Canalesolide A, a Structurally Unique Polyhydroxy Macrolide from the Marine Cyanobacterium Okeania sp. with Potent Antitrypanosomal Activity

Abstract

The discovery of structurally novel natural products remains central to expanding biologically relevant chemical space, particularly within underexplored marine metabolite classes. Herein, we report the discovery and complete structural elucidation of canalesolide A, a new polyhydroxylated macrolide isolated from the marine cyanobacterium Okeania sp. The compound was identified through an integrated workflow combining phenotypic screening against Trypanosoma brucei and LC-MS/MS-based molecular networking, enabling rapid prioritization of bioactive fractions and dereplication of known metabolite families. Spectroscopic analysis revealed that canalesolide A belongs to the bastimolide-related class of macrolides but exhibits a distinct structural architecture. Its structure was established by integrating ultrahigh-resolution NMR spectroscopy, empirical configurational analysis of polyol systems, targeted model compound synthesis, and controlled chemical degradation and derivatization. This combined strategy resolved stereochemical motifs that were inaccessible by direct analysis of the intact macrolide alone, providing a transferable approach for assigning densely oxygenated marine macrolides. Genome mining identified the putative biosynthetic gene cluster and proposed biosynthetic pathway for a bastimolide-related macrolide. Canalesolide A displays potent, low nanomolar antitrypanosomal activity against human-infective subspecies of T. brucei with rapid elimination of parasites within 1 h at 1 μM. Although moderate mammalian cytotoxicity was observed, preliminary in vivo efficacy/toxicity studies in infected mice suggest a narrow therapeutic window highlighting the need for improved selectivity. This study expands the structural and biosynthetic diversity of polyhydroxylated macrolides and establishes a generalizable framework for resolving stereochemically complex natural products.

Introduction

The search for novel chemical scaffolds to combat infectious tropical diseases remains a critical challenge in drug discovery, including for human African trypanosomiasis (HAT), a lethal parasitic infection caused by Trypanosoma brucei, unless adequately treated. , Current therapeutic options, although improved with the recent progress of fexinidazole , and acoziborole, , face persistent threats from emerging drug resistance − and rely on a limited repertoire of molecular mechanisms. These concerns highlight the need to explore new sources of bioactive natural products, which, historically, have been an important source of antiparasitic drugs.

Marine ecosystems, particularly cyanobacteria, are a prolific source of structurally unique secondary metabolites that possess interesting biosynthetic architectures and potent biological activities. , These compounds frequently engage critical biological targets, including proteases , and the proteasome, , providing important leads for antiparasitic drug discovery. Beyond their biological activity, marine cyanobacterial metabolites are of considerable interest from a chemical perspective due to their unusual structural motifs and the synthetic and analytical challenges they present.

Among these, polyhydroxylated macrolides derived from marine cyanobacteria represent a structurally intriguing yet generally underexplored class of natural products. Members of this family, including bastimolides A , and B, , caylobolide A, , (iso)-caylobolide B and its acetylated analogs, , amantelides A and B, nuiapolide, and palstimolide A, are characterized by large macrocyclic frameworks densely decorated with hydroxyl groups, often arranged in repeating 1,5-diol and 1,3,5-triol motifs. This high degree of structural redundancy and stereochemical complexity present significant challenges for structure elucidation using conventional spectroscopic approaches. Indeed, several members of this class remain incompletely characterized, while others, e.g., caylobolide A, have required extensive structural revision through the combined use of advanced NMR methodologies, and total synthesis. ,

Here, we report the discovery and complete structural elucidation of canalesolide A (1), a new polyhydroxylated macrolide isolated from the marine cyanobacterium Okeania sp., that demonstrates potent antitrypanosomal activity. The compound was identified through an integrated workflow combining phenotypic screening and LC-MS/MS-based molecular networking, enabling rapid identification of bioactive fractions with selective antitrypanosomal activity. Spectroscopic analysis revealed that canalesolide A (1) belongs to the bastimolide-related family of macrolides but possesses distinct structural features. Building on recent advances in the analysis of polyhydroxylated macrolides, we sought not only to determine the complete structure of canalesolide A (1), but also to address a recurring stereochemical bottleneck in densely oxygenated macrolides. Our strategy integrated ultrahigh-resolution NMR analysis with empirical polyol configurational methods, targeted synthetic model compounds, controlled hydrolysis and PGME derivatization, enabling assignment of motifs that were not accessible by direct analysis of the intact macrolactone alone. Genome mining of the producing organism revealed a candidate biosynthetic gene cluster (BGC) encoding a modular type I polyketide synthase (PKS) with noncanonical features, representing the first reported putative BGC for a cyanobacterial bastimolide-related macrolide. Notably, the cluster contains signature biosynthetic elements, including an AprA-like loading module and conserved β-branching cassette resembling those of the apratoxin A pathway, consistent with installation of a tert-butyl-derived starter unit and characteristic methyl branch, respectively. Deviations from canonical collinearity, including module iteration and skipping, indicate that canalesolide A (1) is likely assembled via an unconventional PKS logic. Canalesolide A (1) exhibits low nanomolar activity against human-infective subspecies of T. brucei and moderate cytotoxicity across mammalian cell lines. In mice infected with T. brucei, the compound’s toxicity encourages further structure–activity relationship studies to optimize selectivity. Overall, the present findings expand the chemical space of polyhydroxylated macrolides and underscore marine cyanobacteria as a valuable source of chemically novel potential therapeutics.

Results and Discussion

Antitrypanosomal Screening and Molecular Networking-Guided Metabolite Profiling

To identify novel antitrypanosomal natural products, an in-house library of 2,822 marine cyanobacterial and algal extracts, as well as primary fractions obtained from crude extracts (CEs) by silica gel vacuum liquid chromatography (fractions A–I), was screened at 5 μg/mL for inhibition of growth of T. brucei by ≥70% (Figure S1). Forty-eight samples (1.7% of the total number) were active, and of these, nine (0.3%) inhibited the growth of HEK293 cells by ≤50% (Figure S1 and Table ).

1. Antitrypanosomal Activities against T. brucei and Cytotoxicity against HEK293 Cells of Selected Crude Extracts (CEs) and Fractions (Fractions A–I, Eluted with Progressively More Polar Solvents) .

