Machine learning-based evaluation of spontaneous pain and analgesics from cellular calcium signals in the mouse primary somatosensory cortex using explainable features.

Introduction: Pain that arises spontaneously is considered more clinically relevant than pain evoked by external stimuli. However, measuring spontaneous pain in animal models in preclinical studies is challenging due to methodological limitations. To address this issue, recently we developed a deep learning (DL) model to assess spontaneous pain using cellular calcium signals of the primary somatosensory cortex (S1) in awake head-fixed mice. However, DL operate like a "black box", where their decision-making process is not transparent and is difficult to understand, which is especially evident when our DL model classifies different states of pain based on cellular calcium signals. In this study, we introduce a novel machine learning (ML) model that utilizes features that were manually extracted from S1 calcium signals, including the dynamic changes in calcium levels and the cell-to-cell activity correlations. Method: We focused on observing neural activity patterns in the primary somatosensory cortex (S1) of mice using two-photon calcium imaging after injecting a calcium indicator (GCaMP6s) into the S1 cortex neurons. We extracted features related to the ratio of up and down-regulated cells in calcium activity and the correlation level of activity between cells as input data for the ML model. The ML model was validated using a Leave-One-Subject-Out Cross-Validation approach to distinguish between non-pain, pain, and drug-induced analgesic states. Results and discussion: The ML model was designed to classify data into three distinct categories: non-pain, pain, and drug-induced analgesic states. Its versatility was demonstrated by successfully classifying different states across various pain models, including inflammatory and neuropathic pain, as well as confirming its utility in identifying the analgesic effects of drugs like ketoprofen, morphine, and the efficacy of magnolin, a candidate analgesic compound. In conclusion, our ML model surpasses the limitations of previous DL approaches by leveraging manually extracted features. This not only clarifies the decision-making process of the ML model but also yields insights into neuronal activity patterns associated with pain, facilitating preclinical studies of analgesics with higher potential for clinical translation.

Intrinsic plasticity of Purkinje cell serves homeostatic regulation of fear memory.

Two forms of plasticity, synaptic and intrinsic, are neural substrates for learning and memory. Abnormalities in homeostatic plasticity cause severe neuropsychiatric diseases, such as schizophrenia and autism. This suggests that the balance between synaptic transmission and intrinsic excitability is important for physiological function in the brain. Despite the established role of synaptic plasticity between parallel fiber (PF) and Purkinje cell (PC) in fear memory, its relationship with intrinsic plasticity is not well understood. Here, patch clamp recording revealed depression of intrinsic excitability in PC following auditory fear conditioning (AFC). Depressed excitability balanced long-term potentiation of PF-PC synapse to serve homeostatic regulation of PF-evoked PC firing. We then optogenetically manipulated PC excitability during the early consolidation period resulting in bidirectional regulation of fear memory. Fear conditioning-induced synaptic plasticity was also regulated following optogenetic manipulation. These results propose intrinsic plasticity in PC as a novel mechanism of fear memory and elucidate that decreased intrinsic excitability in PC counterbalances PF-PC synaptic potentiation to maintain fear memory in a normal range.

An Automated Cell Detection Method for TH-positive Dopaminergic Neurons in a Mouse Model of Parkinson's Disease Using Convolutional Neural Networks.

Quantification of tyrosine hydroxylase (TH)-positive neurons is essential for the preclinical study of Parkinson's disease (PD). However, manual analysis of immunohistochemical (IHC) images is labor-intensive and has less reproducibility due to the lack of objectivity. Therefore, several automated methods of IHC image analysis have been proposed, although they have limitations of low accuracy and difficulties in practical use. Here, we developed a convolutional neural network-based machine learning algorithm for TH+ cell counting. The developed analytical tool showed higher accuracy than the conventional methods and could be used under diverse experimental conditions of image staining intensity, brightness, and contrast. Our automated cell detection algorithm is available for free and has an intelligible graphical user interface for cell counting to assist practical applications. Overall, we expect that the proposed TH+ cell counting tool will promote preclinical PD research by saving time and enabling objective analysis of IHC images.

