Aberrant alternative RNA splicing regulation is a major cause of genetic diseases, as a recent estimate points out involvement of ~35% of reported mutations in induction of mis-splicing events. Although currently available information for pathogenic mutations are mostly limited to exonic sequences and their vicinity sites owing to a technical limitation of exome sequencing, recent advances in whole genome sequencing shed a light to mutations within deep-intronic sites. Deepintronic mutations may create “pseudo exon”, in which a particular intronic sequence is recognized as an exon due to a cryptic splice site or splice enhancer introduced as a consequence of deepintronic mutations. Inclusion of a pseudo exon induces frameshift and creation of premature stop codon to inhibit expression of a specific gene, leading to pathogenesis. While such pseudo exonic mutations are speculated as major cases for undiagnostic patients of genetic diseases, so far there is no comprehensive investigation for disease-related deep-intronic mutations, and thus their diagnostic applications remain to be achieved. Moreover, pathogenesis-related alternative RNA splicing recently regarded as an attractive therapeutic target, especially after approval of Nusinersen, and characterization of pseudo exonic mutations may also provide a novel therapeutic opportunity. A large scale resources for whole genome sequences, such as Tohoku Medical Megabank Organization (ToMMo), The 1,000 Genome Project, and The GenomeAsia 100K Project, have been developed recently, and current advances in artificial intelligence (AI) technology made it feasible to conduct functional computation for numerous datasets of whole genome sequencing. In this proposal, by taking advantage of whole genome sequence resources and application of AI, we are developing data-driven high-throughput assay system for a comprehensive characterization of deep-intronic mutations, in which AI-based survey for pseudo exonic deep-intronic mutations from ~7,500 whole genome datasets are connected with wet validation in a corresponding donor cells obtained through ToMMo or genome editing. Identified mutations are deposited in a database, and further assessed for diagnostic use by a pseudo exonic mutation array, as well as drug target by a splice-targeting small molecule compounds or antisense nucleotide drugs.

We have been analyzing Percellome data [1], which is the database of gene expression profiles of mice treated with various chemical compounds [1], to infer the mechanism of their toxicity. In this presentation, we will show the effects of estragole (a compound that belongs to phenylpropene suspected to be carcinogenicity in rodents) as a PPARα ligand based on the gene expression profiling, and the differences between the gene expression profiles of estragole and those of other PPARα ligands that have been analyzed in our previous study will be discussed.

GeneChip data of mouse liver after a single oral gavage at 2, 4, 8, and 24 hours after dosing for 4 dose levels, triplicate, total of 48 samples, were compiled on 160 chemicals. And, now we added a new repeated dosing study data from a few chemicals by 14-day repeated dosing to all 48 mice followed by the single dose study protocol mentioned above. From the same sample, whole genome bisulfite analysis (WGBA) and ChIP-seq data against H3K4me3, H3K27me3, H3K27Ac, and H3K9me3 were obtained and analyzed for the effect of repeated dosing on DNA methylation and Histone modification. (Supported by Health and Labour Sciences Research Grant).

Since 2010, the number of successful cases in clinical trials and approved products in gene therapy has increased. Gene therapy can be broadly divided into “in vivo gene therapy”, “in vivo virus therapy” and “ex vivo gene therapy”. As of 2020, as in vivo gene therapy, Glybera, Luxturna and Zolgensma using AAV vector, as in vivo virus therapy, Imlygic using oncolytic virus and as ex vivo gene therapy, Kymriah, Yescarta and Tecartus using CAR-T cell therapy have been approved as gene therapy products. With reference to NDA/BLA review reports, the development status is briefly summarized.

Basic requirements of nonclinical safety evaluation for in vivo gene therapy products are shown in “Ensuring the Quality and Safety of Gene Therapy Products” (PSEHB/MDED Notification No.0709-2). It may, however, not always be appropriate to apply this guideline in a uniform way because types, characteristics, and clinical applications of gene therapy products are diverse, and rapid strides are being made in the scientific advancement and accumulation of experience in this area. Therefore, nonclinical safety evaluations of individual products should be done flexibly on a case-by-case basis.

Approximately 50 clinical trials of genome editing therapies have already been conducted overseas. In addition to issues, such as gene transfer efficiency and immune response to vectors and therapeutic genes, for conventional gene therapy, genome editing therapy has unique challenges including off-target mutations caused by and immune responses against artificial nucleases. In particular, in vivo off-target mutations are difficult to evaluate in preclinical studies due to differences in nucleotide sequences between humans and model animals. In order to establish genome editing as a treatment for a wider range of diseases, it is essential to develop technologies to solve these issues.

