Development of an optical microscopy platform for high-throughput screening of human brain tissue

The human brain volume is several orders of magnitude larger than the mouse one and no technique exists able to image it in its entirety even with μm resolution. To map its structure with sub-cellular resolution over mesoscale (millimeters to centimeters) sized tissue samples, a light sheet fluorescence microscope (LSFM) design is needed. By illuminating a single plane of the sample, it provides intrinsic optical sectioning and direct fast 2D image recording, while minimizing out-of-focus fluorescence background and reducing sample photodamage and photobleaching. To scale up the possibility of imaging in 3D the neuron structure, with a sub-cellular resolution of the whole human brain, we will replicate the innovative apparatus, dual view inverted dual slit confocal light sheet fluorescence microscope, already developed in our lab within the European Human Brain Project and the USA BRAIN initiative projects. The prototype is characterized by having a 4 µm isotropic resolution, fast acquisition rate (47 frames per second, 0.1 mm3/s), digital light-sheet generated by pivoting a galvanometric mirror, suitable for simultaneous illumination at different wavelengths, one photon fluorescence excitation at 4 wavelengths (488, 561, 594, 647 nm), compatible with samples large up to 30x30x0.05 cm3. The new apparatus will be used to analyze a large number of samples to increase the statistical analysis obtainable with this kind of methodology.

Development of optical clearing and staining methods for multi-target investigation of human brain tissue cytoarchitecture

3D reconstruction with LSFM requires the sample to be transparent. However, biological tissues are opaque because of light scattering, therefore a clearing step is necessary. In this project, we developed a new tissue-clearing protocol called SHORT (Pesce et al. 2022), based on the TDE method (Costantini et al. Sci. Rep. 2015) and the SWITCH immunohistochemistry technique (Murray et al. Cell 2015) to stain the cellular structures of clarified samples while highlighting the contrast between nuclei and cytoplasm.

To scale up the analysis for routine use, the sample treatment will be parallelized. Moreover, to enlarge the panel of biological markers analyzable with the LSFM, a large plethora of staining will be tested and applied. Various antibodies for specific staining will be used in addition to small general molecules staining (e.g nuclei staining, membrane staining, etc). In particular, cell markers, both for neuronal and non-neuronal classification, will be analyzed; in addition, vasculature will be detected, and fiber-specific labeling will be pursued. Finally, multi-round and multi-staining approaches will be investigated to obtain different staining on the same sample.

Artificial intelligence to extract information from large-scale microscopy images

We plan to develop and apply novel deep learning architecture for detecting cells and other anatomical entities from 3D images. In particular, we wish to address two important challenges in whole brain images analysis, that emerged from previous studies: (1) the variability in cell size and brightness across different regions of the brain, and (2) the enormous contrast variability that characterizes light-sheet microscopy. Introducing mixtures of convolutional layers can help to construct better convolutional filters that will hopefully specialize to handle different regions and different levels of image quality. In addition, we are considering applying recent Bayesian optimization techniques for automatically configuring image processing pipelines that can improve image quality. Results are very sensitive to configuration parameters and within the model-based sequential optimization strategy it is possible to let the model to depend not only on the parameters tuple but also on features of the image region being considered. This strategy will be compared against the best results achieved with deep learning approaches.
Related to the Zebrafish Lab activity, we have plans to develop a super-resolution algorithm that can exploit the peculiar characteristic of the image acquisition system, where temporal resolution can be traded for spatial resolution, allowing us to compile a datasets of 4D images that have either high spatial resolution or high temporal resolution.

Large-scale mapping of human brain from biobanks

The combination of optical tissue clearing with advanced microscopy techniques can address the issues of more traditional histology protocols, enabling 3D histology, providing a higher level of anatomical insight. The approach presented here will be used to analyze tissues affected by various pathologies to provide insight on the disease, eliminating human biases and obtaining a more precise diagnostic evaluation. But will be used also to investigate the anatomical architecture of healthy samples arrived from biobanks to provide a cell census of the various brain areas, ensuring a complete description of the organ.

