Construction of TUATinsecta database that integrated plant and insect database for screening phytophagous insect metabolic products with medicinal potential

Phytophagous insect larvae feed on plants containing secondary metabolic products with biological activity against other predatory organisms. Phytophagous insects can use their specialised metabolic systems to covert these secondary metabolic products into compounds with therapeutic properties useful to mankind. Some Asians drink tea decoctions made from phytophagous insect frass which is believed to be effective against inflammatory diseases. However, insects that can convert plant-derived secondary metabolic products into useful human therapeutic agents remain poorly studied. Here, we constructed the TUATinsecta database by integrating publicly plant/insect datasets for the purpose of selecting insect species. Using TUAT-insecta we selected the Asian swallowtail butterfly, Papilio xuthus larvae fed on several species of Rutaceous plants and examined whether the plant-derived secondary metabolites, especially those present in frass, were chemically altered or not. We extracted metabolic products from frass using three organic solvents with different polarities, and evaluated solvent fractions for their cytotoxic effects against several human cell lines. We found that chloroform frass extracts from P. xuthus larvae fed on Poncirus trifoliata leaves contained significant cytotoxic activity. Our findings demonstrate that screening of insect species using the ‘TUATinsecta’ database provides an important pipeline for discovering novel therapeutic agents that might be useful for mankind.

Next, using TUATinsecta, we investigated the biological activity of each plant fed on Lepidopteran insects; among the available studies, we encountered reports on diabetes-related (0.18%), cardiotonic (0.85%), antioxidant (1.5%), anti-inflammatory (2.3%), and anticancer activity (2.4%) in these plants, which had already had been used historically as folk medicine or herbal medicine. We were interested in whether or not the insect metabolic system changes second metabolites from these plants. Information in the database suggested that these plant second metabolites could change through the metabolic system in the insect. Some of the uses of Lepidopteran insects for secondary metabolites of plants included changing part of their chemical structure to become pheromones, kairomones, or defensive substances [15][16][17][18] . To compare the biological activity among the varieties of host plants metabolised by insect, we chose the Asian swallowtail P. xuthus, which eats several species of Rutaceae plants.

Comparison of substances in different plants through swallowtail metabolic processes. The
Asian swallowtail, P. xuthus is Lepidopteran insect. Their life stages are shown as Fig. 2a-d. They feed on several species of Rutaceae plants during the larval developmental stage (Fig. 2b), and we examined whether secondary metabolic products including these plants may be altered further through the metabolic process in P. xuthus. Using three types of organic solvents of differing polarity, we extracted the metabolic products from each frass of P. xuthus larvae which fed on Poncirus trifoliata (Fig. 2e), Zanthoxylum piperitum (Fig. 2f), Citrus sudachi (Fig. 2g), Citrus natsudaidai (Fig. 2h), or Citrus junos (Fig. 2i). In Supplementary Table 1 we show the dry weight and extraction weight in each sample. The extraction dry weight of all frass methanol extracts (MeOH ext.) were higher than those of n-hexane extracts (Hex ext.) and chloroform extracts (CHCl 3 ext.) in this study.

