In silico study in toxicity parameters of Pigment Derivated Compounds of Monascus sp. mold as a cervical anti-cancer drugs candidate

: Toxicity prediction is very important for the development and design of new drugs because computational toxicity estimates are not only faster than the determination of toxic doses in animals, but can also help reduce the number of animal trials. The test uses pkCSM, Protox II, Toxtree, preADMET and T.E.S.T. from the results of research that has been carried out on 57 pigment derivated compounds of Monascus sp. mold, the results of the pkCSM application are 39 test compounds for the Protox-II application there is 1 compound, the Toxtree application produces 1 compound, for the PreADMET application 4 safe compounds are produced, and for the T.E.S.T application produces 1 compound because it fulfills one of the aspects of ICH (International Conference on Harmonization) S9: non-clinical evaluation for anticancer pharmaceutical 2010 and has the potential as a candidate for anticancer drugs.


INTRODUCTION
With the existence of modeling predictions by QSAR, it is possible to predict drug toxicity easier and perhaps even before drug synthesis. Quantitative Structure-Activity Relationships (QSAR) has been widely used in toxicology to predict the liability of novel compounds using structural features of known toxicants. During the last 40 years, many QSAR models have been published predicting carcinogenicity 1 and mutagenicity 2 . Toxic effects are generally categorized according to the site of the toxic effect. In some cases, the eff 3activity relationship has been widely used in the world to predict toxicity using computers 4,5 .
One of the new drug candidate alternatives is the pigment produced by Monascus sp. Monascus sp. is one type of mold used for fermentation of Angkak. One of the metabolites produced by Monascus sp. is a pigment. Monascus sp. pigment research has progressed very rapidly, there are 3 main pigments from Monascus sp. namely yellow, orange and red. From these main pigments, several pigment-derived compounds are found which are widely used as natural coloring agents in the textile, food, pharmaceutical and cosmetic industries 6,7,8,9 . According to research by Singgih 10 , 19 pigment derivates from Monascus sp. have been tested to have smaller bond energy value than Genistein as natural ligand. Genistein is an anti-cancer drug that has strong potential in the treatment and prevention of breast cancer because it has anti-proliferative properties. Genistein has an antiesterogenic effect, namely the ability to stop the estrogen hormone binding to cells in order to prevent cancer from growing and dividing. Through binding to ERβ through inhibition of the activity of enzymes involved in estrogen metabolism. Monankarin C with the smallest bond-free energy value is -11.08 Kcal / mol, which means the most stable complex conformation, genetic compounds form hydrogen bonds with Arg346, Glu305, and His475. Monankarin C also has the same interaction so that it is expected to have a better affinity than natural ligands.
The compounds with the next lowest free energy are PP-V Monaskorubin and Monaskin, the best compounds as a candidate for breast cancer drugs, namely Monaskin pigments because they have smaller bond energies than natural ligands. However in PP-V, Monascorubrin, Monankarin C, although they have a small bond energy, this compound is seen from the toxicity test of Monarkarin C which is mutagenic, carcinogenic and toxic to reproduction, whereas in PP-V and Monascorubrin it is toxic to reproduction. Many have been shown to have anticancer activity, but their use is limited due to side effects and high toxic effects 11 . Nevertheless, toxicity can be assessed using computational resources (computational algorithms, software, and data) to organize, analyze, model, simulate, visualize, or predict chemical toxicity 12,13,14,15 .

Materials Preparation
Prior to the in silico toxicity test, 57 pigment derivated compounds of Monascus sp., their molecular structures were drawn using Chem Draw Ultra Version 12, stored in *.sdf or *.pd files and then made in SMILES form using the Chem Draw application.