		T. brucei		HEK293
Sample ID	Species	% inhibition (5 μg/mL)	EC50 (μg/mL)	% inhibition (5 μg/mL)	CC50 (μg/mL)	SI
1966H	Schizothrix sp.	100 ± 2	0.39 ± 0.04	16 ± 5	>5	>13
1774C	Leptolyngbya semiplena	92 ± 1	0.78 ± 0.22	25 ± 6	>5	>6
1701CE	Bostrychia sp.	100 ± 1	1.19 ± 0.56	22 ± 2	>5	>4
1647F	Symploca sp.	100 ± 0	1.44 ± 0.63	39 ± 4	>5	>3
1558E	Symploca sp.	85 ± 7	1.52 ± 0.39	10 ± 1	>5	>3
1738CE	Hydrocoelum sp.	100 ± 1	1.83 ± 0.60	0 ± 1	>5	>3
1944I	Okeania sp.	99 ± 5	2.25 ± 0.72	47 ± 5	>5	>2
1808I	Phormidium sp.	98 ± 6	2.40 ± 0.21	40 ± 5	>5	>2
1775I	Leptolyngbya sp.	100 ± 8	2.92 ± 0.85	0 ± 7	>5	>2
Pentamidine	NA	100 ± 2	0.01 ± 0.00	NA	>4	>400
Bortezomib	NA	100 ± 1	0.002 ± 0.00	100 ± 3	0.01 ± 0.00	5

^aEC₅₀ values are the means ± SD of three biological assays each performed in triplicate.

^bCC₅₀ values are means ± SD of two biological assays each performed in quadruplicate.

^cTested at 4 and 1 μM, respectively.

^dValues in μM.

^eNA: not applicable.

These nine samples were then advanced to concentration response tests with the parasite and HEK293 cells (Table and Figure S2). Against T. brucei, the EC₅₀ values ranged from 0.39 to 2.92 μg/mL, whereas against HEK293 cells, all of the compounds had CC₅₀ values >5 μg/mL. The most active antiparasitic fraction was 1966H with a selectivity index (SI = HEK293 CC₅₀/T. brucei EC₅₀) of at least 13 (Table ). The genera, Leptolyngbya and Symploca, had two active samples each, whereas Okeania, Schizothrix, Bostrychia, Hydrocoelum, and Phormidium had one each.

To prioritize candidate bioactive metabolites associated with the nine antitrypanosomal samples, GNPS-based molecular networking analysis was performed. LC-MS/MS-based metabolite profiling combined with GNPS molecular networking enabled the annotation of a diverse set of natural products across these highly active cyanobacterial extracts and fractions, with further support from PubChem and MarinLit database searches (Figures S3 and S4). This dereplication approach reinforces previous studies that demonstrate the utility of GNPS-guided workflows to identify bioactive metabolites, including antiparasitic compounds. −

Across the active materials, several known cyanobacterial compound classes were annotated by the GNPS library, including hoiamides, microcolins, hectochlorin, malyngamides, phormidolides and hormothamnin A/laxaphycin A (Figures S3 and S4). To evaluate the bioactivity of the annotated compounds, several available compounds and their analogs in the UCSD marine natural product library were tested against T. brucei and HEK293 cells, resulting in a partial assessment of their biological activities (Table ).

2. Activities of Compounds Annotated through GNPS Molecular Networking from Nine Bioactive Crude Extracts and Fractions against T. brucei and HEK293 Cells .

			T. brucei	HEK293
Sample ID	Compound	Class	EC50 (μM)	CC50 (μM)	SI
1966H	Microcolin E	Lipopeptide	>4	>4	>1
All samples except 1774C	Hoiamide A	Cyclic depsipeptide	0.44 ± 0.02	>4	>9
All samples except 1966H, 1774C, and 1944I	Hoiamide B		0.42 ± 0.17	>4	>9
1774C	Hectochlorin	Lipopeptide	0.08 ± 0.02	0.01 ± 0.00	0.12
1774C	Malyngamide C	Lipopeptide	>4	>4	>1
NA	Bastimolide A	Macrolide	0.14 ± 0.02	>4	>28
1944I	Canalesolide A (1)	Macrolide	0.05 ± 0.00	0.74 ± 0.04	15
NA	Pentamidine	NA	0.01 ± 0.00	>4	>400
NA	Bortezomib	NA	0.002 ± 0.00	0.01 ± 0.00	5

^aEC₅₀ values are the means ± SD of three biological assays each performed in triplicate.

^bCC₅₀ values are means ± SD of two biological assay each performed in quadruplicate.

^cNA: not applicable.

Microcolin E was annotated in fraction 1966H and tested against T. brucei and HEK293 cells, but showed no activity (Table ), consistent with the absence of reported antiparasitic activity for this class, despite known immunosuppressive and anticancer properties associated with its phosphatidylinositol transfer protein targeting. − Hoiamides were widely distributed across the extracts and fractions (Figure S3), and hoiamides A and B displayed submicromolar antitrypanosomal activity without detectable cytotoxicity in HEK293 cells (Table ), representing their first evaluation against T. brucei. Given their known activity on voltage-gated sodium channels, , these compounds may act through ion channel-related mechanisms, which is particularly relevant in the context of the unique ion transport systems present in kinetoplastid parasites. Hoiamide A is also reported to induce oxidative stress as well as stress to the endoplasmic reticulum secretory pathway. Hectochlorin was annotated in fraction 1774C and showed potent but nonselective activity against T. brucei and HEK293 cells (Table ): its known mechanism of action through actin hyperpolymerization , suggests a conserved cytoskeletal target in trypanosomes. Although a cluster of malyngamide analogs were identified in fraction 1774C, the compound tested, malyngamide C, was inactive against T. brucei (Table ), indicating that it does not contribute to the fraction’s bioactivity, in spite of reports that malyngamides exhibit cytotoxic and modest antiparasitic effects. , Although phormidolide A was annotated in fraction 1808I (Table ), the compound itself was not evaluated in our bioassays due to its absence from our marine natural products library of purified metabolites. Similarly, hormothamnin A/laxaphycin A were annotated in fraction 1775I, but were not evaluated as pure compounds due their absence from our library. These results are consistent with the limited evidence for antiparasitic activity within these structural classes, which are more commonly associated with antifungal and cytotoxic effects. − Overall, therefore, hoiamides A and B were the most selective antitrypanosomal candidates identified.

