High-throughput targeted screening in triple-negative breast cancer cells identifies Wnt-inhibiting activities in Pacific brittle stars

Pro-proliferative oncogenic signaling is one of the hallmarks of cancer. Specific targeting of such signaling pathways is one of the main approaches to modern anti-cancer drug discovery, as opposed to more traditional search for general cytotoxic agents. Natural products, especially from marine sources, represent a largely untapped source of chemical diversity, which so far have mostly been screened for cytotoxicity. Here we present a pioneering pipeline of high-throughput screening of marine-based activities targeted against the Wnt signaling pathway, which is one of the key factors in oncogenic transformation, growth and metastasis in different cancers, including the devastating triple-negative breast cancer (TNBC) currently lacking any targeted therapies. This pipeline consisted of collection and characterization of numerous invertebrates during the SokhoBio expedition to the Kuril Basin in North Pacific, preparation of extracts from these specimen, and their screening in dedicated assays monitoring Wnt signaling in TNBC cells. This approach yielded a number of promising hits, including highly specific anti-Wnt activities targeting multiple levels within the Wnt pathway from Ophiura irrorata and other Pacific brittle stars.

pathways that are responsible for continuing proliferation of the cancer cells 6 have been one of the major targets for the anti-cancer drug discovery, giving rise to several targeted drugs, such as vismodegib -a selective inhibitor of the Smoothened receptor suppressing aberrant Hedgehog signaling in the treatment of basal cell carcinoma 7 .
Several oncogenic signaling pathways exist. In breast cancer (BC) as an example, three major subtypes of the disease are identified based on the type of the over-activated signaling pathway driving excessive cell proliferation 8 . The first BC subtype is estrogen-receptor positive, receives the command for excessive proliferation from over-activation of the estrogen signaling, and is susceptible to targeted anti-hormonal therapies such as tamoxifen 9 . The second subtype relies on over-production of the receptor tyrosine kinase (RTK) HER2 and resulting over-activation of the RTK pathway for excessive cell proliferation; the targeted treatment with anti-HER2 monoclonal antibodies, marketed under the trastuzumab (Herceptin) trade names, is typically applied against this BC subtype 10 .
The third BC subtype is negative in expression of estrogen and HER2 receptors, as well as progesterone receptor, as is therefore called triple-negative breast cancer (TNBC). Being insensitive to the existing targeted therapies, TNBC accounts for more than half of all BC-related deaths, taking away some 200000 lives annually on the global scale 11,12 . It is widely accepted that one of the most common sources of the pro-proliferation signal in TNBC cells is the over-activation of the Wnt signaling pathway [13][14][15] . The canonical Wnt signaling pathway plays critical roles in embryonic development 16 but is also important for the adult life of all metazoans including humans. The pathway controls cell proliferation and differentiation, and regulates proliferation and self-renewal of various stem cells as well as regenerative processes 17 . Aberrant activation or hyper-activation of the Wnt pathway can lead to carcinogenesis in several adult organs, most notably colon 18 and breast 19 . Regarding the TNBC, the Wnt/β-catenin signaling, along with other developmental signaling pathways including Cripto-1 and Notch/CSL, is of special importance for the regulation of cancer stem cells and for the therapy resistance 15 . Dysregulation of Wnt signaling in TNBC is not only one of the driving forces of initial transformation 20 but also promotes cell migration, colony formation, stem-like features and chemoresistance, playing important roles also in the later stages of tumor progression 14 . Detailed studies on gene expression profiles have subdivided the TNBC into six subtypes, all of which demonstrate aberrant overexpression of certain Wnt-associated genes 21 . Taken together, the modern knowledge suggests that in development of targeted treatments against TNBC, Wnt pathway inhibitors should play an essential though not an exclusive role 14 . The Wnt pathway is a complex signaling cascade. In humans, it operates with 19 types of Wnt proteins as ligands and 10 types of Frizzled proteins as receptors. The downstream components of the pathway involve a co-receptor LRP 5/6, G-proteins, a multiprotein β-catenin destruction complex (containing Axin, APC, glycogen synthase kinase 3β (GSK3β) and casein kinase), β-catenin itself (which serves as a cofactor of the key transcription factor TCF inducing expression of the Wnt target genes) and many more cofactors of the latter. Taken together, it gives a variety of targets for the inhibition of the pathway 22,23 .
