Periodic Reporting for period 3 - NeuroDevo (Spontaneous and sensory-evoked activity shape neural circuits in the developing brain)
Période du rapport: 2022-05-01 au 2023-10-31
The brain of a developing fetus needs to acquire all the important connections than an individual will need later in life. It is still a puzzle how the brain transitions from being weakly and imprecisely connected in early development to being strongly yet precisely connected in adulthood enabling the generation of diverse computations and behaviors. There are many mechanisms that guide this process during development. Some of them are independent of neural activity, and are determined by interactions between different genes and molecules. Other mechanisms depend on specific and non-random patterns of neural activity. While in adulthood this neural activity is generated as we experience the sensory world around us, during early development many of the sensory organs are immature. Nonetheless, neural circuits in diverse brain regions generate neural activity spontaneously. Numerous experiments have demonstrated the importance of this spontaneous activity for organizing and fine-tuning connectivity in the developing brain. Perturbations in the patterns of spontaneous activity have fundamental impact on how neural circuits form in development and could lead to wiring deficits that underlie many developmental and psychiatric disorders including autism spectrum disorders and schizophrenia.
Despite recent technological progress in the recording and manipulation of spontaneous activity in the developing brain, we know little about the structure of spontaneous activity and this activity’s capacity to instruct the organization of local and brain-wide neural circuits. Using experiment-driven theory and modeling, this project aims to understand how spontaneous activity is generated and used to drive circuit organization and computation through a diversity of mechanisms operating at multiple timescales and spatial scales. We focus on activity-dependent mechanisms governing this process in the sensory cortex due to the large amount of available data in this system. This combination of quantitative data analysis, theory and computational models enables us to test the adequacy of specific assumptions one at a time, to explain experimental data from different systems and recording techniques, and to propose hypotheses which can be tested experimentally. Knowing the timing and interaction of mechanisms during normal development, could have important implications for the understanding, treatment and prevention of brain disorders, including intellectual disabilities.
Despite recent technological progress in the recording and manipulation of spontaneous activity in the developing brain, we know little about the structure of spontaneous activity and this activity’s capacity to instruct the organization of local and brain-wide neural circuits. Using experiment-driven theory and modeling, this project aims to understand how spontaneous activity is generated and used to drive circuit organization and computation through a diversity of mechanisms operating at multiple timescales and spatial scales. We focus on activity-dependent mechanisms governing this process in the sensory cortex due to the large amount of available data in this system. This combination of quantitative data analysis, theory and computational models enables us to test the adequacy of specific assumptions one at a time, to explain experimental data from different systems and recording techniques, and to propose hypotheses which can be tested experimentally. Knowing the timing and interaction of mechanisms during normal development, could have important implications for the understanding, treatment and prevention of brain disorders, including intellectual disabilities.
During this reporting period, we have made progress on Aims 2 and 3, which has yielded several published manuscripts and several more are in review. In Aim 2, we investigated the spatio-temporal correlations of spontaneous activity in the visual cortex of rodents using calcium imaging data from our experimental collaborators. We uncovered two types of spontaneous patterns: events with high amplitude involving many of the recorded cells whose source is intrinsic to the cortex, and events with a low amplitude involving few of the recorded cells whose source is the eye. Based on the analyzed properties, we built models at different spatial scales to study the implications of this spontaneous activity on the organization of synaptic inputs on single neurons as well as in networks of neurons.
In Wosniack et al. (eLife, 2021), we studied biological plausible plasticity rules that modify synaptic connection strengths in recurrently connected networks based on activity correlations between individual neurons. We found that the low amplitude events drive topographic connectivity preserving the order in visual scenes, while the high amplitude events regulate overall synaptic strength ensuring that networks maintain activity in normal ranges. In Kirchner and Gjorgjieva (Nature Comms 2021), we investigated how organization can emerge at the subcellular level through mechanisms of synaptic plasticity driven by spontaneous activity. Our model could not only explain the developmental emergence of a local form of synaptic organization, called clustering, but also unify diverse experimental data sets from different species.
Going further along development, in Aim 3 we developed network models to investigate how neural circuits self-organize in the presence of early sensory experience. In Eckmann and Gjorgjieva (bioRxiv 2022), we proposed that inhibition and the plasticity at inhibitory synapses sets up the circuits into highly organized structures that can execute numerous computations previously studied one at a time in circuits with hand-tuned connectivity. In Montangie et al. (PLoS CB 2020), we developed novel mathematical techniques to study in a principled way how network structures such as assemblies emerge from biologically plausible plasticity rules. In Wu et al. (PNAS 2020) and in our ongoing work, we use these plasticity rules to investigate their individual contribution in restoring network function following sensory perturbation.