Biodistribution (BD) of Gene Therapy (GT) Products following on in vivo administration is an important element of the nonclinical development programme for investigational GT Products. In Japan, “Ensuring the Quality and Safety of Gene Therapy Products” (PSEHB/MDED Notification No.0709-2), mentioning BD, was published. In 2018, International Pharmaceutical Regulators Programme (IPRP) Gene Therapy Working Group published a Reflection Paper titled ”Expectations for Biodistribution (BD) Assessments for Gene Therapy (GT) Products”. Based on these background, ICH S12 Expert Working Group was established in Nov 2019. In this presentation, overview of ICH S12 draft guideline will be provided.

For the medical treatment of bacterial infections, antimicrobial drugs play important roles to cure from the diseases or prevent the infections but there are some antibiotics with safety concerns for women or children. The medical package inserts in the antibiotics are described not to recommend to the use of pregnant women or women childbearing potential, or children based on the results of preclinical toxicity studies. I would like to present developmental or juvenile toxicities for the antibiotics with the safety concerns using the medical package insert, interview forms or references.

Viral infections are a common topic in our daily lives. Since clinical trials of antiviral drugs cannot be conducted in pregnant women, most drug package inserts state risks to pregnant women based on results of nonclinical reproductive and developmental toxicity studies. The teratogenic potential of Avigan was reported in the general media due to people's interest in COVID-19 therapeutic medicines. In this session, I would like to introduce the contents of the drug package insert and the reproductive toxicity of antiviral drugs, based on publicly available information.

It is well known that vaccines play an important role in improving the health of mothers and children. Vaccination before pregnancy can prevent diseases that cause birth defects. While vaccination before or during pregnancy is increasing, some people hesitate to get a vaccine. Concern about safety is the most common reason why women do not vaccinate before or during pregnancy. Reproductive and developmental toxicity studies are conducted as a non-clinical evaluation of vaccine. In this presentation, I would like to introduce the new guidelines for vaccines and the results of approved vaccines.

Anti-infective agents are commonly used during pregnancy because treatment of infections is critical to the health of a mother and her fetus. There have been no large-scale studies addressing safety of many classes of anti-infectives and vaccines during pregnancy. Among the antibiotics, penicillins, cephalosporins and macrolides have been used for a long time without much concern about their effects on the fetus. Vaccines are used during pregnancy to prevent maternal infection and to provide protection for the newborn through passive immunization.

Development of alternatives to animal experiments, especially those using in silico methods such as machine learning, is highly demanded. But, since the number of in vivo toxicity data is limited, it is difficult to develop such methods. Then, what can we do with such small data? We have been working on the characterization of physicochemical properties of anemia-inducing and hepatotoxicity-inducing compounds for developing a toxicity prediction model based on physicochemical property with molecular descriptors. In this session, I will introduce our recent results on the topics.

With the remarkable progress in computing resources and machine learning methodologies in recent years, expectations for the development of computational toxicology tools have increased further. The expected contributions of these tools to drug safety research include prediction of toxicity (on-target toxicity/off-target toxicity), elucidation of toxicity mechanisms, and improvement of the accuracy of toxicity assessment.

In this presentation, I would like to introduce the Open-Access Computational Toxicology Tools that we have established and discuss the issues that have been identified through these researches.

Acknowledgment

These researches are supported by AMED under Grant Number 19nk0101103h0005 and 21nk0101111h0102.

In this talk, I will present the first attempt to apply Explainable AI techniques to visualize what AI learns from 2D chemical structure images, which demonstrated that AI sees chemically meaningful substructures in the image to make a prediction as chemists do. In addition, I will talk about recent progress and future prospect of Federated Learning (FL). Astellas is participating in the MELLODDY consortium, aiming to develop privacy-preserved FL platform to leverage each company's data for building predictive models for drug discovery while owners keep control of their proprietary data.

Drugs are small compounds that target specific proteins and induce a perturbation of the entire intracellular biomolecular pathway. While drug administration has the expected effect, it can also cause toxicity and side effects. The molecular mechanism of toxicity and side effects remains largely unidentified. In recent years, a large amount of various information on drugs and compounds has been accumulated in publicly available databases. Such information includes various data such as chemical structure, target protein, drug efficacy, toxicity, side effects, etc., and it is expected that the relationship between them will be clarified by a computer approach. We propose a computational method for associating information on different scales, that is, information on the molecular level of compound-target protein interaction and information on the phenotypic level of compound toxicity / side effect information. We analyzed the correlation between target protein interaction data of compounds and toxicity / side effect data of compounds using machine learning models such as sparse canonical correlation analysis and sparse logistic regression. As a result of protein enrichment analysis of correlation sets using molecular pathway information, proteins working in similar molecular pathways were clustered in the same correlation set even if their molecular functions were different. That is, it can be interpreted that the functional control of proteins clustered in a certain correlation set by a compound may cause similar side effects through the activation and inactivation of the same molecular pathway. The proposed method is expected to be useful for predicting the potential toxicity / side effects of a compound from the target protein profile of the compound, as well as enabling estimation of the molecular mechanism of action of toxicity / side effects.

