**2. Results and Discussion**

To allow for the comparison of LC-MS traces of extracts from different collections of *M. bouillonii*, a new pipeline was created called the Objective Relational Comparative Analysis (ORCA) pipeline

(https://github.com/c-leber/ORCA) (Figure 1). ORCA is a flexible, modular pipeline that includes capabilities for simple and customizable MS1 feature processing. ORCA can also accept any bucket table of samples vs. features as input, allowing for the comparison of data from any source that can be tabulated in such a manner. To accommodate heterogeneous data and to allow for the comparison of diverse datasets, ORCA MS<sup>1</sup> feature processing starts with an input directory of mzXML files, from which the MS1 features are picked and integrated based on the mass-to-charge ratio (*m*/*z*) and a user-selected variant of retention time (rt). Feature picking is parameterized with the user-defined *m*/*z* and rt tolerances, and the peak size and shape parameters. Subsequently, MS<sup>1</sup> features picked from each sample file are consolidated based on the user-defined *m*/*z* and rt tolerance parameters and are organized into a samples vs. features bucket table containing feature integration values, with options to apply transformations based on the goals of the downstream analyses.

**Figure 1.** Illustration of the Objective Relational Comparative Analysis (ORCA) pipeline. The pipeline accepts inputs of either LC-MS datafiles in mzXML format, which can then undergo MS1 feature processing, or an externally created samples vs. features bucket table coming from any data source. Analyses currently offered as a part of the ORCA pipeline include hierarchical clustering, feature selection, and feature dereplication based on user-provided reference data.

After processing of the sample MS<sup>1</sup> features, or the input of an externally generated sample vs. feature bucket table, the vectors of the feature values can then be utilized to initiate a diverse array of analyses, including hierarchical clustering of the samples to gain insights into the relationships between the samples, and univariate feature selection to learn about what specific MS<sup>1</sup> features

are driving the differences between groups of samples. These analyses can then be visualized as dendrograms or heat maps, respectively. ORCA can also be used to generate a list of the most prominent MS<sup>1</sup> features across samples and to assign putative identifications from a user-supplied spreadsheet, allowing one to efficiently detect expected peaks across many samples, and to quickly determine the mass spectral signature of new potential isolation targets. ORCA was designed for assessing the relationships between heterogeneous samples and generating hypotheses regarding which features are driving these relationships; this makes ORCA a useful framework for not only learning about chemogeographical patterns, but also for comparing chemical profiles across different growth conditions [61], detecting contamination of botanical extracts [62], identifying chemotaxonomic patterns [63], and many other potential uses.

Crude extracts of field-collected samples of *M. bouillonii* from American Samoa, Guam, Kavaratti (Lakshadweep Islands, India), Saipan, and the Paracel Islands (Xisha) in the South China Sea, as well as an in-house culture from Papua New Guinea, were profiled via LC-MS/MS; the resultant chromatograms were used as inputs for MS<sup>1</sup> feature processing in ORCA. Hierarchical clustering was performed on the MS<sup>1</sup> features and a dendrogram was produced with a cophenetic correlation coefficient of 0.905, indicating that the displayed structure in the dendrogram is highly correlated to the cosine distances between samples, and thus is representative of the data (Figure 2). The structure in the dendrogram suggests clustering of samples according to geographical region, a phenomenon that has previously been observed across other cyanobacterial samples [59] but has not been specifically reported as a pattern for *M. bouillonii*. It is worth noting that, while samples with shared geographical origin are indeed arranged in clusters together in the dendrogram, the branch points for each geographical cluster are quite large, ranging from 0.4763 cosine distance for the two samples from Guam to 0.7248 cosine distance for where the three samples from Saipan converge. This is likely the result of a combination of the high variability and complexity in the composition of the studied samples, as well as the "curse of dimensionality" that artificially enlarges distance values when large numbers of features are being considered [64]. Classical molecular networking analysis using GNPS provided an orthogonal view, supporting the idea of chemogeographical specificity in these *M. bouillonii* samples, as numerous clusters of location-specific nodes are visible in the resultant network (Figure 3 and Figure S1). Furthermore, hierarchical clustering performed on presence-absence data of the MS2 nodes from GNPS, as visualized with a dendrogram (Figure 4), revealed a chemogeographical clustering similar to that produced from the ORCA MS1 features (Figure 2). The geographically associated structure in the data, as observable via both ORCA dendrograms, along with the presence of numerous location-specific clusters in the molecular network, led us to generate the hypothesis that the geographically specific distributions of natural products in our samples could be leveraged to identify previously unreported metabolites. The clustering of samples by specific geographical location stimulated further analyses to determine which molecular features were driving the observed geographic clusters, and which peaks were regionally specific. Particular attention was paid to Saipan, as it represented one region from which no new natural products from *M. bouillonii* had been reported in the scientific literature.

