic data in full support of the proposed structures and were omitted from this review.⁶¹⁻⁶⁴ Total synthesis of the reported structure of marinoaziridine B ⁶⁵ revealed significant inconsistencies in the ¹H and ¹³C NMR spectroscopic data, suggesting the structure of this MNP should be revised. Total syntheses of other bacteria-sourced MNPs reported in 2024 included bacillimidazole B and discolins A, B, and E,⁶⁶ bacilotetrin C,⁶⁷ dassonmycins A and B,⁶⁸ caerulomycin K,⁶⁹ cihunamide B,⁷⁰ cyclomarin,⁷¹ discoipyrroles A-C,⁷² (+)-dixiamycin,⁷³ homoseongomycin,⁷⁴ ieodomycins A and B,⁷⁵ incarnatapeptins A and B,⁷⁶ iturin A,⁷⁷ lorneic acids C and D,⁷⁸ marformycins A and D,⁷⁹ mavintramycin A,⁴³ moiramide B,⁸⁰ octalactins A and B,⁸¹ rufomycins 1, 2, 4, 11, 22, and 24,⁸² salimabromide,⁸³ and sealutomicin C.⁸⁴

Reviews focused on bacterial MNPs published during 2024 included a summary of metagenomic strategies for the discovery of MNPs,⁸⁵ and actinomycete promoter engineering approaches.⁸⁶ The bacterium B. subtilis was reviewed as a versatile host for heterologous expression of NP BGCs,⁸⁷ as was the regulation of acyl homoserine lactone-based quorum sensing effects on the biosynthesis of bioactive molecules in bacteria.⁸⁸

Structural classes of bacterial NPs reviewed included polyacteylenes,⁸⁹ alkaloids,⁹⁰ and γ-pyrones.⁹¹ Groups of marine bacteria reviewed for biosynthetic potential and the diversity and activity of MNPs produced included the MAR4 clade of the genus Streptomyces,⁹² and marine-sourced Pseudomonas sp.⁹³ Biological activity studies of previously reported bacteria-sourced MNP structures included in vitro and in vivo anti-Toxoplasma gondii activity assessment of a series of marinoquinolines,⁹⁴ mechanism of action (MoA) studies of marinopyrrole, pentachloropseudilin, and pentabromopseudilin,⁹⁵ tartrolon D,⁹⁶ and tunicamycins,⁹⁷ and structure-activity relationship (SAR) reports of thiomarinol A,⁹⁸ and sunshinamide.⁹⁹

2.2. Cyanobacteria

Two linear lipopeptides, pemuchiamides A 198 and B 199 were isolated from a relatively large-scale (3 kg) Japanese collection of the filamentous cyanobacterium Hormoscilla sp.¹⁰⁰ The compounds contain thirteen amino acid residues including four prolines and several unusual ones such as 4-aminobutanoic acid, 4-(4-hydroxyphenyl)butanoic acid and N-methyl-D-phenylalanine. Pemuchiamide A showed moderate activity against Tryponosoma brucei rhodesiense (IC₅₀ = 0.63 μM), while the hydroxylated analogue pemuchiamide B was an order of magnitude less active. This indicated that the 4-aminobutanoic acid moiety may be an important pharmacophore feature involved in the antitrypanosomal activity for this peptide series. Laboratory cultures (some up to six months in length) of three geographically dispersed collections of Lepthotoe strains yielded three new PKS/NRPS hybrid linear peptide metallophones, leptochelins A−C 200, 201 and 202.¹⁰¹ The leptochelin structures were challenging to elucidate since they contain modified amino acids and isolated stereogenic centres. Their structures were deduced using a combination of tandem mass spectrometry, NMR spec troscopy, and bioinformatic analyses of the putative leptochelin BGC lec. The leptochelins bind to zinc, copper, iron, and cobalt and were moderately cytotoxic against four HTCLs. They are also the first reported MNPs from the genus Lepthotoe.

Studies on Moorena spp. yielded seven MNPs. A glycosylated macrolide moorenaside 203 was reported from a Florida collection of Moorena sp.¹⁰² and a Guam-sourced collection yielded two polyketides 204 and 205.¹⁰³ A co-culture of M. producens with Candida albicans led to significant upregulation of the production of a linear depsipeptide hectoramide B 206. Its putative BGC hca was identified and its biosynthetic assembly annotated.¹⁰⁴ Even though the presence of a fungal competitor was crucial to produce hectoramide in liquid culture, when tested against two Candida strains, 206 did not show any antifungal activity. M. bouillonii collected in India, yielded a linear lipodepsipeptide, kavaratamide A 207,¹⁰⁵ and a cyclic depsipeptide, wajeepeptin 208 was isolated from a Japanese collection of Moorena sp.¹⁰⁶ A benthic collection of Moorena sp. (annotated as Lyngbya sp.) in Indonesia yielded a cyclic peptide PM170453 209, the structure and absolute configuration of which was unequivocally confirmed via total synthesis.¹⁰⁷ Six lipopeptides, noducyclamides A1−A4 210−213 and noducyclamides B1 214 and B2 215, were reported from a laboratory culture of Nodularia sp.¹⁰⁸ The dodecapeptide noducyclamide B1 was weakly cytotoxic against the MCF7 breast cancer cell line.

The genus Okeania yielded six new MNPs. A glycosylated, 16-membered macrolide, polycavernoside E 216, was reported from a coastal Okinawan collection of Okeania sp..¹⁰⁹ Related analogues of this MNP, the polycavernosides A−C were originally reported from the red alga Polycavernosa tsudaii^110,111,112 and were implicated in human illness and death from consumption of this edible seaweed in Guam. An additional congener, polycavernoside D, was later identified from a Caribbean collection of the cyanobacterium Okeania sp.,¹¹³ and this latest report of 216 provides increasing evidence that polycavernosides are produced by cyanobacteria rather than red algae. Polycavernoside D did not show significant in vitro cytotoxicity against normal human fibroblast WI-38 cells, was weakly active against T. b. rhodesiense,¹⁰⁹ but demonstrated potent in vivo toxicity in mice with an ip. lethal dose of 0.81 mg/kg.¹¹⁴ Additional Okeania sp. collections yielded a lipopeptide, N-desmethylmajusculamide B 217,¹¹⁵ cyclic peptide okeaniazole A 218,¹¹⁶ and a linear lipopeptide, amantamide C 219.¹¹⁷ Terpenoids continue to be reported from cyanobacterial sources, with kagimminols A 220 and B 221 representing the first examples of cembrene diterpenoids isolated from the genus Okeania.¹¹⁸

An unidentified cyanobacterium with morphological resemblance to Symploca sp. collected in Guam, yielded three enyne-containing chlorinated fatty acid amides, taveuniamides L−N 222−224 and the absolute configuration of the C-8 stereogenic centre was reported for this MNP class for the first time. ¹¹⁹ Using genome-mining and metabolomic strain prioritization strategies, two dialkylresorcinol polyketides, 29-dechloro-29-hydroxybartoloside A 225 and 17-dechloro-17-hydroxybartoloside A 226, were isolated from Synechocystis salina,¹²⁰ with hydroxylation along the aliphatic side chain in both compounds representing a novel structural modification of the bartoloside family of polyketides. Finally, a methyl-branched fatty acid, jobosic acid 227 was reported from an unidentified bacterial biomass collected in Puerto Rico.¹²¹ Cyanobacteria remain one of the few marine taxa where reports of new MNPs from mixed assemblages or incomplete taxonomic identifications continue to be accepted for publication.

The total synthesis of the cyclic depsipeptide lagunamide C was achieved,¹²² resulting in the revision of the proposed structure from a 27-membered macrocycle originally published¹²³ to the 26-membered odoamide,¹²⁴ a known cyanobacterial MNP. Other cyanobacterial MNPs with reported total syntheses included dragocins A−C,¹²⁵ irijimaside A,¹²⁶ jamaicamide B,¹²⁷ majusculamide D,¹²⁸ rivularin A,¹²⁹ serinolamides A and B,¹³⁰ and trichophycin−I.¹³¹

Reviews included summaries of anti-infective,¹³² antiprotozoal,¹³³ and cancer-related¹³⁴ activities of cyanobacterial MNPs, as well as a comprehensive review of the diversity of NP structures and biological activities reported from the genus Oscillatoria.¹³⁵ Notable work on the biosynthesis of cyanobacterial MNPs included the identification and characterisation of the putative BGC for the assembly of dolastatin, reported from Caldora penicillata,¹³⁶ and SAR studies of known cyanobacterial compounds biselyngbyolide B,¹³⁷ carmaphycin B¹³⁸ iezoside,¹³⁹ and majusculamide D.¹⁴⁰ Environmental reports of cyanobacterial MNPs with broad implications to human health included a review on impacts of cyanobacterial blooms on fish farming,¹⁴¹ and the first account of bioaccumulation of lyngbyatoxin A in edible shellfish along the coast of New Zealand.¹⁴² This, together with a report of acute toxicity of a new polycavernoside analogue 216 in mice, accentuates the importance of accurate annotation, full structural characterisation, and monitoring of the occurrence of cyanobacterial MNPs that can impact marine ecosystems and human health.

2.3. Marine-sourced fungi

Co-culture of an Acremonium species with heat-killed Pseudomonas aeruginosa, resulted in isolation of acremosides A−G 228−234, linear sesquiterpenoids with a 2-phenyl acetoxy substituent conjugated to a sugar alcohol.¹⁴³ Other Acremonium strains were the source of the meroterpenoids, acremochlorins A 235 and B 236, pyridine alkaloid acremopyridone A 237 and cyclopentenone derivative acremoketene A 238,¹⁴⁴ sorbicillin-like compound 239 and acremonilactone 240,¹⁴⁵ and ascochlorin glycoside 241.¹⁴⁶

Fermentation of Albifimbria verrucaria yielded the nonenes, verrucanonenes A−C,¹⁴⁷ 242−244 and tetrahydrofuran-containing polyketide derivatives¹⁴⁸ 245−248, the butanolide derivative, alterbutenolide 249 was obtained from Alternaria alternata,¹⁴⁹ and Arthrinium arundinis cultures, respectively derived from a sea anemone and a sponge, yielded arthrinoic acid 250, hexylaconitic anhydride methyl ester 251, nonanoic acid derivative 252, arthripenoids G 253 and H 254,¹⁵⁰ and alternapyrone derivatives, alternapyrones G 255 and H 256. The BGC for alternapyrone biosynthesis was identified and heterologous expression led to isolation of an additional alternapyrone, alterpyranone I 257.¹⁵¹

As has been the case in many previous years, the genus Aspergillus was the most common source of fungal metabolites Co-culture of deep-sea-derived Aspergillus aculeatinus with a mangrove-derived Penicillium strain yielded the paraherquamides, aculeaquamides A 258 and B 259,¹⁵² while addition of sodium bromide to growth media of A. alliaceus resulted in production of brominated anthraquinone monomer 260 and dimers 261−263.¹⁵³ p-Terphenyl 264, 265 and diphenyl 266, 267 ether derivatives were obtained from A. candidus,¹⁵⁴ whilst the benzophenone derivatives, carneusones A−F 268−273 were isolated from a culture of A. carneus.¹⁵⁵ A. chevalieri and A. fischeri respectively yielded prenylated indole diketopiperazine 274¹⁵⁶ and prenylated indole derivative 275,¹⁵⁷ ascidian-derived A. clavatus was the source of 12S-deoxynortryptoquivaline 276,¹⁵⁸ co-culture of A. insulacola with a mangrove-derived Alternaria strain led to isolation of cyclic tetrapeptides, violaceotides B−E, 277−280,¹⁵⁹ miniaturised cultivation profiling enabled isolation of 2,6-diketopiperazines, noonazines A−C 281−283, and azaphilone, noonaphilone A 284,¹⁶⁰ and a nitrobenzoyl sesquiterpenoid 285 was obtained from culture of A ochraceopetaliformis.¹⁶¹

Stephaochratidin A 286, an unusual stephacidin-asperochratide hybrid, was isolated from culture of deep-sea-derived A. ochraceus.¹⁶² Several deep-sea sediment-derived Aspergillus strains yielded a range of metabolites. The sesquiterpenoid malfilanol C 287 was isolated from a culture of A. puniceus, along with known terrestrial but new marine natural product, malifilanol B 288,¹⁶³ phenol derivatives 289−298, (the last a known synthetic compound but new NP), were obtained from A. subversicolor,¹⁶⁴ and separate cultures of A. sydowii yielded sesquiterpenoid 299¹⁶⁵ and sydowimide A 300,¹⁶⁶ the latter by epigenetic modification. Another culture of A. sydowii was the source of xanthone-alkaloids sydoxanthones F−M 301−311,¹⁶⁷ and a prenylated indole alkaloid, aspertaichamide A 312 was obtained from A. taichungensis.¹⁶⁸

