Raster plots machine learning to predict the seizure liability of drugs and to identify drugs.

In vitro microelectrode array (MEA) assessment using human induced pluripotent stem cell (iPSC)-derived neurons holds promise as a method of seizure and toxicity evaluation. However, there are still issues surrounding the analysis methods used to predict seizure and toxicity liability as well as drug mechanisms of action. In the present study, we developed an artificial intelligence (AI) capable of predicting the seizure liability of drugs and identifying drugs using deep learning based on raster plots of neural network activity. The seizure liability prediction AI had a prediction accuracy of 98.4% for the drugs used to train it, classifying them correctly based on their responses as either seizure-causing compounds or seizure-free compounds. The AI also made concentration-dependent judgments of the seizure liability of drugs that it was not trained on. In addition, the drug identification AI implemented using the leave-one-sample-out scheme could distinguish among 13 seizure-causing compounds as well as seizure-free compound responses, with a mean accuracy of 99.9 ± 0.1% for all drugs. These AI prediction models are able to identify seizure liability concentration-dependence, rank the level of seizure liability based on the seizure liability probability, and identify the mechanism of the action of compounds. This holds promise for the future of in vitro MEA assessment as a powerful, high-accuracy new seizure liability prediction method.

Electrophysiological responses to seizurogenic compounds dependent on E/I balance in human iPSC-derived cortical neural networks.

Construction of in vitro functional assay systems using human-induced pluripotent stem cells (iPSCs) as indicators for evaluating seizure liability of compounds has been anticipated. Imbalance of excitation/inhibition (E/I) inputs triggers seizure; however, the appropriate ratio of E/I neurons for evaluating seizure liability of compounds in a human iPSC-derived neural network is unknown. Here, five neural networks with varying E/I ratios (88/12, 84/16, 74/26, 58/42, and 48/52) were constructed by altering the ratios of glutamatergic (E) and GABA (I) neurons. The responsiveness of each network against six seizurogenic compounds and two GABA receptor agonists was then examined by using six representative parameters. The 52% GABA neuron network, which had the highest ratio of GABA neurons, showed the most marked response to seizurogenic compounds, however, it suggested the possibility of producing false positives. Moreover, analytical parameters were found to vary with E/I ratio and to differ for seizurogenic compounds with different mechanism of action (MoA) even at the same E/I ratio. Clustering analysis using six parameters showed the balance of 84/16, which is the closest to the biological balance, was the most suitable for detection of concentration-dependent change and classification of the MoA of seizurogenic compounds. These results suggest the importance of using a human-iPSC-derived neural network similar to the E/I balance of the living body in order to improve the prediction accuracy in the in vitro seizure liability assessment.

Analysis of signal components < 500 Hz in brain organoids coupled to microelectrode arrays: A reliable test-bed for preclinical seizure liability assessment of drugs and screening of antiepileptic drugs.

Brain organoids with three-dimensional structure and tissue-like function are highly demanded for brain disease research and drug evaluation. However, to our knowledge, methods for measuring and analyzing brain organoid function have not been developed yet. This study focused on the frequency components of an obtained waveform below 500 Hz using planner microelectrode array (MEA) and evaluated the response to the convulsants pentylenetetrazol (PTZ) and strychnine as well as the antiepileptic drugs (AEDs) perampanel and phenytoin. Sudden and persistent seizure-like firing was observed with PTZ administration, displaying a concentration-dependent periodic activity with the frequency component enhanced even in one oscillation characteristic. On the other hand, in the administration of AEDs, the frequency of oscillation decreased in a concentration-dependent manner and the intensity of the frequency component in one oscillation also decreased. Interestingly, at low doses of phenytoin, a group of synchronized bursts was formed, which was different from the response to the perampanel. Frequency components contained information on cerebral organoid function, and MEA was proven useful in predicting the seizure liability of drugs and evaluating the effect of AEDs with a different mechanism of action. In addition, frequency component analysis of brain organoids using MEA is an important analysis method to perform in vitro to in vivo extrapolation in the future, which will help explore the function of the organoid itself, study human brain developments, and treat various brain diseases.

Principal Component Analysis to Distinguish Seizure Liability of Drugs in Human iPS Cell-Derived Neurons.

