Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 10 (1), 1319

Advanced Identification of Global Bioactivity Hotspots via Screening of the Metabolic Fingerprint of Entire Ecosystems

Affiliations

Advanced Identification of Global Bioactivity Hotspots via Screening of the Metabolic Fingerprint of Entire Ecosystems

Constanze Mueller et al. Sci Rep.

Abstract

Natural products (NP) are a valuable drug resource. However, NP-inspired drug leads are declining, among other reasons due to high re-discovery rates. We developed a conceptual framework using the metabolic fingerprint of entire ecosystems (MeE) to facilitate the discovery of global bioactivity hotspots. We assessed the MeE of 305 sites of diverse aquatic ecosystems, worldwide. All samples were tested for antiviral effects against the human immunodeficiency virus (HIV), followed by a comprehensive screening for cell-modulatory activity by High-Content Screening (HCS). We discovered a very strong HIV-1 inhibition mainly in samples taken from fjords with a strong terrestrial input. Multivariate data integration demonstrated an association of a set of polyphenols with specific biological alterations (endoplasmic reticulum, lysosomes, and NFkB) caused by these samples. Moreover, we found strong HIV-1 inhibition in one unrelated oceanic sample closely matching to HIV-1-inhibitory drugs on a cytological and a chemical level. Taken together, we demonstrate that even without physical purification, a sophisticated strategy of differential filtering, correlation analysis, and multivariate statistics can be employed to guide chemical analysis, to improve de-replication, and to identify ecosystems with promising characteristics as sources for NP discovery.