Cultures of A. terreus were the source of a range of metabolites including caryophyllene-type sesquiterpenoids, asperporonins A 313 and B 314,¹⁶⁹ ergone derivatives 315−324,¹⁷⁰ (all deep-sea-derived), sulfur containing metabolites asperteretals L 325 and M 326 and terreins A 327 and B 328,¹⁷¹ brominated aromatic butenolides 329−342,¹⁷² meroterpenoids 343−348,¹⁷³ butenolide derivatives 349 and 350 (via epigenetic manipulation and the latter a known synthetic compound but new NP)¹⁷⁴ and a suite of azaphilones, 351−361, which were produced through activation of a cryptic BGC.¹⁷⁵

Some polyketides 362−365, and a nucleoside 366 were isolated from a culture of A. versicolor,¹⁷⁶ while a range of metabolites were obtained from Aspergillus strains from a variety of sources. Sponge-derived Aspergillus strains yielded ergostane-type steroid 367,¹⁷⁷ terminal olefin-containing unsaturated fatty acid 368, 2-carboxy-2’-methyl-azobenzene 369 and 1,3,9-trimethyluric acid 370 (the last two known synthetic compounds but new NPs).¹⁷⁸ Sediment-derived Aspergillus strains yielded aspergilalkaloid A 371,¹⁷⁹ asperindopiperazines A−C 372−374, 5-methoxy-8,9-dihydroxy-8,9-deoxyaspyrone 375, and 12S-aspertetranone D 376,¹⁸⁰ emericelactones F 377 and G 378, preshamixanthones 379 and 380, 25-O-methylarugosin A enantiomers 381 and 382, furanones, 9-hydroxymicroperfuranone 383 and microperfuranone 384, (these last four known terrestrial but new MNPs), phthalimidinic acids 385 and 386, aspergilol G 387, anthraquinone 388, and propionic acid lactone derivative 389,¹⁸¹ (the last a known synthetic compound but new NP). A seawater strain of Aspergillus yielded ceramide 390¹⁸² while a coral-derived Aspergillus strain yielded megastigmanones A−C 391−393 and prenylterphenyllin H 394¹⁸³ and a starfish-derived strain yielded brominated isocoumarin 395. The production of 395 resulted from addition of bromide salts to the culture medium.¹⁸⁴ Aspergillus species from mangrove sediments have yielded a nor-diterpenoid acid, asperbrunneo acid 396,¹⁸⁵ indole alkaloids asperdinones A−H 397−404,¹⁸⁶ isoprenyl phenol 405,¹⁸⁷ and alkyl resorcinols 406−408.¹⁸⁸ Culture of Aureobasidium melanogenum yielded aliphatic δ-lactones 409−411 and fatty acid methyl esters 412−414,¹⁸⁹ sterol 415 was isolated from a culture of Beauveria sp.¹⁹⁰ and triterpenoid 416 was obtained from shellfish-derived Ceriporia lacerata.¹⁹¹ Sesterterpenoids with a 5/8/6/5 tetracyclic ring system, sesterchaetins A 417 and B 418 and diepoxide polyketides 419 and 420, were isolated from deep sea sediment-derived Chaetomium globosum,¹⁹² while sponge-derived C. globosum yielded six new cyctochalasans, marcytoglobosins A 421 and B 422¹⁹³ and marchaetoglobins A−D 423, 424−426, of which 423 is the first furan-containing cytochalasin.¹⁹⁴ Asepterpenedol A 427, an indole sesquiterpenoid obtained from culture of a Chloridium strain, contains a seven-membered ether ring as part of a 7/6/5/5/6/6 hexacyclic framework.¹⁹⁵

Cladosporium strains were the source of a range of new metabolites. Dodecanoic acid derivatives 428−431 (the last two known as semi-synthetic compounds) were isolated from a culture of C. cladosporioides,¹⁹⁶ C. halotolerans yielded pyrone derivatives 432 and 433,¹⁹⁷ drimane sesquiterpenoids 434, isocoumarin 435, coumarin 436 and known synthetic compound, 9-formylfluorene 437 as a new NP,¹⁹⁸ and a further Cladosporium strain was the source of isochromanone 438.¹⁹⁹ Curvularia aeria and C. lunata respectively yielded tricyclic aromatic polyketide 439 (as a racemic mixture)²⁰⁰ and resorcylic acid lactones 440−443²⁰¹ and isolation of curdepsidone A from a Curvularia strain led to revision of the structure to 444.²⁰²

A one strain many compound (OSMAC) approach to cultivation of Emericellopsis maritima resulted in isolation of eremophilane sesquiterpenoids 445−447,²⁰³ 6/7/6 tricyclic diterpenoid eutyditerpenoid A 448 and cytochalasin 449 were obtained from a culture of Eutypella scoparia,²⁰⁴ whilst another Eutypella strain yielded fifteen new sesquiterpenoids, eutypelides A−O 450−464.²⁰⁵ Sesquiterpenoids, including eremophilane derivatives fureremophilanes A−D 465−468 and acorane analogues furacoranes A 469 and B 470, were isolated from culture of Furcasterigmium furcatum,²⁰⁶ Fusarium strains were the source of fusarochromanone derivatives 471 and 472,²⁰⁷ hydroxyphenylacetic acid 473−476 and hydroxyphenylethanol derivatives 477²⁰⁸ and polyketide 478,²⁰⁹ and the cyclic pentapeptides, avellanins D−O 479−490 were obtained from culture of Hamigera ingelheimenensis.²¹⁰ An Irpex strain produced the heptaketides irpetones A 491 and B 492,²¹¹ catechol derivatives, meirols A−C 493−495 were obtained by fermentation of a Meira strain,²¹² whilst a Microascus strain yielded eight decahydrofluorene alkaloids, 496−503.²¹³ Azaphilones 504−506 and a dihydroisocoumarin 507 were isolated from a strain of Neopestalotiopsis,²¹⁴ mangrove sediment-derived Nigrospora oryzae contained four aniline derivatives 508−511,²¹⁵ spiro trione compound 512 and asparvenone derivatives 513−516 were obtained from a culture of Paraconiothyrium sporulosum,²¹⁶ and a culture of Parengyodontium album was the source of aromatic polyketide alternaphenol B2 517.²¹⁷

The Penicillium genus has proven once again to be a prolific source of new metabolites. A phenyl 6,7-dihydroxygeranyl ether derivative 518 was obtained from a P. arabicum culture,²¹⁸ P. brasilianum yielded diketopiperazine alkaloids 519 and 520,²¹⁹ soft coral-derived strains of P. chrysogenum were the source of aromatic heterocycles 521−525, and a sorbicillinoid 526,²²⁰ and of anthranilic acid-peptide derivative 527.²²¹ Dicitrinols A−C 528−530, citrinin derivatives with a 6/5/7/5 core, were obtained from Penicillium citrinum derived from hydrothermal vent sediment,²²² a culture of P. corylophilum yielded enantiomeric hydroxyphenylacetic acid derivatives 531−534 and α-pyrone analogue 535,²²³ and co-culture of deep-sea sediment-derived P. crustosum with mangrove-derived P. citrinum resulted in production of the dihydropyrones, rhytismatones C 536, and D 537.²²⁴

The relative and absolute configuration of the known indolone, notoamide X 538 was assigned for the first time. It was obtained from the mangrove sediment-derived Penicillium janthinellum along with three other indolones with proposed new structures. However, the structures of these additional compounds were only proposed based on MS data and so are not included here.²²⁵ Culture of P. oxalicum derived from a red alga yielded prenylated indole alkaloids 539 and 540²²⁶ and polyoxygenated steroid 541,²²⁷ sediment-derived strains, P. pancosmium and P. rubens respectively yielded meroterpenoids 542 and 543 (P. pancosmium)²²⁸ and polyketide derivatives 544 and 545 and amino-bis-tetrahydrofuran derivatives 546 and 547 (P. rubens),²²⁹ while a coral-derived P. sclerotium culture was the source of azaphilone pigments 548−559.²³⁰ Highly oxygenated meroterpenoids, penisimplinoids A−K 560−570, were isolated from a P. simplicissimum culture,²³¹ steroid 571 and polyketides 572 and 573 were obtained from P. variabile²³² and seagrass-derived cultures of P. velutinum and P. yezoense respectively yielded piperazine derivative 574,²³³ and decalin polyketides 575−584.²³⁴ Meroterpenoids, including 585−591²³⁵ and penicianstinoids F−K 592−597,²³⁶ indole diterpenoids, including 598−604²³⁷ and penpaxilloids F 605 and G 606²³⁸ and diketopiperazine alkaloid 607 (the last as a racemate),²³⁹ were isolated from cultures of various Penicillium strains, sediment-derived strains were the source of azaphilones 608−612²⁴⁰ and clavatol derivative 613,²⁴¹ and Penicillium strains derived from seawater yielded racemic polyketides 614 and 615²⁴² and enantiomeric hydroxyphenylacetic acid derivatives 616 and 617,²⁴³ which were separated by chiral chromatography. Cultures of various Penicillium strains also resulted in production of racemic phenalenone 618,²⁴⁴ quinolone-citrinin hybrids 619 and 620²⁴⁵ and speradines I 621 and J 622.²⁴⁶

Mangrove sediment or root-derived Penicillium spp. have also yielded pyridyl derivatives 623 and 624, indole 625,²⁴⁷ isobenzofuranone derivative 626, α-pyrone 627, quinone 628, phenol derivative 629,²⁴⁸ steroids 630 and 631, benzopyran 632,²⁴⁹ diterpenoid indole alkaloid derivatives 633and 634,²⁵⁰ sesquiterpenoids eupenicisirenins C−O 635−647,²⁵¹ aryl and decalin polyketide acids 648−669,²⁵² and alkyl resorcinols 670−673.²⁵³

Peniotrinins A−F, comprising cytochalasin derivatives 674 and 675, citrinin derivatives 676−678 and tetramic acid derivative 679, were isolated from a culture of a Peniophora strain²⁵⁴ and merocytochalasans, perochalasin A 680, and inseparable C20 epimeric pairs 681 and 682 were obtained from a Peroneutypa species. The BGC responsible for cytochalasin biosynthesis was identified through full genome sequencing and the biosynthesis elucidated through feeding experiments utilising ¹⁵N and deuterium labelled hydroxylamine.²⁵⁵ Pestalotiopsis strains were the source of rearranged pimarane diterpenoid 683 and known terrestrial but new MNPs, nodulisporenones A 684 and B 685,²⁵⁶ and polyketide derivatives, pestalotiopols E−J 686−691 (the last two as a racemic mixture),²⁵⁷ while cultures of deep-sea sediment-derived Phomopsis strains yielded benzoic acid derivative 692²⁵⁸ and 3-carboxy-indole derivatives 693−695.²⁵⁹ Co-culture of two Phomopsis strains obtained from the roots of the Chinese mangrove Rhizophora mangle yielded pyrazine 696, steroid 697 and α-pyrone 698 (from P. asparagi)²⁶⁰ and bis-chromones phomoxanthones L−N 699−701 (from Phomopsis sp.),²⁶¹ none of which were produced in monoculture of the individual strains.

Sediment-derived fungal strains have yielded the sesquiterpenoids 702−711 (from Pseudallescheria boydii)²⁶² naphthopyrone-macrolide hybrids, gymnoasins A−C 712−714²⁶³ and glucolipid 715²⁶⁴ (from Pseudogymnoascus strains) and sesquiterpenoids 716 and 717²⁶⁵ (from Pyrrhoderma noxium). Isolated from the Antarctic, the gymnoasins A−C 712−714 are the first naturally occurring compounds of their type and a biomimetic synthesis of 712 was achieved.²⁶³ Deep-sea sediment-derived fungal strains were the source of bisabolene sesquiterpenoid 718 (from Retroconis fusiformis),²⁶⁶ sesquiterpenoid-hydroquinone hybrids, saromacrophorins A−C 719−721 (from Sarocladium terricola)²⁶⁷ and 4-hydroxyphenyl derivative 722 (from Scedosporium apiospermum), which displayed moderate to potent activity against a panel of Candida auris strains,²⁶⁸ while mangrove sediment-derived Roussoella sp. contained two sesquiterpenoids 723 and 724 and a shikimate 725.²⁶⁹

Dihydroisocoumarins, exserolides L 726 and M 727 were obtained from culture of a Setosphaeria species,²⁷⁰ while various Simplicillium strains yielded a variety of metabolites including nortriterpenoids 728−730,²⁷¹ peptaibiotics 731−741,²⁷² dipeptide 742²⁷³ and acylated valine derivative, siamysin 743 (the last discovered via heterologous expression of a cryptic BGC).²⁷⁴ Eleven brominated depsidones 744−754 were isolated from a culture of Spiromastix sp.,²⁷⁵ Stachybotrys strains yielded atranones V−Z 755−759 and dolabellane diterpenoids 760−762,²⁷⁶ and phenylspirodrimanes, stachybotrins K 763 and L, 764,²⁷⁷ and cyclic tetrapeptides, endolides E 765 and F 766 were obtained from a Stachylidium strain. The endolides contain the unusual amino acid, N-methyl-3-(3-furyl)-alanine which was used in fragment pattern searches of tandem mass spectrometry data, in conjunction with molecular networking, to assist both isolation and structural elucidation of these metabolites.²⁷⁸