Screening for drug discovery targeting the central nervous system requires the establishment of efficient and highly accurate toxicity test methods that can reduce costs and time while maintaining high throughput using the function of an in vitro neural network. In particular, an evaluation system using a human-derived neural network is desirable in terms of species difference. Despite the attention, the microelectrode array (MEA) is attracting among the evaluation systems that can measure in vitro neural activity, an effective analysis method for evaluation of toxicity and mechanism of action has not yet been established. Here we established analytical parameters and multivariate analysis method capable of detecting seizure liability of drugs using MEA measurement of human iPS cell-derived neurons. Using the spike time series data of all drugs, we established periodicity as a new analytical parameter. Periodicity has facilitated the detection of responses to seizurogenic drugs, previously difficult to detect with conventional analytical parameters. By constructing a multivariate analytical method that identifies a parameter set that achieves an arbitrary condition, we found that the parameter set comprising total spikes, maximum frequency (MF), inter- MF interval (IMFI), coefficient of variance of IMFI, and periodicity can uniformly detect the seizure liability of seizurogenic drugs with different mechanisms of action. Seizurogenic drugs were suggested to increase the regularity of the network burst in MEA measurements in human iPS cell-derived neurons.

Neuroprotective efficacy of thymoquinone against amyloid beta-induced neurotoxicity in human induced pluripotent stem cell-derived cholinergic neurons.

The natural antioxidant Thymoquinone (TQ) is the most abundant ingredient in the curative plant Nigella sativa seed's oil. An extensive number of studies have revealed that TQ is the most active and most responsible component for the plant's pharmacological properties. It has been documented in several studies that TQ has a wide range of protective activities and many neuropharmacological attributes. Amyloid beta (Aß) is the major role player peptide in the progression of Alzheimer's disease (AD). Our current study has been implemented to explore the protective possibilities of TQ on Aß1-42 -induced neurotoxicity. To test TQ's effect we used cultured human induced pluripotent stem cell (hiPSC)-derived cholinergic neurons. The obtained results showed that Aß1-42 caused cell death and apoptosis, which was efficiently attenuated by the co-treatment of TQ. Moreover, TQ restored the decrease in the intracellular antioxidant enzyme glutathione levels and inhibited the generation of reactive oxygen species induced by Aß1-42. Furthermore, using the fluorescent dye FM1-43 we demonstrated that TQ was able to reduce synaptic toxicity caused by Aß1-42. Thus, the findings of our study suggest that TQ holds a neuroprotective potential and could be a promising therapeutic agent to reduce the risk of developing AD and other disorders of the central nervous system.

Toxicological evaluation of convulsant and anticonvulsant drugs in human induced pluripotent stem cell-derived cortical neuronal networks using an MEA system.

Functional evaluation assays using human induced pluripotent stem cell (hiPSC)-derived neurons can predict the convulsion toxicity of new drugs and the neurological effects of antiepileptic drugs. However, differences in responsiveness depending on convulsant type and antiepileptic drugs, and an evaluation index capable of comparing in vitro responses with in vivo responses are not well known. We observed the difference in synchronized burst patterns in the epileptiform activities induced by pentylentetrazole (PTZ) and 4-aminopryridine (4-AP) with different action mechanisms using multi-electrode arrays (MEAs); we also observed that 100 µM of the antiepileptic drug phenytoin suppressed epileptiform activities induced by PTZ, but increased those induced by 4-AP. To compare in vitro results with in vivo convulsive responses, frequency analysis of below 250 Hz, excluding the spike component, was performed. The in vivo convulsive firing enhancement of the high γ wave and ß wave component were observed remarkably in in vitro hiPSC-derived neurons with astrocytes in co-culture. MEA measurement of hiPSC-derived neurons in co-culture with astrocytes and our analysis methods, including frequency analysis, appear effective for predicting convulsion toxicity, side effects, and their mechanism of action as well as the comparison of convulsions induced in vivo.

Detection of synchronized burst firing in cultured human induced pluripotent stem cell-derived neurons using a 4-step method.