Scientific Reports
| (2020) 10:17509 | https://doi.org/10.1038/s41598-020-74590-z www.nature.com/scientificreports/ We compared each metabolic product between leaf and frass in MeOH ext., Hex ext., or CHCl 3 ext., by thin layer chromatography (TLC). Consequentially, we found several spots in the frass extracts which not evident in each corresponding plant extract (Fig. 3a- Observation of the morphological changes to human cancer cell lines. We evaluated the cytotoxic activity of the substances extracted from each sample on three human cancer cell lines: the human liver cancer cell line (HepG2), the human uterine cancer cell line (HeLa), and the human pancreatic cancer cell line (MIA Paca-2). First, we examined morphological changes using the phenotypic change score in these human cancer cell lines (Supplementary Table 2, a-g). We found that the frass from P. trifoliata CHCl 3 ext. induced cell death in all three human cancer cell lines (Supplementary Table 2, a, c, e). We observed the strongest phenotypic change based on the phenotypic change score in the frass from P. trifoliata CHCl 3 ext. on HepG2 and MIA-Paca2 (Fig. 4a, c). The frass from P. trifoliata CHCl 3 ext. suppressed HepG2, HeLa, and MIA Paca-2 cell growth and cell viability, or induced morphological changes as assessed by the phenotypic change score. The CHCl 3 ext. from pupae which had been fed on P. trifoliata had no effect on HepG2, HeLa, or MIA-Paca2 cells (Fig. 4, Supplementary Table 2-g).      ; however, we did not detect such a spot in P. trifoliata frass CHCl 3 ext. Taken together, TLC analysis indicates that the composition of the metabolic products in P. trifoliata frass CHCl 3 ext. differed from those in the P. trifoliata leaf CHCl 3 ext.. Comparison of P. trifoliata frass CHCl 3 ext. and C. trifoliata leaf CHCl 3 ext. by high performance liquid chromatography (HPLC) revealed two peaks (Fig. 6a, arrows). These peaks were not observed in P. trifoliata leaf CHCl 3 ext. (Fig. 6b). The peaks of the retention times were 26 min and 39 min (Fig. 6a, arrows).

Discussion
Here, we constructed "TUATinsecta" by data integration to efficiently find the phytophagous insect species to screen for pharmaceutical resources and to experimentally determine whether the metabolic products from the host plants would be changed by insect metabolic system.  Comparison of substances from leaves with those from frass extract in P. trifoliata. We compared CHCl 3 extracts from leaves of P. trifoliata and frass of P. trifoliata using HPLC.

Scientific Reports
| (2020) 10:17509 | https://doi.org/10.1038/s41598-020-74590-z www.nature.com/scientificreports/ First, using several insect databases 19,20 , we entered the information regarding insect species which fed on plants, thereafter adding the information from the KNApSAcK database 21 about their host-plant substances and their biological activity; we then created connections among the various information fields in the TUATinsecta database.
We found 2857 insect-host plants relationships with some biological activity on the organisms. We investigated the information on biological activity for each host plant, and discovered reports of anticancer activity for many of them. We also found reports on diabetes, cardiotonic, antioxidant, anti-inflammatory and antibacterial effects in many host plants based on past records of folk medicines and herbal medicine.
Phytophagous insects take up metabolic products from their host plants during the larval developmental stage. These substances are mainly hydrocarbons, proteins and lipids, which are categorised as primary metabolites 22 . At the same time, phytophagous insects must ingest several types of secondary metabolites-such as alkaloids, terpenoids, flavonoids, polyketides or phenylpropanoids-from their host plants when they feed on them 23 . These secondary metabolites may have affect larval growth, development and eating behaviour 24,25 .
The presence of enterobacteria has been confirmed in Papilio polytes, but most are from plants. These bacteria do not seem to be involved in the metabolic process in Papilio polytes 26 . Furthermore, the administration of antibiotics to larvae of Danaus chrysippus and Ariadne merione does not affect their growth and survival 27 . Thus, we speculated that the enterobacteria are not involved in the metabolism of secondary plant metabolites in Lepidopteran insects. As a further support, the pupal extract does not affect the viability of human cancer cells. Accordingly, we considered that the larva must have a specific metabolic system when feeding on host plants for avoiding the effects of these secondary metabolites on their growth, development and eating behaviour. Presumably, host plant secondary metabolites are processed in the larval body through their special metabolic system and appear in an altered form in frass.
Cytochrome P450 (CYP) plays a role for detoxication of these secondary metabolites in phytophagous insects and has been suggested to relate to the host plant specialisation in phytophagous insects 28 . In the detoxication process, specific plant secondary metabolites are metabolised by CYPs, and the final metabolites are then processed to frass. Both the insect body and the larval frass may also contain metabolised substances not included in the original plant substances. For this reason, we anticipated that frass would contain biologically active agents because the secondary metabolites from the host plant could be changed through the insects' metabolic system. From this standpoint, we examined the biological activity of five kinds of frass obtained from the Asian swallowtail P. xuthus, which feeds on P. trifoliata, Z. piperitum, C. natsudaidai, C. junos or C. sudachi.