One cluster in the GNPS molecular network comprising only nodes originating from the Saipan-collected samples contained a particularly intense node for a feature with *m*/*z* 457.785 (Figure 3). Further investigations in ORCA revealed that this feature was present in high abundance in all samples from Saipan but was undetected or detectable at very low levels in the MS1 spectra of samples from all the other studied locales (Table S2). Additionally, this feature was not dereplicated when queried against all published compounds from *M. bouillonii* at the time (Table S3) and when searched against the MarinLit database (http://pubs.rsc.org/marinlit/). This intriguing chemogeographic pattern prompted prioritization of this feature for isolation and structure elucidation, ultimately resulting in the characterization of a region-specific metabolite. Based on the specific collection site from which the Saipan samples originated (Laulau Bay, Saipan), we originally termed this metabolite "laulauamide". A molecular feature with *m*/*z* 721.10 was found to have a very similar geographic distribution. It was

detected with high intensity in samples from Saipan, while being undetected or detectable at very low levels in other samples (Table S2), and thus was another strong driver of the clustering of the Saipan samples. Isolation and analytical characterization revealed that this MS feature was the sodiated adduct of the known compound lyngbyapeptin A [41] (Table S3).

**Figure 2.** ORCA-generated dendrogram (cophenetic correlation coefficient = 0.905) displaying the results of hierarchical clustering of the MS<sup>1</sup> features from *M. bouillonii* crude extracts. Samples are labeled with aliases comprising a general collection location concatenated to an abbreviated sample code. The structure in the dendrogram suggests that samples collected from the same geographical area are chemically more similar. Colorized for emphasis. Red: Papua New Guinea; Orange: Guam; Gold: American Samoa; Green: Saipan; Blue: Kavaratti (Lakshadweep Islands, India); Purple: Paracel Islands (Xisha) in the South China Sea.

*M. bouillonii* biomass (1 L sample, 132 g dry biomass yielding 10 g of crude extract) from Saipan's Laulau Bay (denoted as Saipan\_32 in Figures 2 and 4) was thoroughly extracted with 2:1 dichloromethane and methanol, and the resulting crude extract was fractionated over silica using vacuum liquid chromatography. LC-MS/MS analysis of the fractions revealed the MS1 feature of interest to be in highest abundance in two relatively polar fractions. Reverse phase HPLC was used to initially isolate 1.5 mg of this compound from the two fractions. 1D and 2D NMR experiments were utilized to establish the planar structure of **1**, with major contributions from the 1H-1H Correlated Spectroscopy (COSY), 1H-13C Heteronuclear Single Quantum Coherence (HSQC), 1H-13C Heteronuclear Multiple Bond Coherence (HMBC), HSQC-Total Correlation Spectroscopy (TOCSY), and long-range 1H-13C Heteronuclear Single Quantum Multiple Bond Coherence (HSQMBC) data.

**Figure 3.** Global Natural Products Social Molecular Networking (GNPS) classical molecular network of fifteen *M. bouillonii* crude extracts with the enlarged inset showing a cluster containing **1** (denoted with precursor mass *m*/*z* 457.785) and seven other nodes representing potential doscadenamide analogs (based on LR-MS/MS data). The green coloring of the nodes indicates that they represent features only detected in samples from Saipan. Nodes are scaled to summed precursor intensity. Grey nodes represent the MS2 features that are present in samples from more than one geographical region. Geographical location of samples is colorized as follows: Red: Papua New Guinea; Orange: Guam; Gold: American Samoa; Green: Saipan; Blue: Kavaratti (Lakshadweep Islands, India); Purple: Paracel Islands (Xisha) in the South China Sea.

**Figure 4.** ORCA-generated dendrogram (cophenetic correlation coefficient = 0.960) displaying the results of hierarchical clustering of the *M. bouillonii* crude extracts presence–absence data regarding GNPS nodes. Samples are labeled with aliases comprising a general collection location concatenated to an abbreviated sample code. Similar to Figure 2, the structure in the dendrogram suggests that samples collected from the same geographical area are chemically more similar. Colorized for emphasis. Red: Papua New Guinea; Orange: Guam; Gold: American Samoa; Green: Saipan; Blue: Kavaratti (Lakshadweep Islands, India); Purple: Paracel Islands (Xisha) in the South China Sea.

