Accelerating Natural Products Research with Ometa Flow
Natural products are molecules created by living organisms. The first human medicines were natural products: willow bark’s salicylic acid gave rise to aspirin, cinchona bark provided the antimalarial quinine, and Penicillium yielded penicillin. While natural products encompass a wide spectrum of compounds, the field of natural products focuses on discovering novel therapeutics from nature’s vast chemical repertoire. This approach has been remarkably fruitful: over 65% of all FDA-approved small molecule drugs are either natural products or derived from them.1
A typical natural products workflow involves several stages. First, crude extracts from an organism are screened for bioactivity. For example, an extract from a marine sponge that kills tumor cells but spares healthy cells would become a candidate for further investigation. Next comes compound identification, where the active component is measured and isolated. Finally, structural elucidation unravels the novel compound’s chemical identity.
Despite its historical achievements and increased rates of success in clinical trials, natural products research occupies a relatively small niche in pharmaceutical R&D.2 Two recurring challenges limit its broader adoption:
Compound identification: Typically, active compounds are identified through a process called library matching, where experimental data is compared to a database of known compounds. While effective in some cases, this method will never identify compounds that don’t already exist in the database. Unfortunately, most compounds aren’t in our best-in-class databases. The typical library search will only identify 5-25% of compounds in an extract, leaving the rest unknown. De novo structure elucidation for those unknowns can take months or years, greatly hindering the drug development pipeline and increasing the cost of the discovery process.
Compound production: Even if a novel bioactive compound can be identified, it can still be very difficult to produce. Natural products can originate from rare or slow‑growing species, and purification protocols may yield only a few milligrams after months of work. A compound with potent bioactivity isn’t viable if it can’t be produced in sufficient quantities.
The following application note demonstrates how Ometa Flow can overcome both challenges by rapidly prioritizing novel bioactive compounds and identifying their distribution across the tree of life. To do so, we analyze MS/MS data of seaweed samples from the genera Dictyota and Laurencia as well as blanks (n=4 each). All Dictyota samples demonstrated bioactivity against three different cancer cell lines, inhibiting the growth of A549 (lung cancer), HT29 (colorectal adenocarcinoma), and PC3 (prostatic adenocarcinoma) cells. Laurnecia samples had no effect on cancer cell proliferation. The raw MS/MS and viability data is publicly available in MassIVE under accession MSV000089070.*
Step One: Molecular Networking
The first step in identifying bioactive compounds is Molecular Networking. This workflow takes MS/MS data and performs clustering, library searching, peak integration, and networking, completing in 11.5 minutes for the samples above.** Results can be accessed here. The workflow identified 609 metabolites in the seaweed samples, 182 (29%) of which were identified using library matching. All library matches are available here. Some of the identified compounds have already been studied for their anticancer effects, such as arachidonoyl ethanolamide phosphate (AEA-P). However, over 70% of the metabolites in these samples remained unidentified. These novel metabolites may include compounds with potent bioactivity, yet without structural information, these data are typically discarded.
The first step to better understand these unknowns is to identify unannotated compounds that are structurally related to known bioactive molecules. Molecular networking rapidly highlights these relationships by graphing metabolites based on their structural similarity. For instance, the subnetwork below contains metabolites with library matches to AEA-P and other fatty acyls. It also contains 10 unknown compounds. The links, or edges, between these known and unknown compounds indicates that they share structural similarities.
Sub-network containing AEA-P and unknown compounds. Each circle (or node) corresponds to an individual metabolite, while edges between nodes indicate structural similarities between those metabolites. This visualization can be accessed here.
Step Two: Bioactivity Networking
A compound may share structural similarity with a bioactive molecule yet lack bioactivity itself. To pinpoint which metabolites are driving bioactivity in a sample, Ometa Flow offers the Bioactivity Networking workflow. This tool correlates sample bioactivity data with metabolite profiles to identify which metabolites are most strongly linked to bioactivity. Multiple bioactivities can be analyzed at once; in this case, it analyzed bioactivities against all three cancer cell lines and completed in 13 seconds.** Results can be accessed here.
These bioactivity results can be mapped onto the molecular network from Step One. For all three cancer cell lines, the unknown compounds in the same subnetwork as AEA-P showed the strongest correlation to bioactivity.