In comparing the biological activity of each extract, we found that CHCl 3 ext. P. trifoliata frass induced strong cell death and morphological change in the HepG2, HeLa and MIA-Paca2 cell lines; CHCl 3 ext. from C. junos frass, however, induced strong cell death and morphological change only in the HeLa cell line. In contrast, CHCl 3 ext. from the frass of Z. piperitum, C. sudachi, and C. natsudaidai did not induce strong cell death and morphological change to the HepG2, HeLa or MIA-Paca2 cell lines. We speculated plants secondary metabolite from P. trifoliata or C. junos might be altered the chemical structure by the metabolic system in P. xuthus.
These altered substances included in the CHCl 3 fraction could induce cell death. CHCl 3 is a less polar solvent. We thus considered these substances potentially to be lipophilic compounds of low molecular weight.
Plants in the Rutaceae include flavonoids, coumarins, carotenoids and limonoids. Some of these substances have effects on the human cancer cell line (Supplementary Table 4) 29,30 . The coumarin derivative, auraptene found in the C. natsudaidai was reported to show anticancer activity 31 . We were curious as to whether the auraptene was included in the frass or the leaf extract of P. trifoliata; HPLC indicated an auraptene peak at 44 min at RT in the Hex ext. of frass and leaf of P. trifoliata (Supplementary Fig. 4a and b). However, neither of the Hex exts. induced cell death in the HepG2, HeLa or MIA-Paca2 cell lines (Supplementary Table 2, and Fig. 5a,d,g). Therefore, we hypothesised that a compound other than auraptene was altered by the P. xuthus metabolic system to induce cell death. According to TUATinsecta, P. trifoliata includes several known biologically active substances such as xanthotoxol and imperatorin. These compounds lead to proliferation in the HeLa cell line. In comparing P. trifoliata frass and leaf CHCl 3 ext. by HPLC, we identified two specific peaks at 26 min and 39 min for CHCl 3 ext. from P. trifoliata frass using HPLC (Fig. 6). However, we could not identify chemical property on these specific peaks. We predict that a very small amount of the substance or complex of several substances in the frass of P. trifoliata can induce cell death.
Some insect frass have a history of use as a folk medicine. In China, San-sha-the dried frass obtained from young silkworm larvae, a known as Chinese herbal medicine-has been used against inflammatory diseases such as rheumatism, joint pain, neuralgia, uterine bleeding, irregular menstruation, and conjunctivitis 32 . In addition, China has a tea called Chongshicha, made using the dried frass obtained from several kinds larvae of Pyralidae or Noctuidae 33 . Malaysia also has a tea made using dried frass obtained from stick insects that feed on guava leaves 34 . Some Malaysians believe this tea is effective against asthma, gastrointestinal disorders, and muscle pain 35 . Thus, Asian countries have a tradition of drinking tea made from frass of phytophagous insects by decoction. Silkworms eats mulberry leaves and stick insects eat guava leaves, and these plants include biologically active substances [36][37][38][39] . In this way, insect frass has long been used as folk medicine; however, the plant substances exhibiting pharmacological actions have, to date, not been clearly identified. Thus, if the plant contains secondary metabolic products that exhibit biological activity, it could be changed by insect metabolism into the metabolic products with even stronger biological activity. In this study, we integrated information on insect-host plants, host plant-secondary metabolites, and metabolite-biological function into TUATinsecta. TUATinsecta will assist investigators to find specific compounds, including plant-derived secondary metabolic products, in specific plants. Therefore, TUATinsecta is a helpful resource for choosing phytophagous insect species which could prove useful in identifying pharmaceutical candidates beneficial to humans.
