Epilepsy is a prevalent neurological disorder characterized by recurrent, unprovoked seizures and altered electroencephalographic patterns. This condition is viewed as a malfunctioning of extensive neural networks due to an imbalance of excitatory and inhibitory signals leading neurons to be excessively excitable and to abnormal synchronized electrical activity. Despite the growing number of new antiepileptic drugs, patients suffering from drug-resistant forms of epilepsy do not respond to pharmacological treatment, and the only effective cure remains the neurosurgical resection of the epileptic focus. Nevertheless, several patients fail to achieve seizure freedom after surgical resection. This emphasizes the urgent need for novel human-relevant models to explore the mechanisms underlying drug-refractory forms of epilepsy. While acute and organotypic slices from resected neurological tissue offer a promising method for studying patient-derived brain tissue mechanisms, this technique is limited by its inherently low throughput and challenges in obtaining appropriate control tissue. Recent advances in organoid technology have allowed for the generation of cerebral dorsal/ventral assembloids, which more accurately model the functional connectivity between excitatory and inhibitory neurons and recapitulate key aspects of cortical circuits. This review summarizes current knowledge on the use of human brain organoids and assembloids to model epilepsy, with a particular focus on organoids harboring focal cortical dysplasia-linked mutations. Human brain organoids and assembloids will allow addressing an important question in the field, namely the relative contribution of neurodevelopmental defects vs. those arising at later stages of CNS development. Limitations of this “neuron-only” in vitro model and potential ways to include non-neuronal cells will also be discussed. Finally, we highlight recent advances in employing these new powerful platforms for investigating network dysfunctions underlying FCDs, screening potential antiepileptic drug candidates, and developing personalized therapeutic strategies.

Brain development and plasticity depend on specific microRNA (miRNA) expression patterns across cell types and subcellular compartments. Nevertheless, comprehensive profiling of localized brain miRNAs is still limited by challenges in isolating individual cell types or compartments and in detection sensitivity. To overcome these limitations, we advanced HIV-1 Gag’s ability to bind host miRNAs within Virus-like Particles to develop Synthetic Nano-Particles for Precise endogenous miRNA loading and export (SNaP). Our data establish SNaP’s modularity and portability to clinically relevant neural cells, with particle yields matching benchmark packaging cells. The integration of SNaP with a cell-specific promoter enabled lineage-restricted miRNA export, while incorporating a dendritic localization signal improved the specificity of post-synaptic miRNA recovery over traditional synaptosomes. Additional engineering with a miRNA-binding module synergistically increased synaptic miRNA packaging in a sequence-independent manner. Collectively, this work positions SNaP as a technological advancement supporting the high-resolution, spatially resolved profiling of miRNAs, adaptable to diverse polarized or heterogeneous culture systems.

In this paper, we present a neuro-inspired approach to reservoir computing (RC) in which a network of in vitro cultured cortical neurons serves as the physical reservoir. Rather than relying on artificial recurrent models to approximate neural dynamics, our biological reservoir computing (BRC) system leverages the spontaneous and stimulus-evoked activity of living neural circuits as its computational substrate. A high-density multi-electrode array (HD-MEA) provides simultaneous stimulation and readout across hundreds of channels: input patterns are delivered through selected electrodes, while the remaining ones capture the resulting high-dimensional neural responses, yielding a biologically grounded feature representation. A linear readout layer (single-layer perceptron) is then trained to classify these reservoir states, enabling the living neural network to perform static visual pattern-recognition tasks within a computer-vision framework. We evaluate the system across a sequence of tasks of increasing difficulty, ranging from pointwise stimuli to oriented bars, clock-digit-like shapes, and handwritten digits from the MNIST dataset. Despite the inherent variability of biological neural responses-arising from noise, spontaneous activity, and inter-session differences-the system consistently generates high-dimensional representations that support accurate classification. These results demonstrate that in vitro cortical networks can function as effective reservoirs for static visual pattern recognition, opening new avenues for integrating living neural substrates into neuromorphic computing frameworks. More broadly, this work contributes to the effort to incorporate biological principles into machine learning and supports the goals of neuro-inspired vision by illustrating how living neural systems can inform the design of efficient and biologically grounded computational models.