The 1H and 13C NMR chemical shifts for the two acetylene groups were highly similar, and by HMBC correlations both had an adjacent methylene group at the same shift (δ 2.18, H2-15 and H2-24). In one of these two cases, sequential correlations deduced from the 1H-1H COSY data, and supported by the results of a 1H-13C HSQC-TOCSY experiment, provided a spin system involving three additional shielded methylene groups at δ 1.44, 1.39, and 1.75 and 1.42 (H2-23, H2-22, and H2-21). The final of these methylene groups was positioned adjacent to a deshielded methine group at δ 3.75 (H2-20). By COSY, the methine was determined to be adjacent to a shielded methyl group at δ 1.12 (H3-27), and its chemical shift was explained by an HMBC correlation placing it adjacent to an ester or amide carbonyl (δ 176.4, C-19). The spin system of the second acetylene-terminating partial structure was highly similar and partially overlapped but terminated with a more shielded methine proton at δ 2.13 (H-11) with an adjacent methyl group (δ 1.11, H3-18) and amide or ester carbonyl (δ 177.0, C-10). Summarizing, two essentially identical 2-methyl-7-octynoic acid structural units were thus defined from highly similar but non-identical data subsets.

The remainder of the molecule was thus composed of C9H14N2O2 with 3 degrees of unsaturation resulting from an enone and one ring structure. Two 1H NMR singlets (δ 5.04, H-2 and 3.84, H3-9) along with a 9-proton connected spin system remained unassigned. The singlet at 3.84 ppm was assignable to a methoxy group at the β-position of the enone by virtue of its relatively deshielded chemical shift and HMBC correlations to the highly deshielded olefinic carbon at δ 179.2 (C-3). The other singlet was thus assigned to the α-position of this enone as it was attached to a shielded olefinic carbon at δ 94.2 (C-2) and showed an HMBC correlation to the carbonyl carbon at δ 170.0 (C-1). As this partial structure accounted for all oxygen atoms in compound **1**, the shielded nature of this carbonyl necessarily required it to be attached to a nitrogen atom, forming an amide. Based on 1H and 13C NMR chemical shift data (δ 4.64, H-4; δ 59.2, C-4), one terminus of the remaining spin system was assigned to a methine with an attached nitrogen atom. 1H-1H COSY data, in conjunction with the 1H-13C HSQC-TOCSY, allowed formulation of four sequential methylene groups. The final methylene was also relatively deshielded (δ 3.22 and 3.13, H2-8; δ 39.3, C-8), consistent with its attachment to a nitrogen atom. At this point, all atoms in the molecular formula of compound **1** were accounted for, except for one proton that was attached to a heteroatom by evaluation of the HSQC data (e.g., only 39 protons were found attached to carbon atoms); this was deduced to be an NH as three of the four oxygen atoms were assigned as carbonyls and one as a methylated enol.

HMBC correlations from the two diastereotopic protons at δ 3.13/3.22 ppm (H2-8) to the carbonyl at δ 177.0 (C-10) connected these two partial structures. The other 2-methyl-7-octynoic acid was therefore connected to the only remaining heteroatom, the N-atom connected to the δ 170.0 (C-1) carbonyl of the enone functionality. Remaining structural features at this point included the formation of one ring, and placement of a proton on one of the two nitrogen atoms; two possibilities emerged (**1a** and **1b**) (Figure 5).

**Figure 5.** Competing structural hypotheses for the two-dimensional structure of compound **1**.

Both structural possibilities had features that were attractive and unattractive from a predicted biosynthetic perspective. In **1a**, the fundamental assembly of the PKS derived octynoic acid; its passage to an NRPS to incorporate a lysine residue, followed by a ketide extension, *O*-methylation of the β-enol, and cyclization to a pyrrolidone ring, is well precedented within cyanobacterial natural products [65–67]. However, the acylation of a second octynoic acid residue to the lysine side chain nitrogen is an unprecedented event. Alternative structure **1b** has the attractiveness of a regular, predicted PKS(4)-NRPS(glycine)-PKS(3)-NRPS(glycine)-PKS architecture; however, it is quite awkward in requiring several unusual adjustments to the oxidation state of the carbon atoms, and creation of the second 2-methyl-7-octynoic acid residue via a completely different set of biosynthetic steps from the first one.