Culture of Talaromyces strains isolated from sediment have yielded a diverse array of metabolites, including oligophenalenone dimers 767−775,²⁷⁹ phenylhydrazone alkaloids 776−779,²⁸⁰ highly oxygenated seco-terpenoids 780−785,²⁸¹ macrolactin 786,²⁸² highly oxygenated phenol derivatives 787−789, isocoumarin glycosides 790 and 791 and naphthalene glycoside 792.²⁸³ Culture of T. purpureogenus yielded six prenylated indole diketopiperazine alkaloids, talaromyines A−F 793−798,²⁸⁴ cyclic heptapeptides 799 and 800, nor-diterpenoids 801 and 802, diterpenoid acids 803 and 804 and triterpenoid 805 were obtained from cultivation of T. scorteus,²⁸⁵ T. siglerae yielded cyclic heptapeptides, talaromides A−C 806−808²⁸⁶ and T. variabilis was the source of hybrid pentaketide sesquiterpenoids 809−811.²⁸⁷ Seawater-derived Talaromyces strains yielded meroterpenoids 812−814²⁸⁸ and benzoquinone 815 and furopyridinone 816 derivatives,²⁸⁹ a strain derived from sediment was the source of ring-opened azaphilone derivative 817,²⁹⁰ thirteen new polyketide derivatives 818−830, (6-hydroxy-4-methoxycoumarin 829 a known synthetic compound but new NP) were obtained from a deep-sea mussel-derived strain,²⁹¹ and duclauxin analogues, taladuxins A−N 831−844 were isolated from a culture of a coral-derived Talaromyces strain from which verruculosin B was also isolated and the configuration revised to 1S as in 845.²⁹² Investigations of mangrove sediment-derived Talaromyces spp. have resulted in isolation of γ-lactams, talaroilactams A−D 846−849,²⁹³ sesquiterpenoids, talaroterpenoids A−D 850−853,²⁹⁴ p-terphenyl derivatives, talaroterphenyls A−D 854−857,²⁹⁵ diketopiperazines, 858−860, a cyclopentenone 861, an α-pyrone 862,²⁹⁶ and linear polyketides 863−868.²⁹⁷ Two of the talaroterphenyls 855 and 856 potently inhibited the enzyme phosphodiesterase-4 (PDE4) linked to inflammation.

Tetramic acid alkaloids, tolypyridones I 869 and J 870 were isolated from a culture of Tolypocladium cylindrosporum,²⁹⁸ Trichoderma strains were the source of sesquiterpenoid derivatives 871−874,²⁹⁹ bergamotene-derived sesquiterpenoids 875−882, (through heterologous expression of a cryptic BGC),³⁰⁰ cyclonerane sesquiterpenoid 883,³⁰¹ γ-butyrolactones 884−889³⁰² and monomeric sorbicillinoids 890−893.³⁰³ Steroid derivatives 894 and 895 and linear polyketide 896 have been isolated from mangrove sediment-derived Trichoderma spp.^304,305 and culture of Xylaria acuta yielded a range of metabolites; lactones 897, 898, 899, glucopyranosides 900 and 901 and mannopyranoside 902, epoxycytochalasins 903 and 904 and butylitaconic acid 905.³⁰⁶

Total synthesis of acremolides A and B has been achieved and the absolute configurations established as 906 and 907,³⁰⁷ while asymmetric syntheses of aspilactonol F and aspiketolactonol were achieved in eight and ten steps respectively and their absolute configurations determined as 908 and 909.³⁰⁸ (±)-Notoamide N was prepared in a biomimetic approach via synthesis of (±)-stephacidin A,³⁰⁹ and several prenylated indole alkaloids of the asperversiamide family, namely asperversiamides B, D, E, G and J and dihydrocarneamide A, were prepared in biogenetically patterned total syntheses. The structure originally proposed for asperversiamide A was shown to be incorrect with a revised structure proposed but not yet proven.³¹⁰ A unified approach was utilised to prepare cyclic tetrapeptide JM-47,³¹¹ cyclocondensation of an electrophilic homoserine lactone was employed in the total syntheses of polonimides A and C³¹² and synthesis of bipenicilisorin was achieved through initial preparation of its monomer penicilisorin and the absolute configuration was established as 910.³¹³ Penostatin D was prepared in an asymmetric synthesis from l-ascorbic acid,³¹⁴ 12β-hydroxy conidiogenone C was also obtained in an asymmetric total synthesis³¹⁵ and total synthesis of the cyclic peptide, talarolide A showed that it (and by inference, talarolides C and D), occur naturally as atropisomers.³¹⁶ Racemic total synthesis of phomopsol B has been reported ³¹⁷ and syntheses of asperisocoumarin A and fusarimatin C were achieved from commercially available starting materials.³¹⁸

Alternariol monomethyl ether was shown to relieve food allergy symptoms in mice and reduce histamine release into blood serum,³¹⁹ the verticillin analogue, leptosin A showed potent inhibition of the parasite Toxoplasma gondii,³²⁰ brevianamide F exhibited antithrombotic activity in a zebrafish model³²¹ and agonodepside B showed potential as a photoprotection agent.³²² Fumetremorgin derivatives 12R,13S-dihydroxyfumitremorgin C and tryprostatin A displayed anti-hypercholesterolemia activity in an in vivo model,³²³ terphenyl HN-001 was shown to attenuate obesity in mice through induction of thermogenesis in adipose tissue³²⁴ and to suppress lipotoxicity from metabolic-associated fatty liver disease through inhibition of phospholipase A₂ (PLA2)³²⁵ and linoleic acid ameliorated metabolic dysfunction-associated steatotic liver disease in mice by targeting fatty acid binding protein 4.³²⁶ The mechanism underlying the antithrombotic activity of the peptide isaridin E was shown to be associated with inhibition of von Willebrand Factor secretion from activated platelets and this leads to protection against sepsis,³²⁷ butyrolactone-I alleviated intestinal heat stress in vivo through inhibition of two key enzymes which are upregulated in heat stress³²⁸ and pannorin was shown to be a selective, if relatively weak (IC₅₀ 1.73 μM) monoamine oxidase inhibitor.³²⁹

The BGC responsible for the biosynthesis of the quinolone alkaloids, the asperalins in Aspergillus alabamensis, was identified and the biosynthesis elucidated through heterologous expression.³³⁰

‘Omics-based methods for isolation and cultivation of marine fungi have been reviewed³³¹ as have genome mining and biosynthetic pathways of biologically active marine fungal metabolites.³³²

2.4. Dinoflagellates

The number of new compounds reported from dinoflagellates and other microalgae is slightly lower than the decadal average.¹ Prorocentrum lima was the source of polyketides prorocentin-5 911 and three okadaic acid derivatives 912−914, although surprisingly, the latter compounds were not assessed for bioactivity.^333,334 An indole methyl ether 915 and a biaryl ester 916 were reported from a Vietnamese Aurantiochytrium strain, although 915 is a likely artefact of isolation.³³⁵

Ectoine plays an important role in controlling osmotic responses in marine planktonic species. Using a hydrophilic interaction liquid chromatography (HILIC) based molecular networking approach, new analogue 2-homoectoine 917 was detected in multiple harmful algal bloom genera including Alexandrium, Karenia, and Prorocentrum. 2-Homoectoine plays a similar osmoadaptive role as archetype ectoine, for the producing strains. The synthesis of 917 was also achieved.³³⁶

Other structures were suggested from MS and transcriptomics data but are not shown here.^337,338

Although only isolated in 2023, the total synthesis of diatom sex pheromone peptide SIP⁺ has been accomplished,³³⁹ as has the production of seco-steroid gibbosterol A.³⁴⁰ The original structure of iriomoteolide-1a was disproven by synthesis in 2010, but efforts have not revealed the correct constitution of the molecule until now. The corrected structures of iriomoteolides-1a 918 and 1b 919 have now been determined following an extensive integration of DFT calculations, NMR analysis, and total syntheses.³⁴¹

A summary of the biomedical application of Alexandrium spp. NPs has been published.³⁴² Compound specific reviews focused on amphidinolides,³⁴³ azaspiracids,³⁴⁴ and ciguatera poisoning, especially related to ciguatoxins.³⁴⁵ The structure of ciguatoxin-5 920, originally proposed based solely upon MS data, has been confirmed by NMR spectroscopic analysis of only 40 µg of isolated material.³⁴⁶ The ambiguity between synthetic and naturally occurring portimine B has been resolved, in part by use of the new i-HMBC NMR experiment that can differentiate ²J_C,H vs ³J_C,H correlations. Differences in the NMR spectra observed were attributed to the formation of a transient hemiaminal hydrate, facilitated by acid modifiers added to aid HPLC separation. The authors advise researchers to be careful when adding acid or base to their chromatography solvents, especially when comparing spectroscopic data to literature values.³⁴⁷

The structure of 44-methylgambierone 921 was revised following extensive NMR analysis. The compound was also found to be unstable in DMSO-d₆ used for spectroscopic analysis hence a microstudy was undertaken to determine its stability towards various solvents over the course of eight months. The effects of the solvent chosen for NMR analysis, and potential storage, should be considered before committing an entire sample.³⁴⁸ A study of the accumulation of domoic acid (DA), the causative agent of amnesic shellfish poisoning, in five species of invertebrate including commercially valuable shellfish, has shown that DA concentrates in scallops (Pecten maximus) at levels 20-fold higher than in clams, slippersnails and sea squirts. The Pectenid species showed a distinctive ability to accumulate, transform and distribute DA within their body tissue.³⁴⁹ Transcriptomics analysis of saxitoxin and non-saxitoxin producing species of the prymnesins, super-carbon chain ladder polyethers produced by Prymnesium parvus, are some of the largest known natural products yet their biosynthetic origins have remained elusive until now. The megadalton PKZILLA-1 and −2 polyketide BGC’s encode for 140 and 99 enzyme domains, respectively, that match the predicted “pre-prymnesin” linear backbone. A smaller PKZILLA-B1 encodes for a smaller analogue, prymnesin-B1. This works expands the state of knowledge of large biosynthetic genes and enzymatic pathways to important biomolecules.³⁵⁰

3. Green algae

A Chinese Caulerpa racemosa var. turbinata yielded caulespiros A 922 and B 923, with neither being active against various bacterial strains.³⁵¹ Mebamamide C 924 was discovered from a Japanese specimen of Bryopsis, but also from the algal gastropod predator Elysia marginata.³⁵²

Two notable reviews of green algal metabolites were published in 2024. One focused on the allelopathic compounds from Ulva prolifera, the algal species that creates “green tides”,³⁵³ while the other summarised the metabolites and bioactivities of compounds from Chlorella vulgaris.³⁵⁴

4. Brown algae

Glyceroether 925 and polyunsaturated glycolipid 926 metabolites were reported from Sargassum sp. (China) and Hizikia fujiformis (South Korea), respectively.^355,356 Macrocarquinoid J 927 and sargasilols J–N 928−931 are meroterpenoid chromanes from Sargassum spp., although all are weak or inactive as anti-inflammatory agents.^{357, 358} An Indian Stoechospermum marginatum was the source of three new spatane diterpenoids 932, 933, 934 that showed weak cytotoxicity against four HTCLs.³⁵⁹

Coriaceumins A–D 935, 936−938 are rare examples of nitrogenous xenicane diterpenoids. Two diterpenoid lactams, dictyolactams C 939 and D 940 were also isolated from the same algal species, Dictyota coriacea, collected at Nanji Island, Zhejiang Province, China, although the lactams are artefacts of isolation as proven by extraction using different alcoholic solvents.³⁶⁰

The total syntheses of ecklonialactones C and D, and of eiseniachloride A, have been achieved for the first time.³⁶¹ Notable reviews include a summary of the phytochemicals produced by Halopteris scoparia,³⁶² and those from the Sargassaceae between 1975 and August 2023.³⁶³ Marine carotenoid fucoxanthin has been shown to reduce the effects of autism spectrum disorder induced by valproate treatment in a mouse model in a dose dependent manner. This reduction is induced by alterations to AKT/GSK-3β signalling.³⁶⁴

5. Red algae

An Okinawan Portieria sample yielded a new halogenated monoterpenoid 941, which also necessitated the reassignment of related compound 942. Both metabolites exhibited antifouling activity at 0.16 µmol.cm⁻².³⁶⁵ The only other report of new red algal metabolites was of diastereomeric sesquiterpenoids laurenolides A 943 and B 944 from a Vietnamese sample of Palisada intermedia.³⁶⁶

The same group has reported the total synthesis of pigments borolithochromes A, D, G, H1, H2, I1 and I2, metabolites of fossilised putative red alga Solenopora jurassica, via two separate publications.^367,368

A review of the metabolites of Gelidium corneum including mycosporine-like amino acids, carotenoids and polyphenols, along with their associated bioactivities, has been published.³⁶⁹ The current status of Japanese Laurencia complex focusing on chemical diversity and morphological characters for identification has been reviewed.³⁷⁰ A comprehensive review of the metabolites of Asparagopsis and their use for chemotaxonomic analysis of the genus has also been released.³⁷¹

