Session E64: Physics of the Brain: Structure and Dynamics

Oscillation onset and wave propagation in neocortex

In this talk, I will present recent findings on spatiotemporal patterns and wave propagation in neocortex based on intracortical microeletrode array recordings in primates during various states including optogenetic stimulation to probe neural dynamics, movement preparation and execution, and human epileptic seizures. I will follow with results based on neural-field models to account for the observed oscillation onset and wave propagation phenomena.
  

A Unification Framework for State Based Control of Seizures and Migraines

There is a several decade history demonstrating that electrical polarization of neurons can modulate neuronal firing, and that such polarization can suppress (or excite) spiking activity and seizures. In recent years, we uncovered a unification in the computational biophysics of spikes, seizures, and spreading depression (Wei et al, Journal of Neuroscience 34:11733-11743, 2014). These findings demonstrated that the repertoire of the dynamics of the neuronal membrane encompasses a broad range of dynamics ranging from normal to pathological, and that seizures and spreading depression are manifestations of the inherent properties of those membranes. Most recently, we have demonstrated that neuronal polarization can suppress (or enhance), block, or prevent spreading depression (Whalen et al, Scientific Reports, 8:8769, 2018), the physiological underpinning of migraine auras. Remarkably, this suppression requires qualitatively different stimulation from that required to suppress spikes and seizures, and is fully consistent with the computational biophysical models of spreading depression. Further unexpected findings from these experiments were that suppression of spreading depression does not generate seizures, and vice versa, that when the brain is in seizure activity suppression does not generate spreading depression. The implications are that in controlling brain dynamics from different states of the brain, there can be state dependent control which is qualitatively very different from that required in other states.   

Patient-Specific Large-Scale Brain Networks in Diseases

Neural network oscillations are a fundamental mechanism for cognition, perception and consciousness. Consequently, perturbations of network activity play an important role in brain disorders. Customization of healthcare with medical decisions tailored to the individual patient is a key aspect of personalized medicine.
When structural information from non-invasive brain imaging is merged with mathematical modelling, then generative brain network models constitute personalized in silico platforms for the exploration of causal mechanisms of brain function and clinical hypothesis testing. In partial epilepsy seizures originate in a local network, the so-called epileptogenic zone, before recruiting other brain regions. We build a Virtual Epileptic Patient (VEP) brain model that integrates patient-specific information, and we demonstrate that VEP derived from diffusion magnetic resonance imaging have sufficient predictive power to improve diagnosis and surgery outcome [1]. Individual brain models for 15 patients are validated against the presurgical electroencephalography data and the standard clinical evaluation, and are used to show that VEP brain models account for the patient seizure propagation patterns and explain the variability in postsurgical success [2].
Finally, we demonstrate how to develop novel personalized strategies towards therapy and intervention, by the example of applying linear stability analysis to the VEP in order to identify the minimal number of links that need to be removed for stopping the seizure propagation [3]. This suggests a less invasive paradigm of surgical interventions to treat and manage partial epilepsy.

References
[1] VK Jirsa et al. NeuroImage 145:377 (2017)
[2] T Proix et al. Brain 140: 641 (2017)
[3] S Olmi et al., Plos CB, [in press] (2019)   

Influence of the Human Amyloidogenic precursor on Epithelial and Neuronal Cell Membrane

The pathology of Alzheimer’s disease (AD) is correlated with the amyloid fiber formation. The Aβ (1-42) amyloid fragment is the principal species associated with senile plaque. There is evidence that the Aβ (1-42) amyloid precursors, such as the monomers, dimers, oligomers, and the proto-fibrils are toxic to the neuronal cell. It is believed that the Aβ (1-42) amyloid precursors perturb the neuronal cell membrane integrity leading to its death; however, the molecular mechanism by which the neuronal cell death occurs is not yet understood. Dielectric relaxation spectroscopy was used to study the interaction of the oligomeric Aβ (1-42) with the cell membrane of epithelial and neuronal cells. We incubated both cell types with the Aβ (1-42) solution of monomers, oligomers, and fibrils. We measured the dielectric response of the incubated cells at different concentrations over a wide range of frequencies (from 10^-2 Hz to 10⁷ Hz). Two dispersion processes (α, β) can be observed for all concentrations of the cell suspensions. We observed changes in the conductivity and permittivity of the neuronal cells when incubated with Aβ (1-42), which suggests the toxicity of the precursors causing marked alterations in the electrical properties of the cell membrane.
  

Impact of Damaged Neurons on Continuous Attractor Network Models of Grid Cells.