Human induced pluripotent stem cell-derived neurons are promising for use in toxicity evaluations in nonclinical studies. The multi-electrode array (MEA) assay is used in such evaluation systems because it can measure the electrophysiological function of a neural network noninvasively and with high throughput. Synchronized burst firing (SBF) is the main analytic parameter of pharmacological effects in MEA data, but an accurate method for detecting SBFs has not been established. In this study, we present a 4-step method that accurately detects a target SBF confirmed by the researcher's interpretation of a raster plot. This method calculates one set parameter per step, in the following order: the inter-spike interval (ISI), the number of spikes in an SBF, the inter-SBF interval, and the number of spikes in an SBF again. We found that the 4-step method is advantageous over the conventional method because it determines the preferable duration of an SBF, accurately distinguishes continuous SBFs, detects weak SBFs, and avoids false detection of SBFs. We found also that pharmacological evaluations involving SBF analysis may differ depending on whether the 4-step or conventional threshold method is used. This 4-step method may contribute to improving the accuracy of drug toxicity and efficacy evaluations using human induced pluripotent stem cell-derived neurons.

Physiological maturation and drug responses of human induced pluripotent stem cell-derived cortical neuronal networks in long-term culture.

The functional network of human induced pluripotent stem cell (hiPSC)-derived neurons is a potentially powerful in vitro model for evaluating disease mechanisms and drug responses. However, the culture time required for the full functional maturation of individual neurons and networks is uncertain. We investigated the development of spontaneous electrophysiological activity and pharmacological responses for over 1 year in culture using multi-electrode arrays (MEAs). The complete maturation of spontaneous firing, evoked responses, and modulation of activity by glutamatergic and GABAergic receptor antagonists/agonists required 20-30 weeks. At this stage, neural networks also demonstrated epileptiform synchronized burst firing (SBF) in response to pro-convulsants and SBF suppression using clinical anti-epilepsy drugs. Our results reveal the feasibility of long-term MEA measurements from hiPSC-derived neuronal networks in vitro for mechanistic analyses and drug screening. However, developmental changes in electrophysiological and pharmacological properties indicate the necessity for the international standardization of culture and evaluation procedures.

Induction of long-term potentiation and depression phenomena in human induced pluripotent stem cell-derived cortical neurons.

Plasticity such as long-term potentiation (LTP) and long-term potentiation depression (LTD) in neuronal networks has been analyzed using in vitro and in vivo techniques in simple animals to understand learning, memory, and development in brain function. Human induced pluripotent stem cell (hiPSC)-derived neurons may be effectively used for understanding the plasticity mechanism in human neuronal networks, thereby elucidating disease mechanisms and drug discoveries. In this study, we attempted the induction of LTP and LTD phenomena in a cultured hiPSC-derived cerebral cortical neuronal network using multi-electrode array (MEA) systems. High-frequency stimulation (HFS) produced a potentiated and depressed transmission in a neuronal circuit for 1 h in the evoked responses by test stimulus. The cross-correlation of responses revealed that spike patterns with specific timing were generated during LTP induction and disappeared during LTD induction and that the hiPSC-derived cortical neuronal network has the potential to repeatedly express the spike pattern with a precise timing change within 0.5 ms. We also detected the phenomenon for late-phase LTP (L-LTP) like plasticity and the effects for synchronized burst firing (SBF) in spontaneous firings by HFS. In conclusion, we detected the LTP and LTD phenomena in a hiPSC-derived neuronal network as the change of spike pattern. The studies of plasticity using hiPSC-derived neurons and a MEA system may be beneficial for clarifying the functions of human neuronal circuits and for applying to drug screening.

Long-term electrophysiological activity and pharmacological response of a human induced pluripotent stem cell-derived neuron and astrocyte co-culture.

Human induced pluripotent stem cell (hiPSC)-derived neurons may be effectively used for drug discovery and cell-based therapy. However, the immaturity of cultured human iPSC-derived neurons and the lack of established functional evaluation methods are problematic. We here used a multi-electrode array (MEA) system to investigate the effects of the co-culture of rat astrocytes with hiPSC-derived neurons on the long-term culture, spontaneous firing activity, and drug responsiveness effects. The co-culture facilitated the long-term culture of hiPSC-derived neurons for >3 months and long-term spontaneous firing activity was also observed. After >3 months of culture, we observed synchronous burst firing activity due to synapse transmission within neuronal networks. Compared with rat neurons, hiPSC-derived neurons required longer time to mature functionally. Furthermore, addition of the synapse antagonists bicuculline and 6-cyano-7-nitroquinoxaline-2,3-dione induced significant changes in the firing rate. In conclusion, we used a MEA system to demonstrate that the co-culture of hiPSC-derived neurons with rat astrocytes is an effective method for studying the function of human neuronal cells, which could be used for drug screening.

