Feature importance of machine learning prediction models shows structurally active part and important physicochemical features in drug design.

The objective of this study was to obtain the indicators of physicochemical parameters and structurally active sites to design new chemical entities with desirable pharmacokinetic profiles by investigating the process by which machine learning prediction models arrive at their decisions, which are called explainable artificial intelligence. First, we developed the prediction models for metabolic stability, CYP inhibition, and P-gp and BCRP substrate recognition using 265 physicochemical parameters for designing the molecular structures. Four important parameters, including the well-known indicator h_logD, are common in some in vitro studies; as such, these can be used to optimize compounds simultaneously to address multiple pharmacokinetic concerns. Next, we developed machine learning models that had been programmed to show structurally active sites. Many types of machine learning models were developed using the results of in vitro metabolic stability study of around 30000 in-house compounds. The metabolic sites of in-house compounds predicted using some prediction models matched experimentally identified metabolically active sites, with a ratio of number of metabolic sites (predicted/actual) of over 90%. These models can be applied to several screening projects. These two approaches can be employed for obtaining lead compounds with desirable pharmacokinetic profiles efficiently.

Predicting drug metabolism and pharmacokinetics features of in-house compounds by a hybrid machine-learning model.

We constructed machine learning-based pharmacokinetic prediction models with very high performance. The models were trained on 26138 and 16613 compounds involved in metabolic stability and cytochrome P450 inhibition, respectively. Because the compound features largely differed between the publicly available and in-house compounds, the models learned on the public data could not predict the in-house compounds, suggesting that outside of a certain applicability domain (AD), the prediction results are unreliable and can mislead the design of novel compounds. To exclude the uncertain prediction results, we constructed another machine learning model that determines whether the newly designed compound is inside or outside the AD. The AD was evaluated multi-dimensionally with some explanatory variables: The structural similarities and the probability obtained from the pharmacokinetic prediction model. The accuracy of predicting metabolic stability was 79.9% on the test set, increasing significantly to 93.6% after excluding the low-reliability compounds. The model properly classified the reliability of the compounds. After learning on the in-house compounds, the reliability model classified almost all (90%) of the public compounds as low reliability, indicating that the AD was properly determined by the model.

Effects of Optogenetic Suppression of Cortical Input on Primate Thalamic Neuronal Activity during Goal-Directed Behavior.

The motor thalamus relays signals from subcortical structures to the motor cortical areas. Previous studies in songbirds and rodents suggest that cortical feedback inputs crucially contribute to the generation of movement-related activity in the motor thalamus. In primates, however, it remains uncertain whether the corticothalamic projections may play a role in shaping neuronal activity in the motor thalamus. Here, using an optogenetic inactivation technique with the viral vector system expressing halorhodopsin, we investigated the role of cortical input in modulating thalamic neuronal activity during goal-directed behavior. In particular, we assessed whether the suppression of signals originating from the supplementary eye field at the corticothalamic terminals could change the task-related neuronal modulation in the oculomotor thalamus in monkeys performing a self-initiated saccade task. We found that many thalamic neurons exhibited changes in their firing rates depending on saccade direction or task event, indicating that optical stimulation exerted task-specific effects on neuronal activity beyond the global changes in baseline activity. These results suggest that the corticothalamic projections might be actively involved in the signal processing necessary for goal-directed behavior. However, we also found that some thalamic neurons exhibited overall, non-task-specific changes in the firing rate during optical stimulation, even in control animals without vector injections. The stimulation effects in these animals started with longer latency, implying a possible thermal effect on neuronal activity. Thus, our results not only reveal the importance of direct cortical input in neuronal activity in the primate motor thalamus, but also provide useful information for future optogenetic studies.

Roles of the Cerebellum in Motor Preparation and Prediction of Timing.

The cerebellum is thought to have a variety of functions because it developed with the evolution of the cerebrum and connects with different areas in the frontoparietal cortices. Like neurons in the cerebral cortex, those in the cerebellum also exhibit strong activity during planning in addition to the execution of movements. However, their specific roles remain elusive. In this article, we review recent findings focusing on preparatory activities found in the primate deep cerebellar nuclei during tasks requiring deliberate motor control and temporal prediction. Neurons in the cerebellum are active during anti-saccade preparation and their inactivation impairs proactive inhibitory control for saccades. Experiments using a self-timing task show that there are mechanisms for tracking elapsed time and regulating trial-by-trial variation in timing, and that the cerebellum is involved in the latter. When predicting the timing of periodic events, the cerebellum provides more accurate temporal information than the striatum. During a recently developed synchronized eye movement task, cerebellar nuclear neurons exhibited periodic preparatory activity for predictive synchronization. In all cases, the cerebellum generated preparatory activity lasting for several hundred milliseconds. These signals may regulate neuronal activity in the cerebral cortex that adjusts movement timing and predicts the timing of rhythmic events.