A study has analysed the metabolic profile of different races of Laurencia nipponica considering their cytotoxicity, genotoxicity and their antifungal properties as it all relates to chemotaxonomy. Ultimately, with the aid of multivariate statistics, three different races can be differentiated using four different compounds, laurencin, laurallene, prepacifenol and laureatin.³⁷² An innovative approach to establishing the absolute configuration of metabolites with vicinal bromo-chloro substituents using chiroptical methods has been published. By applying van’t Hoff’s principle of optical superposition, and with the aid of chiroptical measurements of suitably well characterised model compounds for comparison, the contributions of different chromophores within a compound to the overall molar rotations can be mathematically disentangled to provide the absolute configuration of the metabolites in question. Use of this method has necessitated the revision of the structures of the halogenated meroditerpenoids callyphycols A 945 and B 946 and suggested that the structures of other metabolites within the class should be reinvestigated.³⁷³

6. Sponges

Aurantoside L 947 is an anti-Leishmanial tetramic acid glycoside from Siliquariaspongia japonica.³⁷⁴ A Leucetta sponge from Bohol Province, the Philippines, was the source of five dioxolane ether lipids 948, 949−952. To probe the relative and absolute configurations of their dioxolane cores, model compounds were prepared and their NMR data compared with that of the isolated metabolites. Even then, some compounds were indistinguishable, so the authors utilised europium-based lanthanide chiral shift reagents to ultimately discriminate all the compounds. None of the metabolites exhibited antibacterial activity.³⁷⁵

Endoperoxides, plakortides V 953 and W 954, and related gracilioether M 955, 11,12-dihydrogracilioether M 956, and 9,10-dihydrogracilioether G 957, were reported from Plakinastrella and Plakortis sponges, respectively.^376,377 Somewhat surprisingly, only two new peptides from sponges were reported. Halichondamide A 958 was isolated from a USA-sourced Halichondria bowerbanki, with molecular dynamics simulation used to help delineate the biosynthetic formation of its disulfide bonds; note the name halichondamide should not be confused with that of the much more commonly found halichondramide macrolide family.³⁷⁸ Neopetromin 959 is a cyclic tripeptide that contains a rare C-N biaryl link. Although inactive against two HTCLs, various enzyme targets, and as an antimicrobial, it does cause vacuole fragmentation in an actin-independent manner in tobacco BY-2 cells, like other actin polymerisation inhibitors but without disrupting cellular actin-dynamics. Studies of the impact of this unique bioactivity upon plant physiology are underway.³⁷⁹

PM742 960 is the first example of a polyketide α-pyrone coupled to an oxime-containing, non-ribosomal peptide. Isolated from a Discodermia sponge, PM742 is potently cytotoxic against eight cancer cell lines. Mode of action studies indicate it binds at the colchicine site, causing mitotic arrest via inhibition of microtubule dynamics at the G2/M checkpoint. ³⁸⁰ A synthetic SAR study identified an analogue, PM534, that covers 80% of the centres within the pharmacophore model of the colchicine binding site. PM534 has demonstrated a potent ability to reduce tumour growth in mouse xenograft models and has entered phase 1 clinical trials for the treatment of non-small cell lung cancer.³⁸¹

A new mycothiazole congener 961 was reported from a sample of Cacospongia mycofijiensis collected in Vanuatu. Although a potent cytotoxin, the new analogue is still 400- to 700-fold less active than mycothiazole itself which is cytotoxic in the pM-range. Nematodes (Caenorhabditis elegans) treated with sub-lethal doses of 961 exhibited an extended mitochondrial lifespan. SAR analysis shows that Z-Δ⁵ alkene geometry is critical for pM levels of cytotoxicity but does not impact the mitochondrial life extension bioactivity.³⁸²

Polybrominated 962 and lipidated 963 phenols have been reported from Lendenfeldia and Hemimycale sponges, respectively.^383,384 Several lactam- (964−967) and pyridine-based (968 and 969) alkaloids have been reported from a Chinese (Xisha Is.) Pseudospongosorites suberitoides. All are likely artefacts of extraction with methanol and are inactive as antivirals.³⁸⁵ A pair of enantiomeric indolinone alkaloids, 970 and 971, were sourced from a Tedania sponge.³⁸⁶ Neopetrotaurines A–C 972−974 are taurine-bridged isoquinoline dimer alkaloids that are weak suppressors of oncogenic fusion protein PAX3-FOXO-1 driven transcription. Neopetrotaurine C is a likely artefact of isolation.³⁸⁷

A new bromopyrrole glycoside has been reported from a deep sea Lissodendoryx papillosa specimen. Named in honour of the major contributions of Prof. Valentin Stonik to the field of MNPs, stonikacidin A 975 contains the unusual sugar, L-idonic acid, at its core. Although inactive against numerous mammalian cell lines, 975 has some antibacterial activity and it inhibits biofilm formation in S. aureus.³⁸⁸

Pyrroloiminoquinone alkaloids are commonly encountered in Latrunculid sponges. A dredged Latrunculia specimen yielded 6-chlorodamirone A 976 and its brominated analogue 977; over time the latter converts to the former when exposed to chloride in seawater in a non-enzymatic reaction. The total synthesis of both compounds was also achieved.³⁸⁹

The PIP-HSQCMBC-IPAP NMR pulse sequence was instrumental in helping identify a series of discorhabdin D congeners 978−982 reported from New Zealand Latrunculia spp.³⁹⁰

Multiple pyrrololactam alkaloids and their related metabolites (983−1007) have been reported from Phakellia, Stylissa, and Pseudospongosorites sponges collected around Xisha Is. in the South China Sea.^391,392,393 A New Zealand Dictyodendrilla c.f. dendyi yielded five new sulfated denigrin metabolites 1008−1012 with one rearranging spontaneously upon short term storage in an NMR tube.³⁹⁴

A diverse series of metabolites, including multiple 2-aminoimidazolone derivatives, were isolated from calcareous sponge Pericharax heteroraphis collected from a lagoon in Wallis and Futuna Is.; a new Zn²⁺-containing, heterodimeric clathridine A – clathridimine complex 1013 was one of the isolates. Although inactive as a stimulant of endochondrial ossification, the structure of the compound was confirmed by total synthesis as were those of several other congeners.³⁹⁵

New bromotyrosine derivatives were isolated from Suberea 1014−1017, Ircinia 1018−1021, and Pseudoceratina 1022 sponges collected in Australia, the Philippines, and Okinawa, respectively.^396,397,398 Merosesquiterpenoids have been reported from Pseudoceratina purpurea (1023, 1024, and 1025−1036) and Dysidea arenaria (1037−1042), although all the isolates were only weakly active or inactive altogether in a variety of bioassays.^399,400 Further meroterpenoids were isolated from Hyrtios (1043 and 1044) and Hippospongia (1045−1052) sponges,^401,402 while a number of norditerpenoid cyclic peroxides 1053−1062 were reported from a Bohol Is. (the Philippines) specimen of Diacarnus spinipoculum.⁴⁰³ A Hemimycale sponge was the source of seco-diterpenoid 1063, which was not assessed for bioactivity.⁴⁰⁴ Isolation of ircinialactams 1064−1066 from a Mexican Ircinia felix also provided an opportunity to revise the structure of related variabilin metabolite 1067; the total synthesis of 1064 was also achieved.⁴⁰⁵ An Antarctic Suberites sponge yielded suberitenone sesterterpenoids 1068−1072 although none showed activity as antivirals.⁴⁰⁶ Given how common scalarane sesterterpenoids are in marine sponges, surprisingly only one, lendenfeldarane V 1073, was reported in 2024,⁴⁰⁷ while two nor-sterol epoxide saponins 1074 and 1075 were obtained from a Petrosia sponge collected in the Philippines.⁴⁰⁸ Seco-sterols were reported from Cliona (1076), Hippospongia (1077, 1078) and Spongia (1079) sponges, respectively.^409,410,411 A series of 12 nor-terpenoid peroxides, 1080−1091, all derived from oxidative cleavage of larger metabolites, were isolated from a Diacarnus sponge and all with varying levels of antibacterial and cytotoxic activities.⁴¹² Other structures have been claimed but are inconsistent with the spectroscopic data provided, or were only detected by GNPS-based molecular networking, and hence are not shown.^413,384,404

Numerous first total syntheses were reported and these included halisphingosine A 1092, which necessitated a revision of structure,⁴¹⁴ cyclic peroxide ethyl plakortide Z,⁴¹⁵ peptide neopetromin,⁴¹⁶ and macrolide salarin C, which also validated a biosynthetic oxidation – Wasserman rearrangement to give salarin A in a biomimetic approach.⁴¹⁷ A multiplexed strategy using 96-well plates to optimise conditions for reactions between fatty dialdehydes, a nitrogen source and acrolein has been used to probe and verify the Whitehead – Baldwin biosynthetic proposal for the manzamines.⁴¹⁸ Indole alkaloids dragmacidins G and H, and tulongicin A, have been prepared for the first time,^419,420 as has aminophenol erebusinone, leading to its revised structure 1093.⁴²¹ The first total synthesis of spirodactylone has been accompanied by the synthesis of several related denigrin metabolites that has indicated the published structure of denigrin C is incorrect, although no alternative was proposed.⁴²² Guanidine alkaloids (−)-monanchoradin A and (−)-crambescin A2 393 have been prepared for the first time,⁴²³ as have makaluvamines H, L, M and N.^424,425 A series of BF₂-complexes of ageladine A have been prepared as BODIPY mimics, with their fluorescence being heavily dependent on both pH and halogenation of the imidazole ring.⁴²⁶ Psammaplysins M, O and Q, along with ceratinamide A, have been synthesised as racemates,⁴²⁷ as have sesquiterpenoids halichonadins A, B and C.⁴²⁸

A review of sponge-derived MNPs with activity against lung cancer has been published.⁴²⁹ Taxa-based reviews have focused on compounds from the Spongia and Neopetrosia genera, respectively, ^430,431 while compound specific reviews have summarised the literature pertaining to the convoluted structure elucidation, total synthesis and biological evaluation of PUFA-derivative (−)-mucosin,⁴³² and also of the immunomodulatory effects of halichondrin and its derivatives.⁴³³ Other compound specific reviews have concentrated on aaptamine and its derivatives,⁴³⁴ the anti-cancer potential of marine bis-indoles such the dragmacidins and topsentins,⁴³⁵ and of bromotyrosine-derivatives, mainly from Verongiid sponges.⁴³⁶ A compendium of 413 sponge-derived alkaloids published between 2000 – 2023 has been released,⁴³⁷ with a similar summary of diterpenoids isolated from sponges in 2009 – 2022 covering 228 structures.⁴³⁸

The inseparable endoperoxide pair, plakortinic acids C and D, isolated from the mutualistic symbiotic sponge pair Plakortis symbiotica and Xestospongia deweerdtae, have antiplasmodial activity without causing haemolysis. They exhibit promising absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties making them potential antimalarial leads.⁴³⁹ Chemoproteomic analysis has indicated that ubiquitin carboxyl-terminal lyase 5 (USP5) is the molecular target of endoperoxide gracilioether A (Plakinastrella mamillaris). USP5 is heavily involved in chromatin maintenance, catalyses degradation of abnormal proteins, and is over-expressed in multiple cancer types. Inhibitors of USP5 therefore have promise as anticancer agents.⁴⁴⁰

Peptide stylissatin A and its analogues inhibit membranous neuraminidase-1 functioning by modulating the chaperone activity of cathepsin A, thereby having application in preventing the formation of adipose tissue.⁴⁴¹ Chemoproteomics has also been used to detect agents that protect against inflammatory acute lung injury. Naamidine J interacts with His745 and Phe903 in chromosome segregation 1-like protein (CSE1L) which inhibits nuclear translocation and transcriptional activity of SP1, supressing inflammation. This study also provided more naturally occurring material which facilitated the reassignment of naamidines A, G and J 1094−1096 through X-ray crystallographic analyses.⁴⁴² Although only weak inhibitors of Mycobacterium tuberculosis under standard culturing conditions, axinellamines A and B have moderate to potent activity against the pathogen when grown with limited protein supplementation, mimicking in vivo conditions. Both compounds are inactive against the RAW264.7 cell line as a model host.⁴⁴³ Several bromotyrosine metabolites sourced from Verongiid sponges were found to be weak to moderate inhibitors of prion PrP^SC protein, which in turn reduces ER stress and also the spread of the pathogen in eukaryotic cells. This work provides a lead series of structures for development against a currently untreatable, and often fatal, disease.⁴⁴⁴

A nematocidal screen of over ten thousand fractions from approximately two thousand extracts held in the NatureBank repository highlighted seven extracts for further investigation. Deeper study of two chemotypes with activity against the sheep nematode Haemonchus contortus led to halaminol A and several agelasines as the dominant bioactive metabolites, but only agelasine B showed weak activity against the model roundworm Caenorhabditis elegans, suggesting either a minor component is responsible for the broader activity profile, or there is some instability in the isolated compounds.⁴⁴⁵ Sterol sokotrasterol sulfate (Topsentia sp.) was found to inhibit IFN-γ-induced expression of programmed cell death – ligand 1 (PD-L1) protein via targeting Janus kinase with a downstream activation of signal transducer and activator of transcription (STAT) signalling, making it a new immunomodulatory lead.⁴⁴⁶