In Wosniack et al. (eLife, 2021), we studied biological plausible plasticity rules that modify synaptic connection strengths in recurrently connected networks based on activity correlations between individual neurons. We found that the low amplitude events drive topographic connectivity preserving the order in visual scenes, while the high amplitude events regulate overall synaptic strength ensuring that networks maintain activity in normal ranges. In Kirchner and Gjorgjieva (Nature Comms 2021), we investigated how organization can emerge at the subcellular level through mechanisms of synaptic plasticity driven by spontaneous activity. Our model could not only explain the developmental emergence of a local form of synaptic organization, called clustering, but also unify diverse experimental data sets from different species.
Going further along development, in Aim 3 we developed network models to investigate how neural circuits self-organize in the presence of early sensory experience. In Eckmann and Gjorgjieva (bioRxiv 2022), we proposed that inhibition and the plasticity at inhibitory synapses sets up the circuits into highly organized structures that can execute numerous computations previously studied one at a time in circuits with hand-tuned connectivity. In Montangie et al. (PLoS CB 2020), we developed novel mathematical techniques to study in a principled way how network structures such as assemblies emerge from biologically plausible plasticity rules. In Wu et al. (PNAS 2020) and in our ongoing work, we use these plasticity rules to investigate their individual contribution in restoring network function following sensory perturbation.
Our network models of how different patterns of spontaneous activity influence the organization of network connectivity during development (Wosniack et al., eLife 2021) suggested that changes in spontaneous activity should occur simultaneously to connectivity refinements. To account for this, we postulated that spontaneous activity sparsifies over several weeks of development whereby large amplitude events decrease in frequency. We confirmed this prediction when we re-analyzed the experimental data. Although connectivity refinements and changes in spontaneous activity have been previously independently studied, we established a new link between the two proposing a specific way in which they interact.
In our work on the emergence of subcellular organization of synaptic inputs on the dendritic branches of single neurons (Kirchner and Gjorgjieva, Nature Coms 2021), we were initially only guided by experimental data pertaining to development. Building a mathematical model enabled us to conceptualize the framework and make predictions for how the developmental organization relates to the organization of synapses according to their selectivity to specific stimulus features observed in adulthood of different species. While different laboratories had reported results on different species independently, our theoretical model unified the results into a single framework and generated predictions for other species for which these properties have not yet been measured.
In the next stages of the project, as outlined in Aim 1, we will extend our network models so that they can also generate spontaneous activity from interactions among the neurons. This is in contrast to current approaches where spontaneous activity is introduced into the model with properties measured experimentally without asking how the circuit can produce it on its own, as is the case in biology. We hypothesize that developing neural circuits can generate spontaneous activity due to immature inhibition and specific intrinsic properties of the single neurons. Hence, until the end of the project we expect to have realized a network model that can simultaneously generate spontaneous activity and use it to refine network connectivity. We are also currently analyzing spontaneous activity from a mouse model of autism, and will aim to identify aspects of circuit architecture that are modified in these dysfunctional networks.
In our work on the emergence of subcellular organization of synaptic inputs on the dendritic branches of single neurons (Kirchner and Gjorgjieva, Nature Coms 2021), we were initially only guided by experimental data pertaining to development. Building a mathematical model enabled us to conceptualize the framework and make predictions for how the developmental organization relates to the organization of synapses according to their selectivity to specific stimulus features observed in adulthood of different species. While different laboratories had reported results on different species independently, our theoretical model unified the results into a single framework and generated predictions for other species for which these properties have not yet been measured.
In the next stages of the project, as outlined in Aim 1, we will extend our network models so that they can also generate spontaneous activity from interactions among the neurons. This is in contrast to current approaches where spontaneous activity is introduced into the model with properties measured experimentally without asking how the circuit can produce it on its own, as is the case in biology. We hypothesize that developing neural circuits can generate spontaneous activity due to immature inhibition and specific intrinsic properties of the single neurons. Hence, until the end of the project we expect to have realized a network model that can simultaneously generate spontaneous activity and use it to refine network connectivity. We are also currently analyzing spontaneous activity from a mouse model of autism, and will aim to identify aspects of circuit architecture that are modified in these dysfunctional networks.