Multiplexed Representation of Itch and Pain and Their Interaction in the Primary Somatosensory Cortex.

Itch and pain are distinct sensations that share anatomically similar pathways: from the periphery to the brain. Over the last decades, several itch-specific neural pathways and molecular markers have been identified at the peripheral and spinal cord levels. Although the perception of sensation is ultimately generated at the brain level, how the brain separately processes the signals is unclear. The primary somatosensory cortex (S1) plays a crucial role in the perception of somatosensory information, including touch, itch, and pain. In this study, we investigated how S1 neurons represent itch and pain differently. First, we established a spontaneous itch and pain mouse model. Spontaneous itch or pain was induced by intradermal treatment with 5-HT or capsaicin on the lateral neck and confirmed by a selective increase in scratching or wiping-like behavior, respectively. Next, in vivo two-photon calcium imaging was performed in awake mice after four different treatments, including 5-HT, capsaicin, and each vehicle. By comparing the calcium activity acquired during different sessions, we distinguished the cells responsive to itch or pain sensations. Of the total responsive cells, 11% were both responsive, and their activity in the pain session was slightly higher than that in the itch session. Itch- and painpreferred cells accounted for 28.4% and 60.6%, respectively, and the preferred cells showed the lowest activity in their counter sessions. Therefore, our results suggest that S1 uses a multiplexed coding strategy to encode itch and pain, and S1 neurons represent the interaction between itch and pain.

Development of a spontaneous pain indicator based on brain cellular calcium using deep learning.

Chronic pain remains an intractable condition in millions of patients worldwide. Spontaneous ongoing pain is a major clinical problem of chronic pain and is extremely challenging to diagnose and treat compared to stimulus-evoked pain. Although extensive efforts have been made in preclinical studies, there still exists a mismatch in pain type between the animal model and humans (i.e., evoked vs. spontaneous), which obstructs the translation of knowledge from preclinical animal models into objective diagnosis and effective new treatments. Here, we developed a deep learning algorithm, designated AI-bRNN (Average training, Individual test-bidirectional Recurrent Neural Network), to detect spontaneous pain information from brain cellular Ca2+ activity recorded by two-photon microscopy imaging in awake, head-fixed mice. AI-bRNN robustly determines the intensity and time points of spontaneous pain even in chronic pain models and evaluates the efficacy of analgesics in real time. Furthermore, AI-bRNN can be applied to various cell types (neurons and glia), brain areas (cerebral cortex and cerebellum) and forms of somatosensory input (itch and pain), proving its versatile performance. These results suggest that our approach offers a clinically relevant, quantitative, real-time preclinical evaluation platform for pain medicine, thereby accelerating the development of new methods for diagnosing and treating human patients with chronic pain.

Kdm3b haploinsufficiency impairs the consolidation of cerebellum-dependent motor memory in mice.

Histone modifications are a key mechanism underlying the epigenetic regulation of gene expression, which is critically involved in the consolidation of multiple forms of memory. However, the roles of histone modifications in cerebellum-dependent motor learning and memory are not well understood. To test whether changes in histone methylation are involved in cerebellar learning, we used heterozygous Kdm3b knockout (Kdm3b+/-) mice, which show reduced lysine 9 on histone 3 (H3K9) demethylase activity. H3K9 di-methylation is significantly increased selectively in the granule cell layer of the cerebellum of Kdm3b+/- mice. In the cerebellum-dependent optokinetic response (OKR) learning, Kdm3b+/- mice show deficits in memory consolidation, whereas they are normal in basal oculomotor performance and OKR acquisition. In addition, RNA-seq analyses revealed that the expression levels of several plasticity-related genes were altered in the mutant cerebellum. Our study suggests that active regulation of histone methylation is critical for the consolidation of cerebellar motor memory.

Neural Plasticity in the Brain during Neuropathic Pain.