Neural oscillations in the primate caudate nucleus correlate with different preparatory states for temporal production.

When measuring time, neuronal activity in the cortico-basal ganglia pathways has been shown to be temporally scaled according to the interval, suggesting that signal transmission within the pathways is flexibly controlled. Here we show that, in the caudate nuclei of monkeys performing a time production task with three different intervals, the magnitude of visually-evoked potentials at the beginning of an interval differed depending on the conditions. Prior to this response, the power of low frequency components (6-20 Hz) significantly changed, showing inverse correlation with the visual response gain. Although these components later exhibited time-dependent modification during self-timed period, the changes in spectral power for interval conditions qualitatively and quantitatively differed from those associated with the reward amount. These results suggest that alteration of network state in the cortico-basal ganglia pathways indexed by the low frequency oscillations may be crucial for the regulation of signal transmission and subsequent timing behavior.

Different contributions of preparatory activity in the basal ganglia and cerebellum for self-timing.

The ability to flexibly adjust movement timing is important for everyday life. Although the basal ganglia and cerebellum have been implicated in monitoring of supra- and sub-second intervals, respectively, the underlying neuronal mechanism remains unclear. Here, we show that in monkeys trained to generate a self-initiated saccade at instructed timing following a visual cue, neurons in the caudate nucleus kept track of passage of time throughout the delay period, while those in the cerebellar dentate nucleus were recruited only during the last part of the delay period. Conversely, neuronal correlates of trial-by-trial variation of self-timing emerged earlier in the cerebellum than the striatum. Local inactivation of respective recording sites confirmed the difference in their relative contributions to supra- and sub-second intervals. These results suggest that the basal ganglia may measure elapsed time relative to the intended interval, while the cerebellum might be responsible for the fine adjustment of self-timing.

[Neural Mechanisms of Temporal Monitoring and Prediction].

When waiting for a traffic light or dancing to a musical beat, we unconsciously keep track of elapsed time and precisely predict the timing of forthcoming sensory events. Temporal monitoring and prediction are integral to our daily life, and are regulated by neuronal processes through multiple global networks involving the frontoparietal cortices, the basal ganglia and the cerebellum. These processes are also known to be influenced by a variety of internal state and neuromodulators. Here, we review recent advance of research in the field.

Causal Role of Noradrenaline in the Timing of Internally Generated Saccades in Monkeys.

We recently found that when monkeys performed an oculomotor version of the time production task, the trial-by-trial latency of self-timed saccades was negatively correlated with pupil diameter just before the delay period (Suzuki et al., 2016). Since pupil diameter has been shown to correlate with neuronal activity in the locus coeruleus, the level of noradrenaline (NA) in the brain might regulate the subjective passage of time. To examine this, we orally administered a selective noradrenaline reuptake inhibitor (reboxetine, 0.4-0.8 mg) when animals made a self-initiated memory-guided saccade >1 s following the appearance of a brief visual cue. We found that reboxetine delayed self-timed saccades, while the latency of visually triggered saccades remained unchanged. Because the changes in proportions and latencies of early impulsive saccades were comparable between conditions with and without drug administration, alteration of self-timing might not result from reduced impulsivity. We also assessed other behavioral parameters (saccade accuracy, velocity, and latency variance), but failed to find any drug effect except for the accuracy of visually triggered saccades in the high-dose condition, indicating that reboxetine specifically altered self-timing under our experimental conditions. Our results suggest that NA-related internal states may causally regulate temporal information processing in the brain.

Pharmacokinetic Properties of DNA Aptamers with Base Modifications.

The addition of novel side chains at the 5-position of uracil is an effective means to increase chemical diversity of aptamers and hence the success rate for discovery of high-affinity ligands to protein targets. Such modifications also increase nuclease resistance, which is useful in a range of applications, especially for therapeutics. In this study, we assess the impact of these side chains on plasma pharmacokinetics of modified aptamers conjugated to a 40 kDa polyethylene glycol. We show that clearance from plasma depends on relative hydrophobicity: side chains with a negative cLogP (more hydrophilic) result in slower plasma clearance compared with side chains with a positive cLogP (more hydrophobic). We show that clearance increases with the number of side chains in sequences of ≥28 synthons, but this effect is dramatically diminished in shorter sequences. These results serve as a guide for the design of new therapeutic aptamers with diversity-enhancing side chains.