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
Sampling sites and their chemical characterization. An overview of the geographic origin of the analyzed MeE is indicated in the world map (map from Wikipedia, reuse permitted under the Creative Commons Attribution-ShareAlike 3.0 Unported license (CC-BY-SA 3.0, https://creativecommons.org/licenses/by-sa/3.0/), created by Strebe, https://en.wikipedia.org/wiki/World_map#/media/File:Winkel_triple_projection_SW.jpg, modified) (a). All samples were screened for their chemical composition via FT-ICR-MS analysis, which resulted in chemical fingerprints consisting out of several thousand m/z features per sample and their relative intensities. (b) The combined van Krevelen diagram of all MeE depicts a broad distribution across the chemical space of all elemental compositions (CHO (blue), CHOS (green), CHNO (orange) and CHNOS (red). Bubble sizes represent the summed detected intensities). (c) PCA score plot based on chemical fingerprints, in which samples are colored according to the geographic location of sampling gives an overview of similarities and differences within the samples’ organic content. (d) The chemical space at different sampling sites varied thereby profoundly, exemplarily shown for 31_GL_FB05 (taken in Greenland) and 27_FJO_PILong River (taken in New Zealand). These two samples were chosen as they pose a high antiviral activity, which is later described in detail. The ring charts give the quantitative distribution of elemental compositions (color coding according to b).
Figure 2
Figure 2
Summary of anti-HIV activity of MeE and related m/z features. All samples were tested for their HIV-1-inhibitory potency. A general overview of obtained results is given in the pie chart (a) (full data can be found in Table S1). We categorized the obtained viral inhibition in 4 groups: less than 10%, 10–20%, 20–60% of infection compared to non-treated infected cells. The number of samples belonging to each group is illustrated. Nine out of eleven samples with very high anti-HIV activity (<10% of infection) originate from fjords in New Zealand (map from Wikipedia, reuse permitted under the Creative Commons Attribution-ShareAlike 3.0 Unported license (CC-BY-SA 3.0, https://creativecommons.org/licenses/by-sa/3.0/), https://en.wikipedia.org/wiki/File:New_Zealand_relief_map.jpg, modified). (b) We observed a relation of sampling sites within the fjords and HIV inhibition: for all fjords only surface waters and waters with strong terrestrial input show antiviral activity (HIV infection is given in % in brackets; antiviral samples are highlighted red). In contrast, in waters taken at the same location but from deeper layers the antiviral activity was absent. (c) A correlation analysis of the detected intensity of m/z features and detected HIV inhibition was performed to extract putatively associated m/z features out of the dataset for each sampling site. For all fjords, a set of CHO containing molecules delivered correlation coefficients >0.8. All of these cover the region of polyphenols in the van Krevelen diagram. (d) Following the correlation analysis between presence of m/z features and observed HIV inhibition over different depths at one sampling site, we surveyed the presence of these anti-HIV-associated m/z features along the fjords. This is exemplarily shown for one representative m/z 390.994305. The detected intensity of this m/z in complex extracts sampled along the fjord is illustrated, as well as the detected HIV inhibition of the total extract. The concentration of the molecule is higher deep inside the fjords, while it is absent at the mouth of the fjords and in related off-shore samples.
Figure 3
Figure 3
Obtained HCS results and integration with HIV inhibition and chemical fingerprints. Combined data analysis of the obtained HCS cytological profiles, anti-HIV-1 activity, and chemical fingerprints strongly suggests that the two main groups of antiviral MeE discovered in this study, cause virus inhibition through different MoAs and different sets of molecules. (a) Cluster analysis of all cytological profiles of antiviral samples (>50% inhibition) delivered two main clusters. Importantly, the open water sample 31_GL_FB05 shows a very different cytological profile compared to the other antiviral samples. (Colors indicate positive (yellow) or negative (blue) deviation from the mean of untreated control cells for each cellular feature (control = 1). Spearman rank correlation was used as a distance metric). (b) PCA score plot based on chemical fingerprints, in which samples are colored according to their HIV inhibition, show a close clustering of most antiviral samples in the first quadrant. These samples contain a strong terrestrial input and a related chemical fingerprint. Only one antiviral sample, 31_GL_FB05, does not belong in this cluster, which indicates that this sample contains a different set of molecules. (c) The PCA loading plot of combined data (HCS cytological profiles, HIV-1 inhibition, UHR mass spectrometric fingerprints) shows an overview of cellular core features, which are correlated/anti-correlated with the HIV inhibition and with a subset of m/z features. For easier readability the HCS parameters were limited to a set of core markers, which contains most of the information.
Figure 4
Figure 4
Detection of NRTI-like signatures in a marine antiviral MeE. The marine sample 31_GL_FB05 shows a close clustering to HIV reference compounds (NRTI). M/z features, which are uniquely present in this sample, cover the same chemical space as these NRTIs and show related elemental composition. (a) Secondary analysis integrating reference compounds and cytological profiles obtained from MeE delivers a close clustering of reference nucleosidic inhibitors of the HIV-1 RT with two MeE samples: the anti-HIV active sample 31_GL_FB05 und 56_CAL_ref coast. These two samples differ particularly in the membrane marker region, with the HIV-1 active sample showing strong positive deviation. Colors indicate positive (yellow) or negative (blue) deviation from the mean of untreated control cells for each cellular feature (control = 1). Spearman rank correlation was used as a distance metric. (b) M/z features were filtered for their presence in 31_GL_FB05 and absence in 56_CAL_RefCoast (which clusters with sample 31_GL_FB05 by cytological profiles but does not show HIV-1 inhibition). Filtered m/z features cover the same chemical space as NRTIs (CHNO (orange), CHNOS (red), bubble sizes indicate detected intensity, NRTIs - Nucleoside analog reverse-transcriptase inhibitors, NtRTIs - Nucleotide analog reverse-transcriptase inhibitors, nNRTIs - Non-nucleoside reverse-transcriptase inhibitors). (c) Network analysis of these m/z features delivered three molecular families given by their elemental compositions. All of the correlated molecules contain nitrogen.

Similar articles

See all similar articles

References

    1. Gu J, et al. Use of Natural Products as Chemical Library for Drug Discovery and Network Pharmacology. PLOS ONE. 2013;8:e62839. doi: 10.1371/journal.pone.0062839. - DOI - PMC - PubMed
    1. Newman DJ, Cragg GM, Snader KM. Natural products as sources of new drugs over the period 1981–2002. J Nat Prod. 2003;66:1022–1037. doi: 10.1021/np030096l. - DOI - PubMed
    1. Lam KS. New aspects of natural products in drug discovery. Trends Microbiol. 2007;15:279–289. doi: 10.1016/j.tim.2007.04.001. - DOI - PubMed
    1. Newman DJ, Cragg GM. Natural Products As Sources of New Drugs over the 30 Years from 1981 to 2010. Journal of Natural Products. 2012;75:311–335. doi: 10.1021/np200906s. - DOI - PMC - PubMed
    1. Pye CR, Bertin MJ, Lokey RS, Gerwick WH, Linington RG. Retrospective analysis of natural products provides insights for future discovery trends. Proc Natl Acad Sci USA. 2017;114:5601–5606. doi: 10.1073/pnas.1614680114. - DOI - PMC - PubMed
Feedback