Modeling of these two alternative cyclization products for 13C NMR shifts (see Figures S2 and S3 for the predicted 13C NMR shifts for **1a** and **1b**, respectively) and comparison with those experimentally measured for **1** revealed that both possibilities were reasonably good fits, but the predicted values for 1a tended to be closer to the shifts experimentally derived for compound **1**. For both C-2, the methyl enol carbon (**1a** δ 95.5, **1b** δ 101.4, **1** δ 94.2), and C-3, the deshielded olefinic carbon (**1a** δ 180.7, **1b** δ 171.5, **1** δ 179.2), the fit for alternative **1a** was considerably better. Only at C-6 was the cyclization product proposed in **1b** favored (**1**a δ 24.3, **1b** δ 19.6, **1** δ 20.4). A deeper look into the long-range 1H-13C HSQMBC data was undertaken. The key proton distinguishing these two possible structures, H-4 at δ 4.64, showed correlations to several resonances, including two of the three carbonyl resonances (δ 170.0, C-1; δ 176.4, C-19) and the β-oxygenated enone (δ 179.2, C-3); these correlations were compatible with structure **1a** (one 2-bond and two 3-bond correlations), while in structure **1b** these correlations would result from one 2-bond, one 4-bond, and one 6-bond 1H-13C coupling. Furthermore, analysis of the HMBC correlations observed for H-4 and H2-8, both from the lysine-derived residue, showed mutual signals with only C-5 at δ 29.0 and C-6 at δ 20.4 that would be consistent with either proposed structure. There were no shared correlations observed between these protons and the equally 3-bond proximal carbonyl in **1b**, nor 3-bond correlations from H-4 to C-8 and H-8 to C-4 that would be reasonably expected to be observed from **1b**, lending further support for **1a** as being the correct structure of **1**.

Compound **1** contains three stereocenters—two associated with the two 2-methyl-7-octynoic acid side chains, and one contained in the central heterocycle. A racemic standard of 2-methyloctanoic acid was derivatized with (*S*)-(+)-2-phenylglycine methyl ester. A chiral standard of (*S*)-2-methyloctanoic was generated via the zirconium-catalyzed asymmetric carbo-alumination (ZACA) reaction [68] of 1-octene to stereoselectively install a methyl group at the C-2 position, followed by an oxidation to 2-methyloctanoic acid and derivatization with (*S*)-(+)-2-phenylglycine methyl ester. Configuration of both 2-methyloctynoic acid moieties of compound **1** was established to be *R* through catalytic hydrogenation, acid hydrolysis, derivatization with (*S*)-(+)-2-phenylglycine methyl ester, and comparison via LC-MS to the generated standards of 2-methyloctanoic acid coupled with the same chiral auxiliary group (Figure S4). Ozonolysis with an oxidative work-up [69], followed by acid hydrolysis, was used to open the heterocyclic ring structure and liberate lysine from compound **1**. The lysine was then derivatized with Marfey's reagent (l-FDAA) and compared to racemic and l-lysine standards derivatized with the same Marfey's reagent, indicating an *S* configuration of this residue (Figure S5). The fully elucidated structure of compound **1** was thus determined as in Figure 6.

**Figure 6.** Complete structure of compound **1**.

Low-resolution LC-MS/MS fragmentation data for **1** consistently showed three peaks at *m*/*z* 321, 303, and 168 (Figure S6), which we predicted to represent a side-chain loss, a side-chain loss plus the loss of an amine, and the loss of both side chains plus an amine, respectively. To better understand the fragmentations of compound **1** and use this information for identifying analogs based on repeating the MS2 fragmentation patterns, high-resolution MS<sup>2</sup> fragmentation data were acquired for compound **1**. Numerous fragment peaks were recorded, including peaks observed at

*m*/*z* 321.2171, 303.1901, and 168.1016. These values match very well to the calculated monoisotopic masses of the predicted fragment structures shown in Figure 7 (*m*/*z* 321.217, 303.183, 168.102; allowing for hydrogen rearrangements), lending support to our fragmentation hypothesis, and providing a starting point for understanding and proposing the structures of analogs via their fragmentation patterns.