Assignment of the absolute configuration of antifungal depsipeptide cyclolithistide A 1097 has been completed by the synthesis of all four stereoisomers of the rare amino acid, 4-chloroisoleucine, and then use of Marfey’s method to establish the configuration of the natural residue within the metabolite.⁴⁴⁷

An important cautionary observation regarding isolation of artefacts has been made. Pyrrole-containing MNPs including makaluvamines, discorhabdins, tambjamines and brominated tryptamines, were found to be trideuteromethylated when exposed to DMSO-d₆ NMR solvent in the presence of water, methanol, TFA and trace amounts of iron. Mechanistic investigation indicates that the reaction is caused by the presence of trideuteromethyl radical species generated by Fenton-type chemistry catalysed by the presence of iron or TFA.⁴⁴⁸

Finally, a study has explored the state of knowledge of the currently underexplored terpenoid biosynthesis in Dictyoceratid sponges, which, for example, are well known producers of spongian diterpenoids. While bioinformatic analysis detected analogues of known terrestrial prenyltransferases for one sesterterpenoid and multiple diterpenoid, steroid and carotenoid biosyntheses across various species, a lack of relatives to known terpenoid cyclase enzymes suggests the existence of divergent and unknown biosynthetic pathways in sponges. This study provides a start-point for exploring these currently cryptic biosynthetic genes and enzymes.⁴⁴⁹

7. Cnidarians

Tubastrine, sourced from the hard coral Tubastraea tagusensis, acts as an antioxidant to reduce levels of ROS, which is able to, in turn prevent neuron-like cell death caused by the oligomeric amyloid-beta peptide oAβ42.⁴⁵⁰ While acroamine A 1098, a 2-amino adenine alkaloid reported from the soft coral Acrozoanthus australiae, was essentially inactive as an inhibitor of cAMP-dependent protein kinase A, several semi-synthetic analogues including di- and tri-bromo derivatives were active.⁴⁵¹ One of the more common family of alkaloids reported from cnidaria are the zoanthamines, isolated from Zoanthus sp. A new, efficient synthetic route to norzoanthamine has been reported, notable for including three photochemical reactions used at key steps.⁴⁵² Of a set of thirteen prostanoids isolated from Clavularia sp., two (1099 1100) were new, with the former demonstrating weak in vitro cytotoxicity towards the Ca9–22 oral cancer cell line.⁴⁵³ A new stereoselective approach to the synthesis of prostaglandin A₂ epimers (15S)-PGA₂ and (15R)-PGA₂ has been reported that makes use of only catalytic amounts of heavy metals.⁴⁵⁴ The study also revealed that (15R)-PGA₂ is a weak inhibitor of GABA_A receptors in Xenopus laevis oocytes, while the (15S)-epimer was inactive.

While the structures of sesquiterpenoids 1101−1103, isolated from Lemnalia sp., were established using a combination of NMR and MS techniques,⁴⁵⁵ the structures of six of seven new sesquiterpenoids, bellissinanes A–G 1104−1110 (Stereonephthya bellissima) and each of lemneolemnanes A–D 1111−1114 (Lemnalia sp.) were secured using XRD.^456,457 Two new examples of isoledene-type sesquiterpenoids, 1115 and 1116 were isolated from the soft coral Heteroxenia sp., with both MNPs being inactive against SAR-CoV-2 and Leishmania major.⁴⁵⁸ A deep-sea collection of Anthothela grandiflora afforded unusual dimethylamino substituted cadinene-type sesquiterpenoids, anthoteibinenes A–E 1117, 1118, 1119−1121.⁴⁵⁹ Anthoteibinene B exhibited weak activity towards respiratory syncytial virus in vitro. Asymmetric syntheses of the sesquiterpenoids (−)-lemnalemnane A and (+)-lemnardosinane A, with both routes starting from (S)-carvone,^460,461 and a racemic synthesis of paralemnolin A have been reported.⁴⁶² Using a synthetic route starting from (−)-β-caryophyllene oxide, the structure of the chlorinated norsesquiterpenoid rumphellatin B has been confirmed, while those of rumphellatins A and C have been revised to 1122 and 1123 respectively.⁴⁶³ During the study, it was observed that reaction of rumphelloilide C with HCl in dioxane afforded rumphellatin B, with the authors speculating that the simple transformation may occur in soft coral species.

The structures of meroterpenoids sinudenoids F 1124 and G 1125, isolated from Sinularia densa, are unusual due to the presence of a methylbenzoate core⁴⁶⁴ while more classically functionalized meroterpenoids bearing hydroquinol or naphthoquinone cores 1126−1129 were isolated from Litophyton brassicum.⁴⁶⁵ An asymmetric total synthesis of the pentacyclic benzosesquiterpenoid (+)-verrubenzospirolactone 1130 made use of a late-stage, spontaneous, intramolecular Diels-Alder reaction.⁴⁶⁶

Four examples of diterpenoids embodying three different skeletons, herpetopanone B 1131 bellissimain A 1132 elisabethadienol B 1133 and elisabethadienol C 1134 were reported from extracts of Stereonephthya bellissima, with none showing activity in a pro-angiogenesis assay in zebrafish.⁴⁵⁶ Further examples of biflorane-type diterpenoids, biofloranates E–I 1135−1139, in addition to two glycosides lemnaboursides H 1140 and I 1141, were reported from a South China Sea collection of Lemnalia bournei.⁴⁶⁷ Of the set, only lemnabourside H exhibited antibacterial activity, weakly inhibiting the growth of E. coli. The structures and relative configurations of seco-diterpenoids kallopterolides A–I 1142−1150, isolated from the sea plume Antillogorgia kallos, were secured by combinations of NMR analysis and DFT DU8ML calculations.^468,469 Structures were assigned to polyoxygenated casbane-type diterpenoids sinueracasbanones E–O 1151−1161, isolated from Sinularia erecta, by extensive use of NMR data in conjunction with DFT DP4+ analysis, while absolute configurations were assigned by the use of Mosher’s method and TDDFT-ECD calculations.⁴⁷⁰ Sinueracasbanones I and O exhibited weak anti-inflammatory activity. The capnosane-type diterpenoid 1162, isolated from Lobophytum pauciflorum, was inactive in a zebrafish model of inflammation.⁴⁷¹ Three cycloamphilectane-type diterpenoids, sinucycloamtins A–C 1163, 1164 and 1165 were isolated from a South China Sea collection of Sinularia brassica.⁴⁷² Purification of sinucycloamtins B and C was hampered by the apparent ability of the two compounds to undergo intramolecular ester exchange. Subsequent saponification afforded a diol product, the structure and absolute configuration of which were established by XRD. Pseudopteroxazole and pseudopterosin G are antibacterial diterpenoids previously reported from the Caribbean soft coral Antillogorgia (formerly Pseudopterogorgia) elisabethae. Using an unbiased proteomics approach, the antibacterial mechanism of action of both MNPs appears to be associated with disruption of membrane-associated steps involved in cell wall biosynthesis.⁴⁷³

With over sixty new examples, cembranoids remain the dominant sub-structural class of diterpenoids reported from soft corals. Of a set of eleven cembranoids isolated from a South China Sea collection of Litophyton amentaceum, litoamentenes A–K 1166−1176, only litoamentene I exhibited cytotoxicity, being weakly active against the NCI-H446 cell line.⁴⁷⁴ In addition to sterols (see later), an extract of Lobophytum catalai afforded the cembrane-type diterpenoids lobocalines A–E 1177−1181.⁴⁷⁵ While XRD analysis provided the structure and absolute configuration of lobocaline A, interpretation of NMR data, including DP4+ probability analysis, and TDDFT-ECD calculations, secured the structures and configuration of the remaining examples. The structure of co-metabolite lobophytrol B, previously reported as a bicyclic terpenoid, was revised on the basis of XRD analysis to 1182.

The 3D structure of 7S,8R-dihydrodeepoxysarcophytoxide 1183 has been secured by XRD analysis.⁴⁷⁶ The findings call into doubt the structure and/or stereochemistry previously proposed for (+)-7β,8β-dihydroxydeepoxysarcophytoxide. Pyrane ring-containing cembranoid 1184 and a 4-hydroperoxy derivative of sarcophine 1185 were reported from an extract of Sarcophyton glaucum, collected off the coast of Egypt.⁴⁷⁷ Four sarcophine analogues, sarcoeleganolides H–K 1186−1189 were isolated from a South China Sea collection of Sarcophyton elegans,⁴⁷⁸ while S. crassocaule was the source of meijicrassolins A–E 1190−1194.⁴⁷⁹

A series of capnosane-type cembranoids, sarcocinerenoids A–H 1195−1202 and two unusual variants, sarcocinerenoids I 1203 and J 1204, were reported from extracts of a Xisha Is., South China Sea collection of Sarcophyton cinereum.⁴⁸⁰ XRD analysis was used to secure the structure and absolute configuration of sarcocinerenoids A–C, F, G and I. In a separate publication, details of the characterisation of a further nine cembranoid-type diterpenoids, sarcocinerenolides A–I 1205−1213 from the same S. cinereum biomass were reported.⁴⁸¹ Amongst these structures, sarcocinerenolide A is unusual in being a ring-opened variant, while the remaining congeners are more typical of cembranoids, displaying variation in the location of ether linkages. A structurally-related sarcophine derivative bearing a C-1 to C-12 peroxy linkage, sarcacutumolid A 1214 was reported from the extract of a Red Sea collection of Sarcophyton acutum.⁴⁸²

A Ximao Is., South China Sea collection of Sinularia pedunculata afforded a structurally-diverse set of cembranoid diterpenoids including sinupedunolides A–E 1215−1219 and sinupedunols A 1220 and B 1221.^483,484 The structures of sinupedunolide A and sinupedunols A and B were secured by XRD analysis. Of the five cembrane diterpenoids, sinudenoids H–L 1222, 1223−1226 isolated from Sinularia densa, sinudenoid H is an unusual example of a bisnorfuranocembrenolide skeleton.⁴⁶⁴ The tricyclic structures of 4α-hydroxy-chatancin 1227 and sarcotoroid 1228, isolated from Sarcophyton tortuosum, were proposed to be derived from cembranoid precursors that had undergone transannular Diels-Alder cycloaddition reactions.⁴⁸⁵

The possible role of photochemistry in the biosynthesis of the cyclobutane ring of the cembranoid diterpenoid (+)-providencin 1229 has been supported by its semisynthesis from (−)-bipinnatin E 1230, which was isolated from a Floridian specimen of Antillogorgia kallos.⁴⁸⁶ The Norrish-Yang cyclisation step also afforded a C-14 regioisomeric product, which is likely an anticipated NP. Absolute configuration was assigned to all three compounds based upon XRD analysis of (−)-bipinnatin E. Two accounts of the synthesis of the revised structure of scabrolide B were reported in 2024. In the first publication, (−)-scabrolide B was synthesized in 19 steps (longest linear sequence), and in a simple set of reactions was converted into (−)-sinuscalide C 1231, (+)-ineleganolide and (+)-horiolide 1232.⁴⁸⁷ A second route to (−)-scabrolide B was reported by another research group, with simple conversions again affording related diterpenoids (−)-sinuscalide C and (+)-ineleganolide, as well as a third MNP, (+)-fragilolide A 1233.⁴⁸⁸ A full account of the synthesis of (−)-scabrolide F has been reported, with weak anti-fouling activity observed for the NP and synthetic intermediates.⁴⁸⁹

New biological activities were reported for cnidarian-derived cembranoids including the finding that 8-epi-sarcophinone and ent-sarcophine exhibit anti-fouling activity when included in marine paint preparations,⁴⁹⁰ sinulariolide exhibits anti-inflammatory activity in a rheumatoid arthritis model,⁴⁹¹ 7,8-epoxy-11-sinulariolide acetate also exhibits anti-inflammatory as well as inhibition of osteoclastogenesis,⁴⁹² sinularin induces apoptosis in prostate cancer cells via a mechanism related to up-regulation of FOXO3 forkhead transcription factor,⁴⁹³ and 4-carbomethoxyl-10-epi-gyrosanolide E inhibits oral cancer cell migration and invasion by inhibition of the MAPL pathway.⁴⁹⁴ Sarcotrocheliol exhibits weak cytotoxicity, antimicrobial and larvicidal activities.⁴⁹⁵ An aquaculture specimen of Briareum stechei afforded the epoxybriarane briastecholide O 1234, while wild-type specimens of the same organism yielded the related analogue briastecholide P 1235.⁴⁹⁶