In this paper, we introduce a paradigm for reservoir computing (RC) that leverages a pool of cultured biological neurons as the reservoir substrate, creating a biological reservoir computing (BRC). This system operates similarly to an echo state network (ESN), with the key distinction that the neural activity is generated by a network of cultured neurons, rather than being modeled by traditional artificial computational units. The neuronal activity is recorded using a multi-electrode array (MEA), which enables high-throughput recording of neural signals. In our approach, inputs are introduced into the network through a subset of the MEA electrodes, while the remaining electrodes capture the resulting neural activity. This generates a nonlinear mapping of the input data to a high-dimensional biological feature space, where distinguishing between data becomes more efficient and straightforward, allowing a simple linear classifier to perform pattern recognition tasks effectively. To evaluate the performance of our proposed system, we present an experimental study that includes various input patterns, such as positional codes, bars with different orientations, and a digit recognition task. The results demonstrate the feasibility of using biological neural networks to perform tasks traditionally handled by artificial neural networks, paving the way for further exploration of biologically-inspired computing systems, with potential applications in neuromorphic engineering and bio-hybrid computing.

The mechanisms that determine distinct embryonic pallial identities remain elusive. The central role of Wnt signaling in directing dorsal telencephalic progenitors to the isocortex or hippocampus has been elucidated. Here, we show that timely inhibition of MAPK/ERK and BMP signaling in neuralized mouse embryonic stem cells (ESCs) specifies a cell identity characteristic of the allocortex. Comparison of the global gene expression profiles of neural cells generated by MAPK/ERK and BMP inhibition (MiBi cells) with those of cells from early postnatal encephalic regions reveals a pallial identity of MiBi cells, distinct from isocortical and hippocampal cells. MiBi cells display a unique pattern of gene expression and connectivity, and share molecular and electrophysiological features with the entorhinal cortex. Our results suggest that early changes in cell signaling can specify distinct pallial fates that are maintained by specific neuronal lineages independent of subsequent embryonic morphogenetic interactions and can determine their functional connectivity.

METTL9 is an enzyme catalysing N1-methylation of histidine residues (1MH) within eukaryotic proteins. Given its high expression in vertebrate nervous system and its potential association with neurodevelopmental delay, we dissected Mettl9 role during neural development. We generated three distinct mouse embryonic stem cell lines: a complete Mettl9 knock-out (KO), an inducible METTL9 Degron and a line endogenously expressing a catalytically inactive protein, and assessed their ability to undergo neural differentiation. In parallel, we down-regulated mettl9 in Xenopus laevis embryos and characterised their neural development. Our multi-omics data indicate that METTL9 exerts a conserved role in sustaining vertebrate neurogenesis. This is largely independent of its catalytic activity and occurs through modulation of the secretory pathway. METTL9 interacts with key regulators of cellular transport, endocytosis and Golgi integrity; moreover, in Mettl9KO cells Golgi becomes fragmented. Overall, we demonstrate a developmental function of Mettl9 and link it to a 1MH-independent pathway, namely, the maintenance of the secretory system, which is essential throughout neural development.

The brain is a complex system of interconnected regions that underlie memory, cognition, and perception. Today, our understanding of the brain's dynamic processes remains incomplete, particularly regarding differences in electrophysiological activity and inter-regional connectivity among specific areas. To explore this, we investigated the electrical activity, functional connectivity, and interactions of neural cultures differentiated into hippocampal, isocortical, and entorhinal networks using multi-electrode arrays (MEAs) to record extracellular local field potentials. Our results showed that collective synchronization events, or network bursts, were present in all cultures except for the hippocampal networks. Interestingly, introducing entorhinal neuron spheroids onto hippocampal cultures induced synchronized activity. Furthermore, Self-organized criticality analysis confirmed that all networks, except hippocampal cultures, were in a critical regime. Moreover, we found that entorhinal-hippocampal coupling facilitated criticality, promoting recurrent synchronized activity patterns. The consistent scaling exponents across configurations underscore the universality of criticality in biological networks. Finally, power spectrum analysis revealed a theta band peak in connected entorhinalhippocampal cultures, consistent with in vivo studies, highlighting the role of theta oscillations in memory consolidation. Our findings provide more insights into brain functioning and offer an in vitro model for studying learning and memory.