Neuropathic pain is an intractable chronic pain, caused by damage to the somatosensory nervous system. To date, treatment for neuropathic pain has limited effects. For the development of efficient therapeutic methods, it is essential to fully understand the pathological mechanisms of neuropathic pain. Besides abnormal sensitization in the periphery and spinal cord, accumulating evidence suggests that neural plasticity in the brain is also critical for the development and maintenance of this pain. Recent technological advances in the measurement and manipulation of neuronal activity allow us to understand maladaptive plastic changes in the brain during neuropathic pain more precisely and modulate brain activity to reverse pain states at the preclinical and clinical levels. In this review paper, we discuss the current understanding of pathological neural plasticity in the four pain-related brain areas: the primary somatosensory cortex, the anterior cingulate cortex, the periaqueductal gray, and the basal ganglia. We also discuss potential treatments for neuropathic pain based on the modulation of neural plasticity in these brain areas.

Restoring synaptic plasticity and memory in mouse models of Alzheimer's disease by PKR inhibition.

Alzheimer's disease (AD) is a neurodegenerative disorder associated with deficits in cognition and synaptic plasticity. While accumulation of amyloid ß (Aß) and hyper-phosphorylation of tau are parts of the etiology, AD can be caused by a large number of different genetic mutations and other unknown factors. Considering such a heterogeneous nature of AD, it would be desirable to develop treatment strategies that can improve memory irrespective of the individual causes. Reducing the phosphorylation of eukaryotic translation initiation factor 2α (eIF2α) was shown to enhance long-term memory and synaptic plasticity in naïve mice. Moreover, hyper-phosphorylation of eIF2α is observed in the brains of postmortem AD patients. Therefore, regulating eIF2α phosphorylation can be a plausible candidate for restoring memory in AD by targeting memory-enhancing mechanism. In this study, we examined whether PKR inhibition can rescue synaptic and learning deficits in two different AD mouse models; 5XFAD transgenic and Aß1-42-injected mice. We found that the acute treatment of PKR inhibitor (PKRi) can restore the deficits in long-term memory and long-term potentiation (LTP) in both mouse models without affecting the Aß load in the hippocampus. Our results prove the principle that targeting memory enhancing mechanisms can be a valid candidate for developing AD treatment.

The Brain-Enriched MicroRNA miR-9-3p Regulates Synaptic Plasticity and Memory.

UNLABELLED: MicroRNAs (miRNAs) are small, noncoding RNAs that posttranscriptionally regulate gene expression in many tissues. Although a number of brain-enriched miRNAs have been identified, only a few specific miRNAs have been revealed as critical regulators of synaptic plasticity, learning, and memory. miR-9-5p/3p are brain-enriched miRNAs known to regulate development and their changes have been implicated in several neurological disorders, yet their role in mature neurons in mice is largely unknown. Here, we report that inhibition of miR-9-3p, but not miR-9-5p, impaired hippocampal long-term potentiation (LTP) without affecting basal synaptic transmission. Moreover, inhibition of miR-9-3p in the hippocampus resulted in learning and memory deficits. Furthermore, miR-9-3p inhibition increased the expression of the LTP-related genes Dmd and SAP97, the expression levels of which are negatively correlated with LTP. These results suggest that miR-9-3p-mediated gene regulation plays important roles in synaptic plasticity and hippocampus-dependent memory. SIGNIFICANCE STATEMENT: Despite the abundant expression of the brain-specific microRNA miR-9-5p/3p in both proliferating and postmitotic neurons, most functional studies have focused on their role in neuronal development. Here, we examined the role of miR-9-5p/3p in adult brain and found that miR-9-3p, but not miR-9-5p, has a critical role in hippocampal synaptic plasticity and memory. Moreover, we identified in vivo binding targets of miR-9-3p that are involved in the regulation of long-term potentiation. Our study provides the very first evidence for the critical role of miR-9-3p in synaptic plasticity and memory in the adult mouse.