“Read-across” (RAx), a method of estimating the toxicity based on the structural similarity, is attracting attention as a technology of toxicity prediction that utilizes existing data on safety. In order to refine the accuracy of RAx, it is necessary to accurately reflect the rules between structure and toxicity by considering Adverse Outcome Pathways. We are working on standardization of the RAx strategy by accumulating case studies, and the precise prediction is expected by the Omics analysis, bioassay for key events and pharmacokinetics data. In the lecture, I’d like to introduce the strategy in our company regarding the non-animal testing to realize precise quantitative risk assessment.

Prediction of toxicity in humans is essential in pharmaceutical research and development, and various nonclinical toxicity studies are conducted using cells and animals. Even after various toxicity studies are conducted, some clinical studies are often discontinued due to unexpected toxicity. Especially, nephrotoxicity is difficult to predict due to structural complexity of the kidneys and interspecies differences between humans and experimental animals. Here we report a novel prediction system of nephrotoxicity using gene expression network in human ES/iPS cells and machine learning.

We reported a high accuracy toxicity prediction system using human ES cells in 2016. As next challenges, we report the results of an attempt to apply it to the adult toxins such as nephrotoxins, neurotoxins, and hepatotoxins. Toxicity prediction was performed using gene expression data of about 20 chemicals in human ES cells, and it was found that, for example, hepatotoxins can be predicted with a high accuracy of 90.5%. This prediction system, which uses human ES cells as a reaction device for chemicals, can apply mixtures that have been difficult to predict, and can be used as an unknown primary toxicity screening. As the number of chemicals increases in the future, it is expected that the effects of machine learning will be progressed and the prediction system will become more robust.

Prediction of toxicity of chemical substances in human with alternative testing methods is essential in cosmetic industries. The combined approach of the transcriptomic database in undifferentiated cells and machine learning technology are expected to be a useful tool that can comprehensively evaluate the reactivity of chemical substances in human. (Yamane et al. Nucleic Acids Res. 2016) However, the direct use of human ES cells to assess the safety of chemicals in humans also presents ethical challenges. We selected the representable human iPS cells among 28 cell lines that show a highly similar response to ES cells in databases. Here we report a novel prediction system of toxicities (eg. liver etc) using gene expression network in human iPS cells and machine learning.

We have developed a system that realizes toxicity prediction with a high-precision prediction rate of 90-95% or more by exposing undifferentiated human ES cells to toxic substances regardless of development or adulthood. By expanding the range of applicable target organs and creating a database of nerve, liver, kidney, heart, and carcinogenic (arbitrary organ) toxicity, hazard prediction became possible. Since this system uses undifferentiated cells, it is an epoch-making system that can theoretically predict the toxicity of multiple target organs with one test. For detailed AOP analysis, the conventional system using differentiated cells is effective, when the target is unknown, it is necessary to use as many differentiated cells as the number of organs, leaving problems in standardization and long-term development. This system can be facilitated as a primary screening to find a target, and is expected to be very effective because it can be executed at low cost and in a short time. Therefore, the development of a database is important, and we selected and listed 1,488 chemical substances to register at five levels according to the importance.

The preclinical studies for clarifying a cardiovascular risk of new chemical entity are critical in drug development because cardiovascular effects could be associated with a life-threatening adverse event. To evaluate a cardiovascular risk before FIH, in vivo telemetry study using conscious and non-restricted animals is generally conducted as the safety pharmacology study. Several cardiovascular parameters (e.g. blood pressure, heart rate, ECG) are assessed in the study, and QT prolongation in ECG is regarded as the primary surrogate biomarker to investigate the potential for life-threatening arrhythmia (Torsade de pointes). Concerning the progress of preclinical QT evaluation, a most effective probabilistic rate-correction method (Holzgrefe et al., 2007) was widely adopted during the past decade. Additionally, a translational discussion of drug-induced QT effects from animals to humans is accelerated based on the clinical IQ-CSRC study to characterize the effects of QT reference drugs on the concentration-QT relationship by exposure-response (ER) analysis (Darpo et al., 2014). We would like to introduce a J-ICET (Japan activity for Improvement of Cardiovascular Evaluation by Telemetry system) consortium activity and its results of preclinical QT evaluation by ER analysis in cynomolgus monkeys using QT reference drugs in crinical study (Darpo et al., 2015), and have a discussion about the latest topics (e.g. T wave analysis, automated blood sampling device) and future perspectives of preclinical in vivo studies for evaluating proarrhythmic potential.