In conclusion, we constructed a database for choosing phytophagous insect species which could convert plant-derived secondary metabolites through the insect metabolic system to metabolic products with biological Scientific Reports | (2020) 10:17509 | https://doi.org/10.1038/s41598-020-74590-z www.nature.com/scientificreports/ activity. With the aid of TUATinsecta we chose candidate insect species and identified P. xuthus, which feeds on several species of Rutaceae plants. Using three types of organic solvents, we extracted the metabolic products from the frass produced by the metabolic system of P. xuthus after feeding on each Rutaceae plant. We evaluated the biological activity of these extracts by cell death to the human cancer cell line, and found that P. xuthus produced a special substances included in the frass when P. xuthus fed on P. trifoliata. Our results demonstrate that Lepidopteran insects can alter plant metabolic products, creating new substances with different biological activity. However, identification of these biologically active substances must await development of large-scale insect frass collecting methodologies.
In future studies, we will elucidate the biologically active substance included in the frass when P. xuthus fed on P. trifoliata, and try to convert the biological activity of substances by P. xuthus metabolic enzymes.
Furthermore, we intend to construct a technology safe to elaborate to artificially alter the chemical structure of plant-derived secondary metabolic products by insect metabolic enzymes.
Collectively, the results of our study will contribute to the identification of phytophagous insect species that have the ability to change plant substances into potential therapeutic agents via the insect metabolic system.

Methods
Construction of the dataset for TUATinsecta. To determine the relationships among insects, plants and their substances, we developed a database connecting the information among these fields of study. First, we obtained information for host plants of insects by downloading the Encyclopedia of Life database (https:// eol. org/) 19 for the "eats" attribute as of August 2019. We added further information for host plants to the retrieved data from the database 20 . We also searched field guides and extracted host record information. We then obtained information for substances included in all insects-host plants from the KNApSAcK database (https:// www. knaps ackfa mily. com) 21 , which contains species-metabolite relations for plants and the biological activities and functions of these metabolites. We also extracted descriptions of biological functions from PubMed (https:// www. ncbi. nlm. nih. gov/ pubmed). Finally, we integrated this information-including insect species, insect-host plants, plant substances, and their biological activity-using a tab-delimited formatted file. We also established hyperlinks to the corresponding web pages in the NCBI Taxonomy to facilitate the access to this information from TUATinsecta using the scientific names of insects and plants. We used TogoDB platform (Database hosting service, https:// togodb. org/) for the visualization of the tab-delimited file. We then published the dataset as TUATinsecta on the platform. Imported data are converted to an RDF-ized format. TUATinsecta is freely available at https:// togodb. org/ db/ tuat_ insec ta under Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license (https:// creat iveco mmons. org/ licen ses/ by-sa/4. 0/). All users can search targeted information with specific keywords derived from insect, insect-host plants, substances or biological activity. Users can also download these data in CSV, JSON and RDF format. We will update TUATinsecta when each public database used for the construction of TUATinsecta is updated.

Scientific Reports
| (2020) 10:17509 | https://doi.org/10.1038/s41598-020-74590-z www.nature.com/scientificreports/ solvent until the solvent front reached approximately 7 cm from the origin. We chose chloroform:methanol, 20:1 (v/v) for n-hexane ext., chloroform:methanol 10:1 (v/v) for CHCl 3 ext. and, chloroform:methanol:distilled water = 6: 4: 1 (v/v) for MeOH ext. as the developing solvents on TLC. For visualising each spot indicating an substance, we excited the spots by ultraviolet light (short wavelength; 254 nm, long wavelength; 365 nm) using a UV irradiator (UVGL-15, Funakoshi Co., Ltd., Tokyo, Japan). Thereafter, we also sprayed the TLC plate with 50% sulfuric acid (Wako Pure Chemical Industries, Ltd., Osaka, Japan) and heat-treated it at 120 °C for 2 min after the extract had separated. We recorded the position of each spot relative to the solvent front (Rf). Rf value was calculated as the distance from baseline traveled by the solute divided by the distance from baseline traveled by the solvent (solvent front).