Discovery of a New Bastimolide-Related Macrolide, Canalesolide A (1), from Fraction 1944I

The LC-MS chromatogram of fraction 1944I was dominated by a single major compound (m/z 763.61 [M + H]⁺; Figure S5), suggesting that it might be responsible for the observed bioactivity (Table ). GNPS molecular networking analysis of fraction 1944I revealed that the node corresponding to this major compound was putatively annotated as bastimolide C (an annotated feature in the GNPS database that does not appear in the published literature). The mirrored MS/MS spectra between bastimolide C and the major compound showed high similarity, and another node within the same cluster was annotated as bastimolide A, suggesting that the major metabolite in this fraction likely belonged to the bastimolide family of macrolides (Figure S5). Caylobolide A, the only bastimolide-related macrolide that shares the same molecular weight (exact mass 762.59) as the major metabolite but is absent from the GNPS spectral library, was considered a possible candidate structure. , To determine whether this metabolite represented a known compound or a new analog, the major compound was isolated from fraction 1944I in good yield (45 mg). The ¹H NMR spectrum of the purified compound (Figures S10, S22, and S26) possessed an overall similarity to that of caylobolide A but displayed clear differences, indicating that the metabolite represented a new bastimolide-type compound. ,

Bastimolide-type compounds are a family of, so far, 11 members of polyhydroxylated macrolides from diverse geographic locations and which have been reported for the marine cyanobacteria genera Okeania, Lyngbya, Phormidium, Leptolyngbya and an unidentified genus within the family Oscillatoriaceae, as well as unidentified cyanobacteria (Figure ). ,,− These compounds can be broadly categorized according to the current status of their structural characterization. Bastimolides A and B represent the only members for which structures have been fully established in the original reports (Figure A). , Notably, bastimolide A was unambiguously elucidated through X-ray crystallographic analysis of its nona-p-nitrobenzoate derivative rather than through conventional NMR analysis. Bastimolide B, which shares the same linear precursor as bastimolide A but differs in the position of macrolactonization, was subsequently proposed based on LC-MS comparisons of hydrolyzed derivatives of bastimolides A and B together with biosynthetic considerations, and assuming an identical stereochemistry to that of the co-occurring bastimolide A.

1.

Cyanobacterial bastimolide-type polyhydroxylated macrolides categorized according to their current structural assignment status. (A) Fully elucidated structures (bastimolides A and B), − (B) structures revised through advanced spectroscopic and synthetic studies (caylobolides A and B, and related analogs), − and (C) structures that remain incompletely resolved (amantelides A and B, nuiapolide and palstimolide A). − The originating cyanobacterial taxa, year of report, and collection location are indicated for each compound. Adapted from ref . Available under a CC-BY-4.0 license. Copyright 2025 The Authors, published by Springer Nature. Adapted from ref . Available under a CC-BY-4.0 license. Copyright 2025 The Authors, published by Wiley-VCH GmbH.

Subsequent synthetic studies have provided further insight into the structures of these natural products. , Total synthesis of bastimolide A was accomplished, and comparison of the synthetic material with the natural product showed good agreement with the reported NMR data; however, the study did not attempt a full assignment of the NMR resonances. In contrast, a total synthesis study of bastimolide B enabled a more detailed NMR spectroscopic analysis, leading to the reassignment of several NMR chemical shifts within the polyol backbone, culminating in a complete assignment of all of the NMR resonances.

Caylobolides A and B provide another representative example illustrating the structural challenges associated with this class of polyhydroxylated macrolides (Figure B). Caylobolide A was first reported in 2002 from the marine cyanobacterium, Lyngbya majuscula. Its 2-dimensional structure was established by conventional NMR and MS analysis, revealing a 36-membered polyhydroxylated macrolactone containing multiple polyol motifs, including a pentad of 1,5-diols and a 1,3,5-triol. Although the relative configuration of the 1,3,5-triol unit was assigned using Kishi’s universal NMR database and several remote stereocenters were determined by Mosher’s analysis, much of the polyol backbone remained stereochemically unresolved. Caylobolide B was later reported in 2010 from a Phormidium sp., and its structure was proposed in a manner similar to that of caylobolide A, with only limited stereochemical assignments available for the polyol segments.

In 2025, comprehensive structural revisions of caylobolides A and B resolved these uncertainties through a combination of total synthesis and advanced spectroscopic analyses. , In these studies, ultrahigh-resolution NMR approaches, particularly by increasing the number of data points along the ¹³C dimension in 2D heteronuclear NMR experiments, significantly improved spectral resolution and peak separation, enabling full reassignment of the 2-dimensional structures of caylobolides A and B (Figure B). The absolute stereochemistry of caylobolide A was confirmed primarily through Mosher’s analysis and synthesis, whereas the stereochemical assignment of caylobolide B relied on biosynthetic considerations together with the NMR-based determination of relative configuration. During this structure revision process, several related analogs were also identified, including iso-caylobolide B and its acetylated derivatives, further expanding the structural diversity within this family.

Even following successful structural elucidation campaigns on closely related macrolides, stereochemical assignment within this class remains challenging, with members such as the amantelides, nuiapolide (possibly identical to amantelide A based on 1D NMR data in the literature), and palstimolide still incompletely resolved (Figure C). − Where complete assignments have been achieved, they have often required chemical manipulation beyond direct analysis of the native macrolide, including derivatization, degradation, synthesis or crystallographic analysis of modified analogs. ,,,, These precedents highlight the limited generality of purely nondestructive approaches and underscore the need for transferable strategies that combine high-resolution NMR analysis with targeted chemical and synthetic validation.

Building on these recent advances in the structural analysis of polyhydroxylated macrolides, e.g., caylobolide A, we applied an integrated structural elucidation strategy to characterize the new bioactive metabolite reported herein in 1944I, canalesolide A (1). This strategy combined ultrahigh-resolution NMR spectroscopy, empirical rule-based determination of polyol stereochemistry, including Kishi’s universal NMR database for 1,3,5-triol motifs and the Aggarwal-Butts empirical rule for 1,5-diols, together with synthetic model compounds to extend the analyses of 1,5-diols to tert-butyl-substituted 1,5-diols. In addition, base-mediated hydrolysis followed by PGME derivatization, was employed to establish the absolute configuration of canalesolide A (1).

Structural Elucidation of Canalesolide A (1)

Canalesolide A (1), obtained as a white amorphous powder, had a molecular formula of C₄₂H₈₂O₁₁ as determined by HRESIMS. An intense singlet in the ¹H NMR spectrum, integrating for nine protons, represented three equivalent methyl groups at δ_H 0.93 (9H, s, CH₃-39,40,41) (Table S1). Together with the associated ¹³C NMR resonance at δ_C 26.6 (C-39,40,41) and an HMBC correlation from δ_H 0.93 (CH₃-39,40,41) to a quaternary carbon at δ_C 35.1 (C-38), and an oxygenated carbon at δ_C 80.9 (C-37), the data indicated the presence of a tert-butyl carbinol moiety (Table S1). A methyl signal at δ_H 1.17 (3H, d, J = 6.5 Hz, CH₃-42) indicated the presence of a secondary methyl group, while a carbonyl carbon at δ_C 173.8 suggested the presence of an ester functionality in the molecule. Additionally, sequential losses of nine H₂O (9 × 18 amu) from the m/z 763 ion in the MS/MS spectrum suggested the presence of nine free hydroxy groups in the molecule (Figure S6). Consistent with this observation, the ¹H NMR spectrum displayed ten oxymethine proton signals in the region of δ_H 5.3–3.7, accounting for the nine hydroxyl-bearing methines together with the tert-butyl carbinol moiety. In addition, the ¹H NMR spectrum displayed signals in the aliphatic region (δ_H 2.8–1.3), many of which were highly overlapped, consistent with a number of aliphatic carbon resonances observed between δ_C 47–20 in the ¹³C NMR spectrum.