Drug-induced cardiotoxicity is still a major concern endangering patient health during clinical trials for new chemical entities. Accurate and effective assessment leading to selection of promising candidates without cardiotoxic potential is one of the most important strategies at the preclinical phase of drug development. In safety pharmacology testing, cardiovascular safety profiling is carried out using non-rodent species. In recent years, preclinical in vitro, in vivo, or in silico cardiac electrophysiological safety assessments, typified by predicting drug-induced long QT syndrome, have been enhanced. An additional consideration to drug effects on heart involves cardiac contractility. However, other cardiovascular adverse effects including congestive heart failure are often difficult to detect during the preclinical stage, and the methods for assessment of the effects of the candidates on cardiac function, like their cardiac inotropic and lusitropic potential, still have room for development. The evaluation for cardiac contractility still depends on load-dependent indexes. The LVdP/dt_max is suitable only when loading conditions are unchanged. The QA interval is affected by following factors: electro-pressor latency (e.g. the time from the opening sodium channels to binding of heavy meromyosin head to actin), contraction time during isovolumic systole, pulse wave propagation (distance from LV to transducer, or stiffness of the arterial tree). Many factors other than cardiac contractility are involved, and interpretation being examined only by these parameters remains elusive. The establishment of inotropic/lusitropic properties remains elusive under clinical/screening paradigms because of load-dependency of indices of contractility/lusitropy (e.g., dP/dt_max), and the end-systolic pressure-volume (PV) relationship is known as a gold standard for evaluating the inotropic potential. Recent improvement of PV catheter technology has made possible the accurate measurement of the left ventricular PV by the admittance method. This method incorporates elements of the conductance of blood flow and cardiac muscle combined with the capacitance of the myocardium. The measurement of PV using this method has been conducted mainly in rodents, but not in non-rodents. Preclinical studies in monkeys still occupy an important position in the toxicological field. In addition, the evaluation of cardiovascular safety profile for promising drugs is carried out using monkeys as one of the non-rodent species in safety pharmacology testing. Therefore, there is a need to increase the knowledge and awareness of available and reliable non-rodent models by adopting the PV loop method in order to identify the potential cardiac functional toxicity of new chemical entities. In addition to update cardiac electrophysiological assessments in nonclinical studies, I would like to present new approaches to hemodynamic assessment such as cardiac contractility.

Progress in anti-cancer drugs, ex. HER2 inhibitor, PD-1/PD-L1 inhibitor, and tyrosine kinase inhibitor targeting cancer driver gene mutation have improved the prognosis of cancer patients in a palliative situation with advanced disease to the post-operative situation with localized disease and also in elderly to the pediatric.

The goal of cancer treatment includes the cure, prolongation of survival time, tumor volume reduction, and last but not least improvement or maintenance of QOL, but some anticancer drugs, including the above mentioned, are known to have some cardiac toxicities, including heart failure, myocarditis, or QT prolongation, these events could deteriorate a patient's QOL especially for patients with the early-stage disease or pediatric patients including AYA generation in a long-term perspective.

To mitigate the risk of drug-induced cardiac toxicities that could be fatal, evaluating potential risk factors to determine the risk-benefit balance in each patient is crucial, alongside other pharmacovigilance activities.

In this presentation, I would introduce our clinical point of view in assessing the efficacy and safety of anti-cancer drugs with cardiac toxicities.

The relationship between QT interval prolongation and human proarrhythmic risk in non-clinical studies has been investigated in the drug development. After ICH-S7B guideline was issued, new in vitro test systems have been developed, and test results are being used as reference data for human risk assessment. In this presentation, we will discuss safety considerations in FIH for drugs with QT interval prolongation risk in non-clinical studies, and the reviewers' viewpoints on the human safety considerations in NDA. In addition, the reviewers' viewpoints on human risk assessment will be also reported when the results of new in vitro results are used in NDA, by using a case of a drug approved in Japan.

The 4-HPPD inhibiting herbicides block tyrosine catabolism and induce hyper tyrosinemia for animals. In order to investigate the species difference of 4-HPPDi-induced toxicity, NTBC was administered for 52 weeks to rats and mice via foods. Mean plasma tyrosine concentrations exceeded over 1000 μmol/L in rats but less than in mice. Histopathologically, keratitis in the eyes and vacuolation of molecular layer in the cerebellum was induced in rats but not in mice. There was a clear species difference in the ocular and brain effects associated with a threshold in plasma tyrosine level.