Thymoquinone protects cultured hippocampal and human induced pluripotent stem cells-derived neurons against α-synuclein-induced synapse damage.

The seeds of Nigella sativa are used worldwide to treat various diseases and ailments. Thymoquinone (TQ) that is present in the essential oil of these seeds mediates most of the plant's diverse therapeutic effects. The present study aimed to determine whether TQ protects against α-synuclein (αSN)-induced synaptic toxicity in rat hippocampal and human induced pluripotent stem cell (hiPSC)-derived neurons. Here, we report that αSN decreased the level of synaptophysin, a protein used as an indicator of synaptic density, in cultured hippocampal and hiPSC-derived neurons. However, simultaneous treatment with αSN and TQ protected neurons against αSN-induced synapse damage, as revealed by immunostaining. Moreover, administration of TQ efficiently induced protection in these cells against αSN-induced inhibition of synaptic vesicle recycling in hippocampal and hiPSC-derived neurons as well as against mutated P123H ß-synuclein (ßSN) in hippocampal neurons, as revealed by experiments using the fluorescent dye FM1-43. Using a multielectrode array, we further demonstrated that the treatment of hiPSC-derived neurons with αSN induced a reduction in spontaneous firing activity, and cotreatment with αSN and TQ partially reversed this loss. These results suggest that TQ protects cultured rat primary hippocampal and hiPSC-derived neurons against αSN-induced synaptic toxicity and could be a promising therapeutic agent for patients with Parkinson's disease and dementia with Lewy bodies.

Effects of the antiplatelet agent TA-993 and its metabolite MB3 on the hemorheological properties of rat and human erythrocytes.

In the present study, we investigated the effects of the antiplatelet agent TA-993 and its metabolite MB3 on the hemorheological properties of rat and human erythrocytes in comparison with ticlopidine and aspirin. TA-993 and MB3 concentration-dependently lowered the viscosity of rat erythrocyte suspensions. TA-993 and MB3 inhibited both the hypotonic hemolysis of human erythrocytes and the mechanical hemolysis of rat erythrocytes induced by turbulent flow. Treatment of rats with TA-993 (10 mg/kg/day po) for 10 days significantly increased blood filterability, but ticlopidine and aspirin did not show this effect. TA-993 and MB3 enhanced the interaction of 1-anilino-8-naphthalene sulfonate (ANS), a hydrophobic probe, with human erythrocyte ghosts and reduced the fluorescence polarization in 1,6-diphenyl 1,3,5-hexatriene (DPH, a fluidity probe)-labeled human erythrocyte ghosts. TA-993 and MB3 induced aggregation of liposome suspensions prepared from acidic phospholipids. These findings suggest that TA-993 and MB3 may affect the erythrocyte membrane by interacting with acidic phospholipids and thus improve the hemorheological properties.

Inhibitory effects of TA-993 and its metabolite MB3 on platelet activation induced by collagen and U-46619 in human platelets.

In the present study, we investigated the effects of TA-993 and its metabolite MB3 on platelet activation in vitro. TA-993 and MB3 concentration-dependently inhibited platelet aggregation and ATP release induced by collagen in human platelets. Thromboxane (Tx) A2 formation, as determined by the production of TxB2, and the increase in intracellular Ca2+ concentration ([Ca2+]i) were also suppressed by TA-993 and MB3. TA-993 and MB3 did not inhibit TxA2 formation caused by arachidonic acid. These results suggest that the inhibition of platelet activation by TA-993 and MB3 is partly mediated by an inhibition of TxA2 formation at a step prior to cyclooxygenase. Furthermore, TA-993 and MB3 inhibited U-46619-induced platelet aggregation without blockade of the increase in [Ca2+]i, suggesting that they are likely to exert some additional effects on the intracellular events induced by Ca2+.

Antiplatelet mechanisms of TA-993 and its metabolite MB3 in ADP-induced platelet aggregation.

