Cytotoxic screening of plants from the Brazilian Atlantic Forest has led to the identification of Casearia arborea and Sorocea hilarii as sources of antitumor compounds

Abstract In the present study, we have evaluated the cytotoxic activity of 282 extracts from 72 native plant species of the Brazilian Atlantic Forest biome. As a result, Casearia arborea and Sorocea hilarii leaves extracts showed cytotoxic activity against three tumour cell lines tested (B16F10, SW480 and Jurkat). After bioassay-guided fractionation, the bioactive fractions were submitted to the dereplication study via High-performance Liquid Chromatography, connected to High-resolution Mass Spectrometry (HPLC-ESI-QTOF/MS) analysis, combined with a Global Natural Products Social Molecular Networking (GNPS) tool. A combination of bioactivity-guided and dereplication approaches resulted in the putative annotation of 27 clerodane diterpenes and 9 flavonoids as main compounds present in the cytotoxic fractions of C. arborea. Regarding the active fraction of S. hilarii, 10 megastigmans, 17 spirostane steroids derivatives and 2 lignans were putatively identified. In conclusion, Casearia arborea and Sorocea hilarii are potential sources of antitumor compounds. Graphical Abstract


Introduction
the introduction of the Global natural products Social molecular networking (GnpS), employing tandem mass Spectrometer-coupled Liquid Chromatography data, has contributed to the study of the dereplication of metabolites in complex matrices as plant extracts (Wang et al. 2016). in this study, we report the cytotoxic screening of 282 extracts from 72 plant species native to the Brazilian atlantic Forest and identified extracts from Casearia arborea and Sorocea hilarii leaves with potential cytotoxic activity.Bioguided fractionation of these extracts, combined with dereplication studies, led to the identification of compounds related to the antitumor activity of the species.

Cytotoxic screening in cancer cell lines
the cytotoxic screening of 282 extracts from 72 species of 35 botanical families was evaluated by a mtt assay (Supplementary table S1).preliminary results show that 72 extracts reduced cell viability by more than 70% for at least one tumour cell line.among these, some extracts showed specific cytotoxic activity for a specific tumour cell line, while others were active in common for two or three cancer cells, as shown in the Venn diagram (Supplementary Figure S1).Based on the Venn diagram, 14 extracts inhibited 70% or more cell growth for the three tumour cell lines selected to determine the Gi 50 (Supplementary table S2).From this result of Gi 50 , the dCm:meoH (1:1) extracts from the leaves of Casearia arborea (Rich.)urb.(Salicaceae) and Sorocea hilarri Gaudich (moraceae) showed the lowest Gi 50 values for the three cell lines tested and were, therefore, considered promising for cytotoxic natural product identification (He et al. 2019).

Fractionation bioguided by cytotoxicity of promissory extracts
in the case of C. arborea extract, the dichloromethane fraction concentrated the bioactive compounds, providing a significant increase for both cell lines (p < 0.05) in cytotoxic activity regarding the initial extract.Furthermore, this fraction exhibited a high selectivity index (Si) for Vero cells.there was no significant difference between the Gi 50 of the Sorocea hilarii dCm:meoH (1:1) extract (24.3 ± 3.6 µg/mL) compared with its bioactive n-butanol fraction (23.8 ± 0.6 µg/mL) for B16F10. in SW480, there is a reduction of the Gi 50 from 18.1 ± 0.8 µg/mL to 9.2 ± 1.1 µg/mL from extract to fraction, respectively.However, the n-butanol fraction showed a selectivity index ˂ 1, compared with non-cancerous Vero cells (table 1).