Brain development and plasticity depend on specific microRNA (miRNA) expression patterns across cell types and subcellular compartments. Nevertheless, comprehensive profiling of localized brain miRNAs is still limited by challenges in isolating individual cell types or compartments and in detection sensitivity. To overcome these limitations, we advanced HIV-1 Gag’s ability to bind host miRNAs within Virus-like Particles to develop Synthetic Nano-Particles for Precise miRNA loading and export (SNaP). Our data establish SNaP’s modularity and portability to clinically-relevant neural cells, with particle yields matching benchmark packaging cells. SNaP integration with a cell-specific promoter enabled lineage-restricted miRNA export, while incorporating a Dendritic Localization Signal improved the specificity of postsynaptic miRNA recovery over traditional synaptosomes. Additional engineering with a miRNA-binding module synergistically boosted synaptic miRNA packaging in a sequence-independent manner. Collectively, our work positions SNaP as a technological advancement supporting the high-resolution, spatially resolved profiling of non-coding RNAs, adaptable to diverse polarized or heterogeneous tissues.

Excitatory/inhibitory (E/I) balance is thought to play a key role in cortical activity development. We modeled an in vitro cortical network deployed of the inhibitory neurons normally migrating from the ventral telencephalon and implemented ventral telencephalic (VT) cultures and co-cultures with mixed proportions of dorsal telencephalic (DT) and VT neurons, containing distinct proportions of inhibitory neurons. Interestingly, these pure and mixed cultures developed different patterns of spontaneous activity and functional connectivity. Our findings highlighted a critical role for the inhibitory component in developing correlated network activity. Unexpectedly, networks with 7% of parvalbumin (PV)+ neurons were not able to generate appreciable network burst activity due to the development of a strong network inhibition, despite their lowest E/I ratio. Our observations support the notion that an optimal ratio of PV+ neurons during cortical development is essential for the establishment of local inhibitory networks capable of generating and spreading correlated activity.

In this paper, we present a biologically grounded approach to reservoir computing (RC), in which a network of cultured biological neurons serves as the reservoir substrate. This system, referred to as biological reservoir computing (BRC), replaces artificial recurrent units with the spontaneous and evoked activity of living neurons. A multi-electrode array (MEA) enables simultaneous stimulation and readout across multiple sites: inputs are delivered through a subset of electrodes, while the remaining ones capture the resulting neural responses, mapping input patterns into a high-dimensional biological feature space. We evaluate the system through a case study on digit classification using a custom dataset. Input images are encoded and delivered to the biological reservoir via electrical stimulation, and the corresponding neural activity is used to train a simple linear classifier. To contextualize the performance of the biological system, we also include a comparison with a standard artificial reservoir trained on the same task. The results indicate that the biological reservoir can effectively support classification, highlighting its potential as a viable and interpretable computational substrate. We believe this work contributes to the broader effort of integrating biological principles into machine learning and aligns with the goals of human-inspired vision by exploring how living neural systems can inform the design of efficient and biologically plausible models.

Understanding the dynamical processes driving the functioning of the brain, especially inter-regional connectivity, remains a significant challenge. This study examines methods for analyzing the electrophysiological activity and connectivity of invitro neural networks, which are pivotal for getting insights into brain functions and neurological disorders such as epilepsy and Alzheimer’s disease. Using multi-electrode arrays (MEAs) with 4096 electrodes, we recorded extracellular local field potentials from cultured neural networks. We describe our experimental setup, focusing on high-density MEA technology, and outline protocols for data collection and analysis using Python and Fortran. Our results, on the analysis of MEA signals, contribute to the understanding of the dynamics of cultured neural networks and in the development of new methods for future research.

Differentiation of specific neuronal types in the nervous system is worked out through a complex series of gene regulation events. Within the mammalian neocortex, the appropriate expression of key transcription factors allocates neurons to different cortical layers according to an inside-out model and endows them with specific properties. Precise timing is required to ensure the proper sequential appearance of key transcription factors that dictate the identity of neurons within the different cortical layers. Recent evidence suggests that aspects of this time-controlled regulation of gene products rely on post-transcriptional control, and point at micro-RNAs (miRs) and RNA-binding proteins as important players in cortical development. Being able to simultaneously target many different mRNAs, these players may be involved in controlling the global expression of gene products in progenitors and post-mitotic cells, in a gene expression framework where parallel to transcriptional gene regulation, a further level of control is provided to refine and coordinate the appearance of the final protein products. miRs and RNA-binding proteins (RBPs), by delaying protein appearance, may play heterochronic effects that have recently been shown to be relevant for the full differentiation of cortical neurons and for their projection abilities. Such heterochronies may be the base for evolutionary novelties that have enriched the spectrum of cortical cell types within the mammalian clade.