In TNBC, overexpression of Wnt ligands, the receptor Frizzled7, and the co-receptor LRP6 has been described [24][25][26] . After that, development of agents targeting the Wnt signaling pathway as future anti-TNBC drugs (as well as drugs against other Wnt-dependent cancers) has been initiated. One approach involves development of human antibodies against upstream extracellular components of the pathway, such as an antibody against the cysteine-rich (extracellular) domain of Frizzled7, which later showed cross-reactivity with 4 other Frizzleds but efficiency against several solid cancers in vitro 27 ; phase I clinical studies are currently in place. Screening of libraries of synthetic small molecules is another approach to derive Wnt signaling inhibitors 28 . Several hit compounds have been identified, among them the molecule XAV939 inhibiting tankyrase -a multifunctional enzyme among other things regulating the activity of the β-catenin-destruction complex 29 . Rational drug design and in silico screening is also applicable to the Wnt pathway, especially after the recent description of the crystal structure of a Wnt-Frizzled complex 30 ; an example of such in silico analysis is our recent identification of an anti-leprosy drug clofazimine as a potent inhibitor of the Wnt pathway and growth of TNBC cells in vitro 31 .
Despite these efforts, no anti-Wnt signaling agents have so far made it to the clinic nor even advanced clinical trials 14 . As the demand for such drugs to treat TNBC and other Wnt-dependent cancers is urgent, other approaches to develop Wnt signaling inhibitors should also be considered. We here describe a paradigm of screening for anti-Wnt, anti-TNBC activities from marine sources. Collection of marine invertebrate species performed during the SokhoBio expedition 32 and extract preparation is hereby followed by high-throughput screening for inhibitors of the Wnt pathway in TNBC cells. This approach has resulted in identification of promising activities from brittle stars of North Pacific and broad characterization of their mode of action within the Wnt pathway. Other interesting Wnt inhibitory and activatory activities are also identified in other species. Upscaling and further development of this paradigm may promote the discovery of lead compounds to develop treatments against the Wnt-dependent cancers.

Results
During the SokhoBio expedition 32 of the research vessel "Akademik M.A. Lavrentiev" from 5 th of June to 6 th of August 2015, invertebrate organisms were collected in the Kuril Basin of the Sea of Okhotsk, northwestern Pacific Ocean, at the depths ranging from 1700 to 4750 m ( Fig. 1 and Supplementary Table S1). Specimen from the following invertebrate groups were collected ( Fig. 2 and Supplementary Table S1) Ethanol extracts (see Methods) from 81 specimens ( Fig. 2 and Supplementary Table S1) were screened at different dilutions, using the TNBC cell line BT-20 stably transfected with the TopFlash reporter construct, which expresses a firefly luciferase under the Wnt-dependent TCF-promoter, so that the luminescence activity of the cells reflects the intensity of Wnt signaling 31 , and stimulated by exogenous Wnt3a (provided as Wnt3a-conditioned medium, see Methods). The Wnt3a-conditioned medium is known to be a good substitute for purified Wnt3a for cell culturing purposes and possesses a capability to activate the cascade in the luciferase-based and other cell culture assays and even to support growth of tissue-based organoids in vitro 33 . In addition to the TopFlash firefly luciferase, which measures Wnt3a-induced transcription, the cells were transiently transfected with the CMV-Renilla luciferase reporter a day prior to Wnt stimulation. The Renilla luciferase activity reflects the level of general, Wnt-independent transcription (and thus the well-being of the cells). This dual-luciferase assay allowed us to normalize the values of Wnt-dependent luminescence by the Wnt independent luminescence in the same sample and thus to distinguish between a specific activity of the extracts on the Wnt pathway as opposed to the general effect on cellular transcription and/or general cytotoxicity. The assay used in the study is applicable to track a Wnt-modulating activity working at any level within the cascade, from the upstream ligand-receptor interaction to the downstream transcription factor activity in the nucleus. This allows identifying a broad spectrum of potential inhibitors and modulators during the screening, while their exact molecular target can be deciphered later using other approaches. The specific activity of the cascade in cells treated only with the Wnt3a-conditioned medium (ca. 20-fold over the background), was set to 100%.