Grid cells in the dorsolateral band of the medial entorhinal cortex(dMEX) display strikingly regular firing responses to animal’s position in 2-D space. This helps animals be able to encode relative spatial location without reference to external cues. Within a continuous attractor model of grid cell activity^[1,2], we focus on the question of how two different kinds of damage to the dMEX that can arise from neurodegenerative diseases affect grid cell performance: I) randomly distributed discrete damage and II) diffusing damage that can arise from propagation of neurofibrillary tangles. Drawing on models from the existing literature, we employ 1- and 2-dimensional neural networks with background excitation sensitive to motion and a ring of inhibitory couplings modeled as a difference of Gaussians around each firing neuron. For sufficiently strong inhibition, the model always produces a stable 1-dimensional or 2-dimensional lattice. We will study the impact of the two damage types on the attractor model and contrast this with the impact of damage on an oscillatory grid cell model^[3].

  

Molecular Underpinnings of Postsynaptic Calmodulin-dependent Calcium Signaling

Calcium (Ca²⁺) signaling is a dynamic system where Ca²⁺ concentration fluctuates in range of 0.1-10μM with time. These short transient Ca²⁺ around the entry sites activate Ca²⁺-binding proteins such as calmodulin (CaM). The prototypical pathway describes CaM as encoding a Ca²⁺ signal by selectively activating downstream CaM-dependent proteins through molecular binding. However, CaM’s intrinsic Ca²⁺-binding properties alone appear insufficient to decode rapidly fluctuating Ca²⁺ signals. It has been proposed that the temporally varying mechanism for producing target selectivity requires CaM-target interactions that directly tune the Ca²⁺-binding properties of CaM through reciprocal interactions. I will focus on two unique and distinct CaM binding targets, neurogranin (Ng) and CaM-dependent kinase II (CaMKII), which are abundant in postsynaptic neuronal cells and are biochemically known to tune CaM’s affinity for Ca²⁺ in opposite directions. By employing an integrative approach of quantum mechanical calculations, all-atomistic molecular dynamics, and coarse-grained molecular simulations, we have revealed the molecular mechanisms of CaM’s reciprocal interaction between target binding and Ca²⁺binding.   

Dynamics of neuronal growth on surfaces with controlled geometries

Detailed knowledge of how the surface physical properties, such as mechanics, topography and texture influence neuronal growth and guidance is essential for understanding the processes that control neuron development, the formation of functional neuronal connections and nerve regeneration. I will present experimental results that demonstrate directional neuronal growth imparted by the surface geometry. We quantify axonal alignment and growth dynamics using a general stochastic model based on the Langevin equation. We relate the observed alignment in axonal growth to cellular contact guidance behavior, which results in an increase in the cell-substrate coupling on surfaces with micro-patterned structures. These results provide new insight into the role played by geometrical cues in neuronal growth and could lead to new methods for stimulating neuronal regeneration and the engineering of artificial neuronal tissue.   

Non-parametric discovery of population dynamics from large-scale neural activity recordings

Recent advances in neurotechnology enabled activity recordings from many neurons simultaneously, allowing us to study how neural populations are coordinated to drive behavior. Current population-analysis methods are based on fitting parametric models to data. However, these ad hoc models often distort dynamical features and result in ambiguous model comparisons. To overcome these limitations, we develop a non-parametric framework for discovering population dynamics directly from the data without a priori model assumptions. Our framework is based on latent Langevin dynamics, in which driving forces are directly optimized to effectively search the entire space of possible dynamics. The framework incorporates diverse, non-linear relationships between population-dynamics and firing-rates of single neurons. We derive a gradient descent algorithm for optimization over the space of continuous functions and use cross-validation and early stopping for regularization. Our framework accurately recovers qualitatively different population-dynamics simultaneously with diverse firing-rate profiles of single neurons.   

Tracking multineuronal activity in unrestrained animals with a random access two photon microscopy

Optical recordings of neuronal activity in freely behaving animals can reveal the correlation between the neural activity and behavioral outcome, such as decision making, learning, and multisensory integration. Such recordings require a microscopy that can overcome motion artifacts that accompany behavior. We recently developed a two-photon tracking microscope capable of recording activity from neurons in freely behaving larval fruit flies without motion artifacts[1]. However, due to the inertia of the scanning elements, the current technique is limited to recording only two neurons at a time. In order to study the correlation between neural activities and behavioral outcome across a range of neurons - sensory to motor neurons, it is essential to track multiple neurons at a time.

We extended the microscope using acousto-optic deflectors to relocate the laser beam in a constant time regardless of the distance between the two positions. Combined with improved feedback algorithms implemented in field-programmable gate arrays, this allows us to track and record the activities of many neurons at a high spatio-temporal resolution.

[1] D. Karagyozov, M. M. Skanata, A. Lesar, M. Gershow. Cell Reports. doi 10.1016/j.celrep.2018.10.013