Dereplication analysis of active fractions
the HpLC-ESi-HRmS data analysed in the GnpS platform for dichloromethane fraction of C. arborea and n-butanol fraction of S. hilarii.mS/mS data dereplication was conducted by searching spectral libraries, which allowed the annotation of 39 compounds in C. arborea (Supplementary table S3) from 6 different molecular families, including 27 clerodane diterpenoids, 1 flavan-3-ol, 2 proanthocyanins, 5 flavonols, 3 flavones and 1 cinnamic acid derivative (Figure 1). the main product ions formed by mS/mS involved neutral losses of 18, 42, 60, 150, and 168 da, mainly associated with the cleavage of the ester groups, as acetate and benzoate derivatives, present in the basic structure of clerodane diterpenes (Supplementary Figure S2).this fragmentation pattern, which is in agreement with reports published in the literature, implies the presence of casearin-type clerodane diterpenes (danuello et al. 2020).other   metabolites, present in the active fraction, are phenolic compounds, which are clustered into a molecular family, such as flavonols and proanthocyanidins (Supplementary table S3).Some of the clerodane-type diterpenes and flavonoids (Figure 1), identified in the dichloromethane fraction, were also identified in C. arborea by Beutler et al. (2000) and Santos et al. ( 2021), respectively.nevertheless, this is the first time that the cytotoxic activity of this species has been reported to colon adenocarcinoma and Jurkat cancer cell lines.there are studies reporting the antitumor activity of C. arborea to metastatic melanoma tumour lines a2058 (Garo et al. 2017) andLoX imV i (Beutler et al. 2000), and also to gliosarcoma (SF539) (Beutler et al. 2000).Both of these studies attributed the antitumor activity of C. arborea to clerodane diterpenes.this data reinforces that clerodane-type diterpenes, identified in the active dichloromethane fraction, are responsible for the cytotoxic activity of C. arborea.
the resulting molecular networking of the n-butanol active fraction from Sorocea hilarii allowed annotating 29 compounds (Supplementary table S4), categorised into three different clusters, including 10 megastigmanes, 17 spirostane steroids and their saponin derivatives and 2 lignans (Figure 2). the molecular family referring to megastigmanes was initially constructed from four spectral matches (m/z 373, 387, 389, and 505), and followed by fragmentation analysis, generated by connecting nodes.during mS/mS fragmentation, it was possible to notice that megastigmanes lost sugar units, such as hexoside and pentoside, resulting in aglycone exposure.another molecular family annotated in the S. hilarii fraction was steroidal saponins with spirostane-type aglycone.the generated cluster, based on mS/mS fragments, includes the sapogenins spirostenol and spirostadienol, with an m/z 415.3206 and 413.3048, respectively.the remaining 14 saponins annotated are mainly derived from these two sapogenins, showing losses of sugars.additionally, a minor cluster and isolated nodes were characterised as lignans, annotated with five compounds (Supplementary table S4).
the species S. hilarii is endemic to Brazil, and there are no studies reporting its chemical or pharmacological properties.megastigmanes, spirostane steroids derivatives, and lignans were reported for the first time in S. hilarii (Figure 2).there are several reports of the cytotoxic activity of megastigmane glycosides for various tumour lines, including colon cancer (aati et al. 2022) and melanoma (panza et al. 2011). in the molecular families of lignans, podophyllotoxin stands out, a lignan with high cytotoxic and anticancer power that has been widely studied and used in clinical therapy (Zhao et al. 2021).these natural products, identified in the n-butanol fraction of S. hilarii, can be considered the probable cytotoxic actives of this species.

Conclusions
applying a molecular network for the active fractions of Casearia arborea and Sorocea hilarii, combined with silico dereplication approaches, it was possible to characterise the constituents responsible for the cytotoxicity of these two species.in the case of C. arborea, clerodane diterpenes and S. hilarii, mesgastigmans derivatives, and lignans are possibly the compounds responsible for the cytotoxic activity.
letters in the same column indicate no significant differences between means (p < 0.05) by the one-way aNoVa followed by dunnett's post-test.cytotoxic assays were performed in triplicate with three independent experiments.results were expressed as means ± standard error.DCM:MeOH -dichloromethane: Methanol.**selectivity index (sI) = GI 50 (Vero)/GI 50 (B16F10).

Figure 1 .
Figure 1.(a) Molecular network from uhPlc-esI-hrMs data in its positive setting for the dichloromethane fraction of Casearia arborea leaf extract.(B) detail for the main cluster (in magenta) of the molecular network, referring to the clerodane diterpenoids compounds.*Numbers inside the nodes correspond to the m/z accurate masses (da).**Nodes manually annotated are circled in red.Nodes circled in gray did not show spectral match.

Figure 2 .
Figure 2. (a) Molecular network from hPlc-esI-hrMs data in a positive mode for the n-butanol fraction of Sorocea hilarii leaf extract.(B) orange cluster corresponds to spirostane steroids and their saponin derivatives compounds.(c) the green cluster corresponds to megastimanes compounds.*Numbers inside the nodes correspond to the m/z accurate masses (da).**Nodes annotated through the GNPs spectral library are circled in blue.***Nodes manually annotated are circled in red.**** Nodes circled in gray did not show spectral match.***** saccharide chain consists of pentosides, deoxyhexosides, and hexosides that are linked to the aglycones by O-glycosidic bonds.

Table 1 .
concentration of extracts or fractions from Casearia arborea and Sorocea hilarii leaves that inhibited 50% of cell growth GI 50 (µg/ml) and selectivity index (sI).