Using standard established NMR approaches, including analysis of HSQC, HSQC–TOCSY and HMBC spectra, two partial structures (parts 1a and 1b) were established (Figure B). However, severe signal overlap prevented further structural elucidation beyond these fragments. To address this limitation and determine the complete 2-dimensional structure, ultrahigh-resolution NMR experiments were performed, including pure shift multiplicity-edited HSQC and HSQC–TOCSY with mixing times of 30 and 60 ms, as previously demonstrated in the structure elucidation of caylobolides A and B. , In these experiments, increasing the number of data points in the indirect ¹³C dimension improved the digital resolution (spectral resolution of ∼1.29 Hz), allowing signals that appeared merged under standard acquisition conditions (spectral resolution of ∼24.41 Hz) to be resolved (Figure A). With the improved spectral resolution, detailed analysis of the polyhydroxylated carbon chain became possible. In particular, ultrahigh-resolution pure shift multiplicity-edited HSQC enabled the unambiguous assignment of one-bond ¹H–¹³C correlations for all protonated carbons, even in regions with severe signal overlap. Building upon these assignments, ultrahigh-resolution HSQC–TOCSY spectra recorded with mixing times of 30 and 60 ms revealed a series of sequential correlations between adjacent oxygenated methine and methylene carbons, allowing the extension of the spin systems beyond the partial structures initially proposed after performing NMR under standard conditions (Figure C). The shorter mixing time (30 ms) primarily revealed correlations between directly neighboring carbons (³ J _CH), whereas the longer mixing time (60 ms) enabled the observation of extended connectivity across multiple contiguous centers (⁴ J _CH). These correlations bridged the two initially identified fragments and established the continuous polyol backbone of the macrolide, thereby confirming the complete 2-dimensional framework of canalesolide A (1) (Figure B,C and Table S1).

2.

Comparison of multiplicity-edited HSQC spectra acquired under standard and ultrahigh-resolution conditions, and key 2D NMR correlations used to structurally elucidate canalesolide A (1). (A) Zoomed-in regions of multiplicity-edited HSQC spectra recorded under standard (top) and ultrahigh-resolution (bottom) conditions. Key 2D NMR correlations from standard NMR (B) and ultrahigh-resolution NMR (C) supporting the planar structure of canalesolide A (1).

With the 2-dimensional structure established, the relative configuration of the contiguous stereocenters along the polyhydroxy carbon chain was investigated using primarily empirical NMR rules developed for polyol systems (Figure ). − First, the relative stereochemistry of several 1,5-diol motifs was evaluated based on the chemical shift difference (Δδ_HaHb) between the central diastereotopic methylene protons of 1,5-diol segment (Table S1 and Figure A). As shown in Figure A, the Δδ_HaHb values for position C-7, C-11, C-15, C-23, C-27, and C-31 were readily measured by analyzing the ultrahigh-resolution HSQC spectrum. This allowed a direct comparison of the chemical shifts of the central diastereotopic methylene protons through their correlations to their corresponding carbon signals. According to the empirical rule recently established by Aggarwal and Butts, Δδ_HaHb values ≥0.17 ppm indicate a syn relationship, whereas values ≤0.13 ppm correspond to an anti 1,5-diol configuration. This allowed assignment of the 5,9/9,13/25,29/29,33-diols to be in a syn configuration whereas the 13,17/21,25-diols are in an anti configuration. However, the C-33/C-37 relationship could not be assigned reliably by this empirical rule, because C-37 is incorporated into the lactone ester and the adjacent tertiary alkyl substituent complicates direct application of the free 1,5-diol model. Accordingly, this assignment was deferred and resolved through comparison with synthetic probes, as described below.

3.

Determination of the relative configuration of canalesolide A (1). (A) Assignment of the relative configurations of the 1,5-diol systems based on Δδ_HaHb values of the central methylene protons measured from the HSQC spectrum. (B) Determination of the stereochemical relationship of the 1,3,5-triol system by comparison with Kishi’s universal NMR database and related bastimolide-type macrolides. (C) Assignment of the 5-hydroxy-3-methyl fragment by comparison of Δδ_HaHb values with model fragments F1 and caylobolide A. (D) Determination of the tert-butyl-substituted C-33,37 1,5-diol/hydroxylactone configuration by comparison of seco-canalesolide A (2) with model fragments F2 and bastimolide B. (E) Overall relative configuration of canalesolide A (1).

Next, we examined the relative stereochemistry of the 1,3,5-triol motif (C-17/C-19/C-21) in canalesolide A (1). To apply Kishi’s universal NMR database for polyol systems, the ¹³C NMR spectrum of canalesolide A (1) was analyzed in DMSO-d ₆, and the chemical shift of the central carbinol carbon (C-19) within the 1,3,5-triol unit appeared at δ_C 66.0 (Figure B). In Kishi’s database, the central carbinol carbon resonances of anti/anti triols cluster around δ_C 64, those of anti/syn and syn/anti triols around δ_C 66, and those of syn/syn triols around δ_C 68 in DMSO-d ₆. , Comparison of this value with those reported in Kishi’s universal NMR database suggested that the stereochemical relationship of this 1,3,5-triol motif could correspond to either an anti/syn (δ_C 66.0) or syn/anti (δ_C 65.9) arrangement (Figure B).