Bioactivity network showing sub-network containing AEA-P and unknown compounds. Nodes are colored by the significance of the correlation between bioactivity and metabolite abundance in A549 cells (measured by log p value). The larger and redder the nodes, the more they are predicted to contribute to bioactivity. This visualization can be accessed here.
Step Three: MASST
While networking and bioactivity workflows can rapidly identify lead compounds, downstream development requires that these compounds be produced at scale. While Dictyota can be engineered to produce large quantities of a compound; not all labs are equipped for seaweed cultivation or genetic manipulation. In addition, lead compounds are often discovered in species or strains that are much more difficult to culture. Identifying new source organism(s) that produce a compound of interest can greatly accelerate production and testing of new compounds.
MASST can rapidly identify every species, organism, and strain in which a compound has been detected across hundreds of thousands of samples, even when that molecule is unknown.3 These searches can identify optimal production strains in seconds, greatly accelerating the purification process. In addition, MASST searches can also look for analogs of the compound of interest (eg, methylation or hydroxylation), which can highlight species that produce similar compounds or identify a potential candidate for semi-synthetic production.
MASST searches can be started directly from a molecular network by selecting a node and clicking the “MASST Spectrum” button. This will pull up Ometa’s MASST server where the user can choose which search parameters to use. In this case, a MASST search of node 5857, an unknown compound structurally related to AEA-P that was highly correlated to cytotoxicity against three cancer cell lines, returned similar compounds in several plant species.*** These included species from the genera Eremophila and Euphorbia, which are both used as traditional medicines and have been previously studied for their anti-cancer effects.4,5
plantMASST results from search of node 5857 in the molecular network above.6
Finally, MASST results above searched millions of public MS/MS spectra from domain-specific MASSTs. While these samples cover thousands of species and strains across plants and microbes, the results may not perfectly correlate to the strain libraries you have on hand. In this case, it is possible to build custom MASST databases with public or private data to further tailor results. If you’re interested in creating a custom MASST database, please contact us.
Summary
Here, we demonstrate how Bioactivity Molecular Networking and MASST can rapidly identify bioactive lead compounds and optimize their downstream purification. The workflows highlighted above took MS/MS data with no pre-processing and performed feature extraction, compound identification, peak integration, molecular networking, bioactivity networking, and MASST searches in less than 15 minutes.** This workflow can reduce lead compound generation from months to moments, allowing you to screen a larger number of compounds and produce more drug candidates.
Limitations
There are many methods for linking metabolites with bioactivity. Our workflow uses the method published here, which may not be optimal in all cases.7 Bioactivity is complex, and the myriad ways to predict it can be as well. We’re always considering which new methods to leverage to expand our bioactivity workflows. If you have questions or suggestions, please contact us.
Have more questions? Contact us.
*Cytotoxicity data was transformed to a hundred-point scale for the purposes of demonstration. The original bioactivity data is publicly available in MassIVE.
**Analyses were run on a local server with 20 Cores and 480GB RAM. Runtimes can be significantly improved by increasing computational resources or migrating to high performance computing (HPC).
***MASST search performed with minimum cosine set to 6, minimum matched peaks set to 5, mass tolerance set to 0.05 Da, and search of common analogs turned on.
References
1. Newman, D. J. & Cragg, G. M. Natural Products as Sources of New Drugs over the Nearly Four Decades from 01/1981 to 09/2019. J. Nat. Prod. 83, 770–803 (2020).
2. Domingo-Fernández, D. et al. Natural Products Have Increased Rates of Clinical Trial Success throughout the Drug Development Process. J. Nat. Prod. 87, 1844–1851 (2024).
3. Wang, M. et al. Mass spectrometry searches using MASST. Nat. Biotechnol. 38, 23–26 (2020).
4. Jiménez-González, V. et al. Nature’s Green Potential: Anticancer Properties of Plants of the Euphorbiaceae Family. Cancers 16, 114 (2023).
5. Savigni, D. Evaluation of the Anticancer Potential of Plants Used as Traditional Medicines by Aboriginal People. (The University of Western Australia, 2010).
6. Gomes, P. W. P. et al. plantMASST - Community-driven chemotaxonomic digitization of plants. bioRxiv 2024.05.13.593988 (2024) doi:10.1101/2024.05.13.593988.
7. Nothias, L.-F. et al. Bioactivity-Based Molecular Networking for the Discovery of Drug Leads in Natural Product Bioassay-Guided Fractionation. J. Nat. Prod. 81, 758–767 (2018).