XRD analysis was used to secure the structure and absolute configuration of not only briastecholide P, but also of the known co-metabolite briaexcavatin X 1236. In two separate accounts, a large array of new and known polyoxygenated briaranes were isolated from a collection of Junceella fragilis collected off the coast of Southern Taiwan. In addition to two new chlorine-containing congeners, 1237 and 1238, two previously reported analogues gemmacolide X 1239 and frajunolide I 1240 were re-isolated.⁴⁹⁷ Absolute configurations of the latter two diterpenoids were determined for the first time. In the second publication, one new briarane example was reported, fragilide Z 1241, while the structures and absolute configurations of known co-metabolites 12-epi-fragilide G 1242, fragilide P 1243, junceellolide D 1244, junceellonoid A 1245 and juncin ZI 1246 were secured by XRD analysis.⁴⁹⁸ Excavatolide B and two semisynthetic analogues, one being an ester linked piperidine salt and the second being an oxime-linked terminal alkyne, inhibit inflammation and angiogenesis in models of atopic dermatitis.⁴⁹⁹

As part of a SAR study of transient receptor potential melastatin 7 (TRPM7) ion channel inhibitors, six new xenicane diterpenoids 7S,8S-epoxywaixenicin A 1247, 7S,8S-epoxywaixenicin B 1248, 12-deacetylwaixenicin A 1249, waixenicin E 1250, waixenicin F 1251 and 20-acetoxyxeniafaraunol B 1252 were isolated from a Hawaiian collection of Sarcothelia edmondsoni.⁵⁰⁰ The authors suggested that the epoxide-containing analogues 7S,8S-epoxywaixenicin A and 7S,8S-epoxywaixenicin B may be artefactual. Preliminary elements of an SAR were proposed, indicating the requirement of the 12-acetoxy group in combination with a dihydropyran ring for activity, and that bioactive compounds may act as latent electrophiles. Three examples of xenicane diterpenoids, miolenol 1253, epoxymiolenol 1254 and epoxycoraxeniolide A 1255, were reported from a deep-sea collection of the soft coral Paragorgia arborea.⁵⁰¹

The first asymmetric total synthesis of the xenicane isoxeniolide A 1256 has been reported, confirming the structure and defining the absolute configuration.⁵⁰² Ten new sterols were reported from cnidarians in 2024, including lobocaloids A–D 1257−1260 from Lobophytum catalai,⁴⁷⁵ cholesterols 1261 and 1262 from Litophyton mollis,⁵⁰⁵ capnesterones A 1263 and B 1264 from Capnella imbricata,⁵⁰⁶ and cholestene-7-ones 1265 and 1266 from Lobophytum durum and Capnella imbricata, respectively.^505,506 The study of Lobophytum durum also led to the re-isolation of the known ergosterol derivative 1267, the relative configuration of which was determined by XRD analysis, providing clarification on the configuration of the 24-methyl group.⁵⁰⁵

A secogorgostenone steroid, originally sourced from Sinularia leptoclados, exerts anti-inflammatory activity and brain-protective effects by attenuating matrix metalloproteinase-mediated events.⁵⁰⁷ The MNP chemistry of cnidarians belonging to the genus Capnella, covering the period 1974 to May 2024, has been reviewed.⁵⁰⁸ The background history of the palytoxins, originally isolated from the zoanthid Palythoa sp., as well as their biological activities and mechanism of action, has also been reviewed.⁵⁰⁹ The chemistry and biological activities of ketosteroids, a number of which are isolated from soft corals, have been reviewed.⁵¹⁰ A recombinant variant of Tst2, a 38-residue peptide identified by transcriptomic and proteomic analysis of the Australian sea anemone Telmatactis stephensoni, is an inhibitor of the transient receptor potential subfamily V member 1 (TRPV1) ion channel.⁵¹¹ HCIQ2c1, a Kunitz-peptide identified in cDNA analysis of the sea anemone Heteractis crispa, binds to the transient potential ankyrin 1 (TRPA1) calcium-permeable channel, appearing to stabilize the channel in the open conformation, and reduces pain and inflammation in vivo.⁵¹² Bioassay-directed fractionation of toxins from a Japanese collection of the sea anemone Heteractis aurora led to the identification of one novel class of toxin, denoted Hau I, with as-yet undetermined target(s), two new examples of toxins from the boundless β-hairpin family (Hau II and III) and a new example of a sea anemone type 1 sodium channel toxin (Hau IV).⁵¹³

8. Bryozoans

Brominated indole alkaloids, alternatamides B and D have been synthesised from 3,4-dibromoaniline via the intermediate 5,6-dibromotryptamine⁵¹⁴ and bryostatin 4 has been shown to inhibit the transforming growth factor (TGF)-β signalling pathway, responsible for disease progression in acute erthroleukaemia.⁵¹⁵

9. Molluscs

Compared to previous years, there was a significant drop in new MNPs reported from molluscs in 2024, with the two examples being bursatamide A 1268 and bursatellin B 1269 isolated from the sea hare Bursatella leachii.⁵¹⁶ A general review covering the nutraceutical and medical potential of marine molluscs has been published.⁵¹⁷ The potential of dolastatin MNPs and synthetic analogues as anticancer agents has been reviewed.⁵¹⁸ Exploration of the SAR of (−)-doliculide A led to the identification of some analogues with similar levels of cytotoxicity as the MNP towards the HCT-116 cell line.⁵¹⁹ Furodysinin, isolated from the nudibranch Hypselodoris tryoni, exhibited weak in vitro activity against wild-type Plasmodium berghei, but was considerable less active against an atovaquone-resistant strain.⁵²⁰

Tambjamine analogues show interesting results as antileishmanials, with potent intracellular activity against Leishmania mexicana and L. donovani, and with one analogue demonstrating partial protection in a rodent in vivo model.⁵²¹ The structures, synthetic approaches and SARs of the lamellarin alkaloids have been reviewed.^522,523,524

The global occurrence, toxicity and risk assessment of cyclic imine shellfish toxins have been reviewed.⁵²⁵ Analysis of scallops Argopecten irradians exposed to paralytic shellfish toxin producing dinoflagellate Gymnodinium catenatum identified that the toxins accumulate in the adductor muscle, leading to levels exceeding regulatory limits.⁵²⁶ Diatoms of the genus Pseudo-nitzschia produce the neurotoxin (DA), which in the case of king scallop contamination, can take many months to clear from the mollusc tissue. DA immunodetection has been used to examine the time course of subcellular location of the toxin, finding preliminary evidence of accumulation in autophagosomes in the digestive gland, which may go some way to explain the toxin’s persistence.⁵²⁷ Two fatty acid desaturase genes were identified in the dwarf surf clam Mulinia lateralis, the expression levels of which varied during embryonic development.⁵²⁸

Investigation of the transcriptome of the carnivorous snail Raphitoma purpurea has identified putative conotoxin-like peptides in both salivary gland and venom duct tissue, proving further implication for venom evolution in molluscs of the family Conidae.⁵²⁹ The predatory and defensive venoms associated with 16 different cone snail species have been reviewed,⁵³⁰ while another review has focused more on the effects of κ-conotoxins on potassium channels.⁵³¹ A cyclic octapeptide Am931, identified from analysis of the venom duct transcriptome of Conus amadis, appears to act as a catalyst promoting formation of the natively folded disulfide form of α-conotoxin ImI.⁵³² Replacement of L-lysine or L-arginine residues in α-conotoxin RgIA with their corresponding D-enantiomers has afforded weakly antifungal analogues.⁵³³

Six synthetic α-conotoxins, sequences of which were derived from transcriptome sequencing of venom of C. quercinus, exhibited potent cytotoxicity towards fall army worm cells as well as anti-larval toxicity.⁵³⁴ In two separate studies by the same research group, α-O-conotoxin GeXIVA[1,2] was shown to both reduce cancer chemotherapy induced neuropathic pain in a mouse model and to exhibit in vivo tumour growth suppression of triple-negative breast cancer.^535,536 A disulfide bond deleted and truncated conotoxin analogue, based on the sequences of α-O-conotoxins GeXIVA and GeXXVIIA, demonstrated retained potency of human α9α10 nAChR inhibitory activity, inhibited Ca_V2.2 channels and alleviated pain in a mouse model.⁵³⁷ Conotoxin kM-RIIIJ has been used to explore the physiological functions of hetereodimeric K_V1.2 channels, revealing that the receptors are highly expressed in proprioceptive (sensory motor) neurons in the dorsal root ganglion.⁵³⁸

10. Tunicates (ascidians)

Further investigation of an extract of an unidentified Didemnidae ascidian collected in Palau, afforded guanidine alkaloids 1-carboxydopargimine 1270, mellpaladines D–F 1271, 1272 and 1273, serodopalgimine 1274 and a 4,4’-dimer of serotonin 1275.⁵³⁹ Sulfoxide-containing mellpaladines D and E were isolated in equal amounts, with the authors suggesting they may be products of autoxidation. Mellpaladines D–F were evaluated for effects on behavioural activities in mice, with mellpaladine F being found to induce whole body convulsion.

The original proposed disulfide, homocoupled, cyclic hexapeptide structures of antatollamides A and B have been revised by total synthesis to be simpler monomeric cyclic peptides 1276 and 1277, respectively.⁵⁴⁰ The first enantioselective synthesis of (+)-cylindricine B has been reported.⁵⁴¹ The synthesis and multidrug resistance reversing activities of the pyrrole-containing ningalin MNPs have been reviewed,⁵⁴² as have the cytotoxic properties of cephalostatin and ritterazine dimeric steroidal pyrazine alkaloids.⁵⁴³ The human triggering receptor expressed on myeloid cells 2 (TREM2) is an important regulator of immunological responses. N-Deacetylshermilamine B, as well as a partially characterised mixture of glucosylceramides, isolated from the ascidian Cystodytes dellechiajei, bind to TREM2, with the pyridoacridine alkaloid also exhibiting weak in vitro cytotoxicity towards two multiple myeloma cell lines.⁵⁴⁴

Phosphoeleganin and one semisynthetic analogue decrease interleukin 6 secretion, suggestive of this compound class having potential in the management of diabetes.⁵⁴⁵ A further SAR study of phosphoeleganin has identified the importance of the phosphate and glycine groups, in combination with backbone carbinol stereochemistry, for inhibition of h-15-LOX-1 enzyme.⁵⁴⁶ Of a library of synthetic analogues of rubrolide-related γ-butenolides, a non-halogenated example exhibited weak to moderate cytotoxicity towards a panel of HTCLs, induced ROS production leading to apoptosis, arrested cell cycle at G₂/M and exhibited in vivo activity in a HeLa xenograft assay.⁵⁴⁷

A structurally simplified analogue of diazonamide A exhibited potent antiproliferative activity towards a panel of HTCLs which appeared to be related to inhibition of tubulin assembly, and the simplified analogue was less toxic to noncancerous cell lines.⁵⁴⁸ Screening a library of marine bacterium Tistrella strains revealed the majority were capable of producing didemnin B.⁵⁴⁹ The introduction of a second copy of the didemnin B BGC afforded a modified strain with excellent production capability (75 mg/L), enabling semisynthesis of a series of new analogues. A cytotoxic photoaffinity probe successfully enriched a large number of potential protein targets, including the previously noted depalmitoylating enzyme PPT1. Before being discontinued due to a lack of COVID-19 related hospitalisations, a Phase III trial of plitidepsin showed it to be generally well tolerated and suggested a two-day improvement in the median time to withdrawal of supplementary patient oxygen support.⁵⁵⁰ Screening a library of marine-derived compounds identified polycarpine as a potent inhibitor of the main protease M^pro of coronaviruses, also inhibiting SARS-CoV-2 and PEDV (porcine epidemic diarrhoea virus) viral replication in Vero-E6 cells.⁵⁵¹ The uses of the tetrahydroquinoline alkaloid trabectedin and its related analogue lurbinectedin in oncology, mechanisms of action, and uses in combination therapies have been reviewed.⁵⁵²

11. Echinoderms

Extracts of the arms of the starfish Thromidia catalai afforded three indole derivatives, catalindoles A–C 1278, 1279 and 1280, the structures of which were confirmed by synthesis.⁵⁵³ When collected, the starfish were found feeding on the sponge Theonella swinohei – LCMS analysis of extracts of the starfish digestive gland identified the presence of Theonella-sourced NPs while none were located in the arms, suggesting that the starfish does not accumulate sponge metabolites.

Steroidal disulfates 1281−1284 were isolated from a Far Eastern collection of the slime sea-star Pteraster marsippus.⁵⁵⁴ The biological activities of saponins isolated from sea-stars has been reviewed.⁵⁵⁵ Several examples of polyhydroxylated steroidal saccharides were also reported from starfish, including pectiniferosides A–J 1285−1294 from Patiria (=Asterina) pectinifera,⁵⁵⁶ spiculiferosides A–D 1295−1298 from Henricia leviuscula spiculifera,⁵⁵⁷ and ceramasterosides A, B, D and E 1299−1302, from Ceramaster patagonicus.⁵⁵⁸ Of these MNPs, only the latter, ceramasteroside D, exhibited a notable biological activity, of moderate ability to inhibit the release of nitrite from LPS stimulated microglial cells.