To further refine this assignment, the ¹³C NMR chemical shifts obtained in pyridine-d ₅ were compared with those reported for structurally related bastimolide-type macrolides. , In bastimolide A and caylobolide A, the relative configuration of the corresponding 1,3,5-triol motifs has been firmly established and their ¹³C NMR chemical shifts in pyridine-d ₅ were unambiguously assigned (Table S1). The chemical shifts observed for C-17, C-19, and C-21 of canalesolide A (1) closely matched those reported for the analogous positions in bastimolide A and caylobolide A (Figure B and Table S1). This comparison allowed confident assignment of the stereochemical relationship of the 1,3,5-triol unit of canalesolide A (1) to be anti/syn, summarized in Figure B. Further, the stereochemical assignment of the 1,3,5-triol unit also served to connect the two stereochemically defined fragments established from the 1,5-diol analyses (Figure A,B). While the central methylene protons Δδ_HaHb analysis of 1,5-diols described above allowed the relative configurations within the C-5–C-17 and C-21–C-33 sections to be determined independently, the stereochemical relationship of the central C-17/C-19/C-21 triol unit bridged these fragments. Consequently, the continuous relative configuration along the polyol chain from C-5 to C-33 could be established (Figure A,B).

The relative stereochemistry of the 5-hydroxy-3-methyl group in canalesolide A (1) was next evaluated by comparison with model fragments F1 (anti and syn, (3R)-S7 and (3S)-S7 in the original paper) prepared for caylobolide A total synthesis (Figure C). The chemical shift difference between the central diastereotopic methylene protons (Δδ_HaHb) adjacent to the stereogenic center was shown to be diagnostic when measured in pyridine-d ₅ (Figure S30). In the anti-F1, the Δδ_HaHb value displays a substantial chemical shift difference (Δδ_HaHb ∼ 0.4 ppm), whereas the syn-F1 exhibits nearly identical chemical shifts (Δδ_HaHb ∼ 0 ppm) (Figures C and S30). The Δδ_HaHb value observed for the corresponding methylene group at C-4 in canalesolide A (1) closely matches that of the anti-F1 and is consistent with the value reported for C-4 of caylobolide A measured under identical pyridine-d ₅ conditions (Figure C). These results indicate that the 5-hydroxy-3-methyl unit in canalesolide A (1) adopts the same anti relative configuration as that established for caylobolide A (1).

Lastly, we turned our attention to the problematic case of the tert-butyl-substituted C-33,37 1,5-hydroxylactone (Figure D). The relative configuration of this unit was initially investigated by comparison with bastimolide A, which possesses an analogous terminal tert-butyl-substituted macrolactone motif. Because both compounds originated from marine cyanobacteria and share a closely related structural framework, it was considered likely that the corresponding stereochemical relationship would be conserved. Accordingly, the NMR spectroscopic data of canalesolide A (1) were compared with those reported for bastimolide A (Figure D and Table S1). Contrary to the expectation that the two compounds would exhibit closely similar NMR chemical shift patterns, a direct comparison proved inconclusive. Although both compounds were analyzed in pyridine-d ₅, the Δδ_HaHb values of the central methylene protons of the 1,5-hydroxylactone differed between the two molecules (bastimolide A: Δδ_HaHb ∼ 0.04 ppm; canalesolide A: Δδ_HaHb ∼ 0.09 ppm; Figure D). In DMSO-d ₆ and MeOH-d ₄, canalesolide A (1) displayed the oxymethine proton at δ_H 4.63 (dd, J = 10.5, 3.0 Hz) and at δ_H 4.76 (dd, J = 10.5, 2.0 Hz), respectively, closely matching the signal for bastimolide A at δ_H 4.65 (dd, J = 10.4, 2.4 Hz, DMSO-d ₆) (Figures S22 and S26) , However, in pyridine-d ₅, the signal for canalesolide A (1) appeared at δ_H 5.05 (dd, J = 6.5, 6.5 Hz), clearly different from the signal reported for bastimolide A at δ_H 5.09 (dd, J = 10.4, 2.4 Hz, pyridine-d ₅; Table S1 and Figure S31). These discrepancies suggest that direct comparison of the macrocyclic structures may not provide a reliable basis for stereochemical assignment, as the macrolactone framework may adopt different conformations depending on the solvent, leading to variations in the observed NMR parameters.

In order to eliminate the possibility that these observations were due to the macrolactone form, the macrolactone ring of canalesolide A (1) was cleaved via NaOMe-mediated hydrolysis to generate seco-canalesolide A (2) (Figure S32), enabling comparison with acyclic model fragments F2 (anti and syn forms of tert-butyl-substituted-1,5-diols) synthesized in this study (compounds 5 and 6 in the “model compound F2 synthesis” section of the Supporting Information and corresponding to syn-F2 and anti-F2, respectively; Figures D and A). This strategy follows precedent from bastimolide studies wherein acid-mediated hydrolysis of bastimolide A was unsuccessful, presumably due to steric hindrance from the bulky tert-butyl-substituent, whereas base-promoted hydrolysis successfully produced the corresponding seco derivative.

4.

Determination of the absolute configuration of canalesolide A (1). (A) Base-mediated hydrolysis of canalesolide A (1) to afford seco-canalesolide A (2). (B) PGME derivatization of seco-canalesolide A (2) and analysis of Δδ_(R)‑(S) values for the β,β-disubstituted carboxylic acid moiety. (C) Absolute configuration of canalesolide A (1) based on the PGME analysis.

The stereochemistry of the tert-butyl-substituted 1,5-diol unit was then evaluated by comparing the NMR data of seco-canalesolide A (2) with those of synthetic model fragments F2 representing the anti and syn configurations, as well as with bastimolide B, which contains a free linear tert-butyl-substituted anti-1,5-diol as well as model compound anti-F3 (compound 21 in the original paper) reported in the total synthesis studies of bastimolide B (Figure D and Table S2). The Δδ_HaHb value of the central methylene protons in seco-canalesolide A (2) (Δδ_HaHb ∼ 0.21 ppm) closely matched those observed for the anti-F2 (Δδ_HaHb ∼ 0.21 ppm), which is the same as bastimolide B (Δδ_HaHb ∼ 0.26 ppm) and anti-F3 (Δδ_HaHb ∼ 0.20 ppm), rather than syn-F2 (Δδ_HaHb ∼ 0.58 ppm; Figure D and Table S4). Furthermore, comparison of the ¹H and ¹³C NMR chemical shifts of all carbons and protons within the tert-butyl-substituted 1,5-diol motif revealed strong agreement between seco-canalesolide A (2), bastimolide B, and anti-F3 (all measured in pyridine-d ₅; Table S2). These observations collectively support assignment of the anti-relative configuration of this C-33,37 stereochemical unit (Figure D). This analysis of the synthetic tert-butyl-substituted 1,5-diol model systems supports extension of the Δδ_HaHb method for 1,5-diol originally developed by Aggarwal and Butts to sterically demanding α-tert-butyl-substituted variants, as detailed in the “model compound F2 synthesis” section of the Supporting Information (Figures S47–S49 and Tables S4–S8). Collectively, these comprehensive analyses established the relative configuration of canalesolide A (1) as 3S*, 5R*, 9R*, 13S*, 17S*, 21R*, 25S*, 29R*, 33R*, 37S* (Figure E).