Correlation between Pupil Size and Subjective Passage of Time in Non-Human Primates.

Our daily experience of time is strongly influenced by internal states, such as arousal, attention, and mood. However, the underlying neuronal mechanism remains largely unknown. To investigate this, we recorded pupil diameter, which is closely linked to internal factors and neuromodulatory signaling, in monkeys performing the oculomotor version of the time production paradigm. In the self-timed saccade task, animals were required to make a memory-guided saccade during a predetermined time interval following a visual cue. We found that pupil diameter was negatively correlated with trial-by-trial latency of self-timed saccades. Because no significant correlation was found for visually guided saccades, correlation of self-timed saccades could not be explained solely by the facilitation of saccade execution. As the reward amount was manipulated, pupil diameter and saccade latency altered in opposite directions and the magnitudes of modulation correlated strongly across sessions, further supporting the close link between pupil diameter and the subjective passage of time. When the animals were trained to produce two different intervals depending on the instruction, the pupil size again correlated with the trial-by-trial variation of saccade latency in each condition; however, pupil diameter differed significantly for saccades with similar latencies generated under different conditions. Our results indicate that internal brain states indexed by pupil diameter, which parallel noradrenergic neuronal activity (Aston-Jones and Cohen, 2005), may bias trial-by-trial variation in the subjective passage of time. SIGNIFICANCE STATEMENT: Daily experience of time is strongly influenced by our internal state, but the underlying neuronal mechanism remains elusive. Here we demonstrate that pupil diameter is negatively correlated with subjective elapsed time in monkeys performing an oculomotor version of the time production task. When the animals reported two different intervals depending on the instruction, pupil size was correlated with reported timing in each condition but differed for similar timing under different conditions. Given the close correlation between pupil diameter and noradrenergic signaling reported previously, our data indicate that brain states probed by pupil diameter and noradrenergic neuronal activity might modulate subjective passage of time.

Implications of Lateral Cerebellum in Proactive Control of Saccades.

UNLABELLED: Although several lines of evidence establish the involvement of the medial and vestibular parts of the cerebellum in the adaptive control of eye movements, the role of the lateral hemisphere of the cerebellum in eye movements remains unclear. Ascending projections from the lateral cerebellum to the frontal and parietal association cortices via the thalamus are consistent with a role of these pathways in higher-order oculomotor control. In support of this, previous functional imaging studies and recent analyses in subjects with cerebellar lesions have indicated a role for the lateral cerebellum in volitional eye movements such as anti-saccades. To elucidate the underlying mechanisms, we recorded from single neurons in the dentate nucleus of the cerebellum in monkeys performing anti-saccade/pro-saccade tasks. We found that neurons in the posterior part of the dentate nucleus showed higher firing rates during the preparation of anti-saccades compared with pro-saccades. When the animals made erroneous saccades to the visual stimuli in the anti-saccade trials, the firing rate during the preparatory period decreased. Furthermore, local inactivation of the recording sites with muscimol moderately increased the proportion of error trials, while successful anti-saccades were more variable and often had shorter latency during inactivation. Thus, our results show that neuronal activity in the cerebellar dentate nucleus causally regulates anti-saccade performance. Neuronal signals from the lateral cerebellum to the frontal cortex might modulate the proactive control signals in the corticobasal ganglia circuitry that inhibit early reactive responses and possibly optimize the speed and accuracy of anti-saccades. SIGNIFICANCE STATEMENT: Although the lateral cerebellum is interconnected with the cortical eye fields via the thalamus and the pons, its role in eye movements remains unclear. We found that neurons in the caudal part of the lateral (dentate) nucleus of the cerebellum showed the increased firing rate during the preparation of anti-saccades. Inactivation of the recording sites modestly elevated the rate of erroneous saccades to the visual stimuli in the anti-saccade trials, while successful anti-saccades during inactivation tended to have a shorter latency. Our data indicate that neuronal signals in the lateral cerebellum may proactively regulate anti-saccade generation through the pathways to the frontal cortex, and may inhibit early reactive responses and regulate the accuracy of anti-saccades.

Chemically Modified Interleukin-6 Aptamer Inhibits Development of Collagen-Induced Arthritis in Cynomolgus Monkeys.