Human carcinogenicity of chemicals is assessed mainly from the results of carcinogenicity studies using rodents. However, it has been known that there are many cases that the positive results of rat carcinogenicity studies are not relevant to human. Thus, in human risk assessments of chemicals, it is extremely important that human relevancy is evaluated by clarifying mode of action (MOA) of carcinogenicity and its species differences.

We evaluated human relevancy of rat hepatocellular tumors caused by synthetic pyrethroids (metofluthrin, momfluorothrin) in in vitro hepatocyte culture system, the nuclear receptor constitutive androstane receptor (CAR) knockout rats, and chimeric mice with human hepatocytes (PXB mouse). Using phenobarbital (PB) as a positive control, gene expressions, hepatocellular enzyme activities, and histopathological changes including hepatocellular proliferation were investigated. Consequently, it was revealed that both synthetic pyrethroids induced hepatocellular proliferation and tumor formation in rats by CAR activation in the same manner as PB, while no cell proliferation was induced in human hepatocytes in PXB mouse. Based on these results, we concluded that carcinogenic MOA of both pesticides is not relevant to human, and the European authority accepted our conclusion. Especially, PXB mouse is a useful model to evaluate effects on human hepatocytes.

Butachlor and Alachlor, chloroacetanilide herbicides, induce glandular stomach tumors in rats. We re-evaluated the pathogenesis and human relevance of the tumors. From the re-evaluation of stomach sections stained with neuron-specific enolase and chromogranin A, all tumors were derived from ECL cells. The tumor formation involved a nongenotoxic threshold MOA initially characterized by mucosal atrophy, leading to hypochlorhydria and hypergastrinemia. The mucosal atrophy with hypergastrinemia stimulated the ECL cell proliferation, resulting in neoplasia. We concluded the tumors developed via a threshold-dependent nongenotoxic MOA, under conditions not relevant to humans.

We studied toxic relationship between aromatase inhibitory activity in a novel compound and its effect on reproductive ability in rats.

The compound was orally administered repeatedly to non-pregnant and pregnant rats for 14 or 16 days. Animals in high dose group showed their estrus cycle ceasing and decreased body weight gain. Plasma estradiol concentration decreased. Delayed delivery and increased neonatal mortality were observed.

These reproductive abnormality would depend on the decreased plasma estradiol concentration induced by the aromatase inhibitory activity in the compound.

There is a recent increase in autism spectrum disorder and other neurodevelopmental cases as can be seen in some government reports (https://www.mext.go.jp/content/20200317-mxt_tokubetu01-000005538-02.pdf). Such rapid increase should be attributed to environmental or epigenetic factors. This session welcomes basic, clinical and toxicological scientists to discuss the developmental mechanisms, clinical causal relationships, and toxicological mechanistic researches.

In recent years, the Developmental Origin of Health and Disease (DOHaD) theory, which claims that the causes of offspring’s health and disease can be traced back to the fetal period, has been gaining attention. DOHaD classically focuses on the uterine environment for the fetus, such as maternal nutritional intake and drug exposure. However, recent epidemiological studies have pointed out that paternal factors may also influence the next generation (Paternal Origin of Health and Disease, POHaD). For example, it has been repeatedly reported that there is an increase in low birth weight and neurodevelopmental disorders in children born to older fathers. In the United States, a longitudinal study (EARLI study) was conducted in an attempt to identify risk factors for autism spectrum disorder (ASD), a neurodevelopmental disorder (http://www.earlistudy.org/). The results of this study showed that paternal sperm DNA methylation changes were indeed correlated with ASD-like signs in the offspring, and that several of these DNA methylation changes were also common in autopsied brains of ASD patients. Based on this background, we have been studying the molecular mechanisms by which paternal aging is associated with the risk of neurodevelopmental disorders in children using aged male mice, and have focused on transgenic epigenomic mutations. In addition to paternal aging, it is becoming clear that the male germline epigenome, which can be altered by various factors such as endocrine disruptors, is a risk factor for disease in children. This also implies that offspring can have diverse phenotypes as a result of transgenerational effects of sperm-mediated epigenomics. The perspective of transgenerational effects of epigenomic variation may provide new insights into this issue.

The number of children diagnosed Autism Spectrum Disorder (ASD) has rapidly increasing globally. There are many possible explanations but it can’t explain it well. ASD occur because of both genetic and environmental factor. Since genetic factor wouldn’t change so quickly, environmental factor would be more important.