To determine the absolute configuration of canalesolide A (1), the PGME derivatization method, a modified Mosher-type analysis applicable to carboxylic acids, was employed. Because seco-canalesolide A (2) had already been obtained through base-mediated hydrolysis of 1, this linear derivative bearing a carboxylic acid group was directly suitable for PGME analysis (Figure ). Consistent with previous reports demonstrating that the PGME method is applicable not only to α,α-disubstituted but also to β,β-disubstituted carboxylic acids, both (R)-PGME and (S)-PGME derivatives of 2 were prepared to determine the absolute configuration at C-3 (Figure B). − The ¹H NMR chemical shifts of the protons surrounding C-3 were assigned based on analysis of ¹H and COSY NMR data, and their Δδ_(R)‑(S) values generally exhibited a consistent sign distribution (Figures B and S37–S39 and Table S3). Although the diastereotopic methylene protons at C-4 showed opposite Δδ_(R)‑(S) signs, this behavior can be rationalized by their differential placement within the anisotropic shielding and deshielding regions of the PGME phenyl ring and, therefore, does not interfere with the overall Δδ_(R)‑(S) pattern. Importantly, the large and opposite Δδ_(R)‑(S) values observed for the 42-methyl group and H-4a provide a clear diagnostic pattern supporting assignment of the S configuration at C-3 (Figure B). Because the relative configurations of all stereocenters in canalesolide A (1) were determined as described above, assignment of this stereocenter enabled determination of the complete absolute configuration of canalesolide A (1) as 3S, 5R, 9R, 13S, 17S, 21R, 25S, 29R, 33R, 37S (Figure C). Notably, the resulting stereochemical pattern closely resembles those reported for bastimolide A and caylobolide A at the corresponding positions, further supporting the proposal that these polyhydroxylated macrolides have related biosynthetic origins. ,

Proposed Biosynthesis of Canalesolide A (1)

To investigate the biosynthetic origin of canalesolide A (1), metagenomic sequencing of the RNAlater-preserved sample (PAC17FEB10–2) was performed using the PacBio Revio platform following our previously reported workflow. The HiFi reads were assembled and processed to generate a complete cyanobacterial metagenome-assembled genome (cMAG) with a total genome size of 10.40 Mb. The genome consisted of a single contig and showed high quality, with 99.85% completeness and 0.07% contamination. The contig had an average read depth of 345× and an average gene length of 323.85 bp. Taxonomic classification using GTDB-Tk (v2.1.1) assigned the genome to Okeania sp., with a GC content of 37%. Although the 16S rRNA gene sequence of PAC17FEB10–2 (NCBI accession number HQ900689) from the same sample was originally reported as cf. Oscillatoria margaritifera based on the phylogenetic framework available at the time of publication (2011), current NCBI taxonomy reassigns this sequence to Okeania sp. This updated classification aligns with the genome-based result obtained in this study and supports the reliability of the MAG identification.

Genome mining of the Okeania sp. MAG using antiSMASH 8.0.4 revealed several biosynthetic gene clusters (BGCs), among which a large modular type I PKS gene cluster is proposed to encode the biosynthesis of canalesolide A (1) (Figures A and S40). The cluster begins with a loading module (CanA) highly homologous to AprA (90.1% sequence identity to OQ942863.1 by NCBI BLASTp), comprising the characteristic MT1−ΨGNAT–MT2–ACP architecture responsible for formation of the pivaloyl starter unit in apratoxin A biosynthesis (Figures B and S41). , Consistent with this assignment, the proposed pathway accounts for the characteristic tert-butyl-substituent present in canalesolide A (1). The downstream assembly line contains alternating β-modules (KS–AT–KR–ACP) and δ-modules (KS–AT–DH–ER–KR–ACP), consistent with biosynthesis of the highly oxygenated polyketide backbone containing repeating 1,5-diol and 1,3,5-triol motifs. In addition, the BGC encodes a conserved HMG-CoA-synthase (HCS)-type β-branching cassette, where it comprises a tandem ACP_A tridomain (ACP_I–III), a discrete ACP_D, an HCS, a nonelongating KS-like domain (KS⁰), a dehydratase (ECH₁), a decarboxylase (ECH₂) and an enoyl reductase (ER) (CanG–L), closely resembling the corresponding cassette in the apratoxin A pathway (Figures B,C and S41). , Consequently, this organization is consistent with introduction of a β-methyl branch, corresponding to the C-3 methyl substituent observed in canalesolide A (1).

5.

Overview of the putative canalesolide A (1) BGC and proposed noncanonical biosynthetic pathway. (A) Organization of the canalesolide A BGC, (B) proposed biosynthetic pathway of canalesolide A, (C) proposed β-branching mechanism, and (D) MAFFT-based sequence alignment of KR domains within the cluster visualized in Geneious Prime (v 2025.0.1), highlighting conserved LDD and W motifs associated with stereochemical control and supporting assignment of KR domain stereospecificity. Abbreviations: KS, ketosynthase; KS⁰, nonelongating ketosynthase-like domain; AT, acyltransferase; ACP, acyl carrier protein; KR, ketoreductase; DH, dehydratase; ER, enoylreductase; MT, methyltransferase; ΨGNAT, nonfunctional GNAT-like domain; HCS, 3-hydroxy-3-methylglutaryl-CoA synthase; ECH₁, enoyl-CoA hydratase homologue acting as a dehydratase; ECH₂, ECH acting as a decarboxylase; C, condensation domain; A, adenylation domain.

Following the elongation modules, the assembly line terminates with a KS–AT–ACP–C–A module (CanL, Figure B). Degeneration of the conserved KS catalytic motif suggests that this KS functions as a nonelongating KS-like domain, marking the transition from chain extension to product release (Figure S42). In the absence of canonical thioesterase or reductase domains, the terminal C–A architecture is proposed to mediate product release and macrocyclization, analogous to the mechanism proposed for apratoxin A. The BGC also contains two putative acetyltransferase genes (canM and canN), consistent with the detection of minor acetylated derivatives of canalesolide A (1) by LC-MS in the producing organism (Figures A and S44). Detailed bioinformatic analysis of this putative canalesolide A (1) BGC are provided in the Supporting Information.