Interleukin-6 (IL-6) is a potent mediator of inflammatory and immune responses, and a validated target for therapeutic intervention of inflammatory diseases. Previous studies have shown that SL1026, a slow off-rate modified aptamer (SOMAmer) antagonist of IL-6, neutralizes IL-6 signaling in vitro. In the present study, we show that SL1026 delays the onset and reduces the severity of rheumatoid symptoms in a collagen-induced arthritis model in cynomolgus monkeys. SL1026 (1 and 10 mg/kg), administered q.i.d., delayed the progression of arthritis and the concomitant increase in serum IL-6 levels compared to the untreated control group. Furthermore, SL1026 inhibited IL-6-induced STAT3 phosphorylation ex vivo in T lymphocytes from human blood and IL-6-induced C-reactive protein and serum amyloid A production in human primary hepatocytes. Importantly, SOMAmer treatment did not elicit an immune response, as evidenced by the absence of anti-SOMAmer antibodies in plasma of treated monkeys. These results demonstrate that SOMAmer antagonists of IL-6 may be attractive agents for the treatment of IL-6-mediated diseases, including rheumatoid arthritis.

Chemically modified DNA aptamers bind interleukin-6 with high affinity and inhibit signaling by blocking its interaction with interleukin-6 receptor.

Interleukin-6 (IL-6) is a pleiotropic cytokine that regulates immune and inflammatory responses, and its overproduction is a hallmark of inflammatory diseases. Inhibition of IL-6 signaling with the anti-IL-6 receptor antibody tocilizumab has provided some clinical benefit to patients; however, direct cytokine inhibition may be a more effective option. We used the systematic evolution of ligands by exponential enrichment (SELEX) process to discover slow off-rate modified aptamers (SOMAmers) with hydrophobic base modifications that inhibit IL-6 signaling in vitro. Two classes of IL-6 SOMAmers were isolated from modified DNA libraries containing 40 random positions and either 5-(N-benzylcarboxamide)-2'-deoxyuridine (Bn-dU) or 5-[N-(1-naphthylmethyl)carboxamide]-2'-deoxyuridine (Nap-dU) replacing dT. These modifications facilitate the high affinity binding interaction with IL-6 and provide resistance against degradation by serum endonucleases. Post-SELEX optimization of one Bn-dU and one Nap-dU SOMAmer led to improvements in IL-6 binding (10-fold) and inhibition activity (greater than 20-fold), resulting in lead SOMAmers with sub-nanomolar affinity (Kd = 0.2 nm) and potency (IC50 = 0.2 nm). Although similar in inhibition properties, the two SOMAmers have unique sequences and different ortholog specificities. Furthermore, these SOMAmers were stable in human serum in vitro for more than 48 h. Both SOMAmers prevented IL-6 signaling by blocking the interaction of IL-6 with its receptor and inhibited the proliferation of tumor cells in vitro as effectively as tocilizumab. This new class of IL-6 inhibitor may be an effective therapeutic alternative for patients suffering from inflammatory diseases.

Synthesis and evaluation of a series of 2,4-diaminopyridine derivatives as potential positron emission tomography tracers for neuropeptide Y Y1 receptors.

A series of 2,4-diaminopyridine derivatives was synthesized and evaluated as potential candidates for neuropeptide Y (NPY) Y1 receptor positron emission tomography (PET) tracers. Derivatives bearing substitutions allowing reliable access to radiolabeling were designed, focusing on Y1 binding affinity and lipophilicity. The advanced derivatives 2n and 2o were identified as promising PET tracer candidates.

Design, synthesis and evaluation of a novel cyclohexanamine class of neuropeptide Y Y1 receptor antagonists.

A novel series of cyclohexanamine derivatives was designed and synthesized as potent and selective human neuropeptide Y Y1 receptor antagonists. Modification of high-throughput screening hit compound 1 resulted in the identification of compound 3i, which displays potent Y1 activity and good selectivity towards hERG K(+) channel and serotonin transporter.

Optimization of a series of 2,4-diaminopyridines as neuropeptide Y Y1 receptor antagonists with reduced hERG activity.

The synthesis and evaluation of a series of 2,4-diaminopyridine-based neuropeptide Y Y1 (NPY Y1) receptor antagonists are described. Compound 1 was previously reported by our laboratory to be a potent and selective Y1 antagonist; however, 1 was also found to have potent hERG inhibitory activity. The main focus of this communication is structure-activity relationship development aimed at eliminating the hERG activity of 1. This resulted in the identification of compound 3d as a potent and selective NPY Y1 antagonist with reduced hERG liability.