Using the timely re-activation of WNT signalling in neuralizing human induced pluripotent stem cells (hiPSCs), we have produced neural progenitor cells with a gene expression profile typical of human embryonic dentate gyrus (DG) cells. Notably, in addition to continuous WNT signalling, a specific laminin isoform is crucial to prolonging the neural stem state and to extending progenitor cell proliferation for over 200 days in vitro. Laminin 511 is indeed specifically required to support proliferation and to inhibit differentiation of hippocampal progenitor cells for extended time periods when compared with a number of different laminin isoforms assayed. Global gene expression profiles of these cells suggest that a niche of laminin 511 and WNT signalling is sufficient to maintain their capability to undergo typical hippocampal neurogenesis. Moreover, laminin 511 signalling sustains the expression of a set of genes responsible for the maintenance of a hippocampal neurogenic niche. Finally, xenograft of human DG progenitors into the DG of adult immunosuppressed host mice produces efficient integration of neurons that innervate CA3 layer cells spanning the same area of endogenous hippocampal neuron synapses.

ABSTRACTCerebral cortical development is controlled by key transcription factors that specify the neuronal identities in the different cortical layers. These transcription factors are crucial for the identity of the different neurons, but the mechanisms controlling their expression in distinct cells are only partially known. Here we investigate the expression and stability of the mRNAs of Tbr1, Bcl11b, Fezf2, Satb2 and Cux1 in single developing mouse cortical cells. We focus on Satb2 and find that its mRNA expression occurs much earlier than its protein synthesis and in a set of cells broader than expected, suggesting an initially tight control of its translation, which is subsequently de-repressed at late developmental stages. Mechanistically,Satb23’UTR modulates protein translation of GFP reporters during mouse corticogenesis. Byin vitropull-down ofSatb23’UTR-associated miRNAs, we select putative miRNAs responsible for SATB2 inhibition, focusing on those strongly expressed in early progenitor cells and reduced in late cells. miR-541, an Eutherian-specific miRNA, and miR-92a/b are the best candidates and their inactivation triggers robust and premature SATB2 translation in both mouse and human cortical cells. Our findings indicate that RNA interference plays a major role in the timing of cortical cell identity and may be part of the toolkit involved in specifying supra-granular projection neurons.

Previous work has shown that it is possible to train neuronal cultures on Multi-Electrode Arrays (MEAs), to recognize very simple patterns. However, this work was mainly focused to demonstrate that it is possible to induce plasticity in cultures, rather than performing a rigorous assessment of their pattern recognition performance. In this paper, we address this gap by developing a methodology that allows us to assess the performance of neuronal cultures on a learning task. Specifically, we propose a digital model of the real cultured neuronal networks; we identify biologically plausible simulation parameters that allow us to reliably reproduce the behavior of real cultures; we use the simulated culture to perform handwritten digit recognition and rigorously evaluate its performance; we also show that it is possible to find improved simulation parameters for the specific task, which can guide the creation of real cultures.

Estimating the co-expression of cell identity factors in single-cell is crucial. Due to the low efficiency of scRNA-seq methodologies, sensitive computational approaches are critical to accurately infer transcription profiles in a cell population. We introduce COTAN, a statistical and computational method, to analyze the co-expression of gene pairs at single cell level, providing the foundation for single-cell gene interactome analysis. The basic idea is studying the zero UMI counts' distribution instead of focusing on positive counts; this is done with a generalized contingency tables framework. COTAN can assess the correlated or anti-correlated expression of gene pairs, providing a new correlation index with an approximate p-value for the associated test of independence. COTAN can evaluate whether single genes are differentially expressed, scoring them with a newly defined global differentiation index. Similarly to correlation network analysis, it provides ways to plot and cluster genes according to their co-expression pattern with other genes, effectively helping the study of gene interactions, becoming a new tool to identify cell-identity markers. We assayed COTAN on two neural development datasets with very promising results. COTAN is an R package that complements the traditional single cell RNA-seq analysis and it is available at https://github.com/seriph78/COTAN.