We investigated the antiplatelet mechanisms of TA-993 [(-)-cis-3-acetoxy-5-(2-(dimethylamino)ethyl)-2, 3-dihydro-8-methyl-2-(4-methylphenyl)-1,5-benzothiazepin-4(5H)-one maleate] and its metabolite MB3 (deacetyl and N-monomethyl TA-993) in human platelets stimulated by ADP in vitro. TA-993 and MB3 concentration-dependently inhibited fibrinogen binding to the ADP-stimulated platelets as well as inhibiting platelet aggregation. The antiplatelet effect of MB3 was about 300 times more potent than those of TA-993 and a glycoprotein IIb/IIIa receptor antagonist, Arg-Gly-Asp-Ser (RGDS). Aggregation of ADP-treated fixed platelets caused by the addition of fibrinogen was inhibited by RGDS but not by TA-993 and MB3. TA-993 and MB3 inhibited ADP-induced polymerization of actin filaments. Neither TA-993 nor MB3 affected cyclic AMP and cyclic GMP levels in resting platelets, and nor suppressed the increase in intracellular Ca(2+) concentration induced by ADP. These results suggest that the antiplatelet mechanisms of TA-993 and MB3 may involve inactivation of glycoprotein IIb/IIIa receptors via inhibition of the polymerization of actin.

Fluvastatin, an HMG-CoA reductase inhibitor, protects LDL from oxidative modification in hypercholesterolemic rabbits.

The antioxidative effect of fluvastatin sodium (fluvastatin) on low-density lipoprotein (LDL) was evaluated in vivo and in vitro. Since ex vivo measurement of the LDL oxidizability is reported to reflect the response of the atherosclerotic process, LDL isolated from rabbits fed a high cholesterol diet for 4 weeks with or without fluvastatin, pravastatin or alpha-tocopherol administration was oxidized by copper ions to estimate conjugated diene formation. Fluvastatin but not pravastatin significantly prolonged the lag time of LDL oxidized by copper ions ex vivo without affecting plasma cholesterol levels at a dose of 3 mg/kg after four weeks of treatment. Alpha-tocopherol-treated rabbits showed dramatically elongated LDL oxidation lag time at a dose of 150 mg/kg. In order to assess the mechanism, the content of alpha-tocopherol, a major endogenous antioxidant in LDL was measured, and we found that only LDL isolated from alpha-tocopherol-treated rabbits contained a significantly larger amount of alpha-tocopherol than that from high cholesterol control rabbits. To elucidate the mechanism further, the effect of fluvastatin on conjugated diene formation during copper-induced LDL oxidation in vitro was studied. Fluvastatin not only prolonged lag time, but also suppressed the rate of LDL oxidation, both in a dose dependent manner above 1 microM, while pravastatin showed no effect. These results suggest the direct antioxidative effect of fluvastatin on LDL oxidation in vivo. Since oxidation of LDL is an important step in the initiation and progression of atherosclerosis, fluvastatin may reduce the risk of this condition not only by lowering plasma cholesterol but also by protecting LDL from oxidation.

In vitro inhibitory effects of the optical isomers and metabolites of fluvastatin on copper ion-induced LDL oxidation.

Fluvastatin is a synthetic hypolipidemic drug which inhibits 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase. We compared in vitro the antioxidative effects of two enantiomers (3R, 5S and 3S, 5R) of fluvastatin, which is clinically used as a racemic mixture, on copper ion-induced oxidation of human low-density lipoprotein (LDL). Although 3R,5S-enantiomer of fluvastatin has 30-fold stronger inhibitory activity on HMG-CoA reductase than its optical counterpart, the antioxidative effects of these enantiomers on copper ion-induced LDL oxidation were similar. The antioxidative effects of the metabolites of fluvastatin (M2, M3, M4 and M7) on the copper ion-induced LDL oxidation were also investigated. All the metabolites tested showed an inhibitory effect on this system. Among them, the effects of M2 and M3, which have a phenolic hydroxyl group in each indole moiety, were strong and their potencies were 30-50 times greater than that of fluvastatin. We conclude that not only 3R,5S-enantiomer of fluvastatin but also its optical counterpart and the metabolites also have a potential to show the anti-atherosclerotic effect through their antioxidative activities on lipid peroxidation.

Imidapril, an angiotensin-converting enzyme inhibitor, inhibits thrombosis via reduction in aortic plasminogen activator inhibitor type-1 levels in spontaneously hypertensive rats.