RESULTS AND DISCUSSION
Based The toxicity of drug impurities is also closely related to their structure. Structure-activity relationships have been widely used to predict toxicity using 4,5 . The determination of toxicity testing criteria is based on a protocol issued by the ICH (International Conference on Harmonization) S9: non-clinical evaluation for pharmaceutical anticancer, 2010, which includes: General toxicity includes MTD (Maximum Tolerated Dose), and LOAEL (Low observed adverse effect level), toxicity to reproduction or fetal development (DTP = Developmental toxicity potency), mutagenicity and carcinogenicity.
There are many tools for predicting the toxicity of a molecule, some of which are commercial in nature, some of which are online web servers and some of which can be downloaded freely. In the prediction of toxicity 20 in silico, the type of toxicity assessment uses computational resources (algorithms, software, and data) to organize, analyze, model, simulate, visualize, or predict chemical toxicity20. In predicting the toxicity of in silico, pigment derivated compounds of Monascus sp. mold was performed using pkCSM, Protox II, Toxtree, preADMET and T.E.S.T. The data generated from this in silico test can be used to design further testers.
The performance of the pkCSM software in the external validation dataset showed 83.8% accuracy in the mutagenicity test. There are several pkCSM end points, LD50, AMES test, maximum daily dose, and hepatotoxicity 17 . Based on our research, the toxicity data obtained from the pkCSM application for liver toxicity (hepatotoxic) contained 15 positive compounds that could cause damage to the liver, for mutagenesis data on the pkCSM application used the AMES method namely mutagenesis validation test method. There are 5 compounds that are mutagen positive.
The prediction of ProTox-II hepatotoxicity has a balanced accuracy of 82.00% on cross validation and 86.00% on external validation. In our research for hepatotoxicity, it is not stated that there are compounds that cause damage to the liver. Drug-induced hepatotoxicity is a significant cause of acute liver failure and one of the main reasons for drug withdrawal from the market.
Drug-induced liver injury (DILI) is a chronic process or rare event. The conceptual mechanism of DILI is direct cell stress, direct mitochondrial damage and specific immune reactions for the carcinogenicity test, 19 compounds were identified that could affect genes and damage normal cells so that they could become cancer cells. For mutagenesis data carried out using this application, it was found that there were no compounds that could cause mutations. For LD50, there are 2 compounds belonging to class II, 24 compounds belonging to class III, 28 compounds belonging to class IV and there are 3 compounds belonging to class VI based on the Globally Harmonized System 21,22. Toxtree's performance in the external validation dataset shows 70% accuracy and 78.3% sensitivity in the carcinogenicity test and 78% accuracy for the mutagenicity test. Toxtree represents the end point of various toxicities, namely the Cramer Rules to see from its functional groups, Kroes TTC to estimate the exposure threshold for drug compounds in humans, Benignin Bosa rules for carcinogenicity (genotoxic and non-genotoxic) and mutagenicity in vitro ( AMES test) 23,24. In the Toxtree test according to Cramer's rules, there is 1 compound belonging to class I, which is the lowest level of toxicity, while 56 other compounds are included in class III, which means that the highest level of toxicity contains a functional group, the substituent ring which is a marker for the compound is toxic and even possible to have toxicity which is significant and it is estimated that this compound is not guaranteed its safety because it is a substance with a chemical structure that does not allow initial assumptions of safety or may even suggest significant toxicity or has reactive functional groups 25 . For the Benignin / Bossa rules parameter, which is to determine which compounds can cause carcinogenicity and mutagenicity, 57 compounds tested positive for mutagens were obtained. This is intended as an initial stage of mutagenic in vivo screening, molecular functional groups that are known to be related are aromatic polycyclic hydrocarbons and heterocyclic molecules that can form non-covalent interactions with proteins or DNA bonds that have the potential to be genotoxic 26 . PreADMET provides 62.5% accuracy and 52.2% sensitivity for carcinogenicity tests 27 . PreADMET predicts toxicity based on AMES mutagenicity parameters, the actual predictive value is "positive" or "negative" 21 .
Mutagenicity testing used the PreADMET application using the AMES method and the results obtained were 21 compounds that were positive for mutagens, and for carcinogenicity (Carcino Rat) there were 9 compounds that were negative for carcinogens. The compounds tested using the T.E.S.T application contained 9 positive mutagens and 7 N/A compounds (Not Applicated) and 41 other compounds were negative mutagenic, in this application there were N/A results because using the FDA method: Predictions for each chemical test were made using a new model corresponding to the chemical most similar to the test compound. Each model is generated at runtime which is the reason these 7 compounds cannot be translated into this application.
The toxicity test carried out with 5 different applications for toxicity parameters, it can be concluded that there are 4 compounds: N-glutaryl Monascurobamine, Monankarin F, Monaphilones A and Monaphilones B, that meet the criteria issued by ICH (International Conference on Harmonization) S9: non-clinical evaluation for pharmaceutical anticancer (2010) because they have a low level of toxicity from the 57 compounds tested. Whereas for testing each application, the best compound results for anticancer drug candidates according to ICH is as follows: for the application of pkCSM, the best compounds that have the potential as anticancer drug candidates are 39 test compounds, for the Protox-II application there is 1 safe compound; Xantomiscin B because it is included in class VI for LD50 and negative for carcinogens and mutagens, the Toxtree application produces 1 safe compound; N-glucosyl rubropuctamine because according to Creamer rules it is categorized as class 1 which means has a low level of toxicity, for the application of PreADMET 4 safe compounds are produced, namely Monaspyridine A, Monaspyridine B, Monaspyridine C, and Monaspyridine D, while for the T.E.S.T application produced 1 safe compound, namely Monaphilol A with negative mutagen and LD50 toxicity category including in class V.

CONCLUSION
Research on 57 pigment derivated compounds of Monascus sp. mold, the results were obtained for the application of pkCSM, the best compound that has the potential as an anticancer drug candidate are 39 test compounds, for the Protox-II application there is 1 safe compound, namely Xantomiscin B, the Toxtree application produces 1 safe compound, namely N-glucosyl rubropuctamine, for PreADMET applications 4 Safe compounds which are Monaspyridine A, Monaspyridine B, Monaspyridine C, and Monaspyridine D are produced, while for the T.E.S.T application, 1 safe compound is produced, namely Monaphilol A because it fulfills one of the aspects of ICH (International Conference on Harmonization) S9: non-clinical evaluation for anticancer pharmaceutical 2010 and has potential as an anticancer drug candidate.