This analysis identified 15 extracts from 11 marine invertebrates possessing specific (not affecting general transcription), strong (>2-fold reduction) and significant (p-value by the linear regression test <0.05) Wnt pathway-inhibitory activity (Supplementary Table S2).
Some of these extracts were produced from single specimen of a single species within a larger animal group, such as a decapod crustacean Calocarides quinqueseriatus (reduction of Wnt signaling activity to ca. 10% of control at 1:400 dilution), a large unidentified polychaete (ca. 50% at 1:300), or a crab Munidopsis antonii (ca. 20% at 1:300) (Supplementary Table S2). In some cases, we were able to delineate a specific body part of the invertebrate, which contained the anti-Wnt activity, such as hepatopancreas, but not muscles, of a decapod crustacean Calocarides quinqueseriatus (ca. 10% at 1:400), or skin and muscular sac, but not gonads, of a holothurian Molpadia musculus (see further below, Supplementary Table S2).
Much more confidence is provided when different representatives from the same species or of a bigger group, collected at different sites, could be compared. Five invertebrate groups stood out in this regard. Holothurians Peniagone sp. and Molpadia musculus (skin and muscular sac of the latter), collected at different sites, inhibit the Wnt pathway in a concentration-dependent manner, down to 30% of control at 1:300 for Peniagone and to 40% at 1:100 for Molpadia ( Fig. 3a and Supplementary Table S2). Polychaete Travisia sp., various actinias, and gonads of Molpadia musculus, in contrast, displayed the ability to co-stimulate the Wnt pathway: activation of the Wnt pathway, induced by addition of Wnt3a, became potentiated by 1.5-2 folds if the extract was added together with the Wnt3a conditioned medium (Fig. 3b and Supplementary Table S2). Given the importance of the Wnt pathway activation in stem cell self-renewal and tissue regeneration 17 , these unusual co-stimulatory activities from marine invertebrates can also be of practical interest 22 and will require further elaboration of the exact mechanism of action.
The strongest and the most robust effects on the Wnt pathway were seen with the brittle stars (Ophiura). Of the five specimens collected at three different stations (station 8 at the depth of 2250 m, station 9 at the depth of 3340 m, and station 11 at the depth of 3140 m, Supplementary Table S1 and Fig. 1) and belonging to 3 different species (Fig. 2) including Ophiura irrorata, all showed a near-complete (down to 2-5% of the control at 1:200 dilution) and concentration-dependent inhibition of the Wnt pathway, without noticeable effect on the general cell transcription (Supplementary Table S2 and Fig. 4a,b).
Focusing on Ophiura irrorata, we next roughly assessed where in the Wnt signaling pathway the molecular target of the active component(s) of this brittle star could be located. To address this, we compared the ability of the extract to inhibit the Wnt pathway when activated by Wnt3a vs. LiCl at 15 mM (a concentration within the standard range for induction of the Wnt signaling in cultured cancer cells 31,34,35 ). LiCl inhibits GSK3β, a component of the β-catenin destruction complex, and thus activates the pathway at the cytoplasmic level bypassing the reactions occurring at the plasma membrane and its vicinity 36 . We found that O. irrorata could also inhibit this LiCl-induced Wnt pathway activation (Fig. 4c). Curiously, only the highest tested concentration of the extract (the 1:100 dilution) could elicit a clear inhibition of the LiCl-induced pathway activation; further, the induced reduction was only ca. two-fold (Fig. 4c). This contrasts with the near-complete inhibition of the Wnt3a-induced response, already seen with lower concentrations of the extract, and IC 50 achieved at the ca. 1:400 extract dilution (Fig. 4a,b).