PanNENs are histopathologically classified as well-differentiated pancreatic neuroendocrine tumors (panNETs) or poorly differentiated pancreatic neuroendocrine carcinomas (panNECs) according to the 2010 World Health Organization (WHO) classification system. Pancreatic neuroendocrine tumors (pNETs) occur in 60–70% of multiple endocrine neoplasia (MEN1) patients; about 30% of pNETs undergo a malignant progression, manifesting local or distant metastases, which are one of the main causes of death in MEN1 patients.1, 2 Their molecular characteristics are still undefined. Moreover, despite the common genetic basis, their clinical phenotype is highly variable among patients, even in the presence of the same mutation, suggesting a possible role of other cofactors and/or epigenetic mechanisms in each individual tumorigenesis process.3-5 We analyzed, through next-generation sequencing (NGS), the specific miRNA expression signatures of normal pancreas, gastrinoma, and neuroendocrine pancreatic tumor in a MEN1 patient, whose histopathological examination (via duodenopancreatectomy biopsies) revealed diffuse microadenomatosis with eight macrotumors (>0.5 cm) in the pancreas and four gastrin-secreting tumors (three in the duodenum and one in the gallbladder) (Figure S1). NGS analysis of normal pancreas and MEN1 tumor samples identified the presence of tumor-subtype-specific miRNAs (see Figures 1 and 2). These data open to the possibility of using them for distinguishing different MEN1 gastro-entero-pancreatic neuroendocrine tumors (GEP-NETs). Principal component analysis (PCA) analysis allowed us to confirm that our normalized miRNA expression profiles are able to resolve samples according to their tissue. In particular, while biopsy replicates displayed a low amount of variance (Figure 1A and B), the first two principal components captured 87% of the overall variance and separated the control pancreas from MEN1 pNETs and gastrinomas (Figure 1B). To delve more deeply into these groups, we then proceeded to perform a robust statistical characterization of both subtype-specific and pan-cancer miRNAs. We first performed a group-wise study, in order to extract the miRNAs specifically up-/down-regulated in either pNETs or gastrinomas when compared to control pancreas (“pNET vs. CP” and “Gas vs. CP,” respectively). miRNAs with an adjusted p-value < 0.01 were considered differentially expressed (DE) between pairs of conditions (Tables 2 and 3), and are shown in Figure 1C–F. UpSet plot in Figure 2A shows the degree of intersection between up- (UP) and downregulated (DOWN) miRNAs in “pNET vs. CP” and “Gas vs. CP” contrasts. Our data demonstrate that most of the miRNAs analyzed are homogeneously split into pNET-specific, Gas-specific, and Common classes, while only <4% of them have a divergent behavior. The normalized expression values of miRNAs belonging to the four classes in panel A are shown as heatmaps (Figure 2B–E) and may contain a useful panel of subtype-specific biomarkers. We then focused on the class of miRNAs DE between MEN1 gastro-entero-pancreatic cancer tissues (pNETs and gastrinomas) and control pancreas. miRNAs with an adjusted p-value < 0.01 were considered DE between conditions (Table 4). We observed 17 downregulated and 26 upregulated miRNAs, whose log2 FoldChanges versus average expressions and volcano plots are shown in Figure 3A. These data were also shown in a heatmap containing the normalized expression value of differentially regulated miRNAs (Figure 3B). These miRNAs represent a miRNA signature specific for MEN1 gastro-entero-pancreatic tumors. Then, we applied a correlation network model to identify co-expressed miRNAs and study the impact of their regulation in the context of the network. The reconstructed network originated from MEN1 control and MEN1 tumor gastro-entero-pancreatic samples, and focused on DE miRNAs. The topology of this network evidenced two independent cliques (Figure 3C) corresponding to two clusters of anti-correlated miRNAs, and highlighted some hubs of marked connectivity. We hypothesized that hub miRNAs may act as proxies to help explaining the regulatory function played by the broader regulatory network. Therefore, we examined hub miRNAs from the two different cliques in a deeper analysis. First, to confirm deep sequencing results, we used qRT-PCR to assess the expression of seven hub miRNAs from the two different cliques (C1 clicque: miRNA-378-3p; miRNA-1468-5p; miR-625-5p; miR-625-3p; miR-215-5p. C2 clicque: miR-1301-3p; miR-212-5p). Importantly, the trend of the alteration in the expression was generally