Despite the overall collinear organization of the PKS assembly line, a clear discrepancy exists between the number of elongation modules encoded within the putative BGC and the carbon framework of canalesolide A (1). Such discrepancies have been increasingly recognized as hallmarks of noncanonical modular PKS assembly, in which the canonical “one-module-one-extension” cycle paradigm is relaxed through mechanisms such as iterative module reuse and selective module or domain skipping. , In characterized iterative systems, downstream KS domains function as gatekeepers that selectively accept intermediates only after a defined chain length has been reached. − Module- and domain-skipping events have likewise been reported, further diversifying the biosynthetic outcomes of modular PKSs. , To evaluate this possibility, the stereochemical outcomes predicted from the KR domains were systematically compared with the experimentally assigned configurations of the hydroxy-bearing stereocenters in canalesolide A (1). KR stereochemistry was assigned based on the conserved LDD and W motifs characteristic of B- and A-type KR domains, respectively (Figure B,D). These conserved motifs reliably predict β-hydroxy stereochemistry in modular cis-AT PKSs, with A-type and B-type KRs generally producing l- and d-configured alcohols, respectively, as validated by large-scale bioinformatic analyses of experimentally characterized systems.

Consistent with these predictions, the tert-butyl-substituted hydroxy group at C-37 was assigned as having a d-configuration, in agreement with a B-type KR domain in module 1 of CanB. Downstream of this position, the repeating 1,5-diol motifs extending toward C-25 display a recurring pattern of l-configured hydroxyl groups at C-33, C-29 and C-25. Given the limited number of δ- (KS–AT–DH–ER–KR–ACP) and β- (KS–AT–KR–ACP) modules encoded within the BGC, this pattern cannot be readily explained by a strictly collinear assembly process. Instead, these observations are most consistent with a bimodular iteration involving module 2 (CanC, δ-module) and module 3 (CanE, β-module, A-type KR domain), which are proposed to undergo three rounds of iterative chain extension, accounting for the repeated installation of l-configured 1,5-diol units from C-33 to C-25. An additional iteration of the δ-module (module 2, CanC) is proposed, resulting in a fourth elongation event prior to downstream transfer. The resulting elongated intermediate is then transferred to the downstream CanD region (modules 4–6), whose KR domains are predicted to follow a B-, B- and A-type sequence, in agreement with the stereochemical configuration of the C-21, C-19 and C-17. In this context, the first KS domain of CanD is likely to function as a gatekeeper, selectively accepting the intermediate only after the preceding iterative elongation steps have been completed, thereby contributing to the controlled progression of the biosynthetic assembly line. Following this stage, the repeating pattern of d-configured 1,5-diol units (C-13, C-9 and C-5) spanning C-13 to C-5 likewise requires selective module utilization. We therefore propose that module 7 (CanF β-module) is bypassed, whereas a bimodular iteration involving modules 8 (CanF δ-module) and 9 (CanG β-module, B-type KR domain) proceeds through three rounds of chain extension to generate the C-14 to C-3 segment. These KR-derived stereochemical analyses support the proposed structure of canalesolide A (1) and further suggest a noncollinear biosynthetic mechanism involving selective bimodular iteration and module skipping. Subsequently, module 10 (CanG) is proposed to serve as an additional gatekeeping checkpoint, which likely regulates downstream chain transfer by selectively accepting only fully processed intermediates, thereby reinforcing controlled progression through the latter stages of the assembly line (Figure ).

Mechanistic precedents for such a biosynthetic model have been established in several experimentally characterized modular PKS systems. , Iterative use of modules within otherwise modular PKS systems has been documented, whereas hybrid PKS–NRPS systems such as alpiniamide demonstrate that selective module engagement and recombination of intermediates can generate products through noncollinear assembly. More generally, these noncanonical behaviors are thought to arise from “metabolic decision points” within the assembly line, including selective intermodular recognition and KS-mediated gatekeeping that regulate substrate routing and chain transfer, as exemplified by azalomycin biosynthesis. , Although the proposed biosynthetic pathway for canalesolide A (1) remains hypothetical, the convergence of BGC architecture, KR-derived stereochemical predictions, and the experimentally determined structure provides strong biosynthetic support for the proposed assembly model. Experimental validation will require future biochemical and genetic studies, including enzymatic characterization, heterologous expression, and targeted mutagenesis, to test the proposed roles of module iteration, module skipping, and intermodular chain transfer.

Biological Properties of Canalesolide A (1)

In addition to the nanomolar activity measured for the animal-infective T. b. brucei subspecies (EC₅₀ = 0.05 ± 0.00 μM), canalesolide A (1) demonstrated the same potency against the two subspecies responsible for infection in humans, namely T. b. gambiense (0.07 ± 0.01 μM) and T. b. rhodesiense (0.07 ± 0.02 μM; Figure S45). Further, a time-kill assay that measures the speed at which the parasite dies, revealed that canalesolide A (1.0) rapidly kills T. b. brucei in 1 and 4–8 h at 1 and 0.5 μM, respectively. This is in contrast to the HAT drug, pentamidine, which required ∼24 h at 1 μM for complete kill (Figure S46). In mammalian cell lines, 1 displayed moderate cytotoxicity against HEK293 cells (CC₅₀ = 0.74 ± 0.04 μM, SI = 15; Figure S45), with comparable profiles for medulloblastoma, D283-med (CC₅₀ = 0.95 ± 0.16 μM, SI = 19), glioblastoma, SF188 (CC₅₀ = 0.84 ± 0.10 μM, SI = 17) and nonsmall lung carcinoma, NCI-H460 cells (CC₅₀ = 0.69 ± 0.23 μM, SI = 14; Figure S47). Structurally related cyanobacterial macrolides, such as caylobolides, have previously shown modest cytotoxicity; for example, caylobolide A against HCT116 cells (CC₅₀ = 9.9 μM), caylobolide B against HT29 and HeLa cells (CC₅₀ = 4.5 and 12.2 μM, respectively), and iso-caylobolide B that was ∼2-fold more potent than caylobolide B vs HCT116 cells.

Other polyhydroxylated macrolides have also demonstrated antiparasitic activity, particularly against the malaria parasite, Plasmodium falciparum. For instance, palstimolide A was potent against the asexual blood stage of P. falciparum (EC₅₀ = 0.17 μM) with low toxicity toward HepG2 cells (CC₅₀ = 5.04 μM) while also being active against intracellular Leishmania donovani. Bastimolides A and B similarly display antimalarial activity, although their mechanisms of action remain unresolved. , Other related polyol macrolides, such as amantelide A, exhibit broader bioactivity, including antifungal effects mediated by interactions with sterol-containing membranes and pore formation, reminiscent of polyene antifungals. , The hydroxy group at C-33 of amantelide A likely plays a key role in the compound’s distinctive membrane-binding behavior, as the C-33 acetoxy derivative, amantelide B, loses both sterol-dependent membrane affinity and biological activity. Because canalesolide A (1) also contains a hydroxy group at C-33, in addition to sharing the polyhydroxylated macrolide framework found in bastimolides, amantelides, palstimolide and caylobolides, we hypothesize a similar membrane-targeting mode of action. This is supported by the time-kill experimental data, which demonstrate very rapid parasiticidal activity consistent with, although not diagnostic of, membrane disruption or sterol sequestration.