Discussion
Brittle stars (Ophiuroidae) is the largest class of echinoderms, with >2000 species known, of which ca. 400 species reside in the North Pacific 37 . Our finding that specimens from three different Ophiura species collected at different locations possess promising activities, inhibiting the Wnt pathway in TNBC cells, hints at the possibility that the same (classes of) anti-Wnt compound(s) may exist in other brittle stars.  Supplementary Table S1 and Fig. 1).
Pacific brittle stars have been the source of terpenes and phenylpropanoids (Ophioplocus japonicus 38 , steroid sulfates (Ophiopholis aculeata, Ophiura sarsi, and Stegophiura brachiactis 39 , gangliosides (Ophiura sarsi and Ophiocoma echinata 40 , Ophiomastrix annulosa 41 , and Ophiocoma scolopendrina 42 , and other compounds, some of which are further shared among other echinoderms. Multidimensional fractionation of extracts from O. irrorata or other brittle stars, coupled with activity characterization, followed by structure elucidation, are awaiting to identify the molecular nature of the anti-Wnt components we have described here. Subsequent tests in vitro and in vivo against TNBC (and other Wnt-dependent cancers) will be performed in order to advance these molecules along the path of the anti-cancer drug development 35 . With the recent draft genome sequencing of the brittle star Ophionereis fasciata 43 , potential molecular biology and genetic engineering approaches to characterization and biotechnological applications of the brittle star metabolome will be facilitated.
Extracts from the Persian Gulf Ophiocoma erinaceus have been reported to possess anti-proliferative activity against A2780cp cells, and anti-angiogenic activity in the fertilized Ross egg model 44 . Using HeLa cells, anti-migration and pro-apoptotic activities were reported 45 , and the latter were found to be mediated by the polysaccharide fraction of the O. erinaceus extracts 46 , while the anti-migration activity -by the saponins 47 . Pacific O.japonicus terpenes were also reported to have a certain cytotoxic activity 38 .
However, we must stress here the fundamental difference of our marine natural product-based drug discovery approach from those traditionally applied in the past and also now by many laboratories. We are explicitly not looking for cytotoxic agents. Furthermore, such activities are eliminated during our screening process. Instead, we look for natural compounds targeting one of the hallmarks of cancer -pro-proliferative oncogenic signaling 6 , with a special focus on Wnt signaling responsible for oncogenic transformation, growth and metastasis in a number of tissues, including the breast 22 . Of note, inhibition of Wnt signaling in breast cancer cells causes cell proliferation arrest but not cell death 26,31,35 .
The fact that the Ophiura irrorata extract inhibits LiCl and Wnt3a-induced response with different efficiencies suggests that several Wnt-modulatory components exist in brittle stars, acting at different molecular targets within the pathway. One of these targets lies above, and one -below GSK3β. We are particularly interested in the component(s) active against the higher-located molecular target(s). Indeed, inhibitors of the Wnt pathway acting below GSK3β are known to bluntly suppress all the tissue-specific subtypes of the pathway. Thus, not only the oncogenic Wnt signaling, but also physiologic involvements of the pathway relevant e.g. for renewal of epithelial cells in the gastrointestinal tract will be affected producing strong side effects. Instead, it is more promising to . The Y-axis shows the activity of the Wnt-dependent luminescence, normalized to the Wnt-independent Renilla luminescence, setting to 100% the normalized luminescence activity of cells treated with the Wnt3a-conditioned medium without any extract added. Curves are fit using the equation: , providing c as the IC 50 of the extract-induced inhibition, which is 0.0019 (1:526 dilution) for Peniagone sp. and 0.0121 (1:83 dilution) for the skin and muscular sac of Molpadia musculus. (b) Co-activation of the Wnt pathway can be observed for extracts (all shown for the 1:100 dilution) from several invertebrate groups: polychaete Travisia sp. (specimens 7, 17, 53, 64, 65), various Actinias (specimens 16,44,55,78), and gonads of M.musculus (specimens 9, 60, 90). Data are presented as mean ± sem. "*" signs highlight experimental points with statistically significant pairwise difference from control (no extract added) as determined by the two-way ANOVA (a) and the Student t-test (b); p value < 0.05 is indicated as "*", <0.01 as "**", <0.001 as "***", and <0.0001 as "****". Photographs of the specimens are provided on each panel. "n" stands for the number of experiments.