concordant (six out of seven) between the sequencing data and qRT-PCR (Figure 4A). Moreover, for most of these miRNAs, qRT-PCR results showed a good agreement with the expression levels found in the BON1 cell line (pancreatic carcinoid tumor), commonly used as a reference model for NETs. Next, we checked whether genes predicted to be targets of these hub miRNAs shared a common function. GSEA (Gene-set Enrichment Analysis) revealed that predicted gene targets (according to miRWalk 3.0) were associated to specific KEGG pathways (Figure 4B and C and Table 5) related to cancer establishment and signaling: noteworthy, the list of the top hits includes the “pancreatic cancer” pathway. Exerting their role as key determinants of gene expression, transcription factors and miRNAs are able to co-regulate the expression of targets in the form of feed-forward loops (FFLs) and feedback loops: feed-forward loop analysis evidenced interactions between hub miRNAs and the principal genes involved in the MEN1 gastro-entero-pancreatic neoplasia, as transcription factors (TF) and/or other mRNAs (Figure 4D and E).6 The analysis showed miRNAs “hubs” organized in FFLs with chromatin-remodeling genes (MEN1, ATRX, DAXX) involved in familial and sporadic cancer, as well as with classical oncogenes (RB, TP53) normally mutated in neuroendocrine carcinomas (NEC) (Figure 4D and E). Therefore, it is tempting to suggest that these epigenetic bistable molecular regulatory circuits7, 8 could influence the transition from normal neuroendocrine cells to either well-differentiated GEP-NET cancer or to a poorly differentiated NEC. In summary, by investigating global miRNA expression signatures in matched samples of normal pancreas, pNET, and gastrinoma from a MEN1 clinical case, here we highlighted new specific miRNA signatures and confirmed previously selected microRNAs involved in MEN1 gastro-entero-pancreatic tumors. These miRNAs have been shown to be organized in two gene regulatory networks with common and distinct functional attributes. While this work clearly represents a starting point, and investigation of a greater number of cases will be required to generalize the results here presented, these data make an interesting exploratory contribution to the knowledge of the mechanisms governing MEN1 tumorigenesis. The authors gratefully thank Prof. Stefano Colagrande (Department of Experimental and Clinical Biomedical Sciences, University of Florence). The tissue samples for the study were obtained from a female donor undergoing duodenopancreatic surgery in 2002. Samples were collected by the surgeon after verbal informed consent from the patient, who fully understood that the samples were to be analyzed anonymously for future research on MEN1 syndrome not necessarily related to her specific clinical case. All genetic and molecular analyses for this study were carried out anonymously. Researchers had only access to clinical data of the donor and pathologic characteristics of the samples. Tables containing raw and processed data from the bioinformatic pipeline (including miRNA counts and complete secondary analysis results) can be accessed at the following address: http://doi.org/10.17632/wfm74g24d7.1. The authors declare no conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Neuronal hyperexcitability linked to an increase in glutamate signalling is a peculiar trait of the early stages of Alzheimer's disease (AD) and tauopathies, however, a progressive reduction in glutamate release follows in advanced stages. We recently reported that in the early phases of the neurodegenerative process, soluble, non-aggregated Tau accumulates in the nucleus and modulates the expression of disease-relevant genes directly involved in glutamatergic transmission, thus establishing a link between Tau instability and altered neurotransmission. Here we report that while the nuclear translocation of Tau in cultured cells is not impaired by its own aggregation, the nuclear amyloid inclusions of aggregated Tau abolish Tau-dependent increased expression of the glutamate transporter. Remarkably, we observed that in the prefrontal cortex (PFC) of AD patient brain, the glutamate transporter is upregulated at early stages and is downregulated at late stages. The Gene Set Enrichment Analysis indicates that the modulation of Tau-dependent gene expression along the disease progression can be extended to all protein pathways of the glutamatergic synapse. Together, this evidence links the altered glutamatergic function in the PFC during AD progression to the newly discovered function of nuclear Tau.