Preparation of chemicals and evaluation of cell viability. We dissolved each extract in dimethyl sulfoxide (DMSO; Sigma-Aldrich, MO, USA) to a concentration of 1 mg/mL, 10 mg/mL, and 100 mg/mL to prepare an extract solution. Each extract solution was protected from light by shading. We used cisplatin (CDDP; Wako Pure Chemical Industries, Ltd., Osaka, Japan) as a positive control (PC). We dissolved CDDP in phosphate buffered saline (PBS; 8.1 mM NaHPO4, 137 mM NaCl, 2.7 mM KCl, 1.47 mM KH2PO4, pH 7.4), adjusted to 1 mM, and protected it from light by shading. We used an equivalent concentration (v/v) of DMSO as negative controls.
We obtained the human liver cancer cell line HepG2 (RCB1886), the human cervical cancer cell line HeLa (RCB0007), and the human pancreatic cancer cell line MIA Paca-2 (RCB2094) from RIKEN cell bank (RIKEN Tsukuba, Japan). We cultured each cell line in Dulbecco's modified Eagle's medium (Wako Pure Chemical Industries, Osaka, Japan) supplemented with 10% (v/v) fetal bovine serum (Sigma-Aldrich, MO, USA), 100 U/ml penicillin, and 100 μg/ml streptomycin solution (Wako Pure Chemical Industries, Ltd., Osaka, Japan), at 37 °C under 5% CO 2 . We cultured cell lines until sufficient cell stock obtained and then stored cells at − 80 °C until use. The second passage number of cells was used in all experiments. To assess the effect of the extract on each cell line, we examined cell viability by MTT assay (Merck Millipore, CA, USA).
We seeded each cell line at 2 × 10 3 cells/50 μL/well on a 96-well plate (Nippon Becton Dickson Co., Ltd., Tokyo, Japan), and then cultured them for 24 h at 37 °C under 5% CO 2 . After 24 h, we treated each well with the extract solution at final concentrations of 1 μg/mL, 10 μg/mL, and 100 μg/mL (0.1% (v/v) DMSO), respectively, and continued the culture for a further 72 h. We used cells treated with DMSO at a concentration of 0.1% (v/v) as a negative control. We added 10 μL of MTT reagent to each well of the 96 well plate after culture for 72 h, and performed the culture for a further 3 h. Thereafter, we added 100 μL of a solution of 0.04 N hydrochloric acid-isopropanol solution to each well to dissolve the formazan dye included in the cells, and then measured the absorbance at 540 nm and 620 nm using a plate reader (iMark, Bio Rad, The absorbance CA, USA). We calculated the cell viability with the following equation: Observation of cell morphology. We observed the cell morphology of the extract on each cell line.
We seeded each cell line at 0.5 × 10 5 cells/500 μL/well on a 24-well plate (IWAKI, Shizuoka, Japan) and cultured them for 24 h at 37 °C. under 5% CO 2 . After 24 h, we treated each well with the extract solution at final concentrations of 1 μg/mL, 10 μg/mL, and 100 μg/mL (0.1% (v/v) DMSO), respectively, and continued the culture for a further 72 h. We used CDDP at 1 mM as a PC. After 72 h, we observed the cell morphology using a stereo microscope (Floid Cell Imaging Station, Thermo Fisher). We categorised the cell morphology into five levels and scored it as follows: score 1: the cell proliferation and density are similar to those of the negative control group; score 2: the levels of cell proliferation and density are between those for score 1 and those for score 3; score 3: the cell density has decreased without cell morphological change; score 4: the cell density has decreased to between that for score 3 and that for score 5 by cell death; score 5: the cell density has significantly decreased, similar to that for the PC group, by cell death.
Comparison of both extracts by high performance liquid chromatography. We used CHCl 3 extracts from P. trifoliata leaf and frass, respectively. We adjusted the concentration of each extract to 10 µg/mL with acetonitrile (Sigma-Aldrich co. ltd., MO, USA) and then filtrated them using a 0.2 mm x φ 4 mm filter (AS ONE co. ltd. Osaka, Japan). We collected each filtrate in a 300 μL vial (Waters co. ltd. Tokyo, Japan).
Statistical analysis. Statistical significance was determined by a two-tailed Student's t-test using Excel