selectively target the upper levels of the pathway, where more variability among the signaling subtypes is provided in different tissues and in different contexts (i.e. physiologic vs. pathologic) 22,28,48 .
We consider the pipeline of marine-based drug discovery presented here to represent a pioneering approach to deliver novel targeted drug candidates. This pipeline is composed of invertebrate collection during dedicated sea expeditions, specimen extraction, screening using specific assays monitoring oncogenic signaling in cancer cells, and preliminary characterization of the mode of action of the hit extracts within the pathway. As exemplified in this report, such an approach has resulted in identification of promising anti-Wnt and Wnt pathway-modulating activities in Ophiura species and also in a number of other invertebrates. Subsequent work will advance these activities towards drug candidates, but also will feed more marine organisms into this pipeline.  Table S1 and Fig. 1). The specimens were placed in 10 ml 80% ethanol immediately after collection and extracted by storage at 4 °C for 6 months. Each extract sample was derived from a single individual organism of the size average for its species. In case of separate organs or body parts, they were dissected immediately after catch and placed separately into 10 ml 80% ethanol. Before experiments, the ethanol solution, in which the samples had been stored, was collected and centrifuged to remove the debris; the supernatants were concentrated 100-fold in the following way: 1 ml of the extract was put in a 1.5 ml tube and evaporated under vacuum at 30 °C using the Eppendorf concentrator 5301. The dry pellet was then re-dissolved in 10 μl DMSO and stored at −20 °C.

Methods
Ethics statement. Invertebrate sample collection was performed by the standards approved by the Ministry of Education and Science (Russia), and Federal Ministry of Education and Research (Germany). All efforts were made to minimize suffering.
Luciferase-based assay of the Wnt-dependent transcriptional activity. BT-20 (A.T.C.C. catalogue number HTB-19), a triple-negative human mammary carcinoma line cells stably transfected with the M50 Super 8 × TopFlash plasmid 31 were used to analyze the Wnt modulatory activities of extracts from marine invertebrates. The assay was performed in white tissue-culture-treated 96-well plates (Greiner). The BT-20-Tf cells were seeded in 100 μl of DMEM containing 10% FBS at ca. 10000 cells/well, incubated overnight at 37 °C, 5% CO 2 and subsequently stimulated by 50 μl of the Wnt3a-conditioned medium or LiCl (to 15 mM) in the presence of the invertebrate extracts to the final concentration of DMSO of 0.25%, or DMSO alone, for 12-24 h. If indicated, the cells were additionally transiently transfected by the pCMV-RL plasmid (purified with Quiagen Plasmid Midi Kit 12145) for constitutive expression of Renilla luciferase 50 using XtremeGENE 9 reagent (Roche) according to the manufacturer's protocol 16-24 h prior to treatment with the Wnt3a conditioned medium and the extracts. The medium was subsequently removed and 15 μl of the lysis buffer (25 mM glycylglycine, pH 7.8, 1% Triton X-100, 15 mM MgSO 4 , 4 mM EGTA, 1 mM DTT) were pipetted into each well. After incubation for 5 min at room temperature, the 96-well plate was analyzed using the Victor3 Multilabel Counter (PerkinElmer) or Cytation 5 (Biotek) with the two-channel dispensing unit primed with the buffer solutions for separate activity measurements of firefly and Renilla luciferase, prepared as described 51 Tables S1 and S2), which is available to authorized users. Data availability statement. All data generated or analysed during this study are included in this published article and its Supplementary Information files.