Large-scale genomic analysis of human developmental encephalopathies (DE) for causative genes identification and functional characterization of pathogenic mechanisms

We will expand the knowledge on genetic determinants of DE by performing whole exome sequencing (WES) and, subsequently whole genome sequencing (WGS) in undiagnosed patients with Developmental Encephalopathies (DE). In patients exhibiting focal brain malformations who have been treated surgically for drug-resistant epilepsy, we will perform Unique Molecular Identifiers (UMI)-based high-coverage NGS of candidate gene panels or WGS in paired DNA samples extracted from single cells or homogeneous pools of cells isolated through laser microdissection or cell sorting from dysplastic brain tissue and blood. This approach will allow identifying point mutations in coding and non-coding (i.e. regulatory) regions as well as pathogenic copy number variants.

Comprehensive classification of pathological human brain tissue

We will classify dysplastic brain tissue surgically removed from patients with epileptogenic brain lesions and causing DE by characterizing them at the morphological and immunohistochemical level. We will use conventional markers (including, for example, hematoxylin and eosin, anti-GFAP, and anti-NeuN) as well as antibodies against PI3K/AKT/mTOR pathway proteins and proteins whose involvement in DE pathogenesis will emerge in the framework of the study. For a more comprehensive characterization of the dysplastic tissue, we will integrate data originating from such characterization with those originating from genetic studies.

Transcriptomics analysis of dysplastic human brain tissue

We will use an RNA-Sequencing (RNASeq)-based approach to compare the transcriptomics profile of aberrant cells in focal cortical dysplasia (FCD) caused by mutations in known causative genes versus mutation-negative samples. Upon confirming the type of FCD at the anatomopathological level and dissociating the dysplastic tissue, we will separate single cells or nuclei using a FACS sorter or filtration protocols set up ad hoc. To compare gene expression in WT and dysplastic tissues, we will use the 10X Genomics Chromium technology to generate libraries and Illumina platforms to sequence them. We will integrate results originating from RNAseq analysis with those originating from genetic analysis, to identify novel genes and pathways involved in FCD pathogenesis. Identifying up-regulated or downregulated genes in FCD will also provide indications on possible targets for novel personalized therapeutic strategies to be tested, as a proof of principle, in animal models available within the consortium.

Characterizing, in neuronal and electroporated models, morphologic abnormalities and electrophysiological dysfunction caused by somatic and constitutional mutations

In order to explore the effects of the mutations we will identify through genetic analysis, we will overexpress them in primary cultures of rat neurons via nucleofection or lipofectamine-based transfection and in animal models via in utero electroporation (IUE). We will use rat neurons cultures to perform morpho-functional analysis at the cellular level (e.g. using Sholl analysis to evaluate proper neuronal branching, immunofluorescence to assess synapses number and functionality, and electrophysiology to probe alterations in single neurons electrophysiology). In animal models generated via IUE, we will characterize possible defects in proliferation, migration, and regional organization induced by the mutations electroporating embryo rat brains at E15 and performing immunohistochemical analysis and confocal imaging at different prenatal and postnatal stages (e.g. E16.5, E20, P5, and P28). When indicated, we will also carry out electrophysiological recordings on acute brain slices to search for alterations in cell connectivity. For selected mutations associated with severe encephalopathies, we will reprogram patients’ fibroblasts into human neurons to explore abnormalities in morphology and functional properties induced by different mutations in the genetic background of patients using Sholl analysis and electrophysiological recordings.

Development of high-throughput optical systems for functional and structural mapping of the nervous system in zebrafish larvae

We will make use of a custom-made two-photon light-sheet microscope able to acquire the whole-brain of zebrafish larvae with high spatial (2 μm × 2 μm × 5 μm) and temporal (200 ms) resolution. The employed two-photon illumination is advantageous for zebrafish larva studies since it exploits infra-red excitation that is invisible to the larva and therefore it does not induce visual responses that otherwise would affect the neuronal activity.

By using transgenic lines expressing a fluorescent calcium activity sensor in neuronal cells, it is possible with this setup to record the individual neuronal activity of almost every cell composing the animal brain. Furthermore, we have integrated in the setup a virtual reality apparatus that will allow us to elicit relevant animal behaviours (and therefore correlated neuronal states) by controlling the visual perception of the surrounding environment.