**Figure 7.** Hypothesized fragment structures of compound **1**.

GNPS classical molecular networking placed compound **1** as a node in a cluster with seven other nodes originating from the Saipan *M. bouillonii* samples (Figure 3), suggesting several naturally occurring analogs were present. ORCA revealed that compound **1** is also present in samples from Guam, though detected with a much lower MS<sup>1</sup> intensity than in samples from Saipan. This inspired the generation of a more detailed molecular network composed of both crude extracts and fractions from a Saipan sample and a Guam sample (denoted as Saipan\_32 and Guam\_46 in the above dendrograms), revealing an even larger cluster of potential analogs that contained 33 nodes, including compound **1** (Figure S7). Some nodes in the cluster had very similar masses, which could be the result of an artifact from the particular parameter set selected for the analysis, an artifact of the low resolution MS data analyzed, or be an indicator of isomeric analogs; therefore, further analysis was needed.

Analysis using the GNPS in browser network visualizer suggested that there was a common connection between many of the potential analogs (23 out of 33, including **1**), namely the presence of an MS<sup>2</sup> fragment peak at *m*/*z* 168 (Figure S7). To facilitate further analysis of MS<sup>2</sup> spectra and the presence of potential analogs, the ORCA MS2 auxiliary pipeline was developed. MS<sup>2</sup> scans from the Saipan and Guam crude extract and fractions were binned based on precursor mass, and then filtered to only precursor masses with scans that included a *m*/*z* 168 fragment peak. Clustering scans from each relevant precursor mass by cosine distance, paired with manual analysis, allowed the structures of 9 analogs (**2**–**10**) to be proposed (Figure 8; see Figures S8–S26 for the proposed structures, consensus spectra, and predicted fragment structures). It must be noted that alternative structural proposals are conceivable for these analogs; however, given the literature precedent for cyanobacteria to produce families of natural products with the same array of variations in desaturation and oxidation as proposed here, e.g., [66,70,71], and the predictable MS2 fragmentation spectra observed, these proposals represent the most parsimonious and best supported structural hypotheses. Ambiguities in the remaining related MS<sup>2</sup> spectra prevent the definitive assignment of carbon chain isomers and positional isomers, and the proposal of additional analogs, but suggest a process of combinatorial biosynthesis in generating this expansive natural product family. While quantities of these minor metabolites in our samples were not sufficient for isolation and further characterization, the total synthesis published alongside the characterization of **1** [29] is very amenable to incorporating alternative side chains, and this could be used for generating these proposed analogs for further study.

**Figure 8.** The doscadenamides: compound **1**, along with its analogs whose proposed structures were annotated via informative patterns in the MS2 fragmentation data (see Figures S9–S26). Each analog consists of a heterocyclic core with two fatty acid side chains with the following possibilities: MOYA = 2-methyl octynoic acid; MOEA = 2-methyl octenoic acid; MOAA = 2-methyl octanoic acid; oxo-MOAA = 2-methyl 7-oxo octanoic acid.

Compound **1** contains unusual structural features that, while having precedent in other cyanobacterial natural products, have not previously been seen together. Terminal alkynes can be found in several other natural products from *Moorena* spp., including jamaicamide B [66], carmabin A [71], and vatiamides A, C, and E [72], but having two is notable. While ribosomally synthesized and post-translationally modified peptides (RiPPs) and NRPS-derived natural products with amino acid subunits are common in cyanobacteria, lysine is not often seen, especially in the natural products of marine cyanobacteria [73,74]. The heterocycle in **1**, composed of an acetate extended amino acid, has been observed in the malyngamides [65], jamaicamides [66], gallinamides [67], and other cyanobacterial natural products, but again, never has it been reported involving a lysine residue. Two curiosities of the biosynthesis of compound **1**, namely the origin of the two 2-methyl octynoic acid residues and the formation of the heterocycle, can be explained by analogy to what is known about the biosynthesis of the jamaicamides [66]. To generate 2-methyl octynoic acid, a fatty-acid desaturase analogous to JamB could act upon an octanoic acid precursor, or a smaller precursor that has been PKS-extended to the appropriate size. The placement of the methyl group in the 2 position suggests incorporation via

S-adenosyl methionine (SAM). Formation of the heterocycle likely occurs as the result of an acetate extension of the carboxyl group of lysine, followed by a Claisen-like condensation and cyclization directed by a cyclase analogous to JamQ. As noted above in the discussion of structural possibilities **1**a and **1**b, what is less clear is how 2-methyl octynoic acid is appended to the terminus of the lysine side chain; the peptide bond formed is far from unusual, but its placement suggests enzymatic activity occurring beyond the otherwise linear PKS-NRPS assembly of the molecule.