RNase S complex bearing arginine-rich peptide and anti-HIV activity.

Basic peptide-mediated protein delivery into living cells is becoming recognized as a potent approach for the understanding of cellular mechanisms and drug delivery. We have prepared the conjugates of the S-peptide (1-15) derived from RNase S with membrane-permeable basic peptides, octaarginine and the human immunodeficient virus (HIV)-1 Rev (34-50). The RNase S complexes, formed among these S-peptide (1-15)-basic peptide conjugates and the S-protein and having a dissociation constant in the range of 10(-5) M, efficiently penetrated into the HeLa cells. These RNase S complexes exerted an anti-HIV replication activity. The time-of-drug-addition assay suggested that the site of action for these complexes would reside in the stages between the viral entry into the cells and reverse transcription. The present study exemplified the applicability of the arginine-rich peptides to the intracellular targeting of non-covalent protein complexes and supramolecular assemblies for the research in chemical and cellular biology.

Structural variety of membrane permeable peptides.

Peptide-mediated protein delivery into living cells has been attracting our attention. Among the peptides that have been reported to have carrier activity, the one from the human immunodeficient virus (HIV)-1 Tat has been most often used for the introduction of exogenous macromolecules into cells. We have shown that not only the Tat peptide, but also various arginine-rich peptides showed very similar characteristics in translocation, and the possible presence of ubiquitous internalization mechanisms among the arginine-rich peptides has also been suggested. These arginine-rich peptides includes ones derived from HIV-1 Rev and flock house virus coat proteins. The linear- and branched-chain peptides containing approximately 8 residues of arginine also show a similar ability. In this review, we present the structural variety of membrane permeable peptides and provide a survey of the findings on the translocation of these peptides through the cell membranes.

Generation of reactive oxygen species and activation of NF-kappaB by non-Abeta component of Alzheimer's disease amyloid.

Non-amyloid beta (Abeta) component of Alzheimer's disease (AD) amyloid (NAC) coexists with Abeta protein in senile plaques. After exposure to NAC fibrils, cortical neurons of rat brain primary culture became apoptotic, while astrocytes were activated with extension of their processes. NAC fibrils decreased the activity of reducing 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) in cortical neurons more markedly (IC(50) = 5.6 microm) than in astrocytes (IC(50) approximately 50 microm). The neuron-specific toxicity of NAC fibrils was indicated also by an increased release of lactate dehydrogenase from the cells. Neuronal apoptosis was suppressed by pre-treatment with the antioxidants, propyl gallate (PG) and N-t-butyl-phenylnitrone (BPN), or overexpression of human Bcl-2. Exposure to NAC fibrils enhanced generation of reactive oxygen species (ROS) in neurons and less efficiently in astrocytes, as demonstrated by oxidation of 2',7'-dichlorofluorescin. The site of ROS generation was shown to be mitochondria by oxidation of chloromethyl-tetramethyl rosamine. Exposure to NAC fibrils increased also the nuclear translocation of nuclear factor kappa B (NF-kappaB) and enhanced its DNA-binding activity, which was inhibited by PG and BPN more efficiently in neurons than in astrocytes. These results suggest that NAC fibrils increase mitochondrial ROS generation and activate NF-kappaB, thereby causing a differential change in gene expression between neurons and astrocytes in the AD brain.

Translocation of branched-chain arginine peptides through cell membranes: flexibility in the spatial disposition of positive charges in membrane-permeable peptides.

A basic peptide derived from HIV-1 Tat has been reported to have the ability to translocate through cell membranes and to bring exogenous proteins into cells. We have demonstrated that these features could be observed among many arginine-rich peptides, and the presence of a ubiquitous internalization mechanism for arginine-rich oligopeptides has been suggested. In this report, we report that these features are also applicable to the peptides having branched-chain structures. Peptides that have arginine residues on four branched chains (R(n))(4) [n (number of arginine residues)= 0-6] were prepared. Fluorescence microscopic observation revealed that the (R(2))(4) peptide exhibited the most efficient translocation. The dependence on the number of arginine residues of the translocation efficiency and cellular localization was also observed for the branched-chain peptides as was seen in the linear peptides. Quite interestingly, efficient translocation was also recognized in the (RG(3)R)(4) peptide, where three glycine residues intervened between two arginine residues on each chain of (R(2))(4). The results strongly suggested that a linear structure was not indispensable for the translocation of arginine-rich peptides and that there could be considerable flexibility in the location of the arginine residue in the molecules.