Ten examples of triterpenoid glycosides were reported from sea cucumber species, including pacificusosides L–Q 1303−1308 from Solaster pacificus,⁵⁵⁹ peronioside A 1309 from Psolus peronii,⁵⁶⁰ conicospermiumosides A₃-1, A₃-2, A₃-3, A₇-1 and A₇-2 1310−1314 from Cucumaria conicospermium,⁵⁶¹ and 1315−1318 from Apostichopus japonicus.^562,563 Investigation of extracts of the sea urchin Clypeaster humilis afforded sulfate 1319, methane sulfonate 1320 and boldine 1321, the latter two compounds being reported from a marine organism for the first time,⁵⁶⁴ while bibenzochromenones phanogracilins A–C 1322−1324 were isolated from the crinoid Phanogenia gracilis.⁵⁶⁵ Additional biological properties have been reported for echinochrome A, including anti-asthmatic activity⁵⁶⁶ and the ability to modulate migration of non-small cell lung cancer cells.⁵⁶⁷ 2-Butoxytetrahydrofuran, previously reported from Holothuria scabra, attenuates the aggregation and oxidative properties of α-synuclein, lessening its toxicity in a C. elegans model of Parkinson’s disease.⁵⁶⁸

The effect of the saponin frondoside A on inflammation associated with tooth replantation has been investigated.⁵⁶⁹ Holothurin B and related saponins are able to prevent mast cell degranulation, as measured by the release of β-hexosaminidase, and to alleviate the action of a number of pro-inflammatory cytokines.⁵⁷⁰ The chemistry and biological activities of MNPs reported from the sea cucumber Apostichopus japonicus (2000−late 2023) have been reviewed.⁵⁷¹ Holotoxin A₁, isolated from A. japonicus, inhibits candidiasis by a wide range of mechanisms including increased intracellular accumulation of ROS, mitochondrial membrane depolarisation, and inhibition of mitochondrial complex I.⁵⁷² Further examination of the triterpenoid glycoside cucumarioside A₂-2 has identified it to be active against prostate cancer cell lines, while cucumarioside A₀-1 and djakonovioside A are active against triple-negative breast cancer cell lines.^573,574

12. Miscellaneous

The structure of the bicycloheptane derivative (−)-raikovenal 1325, originally isolated from the marine ciliate Euplotes raikovi, has been confirmed by total synthesis, with the study also assigning absolute configuration.⁵⁷⁵ The potential of seagrass meadows as a source of bioactive molecules has been reviewed,⁵⁷⁶ while species specific reviews of Halophila stipulacea⁵⁷⁷ and Halodule uninervis⁵⁷⁸ covered the phytochemistry and pharmacological activities of metabolites. Further investigation of the diarylheptanoid heterodimer zosterabisphenone B determined it can induce apoptosis in HCT116 cells and reduce the growth of human colon tumour xenografts in mice.⁵⁷⁹

Five pyridyl and dihydroisoquinoline alkaloids, 1326−1330, were reported from the nemertean worm Amphiporus angulatus.⁵⁸⁰ Collections of the intertidal sipunculid worm Phascolosoma granulatum afforded phascolosomines A–F 1331−1336, characterized as linear guanidine amides.⁵⁸¹ A SAR study of the antimicrobial peptide capitellacin, originally isolated from the annelid worm Capitella teleta, identified the importance of one of the disulfide linkages for activity, that the linear peptide was devoid of activity, the D-enantiomer had reduced activity and that most analogues only exhibited weak, at best, activity towards E. coli.⁵⁸² Bioinformatic searching of the transcriptome of the lugworm Arenicola marina identified a new BRICHOS-related antimicrobial peptide, AmBRI-aaa, that was prepared using heterologous expression in E. coli.⁵⁸³ The 44 residue peptide exhibited weak to moderate activity towards some Gram-positive bacterial strains and whole genome sequencing of a resistant mutant identified targets associated with cell wall biosynthesis.

The ribbon worm Cephalothrix simula, known to contain tetrodotoxin, may be a source of toxin in British farmed Pacific oysters Magallana gigas.⁵⁸⁴ A truncated variant of a peptide, bolespleenin^334-347, derived from a gene product of the mudskipper fish Boleophthalmus pectinirostris, exhibited broad-spectrum antibacterial activity, inhibited biofilm formation and was active in a mouse model of skin infection.⁵⁸⁵ Further investigation of the peptide epinecidin-1, originally isolated from the orange-spotted grouper Epinephelus coioides, has determined its ability to suppress lipoteichoic acid-enhanced proliferation of non-small cell lung cancer cells and induce necrotic cell death via a mechanism involving mitochondrial damage and elevated levels of ROS.⁵⁸⁶ The 22-residue peptide piscidin-1, derived from mast cells of the hybrid striped bass Morone saxatilis x M. chrysops, induces apoptosis in oral squamous cell carcinoma cells, and also increases ROS levels leading to mitochondrial dysfunction.⁵⁸⁷

GC/electron capture negative ion-MS analysis of seventeen species of fish from the Seychelles in the Western Indian Ocean identified the predominant polyhalogenated compounds present were halogenated NPs of the methoxylated diphenyl ether class, rather than PCB or DDT anthropogenic persistent pollutants.⁵⁸⁸

13. Perspective – The emerging role of Artificial Intelligence in MNP research

The rapid uptake of artificial intelligence (AI) in commerce, health, transportation and communications is stunning. In the sciences AI is equally transformative. AI is used to analyze large amounts of different kinds of data, to find patterns and relationships, and use those to make predictions, decisions or generate “new” content. Machine learning (ML) is the most widely used approach, and it identifies patterns in data without relying on rules-based programming, allowing tasks to be performed automatically. When dealing with more complex, high-dimensional, unstructured data, a specialized subset of ML called deep-learning (DL) employs multilayered neural networks to identify intricate patterns.

Given that computer assisted analysis have been used for decades in MNP research, how much has the introduction of AI improved productivity in MNP research and increased the discovery of novel and/or potently active MNPs in the last decade?

Two main applications of AI are generative (GAI) and predictive (PAI). GAI is based on large language models (LLN) mostly trained on online data and analyzed using ML or DL algorithms to produce content copied from existing data but stitched together in different ways to construct a “new” amalgam as text, image or video that mimics the training data. GAI can aid researchers to write code for software packages such as Python and R, produce images for graphical abstracts or presentations, help construct grammatically correct text, and provide suggestions for areas to investigate when initiating research on an unfamiliar topic. But GAI needs to be used with caution because it has a propensity to generate wildly inaccurate results. For instance, creating images of chemical structures using GAI is particularly unreliable (Fig 2).⁵⁸⁹

Fig. 2.

Structures of three ubiquitous NPs (caffeine, taxol and penicillin G) and their “structures” generated by ChatGPT. The “structures” generated by Gemini and DeepSeek are presented in the ESI (fig S1)

PAI by contrast can analyse large amounts of orthogonal data to identify patterns that are often hidden or difficult to discern and predict new findings based on this pattern recognition.

The application of AI in the molecular sciences is mostly predictive rather than generative and this approach has resulted in important breakthroughs. The development of AlphaFold, an AI model to predict 3D protein structures based on amino acid sequence, for example, is transformative. Using a transformer neural network architecture trained on ~180 000 experimentally determined 3D protein structures, AlphaFold can now predict the 3D structures of over 200 million proteins.⁵⁹⁰ These predictions are now being incorporated into the drug discovery process.

Computational methods such as TDDFT calculations to assign absolute configurations to molecules, spectroscopic simulations, computer-assisted structure elucidation using rules-based algorithms, compound similarity and substructure searches, dereplication using spectral database matching tools and virtual screening of compound libraries through in silico protein docking are all computer assisted (but not AI) methods developed since the 1990s. Many of these methods are still widely used in NP research and most involve complex calculations that require supercomputers to do time-consuming analyses. The arrival of ML has created opportunities to massively reduce the time required to generate similar results without using supercomputers.^591−594

ML is now used as an aid in MS-based dereplication by predicting MS fragmentation patterns for NPs found in databases to augment the annotation of molecules in molecular networks that were previously only identified through matching experimental spectrometric data.⁵⁹⁵Augmenting lower level DFT calculations with ML generated parametric corrections has resulted in more accurate predictions of NMR data in minutes rather than days.^596,597 ML assisted genome mining has resulted in the discovery of BGCs and the prediction of their products.^598,599

MNP researchers still face many bottlenecks that impede productivity. These include access to biota, accurate taxonomic identification of biota, identification of “interesting” NPs especially those present in complex mixtures, separation and structure identification of NPs, identifying a NP’s bioactivity and understanding its mode of action, identifying the genes and proteins responsible for NP biosynthesis and synthesis of the NP for sustainable production or to generate analogues.

We analyzed 13 582 papers and reviews published in the field of MNP (obtained from MarinLit) over the last 10 years (January 2015-December 2024) to help understand how AI has contributed to solving these MNP research bottlenecks. Keyword searches (Table S1) of all abstracts allowed a connectivity map (using VOSviewer)⁶⁰⁰ to be generated by creating a thesaurus of keywords that we grouped into ontologies based on relatedness of the words. The limitation of this approach is that some abstracts do not necessarily include information regarding the use of AI, but the keyword connectivity maps can help to infer AI usage based on a combination of ontologies. A network from genome mining to bacteria, BGC, molecular network, predict, MS, bioinformatic, model and database for example infers AI is being used in microbial MNP research even though there is no explicit association between the use of AI terms and genome mining in the abstracts (Fig 3).

Fig. 3.

Connectivity map (created in VOSviewer)⁶⁰⁰of network linkages between keywords extracted from MNP paper abstracts. The size of a keyword node is proportional to the number of papers that contain that ontology and the width of the linkage to another node is proportional to the number of instances that the linkage occurs. Each of the top 21 nodes has a individual colour. A: total network; B: highlighted network from “genome mining”

We found that very few studies have used AI to aid in the discovery of new MNPs or bioactivities over the last decade. Overall, the rate of discovery of new MNPs (Fig 1) and potently active MNPs has remained static over the decade.⁶⁰¹

AI has only been explicitly referred to in 82 MNP papers over this period (Fig 4 and 5) and is most highly associated with the prediction of bioactivity through either in silico docking of MNPs to proteins or inferred bioactivity based on structural similarities. ML augmentation of DFT calculations for more accurate NMR chemical shift prediction (DU8ML),⁵⁹⁶ pattern recognition within 2D NMR spectra to predict structures (SMART2.0/DeepSAT),^602,603 pattern recognition within structures to predict NP structure classes (NPclassifier),⁶⁰⁴ prediction of peptide quaternary structures (conotoxins),⁶⁰⁵ and genome mining to generate hypothetical virtual libraries of potential structures of undiscovered molecules for MS molecular network matching (HypoRiPPAtlas)⁵⁹⁹ are examples of other uses. To date, none of these methods have been widely adopted.

Fig. 4.

Trends in number of papers using keywords in their abstracts from January 2015 to December 2024. A: total MNP papers, B: Papers discussing virtual screening and interactions of MNPs with proteins; C: papers discussing AI; D: papers discussing MS molecular Networking; E: Bioinformatics application in MNP research. The bars represent the total number papers.

Fig. 5.

Connectivity map (created in VOSviewer)⁶⁰⁰ of network linkages between keywords extracted from MNP paper abstracts. A: highlighted network from “artificial intelligence” ontology; B: highlighted network from “in silico” ontology.

In silico screening methods (233 papers) have steadily increased over the decade. There have been several studies that have used databases to predict MNPs active against specific targets (such as the SARS-COV 19 virus) through virtual screening. This approach previously used the knowledge of the protein structure (obtained mostly by XRD) and in silico docking techniques to identify molecules that interacted with the protein. However, AI methods now tend to use the knowledge gained from molecules that have been shown experimentally (or virtually through in silico methods) to interact with a specific target protein (or a structurally related protein) and these interactions are used to train a model to identify other compounds that will also interact. This pattern recognition approach is potentially quicker to implement and can be used to analyze large datasets. Unfortunately, where this approach has been adopted using MNPs, the verification of “hits” obtained mostly relies on virtual screening of the “hit” molecules through molecular docking experiments. This seems to be a circular approach given that in some cases, the training data was originally obtained through virtual screening. There has been limited examples where the virtual “hit” compounds have been tested in real laboratory assays.

One of the key advantages of studying marine biodiversity is the propensity to discover unprecedented chemistry. PAI learns from previous discoveries to identify known molecules and their close unknown analogues and to predict new bioactivities for these compounds based on their structural similarity to other molecules with demonstrated bioactivity against a known target. This has the potential to “de-orphan” known MNPs lacking any bioactivity data, but to date this has not occurred.

Linking biosynthetic genes with known NPs or predicting new NPs from bioinformatics associated with BGCs are also effective outcomes from PAI assisted research (BGC is mentioned 561 times over the decade). MNP research increasingly uses ML augmented molecular networking of LCMS data as an aid in the selection of biota to investigate or to link molecular network clusters to BGC predictions. Molecular networking has been discussed in 294 papers over the last 10 years, but invariably the selection of clusters has resulted in the isolation of closely related analogues of known compounds. So, based on this, one might expect that greater effort would now be focused on the unknowns with AI doing the heavy lifting in dereplication by sifting through the large volumes of “known” chemistry to identify the uniquely unknown metabolites that can then become the focus of new research. There is little evidence that this approach is being used yet but hopefully, the future use of this powerful tool can be tailored to highlight the unknowns rather than the knowns.