Finally, to complete our optical neurophysiology approach, we will integrate in this setup an additional custom-made module for optogenetic activation of individual cells in a three-dimensional volume with millisecond-sized temporal resolution. In this way, by using transgenic lines expressing light-activated proteins, we will be able at the same time to record and to finely control the brain activity in the whole-brain of the larva. Moreover, it will be possible also to implement a closed-loop approach, thus controlling the activity patterns of individual neurons as a response to the distributed neural activity and global brain dynamics.

Expansion of zebrafish animal facility

The Department of Biology has developed over the past ten years a facility and the expertise to raise zebrafish for scientific research. The facility is equipped with a standalone Z-Hub (Pentair) system, capable of hosting a total of 48 tanks and about 1500 adult fish. Nursery for raising larvae and dedicated incubators allow efficient growth from larvae to adult stage in the laboratory. The lab is also equipped with a microinjection system for the generation of transgenic lines and fluorescence stereomicroscopes for transgene screening. Within the Brain Optical Mapping (humanbrainmap) project, BIO has implemented an additional standalone system (ZebTec, Tecniplast). The presence of two independent systems provides a strong backup capability making the facility more resilient to instrumentation failures. The new system is capable of hosting 20 more tanks, for as many new fish lines, strongly augmenting the research potential of the facility in collaborations with the project partners. Each system is fully equipped to insure optimal water parameters for fish health and fecundity.

Generation of zebrafish transgenic lines for novel functional and pharmacological studies

The study of specific cell type involved in the development and maintenance of pain is of particular interest in the development of targeted therapy, thus reducing off-target effects. Schwann cells and oligodendrocytes are the glia cells that wrap peripheral and central nerves respectively. In addition to their established role as a protective barrier, Schwann cells/oligodendrocyte have recently emerged as key cellular element to sustain neuronal damage and chronic pain. Recent observations report how intracellular mechanisms in Schwann cells which encompass second messengers including calcium, cyclic AMP (cAMP), nitric oxide, play a fundamental role in the modulation of pain mechanisms. The zebrafish represents an unprecedented tool, hitherto poorly explored, to identify the role of central and peripheral glial cells in chronic pain.

To investigate the role of Schwann cells in chronic pain in zebrafish, we will use a library of fluorescent genetically encoded or fluorescence resonance energy transfer (FRET) biosensors, able to monitor the variation of the second messengers (calcium, cAMP, nitric oxide) involved in pain pathways over time at the cellular level. Specifically, zebrafish larvae (1-5 dpf) will be genetically modified to induce the cell-specific expression of the genetically encoded biosensors. A pan-neuronal and Schwann cells specific promoters will be used for the selective expression of genetically encoded biosensors in sensory neurons and Schwann cells, respectively. Then, the genetically modified zebrafish will be stimulated with proalgesic mediators, including, calcitonin gene related peptide (CGRP), bradykinin, prostaglandin E2 (PGE2), and will be monitored by confocal microscopy, by evaluating the fluorescence variations determined by the genetically encoded biosensors and indicative for a specific intracellular pathway activation.

In addition, transgenic zebrafish larvae expressing two different photoactivated protein (photoactivated adenylyl cyclase-bPAC, and phosphodiesterase-LAPD) in neurons/Schwann cells will be developed. This approach will allow to evaluate the role of cAMP in painful signaling pathways following the stimulation with proalgesic mediators.

Pilot study on the use zebrafish for drug screening in autism related disorders

Creatine transporter deficiency (CTD) is an X-linked inherited metabolic disease presenting with cerebral creatine deficiency due to loss of function mutations of the creatine transporter. The symptoms include early intellectual disability, autistic behavior and epilepsy. Although rare, CTD is a severe health care problem since it is a chronic condition with a strong impact on patients’ quality of life. No cure is available for CTD. Data obtained in subjects with creatine deficiency due to impaired biosynthesis show the potential reversibility of the deficits induced by creatine deficiency. Unfortunately, the approach used for these diseases does not produce substantial improvements in CTD. This project aims at generating a novel CTD fish model carrying mutations of the creatine transporter and providing a thorough characterization of the morphological, functional and behavioural impairments. This platform of analysis will be used for high-throughput screening of novel potential therapeutic molecules.