Today I want to talk about 1) the relationship of neonicotinoids and fetal growth and 2）how difficult I feel against ASD in practice as pediatrician.

At last, I want to introduce latest knowledge about the relationship of growth during infancy and ASD risk.

Two neurodevelopmental disorders, autism spectrum disorder (ASD) and attention deficit/hyperactivity disorder (ADHD), become apparent in infancy and have diverse prognoses.

Both genetic and environmental risk factors contribute to the emergence of ASD and ADHD. Such environmental risk factors include nutrient deficiencies, heavy metals, alcohol, diesel exhaust particles, residual organic pollutants, pesticides and agricultural chemicals in the womb and during neonatal period as well as exposure to digital devices during early infancy.

The author will discuss how these findings should be interpreted from the perspective of child psychiatry and epidemiology.

Although the risk assessments are necessary to prevent the health hazard induced by chemical exposure, several harmful effects on brain functions have been still represented. Recently, chemical exposure to mammals in the prenatal, early postnatal or juvenile period has been recognized as a potential inducer of the effects on the nervous systems in adulthood.

In this symposium, we will report the results of the emotional/cognitive behavior tests and neuroscientific analyses of adult male mice induced by low-level, chronic chemical exposure in early life, focusing on examples of neonicotinoids.

Pesticides for plant protection may affect human health and environment. To minimize the risks, science-based hazard identification, exposure assessment, and risk characterization from various aspects must be conducted. Environmental fate and behavior of pesticides are influenced by the climate and geography, and the types/levels of pesticides used and ecosystems are different depending on the area. Therefore, the methods of safety assessment of pesticides differ by country or region. We will discuss here on the latest situation of safety evaluation of pesticides in the world.

For assessing the risk from the occupational use of plant protection products, the European Union requires that exposure to workers applying products and those either inspecting or harvesting the crop post treatment is evaluated. The tier I risk assessment models used to assess occupational exposure in the EU and refinement using dermal absorption data will be discussed. If exposure still exceeds reference values further refinement is possible through higher tier field studies. Such refinement options will be illustrated with case studies of bespoke experiments designed to determine exposure during application, re-entry and harvest in different agronomic situations.

Part of sprayed pesticides flow into rivers or groundwater and are exposed to humans via drinking water. The exposure assessment should be considered environmental factors such as soil types and meteorological conditions. Drinking water exposure assessment in each country is based on a tier approach that takes into account the environmental factors of each country, and if any risks, exposure is refined in accordance with their individual schemes. The safety criteria is also different in each country. This presentation will explain the difference of pesticide safety assessment through drinking water in Japan, the United States, and Europe.

It is important to ensure the safety for wildlife in the use of agrochemicals. A huge number of toxicity tests with various kinds of species are necessary for environmental risk assessment (ERA). As a basic concept, the ERA based on the comparison between Predicted Environmental Concentration (PEC) and Predicted No Effect Concentration (PNEC) is commonly employed by many countries. In case of PEC/PNEC≧1, further studies and/or restriction of usage will be required. An overview of the ERA process in the world and examples of efforts to develop eco-friendly products will be presented.

For human health of pesticides, carcinogenicity, mutagenicity, and reproductive toxicity (CMR) are important toxicity endpoints. When certain tumors may increase, mechanistic studies for Adverse Outcome Pathway (AOP) using biochemical, molecular biological, and pathological methods can provide useful information for human-relevance. If in vitro genotoxicity is equivocal or positive, in vivo genotoxicity results come to be definitive considering evidence of exposure to the target tissue. Regarding to effects on reproductive performance or next generation, hormone measurements and searches for early key events can be indicators for weight of evidence.

In the recent trends in regulations concerning endocrine disrupting chemicals (EDCs), regulatory authorities in representative countries such as U.S., Europe and Japan has become strictly required various data for the purpose of pesticide registration. There is, however, still no internationally standardized procedure for identifying EDCs, although test methods such as OECD test guidelines have been developed to detect endocrine disrupting property using various cell lines and animals. In this presentation, I would like to introduce the schemes for establishing the evaluation methods of EDCs in each country and discuss the future issues.

With the onset and progression of chronic kidney disease (CKD), the frequency of onset of various diseases such as cardiovascular disease, osteoarthritis, infectious disease, and malignant tumor increases, which affects the prognosis and quality of life (QOL) of CKD patients. These phenomena are especially observed in dialysis patients, and uremic substances related to them have been identified and are being studied daily to lead to prevention or treatment. Among them, the pathophysiology of CKD-related diseases, especially dialysis amyloidosis, has been elucidated, and how much these uremic substances can be removed is considered to be a treatment method for improving the prognosis or QOL of CKD patients.