It is known that angiotensin II (Ang II) exerts an antifibrinolytic effect by stimulating synthesis of plasminogen activator inhibitor type-1 (PAI-1), a specific inhibitor of tissue plasminogen activator (t-PA). The aim of this study was to compare the antithrombotic potency of imidapril, an angiotensin-converting enzyme (ACE) inhibitor, and candesartan, an angiotensin II type 1 (AT1) receptor antagonist, in a model of arterial thrombosis in spontaneously hypertensive rats (SHRs). Oral treatment with 5 mg/kg imidapril 1 h before induction of thrombosis resulted in a significant reduction in thrombus weight, whereas candesartan did not affect thrombus weight under the same treatment conditions. Candesartan lowered blood pressure to the same degree as in the imidapril-treated rats. Imidapril not only reduce the serum and aortic ACE activities, but also reduced aortic PAI-1 protein levels, while candesartan had no effect on theses. These results suggest that imidapril, but not the AT1 receptor antagonist, candesartan, enhances fibrinolysis via a reduction of aortic PAI-1 levels by inhibiting ACE and prevents thrombus formation in SHRs.

An in vitro study of the hydroxyl radical scavenging property of fluvastatin, and HMG-CoA reductase inhibitor.

We investigated the in vitro hydroxyl radical scavenging activity of fluvastatin, a 3-hydroxy-3-methylglutaryl (HMG)-CoA reductase inhibitor. Fluvastatin showed hydroxyl radical scavenging activity as potent as that of dimethylthiourea and alpha-tocopherol, which are well-known respectively, as a hydroxyl radical scavenger and a natural antioxidant. Since this effect was not observed in other HMG-CoA reductase inhibitors, such as pravastatin and simvastatin, the scavenging effect of fluvastatin on hydroxyl radicals would not be a common property of HMG-CoA reductase inhibitors, but is derived from the unique chemical structure of fluvastatin. The hydroxyl radical scavenging activities of human metabolites of fluvastatin were also determined. All the tested metabolites possessing the fluorophenyl indole moiety showed activity. Among them, the metabolites which possess a phenolic hydroxyl group on the indole moiety showed stronger effects than that of fluvastatin. We suggest that the fluorophenyl indole moiety of fluvastatin is important for manifestation of the activity and that the phenolic hydroxyl group enhances the potency.

Antithrombotic action of TA-993, a new 1,5-benzothiazepine derivative, in a canine model of femoral arterial thrombosis.

TA-993 is a novel 1,5-benzothiazepine derivative of l-cis configuration, having a potent antiplatelet action and an increasing action on femoral blood flow. We evaluated the antithrombotic effect of TA-993 in a canine model of femoral arterial thrombosis. Thrombus was induced by both application of direct anodal current to the femoral artery and partial occlusion of the artery. The partial occlusion by placing an adjustable occluder on the artery and the current application were carried out 40 and 60 min after the intraduodenal administration of drugs, respectively. In control dogs, complete sustained occlusion of the femoral artery due to thrombus occurred 55.4 +/- 9.2 min after the onset of current application. TA-993 (3 and 10 mg/kg) dose-dependently prolonged the time for occlusion. Aspirin (30 mg/kg) also prolonged it. TA-993, 10 mg/kg, significantly inhibited whole-blood aggregation 60 min after the administration with a weaker potency than that of aspirin (30 mg/kg), whereas 3 mg/kg of TA-993 did not. The inhibitory effect of TA-993 (10 mg/kg) on platelet aggregation was maintained for >7 h. Moreover, TA-993 (10 mg/kg) increased femoral blood flow in spite of the partially occluded condition. These results indicate that TA-993 has an antithrombotic effect on femoral arterial thrombosis and suggest that an increasing action on femoral blood flow of TA-993 is more relevant than its antiplatelet action to the antithrombotic effect in this model.

Synthesis biological evaluation of alkyl, alkoxy, alkylthio, or amino-substituted 2,3-dihydro-1,5-benzothiazepin-4(5H)-ones.

2,3-Dihydro-1,5-benzothiazepin-4(5H)-ones substituted with an alkyl, alkoxy, alkylthio, hydroxy, or amino group on the fused benzene ring of the 1,5-benzothiazepine skeleton were synthesized and their vasodilating, antihypertensive, and platelet aggregation-inhibitory activities were investigated. (-)-cis-3-Acetoxy-5-[2-(di-methylamino) ethyl]-2,3-dihydro-8-methyl-2-(4-methylphenyl)-1,5-benzothiazepin- 4(5H)-one ((-)-13e) was selected for further studies as a potent inhibitor of platelet aggregation.