Pilot study on zebrafish to understand the role of the Schwann cell/oligodendrocyte lineage in the peripheral and central processing of tissue damage and chronic pain

After the development of the genetic models and the study of the intracellular pathways stimulated by proalgesic mediators, the behavioral activity of transgenic zebrafish larvae will be studied in vivo following administration of proalgesic substances. Given the role of Schwann cells in neuropathic pain models in rodents, a model of neurotrauma will be also developed. The TRPA1 channel, which is abundantly expressed in a subset of peptidergic primary sensory neurons and has been recently identified in Schwann cells/oligodendrocytes, where its activation amplifies neurotoxic and proalgesic processes, will be also explored by using pharmacological and genetic tools in these models. Behavioral test will be carried out by the assessment of spontaneous locomotor activity, light-dark preference (phototaxis), cognitive responses, social behaviors using the DanioVision system.

Pilot study on the use zebrafish and human organoids for getting insight into the brain-heart axis: developmental and (patho)physiological aspects

Between the nervous and the cardiovascular system, a complex network of interactions exists, characterized by a bidirectional communication that is functional in several (patho)physiological contexts. Many experimental and clinical studies showed the involvement of cortical and subcortical regions in controlling the cardiovascular function; similarly, neurological dysfunctions were associated to chronic cardiac conditions as atrial fibrillation and severe cardiac events as myocardial infarction. The molecular and physiological bases of heart-brain axis are studied in our in vivo and in vitro models. We are exploiting a transgenic zebrafish, model for chronic arrhythmias to understand the cardiac parasympathetic/sympathetic systems development, and the human induced pluripotent stem cell (iPSC) technology, that has generated significant enthusiasm for its potential application in basic and translational cardiac research. Human iPSC-derived cardiomyocytes (iPSC-CMs) offer an attractive tool to model cardiovascular diseases, accelerate predictive drug toxicology tests, study the earliest stages of human development and advance potential regenerative therapies. The advantage of iPSC-CMs is to eliminate confounding species-specific differences and, ultimately, pave the way for the development of personalized medicine for cardiovascular diseases. The expression of genes of interests can be manipulated in differentiating cardiomyocytes by infecting cells with engineered lentivirus. Sophisticated physiological analyses will be performed with patch clamp recording and “high throughput” recordings, a system developed in our laboratories. The results will be discussed and deeply related with clinics thanks our analysis performed in human cardiac samples from patients carrying arrhythmias, underwent to surgery, and the results obtained by our collaborators involved in the “Human Brain Mapping” project.

Dissemination activity, Biomedical Devices, Functional data analysis

Every researcher knows the struggle related to the interaction with non-expert public, both in terms of recruitment and dissemination of findings. The effort we are dedicating to public outreach is to soften this barrier and, thanks to this website, social media and other means, we want to increase the interest, and participation, of “non-staff” people.

Our experiments will be focused on biomedical data gathering and analysis. The sensors network will be used to record the following signals:

  • Brain activity
  • Electrodermal activity
  • Heart rate
  • Respiration frequency
  • Pupil and gaze patterns

From the listed signals we will extract information to evaluate cognitive and emotional indices, like workload, arousal, stress state and more. The multimodal data fusion will provide a complete picture of the subject state, allowing us to realize a deep cognitive/emotional profiling that may be used in a variety of fields. Define a personalized experience in an artistic show, highlight the most stressful moments during rehabilitation or during the learning phase are few of the dozens of possible applications.

Depending on the situation, both professional and commercial sensors will be utilized to get high-quality signals and, eventually, evaluate the potential of low-cost screening in low-income regions. Obviously AI will be integrated in many phases of the process, in particular with profile classification.

Finally, we will develop new sensors based on hyperspectral imaging (HSI). This technique has been selected for its ability to monitor and investigate important components of biological tissues. Specifically, biochemical species such as oxyhemoglobin and deoxyhemoglobin can be easily targeted with HSI, allowing to map tissue oxygenation delivery and oxygen consumption, differentiating diseased from healthy tissue in a real-time modality.