Among uremic substances, β₂-microglobulin (β₂-MG), which is a water-soluble medium-molecular-weight substance, is known to be a precursor protein for dialysis-related amyloidosis because its high blood concentration affects the prognosis of CKD patients. In addition, amyloid fibers composed of β₂-MG are deposited mainly in the bone joint tissue of long-term dialysis patients, causing carpal tunnel syndrome, destructive spondyloarthropathies, etc., and as they progress further, they become cardiovascular and other organs, deposits and causes dysfunction.

In dialysis therapy, treatment methods are evolving to improve the removal of β₂-MG, but residual syndrome with uremia symptoms remains a problem. It is still unclear how much of a substance with a large molecular weight that is difficult to remove by dialysis or a substance bound to albumin (Alb) should be removed. Among them, increasing the leakage of Alb does not affect the β₂-MG reduction rate, but the α₁-MG reduction rate increases correlatively. If the removal of α₁-MG increases, the removal of TNF-α, Il-6, Il-1β, etc. with a molecular weight of around 30,000 will also increase. Aggressive Alb removal may be required to remove cytokines such as TNF-α, Il-6, and Il-1β.

[Case report]

The reported cases the experience of renal toxicity due to drug experienced clinical.

Case 1. Diffuse tubular interstitial nephritis induced by cilostazol

Case 2. Alogliptin-induced microvariant nephrotic syndrome and tubular interstitial nephritis

Many drugs are known to cause drug-induced kidney injury (DKI). Understanding the toxicological mechanisms of DKI help to minimize the disadvantages by taking theoretical precautions. The desired procedure for preventing DKIs should be confirmed in clinical practice and an information revolution is also needed to link basic and clinical sciences. Reasonable precautions can be taken during treatment if there is "interest" in the safety of the drugs, however, keeping interest in only one particular area is difficult and insufficient. In this presentation, we will consider the problems related to DKIs in clinical practice.

Cisplatin (CDDP) is an anticancer drug that causes severe kidney damage, especially in the S3 region of the proximal tubule. We showed that cytotoxicity to CDDP is the highest in the S3-derived cells. The S3-specific toxicity could not be explained by accumulation of Pt. Since the highest amount of Pt uptake was found 15 minutes after the exposure, the cytotoxicity by 15 min-exposure to CDDP was examined. The result showed that even the 15-min exposure to CDDP caused the highest cytotoxicity in S3 cells, suggesting the importance of shout-term cellular responses. We are attempting to establish the detection system of of CDDP – glutathione conjugates and investigate their dynamics.

The kidney is exposed to significant drug concentrations via the systemic exposure because of high renal blood flow and high rate of compound clearance. Drug-induced nephrotoxicity is increasingly recognized as a potential issue hampering clinical drug development. The failure of in vitro screening methods to identify nephrotoxic compounds can be attributed to many factors; the major one being the lack of valid in vitro models that fully recapitulate kidney phenotype. The development of in vitro models to successfully predict drug-induced kidney toxicity requires solid understanding of the specific cellular targets, mode of action of nephrotoxicants and setting of reliable biomarkers of nephrotoxicity. The proximal tubule is of particular interest due to its active clearance, reabsorption and intracellular concentration. Many drugs associated with kidney proximal tubule toxicity are hydrophilic suggesting that active uptake via transporters drives the cell accumulation for such low permeable drugs. However, cells commonly used in in vitro toxicity studies lack transporter expression which can lead to false negative results. In this presentation our attempt for proximal tubule toxicity screening using transporter over expression cell line will be introduced and discussed. In addition, we will give an overview of recent approaches for development of novel in vitro screening models which could be utilized to create a more physiologically relevant platform with the potential to improve the prediction value for proximal tubule toxicity screening.

In research and development of pharmaceuticals, critical decision of the developmental strategy determined in each development stage is carried out by the efficacy / safety evaluation and its risk assessment / management. The basis of many of these factors for decision making is the elucidation of the toxic mechanism with mode of action of the drug, which depends on the biomarker setting. Fusion of two factors (Mode of action: mechanism as medicinal effect, Adverse observed pathway: mechanism as toxic action) is indispensable for elucidation of the mechanism of toxicological effects in human body

In terms of “Go”, “No-Go” decision during drug development, supportive utilization of a scientifically interpretable computational toxicology system is required for human safety evaluation and its risk evaluation / management. Pharmaceutical safety evaluator as a related toxicologist who is facing to practical decision does not need a data-driven AI (Artificial Intelligence) system that calls for the final consequence, rather requires an explainable AI that can provide comprehensive information necessary for evaluation and can help decision making. Through the explication and suggestion of information on the mechanism of toxic effects to safety assessment scientists, ultimately a subsidiary partnership system for risk assessment is to be a powerful tool that can indicate project-vector with data weight for the corresponding counterparts.