There is no evidence of an increase in new MNP novelty (i.e. structures possessing new ring systems or features not reported before; information from previous MNP reviews). The term “novel” has been used 2255 times and “unprecedented” 344 times, but virtually never in relation to either truly novel scaffolds or bioactivities. The lack of increased structural novelty might be related to diminishing returns that are anticipated when specific biota resources have been extensively studied.⁶⁰⁶ However, it could also reflect the sustained use of standard purification, biota selection, culture methods, or bioassay methods that unintentionally exclude certain structures. An example of this bias can be seen when normal-phase silica separation methods which were almost exclusively used until the early 1990s started to be supplemented with reversed-phase separation methods, thus expanding the latent chemical diversity accessible within well studied species (Fig 6). This resulted in a significant increase in average polar surface area of MNPs after the early 1990s that also reflected a difference in structural novelty. If a ML approach were adopted, it could potentially identify specific purification methods that have a higher chance of yielding novel chemistry.

Fig. 6.

Box plot of Calculated Polar Surface Area (PSA) for all new MNPs published each year between 1972 and 2024. Black horizontal lines represent median PSA.

There are still very significant pockets of marine biodiversity that remain unstudied. Small animals and plants (species with individuals occupying 1 cm³ or less) are rarely sampled but potentially account for most biodiversity in an area.^607,608 They likely occupy different niches where environmental pressures could drive differences in MNP expression. ML from ecological data could prove insightful. The propensity to isolate and culture microorganisms sourced from marine environments using media suited to fast growing, terrestrial soil-derived microbes, does not help to exploit truly marine biodiversity and their potentially unique chemical diversity. The choice of isolation media and culture media used to access truly marine microbes might benefit from an AI approach. Combining data from microbial ecology, biogeochemistry, aquaculture, food sciences, ecological studies and metagenomics for ML might help to uncover specific nutrient requirements that could enhance the isolation and cultivability of truly marine species.

MNP discovery continues to rely on serendipity. There is no evidence so far that AI has been applied to the biota selection process. At its most basic, MNP researchers have used prior knowledge to target specific species for study (soft bodied marine invertebrates, algae, fungi, bacteria, cyanobacteria). AI could aid in this process. However, we know from experience that the random collection of an invertebrate that was collected simply because it looked different or was found in an unusual place, has led to some of the most unusual molecules we have discovered. Perhaps an AI assistant on a SCUBA dive could help find that elusive organism.

Genome mining helps to identify the vast potential of BGCs (most of which are silent) in microorganisms, but what about the NPs that are not made by enzymes encoded by clustered genes? Pertinent examples are many types of alkaloids which represent a substantial proportion of MNPs and drugs derived from NPs. ML methods using Alphafold could provide insights to predict proteins from genomic data for non-clustered biosynthesis.

The structure determination of NPs has always been challenging, and continued reports of MNP structure revisions show that improved methods to aid in structure determination are always welcomed. Computationally demanding DFT methods to predict NMR chemical shifts have been augmented with parametric rules obtained through ML approaches using NMR data obtained from large libraries of published structures. This has resulted in much faster predictions and resulted in two MNP structure revisions.⁵⁹⁶ The tool can also be used to establish if a proposed structure is correct prior to publication. SMART2.0 is an image-based AI tool that predicts the structure of a compound based on the similarity of its HSQC image to those generated from a training set of over 53 000 experimental HSQC images or HSQC images generated through predictions of H and C chemical shifts for compounds lacking experimental data.⁶⁰² The four MNP studies that have used SMART2.0 to guide prioritization all resulted in the isolation of either close analogues of known compounds or a chimeric molecule containing features from two known molecules. Overall, these ML NMR methods can improve the structure determination workflow, but do not necessarily result in greater novelty of structures being published. Hopefully, it will reduce the number of incorrect structures appearing in the literature (if used judiciously at the time of peer review prior to publication).

Open Molecules 2025 (OMol25), released in mid-2025 by Meta and the US Department of Energy’s Lawrence Berkeley National Laboratory, contains 100 million DFT calculations covering a huge variety of molecular systems that can now be used in ML applications.⁶⁰⁹ This has the potential to massively reduce the time required to predict NMR and ECD data with estimates suggesting a 10-thousand-fold reduction in cpu time to generate equivalent quality predictions to those generated by DFT calculations. This resource has the potential, through ML algorithms, to predict NMR and ECD data for all known MNPs with comparable accuracy to quantum mechanical calculations virtually instantaneously on a laptop or smart phone. Linking this potential capability with other AI platforms such as 2D NMR image recognition, MS² data recognition and NP structure prediction from genomic data and NP bioactivity predictions could provide a paradigm shift in NP discovery efforts.

Since ML methods are trained on known data (such as molecules, NMR spectra, MS² data, genes) there is an obvious bias to match unknowns that mimic knowns. “Rare” structures might not feature in a training set, and undiscovered novel molecules clearly cannot be there at all. Therefore, important potential matches might be missing in the model. This is analogous to the inherent bias used by Amazon to recruit IT staff where female applicants were culled early in the automated selection process because the data used to train the algorithm did not include applications from women.⁶¹⁰ Training models to highlight and prioritize the unknowns will help to counter these biases.

There is a danger when AI is used as a “black box” method by non-experts who accept the results as gospel without any understanding or questioning because of a lack informed knowledge. If these unverified AI results are subsequently used to train new models a misinformation loop is exacerbated, rendering future models even less accurate than their predecessors. There is, therefore, an ongoing need for orthogonal testing of results obtained from AI tools in MNP research. In structure determination, this means that XRD and total synthesis will continue to be important verification tools. Laboratory-based bioassays need to be used to confirm or reject an AI computational result and MS² fragmentation matching for structure annotation will still rely on real MS data from standards.

There have been highly innovative AI methods developed, and some have been widely adopted while others have not. It is somewhat baffling to know why this is the case. For instance, molecular networking is now widely adopted with platforms such as GNPS and applications such as MZmine providing a simple computational engine for upload of raw data and download of clustered and annotated results.^611,612 The barrier to adopting this technology is low and labs using these platforms benefit from their simplicity to access. Equally as innovative is the SMART2.0/DeepSAT platform for interpretation of 2D NMR data to match known compounds and to predict structure classes for unknown compounds. However, the adoption of these methods by the MNP community is low (only four papers refer to SMART2.0 in their abstract). One reason may be that many MNP research groups have NMR “gurus” who can identify a known structure from limited experimental data extremely quickly and do not need to rely on computational methods. However, the most likely reason for slow uptake is that these methods still do not work well with NMR of mixtures (apart from matching major components). The time involved to purify an extract and then obtain HSQC data on individual molecules compared to injecting a minute quantity of an extract into an HPLC column and generating MS² data for thousands of compounds in minutes is likely to be a serious impediment.

Negative results are a major gap in published knowledge. For example, many MNP groups have large inventories of marine species collected or cultured randomly. These collections invariably contain individual samples that have been studied but either produce no MNPs or only known MNPs or the isolated MNPs showed low or no bioactivity in a range of bioassays. These results do not generally make it into research papers but are a gold mine of information that could be used in AI analysis. To make the most of all research outputs therefore requires an acceptable mechanism to capture open access, validated negative research results as well as positive ones.

Linking datasets will be key to seeing the most productive uptake and application of AI in MNP research. The World Register of Marine Species (WoRMs) contains up-to-date taxonomic classifications, collection records and distributions for over 500000 species,⁶¹³ the Ocean Biodiversity Information System contains 167 million geolocation records for over 200000 marine species and 27 million gene sequences,⁶¹⁴ citizen science initiatives such as iNaturalist have images of species linked to taxonomic data and geolocation with close to 280 million observations linked to 536000 species (including 4041 observations from 1280 marine invertebrate species),⁶¹⁵ the Lotus initiative provides taxonomic links to over 750000 NP structure/organism pairs,⁶¹⁶ the newly released Minimum Information about a Tailoring Enzyme (MITE) links biosynthetic enzymes with reaction type and precursor and product structures,⁶¹⁷ SMART2.0 uses a database of over 50000 HSQC images to link with structures, GNPS and other MS based platforms link MS² data to structures. Linking these disparate datasets could provide opportunities to guide new discoveries and to unlock novelty in plain sight.

This perspective has highlighted that MNP researchers are only beginning to embrace the power of AI and it will be fascinating to see where MNPs research in the AI era evolves over the coming decade.

14. Conclusion

New MNP discoveries in 2024 have shown an increasing prevalence of MNPs compounds derived from marine fungi and a decreasing trend in macro-organism studies. Although close to half of all new MNPs were reported from marine fungi in 2024, the majority are minor variations on already well represented structure classes. Actinomycetes are a minor contributor to overall marine procaryote biodiversity, but they continue to be the major source of new bacterial MNPs.

BCGs associated with MNPs obtained from bacteria, cyanobacteria and to a lesser extent fungi continue to be characterised, and genome mining is increasingly being used to find new bacterial MNPs. In some cases, these genomic approaches have resulted in proposed new MNP structures without comprehensive and convincing spectroscopic evidence and from our perspective these compounds have not been structurally verified. Rigorous spectroscopic characterisation including NMR spectroscopy and or XRD analysis should complement MS-based evidence in future studies.

The bioactivities reported for select MNPs continue to be a highlight, with the clinical development of a synthetic derivative of the Indonesian sponge-derived alkaloid PM742 holding promise for lung cancer patients. Synthetic studies continue to be an important technique to support or refute MNP structures and in 2024 over 120 first total syntheses of MNPs were accomplished leading to several structure revisions and over 10 absolute configurational assignments being established.

New sponge-derived MNPs had fewer than expected examples of peptides and alkaloids based on previous trends. Cnidarian studies were predominantly associated with organisms collected from Chinese, Vietnamese and Taiwanese waters in the South China Sea pointing to reduced emphasis on cnidarian biodiversity hotspots. Molluscs, ascidians, bryozoans, red, green and brown algae were, in total, only represented by a handful of studies, while echinoderms were mainly studied from Russian collections. These observations point to diminished geospatial and phylogenetic diversity of MNP discoveries. If these biodiversity trends continue, the overall relevance of MNP research will be diminished. That notwithstanding, the contributions of new MNPs to the structural diversity of organic chemistry remains high and they represent a valuable resource for biodiscovery.

Biographies

Anthony (Tony) Carroll initially studied the alkaloid and lignan chemistry of rainforest plants (BSc (Hons) and PhD, Prof Wal Taylor, Sydney University) but marine natural products became a major focus after postdoctoral fellowships at the University of Hawaii with Paul Scheuer and at James Cook University, Australia with John Coll and Bruce Bowden. Fifteen years as head of natural products chemistry for the AstraZeneca/Griffith University drug discovery project expanded his interests to include high throughput purification and structure determination techniques and cheminformatics. Since 2008 he has held a faculty position at Griffith University, Gold Coast where he is currently a Professor.

Brent Copp received his BSc (Hons) and PhD degrees from the University of Canterbury, where he studied the isolation, structure elucidation and structure–activity relationships of biologically active marine natural products under the guidance of Professors John Blunt and Murray Munro. Two postdoctoral positions with Jon Clardy at Cornell and Chris Ireland at the University of Utah were then followed with a period spent working in industry as an isolation chemist with Xenova Plc. In 1993 Brent returned to New Zealand to take a lectureship at the University of Auckland, where he is currently a Professor.

Tanja Grkovic received her MSc and PhD degrees from the University of Auckland under the supervision of Professor Brent Copp. She then carried out postdoctoral research at the National Cancer Institute with Kirk Gustafson, and Griffith University with Professor Ron Quinn. She is currently a Staff Scientist at the Natural Products Branch and the Molecular Targets Program at the National Cancer Institute where her research is focused on the generation of prefractionated natural product libraries as well as the isolation and structure elucidation of natural products sourced from marine, plant, and microbial biota.

Rob Keyzers carried out his BSc(Hons) and PhD studies at Victoria University of Wellington. His thesis research, carried out under the guidance of Assoc. Prof. Peter Northcote, a former contributor to this review, focused on spectroscopy-guided isolation of sponge metabolites. He then carried out post-doctoral research with Mike Davies-Coleman (Rhodes University, South Africa) and Raymond Andersen (University of British Columbia, Canada) before a short role as a flavour and aroma chemist at CSIRO in Adelaide, Australia. He was appointed to the faculty at his alma mater in 2009 where he is currently an Associate Professor.

Michèle Prinsep received her BSc (Hons) and PhD degrees from the University of Canterbury, where she studied the isolation and structural elucidation of biologically active secondary metabolites from sponges and bryozoans under the supervision of Professors Blunt and Munro. She undertook postdoctoral research on cyanobacteria with Richard Moore at the University of Hawaii before returning to New Zealand to take up a lectureship at the University of Waikato, where she is currently a Professor.