If canalesolide A (1) indeed exerts its antitrypanosomal activity through membrane perturbation, then this mechanism may have significant implications for drug resistance. An important next step, therefore, will be to determine whether canalesolide A (1) retains activity against drug-resistant T. brucei isolates and whether resistance mechanisms associated with current HAT drugs influence susceptibility to this compound. Cross-resistance with current HAT drugs is not expected because the principal resistance mechanisms identified to date involve altered drug uptake or impaired metabolic activation rather than membrane perturbation itself. For example, resistance to pentamidine and melarsoprol is predominantly associated with alterations in the aquaglyceroporin, TbAQP2, − whereas resistance to eflornithine results from loss of the amino acid transporter, AAT6. Also, resistance to fexinidazole has been linked to impaired nitroreductase-mediated activation, , and experimental resistance to acoziborole has been associated with mutations in the endonuclease, CPSF3. , Canalesolide A (1) is a large, neutral, densely polyhydroxylated macrolide that is structurally distinct from the small cationic substrates transported by these uptake systems, ,, further supporting the possibility that its trypanocidal activity may not rely on the resistance mechanisms currently associated with approved HAT therapies. Overall, these observations suggest that canalesolide A (1) represents a mechanistically distinct antitrypanosomal scaffold for future HAT drug discovery. Membrane-acting compounds have often been associated with a higher genetic barrier to resistance compared to agents that act on single protein targets, , although this is not a universal property. , Whether canalesolide A (1) exhibits a high genetic barrier to resistance remains unknown and will require experimental evaluation through in vitro resistance-selection studies.

Given its close structural similarity to canalesolide A (1) and its reported potency against P. falciparum (EC₅₀ = 0.08–0.27 μM), bastimolide A was obtained from the UCSD marine natural product library for comparative evaluation. The compound also demonstrated activity against T. brucei (EC₅₀ = 0.14 ± 0.02 μM) but, notably, lacked detectable cytotoxicity toward HEK293 cells (SI ≥ 28, Table ), indicating improved selectivity relative to canalesolide A (1). Against other kinetoplastid parasites, including T. cruzi and L. donovani, bastimolide A showed moderate activity (EC₅₀ = 6.5 and 3.0 μM, respectively), alongside toxicity toward the mammalian cell lines, MCF-7 and Vero cells (CC₅₀ = 3.1 and 2.1 μM, respectively). Together, these findings highlight canalesolide A (1) and bastimolide A as promising antitrypanosomal scaffolds and support further investigation into their molecular targets and potentially shared mechanisms of action. However, a preliminary in vivo evaluation of 1 in a mouse model of T. brucei infection revealed a narrow therapeutic window with no efficacy at 5 mg/kg, and mortality of infected animals within 24 h at 10 mg/kg, whereas the maximum tolerated dose in uninfected mice was 10–15 mg/kg. This toxicity underscores the challenge of translating potentially membrane-acting macrolides into therapies and the need for further optimization.

Conclusion

By integrating LC-MS/MS-based molecular networking with phenotypic antitrypanosomal screening, we prioritized nine samples from 2,822 marine cyanobacterial and algal extracts and fractions. This led to the discovery and complete structural elucidation of canalesolide A (1), a structurally distinct 38-membered polyhydroxy macrolide from Okeania sp. Its potent low-nanomolar activity against T. brucei and rapid parasiticidal action highlight the therapeutic potential of this underexplored class of marine natural products. Detailed spectroscopic analysis, complemented by empirical configurational methods, model compound synthesis, and targeted chemical degradation and derivatization, allowed for the unambiguous assignment of this highly oxygenated and stereochemical complex macrocycle. Our findings not only underscore the persistent challenges associated with stereochemical resolution in polyol-rich natural products but also establish a practical framework for resoling class-wide configurational ambiguities through the combined use of targeted NMR methodologies and minimal synthetic efforts. Genome mining of the producing organism revealed a modular type I PKS BGC consistent with the structural features of canalesolide A (1), including a loading module capable of installing a tert-butyl group, elongation β- and δ-modules for polyol motif formation and a β-branching cassette for β-methylation. Notably, discrepancies between module architecture and the observed stereochemical pattern support a noncanonical biosynthetic model involving noncollinear module usage, iterative bimodular extension and selective module skipping. These observations underscore the mechanistic flexibility of modular PKS systems and provide a framework for future biochemical validation of the proposed pathway. Ongoing studies focus on elucidating the molecular mechanism of action of canalesolide A (1), including potential membrane-targeting behavior, and defining structure–activity relationships through analogs and semisynthetic derivatives. More broadly, canalesolide A (1) illustrates how the stereochemical degeneracy of densely oxygenated macrolides can be overcome by integrating nondestructive high-resolution NMR analysis with targeted chemical manipulation and synthetic model validation. This combined approach provides a transferable framework for assigning structurally repetitive polyol natural products that remain inaccessible to conventional spectroscopic analysis alone. Ultimately, this study expands the chemical, biosynthetic and functional landscape of polyhydroxylated macrolides and underscores marine cyanobacteria as a rich source of chemically complex antiparasitic scaffolds.

The raw NMR files for canalesolide A (1) have been deposited in the Natural Products Magnetic Resonance Database (NPMRD, www.np-mrd.org) under NP-MRD ID NP0354063. The genome assembly of cyanobacterium Okeania sp. has been submitted to NCBI and is available under Bioproject ID PRJNA1467307.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.6c11673.

• Experimental details, materials, methods, and additional results and spectroscopic data for all compounds (PDF)

¶.

Institut de Chimie de Nice, Université Côte d’Azur CNRS, 06000 Nice, France

◆.

T.R.T and B.R. contributed equally to this work.

The authors declare the following competing financial interest(s): R.K. is a scientific advisory board member, and consultant for BiomeSense, Inc., has equity and receives income. He is a scientific advisory board member and has equity in GenCirq. He has equity in and acts as a consultant for Cybele. M.O.D. has equity in GenCirq. The terms of these arrangements have been reviewed and approved by the University of California, San Diego in accordance with its conflict of interest policies. All other authors declare no conflicts of interest.