To bridge the gaps between the big-data and the knowledge, multi-dimensional thinking based on philosophical ontology theory is necessary to handle heterogeneous data such as interpretable computational toxicology related to drug safety assessment.

People have suffered from emerging diseases such as COVID-19. Viruses change the targets from molecules to organ in the human body. Moreover, excessive human defense functioning makes it more difficult to understand the infectious mechanisms. For tackling this issue, ontology thinking is effective that helps us to explicate the intrinsic natures of entities. This presentation will describe the systematization of COVID-19 infectious process knowledge based on the ontological approach. Moreover, some use cases of ontologies for risk management in new COVID-19 drug development will be discussed.

In recent years, data science has been attracting attention as a science and technology for discovering knowledge and creating new values from the explosive growth of big data in all fields. In the field of drug discovery and life sciences, the data explosion caused by the remarkable progress of high-throughput technology and omics measurement technology has made the research and development of big data science an urgent issue. In the face of such a wide variety and vast amount of data, artificial intelligence (AI) has been attracting worldwide attention as a technology for analyzing such big data. Big data, AI, and IoT are considered to be the core technologies of the fourth industrial revolution, and are about to bring social change to all fields and industries.

The speaker has been working on the application of artificial intelligence and machine learning technology to drug discovery for more than 10 years, and has developed technologies for screening active compounds and automated molecular design. In addition, in November of 2016, the Life Intelligence Consortium (LINC) was launched to promote AI development for the life science field through industry-academia collaboration, aiming to develop a series of AI for drug discovery processes.

However, several issues have become apparent through AI research and development to date. In particular, since AI is developed by learning from existing data, there is the issue of data volume, which greatly affects the performance of AI. Compared to overseas megapharmaceutical companies, Japanese pharmaceutical companies are small in scale and possess inferior amounts of data. Therefore, it is expected to develop high-performance drug discovery AI by sharing data from multiple companies that are in business competition.

In this presentation, I would like to introduce an AI technology for federated learning while maintaining confidentiality of data sharing among multiple institutions, and discuss data sharing among companies, using the development of AI for predicting drug efficacy and ADMET as an example.

We constructed the Ames mutagenicity prediction system YosAI which increases prediction accuracy as well as efficiency of mutagenicity evaluation by integrating the expert knowledge, AI technology based on internal/external Ames data and commercial prediction software, CASE Ultra.

This system can be used to (1) search for the mutagenicity and carcinogenicity risk in internal/external databases, (2) display the presence or absence of CASE Ultra structure alerts, (3) search for similar structures (fingerprint method with Tanimoto coefficient), and (4) to show the presence or absence of electrophilicity and DNA binding ability from the OECD tool box. Utilizing these in silico prediction and chemical considerations for the mechanisms of mutagenicity, we mimicked the expert reviews according to ICH M7 guideline. Furthermore, AI-based mutagenicity prediction is implemented by the Artificial Neural Network method that uses parameters such as alert, and DNA binding ability.

We hope that this prediction system can be utilized as a hybrid knowledge/statistical-SAR model for the spread of computational toxicology in pharmaceutical development.

Since there are still many compounds terminated due to safety concerns in preclinical and clinical studies, compound evaluation and design with safety perspective are strongly needed at the early-stage of drug discovery. We have developed in vitro tiered screening strategy and provided two types of computational support for the purpose. One is a prediction by QSAR models and the other is a structural modification based on matched molecular pair DB, which were built by the large amount of screening data. In this presentation, I would like to introduce the approaches.

Maintenance of appropriate levels of epigenomic modifications is critical for cellular function. Epigenomic alterations are involved in cancer, and are likely to be involved in allergy, autoimmune disorders, and type 2 diabetes. Nevertheless, as inducers of epigenomic alterations, aging and chronic inflammation are almost all the known factors, and little knowledge is available for the influence of chemicals. At the same time, many inhibitors of enzymes that regulate epigenomes have been developed, including those derived from natural compounds. These inhibitors can show teratogenicity, and induce toxicities in bone marrow, the heart, gastrointestinal tract, and testes. Toxicities due to natural and industrial compounds are expected to be much weaker than the inhibitors. However, cumulative effects are known for aberrant DNA methylation, and may be present for other epigenomic alterations. It is important to accelerate our research on how we detect epigenomic toxicity and with what sensitivity.