To further evaluate the relationships of the structural features found together in compound **1** to the known natural product chemical space, we applied a Small Molecule Accurate Recognition Technology (SMART) [75] analysis to search for structurally similar molecules based on HSQC spectra. SMART did not yield any similar compounds with a cosine value higher than 0.84, further revealing the structural uniqueness of compound **1**. We also utilized the structure similarity search function in SciFinder (https://scifinder.cas.org/), which yielded only the sintokamides (Figure S27). The sintokamides share a similar heterocycle and are halogenated natural products from sponges [76]. While not suggested by either structure similarity query, tetramic acids [77] and prostaglandins (PGE2, for example) [78] (Figure S27) are two chemical classes that possess some distant level of structural similarity to compound **1**, and this inspired additional bioactivity testing efforts, as described below.

Structural similarity to tetramic acids inspired in silico antibiotic screening (http://chemprop.csail. mit.edu/) [79]. The known antibacterials C12-tetramic acid and C14-tetramic acid scored over five times greater than the highest scoring doscadenamide (Table S4), providing little incentive to further evaluate the doscadenamides for antibiotic activity.

Compound **1** was assayed for cytotoxicity against human NCI-H460 cells and yielded an IC50 > 22 μM, suggesting negligible cytotoxicity. This lack of cytotoxicity, plus some distant structural similarity to prostaglandins, inspired the screening of compound **1** in a Griess assay for anti-inflammation (as well as cytotoxicity) toward murine macrophages RAW264.7 cells at a range of 7–55 μM. Curiously, rather than producing inflammatory or anti-inflammatory effects, compound **1** yielded dose-dependent synergistic cytotoxicity with lipopolysaccharide (LPS). This anomalous result was confirmed through multiple replicates of the assay (Figures S28–S30).

The doscadenamides were discovered based on global scale patterns in *M. bouillonii* chemical diversity. This illustrates that cyanobacteria harbor intraspecific chemogeographic patterns, and that these patterns can be utilized to direct discovery efforts towards new, regionally specific natural product families. There are many tools available for pursuing chemogeographic and other metabolite patterns in sample sets that can inform discovery efforts, each with their own strengths and limitations. While tools like the ORCA pipeline and GNPS classical molecular networking may be of limited utility in terms of quantitative analyses and effective separation of isomeric features, their flexibility in handling heterogeneous sample sets allows for comparative analyses between samples that could otherwise not be conducted. Furthermore, the intrinsic imperfection of real-world data and the deficiencies inherent to various tools and approaches encourages that an ensemble of tools and approaches be applied. By using ORCA in conjunction with GNPS, we were able to generate convergent results that increased confidence in our conclusions. Converging results from ORCA and GNPS were also helpful in giving confidence to the parameters selected for our analyses; parameter selection is often a challenge when applying computational techniques and requires deep knowledge of the dataset as well as manual validation. The chemogeographic patterns in *M. bouillonii* natural products that are qualitatively presented in this manuscript highlight the opportunity to further explore *M. bouillonii* natural products chemistry and how compounds and compound families are distributed ubiquitously vs. regionally, at different geographical scales. Studying *M. bouillonii* metabolomics in a more controlled, semi-quantitative fashion would allow these patterns to be evaluated more deeply and will be the focus of a future manuscript.

Doscadenamide A (**1**), when considered in isolation, is a structurally intriguing compound. Being composed of a heterocyclized, acetate-extended amino acid core appended with terminal alkyne containing side chains, it blends structural features common among cyanobacterial natural products

with a flair of the unusual: the inclusion of lysine, the dual terminal alkynes, and the acylation of the lysine side chain with one of those terminal alkyne containing side chains. In considering the doscadenamides as a family of cyanobacterial natural products, it is likely they are produced via a seemingly combinatorial addition of different acyl groups to a consistent core structure. From a biosynthetic perspective, this suggests a low level of fidelity in the assembly process. Connecting this family of compounds to the biosynthetic gene cluster responsible for their production would elevate our understanding of how cyanobacteria diversify their natural product arsenals. Since the aforementioned procedure for the total synthesis of compound **1** [29] is amenable to incorporating alternative sidechains, this could be used for generating the nine proposed natural structural analogs reported here (**2**–**10**), as well as for evaluating their activities as quorum sensing modulators [29] and their cytotoxic synergism with LPS.
